Specifying automatic filter press PLC and HMI controls starts with the press sequence, not the controller brand: write down how the machine opens, closes, feeds, washes, blows, squeezes, discharges and returns to ready, which operating modes exist, what inputs and outputs each step needs, and how the press talks to the pumps, dosing and cake conveyors around it. Hardware follows from that. The conditions that change the answer are concrete ones you can look up: whether the press is a chamber or membrane design and therefore needs a squeeze step, whether feed is fixed-duty or varies by slurry case, whether cake transport and feed pumping sit under the same controller or another one, and whether the site already mandates a control platform. A PLC is the programmable controller that executes the sequence and interlocks; the HMI is the operator screen that shows what the press is doing and accepts permitted commands.
Define the press sequence before selecting control hardware
The functional specification comes first: the press sequence, the operating modes, the inputs and outputs, and the interfaces to connected equipment. A list of a controller family, screen size and network card tells a supplier nothing about what the machine must do, who is responsible for each function, or what will be demonstrated at acceptance. Those three gaps — required behavior, responsibility and acceptance checks — are what a functional specification closes.
Work through the sequence as the plates actually move. Closing and pressure holding, feed start conditions, the end-of-feed criterion the project uses, core blow or wash if fitted, the membrane squeeze step on a membrane press, opening, plate shifting and cake discharge, then the return to ready. Each step needs its permissives (the conditions that must be true before it may start), its end condition, and the signals it sets for other equipment. Name the operating modes alongside it: automatic cycling, semi-automatic step-by-step, and manual device operation for maintenance, with the different permissions each mode allows.
The squeeze step is the clearest example of why sequence comes before hardware. A membrane press adds a pressurized diaphragm stage after feed, which means an extra medium supply, extra pressure and timing variables, extra interlocks against opening under pressure, and extra recipe parameters. Whether that stage earns its place depends on the slurry, not on the controls; but once it is in the machine, it is in the sequence, the HMI and the acceptance test.
A site-mandated platform is a normal exception. Some plants fix the PLC family, the HMI brand and the network standard for spare-parts and maintenance reasons, and that constraint is legitimate. It settles the hardware question only. The sequence, permissives, recipe variables, interface signals and failure responses stay project-specific and still need to be written down and agreed.
Show the active step and useful fault information
When a cycle stops, the display should tell the operator four things: which step the press was in, which permissive failed, which device is affected, and what action is permitted next. A fault banner that reads only “cycle fault” or “press stopped” makes the operator reconstruct the sequence from the machine itself, usually while cake is sitting in a closed press and the feed pump is waiting.
Those four elements separate problems that look identical from the aisle. A press that has stopped because feed pressure never reached the start condition is in a different situation from one that stopped because a plate-shifter movement did not complete, and both differ again from a press holding because the cake conveyor downstream is not available to receive material. Each needs a different person and a different response, and the operator can only route the call correctly if the screen names the failed condition and the device behind it.
Specify this as display content, not as a general request for good diagnostics. For each step, state what the screen shows while the step runs and what it shows if the step does not complete: the step name, the permissive or device that blocked it, and the modes or commands that remain available. Alarm text that identifies the physical device by its plant tag is worth more than a numeric fault code the operator has to look up in a manual.
The display tells the operator what happened and what the control system will allow. It does not replace the recovery procedure. Exactly how a stopped cycle is cleared, whether a plate pack may be reopened, and what must be checked first follow the supplied machine design and the site’s approved procedure, and the person that procedure designates decides when the press returns to service.
Control recipe variables, permissions and restoration
A recipe is a named, stored set of adjustable filtration settings, and the specification needs five things for it: variable names, engineering units, permitted ranges, who may change each one, and a version identity. Add a defined way to restore the accepted recipe after someone has edited it. Without those, “adjustable parameters” is an open-ended promise with no way to audit what the press was actually running.
The variable names and units matter more than they look. An operator who sees a labeled time, pressure or volume value with its unit knows which feed case the setting belongs to; an unlabeled number on a settings page does not tell anyone whether it governs feed, squeeze or blow. Permitted ranges keep edits inside what the machine can physically do, and the limits must come from the offered press design and from representative process work on your slurry, not from a generic template. Ask the supplier which parameters are range-limited in the program and where those limits originate.
Version identity and restoration answer a question that comes up on almost every dewatering line: cake came out wetter this week than last, and nobody is sure whether the slurry changed or someone changed a setting. If the HMI shows which recipe version is loaded and the controller holds a protected copy of the accepted configuration, that question is answerable in a minute. Change privileges tied to named users or roles make it answerable with a name attached.
A fixed-duty press feeding one consistent slurry may not need multiple recipes at all. In that case specify one controlled parameter set with the same protections — named variables, units, ranges, restricted edit rights and a restorable accepted copy — rather than building a recipe management system no one will use. Which recipe is approved for production, and who approves a changed one, is decided by the plant’s own procedure and its designated approver, not by the control system.
Specify signals between the press and connected equipment
For every signal crossing between the press and a feed pump, dosing skid, conveyor or screw, define five things: what the signal means, which controller owns it, under what conditions it reads available, how each side behaves if communication is lost, and what states are visible at both ends. This applies to a fieldbus link and to a pair of dry contacts alike; a hardwired interface needs the same functional clarification, just with fewer wires to describe it.
The failure this prevents is a disagreement about the word “ready”. One controller may set a ready signal to mean the equipment is powered and not faulted; the other may read it as the equipment being able to accept material right now. The press then opens and discharges to a conveyor that is energized but already loaded, or holds a full chamber waiting for a dosing confirmation the dosing skid thinks it has already given. Two reasonable designs, one blocked line.
Write the signal definitions as plain statements of condition. A cake transport available signal, for instance, should state whether it means the conveyor is running, is running and empty at the discharge point, or merely has no active fault. A dosing ready signal should state whether chemical is confirmed present and at the specified rate, or only that the pump is enabled. Then state the loss-of-communication behavior for each: whether the press completes the current step, holds at a defined safe state, or does not start the next cycle, and what it displays while it waits.
Visibility at both ends is the part most often left out. If the press shows the conveyor as unavailable but the conveyor panel gives the operator no corresponding indication, the plant loses time proving which machine is actually blocking. Agree during specification which controller is responsible for raising the interface alarm and what each panel displays while the interface is in that state. For chemical dosing interfaces in particular, the signal definition and the dosing control strategy affect clarified water quality and sludge conditioning, so the process engineer responsible for dose rates should review those definitions before they are frozen.
Separate remote viewing, commands and software changes
Remote assistance is not one capability but three, and they should be specified separately: passive viewing of process and alarm data, active operating commands that change what the machine is doing, and software or configuration changes to the program, recipes or interlocks. UK HSE guidance for major-hazard control systems distinguishes passive remote monitoring from active operation and software-change modes on exactly this basis, because their effects on plant operation differ, and expects remote diagnostic access to be justified, risk-assessed, access-controlled, communicated to plant staff, protective of safety functions, and subject to change control.
NIST SP 800-82 Rev. 3, general US operational-technology security guidance, works on the same principle from the security side: least privilege applied separately to viewing, use and alteration roles, remote access that is managed and protected, commands that are authenticated, sessions that can be terminated, logs of what was done, and controlled handling of configuration changes with backup and recovery. Neither document fixes a network architecture or a product for your press; both say what the access arrangement has to control.
Turn that into contract language. For each of the three modes, name who is authorized, how the connection is established and by whom at the plant, whether plant staff must be present and informed while the session runs, how the session is terminated, and what record remains of what was viewed or altered. Software and recipe changes deserve the tightest treatment, since a changed parameter limit or interlock can persist long after the connection drops; the restorable accepted configuration is what lets the plant get back to a known state. Whether any remote mode is permitted at all on a particular press is a plant risk-assessment decision, taken against the site’s own procedure.
If the plant excludes remote connectivity, the specification does not simply go quiet. Document the local support arrangement instead: who holds the program and HMI project backups, in what format, where the licensed development software sits, what diagnostic information can be exported to media from the panel, and what the response arrangement is when the press is down and the supplier cannot see it.
Test the agreed control behavior and failure responses
| Acceptance condition | Agreed behavior to check | Information to observe |
|---|---|---|
| Normal sequence | The specified press sequence and operating modes | Actual sequence against the written cause-and-response list |
| Unavailable interface | The agreed response when connected equipment cannot provide a service or receive material | Interface availability, affected equipment and states visible at both ends |
| Interrupted cycle | The permitted recovery action for the supplied machine and site procedure | Active step, failed permissive, affected device and permitted next action |
| Recipe access | The specified variable names, units, permitted ranges and change privileges | Which settings the authorized user can change and the recipe version in use |
| Configuration recovery | Restoration of the accepted recipe or control configuration | Identity of the restored accepted configuration |
Acceptance testing proves control behavior against a written cause-and-response list, not screen navigation. Demonstrating that an operator can page through menus says nothing about what the press does when a permissive fails mid-cycle or when the cake conveyor drops offline. IEC 62381:2024 covers factory acceptance, site acceptance and site integration testing for process-industry automation against the applicable specification, with agreed scope, activities and responsibilities — which is the useful structural point here: the test is only as good as the specification it is run against, and the parties have to agree in advance who does what and what counts as a result.
So write the cause-and-response list first, during specification, and reuse it as the test record. Each line states a condition to create, the response the control system should give, and the information the tester should be able to observe on the screen while it happens. Behaviors worth covering include the normal sequence in each operating mode, an unavailable connected interface, a cycle interrupted part way, recipe access under each privilege level, and restoration of the accepted configuration. For each one, record what was observed and name it, so the result is identifiable later rather than a tick in a box.
Factory testing has a real boundary. In the supplier’s works, connected equipment is usually simulated: forced signals stand in for the feed pump, the dosing skid and the cake conveyor, and real slurry is absent. That is a legitimate way to test logic, but the factory report should list explicitly which signals were simulated and which conditions are deferred to site testing, so nothing falls between the two visits. Throughput, cake moisture and filtrate clarity are separate project-specific tests on real material; the control acceptance test shows that the sequence, interfaces and recovery behave as written. Whether the tested system is accepted is a decision for the project’s approved procedure and its designated approver.
Bring four things to the acceptance discussion with the supplier of the automatic filter press: the agreed sequence and operating-mode description, the list of interface signals with their meanings and owners, the recipe variable list with units and permitted ranges, and the cause-and-response list you expect to be run. With those in hand the test becomes a check against something, and the same documents carry straight into operator training and handover once the control behavior has passed.
Frequently Asked Questions
Q: What should I provide so controls quotations can be compared fairly?
A: Give each bidder the same description of the required press sequence, operating modes, connected equipment and control responsibilities. Include any site-mandated controller platform, the information operators need during a stopped cycle and the behaviors to demonstrate at acceptance. A hardware list alone does not define comparable control scope.
Q: Does a press handling one established feed need multiple recipes?
A: Not necessarily. A fixed-duty press may need one controlled parameter set. Define its variable names, units, permitted ranges, change privileges and accepted version, together with a way to restore it. If different feed cases require different settings, identify those cases before deciding how many recipes are useful.
Q: How can operators distinguish a press fault from a problem with connected equipment?
A: Specify fault information that identifies the active step, failed operating condition, affected device and permitted next action. For pumps, dosing or cake transport, also define availability signals and what each controller displays. Check the agreed response when a connected service is unavailable; recovery actions must follow the supplied machine design and site procedure.
Q: Can remote troubleshooting be specified without granting remote operating or software-change permissions?
A: Treat viewing, operating commands and software changes as separate permissions. State which activities are allowed, who authorizes access, how a session ends and how changes are recorded. If remote connectivity is excluded, define the local support and configuration-recovery arrangement instead.
Q: What remains to be checked if the factory test uses simulated plant signals?
A: Record exactly which signals and site conditions were simulated and which checks are deferred to the site. The remaining scope should address the actual equipment interfaces and agreed control responses. A successful controls demonstration does not establish throughput, cake condition or liquid quality; those need separate project-specific tests.
