Pergamon Chemical Engineerina Science, Vol. 51, No. 4, pp. 535-547, 1996 Copyright © 1996 Elsevier Science Lid Printed in Great Britain. All rights reserved 0009-2509/96 $15.00 + 0.00 0009-2509(95)00285-5 OBSERVATIONS, MODELING AND OPTIMIZATION OF YIELD, SELECTIVITY AND ACTIVITY DUllING DEHYDROGENATION OF ISOBUTANE AND PROPANE IN A Pd MEMBRANE REACTOR MOSHE SHEINTUCH* Department of Chemical Engineering, Technion, Israel Institute of Technology, Haifa, Israel 32000 and RALPH M. DESSAU Mobil Research and Development Corporation, Central Research Lab., Princeton, NJ, U.S.A. (First received 14 April 1995; revised manuscript received and accepted 8 August 1995) Abstract Dehydrogenation of isobutane and propane was carried out in a membrane reactor made of a Pd/Ru (or Pd/Ag) tube packed with a supported Pt catalyst. The shell side was swept by a stream of nitrogen or its mixture with hydrogen. Significantgains in yield were achieved by separating the hydrogen through the selective Pd membrane: up to 76% butene at 500°C (compared with 32% in equilibrium) and 70% propene at 550°C (23% at equ.). The attained yields, however, were limited at low feed rates by suppressed catalyst activity in the absence of hydrogen. To avoid low activity and fast aging, hydrogen concentration should be kept at about 2% by adjusting the shell or tube flow rates. Fast deactivation was observed with high ratios of shell to tube flow rates. The degree of cracking and of isomerisation increases with conversion. Temperature should be kept below 500°C, during butane dehydrogenation, to avoid cracking and fast aging. Yields under high pressures (18 psi for isobutane and 100 psi for propane) were similar to those obtained under atmospheric conditions. Operation under pressure may be advantageous as high purity hydrogen can be produced. The yield dependence on feed rate and on hydrogen shell-side pressure were adequately described (at 500°C)by a simple model, that incorporates a three-parameter rate expression, that accounts for the accelerating role of hydrogen pressure. The degree of cracking and isomerisation were adequately described by a single-parameter rate expression which assumes that the main and side reactions occur on the same sites. The model was optimized to determine the feed and shell flow rates which maximizethe yield.The optimization suggests that, in the present design,the yield cannot be improved significantlybeyond 90%, but that almost complete conversion could be achieved when the reactor profile of hydrogen pressure is optimized. INTRODUCTION Intensive research in the past decade into membrane reactors has helped to characterize the advantages and pitfalls of the various technologies l-see recent reviews by Hsieh (1991), Shu et al. (1991) and Saracco and Specchia (1994)]. Initial efforts, primarily by the group of Gryzanov in the USSR, focused on using Pd membranes for selective transport of hydrogen and demonstrated the feasibility of this technology, show- ing usually only a small conversion. A large number of hydrogenation, dehydrogenation and dehydrocon- densation reactions were tested (see listings in the reviews cited above). Recent works were aimed at attaining high conversions while employing other, less expensive, membranes. Many studies employed ce- ramic membranes of small pore size (as low as 40/~) which provide only partially selective transport, based *Corresponding author. on molecular weight (Knudsen diffusion); the yield in such units is limited by transport of product to the shell side. Very recent works attempted to combine the mechanical strength of ceramics with the selective transport of Pd by impregnating the ceramic mem- brane with Pd crystallites that plug the pores. Various attempts of attaining high-conversions in dehydrogenation reactions have concluded that the separation of hydrogen, and the high residence times associated with membrane reactors, induce problems of selectivity and of catalyst activity that are more severe than those in regular catalytic reactors. These effects were especially evident during dehydrogena- tion of short alkanes in membrane reactors, since these reactions are conducted at relatively high tem- peratures. Problems of activity and selectivity are less pronounced during low temperature dehydrogena- tion. Many studies have demonstrated almost com- plete cyclohexane conversion, with good selectivity and stable activity, during its dehydrogenation which 535