Every part could be tested for something. The question on a real job is which tests are worth doing, and the answer comes from asking what would go wrong if the part were not tested, and what happens if it is. On a simple bracket for a low-stress location, the answer is usually that nothing is worth validating beyond the dimensions. On a fabricated pressure path that is welded, clamped and exposed to heat cycling, that answer is different.
Thinking in terms of risk rather than in terms of thoroughness keeps a project moving. A test is worth running when the consequence of being wrong is expensive, dangerous or difficult to fix later, and when the test is the cheapest way to reduce that uncertainty. Everything else is process control, which is a different thing from validation, and cheaper.
The highest-value checks are on interfaces, because those are what a customer cannot fix in the workshop. Whether a flange matches the mating flange, whether a bracket lines up with the mounting points, whether a hose end sits on a spigot as intended, whether the run clears the components around it. A part that measures perfectly to a drawing and does not interface correctly is a failed part, and the drawing will not tell you so.
Sealing comes next. Joints that must hold pressure, contain fluid or withstand movement are the ones where a small error becomes a persistent problem. Taking the time to confirm that a joint seals under a representative condition before a batch runs is usually the best-value validation on a custom job, because sealing problems are the ones that recur across a whole batch.
Exposure to heat and to movement rank third and fourth. They matter most where the part is in a hard location, where an earlier version has already failed, or where a material or process has been changed from what has worked before.
A good custom process builds in most of the checking before validation is even discussed. On our own job flow, drawing and negotiation come first, then tooling, and the tooling is tested before anything is produced from it. Samples are made from the tested tooling, an engineer measures their sizes and tests hardness, and the samples go to the customer. Mass production starts only after the customer confirms.
That sequence means a customer is normally confirming a physical part rather than a specification. It also means several questions have been answered before the customer sees anything: whether the tooling produces the intended shape, whether the sample presses together correctly, whether the material behaves as expected. Validation beyond that is aimed at the questions the process cannot answer by itself.
Understanding which route produced a part matters here. A sample mould is smaller and cheaper and makes samples only, while a big mould makes both samples and production parts. Where a project begins on a sample mould and later moves to a big mould, that move is itself a point worth validating, because the tooling that makes the production part is not the tooling the sample came from.
New work deserves more attention than a repeat. The first time a shape, a joint or a combination of materials is made, nobody knows where the difficult part is, and the checking that follows is how that knowledge is acquired. Once a part has run and the customer has confirmed it, the second order is a different kind of job: the questions are about consistency rather than about feasibility.
Repeated work is normally managed by process control rather than by fresh validation. The tooling is known, the sequence is known, and the checks that matter are the ones that catch drift: a dimensional check on a first article, a quick confirmation that the material arriving from stock is the material specified, and a fit check where the vehicle interface is critical. Re-validating everything each time does not improve the part; it adds cost and delay to a job that has already been proven.
The most valuable validation is often the least formal. Fitting a part onto a vehicle and confirming clearances, checking that a flange sits flat against a mating face, holding two pieces together before welding them, confirming that a hose can be pushed onto a spigot by hand without being forced. Those checks cost minutes and catch the errors that matter most.
More formal methods cost more and should be reserved for questions the simple checks cannot answer. Dimensional reporting from an outside service, a pressure test on a completed assembly, exposure testing of a material, and instrumented running are all worth paying for in the right circumstances, and all of them are wasteful when applied to a job where the risk was already low.
The judgement is one the fabricator normally makes with the customer. That conversation is the heart of the work we describe on our customization pages, and it is why the tooling decision comes up so early in a project.
Most of what gets recorded on a job is never needed again. The evidence worth keeping is the material that answers a question later: which drawing revision the part was made to, which material was used, what the confirmed sample looked like, and what was changed after a problem was fixed.
Photographs are underrated here. A picture of the part on the vehicle, of a critical joint, or of a modification made at the customer's request will resolve an enquiry faster than any document. Where a part has been made in titanium or in stainless, the finish and the weld appearance in a photograph will also jog the memory about how that job was run, and material options of that kind are set out across our shop by material listings.
The scope of validation belongs in the initial conversation rather than in a later dispute. These are the points to settle.
Agreeing those points turns validation from a vague expectation into a short, practical list. Related fabricated assemblies sit under other parts and titanium welding parts, with tubing work under exhaust pipe and charge-side components under intercooler.
Experienced fabricators validate less than beginners expect, and it is not because they are casual about risk. It is because they have already seen the same decisions many times, so they know which three things on a job are worth checking and which twenty are not. That judgement is what a customer is really paying for when they place a custom order.
What it should not become is an excuse to skip the checks that have earned their place. The sample confirmation before a batch, the fit check on the vehicle, the measurement and hardness check on a first article: those are not formalities. They are the short list that keeps the expensive mistakes out of a job, and everything else can be decided case by case.
