In this paper, we summarized the concept of multi-scale / multi physics analysis, Digital Twin and Digital Thread, that integrates design/development, production, and measurement in cyber physical space. These methods are necessary for practical CAE and the 21st century type design and development process. Also, the educational methods that are needed to utilize these novel methods are discussed.

Many soft materials show self-assembly structure such as microphase separation of block copolymer, micelle structure of surfactant, liquid crystalline and crystalline structures, and such structures play important roles in the function and properties of materials. Those higher order structures are much larger than single molecular structure, and fine atomistic simulations are not practical in the sense of computational power. Thus, the multiscale simulation including fine atomistic simulation and coarse-grained model is necessary for applying the computational simulation to soft material design. There are many coarse-grained models, and they are based on different physical background. Multiphysics simulation, combining coarse-grained model of different physical background is also necessary. This article introduces basic approaches of coarse-grained simulation tools for soft materials, OCTA, and examples the application. Recently, in addition to the computational simulation, the application of AI for material designs is getting widely used. We also introduce an example of the combination of AI and multiscale / multiphysics simulation.

Micro-macro scale coupling boundary value problem is formulated on the basis of mathematical homogenization theory. This multiscale boundary value problem can be numerically solved using finite element method. However, this approach has a critical problem on the computational effort for industrial applications. Based on numerical material testing, an effective approach called as decoupled scheme is introduced to approximate the corresponding solution of micro-macro scale coupling boundary value problem within practical computational effort.

Several multi-scale and multi-physics simulations were performed to understand the mechanical properties of polymer nanocomposites such as filler-filled rubber system for tire rubber. In this system, the hierarchical structure of filler nanoparticles was estimated from scattering experimental data by Monte Carlo searches. As the first approach, a coarse-grained molecular dynamics model based on the Kremer-Grest type spring bead model was used as the coarsest model that reproduces polymers that do not intersect with each other. Long polymers that were entangled with each other were placed in the space with the hierarchical model of fillers, and crosslinking was set. Virtual experiments such as uniaxial stretching can reproduce various nano-behaviors such as stress-strain curves and fractures with nanovoids.

Computational technique for fluid-structure interactions (FSIs) is one of the elemental techniques in multi-scale and multi-physics modeling and simulations. This is because we frequently face computation objects that require simultaneous simulation of dynamic behaviors of fluid and structures. This article explains two computational FSI methods. One is the interface-tracking (IT) method that suits for computation of FSIs of which computation accuracy of the boundary layer is important, and the other is the interface-capturing (IC) method that is suitable for FSIs with large deformation and multi-body coupled problems. FSI simulations of flutter of a thin membrane and mixing and sedimentation of elastic particles are shown as demonstrational examples.

Simulation of gas-liquid two-phase flows include multi-scale and multi-physics problems. We have developed a weakly compressible computational scheme for incompressible flows with a purely explicit time integration and successfully implemented on GPU (Graphics Processing Unit) providing very high computational performance. AMR (Adaptive Mesh Refinement) method is available for the scheme and high-resolution meshes are assigned dynamically near the region of the gas-liquid interface. The AMR method greatly enhances the computational efficiency and makes it possible to compute the dynamics of thin liquid film such as a soap bubble which is a typical multi-scale problem. At the surface, viscoelasticity becomes more important and transportation of surfactant should be taken into consideration. It has been confirmed that the Marangoni effect stabilizes and avoids breaking-up a thin film covering a bubble on the water surface against the gravity in the simulation.

Non-metallic inclusions in the material of rolling bearings induce subsurface initiated spalling. In the previous paper, from the life test results of the ball bearings made from laboratory melted low cleanliness bearing steels, it was confirmed that the improvement of the bearing life can be derived through bonding of the oxides with the matrix by hot isostatic pressing (HIP). In order to confirm the universality of the life improvement effect, rolling contact fatigue (RCF) tests for HIPed specimens and non-HIPed specimens made from eight heats of mass-produced bearing steel were carried out. From the experimental results, it is confirmed that the RCF life is improved by the HIP even in mass-produced steels with various composition and morphology of inclusions, as in the case of the laboratory melted low cleanliness bearing steel. Furthermore, RCF life of non-HIPed material is found to correspond to the morphology of oxides related to combining with sulfide and the severity of the cavity between oxides and the matrix. It was also suggested that the degree of life improvement by the HIP is influenced by calcium in the simple sulfide.

One of the mechanical learning methods, logistic regression, is applied to the analyses on the rail damage detected on the high rails at the curve sections. The effect of the difference of rail grades on the rail damage severity is estimated to obtain the ideas on the rail grade selection at the curve sections. The logistic regression model with the optimized parameters indicates that an as-rolled rail lowers the rail damage severity compared with a head hardened one on average within the curve radius range from R500 to R800m. From a view point of the damage severity, an as-rolled rail is favorably selected as a high rail in the outer curves with the radius of more than R500 to R600m. Considering the evolution of maximum wear rate of the rail head, a head hardened rail tends to have a little bit smaller damage severity than an as-rolled rail in case the maximum wear rate of an as-rolled rail is quite large. This may attribute to the decrease of the stress loaded by the inhomogeneous worn profile of the rail head.