Increasing Efficiency of Hot Potassium Carbonate CO2 Removal Systems

https://www.uop.com/?document=increasing-efficiency-of-hot-potassium-carbonate-co2-removal-systems&download=1



1 Increasing Efficiency of Hot Potassium Carbonate CO2 Removal Systems Stanislav Milidovich, P.E. and Edward Zbacnik UOP LLC ABSTRACT The UOP Benfield™ Process is the original hot potassium carbonate technology widely used in the ammonia industry for CO2 removal downstream of the steam reformer. With experience in over 700 licensed units, UOP has developed significant advances in the technology and continues to support the industry with opportunities to improve existing unit performance. This technical paper will provide information on three types of unit upgrades: addition of advanced chemical activators, implementation of energy-saving semi-lean solution flash technology, and increasing capacity with high-efficiency tower internals and packing. Introduction The Hot Potassium Carbonate Process (HPC) originated from research work done by the US Bureau of Mines (USBM), between 1940 and 1960. The original justification was to determine how to convert coal to gaseous and/or liquid fuels. If the coal could be gasified, followed by the removal of CO2 and sulfur compounds, the resulting product would be mostly a mixture of hydrogen and CO, also called synthesis gas. The synthesis gas can be used as a chemical plant feedstock, converted to substitute for natural gas, or processed further to produce synthetic gasoline. While gasifying the coal is relatively easy, treating the resulting hot gas is more difficult. Scrubbing this gas while hot is more desirable than cooling it because heavier hydrocarbons and tar-like compounds can condense at cooler temperatures. The solvents available at that time for acid gas removal were water, monoethanolamine (MEA), and other amines. These solvents require operating temperatures below 125°F (50°C) in order to prevent solvent degradation. Water could be used to remove CO2, but it is very inefficient and does not provide acceptable treated gas purity. Caustic solutions, either NaOH or KOH could remove the acid gases very effectively, but could not be regenerated. Using a solution of KOH to first pick up CO2 would generate potassium carbonate in solution, which could absorb still more acid gases. Thus, the “Hot Potassium Carbonate” Process was born. Since the technology was developed by the U.S. Government, the basic process remained the property of the U.S. Government. Several U.S. citizens further developed the technology and started businesses to assist the industry to use this technology. A partnership was formed by Benson, Field and Epes, who were former employees of USBM, to help design some 150 units for use in treating town gas produced from coal at locations throughout Europe, mostly in the United Kingdom before the advent of North Sea gas. Eventually this partnership found patentable improvements to the technology and started designing and licensing their improved versions. This became known as the Benfield™ Process. © 2013 UOP LLC., A Honeywell Company All rights reserved. 2 The improvements and process developments included addition of small amounts of amines/other proprietary additives to the HPC solution to increase the rate of reaction with CO2, and using corrosion inhibitors to permit the use of carbon steel for the majority of the process unit. When evaluating economics, the HPC Process, was a relatively low capital investment option when compared with other CO2 removal process technologies available on the market. The low capital cost can be attributed to the simple unit configuration and to most materials of construction being low-cost carbon steel. This, coupled with relatively low cost of utilities, made the HPC Process popular. With the growing global economy, the demand for many products that are created by utilizing the HPC Process has increased motivating many HPC Process owners to increase capacity and improve process efficiency. During recent years, multiple process improvements developed by UOP have been introduced to the market in order to modernize existing HPC Processes by reducing operating costs or increasing unit capacity. These improvements include a new chemical activator to improve HPC solution performance, LoHeatTM Technology for the reduction of energy required to remove CO2 from the feed gas, and Raschig Super-Ring packing and internals for unit capacity increases. These improvements are discussed in more detail below. Benfield ACT-1TM Activator The first available unit upgrade is addition of much more advanced chemical activator to the HPC Process. Almost all HPC Process uses a chemical solution based on 30% potassium carbonate (K2CO3) dissolved in water, some kind of a chemical activator, and a corrosion inhibitor. The activator is a low-concentration additive designed to improve the rate of CO2 absorption. For many years, diethanolamine (DEA) has been the standard activator and it is still used today at many operating plants around the world. Unfortunately, like most other organic chemicals, DEA is subject to degradation. Some of the reasons DEA tends to degrade are listed below:  DEA will break down from overheating (thermal degradation).  DEA reacts with oxygen from air contact or from overuse of reoxidation agents such as potassium nitrite (KNO2), used to regenerate the corrosion inhibitor (vanadium).  By absorbing CO2, a secondary amine activator such as DEA forms a carbamate chemical that normally is easily regenerated. However, because further reactions can occur, some by-products are formed that are not regenerable, and thus a degradation compound is formed. Typically, these compounds are high molecular weight, polymer-type chemicals. Evidence of extensive DEA degradation can be visually seen. The potassium carbonate solution samples appear black and opaque similar in appearance to liquid coal. Such DEA degradation will cause interference with analytical procedures such as carbonate titrations and vanadium valence determinations. Foaming upsets are also frequent due to degradation products and constant addition of antifoam may be required. Often there is also a rapid reduction of valence state of the vanadium corrosion inhibitor from the active V+5 to passive V+4. 3 One result of the breakdown of the DEA molecule is formation of potassium formate and a few other carboxylic acid salts. These salts can be analytically measured and are usually benign at low concentrations. However, when they are found at concentrations of 5% or more, they interfere with operations by altering the physical properties of the potassium carbonate solution. The solution becomes much harder to regenerate. Most of the other known and unknown DEA degradation compounds are notoriously difficult to analyze since most of these compounds, being large polymer-based molecules, are still reactive. Some amine degradation compounds are even considered to be corrosion accelerators in that they may solubilize iron, keeping it in solution and preventing it from formation of the passivation coating. UOP has found an alternative to DEA which has been commercialized as Benfield ACT-1TM activator. This activator, which is a proprietary chemical from UOP, is also an amine but with a more stable molecular form that is considerably more resistant to degradation. To measure the improved performance of the ACT-1 activator, side-by-side accelerated laboratory degradation tests were performed to compare a potassium carbonate solution with DEA and with ACT-1 activator. The first test was to heat samples of both solutions to 167°F (75°C) and expose them to oxygen by continuously injecting air. The DEA was 15% degraded within 45 days, but the ACT-1 activator was still 100% available. Please refer to Table 1 for laboratory data summary. Table 1: Effects of Oxygen on HPC Process Activators (Lab Test Conditions: CO2 saturated, constant air injection, at 75°C) Days of Test % of Active ACT-1 % of Active DEA Day 0 100 100 Day 10 100 97 Day 18 100 93 Day 37 100 87 Day 46 100 86 In another test, both solutions were heated to 121°C to 132°C (250°F to 270°F) and saturated with CO2 at autoclave pressures of 9 to 14 bar (135 to 200 psi). After 15 days, only 25% of the DEA remained; 100% of the ACT-1 activator remained and was reactive after another 50 days. Please refer to Table 2 for laboratory data summary. 4 Table 2: Effects of Temperature and CO2 on HPC Process Activators (Lab Test Conditions: 121-132°C and continuous exposure to CO2 at 9-14 bar) Days of Test % of Active ACT-1 % of Active DEA % of Active MMEA Day 0 100 100 100 Day 3 100 75 70 Day 8 100 50 50 Day 10 100 42 46 Day 15 100 26 40 Day 18 100 - 33 Day 20 100 - 30 Day 50 100 - - Note: MMEA is 2-methyl-methanolamine. The ACT-1 activator is currently in use in many units worldwide, including ammonia plants. It has been used in new units where no DEA was present and in existing units that had used DEA for more than 20 years and later converted to ACT-1 activator. Concentrations most effectively used in plant solutions are 0.3 to 1.0 weight% ACT-1 activator compared to about 3 weight% for DEA. The performance of ACT-1 activator is far superior to that of DEA when looking at the CO2 absorption rates. In all comparisons, the ACT-1 activator in the potassium carbonate solutions substantially reduced the CO2 slippage typically to about 50% of the levels achieved by DEA activation. This improved unit performance is available at no additional energy demand for solution regeneration and no additional solution circulation is required. In fact, plants frequently find slight reductions in regeneration duty and solution circulation rates when compared to the requirements for the same units operating with DEA. Please refer to Figure 1 for a simple graph comparing relative rate of CO2 absorption between DEA and ACT-1 activator for various CO2 partial pressures. Figure 1: DEA vs. ACT-1 Activator Relative Rate of CO2 Absorption CO2 Partial Pressure 2.5 2.0 1.5 1.0 0.5 0.1 1 10 ACT-1 DEA + ACT-1 DEA 5 The ACT-1 activator benefits are fully achievable in new green-field units and in units fully converted from DEA to ACT-1 activator1 . Further Benfield Process performance improvements with ACT-1 activator are observed through a considerably lower antifoam consumption, a much reduced consumption of reoxidizing agent, and the resulting improved process operation (less foaming upsets). This makes ACT-1 activator very attractive for new and existing HPC units. In summary, the most important benefits of the ACT-1 activator when comparing to HPC solutions activated with DEA are:  Potential for feed gas capacity increases of 5-15% resulting from less CO2 slip to the downstream units  Potential reduction in regeneration duty by up to 15%  Potential reduced solution circulation rates by up to 15%  Much improved operational stability with less foaming upsets. LoHeatTM Technology The second unit upgrade option is the implementation of LoHeatTM Technology which reduces regeneration energy requirement and improves the quality of the semi-lean potassium carbonate solution which is to be recycled to the absorber. The LoHeat Technology operates by recovering heat from the hot semi-lean potassium carbonate solution leaving the upper section of the Regenerator column. The semi-lean solution taken from the mid-section of the Regenerator column, which is typically routed to semi-lean pumps, is sent directly to the new Semi-Lean LoHeat Flash Drum. In this drum, the solution flows through multiple separate compartments in which the operating pressure of the solution is sequentially reduced by steam jet ejectors installed on top of the drum. The multi-stage reduction in pressure flash regenerates CO2 from by the semi-lean solution, improving its absorption capacity in the Absorber column. The liberated CO2 and steam from the semi-lean solution is sent back to the middle of the Regenerator column, in place of reboiler steam. The flash regenerated semi-lean solution is sent back to the Absorber column by the Semi-Lean Solution pumps. The motive steam necessary for the operation of steam jet ejectors is created by routing the reflux water from the top of Regenerator column to a new Condensate Reboiler. Low Temperature Shift (LTS) reactor effluent, which is typically the feed gas to the existing HPC Process in ammonia plants, is used to vaporize the reflux water to create the necessary motive steam for the ejectors. A new Condensate Separator, installed directly downstream of the new Condensate Reboiler, and operated on level control, captures water condensed in the new Condensate Reboiler. The LTS reactor effluent is then routed to the existing Regenerator Reboiler. 6 Since there is no change in the total quantity of stripping steam needed to regenerate the potassium carbonate solution, the process is more energy efficient which ultimately reduces the external heat requirement for the regenerator reboilers. Please refer to the Process Flow Diagram in Figure 2 which graphically illustrating in red a typical scope of upgrading to the LoHeat Technology in an HPC unit at an ammonia plant. Figure 2: LoHeat Technology Integration LoHeat Flash Drum Absorber Reboiler Regenerator Reflux Drum Condenser Reflux Pump Rich Solution Pump Lean Solution Cooler Lean Solution Cooler Semi-Lean Solution Pump Rich Solution Turbine Condensate Feed Gas From LTS Reactor Condensate Reboiler BFW Heater Condensate Condensate Separator HIC Condensate Separator As shown in Figure 2, the LoHeat Technology utilizes three main equipment pieces. These equipment pieces include:  Multiple-compartment LoHeat Flash Drum with ejectors,  Condensate Reboiler,  Condensate Separator. All three pieces have a relatively low capital cost providing the plant owner with a very quick payback time (at times as little as 6 months) on a typical upgrade to LoHeat Technology. The LoHeat Technology integrates well into existing ammonia plants and the newest designs for high-energy-efficient ammonia plants. This technology can achieve a net thermal energy consumption of only 650 kcal/Nm3 (69 Btu/std. ft3 ) of CO2 removed in new units. This is almost a 50% reduction in regeneration energy demand when comparing to a new non-LoHeat HPC unit. 7 For existing plants, the upgrade to LoHeat Technology can reduces energy consumption in the range of 30-40%. Raschig Super Ring Packing and Internals The last upgrade option is tower internals replacement. Increasing unit capacity, while attempting to utilize as much of the existing equipment as possible, may be accomplished by increasing mass transfer efficiency in the Absorber and Regenerator columns. This is typically achieved by replacing the random packing with packing that is significantly more efficient. To better understand why the Raschig Super Ring packing is the best random packing option on the market today, let’s take a look at the packing development history. Please refer to Figure 3. Figure 3: Packing Development History First generation packing was introduced in 1895 and was used into the 1950s. The first generation was rather simple and included two main types; Raschig rings and saddles. The predominant material of construction for the first generation packing was a ceramic. Then, in the late 1950’s, second generation packing was developed. This packing, for the first time, introduced additional surface area by redesigning a standard cylindrical shaped ring. Pall rings and Intelox saddles are still in use at many plants today. First Generation (1895 – 1950) Raschig Ring Berl Saddle Fourth Generation (Late 1990’s) Raschig Super-Ring Second Generation (Late 1950’s– Early 1970’s) Pall Ring Intalox Saddle History of development of characteristic random packing of different generations Fleximax Third Generation (Late 1970’s) CMR Ring Nutter Ring IMTP Ring 8 The third generation packing was developed in the late 1970s and it included CMRTM rings, Nutter RingsTM, IMTP® rings, and Fleximax. The most common random packing used in Benfield Process units that have been either revamped or build within the last 20 years are the CMR rings, Nutter rings, and IMTP rings. Fleximax packing is not recommended as it bears a risk of collapsing at the bed depth of most HPC process palnts. Finally, the fourth generation packing called Raschig Super Rings (RSR), was first sold in 1996. It is the most advanced random packing on the market today offering superior mass transfer capability coupled with the lowest pressure drop available. Over 600 mass transfer columns have been packed with RSR in the various chemical process industries, petrochemicals, refining and environmental applications. Excellent results have been achieved with capacity improvements of as much as 50% over other conventional packing such as Pall Rings and IMTP. Exceptionally good mass transfer efficiencies, pressure drops and loading capacities of the solvent are made possible by the revolutionary idea behind the design of the RSR (Figure 4). The structure of the RSR has been designed to produce turbulent film flows and to prevent formation of drops. The large number of alternating wave swings is the main contributor to the large turbulences in the gas and liquid flows. At the same time it has an extremely open geometry leading to very small pressure drops. Thus, having this fluid-dynamically optimized shape, the RSR packing, while randomly dumped in a packed bed, obtains a structure otherwise found only in the case of structured packing. Figure 4: Raschig Super Ring Random Packing 9 Aspects involved in the design of modern packing elements High-performance packing elements are intended to bring about effective mass transfer between the phases flowing through the columns. Large interfacial area and uniform distribution of the phases over the column cross-section are desirable. A high loading capacity permits high column throughputs, while low pressure drop results in low operating costs. Loading capacity Counter-current packed columns are preferably operated below, or in the immediate vicinity, of the so-called loading point. This is being characterised by the fact that the falling film is backed up by the counter-current gas stream at higher loads. The loading point of a packing element is defined by its fluid dynamic properties. Fluid dynamic studies in the past have repeatedly shown that the droplets forming in a column packing are entrained earlier than down-ward flowing liquid films at high gas loads. In contrast to previous packing element designs, the RSR meets this demand in that it was purposely designed without any projecting metal tongues which could act as dripping points. Liquid and gas distribution The most uniform possible distribution of the liquid and gas phase across the packing element itself and the entire column cross-section is one of the fundamental prerequisites for a column packing that works effectively. If, at the same time, a low resistance to fluid flow of the gas phase is to ensure the minimum possible pressure drop, the structure must be largely open. The alternating wave structure of the RSR has not only created a form which is open on all sides but, at the same time, has also realised a large number of contact points for homogeneous liquid and gas distribution. Mass transfer Effective mass transfer between the phases demands not only a large interfacial area, but also the most turbulent flow conditions possible and frequent renewal of the phase interfaces. With the RSR, several thin films of liquid displaying turbulent flow are formed on the webs and are constantly intermixed as the result of the recurrent contact points within the packing element. A performance gain of as much as 40% can be observed when switching from Raschig rings to RSR, 20-30% when switching from Pall rings to RSR, and 5-10% when switching from IMTP rings to RSR. Case Study To better illustrate the process improvement options described above, let’s take a look at a case study. The case study discussed in this section focuses on revamping an existing ammonia plant in order to increase its capacity from 1000 to 1500 metric tons of ammonia produced per day. In this specific example, the HPC Process unit has been identified as the bottle neck and needed to be revamped. Original HPC Process Unit Please refer to Figure 5 for a simple flow diagram showing the configuration of the original HPC Process unit. 10 The original HPC Process unit was designed to treat 152,900 Nm3 /hr (137 MMSCFD) of shifted syngas feed with approximately 18 mol% CO2. The target treated gas specification was set at 1000 ppmv leaving the HPC Process unit. The unit was also designed to produce acid gas (CO2) with greater than 99 mol% purity. Figure 5: Original HPC Process Unit XXXXXX XXXXXX XXXXXXXX HPC Absorber HPC Regenerator Treated Gas Acid Gas Feed Gas Surplus Water As a side note, the above mentioned specifications are very typical for HPC Process units installed at ammonia plants. Some of the modern HPC Process units require more stringent CO2 specifications for the treated gas, i.e. 500 ppmv or less, which can be achieved. Also, depending if the ammonia plant uses CO2 to produce urea, the CO2 produced by the HPC Process may require even higher purity with restrictions on hydrogen content. The original HPC Process unit has a Carbonate Reboiler duty of 33.6 Gcal/hr (133 MMBtu/hr) which was provided by the hot shifted syngas exiting the Low Temperature Shift (LTS) reactor. This reboiler duty resulted in a process regeneration efficiency of 1221 kcal/Nm3 (130 Btu/ std. ft3 ) of removed CO2. A total cooling duty requirement for the original design was 41 Gcal/hr (163 MMBtu/hr) which was split between the Lean Solution Cooler and the Reflux Condenser. The original HPC Process unit had a lean solution flow rate of 273 m3 /hr (1200 gpm) and a semilean solution flow rate of 818 m3 /hr (3600 gpm). This resulted in a 25% and 75% solution flow rate split between the lean and semi-lean solutions, respectively. The original HPC Process unit utilized 40 mm Pall rings in the upper section of the Absorber, 50 mm Pall rings in the lower section of the Absorber, 50 mm Pall rings in the upper section of the Regenerator, and 40 mm Pall rings in the lower section of the Regenerator. 11 In summary, the original HPC Process unit needed to be revamped in order to process 38% more feed gas and remove 36% more CO2. Please refer to Table 3 for a summary of original design conditions and post revamp conditions. Table 3: Original HPC Process Unit Design Conditions vs. Revamp SI Units Original Design Post Revamp % Change Feed Rate (Nm3 /hr) 152,900 212,100 38% CO2 Removal (Nm3 /hr) 27,411 37,296 36% English Units Original Design Post Revamp % Change Feed Rate (MMSCFD) 137 190 38% CO2 Removal (MMSCFD) 24.6 33.4 36% Revamped HPC Process Unit In order to meet the required specifications for revamp condition and utilize as much of the existing equipment as possible, it was necessary to employ all three HPC Process improvements discussed in the previous sections. Figure 6 shows a simple flow diagram showing of the revamp design. The main change in this flow diagram is addition of equipment associated with LoHeat Technology necessary to reduce the energy demand of the existing unit with much higher syngas feed rate. Figure 6: Revamped HPC Process Unit XXXXXX XXXXXXXX XXXXXX Pure Gas HPC Absorber HPC Regenerator Acid Gas Surplus Water Feed Gas LPS LoHeatTM Flash 12 In this example, the UOP design employs a four-stage LoHeat flash, where CO2 is flashed off in four separate stages and returned into the Regenerator. This has a cooling effect on the solution which improves its capacity for CO2 absorption. From the last flash stage, the solution is pumped into the bulk section of the Absorber. The added LoHeat Technology equipment shown in this example includes: Condensate Reboiler and Pump, LoHeat Flash Drum, and four steam ejectors installed on top of the LoHeat Flash Drum. To meet the regeneration heat requirement for this particular example, UOP also added a small carbonate reboiler, which is heated with low pressure steam (LPS). Since the thermal energy of hot syngas feed was utilized to its full extent, the small reboiler was required in order to boil up any bottoms that we not heated by the main Carbonate Reboiler. To reach the necessary 38% higher feed rate, the original HPC Process unit needed some packing replacement. The revamp design allowed for continued use of the 40 mm Pall ring packing in the upper section of the Absorber, but it was necessary to replace the rest of the random packing. The lower section of the Absorber and the entire Regenerator were repacked with 50 mm RSR packing and internals. To accomplish the necessary 36% higher CO2 removal rate, the total solution flow rate (lean and semi-lean) was increased to 1454 m3 /hr (6400 gpm) from the originally designed 1091 m3 /hr (4800 gpm). The higher total solution flow rate meant that some solution pumps needed to be replaced. In order to salvage the lean solution pumps and to hydraulically unload the top of the Absorber, which allowed for continued use of existing Pall ring packing, the UOP design changed the solution split to 15% lean and 85% semi-lean. Salvaging the lean solution pumps and utilizing existing packing in the upper section of the Absorber was not the only benefit of changing the solution split between lean and semi-lean. This change also resulted in the cooler lean solution improving the treated gas qulity (less CO2 in the treated gas). Similarly by forcing more solution through the LoHeat flash system via the higher semi-lean solution flow rate, there is more heat recovered, which makes the overall unit more energy efficient. Please refer to Table 4 for a comparison summary of major utility requirements between the original HPC Process unit and revamped HPC Process unit. 13 Table 4: Major Utility Comparison, Original vs. Revamp SI Units Original Design Post Revamp % Change Feed Rate (Nm3 /hr) 152,900 212,100 + 38% Lean Solution Flow Rate (m3 /hr) 273 216 - 21% Semi-Lean Solution Flow Rate (m3 /hr) 818 1238 + 51% Lean Solution Cooler (Gcal/hr) 13.2 11.1 - 16% Reflux Condenser (Gcal/hr) 27.8 22.3 - 20% Carbonate Reboiler (Gcal/hr) 33.6 21.6 - 36% New Condensate Reboiler (Gcal/hr) - 11.2 - New LPS Reboiler (Gcal/hr) - 0.4 - HPC Process Regeneration Efficiency (kcal/Nm3 of removed CO2) 1221 884 - 27.6% English Units Original Design Post Revamp % Change Feed Rate (MMSCFD) 137 190 + 38% Lean Solution Flow Rate (gpm) 1200 950 - 21% Semi-Lean Solution Flow Rate (gpm) 3600 5450 + 51% Lean Solution Cooler (MMBtu/hr) 52 44 - 16% Reflux Condenser (MMBtu/hr) 110 88 - 20% Carbonate Reboiler (MMBtu/hr) 133 85 - 36% New Condensate Reboiler (MMBtu/hr) - 44 - New LPS Reboiler (MMBtu/hr) - 1.5 - HPC Process Regeneration Efficiency (Btu/std. ft3 of removed CO2) 130 94 - 27.6% 14 Conclusions Multiple process improvements developed by UOP have been introduced in order to modernize or upgrade the existing HPC Process by reducing operating costs or increasing unit capacity. The improvements discussed in this paper are:  ACT-1 activator designed to improve potassium carbonate solution performance by resisting degradation and improving CO2 absorption rates resulting in up to 5-15% capacity increase or energy savings.  LoHeat Technology engineered to reduce regeneration energy requirement and improve quality of the semi-lean potassium carbonate solution resulting in up to 30-40% energy savings.  Raschig Super-Ring packing and internals designed to maximize mass transfer efficiency and reduce pressure drop in order increase unit capacity resulting in up to 5-40% capacity increase. References 1. Bartoo, R. K., Gemborys, T. M., and Wolf, C. W. (June 1991): “Recent Improvements to the Benfield Process Extended in Use”, UOP technical paper presented at the Nitrogen International Coneference in Copenhagen. 2. Bartoo, R. K., Furukawa, S. K. (January 1997): “Improved Benfield Process for Ammonia Plants”, internal UOP technical paper. 3. Gemborys, T. M. (September 1994): “UOP Benfield Process ACT-1 Activator”, UOP technical paper that covers additional information specific to the ACT-1 activator presented at the AIChE Annual Ammonia Symposium.