Session: 02-03-01: Optimization

Paper Number: 146568

146568 - Development of Openmp Based Parallel Multiphysics Topology Optimization Software for Dc Bus Bar 

In high current-density DC bus bars, Joule heating causes an increase in temperature and leads to a rise in electrical resistivity. This creates a thermal hotspot, which damages the semiconductor devices and adhesive material. The heat transfer area is increased to prevent hotspots. This can cause an oversized design and increase weight and material costs. To overcome these issues, we have developed a topology optimization software for the optimal distribution of material without violating physics. Thermal and electrical models are used for thermo-electric physics analysis. These models are strongly coupled. The heat generated by the electrical Joule heating is taken as the heat generation source term for the thermal model, and the effect of temperature on electrical resistivity has been taken.

We have used an iterative density-based topology optimization method in the present work. In each iteration, the strongly coupled thermo-electric governing equations are simulated by the multiphysics finite element analysis (FEA). An adjoint-based sensitivity method is used for the sensitivity analysis, and a gradient-based optimization algorithm (MMA) is used for optimization.

We have developed an object-oriented C++ program for topology optimization. The code has been verified with commercial finite element software COMSOL Multiphysics. Our topology optimization algorithm can optimize the domain by reducing more than 50% of the material from the initial state. While testing the effect of grid size on the optimal area, it is found that the optimal area converges for medium grid size. However, the decrease in the grid size prevents the checkerboard pattern and grey scale region and shows the finer distribution of optimized material. For small grid sizes, computational time increases. From the time analysis, it has been found that multiphysics FEA, optimization algorithm MMA and filter process take up a significant chunk of computation time. In the present work, we use an application programming interface (API) based on the OpenMP multithread parallel tool to create a parallel program. We have taken different cases with different grid sizes. We tested our parallel program in each case with the all-grid size. The result shows that after speeding up is increasing with increase in threads, but after 8 threads parallel efficiency decreases for coarser grid size. 

We have taken different cases with different grid sizes. We tested our parallel program in each case with the all-grid size. The result shows that after speeding up is increasing with increase in threads, but after 8 threads parallel efficiency decreases for coarser grid size.

Presenting Author: Krishna Mohan Singh Indian Institute of Technology Roorkee

Presenting Author Biography: Krishna Mohan Singh is a Professor in the Department of Mechanical and Industrial Engineering at IIT-Roorkee. He graduated from IIT-BHU, Varanasi, and obtained his M. Tech. and Ph. D. from IIT-Kanpur. He has worked in academia and industry across the globe (India, UK, Japan). He was Senior Visiting Researcher (HIVIPS Fellow) at Hitachi Research Group, Hitachi Ltd., Japan; Honorary Visiting Lecturer and Post-doctoral Research Assistant at Queen Mary, University of London, UK; and JSPS Fellow and Assistant Professor at Shinshu University, Nagano, Japan. His main interest areas are computational mechanics, fluid dynamics, renewable energy, sustainable development and CAD of thermo-fluid systems. He is a Fellow of the Institution of Engineers, a Senior Member of AIAA, Member of ASME, FMFP, and JSME.

Authors:

Hitesh Kumar Sinha Indian Institute of Technology Roorkee
Krishna Mohan Singh Indian Institute of Technology Roorkee