Supervisory authorities

CNRS UPS Enseeiht

Search




Home > News

Non-linear Thermics and Monte Carlo

by Laurence Laffont - published on , updated on

Jean-Marc TREGAN’s thesis defense intitled "Thermique non-linéaire et Monte-Carlo" (Non-linear Thermics and Monte Carlo) will be taking place on Friday, December 4th 2020 at 2 pm at the Amphithéâtre Baillaud of the Paul SABATIER University
.

Zoom:
https://univ-tlse3-fr.zoom.us/j/88051822939?pwd=dXBVMnk3UFNGRzdiK2hBKzg2VitwQT09
Meeting ID: 880 5182 2939
Passcode: 782157
Please mind switching off both microphone and camera.

Abstract:
The work presented adresses the numerical simulation of coupled-heat transfer problems in the presence of four standard non-linear sources: temperature at power 4 in radiation and conductivity, heat capacity and convective exchange coefficient, all of the three functions of temperature.
Our specificity is the use of a Monte Carlo method that preserves a set of strong points in the algorithms used for linearization, most notably for the ability to compute probes in complex geometry.
We begin with a synthesis of the statistical reformulation of the justifying models, within the linear framework, with a randomized reading of the conducto-convecto-radiative coupling which will be the starting point of our proposal. We then group our 4 non-linear questions in the same formal framework, built on transportation physics, in order to exploit the results of a recent revisiting of zero-collision algorithms. The resulting branch algorithms face computational difficulties: the number of branches increases very strongly at low Knudsen numbers. We then propose a workaround strategy that ensures a limitation of the number of branches via a hierarchical rewriting inspired by Picard’s method.