Human behaviour-based safety management in construction

Abstract

The Singapore government has been highlighting the importance workplace safety and health, of which the construction industry had been lacking, with relatively higher workplace incident statistics in the past years. Hence, an analysis to improve the safety performance was conducted. This paper applies the modified CREAM method to develop a general workplace, safety and health model for evaluating the safety performance of constructions workers. The original Cognitive Reliability and Error Analysis Method (CREAM) is a second-generation human reliability analysis method introduced to ¬¬assess the reliability of human actions in a context which includes human behaviour and the work conditions. The CREAM method utilises the common performance conditions (CPC) to assess the effects of circumstances faced by workers on their human reliability. The CREAM method can be adapted using the fuzzy theory and the Bayesian Network with consideration for uncertainty. The CPCs were adapted to fit the context of the construction industry, from which a questionnaire was designed accordingly. Forty-eight workers were surveyed and the data was analysed using the fuzzy Bayesian network. The efficacy of its results was determined by the results of twenty-two workers whose in-house company’s evaluation were provided, with comparison to another result that utilised the fuzzy hierarchical analytical process. Comparatively, the fuzzy belief network produced more conservative predictions. For both methods, adjustments will have to be made to reduce the confidence in worker’s technical experience.