Generative Deep Learning on Images of Thermo-Mechanical Simulation Results

Abstract

This paper presents a novel methodology that combines engineering simulation and machine learning for the thermo-mechanical design of a secondary air system double-sided seal in an aero engine turbine subassembly. Secondary air system seals are crucial in aero engine design as they have a direct impact on specific fuel consumption. The study uses an automated analysis workflow to generate a large dataset of images embedding key design and performance attributes for the seals, such as running clearances at key operating conditions. These images are used to train a conditional Generative Adversarial Network (cGAN), which can then be used for design exploration or optimization. The paper introduces a unique approach to encoding and decoding these images, enabling automatic quality monitoring of the generated images and training processes, as well as extraction of targeted results. The predictability of the Deep Learning models is assessed, demonstrating how this methodology can generate designs in targeted categories and can support decision making both in the preliminary design phase, to enable classification of “good” and “bad” designs, and in the detailed design phase, to support optimization and robust design. Beyond design, these methods can also be used to support the implementation of the Digital Twin.