On efficiency of fire simulation realization: parallelization with greater number of computational meshes

Abstract

Abstract. Current fire simulation systems are capable to utilize advantages of highperformance computer (HPC) platforms available and to model fires efficiently in parallel. In this paper, efficiency of a corridor fire simulation on a HPC computer cluster is discussed. The parallel MPI version of Fire Dynamics Simulator is used for testing efficiency of selected strategies of allocation of computational resources of the cluster using a greater number of computational cores. Simulation results indicate that if the number of cores used is not equal to a multiple of the total number of cluster node cores there are allocation strategies which provide more efficient calculations. IntroductionSimulation of fires belongs to computationally demanding tasks requiring efficient realization on a HPC (high-performance computer) system. Contemporary fire simulation systems are capable to incorporate many physical and chemical processes related to fire such as combustion, pyrolysis, thermal radiation, heat transfer, turbulence, gas spread dynamics, etc. Fire Dynamics Simulator (FDS) [1] is an advanced fire simulator utilizing the knowledge of CFD (computational fluid dynamics). It numerically solves a form of Navier-Stokes equations for low-speed fire induced flows with emphasis on transport of smoke and heat from the fire. FDS has been tested and validated for various types of fires and environments. It supports four programming models: sequential model designed for sequential computers, parallel MPI (Message Passing Interface) model designed for distributed memory systems, multithreading OpenMP (Open Multi-Processing) model designed for shared memory systems and hybrid MPI&OpenMP model designed for distributed shared memory systems. In our previous research we studied the impact of parallelization on efficiency and accuracy of a corridor fire simulation [2, 3] using a single computational node of the HPC cluster at the Institute of Informatics of Slovak Academy of Sciences in Bratislava (Slovakia). The research focused on the parallel MPI model of FDS and on testing the main strategies of allocation of available computational resources to given parallel calculations. In this paper we study efficiency of the parallel MPI model for simulation of the same type of fire using greater numbers of computational meshes and nodes.