Gradient­based multidisciplinary optimization of heat transfer on small spacecraft

Abstract

Due to high demand for performance and cost effectiveness, CubeSats are becoming more technically advanced. This implies increased power consumption for higher performance, which in turn increases the internal heat dissipation. As a result, power-intensive components can quickly overheat due to insufficient heat dissipation, thus limiting the maximum allowable electrical power. Due to the small volume and limited resources, the implementation of active thermal control techniques to increase power on smaller spacecrafts is technically complex and not always possible. Heat transfer depends not only on the parameters of the orbit, power consumption, but also on the structure of the spacecraft itself. As a result, the design of the spacecraft can be optimized so that the internal heat generation could be maximised. In the scope of this thesis a gradient-based method to perform spacecraft structural and mission optimization was developed in order to maximize the allowable heat dissipation of subsystems. The method proposed in this thesis can automatically compute analytic partial and total derivatives of steady state heat-transfer equations with respect to design variables by just using the data of geometrical mathematical model and thermal mathematical model. A pre-processor was developed to interface the open-source optimization framework OpenMDAO with the thermal model outputs obtained in ESATAN-TMS – the main software tool for spacecraft thermal control design in European Space Agency. It was found that simulation results abtained with Open-MDAO agree well with those of Esatan-TMS. The worst case absolute and relative errors for a CubeSat model with solar arrays are below 2 ◦C and 1.2%, respectively. The cause of the error is mainly attributed to external reflections, which, for the sake of computational efficiency, are not accounted for in the OpenMDAO model. Finally, SLSQP algorithm was used to solve the multidisciplinary optimization problem of the interplanetary CubeSat mission, consisting of power and thermal disciplines. A 265% improvement of maximum radio transmitter power was achieved by optimizing 859 thermal, power and attitude design variables with 17 temperature and power constraints. The solution was obtained within less than 3 hours on Intel Core i5-5300U processor. The results also showed that a coupled multidisciplinary modeled that accounted for the dependency of solar cell efficiency on temperature achieved superior results than the uncoupled model, with 7% improvements.