Construction defects and wind fragility assessment for metal roof failure: a Bayesian approach

Post-damage observations reveal that construction error is one of the major contributors to roof damage for houses subjected to extreme winds. In this study, a Bayesian approach was developed to probabilistically quantify the construction defect rates in roof connections, which enables a systematic integration of expert judgement, human reliability analysis (HRA) techniques and limited construction defect data. The reductions of uplift capacities for defective roof connections were also probabilistically modelled based on experimental evidence and engineering judgement. The developed construction defect model was incorporated in a reliability-based fragility method to assess the wind damage to metal roof cladding and timber roof trusses for contemporary houses in non-cyclonic regions of Australia. It was found that, the effects of construction defects are significant for the predicted roof cladding fragility, whereas for roof truss fragility, such effects are lower.