2019/10/30 by Janine Glänzel, Glänzel, Janine, Andreas Naumann +3
Computer Science · Engineering · Physics and Astronomy · #Distributed #FOS: Computer and information sciences #Heat Transfer and Optimization #Model Reduction and Neural Networks #Modeling and Simulation Systems #Parallel #and Cluster Computing (cs.DC)
paper · pdf · doi:10.48550/arxiv.1910.13939
openalex publication_date 2019/10/30 · openalex created_date 2022/07/28 · openalex updated_date 2026/07/28
Decoupling approach presents a novel solution/alternative to the highly\ntime-consuming fluid-thermal-structural simulation procedures when thermal\neffects and resultant displacements on machine tools are analyzed. Using high\ndimensional Characteristic Diagrams (CDs) along with a Clustering Algorithm\nthat immensely reduces the data needed for training, a limited number of CFD\nsimulations can suffice in effectively decoupling fluid and thermal-structural\nsimulations. This approach becomes highly significant when complex geometries\nor dynamic components are considered. However, there is still scope for\nimprovement in the reduction of time needed to train CDs. Parallel computation\ncan be effectively utilized in decoupling approach in simultaneous execution of\n(i) CFD simulations and data export, and (ii) Clustering technique involving\nGenetic Algorithm and Radial Basis Function interpolation, which clusters and\noptimizes the training data for CDs. Parallelization reduces the entire\ncomputation duration from several days to a few hours and thereby, improving\nthe efficiency and ease-of-use of decoupling simulation approach.\n