Application Of Multi-domain Boundary Face Method In Steady-state Heat Conduction And Elastostatic Problems

Since BFM has many advantages, for instance it can obtain good accuracy, easyto deal with problems involve infinite domains, boundary and interface, singularity,moreover, it can realize the integration of CAD/CAE. This paper studies themulti-domain BFM to solve structures with complex boundaries, including differentkinds of materials and large scale models required parallel computing. On the basis ofthe boundary integral equation in BFM, this paper presents the process of assemblyand solution in multi-domain matrix. Furthermore, we make applications ofmuti-domain BFM in steady-state heat conduction and elastostatic problems. Theresults show that our method is correct and has certain feasibility in engineering.In our method, the discrete equations of each domain are assembled into anoverall equation, considering the boundary conditions of each domain and on theinterface. The method makes displacement and surface force (or potential and flux) onboundaries and the interface as unknown quantities. Although the matrix is full rankin each domain, they are assembled into a block matrix.On the application of multi-domain BFM, the article gives several examples toapply the muti-domain BFM to steady-state heat conduction analysis. These structuresinclude mechanical seal rings, crankshaft in engine, linear slide way and a concretedam which is a large scale model. The comparison of the results with FEM presentsthat the multi-domain BFM has feasibility and some advantages in the application ofheat conduction problems.On the application of elastotatic problems, displacement and stress analysis ofkey components in injector are performed. The numerical results of FEM and BFMare basically consistent. Besides, the stress clouds of BFM are smoother in the overalltransition and local small fillet, so BFM can reflect the stress concentration moreclearly. Furthermore, in BFM, using much less nodes than FEM can get equal stresscalculation accuracy than that with many nodes in FEM. In addition, the comparisonof the calculation of the stability of the two methods shows that with the changes ofnodes number in the model, the results of BFM changed little. This certifies that BFMhas better computational stability.