PSI - Issue 62

Tommaso Lazzarin et al. / Procedia Structural Integrity 62 (2024) 625–632 Lazzarin et al./ Structural Integrity Procedia 00 (2019) 000 – 000

626

2

1. Introduction Modelling river flow at bridges is essential in the framework of bridge vulnerability analysis, given that many bridge failures are induced by hydraulic factors such as severe scour around bridge piers and abutments (Shahriar et al., 2021; Wang et al., 2017; Wu et al., 2021). Thus, understanding the mutual interactions between the river flow and the bridge structure is of great importance for the design and maintenance of such structures. Nowadays, the standard tools for performing hydrodynamic analyses at bridges are two-dimensional (2D) depth averaged models. Such models allow characterizing the flow at the bridge site, in terms of the spatial (and temporal) distribution of water depth, averaged velocities, mean bed shear stresses, etc. (Dazzi et al., 2020; Morales and Ettema, 2013). These data are then used to conduct further analyses, such as to estimate the scour depth at equilibrium or the hydrodynamic forces on the bridge piers and deck by means of empirical formulas or simplified models. The assumption of hydrostatic pressure distribution and the vertical averaging, inherent in 2D hydrodynamic models, provide reliable estimations only in cases with relatively regular geometries and where the flow does not become pressurized at the bridge deck. The deformed bathymetry and the abrupt geometrical variations generated by the presence of the bridge structure can limit the accuracy of such calculations in many cases of practical interest. Furthermore, each case study is almost unique, which require conducting detailed analyses based on the actual flow and geometrical conditions to describe the flow fields and to estimate the flow actions on the structure and the riverbed. This is especially true for bridge sites with piers and abutments of complex geometry and/or with complex approaching flow patterns associated with the presence of other channel bank structures upstream of the bridge site and flow shallowness (e.g., Chang et al., 2013; Koken and Constantinescu, 2014; Zeng and Constantinescu, 2017). In the recent years, three-dimensional computational fluid dynamics (3D-CFD) models have been used to carry out advanced analyses of flow at bridges (see e.g., Chua et al., 2019; Erduran et al., 2012). The complex geometry of piers and deck, as well as the deformed bathymetry close to piers and abutments, are naturally accounted for by the 3D nature of these models. Two-phase techniques (e.g., the Volume of Fluid, VoF) allow to track the free surface as part of the solution. Such models can be used to accurately simulate cases with free surface flow at the bridge as well as cases with pressurized flow beneath the bridge deck (Lazzarin et al., 2024). Advanced turbulence models (e.g., eddy-resolving LES and DES approaches) allow to accurately simulate the temporal fluctuations of the flow field. However, describing all the salient features of the flow field at river bridges requires considering relatively long river reaches to account for large-scale flow features, and also using high-resolution computational grids, which leads to huge computational costs. This is the reason why previous studies used turbulence-averaged models (e.g., RANS, see Horna-Munoz and Constantinescu, 2018), or were limited to using eddy resolving techniques for flows with idealized geometry of the piers and abutments (e.g., Chang et al., 2011; Kirkil et al., 2008; Kirkil and Constantinescu, 2015). The present study constitutes a pioneering attempt to use the DES approach and the VoF technique to simulate the flow in a real, large-scale river reach with a realistic multi-pier bridge, represented in full-scale. Besides the free surface regime, the pressure flow ( PF ) regime with deck overtopping is also considered, which cannot be simulated by means of classical depth-averaged models (Kara et al., 2015). The present CFD results are far more detailed than those provided by models typically used in vulnerability analysis of bridges, opening new perspectives for these analyses. The case study and the details of the numerical simulations are presented in Section 2. Section 3 reports the main results. Some conclusions are formulated in Section 4. 2. Materials and Methods The present case study refers to a bridge located at the terminal reach of the Po river (Italy), near the town of Occhiobello (Fig. 1a). The bridge is placed in a meandering reach of the river, where the average width is ~320 m (Fig. 1b). The bridge is part of the motorway A13, and was built in the late ‘ 60s. It has two lanes in each direction and a structure in reinforced concrete. Focusing on the active riverbed, 4 piers (i.e., P1 to P4) sustain the bridge with a center distance of 70 m. The piers have curved edges upstream and downstream, and the footprint is basically rectangular (Fig. 1c). Their width is 2.8 m and their length 21.3 m. In the lower part, the pier width increases by 0.8 m and the length by 2.2 m. The deck is made of two beams with a π cross section; each has a width of 9.8 m and