Freeze the remaining gap before discussing equipment

The previous article solved an operating question: which hours could be recovered without capital expenditure? This article starts only after that recovery has been repeated by relevant shifts and the remaining loss list has been signed.

A common mistake is to begin with a preferred solution. One team asks suppliers to quote a servo retrofit. Another assumes a camera will solve quality loss. A third requests a complete new line because the residual gap still looks large. The request should begin with the unresolved function, not the hardware name.

• State the remaining qualified packed-output gap under the best repeatable schedule.
• Name the station or interface where each hardware-bound loss appears.
• Separate mechanical wear, control limits, quality detection and downstream capacity.
• Record the longest acceptable shutdown window and the production plan that must cover it.

The number carried forward is not permission to multiply the Part 3 recovery twice. Use the verified remaining gap once, then evaluate which physical boundary prevents further recovery.

Compare three investment paths, not two price tags

The lowest purchase value can carry the highest integration uncertainty. Retrofit proposals often look smaller because interface engineering, restart scrap, extended commissioning and production cover are not yet visible in the quotation.

A new line can also be the wrong answer. If the residual loss is confined to one downstream constraint and the main line produces stable, qualified output, replacing the full process spends capital on sections that are not limiting the factory. The comparison must therefore follow the loss path station by station.

Mechanical boundaries decide whether the old frame can carry the new duty

A retrofit requires a stable physical reference. Check frame condition, shaft alignment, bearing seats, transmission wear, guarding, access, foundation and the mounting geometry of the affected station. New controls cannot correct a moving mechanical baseline.

If the unresolved loss appears as position drift at final cutting, do not assume the cutter itself is the only candidate. The product may arrive late because an upstream web path, transfer or waistband relationship has moved. Replacing the last drive can relocate the visible defect without removing its source.

In most cases, retrofit remains credible when the retained frame and transmission can hold the required relationship, access can be made safe, and the changed function has a defined mechanical boundary. Replacement becomes more defensible when several dependent stations need simultaneous redesign or when the retained baseline cannot be measured repeatably.

Control architecture is an interface map, not a servo count

List the installed PLC, drives, encoders, I/O, communication networks, recipe structure, alarm history and available program documentation. Then mark every signal that the proposed change must read, write or track. A module without a data handshake becomes an isolated island.

For a bounded retrofit, confirm who owns the source program, whether spare I/O and network capacity exist, how recipes are versioned, and what happens during a brief stop or power recovery. For an added module, define line-ready, module-ready, fault, bypass, reject, speed reference and product-position signals before the commercial scope is frozen.

The real question is not whether new automation can be installed. It is whether the retained system can prove which product is where, which recipe is active and which device owns the next action.

Quality risk determines whether a module is enough

An added inspection or reject module is appropriate only when the defect can be observed at a stable point and the affected product can be tracked to a confirmed reject location. If the defect is created upstream by unstable formation, tension or registration, detection may contain the symptom without correcting the process.

Define the quality boundary before selecting hardware: defect type, detection point, minimum evidence, reject timing, confirmation method, data record and offline sampling responsibility. Part 5 will turn that output into an inline-inspection scope.

This is where many apparently inexpensive retrofits become weak. The quotation includes a sensor or camera, but not the defect dictionary, product-tracking logic, controlled bypass or verification that the rejected item actually left the line.

The shutdown window belongs in the investment scope

A retrofit touches an operating asset. Survey access, isolation, dismantling, installation, rewiring, dry testing, material trials, first-good release and contingency rollback all consume calendar time. The plan must distinguish physical installation from the time required to return qualified packed output.

A module can be preassembled and tested outside the line, but final interface work still depends on the installed machine. A new line requires a larger project window, utilities and floor space, yet it may protect the existing line during build and reduce the risk of modifying the factory's only productive asset.

Do not compare paths with a single delivery date. Compare the date of last good product before work, the date of first good product after work, and the date the accepted output becomes repeatable across shifts.

Run an interface audit before asking for the final quotation

The audit should place the proposed change inside the real line. Use current drawings, electrical backups, photographs, material routes, SKU records and defect evidence. When documentation is missing, mark the uncertainty as a survey task rather than filling it with an assumption.

Use a decision matrix that can reject all three paths

The matrix should not force a winner. If no path has sufficient interface evidence, the correct status is “survey incomplete.” That protects the factory from turning uncertainty into a cheap-looking quotation.

Do not score purchase cost alone. The commercial comparison should include survey completeness, retained-condition responsibility, required production cover, commissioning boundary and the consequence if acceptance is not reached.

Finish Part 4 with an investment path and acceptance boundary

Part 4 should not end with “retrofit is cheaper” or “a new line is safer.” It must produce a selected path, a list of interfaces still requiring verification, explicit non-retainable boundaries and target acceptance indicators.

• Selected path and the losses it is responsible for removing.
• Retained equipment and interfaces that remain subject to survey.
• Boundaries that cannot be retained or cannot be guaranteed.
• Acceptance units: qualified output, defect evidence, reject confirmation and repeatability by shift.

That output allows Part 5 to answer the next question: which defects need inline detection, what signal should trigger rejection, and which checks still belong offline?

Related project routes

Conclusion: buy the smallest complete solution, not the smallest quotation

Retrofit, an added module and a new line are not three price levels for the same answer. Each path retains a different amount of mechanical condition, control logic, quality risk and shutdown exposure.

The correct path is the smallest complete solution that can remove the signed hardware-bound loss and pass a measurable acceptance boundary. Which part of your proposed upgrade still depends on an interface that nobody has measured?

FAQ