Quasi-autonomous thermal model reduction for steady-state problems in space systems

Abstract

A matrix method is developed to reduce the number of elements of spacecraft thermal mathematical models based on the lumped parameter method. The aim of this method is to achieve a satisfactory thermal model reduction for steady-state problems, in an automatic way, while preserving the physical meaning of the system and the main characteristics of the model. The simplicity of the method, and the computational cost, are also taken into account. The reduction process is based on the manipulation of the conductive coupling matrix, which is treated as a sparse graph adjacency matrix. Then, a depth-first search algorithm is used to find the strongly connected components, which define the condensed nodes. Finally, all the thermal entities are reduced, and the results from the condensed model are compared to those from the detailed one. The entire reduction process is tested on a real thermal model, showing a good performance. In the conclusions section the characteristics and limitations of this method are shown.