Lot-Size Reduction on an Already-Tight Molding Constraint

Start with the changeover load on the constraint.

• Available press time: 24 h × 5 days × 4.33 weeks ≈ 520 hours/month.
• At roughly monthly runs per SKU: 90 changeovers × 4.5 h = 405 hours/month of setup - about 78% of the press's capacity spent changing molds, leaving ~115 hours for actual molding.
• Cutting lot sizes by two-thirds triples run frequency: 270 changeovers × 4.5 h = 1,215 hours/month, more than twice the available time. The proposal is not merely costly - it is physically infeasible.

Conclusion: lot size on this press is constrained by setup time, not by a policy choice. Any cash release has to come from reducing the 4.5 hours first.

• SMED target: getting to 1.5 hours at current frequency frees ~270 hours/month - roughly half the press.
• That freed capacity is what makes smaller lots affordable: at 1.5 h setup, a one-third lot-size reduction costs 270 extra changeover hours, which the SMED gain just covers.
• Sequence it: measure and separate internal vs external setup → attack internal setup → then reduce lots, not the reverse.

At 92% utilization with real variability, queue time is not a side effect - it is the dominant term in lead time. As load approaches capacity, waiting time grows non-linearly; small schedule disturbances become multi-day delays, which is precisely why weekly expedites are 'breaking the schedule'. The 84% on-time figure is largely a utilization decision, not a planning failure.

What to do:

• Set a planned load ceiling of about 80–85% on this press and protect it as a rule, not an aspiration. The apparent 'lost' capacity is recovered through shorter queues, fewer changeover-driven disruptions, and less expediting.
• Buffer the constraint, not every workstation. A time buffer ahead of the press and a small finished-goods buffer on the 11 high-volume SKUs decouple it from upstream variation.
• Schedule the constraint, subordinate everything else to it. Release work to the press at its pace; do not let non-constraint efficiency targets push WIP in front of it.
• Diagnose the 38 days of finished goods before adding any. With 75% of volume in 11 SKUs, a large share of that inventory is probably in the tail - which is both the wrong place and the reason the cash is not being released even as service suffers.

Not uniformly - the 90 SKUs have two very different economics.

• The 11 SKUs (75% of volume): these justify long, stable runs and a finished-goods buffer. Cutting their lot sizes is the most expensive place to chase cash, because each changeover is charged against the highest-volume products. Instead, run them on a fixed, repeating wheel with dedicated molds where possible, and size the buffer to demand variability - this is where service is won.
• The remaining 79 SKUs: short runs, frequent setups, likely most of the 38 days of stock. Options in order of preference: consolidate onto a smaller press with faster changeovers, standardize mold heights and clamping so setups are cheap, quote longer lead times and make to order, or rationalize the tail outright.
• The real lever is the 4.5 hours, applied selectively: halving setup on the tail SKUs is worth more than halving it on the runners, because that is where changeover frequency is highest relative to run time.
• Measure it: track days of inventory and changeover hours by segment. If the 38 days do not fall in the tail within one quarter, the classification is wrong or the tail is being scheduled as if it were a runner.