Generator testing records that do not demonstrate compliance
At a Rated-2 site, the facility engineer states that generators are load-tested monthly. The CMMS export shows 7 of 12 monthly tests in the audit period. The paper log has 12 signatures but no load values, duration or result. A vendor report says 'annual load bank test completed' but the test date is two weeks before the audit period. During the walkdown, the generator day tank gauge reads 60%, while the site procedure requires at least 80%. The auditor also sees that the fuel log is updated only when the tanker delivers, not daily. The audit criteria are TIA-942, the site's maintenance procedure and its fuel-management procedure. You must decide whether the evidence supports a nonconformity and how to report it.
1. Evaluate the reliability and sufficiency of the evidence. What additional evidence would you seek before concluding?
Read the worked answer
The evidence varies in reliability. The CMMS export is system-generated and therefore generally more reliable than memory or a paper signature, but it may be incomplete if technicians do not record work. Seven of twelve entries is objective evidence that the system records only seven tests; it is not, by itself, proof that five tests did not occur. The paper log has twelve signatures but no load values, duration or result. A signature without test data is weak evidence: it shows someone signed, not that a load test was performed to the required parameters. The vendor report is an annual load-bank test dated before the audit period, so it does not cover the period in question.
The day-tank observation at 60% against an 80% procedure minimum is direct, objective evidence at a point in time. It may be transient, so I would check the gauge again, review alarm history and fuel-delivery records, and determine whether the procedure applies continuously or only before a test. The fuel log updated only on delivery suggests the daily check is not being performed, but I would confirm the procedure's frequency and look for other daily records.
Additional evidence: generator run-hour meters at the start and end of each month, fuel consumption records, alarm and event logs, work orders and vendor invoices, maintenance crew interviews, and the next scheduled test observation. I would also check whether the CMMS and paper log are supposed to be reconciled. Only after triangulation would I conclude that testing did not occur.
2. Classify the findings and write report wording that is defensible. What should the auditor avoid saying?
Read the worked answer
I would separate the findings. Finding 1 - maintenance records: the criteria are the site maintenance procedure requiring monthly generator load tests with recorded load, duration and result. The evidence is the CMMS showing 7 of 12 tests, the paper log with signatures but no test data, and the vendor report falling outside the period. The finding should state that the records do not demonstrate that monthly load tests were performed and documented for five months. The risk is that generator reliability is unverified and a real failure could go undetected. Classification: nonconformity, likely major if the procedure is mandatory and the gap is systemic.
Finding 2 - fuel level: the criterion is the fuel-management procedure requiring at least 80% in the day tank. The evidence is the observed 60% reading. The finding is a nonconformity if the requirement is continuous; if it applies only before tests, it may be an observation. The risk is reduced runtime and fuel exhaustion during an extended outage. I would also record the fuel-log practice as a related observation or nonconformity, depending on the procedure's daily-check requirement.
The auditor should avoid saying 'the generators were not tested' or 'the site is noncompliant with TIA-942' without qualification. Better wording is 'the records do not provide evidence that the required tests were performed'. The report should distinguish between absence of evidence and evidence of absence, cite the specific records reviewed, and include the management response and corrective-action requirement.
3. What audit techniques would you use to corroborate the maintenance and fuel evidence, and how would you sample?
Read the worked answer
I would use a mix of document review, data analytics, observation, interview and re-performance where possible. For document review, compare the CMMS export, paper log, vendor reports and work orders for the same dates. For data analytics, extract generator run hours and fuel consumption by month and look for months with no increase or no test activity. A month with no run-hour movement is stronger corroboration than a missing CMMS entry alone.
Observation should include a walkdown of the generator room, day tank, fuel log and gauge, and, if timing allows, witnessing the next scheduled load test. I would check the test parameters against the procedure: load percentage, duration, acceptance criteria and recording requirements. Interviews should be separate with the facility engineer, the maintenance technician and the vendor, asking how tests are recorded and reconciled. I would avoid leading questions and ask for examples.
Sampling should cover all 12 months, not a convenience sample, because the issue is frequency. I would select the five months missing from the CMMS plus two months present in both sources, then trace each to run-hour logs, fuel records and work orders. For fuel, I would sample daily or weekly checks across the period and compare with delivery records. The conclusion should rest on the convergence of independent sources, not on a single missing signature.