SEPARATIONS Pressure Swing Adsorption Cycles for Improved Solvent Vapor Enrichment Yujun Liu and James A. Ritter Dept. of Chemical Engineering, Swearingen Engineering Center, University of South Carolina, Columbia, SC 29208 Bal K. Kaul Exxon Research and Engineering Company, Florham Park, NJ 07932 ( ) ( ) A pressure swing adsorption PSA  sol®ent ®apor reco®ery SVR process simulator was used to in®estigate new PSA cycle configurations designed for higher sol®ent ®apor enrichment. These cycles were modifications of the four-step Skarstrom cycle used com - mercially for PSA-SVR and include the addition of a cocurrent blowdown step, and combinations of cocurrent blowdown and continuous rbatch reflux steps. The reco® ery of gasoline ®apor from tank filling operations was simulated with n-butane, n-heptane, and nitrogen as representati®es of the light and hea®y components in gasoline ®apor, and carrier gas, respecti®ely. Adding a cocurrent blowdown step increased the sol®ent ®apor enrichment, depending mainly on the step ending pressure, not the step time. Both the continuous and batch reflux steps also increased the sol®ent ®apor enrichment, but at the expense of an increased bed capacity factor. For similar increases in the sol®ent ®apor enrichment, batch reflux led to significantly smaller bed capacity factors com - pared to continuous reflux and was thus superior for PSA-SVR. O®erall PSA-SVR pro- cess performance impro®ed markedly by adding cocurrent blowdown and batch reflux steps compared to the con®entional four-step cycle. Introduction PSA has established itself as a competitive technology in the important area of SVR. Existing and potential commer- cial applications include the recovery of many different sol- Ž . vent vapors Holman and Hill, 1992; Hall and Larrinage, 1993 Ž . and hydrocarbon gasoline vapors Pezolt et al., 1997 , with more than 100 gasoline vapor recovery units already operat- ing worldwide. Related theoretical and experimental studies from academia have paralleled this rapidly growing interest Ž in PSA for SVR Suh and Wankat, 1989a; Ritter and Yang, 1991a,b; LeVan, 1995; Liu and Ritter, 1996; 1998 Pigorini and LeVan, 1997; Liu and Ritter, 1997a,b; Subramanian and Ritter, 1997; Ritter and Liu, 1998; Ritter et al., 1998; Subra- manian and Ritter, 1998; Liu et al., 1998a,b; Subramanian et . al., 1999; Liu et al., 1999a,b, 2000 . PSA for SVR is different from conventional PSA processes Ž in that the desirable product is a heavy component or com- . Ž ponents ; however, the purity of the light component usually Correspondence concerning this article should be addressed to J. A. Ritter. . air in SVR processes also has imposed constraints that must be met due to environmental regulations. So the process per- formance is judged not only by the purity, recovery, and pro- Ž ductivity of the light component like in conventional PSA . processes , but also by the enrichment and recovery of the Ž.Ž . heavy component s Liu and Ritter, 1996 . Moreover, almost all of the PSA-SVR processes commercialized so far have uti- Ž lized the simple, twin-bed, Skarstrom-type cycle Holman and . Hill, 1992; Hall and Larrinage, 1993; Pezolt et al., 1997 . While this cycle easily handles feed mixtures to meet the strictest known emission regulations, the enrichment of the Ž. heavy component s is usually far below the thermodynamic Ž limitation, which is governed by the pressure ratio Sub- . ramanian and Ritter, 1997 . Low enrichments necessarily in- crease the cost of downstream processes designed for re- claiming the heavy components. Verylittle effort has been put forth in the development of new PSA cycles that focus on increasing the heavy compo- nent enrichment. A few exceptions include the works by Suh March 2000 Vol. 46, No. 3 AIChE Journal 540