CNC processing at the seabed: how subsea hardware is made to survive

• 1
  Key takeaways
• 2
  What makes CNC processing at the seabed different
• 3
  Pressure, wall thickness and the shapes that survive
• 4
  Alloy choice for seawater: what we machine and why
• 5
  Five-axis setups and tolerance stack-up on long parts
• 6
  Surface finish, passivation and how we verify a subsea part
• 7
  Alloy and process choices for subsea parts

The seafloor is not one environment. A manifold sitting at 500 m sees roughly 5 MPa and 4 °C. The same geometry at 3,000 m sees about 30 MPa, and that pressure does not arrive evenly. It arrives as a cyclic load every time the valve strokes. CNC processing at the seabed starts with reading those numbers before touching a toolpath.

Two failure modes drive most of the design. The first is collapse or buckling of a hollow body under external pressure. The second is crevice corrosion in a flange face where two stainless parts sit in stagnant seawater. The first is a stress problem. The second is an alloy and surface problem.

Subsea parts are also hard to service. Pulling a component back to the surface can cost more than the part itself. That pushes designers toward heavier walls, redundant seals and alloys with a known corrosion history. It also pushes them toward machining, because a forged body with a machined sealing face is easier to certify than a welded assembly.

One more thing separates subsea work from ordinary marine work: documentation. A part may be sound, but if the in-process records are missing, the operator will not install it.

• 1
  Depth sets pressureRoughly 1 MPa per 100 m of seawater.
• 2
  Chloride sets alloy316L and 17-4PH are the common starting points.
• 3
  Service access sets costEvery design choice is weighed against retrieval.

External pressure pushes a hollow housing toward ovality before it pushes it toward yield. Short, thick cylinders resist this well. Long, thin ones do not. When a housing has to be long, designers add external ribs or step the outer diameter, and both of those features have to be machined rather than formed.

Sealing faces are the second constraint. A face seal needs flatness, not just a tight diameter. On a Ø200 mm flange, a few micrometres of waviness across the face can open a leak path once the joint is pressurized from outside. Face flatness is normally called out separately from the dimensional tolerance for this reason.

Threaded ports are a common weak point. A deep tapped hole in a 316L body removes material exactly where the wall is already thin. We usually suggest a shallower port with a spot-faced seat, or a welded-in boss that is machined after welding so the thread stays concentric.

Internal bores that carry hydraulic fluid add a third constraint. A bore that is straight within 0.02 mm over 500 mm will hold a spool valve clearance. One that wanders will not, no matter how good the surface finish is.

• 1
  Avoid long unsupported boresRibs or stepped OD keep ovality in check.
• 2
  Call out face flatnessFlatness is not covered by a diameter tolerance.
• 3
  Watch thread depthDeep ports thin an already loaded wall.

Duplex and super duplex stainless steels are the default for wetted structural parts. They resist chloride stress corrosion cracking better than 316L, but they work-harden quickly, so cutting speeds stay low and tool wear is high. If a print calls for duplex, expect longer cycle times and budget for it.

17-4PH (SUS630) is common for stems, shafts and valve trim. It machines cleanly in the annealed condition and then ages to a high strength. The catch is that aging distorts thin sections. If a shaft has a 0.01 mm concentricity callout, it should be semi-finished, aged, then finished.

Titanium TC4 (Ti-6Al-4V) shows up where weight matters, such as ROV frames. It has poor thermal conductivity, so heat stays in the cut. We run it with high-pressure coolant and lighter radial engagement. Inconel appears in high-temperature wellhead hardware and is slower still.

Aluminium 6061-T6 and 6082 are fine for non-wetted housings and electronics enclosures, but they need anodizing or an isolation barrier before they go near seawater. Bare aluminium next to stainless steel in seawater is a galvanic cell waiting to happen.

• 1
  Duplex: strong, slowWork-hardening raises tool cost and cycle time.
• 2
  17-4PH: age after roughingAging moves thin sections.
• 3
  Titanium: manage the heatHigh-pressure coolant, light radial cuts.
• 4
  Aluminium: isolate itNever couple bare aluminium to stainless.

A subsea housing often has bores on more than one face plus a sealing flange that has to sit square to all of them. On a three-axis machine that means three or four setups, and every setup adds a datum shift. On a simultaneous five-axis center it can be one setup, which removes those shifts entirely.

The benefit is not only accuracy. A part that is set up once cannot be re-clamped out of position. For a 1,200 mm housing with a 0.02 mm perpendicularity callout between two bores, that matters more than spindle speed.

Where five-axis does not help is on simple turned parts. A valve stem is a lathe part. Putting it on a five-axis mill adds cost and nothing else. The judgment call is geometry: if the part has features that need to be true to each other, five-axis pays. If it has one axis of symmetry, turning pays.

Tolerance stack-up is the other side of the same coin. A ±0.005 mm tolerance on a single bore is achievable. Five stacked ±0.005 mm tolerances across a 600 mm assembly usually are not, because the errors add. When a stack-up looks tight, the fix is normally to loosen the non-critical dimensions rather than to tighten the critical one further.

• 1
  Count the setups firstEach one is a new source of error.
• 2
  Turning for round partsDo not mill what a lathe does better.
• 3
  Loosen the stack, not the fitNon-critical dimensions are free to relax.

For a wetted stainless part, the surface finish is a corrosion variable, not a cosmetic one. A rough surface traps chlorides and gives crevice corrosion a place to start. We typically machine sealing faces to Ra 0.8–1.6 μm and functional bores to Ra 0.2–0.8 μm where a seal or a sliding fit requires it.

Passivation follows machining for stainless parts. Free iron left by tooling has to come off before the part goes into service, otherwise the rust appears in weeks. Bead blasting is used on non-sealing surfaces to even out the finish, but blasting near a seal face is avoided because it can round a critical edge.

Inspection is where a subsea part is really decided. We check raw material certificates on arrival, monitor dimensions in process, and run a final inspection on 100% of parts before shipment. Reports are available on request, including dimensional results and material traceability.

The practical limit is measurement access. A deep, small-diameter bore cannot be verified with a plug gauge if the gauge cannot reach the bottom. If a print calls for a tight bore 400 mm deep, we will often ask to open one end so the feature can actually be measured.

• 1
  Finish is corrosion controlRough surfaces start crevice attack.
• 2
  Passivate after machiningFree iron has to be removed.
• 3
  Design for measurementIf it cannot be checked, it cannot be accepted.