OpenHydro Bench

Which water models get the answer right?

An open leaderboard for flood, river, drainage and water-network models. Every test has a known right answer, every model runs the same inputs, and every result can be re-run by anyone.

See the full leaderboardScore your model

Top models by OpenHydro rating

  1. 1=Basilisk Saint-Venant1290
  2. 1=ANUGA 4.0.11219
  3. 1=LISFLOOD-FP 8.3 · DG21139
  4. 1=LISFLOOD-FP 8.3 · FV11139
  5. 1=TELEMAC-2D openTELEMAC 9.11128
  6. 2=Hydra urban drainage1015
  7. 2=LISFLOOD-FP 8.3 · ACC987
  8. 2=HEC-RAS 7.0 1D · 2D SWE-ELM961
  9. 2=HEC-RAS 7.0 2D · SWE-EM951
  10. 3OpenSWMM 6 finite volume876
Higher is better. Each dot is a model's rating and the band its 95 % interval. Production software and research codes, all tests.Full table and per-group boards
39tests, from pipes to floodplains
16models scored, including EPANET, HEC-RAS, TELEMAC-2D, ANUGA
1,396real SWMM models checked
Everyresult re-run from scratch on each change

Start where you are

I use flood or drainage models

Compare models on the same tests

See how close each model gets to the right answer, ranked test by test and overall, with run times.
I build a solver

Score your code

Download the inputs, run them, and check your results in the browser before you submit.
I review model results

Check why a result can be trusted

Fixed inputs, pinned software versions, every result reproducible.

What the tests found

All results →
2.1–105 m
The dam-break front splits the field

A minute after the classic dam break, production codes put the flood front between 2.1 m (ANUGA) and 105 m (LISFLOOD-FP) from the exact one: 0.55 % to 28 % of the 376 m it travels.
SWMM 6
OpenSWMM 6's new routing gets the bore SWMM 5 misses

On a dam break into shallow water, HydroCouple's finite-volume routing gets the bore right and converges; SWMM's classic dynamic wave, in both 5.2.4 and 6, does not.
0.001 ft
HEC-RAS rounds its inputs

HEC-RAS 6.6 stores elevations to the nearest thousandth of a foot, even in a metric model. Its backwater error is tiny (0.1 mm) but cannot shrink on finer grids.
76 %
SWMM 5.1 and 5.2 mostly agree

355 of 470 real models give identical results. Pumps, green roofs and groundwater are where the versions part ways.

Tests built from real engineering questions

Each one isolates something a model has to get right on a real project, from a sewer pipe running full to a dam failing above a city.

All 31 tests →
Rivers and channelsA1F-010

Dam break onto dry ground

How far, and how fast, does the flood travel?

  1. 1Basilisk Saint-Venant0.45 %
  2. 2=Demo 1D suite example0.52 %
  3. 2=ANUGA 4.0.10.55 %
  4. 4=OpenSWMM 6 finite volume0.88 %
  5. 4=GeoClaw Clawpack 5.90.93 %
  6. 6=LISFLOOD-FP 8.3 · FV11.3 %
  7. and 11 more
2D floodsA2C-EA4

Flood spreading over a floodplain

How fast does a breach flood spread over flat ground?

  1. 1=ANUGA 4.0.1<0.01 %
  2. 1=Demo 2D suite example<0.01 %
  3. 1=LISFLOOD-FP 8.3 · ACC<0.01 %
  4. 1=LISFLOOD-FP 8.3 · DG2<0.01 %
  5. 1=TELEMAC-2D openTELEMAC 9.1<0.01 %
  6. 1=HEC-RAS 7.0 1D · 2D SWE-ELM<0.01 %
  7. and 2 more
Rivers and channelsA1F-042

Hump with a hydraulic jump

Does the jump below the hump stand in the right place?

  1. 1Basilisk Saint-Venant0.076 %
  2. 2ANUGA 4.0.10.11 %
  3. 3TELEMAC-2D openTELEMAC 9.10.17 %
  4. 4=OpenSWMM 6 finite volume0.19 %
  5. 4=LISFLOOD-FP 8.3 · DG20.2 %
  6. 4=LISFLOOD-FP 8.3 · FV10.2 %
  7. and 5 more
Sewers and drainageA1D-002

Sewer pipes running full

When pipes surcharge, does the model get the water level in the manholes right?

  1. 1Hydra urban drainage<0.01 %
  2. 2=OpenSWMM 6 dynamic wave0.034 %
  3. 2=EPA SWMM 5.2.4 dynamic wave0.034 %
  4. 4OpenSWMM 6 finite volume23 %
Sewers meeting streetsA12-012

Overflow and return

When a sewer overflows and then drains back, does every drop come out?

  1. 1SWMM + ANUGA coupled 1D/2D0.028 %
  2. 2=OpenSWMM 6 coupled 1D/2D0.22 %
  3. 2=OpenSWMM 6 coupled 1D/2D · cross mesh0.24 %
  4. 4HEC-RAS 7.0 pipe networks · 1D/2D1.3 %
Water mainsA1P-020

Water hammer: valve slammed shut

Is the pressure surge right when a valve shuts instantly?

  1. 1Demo 1D suite example<0.01 %
  2. 2TSNet water hammer · MOC0.13 %
SWMM and EPANET compatibilityB-RTC-001

Pump station switched by rules

Do pumps switch on and off at the same moments as in SWMM 5.2.4?

  1. 1SWMM 5.1.15 previous release<0.01 %
  2. 2OpenSWMM 6 vs SWMM 5.2.451 %
  3. 3Hydra urban drainage90 %
  4. 4OpenSWMM 6 finite volume3e4 %
31tests with a known right answer
Rivers and channels2D floodsSewers and drainageSewers meeting streetsWater mainsSWMM and EPANET compatibility
Browse every test →

How it works

01 — Same inputs

Every model gets identical terrain, flows and settings, each file fixed by a fingerprint so nobody can quietly change it.

02 — A known right answer

An exact mathematical solution, laboratory measurements, or — for compatibility — the output of a pinned SWMM or EPANET release.

03 — A result anyone can check

Every result as a percent difference from the right answer, with the measurement behind it. Every result is re-run automatically and signed.

Want the full method — convergence rates, what each difference is measured against, the audit log? Read how it works.