A Physics-Informed Neural Network Integration Framework for Efficient Dynamic Fracture Simulation in an Explicit Algorithm
Source abstract
Abstract
The conventional dynamic fracture simulation by using the explicit algorithm often involves a large number of iteration computation due to the extremely small time interval. Thus, the most time-consuming process is the integration of constitutive relation. To improve the efficiency of the dynamic fracture simulation, a physics-informed neural network integration (PINNI) model is developed to calculate the integration of constitutive relation. PINNI employs a shallow multilayer perceptron with integrable activations to approximate constitutive integrand. To train PINNI, a large number of strains in a reasonable range are generated at first, and then the corresponding stresses are calculated by the mechanical constitutive relation. With the generated strains as input data and the calculated stresses as output data, the PINNI can be trained to reach a very high precision, whose relative error is about 7.8×10−5%. Next, the mechanical integration of constitutive relation is replaced by the well-trained PINNI to perform the dynamic fracture simulation. It is found that the simulation results by the mechanical and PINNI approach are almost the same. This suggests that it is feasible to use PINNI to replace the rigorous mechanical integration of constitutive relation. The computational efficiency is significantly enhanced, especially for the complicated constitutive relation. It provides a new AI-combined approach to dynamic fracture simulation.
Source: crossref · Quality: baseline
Research context
OpenAlex topics: Hydraulic Fracturing and Reservoir Analysis · Drilling and Well Engineering · Rock Mechanics and Modeling
SciVX discovery tags are terminology-based. OpenAlex topics are source-provided classifications.
Affiliations
College of Information and Electrical Engineering, China Agricultural University, Beijing 100083, China
School of Ocean and Civil Engineering, Shanghai Jiao Tong University, Shanghai 200240, China
References (26)
Show source-provided references
1. On some topics for the numerical simulation of ductile fracture10.1016/j.ijplas.2007.08.004 ↗
2. Continuum models of ductile fracture: A review10.1177/1056789509103482 ↗
3. Using neural networks for fast numerical integration and optimization10.1109/ACCESS.2020.2991966 ↗
4. An artificial intelligence-based model for performance prediction of acid fracturing in naturally fractured reservoirs10.1021/acsomega.1c00809 ↗
5. Combined implicit/explicit algorithms for crashworthiness analysis10.1016/j.ijimpeng.2004.03.004 ↗
6. Computationally optimized formulation for the simulation of composite materials and delamination failures10.1016/j.compositesb.2010.09.013 ↗
7. Efficient and robust constitutive integrators for single-crystal plasticity modeling10.1016/j.ijplas.2006.02.008 ↗
8. Efficiency analysis of numerical integrations for finite element substructure in real-time hybrid simulation10.1007/s11803-018-0426-0 ↗
9. Multiscale graph neural networks with adaptive mesh refinement for accelerating mesh-based simulations10.1016/j.cma.2024.117152 ↗
10. Aldakheel, F., Satari, R., and Wriggers, P. (2021). Feed-forward neural networks for failure mechanics problems. Appl. Sci., 11.10.3390/app11146483 ↗
11. A neural network-based enrichment of reproducing kernel approximation for modeling brittle fracture10.1016/j.cma.2023.116590 ↗
12. Transfer learning enhanced physics informed neural network for phase-field modeling of fracture10.1016/j.tafmec.2019.102447 ↗
13. Neural network-based surrogate model for a bifurcating structural fracture response10.1016/j.engfracmech.2020.107424 ↗
14. WIFF1.0: A hybrid machine-learning-based parameterization of wave-induced sea ice floe fracture10.5194/gmd-15-803-2022 ↗
15. Deep learning applied to computational mechanics: A comprehensive review, state of the art, and the classics
16. Deep learning in computational mechanics: A review10.1007/s00466-023-02434-4 ↗
17. Embedding artificial neural networks into twin cohesive zone models for composites fatigue delamination prediction under various stress ratios and mode mixities10.1016/j.ijsolstr.2021.111311 ↗
18. Using artificial neural network to predict the fracture properties of the interfacial transition zone of concrete at the meso-scale10.1016/j.engfracmech.2020.107488 ↗
19. A shallow physics-informed neural network for solving partial differential equations on static and evolving surfaces10.1016/j.cma.2023.116486 ↗
20. Hidden fluid mechanics: Learning velocity and pressure fields from flow visualizations10.1126/science.aaw4741 ↗
21. Numerical and experimental crack-tip cohesive zone laws with physics-informed neural networks10.1016/j.jmps.2024.105866 ↗
22. Koeppe, A., Bamer, F., Selzer, M., Nestler, B., and Markert, B. (2022). Explainable artificial intelligence for mechanics: Physics-explaining neural networks for constitutive models. Front. Mater., 8.10.3389/fmats.2021.824958 ↗
23. Krizhevsky, A. (2014). One weird trick for parallelizing convolutional neural networks. arXiv.
24. Simulating fracture propagation in rock and concrete by an augmented virtual internal bond method10.1002/nag.1015 ↗
25. Numerical simulation of crack growth in an isotropic solid with randomized internal cohesive bond10.1016/S0022-5096(97)00047-1 ↗
26. Numerical simulations of fast crack growth in brittle solids10.1016/0022-5096(94)90003-5 ↗