WEC-Sim Hydrodynamics for Wave Energy Converters in a Physics-AI Workflow
New1. Overview
Wave energy converters must survive fifty-year storms while extracting power from irregular seas. Physical tank testing cannot cover the design space, so reliable time-domain simulation has become the workhorse. WEC-Sim was initiated by teams at the National Renewable Energy Laboratory and Sandia National Laboratories, first released in 2014, and is distributed under the Apache-2.0 license from the GitHub repository WEC-Sim/WEC-Sim. The repository was created on 3 June 2014 and last pushed on 12 October 2026; MATLAB is the primary language, with roughly 195 stars, 198 forks, and a single-digit number of open issues.
Three reasons justify a dedicated deep entry. First, WEC-Sim is one of the few open-source simulation frameworks that has reached the status of a working reference in its field: a review published at OMAE 2025 counts more than 250 publications that develop or apply the code, the majority written by external authors, and a 2022 review in the International Marine Energy Journal cites the growth of external applications as primary evidence of broad acceptance. Second, the physical organisation of the kernel is directly instructive for anyone building physical-AI tooling: linear hydrodynamic coefficients from a frequency-domain boundary element solver are turned into time-domain convolution integrals through state-space realisation and solved together with rigid-body multibody dynamics, mooring restoring forces and power take-off damping inside a single Simulink model. Third, the software has moved past wave energy: the OMAE 2025 paper on applications beyond wave energy systematically reviews its use for floating wind turbines and hybrid platforms, which turns it from a niche device code into a general marine multibody dynamics framework.
2. How It Works
The modelling strategy is defined explicitly in the official theory manual. WEC-Sim uses a radiation and diffraction approach: a frequency-domain potential-flow boundary element solver solves for the velocity potential and returns a set of linear hydrodynamic coefficients for excitation, added mass, radiation damping and free-surface behaviour; the time-domain simulation then convolves this frequency-domain data back onto the time axis and couples it with structural dynamics to solve the motion equations.
The coordinate system is the precondition for reproducible results. The manual fixes the positive X axis as a wave heading of zero, the positive Z axis vertically upwards, and the Y axis by the right-hand rule. The six degrees of freedom are ordered translation first and rotation second, with pitch about Y and yaw about Z in the fifth and sixth positions. Units are MKS throughout, with angles in radians except wave direction, which is expressed in degrees.
Two paths supply the hydrodynamic coefficients. The main path imports a normalised data structure in the HDF5 format generated by an external boundary element solver, with readers for WAMIT, ANSYS Aqwa, NEMOH and Capytaine. The alternative is for the user to construct the same structure manually. After reading, WEC-Sim rescales the dimensionless coefficients according to the formulas in the theory manual, recovering dimensional excitation forces, the added mass matrix, the radiation damping matrix and the hydrostatic restoring coefficients.
The time-domain equation is the six-degree-of-freedom rigid body equation of motion, containing excitation force and moment, mean drift force, radiation force, power take-off force, viscous damping force, Morison element force, buoyant restoring force and mooring force. The radiation term carries the added mass matrix A(omega) and the radiation damping matrix B(omega), both functions of radian frequency.
Handling fluid memory effects is the technical core. The manual offers two routes: the convolution integral formulation integrates the impulse response directly, which is the most accurate but the most expensive; the state-space formulation uses realisation theory to express the impulse response as a single-input zero-state model, replacing the convolution with a set of constant-coefficient differential equations. The realisation section prescribes the standard algorithm, a singular value decomposition of the Hankel matrix of the impulse response followed by a logarithmic construction of the state-space realisation, with the 1978 paper by Kung as the canonical reference.
On the excitation side the framework covers everything from regular waves to fully directional spectra. The regular wave section derives the dispersion relation and phase velocity. Irregular waves are generated by linear superposition of many regular waves of differing amplitude and period, with Pierson-Moskowitz and JONSWAP spectra including the peak-shape parameter expression, plus frequency-dependent wave direction and directional spreading. Non-linear buoyancy and Froude-Krylov terms extend the forcing, and viscous effects are represented by linear damping, quadratic drag and Morison elements.
3. Scope and Assumptions
The usable scope of WEC-Sim is set by its physical assumptions, and those assumptions determine which questions it can answer.
The most important one is the potential-flow assumption. The boundary element solver solves the Laplace equation under the assumption of inviscid, incompressible, irrotational flow. The framework therefore suits large-scale, high-Reynolds-number bodies of relatively regular geometry. For slender devices, bodies with sharp edges, or very shallow-water operation, the potential-flow solution degrades, and Morison elements only partly compensate for such error.
The second boundary is rigid-body dominance. The manual frames the motion equation around a rigid body, and flexibility is introduced through generalised body modes, with non-linear hydrodynamics on the first six degrees of freedom only reaching the flexible body model in version 7.0.0. Non-rigid structural dynamics arrived as an extension in 2020 and does not yet sit at the same maturity level as the rigid-body path.
The third boundary is the MATLAB dependency. The official getting-started guide names MATLAB, Simulink, Simscape and Simscape Multibody as the required components, and states that compatibility is guaranteed only with the four most recent MATLAB releases, with the Simulink library saved in the R2020b format to widen compatibility. The mooring module MoorDyn, the parallel computing toolbox and Simscape Fluids are additional dependencies that the main repository does not ship by default.
The fourth boundary is engineering discipline. The coefficients returned by boundary element solvers are rescaled before entering the time domain, and an inconsistency in the rescaling formula, the coordinate convention or the unit system produces silently wrong results. The v7.0.0 release notes list items such as resolving a unit discrepancy, fixing a deep water flag defect and fixing phase calculation for large horizontal displacements, which shows these issues arise in real projects.
4. Getting Started
The official guide lays out a four-step path that matches the real order of work.
First, put the source directory on the MATLAB path. There are two options: copy the contents of source/addWecSimSource.m into a new startup.m placed in the MATLAB startup folder so the path is added automatically at every launch, or run addWecSimSource manually each time and remove it with removeWecSimSource. Cloning with Git large file storage is required, because the hydrodynamic data files exist in the repository as HDF5 files; the v7.1.0 notes specifically mention moving those files out of large file storage into ordinary Git to simplify acquisition.
Second, verify the path with the path command and confirm that the source directory and its subdirectories are listed. Third, refresh the Simulink Library Browser so the WEC-Sim library becomes visible. Fourth, run the smoke test: enter the examples/RM3 directory and execute wecSim, which runs the Reference Model 3 point absorber, opens a Mechanics Explorer window and produces plots; examples/OSWEC also runs out of the box. A git-lfs error means large file storage was not installed correctly, and the required file can be regenerated by running examples/RM3/hydroData/bemio.m.
The layout of the examples carries information in itself. The examples directory splits into BEMIO, OSWEC and RM3, with BEMIO holding input samples for the four solvers AQWA, CAPYTAINE, NEMOH and WAMIT in separate subdirectories plus a compareBEMIO.m script for cross-solver comparison. The source/objects directory holds the class implementations, where bodyClass.m at roughly 91 KB, waveClass.m and responseClass.m at roughly 53 KB each are the largest. The source/functions/BEMIO directory concentrates the solver readers, coefficient merging, impulse response computation and spectral processing, with four separate readers readCAPYTAINE.m, readNEMOH.m, readWAMIT.m and readAQWA.m.
The companion WEC-Sim_Applications repository is the best resource after that first step, covering advanced features in more than thirty directories: Body-to-Body_Interactions, Cable, Controls, Desalination, End_Stops, Free_Decay, Full_Directional_Waves, Generalized_Body_Modes, Load_Mitigating_Controls, MOST, Mean_Drift, Mooring, Morison_Element, Nonlinear_Hydro, OWC, PTO-Sim, Paraview_Visualization, Passive_Yaw and Variable_Hydro among them.
5. Comparison with Alternatives
Within time-domain simulation of marine floating structures, WEC-Sim occupies a clear position with three reference points.
The first is commercial hydrodynamics software. Commercial boundary element solvers such as WAMIT and Aqwa cover more geometry and operating conditions, yet they do not provide coupled multibody, power take-off and control simulation, nor scriptable batch execution. The role of WEC-Sim is to take their output and build a complete device model on top; the v7.1.0 notes update the BEMIO examples and add capture of NaN input data from Capytaine, confirming that this division of labour, boundary element solver upstream and WEC-Sim downstream, is the officially recognised arrangement.
The second is peer open-source projects. Floating wind has OpenFAST, volumetric fluid simulation has OpenFOAM, both with comparable maturity and community size in their own directions, yet the three face different physics: OpenFAST addresses wind turbine wake and blade aeroelasticity, OpenFOAM addresses volumetric CFD, and WEC-Sim addresses frequency-domain potential-flow hydrodynamics with device-level multibody dynamics. Notably, the WEC-Sim ecosystem absorbs floating wind capability internally through MOST. The official MOST manual explains that the extension allows simulation of floating offshore wind turbines and hybrid platforms, adding the classes windClass.m and windTurbineClass.m together with the MOST_Lib.slx library, and modifying initializeWecSim, mooringClass and postProcessWecSim to emit turbine power, rotor speed, blade pitch and nacelle acceleration histories.
The third is simplified models and experimental data. The potential-flow framework suits device-level conceptual design and parameter sweeps, while free-surface breaking, spray and strongly non-linear mooring-float coupling still require CFD or tank testing.
6. Benchmarks and Evidence
The credibility of WEC-Sim rests on a documented verification and review chain, and that chain is directly citable.
The engineering verification starts with the OMAE 2014 paper on preliminary verification and validation of WEC-Sim by authors from NREL and Sandia, followed in 2016 by the Phase 1 validation testing paper on numerical modelling of experiments against international tank test models.
Periodic review is carried by the International Marine Energy Journal. The 2020 paper on introducing non-rigid body structural dynamics records how that capability entered the code. The 2022 review of WEC-Sim development and applications, volume 5 issue 3, pages 293 to 303, sorts the literature into feature development, experimental validation, device modelling, control modelling, PTO and grid modelling, and novel applications. The September 2025 review of recent WEC-Sim advanced features applies the same treatment to the v6.1 feature set.
The version history leaves a clear functional trail. Version 4.2 added normal and tangential Morison options, drag bodies, parallel execution of WEC-Sim and end stops on PTOs. Version 4.3 added running directly from Simulink, reading Capytaine output, a more accurate infinite-frequency added mass calculation, and restructured the documentation into four manuals covering introduction, theory, user and development. Version 4.4 added library blocks for cables, spherical constraints and spherical PTOs, a variable time-step solver fix and multi-version documentation. Version 5.0 completed the refactoring of classes and properties, refactored the BEMIO functions and tests, introduced the new PTO-Sim library and warning and error flag mechanisms. Version 6.0 introduced property validation for WEC-Sim objects, the large horizontal displacement option, products of inertia handling and pressure integration for mean drift. Version 6.1 enabled MoorDyn v2 capabilities, the PTO extension feature and NEMOH v3.0 compatibility. Version 7.0.0 is the largest rework, with official counts of 68 issues closed and 52 pull requests merged, concentrating on fully directional spectra, variable hydrodynamics, variable mass, a universal PTO joint, the QTF standing wave Newman approximation, non-linear hydrodynamics for flexible bodies and multiple wave trains. Version 7.1.0 continues with IEC-standard handling of full directional waves, a deep water flag fix, phase calculation for large horizontal displacements and variable hydrodynamics combined with directional waves.
Releases are archived on Zenodo under concept DOI 10.5281/zenodo.3924764, with the v7.0.0 record at 10.5281/zenodo.16950906 listing the resource type as software, the license as Apache-2.0, and 31 credited authors spanning NREL, Sandia and a large number of external contributors.
7. Role in Physics-AI Workflows
For physical-AI practitioners, the value of WEC-Sim lies less in its own computational power than in the fact that it is a parameterised, cleanly structured, batch-runnable solver that works well as a data source and a validation environment.
Generating surrogate training data is the first direct use. The inputs are structured, comprising geometry and mass, mooring parameters, PTO coefficients, spectral parameters, simulation duration and step size; the outputs are equally structured time series covering six-degree-of-freedom displacement and velocity, excitation forces, and PTO power and torque. A form where inputs can be enumerated and outputs can be aligned suits parameter sweeps followed by reduced-order model training or neural operator construction, and the Free_Decay, Multiple_Condition_Runs and Variable_Hydro directories in the official applications repository provide exactly that infrastructure.
Verifying reinforcement learning controllers is the second use. WEC-Sim supports embedding control loops directly in Simulink, and the Controls, Load_Mitigating_Controls and WECCCOMP application directories correspond to general control, load-mitigating control and control competition cases. A controller trained in simulation is validated against a physical response that includes mooring and free-surface effects.
Exploring the engineering design space is the third. Batch simulation yields extreme response statistics for reliability assessment and preliminary fatigue screening, a workflow structurally identical to surrogate modelling in parametric finite element analysis.
Data-driven modelling itself has public precedents. The June 2023 arXiv paper on data-driven modelling of an oscillating surge wave energy converter uses WEC-Sim as the data generator and extracts low-order modes from time-domain responses using dynamic mode decomposition, and the December 2020 arXiv paper on practical optimal control addresses optimal control in a regular wave environment.
Placed in the wider physical-AI landscape, WEC-Sim acts as a ground-truth generator, in the same family as OpenFOAM multiphysics on the fluid side, bridge structural analysis on the structural side and flow control environments on the control side. The surrogate modelling and reduced-order collection and the numerical methods and solvers collection provide cross-domain methodological references.
8. Limitations and Failure Modes
Several situations produce untrustworthy results and should be avoided explicitly.
Cases that need non-linear free-surface effects fall outside the framework. Potential flow does not resolve breaking, spray, wave setup or internal waves, and steepening of gravity waves lies beyond its capability.
Cases that need full volumetric flow information are also out of scope. When the research target concerns internal flow detail, local effects of a drag body or the physical origin of added mass, frequency-domain coefficients give results without mechanism, and a transition to CFD is required.
Cases needing precise capture of fluid damping need care. Viscous resistance is approximated by linear damping and quadratic drag coefficients whose values require experimental calibration, and uncalibrated values directly alter the system damping ratio and resonant response.
Very long horizons and full life-cycle assessment are out of scope as well. WEC-Sim is a time-domain simulation tool, and multi-decade fatigue life assessment requires statistical extrapolation over the design domain rather than running the simulation for the whole life.
Inconsistent coefficient provenance deserves the most attention. Different boundary element solvers can differ in degree-of-freedom ordering, phase convention and reference point, and manually stitching coefficients from multiple sources without a consistency check introduces deviations that are hard to localise. The official repository provides combineBEM, normalizeBEM and cleanBEM to handle such issues, but they cannot substitute for physical judgement by the user.
Software maturity is itself a risk. Versions 7.0.0 and 7.1.0 continue to ship defect fixes, including deep water flag, large displacement phase calculation and variable hydrodynamics with directional waves, and new features and defect fixes often land in the same batch, which indicates that interfaces are still evolving. Reproducing historical results across versions requires pinning both the version and the coefficient files.
9. Version and Source Notes
The verification and provenance notes below record the exact basis of this entry.
Version status comes from the GitHub Releases list, which holds twelve releases: the latest is v7.1.0 dated 30 June 2026, preceded by v7.0.0 on 26 August 2025, v6.1.2 on 14 January 2025, v6.1.1 on 20 September 2024, v6.1 on 16 September 2024, v6.0 on 20 October 2023, v5.0.1 on 28 September 2022, v5.0 on 16 May 2022, v4.4 on 28 October 2021, v4.3 on 22 July 2021, v4.2 on 23 December 2020 and v4.1 on 1 July 2020. The default branch is main, development proceeds on dev through a forking workflow, new feature pull requests target dev, bug fix pull requests target main, and a release merges dev into main and tags the result.
Licensing and attribution follow the COPYRIGHT file in the repository root, which states copyright 2014 Sandia Corporation and NTESS, retains certain rights for the U.S. Government under contract DE-NA0003525, and applies the Apache License 2.0; the LICENSE file is the standard Apache-2.0 text. The GitHub API identifies the repository license as Apache-2.0, consistent with the files.
Archival citation uses the Zenodo concept DOI 10.5281/zenodo.3924764, with the v7.0.0 version record at DOI 10.5281/zenodo.16950906; earlier versions include 10.5281/zenodo.7121186 for v5.0.1 and 10.5281/zenodo.3924765 for v4.1.
Software and data are obtained from https://github.com/WEC-Sim/WEC-Sim , the documentation site is https://wec-sim.github.io/WEC-Sim , and the companion applications repository is https://github.com/WEC-Sim/WEC-Sim_Applications .
The WEC-Sim GitHub organisation publishes nine public repositories: besides the main repository these include bemio, WEC-Sim_Applications, WDRT, MoorDyn under its own GPL-3.0 license as an independent mooring solver, WECCCOMP, FOSWEC, MECC and pioneerWEC.
A boundary on provenance deserves stating. Applications and tutorial cases outside the main repository are maintained separately, so citing an example should name the specific applications repository and directory. Mooring capability depends on the external MoorDyn project, whose version and license are independent of the WEC-Sim main repository.
10. References
- WEC-Sim official repository and documentation, https://github.com/WEC-Sim/WEC-Sim , accessed 11 November 2026.
- WEC-Sim theory manual, docs/theory/theory.rst in the repository, covering coordinate system, unit convention, motion equation, convolution integral and state-space realisation derivations.
- WEC-Sim getting started guide, docs/user/getting_started.rst, listing MATLAB dependencies and the four-step installation procedure.
- WEC-Sim releases page, https://github.com/WEC-Sim/WEC-Sim/releases , carrying release notes from v4.1 through v7.1.0.
- Ruehl et al., Preliminary Verification and Validation of WEC-Sim, an Open-Source Wave Energy Converter Design Tool, OMAE 2014, DOI 10.1115/omae2014-24312.
- Ruehl et al., Beyond Wave Energy: Applications of the WEC-Sim Open-Source Software, OMAE 2025, DOI 10.1115/omae2025-157631.
- Ogden et al., Review of WEC-Sim development and applications, International Marine Energy Journal, volume 5 issue 3, pages 293 to 303, 2022, DOI 10.36688/imej.5.293-303.
- Grasberger et al., Review of recent WEC-Sim advanced features, International Marine Energy Journal, 2025, DOI 10.36688/imej.8.259-263.
- Scriven et al., Introducing non-rigid body structural dynamics to WEC-Sim, International Marine Energy Journal, 2020, DOI 10.36688/imej.3.55-63.
- Tom et al., New Developments and Capabilities Within WEC-Sim, OMAE 2023, DOI 10.1115/omae2023-105030.
- So et al., Development of PTO-Sim for the Open-Source Wave Energy Converter Code WEC-Sim, OMAE 2015, DOI 10.1115/omae2015-42074.
- Carrelhas et al., WEC-Sim wave-to-wire model of a floating oscillating-water-column wave energy converter, Ocean Engineering, 2026, DOI 10.1016/j.oceaneng.2025.123893.
- Zenodo concept archive, WEC-Sim v7.0.0, DOI 10.5281/zenodo.16950906, concept DOI 10.5281/zenodo.3924764.
- WEC-Sim official MOST manual, docs/most, covering floating wind turbine and hybrid platform coupling.
- WEC-Sim_Applications repository, https://github.com/WEC-Sim/WEC-Sim_Applications , containing more than thirty advanced feature application directories.
- arXiv 2306.03861, Data-driven modeling of an oscillating surge wave energy converter using dynamic mode decomposition, 2023.
- arXiv 2012.08049, Practical Optimal Control of a Wave-Energy Converter in Regular Wave Environments, 2020.
- IEC 62600-2, Marine energy - Wave, tidal and other water current converters, marine energy standard series.
- Cummins, The Impulse Response Function and Ship Motions, David Taylor Model Basin report, 1962.
- Li and Yu, A Synthesis of Numerical Methods for Modeling Wave Energy Converter-Point Absorbers, Renewable and Sustainable Energy Reviews, 2012.