Vented gas explosion overpressure calculation based on a multi-layered neural network

The case of a gas explosion occurring in geometrically simple enclosure, equipped with vent is considered. It well known the scientific community that calculation reduced overpressure, determinant safety studies, not trivial. Not only there strong dependency on chemical kinetics combustible but also enclosure geometry, fluid flow, mechanical behaviour, shape, etc … As result, modelling physics at stake challenging, wide range models are proposed literature and this reference situation still object extensive research. A new simulation approach ignoring large part underlying investigated. based use an artificial neural network (ANN). focus given method results obtained ANN rather than itself. Our observations discussed within scope industrial problems. Calculations performed relatively official TensorFlow tutorial, vented database containing 268 tests, led to surprisingly good considering implementation efforts. tool might look promising far from being as trivial it seems first glance: simulations type model must be examined greatest care initial data base very controlled. Routes enhance perform relevant analyses predictions.