System reliability-based bridge assessment using response surface methods
The objective of the present research was to develop a system reliability-based
bridge assessment method for damaged composite bridges. A response surface
method was adopted in combination with the nonlinear finite element analyses,
which provided a powerful tool for the evaluation of the reliability of a bridge
system. Using the method, investigation was made into the effects of corrosion on
the reliability of a bridge system.
A numerical bridge model used for the present study was developed using an actual
composite bridge in the UK. Commercial FE programmes such as ABAQUS and
DIANA were used for the development of the model. In order to validate the use of
these FE programmes in the present study, simulations on the full-scale bridge test
were carried out. Of the three, 40-tonne vehicle loading given in BD 21, the vehicle
loading which caused the worst resistance was selected as a reference loading model.
In the present study, the failure of the load redistribution system due to punching
shear was addressed in detail. It was found that punching shear in the concrete slab
may prevent the total collapse of a whole bridge system taking place. This is because
the system cannot reach the ultimate state in such a case. It was proposed that such
cases be excluded from the evaluation of the failure probability of a whole system.
This approach is expected to provide a more rational basis for the evaluation of the
reliability of a global bridge system than simply suppressing such failure.
The system reliability-based assessment was developed based on the response
surface method in combination with the nonlinear finite element analyses. In the
present study, reliability analyses were carried out for both intact and damaged bridges. Corrosion on steel girders was simulated by reducing the thickness of the
web and the flange. The present study shows that the two-lane loading case governs
the total failure probability (reliability) of the present bridge model. In addition, it
was found that the traffic loading for the evaluation of load effects has a significant
influence on the results. It was also found that the failure of a bridge is most likely to
take place due to extremely heavy trucks for the intact state. It seems that this has a
close connection to the high uncertainty of the traffic loading. However, the failure is
likely to be governed by reduced dimension for severely damaged model. This is
because the uncertainty related to the dimension and ultimately the resistance
becomes bigger as corrosion proceeds.
Results of this research revealed that this assessment methodology is less
conservative. This may lead to rational use of a limited resources. It was clearly
demonstrated through the reliability analyses that a system reliability-based bridge
assessment methodology developed in this research may provide a tool for more
rational bridge assessment.