The question HRA addresses

Human reliability analysis, or HRA, asks which actions could fail during a safety-relevant task, under what conditions and with what consequences. It may combine qualitative methods with quantitative estimates, but it is not a personality test or a ranking of reliable and unreliable employees.

Human and organisational factors help explain why the same person may perform differently depending on design, available information, workload or decision time. An assessment needs to describe that context rather than assign a fixed property to the individual.

Start with actual work

Before calculating, define the scenario, required function and opportunities for error and recovery. Observation and discussion with those doing the work can reveal differences from written procedures, including coordination or checks that the document does not describe.

For example, an alarm response assessment should not simply assume that someone will recognise the signal. It needs to examine whether it is distinguishable, which other demands coincide and what information supports selection of a response within the available time.

Select the method deliberately

HEART, THERP and CREAM represent different approaches. Selection depends on the purpose, available data, team expertise and required detail. A numerical result produced using a recognised technique does not remove the limitations of its assumptions or the need to explain them.

Dependencies between actions, recovery opportunities and uncertainty need explicit treatment. Two checks made using the same incorrect information are not automatically independent barriers merely because they occupy separate boxes in a control record.

Turn analysis into improvement

Useful findings identify changes to design, information, organisation or barriers that reduce error likelihood or consequences. Repeated training may be insufficient when the main problem is confusing controls, ambiguous signals or incompatible priorities.

The report should make the scenario, evidence, assumptions and decisions traceable. Following a modification, review whether expected performance has changed. HRA complements technical system assessment; it does not guarantee accident-free operation or provide a universal percentage for human error across every situation.