Operational technology cannot be patched on an IT schedule, yet unpatched control systems stay exploitable for years. A Windows update that is routine on a laptop can stop an HMI talking to its PLCs, void a vendor support agreement or trip a process that takes days to restart. This guide covers how to run patching for industrial automation and control systems (IACS): the IEC TR 62443-2-3 model of supplier and asset owner duties, vendor qualification, maintenance groups, outage planning, compensating controls for assets that cannot be patched, and how to prove it is working. It is written for OT engineers, plant security leads and IT patch teams taking on industrial sites, and assumes the IT basics in the Patch Management Guide.
Key takeaways
- Qualification comes before deployment. Install only patches the automation supplier has tested against your product versions, or that you have tested yourself on a representative system.
- Group assets, do not list exceptions. NIST SP 800-40 Rev. 4 maintenance groups give every OT asset a maintenance plan, including the ones that will never receive a patch.
- Outage windows set the pace. Tie Level 0 to 2 changes to planned shutdowns and turnarounds, and patch Level 3 and DMZ systems far more often.
- Mitigation is a legitimate answer. Where a patch must wait, a dated, approved mitigation plan (segmentation, disabled services, removed remote access) is what NERC CIP-007-6 and good practice expect.
- Verify and keep a way back. Back up PLC projects and images before every change, check versions afterwards, and measure time to mitigate by asset criticality rather than percentage patched.
Why OT patching is different
OT patch management deals with the same flaws as IT, but under different constraints. The table sets out where the differences come from and what they change in practice.
| Constraint | IT norm | OT reality | Practical consequence |
|---|---|---|---|
| Availability and safety | Reboot overnight, roll back if needed | A reboot may stop a continuous process or a safety function | Changes need operations sign-off and a planned window |
| Supplier support | Vendor ships, you deploy | The automation supplier supports only qualified OS and firmware combinations | Wait for, or demand, compatibility statements |
| Asset life | 3 to 5 years | 15 to 25 years for controllers, drives and safety systems | Large unsupported estates need long-term mitigation |
| Patch delivery | Agent or management tool | Firmware loaded through engineering software, often on site | Patching is a maintenance job, not a push |
| Testing | Pilot ring of users | Few spares, test benches rarely match production | Representative test systems and staged rollout by line or unit |
| Regulation | Contractual SLAs | Sector rules such as NERC CIP set patch evaluation cycles | Evidence of evaluation and mitigation plans matters as much as installs |
NIST SP 800-82 Rev. 3 still recommends deploying security patches as quickly as possible, but only after testing them under field conditions, on a test system where one exists. The goal is speed with evidence, not indefinite delay.
The IEC TR 62443-2-3 model: supplier and asset owner duties
IEC TR 62443-2-3:2015 (Patch management in the IACS environment) is a Technical Report, not a certifiable standard. It sets out requirements for both asset owners and IACS product suppliers running a patch management programme, and defines a common format for exchanging security patch information between them. It also describes what suppliers do to produce patch information and what asset owners do to deploy and install patches. The format was written for security patches but can be used for other updates. For how this part fits the rest of the series, see the IEC 62443 Guide.
- Supplier duties: monitor third-party components (operating system, database, runtime libraries) for vulnerabilities, test vendor patches against each supported product version, and publish whether each patch is approved, not applicable or not compatible.
- Asset owner duties: keep an inventory accurate enough to match patch information to installed versions, decide applicability, schedule installation, test, deploy and record results.
- The shared artefact: a patch compatibility statement per product release. Ask for it in machine-readable form; it is the input that lets you automate applicability checks. See the CSAF Security Advisories Guide.
- Integrators and service providers: IEC 62443-2-4 covers the service provider side, including who applies patches under a maintenance contract. Make that responsibility explicit in contracts.
Maintenance groups and plans
NIST SP 800-40 Rev. 4 recommends grouping assets with similar maintenance needs and writing a maintenance plan per group and scenario, rather than treating hard-to-patch assets as exceptions. Its own examples include legacy OT assets with no patches available, protected by network isolation and physical security, and OT or IoT assets with life-safety impact. A typical plant grouping looks like this.
| Maintenance group | Examples | Routine plan | Emergency plan |
|---|---|---|---|
| OT DMZ and Level 3.5 | Jump hosts, patch relay, remote access gateways, historian replicas | Monthly, close to IT timing | Within days for KEV-listed flaws |
| Level 3 site operations | Historians, MES servers, domain controllers, backup servers | Monthly or quarterly after supplier qualification | Mitigate in hours, patch in the next agreed window |
| Level 2 supervisory | HMIs, SCADA servers, engineering workstations | Quarterly or at planned outages, qualified patches only | Isolate or restrict, patch at the earliest safe window |
| Level 1 controllers | PLCs, RTUs, drives, protection relays | Firmware at turnarounds when a fix applies | Mode switch protection, block engineering protocols, monitor |
| Safety instrumented systems | SIS logic solvers | Only with safety case review and functional test | Compensating controls; change under the safety management process |
| Unsupported legacy | Windows XP or 7 HMIs, end-of-life PLCs | No patches: documented long-term mitigation and replacement plan | Tighten isolation, remove remaining connectivity |
SP 800-40 Rev. 4 also defines four risk responses: accept, mitigate (patch, disable the feature or add controls such as segmentation), transfer and avoid (remove the software or capability). Patching and upgrading are the only responses that remove a vulnerability without removing functionality, which is why they stay the default even in OT. For the zone model behind the groups, see the OT Network Segmentation, Purdue Model and IEC 62443 Zones and Conduits Guide.
Qualifying patches
Supplier qualification
- Read the supplier's compatibility list first: major automation suppliers publish which Microsoft monthly updates and antivirus definitions they have tested with each release of their SCADA, DCS and engineering software. A patch absent from the list is not approved, and installing it may put you outside support.
- Track product version, not product name: approval is usually specific to a release and service pack. Your inventory must hold the exact version of each OT application and controller firmware.
- Ask for timelines in contracts: require the supplier to state its qualification turnaround for critical and known exploited vulnerabilities. See the OT Product Procurement Guide.
Your own testing
- Representative test system: the same OS build, application versions, communication drivers and, where possible, a spare controller of the same model and firmware. A virtual copy of Level 2 and 3 servers is enough for many updates.
- Test cases that matter to operations: HMI to PLC communication, alarm and event flow, historian collection, time synchronisation, redundancy switchover, licence servers and backup jobs, as well as the obvious "does it boot".
- Soak time: run the patched test system for a defined period (often one to two weeks for Level 2) watching for memory leaks, driver faults and timing changes.
- Firmware: confirm the new controller firmware supports the project file version and any function blocks in use, and that the engineering software version on the workstation can still go online to it.
Scheduling, deploying and rolling back
- Collect and match: pull supplier advisories and OS updates into one queue, matched against the OT inventory by product and version. See the ICS Advisory Triage and OT Vulnerability Management Guide for triage.
- Decide the response per maintenance group: patch in the next window, patch urgently, mitigate now and patch later, or accept with a dated review. Record the decision and the approver.
- Stage media safely: download from the vendor, verify hashes or signatures, scan, and move updates into OT through a patch relay in the DMZ (for example a WSUS downstream server or a dedicated repository). Avoid direct internet access from Level 2 and below, and control removable media.
- Back up before change: image servers and workstations, export PLC and DCS project files and configuration, and confirm backups restore. See the OT Backup, Recovery and Golden Image Guide.
- Deploy in rings: start with a redundant or standby node, then one line, unit or cell, then the rest. For redundant servers patch the standby, switch over, observe, then patch the former primary.
- Verify: check installed KB numbers, application versions and firmware revisions against the target, and have operations confirm normal running before closing the change.
- Roll back if criteria are met: pre-agree what triggers a rollback (lost communications, alarm floods, failed switchover) and how: uninstall, restore image or reload the previous firmware and project.
Outage alignment: plants with annual or biennial turnarounds should hold a standing list of firmware and OS work for the next one, so that fixes released mid-cycle are not forgotten. Continuous processes that never stop rely on redundancy to patch one side at a time.
Compensating controls when a patch must wait
Many OT assets will run unpatched for months, and some for life. That is acceptable only when the exposure the vulnerability needs has been removed or watched. NERC CIP-007-6 R2 formalises this for bulk electric system cyber assets: evaluate security patches for applicability at least once every 35 calendar days, then within 35 calendar days of evaluation apply the patch, create a dated mitigation plan or revise an existing one, and implement the plan in its stated timeframe, with changes approved by the CIP Senior Manager or delegate. See the NERC CIP Guide.
| Vulnerability needs | Compensating control | Evidence it is in place |
|---|---|---|
| Network reach to a vulnerable service | Conduit firewall rule restricting the port to named engineering hosts | Rule export and hit counts |
| Internet exposure | Remove direct exposure; access only via brokered remote access | External attack surface scan showing no exposure |
| Remote vendor access | Session-based, approved, recorded access with MFA. See the OT Remote Access Guide | Access logs, approvals |
| Program or firmware download | Controller key switch in RUN, write protection, engineering protocol allowed only from engineering workstations | Mode audit, network monitoring alerts |
| Code execution on a Windows host | Application allowlisting, disabled unused services, removable media control | Allowlist policy in enforce mode, service baseline |
| Default or weak credentials | Change defaults, remove shared accounts | Credential audit |
The Unitronics Vision PLC and HMI flaw CVE-2023-6448 in the CISA KEV catalogue is a reminder that some "patches" are configuration: the vulnerability was an insecure default password, and the decisive mitigations were changing it and removing the devices from direct internet exposure. See the CISA KEV Catalogue Guide.
Where identity fits
Patch infrastructure in OT is a high-value path into the plant. The WSUS relay, the software repository and the accounts that push updates or load firmware can change every Level 2 and 3 host. Use dedicated, non-interactive service accounts per tool, keep them out of IT domains, store any credentials in a vault and rotate them after supplier or integrator staff leave. Firmware download rights on controllers should belong to named engineering accounts, not shared logins. See the Identity for OT Environments Guide and the Engineering Workstation Security Guide.
Detection and measurement
- Log: patch relay approvals and downloads, change tickets, controller mode changes and firmware download events (from passive OT network monitoring), new services or binaries on HMIs and servers, and backup job results.
- Watch for: firmware or program downloads outside approved windows, patch deployment to OT from an IT management server that should not reach it, and assets that revert to an older version after a restore.
SP 800-40 Rev. 4 warns that metrics which are too simple are not actionable and suggests a matrix of vulnerability importance against asset importance, showing the share remediated by deadline and median time. Useful OT metrics:
- Percentage of applicable advisories with a recorded decision (patch, mitigate, accept) within 35 days.
- Median time from supplier qualification to installation, by maintenance group.
- Number of KEV-listed vulnerabilities on OT assets with neither a patch nor a verified compensating control.
- Percentage of OT assets with an exact, current software and firmware version in inventory.
- Count of unsupported assets, and the share with an approved replacement date.
- Rollback rate and unplanned downtime caused by patching (should be low; a rising figure points to weak testing).
Framework mapping
| Framework | Reference | What it asks for |
|---|---|---|
| IEC TR 62443-2-3:2015 | Whole report | Supplier and asset owner patch activities, patch information exchange format |
| NIST SP 800-53 Rev. 5 | SI-2, SI-2(2), CM-3, CM-4, CM-8, RA-5 | Flaw remediation, automated status, change control, impact analysis, inventory, vulnerability monitoring |
| NIST CSF 2.0 | ID.AM-02, ID.RA-01, PR.PS-02, PR.PS-03 | Software inventory, identifying vulnerabilities, maintaining software and hardware commensurate with risk |
| NERC CIP | CIP-007-6 R2 | 35-day evaluation cycle, apply or mitigation plan |
| CIS Controls v8.1 | 7.3, 7.4, 7.7 | Operating system and application patch management, remediation |
| ISO/IEC 27001:2022 | A.8.8, A.8.32 | Management of technical vulnerabilities, change management |
| MITRE ATT&CK for ICS | T0819, T0866, T0883, T1693 | Exploit Public-Facing Application, Exploitation of Remote Services, Internet Accessible Device, Modify Firmware |
| MITRE D3FEND | D3-SU | Software Update |
Common failure patterns
- IT tooling pointed at OT: an enterprise patch server reaches HMIs and installs an unqualified update with an automatic reboot. Separate OT patch infrastructure and approvals.
- "Cannot patch" as a permanent status: exceptions with no owner, no compensating control and no review date. Use maintenance groups and dated mitigation plans.
- Inventory without versions: knowing there are 40 PLCs but not their firmware revisions makes advisory matching guesswork. See the OT Asset Visibility Guide.
- Patching the controllers while the doors stay open: effort goes into Level 1 firmware while the DMZ jump host, VPN and Level 3 Windows servers lag months behind. Most intrusions that disrupt operations start in the IT and DMZ layers: in the Colonial Pipeline ransomware attack the entry point was an unused VPN account without MFA, and the pipeline was shut down as a precaution though OT was not reported as directly compromised.
- No backup before firmware work: a failed update with no exported project leaves the line down while code is rebuilt.
- Forgetting the end of support date: Windows 10 IoT Enterprise LTSC 2021, common on new HMIs, reaches mainstream end on 13 January 2027 and extended end on 14 January 2032 on Microsoft's lifecycle page. Plan replacements years ahead, not when updates stop.
Practitioner checklist
- Record exact OS, application and firmware versions for every OT asset.
- Define maintenance groups by Purdue level and criticality, each with routine, emergency and unpatchable plans.
- Obtain supplier patch compatibility statements and contract for qualification timelines.
- Maintain a representative test system and agreed operational test cases.
- Stage updates through a dedicated OT patch relay with hash or signature verification.
- Back up images and controller projects before each change and test restores.
- Evaluate applicable advisories at least every 35 days and record a decision for each.
- Document compensating controls with evidence and a review date for every deferred patch.
- Protect patch service accounts and firmware download rights as privileged access.
- Report time to mitigate by asset criticality, KEV exposure and unsupported asset counts.
Standards and references
- IEC: TR 62443-2-3:2015 Patch management in the IACS environment
- NIST: SP 800-82 Rev. 3 Guide to Operational Technology (OT) Security
- NIST: SP 800-40 Rev. 4 Guide to Enterprise Patch Management Planning
- NERC: CIP-007-6 Cyber Security, System Security Management
- ISA: ISA/IEC 62443 Series of Standards
- CISA: Known Exploited Vulnerabilities Catalog
- NIST: SP 800-53 Rev. 5
- Microsoft Learn: Windows 10 IoT Enterprise LTSC 2021 lifecycle
Related NHI Mgmt Group resources: OT Security Guide · Patch Management Guide · IEC 62443 Guide · NIST SP 800-82 OT Security Guide · OT Incident Response Guide