Delft3D-FM: Flexible-Mesh Coastal Hydrodynamic Simulation

Deltares' Delft3D Flexible Mesh (Delft3D-FM) suite has become a cornerstone tool for engineers and scientists modeling complex water systems where rigid structured grids fall short. By combining unstructured mesh technology with tightly coupled hydrodynamic, wave, sediment transport, and water quality modules, Delft3D-FM enables high-fidelity simulation of coastal inundation, tidal dynamics, salinity intrusion, and morphological change — all within a single, interoperable framework.

Why Flexible Mesh?

Traditional structured-grid models (including the legacy Delft3D-4 curvilinear solver) require the modeler to choose between coarse resolution everywhere or prohibitively large cell counts to resolve localized features. Delft3D-FM's unstructured mesh engine — the D-Flow FM kernel — eliminates this trade-off by supporting mixed cell types: triangles, quadrilaterals, pentagons, and hexagons can coexist in a single domain. This allows:

Local refinement around harbor entrances, tidal inlets, or bridge piers without propagating fine resolution across the entire domain.
Smooth transitions between offshore coarse cells (500–1000 m) and nearshore fine cells (5–20 m) in a single simulation.
Accurate shoreline representation using triangular cells that conform to irregular coastlines, reducing staircase artifacts common in Cartesian grids.

The mesh is generated and managed through the RGFGRID and D-HYDRO GUI tools, or programmatically via the meshkernel Python library — enabling automated mesh refinement workflows driven by bathymetric gradients or distance-to-shore criteria.

Delft3D-FM module coupling architecture: D-Flow FM hydrodynamic core with D-Waves, D-Morphology, D-Water Quality, and D-Real Time Control

D-Flow FM: The Hydrodynamic Core

The D-Flow FM solver integrates the depth-averaged (2DH) or fully three-dimensional (3D) shallow-water equations on the unstructured mesh using a finite-volume, staggered scheme. Key numerical features include:

Implicit time integration with a Courant-number-independent stability criterion, allowing large time steps (minutes) in deep offshore regions while maintaining accuracy in shallow nearshore zones.
Wetting and drying handled robustly through a threshold-based flooding algorithm, critical for intertidal flat and salt marsh simulations.
Sigma and Z-layer vertical coordinates (or a hybrid combination), giving modelers flexibility to resolve stratification in estuaries or bottom boundary layers in shelf seas.
Parallel computation via MPI domain decomposition, scaling efficiently to hundreds of cores on HPC clusters.

For 3D applications, turbulence closure is provided by the k-ε model or the algebraic eddy-viscosity scheme, with options for Stokes drift and wave-induced mixing when coupled to the spectral wave module.

Module Coupling: Waves, Sediment, and Water Quality

Delft3D-FM's power lies in its modular coupling architecture. The D-Waves module (based on SWAN) exchanges radiation stress gradients and wave-orbital velocities with D-Flow FM at configurable intervals, enabling wave-current interaction in surf zones and storm-surge scenarios. This online coupling is essential for accurately predicting wave setup, longshore currents, and wave-driven sediment transport.

The D-Morphology module extends the hydrodynamic simulation with non-cohesive and cohesive sediment transport:

Van Rijn (2004) and Engelund-Hansen formulas for bedload and suspended load.
Cohesive sediment (mud) transport using the Partheniades-Krone erosion-deposition framework, with user-defined critical shear stress and settling velocity.
Morphological acceleration factor (MorFac) to compress multi-year morphological evolution into feasible simulation times.

Water quality and ecological dynamics are handled by the D-Water Quality (Delwaq) module, which solves advection-diffusion-reaction equations for dissolved oxygen, nutrients, phytoplankton, and contaminants on the same unstructured mesh. Delwaq supports the BLOOM algal growth model and the GEM general ecological module, making it suitable for eutrophication studies in estuaries and coastal lagoons.

Delft3D-FM D-Morphology: bed level change map and vertical suspended sediment concentration Rouse profiles

Practical Workflow: Storm-Surge Inundation Study

A typical coastal inundation study with Delft3D-FM follows this workflow:

Bathymetry and topography assembly: Merge offshore GEBCO/ETOPO data with nearshore multibeam surveys and LiDAR-derived coastal DEMs using the QuickIn interpolation tool.
Mesh generation: Create a coarse offshore mesh (~500 m) refined to 10–25 m in the target coastal zone using meshkernel's refinement-by-polygon feature.
Boundary conditions: Impose tidal water levels from a global tidal model (e.g., FES2014 or TPXO) at the open ocean boundary; apply meteorological forcing (wind, pressure) from ERA5 or operational NWP output.
Calibration: Tune bed roughness (Manning's n or Chézy C) against observed water levels and current velocities from tide gauges and ADCP deployments. The OpenDA data assimilation framework integrates directly with Delft3D-FM for ensemble-based parameter estimation.
Ensemble storm-surge runs: Drive the calibrated model with a suite of synthetic or historical storm tracks to generate probabilistic inundation maps.
Post-processing: Extract maximum water levels, flood extents, and flow velocities using the dfm_tools Python package, which reads native .nc output files and integrates with xarray and matplotlib.

Delft3D-FM storm-surge inundation simulation: maximum water level map and surge hydrographs at coastal gauge stations

Integration with Python and Open-Source Ecosystems

Deltares has invested heavily in Python-based tooling around Delft3D-FM. The hydrolib-core library provides a Pythonic interface for reading, modifying, and writing all Delft3D-FM input files (.mdu, .ext, .bc, .pol), enabling scripted sensitivity studies and automated model setup pipelines. Combined with meshkernel for mesh manipulation and dfm_tools for output analysis, a complete model lifecycle can be managed without the GUI — a significant advantage for operational forecasting systems and CI/CD-driven model validation workflows.

The model's native NetCDF output conforms to CF conventions, ensuring compatibility with standard oceanographic tools such as NEMO's XIOS, Ferret, and the Python netCDF4 and xarray stacks.

Licensing and Access

Delft3D-FM is available under a dual-license model:

Open-source (LGPL): The full source code is hosted on the Deltares GitLab repository, allowing community contributions and custom builds.
Commercial license: Deltares offers support contracts, GUI tools (D-HYDRO Suite), and cloud-based execution through the Delft3D Online platform.

Pre-compiled binaries for Linux and Windows are distributed via the Deltares software portal. The active user community is supported through the Delft3D forum and annual Delft Software Days conference.

Summary

Delft3D-FM represents the state of the art in flexible-mesh coastal and estuarine hydrodynamics. Its unstructured mesh engine resolves complex geometries efficiently, its modular coupling with waves, sediment, and water quality enables integrated environmental impact assessments, and its growing Python ecosystem supports modern, reproducible modeling workflows. For practitioners working on storm-surge forecasting, tidal inlet morphodynamics, estuarine salinity intrusion, or coastal water quality, Delft3D-FM provides a robust, well-validated, and openly accessible simulation platform.

Delft3D-FM: Flexible-Mesh Hydrodynamic Simulation for Coastal, Estuarine, and River Systems