(307f) Multiobjective Optimization of Autothermal Catalytic Membrane Reactor Using Genetic Algorithm

Authors: 
Alwan, G. M. - Presenter, University of Technology
Aldahhan, M. H. - Presenter, Missouri University of Science and Technology


Membrane reactors are suggested to combine reaction with separation to enhance the conversion and yield by overcoming the thermodynamics limitation. Many variables affect the performance of such reactors.Therefore, reactor system needs  to be optimized to  achieve  the maximum benefit. Abo- ghander (2010) studied the optimization of autothermal membrane reactor coupling the  dehydrogenation of ethylbenzene to styrene with the hydrogenation of nitrobenzene to aniline. The total number of the decision variables considered in the optimization problem was 12, representing a set of operating and design parameters.The problem  was solved numerically  by two deterministic multiobjective  approaches: the Normalized Normal Constraint (NNC) method and Normal Boundary Intersection (NBI) method.However, the membrane reactor  for these reactions is highly non-linear and in addition Abo-ghander (2010) used the  dynamic  mode to correlate the objectives with decision variables.  For highly nonlinear system, the stochastic optimization algorithm has been found better than  the deterministic approachesed  by Abo-ghander(2010). Deterministic search techniques, use  characteristics of the problem(e.g.,Jacobian, Hessian) to determine the next sampling point   while stochastic search techniques do not need such information.Instead,the next step in  stochastic methods would be determined based on statistical sampling rules rather than aset  of rigorous deterministic. Furthermore steady state or static formulations that  correlate the  objectives with decision variables are more reliable and accurate compared the dynamic mode  with the process of long cycle time of reaction. Accordingly, the focus of this work is to extend  and improve the optimization effort of Abo-ghander(2010)by  implementing stochastic  optimization  using genetic algorithm.Reducing the number of the decision variables to 6 instead  of 12 used  by  Abo-ghandar(2010) and using steady state(static) correlations  that correlate  the objectives; conversion of nitrobenzene and yield  of styrene ,with  the decision variables; ethyl and nitro benzene  flow rate,pressure and temperature on the shell and tubes side. The noticeable improvement has been found where the yield of styrene has increased within  range of 74- 100%  compared to  49-98% obtained by the algorithms of the optimization used by Abo-ghander(2010), and the finidings will be discussed in this presentation.

Reference:

Abo-ghander, N.S.M,”Coupling Dehydrogenation of Ethylbenzene with Hydrogenation of Nitrobenzene in an Autothermal Catalytic Membrane Reactor”, Ph.D thesis, Chemical and  Biological Engineering Department, University of British Columbia, (2010).