Highway bridges in fire: characterisation of fire loading and structural behaviour

In bridge design, extreme hazards have been considered as design loads for years, including wind, earthquake, snow, and floods; but fire hazard is not usually considered in the design process. However, severe fire accidents occurring near or under bridges are not as rare as generally perceived compared to the other extreme hazards, especially earthquakes and floods. Therefore fire resistance of bridges along the most critical arteries of transport networks, carrying heavy traffic, should be considered. This should ideally be based upon an estimation of the consequences of a particular level of bridge damage in terms of social and economic costs. Since there are no codes or standards relating to fire resistance of bridges, assessment must rely upon a performance-based engineering approach. In conducting performance-based studies of bridge fire resistance, most previous researchers have used code-based fire curves, such as the ISO 834 standard or Hydrocarbon fires, which assume uniform heating along the entire bridge span. However, a real vehicle fire will naturally create a non-uniform, localised fire under the bridge span and the hazard intensity will decay with distance away from the burning vehicle. If such a scenario could be implemented in a more realistic fire model, then more realistic thermal and thermo-mechanical response of structures could be predicted, resulting in more reliable estimates of performance. This thesis consists of three main parts. Part I investigates the structural performance of composite steel-framed bridges and the influence of bridge shape on failure time under code-based Hydrocarbon fire loading. Part II uses the CFD-based fire dynamics simulation code FDS to generate design fire curves for four different classes of vehicles. The design fire curves include the expected decay in the intensity of the heat flux due to the fire along the bridge span. These curves were then generalised as mathematical functions that can be easily used by engineers and designers in the assessment of the performance of existing bridges under realistic hazard scenarios, for fire resistance design. Rectangular bridge models were subjected to the most extreme class of design fire (fuel tanker fires) in order to compare with the Hydrocarbon fire. The analysis showed that, for the bridge structure considered, there is no failure for the model in the fuel tanker fire scenario, even with conservative assumptions. However, failure may occur if a higher heat release rate is used, which is possible for large fuel tanker fires. In Part III the new design curves (developed as mathematical functions) were implemented into the OpenSees software framework to enable a seamless simulation from fire, to heat transfer and structural response