Presets
Playback
Parameters
Camera
Layers
Quality
Model check
The science
What is modelled
Amber (fossil resin, about 1085 kg/m3) lies in sandy layers (quartz, 2650 kg/m3). In hydraulic mining a high-pressure water monitor washes the layer into a slurry, which is then sorted. The sorting step shown here is a vertical up-flow column (an elutriator): grains enter at mid height, an upward current of liquid moves through the column, and every grain either leaves with the current through the top launder or settles out through the cone at the bottom. Which way it goes depends only on its settling speed compared with the up-flow speed. The page follows 700 random grains with that rule and compares the tally with the exact size-distribution formula.
Equations
- Settling speed (Ferguson and Church 2004, an empirical fit between Stokes and Newton regimes): w = R g D2 / (18 ν + √(0.75 C2 R g D3)) with R = |ρp/ρl - 1|; the sign says whether the grain sinks (ρp > ρl) or rises. C2 = 1.0 for natural grains, 0.4 for smooth spheres.
- A grain reports to the overflow if U - w > 0. For sinking grains this means D < D*, the cut size found from w(D*) = U. With a log-normal size distribution the overflow share of the number is Φ(ln(D*/d50)/ln σg); the share of the mass uses the shifted median d50 σg3 ln σg.
- Brine viscosity grows from 1.0 to about 1.9 mPa s between fresh water and 1190 kg/m3 (straight-line interpolation of tabulated values).
- Monitor: v = Cv√(2ΔP/ρ), Q = A v, jet power ½ ρ Q v2.
What it teaches
In fresh water amber (density contrast about 8 %) settles at a speed that overlaps the speed of fine sand, so no up-flow separates them cleanly. Make the liquid denser than amber (brine, about 1100 to 1200 kg/m3) and amber floats whatever its size, while sand still sinks - then a gentle up-flow only has to hold the sand back.
Simplifications
- Hindered settling (crowding) is ignored; the column is dilute. Grains do not touch or influence the flow.
- The velocity profile in the column is uniform, plus a small random walk (0.5 cm2/s) for turbulence. A real column has a velocity profile and wall effects.
- The up-flow is a free setting; the pump in the model is only a picture of where it comes from. Pit digging, pipe losses and pump curves are not simulated; the monitor readout is Bernoulli only.
- The size distributions are illustrative log-normals, not field assays. The settling formula is a fit and is only good to roughly 10 to 25 percent.