It has long been understood that the geotechnical characteristics of the uppermost soil layers can significantly modify the amplitude and the spectral content of the recorded ground motion induced by an earthquake. This effect is generally referred to as local site effect and it plays a fundamental role in determining the spatial variability of the seismic hazard for a given area. Standard 1D or 2D ground response analysis neglects issues related to the dependency of site amplification on source for a specified soil model. Therefore, detailed 3D soil model with the possibility of introducing different source features plays an important role to overcome limitations associated with the 1D or 2D response analysis models. In this work case study of Norcia basin affected by the seismic sequence of 2016 Central Italy is considered to illustrate the variability in ground motion prediction using a physics-based numerical approach. Results of 3D physics-based simulations produced by an open-source high performance spectral element code called SPEED are used. The SPEED simulations are available for the mainshock Mw 6.5 of October 30, 2016 and for two hypothetical rupture scenarios, with variable magnitude and co-seismic slip distribution. Three different soil models are considered for the purpose of this study to check the sensitivity with respect to model parameters. To overcome the limitation associated with the resolution of the numerical model and to predict the response spectral ordinates at short periods a novel approach based on trained Artificial Neural Networks (ANNs) is used to generate BroadBand ground motions. BBs time histories were compared with the strong ground motions records both in time and frequency domain to validate the numerical simulation. Ground shaking maps in terms of Peak Ground Acceleration and Peak Ground Velocity are also constructed and their sensitivity with respect to the different earthquake scenarios soil models is discussed. It has long been understood that the geotechnical characteristics of the uppermost soil layers can significantly modify the amplitude and the spectral content of the recorded ground motion induced by an earthquake. This effect is generally referred to as local site effect and it plays a fundamental role in determining the spatial variability of the seismic hazard for a given area. Standard 1D or 2D ground response analysis neglects issues related to the dependency of site amplification on source for a specified soil model. Therefore, detailed 3D soil model with the possibility of introducing different source features plays an important role to overcome limitations associated with the 1D or 2D response analysis models. In this work case study of Norcia basin affected by the seismic sequence of 2016 Central Italy is considered to illustrate the variability in ground motion prediction using a physics-based numerical approach. Results of 3D physics-based simulations produced by an open-source high performance spectral element code called SPEED are used. The SPEED simulations are available for the mainshock Mw 6.5 of October 30, 2016 and for two hypothetical rupture scenarios, with variable magnitude and co-seismic slip distribution. Three different soil models are considered for the purpose of this study to check the sensitivity with respect to model parameters. To overcome the limitation associated with the resolution of the numerical model and to predict the response spectral ordinates at short periods a novel approach based on trained Artificial Neural Networks (ANNs) is used to generate BroadBand ground motions. BBs time histories were compared with the strong ground motions records both in time and frequency domain to validate the numerical simulation. Ground shaking maps in terms of Peak Ground Acceleration and Peak Ground Velocity are also constructed and their sensitivity with respect to the different earthquake scenarios soil models is discussed.
È noto da tempo che le caratteristiche geotecniche degli strati più superficiali del suolo possono modificare significativamente l'ampiezza e il contenuto spettrale del movimento del suolo registrato indotto da un terremoto. Tale effetto è generalmente denominato effetto sito locale e svolge un ruolo fondamentale nel determinare la variabilità spaziale della pericolosità sismica per una determinata area. L'analisi standard della risposta del suolo 1D o 2D trascura i problemi relativi alla dipendenza dell'amplificazione del sito dalla sorgente per un modello di suolo specifico. Pertanto, un modello 3D dettagliato del suolo con la possibilità di introdurre diverse caratteristiche della sorgente svolge un ruolo importante per superare i limiti associati ai modelli di analisi della risposta 1D o 2D. In questo lavoro si considera il caso di studio del bacino di Norcia interessato dalla sequenza sismica del Centro Italia del 2016 per illustrare la variabilità nella previsione del movimento del suolo utilizzando un approccio numerico basato sulla fisica. Vengono utilizzati i risultati di simulazioni basate sulla fisica 3D prodotte da un codice di elementi spettrali ad alte prestazioni open source chiamato SPEED. Le simulazioni SPEED sono disponibili per la scossa principale Mw 6.5 del 30 ottobre 2016 e per due ipotetici scenari di rottura, con magnitudo variabile e distribuzione co-sismica dello scorrimento. Ai fini di questo studio sono considerati tre diversi modelli di suolo per verificare la sensibilità rispetto ai parametri del modello. Per superare la limitazione associata alla risoluzione del modello numerico e per prevedere le ordinate spettrali di risposta a brevi periodi, viene utilizzato un nuovo approccio basato su RNA (Artificial Neural Networks) addestrate per generare movimenti del suolo a banda larga. Le cronologie temporali dei BB sono state confrontate con le registrazioni di forti movimenti del suolo sia nel dominio del tempo che in quello della frequenza per convalidare la simulazione numerica. Vengono anche costruite mappe di scuotimento del suolo in termini di Peak Ground Acceleration e Peak Ground Velocity e viene discussa la loro sensibilità rispetto ai diversi modelli di scenari sismici del suolo.
Ground motion prediction using a physics-based numerical approach : the case of Norcia, Central Italy
Dsouza, Sachin Alfred
2021/2022
Abstract
It has long been understood that the geotechnical characteristics of the uppermost soil layers can significantly modify the amplitude and the spectral content of the recorded ground motion induced by an earthquake. This effect is generally referred to as local site effect and it plays a fundamental role in determining the spatial variability of the seismic hazard for a given area. Standard 1D or 2D ground response analysis neglects issues related to the dependency of site amplification on source for a specified soil model. Therefore, detailed 3D soil model with the possibility of introducing different source features plays an important role to overcome limitations associated with the 1D or 2D response analysis models. In this work case study of Norcia basin affected by the seismic sequence of 2016 Central Italy is considered to illustrate the variability in ground motion prediction using a physics-based numerical approach. Results of 3D physics-based simulations produced by an open-source high performance spectral element code called SPEED are used. The SPEED simulations are available for the mainshock Mw 6.5 of October 30, 2016 and for two hypothetical rupture scenarios, with variable magnitude and co-seismic slip distribution. Three different soil models are considered for the purpose of this study to check the sensitivity with respect to model parameters. To overcome the limitation associated with the resolution of the numerical model and to predict the response spectral ordinates at short periods a novel approach based on trained Artificial Neural Networks (ANNs) is used to generate BroadBand ground motions. BBs time histories were compared with the strong ground motions records both in time and frequency domain to validate the numerical simulation. Ground shaking maps in terms of Peak Ground Acceleration and Peak Ground Velocity are also constructed and their sensitivity with respect to the different earthquake scenarios soil models is discussed. It has long been understood that the geotechnical characteristics of the uppermost soil layers can significantly modify the amplitude and the spectral content of the recorded ground motion induced by an earthquake. This effect is generally referred to as local site effect and it plays a fundamental role in determining the spatial variability of the seismic hazard for a given area. Standard 1D or 2D ground response analysis neglects issues related to the dependency of site amplification on source for a specified soil model. Therefore, detailed 3D soil model with the possibility of introducing different source features plays an important role to overcome limitations associated with the 1D or 2D response analysis models. In this work case study of Norcia basin affected by the seismic sequence of 2016 Central Italy is considered to illustrate the variability in ground motion prediction using a physics-based numerical approach. Results of 3D physics-based simulations produced by an open-source high performance spectral element code called SPEED are used. The SPEED simulations are available for the mainshock Mw 6.5 of October 30, 2016 and for two hypothetical rupture scenarios, with variable magnitude and co-seismic slip distribution. Three different soil models are considered for the purpose of this study to check the sensitivity with respect to model parameters. To overcome the limitation associated with the resolution of the numerical model and to predict the response spectral ordinates at short periods a novel approach based on trained Artificial Neural Networks (ANNs) is used to generate BroadBand ground motions. BBs time histories were compared with the strong ground motions records both in time and frequency domain to validate the numerical simulation. Ground shaking maps in terms of Peak Ground Acceleration and Peak Ground Velocity are also constructed and their sensitivity with respect to the different earthquake scenarios soil models is discussed.| File | Dimensione | Formato | |
|---|---|---|---|
|
Msc_Thesis_Dsouza.pdf
accessibile in internet solo dagli utenti autorizzati
Descrizione: Ground motion prediction using a physics-based numerical approach: the case of Norcia, Central Italy
Dimensione
17.29 MB
Formato
Adobe PDF
|
17.29 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/182758