IEC 61511 GuideChapter 5 of 12

Safety requirements specification

The blueprint for the SIS

The safety requirements specification (SRS) is the blueprint for the safety instrumented system. It comes after hazard and risk analysis and SIL determination, and before design. But it should never be written once and left on a shelf.

The SRS must be maintained for the whole lifecycle: through design, installation and commissioning, and validation. It is the key document for SIS validation, and it must be updated as part of every SIS modification.

Requirements come in two kinds:

  • Functional requirements say what each safety instrumented function (SIF) must do: which hazardous condition it detects, and what action it takes to reach a safe state.
  • Integrity requirements set the standard the design must meet: the SIL or required risk reduction, and the availability needed to avoid spurious trips.

Why the SRS matters

The UK Health and Safety Executive studied incidents involving control systems. The findings were published as Out of Control: Why control systems go wrong and how to prevent failure. A significant share of the primary causes were traced to poor specification.

Specification errors are especially dangerous because they can go unnoticed for years. The SIS does exactly what it was specified to do, and the error only shows when a real demand occurs.

The SRS may also form part of the contract between the SIS owner and the companies providing engineering services. It can be a single document, but more often it is a set of documents covering different aspects of the system.

Qualities of a good requirement

Complete and precise requirements are hard to write. Every requirement should be:

  • Clear to its intended audience. State the goal, not the method of achieving it, because the method may change during design.
  • Precise: short, specific and not open to interpretation.
  • Verifiable by test, inspection or analysis. If a requirement cannot be verified, it cannot be validated.
  • Maintainable for the life of the SIS, which means thinking about proof testing while writing it.
  • Feasible, not only technically but within cost and schedule.

Compare two integrity requirements. A weak one says “the SIF shall be SIL rated”. A clear one states the demand rate and mode of operation (low demand, for example), the required RRF or PFDavg, and the risk receptor (safety, environment or asset). It also states the SIL to be met for hardware fault tolerance and systematic capability. Where the required risk reduction sits on the boundary between two SIL bands, a conservative specifier chooses the higher one.

What goes into an SRS

The detailed contents vary by project, but a typical SRS covers three levels.

General SIS requirements apply to the whole system, not to any one SIF:

  • environmental conditions;
  • equipment selection rules;
  • logic solver, operator interface and communications requirements;
  • event recording;
  • spurious trip rate;
  • manual shutdown;
  • equipment lifetime;
  • measures against common cause failure.

SIF data sheets, one for each SIF:

  • the SIF tag, description and the hazard it protects against;
  • references to the HAZOP, LOPA, P&IDs and cause-and-effect charts;
  • the safe state, and any hazards caused when several SIFs trip at once;
  • sources of demand and the demand rate;
  • the required SIL and PFDavg (or RRF), and the mode of operation;
  • the process safety time and the required SIF response time;
  • sensor voting, trip points, range and accuracy;
  • final element actions, and valve tight shut-off or leakage requirements;
  • reset, manual shutdown, bypass and override requirements;
  • the proof test interval, and whether testing is online or offline;
  • the maximum spurious trip rate, diagnostics and alarms, and repair time;
  • installation and survivability requirements.

Application program requirements for each logic solver:

  • which SIFs it supports, and their SIL and voting;
  • restrictions from the logic solver’s safety manual;
  • performance, such as scan time and capacity;
  • the action on bad or out-of-range signals: a frozen value, an open or short circuit, or incorrect valve feedback;
  • bypass and online testing functions;
  • how communications with the BPCS and HMI are kept from interfering with the SIFs;
  • output states on loss and return of power.

When several SIFs share one logic solver, its hardware and software must meet the highest SIL of any SIF it carries. SIFs can also share field equipment, such as one shutdown valve used by two SIFs. This is allowed, provided every demand on the shared equipment is assessed and taken into account in SIL verification.

Common SRS problems

The SRS depends on the quality of the information it receives. If the integrity levels from LOPA are poorly defined, those problems carry straight into the SRS.

Completeness is the next challenge. It is easy to miss a critical parameter. For example, if the process safety time never came out of the HAZOP or LOPA, the designer has no basis for the SIF response time.

To avoid problems during design:

  • keep safety and non-safety functions separate, which makes the SIS simpler to engineer and to test;
  • structure requirements as a hierarchy of smaller parts;
  • hold regular design reviews that inspect the SRS;
  • use templates, and checklists for completeness.

Keeping the SRS true in operation

Picture a technician about to proof test a SIF. They notice that the trip setting on the test procedure does not match the setting in the SRS. The test stops, and nobody knows which one is right. Was there an unauthorised change, or is the documentation simply out of date?

To prevent this, the as-built SIS must be checked thoroughly against the SRS and every document it references, including P&IDs, cause and effects and equipment data sheets. Field tags should be cross-checked. The logic solver program version should be recorded, ideally with a date, time and checksum, so that unauthorised changes can be detected.

Once the as-built status matches the documentation, the SRS becomes a controlled document that must be updated for every modification. Keeping requirements in a database rather than in separate files makes this much easier. It also keeps every requirement traceable back to the hazard and the LOPA scenario that created it.

Summary

The SRS turns hazard and risk analysis into requirements that can be designed, built, tested and maintained. Requirements should be clear, precise, verifiable, maintainable and feasible, organised in a hierarchy, and kept in step with the installed system for its whole life.


This chapter summarises the SIS safety requirements lessons in our IChemE-accredited course SIS and Functional Safety for IEC 61511 (eLP003). SIFintegrity keeps every SRS requirement linked to its LOPA scenario and SIL verification in one cloud-based record.