The desire for continuous improvement in safety performance has led the process industry to a situation where the main contribution to accident causation is the actions of people rather than equipment failure. Models of human behaviour and accident causation, and risk assessment techniques aim to improve safety by reducing human error rates. These models require appropriate data and this thesis examines sources of information that could be used to provide accurate data for use in human factors studies. Accident reporting systems are widely used by the process industry to record events resulting in loss. A survey of the systems used by companies has been carried out. This found that some of the information recorded in accident reports was relevant to human factors studies although it was generally limited to details of the behaviour of people "at the sharp end". Little consideration had been given to the actions of people working away from the plant or of the factors that affect human performance. Near miss reporting systems are now used by most companies in the process industry to increase the number of incidents from which they can learn about their safety performance. Most systems lack maturity and at present the provision of data for use in human factors studies is poor. This thesis describes studies carried out to determine the potential of near miss reporting systems to provide appropriate data. It was found that people find it difficult to determine what events and consequences might have happened because there is a lack of evidence. Simple risk assessment based on what people do, the hazards involved and overall unit objectives has been used to provide the required evidence. This has resulted in more effective human factors assessment. Near miss reporting has great potential to provide data for use in human factors studies but it should be considered as a living risk assessment exercise rather than an extension to accident reporting.
|University of Edinburgh
|Electronic Thesis or Dissertation
Page generated in 0.0023 seconds