Tolerances: Thickness Is a Floor, Not a Tolerance Band

A design engineer once explained to me that he could simply mill a missing plate thickness out of the next larger stock size. In general engineering, that would be a perfectly reasonable solution – the next larger thickness is already in stock, the press is standing idle anyway, why wait for a delivery. With armour steel, that exact move is the mistake that costs you the qualification. Not because it is poorly executed, but because it rests on an assumption that simply does not hold for this material.

The reason: with ordinary structural steel, plate thickness is a tolerance band with a plus and a minus. With armour steel, it is a floor. Read both rulebooks the same way, and you end up delivering a plate that conforms to spec on paper and is still ballistically insufficient – a sentence that sounds like a contradiction the first time you hear it, and still describes exactly what happens in practice.

Two rules that do not exist in ordinary steel construction

With structural steel, plate thickness is a dimension with an upper and a lower deviation limit, and both directions are equally valid as long as the static calculation checks out in the end. With armour steel, thickness is not a plain dimension but a protective property. Two rules follow from that which have no equivalent in ordinary steel construction.

The first rule: nominal thickness is never undercut. Every tenth of a millimetre below nominal is lost protection – directly and proportionally, not only past some threshold. A plate that conforms, on paper, to the relevant flat-product standard while sitting at the lower end of its tolerance band can still be noticeably thinner than nominal. That standard knows nothing about protective function; it governs dimensional conformity, not ballistics. Order strictly to that standard without specifying the tighter tolerance class, and you risk a plate that conforms and is still too thin for its intended purpose.

In practice, that means: unless the design has generous margin built in, the order needs to specify the tighter tolerance class, where the lower deviation limit sits at zero and the entire allowed range runs on the plus side. And incoming measurement needs to be documented, not spot-checked and waved through – under quality-assured production, that is a documentation requirement anyway.

Why machining cannot substitute for thickness

The second rule is the actual point, and it is the mistake almost every shop makes when starting out, precisely because it would be entirely correct in classic general engineering: a thinner part is not milled down from a thicker plate. The temptation is understandable – the desired thickness is not currently available, the next larger one is already in stock, the machine is ready to go. It is still not permitted, and for several independent reasons, any one of which is sufficient on its own.

Qualifications are tied to thickness ranges. Technical delivery conditions explicitly link hardness values to specific thickness ranges. A plate milled down from a larger starting thickness is therefore not a qualified plate of the target thickness – it is a plate of the original thickness with an altered cross-section and no valid certification for its new state.

The ballistic proof also applies to the delivered plate in exactly that condition, not to a thinner part machined from it. Ballistic tests are run per batch and per thickness; a plate's certificate says nothing about the ballistic performance of a cross-section that has since been altered.

There is also a metrological point that gets overlooked often: hardness testing is performed near the surface, at a depth of a few tenths of a millimetre up to just over one millimetre below it. Mill away a millimetre, and you remove exactly the zone where conformity was originally verified – the proof is left hanging in the air after that. The hardness profile through the cross-section is also deliberately set and thickness-dependent, not freely reproducible; a machined-down cross-section has a different microstructural state than a rolled and quenched-and-tempered plate of the same final thickness. And one-sided material removal from a quenched-and-tempered plate generates residual stress and distortion that cannot be hot-straightened out with this material – subsequent cold straightening does not turn the part back into a qualified plate either way; the lost certification stays lost.

In the end, it is not an economic win either. Bulk removal of armour steel by machining consumes tooling, machine time, and tool life in a ratio to the material price that simply does not add up – the apparent procurement advantage is usually eaten up after the first couple of insert changes.

What is actually permitted

Not every operation that touches thickness is forbidden – the dividing line does not run between "machining allowed" and "machining forbidden," but along the question of whether the load-bearing protective thickness is altered in the relevant direction.

• Localised machining of mating faces, counterbores, chamfers, fits, or hole patterns is permitted when specified on the drawing, as long as the remaining thickness in the protection-relevant area is untouched.
• Removing the heat-affected zone of a flame-cut edge is not just permitted but required – it concerns the edge, not the protective face.
• Bulk reduction of plate thickness to a different nominal size, by contrast, is not permitted without the customer's explicit consent.

If nominal thickness must be departed from, there is exactly one permitted path, and it runs through the customer, not through your own work preparation: a formal deviation approval, requested before delivery, never after. A request filed after the fact is worthless. A delivery without an approved deviation gets rejected at the customer's incoming inspection.

What it costs

The mistake rarely shows up immediately. A machined-down plate passes your own visual inspection and often the dimensional check too, since the drawing specifies exactly the target dimension. The mistake only becomes visible when a customer traces the part back, an audit asks about the origin of that thickness, or, worst case, ballistic performance is actually tested. By then, the part has been produced and delivered – recall costs, a retroactive deviation request, and the loss of the customer's trust all come on top of what production already cost in the first place.

What to do about it

The most important step is to consciously leave behind the mindset from classic steel construction: with this material, plate thickness is a verified property, not a freely interpretable dimension. Anchor that firmly in work preparation, purchasing, and design, and you avoid the most expensive mistake in this chapter before it ever happens.

What this means for your shop

If your design or work-preparation team is still applying tolerance rules from classic steel construction, that is exactly the point I examine on site – drawings, ordering practice, and incoming inspection. The Defense-Readiness-Check examines this systematically, over two days on site, for a fixed fee, and ends with a written roadmap. If you would like to clarify beforehand, with no obligation, whether that is the right starting point for your shop: initial call, 45 minutes, remote, free of charge – book a first conversation.