Tightening a tolerance is the most expensive line on a drawing

How process cost changes as the tolerance band narrows.

Tightening a single tolerance on an otherwise ordinary drawing can dominate the entire part's manufacturing cost, because cost climbs steeply and unevenly as a tolerance band shrinks, and the steepest part of the climb begins once the tolerance approaches the natural limit of whatever process was making every other feature on the part comfortably and cheaply.

Cheap inside the process, expensive at its edge

A tolerance comfortably wider than a process's natural capability costs almost nothing extra to hold, since the process was already going to land well inside that band without special effort. Once a specified tolerance approaches or crosses the edge of what a given machine, tool or technique can reliably achieve, the cost of holding it rises sharply.

Holding the tighter band usually means three changes at once. The cheap, fast process is abandoned for a slower, more specialised one. Inspection steps appear that were previously unnecessary. Yield falls as more parts land just outside the narrower band and have to be scrapped or reworked. Each of those costs stacks on top of the others, so the total grows far faster than the tolerance shrinks.

The yield loss alone shows how steep this gets. Suppose a process's natural scatter just fits a tolerance band, so that the band's edges sit three standard deviations either side of the target. Only about three parts in a thousand then fall outside. Halve that band while keeping the same process, and its edges now sit at one and a half standard deviations, where roughly one part in every seven or eight falls outside and has to be scrapped or reworked.

The bullseye and the outer rings

Hitting anywhere within the outer rings of an archery target takes modest, ordinary skill, and almost any reasonably competent archer manages it quickly. Hitting the small bullseye demands far more skill, more practice, slower and more deliberate shots, and many more near misses along the way. Each extra millimetre of accuracy costs more effort than the one before it, and the price climbs fastest right at the centre.

Tightening a manufacturing tolerance follows almost the same curve, easy and cheap out in the outer rings where an ordinary process lands comfortably, and disproportionately expensive the closer the requirement pushes toward the centre of what a process can hit reliably. A stabiliser bar, a peep sight and a release aid, the extra equipment a competitive archer adds, do nothing for the outer rings. They exist to shave the last few millimetres off the group around the bullseye, much as a factory brings in specialised tooling only once an ordinary process can no longer hold a tightening tolerance on its own.

From drill to reamer to grinder

Cost against tolerance rarely climbs as a smooth curve. It jumps in steps as a design crosses from one process's comfortable capability into the next, more expensive process's territory. A hole about ten millimetres across, drilled straight into steel, can typically be held to a band of around a tenth of a millimetre. Ask for a band of about two hundredths and the hole has to be drilled undersize and then reamed, a second operation with its own tool. Ask for a band under a hundredth and it usually has to be bored or ground, a third process on a different machine, with its own set-up time and skill requirement.

Each of those jumps is a change of operation, and running the same drill more carefully will not cross it. A designer who tightens a tolerance by what looks like a small, reasonable amount on paper can unknowingly push a feature across one of these hidden process boundaries and add a whole extra operation to every part. The same halving applied to a tolerance that started comfortably loose, well inside the drill's range, barely moves the price at all. The cost of a tightening depends mostly on how close the starting point already sat to a boundary, and much less on the raw size of the change.

Inspection grows alongside the process

A wide, easy tolerance can usually be trusted from a sampled check. Measuring a few parts from a batch and inferring the rest are similar is reasonable when a comfortable process rarely drifts near its limit. A tight tolerance sitting near a process's real capability cannot be trusted that way, because a process running close to its limit does drift past it from time to time, so every part often has to be measured to catch the ones that failed, turning a quick spot check into a full inspection of the batch.

A process at the edge of its capability can sometimes avoid inspecting every finished part by being watched continuously, with each measurement tracked as it comes off the machine so that a drift toward the limit is caught and corrected before it produces a bad part. That trades the cost of checking every part for the cost of watching the process as it runs. Either way the inspection burden is a separate cost layered on top of the process change, and it is often the less visible of the two, buried in a quality department's labour hours instead of appearing as a line on a machining quote.

Redesigning so the feature can relax

Some features need to be held close to a process's limit for the part to function at all, and loosening those simply swaps cost for risk. A tolerance tightened without a specific functional reason is pure, avoidable cost. The cheapest fix is very often found back at the drawing board, by asking whether the assembly could be redesigned so that the feature no longer needs to sit so close to the edge. Loosening the requirement itself ends a cost that a cleverer manufacturing process would only have reduced.

A tolerance tightened out of habit, out of caution, or because a tighter number looked more careful on the drawing can quietly become the most expensive line on the page, dominating a part's cost far out of proportion to its effect on function. Reviewing every tight tolerance against the question of whether the assembly needs it is one of the cheapest design reviews available, because the cost of getting that one decision wrong compounds so steeply once a tolerance has crossed into territory the manufacturing process was never comfortable holding.