Search

Abstract: We present here the enhancement of a parallel incompressible Navier–Stokes solver to be able to manage very large meshes. Mesh generation in engineering applications is often the bottleneck of the complete simulation process. The mesh is the basis of the discretization algorithm and the first “lego” of a simulation. A mesh should approximate well the necessary geometrical elements of the computational domain. In addition, it should be fine enough to capture the relevant physical scales of the engineering problem. Usually, commercial mesh generators do well with the first task. They include refinement tools for boundary layer elements and local adaptivity. However, it is quite difficult to generate very large meshes (say of the order of thousands of millions of elements) with the available tools. The idea of this work is to implement a parallel uniform mesh multiplication in a HPC code developed at Barcelona Supercomputing Center named Alya. [Copyright &y& Elsevier]

Abstract: Domain Decomposition Methods (DDMs) are techniques that divide the solution of a PDE on a domain into smaller solutions on smaller subdomains coupling them using a certain strategy. They are used for essentially two purposes: designing parallel solvers and/or coupling subdomains with different meshes, different numerical approximations, etc. In this paper we are interested in this last category. One example of application is the Chimera method. In that sense, the Chimera method can be viewed as a preprocess technique plus a DDM on overlapping and non-conforming subdomains. The coupling technique of DDM is usually achieved via transmission conditions to impose the continuities of the unknown and its flux across the subdomain boundaries. We propose in this work an alternative coupling strategy, intervening as a preprocess method. It consists in connecting the nodes of one subdomain with the nodes of the adjacent subdomains via newly created elements. In this way, the multi-domain character of a DDM disappears, making it a parallel, implicit and versatile method. We discuss in this paper the relation between the proposed method and the existing coupling strategies. We also present some convergence results as well as some applications to the Navier–Stokes equations and other PDE’s. [Copyright &y& Elsevier]

We present in this paper a numerical strategy for the simulation of rotary positive displacement pumps, taking as an example a gear pump. While the two gears of the pump are rotating, the intersection between them changes in time. Therefore, the computational domain should be recomputed in some way at each time step. The strategy used here consists in dividing a cycle into a certain number of time steps and obtaining different computational meshes for each of these time steps. The coupling between two consecutive time steps is achieved by interpolating the flow unknowns in a proper way. This geometrical decomposition enables one to have a plain control over the mesh, particularly in the zones of interest, which are the gap between the gears and the casing, and the engagement and disengagement zones of the gears. [Copyright &y& Elsevier]

The complexity of combustion simulations demands the latest high-performance computing tools to accelerate its time-to-solution results. A current trend on HPC systems is the utilization of CPUs with SIMD or vector extensions to exploit data parallelism. Our work proposes a strategy to improve the automatic vectorization of finite-element-based scientific codes. The approach applies a parametric configuration to the data structures to help the compiler detect the block of codes that can take advantage of vector computation while maintaining the code portable. A detailed analysis of the computational impact of this methodology on the different stages of a CFD solver is studied on the PRECCINSTA burner simulation. Our parametric implementation has proven to help the compiler generate more vector instructions in the assembly operation: this results in a reduction of up to 9. 39 × of the total executed instruction maintaining constant the Instructions Per Cycle and the CPU frequency. The proposed strategy improves the performance of the CFD case under study up to 4. 67 × on the MareNostrum 4 supercomputer. • A parametric configuration of the CFD data structures is presented for enabling auto-vectorization. • A detailed performance analysis using different compilers is applied to a real CFD case. • A reduction of up to 9.39 times of the total executed instruction is attained, maintaining constant the IPC and the CPU frequency. • The proposed strategy accelerates up to 4.67 the PRECCINSTA burner simulation on the Marenostrum4 nodes. [ABSTRACT FROM AUTHOR]

This paper presents a load balancing strategy for reaction rate evaluation and chemistry integration in reacting flow simulations. The large disparity in scales during combustion introduces stiffness in the numerical integration of the PDEs and generates load imbalance during the parallel execution. The strategy is based on the use of the DLB library to redistribute the computing resources at node level, lending additional CPU-cores to higher loaded MPI processes. This approach does not require explicit data transfer and is activated automatically at runtime. Two chemistry descriptions, detailed and reduced, are evaluated on two different configurations: laminar counterflow flame and a turbulent swirl-stabilized flame. For single-node calculations, speedups of 2.3x and 7x are obtained for the detailed and reduced chemistry, respectively. Results on multi-node runs also show that DLB improves the performance of the pure-MPI code similar to single node runs. It is shown DLB can get performance improvements in both detailed and reduced chemistry calculations. • A load balancing strategy for reaction rate and chemistry integration is presented. • It uses the DLB library to redistribute the computational resources at node level. • Code hybridization improves the performance over MPI-pure implementations. • Single-node and multi-node tests show speedups in reacting flow calculations. [ABSTRACT FROM AUTHOR]