Continuous process environments have inherent constraints that make applying standard OT (operational technology) cybersecurity practices difficult and potentially hazardous. A discrete manufacturer can most often stop a line and restart it at will, while continuous process sites rarely have that option. Typically across refineries, chemical manufacturing facilities, LNG plants, and pulp and paper mills, these continuous processes may run for months between a shutdown and an unplanned interruption, which can impact process stability, plant equipment and on-site people safety.
This is exactly why continuous process manufacturing warrants special treatment. Safety Instrumented Systems (SIS) help bring the operational process to a safe state in the event of a hazardous condition, and thus cannot be treated as normal endpoints. Clearly, any modification or change that affects their availability or communications has to undergo appropriate engineering and safety assessment prior to organizational implementation.
Furthermore, patch management in this scenario becomes a risk-based decision rather than a simple routine, particularly when maintenance windows can be as long as months. CISA’s recommended practices call for separate ICS patch processes and validation, while NIST SP 800-82 suggests testing and compensating controls if the patching is going to be delayed. Such measures usually include passive monitoring, tighter remote access, application allowlisting, backups, and engineered network zones.
Known vulnerabilities that require the next rotation are certainly not ignored. They are managed by mitigating the paths of the attacks. Network segmentation is done on the basis of this same concept, where the process dependencies must first be known by the engineers before allowing any topology changes. Similarly, incident management is restricted since organizations cannot easily isolate a system responsible for managing hazardous processes.
The resilience challenge is absolutely there in the numbers. The SANS 2026 survey found that 34% of the 696 respondents had an ICS/OT incident in the last year and 72% of those incidents resulted in downtime. Only 15% of respondents said they could sustain critical operational processes and safe production in manual mode, without centralized SCADA or distributed control system (DCS) capabilities, for more than 48 hours, while 40% were unsure whether they could do so.
Visibility does appear to make a real difference. Honeywell’s 2026 OT Cybersecurity Benchmark Report disclosed that organizations with stronger asset visibility demonstrated better incident recovery, with 42% reporting downtime of six hours or less compared to 25% among those with weaker asset visibility.
Across continuous process manufacturing sites, containment is never a purely technical decision, but emerges more as a call shared among cybersecurity, control engineering and operations teams. Compromised controllers that are still holding a reaction stable or keeping pressure within limits cannot simply be switched off the way an infected laptop can be unplugged. Moreover, responders have to weigh the cost of letting an attacker linger against the cost of destabilizing operational processes, and oftentimes that trade-off is rarely clear in the first few minutes. Getting it right depends on conversations and relationships that have been built long before the cyber threat incident begins.
Why continuous process industries need different cyber controls
Industrial Cyber asked experts to examine which cybersecurity controls commonly deployed in IT and discrete manufacturing are impractical or potentially unsafe in continuous-process environments. They also looked into what alternative approaches operators are adopting to protect OT without disrupting production or compromising safety.
“Many core cybersecurity controls still apply in continuous process environments. What changes is how, when, and where they are applied to account for process dependencies, safety, and operational constraints,” Andrew McPhee, principal architect for industrial security at Cisco, told Industrial Cyber. “Some conventional IT practices assume that an individual device can be changed or isolated as soon as a threat is identified. Automatic patching, reboots, or immediate quarantine can create problems if applied to a controller, drive, or other asset without understanding its role in the process. A change intended to protect one device can interrupt communications or affect the process it helps control.”
He added that securing these environments requires careful planning; actions cannot be purely reactive. Controls and response actions should be built around consequence boundaries designed to preserve safety.
“While a fix is deferred, operators can reduce exposure by disabling an unnecessary service or restricting which devices can reach it,” McPhee added. “Network controls should be based on observed communications, tested before enforcement, and designed around process zones. If containment is needed, the response and cutoff point should already be understood so a zone can be isolated while its internal control and safety functions remain available.”

Chris McLaughlin, principal advisor at Trusted OT Advisors, LLC, told Industrial Cyber that continuous process environments may operate for years without shutting down, making software patching and technology refreshes difficult to schedule. “Some discrete manufacturing operations also run for extended periods between maintenance windows. This includes industries such as steel or glass production that have assets that run for over a decade or longer without going down. These long operating cycles require a different approach to patching and upgrade planning, rather than relying on frequent shutdowns.”
“Controls that require a restart or interrupt communications can be impractical and potentially unsafe in continuous process manufacturing. Where those interventions cannot be performed safely, alternatives focus on continuous oversight and tightly controlled access,” Christina Hoefer, vice president of OT and IoT strategy at Forescout, told Industrial Cyber. “Continuous network monitoring and anomaly detection help operators identify changes and compare observed activity with expected behavior and known process flows.”
She added that tight controls on both local and remote network access complement that visibility, keeping process networks as isolated as practical. “Where connectivity is necessary, it should be temporary, explicitly approved and limited to the specific systems required, rather than leaving standing access in place.”
Asad Naeem, manager for instrument, control systems and OT cybersecurity at Engro Fertilizers, mentioned that immediate deployment of manufacturer updates and security patches, as commonly practiced in IT, is not feasible in continuous process plants.
“Untested patches can disrupt communications, cause loss of operator visibility, or affect process control. Even representative testing cannot eliminate all risks, as differences between test and production environments may lead to unexpected issues,” Naeem told Industrial Cyber. “A safer approach is to deploy OEM-recommended and validated patches during planned shutdowns. System architecture should also incorporate redundant operator and engineering workstations, allowing patches to be tested on one station while maintaining operational availability through the other.”
Rethinking patching in continuous process operations
The executives address how operators approach patching when the next maintenance window may be months away, particularly if a known vulnerability is being actively exploited elsewhere.
“Operators should not treat patching as a simple yes-or-no decision. A vulnerability creates a potential route for an attacker and applying a patch addresses that known flaw,” McPhee said. “If the firmware cannot be upgraded, a patch is unavailable, or it cannot be applied safely before the next maintenance window, operators can still reduce exposure and lower the risk of exploitation.”
He added that they should first check whether the exploit’s conditions apply to the device, then consider ways to reduce exposure. “For example, if a vulnerable PLC web service is not needed for operations, disabling it may remove that path to exploitation. If a vulnerability only applies when a PLC is out of RUN mode, strict controls over who can change its mode—and when—may reduce the risk until patching is possible. Where the vulnerable service must remain available, network segmentation can limit which devices can reach it. These measures do not remove the vulnerability, but they can substantially reduce the risk of exploitation until the next patch window.”

Operators often use redundancy, clustering, and virtualization to reduce disruption during patching, McLaughlin identified. “Patches are first approved by the industrial control system (ICS) manufacturer and tested in a separate environment. They are then applied in stages: patch the secondary system, switch operations to it, and monitor performance before patching the primary system. When patching must be deferred, organizations use other mitigating controls such as application allowlisting, segmentation and system hardening.”
“In continuous process environments, patching is challenging because many OT systems must remain available around the clock, and updates may require extensive testing, vendor approval, or a planned shutdown to avoid disrupting production,” Hoefer said. “Patching decisions must account for asset criticality, exposure, communication paths, and operational dependencies, rather than vulnerability severity alone.”
She noted that when immediate patching isn’t operationally safe, operators should prioritize systems that are externally exposed, reachable through remote access, or essential to production, using segmentation, restricted access, and continuous monitoring as compensating controls until a maintenance window is available.

Naeem said that when patching must be deferred, operators can adopt virtual patching and other compensating controls to reduce exposure. “These include restricting network access, tightening firewall rules, disabling unnecessary services, and strengthening endpoint protection. If a vulnerable system has unnecessary internet connectivity, it should be disconnected immediately.”
He added that operators should also understand how the vulnerability is being exploited in the wild and implement layered countermeasures to break the attack path, following the defense-in-depth principle. “The objective is to manage exposure and reduce the likelihood of exploitation until a properly tested and validated patch can be deployed during a suitable maintenance window.”
Balancing network segmentation and continuous process safety
The executives examine considerations that shape network segmentation in continuous process environments, where cutting communication between control system components could disrupt production or compromise process safety.
For years, McPhee noted that OT cybersecurity efforts have focused on visibility. “That work now needs to translate into protection. Network segmentation is a key part of that step, but it cannot be implemented blindly. Operators need to understand which assets communicate, how they behave, which services they use, and how those relationships support plant operations.”
He noted that in continuous process environments, routine communication patterns may be stable, giving operators a useful baseline. Operators can use this visibility to build segmentation policies and, with the right tools, simulate their effects before enforcing them in production. This approach helps accelerate segmentation projects by suggesting rules based on observed baseline and using simulation to avoid unintended impacts when actively controlling access.
“Network segmentation must limit the spread of an attack without disrupting essential communications. Placing several production lines in one security zone increases the potential impact: malware introduced on one line could affect the others,” McLaughlin said. “Operators should map dependencies and communications between lines, then assess how losing each connection would affect production and process safety. That analysis should guide zone boundaries and permitted communications.”
Hoefer recognizes that segmentation must reflect how the industrial process actually operates, including which devices communicate, the protocols they use, and the dependencies that support production and safety.
“Segmentation and access control can be more stringent at the perimeter and between supporting services and OT networks, assuming there is no direct process dependency, such as on time servers or user management systems,” she observed. “Then within the control system networks, segmentation will need to be more dynamic and OT-aware with fallback options to allow safety system to run without interruptions.”
Network segmentation must be based on process dependencies, not just cybersecurity requirements, Naeem said. “Consider communication between a Distributed Control System (DCS) and a Safety Instrumented System (SIS).”
He added that while ISA/IEC 62443 requires separate security zones, introducing a firewall between these systems may create additional failure modes or a single point of failure. “Where technically feasible, dedicated point-to-point links can provide an alternative, keeping communication physically isolated from the wider OT network. Any segmentation approach must undergo engineering and safety assessments to ensure reliable communication without compromising process safety.”
Containing cyber incidents without disrupting operations
The executives assess how operators make containment decisions when isolating a compromised system could jeopardize process safety, and who has the authority to approve or override those decisions.
Highlighting that safety should never be compromised, McPhee said that safety procedures work because the boundary and response are understood in advance. “If a person steps across a laser tripwire, the operator knows what will happen and why. Cybersecurity needs the same clarity. For every zone, the plant should know where the cutoff point is, what will be isolated if an incident crosses that boundary, and what that means for the process. If isolation affects production, that consequence should already be understood and accepted. When something goes wrong, people can follow the agreed response rather than improvise under pressure.”
He added that the ISA/IEC 62443 zones and conduits can help define these boundaries. Segmentation should reflect how the process works.
“The cyber ‘big red button’ should cut east-west communications between process zones while preserving the control and safety functions within the isolated zone,” according to McPhee. “Because a PLC or drive may depend on other assets in its process zone, the planned response to a compromised asset may be to isolate the whole zone, not to contain the single asset. Everyone should know which zone-to-zone connections will be cut and what will happen to production. The cutoff point, response, and authority to act should be clear in plant procedures. There should be no surprises when a zone must be isolated.”
“Organizations should identify and test cyber incident scenarios in advance, mapping system dependencies and the safety consequences of isolation. Pre-agreed decisions help teams respond quickly,” McLaughlin pointed out. “For situations not anticipated, the Cyber Incident Response Plan (CIRP) should define how the appropriate operational technology (OT) and IT personnel assess the trade-offs and who has decision-making authority.”
Hoefer observed that an abrupt disconnection could cause greater harm than allowing the compromise to continue while a safer response is coordinated. “An IT/SOC analyst who detects an attack, such as ransomware, therefore needs to consult production staff or the OT security representative before initiating isolation that could affect the process.”
She added that the assessment is joint – IT/SOC explains the attack and the consequences of leaving it uncontained, while OT assesses the affected systems, their process dependencies and what a sudden interruption would mean for production, equipment and people. IT/SOC advises on the cyber threat, but the final decision on actions affecting the operating process rests with the OT team.
“Incident containment decisions require collaboration between Control Systems, Cybersecurity, and Operations teams. Cybersecurity assesses the threat and recommends containment measures, while Operations evaluates process and safety implications,” Naeem said. “The Control Systems team develops a technically safe response based on system architecture, functionality, and recovery procedures.”
For example, he added operator HMIs may be isolated while the control system continues operating autonomously, with field operators maintaining process oversight where feasible. HMIs and servers can then be restored from validated backups. Remote programming access to safety-critical systems should remain disabled to minimize potential cyber risks.
Managing cyber exposure without disrupting production
The executives explain how continuous process operators assess residual cyber risk, and brief leadership on it when patching and other standard fixes are off the table or must wait.
McPhee identified that there is no such thing as zero risk in a running plant. “Risk is about more than whether an asset has a vulnerability. Does it use insecure protocols? Can it accept unauthenticated commands? Does its web server have default credentials—or none at all? The key question is: if the device were compromised, what is the worst credible outcome, and how likely is it? Risk assessment considers both the potential impact and the likelihood of that outcome.”
He added that risk tolerance is the level of risk the organization is prepared to carry, taking into account safety, environmental impact, production loss, and financial cost. Safety must have the highest priority; financial and operational impacts also need to be clear. Risk acceptance is the explicit decision to carry a specific residual risk, with its controls and limits understood.
“Operators should frame the decision in business terms. If a line went down, how quickly could it return to service? Tested backups and recovery plans can reduce the impact and cost of downtime, though they do not remove the vulnerability,” according to McPhee. “What would the likely downtime cost? Is leadership willing to accept that potential cost until remediation is possible, or does the risk justify action sooner? Framing residual risk in terms of business impact gives leadership the information it needs to make an informed decision.”
McLaughlin called for communicating residual cyber risk in terms of likelihood and business impact, explaining how people, process, and technology controls reduce it. “When patching or other conventional remediation must be deferred, explain which alternatives are in place and what risk remains. These may include application allowlisting, which permits only approved applications to run, and other system-hardening measures.”
“When remediation must be deferred, the assessment needs to go beyond the presence of a vulnerability to examine how it could be exploited in the operating environment. That means evaluating network connectivity, local access and remote access by employees or third parties,” Hoefer detailed. “The assessment must also consider what successful exploitation would mean for production uptime, process integrity and safety. The next step is to assess how compensating controls reduce that likelihood and limit the potential impact.”
She also noted that leadership should receive a clear explanation of the exposure, the operational consequences, and the next opportunity for remediation. This makes the decision explicit: whether the residual risk is acceptable until that opportunity.
“When vulnerabilities cannot be immediately remediated due to operational or technical constraints, the residual cyber risk should be formally documented and communicated through the organization’s enterprise risk register,” Naeem said. “This provides leadership with visibility of the exposure, its potential operational and safety consequences, and the compensating controls in place.”
He concluded that where permanent remediation requires additional investment, the requirement should be prioritized and incorporated into upcoming capital expenditure (CAPEX) or operating expenditure (OPEX) planning cycles. “This approach ensures that deferred remediation remains a managed business risk rather than an overlooked technical issue, with clear ownership, visibility, and a planned path toward resolution.”
The post Continuous process manufacturing: Why OT cybersecurity demands different approach when plants cannot stop appeared first on Industrial Cyber.
