Houses are 5 plywood models on the dry beach; they darken when the shoreline reaches them.
Runs 10 waves from H/h = 0.006 to 0.25 on the current slope and plots run-up against the Synolakis law.
Equations. Depth-averaged long-wave (nonlinear shallow-water / Saint-Venant) equations in 1-D:
ht + (hu)x = 0, (hu)t + (hu² + ½ g h²)x = −g h bx − Cf|u|u, solved with finite volumes, an HLL Riemann flux, hydrostatic reconstruction (exactly keeps still water still, handles the moving shoreline) and MUSCL/SSP-RK2.
The wave. A solitary wave η = H sech²(γ(x−x0)), γ = √(3H/4h³), travels at c = √(g(h+H)). On a plane beach of slope 1:cotβ Synolakis (1987) found the maximum run-up of a non-breaking wave R/h = 2.831 √cotβ (H/h)5/4, and breaking on the slope for H/h > 0.8183 cotβ−10/9. Beyond that limit part of the energy is lost in the bore and R falls below the law.
Why a basin. Real tsunamis cannot be measured at the shore in advance; engineers use Froude-scaled flumes (lengths ×λ, speeds ×√λ, times ×√λ). The prototype block above applies that scaling.
Honest simplifications. 1-D (no sideways spreading, so the tank width is only for show); non-dispersive, so steep fronts steepen into a bore a little earlier than in a real flume and the numerical shock stands in for breaking; no turbulence, air entrainment or bed friction (Cf = 0 by default); the wave is initialised next to the paddle instead of being generated by a piston stroke; water and bed are drawn exaggerated vertically (the slider), the tank is not to scale vertically. Models: paddle, gauges, frame, model houses built in Blender; not affiliated with any laboratory.
Independent checks (run on page load with this same solver):