loading...
running...
How to use
Pick a preset or move the sliders: grain size, water depth, slope and temperature set the flow; feed and bar height set the sediment supply. The side view shows the water, a sand bed with its initial profile dashed, moving grains and the logarithmic velocity profile. The header line says which transport regime you are in and how far the Shields number is above its threshold. The four graphs are the bed change along the flume, the Shields diagram, a Hjulström-type velocity-versus-grain-size chart and the Rouse concentration profile.
Try: (1) Gravel at rest then raise the slope until grains start to move; (2) Clear-water scour and watch the inlet hole; (3) Overfed to build a sediment wave; (4) halve the depth while keeping the slope and see the shear stress halve; (5) cool the water to 3 C and see the threshold move. Space pauses, R resets.
The science
A flume is a laboratory channel in which water flows over a bed of loose grains. Whether the grains stay put, roll along the bed or are lifted into the flow decides how rivers carve valleys, how sandbars grow and how a bridge pier gets undermined. This page reduces the problem to a few classical relations and a one-dimensional bed.
Flow. Uniform flow of depth h on slope S gives a bed shear stress and shear velocity
τ = ρ g h S, u* = √(g h S), U = (u*/κ) [ln(11 h / ks) − 1], ks = 2.5 d, κ = 0.41
The mean velocity comes from the rough-wall logarithmic law; the profile u(z) = (u*/κ) ln(z/z0) is drawn in the 2D lab and traced by the flow tracers in 3D. The Froude number Fr = U/√(g h) tells you whether the flow is tranquil (below 1) or rapid.
Threshold of motion. The Shields number compares the drag on a grain with its submerged weight:
θ = τ / [(ρs − ρ) g d], D* = d [(s−1) g / ν²]1/3, θcr = 0.30/(1 + 1.2 D*) + 0.055 [1 − exp(−0.020 D*)]
The last fit is the Soulsby and Whitehouse (1997) curve. For coarse gravel θcr tends to about 0.055; it rises for fine silt because of the viscous sublayer and has a minimum near D* of 10 (about 0.5 mm sand). The viscosity ν depends on temperature (colder water is thicker), so the threshold shifts with the water temperature slider. The Hjulström-type graph plots the mean velocity needed to reach θcr at your depth against grain size; the dot is your flow.
Bedload. Above threshold the Meyer-Peter and Müller relation, written with the variable critical Shields number, gives the volume of solids moved per unit width and time:
qs = 8 (θ − θcr)3/2 √[(s−1) g d³], s = ρs/ρ = 2.65
Suspension. The settling velocity follows Ferguson and Church (2004), ws = (s−1) g d² / (18 ν + √(0.75 (s−1) g d³)). The ratio u*/ws picks the transport mode on this page: below 0.4 grains roll and slide, between 0.4 and 1 they hop (saltation), above 1 turbulence keeps them up. The Rouse number Z = ws/(κ u*) shapes the concentration profile C(z)/C(a) = [ (h−z)/z · a/(h−a) ]Z with reference height a = 0.05 h. The thresholds 0.4 and 1 are textbook rules of thumb, not sharp physical boundaries.
The bed (Exner equation). The bed is a 60-cell strip, 12 m long. The solids balance says the bed rises where transport decreases downstream:
(1 − p) ∂z/∂t = − ∂qs/∂x, p = 0.4, Slocal = S − λ ∂z/∂x
Sediment enters at the inlet at the feed ratio times the capacity of the undisturbed flow. A feed of zero is clear-water scour (the inlet erodes, then the bed degrades downstream); a feed above 1 aggrades; a feed of 1 over a flat bed is equilibrium. The slope term lets a bar relax: a crest steepens the local slope and sheds sediment, which makes the bar spread like a diffusing hump with D = λ (dqs/dS)/(1−p). The factor λ = 0.03 is a reduced-model parameter (the water surface only partly follows small bed changes); it sets the speed of relaxation, not the equilibrium.
Numerical checks run in the page. (1) The measured growth rate of the bar variance equals 2 D from the model's own derivative; (2) the bed volume change equals (sediment in minus out)/(1−p) to round-off during a scour run; (3) a feed equal to capacity leaves a flat bed flat; (4) the Ferguson-Church settling velocity of 0.05 mm silt matches the Stokes law within the expected inertial correction and 1 mm sand lands in the measured range 0.10-0.13 m/s; (5) the Shields number at the solved critical slope equals θcr; (6) the critical shear stress of 0.2 mm sand is close to the textbook 0.15 Pa.
Simplifications. Uniform grain size (no sorting or armouring), no bedforms (ripples and dunes are not generated, so the bed stays smooth), flat free surface with constant depth, bed-shear partition ignored (the skin-friction correction is not applied), the grain speed ~9 u*(1 − 0.7√(θcr/θ)) is only an illustrative animation rate, the water temperature enters only through the viscosity, and the vertical scale of the bed is exaggerated for visibility. Real rivers add width variation, bank erosion, cohesive mud and flood hydrographs.