Prediction of mechanical behavior of reinforced concrete bridges following a fire based on the fire dynamics simulator-kriging model

The performance degradation and degree of damage of bridge materials are more extensive at the fire source compared to other bridge locations. The traditional finite-element model (FEM) modification method leads to a large error and is unable to achieve a good correction effect. In addition, as the traditional static load test method is used to evaluate the bearing capacities of bridges postfire, the risk of bridge damage, or collapse-induced by secondary loading-remains unaddressed. To solve these problems, this paper proposes a modified FEM of a bridge following a fire based on the fire dynamics simulator (FDS)-artificial intelligence algorithm. The static behavior of a bridge is predicted using the modified FEM. First, a numerical simulation of a bridge fire is conducted using the FDS to obtain the temperature field distribution and determine the influence ranges of local design parameters of the bridge at the fire source. Second, a response surface model with a high fitting accuracy is constructed to modify the initial FEM by using the artificial intelligence algorithm proxy model together with dynamic load test results for the bridge postfire. Finally, the static behavior is predicted using the updated model. In this study, an actual hollow girder bridge postfire is taken as the research object, and the static behavior prediction results obtained via the proposed method are compared with the static loading test results. The results indicate that the response-predicted value of the modified FEM is close to the measured value of the bridge postfire.

Automated summary

This paper proposes a modified finite-element model (FEM) of a bridge following a fire based on the fire dynamics simulator (FDS)-artificial intelligence algorithm. The static behavior of the bridge is predicted using the updated model, which is obtained through numerical simulation and dynamic load test results.