Effect of spatial variability of seismic motion, local site conditions and soil-structure interaction on the inelastic dynamic analysis of rc bridges

During strong ground motion, it is expected that bridge structures are subjected to excitation that is spatially variable along their longitudinal axis in terms of amplitude, frequency content and arrival time, a fact primarily attributed to wave passage, loss of coherency and local site conditions. Furthermore, the foundation interacts with the soil and the superstructure, thus affecting the dynamic response of the bridge. Towards this direction, a comprehensive methodology that couples the above phenomena is proposed and the computer code ASINC (Asynchronous Support Input Calculator) is developed. For the generation of spatially correlated seismic motions, the code combines the simulation formula proposed by Deodatis (1986) with a site response analysis of damped multiple-layer soil profiles laying on an elastic bedrock. The motions are then filtered in order to account for the kinematic interaction between the foundation and the surrounding soil, whereas the complex dynamic impedance matrix is derived depending on the foundation type and the potential pile-to-pile interaction (Makris and Gazetas 1992). In order to account for the development of a plastic hinge at the base of the RC pier section, the corresponding rotational springs are further modified (Kappos and Sextos 2001). Having obtained different time histories and spring-dashpot systems for all the support points, inelastic dynamic analysis of the bridge can then be performed with the use of any commercial finite element code, without the requirement of complex soil-structure modeling and special inelastic features. In order to check the reliability of the methodology and code developed, extensive tests were performed for all stages of the proposed process, utilizing complementary finite element analysis, use of alternative computer codes, previous research studies and closed-form solutions, where available. Moreover, two direct comparisons were performed in terms of both seismic input and structural response. In particular, the proposed approach was tested against a) recorded data available at Euroseis-test, a densely instrumented and geophysically well-investigated valley located in the Volvi basin, near Thessaloniki, Greece (Raptakis et al. 2000) and b) the effect of such an asynchronous input on a 300m length bridge studied within the framework of a research project in support of Eurocode 8. The results in both cases show satisfactory agreement and together with the tests carried out at successive analysis stages establish a level of confidence for use of the code parametric analyses and seismic design. Having developed and validated the computer code ASINC for the generation of suitably modified ground motion suites and spring-dashpot foundation systems at each support point of a bridge, an extensive parametric analysis scheme is applied. For that purpose, a well-studied bridge structure (Calvi and Pinto 1996) was considered, consisting of four 50m spans supported on three hollow section piers of unequal heights that vary from 7 to 21m and are designed according to Eurocode 8. The bridge is assumed to be founded on a multi-layer subsoil structure through a 2x2 pile group at each support location and is excited in the transverse direction by the Kallithea record obtained during the 1999 Athens earthquake, scaled to a PGA of 0.24g. Various scenarios are constructed incorporating different combinations of site effects, spatial variability and soil-structure-interaction phenomena, all used as an input for the inelastic dynamic analysis of the bridge. The parametric analysis also involves alternative bridge models in terms of structural system (fundamental period, overall and span length, boundary conditions, pier-deck connection and foundation flexibility). The results are being compared with the superstructure action effects of an elastic fixed base bridge structure that is excited synchronously. The above set of parametric analyses targets to highlight (a) the importance of including/neglecting the spatial variability, site effects and soil-structure-interaction in the inelastic dynamic analysis of bridges (b) their relative effect when they are all included in the analysis and (c) the feasibility of applying the proposed methodology within the context of current practice and code provisions. The results indicate that all the aforementioned phenomena could, under certain circumstances, have a significant role in the modification of the earthquake input motion as well as on the structural response of the bridge itself. Furthermore, they confirm the critical role played by local site conditions either in terms of overall motion amplification or as a source of variability when they substantially differ among the support points (Zerva 1999, Saxena et al. 2000). It is also concluded that the response of the bridge is strongly affected by asynchronous motion, especially with respect to higher modes excitation, increase of the pseudo-static component of the response and potential increase of the relative displacements. Notwithstanding the difficulties to draw general conclusions for such a multiparametric problem, the parametric study contributes to the identification of the extreme cases where the above effects are clearly beneficial or detrimental, while it illustrates the applicability of the overall methodology towards a more comprehensive design process