Multiscale computational simulation of multidimensional chemically reacting flow over washcoated heterogeneous catalysts

Abstract

This work addresses three distinct but interrelated topics: The simulation and evaluation of a novel ceramic microchannel heat exchanger; the adaptation of in situ adaptive tabulation for the acceleration of transient, heterogeneous chemistry in large CFD models; and the development of catalytic washcoat performance relations informed by 3D reconstructions of actual, commercial washcoats. Ceramic microchannel heat exchangers/reactors have several distinct advantages. They are cost effective, chemically inert, are capable of operating at high temperatures, and integrate well with internal catalysts. In this work, a particular design for a ceramic microchannel reactor is evaluated. A 3D computational fluid dynamics model with conjugate heat transfer and detailed kinetics for methane steam reforming was built. The model is used to evaluate the kinetic and hydraulic performance of the reactor as well as explore potential catalyst degradation mechanisms. In situ adaptive tabulation (ISAT) is a technique for accelerating the simulation of detailed kinetic mechanisms. The present work adapts this method, which was originally developed for homogeneous chemistry, to accelerate transient, heterogeneous kinetics. The approach is demonstrated on a series of transient simulations of methane steam reforming within the ceramic microchannel reactor model. For the particular case studied, 10 - 20 times speed-up factors were observed over the internal kinetics solver of ANSYS Fluent with no appreciable decrease in accuracy. Data collected from FIB-SEM imaging provided nanoscale phase information for actual, commercial washcoat structures. Algorithms were developed to process, filter, and discriminate phase information in the FIB-SEM images to facilitate 3D reconstruction. Reconstructed washcoat pores were cast in a dimensionless context and simulated to predict their catalytic performance. Performance relations, first for pores alone, and then for washcoats as a whole, were developed based on good agreement between reconstructed pores and an idealized 2D cylindrical pore model. The generalized relations suggest optimum washcoat depths based on effective kinetic rates and diffusion coefficients. Additionally, a modified washcoat factor was developed to relate the often used geometric multiplicative factor to a particular reaction-diffusion regime. These have important implications not only on the accurate simulation of washcoat performance but also for washcoat manufacturing.