Factory test and site acceptance: what is checked before handover


A factory acceptance test proves the set about to ship; a site acceptance test proves it where it will run. The factory test is a no-load run, then a load bank test at the rating on the order, with the readings recorded on a test sheet. Site acceptance repeats the load steps on the real load.

What the factory test is, and what it is not

Three different exercises get called “the test”, and they prove three different things. A type test is done once on a design, on a sample, in a laboratory, and it says how that design behaves; it is not evidence about a particular machine. A factory acceptance test is done on this set, at the works, before it is packed, and it says this machine met its declared rating and its specification on the day. A site acceptance test is done again after installation, and it adds everything the factory cannot see: the foundation, the exhaust, the fuel supply, the cooling air, the cabling and the real load.

The three exercises that get called a test, what each one proves, and what it leaves behind. They are answered by different documents, which is why they cannot be used interchangeably.
Exercise What it proves Where it happens What it leaves behind
Type test How a design behaves, tested on a sample rather than on your machine In a laboratory, once per design A type test report, quoted by model
Factory acceptance test That this set met its declared rating and its specification before shipment At the works, on the test bay A factory test report and a signed test sheet
Site acceptance test That the installed set and the services around it work together on the real load On site, after installation A site test record and the settings in writing

This page is about the second and the third. The first is a document to ask for and to read carefully, because a type test report describes a design and not a machine; where it sits among the conformity documents is set out in generator export certification.

What the factory acceptance test covers

A factory test is not one measurement, it is a sequence, and the sequence is what makes the report worth reading. The set is run and checked in a fixed order, with the load bank near the end, because a fault found before the load bank is cheaper to correct than one found on it.

The sequence inside a factory acceptance test, and what each stage proves. The order matters as much as the list, because each stage is cheap to correct only until the next one has run.
Stage What is checked What it proves
Visual and identification The model, the serial numbers of the engine and the alternator, and the rating plate against the order and the datasheet The machine on the bay is the machine on the order
Wiring and insulation Insulation resistance and continuity of the alternator, the panel and the cabling No insulation fault and no loose connection before first run
No-load run Starting, run-up, oil pressure, coolant temperature, voltage and frequency at no load, and the charging alternator The engine runs and the alternator builds voltage
Safety devices Emergency stop, low oil pressure, high coolant temperature and overspeed, each operated in turn Each device actually shuts the set down
Protection and control The controller indications, the alarms, and the settings against the commissioning sheet The panel is wired as built and the settings are the ones on the sheet
Load bank test The set at the agreed load steps, holding the rating and the duty The set carries its declared load rather than merely running
Fuel consumption Fuel used at the test load, converted to the specific figure The consumption quoted is the consumption measured
Packing and accessories The accessory list against the packing list, and the documents that travel with the set What was ordered is in the crate

The load bank is the stage that separates a test from a start-up. A set can start, idle and sound healthy while the fuel system, the voltage regulator or a protection setting is wrong, and only load exposes it. Where a load bank is not available, the honest substitute is a resistive or reactive load that actually matches the duty; a set that is started and idled has proved only that it starts.

The load bank test, step by step

A load bank is a bank of resistive elements, with reactive elements added where the specification calls for them, that turns the set’s output into heat so the set can be loaded without a building to power. The test is run in steps rather than as one jump to full load, and each step is held until the readings have settled.

The steps of a load bank test, what each one is for, and what is watched at it. The percentages are of the set’s declared rating at the conditions on the order.
Step The load What is watched Why the step is there
Warm-up No load, until the engine is at normal running temperature Oil pressure and coolant temperature Readings taken cold are not the readings on the sheet
First step About 25 % of the rating Voltage, frequency and stability The first check that the set accepts load at all
Second step About 50 % Voltage and frequency against the declared window The band most sets run in for long periods
Third step About 75 % The same, plus exhaust and coolant temperature Where a cooling or fuel problem begins to show
Full load 100 % of the rating, held until the temperatures stabilise Every reading, and the fuel consumption The only step that proves the rating line
Block load and recovery One large step, where the specification asks for it The voltage dip, the frequency dip and the recovery time The number that decides whether a motor will start
Shutdown and hot restart Stopped and restarted while hot, where the specification asks for it The restart time and the behaviour A hot engine finds a weak starter or a weak battery

The full load step is the one worth insisting on, and it is the one most often cut short. A bank that tops out at 75 % proves three quarters of the rating, and the remaining quarter is where the alternator’s temperature rise, the engine’s fuel delivery and the protection settings are actually exercised. Where the full rating cannot be reached on the bay, say so on the sheet rather than calling the test a full load test. The single step matters for a different reason: it is the transient the set has to ride through when a pump or a chiller starts, and the dip and the recovery time are what a starting study is built on, as set out in generator starting current and motor starting.

What the readings are compared against

A reading on its own means nothing; it means something against a reference. Four references are used, and all four belong on the test sheet, so that the numbers can still be read years later when the set is being serviced or re-sold.

What each group of readings is measured against, and what a difference from the reference means. The acceptance criteria are set by the datasheet, the declared performance class and the project specification, not by this page.
Reading Compared against What a difference means
Voltage and frequency at a steady load The datasheet rating line and the declared performance class to ISO 8528-1 The class sets the steady state window, so the same deviation passes one class and fails a stricter one
Voltage and frequency on a step The transient limit of the declared class A dip that recovers inside the limit is acceptable; one that does not is a sizing or a regulator question
Rated output The declared kVA and kW at the declared power factor, ambient and altitude A rating is declared at a reference ambient, 40 °C in the site’s figures, and the set derates above it, so the comparison has to be at the conditions on the order
Fuel consumption The figure quoted for that load Consumption well above the curve points at the fuel system, the injectors or the load measurement
Insulation resistance The alternator maker’s minimum, from a 500 V test A low reading on a new machine is a stop rather than a note
Protection operation The settings on the commissioning sheet A device that operates outside its setting is a wiring or a setting problem

The performance class is the reference most often left off a quotation. Two sets with the same engine and the same alternator can be declared to different classes, and the class is what the voltage and the frequency are accepted against. Put the class on the enquiry, and the factory test has a reference to be measured against; without it, the only standard is that nothing smoked. The settings that appear on the commissioning sheet itself, and the windows they are drawn from, are worked through in generator control panel and protections.

Attend in person, witness remotely, or accept the report

There are three levels of buyer involvement, and they are a choice rather than a price. What changes between them is how much of the test the buyer sees, and how much can be argued about afterwards.

The three levels of involvement in a factory test, what the buyer does in each, and what comes back. The test itself and the report are the same in all three.
Option What the buyer does What comes back
Attend in person Travels to the works, watches the no-load run and the load steps, and checks the accessory list A signed test sheet, and any question answered while the set is still on the bay
Witness remotely Joins a live video call through the test and asks for specific readings to be shown on camera Photographs, video and the readings sheet, and the same right to see the numbers
Accept the report Lets the test run and reads the report and the photographs afterwards The factory test report, the readings and the test date

At every level the test is booked with the production slot rather than after it, and the set is not packed until it is accepted. The commercial sequence behind that, the factory pre-acceptance step, the instalments and the warranty, is set out on the payment, delivery and warranty page, and this page explains the test that sits inside it rather than restating the terms.

The test sheet, and what should be on it

The test sheet is the document that makes the test auditable years later, and it is the one most often missing when only a report is sent. A report summarises; the sheet carries the readings. Ask for the sheet, and check that these are on it before the set is packed.

  • Identification. The model, the serial numbers of the engine and the alternator, and the rating that was installed.
  • The conditions. The ambient temperature, the altitude and the fuel used, because a rating is only that rating at its declared conditions.
  • The load steps. The load at each step, with the steady readings of voltage, current, kW, power factor, frequency and speed.
  • The step response. The deepest voltage dip, the deepest frequency dip and the time to recover, where the specification asks for them.
  • Fuel consumption at the test load, and the specific figure it works out to.
  • Insulation resistance, with the test voltage used, so that a later reading can be compared with this one.
  • Each protection trip, the setting it operated at, and the result of the test.
  • The start time, and the hot restart where it was tested.
  • The date, the place and the signature of the person who ran the test and the person who witnessed it.

A sheet with a pass or a fail and nothing else is not a test record. The reason to insist on the readings is the same reason the maintenance record matters: a comparison needs a baseline, and the factory test is the only one a set will ever have on a known load, as set out in the maintenance schedule.

Site acceptance: what changes when the set is in place

The factory proves the set, and the site proves the installation and the set together. A machine that passed on the load bank can still fail to carry the load on site, because the foundation, the exhaust, the fuel line, the cooling air, the cabling and the site protection are all outside the factory’s control.

What a site acceptance test adds to the factory test, and who owns each check. The site test is the one that proves the installation, not the machine.
Check What is confirmed Who does it
Installation and services The foundation or the anti-vibration mounts, vibration, the exhaust route and its back pressure, the fuel line and the tank, and the cooling air in and out The installer, with the set running
Electrical connection Cable sizes and terminations, the changeover or the ATS, the phase rotation, and the earthing and neutral arrangement as designed The electrical contractor
Protection settings The site settings against the commissioning sheet, and the grading with the devices downstream The commissioning engineer
Load steps on the real load The set carrying the actual load in the steps the site will see, up to the largest single step Commissioning, with the load available
Transfer and black start The ATS or the AMF transferring on a mains failure, and the set taking the load inside its time Commissioning, with the mains isolated
Paralleling, where fitted Load sharing, the synchronising window and the protection settings under load The switchgear commissioning
Record and handover The settings in writing, the readings, and the instruction of the site staff The supplier or the contractor

Site acceptance is also where the earthing and neutral arrangement is confirmed as built, because the factory never sees the board the set will feed. That arrangement decides the fault current and therefore the protection settings, and it is worked through in generator earthing and the neutral; when the arrangement on site is not the one on the drawing, the settings that were derived from it are wrong, and the two are commissioned together for that reason. Where the set will run in parallel with another, the switchgear adds its own commissioning on top, including a load sharing demonstration under load, as set out in generator paralleling.

What a failed test does to the schedule and the money

A fault found on the test bay is the cheapest fault of the project, and the terms are arranged so that it is found there. If the set fails the factory acceptance test, it is repaired or re-tested before shipment, and the balance instalment does not fall due until the set is accepted; nothing is packed and nothing sails until the test is passed.

The cost of a failure is therefore time rather than money in most cases, and the time appears in one of three places. A re-test needs a fresh test slot, which may sit behind other orders in the programme. A configuration change after the test, such as a different voltage, a different panel or an added transfer switch, invalidates the test, and it is run again. And a fault that is only found on site, after shipment, is handled as a warranty claim against a set that is already installed, which is the most expensive place to find it. The sequence and the money around it are set out in generator lead time, packing and shipping terms, and the terms themselves come from the payment, delivery and warranty page, which sets out the instalments and the acceptance step itself.

That is also why the price of a test is not the price of a load bank. The value of the test is the fault it finds before the crate is closed, and the cost of a load bank hour is a fraction of the cost of the same fault in a plant room.

The mistakes that cost the most at handover

Most of the cost at handover comes from a short list of habits, and each one is avoidable at the enquiry stage rather than on the bay.

The habits that cost the most around a factory test and a site acceptance test, and what prevents each one. All of them are decided before the set is built, not during the test.
Mistake What it costs What prevents it
Testing the set at idle and calling it a test A fault that only appears under load surfaces on site instead A load bank test at the agreed steps, with the full load step actually held
Recording nothing, or only a pass or a fail Nothing can be compared later, and a warranty claim has no baseline A test sheet with the readings, the conditions and the date on it
Booking the witness after the build The test slot is taken and the set ships before anyone sees it Book the test with the production slot, at the order stage
Leaving the protection at the factory default The panel is proved rather than the set, and the grading is untested The commissioning sheet, checked on the bay and again on site
Signing acceptance before the load test The protection the order carries, the balance instalment, is released early Acceptance after the test, which is the sequence in the terms
Skipping site acceptance where the site load is unusual A problem with the installation is blamed on the set A site test on the real load, with the settings handed over in writing

None of this should be taken as a fixed acceptance criterion for a particular project, and a few limits are worth stating. The criteria themselves come from the datasheet, the declared performance class, the project specification and the local installation rules, and a load bank test proves the set rather than the site’s distribution. Where a project asks for a test that is not part of the standard sequence, such as a named transient limit, an endurance run or a witnessed block load at a stated figure, it belongs in the enquiry, because it changes both the test slot and the price. And a test proves the machine on the day it runs, not on the day it is needed; what keeps the second day honest is the maintenance schedule and, on a standby site, the routine no-load and loaded running that keeps the machine exercised.

Questions about factory testing and acceptance

What do FAT and SAT mean?

A factory acceptance test, shortened to FAT, is the test at the works before the set is packed, and a site acceptance test, shortened to SAT, is the test on site after installation. The factory test is a no-load run followed by a load bank test at the agreed steps up to 100 % of the rating; the site test repeats the load steps on the real load and adds the installation, the transfer and the settings. Both produce a record, and they are not alternatives, because each proves something the other cannot.

What is a load bank test?

It is a test in which the set is loaded on resistive elements, with reactive elements added where the specification calls for them, instead of on a building. The set is run in steps, usually about 25 %, 50 %, 75 % and 100 % of the rating, with each step held until the readings have settled and the full load step held until the temperatures stop rising. A 500 kVA set is being asked for about 400 kW at 0.8 power factor, so the bank has to be rated for the real power of the set and not only its kVA.

What is the difference between a factory test and a type test?

A type test is done once on a design, on a sample, and it describes how that design behaves; a factory test is done on the machine you are buying, before it ships. The type test report is a document you ask for and read, and the factory test report is evidence about your set. A type test report naming IEC 60034 tells you how the alternator was classified, and it does not tell you that this set will hold voltage on your load step, which is what the load bank is for.

Can I witness the test without travelling?

Yes. The same test is run and the buyer joins it live on a video call, asking for the specific readings to be shown on camera, and the readings sheet, the photographs and the video are sent afterwards. Remote witnessing is arranged when the order is placed, at the same time as the test slot is booked with the production slot, and it carries the same right to see the numbers as attending in person.

What happens if the set fails the factory test?

It is repaired or re-tested before shipment, and the balance instalment does not fall due until the set is accepted. Nothing is packed and nothing sails until the test is passed, which is why a fault found on the bay costs time rather than money. A fault found only on site is a different matter: it is handled as a warranty claim, under a warranty of 12 months from the works date or 1000 running hours, whichever comes first.

What should be on the factory test sheet?

The identification of the machine, the conditions it was tested at, the load at each step, and the steady readings of voltage, current, kW, power factor, frequency and speed at that step; then the step response, the fuel consumption at the test load, the insulation resistance, each protection trip and its setting, the start time, and the date and the signatures. A sheet that carries a pass or a fail and nothing else is not a test record, because nothing on it can be compared with a reading taken five years later.

Do I still need a site acceptance test if the factory test passed?

Yes, where the installation is anything more than a set on a skid with a cable to it, because the factory cannot test the foundation, the exhaust back pressure, the fuel supply, the cooling air or the site’s own protection. The site test repeats the load steps on the real load, up to the largest single step the site will apply, and confirms the transfer, the settings and the earthing arrangement as built. Where the site load is not yet available, the honest sequence is a set of readings on the load that can be applied, and the rest of the load steps deferred and recorded when the load arrives.

Send the specification, not only the rating

Give us the rating and the configuration, and any clause in the specification that touches the test: the load steps, the hold at full load, the block load and the recovery time, the performance class, whether a witness is wanted, and whether the site load will be available for a site test. The reply carries the test sequence for that set, the readings you will get on the sheet, and the test slot against the delivery date.

Request a quote WhatsApp the sales desk

Written and maintained by the KVA Grid sales engineering desk. The test sequence described here is the standard one for a low voltage set; the acceptance criteria come from the datasheet, the declared performance class to ISO 8528-1 and the project specification, and a project that asks for a different test is quoted against that test rather than against this page. The commercial sequence behind the test, the 30 % advance payment, the 70 % balance before shipment, the factory pre-acceptance step, the warranty of 12 months or 1000 running hours and the 7 working day claim on arrival, follows the payment, delivery and warranty terms, and the documents that travel with the set are listed in generator export certification. How a failed test is priced into the life of the machine is a separate question, set out in total cost of ownership, and what a test report proves about a used set is covered in used generator inspection. Nothing on this page is a quotation for a particular project. Reviewed 10 October 2026.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top
WhatsApp