Jonathan MacArt – Leading a Numerical Experiment

Jonathan MacArt – Leading a Numerical Experiment

 

Before college, Jonathan MacArt had no idea what an engineer did. He did however, progress from building Legos as a kid to disassembling lawn mowers and other devices with motors of different kinds and reassembling them into new things to ultimately building a miniature biodiesel refinery in the backyard with this father.

Seeing a permeating interest in understanding how things work, MacArt’s parents encouraged him to pursue engineering, which he did at the University of Notre Dame. Two undergraduate experiences specifically set MacArt up to continue his studies by pursuing a Ph.D.: he found a natural calling in the engineering approach to problem solving and – partially due to a scheduling conflict – he landed in a sophomore Solid Mechanics class that he not only found challenging, interesting, and rewarding, but in which he found a mentor – his professor, Karel Matous – who from that point forward had MacArt’s future in mind.  Matous’ class “pushed me to think differently than before  –  it required me to truly think through the physics behind what was going on in a given scenario,” said MacArt. “It wasn’t just solving problems using a prescribed method, it was thinking about the physics behind the scenario in question – thinking up from fundamentals to an optimal solution.”

MacArt spent two years doing research in Matous’ lab, where he developed an overarching interest in using computational simulations to better understand physical phenomena. When senior year rolled around, Matous handed him a list of graduate programs to which he should apply, with Princeton on the list.

Now in his fourth year at Princeton, MacArt works with his advisor Professor Michael Mueller in his Computational Turbulent Reacting Flow Laboratory (CTRFL), where he investigates the effects of heat release on turbulence using Direct Numerical Simulation (DNS) of turbulent combustion.

Essentially MacArt has taken his love of things with engines and now applies computational algorithms and fundamental physical principles to studying combustion in those engines, mainly gas turbines and jet engines. “With aircraft engines in particular, combustion is the way we will power those devices for a long time because of the energy density of the fuel,” said MacArt. “It’s our job as engineers and scientists to make these devices as energy-efficient and clean-burning as possible.” But combustion is a very hard phenomenon to predict because of the sheer number of variables involved and building combustion systems is still a very empirical process that relies on experts who build prototypes, test them, tweak it (and then repeat cycle), which – with gas turbines or jet engines – is a very expensive proposition. The expense can discourage experimenting with “aggressive” designs and increases the time to integrate new innovations into products.

MacArt therefore works to develop models to enable those engineers in the field who might design a jet engine to more accurately predict how the engine will perform before they build it; these models  account for certain aspects of the physics in order to reduce the cost of such design predictions. “The computer power to solve the combustion problem in an actual device,” said MacArt “is simply not accessible to engineers designing those devices – there’s not a big enough computer in the world to faithfully simulate combustion in a gas turbine engine.”

Therefore, coarse-grained models must be used to simulate real systems.  Mueller says that the key to success of these coarse-grained models is a solid basis in physics.  Unfortunately, the complex physics required to build these models – complex interactions between chemistry and fluid mechanics – cannot always be probed experimentally.  That’s where MacArt comes in.  Rather than constructing a widget in a lab to do a physical experiment, one can do a computer simulation, very large computer simulations, using the basic fundamentals of physics, what Mueller calls a “numerical experiment.”  MacArt performs these numerical experiments for very small idealized systems that isolate specific features of combustion that might be found in a real system.  The idea is to use the insights gleaned from these experiments to then develop the coarse-grained models that can be used to simulate real systems.

The focus of MacArt’s specific research, however, thinks about things backwards, said Mueller. Engineers typically study how fluid mechanics affects combustion processes, but very few have really given thought to how combustion affects the fluid mechanics, specifically, the complex turbulent flow, which is what MacArt does.  Most of the models used to describe turbulence in combustion are the same as those used to describe turbulence when there is no combustion; MacArt is trying to find when this assumption fails and what ingredients are required for better turbulence models.

The evolution of MacArt’s Ph.D. program has taken him from developing algorithms and tools to take advantage of these very large computers, to trying to understand the combustion processes using various models to create computer simulations. Together with Mueller, he published a paper on his early applied mathematics work; they will also present their current work on the effects of combustion on fluid mechanics at the International Conference on Numerical Combustion in Orlando, Florida and at the Combustion Institute’s National Combustion Meeting in Washington, D.C. this spring.

Mueller said that MacArt realized every graduate student’s dream: “he made the code more accurate, more stable (more robust) and twice as fast. Usually it is a tradeoff, you can achieve more accuracy, but it costs more; Jon got all three at once – more robust, more accurate, and more speed.” As a result, MacArt jokingly told Mueller that he should earn his Ph.D. in half the time, but Mueller advised him that it didn’t work that way, he just had to do twice as much work.

MacArt is up to the task of doing more work than required of him. While he had already completed his graduate student teaching requirement of three semesters, he was needed to help with an additional course. MacArt happily obliged. MacArt says it’s a privilege working with the undergraduates – “I have yet to come across a group of students that are as inquisitive and focused on understanding the fundamentals” he said.

The relationship is reciprocal. “Students were so excited [to have MacArt as an Assistant in Instruction (AI)] they made comments such as, ‘yes, we have Jon; we’re going to understand the course’” said Mueller. You can tell from talking to him, he really enjoys it, takes it seriously, and the students really like him as an AI. The department recognizes this as well, as it awarded MacArt the Crocco Teaching Award in 2016, a departmental award given to a graduate student in recognition of exemplary teaching.

“I do enjoy teaching; it definitely takes time, but it is rewarding and enjoyable and also an important part of our role as graduate students in the college education” said MacArt. On his teaching philosophy, MacArt says that a good teacher knows the material and teaches it, but encourages a student to develop a problem-solving mindset going forward. He describes his method of showing a student the trajectory of a solution by allowing the student to determine the precise steps, and reasoning through approaches to issues that could have more than one solution.

MacArt would like to teach someday; it is clear he is well on his way, just not in half the time.

 

-Femke de Ruyter