TL;DR
This paper tackles the challenge of accurately modeling seashell behavior in discrete element simulations, crucial for optimizing crushing equipment. By combining rigorous physical experiments with a multi-stage calibration process for a bonded-particle model, they achieved a high-fidelity simulation with only a 3.8% error compared to real-world tests.
What Problem Does This Paper Address?
The efficient processing of seashell waste, particularly from aquaculture, is a growing concern. While crushing these shells offers potential for reuse in various industries, their brittle and complex nature makes them difficult to break down uniformly. Discrete Element Method (DEM) is a powerful tool for simulating granular materials, but accurately modeling the mechanical behavior and fracture of shells requires precise parameter calibration. Existing DEM models often struggle with brittle, bonded materials like shells, and there's a lack of established methodologies for their parameterization. This paper aims to fill that gap by developing a reliable DEM model for freshwater mussel shells, enabling better design and optimization of crushing equipment.
Methods
The authors employed a multi-faceted approach:
- Intrinsic Property Determination: Physical experiments were conducted to measure fundamental properties of mussel shells, including density, Poisson's ratio, shear modulus, and elastic modulus.
- Contact Parameter Calibration: Experiments like the inclined plane method and impact tests were used to determine contact parameters (static friction, rolling friction, coefficient of restitution) between shells and between shells and stainless steel.
- Model Selection: The Hertz-Mindlin with Bonding (HMB) model, specifically the BondingV2 variant, was chosen for its ability to represent bonded and brittle materials.
- Multi-Stage Parameter Calibration:
- Plackett-Burman (P-B) Design: This statistical screening method was used to identify the most influential bonding parameters (normal stiffness, tangential stiffness, critical normal stress, critical tangential stress, bonding radius) affecting the ultimate load and displacement in a three-point bending test.
- Steepest Ascent Test: This method was used to refine the parameter ranges identified by the P-B design and move towards the optimal region.
- Response Surface Methodology (RSM) / Box-Behnken Design: This was employed to build a second-order mathematical model and find the optimal combination of the most significant parameters.
- Experimental Validation: A three-point bending test was used as the primary target for calibration, and the final model's accuracy was verified against experimental results, comparing load-displacement curves, fracture behavior, and ultimate load.
Datasets
The primary "dataset" used in this study is derived from experimental measurements on freshwater mussel shells collected from Lianyungang, China. These include:
- Intrinsic Material Properties: Density, Poisson's ratio, shear modulus, elastic modulus.
- Contact Parameters: Static friction coefficient, rolling friction coefficient, coefficient of restitution (for shell-shell and shell-stainless steel interfaces).
- Mechanical Response Data: Load-displacement curves and ultimate load/displacement from three-point bending tests.
- Validation Data: Angle of repose measurements for contact parameter verification.
No pre-existing large-scale datasets were used for training or evaluation; instead, the authors generated their own experimental data to calibrate and validate their DEM model.
Key Insights and Ideas
- Shells are Tricky: The inherent brittleness and complex layered structure of shells make them challenging to model accurately with standard DEM approaches.
- Bonding is Key: The HMB model, with its explicit representation of bonds between particles, is essential for capturing the fracture behavior of shells, unlike simpler contact models.
- Systematic Calibration Pays Off: A hierarchical approach combining statistical screening (P-B), directional search (steepest ascent), and optimization (RSM) is highly effective for identifying and fine-tuning critical DEM parameters.
- Main Effects Dominate (in this range): For quasi-static bending within the tested parameter ranges, the main effects of stiffness parameters were found to be far more significant than interaction effects, simplifying the calibration process.
- Scale Matters: The calibrated parameters are most reliable for simulations using particle sizes similar to those used in calibration (around 1mm). Extrapolating to significantly larger or smaller particles will likely require re-calibration or scaling.
Key Results
- Intrinsic Properties Determined: Density (2200 kg/m³), Poisson's ratio (0.26), shear modulus (1.57 × 10⁸ Pa), and elastic modulus (6.5 × 10¹⁰ Pa) were experimentally measured.
- Contact Parameters Calibrated: Static friction, rolling friction, and restitution coefficients were determined for shell-shell and shell-stainless steel interactions.
- Angle of Repose Validation: The calibrated contact parameters showed good agreement with experimental angle of repose tests, with a relative error of 5.1%.
- Dominant Parameters Identified: Plackett-Burman screening identified normal stiffness per unit area and tangential stiffness per unit area as the most influential parameters for bending failure.
- Optimized Bonding Parameters: Through RSM, an optimal set of bonding parameters was found.
- High Model Accuracy: The final calibrated DEM model reproduced the experimental bending response with a relative error of only 3.8% in ultimate load.
- Realistic Fracture Simulation: The model accurately captured the brittle fracture behavior, load drop, and crack propagation observed in physical tests.
Limitations and Future Work
- Quasi-Static Assumption: The current model and calibration are validated for quasi-static bending. Dynamic crushing and high-speed impact scenarios, where strain rate sensitivity and damping become important, will require further investigation and potential recalibration.
- Spherical Particle Simplification: The use of uniform spherical particles neglects the complex layered architecture and anisotropy of natural shells. While sufficient for global behavior, it may limit the prediction of anisotropy-driven fracture patterns. Future work could explore non-spherical particles or layered bonding strategies.
- Scale Dependence: The calibrated parameters are scale-specific to the particle size used (1mm). For simulations involving significantly different particle sizes (larger fragments or fine powder), re-calibration or scaling of parameters will be necessary.
- Future Directions: The authors suggest extending the framework to dynamic scenarios, investigating scale effects, and incorporating more detailed microstructural features. They also plan to develop scaling relationships and multi-scale calibration strategies to broaden the model's applicability.
Summary updated