Design of a CO2 Absorption System in an Ammonia Plant


Chemical Engineering 403

November 10, 1997


Jason Underwood, Group Leader

Gwendolyn Dawson

Christin Barney

Abstract

The purpose of this project was to design and model a system to separate carbon dioxide from the feed stream for an ammonia synthesis plant. The CO2 is then purified in order to be sold to pay the operating expenses of separation. This design group developed an intricate separation process consisting of an absorber and a stripper which results in only trace amounts of CO2 being left in the feed stream. Furthermore, most of this separated CO2 is recovered, at 99.6% purity, so that it can be sold. This separation process is completed with the help of a hot potassium carbonate solvent that is almost completely regenerated.

Introduction

It is necessary to remove the carbon dioxide from the feed stream because it poisons the iron catalyst found downstream in the ammonia synthesis process. The poisoning is caused by oxygen, from traces of carbon dioxide or water, absorbing onto the iron surface and preventing nitrogen absorption. This deactivation process is only sometimes reversible by reduction with hydrogen. Permanent deactivation can occur due to exposure over the course of several weeks to oxygen-containing compounds such as carbon dioxide. The large expense and difficulty in changing out a catalyst makes preventing their poisoning important.

This carbon dioxide removal is usually done with an absorber-stripper process. The unit should be located between the low-temperature water gas shift and the methanation units, just before the ammonia synthesis beds in a typical ammonia plant. Aspen was used to model the absorber-stripper system. The general design is such that a carbon dioxide absorbing solvent flows down the absorber while the feed gas flows up. Then, the rich solvent is regenerated in the stripper at low pressure before recycling back to the absorber. Next, water is condensed out of the CO2 that was removed in the stripper. Figure 1 in the Appendix shows the flow diagram generated in Aspen of this process. An important decision in the design process was which solvent to use. Historically, water was the first solvent used in this process but is no longer used in new plants due to hydrogen losses along with the carbon dioxide.

Monoethanolamine (MEA) is commonly used in existing plants, but hot potassium carbonate is being used in most new plants. It has a high CO2 capacity and provides negligible hydrogen losses. The choice was mainly between MEA and potassium carbonate. This project group chose potassium carbonate solution as its solvent for reasons which will be detailed later in this report.

Reactions

Several reactions occur in the absorber that remove the carbon dioxide from our feed stream. Because many of these reactions are ionic reactions, this system can only be realistically modeled with the 'elec' option in Aspen. In addition to the reactions described below, the dissociation of water into hydrogen and hydroxide ions is also modeled. First of all in our system, our solvent, potassium carbonate, dissociates by the following reaction:

K2CO3 2K+ + CO32-

This reaction sets up the carbonate system in the solution in the absorber. The carbonate system consists of the three equilibrium reactions of carbonic acid:

H2CO3 + H20 H3O+ + HCO3-

HCO3- + H2O H3O+ + CO32-

CO2 + 2H2O H3O+ + HCO3-

The last reaction describes exactly how the carbon dioxide in the feed stream gets absorbed into the solution. This carbon dioxide reacts to form mostly bicarbonate ion which then exists in an equilibrium relationship with potassium bicarbonate as in the following reaction:

K+ + HCO3- KHCO3

The bottoms stream coming out of the absorber consists of potassium bicarbonate, bicarbonate ion, carbonate ion, and carbonic acid along with slight impurities. In this way, all of the carbon dioxide in the feed stream is absorbed into the bottoms solution that then continues on to the stripper.

In the stripper, essentially, the reverse reactions occur. Equilibrium shifts to the left side of the above carbonate system equations. This shift in equilibrium occurs largely because of the huge difference in pressures between the absorber and the stripper. The stripper removes the carbon dioxide from the solvent and sends in on to be purified and sold. The regenerated solvent is then recycled back to the absorber to complete the cycle again. This solvent can be recycled indefinitely, and new solvent needs to be added to the system only to make up for small solvent losses in the stripper.

Discussion

When designing a carbon dioxide absorption system for an ammonia synthesis plant, several important design elements must be considered. First of all, the type of solvent used in the absorber to remove carbon dioxide is an important decision. A solvent must be chosen based on economics, purity of products, and industry standards. Secondly, the purity of the stream that is to be sent on to the rest of the ammonia synthesis plant must be considered. Small amounts of impurities, especially carbon dioxide, can adversely affect the rest of the downstream process. Thirdly, the purity of the carbon dioxide product stream must be considered. Less pure product carbon dioxide is easier to make but is not worth as much when sold on the open market.

The purity of the two product streams will be discussed here while the solvent choice will be considered in the next section. The first product stream consists of mainly hydrogen and nitrogen which will be combined in a later part of the process to make ammonia. This stream may also contain water, and this will not adversely affect the downstream process. Most importantly, this stream must not contain appreciable amounts of carbon dioxide because this compound is a catalyst poison in the ammonia synthesis process. Ideally, this stream will also contain very little carbon monoxide as well. In this particular project, however, the feed stream into the absorber contains a significant amount of carbon monoxide that will not be removed in the absorber. In a normal situation, this carbon monoxide would already be converted to carbon dioxide before the stream goes into the carbon dioxide absorber.

The second product stream is the pure carbon dioxide stream that comes out of the stripper. Because high purity carbon dioxide yields a high market price, we chose to produce a product gas that is 99.5% carbon dioxide. The impurity in the stream is mostly water. This high purity allows us to sell our product carbon dioxide to industries that demand only the purest carbon dioxide, and because of this, we can get as much as $50/ton for our product carbon dioxide.

Choice of Solvent

The solvent to be used in the absorber is a very important design consideration. Traditionally, monoethanolamine or diethanolamine was chosen to adsorb the carbon dioxide from the feed. Now, however, most new plants are designed to use potassium carbonate for many crucial and convincing reasons.

First of all, potassium carbonate is a more efficient absorber of carbon dioxide than either monoethanolamine or diethanolamine. This means that for a given amount of solvent, potassium carbonate can absorb more carbon dioxide than the other two common solvents, Why is this important? If less solvent is needed, then piping systems and absorber volume can be smaller. In addition, the cost of the solvent is less because less is needed and potassium carbonate is cheaper than the traditional solvents.

Secondly, the use of potassium carbonate over the other two solvents eliminates the need for a heat exchanger between the absorber and the stripper. In a potassium carbonate system the stripper runs cooler than the absorber. This is not true for the traditional solvents. Traditional solvents require the heating of the bottoms stream from the absorber before it reaches the stripper, but a potassium carbonate system requires no such heating. The elimination of this heat exchanger increases profitability of the plant because heat exchangers are both costly to purchase and maintain.

Thirdly, potassium carbonate increases the safety of the carbon dioxide removal system. Traditional solvent absorb not only carbon dioxide but small amounts of hydrogen as well. This hydrogen then continues through the stripper and the out the top stream with the carbon dioxide. Beside decreasing the purity of the product carbon dioxide stream, this hydrogen poses a safety threat, Flashing off hydrogen can cause fires or explosions if proper precautions are not taken. Using potassium carbonate reduces the risk of such accidents.

Finally, the most convincing argument for the use of potassium carbonate as a solvent is that this is the present industry standard. There are many plants that use monoethanolamine and diethanolamine, but these plants were built a long time ago and don't yet have potassium carbonate capabilities. Most new plants use potassium carbonate, and even some existing plants are deciding to convert over to potassium carbonate as their solvent of choice. This evidence shows that potassium carbonate is significantly better than the traditional solvents.

Discussion of Optimal Design

In order to remove CO2 from the feed stream, two main vessels are needed: an absorber and a stripper. Inside the absorber, chemical reactions will occur in which the CO2 is chemically absorbed onto the solvent and this compound, along with water, will flow out the bottom of the absorber. That stream will then go through a flash chamber to lower the temperature and pressure of the liquid (and to separate out some CO2) before proceeding onto the stripper where the CO2 will separate out from the solvent. Both the absorber and stripper were modeled in Aspen using the Radfrac modules.

Absorber Design

The absorber was modeled using three theoretical stages and no condenser or reboiler. The K2CO3 and water stream, the solvent, flowed in from above the top stage while the feed stream from the low-temperature water-gas shift flowed into the column at the bottom. Both the feed stream and the hot potassium carbonate solution entered the absorber at 250 degrees Celsius and 50 bar. The solvent entered as a liquid while the feed stream came in as a gas. The following is a table showing data from the two streams entering the absorber and the streams coming out of it.

Feed StreamSolvent BottomsFeed to Ammonia
Temperature (C)250250 268241
Pressure (bar)5050 3840
Vapor Fraction1.000 01.00
Mole Flow (lbmol/hr)19462 2653937814650
Mass Flow (lb/hr)303920 182133310749175304
Mole Flow (lbmol/hr)
Water62461551 51432353
H27441- .5347440
N22978- .1012978
CO1795- .5821794
CO2919- 625trace
AR37- .0337
KHCO3-- -trace
CH446- .11246
K2CO3- 1102--
K+-- 2205-
H3O+- -trace-
HCO3-- -596-
OH--- 7-
CO32-- -801.-

As the table shows, the absorber does a very good job of removing CO2 from the feed stream coming out of the low-temperature water-gas shift. Only a trace amount of CO2 is remaining in the stream that will be sent on to the ammonia synthesis portion of the plant. The CO2 is all in the bottom stream-mostly in the form of dissolved CO2, but some of it is also in the ionized state. The absorber operates across a temperature range from 268 degrees Celsius to 241 degrees Celsius. There is a pressure drop of 2 bar down the column (from 40 bar at the top to 38 bar at the bottom).

The absorber was sized and an economic analysis completed using the Aspen cost package. It was determined that the optimal size of the column is 36 feet in length and 12 feet in diameter. There will be 11 actual trays, and the cost of the column will be almost $500,000. The absorber, like all parts of the plant, must be made of reinforced stainless steel with thick walls due to the high pressures in the system. Also, a small amount of an anti-corrosive agent must be added to the solvent stream to prevent corrosion in the system. A commonly used agent is vanadium pentoxide. A catalyst can also be added to the system if more CO2 needs to be removed from the system. In this case, a catalyst is not needed. However, one might be necessary depending on how much CO2 is fed into the absorber.

The purified feed stream is sent from the top of the absorber to the ammonia synthesis reactors. Essentially, we have completed the goal of our project: to remove as much CO2 as possible from the feed stream in order to prevent catalysis poisoning downstream. However, for economic reasons it is beneficial to regenerate the solvent and sell the carbon dioxide. The remaining two flash units and the stripper are designed to purify CO2 and regenerate solvent.

First Flash Unit

After leaving the bottom of the absorber, the solvent and CO2 travel to a flash unit which will create a vapor stream consisting of water and some of the CO2 in the system. The flash also serves to lower the temperature and pressure of the solvent stream in order to allow it to enter the stripper at optimal conditions for separation of CO2 from the solvent. The following table shows the data from the stream entering the flash and the two outlet streams.

Stream From Absorber CO2 and WaterStream to Stripper
Temperature (C)268121 121
Pressure (bar)385 5
Vapor Fraction01.00 0
Mole Flow (lbmol/hr)9378 3538718
Mass Flow (lb/hr)310749 1399937565
Mole Flow (lbmol/hr)
Water514357 4785
H2.534.533 .001
N2.101.101 -
CO.582.572 .01
CO2625294 23
AR.03.03 .001
KHCO3-- -
CH4.112.096 .016
K2CO3- --
K+2205- 2205
H3O+trace -trace
HCO3-596 -1205
OH-7- .085
CO32-801 -500

The absorber was sized and an economic analysis completed using the Aspen cost package. It was determined that the optimal flash unit is 6.67 feet in diameter and will cost approximately $40,200.

Stripper Design

After running through the flash, the stream with primarily water and ions is sent to the top of the stripper. In the stripper, the reverse of the reactions in the absorber occur. Here, the CO2 is released from the solution as a vapor and, along with steam, travels out the top of the stripper. The remaining solution, which is almost entirely potassium and carbonate ions, comes out the bottom of the stripper and is sent back to the absorber to be re-used. The solution will react in the stripper with steam coming in from the bottom. The stripper contains 10 theoretical plates and there is almost no pressure drop across the column (2 bar at top to 1.8 bar at bottom). The temperature ranges from 141 degrees Celsius at the top of the column to 147 degrees Celsius at the bottom. The stripper utilizes a condenser. The ratio of the liquid flowing in the top of the column to the distillate is approximately 2.65:1. The following table show the streams coming in and out of the stripper:

Feed to StripperWater CO2 streamRegenerated Solvent
Temperature (C)121120 141147
Pressure (bar)510 21.8
Vapor Fraction01.00 1.000
Mole Flow(lbmol/hr)8718 165332837635
Mass Flow (lb/hr)297298 2978873978253107
Mole Flow(lbmol/hr)
Water47851653 27124264
H2.001- .001trace
N2-- -trace
CO.01- .01trace
CO223- 571.185
AR.001- .001trace
KHCO3-- --
CH4.016- .016trace
K2CO3- --
K+2205- -2205
H3O+trace - -trace
HCO3-1205 --118
OH-.085- -10
CO32-500 --1038

Using the Aspen cost program, the stripper was found to be 7.5 feet in diameter and 28 feet long. It will cost $160,400 including the condenser.

Second Flash Unit

The regenerated solvent is now ready to be sent back to the absorber. Before the CO2 can be sold, though, it must be separated from the water. This is done in a final flash tank. The CO2 and steam streams from both the first flash tank and the stripper are combined and fed into a final flash tank in which the pressure is 5 bar and the temperature is lowered to 10 degrees C. In this manner, the CO2 is removed as a vapor and the water is liquefied so that it can be used elsewhere in the plant. The following table shows the streams into and out of our final flash unit:

From first FlashFrom Stripper Pure CO2Pure Water
Temperature (C)121141 1010
Pressure (bar)52 55
Vapor Fraction1.001.00 1.000
Mole Flow (lbmol/hr)353 3283855.6722780
Mass Flow (lb/hr)13999 739783756550413
Mole Flow (lbmol/hr)
Water572712 2.2262766.786
H2.533.001 .533<.001
N2.101- .101<.001
CO.572.01 .557.025
CO2294571 852.13712.858
AR.03.001 .028.003
KHCO3-- --
CH4.096.016 .089.025
K2CO3- ---
K+-- --
H3O+- --.014
HCO3-- --.014
OH--- -trace
CO32-- --trace

This flash chamber will cost $43,400 according to Aspen.

Analysis of Design

This design clearly meets the stated goals of the project. The ammonia stream is purged of CO2, ensuring that no catalyst will be poisoned downstream. Also, 93% of the CO2 entering the system is recovered at a purity of 99.6%. This CO2 amounts to 164,663 tons produced per year. At the current selling price of $45-50 a ton, the CO2 recovery system generates $8,233,147 in revenue for the plant. Also, virtually all of the K2CO3 is regenerated. A comparison of the regenerated solvent to the original solvent feed indicates there is more water and CO32- in the system and less HCO3-. The water can be removed through a simple separation process. The equilibrium between HCO3- and CO32- can probably be shifted more in favor of HCO3- if necessary.

There is a definite economic benefit for using K2CO3 over MEA as was stated earlier. Most separation systems which use MEA as the solvent use a heat exchanger between the absorber and stripper for the solvent. When using potassium carbonate as the solvent, this heat exchanger is not necessary because the stripper can be run at a lower temperature than the absorber. This means savings on equipment. The price of the K2CO3 necessary to start up the unit is also less expensive than MEA. In bulk and from the same vendor, K2CO3 was $0.5885/lb while MEA was $0.69/lb. Potassium carbonate was also found for $0.4285/lb from another vendor; this price is used to calculate feed costs. The following table summarizes our capital costs. We believe much of the operating costs will be offset by the revenue generated through CO2 purification.

Capital Cost Analysis

Absorber500,000
First Flash40,200
Stripper160,400
Second Flash43,400
Solvent78,073
Total Capital Costs$822,073

Safety Considerations

The safety of an operation is always an important design concern. Thicker-walled vessels and piping will be needed to contain the high pressures in the absorber, but this is true throughout an ammonia production facility. Care must be taken in handling the hot solvent solution. It is a skin and eye irritant and would cause thermal burns on contact. From the material safety data sheets available, K2CO3 appears to be less of a health hazard than MEA. MEA is also more reactive with a variety of materials. When the facility design is closer to completion, the standard process hazard safety analysis will be done.

Conclusion

This project design clearly meets the most important goal-to remove as much CO2 as possible from the ammonia feed stream in order to prevent catalyst poisoning. The project also separates out this CO2 to an extremely high purity, allowing the plant to make as much money as possible off of the system. We believe that those qualities, in addition to the use of potassium carbonate solvent make this plant design an excellent choice for a CO2 absorption system.

Appendix

  1. Aspen Flowchart for our System

References

Strelzoff, Samuel, ìSection 3: Carbon Dioxide Absorption,î Technology and Manufacture of Ammonia, Robert E. Krieger Publishing Company, Inc.: 1988, pp. 193-248.

Twigg, Martyn V., Catalyst Handbook: second edition, Manson Publishing Ltd: 1996, pp. 80, 340-343, 404-409.

Encyclopedia of Industrial Chemistry. Vol A.5, pp. 172-175