Optimizing Mesh and CFD Simulation Performance: A Multivariate Analysis Approach

Abstract

Listen

Translate article

The objective of the study is to analyze mesh parameterization through the weighting of results that can define an increase in precision and a reduction in simulation time. To optimize the performance of a mesh and CFD simulation, we will adopt a systematic approach. Initially, we will employ experimental design, with five factors varied in three replications. These factors directly influence three responses. After collecting the experimental data, an evaluation of the significance between the independent variables was carried out, proposing a regression function capable of explaining the behavior of the factors and their relationship with the defined responses. Next, we will identify the individual optimal points using the Ordinary Least Squares method, representing ideal configurations of factors that maximize or minimize the responses of interest. Based on the results obtained, the Payoff matrix will be assembled, containing the individual optimal points and the reactions of the other variables. Thus, the scaling of functions will be performed to formulate and employ the minimization of the Mahalanobis multivariate distance, considering an adequate weighting between the optimal responses. This approach will guarantee the attainment of balanced and efficient solutions. Finally, we will integrate the results of the Payoff matrix into the mixture arrangement methodology, allowing for the conclusion of the analysis of optimal points for each function weight, providing efficient responses for the continuous improvement of the process. This systematic approach, which combines techniques of experimental design, regression, optimization, and multivariate analysis, will be fundamental to maximize the efficiency of mesh and CFD simulation, ensuring precise and reliable results.