Adjoint-based optimization of thermal components using a CAD-based parametrization

Abstract

Decarbonizing aviation requires the development of novel propulsion systems that would be powered by renewable energy stored in batteries, green hydrogen, and e-SAF (a type of sustainable aviation fuel). To increase the viability of these carbon-neutral solutions, minimizing mission energy consumption will remain the key driver of the design of next-generation aircraft systems and their components. Additionally, increasing importance is placed on thermal energy recovery and thermal management, which necessitates the design of high-performance thermal components, namely heat exchangers and heat sinks.

This dissertation documents research on shape optimization using the discrete adjoint method and CAD-based parametrization for the design of aerospace-grade heat exchangers. The main outcome of this work is the development of the optimization framework to concurrently optimize multiple heat transfer surfaces parametrized using a CAD method based on Non-Uniform Rational Basis Splines (NURBS) and the discrete adjoint method available in the open-source computational fluid dynamics (CFD) software SU2. The application of the design method is demonstrated for two-dimensional and three-dimensional heat transfer surfaces in configurations representative of aircraft condensers and evaporators, as well as heat sinks for thermal management. In this regard, two formulations of surface sensitivity are proposed such that the resulting optimal solutions can feature identical shapes using averaged sensitivities or non-identical shapes when optimized concurrently, albeit independently. Additionally, the feasibility of integrating the CFD-based method in system-level design and its potential for enhancing system performance are investigated.

The results obtained using the design method show that the application of this framework can achieve geometries of thermal components with reduced pressure drop and enhanced heat transfer coefficient compared to conventional designs. The automated design chain applied to a two-dimensional configuration representing tubular heat exchangers reduced the pressure drop significantly while constraining the heat transfer rate. Using three-dimensional shape optimization of pin-fins with conjugate heat transfer resulted in an unconventional fin shape that led to a simultaneous reduction in total pressure losses and an increase in heat transfer coefficient. These performance improvements of about 20% corresponding to optimal geometries obtained from shape optimization can lead to significant gains in the performance of the system, as demonstrated by its application in the early phase of system-level design reported in this work. Future developments on such a design method have the potential to conceive designs of the next-generation heat exchangers that could be deployed in propulsion systems, enabling carbon-neutral aviation.