Comparisons of Amine Solvents for Post-Combustion CO2 Capture: A Multi-Objective Analysis Approach
Anita S. Lee, John C. Eslick, David C. Miller and John R. Kitchin, “Comparisons of Amine Solvents for Post-Combustion CO2 Capture: A Multi-Objective Analysis Approach”, International Journal of Greenhouse Gas Control 18, 68-74 (2013). https://doi.org/10.1016/j.ijggc.2013.06.020
This summary was written by a language model from the paper and has not been checked line by line. Rely on the published paper for details.
Note on the source text: the extracted text is the submitted preprint. The tables are readable, but the figures (Pareto fronts, heat duty and working capacity plots) are not, so trends from figures are taken from the prose.
Question
Solvent discovery for amine scrubbing happens mostly at the laboratory scale, where one measures reaction enthalpy, rates and CO2:amine ratios, yet these quantities do not map simply onto the energy penalty of a full capture plant. Earlier process comparisons of different amines often ran every solvent through a flowsheet tuned for 30 wt% monoethanolamine (MEA), or optimized only cost without a capture target. The paper asks how to compare solvents fairly at the process scale, with each one operated under the design and conditions that suit it best (Lee et al. 2013).
Approach
A fixed absorber/stripper flowsheet plus a ten-stage CO2 compression train was built in Aspen Plus v7.2 with the electrolyte-NRTL model, using flue gas from the NETL baseline 550 MW supercritical pulverized coal plant and a 90% capture specification. Each amine (MEA, diethanolamine DEA, and the hindered 2-amino-2-methyl-1-propanol AMP) was held at 11 mol% amine, equivalent to 30 wt% MEA, so the number of reacting molecules was constant and differences reflect chemistry alone. Both columns were rate-based RadFrac models with structured packing, sized for 80% flooding, with an absorber intercooler and a lean-rich exchanger at a 13.9 C approach; CO2 was compressed to 15.27 MPa and 310 K. Two objectives were used: net power output of the plant (gross output reduced by the low-pressure steam drawn for the reboiler, from a surrogate model, minus compression and pumping loads) and capital cost of the capture and compression units (Lang method, solvent cost excluded). The NSGA-II genetic algorithm in modeFRONTIER, linked to Aspen through the Sinter interface, varied ten design and operating variables (lean amine temperature and loading, column packing heights, intercooler settings, stripper condenser pressure and temperature, compressor cooler temperatures), starting from 25 Latin hypercube points and running 150 generations, about a week per solvent (Lee et al. 2013).
Key findings
For MEA the search produced roughly 3000 converged feasible designs; failures were non-converged runs or cases whose reboiler temperature exceeded the available steam temperature (403.65 K). The Pareto front shows a minimum capital cost for 90% capture, a ceiling on net power, and diminishing returns in efficiency for added capital. The maximum power design gave 394.2 MW at $517.3M with lean loading 0.188, condenser pressure 222 kPa and working capacity 0.315, while the minimum cost design gave 363.5 MW at $244.1M with short columns and lower stripper pressure. Moderate lean loading and high stripper pressure agree with earlier heuristics (Lee et al. 2013).
Simply swapping DEA or AMP into the MEA-optimal design made both look worse on power (373 and 364 MW versus 394 MW), though cheaper ($362M and $353M versus $517M), which would wrongly crown MEA. With solvent-specific optimization, both DEA and AMP had Pareto fronts reaching higher net power at lower capital cost than MEA, and DEA was judged the most promising. The gains trace to a lower reboiler heat duty, which frees steam for power generation and shrinks the kettle reboiler that dominates stripper cost (Lee et al. 2013).
Splitting reboiler duty into reaction enthalpy (from literature measurements) and a process term dominated by stripping steam, the authors find that DEA’s higher working capacity explains its lower duty. They attribute this to a less stable carbamate in DEA than MEA, shifting it toward the bicarbonate route. AMP, which reacts only via bicarbonate, has reaction enthalpy and kinetics near DEA but working capacity near MEA, yet needs less stripping steam (Lee et al. 2013).
Significance
The work offers a template for judging new solvents by plant-level metrics rather than lab-scale ones, and shows that fixed-process comparisons can invert solvent rankings. It suggests two design directions: destabilizing carbamate formation in unhindered amines, and finding hindered or tertiary amines with DEA-like working capacity. The authors caution that quantitative results depend on the quality of the Aspen thermodynamic models for each amine (Lee et al. 2013).