The changeover, which is not a capacity problem

Rennick loses 12.83 hours a week of constraint time to changeover, which is 53% of all downtime and 8.5% of scheduled time. That much is arithmetic. The more important thing is what the 55 minutes does to the schedule.

Worked example 7: reducing the changeover across 14 changeovers a week

First, the mechanics. A changeover has internal elements, which can only be done while the line is stopped, and external elements, which could be done while it is still running but currently are not. The single largest gain in almost every changeover reduction is moving work from internal to external, and it usually costs very little.

Nine minutes of every changeover is somebody walking to stores. That is 126 minutes a week, 2.1 hours of the constraint, GBP 13,043 a week, spent fetching. The redesign costs a second set of dies at GBP 9,400, a changeover trolley at GBP 2,200, pre-staged reel trolleys at GBP 1,600, and standard work and training at GBP 3,500: GBP 16,700 in total.

Now the gain, computed honestly in two steps.

Payback against the net figure is 16,700 divided by 17,319, which is under one week. Be suspicious of that, and say so before somebody else does: a payback of under a week is the signature of a benefit priced at a constraint against demand that is genuinely there. Both conditions have to hold. Check that the demand is real, that it is not conditional on a price the customer will not pay, and that nothing downstream of Line 3 becomes the new constraint once the extra 10,248 units start arriving. If despatch cannot handle it, you have moved the problem 40 metres.

Now the part that matters more than the capacity. Rennick runs 14 recipes and makes 14 changeovers a week, so each recipe runs once a week, in a batch sized to cover a week of call-off.

Product made on Monday cannot lawfully serve a call-off on Friday. At best a single run covers three days of demand. That means every recipe needs to run at least twice a week: 28 changeovers, not 14. The weekly plan Rennick operates is not achievable, and the planner knows it, which is why the mix is wrong three days out of five and why the service level is 97.2%. This is not a scheduling failure by a planner. It is a design failure that the schedule is being asked to absorb.

Doubling the changeover frequency at today's 55 minutes would consume 21% of scheduled time and is unaffordable. Doubling it at 30 minutes costs 1.17 hours a week more than the site spends today, and makes the required schedule possible for the first time. That is the argument to take to the capital committee, and it is not a capacity argument at all. The exam will offer "reduce changeovers to gain capacity" as the plausible answer and "reduce changeover time so the mix can be run more often" as the defensible one. The second one is why the service level moves.