Production sequencing: choosing the order jobs run on a given machine

Sequence-dependent setup turns running order into a cost

On plenty of equipment the changeover time depends entirely on what ran before. Coating lines go light to dark because the reverse needs a full wash; food plants run allergen-free product ahead of allergen-containing; presses group work by tool family; heat treatment groups by temperature band. Where this applies, the ideal sequence is not obvious and the penalty for ignoring it is large and recurring. The first job is to get the actual setup matrix out of the operators' heads and written down — which transitions are cheap, which are expensive, and which are prohibited outright for contamination or safety reasons.

Dispatch rules and what each one quietly sacrifices

Running the earliest due date first protects delivery and ignores setup, so the shift fills with changeovers. Running shortest jobs first clears the queue quickly and starves long jobs indefinitely. Grouping by setup family maximises machine output and pushes some customers to the back. Critical ratio balances remaining work against remaining time and needs due dates people believe in. No rule is correct in general; the useful move is to be explicit about which one the work centre is using and what it is trading away, because an unstated rule always defaults to whichever order avoids the most immediate argument.

Churning the sequence costs more than a better sequence saves

Reordering the queue repeatedly during a shift destroys the preparation that makes a sequence work: kits staged in order, tools drawn in order, material pulled forward, operators briefed. Each reshuffle strands some of that effort. A short frozen horizon — the next few jobs are fixed, everything beyond is provisional — gives the floor something to prepare against while keeping later work flexible. Plants that resequence constantly usually have a due-date problem rather than a sequencing problem, since every order is late and the queue is being reordered by whoever escalated most recently.

Breaking the sequence: authority and evidence

Sequences get broken for real reasons — a machine fault upstream, a material shortage, a genuine customer emergency. The failure is not the break but the absence of a rule about it. Name who can authorise a jump, require the reason to be recorded on the order, and make the displaced job's new position explicit rather than leaving it to fall wherever. Where sales can override the floor directly, the sequence has no owner. Reviewing recorded overrides monthly usually reveals a small number of accounts or products generating most of the disruption, which is a commercial conversation rather than a shop-floor one.

Signals that the running order is working

Watch total setup time per shift at the work centre against its output, and watch how often the executed order differed from the issued one. Rising setup with flat output means the sequence is being broken up. High adherence with poor delivery means the rule is wrong for the mix. Also watch the age of the oldest job in the queue: a sequencing policy that optimises machine efficiency will strand awkward, low-volume orders for weeks, and nobody notices until a customer calls. A standing check on queue age catches that before the complaint does.

Frequently asked questions

Should operators be allowed to reorder jobs at their own machine?

Within a defined window, yes — they see material and tooling problems before anyone else, and forbidding local judgement produces silent workarounds instead. Set the boundary: they may reorder within the jobs already released to the centre, they may not defer anything past its committed date, and any change gets noted. Beyond that window the decision affects other work centres and customers, so it belongs to the scheduler who can see the whole route.

How do we sequence when every order is already late?

Stop sequencing and fix the promise first. A queue where everything is overdue carries no information, so the running order defaults to whoever complained most recently. Re-date the backlog against real capacity, tell customers, then sequence against dates people believe. Doing this is unpleasant and usually resisted, but a plant that sequences on false due dates will keep expediting, keep breaking setups and keep delivering in an order that reflects internal noise rather than commercial priority.

Does sequencing matter on a work centre with quick changeovers?

Less for setup cost, but still for contamination, traceability and delivery. Even where the tool change takes minutes, a food or coatings plant may have transitions that are forbidden or that force a clean-down, and a plant handling controlled materials may need a specific order for segregation. Where none of those apply and changeovers are trivial, sequence by date and stop spending management attention on it — the gain is elsewhere.