Human reliability

I think that one of the most influential statements about human reliability (and human factors more generally) is right at the start of the introduction to Barry Kirwan’s key text (1994, p.1):

“Human error is here to stay”. 

As industries, systems and technologies become ever more complex, human errors can have adverse effects on a larger population (such as errors in the control of nuclear power plants, or in air-traffic control). The challenge for human factors and safety professionals is to assess and manage the risks from human error. But what is human error?

In my work I try not to use the term ‘human error’ (not an easy task . . . ), but instead talk about Human Reliability – which could be seen as the opposite of human error.

There has been a great deal written about human error or human failures. In simple terms, there are two kinds of unintentional failures: physical errors (‘not doing what you meant to do’) and mental or cognitive errors, where you do the wrong thing believing it to be right (i.e. making the wrong decision). In addition, there are intentional failures (sometimes called violations or non-compliances): knowingly taking short cuts or not following known procedures (see my blog post on Preventing Violations).

In my experience, any discussion of ‘human error’ often leads to discussions around blame and culpability. The concepts of blame and just culture are discussed further elsewhere on this website; but at this point it is sufficient to state that Human Reliability Assessment should not be about assigning blame, but about learning.

“But we address human error!”

My experience is that although many organisations address human performance issues in their incident investigations, very few will proactively seek to understand potential human performance problems. This is effectively a human factors risk assessment and is central to improving the human performance aspects of a critical or complex system. I often describe these analyses as ‘pre-incident investigations’.

What are the main steps in Human Reliability Assessment?

“The starting point for the HRA is the identification and understanding of those human activities that are important to safety and how these may fail” (ONR, Technical Assessment Guide, 2013).

The overall HRA process is summarised in a document I produced for the UK HSE (the “7 Steps” approach to Identifying human failures), which also includes 22 key questions that I used to structure an inspection or audit of human reliability on a major hazard facility. This structure is well-established and has been successfully applied in numerous industries, including chemical, nuclear and rail. Other methods are available, but these tend to follow a similar structure to that described below. This approach is often referred to as a ‘human-HAZOP’.

The HRA process can be summarised as follows:

  1. consider the main site hazards;
  2. identify human activities that affect or interact with these hazards;
  3. outline the key steps in these activities;
  4. identify potential human failures in these steps;
  5. identify factors that make these failures more likely;
  6. manage the failures using hierarchy of control;
  7. improving error recovery and mitigation.

A paper presented at the Hazards symposium (Hazards XXI, 2009) provides practical (positive) experience of using this toolkit on a UK major hazard facility.

The importance of Performance Influencing Factors

I’m often asked “why do people behave or think like they do?”, or sometimes “why do smart people do dumb things?”. The answer to this million-dollar question is simple. It’s all about influences. Things to do with the Work (what people are doing), the Organisation (where they are doing it) and the People (who’s doing it) all influence human behaviour. These are known as Performance Influencing Factors, PIFs (or sometimes Performance Shaping Factors). These factors are the context within which human behaviour occurs.

“People’s actions are influenced by the organizations in which they work, shaping their choices in directions that even they may not realize” (NASA: Columbia Accident Investigation Board, 2003).

HumanFactors101 logoIf we can identify the factors around People, Work and Organisations that influence behaviour, we can manage them and optimise human performance. This will deliver benefits to individuals through health, well-being and safety; but will also improve the reliability of systems and processes, and reduce the likelihood or severity of major accidents.

Many of the other topics on this website (such as fatigue, workload, competence, supervision) are actually Performance Influencing Factors. Making them optimal is the key to human reliability. The blog post Human factors and Homer Simpson discusses how these Factors could influence how a control room operator responds to a process upset.

Human Factors 101 - Making it easy for Homer to do the right thing.

To quantify or not to quantify?

Qualitative assessments, such as that outlined above, aim to identify potential human failures and optimise the factors that may influence human performance. In addition to this, quantitative assessments aim to estimate the likelihood of such failures occurring. After having worked as a human reliability analyst for several years, I found various difficulties in quantifying human failures (for example, a lack of robust data regarding the factors that influence performance). I therefore tend to favour the qualitative approach, and this is the direction adopted by the UK HSE for oil, gas and chemical facilities. For some industries, such as nuclear power generation, quantitative HRAs is often applied and the results fed into traditional engineering risk assessment tools and methodologies, such as event and fault tree analysis.

More information on human reliability

Identifying human failures. Extract from UK HSE Inspectors Human Factors Toolkit. Outlines the ‘7-steps’ approach to qualitative human reliability assessment, and provides a series of test-yourself audit questions. Also includes a proforma for recording such assessments.

Human Reliability Assessment. Extract from HSE publication Reducing error and influencing behaviour (HSG48, ISBN 978 0 7176 2452 2, Second edition, published 1999). Contains useful an overview of the main steps in a HRA and discusses Human Error Probabilities .

Human Failure Types. This two-page document published by the UK HSE outlines the different types of human failures, their main characteristics, examples and typical control measures for each type of failure.

Performance Influencing Factors (PIFs). This is a list of characteristics of the People, the Work and the Organisation that influence human performance. Optimising PIFs will reduce the likelihood of human failure and increase human reliability.

Human Reliability Analysis, Office for Nuclear Regulation, (ONR, 2013). NS-TAST-GD-063 (Rev2). This Technical Assessment Guide is intended to support ONR inspectors (particularly Human Factors Specialist Inspectors) in assessing a nuclear licencee’s arrangements. It includes advice on the assessment of HRAs produced by nuclear facilities and so is a useful document against which to benchmark your arrangements. 

Guidance on human factors safety critical task analysis (2011). This document provides an outline to the SCTA process – which is essentially a Human Reliability Assessment. It includes case studies and describes what a good SCTA looks like. It is available to download for free by registering on the Energy Institute site. It is based upon the UK HSE’s ‘7-steps’ approach.

Human factors assessment of safety critical tasks (1999). Offshore Technology Report OTO 1999/092. At the time of this research project, it was recognised that human error accounts for a high proportion of risk in offshore operations, but safety cases provided a hardware-centred approach to the analysis of risks. The same is true today for many organisations. This research, by myself and colleagues, provided an approach that attempted to address that gap.

Evaluation report on OTO 1999-92 – safety critical tasks. HSE Research Report 33 (2002). The evaluation considered the usability of the methodology; the benefits of its use; typical costs to implement; deviations or modifications to the methodology; and opportunities for further development. The general finding of this evaluation is that this tool has many applications for both onshore and offshore installations.

Review of human reliability assessment methods, (2009). HSE Research Report 679. This document  provides a summary of those tools and methods considered to be of potential use to analysts undertaking a HRA assessment in the major hazard sector. It is not intended to be a comprehensive guide to HRA but a useful starting point on which to build knowledge.

A guide to practical human reliability assessment. Kirwan, B. (1994). Taylor and Francis, London. Barry was writing this book when his lectures introduced me to human reliability at The University of Birmingham, UK. Written over a four-year period, this is the seminal work on this topic; and in my library it ranks very closely with ‘Human Error’ by James Reason. It is now considerably more expensive second-hand than it was back in 1994 and has been a much better investment than the stock market. Oh, and if you like science fiction, Barry has also written several successful science fiction novels.

Graeme R Ellis and Andrew Holt (2009), A practical application of ‘human-HAZOP’ for critical procedures, Hazards XXI, IChemE.