Near-source effects on the ground motion occurred at the Conza Dam site (Italy) during the 1980 Irpinia earthquake

Abstract

The seismic performance of geotechnical structures placed close to active faults is influenced by the peculiarities of the near-source seismic wave propagation. This issue is of paramount importance for large and massive structures, such as earth dams, for which the strong variability of the ground motion at the foundation level and the contribution of the vertical component of the motion could induce high differential settlements in the embankment with an increased risk for the structure. When modelling a structure in near-source conditions, the domain of analysis should be considerably wider than that typically adopted in the engineering field, since the seismic source and the propagation medium up to the reference site should be added to the structure and to the soil volume interacting with it. The paper illustrates the procedure developed to define the input motion at the Conza Dam site (Italy) during the 1980 Irpinia earthquake (November 23, 1980; Mw = 6.9). Due to the short distance of the site from the fault area, the Conza Dam is representative of several large dams placed in near-source areas in Italy and worldwide. The paper focuses on (1) the reconstruction of the regional structure of the Campania–Lucania region on the basis of former literature studies and on a new calibration of the 1D velocity model at regional scale; (2) the definition of the reference input motion at the site of Conza Dam with highlights on the spatial variability of the ground motion below the dam foundation and related effects on the structure.

Appendix

1.1 Domain reduction method

The main advantage of the domain reduction method (DRM) is the possibility of substructuring the original problem (Fig. 27a) into two numerical sub-models characterized by different scale dimensions and solved in two different steps (Fig. 27b–c). The methodology is illustrated in Bielak et al. (2003) and Yoshimura et al. (2003) for 3D problems of increasing physical and computational complexity. Referring to the above references for details on the method, only the main features will be here illustrated.

Fig. 27

Real seismic region. a Scheme of the semi-infinite seismic region, including the causative fault, geological structure and local features. b Outer boundary $\varGamma^{ + }$ restricts computation to a finite domain; fictitious interface $\varGamma$ divides region into two subdomains: $\varOmega^{ + }$ which includes the seismic source, represented by nodal forces P _e and $\varOmega$ containing the geological features in the domain of interest. c Scheme of the region partitioned in two subdomains across the interface $\varGamma$ where P _b are the nodal forces transmitted by $\varOmega^{ + }$ into $\varOmega$ (Bielak et al. 2003)

Step 1 The first model, which may span thousands of meters (Fig. 27b), is an auxiliary one and represents a simplification of the real external domain. In this step the earthquake source and propagation path are simulated. As proposed by Bielak et al. (2003), a stratigraphic system (flat-layer scheme), solved by means of 3D Green function, is simulated.

Step 2 The second model (Fig. 27c) contains the domain of interest with reduced spatial dimensions (internal domain), including the structure and the surrounding soil. The seismic source and most of the propagation path from the source to the site are now excluded. The input is given as a set of equivalent nodal forces applied to the interface elements and able to reproduce the seismic source modelled in the first step.

The interaction forces (either contact or boundary forces), P _b , are localized on the fictitious surface $\varGamma$, which divides the entire domain in the two regions $\varOmega$ and $\varOmega^{ + }$, containing respectively the geological features of interest and the semi-infinite exterior subdomain, which includes the fault (Fig. 27b). The vector field of nodal displacements defined ithe interior domain $\varOmega$ the exterior domain $\varOmega^{ + }$ and the boundary between them, $\varGamma$, will be denoted by u _i (interior), u _e (exterior) and u _b (boundary), respectively. In the same figure $\varGamma^{ + }$ is the outer boundary that truncates the original semi-infinite region and where the absorbing boundary conditions are applied.