Iceberg: Buoyancy, Metacentric Stability and Why Melting Bergs Roll Over

A tabular iceberg melts in a time-lapse: freeboard from Archimedes, stability from the metacentric height, and a capsize when the berg becomes too narrow for its thickness.

Switch to 2D

Presets

Playback

Parameters

Camera

Layers

Quality

Model check

The science

What you see. A tabular iceberg floating beside a 150 m ship for scale. In the time-lapse the berg melts: the sides and the waterline are eaten faster than the bottom, so it gets narrower relative to its thickness until it becomes unstable and rolls over. The end view shows the roll. Set the dimensions, ice density, salinity, water temperature, current and sea state; the Nudge button kicks the berg like a wave.

Floating. Archimedes' principle gives the submerged fraction f = ρice / ρwater: about 89 % in sea water, and a little more in fresh water, so roughly a tenth of the thickness shows above the sea.

Stability. For a rectangular cross-section of width W and thickness H the draft is d = fH and the metacentric height is GM = W²/(12 d) − (H − d)/2. The berg is stable only if GM > 0, that is W/H > √(6 f (1 − f)) (about 0.75). The dynamics integrate I θ'' = τ(θ) − c θ' with the exact righting torque τ = ρw g A xB, where the centre of buoyancy xB is the true centroid of the submerged part of the tilted rectangle (heave solved at every angle), so large angles and the capsized equilibrium at 90° come out of the same model. I includes 40 % added water mass.

Melt. Basal and side melt use the empirical Bigg-type form Mb = 0.58 |u|0.8 (Tw − Tice) / L0.2 in m/day with the ice at −10 °C, and wave erosion at the waterline Me = Ss (Tw + 2)/12 m/day. Water density is a simple linear fit in salinity and temperature.

Honest simplifications. The berg is a perfect box of uniform ice (real bergs have a keel, cracks, trapped air and irregular shape). Rolling is two-dimensional, per unit length, with no heave dynamics, drag on the roll beyond a damping ratio of 0.35, or wave-making; melt coefficients are empirical and the time-lapse speed is not real time. The roll period, damping and the sea-state torque are illustrative. Real capsizing bergs are tens of metres of rotating ice and are dangerous, so this model only shows the principle.

Checks. (1) closed-form GM against the numerical slope of the exact righting torque; (2) the critical width-to-thickness ratio by bisection against the formula; (3) the Archimedes area condition and the two equilibria; (4) the small-roll period against a time integration of the exact torque model.

Loading model...
Loading Blender model...
Righting torque against tilt angle
Metacentric height against W/H (H = 200 m)
Freeboard against salinity
Made with 🟠 Blender · Three.js