Abstract:
"AC 2008-2725: DESIGN OF EXPERIMENTS APPROACH TO VERIFICATION AND UNCERTAINTY ESTIMATION OF SIMULATIONS BASED ON FINITE ELEMENT METHOD"
This paper introduces a systematic approach to enhancing the reliability and accuracy of Finite Element Method (FEM) based simulations through the application of the Design of Experiments (DOE) methodology. This approach is presented as a powerful tool for verification and uncertainty estimation in computational modeling. The document highlights the challenges in managing computational resources and analyzing the vast amounts of output data generated by modern FEM software. It proposes that DOE can provide an efficient and structured way to identify the key input variables and parameters that significantly influence simulation outcomes. By using DOE, researchers and engineers can systematically vary these inputs, perform a controlled set of simulations (e.g., full-factorial or fractional-factorial designs), and then statistically analyze the results to understand the impact of each variable and their interactions. This systematic analysis not only aids in verifying the correctness of the simulation code and model but also helps in quantifying the inherent uncertainties in the predictions. The paper illustrates the practical application of this DOE approach with examples, including computational fluid dynamics (CFD) and structural analysis problems using popular FEM software packages such as ABAQUS and ANSYS. The methodology aims to streamline the V&V process, leading to more robust and trustworthy FEM predictions, which is crucial for safety-critical engineering applications.
