Every few weeks a design manager asks us which tool we will use for their drainage study, usually with a preference already formed by what their reviewer last accepted. The honest answer is that we run both — HEC-RAS 2D and InfoWorks ICM — and the choice is a project decision, not a loyalty. Here is how we actually make it.
The one-paragraph version
Rule of thumb: surface-dominated problems → HEC-RAS 2D. Network-dominated problems → InfoWorks ICM. Wadi flood routing, rain-on-grid flood mapping, scour and bridge hydraulics: RAS. Dense urban pipe networks with surcharge, RTC and pipe-surface interaction at scale: ICM. The grey zone in between is where engineering judgement earns its fee.
Where HEC-RAS 2D wins
- Cost and auditability. RAS is free, open-documentation USACE software. Any reviewer anywhere can open your model — no licence negotiation, no version hostage situations. For municipal review in KSA this matters more than people admit.
- 2D engine quality. The sub-grid terrain treatment lets a relatively coarse mesh honour fine terrain detail — curbs, channels, bunds — without exploding cell counts. For rain-on-grid over large desert catchments it is exceptional.
- River and structure hydraulics. Bridges, culverts, inline structures, sediment and scour workflows (with FHWA HEC-18 alignment) are native strengths. Wadi flood studies and crossing designs live here.
- Transparency. Equation sets, solver options and warnings are documented publicly; results are reproducible by anyone with the files. Good science, good liability posture.
Where InfoWorks ICM wins
- Integrated 1D–2D at network scale. ICM's coupled pipe/manhole/surface engine is the most mature on the market. When the question is "what does the network do at 10-year, and where does the exceedance flow go at 100-year", ICM answers both in one model.
- Network data management. Versioned databases, scenario management and audit trails are built for utility-scale models with thousands of assets — a different league from file-based workflows when a model lives for years.
- Urban drainage detail. Gullies and inlet capacity, real-time control, pump logic, water quality and SuDS representations are first-class features, not workarounds.
- Throughput. GPU-accelerated 2D and cloud simulation batches make large scenario matrices (multiple storms × durations × climate uplifts) practical on programme deadlines.
The differences that actually bite
| Decision factor | HEC-RAS 2D | InfoWorks ICM |
|---|---|---|
| Licence cost | Free | Significant; scales with seats/engines |
| Pipe network modelling | Basic (improving in recent versions) | Industry-leading 1D engine |
| Rain-on-grid flood mapping | Excellent, sub-grid terrain | Very good, GPU-fast |
| Structures & scour | Native strength | Workable, not the focus |
| Model handover | Anyone can open it | Receiver needs licences |
| Long-life asset models | File-based, manual discipline | Database, versioned, auditable |
| Reviewer acceptance (Gulf) | Universal | High, esp. on utility frameworks |
Three traps we see in other people's models
- ICM used as a terrain flood mapper with the network switched off — paying licence premiums for a job RAS does free, then struggling at handover because the municipality can't open the model.
- RAS pushed into dense-network territory — dozens of surcharging pipes hacked in as 2D channels or simplistic connections. The flood map looks plausible; the network hydraulics are fiction.
- Mesh convergence never tested in either tool. If halving the cell size moves your peak depth more than a few percent, you don't have a result yet — you have a draft. Budget for the convergence run.
What we actually do on Gulf projects
A common Archeve pattern on KSA work: a HEC-RAS 2D rain-on-grid model establishes regional flows and flood context (cheap, reviewable, defensible), feeding boundary conditions to an ICM model of the urban network where pipe-surface interaction governs design. Rainfall inputs for both come from the process described in our companion note on Saudi rainfall design criteria. The outputs drive network sizing and ultimately the IFC drawing package.
Choosing a tool — or inheriting a model you don't trust? Brief us. A principal who has defended both platforms in municipal review replies within two business days.