Why electrical witness tests fail, and how to pass the first time

By Eng. Mannan Maqsood, Electrical EngineerPublished 6 November 20264 min read

A witness test is meant to confirm results the contractor has already obtained. It is not the first time the installation is tested. When a witness session is stopped after twenty minutes, it is almost never because of one bad reading. It is because the preparation was not done. Every failed witness test costs a re-inspection request, another visit, and often a delay to energisation. This article lists the reasons tests fail most often and how to avoid them, whether you are the contractor preparing the test or the engineer witnessing it.

1. Pre-commissioning was skipped

The contractor should test every circuit internally and record the results before inviting anyone. The witness engineer then repeats a sample, not the full scope. If the pre-commissioning sheets are blank or were filled in the night before, the first circuit checked usually shows it. The test that was supposed to take an hour turns into a full test run with an audience.

Fix: attach the completed and signed pre-commissioning records to the inspection request. If the records are not attached, the witness engineer is entitled to reject the request before travelling to site.

2. Instruments without valid calibration

An insulation resistance tester, loop impedance tester or RCD tester must have a current calibration certificate traceable to a recognised laboratory. The serial number on the certificate must match the instrument on site. Expired certificates, certificates for a different serial number and "borrowed" instruments with no paperwork are among the most common reasons a session is stopped before it starts.

Fix: list the instruments and their certificate expiry dates on the inspection request. Prove each tester on a known source before and after the test. The Meter Lab explains what each instrument measures and the checks before use.

3. No approved method statement or ITP

The test sequence, acceptance values and hold points should come from an approved inspection and test plan. Without one, the contractor and the witness engineer argue on site about what "pass" means. Typical disputes include the insulation resistance test voltage, whether surge protection devices and electronic equipment should be disconnected, and which RCD tests are required.

Fix: get the ITP approved before testing starts, and copy the acceptance criteria onto the test sheets so the pass value is printed beside the measured value.

4. The installation is not ready

A witness test assumes the installation is complete for the part being tested. Common signs that it is not:

• circuit labels and DB schedules missing or still handwritten
• cable glands not earthed, cover plates missing, open knockouts in distribution boards
• termination torque not marked, so tightness cannot be confirmed
• temporary supplies still connected, or the permanent supply not yet stable
• other trades still working in the area, so the test area cannot be made safe

Fix: walk the area the day before with a simple readiness checklist. If it is not ready, move the request. A rescheduled test is cheaper than a failed one.

5. The wrong test, or the right test done the wrong way

Some typical method errors:

• Insulation resistance measured with loads, drivers or SPDs still connected, giving low readings that are blamed on the cable. Or measured with them connected and the result recorded as a pass, hiding a real fault in the cable.
• Continuity of protective conductors not measured end to end, so a missing earth link at an intermediate box goes unnoticed.
• Earth fault loop impedance recorded without comparing it with the maximum permitted value for the actual protective device and disconnection time. A reading of 1.2 Ω means nothing until you know the device it has to trip.
• RCD tests carried out at one test current only, when the ITP requires tripping times at the rated residual current and at five times that value.
• Phase rotation not checked before motors are run, so pumps and fans run backwards on first start.

6. Results that look too good

Experienced witness engineers notice when every circuit on a sheet shows exactly the same value, or when insulation resistance on a long feeder is recorded as ">999 MΩ" on a tester whose range stops below that. Copied results destroy trust, and once trust is lost the engineer will ask to retest everything.

Fix: record the real reading, the instrument used and the test voltage. Real installations give varied results.

7. Interfaces not tested

Many failures happen between systems rather than inside them: the fire alarm does not shut down the AHU, the generator start signal is not wired to the ATS, the BMS shows the wrong status for a pump. Electrical testing passes, but the integrated test fails.

Fix: test each interface point to point with the other trade present before the integrated witness test. Use the cause and effect matrix as the test sheet.

8. Paperwork on the day

Bring the approved drawings, the latest SLD, the approved ITP, the pre-commissioning sheets, calibration certificates and blank witness sheets. Without approved drawings, the witness engineer cannot confirm that the installed breaker or cable matches the design.

A ten point readiness check

1. Inspection request raised with the agreed notice period
2. Approved ITP and method statement attached
3. Pre-commissioning results complete and signed
4. Calibration certificates valid, serial numbers match
5. Area complete, safe and clear of other trades
6. Labels, DB schedules and as-built mark-ups in place
7. Permanent and stable supply, or the agreed test supply
8. Equipment that must be disconnected for testing identified
9. Interfaces pre-tested with the other trades
10. Competent tester present who did the pre-commissioning

Track it, do not remember it

The Electrical & ELV T&C Tracker records each test from first energisation to handover, with witness names and results. The FAT, SAT & Site Testing Lab shows the test methods and acceptance values step by step. Acceptance values must always come from the approved ITP, the governing regulations and the manufacturer's data for your project.

Disclaimer. This article is general technical information for education only. It is not professional engineering, design, legal or safety advice for any specific project, and it does not replace the judgement of a qualified, registered engineer who has reviewed the actual installation. Worked examples and figures are illustrative and do not describe any identified project, client or site. Codes, authority requirements and product data change, so confirm every value against the edition that applies to your project, the authority having jurisdiction and the manufacturer's data. Any testing or work on live systems must be carried out only by competent, authorised people under the approved procedure. The views are the author's own and do not represent any employer, client or authority. To the extent permitted by law, the author, the editor and MEPVirtual accept no liability for any loss, damage or injury arising from the use of, or reliance on, this content. See the full disclaimer or report a correction.