Matthew Cornfield

An Exploration of Flow Control Using Machine Learning and Computational Fluid Dynamics

By Matthew Cornfield

Supervised by Prof. Karen Bradshaw

Although numerous studies relating to computational fluid dynamics and machine learning

have been conducted in relation to automotive development, the majority focus on

early development using completed 3D models, the final testing stages of development, or

machine learning accelerated computational fluid dynamic simulations.

While this approach is helpful in software development and simulation, it is not easily

adaptable to automotive design where the final model is constantly changing and being

modified. Consequently, the aim of the present work is to propose a method for con-

ducting computational fluid dynamics and machine learning concurrently to continue to

accelerate the development process. As a result, the proposed method has been used to

design and improve the aerodynamic efficiency of an object in motion, focusing on 

developing, implementing, and comparing machine learning models capable of generating

three-dimensional objects with the required geometry to direct airflow paths required in

applications such as pressure generation required for both active and passive flow control.

It was determined that both decision tree regression and LSTM auto-encoder models

could be used to optimise the aerodynamic efficiency of solid bodies, with the LSTM

autoencoder performing best overall. An effect of the shape optimisation was the overall

reduction in shape size as optimization increased.

Github - https://tinyurl.com/CFD-FlowControl

Presentation (1 and 2) links:

• Presentation 1

• Presentation 2

and Computational Fluid Dynamics