Making components safe with artificial intelligence Heinrich Büssing Prize awarded to Dr Alexander Henkes

Mechanical engineer Dr Alexander Henkes has been awarded the 2023 Heinrich Büssing Prize for his outstanding scientific achievements. The €10,000 prize is awarded annually by the “Foundation for the Promotion of Science at the Carolo-Wilhelmina” of the Braunschweigischer Hochschulbund (BHB).

In his dissertation, prizewinner Alexander Henkes developed methods based on artificial intelligence that greatly accelerate the highly intensive calculations. As a result, complex simulations can be carried out up to 1,800 times faster, which means that planned lightweight components – for an aircraft or a car, for example – can be tested much more quickly and with fewer resources in the future.

New approaches to computationally intensive simulations

Computer simulations are used in the development of lead components to complement complex experiments and save resources. “However, we encounter a problem with new materials,” explains Dr Alexander Henkes. For example, if researchers change the composition of particle or fibre-reinforced composites at the microscopic level, the behaviour of the entire component changes – and this also depends on the planned shape and material fluctuations. It is difficult to test such components using conventional computer simulations. Dr Henkes: “The calculations can be very computationally intensive – weeks to months. In order to be able to carry out these high-precision calculations, new approaches are needed, especially for smaller companies that cannot afford expensive computer centres. This is where my work comes in.

The methods he has developed to speed up complex simulations circumvent a typical problem of artificial intelligence (AI): to avoid the data hunger of previous AI-based methods, he uses AI-based methods that greatly reduce the time-consuming procurement of data, for example using CT scanners. He has also integrated physical laws into the model. In the future, this will make it possible to test planned lightweight components – for example, for an aircraft or a car – much faster and with fewer resources.

Alexander Henkes completed his doctorate at the Institute for Computational Modelling in Civil Engineering, where he also worked as a research assistant. The title of his thesis is “Artificial Neural Networks in Continuum Micromechanics”. It was proposed by Prof. Henning Wessels, also from the Institute of Computational Modelling in Civil Engineering.

Energetic lightweights

Alexander Henkes is currently working at ETH Zurich as part of the prestigious ETH Postdoc Fellowship: his current research is related to the most challenging but urgent points of his dissertation. AI-based methods often require a lot of energy. Training large models such as ChatGPT can cost the equivalent of several million dollars in electricity. In his dissertation, he has already started to develop “energetic lightweights”. “But there’s still more to do,” says Alexander Henkes. With his future research, he wants to make an even greater contribution to establishing artificial intelligence as a resource-saving tool in the development of high-performance components – such as an aircraft wall.