This presentation from 1974 was a landmark event in the Gas Processing industry.
The concept as well as the desirability of utilization of activated carbon filtration for purification of amine and glycol solutions has been recognized for some time. Most carbon filtration installations have been made as a result of "rule of thumb" design criteria, and considerably more art than technology was required to effectively predict the performance of an activated carbon filter.
by Charles R. Perry President Perry Gas Processors, Inc. Odessa, Texas copyright © 2004 Perry Management, Inc. Presented 1974 the Gas Conditioning Conference University of Oklahoma, Norman, Oklahoma The concept as well as the desirability of utilization of activated carbon filtration for purification of amine and glycol solutions has been recognized for some time. Most carbon filtration installations have been made as a result of "rule of thumb" design criteria, and considerably more art than technology was required to effectively predict the performance of an activated carbon filter. The utilization of diethanolamine for treating purposes in newer plants, as well as conversion of older plants, has resulted in a new sense of urgency to develop a solution purification process for these plants. Since reclaiming has not been successful on diethanolamine solution, filtration is the only readily available method of cleaning DEA solutions. Also dirty glycol solutions have been tolerated for a number of years. Glycol solutions can be reclaimed under vacuum, but this is very seldom done on small field units. Concurrent with the above, certain new processes and new techniques in activated carbon filtration have been developed in the last few years. It now appears that when properly installed and operated, activated carbon filtration may be used exclusively for purification of amine solutions. Activated carbon filtration of glycol systems is not as essential, but can be highly desirable and should be reviewed for possible application in those glycol units where the glycol has become contaminated. This paper will describe and categorize the known filtration processes involving activated carbon for amine and glycol purification.
ACTIVATED CARBON FILTRATION PROCESSES Adsorptive Only Processes In the original concept of using activated carbon filtration for amine and glycol systems, it appeared that conventional mechanical filters (such as "sock" type filter) would be utilized to remove the solids from the stream but not necessarily those contaminants which were in solution or in an emulsion form. These impurities would be removed by a fixed activated carbon bed installed downstream from the mechanical filters. In systems of this type, where the activated carbon was used merely as an adsorptive process to remove contaminants, the carbon bed would be of a relatively small size. Cross-sectional areas would normally be sufficient for flow rates of between 1 and 5 gpm per square foot, with bed depth approximately 3 to 10 feet. The carbon utilized would normally be a high bulk density, gas adsorption grade of carbon, either coke base or coconut shell base. Wood base carbons were seldom used. Carbon particle size would be relatively large (4 x 8 mesh, etc.), with at least one system utilizing crushed activated charcoal briquettes. In operation of this system, the carbon would be used until contaminated and then dumped. Most literature sources report that superheated steam is necessary to regenerate this carbon. These units would normally be installed as side stream filters, processing between 1 and 2% of the total circulation. In general, these systems met with only partial success. Removing solid particulate matter completely from the system required changing the filter elements too often. The activated carbon bed would be too small and too limited in size to remove solid materials from the solution. Their ability to remove dissolved and entrained liquid contaminants depended on bed capacity and the length of time between changes. A typical installation of this type is indicated in Figure 1. Graded Bed Activated Carbon Filters Occasionally activated carbon filters are installed in the form of a graded bed, similar to the standard high pressure rapid rate water filters. In this process, a vessel of the required diameter by approximately 5' shell height contained the carbon filter. A collector header installed in the bottom of the filter is surrounded by a coarse packing such as coarse gravel, crushed anthracite coal, etc. This bottom layer would be approximately 1-3/16" to 1-5/8" in size. On top of this bed would be a layer of finer particle filter support media (either gravel or crushed anthracite coal) of approximately 5/16" to 9/16" diameter. On top of this grade of bed would be another layer of finer support media, approximately 3/32" to 3/16" in diameter. On top of this layer of support media would be a layer of 12" to 18" of fine activated carbon. This activated carbon would normally be a 10 x 30 mesh or smaller. In the operation of these units, the solution enters the top of the filter and moves downward through the bed. The mechanical filtration for removal of solid particulate matter occurs in the top 2" or 3" of the bed, with the remainder of the bed and support media merely supports the fine bed of activated carbon. Dissolved and entrained liquid contaminants are adsorbed on the activated carbon layer. Allowable velocities through such a bed would be a maximum of 2.5 gpm per square foot. Bed life between backwash or recharging would be extremely short, normally no more than 24 hours. In cleaning these beds, backwashing was normally the most effective means. In backwashing, fresh water enters the bottom of the bed at rates of 10 to 15 gallons per minute per square foot, to effectively lift and elutriate the fine bed of activated carbon. Solid contaminants washed from the carbon are carried out in the backwash stream. Liquid contaminants such as oil, corrosion inhibitors, etc. would have to be steamed off periodically to completely regenerate the bed. These graded bed filters are extremely effective in the removal of contaminants from amine and glycol solutions. Unfortunately, due to the very limited amount of volume of activated carbon for removal of contaminants, the life between necessary backwash or changing of the beds is extremely short. Also, it is almost essential that the filtration be a side stream process for 1% to 10% of the total solution stream. A full stream filter would require such a large vessel that it would not be economical or practical to use. Figure 2 indicates a graded bed filter. "Deep Bed" Filtration In the late 1960's, a new process (4) for filtration of glycol and amine solutions was developed and patented, utilizing a deep coarse bed of activated carbon. This process overcomes some of the objections of the above two mentioned processes. It offers long bed life, almost complete removal of solid contaminants, and low pressure drops. In the "Deep Bed" process, a combination of high liquid rates (up to 15 gallons per minute per square foot), a deep bed (minimum 5 feet), relatively large carbon particle size (optimum 4 x 10 mesh) and a low bulk density carbon are combined to accomplish this filtration process. In a typical installation, the contaminated amine is passed downward through the carbon bed in the filter. Even though the carbon particle size is very coarse, the filter will effectively remove extremely finely divided iron sulfide particles and other solid contaminants. The outlet amine solution will be almost water white in an amine system. Heavy hydrocarbons as well as other liquids or emulsified contaminants in the system will be removed by adsorption. Typical bed life for such filters is between 1 and.6 months. Beds may then be regenerated with low pressure steam (40 Psi) for a regenerated bed life being approximately the same as the original new bed life. These beds may be regenerated for a 1 to 2 year period before it is necessary to change the carbon. Although the mechanics of the combined adsorptive purification and mechanical filtration using the large particle size carbon is not fully known, it is believed that liquid contaminants are removed by purely adsorptive processes. Removal of extremely small solid particles (less than 5 micron in diameter) in a bed in which the carbon particle size is a 4 x 10 mesh (approximately 1/8" to 1/4" in diameter) is not what would be anticipated. It is believed that the solid particles are trapped on the jagged edges of the carbon particles as they pass down through the bed. This is the reason for the necessity of the relatively deep bed (minimum 5 foot). In addition, it is believed that solid materials are coagulated with the liquid hydrocarbons removed and adsorbed upon the carbon particles. The result is an amine system with essentially no dissolved or entrained contaminants. Glycol systems can be cleaned almost as well as the amine systems with the properly installed filter. The glycol systems WI normally have a slight discoloration due to discoloration of will glycol itself. These deep bed carbon filters are normally installed on the lean solution stream for full stream service. After a carbon filter is installed, normally 2 to 3 days operation are required before the solution becomes completely clear. In the steam regeneration of the bed, steam is injected into the top of the bed after the vessel had been drained. The outlet header is connected to a drain line, and steam is run into the vessel until the outlet steam condensate is clear and clean. Low pressure steam (as low as 15 Psi) is adequate to completely regenerate the beds sufficiently to obtain good bed life. Normally, steam regeneration is required only every 2 or 3 months. If the solution is quit dirty the steam regeneration may be required once per month. On one installation, the bed was steamed approximately every other week due to an exceptionally high level of contamination. The necessity of steaming of the bed is determined by pressure drop across the bed and the condition of the solution. A newly regenerated bed will have approximately 2 Psi drop; a contaminated bed will have a 10 to 15 Psi drop. In this filtration process, a low bulk density carbon appears be most suitable. Wood base carbons of approximately 16 to 18 pounds per cubic foot will provide excellent service and long bed life. In the use of carbon for. filtration, bed life is closely related to void space, and the lower bulk density carbons have the most void space. These low density carbons are considerably less expensive than the high bulk density, gas adsorptive grade carbons in terms of cost per cubic foot. Figure 3 is a drawing of a "deep bed" filter. LOCATION, INSTALLATION, AND OPERATION OF CARBON FILTERS In most cases carbon filters are installed in the lean solution stream because some contaminants, such as low molecular weight hydrocarbons, are removed in the distillation of the solution. Removal of these contaminants reduces the load on the filter and lengthens bed life. It is also desirable to install the carbon filter downstream of the solution exchangers, and preferably downstream of the solution cooler where the capacity of the activated carbon to adsorb liquid hydrocarbons is greater due to the lower temperature. In the case of small field glycol units, there may not be a suitable place on the unit to install a carbon filter on the lean solution, inasmuch as the filter would have to be installed downstream of the pumps in high pressure service. Most glycol units' filters are installed on a rich solution stream immediately ahead of the glycol-glycol heat exchangers. Certain literature(2) has reported the desirability of installing filters in amine systems on the rich solution where there is a low H2S/CO2 ratio. Our experience has strongly indicated the desirability of such an installation on the low H2S/CO2 ratio streams. This is especially true of diethanolamine. In a diethanolamine system, where there is little H2S, the entrained iron sulfide is converted back to soluble iron in the still due to essentially complete stripping of H2S from the solution in the still. As the sulfide ions are stripped from the solution, the iron sulfide converts back to soluble iron and hydrogen sulfide. If this soluble iron remains in the solution, it will pass through either carbon or mechanical filters and enter the contactor as soluble iron ions which react with the H2S in the gas forming iron sulfide. This iron sulfide precipitates on the trays in the contactor and requires frequent acidizing and cleaning of the contactor. For such installations, it has been found that activated carbon filtration of the rich solution will effectively remove iron sulfide prior to stripping the solution, and will reduce the iron content of the solution to nil, thus eliminating the formation of iron sulfide in the contactor and plugging of the contactor. In the operation of a system with activated carbon filters installed, it is recommended that the operator determine the condition of his filters more by the condition of the solution than by any physical indicator. Pressure drops across the filter bed will give an indication that the bed is becoming contaminated when high pressure drops appear. However, the operator can tell the need to either regenerate or change the carbon beds more readily by the appearance of the solution and the tendency of the plant to foam. When foaming occurs, or when the amine solution appears to be clouded, discolored, or contains solid material, then the operator should make arrangements to steam regenerate the filter (or change the carbon bed). Stand-by filters are not essential since the regeneration of the filter bed normally requires approximately 8 hours. The solution will not become sufficiently contaminated to require a second filter be in service "during the cleaning of the main filter. The filters should be installed with block valves and by-pass valves to allow cleaning without shutdown. If the unit is installed on the lean solution, and if the plant has booster pumps (which are normally located between the amine solution exchangers and the amine coolers), this is an excellent location to install the filters. The filters should be designed for full stream filtration. However, it is not desirable to force the full stream through the filter. A better procedure is to oversize the booster pumps 100% and then allow the equivalent of full stream for the discharge of the booster pump to pass through the filter and back into the booster pump suction, as indicated in Figure 3. If the full stream is forced through the bed, and the bed should become completely plugged for any reason, this will starve the main solution pumps, resulting in a plant shut down. If the filter is installed on the rich solution, it should be installed immediately downstream of the rich solution flash tank before the solution heat exchangers. Also, in this installation, a safety relief valve around the filter should be installed to relieve excess pressure if the filter bed should become completely plugged. (In our experience, we never have encountered a bed becoming completely plugged in a short period of time. Instead, the pressure drop across the bed slowly increases until it becomes excessive to the point that it requires some attention.) In steaming of contaminated beds, merely block and by-pass the filter and drain the solution into a suitable receiver. Low pressure steam (15 to 40 Psi) is run into the top of the bed, and the bed is steamed downward in the same direction as the solution flow. The outlet header line is connected through a valve to a suitable drain. As the steam starts through the bed, dirty condensate, oil, and iron sulfide will discharge through the drain line. As the steaming progresses, the filter bed becomes warm, and the amount of contaminants coming out with the condensate will decrease. When the outlet stream becomes clear water condensate, then the bed is cleaned and may be placed back in service. It is not necessary to cool the bed before placing it back in service. This steam regeneration normally can be accomplished in 2 to 6 hours, depending on the amount of steam available for regeneration. Figure 4 shows a typical installation. CARBON FILTER PERFORMANCE The "deep bed" carbon filter process removes almost all undesirable contaminants from glycol, diethanolamine, and monethanolamine solutions. They may be used without any other purification of the amine and will produce essentially "water white" amine. Table I is a series of analyses for various plants using diethanolamine, indicating the condition of the amine. No detectable iron was in the solution, and the heat stable salts were extremely low. Degradation products were essentially nil. Removing contaminants from the diethanolamine allows extremely high acid gas loadings. Our plants are normally designed to operate at 0.7 Mol per mol loading, and this level provides adequate safety factor. In addition, if the contaminants are removed, very little foaming is encountered in diethanolamine systems, provided they are designed properly. In the design of a unit for high loading of diethanolamine, one must always remember the great volume of acid gas removed in various spots in the system. This starts with the rich amine flash tank. It should be sized for a larger than normal amount of gas release from rich amine solutions. As the solution passes through the heat exchanger, there will be a release of acid gas due to the heating of the solution. It is suggested that the rich amine side of the heat exchangers be equipped with a small vent to let the acid gasses released vent by the remainder of the exchangers to the rich amine line downstream of the exchangers. The rich amine line from the solution exchangers to the still must be oversized to handle the larger amount of acid gas breaking out of the rich amine. The reflux condenser will normally have sufficient area to handle the greater acid gas load, provided excessive pressure drop is not encountered through the condenser. However, the still vent piping and still back pressure motor valve station may have to be increased in size to handle the larger volume of acid gas. ECONOMICS OF ACTIVATED CARBON FILTERS Activated carbon filters are normally the most economical means of purification of amine solution and glycol systems. Capital investments are relatively low (ranging from approximately $800 per gpm capacity for small units 10 gpm or less to approximately $30 per gpm capacity for large units 100 gpm or greater, including the initial carbon charge). Installation cost, including block and by-pass valves will increase these costs by approximately 25%. In operating cost, where steam is available within the plant to steam the filters, steam regeneration is of nominal cost. If it is not available, an oil field portable type steamer may be used for almost all sizes of filters at a cost of approximately $100 to steam regenerate. Activated carbon cost for filters can be estimated at $5 per year per gpm capacity. This will be sufficient to change the carbon at least once a year, which is more often than necessary on most filters. As can be calculated from the above criteria, the cost of filtration, both from initial capital investment and for operating expense, is a very small part of the overall cost of an amine treating plant or glycol dehydration plant. Although activated carbon filters may be installed as a side stream filter, when one considers the improved operation with full stream filtration and exceptionally clean amine or glycol, it becomes ridiculous to attempt to economize in the cost of filtration in a glycol or amine processing plant. CONCLUSIONS Activated carbon filters may be used in lieu of all other types of filters to completely remove all contaminants, both solid and liquid (either dissolved or entrained) from glycol and amine systems. These activated carbon filters, when properly designed and installed will result in the system being cleaner than will be possible with almost all other purification processes for amine and glycol systems. In addition, activated carbon filtration can easily be installed and properly maintained, and will be the most economical method of purification of amine and glycol systems. BIBLIOGRAPHY 1. Gustafson, K. J., "Conditioning Gas - Treating Liquids with Activated Carbon", Proceedings of the Gas Conditioning Conference, 1970. 2. Smith, R. F. and Younger, A. H., "Operating Experience of Canadian Diethanolamine Plants", Proceedings of the Gas Conditioning Conference, 1972. 3. Scheirman, W. L., "Diethanolamine Solution Filtering and Reclaiming in Gas Treating Plants", Proceedings of the Gas Conditioning Conference_ 1973. 4. Perry, Charles R., U.S. Patent No. 3,568,405, "Filtration". Tables and Diagrams: