Hazard and risk analysis
Where the lifecycle starts
Every safety instrumented function (SIF) exists because a hazard analysis found a risk that was too high. Hazard and risk analysis is the first technical phase of the IEC 61511 safety lifecycle, and everything that follows depends on the quality of what it produces. If a hazard is missed here, no amount of good SIS design later will protect against it.
Owners of major accident hazard sites need to identify their hazards and make some grading of the risks. There are several ways to do this, but most companies use a combination of study methods, very often including a HAZOP (hazard and operability study).
Three steps from hazard to SIL target
Hazard and risk assessment that leads to SIL targets typically happens in three steps:
- Identify the hazards. What hazards are present, what harm could they cause, and what sequence of events could lead there? HAZOP is one technique for this; others, such as what-if studies and checklists, are also used.
- Estimate the risk. This is often a qualitative risk ranking, carried out during the HAZOP or separately afterwards.
- Allocate risk reduction to protection layers. Layer of protection analysis (LOPA) is a cost-effective way to bring some quantification to this step, and its result can be a SIL target for a SIF. The next chapter covers it in detail.
Steps 2 and 3 only work if the organisation has already set tolerable risk criteria: the frequency of each severity of consequence that it is prepared to accept. Without them, a team has nothing to measure its results against.
How a HAZOP works
The HAZOP technique was first developed by ICI in the UK in the 1960s. It can be applied at several stages of design, but it is most often used at detailed engineering, when the piping and instrumentation diagrams (P&IDs) are well developed.
A HAZOP must involve a multi-discipline team, led by a competent chairperson who can manage both the people and the process. The team works through the plant systematically:
- Node. The plant is split into sections, usually following the process flow. Each node is marked up on the drawings.
- Design intent. Those who know the process best explain the normal operating envelope, so that deviations are easy to recognise.
- Parameter and deviation. A guideword is applied to a parameter, for example more or less applied to pressure.
- Causes and consequences. If the deviation is credible, the team looks for the human errors, equipment failures or external events that could cause it, and the harm that could follow to people, the environment or assets. Consequences can be delayed, and can happen outside the node being studied.
- Safeguards. The team records the safeguards that exist on the current drawings, such as relief valves, trips and alarms with operator response. Proposed safeguards are recorded as recommendations, not as existing safeguards.
- Recommendations. Everything important is recorded, and actions are raised.
This repeats until every relevant deviation has been applied to every node. A HAZOP only adds value if its actions are closed out afterwards, with a written justification for any recommendation that is not adopted. Actions that reduce risk should take priority over those that improve operations.
Risk ranking with a risk matrix
Many companies rank risk during the HAZOP using a risk matrix. The team makes a qualitative judgement of how likely each cause is, and how severe the consequence would be.
The words on each axis need clear definitions tied to experience. For example:
- possible might mean “has been known to occur at this site”, while remote means “never experienced at this site, but may have occurred elsewhere”;
- serious might mean “reversible injury leading to long-term absence from work”, while major means “permanent disablement or a single fatality”.
Where frequency and consequence meet, the matrix shows whether the risk is low, medium or high. Low risks can be accepted as they are. High risks need more detailed analysis, usually LOPA.
Consequences should first be judged without any safeguards in place. Otherwise, the team is in danger of crediting the same safeguard twice: once in the HAZOP and again in the LOPA.
What good hazard analysis hands on
A well-run hazard analysis gives the next phases of the lifecycle what they need:
- Scenarios with a clear initiating cause and a specific consequence, grouped where one consequence has several causes;
- Existing safeguards, with enough detail to judge whether each could be an independent protection layer (IPL);
- The process safety time: how long it takes from the initiating event to the hazardous outcome. The SIS designer will need this later to set the SIF response time, and it is often missing;
- Recommendations, including candidate SIFs.
In some cases, especially for very high severity scenarios, a quantified risk assessment (QRA) may be needed instead of LOPA. A QRA can properly estimate the size of the area the hazard affects and how many people are exposed.
Summary
Hazard and risk analysis identifies what can go wrong and how bad it could be. HAZOP gives a systematic, team-based way to find hazards, and a calibrated risk matrix filters out the lower risks so that effort goes into the scenarios that matter. The output is a set of well-defined scenarios ready for LOPA and SIL determination.
This chapter summarises the process hazard analysis lessons in our IChemE-accredited e-learning course SIS and Functional Safety for IEC 61511 (eLP003). See also What is HAZOP?