Throughput recovery in a two-plant industrial group

A two-plant industrial group had order books it could not ship. Lead times had stretched from three weeks to seven; overtime was structural; and the board was reviewing a $14M capacity expansion as the obvious answer. The capex case was internally persuasive and externally unexamined.

The group asked for a second opinion on the expansion. The more useful question was where the existing capacity was going — because at nameplate, the two plants already had 26% more capacity than the order book required.

Four weeks of measurement at the line located the loss with some precision: two bottleneck work centres were being starved by upstream sequencing decisions made daily, informally, and differently on each shift. Rework at those centres consumed 19% of their hours, and priority orders — expedited by whoever escalated loudest — reset the schedule a median of six times a week.

None of this was a capacity problem. It was a decision problem wearing a capacity costume: no agreed sequencing rule, no fixed forum to hold it, and no baseline anyone trusted.

The work ran in three phases over nine months, with no capital request. First, the constraint map: losses located and priced line by line, and a frozen month-0 baseline signed by both plant directors. Second, cadence redesign: one sequencing rule for the bottleneck centres, a daily fifteen-minute schedule forum with the authority to hold it, and a weekly operating review of run-rate against baseline.

Third, the hold: three months of the new rhythm under oversight, with expedite requests routed through the forum and priced openly. The loudest-voice escalation channel closed because it stopped working.

Baselines frozen and audited before phase one; month-9 readings verified under MKM-F-003 and re-tested at month 12.