FRI Internal Design vol 3

Issued: 01/15/1998 Revised:

TROUBL ESHOOTING TECHNIQUES

3.01.1

also exhibit vapor maldistribution and in many cases this problem is exacerbated as the vapor flows up the column. Effects such as vapor cross channeling can affect performance in many trays in a column.

4

Hydraulic Limitations

Hydraulic limitations can be experienced in various ways: flooding by inadequate vapor handling capacity, flooding by inadequate liquid handling capacity, excessive pressure drop, excessive entrainment, etc. We will address each one of these.

5

Inadequate Vapor Capacity

Columns are designed to accommodate certain amounts of vapor and liquid flow. Oftentimes they are sized for vapor capacity and then checked for liquid capacity. Generalized methods are used for predicting flood points, load points, and so on. The methods are based on the geometry of the column internals, the properties of the fluid systems, and the practice of the designer in allowing for unforeseen needs for future operating conditions. When a column falls short on capacity, there must be some criterion of what the capacity should be. For vapor flow, the criterion is usually a capacity correlation or a nameplate number. A significantly lower observed capacity can indicate one or more of the following. • • • •

The correlation was not used properly. Too much was expected of the correlation. Operation of the column has introduced agents which could cause foaming. The column internals are not arranged according to design.

Check the capacity correlation applicable ranges - The reasons for limited vapor capacity bear further discussion. So far as proper use of the correlation is concerned, one must realize that all correlations for capacity prediction are largely empirical. They are often based on experimental data for hydrocarbon systems of low viscosity (e.g., the FRI flood data). Unusual properties such as high viscosity and very low surface tension could invalidate the correlation. Also, the correlation involves a curve fit to scattered data, and the probability of premature flooding is inherent in such a fitting procedure. Still another pitfall can be encountered at high liquid loads, beyond the range of the correlations. Watch for foams and the reliability of foaming tests - Unexpected foaming is often a cause of premature flooding. Impurities with surface activity can be generated in the process and cause a foam/froth buildup that results in heavy entrainment or flooding. Suppression of the foam by anti-foam agents may not be feasible in conventional distillation systems. On the other hand, such agents are often used in recirculating systems such as absorbers and extractive distillation columns where the agent can be retained in the solvent phase. Tests for foamability have not been found particularly reliable unless they can be performed at the actual tower conditions and using the actual vapor and liquid compositions. When foaming is anticipated in new designs, allowance can be made in the form of extended tray spacing and/or oversize downcomers. But when it occurs unexpectedly in operation it may be necessary to use a different type of contacting geometry. Packed column contacting is not subjected to the high hole or slot velocities of tray columns, and this seems to make a difference in foam buildup. Little is known of the ability of structured packing to break up foam, but we suspect that they can handle foam better than trays. Beware of second liquid phase formation or appearance - Distillation trays with two liquid phases can

Page 8 of 17



3.01.1

be subject to foaming problems. Studies indicate that when the liquid phases are well established, the total liquid is well-behaved. However, if a tray column operates in the region where the second liquid phase is being formed, there is the distinct possibility of foaming. Check internals installation and/or condition against design - Another reason for limited vapor capacity is that the column internals are not arranged according to design. This can mean that a constriction has been placed to impede liquid flow and thus precipitate a flood condition that is evidenced by limited vapor capacity. As an example, if the clearance under the downcomer baffle is found to be less than the design value; just one location of malfitting can cause the entire column to flood. Another example is that because of support elements the open area (hole area) for a sieve tray is less than design, giving higher hole velocities than expected; this can lead to heavy entrainment or aggravated foaming problems. NOTE: When improperly designed internals are identified and replaced, it is crucial to update drawings so the incorrect drawings won’t be used in later efforts. Are all trays and internals where they are supposed to be - Column internals may be misarranged because of operating upsets. It is not uncommon for trays to be found completely dislodged from their original positions, and in some cases jammed together with resulting vapor flow constrictions. This situation will normally be evidenced in pressure drop and efficiency measurements. An external scan, using gamma ray absorption, can be used to detect building of metal as well as of liquid. The solution to the problem is, of course, to shut down and make repairs. Has fouling occurred and is it a factor - Fouling of the internals of a tower can lead to premature flooding as well since it can reduce available flow areas and increase pressure drops. Do the plant heat and material balance data support the design values for internal flows - Finally, we should recognize that internal flows in an operating column may be higher than thought, because of faulty meter readings, inaccurate simulation results, or erroneous heat and material balances. A column may be handling its design flow rate, but the instruments or calculations don’t indicate that it is.

6

Inadequate Liquid Flow Capacity Check for liquid flow restrictions and excessive pressure drop - Limitations in the ability of tray columns to handle the required liquid flows appear in the downcomers but can be caused by problems in the active area of trays. Excessive downcomer backups can be caused by large liquid flows or by large pressure drop across the trays. Lack of adequate froth collapse time, excessive pressure drop for sufficient flow under the downcomer baffle, foreign materials being left in the downcomer by construction people are examples of problem causes. Dislodged valves from valve trays can accumulate in downcomers and impede flow. A particular point of concern in the column is the bottom seal pan, if arranged such that the reboiler return vapor-liquid mixture can impinge against it. Another point of concern is liquid side draw points where vortexing can occur if sufficient liquid head or proper vortex breakers are not provided. Gamma scans are useful in locating trouble points. Check for restrictions to liquid flow in packed columns - For packed columns, liquid flow restrictions can occur when the distributor or support plate becomes clogged, or when ceramic packings break up and accumulate in some part of the bed. If dirty liquids or slurries are being handled, plugging is certainly possible. Liquid distributors are usually designed to allow for overflow if dispersers become plugged, but the result of such operation is a lowered efficiency. Mention has been made of the use of gamma scans for detecting zones of liquid buildup in an operating column. This is now quite commonly done, and there are companies who specialize in the scanning service. The source/detector combinations can be

Page 9 of 17



3.01.1

moved up and down the column, with varying directional orientations. The results are often indicative of flow limitations or vapor-liquid maldistribution problems. Meaningful scans can be made of columns containing metal structured or random packings.

7

Excessive Pressure Drop

Normally the problem here is excessive pressure drop, especially in those columns where pressure loss is at a premium (e.g. vacuum distillation columns). The result can be a higher than expected bottoms temperature, or a less favorable relative volatility, or a higher energy cost for moving the gas or vapor through the column (especially in the case of absorbers). If, as a corrective action, a shift is made to a lower column pressure, entrainment and flooding problems may develop. Excessive pressure drop can result from the following: • • • • •

Flow constrictions/plugging Bad predictive model Faulty internals Changed flow rates Entrainment

Look for flow constrictions on trays, distributors, and supports - Some of the flow constriction problem areas mentioned above apply equally to the pressure drop problem. Constricted flow can mean higher pressure loss. The constrictions can be due to foreign materials plugging, improper fit-up or installation of internals, accumulated debris, and so on. A gamma ray scan can be helpful in locating zones of constriction. Crushed or deformed packing elements can plug portions of the support plates with a resulting increase in pressure drop. How reliable are the pressure drop correlations being used - As in the case of capacity, the correlations used for predicting pressure drop should be examined carefully. For tray columns, it is necessary to select a discharge coefficient for obtaining dry pressure drop; the coefficient may not properly account for a wetted orifice. The variable area correction for valve trays is another point of concern. For sieve trays, bubble-cap trays or valve trays, estimation of the pressure loss through the aerated zone is always subject to error. And if the column design model does not have the ability to correct for flow conditions and compute pressure drop locally, there may be unexpectedly high pressure drop.

Prediction of pressure drop in packed columns is fairly reliable if operation is below the load point. From the load point on to the flood point, the pressure drop models are relatively unreliable because of the difficulty of predicting holdup as a function of dynamic conditions. Check the pressure drop of peripherals and other internals - Another possible problem can be the pressure drop caused by the internals in packed towers. Support plates are notorious for this particularly when they have open areas less than 100% and when the packed beds use small packings. Another possible problem is high gas velocities (pressure drops) in risers in collectors and distributors. High pressure drops in these devices can also lead to overflowing and poor liquid distribution. Finally, the capacity limits and pressure drop of bed limiters are often neglected when designing a tower and in many cases limit the performance. Check the meters and the material balance - Sometimes an excessive pressure drop results from flows that are higher than thought, due to faulty meter readings. It may not be recognized that an increased reflux ratio results in increased boil up of vapor. There may have been a shift of the column material

Page 10 of 17



3.01.1

balance.

8

Column Instability

Column instability can manifest itself by unsteady flows, temperatures, pressures, compositions, or a combination of these. Stability problems are often the most difficult to diagnose since the information available from the operation of the column will be “contaminated” by control and operator actions taken to mitigate instability. Furthermore, instability problems are by their nature unpredictable and difficult to duplicate. These are some of the things to look at when looking for causes of distillation column instability: Review general control problems - Stability problems can often be caused by improper control schemes, or tuning parameters. A thorough review of the control philosophy and schemes is the first order of business when addressing stability problems in distillation. Frequently, instabilities manifest themselves after hardware changes that produce different fluid inventories and, consequently, column response characteristics. Look for pressure control problems - Lack of consistent operating pressure in a distillation column is the single most important cause of instability. This issue can be very significant when columns ride on header pressure or when the condensing capacity is limited. On the other hand, some systems that use very large air-cooled condensers can be subject to extreme pressure fluctuations caused by weather. Check for water presence in hydrocarbon systems - The sudden appearance of water in a hydrocarbon distillation can have important consequences to stability, especially in systems that exhibit very limited water miscibility. This effect can be catastrophic in cases where the column operates at low pressures and temperatures in excess of 300 ° F. Check for foaming - The appearance of a foam in a distillation column will manifest itself either as a premature capacity limit or, in some less severe cases, as an instability that tends to make the control of the overhead product practically impossible. Foaming, and sometimes entrainment, can usually be suspected when the temperature profiles do not justify the presence of heavy material in the overhead but these materials find their way in large quantities to the top product. Look for feed condition stability - The heat content of the feed can vary slightly but cause tremendous variations in the volumetric ratio of gas to liquid. These changes can cause stability problems even when the swings are minor in cases where the feed is flashing. Columns that take feeds that undergo significant flashing across the feed control valve are very susceptible to instabilities caused by irregular flashing conditions. Look for reboiler surging and stability. Look for proper condenser venting or pressure equalization - Condenser and reboiler stability can have a major impact on the stability of the column itself. The following section addresses some common exchanger problems.

9

Interaction With Problematic Peripherals Check for malfunctioning reboilers, external exchangers, and/or condensers Vapor disengagement - If head room is not provided in a kettle reboiler for disengaging vapor from the liquid, excessive entrainment can choke the vapor return line, causing hydraulic instability and the pressure (and liquid boiling point) in the reboiler to rise. Because the kettle

Page 11 of 17



3.01.1

reboiler's disengagement space is so much smaller than that in the column, this reboiler is not recommended for boiling foamy liquids. Residue accumulation and fouling - The usual overflow baffle in a kettle reboiler can trap solids in the shell, fouling tube surfaces. A nozzle for a slip stream can prevent such accumulation. Other kinds of fouling or plugging can affect reboiler stability by either lowering the process side flow rates or by diminishing heat transfer rates. A common side effect of fouling is an increase in skin temperature that can cause product degradation problems. Level control and distribution - Constant liquid level on the reboiler inlet is necessary for reboiler stability, especially in the case of thermosyphon reboilers. Varying levels will cause reboiler surge and uneven vapor flow into a column with the ensuing instability. A low liquid level accelerates vaporization-induced fouling of the uncovered tubes in kettles. An overflow baffle that sets the liquid level over the tube bundle will prevent this. A high liquid level floods the vapor disengagement space, causing unstable hydraulics and excessive backpressure. As in horizontal reboilers, poor inlet liquid distribution can result in liquid stagnation and high-boilingliquid blanketing of heat-transfer surface area. Improper suppression of vaporization - Although vaporization is often suppressed by an outlet valve or orifice, a large drop in flowrate (hence, pressure drop) may promote vaporization, which can cause hydraulic problems or vapor bind the heat-transfer area, especially in multiple tube pass reboilers. If vaporization is suppressed, a reduction in flowrate will also boost the process temperature rise, sometimes pinching the temperature driving force for heat transfer. Improper reboiler circulation - Thermosyphon reboilers are especially susceptible to diminished circulation caused by hydraulic restrictions. This normally produces and excessive vaporization fraction in the reboiler which can lead to temperature pinches and two phase flow limitations Condenser fouling - Condensers are less likely to be fouled than reboilers because the mass transfer in the column keeps solids out of the overhead stream. The washing action of the process condensate usually keeps the condenser tubes clean. Corrosion products or material sublimation can cause fouling. Process condensate can freeze if the coolant temperature is too low, such as when a process upset allows high-boiling components to be driven overhead. Such a freeze-up can plug the condensate drain and flood the condenser, coat the tubes and reduce heat transfer, or restrict vapor flow. Coolant-side fouling is a common problem. Low water velocity, especially on the shellside, can allow silt to accumulate, and a high water-outlet temperature abets mineral deposition on heat-transfer surfaces. Additionally, algae or fungus and other organisms that thrive in warm water can foul or plug the exchanger. Regular backwashing can minimize some of these problems. Of course, the extent of fouling can be estimated from the change in the heat transfer coefficient. Low coolant flow - As mentioned, a rise in the temperature of the outlet water of a water cooled exchanger may indicate reduced coolant flow. A high coolant inlet temperature will, of course, limit condenser capacity. The temperature can vary seasonally, an even with the time of day, particularly with air-cooled condensers. A poorly located air cooler can recirculate its air curtailing its capacity, or exhaust its air into another air cooler, limiting the latter's capacity. Inert blanketing - Inert gases - such as may flow in through instrument purge lines, and through pipe fittings if the column operates under vacuum, or be present in the column feed -can reduce the condenser's heat-transfer coefficient. The problem is more severe near the exchanger outlet,

Page 12 of 17



3.01.1

where a large concentration of inert gas can create a mass-transfer resistance to vapor condensation and may even block off the flow of vapors, if the inert gas is not vented. Condensers that are operated flooded for pressure control are prone to accumulating inert gas, if the gas is not shunted via a bypass. Vent-system over-loading - If a condenser becomes overloaded, a backpressure may build up from the vent line or vent condenser, which are usually sized for small loads. A hot vapor flow from an overloaded or poorly performing condenser can be detected by measuring the vented gas temperature. Condensate flooding - As can steam condensate in reboilers, process condensate can back up into a condenser and cover the heat-transfer surface. Poor baffle orientation in a horizontal condenser can similarly flood the compartments. Excessive entrainment from the column can overwhelm the condenser's process drain, which should be sized for gravity flow to avoid liquid backup. Exchanger leakage - Water or steam leakage into condensers or reboilers can have a deleterious effect on the capacity or efficiency of distillation columns, especially in low pressure hydrocarbon distillations. The introduction of unwanted water into high temperature towers can even cause severe mechanical damage. Small leaks can often cause column instability as well. Check for malfunctioning overhead accumulators/decanters - Many of the instances of a malfunctioning overhead accumulator can be traced back to the condenser as described above. Most commonly the problems are associated with improper venting and pressure surging. On the other hand, a particular problem with reflux accumulators that also act as decanters is the possible introduction of a second free liquid phase with the reflux. This can causes significant upsets in the column ranging from lower overhead temperatures to severe flooding caused by severe flashing in the presence of a second liquid phase. Check for malfunctioning bottoms pump - A cavitating or undersized bottoms pump can affect column operation significantly by improper bottoms level management. In extreme cases, when the liquid level exceeds the level of reboiler return nozzle, severe column damage can occur.

Page 13 of 17



3.01.1

TROUBLESHOOTING CHECKLIST

The following is a listing of some of the more common causes of problems with distillation, absorption, and stripping columns. This list can be used by the troubleshooting engineer as a checklist that will apply to a particular problem. This checklist can help the engineer cover all the possible causes of performance problems. PROCESS DESIGN □ □ □ □ □

Accuracy of VLE Accuracy of physical properties Proper # of theoretical stages (above minimum) Proper reflux ratio (above minimum) Proper feed location

SIMULATION - (Model match to operation under normal conditions) □ □ □ □ □

Match material balance Match energy balance Match temperature profile Match composition profile Reasonable theoretical/real stages ratio (Efficiency)

SIMULATION - (Model match to operation under upset conditions [if steady]) □ □ □ □

Match material and energy balance Match temperature profiles Actual theoretical/real stages ratio Overall reliability of simulation

PROCESS AND OPERATIONAL ISSUES □ □ □ □ □ □ □ □ □ □ □ □ □ □ □

Has the pressure changed? Has the feed composition changed? Has the feed flow changed? Have the product flows changed? Have the utility conditions changed? Has the feed temperature changed? Are all levels normal in the tower? Is the system foaming? Is the presence of non-condensables accounted for? Can a second liquid phase appear? Undesired reactions within mixture or caused by material of construction Water or steam leakage into tower Appearance of an impurity Can an impurity develop a concentration bulge in tower? Can recycles lead to buildup of undesirables?

Page 14 of 17



3.01.1

EQUIPMENT SIZING □ □ □ □

Reliability of capacity estimation model Reliability of pressure drop estimation model Reliability of efficiency estimation model Is foaming factor selected properly?

MASS TRANSFER DEVICE SELECTION □ □

Is packing size and capacity appropriate? Is tray type and geometry appropriate?

TOWER LAYOUT □ □ □ □ □ □

Is entrainment disengaging space above and below mass transfer zones appropriate? Is vapor distribution height below mass transfer zone appropriate? Are reboiler returns nozzles properly sized and located? Are flashing feed nozzles properly sized and located? Are all outlet nozzles properly sized and located? Are gravity flows and siphoning effects accounted for?

SCAN RESULTS Tray active area Flooding □ □ Heavy entrainment □ □ Uneven distribution (multipass) □ Downcomers Back-up □ Packing Uneven liquid distribution □ Foaming □

Foaming Loss of seal

Choking

□

□ □

Blowing Weeping Loss of trays/panels

□

Loss of seal

□

□

Packing dislodged/damaged

Fouling

Uneven levels Damage

Flashing feed distributors Overflowing □

□

Liquid carryover

Vapor space below trays or bed Liquid carry-over □

□

Evidence of collapsed hardware

Mist eliminator Flooding □

□

Liquid distributors Overflowing □ Foaming □

□

□ □

Dislodged/damaged

Page 15 of 17

□

□

Heavy entrainment

Damage



3.01.1

INTERNALS DESIGN/OPERATION

Is liquid distribution quality proper? Is vapor distribution quality proper? Is flashing feed dealt with properly? Is entrainment accounted for? Is weeping accounted for? Is vapor bypassing possible? Is leakage accounted for?

□ □ □ □ □ □ □

Design Design Design Design Design Design Design

□ □ □ □ □ □ □

Are plugging/fouling/corrosion possible?

________________

Could plugs have been removed by decontamination?

________________

Operating Operating Operating Operating Operating Operating Operating

INTERNALS INSTALLATION (or condition after visual inspection)

Are clearances and tolerances satisfactory? □

Downcomer clearance

_______________ □

Tray/distributor deformation

Evidence of change in flow area by fouling or corrosion □ □

Tray holes Distributor orifices

Levelness and plumbness Tray levelness □ Overflow weir levelness □ Other mechanical damage Dislodged trays or tray panels □ □ □ □

Construction or process debris Fastener reliability Leakage possibility

INSTRUMENTATION AND CONTROL ISSUES □ □ □

□ □ □ □ □ □

□

Valve openings Missing valves or caps

□

Distributor levelness

□

Displaced distributors and collectors

□

Adequacy of control scheme Adequacy of pressure control Poor/improper tuning Resonance Inadequate response time Over-constrained control Inadequate model based control Faulty signals from process Improper meter location

Page 16 of 17



3.01.1

PERIPHERALS Reboiler stability Column and reboiler levels adequate □ Reboiler circulation rate appropriate □

□

Condenser Pressure drop normal □ Condenser properly drained? □

□

□ □ □ □

□

□

Head loss in reboiler piping adequate Proper heat medium flow system

Is condenser properly vented? Is condenser affected by weather?

Ability of pumps to remove streams Operation of vacuum system Are gravity lines leaving tower self venting? Operation of side-strippers normal and stable?

Operation of equipment upstream and downstream How can it affect tower ________________________________________________________

Are upstream/downstream operations normal ______________________________________

Page 17 of 17

FRI VOLUME 5: FRACTIONATION DESIGN HANDBOOK

DESIGNING FRACTIONATION SYSTEMS TO AID TROUBLESHOOTING

Issued:

01/15/1998

3.01.2

Revised:


DESIGNING FRACTIONATION SYSTEMS TO AID TROUBLESHOOTING ................... 1 1.

Flow Measurement ............................................. ................................................... ............. 2

2.

Sampling ............................................................................................................................. 2

3.

Temperature Measurement .................................................. ............................................... 3

4.

Pressure and Pressure Drop Measurement ............................................. ............................ 4

5.

Level Measurements ...........................................................................................................5

6.

Issues .................................................................................................................................. 5

7.

Design for Ease of Inspection/Replacement/Removal .......................................... ............. 6

8.

Miscellaneous ................................................. .................................................................... 7

Page 1 of 8


1


3.01.2

Flow Measurement

Accurate process and utilities flow measurements are essential if an operating material and energy balance is to be calculated about the fractionation system to compare with the expected performance. The following recommendations are made concerning flow measurements for the purposes of troubleshooting: A. Sufficient flow measurement should be available at column inlets and outlets to permit calculating a material balance and energy balance about the column. The more important of these streams should be indicated in the control room. Those that are likely to fluctuate should be recorded. If a density is necessary to properly calibrate the flow metering device, as is the case for an orifice plate with differential pressure measurement, a sample connection and thermowell should be specified for that stream also. B. The hardware for flow measurement should be properly constructed, installed, and calibrated. For example, if orifice plates are used, there must be a sufficient straight pipe run upstream of the orifice to allow time to establish uniform velocity profiles. Orifices should not be used for liquids at their bubble point as flashing across the orifice will distort the flow measurement. C. When orifices are used they should be tagged to indicate the proper direction for installation and the size of the orifice. D. Flow recorders and indicators should show the relationship between % of scale and flow. E. Provision should be made for obtaining the flow rates of utilities. Steam flow should be measured and, when important, indicated in the control room. When fluctuations are likely, steam flows should be recorded in the control room. For troubleshooting purposes, coolant flow can be measured with a portable ultrasonic flowmeter. For accurate flow measurements, the piping should be running full of liquid and the ultrasonic flowmeter clamped onto the piping at least 30 pipe diameters downstream of any flow disturbance such as a valve or bend. F. Provision should be made for obtaining the flow of streams important to the operation of the column but not normally metered, e.g. the vapor vent flow from a partial condenser. If not already available, the addition of appropriate nozzles(s) for radio-isotope or gas tracer flow metering techniques would be very helpful for future troubleshooting.

2

Sampling

Process samples are also essential for calculating an operating material and energy balance about the fractionation system. Sampling and analysis is often the only way to identify the presence of components unaccounted for in the original design. Samples from the column help establish the separation profile and identify poor contacting device efficiencies. The following recommendations are made concerning sampling for troubleshooting: A. Sampling arrangements should be provided on all feed, product, and reflux lines where the stream is all liquid or all vapor. Nozzles for samples from the column should be provided at the top and bottom of each energy and material balance zone, according to reference 11. These zones are defined by points of energy or mass feeds and outlets from the column. B. Sampling valves should be installed and located so an operator can reach them from a platform, grade, or as a minimum from a ladder.

Page 2 of 8

Issued: 01/15/1998

DESIGNING FRACTIONATION SYSTEMS TO AID

Revised:

TROUBLESHOOTING

3.01.2

C. Procedures should be established to ensure representative samples are obtained. Vapor samples should be free of liquid entrainment. Sample lines should be purged to remove a volume of material in excess of the line before sampling. Liquid samples that could flash when discharging into the sample container should be cooled below their flash point before sampling, or sampled under pressure to avoid the flash. Likewise, if the material could react to change composition, the sample should be taken under conditions to stop the reaction. D. For tray columns, nozzles for liquid samples should be located in the trays' downcomers near the bottom, on the centerline, to obtain a rough average of the composition leaving a tray. E. For packed columns, use the FRI liquid sampler. F. The size of piping, which is inserted in columns, associated with sampling devices should be kept to a minimum. A 1/4" diameter is usually adequate. Larger sizes require longer times for sample line flushing and may interfere with flow within the column. G. Since it is difficult to obtain a representative vapor sample, liquid samples should be taken if they will provide the desired information. H. The AIChE Equipment Testing Procedures committee has prepared a comprehensive guide for sampling procedures. (17))

3

Temperature Measurement

Process as well as utilities temperature measurements are also essential for calculating an energy and material balance about the fractionation system. Temperature measurements can also help locate damaged or missing internals by indicating separation problems and occasionally identify poor liquid and/or vapor distribution. The following recommendations are made concerning temperature measurements for the purposes of troubleshooting: A. The temperature of all major feeds and outlets from columns should be measured. For liquid/vapor feeds, the temperature should be measured prior to the point of flashing, or at a point of known pressure. B. Nozzles for temperature measurement should be located sufficiently close to allow detection of problems. As with sample nozzles, reference 11 recommends locating temperature nozzles at the top and bottom of each energy and material balance zone. If a rapid temperature change is expected to occur over a few ideal stages, additional nozzles for temperature measurement would be useful for pinpointing the change. C. For vacuum columns, temperature nozzles should be located with pressure nozzles to maximize the amount of information provided by the measurements. Compositions can frequently be inferred from pressure and temperature, but only knowing pressure or temperature introduces another unknown. D. If the objective of the measurement is to obtain the equilibrium stage temperature, nozzles for temperature measurement should be located out of the path of subcooled or superheated feeds (avoid placing temperature nozzles on trays with external feeds if possible) and liquid descending from the tray above.

Page 3 of 8



3.01.2

E. As the compositions change across a tray, the temperature will also. Therefore, it is difficult to obtain an average tray temperature. An approximate average of the temperature of the liquid leaving a particular tray can be obtained by locating the nozzle for temperature measurement on the centerline of the tray's downcomer near the bottom. The thermowell length should be short enough to avoid striking the downcomer. F. Provision should be made for measuring the temperature of the streams entering and leaving feed preheaters, coolers, and condensers. Steam temperatures in reboilers can be inferred from a measurement of condensing pressure, however, if superheated steam is used, a temperature measurement of the steam should be provided. Coolant header temperatures can provide the inlet temperature of coolant to condensers. Monitoring outlet coolant can help identify fouling problems. Contact pyrometers can be used to measure temperatures, but have limited accuracy for relatively hot, uninsulated surfaces. G. Provisions should be made to ensure that the desired vapor or liquid temperature is measured accurately inside a column. Special care should be taken in the design of the thermowell insertion to ensure vapor temperature measurements when required. Hardware to prevent entrainment on contact with falling liquid should be installed to shield the measuring element. Conversely, assurances of liquid immersion are necessary when liquid temperatures are desired. This can be problematic in particular in packed towers when temperatures within the packing are to be determined

4

Pressure and Pressure Drop Measurement

Pressure drop measurements are useful for evaluating column capacity, fouling, foaming, and locating damaged, missing, or improperly installed internals. The following recommendations are made concerning pressure measurements for the purposes of troubleshooting: A. Pressure nozzles should be included in the column system where knowledge of the pressure is critical to understanding the process. For example, top and bottom column pressures are important in most services, but particularly in vacuum services. Liquid spray nozzles' header pressures control the spray discharge pattern which influences packed bed efficiency. Steam condensing pressure should be measured for reboilers. Pressure drop across feed filters or strainers allows the troubleshooter to monitor solids accumulation. B. Pressure nozzles and transmitters should be located to avoid problems, and ensure good quality data for the troubleshooter. Pressure nozzles should not be within the projected path of feed streams. Two-phase or vapor feeds can distort measurements and damage instrumentation. To avoid liquid entrainment into the piping to the transmitter, nozzles should be located well above the expected spray heights, and/or shielded. Entrainment of liquid into the pressure nozzles can be minimized by locating the nozzles just beneath a tray and over the downcomer of the tray below. C. Nozzles for pressure measurement should be located close enough together to ensure that indicated pressure drops will be sufficient to detect problems. For example, if the pressure drop measurement is across an entire 50 tray column, flooding may go unnoticed, and the exact location of any flooded zone will be very difficult to determine. Flooding within a column can only be pinpointed to as few trays for which a pressure drop measurement is available. Nozzles for pressure measurements should be located wherever the column changes diameter, a different type of contacting device is used, the vapor and/or liquid loads are changing (i.e., locations of material or energy addition or removal), or the tray design significantly changes. In addition to the above, pressure nozzles should be provided above and below each bed in packed columns, and a

Page 4 of 8



3.01.2

minimum of every 10 to 20 trays in tray columns. D. Provisions need to be made to ensure that pressure and pressure drop taps remain clean and unobstructed, especially in fouling services. Some of these provisions might include inert gas purging, proper orientation and drainage of nozzles and impulse lines, and the use of sealed instruments with diaphragms for isolation from the process fluids

5

Level Measurements

Level measurements provide information to detect high base level, explain poor reboiler operation, and check for adequate liquid head at drawoff points. The following recommendations are made concerning level measurements for the purposes of troubleshooting: A. Most level measurement devices are dependent on knowing the density of the material. If the density of the material changes, then the level measurement can be misleading. If the density is lowered, the actual level can be higher than the indicated level. When the liquid level has risen above the upper level nozzle, the indicated level will stay constant. Therefore, if the upper level nozzle is located at the maximum desired liquid level, the troubleshooter can determine when the level has risen too high, even if the density of the material has changed as a result of column upset. B. Level glasses, or some other alternate means of level measurement, should be used in conjunction with level transmitters if: the density of the fluid is unknown, the level transmitter is considered unreliable (as for fouling services), or if operating upsets could significantly affect the density of the fluid. Level glasses used in conjunction with transmitters are also very valuable if the level measurement is essential and/or highly sensitive to the calculated fluid density (as for interface control in decanters). C. High level alarms, operating on a different mechanism than the level measurement technique, should be installed at the 100% level point for critical levels (as in the base of columns). The alarm can be used to calibrate the level transmitter and provide a warning of high level to avoid equipment damage. D. Additional tips regarding level measurement are available in references 18 to 20.

6

Issues

Proper instrumentation is crucial in identifying and solving problems. Nozzles which have only been included for troubleshooting can be blanked, provided with valves, provided with partial instrumentation, or provided with full instrumentation. The ultimate decision is dependent on balancing economics with the risk of needing and not having access to the proper instrumentation. The following circumstances favor the use of more instrumentation at nozzle locations: Large plants Operation at capacity Low cost materials of construction Lack of operating experience Columns with a history of trouble If the columns cannot be easily shutdown to remove blanks and outfit nozzles with the necessary valves, nipples, thermowells, etc. for troubleshooting, it is advisable to provide these arrangements from the start. Page 5 of 8



3.01.2

Valves can be added to an operating column by hot-tapping, but this is a fairly dangerous and relatively expensive procedure. The following shortcuts can be taken to obtain data from nozzle locations without providing full instrumentation: A. Portable or field-mounted recorders or indicators can be used. B. With pressure measurements, different nozzles can be piped to a single transmitter with valves installed to isolate the different nozzle locations. C. With temperature measurements, multiple signals can be sent to a common indicator or recorder. D. A troubleshooting station can be established in the control room where several spare recorders or data loggers are available. The spare recorders can be wired to the transmitters of whichever column might be experiencing difficulties. All shortcut techniques do, however, limit the amount of data that can be taken. If it is necessary to have simultaneous pressure drop readings, multiple pressure transmitters will be required. If frequent problems are expected, such as continually operating a column very near its flood point, recorders should be installed in the control room to monitor key variables. The dynamics of column operation are usually impossible to monitor unless a continuous recorder is used. Economics dictates that the information provided by instrumentation be maximized. This can be accomplished, in part, by proper location and combination of instruments. Frequently, a single nozzle can be used for absolute pressure, differential pressure, and level measurement. By placing pressure, temperature, and sample nozzles at the same location, more information can be obtained than if the nozzles were at three different locations. It is important that instrumentation be accessible, working, and accurate. Transmitters should have vibration free mounting and a watertight housing. Piping to transmitters should be free-draining and traced, or insulated where necessary to prevent plugging or formation of a meniscus. Purging transmitter piping with inerts will also help keep the lines clear. Chart paper should have the proper scale.

7

Design for Ease of Inspection/Replacement/Removal

Provision should be made for access to the column and auxiliary equipment for inspection and maintenance. The following recommendations are made in regard to accessing the fractionation system: A. For tray columns, manholes should be spaced to ensure all sections of the column can be adequately and safely accessed and serviced. Spacing manholes every 20 to 30 trays is typical. Larger spacing is used with clean, non-corrosive services, while smaller spacing is often justified when a considerable amount of maintenance work inside the column is anticipated during short shutdowns. Having several manholes can allow different crews to work in the column together and reduce the number of tray manways that have to be removed to reach the location of the damage. B. Quick opening manways are often justified if frequent tray inspections are anticipated. C. All packed tower internals (particularly distributors and redistributors) should be quickly accessible for inspection, given their importance in the operation of the column. Therefore, manholes for large columns and body flanges for small columns should be located above and below all packed beds.

Page 6 of 8

Issued: 01/15/1998


Revised:

TROUBLESHOOTING

3.01.2

D. Heat exchanger surfaces, which are expected to foul, should be accessible for cleaning. This includes condenser and reboiler tubes and shells, as well as air-cooled condenser's exterior finned surfaces. If an exchanger has a removable bundle, room should be provided around the exchanger for removing the bundle without interference from other equipment. E. Feed points, diameter changes and contacting device changes should have provision to be easily inspected since the chances for errors are greater at these locations. F. Internals ladders should be provided, when necessary, to allow a complete inspection of all theinternals in a column. G. For points in the column that are critical to its performance, such as flashing feed entrances and liquid distributors in packed columns, viewing ports may be justified. When used there should be two viewing ports installed 90 apart. One would be used for viewing while the other would be for a light source. ̊

Gamma scanning and video monitoring are two potentially very useful techniques for troubleshooting fractionation systems. Video monitoring usually require 1-1/2" to 2" nozzles for insertion and require that the columns be shutdown. Other instrument's nozzles can be used, but they should be above the tray deck to avoid any interference from the internals and provide an unobstructed view. Gamma scanning is employed while the column is operating. Since gamma scans provide a density profile of the material in the column, its uses include checking for liquid level, flooding, fouling, foaming, maldistribution, and incorrect installation or damage of internals. It is important that a reasonably clear view of the column is available for its entire length. A 360 platform at the very top of the column simplifies hoisting the source and detector. A good external datum point, both vertically and around the perimeter, simplifies the scanning process. Good, up-to-date drawings are essential. ̊

8

Miscellaneous Heat Exchangers - The following troubleshooting design features are recommended for heat exchangers:

The heating medium flow rate to the reboiler should be measured, as well as some indication of the inlet and exit temperatures. If steam is being used, the condensing pressure should be measured, since it is a good indicator of the available capacity of the reboiler, and can also serve to infer the condensing temperature. For steam, a direct measurement of the temperature is unnecessary if the condensing pressure is measured. If steam is superheated, however, the inlet steam temperature should be measured. A. For feed preheaters and coolers some provision should be made for measuring the streams flows, and inlet and outlet temperatures. See the discussion under flow measurement and temperature measurement for specific guidelines. B. The outlet coolant temperature should be measured for condensers. For a relatively constant duty, the outlet coolant temperature can give insight into the fouling rate of the condenser. C. Nozzles should be provided to vent inerts from exchangers. For thermosyphon type reboilers, an inerts vent nozzle located just below the lowest baffle can also be used to check the exchangerand steam trap's ability to remove condensate. Additional tips regarding inerts venting can be found in Page 7 of 8



3.01.2

reference 21. D. Insulation should be avoided on the bonnets of condensers or coolers where the coolant is on the tube side. For units with multiple tube passes, the troubleshooter can use a contact pyrometer (or hand) on the bonnet to determine whether the pass partitions were properly installed. E. Condenser vent temperatures should be monitored to warn of a loss of coolant or excessive vapor load from the column. Vacuum Systems - The following troubleshooting design features are recommended for vacuum systems:

A. When cost effective, an isolation valve should be installed in the suction line of vacuum pumps or jets. This valve will allow deadhead capacity checks on the vacuum equipment and air leak rate tests on the column. Connections should be installed on the suction piping, between the isolation valve and vacuum equipment, for a load metering manifold (commonly known as a piccolo) and a pressure indicator. B. A connection should be provided at the top of the column, perhaps a tee off the top pressure nozzle, for attaching a highly accurate pressure indicator for air leak rate tests. C. For multistage jets, connections should be installed between the stages to monitor pressure. D. Temperature indicator should be installed in the condensate outlet lines from direct contact condensers.

Page 8 of 8


DIAGNOSTICS USING RADIOISOTOPES

Issued:

12/16/2009

Revised:

3.01.3


DIAGNOSTICS USING RADIOISOTOPES ........................................................................... 1 1.

Introduction ................................................. ................................................... ................... 2

2.

Basics of Radioactive Techniques ............................................................. ........................ 2

3.

2.1

Radiation and Its Detection .................................................................................................. 2

2.2

Radiation Safety ................................................................................................................... 3

2.3

Guidelines for Getting an Accurate and Useful Gamma Scan ............................................. 4

2.4

Gamma Scans and Data Preparation .................................................................................... 5

2.5

Field Test Report ................................................................................................................ 10

Gamma Scans ................................................... ............................................................... 10 3.1

Planning for Gamma Scans ................................................................................................ 11

3.2

Interpretation of Trayed Tower Scans................................................................................ 13

3.3

Interpretation of Packed Tower Scans ............................................................................... 33

3.4

Stationary Monitoring Test ................................................................................................ 43

4.

Other Techniques Using Radiation .................................................................................... 44 4.1

CAT-Scans ......................................................................................................................... 44

4.2

Neutron Backscatter Methods ............................................................................................ 48

4.3

Radioactive Tracers ............................................................................................................ 49

4.3

Mobile Density Gauges ...................................................................................................... 51

Page 1 of 52


1


3.01.3

Introduction

When a distillation column problem develops that causes off-spec product or loss of production, the engineer has an urgent need to know what is happening inside the column. The sooner the problem is understood and corrected, the less loss it will cause. Techniques using radioisotopes enable us to "see" or detect hydraulic or mechanical status inside columns while they are operating. Since the 1960’s, when the first rudimentary radiation surveys were done on commercial-scale vessels, these techniques have become increasingly sophisticated and popular for process diagnostics (1). Modern radioisotope survey methods can be broadly classified into two main categories based on utilization methods of the radiation sources, i.e. the sealed source method and the unsealed source method. The sealed source techniques involve radiation sources that are housed in shielded containers, isolated from process fluids. The unsealed source methods involve radioactive sources that are injected into a process and detected at one or more points downstream. Obviously the unsealed sources are consumed in one test, while the sealed sources are reusable for subsequent tests. This document will focus primarily on the sealed source applications. The above categories can be further subdivided into sub-groups, •

Sealed source methods (commonly referred to as “gamma scans”) Gamma Transmission Gamma Scans • CAT Scans • Neutron Backscatter 



•

Unsealed source methods (commonly referred to as “radioactive tracer tests”) Pulse injection Continuous injection  

Gamma scans and radioactive tracer tests for towers are very much like medical X-rays and radioactive iodine or barium tests for human bodies. These tests help engineers to diagnose tower problems quickly and accurately. However, also like medical X-Rays, the interpretation of the test results can be subjective. Drawing proper conclusions requires both knowledge of the gamma scan technology itself and also of the internals and operating conditions of the column. It should be recognized that it is the engineer (like the medical doctor), not the test provider (like the X-ray technician) who will play the central role for selecting the proper tests, validating the data, analyzing the test results, and recommending corrective action whether through design and revamp or operating changes.

2

Basics of Radioactive Techniques

2.1

Radiation and Its Detection

Radiation sources used for gamma scans must have sufficient photon energy to penetrate through tower vessels. The common sources used are Cobalt-60 (1170 and 1330 KEV, 5 year half-life) and Cesium-137 (662 KEV, 30 year half-life). KEV is kilo (thousands) electron volts, a unit used to scale the photon energies. Half-life is the time it takes for a radiation source to lose half of its activity through decay. Obviously Cobalt-60 has higher energy so it can be used to scan larger diameter and thicker walled towers than Cs-137. For special applications, extremely large Page 2 of 52



3.01.3

diameter towers or very thick walls, Sodium-24 (1370 and 2750 KEV, 15 hour half-life) has been used as a scanning source. During gamma scans, a radiation detector is placed on the opposite side of the tower from the radioactive source along a scanline of interest to detect the gamma signals or number of photons passing through the tower (see Sec. 3.2.1 for more discussion about scanlines). The detector is normally a 2 ″×2″ (or 1″×1″) sodium-iodide crystal optically coupled to a photomultiplier (PM) tube. The 2" detector has greater sensitivity than a 1" detector, but the 1" detector’s compact size is more convenient for scanning a tower with congestion from external interference (e.g. numerous nozzles and manways, and tight platforms). However because of the higher sensitivity of the 2” detector, its use allows better energy calibration for higher resolution even while using smaller radiation sources. When gamma rays penetrate through towers, there are primarily two ways a gamma photons can be attenuated (many folks, including service providers, refer to this as radiation “absorption”) by material in the scan path: the Photoelectric Effect and Compton Effect. The photoelectric effect takes place when a gamma-ray collides with an orbital electron of an atom of the material through which it is passing. The gamma-ray passes all of its energy to the electron and ceases to exist. The gamma-ray is totally absorbed in the process, hence the use of the term radiation “absorption”. The Compton Effect occurs when a gamma-ray strikes a valance electron and transfers only part of its energy. The gamma-ray deflects off in a different direction to that which it approached the atom and this deflected or scattered gamma-ray can undergo further Compton effects within the material. This effect is also known as Compton scattering. Some of the scattered gamma-rays are able to penetrate through the vessel and also reach the detector. These scattered gamma-rays will have lower energy, or in other words, they have been attenuated (2). Ideally all the scattered low-energy signals should be rejected by calibration since the scattered gamma-rays do not have predictable correlation with the material density in the scan path. However if one were to rely only on the Photoelectric Effect then larger and higher-energy radiation sources and/or extremely sensitive (and fragile) detectors would have to be needed. In the real world, commercial scanning systems count a significant part of the scattered signals in order to scan columns with reasonable sized sources and existing detector technology. Very importantly using as small a gamma source as possible improves worker safety. There is a difference between service providers on how much “filtering” of scattered gamma-rays are done. Measuring a large amount of scattered radiation will reduce the “resolution” of gamma scans.

2.2

Radiation Safety

Gamma scans and radioactive tracer tests radiation safety and protection have been countries. The service crews are typically monitoring badges and survey meters, techniques.

have been proven to be very safe techniques, since one of the most rigorously regulated fields in most trained for radiation safety, equipped with radiation and experienced with radiation source handling

The service provider should be familiar with specific regulations, procedures or permits for the specific location (state, city and plant); be licensed or approved by the local regulatory agency for the handling and transportation of radiation sources; and have available on-site documentation of said licensing or approval, proof of worker’s safety training, and written emergency procedures. The service provider should have a standard Job Safety Analysis (JSA) sheet for the client to review and approve. A job safety review meeting might be needed on site before the job is scheduled or started, especially for radioactive tracer tests. The client should ask questions and be comfortable about all the safety concerns before allowing the tests to proceed. In the instance of Page 3 of 52



3.01.3

radioisotope tracer tests special attention should be paid to the tracer half-life, emission volume or concentration of the radioactive tracers, and potential contamination to products, process equipment, or environment. The service crews must perform proper surveys around the job area after the tracer job is finished to make sure there is not any unsafe radiation left behind. The service crews should set up proper barricade tape and radiation warning signs around the job area to prevent undue radiation exposure to plant personnel. The client should provide an on-site radio for the crew to communicate with the unit control room. When plant operators must enter the restricted area for a short period of time, to read a flow meter or to adjust a valve, the operators should first contact the scan crew so they can put the scan source in a safe place to allow the operators access inside the restricted area. Planning and precautions must be taken when gamma scanning will occur in the vicinity* of nuclear-based instrumentation, e.g. nuclear level gauges, and fireye flame control systems. Typically the installed instrument radiation source, being extremely well shielded, will not interfere greatly with the gamma scan data. However it is possible that the nuclear instrument detector system could respond to the scanning source giving a false reading to the control room. While limited measures can be taken to avoid or minimize this possibility it is highly recommended that these instruments be placed in manual-mode or standby during the scan. Plants should have procedures for dealing with this situation, as radiography (weld and pipe Xraying) would be even more risky. * In the vicinity could mean the plant instrument is installed on the process equipment being scanned, installed on neighboring pieces of equipment, direct line-of-sight from instrument to scanning source, etc.

2.3

Guidelines for Getting an Accurate and Useful Gamma Scan

Gamma scans, by design, are quick and after 30+ years of commercial application are also fairly routine. However to ensure that you get the most useful information possible from this service there are several things for you to provide the scan contractor. Prior to the scan the scan contractor will need to be issued the appropriate work permits for your site. The scan contractor should be supplied with a plant or unit radio so constant communication is available between unit operations and the scan crew in case of emergencies and in case operating conditions on the tower change substantially (see below). The scan contractor, to do a competent job, will need good drawings of the tower to be scanned. At a minimum the drawings should show the elevation placement of all major internals and show tray and/or distributor orientations. Detailed drawings showing wall thickness, placement and identification of nozzles, insulation or shell support rings, etc. should allow the scanning contractor to better analyze the scan results. Having good, up-to-date drawings cannot be overemphasized. The interpretation of the scan results, especially concerning the placement and physical condition of tower internals completely depends on using the drawings provided as a reference. In essence the scan results are matched up against the drawings to confirm that internals are in their proper places. One feature of a scan is that it is a “snapshot” of your tower hydraulics at the set of operating conditions during the time of the scan. First, it is important that your tower be operating as stably as possible. Separation towers, by nature, are a dynamic process but on a macro scale the scan should be done when feed/draw flows, column pressure, differential pressure, temperatures, etc. are relatively stable. If any of these parameters varies significantly this should be documented and reported to the scan crew. Secondly, the scan results will reveal the hydraulics for the Page 4 of 52



3.01.3

operating condition condition as they they are during the scan. If the problem of interest, high ΔP, offspecification product, etc. is not occurring at the time of the scan then the scan may not reveal any significant abnormality. abnormality. It is important that that the tower be operating operating at conditions conditions with the problem of interest occurring, or as close to that condition as possible while maintaining stable operation, so the hydraulic conditions seen from the scan will correlate to the problem of interest. Consider making a tower scan under stable operating conditions and an additional tower scan(s) that represents the onset of the deleterious deleterious issue. issue. For example, a scan can can be made of a tower operating stably at a lower feed rate and another at the onset of flooding at a higher feed rate. It is advisable that that repeat scans be done done with the same type of source isotope. isotope. Due to the different energy levels it is not a straightforward comparison between sets of scan data taken with dissimilar sources, e.g., one one scan with a Cesium source source and one with a Cobalt source. However scan sources on repeat scans do not have to be exactly exactly the same source. Scan data taken with two different activity levels but of the same isotope can usually be normalized one to the other. Communicate the symptoms of the problem to the scan contractor at the onset of the discussion about the project scope. Your concerns and what information information you wish to learn learn should set the agenda for how the scan(s) scan(s) are carried out. For example if your concern is tray damage damage and your tower has four-pass trays, do you want only one scan covering one of the four passes, or do you want to scan across across all four passes? Another example, a packed packed tower has an abnormally abnormally high ΔP, do you want a single scan line to look for flooding or a complete grid scan to look at liquid distribution? These issues should be discussed beforehand beforehand with the scan contractor contractor so you are comfortable with the options available. As a good practice in troubleshooting troubleshooting a tower, do not limit your investigation simply to the areas and probable causes where you suspect the problem exists. A gamma scan of one tower should be accomplished in a matter of a few hours, so there should be ample time to discuss the results results with the scanning scanning contractor. Once the scan data collection is complete, the scan contractor should should prepare a preliminary field report report while still at your your site. It is highly recommended that time be spent with the scanning contractor while they are at your site in case circumstances call for additional scans which may be started immediately or planned for a later time. Gamma Scans of a tower that is 6 inches or more in wall thickness are impractical and potentially unsafe. A wall thickness thickness of 3 to 6 inches requires a special radiation source, Sodium-24.

2.4

Gamma Scans and Data Preparation The detector measures the gamma intensity passing through the tower, expressed as a detector count rate, and the gamma counts for each elevation are stored in a computer while the source and the detector move down or up simultaneously along the outside of the tower. Normally the scan speed can be expected to be 20 to 30 feet per hour, depending on the required time for data acquisition and the number of external interferences. For some difficult situations where there are very tight platforms and much external interference in the scan path like piping or stiffening rings, the scan speed will have to be slowed significantly.

Page 5 of 52



3.01.3

The scan system should properly collect or record the following information during the scan: • • • • • • •

Start time Scan purpose (e.g. high rate or low rate) Timing of each elevation measurement Process changes (e.g. bottoms level surge) External interferences (e.g. manway, feed nozzle or platform, etc.) Un-expected interruptions or influences (e.g. windy weather) End time

These field records can be extremely vital for data validation validation and analysis. If some distorted or “fuzzy” scan data is caused by an external interference that was not identified properly, the troubleshooting troubleshooting engineers could be misled into a completely wrong diagnosis. Record the continuous operating conditions conditions (strip charts, DCS, trend lines screen shots) over the time that the scan is taken for additional documentation. documentation. The physical law underlying the gamma scan procedure is Beer’s law on radiation attenuation, - ln (I/Io)

∝ρ

In the above expression, Io and I are the intensity of the incident radiation and intensity through the material, respectively; ρ is density of the material that absorbs the radiation. It should be understood that there are many conditions that need to be fulfilled in order for Beer’s law to be valid. For instance, the absorbing medium (the material the radiation is passing through) must be homogeneously distributed and must not scatter the radiation. In the real world, the tower itself, its internals, and the process fluids are not a homogeneous density body and the gamma ray energy will be scattered significantly significantly by the metal internals internals and frothy liquids. liquids. Due to the “scattered” radiation and its detection Beer’s law does not hold valid for the calculation of density from typical gamma scan data (see earlier discussion discussion in Section 2.1). Nevertheless count rate has proven to be a convenient parameter to approximately correlate to the average material density ( ρ) between the gamma source and the detector. The correlation can be qualitatively expressed as,

ρ ∝

1 ln(count rate)

The formula indicates that the higher the count rate, the lower the average process density density ( ρ), where count rate is the gamma signal rate counted by the detection system. Some scan providers intend to use “density” as the scale of scan plots, by manipulating the gamma counts per many subjective assumptions. When interpreting the results, the engineer should recognize that the reported values of density are not necessarily meaningful. For gamma scans, this clearly shows why the count rate is usually plotted on a logarithmic scale against vertical position to show density variations within the column. A typical gamma scan plot for a single-pass tray tower is shown in Figure 1 on which the following information is indicated.

Page 6 of 52


DIAGNOSTICS USING RADIOISOTOPES • • •

• • • • • • • • •

3.01.3

The Scan started at 9:24 am and ended at 10:51 am. The purpose of the scan was to check the Active Area of the trays at design rate. The X-axis covers a range of gamma counts 100 – 30,000 on a logarithmic scale. The time span for the counts may vary among scanners, so the absolute counts are immaterial. It is important to allow the plots plots to be properly positioned positioned and expanded in the frame. The Y-axis starts from 0' (the top tangent line) and ends at 52' (bottom tangent line of column). There is a mist eliminator in the top of the column. There are total of 18 trays in the column. There is a reflux nozzle with a distributor above the top tray. There are insulation support rings above Tray #3 and between Trays #9 and #10. The measure of aerated liquid levels for each tray is shown. The height at which the bottoms liquid level was detected. The Clear Vapor Bar is located at ~23,000 counts The Liquid Line has about 1,100 counts

In addition it has two extra items included: the Clear Vapor Bar and the Liquid Line for the convenience of scan plot analysis. The Clear Vapor Bar is intended to represent the empty column (devoid of mechanical obstructions) filled with vapor at operating conditions. The lowest density (i.e. the highest detector count rate) measured through a specific column (set diameter and tower wall and insulation thicknesses) thicknesses) establishes the Clear Vapor Bar. It is generally shown on the gamma scan plot as a relatively narrow band that represents the observed minimum absorptions with a modest allowance for variation. Towers with multiple multiple inside diameters, wall thicknesses, thicknesses, or insulation thicknesses can have a different clear vapor bar for each distinct section, or these physical changes must be normalized in the scan data. The Liquid Line is intended to be a reference line that provides an estimate of the height of the froth on the tray deck. The point at which the Liquid Line intersects intersects the scan plot is that height estimate. The placement of the Liquid Liquid Line is specific to each supplier. The Liquid Line is estimated by a couple of different methods. methods. One method averages the highest counts (lowest density) and the typical minimum tray counts (highest tray liquid density) from the scan data. The second method, when a relative relative density plot is used, used, uses the median density as the froth height density. Some practitioners measure tray liquid liquid heights without regard for the effect of tray supports while others do. In any case the tray liquid heights should not be considered as an absolute measure of the tray froth heights – tray froth heights are anything but static, and should be used only in relative comparisons. comparisons. When trays are designed to hold a froth level, the trays likely have a physical or process related problem when a froth level is not detected. Furthermore, early studies studies and field field tests have shown that that when the froth height measurement nearly matches the tray spacing or the density of the vapor space between trays is close to or below the Liquid Line, the trays have severe entrainment and are generally close to flood. The measurement of liquid levels from gamma scanning, whether on chimney trays, fractionation trays, or in the base of the column, has a maximum accuracy or resolution of ± 1 inch (2.5 cm). The typical elevation scan increment is 2 inches (5 cm) with either a 1” or 2” tall detector (size of radiation sensitive crystal crystal in the detector). Tests have shown no improvement improvement in scan resolution resolution at elevation increments smaller than this.

Page 7 of 52



Figure 1 Typical Gamma Scan Plot

Page 8 of 52

3.01.3



3.01.3

An experienced scan crew should be able to identify many mechanical or hydraulic abnormalities of trayed and packed towers, as depicted in Figure 2. More details on interpretation of the scan plots for trayed and packed towers will be discussed in following sections.

Figure 2a Operation Abnormalities of Trayed Tower from Gamma Scans

Figure 2b Operation Abnormalities of Packed Tower from Gamma Scans

Page 9 of 52


2.5


3.01.3

Field Test Report A gamma scan of one tower should be accomplished in a matter of a few hours, so there should be ample time to discuss the results with the scanning contractor. Once the scan data collection is complete, the scan contractor should prepare a preliminary field report while still at your site. It is highly recommended that time be spent with the scanning contractor while they are at your site in case circumstances call for additional scans which may be started immediately or planned for a later time. The field report should contain the following minimum information: •

Tower Identifying Information

•

Specifics on Scan Line Orientation

•

Date and Beginning and Ending Times of the Scan(s)

•

Reason for Conducting the Scan

•

Measurement of Liquid Levels where Appropriate

•

Findings Specific to the Reason for Conducting the Scan

•

Other Major Findings from the Scan

Following the on-site work the scan contractor should issue a formal, written report documenting the job. The final report should cover all the items mentioned above for the preliminary report but go into sufficient detail to explain how the scan was conducted, any event (from the plant operation or from conducting the scan) that may potentially impact the scan results, and expanded details on the scan findings. The scan contractor should offer to include pertinent process data in the final report. If you do not wish to share this information, then you should have it available to file with the final report in your archives. In any case during the scan you should communicate to the scan crew on any change in tower operating conditions such as a change in operating pressure; feed, reflux, or product draw rates; reboiler or condenser duties, etc. Any substantial change in tower operation could have a huge affect on the scan data and lead to misinterpretation if the scan crew is left unaware of these changes.

3

Gamma Scans

Scan plots of trayed towers generally show a series of peaks corresponding to the tray decks and froth, and packed towers show high density bands corresponding to the packing beds, with peaks for distributors, collector trays, etc., separated by low density sections in between. Gamma scans can show a lot of information about a tower and its operation. The following is a list of phenomena that can be determined from gamma scans:

Page 10 of 52



3.01.3

•

Presence or absence of internals or significant damage to internals such as trays, mist eliminator pads, distributors, and packing;

•

Froth or spray height on active trays and the liquid level on collector trays and certain types of liquid distributors;

•

Extent and location of jet or downcomer flooding;

•

Location and density characteristics of foaming;

•

Occurrence of flooding or fouling in a packed bed; and

•

Existence of significant liquid maldistribution on multi-pass and dual-flow trays or in packed beds.

The technique is straightforward and rapid, and can be applied to practically any column to which external access is available. In practice, most columns have convenient platforms or walkways from which source and detector can be suspended; additional scaffolding is rarely required. Where platforms are not available or do not provide adequate access the scan can be performed from a man-basket hoisted by a crane. All sizes of columns can be scanned, from units less than one foot (300 mm) in diameter up to large crude vacuum columns of fifty feet (15 meters) diameter. Nevertheless, meaningful scan data will need thoughtful and strategic planning for the elevation, orientation, and timing of the scans, with proper consideration of the process operating conditions of the tower and the operating problem being studied.

3.1

Planning for Gamma Scans

Since gamma scans involve moving source/detector pairs up or down a column simultaneously, scan lines must be chosen that avoid as many platforms, external supports and nozzles/pipes in the vertical direction as possible. It may be necessary to choose desired scan lines and cut holes for removable plates in the platform decking around the tower. For trayed towers, scan lines that traverse an active tray area or downcomer do not have to orient exactly parallel to the downcomer panels, but the scan lines should not cut across adjacent active and downcomer areas, as this greatly and unnecessarily complicates interpretation of the results. Accurate drawings showing tray and downcomer orientations are essential in order to adequately plan for the scan line orientations. When scanning multi-pass trays, or examining bubbling-area and downcomers, multiple scans are needed in a short period of time while the tower is kept in a steady state. Additional planning for non-conventional (e.g. multi-downcomer type trays) needs to be employed because of the complexity of the particular equipment. In addition, the insight from result of a non-conventional tray type scan can be limited compared to more conventional trays. For example weeping, foaming or tray damage will be harder to detect. Some tests have been performed by simultaneously scanning all desired adjacent areas at the same time by use of multiple gamma sources or multiple detectors. However, the simultaneous scan results were difficult to interpret because of inconsistency and interactions between sources and detectors. Therefore simultaneous scans with multiple sources or detectors are not recommended. When it is difficult to hold the operation of a tower steady over the entire scanning period, another technique ma y be considered. The gamma source and the detector can be stationed at a fixed elevation to monitor the hydraulic dynamics inside the tower. This technique, referred to as Stationary Monitoring, will be discussed later.

Page 11 of 52



3.01.3

Gamma scans are recommended when the tower is in sound condition and operating normally in preparation for future analysis or troubleshooting. These tests, called baseline scans, are very helpful to document the effect of internal supports, external support rings and any other externals/internals in the scan path. It is desirable to have a dry scan or non-operating baseline scan on packed columns before the column is brought online for operations, i.e. the column is scanned while there is no vapor or liquid flows inside. The dry scans would not help very much for trayed columns, but for packed columns, the dry scans will be able to tag following critical information, • • • • •

Start and end elevations of packed beds and the relative density of the dry packing; Elevations and spans of liquid distributors, collectors and redistributors; Characteristics of gamma absorption peaks of the internals due to their metal content only; The effect of any internal or external interference on the scan data that could be due to the column itself or internal piping, etc. Afterwards, the operating scan data will reflect the change in density due to liquid traffic only or lack thereof.

Many internals in packed towers are proprietarily designed and installed, and their effects on gamma signals will need the baseline scans to be properly interpreted. One alternative is to dry scan only the complex columns that have many nozzles and custom or unusual internal hardware, and skip towers that have standard internals and few intermediate nozzles. Another circumstance when dry scans are valuable is when there is concern about bed flooding or fouling or a particular service has a history of being a problem tower. Packed crude vacuum towers are examples of towers that should have a dry scan done due to their large size, relative complexity of some of the liquid distributors, and the typical desire to monitor the beginning of coke buildup in the wash beds. Some towers may need multiple scans periodically for monitoring performance (e.g. plugging or fouling on trays or in packed beds) over their run-length. It is important for the scan crews to understand the purpose of those scans and make sure to repeat the scans on identical paths with an equivalent gamma source. A full-height scan (from tangent to tangent of the tower vessel) should always be performed when the process conditions allow. Very often the root cause of tower trouble happens or starts from other than where expected. For instance, a high liquid level in the bottoms could easily cause a tower to be heavily loaded with entrainment, possibly even flooding from the base liquid being blown upwards (entrained) by a reboiler return or vapor feed. The entrainment accumulates tray by tray and propagates up to a higher section of the tower. If the scans were stopped before reaching the bottoms level, the root cause could be missed or more scans would need to be done later for diagnosing the root cause. Foaming is another example where the location of the foaming problem can be affected by operating conditions and can move around in a column.

Page 12 of 52


3.2


3.01.3

Interpretation of Trayed Tower Scans

3.2.1

Scan Line Orientations A scan line is the gamma ray path between the radioactive source and detector. One or more scan lines are typically needed for investigating specific areas (e.g. bubbling area or downcomer area) of the trays in a tower. Each of the scan lines must be strategically positioned in the right location. Scan line orientation diagrams that are most commonly used for cross-flow trays. The scan lines for 4-pass trays are not shown in Figure 3, since the scan-line orientations would be very similar to those for 2-pass trays.

The diagonal scan lines for tray bubbling areas as show in Figure 3(a) and (d) are generally used for tower diagnosis, since the diagonal line is transmitting through an average froth density and froth height between the inlet and outlet of the bubbling area. Nevertheless when the liquid gradient across the bubbling area is a concern, the parallel scan lines as shown in Figure 3(b) and (e) should be applied. See the section on Liquid Gradient below for more details.

Figure 3 Scan-line Orientation Diagrams for Cross-Flow Trays

Page 13 of 52



3.01.3

If scanning is for the purpose of detecting a liquid gradient across tray bubbling areas, the scan line (b) should be as close to the inlet and outlet of the tray as one can get, but still be assured of staying out of the downcomer. For scanning parallel through side downcomers as shown in (c) the scan line does not necessarily need to get so close to the downcomer wall. However since the scan line is on the very edge of the column, the line should get as far into the column as possible while still keeping the line inside the downcomer. The scan lines for side downcomers as shown in Figure 3 (c) and (f) are typically very close to the tower wall edge, and the scattered gamma signals around the tower can be a significant part of the detector counts. Specially designed collimators (shielding devices) should be used to reduce the scatters.

3.2.2

Tray Damage Tray damage can manifest itself in several ways. If tray panels have been jarred loose, the tray may show a very rapid decline in density just above the tray deck rather than a more normal density decline slope, indicating that the tray is not holding liquid and/or not generating froth. Scans can also show one or more missing tray peaks, indicating that the trays in question have been dislocated. Often an unusual peak at the bottom of a blown-out section will signify tray parts lying on a tray that is supporting them. An example scan showing possible tray damage is given in Figure 4. There are a couple of limitations that should be understood by the client. The scan does not detect the tray deck itself since the usual tray deck is only a few millimeters thick, but instead detects liquid that is sitting on the tray deck, or where the tray deck is supposed to be. If the tray support ring is a "shelf" or ring welded onto the side of the column this usually has minimal affect on the tray "absorption". Heavier supports such as I-beams could have a big response on the scan data if the scan line passes through them. However in any case, especially for the I-beam or similar heavy support, this response should be below the elevation of the tray deck itself. Good practice dictates that any deflection due to supports should not be counted as part of the liquid level. But tray support interference will reduce the vapor space counts just below the tray thus making inexperienced scanners think there is entrainment. Detailed tower drawings with mechanical details of the internals and supports should be carefully checked to avoid misinterpretations. Therefore it is the absence of liquid where liquid should be that is the basis for diagnosing tray damage. Likewise the scan may not detect “partial” tray damage. “Partial” tray damage is defined when most of the tray deck is still in place but perhaps the tray manway is out-of-place, or a section of the tray deck is bent, corroded or cracked, commonly near the walls. Recall from earlier discussion that gamma scan counts are the average of everything in the tower across the scan path. If liquid is detected across the majority of the scan path then the tray froth height may appear normal. Usually “partial” tray damage is localized such that the scan path may not even pass through the damaged area, and liquid flow appears normal across the remainder of the tray deck. So the damage detected by a gamma scan has to be to such an extent that the damage prevents the tray (or liquid distributor or collector tray) from holding a suitable level of froth.

Page 14 of 52



Damaged Trays

Figure 4 Tray Damage as Shown on Gamma Scans Comparison of Damaged Trays (Trays 9 – 18) vs. Normal Trays (Trays 1 – 8)

Page 15 of 52

3.01.3


3.2.3


3.01.3

Entrainment The gamma scan profile for a tray that is entraining is typically characterized by a reduction in the radiation counts in the vapor space between the trays. As seen in Figure 5, the data curve count rate drops in a sloped gradient from the vapor space to the tray peak. This reflects the density gradient through the froth height on the tray – the heavier, larger liquid droplets stay closer to the tray but as one proceeds upward from the tray the liquid droplets are smaller and less dense. The extent to which the minimum observed density is offset from the clear vapor density (as referenced by the Clear Vapor Bar) is considered an indicator of how heavily the tray is entraining. As entrainment increases the vapor space becomes denser with heavier liquid droplets – the tray liquid height increases as the vapor lifts liquid higher off the tray. Once the data curve does not cross back over the tray Liquid Line the entire tray spacing is now filled with aerated liquid and the tray has severe entrainment or flooding.

Figure 5 Diagram showing degrees of entrainment as seen from a gamma scan

In Figure 6 a close look at the bottom trays, Trays 1-20, shows that the entrainment was gradually increasing from Tray 1 to Tray 20. In other words, liquid was being entrained upward tray-by-tray, and the overall liquid rate passing through each tray downcomer (liquid from the tray above plus the entrained liquid) was higher than the tray below. But the entrainment propagation stopped at Tray 20, where the tray spacing increased due to a manway. It was the larger spaces at the manways that saved this column from reaching a fully flooded state.

Page 16 of 52



Figure 6. Entrainment and its Accumulations on Trays

Page 17 of 52

3.01.3


3.2.4


3.01.3

Weeping

The gamma scan profile for trays that are weeping is as distinctly different from entrainment as is the hydraulic cause. Since by definition weeping is a hydraulic state where the tray liquid is not being highly aerated the scan profile for weeping is missing the sloping gradient through the tray froth height, refer to Figure 7. Typically the transition from liquid counts to vapor counts above the tray deck is quite sharp. Above this, the data response through the vapor space tends to be “flat” due to approximately constant density for the vapor space between the tray peaks. It is thought that the scan data also has a jagged appearance through the vapor space since weeping, especially for valve trays, tends to oscillate rapidly so “bands” of liquid droplets are falling through the tray spaces.

Figure 7. Diagram showing degrees of weeping as seen from a gamma scan

Weeping is a frequent problem with the towers operating at low vapor rate or turndown conditions. Figure 8 is a classic example of weeping trays in an Amine Contactor. Note the characteristics from above displayed on Figure 8: the jagged vapor spaces, the rapid decrease in count rate from vapor to tray liquid (i.e. without the sloped gradient typically seen for entrainment), signifying lack of liquid aeration. In this particular example the trays could be classified as severely weeping or dumping. Most trays, especially in the bottom of this Contactor, were holding the bare minimum of liquid. In Figure 8 the discrete data points have been enumerated. Note that on most trays there is only one data point registered within the tray liquid. The elevation increment between data points was 2 inches (5 cm). Less liquid than this would appear as a completely dry or missing tray!

Page 18 of 52



Figure 8 Weeping Trays

Page 19 of 52

3.01.3



3.01.3

3.2.5 Downco mer Backup Downcomer backup can be seen on a properly conducted downcomer scan. A normal downcomer scan will show maximum fluid density at the outlet (bottom) of the downcomer, with a profile of decreasing density proceeding toward the top of the downcomer. When the downcomer approaches flood, high fluid density (non-aerated liquid) will be observed throughout the height of the downcomer. Interestingly, the downcomer density profile may not change much over a wide range of process loads until the flood point is approached, then the material in the downcomer will have a sudden increase in density as liquid stacks up. Figure 9 shows the gamma scans for both active area (blue curve) and the center downcomer area (red curve) of a gas stripper tower with two-pass trays. The liquid level (downcomer backup) can be measured approximately from tray deck elevation line up to the elevation of lowest density (highest counts) between the trays. From the red scan curve on Figure 9, it can be seen that the downcomer backup was ≥ 65% of the tray spacing.

It should be clarified that when reading the downcomer backups, the "lowest density or highest counts between the trays" is not referring to the clear vapor bar. On downcomer scans the vapor space reading never makes it back to "clear vapor" because of the entrained droplets or jumping froth above the downcomer inlet. Reading the downcomer backups sometimes needs more art or experience than science.

Page 20 of 52



Figure 9. Downcomer Backup in Center Downcomers (Red Plot)

Page 21 of 52

3.01.3


3.2.6


3.01.3

Flooding Flooding is a dynamic process of liquid accumulation that develops from the incipient flood point into a fully flooded condition. The flooding process can be fast (in minutes) or slow (in hours), and some flooding symptoms can show up sooner or later than others, depending on column size, type of internals, flow rates and system properties. Incipient flood point (IFP) is the combination of vapor and liquid loads at which liquid downward traffic begins to choke or slow down. Incipient flood may initiate on a tray, in a section of a packed bed, at a feed or draw device, or at any other internal in a column. Most flood-point data in literature were collected from observations of IFP in laboratory test columns. Literature flood-point data typically shows good consistency because most of it was measured from small and well-controlled test columns.

The flooding process starts from the incipient flood point and ends with a fully flooded column, if the vapor/liquid loads in the column are not reduced. Once the flooding process begins, high liquid inventory inside the column will increase the pressure drop across the column, impede phase separation, and lower the separation efficiency. The column may become inoperable or uncontrollable due to excessive retention of liquid inside the column. During the flooding process, one or more trays might become “fully flooded” ( Figure 10), but the column may still be “functional” hydraulically. “Flood point” data of industrial columns tends to be observations of some point during the dynamic flooding process. Refer to section 3.3.4 ahead for flooding in a packed tower. A fully flooded state is the end of the flooding process where it is impossible to obtain net downward flow of liquid, and any liquid fed to the column is carried out with the overhead vapor. A significant part, or the whole volume of the column, is full of liquid and vapor bubbles through the liquid. The column completely loses its operability or function when fully flooded. Figure 11 is a comparison of the gamma scans for a trayed tower in a fully flooded state and a normal operating state. Traditional measurements of pressure drops, temperature profiles, or liquid levels usually cannot tell exactly where flooding originates in a column, particularly with a large number of trays or large height of packing bed. In these cases the incipient flood point will have come and been surpassed before the traditional measurements show any response.

Page 22 of 52



Figure 10. Gamma-Scans of a Partially Flooded Tower Flooding Initiates at the Feed Nozzle Causing the Top Section (Trays 28 – 40) Flooding

Page 23 of 52

3.01.3



Figure 11. Gamma Scans of a Fully Flooded Column (the Blue Curve) as Compared To a “Normal” Operating Column (the Red Curve)

Page 24 of 52

3.01.3


3.2.7


3.01.3

Tray Hydraulic Gradient

The hydraulic gradient or liquid head from inlet to outlet of a tray bubbling area is necessary to overcome the frictional resistance to the passage of the froth across the tray deck. However if the gradient is excessive, the upstream portion or inlet area of the tray bubbling area may be not aerated well, which could be problematic for trays with long liquid flow paths. With an excessive hydraulic gradient a tray deck might not operate in one uniform flow regime. For instance, the inlet area could be operating in a froth flow regime, and the outlet area in a spray regime, while the middle area of the tray could be in a mixed flow regime. These phenomena can get more complicated or troublesome at extreme conditions (Kister, 1992). The parallel scan lines shown in Figure 3 should be used for investigating the possibility or severity of liquid gradients. More details on evaluation of hydraulic gradient and flow regimes on trays by use of gamma scans have been discussed elsewhere. (Xu and Pless, 2001) Figure 12 shows an example of a scan where a liquid gradient was observable. In this case the scanline was positioned parallel near the edge of the active area in order to avoid some external physical interferences. Detecting the liquid gradient was actually accidental. The gamma scan shows that the even-numbered trays (scanline was on the inlet side of the even-numbered trays) above the feed had higher liquid loads and appeared to have a higher degree of entrainment as compared to the odd-numbered trays.

Page 25 of 52



Figure 12. Liquid Gradient across the Bubbling Area of a Tray as Seen From a Gamma Scan

Page 26 of 52

3.01.3


3.2.8


3.01.3

Foaming Foaming is characterized by a sloping, gradual density gradient through the tray liquid or forth height. From a gamma scan “foaming” would appear similarly to “normal” tray entrainment. A diagnosis of foaming depends a lot on the system itself. For example the scan evidence of foaming on an amine contactor would very likely be foaming. However scan evidence of “foaming” on a system not known to have a foaming tendency may be an example of a sustained froth. A sustained froth, where the bubbles of liquid do not readily collapse is more prevalent on scans of light hydrocarbon systems or vacuum systems operating in or near the spray regime. Sometimes the foaming cannot be clearly verified by the gamma scan plots only, and more analysis of fluid properties and/or process conditions is required. When the conditions are available, the best way is to compare the scan plots before and after an injection of anti-foaming chemicals as shown in Figure 13.

It is typically difficult to clearly distinguish between foaming and entrainment in a trayed tower. The scan is not meant to substitute for knowledge of the particular system. For example, a tower may have a predicted jet flood that is far away from current operation, yet the scan shows the tower is fully flooded. Foaming is typically indicated by a step change in the froth height density.

Page 27 of 52



Figure 13. Comparison of Gamma Scan Plots Before (Red) and After (Black) the

Injection of Anti-Foaming Chemicals

Page 28 of 52

3.01.3



3.01.3

3.2.9 Fouling Fouling on trays can occur in hours, days, months, or over several years of tower operation. Fouling plugs the bubbling areas or builds up in downcomers to restrict the throughput and performance of the tower. A baseline scan performed under clean conditions is always advisable, since gamma scans cannot detect or measure the fouling directly. Instead the scan diagnosis depends on the hydraulic abnormalities caused by the fouling. It is very difficult, if not impossible, to differentiate with a gamma scan alone if the abnormalities seen were caused by fouling or by tray design, if there was not a comparative base. An alternative way to get a comparative base is to carry out a gamma scan before and after a chemical or water wash to the suspected fouled trays ( Figure 14).

Page 29 of 52



3.01.3

Figure 14. Diagnosis of Fouled Trays by Comparison of the Gamma Scans Before (Red) and After (Blue) Chemical Wash to the Suspected Fouled Tower

Page 30 of 52


3.2.10


3.01.3

Spray Regime Discussion of the hydraulic gradient normally assumes that the vapor-liquid mixture on the tray deck is in the form of a bubbly liquid, or froth where the liquid phase is continuous. However under high vapor rates and low liquid rates, many studies have shown that the flow regime can invert to a spray, where vapor is the continuous phase and the liquid is dispersed into droplets. Trays operating in the spray regime, while not a common occurrence, when encountered give very distinct characteristics on a gamma scan plot. Figure 15 shows an example of trays operating in the spray regime as seen from a gamma scan. From a gamma scan the peaks of trays in the spray regime show a small layer of dense liquid phase on the tray deck with a lighter or spray phase above, i.e. there is a step change in the phase densities. Please note that in other previous examples the scan peaks typically have a smooth density gradient from tray deck to the froth top, which is typical for trays operating in the froth regime. Tray hydraulics and process properties should be carefully analyzed to validate the flow regimes on trays, since sometimes foaming on trays could generate scan peaks similar to that of the spray regime.

Page 31 of 52



Figure 15. Gamma Scan Plot of the Trays in Spray Regime

Page 32 of 52

3.01.3


3.3


3.01.3

Interpretation of Packed Tower Scans

Tower performance, as measured by separation efficiency or lack thereof, temperature readings, etc., may strongly suggest a packed tower is suffering from liquid maldistribution. A grid scan, as explained below, is a quick and handy screening tool that can be used by the engineer. If the grid scan shows a liquid distribution problem, then one definitely exists. However in some cases even when the grid scan shows what appears to be even liquid distribution there could be a liquid maldistribution pattern, e.g. annular flow. Examples are discussed in the following sections.

3.3.1

Scan Line Orientations

The key to good packing performance is uniform liquid and vapor distribution. Thus the primary information sought from scanning packed towers is whether the density through the packing is consistent by comparing several scan chords. It is normal practice to scan four chords of equal length, which are equidistant from the tower center in pairs that are perpendicular to each other as shown. This layout is commonly referred to as a grid scan. Other combinations of equal length scan lines are also acceptable as long as they cover a good portion of the tower cross section. Figure 16 shows typical orientations for grid scans demonstrating the change in the overall pattern as tower diameters get smaller.

Typical for diameters 5-6 feet (1.5-2 m) and larger.

Typical for diameters 1.5-5 ft (0.5-1.5 m).

Typical for diameters <1.5 ft (0.5m).

Figure 16. Typical Scanline Orientation Patterns for Packed Towers

Page 33 of 52



3.01.3

The multiple scan lines are usually rendered in different colored and/or patterned lines on the plots of the scan data. If the liquid load through the bed is nearly constant and the packing type (density) does not change, the scan lines should follow a ‘vertical’ path down the plot; i.e., they should show approximately constant density from the top of the bed to the bottom. If the liquid loading changes appreciably or the packed bed has two or more different densities of packing, the scan plot curves may be sloped (show a density gradient) or may show sudden step changes. The analyst should verify that the observed behavior makes sense. Regardless of any sloping or step change of the scan plot curves, the data curves should closely overlay each other through the bed, within the density fluctuations seen for individual scan lines. This would be an indicator of good liquid distribution since all the scan curves from the various sides of the tower all show approximately the same bulk density (the underlying presumption is that the packing density is uniform across the tower). When the scan curves separate from each other at the same elevation, this is evidence of differing bulk density or poor liquid distribution. The overall orientation of the “grid” should be positioned to match “squarely” with the liquid distributors, i.e. the grid lines should be oriented to cross either parallel or perpendicular through the troughs, parting boxes, or the gas risers of liquid distributors. In this arrangement the grid scan will provide the most useful information about the hydraulics on the liquid distributors, e.g. composite liquid levels in parting boxes or troughs.

3.3.2

Damage of Packed Towers

Packed beds in towers can be damaged by process surges, corrosion, or fouling. The damage can be seen on gamma scan plots as loss of bed height, dislocation of internals from their expected elevations, variance of bed bulk densities, or maldistribution of liquid. Figure 17 shows a grid scan of a packed tower. Drawings were consulted to ascertain where the internals should be located. The scan showed that the top bed and its reflux distributor were in their correct places. The grid scan also showed the top bed (looking past some external interference effects) had good liquid distribution. By comparison the bottom bed had lost about 3 feet of packing. The liquid distribution through the bottom bed was terrible with the Southwest side (black scan curve) appearing much denser (more liquid) than the other sides. The scans also revealed that the bottom bed support had apparently given way, with packing and liquid filling up the base of the column.

Page 34 of 52



Figure 17. Gamma Scans Showing a Normal Bed (Top Bed) and a Damaged Bed (Bottom Bed)

Page 35 of 52

3.01.3


3.3.3


3.01.3

Liquid Maldistribution

A grid scan of a packed tower is recommended as the first step to diagnose liquid maldistribution. As discussed previously the theory behind this procedure is that a uniform grid scan (comparable data profiles from all four scan lines) is an indication of uniform process density inside the packed tower. Any significant deviation among the scan lines could be an indication of a density maldistribution, i.e. denser readings indicate an excess of liquid while less dense r eadings indicate a liquid deficiency. Figure 18 shows the grid scan results from the same tower – one scan from the time the tower had operating problems and one scan post-turnaround after the problems were repaired. The left-side plot shows the scan results when the tower had operating problems. Actually several hydraulic problems or their appearance can be seen in Figure 18. Bed 1 shows a case of liquid maldistribution with the four scan lines varying from each other. The blue (east) scan line had the highest number of counts; this side of the tower was the least dense or had a deficiency of liquid. Conversely the green (north) scan line had the least counts, so this side of the tower was the densest or had an excess of liquid.

By comparison the right-side plot in Figure 18 shows grid scan results where it appears that the tower had good liquid distribution. However one limitation of a grid scan is that it is possible to have liquid maldistribution patterns that from a grid scan would appear to be uniform. Annular maldistribution is an example of liquid maldistribution that could appear as uniform liquid distribution from a grid scan. In situations where this is a possibility, a CAT-scan could resolve the problem (CAT-scans are discussed in Section 4.1). In addition, in lightly liquid loaded towers (e.g., less than about 1 gpm/ft 2 ), there may not be enough density difference between a dry tower and a lightly liquid loaded tower to see if any distribution (much less maldistribution) exists. Radioisotopic tracer tests are a better alternative for lightly liquid loaded towers.

Page 36 of 52



Figure 18. Liquid Maldistribution in a Packed Tower Bed 1 on left-hand plot had classic liquid maldistribution.

Page 37 of 52

3.01.3


3.3.4


3.01.3

Flooding Flooding in packed towers often initiates at the top or the bottom of a bed, since one of the two locations generally receives the highest vapor/liquid loading in the bed. When flooding occurs, the liquid holdup in the bed will start to increase, or the bed bulk density will increase in the flooding section. Gamma scans can easily detect the high-density section that is holding more liquid than rest of the bed, as shown in Figure 18. For a given size of packing and vapor rate there is a maximum amount of liquid that can pass through the packing. When this limit is exceeded liquid accumulates in the packing causing flooding. Usually below the affected section of packing the operation of the bed appears normal as the hydraulic limitation essentially acts as a metering device. The flooding will cause liquid to build upwards in the tower until the operability limit of the tower is surpassed. If the flooding point is at the bottom of a bed of packing then the entire bed could be completely inundated with liquid. In Figure 18 the top of Bed 2 was flooding. The top 5 – 6 feet of packing was very dense with excess liquid accumulation, but the liquid flooding had not yet reached the distributor above Bed 2.

3.3.5 Fouling Fouling in packed beds could take place anywhere (e.g. top, middle or bottom sections of the bed) over a time period of hours, days or months, depending on the fluid properties and operating conditions. Polymerization, coking, and solids deposit are among the most common types of fouling. The best way to diagnose or monitor the fouling process is to start from a baseline operating scan, where the tower is started up and operating in a known clean status. The tower is then scanned with the same scanning setups periodically to detect the changes of the bed density profiles. For most industrial towers the periods between successive gamma scans can be weeks, months, or up to one year. Scans monitoring the growth of fouling in beds can provide invaluable information on the fouling mechanism itself, and help those responsible to schedule turnarounds or select proper packings for longer run-length. As with fouling in trayed towers, scans of packed towers usually do not detect fouling directly. One possibility to directly detect fouling in packed towers is to scan the tower dry, when not operating. Dense spots in beds of packing, or on or below distributors, bed supports, etc. would be suspect areas where fouling may have built up. However, usually an operating scan showing the origin of problems such as flooding is enough to let operations know the source of the problem. The hydraulic problems (flooding and liquid maldistribution) seen in Figure 18 could easily be the result of fouling. Fouling in the bed could result in liquid maldistribution seen through Bed 1. Likewise fouling material carried into the top few layers of packing in Bed 2 could cause this type of flooding.

3.3.6

Foaming Foaming in packed beds is more difficult to diagnose than other hydraulic phenomena in packed towers, because of its unstable or irregular signatures on gamma scan plots. When liquid foams in a packed bed, the bubbles could be large in some areas and small in other areas. The large bubbles could move around or form and collapse in the bed, which would change the bed bulk densities randomly with time and locations. The foam

Page 38 of 52



3.01.3

bubbles increase resistance to the liquid flow through the packing causing local areas of excess liquid holdup. These random changes may or may not be visible on conventional sensors of temperature or pressure drop on the tower. When a packed tower grid scan shows dramatic spreads among the four scan plots, or there are large density fluctuations down one single scan line, then foaming is suspected. Dramatic fluctuations of the bed density profiles with time, at constant throughput, would be a good indication of foaming in the packed bed. Figure 19 shows a grid scan that found foaming. The left-hand plot in Figure 19 is the scan of the Absorber tower while at low gas rates. The scan showed some liquid maldistribution in the bottom bed and a density gradient of increasing density from bottom to top of bed, signifying there was some liquid holdup in the top section of the bed of packing. The right-hand plot in Figure 19 is the scan of the Absorber while at increased gas rates. Note the liquid maldistribution has worsened and the flooding seems to have propagated up the tower, through the distributor area and into the top bed of packing. But note the response of the green scan line through the bottom bed – there is a decrease in bulk density in the middle of the bed. This is a sure sign of foaming or frothing in the middle of the bed that was not present at the lower gas rate scan. Some of this lighter density liquid or foam also seems to have propagated up into the distributor area. Note that it may be difficult to distinguish between classical maldistribution and foaming induced maldistribution. When foaming occurs in a packed tower, a higher pressure drop is typically observed and the pressure drop may show signs of instability.

Page 39 of 52



Figure 19. Unstable and Irregular Scan Profiles could be an Indication of Foaming in Packed Bed

Page 40 of 52

3.01.3


3.3.7


3.01.3

Other Internals in Packed Towers

As for trayed tower scans, a scan of a packed tower is invaluable for checking the location of all packed tower internals. When collector trays, liquid distributors and redistributors are designed to maintain a liquid level, a gamma scan can be used to measure these liquid levels. When the levels are measured it is critical to know the detail design of the vapor risers. With that information the measured liquid level can be compared to the vapor riser height to determine if liquid is overflowing or nearly overflowing the risers. Any liquid overflow can severely degrade the performance of the liquid distributor device below. Figure 20 shows the scan results when a liquid distributor was found overflowing. In Figure 20 between the 6 and 7 meter elevation the blue and red scan lines cross perpendicular through the distributor parting box so that the liquid level in the parting box can be measured. The measurement on the red scan line showed approximately 120 mm of liquid; however, the blue scan line showed the parting box to be liquid full at that end and likely overflowing liquid. Below the parting box the blue scan line showed liquid flooding at the top of the bed of packing. This distributor was either out-of-level or damaged in some way that prevented it from working properly. As a consequence the liquid distribution through the bed of packing was not good.

Page 41 of 52



Figure 20. Scan Results Showing Overflowing Liquid Distributor

Page 42 of 52

3.01.3


3.4


3.01.3

Stationary Monitoring Test

Conventional gamma scans discussed above presents a density profile of a tower as a function of tower elevation. Stationary Monitoring uses an immobile radiation source and detector to monitor changes in density versus time, instead of density versus elevation, as shown below. This technique has been found to be useful in numerous applications, such as: • • •

Identifying the incipient flood point of trays or in packed beds, Differentiate the mechanisms of tray flooding between bubbling area or downcomers, Calibrate existing level indication instruments.

Stationary monitoring is similar to radioactive level controllers. However, the source for this application is much smaller than what would be commonly used in a permanently mounted device. Stationary monitoring also makes up for one of the other limitations of gamma scanning. A gamma scan is a snapshot of what the tower hydraulics are at that moment. The underlying assumption with the gamma scan procedure is that the tower stays relatively stable so that the scan data is representative of how the hydraulics are doing at the particular set of operating conditions in “steady state”. When it is not possible to hold the tower operation stable long enough to complete a desired scan, then stationary monitoring can be a viable substitute. To offset the limitation that stationary monitoring is only at one tower elevation. Multiple radiation sources and detectors may be employed to monitor several points simultaneously. Figure 21 shows an example when three points on a tower were monitored to see which point went into flooding first. The chart showed that Tray 2 (data represented by the red detector) flooded first. Crosschecking the monitoring time bar with the recorded process variables would show at what operating condition the flooding commenced.

Page 43 of 52



3.01.3

Figure 21. Results from Stationary Monitoring Showing What Point in Tower Flooded First

4

Other Techniques Using Radiation

4.1

CAT-Scans

CAT Scans (Computer Aided Tomography scans) are used to generate three-dimensional “images” of density for particular tower cross-sections, each at a single elevation. Whereas gamma scans (above) produce density profiles versus tower elevation, CAT scans produce density profiles in a horizontal plane. The idea and name are similar to the X-ray CT imaging used on human bodies, but higher energy gamma radiation is required for imaging towers. The basic methodology of CAT scans is shown in Figure 22 . The scan lines fan out from the source and cut across the tower in multiple directions, while the detector picks up the gamma penetration signals on the opposite side of the tower. After a full set of absorption measurements are taken for a chosen source orientation, the source is rotated to a new orientation and another set of absorption measurements are taken. Typically, at least nine source orientations are used, so this requires unhindered 360 ° access around the tower at the elevation of interest. The CAT scan as shown in Figure 22 would supply 27 independent measurements of gamma signals, but most process applications utilize a 9 X 9 or 11 X 11 grid resulting in 81 or 121 independent measurements (3,4).

Page 44 of 52



Figure 22 Layout of the Fan Beams and Image Result of CAT-Scans

Page 45 of 52

3.01.3



3.01.3

To generate a density image or plot from CAT scan data the density is regressed to a 4th order polynomial with respect to x and y directions. Although equations of other forms or degrees can be used, a 4th order polynomial is thought to give a reasonable compromise between sufficiency of detail and extent of data collection / computation. The measurement plane of CAT scans should be at least 1 foot (305 mm) away from the top or bottom of a packed bed (within ( within the packed bed itself), in order to reduce the backscatter signals. For evaluating spray header distributors, a CAT scan could be performed above the top of the bed. A grid scan should be performed on the tower to provide an initial look at the overall internal layout and liquid distribution patterns. The grid scan is also helpful to determine the exact elevations of CAT scans. It is recommended to overlay a liquid distributor sketch on the CAT scan images for the convenience of the liquid distribution analysis, with respect to the orientations of the parting boxes, troughs or vapor risers. The CAT scans compensate compensate for a limitation from only using a grid scan of a packed bed. As mentioned earlier, a grid scan could possibly show what appears to be good liquid distribution when in fact there is liquid maldistribution, e.g. an annular flow pattern. Figure 23 shows a grid scan of a tower that had poor separation efficiency. efficiency. The grid scan seemed to show what appeared to be fairly good liquid liquid distribution. distribution. However, since the tower was was not performing well a CAT scan was done near the top of the bed at the elevation shown in Figure 23. The CAT scan results are shown in Figure 24. The CAT scan showed showed a pattern of liquid maldistribution maldistribution where where more liquid was being distributed in the middle of the tower and less liquid was making it out to the end of the laterals. With the grid scan lines superimposed superimposed on the CAT scan results; one can see that all four scan lines passed through similar areas of density so that the average densities from each scan line were nearly the same.

Page 46 of 52



Figure 23. Grid Scan Shows Seemingly Seemingly Good Liquid Distribution Distribution

Page 47 of 52

3.01.3



3.01.3

Figure 24a – CAT Scan showing liquid distribution relative to the liquid distributor Figure 24b – CAT Scan showing liquid distribution relative to grid scan lines 3 Scale is calculated density in lb/ft

4.2

Neutron Backscatter Methods

The neutron backscatter device consists of a fast neutron source next to a slow neutron detector. Fast neutrons from the source collide with hydrogen nuclei in the material inside the vessel being scanned. The collisions transfer some of the energy to the hydrogen nuclei, reducing the neutrons’ speeds and changing their direction of travel in a process known as neutron moderation. Some of the affected neutrons are deflected back to the detector where they are received and counted. A high concentration of hydrogen nuclei in the material adjacent to the source will result in a large number of slow neutrons detected. In other words, the neutron backscatter technique is basically detecting the “concentration” of hydrogen atoms in the process fluids adjacent to vessel walls. Compared with gamma scans, the neutron backscatter device is easy to use and can usually be operated by a single person. The disadvantages of the technique are: • • •

•

Poor performance with a wall thickness greater than about 1.5" (40 mm); Maximum penetration into a vessel is only about 6" (150 mm); Insulation can make the results unstable or inaccurate because of moisture in the insulation, or by increasing the distance to the vessel walls, making it necessary to remove sections of insulation; It is only useful for detecting hydrogen-containing materials.

The neutron backscatter technique can be used to locate interfaces between materials having different concentrations of hydrogen atoms. In a storage tank, it can differentiate between the interfaces of vapor, oil, water, and sludge ( Figure 25). In a column having trays with downcomers, it can determine the froth height or liquid backup in downcomers. It is particularly useful for measuring the liquid height on chimney trays, on draw sumps, or in tower bases without having to do a full column scan.

Page 48 of 52



3.01.3

Figure 25 Neutron Backscatter Survey Measuring Liquid Interfaces

Generally speaking, the neutron backscatter device can be used to detect or estimate: • • • •

4.3

Liquid levels; Position of material or phase interfaces; Polymer buildup or plugging in pipelines; Relative vapor fractions in two-phase flow pipes, e.g. stratified flow.

Radioactive Tracers

The radioactive tracer technique involves injection of a very small amount (typically a few grams of material) of a radioactive substance into a process stream, which is then monitored with one or more radiation detectors positioned externally on the process equipment downstream of the injection. This technique is useful for measuring flow velocities, reactor residence times for back mixing or dispersion studies, equipment leak tests, and flow distribution. These tracer tests may also be used with non-radioactive tracers; however, sampling requirements can become onerous for some applications.

Selection of radioactive tracers for a specific process is critical for both data quality and environmental safety. An incorrectly chosen tracer may react with the process fluid and do unexpected things, resulting in erroneous conclusions. A tracer should be selected with a short halflife (typically measured in hours) such that the level of residual radioactivity will not interfere with downstream processes or cause a personnel or environmental hazard. Page 49 of 52



3.01.3

In general, the sealed source radiation methods discussed above should be considered first for tower troubleshooting. However, if the problem is not clearly identified by sealed source measurements tracer methods could be invaluable in getting important answers about internal flow behavior. Tracers could be used to detect entrainment in a tower. In this case, a nonvolatile tracer can be injected into the tower in a place where entrainment is suspected. Products above the injection point would then be monitored for presence of the nonvolatile tracer. In one instance a side-draw product from a crude oil distillation tower was off-specification due to high (dark) color. This would mean that heavy components were contaminating the side-draw product. The typical mechanism for this to happen could be tray damage, flooding, or high entrainment. A gamma scan had been done but none of these scenarios seemed to be taking place; only slight entrainment was noticed from the scan results. However since the problem persisted the decision was made to perform a tracer test to determine if entrainment was the problem. A radioactive tracer made from a high-boiling compound was injected into the crude tower below the product draw. The tracer compound was not volatile at the process conditions in the crude tower so the only mechanism for the tracer compound to appear in the side draw would be from entrainment. Radiation detectors were placed on the side-draw downcomers and piping to monitor for any tracer material as shown in Figure 26 . Detectors on the side-draw downcomers and piping showed a positive response to the radioactive tracer material, conclusively proving that high-boiling compounds were being entrained to the side-draw product.

` Figure 26. Radioactive Tracer Injection Showing Entrainment

Page 50 of 52



3.01.3

When applied to towers, radioactive tracers can be used to detect if a total draw tray is leaking or not. Such a test would involve choosing a non-volatile radiotracer, injecting it above the draw tray, and then monitoring a product stream below the draw tray for the presence of any radiation. One limitation to sealed source gamma scans is that they measure the density of the liquid phase in separation equipment. The density of the vapor phase is too low to be detected. An alternative method this is to study the vapor phase flow and distribution using gas radioactive tracers. At times problems in the ancillary equipment around a distillation or separation tower can be exacerbating or causing a problem. For example leaking reboilers or condensers could be the introducing a contaminant into the tower, or overloading it; or a leaking valve may be falsely loading up a tower. A radioactive tracer leak test could prove if such were the case. A material or energy balance required for troubleshooting or debottlenecking may not be closing. Some critical process streams may not have flow meters or the existing flow measurement may be questionable. Radioactive tracers can be injected to measure process flow velocities from which volumetric flows can be calculated.

4.4

Mobile Density Gauges Special mobile gamma density gauges can be applied for a more accurate detection of two‐ phase flow density in pipes than can be acquired with typical gamma scan equipment. The mobile density gauges are based on the same fundamentals as gamma scans; however, the density gauge uses a better gamma source collimator, more sensitive detector, and more deliberate calibration. Convenient access will also be needed to install the density gauge in the appropriate location. Common applications of mobile density gauges with distillation towers are to detect or monitor: • Liquid entrainment into overhead lines • Liquid fraction in reboiler return lines • Vapor carry‐under into liquid draw pipes

Page 51 of 52



3.01.3

Acknowledgement The Design Practice Committee gratefully appreciates the contributions and reviews of Lowell Pless, Business Development Manager of Tracerco to this section. References and Further Readings

1. Severance, W.A.N., “Advances in Radiation Scanning of Distillation Columns”, CEP, 81(9), P.38, 1981. 2. Basic Physics of Nuclear Medicine/Interaction of Radiation with Matter, Wikibooks Web site, http:wikibooks.org, 1 November 2007, Accessed June 12, 2009. 3. Xu, S. X. and G. Kennedy, “Gamma-Ray Computer-Aided Tomography of Industrial Packed Columns”, Paper presented at AIChE Spring National Meeting, Houston TX, March 17, 1999. 4. Xu, S. X., G. Kennedy, C. Conforti, T. Marut, and J. Dusseault, “Troubleshooting Industrial Packed Columns by Gamma-Ray Tomography, Paper presented at CE Expo ’99, Houston TX, June 9-10, 1999. 5. Bowman, J. D., “Use Column Scanning for Predictive Maintenance”, CEP, 87(2), P.25, 1991 6. Bowman, J.D., “Troubleshoot Packed Towers with Radio-isotopes”, CEP, 89(9), P.34, 1993 7. Kister, H., D.E. Grich and R. Yeley, “Better Feed Entry Ups Debutanizer Capacity”, PTQ Revamps and Operations, 2006 8. Kister, H., Distillation Operation, P.424, McGraw-Hill, New York, 1990 9. Xu, S.X. and W. Mixon, “Diagnosing Maldistribution in Towers”, CEP, 103(5), P.28, 2007 10. Xu, S.X., C. Winfield and J.D. Bowman, “How to Push a Tower to its Maximum Capacity”, Chemical Engineering, 105(8), P.100, 1998 11. Xu, S.X. and L. Pless, “Distillation Tower Flooding - More Complex than You Think”, Chemical Engineering, 109(6), P.60, 2002 12. Xu, S.X. and Pless, L., “Understand More Fundamentals of Distillation Column Operation from Gamma Scans”, Paper presented at AIChE Spring Meeting, Houston TX, April 22-26, 2001 13. Kister, H., Larson, K., Madsen, P., “Vapor Cross-flow Channeling on Sieve Trays: Fact or Myth ?”, Chemical Engineering Progress, P. 86, November 1992

Page 52 of 52


DESIGING FRACTIONATION SYSTEMS TO AID MAINTENANCE

Issued:

11/01/1988

3.02

Revised:

DESIGNING FRACTIONATION SYSTEMS TO AID MAINTENANCE

DESIGNING FRACTIONATION SYSTEMS TO AID MAINTENANCE ............................ 1 1.

Introduction ........................................................................................................................ 2

2.

Tower Manholes ................................................ ................................................................. 2

3. Tray Manways .................................................. ................................................................... 4 4.

Ladders and Platforms ........................................................................................................ 4

5.

Heat Exchangers ................................................ ................................................................. 4

6.

Miscellaneous ................................................. .................................................................... 5

Page 1 of 5


1


3.02

Introduction

Maintenance of fractionation systems often requires workers to enter the column to inspect the interior, fix damaged internals, clean plugged internals, and replace missing internals. Repair of fractionation systems can often be simplified by considering maintenance needs during the design phase. By providing cost-effective measures to aid maintenance crews, workers will be able to complete repairs more quickly in a safer environment to reduce column downtime. Fractionation systems require maintenance work due to three major causes: 1. Mechanical failure or deterioration of the column or internals 2. Plugging of the column internals 3. Routine inspections such as annual pressure vessel inspection Tower internals may fail due to process flow surges, vibration,, corrosion, erosion, or coking. Columns should be designed to minimize the need for maintenance when economical. Trays which will experience surging flows should be designed with greater mechanical strength. Hardware should be made of a resistant material to avoid general thinning and weakening due to corrosive attack. Weak or thin hardware can allow trays to slip free of clamps and collapse. Provisions should be made for access to the column and all auxiliary equipment. As a general rule, all internals as well as external hardware (valves, transmitters, etc.) should be accessible for removal and/or repair with the equipment normally used during their routine maintenance.

2

Tower Manholes Locating Manholes - For tray columns, tower manholes should be spaced to ensure all sections of the column can be adequately and safely serviced. Spacing tower manholes every 20 to 30 trays is typical. Larger spacing is used with clean, non-corrosive services, while a smaller spacing is often justified when a considerable amount of maintenance work inside the column is anticipated during short shutdowns. For example, if the trays must be frequently removed for cleaning, the number and size of the manholes should be increased compared to columns where the trays either do not require cleaning or can be cleaned in place. Having several tower manholes can allow different crews to work in the column together and reduce the number of tray manways that have to be removed to reach the desired location.

For very large diameter tray columns operating in dirty services with major support beams, multiple tower manholes at a given elevation may be justified. In general, however, providing more than one tower manhole at a given tray is not economical. Besides the cost of more manholes, it requires larger access platforms, additional lighting, and creates added sources for leakage. With multi-pass trays, removable downcomer sections are preferred over multiple tower manholes at a given tray elevation to provide adequate access. Tray spacing might need to be increased to allow for the manhole size. Therefore, manholes should be installed in the space above the feed trays (where the tray spacing is often raised), above the top tray, and below the bottom tray. When manholes are installed at the feed tray, their orientation and elevation should be specified to avoid feed distributors and other internals that might impede entry. All packed column internals (particularly distributors and redistributors) should be quickly accessible for Page 2 of 5



3.02

inspection given their importance in the operation of the column. Therefore, tower manholes for large columns and body flanges for small columns should be located above and below all packed beds. Elevations where the column has a feed diameter change, or contacting device change should have provisions to be quickly reached since the chances for errors or failures are greater at these locations. Therefore, tower manholes should be located at, or very close to, elevations where there is a feed, diameter change, or contacting device change. Sizing Manholes - Tower manholes are typically 18 inches to 24 inches. As a general rule, tower manholes should be adequately sized to ensure that all repair and/or replacement parts or tools can be passed to the column’s interior. Larger tower manholes have the following advantages:

a.

Allow fabrication of larger panels or beams for internals and may therefore lower the internals’ cost. b. Allow the fewer, larger panels or beams for the internals to be more quickly installed or removed. c. Provide easier and safer access to the column, particularly if a breathing apparatus must be worn. Very large diameter columns with major beams may require 36 to 48 inch diameter manholes for removal of the beams. As the beams are usually attached prior to welding the heads onto the vessel shell, extra large manholes are not, however, required to install the beams during initial construction. Although they will allow removal of the beams without cutting holes in the vessel shell, extra large manholes economics should be carefully evaluated. Drawbacks of these extra large manholes include: a. They are more difficult to seal and provide more surface area for leakage. b. They are more expensive. c. They are much more difficult to open and close due to the greater cover weight which requires larger d. They require an increase in the tray spacing and a subsequently taller column - a significant cost for large diameter columns. Orienting Manholes - For tray columns, tower manholes should be located above their respective tray’s bubbling area, not behind their downcomers.

Tower manholes should be oriented so they can be easily reached by a crane lifting column internals. Also, manholes and their service platforms, should be oriented away from major pipe runs and other possible interferences to provide an unobstructed drop for lowering column internals by rope. Accessing Manholes - All tower manholes should be equipped with a platform if the manholes are not easily accessible from a floor level. The platform should be large enough to allow workers to open or close the tower manholes, enter the column, and remove internals (feed pipes, tray panels, beams, etc.) as needed. Platforms should be located 12 to 36 inches below the bottom of tower manholes to provide easy access. The platform and ladders should be oriented so the manhole cover can be opened away from the ladder leading to the platform. Also, the manhole cover should be capable of being fully opened without interference from rails, nozzles, pipe supports, etc.

Tower manholes should be located at an elevation no more than 36 inches above an internal that provides an adequate footrest. For example, tower manholes located too far above the bottom head would cause a serious safety hazard for workers entering or leaving the column base unless there was an internal ladder.

Page 3 of 5

Issued: 11/01/1988


Revised:

3

MAINTENANCE

3.02

Tray Manways

Tray column manways should be top and bottom removable, so maintenance crews can work up or down the column to open it for initial inspection. Quick-opening tray manways are often justified if frequent tray inspections are anticipated due to a fouling, corrosive, etc. service. Tray manways should be large enough to allow passage of workers and tools and light enough to be lifted by a single worker. Thrift in reference 23 has recommended that manways be a minimum of 12 inches x 16 inches and not exceed 65 lb. Tray manways are usually vertically aligned between tower manhole elevations to provide a clear vertical shaft for workers to easily pass tools and tray parts from one elevation to another. It is preferable to stagger manways in adjacent tray sections (as bounded by tower manhole elevations). Vertically aligned tray manways make it easier to evacuate the column in the event of an emergency and to communicate from one elevation to another. Care must, however, be exercised by mechanics to avoid dropping tools or tray parts on those working below them. In addition, mechanics must be careful to avoid slipping, as they could fall a great distance with vertically aligned tray man ways. Tray manways should be located to provide access to all parts of the tray. Multi-pass trays will require a tray manway in each pass. Large diameter columns with deep beams will require a tray manway in each tray section workers cannot reach by crawling below the he beams. Trays should be spaced to provide a minimum clearance of 14 inches from the bottom of the deepest truss or beam to the top of the tray below (or the top of the caps for a bubble cap tray). This clearance allows most workers to reach all sections beneath trays by crawl ing under the beams.

4

Ladders and Platforms

Internal ladders should be provided when necessary to allow a complete inspection of all the internals in a column. If the expense can be justified, towers can be grouped to allow access to their platforms from a common stairwell with an interconnecting walkway or, preferably, a common floor can be provided to access all columns. Stairs are much easier to climb than ladders and provide a safer passageway for a mechanic carrying heavy tools. A service elevator connecting floor and/or walkway elevations provides mechanics even quicker access to the towers. Valves and pressure, differential pressure, level, and control temperature transmitters should be accessible from a floor, platform, stairway or ladder. If platforms are only accessed by ladder, then the platforms should be no greater than 30 feet apart to avoid fatigue of the climbing worker. Providing platforms to access column body flanges may avoid the need to use two cranes (one to hold a basket with mechanics to unbolt the flange and a second to lift the top section) to disassemble the column sections at a later date.

5

Heat Exchangers

Heat exchanger surfaces which are expected to foul should be accessible for cleaning. This includes Page 4 of 5



3.02

condenser and reboiler tubes and shells as well as air cooled condensers exterior finned surfaces. If an exchanger has a removable bundle, room should be provided around the exchanger for removing the bundle without interference interference from other equipment. As shutdown time is often costly in terms of lost production, consideration should be given to minimizing the time required to access the column or auxiliaries. For example, using “A” or “C” style top heads for large, vertical thermosiphon reboilers only requires that one flanged joint be unbolted to access the tubes for cleaning as opposed to two flanged joints with a “B” style head.

6

Miscellaneous

Sufficient room should be provided to allow maintenance equipment access to the process equipment in the fractionation system. Particular care should be taken in the plant layout to make certain that cranes can reach columns and auxiliary equipment. Occasionally the use of monorails and hoists is economical given maintenance requirements for difficult to access pumps, compressors, etc. When practical, internal, feed distribution pipes, or spargers, should be designed to be removable by mechanics located outside the column. This allows the feed pipe to be removed so interferences with installation or removal of the other tower internals is avoided. Also, maintenance or modification of the feed pipes can be done without having to clean the column sufficiently to allow workers to enter. For columns with body flanges, breaking piping and ladders at the body flange joint greatly eases disassembling the column. In addition, adjacent shell sections should be punch marked to aid accurate alignment when the sections are assembled. A utility station should be installed close to columns to allow mechanics quick access to power for lights and welding machines, air for pneumatic wrenches, water for high pressure cleaning equipment, etc. As pressure relief valves are often heavy and require frequent inspection, consideration should be given to locating them at the condenser’s inlet, which may be accessible from a top floor elevation, as opposed to requiring the use of a crane to remove them from the column’s top head. Tower internals should be designed to aid in their own removal. Material of construction for bolts and nuts should be chosen to avoid galling and corrosion so they can be quickly untightened. It is preferable that nuts be located on the top side of tray decks and the outside of downcomer panels. Tray parts should be clamped together and to support rings to avoid problems in aligning with bolt holes. If through bolting is used, the bolt holes should be oversized or slotted to allow for fit up. For fouling systems which will be chemically cleaned, adequate nozzles must be provided on the column and external piping to allow for the addition, circulation, and draining of the solvent(s). It is usually desirable to design the column shell and foundations for a full hydrostatic load. This allows the column to be hydrostatically tested in-place, filled with water to “rinse” the column, or provides for vessel integrity in the event the column base level rises up into the column proper. Column “rinsing” is often necessary if steaming or other forms of displacement may not completely remove all hazardous materials. To provide a safe working environment, drainage should be provided for areas in the column where liquid can accumulate. Although drain holes may plug, they should still be used. Also, the column operator should adequately prepare the column for entry by required suitable “washing” techniques. A record of potential stagnant areas should should be included in the the operating procedures and and maintenance records.

Page 5 of 5


DESIGNING FOR FOULING SERVICES

Issued:

12/15/1990

3.03

Revised:


DESIGNING FOR FOULING SERVICES .............................. ................................................ 1 1.

Introduction ....................................................... ............................................................................................................. ................................................................. ........... 2

2.

Selection of Device .................................................. .......................................................... .2

3.

Trays ...................................................... ............................................................................................................. ............................................................................. ...................... 3

4.

Packed Columns ................................................ .................................................... ................................................................. ............. 3

Page 1 of 4

Issued: 12/15/1990


Revised:

1

3.03

Introduction

Fouling of column internals can be caused by a number of different mechanisms. These include thermal degradation (coking), polymer formation, corrosion, salt formation, ingress of external matter, etc. Fouling has the common result that column performance deteriorates, often to the point of shut down. In severe cases permanent damage to the shell and internals can occur. Product quality and unit capacity can be adversely affected. In many instances the extent of fouling and its effects can be minimized - or avoided - by appropriate design and selection of internals. This section considers what design practices can be used to achieve that objective.

2

Selection of Device

Whether or not trays or packings are selected will depend on the particular application, previous experience, user preferences, etc. While specific comments are given later on trays and packings, the following general points should be borne in mind. a.

A clear distinction should be made between fouling and corrosion. While both cause fouling in a generic sense, fouling normally refers to blockages resulting in a change in fluid state, whereas corrosion refers to blockages resulting from chemical action of the fluids with the hardware. b. Often the exact nature of fouling (even if known) is not conveyed to the vendor. If the choice of internal is going to rely on vendor experience it is of the utmost importance that the known fouling characteristics or tendencies are conveyed. c. The user should consider the use of internals which better resist fouling even though these may be regarded as being less efficient devices. Continuous operations with a “less efficient” internal may well prove to be more cost effective than intermittent operation with “high efficiency” internal. The use of a larger than normal safety factor may be required. d. While it is difficult to give specific recommendations on suitable internals, the following guide may be helpful, and generally summarizes the data given in this section. DEVICE

All trays

COMMENT Do not use seal pans

Avoid the use of inlet weirs if possible Maintain 1.5 inch downcomer clearance minimum Sieve tray

Usually acceptable with hole size 0.5 inch or greater

Dualflow tray

Commonly used with 1 inch hole size

Valve tray

Caged or fixed valve preferred

Packing

Discuss with vendor Distributor may be the problem

Bubble Cap

Not advised for fouling services

Baffle Tray

Use sloped baffles

Page 2 of 4


3


3.03

Trays

The basic philosophy should be to ensure that there are minimal stagnant areas or areas of abnormally low velocity where either corrosive products can accumulate or where localized residence times can be sufficiently high to cause thermal degradation or polymerization. The residence time profile on a tray is influenced by the shape of the downcomer baffle and specific modifications may be possible. The reader is advised to refer to “reference (30)” for further background on this subject. Relatively stagnant zones can also occur in large downcomers where liquid flow is low. Consider carefully the layout of the tray, particularly with respect to inlet streams and manifolds. The following specific points should be observed: a. b. c.

d.

e.

f.

g. h.

4

Avoid the use of recessed seal pans as these contain stagnant liquid zones. They are also liable to accumulate solid material which can promote corrosion. Avoid the use of inlet weirs. These can cause the accumulation of debris which may reduce the effective under downcomer area. Sieve tray hole size should not be less than 0.5 inch otherwise blockage or partial blockage may occur. It should be noted that even a small diameter reduction (for example, to 0.375 in) can considerably increase the probability of fouling. Use of “larger” hole sizes would not be expected to reduce efficiency. Dualflow trays with a 1” hole size are commonly used in fouling service. In corrosive service tray deck metallurgy should be particularly carefully considered. Corrosion can cause hole size to increase in both sieve and valve trays. In the case of valve trays, caps can then become dislodged. In all cases tray efficiency will deteriorate. In some corrosive services products of corrosion can exert a considerable force on such items as fastenings, bolts, etc. Bolt torque settings recommended by the vendor/manufacturer/licenser should be strictly adhered to. Where valve trays are used, it is important to ensure that the valves are operating fully open. Under part load conditions, valves are generally either fully open or fully closed. It has been specifically noticed that under some fouling conditions (for example, where ammonium salts can deposit), closed valves are likely to become blocked and remain closed. The possible problems of associated flexibility should be discussed with the tray manufacturer. The use of fixed as opposed to movable valves, should be considered where turndown characteristics permit. Where valve trays are used, the caps should have indentations to prevent full contact with the tray deck. If this form of cap is not used, there is a danger of the cap sticking closed. Caged type valves, where the cage legs are less susceptible to frictional wear or corrosion than noncaged valves, are generally preferred.

Packed Columns

While the most critical area in a packed column is the distributor, all internals need to be carefully considered. The following guidelines should be noted which in many cases also constitute good design practice for all systems. Note that in some cases trays would be preferred. a.

Where a spray distributor feed consists of, or contains, recycle streams, 100% duplex filters should be used. It should be particularly noted that although the anticipated contaminant particle size may appear small in relation to the distributor hole size, agglomeration may occur which can ultimately lead to orifice blocking. A filter mesh size of 0.5 mm is commonly used. Filters may be duplex. b. Discharge from gravity distributors should always be from above the distributor base so that orifices do not become blocked with debris or corrosion products. Page 3 of 4


c.

d.

e. f. g.


3.03

Packing size and orientation should be such yes to minimize the accumulation of contaminants. Ordered packings, particularly those with a predominantly vertically oriented structure, are more likely to resist the accumulation of fouling material. Note, however, that the formation of such material will not be avoided - it will merely be passed on to another part of the column. Thin materials are increasingly used in modern packings. Consequently, corrosion which may be tolerated on other internals, can cause considerable problems. Choose the metallurgy carefully. Oxidation, for example, is a form of corrosion which commonly causes problems with carbon steel packings. Where cost considerations prevent the use of alloy steel, ensure that operating procedures are available to handle correct shutdown and start up procedures. Complex internal pipework or other restrictions that may become blocked should be avoided. The assembled distributor should be protected from corrosion products generated within its own environment. While it is not practical to state specific orifice sizes, the minimum diameter should be significantly larger than the normal minimum used in clean services. The use of V-notch distributors is often preferable in fouling services. Where orifice distributors are used, V-notch overflows should always be specified.

Page 4 of 4


DESIGNING FOR EASE OF INSTALLATION

Issued:

1/15/1996

3.04

Revised:

DESIGNING FOR EASE OF INSTALLATION

DESIGNING FOR EASE OF INSTALLATION...................................................................... 1 1.

Introduction ........................................................................................................................ 2

2.

Vessels ................................................................................................................................ 2

3.

Plant Layout ........................................................................................................................ 4

4.

New Installations ............................................................................................................... .6

5.

Order of Installation................................................... .........................................................6

6.

Field Service Companies .................................................................................................... 7

Page 1 of 7


1

DESIGNING FOR EASE OF INSTAL LA TION

3.04

Introduction

Proper installation of tower internals is necessary for a distillation column to operate as designed. If the internals are difficult to install, however, then the time required for installation can lead to high labor costs and unnecessarily extended plant shutdowns. Installation of tower internals can be expedited by 1. Providing easy access to the vessel interior. 2. Providing suitable means of transporting parts to the proper location, both on the plant site and within the tower. 3. Simplifying designs and installation procedures so the internals can be installed quickly. All three of the items listed above can be accomplished, to a large degree, by incorporating certain features into the tower design and plant layout; some of these features will be discussed in the sections that follow. Other ideas for expediting internals installation will be mentioned as well. This discussion focuses on the things that can be done to make internals installation easy. Detailed installation procedures are not covered. Internals suppliers can provide detailed installation procedures for any products they manufacture.

2

Vessels

The vessel should be designed to allow easy access to the tower internals. The most critical item is the point of entry - either a manhole or a body flange. Access to the entry points must also be provided, and the best method of transporting material from the ground to the entry points must be considered. Manhole - The most common method of entering vessels is via a manhole. A manhole should be located at the top and bottom of the tower, and at appropriate locations in between. For trayed towers, this is usually every 10 to 30 trays. If the tower contains two-pass trays, then any manholes in the trayed section should be located above trays with central downcomers. For packed columns, a manhole should be located at each distributor/redistributor.

Several factors must be weighed when determining the optimum number of manholes for a given tower. The benefits of each additional manhole are: 1. Access to the tower - During turnarounds, a crew can be working at each manhole. The more manholes that are available, the shorter the time the tower is out of service. 2. Reduced transportation distance within the tower - Parts have to be transported from the closest manhole to the proper location inside the column. If only a few manholes are available, then the distance that parts must be transported within the column can be quite large; this transportation time can extend the shutdown. The costs associated with each manhole include: 1. Additional tower height and weight - Typical tray spacings do not allow a manhole to be installed; the tray spacing should be 30 to 36 inches (762-915 mm) at an entry point. The manhole nozzle and cover are relatively thick and can add substantial weight to the tower. Unless the condenser and reflux

Page 2 of 7



3.04

drum are mounted above the column, the additional height requires slight increases in the lengths of reflux piping, overhead vapor piping, and access ladders, as well as a slight increase in the head requirements of the reflux pump (and the feed pump, if the manhole is to be located in the stripping section). The additional height may also result in a thicker shell at the bottom of the tower in order to meet pressure rating requirements, and the additional height and weight must be considered in evaluating the foundation/support requirements for the vessel. 2. Access to the manhole must be provided - usually, this means a platform of sufficient size to make the manhole useful as an entry point. Manholes are normally installed on the vessel straight-side, but it may be more convenient in some instances to locate the top manhole in the top head. Manholes on the vessel straight-side should be in line, or approximately so, and facing an open area in the plant, such as a courtyard or alley. The manhole cover is usually supported by a davit or hinge for safety reasons. The cover should swing away from the access ladder. The cover should open fully - no piping, instruments, etc., should be located so that they interfere with the manhole cover, or with access to the open manhole. Adequately sized manholes will greatly simplify internals installation. Most manholes are nominally 24 inches (610 mm) in diameter; occasionally 18 or 20 inches (457-508 mm) is used. Determining the optimum manhole size requires consideration of several cost factors. The larger the manhole, the wider the tray panels or larger the distributor pieces that can be fabricated. This reduces the number of parts that must be transported within the column and the number of fastener assemblies that must be tightened in the tower, shortening installation time. Unless there are some unusual considerations, manhole sizes in excess of 24 inches (610 mm) are not practical since internals designed to pass larger manholes become unwieldy. The cost of an incrementally larger manhole is affected by the tower metallurgy and pressure rating. In addition, an incrementally larger manhole increases the tower height by an incremental amount, so all the factors listed under Item 1 of the list of costs associated with each manhole apply here as well. Whatever the manhole size, it is important to communicate the exact size to the internals vendor when ordering internals. The nominal diameter usually corresponds to the outside diameter of the entry nozzle, but it is obviously the inside diameter that is of interest in designing the internals. If the vessel is clad, then careful attention must be paid to ensure that the diameter through which parts must pass is the diameter reported to the vendor. If the internals are fabricated in parts too large to pass the manhole, the necessary field modifications can be very costly and time consuming. Body Flange - Access to the vessel interior can also be provided by breaking a body flange and lifting the top section of the column off with a crane. This method of access is common for towers too small to enter, but may be used for larger diameter towers as well. This allows internals to be fabricated in pieces of the largest possible size (even single piece, if desired). For towers containing trays clamped to weldedin support rings, the installation of downcomer aprons that extend across a body flange require special consideration. Ladders and Platforms - Worker access to the manhole or body flange is essential. The most common method of providing this access is to locate a platform just below each manhole or body flange.

Access to the platform is by means of ladders which are bolted to clips on the vessel. In the United States, the Occupational Safety and Health Administration (OSHA) requires that a ladder run between platforms not exceed 30 feet (9.1 m). A small platform can be used if its only function is to comply with the OSHA regulation, but platforms that are to be used for tower access should be as large as practical. Towers that are located inside buildings may be accessed from the floors within the building. If this means of access is to be used, the designer should take care to locate the manholes such that they are Page 3 of 7



3.04

roughly 3 to 5 feet (0.9-1.5 m) above a floor. Davits - A davit arm is often installed at the top of a tower to allow the tower internals and other items necessary to install them to be raised to a platform from the ground. If a crane of sufficient capacity is available to the plant site, davit arms may not be necessary. At refineries, davit arms are usually used because of the large number of towers that must be serviced during a turnaround.

Davit holders are sometimes installed inside vessels near the top head. During the shutdown, a davit can be placed in the holder to allow material transport within the tower. This design feature would also facilitate the removal of an injured worker. Other Considerations Related to Vessel Design - Attention should be paid to vessel out-of-roundness; cases have been reported of even large diameter towers being so out-of-round that tray panels, designed and fabricated under the assumption that the tower was within ASME roundness tolerances, would not fit. If a vessel is known to be out-of-round, the internals vendor should be apprised of this fact when internals are ordered.

Some random packed tower designs use a large nozzle near the bottom of each packed bed to allow the packing to be dumped. With the advent of vacuum trucks that can remove the packing from a tower, this design practice is declining in popularity. Small diameter towers may require special considerations. Access is often provided by body flanges, but hand holes are sometimes included near critical internals to allow cleaning or maintenance without disassembling the vessel. If cartridge trays or structured packing in which each layer is a single piece is to be used in the column, then the walls must be smooth; no rings or clips can be used for distributor support, and nozzles may not protrude into the vessel.

3

Plant Layout

There are features that should be incorporated into the plant layout and piping arrangement that can make installation of tower internals much easier. The tower should be sited near an open area, such as a courtyard or alley. This area can be used to park a crane, if necessary for column access, or to loosely assemble trays or distributors prior to installation in the tower. Loose assembly of internals on the ground is a practice recommended by all internals vendors; it ensures that the proper parts and hardware are available, that they fit correctly, and that they are installed in the tower in the proper order. If a body flange is to be used for access, then any piping extending across the flange should either be flanged or easily removable. Any electrical conduit crossing the flange should have a junction box near the flange. Internals - There are also features that can be incorporated into the design of the internals that can expedite the installation procedure. Manways - It is critical that the internals be fabricated in pieces that are sized to pass not only the manhole, but any manways as well. Manways are worker passages located in each tray; the manways should be removable from either the top or bottom of the tray. Manways are usually designed to be inline; that is, the manways on each tray are in the same location as those on the tray above. When all the manways are open, a "mineshaft" is created, giving the greatest open area for worker and material passage. Sometimes, in an effort to minimize the distance that a worker could fall, the manways are designed to be partially in-line; that is, the manways are staggered from tray to tray. This gives a large opening on any given tray, but creates a smaller "mineshaft". Sometimes, shorter "mineshafts" are created by staggering

Page 4 of 7



3.04

the manways every 10 trays or so. For towers that require frequent entry, quick-opening manways can be used. Each quick-opening manway uses only a few heavy-duty fasteners, allowing the manway to be removed much more quickly than conventional manways. Fasteners - Trays are usually attached to a support ring by means of a ledge clamp. The bolt may be loose or welded to the ledge clamp. If the bolt is welded to the clamp, then the top of the bolt has a screwdriver slot or some other directional marking that allows a single worker on top of the tray to install the tray fastener assembly. If the bolt and clamp are separate pieces, then two workers may be required to install a fastener assembly, one above the tray and one below. The separate piece design is less expensive, and allows the bolts to be replaced independently of the clamp, but may incur a higher installation cost than the single piece design.

Tray panel connections are made with fasteners featuring frictional washers or by through-bolting the tray panels. The installation of these tray panel connections can be greatly expedited if the nut is tack welded to the underside of the tray deck or the outside of a truss. In corrosive or fouling services, where frequent replacement of threaded parts will be required, this may not be advisable. Internals vendors will typically use a single nut for all tray fastener assemblies other than manway fasteners. Double nuts can be specified for severe services. Installation costs can be reduced if the spacing between fasteners is increased. This could compromise the leak resistance and structural integrity of the tray or distributor, however, and is not recommended. It is best to allow the supplier to specify the number and location of the fasteners. Other Considerations Related to Internals Design - If beams are used to support the trays or other internals, installation may be more complicated than otherwise. At the very least, a set of manways on each side of the beam should be included in the tray design.

If a beam or other internal part (such as a distributor trough) is to span the entire column diameter, then the manhole nozzle should not protrude into the vessel. It may be impossible to get the part into the tower unless the protrusion is removed. If lattice trusses are used, and the nominal manhole diameter is less than the tray spacing, then one oversized manhole may be justified to avoid building or splicing the trusses in the vessel. Installations that involve internals that vary slightly in some minor details, such as sieve trays or distributors with different hole sizes or metallurgy, can be easily botched. The installation can be simplified by making as many parts identical as possible. When the differences are critical, the part numbers and metal mark, if appropriate, should be clearly stamped on each part and the trays or distributors should be loosely assembled on the ground. Another good idea is to install all trays of one design before the crates containing the other trays are opened. When performing maintenance installations (i.e., replacement-in-kind), care must be taken to order parts of the exact size needed, particularly when obtaining the parts from a source other than the original vendor. Frictional washers must be large enough to adequately overlap both panels, and the lip on the ledge clamps should be of the proper length to allow the assembly to properly clamp to the support ring. Replacement valves should be of the proper thickness and have the proper leg length. Bubble cap/riser assemblies should be of the correct dimensions and provide the desired slot area. Replacement random packing should be of the proper size. The use of the proper tools will greatly expedite the installation of tower internals. Air-driven wrenches, if used, should have torque control to prevent overtightening the fasteners. If new round valves are being installed in existing valve trays, then the vendor's recommendations for spreading the valve legs should be followed, and a tool supplied by the vendor should be used to perform this task. If structured packing is Page 5 of 7



3.04

being installed, suitable shoehorns should be used for installing the final bricks in each layer. The installation of trays can be speeded up by using templates as much as possible. Templates can be made to set the distance between sieve holes, for example, in adjacent panels; this greatly reduces the amount of time that the workers must spend locating the panels for proper overlap of the frictional washers. Templates can also be used to set downcomer clearances, weir heights, etc. Avoid welding (especially seal welding) internals in situ if at all possible. Seal welding is time-consuming and shutdowns could be unnecessarily prolonged if it is required indiscriminately. When seal welding internals, the use of pilot bolting should be considered. Pilot bolting has the advantage of ensuring that all of the parts are in proper alignment prior to welding. The disadvantage is that the pilot bolts must also be seal welded. It is a good idea to notify the internals manufacturer if the internals are to be installed at the vessel fabricator's shop, particularly if the tower is to be a packed column.

4

New Installations

Much of the discussion to this point has been directed to maintenance/revamp concerns related to tower internals installation, primarily because time is of the essence during a shutdown. During the construction of a new plant, internals installation is seldom on the critical path, but there are some things that can be done to speed up installation. For vessels fabricated at a vendors shop, trays can be shipped to the vessel fabricator and installed there so that the tower is delivered completely assembled. Once the tower is set, the manways should be removed and the installation inspected. It is possible to have the vessel fabricator install packed tower internals, but this is generally not recommended. The tower should be in an upright position when it is packed with random dumped packing to minimize non-uniformity in the beds. The structured packing in a tower could become crushed or deformed during shipment in a column. The tower must also be vertical when the distributors are leveled.

5

Order of Installation

Installation of internals inside the tower will go much more smoothly if another crew is working outside the tower on the ground, loosely assembling each tray or distributor. The hardware can be loose-fitted into the appropriate tray panel or distributor part. The parts can then be hoisted into the tower as needed by the installation crew. This procedure ensures that the correct parts can be supplied quickly and in the correct order. The usual order of installation for tray parts is: 1. 2. 3. 4. 5.

Under downcomer pieces Downcomer apron Tray panels farthest from column center Tray panels closest to column center Manways

This order usually allows the parts to be installed with the minimum amount of rework (for improper overlap of frictional washers, etc.). The vendor's recommended installation procedure should be followed as closely as possible.

Page 6 of 7


6


3.04

Field Service Companies

Another means of having tower internals installed quickly is to contract the job to a field service company. Most internals suppliers are affiliated with a field engineering organization, and there are independent companies that install internals. Field service crews are typically more experienced at internals installation than the maintenance crews at a refinery or chemical plant. These crews have the proper tools to perform the necessary tasks. The field service organizations that are affiliated with an internals supplier also have quick access to any parts that must be supplied on an emergency basis.

Page 7 of 7


LOCATION OF INSTRUMENTS

Issued:

1/15/1995

3.05

Revised:

LOCATION OF INSTRUMENTS

LOCATION OF INSTRUMENTS ................................................... ......................................... 1 1.

Introduction ........................................................................................................................ 2

2.

Pressure Measurements ...................................................................................................... 2

3.

Temperature Measurements................................................................... ............................. 3

4.

Level Measurements ...........................................................................................................4

5. Flow Measurements .............................................. ................................................... ............ 5 6. Alarms and Shutdowns ............................................. ........................................................... 6

Page 1 of 16

Issued: 1/15/1995

LOCA TION OF INSTURMENTS

Revised:

1

3.05

Introduction

The purposes of column instrumentation are: 1.

To provide stable operating conditions.

2. To make sure the outflow streams meet the distillation specifications. 3. To minimize the operating cost of the distillation step. 4. To provide column diagnostic data. 5. To help operators startup and shutdown distillation columns. 6. To help operators recover from column upset conditions. While normal fractionating column operation requires a minimum of instrumentation, additional instrumentation is helpful in troubleshooting column problems and in debottlenecking column operations. Where the cost of the additional instrumentation is excessive, consider installing nozzles and thermowells so that additional instrumentation can be installed later without altering the column. Reference should be made the Troubleshooting section of this Design Practices Handbook which discusses the location and use of instrumentation for troubleshooting. "Information supplied by instruments is vital for column control and operation. False information prompts wrong operator (or control loop) action, erratic column operation, and incorrect diagnosis of problems" (49). Instrumentation connections should be designed to avoid hydraulic interference that could cause erroneous measurements or instrument malfunctions (49). H. Kister lists 12 general considerations for instrument connections on pages 123-125 (49). Smart transmitters have greater accuracy (about three times as accurate) compared to conventional transmitters. A smart transmitter automatically compensates for ambient temperature and operating pressure effects on the transmitter output. Their lower maintenance costs often make them a good choice.

2

Pressure Measurements

The minimum instrumentation recommended is the pressure drop across the entire column and the pressure at the top of the column. Additional useful instrumentation includes the pressure drop across each section of the column, the pressure of the reflux drum and in some instances, the reboiler steam chest. Where high pressure drop is the symptom of operating problems, it is very helpful to locate the section of packing or trays having the high pressure drop. In vacuum towers or any mixed phase line where temperature measurements are used to estimate composition, it is important to have a pressure sensor near the temperature sensor. This is because accurate pressure and temperature measurements are required to accurately estimate composition. H. Kister lists seven important considerations for pressure and differential pressure measurements connections on page 130-134 (49). Generally locate the pressure taps to avoid high liquid or vapor velocities and where entrainment into the pressure lines is unlikely. Pigtails ( Figure 1) should be installed in lines to pressure gages measuring condensing vapor pressure, e.g., steam pressure, to generate and keep a liquid seal. This seal prevents condensing vapors from causing vacuum formation that could cause fluctuating pressures and excessive wear on the pressure gages. Page 2 of 16



3.05

Generally locate the top column pressure tap in the top head of the distillation column. Where installation into the top of the head is very difficult, a side mounted tap should be considered. Sometimes the pressure tap is on the overhead line. When a pressure measuring element is on an overhead vapor line that is near a building ceiling, there can be a tendency to mount the element under the line. This makes it more readily accessible. However avoid installing pressure sensing elements such that they do not self drain into the process line. This prevents an altered pressure reading caused by liquid accumulation in the sensor line (Figure 2). Measure the pressure drop across groups of trays or a packed bed with a differential pressure sensor. A less preferred alternative is to infer the pressure drop by difference between two pressure measurements. Figure 3 shows pressure drop measurement across a column.) A differential pressure sensor is especially important when the measured pressure drop is less than ten percent of the absolute pressure. A differential pressure tap should not be located in the overhead vapor line. A tap in the overhead line does not only measure the pressure drop across groups of trays or a packed bed. It adds to that overhead head line entrance loss and pressure drop. Also there can be a velocity effect error caused by flow across the tap. Conventional pressure instruments are accurate to about plus or minus 0.5% of the calibrated span. Thus for a column operating at 100 psia (6.9 bar) with conventional pressure sensors calibrated for 150 psia (10.3 bar) at the top and the bottom of a section having a 5 psi (0.34 bar) pressure drop, the accuracy would be plus or minus 1.5 psi (0.10 bar). A conventional differential pressure instrument should have an accuracy of about 0.5% of the calibrated span. Thus a differential pressure sensor calibrated for 10 psi (0.69 bar) would have an accuracy in this application of 5 psi (0.34 bar) plus or minus 0.05 psi (0.003 bar). Smart transmitters are about three times as accurate as conventional transmitters. If this example used smart pressure transmitters, the accuracy would be plus or minus 0.5 psi (0.03 bar). If the example used a smart differential pressure transmitter, the accuracy would be plus or minus 0.02 psi (0.0014 bar). When measuring differential pressure across a column with a high density vapor (such as a depropanizer), be aware that unlike a vacuum column, the static vapor head is significant compared to the flow induced pressure drop. Consider biasing the transmitter output to eliminate the static vapor head reading. When measuring the pressure in a high vacuum system with a gage, use an absolute pressure (capsule) gage. Regular pressure gages balance the measured pressure against the atmospheric pressure that changes from day to day. An absolute pressure gage has an internal standard. When using a single differential pressure instrument (upper sketch in Figure 3) the transmitter to bottom connection leg must be sufficiently large to permit proper drainage and venting. Insulating the legs will help avoid excessive condensation. If the vapor condensing temperature is above ambient, condensation can be a problem even if well insulated. Heat tracing or gas purging of the legs may be necessary. Figure 4 illustrates problems with pressure tap locations. In Figure 4a, excess weld metal on the column inside deflects the flow pattern causing low pressure readings. The middle drawing illustrates an incorrect pressure reading caused by the disruption of flow on a bend in the piping. The bottom drawing shows a correct pressure measurement.

3

Temperature Measurements

Recommended temperature measurements are the column top and bottom, the reflux drum and in all feed lines to and sidedraw lines from the column. For a single-phase feed, it is best to locate the temperature point close to the tower. For a two-phase feed, the best temperature measurement location is the last point where a single phase exists. If heating or cooling takes place downstream from that point, a temperature and pressure measurement in the two-phase flow is also recommended. It also allows a more accurate

Page 3 of 16



3.05

feed enthalpy measurement needed for a column energy balance. Any temperature measurement in the two-phase feed should be located close to a pressure measurement in the same line. Where temperature measurements are used for control, install sufficient thermowells to cover the entire probable range of points where the temperature could be most sensitive to column upsets. The most sensitive temperature point for a new column can be determined with a dynamic analysis of the column or by perturbations around a steady state design. The designer should take into account all possible modes of future operation. For an existing column, an analysis of temperature profiles for steady state operations plus upsets should show the most sensitive temperature point location. The most sensitive point is where rapid changes can occur and its measurement has a significant effect on column monitoring and control. Additional temperature sensing points should be installed above and below the calculated optimum point, since fractionating columns rarely operate exactly as designed. Consider installing some additional thermowells, suitably spaced, across column sections where none exist. This allows better comparison of the actual column temperature profile with the simulated profile for the same feed and separation. It is better to measure liquid instead of vapor temperatures. This is particularly important in services such as pumparounds in refinery fractionators where the vapor temperature differs from the liquid temperature. The best column measurement location is near the bottom of the downcomer and close to its widest point (Figure 5). If it is difficult to locate the measuring element in the downcomer, then locate it in the tray liquid. Locate temperature sensing nozzles out of the path of subcooled liquids or superheated vapors. Vapor temperatures at the top of the tower are often measured to get a true top temperature, e.g., one not affected by subcooled reflux. In direct heat transfer applications, such as desuperheating catalytic cracker effluent vapor or condensation in refinery pumparounds a vapor temperature measurement is usually needed. It is needed for evaluating heat transfer or for protecting downstream equipment from excessive temperatures. Where measuring vapor temperatures, it is important to shield adequately the thermocouples from any liquid. Temperature sensors respond slower to vapor temperature than liquid temperature changes. Often a thermocouple is put into overhead vapor lines to measure the temperature of the vapor leaving the column. Where the unit is inside a building and the vapor line is close to the ceiling, the thermowell tends to be put in the bottom of the line to improve accessibility. Subcooled condensate can collect in the bottom of the line resulting in an incorrect top column temperature measurement, even if the line is sloped. It is important not to put the thermowell into the bottom of the line and to provide sufficient shielding to prevent temperature loss ( Figure 6). H. Kister lists eight recommended guidelines for column temperature measurement connections on pages 134-136(49).

4

Level Measurements

Losing track of the liquid level is a prime source of column upsets and damage (1) and is often the cause of maloperation. Level instruments are needed in the bottom reservoir of the column, the reflux drum and in condensate drums. In packed towers additional level instruments in liquid collectors and on distributor trays are very helpful in troubleshooting a distillation problem. High level switches assist in avoiding submerging vapor inlets that could result in damaged trays or in lifting of the bottom packing support. Locate column bottom level taps in places that will avoid interference from high velocity streams. When the bottoms liquid is very temperature sensitive, there is a tendency to reduce the inventory of bottoms

Page 4 of 16



3.05

liquid to a minimum. At times one of the two sensing points is located in the pipe carrying liquid from the column to the reboiler pump ( Figure 7). If the flow rate is high, the liquids can produce a reduction of pressure at the lower liquid sensing point and thus an erroneous liquid level reading. Accurately measuring the liquid level in columns having large turnovers of liquid can be difficult unless a stilling chamber or bridle is used. Forced circulation reboilers operated at high rates can result in a high turnover of liquid. Figure 8 shows a configuration with a high liquid turnover with a stilling chamber and a bridle. (The waves shown in the column sump are the result of high liquid turnover.) These usually provide accurate level readings. A less preferred alternative is to use transmitter output damping instead of a stilling chamber or bridle. (The transmitter is then connected directly to the column level taps.) Its disadvantage is slower response to liquid level changes. If the stilling chamber or bridle is heat traced and insulated there can be a tendency to boil out liquid in summer. If the stilling chamber or bridle is not heat traced and covered, the liquid could freeze in winter. Turning off the heat tracing in summer and turning the heat tracing back on in late fall often will correct the problem. The upper and lower level taps should be at least 12 inches (300 mm) apart. In large diameter columns it is a good practice to make this distance larger to handle possible waves, turbulence and frothing above the liquid surface. Usually they are farther apart than 12 inches (300 mm) to provide adequate surge capacity. Surge capacity is generally dictated by process operation or control requirements. H. Kister lists eleven guidelines for level measurement connections on pages 125-130 (49).

5

Flow Measurements

Ideally the reflux and all incoming and outgoing stream flow rates should be measured so that a complete material and energy balance can be made. Heating/cooling medium flow measurements to the reboiler and condenser are necessary for the energy balance. Generally locate a flow measuring device where the liquid pressure is highest and the temperature lowest. This suppresses vapor bubble formation at the flow measuring device that could affect the flow reading. Usually this is between a pump discharge and a downstream flow control valve ( Figure 9). If a cooler is present in that line segment, such as a bottoms product cooler, locate the flow measuring device downstream of it. If orifice plates are used, there must be sufficient upstream straight pipe lengths to establish uniform velocity profiles. See Perry's Handbook Chapter 5 for where to locate the metering device in a line (88). Install gravity flow reflux line flow measurement devices in the low portion of the line so that they are always liquid full. Figure 10 shows the correct and incorrect methods of locating the flow instruments for a gravity flow system. Non-pumped or pressured column bottom product flow measurement also requires special attention. If a bottoms product cooler is present, locate the flow measurement device downstream of it. Again this is to avoid vapor bubble formation as the liquid passes through the flow measurement device. If there is no bottoms product cooler, locate the flow measurement device below the column low liquid level. Also minimize piping pressure drop from the column to the flow measuring device. Always check that vapor bubble formation will not occur in the flow measuring device when the column bottoms level is low.

Page 5 of 16


6


3.05

Alarms and Shutdowns

Alarms may be taken from high or low measurements of control instruments. Since malfunction of the control instrument is often the cause of the alarm, it is good practice to use a separate sensor for the alarm. Install a separate instrument when the measurements are critical. Critical instruments are instruments that must operate properly for a safe operation of the unit. Often duplicate critical instruments are installed with separate taps. Thus if one of the measuring devices malfunctions, the column would not require an unnecessary shutdown to repair it. Install gravity flow reflux line flow measurement devices in the low portion of the line so that they are always liquid full. Figure 10 shows the correct and incorrect methods of locating the flow instruments for a gravity flow system. Non-pumped or pressured column bottom product flow measurement also requires special attention. If a bottoms product cooler is present, locate the flow measurement device downstream of it. Again this is to avoid vapor bubble formation as the liquid passes through the flow measurement device. If there is no bottoms product cooler, locate the flow measurement device below the column low liquid level. Also minimize piping pressure drop from the column to the flow measuring device. Always check that vapor bubble formation will not occur in the flow measuring device when the column bottoms level is low.

References and Further Readings

49. Kister, H. Z.," Distillation Operations", McGraw-Hill Book Company, New York (1990). 88. Perry, R.H. & D.W. Green, eds, "Perry's Chemical Engineering Handbook", 6th Edition, McGraw-Hill Book Company, New York (1984).

Page 6 of 16



Figure 1. STEAM PRESSURE MEASUREMENT

Page 7 of 16

3.05



Figure 2. PRESSURE GAGE INSTALLATION

Page 8 of 16

3.05



3b. Incorrect Method for this Pressure

Figure 3. PRESSURE DIFFERENTIAL MEASUREMENT

Page 9 of 16

3.05



Figure 4. PRESSURE TAP LOCATIONS

Page 10 of 16

3.05



3.05

Generally locate the top column pressure tap in the top head of the distillation column. Where installation into the top of the head is very difficult, a side mounted tap should be considered. Sometimes the pressure tap is on the overhead line. When a pressure measuring element is on an overhead vapor line that is near a building ceiling, there can be a tendency to mount the element under the line. This makes it more readily accessible. However avoid installing pressure sensing elements such that they do not self drain into the process line. This prevents an altered pressure reading caused by liquid accumulation in the sensor line (Figure 2). Measure the pressure drop across groups of trays or a packed bed with a differential pressure sensor. A less preferred alternative is to infer the pressure drop by difference between two pressure measurements. Figure 3 shows pressure drop measurement across a column.) A differential pressure sensor is especially important when the measured pressure drop is less than ten percent of the absolute pressure. A differential pressure tap should not be located in the overhead vapor line. A tap in the overhead line does not only measure the pressure drop across groups of trays or a packed bed. It adds to that overhead head line entrance loss and pressure drop. Also there can be a velocity effect error caused by flow across the tap. Conventional pressure instruments are accurate to about plus or minus 0.5% of the calibrated span. Thus for a column operating at 100 psia (6.9 bar) with conventional pressure sensors calibrated for 150 psia (10.3 bar) at the top and the bottom of a section having a 5 psi (0.34 bar) pressure drop, the accuracy would be plus or minus 1.5 psi (0.10 bar). A conventional differential pressure instrument should have an accuracy of about 0.5% of the calibrated span. Thus a differential pressure sensor calibrated for 10 psi (0.69 bar) would have an accuracy in this application of 5 psi (0.34 bar) plus or minus 0.05 psi (0.003 bar). Smart transmitters are about three times as accurate as conventional transmitters. If this example used smart pressure transmitters, the accuracy would be plus or minus 0.5 psi (0.03 bar). If the example used a smart differential pressure transmitter, the accuracy would be plus or minus 0.02 psi (0.0014 bar). When measuring differential pressure across a column with a high density vapor (such as a depropanizer), be aware that unlike a vacuum column, the static vapor head is significant compared to the flow induced pressure drop. Consider biasing the transmitter output to eliminate the static vapor head reading. When measuring the pressure in a high vacuum system with a gage, use an absolute pressure (capsule) gage. Regular pressure gages balance the measured pressure against the atmospheric pressure that changes from day to day. An absolute pressure gage has an internal standard. When using a single differential pressure instrument (upper sketch in Figure 3) the transmitter to bottom connection leg must be sufficiently large to permit proper drainage and venting. Insulating the legs will help avoid excessive condensation. If the vapor condensing temperature is above ambient, condensation can be a problem even if well insulated. Heat tracing or gas purging of the legs may be necessary. Figure 4 illustrates problems with pressure tap locations. In Figure 4a, excess weld metal on the column inside deflects the flow pattern causing low pressure readings. The middle drawing illustrates an incorrect pressure reading caused by the disruption of flow on a bend in the piping. The bottom drawing shows a correct pressure measurement.

3

Temperature Measurements

Recommended temperature measurements are the column top and bottom, the reflux drum and in all feed lines to and sidedraw lines from the column. For a single-phase feed, it is best to locate the temperature point close to the tower. For a two-phase feed, the best temperature measurement location is the last point where a single phase exists. If heating or cooling takes place downstream from that point, a temperature and pressure measurement in the two-phase flow is also recommended. It also allows a more accurate

Page 3 of 16



Figure 5. THERMOWELL LOCATIONS FOR MEASURING TRAY LIQUID TEMPERATURE

Page 11 of 16

3.05



Figure 6. THERMOWELL LOCATIONS FOR MEASURING COLUMN

Page 12 of 16

3.05



Figure 7. COLUMN SUMP LEVEL TAP LOCATIONS

Page 13 of 16

3.05



Figure 8. USE OF A STILLING CHAMBER OR BRIDLE

Page 14 of 16

3.05



Figure 9. FLOW ELEMENT LOCATION

Page 15 of 16

3.05



Figure 10. MEASUREMENT OF GRAVITY REFLUX TO A DISTILLATION COLUMN

Page 16 of 16

3.05


DESIGN OF SMALL-SCALE COLUMNS

Issued:

10/10/2008

3.07

Revised:

DESIGN OF SMALL-SCALE COLUMNS

DESIGN OF SMALL-SCALE COLUMNS .................. ........................................................... 1 1.

Introduction ........................................................................................................................ 2

2.

Tower Layout ..................................................................................................................... 2

3.

4.

2.1

Bed Heights & Packing Selection ................................................. ............................ 2

2.2

Other Considerations ................................................ ................................................. 3

2.3

Summary ................................................... ................................................................. 3

Tower Internals Design & Selection ................................................ .................................. 4 3.1

Distributors ............................................... ................................................................. 4

3.2

Redistributors.................................................. .......................................................... .5

3.3

Packing Support Plates ................................................ .............................................. 6

3.4

Hold-Down Grids/Bed Limiters ................................................. ............................... 7

3.5

Chimney Trays .............................................. ................................................... .........7

3.6

Feed Inlets............................................... ................................................................... 8

Other Mechnaical Considerations............................................................ ........................... 9 4.1

Vessel Body Flanges.................................................. ................................................ 9

4.2

Hand Holes ......................................................... .......................................................9

4.3

Nozzles ................................................... ................................................................... 9

4.4

Vessel Suppliers............................................................... .......................................... 9

4.5

Tower Attachments.................................................................................................. 10

5.

Case Study ................................................ ........................................................................ 10

6.

References....................................................... ......................................................... ......... 10

Page 1 of 14


1

DESIGN OF SMALL -SCALE COLUMNS

3.07

Introduction Special design considerations must be given to small-scale columns with tower diameters of 36” (915mm) or less. Small-scale columns can be equipped with cartridge trays or packing. Sectional trays have also been provided in tower diameters as small as 29” (740mm). Refer to section 1.19 “Cartridge Trays” for more information. Depending on regional labor and material costs it may be more economical and convenient to pack small diameter columns than to utilize cartridge or sectional trays. In addition, tray efficiencies are likely to be lower in a small diameter column since the flow path length is too small to capitalize on crossflow enhancements. Other design considerations may favor trays such as fouling, two liquid phases, and multiple feed and drawoffs. Small scale packed towers can be installed as standalone pieces of equipment or as part of a skid unit. These small diameter towers may come pre-packed and fully assembled from the manufacturer. Prepacking the vessel is not recommended as this may lead to premature flooding. The packing can compress and internals may become dislodged during shipment. This applies for random packing, structured packing, and most especially ceramic packing. It is strongly recommended that the packing and internals be installed at the field. These issues and other design considerations related to small packed towers are discussed in this section.

2

Tower Layout 2.1

Bed Heights & Packing Selection As with large diameter columns packing selection is an important step in designing a column. The designer must pick the correct packing size and type suitable for the process flow conditions. Consideration must be given to the type of service and process constraints such as; high liquid rate, pressure drop limitations, foaming, fouling, high liquid viscosity, etc.. Corrosion and material of construction must not be forgotten. Fouling due to corrosion of the internals and vessel wall may lead to premature flooding. Ceramic and plastic packings have a smaller void fraction than their metal counterparts resulting in lower capacity for the same ring diameter. Flow data on random packing shows that radial distribution clearly increases as particle diameter increases, so the tendency to develop wall flow will go up as packing diameter increases. Excessive liquid along the column wall will cause poor packing performance. Large random packing rings of >3 inches (> 75mm) are particularly vulnerable to wall flow. Many designers recommend limiting the column diameter to packing size ratio to at least 10 for random rings (1,2). The packing size is defined as the largest dimension of the ring. Some organizations limit this ratio to as high as 15. Smaller than 10 may lead to the packing elements stacking-up along the vessel wall, resulting in liquid channeling (1). This effect is more pronounced with low liquid loads. If large rings are needed to debottleneck a given tower diameter, structured packing should be considered. Radial flow distribution data on structured packing shows that structured packing does not push the liquid outwards as rapidly as random rings. In addition, most structured packings have wiper bands or collars that help collect the liquid off the wall. Crimp height, crimp angle and surface texture have an effect on radial flow. For the moment, there is insufficient data to recommend a maximum packing size based on column diameter for structured packing. However, recent work with Montz B1-250 in air/water concluded that capacity decrease becomes significant when the column diameter approaches the height of a packing element (3). The minimum recommended tower diameter for sheet metal structured packing is 4” (100mm) (4) . The minimum tower diameter actually depends upon

Page 2 of 14



3.07

crimp sizes. Gauze type packings in diameters as low as 2” and sheet metal as low as 4” are in service with no operational problems reported (4) and a 250m2/m3 packing minimum size is 4” ID. It is advocated that small crimp sizes and shorter bed depths be utilized for structured packing. Packing performance can be jeopardized if the bed depth to tower diameter ratio is too large, especially with large size packings. Bed depths for towers in this diameter range should be limited to 20 feet (6100mm) maximum and a bed depth to column diameter ratio of no greater than 10 to 1. In addition, the number of theoretical stages in a single bed should be restricted to a maximum of 10-12. One manufacturer recommends maximum bed height to column diameter of 12 and a maximum of 20 theoretical stages (8). 2.2

Other Considerations Most small-scale towers up to and including 24 inches (600mm) in diameter are constructed from standard pipe sizes. Larger diameters or towers constructed with exotic materials may be rolled from plate. The designer must be aware of the pipe wall thickness and use the true inside diameter of the pipe when determining packing selection and hydraulics. Normal derating factors for foaming systems or uncertainty in the design correlations should be applied. Please refer to section 2.10 “De-Rating Factors – Packed Columns” for more information on this subject. Additional derating factors are not generally applied to small diameter columns. FRI and SRP data comparisons have shown that the capacity and efficiency of structured and random packing is comparable between the 4 feet (1200mm) and 16 5/8 inches (425mm) diameter columns. However, larger size random packings in small-scale column towers tend to have 5 to 10% more capacity than their larger cousins due to the wall flow mentioned above. Economic consideration must be given to the overall tower height versus tower diameter. The relative cost between an 18 inch (450mm) and a 20 inch (500mm) diameter pipe is relative faily small but the tower capacity advantage is close to 25%. Additionally, tower height may be more of a factor in the overall cost of the tower than diameter, due to the more complex support structure needed. There is very little economic incentive to design the column close to the flood point for small diameter columns. In general, it is more economical to design the column with a smaller packing size. This will result in a slightly larger column diameter, but significantly smaller tower height. When increasing the tower diameter, the designer needs to consider minimum wetting rates at turndown conditions. The physical location of the tower may restrict the final size and layout of the column. Small towers may have to fit inside an enclosed building or an existing skid unit. In the first case, the overall tower height is important, while in the second case the tower body flange diameter may be the controlling variable.

2.3

Summary It is recommended that small-scale towers be designed under the following guidelines: 1. The bed depth to tower diameter ratio should be less than 10:1, with a maximum bed depth of 20 feet (6100mm). 2. Limit the bed depth to less than 10-12 theoretical stages. 3. The column diameter to packing diameter ratio for random rings should be at least 10:1. 4. Additional derating factors for small diameter columns are not needed. Normal derating factors should be applied to small columns as with large columns. Page 3 of 14



3.07

5. Minimum tower diameter for structured packing applications is 4” (100mm). 6. Beware of overall vessel height to tower diameter ratio limitations imposed by company specifications or building enclosures. A good rule of thumb is a maximum allowable H/D ratio for the vessel of around 22 to 1.

3

Tower Internals Design & Selection 3.1

Distributors Although the effect of maldistribution is not as severe in small diameter towers as it is in large diameter towers, uniform liquid distribution is critical to obtain maximum packing efficiency. Data from Billet, Stilkkleman, Stoter, Zuiderweg, and Separation Research clearly show that the liquid distribution quality has an impact on HETP in small diameter vessels, especially at low liquid rates. The pan or bucket type is the most common distributor device used for small diameter towers, although troughs can be applied for towers larger than 30” (750mm) in diameter. Pan type distributors provide maximum open area while maintaining an even drip point pitch across the tower. In addition, they can be leveled. Pan type distributors can be fitted with orifices on the floor or with raised orifice drip tubes for fouling services. Drip tubes are also used for high turndown or low liquid rate services. Depending on the size of the column, it may be equipped with or without risers for vapor flow a gap between the vessel wall and distributor is usually provided. Because of this annular space, flow irrigation around the perimeter of the vessel may not be very good. Care must be taken that the orifices are placed as close as possible to the pan wall in a uniform manner. Risers that serve as both vapor and liquid passageways should not be used, as this can lead to liquid entrainment and carryover. Another approach is to use orifice plate distributors that are placed in between body flanges of the vessel. These plates are thicker than normal orifice plates to allow machining on both sides to seat the gaskets between the flanges. Pan distributors are usually constructed in one or two pieces, depending on the size of vessel and accessibility. The distributors can be supported on a continuous support ring or clips. Clips are preferred over a continuous support ring, since it obstruct less tower cross-sectional area. Some designs do not allow the distributor to be affixed to the vessel. It simply rests on top of the packing. Jackscrews are provided on the side of the distributor to keep it centered and provide additional frictional resistance to uplift. This type of support mechanism is not recommended unless the user is sure that the vessel will not be susceptible to transients or upset conditions, which may dislodge the distributor. It is not very expensive or time consuming to provide bolting assemblies to permanently affix the distributor. As with all distributors, considerations must be given to levelness. For assemblies in which the distributor is affixed to a bottom support, washers can be used at each bolt or leg to level the distributor. However, this becomes cumbersome during installation, since the distributor must be removed each time a washer is needed for leveling. Many times installers simply ignore this process. The preferred method is to have the distributor hung from clips above the distributor. A leveling adjustment mechanism can be provided allowing the distributor to be easily leveled from the topside. For structured or gauze type packed tower distributors, support clips or rings should not be installed, since the packing often comes in one piece and may snag on the tower attachments. For

Page 4 of 14



3.07

these cases, a ring is installed between two tower body flanges from which the distributor is suspended. See Figure 1. Tall small-scale columns are more susceptible to swaying than towers with large diameters. Sufficient freeboard must be given between the maximum liquid level and the top of the riser and pan wall to avoid spilling over the wall and risers. A minimum clearance of 4 to 6 inches (100 to 150mm) between the maximum liquid level on the distributor and the edge of the pan is recommended. In addition, a minimum of 3 inch (75mm) liquid head at turndown is needed to minimize the swaying Figure 1 – MTS 109 Distributor with integral bed limiter below and effects on liquid distribution wiper ring above. Courtesy of Sulzer Chemtech quality. A minimum distance of 4 to 6 inches (100 to 150mm) between the distributor and the packing is recommended for vapor/liquid disengagement. See Figure 8. For random and structured packings, a minimum drip point density of 6 to 9 points per square foot (65 to 100 pts/m 2) is recommended, although drip point density as low as 4 points per square foot (40 pts/m 2) has been used successfully. For gauze type packing, a minimum of 9-10 points per square foot (100 to 110 pts/m 2) is recommended. However, if the tower is very small, the point density may have to be increased. For example, using the 4-pts/ft 2 (40 pts/m 2) density guideline in a 6” (150mm) column will result in a single distribution point. Obviously, this is not recommended. For such small towers, a point density of 15 or 20 pts/ft 2 (160 to 215 pts/m 2) should be applied. In tower diameters below 4” (100mm), a distributor may not be required. A single point distributor is all that may be needed. When increasing the point density, smaller diameter orifices will result, increasing the risk of plugged orifices. Depending on the service, a minimum orifice diameter may be required. For fouling services, a drip tube with raised orifices or notches is more appropriate as illustrated in Figure 1. In addition, dual inlet filters feeding the distributor is strongly recommended with a mesh size of about ¼ of the orifice diameter. Please refer to Section 2.02.2 “Liquid Distributors” for more general distributor guidelines and Topical Reports 118 & 122 for orifice sizing equations. 3.2

Redistributors Wall flow can become significant in small diameter columns, which can reduce the efficiency of the packing. The degree of wall flow depends on the packing diameter, type and irrigation rate. Wall wipers alone will not redistribute and re-mix the liquid properly and thus, are not recommended. In addition, a large risk exists that wall wipers will actually reduce tower capacity.

Page 5 of 14



3.07

Section 2.02.3 “Redistributors”, lists a number of rules-of-thumb guidelines to define the maximum bed height. Most of these guidelines are also applicable for small diameter towers, especially the ratio of bed height to column diameter, which takes into account packing size. For this diameter range, the packed bed depth should be limited to 10 times the tower diameter to a maximum height of 20 feet (6100mm) or when the number of expected theoretical stages is 10 to 12 or greater. Orifice pan and plate redistributors can easily be converted to a Figure 2 – Norpro Intallox® Pan Redistributor Model 107 Courtesy of Koch-Glitsch redistributor by simply adding hats to the risers. See Figure 2. With pan distributors, a wall wiper is required around the periphery of the column to redirect any liquid along the tower wall into the redistributor. If the redistributor is hung from a support ring, the support ring can act as a wall wiper. 3.3

Packing Support Plates As with large towers, there are two main types of packing support plates: gas injection and grid bar supports. The gas injection type is used to support packing support plates: gas injection and grid bar supports. The gas injection type is used to support random packing and the grid bar style is used to support structured, gauze or grid packing. The main function of these devices is to support the packing and not to distribute the vapor into the packed bed. Spacing of grids, openings in gas injection plates, and clearance around the periphery of the wall must be sized so that the packing does not fall through. This is a common pitfall when saddle type rings are being used. Since the support span in small towers is Figure 3 – Gas Injection Support Plate EMS 403. Courtesy of Sulzer Chemtech short, a large beam depth is not necessary. For small diameter columns, the gas injection support plate is typically around 4 to 6 inches (100 to 150mm) tall. Similarly, grid supports are typically 2 inches (50mm) high, depending on bed depth. The mechanical integrity of the support plate must take into account the operating weight of the bed, the appropriate corrosion allowance for the service and any future fouling weight. Support plates for small columns are typically constructed in one piece for tower diameters of Page 6 of 14



3.07

12” (300mm) or less and two or three pieces for larger towers, depending on tower accessibility. If tower body flanges are available, most of the supports will be provided in one-piece assembly. The need to secure the support plate to the vessel attachment will depend on how susceptible the vessel is to upset conditions. Some member companies prefer to secure the support plate to the vessel attachment with tray-clips, regardless of service. Access to secure these tray-clips must be considered in the layout of the internals. Thus, support plates should be located near the vessel body flange or a vessel hand hole. Support plates can be affixed to a continuous support ring or clips that are attached to the shell wall or they can be held in place between the vessel body flanges. Continuous rings are not Figure 4 – Gas Injection Support Plate Model 818. Courtesy recommended for small diameter towers as of Koch-Glitsch they can obstruct a large percentage of the vessel cross sectional area. These devices are typically assembled from the bottom side, since it is almost impossible to reach down from the top opening above the bed. For very small tower diameters of 12” or less (300mm), the support plate may limit tower capacity due to the reduced free area available (8). This depends on the style of the support plate. The manufacturer should be consulted to confirm the free area. 3.4

Hold-Down Grids/Bed Limiters For many small towers, bed limiters can be integrated with the liquid distributor to prevent random packing from lifting through the riser or annular space per Figure 1. The dual purpose is very convenient and decreases installation time. Other designs will have the bed limiter sitting on top of the packing. They may have jackscrews that friction fit against the tower wall or they may have an anti-lift device, which impinges against the distributor above. All of these are acceptable configurations for services in which no uplift or major upsets are expected. For services in which transients are known to occur, the bed limiter Figure 5 – Bed limiter fixed to the vessel wall should be attached to the vessel. Again, continuous via small clips rings are not recommended for small towers. Bed limiters can be fixed to the vessel wall with clips as shown in Figure 5.

3.5

Chimney Trays When leakage is a concern such as a reboiler drawoff, total product draw, etc., chimney trays are often seal welded to eliminate leaks. These trays are typically installed at the vessel shop during vessel fabrication. Thus, the tray must be inspected and water tested at the shop prior to the installation of the next vessel section as field inspection is limited. Vessel break flanges,

Page 7 of 14



3.07

manholes, or handholes are required near the chimney tray for periodic inspection. Collector trays that are normally not seal welded, but simply gasketed in large towers, such as above a redistributor, flashing feeds, tray transitions, etc., are simply not necessary in small towers as liquid gradients and liquid re-mixing is not a concern. Pan re-distributors are very suitable to collect the liquid directly from a bed above as shown in Figures 2 & 8. If the system is deemed to be a foamer, it may be prudent to install a collector directly below a feed to provide additional residence time as shown in Figure 8. In Case Study #3 “How To Design an FRI Small Scale Column”, 4.50.03, a seal welded chimney tray is utilized as a total drawoff to feed a kettle reboiler. The chimney tray is this case is located just below a body flange, providing ample space for inspection and a water leak test. 3.6

Feed Inlets Another significant difference between small and large diameter packed columns is the feed inlet arrangements. To permit the installation of single piece devices such as distributors, support plates, etc., through the vessel, feed nozzles should not protrude inside the vessel. Feeds mustbe introduced with flush nozzles or via “stab-in” piping as shown in the attached Figures 11A, B & D. Liquid feeds can be designed so that the liquid is directly discharged into the center of the distributor as shown in Figure 11C. With a center feed discharge, the distributor is usually equipped with a center inlet box to break down liquid momentum. A more elaborate feed arrangement as shown in Figure 6 can also be used. Notice the mist eliminator in the background. Liquid feed inlets must be designed to minimize Figure 6 – Feed pipe with side slots turbulence, while at the same time not limiting access to the tower. A number of design inlets are possible and some are shown in the attached Figure 10. For additional information on feed inlet configurations and distributor design, please refer to section 2.02.2 “Internal Pipework to Packed Tower Distributors”, Section 2.02.1 “Liquid Distributors”, and section 2.02.4 “Liquid Flow Through Gravity”. As with all packed towers, an exterior strainer prior to the liquid feed inlet near the vessel flange is recommended to prevent blockage of the distributor orifices. For flashing feeds, it is recommended that the feed be introduced above a chimney tray with a “V”-baffle deflector above a chimney tray to spread the fluid radially around the vessel wall. Other devices such as feed chambers in Figure 7, which separate the two phases effectively, are available from suppliers. For kettle reboiler vapor returns, a simple flush nozzle is all that is needed, if sufficient space between the inlet nozzle and the packing above is available. For two phase reboiler returns, an inlet baffle should be provided to prevent the liquid from being blown into the Page 8 of 14

Figure 7 – Norpro’s Flashing Feed Chamber. Courtesy of Koch-Glitsch



3.07

equipment above. For gas feeds under a structured packing bed, a gas distributor sparger is recommended to initiate good vapor distribution into the bed. The gas distributor holes should be oriented downward, but far enough from the bottom liquid level to avoid entrainment. A distance of 12 to 18 inches (300 to 450 mm) is recommended.

4

Other Mechnical Considerations The attached Figures 8 through 10 are typical packed tower layouts accommodating a variety of internals. Access and installation of the internals are the main factors when determining the number of body flanges required and location. 4.1

Vessel Body Flanges Vessel body flanges are the most common method of entering a small diameter vessel. Body flanges are positioned so that they can easily provide access to a number of internals through a single flange. The use of body flanges allows easy dismantling of various vessel sections to gain access for maintenance, cleaning and removal of packing. Body flanges are expensive, increase vessel height and provide another source for process leakage. In addition, heat losses are greater in smaller diameter columns due to poor insulation around the body flanges.

4.2

Hand Holes Sometimes hand holes are included as means of providing access to internals during installation. Hand holes can reduce the number of body flanges required if properly positioned, but they provide less access. On occasion, extra hand holes are provided at the bottom of a packed bed section to allow random packing removal without the need to dismantle the bottom body flange. The use of a vessel hand hole for removal of packing should only be considered in clean services with random packings of 1 inch (25mm) or less. Large diameter packings tend to have a difficult time flowing out of a small diameter hole. Hand holes are typically 8 to 10 inches (200 to 250mm) in diameter, not exceeding half of the vessel diameter. A retaining plate can be added to the hand hole flange, as per Figure 8, to keep the packing in place and to minimize liquid & vapor channeling.

4.3

Nozzles For small diameter vessels, most nozzles are flush with the vessel wall to allow easy installation of internals without interference. This is especially true with single piece internals and structured packing. For nozzles that require internal piping to feed a liquid distributor, “stab-in” in piping with a breakaway piping flange can be used. Such nozzles can also be used for access and inspection See detail Figure 11A & 11D.

4.4

Vessel Suppliers Small diameter vessels are more inclined to have a large height to vessel diameter ratio (H/D) which makes them more susceptible to wind motion. For such vessels, flared vessel skirts and guy wires are needed to reduce the moment arm at the base of the skirt and to keep the vessel from swaying. Another technique is to increase the bottom section diameter for liquid hold-up, making the tower smaller in height and easier to support. See Figure 10. Some operating organizations have design limits on the overall vessel height to diameter ratio. This maximum ratio is typically 22 to 1. In low-pressure services, where the tower thickness is not controlled by pressure, it might be more economical to make the tower diameter larger to decrease the vessel

Page 9 of 14



3.07

height and support structure. 4.5

Tower Attachments Clips are preferred over continuous support rings, since they obstruct less tower cross sectional area. For example, in an 18 inch (450mm) column diameter, a 1 inch ( 25mm) wide continuous support ring will obstruct 20% of the cross sectional area of the vessel, i.e. whereas three (3) 1” x 3” (25mm x 75mm) support clips will only block 3.5% of the vessel area. If a continuous ring is to be used, the pan distributor should be equipped with raised legs to provide an open area that is at least equal to the gap between the distributor and vessel wall.

5

Case Study The designer is encouraged to review “Case Study #3 – How to design an FRI Small Scale Column” in section 4.50.03. This case is an actual example of a small diameter deisopentanizer tower using all the available FRI tools and practices.

6

References 1. Billet, R., Distillation Engineering, Chemical Publishing Co., 1979. 2. Kister, Henry S., Distillation Design, McGraw-Hill, 1992. 3. Olujic, Z., Effect of Column Diameter on Pressure Drop of a Corrugated Sheet Structured Packing, TransIChemE, Vol.77 Part A, 2000 4. Pilling, Mark, AIChE Spring 02 presentation 5. Stikkelman, R.M., Gas & Liquid Maldistribution in Packed Columns, PhD Thesis, TU Delft, 1989. 6. Stoter, Frank, Modeling Maldistribution In Structured Packings: From Detail to Column Design, PhD Thesis, TU Delft, 1993. 7. Zuiderweg F.J., Heok P.J., The Effects of Small Scale Liquid Distribution on the Separation Efficiency of Random Packings, I.Chem.E. Symposium Series No. 104, 1987. 8. Norpro’s Small-Tower Design Bulletin, 1990. 9. Norpro’s packed Tower Internals Guide, 2001

Page 10 of 14



Page 11 of 14

3.07



Page 12 of 14

3.07



Page 13 of 14

3.07



Page 14 of 14

3.07


DIMENSIONAL TOLERANCES

Issued:

6/15/2003

Revised:

3.08


DIMENSIONAL TOLERANCES ............................................................................................ 1 1

Introduction ......................................................................................................................... 2

2.

Trays ................................................................................................................................... 2

3.

2.1

Tray Level Tolerance................................................... ............................................. 2

2.2

Tray Deflections ....................................................................................................... 3

2.3

Outlet Weirs............................................................................................................. . 3

2.4

Inlet Weirs ................................................... .................................................. ........... 4

2.5

Downcomer Clearance ............................................................................................. 4

2.6

Downcomer Widths ................................................... ................................................ 5

2.7

Inlet Sump and Seal Pan Widths ................................................. .............................. 5

2.8

Hole Diameter and Area ............................................ ................................................ 5

2.9

Tray Outside Diameter .................................................. ............................................ 6

Packed Towers .................................................................................................................... 6 3.1

Distributor................................................. ................................................................. 6

3.2

Distributor Out of Level .................................................. .......................................... 7

3.3 Distributor Deflection ................................................. ................................................ 7 3.4 Distributor O.D. Tolerance .................................................. ....................................... 7 3.5 Spray Distributors .................................................. ..................................................... 7 3.6 All Other Packing Internals (Support Plate, Bed Limiter, Etc.) ................................. 8 3.7 Pressure Vessel Attachments ................................................ ...................................... 8 3.8 Support Ring Spacing ................................................... .............................................. 8 3.9 Support Ring Levelness ................................................. ............................................. 8 3.10 Support Ring Deviation From Horizontal ........................................................... ....... 8 4.

Appendix A – Original Survey ................................................... ........................................ 9

Page 1 of 19

Issued: 6/15/2003


Revised:

1

3.08

Introduction

Dimensional tolerances can have a very large impact on the installation and manufacturing costs of tower internals. They may also have an equal impact on the performance of the column. The designer must be judicious when applying tight tolerances to various tower internal dimensions. When specifying each tolerance, the designer should take into account its contribution in relationship to the overall performance or mechanical integrity of the internals. In summary, the value of each tolerance should be as practical and as large as possible without compromising the performance of the internal. The most common method of specifying dimensional tolerances is "plus or minus" ( }) a certain value. This expression can be interpreted a number of different ways among installers, inspectors, manufacturers and engineers. It is imperative that this definition is made clear and understood by everyone. For this section, if a downcomer clearance is specified as 2" }1/8", the downcomer clearance can vary between 1 7/8" and 2 1/8", or a total 1/4" deviation from high to low. In 1998, a dimensional tolerance survey was issued to the membership. This section documents the analysis of the survey. The overall response was poor with only 19 out of the 87 members replying. However, this is still a reasonable sample size of the membership population from which applied dimensional tolerances can be extracted. Figure 1 gives a breakdown of the responses by industry type. Out of the eight (8) Engineering survey respondents, only two (2) were from a fabricator.

Response to Tolerance Survey 4

8

Engineering Petroleum Chemical 7

21% Depend Entirely on fabricator's tolerances.

Figure 1

Four out of the nineteen members depend entirely on manufacturer tolerances and had very little input. This represents 21% of the returned surveys, which is probably a true representation of the membership. Two (2) were from the Petroleum Industry and two (2) from Engineering.

2

Trays 2.1

Tray Level Tolerance

Page 2 of 19



3.08

Careful consideration should be taken with columns in which performance can be greatly affected by unlevel trays. The results of this survey are very similar to a smaller survey on tray levelness in early 1980's. In section 1.12 "Tray Levelness", the high and low point allowances of installed trays are represented with maximum and minimum linear equations based on a survey of eight member companies. The new survey shows a wider gap between the largest and smallest allowable tolerances being applied by the members. The criteria given varied a great deal, with many of them dependent on a step change in the tower diameter. One member expressed the criteria in equation form, with the tower diameter as the variable. Two (2) members had a single tolerance value regardless of tower diameter. Figure 2 below is the tolerance data from the respondents. The tolerance given applied to all tray types, i.e. sieve, valve, dualflow, etc.

Tray Level Tolerance 14

m m + , e c n a r e l o T

12 10 8 6 4 2 0 0

1000

2000

3000

4000

5000

6000

7000

8000

Tower Diameter, mm

Figure 2

2.2

Tray Deflections

The tolerances given for this item also varied a great deal between members. However, most of the tolerances specified are dependent on tower diameter. Out of the 16 members that provided feedback on this item, seven (7) expressed tray deflection as the tower diameter divided by a constant for part or all of the their specification. These constants ranged from 1000 to 667, with the median of 900 (Figure 3). These constants apply to units of length. One member expressed the tolerance for deflection being dependent on straight span edge of a tray section. The tolerances applied to both the active and inlet panels of a tray.

Most Common Tray Deflection Criteria

Deflection = θ T /C Where θT = Tower Diameter C = constant, ranged from 1000-667, with a median at 900 Figure 3

2.3

Outlet Weirs

The survey asked for both a level and a height tolerance. Sixty percent (60%) of the members specified a tolerance for both items. It is very interesting to note that most Petroleum companies specified both a level and a height tolerance, while most of the Chemical companies specified only Page 3 of 19

Issued: 6/15/2003


Revised:

3.08

a level tolerance and most of the engineering companies specified only a height tolerance. With regard to adjustable outlet weirs, only 35% of the members use adjustable weirs most of the time (Figure 4). Of those that provided a height tolerance, all of them were within a range of }1.5 to 3mm. The level tolerance was within 1.5 to 3mm from high to low for 76% of those that provided a level tolerance. One member's specification for outlet weir height tolerance was divided into three ranges, depending on tower diameter.

Do You Specify an Adjustable Outlet 29% 6%

Always Most Seldom Never 29% 36%

Height Tolerance: 100% in the range of ? 1.5 to 3 mm Level Tolerance: 75% in the range of 1.5 to 3 mm from high to low Figure 4 2.4

Inlet Weirs

Four out of the sixteen members did not specify criteria for this item. The tolerance given for the inlet weirs is similar to that for the outlet weir and can be summarized as follows: 1. Tolerance on Height: All of the respondents specified heights in the range of ±1.5mm to 3.0mm. 2. Tolerance on Levelness: Most were in the range of 1.5mm to 3mm measured from the highest to the lowest point. One specified 4mm and another 6.4mm. 2.5

Downcomer Clearance

S The survey asked for both a level and a clearance tolerance. Only 29% of the members specified a tolerance for both items, two from each of the Petroleum and Engineering industry and one from Chemical. With regards to adjustable downcomer clearances, only 19% of the members use adjustable downcomer clearance most of the time (Figure 5). Of those that provided a clearance tolerance, 91% were in the range of ± 3mm. Of those that provided guidelines for levelness, 71% were 3mm from high to low.

Page 4 of 19

Issued: 6/15/2003


Revised:

3.08

Do You Specify an Adjustable Downcomer Clearance? 0% 19% Always 31% 50%

Most Seldom Never

Only 29% specify both a leve and a clearance tolerance. Clearance Tolerance: 91% answered ± 3 mm. Level Tolerance: 71% answered 3 mm from high t o low. Figure 5 2.6

Downcomer Widths

Half of the responding members have no criteria or depend on manufacturer tolerances for downcomer widths. The other half provided a variety of tolerances with one respondent providing a very detailed set of criteria, which had consideration for number of passes. One member provided criteria of ±4% of the downcomer width, others provided tolerances of ±3mm, ±6mm, ±9mm between wall and downcomer bar for side downcomer or between downcomer bars for the center and off center downcomers. One member provided tolerances of +9mm -3mm between vessel centerline and downcomer support bar. The tightest tolerance was +4 -2mm for all off-center and center downcomers measured between bars and 0.25% of tower I.D. for side downcomers, with a minimum of ±3mm. 2.7

Inlet Sump and Seal Pan Widths

Sixty seven (67%) rely on manufacturer tolerances. The others had the following results: 1. Both manufacturers tolerance: ± 3mm. 2. One engineering company: ± 9mm. 3. Two petroleum: +5,-3mm and ±3mm. 2.8

Hole Diameter and Area

Half (50%) of the members rely on manufacturer tolerances. The other half gave varying responses and the following summarize the results: 1. Hole Area and valve/cap count: • (2 members) ±1% of area varied and the following summarize the results: • (4 members) ± 2% of area • (1 member) ± 3% of area (manufacturer) and (1 member) +5%, -2% on area. 2. Hole Diameter • (1 member) ±0.2 to +0.4, -0.2mm. Page 5 of 19

Issued: 6/15/2003


Revised:

• • • 2.9

3.08

(3 members) ± 0.1mm. (1 member) ± 0.076mm. (1 member) ± 1% of hole diameter.

Tray Outside Diameter

Thirty-eight percent (38%) of the members depend on manufacturer tolerances. Others provided guidelines as follows: • (1 member) +0, -4mm • (1 member) +0,-6mm for tower diameters <2100mm and +0,-12mm for diameters > 2100mm. • (2 members) ± 3mm, (1 member) ± 4mm, and (1 member) ± 6mm. • (1 member) ± 3mm for tower diameters less than 2000mm and ± 5mm for diameters larger than 2000. • (1 member) ± 3mm for tower diameters less than 3500mm and ± 5mm for diameters up to 9000mm and ±7mm for towers larger than 9000mm.

3

Packed Towers 3.1

Distributor

About half of the returned surveys acknowledged that they “normally” water test gravity distributors and another 17% “sometimes” test distributors. The main factors for NOT testing liquid distributors were evenly split between “the type of service” and “liquid distributor type” regardless of the industry sector. However, most Petroleum and Chemical respondents do normally require liquid distributors to be tested, while only one engineering member responded similarly. Of the members that either sometimes or normally test distributors, (61%) specify their own performance criteria while the rest depend on vendors. See Figure 6.

Do You Water Test Liquid Distributors? 11%

Specify Criteria

17% 6%

55% 17% 33% Normally

Sometimes

seldom

never

61%

Breakdown By Industry f s t o n r e 6 e d b n 4 m o u p s 2 N e r 0

Specify Rely on Vendor Rely on Contractor Normally

1

2

3

Industry

Sometimes seldom never

Page 6 of 19

Figure 6

Issued: 6/15/2003


Revised:

3.2

3.08

Distributor Out of Level

The tolerance given for out of level was indifferent to distributor type. Petroleum and Chemical organizations had a tighter tolerance than Engineering companies. Sixty seven percent (67%) of all Petroleum and Chemical companies had an out of level tolerance of 3mm or less from high to low. One member had the same tolerance as for trays, which resulted in very large tolerances. See attached graphical representation in Figure 7.

Packing Distributor Out of Level Tolerance 14%

3mm or Less 6.4mm

36%

>6.4mm 50%

Tolerance is from high to low point and is the same regardless of distributor type

Figure 7 3.3

Distributor Deflection

Out of the 14 members that furnished detail tolerance data, only 8 provided specific tolerance criteria for deflection. The deflection tolerance given is as follows: 1. Four members (4) at 3mm and two (2) at 1mm from high to low. 2. In addition, three (3) engineering firms and one (1) petroleum organization use tray tolerances to cover distributors.

3.4

Distributor O.D. Tolerance

Very few members specified tolerances for this item. Three (3) at ±3mm and (1) at ± 5mm. 3.5

Spray Distributors

Only 38% of respondents provided tolerance criteria for this distributor type. Comments were made that due to the very nature of this distributor, it does not warrant any special tolerance. Standard manufacturer tolerance is sufficient. Of course, the question naturally arises as to what are standard manufacturer tolerances? Other interesting facts and data are given below: 1. Three (3) of four (4) chemical members don’t use spray distri butors. 2. Tolerances given were: • Out of Level – (2 members) 3mm and (3) 6.4mm from high to low. • Elevation Tolerance – (2 members) ±3mm, (1) ±5mm, (1) ±6.4mm (1) ± 10mm, (1) Page 7 of 19



• 3.6

3.08

±12.7mm. Pitch Tolerance – (2 members) ±3mm, (1) ±1mm and (1) ±6mm.

All Other Packing Internals (Support Plate, Bed Limiter, Etc.)

1. About half (5 of 9) of Petroleum & Chemical producers rely on manufacturer tolerances or have no specific tolerances. 2. The other four (4) provided tolerances of: 2, 3 (2 members) & 5mm from high to low for both deflection and out of level. 3. One engineering firm provided a tolerance of 3mm. Two other engineering companies use the same tolerance as for trays. 4. One manufacturer provided an out of level tolerance of the tower diameter divided by 500.

3.7

Pressure Vessel Attachments

This section of the survey was the least completed by the members. Six members out of the nineteen that responded to this question, or 32%, completely rely on fabricators standards for vessel attachments tolerances. Largest feedback was given to the support ring tolerance, which indicated the highest interest to the membership. Some rely solely on tray levelness criteria while others were more specific. The response was similar across all industries with two exceptions. One manufacturer provided a very tight tolerance for support ring spacing and levelness, while an engineering firm provided a very large tolerance and it seems outside of the norm. Another interesting aspect of this section was that all of the members indicated that the support ring tolerance also applied to dualflow trays.

3.8

Support Ring Spacing

Out of the 13 responses received, 11 members specified a tolerance of ±3mm. specified a tolerance of ±1.5mm and another at ±1mm.

3.9

One member

Support Ring Levelness

A more varied response was received for this specification. In some cases, this tolerance was a function of tower diameter and in others; it was dependent on linear length of ring. Two members just gave a range of 3 to 10mm and 3 to 6mm, depending on tower diameter. The following is a summary of the more specific cases: 1. 2. 3. 4. 5.

(1 member) - 2500mm and less, 3mm, > 2500 4.5mm. (1 member) - 3mm for every 3000mm of tower diameter. (2 member) - 3mm for all diameters. (1 member) - 0.3% of tower I.D. with a maximum of 6mm. 1 member) - less than 1200mm, 9.5mm, 1200 to 2500mm, 12.5mm, 2500 to 4500mm, 16mm, and > 4500mm, 19mm. 6. (1 member) - 1.5mm per every 300mm in support ring length, non cumulative.

Note: The above tolerance is defined from high to low points. 3.10

Support Ring Deviation From Horizontal

Page 8 of 19



3.08

Only five members had a criterion for this specification. The following is a summary: 1. 2. 3. 4.

4

(2 members) - ±6mm. (1 member) - ±3mm. (1 member) - ±1.5mm. (1 member) - 1 degree from normal.

Appendix A – Original Survey

The FRI Design Practices Committee is seeking your assistance in our attempt to determine industry’s best practice on dimensional tolerances and its effects on performance, installation and costs. The FRI Design Practices Committee appreciates your time and effort in completing this survey. There is a great deal of debate of just how important or critical tight tolerances are to the performance of various fractionating devices and its impact on equipment and installation costs. Your input is most valuable in establishing general FRI dimensional tolerance guidelines for trays, distributors and other internals. For your convenience the survey is broken down into three parts: A - Conventional Trays B - Packed Towers C - All Towers We ask that the entire survey be returned by Feb 1, 1998, however, each individual part can be completed and submitted piecemeal at any time prior to Feb 1st. Results will be presented at the next May spring meeting. Please return completed survey to: Fractionation Research, Inc. Design Practice Committee 424 S. Squires St., Ste. 200 Stillwater, OK 74074 In replying to the survey, please place a numerical value above the short heavy lines ”_________”. If you currently do not specify a tolerance for any particular dimension, please insert “none”. Provisions exist for both English and SI units for your convenience. Please feel free to write any comments in the space provided or anywhere in the page or insert additional pages, if needed. Any comments to the value of any tolerance or concern should this tolerance not be met would be very valuable to the membership. Feel free to leave questions unanswered if unable or unwilling to reply to a certain question or part of a question. A partial response is better than none at all. However, please recognize that the more thorough your response is, the better the guidelines will be. FRI Design Practices Committee

Page 9 of 19



3.08

FRI Dimensional Tolerance Survey PART A - CONVENTIONAL TRAYS A. General Information

1. Your Organization’s Primary Business Area. Check “ ”only one. Chemicals Engineering Petroleum 2. Do you rely on fabricators’ standard tolerances? Check “ ”only one. Yes Sometimes Seldom

Never

If “Sometimes”, please describe circumstances:

B. Conventional Valve, Sieve, Bubble Cap, & Dualflow Trays (Proprietary Trays are not included)

1. Out of Levelness Out of levelness is usually related to vessel diameter and may be dependent on the type of tray and service. This tolerance is usual expressed as ± a vertical value or from a maximum to a minimum allowable vertical dimension. For Sieve Trays: For Valve Trays: For Bubble Cap Trays: For Dualflow Trays: Comments or concerns:

2. Tray Deflections a.

Maximum deflection for an active panel. These tolerances may vary with tower diameter and may also vary depending on criticality of service. Note that this is a physical measurement and not a design parameter. For Sieve Trays: For Valve Trays: For Bubble Cap Trays: For Dualflow Trays:

b. Maximum deflection for a tray inlet area. For Sieve Trays: For Valve Trays: For Bubble Cap Trays: Comments or concerns:

Page 10 of 19



3.08

3. Outlet Weirs a.

Height tolerance from tray floor from active panel on the center of the weir, ± ____________inches or ± ____________mm. b. Out of level from high to low point, ____________ inches or ____________ mm. c.

Do you use adjustable weirs? Always Most of the time

Seldom

Never

Comments or concerns:

Inlet Weirs d. Height tolerance from tray floor, ± _____________

inches or ± ____________mm.

e.

Out of level from high to low point, ____________

inches or _____________mm.

f.

Distance between bottom of downcomer panel and weir. ± ___________ inches or ± _____________mm.


4. Downcomer Clearance a.

Distance between bottom of downcomer panel and tray floor. inches or ± _____________mm. ± ___________

b. Out of level from high to low point, ____________ c.

Do you use adjustable bottom downcomer panels? Always Most of the time

inches or ______________mm.

Seldom

Never


5. Downcomer Width The location of the downcomer support bar will control the actual width and area of the downcomer. The designer and installer usually locate these bars by one of the two following dimensions: a.

Side Downcomers Distance between vessel wall and downcomer support bar. ± ___________inches or ± _____________mm or Distance between vessel centerline and downcomer support bar. ± ___________inches or ± _____________mm.

b. Off-center and Center downcomers Distance between downcomer support bars. ± ___________inches or ± _____________mm or Page 11 of 19



3.08

Distance between vessel centerline and downcomer support bar. ± ___________inches or ± _____________mm. Comments or concerns: 6. Downcomer Inlet Sumps and Seal Pan Widths a.

Under Side Downcomers Distance between vessel wall and support angle. ± ___________inches or ± _____________mm or Distance between vessel centerline and support angle. ± ___________inches or ± _____________mm or Distance between bottom of downcomer support bar and sump/seal pan support angle. ± ___________inches or ± _____________mm.

b. Off-center and Center downcomers Distance between sump/seal pan support angles. ± ___________inches or ± _____________mm or Distance between vessel centerline and downcomer support bar. ± ___________inches or ± _____________mm or Distance between bottom of downcomer support bar and sump/seal pan support angle. ± ___________inches or ± _____________mm. Comments or concerns:

7. Hole Diameter & Open Area a.

Sieve hole area tolerance: ± _______________% of total no. of holes or area specified.

b. Sieve hole diameter:

± ___________inches or ± _____________mm.

c.

± _______________% of total number specified.

Valve or Bubble Cap number:


8. Tray Outside Diameter The outside diameter of any internal is a critical installation parameter. This dimension must take into account possible vessel out of roundness and must sit on a support ring with an appropriate overlap. Please check one or both of the tolerances below that is applied: Tray overlap on support ring:

± ___________inches or ± _____________mm.

Tray outside diameter:

± ___________inches or ± _____________mm.

Page 12 of 19



3.08

FRI Dimensional Tolerance Survey PART B - PACKED TOWERS A. General Information

1. Your Organization’s Primary Business Area. Check “ Chemicals Engineering

”only one. Petroleum

2. Do you rely on fabricators’ standard tolerances. Check “ ”only one. Yes Sometimes Seldom

Never


B. Packed Towers

1. Gravity Distributors, Redistributors & Parting Boxes For most gravity type distributors, it is a common practice to measure overall distributor performance via a water test. Tolerances on such items as hole diameter, pitch, and orifice elevation from distributor floor, etc., are imbedded in the performance criteria along with other design parameters. Thus, performance criteria are not considered a dimensional tolerance. Distributor performance can be measured in terms of coefficient of variation, maximum/minimum deviations between individual pour points or area samples and liquid head deviations from the mean between the troughs. Please respond to the following: a. Do you water test liquid distributors prior to installation? Check “ √”only one. ________Normally ________Sometimes ________Seldom __________Never (If you checked “Never”, please skip to section C.1.e.) b. What factor(s) determine if a distributor is not to be water tested, if any? Check “√” only one. _____Cost _____Service ______Schedule _____Size of distributor ____Type of distributor


c.

Do you specify distributor performance criteria or do you rely on outside sources? You may check “ √ “more than one. _________Specific criteria __________ Rely on vendors _________Rely on contractors.

d. Out of Level tolerance from high to low point. This may depend on service and tower diameter. i.

Plate distributors:

___________inches or _____________mm.


ii. Pan distributors

___________inches or _____________mm.


iii. Trough distributors

___________inches or _____________mm Page 13 of 19



3.08


iv. Parting boxes

___________inches or _____________mm


e.

Deflectiont olerances: i.

Plate distributors:

___________inches or _____________mm.


ii. Pan distributors

___________inches or _____________mm


iii. Trough distributors

___________inches or _____________mm

Comments or concerns

iv. Parting boxes

___________inches or _____________mm


f.

Distributor O.D. tolerance:

± __________inches or ± ____________mm.

2. Spray Distributors a. Out of Level tolerance from high to low point across header. This may depend on service, vertical distance from bottom of spray nozzle and bottom of bed, and tower diameter. ___________inches or _____________mm. b. Deflection tolerance, if any: c.

___________inches or _____________mm.

Elevation tolerance from bottom of spray nozzle to top of bed: ± __________inches or ± ____________mm.

d. Pitch or distance between spray nozzles: ± __________inches or ± ____________mm.


3. Support Plates for Random Packing a. Out of Level tolerance from high to low point. This may depend on tower diameter. ___________inches or _____________mm. b. Deflection tolerance

___________inches or _____________mm. Page 14 of 19



3.08


4. Support Plates for Structured Packing a. Out of Level tolerance from high to low point. This may depend on service and tower diameter. ___________inches or _____________mm. b. Deflection tolerance

___________inches or _____________mm.


5. Support Plates for Grid Packing a. Out of Level tolerance from high to low point. This may depend on service and tower diameter. ___________inches or _____________mm. b. Deflection tolerance

___________inches or _____________mm.


6. Galleries for Two Phase Feeds For reference, please see Figure 6 in section 2.02.2, Internal Pipework To Packed Towers, of the Design Practice. a.

Gallery diameter tolerance:

b. Gallery height: c.

___________inches or _____________mm ___________inches or _____________mm

Pitch or distance between orifices: ± __________inches or ± ____________mm

d. Orifice hole area tolerance: ± _______________% of total no. of holes or area specified. e.

Orifice hole diameter: ± ___________inches or ± _____________mm.

f.

Tolerance on the distance between gallery floor and centerline of inlet nozzle: ± __________inches or ± ____________mm

g. Tolerance of height, width and length of feed inlet baffle: ± __________inches or ± ____________mm h. Distance between centerline of feed nozzle to top of baffle tolerance: ± __________inches or ± ____________mm Comments or concerns:

Page 15 of 19



3.08

FRI Dimensional Tolerance Survey PART C - ALL TOWERS B. General Information

1. Your Organization’s Primary Business Area. Check “ ”only one. Chemicals Engineering Petroleum 2. Do you rely on fabricators’ standard tolerances.. Check “ ”only one. Yes Sometimes Seldom

Never


C. Other Internals

1. Chimney/Collector Tray Tolerances a.

Out of Level tolerance from high to low point. This may depend on tower diameter. ___________inches or _____________mm.

b. Deflection tolerance

___________inches or _____________mm.

c.

± __________inches or ± ____________mm.

Riser Height:

d. Hat Clearance:

± __________inches or ± ____________mm.

e.

Riser & hat dimensions such as; width, height or diameter: ± __________inches or ± ____________mm

f.

Drawoff sump dimensions such as; width, height or diameter: ± __________inches or ± ____________mm


2. Vapor, Liquid & Two Phase Orifice Feed Pipe a.

Out of Level tolerance from high to low point across header. This may depend on tower diameter. ___________inches or _____________mm.

b. Deflection tolerance

___________inches or _____________mm.

c.

± __________inches or ± ____________mm

Elevation tolerance:

d. Pitch or distance between orifices: ± __________inches or ± ____________mm e.

Orifice hole area tolerance: ± _______________% of total no. of holes or area specified.

f.

Orifice hole diameter:

± ___________inches or ± _____________mm.

Page 16 of 19



3.08

Comments or concerns: 3. Flashing Feed Baffles For reference, please see Figure 7 in section 2.02.2, Internal Pipework To Packed Towers, of the Design Practice. a.

Distance between baffles:

b. Baffle height: c.

__________inches or _____________mm __________inches or _____________mm

Pitch or distance between orifices (if any): ± __________inches or ± ____________mm

d. Orifice hole area tolerance: (if any) specified.

± _______________% of total no. of holes or area

e.

Orifice hole diameter (if any):

± __________inches or ± ____________mm

f.

Tolerance on the distance between tray or parting box floor and centerline of inlet nozzle: ± _________inches or ± ____________mm

g. Distance between centerline of feed nozzle to top of baffle: ± __________inches or ± ____________mm Comments or concerns:

D. Pressure Vessel

The following are dimension pressure vessel tolerances, which have a direct effect on the installation or performance of various internals: 1. Vessel Out Of Roundness

a.

Per ASME Code section UG-80__________Yes_________No

b. Other codes or standards:___________________________________________________ c.

Other additional tolerances:

2. Vertical Alignment

This tolerance may be dependent on location of equipment such as offshore or onshore and/or on type of fractionating devices such as structured packing or trays. If your tolerances are site or equipment specific, please elaborate on the comment section below. a.

Maximum tolerance for permanentv ertical out of alignment (tilt) at static conditions (no wind loading): _______inches/ft of vertical height or ________mm/m or_ ______degrees from vertical.

b. Is there a maximum limitation? ______________________________________________

Page 17 of 19



3.08

c.

Maximum tolerance for swayingt owers from vertical out of alignment at maximum wind loading conditions: _______inches/ft of vertical height or ________mm/m or_ ______degrees from vertical. Comments or concerns:

3. Overall Vessel Length

a.

Tolerance for overall vessel length from TL to TL ___________inches/ft of vessel length or ____________mm/m.

4. Skirt Base Plate Levelness

a.

Tolerance for bottom skirt base plate out of level from horizontal. ___________inches/ft of bottom skirt diameter or ____________mm/m.

b. Is there a maximum limitation: Comments or concerns

5. Tray Support Rings This includes all conventional trays (except for dualflow trays), chimney and collector trays, seal pans and pots. a.

Tray support ring spacing tolerance: ± ___________inches or ± _____________mm.

b. Out of levelness of support ring, from high to low point (This may vary with tower diameter):

c.

Deviation from perpendicular with respect to the vessel wall (flatness): ± ___________inches or ± _____________mm.

Check “ ” measuring point; inner edge of ring__________ or outer edge_________ d. Please indicate if the above tolerances also apply to Dualflow trays. Yes/No . If other specifications apply, please cite: 6. Packing Gravity Distributor Support Rings & Clips

These specifications may be dependent on service, type of distributor and whether or not the distributor is equipped with a leveling mechanism. a.

Distributor support ring or clip elevation tolerance: ± ___________inches or ± _____________mm.

b. Out of levelness of support ring for distributors withoutl eveling devices. From high to low point: c.

Out of levelness of support ring for distributors withl eveling devices. From high to low point: Page 18 of 19



d. Deviation from perpendicular with respect to the vessel wall (flatness): ± ___________inches or ± _____________mm. e.

Check “√” measuring point; inner edge of ring__________ or outer edge_________.


7. All Other Packing Internals Support Attachments

This includes such items as; support plates, bed limiters, spray nozzles, collectors, etc. a.

Support ring or clip elevation tolerance: ± ___________inches or ± _____________mm. b. Out of levelness of support ring. From high to low point: c.

Deviation from perpendicular with respect to the vessel wall (flatness): ± ___________inches or ± _____________mm.

d. Check “√” measuring point; inner edge of ring__________ or outer edge_________. Comments or concerns::

8. Internal Assembly Bolt Torque Tolerances

This tolerance is a function of material strength Bolt torque

± ___________ft-lbs for 300 series stainless. ± ___________ft-lbs for 400 series stainless. ± ___________ft-lbs for carbon steel. ± ___________ft-lbs for monel.


Page 19 of 19

3.08


DISTILLATION TOWER STARTUP AND SHUTDOWN

Issued:

08/31/2004

3.15

Revised:

DISTILLATION TOWER STARTUP AND SHUTDOWN

DISTILLATION TOWER STARTUP AND SHUTDOWN .................................................... 1 1.

Introduction ........................................................................................................................ 2

2.

Commissioning ................................................. .................................................................. 2 2.1

Safety ......................................................................................................................... 2

2.2

Vessel Isolation and Entry Issues ............................................. ................................. 2

2.3

Tower Inspection ....................................................................................................... 2

2.4

Instrument Calibration/Loop Checks ................................................... ...................... 3

2.5

Clean Out .................................................. ................................................................. 3

2.6

Leak Testing .............................................................................................................. 4

3.

Procedures .......................................................................................................................... 4

4.

Routine Startups ................................................................................................................. 5

5.

First Time Startup-Revamped Towers .............................................. ................................. 6

6.

First Time Startup – New Towers ................................................. ..................................... 7

7.

Special Startup Lines and Equipment................................................ ................................. 8

8.

Early Operation................................................. .................................................................. 8

9.

Shutdown ................................................ ............................................................................ 8

Page 1 of 9


1

DISTILL ATION TOWER STARTUP AND SHUTDOWN

3.15

Introduction

Distillation towers are designed to effect a desired separation and to handle a certain throughput. The design basis, however, is almost always steady-state operation as defined by a heat and material balance. To take a tower from an empty piece of equipment to a properly operating production unit requires a successful startup. Procedures must be written and accurate design documents developed to instruct the operators of the tower on the proper procedures and operations, and checklists should be developed to ensure that the procedures are executed correctly. In addition, the fractionation system must be supplied with adequate instruments, piping, and auxiliary equipment to allow it to be started up. The tower must also be shutdown periodically for maintenance, repairs, or upgrades. Some of the issues related to starting up and shutting down a tower will be discussed here; the focus will be on planned operations – emergency shutdowns will not be covered.

2

Commissioning

The process of preparing a tower for its initial startup is known as commissioning. Kister 1 offers some valuable guidance on the commissioning process. Some of the important steps taken during tower commissioning are discussed below. 2.1

Safety

A process hazard and operational analysis (i.e., HAZOP) review of new or modified equipment must be done prior to commissioning. The piping and instrumentation diagrams (PIDs) of the system should be reviewed and walked down and physically verified to ensure that subsequent commissioning operations can be conducted safely. All commissioning and startup procedures should be reviewed and approved as well. 2.2

Vessel Isolation and Entry Issues

Many steps in the commissioning process involve isolating the fractionation system (the column, reboiler, condenser, reflux drum, and any other closely connected auxiliary equipment) from the remainder of the process. Some of these steps, such as tower inspection, will require personnel to enter the vessel. Other steps, such as leak testing, simply require that the fractionation system be valved out from the remainder of the process to prevent material from leaving the system via the process piping. For the latter case, closing valves in the process piping (or blanking, if necessary) should be sufficient. The requirements for vessel entry are a bit more stringent, however. In the United States, the Occupational Safety and Health Administration (OSHA) requires that a vessel be isolated from all material (other than fresh air) and energy sources prior to personnel entry. It is the duty of the "employer" (presumably the owner of the vessel) to isolate the vessel and prepare an entry permit documenting the measures that were taken. All material and energy sources must be isolated prior to entry; because isolation valves can leak through, it is good practice to install blanks in addition to closing, locking, and tagging the isolation valves. Many companies require inert gas lines to have removable spool pieces, so that there is no possibility of introducing inert gas while workers are in the column. The confined space entry regulations also require that an individual, designated as the entry supervisor, inspect the isolation points to be sure that the vessel is truly isolated prior to entry. The OSHA regulations covering confined space entry apply not only to employees of owners of the tower, but to contract employees who might be hired to work inside or inspect the column as well. For complete confined space entry regulations, refer to Title 29, Code of Federal Regulations, Part 1910, Section 146 (29 CFR 1910.146). 2.3

Tower Inspection

Page 2 of 9



3.15

Do not assume that internals or even the weld-ins were fabricated or installed correctly. If internals were installed at the vessel fabricator's shop, they may have become dislodged or damaged in transit to the plant site. It is vital to ensure that all internals are installed securely and within level tolerances per the internals supplier's drawings. Refer to FRI Handbook Volume 5, Section 5.01 for more information on column inspection, and to Section 5.05 for a discussion of leveling tower internals. Care should be taken during initial entry to a transported tower. Tower internals may appear to be secure but may in fact be unsafe for worker entry and support. 2.4

Instrument Calibration/Loop Checks

To operate any chemical process, the level transmitters must be spanned, the calibration of temperature and pressure transmitters verified, and the control loops checked. The settings of any relief devices should also be verified. This should be done prior to any cleaning or leak testing so that the instruments may be used to monitor conditions during these tests. 2.5

Clean Out

Following construction of a plant and installation of tower internals, the column and adjoining piping will undoubtedly contain dirt, weld spatter, rags, soft drink cans, food packaging, and other assorted debris. This extraneous matter must be removed before the tower can operate properly. The lines should be blown out using nitrogen or air. However, it is not a good idea to blow lines into a distillation tower, since the items that were blocking a line may now block a downcomer or distributor. A distillation tower is often used as a reservoir for the gas used for line blowing. If this is done, great care must be taken to avoid damaging the column internals when the tower is depressurized. The column may be washed to remove dirt and other undesirable material. Washing is a potentially troublesome operation, especially when chemicals are used. Several pitfalls are discussed at length elsewhere1. Some of the major trouble spots are summarized below. Clean water should be used so that the washing operation does not deposit more solids in the tower than it removes. If river water is used to wash the tower, then mud deposits may foul the internals. If the wash water is excessively hard, then salt deposits may remain after the wash. Water used to wash stainless steel columns should be free of chlorides. The internals may be covered with machine oil that was used in the fabrication process; if the oil must be removed before the column is started, then the tower should be steamed before startup. It is also possible to have the internals fabricator remove the machine oil prior to shipment. It is a good idea to leak test the tower prior to steam cleaning in order to minimize the risk of personnel injury or equipment damage during the steaming operation. During any steaming operation, it is important to vent the tower, or pressure it with inert gas, to prevent vacuum from being inadvertently drawn. The transition from normal operation to shutdown conditions can loosen scale in the tower, piping, and auxiliaries. If the piping is carbon steel, then exposure to air during shutdown can form rust. In situations such as these where the shutdown may result in the formation or dislodging of solids, the lines should be cleaned prior to restarting the column. If these lines are blown into the column, however, the holes in liquid distributors or the downcomers in trayed columns may become plugged. For this reason, it is not advisable to blow lines into the column; strainers should be installed if this is the only way to blowout the piping. Stainless steel piping is sometimes used in services where carbon steel would otherwise be acceptable, for the sole purpose of avoiding rust formation during shutdown. Carbon steel column internals can also rust if steam-out procedures call for drying with air following steam-out. While it is unlikely that the material of construction of the internals would be upgraded to stainless steel based on this consideration alone, this effect Page 3 of 9



3.15

should be considered when developing the steam-out procedures. This is particularly important when revamping from trays to ra ndom packings. Towers that are prone to coking will probably not be adequately cleaned by conventional washing procedures. In these services, the location and severity of any plugging must be assessed by visual inspection, and the appropriate steps taken to clean the internals. If any of the pumps that will be used to operate the column are used during the cleaning operation, it is recommended that startup strainers ("witches hats") be installed in the pump suction piping. The strainers will serve to protect the pumps from any large objects, such as loose hardware, that might be washed out of the tower. Once the tower has been successfully started, the strainers should be removed; if they are not, the pumps may cavitate as scale builds up. (Note: if solids removal is desired during normal column operation, then strainers or filters should be installed on the discharge side of the feed or bottoms pumps, and the system should be configured to allow online cleaning of the solids removal device). 2.6

Leak Testing

The distillation system must be tested to be sure that no process material can leak out (for separations carried out at atmospheric pressure and above) or in (for vacuum operations). The usual means of leak testing is to isolate the system, fill it with inert gas, and monitor the rate at which pressure drops. All gasketed connections are checked for leaks by applying a soap solution to the exterior of the gasketed connection and watching for bubbles to form. More sophisticated technology such as ultrasonic leak detection may be required. Leak detection is sometimes accomplished by filling the tower with water and inspecting the exterior of gasketed connections for water leaks. If this method is used, the column and its foundation must be designed to allow filling with water in the vertical position, and care must be taken no to damage the internals while filling or emptying the tower. Steam is also used occasionally, but is more troublesome due to the high temperatures involved. Towers that will operate under vacuum are subjected to a pressure test, then vacuum is drawn on the system, the system isolated again, and the pressure rise as a function of time is once again monitored; this is commonly called a drop test. Detection of large leaks can be accomplished via hydrostatic testing, ultrasonic testing, or the soap-bubble method described earlier. Detection of small leaks may require tracer gas methods. Shaving cream has also been used to test for leaks in vacuum services; the flanges are taped, a small hole punched in the tape, and a dollop of shaving cream applied. A leak will form a small crater in the shaving cream. Before performing a vacuum test on a tower, the vacuum rating of all vessels and instruments should be verified. The diaphragm that isolates some pressure transmitters or differential pressure cells from the process may not be rated for full vacuum on the process side.

3

Procedures

The process of starting up a distillation column can be quite complicated. A detailed procedure must be followed in the correct sequence to ensure that the tower is brought up to operating conditions safely and quickly. A startup procedure for each fractionation system should written by the plant supervisors or engineers with substantial input from the plant operators. Start/Stop procedures should also be developed for emergency situations. The procedures should be subjected to the appropriate safety review process, and the plant operators should be trained in these procedures. The startup procedure should be sufficiently detailed to enable the operators to efficiently start the column.

Page 4 of 9



3.15

The procedures should take into account any special situations that exist during startup but not during normal operation. If any process or safety interlocks must be overridden to start the column, this should be noted, and reactivation of the interlock included at the appropriate point in the procedure. In some situations, the column’s instruments do not function properly during startup conditions. One common example is the level instrumentation on towers that start-up at ambient temperatures but operate at elevated temperatures; the fluid density will change substantially as the temperature increases, so level instruments based on liquid density will be misleading during the initial period of operation. Some columns are started on fluid mixtures that are less hazardous than those that will be processed during normal operation; these alternate fluids may not have physical properties similar to the normal process fluids, so instruments such as level switches that are based on electrical conductivity or level transmitters that use radar signals may not function well, or at all, until substantial inventories of the normal fluids are accumulated in the tower. During the initial startup of a column or a plant, changes to the system configuration and the startup procedures may occur relatively rapidly. It is vital to keep the procedures and design documents up-todate. Any valves, piping, instrumentation, equipment or interlocks that are added should be shown on the PIDs. Equipment drawings should be updated to reflect changes to the tower or internals, and the procedures should be modified accordingly. Although it is impractical to subject each change made in this situation to the same process hazard analysis procedure that was performed initially, all changes made during startup must be made in accordance with the applicable management-of-change procedure to ensure safety of the workers and the protection of the physical plant. Some companies conduct revalidation studies of the startup and shutdown procedures prior to each turnaround. To the extent possible, the individuals involved in the original development and review of the procedures should participate in the revalidation effort. The revalidation focuses on changes to the system that have been implemented since the last formal review of the procedures. Typically, only a couple of hours per fractionation system are required to conduct this level of safety analysis. All of the above considerations apply to column shutdown procedures as well. Procedural checklists are invaluable for executing routine startups and shutdowns. 4

Routine Startups

Towers must periodically be shutdown for maintenance and upgrades. They are also shutdown occasionally by emergency safety systems during process upset conditions. Following these relatively routine occurrences, the tower must be restarted to resume plant production. A procedure must be followed to restart the column; the details of the procedure may depend upon the circumstances of the shutdown. There are a couple of problems that commonly arise if the procedure involves washing the tower prior to the introduction of process material. If the wash liquid is boiled during the washout, then the washing step should be considered during the sizing of the column relief devices (both pressure and vacuum relief). Also, if the material processed in the tower is a high-boiler that is immiscible with water, then a detailed water removal procedure specific to that tower should be developed. Generic instructions, such as including a step in the startup procedure that says, "remove all water from the system" have been proven ineffective, often with catastrophic consequences. A water removal procedure is essential if even small amounts of water can be present in the system at startup, either as a result of washout or from any other source. In most cases, it is advisable to bring the tower to operating pressure prior to the introduction of feeds. For pressure distillations, feeding at atmospheric pressure will probably allow the feed to flash; the cooling that results from this flash could cool the vessel below its embrittlement temperature if it is fabricated from carbon steel. For vacuum distillations, it may not be possible to boil the feed materials with the available heating utility until vacuum is established. More details concerning the introduction of material at startup can be found in Kister 1. Page 5 of 9



3.15

When the operating pressure is reached, it is necessary to establish vapor and liquid traffic in the column. Many towers are started on total reflux; this involves introducing a quantity of material into the column, turning on the hot and cold utilities, boiling some material overhead, condensing the material, collecting it in a reflux drum, and returning all of it to the column to be boiled again. Once vapor and liquid flows are established, continuous feed can be introduced and products withdrawn. It is possible to achieve hydraulic stability without achieving the desired separation; once the tower has “lined out” (i.e., established relatively constant temperature and pressure profiles and fairly steady levels in the column base and reflux drum), operating conditions can be adjusted as necessary to improve the separation. If the tower is part of a process with significant recycle, it may take a long time to establish the composition profiles for which the tower was designed. Some towers cannot be started on total reflux. In these cases, it may be necessary to simply begin feeding the tower and operating normally. Other towers may require starting on a fluid similar to the process fluid. Towers in long refining trains with no intermediate storage can be particularly troublesome to start; such columns may be started more easily if the reflux drum is filled with on-spec material prior to the startup. Difficulties can also arise when starting towers in which the base temperature is controlled by bottoms outflow. If the material in the bottom of the column is too cold, as it is likely to be during startup, then the control system will not allow any material to leave the bottom of the column. This will result in high base level; if the alarms are ignored, tower flooding and internals damage can result. To start columns that are controlled in this fashion, it will probably be necessary to put the base temperature loop in manual control and allow some off-spec material to leave the tower until the base is heated to operating temperature. It is important to pay attention to the column material balance while non-steady state conditions prevail during startup. The flow rates of the feeds and products should be compared frequently to be certain that the accumulation of material in the tower does not exceed the desired inventory. The startup of each column must be considered carefully. Kister to start a tower. 5

1

offers some guidelines on the best ways

First Time Startup-Revamped Towers

The first startup of a distillation column following a revamp offers many of the same challenges as a new tower, as well as several special considerations. Trayed towers revamped with structured packings will exhibit a pressure drop that is lower by approximately an order of magnitude; if operators attempt to push the tower until the previous pressure drop is reached, the tower will be flooded and the internals possibly damaged. Because structured packings have very little liquid holdup compared to trays, the tower will establish steady state about ten times faster than with trays, and control loop responses will be much faster in packed towers than in trayed towers. If a chimney tray was seal welded as part of a revamp, liquid level will be established much more quickly on the tray. Revamps usually involve changes to internal tower piping; such changes may eliminate old internal piping problems or generate new ones, and thus affect tower response at startup. Sloley2 summarized several experiences with revamping towers. Many of these involved instrument nozzles that were not relocated. For example, if a tower is revamped from trays to packing, the pressure taps should be relocated so they are between packed beds. If the tray spacing is modified as part of a revamp, temperature nozzles should be relocated to the desired location (usually the bottom of a downcomer for liquid temperatures, and just below a tray for vapor measuring thermowells). This is particularly important for temperature measurements that are used as part of the column control strategy. If the tower was revamped with more efficient internals, it will probably be necessary to change the temperature control point; it may be necessary to add nozzles to the tower if there are none available at the desired location. Because the temperature control loops will respond much faster, it is a good idea to retrain the operators prior to starting a revamped tower. Sidedraw nozzles should be relocated to the

Page 6 of 9



3.15

appropriate location as part of the revamp. Replacing column auxiliaries, such as the reboiler, condenser, or pumps, can prove problematic if not handled properly. Larger pumps may cavitate if suction head requirements are not satisfied. New reboilers and condensers may respond differently than the old units; these differences must be considered when developing the startup procedures. Modifications to other equipment in a process can affect the operation of distillation equipment if the modifications alter the composition, quality, or rate of feed to the column. The effect of proposed modifications on the entire process should be considered before they are implemented. It may prove necessary to modify the tower or internals as well to take full advantage of any expected improvement. It is also important to be sure that all process PIDs affected by the revamp are up-to-date and accurately reflect the process as configured. 6

First Time Startup – New Towers

The “maiden voyage” of any distillation system presents some special challenges in addition to the usual concerns with tower startup. The column itself may only be a small part of the scope of the startup; there may also be a reaction system, other separation systems, and auxiliary equipment associated with the column of interest. At a new plant site, all the relevant utility systems must be commissioned and fully operational prior to tower startup. The operators will have to travel a learning curve with respect to the operation of the column. Experience with other distillation systems will be of great benefit, but the new system may have characteristics that take some time to appreciate. A simple system may take only a few hours to start once commissioning is completed; a reactive distillation column, on the other hand, may take many months to fully understand. It is absolutely essential that the tower instruments be fully commissioned prior to startup. Many safety systems depend upon accurate information from the process instrumentation, so operating with uncommissioned instruments can endanger life and property. Even if no hazardous situations arise, it is possible to damage the column internals or produce large amounts of off-spec material 3. The risks associated with operating the tower with uncommissioned instrumentation far outweigh any perceived benefits. When the tower is to process hazardous chemicals, it may be possible to first start it up on similar but less hazardous chemicals, and then when the system is operational and tested, bring in the intended chemicals. This procedure permits debugging the tower and giving the operators some initial experience with it without risking the consequences of mishandling hazardous materials. During the first startup of a tower, factors that were ignored or inadequately considered in the plant design will come to the surface. Tower internals, auxiliaries, and relief systems that were designed based on normal operating conditions only may be inadequate for startup or shut down conditions. The sizing of control valves or flow orifices, if not done correctly, will restrict flow into or out of the column. If the pressure loss across control valves and other equipment was not accounted for, feeds to a column may flash unexpectedly, to the detriment of tower capacity, efficiency, or both. In cold services, such flashing may cause chilling below the pipe metal safe working temperature. Pumps may cavitate if adequate suction head is not available, or if vortex breakers were omitted from outflow nozzles. Operators may target erroneous flows into the column if energy balance effects are not taken into consideration; one example is a greatly subcooled reflux resulting in a much lower external reflux flowrate than predicted by the design simulation (which probably used reflux at its bubble point). Initial attempts to start a column may be thwarted by downcomers that are not sealed at low liquid rates; failure to establish a seal may not allow the tower to be brought up to operating rates. This is particularly Page 7 of 9



3.15

problematic with sieve trays. Kister 4 developed a method for generating a startup stability diagram to allow this problem to be identified at the design stage (updated equations for this method are available in Appendix B of Kister 1). Downcomer designs that ensure a positive seal under all conditions eliminate this concern. If this problem is not discovered until the tower is constructed, the trays can be fitted with inlet weirs or the decks replaced with valve trays. Many modern high capacity tray designs feature dynamically sealed downcomers. To successfully start towers containing these trays, it may take longer to establish sufficient inventory to seal the downcomer openings. All tower internals, auxiliaries, and relief systems should be sized for the maximum expected rates. 7

Special Startup Lines and Equipment

Some towers may require special piping and equipment to be brought up to operating rates. Some reboilers need equalization lines to be started. Special piping will be needed if material other than the normal column feed will be used to start the tower. In heat integrated systems, startup reboilers or heaters may be required to establish flows so that the heat-integrated units can begin operation. Compressors usually require spill-back lines, after their trim coolers, to enable startup. Once the tower has been successfully started, all special piping and equipment must be properly isolated. Thermosiphon reboilers are a common means of generating boilup in a tower, but can be difficult to start. Startup lines or special startup operations are sometimes required. Columns with baffled sumps are usually equipped with a dump line between the reboiler inlet line and the sump outlet; it may be necessary to open this line during startup to establish thermosiphon. The same is true for reboilers that are fed from a trapout. In vacuum services, where thermosiphon action is particularly sensitive to liquid level, lowering liquid level or throttling liquid flow to the reboiler may be needed before the unit will start properly. If the reboiler is mounted far below the return nozzle, it may be necessary to inject gas in the liquid return line; this lowers the density of the material in the return piping and helps to initiate thermosiphon action. Refer to the FRI Handbook, Vol. 5, Sections 1.06 and 4.03.1 for more information on thermosiphon reboiler circuits. 8

Early Operation

During the first few weeks or months of tower operation, operating conditions may change unexpectedly. Impurities that were not identified during bench scale or pilot studies may accumulate in the system, disrupting column operation. Corrosive conditions that were previously undetected may also manifest themselves. Tray vibrations at low operating rates (such as those experienced in startup) may hamper startup, or prevent it altogether. See the FRI Handbook, Vol. 5, Section 1.21 for more information on the causes and prevention of tray vibration. Changing tower internals (revamping trays with packing, for example) can allow conditions such as corrosion, fouling, or mechanical strength deficiencies that were not previously problematic to manifest themselves. Some designers say that the HETP of plastic packings improves after a few weeks of operation. Changes in the packing surface due to etching or solids deposition may bring about this improvement, or it may be the removal of mold release agents and plasticizers that could promote foaming. 9

Shutdown

As with tower startups, the procedure for shutting down a tower must be executed in a definite sequence. In most cases, it is advisable to remove liquids from the tower before returning it to atmospheric pressure. Washout, steaming, and inert gas purging may be needed to cool the vessel and prepare it for entry; it may be necessary to bring the column to atmospheric pressure to perform these operations. If the wash liquid is boiled during the washout, then the washing step should be considered during the sizing of the column relief devices. If the column contains combustible or pyrophoric liquids or deposits, then it should not be Page 8 of 9



3.15

opened to the atmosphere immediately following steaming, but should first be cooled down with some inert gas. This is particularly true for towers with internals having large surface area (such as structured packings or mist eliminators). Failure to cool such columns prior to air introduction has resulted in fires. See the FRI Handbook, Vol. 5, Section 2.11 for more information on the causes and prevention of packing fires. These towers are sometimes cooled by filling with water, or a small flow of water may be left on the packing through the duration of the shutdown. Adequate time should be allowed for shutdown operations; it may take up to two days to steam out a tower and prepare it for entry. The underflow line should be sized to drain the tower in a timely fashion (or a separate drain line provided) on towers that normally operate with a very small bottoms/feed ratio. Depending upon the hazards associated with the service, additional measures may be required to decontaminate the vessel. Isolation of the vessel will require still more time. All factors related to the column shutdown should be considered when planning maintenance or revamps.

References

1. Kister, H.Z., Distillation Operation, McGraw-Hill, New York, 1990 2. Sloley, A.W., “Avoid Problems During Distillation Column Startups,” Chem.Eng.Prog., 92(7), p. 30, 1996 3. Hower, T.C., and H.Z. Kister, “Solve Column Process Problems, Series 2,” Hydrocarbon Processing, June 1991 4. Kister, H.Z., “When Tower Startup Has Problems,” Hydrocarbon Processing, 58(2), 1979, p. 89 5. Title 29, Code of Federal Regulations, Part 1910, Section 146 (29 CFR 1910.146).

Page 9 of 9


PERFORMANCE TESTING

Issued:

05/24/2006

3.20

Revised:

PERFORMANCE TESTING

PERFORMANCE TESTING .................................................................................................... 1 1.

Introduction ........................................................................................................................ 2

2.

Types of Performance Testing ............................................................................................ 2

3.

Strategic Preparation – Setting Expectations............................................... ....................... 2

4.

Strategic Preparation - Measurements ................................................................................ 3

5.

Site Preparation................................................ ................................................................... 3

6.

Gathering the Data .............................................................................................................. 4

7.

Reconciling the Data .......................................................................................................... 4

8.

Interpreting the Data ................................................. ......................................................... . 4

9.

Post Processing ................................................ ................................................................... 5

10. Example ................................................... ........................................................................... 5

Page 1 of 6


1

•

•

General introduction / value of performance testing troubleshooting existing columns o establishing a baseline for new columns o validating a performance guarantee o o validate and improve computer models (so they are more predictive) importance of preparation and evaluating data to get meaningful results o Scope of this section provide information to help define actual operating performance of columns o focus is on single columns, rather than overall refining trains o hydraulic capacity testing o efficiency testing o interactions between hydraulics and efficiency o What is this section ‘not’ not a detailed guide on site preparation, since those topics are covered well elsewhere o not a detailed guide on computer simulation, though useful to refer to that design practice o doesn't deal directly with unsteady state operation o not intended to be a guide on troubleshooting, though information can be used as part of such an effort o

Types of Performance Testing •

3

3.20

Introduction •

2

PERFORMANCE TESTING

Brief discussion of broad scenarios for performance testing total reflux o continuous operation (normal operation - making sense of available data) o continuous operation (planned performance test) o

Strategic Preparation – Setting Expectations •

•

•

•

Two broad categories of strategic preparation: insure that “expected” performance is understood o insure that the measurements made will be adequate o Goal: Defining “Expected” performance clarify distinction between “expected”, “desired”, and “guaranteed” performance o be clear up front what you’re looking for: hydraulic capacity, efficiency, both? turndown? o o to get a good test, need to have some feel for what results should look like computer simulation o experience (what is achieved elsewhere) o hydraulic evaluation o Computer simulation baseline computer simulation (reference DPC section on Computer Simulations) o very helpful to have sensitivity studies done in advance too (if performance is too sensitive to a o parameter, might want to avoid using that one as a performance match criterion) establishing expected stage efficiency o selecting VLE (essential to get meaningful efficiency results) o understand accuracy of underlying data (difficulty matching base temperatures if high boilers present) o identifying pinches (can make it difficult to get a reasonable match) o Experience known performance of similar facilities elsewhere o need to make sure that the separation they think they are achieving is really what they areachieving. o (sometimes analytical methods can leave out components.)

Page 2 of 6


o o

•

•

4

3.20

benchscale / laboratory data pilot data

Hydraulic rating approach to maximum capacity (can influence performance) o what were original internals designed to do? o Other factors that could influence performance performance diagrams o other information sources - gamma scans, etc. o

Strategic Preparation - Measurements •

•

•

•

•

5

PERFORMANCE TESTING

Goal: insuring that measurements made will be adequate definining what to measure o recording operating data o recording sample data o Defining what will be measured importance of material / energy balance o understanding which measurements are available o possibility of portable instrumentation o “Making do with less data”... pitfalls and possibilities checking material / energy balances in advance, using available measurements (‘mini’ performance test) o being clear on the consequences of inadequate data o sometimes you can draw meaningful conclusions even when data are flawed (Kister 1998 Nye-tray o example - no energy balance, poor mass balance - still meaningful) when is a test simply not worth running? o Recording operating data understanding data historians o understanding source of all data (elevation of instrument taps, etc. mention trays labeled backwards, o instruments hooked up incorrectly to DCS etc.) understanding nature of all data (average, compressed, or real data) o understanding calibrations (liquid density impact on flows, in particular) o use real data if at all possible - averages may be meaningless - distillation is not a linear process o Understanding sample data o time taken vs. time measured where will samples be taken (representative of instantaneous performance, or considerable lag?) o what are methods really reporting (do analytical results add up to 100%? If not, what is missing?) o challenges with multiple liquid phases o challenges with reactive systems o challenges with vapor samples when condensing and/or entrainment can occur o

Site Preparation •

•

Acknowledge wealth of other resources available on this particular topic (Include only a high level nod to the subject in this section – one or two paragraphs mentioning the types of ‘site prep’ subject matter covered in the references.) Advance preparation lining up resources o

Page 3 of 6


o

•

6

•

•

•

sampling load

Immediate preparation calibrating instrumentation o safety checks o etc. o

Running the test Recording the data getting best operator on shift to take the data o Validating data make sure it makes sense before test ends o o checking material balances Defining steady state what to look for o o importance of good process control look at response of output variables - cycles? o need to let things line out (may need to turn off multivariable control and go with regulatory controllers) o residence time / turnovers o some processes may never come to steady state (large reflux drums with tiny draws, tar purge columns, o etc.)

Reconciling the Data •

•

•

8

3.20

Gathering the Data •

7

PERFORMANCE TESTING

Deciding what to believe when there are conflicting data heirarcy of reliability and importance o process specific o Instrumentation accuracy (refer to DPC section 3.05 - Location of Instruments) inherent accuracy of various instruments o common sources of incorrect readings o installation problems o o wrong instrument for the application Selecting which data to use forcing the balances to close o o using component balances to reconcile flows correcting sample results (example of condensing liquid in vapor sample) o

Interpreting the Data •

Absolute performance, and also comparison to baseline expectation Total reflux o o Fenske equation very useful because it reduces the number of variables that can influence results o some simulation packages can handle this, many don’t o look for changes in efficiency as absolute boilup rate changes. any changes so highlighted are o indications of changes in underlying efficiency.

Page 4 of 6


•

•

•

•

•

9

10

PERFORMANCE TESTING

Continuous operation o additional complications Computer simulation match reference DPC section on computer simulation o setting initial efficiency expectation (constant throughout column? variable?) o strategies for defining what actual efficiency is o Validating assumptions comparing temperature profile versus expectations o comparing pressure profile versus expectations o luxury of in-column samples, if available (mention FRI data, as example) o Comment on conservatism dealing with a performance match that is not exact o Interpretation pitfalls pinches o proximity to minimum reflux (changes in stages can be profound) o differences in VLE o unsteady state operation o o position on operating diagram

Post Processing •

Communicating results

•

Possible ties to troubleshooting support

•

Definining when the job is done

Example •

Total reflux performance test example

•

Continuous operation (normal operation) performance test example

•

Continuous operation (planned test) performance test example

Page 5 of 6

3.20


PERFORMANCE TESTING

3.20

References 1.

“Guide to Trouble-Free Distillation”, D. B. McLaren & J. C. Upchurch, Chemical Engineering, June 1, 1970

2.

“Using Performance Data to Improve Plant Operations”, L. Yarborough, et. Al, Proc. 59 th Annual

3.

Convention of the Gas Processors Association

4.

“Distillation Column Performance Testing: Continuous and Batch Approaches”, V. G. Kalthod, J. R. Fair, et. al., AIChE Nov. 1997.

5.

DPC Section on Computer Simulations

6.

AIChE Equipment Testing Procedure – Tray Distillation Columns, 2nd ed. August 23, 1986.

7.

Distillation Operation, H. Z. Kister

8.

“Efficient Test Runs”, G. J. Gibson, Chemical Engineering, May 11, 1987

9.

“High Performance Trays Increase Column Efficiency and Capacity”, D. R. Summers et. al., AIChE Apr. 2003

10. “Ethylene fractionator revamp results in 25% capacity increase”, D.R. Summers, et. al, Oil & Gas Journal, Aug 10, 1992 11. Kister 1998 paper on performance matching of Nye trays with limited / poor data / TBD 12. “Push Valve Experience on Distillation Trays”, D. R. Summers, AIChE Apr. 2005 13. “Enhanced V-Grid Trays Increase Column Performance, D. R. Summers & D. Ehmann, AIChE Nov. 2002 14. “Trayed Revamp Yields a Significant C3 Splitter Capacity Increase”, D. R. Summers & S. T. Coleman, AIChE Nov. 1990

15. “High Capacity Trays improve RVP”, S. E. Grill et. al, AIChE Apr. 1993.

Page 6 of 6


PROCESS SIMULATIONS

Issued:

08/04/2003

Revised:

3.50

10/10/2008

PROCESS SIMULATIONS

PROCESS

SIMULATIONS.....................................................................................................................................1

1.

Introduction .................................................................................................................................................... 2

2.

Review of Existing Equipment and Process Flows ......................................................................................... 2

3.

Data Gathering ................................................................................................................................................ 2

4.

5.

3.1

Sample Collection & Lab Data ............................................................................................................... 3

3.2

Heat and Material Balance ...................................................................................................................... 3

Selection of Property Packages ....................................................................................................................... 4 4.1

Non-Ideality ............................................................................................................................................ 4

4.2

Dual Property Packages .......................................................................................................................... 4

4.3

Azeotropes............................................................................................................................................... 4

4.4

Checking of VLE Performance ............................................................................................................... 4

Model Construction Considerations ................................................................................................................ 5 5.1

Stage Efficiencies .................................................................................................................................... 5

5.2

Equilibrium Stage versus Rate Based Calculations ................................................................................. 5

5.3

Graphical Solutions – Graphical Troubleshooting, Composition Profiles .............................................. 6

5.4

Dynamic Simulation................................................................................................................................ 6

5.5

Feeds and Draws ..................................................................................................................................... 6

5.6

Reboilers, Condensers, Pumparounds, Heat Pumps ................................................................................ 7

6.

Model Validation ............................................................................................................................................ 7

7.

Interpreting Results ......................................................................................................................................... 7

8.

Selecting Data for Rating Internals ................................................................................................................. 8

9.

Internal Simulation Package Rating Programs................................................................................................ 8

10. Economic Evaluations & Sensitivity Analyses .............................................................................................. 8 11. Bibliography...................................................................................................................................................9

Page 1 of 9

Issued: 08/04/2003 Revised: 10/10/2008

1

PROCESS SIMULA TIONS

3.50

Introduction Process simulation packages are an invaluable tool for simulating process operations. When used properly, a simulation will accurately model a process so that output information can be used to calculate operating conditions that are otherwise unmeasurable, such as internal column loadings and physical properties. The bases used to converge all process simulations are Vapor-Liquid Equilibrium (VLE) data and the heat and material balances. The calculated compositions of the vapor and liquid are based on the specified VLE data. The heat balance is calculated using a specified thermodynamic equation package. Therefore, the final simulation results are only as good as the packages that were selected and the data that were input. Although process simulations are used to simulate entire operating units, the scope of this section will be limited to column operations. The basic steps required to perform a process simulation are: • • • • • •

Review of Existing or Future Equipment to be Simulated Data Gathering Selection of the Thermodynamic and Vapor – Liquid Equilibrium (VLE) Packages Construction of the Model Validation of the Model Interpretation of Results

All of these elements are critical to the success of a simulation. The ideal simulation will match all aspects of how the column actually operates or will operate. If the basis for the simulation construction is not representative of the actual column operation and design, the simulation cannot be accurate.

2

Review of Existing Equipment and Process Flows The first step of the simulation process is to make sure that the process flows are understood and accurately represented. All feed, product, and recycle streams as well as energy sources need to be identified and accounted for. It is most important to note at what tray or between what beds a feed or product is located. Depending upon what the simulation is to be used for, minor streams may or may not need to be included in the simulation. However, when excluding minor streams, you must be very careful to ensure that this omission is truly inconsequential.

3.

Data Gathering Data gathering is extremely important when constructing a simulation. Errors in data gathering can lead to simulations that will not converge or even worse, simulations that do converge and produce incorrect outputs that go undetected. One of the most important steps for gathering data is planning. During a hectic test run, it is often difficult, if not impossible, to get additional samples or data points taken. The fewer unexpected requests during the test run, the more smoothly and accurately the data will be collected. Safety considerations must always be carefully evaluated when sampling streams that are at high pressures and temperatures or are flammable or toxic. Proper sampling technique must be observed when collecting samples. For further information on proper sampling procedures refer to section 2.08 of this manual or the AIChE equipment testing procedures.1 A list of the typical data required for a column simulation is shown below.

Page 2 of 9

Issued: 08/04/2003 Revised: 10/10/2008


3.50

Temperatures & Pressures Overhead Receiver Column Overhead Any Column Midpoints Available Feeds Bottoms Heaters/Exchangers inlets and outlets • • • • • •

Compositions Feeds Products/Product Specifications Column Tray or Bed Samples Heating or Cooling Streams Heater Fuel Streams • • • • •

Flows • • • •

Feeds Products Flows to Heaters/Exchangers Reflux and Pumparound

For new columns, these values will be defined as part of the design scope. F eeds and product specifications should be defined there. If the simulation is being used for a feasibility study, perhaps a range of feed and product rates and specifications will be used to determine an economic optimum. Temperatures are typically set by available heating and cooling sources. Pressures are normally set by the temperature and composition requirements. Unless the process being simulated is uncommon, it is a good idea to look at the operation of similar columns to review typical operating conditions. For column revamps, operating data can be gathered from the existing column operation. This can often be accomplished effectively with the use of a distributed control system (DCS). Collecting data from a DCS will normally require a tag list that will be used to download pertinent data on a periodic or time averaged basis. It is important to have up to date Process and Instrumentation Diagrams (P&IDs) and master tag lists from which to construct the test run tag list. In some data logging systems, data which are not included beforehand on a test run tag list cannot be retrieved afterwards and are therefore lost. Be sure that your tag list is complete and will provide you with enough data to construct a full heat and material balance around the column. When in doubt as to whether or not a tag is relevant, include it on your list.

3.1

Sample Collection & Lab Data When collecting data, it is imperative that the column be running at steady state conditions when a data set is collected. It is always important to know what the response time of the column is when process changes are being made. Smaller columns may equilibrate to a process change in a matter of minutes while larger columns may take several hours. Accurate data cannot be collected until the column has had time to equilibrate. The response time will be a function of the column size, liquid holdup, instrumentation, and complexity. The best way to determine the column response time of an existing column is to talk to the operators and then verify this by making a moderate change to the column and observing the trends. When available, a dynamic simulation can sometimes be used to model the response time.

3.2

Heat and Material Balance When taking data, it is important to also include heat balance data to cross check the rest of the column information. Without a heat balance, there is no way to verify the reflux or pumparound flow

Page 3 of 9

Issued: 08/04/2003 Revised: 10/10/2008


3.50

rates. This information is extremely important with multi-draw columns where internal loadings change significantly in different sections of the tower.

4

Selection of Property Packages Existing VLE correlation packages may be modified with research data. These data can be found in literature or may be available from the simulation vendor. Many large operating companies have in-house VLE data. Of the various process simulation software packages available, some are better suited for different operating systems than others. Most packages handle ideal hydrocarbon systems readily. Non-ideal, aqueous, and electrolytic systems can be more difficult to simulate. Proprietary chemicals can also be difficult to simulate due to a lack of published physical property data. When dealing with these systems, more accurate results can usually be obtained by using externally gathered data. Some chemical systems have simulation packages designed specifically for them. For example, there are several specialized simulation and property packages available to deal with amine systems . Many process simulation software packages provide an oil package to evaluate hydrocarbons based on laboratory distillations, densities, UOP “K factor”, viscosities, and other available properties. These oil packages then divide the hydrocarbon streams into pseudo components and then perform the simulation with these components. 4.1

Non-Ideality Equation of state models such as Peng-Robinson, SRK, etc., are reliable in predicting properties of ideal systems but are limited in their ability to handle non-ideal or polar systems. These systems normally require the use of a dual model system where an equation of state is used to predict the vapor fugacity components and an activity model is used for the liquid phase. However, activity models are much more empirical in nature. Therefore, more caution should be used when selecting these models for your simulation.

4.2

Dual Property Packages In some columns, the equilibrium conditions in the top section of the tower can be different enough from the bottom section to warrant using different VLE packages in each section. In this case, the column sections are initially modeled separately and linked together in the simulation using recycle streams. This technique can be used to obtain more representative simulations. 2

4.3

Azeotropes Care must be used when simulating systems with azeotropic components. Because of their nature, these systems present a severe test for simulation programs. Residue curves, which have been around since the early 1900’s, can be used effectively to deal with azeotropes. A residue curve is a plot of the liquid phase composition as it varies with time during a simple distillation. Residue curve maps can be pictured graphically for up to four components. Residue curves can be constructed from experimental data or can be calculated analytically if equations of state or activity coefficient expressions are available. Some examples of residue curves and general information on how to use them can be found in Perry’s Handbook.3 The use of a residue curve map of the azeotropic components can provide a good indication of what separations are possible.

4.4

Checking of VLE Performance Once a VLE package has been selected, it is good practice to print out a vapor-liquid plot at conditions similar to available reference data to ensure that it corresponds well, particularly if one is using default interaction parameters from the simulation package.

Page 4 of 9

Issued: 08/04/2003 Revised: 10/10/2008

5


3.50

Model Construction Considerations There are various factors to be considered for the actual construction of the model. Once again, the primary concern with the model construction is accuracy of the model with respect to the actual column operation. Care must be taken to ensure that the scope and complexity of the model is such that it will provide the full range of desired information. Various factors pertinent to the model construction are discussed below. 5.1

Stage Efficiencies Many simulation programs allow stage efficiencies to be used to simulate the actual tray efficiencies. This technique is convenient, but can produce incorrect results if not used properly. When stage efficiencies are used, the resulting simulation will often indicate that the heavier components will carry higher up in the column than in actual operation. This will also lead to an increase in predicted 4 stage temperatures . Because of this, the use of tray efficiencies is generally not recommended. If stage efficiencies are used, care should be taken not to apply them on stages that contain product draws. It should also be noted that rate based simulations rely on calculations of interphase mass transfer for actual internals. Since they do not use the concept of theoretical stages, the concept of tray efficiency does not apply

5.2

Equilibrium Stage versus Rate Based Calculations The vast majority of process simulations are conducted using an equilibrium stage model. This method models the column as a stack of equilibrium stages or theoretical plates. The key assumptions required for this model are perfect mixing and thermodynamic equilibrium. In reality, these assumptions are never truly accurate. As a result, the concepts of tray efficiency and HETP are used to account for less than ideal mixing and departure from equilibrium. Because of the above-mentioned limitations, a rate based method may be better suited for the simulation of some processes. This method models the column as a group of segments where equilibrium is assumed only at the vapor-liquid interface. These segments can be used to represent a tray or a portion of a packed bed of a specified height. With the rate based approach, the degree of separation achieved depends on the rates of mass and heat transfer between phases contacted on a tray or section of packing. The advantages of the rate based method are that it eliminates the uncertainty of using stage efficiencies and HETP, predicts the departure from temperature equilibrium, and can account for interactions between diffusing components. Rate based simulations are recommended for nonideal multicomponent systems, reactive distillation systems, or any system where there is an interest in the distribution of non key components. Specific cases where rate base simulation can be beneficial are applications where mists or bubbling may form from the presence of vapors at temperatures below their dew point or liquids at temperatures above their bubble point. 5 With respect to tower internals, rate based simulations have their greatest advantage with packed towers because in packed towers, there is no limitation on the number of stages into which the tower can be divided. Rate based simulations have a major advantage where conditions are changing rapidly, especially in short beds where fractions of a theoretical stage are important. In depth questions regarding the applicability of certain processes for rate based modeling should be addressed to the simulation software vendors.

Page 5 of 9

Issued: 08/04/2003 Revised: 10/10/2008

5.3


3.50

Graphical Solutions – Graphical Troubleshooting, Composition Profiles Most simulation packages will allow you to print out McCabe-Thiele diagrams of columns as well as reflux ratio versus theoretical stages. These diagrams provide a graphical view of the column operation and can many times help identify possible problems or limitations of the column operation. These graphs are particularly useful for identifying column pinch points and approaches to infinite reflux ratio or theoretical stages. A plot of stages versus reflux ratio is an invaluable tool that can be used to ensure that the column design is optimized with respect to energy cost and capital investment as well as indicate how sensitive the design is to a loss of theoretical stages.

5.4

Dynamic Simulation Dynamic simulation simulation can be a useful tool for simulating tower operations that are not considered steady state. Dynamic simulation has been s uccessfully used for reviewing relief scenarios for distillation columns and also for control analysis. Instrument perturbation simulations simulations can be useful for evaluating the instrumentation instrumentation performance. Dynamic Simulations may help identify control problems that have have substantial impact impact on performance performance of tower internals. internals.

5.5

Feeds and Draws The simulated introduction introduction of a mixed feed to a column can be a source of uncertainty. This is an important issue because it determines determines what the vapor and liquid loadings will be above and below that point. The standard standard handling of mixed phase phase feeds is for the vapor portion portion to be mixed with the the vapor stream from the stage below the feed stage and the liquid portion to be mixed with the liquid from the stage above. These combined streams are then mixed on the feed stage. An example of this is shown below.

Page 6 of 9

Issued: 08/04/2003 Revised: 10/10/2008


3.50

Superheated and subcooled feeds produce unique loading conditions within a column. A superheated feed, which is commonly found in an FCC Main Fractionator, will produce a “vapor bulge” in the stages above the feed. This is the result the superheated vapor heat causing a greater than mole for mole vaporization of the liquid that it contacts. The number of stages in this section of the tower must be correctly specified specified so that the bulge appears in the simulated simulated tower loadings. loadings. One method of accounting for the bulge is to first simulate the tower with the correct number of stages required to match the actual or desired tower operation and then add enough stages to a second simulation to produce the bulge loadings. The first simulation simulation will give correct correct loadings for all sections sections of the tower except for the bulge section. The second simulation will be used only to get the hydraulic loads in the bulge section. In the example example case of the the FCC Main Fractionator, Fractionator, two stages are normally normally required for the actual simulation and three stages are required to properly yield the bulge loadings. Internals in the bulge section of the the tower must be sized to to adequately handle handle these loads. Correspondingly, Correspondingly, a subcooled liquid feed or reflux will produce a liquid bulge in the stages below. This too, must be accounted for when designing the tower internals. 5.6

Reboilers, Condensers, Pumparounds, Heat Pumps When constructing or reviewing simulations, the reboiler and condenser operations need to be evaluated to determine as to whether or not they constitute a full equilibrium stage and should be modeled accordingly. Condensers must be specified as total, partial, or subcooled. More detailed information on this subject can be found in section 4.03 of this manual entitled , Exchangers in Fractionation Services. In some cases, the reboilers and condensers are simulated as an integral part of the tower operation. In others, the reboiler and or condenser can be more accurately modeled as a separate piece of equipment that is operating in conjunction with the tower operation. Pumparounds are sometimes treated simply as a side exchanger attached to a particular theoretical theoretical stage. This may be convenient to the simulator but it provides nothing for the process engineer working up the design of the pumparound circuit. A more thorough way of simulating a pumparound is to withdraw the fluid out of the tower, provide the necessary heat exchange and place it back in the tower as another feed. This obviously is more complex and may result in considerable convergence convergence difficulties, but the results will be meaningful. The addition of a heat pump to a column adds another degree of complexity to the simulation. With these configurations, special emphasis must be placed on the integrated system. The compressor operation including the compression ratio, hydraulics, and heat sink must be thoroughly evaluated to ensure that it functions efficiently with the column operation.

6

Model Validation Once the simulation has been run and converged, it should be validated against other data. In most cases, there will be an existing column or other s imilar operations, which can be used as the basis for the validation. The model should predict, as closely as possible, the correct composition, temperature and pressure profiles, and energy requirements. Variances between the model and the actual operation or predicted operation need to be reconciled. If possible, a simulation validation validation for more than one operating point will provide more confidence in the applicability of the simulation over a range of operations. This is especially important important when extrapolating data for revamp cases.

7

Interpreting Results Column internal loadings can be generated from the validated simulation and used for rating of the internals. Most simulations print out the physical and transport properties of the vapor and liquid stream leaving each stage. When using these data for internal ratings, it is important to remember to select the liquid stream leaving

Page 7 of 9

Issued: 08/04/2003 Revised: 10/10/2008


3.50

one stage and the vapor stream leaving the stage below. However, the designer s hould be aware that loadings are occasionally increased across a given theoretical stage. The designer must then prorate the loadings to and from the theoretical stage in question. Flashing feeds, for example, may artificially increase a stage’s loads when in reality the extra load is above it. When working with reboilers and condensers, you must identify the flows shown that are outside the actual column and not incorporate them in the rating process. Be careful to verify the tray numbering system versus the simulation numbering system. Most simulations number stages from top to bottom with the condenser being stage number 1 and the reboiler being the last stage. By convention, actual column trays are numbered without consideration of condensers and many columns are numbered from bottom to top. Pumparound sections can have their own nuances. These sections are typically situated in a tower to simply achieve heat transfer. Two or three theoretical stages are needed to just to establish (be far enough away from the feed and withdraw trays to see) what the liquid and vapor loads are in this section of the tower. In addition, the mass transfer rate and the heat transfer rate are typically different in a real tower. Care should be taken to ensure that the material balance is closed and the temperatures should be checked in a pumparound section.

8

Selecting Data for Rating Internals When selecting data for rating of internals, review the entire column loadings to get a feel for variances in the internals loads. Check vapor and liquid volumetric flow rates and physical properties throughout the column. Stages with significant variances in transport properties will require additional ratings account for these changes. Normal practice is to select high and low loading cases for each similar section of the column to rate or size the internals. Some simulation packages will do this automatically. The next step is to adjust the simulation or flows to account for desired spare capacity or turndown. In some cases the turndown can just be simulated as 50% of design flow or whatever criteria is required. In some cases, other factors may require that a separate turndown case be simulated. For Reflux distributors it is important to size them for the proper temperature. If a reflux is subcooled, then the design of the distributor should be made using the subcooled reflux temperature. temperature.

9

Internal Simulation Package Rating Programs Rating programs for column internals are a feature provided by most simulation packages. These rating programs can be useful for initial initial sizing but often often have limited range range of applicability. applicability. Simulation hydraulic hydraulic ratings will frequently not flag results that are outside the range of applicability of the applied correlations. Check with the simulation package vendor to identify these limitations. When proceeding beyond the initial estimate stage, it is recommended recommended that the internals vendor verify the internal ratings.

10

Economic Evaluations & Sensitivity Analyses One of the major benefits of using a simulation package is the ability to experiment and perform sensitivity analyses with various column operations and configurations in an effort to optimize the process. Simulations can be used to evaluate the effect of varying the order of columns in a multi –column process. Evaluations of reflux ratio versus required stages are also very useful. This can be used to determine the optimum column size and internals required. Another good analysis is to review the effect of reflux and internal efficiencies efficiencies with product specifications. specifications. Studies can also be conducted conducted to determine determine optimum feed location location and temperature temperature as well as pressure minimization. minimization. These reviews will help the engineer design an optimized column with maximum operating flexibility. flexibility.

Page 8 of 9

Issued: 08/04/2003 Revised: 10/10/2008

11


3.50

Bibliography 1.

AIChE Equipment Testing Procedure. Tray Distillation Columns. A Guide to Performance Evaluation. 2nd Edition 1987

2.

Pilling, M.W. and P. Mannion, “Simulation Aid to Main Fractionator Expansion”, Petroleum Technology Quarterly, 2(2) pp. 37-43, Summer 1997

3.

Perry, R.H., Green, D.W., and Maloney, J.O., “Perry’s Chemical Engineers’ Handbook”, 7th Edition, McGraw-Hill, 1997.

4.

“Theoretical Trays and Efficiencies in Hysim”, Hyprotech Technical Updat

5.

Ryan, J. M., Chang-Li Hsieh, and M.S. Sivasubramanian, “Predict Misting and Bubbling in Towers”, Chemical Engineering Progress, pp. 83-90, August 1994.

Page 9 of 9


ACCUMULATION IN DISTILLATION COLUMNS – MIDDLE BOILERS

Issued:

10/06/2011

3.61

Revised:

ACCUMULATION IN DISTILLATION COLUMNS – MIDDLE BOILERS

ACCUMULATION IN DISTILLATION COLUMNS – MIDDLE BOILERS ....................... 1 1. Introduction ......................................................................................................................... 2 2. Middle Boiler Accumulation Mechanisms and Symptoms ........................................... ...... 2 3. Enablers of Middle Boiler Accumulation ................................................. ........................... 3 3.1 Tower Controls .......................................................................................................................... 3 3.2 Increased Staging ....................................................................................................................... 3 3.3 Liquid Accumulation From Sources Other Than Middle Boilers.................................... .......... 4

4. Diagnosing and Detecting Middle Boiler Accumulation ............................................... ..... 4 4.1 Detection .................................................................................................................................... 4 4.2 Diagnosing ................................................................................................................................. 5

5. Middle Boiler Accumulation Mitigation ............................................... .............................. 5 6. References .................................................. ......................................................................... 6

Page 1 of 6


1

ACCUMULA TION IN DISTILL ATION COLUMNS MIDDLE BOILERS

3.61

Introduction

Middle boilers are common throughout the chemical and refining industries. They have been known to exist for decades in such common separations as argon in air separation, methyl-acetylene in C 3 Splitters, Green Oil in C 2 Splitters, Fusels in Alcohol rectification and water in FCC Deethanizer/Strippers. In established processes like these, the middle boilers are expected. Process designs take them into account, and they therefore cause little difficulty. Unexpected middle boilers, however, can be very troublesome. This design practice is intended to help the engineer deal with previously unknown middle boilers that disrupt tower performance and capacity. Accumulation of middle boilers in distillation columns is a result of the inability to remove the component(s) from the top or bottom of the tower. The buildup can result in a number of transient problems, including regular periodic temperature swings, or pressure drop and/or composition spikes. These can be accompanied by the column briefly flooding and then discharging (hiccupping) the liquid to the overhead or to the base. After the hiccup, the process begins again. The first step in correcting liquid accumulation in a distillation column is to identify which mechanism is responsible. Appropriate measures can then be implemented to mitigate the problem.

2

Middle Boiler Accumulation Mechanisms and Symptoms

There are several sources for middle boilers. The most common is when a mid-boiling impurity is present in the feed in trace amounts. This trace impurity boils at a temperature lower than the major component in the bottoms product, and higher than the major component in the overhead product. If the overhead and bottoms products are pure, the impurity gets trapped, and its concentration builds up over time until it can break out of that trap. Possible sources of middle boilers include: 1. True middle boiling impurities. 2. Middle boiling impurities due to non ideal behavior 3. Unbalanced azeotropes (excess entrainer) in azeotropic distillation 4. Impurities forming middle boiling azeotropes with one of the product components 5. A reaction of an impurity with one of the main components forming a middle boiler The main symptom of middle boiler accumulation is that, over time, at regular intervals, the impurity accumulates until it causes liquid to stack up in the column. It may then overflow the top or overcome the vapor traffic and suddenly go out the bottom of the column. The main point here is that if the total feed rate and composition of the middle boiling impurity to the column are constant then the rate of accumulation will be constant resulting in periodic column upsets. Factors that interact with middle boiler accumulation and aggravate the situation: 1. Freezing (i.e. formation of hydrates) of middle boilers 2. Foaming 3. Control schemes With each of these interacting factors the system may lose its ability to recover from the upset by itself. In C2 Splitters, water (if present) will azeotrope with the hydrocarbons until it finds the tray with a cold enough temperature to freeze or form hydrates (1). This can prematurely flood the tower by plugging the

Page 2 of 6

Issued: 10/06/2011

ACCUMULA TION IN DISTILL ATION COLUMNS

Revised:

MIDDLE BOILERS

3.61

open area or downcomers. Another example would be an FCC (or coker) absorber/deethanizer where water can build up in the tower and it cannot go overhead because of the cold temperature and cannot exit the bottom because of the hot temperature. This can result in corrosion (2), internals damage (3), potentially two phases and Ross (4,5,6) type foaming.

3

Enablers of Middle Boiler Accumulation 3.1

Tower Controls

Tower control systems can promote accumulation of mid-boilers, especially in systems where relatively pure components are being produced in the overhead and bottoms. Usually, this will have one product on composition control using temperature to control the steam or reflux with the other stream on flow control cascaded to temperature. In the case of azeotropic distillation there may also be a material balance control for reflux to satisfy the azeotrope. Control systems can also be misled by middle boiler accumulation. Systems which control a temperature in a column are really trying to hold a composition profile. Middle boilers can change what the composition actually is, at that given temperature. As a hypothetical example, consider a binary mixture at 15 bar pressure of propylene (Tb =36ºC) and Iso-butane (Tb=85ºC). The dew point of a 50/50 molar mixture is 65ºC. If a middle boiler such as propane (Tb=44ºC) is introduced, that 65ºC temperature no longer represents a unique composition. A mixture of 18% propylene, 37% propane and 45% Iso-butane would have the same dew point, as would many other combinations. Control systems can also be adversely affected when the middle boiler has different physical properties than the main constituents. For example, differences in latent heat of vaporization can change vapor traffic in columns, and have an impact on pressure drop or approach to flood. Impurities have also been known to cause foaminess in otherwise well-behaved systems, which can lead to control instability and premature flood. 3.2

Increased Staging

Increasing staging by adding more trays or packing or improving the performance of existing equipment to improve separation efficiency (higher product purities) can also bring about trapping of middle boilers that were previously able to escape. In one large absorber column in a chemical plant, the top trays in a column were replaced with structured packing. This met the project goal of improved absorption efficiency, but also reduced overhead entrainment. In many situations this would be a good thing, but in practice the entrainment had been the purge point for trace amounts of an acid in the feed. After the retrofit, the concentration of acid built up enough to significantly increase the corrosion rate in the column. In another instance, trays were added to a column to reduce the water content of an overhead product. This had the side effect of increasing the concentration of some nitrile impurities in the column. Even though they were not fully trapped, the increase in concentration and extra residence time allowed them to hydrolyze, forming carboxylic acid impurities. These acids significantly increased the corrosion rate in the column.

Page 3 of 6


3.3

ACCUMULA TION IN DISTILL ATION COLUMNS MIDDLE BOILERS

3.61

Liquid Accumulation From Sources Other Than Middle Boilers

In some cases, it is possible to get regular periodic loading and dumping behavior in columns, even when no middle boilers are present. It is important to recognize the circumstances when this is likely to occur, as it requires entirely different corrective action strategies than when middle boilers are the cause. One situation in which liquid accumulation and dumping is likely to occur when there are abrupt changes in the hydraulic capacity of column internals when going from the rectification to the stripping section. If the components have widely different latent heats, and the control system allows the lower latent heat component to move into the more restrictive hydraulic section, then liquid will accumulate resulting in flooding. Adjusting the temperature (or analyzer) set point to move the lower latent heat component will solve this situation. For example, this can occur in systems separating high latent heat water from low latent heat organic compounds to achieve high purity water in the bottoms and high purity organic in the overhead. In this situation there will be major differences in needed hydraulic capacity in the stripping section versus the rectification section. If the control system allows the organic profile to slip down the column then flooding could occur. This can happen with overly aggressive advanced control systems. A poor control philosophy can lead to a mass, component, or heat imbalance resulting in liquid or component accumulation in the tower. A poor temperature control point or lack of adequate temperature control can also lead to instability in the system which appears somewhat similar to liquid or component accumulation. Another possible source of periodic loading and dumping behavior is when fouling or some other mechanism severely reduces the open area of trays or column internals. This mechanism is typically characterized by extremely high pressure drops. In one example of this, a site allowed very hard water to leak into the reflux drum of a column designed to remove small amounts of water from an organics stream. Over time, the hardness totally plugged the openings of the bubble-cap trays in the top of the column. Vapor could only pass up through the downcomers on the trays. The high pressure drop associated with this caused liquid to build up on the trays until it reached a high level, at which point it rapidly dropped, then began building again.

4

Diagnosing and Detecting Middle Boiler Accumulation 4.1

Detection

Detection of liquid accumulation can be done as follows: 1. Examine process data such as pressure drop. The pressure drop will build in regular cycles until the column can no longer hydraulically support the liquid and it will briefly flood sending large quantities of the liquid overhead or out the bottoms. The pressure drop builds although the steam or heat input rate, reflux, and feed are constant. In this case, the column seems to flood at regular but lengthy intervals. The length depends on the concentration of the middle boiler in the feed and how fast it accumulates in the tower. This length can range between a few hours to several days. In between the flooding incidents, the tower will behave normally. A hydraulic analysis may reveal that Page 4 of 6

Issued: 10/06/2011

ACCUMULA TION IN DISTILL ATION COLUMNS

Revised:

MIDDLE BOILERS

2. 3.

4. 5.

6.

4.2

3.61

the column is not close to flood at the operating conditions. Study the temperature profile over a period of time for evidence of a significant change for no apparent reason. This temperature change is consistent and cyclic. Sample the column feed and product streams to perform a material balance which may show accumulation of a particular component. Even if the component is below the detection limit it can still cause a problem if its flow is high enough. Sample intermediate points on the column. Simulate the column using a process simulator with verified physical property package for all components found during sampling, and then plot the composition profile to find compositional bulges. In the case of a middle boiling component the simulation may be very difficult to converge. Trick the simulation by artificially increasing (or decreasing) the middle boiling component concentration in the feed, or by artificially taking a side draw, to facilitate a converged solution. Initial estimates of temperatures and compositions many times are keys to attaining a converged solution. Gamma scan the column to find the middle boiler accumulation area which may show up as an intermediate flood location different than expected. Repeat gamma scans to help detect the middle boiler buildup especially during cyclic operation.

Diagnosing

Samples taken from the column will confirm the existence of components which may be middle boilers. If possible, change the control point for a period of time to a lower purity specification at one end of the column. This will allow the middle boiler to escape, reversing the accumulation and reducing the buildup. In the case of masquerading middle boiler accumulation such as in the separation of water as a high boiler from a low latent heat organic, the internals are usually designed with very different hydraulics in the stripping section versus the rectification section. If the composition control for the column bottoms is temperature resetting the steam rate, then increasing the temperature set point moves the organic profile up the column. Initially, this will cause the steam rate and pressure drop to increase. As soon as the organic profile moves out of the hydraulically restrictive area in the stripping section then the pressure drop will decrease significantly.

5

Middle Boiler Accumulation Mitigation

Middle boiler accumulation can be overcome. There are several methods listed below: 1. Change the control to allow the middle boiling component to leave with either the overhead or bottoms product. This could also be accomplished by changes in operation, such as reducing the reflux ratio. This method requires a compromise that may be unacceptable to operations or the ultimate customer because the purity of either the bottoms or overhead product is reduced. 2. Remove the middle boiling component from the feed. 3. Add a side draw and remove the middle boiling component. The side draw rate does not have to be large, but it must be high enough to maintain material balance of the middle boiling component. The location, phase, rate and number of side draws can be determined from simulation results. Where practical, a number of extra side draw points should be installed to account for simulation inaccuracies. 4. In the case of azeotropic distillation ensure the entrainer material balance is satisfied. This needs to be accounted for in the control system.

Page 5 of 6

Issued: 08/04/2003 Revised: 10/10/2008

1


3.50

Introduction Process simulation packages are an invaluable tool for simulating process operations. When used properly, a simulation will accurately model a process so that output information can be used to calculate operating conditions that are otherwise unmeasurable, such as internal column loadings and physical properties. The bases used to converge all process simulations are Vapor-Liquid Equilibrium (VLE) data and the heat and material balances. The calculated compositions of the vapor and liquid are based on the specified VLE data. The heat balance is calculated using a specified thermodynamic equation package. Therefore, the final simulation results are only as good as the packages that were selected and the data that were input. Although process simulations are used to simulate entire operating units, the scope of this section will be limited to column operations. The basic steps required to perform a process simulation are: • • • • • •

Review of Existing or Future Equipment to be Simulated Data Gathering Selection of the Thermodynamic and Vapor – Liquid Equilibrium (VLE) Packages Construction of the Model Validation of the Model Interpretation of Results

All of these elements are critical to the success of a simulation. The ideal simulation will match all aspects of how the column actually operates or will operate. If the basis for the simulation construction is not representative of the actual column operation and design, the simulation cannot be accurate.

2

Review of Existing Equipment and Process Flows The first step of the simulation process is to make sure that the process flows are understood and accurately represented. All feed, product, and recycle streams as well as energy sources need to be identified and accounted for. It is most important to note at what tray or between what beds a feed or product is located. Depending upon what the simulation is to be used for, minor streams may or may not need to be included in the simulation. However, when excluding minor streams, you must be very careful to ensure that this omission is truly inconsequential.

3.

Data Gathering Data gathering is extremely important when constructing a simulation. Errors in data gathering can lead to simulations that will not converge or even worse, simulations that do converge and produce incorrect outputs that go undetected. One of the most important steps for gathering data is planning. During a hectic test run, it is often difficult, if not impossible, to get additional samples or data points taken. The fewer unexpected requests during the test run, the more smoothly and accurately the data will be collected. Safety considerations must always be carefully evaluated when sampling streams that are at high pressures and temperatures or are flammable or toxic. Proper sampling technique must be observed when collecting samples. For further information on proper sampling procedures refer to section 2.08 of this manual or the AIChE equipment testing procedures.1 A list of the typical data required for a column simulation is shown below.

Page 2 of 9

Issued: 08/04/2003 Revised: 10/10/2008


3.50

Temperatures & Pressures Overhead Receiver Column Overhead Any Column Midpoints Available Feeds Bottoms Heaters/Exchangers inlets and outlets • • • • • •

Compositions Feeds Products/Product Specifications Column Tray or Bed Samples Heating or Cooling Streams Heater Fuel Streams • • • • •

Flows • • • •

Feeds Products Flows to Heaters/Exchangers Reflux and Pumparound

For new columns, these values will be defined as part of the design scope. F eeds and product specifications should be defined there. If the simulation is being used for a feasibility study, perhaps a range of feed and product rates and specifications will be used to determine an economic optimum. Temperatures are typically set by available heating and cooling sources. Pressures are normally set by the temperature and composition requirements. Unless the process being simulated is uncommon, it is a good idea to look at the operation of similar columns to review typical operating conditions. For column revamps, operating data can be gathered from the existing column operation. This can often be accomplished effectively with the use of a distributed control system (DCS). Collecting data from a DCS will normally require a tag list that will be used to download pertinent data on a periodic or time averaged basis. It is important to have up to date Process and Instrumentation Diagrams (P&IDs) and master tag lists from which to construct the test run tag list. In some data logging systems, data which are not included beforehand on a test run tag list cannot be retrieved afterwards and are therefore lost. Be sure that your tag list is complete and will provide you with enough data to construct a full heat and material balance around the column. When in doubt as to whether or not a tag is relevant, include it on your list.

3.1

Sample Collection & Lab Data When collecting data, it is imperative that the column be running at steady state conditions when a data set is collected. It is always important to know what the response time of the column is when process changes are being made. Smaller columns may equilibrate to a process change in a matter of minutes while larger columns may take several hours. Accurate data cannot be collected until the column has had time to equilibrate. The response time will be a function of the column size, liquid holdup, instrumentation, and complexity. The best way to determine the column response time of an existing column is to talk to the operators and then verify this by making a moderate change to the column and observing the trends. When available, a dynamic simulation can sometimes be used to model the response time.

3.2

Heat and Material Balance When taking data, it is important to also include heat balance data to cross check the rest of the column information. Without a heat balance, there is no way to verify the reflux or pumparound flow

Page 3 of 9

Issued: 08/04/2003 Revised: 10/10/2008


3.50

rates. This information is extremely important with multi-draw columns where internal loadings change significantly in different sections of the tower.

4

Selection of Property Packages Existing VLE correlation packages may be modified with research data. These data can be found in literature or may be available from the simulation vendor. Many large operating companies have in-house VLE data. Of the various process simulation software packages available, some are better suited for different operating systems than others. Most packages handle ideal hydrocarbon systems readily. Non-ideal, aqueous, and electrolytic systems can be more difficult to simulate. Proprietary chemicals can also be difficult to simulate due to a lack of published physical property data. When dealing with these systems, more accurate results can usually be obtained by using externally gathered data. Some chemical systems have simulation packages designed specifically for them. For example, there are several specialized simulation and property packages available to deal with amine systems . Many process simulation software packages provide an oil package to evaluate hydrocarbons based on laboratory distillations, densities, UOP “K factor”, viscosities, and other available properties. These oil packages then divide the hydrocarbon streams into pseudo components and then perform the simulation with these components. 4.1

Non-Ideality Equation of state models such as Peng-Robinson, SRK, etc., are reliable in predicting properties of ideal systems but are limited in their ability to handle non-ideal or polar systems. These systems normally require the use of a dual model system where an equation of state is used to predict the vapor fugacity components and an activity model is used for the liquid phase. However, activity models are much more empirical in nature. Therefore, more caution should be used when selecting these models for your simulation.

4.2

Dual Property Packages In some columns, the equilibrium conditions in the top section of the tower can be different enough from the bottom section to warrant using different VLE packages in each section. In this case, the column sections are initially modeled separately and linked together in the simulation using recycle streams. This technique can be used to obtain more representative simulations. 2

4.3

Azeotropes Care must be used when simulating systems with azeotropic components. Because of their nature, these systems present a severe test for simulation programs. Residue curves, which have been around since the early 1900’s, can be used effectively to deal with azeotropes. A residue curve is a plot of the liquid phase composition as it varies with time during a simple distillation. Residue curve maps can be pictured graphically for up to four components. Residue curves can be constructed from experimental data or can be calculated analytically if equations of state or activity coefficient expressions are available. Some examples of residue curves and general information on how to use them can be found in Perry’s Handbook.3 The use of a residue curve map of the azeotropic components can provide a good indication of what separations are possible.

4.4

Checking of VLE Performance Once a VLE package has been selected, it is good practice to print out a vapor-liquid plot at conditions similar to available reference data to ensure that it corresponds well, particularly if one is using default interaction parameters from the simulation package.

Page 4 of 9

Issued: 08/04/2003 Revised: 10/10/2008

5


3.50

Model Construction Considerations There are various factors to be considered for the actual construction of the model. Once again, the primary concern with the model construction is accuracy of the model with respect to the actual column operation. Care must be taken to ensure that the scope and complexity of the model is such that it will provide the full range of desired information. Various factors pertinent to the model construction are discussed below. 5.1

Stage Efficiencies Many simulation programs allow stage efficiencies to be used to simulate the actual tray efficiencies. This technique is convenient, but can produce incorrect results if not used properly. When stage efficiencies are used, the resulting simulation will often indicate that the heavier components will carry higher up in the column than in actual operation. This will also lead to an increase in predicted 4 stage temperatures . Because of this, the use of tray efficiencies is generally not recommended. If stage efficiencies are used, care should be taken not to apply them on stages that contain product draws. It should also be noted that rate based simulations rely on calculations of interphase mass transfer for actual internals. Since they do not use the concept of theoretical stages, the concept of tray efficiency does not apply

5.2

Equilibrium Stage versus Rate Based Calculations The vast majority of process simulations are conducted using an equilibrium stage model. This method models the column as a stack of equilibrium stages or theoretical plates. The key assumptions required for this model are perfect mixing and thermodynamic equilibrium. In reality, these assumptions are never truly accurate. As a result, the concepts of tray efficiency and HETP are used to account for less than ideal mixing and departure from equilibrium. Because of the above-mentioned limitations, a rate based method may be better suited for the simulation of some processes. This method models the column as a group of segments where equilibrium is assumed only at the vapor-liquid interface. These segments can be used to represent a tray or a portion of a packed bed of a specified height. With the rate based approach, the degree of separation achieved depends on the rates of mass and heat transfer between phases contacted on a tray or section of packing. The advantages of the rate based method are that it eliminates the uncertainty of using stage efficiencies and HETP, predicts the departure from temperature equilibrium, and can account for interactions between diffusing components. Rate based simulations are recommended for nonideal multicomponent systems, reactive distillation systems, or any system where there is an interest in the distribution of non key components. Specific cases where rate base simulation can be beneficial are applications where mists or bubbling may form from the presence of vapors at temperatures below their dew point or liquids at temperatures above their bubble point. 5 With respect to tower internals, rate based simulations have their greatest advantage with packed towers because in packed towers, there is no limitation on the number of stages into which the tower can be divided. Rate based simulations have a major advantage where conditions are changing rapidly, especially in short beds where fractions of a theoretical stage are important. In depth questions regarding the applicability of certain processes for rate based modeling should be addressed to the simulation software vendors.

Page 5 of 9

Issued: 08/04/2003 Revised: 10/10/2008

5.3


3.50

Graphical Solutions – Graphical Troubleshooting, Composition Profiles Most simulation packages will allow you to print out McCabe-Thiele diagrams of columns as well as reflux ratio versus theoretical stages. These diagrams provide a graphical view of the column operation and can many times help identify possible problems or limitations of the column operation. These graphs are particularly useful for identifying column pinch points and approaches to infinite reflux ratio or theoretical stages. A plot of stages versus reflux ratio is an invaluable tool that can be used to ensure that the column design is optimized with respect to energy cost and capital investment as well as indicate how sensitive the design is to a loss of theoretical stages.

5.4

Dynamic Simulation Dynamic simulation simulation can be a useful tool for simulating tower operations that are not considered steady state. Dynamic simulation has been s uccessfully used for reviewing relief scenarios for distillation columns and also for control analysis. Instrument perturbation simulations simulations can be useful for evaluating the instrumentation instrumentation performance. Dynamic Simulations may help identify control problems that have have substantial impact impact on performance performance of tower internals. internals.

5.5

Feeds and Draws The simulated introduction introduction of a mixed feed to a column can be a source of uncertainty. This is an important issue because it determines determines what the vapor and liquid loadings will be above and below that point. The standard standard handling of mixed phase phase feeds is for the vapor portion portion to be mixed with the the vapor stream from the stage below the feed stage and the liquid portion to be mixed with the liquid from the stage above. These combined streams are then mixed on the feed stage. An example of this is shown below.

Page 6 of 9

Issued: 08/04/2003 Revised: 10/10/2008


3.50

Superheated and subcooled feeds produce unique loading conditions within a column. A superheated feed, which is commonly found in an FCC Main Fractionator, will produce a “vapor bulge” in the stages above the feed. This is the result the superheated vapor heat causing a greater than mole for mole vaporization of the liquid that it contacts. The number of stages in this section of the tower must be correctly specified specified so that the bulge appears in the simulated simulated tower loadings. loadings. One method of accounting for the bulge is to first simulate the tower with the correct number of stages required to match the actual or desired tower operation and then add enough stages to a second simulation to produce the bulge loadings. The first simulation simulation will give correct correct loadings for all sections sections of the tower except for the bulge section. The second simulation will be used only to get the hydraulic loads in the bulge section. In the example example case of the the FCC Main Fractionator, Fractionator, two stages are normally normally required for the actual simulation and three stages are required to properly yield the bulge loadings. Internals in the bulge section of the the tower must be sized to to adequately handle handle these loads. Correspondingly, Correspondingly, a subcooled liquid feed or reflux will produce a liquid bulge in the stages below. This too, must be accounted for when designing the tower internals. 5.6

Reboilers, Condensers, Pumparounds, Heat Pumps When constructing or reviewing simulations, the reboiler and condenser operations need to be evaluated to determine as to whether or not they constitute a full equilibrium stage and should be modeled accordingly. Condensers must be specified as total, partial, or subcooled. More detailed information on this subject can be found in section 4.03 of this manual entitled , Exchangers in Fractionation Services. In some cases, the reboilers and condensers are simulated as an integral part of the tower operation. In others, the reboiler and or condenser can be more accurately modeled as a separate piece of equipment that is operating in conjunction with the tower operation. Pumparounds are sometimes treated simply as a side exchanger attached to a particular theoretical theoretical stage. This may be convenient to the simulator but it provides nothing for the process engineer working up the design of the pumparound circuit. A more thorough way of simulating a pumparound is to withdraw the fluid out of the tower, provide the necessary heat exchange and place it back in the tower as another feed. This obviously is more complex and may result in considerable convergence convergence difficulties, but the results will be meaningful. The addition of a heat pump to a column adds another degree of complexity to the simulation. With these configurations, special emphasis must be placed on the integrated system. The compressor operation including the compression ratio, hydraulics, and heat sink must be thoroughly evaluated to ensure that it functions efficiently with the column operation.

6

Model Validation Once the simulation has been run and converged, it should be validated against other data. In most cases, there will be an existing column or other s imilar operations, which can be used as the basis for the validation. The model should predict, as closely as possible, the correct composition, temperature and pressure profiles, and energy requirements. Variances between the model and the actual operation or predicted operation need to be reconciled. If possible, a simulation validation validation for more than one operating point will provide more confidence in the applicability of the simulation over a range of operations. This is especially important important when extrapolating data for revamp cases.

7

Interpreting Results Column internal loadings can be generated from the validated simulation and used for rating of the internals. Most simulations print out the physical and transport properties of the vapor and liquid stream leaving each stage. When using these data for internal ratings, it is important to remember to select the liquid stream leaving

Page 7 of 9

Issued: 08/04/2003 Revised: 10/10/2008


3.50

one stage and the vapor stream leaving the stage below. However, the designer s hould be aware that loadings are occasionally increased across a given theoretical stage. The designer must then prorate the loadings to and from the theoretical stage in question. Flashing feeds, for example, may artificially increase a stage’s loads when in reality the extra load is above it. When working with reboilers and condensers, you must identify the flows shown that are outside the actual column and not incorporate them in the rating process. Be careful to verify the tray numbering system versus the simulation numbering system. Most simulations number stages from top to bottom with the condenser being stage number 1 and the reboiler being the last stage. By convention, actual column trays are numbered without consideration of condensers and many columns are numbered from bottom to top. Pumparound sections can have their own nuances. These sections are typically situated in a tower to simply achieve heat transfer. Two or three theoretical stages are needed to just to establish (be far enough away from the feed and withdraw trays to see) what the liquid and vapor loads are in this section of the tower. In addition, the mass transfer rate and the heat transfer rate are typically different in a real tower. Care should be taken to ensure that the material balance is closed and the temperatures should be checked in a pumparound section.

8

Selecting Data for Rating Internals When selecting data for rating of internals, review the entire column loadings to get a feel for variances in the internals loads. Check vapor and liquid volumetric flow rates and physical properties throughout the column. Stages with significant variances in transport properties will require additional ratings account for these changes. Normal practice is to select high and low loading cases for each similar section of the column to rate or size the internals. Some simulation packages will do this automatically. The next step is to adjust the simulation or flows to account for desired spare capacity or turndown. In some cases the turndown can just be simulated as 50% of design flow or whatever criteria is required. In some cases, other factors may require that a separate turndown case be simulated. For Reflux distributors it is important to size them for the proper temperature. If a reflux is subcooled, then the design of the distributor should be made using the subcooled reflux temperature. temperature.

9

Internal Simulation Package Rating Programs Rating programs for column internals are a feature provided by most simulation packages. These rating programs can be useful for initial initial sizing but often often have limited range range of applicability. applicability. Simulation hydraulic hydraulic ratings will frequently not flag results that are outside the range of applicability of the applied correlations. Check with the simulation package vendor to identify these limitations. When proceeding beyond the initial estimate stage, it is recommended recommended that the internals vendor verify the internal ratings.

10

Economic Evaluations & Sensitivity Analyses One of the major benefits of using a simulation package is the ability to experiment and perform sensitivity analyses with various column operations and configurations in an effort to optimize the process. Simulations can be used to evaluate the effect of varying the order of columns in a multi –column process. Evaluations of reflux ratio versus required stages are also very useful. This can be used to determine the optimum column size and internals required. Another good analysis is to review the effect of reflux and internal efficiencies efficiencies with product specifications. specifications. Studies can also be conducted conducted to determine determine optimum feed location location and temperature temperature as well as pressure minimization. minimization. These reviews will help the engineer design an optimized column with maximum operating flexibility. flexibility.

Page 8 of 9

FRI Internal Design vol 3

Recommend Documents