On Analyzing Quality of Data Influences on Performance of Finite Elements Driven Computational Simulations

Abstract

Abstract. For multi-scale simulations, the quality of the input data as well as the quality of algorithms and computing environments will strongly impact the intermediate results, the final outcome, and the performance of the simulation. To date, little attention has been paid on understanding the impact of quality of data (QoD) on such multi-scale simulations. In this paper, we present a critical analysis of how QoD influences the results and performance of basic simulation building blocks for multi-scale simulations. We analyze the impact of QoD for Finite Element Method (FEM) based simulation building blocks, and study the dependencies between the QoD of input data and results as well as the performance of the simulation. We devise and implement novel QoD metrics for data intensive, FEM-based simulations and show experiments with real-world applications by demonstrating how QoD metrics can be efficiently used to control and tune the execution of FEM-based simulation at runtime.

An Artifact-based Workflow for Finite-Element Simulation Studies

“…In the field of FEA, support has been focused on automating simulation experiments using scientific workflows and scripts [56,57]. To those simulation experiments belong traditional types of simulation experiments such as sensitivity analysis or optimization [58], but also FEA specific experiment types to assess questions of mesh quality, energy balance, or simulation accuracy [13,59], as well as more recent developments, e.g., uncertainty quantification based on surrogate models [60].…”