Service · Factory execution · FAT
FAT witnessing for EPC procurement packages in China
On an EPC package, the expensive gap is rarely between good equipment and bad equipment. It is between the test the contract bought and the test the shop actually ran — a routine verification relabelled as a performance FAT, a prototype tested instead of the shipping serial number, or logic proven by forced signals without anyone writing that down. Sinospect witnesses and challenges FAT on bought-out packages at Chinese factories: readiness gated before travel, execution logged independently, the FAT mode classified honestly, punch items coded by consequence, and a release recommendation the project can defend in a data-book review.
- Package classes
- Rotating equipment, switchgear/MCC/VFD, transformers, motors, control panels and instrumented skids
- Acceptance basis
- Purchase order, final data sheets, approved drawings, ITP and the FAT procedure — read against the governing standard family and edition
- Where it runs
- OEM, packager and subcontractor sites across China, including the ones not named in the contract
- Engagement output
- Clause-by-clause compliance matrix, raw data, coded punch list and shipping-release recommendation
Quick answers
What does FAT witnessing for an EPC procurement package cover?
Pre-FAT readiness review against the PO, data sheets, approved drawings and ITP; witnessed execution of the declared tests with independent logging of variables; classification of the FAT mode actually executed; a coded punch list; and a written release recommendation tied to the serial number that is shipping. The scope is witness-and-challenge, not attendance.
| What the contract says | What it can mean in the shop | What the witness records |
|---|---|---|
| Witness FAT | A review of the routine test report already produced last week | Whether the test was performed in front of the witness or replayed on paper |
| Full performance test | A partial-load run, or logic exercised by forced signals | The executed mode: full-load, limited-load, routine-only or simulation-based |
| IEC compliant | Tested to an equivalent national route, or outside the laboratory's accredited scope | Standard family, edition, test route, and whether the scope covers the ordered rating |
| Package complete | Subcontractor-built auxiliaries excluded from the witnessed event | Which scope items were physically present and which were tested elsewhere |
| Passed FAT | Passed with open punch items and a deferred retest | Punch items coded by consequence, with the retest status of each |
| Released for shipment | Equipment re-opened, reworked or re-packed after the witnessed event | Post-FAT interventions and whether they invalidate a witnessed result |
Which equipment classes does the witness scope cover, and against what?
Rotating equipment, electrical distribution and drives, transformers, motors, and control and instrumented packages — each with its own governing standard family, its own minimum test content and its own list of tests that are legitimately simulated. Naming the acceptance basis per class is what stops a package FAT collapsing into one generic pass.
| Package type | Governing basis commonly cited | The variable most often left vague |
|---|---|---|
| Pumps and pump skids | ISO 9906 hydraulic acceptance grades; ISO 10816 for mechanical vibration | Which acceptance grade applies — Grade 3B allows ±9% on flow and ±7% on head, Grade 1B far less |
| Compressors and auxiliaries | API mechanical run test and rotor-response criteria | Run duration and load condition, and whether coupling match marks were re-checked after the run |
| MV/LV switchgear, MCCs, VFDs | IEC 62271 and IEC 61439 routine verifications | Whether the drive was tested integrated at rated load or as isolated subassemblies |
| Transformers and dry-type packages | Routine tests plus partial-discharge measurement where specified | That resistance and impedance measurements say nothing about solid-insulation condition |
| Motors and package drives | Temperature-rise class and a named vibration measurement method | "No abnormal vibration" in place of a stated method, axis and load condition |
| PLC, HMI and instrumented packages | Approved cause-and-effect matrix and I/O schedule | The boundary between hardwired signals and forced ones, and the frozen software revision |
01 · Scope
When a package FAT needs challenge, not attendance
EPC procurement carries a structural asymmetry: the supplier writes the FAT procedure, owns the instruments, controls the shop schedule and knows what the buyer's representative is likely to check. Attendance alone does not correct that — it produces a signature on the supplier's own agenda. Challenge does: reconciling documents before energization, insisting the shipping serial number is the unit on test, logging variables independently, and red-tagging blocked tests, substituted conditions and undeclared simulations as they happen. The value of that discipline scales with how far the package sits from a catalogue item and how badly the commissioning schedule would suffer from finding out later.
It becomes useful when
- The package is engineered to project data sheets rather than bought from a catalogue, so no prior unit proves anything about this one.
- The FAT procedure and the ITP disagree, or the ITP hold points were never converted into a procedure at all.
- Test media, load banks or downstream systems do not exist in the supplier's shop and nobody has written down what that means.
- Part of the package is built or tested by a subcontractor whose site is not the address on the purchase order.
- A milestone payment or a shipping release is tied to the FAT result.
- The receiving team will commission equipment it did not specify, months later and on another continent.
02 · Verification
What the witness controls across the package
Control point
Pre-FAT readiness gate
Approved FAT procedure issued, drawings and data sheets frozen, serial numbers assigned, utilities and test media confirmed, calibration pack issued, software and firmware revisions listed, and earlier punch items either closed or explicitly carried in. Attending an unready shop produces cost and weak evidence, so readiness is confirmed before travel.
Control point
Document reconciliation and physical walkdown
Purchase order, line-item deviations, final data sheets, general arrangement, P&IDs and wiring diagrams reconciled against the assembly before energization; nameplates, tag numbers and terminations walked down against the document set rather than assumed from the drawing register.
Control point
Witnessed execution and independent logging
Declared routine, performance and special tests witnessed with variables, times and instrument references logged independently of the supplier's own sheets; instrument photos, nameplate photos and alarm or trip events captured as they occur.
Control point
Test-boundary and simulation disclosure
The executed FAT mode classified as full-load, limited-load, routine-only or simulation-based; the hardwired-versus-forced signal boundary recorded; substituted conditions, borrowed instruments and temporary arrangements red-tagged when they appear rather than discovered in the report.
Control point
Punch coding and release decision
Findings coded by consequence — shipment blocked, ship only against a written concession, or documentary closure before data-book handover — with the release recommendation tied to the shipping serial number, the tests skipped and the residual risk transferred to site.
03 · Evidence
Evidence captured across the witnessed package FAT
- Baseline and readiness record
- Document baseline frozen: PO and deviations, final data sheets, approved drawings, P&IDs, ITP, FAT procedure, document register and nameplate schedule.
- Serial numbers and the production status of every item on test, plus the exact site — including whether any stage runs at a subcontractor or rented facility.
- Instrument serial numbers, calibration validity and traceability, and the standard family, edition and accredited scope behind any compliance claim.
- Witnessed execution record
- Independently logged variables and times for each declared test, against the acceptance criterion actually stated in the procedure.
- Photo and video capture of instruments, nameplates, alarm and trip events, and any condition that departs from the written procedure.
- Software and firmware revisions recorded as frozen on the day, with the I/O simulation coverage identified test by test.
- Disposition and handover record
- Clause-by-clause compliance matrix separating tests performed, tests deferred and tests not run.
- Coded punch register with owner, evidence required for closure and retest status.
- Release recommendation, list of skipped tests with reasons, and the residual risks explicitly transferred to site acceptance and commissioning.
04 · Deliverables
Deliverables issued
Deliverable
For · Package engineering and procurementPre-FAT readiness report
What is ready, what is not, and whether the visit should proceed — issued before travel is committed.
Deliverable
For · Discipline engineers and QA/QCClause-by-clause FAT compliance matrix
Each procedure clause against what was actually executed, with the acceptance criterion and the observed result side by side.
Deliverable
For · Engineering and the data bookRaw data and calibration pack
Recorded instrument readings, calibration certificates used on the day and the frozen software-revision record — not a summary sheet.
Deliverable
For · Procurement and the supplierCoded punch and NCR register
Findings classified by consequence, with the evidence each one needs before it can close.
Deliverable
For · Procurement, project management and financeShipping-release recommendation
Accept, conditional accept against a locked punch list, reject and retest, or witness aborted for non-readiness — with the FAT mode and residual risk stated.
05 · Risks reduced
Risks this control closes
Risk
A routine verification is relabelled as the performance FAT the contract paid for.
How the witness closes it
The executed mode is classified explicitly and the compliance matrix separates tests performed from tests referenced, so the difference is on the record rather than in the wording.
Risk
The FAT runs on a prototype or a representative unit, not the serial number that ships.
How the witness closes it
Serial numbers and production status are confirmed before energization and every released item is tied to the unit actually tested.
Risk
Logic passes on forced signals and the simulation boundary never reaches the report.
How the witness closes it
The hardwired-versus-simulated boundary is recorded test by test, with the residual behaviour listed as a site-acceptance item.
Risk
Subcontracted skids and auxiliaries sit outside the witnessed event but inside the certificate.
How the witness closes it
The physical location of every scope item is established in the pre-FAT review, and uncovered items are named as uncovered.
Risk
Calibration records, instrument identities and software revisions are reconstructed after the fact.
How the witness closes it
They are captured on the day and frozen into the evidence pack, where a later data-book review can still check them.
Risk
"Passed FAT" masks open punch items and a retest that never happened.
How the witness closes it
Punch items are coded by consequence and the release recommendation states what remains open and what closing it requires.
06 · Background resources
Related
- Resource · China-side procurement support for EPC teamsThe full package lifecycle — supplier qualification, technical review, manufacturing and handover — that this FAT scope sits inside.
- Service · Technical review & procurementWhere package specifications and offers are challenged before award, so the FAT has criteria worth witnessing.
- Service · Factory acceptance test in ChinaThe parent FAT service: scope, control points and deliverables for any equipment class.
- Service · Switchgear & panel FAT witnessingThe deeper scope when the package is LV/MV switchgear, MCCs or control cabinets.
- Resource · Factory acceptance testing guideThe full FAT guide: sequence, evidence and the release decision.
- Resource · Factory acceptance test checklistThe twelve-section FAT checklist, downloadable as a printable PDF.
07 · Questions
Frequently asked questions
How is this different from expediting or a pre-shipment inspection?
Expediting tracks whether the package will be ready; pre-shipment inspection verifies that what is in the crate matches the order. FAT witnessing for an EPC package does something narrower and harder: it challenges whether the test that is being run is the test the contract bought, on the unit that is actually shipping, with results that survive a data-book review. All three have a place in a package plan — they answer different questions and none substitutes for another.
What if the factory cannot run a full-load test?
That is common and often legitimate — process media, load banks, downstream systems or ambient conditions may simply not exist in a supplier's shop. The failure is not the limitation; it is leaving it undeclared. The witness record classifies the executed mode as full-load, limited-load, routine-only or simulation-based, states which signals were hardwired and which were forced, and lists what consequently transfers to site acceptance as residual risk.
Can one visit cover a package built across several suppliers?
Sometimes, and where it cannot, that itself is the finding. Packages routinely stop at the OEM's cabinet while subcontracted skids, auxiliaries or panels are built and tested elsewhere. The pre-FAT review establishes where each scope item is physically made and tested, and the witness plan either extends to those sites or records explicitly that they were outside the witnessed event — rather than letting a single certificate imply coverage it does not have.
What does an "IEC compliant" claim actually need to be checked against?
Three things: the standard family and edition actually used, whether the test route was the IEC method or an equivalent national route, and whether the testing facility's accreditation covers the ordered rating. Laboratory accreditation under ISO/IEC 17020 or ISO/IEC 17025 is always scoped — by product type, voltage class, current rating and measurement range — so a certificate can be genuine and still not cover the unit you bought. The witness checks the scope, not just the logo.
What is the output the project team can actually use?
A clause-by-clause compliance matrix against the FAT procedure, raw recorded data rather than summary sheets, the calibration and software-revision records frozen as they stood on the day, a coded punch list, and a release recommendation tied to the shipping serial number. It is written to be usable in a data-book review, a payment-milestone decision and a supplier escalation — not just to confirm that somebody attended.
08 · Getting started
What to send — start with what you have
The FAT procedure and the ITP together show most of what is wrong before anyone travels. If they do not exist yet, the purchase order and data sheets are enough to start.
Purchase order and line-item deviations
What was actually bought, including anything conceded during negotiation.
Final data sheets and approved drawings
The frozen technical baseline the assembly is verified against.
ITP and FAT procedure
The hold points and the test content — the two documents that most often disagree.
Cause-and-effect matrix and software revision list
For control and instrumented packages, so simulation coverage can be judged rather than assumed.
Factory city, planned FAT date and utility availability
Where the shop is, when the package is ready, and whether full-load test media exist on site.
Partial information is fine. Sinospect states what is missing and what it changes about the achievable witness scope before any commitment.
Put a challenge function on your next package FAT
Send the purchase order, the ITP or FAT procedure if they exist, and the planned test window. Sinospect replies within one business day with the proposed witness scope, the readiness gaps already visible in the documents, a quotation and the earliest feasible attendance.