Changing a Garment Production Schedule: Readiness, Freeze Windows and Approval

A late fabric lot changes more than one start time

The next order is due to enter a sewing line, but its material has not been released. Another order is ready. Moving that second order forward seems sensible. The line can keep working while the first order waits.

That choice can still make the more urgent order late. The substitute may take hours to finish and need two setup changes. It may also use a finishing machine that both orders need. A new colour on the planning board does not settle those effects.

Garment production schedule change control gives the team a clear decision process. The team compares choices, agrees on the tradeoff and issues one current plan. People on the line then know which instructions apply and why the sequence changed.

The factory case below is fictional. Its quantities, rates, deadlines and approval fields are teaching examples. They are not standard planning allowances or rules for every factory. The purpose is to make the consequences of a change visible before people act on it.

Keep the original promise and the current forecast

A useful planning record distinguishes the agreed baseline from the latest forecast. The baseline shows what the team intended to achieve. The current schedule shows what the team now expects to happen. It uses actual progress and the latest limits on the work.

These records answer different questions. A planner needs a current forecast to direct the next shift. Managers also need the earlier promise. Comparing it with the forecast helps them trace when an order moved and why.

The US Government Accountability Office (GAO) makes this distinction in its Schedule Assessment Guide. Best Practice 10 also explains why approved versions and change controls matter. The guide concerns project schedules. A factory can use its approach to records without taking on the full approval process of a large public project.

Do not erase yesterday's plan to make today's results appear on time. If a baseline change is justified, retain the earlier version, approval and reason. A revised promise should remain traceable to the decision that changed it.

Not every forecast update needs the same approval as a change to a customer promise. Define which decisions each role may make. A planner might record that fabric is now expected at noon. That entry alone does not approve a new shipment date or a change to work already released.

Establish when an order is actually ready

“Fabric arrived” is incomplete evidence for a sewing start. The relevant material may still await inspection, shade allocation, cutting or release. A partial delivery may support only part of the required size mix.

The readiness check should name the exact order, style revision, colour, quantity and production stage. It should point to the records that permit work to begin. A substitute order is ready only for the scope supported by those records.

A sewing start may need approved cut panels and trims, plus the method and product details. Other items may be needed only at a later stage. Set dates for those inputs too. Requiring every future item at the line today can delay work without solving the real readiness gap.

Microsoft's operations-scheduling documentation describes finite material and finite capacity as separate limits. When both apply, the needed materials and resources must be available together. A material forecast alone therefore cannot establish a feasible start.

ApparelDiary's supplier lead-time guide explains the difference between arrival and usable stock. Use the same distinction in a rescheduling decision. In the example below, “ready” means that inputs have been released for sewing. A truck at the gate does not meet that condition.

Give the near-term schedule a clear change rule

A factory may call its protected near-term period a freeze window. Within that window, a planner cannot move work freely without the defined approval. The purpose is to protect preparation and released work while allowing necessary exceptions.

Suppose a site protects its next twelve working hours. A sequence change within that period needs the production manager's approval. This is an invented policy. A real site would choose the period based on how much preparation it needs and how quickly it can respond to change.

Outside the window, the planner may have more freedom, subject to capacity and due-date limits. Inside it, the change request needs a reason and an assessment of affected work. The rule must still allow an urgent safety response. It must never force continued use of material on hold.

Software terms need separate care. Oracle's 26B supply-plan guide describes a planning time fence. This feature restricts certain actions within a set period. It also allows some reschedule-out or cancellation suggestions for excess supply. A planning time fence is therefore not a universal ban on every change.

Document the actual software settings as well as the human approval rule. A field called “firm” or “frozen” does not by itself prove that every planning run will preserve every order. Test the relevant behaviour before relying on it to protect released work.

Look at the work affected, not just the new start date. An order may move to a start outside the window even though its panels are already cut. Its attachments may also be prepared. The rule should cover this work too. Moving the date on the board does not undo the preparation or remove the cost of changing it.

Set up a small case with explicit boundaries

Consider one sewing line that can process only one of two styles at a time. Its current setup is already suitable for order A. There is no unfinished quantity from an earlier order occupying the line.

Both styles run at an assumed rate of 100 accepted garments per working hour once setup is complete. Each change between styles takes one working hour, including the required startup check. All required staff and support are available for those planned hours.

Hours are cumulative working hours from the decision point. They are not clock times or calendar days. Breaks and closed periods do not add productive time. The rates already measure accepted output at this stage. Applying another efficiency factor would change the stated basis.

These deadlines concern sewing-stage completion. Later washing, finishing, inspection and shipment must still be planned. The example does not equate a completed sewing lot with a shipment ready to leave.

The original sequence was A from hour 0 to 8, a setup change from 8 to 9, then B from 9 to 15. Both stage deadlines were met. Now A's ready time moves to hour 4, so the old sequence cannot run as drawn.

Option one: retain the sequence and wait for A

If the team keeps A first, the line waits from hour 0 to 4. A then takes eight hours and finishes at hour 12. After the one-hour setup change, B runs from hour 13 to 19.

A meets its hour-14 deadline. B finishes three working hours after its hour-16 deadline. The last order leaves this stage at hour 19.

The total is easy to reconcile: fourteen hours of running, one hour of setup and four hours of waiting. Together they occupy nineteen working hours.

This option preserves the sequence and avoids disturbing A's prepared setup. But B's three-hour delay may still be unacceptable. Waiting is a choice with consequences, but it is not automatically worse than every substitute plan.

The four idle hours are not automatically four hours of avoidable cash expense. Some paid costs may remain under every option. Record the operational delay first, then assess any real cost differences on the appropriate basis.

Option two: finish B before returning to A

Moving B forward requires a setup change from hour 0 to 1. B then runs for six hours and finishes at hour 7. The line changes back to A from hour 7 to 8. A finishes at hour 16.

There is no planned waiting in this version. Both orders finish by hour 16, which is earlier than hour 19 under the first option. However, A is now two hours late against its own deadline.

The work uses fourteen running hours and two setup hours. It ends sooner overall because it avoids the four-hour wait, even though it adds one setup change.

That does not make it the right business decision. If A feeds a fixed downstream slot at hour 14, missing that slot could create a much larger delay. B's early completion might provide little benefit if its next stage cannot receive it yet.

The planner must show both orders' results. Reporting only that the line stayed busy would hide the delay imposed on A.

Option three: split B around A

Suppose B can be split into two identified batches without breaking its approved method or quality controls. This is a stated case assumption, not permission to split any garment order.

The line changes to B from hour 0 to 1, then makes 300 garments from hour 1 to 4. It changes to A from hour 4 to 5. A runs from hour 5 to 13. After another setup change from hour 13 to 14, the remaining 300 garments of B run from hour 14 to 17.

A now finishes one hour before its deadline. B finishes one hour late. The plan uses fourteen running hours and three setup hours, with no planned wait.

The split improves B's completion by two hours compared with waiting for A first. It also keeps A within its deadline. Compared with completing B first, it protects A at the cost of another setup and a later finish for B.

The physical split needs a record. List B's sizes and bundles in each batch, and name the place that will hold the unfinished balance. Someone must check that balance. Without a clear owner, both teams may assume that the other will finish it.

If the required batch split is not approved, this option is unavailable. A mathematically attractive sequence cannot override a technical restriction. The planner must remove infeasible choices before asking management to choose among them.

Reconcile the split by size and status

To make the example more concrete, suppose B contains 150 small, 300 medium and 150 large garments. An approved equal split would place 75 small, 150 medium and 75 large garments in each batch. Each group totals 300, and both groups together match the full 600-piece order.

That balance must also exist in the physical bundles. Give each group a clear identity so that two instructions do not refer to the same goods. Count the first batch's completion once against the open order. Keep the second batch visible as unfinished work until it is complete.

Suppose the first batch has only 280 accepted garments. The order still needs 600 minus 280, or 320 accepted garments. Its open balance is no longer 300. However, this does not automatically mean cutting twenty extra garments. Those twenty may still be on hold or await a decision. They have already used material from the order.

Reconcile that status before issuing a replacement quantity. The time left may also change, but a count does not show how long the work will take. Repair, release and replacement can follow different routes. Preserve those routes in the plan. Do not treat every piece awaiting acceptance as a new garment to make.

Compare the consequences on the same basis

The table compares three candidates under one set of assumptions. It does not prove that no better plan exists. A second qualified line might create another option. So might an allowed batch transfer or a new, confirmed material date.

Do not add the late hours and call the smallest sum the best outcome without agreement on the objective. A one-hour miss on one order may matter more than a longer delay on another. The table preserves those separate effects so the approver can assess them.

ApparelDiary's guide to EDD, SPT and critical ratio explains common sequencing rules. A dispatch rule can help select a sequence. It cannot supply missing readiness evidence or authorise a change in a customer promise.

Check the assumptions that could reverse the choice

The split plan assumes A is ready at hour 4. Suppose release slips to hour 6. Its startup check needs the released inputs. There is no checked time split between advance preparation and that check, so this forecast credits no completed setup before release.

The first B batch still finishes at hour 4. The line then waits until hour 6 and changes to A from hour 6 to 7. A runs from hour 7 to 15. After the final setup from hour 15 to 16, B finishes at hour 19.

A is now one hour late, and B is three hours late. Fourteen running hours, three setup hours and two waiting hours total nineteen hours. A validated preparation plan might improve this forecast, but the planner must establish that split before claiming the gain. The decision record should show the sensitivity before approval.

A ready-time estimate should therefore have an evidence owner and a next update. A message saying “likely soon” is too weak to support a fixed start. Where the release remains uncertain, preserve that condition in the forecast.

Run time also needs scrutiny. A recently introduced style may not sustain the rate from a mature style. The rate may assume trained workers, stable quality and suitable machines. State those conditions beside the time estimate so that the planner can check them.

The Microsoft resource guidance distinguishes resource capability from availability. A free machine or worker is not necessarily suitable for the required operation. Its scheduling model matches task requirements to resource capabilities and considers existing reservations.

For a factory decision, verify the actual machine, attachment, method and worker competence. A spare total of labour hours is not enough. The people and equipment must be able to perform the work when it is needed.

Include setup preparation without counting it twice

Some setup work can happen while the line keeps running. Other tasks need the line to stop. The Lean Enterprise Institute's SMED description separates these external and internal setup activities.

The planner can then ask which preparation is safe to do before the change. The rest still needs time in the schedule. Calling a task external does not make its labour or equipment needs disappear.

In the case, the one-hour change includes the required startup check. Adding another hour for the same check would double count it. Leaving it out of both estimates would make the change appear too short.

External preparation still uses resources. A mechanic preparing B's attachment may already be committed elsewhere. If that work is not ready, the assumed one-hour interruption may no longer be valid.

Do not use the name of an improvement method as evidence that a change will take less time. Use a checked estimate for the relevant sequence and conditions. A-to-B and B-to-A change times need not be equal in a real factory.

Trace the change through the rest of the route

Before approving a substitute order, follow its output into the next stage. A washing batch may need a minimum load or a booked machine. Finishing may have a fixed style sequence. Quality may have limited inspection capacity at the proposed release time.

These constraints can alter the result without changing sewing's arithmetic. B may finish early at sewing but wait at its next stage. In that case, the factory has moved stock without bringing shipment forward. If A misses a booked process slot, its full delay may exceed the two sewing hours in the table.

Check shared resources in both directions. A borrowed operator may reduce another line's feasible output. A second line may be free only until a different order arrives. Any recovery claim should include the displaced work.

Released work also needs protection. Identify which order owns the cut panels, labels and work tickets. Prepared attachments also need a clear assignment. If a transfer is approved, record the quantity and destination. Updating the schedule must not silently change stock identity or product revision.

Keep the final commercial commitment separate from the internal stage forecast. Only the authorised commercial route should communicate a revised delivery promise. Production provides the evidence; it does not make every planner a buyer-facing decision-maker.

Approve a specific change with named conditions

A useful change request is short enough to review but complete enough to act on. It names the old version, the proposed version and the affected orders. It explains the triggering fact and compares the feasible choices.

The example request could propose the split plan with two conditions. A must be released at hour 4, and B's batch split must be approved. It should state A's hour-13 finish and B's hour-17 finish, including B's one-hour miss. Hiding the miss would prevent an informed decision.

The approver needs to know who accepts each consequence. Planning might check the sequence while production confirms the line conditions. Quality would check the required basis for release. A change in a customer commitment may require a different authority.

These roles are examples to adapt to the site's actual system. A single signature does not waive every technical or commercial check. A conditional approval remains conditional until the stated evidence is present.

Give the decision an effective time and a next review point. Retain rejected options and their main reasons when they explain the choice. That record makes a later review more useful than a note saying “plan changed due to urgency.”

Issue one current version to the people using it

Send the approved plan to cutting, sewing and material handling. Include any later stage whose work changes. Remove obsolete working copies from points where they could still direct work. Retain the archived version for history.

The message should identify the exact change. “Use version 8” gives too little context. Explain that B's first 300 garments move ahead of A, with the other 300 after it. Both the quantity split and its order need to be understood.

The HSE's shift-handover guide sets out three parts: preparation, two-way exchange and an incoming-team check. These communication principles help when a revised plan passes between shifts or functions. A sent file alone does not show that the receiving team understood it.

The receiving owner should confirm the next order and its quantity. Ask about setup status and any hold that still applies. Resolve a mismatch before the next start. The correct version number cannot compensate for instructions that people interpret differently.

Keep the forecast current after the decision

Approval does not make the assumptions come true. Record when material is released and when setup is complete. Track accepted output and the state of the split balance too. Compare these with the conditions used in the approved plan.

Suppose A is still on hold at the agreed check. Return that fact to the people responsible for the change decision. Do not continue publishing its earlier finish as if nothing changed. The team may need another sequence, another resource or an honest revised forecast.

At closure, retain what happened alongside what was approved. Separate a late material release from a longer setup or a lower running rate. Those are different reasons for variance and may need different improvements.

Review repeated changes over several orders. A readiness step that fails often deserves its own review. Fixing it may prevent some future schedule changes. If approved changes work but teams receive them too late, improve the handover. The change log should help the factory learn where its plan loses contact with the work.

References

• GAO: Schedule Assessment Guide, especially Best Practice 10 on baseline schedules and change control.

• Microsoft: operations scheduling, finite capacity and finite material.

• Oracle 26B: supply-plan attributes and planning time fences.

• Microsoft: operations resources, capabilities and capacity reservations.

• Lean Enterprise Institute: internal and external setup work in SMED.

• HSE: reliable shift-handover communication.