444981 Stable Membrane Processes for Enhanced Olefin-Paraffin Separations in Refineries and Polyolefin Plant

Monday, April 11, 2016: 2:00 PM
344AB (Hilton Americas - Houston)
Hannah Murnen1, Sudip Majumdar2, Yosuke Koizumi2, Ken Loprete2, Kenneth J. Pennisi2 and Stuart Nemser2, (1)Chemical Engineering, University of California, Berkeley, Berkeley, CA, (2)Compact Membrane Systems, Wilmington, DE

Olefins are primary feedstocks for manufacture of petrochemicals. There is significant economic benefit in increasing (1) capacity of existing distillation operations while simultaneously reducing costs and (2) minimizing purge from polyolefin plants. Existing distillation for separation of olefins from paraffins (e.g., propylene from propane) is very difficult requiring large distillation columns and significant energy consumption. In principle membranes are small and should be easily retrofitable into distillation columns.

Membrane processes have been previously evaluated for separating olefins from paraffins.  In many cases, silver salt facilitating agents have been incorporated into membranes to preferentially transport olefins.  Historically with earlier membranes, while good separations have been initially demonstrated in the laboratory, membrane stability problems have prevented development of commercial systems.

Using customized amorphous fluoropolymer (CAF) membranes, Compact Membrane Systems has developed stable olefin-paraffin separation membranes with very good propylene flux and propylene/propane selectivity. The permeance and selectivity combination of the membranes are above the industry Robeson curve.  In addition results were stable when exposed to high levels of conventional foulants (e.g., sulfur containing gases) in the laboratory. Using basic data from these laboratory results payback times for hybrid distillation-membrane processes were calculated to be less than 1-2 years in refinery operations. Similarly related to purge lines in polyolefin plants payback times are less than 12 months. Experimental data and modeling results will be discussed more extensively at the presentation.

Extended Abstract: File Uploaded