Title:	Topology Optimization considering high-resolution 3D Multiphysics Fluid-Thermal-Structural models
DNr:	LiU-compute-2021-40
Project Type:	LiU Compute
Principal Investigator:	Jonas Lundgren <jonas.lundgren@liu.se>
Affiliation:	Linköpings universitet
Duration:	2021-09-06 – 2022-10-01
Classification:	20301
Keywords:

Abstract

In a previous project (SNIC 2020/13-47), access to the supercomputing resources of NSC was shown to have a great effect on the possibilities of developing a state-of-the-art 3D Topology Optimization (TO) design model that fully utilizes the design freedom of Additive Manufacturing (AM), a manufacturing method capable of containing a very complex interior geometry. The model is customized for designing gas turbine components subjected to very hot gas flows, but the general concept is applicable to a vast range of design challenges involving structural integrity and/or temperature/coolant distribution issues. Previously, a Matlab code has been developed for this purpose, and it has successfully shown the general concept of this model. However, it is believed that the PETSc suite constitutes a more solid base for this project, since it enables the model to take care of higher resolutions. These high-resolution geometry representations are needed to utilize the enormous design flexibility that is introduced by the AM printers, where every point in space either can be filled or empty. If such a complex geometry should be the basis of a finite element (FE) model, the computational resources needed are very large. This complexity is further extended by the fact that several physical domains is involved in this model, since both structural integrity, flow properties and temperature performance is of interest to the final component. A high-resolution three-field model is believed to have a great impact on the design process of gas turbine components. This TO-based model introduces the possibility to design components that can handle an increased temperature, which gives an opportunity to introduce more environmentally friendly fuels, such as hydrogen, for the operation of the turbine, as well as a possibility to increase the efficiency of the turbine. To further develop this high-resolution three-field model in 3D, it is imperative that supercomputing resources are further available to this project. If so, a migration from Matlab to a PETSc environment is possible, which is believed to be the key to obtaining a model where the resolutions are good enough for actually building them in the AM printers, something that later can be used to validate the model.