The claim under test
“Wet sand sticks together, so the wetter the sand the stronger the castle.”
One bridge. The pull of a single liquid neck between two grains,
against the gap it has to span, for three bridge volumes. Computed live in this page from the
same code the pack uses.
The headline. Tensile strength of the pack against liquid content,
measured over independent random packings.
How tall can it go. The two competing tower limits and their
minimum.
Why it plateaus. Bridges per unit cell and the mean force per
bridge, normalised to their values at the knee.
Nobody measures this and it is the single biggest lever in the whole model.
Clean laboratory quartz is a few degrees; untreated natural sand has been measured at 61°.
This re-runs the packing and the plane-cut strength measurement on
your machine, with a packing this page has never seen, and plots the result on top of the shipped
curve. It is the same code that produced the curve.
Numbers this page will not let itself round off
How to play
- Pick a tool, set how much water is in your bucket, and click on the beach to build.
- The dashed outline turns red when the piece you are about to place will not stand. The
readout tells you which way it fails and by how much.
- Start the clock. The sand dries, and dry sand has no cohesion — a tower that
stood five minutes ago will slump.
- The tide comes in near the end. Saturated sand has no cohesion either, for the opposite
reason: once the pore space is full there are no curved water surfaces left to pull the grains
together.
What the model actually says
What is under the hood
Where this is weak
Credit
Building a sandcastle is a folk activity with no author, no publisher and no owner, and this is
not a replica of anyone’s software. What it is derived from is published physics, and
those people are named. The measurements and formulas this app leans on are due to
H. Rumpf (1962); A. G. Greenhill (1881); Culmann’s wedge method as set out in
M. Murthy’s Geotechnical Engineering; V. Richefeu, M. S. El Youssoufi and F. Radjaï,
Physical Review E 73, 051304 (2006); M. Scheel and co-workers,
Nature Materials 7, 189 (2008), reached here through the open restatement
by C. Semprebon, M. Scheel, S. Herminghaus, R. Seemann and M. Brinkmann; M. Pakpour, M. Habibi,
P. Møller and D. Bonn, Scientific Reports 2, 549 (2012); N. Lu,
T.-H. Kim, S. Sture and W. J. Likos, Journal of Engineering Mechanics 135,
1410 (2009); and the IAPWS surface-tension release R1-76(2014). Every constant taken from them is
listed with its source in CREDITS.txt.
This is an independent implementation. Nothing is copied from any other simulation. The packing
generator, the bridge solver, the Young–Laplace oracle, the wedge minimiser, the buckling
eigensolve and the renderer are all written from the equations.
Where the numbers on this page come from
Source, licence and machine-readable notes:
CREDITS.txt ·
LICENSE.txt ·
llms.txt