Fossil Plant Cycle Chemistry

Effective December 6, 2006, this report has been made publicly available in accordance with Section 734.3(b)(3) and published in accordance with Section 734.7 of the U.S. Export Administration Regulations. As a result of this publication, this report is subject to only copyright protection and does not require any license agreement from EPRI. This notice supersedes the export control restrictions and any proprietary licensed material notices embedded in the document prior to publication.

Fossil Plant Cycle Chemistry Instrumentation and Control— State-of-Knowledge Assessment SED

R I A L

LICE

N

M AT E

Place Image Here

Fossil Plant Cycle Chemistry Instrumentation and Control— State-of-Knowledge Assessment 1012209

Final Report, March 2007

EPRI Project Manager K. Shields

ELECTRIC POWER RESEARCH INSTITUTE 3420 Hillview Avenue, Palo Alto, California 94304-1338 • PO Box 10412, Palo Alto, California 94303-0813 • USA 800.313.3774 • 650.855.2121 • [email protected] • www.epri.com

DISCLAIMER OF WARRANTIES AND LIMITATION OF LIABILITIES THIS DOCUMENT WAS PREPARED BY THE ORGANIZATION(S) NAMED BELOW AS AN ACCOUNT OF WORK SPONSORED OR COSPONSORED BY THE ELECTRIC POWER RESEARCH INSTITUTE, INC. (EPRI). NEITHER EPRI, ANY MEMBER OF EPRI, ANY COSPONSOR, THE ORGANIZATION(S) BELOW, NOR ANY PERSON ACTING ON BEHALF OF ANY OF THEM: (A) MAKES ANY WARRANTY OR REPRESENTATION WHATSOEVER, EXPRESS OR IMPLIED, (I) WITH RESPECT TO THE USE OF ANY INFORMATION, APPARATUS, METHOD, PROCESS, OR SIMILAR ITEM DISCLOSED IN THIS DOCUMENT, INCLUDING MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE, OR (II) THAT SUCH USE DOES NOT INFRINGE ON OR INTERFERE WITH PRIVATELY OWNED RIGHTS, INCLUDING ANY PARTY'S INTELLECTUAL PROPERTY, OR (III) THAT THIS DOCUMENT IS SUITABLE TO ANY PARTICULAR USER'S CIRCUMSTANCE; OR (B) ASSUMES RESPONSIBILITY FOR ANY DAMAGES OR OTHER LIABILITY WHATSOEVER (INCLUDING ANY CONSEQUENTIAL DAMAGES, EVEN IF EPRI OR ANY EPRI REPRESENTATIVE HAS BEEN ADVISED OF THE POSSIBILITY OF SUCH DAMAGES) RESULTING FROM YOUR SELECTION OR USE OF THIS DOCUMENT OR ANY INFORMATION, APPARATUS, METHOD, PROCESS, OR SIMILAR ITEM DISCLOSED IN THIS DOCUMENT. ORGANIZATION(S) THAT PREPARED THIS DOCUMENT Electric Power Research Institute (EPRI) Scientech, LLC

NOTICE: THIS REPORT CONTAINS PROPRIETARY INFORMATION THAT IS THE INTELLECTUAL PROPERTY OF EPRI. ACCORDINGLY, IT IS AVAILABLE ONLY UNDER LICENSE FROM EPRI AND MAY NOT BE REPRODUCED OR DISCLOSED, WHOLLY OR IN PART, BY ANY LICENSEE TO ANY OTHER PERSON OR ORGANIZATION.

NOTE For further information about EPRI, call the EPRI Customer Assistance Center at 800.313.3774 or e-mail [email protected]. Electric Power Research Institute, EPRI, and TOGETHER…SHAPING THE FUTURE OF ELECTRICITY are registered service marks of the Electric Power Research Institute, Inc. Copyright © 2007 Electric Power Research Institute, Inc. All rights reserved.

CITATIONS This report was prepared by Electric Power Research Institute (EPRI) 3420 Hillview Avenue Palo Alto, CA 94304 Principal Investigators K. Shields B. Syrett Scientech, LLC 2650 McCormick Drive Suite 300 Clearwater, FL 33759 Principal Investigators D. Meils J. Witherow This report describes research sponsored by the EPRI. The report is a corporate document that should be cited in the literature in the following manner: Fossil Plant Cycle Chemistry Instrumentation and Control—State-of-Knowledge Assessment. EPRI, Palo Alto, CA: 2007. 1012209.

iii

PRODUCT DESCRIPTION

Effective monitoring of the purity of water and steam is an integral part of a productive cycle chemistry monitoring program. EPRI’s cycle chemistry guidelines for fossil plants identify a group of core monitoring parameters that are considered the minimum requirements. Meeting the core monitoring requirement is part of EPRI’s cycle chemistry benchmarking criteria for plant cycle chemistry programs. In addition to the core parameters, many other chemistry parameters may be measured—either routinely or as needed for diagnostic and troubleshooting purposes. On-line monitoring of cycle chemistry is preferable to grab sample analysis. The EPRI report Reference Manual for On-Line Monitoring of Water Chemistry and Corrosion: 1998 Update (TR-112024) marked the last time this subject had been addressed. That state-of-art assessment was identified as the initial activity in a new project that considers both the need and opportunity for advancements in monitoring technology. Results and Findings This report describes the available technology options for monitoring a number of cycle chemistry parameters. The main focus is on methods available and in common use at fossil stations; however, information is also provided on a number of recently developed techniques as well as others that show promise. The report also considers techniques to monitor electrochemical corrosion potential and corrosion rate—two parameters not generally monitored in fossil plants and for which further development could lead to improved monitoring tools for fossil plants in the future. Challenges and Objectives The real-time monitoring of cycle chemistry supports operator oversight of water and steam purity, minimizing the time needed to 1) identify out-of-specification chemistry conditions and 2) implement appropriate corrective actions. It also provides a warning of plant equipment malfunction, facilitates the control of chemical additions, optimizes maintenance and repair schedules, and improves corrosion control. Fossil plant personnel involved in the evaluation, selection, operation, and maintenance of chemistry analyzers will find the information in this report useful. Information addressing the individual EPRI core monitoring parameters should be of particular value to users whose plants do not currently meet all of the core requirements. Applications, Value, and Use Despite the many advances in fossil plant cycle chemistry monitoring that have been achieved over the last 30 years, the current techniques have their limitations. In addition, resources needed to obtain, install, operate, and maintain analyzers to provide the information needed to control cycle chemistry are often an area of concern. Familiarity with the technology involved is therefore important when limited resources are invested in chemistry surveillance and control. v

EPRI Perspective Chemistry analyzers now in use require that samples be collected and conditioned. Information provided by the analyzers is useful for assessing conditions under which the potential exists for corrosion or other chemistry-related damage to cycle components. However, this information is indirect in that it does not determine the presence or magnitude (that is, rate) of corrosion. Research efforts are now investigating corrosion in boilers and turbines to enable a better understanding of its mechanisms and root causes in order to further the development of cycle chemistry guidelines. These future guidelines could likely be more effectively implemented if improved monitoring techniques were available. Therefore, investigations are planned to identify and develop monitoring techniques for the direct measurement of chemistry environments in which corrosion or deposition activity occurs. Approach The project team collected information from various sources for this report. These included the published literature, instrument manufacturer personnel and materials, consultants knowledgeable in chemistry instrumentation, the Internet, and relevant EPRI reports. Keywords Cycle chemistry Core monitoring parameter Instrumentation Analyzer Corrosion Surveillance

vi

EPRI Proprietary Licensed Material

CONTENTS

1 CYCLE CHEMISTRY MONITORING IN FOSSIL PLANT CYCLES ......................................1-1 1.1 Introduction ....................................................................................................................1-1 1.2 Monitoring Requirements and Choice of Cycle Chemistry.............................................1-1 1.3 Cycle Chemistry Monitoring Approaches .......................................................................1-3 1.3.1 Core Monitoring Parameters ..................................................................................1-3 1.3.2 Diagnostic Monitoring Parameters .........................................................................1-4 1.3 Future Cycle Chemistry Guidelines and Monitoring Approaches...................................1-5 1.4 References .....................................................................................................................1-7 2 AIR IN-LEAKAGE ..................................................................................................................2-1 2.1 Effects of Air In-leakage on Unit Performance and Cycle Chemistry .............................2-1 2.2 Monitoring Methods........................................................................................................2-2 2.2.1 Rotameters .............................................................................................................2-2 2.2.2 Early Advanced Air Removal Monitoring Systems .................................................2-3 2.2.3 Multisensor Probe Design ......................................................................................2-3 2.3 References .....................................................................................................................2-7 3 CARRYOVER IN DRUM BOILERS........................................................................................3-1 3.1 Introduction to Carryover................................................................................................3-1 3.2 Consideration of Carryover in EPRI Cycle Chemistry Guidelines ..................................3-2 3.3 Determination of Total Carryover from Drum Boilers .....................................................3-4 3.4 Sampling and Data Collection Considerations ...............................................................3-4 3.5 References .....................................................................................................................3-5 4 CONDUCTIVITY .....................................................................................................................4-1 4.1 Purpose and Use ...........................................................................................................4-1 4.2 Description .....................................................................................................................4-1 4.3 Technical Considerations ...............................................................................................4-9

vii


4.3.1 Cell Constants ........................................................................................................4-9 4.3.2 Cell Constant Determination or Verification..........................................................4-10 4.3.3 Cell Construction and Installation Considerations ................................................4-10 4.3.4 Cell Polarization....................................................................................................4-11 4.3.5 Temperature Effects .............................................................................................4-11 4.3.5.1 Solute Effects................................................................................................4-12 4.3.5.2 Solvent Effects ..............................................................................................4-12 4.3.6 Cation and Degassed Cation Conductivity Temperature Compensation .............4-13 4.3.7 Cation Conductivity Column Connections, Size, Flow and Flow Rate Considerations................................................................................................................4-13 4.3.8 Cation Conductivity Resin Exhaustion, Regeneration and Rinse-in .....................4-13 4.4 Interferences ................................................................................................................4-14 4.4.1 Organic and Strong Acids Interferences...............................................................4-14 4.4.2 Sample Line Interference .....................................................................................4-14 4.5 Calibration ....................................................................................................................4-15 4.6 Calibration Checks .......................................................................................................4-15 4.7 Alternative Methods for Determining Conductivity .......................................................4-15 4.4 End User Considerations .............................................................................................4-21 4.9 References ...................................................................................................................4-21 5 OXYGEN.................................................................................................................................5-1 5.1 Purpose and Use ...........................................................................................................5-1 5.2 Description of Methods...................................................................................................5-1 5.2.1 Galvanic Method.....................................................................................................5-2 5.2.2 Polarographic Method ............................................................................................5-3 5.2.3 Equilibrium Method.................................................................................................5-4 5.3 Technical Considerations ...............................................................................................5-5 5.3.1 Membrane Replacement ........................................................................................5-5 5.3.2 Electrode Cleaning .................................................................................................5-5 5.3.2.1 Polarographic..................................................................................................5-5 5.3.2.2 Galvanic ..........................................................................................................5-6 5.3.2.3 Equilibrium ......................................................................................................5-6 5.3.3 Temperature and Pressure Compensation for Sensors .........................................5-6 5.3.4 Flow Rate Sensitivity ..............................................................................................5-8

viii


5.3.5 Response Time ......................................................................................................5-8 5.3.5.1 Response Time After Air Calibration...............................................................5-8 5.3.5.2 Response Time Due to Changes in Dissolved Oxygen Concentration in the Process Stream .....................................................................................................5-9 5.4 Interferences ..................................................................................................................5-9 5.4.1 Oxygen Contamination ...........................................................................................5-9 5.4.2 Sample Conditioning ............................................................................................5-10 5.4.3 Electrolyte.............................................................................................................5-10 5.4.4 Stray Current ........................................................................................................5-10 5.4.5 Membrane Fouling, Positioning and Tension .......................................................5-10 5.4.6 Hydrogen ..............................................................................................................5-10 5.4.7 Other Cycle Chemistry Additives..........................................................................5-11 5.4.8 TDS ......................................................................................................................5-11 5.5 Calibration ....................................................................................................................5-11 5.5.1 Polarographic .......................................................................................................5-11 5.5.2 Galvanic................................................................................................................5-11 5.5.3 Equilibrium............................................................................................................5-12 5.5.4 Air Calibration for All Sensor Types......................................................................5-12 5.6 Calibration Checks for All Sensor Types......................................................................5-13 5.6.1 Zero Point for All Sensor Types............................................................................5-14 5.6.2 Maintenance for All Sensor Types........................................................................5-15 5.7 Alternative Methods......................................................................................................5-15 5.7.1 Luminescent Oxygen Sensors [17].......................................................................5-15 5.7.1.1 µg·kg-1–Resolution Optical Sensor to Monitor Dissolved Oxygen ...............5-18 5.8 End User Considerations .............................................................................................5-19 5.9 References ...................................................................................................................5-20 6 OXIDATION-REDUCTION POTENTIAL ................................................................................6-1 6.1 Purpose and Use ...........................................................................................................6-1 6.2 Description of Method ....................................................................................................6-1 6.3 Technical Considerations ...............................................................................................6-4 6.3.1 Voltmeter Selection ................................................................................................6-4 6.3.2 Reference Electrodes .............................................................................................6-5 6.3.3 ORP Sensing Electrode..........................................................................................6-5

ix


6.4 Interferences ..................................................................................................................6-7 6.5 Calibration ......................................................................................................................6-7 6.5.1 Calibration Checks .................................................................................................6-9 6.6 End User Considerations ...............................................................................................6-9 6.7 References ...................................................................................................................6-10 7 pH ...........................................................................................................................................7-1 7.1 Purpose and Use ...........................................................................................................7-1 7.2 Description .....................................................................................................................7-1 7.3 Technical Considerations ...............................................................................................7-2 7.3.1 Sensing (Glass) Electrode......................................................................................7-2 7.3.2 Reference Electrode...............................................................................................7-3 7.3.3 Temperature Effects ...............................................................................................7-5 7.3.3.1 Electrode Effects.............................................................................................7-6 7.3.3.2 Solution Additive Effects .................................................................................7-7 7.4 Interferences ..................................................................................................................7-8 7.4.1 Interfering Ions........................................................................................................7-8 7.4.2 Interfering Stray Currents .......................................................................................7-8 7.5 Calibration ......................................................................................................................7-9 7.6 Calibration Checks .......................................................................................................7-10 7.7 Alternative Methods for Determining pH ......................................................................7-11 7.8 End User Considerations .............................................................................................7-14 7.9 References ...................................................................................................................7-15 8 SODIUM..................................................................................................................................8-1 8.1 Purpose and Use ...........................................................................................................8-1 8.2 Description of Method ....................................................................................................8-1 8.3 Technical Considerations ...............................................................................................8-4 8.3.1 Sensing Electrode ..................................................................................................8-4 8.3.2 Reference Electrode...............................................................................................8-7 8.4 Interferences ..................................................................................................................8-8 8.5 Calibration ......................................................................................................................8-9 8.6 Calibration Checks .......................................................................................................8-11 8.7 Alternative Methods......................................................................................................8-12

x


8.8 End User Considerations .............................................................................................8-13 8.9 References ...................................................................................................................8-13 9 AMMONIA ..............................................................................................................................9-1 9.1 Purpose and Use ...........................................................................................................9-1 9.2 Description of Methods...................................................................................................9-2 9.2.1 Background ............................................................................................................9-2 9.2.2 Conductivity Approximation ....................................................................................9-3 9.2.2.1 Conductivity Approximation Limitations ..........................................................9-3 9.2.3 Ion Selective Electrode...........................................................................................9-4 9.2.4 Colorimetry .............................................................................................................9-5 9.3 Technical Considerations ...............................................................................................9-6 9.3.1 Ion Selective Electrodes (ISE)................................................................................9-6 9.3.1.1 Ammonium Sensing Electrode........................................................................9-6 9.3.1.2 Reference Electrode .......................................................................................9-6 9.3.2 Colorimetric Analyzers............................................................................................9-7 9.3.2.1 Colorimetric Limitations...................................................................................9-8 9.3.2.2 Sample Considerations.................................................................................9-11 9.3.2.3 Time Delay....................................................................................................9-11 9.3.2.4 Sample Temperature ....................................................................................9-11 9.3.2.5 Sample Volume ............................................................................................9-12 9.4 Interferences ................................................................................................................9-12 9.4.1 ISE Interferences..................................................................................................9-12 9.4.2 Colorimetric Interferences ....................................................................................9-13 9.5 Calibration ....................................................................................................................9-13 9.5.1 ISE Calibration......................................................................................................9-13 9.5.2 Colorimetric Calibration ........................................................................................9-14 9.6 Calibration Checks .......................................................................................................9-14 9.7 Alternative Methods......................................................................................................9-15 9.7.1 Direct Nesslerization.............................................................................................9-15 9.7.2 Titrimetric Ammonia Determination ......................................................................9-16 9.7.3 ISE – Gas Permeable Membrane.........................................................................9-16 9.8 End User Considerations .............................................................................................9-17 9.9 Field Experience...........................................................................................................9-17

xi


9.9.1 Sample Delivery ...................................................................................................9-17 9.9.2 Analyzer Malfunction ............................................................................................9-18 9.10 References .................................................................................................................9-18 10 CHLORIDE .........................................................................................................................10-1 10.1 Purpose and Use........................................................................................................10-1 10.2 Description of Method ................................................................................................10-2 10.3 Technical Considerations ...........................................................................................10-4 10.3.1 Sensing Electrode ..............................................................................................10-4 10.3.2 Reference Electrode...........................................................................................10-4 10.3.3 Temperature Considerations ..............................................................................10-5 10.4 Interferences ..............................................................................................................10-6 10.5 Calibration ..................................................................................................................10-6 10.6 Calibration Checks .....................................................................................................10-9 10.7 Alternative Methods ...................................................................................................10-9 10.7.1 ASTM D512 Test Method A—Mercurimetric Titration ........................................10-9 10.7.2 ASTM D512 Test Method B—Silver Nitrate Titration .........................................10-9 10.7.3 Standard Methods: Method 4500 - Cl D. Potentiometric Method [11] .............10-10 10.7.4 Standard Methods: Method 4500 - Cl E. Automated Ferricyanide Method [11]................................................................................................................................10-10 10.7.5 Ion Chromatography.........................................................................................10-10 10.8 End User Considerations .........................................................................................10-10 10.9 References ...............................................................................................................10-11 11 HYDRAZINE .......................................................................................................................11-1 11.1 Purpose and Use........................................................................................................11-1 11.2 Description of Methods...............................................................................................11-2 11.3 Technical Considerations ...........................................................................................11-2 11.3.1 Colorimetry .........................................................................................................11-2 11.3.1.1 Colorimetric Limitations...............................................................................11-3 11.3.2 Amperometry ......................................................................................................11-4 11.3.2.1 Two Electrode Method ................................................................................11-4 11.3.2.2 Three Electrode Method .............................................................................11-5 11.3.3 Iodide Ion Selective Electrode Method ...............................................................11-6 11.4 Interferences ..............................................................................................................11-7

xii


11.4.1 Colorimetry .........................................................................................................11-7 11.4.2 Amperometry ......................................................................................................11-8 11.4.3 Ion Selective Electrode Method..........................................................................11-8 11.5 Calibration ..................................................................................................................11-8 11.5.1 Colorimetry .........................................................................................................11-8 11.5.2 Amperometry ......................................................................................................11-9 11.5.3 Ion Selective Electrode Method..........................................................................11-9 11.6 Calibration Checks ...................................................................................................11-10 11.7 Alternative Methods .................................................................................................11-10 11.8 End User Considerations .........................................................................................11-11 11.8.1 Recognizing Instrument Malfunctions...............................................................11-12 11.8.1.1 Sample Delivery........................................................................................11-12 11.8.1.2 Analyzer Malfunction.................................................................................11-12 11.9 References ...............................................................................................................11-13 12 HYDROGEN .......................................................................................................................12-1 12.1 Purpose and Use........................................................................................................12-1 12.2 Discussion of Methods ...............................................................................................12-1 12.2.1 Electrochemical / Polarographic .........................................................................12-2 12.2.2 Thermal Conductivity..........................................................................................12-2 12.2.3 Gas Density Meter..............................................................................................12-2 12.3 Technical Considerations ...........................................................................................12-3 12.3.1 Electrochemical / Polarographic .........................................................................12-3 12.3.2 Thermal Conductivity..........................................................................................12-5 12.3.3 Gas Density Meter..............................................................................................12-8 12.4 Calibration ..................................................................................................................12-9 12.5 Calibration Check .......................................................................................................12-9 12.6 Alternative Methods ...................................................................................................12-9 12.7 End User Considerations .........................................................................................12-10 12.8 References ...............................................................................................................12-11 13 ION CHROMATOGRAPHY ................................................................................................13-1 13.1 Purpose and Use........................................................................................................13-1 13.2 Description of Method ................................................................................................13-1

xiii


13.2.1 Ion Exchange Chromatography..........................................................................13-3 13.2.2 Ion Exclusion Chromatography ..........................................................................13-7 13.2.3 Ion Pair Chromatography ...................................................................................13-7 13.3 Technical Considerations ...........................................................................................13-7 13.3.1 Sample Injection .................................................................................................13-7 13.3.2 Column Selection .............................................................................................13-10 13.3.2.1 Concentrator Column................................................................................13-11 13.3.2.2 Guard Column...........................................................................................13-11 13.3.2.3 Separator Column.....................................................................................13-11 13.3.2.4 Eluent Suppressor Column .......................................................................13-12 13.4 Eluent Selection .......................................................................................................13-13 13.5 Detectors ..................................................................................................................13-17 13.5.1 Interferences.....................................................................................................13-19 13.6 Calibration ................................................................................................................13-21 13.6.1 Calibration Checks ...........................................................................................13-21 13.7 Alternative Methods .................................................................................................13-21 13.8 End User Considerations .........................................................................................13-22 13.9 References ...............................................................................................................13-22 14 IRON AND COPPER ..........................................................................................................14-1 14.1 Purpose and Use........................................................................................................14-1 14.2 Description of Methods...............................................................................................14-1 14.3 Technical Considerations ...........................................................................................14-2 14.3.1 Integrated Sampling ...........................................................................................14-2 14.3.1.1 Wet Chemistry Analysis of Integrated Samples..........................................14-3 14.3.1.2 XRF Analysis of Integrated Samples ..........................................................14-3 14.3.2 Turbidity..............................................................................................................14-4 14.3.3 Particle Counter..................................................................................................14-5 14.3.4 Dynamic Light Fluctuation Based Particle Monitors ...........................................14-6 14.3.5 Acoustic Detection..............................................................................................14-6 14.3.6 Colorimetric ........................................................................................................14-7 14.3.6.1 Colorimetric Limitations...............................................................................14-8 14.4 Calibration ..................................................................................................................14-9 14.4.1 Wet Chemistry Analysis......................................................................................14-9

xiv


14.4.2 XRF Analysis ....................................................................................................14-10 14.5 Calibration Check .....................................................................................................14-10 14.6 Alternate Methods ....................................................................................................14-11 14.7 End User Considerations .........................................................................................14-11 14.8 References ...............................................................................................................14-12 15 PHOSPHATE......................................................................................................................15-1 15.1 Purpose and Use........................................................................................................15-1 15.2 Description of Method ................................................................................................15-1 15.3 Technical Considerations ...........................................................................................15-3 15.3.1 Colorimetric Limitations ......................................................................................15-3 15.3.2 Sample Considerations ......................................................................................15-4 15.3.3 Time Delay .........................................................................................................15-5 15.3.4 Sample Temperature..........................................................................................15-5 15.3.5 Sample Volume and Introduction of Chemicals..................................................15-5 15.3.6 Light Intensity .....................................................................................................15-5 15.4 Interferences ..............................................................................................................15-6 15.4.1 Particulate Matter ...............................................................................................15-6 15.4.2 Sample Discoloration..........................................................................................15-6 15.4.3 Other Substances...............................................................................................15-6 15.5 Calibration ..................................................................................................................15-7 15.6 Calibration Check .......................................................................................................15-7 15.7 Alternative Methods ...................................................................................................15-7 15.8 End User Considerations ...........................................................................................15-8 15.9 Field Experience.........................................................................................................15-8 15.9.1 Sample Delivery .................................................................................................15-9 15.9.2 Analyzer Malfunction ..........................................................................................15-9 15.10 References .............................................................................................................15-10 16 SILICA ................................................................................................................................16-1 16.1 Purpose and Use........................................................................................................16-1 16.2 Description of Method ................................................................................................16-1 16.3 Technical Considerations ...........................................................................................16-3 16.3.1 Colorimetric Limitations ......................................................................................16-3

xv


16.3.2 Sample Considerations ......................................................................................16-4 16.3.3 Time Delay .........................................................................................................16-5 16.3.4 Sample Temperature..........................................................................................16-5 16.3.5 Sample Volume and Introduction of Chemicals..................................................16-5 16.3.6 Light Intensity Check ..........................................................................................16-6 16.4 Interferences ..............................................................................................................16-6 16.4.1 Ortho-phosphate.................................................................................................16-6 16.4.2 Non-reactive Silica..............................................................................................16-6 16.4.3 Particulate Matter ...............................................................................................16-7 16.4.4 Reagent Contamination......................................................................................16-7 16.4.5 Sample Discoloration..........................................................................................16-7 16.4.6 Sample Temperature..........................................................................................16-7 16.5 Calibration ..................................................................................................................16-7 16.6 Calibration Checks ................................................................................................16-8 16.7 Alternative Methods ...................................................................................................16-9 16.8 End User Considerations ...........................................................................................16-9 16.9 Field Experience.......................................................................................................16-10 16.9.1 Sample Delivery ...............................................................................................16-10 16.9.2 Analyzer Malfunction ........................................................................................16-10 16.10 References .............................................................................................................16-11 17 TOTAL ORGANIC CARBON .............................................................................................17-1 17.1 Purpose and Use........................................................................................................17-1 17.2 Forms of Organic Carbon and Description of Method................................................17-2 17.3 Technical Considerations ...........................................................................................17-3 17.3.1 High Temperature Combustion Method .............................................................17-3 17.3.2 Persulfate Oxidation ...........................................................................................17-4 17.3.3 Conductometric-type TOC..................................................................................17-6 17.4 Calibration ..................................................................................................................17-8 17.5 Calibration Check .......................................................................................................17-8 17.6 Alternative Methods ...................................................................................................17-9 17.6.1 Closed-Loop Photocatalytic Oxidation................................................................17-9 17.6.2 TOC Data Comparison .....................................................................................17-11 17.7 End User Considerations .........................................................................................17-11

xvi


17.8 References ...............................................................................................................17-12 18 ELECTROCHEMICAL CORROSION POTENTIAL ...........................................................18-1 18.1 Purpose and Use........................................................................................................18-1 18.2 Description of Method ................................................................................................18-2 18.2.1 Use of Reference Electrodes..............................................................................18-3 18.3 Technical Considerations ...........................................................................................18-4 18.3.1 Reference Electrode Selection ...........................................................................18-4 18.3.2 Quasi-Reference Electrodes Used in Corrosion Rate Probes............................18-5 18.3.3 Reference Electrode Issues ...............................................................................18-6 18.3.4 Voltmeter Selection ............................................................................................18-7 18.4 Calibration and Maintenance Procedures ..................................................................18-7 18.4.1 Reference Electrodes .........................................................................................18-7 18.4.2 Electrochemical Corrosion Potential Probe ........................................................18-8 18.5 Field Experience.........................................................................................................18-8 18.5.1 ECP Measurements in Condensers ...................................................................18-8 18.5.2 ECP Measurements in Boiling Water Reactors ..................................................18-9 18.6 Possible Future ECP Measurements in Fossil Plants ................................................18-9 18.7 References ...............................................................................................................18-10 19 CORROSION RATE ...........................................................................................................19-1 19.1 Purpose and Use........................................................................................................19-1 19.2 Description of Traditional Methods.............................................................................19-1 19.3 Advantages of On-Line Corrosion Monitoring ............................................................19-2 19.4 Description of Physical Methods ................................................................................19-3 19.5 Description of Electrochemical Methods ....................................................................19-5 19.5.1 General Considerations......................................................................................19-6 19.5.2 Linear Polarization Resistance (3-Electrode) .....................................................19-8 19.5.3 Linear Polarization Resistance (2-Electrode) ...................................................19-11 19.5.4 Electrochemical Impedance Spectroscopy.......................................................19-13 19.5.5 Coupling Current (Zero Resistance Ammetry) .................................................19-14 19.5.6 Electrochemical Noise ......................................................................................19-15 19.5.6.1 Electrochemical Potential Noise ...............................................................19-15 19.5.6.2 Electrochemical Current Noise .................................................................19-17

xvii


19.5.6.3 Electrochemical Noise Resistance............................................................19-18 19.5.7 Galvanic Corrosion Monitoring Using Zero Resistance Ammetry.....................19-18 19.5.8 Coupling Current Between Two or More Metals...............................................19-19 19.5.9 Coupling Current Between Segmented Weld Electrodes .................................19-20 19.5.10 Electrochemical Methods in Combination ......................................................19-20 19.6 Calibration ................................................................................................................19-21 19.7 End User Considerations .........................................................................................19-21 19.8 Possible Future Corrosion Rate Monitoring in Fossil Plants ....................................19-22 19.9 References ...............................................................................................................19-22

xviii


LIST OF FIGURES Figure 2-1 Multisensor Probe.....................................................................................................2-5 Figure 2-2 Multisensor Probe Instrument Schematic .................................................................2-6 Figure 3-1 Representative Drum Boiler Mechanical Carryover..................................................3-3 Figure 4-1 A Typical Cation Conductivity Flow Diagram............................................................4-4 Figure 4-2 Chloride (Cl) Concentration vs. Specific Conductivity ..............................................4-5 Figure 4-3 Sulfate (SO4) Concentration vs. Specific Conductivity..............................................4-6 Figure 4-4 Carbon Dioxide (CO2) vs. Specific Conductivity .......................................................4-7 Figure 4-5 Typical Degassed Cation Conductivity Schematic Diagram.....................................4-8 Figure 4-6 Relationship Between Ammonia Concentration mg/L (ppm) and Specific Conductivity (µS/cm) at 25°C (77°F) ................................................................................4-16 Figure 5-1 Typical Oxygen Sensing Probe ................................................................................5-2 Figure 5-2 Solubility of Dissolved Oxygen (mg/L (ppm)) vs. Temperature (°C) .........................5-7 Figure 5-3 Response of a Polarographic Oxygen Sensor to a 20 µg/L (ppb) Oxygen Addition Generated by a Faraday Cell .............................................................................5-14 Figure 5-4 Principle of Optical Oxygen Detection Using Fluorescent Dye...............................5-16 Figure 5-5 Fluorescence Density Decay Time as a Function of Oxygen Concentration..........5-16 Figure 5-6 Phase Shift of Modulated Signals...........................................................................5-17 Figure 5-7 Stern-Volmer Calibration Curve..............................................................................5-18 Figure 5-8 Luminescence Sensor Design................................................................................5-19 Figure 7-1 Flowing Junction Reference Electrode Head Cup Configuration..............................7-5 Figure 7-2 Combination Electrode .............................................................................................7-6 Figure 7-3 Standardization (Zero Intercept) Adjustment............................................................7-9 Figure 7-4 Slope (Span) Adjustment........................................................................................7-10 Figure 7-5 Specific Conductivity, Ammonia, pH at 25°C..........................................................7-12 Figure 7-6 Ammonia Concentration vs. pH for Various Carbon Dioxide Concentrations.........7-13 Figure 8-1 Measured Sodium Concentration vs. Concentration of Sodium Added....................8-3 Figure 8-2 Response Time of Sodium Ion Selective Sensors: Time vs. Sodium Concentration in µg/L (ppb) Before and After Etching .......................................................8-5 Figure 8-3 Elapsed Time After Known Standard Addition: Time vs. Sodium Concentration in µg/L (ppb)................................................................................................8-6 Figure 8-4 Sodium Analyzer Response: Elapsed Time After Addition of 20µg/L (ppb) Sodium vs. Sodium Concentration.....................................................................................8-7

xix


Figure 8-5 Calibration Slope (Millivolt Response) of Sodium Ion Selective Electrode at Varying pH Values .............................................................................................................8-9 Figure 8-6 Verification Results Immediately After a 200 and 2,000 µg/L (ppb) Sodium Calibration ........................................................................................................................8-10 Figure 9-1 Percent of Ammonia and Ammonium Ions as a Function of Solution pH .................9-3 Figure 9-2 Relationship between Ammonia, Specific Conductivity and pH in the Absence of Carbon Dioxide at 25ºC..................................................................................................9-4 Figure 9-3 Chemical Formula and Structure for Phenol ............................................................9-7 Figure 9-4 Chemical Formula and Structure for Salicylic Acid...................................................9-8 Figure 9-5 Absorbance vs. Concentration ...............................................................................9-10 Figure 10-1 Thermo Orion Electrode Reservoir System ..........................................................10-5 Figure 10-2 Calibration Curve for Low Level Chloride SIE ......................................................10-8 Figure 11-1 Absorption vs. Concentration ...............................................................................11-4 Figure 12-1 Polarographic Hydrogen Sensing Electrode with “Guard Ring”............................12-4 Figure 12-2 The Wheatstone Bridge Circuit Showing Schematically How the Reference Gas Flows Over Filament Resistances Ra and Rc, while the Sample Flows Over Filament Resistances Rd and Rb. Ra = Rb = Rc = Rd when Reference and Sample Gases have the Same Thermal Conductivity ...................................................................12-6 Figure 12-3 Diagram of Gas Chromatograph with Thermal Conductivity Detector..................12-7 Figure 12-4 Hydrogen Sensor with Thermal Conductivity Detector .........................................12-8 Figure 13-1 Schematic of a Single Column Ion Chromatograph .............................................13-4 Figure 13-2 Illustration of How Ions Elute (Leave the Separator Column) at Different Rates Resulting in a Separation of Ionic Species in the Flowing Eluent ..........................13-5 Figure 13-3 Gradient Separation of Common Anions Using a Hydroxide Gradient.................13-6 Figure 13-4 Anion Analysis Using Ionpac® AS17 Separator Column with Eluent Generator; Method 1 ........................................................................................................13-9 Figure 13-5 Anion Analysis Using Ionpac® AS17 Separator Column with Eluent Generator; Method 2 ......................................................................................................13-10 Figure 13-6 Chemistry and Ion Movement in Continually Regenerated Eluent Suppressor.....................................................................................................................13-14 Figure 13-7 The KOH Generator Cartridge Consists of a KOH Generating Chamber and + K Electrolyte Reservoir Connected by a Cation Exchange Connector .........................13-16 Figure 13-8 Carbonate Removal Device................................................................................13-17 Figure 13-9 Comparison of Spectrum with (Top Spectrum) and without Carbonate Removal Cevice ............................................................................................................13-18 Figure 13-10 Chromatogram of a Sample Containing 0.022 µg/l Sodium and 3000µg/l Ethanolamine .................................................................................................................13-20 Figure 14-1 Particle Counter Schematic, Illustrating 1 Particle at 3 microns and 2 Particles at 1 micron.........................................................................................................14-5 Figure 14-2 Dynamic Light Fluctuation Schematic ..................................................................14-6 Figure 14-3 Illustration of Self Absorption at High Concentrations ..........................................14-9

xx


Figure 15-1 Typical Beer’s Law Absorption (A) vs. Concentration Curve ................................15-3 Figure 16-1 Graph Showing 100 Percent Absorption ..............................................................16-4 Figure 17-1 High Temperature Oxidation TOC Analyzer Flow Diagram..................................17-5 Figure 17-2 Persulfate Oxidation, Ultraviolet Lamp TOC Analyzer Flow Diagram...................17-7 Figure 18-1 An Analogy between Measuring Mountain Height with Respect to Sea Level and Measuring ECP with Respect to a Reference Electrode...........................................18-4 Figure 19-1 Equivalent Electric Circuit (Upper Figure) and Corresponding Schematic Diagram of the 3-Electrode Corrosion Probe Immersed in the Corrosive Environment (Lower Figure)...........................................................................................19-10 Figure 19-2 View of the Exposed End of a Corrosion Probe with Three Concentric Flushmounted Electrodes Arranged to Minimize Solution Resistance Errors.........................19-11 Figure 19-3 Equivalent Electric Circuit for the 2-Electrode Corrosion Probe Immersed in the Corrosive Environment.............................................................................................19-12

xxi


LIST OF TABLES Table 1-1 EPRI Cycle Chemistry Core Monitoring Parameters for Conventional Fossil Units ...................................................................................................................................1-3 Table 2-1 Example Diagnostic Capabilities of a Five-Probe Air In-leakage Measurement System ...............................................................................................................................2-6 Table 2-2 MSP Probe Indications for Various Probe Positions..................................................2-7 Table 4-1 Typical Conductivity Range Limits as a Function of Cell Constant............................4-9 Table 4-2 Equivalent Conductances of Separate Ions at Various Temperatures ....................4-18 Table 6-1 To Convert ORP or ECP Values Measured Using Reference Electrode #1 to Values on Reference Electrode #2 Scale, Add the Indicated Conversion Factor to the Measured Potential ......................................................................................................6-6 Table 6-2 Expected ORP Values for Reference Quinhydrone Solutions at pH 4 and pH 7.......6-8 Table 7-1 Various Solution Additive Temperature Correction Factors for Power Plant Steam/Water Cycle pH Measurements ..............................................................................7-7 Table 7-2 Example Calculated pH by Differential Conductivity from One Instrument Supplier ............................................................................................................................7-14 Table 8-1 Typical Results from Known Addition Method for Calibration Check in the nanograms/L (ppt) range..................................................................................................8-12 Table 9-1 Precision Data for Manual Phenate Method Based on Triplicate Analyses of Ammonium Sulfate.............................................................................................................9-9 Table 9-2 Precision and Bias Data; Direct Nesslerization .......................................................9-15 Table 10-1 Boiler Water Chloride Limits in µg/L (ppb) @ 15858 kPa, (2300 psi) ....................10-2 Table 12-1 Some Performance Ranges for Aqueous EC and TC Detectors .........................12-10 Table 17-1 Summary of Some Typical On-Line Instrument Capabilities ...............................17-12 Table 19-1 Typical Examples of the Relationship Between the EPN “Fingerprint”, Statistical Parameters, and the Corrosion Mechanism in Progress ...............................19-17

xxiii


1 CYCLE CHEMISTRY MONITORING IN FOSSIL PLANT CYCLES

1.1 Introduction Technology advancements made during the twentieth century enabled substantial improvement to be made in the efficient and cost effective production of electric power from fossil fuels. In conventional designs in which steam is produced and expanded in a turbine, the ability to provide water at higher purity levels than previously possible enabled designers to increase the operating temperatures and pressures of fossil units. These improvements were accompanied by increasingly tighter chemistry limits and a need for reliable monitoring approaches that could be used for surveillance and control purposes. This need shifted the emphasis on chemistry monitoring from infrequent collection and analysis of grab samples to reliance on analyzers that provided semi-continuous or real time monitoring of the chemistry parameters of interest. Also, a trend of reductions in fossil plant staff size and experience levels have dictated increasing reliance on the performance of chemistry analyzers and the reliability of chemistry monitoring data. This is of some concern in all situations and, in particular, those where responsibility for day to day chemistry activities has been assigned to plant operators without extensive relevant education or training. Advancements in data transmission, display, storage and assessment (including expert systems) as well as general communications have made these changes in staffing possible. However, effective use of these capabilities remains completely dependent on the suitability of the cycle chemistry treatments selected and the associated monitoring capability of the selected analyzers.

1.2 Monitoring Requirements and Choice of Cycle Chemistry Over the last 20 years EPRI has taken a leadership position in the area of cycle chemistry control in fossil generating units. The third generation of EPRI Cycle Chemistry Guidelines was introduced over the period 2001-2005. The guidelines are widely accepted as the de facto worldwide standard; available options for feedwater and boiler water treatment and operational chemistry target values and action levels for conventional fossil units are summarized in a series of reports [1-3]. Effective cycle chemistry programs require proper selection of treatments consistent with unit characteristics. Treatment control and optimization requires provision of sampling and analysis capabilities consistent with the treatments selected for use. The current guidelines indicate that 1-1

EPRI Proprietary Licensed Material Cycle Chemistry Monitoring in Fossil Plant Cycles

there are three options for the feedwater treatment and four choices for boiler water treatment. The feedwater treatments are designated and may be generally characterized as follows. •

Reducing all-volatile treatment, AVT(R), in which the feedwater must be dosed with a reducing agent to minimize corrosion of copper alloy components

•

Oxidizing all-volatile treatment, AVT(O), in which no reducing agent is added or needed since there are no copper alloys in the feedwater system

•

Oxygenated treatment, OT, in which the feedwater is dosed with oxygen; OT can only be used in cycles where the feedwater meets necessary purity criteria, employs condensate polishing and no copper alloys are present following the condensate pump discharge.

The available boiler water treatments are identified and generally characterized as follows. •

All-volatile treatment [1], AVT, in which the feedwater treatment is either AVT(R) or AVT(O) as appropriate and no further chemical treatment is applied within the boiler.

•

Phosphate Continuum Treatment [2], PC, in which a drum boiler is dosed principally with trisodium phosphate (TSP); dosing of the boiler water with sodium hydroxide is also allowed as a supplement to TSP when needed for pH control. Low and high level TSP dosage variants of PC, termed PC(L) and PC(H), respectively may be considered depending on the characteristics and needs of the unit.

•

Caustic Treatment [2], CT, in which a drum boiler is dosed with sodium hydroxide.

•

Oxygenated Treatment [3], OT, in which the feedwater dosed with oxygen as needed for OT is used in the boiler without further treatment; OT may be used in cycles with once-through or drum type boilers, however, with the latter feedwater oxygen dosing must be more carefully controlled so as to avoid possible corrosion in the boiler.

Additional details concerning selection, use and optimization of each treatment is provided in the corresponding guideline report. These reports also provide guidance concerning required sample points and chemistry parameters that should be monitored for the various chemistry treatment options. This guidance is based on worldwide experience in ensuring minimized levels of corrosion, impurity ingress and deposition as needed to eliminate chemistry related damage in boilers and turbines while simultaneously minimizing, or even eliminating, the need for operational chemical cleaning of boilers. The chemistry parameters that may be monitored under the cycle chemistry guidelines fit into two general categories: 1) those parameters which all fossil plants should have for optimum chemistry control (termed the core parameters) and 2) those parameters which are regarded as diagnostic parameters that may be monitored as needed for troubleshooting or during commissioning. The rationale for each important sample point and chemistry parameter combination is discussed in detail in the individual guideline reports to explain its designation as an EPRI core or optional (diagnostic) monitoring requirement. The findings of EPRI cycle chemistry research for conventional fossil plant cycles are also suitable for application to combined cycle units with heat recovery steam generators since the 1-2


underlying science is the same. A separate guideline report [4] is available that considers differences in design and operation of combined cycle unit as they effect the chemistry.

1.3 Cycle Chemistry Monitoring Approaches This report provides a review and discussion of on-line analyzer techniques currently available and in use in fossil power plants, updating many technical presentations from a 1998 report [5]. The focus is on analyzers used for fossil plant chemistry monitoring as described in the latest guidelines [1-4]. Monitoring techniques that may be considered for use in future plants are also presented as a precursor to possible future research being planned by EPRI to address possible improvements in monitoring and control of cycle chemistry in fossil steam-water cycles. 1.3.1 Core Monitoring Parameters An overall summary of the EPRI Core Monitoring Parameters and sampling points, adapted from References 1-3, is provided in Table 1-1 [1-3]. Except in situations where a specific boiler design, feedwater chemistry or boiler water chemistry is indicated in the notes, the core requirements indicated apply to all conventional plant designs and chemistries. Table 1-1 EPRI Cycle Chemistry Core Monitoring Parameters for Conventional Fossil Units [1-3] Parameter

Monitoring Points

Cation Conductivity

Condensate Pump Discharge, Condensate Polisher Outlet or Economizer Inlet, Reheat (or Main) Steam, Boiler Water a

Specific Conductivity

Treated Makeup, Boiler Water b

pH

Boiler Water c

Dissolved Oxygen

Condensate Pump Discharge, Economizer Inlet, Boiler Water d

Sodium

Condensate Pump Discharge, Economizer Inlet, Boiler Water e

Oxidation-Reduction Potential (ORP)

Deaerator Inlet f

Air in-leakage

Condenser Air Removal System Exhaust

Carryover

Calculated from Boiler Water and Saturated Steam Readings g

Notes: a – Drum boiler units only; measured on blowdown for boilers on PC or CT, and either blowdown or downcomer for boilers on AVT or OT b – blowdown of drum boilers on PC and CT c – Drum boiler units only; measured on either pH or downcomer of units on AVT; on either blowdown or downcomer of units on PC or CT; on downcomer of units on OT d – Only required in drum boiler units on OT; measured on downcomer e – Only required in drum boiler units on PC or OT; measured on downcomer f – Only required on units employing AVT(R) as feedwater chemistry g – Based on either on-line monitor readings (if available) or analysis results for grab samples

1-3


The Core Monitoring Parameters are considered the minimum level of surveillance needed for all conventional fossil generating units. They are purposely included as an integral part of the EPRI Cycle Chemistry benchmarking criteria used to assess individual cycle chemistry programs. In general, use of on-line analyzers for continuous analysis of chemistry is preferred. In the ensuing discussions of on-line analyzer technology, greatest emphasis has been placed on the EPRI Core Monitoring Parameters, which are presented in Sections 2-8 of the report. •

Section 2, Air in-leakage

•

Section 3, Carryover in Drum Boilers

•

Section 4, Conductivity; includes discussions of specific, cation and degassed cation conductivity

•

Section 5, Dissolved Oxygen

•

Section 6, Oxidation-Reduction Potential (ORP)

•

Section 7, pH

•

Section 8, Sodium

1.3.2 Diagnostic Monitoring Parameters In customizing their cycle chemistry programs, the organization may elect to monitor one or more of the diagnostic parameters with on-line analyzers. The more common monitoring techniques typically used for diagnostic purposes are presented in Sections 9-17 of the report. These are listed as follows. •

Section 9, Ammonia

•

Section 10, Chloride

•

Section 11, Hydrazine

•

Section 12, Hydrogen

•

Section 13, Ion Chromatography; can be used to monitor various ions in water and steam.

•

Section 14, Iron and Copper; includes a discussion of indirect means of monitoring suspended solids and particles in water and steam

•

Section 15, Phosphate

•

Section 16, Silica

•

Section 17, Total Organic Carbon

For many of the parameters and techniques more than one approach may be available. In some instances, particularly when chemistry data are needed for diagnostic purposes, collection of grab samples for laboratory analysis may be sufficient to satisfy the needs of the operator. These considerations are included in the Section 2-17 discussions. 1-4


1.3 Future Cycle Chemistry Guidelines and Monitoring Approaches Conventional sampling and analysis for chemistry surveillance and control in fossil generating units requires extraction of a representative sample of the fluid and subsequent conditioning to standard conditions, which are also assumed when establishing limit values for the various parameters of interest. Present day on-line analyzers are sensitive to inlet sample parameters such as flow, temperature, pressure, etc. and available designs can compensate for some but not all deviations from standard conditions. Thus effective sample conditioning is of great importance when using available monitors for cycle chemistry surveillance and control. The challenges of effective sample extraction and conditioning are considerable. Economic considerations often lead to compromises in design of sample conditioning system. For example, it is generally preferable with respect to provision of representative samples that the length of sample lines be minimized. However, such practice is frequently not followed as the cost of centralized sample conditioning and analysis facilities is less than that of remote sample stations. Direct measurement of fluid properties would eliminate the need of sample conditioning and associated error sources. This is now possible for some parameters such as pH and specific conductivity but only over a limited range of temperatures. Effective use of such instrumentation around the steam water cycle would require significant expansion of the working temperature range and development of new control limits that correspond to such temperatures. This need is greatest in high temperature areas such as boilers and turbines in which direct monitoring is not currently possible. The purpose of chemistry monitoring is to protect the components in contact with steam and water from damage and efficiency loss. The mechanisms and associated root causes responsible for most chemistry related damage and efficiency are now very well understood but the precise conditions at which corrosion and deposition activity begin are still not known. The past and present chemistry guidelines all serve to provide the operator with a warning as to when corrosion and/or deposition activity are thought to place the unit at risk. A focal point of improved chemistry guidelines is to provide operators with direct indications that the chemistry in boiler water supports corrosion activity since chemistry related boiler tube failures continue to represent a substantial source of availability loss across the industry. In consideration of this, EPRI has initiated laboratory research to better understand the environmental conditions in boilers which lead to corrosion damage and tube failure by mechanisms such as hydrogen damage. This work involves making direct measurements of surface potential and pH at temperature while adjusting the chemical composition of the water. Initial work will consider volatile treatments at various contaminant levels and subsequently consider the effect of solids based treatment chemicals used in drum type boilers. The apparatus used to make the laboratory measurements is not suitable for use in a working boiler. Underdeposit corrosion activity in boilers is also dependent on tube surface cleanliness. Research has been undertaken to develop a comprehensive model that can be used to understand the 1-5


various parameters which influence deposition. The ultimate aim of this work is to consider chemical activity under various deposit conditions so as to define the conditions under which corrosion activity becomes significant. The latest cycle chemistry guidelines include target values for boiler water in drum-type units that minimize the risk of corrosion damage. However, these values are empirical in that they reflect operating experience in fossil plants. Similarly, allowable levels of boiler tube deposition used to set criteria on the advisability of chemical cleaning are empirical, and represent levels of waterside solids that field experience suggests will not lead to tube damage and failure. It is envisioned that future chemistry guidelines will reflect the findings of research by EPRI in the corrosion and deposition areas. For waterside corrosion, the ultimate chemistry surveillance capability would be to measure corrosion activity directly in boilers and turbines, the major components at greatest risk of corrosion damage. However, there is currently no means to monitor corrosion rates under fossil plant operating conditions, in particular those which exist in these high temperature components. Sections 18 (on Electrochemical Corrosion Potential (ECP)) and 19 (on Corrosion) of this report review the science of corrosion monitoring and requirements for practical use of corrosion measurement techniques in fossil plants. From inspection of the Section 18 and 19 discussions it becomes clear that the potential for use of ECP and corrosion monitors in steam-water cycles exists but this remains a developing area. There are constraints on the applicability and reliability of available corrosion measurement devices that must be resolved in order to produce instruments suitable for use in the more demanding components of working fossil plants. Possible changes in cycle chemistry guidelines to prevent or control corrosion, based on the currently used indirect direct measurement methods, could initially involve modification of chemistry limits for impurities in the boiler water. These revised chemistry limits could perhaps be further refined to reflect the state of deposits on heat transfer surfaces as the effects of deposits on corrosion activity become more fully understood. Assuming that a robust corrosion monitor capable of operating in the boiler and/or turbine environment could be produced, extensive testing would be required as part of the development process. Location of its use would need further consideration due to concerns about compliance with design codes, installation costs, and possible consequences of probe damage during service. These concerns are significant and probably dictate installation of the monitor in a location external to the unit so it can be reliably isolated if needed. Such an approach may in fact hold many advantages over use of in situ probes.

1-6


1.4 References 1. Cycle Chemistry Guidelines for Fossil Plants: All-volatile Treatment, Revision 1. EPRI, Palo Alto, CA: 2002. 1004187. 2. Cycle Chemistry Guidelines for Fossil Plants: Phosphate Continuum and Caustic Treatment. EPRI, Palo Alto, CA: 2004. 1004188. 3. Cycle Chemistry Guidelines for Fossil Plants: Oxygenated Treatment. EPRI, Palo Alto, CA: 2005. 1004925. 4. Cycle Chemistry Guidelines for Combined Cycle/Heat Recovery Steam Generators (HRSGs). EPRI, Palo Alto, CA: 2006. 1010438. 5. Reference Manual for On-Line Monitoring of Water Chemistry and Corrosion: 1998 Update, EPRI, Palo Alto, CA: 1999. TR-112024.

1-7


2 AIR IN-LEAKAGE

2.1 Effects of Air In-leakage on Unit Performance and Cycle Chemistry Fossil unit operation requires that the condenser be operated under vacuum. The air removal system is designed to remove noncondensable gases from the condenser. In so doing vacuum conditions are established during fossil unit startup and maintained during unit service. Excessive air in-leakage rates can result in reduced condenser vacuum, thereby reducing turbine backpressure and the efficiency of the cycle [1]. Air ingress can also increase the concentration of dissolved oxygen in the steam-water cycle. All EPRI Cycle Chemistry Guidelines for conventional fossil and combined cycle plants now feature a Target Value of ≤10 ppb for dissolved oxygen at the condensate pump discharge [2-5]. Meeting the target provides maximum flexibility in treatment and generally ensures that contamination by carbon dioxide, which contributes to measured cation conductivity, is minimal. By meeting the target for dissolved oxygen at the condensate pump discharge and by using ammonia as the feedwater pH conditioning agent, there is generally no need to rely on techniques other than cation conductivity as the primary indicator of cycle contamination by inorganic ions such as chloride and sulfate that are of concern in boilers and turbines. Condensate dissolved oxygen and carbon dioxide levels are affected by the extent and location of air in-leakage in the condenser and in other parts of the cycle that operate at subatmospheric pressure. Air in-leakage in excess of that removed by the condenser air removal system will result in increased condensate dissolved oxygen and carbon dioxide levels, which may cause corrosion within the condenser. Contamination of the feedwater by oxygen and carbon dioxide may also lead to an increase in corrosion-product generation within the feedwater part of the cycle [1]. Effective monitoring and control of condenser air in-leakage is considered so important that it has been designated a Core Monitoring Parameter in all EPRI cycle chemistry guidelines [2-5]. Control of condenser air in-leakage to reduce condensate oxygen to ≤10 ppb is of greatest importance in those units that must employ AVT(R) so as to minimize corrosion of copper alloys and corrosion product transport in the feedwater part of the cycle. Protection of copper alloys in the feedwater environment requires that a reducing condition is established and maintained at all times. EPRI research has shown and field experience has confirmed that use of hydrazine or another reducing agent to establish a negative oxidation-reduction potential (versus silver/silver chloride; see Section 6) condition consistent with the requirements of AVT(R) cannot be reliably

2-1

EPRI Proprietary Licensed Material Air In-leakage

achieved when the condensate oxygen exceeds 10 ppb, regardless of the concentration or type of reducing agent applied. In units that employ oxidizing feedwater treatments, maintaining low levels of cycle air inleakage is not needed for protection of copper alloys in the feedwater. However, control of this parameter is desirable to limit dissolved oxygen and carbon dioxide contamination, which provides maximum flexibility in control of the feedwater treatment with minimum interference in measurement of two important chemistry monitoring parameters, namely conductivity and pH. In fossil unit cycles with condensate polishers, carbon dioxide will be exchanged by the anion media, potentially increasing the required frequency of regeneration (in deep bed system designs) or replacement of the precoat media (in precoat filter/demineralizer system designs). The Heat Exchange Institute recommends that the condenser design restrict air in-leakage to no more than 1.0 scfm per 100 MW of generating capacity. Constant vigilance must be exercised to prevent air in-leakage to the cycle from those equipment elements under vacuum. Particular problem areas include pump seals, valve bonnets, threaded joints, and especially the expansion joint between the turbine and condenser. Monitoring and limiting the amount of air in-leakage (condenser air removal system flow) is essential for proper control of dissolved oxygen and carbon dioxide in the cycle. Such monitoring will determine when an exhaustive effort must be made to find and fix the source of air leakage. Air in-leakage control is a continual process over the working life of any steam generating unit. The time and effort required to control air inleakage is justified by the beneficial effects on unit startup times, efficiency and compliance with cycle chemistry guidelines [1].

2.2 Monitoring Methods The earliest and still most common means of monitoring air removal at the condenser air removal system exhaust is by means of rotameters. Other monitoring techniques are also available. Some of these devices measure the volumetric flow rate while others measure volumetric flow plus some additional parameters [1]. 2.2.1 Rotameters Typically, a rotameter installed at the exhaust of the air removal system is relied upon to quantify air in-leakage. This was the standard approach at the time plants were designed and constructed whether the air removal devices employed were steam jet air ejectors or mechanical vacuum pumps. Manual rotameter readings were intended to be taken at fixed intervals, typically once a day. However, low manpower levels in many fossil plants often prevent this. Air in-leakage can increase and remain at potentially damaging levels for hours or possibly days before the excess in-leakage is discovered, since turbine backpressure may not have increased significantly or noticeably.

2-2


Under certain conditions, rotameters provide adequate flow rate measurements of air in-leakage. However, many factors can affect the accuracy of the readings so obtained, such as debris, exhauster pulsations, water carryover, etc. Rotameters also have limited range, generally 10:1. Therefore, most problematic air in-leakages cause the rotameter to be pegged against the high level stop. Rotameters receive essentially very little maintenance and are checked very infrequently at many plants. In these situations, the operator has no idea of the extent of the air in-leakage, given the fact that the turbine backpressure rise may appear to be rather small and insignificant [1]. 2.2.2 Early Advanced Air Removal Monitoring Systems Rotameters have long been recognized as subject to a number of limitations and deficiencies. Their accuracy is questionable and readings are often unstable and subject to either high or low bias [1-6]. The data are only available intermittently and an operator is required to adjust valve settings and collect readings; there is no means to collect readings in real time or automatically transmit them to the control room operator. In an attempt to mitigate these inherent problems with rotameters, some improved instrument designs, which featured continuous readouts of removal rates, were developed for use in power plants [1,6,7]. Some of these designs had some serious drawbacks, such as not being able to distinguish between water vapor and air; and, operating under the principle that the condenser exhaust gases were always at saturation, when in fact, they were not. Some of these devices required a high level of maintenance, which often exceeded the resources of the plant [1]. 2.2.3 Multisensor Probe Design Subsequent developments have produced more advanced instrumentation, which employs a number of sensors that collectively overcome these earlier limitations and is available commercially [1,7,8]. One such instrument features a multisensor probe (MSP) which is installed in the exhaust line from the condenser. This device is depicted in Figure 2-1 [1,8]. This instrument has the ability to directly measure the mass properties of gas in the condenser vacuum line, as well as compute other parameters based on thermodynamics. The instrument combines a gas analyzer with a total mass flow meter. The instrument utilizes a velocitymeasuring device, and other facilities for measuring temperature, pressure, and relative saturation. By including the measurement of pipe diameter, the instrument can separate the effects of non-condensable gas from water vapor and compute individual and combined properties of the gas mixture. As a result, the following parameters of the gas flowing in the exhauster suction line are either directly measured or are determinable by calculation using direct measurement results and the piping diameter: •

Air in-leakage rate

•

Volumetric flow rate

2-3


•

Total mass flow

•

Water vapor mass flow rate

•

Water vapor/air mass ratio (relative saturation)

•

Water vapor density

•

Air density

•

Partial pressure of water

•

Partial pressure of air

•

Total pressure

•

Temperature

Air in-leakage, as measured by the MSP is the sum of all leak sources entering the subatmospheric pressure system and subsequently flowing through the exhauster line upstream of the sensor location. The MSP, however, does not measure air entering the low-pressure system beyond (downstream of) the probe, such as at valves, input connections, and shaft seals. The MSP does measure the exhauster capacity, which is useful to gauge the capacity of the exhausters to remove the gas mixture from the condenser [4]. Evaluation and analysis of condenser system performance and measurement of in-leakage is best accomplished with the use of multiple probes or alternatively by collection of data from desired locations with single portable probe. An example system utilizes five MSP probes as shown schematically in Figure 2-2. As indicated in the example depicted in the figure, each probe is identified by a number that is used to categorize performance information and to help pinpoint possible leak locations, as illustrated in Table 2-1 and Table 2-2 [4].

2-4


Figure 2-1 Multisensor Probe

2-5


3

2

4

1

5

Exhauster 2

Exhauster 1

Condenser Figure 2-2 Multisensor Probe Instrument Schematic

Table 2-1 Example Diagnostic Capabilities of a Five-Probe Air In-leakage Measurement System Leak Location

Probe Indications of Air In-Leakage (SCFM) and Plant Dissolved Oxygen (DO)

Below water line, left side of condenser

Much higher than normal DO, 1>2 1 + 2 = 3 = 4 + 5, 4=5

Above water line, right side of condenser

Slightly higher or normal DO, 2 > 1, 1 + 2 = 3 = 4 + 5, 4=5

Small leak or faulty exhauster down stream of probe 4

Slightly higher or normal DO, 1 = 2, 1 + 2 = 3 = 4 + 5, 4<5

Large leak of failed exhauster down stream of probe 4

Slightly higher or normal DO, 1 = 2, 1 + 2 = 3 < (4 + 5), 5 > 4 (back flow at 4)

Center joint seal, LP bearing seal, or other central location

Slightly higher or normal DO, 1 = 2, 1+2=3=4+5

2-6

EPRI Proprietary Licensed Material Air In-leakage Table 2-2 MSP Probe Indications for Various Probe Positions Leak Location

Probe Indications of Air In-Leakage (SCFM)

Normal tight system

1 = 2.5 SCFM 2 = 2.5 SCFM 3 = 5 SCFM 4 = 0 SCFM (exhauster 2 not in service) 5 = 5 SCFM

Abnormal, need to locate and fix leaks (central joint seal)

1 = 15 SCFM 2 = 18 SCFM 3 = 33 SCFM 4 = 16.5 SCFM 5 = 16.5 SCFM

Abnormal, need to locate leaks and fix vacuum pump shaft seal in exhauster 2

1 = 20 SCFM 2 = 20 SCFM 3 = 40 SCFM 4 = 20 SCFM (Differences due to reversed flow 5 = 60 SCFM through shaft seal being sensed)

2.3 References 1. Condenser Tube Failures: Theory and Practice: Volume 1: Fundamentals. EPRI, Palo Alto, CA: 2005. 1010188. 2. Cycle Chemistry Guidelines for Fossil Plants: Phosphate Continuum and Caustic Treatment. EPRI, Palo Alto, CA: 2004. 1004188. 3. Cycle Chemistry Guidelines for Fossil Plants: Phosphate Continuum and Caustic Treatment. EPRI, Palo Alto, CA: 2004. 1004188 Cycle Chemistry Guidelines for Fossil Plants: Oxygenated Treatment. EPRI, Palo Alto, CA: 2005. 1004925. 4. Cycle Chemistry Guidelines for Fossil Plants: Oxygenated Treatment. EPRI, Palo Alto, CA: 2005. 1004925. 5. Cycle Chemistry Guidelines for Combined Cycle/Heat Recovery Steam Generators (HRSGs). EPRI, Palo Alto, CA: 2006. 1010438. 6. K. Weaver, M. Twerdochlib and J. Bellows, “On-line Monitoring of Steam Turbine Air inleakage”, presented at the Joint ASME-IEEE Power Generation Conference, Philadelphia Pennsylvania, September 25-29, 1998: ASME, NY, NY. 88-JPGC-PWR-3. 7. J. A. Ferrison and P. Baker, “Air Inleakage Measurement Experiences with the Chell Air Ingress Monitor”, presented at the EPRI Condenser Technology Conference, Saint Petersburg Florida, September 30, 1983.

2-7


8. J. W. Harpster, “On Understanding Mechanisms that Control Dissolved Oxygen in Condenser Condensate.”, Proceeding of the 21st Annual Electric Utility Workshop, University of Illinois, Champaign, IL, May 8-10, 2001.

2-8


3 CARRYOVER IN DRUM BOILERS

3.1 Introduction to Carryover In cycles with once through boilers, the steam purity is essentially equal to that of the boiler feedwater providing that impurity levels are maintained below those at which deposition (and possible subsequent dissolution of the deposited solids) could occur. In cycles with drum type boilers the composition of the drum steam is dependent on the phenomenon known as carryover. Carryover is defined as the ratio of concentration of a chemical species in the saturated steam exiting the boiler drum to the concentration of the same species in the boiler water (usually measured in the blowdown). Total carryover (T) is defined in Equation 3-1 as follows. T=M+V

Equation 3-1

In this equation, M is mechanical carryover (due to steam moisture) and V is the vaporous carryover (due to volatile partitioning into the vapor phase). Mechanical carryover represents any boiler water fine droplets or mist that exits the boiler drum with the saturated steam. It is dependent on factors such as boiler pressure, drum water level, and the design and integrity of the internal separator devices employed to prevent boiler water from entering the steam. Vaporous carryover represents those impurities which partition from the liquid to the vapor phase during the boiling process. The partitioning tendency of individual chemical species is dependent on operating temperature and pressure. These partitioned impurities cannot be removed by the steam separator devices which are provided to control mechanical carryover. The extent to which a species partitions to steam often exceeds its solubility in steam, which can lead to deposition in superheaters, reheaters and steam turbines. For example, silica has long been recognized as a constituent subject to vaporous carryover. If not controlled, silica in steam will form deposits on the low pressure turbine. EPRI research investigating the properties of copper around the cycle determined that copper is highly volatile in boiling water, even at low pressures. However, the low solubility of copper results in significant deposition of copper in the primary superheater.

3-1

EPRI Proprietary Licensed Material Carryover in Drum Boilers

3.2 Consideration of Carryover in EPRI Cycle Chemistry Guidelines While previous EPRI chemistry guidelines suggested that unit specific steam purity guidelines be established based on assessment of predicted mechanical and volatile carryover relationships, the latest guidelines [1-3] include target values that reflect results of over a decade of research to define vaporous carryover based on a series of experimental assessments of the various constituents of interest and concern [4-10]. This work provides very accurate predictions of vaporous carryover over the operating pressure ranges of interest and considers the interaction of other species in the boiler water. For potentially corrosive contaminants such as sodium, chloride and sulfate, the vaporous carryover is pressure- and solution pH-dependent; in general, volatile carryover of these impurities is very minor except at high pressure providing the boiler water pH is maintained in expected ranges. By contrast, earlier volatility studies were based on single component solutions, at impurity concentrations in the liquid phase which greatly exceeded those allowable in boiler water [11]. Thus the latest control curves for boiler water are much more reliable with respect to vaporous carryover. However, they are based on assumed levels of mechanical carryover, which has a dominant influence on total carryover for most impurities. The primary exceptions are silica and copper. Figure 3-1 indicates the relationship used to define mechanical carryover in the guidelines [2]. Note that the values in this curve include a safety margin. Plant operating experience has shown that the performance guarantees of the manufacturer are generally readily met during performance testing normally conducted during commissioning of new units. However, performance often declines as a consequence of deterioration of the steam separator devices over time. Also, carryover is sometimes observed to increase during unit startup as a consequence of variations in steam drum water level. Such steam purity excursions may go unnoticed in cases where the unit sampling and analysis system is not in service during startup. Regrettably, a substantial number of steam purity problems in fossil units with drum boilers are ultimately determined to have been caused by carryover. These events are noted across the worldwide fleet of units. Disturbingly, it is very clear that many of these events could have been identified sooner had the plant staff assessed the carryover activity but this is not checked in many boilers after completion of the initial performance test conducted during commissioning of the unit. To address this gap in chemistry monitoring, the most recent EPRI cycle chemistry guidelines have designated carryover from drum boilers as a Core Monitoring Parameter. Carryover is the only core parameter that is a relationship of measured parameters. It is determined by calculation, using data measured in boiler water and saturated steam. Also, it need not be checked continuously or even with on-line instrumentation. The most recent cycle chemistry guidelines [1-3] indicate assessment of carryover at six month intervals so as to provide plant operators with an early indication of an ongoing or developing problem with steam purity. This approach also allows for inspection of plant operating and chemistry records and for planning of boiler drum internal inspection activities when assessment carryover assessment findings suggest such action is warranted.

3-2


Total carryover is usually measured by analysis of sodium in the boiler blowdown and saturated steam. Sodium is generally the dominant cation of possible concern in the boiler water, particularly when the boiler water is treated with trisodium phosphate or caustic. Sodium can be monitored reliably and at low concentrations by either on-line analyzers or laboratory analysis of grab samples. Section 8 of this report provides information on sodium analysis options. 4.8

6.2

7.6

9.0

Drum Pressure (MPa) 10.3 11.7 13.1 14.5

15.8

17.2

18.6 19.6

0.3

Mechanical carryover (%)

0.2

0.1 0.09 0.08 (Note: This curve includes a safety factor of 2)

0.07 600 700

900

1100 1300 1500 1700 1900 2100 2300 2500 2700 2850 Drum Pressure (psig)

Figure 3-1 Representative Drum Boiler Mechanical Carryover [2]

Carryover is also a Core Monitoring Parameter in combined cycle plants with drum-type generating circuits [12]. In multi-pressure HRSG designs, drum carryover should be monitored in each circuit. 3-3


3.3 Determination of Total Carryover from Drum Boilers Total carryover, the ratio of sodium concentrations in the saturated steam and boiler water, may be determined by calculation as shown in Equation 3-2.

T=

Na in saturated steam Na in blowdown

Equation 3-2

During carryover assessment, it is preferable to evaluate a broad range of operating conditions, particularly in cycling and peaking units.

3.4 Sampling and Data Collection Considerations Calculation of total carryover requires determination of a low volatility parameter (typically sodium though other parameters, such as chloride or sulfate, could be considered as an alternative or in addition to sodium) in the saturated steam and boiler water of the unit. Ideally, the readings should be collected as close to simultaneously as practical. When one or more on-line analyzers will be used to collect the data the outputs of the instruments should be checked and calibration performed if needed. If a single online analyzer will be used for both sample points, sufficient time must be provided to allow for switching of samples and stabilization of readings. In cases where on-line analyzers are not used, grab samples should be collected as close to simultaneously as practical. The boiler blowdown is the preferred sample point for carryover assessment as it is more representative of the liquid phase which may enter the steam with the saturated vapor. It should have a somewhat higher concentration of the constituent of interest than a boiler downcomer sample, which will tend to be slightly diluted by the feedwater entering the boiler. For carryover assessment purposes, saturated (drum) steam is the only acceptable sample point for steam. Superheated and reheat steam samples are not representative for carryover assessment since impurities in the saturated steam may form deposits prior to the superheater and reheater. Also, impurities from previously deposited solids may later dissolve and be present in the superheated and reheat steam. In addition, the composition of superheated and reheat steam is influenced by use of feedwater for attemperation. Monitoring of the saturated steam provides verification of compliance with the boiler manufacturer’s performance guarantee for steam purity, which often applies only to the saturated steam. Depending on boiler design the saturated sample source may be from single or multiple steam leads (off-take lines). The boiler operator should ensure that a representative sample is taken. In older boilers with multiple steam lead arrangements, some of the leads may be isolated and no longer operable. In such cases, resultant steam purity data does not provide complete assurance that contamination is not entering the steam path.

3-4


With proper sampling and analysis, it is thus possible to effectively monitor the total carryover of impurities into the high pressure turbine. As carryover is influenced by boiler water pH, the operator should note this at the time the sodium or other constituent readings are checked. Also, if a downcomer sample of boiler water must be used, the condition of the feedwater should be noted. These additional details will be helpful in interpretation of results including comparison of findings from carryover assessments made over time. Such an approach greatly improves the chance for identifying and correcting problems before target values for impurities allowed in the reheat steam are exceeded. The saturated steam chemistry data and trends over time can also be related to the performance of the steam drum moisture separators by measuring the carryover of impurities from the boiler into the steam. Excessive carryover may indicate poor moisture separator performance. When the design includes multiple steam leads and these can be sampled individually, it should be possible to identify areas of greatest concern. The latest cycle chemistry guidelines are based on findings of a series of EPRI sponsored investigations of volatility [4-10]. In the case of constituents such as silica and copper, volatile carryover is a significant contributor to total carryover over a broad range of operating pressures [8,9]. However assessment results for most impurities including sodium, chloride and sulfate has confirmed that mechanical carryover of has the greatest impact upon total carryover and steam purity, particularly at boiler drum operating pressures below 2500 psi (17.2 MPa). As discussed here and in the individual cycle chemistry guidelines, total carryover from drum boilers is considered now a Core Monitoring Parameter which should be routinely checked at intervals of about every six months so as to minimize the risk of steam contamination and possible damage to the turbine.

3.5 References 1. Cycle Chemistry Guidelines for Fossil Plants: All-volatile Treatment, Revision 1. EPRI, Palo Alto, CA: 2002. 1004187. 2. Cycle Chemistry Guidelines for Fossil Plants: Phosphate Continuum and Caustic Treatment. EPRI, Palo Alto, CA: 2004. 1004188. 3. Cycle Chemistry Guidelines for Fossil Plants: Oxygenated Treatment. EPRI, Palo Alto, CA: 2005. 1004925. 4. Behavior of Ammonium Salts in Steam Cycles. EPRI, Palo Alto, CA: December 1993. TR102377. 5. Volatility of Aqueous Sodium Hydroxide, Bisulfate and Sulfate. EPRI, Palo Alto, CA: February 1999. TR-105801.

3-5


6. Vapor-Liquid Partitioning of Sulfuric Acid and Ammonium Sulfate. EPRI, Palo Alto, CA: February 1999. TR-112359. 7. Volatility of Aqueous Acetic Acid, Formic Acid and Sodium Acetate. EPRI, Palo Alto, CA: July 2000. TR-113089. 8. Behavior of Aqueous Electrolytes in Steam Cycles: The Solubility and Volatility of Cupric Oxide. EPRI, Palo Alto, CA: November 2000. 1000455. 9. The Volatility of Impurities in Water/Steam Cycles. EPRI, Palo Alto, CA: 2001. 1001042. 10. Vapor-Liquid Partitioning of Phosphoric Acid and Sodium Phosphates. EPRI, Palo Alto, CA: 2003. 1007291. 11. Assessment of the Ray Diagram. EPRI, Palo Alto, CA: 1996. TR-106017. 12. Cycle Chemistry Guidelines for Combined Cycle/Heat Recovery Steam Generators (HRSGs). EPRI, Palo Alto, CA: 2006. 1010438.

3-6


4 CONDUCTIVITY

4.1 Purpose and Use Conductivity, more specifically cation conductivity is an EPRI Guideline Core Monitoring parameter [1-4]. As such, cation conductivity is continually monitored on-line to indicate the ionic concentration in the solution being monitored and to warn of contaminant in-leakage. Three types of conductivity measurements are discussed:

•

Specific conductivity

•

Cation (acid) conductivity

•

Degassed cation conductivity

These three types of conductivity are also monitored for one or more of the following reasons:

•

To facilitate the correlation of a water chemistry parameter (e.g., pH, conductivity, ammonia correlation).

•

To check the accuracy of water chemistry control (such as ammonia or pH).

•

To warn of condensate polisher malfunction.

•

To provide feedback signal for automated process control.

•

To facilitate demineralizer system operation and/or regeneration.

•

To monitor for the intrusion of volatile contaminates (e.g., CO2 or volatile organics).

•

To monitor the laboratory pure water system.

•

To determine the concentration of ions in solution.

The data generated by the on-line monitoring equipment are used by plant chemistry and operations department personnel. The goal for plant personnel is to identify the intrusion of impurities into the steam-water cycle and take appropriate corrective actions to minimize the duration and severity of the upset conditions.

4.2 Description On-line conductivity monitoring equipment measures the ability of water to conduct electricity (i.e., the electrical conductivity of water). Water conducts electric current by the movement of ions dissolved in the water. Consequently, the conductivity value can be used to assess the 4-1

EPRI Proprietary Licensed Material Conductivity

purity of water or, in cases where the ions are corrosive, conductivity can be used to assess the corrosivity of water. Conductivity of power plant waters in the boiler or steam generator cycle is usually measured in units of micro Siemens per centimeter (µS/cm), formerly called micro mho per centimeter, which, by definition, is the inverse of resistivity in units of megohm-cm. The conductivity, C, of an aqueous solution depends on the concentration (Q) of the dissolved substance, the degree of dissociation (α) into positively and negatively charged ions, and the charge on those ions which is dependent on their valence (n):

C = α QnF ( I + + I − )

Equation 4-1

Where F is Faraday’s constant (96,463 coulombs), and I+ and I- represents the mobility of the positive and negative ions, respectively. The mobility is dependent on the type of dissolved substance. If more than one substance is present, Equation 4-1 is used to determine the contribution of each substance, and then the total conductivity is obtained by adding the individual contributions together. Three types of conductivity are routinely measured in power plants: specific conductivity, cation conductivity, and degassed cation conductivity. Common to all three types is the measurement of the resistance of a known volume of water in a water sample cell. There are two common approaches to making these measurements: either with an on-line flow-through cell or dip cell fitted with electrodes. For either application, electrode sensors are immersed in the water and the resistance between the electrodes is measured directly. In this section, only on-line flow-through cell measurements are discussed further. Electrode materials must be corrosion-resistant in the expected environment. Typical materials include nickel, stainless steel and titanium. The body of the flow-through cell may be fabricated from such materials as stainless steel (e.g. type 316) or polyethersulfone (PES). Specific conductivity is one of the most frequently monitored parameters in power plant waters. In its most common form, the technique involves measuring the electrical conductivity (or conductance) of a volume of water which occupies the space between two electrodes of known dimensions. For instance, if the electrodes are flat, parallel plates a distance L apart, and each has a surface area, A, facing the opposite plate, the conductivity, C, of a water sample between the two electrodes is given by:

C = 1/R = (κ × A)/L

Equation 4-2

or by:

κ = L/(A × R) = (L × C )/A

4-2

Equation 4-3


where R is the resistance of the sample and κ is the specific conductivity. By definition, κ is the conductivity of a column of the sample water having a length of 1 cm and a cross-section of 2 1 cm measured at 25°C (77°F). In on-line conductivity monitoring equipment, the electrodes are incorporated into a flowthrough conductivity cell allowing the water sample to be constantly (or frequently) refreshed, thereby allowing the measured conductivity value to be continually updated. Cation conductivity is another common conductivity measurement which can provide useful additional information when compared with specific conductivity. Here, before measuring the conductivity, the water sample first passes through a cation resin bed which exchanges any cations (positive ions) in the water for hydrogen ions (Figure 4-1). For instance, if the water contains NaCl, the dissolved Na+ ions are held in the bed and H+ ions are released in their place. Thus, NaCl (equivalent conductance 126 S-cm2/equivalent) is converted into HCl (equivalent conductance 426 S-cm2/equivalent) yielding a conductivity increase of about a factor of 3. In a similar manner, other salts are converted to their corresponding acids with an accompanying approximate 3-6 fold increase in conductivity. The conductivity of the water leaving the cation resin bed—the cation conductivity—is measured in a second flow-through cell, similar to the cell used for specific conductivity measurements. Any acids in the sample would, of course, pass through the cation bed unchanged because the positive ions involved are already H+ ions. Conversely, if the water contains chemicals that are alkaline in nature, such as NH4OH or NaOH, the NH4+ or Na+ ions would be replaced by H+ ions and result in the formation HOH (that is H2O, or water) with a corresponding large decrease in conductivity. If the cation conductivity is lower than the specific conductivity, it would reflect the removal of alkaline species by the cation bed, such as those that might be used for water treatment or demineralizer regeneration (e.g., NH4OH, NaOH respectively). If the cation conductivity is higher, it would indicate the presence of salts, like NaCl, in the original water sample, possibly caused by cooling water in-leakage in the condenser. Interpretation of cation conductivity values becomes more difficult when opposing effects of salts and alkaline species are seen in the same sample. Carbon dioxide (CO2) may also be in the sample and will be converted to carbonic acid as the sample stream passes through the cation column. Elevated cation conductivity from carbon dioxide may not be a corrosion concern and its contribution to cation conductivity may be removed using a degassed cation conductivity monitor.

4-3


Figure 4-1 A Typical Cation Conductivity Flow Diagram [11] Source: Adapted from Reference 11, Courtesy American Society for Testing and Materials

Various cation conductivity measurements can be related to the concentrations of specific chemical species. Figures 4-2, 4-3, and 4-4 show conductivity as a function of a single species; HCl, H2SO4 and H2CO3 respectively [5]. Degassed cation conductivity is the third type of conductivity measurement. During startup the main concern is the presence of salts in the steam/water cycle. The detection of these salts is enhanced as they are converted to strong acids in the cation column as previously discussed but the presence of carbon dioxide also increases cation conductivity thereby reducing the sensitivity of the conductivity measurement to dissolved salts. To eliminate the effects of carbon dioxide and to minimize the hold times, the water may be degassed prior to measuring the cation conductivity. The resulting degassed cation conductivity of the feedwater or steam more accurately reflects the effects of the dissolved salts of interest during startup. This is achieved by delivering the water sample leaving the cation bed to an electric reboiler where the sample is heated to near boiling temperatures to drive off the dissolved gases. The conductivity of the water leaving the reboiler is then measured in a third flow-through cell. With the contribution from dissolved gases removed, this conductivity measurement can be employed as an extremely sensitive detector of small condenser leaks or malfunctions in the condensate polisher or demineralizer. Figure 4-5 shows a schematic diagram of a typical degassed cation conductivity monitor.

4-4


Figure 4-2 Chloride (Cl) Concentration vs. Specific Conductivity [5] Source: Adapted from Reference 5, Courtesy American Society for Testing and Materials

4-5


Figure 4-3 Sulfate (SO4) Concentration vs. Specific Conductivity [5] Source: Adapted from Reference 5, Courtesy American Society for Testing and Materials

4-6


Figure 4-4 Carbon Dioxide (CO2) vs. Specific Conductivity [5] Source: Adapted from Reference 5, Courtesy American Society for Testing and Materials

4-7


Figure 4-5 Typical Degassed Cation Conductivity Schematic Diagram [5] Source: Adapted from Reference 5, Courtesy American Society for Testing and Materials

For accurate results, the temperature of the water sample entering this cell must be closely controlled, and the instrument must be capable of correcting for temperatures near boiling. Temperature compensation is very important in degassed cation conductivity measurements, and is complicated by the fact that measurements must be performed near the boiling point of water. With adequate insulation of the cell and temperature measuring thermometer, and with careful adjustment of the reboiler temperature controller, the reboiler can provide a supply of flowing water at a constant temperature of about 95°C. For these acidic solutions at temperatures near boiling temperature correction factors are complex. End users should evaluate the proposed correction factors to assure the instrument suppliers have accommodated for these unique conditions into their temperature correction factors (See discussion which follows on Cation Conductivity Temperature Compensation).

4-8


Other methods of stripping the dissolved gases from the sample (e.g. sparging with high purity nitrogen) are also available, but they are more labor-intensive and not as widely used.

4.3 Technical Considerations Water samples taken from the steam/water cycle typically have low ionic strength (i.e., 0.1 to 100 µS/cm specific conductivity), leading to several analytical challenges. These challenges can be overcome provided appropriate conductivity cells with appropriate cell constants are selected; appropriate temperature compensation is applied; cation conductivity resin and columns are properly designed and used; carbon dioxide and organic acid interference is understood and, appropriate cell constant checks are performed. Without appropriate consideration to these factors, conductivity measurements may be erroneous. 4.3.1 Cell Constants The dimensions of the cell determine the value of the cell constant. In equation 2, for instance, the cell constant is the geometric factor, L/A. In practice, electrodes are seldom plate-shaped, rather they are rings, cylinders or pins; but, whatever design is used, the geometry of the system determines the cell constant. The units of the cell constant are reciprocal length, e.g. cm-1. The dimensions of the cell, and hence the value of the cell constant, must be selected to provide enough sensitivity over the conductivity range of interest. Low conductivity measurements (high resistance) require a low cell constant—that is, large electrodes that are relatively close together. Conductivity cells for high conductivity (low resistance), on the other hand, have a high cell constant where electrodes are smaller and spaced further apart. In most commercial instruments, the value of the cell constant must be manually selected on the associated instrumentation before the measurements are made. Typical conductivity range limits are listed in Table 4-1 as a function of cell constant. Table 4-1 Typical Conductivity Range Limits as a Function of Cell Constant Cell constant (cm-1)

Optimum Conductivity Range (µS/cm)

Optimum Resistively Range (ohm-cm)

0.01

0.055 to 20

50,000 to 18,000,000

0.10

0.5 to 200

5000 to 2,000,000

1.00

10 to 200

500 to 100,000

10.00

100 to 20,000

50 to 10,000

100.0

1000 to 200,000

5 to 1000

4-9


4.3.2 Cell Constant Determination or Verification Cell constants stated by the manufacturer with each cell are typically determined experimentally. Cells are manufactured to specific geometries to represent a specific cell constant. Subsequently, the cell constant may be influenced by minor dimensional variations during manufacture, shockinduced dimensional changes during shipping, mounting orientation, or flow velocity. Therefore it is recommended that the cell constant be determined with a stable solution of a known specific conductivity. The cell constant is calculated based on the observed specific conductivity compared to the expected specific conductivity [7]. For low conductivity measurements (cell constant 0.01) a stable solution with a known specific conductivity in a range of 0.55-20 µS/cm would be required to calculate the cell constant. Solution with specific conductivity <100 µS/cm specific conductivity are considered unstable due to rapid contamination from CO2. Therefore it is recommended that a cell constant for a reference conductivity cell with a higher cell constant (0.1 or 1) be determined first by using a higher specific conductivity solution (i.e., > 100 µS/cm specific conductivity). That reference conductivity cell is then used to determine or verify the lower cell constant cell (0.01) by comparison of a low conductivity sample stream as described ASTM D-5391 [8]. The lower cell constant cell being determined or verified should be connected to the same low conductivity (i.e., in the range of 0.55-20 µS/cm specific conductivity) sample stream as, and in close proximity to, the reference cell to avoid sample contamination and to minimize temperature differences between the cells. 4.3.3 Cell Construction and Installation Considerations Cells used for low ionic strength water applications (<10 µS/cm) should not be platinized cells due to the tendency of the porous platinum surfaces to hold trace ions. Trace or flash platinumblack is permissible. Platinization was traditionally used to create a roughened surface on smooth platinum electrodes to make intimate contact with the sample and give the cell wide rangeability, especially at the high end, with fairly unsophisticated measuring circuits. Traditional platinizing left a coating of platinum black like the heat shield on top of a candle or kerosene lamp. That coating was like an ion sponge that leached ions for days in rinsing down, -1 especially in the closely spaced series of electrode plates of a 0.01 cm cell constant. Manufacturers recognized this problem and eliminated the platinization on sensors for pure water and limited their range. Beckman/Rosemount arrived at a compromise of applying just a trace (flash) of platinum black that gave the metal a slightly dingy appearance but still fairly shiny and could rinse down fairly quickly. Today all but the academic purists use the concentric titanium sensors (invented by Thornton and Balsbaugh for the semiconductor industry in the late 1960s) with mechanically roughened surfaces to achieve the same electrode performance. This design coupled with improved measuring circuitry allows use of more widely spaced electrodes in a 0.1 cm-1 cell constant and even wider rangeabililty. The wider spacing and absence of platinum allows very rapid rinse down plus much reduced chance of collecting DI resin beads or corrosion products. Cells for use 4-10


downstream of ion exchange systems (cation columns, condensate polishers or demineralizers) should have a physical size to minimize the entrapment of fine resin particles between sensing plates (>1.5 mm, 0.06 in., between plates) or be located downstream of installed resin traps. Cells should never be installed downstream of pH monitors due to sample contamination from KCl ions released through the pH reference electrode. Cells should be designed to minimize the capacitance of the cell and the lead wires. Lead wires should be limited to 7 m (20 ft.). Excessive capacitance can lead to a positive bias conductivity. 4.3.4 Cell Polarization R (or C) in Equation 4-2 is determined by measuring the current resulting from applying a known potential across the electrodes of the conductivity cell. An important factor in determining the accuracy of conductivity measurements is the extent of polarization caused by the passage of this current. If a direct current (DC) potential were to be applied across the electrodes, positively charged ions in the sample water would move to the cathode and negatively charged ions would move to the anode. This movement and accumulation of ions at the electrodes would result in polarization of the electrodes, which would oppose (be of opposite polarity to) the applied DC potential, and would result in a reduction in the current. In commercial 2-electrode systems, the effects of polarization are greatly reduced by applying an alternating current (AC) potential instead of a DC potential. The reversing polarity of the AC current counters the tendency for ions to accumulate at one electrode or the other. In principle, the higher the frequency the lower is the polarization effect, but capacitance effects in the connecting cables provide an upper limit to the frequency that can be used in a practical system. While the use of an AC potential to energize the conductivity cell is standard practice, it may not completely eliminate polarization. However, the effects of polarization in a 2-electrode conductivity cell can be minimized by ensuring that the optimum cell constant is selected, as previously discussed, so that the current density flowing between the electrodes is as low as possible without sacrificing adequate measurement sensitivity. 4.3.5 Temperature Effects Specific conductivity, cation conductivity, or degassed cation conductivity measurements are affected by the temperature of solution being measurement. By convention, conductivity is routinely reported corrected to 25°C (77°F). Even minor temperature differences from 25°C (77°F) can have a significant impact on observed conductivity measurements. An error of only 0.1°C in the temperature measurement can result in an error of 0.0006 µS/cm in water that has a conductivity of 0.055 µS/cm; an error of over 1 percent. Because of this error, samples outside 25 ± 0.2°C must have an appropriate temperature compensation factor applied. In many modern microprocessor-based conductivity instruments, temperature is measured directly in the conductivity cell and the instrumentation automatically corrects the conductivity 4-11


value for temperature. To assure accurate temperature measurement, microprocessor-based conductivity instruments may also feature a simple electronic temperature calibration feature. However, a more precise temperature measurement calibration, using separate certified thermometers, may provide improved performance. The conductivity at 25°C (77°F) is normally the value calculated and displayed by the instrument. However, it is important to note that the temperature coefficient, “a”, in Equation 43 is dependent on the nature and concentration of the ions present (solute) in the solution being monitored, temperature coefficients can range from about 0.5% per °C to 3.0% per °C in strong electrolyte, highly conductive solutions [9]:

κ

T

=

κ

25

[1 + a(T - 25)]

Where: κ T and respectively.

κ

25

Equation 4-4

are the specific conductivities (µS/cm) at temperatures T°C and 25°C,

4.3.5.1 Solute Effects For power plant steam/water cycle specific conductivity measurements > 10 µS/cm, the coefficient “a” in Equation 4.3 is typically set to 2% per °C different from 25°C. The 2% temperature coefficient is appropriate for the most common ions present in a power plant steam/water cycle. In more sophisticated, microprocessor-controlled instruments, the value of “a” in Equation 4-3 can be adjusted to suit the particular solution being monitored. 4.3.5.2 Solvent Effects For conductivity measurements > 10 µS/cm, hydrogen (H+) and hydroxyl ions (OH-), formed by dissociation of the solvent (water), contribute very little to the overall conductivity of the solution. However, for conductivity measurements < 10 µS/cm, the relative importance of these ions increases until in ultra high purity waters approaching 0.055 µS/cm, they become the dominant factor as the only contributors to conductivity. At 25°C (77°F), pure water has a conductivity of 0.055 µS/cm. In these pure waters, the linear relationship given in Equation 4.3 is no longer valid, and more accurate empirical equations must be used. The three most widely used equations are known as the General Electric (GE), Marsh and Stokes, and Truman Light equations [10]. Some microprocessor-controlled instruments provide all three of these options or a proprietary option to allow the user to select the preferred temperature compensation equation.

4-12


4.3.6 Cation and Degassed Cation Conductivity Temperature Compensation Sample temperature compensation for cation conductivity requires a unique correction factor for the acidic composition of the cation effluent. Cation conductivity samples have had their cations + + replaced with H ions and therefore are acidic solutions. The H ions have a direct impact on the ionization of water. The additional H+ ions tend to suppress the ionization of water. This interaction and its variation with temperature require more sophisticated temperature compensations to provide accurate cation conductivity measurement, particularly at temperatures outside 25 ± 0.2 °C [9]. End users should evaluate cation conductivity monitors to assure appropriate temperature compensation is applied. 4.3.7 Cation Conductivity Column Connections, Size, Flow and Flow Rate Considerations Cation columns come in a variety of shapes and sizes, making it important to match the appropriate sample flow with the cross-sectional area of the column. Sample flow through the column may be either upward or downward. In either case, air needs to be vented from the column to avoid incorrect readings. The columns must be fitted with an end screen to distribute flow across the cross section of the column and to prevent resin fines from leaking out of the column. Upward flow provides for automatic venting of air during initiation of sample flow which may be helpful for establishing faster accurate readings during startup of cycling plants. However, the cation resin in the column must be packed full to prevent fluidizing or channeling within the cation column. For downward flow cation columns fluidizing is not a concern since the resin is compressed in the cation column from the sample flow. However, channeling or incomplete ion exchange may occur when flow velocity through the column is too fast or too slow. Fast downward flow causes channeling where a portion of the flow to move faster than the bulk flow through the column thereby not allowing adequate contact time for complete ion exchange. The most common problem with cation conductivity is not having adequate flow velocity to maintain the turbulence to bring the sample into close contact with the resin resulting in incomplete exchange. Columns should have an inside diameter of < 60 mm, (2.3 in) and be designed to have a flow velocity of > 300 mm/min, (1 ft/min) and be constructed of non-leaching materials. Tubing used to connect the sample stream to the cation column should be minimized and ideally not be permeable to gases (i.e., CO2) [11]. 4.3.8 Cation Conductivity Resin Exhaustion, Regeneration and Rinse-in The measurement of cation conductivity involves passing the sample stream through a bed of strong acid cation exchange resin. Resin used in the cation column should be sulfonated styrenedivinylbenzene 8% cross-linked, strong acid gel resin in the hydrogen form. The resulting cation4-13


exchanged sample stream is then directed to the conductivity cell for measurement. Successful measurement of cation conductivity requires very careful packing of the resin bed to avoid channeling of the flow. In previous validation studies [12] nuclear grade “indicating resin” (type TDC-1 hydrogen form cation resin) was used, and the data indicated that the beds began to lose efficiency as soon as the top of the bed began to change color. Since the beds must be replaced as soon as the efficiency begins to drop, the color warning provided by these indicating resins is helpful but not an absolute indication of resin efficiency. For new columns a rinse-in period is needed to establish a stable reading. Residual regeneration chemicals and ions that leach out from the resin while the resin is not in use must be flushed out of the resin column before a stable accurate measurement can be obtained. A rinse-in time of a few minutes to several hours may be experienced depending on the condition of the resin being rinsed. If a stable measurement has not been achieved after a few hours of rinse-in the resin may be unacceptable for use. Consult the cation resin supplier’s information on rinse-in times. Alternatively, two resin columns can be used in series with a conductivity cell after each column; a difference in the two conductivity readings signals depletion of the upstream column. If the cation resin is not replaced before exhaustion, cations in the sample will not be exchanged completely, leading to an error in the cation conductivity measurement. This series flow schema also assures the second column is being adequately rinsed to minimize flush time when it is placed in service [12].

4.4 Interferences 4.4.1 Organic and Strong Acids Interferences It is worth noting that interpretation of the difference between cation conductivity and degassed cation conductivity values could be complicated by the presence of low molecular weight organic acids, such as formic acid and acetic acid. However, studies have shown that the actual loss of formic or acetic acid from a degassed cation conductivity system via the system vent is so low that this loss can be disregarded [13]. Further it has been shown that the strong acids formed from neutral salt conversion in the cation column (i.e., HCl or H2SO4 ) are not volatile at degassed cation conductivity tempertures [14]. 4.4.2 Sample Line Interference For low ionic strength water measurements (< 10 uS/cm) long sample lines should be avoided. Ions in the sample steam tend to interact with the surface of the sample lines. These interactions have ion-exchange like characteristics (so called inorganic ion exchange). To the extent practical, all joints on sample lines should be air-tight. Sample line flow velocity should be maintained > 2 m, (6 ft.) per second to prevent sample deposition. Sample flow through the cell should typically be 200 mL/min, unless otherwise specified by the supplier.

4-14


4.5 Calibration The analytical capabilities of on-line conductivity instruments should be checked periodically to demonstrate calibration stability. One method exists for verifying instrument stability; the Line Method [15]. For the Line Method, a calibrated separate conductivity meter, typically a portable conductivity meter, is used to analyze the same sample steam as the installed on-line instrument. The results of the comparison are evaluated against the acceptance criteria (e.g., the measured values agree within ± 0.1 µS/cm). Provided the on-line analyzer agrees within the acceptance criteria the on-line instruments calibration has not changed. If the results are outside the acceptance criteria the on-line instrument should be recalibrated.

4.6 Calibration Checks The conductivity measurement involves two main components; the conductivity cell and the conductivity measuring instrument. The determination and verification of cell constants has been previously discussed. Evaluating a conductivity measuring instrument requires the use of precision resistors with a tolerance of ± 0.1% [8]. The precision resistors are used in place of the measuring cell and temperature compensator. The conductivity check resistance in ohms equals -1 the cell constant (cm ) divided by the conductivity to be indicated (S/cm). The difference between the stated precision resistors resistance and the instrument read out converted to resistance should be compared to the instrument suppliers recommended acceptance criteria. If the difference between these two resistances is outside the instrument suppliers recommended acceptance criteria the conductivity measuring instrument should be recalibrated or replaced. Additional maintenance and calibration activities are recommended in EPRI Report GS-7556 [11].

4.7 Alternative Methods for Determining Conductivity Because of the interrelationship between pH, specific conductivity and ammonia, knowing any two of these variable can lead to an estimation of the third. If ammonia and pH are known then specific conductivity may be estimated. Figure 4-6 shows the relationship between ammonia and specific conductivity.

4-15


Figure 4-6 Relationship Between Ammonia Concentration mg/L (ppm) and Specific Conductivity (µS/cm) at 25°C (77°F) [1] Source: Reference 1

Specific conductivity is a measure of the concentration of ions in solution. Assuming only one ion is in solution then specific conductivity can be calculated from the equivalent conductance [6] for that ion and the conductivity of the solvent, water. The equivalent conductance, Λ, is defined as the conductivity of 1.0 gram equivalent of solute contained between two electrodes spaced exactly 1.0 cm apart. The solution volume and the electrode area are not specified. The 4-16


equivalent conductance is obtained by multiplying the conductivity (specific conductance), κ, by the volume, V, (in milliliters) containing exactly 1.0 gram equivalent weight.

Λ =κ V

Equation 4-5

Since the normal concentration (normality), N, of a solution is the number of equivalents per liter (or 1000 mL, which is 1000 cm 3 ) equivalent conductance is:

Λ=

1000κ N

Equation 4-6

The equivalent conductance ( Λ ) has units of S - cm 2 - equivalent -1 . Conductivity (specific conductance) can be obtained from equivalent conductance as follows:

κ (µ S − cm ) = −1

(

)

Λ S − cm 2 − equivalent −1 N ( equivalent/L ) 3

1000 cm /L

106 µ S/S Equation 4-7

κ = 1000 Λ N Equation 4-6 shows that specific conductivity can be determined from equivalent conductance and the solution normality. The concentration of ions has an effect on conductance; the higher the concentration the lower the equivalent conductance because of the interaction of ions in carrying charges (i.e., “activity” effects). The equivalent conductance at infinite dilution is the value used to calculate the conductivity of most power plant solutions in which ionic compounds are present in the mg/L (ppm) level. Table 4-2 gives the equivalent conductance values for selected ions at infinite dilution. The solution conductivity is the sum of the individual component conductivities, including that of the solvent. This can be stated as follows:

κ (S − cm −1 ) = ∑ Where

Λc Λ Cc + ∑ a Ca Ea Ec

Equation 4-8

Λ c = equivalent conductance of an individual cation, S - cm 2 − eq −1

E c = gram equivalent weight of an individual cation, g - eq −1

C c = concentration of an individual cation in mg/L (ppm) Λ a = equivalent conductance of an individual anion, S - cm 2 − eq −1

E a = gram equivalent weight of an individual anion, g - eq −1 C a = concentration of an individual anion in mg/L (ppm)

4-17


This equation is the same as:

κ (µ S − cm −1 ) = ∑1000 Λ c N c +

∑1000Λ N a

Equation 4-9

a

Where N c and N a are the normalities of the individual cations and anions, respectively. The following example illustrates the use of equivalent conductance of ions in calculating conductivity. Table 4-2 Equivalent Conductances of Separate Ions at Various Temperatures Equivalent Conductance, S-cm2 - equivalent-1 (at infinite dilution) Ion

0°C

18°C

25°C

50°C

75°C

100°C

Ammonium ( NH +4 )

40.02

64.5

74.5

115

159

207

½ Calcium (Ca+2)

30.0

51.0

60.0

98.0

142

191

½ Barium (Ba+2)

33.0

55.0

65.0

104

149

200

Hydronium (H3O+)

240

314

350

465

565

644

Potassium (K+)

40.4

64.6

74.5

115

159

206

Sodium (Na+)

26.0

43.5

50.9

82.0

116

155

41.1

65.5

75.5

116

160

207

Hydroxide (OH )

105

172

192

284

360

439

Nitrate ( NO −3 )

40.4

61.7

70.6

104

140

178

½ Sulfate (SO −42 )

41.0

68.0

79.0

125

177

234

Chloride (Cl-) -

Example: Calculate the conductivity of a 1 mg/L (ppm) NaCl (sodium chloride) solution at 25°C (77°F) from data in Table 4-2. Solution: a. The first step is to calculate the ionic mg/L (ppm) values:

4-18


ppm Na + = ppm NaCl x

= 1 ppm x

Atomic weight of Na Molecular weight of NaCl

22.99 g Na/mole NaCl 58.44 g NaCl/mole NaCl

= 0.393ppm ppm Cl = ppm NaCl x

= 1 ppm x

Atomic weight of Cl Molecular weight of NaCl

35.45g Cl/mole NaCl 58.44 g NaCl/mole NaCl

= 0.607 ppm b. The equivalent weights must be determined: The equivalent weight for Na + is its atomic weight since it has a valence of +1.

E c = 22.9898 g/eq The equivalent weight for Cl − is the atomic weight because it has a valence of -1.

E a = 35.453 g/eq The conductivity can be calculated from the ionic equivalent conductivities from Table 4-2 assuming 25°C (77°F) Λ c = 50.9 S − cm 2 − equivalent −1 Λ a = 75.5S − cm 2 − equivalent −1

4-19


) ∑ ΛE

(

κ µS − cm −1 =

c

Cc +

c

=

Λa

∑E

Ca

a

50.9 75.5 x 0.393 + x 0.607 22.9898 35.45

= 0.870 µS − cm −1 + 1.23 µS − cm −1 = 2.163 µS − cm -1 Solvent Effects: The Conductivity of Pure Water Water ionizes to yield hydronium and hydroxyl ions:

2H 2 O

→ ←

H 3O + + OH −

Equation 4-10

The hydronium and hydroxyl ions contribute to the conductivity of the solution. At 25°C (77°F), the conductivity of pure water is:

κ (µS − cm −1 ) = κ H O+ + κ OH− 3

= 1000 Λ H O+ N H O+ + 1000Λ OH− N OH− 3

3

= (1000 µS − 1cm −3 − S−1 x 350 S − cm 2 − eq −1 x 1.004 x10−7 eq/l )

+ (1000 µS − 1cm -3 − S-1 x192 S − cm 2 − eq −1 x1.004 x10−7 eq/1)

= 0.0351 µS − cm −1 + 0.0193 µS − cm −1 = 0.0544 µS − cm −1 Thus, a sample of pure water will always still have a background conductivity because of the ionization of water:

κ m = κ s + 0.0544 µ S − cm −1 (25°C)

Equation 4-11

where κ m is the measured conductivity and κ s is the conductivity of the ionic constituents in the sample (i.e., the solute). 4-20


4.4 End User Considerations The performance characteristics (range of measurement, accuracy, precision, bias, drift, response time, and signal change) for the monitoring equipment provided by the manufacturer or supplier should be considered when selecting a suitable on-line conductivity monitor. In general, manufacturers determine these characteristics using their own in-house methods. End users should select the instrument that is best suited for the intended analytical application. Other on-line conductivity monitor considerations for pure water include:

•

Select conductivity cells with appropriate cell constants

•

Assure appropriate temperature corrections are applied for each conductivity measurement

•

Select cation columns with appropriate flow characteristics

•

Ensure cation resin is properly regenerated and rinsed for use

•

Be aware of potential carbon dioxide or organic acid interferences with cation and degassed cation conductivity

•

Minimize sample line lengths and use appropriate tubing for connections

•

Validate conductivity cell constants periodically

4.9 References 1.

Cycle Chemistry Guidelines for Fossil Plants: All-volatile Treatment, Revision 1. EPRI, Palo Alto, CA: 2002. 1004187.

2.

Cycle Chemistry Guidelines for Fossil Plants: Phosphate Continuum and Caustic Treatment. EPRI, Palo Alto, CA: 2004. 1004188.

3.

Cycle Chemistry Guidelines for Fossil Plants: Oxygenated Treatment. EPRI, Palo Alto, CA: 2005. 1004925.

4.

Cycle Chemistry Guidelines for Combined Cycle/Heat Recovery Steam Generators (HRSGs). EPRI, Palo Alto, CA: 2006. 1010438.

5.

ASTM D4519-94, “Standard Test Method for On-Line Monitoring of Electrical Conductivity to Determine Anions and Carbon Dioxide in High Purity Water”. American Society for Testing & Materials, Philadelphia, PA.

6.

John Dean ed., Lange’s Hand Book of Chemistry, 12th edition, McGraw-Hill Book Company, 1979, Section 6, pp. 34.

7.

ASTM D1125-91, “Standard Test Methods for Electrical Conductivity and Resistivity of Water”. American Society for Testing & Materials, Philadelphia, PA. 4-21


8.

ASTM D5391-93, “Standard Test Method for Electrical Conductivity and Resistivity of a Flowing High Purity Water”. American Society for Testing & Materials, Philadelphia, PA.

9.

D. Gray, “Cation Conductivity Temperature Compensation”, Ultrapure Water, Volume 16, No. 4, April 1999, pp. 60-63.

10.

O. Zabarsky, “Temperature Compensated Conductivity—A Necessity for Measuring HighPurity Water,” Ultrapure Water, pp. 56-60 (January/February 1992).

11. ASTM D D6504-00, “Standard Practice for On-Line Determination of Cation Conductivity in High Purity Water”. American Society for Testing & Materials, Philadelphia, PA. 12.

Monitoring Cycle Water Chemistry in Fossil Plants, Vol. 3 Project Conclusions and Recommendations, by A.F. Aschoff, D.M. Sopocy, D.T. Eglar, O. Jonas, J.K. Rice, C.C. Stauffer, and W.E. Allmon. EPRI, Palo Alto, CA: October 1991. GS-7556, Volume 3.

13.

M Gruszkiewicz, “Degasssed Conductivity—Comments on an Interesting Reasonable Plant Cycle Chemistry Monitoring Technique, Part 1: Degassing of Low-Molecular-Weight Organic Acids in Technical Degassed Cation Conductivity Monitors”, PowerPlant Chemistry, Volume 6(3), 2004, pp. 177-184.

14.

M. Gruszkiewicz, “Degasssed Conductivity—Comments on an Interesting Reasonable Plant Cycle Chemistry Monitoring Technique, Part 2: Degassing of Carbon Dioxide in Technical Degassed Cation Conductivity Monitors and Temperature Conversion of Cation Conductivity Measured at nearly 100°C to 25°C ”, PowerPlant Chemistry, Volume 6(5), 2004, pp. 279-289.

15.

ASTM D3864-96(2000), “Standard Guide for Continual On-Line Monitoring Systems for Water Analysis”., American Society for Testing & Materials, Philadelphia, PA.

4-22


5 OXYGEN

5.1 Purpose and Use Dissolved oxygen (DO) is an EPRI Core Monitoring Parameter [1-4]. As such, the reason for continuous on-line monitoring of DO is to check the acceptability of water chemistry, thereby ensuring that dissolved oxygen is maintained at acceptable levels—with lower levels in feedwater for cycles employing reducing all volatile treatment-reducing (AVT (R)) and somewhat higher levels when using oxidizing all volatile treatment-oxidizing (AVT(O)) or oxygenated treatment (OT). Dissolved oxygen may also be monitored for one or more of the following reasons:

•

To check the accuracy of water chemistry control, so ensuring that corrosion rates are kept at acceptable low levels.

•

To facilitate the correlation of a water chemistry parameter with plant operating variables, with an aim to optimizing operations (e.g., condenser air removal or deaerator operations).

•

To provide feedback stimulus for automated process control, e.g., for oxygen control on oxygenated treatment (OT).

•

To monitor for condensate pump seal leakage.

•

During and following changes in feedwater treatment.

The data generated by continuous on-line monitoring of dissolved oxygen is used by plant chemistry and operations department personnel. The goal for plant personnel is to maintain dissolved oxygen within prescribed limits.

5.2 Description of Methods In typical fossil power plant steam/water cycles, dissolved oxygen concentrations may be temporarily high during start-up but rarely exceed 400 µg/L (ppb) (parts per billion) during normal operation and are often much less. On-line monitoring techniques that are particularly accurate over the concentration range of 0-400 µg/L (ppb) are therefore of great interest. Three approaches are discussed below. Two are based on diffusion—the galvanic and the polarographic methods—and the third approach is based on establishing equilibrium between dissolved oxygen concentrations on the inside and outside of the sensing probe. 5-1

EPRI Proprietary Licensed Material Oxygen

All three methods make use of a sensing probe immersed in the water sample of interest. In each case, the probe consists of two metal electrodes—an anode and a cathode—in contact with an internal electrolyte that is separated from the water sample by a semi-permeable membrane. Because all three methods measure current (amperes) flowing between the cathode and anode, they may be termed amperometric methods. Figure 5-1 shows a typical oxygen sensing probe.

Figure 5-1 Typical Oxygen Sensing Probe [5] Source: Adapted from Reference 5, Courtesy Mettler-Toledo Thornton

5.2.1 Galvanic Method In the galvanic method (also known as the Mackereth cell method), the anode material is typically composed of lead and the cathode is a much more noble material, such as silver, gold or platinum. The galvanic method is polarographic although it does not require external polarization. Consequently, these electrodes have different electrochemical potentials in the internal electrolyte, and they form a galvanic cell if they are electrically coupled through an external ammeter. The reactions that take place at, say, a platinum cathode and a lead anode in a potassium iodide internal electrolyte are as shown in equations 5-1 and 5-2. 5-2


Cathode:

O 2 + 2H 2 O + 4e - → 4OH -

(Reduction Reaction)

Equation 5-1

Anode:

2Pb + 4I - → 2PbI 2 + 4e -

(Oxidation Reaction)

Equation 5-2

The lead is oxidized to lead iodide and the oxygen is reduced to hydroxyl ions. This results in the flow of an electrochemical current between the anode and cathode. The charge is carried by free electrons in the wires and ammeter within the external circuit, and by charged ions, such as I- and OH-, in the internal electrolyte. Another design [5] uses a sodium bicarbonate electrolyte, a platinum cathode and a lead anode. The electrochemical reactions can be summarized as shown in equations 5-3 and 5-4. Cathode:

O 2 + 2H 2 O + 4e - → 4OH -

Anode:

2Pb + 4OH - → 2PbO + 2H 2 0 + 4e - (Oxidation Reaction) Equation 5-4

(Reduction Reaction) Equation 5-3

In this case, there is no consumption of electrolyte and only the anode is oxidized. Consequently, with a suitable electrolyte composition the anode surface can be kept clean so maintenance frequencies can be extended to upwards of three years. In either case, the magnitude of this current is controlled by the rate at which oxygen arrives at the cathode surface, which is determined by the rate of oxygen diffusion through the semipermeable membrane and the internal electrolyte. The rate at which oxygen diffuses through the semi-permeable membrane is, in turn, proportional to the dissolved oxygen content of the water sample flowing past the membrane. Thus, the electrochemical current flow is proportional to the dissolved oxygen content in the water sample. The current can be converted with amplifiers and appropriate circuitry to provide a current or voltage output for a strip chart recorder, data logger, or control equipment. 5.2.2 Polarographic Method In the polarographic method (also known as the Clark cell method), the anode and cathode materials have different nobilities but not as widely different as in the galvanic method. The anode material is typically silver and the cathode is a more noble material, such as gold or platinum. Consequently, the natural galvanic effect is much smaller, and another approach is used to achieve the necessary sensitivity to oxygen. The potential difference between the cathode and anode (the cell voltage) is increased and held constant by external instrumentation. This cell voltage is typically held constant at a value in the range 0.8 to 1.0 volt. The current flowing between the cathode and anode is measured and, as in the galvanic method, the current is directly proportional to the dissolved oxygen concentration in the water sample flowing past the probe. The reactions that take place at, say, a gold cathode and a silver anode in a buffered potassium chloride internal electrolyte are as shown in equations 5-5 and 5-6. 5-3


Cathode:

O 2 + 2H 2 O + 4e- → 4OH −


Equation 5-5

Anode:

4Ag + 4Cl - → 4AgC1 + 4e -


Equation 5-6

The silver is oxidized to silver chloride and the oxygen is reduced to hydroxyl ions. This results in the flow of an electrochemical current between the anode and cathode. The charge is carried by free electrons in the wires and ammeter within the external circuit, and by charged ions, such as Cl- and OH-, in the internal electrolyte. The magnitude of this current is controlled by the rate at which oxygen arrives at the cathode surface, which is determined by the rate of oxygen diffusion through the semi-permeable membrane and the internal electrolyte. The rate at which oxygen diffuses through the semi-permeable membrane is, in turn, proportional to the dissolved oxygen content of the water sample flowing past the membrane. Thus, the electrochemical current flow is proportional to the dissolved oxygen content in the water sample. The current can be converted with amplifiers and appropriate circuitry to provide a current or voltage output for a strip chart recorder, data logger, or control equipment. 5.2.3 Equilibrium Method In the equilibrium method, the anode and cathode materials are typically platinum. Equilibriumtype probes [6] rely on establishing equilibrium between dissolved oxygen in the sample and inside the probe. When the probe is immersed in the sample, oxygen diffuses through the membrane and is reduced at the cathode as shown in Eq. 5-7. Simultaneously, an equal amount of oxygen is generated at the anode as shown in Eq. 5-8. The diffusion continues until the partial pressure of oxygen on both sides of the membrane is equal and balanced. The current necessary to maintain this equilibrium is converted to the dissolved oxygen concentration in the solution being monitored. Cathode:

O 2 + 4H + + 4e - → 2H 2 O


Equation 5-7

Anode:

2H 2 O → O 2 + 4H + + 4e -


Equation 5-8

As in the diffusion-type approaches, oxygen is reduced at the cathode. However, in equilibrium sensors a measuring circuit causes electrical current to flow through the anode resulting in an oxidation reaction that produces oxygen. The net reaction of oxygen being reduced at the cathode and oxygen being produced at the anode leads to no net consumption of oxygen or water in the overall reaction. The equilibrium is maintained until the dissolved oxygen concentration changes in the sample stream and then a new equilibrium is reached. The resultant current flow is proportional to the dissolved oxygen concentration in the sample stream. The current can be converted with amplifiers and appropriate circuitry to provide a current or voltage output for a strip chart recorder, data logger, or control equipment.

5-4


5.3 Technical Considerations Water samples taken from the steam/water cycle typically have low dissolved oxygen (i.e., a few µg/L (ppb) to several hundred µg/L (ppb)), leading to several analytical challenges. These challenges can be overcome provided appropriate membranes are utilized, electrodes are kept clean, appropriate temperature and pressure compensations are applied, flow rate sensitivity is understood and interferences are minimized. Response times are optimized and calibration and calibration checks are performed as required. Without appropriate consideration to these factors dissolved oxygen measurements may be erroneous. 5.3.1 Membrane Replacement The semi-permeable membrane is usually polytetrafluoroethylene (PTFE), perfluoroalkoxy (PFA), ethylene-tetrafluoroethylene copolymer, Teflon®, high density polyethylene, silicone or a similar material. Most often one type of membrane is used more typically with one type of sensing electrode. Polarographic type sensing electrodes tend to favor the ‘fluoro-type’ membranes. Galvanic type sensors tend to favor the high density polyethylene membranes. Equilibrium-type sensing electrodes tend to favor silicone membranes. One manufacturer [5] claims the thicker Teflon® membrane design reduces the amount of oxygen permeating the membrane thereby reducing sensitivity to coatings and low flow. Teflon® membranes are thicker but allow for faster transfer rate for oxygen. Since conventional fluoro-type membranes allow large volumes of oxygen through the membrane, the potential for depletion of oxygen near membrane surface requires a higher flow rate to assure oxygen levels are not depleted near the sensing membrane surface. Fouling of the fluoro-type membrane surface leads to slower permeation and lower readings. For fluoro-type membranes, the relative tension, thickness and smoothness of the membrane affect its gas permeability. Membrane replacement is required on fluoro-type commercially available DO analyzers after periodic electrode cleaning. Some manufacturers provide aids for installing membranes in a specific configuration. Other manufacturers provide a disposable DO probe to eliminate the need for, and variability in, membrane installation [7]. 5.3.2 Electrode Cleaning 5.3.2.1 Polarographic For polarographic sensors, the calibration slope can change if the anode turns gray from the deposition of silver that is dissolved in the electrolyte as silver chloride (an anodic reaction product). Polarographic type sensors therefore require periodic cleaning to remove silver chloride deposits from the anode.

5-5


5.3.2.2 Galvanic Depending on the type of electrolyte and the size of the anode, galvanic sensing electrodes may require periodic cleaning. Specifically, galvanic sensing electrodes with a potassium iodide or potassium chloride based electrolyte may require infrequent cleaning of the anode when the electrolyte is being replenished. Deposits on the anode should be removed using a soft cloth and gentle cleaning action. Electrodes with sodium bicarbonate electrolyte and large surface lead (Pb) anode may require infrequent cleaning since the anode product, lead oxide (PbO), deposits over a large area. Cleaning for these large surface area anodes should be only be done per the instrument suppliers recommendations when the anode appears fouled. Electrodes with sodium bicarbonate electrolyte and platinum electrodes do not require cleaning [5]. 5.3.2.3 Equilibrium Equilibrium sensors are not typically cleaned or refurbished. Once they have reached their end of life they are replaced with a new sensor. 5.3.3 Temperature and Pressure Compensation for Sensors Oxygen has a temperature-dependent solubility in water at atmospheric pressure as shown in Figure 5-2. If the curve were extrapolated, the solubility at 100°C would be zero as water vapor pressure reaches 100% at boiling—again, at atmospheric pressure. The entire curve would be shifted upwards if the pressure over the water were increased. Working in the opposite direction, deaerators depend on partial vacuum and/or elevated temperatures to remove all dissolved gases. DO sensors are electrochemical devices that take advantage of the gas permeability of polymer membranes to separate the sensing electrodes from the sample. This separation enables a sensor to provide a controlled environment for the electrodes and electrolyte while allowing oxygen to enter from the sample and react. It keeps the electrochemistry fairly well contained and clean. The diffusion rate of oxygen through a membrane is proportional to the partial pressure of oxygen in the sample. Of course the membrane material and thickness also affect the diffusion rate, but they are fixed and those properties are normalized in calibration.

5-6


Figure 5-2 Solubility of Dissolved Oxygen (mg/L (ppm)) vs. Temperature (°C) [5] Source: Adapted from Reference 5, Courtesy Mettler-Toledo Thornton

The oxygen which permeates the membrane reacts at the cathode, producing a current in direct proportion to the quantity of oxygen. That current is the measurement signal which matches the oxygen partial pressure and the concentration of DO, at least at constant temperature. To derive a concentration measurement from partial pressure with varying temperature, the signal must be temperature compensated, based on the relationship in Figure 5-2. That is, the DO concentration in water that a partial pressure represents depends on temperature. The sensor’s RTD (resistance temperature detector) signal is used by the instrument microprocessor to temperature compensate the measurement. In addition, the instrument must compensate for the temperature-dependent diffusion rate of oxygen through the membrane material. Some utility personnel prefer to avoid the need for temperature compensation by ensuring that the temperature of the water sample is maintained at 25°C (77°F) as it passes over the oxygen sensor; in this way, sources of error associated with the temperature compensation algorithm can be avoided. During calibration, air is used as the standard and its oxygen partial pressure depends on the ambient barometric pressure at the time of calibration. Compensation for actual barometric pressure during calibration assures this dependency is accounted for. Pressure compensation is important because the tension in the semi-permeable membrane affects its diffusion characteristics. Compensation is usually handled electronically by the instrumentation, but when the water sample pressure is low (say < 345 kPag (50 psig)), pressure compensation can be done mechanically instead. This may be achieved by incorporating a 5-7


flexible membrane in the walls of the oxygen probe housing so that the external water sample pressure can be transmitted immediately to the internal electrolyte, thereby maintaining equal pressure on both sides of the semi-permeable membrane. Another strategy to eliminate the need for pressure compensation is to provide an overflow head-cup on the sample supply line to assure a low constant pressure on the membrane. 5.3.4 Flow Rate Sensitivity For accurate measurements, it is essential that the oxygen content of the water at the outer surface of the semi-permeable membrane is representative of the bulk water passing through the flow cell. For diffusion type sensors, oxygen diffuses out of the sample through the semipermeable membrane and is reduced in the sensor. Therefore, a continuous flow of fresh sample must be supplied to the membrane surface. Minimum flow rates (e.g., 50-200 mL/min) are specified by the manufacturer of the monitoring equipment to ensure that the rate of oxygen removal by the sensor is negligible compared with the oxygen supplied to the flow cell. For equilibrium type sensors, oxygen diffuses out of the sample and into the semi-permeable membrane only until equilibrium is reached. Therefore, the sample must only be supplied to the membrane at a flow rate sufficient to allow for representative sampling of the process being monitored. In either case, an upper flow rate limit may also be specified, above which the water flow becomes so turbulent that the membrane vibrates and induces convective transport of oxygen and other species through the internal electrolyte. This condition would lead to an overestimate of the dissolved oxygen concentration. If there is a period when flow to the analyzer may be completely interrupted (such as when a unit is not in service), it is advisable to avoid prolonged sensor response to high oxygen levels by terminating power to the analyzer. 5.3.5 Response Time Generally it is better to have a fast responding sensor. Fast response time minimizes the time a sensor is out of measuring range after calibration and minimizes the time to see a change in DO concentration in the process stream. 5.3.5.1 Response Time After Air Calibration Most modern sensors can achieve 90% reduction from air saturated conditions within 60 seconds after calibration and be fully within normal operating ranges within 20-30 minutes. For a polarographic-type oxygen sensor, an auxiliary ring-shaped cathode (“guard ring electrode”) may be installed to surround the main centrally located disk-shaped cathode. One purpose of the auxiliary cathode is to reduce residual oxygen present in the electrolyte after air calibration. The auxiliary cathode is also designed to reduce or eliminate the adverse effect of 5-8


interfering species that may initially be present in the internal electrolyte or that are created by the anodic reaction. In the absence of the auxiliary cathode, these species would diffuse to the main cathode and contribute to the total measured current, thereby leading to an overestimate of the dissolved oxygen concentration. With the auxiliary cathode in place, these species are removed before they can reach the main cathode. For all types of oxygen sensors, sample line and sample flow cell considerations should be optimized to minimize the time it takes to purge residual DO out of the sample chamber and lines. After air calibration it is important to understand that all of the sample lines may be full of air, and that all the residual oxygen must be purged out and/or dissolved into the sample liquid, and then dissipated before the readings will read the correct “system” value. The larger volume of the sample line that is opened to air during the air calibration, the longer is the recovery time. Also, the slower the flow rate after calibration, the longer the recovery time. So it is best to first purge the air out with a relatively low (gentle) flow rate, then after the lines are full of water, increase the flow rate to reduce the time required to dissipate the remaining DO. 5.3.5.2 Response Time Due to Changes in Dissolved Oxygen Concentration in the Process Stream Once residual DO is purged from the sample chamber and lines, a change in the process stream DO can be detected by all types of sensing electrodes in a matter of a few minutes. Response time tends to increase for polarographic type sensors over a calibration cycle due to silver deposition on the cathode. See section on Interferences which follows for more detailed discussion of this phenomenon.

5.4 Interferences 5.4.1 Oxygen Contamination Oxygen (in the form of air) is an omnipresent contaminant. Sample line and instrument sample chamber leaks are likely sources of oxygen contamination. All sample lines and flow chambers should be air-tight and be constructed of stainless steel (e.g., Type 316SS). Polyvinylidene flouride (PDVF) or Nylon tubing are suitable for temporary connections if needed. One helpful strategy to determine if dissolved oxygen readings are being contaminated is to adjust sample flow. Flow rate increases of up to 50% may be appropriate for this diagnostic check; however, do not exceed instrument suppliers maximum flow rate. If increasing flow produces a decrease in DO, then some of the DO is likely coming from a leak since, as the sample flow increases and the source leak remains the same, the resultant DO goes down because it is being diluted by the additional sample flow. If the DO does not decrease with an increase of sample flow then the DO reading is correct. If the DO increases as the sample flow rate is increased then the original sample flow rate was most likely too low.

5-9


5.4.2 Sample Conditioning Hot samples should be cooled as close as possible to the sampling point to inhibit the reaction of dissolved oxygen with hydrazine or other reducing agents as the sample flows through the sample line to the monitoring instrument. New sample lines should be conditioned in accordance with ASTM D 3370 [8]. 5.4.3 Electrolyte Polarographic type sensors and galvanic type sensors that use an iodide or chloride salt containing electrolyte consume electrolyte during operation. Periodic recalibration is necessary to adjust for this consumption and for other factors such as the electro-deposition of silver on the cathode. Periodic electrolyte replacement is also required. 5.4.4 Stray Current For polarographic sensors, dissolved silver ions from the anode may contribute to stray current from the anodic oxidation reaction shown in equation 5-9.

4Ag + 4Cl - → 4AgC1 + 4e -

Equation 5-9

5.4.5 Membrane Fouling, Positioning and Tension Deposits on the permeable membrane (e.g., iron oxides) will slow instrument response and can lead to a negative bias. For equilibrium sensors this has less impact on DO measurements since only a small amount of oxygen is needed to establish equilibrium. Buildup of organics in sample lines or sample chambers will impact DO measurements. Affected surfaces should be cleaned with a dilute biocide. Membrane tension and positioning can affect sample read out. Manufacturer’s instructions for membrane installation should be followed. 5.4.6 Hydrogen For sample streams containing hydrogen, i.e., stator cooling water, dissolved hydrogen can cause significant negative errors in equilibrium probes. Equilibrium type probes are not recommended for these applications. Galvanic type probes are not affected by dissolved hydrogen (H2) and are recommended for use when hydrogen is present. Polarographic type sensors must be compensated for dissolved hydrogen concentrations in the range of 25-50 cc/kg at STP in the sample stream.

5-10


5.4.7 Other Cycle Chemistry Additives Volatile additives in the sample stream (reducing agents, ammonia, hydrogen and other neutralizing amines) may pass through the probe and cause unwanted reactions leading to negative errors. The instrument supplier’s manual should be consulted for relevant cautions and limitations. 5.4.8 TDS Although not typically an issue with power plant steam water cycle samples, high dissolved solids (> 1000 mg/L (ppm)) affect the solubility of oxygen. Some analyzers provide a salinity correction which allows the user to input a correction factor for salinity in parts per thousand [6].

5.5 Calibration Because of the difference in operating principle for each of the three sensing electrodes, there are different calibration needs. 5.5.1 Polarographic Polarographic sensing electrodes require calibration each time the anode and cathode are cleaned and each time the internal electrolyte is replenished. The time between calibrations is dependent on the DO concentration being measured. Higher DO concentrations generally mean that more frequent calibrations are necessary. For instance based on the acceptability of calibration check results, monthly or quarterly calibrations may be required. 5.5.2 Galvanic For galvanic sensors that use a salt containing electrolyte (e.g., iodide or chloride) the anode is oxidized as follows:

2Pb + 4I - → 2PbI 2 + 4e -

Equation 5-10

(Oxidation Reaction with potassium iodide salt electrolyte) Periodic cleaning of the anode is required. Additionally since electrolyte is consumed in this galvanic sensing electrode design, periodic replenishment of electrolyte is necessary. The time between calibrations is dependent on the DO concentration being measured. Higher DO generally means more frequent calibrations. Monthly or quarterly calibrations may be required. For galvanic sensors using a sodium bicarbonate electrolyte, the electrolyte is in dissociation equilibrium with water which produces hydroxide ions. 5-11


Na + + HCO3- + H2O ←⎯⎯→ Na+ + H2CO3 + OH-

Equation 5-11

The anode is oxidized as follows:

2Pb + 4OH - → 2PbO + 2H 2 O + 4e -

Equation 5-12

(Oxidation Reaction with sodium bicarbonate electrolyte) Hydroxide ions are also generated by the cathode counter reaction (Eq. 5-1) proportional to dissolved oxygen concentration. Since electrolyte is not consumed in this reaction, the electrolyte is not required to be replenished. With proper design the anode surface is large in order to reduce the affects of oxidation and extend the intervals between probe cleaning. The time between calibrations is dependent on the DO concentration being measured. Higher DO generally means more frequent calibrations. Yearly or bi-yearly calibrations may be required. For galvanic sensors the anode can be cleaned with simple scrubbing, the electrolyte is not a proprietary solution and can be made in-house, and the replacement of the Teflon® membrane is easily accomplished without special tools or training. 5.5.3 Equilibrium For equilibrium sensors, since there is no consumption of the electrolyte and no oxidation of the anode frequent calibrations are not necessary. Consequently, based on the acceptability of calibration check results, only yearly or bi-yearly calibrations may be required. 5.5.4 Air Calibration for All Sensor Types In all of these cases, the on-line oxygen monitor allows span calibration by exposure of the probe to either air-saturated water or water vapor-saturated air. This is possible because the composition of air is constant over the entire globe, having 20.94% oxygen in dry air. This span calibration can be performed by plant personnel by manually turning a calibration knob or by pressing a button to activate a self-calibration function of the instrument. Some instruments can also be set to perform automatic span calibration periodically (such as every 7 days). During this procedure, the flow cell is automatically drained, thereby exposing the probe to moist air. Because barometric pressure has a direct affect on oxygen concentration in the atmosphere, a correction for barometric pressure is necessary during DO sensor calibration. Air, which contains 20.94% oxygen, will have an oxygen partial pressure of 159 mm Hg if the total pressure is 760 mm Hg. Data tables [9] of oxygen solubility are normally referenced to obtain this value. When conditions differ from this, an altitude or pressure correction must be made. The correction is made using equation 5-10.

5-12


S=S’ (P-p)/ (760-p)

Equation 5-13

Where: S is the solubility of oxygen at the barometric pressure of interest, S’ is the solubility at 760 mm Hg at a given temperature, P is the barometric pressure in mm Hg, and p is the partial pressure of water vapor at the given temperature. Due to the large difference between the calibration concentration of oxygen and the routinely measured concentration of oxygen in power plant systems it may take hours for all trace oxygen to diffuse out of the probe allowing accurate measurements of the process stream.

5.6 Calibration Checks for All Sensor Types The analytical capabilities of on-line DO instruments should be checked periodically to demonstrate calibration stability. Three methods exist for verifying instrument stability; the Standard Injection Method [10], the Line Method [11] or by using a Faraday cell [12]. For the Standard Injection Method, a known DO near the expected DO of the process being monitored is analyzed by the on-line instrument and the results are compared to the acceptance criteria (e.g., agree within ± 3 sigma or ± 10%). Provided the on-line analyzer agrees within the acceptance criteria, the on-line instruments calibration is considered to be acceptable. If the results are outside the acceptance criteria, the on-line instrument should be recalibrated. Introducing a known standard near the expected DO is no trivial task. One approach is to prepare a calibration gas comprising a low concentration of oxygen in nitrogen and to replace the normal sample flow to the DO analyzer with the calibration gas for the calibration period. However any source of air in-leakage would bias the test results high. A second approach [13] is to bubble a calibration gas into a capped, vented sample container with the appropriate DO probe immersed in the sample. The expected DO can be calculated based on atmospheric pressure, fugacity and solubility of the calibration gas. Again, any source of air in-leakage would bias the test results high. For the Line Method a calibrated separate DO meter, typically a portable DO meter, is used to analyze the same sample stream as the installed on-line instrument. The results of the comparison are compared to the acceptance criteria (e.g., agree within ± 3 sigma or ± 10%). Provided the comparison results agree within the acceptance criteria, the on-line instruments calibration is considered to be acceptable. If the results are outside the acceptance criteria, the on-line instrument should be recalibrated. Alternatively, a suitable wet chemistry method (Ampoules ASTM D 5543 [14], Indigo Carmine ASTM 888 [15]) may be used to determine DO in the sample stream and the results compared as above. For the third approach, a Faraday cell is used to generate a known concentration DO in the sample inlet by dissociating water. This is done by passing a current from one electrode through the sample water to a second electrode positioned some distance away from the first. The 5-13


Faraday cell is designed so that the dissociated oxygen dissolves in the flowing sample. The corresponding dissociated hydrogen produced is not in sufficient quantity to affect instrument readings. The calculated increase in O2 can be measured with the DO meter. If there is agreement to the acceptance criteria (e.g., agree within ± 3 sigma or ± 10, the on-line instruments calibration is considered to be acceptable. If the results are outside the acceptance criteria, the on-line instrument should be recalibrated. Figure 5-3 shows the response of a polarographic oxygen sensor to a 20 µg/L (ppb) oxygen addition generated by a Faraday cell.

Figure 5-3 Response of a Polarographic Oxygen Sensor to a 20 µg/L (ppb) Oxygen Addition Generated by a Faraday Cell [12] Source: Adapted from Reference 12

5.6.1 Zero Point for All Sensor Types Because the current flowing in an on-line amperometric dissolved oxygen analyzer is directly proportional to the dissolved oxygen concentration (more precisely, the partial pressure), a zero current should always represent zero oxygen. Thus, a zero-point calibration is not strictly necessary, although periodic correction for zero drift (to maintain electronic zero) may be necessary on some instruments.

5-14


5.6.2 Maintenance for All Sensor Types On-line oxygen monitors may be checked using substitute resistors in place of the probe. The instrument supplier’s instructions should be consulted for specific details. Since dissolved oxygen is typically corrected for the temperature of the sample stream, periodic verification of the temperature sensor is recommended. The instrument supplier’s instructions should be consulted for specific details. Temperature should be correct to within +/- 1°C (1.8°F). Calibration and maintenance procedures are typically described in literature supplied with the monitoring equipment by each manufacturer but, in some instances, it may be more appropriate to follow the maintenance and calibration activities recommended in EPRI Report GS-7556 [14].

5.7 Alternative Methods ASTM D5543-94 [15], and ASTM D 888-87 [16] provide alternative methods for determining DO. However, neither colorimetric method is suitable for on-line determination. Following is a discussion of a recently developed oxygen sensor technology. 5.7.1 Luminescent Oxygen Sensors [17] Optical sensing of oxygen originates from the work of Kautsky in 1939 where he demonstrated that oxygen can dynamically quench the fluorescence of an indicator (decrease the quantum yield). This principle has been reported in various fields of application such as monitoring aquatic biology in waste water tests for blood gas analysis and cell culture monitoring. The method is now recognized by ASTM for the measurement of oxygen in water. Compared to classical oxygen detection using electrochemical sensors, luminescent technology offers several advantages such as no oxygen consumption, independence from sample flow velocity, no electrolyte and low maintenance. Optical sensing of oxygen is based on the measurement of the red fluorescence of a dye/indicator illuminated with a blue light as show in Figure 5-4. The dye fluorescence is quenched by the presence of oxygen. The oxygen concentration can be calculated by measuring the decay time of the fluorescence intensity as shown by Figure 5-5. The higher the oxygen concentration is, the shorter the decay time will be. By modulating the excitation, the decay time is transformed into a phase-shift of the modulated fluorescence signal, which is independent of fluorescent intensity and thus of potential aging (Figure 5-6).

5-15


Figure 5-4 Principle of Optical Oxygen Detection Using Fluorescent Dye [17] Source: Adapted from Reference 17

Figure 5-5 Fluorescence Density Decay Time as a Function of Oxygen Concentration [17] Source: Adapted from Reference 17

5-16


Figure 5-6 Phase Shift of Modulated Signals [17] Source: Adapted from Reference 17

The oxygen partial pressure (pO2) is then linked to the corresponding phase-shift measurement (Φ) to build the sensor calibration curve (Figure 5-7). This curve is described by the SternVolmer equation, (Eq. 5-14), where Ksv is the indicator quenching constant (in mbar-1) representing the quenching efficiency of the oxygen and thus the sensor sensitivity, f0 is a constant and Φ0 is the phase-shift at zero oxygen representing the unquenched fluorescence decay time of the dye. The calibration curve thus relies on two parameters: the phase-shift at zero oxygen and the luminescent spot sensitivity, Ksv. The dissolved oxygen concentration is then calculated with Henry’s law using the water solubility curve as a function of the temperature.

pO 2 =

(Φ0 − Φ ) K SV [Φ − Φ 0 (1 − f 0 )]

Equation 5-14

5-17


Figure 5-7 Stern-Volmer Calibration Curve [17] Source: Adapted from Reference 17

5.7.1.1 µg·kg-1–Resolution Optical Sensor to Monitor Dissolved Oxygen The measuring system consists of four key components. First, a specific sensitive luminescent spot has been developed to reach µg/L (ppb) resolution (Figure 5-8). This exchangeable spot is fixed on an optical fiber probe connected to the measuring electronics. In the instrument, a high resolution digital phase meter is integrated together with the opto-electronics component (excitation light and light detection device). Finally, a flow chamber equipped with a solenoid valve is connected to the sample line. The role of the solenoid valve is to switch from the measurement sample to the calibration sample at user programmable intervals to either verify or calibrate the system. The calibration procedure is therefore completely automatic and operator independent.

5-18


Figure 5-8 Luminescence Sensor Design [17] Source: Adapted from Reference 17

5.8 End User Considerations The performance characteristics (range of measurement, accuracy, precision, bias, drift, response time, and signal change) for the monitoring equipment, provided by the manufacturer or supplier should be considered when selecting a suitable on-line DO instrument. In general, manufacturers determine these characteristics using their own in-house methods. End users should select the instrument that is best suited for the intended analytical application. Other on-line DO instrument considerations include:

•

Adequate temperature and pressure compensation

•

Appropriate membrane selection

•

Frequency of electrode cleaning

•

Sensitivity to flow rate

•

Hydrogen sensitivity

•

Frequency of electrolyte replacement

•

Membrane fouling considerations

•

The use of a guard ring in polarographic sensors to a)reduce residual oxygen present in the electrolyte after air calibration and b)reduce or eliminate the adverse effect of interfering species that may initially be present in the internal electrolyte or that are created by the anodic reaction

•

Response time

5-19


•

Ease of servicing/cleaning sensor internals and membranes

•

Appropriate air-tight sample lines

•

Ease of calibration and on-going verification of instrument performance

5.9 References 1.

Cycle Chemistry Guidelines for Fossil Plants: All-volatile Treatment Revision 1. EPRI, Palo Alto, CA: 2002. 1004187.

2.


3.


4.


5.

“A New Approach to Low-Level Dissolved Oxygen Measurements”, David Gray, Thornton Inc., Proceedings of the International Water Conference October 2001, Pittsburgh, PA.

6.

“Honeywell DL5 PPM and PPB Dissolved Oxygen Probe”, Personal Correspondence, Alan Hatcher, Honeywell, Dan Meils, Scientech, June 2006.

7.

Waltron LLC, µAI-9437, Dissolved Oxygen Instruction Manual.

8.

ASTM D3370-95a (2003) e1, “Standard Practices for Sampling Water from Closed Conduits”. American Society for Testing and Materials, Philadelphia, PA.

9.

th John Dean ed., Lange’s Hand Book of Chemistry, 12 edition, McGraw-Hill Book Company, 1979.

10.

Advanced Laboratory QA/QC Practices, Daniel Meils, SCIENTECH, Inc., 2006.

11.

ASTM D3864-96(2000), “Standard Guide for Continual On-Line Monitoring Systems for Water Analysis”. American Society for Testing and Materials, Philadelphia, PA.

12.

“Recent Progress in the Design and Operation of Fully Automated Oxygen Monitoring Systems.” H. Maurer, Swan Analytical Instruments, Scientech On-line chemistry Process Instrumentation Seminar, Clearwater, Fl, 1995.

13.

ASTM D 5462-02, “Standard for On-Line Measurement of Low-Level Dissolved Oxygen in Water”. American Society for Testing and Materials, Philadelphia, PA.

5-20


14.

Monitoring Cycle Water Chemistry in Fossil Plants”, Vol. 3 Project Conclusions and Recommendations by A.F. Aschoff, D.M. Sopocy, D.T. Eglar, O. Jonas, J.K. Rice, C.C. Stauffer, and W.E. Allmon. EPRI, Palo Alto, CA: October 1991. GS-7556.

15.

ASTM D5543-94, “Standard Test Methods for Low-Level Dissolved Oxygen in Water, Test Method A—Color Comparator Test Method Using Self-Filling Glass Ampoules”. American Society for Testing and Materials, Philadelphia, PA.

16.

ASTM D 888-87, “Standard Test Method for Colorimetric Indigo Carmine, Test Method A”. American Society for Testing and Materials, Philadelphia, PA.

17.

Dunand, Frank A., Ledermann, Nicolas, Hediger, Serge, Power Plant Chemistry, 2006 8(10), 603.

5-21


6 OXIDATION-REDUCTION POTENTIAL

6.1 Purpose and Use Oxidation-reduction potential (ORP) [also known as the reduction-oxidation (redox) potential] is listed as an EPRI Core Monitoring Parameter [1], in mixed metallurgy systems using all volatile treatment including a reducing agent—ATV(R). The purpose for monitoring ORP is to assure the feedwater is in a reducing condition as needed to minimize copper transport when operating with AVT(R) Chemistry. ORP has also been used to monitor changes to AVT(O) and OT but this is not required. This chemical treatment scheme refers to the feedwater treatment. In other power plant systems such as; cooling tower systems and drinking water / service water systems, ORP reflects the balance between oxidizing disinfectants and dechlorinating agents. Oxidizing biocides include gaseous chlorine, bleach, chlorine dioxide, chlorinated and brominated isocyanurates, calcium hypochlorite, bleach-bromide mixtures, halogenated hydantoins, and peroxide. Release of these oxidizing biocides into surface waters is potentially harmful to the environment and is closely controlled. As a result, chemical dechlorination is often required prior to any surface discharge. Many water purification components are also intolerant of oxidizing biocides (reverse osmosis membranes and ion exchange resins) and must be protected. Typical dechlorination products include sulfur dioxide, sodium meta-bisulfate, and activated carbon. ORP serves as a useful indicator that has been used to regulate the addition of oxidizing biocide to the system and as an indicator of proper dechlorination after the disinfection is complete. ORP is continually monitored on-line in units with copper alloy feedwater system components to monitor the chemistry environment. Negative ORP values indicative of a reducing environment are needed under all conditions to minimize copper corrosion and transport.

6.2 Description of Method This subsection presents basic background information on the principles of making corrosion potential (also called electrochemical corrosion potential (ECP)) and ORP measurements. Since these potentials have been measured in power plants only in recent years, it is appropriate to briefly review the terminology relating to these potentials. ORP and ECP are specific types of electrochemical potential measurements and are not synonymous. Both these types of measurements are quantifications of the potential difference between a metal electrode and a reference electrode immersed in the environment of interest. In the case of the ORP 6-1

EPRI Proprietary Licensed Material Oxidation-Reduction Potential

measurement, the metal electrode is a noble metal (e.g. Pt) and, in the case of the ECP measurement, the metal electrode is a metal used to fabricate one of the power plant components of interest. The value of the measured potential is determined by the specific oxidation and reduction reactions occurring at the metal/environment interface. ORP values and ECP values can differ by hundreds of millivolts and can respond quite differently to changes in system conditions. The corrosion reactions occurring on the system component will control the value of ECP while the platinum electrode used for the ORP measurement—being a noble metal and presumed to be immune to corrosion—will not experience the same change. Instead, ORP is more a reflection of the oxidative power of the aqueous environment (see below). ORP is defined as electromotive force between a noble metal electrode and a reference electrode when immersed in a solution [2,3]. Generally the system is not in equilibrium and the redox 2+ 3+ species are generally not conjugately related (e.g. made up only of Fe and Fe ions that are related by the charge transfer reaction Fe3+ + e- → Fe2+ ). Indeed, complex aqueous systems contain many conjugate pairs such as Fe2+/Fe3+, H2/H+, HOCl/OCL- and O2/H2O in aerated water. The normal ORP convention is to use a platinum sensing electrode although gold or silver is sometimes used. In the case of an ECP measurement, the metal electrode may be the power plant component itself (such as a pipe or heat exchanger tube) or it may be a separate piece of metal (a “probe”) having a chemical composition similar to the component of interest and immersed in the environment of interest. When a metal, such as iron, is corroding in an environment, such as water containing dissolved oxygen, the primary oxidation reaction is irreversible and heterogeneous:

e.g., Fe → Fe 2+ + 2e _

Equation 6-1

This reaction, together with the primary reduction reaction, e.g., H 2O +

1

2

O 2 + 2e − → 2OH −

Equation 6-2

control the corrosion potential measured. Theoretically, each corrosion product is thermodynamically stable only in a specific corrosion potential range. A change in the corrosion 2+ reaction product from Fe to Fe3O4 to Fe2O3, for instance, would be reflected in a change in the ECP. When the metal is platinum or some other noble, non-corroding metal electrode, the potential measured can still be considered a corrosion potential of sorts but, more correctly, it is an ORP measurement. Under such circumstances, the primary oxidation reaction, occurring at the metal/environment interface is simply the reverse of the reduction reaction. For instance, the reduction reaction in Eq. 6-2 would be in equilibrium with the following oxidation reaction: 2OH - → H 2O + 1 2 O 2 + 2e −

6-2

Equation 6-3


and the rates of the forward and backward reactions are usually quite small. Here, the oxidation and reduction reactions do not involve (irreversible) corrosion; rather they are reversible, homogeneous reactions reflecting the oxidative power of the environment. The platinum (or other noble metal) electrode merely provides a surface upon which the reversible reactions can be detected and monitored as an ORP. The oxidation reaction (e.g. Eq. 6-3) also occurs on corroding metals but, here, the corrosion reaction (e.g. Eq. 6-1) normally occurs at a much higher rate and it controls the value of the ECP. In summary, while ORP and ECP (corrosion potential) are measured in a similar way, the former reflects the oxidative power of the environment while the latter reflects the corrosion reaction occurring on the metal surface. The ECP may be dependent on the ORP in that an increase in ORP may be accompanied by an increase in the corrosion potential, but ORP and corrosion potential seldom have the same value. Two important reversible reactions that influence the ORP of a power plant environment can be written as follows:

H 2O + 1 2 O 2 + 2e- ↔ 2OH − , 2H+ + 2e- ↔ H2 ,

E = 1228 . − 0.0591pH + 0.0148 log pO 2

E = 0.000 − 0.0591pH − 0.0296 log p H 2

Equation 6-4 Equation 6-5

where E is the ORP at 25°C (77° F) measured with respect to a standard hydrogen electrode (SHE); and pO 2 and p H 2 are respectively the partial pressures of oxygen and hydrogen in the environment. The 2-directional arrows in Eq. 6-4 and Eq. 6-5 indicate the reversibility of the reactions. Similar relationships can be derived for other temperatures. From these equations, it is clear that environments containing dissolved oxygen have a higher ORP than those containing hydrogen ions (or hydronium ions) at all pH values. This means that oxygen is a more powerful oxidant than hydrogen ions. Furthermore, in both equations, the ORP is decreased by about 59 mV for each increase of 1 pH unit at 25°C (77° F). The ORP also decreases as the partial pressure of oxygen decreases and the partial pressure of hydrogen increases. ORP is only used to confirm reducing conditions with AVT(R) in units with copper in feedwater. As might be expected, the ORP is also influenced by the presence of reducing agents like hydrazine, and ORP measurements have been used to ensure that hydrazine is maintained at the correct level. However, ORP depends on far too many factors to sensibly exhibit and predict any sort of “cause and effect” relationship for the parameter of interest. The best example of this: if the sample contains oxygen in quantities of >10 µg/L (ppb), it is generally not possible to add enough reducing agent to establish a negative ORP indicative of a reducing environment. ORP is a function of the system pH, the partial pressures of hydrogen and oxygen, the mass transport properties, the flow rate, and the materials in the system.

6-3


6.3 Technical Considerations An ORP analyzer has similar components as a pH analyzer. The ORP analyzer is comprised of a volt meter, reference electrode and sensing electrode. The potential between t he reference electrode and the sensing electrode are measured by the volt meter and a readout in mV is displayed. 6.3.1 Voltmeter Selection

Most laboratory and process pH meters are also suitable for measuring ORP by utilizing an appropriate set of electrodes and choosing a suitable meter output scale. The ORP readout will be in millivolts (mV) units with a typical response of -1000 to +1000 mV being adequate. Temperature compensation of ORP measurement is related only to the electrode properties described by the Nernst Equation. 0 E = E + 2.3 (RT/nF) logQ

Equation 6-6

where: E

=

measured potential

E

=

potential when all components involved in the reaction are at unit activity and 25 °C (77° F)

R

=

gas constant

T

=

absolute temperature, °K (°C = T – 273.15)

F

=

Faraday constant

n

=

number of electrons involved in the reaction

Q

=

product of all the activities of the oxidants divided by the product of all the activities of the reductants

0

Automatic temperature compensation is normally not attempted in ORP measurements in cycle chemistry measurements. Every attempt should be made to measure the sample near the reference temperature of 25 °C (77°F). For readings not made at this standard temperature, the error is fairly minor and is not considered in power plant ORP measurements. These effects are on the order of 0.1 to 0.2 mV/ °C when using the most common Ag/AgCl, Sat. KCl reference electrode. It is important to mention that the ORP of an aqueous solution is almost always sensitive to pH variations, even to reactions that do not appear to involve hydrogen or hydroxyl ions. The ORP 6-4


generally tends to increase with increasing hydrogen ion concentration (lowering pH) and decrease with increasing hydroxyl ion concentrations (elevating pH values). At 25°C (77°F), ORP values vary approximately 59 mV/pH unit—this property is useful during electrode calibration but is troublesome when taking readings in samples with varying pH values, such as a feedwater system. 6.3.2 Reference Electrodes

Reference electrodes used for ORP measurements must be maintained and periodically checked for proper operation. It is common practice to check reference electrode potentials against the potentials of other nominally identical reference electrodes. All measurements should agree within one millivolt and any reference electrode not meeting this standard should be withdrawn from service. Some reference electrodes can be rejuvenated and brought back into service simply by replacing the internal electrolyte, where possible. Other reference electrodes are not designed for rejuvenation and must be discarded, as in sealed gel type electrodes. Ideally, all reference electrodes should be calibrated periodically because their response characteristics tend to change with time. Here, calibration involves checking that the measured reference electrode potential [measured with respect to a standard hydrogen electrode (SHE)] is equal to the theoretical potential. The apparatus needed to make such a calibration using a hydrogen electrode as the standard is described in ASTM G5-87 [4]. The rate that a reference electrode degrades depends partly on the type of service it sees and the contaminants it picks up from the service environment (see Section 18). For instance, a silver/silver chloride (Ag/AgCl) electrode, which can be used at temperatures up to 300°C (572°F), is susceptible to irreversible damage if contaminated with sulfide. The formation of silver sulfide on the electrode surface causes a permanent change to occur that cannot be fixed by recalibration. Calibration may also be impossible if the silver chloride component has been reduced to silver by hydrogen that has diffused through the reference electrode internal electrolyte from the bulk environment. Dilution of the internal electrolyte (usually 3 molar or saturated potassium chloride solution) with the bulk environment may also occur, especially at high temperatures (e.g. above 200°C (363°F) and when there is a pressure differential between the inside and outside of the reference electrode chamber. Although ORP is not measured at these temperatures in power plant applications, dilution of the KCl will cause the reference potential to drift from the initial value, so periodic replacement of this solution is recommended in electrode designs that allow this. 6.3.3 ORP Sensing Electrode

Corrosion scientists and electrochemists in the research world typically report ORP values on the SHE scale even though the measurements are rarely actually made using an SHE. In practice, a more robust, more convenient reference electrode is used and the value measured is converted to the SHE scale either by calculation or by adjustment of instrument zero during calibration. In the power industry, however, ORP values are usually reported with respect to the reference 6-5


electrode contained in the ORP probe. Such a practice is potentially confusing only because the reference electrode used is not routinely mentioned when reporting the ORP value. Fortunately, the great majority of ORP probes used in power plants use one type of reference electrode: the Ag/AgCl reference electrode. Also fortunately, the internal electrolyte in the reference electrode, which influences the electrode potential, is typically potassium chloride with a concentration ranging from 3M to a saturated solution. An increase in the chloride concentration over this range would decrease the reference electrode potential by no more than 11 mV for temperatures in the range 20-30°C (68-86°F) (see Table 6-1). Since errors of 11 mV would be considered small for most practical ORP measurements, knowledge of the precise KCl concentration is usually not an issue at near-ambient temperatures. Corrections for KCl concentration may be necessary if the ORP is measured at higher temperatures. Nevertheless, confusion has arisen when ORP values measured versus a Ag/AgCl reference electrode are inadvertently compared with others measured versus an SCE or SHE. Consequently, it is recommended that the reference electrode used for the measurement is always mentioned along with the measured ORP values (e.g., ORP = 0.100 V(Ag/AgCl, Sat. KCl)). Table 6-1 To Convert ORP or ECP Values Measured Using Reference Electrode #1 to Values on Reference Electrode #2 Scale, Add the Indicated Conversion Factor to the Measured Potential [3] Convert From Potential (mV) Versus Reference Electrode #1* (Listed Below)

Add the Conversion Factor Below to Convert To Potential (mV) Versus Reference Electrode #2* To SCE

To Ag/AgCl (3M KCl)

To Ag/AgCl (Sat. KCl)

To SHE

20°C 25°C 30°C 20°C 25°C 30°C 20°C 25°C 30°C 20°C 25°C 30°C (68° F) (77° F) (86° F) (68° F) (77° F) (86° F) (68° F) (77° F) (86° F) (68° F) (77° F) (86° F)

From SCE

—

—

—

+34

+35

+36

+45

+45

+45

+247 +244 +241

From Ag/AgCl (3M KCl)

-34

-35

-36

—

—

—

+11

+10

+9

+213 +209 +205

From Ag/AgCl (Sat. KCl)

-45

-45

-45

-11

-10

-9

—

—

—

+202 +199 +196

From SHE

-247

-244

-241

-213

-209

-205

-202

-199

-196

—

—

* Ag/AgCl = silver/silver chloride; SCE = saturated calomel electrode; SHE = standard hydrogen electrode. Note: the presence of liquid junction potentials may result in the listed conversion factors being in error by 1 or 2 mV.

When reporting and comparing comparing ORP measurements, they should be referred to the same reference electrode scale. As described above, the conversion factors shown in Table 6-1 can be used to convert from one reference electrode scale to another. For instance, an ORP of

6-6

—


+100 mV measured at 25°C (77°F) versus a Ag/AgCl, Sat. KCl reference electrode can be converted to an ORP versus an SHE, by adding 199 mV:

Therefore,

ORP (Ag/AgCl, Sat. KCl) = 100 mV

Equation 6-7

ORP (SHE) = 100 mV + 199 mV = 299 mV

Equation 6-8

6.4 Interferences In general, ORP measurements are not subject to solution interferences from color, turbidity, colloidal matter and suspended matter. Deposits on the surface of the metallic portion of the sensing electrode can cause the electrode to be unresponsive or exhibit a memory effect. ORP measurements are temperature sensitive as defined by the Nernst Equation. However, the magnitude of the error is small compared to the variability of the system being measured and automatic temperature compensation is not normally employed. ORP measurements are also pH dependent as previously discussed. The pH/ millivolts correlation is approximately 59 mV per standard pH unit at 25°C (77°F). Again it is important to remember that ORP readings are by no means absolute determinations that can be correlated across multiple systems. While a +100mV (Ag/AgCl, 3M KCl) reading may be highly oxidizing in a feedwater system, a cooling tower at lower pH may have very little oxidizing power at this value. For true disinfecting conditions using chlorine, an ORP reading of +400 mV to +600 mV (Ag/AgCl, 3M KCl) may be desired.

6.5 Calibration The instrument used to measure ORP may be calibrated using standard solutions that have known ORP values. For instance, one possible set of standard ORP reference solutions are based on pH 4 and pH 7 buffer solutions, similar to those used for pH meter calibration. However, for ORP standards, quinhydrone (formulated to provide an equimolar solution of quinone and hydroquinone) is added to the buffer solutions, which forms a reversible oxidation-reduction couple when dissolved in water. Hydrogen ions participate in the reaction between the quinone and hydroquinone, Equation 6-11, creating a pH dependent equilibrium: C6 H 4 O 2 + 2H+ + 2e- ↔ C6 H 6 O 2

Quinone

Equation 6-9

Hydroquinone

E = E°-[2.303RT/(2F)] × log a hq /a q − [2.303RT/F] × pH

Equation 6-10

When the activity of quinone (aq) is equal to the activity of hydroquinone (aaq), the second term ( log a hq /a q ) on the right hand side of Eq. 6-10 drops out so that the electrode potential, E, is dependent only on pH. 6-7


The quinhydrone solutions are prepared by dissolving 10 grams of quinhydrone in one liter of pH 4 or pH 7 buffer solution [5]. Quinhydrone is not very soluble, so only a small amount will dissolve in the buffer solution, changing it to an amber color. However, it is important that excess quinhydrone is used so that solid crystals are always present. The quinhydrone powder poses a moderate health risk, causing irritation of the lungs with prolonged exposure to the dust. The calibration solutions are fairly innocuous unless ingested in large amounts. These hazards are minimized or can be avoided by using safe handling practices. Refer to the relevant MSDS for appropriate information. For the pH 4 and 7 quinhydrone solutions, ORP values of about 263 mV and 86 mV, respectively, can be expected at 25°C (77°F) when measured against a Ag/AgCl, Sat. KCl reference electrode [2,5,6]. These values are consistent with a change of 59.0 mV per pH unit, which are consistent with the calculated value of 2.303 RT/F (= 59.2 mV at 25°C (77°F)) in Eq. 6-10. ORP values for the standard solutions at other temperatures and for other reference electrodes are provided in ASTM D1498-00 [2] and are reproduced in Table 6-2. Table 6-2 Expected ORP Values for Reference Quinhydrone Solutions at pH 4 and pH 7 ORP Value (mV) Reference Electrode*

pH 4 Buffer Solution

pH 7 Buffer Solution

20°C (68°F)

25°C (77°F)

30°C (86°F)

20°C (68°F)

25°C (77°F)

30°C (86°F)

Ag/AgCl

268

263

258

92

86

79

SCE

223

218

213

47

41

34

SHE

470

462

454

295

285

275

* Ag/AgCl = silver/silver chloride (Sat. KCl); SCE = saturated calomel electrode; SHE = standard hydrogen electrode.

The quinhydrone standards are easily made but they are stable due to reactions with air, primarily oxygen contamination, for only about eight hours and are not expected to yield highly reproducible ORP values. Nevertheless, ORP values in freshly prepared solutions are expected to be within 10 mV of the values listed in Table 6-2; and an ORP value measured with a probe in one of these standards can usually be considered acceptable if it lies within 30 mV of the listed value. If the electrode does not respond as expected or has a slow response, it should be cleaned and the calibration procedure repeated. Removal of oily or organic deposits can be achieved with a detergent or, if necessary, methanol or isopropyl alcohol. For removal of contaminants or mineral deposits, the electrode should be soaked in 10% nitric acid for 10 minutes. Alternatively, ASTM D1498-00 suggests the use of warm (70°C) aqua regia for about 1 minute [2]. As a last resort, the platinum surface can be polished with a 600 grit wet-dry emery cloth or a 1-3 micron alumina polishing powder to remove any stubborn coatings or particulates. The electrode will need to soak in purified water for at least 30 minutes after performing any of these cleaning procedures.

6-8


A single point calibration (or “standardization”) is often considered adequate. ORP meters, unlike pH meters, do not have “slope calibration” to allow adjustment of the response to a specified change in ORP. Nevertheless, it is usually good practice to verify that the ORP probe (platinum/reference electrode combination) is operating in a predictable fashion by measuring the ORP in a second standard solution. If the electrode potential does not change by the expected amount, the electrode should be cleaned and re-calibrated, as described above. Other ORP calibration solutions are available based on the ferrous/ferric cyanide or ferrous/ferric sulfate equilibrium reactions, but the quinhydrone standards are more widely used. 6.5.1 Calibration Checks

On-line ORP instruments should be checked periodically to demonstrate calibration stability. The Line Method [7] is appropriate for verifying instrument stability. Here, a calibrated separate ORP monitor is used to analyze the same sample steam as the installed on-line instrument. The two results are compared to the acceptance criteria (e.g., agree within ± 3 sigma or ± 10%). Provided the on-line analyzer agrees within the acceptance criteria, the on-line instrument’s calibration is considered to be acceptable. If the results are outside the acceptance criteria the on-line instrument must be recalibrated.

6.6 End User Considerations Manufacturers of on-line pH monitors typically include the ability to monitor ORP in their instrument design. In fact, the two measurements are closely related and only differ in the design of the sensing electrode and manner in which the mV output signal is manipulated. For the most accurate ORP determinations (±5 mV precision), samples should be maintained at 25 ±1 °C (77 ±2 °F) to preclude ion activity discrepancies [2]. ORP sensing probes in power plant feedwater systems do not suffer from foreign material buildup. In cooling water or service water applications, ORP sensing probes are susceptible to foreign material plating the noble metal surface. Some of these applications require almost daily electrode cleaning, stabilization, and calibration verification. Systems are presented by several manufacturers that provide in-situ on-line cleaning of the probe. Calibration is performed with one or more pH buffers to which has been added a quinhydrone compound (see section 6.4). Most calibrations are only one point (the mV reading of a pH 4 buffer with added quinhydrone) adjustments but making a cross check with a second buffer is advisable. The span value of the output signal is also a user defined variable with many of the instruments. Full scale deflections are possible with as little as 100mV and can be ranged up to 1000mV. ORP readings are empirical at best but satisfactory for their intended use in fossil feedwater cycles. Absolute values are not transferable from one system to another and may undergo changes with various water quality parameters.

6-9


6.7 References 1. Cycle Chemistry Guidelines for Fossil Plants: All-volatile Treatment, Revision 1. EPRI, Palo Alto, CA: 2002. 1004187. 2. ASTM D 1498-00; “Standard Practice for Oxidation-Reduction Potential in Water”. ASTM International West Conshohocken, PA 19426; 2006. 3. Barry Dooley, Digby Macdonald, and Barry C. Syrett, “ORP—The Real Story for Fossil Power Plants”, Power Plant Chemistry, 2003, Volume 5(1). 4. ASTM G5-87, “Standard Reference Method for Making Potentiostatic and Potentiodynamic Anodic Polarization Measurements”, ASTM, Philadelphia, PA. 1987. 5. L. McPherson, Chemical Engineering, p. 143-145, March 1994. 6. S. Filer, A.S. Tenney III, D. Murray and S.J. Shulder, “Power Plant ORP Measurements in High Purity Water”, NUS International Chemistry On-Line Process Instrumentation Seminar, Clearwater Beach, FL. November 1997. 7. ASTM D3864-96(2000), “Standard Guide for Continual On-Line Monitoring Systems for Water Analysis, American Society for Testing & Materials”. American Society for Testing and Materials, Philadelphia, PA.

6-10


7 pH

7.1 Purpose and Use In boiler water, pH is an EPRI Core Monitoring Parameter [1-4]. As such, pH should continuously be monitored on-line to check the acceptability of the water chemistry, thereby ensuring that corrosion rates are kept at low levels. In fossil plant steam-water cycles pH may also be monitored for one or more of the following reasons:

•

To facilitate the correlation between two or more water chemistry parameter (e.g., pH, conductivity, ammonia correlation).

•

To provide a feedback signal for automated process control.

•

To warn of in-leakage of contaminants.

•


•

To troubleshoot or verify the accuracy of other on-line pH monitors.

•

To check the pH of water streams not routinely monitored continuously by on-line monitors.

The data generated by continuous on-line monitoring of pH is used by plant chemistry and operations personnel. The goal for plant personnel is to maintain pH within prescribed limits. The pH method and instrument described below is substantially different from the pH method and instrument used for environmental or other high ionic strength solutions. Although the technical understanding for the measurement of pH is essentially the same, other considerations for these high ionic strength samples are significantly different. For a discussion on environmental pH analysis refer to APHA Standard Methods for Examination of Water and + Wastewater, Part 4500-H [5].

7.2 Description The term pH is a measure of the acidity (or alkalinity) of an aqueous fluid. The factor that most influences the level of acidity is the activity of the hydrogen (H+) ions in the fluid. By definition, pH is the logarithm of the reciprocal of the H+ activity:

7-1

EPRI Proprietary Licensed Material pH

pH = log10 [1/H + ] = - log10 H +

Equation 7-1

In weak solutions typical for power plant on-line instrumentation applications, H+ activity is approximately equal to H+ concentration in moles/liter. The pH is obtained by measuring the potential difference between a pH-sensitive electrode and a reference electrode, both of which are immersed in the solution of interest. Measurement of pH is accomplished by means of an electrode that develops an electrochemical + potential directly related to the H activity of the solution in which the electrode is immersed. The purpose of the pH-sensing electrode is to provide a varying electrochemical signal corresponding to the H+ ion concentration in the solution being tested. Although it is not feasible to measure the electrochemical potential directly, it is possible to measure differences in potential with a voltmeter or electrometer. Consequently, a reference electrode is needed in addition to the pH-sensitive electrode to complete the circuit. The purpose of the reference electrode is to provide a constant electrochemical reference potential (a baseline) against which the potential of the pH-sensing electrode can be compared. The reference electrode + potential is not affected by H ion concentration. Therefore, when pH instrumentation measures the potential difference between the pH-sensitive + electrode and the reference electrode, the potential difference can be directly related to the H ion activity at the pH electrode surface. Instrument manufacturers supply the pH electrodes and reference electrodes separately or in a single combined unit. For low-conductivity pH applications separate electrodes provide some technical advantages discussed below. For routine pH applications combination electrodes are adequate.

7.3 Technical Considerations Water samples taken from the steam/water cycle typically have low ionic strength (i.e., 0.1 to 100 µS/cm), leading to several analytical challenges. These challenges can be overcome provided appropriate pH sensing and reference electrodes are selected; appropriate temperature compensation is applied; interfering ions are minimized; interfering stray current is minimized; and, appropriate calibration and calibration checks are performed. Without appropriate consideration of these factors it is not unusual for the pH electrode response to be slow, “noisy”, and non-reproducible under such circumstances. 7.3.1 Sensing (Glass) Electrode

For power plant cycle chemistry pH measurements, the sensing electrode is normally made from a special glass, fashioned into a thin-walled, small bulb, and fitted with a suitable electrical contact on its inner surface, the so-called ‘glass electrode’. The electrical contact takes the form 7-2


of a reference half cell, normally a silver wire partially coated with (or chemically converted to) silver chloride (Ag/AgCl), which dips into a pH-buffered solution held within the glass bulb. The external surface of the glass bulb is immersed in the sample water being measured. 7.3.2 Reference Electrode

While the actual potential measured will depend on the specific type of reference electrode used, the change in potential resulting from a known change in pH will be independent of the reference electrode. For instance, a change of 1 pH unit at 25°C (77°F) will produce a change in potential difference of 59.16 mV. The pH monitoring instrumentation is simply a voltmeter, more specifically a high impedance volt meter, calibrated to read pH in specific units (SU), instead of potential in millivolts (mV). Variations in reference electrode potentials are simply adjusted by an electronic zero adjustment. NOTE: By convention, the instrument is adjusted so that electronic zero (zero mV; also called the isopotential point) represents the mid-point of the pH scale; pH 7. For pH measurements to be accurate the reference electrode must provide a stable reference point against which the pH-sensitive electrode potential is measured and there must be electrolytic contact between these two electrodes. Reference electrodes are often composed of a silver (Ag) wire partially coated with silver chloride (AgCl, thus creating a Ag/AgCl junction) surrounded by a saturated potassium chloride (KCl) solution contained in a non-conductive glass tube. Electrolytic contact between the KCl solution on the inside of the glass tube and the water being measured on the outside of the tube is often maintained by means of a small porous plug or frit that passes through the tube wall. + A chemical potential gradient promotes migration of KCl solution (K and Cl ions) through the frit. As a result, the KCl solution becomes diluted over time, leading to a change in the electrochemical potential of the reference electrode and—without replacement of the saturated KCl solution or re-adjustment of the voltmeter zero setting—consequently leading to an inaccurate pH measurement. Dilution may also occur when the solution being monitored flows through the frit into the saturated KCl solution because of a positive pressure gradient through the frit. Similarly, a negative pressure gradient can drive the KCl solution into the solution being monitored. To prevent this type of dilution or loss of KCl solution, the electrodes must be placed in a vented sample flow-through sample chamber. Additionally, the reference electrode fill solution (saturated KCl) should be topped off routinely to maintain a level nearly full. The presence of a few KCl crystals at the bottom of the reference electrode tube helps ensure that the solution is always saturated. (Note: Consult the reference electrode manufacturer’s manual for specific instructions.)

For pH measurements of low ionic strength solutions (i.e., specific conductivity <10 µS/cm and pH~7) additional reference electrode considerations are necessary. The uncertainty of the liquid junction potential of the reference electrode is the source of the largest pH measurement error in water with low ionic strength. The liquid junction potential stems from the difference in the ionic mobility of the potassium ion and the faster chloride ion. This leads to a charge separation. 7-3


This effect becomes more pronounced in diluted aqueous solutions. The liquid junction potential amounts to roughly 1 mV per decade in concentration difference between reference and measuring solutions. The liquid junction potential in pH measurements of low ionic strength water will always lower the corrected pH value. The largest measurement errors are produced by the variable liquid junction potential of diffusion driven gel electrodes because of the poor connection between the reference and measuring solutions. The best results are provided by freely flowing liquid reference junctions. The flowing junction reference electrode is fitted with a reservoir filled with saturated KCl. The reservoir is maintained at a level above the level of the sample container thereby providing a positive head pressure assuring flow (flowing junction reference electrode) of saturated KCl solution through the frit. The saturated KCl solution reservoir must be continually refreshed to makeup for this loss. Figure 7-1 shows a flowing junction reference electrode head cup configuration. For higher ionic strength solutions (specific conductivity > 10 µS/cm, as is the case in some boiler blowdown samples) a separate flowing junction reference electrode is not necessary since + the ionic strength in the solution is adequate to provide H ion mobility on its own. A typical combination electrode is shown in Figure 7-2. Combination electrodes incorporate the sensing and reference electrodes into one electrode assembly.

7-4


Figure 7-1 Flowing Junction Reference Electrode Head Cup Configuration Source: Adapted from a Drawing Provided Courtesy of Honeywell Corporation

7.3.3 Temperature Effects

The potential difference measured between the pH sensing and reference electrodes is also influenced by temperature. As a consequence, the calibration of the instrument (the voltmeter) is affected by the sample temperature. The sensitivity to temperature results from electrode effects and solution additive (composition) effects. It is easy to compensate for the first effect and relatively easy to compensate for the second effects provided the solution additives are known.

7-5


Figure 7-2 Combination Electrode

7.3.3.1 Electrode Effects

One effect relates to the dependency of the glass electrode and reference electrode potentials to temperature: the potential difference changes according to the Nernst equation by 2.303 RT/F for each pH unit of change, where R is the gas constant, F is the Faraday constant, and T is the temperature in degrees Kelvin. For instance, a change of 1 pH unit at 0°C (32°F) will cause a change of 54.20 mV, whereas at 25°C (77°F) the change will be 59.16 mV, and at 100°C (212°F) the change will be 74.05 mV. Many modern pH monitoring systems measure the water sample temperature along with the pH and automatically compensate for the temperature. For on-line monitoring, the pH-sensitive electrode, the reference electrode, and the temperature sensor are often incorporated in a flow-through cell through which the water sample flows continuously or intermittently. In other systems, the temperature compensation must be performed manually, or the pH value must be corrected using calibration charts provided with the instrumentation. The temperature compensation only corrects for variations in the electrode slope factor of the Nernst equation; it does not account for the effect on the pH value to 25°C (77°F) of the temperature-dependent dissociation of chemical species in the water sample. This requires an additional compensation.

7-6


7.3.3.2 Solution Additive Effects

The dissociation of pure water itself is dependent on temperature, such that the pH changes from 7.47 at 0°C (32°F) to 6.12 at 100°C (212°F). The dissociation of other chemical additives in the water is also dependent on temperature. Consequently, it is essential that the temperature of measurement be reported along with the pH value. Many utilities find the best solution to this problem is to ensure that the water sample contacting the glass and reference electrodes is maintained at 25°C (77°F) at all times. Otherwise the composition of the solution being measured should be used to determine the appropriate correction factor. The mathematical calculation for determining the corrected pH to 25°C is given in equation 7-2. pH@25C = pHm + (T-25) Correction Factor

Equation 7-2

Where: pH@25C =

the pH corrected to 25°C

pHm

=

the pH measured

T

=

the temperature of the pH sample being analyzed

=

the appropriate correction factor from Table 7-1

Correction Factor

Table 7-1 shows typical temperature correction factors used in power plant applications [9]. Table 7-1 Various Solution Additive Temperature Correction Factors for Power Plant Steam/Water Cycle pH Measurements Solution Additive

System Monitored

Correction factor per degree centigrade different from 25°

Ammonia

Condensate and feedwater (or boiler on AVT)

0.0323 SU

Strong bases

Boiler chemistry (PO4, NaOH treatment)

0.0332 SU

None

Ion exchange or water treatment effluent

0.0165 SU

Example:

Assume: An all volatile treatment (AVT) plant with ammonia as the primary pH additive Observed pH = 9.55 Observed temperature = 22°C Correction factor for ammonia = 0.0323 SU degree centigrade different from 25° 7-7


Solution: From Equation 7-2: pH@25 = pHm + (T-25) Correction Factor pH@25 = 9.55 + (22-25) 0.0323 pH@25 = 9.55 + (-3) 0.0323 pH@25 = 9.55 + (-0.969) pH@25 = 9.46

7.4 Interferences 7.4.1 Interfering Ions

Other errors may occur in the pH reading under some circumstances. For instance, at high pH values where the H+ ion concentration is very low, other positively charged ions may greatly outnumber the H+ ions and may have a significant effect on the potential of the glass electrode. A prime example is the sodium (Na+) ion error which is evident at pH values above about 10 if the Na concentration is greater than about 10 gram/liter or 10,000 mg/L (ppm). However, this level of Na+ concentration is not typically encountered in power plant cycle chemistry. None the less low sodium error electrodes are recommended for measurements over pH 10. 7.4.2 Interfering Stray Currents

The high resistance of the sensing electrode can also cause problems, particularly at low temperatures (below 10°C or 50°F) where the resistance is even greater. The high resistance can stimulate stray currents (AC pick-up) in the circuit unless the integrity of the insulation is rigorously maintained. Essentially no current can flow between the glass electrode and reference electrode during measurement if the pH value is to be measured accurately. This requirement dictates the use of a high impedance (perhaps >10,000 megohm) voltmeter and the need to minimize the length of the leads from the electrodes to the instrument. Low impedance outputs, available on many instruments, can be used to transmit the pH data over great distances without suffering degradation or interference from external voltage sources. Flow sample chambers provide a mounting point for the sensing and reference electrodes. The flow chambers for high purity pH measurement electrodes should be designed to prevent stray current that may affect pH measurements. Stray current may be created when high purity (low conductivity) water flows past glass electrodes, thereby creating a static-like charge. One strategy to avoid stray current interference is to manufacture the flow chamber from a conductive 7-8


material, such as 316L stainless steel, and to provide appropriate electrical grounding for the instrument. Electrical charges may also be developed when the electrodes are wiped with an absorbent paper towel. Electrodes should always be blotted not wiped to prevent this charge buildup.

7.5 Calibration The pH instrument may be calibrated by immersing the sensing and reference electrodes first in one buffer solution of known pH for standardization (zero intercept; neutral pH) adjustment shown in Figure 7-3 of known pH, then in another buffer of different pH to set the instruments slope (span) shown in Figure 7-4. For instance, buffer solutions of pH 7 and pH 10 may be used to calibrate the instrument in preparation for measuring the pH of solutions expected to be alkaline (basic), as in most power plant operations. The instrument uses these accurately known buffer values to set the slope and intercept of the electronically stored “curve” of pH versus potential. In principle the pH of the buffers used should span, or be close to, the pH of the solution being monitored. It is generally undesirable to standardize with pH 7 (isopotential point) and pH 4 (slope / span) buffers for power plant process samples since most samples are in the alkaline range. Furthermore, standardization in conventional buffers of relatively high ionic strength will increase the time for subsequent stabilization in a low ionic strength solution; and the possibility of sample contamination by the buffer will also be increased. Therefore, low ionic strength buffers are recommended for use with low conductivity water (i.e., specific conductivity < 10 µS/cm) samples.

Figure 7-3 Standardization (Zero Intercept) Adjustment

7-9


7.6 Calibration Checks On-line pH instrument analytical capabilities should be checked periodically to demonstrate calibration stability. Two methods exist for verifying instrument stability; the Standard Injection Method [6] or the Line Method [7]. For the Standard Injection Method, a buffer of known pH, near the expected pH of the process being monitored, is analyzed by the on-line instrument and the results are compared to the acceptance criteria (e.g., agree within ± 3 sigma or ± 0.1 pH unit). Acceptance criteria are either established based on statistically derived limits. i.e., ± 3 sigma [6] or based on some predetermined limits established from experience, i.e., ± 0.1 pH unit. Provided the on-line analyzer agrees within the acceptance criteria the on-line instrument’s calibration is deemed not to have changed. If the results are outside the acceptance criteria the on-line instrument must be recalibrated.

Figure 7-4 Slope (Span) Adjustment

For the Line Method a calibrated separate pH meter, typically a portable pH meter, is used to analyze the same sample stream as the installed on-line instrument. Provided the on-line analyzer agrees within the acceptance criteria (e.g., matches the results of the on-line analyzer within ± 3 sigma or ± 0.1 pH unit) the on-line instruments calibration is deemed not to have changed. If the results are outside the acceptance criteria the on-line instrument must be recalibrated. Additional technical discussions are presented in ASTM D5128-90 (1995) ε1, Standard Test Method for On-Line pH Measurement of Water of Low Conductivity [9].

7-10


Additional maintenance and calibration activities for pH instruments are described in EPRI Report GS-7556 [8].

7.7 Alternative Methods for Determining pH Although the preferred method for monitoring pH in an on-line application is by direct measurement as described above there are alternative methods for determining pH in low ionic strength solutions. Specifically, if the amount of contaminants is low and the amount of the additive affecting pH is known, then the pH may be estimated. There are at least two techniques for this: determining pH from known ammonia concentration and specific conductivity, and determining pH from the calculated difference between specific conductivity and cation (acid) conductivity. Determining pH from a known ammonia concentration and specific conductivity can be accomplished by using a curves developed for these parameters as shown in Figure 7-5. This calculation is only valid in the absence of appreciable carbon dioxide. Example:

Assume: Specific conductivity at 25°C = 8 µS/cm Ammonia concentration = 1.5 mg/L (ppm) So: pH at 25°C = 9.50

7-11


Figure 7-5 Specific Conductivity, Ammonia, pH at 25°C [1] Source: Reference 1

Carbon dioxide is often present in samples due to some air in-leakage. Determining pH from a known ammonia concentration, carbon dioxide and specific conductivity can be accomplished by using a family of curves developed for these parameters as shown in Figure 7-6.

7-12


10 9 8 7 6 5 4

Ammonia (ppm NH3)

3 2

1 0.9 0.8 0.7 0.6 0.5

CO2 (ppm) 4.00 3.50 3.00 2.50 2.00 1.50 1.00 0.50 0.20 0.10 0.05 0.00

0.4 0.3 0.2

0.1 8.8

9.0

9.2

9.4

9.6

9.8

10.0

pH at 25 °C Figure 7-6 Ammonia Concentration vs. pH for Various Carbon Dioxide Concentrations [1] Source: Reference 1

Some instrument suppliers provide capability to calculate pH from specific and cation conductivity. This can be done relatively simply provided several assumptions are met. Determining pH from the calculated difference between specific conductivity and cation conductivity requires the following assumptions are met:

•

Two channel conductivity instrument is available providing specific conductivity and cation conductivity.

•

Only one pH adjusting chemical additive is used.

•

The correct non-linear temperature compensation for the corresponding additive affecting pH (ammonia, ethanolamine, morpholine, etc.).

•

No phosphates are present in the sample stream.

•

At pH < 8, impurities (such as NaCl) must be small compared to the additive affecting pH.

•

Optimally the pH range of the sample of interest should be between 7.5–10.5. 7-13


Provided these assumptions are met the calculated pH is remarkably close to the observed pH. An example in Table 7-2 shows very good agreement [10]. Table 7-2 Example Calculated pH by Differential Conductivity from One Instrument Supplier [10] Impurities

1000 µg/L (ppb) Na+, (4.35 x 10-5 M/L)

pH additive (Ammonium Hydroxide)

3.5 x 10-5 M/L

Specific conductivity

12.9 µS/cm

Cation conductivity

18.3 µS/cm

Theoretical pH

9.5

Calculated pH by differential conductivity

9.47

7.8 End User Considerations The performance characteristics (range of measurement, accuracy, precision, bias, drift, response time, and signal change) for the monitoring equipment, provided by the manufacturer or supplier should be considered when selecting a suitable on-line pH instrument. In general, manufacturers determine these characteristics using their own in-house methods. End users should select the instrument that is best suited for the intended analytical application. End users should also verify that the instrument meets the requirements for its intended purpose and use (i.e., it demonstrates the required level of accuracy and precision when installed and functioning at your facility). Other on-line pH instrument considerations for low ionic strength (<10 µS/cm, specific conductivity) include:

•

Sensing electrode should be made from high resistance, low sodium glass for pH values greater than 10.

•

Flowing junction reference electrodes should be considered for measuring pH of low ionic strength (<10 µS/cm, specific conductivity) water samples.

•

Electrically grounded conductive flowing sample chamber should be considered when stray currents are a problem.

•

The instrument should be capable of applying sample specific temperature compensation.

•

Capable of analyzing a “grab sample” for periodic buffer check verification or analysis of grab samples.

•

Interfaces and diagnostic capabilities of the instrument should match specific needs.

•

Output type compatible with data acquisition needs.

7-14


7.9 References 1.


2.


3.


4.


5.

Standard Methods for the Examination of Water and Wastewater, Part 4500-H+, pH Value, American Public Health Association, Washington, DC.

6.

Advanced Power Plant Chemistry QA/QC Practices, Scientech, LLC., Clearwater, FL, 2006.

7.

ASTM D3864-96(2000), Standard Guide for Continual On-Line Monitoring Systems for Water Analysis, American Society for Testing & Materials, Philadelphia, PA.

8.

Monitoring Cycle Water Chemistry in Fossil Plants, Vol. 3 Project Conclusions and Recommendations, by A.F. Aschoff, D.M. Sopocy, D.T. Eglar, O. Jonas, J.K. Rice, C.C. Stauffer, and W.E. Allmon. EPRI, Palo Alto, CA: October 1991. GS-7556, Volume 3.

9.

ASTM D5128-90 (1995) ε1, Standard Test Method for On-Line pH Measurement of Water of Low Conductivity. American Society for Testing & Materials, Philadelphia, PA.

10. H. Maurer, “On-line pH Measurement by Differential Cation and Specific Conductivity”, Proceedings of the 1997 International Chemistry On-line Process Instrumentation Seminar, Clearwater, FL, 1997.

7-15


8 SODIUM

8.1 Purpose and Use Sodium is an EPRI Core Monitoring Parameter [1-4]. As such, it should be monitored continuously on-line to check the acceptability of water chemistry, thereby ensuring that corrosion rates are kept at low levels. Sodium may also be monitored for one or more of the following reasons:

•


•

To warn of boiler water carryover.

•

To identify cooling water in-leakage at the main steam condenser.

•


The data generated by continuous on-line monitoring of sodium is used by plant chemistry and operations department personnel. The goal for plant personnel is to maintain sodium below prescribed limits.

8.2 Description of Method An ion selective electrode (sometimes called a specific ion electrode) is typically used in conjunction with a reference electrode to continuously monitor sodium concentrations at various locations in the steam/water cycle. Values below 10 µg/L (ppb) are common during normal operation, and changes of 1 to 2 µg/L (ppb) are often considered significant. Consequently, sensitive instrumentation is essential for this application. To allow measurements below 1 µg/L (ppb), the glass used in the ion selective electrode must be extremely insoluble; and the reference electrode, typically silver/silver chloride or silver/silver bromide, is designed and positioned in the flow cell downstream of the sensing electrode to minimize sample contamination from the reference electrodes internal fill solution.

8-1

EPRI Proprietary Licensed Material Sodium

The response of the ion selective electrode is described by the Nernst equation: E = E o + [2.3026 R T / n F] × log10 [(CNa + CB )/ CIso ]

Equation 8-1

Where: E

=

the measured potential of the electrode pair (V)

Eo

=

the measured potential when the iodide concentration in the sample quals CIso (V)

R

=

the ideal gas constant (8.317 joules/°Kxmole)

T

=

temperature of the sample (°K)

n

=

the valence of the ionic species (equals 1 for sodium)

F

=

Faraday’s constant (96,486.7 coulombs/gram-equivalent)

CB =

the detection limit of the system (µg/L (ppb)). This term is used to define the curvature of the electrode response near the detection limit. It depends on pH, temperature and interferences (such as potassium, ammonium, and other ions).

CIso =

the sodium ion concentration (µg/L (ppb)) in the sample that produces an electrode potential that is independent of temperature; the isopotential point. This is the reference point for temperature compensation.

The term, 2.3026RT/nF, in Equation 8-1 is essentially the temperature-dependent slope of the electrode response. Most current models of sodium analyzers monitor the temperature of the sample and automatically compensate for temperature variations. The associated instrumentation converts the potential of the ion selective electrode to a sodium ion concentration and displays this information. The sodium ion electrode is not subject to interference from color, turbidity, colloidal matter, oxidants, or reducing agents. Ion selective electrodes have very linear characteristics over a working range of many decades. Manufacturers of sodium analyzers that incorporate ion selective electrodes have shown that the slope of these electrodes is linear to at least 0.01 µg/L (ppb), and some manufacturers have proposed linearity to 0.005 µg/L (ppb) [5]. Because of this linearity, calibrations at concentrations 100 or 1,000 or 10,000 times the routine measurement range can provide accurate measurements at lower routine concentrations. An alternative approach is to dilute the standards mechanically within the instrument. Data from one instrument supplier [6] shows an accuracy of 90% recovery at 0.01 µg/L (ppb) from a calibration by double known addition at 10 and 100 µg/L (ppb). Figure 8-1 plots the concentration of sodium added to the sample against the concentration measured by the instrument. There is a linear relationship between these values correct to within 0.01 µg/L (ppb).

8-2


Figure 8-1 Measured Sodium Concentration vs. Concentration of Sodium Added [5] Source: Adapted from Reference 5, Courtesy Swan Instruments

Typically, the water sample is filtered, its pressure is regulated, and its flow rate and temperature are controlled. The sodium selective electrode and reference electrode are positioned in a flow cell through which the sample passes. The reference electrode is always positioned downstream of the sensing electrode to avoid possible sample contamination from the reference electrode filling solution. The potential difference, in millivolts (mV) between these electrodes responds logarithmically to changes in the dissolved sodium concentration, as indicated in Equation 8-1. Since sodium electrodes are sensitive to hydrogen ions, determining sodium concentrations <10 µg/L (ppb) requires the potential interference from hydrogen ions (H+) be eliminated by increasing the pH to 10.5 or higher so that the hydrogen ion concentration is at least three orders of magnitude lower than the sodium concentration. pH adjustment is typically performed by introducing diisopropylamine (DIPA), dimethylamine (DMA) or monoethylamine (MEA) into the sample stream via diffusion. Some instrument suppliers recommend a pH of 12 for sodium concentration below 1 µg/L (ppb). Typically DIPA is used to achieve this higher pH. Alternatively, for determining sodium concentration above 10 µg/L (ppb), a pH of 9.0 or higher is required. In this case a buffer solution containing ammonium hydroxide or morpholine may be introduced into the sample upstream of the sensing electrode or if the sample stream itself has a pH >9.0 no external pH adjustment may be required.

8-3


Other positively charged monovalent cations (silver, lithium, potassium, etc.) can potentially cause interference but, in the high purity fluids of interest in the steam/water cycle, the risk of such interferences is very low. Sodium sensing and reference electrodes undergo changes over their useful lives. Therefore periodic re-calibration is necessary. Calibrations are performed by the successive replacement of the water sample with two standards at known concentrations. The first standard is prepared at a high enough concentration (> 100 µg/L (ppb)) to avoid standard contamination leading to false calibrations. A second standard is introduced at a concentration ten times the first standard’s concentration and the instrument determines the calibration slope for that sensing electrode. Alternatively, dilute standards are created within the instrument by pipetting concentrated standards into a flow cell. The flow cell dilutes the concentrated standards to concentrations closer to the sample stream concentration for calibration. Specifically, standards containing 19.1 mg/L (ppm) and 192 mg/L (ppm) sodium are pipetted into the flow cell and are diluted by the instrument to 20 µg/L (ppb) and 200 µg/L (ppb) respectively. This “double known addition” method of calibration provides accurate, reproducible results. Zero intercept is also determined during the calibration process. Sodium sensing electrodes also undergo continuous sodium leaching in water resulting in a thick leached layer and an associated slow response to a step concentration change. The leached layer can be readily etched away with dilute HF. Etching solutions, used as directed, will not harm electrodes or adversely affect calibration values.

8.3 Technical Considerations 8.3.1 Sensing Electrode

Sensing electrodes are constructed of special glass. Two glass formulas are used: 1.

Pure sodium aluminosilicates. This formula is best suited for sodium measurements < 10 µg/L (ppb). It is difficult to work with in manufacturing due to its very high melting point, but has very good analytical performance. This formulation has very good Nernstian response and does not need to see high sodium concentration in solution to maintain its response characteristics. Sodium aluminosilicates can also be cleaned and rejuvenated easily by etching exposed surfaces in HF.

2.

Sodium/lithium aluminosilicates. This formula is best suited for sodium measurements >10 µg/L (ppb). It is easier for manufacturing the electrode, but needs regular exposure to sodium in solution or it becomes “super-Nernstian,” resulting in erroneously low sodium results at low (<0.1 µg/L (ppb)) concentrations. Lithium glass response can be improved over the short term by conditioning in reasonably concentrated (several mg/L (ppm) NaCl solutions.

8-4


Sensing electrode response time increases over time due to the nature of the operating environment but periodic cleaning minimizes this effect. Sodium sensing electrodes undergo continuous leaching in water. In a layer near the glass/solution interface, sodium ions leach into the solution and are replaced by hydrogen ions, changing the glass surface structure. As more sodium ions are leached into the solution, the layer of leached/ sodium ions gets thicker. The thicker the leached layer the slower the sensing electrode responds to changes in sodium ion concentration. This phenomenon does not influence the stability or correctness of the reading. It only affects the response time to step concentration changes. The leached layer can be readily etched away with dilute HF. Etching solutions, used as directed, will not harm electrodes or affect calibration values. Etching is always recommended before calibration to speed up calibration equilibration times.

Figure 8-2 Response Time of Sodium Ion Selective Sensors: Time vs. Sodium Concentration in µg/L (ppb) Before and After Etching [5] Source: Adapted from Reference 5, Courtesy Swan Instruments

Figure 8-2 [5] shows the response time before etching and after etching. Before etching the response time to obtain stable readings was observed to be approximately 45 minutes. After etching the response time to obtain stable readings was observed to be approximately 2 minutes.

8-5


Figure 8-3 [6] shows similar results from an installed sodium analyzer on a full-flow condensate polisher effluent sample line. The sample was spiked with 20 µg/L (ppb) sodium and response time was observed at various intervals. Notice how in this test the elapsed time to reach the spiked value was about two minutes immediately after etching and about ten minutes after 13 days of continuous use following etching. Figure 8-4 [6] shows similar results from an installed sodium analyzer on a full-flow condensate polisher effluent sample line. Again the sample was spiked with 20 µg/L (ppb) and response time was observed at various intervals. Notice how in this test the elapsed time to reach the spikes value is about 5 minutes immediately after etching and about 35 minutes after 7 days.

Figure 8-3 Elapsed Time After Known Standard Addition: Time vs. Sodium Concentration in µg/L (ppb) [6]

8-6


Figure 8-4 Sodium Analyzer Response: Elapsed Time After Addition of 20µg/L (ppb) Sodium vs. Sodium Concentration [6]

One etching solution utilizes dilute NaF, adjusted to a pH that releases a modest activity (<5% by volume) of HF. A 1-minute immersion in this solution, once per month, maintains optimal performance without deterioration of electrode life. For lithium aluminosilicate sensing electrodes, a similar improvement in response time can be obtained by exposing the sensing electrode to relatively high sodium concentrations (i.e., several mg/L (ppm) NaCl). 8.3.2 Reference Electrode

While the actual potential measured will depend on the specific type of reference electrode used, the change in potential resulting from a known change in sodium concentration will be independent of the reference electrode. For instance, a 10 fold increase of sodium concentration at 25°C (77°F) will produce a change in potential difference of 59.16 mV. The sodium monitoring instrumentation is simply a voltmeter, more specifically a high impedance voltmeter, calibrated to read sodium concentration units (µg/L (ppb)), instead of potential in millivolts (mV). For sodium measurements to be accurate, the reference electrode must provide a stable reference point against which the sodium concentration is measured and provide a salt bridge for electrolytic contact with the solution being measured. Reference electrodes are typically composed of a silver/silver chloride electrode (Ag/AgCl) surrounded by a saturated potassium chloride (KCl) solution contained in a non-conductive glass tube fitted with a porous plug (frit). This porous plug maintains electrolytic contact between the solution being measured and the Ag/AgCl reference electrode. 8-7


KCl solution (containing dissolved ions) migrates through the frit to assure continuous electrolytic contact with the solution being measured. However, if the KCl solution becomes depleted or diluted, sodium measurements may become sluggish or inaccurate. Depletion occurs by normal migration of KCl through the frit or by evaporation. Periodic refilling of reference fill solution is necessary to ensure continued operation. Dilution occurs when the solution being monitored backflows through the frit into the saturated KCl solution. To prevent backflow dilution, the electrodes must be placed in a vented sample flow-through sample chamber. Additionally the reference electrode fill solution (saturated KCl) should be topped off routinely to maintain a level nearly full. The reference electrode manufacturer’s manual should be consulted for specific instructions.

8.4 Interferences Hydrogen ions (H+) are the major interfering species commonly encountered in power plant sodium measurements. Silver, lithium and potassium also interfere but they are not present in high enough quantities in power plant sample streams to cause a significant interference. For + sodium measurements < 10 µg/L (ppb), H ion interference is minimized by adjusting pH to >10.5 by introducing DIPA, DMA or MEA pH adjuster into the sample stream by diffusion through permeable tubing. The sample stream passes through a permeable tube that is enclosed in a container of pH adjuster. The pH adjuster permeates the tube and interacts with the sample stream to elevate the pH. The length of the diffusion tubing is adjusted to yield the desired pH of the sample stream exiting the closed container of pH adjuster. These pH adjusters have a strong impact on pH and can be used to easily achieve pH of >10.5. Figure 8-5 [5] shows that the slope of the calibration line deviates from Nernstian response for pH < 9 for Sodium values between 10 and 100 µg/L (ppb). Therefore, higher pH is needed to get Nernstian response at sodium concentrations <10 µg/L (ppb). DIPA is recommended as the pH adjuster for detecting low sodium concentrations (< 10 µg/L (ppb) sodium). In contrast, for determining sodium concentration >10 µg/L (ppb), a pH of 9.0 or higher is acceptable. In this case a buffer solution containing ammonium hydroxide or morpholine may be introduced into the sample upstream of the sensing electrodes or if the sample stream itself has a pH >9.0 no external pH adjustment may be required. The sensing electrode is made of glass containing low concentrations of sodium. At extremely low sample concentrations of sodium, the dissolution of the electrode glass, containing sodium affects the instrument readout. At about 5 nanograms/L (ppt) the glass electrode establishes an equilibrium with the surrounding sample stream such that sodium ions in the sample stream are balanced with sodium ions being sloughed off the sensing glass. At sample stream concentrations below about 5 nanograms/L (ppt), the equilibrium shifts such that the sensing electrode is supplying sodium ions, by dissolution, to the sample to maintain an equilibrium of about 5 nanograms/L (ppt)) sodium. In effect, the measured sodium concentration will never be <5 nanograms/L (ppt) due to the dissolution of sodium from the sensing electrode.

8-8


Figure 8-5 Calibration Slope (Millivolt Response) of Sodium Ion Selective Electrode at Varying pH Values [5] Source: Adapted from Reference 5, Courtesy Swan Instruments

8.5 Calibration The sensitivity of a sodium analyzer system may change over time due to changes affecting the glass tip. For instance, sample contaminants may deposit on the sensing electrode in the form of surface films or particulates. When this occurs, the system must be recalibrated. Etching is always recommended before calibration to speed up response times. The sensitivity of a sodium analyzer system may also change over time if the reference electrode junction becomes blocked. In such cases, the junction must be unblocked or the reference electrode must be replaced before the system is recalibrated. Calibrations are typically accomplished monthly or weekly depending on sample and probe conditions. Newer probes (< 6 months old) are often stable and may only need calibration monthly. Older probes may need more frequent (i.e., weekly) calibrations. Many on-line sodium monitors allow automatic or semi-automatic two- or three-point calibration in which the sample is replaced successively by two or three standard solutions, with concentrations a decade apart (e.g., 1 and 10 mg/L (ppm), or 200 and 2,000 µg/L (ppb)). This allows the instrument to check and, if necessary, correct the span and calibration. In any case 8-9


externally diluted standards should be > 100 µg/L (ppb) to minimize the potential for standard contamination. It is also necessary to check the flow of the reference probe electrolyte. Unstable signals very often have their origin in the reference system. Figure 8-6 [5] shows the performance of a sodium ion selective electrode following calibration with 200 and 2,000 µg/L (ppb) sodium solutions. The data in the figure show:

•

95.8%recovery at 2 µg/L (ppb),

•

98.7 % recovery at 20 µg/L (ppb), and

•

99.1 % recovery at 200 µg/L (ppb).

Figure 8-6 Verification Results Immediately After a 200 and 2,000 µg/L (ppb) Sodium Calibration [5] Source: Adapted from Reference 5, Courtesy Swan Instruments

A modification of this technique, the double known addition (DKA) method, involves the addition of a known amount of a standard solution to the sample on two successive occasions. The potential, ES, recorded by the instrument at the start of this procedure corresponds to the starting concentration, CS, as shown in equation 8-2. This equation is a simplified version of equation 8-1, in which CB is assumed to be small compared to CS, and 2.3026RT/nF equals a factor, S:

8-10


E = E O + S log (CS / CIso )

Equation 8-2

The addition of the standard increases the concentration by a known amount, dC1, to CS+dC1, and changes the measured potential to E1, as shown in equation 8-3: E1 = E O + S [log (CS + dC1 ) / CIso ]

Equation 8-3

Then, a second addition of a standard solution is made to the sample. This standard solution is preferably about ten times more concentrated than the first, causing a further increase in concentration, dC2, and a new potential, E2, as shown in equation 8-4: E 2 = E O + S [log(CS + dC1 + dC 2 ) / CIso ]

Equation 8-4

The instrumentation automatically solves the three equations (8-2 through 8-4) for the three unknowns, and EO and S are stored for subsequent use in the on-line monitoring mode. Some instruments also correct for the detection limit of the system, CB, shown in equation 8-1. This so-called blank correction improves detection of very low levels of sodium. If there are any questions regarding the effectiveness of the pH adjusting solutions, checking the pH of the sample discharge from the analyzer may be appropriate to ensure that the buffer is elevating the pH as desired. It may be more appropriate to follow the maintenance and calibration activities recommended in EPRI Report GS-7556 [7].

8.6 Calibration Checks On-line sodium instruments should be checked periodically to demonstrate calibration stability. Two methods exist for verifying instrument stability; the Standard Injection Method [8] or the Line Method [9]. For the Standard Injection Method, a known sodium standard, in the concentration range where the sodium calibration can be readily verified (typically 10 µg/L (ppb)), is analyzed by the online instrument and the results are compared to the acceptance criteria (e.g., the results should agree within ± 3 sigma or ± 10%). Provided the on-line analyzer agrees within the acceptance criteria, the on-line instruments calibration is considered still acceptable. If the results are outside the acceptance criteria, the on-line instrument must be recalibrated. One approach is to prepare a standard and introduce this standard to the sodium analyzer replacing normal sample flow for the test period with the standard. Unfortunately, any source of sodium contamination would bias the test results high. A second approach [6] involves injection of a standard as a known addition into the sample stream and the mixture is analyzed by the instrument. The results are compared to the acceptance criteria (e.g., they should agree within ± 3 sigma or ± 10%). Again, any source of 8-11


sodium contamination would bias the test results high. If the results are outside the acceptance criteria the on-line instrument should be recalibrated. Some instruments [5] provide fully automated calibration verification; a standard solution is always available. The standard is diluted down to 10 µg/L (ppb) for automatic verification. As the diluted solution is injected, the sensor will report the increase. If it is within limits of recovery and response time specified by the supplier, the instrument is considered to meet acceptance criteria. One instrument supplier has proposed a calibration check by a known addition method using a standard and dynamic on-line addition to provide a ppt (nanograms/L) level standard solution for low-level verification of on-line analyzers [6]. Assuming the process concentration does not change during the evaluation period, Table 8-1 results can be used to demonstrate the calibration has not changed by following the Standard Injection Method described above. Table 8-1 Typical Results from Known Addition Method for Calibration Check in the nanograms/L (ppt) range Spike (nanograms/L Na (ppt))

Recovery

10 ppt

90%

42 ppt

93%

64 ppt

103%

140 ppt

105%

1,100 ppt

104%

Note: baseline sodium concentration stated to be about 12 ppt.

For the Line Method, a calibrated separate sodium monitor, typically a portable sodium monitor is used to analyze the same sample steam as the installed on-line instrument. The two results are compared to the acceptance criteria (e.g., agree within ± 3 sigma or ± 10%). Provided the online analyzer agrees within the acceptance criteria, the on-line instrument’s calibration is considered to be acceptable. If the results are outside the acceptance criteria the on-line instrument should be recalibrated.

8.7 Alternative Methods Various other methods exist for determining sodium including; Ion Chromatography (IC), Atomic Absorption (AA) spectroscopy [10,11], Inductively Coupled Plasma (ICP), Inductively Coupled Plasma-Mass Spectrometry (ICP-MS), and Graphite Furnace, flame photometry. However, IC is the only suitable alternative method for continuous on-line determination.

8-12


8.8 End User Considerations The performance characteristics (range of measurement, accuracy, precision, bias, drift, response time, and signal change) for the monitoring equipment, provided by the manufacturer or supplier should be considered when selecting a suitable on-line sodium instrument. In general, manufacturers determine these characteristics using their own in-house methods. End users should select the instrument that is best suited for the intended analytical application. Other on-line sodium instrument considerations include:

•

Appropriate pH adjustment for analysis range of interest to minimize H+ ion interference with sodium measurements

•

Ease and robustness of calibration for the intended use

•

Ease and robustness of calibration verification

•

Ease of reestablishing sensing electrode response time by etching

•

Appropriate formula for ion selective sensing electrode glass

•

Appropriate sensing electrode linearity characteristics in the analytical range of interest

•

Reference electrode stability

8.9 References 1.


2.


3.


4.


5.

“On-Line Sodium Monitoring with Ion-Selective Glass Electrodes”, Dr. Peter Wuhrmann, Swan Analytical Instruments, ESKOM International Conference on Process Water Treatment and Power Plant Chemistry, November 1997.

6.

“Theoretical and Practical Aspects of Glass Electrodes in On-Line Applications, Steve West, Thermo-Orion, Scientech Chemistry On-Line Chemistry On-line Process Instrumentation Seminar, Clearwater Florida, November 2000.

8-13


7.

Monitoring Cycle Water Chemistry in Fossil Plants, Vol. 1 Monitoring Results, by A.F. Aschoff, D.M. Sopocy, D.T. Eglar, O. Jonas, J.K. Rice, C.C. Stauffer, and W.E. Allmon. EPRI, Palo Alto, CA: October 1991. GS-7556, Volume 1.

8.

Advanced Power Plant Chemistry QA/QC Practices, Scientech, LLC., Clearwater, FL, 2006.

9.

ASTM D3864-96(2000), “Standard Guide for Continual On-Line Monitoring Systems for Water Analysis”, American Society for Testing & Materials, Philadelphia, PA.

10. ASTM D4191-93, “Standard Test Method for Sodium in Water by Atomic Absorption Spectrophotometry”, 1992 Annual Book of ASTM Standards, Vol. 11.01 Water. American Society for Testing and Materials, Philadelphia, PA. 11. ASTM D6071-96(2000), “Standard Test Method for Low Level Sodium in High Purity Water by Graphite Furnace Atomic Absorption Spectroscopy”, 2000 Annual Book of ASTM Standards, Vol. 11.01 Water. American Society for Testing and Materials, Philadelphia, PA.

8-14


9 AMMONIA

9.1 Purpose and Use Ammonia is not defined as a Core Monitoring Parameter under current EPRI Cycle Chemistry Guidelines [1-4] but is commonly checked, usually as a grab sample at a location such as the economizer inlet. Ammonia is one of several alkaline chemicals that may be added to increase the pH of boiler feedwater in order to control corrosion of carbon-steel and other ferrous alloys in the steam/water cycle. When system metallurgy contains copper alloys, accurate determination of ammonia content is necessary to minimize formation of the copper/amine complex which leads to rapid localized corrosion. Frequent ammonia analyses or on-line ammonia monitoring may be appropriate in these mixed metallurgy systems. The pH of the feedwater in all-ferrous systems depends on the mode of operation. In plants using all volatile treatment (AVT) chemistry control, the pH is typically 9.2-9.6, measured at 25°C (77°F). For units where single phase flow accelerated corrosion is a concern, the feedwater pH is often elevated to a minimum 9.4 with control ranges approaching 9.8 on the top end. For once-through units using oxygenated treatment (OT), the pH is 8.0-8.5, and, for drum units using OT, the pH is in the range 9.0-9.6. For systems with both ferrous and copper alloy components, OT is not recommended, and the pH range for AVT is usually reduced to 8.8-9.3. This lower pH range, corresponding to a lower ammonia concentration, is preferred for mixed metallurgy systems because copper alloys corrode increasingly more rapidly at higher ammonia concentrations, particularly when the oxidation-reduction potential (ORP, Section 6) is oxidizing [1]. Plants where ammonia is continually monitored on-line use the information to:

•

Check the accuracy of water chemistry control, so ensuring that corrosion rates are kept at acceptable low levels.

•

Facilitate the correlation of ammonia with other chemistry parameters (i.e., pH and specific conductivity).

The data generated by continuous on-line monitoring of ammonia is used by plant chemistry and operations personnel. The goal for plant personnel is to maintain ammonia within prescribed limits.

9-1

EPRI Proprietary Licensed Material Ammonia

9.2 Description of Methods Ammonia content can be approximated from Specific Conductivity curves when analyzing high purity water. There are two common methods for on-line determination of ammonia. One method utilizes an ion selective electrode which is sensitive to the ammonium ion concentration in the sample [5]. The second method is colorimetric based on Berthelot’s Phenate Method of detection and quantification [6]. 9.2.1 Background

For accurate ammonia determination, some form of bench analysis or on-line instrumentation is required. On-line analyzers provide frequent and accurate information and can be used at several locations in the steam/water cycle. Some plants measure the ammonia content in the saturated steam. Current EPRI cycle chemistry guidelines recommend monitoring ammonia (when in use) at the economizer inlet. Sampling of saturated steam, however, will aid in determining the amount of ammonia that has volatilized and could be transported with the steam to the condenser. This may be an important consideration in plants with copper alloy condenser tubes. +

When ammonia (NH3) dissolves in water, it equilibrates with hydronium ion (H3O ) or hydroxyl ion (OH-), depending on the pH, to form ammonia (NH3) or ammonium ions (NH4+). The relative amounts of ammonia and ammonium are a function of the solution pH (Figure 9-1). At pH values below 6 (high H3O+ concentrations), the first reaction is promoted and virtually all of the ammonia is converted to ammonium ions (Eq 9-1). At pH values above 12 (high OHconcentration), the second reaction shifts to the left (reactant side) and the ammonium ion concentration is virtually zero (Eq. 9-2). Approximately equal concentrations of ammonia and ammonium ions exist at pH 9.3 at 25 °C (77 °F)

←⎯⎯→ NH4+ + H2O

Equation 9-1

NH3 + H2O+ ←⎯⎯→ NH4+ + OH-

Equation 9-2

NH3 + H3O+

Ammonia may also be formed in the steam/water cycle due to the decomposition of other volatile amines (e.g. morpholine, cyclohexylamine, etc.) and hydrazine (N2H4), a chemical reducing agent frequently used in high-pressure boiler operations. At temperatures between 111270°C (232 and 518°F), hydrazine volatilization increases with only slight chemical breakdown. However, above 270°C (518°F), hydrazine undergoes rapid thermal degradation, defined by the following reaction:

3N 2 H 4 + Heat → N 2 +4NH 3

9-2

Equation 9-3


Figure 9-1 Percent of Ammonia and Ammonium Ions as a Function of Solution pH

9.2.2 Conductivity Approximation

Clearly, the concentration of ammonia in the feedwater is a critical parameter for controlling corrosion in the cycle. If the pH of the water is controlled entirely by the concentration of ammonia present, the concentration can be estimated quite accurately by interpolation from a graph, knowing only the sample pH and conductivity (Figure 9-2). 9.2.2.1 Conductivity Approximation Limitations

The conductivity measurement is also influenced by the presence of other ionic species, and the pH measurement is affected by carbon dioxide and other chemicals that may be present. Consequently, all changes in conductivity or pH are not necessarily indicative of changes in the ammonia concentration. Furthermore, because pH is a logarithmic scale, as the ammonia levels increase to high values, the changes in pH become increasingly small and the sensitivity of measurement decreases. At low levels of ammonia, both pH and conductivity rapidly approach lower limits reflecting the values for the ammonia-free water, and again the sensitivity of measurement decreases. Consequently, at best, accurate estimates can be made in a fairly narrow range of ammonia concentrations using this approach; i.e. between 0.1 and 2.0 mg/L (ppm) ammonia. 9-3


Figure 9-2 Relationship between Ammonia, Specific Conductivity and pH in the Absence of Carbon Dioxide at 25ºC [1] Source: Adapted from Reference 1

9.2.3 Ion Selective Electrode

One type of on-line ammonia monitor utilizes an ion selective electrode for analyte measurement [5,6]. The ion selective electrode detects NH4+ ions, not ammonia, so measurement must be preceded by an adjustment of the sample pH to a value that is low (pH<6) to convert virtually all ammonia to NH4+ ions. This is done by feeding the water sample at a controlled pressure and 9-4


flow rate through a special tube that passes through a reagent diffusion bottle. Acetic acid contained in the bottle passively diffuses through the special tube and reduces the sample pH. The treated sample then flows on to a chamber housing the ion selective electrode, reference electrode, and temperature probe. Here, the measurement of ammonium ion concentration is made before the sample is finally discarded to waste. The potential of the ion selective electrode, E, measured with respect to the reference electrode is described by the Nernst equation, as follows: E = E

O

+ [2.3026 R T / n F] × log [(C 10

NH4

+C )/C ] B

Iso

Equation 9-4

E

=


EO

=

the measured potential when the ammonium concentration in the sample equals CIso (V)

R

=


T

=


n

=

the valence of the ionic species (equals 1 for ammonium)

F

=


CNH4

=

effective ammonium ion concentration, i.e. activity in µg/L (ppb)

CB

=

the detection limit of the system in µg/L (ppb). This term is used to define the curvature of the electrode response near the detection limit. It depends on pH, temperature and interferences (such as potassium, sodium, and other ions).

CIso

=

the ammonium ion concentration in µg/L (ppb) in the sample that produces an electrode potential that is independent of temperature; the isopotential point. This is the reference point for temperature compensation.

The term, R T/n F is essentially the temperature-dependent slope of the electrode response. The analyzer monitors the temperature of the sample and automatically compensates for temperature variations. After calibration of the analyzer, the measured potential (E) is easily converted to the ammonium ion activity (concentration) using the above equation, and displayed as ammonia concentration on a built-in meter. 9.2.4 Colorimetry

On-line ammonia instrumentation is also available utilizing colorimetric analysis [7,8]. The colorimetric instruments are a batch type process based on Berthelot’s Phenate Method [9]. A sample containing ammonia is reacted with sodium hypochlorite resulting in a rapid conversion of the ammonia to monochloramine. 9-5


NH4+ + Na+ + OCl-

←⎯⎯→ NH2Cl + Na+ + H2O

Equation 9-5

Subsequently the monochloramine reacts with salicylic acid (C7H6O3) in an alkaline environment to form an indophenol dye (Eq. 9-6). This dye dissociates into the blue indophenol anion at high pH and can be detected in very low concentrations by a colorimeter or photometer. Salicylic Acid + NH2CL →

Indosalicylate (Indophenol Dye)

Equation 9-6

9.3 Technical Considerations 9.3.1 Ion Selective Electrodes (ISE) 9.3.1.1 Ammonium Sensing Electrode

Ammonium sensing electrodes are primarily made with a polypropylene case and a glass sensing bulb. Some designs have a polyvinylchloride (PVC) ion sensitive membrane. Electrolytes are frequently potassium chloride (KCl) gels which give maintenance free operation; i.e. the electrolyte does not have to be replenished. The nominal maximum operating temperature of the electrode is 50°C (122°F) but accurate temperature compensation is required due to the temperature dependency of the Nernst Equation. Minimum detection values of the sensing electrode can extend down to 0.01 mg/L (ppm) but most literature describes a minimum quantification range of 0.1 to 1 mg/L (ppm). Optimal electrode response is in the 3 to 8.5 pH + range; practical measurement range should be pH <6 to maximize the ammonium NH4 cation species rather than the ammonia NH3 as dissolved gas. 9.3.1.2 Reference Electrode

While the actual potential measured will depend on the specific type of reference electrode used, the change in potential resulting from a known change in ammonia concentration will be independent of the type of reference electrode. For instance, a ten fold increase in ammonia content at 25°C (77°F) will result in a 59.16 mV difference. The ammonia monitoring instrument is simply a volt meter, more specifically a high impedance voltmeter, calibrated to read ammonium in units of µg/L (ppb) or mg/L (ppm) instead of potential in millivolts (mV). For ammonium measurements to be accurate, the reference electrode must provide a stable reference point against which the ammonium concentration is measured and provide a salt bridge for electrolyte contact with the solution that is being measured. The most common reference electrode used in ammonium ISE tests is the sleeve type double junction design based on the silver/silver chloride (Ag/AgCl) half cell. The inner chamber is filled with potassium chloride (KCl) saturated with silver chloride (AgCl). The outer chamber can be filled with various solutions including potassium chloride (KCl) or sodium chloride (NaCl). Other outer chamber fill solutions can be gel type which minimizes the maintenance required for electrolyte replacement. The outer chamber potassium or sodium chloride solution serves as an electrolytic 9-6


conductor (salt bridge) facilitating detection of the ammonium ion concentration. Electrolytic contact is maintained between the solution being measured and the Ag/AgCl electrode inside the glass tube through the frit in the glass via the saturated potassium or sodium chloride. Potassium or sodium chloride solutions (ions) migrate through the reference electrode frit to assure continuous electrolytic contact with the solution being measured. However, if the KCl solution becomes depleted or diluted, response to the ammonium ion may become sluggish or inaccurate. Depletion occurs by normal migration of the solution through the frit or by evaporation so periodic refilling of the reference fill solution is necessary. Using a gel filled reference electrode eliminates this refilling requirement but usually dictates yearly electrode replacement to assure adequate electrolyte strength. Dilution occurs when the solution being monitored backflows through the frit into the KCl solution. To prevent backflow dilution, the electrodes must be placed in a vented flow-through sample chamber. 9.3.2 Colorimetric Analyzers

When the ammonia test is a colorimetric determination, the addition of chemical reagents to provide development of the proper color species will be required. The need to add the reagents in the proper sequence and amount, assure adequate mixing and necessary reaction times, accurately determine the amount of light absorption at the proper wavelength, correlate this light absorption to some standard curve, and then prepare for the upcoming sample all add complexity to this instrument compared to a simple and continuous potentiometric test such as pH. For this reason, some would argue that an ion selective electrode system is less maintenance intensive than a corresponding colorimetric analyzer. The colorimetric on-line ammonia analyzers perform a batch-wise analysis on the sample. The chemistry of this determination is based on the Berthelot’s Phenate method [9] previously discussed. The classical method uses a mixture of liquefied phenol in ethyl alcohol as the color indicator.

Figure 9-3 Chemical Formula and Structure for Phenol

The manufacturers of on-line analyzers modify the method to use salicylic acid (a slightly different compound in the family of phenols) as the indicator. 9-7


Figure 9-4 Chemical Formula and Structure for Salicylic Acid

Independent of the specific molecule used, the intense blue color due to indophenol formation is monitored at 650 nm wavelength. Table 9-1 displays the precision data for the Standard Method upon which this analyzer is based. This data was developed using a 1 cm path length and suggests that ammonia content at 100 µg/L (ppb) has a 2 sigma RSD of less than 20% on an inter-laboratory basis. Longer path lengths would be expected to improve the low end detection limit of this method. 9.3.2.1 Colorimetric Limitations

All colorimetric tests, such as the test for ammonia, rely on the principle of Beer’s Law (also known as the Beer-Lambert Law) which states that the amount of light absorbed by a sample (A) is proportional to some absorbtivity constant (a), the path-length of light (b), and the concentration of the analyte species(c). A=axbxc

9-8

Equation 9-7

EPRI Proprietary Licensed Material Ammonia Table 9-1 Precision Data for Manual Phenate Method Based on Triplicate Analyses of Ammonium Sulfate [9]

Lab/Analyst

NH3-N Concentration Mg/L

Optical Density

Relative Standard Deviation %

1/1

0.1

0.129

1.55

1/2

0.1

0.114

9.66

2/1

0.1

0.100

10.2

2/2

0.1

0.122

2.36

3/1

0.1

0.112

3.61

3/2

0.1

0.107

1.94

1/1

0.3

0.393

0.39

1/2

0.3

0.364

0.32

2/1

0.3

0.372

2.64

2/2

0.3

0.339

0.90

3/1

0.3

0.370

0.31

3/2

0.3

0.373

0.46

1/1

0.5

0.637

0.77

1/2

0.5

0.630

0.56

2/1

0.5

0.624

1.65

2/2

0.5

0.618

0.86

3/1

0.5

0.561

0.27

3/2

0.5

0.569

0.91

The absorbtivity constant is determined by the color species and test conditions of the method and remains constant as long as the reagents, test conditions, and wavelength of light do not change. The path length (the amount of sample through which the light passes) is also constant for a given sample cell. The mathematical formula then infers that the amount of light absorbed by a sample (i.e. the intensity of the color developed in the test) is directly proportional to the concentration of the analyte—in our case, ammonia. As the ammonia concentration increases, the intensity of the blue color will increase and the color change can be quantified. Limitations on this analytical principle arise at both the low end and high end of the useable range. The low end limit is caused by the detection limit of the test—this is primarily a function of the photo-multiplier (detector) sensitivity and stability. How small a change can be detected by the sensor and how stable is the baseline (or the zero reading)? When trying to quantify a 9-9


concentration near the detection limit, it is not uncommon to see negative readings or duplicate readings that have large relative errors although the absolute error may only be 10 or 20 µg/L (ppb). As a result, most ammonia analyzers will list a detection limit, an accuracy limit, and a precision limit. This accuracy limit is typically an absolute value or some percentage of the reading, which ever is bigger. A typical analyzer has the following specifications which demonstrate this technical consideration [7]. Minimum Detection Limit: .02 mg/L (ppm)as N Accuracy: ±5% of reading or ±0.02 mg/L (ppm) as N, whichever is greater Precision: ±3% of reading or ±0.01 mg/L (ppm) as N, which ever is greater Evaluation of this specification suggests that the analyzer should “see” 0.02 mg/L (ppm) of ammonia but that the accuracy of the reading is ±0.020 mg/L (ppm) (a 200% relative error at this very low value) and that the precision of the reading is ± 0.01 mg/L (ppm). A different analyzer may have a lower detection limit as a result of using another analytical method (changing the absorbtivity constant) and using a larger sample cell (obtaining a longer path length). Limitations on the high end of the analytical range, shown in Figure 9-5, a rise from a property called self-absorption. As the intensity of the blue color increases, the amount of light coming into the sample cell is almost totally absorbed.

Figure 9-5 Absorbance vs. Concentration

At some point there is 100 percent absorption, the line flattens out, and the analytical test is no longer usable. However, the area where linearity first starts to deviate is also an area of concern. The molecules imparting color (absorbing light) are so numerous that some molecules are shaded from the incident light and not accounted for. The ionic solution strength at higher concentrations also starts to affect the color developing molecules and leads to non-linear 9-10


response. Most instruments therefore have a maximum analytical range which corresponds to the top portion of the straight-line relationship. Analytical readings above this linear region are disallowed and the analyzer produces an “over range” alarm. 9.3.2.2 Sample Considerations

Process samples for ammonia analysis need to be representative of the sample stream being measured. As such, proper flushing of sample lines, temperature conditioning, particulate removal, and pressure conditioning are all required. Many analyzers have a failsafe diagnostic which senses a loss in sample flow and triggers an alarm. In addition, because the sample analysis is based on a color development technique, proper flushing of the sample cell is required between samples to remove the previous test solution. This cell flushing step is usually accomplished by regulating the incoming sample pressure and allowing a predetermined time to completely exchange the new sample for the prior test solution. 9.3.2.3 Time Delay

Another technical consideration is the time lag required for reagent addition and color development, which produces a batch type process. A typical colorimetric ammonia analyzer has a nominal 5 to 15 minute cycle time from introduction of one sample to the next. This analysis delay could be troublesome if the process stream ammonia concentration is changing rapidly. The amount of time required for color development is temperature dependent. Some colorimetric ammonia instruments try to account for varying temperatures by adding sample heaters or modifying the delay time set aside for color development as a function of sample temperature. Most ammonia analyzers without heaters make allowances for this color development time and will accept inlet sample temperatures in the nominal 5-50°C (41-122 °F) range. More rapid cycle time is obtained with a heater which elevates the incoming sample temperature to 55°C (131 F) and reduces the cycle. 9.3.2.4 Sample Temperature

Further sample temperature consideration must be given if the sample temperature is below the dew point of the surrounding environment. Under these conditions, moisture can condense on the surface of the sample cell which will lead to unstable results. Many instruments are designed to accept an instrument air purge stream to lower the relative humidity of the sample compartment. Any leaks in the sample line or reagent lines within the compartment must also be immediately fixed to prevent condensation of water vapor on the sample cell.

9-11


9.3.2.5 Sample Volume

A fixed volume of sample is required for the colorimetric determination. The fixed volume is provided by a sample cell of known volume or a specific length of small bore tubing (a flow loop) that contains a known volume of sample. The individual chemical reagents can either be added by a mechanical pump, an eductor which creates a low pressure region in the flowing liquid and sucks the chemical in, or through a pressurized system which admits the reagent through a solenoid valve. These chemical reagents must be replenished on a regular basis. Most instruments are designed for a 30–40 day supply of chemicals. Intimate mixing of the chemical and the sample is attained with either a magnetic stirrer or by turbulent flow through the flow loop design. The proper wavelength of incident light is obtained by an interference filter or monochrometer and the differential light absorption is determined by a photo-multiplier detector. This photometric response is then compared to a calibration curve.

9.4 Interferences 9.4.1 ISE Interferences

Ammonium ion selective electrodes also respond to potassium (0.11), sodium (0.0016), calcium (0.00006), and lithium (0.00003) where the numbers in parentheses represent the selectivity coefficient of each ion. This coefficient is a measure of the sensing electrode’s response to an interfering ion compared to the response for the ion being measured [10]. Upon review of the selectivity coefficients of these interfering ions, only potassium is likely to cause a potential interference to the ammonium determination. In the process streams of the feedwater/boiler/steam it is unlikely that sufficient potassium is present to pose a measuring complication. Literature describes a procedure where a second electrode (potassium sensitive) is included in the measurement process to mathematically remove the electrochemical interference of the potassium ion [11]. One manufacturer of on-line ammonia analyzers using ion selective electrode technology allows for elevated potassium content by offering an optional potassium sensing electrode and software modification [12]. +

It is important to remember that the current on-line ammonia analyzers are measuring the NH4 content in the sample. The Description of Methods Section of this document explained the pH relationship between the dissolved gas ammonia and the cationic ammonium ion. At pH = 9.3 the equilibrium reaction is such that 50% of the ammonia present is in each species. At higher pH values (often found in feedwater and steam samples) successively more of the ammonia exists as the dissolved gas rather than the ionic species. This dissolved gas is not measured by the ammonium sensing electrode and instrument results would be erroneously low. On-line instruments need to pH adjust the sample to pH <6 in order to assure an equilibrium shift toward the ammonium cation. It’s interesting to note that the ion selective electrode method described in reference material relies on the ammonia gas sensing electrode [13,14]. This electrode uses a hydrophobic gas permeable membrane to separate the sample solution from an internal solution of ammonium 9-12


chloride. The sample is pH adjusted with a strong base (sodium hydroxide) to pH> 11 which converts any ammonium ions to the dissolved ammonia gas species. The ammonia diffuses through the membrane and changes the internal pH that is sensed by a pH electrode. The fixed level of chloride in the internal solution is sensed by a chloride ion-selective electrode that serves as the reference electrode. Potentiometric response to dissolved ammonia is measured as in the ammonium ion selective process and correlated to ammonia content. 9.4.2 Colorimetric Interferences

All colorimetric tests must also account for particulates in the sample which result in light scatter across the sample cell. The particulates cause an opacity or cloudiness in the sample which can hinder the analytical measurement. This light scatter can be misinterpreted as light absorption which will lead to an erroneously high reading of ammonia content. Most instruments try to minimize this interference by taking a “zero reading” or “base-line reading” of the sample immediately before adding the color developing reagent. This is certainly a valid and desirable step in the measuring process. However, the analytical reading is usually made after the sample has been allowed to stand undisturbed for some length of time. If particles that were suspended at the time of the “zero reading” have settled at the time of the analytical reading, an erroneous quantification will occur. Particulates can thus lead to inconsistencies in analytical results. While instruments advertise their ability to not have particulate interference, the user should make every effort to filter particulates before doing the colorimetric test.

9.5 Calibration 9.5.1 ISE Calibration

The ammonia ion selective electrode analyzer is designed to be calibrated using the double known standard addition (DKA) method. The analyzer uses two standards to calibrate the logarithmic response of the electrode to changing ammonia concentrations [15]. These two calibration points are at approximately 0.19 and 1.90 mg/L (ppm) NH3. Prior to the addition of the first standard, the instrument measures the potential (E0) and stores this value in the microprocessor. A known amount of Standard Solution 1 is added to the sample reservoir which increases the concentration (Cs) by a corresponding amount (dC1). The new potential (E1) is measured and stored when electrode stability is reached. Standard 2 (ideally 10 times more concentrated than Standard 1) is added, which again increases the concentration in the sample reservoir (dC2). Again the new potential (E2) is measured and stored when stable. The three following equations with three unknowns coming from the Nernst Equation can now be solved. E = E O + S log (CS / CIso )

Equation 9-8

where S is the temperature dependent slope and CIso is the ammonium ion concentration in µg/L (ppb) in the sample that produces an electrode potential that is independent of temperature; the isopotential point. This is the reference point for temperature compensation. The addition of the 9-13


standard increases the concentration by a known amount, dC1, to CS+dC1, and changes the measured potential to E1, as shown in equation (9-9):

E1 = E O + S [log (CS + dC1 ) / CIso ]

Equation 9-9

Then, a second addition of a standard solution is made to the sample. This standard solution is preferably about ten times more concentrated than the first, causing a further increase in concentration, dC2, and a new potential, E2, as shown in equation (9-4): E 2 = E O + S [log(CS + dC1 + dC 2 ) / CIso ]

Equation 9-10

The instrumentation automatically solves the three equations (9-8 through 9-10) for the three unknowns, and E0 and S are stored for subsequent use in the on-line monitoring mode. Values for E0, E1, and E2 have been determined during the calibration procedure. The microprocessor solves the three equations giving the values of S and E0 which allows a calibration curve to be established. This curve is stored in the microprocessor for determining unknown ammonia values. Refer to the discussion in 9.2.3 for further explanation of the Nernst Equation and sensing electrode operation. 9.5.2 Colorimetric Calibration

The colorimetric analyzers undergo a one point calibration to establish the slope of the BeerLambert equation. Some systems also have a “zero” solution to begin the calibration sequence. Ideally this calibration point is high enough to minimize uncertainty and bias and yet be close to the concentration range of the ammonia being analyzed. From a practical standpoint, the standard often is much higher in concentration than the sample to provide for stability of the standard. It is very difficult to assure stability of a standard in the µg/L (ppb) range. One manufacturer has its calibration standard at 2 mg/L (ppm) NH3 as nitrogen (N). This calibration sequence is usually performed weekly and often is programmed to occur automatically. Analyzers allow flexibility to perform this calibration more often as well as “on demand”.

9.6 Calibration Checks On-line ammonia instruments should be checked periodically to demonstrate calibration stability. The Line Method [9] is appropriate for verifying instrument stability for either ISE or colorimetric methods. For the Line Method, a calibrated separate ammonia monitor, typically a bench top analyzer is used to analyze the same sample steam as the installed on-line instrument. Provided the bench top analyzer agrees within the acceptance criteria (e.g., matches the results of the on-line analyzer within ± 3 sigma or ± 10%) the on-line instruments calibration is deemed to be

9-14


acceptable. If the results are outside the acceptance criteria the on-line instrument must be recalibrated. For colorimetric methods, instrument analytical capabilities may also be checked periodically to demonstrate calibration stability using the Standard Injection Method [9]. For the Standard Injection Method, a known standard solution, near the mid-point of the calibration curve, is analyzed by the on-line instrument and the results are compared to the acceptance criteria. Acceptance criteria are either established based on statistically derived limits. i.e., ± 3 sigma or based on some predetermined limits established from experience, i.e., ± 10%. Provided the online analyzer agrees within the acceptance criteria the on-line instruments calibration is deemed to be acceptable. If the results are outside the acceptance criteria the on-line instrument must be recalibrated.

9.7 Alternative Methods 9.7.1 Direct Nesslerization

The most common colorimetric ammonia determination (especially from an historical perspective) is Direct Nesslerization [15]. This method is suitable for rapid routine determination of ammonia nitrogen in steam condensate and demineralized water. The classic Nessler reagent is an aqueous mixture of anhydrous mercuric iodide, potassium iodide, and sodium hydroxide. This reagent reacts with ammonia to yield a yellow mercury amine complex. NH4+ + 2[HgI4]2• + 4OH• → HgO·Hg(NH2)I + 7I• + 3H2O

Equation 9-11

The intensity of the yellow color is measured by a photometer at 425 nm wavelength and is compared to a calibration curve developed under the same test conditions. Some precision and bias data are given in Table 9-2. Table 9-2 Precision and Bias Data; Direct Nesslerization [15] Amount Ammonia N Added mg/L (ppm)

Mean Recovery %

Precision mg/L (ppm) St

Precision mg/L (ppm) So

Bias %

0.120

89

0.011

0.003

-10.8

0.200

98

0.013

0.002

-2.5

0.350

98

0.021

0.002

-1.7

1.000

101

0.042

0.014

+1.4

The Nessler determination has several interferences, many of which can be removed by a predistillation step under alkaline conditions. The ammonia vaporizes and is captured in an acidic 9-15


absorber liquid (often boric acid) while leaving many of the interfering ions behind. Most of these interferences are not found in power plant samples. One exception is a positive interference from volatile alkaline compounds such as hydrazine and other amines. These will give a false positive interference but usually the concentration of these compounds is significantly less than ammonia itself. One last comment about the Nessler ammonia test pertains to the chemical reagent itself. The presence of mercury in the test reagent has come under increasing scrutiny in recent years. Disposal of unused reagent and the discard test solutions after the determination can present environmental liability. Most plants using this reagent capture all wastes from the ammonia test and process them through a hazardous waster treatment facility. 9.7.2 Titrimetric Ammonia Determination

Although not suitable for on-line applications, a second chemical test procedure involves a distillation and titrametric determination of ammonia [16]. A sample (nominal 500 mL) is pH adjusted with borate buffer and 6N sodium hydroxide to a pH=9.5 and distilled at 6–10 mL/min into a scrubber solution of indicating boric acid. At least 200 mL of distillate is collected. The presence of ammonia will be indicated by the buffered acid changing from a purple color to green as the condensate caries over. The amount of ammonia which has distilled into the indicating absorber solution can now be titrated with 0.02N Sulfuric Acid. The endpoint will be marked by a color change from green to purple caused by the indicator in the absorber solution. A blank sample is run to determine the presence of any ammonia in the chemical reagents and to duplicate the threshold volume necessary to change indicator colors. The amount of ammonia in the sample is then calculated by: mg NH3 – N / L = [(A-B) 280] / mL sample

Equation 9-12

A = volume of 0.02N sulfuric acid titrated for the sample in mL B = volume of 0.02N sulfuric acid titrated for the blank in mL The distillation and titrimetric test is seldom used in the power plant cycle due to the poor level of sensitivity of the method. A sample containing 200 µg/L (ppb) ammonia was analyzed by 21 laboratories with a standard deviation of 69.8% and a relative error of 20%. Analysis of a similar synthetic sample containing 800 µg/L (ppb) ammonia yielded a standard deviation of 28.6% and relative error of 5%. 9.7.3 ISE – Gas Permeable Membrane

This ion selective electrode method utilizes a gas permeable membrane and differs from the ammonium ion selective electrode used with the on-line analyzers. A sample is pH adjusted to a value > 9.6 and the ammonia gas sensing electrode is used for the quantification. The gas which

9-16


diffuses across the membrane reacts with an internal electrolyte. The resultant pH change of this internal buffer is then correlated to the gaseous ammonia content [17].

9.8 End User Considerations The performance characteristics (quantification range, accuracy, precision, bias, drift) and design characteristics (cycle time, selection of reagents, reagent consumption, sample manipulation, sample conditioning, and chemical interferences) for the ammonia monitor as provided by the manufacturer or supplier should be considered when selecting an suitable instrument. An obvious difference between the two measurement techniques is the continuous readout of the ion selective electrode monitor versus the batch type analysis of the colorimetric monitor. Other on-line ammonia monitor considerations for condensate/steam samples include: •

Inlet sample flow and pressure requirements. – Sensing electrode capable of analyzing ammonia in the range of interest. – Stable reference electrode. – Ease of calibration.

•

Physical space requirements and mounting configuration.

•

Digital Control System (DCS) interface compatibility.

•

Provision for adequate sample and spent reagent drain.

•

Provision for instrument purge air / pressurization air if required.

•

Ability to perform external validation with grab samples or standards.

9.9 Field Experience Several errors are most likely to occur with a colorimetric ammonia analyzer. They can be collected into two groups—Group 1 pertaining to sample delivery and Group 2 related to analyzer malfunction. A brief discussion of these malfunctions follows. 9.9.1 Sample Delivery

•

All analyzers must rely on sample delivery at adequate flow. One design is to monitor the inlet sample pressure to assure adequate flushing between samples. This requires inlet sample pressure at 3.45 kPa ± 2.07 kPa (5 ±3) psi. Another manufacturer has a constant head system with an overflow weir in the sample reservoir. The reservoir also has a float switch installed to indicate a “no sample’ condition. Both designs effectively stop the analysis routine if sample flow is not detected. A no flow condition alarm signal and read out is recommended for the control room / lab.

9-17


•

As discussed previously, the colorimetric ammonia monitor attempts to cancel out the effects of suspended particulates in the sample by taking a “zero reading” immediately prior to color development. However, accumulation of particulates in the sample lines and colorimeter cells can cause very erratic and erroneous readings. Special precautions should be taken during cyclic operation and startup to filter the suspended material prior to the silica analyzer.

9.9.2 Analyzer Malfunction

•

The most common problem with colorimetric ammonia analyzers has to do with liquid handling within the instrument. Depending on the analyzer design, the failure list includes imperfect pinch valves that allow leak through, sample lines that plug or develop pin holes, reagent lines that become unattached or develop leaks, eductors that become plugged and peristaltic pumps that develop leaks. Many of these malfunctions will be evident by liquid lying in the bottom of the analyzer or running out onto the floor. Less obvious malfunctions will be evident by the analyzer failing to calibrate, producing negative values, or yielding wildly fluctuating readings.

•

Reagent and standard replenishment continues to cause malfunctions in all designs. One design also has an internal reservoir of demineralized water used for dilution of concentrated samples. Diligence is required to keep this diluent bottle filled and allow the analyzer to work properly.

•

Electrode condition is certainly an issue for the ion selective ammonium determination. Weekly calibrations are recommended by one manufacturer. Similarly, replenishment of the acidic buffer chemicals must be performed to ensure the ammonia is in the proper ionized species for the ion selective electrodes. Gel filled electrodes have a nominal one year useful life.

9.10 References 1.


2.


3.


4.

Cycle Chemistry Guidelines for Combined Cycle/Heat Recovery Steam Generators (HRSG’s). EPRI, Palo Alto, CA: 2006. 1010438.

5.

Orion Ammonia Monitor 1810 Product Description; Thermo Electron Corporation, Beverly, MA 09151. 2006.

9-18


6.

Standard Methods for the Examination of Water and Wastewater Analysis, Method 4500-F; 21st Edition, 2005.

7.

Swan Monitor FAM Ammonium 2711 Data Sheet; Swan Analytische Instrumente AG, Riedikon, Switzerland. 2005.

8.

Hach APA 6000 Low Range Ammonia Analyzer, Hach Company, Loveland, CO 2001.

9.

Waltron µAI-9046 Ammonia Analyzer, Waltron LLC, Whitehouse, NJ, 08888. Rev 2.08.

10. ASTM D 4127, “Standard Terminology Used with ion Selective Electrodes”, ASTM International, West Conshohocken, PA 2002. 11. “Procedure for Determining the Ammonium Content in Aqueous Solutions with High Potassium Content”, NICO2000 Ltd, London, UK. 2006. 12. FAM Ammonia Monitor Datasheet, Swan Analytische Instrumente AG; Reidikon, Switzerland 2005. 13. ASTM D1426, “Standard Test Methods for Ammonia Nitrogen in Water”, ASTM International, West Conshohocken, PA. 2002. 14. Method 4500-NH3 D, Standard Methods for the Examination of Water and Wastewater, 21st Ed,. 2005. 15. ASTM D1426, “Standard test Methods for Ammonia Nitrogen in Water—Test Method A”, ASTM International, West Conshohocken, PA. 2003 16. Standard Methods for the Examination of Water and Wastewater Analysis, Method 4500-B and C; 21st Edition, 2005. 17. Orion Guide to ISE Analytical Methods. 1977.

9-19


10 CHLORIDE

10.1 Purpose and Use While chloride is not listed as an EPRI Guideline Core Monitoring Parameter [1-4] its analysis is recommended as a troubleshooting tool and for commissioning purposes. The most recent EPRI documents have replaced the chloride monitoring requirement with cation conductivity as the Core Parameter, partially due to the difficulty of on-line measurement of low level anions including chloride. Elevated chloride concentrations in the boiler can lead to corrosive conditions which can damage the waterwall tubes. Carryover can degrade steam purity, lead to build-up on turbine components and promote corrosion mechanisms. Minimizing this carryover is an additional reason to control the chloride content of boiler water. Chloride is most frequently monitored in the plant for the following reasons: •

To warn of in-leakage of contaminants (primarily condenser cooling water ingress).

•

To facilitate the correlation with other chemistry parameters (i.e., cation conductivity).

•

To check the accuracy of water chemistry control (for chloride), so ensuring that carryover and deposit rates are kept at acceptable low levels.

•


•

To warn of make-up demineralizer malfunction.

Although on-line ion chromatography (discussed in Section 13) can be used to monitor chloride in the steam/water cycle [5,6] the chloride analyzer discussed here is most commonly used and employs an ion selective electrode (sometimes called specific ion electrode) and a reference electrode. When these electrodes are immersed in the water sample, the potential difference between them responds logarithmically to changes in the chloride concentration. This potentiometric method is based on ASTM D512 [7]. Chloride concentrations are monitored at various locations in the steam/water cycle. Based on turbine deposition studies and the solubility of sodium chloride in superheated steam a Target Value of 2 µg/L (ppb) was set by EPRI for chloride in boiler feedwater and steam. Compliance with this limit should result in avoidance of damage to the boiler and turbine and help to minimize deposition in the LP turbine. Similarly low levels of chloride < 3 µg/L (ppb) are recommended in the makeup water. The boiler water chloride control limit is dependent on the treatment chemistry and operating pressure. Target values were derived to provide corrosion protection for both the boiler and the 10-1

EPRI Proprietary Licensed Material Chloride

turbine. Table 10-1 illustrates typical control limits of drum boiler chloride in µg/L (ppb) for a 15,858 kPa (2300 psi) operating pressure under different water treatments. Table 10-1 Boiler Water Chloride Limits in µg/L (ppb) @ 15858 kPa, (2300 psi) [1-4] Boiler Water Treatment

All Volatile

Phosphate Continuum Low

Phosphate Continuum High

Caustic

Chloride Limit

65

190

700

340

Chloride may also undergo volatile transport to the steam as both HCl and NH4Cl, the ratio of which depends largely on the pH of the liquid [8]. While the dominant species for NH4Cl solutions in both the high and low temperature phase are NH4+ and Cl-, the species transported to the steam are predominately HCl and NH3. At low ammonia concentrations, the major source of chloride in the steam is partitioning of HCl from the liquid. On the other hand, the primary contribution of HCl partitioning to the chloride content of the steam does not imply that the subsequent appearance of chloride, in the early or first condensate, will be as HCl. For the volatility and speciation in ammonium chloride solutions, it is important to remember that the solute transports to the steam as mixtures of HCl, NH3, and NH4Cl. The relative portions of each species are dependent on temperature, total chloride, solution pH and the presence of other cations (Na+). The difference in pH and the increasing volatility of HCl accounts for the increased importance of HCl at high temperatures. Moreover, under AVT conditions, the high volatility of NH3 may lead to lower boiler water pH values than predicted and hence an increase in chloride as HCl. Monitoring and maintaining the steam chloride ≤ 2 µg/L (ppb) will protect the steam turbine from chloride induced corrosion. Finally, the low buffering capability of ammonia in an AVT program can lead to localized low pH areas in the boiler due to HCl concentration. Hydrogen damage and resultant tube failures are common in All Volatile Treatment/ Equilibrium Phosphate Treatment/ Oxygenated Treatment programs if the chloride content is not kept very low. Consequently, sensitive instrumentation is essential for monitoring chloride in the cycle.

10.2 Description of Method Chloride ion selective electrodes (ISE) are the most common form of on-line analysis for the steam/water cycle. Specifically, the potential difference between a chloride ion selective electrode and a suitable reference electrode in the water sample of interest can be used to calculate the chloride concentration. The response of the ion selective electrode is described by the Nernst equation: E = E o + [2.3026 R T / n F] × log10 [(CCl + C B )/C Iso ]

10-2

Equation 10-1


Where: E = the measured potential of the electrode pair (V) Eo = the measured potential when the chloride concentration in the sample equals CIso (V) R = the ideal gas constant (8.317 joules/°Kxmole) T = temperature of the sample (°K) n = the valence of the ionic species (equals 1 for chloride) F = Faraday’s constant (96,486.7 coulombs/gram-equivalent) CB = the detection limit of the system in µg/L (ppb). This term is used to define the curvature of the electrode response near the detection limit. It depends on pH, temperature and interferences (such as sulfate, iodide, hydroxide and other ions). CIso = the chloride ion concentration in µg/L (ppb) in the sample that produces an electrode potential that is independent of temperature; the isopotential point. This is the reference point for temperature compensation. The term, 2.3026RT/nF, in Equation 10-1 is essentially the temperature-dependent slope of the electrode response. Most current models of chloride analyzers monitor the temperature of the sample and automatically compensate for temperature variations. The newest generation of on-line chloride monitor uses a water sample that is filtered, its pressure is regulated, and its flow rate is controlled by a restrictor tubing [9]. The sample then flows through a fluid connector block, into a reagent diffusion container where the sample is acidified and into a refrigeration unit. Lowering the sample temperature to the 5-15 °C (41-59 °F) range allows increased electrode sensitivity and produces a detection limit of 10 µg/L (ppb) Chloride. Once cooled, the sample is transferred through insulated tubing to the flow cell. Within the flow cell, the cooled sample passes first the chloride sensing and then the reference electrode, the temperature sensing probe, and then is discharged to waste. The potential difference between these electrodes responds logarithmically to changes in the dissolved chloride concentration, as indicated in Equation 10-1. Potential interference from hydroxyl ions (OH-) is eliminated by decreasing the pH to 4 or lower so that the hydroxyl ion concentration is at least three orders of magnitude lower than the chloride concentration. This sample conditioning is performed by introducing formic acid into the sample upstream of the sensing electrodes. Associated instrumentation converts the potential of the ion selective electrode to a chloride ion concentration and displays this information. Addition of a recorder or data logger allows a continuous record of the chloride concentration to be kept. Manufacturers of chloride analyzers that incorporate ion selective electrodes claim that the limit of detection is as low as 5 µg/L (ppb) for the low level instrument and 100 µg /L (ppb) for the high level instrument.

10-3


10.3 Technical Considerations 10.3.1 Sensing Electrode

The chloride sensing electrode is classified as a crystalline membrane. It is designed such that a pellet of silver chloride is in direct contact with the water sample. Because silver chloride has extremely low solubility in water, the silver chloride pellet never reaches chemical equilibrium with the sample water. Instead, a small amount of chloride ion dissolves into the sample. The resulting relative surplus of silver ions at the surface of the pellet creates a measurable electrical potential that varies with the concentration of the chloride ions in the sample. This potential is measured with an external reference electrode and quantification compared to a standard of known chloride content is made. The surfaces of solid-state electrodes passivate in time, causing drift, poor reproducibility, and loss of response at low levels. The electrode can be restored to good operating condition by repolishing with abrasive strips, usually supplied with the electrode. Polishing the electrode is recommended before its first use and when performance deteriorates. This abrasive strip can also be used if the electrode surface has been etched or chemically poisoned. Many on-line instruments include this polishing step in their monthly maintenance protocol. Detailed procedures are included with the sensing electrode. 10.3.2 Reference Electrode

While the actual potential measured will depend on the specific type of reference electrode used, the change in potential resulting from a known change in chloride concentration will be independent of the type of reference electrode. For instance, a ten fold increase in chloride content at 25°C (77°F) will result in a 59.16 mV difference. The most common reference electrode used in chloride ISE tests is the sleeve type double junction design based on the silver/silver chloride half cell. The inner chamber is filled with potassium chloride (KCl) saturated with silver chloride (AgCl). The outer chamber can be filled with various solutions including 10% nitric acid, potassium nitrate, and cesium bromide. Other outer chamber fill solutions can be used provided that there are no species present (such as sulfide or iodide) which will react with the potassium, sodium, chloride, nitrate, or silver ions contained in the inner chamber filling solution. The following diagram shows a design for maintaining the proper amount of filling solution in the outer chamber. This design eliminates the need for frequent maintenance and electrolyte replacement. The 2 oz. (59 mL) reservoir of electrolyte lasts several months and assures constant electrolyte flow through the reference electrode which improves electrode stability.

10-4


Figure 10-1 Thermo Orion Electrode Reservoir System [9] Source: Adapted from Reference 9, Courtesy of Thermo Scientific Corporation

10.3.3 Temperature Considerations

The Nernst equation has a pronounced temperature dependency. Low level chloride results can change by as much as 4% per °C (2.2% per °F) so accurate temperature compensation is required. All current analyzers have temperature compensation included. Literature states that a low end quantification limit of 2 mg/L (ppm) is expected with ASTM Method D512. Past online chloride analyzers have published a measuring range down to 100 µg/L (ppb). However, the newest generation of chloride analyzers actually chills the sample to ≤ 12 °C (54°F); optimum detection limits are achieved at a nominal 5°C (41°F). Published detection limits with this analyzer are 5 µg/L (ppb) chloride; field results suggest 10 µg/L (ppb) as more realistic.

10-5


10.4 Interferences The chloride ion selective electrode is relatively free from interferences. Considerations for low level analyses include minimizing the presence of hydroxyl anions—this is accomplished by the addition of an acidic compound via the diffusion tubing to lower the pH below 4. The chloride electrode can be degraded by sulfide anions and it can also respond to both bromide and iodide halogen anions. However, literature states that to have a measurable effect these interfering ions must be present in the hundreds of mg/L (ppm) concentration—a condition that is not likely to occur in normal power-plant process streams.

10.5 Calibration The chloride analyzer is designed to be calibrated using the double known addition (DKA) method. The high range chloride analyzer uses two standards to calibrate the logarithmic response of the electrode to changing chloride concentrations [10]. These two calibration points are at approximately 1.54 and 15.4 mg/L (ppm) chloride. Prior to the addition of the first standard, the instrument measures the potential (E0) and stores this value in the microprocessor. A known amount of Standard Solution 1 is added to the sample reservoir which increases the concentration (Cs) by a corresponding amount (dC1). The new potential (E1) is measured and stored when electrode stability is reached. Standard 2 (ideally 10 times more concentrated than Standard 1) is added, which again increases the concentration in the sample reservoir (dC2). Again the new potential (E2) is measured and stored when stable. Starting with the Nernst Equation above (Eq. 10-1) and assuming Cs>> CB, (i.e. the concentrations of the standards are much larger than the chloride content at the detection limit) there are three equations with three unknowns: E = E O + S log (CS / CIso )

Equation 10-2

where S is temperature dependent slope and CIso is the chloride ion concentration in µg/L (ppb) in the sample that produces an electrode potential that is independent of temperature; the isopotential point. This is the reference point for temperature compensation. The addition of the standard increases the concentration by a known amount, dC1, to CS+dC1, and changes the measured potential to E1, as shown in equation 10-3: E1 = E O + S [log (CS + dC1 ) / CIso ]

Equation 10-3

Then, a second addition of a standard solution is made to the sample. This standard solution is preferably about ten times more concentrated than the first, causing a further increase in concentration, dC2, and a new potential, E2, as shown in equation 10-4: E 2 = E O + S [log(CS + dC1 + dC 2 ) / CIso ]

10-6

Equation 10-4


The instrumentation automatically solves the three equations (10-2 through 10-4) for the three unknowns, EO and S are stored for subsequent use in the on-line monitoring mode, and a calibration curve is established. This curve is stored in the microprocessor for determining unknown chloride values. The newest generation of monitor employs a two segment calibration method to cover a wide range of concentrations. The first segment is an approximately liner range of 0 to 125 µg/L (ppb) chloride. The second segment is a range of 75 µg/L (ppb) to 1000 µg/L (ppb) where the electrode response is logarithmic with changing concentrations. Calibration is carried out at three chloride levels with concentrations being in the range of 0-20 µg/L (ppb), 75–125 µg/L (ppb), and 100–1000 µg/L (ppb) ranges, starting with the lowest concentration. The first solution can be the “zero chloride” solution and the second and third levels can be generated by introducing known chloride concentrations into the sample chamber. Calibration parameters of both segments are computed by the microprocessor and effects from the temperature fluctuation are constantly corrected. Based on the potential measured on the sample solution, the microprocessor makes a judgment of which segment of calibration is to be used to read the chloride concentration. The electrode response to these two segments can be characterized by the following equations: Low Level:

E = Eo(T) + S1(T)*(C/C2)

Equation 10-5

High Level:

E = Eo(T) + S2(T)*log (C/C2)

Equation 10-6

where: E = measured electrode potential Eo(T) = temperature dependent potential value S1(T), S2(T) = temperature dependent slope values C = concentration (activity) of chloride ion in the sample C2 = concentration (activity) of chloride ion of the second standard By doing a three point calibration, the microprocessor determines the actual values of all the parameters and enables measurement of chlorides at both low and high levels. The monitor incorporates a cooling system to achieve the stated 5 µg/L (ppb) detection limits and uses its microprocessor to constantly update temperature corrections for data supplied by the automatic temperature probe (ATC) probe. Figure 10-2 demonstrates a typical Ion Selective Electrode calibration. Note the theoretical slope of 59.6 mV for each decade of concentration change. A typical working range comprises three or more decades of concentration where the electrode response conforms to the Nernst equation. Note the deviation from Nernst behavior and how the low detection limit is estimated by extrapolating the linear portion of the curve and baseline mV.

10-7


Figure 10-2 Calibration Curve for Low Level Chloride SIE [16]

10-8


10.6 Calibration Checks On-line chloride instrument analytical capabilities should be checked periodically to demonstrate calibration stability. Two methods exist for verifying instrument stability; the Standard Injection Method [9] or the Line Method [10]. For the Standard Injection Method, a known standard solution, near the mid-point of the calibration curve, is analyzed by the on-line instrument and the results are compared to the acceptance criteria. Acceptance criteria are either established based on statistically derived limits. i.e., ± 3 sigma or based on some predetermined limits established from experience, i.e., ± 10%. Provided the on-line analyzer agrees within the acceptance criteria, the on-line instruments calibration is deemed to be within acceptance limits. If the results are outside the acceptance criteria the on-line instrument must be recalibrated. For the Line Method a calibrated separate chloride analyzer, typically a bench top analyzer is used to analyze a sample from the same sample stream as the installed on-line instrument. Provided the bench top analyzer agrees within the acceptance criteria (e.g., matches the results of the on-line analyzer within ± 3 sigma or ± 10%) the on-line instruments calibration is deemed to within acceptance limits. If the results are outside the acceptance criteria the on-line instrument must be recalibrated.

10.7 Alternative Methods There are a multitude of wet chemical analyses available for chloride determination [7]. Although these methods are not currently used for on-line analyses, a listing is provided for reference. 10.7.1 ASTM D512 Test Method A—Mercurimetric Titration

This method describes a colorimetric end-point titration whereby a dilute mercuric nitrate solution is added to an acidified sample in the presence of di-phenylcarbazone-bromphenol blue indicator. The end point of the titration is the formation of the blue-violet mercury diphenylcarbazone complex. This test method is validated for chloride concentrations from 8 to 250 mg/L (ppm). Commercial test kits based on this method have a workable range of 10-8000 mg/L (ppm). 10.7.2 ASTM D512 Test Method B—Silver Nitrate Titration

This method describes a colorimetric end-point titration whereby a sample adjusted to pH 8.3 is titrated with silver nitrate in the presence of potassium chromate indicator. The end point is indicated by the persistence of the brick red silver chromate color. This test method is validated for chloride concentrations from 8 to 250 mg/L (ppm). Commercial test kits based on this method have a workable range of 5-1000 mg/L (ppm).

10-9


10.7.3 Standard Methods: Method 4500 - Cl D. Potentiometric Method [11]

Chloride is determined by a potentiometric titration with silver nitrate solution and a silver-silver chloride electrode system. During titration an electronic voltmeter is used to detect a change in potential between the glass reference electrode and the silver-silver chloride sensing electrode. The end point of the titration is that instrument reading in which the greatest change in voltage has occurred for a small and constant increment of silver nitrate added. Commercial test kits based on this method have a workable range of 5-1000 mg/L (ppm) 10.7.4 Standard Methods: Method 4500 - Cl E. Automated Ferricyanide Method [11]

Thiocyanate ion is liberated from mercuric thiocyanate by the formation of soluble mercuric chloride. In the presence of ferric ion, free thiocyanate ion forms a colored ferric thiocyanate, whose intensity is proportional to the chloride concentration. This method describes an automated process employed by the Technicon AutoAnalyzer II [12]. Using standard sample volumes, the detection limit for this instrument was published as 70 µg/L (ppb) chloride. 10.7.5 Ion Chromatography

An alternative on-line method for chloride determination, ion chromatography, discussed in Section 13 of this manual, uses a specific conductivity detector [13,14]. The trace anion method utilizes a sample pre-concentration column where the anions of interest are collected on an ion exchange column. After a suitable volume of sample has been concentrated, eluent is pumped through the concentrator column to remove the trapped anions. Eluent then flows through the analytical column where the anions are separated based on the retention characteristic of each anion relative to the eluent used. The eluent stream containing the anions of interest then passes through a suppressor device where the cations from the eluent are exchanged for hydrogen ions, converting the anions to their acid form. After the suppressor device, the eluent solution passes through a conductivity detector where the separated anions are detected. Detection limits for the anions are enhanced because the anions are in the acid form rather than the sodium salt. This method reports a single operator detection limit of 0.8 µg/L (ppb). The common practical range of the method is listed as 1-100 µg/L (ppb) for chloride. At least one instrument manufacturer has commercially available on-line ion chromatography equipment operating in the electric utility setting [15].

10.8 End User Considerations The performance characteristics (range of measurement, accuracy, precision, bias, drift, response time, and signal change) for the monitoring equipment, provided by the manufacturer or supplier should be considered when selecting a suitable on-line sodium instrument. In general,

10-10


manufacturers determine these characteristics using their own in-house methods. End users should select the instrument that is best suited for the intended analytical application. Other on-line chloride instrument considerations include: •

Appropriate pH adjustment for analysis range of interest to minimize OH- ion interference with chloride measurements.

•

Ease and robustness of calibration for the intended use.

•

Ease and robustness of calibration verification.

•

Appropriate sensing electrode linearity characteristics in the analytical range of interest.

•

Reference electrode stability.

10.9 References 1.


2.


3.


4.


5.

ASTM D5542-04, “Standard Test Method for Trace Anions in High Purity Water by Ion Chromatography”. 2004 Annual Book of ASTM Standards, Vol. 11.02 Water. American Society for Testing and Materials, Philadelphia, PA.

6.

ASTM D5996-05, “Standard Test Method for Measuring Anionic Contaminants in HighPurity Water by On-Line Ion Chromatography”. 2005 Annual Book of ASTM Standards, Vol. 11.02 Water. American Society for Testing and Materials, Philadelphia, PA.

7.

ASTM D512-04, “Standard Test Method for Chloride Ion in Water”. 2004 Annual Book of ASTM Standards, Vol. 11.01 Water. American Society for Testing and Materials, Philadelphia, PA.

8.

The Volatility of Impurities in Water/Steam Cycles. EPRI Palo Alto, CA: 2001. 1001042.

9.

Model 1817LL Chloride Analyzer Instruction Manual, Thermo Scientific Corporation, Beverly, MA. 2003 S-1817LL-E-0506 RevB.

10-11


11. ASTM D512-04, “Standard Test Method for Chloride Ion in Water”. 2004 Annual Book of ASTM Standards, Vol. 11.01 Water, American Society for Testing and Materials, Philadelphia, PA. 12.

Standard Methods for the Examination of Water and Wastewater, 21st Edition. Baltimore, MD. 2005.

13.

ASTM D 5996. “Standard Test Method for Measuring Anionic Contaminants in HighPurity Water by On-Line Ion Chromatography”. ASTM International, West Conshohocken, PA. 2005.

14.

ASTM D 5542. “Standard Test Method for Trace Anions in High Purity Water by Ion Chromatography”. ASTM International, West Conshohocken, PA. 2004.

15.

Dionex Process Analytical DX-800 Product Information, Dionex Corporation, 1228 Titan Way, Sunnyvale, CA 94088-3603. 2004.

16.

Fossil Power Plant Chemistry, Scientech, LLC, 1060 Keene Road, Dunedin, FL 34698-6300, 2001.

10-12


11 HYDRAZINE

11.1 Purpose and Use Hydrazine is a reducing agent that is frequently used in boiler feedwater treatment to promote oxygen removal. Hydrazine reacts with oxygen to form water and nitrogen and, under certain conditions (e.g. temperatures > 270°C (518°F)), it can also decompose to form ammonia and nitrogen. Hydrazine is also instrumental in promoting magnetite (Fe3O4) formation which provides an iron oxide surface that consumes dissolved oxygen. Reducing agents such as hydrazine are critical in mixed metallurgy systems (systems containing copper alloys) to prevent copper corrosion due to amines in the presence of dissolved oxygen. While hydrazine is not listed as an EPRI Core Monitoring Parameter [1-3], it is often monitored in mixed metallurgy feedwater cycles using reducing All Volatile Treatment—AVT(R). This chemical treatment scheme refers to the feedwater treatment and is not influenced by further addition of phosphate or caustic in the boiler drum. The prescribed location for this sample point is the deaerator inlet, although many plants sample the economizer inlet as well. Proper control of hydrazine as a reducing agent is necessary to protect the copper alloys in the mixed metallurgy system from oxygen attack. Overfeeding hydrazine can lead to excess ammonia in the system, which can also be damaging to the copper alloys. Hydrazine feedrate can be indirectly controlled by measurement of the oxidation reduction potential (ORP) of the system—discussed in detail in Chapter 6. Hydrazine is continually monitored on-line in the plant for the following reasons, listed in order of importance: •

To check the accuracy of water chemistry control, so ensuring that corrosion rates are kept at acceptably low levels.

•

To evaluation of other chemistry parameters (i.e., ORP and dissolved oxygen).

•

To provide feedback stimulus for automated process control.

The data generated by continuous on-line monitoring of hydrazine is used by plant chemistry and operations personnel. The goal for plant personnel is to maintain hydrazine within prescribed limits.

11-1

EPRI Proprietary Licensed Material Hydrazine

11.2 Description of Methods Several methods are available for monitoring hydrazine concentration on-line: ion chromatography, colorimetry, amperometry, and the ion selective electrode method. Since ion chromatography is not typically used for on-line surveillance and control of hydrazine, it will be discussed in subsection 11.7, Alternative Methods of Analysis. The colorimetric method is an automated version of the ASTMD 1385 [4] grab sample technique, based on the reaction of hydrazine with p-dimethylaminobenzaldehyde to produce a yellow reaction product in an acidic environment. The intensity of the yellow color, which is proportional to the hydrazine concentration in the sample, is measured with a spectrophotometer at a wavelength of 460 nm. The amperometric methods fall into two variants: a 2-electrode system in which an anode and a cathode are made from dissimilar metals, and a 3-electrode system in which anode, cathode and reference electrodes are connected to a potentiostatic circuit. In each case, the anode contacts the water sample containing hydrazine, and the potential difference between the anode and cathode promotes the oxidation of hydrazine at the anode surface. The current that flows between the anode and cathode as a result of the electrochemical reactions is proportional to the hydrazine concentration. The ion selective electrode (ISE) method involves adding iodine to the water sample which reacts with the hydrazine to produce iodide. The concentration of iodide produced, which is proportional to the concentration of hydrazine in the water sample, is measured with an iodide ion selective electrode (sometimes called specific ion electrode).

11.3 Technical Considerations 11.3.1 Colorimetry

Typically, the water flow rate to the analyzer inlet exceeds the minimum flow required and the excess flows to waste. This arrangement ensures that a fresh sample is always available to the analyzer. Often, the sample is also conditioned to regulate the inlet water pressure. At the beginning of each analysis cycle, the measurement cell is flushed thoroughly with a fresh sample. An initial sample blank absorbance reading is taken with the spectrophotometer at a wavelength of 458 to 460 nm to provide the zero reference measurement. Then a solution of p-dimethylaminobenzaldehyde in methyl alcohol and hydrochloric acid is added to the sample. A yellow color characteristic of p-dimethylaminobenzalazine is formed (Eq. 11-1), the intensity of which depends on the amount of hydrazine present. A second absorbance measurement is made at 458 to 460 nm and compared with the zero reference measurement. The difference between the two measurements, which is proportional to the hydrazine concentration in the sample, is displayed directly in units of hydrazine concentration.

11-2


2 ρ − (CH 3 )2 NC6 H 4CHO + N 2 H 4 → (CH 3 )2 NC6 H 4CHN − HCHC6 H 4 (CH 3 )2 + 2H 2O Equation 11-1

11.3.1.1 Colorimetric Limitations

All colorimetric tests such as hydrazine rely on the principle of Beer’s Law (also known as the Beer-Lambert Law) which states that the amount of light absorbed by a sample (A) is proportional to some absorption constant (a), the path-length of light (b), and the concentration of the analyte species(c). A=axbxc

Equation 11-2

The absorption constant is determined by the color species and test conditions of the method and remains constant as long as the reagents, test conditions, and wavelength of light do not change. The path length (the amount of sample through which the light passes) is also constant for a given sample cell. The mathematical formula then infers that the amount of light absorbed by a sample (i.e. the intensity of the color developed in the test) is directly proportional to the concentration of the analyte—in this case, hydrazine. As the hydrazine concentration increases, the intensity of the yellow color will increase and this color change can be quantified. Limitations on this analytical principle arise at both the low end and high end of the useable range. The low end limit is caused by the detection limit of the test—this is primarily a function of the photo-multiplier (detector) sensitivity and stability. How small a change can be detected by the sensor and how stable is the baseline (or the zero reading)? When trying to quantify a test right at the detection limit, negative readings are sometimes observed or duplicate readings are seen that have large relative errors although the absolute error may only be a few µg/L (ppb). As a result, most hydrazine analyzers will list a detection limit, an accuracy limit and a precision limit. This accuracy limit is typically an absolute value or some percentage of the reading. The colorimetric analyzers reviewed did not list an absolute accuracy value. A commonly used analyzer has the following specifications which demonstrate this technical consideration [5]. •

Minimum Detection Limit: 1 µg/L (ppb).

•

Accuracy: ±2% of reading.

•

Precision: ±1 µg/L (ppb) or ±2% of reading whichever is greater.

Evaluation of this specification shows that the uncertainty of the measurement is large relative to the measurement itself at or near the minimum detection limit. At lower concentrations (1–5 µg/L (ppb)) end users are cautioned about using data for critical decision making since the inaccuracy is large relative to the measurement itself. At concentrations >50 µg/L (ppb) the precision limit of ± 5% is used since it becomes larger than ± 1 µg/L (ppb). A different analyzer may have a different specification with a lower detection limit as a result of using another analytical method (changing the absorption constant) and using a longer path length sample. 11-3


Limitations on the high end of the analytical range (Figure 11-1) arise from a property called self-absorption. As the color intensity becomes more and more yellow, the amount of light coming into the sample cell is almost totally absorbed.

Figure 11-1 Absorption vs. Concentration

At some point there is 100 percent absorption, the line flattens out, and the analytical test is no longer usable. However, the area where linearity first starts to deviate is also an area of concern. The molecules imparting color (absorbing light) are so numerous that some molecules are shaded from the incident light and not accounted for. The subsequently curved line can no longer be used for accurate analytical determination. The ionic solution strength at higher concentrations also starts to affect the color developing molecules and leads to non-linear response. Most instruments therefore have a maximum analytical range which corresponds to the top portion of the straight-line relationship. Analytical readings above this linear region are disallowed and the analyzer produces an “over range” alarm. Turbidity or color in the water sample may interfere with this analysis, but this is not normally of concern in the sample streams that are typically analyzed in power plants. 11.3.2 Amperometry 11.3.2.1 Two Electrode Method

As indicated above, there are two types of amperometric methods: the 2-electrode and the 3-electrode methods, and details of both methods vary somewhat from one manufacturer to the next. For instance, in one manifestation of the 2-electrode technique, the water sample flows at a constant rate through a tube made from a porous ceramic [6]; other manufacturers make the tube from a semi-permeable membrane instead. A platinum wire anode is positioned along the axis of 11-4


this tube in direct contact with the water sample, and a silver wire cathode is wound around the outside of the tube. The electrodes and ceramic tube are surrounded by a non-porous outer jacket. The space between the jacket and tube is filled with a mixture of gel and silver oxide. The silver oxide, being in intimate contact with the spiral wound silver wire, becomes part of a composite cathode. Electrolytic contact between the platinum anode and silver/silver oxide cathode occurs because ionic transport is possible through the porous ceramic tube. The potential difference between the anode and cathode (essentially a galvanic couple) stimulates oxidation of hydrazine at the anode and reduction of silver oxide and plating of silver at the cathode. The resultant current depends on how rapidly hydrazine is transported to the anode and, for a given flow rate, flow geometry, pH, and temperature, the current is proportional to the concentration of hydrazine in solution. A thermistor or equivalent device may be incorporated into the measurement cell to provide automatic temperature compensation. Some manufacturers suggest that, for optimum performance, the sample should be made highly alkaline by adding sodium hydroxide to the water sample upstream of the measurement cell. The anode reaction is as follows: N2H4 + 4OH-

→

N2 + 4 H2O + 4 e-

Equation 11-3

The following reaction simultaneously occurs at the cathode: Ag2O + H2O + 2e-

→

2Ag + 2 OH-

Equation 11-4

11.3.2.2 Three Electrode Method

Like the 2-electrode method, the 3-electrode method is configured to generate a mass transport limited current that is proportional to the hydrazine concentration. Design details vary from one manufacturer to the next, but in one manifestation, the sample must be pre-conditioned if the conductivity of the water sample is < 8 µS/cm. If pre-conditioning is necessary, the water must flow through a bed of granulated marble to increase the conductivity and adjust the pH. The water sample then flows at a constant rate through a stainless steel tube that also serves as the counter-electrode (cathode) of the measurement cell. A gold-plated stainless steel sensing electrode (anode) is positioned along the axis of this tube; a silver/silver chloride reference electrode is mounted nearby, and brought into electrolytic contact with the water sample by means of an electrolyte bridge. The bridge is typically a plastic tube, filled with potassium chloride solution, and fitted at its end with a porous ceramic diaphragm to minimize mixing of the potassium chloride and the water sample. Another manufacturer pre-conditions (increases the conductivity of) the flowing water sample by injecting potassium nitrate; the water then flows to a specially designed flow cell in which a silver counter-electrode (cathode), a platinum sensing electrode (anode), and a calomel reference electrode are mounted. In all 3-electrode designs, the sensing electrode, counter-electrode and reference electrode are connected to a potentiostat. The potentiostat controls the electrochemical potential difference between the sensing electrode (anode) and reference electrode by supplying, and automatically adjusting, the flow of direct current between the counter- and sensing electrodes (cathode and 11-5


anode). Thus, the potentiostat replaces the stimulus of the galvanic couple in the 2-electrode method, and ensures that the electrochemical conditions prevailing at the anode surface lead to a mass transport limited current. One manufacturer that uses a platinum anode (working electrode) and stainless steel cathode (counter-electrode) reports a linear relationship between cell current and hydrazine concentration in the range of 0 to 500 µg/L (ppb) when the anode potential is controlled at +480 mV versus the Ag/AgCl reference electrode [5]. Following are the reactions at the two electrodes: Anode:

N2H4 + 4 OH-

Cathode: 4 H2O + 4 e-

→ →

N2 + 4 H2O + 4e2 H2 + 4 OH-


The reducing agent is oxidized on the platinum electrode into nitrogen and water. Simultaneously the counter electrode decomposes the water. For a given flow rate, flow geometry, pH, and temperature, the measured current is proportional to the concentration of hydrazine in solution (just as it is in the 2-electrode method). One advantage of the 3-electrode system is that the cathode needs very little maintenance whereas, in the 2-electrode system, the silver oxide gel mixture must be replaced when it has been consumed by the cathodic reaction. In addition, the 3-electrode system theoretically has a faster response because both anode and cathode are in direct contact with the water sample. Teflon® beads, driven by the sample flow, circulate on the surface of the platinum anode to prevent deposition. In all amperometric methods, the accuracy of measurements can be degraded by the presence of other oxidizable species in the water. Sulfite, for instance, must be removed by passing the sample through an anion bed before the hydrazine concentration can be measured. In addition, the concentrations of morpholine, cyclohexylamine, and ferrous ion should each be maintained below 1 mg/L (ppm) to avoid significant errors. Amperometric methods also respond to carbohydrazide (a hydrazine substitute) which may be of analytical interest or may be an interference depending on the type of reducing agent being used. The reaction of hydrazine in amperometric determinations is enhanced by an elevated pH. Amines such as ammonia, diisopropylamine (DIPPA), diethylamine (DEA), or monoethylamine (MEA) are used to condition the sample to a pH >10.2 before it enters the measuring cell. 11.3.3 Iodide Ion Selective Electrode Method

In this method, the pressure and flow rate of the water sample are controlled at the inlet of the analyzer. The water flows at a constant rate through special plastic tubing that is coiled in a reagent bottle containing both acid and iodine. Both substances diffuse through the tubing wall and dissolve in the sample, resulting in a reduction of the sample pH and, at the same time, a controlled addition of iodine [9]. The hydrazine in the water sample and the iodine react to form iodide, as follows: N 2 H 4 + 2 I 2 → 4H + + 4I - + N 2

11-6

Equation 11-7


The acid is added to stifle the following competing reaction which, if it occurred, would lead to production of iodide by a route that is unrelated to the hydrazine concentration: 3H 2O + 3 I 2 → 6H + + 5I - + IO 3-

Equation 11-8

Thus, each molecule of hydrazine in the original sample produces four iodide ions and the concentration of iodide is directly proportional to the original hydrazine concentration. The sample flows on to the measurement cell which houses an iodide ion selective electrode and a reference electrode. The iodide electrode responds logarithmically to iodide concentration, as described by the Nernst equation: E = E o + [2.3026 R T / n F] × log10 [CI /C Iso ]

Equation 11-9

where: E

=


Eo

=

the measured potential when the iodide concentration in the sample equals CIso (V)

R

=


T

=


n

=

the valence of the ionic species (equals 1 for iodide ion)

F

=


CI

=

effective iodide concentration, i.e. activity, µg/L (ppb)

CIso =

the iodide concentration in µg/L (ppb) in the sample that produces an electrode potential that is independent of temperature: the isopotential point. This is the reference point for temperature compensation.

The measured potential, E, may be converted directly to units of hydrazine concentration and is usually displayed by the instrumentation in this way. It is clear from the Nernst Equation that the measured potential is dependent on temperature as well as iodide concentration. Typically, hydrazine analyzers monitor the temperature of the sample and automatically compensate for changes in temperature.

11.4 Interferences 11.4.1 Colorimetry

The colorimetric method, which is based on the acidified reaction of hydrazine with p-dimethylaminobenzaldehyde is relatively free from chemical interferences. It’s unlikely that the ammonia in the feedwater sample would be strong enough to overcome the hydrochloric acid in the color reagent but note that the color developing reaction is dependent on acidic conditions. As with any colorimetric test, yellow colored species that absorb in the same wavelength of light (460 nm) have the potential to interfere with quantification. Most on-line analyzers attempt to 11-7


avoid this interference by setting the baseline (zero) of the reading immediately before adding the color developing reagent. In this way, any background color will be accounted for prior to color development. The same caveat is true for suspended solids and turbidity in the sample. 11.4.2 Amperometry

The amperometric methods are influenced by other reducing agents such as carbohydrazide. Reducing agents are not necessarily a source of interference in that several instruments actually have a customizable program which allows readout in µg/L (ppb) carbohydrazide rather than hydrazine. Use of an amperometric method allows flexibility to use alternate reducing agents. The sensing electrode tends to suffer surface degradation over time and this often leads to sluggish response. Some manufacturers have designed a measuring cell with free-flowing Teflon beads that are agitated by the sample flow and serve to scour the platinum electrode surface clean. 11.4.3 Ion Selective Electrode Method

The ion selective electrode method is susceptible to any other reducing agents that would react with the iodine reagent to produce iodide, the analyte that is detected. Other halogen ions such as chloride also cause a sensor response. As in the amperometric method, molecules of carbohydrazide or other reducing agents can either be an interference source or the analyte species of interest. Use of an ISE method allows flexibility to use alternate reducing agents.

11.5 Calibration 11.5.1 Colorimetry

Most colorimetric on-line hydrazine monitoring instrumentation allows fully automated 2-point calibration. When activated, the instrument pumps calibration standards through the system instead of the normal water samples. The zero and span are then automatically adjusted to be consistent with the concentrations of hydrazine in the standard solutions. The frequency of calibration can be pre-programmed on some instruments. Calibration is traditionally performed by plant staff, typically by activating a built-in function of the instrument that provides an automatic two-point calibration. However, some older instrumentation requires manual calibration. Whatever the age of the instrumentation, a typical calibration would make use of demineralized water (0 µg/L (ppb) hydrazine) and a standard solution containing a known concentration, for instance, 200 µg/L (ppb) hydrazine. Another instrument performs its baseline calibration automatically (perhaps once every 8 hours) and the full scale calibration on another schedule (every two days). Performing additional calibrations while an individual is observing the instrument can be valuable for validity assurance. If sample valves, reagent lines, mixing mechanisms, light 11-8


source, detectors, or amplifiers are not operating properly, the instrument will likely not calibrate to manufacturer’s specifications. The individual commanding the manual calibration can then troubleshoot the analyzer and resolve the discrepancy. Calibration and maintenance procedures are typically described in literature supplied with the monitoring equipment by each manufacturer. Additional insight on maintenance and calibration is included in EPRI Report GS-7556 [7]. 11.5.2 Amperometry

Only one amperometric on-line hydrazine monitoring instrument allows fully automated 2-point calibration in a fashion very similar to the colorimetric design [8]. When activated, the instrument pumps calibration standards through the system instead of the normal process sample. The zero and span are then automatically adjusted to be consistent with the concentrations of hydrazine in the standard solutions. The frequency of calibration can be pre-programmed and manual calibration can also be initiated at any time. This instrument is also designed to accept a grab sample. All of the amperometric on-line hydrazine analyzers allow calibration to a sample value that is determined by a laboratory bench method. These reference methods are either the ASTM D13854 (p-dimethylaminobenzaldehyde) previously discussed or a proprietary Hach method [12]. Bench analysis of this sample and correlation of the on-line instrument to this reading would then theoretically establish the slope for the electrode response. The disadvantage of this approach is that hydrazine values in the process sample are often fairly low. The calibration obtained by correlating the bench test and the analyzer in this low range creates a compressed slope region. This narrow calibration range can result in large inaccuracies when the hydrazine content is high. It is often not feasible to elevate the process hydrazine content in order to determine a response closer to full scale on the analytical curve. 11.5.3 Ion Selective Electrode Method

The hydrazine analyzer based on the iodide ion selective electrode design is calibrated using the double known standard addition (DKA) method [9]. The calibration process makes use of a precisely controlled syringe called a dynamic calibrator. This syringe injects known volumes of hydrazine standard into the flowing sample stream and allows a two point DKA calibration. These two calibration points are at approximately 20 µg/L (ppb) and 200 µg/L (ppb) hydrazine. Prior to the addition of the first standard, the instrument measures the potential (E0) and stores this value in the microprocessor. As shown in Eq. 11-10, the measured potential (E) is equal to E0 when CS = zero: E = E O + S log (CS / CIso )

Equation 11-10

11-9


A known amount of Standard Solution 1 is added to the sample reservoir which increases the concentration (Cs) by a corresponding amount (dC1). The new potential (E1) is measured and stored when electrode stability is reached (Eq. 11-11). E1 = E O + S [log (CS + dC1 ) / CIso ]

Equation 11-11

Standard 2 (ideally 10 times more concentrated than Standard 1) is added which again increases the concentration in the sample reservoir by a corresponding amount (dC2). Again the new potential (E2) is measured and stored when stable (Eq. 11-12): E 2 = E O + S [log(CS + dC1 + dC 2 ) / CIso ]

Equation 11-12

There are now three equations (Eqs. 11-10, 11-11, and 11-12) with three unknowns coming from the Nernst Equation. The instrumentation automatically solves the three equations (11-10 through 11-12) for the three unknowns, and EO and S are stored for subsequent use in the on-line monitoring mode. The working range for the ion selective electrode method is 0-200 µg/L (ppb) with accuracy listed in two different ranges [9]. When the instrument is spanned at 0-50 ppb (µg/L), the accuracy is ±2 µg/L (ppb) or ±10% of the reading, whichever is greater. When spanned over the broader 0-200 µg/L (ppb) range, the accuracy is ±2 µg/L (ppb) or ±15% of the reading. Some analyzers can also be single point calibrated against an external sample analysis (off line calibration to a value determined on the bench) as in the amperometric design.

11.6 Calibration Checks On-line hydrazine instruments should be checked periodically to demonstrate calibration stability. The Line Method [10] is appropriate for verifying instrument stability for either ISE, colorimetric, or amperometric methods. For the Line Method, a calibrated separate hydrazine monitor, typically a bench top analyzer is used to analyze the same sample steam as the installed on-line instrument. The two results are compared to the acceptance criteria (e.g., agree within ± 3 sigma or ± 10%). Provided the online analyzer agrees within the acceptance criteria, the on-line instrument’s calibration is considered to be acceptable. If the results are outside the acceptance criteria the on-line instrument must be recalibrated.

11.7 Alternative Methods The wet chemistry method used for laboratory analysis is invariably based on the p-dimethylaminobenzaldehyde color complex [10]. When a solution of this indicator in methyl

11-10


alcohol is added to hydrazine in a diluted hydrochloric acid environment, a characteristic yellow color of p-dimethylaminobenzalazine, as shown earlier in Eq. 11-1, is formed. The intensity of the yellow color formed is proportional to the hydrazine present and is in good agreement with Beer’s Law in the range of 5–200 µg/L (ppb) hydrazine. The analytical wavelength for this determination is 458 nm with a 50 cm sample cell. Higher concentrations can be determined by making appropriate sample dilutions. Background color in the prescribed wavelength interferes with the test, as do turbidity and other dark colors. Some canceling of these effects is available through suitable manipulation of the colorimetric blank. Ion Chromatography is also available for an alternative method of hydrazine detection in the laboratory [11]. Hydrazine can be separated from other monovalent cations. The cation-resin column is packed with poly (styrene-divinylbenzene) based action exchanger with sulfonic acid functional groups. Silica-based columns are not suitable for hydrazine analysis. The mobile phase is made up of 3.2 mM nitric acid. One problem when using mobile phase nitric acid is that trace amounts of transition metals can irreversibly retain on the column. The symptom is identified by loss of retention time as the exchange sites are tied up by the strongly bound metal ions. To avoid this problem, a scavenger column is connected between the pump and injection valve to remove traces of transition metals form the mobile phase before it reaches the separation column. The scavenger column is packed with high capacity cation exchanger and eliminates frequent regenerations of the separation column and prolongs column life. Reported detection limits for this ion chromatographic analysis of hydrazine are 2 mg/L (ppm) with a 100µL injection volume. The limit of detection can be improved by increasing the injection volume.

11.8 End User Considerations The performance characteristics (quantification range, accuracy, precision, bias, drift) and design characteristics (cycle time, selection of reagents, reagent consumption, sample manipulation, sample conditioning, and chemical interferences) for the hydrazine monitor as provided by the manufacturer or supplier should be considered when selecting an suitable instrument. Other on-line hydrazine monitor considerations for feedwater/condensate samples include: •

Inlet sample flow and pressure requirements

•

Digital Control System (DCS) interface compatibility

•

Provision for adequate sample and spent reagent drain

•

Provision for instrument purge air / pressurization air if required

•

Ability to perform external validation with grab samples or standards

11-11


11.8.1 Recognizing Instrument Malfunctions

Several measurement errors may occur when using a colorimetric hydrazine analyzer. They can be collected into two groups: Group 1, pertaining to sample delivery and Group 2, related to analyzer malfunction. A brief discussion of these malfunctions follows. 11.8.1.1 Sample Delivery

•

All analyzers must rely on sample delivery at adequate flow. One approach is to monitor the inlet sample pressure to assure adequate flushing between samples. Required inlet sample pressure is from 3.45 kPa ± 2.07 kPa (5 ±3 psi). Another manufacturer has a constant head system with an overflow weir in the sample reservoir. The reservoir also has a float switch installed to indicate a “no sample’ condition. Both designs effectively stop the analyses routine if sample flow is not detected. It is desirable to have an alarm contactor available to signal this “trouble” condition to the DCS and subsequently to the control room / lab. On systems without this relay, a hydrazine analyzer can effectively sit idle for long periods of time and the malfunction may go unnoticed.

•

As discussed previously, the colorimetric monitor attempts to cancel out the effects of suspended particulates in the sample by taking a “zero reading” immediately prior to color development. However, accumulation of particulates in the sample lines and colorimeter cells can cause very erratic and erroneous readings. Special precautions should be taken during cyclic operation and start-ups to filter the suspended material prior to the hydrazine analyzer.

11.8.1.2 Analyzer Malfunction

•

The most common problem with colorimetric analyzers has to do with liquid handling within the instrument. Depending on the analyzer design, the failure list includes pinch valves that leak through, sample lines that plug or develop pin holes, reagent lines that become unattached or develop leaks, eductors that become plugged and peristaltic pumps that develop leaks. Many of these malfunctions will be evident by liquid lying in the bottom of the analyzer or running out onto the floor. Less obvious malfunctions will be evident by the analyzer failing to calibrate, producing negative values, or yielding wildly fluctuating readings.

•

Reagent and standard replenishment continues to cause malfunctions in all designs. Purchase of prepared reagents and standards from the instrument supplier is a good practice. One instrument design has an internal reservoir of demineralizer water used for dilution of concentrated samples. Diligence is required to keep this diluent bottle filled and allow the analyzer to work properly.

•

Properly maintained and calibrated, amperometric analyzers seldom have malfunctions. The majority of these instruments have a pH buffering reagent which needs periodic replenishment. Fortunately, the reagent consumption is fairly small and requires attention only once a month. Most systems use incoming sample pressure to provide the motive force

11-12


for the reagent addition (either via an eductor that draws buffering vapor into the sample or via diffusion tubing that is immersed in the chemical) so external pumps and isolation valves are not required. It appears that surface cleanliness and stability of the noble sensing (working) electrode is a critical design consideration. Three of the four amperometric hydrazine analyzers that were reviewed have self-cleaning working electrodes to address this concern. The fourth analyzer (which is unique in that it does not require a pH buffering reagent) is not self-cleaning and requires electrode servicing on a three month schedule. Analyzers in this design category all measure and quantify weekly or monthly drift tolerances which also is an indication of the electrode condition. •

The iodide selective ion electrode analyzer has no moving parts, similar to the amperometric technology. Addition of the single reagent is made by way of diffusion tubing using only the inlet sample pressure and no isolation valves. Interferences to this process are minimal since the samples are fairly pure. Chloride ingress during a condenser leak on sea-water cooling could create erroneous hydrazine results since the electrode also responds to other halogen ions.

11.9 References 1.

Cycle Chemistry Guidelines for Fossil Plants: All Volatile Treatment, Revision 1. EPRI, Palo Alto, CA: 2002. 1004187.

2.


3.


4.

ASTM D1385, “Standard Test Method for Hydrazine in Water”. American Society for Testing and Materials, Conshohocken, PA 19428-2959. 2001.

5.

Hach 9186 Oxygen Scavenger Instrument Manual ,1st Edition; Hach Company, Loveland, CO 2003-04.

6.

Amatek / Thermox 7025 Hydrazine Analyzer; Bulletin P-57; Pittsburgh, PA.

7.

Monitoring Cycle Water Chemistry in Fossil Plants: Vol. 1 Monitoring Results, by A.F. Aschoff, D.M. Sopocy, D.T. Eglar, O. Jonas, J.K. Rice, C.C. Stauffer, and W.E. Allmon. EPRI, Palo Alto, CA: October 1991. GS-7556, Volume 1.

8.

Waltron Aqualyzer 9071 Hydrazine Analyzer Product Information Sheet; Waltron LLC, Whitehouse, NJ 08888-0070.

9.

Thermo Orion 1818AO Oxygen Scavenger Monitor Data Sheet; Thermo Electron Corporation, Beverly, MA, 01915. 2006.

11-13


10. ASTM D1385, “Standard Test Method for Hydrazine in Water”, American Society of Testing and Materials, Conshohocken, PA 19428. 2001. 11. Alltech Application Note: A0034; Alltech Associates, Inc. Deerfield, IL 60015. 1997. 12. Hach Method 8140, Iron Reduction Method. Hach Water Analysis Handbook.

11-14


12 HYDROGEN

12.1 Purpose and Use Hydrogen is not widely monitored in fossil stations [1] but hydrogen analyzers may be used in a variety of power plant applications [1-3]. For instance, in the steam/water cycle of power plants, a layer of magnetite forms on the surfaces of carbon steel tubing and piping which protects these components from further corrosion. Under some circumstances, the magnetite is disrupted, leading to accelerated corrosion (the anodic reaction) at the sites of the disruption and an accompanying production of hydrogen (the cathodic reaction). Even though other possible sources of hydrogen exist, the concentration of hydrogen in the steam, feedwater, etc. has sometimes been measured to provide an indication of active corrosion in the feedwater or steam generator of nuclear plants [4]. Many large electrical generators are cooled with hydrogen gas due to its high thermal coefficient of heat transfer. Maintaining a high hydrogen purity (usually >98%) is necessary for efficient heat transfer. Hydrogen has a combustible range in air of 4%-75% and an explosive range of 13%–59%. Accurate on-line analysis of hydrogen purity is necessary as a safety issue both during normal operations and purging. The data generated by the monitoring equipment is used by the plant chemist and operations department personnel. The goal for plant personnel is to maintain hydrogen within prescribed limits for electrical generators.

12.2 Discussion of Methods Three types of hydrogen analyzers are in use: the polarographic sensor, the thermal conductivity sensor, and a vibratory gas density meter. Both electrochemical (polarographic) and thermal conductivity detectors have experienced widespread use in the past 20 years. Hydrogen analyzers based on polarographic detectors are used for quantifying dissolved hydrogen in aqueous samples. Hydrogen analyzers based on the thermal conductivity are used in gaseous hydrogen analysis (gas chromatography, generator hydrogen purity) and in one design of aqueous sensor [5]. An analyzer designed as a hydrogen purity meter for the gas cooled generator application uses a pair of vibrating piezo-electric elements to calculate the hydrogen concentration [6].

12-1

EPRI Proprietary Licensed Material Hydrogen

12.2.1 Electrochemical / Polarographic

The polarographic method, which is similar to the polarographic method for oxygen (see Dissolved Oxygen, Section 5), makes use of a probe which incorporates two metal electrodes— an anode and a cathode—in contact with an internal electrolyte that is separated from the water sample by a semi-permeable membrane. Hydrogen from the sample diffuses through the membrane and is oxidized at the anode within the probe. The current flowing between the anode and cathode is directly proportional to the partial pressure of hydrogen in the water sample flowing over the outer surface of the membrane. Because a current (amperes) is measured, the polarographic technique may also be termed an amperometric method. 12.2.2 Thermal Conductivity

This method employs a gas chromatograph in conjunction with a thermal conductivity sensor. Hydrogen purity meters for monitoring hydrogen cooled generators also use thermal conductivity sensors. Thermal conductivity (TC) sensors use the high heat transfer coefficient of hydrogen as their basis of operation. The variation in heat transfer ability between hydrogen and some reference gas causes a heated element (filament or thermistor bead) to cool at a differing rate. The electrical resistance of the heated element is proportional to the temperature of that element. Therefore, the resistance to current flow can be a direct measure of the thermal conductivity of the gas and indirectly can quantify the amount of hydrogen in the sample. Detailed discussion of this detector follows in the Technical Considerations discussion later in this section. In gas chromatography the various components of the gas mixture are separated by a packed media (i.e. a molecular sieve) and elute off the separation column at varying times. Each gas eluant is then passed through a detection device. Instruments using thermal conductivity detectors direct the gas through a series of filaments referred to as a Wheatstone bridge design. The thermal conductivity of the gas component is compared to the thermal conductivity of a known carrier gas and this difference in the ability to transfer heat (thermal conductivity) can be used to quantify the individual component. It is important to understand that in gas chromatographic analysis two distinct steps are involved. First is the separation of the gas mixture into the individual gas components by some form of selective column. The retention time of the component as it elutes off the separation column is indicative of the gas identity. After separation, a variety of detection methods can then be used to quantify the individual component; e.g. a thermal conductivity detector. 12.2.3 Gas Density Meter

A gas density meter can also be used to quantify hydrogen content. Hydrogen has the lowest density of all gasses. This physical property can be used to differentiate hydrogen from other gases and gas mixtures such as carbon dioxide and air. The gas density monitor under consideration uses vibrating piezo-electric driver elements to cause a thin walled cylinder to 12-2


vibrate [6]. The resonance frequency of this vibration varies with the density of the gas within the cylinder. A second set of elements located in the same plane senses this resonance frequency and correlates the frequency to the density of the gas filling the cylinder. The vibrating gas density meter is designed to determine the hydrogen purity during normal generator operations. It can also determine the amount of hydrogen in carbon dioxide during the generator purge sequence and the amount of carbon dioxide in air during the final purge step.

12.3 Technical Considerations 12.3.1 Electrochemical / Polarographic

The probe used in the polarographic method contains an anode and a cathode, separated from the water sample by a semi-permeable membrane. The anode material is typically platinum and the cathode is a material such as silver/silver chloride (produced by chloridizing a silver electrode). The semi-permeable membrane may be polytetrafluoroethylene (PTFE), perfluoroalkoxy (PFA), SARAN™, Tedlar® polyvinyl fluoride (PVF) or a similar material; this membrane allows hydrogen to pass through it from the water sample to the internal electrolyte, but prevents the transfer of dissolved solids which, if it occurred, could interfere with the measurement. The internal electrolyte contacting the anode and cathode may be sodium chloride, potassium chloride, or other salt solution, and would normally be buffered. The potential difference between the cathode and anode (the cell voltage) is held constant by external instrumentation. This cell voltage is typically held at about -0.4 volt. The reactions that take place at, say, a silver/silver chloride cathode and a platinum anode in a buffered potassium chloride internal electrolyte are as follows: Cathode:

2AgCl + 2e- → 2Ag0 + 2C1-

(Reduction)

Equation 12-1

Anode:

H 2 + 2OH - → 2H 2O + 2e-

(Oxidation)

Equation 12-2

[or H 2 → 2H + +2e- ]

(Oxidation)

Equation 12-3

The hydrogen is oxidized to water (or protons) and the silver chloride is reduced to metallic silver. This results in the flow of an electrochemical current between the anode and cathode. The charge is carried by free electrons in the wires and ammeter within the external circuit, and by charged ions, such as Cl- or OH-, in the internal electrolyte. The magnitude of this current is controlled by the rate at which hydrogen arrives at the anode surface, which is determined by the rate of hydrogen diffusion through the semi-permeable membrane and the internal electrolyte. The rate at which hydrogen diffuses through the semi-permeable membrane is, in turn, proportional to the dissolved hydrogen content of the water sample flowing past the membrane. Thus, the electrochemical current flow is proportional to the dissolved hydrogen content of the water sample. The current can be converted with amplifiers and appropriate circuitry to provide a current or voltage output for a strip chart recorder, data logger, or control equipment. 12-3


Temperature and pressure compensation are common features of on-line hydrogen monitoring equipment. Temperature compensation is necessary because the permeability of the semipermeable membrane and the hydrogen diffusion rate through water increase as the temperature increases. Temperature compensation is accomplished by incorporating a suitable sensor, such as a thermistor, into the hydrogen probe housing, and including appropriate circuitry in the instrumentation. Optimum temperature compensation can be provided over a wide range by incorporating more than one thermistor, each accurate over part of this range. In this way, the value of dissolved hydrogen that would have been measured at a pre-set standard temperature, usually 25°C (77°F), is automatically calculated and displayed. Pressure compensation is important if the fluids and components within the probe are compressible because the tension in the semi-permeable membrane affects its diffusion characteristics. When necessary, compensation is usually handled electronically by inputting pressure sensor data to the instrumentation from an external pressure sensor in the sample line. Some monitors may have other special features. For instance, in one polarographic hydrogen sensor, an auxiliary ring-shaped anode (“guard ring electrode”), see Figure 12-1, surrounds the main centrally located disk-shaped anode [5]. The auxiliary anode, fabricated from platinum, is designed to reduce or eliminate the adverse effect of interfering species that may initially be present in the internal electrolyte or that are created by the cathodic reaction. In the absence of the auxiliary anode, these species would diffuse to the main anode and contribute to the total measured current, thereby leading to an overestimate of the dissolved hydrogen concentration. With the auxiliary anode in place, these species are removed before they can reach the main anode.

Figure 12-1 Polarographic Hydrogen Sensing Electrode with “Guard Ring” [7]

For accurate measurements, it is essential that the hydrogen content of the water at the outer surface of the semi-permeable membrane be representative of the bulk water passing through the 12-4


flow cell. Because hydrogen diffuses out of the water sample and into the semi-permeable membrane, a continuous flow of fresh water sample must be maintained over the membrane surface. Minimum flow rates (typically 20-50 mL/min) are specified by the manufacturer of the monitoring equipment to ensure that the rate of hydrogen removal by the sensor is negligible compared with the flux of hydrogen through the flow cell. An upper flow rate limit may also be specified, above which the water flow becomes so turbulent that the membrane vibrates and induces convective transport of hydrogen and other species through the internal electrolyte. This condition would lead to an overestimate of the dissolved hydrogen concentration. 12.3.2 Thermal Conductivity

Thermal conductivity determination of hydrogen content is an on-line technique for gaseous environments such as for hydrogen in the generator cooling systems. A critical feature of the thermal conductivity method is the Wheatstone bridge network comprising four filaments, connected together as shown in Figure 12-2. Each filament is located within its own cavity within a single metal block—the thermal conductivity cell block. Two of the cavities (containing filaments Rd and Rb, Figure 12-2) are connected and form part of a flow channel for the sample gas, and the other two cavities (containing filaments Ra and Rc) form part of a separate flow channel for the reference gas. In some applications, the reference flow channel is plugged at its inlet and outlet, and is filled permanently with a non-flowing reference gas. A constant voltage is applied to the bridge to heat the filaments. The heat from the filaments is dissipated by conduction through the surrounding gas to the adjacent block of metal which serves as a heat sink. By keeping the temperature of the heat sink constant at, say, 50°C (122°F), the temperature of each filament is controlled only by the thermal conductivity of the gas surrounding that filament. Since the resistance of the filament is determined by its temperature, the resistance is also controlled by the thermal conductivity of the surrounding gas. The thermal conductivity of a gas is dependent upon its molecular weight: the lower the molecular weight, the higher the mobility of the molecules, and the higher the thermal conductivity. In gas mixtures, the overall thermal conductivity is determined by adding together the products of the individual thermal conductivities of the gases and their partial pressures. Thus, a mixture of nitrogen and hydrogen has a higher thermal conductivity than nitrogen alone, because hydrogen has the lowest molecular weight of any gas. When the thermal conductivities of the sample and reference gases are identical, the heat is conducted away from all four bridge filaments equally, the bridge is balanced electrically, and no output is seen on the voltmeter (or other bridge indicator). When the thermal conductivities of the sample and reference gases are different, the resistances of the filaments exposed to the sample gas and reference gas are different, leading to an unbalance in the bridge which is detected by the indicator. If the concentration of only hydrogen in the sample gas mixture changes with time, the extent of the unbalance can be converted directly to the concentration of hydrogen.

12-5


Figure 12-2 The Wheatstone Bridge Circuit Showing Schematically How the Reference Gas Flows Over Filament Resistances Ra and Rc, while the Sample Flows Over Filament Resistances Rd and Rb. Ra = Rb = Rc = Rd when Reference and Sample Gases have the Same Thermal Conductivity. [Solid lines = electrical circuit; dashed lines = sample or reference gas flow]

This same Wheatstone Bridge design is used in thermal conductivity detectors for gas chromatographs. Chromatography illustrated in Figure 12-3 is a technique for separating chemical substances that relies on differences in partitioning behavior between a flowing mobile phase and a stationary phase to separate the components in a mixture. The sample is carried by a moving gas stream through a tube packed with a finely divided solid that may or may not be coated with a film of a liquid. Because of its simplicity, sensitivity, and effectiveness in separating components of mixtures, gas chromatography is one of the most important tools in chemistry. It is widely used for quantitative and qualitative analysis of mixtures. The method consists of, first, introducing the test mixture or sample into a stream of an inert gas, commonly helium or argon, that acts as carrier. Liquid samples are vaporized before injection into the carrier stream. The gas stream is passed through the packed column, through which the components of the sample move at velocities that are influenced by the degree of interaction of each constituent with the stationary nonvolatile phase. The substances having the greater interaction with the stationary phase are retarded to a greater extent and consequently separate from those with smaller interaction. As the components elute from the column they can be quantified by a detector and/or collected for further analysis.

12-6


Figure 12-3 Diagram of Gas Chromatograph with Thermal Conductivity Detector

Separation of a gas into its multiple components is an example of gas-solid chromatography (GSC). Gas-solid chromatography is based upon a solid stationary phase on which retention of analytes is the consequence of physical adsorption The choice of carrier gas depends on the type of detector that is used and the components that are to be determined. Carrier gases for chromatographs must be of high purity and chemically inert towards the sample e.g., helium (He), argon (Ar), nitrogen (N2), carbon dioxide (CO2) and hydrogen (H2). The carrier gas system can contain a molecular sieve to remove water or other impurities. The carrier gas will always have a thermal conductivity far removed from that of the analyte in question. Therefore, nitrogen and argon are typical carrier gases for hydrogen determination. The thermal conductivity detector described previously can now determine the difference between the eluting component and the pure carrier gas. The identity of the eluting component is determined by the retention time in the column and the concentration is quantified by the height of the peak or the area under it. Modern detectors create a millivolt signal which correlates to the area under the eluant peak and is totalized for quantification against a standard of known hydrogen concentration. On-line analyzers using an in-situ thermal conductivity sensor as in Figure 12-4 are also used for determining hydrogen content in both aqueous and gaseous samples

12-7


Figure 12-4 Hydrogen Sensor with Thermal Conductivity Detector [7]

Rather than utilizing a true Wheatstone Bridge design with four elements and continually flowing reference gas, the small enclosed volume between the diffusion membrane and the thermal conductivity sensor is periodically flushed with a purge gas. After each purge, the gas to be measured diffuses through the membrane, changing the thermal conductivity of the gas surrounding the detector. It is the rate of change of the thermal conductivity that allows the concentration of the gas to be calculated [7]. Selection of membrane material allows flexibility for different sample environments and sensitivity limits. 12.3.3 Gas Density Meter

This hydrogen measurement technique is not intended for samples from the steam-water cycle. It measures the density of a gas sample and compares the resultant value to that of the pure material (be it a pure gas such as hydrogen or carbon dioxide, or a gas mixture such as air) [8]. The technology developments to use the density monitor as a hydrogen purity analyzer have come about since the year 2000. The principle of measurement relates to the relationship between gas density and frequency ratio. [6] The resonant frequency of the thin walled cylinder varies with the density of the gas flowing inside and outside of it. Hence, measuring the resonant frequency will provide the density of the flowing gas. Two pairs of piezo-electric elements are fitted to the cylindrical resonator along the same circumferential plane: one pair for drive and the other pair for detection. The former pair generates two vibration modes of the resonator independently. It is difficult to obtain the gas density from a single resonant frequency since the temperature effect is too large to be 12-8


compensated for completely. Hence, the density of the gas is based on the ratio between the resonant frequencies so as to reduce the temperature effects. Another merit of applying the frequency ratio is that the effect of piezo-electric element deterioration with age is reduced. Fundamentally, a vibratory gas density meter can only measure the density of the gas sample. [6] Nevertheless, if the densities of the individual components of a mixed gas containing two components are known, the concentration of those individual components can be obtained from the measured and corrected density of the mixed gas. The concentrations of component gases cannot be obtained from a mixed gas containing three or more components, except in special cases. Performance specifications for the vibratory gas density meter are available in the literature [8]. The structure of the measuring sensor is designed to minimize interference [6]. Multiple o-rings serve both to provide a gas tight seal and to dampen the resonator from external vibration. A thin platinum resistance temperature detector (RTD) is located in the gas path to accurately measure and compensate for the sample temperature. The variation in the vibration frequency of the resonator is a function of the gas density. This frequency change is used to determine the hydrogen purity of the gas sample.

12.4 Calibration Calibration is accomplished with a zero and span gas which are normally piped in parallel with the normal sample point and controlled via a multi-port switching valve. Automatic, semiautomatic and one-touch manual calibration modes are available. No calibration frequency was implied in the technical literature.

12.5 Calibration Check On-line hydrogen instrument analytical capabilities should be checked periodically to demonstrate calibration stability. One method is recommended for verifying instrument stability; the Standard Injection Method [10]. For the Standard Injection Method, a known standard gas, near the mid-point of the calibration curve, is analyzed by the on-line instrument and the results are compared to the acceptance criteria. Acceptance criteria are either based on statistically derived limits. i.e., ± 3 sigma or based on some predetermined limits established from experience, i.e., ± 10 %. Provided the on-line analyzer agrees within the acceptance criteria, the on-line instruments calibration is considered still acceptable. If the results are outside the acceptance criteria, the on-line instrument must be recalibrated.

12.6 Alternative Methods Although not a continuous monitoring method, a grab sample thermal conductivity approach is an improved version of ASTM D1588 [9] Method A, a standard method that was adopted in 1960, was used for over 20 years, but was withdrawn by ASTM in 1984. In the improved 12-9


version, dissolved and gaseous hydrogen are stripped from the water sample, nitrogen is added to the stripped gas mixture to adjust the pressure to atmospheric, and the gas mixture is cooled (or warmed) to a standard temperature. The thermal conductivity of this water-saturated mixture of nitrogen and hydrogen is compared with the thermal conductivity of a reference gas, such as water-saturated, hydrogen-free nitrogen. This is achieved by exposing the sample and reference gases to metal filaments that are heated by passing a current through them. The resistances of the filaments are controlled by their temperatures which are, in turn, determined by the thermal conductivities of the gases surrounding them. The thermal conductivity is a function of the partial pressure of hydrogen present, so the hydrogen concentration can be related to the measured resistances of the filaments.

12.7 End User Considerations Both the aqueous electrochemical (EC) and thermal conductivity (TC) sensors use a gas permeable membrane to separate the sample liquid from the sensing surfaces. The choice of the membrane material is a function of the desired detection limits, the measurement range, the sample matrix and the response time. Table 12-1 lists some performance ranges for both EC and TC detectors; the membrane materials have some commonality but are not perfectly matched so direct membrane comparison is not intended. The values are intended to be for example only and do not represent all available sensors. Table 12-1 Some Performance Ranges for Aqueous EC and TC Detectors EC Trace

EC Low Range

EC Mid Range

EC High Range

TC Trace

TC Mid Range

TC High Range

Membrane Material (1)

PFA

Tefzel® (ETFE)

Tedlar® (PVF)

Saran™ (PVDC)

Silicone rubber

PFA

ETFE

Measurement Range

0-75 µg/L (ppb)

0-300 µg/L (ppb)

0–3200 µg/L (ppb)

0 µg/L (ppb) to 32 mg/L (ppm)

0–1000 µg/L (ppb)

0–2000 µg/L (ppb)

0µg/L (ppb) to 10 mg/L (ppm)

Accuracy

±1% or 0.03 µg/L (ppb)

±1% or 0.09 µg/L (ppb)

±1% or 1 µg/L (ppb)


±1% or 1µg/L (ppb)



Response time

2 sec

5 sec

6 sec

50 sec

12 sec

17 sec

17 sec

Minimum Sample Flow

50–220 ml/min

40–200 ml/min

20–70 ml/min

20–40 ml/min

250 ml/min

220 ml/min

200 ml/min

Cal Gas

1 % H2

10% H2

100% H2

100% H2

10% H2

100% H2

100% H2

(1) ETFE (ethylene-tetrafluoroethylene), PVF (polyvinyl fluoride), PVDC (polyvinylidene chloride), PFA (Perfluoroalkoxy)

The EC sensor relies on the surface of the platinum anode to generate an electric current in the presence of the hydrogen. For this to take place, an extremely clean metal surface is essential. If 12-10


any film, grease, or other impurity covers the metal surface this reaction may be impeded or even stopped. Also, the reaction that takes place on the chloridized silver cathode leads to loss of performance after a certain operational time. Periodic sensor maintenance must occur to restore original performance. The main steps involved include: •

Dechloridization of the cathode—removing the chloride film from the silver cathode surface.

•

Rechloridization of the cathode—a new layer of silver chloride is grown on the cathode’s surface.

•

Activation of the platinum anode—The center anode surface is polished and treated with nitric acid.

Both the electrochemical and thermal conductivity sensors utilize a membrane to separate the liquid sample from the detecting surfaces and allow hydrogen diffusion into the cell. Periodic replacement of this membrane is to be expected to assure proper sensor operation. Hydrogen purity meters using direct thermal conductivity sensors are available and come as OEM equipment with many hydrogen cooled generators. Gas chromatographic analysis using the thermal conductivity sensor is primarily used as a laboratory technique, but there is no reason that on-line GC could not be applied for a sample from the generator hydrogen cooling system. Using the GC for measuring dissolved hydrogen in an aqueous sample would require a membrane and sweep gas to deliver the sample to the column inlet. Automating this process may be fairly delicate. It is more likely that a manual sample scheme would be utilized. An aqueous sample is pulled in a pressurized container, the sample is purged with a gas, the pressure and /or volume is closely adjusted and the mixture is admitted to a gas chromatograph. The most likely application for the vibratory gas density meter is analyzing gaseous hydrogen (hydrogen purity) in a generator hydrogen cooling system. The system is relatively uncomplicated compared to an on-line gas chromatograph. Performance data was given in an earlier section. It should be noted that hydrogen purity analyzers are also available using thermal conductivity sensors as described earlier. The thermal conductivity of the process gas passes through one half of the sensor and is compared to the thermal conductivity of a reference gas.

12.8 References 1.

Guideline Manual on Instrumentation and Control for Fossil Plant Cycle Chemistry. EPRI, Palo Alto, CA: April 1987. CS-5164.

2.

PWR Primary Water Chemistry Guidelines: Revision 3, prepared by PWR Primary Water Chemistry Guidelines Revision Committee. EPRI, Palo Alto, CA: November 1995. TR105714.

12-11


3.

BWR Water Chemistry Guidelines: 1996 Revision, prepared by BWR Water Chemistry Guidelines Revision Committee. EPRI, Palo Alto, CA: December 1996. EPRI Report TR-103515-R1.

4.

Generation and Retention of Helium and Hydrogen in Austenitic Steels Irradiated in a Variety of LWR and Test Reactor Spectral Environments, F. A. Garner, B. M. Oliver, L. R. Greenwood, D. J. Edwards and S.M. Bruemmer (Pacific Northwest National Laboratory)∗M. L. Grossbeck, (Oak Ridge National Laboratory) 2001.

5.

Orbisphere 3600 series Hydrogen Sensor; Sales Information by Hach Ultra Analytics, Geneva, Switzerland 2003.

6.

GD Series Vibratory Gas Density Meters, Yokogawa Technical Report No. 29 SUZUKI Jun-ichi 2000.

7.

Orbisphere Model 31xxx Sensor Datasheet, Hach Ultra Company—2005.

8.

Yokogawa GD402 Gas Density Meter; Product Bulletin GS 11T3E1-01E; Yokogawa Electric Corporation, Tokyo, Japan. 2000.

9.

ASTM D1588-60 (Reapproved 1974; withdrawn in 1984), “Standard Test Methods of Dissolved and Gaseous Hydrogen in Water”. American Society for Testing and Materials, Philadelphia, PA.

10. Advanced Power Plant Chemistry QA/QC Practices, Scientech, LLC, Clearwater, FL, 2006.

12-12


13 ION CHROMATOGRAPHY

13.1 Purpose and Use Ion chromatography (IC) is a detection technique that is applicable to quantification of both cations and anions. Sodium is designated as an EPRI Core Monitoring Parameter [1-4] while anions such as chloride and sulfate are monitored as needed for trending and troubleshooting purposes. The reader is referred to individual chapters on sodium and chloride to survey the implications of improper contaminant control. Note that sodium is typically monitored on-line by the ion selective electrode (ISE) method. While ion chromatographs are more complex than many of the on-line instruments, they have the advantage of quantifying multiple analytes in one analysis test. Still, use of on-line IC is rare; it is much more common to rely on laboratory IC for analysis of grab samples for trace ion quantification. Ion chromatographs are used on-line in the plant for the following reasons: •


•


•

To check the accuracy of water chemistry control (such as sodium, ammonia, chloride, sulfate, phosphate), so ensuring that corrosion rates in both the boiler and steam system are kept at acceptable low levels.

•

To detect and quantify organic acids which may be the result of water treatment plant inadequacy, condenser in-leakage, or degradation of organic boiler treatment chemicals.

•

To facilitate the correlation of a water chemistry parameter with plant operating variables, with an aim to optimizing operations.

The data generated by continuous on-line monitoring of sodium is used by plant chemistry and operations department personnel. The goal for plant personnel is to maintain sodium below prescribed limits.

13.2 Description of Method In the mid-1970s, it became apparent that critical components in the steam/water cycle of power plants were undergoing localized corrosion because of the presence of very low concentrations (only a few parts per billion, µg/L (ppb)) of ions such as sulfate, sodium and chloride. The corrosive conditions were often transient and occurred at times when grab sample analyses were 13-1

EPRI Proprietary Licensed Material Ion Chromatography

not performed, so the corrosive conditions sometimes went unchecked for long periods. Shortcomings of the on-line instruments then available, such as conductivity meters, led to consideration of IC, a method first described in 1975 [5]. By the end of the 1970’s, rugged online ion chromatographs had been developed that were suitable for use in power plants [6]. Three plant demonstrations of on-line IC were supported by EPRI [7,8] in the 1980s and led to widespread acceptance of the technique by the electric power utility industry. Plants utilizing ion chromatography in this era were limited to bench top analyzers and grab samples. The ion chromatograph is unlike other on-line chemical analyzers discussed in this manual in that it is capable of monitoring the concentrations of many ionic species in a single water sample. For instance, it can monitor trace amounts of anions such as F-, Cl-, Br-, NO2-, NO3-, SO42-, PO43-, and CrO42-, or cations such as Na+, NH4+, K+, Ca2+, Cu2+, Ni2+, Zn2+, and Fe3+, or soluble silica. Ion chromatographs come in a wide variety of models, each customized for a specific task. In principle, a water sample is injected into the IC flow system and is pumped through a resinpacked column along with an eluent, a solution that facilitates the separation of the sample species in the column. The sample species are separated because they move at different rates through the column, depending on their relative affinities for the resin. The eluent and separated sample flow to a detector, such as a conductivity cell, that monitors the passage of the sample species and transmits a signal to a computer. The computer compares the sample data with data from standard solutions, and identifies the various species present, together with their concentrations. IC is a sub-set of a separation process called High Performance Liquid Chromatography (HPLC). HPLC separations rely on a stationary phase and a mobile liquid phase. The various components of the sample (analytes) are attracted to the stationary phase to a differing degree which effects a time-based separation (elution) from the stationary phase column. The liquid phase transports the analytes through the column and to the detecting device. The type and concentration of the liquid phase affects the rate of elution and subsequently the degree of separation. All IC techniques rely on the fact that the species being analyzed in the water sample are capable of separation in one of three basic types of separator columns: •

Ion exchange chromatography, also known as high performance ion chromatography (HPIC).

•

Ion exclusion chromatography, also known as high performance ion chromatography exclusion (HPICE).

•

Ion pair chromatography, also known as mobile phase ion chromatography (MPIC).

Although on-line IC equipment requires an elevated degree of maintenance and involves relatively complex procedures, it analyzes many ions simultaneously, and does so quite rapidly with very good sensitivity and selectivity. Furthermore, when the sample contains only trace amounts of ions, grab sample analysis tends to be imprecise because of contamination during sampling and transfer. The automated water sample handling feature of on-line IC allows much greater precision.

13-2


13.2.1 Ion Exchange Chromatography

This is the most widely used type of IC in power plants. It involves an ion exchange process in the separator column, and it is useful for separating both inorganic and organic anions and cations. The first step in the analysis is injection of a known volume of the water sample into a flowing liquid, called the eluent or mobile phase (see Figure 13-1). The eluent carries the sample into a separator column that is packed with resin beads made from materials such as polymethacrylate or polystyrene-based resin crosslinked with divinyl benzene. The resin is formed with exchange functions or separation groups that interact either with negatively or positively charged ions, depending on whether analysis of anions or cations is desired. When analysis of anions is needed, the exchange function is generally a quaternary ammonium group whereas, in cation chromatography, it is a sulfonate group. An anion exchange resin, for example, may be pre-treated with a bicarbonate (HCO3-) solution to completely convert the fixed -N+R3 groups to the bicarbonate form (resin-N+R3 HCO3-). As the water sample passes through the column, the ions interact with the resin functional groups in different ways depending on their unique chemical and physical properties. For instance, an anion exchange resin in bicarbonate form can exchange a bicarbonate ion for, say, a chloride ion from the flowing water sample: resin − N+ R3HCO3− + Cl − → resin − N+ R3Cl − + HCO3−

Equation 13-1

Adsorption processes may also be involved in the process (especially for hydrophobic ions). The resin will bind the chloride ion temporarily, but eventually the eluent, flowing behind the sample, reverses the procedure by providing a replacement bicarbonate ion and causing the chloride ion to be released or eluted into the eluent. As a result, each type of ion is retained for a discrete, reproducible period of time (retention time) which is characteristic of that ion. This causes a temporal separation of the ions in the eluent so that, as it flows out of the bottom of the separator column, first one ion appears, then another, and so on, as illustrated in Figure 13-2. In this idealized example, the dead time, tm, is the time taken for a compound which does not interact with the separator column to migrate through the column with the eluent. The gross retention time, tms, is the time taken from first introducing the eluent at the top of the column to achieving the peak concentration of an interactive ion at the bottom of the column (peak 1 in Figure 13-2). The net retention time, ts, is equal to the gross retention time less the dead time (ts = tms – tm). Another ion may appear at a greater net retention time, as represented by peak 2 in Figure 13-2.

13-3


Figure 13-1 Schematic of a Single Column Ion Chromatograph

By injecting standard solutions of each ion of interest into the chromatograph, the retention times are established for each, which allows a correlation to be made subsequently between the ts value and the type of ion present. The eluent then flows to a detector that allows the concentration peaks in Figure 13-2 to be detected by one of a number of methods. The detector is coupled to a microprocessor-controlled instrument that monitors the retention time and the height of (or area under) the concentration peak. Comparing the sample data with calibration data, the computer then identifies and quantifies the dissolved ions in the original water sample. The most widely used detector is a conductivity cell, but amperometric, fluorescence, and ultraviolet/visible absorbance detectors are also used on occasion.

13-4


Figure 13-2 Illustration of How Ions Elute (Leave the Separator Column) at Different Rates Resulting in a Separation of Ionic Species in the Flowing Eluent

When the detector is based on conductivity, the higher the concentration of the ions in the eluent, the higher is the conductivity reading. The precise correlation between conductivity and concentration is dependent on the relative mobilities of the ions, and may be determined by calibration with standard samples. Although the separated ions contribute to the conductivity, the eluent itself is often highly conductive. Consequently, the effects of the ions of interest would be overwhelmed unless methods are employed to suppress the eluent conductivity. One approach is to pass the effluent from the separator column through a second column—a suppressor column—before it reaches the detection system. The function of the second column is to convert the highly conductive eluent into a weakly dissociated solution of low conductivity. For instance, in the case of a sodium bicarbonate eluent carrying chloride anions, the suppressor column would exchange the sodium ions for protons to produce weakly dissociated carbonic acid. In this way, the ability of the conductivity cell to detect chlorides is increased dramatically. The suppressor column is also invaluable when a technique called gradient elution is employed. For ion chromatography to be a practical analytical tool, the analyte peaks must be completely separated prior to detection and quantification. Several anions such as fluoride, acetate, formate, chloride and nitrite have short retention times on the stationary phase and a weak eluent (low solution strength) must be used to separate the peaks. However, this same solution strength would cause exceedingly long retention times for anions such as nitrate and phosphate. This slow elution would limit the practicality of the test for on-line utilization. Gradient elution describes a process where the early ions are separated under a low solution strength and then the eluent concentration is increased (other analytical schemes even change the chemical nature of the eluent) to shorten retention time of the later analytes. This eluent change however increases 13-5


the specific conductivity of the flow to the detector and elevates the background conductivity. Use of the suppressor column provides more stable and lower background conductivity to the detector which improves detection limits and response. Gradient elution is further enhanced by a device called an eluent generator. This system electrolytically produces high purity potassium hydroxide eluent from demineralized water. The eluent concentration can be changed by altering the carrier flow and the applied current. In addition to producing a highly repeatable eluent concentration, the KOH eluent is free of carbonate ions to provide the exceptionally flat detector baseline seen in Figure 13-3 [9]. Acidic eluent can also be produced with the eluent generator.

Figure 13-3 Gradient Separation of Common Anions Using a Hydroxide Gradient [9] Source: Reference 9, Courtesy Dionex Corp.

Problems can occur if the water sample contains foulants, such as corrosion product particulates. In such cases, the sample should first pass through a so-called guard column before reaching the separator, as illustrated in Figure 13-1. The guard column removes these species and prevents fouling and interferences in the separator column. If the sample contains a significant concentration of particulate matter, a 0.22 µm or 0.45 µm filter should be installed upstream of the columns to remove the particulates thereby preventing plugging of the columns. 13-6


Modern instruments require only 10 to 100 microliters of a water sample to identify ions present in the parts per billion (µg/L (ppb)) range. The detection limit can be reduced to a few parts per trillion (ppt) or less by enriching the samples in a concentrator column. This is done by passing relatively large volumes (25 milliliters or more) of sample through the column to accumulate the ions of interest before passing the eluent through the analytical column in the normal way. 13.2.2 Ion Exclusion Chromatography

This form of IC is not commonly used for analysis of water samples in power plants. It involves Donnan exclusion, steric exclusion, and adsorption processes on a totally sulfonated, high capacity, cation exchange resin. Ion exchange does not occur in the separation process. Donnan exclusion, for example, relies instead on the ability of undissociated species to penetrate the pores of the resin, while totally dissociated, inorganic acids cannot penetrate the Donnan membrane so are not retained. Thus, dilute hydrochloric acid, which is used as an eluent, is completely excluded from the resin because it is totally dissociated, whereas neutral water molecules and other undissociated species can diffuse unhindered in and out of the pores of the resin. Ion exclusion chromatography is able to separate weak organic acids from totally dissociated acids [19]; separating oxalic, formic and acetic acids from other long chain hydrophilic acids. Ion exclusion chromatography can also be used for determining carbonate and borate concentrations [20]. 13.2.3 Ion Pair Chromatography

This method of IC can separate metal complexes and surface active (exchangeable) anions and cations. However it is rarely used by the electric power utility industry. Ion pair chromatography involves pairing of two ions followed by adsorption on neutral, non-polar, macroporous resin, such as polystyrene/divinyl benzene. Ion pairing usually enhances the separation of ions with similar exchange properties. Although a couple of mechanisms have been proposed, the physicochemical phenomena on which the retention mechanism is based have not yet been fully explained [10].

13.3 Technical Considerations 13.3.1 Sample Injection

Delivery of the sample onto the separator column is accomplished either by direct injection or by sample pre-concentration. As a rule of thumb, direct sample injection onto the separator column (10 to 100 µL) can achieve ion detection limits in the 3-7 µg/L (ppb) levels. Increasing the sample volume lengthens the amount of time required to load the sample onto the separator column. The long injection time causes a broadening of the peak and leads to a poorer signal to noise ratio which is detrimental to low level quantification. The larger sample volume also 13-7


causes a larger “water peak” (negative conductivity spike) which can mask some of the early eluting ions. Lower detection and quantification levels are achieved by loading the sample onto a preconcentrator column. Many ions can be quantified in the few µg/L (ppb) and sub µg/L (ppb) range using a concentrator column without complex chemical pretreatment. This short column (35 -50 mm (0.14-0.20 inches) in length) has a stationary phase similar, if not identical, to that on the separator column. The function of this short column is to strip ions from a measured volume of high purity sample. After the concentration step is complete, eluent flows through the concentrator column and the ions of interest move onto the larger separation column. The concentrator column eliminates the disadvantages of a large water dip and broad peaks that result from large direct sample injection volumes as were just discussed. However, there are also some technical issues with the concentrator column. Additional valves are required for directing a sample through the concentrator and then switching to eluent flow. Additional time must also be built into the sample cycle to allow for pre-concentration. It is important that the direction of sample concentration on the concentrator column and the analyte elution from the concentrator column occurs in opposite directions. This produces a rapid release of the ions onto the separator column. Another consideration is the exchange capacity of the concentrator column which can limit the amount of sample that is exchanged. The situation is further complicated when ions of widely different affinities are present in the sample. An ion with a high exchange affinity can act as an eluent and strip off an anion of lesser affinity. For example, anions such as sulfate can act as an eluent and displace more weakly retained ions such a chloride or fluoride when concentrations are close to the dynamic capacity of the concentrator column. Breakthrough volume, which is defined as the volume of sample that causes an ion of interest to be eluted from the concentrator column, must be determined for a simulated sample matrix to ensure quantitative retention of all anions and accurate results. Procedures for calculating the breakthrough volume are available in the technical literature. Exceeding the breakthrough volume on a cation analysis is very possible when high concentrations of amines (ammonia, mono-ethanolamine, etc.) are present with low amounts of sodium. The hydroxide formed by the neutralizing amine can also act as an eluent when concentrating anions. A high volume direct sample injection technique has recently been developed and incorporates the use of gradient effluent elution [11]. A large volume of sample (up to 1000µL) is injected onto the separator column and the column is allowed to stabilize with a very low eluent concentration for approximately 5 minutes. This displaces the water peak and slowly elutes the weakly attracted ions such as fluoride, acetate, and formate. Gradient increases to higher KOH concentrations are then used to separate the more strongly retained ions such as phosphate and sulfate. The two methods that were put forth have a separation time of 27 minutes (method 1) and 42 minutes (method 2) [11]. This analytical method is very well suited to on-line ion chromatography because valve manipulation is minimized compared to pre-concentration. The eluent generator is highly repeatable, is relatively free from carbonate contamination, and able to run unattended for long periods of time. Conductivity suppression produces a very flat baseline which promotes low detection limits. Chromatograms showing the results of this technique are presented in Figure 13-4 and Figure 13-5. 13-8


Figure 13-4 Anion Analysis Using Ionpac® AS17 Separator Column with Eluent Generator; Method 1 [11] Source: Reference 11, Courtesy Dionex Corp.

13-9


Figure 13-5 Anion Analysis Using Ionpac® AS17 Separator Column with Eluent Generator; Method 2 [11] Source: Reference 11, Courtesy Dionex Corp.

13.3.2 Column Selection

There are four types of columns used in typical cycle chemistry IC applications. They are the concentrator columns, guard columns, separator columns, and suppressor columns. The concentrator, guard, and separator columns are chemically similar for a given separation method. 13-10


When analysis of anions is needed, the functional group is generally a quaternary ammonium group whereas, for cation separation, the active site is a sulfonate group. 13.3.2.1 Concentrator Column

The concentrator column is designed to improve the detection limit of a given analyte. Direct injection is often limited by the volume of sample—as the volume becomes larger, the water dip becomes more of a hindrance and peak shape often broadens. The concentrator column has a similar exchange functionality to the separator column but the liquid phase is the sample (usually fairly pure water). In most cases, the liquid phase of the sample won’t elute the analyte from the concentrator column and the low level ions accumulate near the entrance of the concentrator column. Exceptions to this inert behavior (liquids containing high concentrations of strongly absorbed cations such as amines or non-neutral pH solutions) can create a loss of the first anions to elute from the column. Determination of the breakthrough volume of the concentrator for a given sample matrix is always a consideration. As the sample flows through the concentrator column, the ions of interest occupy an exchange site on the resin surface and remain somewhat firmly attached. At the end of the concentration cycle, chemical eluent flows through the concentrator in a reverse flow and elutes off the trapped ions. Ideally, this elution should be a rapid event to place all of the ions onto the separator column in a plug flow. When concentrating 10 mL or less of sample, a sample loop is the preferred introduction method. Larger volumes of sample are measured as they flow through the concentrator column either with a rotameter (perhaps using the pressure of a process sample stream to facilitate flow) or with a calibrated pump pulling a known volume through the column. One should never use a pump to push a sample through the concentrator column due to the possibility of contamination. If the samples contain particulate material, the concentrator column will become fouled and a high pressure drop can develop. In this application, concentrator columns may have a short useful life. 13.3.2.2 Guard Column

The guard column is really part of the analyte separation process; it is exactly the same construction and material as the separator column but approximately 10%- 20% of the column length. Its function is to be a sacrificial component in the event of particulate contamination from the sample. While the concentrator column in principle performs this same filtering capacity, its filtering capability is compromised because of the reverse directional flow during the elution step. Any particulates that were deposited on the separation media can now be flushed toward the separator column. For this reason, a guard column is employed even in applications where a concentrator column is being used. 13.3.2.3 Separator Column

The separator column is responsible for sequentially eluting the analytes of interest in a very repeatable manner. Separation is always a compromise between degree of peak resolution and the time required for the last peak to elute. The improved performance of HPLC over classical 13-11


liquid chromatography results from the reduction in particle size of the separation media (stationary phase). As the size decreases, more active sites (sometimes called theoretical plates in liquid chromatography terminology) are available for a given length of column. Typical separation columns have internal diameters of 4-10 mm (0.15-0.39 inches) and are 10-30 cm (3.9-11.8 inches) in length. Larger internal diameters and particle dimensions create more diffusion during the separation process and result in broader peaks. Recent advancements have resulted in micro-bore columns with internal diameters of 1-5 mm (0.04-0.20 inches) which are packed with stationary phase particles in the 3-5 µm range. Smaller diameters create faster and sharper separation with less eluent consumption but require higher operating pressures to overcome the internal flow resistance. 13.3.2.4 Eluent Suppressor Column

The final column in a typical IC analyzer is the eluent suppressor column. The most common detector for ion chromatography is one based on specific conductivity. This detector is highly sensitive, is universal for all charged species, and responds in a predictable way to increases in ionic concentration of the analyte. The largest limitation to the conductivity detector is the high background conductivity of the liquid phase eluent. This high background value tends to mask the small changes in conductivity that occurs when the analyte elutes off the separator column. This problem of high background conductivity was solved in 1975 with the introduction of the eluent suppressor column immediately following the separation column [12]. The suppressor column is packed with a second ionexchange resin that converts the ions of the eluting mobile phase to a weakly ionized species without affecting the conductivity due to the analyte. For example, when cations are being determined in the ion chromatograph, hydrochloric acid is a common eluting reagent and the suppressor column is an anion resin in the hydroxide form. The product of the suppressor column reaction is water: H+ (aq) + Cl- (aq) + resin+OH- (s)

→

resin+Cl- (s) + H2O

Equation 13-2

The analyte ions (cations) are not retained or altered by this suppressor column and the background conductivity from the eluent is very low. For anion separations, the suppressor column is a cation exchange resin in the hydrogen ion form. Common anion eluents are sodium bicarbonate, sodium carbonate, and potassium hydroxide solutions. The conductivity of these eluents is suppressed as follows. The largely undissociated carbonic acid does not contribute significantly to the conductivity. The product of the potassium hydroxide suppression exchange reaction is water. Na + (aq) + CO3-2 (aq) + resin-H+ (s) → resin-Na+ (s) + HCO3-1

Equation 13-3

Na + (aq) + HCO3-1 (aq) + resin-H+ (s) → resin-Na+ (s) + H2CO3

Equation 13-4

K+ (aq) + OH- (aq) + resin-H+ (s) → resin-K+ (s) + H2O

Equation 13-5

13-12


After continued operation, the active sites of the suppressor ion exchange resin become saturated with the eluent ions and must be restored to their hydrogen or hydroxide form. Typically this is accomplished with a regeneration cycle—the anion resin is regenerated with a caustic solution and the cation exchange resin is regenerated with an acid solution. This regeneration cycle is either performed external to the ion chromatograph or with switching valves that re-routes the eluent and admits regenerant to the suppressor column. In both cases, there is an interruption to the analytical cycle while the suppressor is being regenerated or reaching equilibrium if it is externally regenerated. Recent advancements include a continually regenerated eluent suppressor column [13]. This regeneration uses an electrochemical water splitting process where DC current is used to generate low level hydronium and hydroxide ions and the suppressor is continually removing eluent ions. Figure 13-6 shows the suppression mechanism inside a suppressor when KOH is the eluent for anion-exchange separations. Analyte ions elute from the separator column with potassium counter ions. Two electrodes, one beside each membrane (on the side opposite the eluent), hydrolyze water to hydronium and hydroxide ions. Hydrogen ions diffuse across the membrane next to the anode, neutralizing the hydroxide eluent to water, while potassium ions from the eluent diffuse across the other membrane, providing counter ions to the hydroxide being generated at the cathode. In effect, potassium hydroxide from the eluents is transferred across the membrane and does not reach the detector. The suppressor thus lowers the background conductivity from the eluent to almost zero and significantly improves the sensitivity of the conductivity detector.

13.4 Eluent Selection Determining which eluent is required for a given ion chromatographic analysis is primarily dependent on the ions of interest and their expected concentration. Recommendations often come from the instrument manufacturer as part of the methodology for a given separation column. In general, the higher the eluent molar concentration, the more rapid the analytes elute from the separation column. The goal for any ion chromatographic analysis is to have complete resolution between analyte peaks and minimize the time required for the last peak to elute from the separation column.

13-13


Figure 13-6 Chemistry and Ion Movement in Continually Regenerated Eluent Suppressor [13] Source: Reference 13, Courtesy Dionex Corp.

The most common eluents for anion separation are alkaline solutions of sodium hydroxide, potassium hydroxide, lithium hydroxide, sodium tetraborate, sodium carbonate, and sodium bicarbonate. Review of current literature shows ionic solution strength of these alkaline eluents ranging in concentration from 0.5 -65 millimolar. Cation separations are typically performed using methanesulfonic acid (MSA), pyridine dicarboxylic acid, ammonium hydroxide, disodium hydrogen phosphate, sodium iodide, nitric acid, ethylene diamine tetra-acetic acid (EDTA), boric acid, diaminoproprionic acid (DAP), pyridine dicarboxylic acid (PDCA), oxalic acid, and citric acid. Literature again shows eluent ionic strength to be in the 1-50 millimolar range. As previously discussed, gradient elution involves changing the eluent composition and /or concentration throughout the course of the analytical run. A low strength eluent is used for perhaps the first five minutes of the run to separate and deliver the low affinity analytes to the detector. At some precise and repeatable time, the eluent solution strength is increased to promote a faster elution of the more strongly attached analytes. In some analytical schemes, as many as four different eluent compositions are used throughout a single run. Gradient elution is made possible largely by the suppressor column described previously. Without the eluent suppressor, the increasing solution strength would produce correspondingly higher baseline conductivity. This baseline shift would make quantification of low level ions very difficult. Efficient eluent suppression results in a flat baseline even as the eluant strength increases and promotes accurate quantification. The quality of water used for eluent preparation is very important. Any contamination in the eluent will occupy ion exchange sites on the separator column and reduce peak resolution. A gradual increase in baseline will often occur when the eluent is contaminated. One 13-14


decontamination technique is to install an eluent trap column in the flow path prior to introducing the eluent into the separator column. This trap will remove trace levels of anion contamination but it requires periodic regeneration to restore its active sites. A recent development in eluent technology is the eluent generator [13] composed of an electrolyte reservoir and an eluent chamber separated by an ion exchange connector. Passage of a current through this system allows electrolytic production of the eluent, as described below. This apparatus allows automatic production of high quality eluents on a real time basis (eluent storage and subsequent contamination is eliminated). Generators are available for a range of eluents including potassium hydroxide, sodium hydroxide, lithium hydroxide, sodium carbonate and sodium bicarbonate, methanesulfonic acid (MSA). By adjusting the electrolytic current supplied to the generator, differing solution strengths can be produced, greatly simplifying gradient elution. One generator can actually change the ratio of sodium carbonate / sodium bicarbonate for further eluent modification. The eluent generator cartridge is the heart of the eluent generation process. Figure 13-7 illustrates the operation principle of the KOH cartridge. The cartridge consists of a high pressure KOH chamber and a low pressure K+ ion electrolyte reservoir. The KOH chamber contains a cathode where the hydroxide ions are generated through an electrolytic water splitting reaction. The K+ electrolyte reservoir contains an anode and an electrolyte solution of K+ ions. The KOH generation chamber is connected to the electrolyte reservoir by means of a cation exchange connector that permits passage of the K+ ions from the electrolyte reservoir into the generating chamber while preventing the passage of anions from the electrolyte chamber into the generating chamber. The cation exchange connector also serves the critical role of a high pressure physical barrier between the low pressure electrolyte reservoir and the high pressure generating chamber. To generate KOH eluent, demineralized water is pumped through the KOH generating chamber and a DC current is applied between the anode and cathode of the eluent generator cartridge. Electrolysis of water occurs at both the cathode and anode of the device. Water is oxidized to form H+ ions and oxygen gas at the anode in the K+ electrolyte reservoir. H2O

→

2H+ + ½ O2 + 2e- (anodic reaction)

Equation 13-6

Water is reduced to form hydroxide ions and hydrogen gas at the cathode in the KOH generating chamber. 2H2O + 2e-

→

2OH- + H2 (cathodic reaction)

Equation 13-7

Hydrogen ions, generated at the anode, displace K+ ions in the electrolyte reservoir. The displaced K+ ions migrate across the cation exchange connector into the KOH generating chamber. These K+ ions combine with the OH- ions generated at the cathode to produce KOH solution, which is used as the eluent for the anion-exchange chromatograph. The concentration of the generated KOH is dependent on the current applied to the KOH generator and the flow rate through the KOH generating chamber. Therefore with a constant flow rate, the eluent 13-15


generator will precisely control the applied current to accurately and reproducibly generate KOH at the desired concentration.

Figure 13-7 The KOH Generator Cartridge Consists of a KOH Generating Chamber and K+ Electrolyte Reservoir Connected by a Cation Exchange Connector [13] Source: Reference 13, Courtesy Dionex Corp.

This same principle is used for generation of NaOH and LiOH alkaline eluents and acidic eluents. A K2CO3 cartridge allows generation of various solutions of this potassium carbonate eluent. Adding the Electrolytic pH Modifier (EPM) to the potassium carbonate generator allows production of bicarbonate and carbonate ions. The ratio of carbonate to bicarbonate can be precisely controlled by adjustment of the current applied to the EPM. As previously discussed, regenerant contamination by trace levels of anions can be overcome with the use of a trap column prior to admission of the eluent onto the separation column. A conventional trap column needs to be periodically regenerated which hampers the continuous operation of an on-line analyzer. Recent developments include a Continually Regenerated Trap Column (CR-TC) [13] for use with hydroxide and MSA eluent generators which allows for operation with very low baseline drift during gradient elution, improved day to day reproducibility and continual on-line operation. The Continuously Regenerated Anion Trap Column (CR-ATC) [13] removes all anionic contamination (including carbonate) for anion exchange applications. The Continuously Regenerated Cation Trap Column (CR-CTC) [13] removes trace level cationic contamination (including ammonia) from MSA applications.

13-16


Both samples and stored eluents absorb carbonate upon contact with air that can compromise the chromatographic separation. The large peak caused by carbonate elution can interfere with accurate determination of sulfite and nitrite which elute on the tail of the carbonate peak. Minimizing the carbonate interference leads to improved quantification. Use of a Carbonate Removal Device (CRD)[14] can reduce this carbonate interference. Figure 13-8 shows a diagram of the CRD.

Figure 13-8 Carbonate Removal Device [13] Source: Reference 13, Courtesy Dionex Corp.

The CRD makes use of a thin film of carbon dioxide-permeable coating on the exterior surface of a narrow bore capillary membrane tube. The CRD is inserted into the system flow path after the anion suppressor. Suppressed eluent including the analytes of interest are routed inside the coated capillary membrane tube. A self-sustaining source of alkaline solution originating from the suppressor sweeps the outside of the coated capillary membrane. The carbonic acid /carbon dioxide diffuses across the membrane, is quickly converted to carbonate by the alkaline material and is swept to waste. Figure 13-9 below shows the improvement when using the CRD [14].

13.5 Detectors The most common detector for on-line ion chromatography is based on measuring the conductivity of the eluent and the subsequent conductivity increase associated with each analyte as it elutes off the separation column. According to published standards [15], the detector should have a low dead volume (1µL). The temperature compensated or corrected flow-through conductivity cell should be capable of measuring from 0 to 1000 µs/cm. Temperature control should be at ±0.5 °C or better if the cell is temperature controlled. Conductivity detection is 13-17


applicable to cations, anions, and several organic acids. When utilizing an eluent suppressor column, the range of conductivity measurement is significantly compressed and a more sensitive cell can be utilized.

Figure 13-9 Comparison of Spectrum with (Top Spectrum) and without Carbonate Removal Cevice [14] Source: Reference 14, Courtesy Dionex Corp.

Another detection module routinely used in ion chromatography but with less versatility than conductivity detectors is the post column reactor. In the post column reactor, color developing reagents are added to the eluting analytes. The degree of color intensity follows the LambertBeer Law as follows: A=axbxc

Equation 13-8

where the color absorbance (A) is proportional to the absorbance constant (a), the sample cell path length (b) and the concentration of the analyte (c). The post column reactor is then followed by a photometric detector operating either in the UV or visible wavelength range. This detector applies equation 13-8 to compare the unknown analyte concentration to a standard of known concentration and thus to quantify the sample. Classic ion chromatography analytical methods using the post column reaction and photometric detection include the speciation of chromium (III) and chromium (IV), speciation of nitrate and nitrite and quantification of organic compounds such as methylcarbamates.

13-18


Fluorescence detectors make use of a physical characteristic of many organic compounds. When bombarded with incident light, these molecules emit a characteristic wavelength of light. The intensity of the emitted light is proportional to the quantity of molecules being bombarded. Therefore, the ion chromatograph separates a group of poly-aromatic hydrocarbons (PAH) into the specific components and the fluorescence detector quantifies the individual components. The difference between the UV / Vis detector described above and the fluorescence detector has to do with the orientation of the photomultiplier device (PMD). In the UV / Vis system, the PMD is in alignment with the incident light and the detector monitors the amount of light that is absorbed. The PMD in the fluorescence detector is at right angles to the incident light and measures the emitted light from the molecule. Electrochemical detectors are used in the analysis of organic acids. They measure the current produced by the oxidation of the compound at the gold electrode. The specific technique used is called pulsed amperometry. A possible application would be ion exclusion separation of organic acids followed by electrochemical detection. Only the conductivity detector is used in current on-line chromatography instruments in the power generation industry. 13.5.1 Interferences

Interference occurs when analyzing anions in power plant water samples when the feedwater is pH adjusted using volatile amines. The eluent used for anion separation is an alkaline liquid phase with closely controlled solution strength in the millimolar range. The strong amine in the sample can cause problems with breakthrough volumes if a concentrator column is used. If direct injection of a large sample volume is utilized, the pH shift caused by the amine can cause premature elution and insufficient peak resolution. To overcome this interference, ASTM D 5996 recommends that the sample for ion chromatographic anion analysis be taken from the effluent of a strong acid resin cation exchange column. There is a risk of sample and standard contamination when working with trace level analysis. Just the act of pulling a sample and transporting it to the IC instrument can contaminate the samples. On-line instruments are very advantageous since sample handling is kept to a minimum and the instrument pulls a sample from a flowing stream without contact with the environment. When analyzing trace ions via grab samples, the sample bottles should be rinsed with deionized (DI) water, filled with DI, and allowed to stand at least overnight before emptying, drying and filling with sample. For ultra-low level analysis, this process may need to be repeated until acceptable blank values are obtained. Poly-carbonate and poly-styrene tissue culture flasks with plug seals have proven to be an excellent material for trace ion grab samples. When a flask is not in use, it is best to keep it full of DI water and rinsed periodically. Disposable, powder-free PVC gloves should be worn when working on the IC system and care should be taken to minimize contact with all solutions, even when wearing gloves.

13-19


Another potential interference is when one analyte is present at such a high concentration that it masks adjacent peaks on the chromatogram. The previous discussion about removal of the carbonate peak (using a Carbonate Removal Device) is an example of this type of interference when analyzing anions. Ideally, one manipulates the chromatographic separation such that the interfering ion elutes later than the analyte of interest, which minimizes the interference. The following chromatogram (Figure 13-10) shows the analysis of low level sodium in the presence of high concentrations of monoethanolamine [18]. The large tailing peak of the monoethanolamine is not a source of interference in this example since it elutes later than the analyte of interest. A boiler water anion analysis containing high phosphate concentration would also benefit from the slow elution time of the phosphate peak (Figure 13-4).

Figure 13-10 Chromatogram of a Sample Containing 0.022 µg/l Sodium and 3000µg/l Ethanolamine [18] Source: Reference 18, Courtesy Dionex Corp.

13-20


13.6 Calibration Ion chromatography calibration usually involves analyses of known standards leading to the development of a 1-, 2-, or 3-point calibration curve. In some instruments, the calibration procedure is automated by using a built-in function; activation may be either manual or computer-controlled to occur at pre-set intervals. 13.6.1 Calibration Checks

On-line ion chromatography instruments should be checked periodically to demonstrate calibration stability. Two methods exist for verifying instrument stability; the Standard Injection Method [16] or the Line Method [17]. For the Standard Injection Method, a known standard composed of the analytes of interest (e.g., Cl, SO4, Na, etc.) in the concentration range where the calibration can be readily verified, is analyzed by the on-line ion chromatograph and the results are compared to the acceptance criteria (e.g., the results should agree within ± 3 sigma or ± 10%). Provided the on-line analyzer agrees within the acceptance criteria, the on-line instruments calibration is considered still acceptable. If the results are outside the acceptance criteria, the on-line ion chromatography must be recalibrated. Unfortunately, any source of contamination would bias the test results high. See the discussion on interferences in Section 13.4 for more information on sample handling. A second approach involves injection of a standard as a known addition into the sample stream and the mixture is analyzed by the ion chromatograph. The results are compared to the acceptance criteria (e.g., they should agree within ± 3 sigma or ± 10%). Again, any source of contamination would bias the test results high. If the results are outside the acceptance criteria the on-line ion chromatograph must be recalibrated.

13.7 Alternative Methods Ion chromatography’s greatest advantage over other on-line detection methods is its ability to analyze for several components in one analytical run. The majority of the components of interest in the power plant have on-line analyzers which are based on alternate methods. The reader is directed to the following sections for further information on these alternate methods. •

Sodium – Section 8 – ISE.

•

Ammonia – Section 9 – ISE and colorimetric.

•

Chloride – Section 10 – ISE.

•

Hydrazine – Section 11 – ISE, amperometric, colorimetric.

•

Phosphate – Section 15 – Colorimetric.

•

Organics – Section 17 – Combustion/Oxidation.

13-21


Note that there are no alternative methods for on-line analysis of low level sulfate. Grab samples for low level sulfate are analyzed on a laboratory IC for quantification and verification of the online instrument. While Section 17 lists alternate methods for TOC determination, only IC has the ability to identify and quantify the individual organic components. Differentiating between organic carbon species may become increasingly important to meet future steam purity limits for the steam turbine.

13.8 End User Considerations The end user needs to understand the degree of maintenance and technical commitment required for on-line ion chromatographic analysis. Several new developments such as eluent generators, continuously regenerated suppressors and trap columns, and automatic calibration should significantly reduce the amount of day to day attention that is required. Without these enhancements, daily attention to suppressor regeneration is likely—this activity usually lends to several hours of down-time while the instrument returns to a stable condition. Most on-line instruments now provide for monthly supplies of eluent in their storage compartments.

13.9 References 1.


2.


3.


4.


5.

H. Small, T.S. Stevens, and W.C. Bauman, “Novel Ion Exchange Chromatographic Method Using Conductometric Detection”, Analytical Chemistry, Vol. 47, No. 11, (1975).

6.

T.O. Passell, “Use of On-Line Chromatography in Controlling Water Quality in Nuclear Power Plants”, J. of Chromatography A, Vol. 671, p. 331. (1994).

7.

On-Line Chromatography at Three PWR’s, by M.N. Robles and J.L. Simpson. EPRI, Palo Alto, CA: September 1985. NP-4121.

8.

In-Plant Measurement of Corrosive Ions in Water, by M.N. Robles, J.L. Simpson, D. Dutina, and T.O. Passell. EPRI, Palo Alto, CA: September 1989. NP-6308.

13-22


9.

Dionex EG50 Product Information Brochure.

10. J. Weiss, Handbook of Ion Chromatography, edited by E.L. Johnson. Dionex Corporation, Sunnyvale, CA (1986). 11. Determination of Trace Anions in High-Purity Waters by Ion Chromatography with the IonPac® As17 Using High-Volume Direct Injection with the EG40. Dionex Corporation, Sunnyvale, CA 2003. 12. H. Small, T. S .Stevens, W .C. Bauman, Anal, Chem,, 1975,47,1801. 13. Dionex Reagent-Free Ion Chromatography (RFIC®) System; Product Brochure, LPN1819 10M, Dionex Corporation , Sunnyvale, CA 2006. 14. Reducing Carbonate Interference in Anion Determinations with Carbonate Removal Device, Dionex Technical Note 62, Dionex Corporation, Sunnyvale, CA 2006. 15. ASTM D5996-05, “Standard Method for Measuring Anionic Contaminants in High-Purity Water by On-Line Chromatography”. ASTM International, West Conshohocken, PA. 2005. 16. Advanced Power Plant Chemistry QA/QC Practices, Scientech, LLC., Clearwater, FL, 2006. 17. ASTM D3864-96(2000), Standard Guide for Continual On-Line Monitoring Systems for Water Analysis, American Society for Testing & Materials, Philadelphia, PA. 18. Determination of Sodium at the Parts-Per-Trillion Level in the Presence of High Concentrations of Ethanolamine in Power Plant Waters—Dionex Application Note 152, Dionex Corporation, Sunnyvale CA. 19. Acclaim OA (Organic Acid) Analytical Column Product Manual; Dionex Corporation, Document 031996-01. 2004. 20. IONPAC® TBC-1 Borate Concentrator Column and IONPAC® ICE Borate Separation Column; Dionex Corporation LPN 1001 3M. 1998.

13-23


14 IRON AND COPPER

14.1 Purpose and Use Iron and copper are not defined as Core Monitoring Parameters in current EPRI Guidelines [1-4] but it is suggested that samples be taken and analyzed periodically to measure corrosion product levels in the steam-water cycle. Performance of corrosion product monitoring campaigns is a useful means of verifying that the levels of corrosion product transport are consistent with those which can be attained when the feedwater treatment has been properly selected and optimized. Transport of feedwater corrosion products can cause problems in both fossil and nuclear power plants. Iron and copper corrosion products are of primary concern because of their relative abundance in power plant cycles. In fossil power plants, transported corrosion products can form deposits on water-wall tubes under high heat flux conditions, providing a site for concentration of impurities and potentially leading to tube failures by caustic gouging, hydrogen damage, and low cycle corrosion fatigue [5]. Corrosion product monitoring in the plant is conducted primarily for the following reasons: •

To facilitate the correlation of a water chemistry parameter with plant operating variables.

•

To check the accuracy of water chemistry control (such as reducing agent, oxygen, ammonia or pH), so ensuring that corrosion rates are kept at acceptable low levels.

The data generated by continuous on-line monitoring of iron and copper is used by plant chemistry and operations personnel. The goal for plant personnel is to maintain iron and copper within prescribed limits. Iron and copper may exist in dissolved, colloidal and suspended forms. Some of the available methods are capable of a total metal determination while others assess the concentration of the metal in a particular form. For example, it is common to test grab samples of boiler feedwater for suspended iron oxides during startup of fossil boilers of once-through design.

14.2 Description of Methods Several methods of analysis are available and in use. Monitoring techniques range from grab sampling and laboratory analysis to on-line automated techniques that often utilize the grab analysis methods, either periodically or with integrated samples. The on-line integrated sampling device allows concentration of filterable species on a 0.22 µm or 0.45 µm membrane and concentration of non-filterable colloidal and soluble species on cation (and, if needed, anion) 14-1

EPRI Proprietary Licensed Material Iron and Copper

ion exchange membranes. The accumulated waterborne particulates can then be analyzed in one of two ways. Either the filters are removed and the samples are digested for subsequent analysis by flame atomic absorption, graphite furnace atomic absorption, or atomic emission spectroscopy, or the particulates on the filters are irradiated on-line with X-rays that cause elements to emit a fluorescent radiation that can be correlated with elemental concentrations (an analytical technique known as X-ray fluorescence or XRF). On-line integrated sampling does not allow real time on-line analysis of iron, copper, and other corrosion products. But the on-line integration of the sample does circumvent the difficulties in obtaining representative samples for standard wet chemistry laboratory methods. As an example of these difficulties, a conventional grab sample containing iron particles with an average diameter of 5 µm and a density of 5 g/mL could give erratic results when directly analyzed by graphite furnace atomic absorption: if the average iron concentration were 10 µg/L (ppb), no more than one particle (and possibly none) would be present in the 20 µL aliquot typically injected into the furnace [5]. Since iron levels in most sample streams of interest are typically below the detection limit of direct XRF analysis, a concentration mechanism is required for XRF analysis as well. Other possible approaches to monitoring corrosion product particulates include: turbidity, particle counters, particle monitors, acoustic detection, and colorimetric.

14.3 Technical Considerations 14.3.1 Integrated Sampling

Several integrated sampling devices are commercially available. They require access to a representative water sample, flowing at rates high enough (flow velocity > 1 m/sec (3 ft/sec)) to avoid corrosion product particle dropout. Significant pressure drops in sample coolers should be minimized for the same reason. Detailed information on sampling procedures is available in EPRI reports CS-5164 and GS-7556 [6,7]. The sample flow is split so that water is delivered continuously to the integrated sampling device at a constant flow rate in the range of about 50 to 200 mL/min, and the remainder flows to the drain. The filtration unit holds a 0.45 µm or a 0.2 µm membrane to collect the particulate matter. If dissolved iron analysis is desired, one or more cation ion exchange membranes are positioned beneath the particulate filter to collect colloidal and soluble species passing through the membrane filter. A constant sample flow rate is maintained through the filter and ion exchange membranes by incorporating a pressure regulator downstream of the filtration unit. The volume of water passing through the integrated sampling device is measured with a sample flow totalizer before the sample flows to waste. For samples that require wet chemistry analysis of the collected material, sampling periods range from one to seven days, equivalent to total sample volumes in the range 144 to 1008 liters for a

14-2


flow rate of 100 mL/min. The filter pad and ion exchange membranes are then removed from the filtration unit for chemical analysis. For samples analyzed using an XRF spectrometer analysis, the sampling period continues until it becomes necessary to change the filter for one of two reasons: either the amount of particles deposited reduces the flow rate below a useful range or the amount of iron in the deposit produces a fluorescent signal beyond the linear response curve of the detector. The total amount of flow through these samplers is critical when calculating or comparing results achieved from wet chemical or XRF spectrometer analysis [8]. 14.3.1.1 Wet Chemistry Analysis of Integrated Samples

For wet chemistry laboratory analyses of integrated samples of iron and copper, the membrane filter (carrying the particulates) and cation resin impregnated membranes (carrying the soluble metals) are digested separately in hydrochloric-nitric acid mixtures to produce the integrated and solubilized samples needed for reliable chemical analysis [9]. A concentration factor of 1000 or more is typical using this approach. Thus, these samples can be used to analyze separately for particulate loading and for dissolved corrosion products (iron and/or copper) in the original sample. Various standard laboratory methods are available for these analyses. For instance, the flame atomic absorption technique is described in D1068-05 Standard Test Methods for Iron in Water [10] and in D1688-02 Standard Test Methods for Copper in Water [11]. Graphite furnace atomic absorption techniques are addressed in EPA 236.2 for iron [12] and EPA 220.2 for copper [13]. Analysis by direct current atomic emission spectroscopy is covered in D4190-03 Standard Test Method for Elements in Water by Direct-Current Argon Plasma Atomic Emission Spectroscopy [14] and analysis of metals in drinking water by atomic emission spectroscopy utilizing inductively coupled plasma is described in EPA 200.7 [15]. Ion chromatography may also be used but requires post-column determination using a chelating agent and is detected with a suitable visible absorbency detector. 14.3.1.2 XRF Analysis of Integrated Samples

In the XRF analysis of an integrated sample, a sample is irradiated with X-rays of an appropriate energy to excite, through ionization, the elements of interest in the sample. Upon spontaneous de-excitation, sample elements emit fluorescent radiation, the component energies of which are characteristic for each element, and the intensities of which may be correlated with the elemental concentrations in the sample. The concentrations of particulate or dissolved metals, or both, in water streams are determined through accumulation on appropriate collection media (filters or ion exchange materials) and detection by x-ray fluorescence spectroscopy, providing real time determination of iron and other metals found in water streams. The water sample delivered into the monitoring system passes 14-3


through a flow sensor, and then, to a flow cell assembly containing a membrane or resin filter, depending on the application of interest. For an application where only dissolved metals are to be analyzed, the sample needs to be filtered upstream of the sample chamber to prevent particulate contamination of the resin membrane surface. A sample bypass valve is used for flow control through the sample chamber. Two sample chambers in sequence can be used to determine both particulate and dissolved components of the metal(s) of interest. X-ray fluorescence is used to determine the concentration of the captured material. XRF analysis gives a measure of total elemental concentration independent of the oxidation state or molecular configuration of the element. Elements with atomic numbers 13 through 92 can be detected. The filter chamber is essentially a variation of the traditional corrosion product sampler used to collect integrated samples. The main difference in the design of the flow cell in the on-line monitor is that the sample enters the filter chamber in a way that allows an x-ray probe to be positioned in close proximity to the filter or resin membrane surface. Since even a small quantity of water covering a sample significantly attenuates the excitation and emission radiation, a computer controlled valve switching system is incorporated into the monitor. In one position, this valve allows sample flow to proceed through the monitoring unit and metals to accumulate on the filter or resin membrane while the total flow is monitored. In the other position, the valve introduces air or other another gas (e.g., nitrogen) to purge the filter chamber of liquid while the sample is diverted to waste. It is during the air purge that the x-ray measurement takes place. In this way, the monitor operates by continuously alternating between two modes: a sample accumulation mode and an analysis mode. Typical time assignments for these modes for sample concentrations in the low µg/L (ppb) range are five minutes each; thus, in one cycle, sample accumulates for five minutes followed by a five minute x-ray measurement. With various delays for valve switching operations, computer extraction of x-ray data, and data manipulation, the measurement cycle in this case lasts approximately 14 minutes. Single-channel instruments can be configured to monitor iron, copper, or any other solid species. A two-channel instrument (with one sample chamber per channel) can provide greater monitoring flexibility. For instance, one channel can be equipped with an iron-probe that contains curium-244 while the second channel is equipped with a sulfur-probe (for lighter elements) that contains iron-55. Alternately, both channels can use iron-probes so that one channel can be collecting a sample while the other channel is analyzing for iron [16]. 14.3.2 Turbidity

Turbidity is defined by ASTM as “the optical property that causes light to be scattered and absorbed rather than transmitted in straight lines through the sample”. In simple terms, the interaction between light and suspended particles in a liquid is called turbidity. This interaction is affected by the size, shape, and refractive index of the particles, and the wavelength of the incident light. The colors of the suspended material and the sample fluid also affect the measurement. Turbidity is an “index” of water clarity. A common method used for turbidity measurement is nephelometry in which a detector measures light reflected, or “scattered” from 14-4


particles at an angle 90 degrees to the incident (source) light. In liquids with only a few particles per milliliter, or if the particles are only slightly reflective, this is a very small signal. Since metal oxide particles are usually dark they absorb rather than reflect light so nephelometry is not a preferred method for this application. 14.3.3 Particle Counter

Particle counters detect particles in liquids, determine their size, and put them into “bins” of defined size ranges. There are various optical and electrokinetic methods used in particle counting. One of the simplest and most sensitive for particles above 1 micron is light blocking, or “extinction” (Figure 14-1). The light source is a laser diode. As sample flows through the sensor, particles create shadows on the detector which are converted to electrical (millivolts (mV)) “blips” proportional to the particle size. A high speed microprocessor interprets the data and directs information to “bins” of selected size ranges, e.g. 2-5 microns, 5-10 microns, etc. Since this is a continuous sample flowing at a constant rate, the data can be mathematically expressed in particles per milliliter, or “counts per mL” for each size bin. Particle sizes are expressed as diameters, although it’s obvious that all particles aren’t spheres, or even circular. In many practical applications this doesn’t detract from the usefulness of the data since an assumption can be made that size “distribution” would be the same if all particles were spherical or circular. A great advantage for light blocking counters in boiler cycle water is that the particles are usually opaque and dark making very distinct, easily detectable “shadows”. On line particle counters are being used in potable and industrial applications and are capable of detecting one particle per milliliter of 2 micron size [20].

Figure 14-1 Particle Counter Schematic, Illustrating 1 Particle at 3 microns and 2 Particles at 1 micron [21] Source: Adapted from reference 21

14-5


14.3.4 Dynamic Light Fluctuation Based Particle Monitors

The particle monitor (PM) operates on the principle of “dynamic light fluctuation” (DLF), which is distinctly different from, the particle counter [20]. Figure 14-2 is a simple schematic of the measuring cell. As particles pass through the cell they cause disruptions, or “fluctuations” in intensity of the infrared LED light beam. More particles increase the magnitude of the fluctuations. The PM derives information on particulate concentration from the magnitude of these fluctuations, not from the total intensity of scattered or blocked light as in turbidimeters or particle counters. This information is converted to a reading proportional to the particulate concentration, and is displayed as the “particle index” (PI). The instrument does not measure particle size or number, but tracks “trends” in particulate concentrations. Sensitivity is equal to, or better than the particle counter. A unique feature of particle monitors is that the water sample never contacts the sensor cell optical components. Maintenance consists of simply replacing the plastic sample tubing. Also the PM does not require “calibration”, but uses a simple procedure that verifies instrument “sensitivity and repeatability”. The PM has the capability of detecting particulate concentration as low as one µg/L (ppb) and the capability of automatically obtaining a sample when the PI exceeds a predefined “threshold”. This enables plant personnel to analyze for specific contaminants and particulate concentrations when an “excursion” occurs [21].

Figure 14-2 Dynamic Light Fluctuation Schematic [21] Source: Adapted from reference 21

14.3.5 Acoustic Detection

This technique involves the use of a transducer that transmits a focused, ultrasound beam into the liquid containing particles. As this energy pulse strikes the corrosion product particles, some of it is reflected back to the transducer where it is detected. Each pulse lasts for 1 microsecond and 14-6


200 pulses per second are transmitted. This allows about 500 microseconds to receive the reflected echoes to “count” particles. The instrument’s output consists of a digital meter, optional bar graph, and current or voltage process outputs. The output is a relative concentration indication. It is simply referred to as “counts,” but may often be correlated to mg/L (ppm), µg/L (ppb), percent solids, or some other unit meaningful to the user. Some newer instruments provide enhanced performance via computer control and advanced signal processing. Primary differences for these monitors include independent bubble and particle response, and particle size compensation [22]. 14.3.6 Colorimetric

Soluble copper concentration can be determined colorimetrically. Colorimetric determination requires the addition of chemical reagents to provide development of the proper color species. To operate properly, this instrument needs to add the reagents in the proper sequence and amount, assure adequate mixing and necessary reaction times, accurately determine the amount of light absorption at the proper wavelength, correlate this light absorption to some standard curve, and then prepare for the upcoming sample. The colorimetric on-line copper analyzer performs a batch-wise analysis of the sample using a chemical approach that is based on the Bathocuproine method [17,23]. Cuprous ions form a water soluble orange colored chelate with bathocuproine disulfonate. The sample is buffered at a pH of about 4.3 and reduced with hydroxylamine hydrochloride. The absorbance is measured at 484 nm. Soluble iron is determined colorimetrically by two methods. The first is based on the reaction of ferrous ions with a 1,10 phenanthroline reagent at a pH of 3.2 [34]. Ferric ions can also be determined with this method if a preliminary chemical reduction step is performed. Reagents such as hydroxylamine or ascorbic acid are used to convert ferric to ferrous ions. The orange-red complex formed by this reaction is then measured at 510nm. On-line iron analysis based on this method is available [25] but the 3 µg/L (ppb) detection limit gives marginal suitability for condensate / feedwater applications. Test kits for grab sample total iron determinations are also available using the phenanthroline color complex [26]. The stated iron detection limit for the test kit is 20 µg/L (ppb). A second analytical method is based on the addition of ammonium thioglycolate in an acidic pH to form a red / purple complex [27]. Color intensity is measured at 562 nm with a reported detection limit of 9 µg/L (ppb) and a working range to 1400 µg/L (ppb). The colorimetric test is specific to ferrous iron and a digestion step must be added for conversion of ferric iron; after the digestion, test results are given as total iron and speciation is determined by difference. On-line colorimetric determination for iron is not currently available for the concentration of interest in power plant condensate and feedwater applications. 14-7


14.3.6.1 Colorimetric Limitations

All colorimetric tests rely on the principle of Beer’s Law (also known as the Beer-Lambert Law) which states that the amount of light absorbed by a sample (A) is proportional to some absorbtivity constant (a), the path-length of light (b), and the concentration of the analyte species(c). A = a x b x c The absorbtivity constant is dependent on the color species and test conditions of the method and remains constant as long as the reagents, test conditions, and wavelength of light do not change. The path length (the amount of sample through which the light passes) is also constant for a given sample cell. The mathematical formula then infers that the amount of light absorbed by a sample (i.e. the intensity of the color developed in the test) is directly proportional to the concentration of the analyte—in this case, copper. As the copper concentration increases, the intensity of the orange color will increase and the color change can be quantified. Limitations on this analytical principle arise at both the low end and high end of the useable range. The low end limit is caused by the detection limit of the test—this is primarily a function of the photo-multiplier (detector) sensitivity and stability. How small a change can be detected by the sensor and how stable is the baseline (or the zero reading)? When trying to quantify a concentration near the detection limit, it is not uncommon to see negative readings or duplicate readings that have large relative errors although the absolute error may only be 10 or 20 µg/L (ppb). As a result, most copper analyzers will list a detection limit, an accuracy limit, and a precision limit. This accuracy limit is typically an absolute value or some percentage of the reading, which ever is bigger. A typical analyzer has the following specifications which demonstrate this technical consideration [18]. •

Minimum Detection Limit: 1 µg/L (ppb).

•

Accuracy: ±2% of reading (µg/L).

•

Precision: ±2% of reading (µg/L).

Limitations on the high end of the analytical range, shown in Figure 14-3, arise from a property called self-absorption. As the color intensity becomes more and more blue, the amount of light coming into the sample cell is almost totally absorbed.

14-8


Figure 14-3 Illustration of Self Absorption at High Concentrations

At some point there is 100 percent absorption, the line flattens out, and the analytical test is no longer usable. However, the area where linearity first starts to deviate is also an area of concern. The molecules imparting color (absorbing light) are so numerous that some molecules are shaded from the incident light and not accounted for. The absorption coefficient can change at high analyte concentrations and stray light due to monochrometer imperfections can also skew the linearity. The subsequently curved line can no longer be used for accurate analytical determination. Most instruments therefore specify a maximum concentration that can be measured accurately corresponding to the top portion of the straight-line relationship.

14.4 Calibration 14.4.1 Wet Chemistry Analysis

Membrane filter standards for iron are prepared from an atomic absorption standard for iron. Alternatively, users may make up their own standard solution. The atomic absorption (AA) standard is 1000 mg/L (ppm) iron dissolved in dilute acid solution. Sodium hydroxide or ammonia solution is used to adjust measured aliquots of this standard to a pH of 10 to 11. Under these conditions, iron is insoluble and filterable. The resulting iron suspension is filtered through a 0.45 micron membrane filter and the filtrate analyzed by AA after appropriate acidification and digestion. Any iron determined in the filtrate is subtracted from the amount contained in the volume passed through the filter to determine the amount deposited on the filter. The aliquots of AA standard may be diluted with high purity water prior to pH adjustment. The dilution provides more conveniently handled volumes, and the larger volumes will promote more uniform distribution of iron on the filter surface. The amount of dilution does not affect the total iron to be filtered. 14-9


This procedure is repeated to prepare filter standards covering a range of total iron from zero (blank filter) to three milligrams. The lowest prepared standard should be near the minimum detectable amount (approximately 3 µg) distributed over the surface of a membrane filter with a diameter of 47 mm. This procedure, without adjustment to high pH, may be used to prepare cation resin membrane standards covering a range of total dissolved iron. For colorimetric analysis a series of standards are prepared and analyzed to determine the calibration curve for the on-line analyzer. On-line copper analyzers are equipped to automatically analyze standards and determine the calibration curve to be used in the on-line instrument. 14.4.2 XRF Analysis

Filter or resin membrane standards are analyzed using the on-line x-ray probe to collect intensity data under the exact conditions of on-line service. This includes installing the filter or resin membrane in the flow chamber of the on-line system and analyzing through the Kapton® window. The intensity data collected in this way is plotted against the known amounts of total iron and the equation for the best line through the data is determined by standard least squares regression techniques. For up to approximately 3000 µg/L (ppb) of total iron, the relationship is nearly linear. The resulting calibration equation is then used to convert intensity measurements of filters or resin membranes to iron mass during on-line operation. The calibration curve parameters may be inputted to the x-ray control electronics unit or directly to the PC depending on the software configurations of the monitor. In either case, the intensity measurements during an on-line session are converted automatically to mass values and both are stored in the data file, which is updated continuously during operation. This mass measurement (in micrograms, µg) is used with the incremental volume total (L) between measurements to give the desired concentration in µg/L (ppb).

14.5 Calibration Check For those devices which use off-line wet chemistry analysis of collected particulates, the validation should be on the analytical method (rather than the sample collection technique) and is typically defined by the lab’s QA/QC procedures. For the device that incorporates an on-line XRF monitor, validation is accomplished by removing the particulate-laden filter and conducting off-line analysis using a laboratory XRF system. For XRF analytical capabilities should be checked periodically to demonstrate calibration stability. One method exists for verifying instrument stability; the Line Method [19]. For the Line Method, a calibrated separate XRF instrument is used to analyze the same sample as the installed on-line XRF instrument. The results of two instruments are compared to the acceptance 14-10


criteria (e.g., the differences agree within ± 3 sigma or the measured values agree within ± 10%). Provided the on-line analyzer agrees within the acceptance criteria, the on-line instrument’s calibration is deemed to be acceptable. If the results are outside the acceptance criteria the online instrument should be recalibrated.

14.6 Alternate Methods Historically, the quantity of particulate matter has been estimated by comparison with a set of filter pad “standards”. The test protocol was developed by Babcock and Wilcox, a boiler manufacturer, to predict the level of “cleanliness” in the condensate / feedwater cycle during plant start-up. Two series of test pads with increasing particulate loading were created; one using red iron oxide (i.e. hematite) and another using black iron oxide (i.e. magnetite) material. A photograph showing this sequence of increasing particulate loading was then used to as a comparison against the analytical sample. This comparison was used to monitor start-up and predict the likely corrosion product transport.

14.7 End User Considerations The performance characteristics (range of measurement, accuracy, precision, bias, drift, response time, and signal change) for the monitoring equipment, provided by the manufacturer or supplier should be considered when selecting a suitable on-line iron and copper instrument. End users should select the instrument that is best suited for the intended analytical application. Integrated corrosion product samplers are best suited for a comprehensive monitoring program. The main emphasis is on quantifying the filterable material although some users also install the ion exchange membrane. Other on-line iron and copper instrument considerations include: •

Analytical capability to digest and analyze samples from an integrated sampler.

•

Need for continuous or semi-continuous monitoring.

•

Need for speciation.

•

Suitability of method to meet the defined accuracy, precision, bias and drift requirements.

•

Ease and robustness of calibration for the intended use.

•

Ease and robustness of calibration verification.

14-11


14.8 References 1.


2.


3.


4.


5.

Interim Consensus Guideline on Fossil Plant Chemistry, by A.F. Aschoff, Y.H. Lee, D.M. Sopocy, and O. Jonas. EPRI, Palo Alto, CA: June 1986. CS-4629.

6.

Guideline Manual on Instrumentation and Control, by R.D. Hopkins, E.H. Hull, K.J. Shields, and S. Yorgiadis. EPRI, Palo Alto, CA: April 1987. CS-5164.

7.

Monitoring Cycle Water Chemistry in Fossil Plants, Vol. 1, by A.F. Aschoff, D.M. Sopocy, D.T. Eglar, O. Jonas, J.K. Rice, C.C. Stauffer, and W.E. Allmon. EPRI, Palo Alto, CA: October 1991. GS-7556, Volume 1.

8.

Connolly, D. and S. Harvey. “On-Line Corrosion Product Monitoring of Nuclear Plant Feedwater”, Ultrapure Water, November 1995, pp. 28-31.

9.

PWR Secondary Water Chemistry Guidelines: Revision 4, prepared by PWR Water Chemistry Guidelines Revision Committee. EPRI, Palo Alto, CA: November 1996. TR102134-R4.

10. ASTM D1068-05, “Standard Test Methods for Iron in Water”, American Society for Testing & Materials, Philadelphia, PA. 11. ASTM D1688-02, “Standard Test Methods for Copper in Water”, American Society for Testing & Materials, Philadelphia, PA. 12. Iron, Method 236.2 (AA Furnace), Methods for Chemical Analysis of Water and Wastes, EPA-600/4-79-020. U.S. Environmental Protection Agency, Cincinnati, Ohio, (Revised March 1995). 13. Copper, Method 220.2 (AA Furnace), Methods for Chemical Analysis of Water and Wastes, EPA-600/4-79-020. U.S. Environmental Protection Agency, Cincinnati, Ohio, (Revised March 1995). 14-12


14. ASTM D4190-03, “Standard Test Methods for Elements in Water by Direct-Current Argon Plasma Atomic Emission Spectroscopy”, American Society for Testing & Materials, Philadelphia, PA. 15. Inductively Coupled Plasma, Method 200.7 (AA Furnace), Methods for Chemical Analysis of Water and Wastes, EPA-600/4-79-020. U.S. Environmental Protection Agency, Cincinnati, Ohio, (Revised March 1998). 16. Connolly, D. and P. Millett. “On-Line Particulate Iron X-Ray Monitor”, Ultrapure Water, February 1994, pp. 61-65. 17. Part 3500-Cu C, Bathocupric Method, Standard Methods for the Examination for Water th and Wastewater, 20 Edition, 1999, American Public Health Association, Washington, DC. 18. Waltron, LLC, Aqualert Division, 9040 Series Copper Analyzer, product information manual, Whitehouse, NJ. 19. ASTM D3864-96(2000), “Standard Guide for Continual On-Line Monitoring Systems for Water Analysis”, American Society for Testing & Materials, Philadelphia, PA. 20. Chemtrac PC 2400D Particle Counter and PM 2500 XRD Particle Monitor, Chemtrac Systems, Inc. Norcross, GA 30092. 21. “Continuous Metal Transport Monitoring using On-Line Particulate Determination”, L. Joseph Hancock, Richard A. Breckenridge—Arizona Public Service Company, presented at Eighth International Conference on Cycle Chemistry in Fossil and Combined Cycle Plants with Heat Recovery Steam Generators, Calgary, Canada 2006. 22. Acoustic Particle Counter, Innovative Dynamics Inc, Ithaca, NY. 23. Waltron µAI9045 On-Line Copper Analyzer, Waltron LLC, Whitehouse, NJ. 24. Standard Methods for the Examination of Water and Wastewater, 21st Edition. Method 3500 Fe B. Phenanthroline Method, 2005. 25. Model CFA-1026 On-Line Iron Analyzer, Scientific Instruments, 200 Saw Mill River Road, Hawthorne, NY 10532. 26. Hach Method 8008 FerroVer® Powder Pillows, Hach Company, Loveland, CO. 27. Hach Method 8147 FerroZine® Reagent, Hach Company, Loveland, CO.

14-13


15 PHOSPHATE

15.1 Purpose and Use Phosphate is listed as an EPRI Guideline Core Monitoring Parameter [1,2]. The on-line monitoring equipment is used most frequently to monitor phosphate in the continuous blowdown of fossil boilers. Optimal phosphate concentration is both a function of the operating pressure of the steam/water cycle and the likelihood of boiler contamination. Phosphate is continually monitored on-line in the plant for the following reasons: •

To check the accuracy of water chemistry control (such as the sodium-to-phosphate molar ratio).

•

To facilitate the correlation of phosphate content with plant operating variables.

•


The data generated by continuous on-line monitoring of phosphate is used by plant chemistry and operations department personnel. The goal for plant personnel is to maintain phosphate within prescribed limits.

15.2 Description of Method Cycle chemistry guidelines for fossil power plants indicate that, when phosphate is used for pH control, the sodium-to-phosphate molar ratio should be controlled to minimize corrosion rates and scaling of heat transfer surfaces [1]. The resulting need for continuous surveillance of phosphate (and sodium) levels has led to the introduction of several on-line monitors. One relevant technique is on-line ion chromatography: further details of this approach can be found in Section 13. More frequently, an on-line colorimetric technique is used to monitor phosphate. Although the details of the method vary from one manufacturer to another, they all result in reaction of the phosphate with reagents to produce a colored solution. The intensity of the color, which is proportional to the concentration of phosphate, is then measured with a spectrophotometer. The colorimetric methods are based on reactions that are specific to the orthophosphate ion so other forms of phosphorus (organic phosphorus and such hydrolyzable forms as metaphosphate, pyrophosphate, and tripolyphosphate) will not be detected by this method. It is possible that a small percentage of the hydrolyzable phosphorus is converted to orthophosphate when an acidic reagent is added to the sample during the analysis procedure. 15-1

EPRI Proprietary Licensed Material Phosphate

Several manufacturers choose to add acidified ammonium molybdate and a reducing agent to produce a blue phosphomolybdate complex. Others add an acidified mixture of ammonium molybdate and ammonium metavanadate to produce a yellow molybdovanadophosphoric acid complex. The absorbance of the blue solution is measured with a spectrophotometer at 880 or 625 to 670 nm depending on the details of the method. In general the blue phosphomolybdate complex is a more sensitive method than the yellow molybdovanadophosphoric acid method. Similarly, the absorbance of the yellow solution is measured at 400 to 480 nm depending on the details of the method. One manufacturer uses the blue phosphomolybdate method for measuring phosphate in the low range (0-5000 µg/L (ppb) as PO4) and the yellow molybdovanadophosphoric acid method for measuring phosphate in a higher range (0-50 mg/L (ppm) as PO4). The on-line analysis methods are similar, though not identical, to the laboratory analysis techniques described in ASTM D515 Method A [3] (blue complex) and the now discontinued Method C [4] (yellow complex). Similar references are found in Standard Methods for the Analysis of Water and Wastewater 4500-P C (yellow complex) and 4500-P D (blue complex) [5]. However, unlike laboratory units, on-line analyzers are equipped with automatic fluid handling systems. A sample bypass flow system is provided so that the on-line analyzer has continuous access to a fresh sample of the water containing phosphate. The water sample is typically conditioned so that the pressure and temperature are within set limits before it enters the analyzer. Necessary reagents and standards are often purchased from the instrument supplier which eliminates the risk of field contamination during the preparation process. An instrument that uses the molybdovanadate reagent is described below as an example of the methods available. At the beginning of each cycle, the measurement cell is flushed thoroughly with fresh sample. Sulfuric acid is added to the sample to dissolve particulate matter and an anionic surfactant is added to minimize bubble formation on the measuring cell walls. An initial sample blank absorbance is measured at 480 nm to serve as a zero reference, and then the molybdovanadate reagent is added to react with the orthophosphate to form the yellow molybdovanadophosphoric acid complex. The sample is then allowed to stand undisturbed for a pre-set time to allow for proper color development. The light absorbance through this colored solution is measured at 480 nm. The difference between this absorbance and the initial blank absorbance is converted to the orthophosphate (PO4) concentration and displayed as mg/L (ppm) PO4. Silica is a potential interference in this analysis, but only at concentrations not typically found in fossil power plant samples, i.e. above 500 mg/L (ppm). Sulfide present in the sample can also interfere with quantification. Sulfides are not typically found in fossil power plant samples.

15-2


15.3 Technical Considerations 15.3.1 Colorimetric Limitations

All colorimetric tests such as phosphate rely on the principle of Beer’s Law, as shown in Equation 15-1 (also known as the Beer-Lambert Law). This law states that the amount of light absorbed by a sample (A) is proportional to some absorption constant (a), the path-length of light (b), and the concentration of the analyte species (c). A=axbxc

Equation 15-1

The absorption constant is determined by the color species and test conditions of the method and remains constant as long as the reagents, test conditions, and wavelength of light do not change. The path length (the amount of sample through which the light passes) is also constant for a given sample cell. The mathematical formula infers that the amount of light absorbed by a sample (i.e., the intensity of the color developed in the test) is directly proportional to the concentration of the analyte—in this case, phosphate. As the phosphate concentration increases, the intensity of the color will increase and this color change can be quantified as illustrated in Figure 15-1.

Figure 15-1 Typical Beer’s Law Absorption (A) vs. Concentration Curve

Limitations on this analytical principle arise at both the low end and high end of the useable range. The low end limit is caused by the detection limit of the test—this is primarily a function of the photo-multiplier (detector) sensitivity and stability. How small a change can be detected by the sensor and how stable is the baseline (or the zero reading)? When trying to quantify a test right at the detection limit, sometimes negative readings or duplicate readings are seen that have large relative errors although the absolute error may only be a fraction of a mg/L (ppm). As a 15-3


result, most phosphate analyzers will list a detection limit, an accuracy limit and a precision limit. This accuracy limit is typically an absolute value or some percentage of the reading. A commonly used analyzer has the following specifications which demonstrate this technical consideration [6]. •

Minimum Detection Limit: 0.2 mg/L (ppm).

•

Accuracy: ±0.5 mg/L (ppm) or ±5% of reading, whichever is greater.

•

Precision: ±0.5 mg/L (ppm) or ±5% of reading whichever is greater.

Evaluation of this specification shows that the accuracy uncertainty of the measurement is large relative to the measurement itself at or near the minimum detection limit. At lower concentrations (0.2-1.0 mg/L (ppm)) end users are cautioned about using data for critical decision making since the inaccuracy is large relative to the measurement itself. At concentrations greater than 10 mg/L (ppm) the accuracy limit of ± 5% is used since it becomes larger than ± 0.5 mg/L (ppm). A different analyzer may have a different specification with a lower detection limit as a result of using another analytical method (changing the absorption constant) and using a longer path length sample. At some point there is 100 percent absorption, the line flattens out, and this analytical approach is no longer usable. However, the area where linearity first starts to deviate is also an area of concern. The molecules imparting color (absorbing light) are so numerous that some molecules are shaded from the incident light and not accounted for. The absorption coefficient can change at high analyte concentrations and stray light due to monochrometer imperfections can also skew the linearity. The subsequently curved line can no longer be used for accurate analytical determination. Most instruments therefore have a maximum analytical range which corresponds to the top portion of the straight-line relationship. Analytical readings above this linear region are disallowed and the analyzer produces an “over range” alarm. 15.3.2 Sample Considerations

Process samples for phosphate analysis need to be representative of the sample stream being measured. As such, proper flushing of sample lines, temperature conditioning, particulate removal, and pressure conditioning are all required. Many analyzers have a failsafe feature which senses a loss in sample flow and triggers an alarm. In addition, because the sample analysis is based on a color development technique, proper flushing of the sample cell is required between samples to remove the previous test solution. This cell flushing step is usually accomplished by regulating the incoming sample pressure and allowing a predetermined time to completely exchange the new sample for the prior test solution. Inlet sample pressure is often regulated at 3.45 kPa ± 2.07 kPa (5±3 psi) to assure adequate flushing; sample pressure control becomes an issue when process stream pressure changes due to load changes or sample sequencing.

15-4


15.3.3 Time Delay

Another technical consideration is the time lag required for reagent addition and color development which necessarily results in a batch type process. A typical phosphate analyzer has a nominal ten minute cycle from introduction of one sample to the next. This analysis delay could be troublesome if the process stream phosphate concentration is changing rapidly. Fortunately, in the typical fossil plant applications, the phosphate content is a slowly changing value and the batch process concept does not significantly hinder its usefulness. The amount of time required for color development is temperature dependent. Some colorimetric instruments try to account for varying temperatures by adding sample heaters or by modifying the delay time set aside for color development as a function of sample temperature. Most phosphate analyzers make allowances for this color development time and will accept inlet sample temperatures in the nominal 5-50°C (41-122°F) range. 15.3.4 Sample Temperature

Further sample temperature consideration must be given if the sample temperature is below the dew point of the surrounding environment. Under these conditions, moisture can condense on the surface of the sample cell, which will lead to unstable results. Many instruments are designed to accept an instrument air purge stream to lower the relative humidity of the sample compartment. Any leaks in the sample line or reagent lines within the compartment must also be immediately fixed to prevent condensation of water vapor on the sample cell. 15.3.5 Sample Volume and Introduction of Chemicals

A fixed volume of sample is required for the color determination. This is provided by a sample cell of known volume or a known volume sample loop (small bore tubing). The individual chemical reagents can either be added by a mechanical pump, an eductor which creates a low pressure region in the flowing liquid and sucks the chemical in, or through a pressurized system, which admits the reagent through a solenoid valve. These chemical reagents must be replenished on a regular basis—most instruments are designed for a 30-40 day supply of chemicals. Intimate mixing of the chemical and the sample is attained with either a magnetic stirrer or a turbulent flow mixing device in a loop flow technology. The proper wavelength of incident light is obtained by an interference filter and the differential light absorption is determined by a photomultiplier detector. This photometric response is then compared to a calibration curve derived using samples of known phosphate concentration. 15.3.6 Light Intensity

Many instruments also measure the intensity of light that passes through the sample before the color developing reagent is added (i.e. during the “zero reading”) to assure some minimum value

15-5


is obtained. If this light intensity is not adequate, an error message is often displayed that directs to user to “clean sample cell” or “replace lamp”.

15.4 Interferences 15.4.1 Particulate Matter

All colorimetric tests must also account for particulates in the sample which result in light scatter across the sample cell. The particulates cause an opacity or cloudiness in the sample which can hinder the analytical measurement. This light scatter can be misinterpreted as light absorption which will lead to erroneously high phosphate content. Most instruments try to minimize this interference by taking a “zero reading” or “base-line reading” of the sample immediately before adding the color developing reagent. This is certainly a valid and desirable step in the measuring process. However, the analytical reading is usually made after the sample has been allowed to stand undisturbed for some length of time. If particles that were suspended at the time of the “zero reading” have settled at the time of the analytical reading, an erroneous quantification will occur. Particulates can thus lead to inconsistencies in analytical results. While instruments claim their ability to not have particulate matter interference, the user should make every effort to filter particulates before doing the colorimetric test. 15.4.2 Sample Discoloration

Sample discoloration can also affect on-line phosphate analyzers. While establishing the “zero reading” as mentioned above tends to cancel out the effect of background color, discoloration in the same approximate wavelength (e.g. blue or yellow) can shift the calibration line and limit the maximum range of quantification. The detector senses that the instrument is no longer in the straight line region of the calibration curve and refuses to quantify the result; it gives an “overrange” message which disallows the test. 15.4.3 Other Substances

Positive interference is caused by silica and arsenate only if the sample is heated. Negative interference is caused by arsenate, fluoride, thorium, bismuth, thiosulfate, thiocyanate, or excess molybdate. Blue color is caused by ferrous ion but this does not affect results if ferrous ion concentration is less the 100 mg/L (ppm). Another potential interference in the molybdovanadate based phosphate test is the presence of sulfide. No provision is made for counteracting this negative interference in the on-line instrument but the presence of sulfide in most process streams should not occur. A bench test with a Bromine Water pre-step should be used as a cross check if sulfide is suspected to quantify the level of the interference [7].

15-6


15.5 Calibration On-line phosphate monitoring instruments have both manual and automatic calibration capabilities. A typical automatic calibration routine is to have a weekly verification using an internal standard. Frequently, this calibration uses a standard purchased from the instrument manufacturer/supplier. The automatic calibration is usually an analysis of a single standard within the linear range according to the Beer-Lambert Law to establish the response slope. The second point of the calibration line is the x-y origin prior to addition of the color development reagent. The standard concentration is traditionally rather high on the absorbance / concentration curve to assure stability of the standard. This can lead to accuracy issues if the analytical result of the process stream is only 1 or 2 % of the standard concentration. A recommended second step would be to introduce a freshly-made standard into the instrument closer to the analytical value of the process stream for accuracy verification. Several of the on-line phosphate monitors have provision to analyze a grab sample or verification sample independent of the internal solution which is used for the automatic calibration. These on-line results are then compared to those obtained from standard laboratory analysis techniques (such as ASTM D515) [3].

15.6 Calibration Check On-line phosphate instrument analytical capabilities should be checked periodically to demonstrate calibration stability. Two methods exist for verifying instrument stability; the Standard Injection Method [8] or the Line Method [9]. For the Standard Injection Method, a known standard solution, near the mid-point of the calibration curve, is analyzed by the on-line instrument and the results are compared to the acceptance criteria. Acceptance criteria are either based on statistically derived limits. i.e., ± 3 sigma or based on some predetermined limits established from experience, i.e., ± 10%. Provided the on-line analyzer agrees within the acceptance criteria, the on-line instruments calibration is considered still acceptable. If the results are outside the acceptance criteria, the on-line instrument must be recalibrated. For the Line Method, a calibrated separate phosphate analyzer, typically a bench top analyzer using ASTM D515 is used to analyze a sample from the same sample stream as the installed online instrument [3]. Provided the bench top analyzer agrees within the acceptance criteria (e.g., matches the results of the on-line analyzer within ± 3 sigma or ± 10%) the on-line instruments calibration is considered still acceptable. If the results are outside the acceptance criteria the online instrument must be recalibrated.

15.7 Alternative Methods While different manufacturers use slightly different color development reagents and protocol (adding ammonium molybdate and a reducing agent to produce a blue phosphomolybdate complex or adding an acidified mixture of ammonium molybdate and ammonium metavanadate 15-7


to produce a yellow molybdovanadophosphoric acid complex) on-line phosphate analysis is typically done by colorimetric comparison. It should be noted that ion chromatography (IC) is also a widely recognized technique for anion analysis and phosphate content may be determined by this technique [10]. From a practical standpoint, ion chromatography is a more sophisticated analytical technique which relies on a partitioning of the various anions in a sample and a sequential arrival of each anion at the detector over a period of time. The technique can be highly sensitive with detection levels in the 100 µg/L (ppb) range for phosphate. IC would most likely be used for on-line phosphate determination when several trace anions were of analytical interest (e.g. acetate, formate, chloride, and sulfate) and obtaining a phosphate value would be incidental in the scan. The elution time (delay time from sample injection until arrival at the detector) for phosphate is typically from 10 to 30 minutes depending on chromatographic column conditions. Similarly, the phosphate concentration in most boiler samples is one to two orders of magnitude greater than the trace ions of analytical interest. Analytes of widely differing concentrations can often be a problem in IC as the high concentration causes severe tailing, destabilizes the column, and requires a long rinse time. Ion chromatography continues to be developed for the electric utility industry and may fill special needs in anion quantification.

15.8 End User Considerations The performance characteristics (quantification range, accuracy, precision, bias, drift) and design characteristics (cycle time, selection of reagents, reagent consumption, sample manipulation, sample conditioning, and chemical interferences) for the phosphate monitor as provided by the manufacturer or supplier should be considered when selecting a suitable instrument. Other on-line phosphate monitor considerations for boiler water samples include: •

Inlet sample flow and pressure requirements.

•

Physical space requirements and mounting configuration (surface mounted or recessed into a sample panel).

•


•


•

Provision for instrument purge air/pressurization air if required.

•


15.9 Field Experience Several errors are most likely to occur with a colorimetric phosphate analyzer. They can be collected into sample delivery or analyzer malfunction errors.

15-8


15.9.1 Sample Delivery

•

All analyzers must rely on sample delivery at an adequate flow rate to assure adequate flushing between samples. The approach used by one manufacturer is to monitor the inlet sample pressure. Required inlet sample pressure is from 3.45 kPa ± 2.07 kPa (5 ±3 psi). Another manufacturer has a constant head system with an overflow weir in the sample reservoir. The reservoir also has a float switch installed to indicate a “no sample’ condition. Both designs effectively stop the analyses routine if sample flow is not detected. It is desirable to have an alarm contactor available to signal this “trouble” condition to the DCS and subsequently to the control room / lab. On systems without this relay, a phosphate analyzer can effectively sit idle for long periods of time and the malfunction will go unnoticed.

•

As discussed previously, the phosphate monitor attempts to cancel out the effects of suspended particulates in the sample by taking a “zero reading” immediately prior to color development. However, accumulation of particulates in the sample lines and colorimeter cells can cause very erratic and erroneous readings. Special precautions should be taken during cyclic operation and start-ups to filter the suspended material prior to the phosphate analyzer.


•

The most common problem with colorimetric phosphate analyzers has to do with liquid handling within the instrument. Depending on the analyzer design, the list of possible problems includes pinch valves that leak, sample lines that plug or develop pin holes, reagent lines that become unattached or develop leaks, eductors that become plugged and peristaltic pumps that develop leaks. Many of these malfunctions will be evidenced by liquid lying in the bottom of the analyzer or running out onto the floor. Less obvious malfunctions will be evident by the analyzer failing to calibrate, producing negative values, or yielding wildly fluctuating readings.

•

Reagents and standards must be replenished as needed. Reagent and standard consumption without appropriate replenishment may cause malfunctions in all designs. One design also has an internal reservoir of demineralizer water used for dilution of concentrated samples. Diligence is required to keep this dilution bottle filled to allow the analyzer to work properly.

Calibration and maintenance procedures and schedules are typically described in literature supplied with the monitoring equipment by each manufacturer. In addition, guidelines for maintenance and calibration activities are recommended in EPRI Report GS-7556 [11].

15-9


15.10 References 1.

Cycle Chemistry Guidelines for Fossil Plants: Phosphate Continuum and Caustic Treatment, EPRI, Palo Alto, CA:2004. 10004188.

2.

Cycle Chemistry Guidelines for Combined Cycle / Heat Recovery Steam Generators (HRSGs), EPRI, Palo Alto, CA: 2006. 1010438.

3.

ASTM D515-88, Standard Test Methods for Phosphorous in Water, Method A: Colorimetric Ascorbic Acid Reduction, American Society for Testing and Materials, Philadelphia, PA.

4.

ASTM D515-88, Standard Test Methods for Phosphorous in Water, Method C: Colorimetric Molybdovanadophosphate Method, (method discontinued in 1988) American Society for Testing and Materials, Philadelphia, PA.

5.

Standard Methods for the Examination of Water and Wastewater, 21st Edition, 2005.

6.

Hach Series 5000 High Range Phosphate Analyzer; Model 60001 Instrument Manual, Second Edition, Rev.2, 11/2000.

7.

Standard Methods for the Examination of Water and Wastewater, 21st Edition, 2005.

8.

Advanced Power Plant Chemistry QA/QC Practices, Scientech, LLC, Clearwater, FL, 2006.

9.

ASTM D3864-96(2000), “Standard Guide for Continual On-Line Monitoring Systems for Water Analysis”, American Society for Testing & Materials, Philadelphia, PA.

10. Application Note 146—Determination of Trace Anions in High Purity Waters by Ion Chromatography with the IonPac AS17 Using High Volume Direct Injection with the EG40, Dionex Corporation. 11. Monitoring Cycle Water Chemistry in Fossil Plants, Vol. 3 Project Conclusions and Recommendations, by A.F. Aschoff, D.M. Sopocy, D.T. Eglar, O. Jonas, J.K. Rice, C.C. Stauffer, and W.E. Allmon. EPRI, Palo Alto, CA: October 1991. GS-7556.

15-10


16 SILICA

16.1 Purpose and Use While silica is not listed as an EPRI Guideline Core Monitoring Parameter [1-3] for fossil plants it is widely monitored for diagnostic and troubleshooting purposes. It is included as a Core Monitoring Parameter for combined cycle plants [4]. Elevated silica concentrations in steam can lead to silica deposition in the low pressure turbine so it is desirable to control the silica content of boiler water. Silica content is also the primary indication of anion resin exhaustion in both makeup and condensate polishing equipment: under most conditions, an increase in silica content at the anion resin outlet will occur prior to a change in either pH or conductivity. Silica is most frequently monitored in the plant for the following reasons: •


•

To facilitate the correlation of a water chemistry parameter with plant operating variables, with an aim to optimizing operations.

•

To check the accuracy of water chemistry control (for silica), so ensuring that carry-over and deposit rates are kept at acceptable low levels.

•


The data generated by continuous on-line monitoring of silica is used by plant chemistry and operations department personnel. The goal for plant personnel is to maintain silica below prescribed limits.

16.2 Description of Method On-line silica analyzers make use of the heteropoly blue method—a colorimetric analysis procedure described in ASTM D859-05 [5] in which the soluble silica reacts with molybdate ion to form a greenish-yellow silicomolybdate complex, which in turn is converted to a blue complex by reduction with 1-amino-2-napthol-1-sulfonic acid or ascorbic acid. A spectrophotometer is used to measure the absorbance of this blue sample at a wavelength of about 815 nm. The difference between this absorbance and that of a reagent blank is directly related to the silica content of the sample. According to ASTM D859, the useful measurement range is 20 to 1000 µg/L (ppb) but manufacturers of on-line analyzers have extended this range to 0 to 5000 µg/L (ppb).

16-1

EPRI Proprietary Licensed Material Silica

The molybdate-reactive forms of silica measured by this method include dissolved simple silicates, monomeric silica and silicic acid, and an undetermined fraction of polymeric silica. These forms generally predominate in power generation cycles although non-reactive (see Section 16.4.2 Non-reactive Silica) colloidal (crystalloidal) silica is sometimes present in the make-up water supply. On-line silica analysis is similar to the laboratory analysis technique described in ASTM D859 but on-line analyzers are equipped with an automatic fluid handling system. A sample bypass flow system is provided so that the on-line analyzer has continuous access to a fresh sample of the water containing reactive silica. The analyzer transports the sample through the system with inlet sample pressure, with a peristaltic pump or with a positive displacement piston pump; reagents may be transported by similar pumps or by compressed air. The reagents are added sequentially in three stages and, at each stage, the solutions are mixed and are given a specific time to react before continuing to the next stage. Not all analyzers are equipped to null the effects of phosphate, so it is important to select a model that handles this interference if samples are likely to contain phosphate; for instance, for boiler water samples taken from drum units on phosphate treatment. As the sample enters the analyzer, it is mixed with ammonium molybdate solution and either hydrochloric or sulfuric acid. This acidic molybdate solution is allowed to react for a specific period (a few minutes) with any dissolved silica and phosphate to form molybdosilicic and molybdophosphoric acid complexes, respectively. For instance, hydrated silicic acid reacts with the acidified ammonium molybdate solution to form the molybdosilicic acid complex and ammonium sulfate byproduct as follows: H 8SiO 6 + 12(NH 4 ) 2 MoO 4 + 12H 2SO 4 → H8 [Si(Mo 2O 7 ) 6 ] + 12(NH 4 ) 2 SO 4 + 12H 2O Equation 16-1

Then oxalic acid is added and thoroughly mixed with the sample to halt the development of the molybdosilicic acid complexes and destroy any molybdo-phosphoric acid complexes. Citric acid may be used instead of oxalic acid, although it is not as efficient at masking the phosphate interference. A light absorbance measurement at a wavelength in the range of 810–820 nm is made on a reagent blank to provide a zero reference for the sample. Then 1-amino-2-napthol-1sulfonic acid or ascorbic acid is added to the sample to reduce the yellow colored molybdosilicic acid complex to a heteropoly blue colored solution. A second light absorbance measurement at the same wavelength is made and compared with the reference measurement. The difference between the two measurements determines the silica concentration, and the result is displayed directly in units of concentration (for instance, µg/L (ppb)). Phosphate, at concentrations >50 µg/L (ppb), is a primary interference in this analysis but it can be chemically eliminated by the addition of oxalic acid or citric acid.

16-2


16.3 Technical Considerations 16.3.1 Colorimetric Limitations

All colorimetric tests such as silica rely on the principle of Beer’s Law (also known as the BeerLambert Law) which states that the amount of light absorbed by a sample (A) is proportional to some absorption constant (a), the path-length of light (b), and the concentration of the analyte species(c). A=axbxc

Equation 16-2

The absorption constant is determined by the color species and test conditions of the method and remains constant as long as the reagents, test conditions, and wavelength of light do not change. The path length (the amount of sample through which the light passes) is also constant for a given sample cell. The mathematical formula then infers that the amount of light absorbed by a sample (i.e., the intensity of the color developed in the test) is directly proportional to the concentration of the analyte—in this case, silica. As the silica concentration increases, the intensity of the blue color will increase and this color change can be quantified. Limitations on this analytical principle arise at both the low end and high end of the useable range. The low end limit is caused by the detection limit of the test—this is primarily a function of the photo-multiplier (detector) sensitivity and stability. How small a change can be detected by the sensor and how stable is the baseline (or the zero reading)? When trying to quantify a test at the detection limit, sometimes a negative reading is observed or duplicate readings may have large relative errors. As a result, most silica analyzers will list a detection limit, an accuracy limit, and a precision limit. This accuracy limit is typically an absolute value or some percentage of the reading. A commonly used analyzer has the following specifications which demonstrate this technical consideration [6]. •

Minimum Detection Limit : <0.5 µg/L (ppb).

•

Accuracy: 0.00 to 500 µg/L; ± 1 µg/L (ppb) or ± 5% of reading, whichever is greater, 500 to 5000 µg/L (ppb); ± 7% of reading.

•

Precision: ±0.5 µg/L (ppb) or 5% of reading which ever is greater.

Evaluation of this specification shows that the accuracy uncertainty of the measurement is large relative to the measurement itself at or near the minimum detection limit. At lower concentrations (0.2-10 µg/L (ppb)) end users are cautioned about using data for critical decision making since the inaccuracy is large relative to the measurement itself. At concentrations greater than 20 µg/L (ppb) the accuracy limit of ± 5% is used since it becomes larger than ± 1 µg/L (ppb). A different analyzer may have a lower detection limit as a result of using another analytical method (changing the absorption constant) and using a longer sample cell (obtaining a longer path length).

16-3


Limitations on the high end of the analytical range arise from a property called self-absorption. As the color intensity becomes more and more blue, the amount of light coming into the sample cell is almost totally absorbed.

Figure 16-1 Graph Showing 100 Percent Absorption

At some point there is 100 percent absorption, the line flattens out, and the analytical test is no longer usable (see Figure 16-1). However, the area where linearity first starts to deviate is also an area of concern. The molecules imparting color (absorbing light) are so numerous that some molecules are shaded from the incident light and not accounted for. The absorption coefficient can change at high analyte concentrations and stray light due to monochrometer imperfections can also skew the linearity. The subsequently curved line can no longer be used for accurate analytical determination. Most instruments therefore have a maximum analytical range which corresponds to the top portion of the straight-line relationship. Analytical readings above this linear region are disallowed and the analyzer produces an “over range” alarm. 16.3.2 Sample Considerations

Process samples for silica analysis need to be representative of the sample stream being measured. As such, proper flushing of sample lines, temperature conditioning, particulate removal, and pressure conditioning are all required. Many analyzers have a failsafe diagnostic which senses a loss in sample flow and triggers an alarm. In addition, because the sample analysis is based on a color development technique, proper flushing of the sample cell is required between samples to remove the previous test solution. This cell flushing step is usually accomplished by regulating the incoming sample pressure and allowing a predetermined time to completely exchange the new sample for the prior test solution. Inlet sample pressure is often regulated at 3.45 kPa ± 2.07 kPa (5±3 psi) to assure adequate flushing; sample pressure control

16-4


becomes an issue when process stream pressure changes due to load changes or sample sequencing. 16.3.3 Time Delay

Another technical consideration is the time lag required for reagent addition and color development which produces a batch type process. A typical silica analyzer has a nominal 11 to 15 minute cycle time from introduction of one sample to the next. This analyses delay could be troublesome if the process stream silica concentration is changing rapidly. Fortunately, the silica content is a somewhat steady value and the batch process concept does not significantly hinder the test. The amount of time required for color development is temperature dependent. Some colorimetric silica instruments try to account for varying temperatures by adding sample heaters or modifying the delay time set aside for color development as a function of sample temperature. Most silica analyzers without heaters make allowances for this color development time and will accept inlet sample temperatures in the nominal 5-40 °C (41-104 °F) range. More rapid cycle time is obtained with a heater which elevates the incoming sample temperature to 30-50 °C (86122 °F) and reduces the cycle time to nine minutes. One manufacturer claims a six minute response time without the use of a heater with sample temperatures of 5-45 °C (41-113 °F). 16.3.4 Sample Temperature

Further sample temperature consideration must be given if the sample temperature is below the dew point of the surrounding environment. Under these conditions, moisture can condense on the surface of the sample cell which will lead to unstable results. Many instruments are designed to accept an instrument air purge stream to lower the relative humidity of the sample compartment. Any leaks in the sample line or reagent lines within the compartment must also be immediately fixed to prevent condensation of water vapor on the sample cell. 16.3.5 Sample Volume and Introduction of Chemicals

A fixed volume of sample is required for the color determination. This is provided by a sample cell of known volume or a known volume sample loop (small bore tubing). The individual chemical reagents can either be added by a mechanical pump, an eductor which creates a low pressure region in the flowing liquid and sucks the chemical in, or through a pressurized system which admits the reagent through a solenoid valve. These chemical reagents must be replenished on a regular basis—most instruments are designed for a 30-40 day supply of chemicals. Intimate mixing of the chemical and the sample is attained with either a magnetic stirrer or a turbulent flow mixing device in a loop flow technology. The proper wavelength of incident light is obtained by an interference filter and the differential light absorption is determined by a photomultiplier detector. This photometric response is then compared to a calibration curve of a known silica concentration. 16-5


16.3.6 Light Intensity Check

Many instruments also measure the intensity of light that passes through the sample during the “zero reading” to assure some minimum value is obtained. If this light intensity is not adequate, an error message is often displayed that directs the user to “clean sample cell” or “replace lamp”.

16.4 Interferences 16.4.1 Ortho-phosphate

The primary interference with the hetero-poly blue silica test is a false positive caused by the presence of ortho-phosphate ions. This can be especially troublesome when determining the silica content in a high pressure boiler sample since the silica content may be in the 100 µg/L (ppb) range and the phosphate content could be several mg/L (ppm). As discussed previously, either oxalic acid or citric acid is added to the reaction sequence to destroy the molybdophosphoric acid complex and eliminate this interference. Failure to add this chemical would cause a high bias in the silica readings in the presence of phosphate. 16.4.2 Non-reactive Silica

It’s important to remember that some forms of silica may not react with the color developing chemicals of the analyzer—the test determines only “reactive silica”. In at least one of its forms, silica does not react with molybdate even though it is capable of passing through a filter paper and does not make the water noticeably turbid. It is not known to what extent such “unreactive” silica occurs in water. Terms such as colloidal, crystalloidal, and ionic have been used to distinguish between the various forms of silica but the use of such terminology cannot be fully justified. “Molybdate-unreactive” silica can be converted to “molybdate-reactive” silica by heating or fusing with alkali. The significance of this chemical change can be troublesome in the power plant. “Unreactive silica” (often called colloidal silica) can be present in the make-up water which is not detected by the analyzer. However, under high temperature conditions of the boiler the silica changes its structure and becomes reactive silica. This explains how the silica concentration may appear to increase in the system when there was no apparent increase in the make-up water silica. To troubleshoot this scenario, a grab sample of the make-up water is typically acidified and boiled, cooled, and either poured through the on-line analyzer or tested on the bench. A significant increase in silica content compared to the raw sample indicates that non-reactive silica is present. In general, it is best to avoid contact of low level silica samples (especially with pH values greater than 8) with the borosilicate glassware that is typically found in the lab. Silica reagents and standards, sample tubing, and grab sample containers should all be made of polystyrene, Teflon, or other plastic components. Freezing to preserve silica samples can lower silica content by 20-40% in waters that have pH <6.

16-6


16.4.3 Particulate Matter

All colorimetric tests must also account for particulates in the sample which result in light scatter across the sample cell. The particulates cause an opacity or cloudiness in the sample which can hinder the analytical measurement. This light scatter can be misinterpreted as light absorption which will lead to an erroneously high measured silica content. Most instruments try to minimize this interference by taking a “zero reading” or “base-line reading” of the sample immediately before adding the color developing reagent. This is certainly a valid and desirable step in the measuring process. However, the analytical reading is usually made after the sample has been allowed to stand undisturbed for some length of time. If particles that were suspended at the time of the “zero reading” have settled at the time of the analytical reading, an erroneous quantification will occur. Particulates can thus lead to inconsistencies in analytical results. While suppliers advertise the ability of their instruments to have no particulate matter interference, the user should make every effort to filter particulates before doing the colorimetric test. 16.4.4 Reagent Contamination

The ammonium molybdate reagent used in the silica test is particularly prone to silica contamination and the amount of this silica can vary from one batch of reagent to the next. Therefore, a blank value is usually provided with each bottle of molybdate reagent and this number is programmed into the analyzer each time the reagents are changed. Incorrect blank values will cause a bias in all readings. 16.4.5 Sample Discoloration

Sample discoloration can also affect on-line silica analyzers. While establishing the “zero reading” as mentioned above tends to cancel out the effect of background color, discoloration in the same approximate wavelength (e.g. blue in this example) can shift the calibration line and limit the maximum range of quantification. The detector senses that the instrument is no longer in the straight line region of the calibration curve and refuses to quantify the result; it gives an “over-range” message which disallows the test. 16.4.6 Sample Temperature

The sample temperature as it corresponds to the relative humidity of the sample compartment must also be considered. Samples that are cooler than the surrounding dew point will cause condensation on the colorimetric cell and lead to incorrect and erratic results.

16.5 Calibration On-line silica monitoring instruments have both manual and automatic calibration capabilities. A typical automatic calibration routine is to perform a weekly verification using an internal 16-7


standard. Frequently, this calibration uses a standard purchased from the instrument manufacturer/supplier. The automatic calibration is usually an analysis of a single standard within the linear range according to the Beer-Lambert Law to establish the response slope. The second point of the calibration line is the x-y origin prior to addition of the color development reagent. The standard concentration is traditionally rather high on the absorbance / concentration curve to assure stability of the standard. This can lead to accuracy issues if the silica content of the process stream is only 1 or 2% of the standard concentration. A recommended second step would be to introduce a freshly-made standard into the instrument closer to the analytical value of the process stream for accuracy verification. Several of the on-line silica monitors have provision to analyze a grab sample or verification sample independent of the internal solution which is used for the automatic calibration. These on-line results are then compared to those obtained from standard laboratory analysis techniques (such as ASTM D859 [5]). 16.6 Calibration Checks

Manufacturers of on-line silica monitoring instrumentation often do not specify the in-house techniques used to validate their equipment. However, it is common for manufacturers to indicate that their monitors had been validated in the field by comparing the on-line data with measurements made on grab samples using the heteropoly blue bench method [5]. In addition, some manufacturers suggest that validation is also possible by introducing a standard solution into the on-line system and demonstrating that the measured value is equal to the concentration of the standard. EPRI Report TR-103515 [7] indicates that on-line silica analyzers should be checked against laboratory analytical results approximately weekly. A method, such as ASTM D4517 [8], that is significantly different from the heteropoly blue technique can offer a good independent check. ASTM D4517 is a graphite furnace atomic absorption method that determines “total” silica in the range 25 to 250 µg/L (ppb). On-line silica instrument analytical capabilities should be checked periodically to demonstrate calibration stability. Two methods exist for verifying instrument stability; the Standard Injection Method [9] or the Line Method [10]. For the Standard Injection Method, a known standard solution, near the mid-point of the calibration curve, is analyzed by the on-line instrument and the results are compared to the acceptance criteria. Acceptance criteria are either established based on statistically derived limits. i.e., ± 3 sigma or based on some predetermined limits established from experience, i.e., ± 10%. Provided the on-line analyzer agrees within the acceptance criteria the on-line instruments calibration is deemed to be within acceptance limits. If the results are outside the acceptance criteria the on-line instrument must be recalibrated. For the Line Method a calibrated separate silica analyzer, typically a bench top analyzer is used to analyze a sample from the same sample stream as the installed on-line instrument. Provided the bench top analyzer agrees within the acceptance criteria (e.g., matches the results of the on16-8


line analyzer within ± 3 sigma or ± 10%) the on-line instruments calibration is deemed to be within acceptance limits. If the results are outside the acceptance criteria the on-line instrument must be recalibrated.

16.7 Alternative Methods While various manufacturers use slightly different color development reagents and test protocol, on-line silica analysis is typically done by colorimetric comparison through development of the heteropoly blue complex. Alternate methods for silica determination are available, although not as on-line techniques. Graphite furnace Atomic Absorption Spectroscopy (GFAAS) has already been referenced in section 16.6. In addition to different colorimetric analyses, another alternate technique is a method utilizing an Inductively Coupled Plasma (ICP) [11]. An ICP source consists of a flowing stream of argon gas ionized by an applied radio frequency typically oscillating at 27.1 MHz. This field is inductively coupled to the ionized gas by a water-cooled coil surrounding a quartz torch that supports and confines the plasma. A sample aerosol is generated in an appropriate nebulizer and spray chamber and is carried into the plasma through an injector tube located within the torch. The sample aerosol is injected directly into the ICP, subjecting the constituent atoms to temperatures of 5727- 7727 °C (10,340-13941 °F). Because this results in almost complete dissociation of molecules, significant reduction in chemical interferences is achieved. The high temperature of the plasma excites atomic emission efficiently. Ionization of a high percentage of atoms produces ionic emission spectra. The ICP produces an optically “thin” source that is not subject to self-absorption except at very high concentrations. Thus liner dynamic ranges of up to four orders of magnitude are observed for many elements. The efficient excitation provided by the ICP results in low detection limits and the ability to determine many elements simultaneously. The light emitted from the ICP is focused on a grating refractometer which separates the incident light into discrete wavelengths. These narrow columns of light now strike an array of photo-multiplier detectors enabling the emission lines for each element to be quantified. Of interest to this discussion is the resultant working range for silica; measurement of concentrations ranging from 20 µg/L (ppb) to 100 mg/L (ppm) are possible. ICP is traditionally a laboratory based instrument rather than an on-line piece of equipment. The ability to quantify many elements simultaneously leads to high productivity in the analytical laboratory. This capability may have similar future advantages in the power plant laboratory.

16.8 End User Considerations The performance characteristics (quantification range, accuracy, precision, bias, drift) and design characteristics (cycle time, selection of reagents, reagent consumption, sample manipulation, sample conditioning, and chemical interferences) for the silica monitor as provided by the manufacturer or supplier should be considered when selecting a suitable instrument. Other on-line silica monitor considerations for boiler water samples include: 16-9


•

Inlet sample flow and pressure requirements.

•

Physical space requirements and mounting configuration (surface mounted or recessed into a sample panel).

•


•


•

Provision for instrument purge air / pressurization air if required.

•


16.9 Field Experience Several errors may be introduced when using a colorimetric silica analyzer. They can be collected into two groups: Group 1 pertains to inadequate sample delivery and Group 2 relates to analyzer malfunction. A brief discussion of these malfunctions follows. 16.9.1 Sample Delivery

•

All analyzers must rely on sample delivery at an adequate flow rate to assure sufficient flushing between samples. The approach used by one manufacturer is to monitor the inlet sample pressure. This manufacturer requires that inlet sample pressure be maintained from 3.45 kPa ± 2.07 kPa (5 ±3 psi). Another manufacturer has a constant head system with an overflow weir in the sample reservoir. The reservoir also has a float switch installed to indicate a “no sample’ condition. Both designs effectively stop the analyses routine if sample flow is not detected. It is desirable to have an alarm contactor available to signal this “trouble” condition to the DCS and subsequently to the control room / lab. On systems without this relay, a silica analyzer can effectively sit idle for long periods of time and the malfunction will go unnoticed.

•

As discussed previously, the silica monitor attempts to cancel out the effects of suspended particulates in the sample by taking a “zero reading” immediately prior to color development. However, accumulation of particulates in the sample lines and colorimeter cells can cause very erratic light adsorption and thus lead to erroneous readings. Special precautions should be taken during cyclic operation and startups to filter the suspended material prior to the silica analyzer.


•

The most common source of problems with colorimetric silica analyzers has to do with liquid handling within the instrument. Depending on the analyzer design, the list of possible problems includes pinch valves that leak, sample lines that plug or develop pin holes, reagent lines that become unattached or develop leaks, eductors that become plugged and peristaltic pumps that develop leaks. Many of these malfunctions are evidenced by liquid lying in the bottom of the analyzer or running out onto the floor. Less obvious malfunctions will be

16-10


evident by the analyzer failing to calibrate, producing negative values, or yielding wildly fluctuating readings. •

Reagent and standard replenishment continues to cause malfunctions in all designs. Purchase of prepared reagents and standards from the instrument supplier is a good practice. One instrument design has an internal reservoir of demineralizer water used for dilution of concentrated samples. Diligence is required to keep this diluent bottle filled to allow the analyzer to work properly.

•

There is a current trend to “share” analyzers by sequencing multiple sample points through one on-line silica analyzer. This decision can minimize capital dollars required for on-line instrumentation. Sharing silica analyzers has a potential downfall in that the silica content may be widely different in the various streams. A typical make-up water or steam silica value would be 10 µg/L (ppb) or less compared to a boiler sample which would contain up to 500 or 1000 µg/L (ppb). There is a real risk of cross contamination when using one instrument to analyze process samples with widely varying silica content. Many plants that wish to share analyzers dedicate one system to low level silica determination (make-up, condensate, steam or polisher outlet) and another instrument for high level silica (boiler blowdown).

Calibration and maintenance procedures and schedules are typically described in literature supplied with the monitoring equipment by each manufacturer. In addition, guidelines for maintenance and calibration activities are recommended in EPRI Report GS-7556 [12].

16.10 References 1.

Cycle Chemistry Guidelines for Fossil Plants: Phosphate Continuum and Caustic Treatment, EPRI, Palo Alto, CA: 2004. 10004188.

2.

Cycle Chemistry Guidelines for Fossil Plants: Oxygenated Treatment, EPRI, Palo Alto, CA: 2005. 1004925.

3.

Cycle Chemistry Guidelines for Fossil Plants: All-volatile Treatment, Revision 1, EPRI, Palo Alto, CA: 2002. 1004187.

4.

Cycle Chemistry Guidelines for Combined Cycle / Heat Recovery Steam Generators (HRSGs), EPRI, Palo Alto, CA: 2006. 1010438.

5.

ASTM D859-05, “Standard Test Method for Silica in Water”, American Society for Testing and Materials, Philadelphia, PA.

6.

Hach Series 5000 Silica Analyzer; Model 6000 Instrument Manual, Hach Company 3rd Edition; 11/2004.

16-11


7.

BWR Water Chemistry Guidelines: 1996 Revision, prepared by BWR Water Chemistry Guidelines Revision Committee. EPRI, Palo Alto, CA: December 1996. EPRI Report TR-103515-R1.

8.

ASTM D4517-04, “Standard Test Method for Low-Level Total Silica in High-Purity Water by Flameless Atomic Absorption Spectroscopy”, American Society for Testing and Materials, Philadelphia, PA.

9.

Advanced Power Plant Chemistry QA/QC Practices, Scientech, LLC, Clearwater, FL, 2006.

10. ASTM D3864-96(2000), Standard Guide for Continual On-Line Monitoring Systems for Water Analysis, American Society for Testing & Materials, Philadelphia, PA. 11. Standard Methods for the Examination of Water and Wastewater Analysis, 21st Edition, Method 3120. 2005. 12. Monitoring Cycle Water Chemistry in Fossil Plants, Vol. 3 Project Conclusions and Recommendations, by A.F. Aschoff, D.M. Sopocy, D.T. Eglar, O. Jonas, J.K. Rice, C.C. Stauffer, and W.E. Allmon, EPRI, Palo Alto, CA: October, 1991. GS-7556.

16-12


17 TOTAL ORGANIC CARBON

17.1 Purpose and Use Total organic carbon (TOC) is not defined as a “Core Monitoring Parameter” in current EPRI Guidelines [1-4] but it is sometimes used for continuous monitoring of organics in make-up water treatment systems and within the steam-water cycle. Laboratory testing of grab samples for TOC is also practiced as needed for diagnostic and troubleshooting of chemistry problems. A wide variety of organic compounds can enter the steam/water cycle in both nuclear and fossil power plants. Sources of organics include ion exchange resins, lubricating oils, cleaning solvents, makeup water, cooling water from condenser leaks, and deliberate additions (for example, morpholine) made for pH control. Depending on the compounds present, they may cause foaming in boilers, increase mechanical carry-over, and foul ion exchange resins. Feedwater organic contamination can lead to excessive cation conductivities in the steam cycle. There is no definitive and conclusive evidence that these organics have a damaging effect, but it has been long hypothesized that organics may play a significant role with respect to performance and corrosion [5]. The effect of organics on conductivity is of concern as it is not possible to differentiate conductivity caused by inorganic and organic species. As a result, steam cation conductivity limits are becoming increasingly more restrictive in new installations. Organic compounds in feedwater may undergo thermal or radiolytic decomposition and form undesirable corrosive species. For instance, trichloroethane (a degreasing solvent) and sulfonic acid (from cation resins) break down in boilers or steam generators to form hydrochloric acid and sulfuric acid, respectively, together with carbon dioxide. Consequently, some utilities monitor total organic carbon (TOC) on a routine basis to help ensure that potentially damaging levels of organics are avoided. TOC is monitored by means of grab sampling and analysis or by on-line monitoring in the plant for one or more of the following reasons: •

To provide additional temporary information for troubleshooting unusual problems.

•


•

To monitor make-up water quality for organic carbon content.

The data generated by the monitoring equipment are used by plant personnel (chemists and chemistry staff, engineers, and operators). To a lesser extent, this information may also be used by system/corporate chemists and engineers. 17-1

EPRI Proprietary Licensed Material Total Organic Carbon

17.2 Forms of Organic Carbon and Description of Method Continuous on-line TOC analyzers are automated versions of grab sample analyzers. In many cases grab sample analysis capability may be adequate to meet the needs of most fossil power plant applications. The following discussion pertains specifically to on-line analyzers; however as mentioned grab sample analyzers use similar analytical techniques. Carbon can occur in water as inorganic compounds as well as organic compounds. Inorganic carbon exists as carbon dioxide, carbonate ions, and bicarbonate ions, and the amount present in water is known as total inorganic carbon (TIC). The total carbon (TC) present is equal to the sum of the organic and inorganic components (TOC plus TIC). TOC is defined as all carbon atoms covalently bonded in organic molecules. Subsets of TOC also include •

Dissolved organic carbon (DOC)—that fraction which passes through a 0.45 µm filter.

•

Suspended organic carbon—also referred to as particulate carbon—the fraction of TOC that is retained by a 0.45 µm filter.

•

Purgeable organic carbon (POC)—also referred to as volatile organic carbon—the fraction of TOC removed from an aqueous solution by gas stripping under specified conditions.

•

Nonpurgeable organic carbon—the fraction of TOC not removed by gas stripping.

Current on-line TOC instruments are described in ASTM D5997-96 (2005) Standard Test Method for On-Line Monitoring of Total Carbon, Inorganic Carbon and Organic Carbon in Water by Ultraviolet, Persulfate Oxidation, and Membrane Conductivity Detection [6]. TOC methods involve the oxidation of organic substances to carbon dioxide by •

Heating in the presence of oxygen and oxidation catalysts.

•

Adding persulfate in the presence of ultraviolet light.

•

Similar chemical oxidation methods.

Subsequently, the amount of liberated carbon dioxide is quantified by one of several methods including membrane diffusion and conductivity detection and non-dispersive infrared analysis. The amount of carbon dioxide is proportional to the TC, TIC or TOC in the sample depending on the details of the method. The selection of a specific on-line TOC method and instrument often depends on the level of TOC in question, the amount of inorganic carbon present, the matrix of the sample, the need for portability, the requirement for chemical reagents and carrier gas, the need for rapid response, and ease of maintenance. Instrument manufacturers employ specific oxidation techniques, detection devices, internal sample flow designs, purge or vacuum removal processes, and data handling systems to address these measuring requirements.

17-2


17.3 Technical Considerations As mentioned above, TOC and TIC are added together to determine the TC found in a sample. In all of the TOC analytical methods, some differentiation must be made to separate these two carbon fractions for accurate quantification. To determine the quantity of organically bound carbon, the organic molecules must be broken down and converted to a single carbon atom molecular form that can be measured quantitatively. TOC methods utilize high temperature, catalysts and oxygen, or lower temperatures (<100°C (212°F)) with ultraviolet irradiation and /or chemical oxidants to convert organic carbon to carbon dioxide (CO2). The CO2 may be purged from the sample, dried, and transferred with a carrier gas to a non-dispersive infrared analyzer (NDIR). Alternatively, the carbon dioxide may be separated from the sample liquid phase by a membrane selective to gaseous CO2. The CO2 is absorbed into a volume of high purity water and the corresponding increase in conductivity is related to the amount of CO2 that diffuses through the membrane. In most power plant water samples, the inorganic carbon fraction is many times greater than the TOC fraction. Inorganic carbon interferences can be eliminated by acidifying samples to pH ≤ 3 to convert the inorganic species to CO2. Subsequent purging of the sample with a purified gas or vacuum degassing removes the CO2 by volatilization. Sample purging also removes the purgeable organic carbon fraction which can lead to a low TOC value in some designs. Fortunately, in many surface and ground waters, the purgeable organic carbon contribution to TOC is minimal and this process discrepancy can be ignored. Alternatively, some instrument designs measure the TC and the total inorganic carbon (TIC) and calculate the TOC by difference. This method is not optimal when low levels of TOC must be determined. 17.3.1 High Temperature Combustion Method

The high temperature combustion method is suitable for samples containing high levels of TOC with a nominal detection level of 1 mg/L (ppm) carbon although some instruments report a detection level of 10 µg/L (ppb). This method may be particularly well suited for samples having a high amount of suspended organic material since the persulfate oxidation process (see below) may not completely decompose these particles. The persulfate oxidation process may similarly be limited by the ability to effectively introduce these suspended particles into the instrument. The sample is introduced into a heated reaction chamber (approximately 700 °C (1292 °F)) packed with an oxidative catalyst such as cobalt oxide, platinum group metals, or barium chromate. The water is vaporized and the organic carbon is oxidized to CO2 and H2O. The CO2 from the oxidation of the organic and inorganic carbon is transported in a stream of carrier gas and measured by means of a nondispersive infrared (NDIR) detector. Because TC is measured in this process, inorganic carbon must be measured separately or removed by acidification and sparging.

17-3


Several on-line instruments are available which employ the high temperature oxidation technique and NDIR carbon dioxide quantification. The flow diagram shown in Figure 17-1 shows an instrument design where the inorganic CO2 is sparged and removed from the sample prior to the high temperature oxidation, thus leading to a true TOC measurement [7]. A modification to the 700 °C (1292 °F) catalytic oxidation reaction is provided by one manufacturer [12]. This instrument uses a batch process rather than a continuous flow of sample. By increasing the reaction chamber temperature to ≥ 1200 °C (2192 °F), the organic and inorganic carbon molecules are completely combusted using only air. The resultant gas mixture is dehumidified and delivered to the NDIR detector to yield a total carbon (TC) content. A separate sample is acidified and sparged (no reaction chamber) to generate an inorganic carbon (IC) value and TOC is determined by difference. A response time of 3 minutes is quoted in the literature with an operating range of 0.01 to 50,000 mg/L (ppm) TOC. 17.3.2 Persulfate Oxidation

This method is also used to chemically decompose the organic carbon molecules to carbon dioxide for further quantification. Variant devices using this method employ either heat or ultraviolet radiation to activate the chemical reaction. These oxidation methods provide rapid and precise measurement of trace levels of organic carbon in water. Some instruments utilize an ultraviolet lamp submerged in a continuously gas-purged reactor that is filled with a persulfate solution which is continually refreshed at a constant rate. The samples are added serially (a batch process) into the reactor and the resultant CO2 is directed to an NDIR analyzer that is tuned to the absorptive wavelength for CO2. The heated persulfate instruments utilize a digestion vessel that is heated to 95–100°C (203–212 °F). A second design for the persulfate oxidation process uses a continuous-flow injection of the sample into the instrument. The sample is acidified and persulfate solution is added. The sample flow is then split—one stream passes to a delay coil while the other stream passes through the UV reactor. The CO2 is separated from each stream by a membrane that is selective to that gas and allowed to pass into high purity water where the rise in conductivity is measured. The CO2 from the non-irradiated sample represents the inorganic carbon and the CO2 from the irradiated sample represents total carbon. TOC can then be calculated by difference.

17-4


Figure 17-1 High Temperature Oxidation TOC Analyzer Flow Diagram [7] Source: Adapted from Reference 7

A third design variant of the persulfate oxidation process, Figure 17-2, also uses a continuousflow injection of the sample into the instrument. Removal of inorganic carbon is provided by either vacuum degassing or by sparging. The sample containing only organic carbon is then 17-5


transferred to a mixing chamber where the persulfate solution is added. The mixture advances to one or more UV reactor vessels where the carbon is decomposed to CO2. The gas is dried and then admitted to the NDIR for quantification [8]. 17.3.3 Conductometric-type TOC

Two designs of conductometric-type TOC analyzers are commercially available. The first design is a direct conductivity measurement system. This instrument first measures the conductivity of the water sample, and assumes that any conductance is due to the dissociation of carbon dioxide in water: CO 2(aq) + H 2O ↔ H + + HCO3 H 2O ↔ H + + OH −

-


A built-in computer notes the sample temperature and calculates the TIC that would give the measured conductivity. Then, the organic compounds are oxidized by exposing the water to UV light (185 nm) in the presence of a titanium oxide catalyst. This treatment converts the organic carbon to carbon dioxide which changes the equilibrium of Equations 17-1 and 17-2 and increases the conductivity of the sample. The conductivity profile is monitored until no further change occurs indicating the reaction is complete. The computer uses the new conductivity value to calculate the TC and, by difference (TC minus TIC), it also calculates the TOC value. A feature of this instrument is that the electrodes used for the conductivity measurement are fabricated from titanium and are positioned in the UV oxidation reaction chamber. A layer of titanium oxide on the surfaces of the titanium electrodes serves as the catalyst for the oxidation reaction. This technique has a detection limit of less than 1 µg/L (ppb) making it suitable for determination of TIC and TOC in pure waters. However, it cannot make accurate measurements in waters containing other ionic species because the conductivity measurements would be affected by these species. Consequently, all but pure waters must be deionized before the sample is directed to the TOC analyzer. Clearly, it is essential that this deionizing column not release organic matter itself, which would lead to an anomalously high TOC reading.

17-6


Figure 17-2 Persulfate Oxidation, Ultraviolet Lamp TOC Analyzer Flow Diagram [8] Source: Adapted from Reference 8

The second design is a membrane conductometric-type TOC analyzer which has the lowest detection limit of the current detection methods with a reported limit of 30 ng/L (ppt) [9]. This analyzer is also designed for testing high purity water and assumes a low /negligible TIC value. 17-7


A high purity distilled water loop within the instrument becomes the detection and quantification media. The loop is internally purified with a deionization column between tests to assure the highest quality water and minimize the need for low level conductivity standards. The analytical sample is injected into the analyzer at the start of each TOC determination. A reagent-free UV reactor is used to decompose the organic carbon in the sample. This instrument utilizes gas permeable membranes to selectively pass only the CO2 produced from the oxidation of organic carbon into the high purity water loop. Using a CO2 permeable membrane removes the interferences of acids, bases, and halogenated oxidation byproducts from the conductivity measurement. The conductivity increase is due only to the dissociation of carbon dioxide in water (Eq. 17-1). This UV oxidation converts the organic carbon to carbon dioxide which increases the conductivity of the sample. At the end of a predetermined reaction time, the CO2 is at equilibrium on both sides of the membrane, and the water containing the bicarbonate ions is transferred to a conductivity detector. The water is then passed through the purifying ionexchange cartridge and returned to the membrane cell for the next sample. A six minute analysis cycle is possible with this design.

17.4 Calibration Calibration of the TOC analyzers varies significantly with the type of detector. The NDIR detector requires calibration gases on an annual basis (or after detector maintenance). Many of the analyzers provide for one and two point calibration via analysis of standard TOC solutions. Aqueous calibration standards are typically a simple organic molecule such as isopropyl alcohol or potassium acid phthalate. The degree of complete oxidation in these standards may be greater than actual samples and this resistance to oxidation is always a concern with the actual samples. Launching the calibration process can be either manual or automatic. Programmable autocalibration features can vary the calibration frequency from hourly to every 999 hours. Low level conductivity standards are required for some of the conductometric type detectors; these low level standards are often stable for only short times and can introduce errors into the analysis. Some manufacturers recommend factory calibration on an annual schedule. This process involves a complete service diagnosis and instrument calibration to original specifications following the same procedure used during new unit assembly. When the instrument cannot be returned for factory calibration, a field calibration may be used where a manufacturer’s representative installs a reference instrument in parallel with the TOC monitor and compares the results from a series of measurements for temperature, conductivity and TOC.

17.5 Calibration Check On-line TOC instrument analytical capabilities should be checked periodically to demonstrate calibration stability. One method is recommended for verifying instrument stability; the Standard Injection Method [11]. For the Standard Injection Method, a known standard solution, 17-8


near the mid-point of the calibration curve, is analyzed by the on-line instrument and the results are compared to the acceptance criteria. Acceptance criteria are either based on statistically derived limits. i.e., ± 3 sigma or based on some predetermined limits established from experience, i.e., ± 10 %. Provided the on-line analyzer agrees within the acceptance criteria, the on-line instrument’s calibration is considered still acceptable. If the results are outside the acceptance criteria, the on-line instrument must be recalibrated.

17.6 Alternative Methods Several wet chemistry bench methods for TOC analysis are the basis for the on-line analyzers. They include: •

ASTM D4129-05, Standard Test Method for Total and Organic Carbon in Water by HighTemperature Oxidation and by Coulometric Detection [6].

•

ASTM D-6317-98 (2004) Standard Test Method for Low Level Determination of Total Carbon, Inorganic Carbon and Organic Carbon in Water by Ultraviolet, Persulfate Oxidation, and Membrane Conductivity Detection [6].

•

ASTM D5904-03 Standard Test Method for Total Carbon, Inorganic Carbon and Organic Carbon in Water by Ultraviolet, Persulfate Oxidation, and Membrane Conductivity Detection [6].

•

Part 5310, Total Organic Carbon: (B) High Temperature Combustion, (C) PersulfateUltraviolet or Heated Persulfate Oxidation, (D) Wet-Oxidation Method [6].

17.6.1 Closed-Loop Photocatalytic Oxidation

In addition, a new process for on-line TOC has been developed and introduced [10]. The following paragraphs describe the features of the flow loop process, the sample reaction mechanics, and the direct quantification of both TOC and TIC. The goal in developing this new TOC methodology was to design an analysis system that was easier to operate and used less hazardous chemicals than the current combustion and wet oxidation TOC methods. Combustion TOC methods require a high temperature furnace and frequent replacement of the catalyst. Wet oxidation TOC methods require the use of UV light, and hazardous oxidizing chemicals like sodium persulfate. Both methods require a supply of CO2-free carrier gas and are designed as flow-through systems. The disadvantage of the flow-through system is that the duration of the oxidation step must be predetermined. If the time is too short, the TOC results could be incorrect due to incomplete oxidation of the sample. An alternative approach would be to use a closedloop design [10]. For the closed-loop system, the loop is initially open to allow atmospheric air to be pumped into the loop. This eliminates the need for a carrier gas. After the loop is closed, the background CO2 in the loop is used as the baseline. An injection is made through the autoinjector into the reactor and the oxidation process begins. As the oxidation proceeds there is an increase in CO2 in the loop. The CO2 is continuously circulated through the loop and its concentration is monitored 17-9


using NDIR. The concentration of CO2 will continue to increase until the sample has been completely oxidized. The time it takes for complete oxidation will depend on the type and concentration of carbon species being analyzed. There is no fixed oxidation time for the closedloop system. The closed-loop TOC system uses an approach called dynamic endpoint detection. The system will establish a baseline CO2 level. It will then monitor the oxidation until there is no further increase in CO2. The resulting TOC concentration is calculated by subtracting the initial baseline CO2 level from the final measured CO2 level. The system is then vented to re-establish a baseline for the next sample. This reaction to completion is significant since many water samples contain components not easily converted to CO2. To eliminate the use of hazardous chemicals and accessories like high temperature furnaces and UV lamps, this new TOC methodology uses a photocatalytic oxidation process that employs a titanium dioxide slurry and a 400 nm lamp. The sample and a titanium dioxide slurry are injected into the reactor. The solution is exposed to a 400 nm UV light. When exposed to the 400 nm light, the titanium dioxide particles undergo electron transitions that, in an aqueous suspension, catalyze reaction of water molecules near the particle surfaces to produce hydroxyl radicals. The hydroxyl radicals so produced oxidize organic compounds that are present in the water sample. Organic compounds are rapidly oxidized to carbon dioxide as shown in equation 17-3: C 6 H 5 X + 7O 2 → 6CO 2 + HX + 2H 2 O

Equation 17-3

The CO2 produced is continually pumped through the closed-loop system and its concentration is measured by the NDIR detector until the oxidation process is complete. The major advantage of this closed-loop photocatalytic system over the combustion and wet oxidation flow-through system designs, is the ability to measure total organic carbon directly. All other TOC systems must measure TOC indirectly as the difference between TC and TIC. For the closed-loop system both TIC and TOC can be measured in a single sample analysis. This system being completely closed also eliminates the loss of purgable organic carbon (POC) during the TIC analysis, which has been a serious problem with the flow-through systems. A water sample is injected into the reactor and is combined with the titanium dioxide suspension that is pH adjusted to 3. At a pH of 3, any inorganic carbon present in the sample is converted to CO2 and measured with the NDIR detector. The change in CO2 concentration in the closed-loop is monitored until an endpoint is detected. The lamp is switched on to initiate the oxidation of all the organic carbon in the sample. The change in CO2 is again continuously measured until an endpoint is detected. At the end of the oxidation process the system is vented to re-establish the baseline for the next sample.

17-10


17.6.2 TOC Data Comparison

Reference 10 indicates that a variety of compounds were analyzed using a closed-loop photocatalytic oxidation analyzer and the results compared to those produced by analysis with combustion technology. The results showed that in almost all cases the closed-loop photocatalytic oxidation technology produced better percent recoveries than the combustion technique [10].

17.7 End User Considerations There are many combinations of oxidative and detection techniques to be considered. Perhaps a process to use for determining which instrument is best for a given user application is to consider the following questions. •

What is the sample matrix? If the sample is raw water which may include high and varying levels of all forms of carbon (inorganic, organic, purgeable), the high temperature combustion method is probably preferable. Instruments that utilize the various wet oxidation with persulfate methods usually require a dilution provision to extend their range upward. This dilution is to assure adequate chemical oxidant for the high amount of dissolved solids. The high temperature combustion will also generally determine organic carbon from compounds that are chemically refractory and not detected with the wet oxidation methods. High temperature combustion may also be desirable for samples containing high levels of suspended organic carbon which may not be efficiently oxidized with the persulfate or UV methods. If conductometric detection is employed, samples that contain a large amount of total dissolved solids may experience interference with a direct conductivity measurement. Utilization of a CO2 permeable membrane will help eliminate the effects of acids, bases, and inorganic ions resulting from oxidation that would cause false positive conductivity increases. Direct conductometric detection can only be used with fairly pure water

•

What is the desired detection level and range of analysis? Some high temperature combustion instruments are not designed for TOC analyses below 1 mg/L (ppm) but the high limit of their working range may reach 20,000—50,000 mg/L (see Table 17-1). The high temperature methods can accumulate non-volatile residues in the reactor, whereas the wet oxidation / UV methods tend to be rinsed at the end of each cycle, thus eliminating accumulation. Instruments based on persulfate and UV oxidation methods frequently report detection limits down to 10 µg/L (ppb) and one instrument actually reports a TOC detection range of 30 ng/L (ppt) [10].

•

What type of sample is the instrument designed for? Several TOC analyzers are designed only for ultra-pure water (UPW) applications with minimal inorganic carbon content. Their design is very simple (usually ambient UV 17-11


oxidation with no chemical or catalytic components), detection is by direct conductometric response, and their detection limits are in the parts per trillion range (typically 0.05 µg/L (ppb) to 0.1 µg/L (ppb)). However, one must recognize the limitation of these instruments since they have no provision for acidifying the sample and either removing the inorganic carbon through sparging or measuring the inorganic carbon through a separate detection step. Thus, values reported as TOC should more properly be reported as TC. In UPW, the contribution of carbonate / bicarbonate inorganic material should be very low and this value reported as TOC is a good approximation. However, in naturally occurring waters, the inorganic carbon can be many times the TOC value and one would get very erroneous values with these instruments. Table 17-1 Summary of Some Typical On-Line Instrument Capabilities (Data compiled from Manufacturer’s Literature) Oxidation Type

Detection Type

Minimum Detection (mg/L)

Detection Range (mg/L)

Precision

Accuracy

High Temp (1200 °C)

NDIR

0.1

0.1–50,000

Not Reported

Not Reported

High Temp (680°C/ Pt)

NDIR

0.1

0.1–20,000

±5% or reading

±5% of reading

UV/Persulfate (100°C)

NDIR

0.002

0.002– 30,000

±1.5% or 2 µg/L (ppb)

Not Reported

UV/Catalyst TiO2 (ambient)

NDIR (closed loop)

0.05

0.05–5000

± 2%

Not Reported

UV Ambient

Membrane Conductivity (Closed Loop)

0.00003 (30 ppt)

0.00003– 2.5

≤1% RSD or 0.03 µg/L (ppb)

± 5% of reading

UV Ambient (UPW only)

Direct Conductivity

0.00005 (50 ppt)

0.00005 – 1.0

±0.05 µg/L (ppb) or ± 5%

±2 of reading

17.8 References 1.


2.


3.


17-12


4.


5.

Maughn, E.V. and Staudt U. “TOC—The Contaminant Seldom Looked For in Feedwater Makeup and Other Sources of Organic Contamination in the Power Plant”; PowerPlant Chemistry, Vol. 8 (2006).

6.

Annual Book of ASTM Standards, American Society for Testing and Materials, Philadelphia, PA.

7.

Hach-Astro TOC HT Process TOC Analyzer; User’s Manual; Hach Company, Loveland, CO.

8.

Hach 1950plus TOC Analyzer; User’s Manual, Hach Company, Loveland, CO.

9.

Sievers 500RL On-Line TOC Analyzer; GE Water and Process Technologies, Product Information Bulletin.

10. A New Approach to Total Organic Carbon Analysis using Closed Loop Photocatalytic Oxidation; Anatel Corporation. Presented at ISA Analytical Division Symposium, Denver, CO 2002. 11. Advanced Power Plant Chemistry QA/QC Practices, Scientech, LLC, Clearwater, FL, 2006. 12. LAR Process Analyzers; Quick TOC-Series; Berlin, Germany.

17-13


18 ELECTROCHEMICAL CORROSION POTENTIAL

18.1 Purpose and Use The electrochemical corrosion potential (ECP) is also known by several other names, including the corrosion potential, the reduction potential, or, simply, the potential. The ECP of metal components is not an EPRI Guideline Core Monitoring parameter [1,2,3,4] in fossil power plant steam-water cycles but is measured in some nuclear plants to verify that water chemistry conditions are optimal. For instance, ECP may be measured in the recirculation system or in the core of a boiling water reactor (BWR) to help in the control of either hydrogen water chemistry or the noble metal chemical addition process. Like oxidation/reduction potential (ORP, see Chapter 6), ECP is an example of an electrochemical potential that provides valuable information about the reactions occurring at a metal/environment interface. However, ECP should not be confused with ORP. ECP reflects the corrosion reaction occurring on the metal surface whereas ORP reflects the oxidizing or reducing power of the environment. While both ORP and ECP involve the measurement of a potential difference between a metal and a reference electrode, the metal (indicator) electrode is a noncorroding, noble metal (typically platinum) in the case of ORP and is the engineering metal (plant component) of interest in the case of ECP. Both ECP and ORP may change with a change in pH, temperature, partial pressure of oxygen or hydrogen in the aqueous solution, mass transport properties, flow rates, time of exposure, etc., but ECP and ORP do not normally have the same value. Typically, ORP measurements are made at ambient temperature but ECP is normally measured at the relevant operating temperature of the plant component so that the ECP value reflects the (often temperature-sensitive) corrosion reaction of interest. While ECP measurements are not widespread in either fossil or nuclear power plants, they are potentially useful for the following reasons: •

To confirm that environmental conditions are optimum for minimizing corrosion, such as ensuring that dissolved oxygen concentrations are optimum for allowing formation of a protective surface oxide film.

•

To warn of in-leakage of contaminants, such as air in feedwater systems and air or hydrogen in the stator cooling water systems of electric generators, which might result in excessive corrosion and accelerated failure of components.

•

To assess the risk of potential-dependent corrosion mechanisms, such as intergranular attack, stress corrosion cracking, corrosion fatigue, dealloying, pitting and crevice corrosion. 18-1

EPRI Proprietary Licensed Material Electrochemical Corrosion Potential

•

To assess the effectiveness of corrosion control techniques, such as cathodic protection and corrosion inhibitor additions.

18.2 Description of Method When a metal corrodes, it reacts with the environment to produce either an insoluble corrosion product that accumulates at the metal surface (such as rust on carbon-steel immersed in ground water) or a soluble corrosion product (such as zinc [Zn2+] ions dissolving from zinc immersed in acid). In any particular aqueous environment, only a limited number of reactions are thermodynamically possible. The two parameters that have the largest control over the thermodynamic possibilities are the ECP of the metal and the pH of the aqueous solution. Plant chemists and operators have monitored the pH of the various waters in power plants for many years to help establish and control the corrosivity of these fluids (see Chapter 7). More recently, and especially in nuclear systems, the ECP of metals has also been monitored. Monitoring the ECP of a component helps establish whether the component is susceptible to certain types of corrosion. For instance, it can indicate whether stress corrosion cracking or intergranular attack is likely to occur or whether the formation of a particular corrosion product is thermodynamically possible. For the same reasons, it provides a measure of the efficacy of some corrosion control techniques, such as cathodic protection or water treatments that are designed to control corrosion by altering the ECP. When monitoring ECP, the metal of interest may be the power plant component itself (such as a pipe or heat exchanger tube) or it may be a separate piece of metal (a “probe”) having a chemical composition similar to the component of interest and immersed in the environment of interest. When a metal—such as iron—is corroding in an environment—such as water containing dissolved oxygen—the primary oxidation reaction is irreversible and heterogeneous: e.g. Fe → Fe2+ + 2e-

Equation 18-1

This reaction, together with the primary reduction reaction, e.g. H2O + ½ O2 + 2e- → 2OH-

Equation 18-2

controls the ECP measured. Theoretically, each corrosion product is thermodynamically stable only in a specific corrosion potential range. A change in the corrosion reaction product from Fe2+ to Fe3O4 to Fe2O3, for instance, would be reflected in corresponding increases in the ECP. The ECP is an experimental observation and is the potential in volts of an electrode in an electrolyte (an environment containing ions) measured against a reference electrode, as described below. It represents the reversible work to move unit charge from the electrode surface through the electrolyte to the reference electrode. The ECP is measured at the electrode/electrolyte interface and may be influenced by many factors, including the composition of the metal; prior metallurgical history, heat treatment, or deformation history; orientation of the grains of the 18-2


metal with respect to both the environment and the stress; surface roughness; scratches on the metal surface; cut edges; the shape of the metal electrode; and film formation [5,6]. It is also strongly influenced by the makeup of the environment, such as the concentrations of dissolved oxygen and metal ions in the electrolyte. 18.2.1 Use of Reference Electrodes

It is not feasible to measure the ECP of a metal directly, but it is possible to measure differences in potential with a voltmeter. This is why reference electrodes are so important. A reference electrode provides a constant electrochemical reference point (a baseline potential) against which the potential of a metal can be compared. Just as sea level is commonly used as the reference point to judge the height of mountains, so a reference electrode is used to judge the magnitude and variability of the potential of a metal. Figure 18-1 illustrates this concept. If the potential difference measured between the metal and the reference electrode (REF. 1) is 365 mV, the ECP of the metal (Ecorr in Figure 18-1) is stated to be 365 mV versus REF. 1, or 365 mV with respect to REF. 1; and it is written simply as 365 mV (REF. 1). A second type of reference electrode (REF. 2) can be just as stable and serve equally well as a reference point, yet it may have a different baseline potential value from REF. 1, as illustrated in Figure 18-1. This results in a different measured potential difference. In the example given in Figure 18-1, the potential difference has been reduced to 125 mV, and the ECP of the metal (Ecorr) would be expressed as 125 mV versus REF. 2. Clearly, it is important to specify the reference electrode used when reporting ECP values. At elevated temperatures, it is common to report potentials versus the standard hydrogen electrode (SHE) even though the actual measurement may have been made versus a different reference electrode. The conversion is achieved by adding the appropriate reference potential difference, ∆Eref, to the measured ECP. For instance, in Figure 18-1, if REF. 1 is an SHE, and the hypothetical REF. 2 was used for the actual potential measurement, ∆Eref = +240 mV. The potential versus SHE is given by Equation 18-3: ECP (SHE) = ECP (REF.2) + 240 mV

Equation 18-3

By definition, the SHE scale is independent of temperature, but this is not true for other reference electrodes. Consequently, to make the conversion to the SHE scale, the value of ∆Eref must be known for the temperature at which the measurement was made. This approach may be used to convert potentials from one reference electrode scale to the SHE (or any other reference electrode) scale if the value of ∆Eref is known. Tables of ∆Eref for commonly used reference electrodes at room temperature (77°F or 25°C) are readily available in standard electrochemistry and corrosion texts [7,8]. A few examples are also provided in Table 6-1 in Chapter 6.

18-3


When a reference electrode is properly designed, its potential does not change with time when immersed in the aqueous environment of interest. Thus, when the voltmeter detects changes in the potential difference between the reference electrode and a metal immersed in the same environment held at constant temperature, changes in potential are occurring only at the metal surface, not at the reference electrode.

Figure 18-1 An Analogy between Measuring Mountain Height with Respect to Sea Level and Measuring ECP with Respect to a Reference Electrode

18.3 Technical Considerations 18.3.1 Reference Electrode Selection

A silver/silver chloride (Ag/AgCl) electrode is an example of a reference electrode used widely by the electric power industry in aqueous environments at temperatures from ambient up to 575°F (300°C) [9,10]. The potential of the Ag/AgCl electrode is constant because a single reversible reaction occurs on the electrode surface (the forward and backward reactions occur at the same rate): AgCl + e- → Ag + Cl −

18-4

Equation 18-4


The potential of the electrode is controlled only by the activity of the chloride ions (and the temperature). The heart of the electrode is a silver wire that is partially coated with, or chemically or electrochemically converted to give, a layer of silver chloride. In many commercial designs, this coated wire is surrounded by a potassium chloride solution of known concentration, and contained within a tube made from an electrically non-conductive material, such as glass. Electrolytic contact must be made between the potassium chloride solution on the inside of the tube and the aqueous solution of interest on the outside without one liquid significantly mixing with, or contaminating, the other. One simple way of achieving this liquid junction is to place a solid porous plug in the wall of the tube. In this way, the electrode “sees” the aqueous solution of interest because of the electrolytic path through the porous plug. Reference electrodes may also be designed with a gel junction, although a flowing liquid junction is usually preferred for high purity samples. In some power plant applications, absolute precision is not necessary. For instance, when monitoring the potential of a cathodically protected condenser tubesheet or waterbox to establish and maintain the appropriate level of protection in seawater, a measurement correct to within 10 mV is perfectly acceptable. Under these circumstances, a rugged reference electrode can be made by eliminating the tube containing the internal potassium chloride solution, and immersing the Ag/AgCl electrode directly into the seawater. Here, the chloride activity in seawater controls the electrode potential at about +0.250 V (SHE) at 77°F (25°C); this value may need some correction during the year to allow for seasonal changes in chloride activity, depending on the proximity to sources of fresh water run-off (e.g. a river mouth). An Ag/AgCl/seawater electrode is inappropriate in sulfide-polluted seawater because its potential would be affected by the reaction between sulfide and silver. In high temperature waters, other reference electrodes, such as copper/copper oxide and platinum/hydrogen electrodes [11,12,13], have been used on occasion. In condenser waterboxes and other lower temperature applications, copper/copper sulfate, saturated calomel (SCE), and other electrodes have been used [14]. All of these electrodes rely on a reversible reaction— equivalent to the Ag/AgCl reaction above—to establish a stable reference potential; and all have shortcomings that limit their use in certain environments. 18.3.2 Quasi-Reference Electrodes Used in Corrosion Rate Probes

Reference electrodes are not only used for measuring ECP values, they can also be used for other electrochemical measurements including corrosion rate measurements. Chapter 19 describes electrochemical methods of measuring corrosion rate and, in each of these electrochemical methods, a reference electrode is required for the measurement. The reference electrode can be similar to those described earlier in this chapter; that is, it may provide a constant potential on a thermodynamically meaningful scale, such as the SHE or SCE scale. Alternatively, it may be another category of “reference electrode”: the type used in many robust field-worthy corrosion rate probes. Here, the quasi-reference electrode is usually a solid metal rather than a 18-5


thermodynamically valid reference electrode. It is critical only that the potential of the quasireference electrode be near-constant for the short period during each measurement when the neighboring “working electrode” (the corroding metal of interest) is polarized by passage of a small current from a counter-electrode. This period may be as short as a few seconds to as long as a few minutes. The quasi-reference electrode potential must serve as a reference point for measuring the magnitude of the induced change in potential of the working electrode but it is not critical to know the absolute potential on a thermodynamically meaningful scale. Occasionally, a corrosion-resistant alloy is used as the quasi-reference electrode in corrosion rate probes, but it is not uncommon for the reference and working electrodes to be the same material. Whatever metal is used for the quasi-reference electrode, its potential may change sufficiently quickly for a period immediately after immersion that it is initially not stable enough to serve as a reference. During such periods, accurate corrosion rate measurements are more difficult, if not impossible. However, once near-steady state conditions have been established—say, its potential changes by less than 1mV in 30 minutes—the metal reference electrode can serve its intended purpose. Furthermore, a metal electrode is an unacceptable reference if pitting or some other form of localized corrosion is occurring at the reference electrode surface. Under such circumstances, the electrode potential may fluctuate and lack the necessary stability. 18.3.3 Reference Electrode Issues

As indicated in section 18.3.1, the potential of the Ag/AgCl electrode is constant because a single reversible reaction occurs on the electrode surface and the potential is controlled only by the activity of the chloride ions (and the temperature). However, the stability of this reference potential can be upset by one or more of the following conditions that can develop in power plant applications: •

The activity of the chloride at the Ag/AgCl junction changes, e.g. by mixing of the internal potassium chloride solution with the external aqueous environment, caused by diffusion through the porous plug.

•

The AgCl is converted to another species by reaction with a component of the external environment that diffuses through the porous plug, e.g. reaction with hydrogen to form Ag.

•

The Ag is converted to another species by reaction with a component of the external environment that diffuses through the porous plug, e.g. reaction with sulfide to form Ag2S.

•

A current flows through the reference electrode into the aqueous environment.

The first three conditions may be inhibited by increasing the diffusion path length between the external environment and the Ag/AgCl junction. For instance, a tortuous path can be provided between the porous plug and the Ag/AgCl junction by packing the tube with inert, nonconductive particles, such as zirconia sand, before filling the interstices with potassium chloride solution. Eventually, however, the reference electrode potential will be affected so that rejuvenation or replacement of the electrode is necessary. The last condition (condition d) arises when the potential difference between the reference electrode and the metal of interest is 18-6


measured with a voltmeter that has too low of an input impedance. Even currents significantly less than 1 µA can polarize the reference electrode and affect its potential. Consequently, electrode potential measurements should be made with a voltmeter or electrometer having an input impedance of at least 108 ohms and as high as 1014 ohms, depending on the application (see Section 18.3.4 for more details). Similar issues can arise when using reference electrodes other than Ag/AgCl so it is imperative to understand the strengths and weaknesses of the reference electrode selected and to provide appropriate and timely maintenance. 18.3.4 Voltmeter Selection

When making electrochemical measurements like ECP and ORP, it is important to use a voltmeter with an input impedance that is sufficiently high that a significant current does not flow through the reference electrode. Current flow would cause the reference electrode to “polarize” and give inaccurate readings. This is particularly important in test environments (such as high purity water) that have very low conductivities. An electrometer with an input impedance exceeding 108 ohms (and preferably 1010 ohms or more) is usually adequate in these low conductivity solutions, but higher input impedances may be necessary in some cases. The selection of a higher input impedance is often accompanied by an increased sensitivity to AC pick-up and “noisy” measurements. Where the electrometer allows selection of the input impedance, the optimum setting can be achieved by starting at a very high input impedance, then reducing the input impedance stepwise until the ECP reading begins to drift or change in an unexpected fashion, indicating polarization of the electrodes. At this juncture, the input impedance should be increased immediately to a somewhat higher value so as to eliminate the polarization.

18.4 Calibration and Maintenance Procedures 18.4.1 Reference Electrodes

All electrochemical potential measurements, including ECP and ORP (Section 6), utilize a reference electrode that must be maintained and periodically checked for proper operation. It is common practice to check reference electrode potentials against the potentials of other nominally identical reference electrodes. All measurements should agree within one millivolt and any reference electrode not meeting this standard should be withdrawn from service. Some reference electrodes can be rejuvenated and brought back into service simply by replacing the internal electrolyte (see below), but some reference electrodes are not designed for rejuvenation and must be discarded. Ideally, all reference electrodes should be calibrated periodically because they tend to change with time. The apparatus needed to make such a calibration using a hydrogen electrode as the standard is described in ASTM G5-94(2004) [15]. The rate that a reference electrode degrades 18-7


depends partly on the type of service it sees and the contaminants it picks up from the service environment. For instance, a silver/silver chloride (Ag/AgCl) electrode, commonly used for both ECP and ORP measurements at temperatures up to 300°C (572°F), is susceptible to irreversible damage if contaminated with sulfide. The formation of silver sulfide on the electrode surface causes a permanent change to occur that cannot be fixed by rejuvenation. Calibration may also be impossible if the silver chloride component has been reduced to silver by hydrogen that has diffused through the reference electrode internal electrolyte from the bulk environment. Dilution of the internal electrolyte (usually potassium chloride, KCl) with the bulk environment may also occur, especially at high temperatures [e.g. above 200°C (392°F)] and when there is a pressure differential between the inside and outside of the reference electrode chamber. Dilution of the KCl will cause the reference potential to drift from the initial value, so periodic replacement of this solution is recommended in electrode designs that allow this. 18.4.2 Electrochemical Corrosion Potential Probe

If the reference electrode has been properly maintained and calibrated, nothing further can be done to calibrate the ECP probe because there is no “correct value” or standard ECP available for calibration purposes. As suggested earlier in this chapter, when reporting ECP values, the reference electrode should always be cited. For instance, if a potential measurement, E, is made using an Ag/AgCl reference electrode that contains an internal 3.0M KCl solution, the potential would be written as E volt (Ag/AgCl, 3.0M KCl). Conversion of ECP values from one reference electrode scale to another can be achieved by adding the appropriate conversion factor to the measured potential (see Table 6-1 in Chapter 6 for examples of conversion factors).

18.5 Field Experience 18.5.1 ECP Measurements in Condensers

As mentioned previously, potentials may be measured in condenser water boxes to ensure that the cathodic protection system has lowered the potential sufficiently to halt various forms of corrosion—primarily galvanic corrosion, but possibly also inlet end erosion-corrosion and sulfide attack. Monitoring the potential is particularly critical when a cathodically protected condenser is fitted with titanium or ferritic stainless steel tubes. These materials can be adversely affected if the potential drops below about -0.75 V (SCE) [i.e. -0.51 V (SHE)], a condition that can arise if too great a cathodic protection current is applied [16]. At potentials more negative than -0.75 V (SCE), electrolytic decomposition of water may result in the production of sufficient hydrogen on the metal surface that ferritic stainless steels are susceptible to hydrogen stress cracking, and titanium is susceptible to hydriding and possibly cracking. A potential monitoring program can help ensure cathodic protection is adequate while preventing the production of damaging hydrogen. 18-8


18.5.2 ECP Measurements in Boiling Water Reactors

In a boiling water reactor (BWR), the ECP of sensitized stainless steel is a key factor in determining its susceptibility to intergranular stress corrosion cracking (IGSCC). Under normal water chemistry (NWC) operation, ECP for stainless steel in the recirculation system piping is typically in the range -0.05 to +0.15 V (SHE) and the sensitized alloy is susceptible to IGSCC [17,18]. However, under hydrogen water chemistry (HWC) conditions, hydrogen is injected into the feedwater which depresses the ECP to values more negative than -0.23 V (SHE), and IGSCC is mitigated [19]. Similar benefits result from a lower addition of hydrogen when used in conjunction with noble metal chemistry addition (NCMA) process. ECP values can thus provide a means of checking that sufficient hydrogen is being injected into the feedwater to mitigate cracking. The ECPs of stainless steel and other alloys used in BWRs have been monitored both in situ and outside of the reactor vessel using various reference electrodes [19]. In situ measurement locations have included residual heat removal system piping and the core. Measurements out of containment can be performed in a sidestream mode in autoclaves fed by water taken from the recirculation system. The latter measurements are generally performed more easily than those done in situ, and maintenance of the electrodes is facilitated. However, stainless steel and other alloy electrodes used in the autoclave must be carefully designed to ensure they are representative of the actual components of interest. Similarly, the environment must closely simulate the service environment of interest; a problem can arise, for instance, if H2O2 or other species in the water decomposes as it travels from the sample point to the autoclave. For this reason, sample lines should be kept as short as possible and flow rates should be high enough to maintain the desired water chemistry in the autoclave.

18.6 Possible Future ECP Measurements in Fossil Plants Although ORP has proven to be an excellent tool for controlling the oxidizing power of the feedwater in all-ferrous systems, or the reducing power of the feedwater for mixed-metallurgy feedwater systems, it is not as direct an indicator of flow-accelerated corrosion (FAC) and other forms of corrosion in feedwater systems as ECP. It is ECP, rather than ORP, together with temperature and pH that have the largest influence on determining the corrosion reactions that are thermodynamically possible. Consequently, consideration should be given in the future to controlling water chemistry and corrosion in fossil plant feedwater systems through the use of in situ measurements of ECP. ECP measurements in other parts of a fossil power plant may also be useful to forecast and control corrosion. A few possible applications are listed at the end of Section 18.1.

18-9


18.7 References 1.

Cycle Chemistry Guidelines for Fossil Plants: All-volatile Treatment, Revision 1. EPRI, Palo Alto, CA: 2002. 1004187

2.


3.


4.


5.

R.B. Mears and R.H. Brown, Industrial and Engineering Chemistry, vol. 33, p. 1001 (1941).

6.

E.D. Verink, Corrosion Testing Made Easy: The Basics. NACE International, Houston, TX, 1994, p. 38.

7.

L.L. Shreir, R.A. Jarman, and G.T. Burstein, Editors, Corrosion, Third Edition. Butterworth Heinemann, 1994 .

8.

R.S. Treseder, Editor, NACE Corrosion Engineer’s Reference Book. NACE International, Houston, TX, 1980, p. 60.

9.

Improved Electrodes for BWR In-Plant ECP Monitoring, EPRI, Palo Alto, CA: August 1982. NP-2524.

10. Measurements of Electrochemical Potential in BWR Water, EPRI, Palo Alto, CA: May 1984. NP-3521. 11. M.E. Indig, B.M. Gordon, R.B. Davis, and J.E. Weber, “Evaluation of In-Reactor Intergranular Stress Corrosion Cracking via Electrochemical Measurements”, p. 411 in Proceedings of Second International Symposium on Environmental Degradation of Materials in Nuclear Power Systems—Water Reactors. American Nuclear Society, LaGrange Park, IL: 1986. 12. E.L. Burley, “Oxygen Suppression in Boiling Water Reactors—Phase 2”. General Electric Company: October 1982. Report No. NEDC-23856-7. 13. B. Rosborg and A. Molander, “The Corrosion Potential of Type 304 Stainless Steel in Swedish LWRs During Steady Reactor Operation”, in Proceedings of Second International

18-10


Symposium on Environmental Degradation of Materials in Nuclear Power Systems—Water Reactors. American Nuclear Society, LaGrange Park, IL: 1986.

14. Current Cathodic Protection Practice in Steam Surface Condensers, EPRI, Palo Alto, CA: May 1983. CS-2961. 15. ASTM G5-94(2004), “Standard Reference Test Method for Making Potentiostatic and Potentiodynamic Anodic Polarization Measurements”, ASTM, Philadelphia, PA (2004). 16. Effects of Selected Water Treatments and Cathodic Protection on Corrosion and Embrittlement of Condenser Tubes, EPRI, Palo Alto, CA: January 1988. CS-5589. 17. Improved Electrodes for BWR In-Plant ECP Monitoring, Electric Power Research Institute, Palo Alto, CA: August 1982. EPRI Report NP-2524. 18. Measurements of Electrochemical Potential in BWR Water, Electric Power Research Institute, Palo Alto, CA: May 1984. EPRI Report NP-3521. 19. BWRVIP-130:BWR Vessel and Internals Project, BWR Water Chemistry Guidelines—2004 Revisions, EPRI, Palo Alto, CA: October 2004. 1008192.

18-11


19 CORROSION RATE

19.1 Purpose and Use On-line corrosion rate monitoring is not widely performed in fossil and nuclear power plants, but previous surveys have indicated that on-line corrosion monitors are occasionally used in main steam condenser cooling water systems and service water systems and, to a lesser extent, in turbine building cooling water, bottom ash water, and reactor building cooling water systems [1]. Corrosion rate monitoring is rarely, if ever, performed in fossil plant steam-water cycles and, more specifically, corrosion rate monitoring is not an EPRI core monitoring parameter [2,3,4,5]. When corrosion rate is monitored on-line, the data generated may warn of a corrosive environment so stimulating corrective corrosion control actions; or they may be used to track the history of metal loss to facilitate the scheduling of maintenance, repair, or replacement of plant components. In other applications, corrosion monitoring may be used to check the efficacy of corrosion control methods such as cathodic protection or corrosion inhibitor additions.

19.2 Description of Traditional Methods Corrosion monitoring techniques have been used for over 50 years, particularly by the oil and gas industry, but increasingly by other industries. The traditional method is not on-line and, instead, involves measuring the weight (or mass) loss experienced by small pieces of metal—socalled corrosion coupons—in the environment of interest after set exposure periods. A corrosion coupon can take one of many forms but often it is a rectangular or circular plate or sheet having an exposed surface area in the range of 10 to 100 cm2 (1.5 to 15 in2). Each corrosion coupon is cleaned, dried, and weighed prior to exposure. It is then mounted in, and exposed to, the environment in a way that prevents it from contacting other metals, i.e. without any unintentional influence of galvanic corrosion. After each exposure period, one or more corrosion coupons are removed from the environment, and corrosion products adhering to the surfaces are removed mechanically or chemically using standard procedures (e.g. see references 6, 7 and 8). The coupons are then dried, re-weighed, and the weight loss of each is calculated. The average corrosion rate (R) for that exposure period is calculated from the weight loss due to corrosion (W grams), the time of exposure (t years), the surface area of the corrosion coupon exposed to the environment (A cm2), and the average density of the metal (ρ gm/cm3), as follows: R, µm/year = (10,000 x W) / (ρ x A x t)

Equation 19-1

19-1

EPRI Proprietary Licensed Material Corrosion Rate

or R, mils/year = (393.7 x W) / (ρ x A x t)

Equation 19-2

where mils/year means thousandths of one inch per year. This weight-loss approach to monitoring has a number of advantages. It involves a very simple, easily understood principle; it has wide application and can be used in virtually any plant environment; the exposed coupons can be used for visual examination and for a relatively easy assessment of resistance to pitting, crevice corrosion, and other forms of localized corrosion; there are no sophisticated electronic instruments to adjust or maintain; and the coupons themselves are usually low cost items. However, this traditional approach to corrosion monitoring has important disadvantages in many applications for the electric power industry. The period of exposure must be long enough to allow sufficient weight change for the corrosion rate to be measured with reasonable accuracy. Unless the metal has very poor corrosion resistance and is totally unsuitable for the application, the minimum exposure period needed to give a measurable weight loss is typically about three months. This leads directly to a major shortcoming of this monitoring approach: the calculated corrosion rate is the average rate for this relatively long period of exposure, and short periods of high (or low) corrosion rates would pass unnoticed. This prevents correlation of transient corrosion rates with operating conditions, knowledge of which could facilitate the development of improved operating practices or corrective actions to reduce corrosion. Another shortcoming of the weight-loss approach is that it can be time consuming to examine, weigh, remove corrosion products from, and re-weigh, the large numbers of coupons needed to monitor corrosion for extended periods.

19.3 Advantages of On-Line Corrosion Monitoring As noted in the preceding discussion, a major shortcoming of the traditional weight loss method of corrosion monitoring is that short term process upsets, or other corrosive conditions, pass unnoticed. Even longer term increases in corrosion rate may not be perceived quickly enough to prevent serious damage to plant equipment if the next withdrawal of corrosion coupons is scheduled for several weeks or months after the event causing the corrosion rate increase. Online monitoring techniques overcome these shortcomings, and allow rapid detection of process upsets. Furthermore, on-line measurements can be made without the need to physically enter the system to withdraw coupons, a process that, in some instances, would require shut down of operating equipment. Other potential advantages and uses of on-line corrosion monitoring equipment include: •

Effective and speedy comparison of candidate corrosion inhibitors, and optimization of inhibitor dosage (e.g. in cooling water systems)

•

Optimization of cleaning schedules for heat transfer surfaces (e.g. in condensers)

•

Improved control of corrosion by cathodic protection (e.g. in condensers)

19-2


•

Identification of optimum operating temperatures and operating procedures for minimizing corrosion, thereby extending component lifetime, reducing the number of forced outages, and improving unit reliability (e.g. in flue gas desulfurization systems)

•

Rapid field trials of candidate replacement materials with minimum plant disruption.

On-line corrosion rate monitors fall into two major categories: physical methods and electrochemical methods. In general, physical methods are less sensitive than electrochemical methods so, for any given incremental change in corrosion rate, an accurate measurement of that change may be obtained after an exposure of only seconds or minutes using electrochemical methods but may take hours or even days using physical methods. The higher sensitivity of electrochemical methods can be a huge advantage in situations where a rapid response to process upset conditions can prevent major damage to plant components. However, there are some circumstances—for instance, in some high temperature high-purity waters—where electrochemical measurements are difficult to interpret and they yield inaccurate corrosion rate values. In these cases, physical methods may be the better option.

19.4 Description of Physical Methods Just as the traditional weight-loss method of monitoring corrosion measures the physical loss of metal directly, so too do some on-line methods. The three most important such methods are: •

Electrical field mapping

•

Metal activation

•

Electrical resistance

These physical methods have seldom been used in power plants although, in principle, the electrical resistance method appears to be quite suitable for some plant applications, especially those where corrosion rates are moderately high (>10 mils/year or >254 µm/year). The electrical field mapping method is used for pipes and other structures that have exposed external surfaces, and carry the corrosive environment internally. In this method, an external power source is used to induce a voltage drop along the critical section of pipe, and the local variations in this induced voltage are monitored through an array of electrical connections on the pipe surface. These connections are made externally using welded or spring-loaded contact pins. The voltage drops between pins are measured and compared with each other, and with the voltage drop across an unattacked reference element, to detect non-uniformity due to corrosion, cracking, pitting, erosion, or other types of metal loss. For general corrosion monitoring, the sensitivity of measurement is claimed to be about 0.1% of the pipe wall thickness; so, for a pipe with a 0.5 inch (12.7 mm) wall thickness, a corrosion rate of 10 mils/year (254 µm/year) would be detected in about 1/20 year, or 18 days. Although this method is not currently used in power plants, it is potentially useful for monitoring flow accelerated corrosion in high risk areas of piping systems such as at elbows and tees. However, application is limited to pipe locations that allow attachment of an array of electrical connections on the pipe surface. 19-3


The metal activation method also has had little or no use in power plants. This method involves activating the metal before exposure to the corrosive environment by bombarding it with either high energy charged particles (in the case of thin layer activation) or with neutrons (in the case of neutron activation). The activated metal may be a sample that is incorporated into a corrosion probe or it may be the component itself. Only a thin surface layer of the metal—in the range 0.001 to 0.012 inch (25 to 300 µm)—is affected by the thin layer activation (TLA) process whereas neutrons may activate metal to a depth of 1 inch or more. In either case, radionuclides are formed that emit gamma radiation. As the volume of activated metal is decreased by corrosion, the gamma radiation level decreases, and the decrease can be monitored with a nearby detector. Alternatively, as the corrosion products enter the process fluid (corrosive environment), the resulting increase in gamma radiation in the fluid may be monitored. The measured radiation level of the corroding metal must be compared with the natural radiation decay of the activated metal which would occur even in the absence of corrosion. It is the difference that provides a measure of the corrosion loss. Clearly, this method is only effective if all, or most, of the corrosion products are soluble in the environment and do not remain in the area being monitored. Insoluble corrosion products remaining at the surface would continue to emit gamma radiation. There are many variables involved in this approach, but a corrosion rate of 10 mils/year (254 µm/year) would be detectable within a few hours using TLA and in less than a day using neutron activation. Use of these metal activation methods has probably been constrained by concerns about radiation exposure and clean-up requirements. Of the three physical methods discussed here, the electrical resistance method appears to offer most promise for future use in power plants because of its simplicity and adaptability to many applications. This approach relies on measuring the increase in the electrical resistance of a metal as its cross-sectional area is reduced by corrosion. The metallic element monitored may be a wire, strip, or tube incorporated into a probe inserted into the environment of interest. Since the temperature of most plant equipment fluctuates, the electrical resistance values measured must be continually corrected for temperature. In electrical resistance probes, this is often achieved by incorporating the metallic element into a Wheatstone or similar bridge, together with a reference element protected from the environment and exposed to the same process temperature. Corrosion rates are computed as metal loss over a known time, with measurement resolution about 0.1% of the probe span. Probe spans typically range from 0.002 to 0.025 inch (50.8 to 635 µm), so a corrosion rate of 10 mils/year (254 µm/year) could be detected in as little as 1/5000 year, or about 2 hours. In general, the greater the resolution, the smaller the cross-section of the metallic element in the probe, and the shorter the useful life of the probe. Obtaining the higher levels of resolution is not easy: it requires the measurement of very small resistance changes (of the order of a few micro-ohms) with instrumentation that is potentially hundreds of feet from the probe. The cable through which the measurement signals pass may itself have significant resistance (of the order of 50 ohms) and may behave like an antenna. Electrical noise may be compounded by thermal noise leading to the necessity for signal processing and a corresponding reduction in resolution.

19-4


Recent advances in digital temperature monitoring technology have allowed the electrical resistance technique to be used at temperatures as high as 1472°F (800°C) with adequate compensation for temperature.

19.5 Description of Electrochemical Methods There are a wide range of electrochemical methods used for on-line corrosion monitoring including the following: •

Linear Polarization Resistance

•

Electrochemical Impedance Spectroscopy

•

Coupling Current (Zero Resistance Ammetry)

•

Electrochemical Current Noise

•

Electrochemical Potential Noise

•

Electrochemical Noise Resistance

•

Galvanic Corrosion Monitoring Using Zero Resistance Ammetry

•

Coupling Current Between Two or More Metals

•

Coupling Current Between Segmented Weld Electrodes

In power plants, the most widely used electrochemical technique is linear polarization resistance (LPR), also known as polarization resistance or linear polarization. LPR has been used occasionally to monitor corrosion rates in main steam condenser cooling water systems and service water systems and, to a lesser extent, in turbine building cooling water, bottom ash water, and reactor building cooling water systems. Many of the other methods listed above have also been used under field conditions although more typically in other industrial settings (such as the oil and gas industry) than in electric power plants. Most of the methods listed are commercially available and, in principle, suitable for use in power plant applications. Because of the complexity and high cost of some of the instrumentation or the difficulty in interpreting the data, some methods, like electrochemical impedance spectroscopy, have previously been used more frequently in the laboratory. However, these techniques are starting to be used for special applications in the field. For instance, one supplier of on-line monitoring equipment offers a suite of electrochemical methods tailored for the oil and gas industry that includes LPR, electrochemical current noise, and an adaptation of electrochemical impedance spectroscopy called harmonic distortion analysis. The electrochemical methods listed above are described in the following sections of this chapter.

19-5


19.5.1 General Considerations

On-line electrochemical methods of corrosion monitoring involve the introduction of a corrosion probe into the environment of interest. The corrosion probe comprises a sample of the metal of interest (the so-called “working electrode”) together with one or more additional electrodes which serve as reference and counter electrodes. In some approaches, the reference electrode and counter electrode is one and the same electrode. Electrical contact is made between each electrode and an electrical lead wire in such a way that the contact area and wire are shielded and electrically insulated from the corrosive environment. The wires are connected to the instrumentation to allow the appropriate electrochemical measurements. The probe design and location in the corroding system can greatly influence the accuracy of measurement. First and foremost, the working electrode material must be representative of the metal of interest in terms of chemical composition, thermo-mechanical treatment, surface finish, and orientation of the exposed surface relative to microstructural features. In some applications, it may be necessary to heat, cool, or stress the working electrode to reproduce service conditions properly; and, in flowing environments, the probe should be installed so that flow over the working electrode surface closely simulates the flow over the metal of interest, in terms of flow rate, geometry, turbulence intensity, etc. Electrochemical methods of corrosion monitoring, in general, allow measurements that are orders of magnitude more sensitive than any of the physical methods mentioned earlier in this chapter. In corrosive environments that have medium to high conductivity, general corrosion rates of less than 0.01 mils/year (< 0.25 µm/year) can be measured essentially instantaneously (within a few seconds or minutes). Accuracy and sensitivity of measurement are limited in practical applications by a number of factors, e.g., resistive potential drops in the electrochemical cell and electrical noise in the measured signals, which are particularly noticeable in high impedance systems such as those associated with high resistivity water and low corrosion rates. For instance, the accuracy of corrosion rate measurements in actual or simulated fossil or nuclear plant feedwaters is often low. The accuracy of most, but not all, electrochemical monitoring methods is also controlled by the ability to determine accurately the factor that relates the measured electrochemical parameter to the actual corrosion rate. This factor, sometimes incorrectly referred to as a proportionality constant, is, in fact, usually not constant. Instead, it depends on still other electrochemical parameters which may vary somewhat with time. Nevertheless, in conductive waters, this factor can usually be determined or predicted with sufficient accuracy that corrosion rates are correct within a factor of 3. In most applications of interest to the electric power industry, this level of accuracy is perfectly acceptable because it is the relative change in corrosion rate that is of interest, not the absolute value. All on-line, electrochemical, corrosion monitoring techniques make measurements that reflect the rates or indicate the mechanisms of the reactions taking place at the metal/environment interface. These methods rely on the concept of a corrosion current, a parameter that is directly proportional to the corrosion rate. While it is impossible to tap into the electrochemical cell 19-6


itself with an ammeter, many of the electrochemical monitoring techniques make measurements that can be directly related to the corrosion current. Once determined, the corrosion current can be related to the more familiar corrosion rate units through Faraday’s Law, which is simply a translation of the corrosion reaction into quantitative terms. For instance, an anodic reaction that may occur on the surface of copper or a copper alloy immersed in water can be expressed as follows: Cu → Cu + + e −

Equation 19-3

The production of each univalent Cu+ ion (the corrosion product) is accompanied by the production of one electron. In the case of the corrosion of an iron alloy, the anodic reaction might involve the production of Fe2+ with a valence of 2: Fe → Fe2+ + 2e-

Equation 19-4

In this case, the production of one ion is accompanied by the production of two electrons. It is clear, therefore, that the higher the valence of the ion produced, the greater is the electron flow. This leads to the conclusion that the corrosion current is proportional to the number of electrons (n) involved in the anodic reaction, e.g. n = 1 in Reaction 19-3, and n = 2 in Reaction 19-4. Furthermore, for each electron produced in the anodic reaction, the weight of metal lost is proportional to the atomic weight of the metal in the corrosion product. In other words, for a given value of the corrosion current, the corrosion rate is proportional to its atomic weight. Faraday determined that one gram equivalent of metal (one gram atom divided by n) was associated with the passage of 96,463 coulombs of electrical charge. Therefore, a metal loss rate of AW/n grams per second generates a corrosion current of 96,463 amps, where AW is the gram atomic weight of the corroding metal. In general, the metal loss rate, R, can be calculated from the corrosion current density (Icorr amps/m2) as follows: R, grams / second = [I corr × A W × A t ] / [96,463 × n]

Equation 19-5

where At square meters is the total area exposed to the environment. Or, in the more familiar units: R, µm/year = [326.9 x Icorr x Aw] / [n x ρ]

Equation 19-6

R, mils/year = [12.87 x Icorr x Aw] / [n x ρ]

Equation 19-7

or

where ρ is the density of the corroding metal in grams/cm3.

19-7


19.5.2 Linear Polarization Resistance (3-Electrode)

The most widely used on-line electrochemical corrosion monitoring technique is linear polarization resistance (LPR), also known as polarization resistance or linear polarization. Instrumentation and corrosion probes for the LPR technique are available in various forms, but they either support the 2-electrode approach or the 3-electrode approach. In addition, a major variable in the instrumentation is how it is designed to compensate for a potentially important source of error—the solution resistance (see below). In the 3-electrode approach, the three electrodes contained within the corrosion probe are the metal sample of interest (the “working electrode”), a counter electrode, and a reference electrode. All three electrodes must be placed in a common corrosive environment (the electrolyte or aqueous solution of interest). The working electrode should be fabricated from material that has a chemical composition, thermo-mechanical treatment, surface finish, and orientation in the corrosive environment as similar as possible to the service component of interest. The counter electrode is normally made from a metal but a conducting nonmetal, such as graphite, could be used. Its sole purpose is to deliver a small current to or from the working electrode whenever the corrosion rate is measured. In robust field-worthy corrosion probes, the reference electrode is usually a metal rather than a thermodynamically valid reference electrode that might be used in a laboratory setting (see Chapter 18, Electrochemical Corrosion Potential). In practice, the counter, reference, and working electrodes may be made from the same material, though this is certainly not always the case. During a corrosion rate measurement, an instrument, called a potentiostat, passes a current, ∆Imeas amps, between the counter and working electrodes. The potential of the working electrode changes by an amount ∆E volts as a result of this current flow, and this is recorded as an identical change in potential difference, ∆E, between the reference and the working electrode. ∆E is typically kept quite small, usually 10 mV but sometimes higher. The corrosion rate of the metal is directly proportional to ∆Imeas /∆E. The reciprocal of this value, i.e. ∆E/∆Imeas, has units of resistance (ohms) and is known as the polarization resistance, RP. Stearn and Geary [9,10] developed a relationship between RP and the corrosion current, Icorr amps/m2 as follows: I corr = B /[A t × R P ] = [B × ∆I meas ] / [A t × ∆E]

Equation 19-8

where At square meters is the total area exposed to the environment, and B is a factor that depends on the other electrochemical parameters—the anodic Tafel coefficient, ba, and the cathodic Tafel coefficient, bc. A full discussion of these coefficients can be found elsewhere [11]. While the anodic and cathodic Tafel coefficients may vary a little with time as corrosion rates and mechanisms change, typically the value of B/At is assumed to be constant. The value of B may be determined experimentally or simply estimated. Dividing B/At by the RP value, determined by the LPR technique gives Icorr (Equation 19-8). The value of Icorr can then be converted to conventional corrosion rate units using such relationships as those shown in Equations 19-6 and 19-7. While commercial LPR units monitor RP, they may provide an output 19-8


that is calibrated directly in mils/year or µm/year for some common combinations of metals and environments. While a detailed discussion is outside the scope of this report, the reader is cautioned that, under some conditions, exchange currents for oxidation-reduction (also called “redox”) reactions occurring on the working electrode surface are comparable in magnitude to the corrosion current and can lead to an overestimate of ∆Imeas. For instance, in high temperature nuclear steam generator waters, correction for the effect of redox reactions must often be made to obtain accurate measurement of the corrosion current. Equation 19-8 assumes that ∆E is a true indication of the electrochemical potential change at the working electrode surface. In fact, ∆E equals the sum of this potential change and the resistive potential drop in the solution between the reference electrode and the working electrode. The resistive potential drop is equal to ∆Imeas × Rs, where Rs is the resistance of the solution between the reference and working electrodes. Consequently, Equation 19-8 must be rewritten as follows: B/[A t × (R P + RS )] = [B × ∆I meas ]/[A t × ∆E]

Equation 19-9

The 3-electrode probe immersed in the corrosive solution often behaves electrically like the electrical circuit shown in the upper section of Figure 19-1. In other words, the response of the circuit to the passage of the current, ∆Imeas, is the same as the response that would occur in the equivalent circuit shown in Figure 19-1. Cdl is the capacitance of the working electrode/solution interface, and is known as the double layer capacitance. RP* and Cdl* are the polarization resistance and double layer capacitance of the counter electrode, and Rs* is the resistance of the solution between the counter and reference electrodes. When using the 3-electrode probe with a potentiostat, the value of Rs* is irrelevant but, in a 2-electrode system (see below), this parameter must be considered. In situations where the corrosion rate is relatively low (RP is high) and the conductivity of the corrosive environment is relatively high, Rs is negligible compared with RP and Equation 19-8 can be used directly. However, in some applications of interest to the electric power industry, this is not true. Here, the resistive potential drop, Rs, must be measured and subtracted from the RP + Rs term in Equation 19-9 before Icorr [=B/(At× RP)] can be calculated. In extreme situations, where the water conductivity is very low (e.g. boiling water reactor steam generator water), Rs is so much larger than RP that RP + Rs (measured) is approximately equal to Rs (measured), and the difference between these two large numbers (the needed value, RP) is a relatively small number with poor accuracy.

19-9


Figure 19-1 Equivalent Electric Circuit (Upper Figure) and Corresponding Schematic Diagram of the 3-Electrode Corrosion Probe Immersed in the Corrosive Environment (Lower Figure)

There are several approaches to solution resistance compensation. The easiest approach is more properly termed error minimization. For instance, in some situations, placing the reference electrode as close as possible to the working electrode in order to minimize the value of Rs may provide adequate accuracy. Configurations that minimize current flow in the solution near the reference electrode will also minimize solution resistance error. For instance, a flush-mounted triangular array of electrodes will give a larger error than a flush-mounted concentric ring electrode configuration where the central electrode is the reference electrode, the intermediate 19-10


ring is the working electrode, and the outer ring is the counter electrode (Figure 19-2). The current flow from the counter electrode to the working electrode will have highest density in the solution situated between these electrodes, and will be much lower in the solution situated near the reference electrode.

Figure 19-2 View of the Exposed End of a Corrosion Probe with Three Concentric Flush-mounted Electrodes Arranged to Minimize Solution Resistance Errors

One approach to solution resistance compensation involves passing a high frequency ac voltage between the counter and working electrodes and monitoring the corresponding current. As described in TR-112024 [1], these data may be used to determine the complex impedance of the system which, at high frequencies, is equal to the total solution resistance between these * electrodes (Rs + Rs ). The total solution resistance can then be related geometrically to Rs. For instance, if the reference electrode is equidistant between the working and counter electrodes, Rs is equal to (Rs + Rs*)/2. A third method, the current interrupt method, involves monitoring the instantaneous potential drop, ∆Vinst, that occurs when the current, ∆Imeas, flowing between counter and working electrodes is interrupted: Rs is equal to ∆Vinst/∆Imeas. The error minimization approaches are more commonly used than the solution resistance compensation approaches in industrial LPR monitoring instrumentation. The former, while less costly than the latter, could lead to inaccurate corrosion rate measurements in some low conductivity waters. 19.5.3 Linear Polarization Resistance (2-Electrode)

The 2-electrode LPR method is quite similar to the 3-electrode method but, here, the reference electrode is eliminated, and the counter electrode must be made from the same material as the working electrode. As before, when the potentiostat delivers the current, ∆Imeas, the working 19-11


electrode is polarized (its electrode potential shifts from its natural, or free corrosion, potential) usually by about 10 mV. The counter electrode is polarized by a similar amount, but with opposite polarity, such that the total potential difference, ∆E′, between the working and counter electrodes is about 20 mV. It is ∆E′ that is measured and controlled by the potentiostat in the 2-electrode system. The modified equivalent electric circuit is shown in Figure 19-3.

Figure 19-3 Equivalent Electric Circuit for the 2-Electrode Corrosion Probe Immersed in the Corrosive Environment

Equation 19-8 and 19-9 must also be modified because ∆E′ is measured instead of ∆E, RP equals RP*, and Rs′ (= Rs + Rs*) is the relevant solution resistance, not Rs: Icorr = B / [Rp x At]

Equation 19-10

B/[(2Rp + Rs') x At] = [B x ∆Imeas] / [∆E' x At]

Equation 19-11

It is generally accepted that the 2-electrode approach results in larger solution resistance errors than the 3-electrode approach, but this is irrelevant in highly conducting solutions and at relatively low corrosion rates. For instance, solution resistance errors would be minimal when using either approach for copper alloy condenser tubes in brackish water or seawater but may be significant, especially when using the 2-electrode approach, for copper alloy condenser tubes in low-conductivity lake or river water. The shape of the corrosion probe in both the 2-electrode and 3-electrode approaches can be chosen to suit the application. Frequently, a cylindrical housing is fitted at one end with the two or three electrodes, separated from each other by non-conductive material. The electrodes may be fabricated flush with the flat end of the cylindrical housing (an example is shown in Figure 19-12


19-2) or they may project finger-like from the end. These electrode assemblies can be mounted in the service component of interest (for instance, through the wall of a pipe, tank, or waterbox) so that the electrodes are either flush with the wall of the component or simply protrude into the corrosive environment. Flush mounting is preferable when flow conditions influence the type and rate of corrosion. Corrosion probes are available for specialized applications too. For instance, flow-through ring-shaped electrodes can be fabricated from condenser tube materials and mounted on the end of a condenser tube so that the flow conditions in the tube are reproduced on the electrode surfaces [12,13]. 19.5.4 Electrochemical Impedance Spectroscopy

Electrochemical impedance spectroscopy (EIS)—sometimes called ac (alternating current) impedance—is an ac method that is analogous to the LPR dc method. The estimate of corrosion rate is also based on the measurement of RP which is substituted in Equation 19-8 to calculate the corrosion current, then in Equation 19-6 or 19-7 to calculate the corrosion rate. In the laboratory, EIS is performed using three electrodes, similar to the 3-electrode LPR configuration (see above); an ac current is passed between the counter and working electrodes to control the potential of the working electrode, measured with respect to the reference electrode. More precisely, a small ac voltage signal is usually applied and the resulting current measured. A frequency response analyzer can be used to apply the sinusoidal voltage (typically having an amplitude of 10 to 20 mV) to the working electrode, and to monitor the current response. The resulting data can be stored in, and subsequently analyzed by, an on-line microcomputer. The current response is compared with the applied potential to yield the complex impedance at that frequency. By measuring the complex impedance over a broad range of frequencies (typically 1 mHz to 10 kHz), it is possible to obtain information about the corrosion mechanism, and to extract values of Rp and Rs. While invaluable for research purposes, a major disadvantage of full frequency EIS is the complexity and high cost of the instrumentation, and the complexity of data analysis. EIS measurements in the field may be greatly simplified: here, a 2-electrode approach is typically used and measurements are made at only two frequencies. In principle, a high frequency measurement is made to determine Rs′ and a low frequency measurement to determine Rs′+2RP. Then, RP is determined by subtracting the high frequency measurement from the low frequency value and dividing by 2. This process may be clarified by referring to Figure 19-3. At very high frequencies, the double layer capacitance, Cdl, behaves like a zero resistance so that the resistance to charge transfer is equal to Rs′; whereas at very low frequencies, Cdl tends toward an infinite resistance, so that the resistance to charge transfer becomes Rs′+2RP. A much more complete discussion of EIS can be found in a primer, published by NACE International [14]. This method has the great advantage over most commercial LPR instruments that there is built in solution resistance compensation. Experience has shown that these measurements can be used reliably to follow trends in general (uniform) corrosion.

19-13


19.5.5 Coupling Current (Zero Resistance Ammetry)

This technique makes use of a so-called “zero-resistance ammeter” (ZRA), which is an operational amplifier circuit that allows measurement of the current, I(t), flowing between two electrodes while imposing essentially zero resistance to its flow. A ZRA may be used to measure the relatively large galvanic currents flowing between dissimilar metals, but it can also be used to monitor the much smaller currents flowing between nominally identical metal electrodes. It is this latter application that is the basis of the ZRA coupling current method of corrosion rate monitoring. Two nominally identical electrodes might be expected to have identical corrosion potentials when immersed uncoupled in a corrosive environment, and hence no current would be expected to flow between them if they were then electrically coupled. In practice, the corrosion potentials of the electrodes would normally be slightly different, perhaps by a few microvolts or a few millivolts, and would tend to fluctuate with time (t). If the two electrodes are coupled together (short-circuited) through a ZRA, they are forced to maintain the same potential. A current flows from one electrode to the other that polarizes (i.e. changes the electrode potential of) the slightly more noble electrode in the active direction, and polarizes the slightly more active electrode in the noble direction. This coupling current, I(t), continuously monitored with the ZRA, fluctuates with time in response to the fluctuating electrochemical nature of each electrode—corresponding to the fluctuating potential difference that would occur if the electrodes were not coupled through the ZRA. Experience has shown that the magnitude of the coupling current is directly related to the corrosion current, Icorr, in many (but not all) situations. Commercial units that employ this technique incorporate a circuit that automatically adjusts zero, and that provides an output, IZRA, that is equal to the root mean square (rms) value of I(t) [15]: I corr = K ZRA × I ZRA

Equation 19-12

where KZRA is a constant. The value of Icorr can then be substituted in Equation 19-8 or 19-9 to determine the corrosion rate. In addition to determining corrosion rates, the recording of IZRA versus time may show patterns that indicate the type of corrosion occurring. For instance, the localized breakdown of a passive film, a precursor to pitting, is characterized by sporadic transients in the current flow. If pits then nucleate, propagation of the pits produces a sustained larger current flow. In determining the corrosion current from Equation 19-12, KZRA is assumed to be constant. This assumption is reasonably accurate over a wide range of corrosion rates but, at very low rates of attack, the measured current is influenced by the resistivity of the electrolyte (corrosive environment) as well as the impedance of the electrodes. Under these low corrosion rate conditions, the ZRA is operating outside its normal calibrated range, and substituting the measured current in Equation 19-12 will lead to an underestimate of corrosion rate.

19-14


19.5.6 Electrochemical Noise

Electrochemical noise measurements are made with two electrodes in a similar way to the coupling current measurements just described but the raw data are processed in a different fashion. The terminology used to describe electrochemical noise phenomena has not yet been standardized, a situation that has led to some confusion. Therefore, it is appropriate to define the terms used in this report. The raw parameters, measured as a function of time (t), are the electrode potential (E(t)), measured with respect to a reference electrode, and the current (I(t)) flowing between that electrode and a second electrode. The E(t) (or I(t)) signal recorded has two components: a dc component and, as described in more detail later, a time-varying noise component with a mean value of zero. In a specified time interval, the mean values of the dc components of potential and current are written as Em and Im, respectively. Similarly, the electrochemical potential noise (EPN) and electrochemical current noise (ECN) components are written as EEPN and IECN, respectively, where

and

EEPN

= E(t) – Em

Equation 19-13

IECN

= I(t) – Im

Equation 19-14

The average amplitude of a potential noise signal can be described by the variance of that potential, sometimes called the voltage noise power, and written as mEEPN2 where the subscript, m, denotes the mean value over a specified time interval. Similarly, the variance of current or current noise power is written as I 2. To avoid working with units of V2 (or A2), it is usual to m ECN report noise amplitudes as the positive square root of the variance, i.e. the standard deviation, √(mEEPN2) with units of volts [or √( mIECN2) with units of amps]. 19.5.6.1 Electrochemical Potential Noise

EPN is the low amplitude, low frequency, random fluctuations of the corrosion potential resulting from stochastic processes occurring at the metal surface: amplitudes are typically in the range of 1 µV to 1 mV and frequencies are 1 Hz and less. Experience has shown that the amplitude and frequency characteristics of these fluctuations can be correlated with the type of corrosion occurring. Indeed, the main advantage of the EPN technique is its ability to identify on-line when localized corrosion (pitting or crevice corrosion) is occurring. There is some evidence that noise measurements can also identify stress corrosion cracking processes on-line. EPN may be monitored as the fluctuations in potential difference between a single metal electrode and a stable reference electrode or between two nominally identical metal electrodes. In the latter case, the two electrodes normally will each produce about the same amount of noise, so the noise powers ( E 2 values) are additive and will be √2 times the noise power generated m EPN by the single electrode. Furthermore, when a reference electrode is used, the dc potential 19-15


difference, Em, between the electrodes is often hundreds of millivolts, so measurement of the much smaller noise component, EEPN, can be quite difficult. Consequently, commercial noise measuring equipment usually uses the two-electrode approach. The data may be gathered using either digital or analog circuitry [15]. For instance, a computercontrolled digital voltmeter can be used with data transform techniques, such as fast Fourier transform or the maximum entropy method. Alternatively, a less complex analog instrument can be used that outputs EEPN or log EEPN. The output of one commercial unit, used in an EPRI project, was the logarithm of the standard deviation, i.e. log √mEEPN2, measured over a narrow band of frequencies [16]. The trace of EEPN versus time can be inspected for “fingerprints” (recognizable patterns) that are characteristic of a particular type of corrosion. General corrosion, pitting and crevice corrosion all have recognizable EPN fingerprints so it is possible to determine on-line when localized corrosion is occurring. With a little training, this analog, real-time monitoring technique can be very instructive. However, if daily or twice daily updates on the system condition are acceptable, gathering the EPN data using digital equipment is advantageous in that it facilitates a statistical analysis of the data. The mean EEPN signal over a 12- or 24-hour period (Em), the standard deviation over the same period [√(mEEPN2)], and the coefficient of variation [√(mEEPN2)/( Em)] are the statistical equivalents of the EPN fingerprints and have been interpreted as shown in Table 19-1. Correlation of the corrosion mechanism with the statistical data (Table 19-1) is feasible only while operating conditions are relatively constant over the 12- to 24-hour data collection period. Changes in unit operating conditions that occur during start-up, shutdown, or load change can also lead to a high standard deviation and a high coefficient of variance. Consequently, the EPN statistical data can be interpreted accurately only after assessing the potential influence of transients in the operating conditions. In general, EPN measurements can be interpreted more accurately when compared and correlated with ECN data gathered concurrently (see section 19.5.10).

19-16

EPRI Proprietary Licensed Material Corrosion Rate Table 19-1 Typical Examples of the Relationship Between the EPN “Fingerprint”, Statistical Parameters, and the Corrosion Mechanism in Progress Type (Mechanism) of Corrosion

EPN “Fingerprint”

Mean (Em)

Coefficient of Variance [√( mEEPN2)/( Em)]

General corrosion

Low signal, with only low amplitude fluctuations.

Low (e.g. 20 µV)

Low (e.g. 5)

Pit initiation

Low baseline signal with frequent large fluctuations.

Low to high (e.g. 20 to 80 µV)

High (e.g. 45)

Pit propagation

High baseline signal with frequent small to medium fluctuations.

High (e.g. 80 µV)

Low (e.g. 5)

Crevice corrosion*

Low baseline signal with Very low (e.g. 5 µV) periodic individual transients that decay exponentially and relatively slowly.

Very high (e.g. 165)

* EPN measurements have accurately identified crevice corrosion in some systems but they are cannot be relied upon to do so in all systems.

19.5.6.2 Electrochemical Current Noise

The ECN technique is analogous to the EPN technique, except that the noise current, IECN, flowing between two nominally identically electrodes, is monitored instead of EEPN (see Equation 19-14). Similarly, the standard deviation √( mIECN2), may be recorded as a measure of the average ECN amplitude over a selected period. This ECN amplitude differs from the current measured by zero resistance ammetry, IZRA, in that the latter is the rms value of the total current (I(t)) flowing, whereas the ECN value is the rms value of only the noise component (IECN). The value of IZRA is typically 100 to 1000 times larger than IECN or √(mIECN2).

Like EPN, ECN can be used on-line to identify when localized corrosion is occurring. Experience has shown that the corrosion current is proportional to the ECN amplitude, as follows: Icorr = KECN x IECN

[or, for digital systems, Icorr = K'ECN x √(mIECN2)]

Equation 19-15

where KECN (or K′ECN) is a constant that depends on the material/environment combination. When using analog instrumentation, sensitivity to pitting can be assessed by calculating the ratio, IECN/ IZRA; and, in the case of digital instrumentation, the equivalent pitting index is given by √(mIECN2)/Im, where Im is the mean dc component of the total current, I(t), over a limited period (e.g. 19-17


12 or 24 hours), and √( mIECN2) is the standard deviation of the ECN over the same period. The higher the pitting factor, the more likely that pitting is occurring. 19.5.6.3 Electrochemical Noise Resistance

The values of EPN and ECN can be combined to produce electrochemical noise resistance (ENR), as follows [17]: R ENR = E EPN / I ECN

Equation 19-16

Alternatively, the digital data can be used to calculate a similar, though not identical, value of RENR, as follows: RENR = [√ (E EPN 2 )] / [√ (m I ECN 2 )]

Equation 19-17

The value of RENR is analogous to RP in Equations 19-10 and 19-11, and can be used to calculate the corrosion current, Icorr, in the same way. Proponents of the ENR method point out that measurement of EEPN and IECN [or √(EEPN2) and √( mIECN2)] do not disturb the electrochemical (corrosion) system at all, since EPN and ECN are natural fluctuations of the electrode potential and current. The LPR method, on the other hand, involves the passage of a small external current and a corresponding small departure from the natural condition. Although substitution of R into Equation 19-10 and 19-11 leads to reasonable estimates of I , the theoretical meaning ENR corr of RENR is still debated. Compact computers have been used in recent years to store and process digital data, and to provide continuous outputs of both RENR and the pitting index, either for consecutive increments of time or for a moving window of time. 19.5.7 Galvanic Corrosion Monitoring Using Zero Resistance Ammetry

By definition, galvanic corrosion may occur when two or more, dissimilar, electrically conducting materials are coupled while immersed in an electrolyte. The current flowing between any two of these dissimilar materials, called the galvanic current, Igalv, normally results in an increase in the corrosion rate of the less noble material, compared with its corrosion rate in the uncoupled state. Similarly, the passage of I results in a reduction of the corrosion rate of the galv more noble component—this phenomenon being the basis of cathodic protection by sacrificial anodes. For instance, if copper is immersed in aerated seawater, it will corrode at a measurable rate, Ru. However, if the same copper is then electrically coupled to zinc, the copper corrosion rate would be reduced significantly below Ru. While the anodic reaction on the copper surface (such as Cu → Cu2+ + 2e-) is inhibited, the cathodic reaction (such as oxygen reduction, O2 +

19-18


2H2O + 4e- → 4OH-) is stimulated. In contrast, the anodic reaction on the zinc surface (such as Zn → Zn2+ + 2e-) is stimulated, while the cathodic reaction (oxygen reduction) is inhibited. Typically, the dissimilar materials are metals, but other electrically conducting materials, like graphite, may form part of the galvanic cell. Galvanic corrosion is often seen in such power plant components as steam surface condensers where combinations of stainless steels, carbonsteel or cast iron, copper alloys, and titanium components can lead to severe corrosion of the less noble components, especially in highly conductive cooling waters like seawater. However, galvanic corrosion can also occur when apparently similar metals or alloys are in electrical contact in a corrosive environment. Such circumstances may arise, for instance, in a welded structure, as described later. A detailed explanation of galvanic corrosion monitoring methodology can be found in a primer, published by NACE International [18]. In brief, the current (Igalv) flowing from a more cathodic material to a more anodic material can be measured if the two materials can be electrically isolated from each other, then coupled through a zero resistance ammeter (ZRA). In some cases, the value of Igalv can be substituted for Icorr in Equation 19-6 or 19-7 to give an acceptably accurate measurement of the corrosion rate of the more anodic material. However, in other cases, this approximation leads to a significant underestimate of the corrosion rate. For instance, for the case of galvanic corrosion of steel coupled to brass in seawater, the local action current, Ilocal, must be added to Igalv to obtain the total anodic current, Itotal: I total = I galv + I local

Equation 19-18

Ilocal is the (unmeasured) current that flows between anodic areas and the cathodic areas remaining active on the steel surface; and it is the value of Itotal that determines the corrosion rate of the steel. Only when Igalv >> Ilocal, can Igalv be substituted in Equation 19-6 or 19-7 to provide an acceptable measure of the steel corrosion rate. [Note that Ilocal cannot simply be determined separately using such conventional electrochemical techniques as LPR (see earlier in this chapter). These techniques can only be applied to unpolarized metals, and are invalid when the electrode is galvanically coupled to a dissimilar metal.] In some instances, coupling a more active material to a more noble material can lead to a reduction of the corrosion rate of the more active material. Specifically, it can occur when galvanic coupling results in passivation of a previously actively corroding metal. 19.5.8 Coupling Current Between Two or More Metals

When galvanic corrosion occurs in real structures, Igalv flowing between dissimilar metal components can rarely be measured directly using a ZRA because it is not practical to electrically isolate the various components. Instead, galvanic corrosion rates can be measured using a multi-element probe immersed in the aqueous environment of interest. The elements are 19-19


metallurgically similar to the alloys in the real structure. The exposed area ratios and geometries of the elements should be as close as possible to those found in service. Such probes are commercially available, and have been used for monitoring galvanic corrosion rates. However, the probes are probably used more frequently for monitoring the level of dissolved oxygen in treated water. For instance, a two element probe (combination of carbonsteel and admiralty brass) is suited to monitoring the dissolved oxygen in the waterflood used to enhance production from oil and gas reservoirs. As described above, when dissimilar metals are coupled through a ZRA, the anodic reaction is stimulated on the more active metal (in this case, carbon-steel) while the cathodic reaction is stimulated on the more noble metal (in this case, brass). If Igalv >> Ilocal, the galvanic current flowing not only provides a measure of the carbonsteel corrosion rate but also the rate of oxygen reduction on the brass surface. In this application, the rate of oxygen reduction is proportional to the dissolved oxygen concentration in the waterflood. 19.5.9 Coupling Current Between Segmented Weld Electrodes

A special form of galvanic corrosion may occur in either weld metal or the adjacent heat-affected zone (HAZ). Minor differences in composition and microstructure across the weld joint can lead to corresponding differences in the local electrochemical potential, which promotes dissolution of the least noble areas by galvanic corrosion. The precise location and rate of this preferential attack can be influenced by the composition of the weld electrode, the welding procedure, and the environment. Conventional electrochemical techniques are unsuitable for monitoring weld corrosion but a specialized ZRA method was developed and used in the early 1990s [19]. First, a sample weld is cut from the component of interest, or a dummy weld is manufactured using the same welding procedure and using materials with the same composition. Five rectangular elements are cut from this weld to form the basis of a 5-element sensor probe: two elements are cut from the parent material, one from each HAZ, and one central element from the weld root run material. The five elements are sandwiched together in the probe, separated from each other by non-conducting material, and arranged as they were in the original weld: parentHAZ-weld-HAZ-parent. Only the front surfaces of the five elements contact the environment of interest, and each of the back surfaces are connected to multi-ZRA instrumentation so that the current flow (either anodic or cathodic) to each element can be monitored. The individual currents can be used to calculate the corrosion rates of each element, thereby establishing the location and rate of attack on-line. Corrosion countermeasures, such as the addition of chemical inhibitors to the environment, can also be evaluated using this technique. 19.5.10 Electrochemical Methods in Combination

While the methods described above may be used singly to monitor corrosion behavior, in some instances they are more effective when used in parallel. Several EPRI studies performed back in the 1980s made use of a monitoring system that employed a multi-element corrosion probe and supporting instrumentation [15,16,20,21,22]. This system collects four sets of data concurrently 19-20


by EIS, ZRA, EPN, and ECN (electrochemical techniques described earlier in this chapter). While detailed interpretation of the data requires some experience, the advantage of the multisystem is that the weaknesses of one method are compensated by the strengths of another. If an analog measurement system is used, for instance, the EIS trace of RP is inversely proportional to the corrosion rate assuming the corrosion reaction is charge transfer controlled (see Equation 1910). However, the EIS signal gives no indication of whether the attack is uniform or localized. Clearly defined peaks in the E traces, on the other hand, indicate initiation of pitting or crevice EPN corrosion. Furthermore, the rate of this localized corrosion is provided by the IECN data. As pitting or crevice corrosion propagates and becomes more widespread, the EEPN and IECN traces show fewer well defined peaks. Finally, the IZRA value most accurately reflects the corrosion rate at moderate to high rates of attack, and tends to underestimate corrosion at low rates of attack. By evaluating all the traces in parallel, it is possible to interpret the data correctly and determine the type and rate of corrosion occurring at the surface of the corrosion probe. The superior analytical capabilities of digital, parallel processing equipment may also be exploited to track several corrosion and process parameters simultaneously, and to search for relationships between these parameters. Such relationships potentially allow development of fault-specific alarms and automated control of operating parameters (such as chemical additions and temperature). One supplier of on-line monitoring equipment offers a suite of electrochemical methods (LPR, electrochemical current noise, and an adaptation of electrochemical impedance spectroscopy called harmonic distortion analysis) that are monitored simultaneously and, after on-line analysis, are capable of providing such automated process control [23].

19.6 Calibration Calibration of corrosion monitoring instruments is typically done by comparing weight loss data using the traditional coupon method with the weight loss calculated by integrating the corrosion rate data measured with the corrosion monitoring instrument over the same period. Integration of the corrosion rate data can be achieved by measuring graphically (or by using a simple computer program) the area under the corrosion rate-versus-time curve. A retroactive correction can then be made to the corrosion rate curve so that the area under the curve equals the weight loss determined using the traditional coupon method. This “calibration” process is essentially modifying (correcting) the B value in Equations 19-8 and 19-9.

19.7 End User Considerations The commercially available corrosion monitoring instruments that are well established and relatively inexpensive include the electrical resistance method and the 2-electrode and 3-electrode LPR methods. Consequently, these techniques are often favored if the selected application allows accurate corrosion rate measurements. The electrical resistance method is inherently less sensitive than the electrochemical methods but has the advantage that accuracy of measurement is not influenced adversely by low conductivity environments. 19-21


For special applications or for failure analysis purposes, some of the other electrochemical methods may be justified. The electrochemical potential noise technique, for instance, might be favored when warning is needed of the onset of pitting or crevice corrosion. Also, galvanic corrosion rates might be monitored using zero resistance ammetry.

19.8 Possible Future Corrosion Rate Monitoring in Fossil Plants On-line corrosion rate monitoring techniques have occasionally been used in power plants but, so far, there has been much more widespread use of these techniques in the oil and gas industry. The current power plant applications (for example, condenser cooling water systems, service water systems, turbine building cooling water, bottom ash water, and reactor building cooling water systems) may see greater use in the future as the benefits of monitoring are more fully realized. For instance, such techniques may be justifiable for use during normal plant operation to help control corrosion inhibitor additions in service waters or level of cathodic protection in condenser water boxes. In addition, there may be new opportunities. For instance, corrosion rate monitoring equipment might be used advantageously to ensure proper control of corrosion rates during chemical cleaning operations. With appropriate analysis and response to the corrosion rate data, it is possible to minimize corrosion rates, reduce maintenance and replacement costs, and increase plant safety.

19.9 References 1.

Reference Manual for On-Line Monitoring of Water Chemistry and Corrosion: 1998 Update, EPRI, Palo Alto, CA: 1999. TR-112024

2.


3.


4.


5.


6.

NACE Standard Recommended Practice RP0775-2005, Preparation, Installation, Analysis, and Interpretation of Corrosion Coupons in Oilfield Operations. NACE International, Houston, TX.

7.

ASTM G4-01, Standard Guide for Conducting Corrosion Tests in Field Applications. ASTM International, West Conshohocken, PA.

19-22


8.

ASTM D2688-05, Standard Test Methods for Corrosivity of Water in the Absence of Heat Transfer (Weight Loss Methods). ASTM International, West Conshohocken, PA.

9.

M. Stern, Corrosion, Vol. 13, p. 755 (1957).

10. M. Stern and A.L. Geary, J. Electrochem Soc., Vol. 104, p. 56 (1957). 11. “Kinetics in Aqueous Corrosion” p. 29-36, and “Laboratory Testing” p. 212-228, chapters in Metals Handbook, Volume 13, Corrosion, 9th edition. ASM International, Metals Park, Ohio: 1987. 12. CISE Newsletter, No. 35. CISE spa, Via Reggio Emilia, 39, Casella Postale 12081, I20134 Milano, Italy (January 1992). 13. “MACC: An Automatic Corrosion Monitoring System For Steam Condenser Tubes”, CISE Technical Description No. DMT-DT-89-001. CISE spa, Via Reggio Emilia, 39, Casella Postale 12081, I-20134 Milano, Italy (1989). 14. S. Turgoose and R.A. Cottis, Corrosion Testing Made Easy: Electrochemical Impedance and Noise. NACE International, Houston, TX (1999). 15. Multisystem Corrosion Monitoring in FGD Systems, Phase 2, by D. M. Farrell, W. M. Cox, and D. Gearey. EPRI, Palo Alto, CA: April 1988. CS-5734. 16. Multitechnique Corrosion Monitoring in Flue Gas Desulfurization Systems, Phase 1, by D. Gearey and W. M. Cox. Electric Power Research Institute, Palo Alto, CA: January 1988. EPRI Report CS-5605. 17. A.N. Rothwell and D.A. Eden, “Electrochemical Noise Techniques for Determining Corrosion Rates and Mechanisms”, Paper No. 223, CORROSION/92, NACE Annual Conference. NACE International, Houston, TX (1992). 18. H.P. Hack, Corrosion Testing Made Easy, Volume 2: Galvanic Corrosion Test Methods. NACE International, Houston, TX: 1993. 19. A.N. Rothwell, “Weld Corrosion: Causes and Solutions”, Corrosion Prevention & Control, Vol. 39, pp. 113-118 (October 1992). 20. On-Line Corrosion Monitoring at the High Sulfur Test Center, EPRI, Palo Alto, CA: June 1993. TR-102538. 21. Multi-System Corrosion Monitoring in a Cyclic Reheat Test Facility: Phase 1, EPRI, Palo Alto, CA: April 1988. CS-5776.

19-23


22. Multi-System Corrosion Monitoring in a Condensing Flue Gas Heat Exchanger: Phase 2. EPRI, Palo Alto, CA: October 1991. GS-7540. 23. D.A. Eden, “Real-Time Solutions Integrate Corrosion Monitoring with Process Control”, presented at Scientech’s 15th Annual International On-Line Instrumentation Seminar, Clearwater Beach, Florida (November 15-17, 2006).

19-24

Export Control Restrictions

The Electric Power Research Institute (EPRI), with major

Access to and use of EPRI Intellectual Property is granted with the

locations in Palo Alto, California, and Charlotte, North Carolina, was

specific understanding and requirement that responsibility for ensur-

established in 1973 as an independent, nonprofit center for public

ing full compliance with all applicable U.S. and foreign export laws

interest energy and environmental research. EPRI brings together

and regulations is being undertaken by you and your company. This

members, participants, the Institute’s scientists and engineers, and

includes an obligation to ensure that any individual receiving access

other leading experts to work collaboratively on solutions to the

hereunder who is not a U.S. citizen or permanent U.S. resident is

challenges of electric power. These solutions span nearly every area

permitted access under applicable U.S. and foreign export laws and

of electricity generation, delivery, and use, including health, safety,

regulations. In the event you are uncertain whether you or your com-

and environment. EPRI’s members represent over 90% of the electricity

pany may lawfully obtain access to this EPRI Intellectual Property, you

generated in the United States. International participation represents

acknowledge that it is your obligation to consult with your company’s

nearly 15% of EPRI’s total research, development, and demonstration

legal counsel to determine whether this access is lawful. Although

program.

EPRI may make available on a case-by-case basis an informal assessment of the applicable U.S. export classification for specific EPRI

Together...Shaping the Future of Electricity

Intellectual Property, you and your company acknowledge that this assessment is solely for informational purposes and not for reliance purposes. You and your company acknowledge that it is still the obligation of you and your company to make your own assessment of the applicable U.S. export classification and ensure compliance accordingly. You and your company understand and acknowledge your obligations to make a prompt report to EPRI and the appropriate authorities regarding any access to or use of EPRI Intellectual Property hereunder that may be in violation of applicable U.S. or foreign export laws or regulations.

Program: Boiler and Turbine Steam and Cycle Chemistry

© 2007 Electric Power Research Institute (EPRI), Inc. All rights reserved. Electric Power Research Institute, EPRI, and TOGETHER...SHAPING THE FUTURE OF ELECTRICITY are registered service marks of the Electric Power Research Institute, Inc.

Printed on recycled paper in the United States of America

1012209

Electric Power Research Institute 3420 Hillview Avenue, Palo Alto, California 94304-1338 • PO Box 10412, Palo Alto, California 94303-0813 USA 800.313.3774 • 650.855.2121 • [email protected] • www.epri.com

Fossil Plant Cycle Chemistry

Recommend Documents