PYROLYSIS MODEL PARAMETER OPTIMIZATION USING A CUSTOMIZED STOCHASTIC HILL-CLIMBER ALGORITHM AND BENCH SCALE FIRE TEST DATA

Abstract

This study examines the ability of a stochastic hill-climber algorithm to develop an input parameter set to a finite difference one-dimensional model of transient conduction with pyrolysis to match experimentally determined mass loss rates of three sample materials exposed to a range of constant incident heat flux. The results of the stochastic hill-climber algorithm developed as part of the present study are compared to results obtained with genetic algorithms. Graphical documentation of the impact of single parameter mutation is provided. Critical analysis of the physical meaning of parameter sets, and their realistic range of application, is presented. Criteria are also suggested for stability and resolution of solid phase temperature and fuel mass loss rate in an implicit Crank-Nicolson scheme with explicit treatment of the heat generation source term.