Rainfall is not constant — a real storm rises to a brief peak and tapers over a long tail.
Treating the 300 mm/hr peak as if it held steady massively overstates the water. This simulator plays a
realistic hyetograph (intensity vs. time, with natural minute-to-minute pulsing) and runs each
instant through the filter — runoff → flow-dependent capture → harvest + waste — accumulating the
true totals. The honest headline number is the volume-weighted capture over the whole storm,
which no single constant-intensity figure can give you.
Naive assumption: peak inflow × full storm duration. This is what a single-intensity model implies if you forget the storm tapers.
The true area under the hyetograph — the rain that really fell. —
Volume-weighted capture over the whole fluctuating storm — the only single number that is actually defensible.
Method: hyetograph = gamma-pulse envelope (rising limb, sharp peak,
recession tail) with ±15 % stochastic minute-to-minute jitter. Each timestep: runoff
Q = C·i·A (C = 0.90, A = 46.5 m²); capture
η(Q) = 0.65 + 0.31·e^(−Q/150) (falls as flow rises — contact-time law, fitted to measured data for comparable fine-mesh inline filters); harvest
= min(Q·η, 240 L/min outlet cap); waste = remainder. Totals are trapezoidal integrals
over the storm. Synthetic design storm for illustration — replace with gauge data when available.