Innovative approach for soil-structure interaction assessment

The severe damage caused by the Ms 8.1 Michoaćan earthquake in Mexico City in 1985 highlighted the potential effects of soil-structure interactions (SSI) during strong ground motion. Although SSI studies have been carried out for years, so far little attention has been paid to the analysis of the polarization of the wavefield radiated from a vibrating structure into its surroundings, which is necessary for a better characterization of the seismic wavefield. In the framework of this thesis, a novel approach for soil-structure interaction assessment based on waveform analysis is proposed. The approach is an innovative combination of deconvolution and polarization analysis of earthquakes recorded in a building and its surroundings. It allows the identification of the wave types of the radiated waves and the estimation of the energy of the radiated wavefield. The approach consists of four main steps: 1) evaluation of the dynamic behavior of the building, 2) deconvolution of the data recorded in the building and its surroundings, 3) identification of the seismic phase associated with the energy transmitted from the building to the ground and reconstruction of the radiated wavefield, and 4) polarization analysis. The proposed approach was tested using earthquake recordings from two experiments conducted in Italy. The first was carried out in 2019 in Matera, Italy, where a 7-story building and a nearby sports field were instrumented with three-component sensors. The second was conducted in 2022 at the test site in Piana di Toppo, Italy, where a single-degree-of-freedom (SDOF) structure was built to validate the approach in a simpler and more controlled environment. The frequency band containing most of the vibrational energy of the building in the Matera building was estimated using the spectral ratio method. Then, the earthquake data were deconvolved using a sensor at the top of the building as a reference. In order to identify the seismic phases of the complex deconvolved wavefield, a simple analytical transfer function was calculated based on a simplified geometry of the test site. In the analytical deconvolved wavefield, a peak related to the energy transmitted from the building to its surroundings was identified. The reconstructed radiated wavefield was significant compared to the signal recorded in the surroundings of the building and its energy was calculated to be up to 59~\% of the field signal. The polarization of the wavefield transmitted from the building to its surroundings was estimated as mostly linear in the analyzed frequency band. This could be explained, for example, by quasi-Rayleigh waves characterized by three planes in which radial and transverse components have a phase shift and the particle motion in the horizontal plane is elliptical. The built structure of the Piana di Toppo experiment did not transmit shaking energy back to the ground. This prevented the successful identification of the types of waves radiated and the amount of related energy amount. The results suggest that such an experimental design may not be suitable for an SSI experiment. The wavefield radiated from the building in Matera consisted of unconventionally polarized surface waves. Moreover, the energy radiated back from the building showed that the influence of the building on the ground motion was significant for the horizontal components in the considered frequency band. However, for the second experiment, no wavefield radiated from the built structure could be identified, and the results obtained in this thesis are limited to the recordings of only one earthquake. Therefore, more data need to be analyzed to confirm these observations. Further analysis with different data sets will be performed in the future to validate the proposed approach and the obtained results.