Background
Heat treatment is an important post-treatment process for cast metal components with the purpose of increasing their mechanical properties. After casting, the components are placed in a heat treatment furnace, which is heated using either electrical heaters or gas burners. The components are held at elevated temperatures for a prescribed amount of time after which they are removed from the furnace and then cooled by air or by being submerged in water. This process – holding at high temperatures and then cooling the components in a controlled way – facilitates the necessary metallurgical transformations that give the components their final properties.

Since heat treatment is conducted at high temperatures, typically in the 900-1100 ℃ range, it is very energy intensive. For this reason, foundries try to optimize this process in terms of temperature and time. While optimization can be done to some extent by using experimental trial-and-error, a complementary but less developed route is to set up and model a digital twin of the heat treatment process, which allows for much greater flexibility and quicker screening of new concepts.

Description
The goal of this thesis project is to set up a digital twin of a heat treatment process for cast components using OpenFOAM, an open source CFD and FEM environment. Tasks include:

• Building and meshing the geometries of heat treatment furnaces and simplified cast metal components.
• Setting up a functional, physics-based heat treatment model in OpenFOAM, reflecting real process conditions such as conduction, convection, radiation, and temperature-dependent materials properties.
• Analyzing parameter sensitivity for the model:
  • What level of mesh detail is necessary for convergence and getting a sufficiently small simulation error (compared with existing experimental data)?
  • What model parameters will have the greatest influence on simulation results?
• Use physics-informed neural networks such as neural operators to speed up the simulations in order to enable iterative improvements of the model in an optimization loop.
• Optional: additional experimental verification.
• Dissemination of results.

Key Responsibilities

• Learning how to work in OpenFOAM and generating data.
• Literature review on heat treatment simulations and attempts at leveraging AI to speed up the simulations.
• Training neural networks on generated data and analyzing the results.
• Writing and defending the thesis.

Qualifications

• At least high school-level knowledge of physics.
• Basic knowledge of numerical simulations.
• Basic skills in Python programming.
• Basic knowledge of deep neural networks.
• A passion for science!

Terms

Start and end date: 2026-01-12 – 2026-06-05

Degree level: The project is carried out at Master’s level.

Number of students: 1 or 2.

Place: Partly remote and partly in-office (flexible). RISE has offices in several cities, for example Västerås, Jönköping, Mölndal and Lund. However, practical parts of the work must be conducted at RISEs laboratory in Jönköping.

Compensation: 1330 SEK/hp/student if one student, 1000 SEK/hp/student if two.

Welcome with your application! 
Last day of application: December 31, 2025.

Supervisors:
Andreas Thore, Ph.D., RISE, andreas.thore@ri.se
Johan Wendel, Ph.D., RISE, johan.wendel@ri.se

Examiner: RISE