Four-Engine Pipe Comparator

Same St. Venant physics, four ways of computing it

All four engines route unsteady flow through closed conduits from the same 1D conservation laws. What separates them is how pressurization is handled, implicit vs explicit time-stepping, and how the pipe network talks to the surface. This app is a working system for comparing them: a shared benchmark suite (SWMM5 .inp as the interchange format), a solver-numerics lab, a capability matrix, and a terminology map.

What “SWMM6” means here: the next-generation open track — SWMM5+ (CIMM.org, finite-volume, parallel) and OpenSWMM — rather than a shipped EPA release. Its rows are roadmap-level and marked accordingly; edit COMPARISON_DATA at the top of this file as the alphas ship.

Numerics Lab — the pressurization wave and the timestep it costs

A circular pipe’s top width T → 0 as it fills, so the gravity-wave celerity c = √(gA/T) blows up near the crown. The Preissmann slot caps that blow-up with a narrow virtual slot of width b = f·D — but an explicit scheme must then resolve the slot wave, and Δt collapses. Implicit and semi-implicit engines don’t pay that bill. Drag the sliders and watch what each engine’s timestep philosophy does with the same pipe.

The comparison system — one model, four engines, one scoreboard

Tip: HEC-RAS can also convert a SWMM .out file to HEC-DSS and overlay it as observed data — so the SWMM5 reference run plots directly against the HEC-RAS pipe results.

import h5py, numpy as np, pandas as pd
f = h5py.File('MyModel.p01.hdf', 'r')
# browse: f.visit(print)  → find the pipe-network Results path for your conduit,
# then pair it with the matching Time dataset and save t,Q rows:
# pd.DataFrame({'t_hr': t, 'q': q}).to_csv('c1.csv', index=False)

Which engine for this job?

Tick what the project actually requires. Scores are 0–2 per requirement per engine (the scoring table is ADVISOR in the source — argue with it there).