A multiple defect approach to bridge life cycle modelling

Bridges are critical assets for the safe, reliable and functional operation of transportation networks. Infrastructure asset managers are responsible for ensuring that these bridges adhere to rigorous safety standards using the finite resources available to transportation agencies. Predicting future condition, forecasting required interventions and developing equitable resource allocations from limited budgets is a challenging task for bridge asset managers. To facilitate the decision making process at network level and to present decisions to stakeholders, it is common that a life cycle analysis is performed to evaluate the outcomes of different potential management strategies. An accurate bridge life cycle analysis is contingent on an appropriate bridge deterioration model being employed. In many jurisdictions, stochastic deterioration models are calibrated using condition records from visual bridge examinations, however, condition records typically report bridge condition on a single condition scale. In this thesis, defect specific condition scales are defined to incorporate multiple defect specific indicators in the modelling of deterioration. These additional indicators enable the modelling of the interactions between defects during bridge deterioration, with examples shown for bridge components constructed out of masonry and metal. The multiple defect deterioration models are presented as Dynamic Bayesian Networks, which are calibrated using condition records from railway bridges in the United Kingdom. Decision modelling in a life cycle analysis enables the evaluation of different asset management strategies. Typical specifications of intervention types in life cycle analysis models are often arbitrarily defined as qualitative actions, e.g. minor and major repair. However, the additional condition indicators developed in this thesis enable the modelling of targeted defect specific maintenance intervention types. Modelling defect specific interventions provides scope to quantitatively assess the effects of strategies that favour increased volumes of preventative maintenance. A Petri net model is used to perform a life cycle analysis, which incorporates a novel dynamic conditional approach to utilise the multiple condition indicators. The model is enhanced by incorporating intrinsic structural and material properties of bridge components alongside local factors such as coastal proximity