Deep combustion

In this short article, we’ll try to model the combustion using deep learning. A demo applet is proposed to study combustion of 4 components formulations. Usually the chemical reaction equilibrium is found using the point at minimal free enthalpy .

To find it, we need to take into account how enthalpy and entropy of reaction products vary according to temperature and the equation of state.

For detonation modeling, a feedforward neural network approach was efficient, but for combustion the results of various machine learning algorithms, including gradient boosting, gave poor results.

Using Tensorflow and a 3 layers neural network, the model seems to be sufficiently efficient to try inference.

Hereafter, you will be able to make some calculations for Equilibrium temperatures of organic compounds based energetic formulations at 1 atm.

There server is not very powerful. So the display may take a while. Select the components ant the amount of each of them. Then RUN to get the infered equilibrium temperature


NC13.2% is nitrocellulose (celluloid) and PE is a polyethylen polymer. You may chose among the prefilled materials or add your own molecule.

This applet is a proof of concept for Research purpose. So some results may not be relevant.

Télécharger cet article au format PDF ou ePub

Laisser un commentaire

Votre adresse e-mail ne sera pas publiée. Les champs obligatoires sont indiqués avec *

Ce site utilise Akismet pour réduire les indésirables. En savoir plus sur comment les données de vos commentaires sont utilisées.