Geodeco S.p.A. - Research & Development - GNDT Project Description

INTRODUCTION In this study a series of advanced Probabilistic Seismic Hazard Analyses (PSHA) were carried out keeping into account site amplification effects. In particular: 1) Seismic hazard maps were developed for the Arma di Taggia area (i.e., “Small Area”), where specific investigation data were available. 2) Seismic hazard analyses were carried out at about 4000 different locations throughout the Imperia area (i.e., “Large Area”). In this case no specific soil investigation data were available and the study was performed based on geological maps only. This note outlines the adopted methodology and presents results obtained during the study.
METHODOLOGY The adopted methodology couples conventional Probabilistic Seismic Hazard Analysis (PSHA), whose seminal work was developed some 35 years ago by Cornell, with non-linear dynamic analyses of the soil column subject to real rock ground motions. For each soil column configuration the uncertainty in the soil characteristics (and statistical correlation among properties in different layers) can be incorporated by randomization of the soil properties in each analysis. The non-linear effects of the soil layers on the intensity of the ground motion at the surface is captured by a site-specific, frequency-dependent amplification function, AF(f), where f is a generic oscillator frequency. AF(f), which varies with the intensity of the bedrock motion, is defined as the ratio of the spectral acceleration at the surface to the spectral acceleration at the bedrock, both computed at the same frequency f. However, since the purpose of the project was to provide hazard evaluations over relatively large areas, with limited geological/geotechnical information, we used response surfaces that yield the AF(f) of a soil column at a given location (or of multiple soil columns that are plausible for the site based on the available information) based on a family of pre-computed AF(f) for a large number of soil columns. The response surface is defined by a polynomial functional form, and the only independent variable that is used in the response surface as a predictor for AF(f) of a new soil column is the elastic fundamental frequency of vibration. The latter is related to the soil behavior at very small strain levels and reflects the initial state of the soil. It is correlated with the shear wave velocity of the soil, a parameter widely used to distinguish among different soil types (e.g., NEHRP soil classification), and in many cases it can be obtained from relatively inexpensive measurements of microtremors. Soils behaving similarly at low strain levels but differently at large strain levels, including those with different susceptibility to cyclic mobility effects, can be represented by soil-type-specific response surfaces. In this study the analyses over the investigated area were carried out adopting response surfaces, whereas detailed non-linear finite element analyses were carried out, for comparison purposes, at a few locations where specific geotechnical investigations were available. The methodology allows to keep into account both seismic and geotechnical epistemic uncertainties via a logic-tree approach. RESULTS PSHA in Rock Seismic hazard analyses were carried out at about 4000 different locations throughout the Imperia area (i.e., “Large Area”), and included also the Arma di Taggia Area (i.e., “Small Area”). They were carried out keeping into account the epistemic uncertainty by a logic-tree approach. In particular: Variable parameters were considered for the Gutenberg-Richter recurrence law. Two different sets of seismotectonic provinces were adopted. Two different attenuation laws were used. All results were obtained for a return period of 475 years. Both epistemic and non-epistemic analyses were carried out in the Small Area. Typical response spectra at the rock surface, defined for different confidence levels (i.e., 15, 50, 84 and 95%), are shown in Figures 2 and 3. At locations where the rock units underlie a soil deposit the response spectra at the ground surface were obtained keeping into account local amplification effects (see the following sections). Small Area: Arma di Taggia Results obtained at one location of the Arma di Taggia area (i.e., “Small Area”) are shown in Figure 2. In this case the elastic fundamental frequency of the soil deposit, required as input for selecting the appropriate amplification parameters, was provided by measurements of microtremors (measurements locations shown in Figure 1). This procedure was found to provide reliable results based on comparisons, at two different locations, with V_s values obtained from seismic downhole records and with bedrock depths estimated from available geophysical information. The response spectra shown in Figure 2 were defined for different confidence levels (i.e., 15, 50, 84 and 95%). In this case the epistemic uncertainty is related only to the seismic input parameters (see previous section), whereas no epistemic uncertainty was considered for the soil deposit (i.e., a single soil stratigraphy was adopted based on the available information). A map depicting PGA values at the ground surface over the entire area were also developed. Similar maps can be obtained also for any other relevant parameters (e.g., spectral accelerations at given frequencies). Large Area: Imperia Results obtained at one location of the Imperia area (i.e., “Large Area”) are shown in Figure 3. Since no specific information on local soil stratigraphy was available, in this case the elastic fundamental frequencies of the soil deposit were estimated using the available geological maps. In this particular case, related to a recent marine deposit, we selected response surfaces applicable to saturated sands, loose to medium dense, with variable depth of the bedrock, and estimated two alternative fundamental frequencies. The epistemic uncertainty related to the seismic input parameters is also considered, similarly to what done for the small area (i.e., Arma di Taggia). This procedure allows a quantification of the uncertainty level which could be reduced if additional investigations were carried out. The response spectra shown in Figure 3 were defined for different confidence levels (i.e., 15, 50, 84 and 95%).
Published papers related to the projects: Pelli F., Mangini M., Bazzurro P., Eva C., Spallarossa D., Barani S. (2004), "Site Amplification Effects for Probabilistic Seismic Hazard Mapping in North-West Italy", XXIX General Assembly of the European Seismological Commission, 12-17 Sept., Potsdam, Germany. Eva C., Spallarossa D., Barani S., Pelli F., Mangini M., Bazzurro P. (2004), "Probabilistic Seismic Hazard Analysis Using a Logic Tree Approach: an Application to Western Liguria (North-Western Italy)", XXIX General Assembly of the European Seismological Commission, 12-17 Sept., Potsdam, Germany.
For more information on Geodeco's research and development activities please contact Mauro Mangini, M.S.