In-Situ Processing and Visualization for Direct Numerical Simulation of Coolant Flow through Mixing Vanes

摘要

As we move into peta- and exa-scale high performance computing, the researchers have unprecedented capability to simulate and study complex problems with high fidelity. However, at the same time, there is a tremendous challenge to analyze and process the vast amount of data generated by these high-fidelity simulations in order to understand and gain meaningful insight into complex phenomena such as turbulent two-phase flows. The traditional workflow, which consists in conducting simulation on massively parallel supercomputers and recording vast amount of raw simulation data to disk storage for further post-processing and visualization, is no longer a viable approach due to prohibitive cost of disk access and considerable amount of time spent on data transfer. One approach to tackle this issue is in-situ processing and visualization that couple a simulation with post-processing so visualization occurs while the simulation is running. This in-situ approach minimizes data storage by extracting and visualizing important features of the data directly within the simulation without saving the raw data to disk. In this paper, we present our approach to in-situ processing and visualization of simulation data generated by massively parallel finite-element computational fluid dynamics solver (PHASTA) instrumented and linked with the ParaView Catalyst co-processing library. We demonstrate our in-situ processing and visualization capability with a fully resolved turbulent flow through 2x2 reactor subchannel geometry with mixing vanes and spacer grids. The in-situ visualization for our turbulent flow simulation has been conducted on the supercomputer Cray XK7 "Titan " (located at Oak Ridge National Laboratory) using 18,432 computing cores.