Human Error and the Implications for Industry


Table of Contents

Introduction: does human error cause accidents?
What is human error?
What types of errors do humans make?
How errors cause accidents.
What factors influence human reliability?
Human variability.
The effect of stress.
The Consequences of Human Error.
The Prevention of Human Error.

Motivational Campaigns.
Increased Discipline.
Safety Audits.
Increased Automation.
Improved Training.
Redesign the Job, the Equipment, or the Procedures.

The Management Issues in Prevention.
Collecting Data.
References:


Human Error and the Implications for Industry.

Andrew J. Brazier.

A literature survey was carried out to find the current knowledge of human errors with reference to safety in industry. This paper covers the types of errors, their consequences, the factors that effect human reliability, and how to prevent accidents. The main message is that safety is the responsibility of every person and must start with management.

Introduction: does human error cause accidents?

Baron [1] states that, on average, about 80% of industrial accidents are caused wholly or partly by human actions. Dhillon [2] collected the following data of failures due to human error in different organisations.

Further to this, 20% of defects are missed by inspectors who are specifically looking for them and the U.S. Strategic Air Command claim that 16% of all critical events are made worse by human intervention.

The British Health and safety executive (HSE [20]) made a study of 135 vessel failures between 1926 and 1988. They collected information about the causes of failure from reports in journals and from data bases.

They found that 24.5% of failures were caused directly by human error. The main errors included leaving valves in wrong status, filling the wrong vessel, poor procedures and verbal instruction, unsafe vessels thought to be safe, poor permit to work systems, and the most frequent was simple overfilling.

They found the total human contribution was 32.8%. The further contributions were through repeated overfilling, vehicle impact, vessel hitting objects or being dropped when moved, layers of reactive chemicals when agitator not used, wrong material fed to vessel, and incorrect installation.

A similar survey was also carried concerning the failure of valves in nuclear reactors. It was found that human error was responsible for 47.4% of the failures in Boiling Water Reactors and 45.7% in Pressurised Water Reactors. The main causes of failure were design and maintenance errors. Administration, fabrication, installation, and operator errors were the other human causes of valve failure.

All this evidence shows that human error is a major cause of accidents. It should also be realised that the information used relies on the reporting of incidents. It would seem likely that not all incidents are reported. Even when incidents are reported it not always possible to find the real cause, errors in design or fabrication may not always be apparent even after a major incident.

If we are to reduce accident rates, in order to prevent injury and property damage, reducing the risk and the consequences of human error is an important task.

What is human error?

Human error is defined by Dhillon [11] as, "A failure to perform a prescribed task or the performance of a forbidden actions." The consequences include serious injury and property damage, less serious injury or damage, and events with no real effect, these can be considered as "Near miss incidents." (Knox [21])

The role of humans in engineering systems include design and maintenance. They all make errors and the consequences will be similar. The root cause of errors that occur may be less obvious.

Managers also make errors. The consequences in their case are different, they allow error to appear in the engineering system and operation by not supervising their personnel correctly.

What types of errors do humans make?

It is important to realise that humans will make errors no matter what their level is of skill, experience, or training (Kim [4]). The working environment effects human reliability and is greatly influenced by equipment design and management policies. A knowledge of the types of errors that will be made is required if accidents are to be prevented.

Norman [5] suggests two ways that human errors occur, mistakes and slips. A mistake is a failure in the planning of a task, a slip is due to the execution of a task, a lapse in concentration. "If the intention is not appropriate, this is a mistake. If the action is not what was intended, this is a slip." Kletz[6] adds violations (a decision is made not to do something) and mismatches (someone is not able to do something), as further types of human error.

Kontogiannis et. al. [7] have summarised the types of errors that humans make. The possibilities can be summarised as doing nothing, doing something wrong, or doing something right but in the wrong place. The main possibilities are:

(ACE safety services [8]) list the following as unsafe acts that are likely to cause accidents:

How errors cause accidents.

A major equipment failure or a serious human error during operation can cause an accident. Most of these failures are foreseen in design, investigated and safety devices or procedures are developed before operation commences. Most accidents, however, progress from a series of relatively minor incidents that defeat safety systems. (Lewis et. al. [9]). Good design is vital to prevent chains of events that cause such accidents.

When talking about human error, operators are usually considered to be the main factor. Humans, however, have a much greater influence due to "Resident errors" in the engineering system. These errors have very serious implications and should be investigated in great detail when considering reliability and safety.

HSE [3] suggest the following resident errors:

These resident errors will wait in the system and not show until some event, during operation, triggers them.

These local triggers include:

A lot of accidents occur during unusual operation such as start-up, shut-down, maintenance, or testing. The resident errors remain the same but unsteady conditions increase the number of local triggers present.

What factors influence human reliability?

Collecting data has allowed theorists to categorise errors and use this information to provide clues as to the processes involved in routine human action (Reason [10]). For practical application it is important to understand the errors personnel are likely to make. Steps can be taken to eliminate them or, if this is not possible, to minimise the consequences.

Humans can make errors at any time. The lack of precision in physical work or inappropriate timing of actions that all people show at sometime means that something unpredictable will happen (Human reliability associates [11]).

Whilliams [12] has summarised the main causes of human error:

Human variability.

A lot of design work assumes good engineering practice will be followed by all humans with any influence to the system. The problem is that this cannot be guaranteed. Human performance depends on a lot of factors which means they perform differently in different situations. Lee et. al. [14] list the principle factors that will effect a persons performance as:

These factors should be taken into account when considering safety. What is the possibility that good engineering practice will not be followed?

It is also important to consider these factors when selecting personnel or setting up teams.

The effect of stress.

Stress is the demand placed on a persons physical or mental energy. It effects the way people react and is important when considering human performance and reliability. It can cause errors and effect the response to incidents when they occur. Dhillon et. al. [13] and Papazoglou et. al. [14] list the factors at work that cause stress:

It is important to remember that an incident occurring will probably increase stress. This is likely to effect human reaction times and reliability at a critical time.

The Consequences of Human Error.

The consequences of human errors being made range from serious accidents through to an event with no apparent lasting effect. Serious accidents tend to be investigated whereas less serious incidents may not even be reported. It should be remembered that any incident has the potential to lead to something more serious. Knox [21] has represented the number of incidents compared to their seriousness as a pyramid.

Time lost injury
Non-disabling injury
Property damage and financial loss.
Unreported incidents or near misses.

Of course the consequences of human error depends on who makes the error, and what error has been made. If an operator makes a mistake or slip, the consequence will either be that the equipment required will not operate or its operation will be delayed. Latent errors are much more important. Design errors may mean that the system will not respond to multiple failures or that a series of events will prevent a safe state from being reached. (Brazendale [16]). Design is also important on assuring that operator stress is kept to a minimum. Fabrication and maintenance errors will lead to poor reliability. Management errors can lead to a generally poor attitude to safety which will spread to all areas of the company.

The Prevention of Human Error.

It has been suggested that human error causes up 80% of all accidents (Baron [15]). If we eliminate human error or at least reduce the consequences, we will have gone a long way to preventing accidents. To improve human reliability we need to understand what affects it. Unfortunately, operators are usually blamed for making errors that cause accidents. Human's will always make errors, the reason that accidents happen is that latent errors are present in all systems which do not give the operator a chance, accidents are waiting to happen. These latent errors are the root cause of most accidents. The way to reduce human error and prevent accidents is through effective management such that safety is considered important by everyone involved.

Lucas [3] has produced a list of ways to reduce human error.

Motivational Campaigns.

This is a system where some-sort of reward is offered for operating in a safe manner. This usually involves analysing accident rates. The problem with this is that people are effectively punished for the accidents they are involved in thus, rather than reducing accidents, the number of incidents reported is reduced.

Increased Discipline.

Here people are punished for the accidents they are involved in. Once again this is more likely to reduce reporting rather than necessarily the number of accidents. It also requires placing the blame on certain people. Although an operator might have been directly involved, the accident is more likely to have been caused by latent human errors which may not be so obvious and are out of their control.

Of course it is usually the management of a company that would decide who to punish, blaming an operator is a lot easier than blaming a manager who is really more responsible by allowing latent errors to be present in the system.

Safety Audits.

These will generally be regular checks made by independent assessors covering a wide range of safety features. They provide a good indication of the obvious problem areas and hazards. The problem is that they will generally be limited to a check list of items thus the problem of identifying critical factors may not be fulfilled.

The timing of safety checks is important to maintain system reliability effectively. (Lewis [9]). Thorough start-up checks and verifications have been shown to be more effective than inspections during normal operation.

Increased Automation.

Here human manual control is replaced by automatic controls, generally electronic devices. These devices will do as instructed without the problems of human variability and unpredictabilty. The problem is that automation will generally make the system more complex and introduce more latent error, accidents waiting to happen. With a high amount of automation, the human has different tasks to perform. They are basically there to deal with unforeseen circumstances or to perform tasks that the designer cannot automate. (Brazendale [16]). This generally means that during normal operation there is less to be done. This leads to deterioration in skill, reduced attention, and a loss of understanding of what is actually happening. Operators probably need more training to operate automated plant although it would appear they have less to do than on a manually operated plant.

Automation can be useful but must be designed correctly. The information should match the operators own mental idea of what is happening. The information should be useful. Important information should be given priority, cross checking should be possible to validate information, alarm analysis and decision aids should be included.

Improved Training.

Training is very important in the effort to reduce human errors and hence accidents. Safety training is vital for everybody involved in the system. Operator training will not, however, improve reliability when the root cause is bad design or poor management.

Training given should be well planned and appropriate to the job. Realistic simulation and role-play exercises are some of the best ways to train people. Everybody must be familiar with the system and made aware of the risks involved and how their actions effect safety. Training should cover the use of all job aids including procedures, and other ancillary and emergency equipment. Recovery procedures should be explained for use after errors have been made.

Personnel performance checks and evaluations should be used and good, constructive feedback given at regular intervals. Refresher training should also be used to prevent behaviour patterns building up such that variations of equipment and procedures can not be handled. (HSE [20]).

Redesign the Job, the Equipment, or the Procedures.

This really is the main way to reduce accidents caused by human error. A more simple job will reduce confusion, reduce stress and mistakes will be less likely. Human errors will still occur but improved equipment design minimise the consequences. The ability to control dangerous situations, however, is vital. This requires plenty of useful, easy to interpret, information.

Procedures are very important and should be well thought out.

The Management Issues in Prevention.

I think I have shown that human error does account for a large number of accidents, however it is the latent errors that are the real root cause. It is far too easy to blame operators for causing accidents but it must be appreciated that all humans will make errors. It is the job of the management to ensure that safety is a priority. The responsibility starts at the very top, with the managing director, and must work its way to all levels in the company. That way the company culture will support the concept that safety comes first.

Whalley and Lihou [17] have summarised the common characteristics observed at companies with low accident rates:

Managers are subject to a variety of factors, like any person, that will effect their performance and which can ultimately effect safety. They are subject to all the stresses that effects every persons reliability, along with the extra pressures of management such as:

It is important to remember that all managers personalities, management styles and judgement will have far reaching effects where safety is concerned.

Collecting Data.

The problem with assessing human error tends to be collecting useful data. "Data Banks" do exist but many people consider the information available in them to be of little practical use (Lee et. al. [14]). The problem is that once a method for collecting data has been formulised, information starts to be lost. It is also very difficult to validate the information contained.

An effective system for collecting data (Dhillon [10]) must:

Collection can be made through:

At present, a lot of time is spent investigating major accidents. It has been pointed out that major accidents do not tend to recur, they are usually the worst case situations involving a chain of minor events. It should be remembered that any incident has potential to develop into a major event. As has been said already, for every major accident, there will probably be hundreds of minor incidents, many of which will probably not be reported. Investigation time would be better spent looking at all incidents and thoroughly investigating the ones with the serious consequences even if a major accident was avoided at the time.

HSE [20] suggest the following factors are essential if accident reporting is to be of any use:

With a good, company wide policy for safety, the chances are that all incidents will be reported, more information will be available and it is likely to be far more useful for industry than data collected by other methods.

References:

[1] Kontogiannis, T. and Embrey, D. (1992). Human Reliability Assessment.Human reliability Associates' course, Practical Techniques for Assessing and Reducing Human Error in Industry. (sec14)

[2] Kletz, T. (1991). An Engineer's View of Human Error. IChemE.

[3]Lucas, D. (1992). Tackling the Problem of Human Error. Human reliability Associates' course, Practical Techniques for Assessing and Reducing Human Error in Industry. (sec3).

[4] Practical Application of Human Error Concepts. Human reliability Associates' course, Practical Techniques for Assessing and Reducing Human Error in Industry. (sec7)

[5] Papazoglou, IA. Nivolianitou, Z, Aneziris, O, Christou, M. (1992). Probabilistic Safety Analysis in Chemical Installations. Journal of Loss Prevention in the Process Industries (Vol.5, No.3).

[6] Concise Oxford Dictionary.

[7] Dhillon, BS. and Yang, N. (1992). Reliability and Availability Analysis of Warm Standby Systems With Common-Cause Failures and Human Errors. Microelectronics and Reliability (vol.32, No.4).

[8] ACE Safety Services, Consultant in Safety Training, Advice & Protective Equipment.

[9] Lewis, CM. and Stine, WW. (1989). Hidden Dependence in Human Errors. IEEE Transactions on Reliability. (vol.38, No.3, pp 296-300)

[10] Dhillon, BS. (1990). Human Error Data Banks. Microelectronics and Reliability. (vol.30, No.5, pp.963-971).

[11] Dhillon, BS. and Rayapati, SN. (1988). Human Performance Reliability Modelling. Microelectronics and Reliability. (vol.28, No.4, pp.573-580)

[12] Kim, K. (1989). Human Reliability Model With Probabilistic Learning in Continuous Time Domain. Microelectronics and reliability. (Vol.29, No.5, pp.801-811)

[13] Dhillon, BS. (1989). Human Errors: A Review. Microelectronics and reliability. (Vol.29, No.3, pp.299-304)

[14] Lee, KW. Tillman, FA. Higgins, JJ. (1988). A Literature Survey of the Human Reliability Component in a Man-Machine System. IEEE Transactions on Reliability. (Vol.37, No.1, pp.24-34)

[15] Baron, RG. (1988). Human Factors in the Process Industries. Human Factors and Decision Making: Their Influence on Safety and Reliability, Symposium for the Safety and Reliability Society. ed. Sayers, BA. (pp.1-9)

[16] Brazendale, J. (1988). Allocation of Function Between Man and Programmable Electronic Systems in Safety-Related Applications. Human Factors and Decision Making: Their Influence on Safety and Reliability, Symposium for the Safety and Reliability Society. ed. Sayers, BA. (pp.51-70)

[17] Whalley, S. and Lihou, D. (1988). Management Factors and System Safety. Human Factors and Decision Making: Their Influence on Safety and Reliability, Symposium for the Safety and Reliability Society. ed. Sayers, BA. (pp.172-188)

[18] Williams, JC. (1988). A Human Factors Data-Bank to Influence Safety and Reliability. Human Factors and Decision Making: Their Influence on Safety and Reliability, Symposium for the Safety and Reliability Society. ed. Sayers, BA. (pp.223-240)

[19] Homewood, S. and White, S. (1989). Zeebrugge- a Hero's Story. Bloomsbury.

[20] HSE Contract Report No. 33/1992. Organisational Management and Human Factors in Quantitative Risk Assessment, Report 1.

[21] Knox, BJ. (1990). Safety Standards - a Time For Change. Piper Alpha, Lessons foe Life-cycle Safety Management. (I.Chem.E Symposium Series No. 122, pp.77-81).

[22] Reason, J. (1990). Human Error. Cambridge University Press.

[23] Norman, DA. (1983). Position Paper on Human Error. NATO Advanced Research Workshop on Human Error, Bellagio, Italy.

[24] Wilson, KC. (1990). Process Control, Plant Safety and the Human Factor: Lessons to be Learned from a Major Refinery Incident. Transactions IChemE. (Vol.68, Part B, pp.31-36)

[25] Aplleton, B. (1992). International Management of Safety Course. Video of talk given at ICI.

[25] Tweeddale, HM. (1992). Balancing Quantitative and Non-Quantitative Risk Assessment. Process Safety and Environmental Protection. (Vol.70, Part B, pp.70-74)

[26] Galagher, AJ, and Tweeddale, HM. (1992). A Development in Auditing of Master Hazard Plants. Process Safety and Environmental Protection. (vol.70,part B1, pp. 18-21)