Application of the simulated annealing algorithm to the correlated WMP radiation model for flames

Several alternatives have been proposed to model the radiation profile emitted by flames. Among them there is the weighted-multi-point-source-model (WMP), which is able to provide good predictions for medium and far distances. Despite being a simple model, the determination of its parameters can be difficult, which creates the opportunity to apply optimisation methods in order to find the best answer. Moreover, correlations of these parameters as function of flame variables can be applied aiming to relate the model to the physical phenomenon, making the problem even harder to solve. Previous studies applied the Generalised Extreme Optimisation (GEO) to find the best possible combinations of these parameters, but there is a chance that other methods could improve the results considerably. This study has the objective to further explore the WMP model as formulated previously in the literature and apply the Simulated Annealing optimisation method to assess if the answers could be improved. The same correlation and experimental data were used as well. A total of 77 different cases were explored over 50 runs each, configuring the answers statistically significant. The SA method had better performance, showing similar average results for the objective function, but overall best answers.