Generative Thermal Design Through Boundary Representation and Multi-Agent Cooperative Environment

TL;DR

This paper introduces a novel generative thermal design method utilizing multi-agent deep reinforcement learning and boundary representation, enabling efficient exploration of complex heat transfer geometries without relying on shape derivatives.

Contribution

It presents a new multi-agent reinforcement learning framework with a neural surrogate model for thermal design, handling complex boundary interactions and multi-objective optimization.

Findings

The framework effectively predicts heat transfer and pressure drop.

It learns design strategies without shape derivatives.

Demonstrates multi-objective optimization capabilities.

Abstract

Generative design has been growing across the design community as a viable method for design space exploration. Thermal design is more complex than mechanical or aerodynamic design because of the additional convection-diffusion equation and its pertinent boundary interaction. We present a generative thermal design using cooperative multi-agent deep reinforcement learning and continuous geometric representation of the fluid and solid domain. The proposed framework consists of a pre-trained neural network surrogate model as an environment to predict heat transfer and pressure drop of the generated geometries. The design space is parameterized by composite Bezier curve to solve multiple fin shape optimization. We show that our multi-agent framework can learn the policy for design strategy using multi-objective reward without the need for shape derivation or differentiable objective function.