Best Practice Guide Calibration of Flowmeters

Best Practice Guide Calibration of Flowmeters

The Calibration of Flowmeters Best Practice Guide

NEL East Kilbride, Glasgow, G75 0QU, United Kingdom

March 2002

Executive Summary This guide covers the general principles of calibration for flow measurement. It gives an overview of calibration methods used in a variety of situations from calibrations in standards laboratories to calibrations in the field, and verification of a flowmeter in a non-laboratory situation. It is produced for operators of calibration facilities, users requesting calibrations and engineers having to establish a calibration method. The guide is divided into two parts. The first part covers general principles of calibration when applied to the calibration of flowmeters or devices to measure flowing fluids. The second part describes various individual techniques and methods which may be employed.

Acknowledgement The production of this Best Practice Guide was funded by the Department of Trade and Industry's National Measurement System Directorate as part of the 1999-2002 Flow Programme. Author: Richard Paton Date: March 2002

© TUV NEL Limited 2002 ii

Guide to the Calibration of Flowmeters


CONTENTS

Part 1 - General Principles Page

EXECUTIVE SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

ii

PART 1 - GENERAL PRINCIPLES 1 1.1 1.2

W HAT IS C AL IB RAT IO N? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rate, Quantity and Tim e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Resolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1 2 3

2 2.1 2.2 2.3 2.4 2.5 2.6 2.7

THE IMPORTANCE OF CALIBRATION FLUID AND CONDITIONS Fluid Properties . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Flow Profile . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Traceability, Accuracy and Uncertainty . . . . . . . . . . . . . . . . . . . . . . . . . . Accreditation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Reporting the Result . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Calibration Frequency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cost and Specification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

3 4 4 6 6 8 8

PART 2 - CALIBRATION METHODS 3 3.1 3.2 3.3

C AL IB RAT IO N M ET HO DS F OR L IQ UI DS Liquid Collection Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Measurement Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pip e Provers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

10 13 16

4 4.1 4.2 4.3

C AL IB RAT IO NS FO R G AS FL OW ME TE RS . . . . . . . . . . . . . . . . . . . . . . Dis placement Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Critical Flow Venturi-Nozzle (Sonic Nozzle) . . . . . . . . . . . . . . . . . . . . . . . Gravimetric / P.V.TMethods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

20 20 22 23

5 5.1 5.2 5.3

O N- SI TE CA LI BR AT IO N M ET HO DS . . . . . . . . . . . . . . . . . . . . . . . . . . . Tracer Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Insertion Meters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cla mp-on Ultrasonic Meters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

24 24 25 26

6

E XP EC TAT IO NS F OR A C A LI BR AT IO N

. . .. . . . . . . .. . . . . . .. . . . . .

26

APPENDIX 1: CALIBRATION CHECK LIST . . . . . . . . . . . . . . . . . . . . .

27

APPENDIX 2: REFERENCES AND BIBLIOGRAPHY . . . . . . . . . . . . . .

29

iii

Many of the principl es of calibration apply equally across metrology. This part of the guide re-iterates these principles and shows how they apply to flow calibrations. Vocabulary is important to the understanding of the principles, and the important definitions of terms have been reproduced from the 'International Vocabulary of Basic and General Terms in Metrology' 1 (VIM). The VIM will be used throughout this guide to establish terminology to provide a clear base understanding of what is meant when various terms are used in defining a calibration.

1

WHAT I S CA LI BRATI ON?

Calibration:

The VIM definition of calibration is given opposite, but how is it applicable to flow measurement? Flow measurement does not rely on a single operation. Neither does the calibrati on process. Measurement of the quantity of fluid depends on establishing the basic quantity measured and a number of influence factors. Moreover, the flow device has to be calibrated across a range of flowrates. This all combines to give a set of operations which come together to provide the calibration.

"The set of operations that establish, under specified conditions, the relationship between values of quantities indicated by a measuring instrument or measuring system and the corresponding values realised by standards"

As the fluid and influence factors (e.g. temperature and viscosity) all affect the meter performance, the calibration is carried out 'under specified conditions' and these must be defined. A calibration is not an absolute operation. It is a comparison between the device being calibrated and the standard. Through this comparison, a relationship between what the device or flowmeter measures and what the standard measures is established. This must be expressed in some way which gives a meaningful expectation of how the device will perform in use. Standard: "Measuring instrument or measuring system intended to realise or reproduce a unit or one or more values of a quantity to serve as a reference"

The comparison during a calibrati on is with a standard. The standard comprises the system of pumps, pipes, fluids, instrumentation, quantity reference measurement, calculations and operators, all combined to measure the quantity of fluid passing through the flowmeter being calibrated.

As fluid flow is dynamic and all measurement devices are affected in some way by the conditions of use, it will be impossible to have a standard that fully reproduces the conditions under which the meter will be used in practice. Flowmeters are affected by temperature, viscosity, flow profile, flow rate fluctuations and pulsations. They are also affected by the external environment, vibrati on, stress temperature etc. Different devices are affected in different ways.

1


As a calibration is a comparison between the measurement made by a flowmeter and that realised by the standard, the resultant relationship will be for the specified conditions, and a further assessment of the relevance to the final application must be carried out. Selecting the standard will be a compromise to replicate the conditions of use as closely as possible or practicable. The standard must also be compatible with the performance and characteristics of the meter to be tested and the result desired. 1.1

Rate, Quantity and Time

The mechanism by which a flow measurement device gives a reading of flow is dynamic. The sensor reacts to the flow of fluid through or past it to realise an output related to the flowrate or the quantity passing through it. Clearly the measurement of flowrate and that of quantity are related through the measurement of the time interval across which the quantity is measured. In practice the end use of the device leads to different expectations for the behaviour and hence the calibration. In establishing this relationship it is vital to relate the response time of the device to the calibration method.

1.2

Resolution

Resolution: "Smallest difference between indications of a display or output device that can be meaningfully distinguished"

Response Time: "Time interval between the instant when a stimulus is subjected to a specific abrupt change and the instant when the response reaches and remains within specified limits around its steady value"

The interpretation of response time is reasonably straightfo rward for mechanical meters. The mechanical interface between the fluid and the indicator can be pictured and defined in terms of momentum and drag affecting the meter when the flow changes. With the introduction of electronic devices attached to mechanical meters, or particularly as an integral part of the flowmeter, this relationship has become more difficult to establish. To give some examples: a positive displacement meter in liquid responds very quickly to changes in flowrate, even very abrupt changes; the flow stops, the rotor stops and the register stops. If a pulse generator is fitted, when the flow stops, the generated pulses stop, but a frequency counter will not reflect this until it completes its measurement cycle, which may be 1 second or 10 seconds. During that time, a totaliser will correctly indicate quantity, but the flowrate indicator will not be showing the correct (instantaneous) flowrate. To become more sophisticated, the pulses can be scaled by a microprocessor which calculates the output quantity, corrects it for temperature etc, and outputs a proportional pulsed output, mAcurrent output and a digital output of quantity and flowrate. All these outputs will have a different response time and will all lag behind the change in flow by the processing time. In this case however, the lag, although potentially long, may reflect a delay in outputting real measured values. These can be missed if the calibration method has not been chosen to recognise what is happening. With many new technologies such as Coriolis, Electromagnetic, or Ultrasonic meters being totally dependent on a microprocessor based output, and the variety of settings to average, damp or cut off unacceptable results, the definition and understanding of response time is vital to the calibration process. If the device response time does not match the time within which a calibration test point is taken, poor repeatability or calibration offsets may be seen although these effects might not occur in service.

2


pulsed and analogue has to be considered

Although it may seem obvious, the resolution of the device must be adequate to allow a calibration to match the uncertainty requir ed. To achieve this, the standard must be able to measure enough fluid to match the resolution and uncertainty of the device. For example, if a flowmeter has a resolution of 1 litre, any standard used to calibrate it must have a volume of significantly more than 1,000 litres to achieve an uncertainty of 0.1 per cent. To meet oil industry norms, a volume of 10,000 litres would be expected to ensure insigni ficant (0.01%) resolution uncertainty. The requirement for resolution to be assessed applies equally to meters with outputs. In the latter case the effect of sampling and averaging along with the resolution of the instruments.

2

THE IMPORTANCE OF CALIBRATION FLUID AND CONDITIONS

2.1

Fluid Properties

All flowmeters are transducers interacting in some way with the flowing fluid. The nature of this interaction is affected by the properties of the fluid or the velocity distribut ion of the fluid passing through the meter. Changes in this interaction alter the ability of the transducer to give an accurate representation of the quantity. The magnitude of the error is different for different meter types and fluids. For this reason it is desirable to calibrate using the same fluid and pipework for which the meter will normally operate. This is not often possible and hence the best economic compromise must be established when choosing the calibration. This choice will be based on the final duty of the meter, the required uncertainty and a knowledge of the performance expectation of the type of meter. For some meters, for example orifice plates (within accepted limits), the performance can acceptably be related to Reynolds number, allowing a calibration to be carried out in a different fluid to the meter's operating fluid. This may even allow a liquid calibration to relate to a meter for gas duty provided the Reynolds numbers are the same. For other meter types such as turbine meters, especially those required to meter hydrocarbons, the choice of calibration fluid is particularly important. Turbine meters

3


are viscosity-sensitive, and the figure above gives some typical calibration results from a turbine meter for water and three petroleum products. The meter is primarily viscosity-sensitive, and hence it is important to calibrate these meters using the working liquid (or substitute) if possible. For this reason, among others, fiscal meters for oil are often calibrated on site using a dedicated pipe prover. For gas meters, air is often used as the calibration fluid for safety reasons. When used with different gases the performance related to Reynolds number provides a good relationship for a wide range of meters, the notable exceptions being variable area meters and certain thermal flowmeters. If the gas viscosity is significantly different from that of air, then the performance of some positive displacement meters may be affected, and in these cases calibrations should be carried out in the gas on which they are to be used. Properties of the fluid such as density, temperature, conductivity and pressure etc, may also have to be considered when replicating the use of the meter in a calibration. 2.2

Flow Profile

As a fluid passes along a pipe, the distribution of velocity across the pipe alters to approach a 'fully developed' profile that is dependent on the pipe diameter, roughness and fluid Reynolds number. The presence of any change in pipe configuration from a straight pipe will alter the profile drastically: bends, valves, double bends etc, all introduce asymmetry of the velocity distribution, and some of them introduce swirl. As the way the fluid interacts with the sensor can be highly dependent on the velocity profile, these effects must be considered in the calibration. Most calibration facilities allow adequate straight lengths of pipe upstream and downstream of the flowmeter, combined with flow straighteners to provide close to the ideal flow conditions to suit the types of meter being calibrated. It must be noted that such lengths of pipe must have the same internal diameter as the bore of the meter inlet, must not have step changes or mis-alignment of joints, and must not have protrusions from gaskets or joints into the flow. 2.3

Traceability, Accuracy and Uncertainty

As a calibration is a comparison of the reading from a device with that of a standard, it is necessary to consider what properties are required in a standard. Firstly and most importantly, the standard has to measure the same quantity as the device. There is no point in comparing a mass meter output with that of a volume tank without being able to measure density. In flow Accuracy: measurement, the standard is a system comprising a measurement of quantity and subsidiary measurements of "Closeness of the the fluid conditions, properties and influence factors. Time agreement between the result of a also has to be measured to establish flow rate. measurement and a true value of the It is contentious to use the word 'accuracy' in relation to measurand" calibration work as it has little scientific meaning and many would argue that the term has no place in a discussion of calibration. In practice however, 'accuracy' is the term most users relate to and can usefully express an expectation, and general specification in a manner we understand. Accuracy is a qualitative term and the number associated must be taken in the spirit of this concept and used for only indicative purposes.

4


Traceability: "Property of the result of a measurement whereby it can be related to stated references, usually national or international standards, through an unbroken chain of comparisons with stated uncertainties"

Another feature of the standard is that it must be shown to be able to reproduce the measurement that it claims to make with some degree of confidence. To do this, all the measurements in the system have to show traceability to higher level measurements and ultimately to national and international standards . The definition is given opposite and expresses the process by which a measurement can be shown to possess a calibrati on to a higher standard. It would be expected that the uncertainty of each calibration higher in the chain would have a smaller uncertainty at each step. It must be noted however that providing or claiming traceability makes no statement regarding the quality or uncertainty of the final calibration, and only satisfies one aspect of the quality requirements for an accredited calibration.

To express correctly the 'accuracy' of a standard or a calibration it is the 'uncertainty' which must be stated. Uncertainty provides a confidence that the determination lies within the quoted limits. The value of the limits and the confidence in these limits may vary. For flow measurement the confidence in the result lying within the quoted uncertainty is normally k=2 which is approximately 95% confidence. A full explanation of this concept is given in Reference 2. Every standard must be assessed for the uncertainty in the determination of its measured quantity, as indeed must the result of a calibration derived from the standard.

Uncertainty: "Parameter associated with the result of a measurement that characterises the dispersion of the values that could reasonably be attributed to the measurand"

The uncertainty quoted for a calibration or a standard will be evaluated from a detailed examination of all the components of the system, the use of the system and its history. It will specifically state for what parameter the uncertainty is quoted. This may be the quantity measured by the standard or the quantity passed through the flowmeter. It is stressed this is not the uncertainty of the calibration result. The resolution of the meter, the influence factors and finally the repeatability and linearity of the calibration results must all be included to provide the uncertainty of the calibration. The purpose of a calibration is to provide one component of the evaluation of the uncertainty associ ated with measurements from the meter in its final application. A responsibility remains with the end user to use the calibration uncertainty along with an understanding of the use of the meter compared with the calibration conditions, to provide the uncertainty in this final result. It is to be noted the 'error' shown by the calibration must be allowed for. All calibration results should have a stated uncertainty, and this should be on the certifi cate giving the result. The statement should be clear and unambiguous as to what is included and which result is quoted. Uncertainty can be expressed on the certificate as being the uncertainty of the measured quantity (flow, volume or mass), or the uncertainty of the meter's estimate of this quantity. The uncertainty of an equation fitted to the data may be included which includes meter performance during calibration. The uncertainty will not include the estimate of uncertainty at a different time or condition.

5


It is also worth noting that uncertai nty may vary across the flow range of the meter. The quantity of fluid collected by the standard may contribute to different uncertainties, or the meter performance may vary. In specifying the required uncertainty of a standard relative to that of the meter, it is general advice that the standard should have an uncertainty 10 times smaller than the requirement of the device to be calibrate d. Although this is a good principl e, in flow measurement it is often not possible to achieve this due to the high accuracy expectations of flowmeters. A standard with an uncertainty of a factor of three lower than the requirement may be all that can be achieved. In some situations, especially in field testing or in-situ calibrations, the uncertainty of the reference or standard may be poorer than the expected uncertainty of the meter. The achieved uncertainty of any calibration is larger than that of the reference or standard used. Therefore the uncertainty of the final measurement above the level expected prior to the calibration will be increased. When this occurs the calibration may best be described as a 'verification', i.e. the result is used to confirm the meter is (probably) performing within specification. 2. 4

Another option is to use electronic outputs through serial or 'field bus' links where many different options of output may be available incl uding display of pulses or mA. Thi s type of output cannot usually be gated dynamically to synchronise with a standard and care must be taken over update and processing times in the system. The result of a calibration is normally given in tabular form listing the measurements from the standard and the device. The extent to which the influence factors and raw data are given will vary depending on the calibrat ion specificati on. The presentation of meter and standard readings is not the most helpful way to interpret the result of the calibrati on. It is therefore normal to calculate some form of difference or factor. This performance indicator can be used to display the result in a manner which best reflects the performance of the meter across the flow range and enables the user to correct the meter output in future use. A number of different performance indicators are commonly used: K-factor: Used for meters with pulsed outputs proportional to quantity passed. K-factor is expressed as Pulses per unit quantity. (pulses/m3, pulses/kg etc) Factor: Where a flowrate-based output is found (flowrate, volts, frequency, mA etc), a meter factor may be computed.

A ccred it at ion

Accreditation is the process that a calibration laboratory undergoes to ensure the result provided to a client meets the standard of expectation as stated in the scope of the work. Accreditation is a process by which the equipment, technical methods, contractual methods, and quality of results are examined to give confidence to the client in the delivery of the final result. Strictly speaking accreditation may be carried out by the laboratory, but in practice it is normally carried out by a third party or the client. Before placing a contract for work a client may accredit a laboratory by examining its methods. If this process is carried out by the client, it is not uncommon to allow the laboratory to inform other clients of this accreditation suggesting that they may accept a result with a higher confidence without applying their own accreditation procedure. To avoid multiple accreditati on to different specificati ons, third-party accreditat ion is offered. This is normally provided by an accreditati on body approved at National Government level that will accredit laboratories to a defined quality standard. Within UK the accreditation organisation is the United Kingdom Accreditation Service (UKAS). Calibrati on laboratories are now accredited to the international standard ISO 17025. Inter-govern mental agreements have been established to allow the accreditation given in different countries to be recognised internationally. 2.5


Reporting the Result

To report the result of a calibration, the nature of the meter output has to be understood. Flowmeters may indicate flowr ate or quantity in a number of different ways. There may be a mechanical or electronic display indicating quantity or flowrate, or an electronic output based on pulses, frequency or current (mA). The output may be in the form of a differential pressure. Where the output or display is based on the flowrate (i.e. frequency, flowrate, differential pressure or mA), readings of the output must be sampled and averaged during each calibration test point. If the output is based on quantity passed (i.e. total pulses or display of quantity), the reading of the display has to be compared with a quantity of fluid measured by the standard. If the display is a visual one, clearly the flow has to be stopped to read the display, but if the output is electrical, electronic gating can coincide with a measurement from the standard.

6

where F is the meter factor, Q is flowrate, and V is volume: i identifies the value from the device, and s identifies the value from the standard. Error: Error is the difference between the indicated value and the value determined by the standard. Relative error, the error divided by the value determined by the standard, is normally used and expressed as a percentage.

It is very important always to define this equation as some industries use a different convention best described as the inverse or negative error. This is based on the standard minus the indicated values. Error can also be defined for meters with electrical outputs of pulses, frequency, volts or mA. In this case the indicated value is calculated from the output reading and the predetermined relationship (normally linear) between the output value and the equivalent quantity or flowrate. For instance it may be assumed that 20 mA corresponds to 10 l/s and 4 mA to 0 l/s. The value of Qi would be calculated from the measured current and the linear relationship.

7


Discharge coefficient (C): For differential pressure meters such as orifice plates and nozzles, the performance indicator C is usually defined. C is the ratio of the actual flow to the theoretical flow. The theoretical flow is however defined on the assumption that the minimum area of the jet of fluid in or downstream of the throat of the device is equal to the area of the throat (or orifice). Calculated C ratios vary from 0.5 to 1 depending on the device. Typical orifice plates give a C just over 0.6 and nozzles between 0.9 and 1. C is relatively constant for any particular device, only varying slightly over the flow range.


Low uncertainty will attract a higher price as will a large number of test points although a small number of test points does not reduce the price proportionally due to the overhead of fitting the meter and reporting. Being able to supply multiple meters of the same or similar size should attract reduced costs.

Flowrate: To express the performance of a device, its performance across its flow

A check list is given in Appendix 1. This shows the main pieces of information required when accepting or requesting a calibration. This ranges from the contact person information to the meter size and type through to the fluid and test specification. The connections and pipework are a vital piece of information to allow the laboratory to fit the meter.

range has to be expressed. The flowrate is normally express ed in terms of quantity per unit time with the units chosen to suit the application. Alternati vely a more complex flow-based parameter may be used such as Reynolds number which can generalise a performance curve by accounting for viscosity and density.

At all times good communications between the laboratory and the client are vital to ensure the result is applicable to the end purpose and that the calibration is carried out in a timely manner at the lowest cost to the client and acceptable profit to the laboratory.

2.6

Calibration Frequency

A common question to ask related to calibration is how often a flowmeter should be calibrated. This is important since calibration can be an expensive exercise, but a flowmeter with an incorrect reading can be even more expensive. Unfortunately there is no correct answer to the question. In some applications an answer is easy. A third party or a standard will define the calibration frequency. For most applications it is the user who must examine and define their own reasoning to justify calibration intervals. The decision is complex and will be based on a number of factors. The premise is to define a calibration interval that minimises the risk of an incorrect meter reading making a significant impact on the final process. If, for example, high flowrates of crude oil attract huge tax liabilities, weekly calibrations (in product and in-situ) will be needed. The product value is high; the risk of meter damage is high. Alternati vely metering waste water with a Venturi may only require 5 yearly inspection of the Venturi, yearly calibration of the differential pressure measurement device and no flow calibrati on ever. The risk of the pressure instrument changing is medium, the risk of fouling is medium, the product value is low, and the cost of a flow calibration prohibitive relative to the risk of a change in C. Other factors affecting the decision may be based on the history of the meter. This is shown by keeping a control chart of past calibrations, and the comparison of measurements within the process. If the process is closed for maintenance this will allow the meter to be calibrated at a lower cost than a specific shut down. A customer or partner in the process or a regulator may impose a third-party requirement. 2.7

Cost and Specification

The cost of a calibration is outwith the scope of this document. Some relative levels of cost and guidelines for the provision of the lowest prices can however be suggested. A laboratory providing calibrations of one type of flowmeter only and providing a limited range of sizes will be able to provide the lowest cost calibrations. Laboratories covering all types of meters and a wide range of sizes and fluids will generally be more expensive.

8

9



A number of criteria have to be met. Firstly, the pump and the flow circuit have to be arranged and designed to allow the flow through the meter to be stopped without damage to the pump or pipework. A pump bypass is usually fitted. Secondly no air may be left trapped in dead ends or T pieces as this will provide a spring effect causing the flow to oscillate when stopped, causing incorrect meter readings.

Part 2 - Calibration Methods A number of quite specific methods and systems are recognised for the calibration of flow devices. Although both follow the same principles major differences exist between liquid and gas methods. Two main differences exist between gas and liquid flow methods. The first is that liquids will remai n in an open container while gases need to be contained. Moreover, gases are highly compressible while liquids, for most practical purposes, may be assumed incompressible except for some small corrections. This causes fundamental differences in the approach to calibrating in gas and liquid, and for this reason they are considered separately.

3

CALIBRATION METHODS FOR LIQUIDS

3.1

Liquid Collection Methods

One characteristic of a liquid is that it can usually be contained in an open vessel, although if the liquid is volatile or hazardous, suitable precautions have to be taken. As a result, calibration standards are usually of the 'bucket and stopwatch' type. The bucket is a collection container, which is weighed or has a known volume, while the stopwatch is a method of timing the filling of the container. Static methods of calibration are based on collecting fluid in the bucket and determining its quantity by a static measurement; dynamic methods try to establish the quantity in the measure dynamically.

3.1.1 Standing start and stop method This method is generally preferred for meters used for measuring quantity of liquid, especially meters for batch quantities. The 'standing start and stop' method is the simplest method available and can be used for both high and low accuracy calibrations. The flow system is filled, all air purged and the required flowrate established. The flow is then stopped using a fast-acting valve. When the container is drained, the drain valve is closed, the flow started and the container filled. When the container is full the flow is stopped. The quantity collected is measured and compared with the meter reading. The time to fill gives the flowrate during the fill.

10

The flow has to be started and stopped as quickly as is practical to minimise the rise and fall time errors. Stopping too fast will create pressure fluctuations and 'water hammer'; these must be reduced by slowing down the valve until an acceptable performance is found. The stop valve should have an equal opening and closing time. The meter being calibrated has to have a fast response time to match the start and stop of the flow. T he test time has to be sufficiently long in comparison with the acceleration and deceleration periods to give insignificant error. This is illustrated with meter 2 showing the effect of microprocessor or an electronically enhanced meter where the sensor may start and stop quickly but the electronics take a significant time to 'catch up' with real time. Meter 1 shows a 'conventional' meter with a slow response to a change in flow. Many meters will however follow the flow response very closely. As the level of liquid downstream of the shut-off valve in the pipework may vary due to the surge in the flow, a constant level or transfer point must be established. This is done using a weir arrangement as shown in the schematic diagram for a top filling arrangement It is unlikely to achieve high accuracy with test times less than 60 seconds, and for large flows, with a slow operating valve, longer times may be required. Switches on the stop valve may be used to time the tank filling and gate start and stop pulse counters.

11


What is discussed is a standing start and finish method based on a gravimetric or weighing tank. If a volume tank is used, filling is usually from the bottom, and the valve closing time is slow or stepped at the last stages to ensure the level in the tank lies in the measuring neck.


A three port valve can be used as the diverter. As this will probably introduce a pressure surge at the cross over point and a different back pressure (hence flowrate) in the two positions it i s not normally used for high accuracy systems.

3.1.3 Dynamic methods 3.1.2 Flying start and finish method This is sometimes called the diverter method. In this method the flow through the meter is not stopped, but the flow is diverted between a return to the supply and the container. A switch on the diverter mechanism starts and stops a timer and counter. These time the filling of the collection measure and count pulses from the test device. In this method the key to accurate measurement is a clean separation between fluid entering the container and fluid returning to the supply. This should be accomplished without any change of flowrate through the device. For this reason the flow into the diverter is normally conditioned by creating a long thin jet impinging on a splitter plate. This will be open to atmosphere ensuring no change in pressure occurs which may give a change in flowrate when diverting. The diverter mechanism is operated as quickly as possible to reduce 'timing errors' to a minimum. With a well-designed divert er, test times down to 30 seconds can be achieved with divertors operating at less than 0.5 seconds.

Dynamic methods are techniques where, not only is the flow continuous, but the standard also measures the quantity, without stopping the flow passing through it. The static methods outlined above can be modified to make a dynamic measurement technique possible . The changing weight or volume in the container can be detected and used to trigger the test point measurements of time and meter output. Flow is establish ed into a tank with the drain open. To initiate a test, the drain is closed and the rising level or increasing weight is detected and used to initiate a test point by starting a timer and reading the meter. When the tank is full, and after a signal has been sent to stop collection of data, the tank drain is opened and the tank emptied. The drain has to be large enough to allow the full flow to pass through when open. Although there are exceptions the technique is really only used for low accuracy calibrations as repeatability is normally 1 per cent at best. The method is sometimes employed using very large tanks in the field where long test times can give reasonable accuracy for in-situ calibration or verification of a meter. Other dynamic techniques include the use of reference meters, pipe provers or other reference devices such as sonic nozzles or insertion meters. All are discussed later.

3.2

Measurement Methods

Two principles are used to measure the quantity of liquid in the container. The volume can be defined, or the mass can be determined.

3.2.1 Gravimetric calibrators The main source of uncertainty lies in the timing error. This is shown diagrammatica lly. T he 'hydraulic centre' of the diverter is found by calibrating a high quality reference meter at a constant flowrate. Long and short diversion times are used. The difference between the calibration using short diversions and that using long diversions defines the timing error. The timing sensor is moved until the difference between calibration points is minimised. This is repeated at different flowrates and a best compromise position found. The residual scatter and difference between long and short diversion calibrations gives the uncertainty due to timing error. Flying start and finish methods are used primarily for meters with slow response times and where flowrate is the primary measurement rather than quantity passed. Meters with visual displays cannot be calibrated by this method.

12

A flowmeter can be calibrated gravimetrically by weighing the quantity of liquid collected in a vessel. The vessel is weighed empty, again full, and the difference calculated. This gives the weight (in air) of the fluid collected. As the quantity collected must be expressed as mass, then converted to volume, the weight collected has to be corrected for the effect of air buoyancy. As a weighing machine is calibrated using weights with a conventional density of 8000 kg/m3, and the fluid collected will have a significantly different density, the up-thrust of the air on the tank will have a signific ant effect. This amounts to around 0.1 per cent for water. The correction is given below.

where M W rair rf rw

is the mass, is the measured weight is density of air, is the density of the fluid, and is the density of the calibration weights (8000 kg/m3).

13


The term in the large brackets is called the buoyancy correction factor and can be predetermined and treated as a constant in some applications where the highest accuracy is not required. To determine the volume, the mass collected is divided by density. The density is determined at the flowmeter, to give the volume passed through the flowmeter. Density can be measured using a densitometer but is more often calculated from the temperature and pressure at the meter and knowledge of the fluid properties. If a densitometer is used, any difference in temperature (hence density) between the densitometer and the meter has to be allowed for. The weighing machines used must be calibrated using recognised standards of mass. Normal platform machines fitted with steelyards provide measurements of weight to high accuracy, provided they are carefully maintained. Electroni c 'force balance' machines provide a better performance with an electronic output. Gyroscopic weighing gives the ultimate resolution but probably exceeds the requirements of flow measurement. Combined with the other uncertainties in density etc, uncertainties down to 0.03% can be achieved. Strain gauge load cell weighing techniques may be used, but generally will not provide an uncertainty capability much better than 0.1%.


For all volumetric methods, a number of corrections and conventions have to be observed due to the expansion and contraction of both the standard, and the device being calibrated. The expansion and contraction of the fluid between the standard and the flowmeter have also to be recognised. Expansion due to temperature is the most important, but expansion in a pressurised system must also be accounted for. Reference volume tanks, and pipe provers, have their volume defined at a stated reference temperature (and pressure). Normal reference temperatures are 15 or 20°C. Other references can be defined for special purpose s to minimise the size of corrections. Similarly reference pressure is normally atmospheric pressure (1.01325 bar(a)). How this standard or base volume is measured and defined is outwith the scope of this document but this provides the starting point for the calibration. The volume contained in the standard at the temperature of the standard during a calibrati on is not the base volume. It is the base volume increased or decreased by the expansion of the material of the standard. As the container makes up a volume, it is the cubical expansion of the material used. The equation is fundamentally

3.2.2 Volumetric calibrators The measurement of the quantity of liquid collected may be carried out volumetrically, i.e. by collecting a known volume of liquid in a container. In the volumetric method the standard vessel takes the form of a container with calibrated volume. Normally this will be a pipette with conical ends to facilitate drainage and to reduce the risk of air entrapment. The neck of the pipette is normally fitted with a sight glass and a scale marked in volumetric units. A typical volumetric tank is shown below. The tank is not itself a primary calibration device and its volume must be determined by calibration. This can be carried out by weighing the water contained in the vessel, or, for larger vessels, carried out using smaller volumetric measures which are themselves traceable to national standards by weighing methods. Volumetric systems are normally used with standing start and finish methods due to the difficulty of diverting flow into the tank end. The technique gives a very high level of repeatability but is by necessity lower down the traceabil ity chain. Tank volumes are expressed at a reference temperature (normally 15 or 20°C) and corrections have to be applied for the expansion of the material of the tank. Drainage time (after the tank is empty) is vitally important, as liquid clingage to the wall can be significant. Each tank has a calibrated drain time and this must be maintained. For this reason high viscosity liquids above 10 cSt start to give problems of both accuracy and repeatability due to the unpredictable quantity of liquid left attached to the walls of the tank.

14

where VS is the volume contained, αS is the linear expansion of the material of construction of the standard (prover or tank), t S is the temperature of the standard and tR is the defined reference (base) temperature. To define the volume of fluid which has passed through the flowmeter into the standard, the expansion of the fluid due to the temperature difference has to be calculated.

where V is volume passed through the meter, α is the cubical expansion of the fluid, t S is the temperature of the standard and t M is the temperature of the meter. Similar corrections have to be applied for pressure. It is noted however that a volume tank will always be at atmospheric pressur e. It is sometimes found more practical to reduce the volume of everything to that at the reference temperature rather than correcti ng to actual conditions and then calculating the error or k-factor. Both approaches should give the same answer. In the oil industry these corrections are calculated individually in a formulaic manner and are given 'Correction factor' nomenclature. Ctsp = Temperature correctio n for the Steel of the Prover (standard). Cplp= Pressure correction of the Liquid for the Prover etc. The correction of the flowmeter to a reference condition is contentious. If this is done, the calculations are the same as above. The difficulty is defining the expansion coefficient . Flowmeters are complex devices where not only does the volume or area of the device change with temperature, but internal clearances and friction change to give a rather ill defined coefficient. For this reason it is normally advised not to apply correcti ons to the flowmeter, but to quote the result at actual conditions . Some industry practice does however apply corrections to defined coefficients. 15


Meters can be calibrated against tanks with a repeatability of better than ±0.02% and uncertainties of 0.05%. This of course assumes the volume is adequate and the standing start and finish method is employed.


Provers are often installed at metering stations when the product is of high value and dedicated to the application, or sometimes are mobile and taken to different metering stations. Four main classifications of provers are found.

3.3.1 Unidirectional sphere prover 3.3

Pipe Provers

The pipe prover also is a volume method. The difference between a volume tank and a prover is that the prover may be used at pressure and the calibration is dynamic. This will require a correction of the base volume at reference temperature and pressure for both temperature and pressure related to the expansion of the pipe and the fluid.

A unidirectional prover has a displacer which only travels in one direction along the pipe. The displacer consists of an elastomer (neoprene, viton, polyurethane, etc) sphere which is hollow. The centre is filled with liquid and pressurised to inflate the sphere until it is larger than the pipe bore. A typical inflation is around 2 per cent larger than the pipe bore. When the sphere is inserted into the pipe it takes up an elliptical shape and makes a good seal to the pipe wall. The pipe itself is a l ong length of steel pipe with a smooth bore. The internal surface is usually coated with phenolic or epoxy resin to provide a smooth low-friction lining and to protect against corrosion. As the pipe can be extremely long, it is usually constructed in a series of loops. The radius of the bends is chosen to allow the sphere to pass without either sticking or leakage.

Pipe provers probably provide the best calibration devices for truly dynamic calibration. They are used in a sealed system, provide high accuracy calibrations, and can be used in-situ as well as in laboratories.

A volume tank or seraphim being levelled prior to use (Courtesy of Alderley Systems)

It is to be noted that the term 'proving' is used extensively in the oil industry for the set of operations to 'prove' the accuracy and fitness for purpose of a flowmeter. It is recognised that the term is synonymous with calibration.

The pipe prover principle is shown opposite. A length of pipe is fitted with switches such that the volume between the switches is known. If a displacer or pig is introduced to the flow, the time it takes to travel between the switches will give a measure of the flowrate. If the switches are used to gate a pulse counter, totalising pulses from a flowmeter, a measure of the meter factor (pulse per litre) can be found. The technique illustrates the ingenuity brought to bear on a calibration problem. The first prover was a mile long pipe l inking two oil refineries in dispute over the flowmeters measuring product. With no ability to calibrate the meters independently, the length and diameter of the pipe were estimated and the time for a cleaning 'pig' to travel the distance provided an adequate measurement of volume to verify the transfer meters.

At each end of the calibrated length of pipe a detector switch is located through the pipe wall. This usually takes the form of a plunger triggering a switch when the sphere passes under it. Unidirectional Prover (Courtesy of Alderley Systems)

At the end of the prover is the sphere handling valve. This arrangement is designed to hold the sphere. At the start of a test the sphere is launched into the flow and carried round the loop. At the end of the loop the sphere is captured and returned to the launch position ready for another run. The valve also has to allow for removal of the sphere. The design of the valve is critical, and must not only be leak tight but must have mechanisms to prove it is leak tight.

This concept has been refined to give the measuring device called a 'pipe prover'. These devices are used extensively to measure all types of high value fluid from LPG to high viscosity crude oil and are produced in all sizes from 2-48 inches diameter. The fluid remains contained and sealed in the system, the calibration fluid can be the product at normal conditions, and calibration can take place without interrupting the process. 16

17


3.3.2 Bi-directional sphere prover Because of the complexities of sphere handling and to reduce the turn round time of the sphere, the bi-directional prover was developed. Similar in layout to the unidirectional type, the main difference is that flow can circulate around the loop in both directions. A four-way valve, of very high integrity, changes the flow path without breaking the flow. The sphere is held in special end chambers. These are designed to launch the sphere and absorb the shock of capture. One chamber also provides a means of removing the sphere. Note from the drawing that two switches are provided at each end. This provides better integrity of the measurement by giving redundancy and a means of checking results by developing four separate volumes.

3.3.3

Piston provers

For difficult fluids, which may damage a lining material, or leak past the conventional sphere displacer, a piston may be used. The pipe must be straight to allow a piston to pass. The pipe is normally a smooth, honed pipe of stainless steel or plated carbon steel. The Bi-directional Prover with Associated Flowmetering Skid displacer is a piston with multiple (Courtesy of Alderley Systems) seals. Switches can be conventional plungers or high integrity, noncontacting types. These provers are bi-directional using a four-way valve. This type of prover is not so common for the usual liquid applications but finds a particular application with LPG, LNG and other difficult high value products.

As a small volume will not allow the generation of enough pulses from the flowmeter a technique called pulse interpolation is used to increase the resolution of the pulse counting. For the oil industry the minimum number of pulses has to be 10 000. If fewer are collected, pulse interpolation may be used providing the signal stability is suitable. Examples can be shown where down to 100 pulses are collected during one pass of the displacer.

3.3.5 Operation and calibration of a prover To use a prover, the flow is directed through the prover and then the meter. The displacer is launch ed into the flow. When the first detector is actuated, a counter and timer are started. When the second detector is actuated the timer and counter are stopped. From the known volume between the detectors, the pulses counted and the time, a calculation of volumetric flow rate and K-factor are derived. In oil industry standards 3 or 5 calibrations are carried out at each flowrate specified. These have to fall in a range of 0.02%.

Any prover but smaller than a 'conventional' design for any particular application. Usually this will mean the prover will have a volume too small to allow the collection of enough pulses to give insignificant meter resolution uncertainty (<10000 pulses is the standard criteria). In other respects it will be of the same design as a 'conventional' prover but require pulse interpolation to be employed to improve meter resolution.

A prover is a volumetric calibrator and the calculations have to take into account the correcti ons for temperature and pressure of the prover and liquid. Codes of Practice governing the design, calibration and use of pipe provers, including the small volume versions, are available from ISO, Institute of Petroleum and API.

Depending on the definition, this type of prover can be one of two concepts.

b)

The design illustrated shows a unidirecti onal piston prover. To allow a short length and retain accuracy the optical detectors are mounted external to the pipe. The piston has an integral 'poppet' valve, which allows the flow to pass through the piston when held open. An external rod allows the piston to be pulled, using hydraulic pressure, to the upstream end of the cylinder while holding the valve open. Releasing the hydraulic pressure allows the valve to be shut by a combination of a spring, gas pressure on the end of the rod, and the force of the flow. The flow carries the piston down the pipe. At the downstream end, the hydraulic pressure is restored and the valve opens and is returned to the start position. A second rod carries flags which, through optical detectors, give very precise start and stop signals across the measured volume.

The volume of the prover between the switches is determined by calibration. This will be called the 'base vol ume' and is quoted as being the volume at 15°C (or other reference temperature) . The volume is found by displaci ng water into a volume (or mass) standard measure. For larger provers, a reference meter is used with water or product to measure the volume and the meter calibrated as part of the same operation against a volume measure, small pipe prover or small volume prover.

3.3.4 Small volume (or compact) provers

a)


A custom-designed pipe prover with a volume about one-tenth of a conventional design for the same duty. It is normally a piston prover.

18

3.3.6 Pulse interpolation Mentioned above is the concept of pulse interpolation. This technique effectively increases the resolution of a pulsed output by estimating the fraction of a pulse missed at the beginning of a test and that gained at the end of a test.

19


This can be done electronically using frequency multipliers, or by pulse timing using two or three techniques. By far the most common method is the double chronometry timing method illustrated below. To estimate the fraction of a pulse lost or gained at the start and finish of a pass, the whole number of pulses is counted. This number is multiplied by the ratio of the time between the switches to the time between the first pulse after the start switch and the first pulse after the stop switch.

The technique works well when the pulses have a constant frequency or period. If the period of the individual pulses varies by more than 5-10% lack of repeatability is found.

4

CALIBRATIONS FOR GAS FLOWMETERS

The choice of calibration method for any particular flowmeter is governed by the meter type, the ranges of flow and flow conditions, the pressure and the accuracy of calibration required. In general all the methods have analogies with the liquid methods. The main difference between the calibration of a gas flowmeter and a liquid device is the compressibility of the fluid. When calibrating a gas flowmeter, the temperature, pressure and hence volume of gas measured by the standard will be different. Correctio ns to common conditions must be made. It is often best to convert to mass flow at each stage and then back to the conditions at one position or to 'standard' conditions.

4.1


Some piston pipe provers have been produced for gas service. These are generally for higher pressures where the gas density is high or the piston is driven or assisted in some manner. There are however, a number of specialised and proprietary piston provers available for use with low pressure gas where the piston is driven by a precision lead screw, taking out the need for the gas pressure to overcome friction.

4.1.1 Mercury seal prover For low flows, mercury seal provers use a very light displacer in a vertical glass tube. The piston runs vertically upwards in the tube. The seal is a mercury ring formed in a recess in the piston. This reduces friction to a minimum.

Soap film burettes are again a form of pipe prover used for both calibration and measurement. In this case, a glass tube is vertically mounted with a reservoir of soapy water below the gas inlet. Gas flow from the meter on test passes through a vertically mounted burette. As the gas enters the burette a soap film is formed across the tube and travels up the tube at the same velocity as the gas. By measuring the time of traverse of the soap film between graduations at either end of this accurately calibrated burette the rate of flow of the gas may be obtained. What is created is a pipe prover with the displacer formed by the soap film. This method is usually used to measure gas flows within the range 10-7 to 10-4 m3/s at conditions close to ambient, and under very carefully controlled conditions reference flows can be determined to within ±0.25 per cent using soap film burettes.

Displacement Methods

A number of proprietary standard devices are used for gas calibration based on the principl es of the pipe prover. The biggest drawback of any prover system for gas is the friction generated by the displacer seal. This friction will require the gas to compress until the pressure difference overcomes the friction. Variations in the friction can prevent the displacer moving smoothly, causing sticky or juddering movement hence giving poor results.

20

4.1.2 Bell provers The 'Bell prover' is the standard for calibrating low-flow gas meters such as domestic gas meters. A cylinder (or bell), open at the bottom and closed at the top, is lowered into a liquid bath. The weight of the cylinder is supported by a wire, string or chain and counterbalanced by weights. A smaller counterbalance is fitted to compensate for the changing buoyancy as the cylinder is submerged. All pulleys etc, are on low friction bearings. By altering the counterbalance weight, a pressure can be generated in the cylinder.

21


A pipe passing through the liquid communicates with the trapped volume, and, as the cylinder is lowered, gas is displaced from the cylinder to the meter on test. By timing the fall of the cylinder and knowing the volume/length relationship for the cylinder, the volume flow of gas through the meter may be determined and compared with the meter reading. By closing a valve leading to the meter, and opening a valve (not shown) from a gas supply, the cylinder can be returned to the start position. Original designs used water as sealing liquid. As this saturates the air as it evaporates, giving concerns over humidity measurement, most bell provers are now filled with low-vapourpressure/ low-viscosity oil.

The expression for the mass flowrate of the gas is:

where Cd is discharge coefficient, C* is critical flow factor, and A is area of the nozzle throat. PO and TO are the upstream pressure and temperature. The mass flowrate under sonic conditions is independent of downstream pressure and temperature and dependent only on the geometry of the nozzle, the properties of the gas, and the upstream pressure and temperature. This feature makes the device particularly suitable for calibrating meters, like some rotary displacement meters, which can introduce pressure pulsation into the flow. A standard toroidal throat sonic Venturi as specified in an ISO standard is shown. Other designs based on conical entries or parallel throat orifice plates can be used but with larger pressure drops.

In order to minimise expansion or contraction of the gas, the liquid, gas and air temperatures should not differ by more than 1°C. Errors can also arise due to incorrect compensation for change in buoyancy of the bell as it is immersed and the fact that the gas is not fully saturated. At present, for flows up to some 10-2 m3/s, bell provers can be used to measure flows to within ±0.2 per cent if strict precautions are taken to minimise the errors.

4.2


Critical Flow Venturi-Nozzle (Sonic Nozzle)

Although not a primary method of calibration sonic nozzles can form part of a system when combined with primary methods. Sonic nozzles also provide the reference system for many calibration facilities where their stability requires infrequent calibration of the nozzle. If the pressure drop between the inlet and the throat of a nozzle or restriction is increased until sonic velocity is reached at the throat, then for a given value of the upstream pressure and temperature, the mass flowrate through the nozzle will be constant.

22

One disadvantage of the critical flow Venturi-nozzle is the large pressure drop, which is normally much greater than that for subsonic nozzles or other flowmetering devices. Moreover, an accurate knowledge of the thermodynamic properties of the gas is required, and this may cause difficulties in gases such as natural gas where the composition may be complex and variable. The device is however particularly suitable for calibrating flowmeters in high-pressure gas flows at flowrates where the throat Reynolds number exceeds 10 5: uncertainties of 0.2 per cent may be achieved. The large pressure drop can in some situations lead to long times being required to establish stable temperature conditions at the test meter. One feature of the nozzle is that, as pressure changes cannot travel faster than the speed of sound, the nozzle effectively de-couples any downstream pressure changes from upstream conditions and hence provides a constant mass flow when upstream pressure is constant. This applies even when downstream pressure changes. Although for measurement purposes a nozzle is used to de-couple up and downstream pressures, any suitable restriction, such as a valve run at sonic velocity, performs the same function. It is this feature that allows nozzles to be calibrated against primary mass or volume static measurement systems, and in fact nozzles are integrated to the operation of primary standards.

4.3

Gravimetric / P.V.T Methods

A simple primary gas calibration system is outlined. This is similar to a liquid collection method and can be found as either a volumetric or a gravimetric system. For gas, the collection tank is sealed and gas is introduced through valves. In gravimetric systems, the tank has to be disconnected from the supply to allow weighing. For volumetric systems the pressure, temperature and tank volume must be known. An obvious drawback of such a system is found. As gas enters the tank, the pressure rises. As the pressure rises, the flow rate must reduce. To get round this problem, these systems are usually used in conjunction with a sonic nozzle to 'de-couple' the pressure. As explained in Section 4.2, the mass flow through a critical nozzle is dependent on the upstream pressure and independent of the downstream (tank) pressure. This allows the mass flow through an upstream device to remain constant while the tank fills. 23



Alternatively the nozzle can be used as a transfer device to calibrate a lower pressure meter whereas it is itself calibrated against the primary tank. The measuring vessel may be a sealed vessel which is weighed. To give some sizes, a vessel may weigh some 2.5 tonnes and filled to 30 bar will contain about 40 kg of gas. Alternatively, the vessel may be of known volume. In this case by measuring the pressure and temperature and knowing the gas properties, the quantity of gas collected is derived. Due to gas compressibility effects, problems encountered in maintaining a closely controlled flowrate are considerably greater in gas flow systems than in liquid flow systems. Also since gases are very much less dense than liquids further difficulties are encountered in accurately weighing the diverted mass. The NEL gravimetric system can be used to measure air flowrates of up to 4 kg/s, at pressures of up to 50 bar, to within an estimated uncertainty of 0.15 per cent.

5

For the dilution method a tracer fluid which is detectable in low concentrations, is injected into the flow (see the Figure 3 opposite) at a known rate q (m /s). The mainstream flow is then sampled at a distance downstream of the injection point far enough to have allowed homogeneous mixing to have taken place, and the concentration, C, of the tracer is measured. Since the rate q is usually very small compared with the main flowrate Q, the flowrate can be derived from

ON-SITE CALIBRATION METHODS

In this Section three methods, suitable for calibrating flowmeters on site, are described. These are in addition to using any of the methods previously described. Generally these methods are used to calibrate meters where the standard laboratory methods are not suitable due to the product, the meter size, or an inability to stop the process. Generally they are methods which do not give the best uncertainty, and may in some cases provide poorer uncertainty than that expected of the meter. If doubt exists these may be the methods which have to be employed to verify meter performance.

5 .1

In transit-time methods a pulse of tracer fluid is injected into the main flow stream, and the time taken for the tracer to pass between two detection points is noted. If the volume of pipe between the detectors is known the volumetric flow can be determined. At present, tracers used in this method are usually radioactive isotopes, and radiation detectors are used to determine the tracer transit time.

Tracer methods are not suitable for sluggish, slow moving flows. In dilution methods the main source of error occurs in obtaining accurate determination of the tracer concentration, and in tracer velocity methods difficulties are encountered in determining the volume between detectors. However it is claimed that, by incorporati ng recently developed radioactive techniques, an experienced team can determine the flowrate under the most favourable conditions to within 0.5 per cent.

5.2

Insertion Meters In these methods, the flowrate in the pipeline is estimated by measuring a number of point velocities at discrete positions in a cross-section of the flow, and then integrating these over the crosssection. The device used to measure the point velocities may be a pitot tube, insertion turbine or an insertion electromagnetic meter.

Tra ce r M et ho ds

Tracer techniques can be divided into two methods: a) b)

Transit time (velocity methods), and D il ut ion met hods .

The main disadvantages of these methods are that they are time consuming and that serious difficulties are encountered with unsteady flows. For gas velocities in the range 0.3 to 3.0 m/s uncertainties of 4 per cent are attainable using vane anemometers and for velocities in the range 6-120 m/s uncertainties of within 2 per cent can be achieved using pitot tubes. For water flows in large diameter water pipes, care must be taken to ensure the measurement is unaffected by flow disturbances, bends or T pieces. Of course, one of the issues is how to calibrate the insertion meter.

24

25


5.3


Clamp-on Ultrasonic Meters

Clamp on ultrasonic meters operate by knowing the internal diameter of the pipe, and measuring the velocity across a diameter by using a time of flight ultrasonic meter. The transducers are clamped on to the outside wall of the pipe and measurements taken. Many factors have to be considered such as flow profile, pipe material and internal condition and fluid properties. Uncertainti es of no better than 1-10% can be assumed but 5% is a more probable expectation.

6

EXPECTATIONS FOR A CALIBRATION

The calibration curve of a meter applies to that meter only, operating under the conditions in which it was calibrated. If in service these conditions are changed the calibrati on may not apply. What then are the real orders of uncertainty which might be reasonably obtained from calibrated meters? First, the meter cannot be calibrated to an uncertainty level better than its repeatability and the uncertainty of the standard. The random uncertainties of a calibration can be calculated statistically from the results of a calibration, whereas the systematic uncertainties can only be estimated from a knowledge of the calibration system and its method of traceability. The absence of systematic errors can often only be checked by an intercomparison of facilities using a transfer standard. Liquid flowmeter calibration facilities, having a known traceability path, should be able to measure flowrates to uncertainty levels between 0.5 and 0.05 per cent depending upon the size, cost and complexity of the system, and to measure volumes with a somewhat higher accuracy.

Appendix 1 Calibration Check List

Calibration systems for gas flowmeters should be able to measure flowrate to uncertainty levels of 0.5 per cent. A primary gravimetric system such as that at NEL can be used with an uncertainty of 0.15 per cent, and when critical nozzles, directly traceable to the gravimetric system, are used as working standards, the uncertainty level drops to 0.25 per cent.

26

27



APPENDIX 1 CLIENT CALIBRATION CHECK LIST 1.

Typ e of met er Turbine/DP/Coriolis/Ultrasonic/etc

2.

Make/Model

3.

Size of meter (Length, diameter) (weight) (other sizes)

4.

Typ e of Fl ui d Water/Oil/Gas(air?)/Multiphase (get details i.e. what is the viscosity of the oil etc)

5.

Flowrate/Flowrange (remember to note Units!)

6.

Operating Pressure (especially for gas)

7.

Operating Temperature

8.

Signal Output: Pulsed/mAmps Pulsed = max frequency Resolution: is it a scale, pulses/unit What's the electrical characteristics (volts etc)

Appendix 2

9.

K-factor (below 4 no verbal quote before checking)

References

10.

Is the pipework included? Are all electronics included?

11.

What uncertainty is required

12.

What Flanges: screw etc Are the flanges raised or RTJ Some standard fittings are: ANSI 150 PN10 BSP ANSI 600 PN16 BSP(T) ANSI 300

NPT

13.

Measured Points required e.g. (3 @5 flowrates) (1 at 10 flows) etc

14.

Timescale required

15.

Have you had the meter calibrated before

Contact Name: Company: Address: Tel:

Email:

Fax: 28

29


APPENDIX 2 REFERENCES 1

International Vocabulary of Basic and General Terms in Metrology (VIM). BS PD6461, 1995 (Also ISO).

2

Guide to the Expression of Uncertainty in Measurement 1995 (GUM). BIPM/ISO/OIML/IEC/IFCC/IUPAC/IUPAP.

3

ISO 12916. Liquid Hydrocarbons - Dynamic Measurement - Volumetric Proving Tanks Or Measures.

4

ISO 8222. Petroleum Measurement Systems -Calibration - Temperature Corrections For Use When Calibrating Volumetric Proving Tanks.

5

ISO 7278-2. Liquid Hydrocarbons - Dynamic Measurement - Proving Systems For Volumetric Meters - Methods For Design, Installation and Calibration of Pipe Provers..

6

ISO 7278-3. Liquid Hydrocarbons - Dynamic Measurement - Proving Systems For Volumetric Meters - Pulse Interpolation Techniques.

7

ISO 7278-4. Liquid Hydrocarbons - Dynamic Measurement - Proving Systems For Volumetric Meters - Guide For Operators of Pipe Provers.

8

ISO 9300. Method of Measurement of Gas Flow By Means of Critical Venturi Nozzles.

9

ISO 91-1. Schedule for Petroleum Measurement Tables.

10

Expression of Uncertainty and Confidence in Measurement. NAMAS M3003, UKAS, London, 1997.

11

NEL Report No 367/99. ISO 5168 (Draft 5). Measurement of fluid flowestimation of uncertainties. A Report for NMSPU, DTI, London. September 1999, East Kilbride, Glasgow, National Engineering Laboratory.

12

Petroleum Measurement Manual. Part X: Meter Proving. Section 3: Code of Practice for the Design, Installation and Calibration of Pipe Provers.

13

ISO 4185:1980. Measurement of Liquid Flow In Closed Conduits -- Weighing Method.

14

ISO 9368-1:1990. Measurement of Liquid Flow In Closed Conduits By The Weighing Method -- Procedures For Checking Installations -- Part 1: Static Weighing Systems.

15

ISO 8316:1987. Measurement of Liquid Flow In Closed Conduits -- Method By Collection of The Liquid In A Volumetric Tank.

16

ISO/TR 5168:1998. Measurement of Fluid Flow -- Evaluation of Uncertainties.

30

NEL, East Kilbride, Glasgow, G75 0QU Tel: +44 (0) 1355 220222 Fax: +44 (0) 1355 272999 e-mail: [email protected] Web site: www.nel.uk

Best Practice Guide Calibration of Flowmeters

Recommend Documents