Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
173002 | Computers & Chemical Engineering | 2011 | 11 Pages |
In this article, the In Situ Adaptive Tabulation (ISAT) procedure, originally developed for the efficient computation of homogeneous reactions in chemically reacting flows, is adapted and demonstrated for reacting flow computations with complex heterogeneous (or surface) reactions. The treatment of heterogeneous reactions within a reacting flow calculation requires solution of a set of nonlinear differential algebraic equations at boundary faces/nodes, as opposed to the solution of an initial value problem for which the original ISAT procedure was developed. The modified ISAT algorithm, referred to as ISAT-S, is coupled to a three-dimensional unstructured reacting flow solver, and strategies for maximizing efficiency without hampering accuracy and convergence are developed. These include use of multiple binary tables, use of dynamic tolerance values to control errors, and periodic deletion and/or re-creation of the binary tables. The new procedure is demonstrated for steady-state catalytic combustion of a methane–air mixture on platinum using a 24-step reaction mechanism with 19 species, and for steady-state three-way catalytic conversion using a 61-step mechanism with 34 species. Both reaction mechanisms are first tested in simple 3D channel geometry with reacting walls, and the impact of various ISAT parameters is investigated. It is found that the temperature of the reacting wall dictates the retrieval rate from the ISAT table. As a final step, the catalytic combustion mechanism is demonstrated in an laboratory-scale monolithic catalytic converter geometry with 57 channels discretized using 354,300 control volumes (4.6 million unknowns) after employing quarter symmetry. For this particular case, the use of ISAT-S resulted in reduction of the overall CPU time from 19.3 to 13.6 h. For all of the cases considered, the reduction in the time taken to perform surface chemistry calculations alone was found to be a factor of 5–11.