Manufacturing Rework Control System: A Thailand Implementation Guide

When introducing a rework management system in manufacturing, it is not enough to simply record “reworked” on a screen. You must be able to track the process from isolating nonconforming items from good stock, to who authorized which action, what was fixed according to which version of the procedure, and which inspections were passed before the item became eligible for shipment. In Thai factories, especially when headquarters, quality, production, and warehouse departments have differing opinions, unclear system boundaries often leave paper or verbal communication on the shop floor. This article clarifies state transitions and data integration for selection, requirements definition, estimation, and acceptance testing.

Rework Management Is Not Just “Defective Item Recording”—It Controls Shipment Eligibility

Rework refers to additional work performed on nonconforming products or work-in-process to restore them to specified requirements. It is distinct from concession/use-as-is (special approval to use or ship items that do not meet requirements), repair (which may involve specification changes), and scrap. Since terminology varies by customer contract and industry standard, requirements definitions should specify which requirements are met after each action and whose approval is needed, rather than focusing on names.

While registering nonconformities can be handled by a Quality Management System (QMS), not all systems can consistently manage the movement of physical items, reintegration into processes, recording of parts and labor, re-inspection, and restoration of inventory status. Conversely, a Manufacturing Execution System (MES) may record rework results but miss approvals from the quality or customer side. The focus should not be on “adding new systems,” but on deciding which system (QMS, MES, ERP, WMS) holds the master record and which state changes are communicated to other systems.

ISO guidance on quality management emphasizes records of the nature of nonconformity, actions taken, concessions, and decision-makers. While this does not mandate specific software, it justifies designing systems to explain approval histories. GS1 traceability documentation also recommends recording the relationship between manufacturing lots or serials and input/output when reworked intermediates are traceable. Conformance to standards is determined individually based on customer requirements and audit scope.

First, Decide on the “Scope” and “Isolation Unit”

If you only target finished goods, defects found in work-in-process remain on paper ledgers outside the process. Nonconformities in raw material receipt, in-process dimensions, post-assembly functional tests, or pre-shipment inspections all involve different owning departments and permitted actions. Decide how to identify items—by part number, serial, lot, container, or pallet—based on detection points and downstream impact.

For example, if 7 out of 100 items in a lot require rework, treating the entire lot as one unit forces you to either hold all 93 good items or risk shipping the 7 nonconforming ones as good. This is an illustrative example for explaining quantities, not a recommended value. If you handle partial lots, you need rules linking parent lots, split child lots, individual IDs, quantities, and storage locations, ensuring the sum before and after splitting matches. Even if you use the same part number after rework, you must retain the relationship between the original and reworked items in the history.

For large items that cannot be physically isolated, system holds alone may not suffice. Combine tags, floor markings, shelves, container seals, or equipment-side input bans so operators can visually confirm item status. Conversely, physical tags alone cannot stop ERP allocation or WMS picking. Define isolation completion as the point when item display, storage location, inventory status, and shipment ban are all consistent, not just when a tag is attached.

Manufacturing Rework Control System: A Thailand Implementation Guide - figure 1

Designing State Transitions from Nonconformity Detection to Inventory Restoration

At a minimum, states can be divided into: ① Detection, ② Hold, ③ Disposition Review, ④ Rework Approval, ⑤ Rework in Progress, ⑥ Awaiting Re-inspection, ⑦ Re-inspection Passed, ⑧ Ready for Shipment. Branches include rejection, scrap, concession, supplier return, or awaiting customer approval. Companies do not need to use the same names, but you must define “who can move to which next state, based on which input” for each state, or it becomes just a status field.

When detecting nonconformity, record the detection date/time, equipment/process, inspection item, measurement result, specification or criteria version, quantity found, suspected range, and discoverer. If measurement values are manually entered, keep the operator and edit history; if acquired from instruments, record instrument ID and timestamp. Selecting just one defect code may not be enough to determine re-inspection items later. Allow free text, but separate structured items needed for analysis.

When moving to hold, cross-check item ID and inventory ID to confirm the relevant quantity is removed from available inventory. If ERP, WMS, and MES update at different times, clarify which system’s hold is the final shipment gate. If communication fails, do not display “hold applied”; leave it in an exception queue as a sync failure. Use event IDs and idempotent processing to prevent double reduction of quantities due to duplicate transmissions.

Define roles for quality, manufacturing engineering, production, and, if needed, the customer during disposition review. Do not use the same approval button for both standard rework and exceptional actions without procedures. For automotive parts with customer-specific requirements, unplanned rework may require prior approval. For example, Stellantis’s published customer-specific requirements state that planned rework must be included in process flows, FMEA, and control plans, while unplanned rework requires prior approval. Do not generalize this requirement to other products; always check the applicable contract.

After rework approval, derive work instructions from the original manufacturing order. Decide whether to return to the original process, move to a dedicated cell, limit the number of rework cycles, and where to record added/replaced parts, scrap, labor, equipment, and operators. Re-inspection after rework may require more than just repeating the original inspection. Re-evaluate characteristics affected by rework and include additional inspections or functional tests in the inspection plan if needed.

Restoration to shipment eligibility is not triggered by “rework completed,” but only after all specified re-inspection results are in and an authorized person releases the item. If WMS hold release, ERP inventory status change, MES item state, or QMS disposition completion are delayed, different screens may show different conclusions. Confirm the processing order between systems and correction procedures for failures during acceptance testing.

Do Not Confuse Rework with Repair, Concession/Use-as-is, or Scrap

The purpose of rework is to restore original requirements. Repair may restore usability but not all original requirements; concession/use-as-is is an authorized decision to use or ship a known nonconforming item. Scrap prohibits reintegration and may include processes to render items unusable. It is not enough to simply list these options on a screen; define separate approvers, re-inspection, customer notification, labeling, inventory status, and accounting linkage for each action.

For the same item, if you restore it to specification by additional machining, it is rework; if you use it with customer approval under an alternative specification, it is closer to concession. These examples are for classification purposes only and do not substitute for legal or regulatory judgment. Collect real cases where factory-side opinions differ, and clarify priorities among customer drawings, contracts, quality agreements, and internal procedures with the quality manager.

Changes in disposition must also be recorded in the history. If an item initially planned for rework fails re-inspection and is switched to scrap, do not overwrite the original approval—record the reason for invalidation, new approver, item quantity movement, and additional labor in chronological order. Saving only the “final status” distorts analysis of defect rates and rework costs.

Role Distribution of QMS, MES, ERP, and WMS

QMS typically handles nonconformance reporting, root cause analysis, disposition decisions, approvals, and corrective actions. MES handles process routes, work records, reintegration, and re-inspection at the shop floor. ERP manages manufacturing orders, items, costs, accounting, and inventory valuation. WMS manages locations, item movement, and picking bans. However, boundaries differ by product, so do not assign responsibility by product name alone. Create a “responsibility matrix” specifying the master for each data item.

Recommended minimum events are: NONCONFORMANCE_OPENED, HOLD_APPLIED, DISPOSITION_APPROVED, REWORK_STARTED, REWORK_COMPLETED, REINSPECTION_PASSED, RELEASE_APPROVED. These are example names, but the key is to include event ID, target item, quantity, timestamp, operator, approval version, source, and related prior event. Ensure different systems do not double-update when receiving the same event.

Handling integration failures is central to implementation decisions. For example, if a hold is confirmed in the quality system but ERP connectivity is lost immediately after, the quality screen shows hold, but ERP still allows allocation. In RFPs, ask not only “Can you resend?” but also “Which data stops the shipment gate?”, “Who is notified of sync failures and within how many minutes?”, and “How are quantities reconciled after recovery?” Set notification times according to your operation and risk, and decide based on actual measurements.

The article How to Eliminate Duplicate Data Entry in Thai Factories organizes master records and synchronization for MES, ERP, and WMS. Implementing an Outsourced Processing Control System in Thailand covers design for reconciling issued and received quantities at external processes. If rework is outsourced, connect internal hold/disposition approvals and external handover using both articles’ perspectives.

Manufacturing Rework Control System: A Thailand Implementation Guide - figure 2

Implementing Traceability and Lot Splitting

Even for simple rework without adding raw materials, you must be able to trace when, under what conditions, by whom, and under which inspection criteria the item was handled. When parts are replaced, mixed, repackaged, relabeled, or lots are split, you need to link input and output identifiers. The GS1 EPCIS TransformationEvent is a standard concept for representing input-output relationships. This does not mean your rework process must be implemented in EPCIS, but it is a useful reference for designing input-output record linkage.

Quantity reconciliation can be explained, for example, as: received 100, isolated 7, rework completed 6, scrapped 1, restored to good stock 6. However, the numbers alone from “isolated 7” to “restored 6” do not tell you which 1 was scrapped or which 6 underwent which procedures. For products requiring individual identification, track by serial; for lot-managed products, track by split lot. If items are reworked multiple times from the same process, save the count and cause. The numbers here are illustrative, not actual or standard values.

For post-shipment defect investigations, you need both “reverse lookup” to search only reworked items and “forward lookup” to trace destinations from a specific lot. If you only look at standard flow defect rates, reworked items become indistinguishable from regular items. Test whether you can reproduce links among added materials, re-inspection results, customer approvals, and shipment lots with sample data.

Estimating Costs and Benefits

The cost of a rework management system is not determined by the number of screens. First, check how much of your existing QMS, MES, ERP, and WMS can be used, and separate out additional functions. Estimates should include item ID granularity, terminals/scanners/labels, multiple sites and languages, approval flows, process route branching, data migration, interfaces, offline processing, audit logs, maintenance, and training. Individual prices depend on connection environment and contract terms; do not rely on baseless standard rates.

ROI cannot be evaluated solely by claims like “defect rate will decrease.” Compare current and post-implementation values for hold release lead time, shipment error prevention, rework labor, re-inspection waiting, recurrence of the same defect, scrap rate, WIP, and investigation time. Even if quality data visualization speeds up root cause analysis, defects will not decrease unless improvement actions are implemented. Report system effects and process improvement effects separately.

For example, in a factory with 200 nonconformance tickets per month, if 40 proceed to rework, it is easier to measure detection-to-release time, rework loops, and inventory discrepancies for rework cases before investing in expensive individual serial management for all cases. The numbers are hypothetical to illustrate estimation methods. For actual investment decisions, use your own case counts, labor rates, customer requirements, audit risks, and system modification costs.

Functions and Data Items to Specify in the RFP

In RFPs, avoid vague statements like “must manage nonconformities” and include actual abnormal scenarios as business cases. At a minimum, cover cases such as partial lot failures in in-process inspection, holding at material receipt, tracking items already sent to the next process, requiring approval for unplanned rework, re-inspection failures, syncing after communication loss, and shortages on return from subcontractors. You do not need to require all cases that do not apply to your company.

The function list should include: ① Item/lot/quantity identification, ② Hold and shipment ban, ③ Disposition types, ④ Approval authority, ⑤ Procedure versioning, ⑥ Input materials and labor, ⑦ Re-inspection planning, ⑧ Release authority, ⑨ ERP/WMS inventory sync, ⑩ Search and audit logs. For each, require a classification of “standard function,” “configurable,” “custom development,” or “not supported,” and request demo evidence. A checkmark in the spec sheet alone does not guarantee feasibility.

Also specify data ownership in the RFP. For example, nonconformance ticket number in QMS, actual work start/end in MES, allocatable inventory in ERP or WMS, customer approval number in QMS—specify by item. Ask not only about API or file connections, but also about identifier mapping, resend order, duplicate elimination, failure correction, and audit data retention periods.

Do not decide vendor selection by price alone; test with sample data whether disposition approval and item control work together. Even if existing products lack features, it may be easier to add peripheral workflows or event integration than to deeply modify the core product. However, the more boundaries you add, the more complex responsibility becomes during failures, so agree on division of responsibility among the implementing company, operations, and quality manager before contracting.

Acceptance Criteria for PoC, FAT, and SAT

For PoC, select representative cases that cross state and system boundaries, not just the most frequent nonconformities, to better detect defects. Run through the entire flow: isolate a partial lot, approve disposition, issue rework instructions, record input parts and labor, pass only some items in re-inspection, and restore only passed items to shippable inventory. Also check that items failing re-inspection return to hold and that history is not lost.

For FAT, check role permissions, screens and reports, APIs, quantity invariance, event duplication, communication loss, timestamps, and audit logs. For example, if an operator tries to start rework before quality approval, the expected result is prohibition. If another operator opens an outdated procedure version, confirm that execution is blocked or that there is clear guidance to the updated version. Set test quantities and times according to factory conditions.

For SAT, check actual factory terminals, barcodes, labels, wireless, equipment operation, physical storage, and multilingual screens (Thai, Japanese, etc.). Intentionally create mismatches—such as a screen showing hold but no tag on the item, or an item physically isolated but still allocatable in ERP—and test who notices and how recovery occurs. Do not enable the production shipment gate just because sample data succeeded.

Acceptance decisions must be documented, including allowable exceptions, on-site recovery time, manual fallback procedures, and responsible persons. “The screen worked” alone does not constitute acceptance. Completion criteria require reproducing the business scenario from start to finish, with matching records for target items, quantities, approvals, inspections, inventory, and shipments.

Manufacturing Rework Control System: A Thailand Implementation Guide - figure 3

Common Pitfalls and How to Avoid Them

First, do not equate completion of the quality ticket with item release. Even if disposition is decided, shipment is prohibited until rework and re-inspection are complete. Separate states and segregate change authority. Second, if item IDs are reassigned mid-process, the link between the original nonconformity and the reworked item is lost. Test parent-child relationships during label reissue, splitting, and repackaging.

Third, if you only create standard rework routes, approvals for unplanned actions may remain on paper. Even rare but high-risk exceptions must be included in the design. Fourth, if re-inspection is only recorded as “pass/fail,” you cannot track which items were rechecked against which version of the criteria. Ensure you can trace reasons for additional inspections and measurement values. Fifth, do not treat system sync failures as successes. Especially for hold and release, make failures visible and design shipment gates to default to the safe side.

Implementation Sequence and Operations in Thai Factories

When headquarters sets quality procedures and the Thai factory handles physical items, language differences are not just a display issue. If “hold,” “rework,” “repair,” “concession,” and “release” mean different things to different departments, the same case may be reported in different states. Create a glossary in Japanese, English, and Thai, and agree on entry conditions, actions, and exit conditions for each state in a single table. Use short terms for shop floor screens, but leave sufficient explanations for approvers.

In the early phase, do not just pick processes with the most cases; choose processes where the boundary between quality and warehouse is clear. For example, for post-assembly inspection failures, you can cycle through moving items to the hold shelf, deciding disposition, sending to the rework cell, re-inspecting, and returning to the good shelf. Expanding to all sites, items, and actions from the start makes it hard to define exceptions, and the shop floor may revert to paper. For cases excluded from PoC, list the conditions for moving to the next phase.

Divide roles into discoverer, isolation handler, disposition reviewer, rework approver, operator, inspector, release approver, and inventory adjuster. Even if one person has multiple roles, log which authority was used for each action. If independence of quality assurance is required by contract or internal rules, set permissions so the person completing work cannot self-approve final release. For night shifts or holidays when approvers are absent, predefine proxy approvers and approval deadlines to avoid unauthorized progression to the next process.

When migrating from paper, handling items already on hold before migration is critical. If all inventory is automatically considered “good” on the new system’s go-live date, nonconforming items from old ledgers may be restored. Conduct a physical inventory before switchover and reconcile open nonconformance tickets, held lots, pending customer approvals, and pending scrap. Record the original ticket number and migrator in migration data to distinguish pre- and post-migration cases.

Training should cover not just operations but also why holds must not be released, and why “completion” after rework is separate from “shipment eligibility,” using actual items. Training for expected errors is also effective. Define what operators should do and who to contact when scanning outdated procedure barcodes, scanning labels from other lots, or losing connectivity. Even after operations are established, monitor the number of cases in the exception queue and parallel paper records.

Dashboards: Go Beyond Just “Defect Rate”

Nonconformity rate is important, but bottlenecks in rework management often lie in approval or re-inspection waiting. Aggregate dwell times by status: from detection to hold, hold to disposition approval, approval to work start, completion to re-inspection, and re-inspection to inventory restoration. Averages alone hide long-stalled cases, so also show distribution and oldest cases. Do not blame all delays on the shop floor; separate customer approval waits and parts shortages.

Fix the denominator for rework rate calculations. The value changes depending on whether you use input quantity, finished quantity, or inspected quantity. Also distinguish whether you count a second rework on the same item as two cases or as one item. Mixing management and process improvement metrics under the same name leads to errors in site comparisons. Clearly define item definitions and aggregation periods on the dashboard, and record the impact on past values when changes are made.

For cost analysis, look not only at rework labor but also additional materials, re-inspection, equipment occupancy, outsourcing costs, scrap, and WIP dwell. Immediately after adding new measurable items, apparent rework costs may increase as previously invisible costs become visible. When evaluating improvement effects, separate changes in measurement methods from actual process improvements. For root cause analysis, distinguish between the process where the nonconformity was “detected” and where it “may have occurred.”

For quality departments, it is not just the most frequent defect codes that are useful. Combining the rate of rework failures, number of re-dispositions on the same lot, links to the same equipment/mold/material lot, and customer approval wait times makes it easier to discuss whether the issue is insufficient work instructions or unstable process conditions. Even if correlations are found, do not assume causality—return to on-site checks and validation of measurement conditions.

Deliverables to Provide at Implementation

If you only request a screen list from the vendor, successors cannot read state and responsibility. At a minimum, require a state transition diagram, authority table, item ID rules, master data table, system event list, recovery procedures for integration failures, inspection plan version control, audit log retention policy, RFP responses, and FAT/SAT results. Retention periods differ by product law or contract, so have your quality/legal departments decide, and ensure the system allows configuration and extraction.

Attach a flowchart separating standard nonconformities and exceptional actions to shop floor procedures. For temporary paper records when screens are unavailable, decide on numbering, hold display, shipment stop, post-recovery input, and double-entry checks. If you stop the system without a paper fallback, night shifts cannot determine item location or shipment eligibility. Conversely, running paper records indefinitely in parallel creates two masters, so clarify which is the final record after recovery.

Include change management after go-live in the design. If customer specs or inspection criteria change, you must decide whether to apply the old or new version to items already on hold. Do not overwrite procedures; record effective start date/time, target items/customers, and approvers, and approve switching conditions for ongoing cases. After changes, retest representative scenarios to confirm rework approval and shipment gates work as expected.

Maintenance contracts after acceptance should cover reconciliation of items and inventory during failures, regression testing for interface changes, label spec updates, periodic authority checks, and log extraction support. If only the person in charge manually assembles histories before a quality audit, the benefits of implementation will not last. At monthly operations meetings, review unresolved integration errors, long-term held items, proxy approvals, and attempts to use outdated procedures, and improve both procedures and settings.

Frequently Asked Questions

What factors affect the cost of a rework management system?

Costs depend on existing QMS/MES/ERP/WMS functions, number of integrations, item identification granularity, approval and customer requirements, terminals/labels, multilingual support, and offline capability. Start by selecting one current flow and request estimates separating additional functions and integrations needed from hold to release. Price alone does not allow comparison of data migration or exception handling scope.

Can we start from an Excel nonconformance ledger?

Yes, you can. However, do not simply digitize the ledger—define item ID, quantity, disposition authority, re-inspection, inventory status, and shipment ban. Limit the initial scope to one process or item group, and set a period and exit condition for parallel operation of paper/Excel and the system to ease migration.

Should lot numbers change after rework?

There is no one-size-fits-all answer. Decide based on product standards, customer requirements, individual identification, rework content, split quantities, and traceability scope. Whether you change the number or not, it is important to be able to trace the original item, disposition, input materials, re-inspection, and shipment destination. Decide in consultation with the quality manager and customer contract.

Should we modify QMS or MES first?

If approval history for nonconformities is missing, start with QMS; if you cannot track physical reintegration and work records, start with MES; if shipment bans are ineffective, start with ERP/WMS. Trace a single nonconformity case from the shop floor and address the most critical control gap first. Do not fix system priorities from the start—this helps avoid double entry.

Conclusion

The success of a rework management system depends not on digitizing rework tickets, but on seamlessly connecting item isolation, disposition approval, procedure versioning, re-inspection, inventory restoration, and shipment gates. Start by tracing a real nonconformity case from detection to shipment eligibility, and identify where people are bridging gaps with paper or verbal communication. Translate these findings into state transitions, responsibility matrices, RFPs, and FAT/SAT acceptance criteria to make product comparisons concrete.

Even when organizing QMS, MES, ERP, and WMS boundaries or rework approval routes in Thai factories, you can consult TOMAS TECH. We will work with you to define the minimum target scope and validation scenarios based on your current nonconformance tickets and item flows.