MechSE Seminars

View Full Calendar

Exploring Multiphase Flow Dynamics and Turbulence: Insights from Simulations and Data-Driven Modeling

Event Type
Mechanical Science and Engineering
4100 Sidney Lu Mechanical Engineering Building
Mar 21, 2024   4:00 pm  
Dr. C. Ricardo Constante-Amores, Chemical & Biological Engineering, University of Wisconsin-Madison
Amy Rumsey


Multiphase flows involve the simultaneous movement of multiple fluid phases, and play a crucial role in numerous applications ranging from oil and gas production to pharmaceutical manufacturing. Topological transitions in multiphase flows, exemplified by scenarios where interfacial distances vanish and velocity fields diverge, pose challenges that demand specialized numerical methods. Previous research to my work has often been confined to either pre-breakup/coalescence flows or axisymmetric studies, limiting their applicability to diverse real-life situations. I have used a hybrid front-tracking/level-set method for interfacial modeling that not only considers surface tension forces but also accounts for the impact of interfacial surfactant concentration gradients, giving rise to Marangoni stresses. One of the examples that I will discuss is the breakup of a turbulent jet through a cylindrical nozzle in the absence and presence of surfactants. I will show the crucial role of surfactants in the interfacial dynamics, such as the inhibition of capillary singularities. 

In the second part of the seminar, I will focus on single-phase turbulent flows. Turbulence is a common factor in various fields like aviation, manufacturing, and even human health (arterial flow). Gaining a fundamental understanding and controlling turbulence is crucial for improving energy efficiency for fluid transportation. The challenge lies in that the accurate simulation of complex dynamics in turbulent flows demands a substantial number of degrees of freedom, i.e. a high-dimensional state space. Then, there is a need for the development of low-dimensional reduced-order models. I will describe a data-driven reduced-order modeling method, that finds a nonlinear coordinate representation of a low-dimensional space using a machine-learning architecture called an autoencoder, and then learns an ordinary differential equation for the dynamics of this low-dimensional space. I apply this framework to learn the complex dynamics of turbulent pipe flow and other large-scale spatiotemporal turbulent systems.

 About the Speaker 

C. Ricardo Constante-Amores is a postdoctoral associate in the Department of Chemical & Biological Engineering at University Wisconsin-Madison. He received his BEng in Chemical Engineering from the Complutense University of Madrid in 2014, followed by his MSc and Ph.D from the Department of Chemical Engineering at Imperial College London in 2015 and 2021, respectively. After earning his Ph.D., he did a short postdoc between the Mathematical Institute and Engineering Science Department at the University of Oxford. Among Ricardo’s professional distinctions are the ERCOFTAC Da Vinci Medal, the Milton Van Dyke Award from the American Physical Society, the Osborne Reynolds Prize, and the Dudley Newitt Prize.

Host: Professor Leonardo Chamorro

link for robots only