The Machine That Passed FAT and Failed SAT
We built an assembly machine. The FAT (factory acceptance test) at our shop went perfectly — it ran 8 hours straight at the design cycle time, all stations passed, the customer signed off. Two weeks later, at the customer’s plant (SAT), the machine didn’t run right. The air pressure was different (our shop had 7 bar, theirs had 5.5 bar). The foundation wasn’t level (our floor was precision-ground, theirs was a standard concrete slab). The parts coming off their line had slightly different dimensions than the sample parts we tested with. We spent a week on site fixing issues that FAT should have caught. The problem wasn’t the machine — it was that FAT tested the machine in ideal conditions, not in the customer’s actual environment.
FAT and SAT for custom machinery is the process of verifying the machine before and after delivery. Done right, it catches issues at the builder’s shop (where they’re cheap to fix). Done wrong, it pushes them to the customer’s floor (where they’re expensive and embarrassing). This article is how I run FAT and SAT so the machine actually works when it arrives.
FAT vs. SAT: What Each One Does
FAT (Factory Acceptance Test)
FAT is the test at the builder’s facility, before the machine ships. The customer comes to the shop (or watches via video) and verifies that the machine meets the spec.
What FAT verifies:
- The machine runs the designed cycle time.
- All stations perform their functions correctly.
- The safety systems work (e-stop, guards, light curtains).
- The HMI, alarms, and documentation are in order.
- The machine builds good parts (using production-like samples).
What FAT can’t verify: The customer’s actual air pressure, foundation, parts, and operators. Those are SAT territory.
SAT (Site Acceptance Test)
SAT is the test at the customer’s plant, after installation. The machine is run in its real environment with real production parts and operators.
What SAT verifies:
- The machine runs on the customer’s utilities (air, power, water).
- The machine works on the customer’s actual parts (not sample parts).
- The operators can run the machine (training, HMI usability).
- The throughput holds over a longer run (not just an hour).
- The safety systems work in the installed layout.
| Test | Where | When | Verifies |
|---|---|---|---|
| FAT | Builder’s shop | Before shipment | Mechanical function, cycle time, safety, documentation |
| SAT | Customer’s plant | After installation | Real utilities, real parts, operators, sustained throughput |
What FAT Actually Tests (and What It Misses)
FAT is the builder’s last chance to catch problems before the machine leaves the shop. A thorough FAT tests more than just “it runs.”
Cycle Time and Throughput
Run the machine at the production rate for a sustained period (4–8 hours). Measure the actual cycle time, not just the theoretical. If the target is 6 seconds per cycle, time 100 cycles and calculate the average. Does it hold? Do the stations drift (sensor timing changes, accumulators fill)?
Most builders run the machine for 30 minutes, see it work, and call it good. The problems show up after hours of running — heat soak, oil warm-up, sensor drift. Run the sustained test.
Bad Parts and Fault Recovery
Test what happens when things go wrong:
- Part misfeed: Push a part in crooked. Does the machine detect it and stop? Does it recover automatically or need operator intervention?
- Sensor timeout: Block a sensor. Does the machine fault safely? Does the alarm on the HMI make sense?
- Missing part: Skip a station. Does the downstream station detect the missing part?
FAT often tests only the happy path. The fault recovery is where machines fall apart on the floor. Test the exceptions in FAT, not at the customer’s plant.
Safety Systems
Press the e-stop. Does it stop in time? Open the interlocked gate. Does the machine stop? Trip the light curtain. Does the cycle halt? These are mandatory tests, not optional.
Documentation
FAT is when the customer checks that the documentation is complete:
- Electrical schematics (as-built, not the preliminary).
- PLC program (commented, backed up).
- pneumatic and mechanical drawings.
- Operator manual (in the customer’s language, with photos).
- Spare parts list (recommended spares, part numbers).
- Safety file (risk assessment, CE documentation if applicable).
Missing documentation at FAT is easier to fix than missing documentation after the machine ships. Don’t let the customer sign off with “we’ll send the manuals later.”
The FAT rule: If the customer signs off FAT, they’re accepting the machine as-built. Anything that shows up later (beyond warranty) is a change order. Be thorough at FAT. The problem you find at the shop costs hours. The same problem at the customer’s plant costs days plus travel.
What SAT Covers (The Real-World Test)
SAT is where the machine meets reality. The customer’s utilities, parts, and operators are different from the builder’s shop.
Utilities Check
Before running the machine, verify the customer’s utilities match the requirements:
- Air pressure and quality: Is the pressure what the machine needs? Is the air dry and filtered? (A builder’s shop has clean, dry air. The customer’s plant may have wet, oily air.)
- Electrical supply: Voltage, phase, grounding. A machine that runs on 220V 3-phase won’t work on 208V without adjustment.
- Floor condition: Is the floor level? Is it strong enough for the machine’s weight? A wobbly foundation causes precision problems.
Real Parts vs. Sample Parts
FAT used sample parts (the customer sent a few). SAT uses production parts — actual parts off the line. These may have dimensional variation that the samples didn’t. The machine must handle the real variation, not just the perfect samples.
If the machine jams on a real part that worked on a sample, that’s a SAT finding. The builder fixes the tooling or adjusts the program on site.
Operator Training
SAT is when the customer’s operators learn the machine. The builder’s technician trains them:
- Normal operation (start, run, stop).
- Changeover (how to switch between part variants).
- Fault recovery (what to do when the alarm shows).
- Preventive maintenance (daily/weekly tasks, lubrication points).
Operators who can’t recover from a simple fault will call the builder every shift. Train them during SAT, not after the builder leaves.
The Acceptance Criteria: Written Before FAT
The worst FAT is the one where nobody knows what “passing” means. The acceptance criteria must be written in the contract before the machine is built.
What the Criteria Should Specify
- Cycle time: Maximum seconds per cycle, sustained for X hours.
- Yield: Percentage of good parts (e.g., ≥98%).
- Up time: Machine running at least X% of the test period.
- Safety: All e-stop and guard functions verified.
- Documentation: Listed above.
Without written criteria, the customer’s engineer at FAT can say “it doesn’t feel fast enough” and you’re arguing about opinions. With written criteria, you measure against numbers. Agree on the numbers in the contract.
Common FAT/SAT Pitfalls
1. Testing on Perfect Parts Only
The builder tests on the best samples. The customer runs on real parts (with variation). The machine that works on perfect parts jams on the first real part. Test with parts that have the full expected variation range.
2. Not Running Long Enough
A 30-minute test doesn’t show heat soak, oil warm-up, or sensor drift. Run the sustained test (4–8 hours). The problems that show up after an hour are the ones that hurt on the floor.
3. Skipping Fault Recovery
Testing only the happy path. The machine runs fine until the first jam — then nobody knows how to recover. Test the faults in FAT.
4. Documentation As an Afterthought
“We’ll send the manuals later” becomes manuals that never arrive. The customer can’t maintain the machine without documentation. Verify documentation at FAT.
5. SAT Without a Technician
Shipping the machine and saying “good luck” means every small problem becomes a service call. The builder’s technician should be on site for SAT — install, train, and fix issues while they’re still fresh.
A FAT/SAT Checklist
- Are the acceptance criteria written in the contract (cycle time, yield, uptime)?
- Has the machine run a sustained test (4–8 hours) at FAT?
- Have fault recovery scenarios been tested (misfeed, sensor timeout, missing part)?
- Has the safety system been verified (e-stop, guards, light curtains)?
- Is the documentation complete (schematics, PLC program, manuals, spares)?
- At SAT: have the customer’s utilities been checked (air, power, floor)?
- Has the machine run on production parts (not just samples)?
- Have the operators been trained (operation, changeover, fault recovery)?
- Is there a punch list of items to fix? (Track to closure.)
- Who signs off FAT and SAT? (Both parties, with the punch list attached.)
- What is the warranty period and what does it cover?
- Is a builder technician on site for SAT?
The Bottom Line
Custom machine delivery and acceptance isn’t shipping the box and waiting for the check. FAT is where you test the machine in your shop — sustained run, fault recovery, safety, documentation. SAT is where it meets the customer’s real utilities, parts, and operators. The acceptance criteria must be written before the machine is built, so “passing” is a number, not an opinion. The machine that passed FAT in the builder’s clean, ideal shop but failed SAT on the customer’s wet air and real parts wasn’t a bad machine — it was a FAT that tested the wrong things. Test the exceptions, run the sustained cycle, verify the documentation, and have a technician on site for SAT. The handover that goes smoothly isn’t luck. It’s a FAT that actually tested what the customer will live with.
