Back to all posts
Control Systems7 min read·October 25, 2026

How Should Vendors Get Remote Access to Plant Controls?

TB
ThynkBlox Automation Team
Automation Engineering

For secure vendor remote access to PLC and SCADA systems, require named accounts, site approval and a time-limited connection through a controlled entry point. Restrict each session to the equipment and work agreed in advance, keep a session record and give the site a tested way to revoke access. Treat permission to diagnose a fault separately from permission to change operating equipment.

What should the site approve before a vendor connects?

Start with a specific support request. “Help with the line” is too broad to define an access boundary or decide when the work is finished. For the wider service terms, see what a plant automation support agreement should cover.

Use a short approval record that answers:

  • Who is the named vendor engineer?
  • Which site employee owns the request?
  • Which systems may the engineer reach?
  • Is the work observation, diagnosis or an approved change?
  • When does access start and expire?
  • Who will supervise and close the session?
  • What should happen if the connection fails?

Have the site owner confirm the operating condition before access starts. Record whether equipment is running, stopped or released for maintenance, along with any restrictions.

Approval should apply to that request and access window. A support contract should not serve as standing permission to connect whenever the vendor chooses.

What should the controlled entry point do?

Specify a site-managed route that checks identity and approval before allowing access to selected plant systems. Depending on the design, this could include a remote access gateway and a dedicated support workstation.

Require the proposed arrangement to demonstrate these controls before accepting it:

ControlAcceptance check
Named identityEach engineer signs in using an individual account
Additional identity checkMulti-factor authentication is required at the entry point
Restricted destinationsThe account reaches only the approved systems
Restricted toolsAvailable tools match the approved support task
Session visibilityThe site can identify active connections
ExpiryAccess ends at the agreed limit
RevocationAn authorised site employee can terminate access

Include every connection route in the design review, including vendor appliances, cellular connections and unattended support software. Require them to follow the same approval and closure rules.

Do not accept a product demonstration as proof of the installed arrangement. Test the actual route from an external connection to the approved destination.

How should named accounts and permissions work?

Give each engineer an individual identity. Avoid a shared “vendor” login passed between support staff. Require a new approval when a different engineer takes over.

Separate the account used to enter the site from permissions inside the destination system. Approving a connection should not automatically grant controller programming, SCADA administration or access to unrelated equipment.

Start with the narrowest permissions needed for the request. For observation, specify viewing rights and identify any functions that must remain unavailable. Verify that the installed software enforces those limits.

Where an older application cannot provide individual accounts or suitable restrictions, document the limitation. Define compensating controls, such as supervised use of a dedicated workstation, restricted destinations and separate approval before opening engineering tools. Have the site explicitly accept the remaining risk before enabling access.

Why must support approval and change approval stay separate?

Make diagnosis the default scope unless the site has approved a defined change.

For example, permission to inspect an alarm history should not include permission to download controller logic, alter a setpoint, acknowledge alarms, force an input or output, or restart equipment. List permitted and prohibited actions in the request.

When diagnosis identifies a proposed change, pause and obtain approval from the person responsible for that equipment. The change record should describe:

  • What will change and why.
  • Which equipment and operating functions are affected.
  • The required plant condition and local attendance.
  • The backup and recovery arrangements.
  • The checks required before returning equipment to service.

Keep these responsibilities in the functional design specification. For changes requiring planned downtime, connect the support procedure to the PLC retrofit shutdown plan.

Arrange any safety-related assessment and design work with a qualified assessor.

What should happen during a support session?

Assign a site contact who can see the equipment’s condition, communicate with the vendor and stop the work. Agree which decisions that person can make and which require escalation.

Use a brief opening check: confirm identities, the support request, permitted destinations, operating restrictions and the access expiry. Confirm that the expected recording and logging controls are active.

Keep a session record covering:

  • The engineer and approving site employee.
  • Connection start, finish and destination.
  • The support request and approved scope.
  • Actions taken, files transferred and changes proposed.
  • Any additional approval granted during the session.
  • The outcome and outstanding work.

Where screen or command recording is required, define what it captures, who may review it and how long it is retained. Set access and retention rules that account for sensitive production information.

Do not treat a recording as permission to act. Keep the approval record and session evidence together.

How should access expire and be revoked?

Set an expiry when approving the request. Require an explicit extension if the work runs beyond it.

At closure, end the active session and remove the temporary permissions used for the job. Decide separately whether the named account should remain disabled for future support or be removed entirely.

Give an authorised site employee a clear revocation procedure. Include both blocking new connections and terminating existing ones. Test these separately: disabling an account is not sufficient evidence that every active connection has ended.

Also check any temporary credentials, transferred files and alternative access routes created during support.

Plan for an interrupted session before work begins. Specify how the local team will assess equipment status and handle an incomplete change. Treat connection termination and equipment shutdown as separate decisions; neither should substitute for an agreed local procedure.

What does an approved support request look like?

This illustrative example concerns a packaging line with repeated SCADA communication alarms.

The maintenance lead requests remote diagnosis. The approval names the vendor engineer, permits access to the SCADA support workstation and expires at the end of the agreed support window. The permitted work is viewing diagnostics and collecting an agreed log file.

Controller downloads, alarm configuration edits and equipment commands are excluded. A site technician attends locally.

During diagnosis, the vendor proposes changing a communication setting. The maintenance lead pauses the work and requests a separate change approval. That approval defines the affected connection, the required plant condition, the recovery method and the checks needed afterwards.

Following verification, the site closes the session, removes temporary access and records the result. The diagnosis request and change approval remain distinct records.

What common mistakes should plants avoid?

Leaving access available between support jobs. Make each connection depend on a current request, with expiry and closure checks.

Approving the vendor company instead of the person. Name the engineer and repeat the identity check when support changes hands.

Allowing diagnosis to expand into modification. Require a deliberate pause when the proposed action crosses the approved boundary.

Collecting logs without reviewing closure. Assign someone to confirm what happened, whether changes were made and whether access ended.

Writing restrictions that the system cannot enforce. Test permissions against the installed tools and document limitations.

Creating an undefined emergency exception. Prepare an emergency route with named approvers, a restricted scope and a required review after the event.

What should the site agree with its support provider?

Before enabling remote support, agree ownership of accounts, approval, supervision, records and revocation. Include how engineer substitutions, urgent requests and failed connections will be handled.

Use the automation services overview to frame the support scope, then document the site’s actual responsibilities and acceptance checks. The useful deliverable is an arrangement the maintenance team can operate and verify during a real support call.

Frequently asked questions

Can a vendor keep a permanent account?

A named account may remain available for future requests, but set it to disabled between approved sessions. Define who enables it, which permissions apply and when it expires.

Is a secure connection enough?

Make connection security one acceptance item. Also require identity checks, site approval, restricted destinations, session evidence and tested revocation.

Can the vendor change a PLC while production is running?

Do not grant blanket permission. Require a specific change assessment, the appropriate site approval and an agreed operating condition. Arrange safety-related work with a qualified assessor.

What if the plant cannot record the screen?

Document that limitation before granting access. Agree the evidence required for the task, such as connection logs, a supervised action record and retained change files. Have the site decide whether that evidence is sufficient.

Next step

Want to apply this to your plant?

Every project starts with a fixed-fee assessment, and you can stop after the report.