Dedicated machine tools

Why does a factory making one part forever buy a machine that can make nothing else?

A modern machining centre can make almost anything: change the program, change the tooling, and the same machine cuts a gearbox housing this week and a bracket the next. It is the more capable machine by any obvious measure.

So why does a plant committed to one component — the same cylinder head, millions of times — spend more on a machine that can produce that part and literally nothing else? Buying less capability for more money looks like a mistake. It is worth working out what the plant is buying instead, because the answer is not sentiment about old technology.

Start with where the time goes on the flexible machine. Cutting metal is only part of the cycle. The spindle must change tools between operations. The part must be clamped in a fixture that can hold many shapes, which means it is clamped less rigidly than a shape-specific fixture could manage. The machine must move between features under program control, which means moving one head through many positions in sequence.

Now ask what falls away if the machine only ever makes one part. The tool changes vanish, because you can mount every tool at once, each on its own head, each permanently at the right angle. The fixture becomes a casting shaped to that exact component, gripping it in the places it is stiffest. The sequencing largely vanishes too, because heads that never need to move elsewhere can work simultaneously — drill eight holes at once rather than one after another.

So the dedicated machine is not simply a general machine with its options removed. Every degree of freedom given up was traded for something: rigidity, parallelism, or eliminated motion. Does that change the cost picture, or only the speed?

Both, and in a way worth being precise about. Compare the two on cost per part. The dedicated line has a much higher fixed cost — it is bespoke, engineered for one geometry, and worthless for anything else — and a much lower cost per unit, because cycle time is short, setup is zero, and the fixture holds tolerance without operator skill. Two lines, one steep and low, one flat and high, and they cross at some volume. Below the crossover the flexible machine wins outright. Far above it, the dedicated machine wins by an enormous margin, because the fixed cost is divided by millions.

That gives us the shape of the bargain, but not yet its full price. Ask what else was given up along with the flexibility. If the part changes — a new engine design, a revised tolerance, a customer specification — the flexible machine is reprogrammed in an afternoon. The dedicated line is scrap, or an expensive retooling that may cost most of a new line. What the plant actually sold, then, is not just versatility. It sold the option to change its mind, and it was paid for that option in cycle time.

Which reframes the decision honestly. The right question is never "which machine is better" but "how confident am I in the volume and the stability of this design, over the life of the asset?" The dedicated machine is a bet that the part will not change. The flexible machine is insurance against it changing, bought at a premium of a few seconds on every part forever.

And notice that the market's answer has shifted over time, which is good evidence that this is a real trade-off and not a fixed truth. Ford's Highland Park plant took specialization about as far as it can go and produced the Model T at a price nobody could match — and famously found the same line an obstacle when the Model T finally had to be replaced. Modern vehicle plants, facing many variants and shorter model lives, have moved substantially toward flexible body shops and multi-model lines, accepting a slower cycle to keep the option open. Dedicated equipment did not lose an argument; the volumes and product lifetimes changed, and the crossover moved.

One caveat about the arithmetic. Real comparisons are messier than two straight lines: a dedicated line can have poorer availability, since one failed station stops everything, whereas a cell of general machines degrades gracefully. The honest version of this calculation uses the actual numbers, not the shape.

The misconception worth correcting is that a dedicated machine is a primitive one. It is usually the more sophisticated object: multi-spindle heads working simultaneously, a fixture engineered around one casting's stiffness, in-process gauging, and error-proofing that makes it physically impossible to load the part backwards. What it lacks is generality, and generality is not the same as sophistication.

Second, the specialization is not only in the metal. It reaches into the workforce, the maintenance regime and the supply chain, all of which become expert in one thing only — the same bargain repeated at a larger scale, and why a plant's ability to change products is more constrained than its machine list suggests.

Third, the space between the two extremes is real and is where most of the interesting engineering now happens. Flexible manufacturing systems, and the reconfigurable machines that let a line be re-laid for a related family of parts rather than one part, are attempts to buy some of the dedicated machine's speed while keeping some of the option to change. They are compromises, and they carry the costs of both.

• Where the cost curves actually cross, and how a plant estimates that volume before committing.
• Reconfigurable manufacturing systems as a middle path, and what they give up at both ends.
• Why a dedicated line's availability behaves differently from a cell of general machines.

Every term the collection defines is gathered in the glossary.