Development of a single-parameter spring-dashpot rolling friction model for coarse-grained DEM

This study proposes a novel single-parameter spring-dashpot rolling friction model based on the critical rolling angle to simplify calibration and enhance the computational efficiency of large-scale coarse-grained DEM simulations for non-spherical particles, as validated by its ability to accurately reproduce macroscopic behaviors in industrial incinerator systems.

The Gist

Imagine you are trying to simulate a massive pile of sand, gravel, or trash in a computer. In the real world, these particles aren't perfect spheres; they are jagged, flat, and irregular. They get stuck, interlock, and resist rolling over each other like a pile of rocks.

However, computers are terrible at calculating the complex math of jagged rocks bumping into each other. It's like trying to solve a Rubik's cube while juggling chainsaws. To make the simulation run fast enough to be useful, scientists usually pretend all the particles are perfect, smooth marbles.

The Problem:
If you use smooth marbles, they roll too easily. A pile of marbles will slide down a hill much faster than a pile of real rocks. To fix this, scientists add a "fake" force called rolling friction. Think of this as a digital brake pad that stops the marbles from rolling too freely, mimicking the resistance of real, jagged rocks.

The Old Way (The "Brake Pedal" Problem):
The old method for applying this brake was like a car with a sticky brake pedal. Once you pressed it, it applied a constant, unchanging force.

• The Issue: If the car (particle) was trying to stop, the brake would sometimes push back too hard or not hard enough, causing the car to vibrate or "jitter" endlessly. It was also very hard to tune; you had to adjust four or five different knobs (parameters) to get the simulation to look right, and changing one knob messed up the others.

The New Solution (The "Smart Shock Absorber"):
This paper introduces a new, smarter way to model that friction. The researchers built a "spring-and-dashpot" model.

• The Analogy: Imagine the particle isn't just hitting a wall; it's sitting on a shock absorber (like on a car).
  • The Spring part pushes back when the particle tries to roll.
  • The Dashpot (damping) part acts like oil in the shock absorber, smoothing out the movement so it doesn't bounce or jitter.
• The Magic: The best part is that this new model only needs one single setting to work: the "Critical Rolling Angle."
  • Think of this as the "tipping point." It's the exact angle of a hill where a rock just barely starts to roll. If you know that angle from a simple real-world experiment, the computer knows exactly how to behave. No more guessing with five different knobs.

Why This Matters (The "Big Picture"):
The researchers didn't just stop at the math. They tested this new model in two ways:

1. The Stability Test: They simulated a pile of particles collapsing. The old model made the particles jitter and vibrate like a nervous dog. The new model made them settle down smoothly, just like real rocks.
2. The Industrial Test: They simulated an incinerator (a trash-burning furnace) with a control plate inside.
   • They compared a simulation with millions of tiny particles (the "Original") against a simulation with fewer, larger "super-particles" (the "Coarse-Grained" model).
   • The Result: The new model allowed the "super-particles" to behave almost exactly like the millions of tiny ones. This means engineers can simulate massive industrial systems (like power plants or chemical reactors) on their computers in a reasonable amount of time, without losing accuracy.

In a Nutshell:
This paper gives engineers a new, simpler, and more stable "remote control" for simulating how non-round particles behave. Instead of wrestling with complex, jittery math, they can now use a single, intuitive setting (the tipping angle) to make their computer simulations run faster, smoother, and more accurately—allowing them to design better industrial machines without needing a supercomputer the size of a house.

Technical Summary

1. Problem Statement

Simulating granular flows composed of non-spherical particles using the Discrete Element Method (DEM) is computationally prohibitive for large-scale industrial applications. While non-spherical shapes (e.g., ellipsoids, polyhedra) are physically accurate, they require complex contact detection and rotational dynamics calculations.

To address this, researchers often use spherical particles with rolling friction to approximate the resistance caused by non-spherical shapes. However, existing rolling friction models suffer from two main limitations:

• Numerical Instability: The widely used Directional Constant Torque (DCT) model applies a constant opposing torque, which can cause spurious rotational oscillations when particles approach a stationary state.
• Parameter Complexity: The standard Spring–Dashpot (S–D) models require multiple empirical parameters (e.g., rolling stiffness, damping, friction coefficients) that must be calibrated in an interdependent manner. This leads to high experimental effort, parameter ambiguity, and uncertainty in practical applications.

2. Methodology

The study proposes a novel framework integrating a new rolling friction model with a Coarse-Grained (CG) DEM approach within a coupled DEM-CFD (Computational Fluid Dynamics) framework called FELMI (Flexible Eulerian–Lagrangian Method with an Implicit algorithm).

A. The Proposed Rolling Friction Model

The authors developed a modified S–D type rolling friction model with the following characteristics:

• Single Parameter: The model relies on a single physically meaningful parameter: the critical rolling angle ($\phi_c$). This represents the slope angle at which a particle begins to roll.
• Theoretical Derivation: The model is derived by decomposing the contact area into infinitesimal spring-dashpot elements. By integrating the stress distribution, the authors established a direct relationship between the rolling resistance moment and the normal contact force.
• Formulation: The rolling resistance moment ($M$) is defined as:
  $$M = -k_r \theta_{rel} - \eta_r \omega_{rel}$$
  where $k_r$ (angular stiffness) and $\eta_r$ (angular damping) are derived from the normal contact parameters and the critical rolling angle. A yield criterion limits the moment to a maximum value ($M_{max}$) to prevent unphysical rotation.
• Stability Mechanism: Unlike the DCT model, the proposed model includes a viscous damping term dependent on angular velocity, ensuring that particles reach a true equilibrium state without persistent oscillation.

B. Coarse-Grained (CG) Scaling

To enable large-scale simulations, the model was integrated into a CG-DEM framework where groups of original particles are represented by a single larger "coarse-grained" particle.

• Scaling Laws: The study derived specific scaling relationships for the rolling resistance parameters to ensure dynamic similarity between the original and coarse-grained systems.
  • Angular stiffness ($k_r$) and damping ($\eta_r$) scale with the cube of the coarse-grain ratio ($l^3$).
  • Relative angular displacement and velocity scale inversely with the ratio ($1/l$).
• Result: The rolling resistance moment in the CG system scales linearly with the original system ($M_{CG} = l \cdot M_{orig}$), preserving the macroscopic behavior.

C. Validation Setup

The model was validated using two scenarios:

1. Stability Test: A quasi-2D collapse of a particle assembly on a flat surface to compare rotational stability against the DCT model.
2. Industrial Application: A DEM-CFD simulation of an incinerator system with a control plate. This involved comparing the original particle system (2.56 million particles) against coarse-grained systems (40k–320k particles) with and without the proposed rolling friction model.

3. Key Contributions

1. Single-Parameter Rolling Model: Development of a robust S–D rolling friction model that requires only the critical rolling angle for calibration, eliminating the need for complex, multi-parameter tuning.
2. Theoretical Consistency: Derivation of angular stiffness and damping coefficients directly from contact mechanics theory, ensuring the model parameters are physically grounded rather than purely empirical.
3. CG-DEM Integration: Successful derivation of scaling laws to integrate the new rolling friction model into coarse-grained DEM, allowing for the simulation of industrial-scale non-spherical particle systems with spherical proxies.
4. Numerical Stability: Demonstration that the proposed model eliminates the spurious rotational oscillations inherent in constant torque models, allowing particles to reach a physically consistent static equilibrium.

4. Results

• Stability Analysis:
  • In the particle collapse test, the proposed model (Case 1-1/1-3) resulted in particles reaching a steady state with near-zero angular velocity.
  • In contrast, the existing DCT model (Case 1-2/1-4) exhibited persistent rotational oscillations and fluctuations in mean angular velocity, even when particle positions appeared static.
• Incinerator Simulation (Macroscopic Behavior):
  • Without Rolling Friction: Coarse-grained models failed to accurately reproduce particle accumulation on the control plate, resulting in an angle of repose significantly lower than the original system.
  • With Proposed Model: The coarse-grained simulations incorporating the new rolling friction model successfully reproduced the macroscopic behavior of the original system.
  • Quantitative Agreement: The simulations showed excellent agreement in:
    • Transient total kinetic energy.
    • Particle fraction above the control plate.
    • Pressure drop profiles.
    • Angle of Repose: The deviation between the coarse-grained model with the new friction and the original system was within the acceptable 10% threshold (unlike the model without rolling friction).

5. Significance

This study provides a practical and computationally efficient solution for simulating industrial granular flows involving non-spherical particles. By reducing the calibration burden to a single, experimentally accessible parameter (critical rolling angle) and ensuring numerical stability, the proposed model makes large-scale, industrial-scale DEM-CFD simulations feasible. The successful validation in an incinerator system demonstrates that this approach can accurately predict complex gas-solid interactions and particle accumulation patterns, offering a reliable tool for equipment design and process optimization in sectors like waste management, pharmaceuticals, and chemical engineering.