Abstract

The numerical modeling of multi-dimensional reactive flows with realistic/detailed kinetic mechanisms (including hundreds of species and thousands of reactions) represents a challenging problem and places severe demands on computational resources, because of the number of strongly coupled equations to be solved and their high non-linearity and stiffness. This talk presents and discusses the main numerical techniques adopted in the CRECK Modeling Lab at Politecnico di Milano for accelerating the CFD simulations of multi-dimensional laminar and turbulent reactive flows in the OpenFOAM framework. Starting from an introduction about the operator-splitting method, we discuss the relevance of ODE solvers for the integration of the chemical step and their links with the specific features of a detailed kinetic mechanism. Then, we proceed by introducing the SPARC (Sample-Partitioning Adaptive Reduced Chemistry) technique, a novel adaptive chemistry methodology for mitigating the computational costs of reactive-flows simulations, based on machine-learning algorithms which automatically classify the composition space via a priori defined classifiers. Finally, we present P(CA)2, a new method for Cell Agglomeration (CA), based on Principal Component Analysis (PCA), which automatically follow the temporal evolution of the thermo-chemical state of the system, identifying the most relevant species to be used in the CA process. To better show the potential benefits and limitations of the proposed acceleration techniques, several examples will be presented and discussed: laminar coflow and counterflow diffusion flames, temporally-evolving planar jet-flames, heterogeneous catalytic reactors, etc.

Alberto Cuoci

Full Professor of Chemical Engineering

My research interests include Computational Fluid Dynamics (CFD) and numerical modeling of chemical reactors.