Computational fluid dynamics
Solver selection
We use STAR-CCM+ for production resistance and propulsion work where mesh control, overset grids, and free-surface VOF are the priority. We use OpenFOAM where the model has to be open, reproducible, and shipped to the client at the end of the project. We use WAMIT for linear potential-flow seakeeping and offshore-structure diffraction-radiation. The choice is project-driven, not preference-driven.
Meshing standards
Hull meshes use a wall-normal first-cell y+ in the range 30–100 for wall-function turbulence treatment, or below 1 for low-Reynolds models on appendages and propellers. Free-surface refinement is targeted at 80–120 cells per wavelength in the wave-making region. Refinement zones are defined parametrically — a coarse, medium, and fine mesh are run on every project so we have a documented mesh-independence assessment in the report. Cell-count is whatever the physics requires; we do not run small meshes for speed when the result is the deliverable.
Turbulence and physics
Default turbulence model: k-omega SST for resistance and external aero. k-epsilon realizable when modelling internal flows or where SST has known weaknesses. Free-surface treatment: VOF with HRIC for sharp interfaces; level-set when the interface tangle becomes significant. Compressibility is treated explicitly when Mach number exceeds 0.3, which in the marine context means propeller tip vortices and air entrainment studies — almost never the global hull problem.
Validation
Every CFD report includes a validation section. Where towing-tank data exists for the vessel or a near-sister, we calibrate against it and report the residual. Where it does not, we calibrate against ITTC benchmark hulls (KCS, KVLCC2, Wigley) at comparable Froude numbers and report the residual against published data. A study with no validation reference is flagged as such; we do not bury that fact in the executive summary.
Finite element analysis
Software and rule sets
SESAM (DNV) is our default for offshore structures and hull-girder analyses; Ansys is used where coupled phenomena (thermal, fluid-structure interaction) dominate. Class rule sets in active use: DNV-RP-C203 for fatigue, DNV-OS-C101 for offshore steel structures, and the Common Structural Rules (CSR) for tankers and bulkers in IACS class.
Mesh and element selection
Shell elements (4-node quadrilateral) for plating and webs in global models; 20-node hexahedral solids for hot-spot stress analysis at fatigue-critical details. Mesh refinement at hot-spots follows DNV-RP-C203 conventions: t × t element size at the hot spot, with surface-extrapolated stress reported. Coarser mesh sizes are acceptable for global response provided the local hot-spot mesh exists and is referenced in the report.
Load cases
We start from the rule load cases the relevant class society prescribes, then add project-specific cases — a transit weather case for a delivery voyage, a site-specific 100-year storm for an offshore deployment, a damage-stability case with the corresponding global-bending response. Load-case lists are agreed in the technical proposal and locked at kick-off; mid-project additions go through a change order with named consequences.
Fatigue
Rainflow counting on stress histories from time-domain seakeeping; S-N curves per DNV-RP-C203 (D-curve for as-welded details, others where applicable); Miner sum to demonstrate fatigue lives meet or exceed the design life with the relevant safety factor. Where fatigue is a binding constraint we report damage per detail rather than just an overall pass/fail.
Mooring analysis
Workflow
Quasi-static screening in ARIANE-3D across candidate configurations — spread, taut-leg, single-point — before committing to a design. The screening uses simplified vessel hydrodynamics and the metocean intensity curves; output is a candidate-shortlist with line tensions and footprint envelopes. The shortlisted configuration then goes into time-domain analysis in OrcaFlex or MOSES with full 6-DoF vessel hydrodynamics and line dynamics.
Design environment
Operational, survival, and ULS / ALS conditions per DNV-OS-E301 or API RP 2SK, depending on the rule set the operator commits to. Metocean inputs come from site-specific hindcast plus measurements where available; uncertainty bands are propagated through to line-tension envelopes rather than absorbed into a safety factor.
Anchor design
Anchor type follows from soil characterisation and required holding capacity — drag embedment for soft cohesive soils, suction caissons for deep cohesive layers, gravity or pile anchors for hard sites. Holding-capacity calculations follow API RP 2SK or DNV-OS-E301. Installation analysis is included in the deliverable when a contractor has not been pre-selected.
Time-domain dynamics
Linear core
WAMIT for the linear seakeeping problem — added mass, radiation damping, exciting forces, motion RAOs. Mesh density on the wetted hull at 5–8 panels per shortest wavelength of interest. Output is RAOs per heading and frequency, validated against published benchmarks where they exist (Wigley, S-175, SR108) and against tank data where the project commissions or accesses it.
Non-linear time domain
Where roll motion is large, where parametric excitation is suspected, where the linear assumption breaks — we move to Ship@Sea, our in-house six-degree-of-freedom non-linear simulator that has evolved from the principal\'s 2005 PhD work and continues to be maintained and validated [Silva and Guedes Soares, 2005, 2013]. It couples non-linear restoring (instantaneous wetted surface integration), Froude-Krylov forces on the exact-wetted surface, and roll damping from a viscous correction calibrated against decay tests. It is our reference tool for parametric rolling and for any seakeeping result where roll exceeds 5–8°.
Manoeuvring
Manoeuvring is run in MatLab/Simulink, using a hydrodynamic-derivative model calibrated against CFD or sea-trial data. Standard manoeuvres (turning circle, zigzag, spiral, crash-stop) are validated against IMO Resolution MSC.137(76) acceptance criteria. Probabilistic operability assessments (e.g. the Corvo and Silver Mary studies in Vila do Porto) wrap the deterministic simulator inside a metocean joint-probability distribution to estimate days-per-year operability with credible intervals — see the Corvo case note for a worked example.
Write everything down
Every assumption, every input, every result is documented in the project file. Six months later, when the operator has a new question, we can answer it because the model is still alive and the assumptions are findable. This is one of our four founding principles, and it is the discipline that distinguishes consulting work that ages from work that decays.
Reports follow a standard structure: scope, methodology, inputs, validation, results, sensitivity, limitations, conclusions. Every numerical claim has a traceable run reference; every figure has a caption with units and conditions; every assumption that affects the result is named in the limitations section, not buried.
How we work with class
Class engagement starts at the technical proposal: we name the rule set, the expected approval path, and the known interpretation risks. During design we hold regular working sessions with the class reviewer rather than waiting for formal comments — most disagreements are easier to resolve in conversation. During construction we attend the surveys; during sea trials we witness against the class criteria, not just against the contract. Active liaison files: DNV, Lloyd\'s Register, Bureau Veritas, RINA.