Marine and subsea

Four capabilities for additively-manufactured subsea parts: hydrostatic load cases, pressure-hull collapse margin, seawater corrosion, and an advisory qualification dossier.

ADVISORY ONLY

souxmar is not a classification society, a notified body, or a certifying authority. Nothing here approves anything, and running it qualifies nothing.

The collapse pressures are preliminary sizing arithmetic from closed-form shell formulae with blanket knockdown factors. The corrosion numbers are indicative literature values chosen so alloys can be compared — not design allowances. The qualification report is a checklist derived from publicly documented AM qualification practice, not a reproduction of anyone's rule and not a demonstration of compliance with it.

A hull, a propeller, or a pressure-boundary fitting is approved against a class rule or a naval standard, with the surveyor's own formulae and measured weld, out-of-roundness and material data.

The capabilities

Capability id	Kind	Gives you
solver.marine.hydrostatic	solver	Nodal hydrostatic_pressure (Pa), one step per depth factor
solver.marine.hull_collapse	solver	Per-cell [collapse_pressure_Pa, margin, governing_mode_code]
solver.marine.corrosion	solver	Per-cell [thickness_loss_mm, pitting_risk, galvanic_risk]
writer.marine.qualification_report	writer	Markdown advisory dossier

Depth and sign conventions

build_direction is the "up" axis (default [0, 0, 1]) — the same key the AM capabilities use, because a subsea AM part is analysed in the frame it was built in.

The mesh's top sits at design_depth × factor; every other node sits deeper by its own distance below that node:

p = rho · g · (depth_of_top + (z_top − z_node))  [+ p_atm]

Depth is positive downward. Pressure is a positive magnitude, compressive on the wetted surface — gauge unless include_atmospheric is set. The solver does not know which faces are wetted: every node is loaded, and the consumer applies it as a follower load along the inward normal. A sign error here is silent, so check the deformed shape.

Load cases arrive as field time steps, one per depth_factors entry. The default [1.0, 1.5, 2.25] is operating / test / collapse.

Hull collapse

  - id: collapse
    plugin: solver.marine.hull_collapse
    input:
      mesh: { from: ring }
      hull_type: ring_stiffened_cylinder   # or cylinder | sphere
      diameter: 1.0            # OUTER diameter, m
      thickness: 0.012
      unsupported_length: 0.5  # frame spacing
      youngs_modulus: 1.9e11
      poisson_ratio: 0.28
      yield_strength: 5.0e8
      design_depth: 300.0
      seawater_density: 1025.0
      imperfection_knockdown: 0.75
      am_anisotropy_knockdown: 0.90
      safety_factor: 1.5

Four modes are evaluated and the minimum after knockdown governs:

Code	Mode	Closed form
0	Interframe shell instability	Windenburg–Trilling p = 2.42·E·(t/D)^2.5 / [(1−ν²)^0.75·(L/D − 0.45·√(t/D))]
1	Membrane yield	p = 2·σy·t/D (cylinder), 4·σy·t/D (sphere)
2	General (long-cylinder) instability	p = 2·E·t³ / [(1−ν²)·D³] — unstiffened cylinder only
3	Sphere elastic buckling	p = 2·E·(t/R)² / √(3(1−ν²))

Then margin = p_collapse / (p_design · safety_factor), with p_design = rho·g·design_depth as gauge.

General instability of a ring-stiffened hull is not evaluated. It needs frame area, frame inertia and bulkhead spacing, which this capability's inputs do not carry — and it is frequently the governing mode for a real framed hull. A ring_stiffened_cylinder result that reports mode 0 or 1 is not telling you the frames are adequate; it never looked.

The result is uniform over the mesh: an analytical answer keyed to the input geometry, not a mesh-resolved one. Refining the mesh will not change it.

Corrosion, PREN and galvanic risk

  - id: corrosion
    plugin: solver.marine.corrosion
    input:
      mesh: { from: ring }
      alloy: "316L"            # 2507 | NAB | Ti6Al4V | IN625 | AlSi10Mg | CuNi90-10
      mating_alloy: "Ti6Al4V"  # noble mate: the part is the anode
      area_ratio_cathode_anode: 4.0
      seawater_temperature: 10.0
      flow_velocity: 0.1       # near-stagnant
      service_life_years: 25
      cathodic_protection: true
      as_built_surface: true

thickness_loss_mm sums a uniform term (rate scaled by temperature, oxygen, salinity, velocity and surface condition), an oxygen-limited galvanic term (only the anodic member loses metal), and a sqrt(time) pitting term weighted by the pitting risk. A coating scales the first two by 1 − coating_efficiency; cathodic protection scales them to a small residual, never zero.

Pitting risk for the stainless and Ni-Cr-Mo alloys keys off

PREN = %Cr + 3.3·%Mo + 16·%N
CPT  = 2.24·PREN − 47.1   [°C]

which is why 316L is marginal and 2507 is not:

Alloy	PREN	Critical pitting temperature
316L	≈ 25.5	≈ 10 °C
2507 super-duplex	≈ 41.9	≈ 47 °C

316L's critical temperature lands inside the range of ordinary seawater — and stagnant water drops the effective threshold by up to 20 K more, because deposits and biofilm make crevices. That is why 316L subsea hardware is normally cathodically protected, coated, kept in flowing water, or accepted with an inspection regime. 2507 sits about 35 K clear of any natural seawater temperature, which is why it is the default for permanently wetted unprotected stainless hardware.

Galvanic risk scales with the driving force and the square root of the cathode/anode area ratio. Area ratio dominates: a small stainless fastener in a large bronze plate is a nuisance, a small bronze fitting in a large stainless structure is a wear part. Note that carbon-fibre composite is noble to every metal in the series — a metal fitting bolted into a CFRP AUV structure is the anode.

Material shortlist

Material	Use it for	Watch out for
NAB (nickel-aluminium bronze)	Propellers, impellers, pump/valve bodies, tidal blades	Erosion-corrosion above ~4.3 m/s; anodic to stainless
316L	General subsea hardware, brackets, housings	Pitting/crevice — marginal unprotected in warm or stagnant water
2507	Permanently wetted unprotected hardware, pressure-boundary fittings	AM phase balance and nitrogen retention must be demonstrated
Ti-6Al-4V	Deep hulls and housings, mass-critical parts	Noble — everything attached corrodes; hydrogen embrittlement of the cathode
CuNi 90-10	Seawater piping, heat-exchanger parts, spools	Flow-limited (~3.5 m/s); needs a corrosion allowance
Alloy 625	High-integrity and hot-service subsea parts	Cost; noble to almost everything
AlSi10Mg	Topside and dry internals only	Not for permanent immersion
PEKK / PA12-CF	AUV fairings, flooded structures, connector bodies	Z-strength is the design driver; CF fill is noble to metals

Qualification dossier

  - id: dossier
    plugin: writer.marine.qualification_report
    input:
      mesh:  { from: ring }
      field: { from: collapse }
      path:  hull-ring-qualification.md
      part_name: "Pressure-hull ring frame"
      process: lpbf
      alloy: "316L"
      application: hull        # | propulsion | piping | structural | non_structural
      criticality: 1           # 1 = HIGHEST consequence of failure
      design_depth: 300.0
      service_life_years: 25

It renders part identification, a criticality rationale, the evidence checklist (feedstock traceability, machine qualification, witness coupons, NDT method and coverage, mechanical test matrix, build-direction property declaration, post-processing / HIP record), a summary of whichever field it was handed, and the advisory-only disclaimer.

No certificate number, stamp, surveyor name, approval reference or test result is generated — every such field is left blank for the responsible engineer. The traceability digest is FNV-1a 64-bit: a non-cryptographic content hash, not a signature.

From chat

apply_hydrostatic_load    stage a depth-derived pressure load case  (prompts once)
check_marine_integrity    corrosion / galvanic / collapse-margin summary

Next

• Additive manufacturing
• docs/MARINE.md — full conventions, seawater properties, every input key, and "What would make this class-credible": the honest gap list between this block and a class submission.