IEC 61511 GuideChapter 4 of 12

LOPA and SIL determination

From scenario to SIL target

After the hazards and safeguards have been identified, the risk reduction needs to be analysed further. SIL determination decides whether a safety instrumented function (SIF) is needed for each hazardous scenario and, if so, what safety integrity level (SIL) it needs.

IEC 61511 does not prescribe a single method. The common ones are:

  • Risk graph: a qualitative method using consequence, occupancy, possibility of avoidance and demand rate;
  • Layer of protection analysis (LOPA): a semi-quantitative method, and the most widely used;
  • Quantified risk assessment (QRA): detailed techniques such as consequence modelling, event trees, fault trees and bow-tie diagrams, used for the most hazardous scenarios.

Whatever the method, it must be applied by a multi-discipline team with expertise in the process, its operation, and control and instrumentation.

SIL and risk reduction

SIL measures the risk reduction that a SIF provides against one specific hazardous event. For a SIF in low demand mode (demanded less than once a year, as most process industry SIFs are), each SIL is a band of risk reduction factor (RRF). The average probability of failure on demand (PFDavg) is the reciprocal of the RRF:

SIL Risk reduction factor (RRF) PFDavg
1 more than 10, up to 100 0.1 to 0.01
2 more than 100, up to 1,000 0.01 to 0.001
3 more than 1,000, up to 10,000 0.001 to 0.0001
4 more than 10,000, up to 100,000 0.0001 to 0.00001

The principle is to reduce risk until the target is met, and no further. If SIL 1 meets the target, it is also the lowest-cost option. Specifying SIL 3 where SIL 2 is enough is over-engineering, and it costs more for the whole life of the SIF: in design, testing and maintenance.

Risk graphs: a screening tool

A risk graph is simple to use, but it must be calibrated by a safety professional before use. Calibration makes sure the consequence and frequency categories are appropriate, and that the outcome meets the organisation’s tolerable risk criteria.

The team must also take care not to credit the same factor twice. For example, if the consequence comes from a study that already accounts for occupancy, the team must not reduce the result again for low occupancy.

A correctly calibrated risk graph gives conservative results. If it produces very few SIL 1 results, LOPA may be overkill. If it produces many SIL 2 or SIL 3 results, the time taken to do LOPA is usually well spent.

How LOPA works

LOPA analyses one hazardous scenario at a time. The scenario starts with an initiating event, such as an equipment failure or a human error. The analysis then looks at each independent protection layer (IPL) that could stop the event escalating to the consequence.

  1. Initiating event frequency. The cause is given a frequency per year, from site experience or generic data.

  2. Conditional modifiers. Where the procedure allows it, probabilities such as an enabling condition, or the chance that a person is present, can be applied.

  3. Independent protection layers. Each IPL is given a PFD. To count as an IPL, a layer must be:

    • specific to the scenario;
    • effective within the process safety time;
    • independent of the initiating event and of every other IPL credited;
    • auditable.

    Examples are operator response to an alarm, a basic process control system (BPCS) function, and a mechanical relief valve.

  4. Mitigated frequency. The initiating event frequency is multiplied by every PFD and modifier.

  5. Compare with the target. If the mitigated frequency is above the tolerable frequency for that consequence, more risk reduction is needed.

Layers can either prevent the hazardous event or mitigate it after it happens (secondary containment, for example). LOPA can consider both, but it should focus first on prevention.

A written LOPA procedure is strongly recommended. It sets typical initiating event frequencies, IPL credits and the rules for independence before the study starts, so that every study is consistent. Every assumption must be recorded, so that others can follow the results later.

Worked example

A scenario begins with a pump failure, expected about once in 10 years: 0.1 per year. Three independent protection layers can each stop it escalating:

Item Frequency or PFD Running frequency (per year)
Initiating event: pump failure 0.1 per year 0.1
IPL 1: operator response to an independent alarm PFD 0.1 0.01
IPL 2: BPCS control action PFD 0.1 0.001
IPL 3: relief valve, sized for the full case PFD 0.01 0.00001

If every layer fails, the hazardous outcome happens about 1 × 10⁻⁵ per year. Note that the result has the same time base as the initiating event.

The tolerable frequency for this consequence, taken from the company’s risk criteria, is 1 × 10⁻⁶ per year. The required risk reduction is:

RRF = mitigated frequency ÷ target frequency = 1 × 10⁻⁵ ÷ 1 × 10⁻⁶ = 10

So a SIF is needed with a risk reduction of at least 10. This puts it at the boundary of SIL 1. Where a result sits on the boundary between two bands, the conservative choice is the higher band.

The team could take this target forward to SIS design, or look for another genuine IPL instead. If another IPL is credited, it must really be independent. In this example, the alarm must not share a sensor or logic solver with the BPCS action.

What LOPA hands on

For each SIF it requires, LOPA should hand on:

  • the target SIL and required RRF;
  • the hazard the SIF protects against, and the scenario references;
  • the demand rate on the SIF;
  • the independent protection layers (IPLs) credited in the scenario, each with its full justification: the PFD claimed and its source, and why it meets the IPL criteria of being specific, effective, independent and auditable. Each IPL then has to be kept in place and tested for the risk reduction to hold.

These all become inputs to the safety requirements specification. If any are missing or vague, the problem is carried into the SRS.

Summary

LOPA brings some quantification to risk assessment after HAZOP, at reasonable cost. It works best with a clear procedure, a competent multi-discipline team and recorded assumptions. Its output, the required risk reduction for each SIF, is the integrity requirement that the rest of the lifecycle must deliver and keep delivering.


This chapter summarises the protection layer analysis and SIL determination lessons in our IChemE-accredited course SIS and Functional Safety for IEC 61511 (eLP003). SIFintegrity is our cloud-based software for running LOPA studies and carrying each SIL target through to the SRS and SIL verification. See also What is LOPA?