A factory acceptance test (FAT) and a site acceptance test (SAT) on a filtre-presse entièrement automatique only mean something if the two events are written against a single list of requirements, each one assigned to the place where it can actually be demonstrated. The factory can show that the press is built as specified, that the automatic sequence runs, that interlocks and fault responses behave, and that interfaces respond to signals. Feed-specific dewatering results — cake condition, filtrate quality and throughput on the plant’s real slurry — depend on the installed feed system and the actual sludge, so they belong at site unless the factory can run representative test material under documented process conditions. The deciding conditions are concrete: what feed material is available at each location, which connected equipment exists at each stage, and which requirements were written into the purchase order.
Agree what FAT and SAT must demonstrate
Settle the content of both events before either one is scheduled: the list of requirements, who performs and who witnesses each test, the test cases, the evidence that counts as proof, and the criteria that close each line out. An acceptance event without that agreement turns into a demonstration of whatever the press happens to do that day, and the buyer ends up signing a document that records attendance rather than conformity.
Split the requirement list into two families from the start. One family covers automation and function: plate shifting, closing and pressure holding, valve sequencing, filtrate and core blow steps, alarm and fault handling, operator interface behavior, and the response to loss of air, power or an external permissive. The other covers process duty: cake moisture, filtrate clarity, solids throughput per cycle or per shift, and cycle time at the agreed feed. These two families fail for different reasons and are proven by different evidence, so mixing them in one pass/fail line hides which one actually passed.
IEC 62381:2024 covers factory acceptance testing, site acceptance testing and site integration testing for process-industry automation, structured around agreed scope, activities and responsibilities, and tested against the applicable specification. That structure is useful for the automation family, but it supplies no cake-moisture, throughput or filtrate-quality target for a filter press — those numbers come from the project agreement and nowhere else. If a specification line says “low moisture cake” without a measurement method and a feed condition, it is not testable, and that is worth fixing before fabrication rather than at the witness event.
Contracts sometimes combine the two events, for example where a skid-mounted automatic filter press is tested with site personnel present and then re-verified after installation. That is workable provided each claim stays attached to its own evidence: what was observed with which feed, under which utility conditions, with which connected equipment present.
Separate factory demonstrations from site-only questions
| Acceptance question | What factory testing can establish | What may still need site testing |
|---|---|---|
| Specified assembly, automatic sequence and fault responses | Demonstrations within the agreed factory scope, including checks using simulated signals or substitute feed. | Requirements that could not be demonstrated remain open, with recorded evidence, responsibility and a retest condition. |
| Cake condition, filtrate quality and throughput | Representative factory test material can support only the documented process conditions; simulated signals alone do not establish dewatering performance. | Actual-sludge questions not represented at the factory, tested against agreed feed conditions, methods and pass criteria. |
| Connected equipment and installed operation | Specified interfaces and responses represented by the factory setup. | Actual tank, pump or conveyor connections, utility conditions, operator controls and normal interruption paths in a coordinated sequence. |
Treat the factory event as proof of construction and behavior, not proof of dewatering. Most factory floors can supply compressed air, power, water and a hydraulic power pack, but not the plant’s actual ceramic body slurry, glaze waste or mineral tailings at the right solids concentration and temperature. Where the factory runs simulated signals or a substitute feed such as clean water or a bench-prepared slurry, the event verifies the specified assembly, the automatic sequence, the interfaces and the fault responses, and the remaining questions about actual sludge and installed connections move to the site list by name.
That boundary has a physical reason. A recessed-plate press builds cake by pumping slurry into closed chambers against the resistance of the filter cloth and the accumulating cake itself, so feed solids concentration, particle size distribution, compressibility and any conditioning chemistry determine filtration rate and the cake that forms. The EPA’s fact sheet on recessed-plate filter presses describes exactly these interacting elements — feed pump and plate-cloth interfaces, filtrate removal, cake discharge and the need for representative testing — in the context of municipal biosolids; the mechanisms carry over to industrial slurries, while the numbers do not. Clean water through a press proves the hydraulic circuit and the sealing, and it tells you nothing about how a glaze filtrate will look or whether a mineral cake will release from the cloth.
Representative factory test material is a real option and worth asking for when the plant can ship feed. When it is used, the result supports only the process conditions that were documented at the time: the solids concentration, the conditioning dose, the feed pressure profile and the cycle length actually run. Change the feed case at site and the factory cake result no longer stands as evidence for the new case.
The practical discipline is to write the factory report so that a requirement either carries observed evidence or carries an open flag with a named responsible party and a retest condition. An unmarked blank is the thing that causes trouble six months later, because nobody can tell whether the test passed, was waived or was simply forgotten.
Define representative dewatering performance tests
If cake moisture, filtrate quality or throughput are acceptance items, define the whole test — feed, sampling, measurement, operating state, cycle boundaries and pass criteria — as one specification rather than as three separate numbers. These three results are produced by the same cycle and trade against each other, so testing them independently invites a result that is technically true and practically meaningless.
Start with the feed condition, because it sets everything downstream. Record the solids concentration and how it was measured, the particle size character, the temperature, and any conditioning chemical, its dose and its point of addition. Agree whether the test feed is drawn from a mixing tank at steady state or from a settled sump, because a press fed from the bottom of a thickener behaves differently from one fed a diluted upper layer. The feed pump and its pressure ramp belong in the same description; a press that reaches full feed pressure slowly produces a different cake than one that reaches it quickly, even on identical slurry.
Then fix the mechanics of measurement. Name the sampling points — which plates, how many chambers, whether samples are taken from the cake face or the core — and the method used to determine moisture, since cake moisture varies across a pack and a single sample from a favorable chamber says little about the solids actually processed. Do the same for filtrate: sample from the common discharge rather than one port, and state the measurement, whether that is suspended solids, turbidity or clarity against a defined reference, because the filtrate requirement is usually driven by what the plant intends to do with the water. Reuse water for pump seals and floor washdown is a different target from reuse in body preparation.
Define the operating state and the cycle boundaries explicitly. Throughput means nothing without agreement on where the cycle starts and ends: whether cake discharge, cloth wash and refill time count, how many consecutive cycles are averaged, and whether the press must be at thermal and hydraulic steady state before measurement begins. One cycle from a cold press after a long idle period is not comparable to the fourth cycle of a continuous run. Numeric pass criteria themselves come from the project agreement and the applicable measurement requirements, not from a general expectation about what presses achieve — a supplier who names a cake moisture figure without naming the feed case has described a hope, not a test.
Test the installed press and connected equipment
Site testing exists because the press at site is one element of a train, and the train fails in places the press alone cannot reveal. Run the installed interfaces, utility conditions, operator controls and the normal interruption paths as one coordinated sequence, with the feed tank, feed pump, filtrate return, cake conveyor or bin, air supply and control network all live.
The failures this catches are ordinary. A press that passed every factory check can sit waiting because the feed pump cannot reach filtration pressure through the installed pipe run, or because the upstream tank cannot hold enough slurry to fill the pack in one go. It can overflow a receiving boundary when the cake bin fills faster than the plant expected, or when the filtrate return line backs up into a tank already at level. It can stop and refuse to restart because an external signal — a conveyor running permissive, a tank level, a plant air pressure switch — was never wired or was wired with inverted logic. None of these are defects in the press; all of them are acceptance failures of the installed system.
Include the interruption paths that occur in normal operation, not just the fault cases. What happens when an operator pauses mid-filtration, when feed is lost partway through a cycle, when the downstream conveyor stops, when air pressure dips and recovers. Each should have a defined behavior and a defined way of resuming, and site is where that behavior gets observed against the control description.
Utility conditions deserve their own measurements at site: available air pressure and flow at the press during a plate shift rather than at the compressor, supply voltage under load, water pressure for cloth washing, and ambient conditions where they affect hydraulics. Where a third-party interface cannot be tested — a plant historian link, a supervisory system not yet commissioned, a downstream unit still under construction — leave it explicitly open with a named completion action and owner rather than closing the line because the press end of the connection looked correct.
Record failed tests and unfinished obligations
A test that fails or cannot be performed gets recorded in full: the requirement as written, the actual observed result, the evidence collected, the party responsible for resolution, and the condition under which it will be retested. Describing the event as complete while a line sits unresolved is the single documentation habit that costs buyers the most, because it erases the difference between a defect that was fixed and a promise that was never checked.
Commissioning documentation practice, as set out in the WBDG guidance for building systems, links requirements, responsibilities, schedules, checklists and acceptance into one traceable set — the framework is for building commissioning rather than industrial procurement, but the traceability principle transfers directly. Applied to a press, it means the buyer can later open the record and see, for each requirement, whether it passed, failed and was corrected, was deferred to a later event, or fell outside the agreed scope. Those four outcomes have different consequences for payment, warranty and plant readiness, and they cannot be reconstructed later from memory.
Separate retest conditions from retest dates. “Retest after cloth change” and “retest on 12 March” are not the same obligation; the first ties the test to a physical precondition that may or may not have been met by the date. For process items, the retest condition usually names the feed case, because a cake moisture line can only be retested against the slurry it was written for.
Where the project chooses to accept the press conditionally and start production with items outstanding, that is a contractual decision made by whoever the contract names as the decision authority, supported by an explicit list of retained obligations with owners and completion conditions. The acceptance record should show the decision and its basis rather than absorbing the open items into a general sign-off.
Build a requirement-by-requirement acceptance record
Build the record as one line per requirement, carrying what was demonstrated, where, with what evidence, and what remains. A line such as “plate shifting: factory sequence observed on simulated signal; actual cake release: site trial pending, 18 percent solids feed case” tells the site team exactly what it inherits, while a single tick against “automatic operation” tells it nothing. Each line should also carry a revision number, the feed case it applies to, and sign-off fields for the parties the contract names.
The reason for this granularity is specific to presses. Automatic plate shifting is a motion test — the shifter indexes plates at the specified rate without jamming — and it can pass perfectly while cake release fails, because release depends on cloth condition, cake cohesion and the geometry of the discharge, none of which a dry-cycle motion test touches. The same split applies elsewhere: closing and pressure holding is not the same requirement as sealing against a particular slurry at filtration pressure; valve sequencing is not the same as filtrate flow behavior; cloth washing motion is not the same as recovered permeability. Writing them as separate lines prevents the easy test from standing in for the hard one.
Keep the revision field honest. When a feed case changes, when conditioning chemistry is substituted, or when a connected unit is redesigned, the affected lines revert to open rather than carrying forward a result obtained under different conditions. A record that only ever gains ticks is a record that has stopped tracking the project. Carry the closed-out record forward into the operator-training and handover checks that follow acceptance, since the open items define what the operating team must watch during the first production weeks.
Before approaching a supplier about acceptance scope, have the specific press configuration identified — plate size and count, chamber depth, membrane or recessed, shifter and cloth-wash arrangement — along with the feed characterization (solids concentration, particle character, any conditioning), the target cake and filtrate requirements with their measurement methods, the connected equipment at the press battery limits, and the utility conditions available at site. With those in hand, the test list can be assigned to the factory or the site on evidence rather than on convenience, and the acceptance record will reflect what the press actually did.
Questions fréquemment posées
Q : Will a successful test on one sludge cover all the feeds we plan to process?
A : It demonstrates only the documented test conditions. Identify materially different planned feed cases in the acceptance scope and specify how each will be tested or kept outside the demonstrated results. Record the feed case, operating state and cycle boundaries with each result.
Q : Can we trade lower throughput for a drier cake during acceptance?
A : Only if the project agreement permits that trade-off. Judge cake condition, filtrate quality and throughput together under the agreed feed and cycle conditions; a favorable cake sample does not establish an overall pass. Record any missed requirement separately, and use the contract’s decision authority for changes to acceptance terms.
Q : What can we complete if the receiving conveyor is not ready for the site test?
A : Complete and document only the agreed checks that can be demonstrated without that connection. Keep the untested conveyor interface and coordinated operating sequence open, with a responsible party and completion action. A successful press-only check does not close the missing interface test.
Q : Can we sign an acceptance record while some tests remain unfinished?
A : Only under the contract’s provisions for conditional acceptance and its designated decision authority. Separate passed, failed and unperformed requirements, and record the retained obligations and their completion or retest conditions. The sign-off should make those open items explicit rather than imply that every requirement has been met.
