عنوان مقاله [English]
In recent years, great interest has been shown towards the effective modeling of masonry arch bridges. However, the issue of an efficient model is a controversy among researchers with contrasting strategies. The fact that there are a great number of stone arch bridges in Iran (about 3300), most of which serve the railway network, makes the issue very crucial in terms of road network vulnerability, due to their unknown behavior against usual and unusual loads. In this paper, a nonlinear 3D finite element method is employed in order to determine the ultimate failure load of stone arch bridges. Most of these bridges are composed of three structural parts; arch barrel, backfill and spandrel walls. However, because of simplification in modeling masonry arch bridges in some research, spandrel walls are neglected. An efficient description of the material properties of these parts has great influence on the accuracy of the resulting ultimate failure load. Experience has shown that the elastic modeling of these systems could not yield a reasonable behavior.
Also, even if only nonlinear models of different contacts are used, in some cases, the analysis results would not be satisfactory. It is understood that accurate results could be achieved even by simple Mohr-Coulomb models for the barrel arch and spandrel walls. At the same time, the Drucker-Prager material law for the backfill, along with appropriate modeling of the contact surfaces of different materials, should be used. It is shown that hardening stiffness in the pressure over-closure of hard contacts should not be neglected. In addition, the important role of spandrel walls has to be accounted for in a 3D analysis model. Indeed, the former improves the bridges failure mechanism, whereas the latter restrains the bridges lateral deformations. To validate the proposed model, an actual failure test on the Prestwood Bridge is considered.
According to the actual failure test, the proposed model is subjected to a load at the quarter point of the bridge span. Despite other models, the aforementioned modeling strategy could yield a similar collapse mechanism to
that of the prototype, and the ultimate failure load is achieved with only 1.3 percent error, in respect to the experimental one.