Researchers Use XSEDE Allocated Resources to Simulate Turbulent Flows
|Lagrangian Coherent Structure (LCS) models provide insight into turbulence. In this study, the authors used high-fidelity large eddy simulation models generated by Comet and Bridges to accurately predict the turbulent flow field. Image courtesy of Bashar Attiya, Lehigh University|
By Kimberly Mann Bruch, SDSC Communications
Turbulent flows are everywhere, even though they are often invisible to the human eye. Using supercomputers such as the ones allocated through XSEDE, researchers can now create comprehensive simulations of such flows involved in the engineering of an array of structures, such as off-shore construction facilities.
Researchers from Lehigh University did just that, publishing their findings in a Canadian Journal of Physics article called Vortex Identification in Turbulent Flows Past Plates Using Lagrangian Method, after using Comet at the San Diego Supercomputer Center (SDSC) and Bridges at the Pittsburgh Supercomputing Center (PSC).
"To develop these descriptive simulations, we used Comet and Bridges to run three-dimensional simulations to better understand the physics of fluid motion and the vortex dynamics in the vicinity of specific plates and surfaces," said Bashar Attiya, a Ph.D. candidate at Lehigh University and first author of the paper.
Such visual simulations let researchers better understand the pattern of turbulent flows. The scientists used a specific type of flow visualization called Lagrangian Coherent Structure (LCS) to show trajectory patterns of fluid dynamic systems over a specific period of time. The skeleton of the flow pattern that is sustained in the flow is considered a coherent structure.
"Geophysical flows such as weather patterns, hurricanes, oil spills, and ocean currents are examples of LCSs," explained Attiya. "Because we interact with the fluid flows on a daily basis, using LCSs to understand the complex nature of such fluids is essential, and will help to convert complex issues into more tangible and clear problems that can be solved."
In contrast to other visualization techniques such as the Eulerian method that provides instantaneous flow structures with user-defined thresholds, the LCS model provides temporal characteristics of the flow field with an objective threshold. That is, LCS allows for the detection of the fluid motion level by observing and recording velocity field changes. Identifying LCSs in the fluid dynamic systems can help with engineering design in terms of characterizing fluid-structure interactions.
While the LCS model provides a more accurate flow physics prediction, this type of transient simulation requires large-scale computational resources such as the ones allocated through XSEDE, as well as enough storage space for the large amount of three-dimensional data that is generated to obtain flow field information.
"Thanks to Comet and Bridges, we were able to complete these high-resolution simulations in a very timely manner," said Attiya. "In the past, our computer resources were not capable of modeling three-dimensional problems with high enough resolution, but in this study we were able to use supercomputers to accurately predict the turbulent flow field."
Alparslan Oztekin is Bashar's current Ph.D. supervisor and senior author of the new paper. It is the second year that the group received allocations from XSEDE. "We submitted the original proposal in 2017 and got an allocation for all of 2018," said Bashar. "We renewed the allocation by submitting another proposal last year, so we have all of 2019 to continue our work."