Computational optimization of catalyst distributions at the nano-scale

Abstract

Catalysis is a key phenomenon in a great number of energy processes, including feedstock conversion, tar cracking, emission abatement and optimizations of energy use. Within heterogeneous, catalytic nano-scale systems, the chemical reactions typically proceed at very high rates at a gas-solid interface. However, the statistical uncertainties characteristic of molecular processes pose efficiency problems for computational optimizations of such nanoscale systems. The present work investigates the performance of a Direct Simulation Monte Carlo (DSMC) code with a stochastic optimization heuristic for evaluations of an optimal catalyst distribution. The DSMC code treats molecular motion with homogeneous and heterogeneous chemical reactions in wall-bounded systems and algorithms have been devised that allow optimization of the distribution of a catalytically active material within a threedimensional duct (e.g. a pore). The objective function is the outlet concentration of computational molecules that have interacted with the catalytically active surface, and the optimization method used is simulated annealing. The application of a stochastic optimization heuristic is shown to be more efficient within the present DSMC framework than using a macroscopic overlay method. Furthermore, it is shown that the performance of the developed method is superior to that of a gradient search method for the current class of problems. Finally, the advantages and disadvantages of different types of objective functions are discussed.

Clean, efficient and affordable energy for a sustainable future

“…In paper [136], the performance of a Direct Simulation Monte Carlo (DSMC) code with a stochastic optimization heuristic for evaluations of an optimal catalyst distribution was investigated, which is useful to improve the performance of processes for biomass conversion. This study is important because the statistical uncertainties characteristic of molecular processes pose efficiency problems for computational optimizations at the nano-scale.…”