Enlarge picture

Mount Rainier

Close menu

(26.04.2024) Through its research fellowship program and Walter Benjamin Fellowship, the DFG supports researchers in early career phases by funding an independent research project abroad and, since 2019, in Germany too. Many of these fellowships are taken up in the USA and to a lesser extent in Canada. In a series of talks, we aim to give you an impression of the range of researchers in receipt of DFG funding.

DFG: Dr. Lagemann, thank you so much for taking the time to talk to the DFG Office North America.

Esther Lagemann (EL): Thank you for the opportunity to take part in this interview and of course for the fellowship, which gives me the freedom to pursue a largely independent and hopefully productive research stay here at the University of Washington in Seattle.

DFG: You took your upper secondary school examinations in Neuss near Düsseldorf. How did you end up going from there to studying mechanical engineering at RWTH Aachen University?

EL: The two towns are 70 kilometres apart, and since they’re both situated in the Catholic Rhineland, Neuss and Aachen are not very far removed from each other culturally either, but that’s probably not what you were getting at. My father’s a mechanical engineer, too, and he studied at the University of Applied Sciences in Aachen. Since I’m an only child, I got to be involved in a lot of my father’s DIY activities – he did an apprenticeship as a locksmith before going to university. I didn’t necessarily always get to do the critical stuff – I’d normally be holding the vacuum cleaner rather than the drill, for example, but he certainly gave me the opportunity to develop an interest in technical things. My mother is a specialist in remedial education and a nursery school teacher, so she’s more responsible for the musical and artistic side of our family. If you take a look at the subjects I took in the last two years of upper secondary school, you’ll see that they roughly reflect my parents’ interests – math for my father and art for my mother. So with their different inclinations, my parents definitely left their mark on me. But I didn’t just develop my interests to please my parents, of course. They supported me in trying out everything I wanted to do and I was free to find out what I really enjoyed. My school was a contributing factor here, too – it was a school for girls with a focus on STEM subjects. We had two advanced groups in maths and physics in my year, and the support we had was very effective and helpful. I can’t say for sure how I would have fared at a different school, but it certainly suited my rather reserved nature during my time at school that all-girls classes are probably quieter than mixed classes – that’s the best learning atmosphere as far as I’m concerned. 

DFG: Is mechanical engineering such an obvious choice if your specialized subjects in the last two years of upper secondary school were math and art?

EL: Perhaps not that obvious, but I didn’t just enroll at RWTH for mechanical engineering. I also signed up for architecture, and it wasn’t until I actually started my degree program that I opted for mechanical engineering. 

DFG: Don’t the first semesters of a subject area like mechanical engineering seem more daunting than in architecture because of the kind of “screening” process that takes place with courses in subjects like physics and mechanics? 

Enlarge picture

Whidbey Island

Close menu

EL: Even though I tend to be a bit reserved, that doesn’t mean I’m instantly frightened off. And I wasn’t entirely on my own at the start of my studies in Aachen: there were several other girls from my school with me there as well. Of course, the first few weeks and months of a degree program are very different to upper secondary school, but we all rose to the new challenges quite well. Then there was the fact of living in an unfamiliar city, since I was keen to stand on my own two feet after passing my school-leaving exams, although I do have to admit that it took until my second semester for me to get my own washing machine in Aachen. But when it came to installing it, it was my father who helped me rather than the other way round.

DFG: What were your personal success factors for a successful degree program?

EL: To begin with, the fact that my father was already an engineer. That helped me because of his basic understanding of the course content and methodology, and of course he was very interested in what I was up to. But my specialization during my studies was different from his, so his collection of specialist books and expertise was actually only of limited use as far as I was concerned.

DFG: In terms of your specialization, the term “wall shear stress” stands out in particular – something that non-experts may erroneously imagine to be linked to the field of structural calculation or even architecture. Can you explain it for us?

EL: Wall shear stress is a key parameter in fluid mechanics: it doesn’t describe the tension within an object, but the friction or resistance that occurs when liquids and gases flow past objects. This resistance is greatest directly at the boundary between the surface and the flow due to friction, because the flow is slowed down to the speed of the wall, which in most cases is immobile. We refer to this area of “deceleration” as the boundary layer. Although shear stress decreases as the distance from the wall increases, it is subject to constant interaction with the turbulence resulting from the flow. And that’s precisely what makes measuring and predicting wall shear stress very tricky. Now you could say that turbulent flows are just chaotic like a lot of things in life, but neither boundary layer theorists nor engineers in the many fields that deal with aerodynamics and fluid dynamics would be satisfied with that. They’re interested in optimizing aircraft wings and wind turbines, for instance, or gaining a better understanding of the processes at work in the human vascular system. Wall shear stress has a crucial role to play in all these areas.

DFG: What methods do you use to come to grips with this apparent chaos?

EL: One very useful and now widely-used method in experimental fluid mechanics is particle image velocimetry – or PIV for short. It’s an optical method for determining velocity fields in flowing gases and liquids. You put tiny particles into the flow, illuminate them with a laser, then photograph the flow field at short intervals: in this way you can track the movement of the particles to determine approximately which point of the flow was how fast at what time. If you achieve a high spatial resolution in the boundary layer, you can even calculate the wall shear stress from the velocity distribution near the wall. 

DFG: Don’t these added factors alter the flow behavior and distort the result?

EL: Yes, that’s precisely why it’s very important to match the properties of these particles to the flow field you’re measuring. We use tracer particles of about one micrometer in size such as oil droplets or other “ideal” substances – ideal in the sense that they have no impact on the flow behavior, or at least not an impact we can’t eliminate in our calculations and therefore measure the flow. Since the PIV method is so popular in fluid mechanics, there are actually commercial suppliers for this kind of substance.

DFG: But measuring velocity fields in flows is not what you’re ultimately trying to achieve, right?

Enlarge picture

Space Needle

Close menu

EL: No, measuring velocity fields using PIV is more of a standard procedure, at least in many areas of application where we have optical access to the relevant flow. A much more sophisticated issue is to understand these velocity fields and, as I briefly mentioned earlier, to look at how the dynamics of the turbulent flow structures affect the wall shear stress. It would be very nice – and now we come to the real Holy Grail in my field of research – if we could express this understanding as a mathematical model that is as general as possible. In other words, a model that is capable of describing the interactions that occur in a blood vessel and on the wing of an aircraft. But of course that’s not something that’s really tangible for me yet – it’s still somewhere at the back of my mind.

DFG: You’re looking to use machine learning techniques to search for this Holy Grail. How does that work exactly?

EL: My Walter Benjamin proposal is actually entitled “Prediction and understanding of the wall-shear stress modulation by non-linear interactions based on novel machine learning techniques”: you might translate that for the non-expert to read “Understanding turbulence chaos using an artificial brain.” In a number of ways, Steve Brunton’s group offers an excellent basis for success on a project like this. There’s an AI Institute in Dynamic Systems here with experts in artificial intelligence and machine learning, as well as experts in dynamic systems such as turbulent flows. The aim is to use state-of-the-art methods like neural networks to gain a better understanding of dynamic systems and to be able to model them and control them. In my case we’re currently using a so-called “autoencoder” – a neural network made up of more than a million neurons that is effectively a machine “brain.” We feed this brain with information, which in our case is PIV data. The idea is for the neural network to learn the dynamics of wall shear stress based on the data. Ideally, we also want it to tell us how it learned the relationship between the velocity fields and the wall shear stress. After all, just staring at a black box isn’t going to be of much use to us. But for this to work, we have to find the right architecture for the network, such as how the different neurons have to be connected to each other. So architecture has actually come to play a role in my life after all, though a slightly different one from what I might have expected.

DFG: How can we imagine a neural network being trained? It doesn’t respond to training stimuli in the same way as a human being, does it?

EL: The underlying idea is actually the same. Our aim is to feed our neural network with information so that it engages in a learning process – just as the human brain accumulates experience and forms new synapses based on visual impressions, for instance. But in our case, we want to train it to become a specialized expert; in other words, we only train our network with PIV data. For this purpose we use synthetically generated PIV images as well, i.e., produced by simulation software rather than an experiment. Firstly this gives us more ways of generating different flow cases and varying the properties of the particles, the lighting conditions, etc., which is much more complex to do in an experiment. Secondly – and this is the more crucial point – we know exactly the underlying flow properties of this synthetic data and how they interact with the wall shear stress. In the case of experimental PIV data we generally don’t have this information (that’s the problem, of course) so we’re unable to judge whether our neural network is learning a physically correct correlation in the first place. Using this synthetic data, we can tell the network what the correct result is, and in this way – with more and more training sessions – it will hopefully learn to make increasingly accurate predictions. Similar to the way in which you might repeatedly give a child an apple and tell them it’s an apple, until finally they know what an apple is. Training a neural network takes about the same amount of patience and dedication, since it can also be quite stubborn and unwilling to learn.

DFG: What are your career plans once your current project is completed?

Enlarge picture

Research Group

Close menu

EL: First and foremost, I hope to achieve very good results so that I can successfully apply for a junior research group leader position in Germany, or perhaps even a tenure track position here in the US, such as an assistant professor post. Since I really enjoy both teaching and research, I’d very much like to stay in academia as a researcher. When I did an internship at a kindergarten during my school days, I quickly reached the limits in terms of age-appropriate didactics and pedagogy, but things went much better during the four years I spent teaching and supervising students at RWTH Aachen.

DFG: You’ve been in Seattle, Washington, for almost six months now. Have you had a chance to explore the surrounding area at all?

EL: In the winter months it’s really too rainy to explore the area extensively, but apart from the typical sightseeing spots like the Space Needle and Pike Place Market, my husband and I have been to Whidbey Island, Discovery Park, and the small neighboring beaches. For the summer we’re already planning more trips to the nearby mountains and national parks. But we live on one of the many inland waterways in Seattle anyway, and being so close to the Pacific Ocean means we get to hear seagulls and go for walks by the water every day. In addition to being close to the sea, I particularly love being able to see the mountain ranges and vast forests in the distance – in good weather they’re always visible in the background. So by Aachen standards, we’re already living in exotic climes. And during the first few months you’re really busy settling in – setting up your new home, finding your feet in a foreign country and getting your research project established in the working group.

DFG: How would you describe your current working group as compared to those you’ve experienced in Germany?

EL: Steve Brunton’s group is at one of the best universities in the US, the University of Washington in Seattle. Seattle’s tech landscape is huge due to all the innovation-driven companies like Microsoft, Google, Amazon, and Boeing, and it’s sometimes even referred to as the next Silicon Valley. Christian and I are very happy to be able to do our postdocs with Steve and work on our own projects thanks to the DFG fellowship. Steve’s group is quite large by American standards, consisting of about 12 postdocs and ten doctoral researchers, with probably about half from the US. The others are from India, Korea, Canada and Eastern Europe. At RWTH, only four out of the 40 doctoral researchers at my institute had not been educated in Germany; they came from China, India and Spain. This aspect of internationalism is something where there’s still a lot of room for improvement in Germany. The other aspect that struck me very clearly here is the different levels in the degree programs on each side of the Atlantic. In Germany, it’s normal for people to do a master’s degree between doing their bachelor’s degree and their doctorate: that involves learning a lot of content as well as taking key steps towards independence. In the US, doctoral students start their doctoral training straight after four years of a bachelor’s degree programme, so they’re obviously less independent and less well trained. 

DFG: Are you happy with your decision to pursue a career in engineering?

Enlarge picture

Portage Bay

Close menu

EL: Absolutely! I couldn’t imagine what I would have become if I’d studied architecture in Aachen rather than mechanical engineering. Although it does have to be said that my research in fluid mechanics doesn’t fit in with the classic career choice in engineering that a lot of people might go for. Nor do I consider it a bad decision to have turned my other more artistic interests such as playing the flute, volleyball, jazz dance and modern dance into hobbies rather than my profession.

DFG: To conclude: do you have any particular advice for women who might be considering studying mechanical engineering? 

EL: Yes, I do: a degree program in this field is a highly rewarding mixture of physical-mathematical fundamentals and extremely diverse applications. It’s still very much a male domain, however, so there isn’t much feedback in the teaching and learning process – something that women perhaps need a little more on average than men. As a woman, you find yourself amongst a lot of men at lectures; at my university in Aachen the share of women in the bachelor’s degree program was around 12%. What is more, the vast majority of men like to give the impression that they’re already familiar with the subject matter before it’s even taught, which can be rather intimidating. But that certainly shouldn’t stop you from finding your own path and studying whatever suits your interests. With over 1,000 students per year – as was the case in Aachen, at least – there’s plenty of choice to find like-minded friends and study partners. One other piece of advice I’d pass on: as a woman, never arrive late for a lecture on a general topic. Not for fear of missing out on anything, quite the opposite: you enter the lecture theatre and then have the undivided attention of several hundred male students, all of whom will seek to outdo each other with niceties and comments. But apart from that, I have to say that I never had the feeling of being discriminated against or directly disadvantaged as a woman. 

DFG: Thank you very much for that advice and for the entertaining and informative interview, and let’s hope for an increased share of women in engineering sciences. We wish you and your husband all the very best for your professional and personal future, and may it hopefully lead you back to Germany one day.