Using Bayesian Network to Develop an Approach for Construction Safety Risk Assessment

Risk assessment for construction safety is critical for effective project control. Traditional risk assessment approaches mostly rely on expert opinion because of limited accident data. Therefore, assessment results may lack precision because of experts‟ subjectivity and limited cognition when multiple safety risks are assessed simultaneously. This study developed a new assessment approach based on the Bayesian Network for construction safety risks. Both expert opinion and real data were incorporated by this approach. The approach has been developed for major safety risk events (Scaffolding Collapse, Fire Disaster, Falling, Machinery Injury, Electric Shock and Strike By/Against) based on expert opinion from construction companies and real accident data in Beijing. Risk probability analysis was based on both subjective and objective information, while risk influence analysis was based on objective information from real accident data. The assessment results could be updated easily as new data was collected. Using this approach, project managers could comprehensively assess risks or understand the causes of accidents where there were complex interrelationships between safety factors and events.