OT Cybersecurity Risk Assessment: A Consequence-Driven Method

How is an OT risk assessment carried out? Scenario steps based on safety, environmental and production consequences, the link to HAZOP and SIS, an example matrix and the relationship with IEC 62443-3-2.
Technical summary
Technology
IEC 62443-3-2, HAZOP, LOPA, SIS, IEC 61511
Topic
Scenario-based, physical-consequence-driven cyber risk assessment in OT
Level
Advanced
Prerequisites
  • An up-to-date OT asset inventory and network diagram
  • Access to the plant’s process safety studies (HAZOP, LOPA)
  • Participation of the operations and process safety teams
In this article8 sections

IT risk assessments are usually built on the confidentiality, integrity and availability of information. In a manufacturing plant, the real question is different: as a result of a cyber incident, could someone be injured, would there be an environmental release, how long would production stop, would a defective product reach the customer? Below we explain step by step a scenario-based method that runs the OT cyber risk assessment on these physical consequences and describe its relationship with IEC 62443-3-2. If an up-to-date device and connection inventory is missing, record the current state of the network with a factory OT infrastructure discovery.

Why IT risk methods fall short in OT

In the asset-based IT approach, threats and vulnerabilities are listed for each server or application, and assets with high confidentiality impact come to the fore. In OT, this approach misleads in three ways:

  • Impact is measured in the wrong place: There’s no secret information inside a PLC; but if its integrity is compromised, a valve can open at the wrong moment. Confidentiality-weighted scoring shows this asset as low importance.
  • Assets don’t mean much on their own: An HMI’s risk depends on what the connected process does and whether the operator can safely stop the process without the HMI.
  • Engineering safeguards aren’t taken into account: A pressure relief valve, mechanical limits or an independent safety system limit the physical consequence of a cyber incident. An assessment that doesn’t see these either exaggerates the risk or directs investment to the wrong place.

A common picture: in IT’s risk register, the email server appears as “critical” and the PLC driving the dosing pump as “medium.” The method is wrong, and so is the result. That’s why the consequence-driven method starts from the other end: it first defines the bad outcome that must be prevented, then looks for the cyber scenarios that could lead to it.

Consequence categories: safety, environment, production, quality

The first step is adapting the risk language the organization already uses to OT. Most plants have consequence scales prepared for OHS, environment and operations; the cyber assessment shouldn’t reinvent them but use them. The table below is an example scale structure; fill in the threshold values with the organization’s own criteria.

Level Safety Environment Production / quality
1 – Low No injury On site, no record required Short stoppage, product not affected
2 – Medium First aid level On site, can be cleaned up Shift-level stoppage
3 – High Lost-time injury A release requiring notification Days of stoppage or defective shipments
4 – Critical Serious injury or death A release spreading off site Long-term stoppage, recall

A scenario’s consequence level is the highest value across the categories. The quality category carries extra weight for food, pharmaceutical and automotive suppliers: a recipe parameter changed without anyone noticing can cause serious consequences on the customer side without any physical damage. No damage, but a recall.

Steps of a scenario-based assessment

The assessment is carried out for a system boundary: a line, a process unit or a plant. The steps:

  1. Define the scope: The system boundary, the assets within it and external connections. Without an up-to-date inventory, the assessment rests on assumptions; for the inventory method, see the passive asset discovery in OT networks article.
  2. List the bad outcomes: With the operations and process safety teams, identify the most serious physical consequences that could occur in this system.
  3. Map the control points: For each outcome, find which control function (PLC logic, HMI command, setpoint, safety interlock) would need to change.
  4. Work out the attack path: Which route reaches that control function: remote access, the engineering workstation, a crossing from the corporate network, removable media, a vendor.
  5. Count the existing safeguards: Both cyber controls (segmentation, authentication, monitoring) and engineering safeguards (mechanical protection, an independent safety system).
  6. Score likelihood and consequence: Determine the residual risk taking the safeguards into account.
  7. Compare with acceptable risk: For scenarios exceeding the threshold, recommend additional controls and rescore the risk after the recommendation.

Likelihood scoring should rest on qualitative criteria. For example: if the attack path starts from a service directly reachable from the internet, high; if it requires compromising a privileged account on the corporate network and then getting through the IDMZ, medium; if it’s possible only with physical access to the plant, low. If the criteria are written down, different people give similar scores. If they aren’t, the score is the score of whoever talks loudest in the workshop.

In process industry plants, process safety studies such as HAZOP and LOPA already document bad outcomes and the protection layers against them. The cyber assessment should use these studies as input:

  • HAZOP deviations: Deviations such as “high pressure”, “no flow” and “reverse flow” and their causes are examined. Deviations whose cause is a control loop failure can also be triggered by cyber means; these are scenario candidates.
  • LOPA protection layers: Whether each layer is programmable is determined. A mechanical relief valve isn’t affected by a cyber attack; an interlock logic running on a PLC can be.
  • Safety instrumented system (SIS): The independence of an SIS under IEC 61511 from the control system should also be verified from a cyber standpoint. In the TRITON/TRISIS incident made public in 2017, attackers targeted the safety controllers at a facility; this incident shows why the SIS should be included in the cyber risk assessment.

The link works both ways. The cyber assessment may reveal that a protection layer considered “independent” in the HAZOP is actually programmed from the same engineering workstation as the control system. Independent on paper, not on site. This finding should be fed back to the process safety team.

Example scenarios and assessment table

The scenarios below are generic examples prepared to show how the method is applied; the scores are for a hypothetical plant.

Scenario Attack path Existing safeguards Consequence / likelihood
Raising the furnace temperature setpoint Vendor remote access → HMI Independent over-temperature relay 2 / Medium
Changing a filling line recipe parameter Corporate network → MES → PLC Final product quality control (sampling) 3 / Medium
Running the chemical dosing pump continuously Engineering workstation → PLC program Interlock inside the PLC (programmable) 4 / Low–medium
All HMIs encrypted by ransomware Shared AD → OT Windows servers Manual shutdown procedure 3 / High

These scores are placed in an example likelihood–consequence matrix. The matrix cells are divided into “acceptable”, “ALARP / improvement should be considered” and “unacceptable” zones according to the organization’s risk appetite. The dosing scenario in the example usually falls into the unacceptable zone even though its likelihood is low, because its consequence level is critical; the only protection layer is inside the PLC the attack targets. The safeguard and the target are in the same place. The ransomware scenario, on the other hand, has a low safety consequence, a high production consequence and high likelihood; its mitigation priority is high too, for a different reason. The spread paths of this scenario are described in detail in the how ransomware spreads to OT article.

The relationship with IEC 62443-3-2

IEC 62443-3-2 uses risk assessment to partition a system into zones and conduits and to set a target security level (SL-T) for each. The standard’s workflow is roughly: identifying the system under consideration, an initial risk assessment, partitioning into zones and conduits, comparing initial risk with acceptable risk, a detailed risk assessment, documenting security requirements and approval by the asset owner.

The scenario method above corresponds most closely to the initial and detailed risk assessment steps. In the initial assessment, worst-case consequences are determined without considering controls; in the detailed assessment, scenarios, existing safeguards and likelihood come into play. The network engineering side of zone and conduit design is outside the scope of this guide; for that topic, see the IT/OT segmentation guide and the IEC 62443 industrial security consulting page.

Turning findings into control and investment decisions

The value of an industrial cybersecurity assessment depends on decision-makers being able to understand its output. The language of physical consequences is an advantage here: instead of “authentication is missing on the PLC”, the sentence “the dosing pump running uncontrolled is possible by compromising a single engineering workstation” makes management’s prioritization easier. When turning findings into decisions, follow these steps:

  • For each scenario in the unacceptable zone, at least one additional control and the expected risk level after that control
  • Also considering non-cyber options: a mechanical limit or an independent relay can be more reliable than a complex cyber control
  • The owner, budget and schedule of controls; compensating controls for systems that can’t be patched (OT vulnerability management and patching decisions)
  • Recording accepted risks with the asset owner’s signature and defining reassessment triggers
  • Carrying the highest-risk scenarios over into the incident response plan and exercise scenarios

If the report stays on the shelf, the assessment counts as not done. Every finding should have an owner and a date. To set up the risk assessment process at the organizational level or to merge the OT scope with existing information security risk management, see the cybersecurity risk assessment service. For the general framework of the method, the OT risk management section of NIST SP 800-82 Rev. 3 is also a useful reference.

Frequently asked questions

How often should the OT risk assessment be renewed?

Instead of a fixed calendar, tie it to triggering events: commissioning a new line or system, a change in network architecture, an update to the process safety study, a significant incident or new threat information. In addition to these, a review is done at the interval the organization sets.

Is this method enough to comply with IEC 62443-3-2?

Not on its own. The method is consistent with the standard’s initial and detailed risk assessment steps; but the standard also requires defining the system boundary, partitioning into zones and conduits, setting target security levels and having the requirements approved by the asset owner. A compliance claim requires all steps to be documented.

How should we score likelihood in OT?

Historical frequency data is usually insufficient for cyber incidents. Score likelihood on a qualitative scale based on factors such as the access point the attacker must reach, the knowledge and skill required, existing controls and the length of the attack path. Write down the scale’s criteria so that different assessors reach similar results.

If there’s a safety instrumented system, can cyber risk be considered low?

Only if its independence is preserved. An independent, robustly designed SIS is a strong protection layer; but being on the same network as the control system, being programmed from the same engineering workstation or being remotely accessible weakens this protection. In the assessment, verify the SIS’s cyber independence separately.

Support

Contact us for support or a discovery session on this topic.

Contact form
[email protected]

Eretek Cybersecurity

Firewall, VPN, endpoint security and threat prevention. We reduce the attack surface and measure and report the risks.

Knowledge Center

Other articles