Thermal Model Automatic Reduction: Algorithm and Validation Techniques 2006-01-2112
This paper reports on an automated technique for reducing lumped-parameter thermal models. The method is applied in case of a simple geometry (a flat radiator panel), first in steady state conditions and then a comparison in time-variable environment is carried on, assessing how the reduction procedure affects the effective thermal mass of the radiator.
More complicated shapes are studied, including concave corners and holes. The technique is implemented by a Matlab® code and it is applied to reduce the heat-pipes- embedded-radiator model of AMS-02, an ISS external payload.
The detailed and reduced models behaviour is simulated by SINDA® and the results are compared: existing specification for test-model correlation, regression line and histograms are used for the verification. Furthermore radiator functioning extreme situations are simulated to see the method validity limits in steady state, introducing the concept of ‘working volume’.
A new method for evaluating the performance of the reduction technique is presented as well, introducing a metric to quantify synthetically the distance between the complete and the reduced thermal model. The method is shown to be suitable for being extended to non-radiator geometries.