Service · Factory execution · FAT
Switchgear and GIS partial-discharge FAT witnessing in China
A partial-discharge certificate for switchgear can be technically true and commercially worthless at the same time — because the number belongs to an object nobody named. Sinospect witnesses PD testing on GIS, metal-clad and ring-main assemblies at the Chinese factory: the vendor's procedure red-lined before travel, calibration and noise floor verified on the bay, the test object recorded by level, phase-resolved patterns interpreted against the defect classes that actually occur in switchgear builds, and a witnessed retest after any rework that disturbs the geometry.
- Equipment class
- GIS and HGIS, metal-clad and metal-enclosed MV switchgear, ring-main units, switchgear–cable interfaces
- Acceptance method
- IEC 60270 conventional measurement; UHF and acoustic used for noise discrimination and localization only
- Specification basis
- Your project specification, read against the IEC 62271 and Chinese GB/T routes switchgear is commonly built and routine-tested to
- Engagement output
- Witness record with test-object level, pattern captures, defect register and retest matrix
Quick answers
What does switchgear partial-discharge FAT witnessing in China cover?
An independent engineer red-lines the vendor's PD procedure before travel, then witnesses the test on the factory floor: calibration of the complete measuring circuit, the background-noise floor, the voltage sequence, and the measured magnitude read against the limit for the object actually under test. Phase-resolved patterns are interpreted against known defect classes, and any rework triggers a witnessed retest.
| Test object | What a clean result proves | What it does not prove |
|---|---|---|
| Single insulator or component, before assembly | The component left its own production line without a detectable defect | Nothing about the assembled bay, the barriers around it, or the joints made later |
| Partition or sub-assembly | The sub-assembly's own insulation and field grading | The complete current path, adjacent compartments, or the cable interface |
| Shipping module or complete bay | The assembly as tested in the factory circuit | The electromagnetic and mechanical environment of the final installed lineup |
| Installed lineup at site | The as-built condition after cabling and termination | Nothing at FAT — this level is a site activity, not a factory one |
Published limits are stated per object and are not interchangeable between levels. A certificate that does not name the object cannot be read against any of them.
Which defects does PD reveal in switchgear that a routine FAT does not?
The defect classes that survive a visual inspection and a withstand test: loose metallic particles created during assembly, metal parts that have lost their earth connection, sharp protrusions on live or enclosure surfaces, voids and delamination inside solid insulation, and workmanship faults at the cable interface. Each has a characteristic phase-resolved signature, which is why interpretation matters more than a single magnitude.
| Defect class | Typical origin in a switchgear build | Why magnitude alone misleads |
|---|---|---|
| Moving or hopping metallic particles | Swarf and fragments generated by drilling, threading and assembly work inside the enclosure | Activity is intermittent — a short measurement window can miss it entirely |
| Floating-potential discharge | A metallic part that has lost galvanic contact and now sparks across a very small gap | Tends to show high amplitude and high pulse count, which can be misread as a measurement artefact |
| Protrusions and sharp points | Burrs, unrounded conductor edges and hardware left inside the high-field region | May sit below the limit at test voltage while remaining a growth site in service |
| Voids, cracks and delamination in solid insulation | Supports, drive shafts, bushings and cast components | Small isolated voids can take longer to initiate than a routine test window allows |
| Surface contamination and cable-interface workmanship | Semi-conductive residue, tape or foreign material in the stress region; stress-cone position; screen cutback | Different causes produce superficially similar patterns and need envelope discrimination, not a number |
01 · Scope
When a switchgear PD witness earns its cost
Partial discharge is where switchgear quality either is or is not real, and it is also where a factory has the most room to be technically accurate and practically unhelpful. Three habits do the damage: testing a component and reporting it as if the assembly passed, letting a screening instrument stand in for a calibrated measurement, and closing a compartment after rework without repeating the test. None of these is fraud; all three are normal factory economics. A witness who understands the measurement — not just someone present while it runs — is what converts the PD line on your certificate into evidence you can defend at handover, to a lender, or in a dispute.
It becomes useful when
- The order is GIS, HGIS, metal-clad or metal-enclosed MV switchgear where the specification sets a PD acceptance limit.
- The vendor's FAT procedure does not state which object each PD limit applies to — insulator, partition, bay or shipping module.
- The assembly includes cable terminations or interfaces, where workmanship defects are common and are completed after the electrical tests.
- The factory proposes handheld TEV or ultrasonic screening in place of a calibrated IEC 60270 measurement.
- A previous unit from the same builder produced marginal, unstable or unexplained readings.
- Rework has already happened once and nobody has decided in writing what has to be re-tested.
02 · Verification
What the witness controls on the switchgear bay
Control point
Pre-FAT procedure red-line
The vendor's FAT procedure, approved drawings, interlock logic, CT/VT list, cable-interface design and declared PD limits reviewed before travel; the limit is tied to a named object level, and gaps between what the contract promises and what the procedure actually executes are listed before anyone books a flight.
Control point
Test-object identity and level
Serial numbers and production routing confirmed against the units being bought; whether the item on the bay is a production unit, a pre-series build or a type-test specimen established and recorded, along with the object level the measurement will belong to.
Control point
Calibration and noise floor
Calibration of the complete measuring system in the complete test circuit witnessed in the sense of IEC 60270, repeated after any material change to the setup, with the background-noise level recorded before energization and challenged when it sits close enough to the limit to make a pass meaningless.
Control point
Measurement and pattern interpretation
Voltage steps and durations witnessed against the agreed sequence; inception and extinction behaviour observed; phase-resolved captures interpreted against known defect classes, with repeated captures rather than a single screenshot where the signature is unstable.
Control point
Localization, internal inspection and retest
Corroborative UHF or acoustic localization used to narrow a real indication to a compartment; internal photographic evidence captured before compartments are closed; any rework that disturbs geometry followed by a witnessed retest recorded against the original result.
03 · Evidence
Evidence captured during the witnessed PD test
- Test basis and readiness
- Red-lined FAT procedure with the object level, limit, voltage sequence and measurement conditions agreed in writing before the visit.
- Serial numbers, drawing revisions and the production status of the unit on the bay recorded.
- Instrument references, calibration records and the ambient conditions — temperature, humidity, external interference — logged at the setup.
- Witnessed measurement
- Circuit calibration and background-noise floor recorded before energization, and again after any material setup change.
- Voltage steps, durations and measured magnitudes recorded as observed, including unstable readings and repeated runs.
- Phase-resolved captures and raw traces retained with the setup notes needed to interpret them later.
- Defect handling and disposition
- Each indication logged with its location, the defect class it fits, and whether it was resolved as real discharge or as external interference.
- Internal photographic evidence of compartments before closure, where access allows.
- Retest matrix showing original result, intervention and post-rework result — with a separate pass or fail statement for each object level, never one blended sentence.
04 · Deliverables
Deliverables issued
Deliverable
For · Engineering and procurementPre-FAT red-line and witness plan
Comments on the vendor's PD procedure and the agreed object level, limit and sequence — so the argument about criteria happens before the test, not during it.
Deliverable
For · HV engineering and external reviewersWitnessed PD record
Calibration, noise floor, voltage steps and measured magnitudes tied to serial numbers, with the object level stated for every result.
Deliverable
For · EngineeringPattern captures and raw evidence index
Phase-resolved captures, traces and setup photographs, indexed so a third party can re-read them without being in the room.
Deliverable
For · Procurement and the supplierDefect register and retest matrix
Each indication with its likely defect class and location, the corrective action taken, and the witnessed result after rework.
Deliverable
For · Procurement and project managementContract-gap register
Where the executed FAT differs from what the contract promised — scope not run, objects not tested, evidence not produced — separated from the technical result.
05 · Risks reduced
Risks this control closes
Risk
A clean component result is presented as if the assembled bay had passed.
How the witness closes it
The object level is recorded for every measurement and the report carries a separate pass or fail statement per level, so a blended acceptance sentence cannot be written.
Risk
A screening reading from a handheld instrument is offered as the acceptance value.
How the witness closes it
Acceptance is held to the calibrated IEC 60270 measurement; screening and UHF results are recorded as what they are — localization and noise discrimination, not pC-calibrated magnitudes.
Risk
Factory-floor interference is recorded as a defect, or a real defect dismissed as noise.
How the witness closes it
Noise floor, setup control, repeated captures and phase-resolved interpretation are used to separate the two, with the reasoning written into the record rather than asserted.
Risk
The measuring circuit is calibrated once, then the setup changes and the numbers drift.
How the witness closes it
Calibration is witnessed in the complete circuit and renewed after material setup changes, with each check recorded against the measurements it covers.
Risk
A compartment is opened for rework and closed again without repeating the test.
How the witness closes it
Retest is treated as mandatory wherever geometry, barriers, bushings, terminations or shields have been disturbed, and the retest matrix keeps both results visible.
Risk
The cable interface — the part most often finished after the electrical tests — is never covered at all.
How the witness closes it
The witness scope states explicitly which interfaces were in the tested object and which remain for site, so nobody assumes the other test proved what neither tested.
06 · Background resources
Related
- Service · PD test witnessing at FATThe general PD witness scope across transformers, instrument transformers and HV cables, focused on measurement credibility.
- Service · Switchgear & panel FAT witnessingThe full switchgear witness scope: construction, wiring, dielectric tests, interlocks and protection function.
- Service · Factory acceptance test in ChinaThe parent FAT service — PD witnessing usually runs inside this wider witnessed scope.
- Resource · Partial discharge testing at FATThe specifier's guide: when PD belongs in the FAT and what a credible result records.
- Resource · Factory acceptance testing guideThe full FAT guide: sequence, evidence and the release decision.
- Sector · EnergyHow Sinospect supports energy-sector buyers across supply, inspection and delivery.
07 · Questions
Frequently asked questions
What is the difference between this and general PD test witnessing?
General PD witnessing verifies that the measurement itself is credible — calibration, noise, voltage sequence, the right limit — for any equipment class. This scope adds the switchgear-specific layer: which object in the assembly was actually tested, whether the phase-resolved pattern matches a defect class known to occur in GIS and metal-clad designs, and what has to be re-tested once a compartment has been opened. If your order is transformers or cables, the general PD scope is the right page; if it is switchgear, this one adds the part that decides whether a clean number means anything.
Can a handheld TEV or ultrasonic check replace an IEC 60270 measurement?
No, and treating it as a substitute is the most common technical error on this subject. Handheld transient-earth-voltage and ultrasonic instruments are screening and trending tools: the TEV amplitude varies with the enclosure's grounding arrangement and the ultrasonic response varies with the path between the source and the sensor, so neither yields a value that can be compared with an acceptance limit expressed in picocoulombs. Conventional IEC 60270 measurement is the method calibrated in apparent charge, and it is the method acceptance should rest on.
Is UHF measurement useful on GIS if it is not calibrated in picocoulombs?
Yes — for the job it is good at. UHF sensing is sensitive and comparatively immune to the electrical noise of a live factory, which makes it valuable for deciding whether an indication is real discharge or interference, and for narrowing down which compartment it comes from. What it does not provide is a pC-calibrated magnitude, so it supports the IEC 60270 result rather than replacing it. The witness record states which method produced the acceptance value and which produced the localization.
The factory says the insulators already passed a routine PD test. Is that enough?
It depends entirely on what the contract bought. A routine PD test on a single insulator before assembly proves the component; it does not prove the partition, the assembled bay, or the shipping module — and it certainly does not prove the cable compartment, which is usually completed after the electrical tests. The witness records the test-object level explicitly and refuses a blended acceptance sentence that reads as if one clean result covered all of them.
What happens after a defect is found and the compartment is opened?
Opening a compartment changes the geometry that PD behaviour depends on, so the previous result no longer applies to the reworked object. The witness scope treats retest after intervention as mandatory wherever barriers, bushings, terminations, shields or conductor geometry have been disturbed, and records original result, intervention and post-rework result side by side rather than replacing the first number with the second.
08 · Getting started
What to send — start with what you have
The vendor's FAT procedure is the most useful single document, because it is where the object-level ambiguity usually lives. A specification or even the quotation is enough to start.
Vendor FAT procedure or test plan
The document that will govern the day — the one that gets red-lined before travel.
Specification and declared PD limits
The limit your contract sets and the object it applies to, or the standard it defers to.
Single-line diagram and general arrangement
What the assembly is meant to be, so the object under test can be identified against it.
Cable-interface design, where applicable
Terminations and accessories, so the scope states what the factory test can and cannot reach.
Manufacturer, serial quantities and expected test window
Where the bay is and when the units reach it, so the visit lands before dispatch pressure builds.
Partial information is fine. Sinospect confirms the governing limit, the object level and what is missing before any commitment.
Put an independent witness on your switchgear PD test
Send the FAT procedure or the specification, the switchgear type and the expected test window. Sinospect replies within one business day with the proposed witness scope, the red-line points already visible in the procedure, a quotation and the earliest feasible attendance.