At Andy Marine, we believe that practical manufacturing knowledge is one of the most valuable assets built over time.
Inside our workshop, there are engineers, production specialists, and experienced technicians who have spent years working with stainless steel marine hardware — from material selection and casting processes to finishing and quality control.
Many of these insights are learned through daily production experience: understanding why a component fails, how different materials perform in marine environments, and what details truly matter when developing reliable hardware.
However, this practical knowledge is often shared only within the workshop and rarely reaches customers who are making sourcing and engineering decisions.
ENGINEER TALKS is our effort to organize these conversations and share the manufacturing experience behind marine hardware.
Through this series, we discuss materials, production processes, design considerations, and real-world lessons from the perspective of a marine hardware manufacturer.
Today's ENGINEER TALKS Ep.36 is presented by Engineer Wu, written by me the editor.
Two Bases, One Drawing Envelope
Take a hypothetical pair of cleat bases: same envelope, same bolt pattern, same load path. One is welded up from plate, the other cast in one piece. On the bench they look alike.
The instinct is to ask which is stronger. The better question is which failure this project can inspect. Final suitability still depends on the vessel structure and a qualified review — this episode is about the route, not the boat.
What the Route Actually Decides
Where the risk sits. In a casting it is in solidification: whether the mould filled, whether the section fed as it cooled. None of that is visible on the finished part, and all of it is controlled by gating, risering and section design. In a weldment the risk is in the joints — places where metal was melted, where properties beside the weld changed, and where geometry concentrates stress. A weldment has one such place per joint; a casting has none of those, though every section change carries its own solidification question.
Tooling and joints. Casting carries a one-time tooling cost amortised over volume. Fabrication carries little tooling but adds per-unit welding, fixturing, distortion correction and inspection, and that per-unit cost tracks joint count and joint access rather than part weight, because every joint adds fit-up, welding, cleaning and an inspection gate. A casting replaces the joint set with one body, which changes which geometries are cheap to make. Where that crossover falls is your own volume and design churn.
Change and repair. A weldment absorbs a geometry change by re-cutting; a casting absorbs it through pattern modification and a new sample, which is why welded assemblies dominate prototypes. Repair runs both ways: a weldment's is local, gouged back to sound metal; a casting's is a qualified operation on a section whose thermal history the repairer may not know — constrained, not worse. Drawing conventions belong to the ET27 discussion: custom marine hardware drawings.
The weldment's risk is localised at its joints; the casting's is distributed through its sections. Neither is inherently better: they fail in different places and are inspected by different means.
The Weldment's Failure Map
A weldment's failures are few, well documented and unevenly visible. Each has a gate that would catch it.
| Risk | What it physically is | Gate that addresses it |
|---|---|---|
| Incomplete root fusion / incomplete root penetration | One side of the joint unfused at the root, or both root faces unfused, with different causes | Joint preparation and a qualified procedure, plus penetrant or magnetic particle inspection where the root is reachable, radiography or ultrasonic where not |
| Lack of sidewall or inter-run fusion | Deposited metal fails to bond to base metal or a previous pass, leaving a planar interface that acts as a stress raiser, often subsurface | Procedure and technique, interpass cleaning, volumetric inspection where warranted |
| Undercut | A groove melted into the base metal at the toe, reducing effective section and raising stress where it is already highest | Visual examination against the named acceptance level, toe blending where specified |
| Distortion and residual stress | Non-uniform heating and cooling; austenitic stainless conducts heat poorly and expands more, so heat and stress concentrate in a narrower zone | Weld sequence, fixturing, heat sinks, and design that keeps long welds away from thin sections |
| Heat-affected zone condition | Base metal beside the weld changed by the heat cycle, with no visible sign at the surface | Grade selection, the qualified procedure, post-weld surface restoration |
Where the root cannot be reached by penetrant or magnetic particle testing, detection moves to radiography or ultrasonic inspection, and repair means gouging back to sound metal and re-welding to the original procedure. An inaccessible root is asserted, not inspected — root access is a drawing decision before it is a welding decision.
Imperfection is not defect: acceptance is judged against a quality level named by the application code, and which level applies is the buyer's decision, not a property of the weld. What an individual indication means belongs to the ET34 discussion: dye penetrant testing of marine castings.
The Heat-Affected Zone Is a Heat-Cycle Question, Not a Grade Question
In austenitic stainless steel, carbides beside the weld dissolve on heating and re-form on cooling as chromium carbides at the grain boundaries, taking chromium out of the metal next to them. Where the depletion is sufficient the steel is sensitised, and the certificate does not change.
TWI places that band at roughly 600–900 °C, with exposure times as short as tens of seconds. The message is not a number to design to: ordinary welding heat sits inside it, so the zone is a normal consequence of joining.
Solution treatment re-dissolves the carbides, but it is rarely practical on a finished structure. A low-carbon grade moves the time axis rather than abolishing the phenomenon. Stabilised grades tie up carbon with titanium or niobium, yet a narrow band near the fusion line can still be attacked in hot acids.
None of this predicts what a joint will do in service; the outcome depends on alloy, joint, exposure and cleaning regime. A sensitised zone, at least, shows nothing at the surface.
Post-Weld Cleaning Is Three Operations, Not One
"Pickled and passivated" appears on drawings as one line item; it is three operations. Descaling removes thick, visible oxide scale. Pickling removes a thin layer of metal, and it is what takes off the heat-tinted layer where surface chromium has been reduced. Passivation removes no metal; it develops the passive film on a clean surface that already carries its chromium. They are sequential: passivating over an intact heat tint restores nothing.
Acid does not degrease — oil and shop contamination come off first, or the acid reaches a contaminated surface instead of the steel. Heat tint on a face that will be closed up still has to go if that face sees the service environment.
Pickling is not harmless in excess: over-long contact with hydrofluoric-acid-bearing products can pit the surface, so the supplier's instruction governs.
What the Paperwork Can Prove
A welding procedure specification is a written instruction. A procedure qualification record covers a range: a procedure within that range produced joints meeting the required test results. A welder certificate covers a test piece welded under test conditions. None proves what happened on a particular joint. The batch is released on production records — which procedure revision, which welder, which joint, which inspection result.
The casting side raises a parallel question. Repair welding of a steel casting is a qualified operation under a standard of its own: ASTM A488/A488M covers qualification of procedures, welders and operators for fabrication and repair of steel castings by electric arc welding.
The record's content matters more than the fact of repair: which procedure, which welder, what was excavated, what the inspection found, and whether it sat inside the qualified range for that thickness and position. "Welded and dressed" fails at a customer audit, not at despatch.
What to Put in the Enquiry, and What to Require at Release
- If welded: state the weld length, joint access, and whether the root can be reached for inspection.
- If cast: state annual volume and expected design churn, so tooling can be amortised honestly.
- Either route: name the acceptance level, or say the application code names it.
- Either route: ask for the record set at release, not certificates at first article.
- Either route: ask for cleaning as three line items — degrease, pickle, passivate — so a brush-off is not invoiced as chemical restoration.
At release, ask not whether the paperwork is complete but whether it identifies the joints you are accepting.
The argument that survives is narrower: which failure can be inspected, by what method, on what record. The broader case for durability is in an earlier episode: structural design, surface finishing and installation.
