What the two have in common
Both build the contact the same way: a spring, which produces the force from the overlap, and a dashpot, which dissipates energy and gives the coefficient of restitution. That holds in the normal and in the tangential direction. The tangential branch also has the friction limit: the force never exceeds μ·Fn, and beyond that it slides.
The difference: how the force grows
The distinction is in the normal spring. In Linear Spring-Dashpot the force is proportional to the overlap, so the stiffness is constant. In Hertz-Mindlin the force grows with the overlap raised to 1.5, which comes from the elastic contact theory between two spheres. In practice, the Hertzian contact is softer on a light touch and stiffens as the load increases.
Hertz-Mindlin
It is the default for most granular materials. The stiffness comes from the material properties (Young's modulus and Poisson's ratio), so the parameters carry direct physical meaning.
- Advantages: it describes the real contact force better, especially under high load, and the packing under self weight. It is the natural choice when you are going to report forces, stresses or wear.
- Limitations: it costs more per contact and the stiffness changes with the load, which demands more care with the time step in heavily compressed regions.
- Use it when: the result that matters is force, wall stress, wear or bed compaction.
Linear Spring-Dashpot
The stiffness is set directly and does not change. It is the simplest and most predictable model.
- Advantages: faster, with a stable time step that is easy to estimate. Because you control the stiffness directly, it is excellent for parameter sweeps and for calibration loops.
- Limitations: constant stiffness does not reproduce the stiffening of the real contact, so force peaks under high load are less faithful.
- Use it when: the result that matters is the flow pattern, the throughput or the process time, and you want to run more cases in less time.
Quick comparison
| Hertz-Mindlin | Linear Spring-Dashpot | |
|---|---|---|
| Normal force | nonlinear (δ1.5) | linear (kn·δ) |
| Stiffness | grows with load | constant |
| Main input | material properties | stiffness set directly |
| Cost per contact | higher | lower |
| Best for | force, stress, wear | flow, calibration, large models |
How to choose in practice
Ask what you are going to report. If the answer is a force number (wall load, torque, wear), go with Hertz-Mindlin. If the answer is how the material moves (does it clog or not, how much comes out per hour, how long it takes), the linear model delivers the same behavior at a lower cost.
Worth noting: in dry granular flow, both calibrated models usually give similar flow patterns, because the flow is dominated by friction and geometry, not by the shape of the spring. The difference shows up when the material gets compressed and the magnitude of the force starts to matter.
Frequently asked questions
Which one should I use if I'm starting out?
Start with Hertz-Mindlin, the default for most granular materials, using parameters with physical meaning. Switch to the linear model if you need more speed in studies and calibration.
Do the two give the same result?
In well-calibrated dry flow, the flow pattern is usually similar. The contact force values are not. If you report force, stress or wear, the choice of model changes the number.
Does the contact model change friction and restitution?
No. Restitution, sliding friction and rolling friction remain parameters of the material pair in both models. What changes is how the normal spring computes the force.
Do I need to recalibrate when switching models?
Yes. Calibrated parameters are valid for the model they were tuned in. Rerun the calibration tests in the new model before trusting the result.
Try both models
BrixDEM ships Hertz-Mindlin and Linear Spring-Dashpot in the same solver. Switch the model and compare on your own case.