Optimisation of Reaction Mechanisms for Aviation Fuels Using a Multi-objective Genetic Algorithm

Abstract

Abstract. In this study a multi-objective genetic algorithm approach is developed for determining new reaction rate parameters for the combustion of kerosene/air mixtures. The multi-objective structure of the genetic algorithm employed allows for the incorporation of both perfectly stirred reactor and laminar premixed flame data into the inversion process, thus producing more efficient reaction mechanisms.

An Informed Operator Based Genetic Algorithm for Tuning the Reaction Rate Parameters of Chemical Kinetics Mechanisms

“…It can be seen that if a mechanism is only optimised for SENKIN then it does not produce good results for PSR and PREMIX. However, mechanisms optimised for PSR and/or PREMIX, produce reasonably accurate results for SENKIN simulations as well, see [2]. Similarly, if an objective function based only on SENKIN evaluations is used in the informed operators then the search is possibly guided toward some false optima which correspond to reaction mechanisms efficient for ignition delay time predictions but not for species concentrations.…”

“…In this study we use a constrained genetic algorithm similar to the one proposed in [2]. The genetic operators and the parameters used for this genetic algorithm were taken to be population size n pop = 20, number of offspring n child = 30, non-uniform arithmetic crossover, crossover probability p c = 0.65, tournament selection, tournament size k = 2, tournament probability p t = 0.8, non-uniform mutation, mutation probability p m = 0.5.…”

“…In particular, PREMIX evaluations require a large CPU time. For complex hydrocarbon fuels the CPU time can build up to several months even if only one PREMIX operating condition is used, see [2].…”

“…A powerful inversion technique, based on a genetic algorithm, was developed in [2] in order to determine the reaction rate parameters for the combustion of a kerosene/air mixture in a perfectly stirred reactor (PSR) and in premixed laminar flames (PREMIX). In order to obtain a reaction mechanism which can be used for a wide range of practical problems we also extend the work of [2] to include data from ignition delay time measurements.…”

“…In order to obtain a reaction mechanism which can be used for a wide range of practical problems we also extend the work of [2] to include data from ignition delay time measurements. It is important to use this approach since in many practical situations the amount of experimental data available is limited and it is important to use all the available experimental data, which may come from various types of measurements.…”