IEC 61511 GuideChapter 6 of 12

SIS design and SIL verification

Designing to the SRS

When the safety requirements specification (SRS) is complete and stable, design can begin. The handover should start with a review of the SRS against a checklist for completeness, clarity, ambiguity and traceability. Any requirement that is missing or unclear should be logged and tracked until it is resolved. A design built on a weak SRS will fail validation, or worse, pass it.

To meet its SIL, each safety instrumented function (SIF) must satisfy three separate requirements:

  1. Systematic capability: the equipment, and the way it is applied, is suitable for the SIL;
  2. Hardware fault tolerance (HFT): there is enough redundancy for the SIL;
  3. Random failure probability: the calculated PFDavg (or PFH) meets the target.

Showing that all three are met is called SIL verification. It is not something a basic process control system needs.

Selecting equipment: systematic capability

IEC 61511 requires every SIS device to be justified, either by prior use or by assessment against IEC 61508.

Prior use is the user’s documented judgement that a device is fit for use up to a limited SIL. It is based on operating experience with the same device in similar duty and environment, after poorly performing equipment has been removed from the data. It is easiest to justify for non-complex devices with no firmware or software, such as relays, solenoid valves, barriers, actuators and valves. Devices with fixed programming languages, such as most smart transmitters, need more evidence. Logic solvers that use limited variability languages need more evidence still, and prior use is not accepted for them at SIL 3.

IEC 61508 assessment is carried out by the manufacturer, or by an independent third party, usually called certification. Certification is not required by either standard. But it is strongly recommended for complex programmable devices at SIL 3, given the effort of justifying them any other way. A “SIL certified” label must never be the only justification: the device must still suit the application, the environment and the restrictions in its safety manual.

Failure: random and systematic

There are two kinds of failure:

  • Random failures come from degradation such as vibration, corrosion, temperature and electromagnetic interference. They can be estimated with failure rates.
  • Systematic failures come from human error in specification, design, software, installation or procedures. They cannot be calculated. They are controlled by functional safety management, verification and competence.

A failure rate is the number of failures in a population of devices divided by their total time in service. It is assumed to be constant during the device’s useful life, which excludes early-life failures and wear-out. So every calculation assumes equipment is replaced before it wears out.

Failures are split into four modes: safe or dangerous, and detected or undetected by diagnostics. The one that matters most for SIL verification is dangerous undetected (λDU): a hidden failure that stops the SIF working when a demand comes.

The failure mix varies between device types:

  • A SIL-capable transmitter usually has a lower proportion of dangerous undetected failures than a standard one, even if its total failure rate is the same or higher.
  • Safety PLCs have very high diagnostic coverage, so the logic solver is rarely the weak link.
  • Valves usually are the weak link. Without partial valve stroke testing (PVST), all their dangerous failures are undetected until the next proof test.

Hardware fault tolerance

HFT is the number of faults a subsystem can tolerate and still perform its safety function. One transmitter has HFT 0. Two transmitters voting one-out-of-two (1oo2) have HFT 1.

IEC 61511 sets a minimum HFT for each SIL:

  • HFT 0: SIL 1, and SIL 2 in low demand mode;
  • HFT 1: SIL 2 in high demand or continuous mode, and SIL 3;
  • HFT 2: SIL 4.

The requirement applies to each subsystem separately: sensors, logic solver and final elements. Redundancy brings its own risk of common cause failure, so redundant devices need to be diverse, separated or protected from shared stresses. That must be accounted for in the calculation.

Calculating PFDavg

For a simple single-channel (1oo1) device in low demand mode, a common approximation is:

PFDavg ≈ λDU × TI ÷ 2

where TI is the proof test interval. The PFDavg of a SIF is approximately the sum of the PFDavg of its sensor, logic solver and final element subsystems.

Worked example

A SIF needs SIL 2: a PFDavg below 0.01, or an RRF above 100. It has one transmitter, a safety PLC and one shutdown valve assembly. The dangerous undetected failure rates (per year) and the resulting PFDavg are:

Subsystem λDU per year PFDavg, 1-year test PFDavg, 3-year test
Transmitter 0.005 0.0025 0.0075
Safety PLC 0.001 0.0005 0.0015
Valve assembly 0.010 0.0050 0.0150
SIF total 0.008 (RRF 125) 0.024 (RRF 42)

With annual proof testing, the SIF reaches an RRF of 125 and meets SIL 2, with HFT 0 allowed in low demand mode. If the test interval is stretched to three years, the RRF falls to 42 and the SIF only reaches SIL 1.

Two things follow from this example:

  • The valve contributes most of the total, which is typical. A better valve, or PVST, is often the most effective improvement.
  • The proof test interval in the SRS is an input to the calculation. Deferring a test reduces the SIL actually achieved.

Real calculations must also include common cause failure for redundant subsystems, proof test coverage (a test that cannot find every dangerous failure), mission time, repair time and high-demand cases. In high demand or continuous mode, the measure is PFH (probability of dangerous failure per hour), not PFDavg. Every failure rate needs a stated source.

Application software

The logic solver application program is designed in this phase too, using the safety manual’s restrictions and the application program requirements in the SRS. It must be verified by review and testing before it is accepted. Keep safety logic simple, and separate it clearly from non-safety functions.

Summary

SIL verification shows that each SIF meets its SIL on three counts: systematic capability, hardware fault tolerance and PFDavg or PFH. The calculation is only as good as its failure data and its assumptions about proof testing. Those assumptions become operational requirements that must be met for the life of the SIS.


This chapter summarises the equipment selection, equipment failure and SIF design lessons in our IChemE-accredited course SIS and Functional Safety for IEC 61511 (eLP003). SIFintegrity carries out SIL verification with transparent PFD and PFH calculations and a stated source for every value. See also SIL verification calculation and What is hardware fault tolerance?