“Robust optimization of a combustion chamber”
by Vincent Baudoui, CENAERO FRANCE
Robust optimization aims at taking uncertainties into account in order to get optimal solutions that are relatively insensitive to parameter variations. For instance, the minimization of pollutant emissions for a turbomachine combustion chamber whose injectors can clog unpredictably is presented here. This multiobjective optimization is led thanks to 2D axisymmetric numerical simulations that are quite expensive. Fortunately, the use of surrogate models keeps the number of simulations as low as possible and eases robustness estimations relatively to the problem uncertainties. Model error is managed thanks to a new design of experiment enrichment strategy called PareBRO (for “Pareto Band Robust Optimization”) which allows finding several robust designs of the combustion chamber studied.