Analysing accidents

In this account, we consider the use of empirical Bayes methods for rare events, as discussed in Quigley et al. (2007). Empirical Bayes methods have been applied in the fields of reliability (Sarhan 2003) and risk analysis (Martz et al. 1999) and are regularly used to analyse accident occurrence patterns in road safety applications (Persaud and Lyon 2007). Empirical Bayes models have been shown to perform well against full Bayes models (Camara andTsokos 1999), and even favourably when data is scarce (Vaurio and Jankala 2006, Srivastava and Kubokawa 2007). 

In actual practice, all of these controls are frequently used in combination. In performing system safety analyses, accident investigations, and design... 

The initial training includes safe work practices, the hazardous materials of the process and safety measures, emergency procedures, operating procedures and job hazard analyses, accidents and near misses, operating limits and actions and the results of deviation, and equipment and process parameters. 

In order to assess the efficiency of the safety design and the planned protective measures, the occurrence of accidents is assumed in analyses and the resulting plant responses are examined. The analysed accidents are characteristic with respect to the possible release of radioactive materials and to the loads they impose on components and systems. 

Another type of computer is the microcomputer which can be put on a desk and may be used by people with no specialised computer training. The processing capacity is much less than the main frame computer but is quite adequate for a single application, for example, recording and analysing accident information. 

Software. Any computer, main frame or micro, requires specially written instructions before it can carry out any task. These instructions, in the form of computer programmes, are known as software. Note that any computer can run a number of different programmes which means, for example, that a microcomputer normally used for analysing accident records can also be used for word processing, financial modelling and for keeping other types of record. 

Setting up the database so that it will do the recording and analysis required. If, for example, a general database is to be used to record and analyse accident and incident data it would be necessary to set up the fields for recording such things as name of person injured, time of... 

Identify what information (line drawings, calculations, test reports, operations procedures, other safety analyses, accident investigation reports, etc.) is needed to produce the desired results How easy is the information to obtain ... 

The information stored in the system safety portion of the safety knowledge management system should include Prior and current safety analyses Accident and near-miss histories Safety standards Identified hazards Known causes of identified hazards Proven hazard controls List of hazard consequences Hazard logs and risk registers Your hazard tracking system... 

III-31. A method for analysing accidents of this type is described in Ret [111-9] and supporting references. 

Streff, F.M. and Geller, E.S. 1988. An experimental test of risk compensation Between-subject versus within-subject analyses. Accident Analysis and Prevention, 20(4), 277-287. 

Care needs to be taken in combining individual events in analysing accidents to ensure that there is some rationale for the particular combinatioa A random combination of events may represent an extremely unlikely scenario that should be shown in the probabilistic safety analysis to be sufficiently rare as to be discounted rather than being taken as a postulated accident. In probabilistic safety analysis, an approach using best estimate analysis is adopted for severe accidents while conservatism should be applied in the analytical approach for postulated accidents that have a relatively higher likelihood of occurrence. 

The principles and methods presented in Part III are reactive in the sense that they are used in an approach where we learn from incidents and deviations after the occurrence. In Part V we will look into a proactive approach, where we identify and analyse accident risks before they have been manifested in actual occurrences. 

Computer support is not necessary when analysing accident data. For companies with only a few accidents and near accidents per year, paper-based archives will be adequate. 

Identification of critical behaviour by analysing accident reports, safety instructions, inspection reports etc. 

---