30 hp Master Thesis: PhysicsAI versus Design of Experiments

30 hp Master Thesis: PhysicsAI versus Design of Experiments

Introduction

Scania’s purpose is to drive the shift towards a sustainable transport system, creating a world of mobility better for business, society, and the environment. Exploring and accelerating the development of tomorrow's transport system is one of the driving forces. Imagine you could impact the next generation Scania vehicles by working in a dynamic and diverse workplace. Yes, this opportunity gives you significant exposure to the vehicle development process.

You will have the opportunity to develop your engineering skills, to learn if the Scania work culture suits your future career goals and to show your abilities and talents over 20 weeks of thesis work.

Background

Simulation and testing are core parts of the development process for next generation of vehicles and have been so for many years. To accurately predict or verify results is key. Over the years simulation has played a larger role in the development process and in providing faster design iterations.

A traditional way of exploring the design space is to use Design of Experiments, (DOE). In DOE, selected input parameters are varied to explore their effect on selected system outputs. In contrast to trial-and-error, this allows for systematic studies of different design parameters. For complex models with many parameters, this is usually a time-consuming process.

The emergence of AI and machine learning could possibly offer an interesting alternative to DOE. Training an AI model on design parameters could allow for the optimization of selected system outputs. However, due to the maturity state of the technology, several questions regarding how this compares to the more traditional method remains to be fully understood.

Objective and job description

Main focus of the work will be to train and evaluate how the AI model performs with respect to design optimization as compared to a traditional DOE approach/adjoint based optimization. This will be done by optimizing a fan installation with geometric deep learning in PhysicsAI and then evaluate the methodology and the performance of the fan installation to that obtained by DOE through Computational Fluid Dynamics (CFD) simulations.

The thesis project will require work in both Windows and Linux environments, work with python code and running CFD simulations through STAR-CCM+ on high-performance computing clusters.

Education

Final year pursuing a master's degree in Mechanical/Aerospace Engineering or equivalent and with a specialization in Fluid Dynamics.

Experience from working on computer cluster environments with Linux interface, unsteady state simulations, and exposure to scripting simulation workflow are desirable. Knowledge in STAR-CCM+ and Python is beneficial as well as having a genuine interest in CFD.

Number of students

1 - 2

Start date

January (exact dates can be discussed later) and estimated time required is 20 weeks.

Contact persons and supervisors

Supervisor, Eric Axtelius (TGRMST1), [email protected]

Manager, Joakim Lindholm (TGRMST1), [email protected]

Application

Your application must include a CV, personal letter, and transcript of grades.

A background check might be conducted for this position. We conduct interviews continuously and may close the recruitment earlier than the date specified.

Publication date from - to.

2026-10-07–2026-11-30