Gas 2025 Issue

3.0

10 , 000

Gas from 1/4 of area Gas from 3/4 of area Gas from absorber

Excess gas

1 , 000

Excess solvent

2.5

100

Uniform distribution

2.0

Solvent ow to 1/4 of area (gpm)

where one-quarter of the area carries the excess. The extent of liquid maldistribution required to produce equal pressure drops across the two parallel columns is then cal- culated. In the second scenario, varying degrees of liquid maldistribution are considered, and the corresponding gas maldistribution is calculated in the same way. Figure 1 shows a typical parallel-column simulation model when 27.5% vs 25% of the total gas flow is to one-quarter of the absorber cross-section. Note that this seemingly small amount of gas maldistribution pushes almost all the liquid from that part of the column so that only 3.7% vs 25% of the liquid contacts this gas. A corre- sponding case is one in which the liquid flow to one-quar - ter of the absorber cross-section is purposely raised from 25% to 30% of the total liquid flow (45 vs 37.5 USgpm). In this case, the increased solvent flow forces only quite a small corresponding reduction in gas flow to maintain a balanced pressure drop, and the effect on H 2S treating is not nearly as significant. It is, perhaps, worth noting that maldistributed gas seems to have a more severe impact on performance than a similar amount of liquid maldistri- bution. This makes sense when one realises that pressure drop is a lot more responsive to gas flow than liquid rate. Liquid films on packing surfaces somewhat narrow the size of passages available for gas flow; however, it is the gas flow that directly causes pressure drop. Balancing gas and solvent flows The balanced hydraulic distribution of gas and solvent plays an important role in setting TGTU absorber perfor - mance. Since the gas flow splits into two parts at the base of the actual absorber, the split streams in the model pass through two parallel columns and then recombine at the top of the absorber (solvent flow splits similarly). The pres - sure drop across these two columns must be identical; oth - erwise, the flows will redistribute. Figure 2 shows how the gas and solvent flow rates are related in the one-quarter- area column under the overall conditions of Table 1. Similar calculations have been done for flows that maldistribute to each half the absorber area. The results are substantially the same, so they are omitted here for the sake of brevity. Figure 2 How the flow rate of one phase responds to flow rate changes in the other phase when the absorber is split into ¼- and ¾-area parallel columns

The effect on mass transfer performance is qualitatively the same, too. It is useful to appreciate from Figure 3 that when, for example, a poorly constructed or maintained distributor dis- charges too high a solvent flow into one part of the column, the gas flow there sees more resistance, and it will respond by reducing its local flow rate and divert the excess to some another part of the column. The net result is that the higher solvent rate combines with the lower gas rate, so the liquid-to-gas (L/G) ratio becomes even more elevated; thus, one might expect some - what better treatment (lower residual H2S). However, the L/G ratio decreases in the other part of the column, which can be expected to result in poorer treatment (increased H 2S leak). The question then is: What is the net effect on the recombined gas (on the real absorber’s performance) of these competing effects? Effect of maldistribution on mass transfer performance One set of simulations was done at several solvent rates deviating positively from the uniform distribution value (37.5 gpm flow to one-quarter of the area of the actual absorber), with the gas rate to that quarter adjusted to give equal pres- sure drop across the one-quarter- and three-quarter-area sections. A second set of simulations was done with gas rates deviating positively from the uniform distribution value of 2.5 MMscfd. Here, equal pressure drops across the parallel columns required the solvent rate to be reduced. The results of all simulations were consolidated into the combined plots in Figures 3 and 4 . Figure 3 shows the H 2S content of the treated gas from the one-quarter- and three- quarter-area columns and the treated gas from the com- bined columns (from the real absorber, as a function of the actual solvent flow to the one-quarter-area column). Figure 4 combines liquid and gas maldistribution into the L/G ratio, which explicitly contains the effect of liquid maldistribution and the concomitant gas maldistribution. As expected, optimal absorber performance is real- ised when both phases are perfectly distributed over the Figure 3 How solvent maldistribution in terms of actual flow rate (gpm) to the ¼-area column affects H 2S content of the treated gas from the ¼- and ¾-area parallel columns and from these columns when combined into the real absorber

Gas 2025