Gas 2023 Issue

by MDEA. Figure 5 shows the same for MEA and DEA sol- vents as measured in commercial absorbers. In the case of MEA and DEA, the measured and predicted per cent remov- als match well with the exceptions where the predictions go either side. For MDEA, however, a few cases suffer from an underprediction by the model. It is important to note that most disagreement occurs for absorbers containing feed gas COS ppmv levels in the range of single to low double dig- its. In these ranges, the calculation of per cent removals will be susceptible to error magnification from uncertainties and inaccuracies of gas chromatographic measurements. In addition, going back to the COS kinetic mechanism, it is impossible for MDEA to absorb anything like over 40% of the inlet COS. However, 50% and higher removal by DEA, and even more by MEA, is to be expected (and is, in fact, observed). These expectations are supported by the relatively rapid kinetics of deprotonation to thiocarbamate for MEA and DEA and the complete absence of thiocarbamate formation by MDEA; hence slow absorption by MDEA and increasingly rapid absorption by DEA and MEA, in that order. MDEA is likely to suffer from degradation that leads to a significant accumulation of primary and secondary amine contaminants in the solvent, thus aiding and increasing the removal of COS. The original source acknowledges that several of the cases presented in this study lacked a full solvent analysis. In comparing the predicted mercaptans removal against plant data in MDEA, MEA, and DEA solvents, data collected from absorbers showed H2 S and CO 2 being simultaneously removed in much larger amounts than the mercaptans. Typical mercaptans concentrations in the feed gas to the absorbers are 100 ppmv, although there are data showing feed concentrations from 3 ppmv to 1,000 ppmv. Plots generated indicate considerable scatter in the results. Perhaps this should not be too surprising since it is next to impossible even to reproduce removal rates when inlet con- centrations are 100 ppmv and lower in many cases, and the mercaptan absorption rate is greatly affected by the absorp- tion of much more acidic H2 S and CO 2 at several orders-of- magnitude higher concentration. Small changes in acid gas absorption rates can be expected to have a profound effect on the absorption rate of such weak acids as mercaptans. However, there is a more disturbing trend – except for C1SH in MEA and DEA, simulation results are almost uniformly biased towards underprediction of mercaptans removal rates. One piece of data missing from information on mercaptans is an assessment of the pK a of each individual mercaptan and its temperature dependence. pK a has a direct effect on physical solubility and the dissociation of dissolved mercap- tan. This is directly affected by the amine in question as well as the carbamate or bicarbonate ion concentration and the bisulphide and sulphide concentrations. However, enough data are available only for pK a of EtSH, but not for MeSH, PrSH, and BuSH. As shown in Figure 6 , the pK a of these mercaptans are not monotonic with respect to carbon number, making it difficult to make a generic estima - tion of pK a based on carbon number. Only one data point in Dean’s Handbook of Organic Chemistry 1 is available for pK a of MeSH, whereas two data points, one each by Dean1 and Yabroff², have been reported for PrSH, which are inconsistent

100

0.1

0.01

0.001

0.0001

Absorbers % removal (ProTreat)

% removal (data)

Feed concentration

Figure 4 Removal of COS in MDEA solvent

a total acid gas loading of only 0.02-0.03 mol/mol of amine. This case study also shows that, in absorbers that show such a trend, the treating levels of mercaptans can be increased by raising the solvent circulation rate, which gives more absorption region for mercaptans in the column. Likewise, a reduction in the lean loading will also aid mercap- tans removal by providing more solvent capacity in the active part of the absorption profile. However, contrary to the usual expectation, oversizing the absorber with more trays/packing height would have zero to negative benefits on mercaptans removal. Accurate simulation tools can aid designers and operators to identify opportunities for meeting the treated gas specifications. For simulators to reliably predict the treat - ing of mercaptan, accurate VLE data and models that repre- sent the solubility of mercaptans in amines across a range of temperature and acid gas loading values are required. Validation against plant data Figure 4 compares the measured (dark blue columns) and simulated (light blue columns) percentages of COS removal

100

0.1