Systematically identifying unertainties
Portrait of Athene Young Investigator Dr Henning Bonart
2024/06/26 by Astrid Ludwig
Dr. Henning Bonart is the head of a research group at the Institute for Nano- and Microfluidics in the Mechanical Engineering Department and a newly appointed Athene Young Investigator at TU Darmstadt. The fluid process engineer wants to apply the statistical methods used in the field of uncertainty quantification and physics-based machine learning to reconcile computer simulations with time-consuming laboratory experiments or processes in microfluidics more reliably than before.
Henning Bonart specialises in microfluidics, i.e. the study of sliding droplets on microstructured surfaces and thin films, as well as the flow of fluids in microchannels. There is a huge range of applications for this research. Microfluid biosensors can detect viruses or sliding droplets can serve as electrical bioreactors. “There is a lot to discover in this field and many effects still need to be explained”, says the researcher. Bonart really enjoys it when things get complicated – he is motivated by a challenge. “I enjoy applying advanced mathematics and statistics on fluid dynamics problems and develop and combine various different methods and computer algorithms with one another to find a solution”, he explains.
The 35-year-old researcher goes on to point out that there are a great deal of technical systems in engineering that are difficult to measure experimentally. “It is usually very expensive and time-consuming or even sometimes impossible to observe certain processes in the laboratory or in industrial applications in great detail. Although modern simulations enable us to depict interesting processes and interactions on the computer, it is unfortunately not very clear how reliable these models and predictions actually are.” The Athene Young Investigator is focussing on combining experimental data and detailed simulation models, which in combination can provide the necessary information about the system together with the level of uncertainty. Bonart is convinced that “this information can then be used as the basis for any further procedures – making the whole process quicker and cheaper”.
Modern simulations enable us to depict interesting processes and interactions on the computer.
A systematic combination of computers, experiments and measurements that deliver reliable information are already being used in practice. “However, it is very difficult to apply complex models due to limited computing capacity and methods because it simply takes too long. We usually rely on simpler but less precise models for this reason”, explains the researcher. This is something that the Athene Young Investigator aims to change. He wants to design software sensors using expensive simulations in his AYI research: “This means taking a variable from a real system that can be easily measured and inserting it into a detailed, complex model to extract information that itself is difficult to measure. I can then use these virtual measurements to better understand, design and control the real system”, he continues.
Coordinating computer simulations and complex measurements
He wants to use these methods primarily for very elaborate simulations of droplets colliding with surfaces that can take several days. To this end, he uses the latest machine learning methods, such as physics-based model reduction, and stochastic surrogate models. In this way, Bonart also hopes to draw conclusions about how to optimize the design of experiments. This would make it possible to reconcile computer simulations with time-consuming measurements more reliably than before.
And this is where Thomas Bayes and uncertainty quantification (UQ) enter the picture. Bayes was an English statistician, philosopher and Presbyterian minister who lived in the 18th century. He developed a formula to calculate conditional probabilities, a systematic way to adapt hypotheses or statements to lived experiences or events. According to Bonart, uncertainty quantification provides statistical methods for quantifying and then systematically reducing measurement errors and uncertainties. “For example, we can identify where we need to make repeat measurements so that we can make reliable statements about an observed effect. And we can also say how certain we are about our predictions.”
He has immersed himself in the philosophy of Bayesian statistics for this purpose. The impetus to take this path came from his first research stay in 2019 in the Uncertainty Quantification Group headed by Professor Youssef Marzouk at the Massachusetts Institute of Technology (MIT) in Cambridge. Bonart believes that Bayesian statistics leads to easily interpretable probability statements and are perfect for applications in engineering and teaching.
“It is a signal to me that it is worthwhile to stay at the university and in the academic world.
Henning Bonart completed his bachelor’s degree in chemical engineering and process technology at the Karlsruhe Institute of Technology and his master’s degree and PhD at the Technical University of Berlin in the Process Dynamics and Operations Group. He moved to TU Darmstadt and the Max Planck Institute for Polymer Research in Mainz in 2020. Bonart was than awarded his own Walter Benjamin position at TU Darmstadt in 2021 and spent another eight months of research at MIT in 2022 as a Walter Benjamin fellow. Since 2023, Bonart has been the head of a group at the Institute for Nano- and Microfluidics at TU Darmstadt and is involved in the ETOS future cluster funded by the BMBF. Recently, he has been selected as an Athene Young Investigator. He is delighted with this “recognition of my work and research that also means external feedback for my ideas.”
The father of three children hopes that this award will improve his chances of obtaining a DFG Junior Research Group and becoming a professor. “It is a signal to me that it is worthwhile to stay at the university and in the academic world.”
The Athene Young Investigator Programme
The Athene Young Investigator (AYI) Programme at TU Darmstadt supports exceptional researchers on their career path for a period of five years. The aim is to promote the scientific independence of early career researchers and give them the opportunity to qualify for the post of university professor by leading an independent junior research group. The heads of these junior research groups are given certain professorial rights and their own budget.
In 2024, the TU Darmstadt has awarded another three excellent young researchers as “Athene Young Investigators”. In the coming weeks, we will introduce the three researchers on the TU Darmstadt website.
