A fourth state of matter
Pour sand into a pile and it behaves like a solid, supporting a footprint without deforming. Tip the container and it flows like a liquid through an hourglass neck. Shake it hard enough and it churns like a gas. None of the three descriptions is right on its own, because dry sand grains obey rules that classical states of matter do not: they interact only at contact, through friction and normal force, with no cohesion pulling them together. Every collision between grains dissipates energy — granular media are inherently dissipative and have no equilibrium state without continuous energy input. And because grains are macroscopic, thermal fluctuations (kT) are utterly negligible next to a grain's gravitational potential energy, so the statistical-mechanics toolkit built for atoms does not transfer directly. Below a critical packing density the material flows; above it, force chains percolate through the pile and it locks rigid — the jamming transition.
The angle of repose
Pour sand onto a flat surface and a cone forms whose slope stabilises at the angle of repose θᵣ — set almost entirely by interparticle friction:
tan θᵣ = μ_s // μ_s = coefficient of static friction dry quartz sand: θᵣ ≈ 30-35° (μ_s ≈ 0.58-0.70) glass beads: θᵣ ≈ 22° (smooth, low friction) wet sand: θᵣ ≈ 45°+ (capillary bridges add cohesion)
There is a subtlety worth knowing: a pile can hold briefly at a steeper maximum angle of stability θₘ > θᵣ until disturbed, then avalanches back down to θᵣ. That bistability — stable-until-triggered, then a sudden slip — is the microscopic root of every avalanche discussed below.
Force chains and the Janssen effect
Stress inside a granular pile is not distributed evenly the way it is in a fluid — it concentrates along force chains, quasi-linear paths of grains carrying three to ten times the average load, typically spanning 5-20 grain diameters and constantly breaking and re-forming as grains shift. Photoelastic-disc experiments make them visible as bright lines under polarised light. Force chains are the microscopic mechanism behind the Janssen effect: in a tall silo, grains arch against the walls and redirect load sideways through friction, so the pressure at the bottom saturates with height instead of growing linearly the way hydrostatic fluid pressure would. Silo and hopper engineers have to design around that non-uniform, wall-mediated stress rather than treating stored grain like a liquid.
The cellular automaton behind falling sand
The classic "falling sand" simulation sidesteps continuum mechanics entirely and represents the world as a 2D pixel grid, applying a simple rule to every sand cell each tick:
for each sand cell, in priority order: 1. if cell below is empty → move down 2. else if below-left is empty → move down-left 3. else if below-right is empty → move down-right (coin-flip if both open) 4. else → stay (part of the pile)
The random left/right tie-break at step 3 is essential — without it the grid's axes bias every pile into an unnaturally symmetric shape. This simple rule set is enough to produce believable pile formation, flowing streams and arching over small gaps, though the angle it settles at (~45° on a plain 4-directional grid) is a lattice artefact rather than physically accurate; giving grains 8-directional movement options brings the automaton's angle of repose down toward the 30-35° real sand exhibits.
Avalanches and self-organised criticality
In 1987 Per Bak, Chao Tang and Kurt Wiesenfeld showed that a slowly driven sandpile self-organises, with no external tuning, into a critical state where avalanche sizes follow a power law rather than clustering around any typical size — self-organised criticality (SOC):
BTW sandpile model:
drop a grain: z(x,y) += 1
while any z(x,y) >= 4:
z(x,y) -= 4
z(x±1,y) += 1; z(x,y±1) += 1 // topple to 4 neighbours
P(s) ~ s^-τ // avalanche-size distribution, τ ≈ 1.0-1.5
SOC's power-law signature — no characteristic scale, so a single grain can trigger an avalanche of almost any size — turns up far outside sandpiles: earthquake magnitudes (the Gutenberg-Richter law), forest fire sizes, species extinction events and even financial market crashes all show the same statistical fingerprint of a system that has tuned itself, without anyone tuning it, to the edge of instability.
Frequently asked questions
What sets the angle of repose of a sand pile?
The angle of repose θᵣ is set almost entirely by interparticle friction: tan θᵣ = μₛ, the coefficient of static friction between grains. Dry quartz sand sits around 30–35°, smooth glass beads as low as 22°, while wet sand or cornstarch can exceed 45–50° because capillary bridges or irregular grain shapes add extra resistance to sliding. A pile can actually stand briefly at a steeper "maximum angle of stability" before avalanching back down to θᵣ.
Why does pressure at the bottom of a grain silo not increase linearly with depth?
Unlike a fluid, where hydrostatic pressure grows linearly with depth, granular stress concentrates into force chains — arching paths of grains that redirect load sideways onto the container walls instead of straight down. This is the Janssen effect: friction between the grains and the silo wall means the pressure at the bottom saturates rather than growing indefinitely with height, which is why silo and hopper design cannot use simple fluid-pressure formulas.
What is self-organised criticality and how does the sandpile model show it?
Self-organised criticality (SOC) is when a slowly driven system settles, without any tuning, into a critical state where events of all sizes occur with no characteristic scale. Bak, Tang and Wiesenfeld's 1987 sandpile model shows this directly: grains are added one at a time, cells above a threshold topple onto their neighbours, and the resulting avalanche sizes follow a power law P(s) ~ s⁻ᵗ. The same statistical signature shows up in earthquake magnitudes, forest fire sizes and other naturally "self-tuning" systems.
Try it live
Open Granular Materials to drop sand grains, watch a pile build up to its angle of repose, and switch to the Bak-Tang-Wiesenfeld cellular automaton to watch power-law avalanches unfold in real time.
▶ Open Granular Materials simulation