Two techniques, Artificial Neural Network (ANN) and Repro-Modelling (RM), are successfully used to represent the chemistry in turbulent combustion simulations. This is a novel application of both methods which show satisfactory accuracy in representing the chemical source term, and good ability in capturing the general behaviour of chemical reactions. The ANN model, however exhibits better generalization features over those of the RM approach. In terms of computational performance, the memory demand for handling the chemistry term is practically negligible for both methods. The total Central Processing Unit (CPU) time for Monte Carlo simulation of turbulent jet diffusion flame, which is dictated mainly by the time required to resolve the chemical reactions, is smaller if the RM method is used to represent the chemistry, in comparison to the time required by the ANN model. The potential and capabilities of these techniques are extendable to handle the chemistry of different fuels, and more complex chemical mechanisms.
|Number of pages||6|
|Publication status||Published - Dec 1 1995|
|Event||Proceedings of the 1995 IEEE International Conference on Neural Networks. Part 1 (of 6) - Perth, Aust|
Duration: Nov 27 1995 → Dec 1 1995
|Other||Proceedings of the 1995 IEEE International Conference on Neural Networks. Part 1 (of 6)|
|Period||11/27/95 → 12/1/95|
ASJC Scopus subject areas