Segregation of thermal radiation problems with COMSOL Multiphysics®

Keywords

 

Heat transfer, surface-to-surface thermal radiation, problems segregation, optimization, memory gain, hemicube, ray tracing


 

Context / Goal

 

High-temperature industrial applications most often require consideration of heat transfer by thermal radiation. A complex geometry with multiple radiating surfaces can quickly require large amounts of RAM memory and can greatly penalize the computation time. The main reason for this performance degradation is a greater filling of computational matrices caused by the non-local coupling intrinsic to the radiation.

Here we are interested in segregating the thermal radiation problem into a set of smaller subproblems with the COMSOL Multiphysics® software. This approach is based on dividing the radiating entities into groups, in order to optimize the coupling between these surfaces, thus making calculations possible on current computers.

SIMTEC can thus help its customers to develop models adapted to their computational resources.

 

SIMTEC's Achievements / Results

For confidentiality reasons, only a simplified model is illustrated here.

This model is composed of a row of 3 radiating tubes placed in front of a diffuse mirror (zero hemispherical emissivity). A heat flux (in W∕m2) is imposed on the front faces of the tubes.

 

SIMTEC actions consisted in

  • Developing the segregation theory, by making radiation groups.
  • Segregating the irradiance G, composed of the mutual part Gm (coming from other surfaces) and the ambient one Gamb (coming from the ambient environment).
  • Segregating the radiosity  composed of the reflected part and the emitted one.
  • Add and update the cross-module contributions Gm in the course of segregated iterations.

The following figures present a comparison between the fully coupled approach and the segregated one. The results are identical.

 

 

 

 

The following figures present a comparison between the fully coupled approach and the segregated one. The results are identical.

 

 

 

 

This method makes it possible to solve problems with a large number of radiating surfaces using a reasonable amount of computational resources.

 

Pour un grand modèle montré dans la dernière image :

  • Computation is about 10 times faster with a segregated model as compared to a coupled one.
  • About 10 times less RAM is used with a segregated model compared to a coupled one.
  • This type of segregation can be achieved using radiation groups integrated in the COMSOL Multiphysics® software, validated by our approach.