Site acceptance test SAT

What it is (precise)

IEC 62381 defines the site acceptance test at 3.1.19 as an activity, including inspection and testing, conducted at the site of the installation, to demonstrate that the installation of the automation system, any subsystem or any component is in accordance with the applicable standards, codes, specifications and installation instructions.

Compare the factory acceptance test at 3.1.8, which demonstrates that the system, subsystem or component is in accordance with the specification and is typically conducted at the vendor's facility. The difference is the subject of the sentence: FAT tests the equipment against its specification, SAT tests the installation against codes, specifications and installation instructions. A SAT scoped as "repeat the FAT on site" has missed the definition.

That leaves a gap, and the same standard names the test that fills it. The site integration test at 3.1.20 is inspection and testing at the site to demonstrate that the merging of the various subsystems and components into one overall automation system is completed and that all components work together in accordance with the specification.

Its factory-side twin, the factory integration test at 3.1.9, arrived with Edition 3.0 in July 2024. On a battery plant with a separately supplied battery management system, power conversion system and plant controller, the integration question is usually the one that decides the schedule, and it is a SIT question rather than a SAT question.

Clause 7 gives SAT the same closure machinery as FAT: 7.3 SAT punch list, 7.4 documentation of SAT with a SAT report at 7.4.2 and final documentation at 7.4.3, and 7.5 SAT completion — where a punch list is defined at 3.1.16 as a list of all problems found, the corrective action to be taken, party responsible and date completed.

Then comes the boundary that matters most on a BESS project. Clause 1.2.1 states that activities prior to or after the FAT, FIT, SAT and SIT, such as loop checks and commissioning, are not covered by the document. SAT completion is not commissioning completion, and the standard says so itself.

Why it matters in a real grid-scale project

EPRI's ESIC Energy Storage Commissioning Guide places site acceptance testing precisely: it is one of the final steps before the owner accepts the system and the title, meaning ownership, is transferred. Read the preposition. SAT sits immediately upstream of the transfer; it is not itself the transfer.

The outcome EPRI expects at the end of the phase is formal documentation indicating the system is acceptable and is transferred from the vendor to the utility, owner or operator. The 2025 Morgan Lewis survey of US storage procurement describes the parallel commercial shape — partial completion at delivery, substantial completion at satisfactory commissioning, with specified operational parameters that must be achieved for substantial completion.

The data has a second life. EPRI stresses documenting the explicit test conditions and protocols used to generate the baseline performance data, so the procedures can be repeated later to assess degradation or other issues relevant to warranty claims.

On a grid-scale battery that baseline is the number every subsequent capacity warranty argument is measured against, so temperature, state of charge window, C-rate, auxiliary load and metering point belong in the SAT report rather than in someone's memory. A site test run to a vague protocol produces a baseline nobody can reproduce, and an unreproducible baseline is worth very little when a capacity claim is filed in year six.

Scope it wider than an installation check if the supply is split. EPRI records a case where a battery subsystem passed its factory acceptance test and shipped, a power conversion subsystem from a different vendor passed its own factory acceptance test and shipped, and the integrated system did not operate — extensive field work followed, and each vendor blamed the other.

EPRI's recommendation is to buy an integrated system or otherwise plan for significant on-site integrated system acceptance testing. Translated into a specification: on a multi-vendor BESS the on-site test scope has to reach the integrated behaviour, not just installation conformity box by box.

Typical values and standards

EPRI describes what has to be true before site testing can start: all equipment installed and interconnected to the power system, with permission to exchange power — often a temporary permission first and an operational permission later — so that correct and intended operation can be confirmed.

If a sequence-of-operations test has been defined earlier in the project, EPRI puts it here. The guide also flags that ISO or market-participation testing can add one to two weeks, which is the kind of slip that eats whatever float remains between site testing and the commercial operation date on a project whose revenue model already assumed the plant was earning.

For a concrete picture of what a BESS on-site test list contains at the electrical-plant level, Entergy Louisiana's model BESS scope book lists grounding system testing, megger tests, high-pot or VLF tests, functional tests of all controls, protection relays and interlocks, functional tests of all safety devices and alarms, AC/DC motor tests, battery and UPS tests, PCS test, switchgear test, control circuit checkout, instrument and loop calibration, fire protection test, and all manufacturer-recommended equipment tests.

Most of that is installation conformity, exactly as the IEC definition implies — insulation, terminations, phasing, interlocks — with a layer of functional checks on top.

In the US the commissioning that follows is a code requirement, not only a contract term. IFC 2021 Section 1207.2.1, as adopted in Colorado's DFPC Fire Code 2021, requires commissioning of newly installed ESS, and of existing ESS retrofitted, replaced or returning to service, prior to the ESS being placed in service, in accordance with a commissioning plan approved before commissioning starts.

Section 1207.2.1.2 makes the commissioning report, including the results of the initial acceptance testing, a precondition of final inspection and approval by the fire code official, kept at an approved on-site location. Adoption is not uniform: Virginia's SFPC 2021 instead ties commissioning to operational permit issuance and the applicable building code, and requires the report only on request.

Performance, P-Q and controls on the site test list

The Entergy list is the electrical layer. Above it sits the performance layer, and that is the one the money is attached to. A capacity test produces a single integrated number — the energy delivered between two defined endpoints, normally a full charge, a defined rest, then a discharge at rated power down to the minimum operating state of charge — and the canon used across this site reads it at the point-of-interconnection revenue meter rather than at the PCS terminals.

A round-trip efficiency test produces a ratio instead, which adds an obligation: the charge leg and the discharge leg have to be metered at the same boundary, or the quotient describes two different plants. The measurement-boundary entry covers why that rule bites; what belongs in the site procedure is the naming — which meter, which reference temperature, which state-of-charge window, and whether auxiliary load counts.

Station service is where two honest readings part company: fed from upstream of the named meter it sits inside the result, fed from downstream through the auxiliary transformer it does not, and the same plant can pass or miss on that one line.

Reactive capability changes character when it moves to site. Each converter arrives with a P-Q envelope demonstrated at its own AC terminals, while the obligation is written at the point of interconnection, and everything in between — the medium-voltage collection system, the step-up transformers and the main power transformer — takes vars off the total before the plant number is read. A fleet whose units all cleared their factory curves can still come up short at the high side, and the shortfall surfaces as a failed witness test rather than as a design-review comment.

In the US, FERC Order 827 requires newly interconnecting non-synchronous resources to provide roughly 0.95 leading to 0.95 lagging at the high-side point of interconnection, IEEE 2800-2022 sets performance and verification for transmission-connected inverter-based resources, and IEEE 1547-2018 governs distribution-connected resources and is not a substitute for either.

In the EU the duty arrives through each national grid code's implementation of Regulation (EU) 2016/631, so the envelope is a jurisdiction question before it is an equipment question. The dedicated P-Q capability test is where the plant demonstrates it; the site acceptance scope's contribution is having installed the plant that will be demonstrated.

The controls layer is where site schedules actually go. A battery plant carries a control hierarchy — cell-level battery management, converter controls, an energy management system, a plant controller, plant SCADA, and a remote terminal unit facing the utility — and every measurement and command crossing one of those interfaces is a point on a list that at least three vendors contributed to.

Proving that list is test work rather than document review: a point is proven when a real field quantity appears on the operator's screen with the right scaling, the right engineering units and the right sign convention, and when a command travelling the other way lands where it was addressed.

Telemetry to the grid operator and the automatic generation control path each need their own end-to-end proof, because market qualification depends on them and no balancing authority accepts a mapping table as evidence. Underneath everything sits time synchronization: one clock reference distributed to the battery management system, the converters, the plant controller, the protection relays and the meter, typically by GNSS-disciplined IRIG-B or IEEE 1588 precision time protocol.

Without it a sequence-of-events record cannot be reconstructed, a ride-through response cannot be aligned with the disturbance that caused it, and in the US the NERC disturbance-monitoring and reporting requirements have nothing consistent to draw on.

How it shows up in specs, studies and contracts

Utility specifications sequence the two explicitly. Entergy Louisiana's scope book requires the seller to perform and successfully complete site acceptance testing followed by commissioning tests on the systems and equipment in its scope of supply, to demonstrate the safety, operability and reliability of those systems within specified design limits according to the contract, engineering drawings, documents and specifications.

The same document schedules commissioning turnover documentation at substantial completion. If a specification names SAT and leaves commissioning tests undefined, half the scope is missing, and the gap gets filled by whoever is holding the pen when the argument starts.

Standard-form plant contracts use a different vocabulary for the same chain.

In the World Bank Standard Procurement Document for Plant, Precommissioning is the testing and checking the contractor carries out in preparation for Commissioning; Completion means precommissioning is finished and the facilities are ready for commissioning; Commissioning is operation of the facilities by the contractor following Completion, for the purpose of carrying out the Guarantee Tests; and Operational Acceptance is the employer's acceptance certifying the contractor's fulfilment of the contract in respect of the Functional Guarantees.

The words FAT and SAT appear nowhere in that document. Map your test plan onto the contract's defined terms rather than assuming the contract will recognise yours.

Those definitions turn into gates, and a regulatory chain runs alongside them. GCC 24.5 gives the project manager fourteen days after the contractor's notice either to issue a Completion Certificate or to notify defects and deficiencies, which is the mechanism a SAT punch list actually feeds into, and 25.3.1 sets when Operational Acceptance occurs.

In the US, the FERC pro forma Large Generator Interconnection Agreement defines Trial Operation as the period during which the interconnection customer is engaged in on-site test operations and commissioning of the generating facility prior to Commercial Operation, beginning at the Initial Synchronization Date, and defines Commercial Operation as generating electricity for sale excluding electricity generated during Trial Operation. Commissioning energy is legally distinct from commercial output.

Acceptance testing produces the numbers the commercial documents then live on, so read the baseline clause with the test procedure open beside it. The capacity result taken at commercial operation is either the baseline the later retention checks are measured against, or it is not — and where the warranty is baselined on nameplate instead, a test that comes in above nameplate hands that surplus to the supplier as headroom against the guaranteed curve rather than to the owner as delivered energy.

The same question repeats for round-trip efficiency and for availability, where the acceptance value and the guaranteed value can sit on different boundaries and different measurement windows with neither document being wrong on its own terms. The remedy ladder — cure period, retest, liquidated damages, price adjustment — is contract work rather than test work: the SAT report supplies the measurement and is silent on what it costs.

The artefact that holds all of this together is a commissioning matrix: one row per test, with columns for the scope owner, the clause or standard the test is run against, the witnessing party, the acceptance criterion written as a number, the document the test produces, and the payment or permission gate that document releases.

Build the rows from the contract's defined terms rather than from acronyms, because a single row can be a site acceptance line for the integrator, a code-required acceptance test for the fire code official and a market-qualification test for the system operator at the same time, needing three signatures on one result. Empty witness or criterion cells are the useful early warning — those are the rows that turn into arguments inside the outage window.

Common pitfalls

The first pitfall is treating site acceptance testing and commissioning as one activity.

IEC 62381 Clause 1.2.1 excludes commissioning and loop checks from its scope; Entergy's scope book sequences site acceptance testing and then commissioning tests as separate obligations; and EPRI, adopting the IEEE definition of commissioning as a process that assures a component, subsystem or system will meet the intent of the designer and the user, notes that commissioning closeout typically marks the point where the purchaser takes ownership responsibility of the equipment and the equipment warranties begin.

Two scopes, two sets of acceptance criteria, two sign-offs. Write them separately or find out later which one the money was attached to.

The second is terminology drift, and energy storage has it badly. The DOE Energy Storage Handbook, Chapter 21, never uses the term SAT. Its first formal step of commissioning is the Operational Acceptance Test, which occurs once all the equipment has been installed on site and determines that the individual components operate as specified — insulation resistance, torque on bolts, rotation and phasing of circuits and machines, security of covers and barriers, smoke tests, and point-to-point testing of the electrical controls.

The handbook then reserves FAT for a later on-site Functional Acceptance Test at the completed installation connected to the utility grid, after which the system is considered commissioned and handed off. Read the definitions clause; assume nothing from the acronym.

The third is specific to the controls scope: signing off the point list as a document. A mapping table reviewed in an office proves that two vendors agree on register numbers. It does not prove that a positive number on the operator's screen means export rather than import, that a Mvar reading is not scaled by ten, or that a plant-controller setpoint still reaches the converters when the primary communications path is down.

Each of those survives a table review and fails on first dispatch. Verify points end to end, from the field quantity to the screen and back, and treat the sign convention as a test case rather than as a convention.

Three smaller traps close the list. Routine tests are not automatically factory tests — IEC 62271-200:2021+AMD1:2024 lists 8.104 tests after erection on site and 8.105 measurement of fluid condition after filling on site among its routine tests, although only the contents page is publicly visible, so do not assume they bind unconditionally on every project.

IEC 62933-2-1:2017 is scoped to unit parameters and testing methods and sets no acceptance framework; that comes from the contract, from IEC 62381, or from the owner's specification. And under the FERC pro forma LGIA, the Article 6.3 right to observe testing is scoped to each party's Interconnection Facilities, so the transmission provider has no pro forma right to witness testing of the battery plant itself.