Optimal design of aircraft thermal systems and their heat exchangers leveraging a data-driven surrogate model
F. Beltrame (TU Delft - Flight Performance and Propulsion)
Piero Colonna di Paliano (TU Delft - Flight Performance and Propulsion)
C. M. De Servi (Vlaamse Instelling voor Technologisch Onderzoek, TU Delft - Flight Performance and Propulsion)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
Thermal energy recovery is being investigated by leading aerospace companies as a means to improve the efficiency of next-generation propulsion systems. The organic Rankine cycle (ORC) system, due to the flexibility of the concept, is arguably the best technology for waste heat recovery and, thus, a promising solution to develop recuperated engines. In such systems, heat exchangers are arguably the most critical components, as their design must balance thermal performance with constraints on weight and volume. Consequently, integrating the optimization of heat exchangers into the overall system design may lead to substantial performance enhancement compared to more traditional iterative design methods. The objective of this study was the development of a systematic methodology for optimizing airborne thermal systems, with a focus on addressing the computational challenges of integrated design. Three design strategies are compared: (i) optimization of the sole cycle parameters while performing heat exchanger sizing for values of the geometrical characteristics defined a priori based on a preliminary investigation of the design space of these components, (ii) concurrent optimization of both the thermodynamic cycle and of the most critical heat exchanger, e.g., the condenser, and, (iii) use of a data-driven surrogate model of the condenser to predict the optimal heat exchanger geometry as a function of any feasible thermodynamic conditions to reduce the number of optimization variables of the integrated design problem. The surrogate model is constructed based on datasets of Pareto-optimal HX designs in the objective space defined by heat exchanger weight and pressure drops. The three design strategies are applied to two case studies featuring supercritical ORC systems utilizing cyclopentane as the working fluid: a combined cycle auxiliary power unit (CC-APU) and a combined cycle turboshaft (CC-TS) engine. Findings indicate that integrated optimization yields performance gains that vary depending on the heat exchanger topology, application, and thermodynamic cycle. For instance, CC-APU designs obtained with the integrated design optimization method are up to 15% lighter than designs obtained with the optimization of the thermodynamic cycle parameters alone, for the same net power output. Microchannel condenser designs with offset strip fins allow for obtaining a better performance than louvered fin-based designs if a low-pressure drop is targeted, whereas louvered fins are advantageous if a higher pressure drop is allowed. The design strategy employing the surrogate model considerably reduces the computational cost, without significantly affecting accuracy: the relative deviation between the Pareto front obtained with the surrogate model and that obtained with the integrated optimization strategy ranges between 1% and 2.9%. These values are comparable to the uncertainty of the predictions of the heat exchanger model. The reduction in computational time required to generate the Pareto fronts associated with the two case studies is up to 200%.