Convenient access to efficient thermal design

HPC-Competence Center

The Flemish Supercomputer Center (VSC) is a partnership between the five Flemish universities and their university associations.The VSC encourages the use of scientific and technical computing in the Flemish academic and industrial landscape. To this end it offers infrastructure, training and services. In addition, VSC acts as a lever to promote the importance of scientific and technical computing and its added value to society.

The Enterprise

Diabatix is a Flemish start-up that focuses on the design, simulation and optimization of refrigeration components. This encompasses cooling components for lasers, (power) electronics, electric car batteries, internal combustion engines, etc. Manufacturers work together with Diabatix to make their machines, appliances and installations more efficient, compact and environmentally friendly.

How HPC makes the difference

Diabatix offers an innovative, easily-accessible platform solution, which assists every engineer to design critical thermal applications in a faster and better way. The generative design method used removes the typical hurdles of trial-and error- engineering, leading to disruptive cooling performance enhancements up to even 30%.

The platform that Diabatix offers runs on proprietary software. It gives the customer easy access to an in-depth analysis of the complex physics of heat transfer in liquids, solids and gas flows during design. In this way, customers need not be experts in thermal design, but still get access to optimized scientifically-sound design.

The calculations behind the design require a lot of computing power, for which the VSC compute clusters are very useful. With the aid of HPC, Diabatix is able to master more complex thermal applications in a faster and more profound manner. Diabatix founders Lieven Vervecken and Joris Coddé: “During our doctoral research, we have gained experience with the use of HPC, also in the VSC. This allows us to process thermal design studies on hundreds of processors, which typically makes the computation 50 times shorter.”