To troubleshoot a recurring packaging-line fault, define the symptom, preserve the relevant records, align the event timeline, separate observations from possible causes, and document the next evidence needed. A final alarm identifies what was reported; it does not, by itself, establish why the line stopped.
1. Start with a bounded symptom
Write down the asset, operating context, reported behavior, and time window. “The line keeps failing” is hard to investigate. “The sequence remained in its waiting state after a start request during this recorded interval” gives the team something specific to check.
Identify who reported the symptom and how it was observed. A technician’s direct observation, an operator’s account, and a controller export are different kinds of evidence. Keep those labels in the record. Leave duration, frequency, and physical behavior unknown when they have not been established.
2. Gather the records around the event
Begin with approved copies of information the plant already holds. Capture enough time before and after the reported event to see relevant changes; choose the window from the process context rather than an arbitrary fixed number of seconds.
- Alarm and event exports: original timestamps, identifiers, messages, and status changes.
- Machine-state records: relevant state, request, command, and feedback signals with their source and quality.
- Maintenance history: the prior symptom, work performed, and evidence that the outcome was verified.
- Approved references: the applicable manual or site procedure, including its revision and the relevant section.
Keep original files intact and link each extracted record to its source. Note missing intervals, changed configurations, and unknown clock settings. Those limits can explain why apparently conflicting records do not line up.
3. Rebuild the sequence before naming a cause
Put events on a common time basis, retaining the original timestamps. Check whether a timestamp describes the source event or receipt by another system. If two systems’ clocks cannot be aligned, show the uncertainty instead of presenting their events as an exact order. Our alarm history analysis guide explains these checks.
Write an observation next to its supporting record. Put interpretations in a separate hypothesis list. For each hypothesis, identify the evidence that supports it, anything that conflicts with it, and the missing information that would distinguish it from an alternative. A repeated alarm label alone does not prove that a previous repair applies.
4. Worked example: a start request that does not advance the sequence
This is a synthetic example from the Reliabotix demo, with no equipment connected. The export reports a start request changing to TRUE. It reports the sequence as WAIT_START at the same timestamp and again one second later, with its run command FALSE.
The supplied records show a request without a recorded transition out of WAIT_START during this interval.
Unresolved questionThe existing transition logic, command-source binding, and detailed diagnostics are absent. The exact cause and physical operating condition remain unconfirmed.
The next evidence request is the relevant existing transition logic and binding, reviewed through approved read-only engineering access. An incorrect binding or unmet condition is a possibility to investigate, not a fact established by the export. The sample does not justify a reset, a logic change, or a claim that a repair succeeded. Explore the Reliabotix example.
5. Leave the next shift a useful handoff
Record the current symptom, the evidence reviewed, the open hypotheses, and one named next step with an owner. When work has been performed, keep the corrective action separate from its verification. “Running again” does not automatically explain the original fault or demonstrate that it will not recur.
Use the maintenance incident report template to retain source references, unresolved questions, and review decisions. Mark cause and restoration status separately so an incomplete diagnosis is not silently converted into a verified past fix.
Physical servicing belongs under the plant’s approved procedures and authorized personnel. OSHA’s hazardous energy guidance describes why unexpected energization and stored energy require control during maintenance. This evidence checklist is not a servicing or restart procedure.
PUT ONE FAULT ON RECORD
Bring the history. Start with the evidence.
Tell us which recurring fault costs your team time. A first review can start with sanitized records and no live plant connection.
Review a recurring fault ↗