The present study proposes a multi-material topology optimization method considering the strengths both of solid materials and interface. In order to consider multiple factors of Young’s modulus and strength at the same time, the DMO material interpolation function that the coefficients are given equivalently is applied. To resolve the singularity problem, we propose the stress definition adapted to the DMO material interpolation function. Furthermore, the graded interface with a two-step density filter is introduced to represent the interface on a density-based method. The tension/compression asymmetric interfacial strength criteria makes it possible to obtain an optimal layout in which the interface is not subjected to tensile strain. The validity of the method proposed in this study is verified with three structural models.

A Faster and data-driven method by deep learning, compare with a traditional method by FEM, is proposed for thermal simulation of Wire Arc Additive Manufacturing (WAAM). In this paper, cross sectional temperature distributions of WAAM-made parts are predicted and shows a good agreement with FEM method.

This study develops a numerical method for large-scale parallel analysis in Euler description based on the mesh-free method for optimal topology design. Therefore, we propose an upwind scheme for the least squares moving particle semi-implicit (LSMPS) method to stabilize numerical oscillations associated with the Euler description. In addition, we extend the distributed memory parallelization method based on a conventional overlapping domain decomposition to the graph structure of the particles, and develop a parallel convection–diffusion analysis solver based on our proposed upwind LSMPS method. Furthermore, we perform a verification and scaling tests of the proposed method by solving the 3D convection–diffusion problem.

This study proposes a velocity-based space-time finite element method (v-ST/FEM) that enables a large deformation analysis of solids over arbitrary moving mesh (AMM) via simple and rigorous formulation in the space-time domain. The proposed method solves quasi-static problems with a hypoelastic constitutive model. The Cauchy stress convected through AMM is incrementally updated following the relative velocity between the moving mesh and the solid deformation, which results in the weak form for the primary unknown of velocity. An iterative algorithm for solving the discretized system equation is implemented, whereby geometrical nonlinearity is considered by the iterative update of the stress over the moving mesh. Numerical analyses of benchmark problems such as simple tensile deformation, rotating cylinder under compression, and non-uniform large tensile deformation have been conducted. The numerical results have shown that v-ST/FEM achieves as accurate computation over AMM as over Lagrangian mesh and have demonstrated the applicability of AMM.

A numerical method is proposed in the present paper for structural analysis of large-scale assembly structures. The mortar finite element method (FEM) has been developed for assembling structural components modeled by finite elements and a set of constraints in a weak form, which is formulated using Lagrange multipliers. In the dual Lagrange multiplier method proposed by Wohlmuth, a set of biorthogonal shape functions is used to discretize Lagrange multipliers. One of the present authors proposed a method to incorporate multi-point constraints (MPCs) into the balancing domain decomposition (BDD) method, which was proposed by Mandel. The method, which is called the BDD-MPC method in this paper, can solve large-scale structural problems having many MPCs at high speed using parallel computers. The proposed method for large-scale assembly structures combines the above two methods, i.e., the mortar FEM using the dual Lagrange multipliers and the BDD-MPC method. A numerical integration method using back ground cells for integrating the constraints in a weak form is also proposed. In this method, square and fine integration cells are arranged as a grid without considering the shapes of surface elements on the surfaces to be connected. The integration method is verified by investigating the levels of details of divisions of both meshes to be connected and the integration cells. In the illustrative example of two cubes that are connected by the dual Lagrange multipliers, very fine mesh division is necessary to obtain a solution with sufficient accuracy. It is demonstrated that the BDD-MPC method is a powerful tool to solve such a problem. Computation performances of the method are also investigated.

A numerical method to simulate fatigue crack propagation of a C(T) test specimen with cladding was developed. The method employs an XFEM using hexahedral continuum elements with only Heaviside step function, which can model planar crack independently of finite elements. In the method, the stress intensity factor for crack across dissimilar materials, are calculated without crack tip asymptotic fields, and the 3rd order Bezier curve is used to smooth the crack front geometry. Initially, to verify the stress intensity factor evaluation method used in this study, stationary crack analyses of C(T) test specimen with cladding are performed and the results are compared with those by FEM. And then, crack propagation analyses are executed by repeating XFEM analyses. The transition of the crack front geometry, and the relationship between crack length and number of load cycles are evaluated. Consequently, non-uniform crack propagation behavior can be numerically simulated.

In recent years, topology optimization methods have been applied not only to structural problems but also to fluid flow problems. Most of the previous studies assume steady-state flow. In contrast, this paper focuses on topology optimization for unsteady flow, which is more general from an engineering point of view. However, unsteady flow topology optimization involves solving the governing and adjoint equations of a time-evolving system, which requires a huge computational cost for topology optimization with a fine mesh. Therefore, we propose a large-scale unsteady flow topology optimization based on the building-cube method (BCM), which is suitable for massively parallel computing. BCM is one of the hierarchical Cartesian mesh methods and is confirmed to have very good scalability. The governing equations are discretized by a cell-centered finite volume method based on the BCM. The objective sensitivity is obtained by the continuous adjoint method. Several numerical examples are demonstrated to discuss its applicability to large-scale computations.

We propose a new multiscale method that combines the radial basis function (RBF)-based surrogate homogenization for elastoplastic composites and numerical material test (NMT) that allows for nonperiodic distributions of constituent materials within a representative volume element (RVE). In the offline process, we construct a constitutive database of the macroscopic stress-strain responses by conducting NMTs with a special constraint such that the volume integral of gradient of fluctuation displacement over the RVE domain is zero. Then, a surrogate model is created by using the constitutive database. In the online process, we conduct a numerical simulation of a macrostructure with the created surrogate model, followed by localization analysis using arbitrary strain histories. The validity of the proposed multiscale analysis method using RBF-based surrogate modeling is confirmed through these offline and online processes. In numerical examples, to demonstrate the modeling capability of nonperiodic microstructures, cubic and circular RVEs are targeted.

The present study focuses on combining data assimilation with a simple heat exchanger model to enable real-time simulations. First, a 0-dimensional (0D) heat exchanger model is proposed and then an impact assessment of applying different ensemble Kalman filter (EnKF) conditions on the 0D model is presented. Finally, the performance of the combined 0D model and EnKF is benchmarked against a 1-dimensional (1D) model. The survey of EnKF conditions reveals that an ensemble size of at least 10 members and a parameter noise of 5% standard deviation are crucial to balance the accuracy and convergence speed of parameter and state estimation of the 0D model. Under these conditions, the 0D model accurately predicted the behavior of an actual heat exchanger with higher accuracy than the 1D model. The results also showed that the proposed method completed 1s of the simulation period almost 2000 times faster than the 1D model.