This presentation was held at the 2020 NAFEMS UK Conference "Inspiring Innovation through Engineering Simulation". The conference covered topics ranging from traditional FEA and CFD, to new and emerging areas including artificial intelligence, machine learning and EDA.

Resource Abstract

Electrification is a major initiative across the aviation sector. Rolls-Royce is leading a small team of companies, including Electroflight (Gloucester), aspiring to break the record for the world's fastest (300+ mph) full electric aircraft: project ACCEL (Accelerating the Electrification of Flight). To succeed, the ACCEL project must deliver ground-breaking new technology across electrical systems, energy storage, systems integration and controls.

The battery assembly for ACCEL is the largest, most energy dense assembly ever used to an aviation application. A crucial aspect of the project success is managing battery thermal performance, throughout the aircraft operation.

The batteries are assembled from individual, high performance cells, into multiple sub-assemblies and strings, each with active cooling systems. The cooling system comprises a series of heat-exchanger plates, feed by cooling water, through a series of inlet and outlet manifolds. The design of the inlet and outlet manifolds, in both forward and reverse flow, with a limited space envelope, an even flow split and with minimum pressure loss, is essential for the cooling system performance. In addition, the project, as a whole, is running to very tight timescales, so a fast turnaround at all stages is essential.

Electroflight partnered with ANSYS to assist in the optimisation of the cooling system, specifically the cooling manifolds design. Within the overall project constraints, this required a new engineering design approach.

A two-stage interactive and smart optimisation workflow was used for this study:

- Firstly, an interactive GPU accelerated physics solver was used to explore the fluid volume shape, to optimise the manifold design and flow splits. This technique allowed manipulation of geometry, fluid types and physical inputs, with instantaneous feedback on changes in system performance.

- A second stage using traditional CFD, with conventional meshing and discretisation, was used to validate the predicted manifolds performance and an Adjoint optimisation was performed, to tune the design and minimize the pressure drops. The adjoint solver calculates the best-performing shape with respect to a target variable(s) and automatically morphs the fluid volume shape. The results of the adjoint optimization can then be exported and reverse engineered, for manufacture.

This workflow established optimal manifold concepts, that would achieve an even flow split, with the Adjoint optimisation further reducing the pressure drop by up to 46%.

Subsequent adjustments to the manifold arrangement, required due to changes in the feed pipes and adjustments to the space envelope, were easily addressed by revisiting the workflow process.

This project has demonstrated that a combined approach of fast-interactive physics simulation and Adjoint optimisation, can rapidly derive an optimal solution, replacing the need for otherwise lengthy parametric study.