Analysis of categorical incident data and design for safety interventions using axiomatic design framework
MetadataShow full item record
Although analysing categorical data from incident investigation reports provides meaningful associations amongst causal factors of incidents, however, to date, no studies considered these associations in designing actionable interventions for safety improvement. We propose a methodology using descriptive analytics and axiomatic design framework. In this study, we have analysed injury, and ‘property-damage’ data, collected for 45 months from a large integrated steel plant. The data are analysed using the contingency table, Cramer's V, Phi coefficients (ϕ) and Fisher's exact test. The ‘wire-making division’ is the most injury-prone. Unsafe acts done by fellow workers are significantly causing injuries in ‘support services’, maintenance and ‘steel-making’. The property-damage cases are mostly reported in ‘steel-making division’, and caused by material-handling, crane-dashing, toxic-chemical, hot-metal and process-related incidents. It is also found that SOP inadequacy and non-compliance are significantly associated with ‘property-damage’ incidents. The key interventions from axiomatic design are as follows. For process-related incidents, regular inspection and maintenance of safety-critical equipment should be done. Safety-critical instrument and alarms can also be used to monitor safe operating limits of processes. Unsafe acts by fellow workers are the result of lack of coordination and communication. So, the management should identify and provide the types of safety training necessary to improve the same. The material-handling related problems can be handled through improved staff competency and communication. To address the SOP related issues, operating procedures should be reviewed, revised and communicated regularly.