Coupling time-varying modal analysis and FEM for real-time cutting simulation of objects with multi-material sub-domains

Abstract

Powerful global modal reduction techniques have received growing recognition towards significant performance gain in physical simulation, yet such numerical methods generally will fail when handling deformation of heterogeneous materials across multiple sub-domains involving cutting simulation. This is because the corresponding topological changes (due to cutting across multiple sub-domains) and/or drastic local deformations tend to invalidate the global subspace techniques. To ameliorate, this paper systematically advocates a novel deformation and arbitrary cutting simulation approach by adaptively integrating FEM-based fully-physical simulation and local deformation's modal reuse into a CUDA-enabled parallel computation framework. This paper's originality hinges upon the maximal reuse of the space-time-varying local modes from prior fully-physical simulations and the adaptively coupling of sub-domain behaviors, which give rise to great improvement of computational complexity while guaranteeing high-fidelity simulation effects. Other key advantages include, being independent of underlying physical models (e.g., either FEM or meshless methods), being flexibly accommodating sub-domains' heterogeneous material distributions, and being accurately responding to local user interactions. During the initialization stage, we partition the object into multiple sub-domains according to its material distributions and/or geometric structures, and respectively employ FEM for physics-based representation/simulation. During the dynamic stage, for each sub-domain, we leverage its local modal reduction in order to project complex deformations onto a low-dimensional subspace. We dynamically determine the sub-domain-specific switch between deformation reconstruction based on modal reuse and FEM-based physical simulation according to the physically-consistent error estimates, and couple all the sub-domains' physical behaviors together by imposing adjacent sub-domains' geometric-continuity constraints. To validate our method, we conduct extensive and quantitative evaluations over comprehensive and well-designed experiments, and all the experimental results have confirmed the advantages of our method in terms of efficiency, accuracy, and unconditional stableness in practical graphics applications.

A multi‐GPU finite element computation and hybrid collision handling process framework for brain deformation simulation

“…At the same time, a method based on the model order reduction technology has been proposed. 26 The computational accuracy of such method is highly dependent on the template library. Besides that, a fully nonlinear method is proposed for brain deformation, 8 which can only deal with coarse brain model because of their computation complexity.…”

“…In this method, the stiffness of each element is assumed linear in the local frame described by its rotated state, which enables us to simulate geometric nonlinearities (i.e., large displacements and large rotations but small strains), but the material nonlinearity is simplified. At the same time, a method based on the model order reduction technology has been proposed . The computational accuracy of such method is highly dependent on the template library.…”

A new ChainMail approach for real-time soft tissue simulation

“…2,3 The former is computationally efficient but does not allow accurate modelling of soft tissue material properties, whereas the latter is computationally extensive and formulations must be simplified to reduce runtime computation. 4 Various techniques have been reported to improve the computational performance of FEM. The matrix condensation reduces the computational time by confining the full computation of a volume mesh to the surface nodes; however, this simplification significantly reduces the simulation accuracy.…”