A canonical flow in fluid mechanics is flow past a circular cylinder. At about Reynolds number ~ 300,000 a curious phenomenon is observed: this is the drag crisis. We report recent progress on wall-resolved large-eddy simulation (LES) of flow past a circular cylinder. Three configurations are considered: a smooth cylinder, a grooved cylinder and a rotating cylinder. An examination of the local skin friction and flow separation details indicates that the drag crisis is not necessarily attributed to the boundary layer transition from a laminar to a turbulent state. In addition, we discuss recent results of wall-modeled LES of flow past airfoils with an emphasis on the strong validation against experiments. The wall model is essential to avoid resolving the flow at the wall and involves the derivation of a Dirichlet boundary condition for the velocity at a virtual wall.
Acknowledgement: The Shaheen-Cray XC40 at KAUST was utilized for all the simulations.
Finding the solutions of large sparse linear systems is part of our daily work here at the Lab. These linear systems are ubiquitous, and appear in a wide range of applications in computational science. The goal of this seminar is to impart a working knowledge of the sparse direct and iterative methods used to solve these systems. This presentation will provide an overview of the algorithms, data schemes, and available software that one can use to both understand the methods and know how to best use them.
Scientific visualization enables insight, verification, and communication for presentations and publications. Visualization is closely tied to analyzing and exploring data generated by simulations and acquired in experiments. Animation is effective for representing complex behavior of variables over time. I will present software tools and techniques available for scientific visualization. Interactive programs such as VisIt and Paraview have a graphical user interface for exploring and displaying data. Visualization workflows developed in Python for PPPL projects will also be presented.
Optimization and testing for the gyrokinetic PIC codes ORB5 and XGC
Like many legacy codes, the gyrokinetic PIC code ORB5 was written without GPUs or multi-core processors in mind and was thus unable to take full advantage of modern supercomputer architecture.
Here, I summarize my work in modularizing, introducing new algorithms, and implementing OpenMP into ORB5.
I then summarize recent progress with XGCa GPU functionality and unit testing.
Engineering GFDL’s Climate Models For Future Architectures
This tutorial will introduce the Message Passing Interface (MPI), the most widely used method for distributed parallelism on small departmental clusters as well as on large leadership class computers. Practical use of MPI will be discussed, along with more advanced capabilities.