SEA-Stack

Cross-fidelity simulation software for marine energy technology development

OREC 2026 Workshop

Hosted by NLR — National Lab of the Rockies

David Ogden (Simocean) Allison Riley (NLR · Mines) Salman Husain (NLR) Stephen Chamot (NLR)

SEA-Stack partners: Simocean · UW–Madison · Colorado School of Mines (Mines)

What this session is about

Goals — the workflow problems we're trying to solve for marine energy (wave, tidal, and beyond).
Our approach — how SEA-Stack is put together, what is in the current release, and example studies.
How you use it — one system definition (YAML, hydro data, and related inputs) for interactive runs, campaigns, and planned development.
Your input — in breakout: would this be useful for your work? What would help or stop you using it? What capabilities, tools, or benchmarks matter?

How today is structured (120 minutes)

35 min — SEA-Stack software

motivation for building SEA-Stack
current capabilities
workflows, TALOS example, collaborator validation
what we are focusing on next

35 min — HIL demo

Hardware-in-the-loop: coupling the model to real-time control and test hardware.

50 min — breakout

small groups; all four topics
~10 min per topic (flexible)
10 min wrap-up in the room

Breakout topics

In breakout, your group works through all four topics (~10 minutes each). We return to this slide before you start.

1. Problem & priorities

fidelity and physics gaps

Are we focused on the right problems for your field (wave, tidal, other)? What physics or fidelity gaps matter for devices like yours?

2. Using the software

barriers to use

Would this fit your work — what would help or stop you? After the HIL demo: does that workflow fit your control or test setup? Docs, examples, install, training?

3. Tools & interoperability

workflows, formats, integrations

Which tools, formats, and integrations matter in your workflow? What capabilities are still missing for the studies you actually run?

4. Evidence & field data

validation and operational data

What benchmarks or verification would you need before relying on results? What field data, tests, or reports would help?

Constructive criticism is welcome — a clear “not for us because…” is as helpful as concrete suggestions for what would work.

Motivation for Building SEA-Stack

Wide variety of WEC concepts under exploration

Each concept sits in a huge design space. We need effective tools to explore that space and compare variants before committing to detailed studies.

High fidelity for loads and survivability

https://www.youtube.com/watch?v=ARIg3oQtZ0w

Impulsive, strongly nonlinear loading — often in operational sea states, not only in storms.

Potential-flow and other low-fidelity models generally cannot resolve this loading physics directly. They are still essential for broad design exploration.

For slamming, impact loads, and similar events you need targeted high-fidelity modelling (CFD, SPH, and related tools) where it matters.

What the workflow needs to support

Unusual concepts

rigid / flexible systems
novel PTOs
closed loops
unusual topologies

Broad exploration

sweeps
optimization
many sea states
parametric studies

Loads and survivability

impulsive loads
survivability
nonlinear events
targeted high fidelity

Simulation software should not constrain design choices or exploration of new concepts.

One integrated workflow vs siloed tools

How SEA-Stack is built

A WEC model has to keep everything coupled

A useful WEC model carries hydrodynamics, structures, mooring, PTO, and control together — coupled, not as independent pieces.

Requirements and how we addressed them

What we wanted	How we built it	What it uses
Explore unusual WEC concepts	Flexible multibody foundation	Project Chrono — native C++ multibody, constraints, joints, contact, FEA-capable host
Short turnaround in interactive use	Interactive visual workflows	VSG GUI + packaged demos
Run large campaigns and optimization studies	Headless scalable workflows	YAML / CLI / HDF5 — reproducible study automation and batch I/O; native C++ underneath
Carry the same system definition across fidelities	Solver-independent domain models	Thin adapter layers — same YAML/device definition when hydro engine changes
Extend physics over time	Modular explicit interfaces	Chrono-free domain libraries
Couple to external solvers	Shared C++ interfaces	Chrono · MoorDyn · OpenFOAM · DualSPHysics — same codebase for thin adapters

Architecture — domain, adapters, and external solvers

Model definition — YAML, assets, outputs

example.setup.yaml

setup.yaml wires the case together

model_file: example.model.yaml
simulation_file: example.simulation.yaml
hydro_file: example.hydro.yaml

output_directory: outputs

example.model.yaml

One segment + one universal joint (excerpts) · TSDAs omitted on slide

chrono-version: 10.0

model:
  name: "example_model"
  angle_degrees: false

  data_path:
    type: RELATIVE
    root: "."

  bodies:
    # Segment 1 (Nose)
    - name: body1
      location: [18.0, 0.0, -1.8]
      mass: 438293
      fixed: false
      inertia:
        moments: [876585.0, 47773884.0, 47773884.0]

  joints:
    - name: joint_12
      type: UNIVERSAL
      body1: body1
      body2: body2
      location: [36.0, 0.0, -1.8]
      axis1: [0, 1, 0]
      axis2: [0, 0, 1]

example.hydro.yaml

Waves, spreading, excitation, MoorDyn · H5 from shared BEM data

hydrodynamics:
  bodies:
    - name: body1
      h5_file: ../assets/hydroData/example_directional.h5

  waves:
    type: irregular
    height: 3.0
    period: 10.0
    spectrum: jonswap
    spreading:
      type: cos2s
      s: 12

  moordyn:
    enabled: true
    input_file: mooring/lines_example.txt
    bodies: [body1, body3]

example.simulation.yaml

Chrono time step, integrator, linear solver

chrono-version: 10.0

simulation:
  time_step: 0.02
  end_time: 600.0
  integrator:
    type: HHT
    use_stepsize_control: true
  solver:
    type: SPARSE_LU

Same definition, multiple solver paths

Workflows and examples

Workflows — current and planned

Interactive and headless workflows are in the current release. Higher-fidelity coupling is under development.

Current

Interactive

Try layouts, debug topology, watch coupled motion.

Current

Headless

Sweeps, campaigns, optimization — from a CLI or script.

In development

Higher fidelity

CFD / SPH for nonlinear loads and survivability — same model definition, when the coupling is ready.

Interactive — does the concept behave?

Quick qualitative checks before you commit to a big parametric run.

New topologies, joints, mooring, sea states
Watch the motion, debug, ask “does this look right?”

Five-segment attenuator — directional spreading, MoorDyn (illustrative run)

5sa walkthrough — workflow overview

Generate
mesh

→

Run
BEM

→

Write
YAML files

→

Run
SEA-Stack

→

View
outputs

Pelamis P2 at EMEC (Orkney) — image courtesy of EMEC.

The 5sa demo is loosely modelled on a five-segment attenuator like this. Same five-step chain top to bottom — we will step through it next.

5sa walkthrough — geometry exploration

Generate
mesh

→

Run
BEM

→

Write
YAML files

→

Run
SEA-Stack

→

View
outputs

Increase body diameter 3 m → 5 m, regenerate the five-body mesh → written into 5sa/assets.

Capytaine

BEM solve on the updated mesh — added mass, radiation damping, and excitation written to an H5 file in 5sa/assets.

hydro.yaml

WriteYamlModel_5sa

→

model.yaml

sim.yaml

setup.yaml

model.yaml is regenerated from the new H5 hydro data by WriteYamlModel_5sa. The other YAML files are unchanged for this sweep.

5sa running in SEA-Stack — irregular waves, MoorDyn mooring, line tensions coloured by load.

Absorbed power across diameter sweep 3-5 m

Absorbed power across the diameter sweep (3.0 → 5.0 m) and per-body heave — read via h5outputsReader.

TSDA reaction forces on body 1 and MoorDyn line tensions.

Mean absorbed power and total absorbed energy over a sea-state grid (Hs vs Tp).

Example cases included with the release

RM3 with a rectified hydraulic PTO + PI controller.

Trimaran with Euler-beam FEA cross-arms.

How much PTO or structural detail you need depends on the device and the question — these show what is possible.

Headless — sweeps, optimization, campaigns

Same system definition, run from a CLI. Sweep, optimize, or batch many sea states without rebuilding the model for each design point.

TALOS — exploring thousands of designs

The TALOS project explored thousands of WEC design variants in SEA-Stack. The figure shows one slice — a taper-draft sweep at regular-wave periods, with six-DOF response and capture width. What design exploration or optimization studies matter for your work — and what would you need a tool to support?

Validation with Strathclyde, Oxford, and PNNL

Strathclyde · Oxford — VLFP

PNNL — WITT

Building models in SEA-Stack and validating against tank-test and benchmark data — validation reports in preparation.

Higher fidelity — direction we're building (CFD / SPH)

The goal: couple to CFD / SPH for slamming, survivability, or complex free-surface behaviour — without redefining the device. This path is in development, not a routine workflow yet.

https://www.youtube.com/watch?v=ebtOX1pSci4

Next steps and discussion

Current capabilities

System dynamics

rigid multibody hydrodynamics
wide range of joints and constraints
closed kinematic loops
dry FEA bodies / flexible members

Hydrodynamics and waves

first-order potential-flow forces
radiation damping via convolution or state-space
regular, irregular, directional, and bimodal seas
nonlinear hydrostatics from submerged volume

Subsystems and workflows

MoorDyn coupling
linear and hydraulic PTO components
PI control and scalar control interfaces
YAML / CLI / HDF5 campaign workflows

Each public release includes verification reports. What benchmark or validation cases would you want included?

What we are working on next

Following the current release — planned development priorities.

Simulation is only useful if it connects to real hardware. HIL testing is one of those bridges — over to the next section.