AI tool aims to automate the design of finned heat sinks

A new project at the Frankfurt University of Applied Sciences is researching an AI-based software tool for the more efficient design of finned heat sinks.

Finned heat sinks improve heat dissipation from a component to the surrounding air through their enlarged surface area. As the performance of processors and power electronics continues to increase, their thermal design is becoming more complex.

In the research project “ThermoPro”, the research group for sustainable thermal management at the Frankfurt University of Applied Sciences, together with Seifert electronic, is developing an AI-based software tool for heat sink design.

The planned system will calculate the thermal resistance and determine suitable geometric parameters for the heat sink once defined input parameters are provided. These include the required cooling performance, the available installation space, and the manufacturing technology used.

According to project lead Boris Schilder, the focus is not on the cooling concept itself but on determining the optimal geometry for heat sink production.

More efficient, smaller heat sinks

The starting point of the project is new manufacturing processes for finned heat sinks used by the project partner Seifert electronic. These processes offer greater design freedom than the extrusion process commonly used in industry.

This allows cooling fins to be arranged more densely and made thinner, enabling more compact and efficient heat sinks. At the same time, however, the number of possible geometry combinations increases significantly. In addition to the mechanical design, factors such as the heat to be dissipated and the expected ambient temperature also play an important role.

AI model as the basis for the design software

The project team is developing an AI model that will serve as the foundation for a future design software tool. The goal is a system capable of automatically calculating suitable heat sink geometries for a specific component.

Schilder explains: “We want to develop an AI model that will serve as the basis for a design software tool. This software will calculate the optimal heat sink for a given component based on input parameters such as the required cooling performance, the available installation space, and the manufacturing technology used.”

The ThermoPro project runs for two years and is funded under the Central Innovation Programme for SMEs by the Federal Ministry for Economic Affairs and Energy. Total funding amounts to €550,000, of which €280,000 goes to the Frankfurt University of Applied Sciences.

ThermoPro originated within the ZIM innovation network FREEM, which focuses on manufacturing technologies for resource- and energy-efficient electrical machines. Previously, the project partners had already collaborated in the ZIM project EvapoCool, which investigated evaporative cooling for high-performance electronics.