Identification of Catalytic Converter Kinetic Model Using a Genetic Algorithm Approach

The need to deliver fast-in-market and right-first-time ultra low emitting vehicles at a reasonable cost is driving the automotive industry to invest significant manpower in the computer-aided design and optimization of exhaust aftertreatment systems. To serve the above goals, an already developed engineering model for the three-way catalytic converter is linked with a genetic algorithm optimization procedure, for fast and accurate estimation of the set of tunable kinetic parameters that describe the chemical behavior of each specific washcoat formulation. The genetic algorithm-based optimization procedure utilizes a purpose-designed performance measure that allows an objective assessment of model prediction accuracy against a set of experimental data that represent the behavior of the specific washcoat formulation over a typical test procedure. The identification methodology is tested on a characteristic case study, and the best fit parameters produced demonstrate a high accuracy in matching typical test data. The results are far more accurate than those that may be obtained by manual or gradient-based tuning of the parameters. Moreover, the set of parameters identified by the GA methodology, is proven to describe in a valid way the chemical kinetic behavior of the specific catalyst. The parameter estimation methodology developed, fits in an integrated computer aided engineering methodology assisting the design optimization of catalytic exhaust systems, that extends all the way through from the model development to parameter estimation, and quality assurance of test data. * corresponding author, Laboratory of Thermodynamics & Thermal Engines, Mechanical & Industrial Engineering Department, University of Thessaly, 383 34 Volos, Greece. Tel: +30-24210-74067, Fax: +30-24210-74050, e-mail: [email protected]