AMP for carbon capture?
Process simulation is used to evaluate the potential for using 2-amino-2-methyl-1-propanol (AMP) as a solvent for CO2 capture. In terms of regeneration energy (the main energy consumer in capture plants), the performance of a rudimentary MEA-based plant is established first.
Ralph H Weiland, Nathan A Hatcher and Jaime L Nava
Optimized Gas Treating, Inc.
Viewed : 4143
Pilot plant data on CO2 absorption into MEA are shown to be well-reproduced using the ProTreat‘ mass and heat transfer rate process simulation tool. The ability of the simulator to model accurately AMP absorption data taken in the same pilot plant is demonstrated and the model is then used to show that AMP should enjoy at least a 15% energy advantage over MEA.
Introduction and Background
Carbon capture (CC) commonly means the removal of CO2 from gases, primarily associated with power generation plants, mostly burning coal. Fuels can be combusted at high pressure with or without oxygen enrichment, or even using pure oxygen, or in atmospheric pressure boilers. The latter is the concern of this paper. Oxidation of the carbon and hydrogen contents of fuels generates almost all the heat developed by combustion. The C:H ratio of coals is considerably higher than fuel oils, which in turn is higher than natural gas (primarily methane) so coal firing is a much more prodigious producer of CO2 than other fuels. This makes CO2 emission from conventional, coal-fired power plants the major immediate concern in carbon capture and sequestration (CCS), and CC from this kind of plant is our exclusive focus here.
Existing technological approaches to CO2 removal from conventional power plant flue gases are almost all based on solvents of one kind or another. At present, high-strength MEA is receiving the most attention and is the solvent against which other technologies are usually benchmarked. However, by no means is this the only solvent being studied. Others include 2-amino-2-methyl-1-propanol (AMP), cold ammonia, a variety of caustic-potash-neutralised amino acids, amine-promoted potassium carbonate, promoted MDEA, and physical solvents including ionic liquids, a new and interesting class of fluids being developed at the University of Notre Dame and elsewhere. Our focus in the paper is with AMP and the MEA benchmark.
CC from atmospheric-pressure flue gas presents a unique set of difficulties, almost none of which is experienced in more-conventional gas treating. The main source of these problems as far as the absorption process itself is concerned stem from two factors: the low pressure of the gas and its unavoidable oxygen content. Even a small power plant generates an enormous volume flow of flue gas at atmospheric pressure. The driving force for CO2 absorption is the CO2 partial pressure and in an atmospheric-pressure gas, this is by its nature quite low. Furthermore, the capacity of the solvent also depends directly on the CO2 partial pressure. Thus, low partial pressure means potentially slower absorption rates into a reduced-capacity solvent even if the solvent is reactive. But the story does not end here.
Except possibly in tail-gas treating, pressure drop across contacting columns is not usually of great concern in acid gas removal because the gas itself is of adequate pressure. However, most flue gases enter a scrubber with only a few inches of water column (w.c.) pressure, and at the gas and liquid loads typical of CC, 50 ft (15 m) of even large-crimp structured packing will exhibit pressure drop of many tens of inches w.c. The blower power to boost the pressure of enormous gas volumes by 50 to 100 in w.c. (125 to 250 mbar) is anything but small. As will be shown later, just the blower(s) alone can consume one or two percent of the power plant’s total output.
High gas volumes make for very large diameter columns even for power plants of quite modest output. The choice of column internals is limited to packing for several reasons. Trays exhibit much more pressure drop (per unit height of tower) than packing, and pressure drop is a major consideration in these plants. For this reason alone, trays are out of the question for scrubbers in CC applications. Additionally, however, once column diameter starts to exceed 25 or 30 ft, (10 m) the number of tray passes becomes large (five and six passes become necessary), the trays become much harder to design properly, they become increasingly difficult to support structurally, and their cost escalates. Packing becomes the only realistic option. In and of itself, packing is not a bad choice, but there are negative consequences. One is reduced CO2 absorption rate. Mass transfer resistance in the liquid phase controls CO2 absorption and liquid flow over packing is much less turbulent than flow across trays. The implication is increased resistance to CO2 absorption when using packing. There are other negatives such as an increased plugging tendency and the near impossibility of cleaning packing and repairing damage without emptying a huge column. On the positive side, packing, especially structured packing, has a very substantial capacity advantage over trays — this is one of the reasons structured packing is being more frequently specified in gas treating generally, and especially in revamps and expansions.
The energy required for solvent regeneration is the dominant operating cost associated with CC. Fast-reacting carbamate-formers like MEA and ammonia, although very effective at reacting with and removing CO2, also have high associated heats of absorption, making them energy-intensive to regenerate. However, the technology of using MEA in gas treating is rather well established, and ammonia is a very low cost chemical, so both components remain high on the list of interesting CC solvents. This is despite the difficulties of keeping highly-volatile ammonia in the system and preventing excessive MEA degradation by the oxygen always present in flue gases. Incidentally, the high regeneration-energy requirement is also an important factor in setting 85 to 90% CO2 recovery as the CC standard. As will be seen later in this paper, when recoveries are pushed upwards of 95%, the energy consumption escalates very rapidly.
The lower loading potential of carbamate-forming amines (two molecules of amine are used for each molecule of CO2 absorbed) has made the moderately hindered amine AMP more interesting. AMP is a primary amine, but the secondary methyl group shields the amino group to a significant extent, and carbamate formation is made more difficult. Thus, because the reaction product is carbonate rather than carbamate, regeneration energy ought to be lower than for MEA. Steric hindering or shielding also means that, at least in theory, each CO2 molecule uses only one AMP molecule, potentially doubling the capacity of the solvent. Whether the capacity potential can be realised in practice, of course, depends on phase equilibrium at scrubber conditions of temperature, and especially of available CO2 partial pressure.
Academic laboratories began to characterise AMP almost as soon as Exxon’s first Flexsorb patents were issued more than 20 years ago. In the literature today there are enough phase equilibrium, kinetics, and physical property data of good quality and reliability to allow AMP to be process simulated with high accuracy. AMP was added to the ProTreat amine simulator’s solvent offerings in mid-2009. In a later section, first AMP simulations will be benchmarked against pilot plant data and then AMP itself will be benchmarked against MEA via simulation of commercial-scale operations using ProTreat’s mass and heat transfer rate-based tower model.
Add your rating:
Current Rating: 1