https://doi.org/10.1016/j.seppur.2020.118193
“There is surprisingly a fair amount of data concerning the mixture of methanol plus MEA, also known as the original Amisol® formulation. From bench-scale absorber columns [9], [10], [11], [12], [207], [208], to bench-scale absorber strippers [209], to complete bench scale-absorber-desorber loops [13], [14], one could easily claim that this process is quite well understood. However, it is difficult to make a case for reviving the Amisol® solvent after it has been picked up industrially and steadily lost favor over the years.
A single instance of a bench-scale absorber operating with a novel solvent, NMP–MEA–water, has been found in Tan et al. [16]. However, the authors did not record data such as temperature profiles or mass transfer rates for their solvents, simply reporting CO2 capture efficiencies as function of inlet gas pressures. Conversely, there are interesting studies showing the operation of water-lean solvents in purely theoretical basis through the use of computer simulations [210], [211], [212], [213], and thus lacking the hard credibility of a true report on a pilot plant campaign.
With that being said, we find it particularly instructive to fully reproduce in Table 9 the only comprehensive pilot plant data that we could find regarding CO2 absorption with water-lean solvents. These values were reported by Semenova and Leites [17] in Russian. We shall analyze some aspects of their data and how they relate to what has been discussed so far.
Table 9. Data regarding pilot plant operation with water-lean solvents as reported in Semenova and Leites [17]. The water-lean solvents are based on MEA plus N-methyl-2-pyrrolidone, tetrahydrofurfuryl alcohol and ethylene glycol.
| Parameter | H2O | NMP | THFA | MEG |
|---|---|---|---|---|
| Solvent concentrations/%wt. | ||||
| MEA | 20 | 19 | 20 | 21 |
| H2O | 80 | 3 | 9 | 6 |
| Diluent | 0 | 78 | 71 | 73 |
| CO2 concentrations in gas/%v. | ||||
| Flue gas | 20.4 | 19.2 | 19.1 | 20.8 |
| Treated gas | 2.6 | 1.3 | 1.7 | 9.1 |
| CO2 concentrations in solvent/mol CO2∙mol MEA−1 | ||||
| Lean solvent | 0.210 | 0.070 | 0.080 | 0.143 |
| Rich solvent | 0.423 | 0.348 | 0.286 | 0.331 |
| Solvent temperature in absorber/°C | ||||
| Inlet | 23 | 39 | 40 | 38 |
| Outlet | 39 | 74 | 70 | 57 |
| Temperature difference in heat exchanger/°C | ||||
| Hot end | 12 | 25 | 23 | 24 |
| Cold end | 9 | 4 | 10 | 15 |
| Solvent temperature in desorber/°C | ||||
| Inlet | 103 | 110 | 112 | 113 |
| Bottom | 115 | 135 | 135 | 140 |
| Top | 99 | 103 | 85 | 105 |
| Heat consumption/MJ∙m3 CO2−1* | ||||
| Sensible heat | 2.24 | 0.544 | 1.34 | 3.10 |
| Latent heat | 2.39 | 0.335 | 3.25 | 0.109 |
| Absorption heat | 3.35 | 4.48 | 3.25 | 3.77 |
| Overall heat | 7.98 | 5.34 | 5.38 | 6.97 |
*1 MJ∙m3 CO2−1 ≈ 0.509 GJ∙ton CO2−1 (assuming 1 mol CO2 = 22.4 l CO2).
Before proceeding to a proper analysis of the results, one should notice that the water-lean solvent based on ethylene glycol was not able to deliver a CO2 removal comparable to that of the other solvents.
All water-lean solvents analyzed by Semenova and Leites [17] achieved rich loadings below that of aqueous MEA, which is consistent with what has been discussed in Section 4.1. Solvent temperatures were overall higher in water-lean solvents, a fact that perhaps stems from the authors aiming to compensate for the low volatility of these absorbents in the reboiler as discussed in Section 8.2. As such, the temperatures of the rich solvent leaving the absorber were higher in water-lean solvents, and thus the temperatures of the desorber feeds were all comparable, even though the cross-heat exchanger efficiencies (as evaluated by the temperature differences in the hot and cold ends) seem to have been indeed lower for water-lean solvents.
The overall heat duties in aqueous MEA and in solvents with NMP, THFA and MEG as reported by Semenova and Leites [17] are respectively 4.1, 2.7, 2.7 and 3.5 GJ∙ton CO2−1. Moreover, although NMP and THFA have reduced the overall reboiler heat duties, they did so for different reasons. Shifting from water to NMP provoked a reduction in latent heat, whereas the shift to THFA brought a reduction in sensible heat, both leading to overall lower parasitic costs of solvent regeneration. Meanwhile, the heat of desorption of CO2 is quite comparable among all solvents, with those of water-lean formulations with NMP and MEG being higher than that of aqueous MEA, as discussed in Section 8.1.
There is a final aspect to be mentioned. Recent CO2 capture plants have been slowly approaching the 2 GJ∙ton CO2−1 mark for reboiler duties [214]. Rochelle et al. [215] report a similar performance during pilot plant operations. They achieve these reboiler duties not solely because of their amine solvent (aqueous 5 m piperazine), but because of intelligent process modifications. Plant design seems to be capable of providing an effective reduction in energy costs. This poses a complicated problem. Not all researchers have access to these improved equipments for CO2 capture, but one still has to compare very distinct solvents somehow. And though it is tempting to perform these comparisons in a purely percentual basis, the fact is simply that one cannot analyze solvents in conditions that do not enable their peak performance. As one might argue that shifting from aqueous MEA to a hybrid solvent with NMP reduces reboiler duties in 33% based on the data from Semenova and Leites [17], another might correctly reply that there is margin to reduce these duties in aqueous MEA with clever process modifications (for example, their sensible heat duties are unreasonably high). Whether these process modifications would work equally as well in water-lean solvents remains to be seen. We do not believe that pointing this out is being particularly harsh on new solvents. Quoting Leites [31] himself, thermodynamical analyses of CO2 capture processes show that sensible and latent heat expenses can be reduced almost to zero with sophisticated plant designs. Rochelle et al. [215] prove this point by almost eliminating latent heat duties with the advanced flash stripper.
To quickly summarize the conclusions of Section 8:
- •
-
Water-lean solvents will generally incur in similar heats of CO2 absorption as their aqueous counterparts;
- •
-
Water-lean solvents can reduce latent heat expenditures if they are less volatile than water, though they will probably result in more inefficient heat transfer phenomena;
- •
-
There is indeed a potential for reduction of overall energy duties in CO2 capture plants with water-lean solvents, yet it is discussable if these same reductions could not be obtained with proper reconfiguration of the process operating with aqueous amines.”