Carry the Part 2 loss tree into a weekly production board

Part 2 solved a diagnostic problem: it identified which SKU, shift and station owned the lost hours. This article continues with a different decision. It asks which losses disappear when the same demand is scheduled in a better order, and which losses remain even after the schedule is disciplined.

A common mistake is to move directly from a large downtime total to a machine investment request. First mark every event in the loss tree as recurring, avoidable, movable or hardware-bound. A changeover required by customer demand is recurring. A second changeover created by splitting the same SKU into two short campaigns may be avoidable.

• Use one planning horizon that reflects the real order cycle.
• Keep the same qualified packed-output boundary used in Part 2.
• Assign each loss to an SKU, shift and reason code.
• Separate a scheduling opportunity from an assumed engineering improvement.

Classify schedule losses before trying to remove them

Not every stopped minute is available for recovery. Product demand may require a size change. Hygiene rules may require cleaning. A roll may reach its safe change point. The planning team should therefore classify the event before placing a recovery value beside it.

Actually, the label “planned stop” does not make a loss efficient. A stop can be planned and still be duplicated, badly sequenced or scheduled after the line has already been interrupted for another compatible task.

Fix campaign length before adding another shift

A campaign is the uninterrupted quantity of one SKU produced before the line changes to another. Short campaigns improve responsiveness, but they increase the number of transitions paid by the factory. Long campaigns protect runtime, but they consume working capital and warehouse space if demand cannot absorb the output.

The planning decision is therefore not “run longer.” It is to set a minimum economic campaign by SKU using verified changeover loss, confirmed demand, shelf-life policy, warehouse capacity and delivery dates. The campaign should be long enough to avoid repeating a preventable transition but not so long that inventory becomes the new constraint.

The board above is a visual method, not a recommended schedule. Its purpose is to make repeated runs and transition count visible before the production meeting approves the week.

Sequence SKUs by observed change burden, not by label order

Alphabetical order and sales-entry order rarely produce the lowest-change route. Build a matrix from observed changeovers: outgoing SKU on the left, incoming SKU across the top, with total stopped minutes and first-good release time in each cell. The matrix should show direction because A-to-B may not equal B-to-A.

The earlier Welldone article on size changeover explains what changes physically inside one transition. This article does not repeat that tooling discussion. It uses the observed duration of each SKU pair as planning input, then chooses a weekly path that satisfies due dates with fewer high-burden transitions.

In most cases, the best sequence is constrained by more than changeover time. Customer due dates, material availability, packing format and warehouse limits remain hard boundaries. A schedule is improved only when it protects service as well as runtime.

Combine cleaning, roll replacement and inspection into controlled windows

Production teams often stop separately for work that could occur inside one safe, documented window. A planned SKU transition may already require access, cleaning and recipe verification. If a roll is near its approved change point and a routine inspection is due, moving compatible tasks into the same window can prevent a second interruption.

This is not permission to overlap unsafe work. Lockout requirements, access zones and quality-release steps still control the sequence. The planning gain comes from preparing tools, materials and people before the window starts, then executing a written parallel-work plan rather than discovering tasks after the line stops.

• Stage the next material set and verified recipe before the stop.
• List tasks that can run in parallel and tasks that must remain sequential.
• Name the person who releases each task and the person who releases the first good product.
• End the clock at qualified packed output, not at the first machine movement.

Protect shift handover from becoming an unnecessary restart

A stable line should not need to be stopped simply because the crew changes. Yet a weak handover can recreate startup loss: the incoming team rechecks settings, changes an unrecorded adjustment or pauses while waiting for material and quality status.

Use a running handover when site safety and staffing allow it. The outgoing operator records current SKU, material lot, active recipe, open alarms, last quality check, roll condition and expected next intervention. The incoming shift accepts the line against the same record rather than rebuilding the process state from memory.

The real question is whether the best shift can be repeated by the next team. If recovery exists only when one experienced operator is present, the capacity is not yet robust enough to enter the investment comparison as a permanent gain.

Validate recovered capacity with qualified packed output

A better-looking schedule is not evidence of recovered capacity. Run the revised sequence for enough comparable cycles to record transition count, stopped minutes, restart rejects, qualified packed output and late-order impact. Keep the same SKU, BOM and release boundary wherever comparison is intended.

Do not convert every saved minute into sellable output automatically. The recovered window has value only when demand, materials, labor, inspection and packing can use it. If the packing constraint identified in Part 2 remains full, main-line minutes may move the queue without increasing warehouse-released product.

Worked example: convert avoided transitions into a no-capex recovery

Assume the Part 2 loss sheet shows that campaign consolidation removes three transitions per week. Each avoided transition, measured from last good packed product to first good packed product, averages 50 minutes. Combining compatible cleaning and roll work saves another 30 minutes per week, while a running shift handover prevents 60 minutes of repeated checks per week.

Using the Part 1 illustrative planning rate of 24,000 qualified packed pieces per hour, 16 recovered hours equal 384,000 pieces per four-week period. Spread across the original 416 scheduled hours, that is an equivalent recovery of about 923 pieces per hour. The original illustrative gap of 7,731 pieces per hour therefore falls to about 6,808 pieces per hour, not to zero.

This calculation produces a decision boundary. Scheduling can recover a measured portion of the gap, while the remaining gap moves forward as a candidate for retrofit, an added module or another line.

Decision gate: freeze the operational gain before specifying hardware

Part 3 should finish with three signed outputs: recoverable capacity without capital expenditure, the remaining capacity gap and the losses that persist under the best repeatable schedule. Do not send a total downtime figure to Part 4. Send the losses that operations could not remove.

Part 4 can then compare retrofit and replacement against a cleaner requirement. Mechanical limits, control architecture, quality risk and shutdown window belong in that next decision. Scheduling losses do not.

Related project routes

Conclusion: recover the calendar before expanding the machine room

A production schedule can create capacity loss without a mechanical failure. Fragmented campaigns, high-burden SKU order, separate service stops and weak handovers repeatedly consume the same planned hours.

The recovery is credible only when qualified packed output rises under the same measurement boundary. After that test, the factory has a smaller and more defensible hardware problem. Which losses still appear when your best weekly sequence is repeated by every shift?

FAQ