Bubble Tea Shake

Sloshing at the natural frequency and tapioca pearls stirred up from the bottom.

Switch to 2D

Presets

Playback

Shake motion

Choose how the cup is moved.

Shaking

Cup and tea

Changing radius, depth, pearl count or size pours the pearls back.

Pearls

Tea

Actions

Camera

Quality

Layers

Model checks

The science

The model

The cup is a vertical cylinder of radius R holding tea of depth h. In the frame of the cup the liquid feels gravity plus the inertial force of the shaking, so the effective gravity is g_eff = (-a_x, -a_y, -(g + a_z)).

Sloshing mode

For small waves the free surface of an inviscid liquid in a cylinder has the lowest azimuthal mode eta(r, theta, t) = [q_x cos(theta) + q_y sin(theta)] J1(xi r/R) / J1(xi) with xi = 1.8412 the first zero of J1' (Lamb, Hydrodynamics, art. 257). Its natural frequency is omega^2 = (g xi / R) tanh(xi h / R) - about 3.4 Hz for a 4 cm radius cup filled 11 cm deep. The amplitude q obeys a damped driven oscillator q'' + 2 zeta omega q' + omega^2 q = -Gamma a where Gamma = omega^2 R c0 / g and c0 = 2/(xi^2 - 1) is the projection of the tilted flat surface eta = -a x / g on the mode (checked here by numerical quadrature). Large waves lose energy to breaking, modelled by zeta = zeta0 + kappa |q| / R.

Flow under the surface

The same velocity potential gives the liquid velocity: horizontal speed u = q' (J0(kr) - J2(kr) cos 2 theta, ...) / (2 J1(xi) tanh(kh)) cosh(k(z+h)) / cosh(kh) and vertical speed w = q' J1(kr) cos(theta) sinh(k(z+h)) / (J1(xi) tanh(kh) cosh(kh)), with k = xi / R. It decays with depth like cosh, which is why a plain sideways shake barely moves pearls resting on the floor of a deep cup.

Pearls

Each pearl (diameter d, density rho_p) obeys (rho_p + rho_f/2) dv/dt = (rho_p - rho_f) g_eff + (3/4) rho_f (C_D/d) |u - v| (u - v): net weight (buoyancy included), added mass of the liquid it drags along, and drag with the Schiller-Naumann coefficient C_D = 24/Re (1 + 0.15 Re^0.687) + 0.42 / (1 + 42500 Re^-1.16), Re = rho_f |u - v| d / mu. In still tea a pearl falls at the terminal speed where weight equals drag. Pearls collide with each other (restitution e), the floor, the wall and the surface.

Turbulence closure (empirical)

Violent shaking breaks waves and mixes the liquid down to the floor, which a linear potential flow cannot describe. The model adds a random velocity (Ornstein-Uhlenbeck process, eddy size 0.8 R) with rms sigma = mix x sqrt(surface speed^2 + (A Omega / 2)^2) to the liquid velocity seen by every pearl. The factor mix is tuned, not derived; set it to 0 to see the pure potential-flow result (pearls then stay put).

Simplifications

Straight cylinder, no lid, no straw; linear wave theory with an amplitude-dependent damping patch; the liquid is not disturbed by the pearls (one-way coupling); no pearl rotation; pearls are identical rigid spheres. The surface is clamped at 80 % of the depth. Trends and orders of magnitude are right, individual trajectories are not predictions.

What to try
  1. Press the first preset: a side-to-side shake at the natural frequency makes the surface rise by centimetres at the wall (graph 1) and the resonance curve (graph 4) shows why.
  2. Move the shake frequency away from the natural one with the slider: the wall rise collapses, as the resonance curve predicts.
  3. Set the turbulence efficiency to 0: the potential flow alone does not lift the pearls from the floor. Raise it and watch the suspended share and the depth histogram change.
  4. Vertical shake above 1 g (the fourth preset): pearls hop off the floor because their effective weight drops.
  5. Make the tea thick (viscosity slider) or the pearls nearly as dense as tea: the terminal speed drops and the suspension lasts longer after you stop.
Loading Blender model...
Cup motion and surface rise at the wall
Pearl height over time
Where the pearls are: depth histogram
Resonance curve of the sloshing mode
Made with 🟠 Blender · Three.js