Data driven forecast of droplet combustion

Abstract

The characteristics of a diffusion flame resulting from the gasification of a condensed fuel are predicted from the synthesis of simple models and data. Combustion of a droplet in microgravity is used as a canonical configuration to illustrate the methodology. The simplicity of the spherical configuration and the detail of the measurements make the available experimental data ideal for this study. The approach followed combines the classical analytical solution first proposed by Spalding to describe the condensed phase gasification with a numerical method that describes the gas phase.Available data on flame geometry and regression rates are used to initialize the model and produce adequate predictions of the time evolution of all relevant variables. The method was shown to make proper predictions under numerous configurations and with very small computational cost.

Data driven forecast of concurrent flame spread in micro-gravity

“…In the present study, the work of Xi et al [22] is extended to the two-dimensional situation of concurrent flame spread over a flat surface. The analytical and numerical models implemented over the flat fuel sample are described first, together with the boundary conditions through which information is transferred.…”

“…In order to combine the benefits of analytical and CFD modelling, and to improve on the quality of the prediction using data assimilation, a data-driven hybrid model has already been successfully derived for the droplet combustion system [22]. A steady-state analytical gas phase model adequately describes the heat and mass transfer at the fuel surface, providing self-similar fields of velocity, temperature, and species in the vicinity of the droplet.…”