The discrete element method (DEM) is largely used to simulate the geotechnical behavior of granular materials. However, numerical modeling with this type of code is expensive and time consuming, especially when fine particles are involved. This leads researchers to make use of different approaches to shorten the time of calculation without verifying the stability and reliability of numerical results, even though a compromise between the time of calculation and accuracy is commonly claimed. The particle size distribution (PSD) curve of studied granular material is completely ignored or arbitrarily cut. It is unclear if the ensued numerical results are still representative of the studied granular materials. Additionally, one can see a large number of numerical models established on a basis of calibration by ignoring the physical meaning and even measured values of some model parameters. The representativeness and reliability of the obtained numerical results are questionable. All these partly contribute to reducing the public’s confidence in numerical modeling. In this study, a methodology is illustrated to obtain an optimal DEM model, which minimizes the time of calculation and ensures stable and reliable numerical results for the mechanical behavior of a waste rock. The results indicate that the PSD curve of the studied waste rock can indeed be cut by excluding a portion of fine particles, while the Young’s modulus of the waste rock particles can also be decreased to accelerate the numerical calculations. A physical explanation of why the time of calculation can be shortened by reducing the Young’s modulus of waste rock particles is provided for the first time. Overall, the PSD cut, reduction in Young’s modulus, and time step must be determined through sensitivity analyses to ensure stable and reliable results with the shortest time of calculation. In addition, it is important to minimize the number of model parameters determined through the process of calibration, especially for those having physical meanings. In this study, the only model parameter having a clear physical meaning but difficult to measure is the rolling resistance coefficient for repose angle tests on the studied waste rock. Its value has to be obtained through a process of calibration against some experimental results. The validity and predictability of the calibrated numerical model have been successfully verified against additional experimental results.

Accurate calibration of mesoscopic contact model parameters is essential for ensuring the reliability of Particle Flow Code in Three Dimensions (PFC3D) simulations in geotechnical engineering. Trial-and-error approaches are often used to determine the parameters of the contact model, but they are time-consuming, labor-intensive, and offer no guarantee of parameter validity or simulation credibility. Although conventional machine learning techniques have been applied to invert the contact model parameters, they are hampered by the difficulty of selecting the optimal hyperparameters and, in some cases, insufficient data, which limits both the predictive accuracy and robustness. In this study, a total of 361 PFC3D uniaxial compression simulations using a linear parallel bond model with varied mesoscopic parameters were generated to capture a wide range of rock and geotechnical material behaviors. From each stress–strain curve, eight characteristic points were extracted as inputs to a multi-objective Automated Machine Learning (AutoML) model designed to invert three key mesoscopic parameters, i.e., the elastic modulus (E), stiffness ratio (k_s/k_n), and degraded elastic modulus (E_d). The developed AutoML model, comprising two hidden layers of 256 and 32 neurons with ReLU activation function, achieved coefficients of determination (R²) of 0.992, 0.710, and 0.521 for E, k_s/k_n, and E_d, respectively, demonstrating acceptable predictive accuracy and generalizability. The multi-objective AutoML model was also applied to invert the parameters from three independent uniaxial compression tests on rock-like materials to validate its practical performance. The close match between the experimental and numerically simulated stress–strain curves confirmed the model’s reliability for mesoscopic parameter inversion in PFC3D.