Design optimization of thermal paths in spacecraft systems

Abstract

This thesis introduces a thermal design approach to increase thermal control system performance and decrease reliance on system resources, e.g., mass. Thermal design optimization has lagged other subsystems because the thermal subsystem is not thought to significantly drive performance or resource consumption. However, there are factors present in many spacecraft systems that invalidate this assumption. Traditional thermal design methods include point designs where experts make key component selection and sizing decisions. Thermal design optimization literature primarily focuses on optimization of the components in isolation from other parts of the thermal control system, restricting the design space considered. The collective thermal design optimization process formulates the thermal path design process as an optimization problem where the design variables are updated for each candidate design. Parametric model(s) within the optimizer predict the performance and properties of candidate designs. The thermal path parameterization captures the component interactions with each other, the system, and the space environment, and is critical to preserving the full design space. The optimal design is a thermal path with higher performance and decreased resource consumption compared to traditional thermal design methods. The REgolith X-ray Imaging Spectrometer (REXIS) payload instrument serves as a case study to demonstrate the collective thermal design optimization process. First, a preliminary thermal control system model of a point design is used to determine the critical thermal path within REXIS: the thermal strap and radiator assembly. The collective thermal design optimization process is implemented on the thermal strap and radiator thermal path. Mass minimization is the objective and the REXIS detector operational temperature is a constraint to the optimization. This approach offers a 37% reduction in mass of the thermal strap and radiator assembly over a component-level optimization method.