Probability Of Failure On Demand Explained

Design Hazard Review for New Processing Facility

Probability Of Failure On Demand Explained is more than a formula—it is the shared language that functional safety, instrumentation, and operations teams use to judge whether a safety instrumented function (SIF) will work when their facility most needs it. In refineries, gas processing plants, offshore platforms, power stations, chemical terminals, and high-hazard manufacturing lines across the UAE, the ability to quantify the chance that a protective layer may not act on demand is central to risk decisions, compliance under IEC 61511, and the day-to-day assurance expected by regulators and corporate leadership alike.

Professionals often first meet the concept of PFDavg (average probability of failure on demand) in a LOPA or SIL determination workshop. But the topic quickly widens: it touches architecture (1oo1, 1oo2, 2oo3), proof test intervals, diagnostic coverage, common cause failures, bypass/override management, and the traceability of assumptions back to a safety requirements specification (SRS). Getting these pieces right—practically and defensibly—is what turns a number in a spreadsheet into credible risk reduction in the field.

Because many facilities operate in harsh environmental conditions, from high temperatures and humidity to corrosive atmospheres and airborne sand, assumptions that might be reasonable in temperate environments need careful validation in the Gulf. That is why PFDavg is not just an engineering exercise; it is an operational discipline that bridges design intent, maintenance realities, and incident learning.

Who This Concerns

Probability Of Failure On Demand Explained

The topic sits at the intersection of process safety and controls. It directly concerns teams who specify, install, operate, maintain, and govern safety instrumented systems (SIS), and those accountable for overall risk management. In the UAE context, this typically includes:

  • HSE and Process Safety Managers responsible for ALARP demonstrations, corporate standards, and assurance against IEC 61511 and ADNOC HSE requirements.
  • Instrumentation and Control (I&C) Engineers who select sensors, logic solvers, final elements, and define proof test coverage and intervals.
  • Operations and Maintenance Leaders who manage overrides, impairment planning, and turnaround windows affecting SIF availability.
  • Project and Engineering Managers on greenfield and brownfield modifications, where SIL assignment, verification, and validation must be coordinated with EPC contractors and OEMs.
  • Industrial Risk, QRA, and LOPA practitioners who translate consequence and frequency estimates into target risk reduction and SILs.
  • Compliance Officers and Auditors engaging with OSHAD (Abu Dhabi), Dubai Municipality, Civil Defence, and sector regulators to evidence functional safety lifecycle controls.

Contractors, integrators, and third-party inspectors are also affected, because their deliverables—FMEDA data packages, proof test procedures, FAT/SAT records—feed directly into the integrity arguments supporting PFDavg claims.

Why This Matters

Probability Of Failure On Demand Explained

Understanding PFDavg matters because it quantifies how much risk a SIF actually reduces during the period between proof tests. When leadership signs off that a refinery’s overpressure protection meets SIL 2, or a pipeline’s HIPPS meets SIL 3, they are implicitly accepting that the achieved average probability of failure on demand sits within a defined range. If the data are weak, assumptions are optimistic, or maintenance realities drift from plan, the gap between required and achieved risk reduction widens—sometimes invisibly—until a demand occurs.

From a compliance perspective, IEC 61511 requires not only that SIFs are designed to meet target SILs, but that the achieved performance is verified, validated, and maintained. ADNOC operators, power and water utilities, and major chemical firms increasingly expect evidence that PFDavg figures are traceable to credible data and that operational feedback (near-misses, spurious trips, failed proof tests) is used to refine those figures. Regulators and insurers look for a living lifecycle: competent people, controlled changes, and measured performance that aligns with the safety requirements specification.

Practically, PFDavg choices drive cost, availability, and human workload. Opting for 1oo2 sensors may reduce PFDavg, but it can increase common cause vulnerability or create a higher spurious trip rate that disrupts plant stability. Extending a proof test interval from one to three years might fit a turnaround cycle, but may sharply erode risk reduction if diagnostic coverage and partial-stroke tests do not compensate. Knowing where these trade-offs sit—quantitatively—prevents surprises later in audits, in management of change reviews, or after an incident.

Key Points to Understand

Probability Of Failure On Demand Explained

Having framed why the topic is vital, it helps to unpack its core elements in a structured way. At its core, Probability Of Failure On Demand Explained boils down to one question: over the period of interest (usually one proof test interval), what is the average likelihood that a particular SIF will fail dangerously when called upon? The answer depends on a set of well-defined variables and assumptions.

Low demand versus high/continuous demand

Probability Of Failure On Demand Explained

PFDavg applies to low-demand functions—those expected to be called upon no more than once a year, typically many times less. For high/continuous demand functions, risk targets are expressed as PFH (probability of dangerous failure per hour). IEC 61511 mirrors IEC 61508 in mapping PFDavg (for low demand) and PFH (for high/continuous demand) to SIL ranges. Most process sector SIFs—ESD, HIPPS, burner management trip, chlorine dosing shutdown—are low demand, hence PFDavg is the relevant measure.

Relationship between SIL and PFDavg

Probability Of Failure On Demand Explained

SIL is a discrete level that maps to bands of risk reduction. In low-demand mode, the mapping is commonly summarized as below:

SIL PFDavg range Risk Reduction Factor (approx.)
SIL 1 10^-2 to < 10^-1 10 to 100
SIL 2 10^-3 to < 10^-2 100 to 1,000
SIL 3 10^-4 to < 10^-3 1,000 to 10,000
SIL 4 10^-5 to < 10^-4 10,000 to 100,000

A LOPA determines the required risk reduction; SIL verification then shows the achieved PFDavg with the proposed architecture, proof test strategy, and component choices. The two must match or exceed the target.

What drives PFDavg

Probability Of Failure On Demand Explained

Several variables combine to produce an average probability for the interval of interest. The dominant contributors in the process sector are often the final elements (e.g., shutdown valves) because they may have lower diagnostic coverage and are tested less frequently. Key terms include:

Term Meaning Notes
λD Dangerous undetected failure rate Per hour; often derived from FMEDA or field data and adjusted for environment.
DC Diagnostic coverage Fraction of dangerous failures detected automatically by built-in diagnostics.
T1 Proof test interval Average time between periodic tests capable of revealing hidden dangerous failures.
β-factor Common cause fraction Fraction of failures that affect redundant elements simultaneously.
MTTR Mean time to repair Affects exposure after a detected failure or a proof test reveals a fault.
Architecture Voting arrangement 1oo1, 1oo2, 2oo3 etc.; affects combinatorial probability of failure.

The simple 1oo1 approximation PFDavg ≈ λD × T1 / 2 is often a starting point, but real verifications go further, accounting for diagnostics (which reduce λD to λDU), partial-stroke tests on valves (which increase effective coverage), repair times, proof test coverage, and β-factor for redundant architectures. For example, a 1oo2 sensor set can dramatically lower PFDavg if common cause is well controlled; if not, the benefit is muted.

Data sources and credibility

Probability Of Failure On Demand Explained

IEC 61511 expects component failure rate data to be credible and justified. Typical sources include vendor FMEDA reports, industry databases (OREDA), and site-specific field return data. Field feedback is particularly valuable in the UAE, where heat, dust, and corrosive marine air can shift failure rates relative to generic assumptions. Where claims rely on diagnostics or partial-stroke testing, the corresponding coverage figures must reflect actual device capabilities and test procedures—not catalog aspirations.

Traceability to the SRS and lifecycle

Probability Of Failure On Demand Explained

PFDavg targets and their basis must be documented in the SRS, then carried through design, FAT/SAT, commissioning, and into operation where proof testing, bypass control, and maintenance keep the function within its validated envelope. Periodic functional safety assessment (FSA) checkpoints verify that assumptions still hold and that performance metrics (spurious trip rates, overdue tests, failed proof tests) are monitored and acted upon. For additional reading on governance and policy insights relevant to industrial risk and safety management, see resources periodically curated under CABINET NPM.

How This Works in Practice

Translating the concepts into operating life starts before hardware is ordered—typically in hazard and risk studies such as HAZOP, HAZID, and LOPA. Once a target risk reduction is agreed for an initiating event and consequence, the SIF’s required SIL emerges. Designers then propose an architecture and device set that can achieve the target, documenting assumptions in the SRS, including proof test intervals, bypass limits, diagnostic responses, and independence from the BPCS per IEC 61511.

During design, teams may evaluate alternative architectures against availability and operability constraints. A gas compressor’s high-high discharge pressure trip, for example, may use 2oo3 transmitters to balance reliability and spurious trip exposure. A pipeline HIPPS may require dual final elements in series to achieve the necessary PFDavg, with partial-stroke tests and leak-tightness verification baked into the maintenance plan. For a power boiler, burner management system trips must align with NFPA codes while maintaining verified PFDavg across sensors, logic, and fuel shutoff valves.

Proof test intervals often anchor to turnaround cycles, but that is not the only option. Many facilities adopt staggered partial-stroke testing on shutdown valves to lift effective coverage between full turnarounds, while keeping plant impacts manageable. Others deploy on-line proof tests for redundant transmitters, using bypass permits and override controls to maintain protection while an element is isolated. The assumptions behind these choices—test effectiveness, bypass duration, diagnostic detection—must be quantified in the PFDavg calculation and controlled procedurally in operations.

Once in operation, performance monitoring keeps the numbers honest. Sites increasingly use barrier health dashboards to visualize overdue proof tests, instrument impairments, open overrides, and SIF trip demands. These metrics flow into monthly management reviews and into the functional safety assessment schedule. When proof tests reveal hidden failures at rates higher than expected, teams update their data assumptions and adjust proof test intervals or component choices accordingly—closing the loop demanded by IEC 61511.

In the UAE, environmental stresses add an extra layer: sand ingress can compromise valve actuator seals, marine atmospheres accelerate corrosion in terminals, and extreme heat pushes device specifications. These realities influence both component selection (e.g., IP ratings, corrosion-resistant materials, suitable valve packing) and maintenance strategies (more frequent inspections, enhanced protective enclosures), which in turn shape credible PFDavg claims.

Benefits and Practical Use Cases

Clear, defensible PFDavg thinking yields benefits that reach well beyond compliance. It supports better design, steadier operations, and more confident decision-making. The most visible payoffs typically include:

  • Risk transparency: Decision-makers see the quantitative gap between required and achieved risk reduction for each SIF.
  • Smarter architecture choices: The cost and complexity of redundancy are weighed against demonstrable risk reduction.
  • Operational fit: Proof test intervals and procedures integrate with turnaround windows and staffing, without eroding integrity.
  • Credible audits: Evidence lines from SRS to field maintenance show that IEC 61511 requirements are being lived, not just written.
  • Continuous improvement: Field failure data, spurious trip histories, and near-miss analyses feed parameter updates and actions.

Typical use cases where PFDavg clarity makes a decisive difference include:

  • Upgrading an ageing ESD system at a gas plant, where moving from 1oo1 to 1oo2 sensors plus enhanced diagnostics achieves SIL 2 without adding untenable spurious trips.
  • Defending HIPPS design for a trunkline under ADNOC oversight, where dual valves in series and quarterly partial-stroke tests close the final-element risk gap.
  • Optimizing storage terminal overfill protection by allocating SIL 2 to independent high-level shutdown, while re-baselining proof test intervals to reflect actual crew capacity and tank utilization patterns.
  • Harmonizing a power plant’s burner management and fuel gas shutoff arrangements with NFPA and IEC 61511, confirming that diagnostic coverage plus targeted spares reduce PFDavg within the SIL 1 range.

Best Practices and Points of Caution

Having outlined benefits and applications, it is important to anchor them in doable, evidence-based practices and the common pitfalls to avoid. The following recommendations reflect lessons seen across the oil and gas, petrochemical, power, and logistics sectors.

  • Start with LOPA quality: A precise target risk reduction simplifies everything downstream. Double-check initiating event rates, IPL credits, and consequence modelling before locking SIL targets.
  • Choose data carefully: Prefer vendor FMEDAs and industry-validated sources; adjust for environment. Avoid unreferenced internet tables or mixing datasets without reconciliation.
  • Model common cause explicitly: In redundant architectures, set β-factors credibly and design to reduce them (diverse technologies, separation, environmental controls).
  • Treat final elements as dominant: Shutdown valves and actuators often drive overall PFDavg; invest in partial-stroke testing, valve condition monitoring, and realistic proof test coverage.
  • Control bypasses and overrides: Quantify their effect on PFDavg and limit duration with permits, alarms, and independent approvals.
  • Integrate proof tests operationally: Stagger tests for redundancy, train technicians on coverage-critical steps, and capture as-found vs. as-left conditions.
  • Keep the SRS alive: Reflect changes from MOC, instrument substitutions, and revised intervals; update verification math accordingly and revalidate where needed.
  • Measure and learn: Track spurious trips, failures on demand, overdue tests, and diagnostic alarms. Feed these into periodic functional safety assessments.
  • Mind independence: Separate SIS from BPCS per IEC 61511; ensure power, I/O, and networks do not create hidden common causes.
  • Engage stakeholders: Align engineering, operations, maintenance, and HSE early, so that SIL claims match real-world capacity and culture.

Frequently Asked Questions About Probability of Failure on Demand

What is the difference between PFDavg and PFH?

PFDavg (average probability of failure on demand) applies to low-demand functions—those expected to be called upon infrequently (generally less than once per year). PFH (probability of dangerous failure per hour) is used for high- or continuous-demand functions. The choice aligns with IEC 61511/IEC 61508 definitions, and SIL bands are mapped differently for each regime. Most process safety SIFs in refineries, terminals, and power plants are low demand and therefore use PFDavg.

How does proof test interval affect PFDavg?

All else equal, longer proof test intervals increase PFDavg because hidden dangerous failures can accumulate longer before detection. The simple 1oo1 approximation shows a linear relationship with interval length. In practice, diagnostics and partial-stroke tests can offset some of this increase by revealing failures earlier, but the assumptions behind their effectiveness must be credible and documented in the SRS and maintenance procedures.

Do partial-stroke tests meaningfully reduce PFDavg for shutdown valves?

Yes—if the test is designed and executed to reveal a relevant subset of dangerous failures and if test coverage claims are reasonable. Partial-stroke tests cannot detect every failure mode (e.g., tight shutoff capability), but they can significantly reduce the exposure time for many mechanical and actuator-related failures. The resulting improvement must be reflected in the model by adjusting proof test coverage and effective intervals.

Which data sources are acceptable for failure rates?

IEC 61511 expects data to be credible, justified, and appropriate to the context. Typical sources include vendor-provided FMEDA reports, industry databases such as OREDA, and site-specific field return data. In the UAE, environmental factors (heat, dust, marine atmosphere) may diverge from generic assumptions, so local experience and failure reporting are valuable to calibrate parameters.

How do common cause failures change redundancy benefits?

Redundant architectures like 1oo2 or 2oo3 assume elements fail independently. Common cause mechanisms (shared environment, common power, identical firmware bugs) mean some proportion of failures affect multiple elements simultaneously. The β-factor models this. If β is high, the benefit of redundancy shrinks. Designing for diversity and independence—separate enclosures, staggered proof tests, different manufacturers—helps keep β low and preserves the risk reduction advantage.

What does IEC 61511 require regarding PFDavg evidence?

IEC 61511 requires assignment of SIL targets based on risk, verification that the design can achieve those targets (including PFDavg calculations with justified data and assumptions), and validation that the implemented SIF functions correctly. The standard also demands maintaining performance over the lifecycle through proof testing, competent personnel, management of change, periodic functional safety assessment, and feedback of operational experience into the data and intervals.

How should bypasses and overrides be managed?

Bypasses and overrides should be minimized, controlled via permit-to-work or dedicated override management systems, time-limited, and alarmed to operations. Their effect on risk exposure should be quantified, and cumulative bypass hours tracked. Many organizations set explicit limits in the SRS for maximum allowed bypass duration, approval levels, and monitoring requirements to control the added risk.

How do HAZOP and LOPA relate to PFDavg?

HAZOP identifies hazardous scenarios and initiating causes; LOPA quantifies the frequency and evaluates layers of protection to determine the required risk reduction for each scenario. From this, a target SIL arises for any SIF credited. PFDavg is the metric used to verify that the proposed SIF achieves that target in low-demand mode. The logic is sequential: hazard identification, risk evaluation, SIL assignment, and then PFDavg verification against the target.

Conclusion and Next Steps

Bringing the strands together, the ability to quantify average probability of failure on demand connects technical design, human procedures, and the realities of plant operation. Earlier we saw how data credibility, diagnostics, proof test coverage, and common cause can swing achieved risk reduction by orders of magnitude, and how operational constraints in the UAE—turnaround schedules, environmental stressors, staffing—shape what is realistically maintainable. When PFDavg is treated as a living parameter tied to your SRS and measured in the field, a SIL claim becomes more than a label; it becomes a reliable promise of protection aligned with IEC 61511 and with stakeholder expectations.

To go further on this topic, NPM’s broader SIL Verification Services UAE page covers this in more depth.

Explore NPM’s References

NPM’s track record with functional safety and risk reduction spans refineries, offshore and onshore hydrocarbons, petrochemical complexes, midstream pipelines and terminals, utility-scale power and water, as well as high-hazard manufacturing. Across these environments, our teams have engaged with the full lifecycle demanded by IEC 61511: hazard and risk studies that define credible SIL targets; architecture and device selections grounded in justified data; SIL verification that treats PFDavg as a decision tool rather than a paperwork exercise; validation and commissioning that prove functions in the field; and ongoing operational assurance that keeps PFDavg aligned with reality. This body of work also intersects with adjacent disciplines—alarm management, human factors, maintenance strategy, and cybersecurity—because in practice these domains interact with how SIFs behave and are maintained. The references reflect a spectrum of objectives: sometimes to demonstrate that existing protection was already adequate if documented properly; sometimes to recalibrate proof test intervals after failure data diverged from generic assumptions; and sometimes to redesign SIFs where common cause or environmental stressors were undermining risk reduction. The portfolio shows how rigorous but pragmatic approaches help clients balance safety integrity with operational availability and regulatory confidence—worldwide, across diverse regulatory and corporate frameworks—and how lessons learned in one sector can usefully inform another without diluting context-specific detail.

For more information about our references, visit Our References.

Start a Conversation

If you would like to explore how the ideas in this article apply to your facility—whether you are reassessing proof test intervals, checking SIL alignment, or sharpening your lifecycle governance—NPM is available to discuss practical next steps. We are happy to answer questions, share perspective, and help you weigh options in context. For any further questions or to speak with our team, Contact Us.