What Should a PLC Do When Communication Fails?
A PLC should follow an agreed response for each failed communication interface: which operations may continue, which commands become invalid, what operators must see and what must happen before control resumes. Define that response for each equipment interface instead of relying on default settings. Include watchdogs, stale-command handling, local control and recovery; safety-related behaviour requires a qualified assessor.
Why should each interface have its own failure response?
Start with the work that depends on the connection. Losing a production dashboard should not automatically receive the same response as losing the feedback needed to control a machine. For a planned changeover, see when to roll back a control system cutover.
For every interface, ask: what decision becomes unreliable when this information stops arriving?
Then document the permitted operating state. Specify whether equipment should continue under defined conditions, complete a bounded action, enter a controlled stop or inhibit further operation. Treat these as design decisions to assess against the actual process, not universal recommendations.
Do not write only “stop on communication failure”. Identify what stops, in what sequence, which supporting functions remain available and who can restart it. Where the response affects safety, have a qualified assessor establish the required behaviour and validation.
What should the interface specification contain?
Create a communication-loss schedule alongside the control sequence. Give each connection its own entry, including connections between controllers.
| Interface detail | Decision to record |
|---|---|
| Connected equipment | Which controller and device exchange information |
| Control owner | Which system may issue commands in each operating mode |
| Required information | Which values, states and acknowledgements permit operation |
| Loss detection | What counts as missing, invalid or stale information |
| Detection timing | How long the agreed condition may persist |
| Failure response | What continues, stops, becomes inhibited or requires intervention |
| Operator indication | What the operator sees and what action is required |
| Recovery | What must be checked before commands are accepted again |
Record the behaviour at both ends. The controller's intended response and the device's configured response must form one agreed sequence.
Put this schedule in the functional design specification, where operations, maintenance and the controls team can review it together.
How should watchdogs detect a lost connection?
Define what each watchdog proves. Distinguish between evidence that a connection exists and evidence that the required application data is still being refreshed.
For example, specify a changing heartbeat or message sequence where appropriate. State which application updates it, how the receiver checks it and what happens if it stops changing. Do not accept an unchanged “healthy” flag as sufficient evidence of fresh data.
Choose the timeout from the process requirements and expected communication behaviour. Document why it is acceptable, then verify it under representative conditions. Avoid copying one timeout across unrelated interfaces.
Also define startup behaviour. The receiver should have an explicit state before the first valid exchange, with clear conditions for becoming ready. Include invalid data, missed acknowledgements and repeated interruptions in the design rather than treating every failure as a disconnected cable.
If the controller itself is running slowly, see how to troubleshoot high PLC scan time.
How should stale commands be handled?
Treat permission to communicate and permission to execute as separate decisions.
Specify the validity conditions for every remote command. These might include the current operating mode, the active control owner, a command identifier, an acknowledgement and an agreed expiry rule. Choose only the mechanisms the interface needs.
Make command handling explicit:
- Decide whether a command remains valid after communication is lost.
- Define how the receiver recognises a command it has already processed.
- State whether queued commands are discarded or reviewed after reconnection.
- Require a fresh request where an old start, reset or setpoint must not be reused.
- Define what happens when the sender receives no acknowledgement.
Avoid an instruction such as “retry until successful” without defining how repeated delivery is handled. Recovery should establish the current equipment state before another action is authorised.
For the handshake that stops a retried request running twice, see how to prevent duplicate MES commands to a PLC.
When should local control remain available?
Define local control as an operating mode with a clear owner, permitted actions and transfer conditions.
For each interface, decide whether local operation is allowed during communication loss. If it is, document which controls remain available, which process conditions must be satisfied and how remote commands are inhibited while local control owns the equipment.
Specify what the local operator can see. If a required indication is unavailable, decide whether that limits the permitted task.
Returning communication should not silently transfer ownership. Define whether the operator must select remote mode, acknowledge the transfer or make a fresh start request. Apply the same clarity when moving from remote to local control.
Any safety-related permissions, interlocks or mode changes require a qualified assessor. A local selector alone is not a complete operating procedure.
What would an equipment-level example look like?
Illustrative example: consider a transfer pump controlled by a PLC, with a remote drive and a supervisory operator screen. The following is a discussion template, not a recommended pump-control design.
For loss of the operator-screen connection, the specification might allow an already authorised transfer to continue under PLC control, provided all required process information remains valid. It would inhibit new remote requests and identify where operators can check the transfer.
For loss of the drive connection, the specification would separately define the drive's configured response, the PLC's response to unavailable feedback and the conditions for declaring the pump stopped. It should not substitute the last displayed status for confirmed equipment state.
If local operation is permitted, the procedure would define who takes ownership and which indications are required.
After reconnection, the sequence would compare the actual operating state with the intended state, discard expired requests and require the agreed restart action. A qualified assessor must determine any safety-related behaviour.
What should operators see during communication loss?
Write the indication around the affected function and required action.
Name the equipment, identify the unavailable connection and state the operating restriction. Where relevant, tell the operator whether an existing operation is continuing, whether new requests are inhibited and where local control is available.
Define how stale values appear. Require unavailable or outdated information to be visibly distinguished from current information, including in trends and summaries where operators use them to make decisions.
Separate restoration from readiness. “Communication restored” should describe the connection; “ready for remote operation” should depend on the recovery checks.
For deciding which events need an alarm and an operator response, see our guide to alarm management.
How should recovery be tested?
Test the specified behaviour through to restored operation, not just the initial alarm.
Use an authorised test plan with agreed precautions. Have a qualified assessor define the validation needed for safety-related functions.
Include these checks:
- Lose communication while idle and during an active sequence.
- Interrupt communication after a command is sent but before acknowledgement.
- Restore the connection while equipment remains under local control.
- Reconnect with an old command still present at the sender.
- Restart either endpoint and verify its initial state.
- Check repeated interruptions and unavailable required data.
- Confirm indications, command acceptance and restart conditions.
Record expected and observed results. Acceptance should depend on equipment behaviour and operator visibility, not merely on a healthy connection indicator.
What common mistakes should the review catch?
One response for every connection. Review each interface against the function it supports.
An alarm without a control response. State what equipment and command handling do while the fault remains active.
Automatic return to the previous command. Decide whether that command is still valid and whether a fresh request is required.
Local control without ownership rules. Define transfer, permissions and the return to remote operation.
Assuming a stop is always the correct response. Assess the process consequences and supporting functions. Safety-related decisions belong with a qualified assessor.
Related reading: buffering plant data during network outages and defining machine restart after power loss.
Frequently asked questions
Should every communication failure stop the PLC?
Specify the affected equipment's response rather than using controller shutdown as a blanket rule. Base the decision on which functions have lost the information they require.
Is a watchdog enough?
Use it as one part of the design. Also define data validity, command ownership, acknowledgements, operator indications and recovery.
Should equipment restart automatically when communication returns?
Make that an explicit, assessed requirement. Require valid state information and the agreed authorisation before accepting further commands.
Where should a plant start?
List the equipment interfaces and review one complete failure-and-recovery sequence at a time. Our automation services cover control-system assessment and integration planning.