Troubleshooting Checklist

The checklist

1. 1

   Define the problem

   Write it down: what is wrong, on which equipment or tag, since when, how often, and what is working normally around it. A level that reads high on one wet well since Tuesday, with the other stations normal, is a different search from every level reading high since the server was patched.

2. 2

   Ask what changed

   Change records, the vendor session log, the alarm journal for the hour before, the weather, the maintenance schedule, the last person in the panel. If something changed, start there.

3. 3

   Make it safe

   The affected equipment in a known state: manual, off, or hand at the panel. The manual operation procedure ready. Lockout where hands go into equipment. Operations told what is being done.

4. 4

   Gather the evidence

   Trends of the affected tag and the related ones over the period; the alarm journal; the controller fault log; the SCADA event log; the communication statistics; the switch and firewall logs if the network is suspect. Look before touching.

5. 5

   Divide the chain

   From the field device to the display: the device display, the loop current, the controller input, the controller tag, the SCADA tag, the historian, the client. Compare the value at each layer; the fault is between the last true layer and the first false one.

6. 6

   Form one hypothesis

   From the evidence, the most likely single cause. Write it down with the test that would confirm it.

7. 7

   Test it, one change at a time

   The test that confirms or eliminates the hypothesis, with the result recorded. If confirmed, fix. If not, the next hypothesis. Never two changes without a check between.

8. 8

   Verify

   The symptom gone under the conditions that produced it, not just at the moment of the fix. Watch the trend through a cycle.

9. 9

   Restore

   Every temporary change reversed: forces, bypasses, shelved alarms, manual modes, test jumpers, disconnected shields, changed setpoints. Walk the list.

10. 10

    Record

    Symptom, evidence, cause, fix, parts, time, and the person, in the work order and the plant notes. Update the drawing or the schedule if it was wrong.

Questions that shorten the search

Escalation

The checklist also says when to stop and call: when the process is unsafe and manual operation is not holding it, when the evidence points at a compromise, when a change would affect equipment or people beyond the technician authority, or when the search has gone two hours without a hypothesis that survived a test. The call list is on the same sheet: the on-call engineer, the integrator support line, the vendor, the supervisor. Calling early is cheaper than a night of guessing, and the evidence gathered in step four is what the person on the phone will ask for.

Where to go next

The specific troubleshooting pages on this site take the search from the checklist to the symptom: a transmitter reading wrong, a level that jumps, a device that times out, a drive that trips, a value frozen on the screen, an alarm that did not sound. Each follows the same shape, with the causes ranked and the checks for each, and the general checklist is the frame they fit in.