Optimization of material properties for CFD Fire modelling

This project was completed during the intern ship in Victoria University, Melbourne, Australia in the duration of 3 months.

Introduction : The material data obtained from various standards test procedures, accumulates the errors from each of the tests. To prepare a standard material properties data, this error has to be reduced or eliminated. The GA and ANN for this purpose were a favourable choice, to obtain the optimized material properties and reaction parameters data, which were obtained experimentally, so that these properties could further be used for CFD Fire Modelling.

Fire Modelling : 

These are the few pictures of fire modelling and fire dynamic simulations. 

Work Done: GPYRO, a pyrolysis software, and FDS was used for fire modelling. The material properties obtained from various test results were optimized the GA routine in the GPYRO code and then the optimized results were used in FDS for fire modelling of various kinds of materials which include foam, wood, etc. The FORTRAN code of GYPRO was studied extensively, with the use of GA in it, to check the convergence problem and to carry out the transformation from Genetic Algorithm to Artificial Neural Networks. The problem of failure of GPYRO code output to converge to an optimal solution was solved. The various cases were successfully tested for the edited code and the code converged.

Want to know more about GA and ANN?