Large-Scale Parallel Thermal Elastic-Plastic Welding Simulation Using Balancing Domain Decomposition Method

Author(s):  
Yasunori Yusa ◽  
Yuma Murakami ◽  
Hiroshi Okada

Abstract From the mechanical point of view, welding processes can be modeled as a coupling problem between heat conduction and thermal elastic–plastic problems. Such a welding mechanics problem generally requires a large amount of computational time due to its nonlinearity as well as a lot of time steps with a moving heat source. To overcome this difficulty, we are developing a large-scale welding simulator based on the domain decomposition method, which is one of parallel finite element methods. In the present paper, the methodology of the domain decomposition method that is applied to welding analysis is presented, followed by the algorithm. Then, a bead-on-plate problem, which is a popular benchmark problem in the field of computational welding mechanics, was analyzed by the simulator. The bead-on-plate problem was successfully analyzed within very small numbers of iteration steps of the Newton–Raphson and conjugate gradient methods.

2011 ◽  
Vol 462-463 ◽  
pp. 605-610 ◽  
Author(s):  
Hiroshi Kawai ◽  
Masao Ogino ◽  
Ryuji Shioya ◽  
Shinobu Yoshimura

To solve a large scale elasto-plastic dynamics analysis of a complicated structure, such as a seismic analysis of a nuclear power plant and a skyscraper, a new implementation strategy for a parallel finite element code, suitable on a parallel supercomputer with modern multi-core / many core scalar CPUs, has been required. In this work, we propose a new design and programming style to optimize the performance of a parallel finite element code based on the domain-decomposition method (DDM) on multi-core CPUs, considering their cache hierarchy. Instead of a traditional, memory access-intensive approach, DS (Direct solver-based matrix Storage), two new matrix storage-free approaches, DSF (Direct solver-based matrix Storage-Free) and ISF (Iterative solver-based matrix Storage-Free), are proposed. Our new DSF/ISF-based DDM solver is not only more efficient in memory usage but also comparable in computational time against existing DS-based DDM solvers on multi-core CPU architectures.


Sign in / Sign up

Export Citation Format

Share Document