FAT and SAT done right — a 14-point field checklist

If you are the quality engineer, procurement manager, or installation supervisor witnessing a FAT at a supplier's factory next month, this post is written for you. The assumption is that you have read the specification, the PO is placed, the rig exists, and you are about to get on a plane — or, increasingly, drive four hours to the supplier's Pune or Bengaluru facility — to witness the acceptance test that decides whether the rig ships to your site or goes back into rework.

Most FAT visits we have seen are under-prepared, under-scoped, and over-trusting. The witnessing team arrives the morning of the test, reviews a protocol they have not pre-read, watches the supplier's engineers demonstrate what the supplier chose to demonstrate, signs the form, and flies home. Four months later, when the rig misbehaves on site, they discover that three-quarters of the acceptance criteria were never actually tested. This post is the corrective.

1. What FAT and SAT actually are

Before anything else, the definitions. FAT and SAT are contractual gates, not inspections. The distinction matters because it decides what the supplier is obliged to do when something fails.

Factory Acceptance Test (FAT)

FAT is a witnessed test of the complete rig, conducted at the supplier's factory, against a written acceptance protocol agreed in advance. It takes place before the rig is crated for dispatch. The supplier provides the factory space, the utilities, the reference instrumentation, the test fluid, and their commissioning engineers. The buyer provides a witnessing team with written authority to sign acceptance or refuse it. A failed FAT means the rig does not ship; the supplier remediates at their own cost and the FAT is re-witnessed. A signed FAT triggers the dispatch payment tranche — typically 40 to 60 percent of the contract value.

Site Acceptance Test (SAT)

SAT is a witnessed test of the installed rig, conducted at the buyer's site, against a protocol that overlaps the FAT protocol but adds site-specific tests that FAT cannot cover — real incoming power quality, real ambient conditions, real interface to your upstream and downstream systems, and real operator use. A failed SAT means the supplier does not get their final payment; they remediate until SAT is re-witnessed and signed. A signed SAT starts the warranty clock, triggers the final payment, and hands legal custody of the rig to the buyer.

Why this is not an inspection

"Inspection" is a weaker word. An inspection finds problems; someone still has to decide what to do about them. FAT and SAT are gates: the rig either passes or does not, and if it does not, a defined commercial consequence follows. That consequence — the supplier's unpaid balance — is the only real leverage you have in the entire contract after the PO is placed. Everything before FAT is preparation. Everything after SAT is warranty claims. The two gates are where the rig is actually bought.

Most of what follows is about not wasting those two gates.

2. Before you fly to the factory — 5 checklist items

The FAT is won or lost in the two weeks before you arrive at the factory. A witnessing team that walks in unprepared is witnessing the supplier's demonstration, not running their own test.

1. Get the acceptance test protocol in writing, 2 weeks before the visit

The acceptance test protocol is the line-by-line script the rig will be tested against. Every requirement in the specification should map to at least one test step in the protocol. You want this document on your desk at least 14 days before the scheduled FAT date, so you have time to read it, red-line it, and negotiate changes before you are in the factory with your engineers waiting. If the supplier's answer is "we'll finalise the protocol when you arrive", the FAT is already compromised. The protocol is a pre-agreed document, not an on-the-day creation.

Specifically ask for: the test step numbering, the acceptance criteria per step (numerical, with units), the reference instrumentation each step relies on, the expected duration per step, the sequence of steps, and the supplier's staff assigned to each step. A protocol without all of these is a draft, not a protocol.

2. Bring your own reference instrumentation (if budget allows)

Every number that comes off the rig at FAT will be read by a sensor installed by the supplier, calibrated by a laboratory of the supplier's choosing, traced to a standard the supplier nominated. That is not a conspiracy; it is the default. But where the stakes justify it, bring at least one independent reference — a calibrated deadweight tester for pressure, a master flow meter, a traceable thermocouple, a load cell with a current NABL certificate. At a minimum, ask for the calibration certificate of every primary reference instrument on the rig, issued no more than 6 months before the FAT date, and check that the uncertainty declared is adequate for the accuracy class being demonstrated.

On a 1,800-bar rig with a claimed 0.25 percent FS accuracy, a reference transducer with 0.5 percent FS uncertainty cannot prove compliance. The reference has to be at least 3 to 4 times more accurate than the device under test.

3. Pre-read every standard cited in the spec

The specification references standards — ASME BPVC Section VIII, ISO 10012, MIL-STD-1522, API 520, NABL, ISO 17025, whatever applies. Every one of those references is a promise the supplier has made to deliver against a body of technical requirements. The supplier's engineers will assume you have not read them. In most cases they will be right.

Pre-reading gives you two things. It tells you what the clause actually requires — often narrower or broader than the specification's paraphrase. And it tells you what test method is expected. "Per ISO 10012" carries a specific meaning about measurement management systems that the supplier may be interpreting loosely. If you know the standard, you can ask the specific question at the specific moment. If you do not, you will watch the demonstration and sign the form.

4. Plan for 1.5 times the scheduled FAT duration

The supplier will schedule the FAT as "3 days, 9 AM to 6 PM". Book your travel, accommodation, and return ticket for 4.5 to 5 days. Rigs are paused mid-FAT for calibration corrections, sensor replacements, PLC patches, software bugs that only surface under witness, and utility issues at the factory. An FAT that runs exactly to schedule is a red flag in its own right — either the protocol is thin or the team is skipping steps to finish on time.

Your organisation's procurement and travel desk will push back on this. Hold the line. A two-day overrun spent on the supplier's floor, before dispatch, costs less than a two-month overrun on your own floor, after commissioning.

5. Line up your backup resources in-country for the day you are stuck

At some point during the FAT, you will need to make a call the protocol did not anticipate. An instrument is missing its calibration certificate. A test sequence produces a result just inside acceptance but outside what you would accept in service. The supplier is proposing a workaround that looks reasonable and you cannot be sure. Before you leave, know who you can call — a senior QA colleague, a design authority, a standards specialist — who can pick up the phone at short notice and help you make the decision. Write their numbers down on paper. Factory Wi-Fi in industrial estates is unreliable. Indian mobile coverage in large fabrication halls is worse.

3. On-site FAT — 9 things to verify

These are the 9 checkpoints that, in our field experience, separate a defensible FAT from a rubber-stamped one. They are not in priority order; in any given FAT, all 9 apply.

6. Physical inspection: welds, finishes, electrical workmanship, labelling

Before the rig is powered up, walk around it for 45 minutes. Look at the welds on the main frame, the hydraulic manifold, the reservoir, and any pressure-retaining components. Look for undercut, porosity, spatter not cleaned off, and any visible grinding on a weld that is supposed to be as-welded. Look at the painting and finish — runs, thin spots, overspray onto instrumentation. Open the electrical enclosure and look at cable dressing, gland tightness, ferrule presence on every conductor, and the separation of high-voltage and low-voltage cable runs. Check that every pipe, cable, and component is labelled per the drawing and that the labels match the as-built document.

A rig that looks untidy at FAT will not tidy itself up in transit.

7. Safety systems: e-stops functional from every workstation, interlocked doors

Before any pressurisation, test every emergency stop button on the rig. Press each one while the system is idle and verify the control system annunciates it, the hydraulic power pack depressurises safely, and the reset sequence requires positive action at the local station (not just a remote reset from the HMI). Test every interlocked door, guard, and safety gate. Open each during an operating condition and verify the rig reverts to a safe state within the time specified. If the rig has light curtains, two-hand controls, or pressure vent valves, test each one independently.

Safety systems are the one area where the correct posture is active distrust. They are tested infrequently in service and failures are not self-reporting. FAT is the single best time to prove every safety device, one at a time, in isolation. Budget a full half-day.

8. Calibration certificates physically present and current

Ask to see the calibration certificate — the physical paper, not a scan on someone's laptop — for every primary transducer, reference instrument, and master gauge on the rig. Check three things on each: the issue date (not more than 12 months old, ideally not more than 6 months at FAT), the laboratory accreditation number (NABL or ILAC-MRA signatory), and the uncertainty declared. Cross-check the serial number on the certificate against the serial number on the instrument. Mismatches are common and are usually clerical, but until the paperwork matches the hardware, nothing is certified.

9. PLC code version matches the signed-off revision

The PLC code has a version number. The SCADA project has a version number. Both should match the revision that was signed off in the design review. Ask the supplier to display the version string on the HMI or on the programming laptop. Cross-check against the revision number in your design documentation. This takes 3 minutes and catches the most common silent substitution in any FAT — a debug build, an unapproved patch applied on the factory floor, or an older version that had not been replaced. If the supplier cannot produce a version string on demand, their configuration management is not real.

10. Alarm testing: deliberately trigger each alarm condition

Every alarm the rig is supposed to raise needs to be triggered, at least once, during FAT. Not simulated via the HMI's "force alarm" debug screen — actually induced at the physical sensor. Low reservoir level: drain to the trip point. High fluid temperature: run the heaters to the trip. Low pilot pressure: close the pilot supply. High vibration, over-current on a motor, loss of communication to a remote I/O module — each one gets a deliberate trigger and the alarm is verified in the HMI, the audit log, and any remote annunciation. This single checkpoint routinely surfaces 3 to 6 defects that were invisible in normal running.

11. Automated test sequence: let the bench run its own test-cycle autonomously

Ask the supplier to set up the rig to run one of your standard test recipes in full automatic mode, from part loading (or a dummy load representation) through to part unloading, and observe it running for at least three complete cycles without intervention. Your job during this hour is not to watch the rig — it is to watch the operators. If the supplier's engineer stays within arm's reach the whole time, fiddling with the HMI, clearing minor alarms, restarting the SCADA, the rig is not actually autonomous. The "fully automated" claim on the spec sheet needs to survive an observation window with no human hands.

12. Manual override: verify every manual bypass works

Every rig has manual overrides — handpump backup, local valve actuation, bypass switches, maintenance mode. These are the things the maintenance team will use at 2 AM on a Sunday when the automation fails. They get tested at FAT and essentially never again. Walk through each one: manual pressurisation via handpump to verify the circuit holds, local/manual selection on each motor starter, maintenance bypass for each interlock (verifying the safety system still engages the critical protections even in maintenance mode). Write each one on the protocol as tested.

13. Data logging: stress-test the data system

Data logging fails in three modes that do not show up in normal operation: high-rate capture (where the system starts dropping samples), power interruption (where the last several minutes of data are lost or corrupted), and disk full (where the system either overwrites the oldest data without warning or stops logging silently). Test all three at FAT.

• High rate. Configure the rig to log at its stated maximum rate on all channels simultaneously. Run for at least 5,000 samples. Export the data and check the timestamp continuity. Missing samples or timestamp drift are defects.
• Power interruption. Mid-test, trip the control panel main isolator. Restart. Check that the log file contains a clean boundary — either the last flushed sample or a marked end-of-file — rather than a corrupted tail record.
• Disk full. Configure the logging directory near its capacity and run until the disk fills. The rig should either annunciate, rotate old logs per a documented policy, or halt — but it must not silently lose data.

14. Documentation package: every drawing, every BOM, every calibration certificate, every software backup

At the close of FAT, the documentation package should be physically handed over — not promised in the shipping crate. Specifically: the as-built general arrangement and hydraulic/pneumatic/electrical schematics, the bill of materials with supplier part numbers, every calibration certificate from checkpoint 8, a USB or external drive with the PLC source code, the SCADA project file, the HMI screens, the version-controlled configuration files, the operator manual, the maintenance manual, and the spare parts list. If any of these items is "coming with the shipment" or "to be issued next week", do not sign the FAT.

The documentation package is the rig's long-term maintainability. Once the final payment has cleared, the commercial pressure to complete it has evaporated. Collect it at FAT.

4. The "reject on the spot" conditions

There are specific situations in which signing FAT acceptance is the wrong choice, regardless of schedule pressure or travel cost. If any of the following occurs and is not remediated during your visit, refuse to sign and require a re-witness.

In 20 years of witnessing and being witnessed, the single most common recovery from a serious FAT defect is a re-visit four to eight weeks later. It feels painful at the time. It is far less painful than discovering the same defect at SAT, or in the third month of warranty, when the supplier has been paid.

5. Site Acceptance Test (SAT) — the integration layer

FAT proves the rig works on the supplier's floor, with the supplier's utilities, at the supplier's ambient conditions, operated by the supplier's engineers. None of those variables are yours. SAT is where the rig is tested against your variables. It is a materially different test, and the protocol should reflect that.

What SAT uniquely tests

• Your utilities. Your incoming power quality, your cooling water flow and temperature, your compressed air pressure and dew point, your nitrogen purity if applicable. The rig's specification states required utility quality; SAT proves it operates correctly on what you actually supply.
• Your environmental conditions. Ambient temperature range, humidity range, particulate loading, vibration from adjacent equipment. A rig that ran happily in a climate-controlled factory at 24°C may misbehave on a Nagpur shop floor in May.
• Your operator ergonomics. The rig is installed in a specific position, with specific sight lines, specific reach distances, and specific foot traffic patterns. Controls that were comfortable in the supplier's test bay may be unusable when the rig is boxed in between two other benches.
• Your interface to upstream and downstream systems. Parts arrive from a specific conveyor, a specific jig, a specific handling device. Data goes out to a specific MES, a specific historian, a specific QMS. None of these interfaces existed at FAT. All of them must work at SAT.

The SAT protocol typically re-runs a subset of the FAT acceptance steps on the installed rig — because utility quality and environmental factors can shift a compliant FAT result to a non-compliant SAT result — and then adds site-specific tests that FAT could not cover.

6. Turning SAT into a force multiplier

The way most teams run SAT, it is a box-ticking exercise: re-verify a few numbers, sign the form, release the final payment, move on. This is a waste. A well-run SAT generates four things that pay back for the next 15 years of the rig's life.

Operator training hours that actually stick

SAT is the one and only time the supplier's commissioning engineer, your maintenance team, and your operators are all physically present at the rig, with every subsystem in working order, with time allocated and no production pressure. This is the window in which real training happens — not the classroom sessions, not the manual handover, but the "walk me through how you would respond if X happens" conversations on the rig itself. Insist on at least 2 full days of supplier-led operator walkthrough during SAT, and 1 full day of maintenance walkthrough. Have the participant list signed by both sides.

A baseline data set for future drift comparison

During SAT, capture a reference data set: 10 cycles of each production test recipe, stored on durable media, labelled with date, operator, utility conditions, and calibration state. Three years from now, when the rig is suspected of drifting, this is the baseline you will compare against. Every serious facility keeps one. Most do not, and when they need it, they cannot reconstruct it.

A punch-list of minor deviations the supplier is still incentivised to fix

No SAT is perfect. Every SAT produces a punch-list of minor items — a mislabelled indicator, a cable tray a little loose, a software screen that displays the wrong unit, a sluggish alarm acknowledgement. Capture every one. Negotiate a written closure plan with the supplier before signing the SAT. They have not yet been paid the final 10 to 20 percent of the contract; that is your leverage to close these items within weeks, not years.

Relationship capital with the supplier's commissioning team

The commissioning engineer who spent three weeks installing and testing your rig knows it better than anyone will for the next decade. Get their direct number. Build a working relationship. Three years from now, when something misbehaves and the supplier's front office routes you through two layers of tier-1 support, the person you want on the call is that commissioning engineer. Relationships built during SAT are one of the most underrated asset classes in industrial procurement.

7. The commonly-skipped SAT tests

The tests most often skipped at SAT are the ones most likely to fail in service. In order of how often we see them omitted:

Long-duration soak test

A lazy SAT runs one or two cycles and calls the dynamic testing complete. A real SAT runs the rig continuously for at least 8 hours, ideally a full 24-hour period including a thermal soak. Hydraulic systems drift as oil temperature stabilises, seals settle, and control loops find their working point. Repeatability over 24 hours is the number that actually predicts production behaviour. Two cycles predict nothing.

Full environmental envelope

The specification states an ambient operating range. SAT should demonstrate at least the cold start at minimum ambient. If the rig is rated for 5°C to 45°C operation and SAT is conducted in April at 28°C, the low-end and high-end behaviour is untested. Where a full-envelope test is impractical, at minimum capture the ambient conditions at SAT and insist on a written supplier commitment covering drift across the specified envelope.

Communication loss and recovery

Any rig with remote I/O, a PROFINET or EtherCAT fieldbus, a SCADA running on a separate PC, or a connection to a site historian has a communication layer that can fail. Test it. Physically disconnect the Ethernet cable between PLC and SCADA during a running test. Verify the rig enters a safe state, the disconnection is logged, the reconnection is clean, and any buffered data is flushed without corruption. Five minutes of disconnection should not produce unexplained behaviour.

Power sag and surge recovery

Indian industrial power is not European industrial power. Voltage sags, brownouts, phase imbalance, and brief interruptions are normal. The rig's UPS and power conditioning should handle events within its specified envelope without dropping the test. At SAT, deliberately sag and interrupt the supply (within controlled limits your electrical team signs off) and verify the rig recovers. If the supplier is not willing to do this, the UPS is decorative.

Backup generator transition

If the rig is on a DG set-backed supply, the transition from utility to DG and back produces a brief voltage and frequency disturbance. The rig should ride through it. This test is skipped at nearly every SAT because it requires coordination with the plant electrical team. Coordinate it. A rig that cannot survive a DG transition will lose mid-test data every time the utility hiccups.

8. Post-SAT handover discipline

SAT is signed. The final payment is about to release. Before it does, the following should be in writing and physically in your possession:

• The signed SAT protocol with every test step initialled, with the actual measured values recorded next to the acceptance criteria.
• The final calibration certificates against your site-installed reference standards, issued within the 30 days preceding SAT.
• The as-built drawings, with any site modifications marked in red and countersigned by the supplier's project engineer.
• The final PLC and SCADA code backup, on physical media (USB stick or external drive), with a readable version string and a date.
• The full operator and maintenance manuals, in the language of your operating staff, reflecting the as-installed configuration.
• The spare parts list with manufacturer part numbers, recommended stock levels, and lead times.
• The operator training attendance register, signed by every attending operator and by the supplier's trainer.
• The punch-list from checkpoint 6, with a committed closure date for each item and the retention amount linked to its closure.

Any item missing from this list is a reason to hold final payment. The supplier's counter-argument will be that the missing items are "administrative" and will follow in the coming weeks. They will not. Every document not handed over at SAT has a vanishing probability of ever being handed over, because the commercial leverage to produce it has gone.

FAT and SAT are not procedural. They are the two points in the contract where the buyer pays for something and where the buyer can still change what they get. The rigs that work for 15 years and the rigs that spend their lives in partial service are separated less by manufacturer and more by how disciplined the buyer was at those two gates.

The 14 checkpoints above are not novel. They are what good commissioning engineers do instinctively. This post exists because in most organisations the instinct is not written down, the experienced person is not on the team that goes to this particular FAT, and the checklist ends up being whatever the supplier emailed two days before the visit. Write your own checklist, pre-read it, and walk into the factory knowing what you are there to verify. The rig will be better for it and so will the next twenty years of running it.

If you are still at the earlier stage — choosing which supplier's FAT you will eventually witness — our pillar guide on how to choose a test bench manufacturer walks through supplier evaluation from installed-base checks through factory audit to PO placement. It is the upstream companion to this post.