Adaptive Material Point Method for Solid Simulation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The high computational cost of traditional Material Point Methods (MPM) limits their use in simulating deformable solid materials with memory, as they require accounting for additional parameters and forces compared to memory-free materials like liquids.
Innovation Solution
An adaptive MPM approach that focuses computational resources by splitting and merging particles based on grid resolution, using different transfer functions for data transfer, and dynamically adjusting grid and particle resolution to improve efficiency, particularly around material surfaces where visual detail is concentrated.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If traditional Material Point Method is used to simulate solids with memory, then visual fidelity is improved, but computational cost increases significantly
Solution Approach 1:
The patent applies different levels of grid resolution to different spatial regions based on their importance. Fine grid resolution is applied only to regions containing material or regions of interest, while coarse grid resolution is applied to empty or less important regions. This local differentiation maintains visual fidelity in critical areas while reducing computational cost in less critical areas.
Solution Approach 2:
The simulation domain is segmented into multiple grid levels with different resolutions. The system dynamically manages multiple grid hierarchies, transferring data between different resolution levels. This segmentation allows the simulation to focus computational resources on specific regions rather than uniformly processing the entire domain, thereby reducing overall computational cost while maintaining necessary visual fidelity.
2Measurement precision
If high grid resolution is applied throughout the entire domain, then simulation accuracy is improved, but computational resources are wasted in empty regions
Solution Approach 1:
The patent implements adaptive grid refinement where the grid resolution is locally adjusted based on the presence and importance of material. Regions containing deformable solids or regions of visual interest receive fine grid resolution for accurate simulation, while empty or less important regions use coarse grid resolution. This local quality differentiation eliminates computational waste in empty regions while maintaining simulation accuracy where needed.
Solution Approach 2:
The grid resolution is dynamically adjusted during the simulation based on the material's position and deformation state. The system can refine or coarsen grids adaptively as materials move and deform, ensuring that high computational resources are allocated to regions that currently require accuracy rather than statically allocating resources throughout the entire domain.
3Ease of manufacture
If uniform grid resolution is used, then implementation is simpler, but computational resources are not optimized for regions of visual interest
Solution Approach 1:
The patent segments the computational domain into multiple grid levels with different resolutions, organized in a hierarchical structure. This segmentation allows the system to manage complexity through organized layers of detail rather than attempting uniform processing, achieving both implementation feasibility and computational optimization through structured multi-resolution management.
Data Source
AI summary
An adaptive material point method for simulating and rendering of solid materials is disclosed. During each time step, a simulation application splits and/or merges particles according to a resolution of a grid and assigns properties to the split/merged particles. The simulation application then rasterizes particle information, including masses and velocities, to the grid, on which forces and/or collisions are computed to obtain updated velocities. Grid information, including the updated velocities, is then transferred back to the particles. The transfer from particles to grid and back from grid to particles may employ different transfer functions. In particular, a quadratic transfer function may be used to transfer information from the particles to the grid, while a linear transfer function may be used elsewhere, including to transfer information back from the grid to the particles. After updated velocities are transferred to the particles, the simulation application advects the particles using the updated velocities.