Flexible automation: paying for variety you may or may not end up using

Flexibility means the cost of change, not the breadth of the brochure

Ask a supplier what it costs to introduce a new part and the answer separates real flexibility from marketing. The questions are concrete: how long the change takes, who can perform it, whether new tooling is needed, whether a program must be written or merely selected, whether the safety assessment alters, and whether the change can be reversed. A machine with a wide theoretical envelope and days of engineering per new part is a dedicated machine that occasionally gets re-dedicated. Put those questions into the requirement document and make the supplier demonstrate a product change during acceptance.

Part families are what make flexibility achievable

Flexibility is created upstream in design more than at the cell. Parts sharing datum features, gripping surfaces, fixture interfaces and roughly similar mass can pass through common tooling with a program change and a modest set of change parts. Parts designed independently by different engineers over a decade cannot, whatever equipment gets bought. A flexible cell therefore often implies a design standard for future parts, agreed with engineering and enforced through design review. Retrofitting commonality onto an existing range works for some features and is costly for others, so establish which parts are worth changing before specifying anything.

Changeover time is the honest measure

Measured properly, changeover is the figure that tells you what a cell can genuinely do. Measure from the last good part of one product to the first good part of the next, performed by the people who will do it on a night shift rather than by an engineer during commissioning. That interval sets the minimum economic batch, and therefore the order sizes the business can accept and the responsiveness it can promise customers. Where change parts live, whether recipes are stored in the control, and whether manual adjustment or re-teaching is needed influence it far more than the mechanical design does.

The capability bought and never exercised

A recurring pattern: the plant buys extra axes, spare input and output capacity, a general-purpose gripper and sensing able to handle parts it does not yet make, and none of it ever gets used, because using it requires engineering time nobody budgets once the project closes. The capability is real; the capacity to exploit it is not. Before paying for optional flexibility, identify who would implement the change, from whose budget, and how quickly. Where that answer is vague, the money often does more good making current products run reliably, with flexibility purchased later against a real requirement.

Where the premium is worth paying

Reconfigurable arrangements earn their premium where volume per variant is modest, the range churns, product life is short, or the plant sells capacity rather than a fixed product, as contract manufacturers do. They pay again where forecast confidence is weak, since committing to dedicated equipment can strand capital when demand moves. The accompanying requirement is skill: equipment is only flexible in the hands of people able to program and re-tool it, so the staffing plan and the equipment choice belong in the same discussion rather than in sequence. Where that skill cannot be recruited or held onto, the honest response is to buy less adaptability and narrow the product range instead.

Frequently asked questions

How much flexibility should we specify up front?

Specify the structural things that cannot be added afterwards and defer the rest. Reach, payload, machine envelope, available utilities and spare room inside the enclosure are effectively fixed at purchase, so size them against parts you can credibly foresee rather than against today's range alone. Programming capacity, extra grippers and further fixtures can follow once a product is confirmed. The expensive error runs the other way: tooling for hypothetical products inside an envelope that cannot grow.

Who should be able to change the cell over?

Production people on the shift that needs the change, otherwise flexibility exists only while engineering is available, which in practice means weekdays. Achieving that constrains the design: recipes selected rather than parameters typed, change parts that locate positively and cannot be fitted the wrong way round, no re-teaching of positions, and an opening check confirming the change worked. Where a change genuinely needs an engineer, say so in the planning assumptions instead of pretending the cell turns around at will.

Does flexible automation always cost more than dedicated equipment?

Per unit of output usually yes, because the mechanisms are more general, cycles run longer and the control is more involved. The comparison only makes sense across the whole life, including what happens when a product changes. Dedicated equipment is cheaper per part while its product runs and worth little afterwards. Reconfigurable equipment costs more per part and survives a product change. Which wins turns on your confidence that the product still exists in the same form in several years.