Choosing a method: legal duties, standards and the technique toolbox

There are dozens of risk assessment techniques. IEC 31010:2019, the international companion to ISO 31000, describes more than forty, from brainstorming and checklists to Monte Carlo simulation and Bayesian networks. No organization uses them all, and choosing the wrong one wastes time or, worse, gives false confidence. A full quantitative risk assessment of an office photocopier is a waste of money; a quick checklist for a new chlorine unloading station is negligent. This lesson explains what the law and the main standards require, how to sort techniques into families, and how to match the method to the question being asked.

What the law requires

In Canada, every provincial and territorial OHS law and the Canada Labour Code require employers to identify hazards and take every precaution reasonable in the circumstances. Some jurisdictions spell out a formal hazard assessment. Alberta's OHS Code Part 2 requires a written hazard assessment before work begins at a new work site, when a new process or equipment is introduced, and at reasonably practicable intervals, with workers involved and the results communicated. British Columbia's OHS Regulation requires risk assessments for specific hazards such as confined spaces, workplace violence and musculoskeletal injury, and Ontario's OHSA requires employers to assess the risk of workplace violence and to take every precaution reasonable. Federally, COHSR Part XIX requires a hazard prevention program with a hazard identification and assessment methodology, reviewed at least every three years. None of these laws names a particular technique; the employer chooses a method that suits the hazard.

Process safety: where methods are named

For facilities that handle large quantities of hazardous substances, Canada relies on a mix of federal environmental law, provincial OHS law and voluntary standards. The federal Environmental Emergency Regulations, 2019 (E2 Regulations) under the Canadian Environmental Protection Act require facilities storing listed substances above threshold quantities to notify Environment and Climate Change Canada and, above higher thresholds, to prepare and test an environmental emergency plan based on an assessment of potential incidents. CSA Z767 Process safety management, updated in 2024, sets out a full process safety management system including hazard identification and risk analysis, and many companies and some regulators reference it. The Chemical Institute of Canada's process safety community has also published guidance on risk assessment and land use planning. Canada has no single equivalent of the US PSM standard.

The main standards

CSA Z1002 Occupational health and safety: hazard identification and elimination and risk assessment and control is the main Canadian standard on this subject. It describes how to identify hazards, assess risk, choose controls using the hierarchy and review the results, and it includes informative annexes with example tools. CSA Z1000, the Canadian OHS management system standard, requires a risk assessment process as part of the system. Internationally, ISO 45001 clause 6.1.2 requires organizations to identify hazards and assess OHS risks with a methodology that is proactive rather than reactive.

Two international standards apply everywhere. ISO 31000:2018 gives the overall risk management principles and process. IEC 31010:2019 is the technique catalogue: for each method it describes the inputs, outputs, strengths and limitations, and it includes a table showing which techniques suit identification, analysis of consequences or likelihood, evaluation and recording. For process safety and instrumented systems, IEC 61511 (with IEC 61508 for manufacturers) adds LOPA and safety integrity levels, covered in Lesson 10.

Families of techniques

It helps to group methods into a few families. Qualitative methods describe risk in words or categories, such as high, medium and low. Semi-quantitative methods use numbers as scores or orders of magnitude, such as a 5 by 5 matrix or a LOPA that works in powers of ten. Quantitative methods estimate actual frequencies and consequences, such as individual risk of death per year or the number of people exposed to a toxic cloud. The more quantitative the method, the more data and expertise it needs, and the more its results can be misused if the inputs are weak.

How to choose

IEC 31010 suggests several questions. What is the purpose: finding hazards, ranking them, sizing a safeguard or proving a risk meets a numerical target? What stage is the project at: concept, detailed design, operation or decommissioning? What information exists: a sketch, piping and instrumentation diagrams (P&IDs), procedures, failure data? How severe could the worst outcome be? What resources, time and expertise are available? And what does the regulator or the company standard require? The answers usually point clearly to one or two methods.

A practical rule is to start simple and escalate. A team screens the activity with a checklist or what-if review and a risk matrix. Most risks fall clearly into the acceptable or clearly-needs-action categories and can be handled there. A few high-consequence scenarios, especially those involving multiple failures or relying on instrumented safeguards, are passed to a more detailed method such as LOPA or a fault tree. Only the highest-consequence, most complex cases need a full quantitative risk assessment. This approach spends effort where it matters.

Combining methods

Most mature organizations use a set of methods that feed one another. A HAZOP identifies a deviation, such as high pressure in a vessel, with a severe consequence. The matrix ranks it as high. LOPA checks whether the existing safeguards reduce its frequency enough and, if not, sets the required integrity of a new safety instrumented function. A bowtie then shows the barriers on one page for operators and managers, and the critical barriers become the subject of inspections and audits. Lesson 12 returns to how these pieces fit into a management system.

It is equally important to know what each method does not do. A HAZOP does not rank risk unless a matrix is attached. An FMEA looks at one failure at a time. A bowtie does not calculate anything. A risk matrix does not identify hazards; it only ranks the ones you have already found. Good practice is to state in the report which methods were used, why, and what their limits are, so that readers do not take more assurance from the study than it can give.

Proportionality in practice

Regulators in both countries expect the depth of an assessment to match the size of the risk. An inspector reviewing a small shop does not expect a fault tree, but will expect to see that hazards were identified with the workers, that the obvious controls are in place and that the assessment was reviewed after changes. An inspector or auditor at a major hazard facility will expect systematic methods, competent teams, documented recommendations and evidence that the critical safeguards are tested. A useful test is to ask whether a reasonable, informed person, reading the assessment after a serious incident, would agree that the effort matched the hazard.

Proportionality also applies within a single facility. A refinery may use HAZOP and LOPA for its process units, JHAs for maintenance tasks, checklists for office areas and a bowtie for each major accident hazard. Writing these choices into the company procedure, with the reasons, makes the approach consistent and defensible.