Abstract
Box culvert structures are widely used nowadays in infrastructures such as roads or bridges worldwide. The skewness in the box culvert bridges is usually inevitable to achieve the desirable design of their layout. Based on the limited knowledge in proposing a simple mathematical model to predict the load-carrying capacity of these skew structures, more studies are regarded essential in this direction. The work presented is dealing with the modeling of the skewed reinforced concrete (RC) box culverts using a valid finite element modeling in ANSYS-11. In addition, artificial neural network (ANN) was adopted to formulate a mathematical model for the response of these bridges depending on the outputs of parametric study of the present finite element simulation. Specific nonlinear relationships have been employed in the numerical analysis to model the behavior of both concrete and embedded steel bars. A database of the performance of RC skew culverts, established from the outputs of over 285 finite element models, was utilized for the development of ANN-based models. The skew angle of the culvert, the span of its top slab, thickness of the slab and concrete compressive strength have been used as input variables to predict the behavior of the culvert via the developed ANN. Simple equations were derived depending on the results of this network to compute the failure load and maximum deflection of the top slab of the skew culvert. The proposed models showed superior efficiency over existing tedious analytical or numerical solution with consideration of combined effect of simplicity and accuracy in the output prediction. Moreover, the comparison of the finite element outcomes with those of previous experimental work confirms the validity of the used constitutive relationships in the analysis of single-cell concrete culverts.
Published Version (
Free)
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call