Classifying bridges for the risk of fire hazard via competitive machine learning

Abstract

This study presents a machine learning (ML) approach to identify vulnerability of bridges to fire hazard. For developing this ML approach, data on a series of bridge fires was first collected and then analyzed through three algorithms; Random forest (RF), Support vector machine (SVM) and Generalize additive model (GAM), competing to yield the highest accuracy. As part of this analysis, 80 steel bridges and 38 concrete bridges were assessed. The outcome of this analysis shows that the ML based proposed approach can be effectively applied to arrive at the risk based classification of bridges from a fire hazard point of view. In addition, the developed ML algorithms are also capable of identifying the most critical features that govern bridges vulnerability to fire hazard. In parallel, this study showcases the potential of integrating ML into structural engineering applications as a supporting tool for analysis (i.e. in lieu of experimental tests, advanced simulations, and analytical approaches). This work emphasizes the need to compile data on bridge fires from around the world into a centralized and open source database to accelerate the integration of ML in to fire hazard evaluation.

Probabilistic evaluation of the axis distance’s influence on the flexural strength deterioration of reinforced concrete beams under ISO 834 fire

“…Fu [15] developed a ML-framework specially designed for assessing the progressive collapse resistance of steel frame structures under fire, showing that the neural network-based model achieved better results than the counterparts given a sufficiently large dataset. Kodur et al [16] explored a data-driven method to assess the fire hazard of bridges based on their geometric configurations (span, number of lanes), materials, and current operation states (damage, age). It was shown that the proposed method could be used as a low-budget tool to assess the fire vulnerability of bridges with similar patterns.…”

“…As pointed out by the aforementioned studies [10][11][12][13][14][15][16], a common major obstacle to the data-driven method is the scarcity of relevant data; this problem is more accentuated when studying structural members under fire compared to ambient condition. Even with data in hand, there exist unavoidable deviations between them because experiments and simulations were carried out by different authors in various conditions.…”

Prediction of Bridge Fires Characteristics Using Machine Learning

“…In latest years multiple bridge fires have occurred with some resulting in major destruction or bridge failure [3]. Since most bridge fires are triggered by vehicle collisions with other vehicles or with bridge structural elements, bridge fires can be destructive [5][6][7]. This has been because of high-speed crashes resulting in the combustion of highly flammable hydrocarbon-based fuels.…”

“…This has been because of high-speed crashes resulting in the combustion of highly flammable hydrocarbon-based fuels. Fires can cause severe strength deterioration in structural elements due to the lack of load-carrying capacity of components, which can lead to partial or complete bridge failure [6,7]. Except in small fire cases where the bridge does not collapse, proper review, inspection, and repairs are expected before reopening.…”

“…Bridge structural members are currently configured for seismic loads not for fire hazards in usual approaches of design and construction. Furthermore, Wardhaua and Hadipriono [6], and Scheer [7], showed that fire caused 3.2% and 4.9% of the overall bridge population to collapse, respectively [8]. However, there are no provisions in existing codes and guidelines for structural fire construction for bridges [4,9].…”