The contact and non-contact behaviors of material interfaces generate harmonics arising from interactions with large amplitude incident waves. In this phenomenon, which is called contact acoustic nonlinearity (CAN), intermittent impacts by an incident wave cause the interface contact phase to alternate between contact and separation. This paper proposes a method for numerical simulation of CAN using an elastodynamic finite integration technique (EFIT). The accuracy of the EFIT simulation was verified through comparison with an analytical solution. In one-dimensional CAN theory, a wave penetrating at an interface demonstrates a sawtooth waveform indicating that the interface is closing with a constant velocity. Simulation results revealed that the closing velocity is determined by the compressive stress of the material, and experimental measurements on a polymethylmethacrylate specimen revealed the harmonics caused by the sawtooth wave.
This paper deals with finite element analysis involving tens of billions of degrees of freedom (DOF) in a high-frequency electromagnetic field. The iterative substructuring method has been considered to be an efficient parallel computing method. To show the possibility of analyzing electromagnetic field problems with complex numbers and tens of billions of DOF, problems with up to 30 billion DOF are analyzed by all nodes of an Oakleaf-FX supercomputer. As a result, the human model has been successfully solved in approximately 10 minutes, and the simple hyperthermia applicator model with 30 billion DOF has been successfully solved in approximately 19 minutes. There is a problem in the output analysis results, and bugs relating to the limits of the 32-bit integer data type have been found and fixed through actual analysis of problems with tens of billions of DOF.
The practical use of the fluid-structure interactions (FSI) analysis system requires the robustness for the parametric design of artifacts. In the FSI analysis employing interface tracking methods, mesh moving technique is significant for avoiding the failure of the analysis due to the mesh distortion. Although many conventional techniques have been introduced in the system, there still remain some cases where the analysis fails because of the mesh distortion. For the further improvement of the robustness, we propose a new mesh moving technique, minimum-height-based stiffening technique, where the mesh deformation of the fluid domain is virtually governed by the linear elastic equations and the stiffness of each element is determined according to its minimum height. The proposed technique is applied to two-dimensional benchmark problems with three types of prescribed motions or deformation: translation, rotation, and bending. The results were compared with those with Jacobian-based-stiffening technique, which is one of the most effective approaches, in terms of mesh quality factors. As a result, the proposed technique shows better performance, i.e. the improvement by more than 10% for our mesh quality factor. In addition, low sensitivity of the mesh quality factors to the optimum value of the control parameter was observed. This low sensitivity can contributes to the usability of our proposed technique.