The Fourteen-Hour Gap at Outfall 2
The continuous fluoride analyser at Outfall 2 logs a reading every 15 minutes and feeds the discharge permit report automatically. On the night of 3 March a scheduled server migration overran and the logger wrote nothing between 01:15 and 15:30 - fourteen hours and fifteen minutes. The plant ran its backup neutralisation line for most of that window because the primary dosing pump was out for repair. Shift grab samples at 06:00 and 12:00 returned 2.9 mg/L and 3.4 mg/L against a permit limit of 4.0 mg/L, but both were run on a bench meter calibrated quarterly rather than daily. The regulator's annual audit is in six days. Your supervisor, who is also the permit signatory, tells you to smooth the gap with the grab results so the report looks continuous: nobody cares about a server fault, and an incomplete submission triggers a mandatory site visit.
1. Should the permit report be submitted with the gap filled from the grab samples? Show how you reach the decision.
Read the worked answer
Two separate questions are hidden in one instruction: is the estimate defensible, and is presenting it as measurement honest?
Defensibility first. Two grab samples on a quarterly-calibrated bench meter cannot reconstruct fifteen hours of continuous data. The calibration interval alone puts an unknown bias on both values, and the samples are instantaneous points taken during a non-routine bypass configuration - they miss the dosing pump restart entirely. Any interpolated curve would be a shape no instrument recorded.
Honesty second. A report that shows continuous values implies continuous measurement. Filling the gap without flagging it turns a technical failure into a false record, whatever the intent.
The decision: submit on time, show the gap as a gap, attach the two grab results as supporting evidence with method and calibration status stated, and add a note on cause and corrective action. A flagged report can be defended in the audit room; a backfilled one cannot be defended once the logger system log is produced.
2. The supervisor repeats the instruction in writing and copies the operations manager. What do you do next?
Read the worked answer
A written instruction changes the problem from technical to records-integrity, and repetition does not make it acceptable.
Reply in writing, once, calmly and factually: the gap is fifteen hours fifteen minutes; the supporting samples are two points on a quarterly-calibrated meter; presenting them as continuous would misrepresent the record to a regulator. Do not lecture and do not speculate about motive - the email may be read by people who were not in the conversation.
Then offer the workable alternative rather than only a refusal: submit flagged, check whether the permit has a data-loss or estimated-data provision, and attach a short incident note. Most permits do have such a provision; using it is legitimate where backfilling is not.
If the instruction stands after that, escalate to whoever owns environmental compliance above the signatory, or use the internal reporting channel. What you must not do is go quiet and let the report be filed unaltered, and you must not edit the logger export or the bench records to make the story consistent.
3. At the audit, the regulator asks what happened on 3 March. What do you say, and what should the file contain?
Read the worked answer
Answer directly: the logger did not record between 01:15 and 15:30 because a server migration overran, and it happened during a period of non-routine operation. Volunteer the bypass and the pump outage before the auditor finds them. A disclosed deviation is a finding; a concealed one is a credibility problem that contaminates everything else in the audit.
The file should hold: the raw analyser export and the logger system log showing the outage window; the migration change record and its approval; the two grab results with the bench meter's calibration certificate and date; the pump repair work order showing the plant was in bypass; the corrective actions and their status - local buffering on the logger, a migration blackout window, daily bench calibration during bypass operation; and the correspondence showing that the decision was to submit flagged rather than backfill. That last item is the evidence that the gap was a technical failure, not a reporting choice.