TEG Dehydration Design Basis and Design Guidelines

Gas Dehydration Pretreatment This should include adequate vapor/liquid separation and solids filtration. I feel stronger about the adequate vapor/liquid separation than the solids filtration. If you have some liquid formation, that should help in the washing out of solids. If your operation has very little condensation upstream, then the need for solids filtration is stronger. The need for good upstream vapor liquid separation cannot be over emphasized. If you have liquid hydrocarbon entrainment into your system, the entire operation will be negatively impacted. If you have a trayed column, it will foam, with the accompanying heavy entrainment losses and lower column efficiency. The liquid hydrocarbon will carry through to the flash drum, also causing potential foaming problems there, and possible overloading the flash drum overhead system. With flashing hydrocarbons, the amount of gas coming off in the flash drum will be many times the normal operating rate. If you have a good flash drum it may be able to remove the liquid hydrocarbon here, if not it is off to the reboiler, to be vaporized there and possible overload the gas system there. Not all the hydrocarbon will be vaporized. The heavy ends will form a tar like substance to foul your surge drum, lean/rich exchanger and filters. It is far easier to do a good job upstream on the original vapor liquid separation, than to deal with the far more difficult job of liquid-liquid separation. The upstream vapor liquid separator should be sized to handle more gas than the glycol contactor can handle. Eventually operations will push the system to a gas rate it cannot handle. May as well let the thing which fails be less annoying. If the upstream separator fails first the operation will require much attention, and have heavy TEG losses. The upstream separator should be sized to handle 20% more gas the rest of the TEG system. Also the separator should be simple in nature, and less likely to fail. A vertical separator, with a horizontal mesh pad fits this criteria. Putting in an extra thick pad will help reduce the entrainment load. A 12 inch, multidensity wire mesh pad should do good. With a dp cell across the mesh pad to warn of its flooding. The TEG Contactor can be either structure packing or bubble cap trays. I dont think having an integral vapor-liquid separator below the contactor is the most reliable system. A failure of the seals between the glycol contactor and the vapor-liquid separator can result in losses of TEG, which would be hard to trouble shoot. Also now with the use of structured packing the capacity of the vapor liquid separator will not be as high as the TEG Contactor. Remember the liquid density in the vapor liquid separator will be lower than the TEG, resulting in a lower allowable velocity. The only enhancement I would make is a thicker pad and make it composite (6 inches of type 421 and 6 inches of type 326). A gas filter downstream of the knockout drum may be nice, but I am not sure it is that necessary. If you do install one, do not count on it to also remove liquids. They typically are designed for a very high V(150) and do not reliable remove liquid. A good vapor/liquid separator is absolutely essential for the TEG unit to operate well and trouble free. The allowable velocity calculated for the mesh pad should be based on the vapor density and the liquid hydrocarbon density (not the water density or the average density of the liquid hydrocarbon and liquid water) The maximum inlet temperature to the TEG system should be considered carefully. Just going from 90F to 100F can increase the water load on the reboiler system by 35%. Also the obtainable dewpoint will be effected directly (10F higher)

TEG Contactor If structured packing is used here it should be designed for a maximum Fs of 2.8. Internals for the contactor should include:

York Reid Demister

This is a Dacron/316SS composite mesh pad designed specifically for TEG Dehydration service. I recommend a 16 inch thick pad, with 1 inch top and bottom grids. The contactor should have top and bottom rings to hold the mesh pad in place. Omission of the top ring sometimes results in loss of the mesh pad and increased TEG losses. The mesh should also be wired into place on the bottom ring. The top of the mesh pad needs to be far enough away from the gas outlet nozzle to prevent channeling. The distance is equivalent to drawing a 45 line from the outer edge of the mesh pad to the outlet nozzle. This results in a vertical distance = (mesh pad diameter - outlet Gas diameter)/2 The bottom of the mesh pad should be at least 3 feet above the top of the TNT 727 distributor (not counting the parting box).

Inlet TEG Distributor pipe

This is merely an internal extension of the lean TEG line to the contactor. It is routed to the parting box above the TEG Distributor

Glitsch TNT 727 Narrow Trough Distributor (316SS), with parting box

This will provide the proper distribution of the TEG to the structured packing. It is constructed with small orifices for distribution. If the orifices plug, the TEG will rise and flow into the same drip tube via a V notch. The drip tube can be extended to be close to the packing surface. The final connection to the packing should be wire gauze which is tack welded to the drip tubes of the TNT 727 distributor. Glitsch can provide their distributor already configured in this manner. It is important that the liquid glycol does not free fall to the packing surface. If it is allowed to free fall, entrainment into the gas phase can take place, with a corresponding increase in glycol losses. Drip points/ft 2 should equal 4.

Glitsch HDG 421 Packing Holddown Glitsch HPD 202 B High Performance Distributor

This is structured packing which is in 1/2 the normal thickness of 9.75 inches. Two layers of 4.875 inch thickness are used to speed up the distribution of the glycol. To prevent flooding and entrainment from the top of the packing, Glitsch normally uses one size larger of structured packing. In this case 1/2 inch crimp instead of the 1/3 inch crimp of the packed bed. I also recommend using a 60 crimp angle to further decrease the gas velocity at the top of the bed. This reduces the velocity by over 20%, which should in turn reduce the tendency to entrain.

The Packed Section

21.25 feet of Gempak 3A structured packing. The original PBU contactors had their bubble cap trays replaced with about 17 feet of packing.

Glitsch HPS 121 packing support Inlet gas distributor

Im not terribly knowledgeable on inlet gas distributor requirements. I have seen contactors with no distributor, half pipes, spreader pipe style distributors and chimney trays. I think if your inlet velocity is reasonable (in the area of 50 feet/sec) you should be able to use a half pipe, located about two feet from the bottom of the packing. Glitsch may have some comments in this area.

Contactor Surge Volume

This volume should be about 20 minutes. This will take care of normal upsets, etc and also allow for some volume for the holdup of TEG in the contactor. I recall the liquid holdup being about 5% at the high design gas rates. By putting the surge volume control on the contactor you maintain a even flow of TEG to the

Reboiler and minimize upsets. Anytime you vary the flow to the reboiler, you stand the chance of changing the reboiler temperature. It is important to maintain the 400F reboiler temperature. Even small upsets to this can throw off the lean teg. The original system of using a surge drum under the reboiler, and allowing its level to float up and down, leaves the TEG pump subject to running dry and cavitating. Having a large reciprocating pump cavitating can be quite exciting.

Bubble Cap Tray Contactor If bubble cap trays are used, you just need to set the number of trays, and install a mesh pad. Again I recommend a York Reid Demister. I am aware of 12 foot diameter bubble tray tower operating at a C of 0.27 and a Demister K value of 0.3. The unit is operating well, except it is starting to carry over some glycol, which is causing trouble downstream in the Brazed Aluminum Heat Exchanger. The TEG freezes in the downstream heat exchanger, causing the pressure drop to increase. The frozen teg is removed by a high dosage of methanol. (15 gpm for 5 minutes). Please note the above quoted C is based on the tray area (with the downcomer area subtracted from the tower area). This is equivalent to about a C of 0.23 using the entire column area. The efficiency of the column is expected to fall off as the entrainment goes up. Since the total flow of glycol is very low for a contactor, even low entrainment rates between trays can dramatically hurt your efficiency. A 10 foot diameter column with 10 gpm internal entrainment has substantially decreased its efficiency since the total liquid rate to the column may only be 30 to 60 gpm. The entrainment of the rich teg from the tray below to the tray above, essentially makes the upper trays have richer teg, reducing their ability to dehydrate. TEG Regeneration System

Rich TEG/Still column Reflux exchanger This unit allows for some reflux (water) in the still column to reduce the TEG losses from the regeneration system. Typically you need to allow for about 20 degrees of temperature change in the rich teg to provide enough reflux in the still column to reduce the losses. I often see a temperature spec on the still column overhead. A temperature spec does not make a lot of sense if you are using stripping gas. Once you begin condensing water, the temperature of the overhead will remain relatively constant even with large changes in reflux duty. It is essentially a two component system: highly volatile gas and water. Once the gas rate is set, the overhead pressure of the still column, and the amount of water removed by the TEG system, then the mole fraction of water in the Still Column overhead gas is set. This essentially sets the overhead temperature, since the gas will be in equilibrium with liquid water on the top section of the Still Column. Once the reflux is high enough to reduce the TEG losses, you are only increasing the Reboiler Duty when you reflux more. Rich/Lean Heat Exchanger This heat exchanger adds more heat to the rich TEG to bring up the flash drum temperature to 150F. Running the flash drum temperature up, reduces any foaming tendency. This exchanger also recovers heat from the lean TEG. Flash Drum The purpose of this drum is to degas the TEG and allow for decanting of any liquid hydrocarbon. As mentioned earlier, decanting the liquid hydrocarbon is often difficult. If you do a poor job of it, you lose TEG as well. Best option is to not let any liquid hydrocarbons into your system. This unit runs best at a relatively warm temperature. Low pressure also reduces foaming. Another Lean/Rich heat Exchanger This final lean/rich heat exchanger is for heat recovery. You should expect to be able to get the rich TEG up to 275 to 300F. The remainder of the heat will have to be added by the reboiler. Reboiler Reboiler should be operated at 400F and low pressure (as close to atmospheric as feasible). I am familiar with units that have operated at these temperatures for over 20 years with no degradation problems. Running at a lower temperature will make it more difficult to get the required TEG purity. The temperature controller should be a PID type controller (proportional, integrating, derivative). On-off control of the TEG Reboiler temperature is totally unacceptable at the required TEG purity of this system. Fire tube flux rate can be designed for 8,000 Btu/hr-ft 2 . Again I am familiar with units that have operated for years at 10,000 Btu/hr-ft 2 with no significant problems . As I said the pressure of the reboiler should be low. This makes it easier to make high purity TEG. Higher operating pressures can be offset with longer packed sections in the stripping column and more stripping gas. I would not be concerned with running up to 10 or 12 psig. We had a unit which was designed to run at 1.5 psig, but a badly designed overhead condenser had a 10 psig pressure drop, resulting in a high reboiler pressure. The unit still made its 99.98wt% TEG spec as a result of a 8 foot long structured packing section and design stripping gas rate of 8 scf/gal

Stripping Column

The stripping column (the stand pipe that the glycol flows down from the Reboiler to the Surge Drum). I recommend using structured packing for the stripping column. The column will be highly liquid loaded (20 to 40 gpm/ft 2) The column should be designed with the upset liquid rates in mind. To avoid too large of an upset rate, do not over size the level control valve from the TEG Flash Drum. Structured packing is about twice as efficient as pall rings (see attached graph) and doesnt have the problem of packing leaving the column as pall rings sometimes do. I also recommend using 8 feet of packing. This allows for much easier attainment of the required lean TEG Concentration of 99.99 wt%. I have attached a graph depicting the TEG Reboiler/Stripping column performance for various numbers of theoretical trays. I have plotted the performance in terms of wt % H2O, instead of the usual wt % TEG, as it results in straighter lines, and are easier to read in the high TEG Concentrations. While it can be argued that you dont need 99.99 wt % , it really doesnt cost much to get the extra lean concentration once you are committed to putting in stripping column. The graph I have attached is based on a 2 psig reboiler pressure. This will likely require a blower. If your LP Compressor (say 5 psig), we should be able to operate at the higher pressure of 6 or 7 psig and compensate with more stripping column. The current specd 8 foot height might even do the job. The bottom of the stripping column should not extend far into the surge drum as is commonly the case. The column should end near the top of the surge drum. This will allow the stripping gas to flow from the surge drum to the bottom of the stripping column.

Stripping Gas This gas should be routed to the bottom of the stripping column via the surge drum. Do not route this gas pipe through the reboiler. Since the stripping column will be using structured packing the gas pipe can not be routed down the column as is commonly done. Also the gas can be preheated by routing it along the bottom of the surge drum in the liquid phase. A rotameter should be provided for metering of the stripping gas prior to entering the surge drum. A manual throttle valve down stream of the rotameter will be adequate for control. Surge Drum Flow to the Surge drum should be in one end and out the opposite end of the vessel to ensure good mixing and discourage static areas. This vessel should be used to protect the pumps. The majority of the system surge should be taken at the contactor, as explained earlier. Rich/Lean TEG Exchanger Traditionally these units were shell and tube. More common is now to see plate and frame heat exchangers. The plate and frame heat exchangers are much more compact. Still Column The purpose of the still column is to reduce the TEG losses. It achieves this by having 2 trays (bubble cap) above the rich TEG feed point. The liquid for the trays typically comes from an internal still column condenser (a coiled tubing exchanger). This provides a small flow of liquid water for reflux. This is generally enough since the boiling points between water and TEG is so large (546F versus 212F). Still Column Reflux Condenser

This is an area that has resulted in annoying TEG losses. The typical reflux condenser is a coiled tube. The tube can crack/fail and then rich TEG is sprayed into the outlet still column gas stream. If this system is still used, at least make it easy to pull the coil and replace it. Alternative overhead condensers should be considered which are more robust. I once had to trouble shoot a TEG system that was experiencing high losses. I did the usual inspection and found 20 volume % TEG in the Still Column Effluent Drum. I reviewed the stripping gas rate, etc to determine if the unit could be flooding, but it appeared alright. A performance test where I varied the amount of reflux in the overhead resulted in the losses increasing with increasing reflux. This is just the opposite of what one would expect. This operation ran parallel dehydration trains, so I repeated the test on the parallel unit and got the system response I originally expected. Shutdown and inspection of the reflux coil reviewed it to be cracked at the point it hangs from the top of the still column. This units can be difficult to reweld.

Still Column Effluent Condenser Not all units will have one of these. Frequently this is the end of the line for TEG units. The effluent just vents into the air. This is becoming more and more unexceptable for more units with concerns and regulations regarging BTEX ( Benzene, Tolulene, ethlybenzend and xylene). Typically units with gas rates 25 MMSCFD may come under the regulations. This condenser cools the effluent gas down from around 200 to 100F. Still Column Effluent KO Drum Same comments as above. This allows for the knock out of liquids, prior to routing the gas to a blower or compression system. Effluent Blower This unit recovers the stripping gas and any absorbed hydrocarbons and routes them else where in the plant for reuse. Some care is needed in specifying this unit. TEG absorbs a

lot of heavy hydrocarbons and CO 2 . If the stripping gas is turned off or reduced, the mw of the gas going to the effluent blower can change dramatically. I was starting up a unit once, and to tune the recycle system I reduced the system flow by turning the stripping gas way down. The recycle system responded by opening the recycle valve. Shortly after that the blower motor tripped on high amps. This particular blower did not have a pressure gauge downstream of it, so I did not notice what must have been a substantial increase in mw as a result of the reduction in stripping gas. I expect the mw went from 22 to around 36 or so. To make matters worse the breaker broke and we were out of commission for a day. I did not realize why the amps went up on reduced gas flow until considerable later.

TEG Additions

This can be a problematic area for TEG Systems. The TEG should be added upstream of the Regenerator as the storage TEG is no where close to 99.99 wt%. I recently contacted a lab regarding the concentration as they receive it. It was 99.7 wt %. The TEG also must be added slowly to the system, otherwise the Reboiler will suffer a significant drop in temperature, with a corresponding drop in lean TEG concentration. I recommend a normal addition rate of no more than 10% of the circulation rate. I have been running performance tests on TEG Systems, and noted 30 to 40F drops in the reboiler temperature, and find out the operator is adding teg at a rate similar to the normal circulation rate, greatly overloading the TEG Reboiler, and leaving the system upset for some time. Since I dont expect your TEG loss rate to exceed 0.1 gal/mmscf, if you add TEG weekly, it will take about one hour to make up the losses. A transfer pump may have to be included for high rates (or use the spare TEG Circulation pump).

Lab Tests Typical lab tests for the TEG Units are:

Water contents of Rich and Lean TEG - (Karl Fischer) dependent on your dewpoint spec PH of rich and lean TEG 6.8 to 7 on the lean particulates <1mg/liter Still column Effluent (Refractive Index) <1 volume % TEG Water content of Gas - Either Bureau of Mines Dewpoint Testor or Lockwood and McLorie Analyzer

On Line Analyzers The only on line analyzer I ever see on a TEG system is a moisture analyzer and they usually dont work well. The probes are easily fouled with TEG. Also they can be quite sensitive to pressure variations. I once was installing a new probe and thought it would be a good idea to pop and purge the sample loop to speed up the drying out of the sample system. Unfortunately the pop and purge cycle causing the gold foil on the probe to come apart. Another failed probe. My best experience with moisture analyzers for the gas is the Lockwood and McLorie analyzer. This is quite a complicated device that involves flowing a know quantity of gas through a packed column of glycerol. The glycerol column is taken to the Lab and eluted into a special chromatograph. The chromatograph output is two peaks on graph paper, which is measured either manually with a planimeter or electronically with an integrator. This area of the peak is compared with a standard which is injected into the chromatograph before and after every set of samples is run. The standard is a 7 micro liter sample of ethyl ether that was in equilibrium with liquid water. Knowing the temperature of the ethyl ether and the solubility of water in it, allow you to calibrate the area of the chromatograph peaks to micro grams of water. The end result is a direct measurement of the lb water /MMSCF of gas. A very accurate instrument, which is able to reliable measure down to 0.02 lb water/MMSCF. It does require a reasonable lab to run and is some what labor intensive. If you dont need this high accuracy a Bureau of Mines dewpoint testor does a very nice job. The Bureau of Mines Analyzer is the original dewpoint tester. This device routes a sample of your gas past a chilled mirror. By viewing the chilled mirror you can determine the dewpoint of your gas. The temperature of the chilled mirror is varied by running a refrigerant by it. This device is rugged and reasonable accurate, but requires a skilled operator and is not an online device. This device can be purchased from Chandler

Engineering, Ph: 918-250-7200; fax 918-459-0165. Cost runs about 3 to $5,000. An automatic/online version of the Bureau of Mines testor has been developed by Bovar Western Research. They can be contacted in Calgary, Canada at 403-235-8300.The cooling capacity of this device is 90F (50C) below the temperature at the monitor installation. The cost is about $25,000. Conversion of existing Bubble Cap Tray units to Structured Packing This can work out very nicely. You may increase your capacity as much as 100%. Of course you will be loading up your regeneration system somewhat. The majority of the heat load in a regen system is in the heating up of the glycol. In one of these conversion jobs, the water load of the rich TEG starts to add up. To compensate for this, you may reduce the TEG circulation rate. The structured packing needs a wetting rate of at least 0.3 gal/ft. Additional problem areas are if you have an integral vapor liquid knock out drum in the bottom of the contactor, it will likely limit your capacity. Conversion of the unit will require removal of all of the trays, and cutting of the tray support rings (dont get carried away, you will need the bottom ring to support the packing and the top ring for the packing hold down and liquid distributor). The very first installation of structured packing on the North Slope did not include the cutting off of the support rings. This was likely because structured packing was untested in glycol dehydration and removing the support rings would prevent the operator from reinstalling the trays. Unfortunately it also guaranteed the unit would fail. Leaving the support rings in place allowed wet gas to bypass the structured packing. When the inlet gas moisture content is 45 lb/MMSCF, and the outlet spec is 0.1 lb/MMSCF you dont need to bypass much gas to be off spec. The recommended procedure is to remove the support rings to 3/8 of the wall. Any downstream TEG Knock drum will likely need to be upgraded to handle the higher gas rates. I recommend the York Reid Demister for this service. Ideally you should handle the mist removal at the top of the contactor. I dont think you gain anything by exiting the contactor and shearing the teg entrainment particles into smaller particles to be removed downstream! Also proper use of a liquid distributor for the TEG and York Reid Demister should allow (if there is space in the contactor) you to recover the TEG in the Contactor, which is simpler