September 14, 2021

Turbulent air flows – vortices and eddies – can create challenging problems in the fields of aerospace engineering and transportation. These flows can cause drag on vehicles such as airplanes, spacecraft, trains and trucks, reducing their performance and increasing their fuel consumption. In other scenarios, however, turbulence can help keep the flow attached to the vehicle and prevent or delay separation and the associated drag. But even as turbulence can cause a myriad of issues, researchers have recently suggested that harnessing the kinetic energy from the largest kinds of turbulent eddies near a vehicle’s surface – Large Scale Motions (LSMs) – and redirecting them, can actually enhance a vehicle’s performance.

Two faculty members, Efstathios Bakolas and David Goldstein in the Department of Aerospace Engineering and Engineering Mechanics (ASE/EM), aim to better understand these problems and develop solutions to manipulating LSMs with two newly-funded research projects at the intersection of computational fluids and control theory. With a combined total of nearly $1M in funding from the National Science Foundation (NSF), the researchers will lead two interdisciplinary teams to study, simulate and control the behavior of large-scale turbulent flows.

Early discussions between the two professors, whose research specializations are in very different areas – computational fluids (Goldstein) and control theory (Bakolas) – began several years ago in the hallways of W.R. Woolrich Laboratories, the former home of the ASE/EM department. It was here that the two realized how complimentary their research methods just might be. And now, thanks to two new grants, their collaborative ideas will begin to come to life over the next few years.

“These two fields, at first glance, probably don’t seem to be very relevant to each other,” Bakolas said. “But once David and I started talking about them, we realized that some of the techniques I use in the area of controls could help solve some difficult fluids problems. Similarly, a fluids problem might provide me with some insight about the types of applications I should or shouldn’t be trying in my theoretical/analytical work.”

Bakolas is leading the project, “Data-Driven Model Reduction and Real-Time Estimation and Control of Coherent Structures in Turbulent Flows.” The goal of this research is to develop data-driven algorithms for modeling and control of LSMs, with the long-term goal of developing a holistic framework for modeling, estimation and control of turbulent flows. Goldstein is a co-principal investigator on the project.

“Traditional methods for control and estimation of dynamical systems are model-based. One of the current trends in control theory is to replace the model-based paradigm with a data-driven paradigm. By harnessing the contemporary data science revolution, one can discover solutions to complex problems in control theory which cannot be addressed by traditional model-based methods,” Bakolas said. “This research will promote the transitioning to the model-free paradigm in control theory and will demonstrate its applicability to estimation and control problems of a notoriously challenging and complex class of systems, namely turbulent flows.” 

turbulent flow diagram
Targeting and manipulating multiple volumes: a) a Gaussian mixture model (GMM) is fitted to a detected volume (green ellipsoid) upstream of the actuator (black sphere); b) the actuator pushes the target downward (green) while a new volume is detected and added to the GMM (yellow); and c) volumes downstream of the control grid (x > 6) are deleted from the GMM while newly detected volumes (red) are added.

Goldstein is leading the project, “Reaping the Whirlwind: Re-energizing Boundary Layers by Targeted Manipulation of Coherent Structures.” The goal of this research is to develop physics-based simulations that will allow researchers to manipulate and redirect these large-scale turbulent flows. Goldstein’s team is also collaborating with researchers from the Rensselaer Polytechnic Institute who will use sensors to track LSM structures and actuators to try to deflect them in a wind tunnel to validate the UT group’s numerical simulations. Saikishan Suryanarayanan, a research associate in ASE/EM, is a co-principal investigator on the project, as well as Bakolas.

“We have taken a different approach to trying to re-energize the flow near the surface of a body like an airplane wing. We do not want to simply add energy to the flow, say with some little jet. We want to leverage the existing eddies in the natural turbulence to do that work – to reap the energy in those whirlwinds, quite literally,” Goldstein said.

Alexandros Tsolovikos, an award-winning Ph.D. graduate student, and Akshit Jariwala, a new graduate student joining the group, will play instrumental roles in bridging the gap between control theory and computational fluid dynamics. These students, both co-advised by Bakolas and Goldstein, alongside Suryanarayanan, will be creating direct numerical simulations of the turbulent flows, data-driven models describing the evolution of LSMs, as well as detection and control algorithms.

Real-world applications of this work could include advancing the performance of vehicles, resulting in less fuel consumption and a reduction in greenhouse gas emissions, or developing new approaches to extract renewable energy from large arrays of wind turbines. The work will also promote nationwide efforts to integrate data sciences and traditional engineering fields such as control theory and fluid dynamics.