The wind on the site’s rain now behaves as wind near the ground does: it swells and lulls, gusts and veers, and pushes small drops harder for their weight than big ones. On 10 October I wanted to know how realistic the wind was: whether it starts, stops and changes direction, and whether it can blow the rain anywhere from a little sideways to fully across. Each of those became a control.
A steady sideways push would be wrong. Real wind near the ground has a mean speed with slow swells and lulls over tens of seconds, gusts on top that rise in a second or two, hold and fall more slowly, with gust factors of 1.3 to 2 over the mean, and turbulence between.
The direction wanders and veers in the gusts. A gust sweeps across the pane at about 1,500 mm a second rather than hitting all of it at once. It is deterministic from a seed, so a clip renders the same twice.
A vertical window catches more rain as the wind rises
The first consequence is how much rain reaches the glass at all. A vertical window only catches what the wind drives at it. In the code that rate is 0.3 times the breeze rate in a calm, 1 at a breeze of 8 mph, rising in a straight line to a cap of 3.5. The presets are Calm 0, Breeze 8, Windy 20, Strong 35, Gale 50 and Storm 70 mph.
Small drops are pushed harder for their weight
The second is the push on a drop. Air drags on a drop in proportion to its frontal area times the speed squared, while its weight goes as its width squared times its height, so the push over the weight goes as speed squared over size. Small water is shoved harder for its weight than big.
The code sets a 1.5 mm drop at about 0.35 of its weight in a 20 mph wind and 2.2 weights at 50, and caps the push at 6 weights, because a gale tears water off before it pushes harder. Those are marked as estimates. One check from the literature: in a wind up the glass, small drops climb, big ones still fall and the middling ones hover.
In a gale more effects switch on, none of them calibrated
In a gale the other controls switch on. Above 15 mph a raindrop bursting on the glass throws a mist that the wind carries on downwind. In a strong wind air tears water off a big drop’s downwind edge and flings it as small drops. A hard gust can flex the pane so every drop close to letting go does so at once, a sudden cascade of new streams.
None of those strengths is calibrated against footage, and the project log labels the wind work as work in progress.
Against a frame, water pools along the bottom edge and drips
I also wanted streams to reach the bottom edge of the glass. Where a pane is a section of a longer window, as most of the site’s panes are, the water runs on out of sight.
Against a frame, as on the drop page, water comes to rest along the frame and gathers as drops join. When a pool holds about 28 µL (an estimate, with a variation of 20% either way from drop to drop) it hangs and lets one fall, and 35 to 55% of the water stays. Behind a frame the wind has a lee and a swirl, reaching 14 mm.
In the code, the module called drip is the eye dropper, the pipette tool added on 9 October. Dripping off the edge lives in the simulation.
Sources and notes (9)
- Blocken & Carmeliet 2004, wind-driven rain — wind-driven rain on a window
- Wind- and gravity-forced drop depinning — wind strength needed to move a drop
- Wikipedia: Rain — rain classes
- Wikipedia: Raindrop size distribution — drop sizes in rain
- Gaylard et al. 2017, rivulets in airflow — rivulets pushed by airflow
- Wind code — the notes this was written from
- Drop simulation code — the notes this was written from
- Tuning constants — the notes this was written from
- Rain types code — the notes this was written from
Water, rain & windAll 11 parts →
