Calibration of Discrete Element Method (DEM) models is an iterative process of adjusting input parameters such that the macroscopic results of simulations and experiments are similar. Therefore, selecting appropriate input parameters of a model effectively is crucial for the efficient use of the method. Despite the growing popularity of DEM, there is still an ongoing need for an efficient method for identifying contact law parameters. Commonly used trial and error procedures are very time-consuming and unpractical, especially in the case of models with many parameters to calibrate. It seems that machine learning may offer a new approach to that problem. This research aims to apply supervised machine learning to figure out the dependencies between specific microscopic and macroscopic parameters. More than 6000 DEM simulations of uniaxial compression tests gathered the data for two algorithms - Multiple Linear Regression and Random Forest. Promising results with an accuracy of over 99% give good hope for finding a universal relation between input and output parameters (for a specific DEM implementation) and reducing the number of simulations required for the calibration procedure. Another pertinent question concerns the size of the DEM models used during calibration based on the uniaxial compression test. It has been proven that calibration of certain parameters can be done on smaller samples, where the critical threshold is around 30% of the radius of the original model.
Citation: Piotr Klejment. Application of supervised machine learning as a method for identifying DEM contact law parameters[J]. Mathematical Biosciences and Engineering, 2021, 18(6): 7490-7505. doi: 10.3934/mbe.2021370
Calibration of Discrete Element Method (DEM) models is an iterative process of adjusting input parameters such that the macroscopic results of simulations and experiments are similar. Therefore, selecting appropriate input parameters of a model effectively is crucial for the efficient use of the method. Despite the growing popularity of DEM, there is still an ongoing need for an efficient method for identifying contact law parameters. Commonly used trial and error procedures are very time-consuming and unpractical, especially in the case of models with many parameters to calibrate. It seems that machine learning may offer a new approach to that problem. This research aims to apply supervised machine learning to figure out the dependencies between specific microscopic and macroscopic parameters. More than 6000 DEM simulations of uniaxial compression tests gathered the data for two algorithms - Multiple Linear Regression and Random Forest. Promising results with an accuracy of over 99% give good hope for finding a universal relation between input and output parameters (for a specific DEM implementation) and reducing the number of simulations required for the calibration procedure. Another pertinent question concerns the size of the DEM models used during calibration based on the uniaxial compression test. It has been proven that calibration of certain parameters can be done on smaller samples, where the critical threshold is around 30% of the radius of the original model.
| [1] |
H. Q. Do, A. M. Aragon, D. L. Schott, Automated discrete element method calibration using genetic and optimization algorithms, EPJ Web of Conferences, 140 (2017), 15011. doi: 10.1051/epjconf/201714015011
|
| [2] | L. Benvenuti, C. H. Kloss, S. Pirker, Characterization of DEM particles by means of artificial neural networks and macroscopic experiments, 16th EANN workshops, 2015. |
| [3] |
E. Onate, J. Rojek, Combination of discrete element and finite element methods for dynamic analysis of geomechanics problems, Computer Meth. Appl. Mech. Eng., 193 (2004), 3087-3128. doi: 10.1016/j.cma.2003.12.056
|
| [4] |
B. Yang, Y. Jiao, S. Lei, A study on the effects of microparameters on macroproperties for specimens created by bonded particles, Eng. Comput., 23 (2006), 607-631. doi: 10.1108/02644400610680333
|
| [5] |
A. Fakhimi, T. Villegas, Application of dimensional analysis in calibration of a discrete element model for rock deformation and fracture, Rock. Mech. Rock. Eng., 40 (2007), 193-211. doi: 10.1007/s00603-006-0095-6
|
| [6] |
Y. Wang, F. Tonon, Calibration of a discrete element model for intact rock up to its peak strength, Int. J. Numer. Anal. Met., 34 (2010), 447-469. doi: 10.1002/nag.811
|
| [7] |
H. Q. Do, A. M. Aragón, D. L. Schott, A calibration framework for discrete element model parameters using genetic algorithms, Adv. Powder. Technol., 29 (2018), 1393-1403. doi: 10.1016/j.apt.2018.03.001
|
| [8] | M. Wang, P. Cao, Calibrating the micromechanical parameters of the PFC2D (3D) models using the improved simulated annealing algorithm, Math. Probl. Eng., 1 (2017), 1-11. |
| [9] |
L. Benvenuti, C. Kloss, S. Pirker, Identification of DEM simulation parameters by artificial neural networks and bulk experiments, Powder Technol., 291 (2016), 456-465. doi: 10.1016/j.powtec.2016.01.003
|
| [10] |
M. J. Sun, H. M. Tang, X. L. Hu, Y. F. Ge, S. Lu, Microparameter prediction for a triaxial compression PFC3D model of rock using full factorial designs and artificial neural networks, Geotech. Geol. Eng., 31 (2013), 1249-1259. doi: 10.1007/s10706-013-9647-1
|
| [11] | A. S. Tawadrous, D. De Gagné, M. Pierce, D. Mas Ivars, Prediction of uniaxial compression PFC3D model micro-properties using artificial neural networks, Int. J. Numer. Anal. Met., 33 (2009), 1953-1962. |
| [12] |
M. Chunhui, J. Yang, G. Zenz, E. J. Staudacher, L. Cheng, Calibration of the microparameters of the discrete element method using a relevance vector machine and its application to rockfill materials, Adv. Eng. Software, 147 (2020), 102833. doi: 10.1016/j.advengsoft.2020.102833
|
| [13] | C. O'Sullivan, J. D. Bray, Selecting a suitable time step for discrete element simulations that use the central difference time integration scheme, Eng. Comput., 21 (2004), 2-4,278-303. |
| [14] | S. Abe, V. Boros, W. Hancock, D. Weatherley, ESyS-particle tutorial and user's guide, Version 2.3.1, https://launchpad.net/esys-particle, 2014. |
| [15] | D. K. Weatherley, V. E. Boros, W. R. Hancock, S. Abe, Scaling benchmark of ESyS-Particle for elastic wave propagation simulations, 2010 IEEE Sixth Int. Conf. e-Science, Brisbane, Australia, 277-283, 2010. |
| [16] | Y. C. Wang, S. Xue, J. Xie, Discrete element method and its applications in earthquake and rock fracture modelling, In Y.-G. Li (ed.) Imaging, Modeling and Assimilation in Seismology, De Gruyter, 235-262, 2012 |
| [17] | P. A. Cundall, A computer model for simulating progressive, large-scale movement in blocky rock systems, Proc. Symp. Int. Soc. Rock Mech., 2 (1971), 8. |
| [18] | P. A. Cundall, A computer model for rock-mass behavior using interactive graphics for the input and output of geometrical data, National Technical Information Service, Report no. AD/A-001 602, 1974. |
| [19] |
P. A. Cundall, O. D. L. Strack, A discrete numerical model for granular assemblies, Geotechnique, 29 (1979), 47-65. doi: 10.1680/geot.1979.29.1.47
|
| [20] |
D. O. Potyondy, P. A. Cundall, A bonded-particle model for rock, Int. J. Rock Mech. Min. Sci., 41 (2004), 1329-1364. doi: 10.1016/j.ijrmms.2004.09.011
|
| [21] | Y. Wang, S. Abe, S. Latham, P. Mora, Implementation of particle-scale rotation in the 3-D lattice solid model, Pure Appl. Geophys., 163 (2000), 1769-1785. |
| [22] |
Y. C. Wang, A new algorithm to model the dynamics of 3-D bonded rigid bodies with rotations, Acta Geotech., 4 (2009), 117-127. doi: 10.1007/s11440-008-0072-1
|
| [23] | T. L. Anderson, Fracture Mechanics: Fundamentals and Applications, CRC Press, Boca Raton, 1991. |
| [24] | R. M. Christensen, The theory of materials failure, Oxford University Press, Oxford, 2013. |
| [25] | A. Geron, Hands-On Machine Learning with Scikit-Learn and TensorFlow, O'Reilly Media, Inc, USA, 2017. |
| [26] | S. Makridakis, S. C. Wheelwright, R. J. Hyndman, Forecasting methods and applications, John Wiley & Sons, Hoboken, NJ, USA, 2008. |
| [27] | L. Breiman, Random forests, Mach. Learn., 45 (2001), 5-32. |
| [28] |
P. Mora, D. Place, Stress correlation function evolution in lattice solid elasto-dynamic models of shear and fracture zones and earthquake prediction, Pure Appl. Geophys., 159 (2002), 2413-2427. doi: 10.1007/s00024-002-8741-8
|
| [29] | P. Klejment, The microscopic insight into fracturing of brittle materials with the discrete element method, Publications of the Institute of Geophysics Polish Academy of Sciences—Geophysical Data Bases, Processing and Instrumentation, 427 (A-31), 2020. |
| [30] | Y.G. Li, ed., Imaging, Modeling and Assimilation in Seismology, De Gruyter, Berlin, 2012. |
Figures(9) / Tables(1)
Piotr Klejment. Application of supervised machine learning as a method for identifying DEM contact law parameters[J]. Mathematical Biosciences and Engineering, 2021, 18(6): 7490-7505. doi: 10.3934/mbe.2021370
DownLoad: