How Should a Tightening Workstation Handle NOK Results and Rework?

A tightening tool reports NOK, but the operator still has a partly assembled product in the fixture and a queue building behind it. Should they pull the trigger again, loosen the fastener, or call a supervisor? A tightening workstation needs an NOK and rework workflow that answers those questions without losing the failed result or allowing the wrong point to be worked. This guide explains how to distinguish a failed attempt from product disposition, contain the affected operation, route the cause to an approved action, verify the rework, and retain a traceable record. It also shows which workstation-control functions to specify when the existing tool, fixture, and production system must work together.

What Does an NOK Result Mean at a Tightening Workstation?

Separate the Tool Result From the Status of the Fastening Point

NOK means that a tightening attempt did not meet the conditions configured for that tool and program. The failed condition could concern torque, angle, time, the tool trigger, or another monitored event. Atlas Copco’s Power Focus NOK result list includes both out-of-limit values and interrupted attempts. One NOK event does not, by itself, tell you whether the fastener can be reworked, whether it has been damaged, or whether the product must be scrapped. Those decisions belong to the approved joint and product process.

It helps to keep three states distinct: the attempt result reported by the tool; the point status of the identified fastening location; and the product disposition for the assembled unit. A failed attempt normally leaves the affected point unresolved until an authorized action and acceptance check close it. The product may therefore be on hold even if other points have valid OK results. Conversely, a tool may report an NOK because the trigger was released prematurely, which calls for diagnosis rather than an automatic assumption that the joint has failed in the same way as an over-torque event.

Define Completion at the Point and Product Levels

For a multi-point job, a batch count is useful but is not a substitute for knowing which point produced each result. The station should associate the result with the active product, point ID, tool program, and operation attempt wherever the installed equipment supports those signals. Atlas Copco’s Live Results documentation distinguishes tightening-level and batch-level status and exposes a detailed reason for an NOK.

Define “point complete” as a validated, accepted result for the correct point—not as an alarm acknowledgment or a count increment. The corresponding product-complete rule can then require all mandatory points to be accepted and any outstanding holds to be resolved. The exact acceptance rule, including whether a subsequent attempt is sufficient, must be approved for the particular joint.

What Should the Workstation Do Immediately After an NOK Result?

Hold the Affected Task and Communicate a Clear State

When a trustworthy NOK result arrives for the active point, the workstation should retain that event and mark the point as pending resolution. For a controlled sequence, a sensible design is to prevent automatic advancement or release of the product while that mandatory point remains unresolved. Whether unrelated tasks on the same unit may continue depends on the process plan and how the line handles held workpieces; it should not be decided silently by a generic alarm routine.

Tell the operator what happened in terms they can act on: product ID, point ID, program, failure category, and required next step. A red or warning signal can draw attention, but a lamp alone cannot explain whether the operator should stop, inspect, or request authorization. A configurable KURAN KR004-3SD three-color tower light can display workstation status; the corresponding message and interlock logic still need to be defined in the control design.

Keep Acknowledgment Separate From Permission to Rework

If the operator presses reset, has the failed bolt been cleared? No: acknowledging a displayed error should remove or silence the notification only when the station’s configured conditions permit it. It should not silently change the point to OK, erase the original tool result, or grant permission to repeat the operation.

An operational state model can use OK / unresolved NOK / held for review / authorized rework / rework verification / accepted or rejected. This is a proposed design pattern, not a universal controller feature or mandatory standard. If the installed tool permits another trigger pull while the workstation says “hold,” verify the physical and logical interlocks together; a screen message without effective process control may not contain the error. Once the affected task is held, the station needs to distinguish the reason for failure.

How Should You Diagnose the Type of NOK?

Use the Controller’s Detailed Result Before Choosing an Action

A generic red NOK signal throws away information. Retrieve the tool controller’s detailed status, measured values, limits, selected program, and available trace for the attempt. Atlas Copco’s NOK code descriptions include torque and angle high or low, timeout, trigger loss, slip-off, and rehit. Its result view can display torque/time and torque/angle traces where supported. The same red indicator can therefore correspond to physically different situations.

Source basis: The event categories come from Atlas Copco’s Power Focus NOK result list and target torque limits documentation. The review actions are proposed workstation-process guidance, subject to the approved joint specification; “no valid result” is deliberately distinguished from a controller-reported NOK.

Distinguish a Process Error From a Connection Error

Before releasing the tool again, compare the product variant, point ID, program selection, tool/socket combination, and actual workpiece position. A wrong-program event may arise upstream of the fastener; simply repeating the same program could reproduce the error. A signal that never reaches the station is a different fault: the tool may have completed an operation, but the supervisory layer cannot yet substantiate its status.

Do not convert “unknown” to OK for throughput. Hold the workpiece while a defined recovery procedure checks the controller record, product identity, and any possible duplicate attempt. The detailed diagnosis leads to an authorization decision rather than directly to a second tightening.

When Can an Operator Rework a Failed Fastening Point?

Let the Joint Instruction Set the Physical Action

There is no single retightening rule for all joints. The assembly engineering and quality teams should specify, for each relevant failure category, whether to continue an interrupted cycle, loosen and retighten, replace the fastener, inspect mating threads, reassemble the joint, or quarantine the unit. A second trigger pull is a process step only if the approved instruction permits it for that failure and joint condition.

This distinction is important for over-torque. Atlas Copco’s torque-limit guidance notes that excessive torque can deform a screw or strip threads, while insufficient torque may fail to deliver the required clamping force. Those are possible mechanisms, not a diagnosis of every NOK joint. The acceptable response depends on the fastener, materials, tightening strategy, and any observed damage. Do not infer from an acceptable torque on a later attempt that an earlier potentially damaging event has been reversed.

Define Authorization, Retry Limits, and Escalation

The control specification should identify who may release a held point into rework, which reason codes each role may handle, and when a quality or engineering review is required. For example, a trained operator might be permitted to repeat a clearly interrupted attempt under an approved instruction, while an over-torque or repeated NOK might require a supervisor. These are examples of an authorization model, not recommended limits for every production line.

Make the permitted attempt count and any time window explicit in the joint-specific procedure. When the limit is reached or a forbidden failure category occurs, the station should move to a defined escalation or quarantine state rather than offering endless retries. Atlas Copco’s ToolsNet Rework documentation describes operator identification and authorization together with guided repair tasks. Once permission is established, the station can guide a controlled rework cycle.

What Steps Make a Rework Cycle Controlled and Verifiable?

Guide Only the Affected Point Through the Approved Procedure

How does the operator know which bolt to work on next? Start with the held product and point identity, not merely the last red result on a shared screen. A practical controlled sequence is:

1. Identify and lock the transaction. Confirm the workpiece ID, held point ID, original attempt ID, NOK reason, and currently loaded variant. Block selection of another point if the rework instruction requires exclusive access.
2. Authorize the prescribed route. Record the approving role or operator, the approved instruction revision, and whether this case calls for inspection, replacement, loosening, or another defined action.
3. Verify the physical setup. Confirm the correct tool, socket or bit, program, and part position. If the original issue was a wrong program or wrong accessory, correct and record that cause before rerunning.
4. Perform and capture the action. Log any loosening, replacement, or other required intermediate operation where equipment supports it, then capture the new tightening result as a new attempt linked to the original NOK.
5. Apply the acceptance rule. Accept the point only after the prescribed verification is satisfied. If it fails again, follow the configured limit and escalation path; do not overwrite the prior result with the latest one.

Desoutter describes a position-aware rework application that receives NOK screw information and guides action on the specific affected fastener. Atlas Copco’s rework workflow likewise associates product ID, required tools and parameters, operator guidance, and repair results. Those implementations illustrate useful design principles; they do not establish that every existing workstation already has these capabilities.

Close the Point Only After an Accepted New Result

The rework cycle should end with an explicit disposition: point accepted under the approved rule, still held pending further work, or rejected/quarantined. A new OK tightening may be necessary, but not necessarily sufficient if the approved procedure also requires an inspection or replacement check. Keep the original NOK, intermediate actions, and final acceptance decision as separate events.

If rework occurs at a separate station, transfer the product and unresolved point status with the unit rather than treating the transfer as completion. Atlas Copco describes a dedicated rework checkpoint that queries the product’s tightening history and returns repair data. At the original or separate station, reliable point identity is the next control requirement.

How Do Point Identification and Position Control Prevent Wrong-Point Rework?

Bind the Result to a Fastener, Not Just a Count

Similar fasteners in one fixture create a practical risk: the operator may repeat the failed action on a neighboring point or move to a new point while the NOK remains unresolved. Map every mandatory point to a unique ID, its allowed tool-position region, applicable variant, and tightening program. When the position method is accurate enough to distinguish adjacent points under real fixture variation, it can be used to enable the tool only at the held point during rework. Position sensing confirms a location condition; it does not itself measure tightening torque or decide whether a damaged joint is acceptable.

Desoutter’s nexonar example explicitly addresses reworking only the identified NOK screw. For a KURAN-based station, the KR002-B01 field controller provides field I/O and multi-point sequence control, which can be evaluated as part of the point-enable and status logic. Whether a particular arm or positioning arrangement can resolve the real fastener spacing must be validated on the application; do not assume that any arm automatically provides reliable point identification.

Check the Edge Cases During Commissioning

Test neighboring points, the tool positioned between regions, product changeover, a moved fixture, and a rework attempt started from the wrong point. Review what happens when the operator removes and reinserts the socket without completing a tightening. The station should not award a point OK merely because the tool was in the right place when an unrelated OK signal arrived. The position, product, program, active transaction, and result must refer to the same authorized operation.

With the correct point controlled, the next question is whether the evidence of both the failed and repaired operations can be reconstructed later.

What Records Must Link the Original NOK to the Rework Result?

Preserve the Full Event Chain

For each affected point, capture the product or serial identifier, variant, point ID, timestamp, tool and program IDs, attempt number, measured result and limit status, detailed NOK reason, and available torque/angle trace or its retrievable reference. Then link the approved rework instruction, operator or approver identity, intermediate action, replacement-part information where applicable, new result, final point disposition, and final product release decision.

Keep the failed record rather than replacing it with a later OK. Atlas Copco’s ToolsNet Rework description describes consolidating production and rework data by product ID. Its ToolsNet 8 information discusses traceability and analysis of tightening data. These examples support linking events, while the precise fields and retention period should follow customer, contractual, regulatory, and internal requirements.

Define What the Field and Supervisory Systems Own

A tool controller may own the measured tightening result; a field controller may manage point and device I/O; a supervisory system or MES may associate those events with a serial number and final product disposition. Agree on which system is the source of truth for each state and what happens if its message is delayed or unavailable.

KURAN’s KR002-H01 supervisory controller describes per-point result collection and connections to higher-level systems. That makes it a relevant product page when a buyer needs workstation-level traceability and coordination. Confirm the actual interface, event fields, buffering behavior, and integration scope for the tool and MES in the proposed configuration. This leads directly to the recovery behavior needed when one of those links is interrupted.

How Should the Station Recover From Missing Results or an Interrupted Cycle?

Treat an Unknown Transaction as Unresolved

A cable disconnect, controller restart, lost network message, or power interruption can leave the station without a valid result even if the tool physically contacted the fastener. “No result received” is an uncertainty state, not evidence that the attempt was OK or that the tool controller itself classified it as NOK. Hold the affected product and point until the station can reconcile the tool’s stored result with the active product, point, program, and attempt.

On restart, restore the last committed status and unresolved holds before enabling the next operation. If the tool result can be retrieved unambiguously, link it to the original transaction; if it cannot, use the approved inspection or rework route. Avoid creating a new OK event by replaying a delayed message or counting the same tightening twice. The exact implementation depends on the controller’s event IDs, local storage, protocol, and network architecture.

Test Recovery as a Scenario, Not a Checkbox

During commissioning, deliberately test a lost result immediately before and after the tool completes its cycle, a restart with a held point, an operator reset during an NOK, and an MES outage while a repaired unit awaits release. Record which system can still identify the part and the active point, who may unlock it, and how duplicate results are detected.

Atlas Copco’s Live Results guide shows that a controller can retain more detailed tightening information than a simple lamp or batch indicator displays. Use the recovery test to prove how that information is retrieved and matched in the actual installation, not to assume that all controllers expose the same fields. A tested recovery protocol becomes an essential part of the supplier specification.

What Should You Specify Before Buying or Retrofitting a Rework Solution?

Start With the Failure and Acceptance Matrix

Bring a list of product variants, fastening point IDs, tightening programs, tool/controller models, NOK reason codes, and the approved response for each cause. Add the roles allowed to authorize rework, maximum permitted attempts where applicable, product-hold conditions, and the exact evidence required for point acceptance and release. This matrix is more useful to a supplier than a request for a generic “NOK alarm.”

Then describe the physical workstation: fixture and point layout, tool access, existing torque reaction arm, available sensors, socket or bit changes, I/O interfaces, network/MES requirements, and expected behavior during signal loss. If you are retrofitting position control to an existing workstation, document the current controller’s available result and enable signals before assuming that a new field device can infer everything from them.

Match the Control Layer to the Actual Gap

A local multi-point station may primarily need point-enable logic, field I/O, and visible feedback. A multi-station process that must link serial numbers, authorization, results, and final disposition may need supervisory coordination and a defined interface to MES. KURAN’s KR002-B01 field controller and KR002-H01 supervisory controller address different layers; the KR004-3SD status light can make the chosen state visible. These products should be evaluated against the specified workflow, compatible tightening tool, and integration tests; their product pages do not replace a joint-specific rework procedure.

For a useful first discussion, send KURAN the point map, controller model/protocol, sample OK and NOK records, failure-and-action matrix, and an example of how a held product is released. You can use KURAN’s contact page to request an integration review focused on the actual NOK and rework path rather than only the hardware list.

Conclusion: Close the Loop at the Point, Not the Alarm

An effective workstation response to NOK has five linked decisions: identify the failed attempt, hold the affected point, determine the permitted action, verify the rework, and retain both results with the product record. The appropriate physical remedy and acceptance criteria vary by joint, while the controls must match the installed tool and line architecture. When those decisions are specified before selecting equipment, the operator gets a clear path through the exception and the buyer can evaluate a solution against an observable outcome: no unresolved point is silently released as complete.

FAQ: NOK Results and Rework at Tightening Workstations

Should rework use the same tightening program as the original cycle?

Use the program specified in the approved rework instruction; it may or may not be the original production program. The correct choice depends on the joint’s condition and any loosening, replacement, or inspection step. Atlas Copco’s ToolsNet Rework description includes selecting the tool and parameters required for the repair. Record the actual program ID with the new attempt.

How should a station handle several NOK points on one product?

Maintain a separate unresolved status and action history for each affected point. The product remains on hold until every mandatory point has an approved disposition; the order of several rework actions should follow the product’s assembly instruction. Atlas Copco’s ToolsNet Rework example compares a product’s recorded results with required tasks and guides the remaining repair steps.

Does ISO 9001 prescribe a universal number of rework attempts?

No universal tightening retry count should be taken from ISO 9001. ISO 9001:2026 is a quality-management-system requirements standard, not an application-specific fastener program. This is an inference from the scope of the standard, not a substitute for reviewing its full licensed text. Define attempt limits from the joint design, approved work instruction, and relevant customer requirements; document who approves exceptions.

Can a rejected tightening point be reworked at a separate station?

Yes, if the transfer preserves the product ID, unresolved point, approved task, and original result. A dedicated station must return its inspection or tightening results before the product is released. Atlas Copco’s ToolsNet Rework example describes a checkpoint that reads production data by product ID and records repair operations.

How long should NOK and rework data be retained?

There is no single retention period that fits every fastening application. Establish it from applicable customer contracts, product and regulatory requirements, internal quality policy, and the capability of the storage system. Preserve the link between the original NOK and the approved rework for the selected period; the Atlas Copco ToolsNet 8 overview describes traceability and data analysis but does not set a universal retention duration.

What if the tool shows OK but the workstation receives no result?

Keep the point unresolved until the two records can be reconciled. Check the tool’s stored result, product and point identity, program, timestamp or event ID, and whether a delayed message or duplicate attempt is possible. A controller’s detailed live-result views illustrate the information that may be available; recovery depends on what the specific installed tool and interface actually expose.