HEC-RAS is the default flood modelling tool for most of the world, and that is a reasonable state of affairs rather than an accident. It is developed by the US Army Corps of Engineers, it is free, it does 1D and 2D, and its results are accepted by regulators almost everywhere. If you are choosing a flood model and have no specific reason to do otherwise, start here.
The commercial packages are not competing on correctness. They compete on specific capabilities, on how well they handle very large or complex domains, and on workflow.
HEC-RAS, and why it is the default
HEC-RAS covers the great majority of riverine flood work: steady and unsteady 1D hydraulics, 2D areas, coupled 1D-2D, bridges and culverts, dam and levee breach, and sediment transport.
It is the right choice when:
- The problem is riverine. A river, a floodplain, structures along it.
- The result must be defended. The USACE pedigree and open availability mean a reviewer can rebuild your model, which is a strength rather than a risk.
- Budget matters. The money saved is better spent on survey and terrain data, which affect the answer far more than the solver does.
Its practical limits show up on very large 2D domains, on complex coastal processes, and in team workflow on big programmes.
Where MIKE earns its licence
The DHI MIKE family is the usual choice when water is not behaving like a river. MIKE 21 is a general 2D free surface flow model, and MIKE FLOOD couples 1D and 2D.
It is worth the cost when the problem involves coastal and estuarine processes, where tides, waves and salinity matter, or where the study is genuinely a hydrodynamic one rather than a floodplain mapping exercise. DHI also has a deep bench of coupled modules, which matters when flooding is one part of a wider water system question.
Where TUFLOW fits
TUFLOW is a 2D and 1D-2D hydraulic modelling package with a strong following, particularly in Australia and the United Kingdom, and it is well regarded for large 2D flood studies and for its computational performance on big domains.
Where a regional flood mapping programme has to run many scenarios over a large area, that performance is a real project constraint rather than a technical nicety. Client and regulator familiarity in a given market is also a legitimate reason to choose it, and worth asking about early.
Where Delft3D belongs
Delft3D from Deltares is a 3D hydrodynamic, morphological and water quality modelling system. It is the right tool for coastal and estuarine work where sediment transport, morphology and stratification are part of the question rather than a boundary condition.
For a straightforward fluvial flood map it is more model than the problem needs.
The choice is usually made by the water body
As with groundwater, the deciding factor is physical rather than commercial. A river and its floodplain points to HEC-RAS. A coast, an estuary or a tidal system points to MIKE or Delft3D. A very large 2D domain with many scenarios points to whichever package your team can run fastest, which is often TUFLOW.
What almost never decides it is the feature list. Every one of these will solve the shallow water equations competently.
What actually determines the answer
Flood model accuracy is dominated by terrain and by the flow record, not by the solver. A LiDAR survey at the right resolution, correct structure geometry, and a gauge record that has been checked rather than assumed will move your result far more than switching packages.
This is the same point we make about designs failing on their inputs rather than their hydraulics. A model calibrated against an observed flood in a free package beats an uncalibrated model in an expensive one, and it is not a close contest.
We work in these tools on surface water and flood projects and resell none of them. If you are scoping a flood study and are unsure whether your problem is fluvial or genuinely coastal, that distinction is worth settling before anything is procured, and it is part of what our consulting and modelling services do. Where the result has to become a live warning or operating tool, it becomes a decision support system rather than a study.