Engineering-Based Thermal CFD Simulations on Massive Parallel Systems

Abstract

The development of parallel Computational Fluid Dynamics (CFD) codes is a challenging task that entails efficient parallelization concepts and strategies in order to achieve good scalability values when running those codes on modern supercomputers with several thousands to millions of cores. In this paper, we present a hierarchical data structure for massive parallel computations that supports the coupling of a Navier-Stokes-based fluid flow code with the Boussinesq approximation in order to address complex thermal scenarios for energy-related assessments. The newly designed data structure is specifically designed with the idea of interactive data exploration and visualization during runtime of the simulation code; a major shortcoming of traditional high-performance computing (HPC) simulation codes. We further show and discuss speed-up values obtained on one of Germany's top-ranked supercomputers with up to 140,000 processes and present simulation results for different engineering-based thermal problems.

Design and optimisation of an efficient HDF5 I/O Kernel for massive parallel fluid flow simulations

“…The main contribution is a hierarchic data structure that inherently supports distributed computing and already allows for in situ data exploration during runtime. This follows the explanations of Frisch et al…”

Measuring and Comparing the Scaling Behaviour of a High-Performance CFD Code on Different Supercomputing Infrastructures

“…The mathematical background of the code is described in detail in Frisch et al [1]. This section aims at giving a concise introduction into the mathematical modelling and into the data structure in order to bring the reader up to speed and to motivate the usage of HPC methods.…”

“…Figure 8 shows a classroom setup, where the human occupants are coupled to a thermoregulation model imposing thermal boundary conditions onto the surfaces, and acting as driving forces for the natural convection scenario. Further information and result evaluation can be found in Frisch et al [1]. Thus, the code is able to handle quite different physical scenarios on different scales while running on more than 100,000 cores.…”