NaTech events (Natural Hazard Triggering Technological Disasters) are industrial accidents triggered by natural hazards which may lead to potentially tremendous impact on the environment and the population. Seismic-induced tsunami NaTech risk assessment entails the seismic sources to be characterised and modelled to provide the Peak Ground Acceleration (PGA) information as input to the seismic-induced tsunamis models and simulation needed for a Seismic Probabilistic Tsunami Hazard Analysis (SPTHA). In this thesis, we propose two Sensitivity Analysis (SA) methods to deal with the computational issues related with: 1. The identification of the model parameters most affecting the PGA; 2. The identification most relevant features of the seismic model, for deciding a priori the seismic scenarios to be simulated. With respect to the first issue, we propose a novel Bootstrapped Modularised Global Sensitivity Analysis (BMGSA) method. The method is tested on a benchmark case study. The results are compared with a standard variance-based Global SA method. The strength of the proposed method is that its application only requires input-output data and not the direct accessibility to the code. With respect to the second issue, we propose a wrapper-based heuristic approach to select the set of most relevant features of the seismic model, for deciding a priori the seismic scenarios to be simulated. The proposed approach is based a Multi-Objective Differential Evolution Algorithm (MODEA) and is developed with reference to a case study whose objective of the analysis is calculating the annual rate of a threshold exceedance of the height of tsunami waves caused by subduction earthquakes that might be generated on a section of the Hellenic Arc and propagated to a target site on the eastern coast of Sicily (Siracusa). The comparison between the mean values of annual rate of exceedance of the tsunami wave height estimated considering only the selected scenarios and the full set of scenarios shows that the proposed approach allows a reduction of 95% of the number of scenarios with half of the features to be considered, and with no appreciable loss of accuracy.
Gli eventi NaTech (Natural Hazard Triggering Technological Disasters) sono incidenti industriali causati da calamità naturali con possibili impatti disastrosi su ambiente e popolazione. La valutazione del rischio di NaTech dovuti a tsunami sismogenerati implica la caratterizzazione e la modellazione delle sorgenti sismiche per calcolare Peak Ground Acceleration (PGA), l’input di modelli e simulazioni di tsunami, necessaria per una Seismic Probabilistic Tsunami Hazard Analysis (SPTHA). In questa tesi proponiamo due metodi di Sensitivity Analysis (SA) per affrontare due problemi computazionali: 1. Individuazione dei parametri del modello che influenzano maggiormente la PGA; 2. Individuazione delle feature del modello sismico più rilevanti, per decidere a priori gli scenari sismici da simulare. Riguardo il primo problema, proponiamo un nuovo metodo di Bootstrapped Modularised Global Sensitivity Analysis (BMGSA), testandolo su un caso di studio di riferimento e confrontando i risultati con un metodo standard di Global SA variance-based. Il punto di forza del metodo proposto è che la sua applicazione necessita solo di dati di input-output e non dell’accesso diretto al codice. Riguardo il secondo problema, proponiamo un approccio euristico wrapper-based, per selezionare l'insieme delle caratteristiche più rilevanti del modello sismico, per decidere a priori gli scenari sismici da simulare. L'approccio proposto si basa su un Algoritmo di Evoluzione Differenziale Multi-Obiettivo (MODEA) ed è sviluppato con riferimento ad un caso studio il cui obiettivo è il calcolo del tasso annuale di superamento di una altezza soglia delle onde di tsunami causate da terremoti di subduzione che potrebbero essere generate su un tratto dell'Arco Ellenico e propagate ad un sito target sulla costa orientale della Sicilia (Siracusa). Il confronto tra i valori medi del tasso annuale di superamento dell'altezza dell'onda di tsunami stimata considerando solo gli scenari selezionati e l'insieme completo degli scenari mostra che l'approccio proposto permette una riduzione del 95% del numero di scenari con la metà delle feature da considerare, e senza apprezzabili perdite di precisione.
Advanced sensitivity analysis methods for seismic-induced tsunami NaTech risk assessment
Gallo, Nicola
2020/2021
Abstract
NaTech events (Natural Hazard Triggering Technological Disasters) are industrial accidents triggered by natural hazards which may lead to potentially tremendous impact on the environment and the population. Seismic-induced tsunami NaTech risk assessment entails the seismic sources to be characterised and modelled to provide the Peak Ground Acceleration (PGA) information as input to the seismic-induced tsunamis models and simulation needed for a Seismic Probabilistic Tsunami Hazard Analysis (SPTHA). In this thesis, we propose two Sensitivity Analysis (SA) methods to deal with the computational issues related with: 1. The identification of the model parameters most affecting the PGA; 2. The identification most relevant features of the seismic model, for deciding a priori the seismic scenarios to be simulated. With respect to the first issue, we propose a novel Bootstrapped Modularised Global Sensitivity Analysis (BMGSA) method. The method is tested on a benchmark case study. The results are compared with a standard variance-based Global SA method. The strength of the proposed method is that its application only requires input-output data and not the direct accessibility to the code. With respect to the second issue, we propose a wrapper-based heuristic approach to select the set of most relevant features of the seismic model, for deciding a priori the seismic scenarios to be simulated. The proposed approach is based a Multi-Objective Differential Evolution Algorithm (MODEA) and is developed with reference to a case study whose objective of the analysis is calculating the annual rate of a threshold exceedance of the height of tsunami waves caused by subduction earthquakes that might be generated on a section of the Hellenic Arc and propagated to a target site on the eastern coast of Sicily (Siracusa). The comparison between the mean values of annual rate of exceedance of the tsunami wave height estimated considering only the selected scenarios and the full set of scenarios shows that the proposed approach allows a reduction of 95% of the number of scenarios with half of the features to be considered, and with no appreciable loss of accuracy.File | Dimensione | Formato | |
---|---|---|---|
2021_12_Gallo_01.pdf
accessibile in internet per tutti
Descrizione: testo tesi
Dimensione
2.78 MB
Formato
Adobe PDF
|
2.78 MB | Adobe PDF | Visualizza/Apri |
2021_12_Gallo_02.pdf
accessibile in internet per tutti
Descrizione: executive summary ITA
Dimensione
934.4 kB
Formato
Adobe PDF
|
934.4 kB | Adobe PDF | Visualizza/Apri |
2021_12_Gallo_03.pdf
accessibile in internet per tutti
Descrizione: executive summary ENG
Dimensione
898.94 kB
Formato
Adobe PDF
|
898.94 kB | 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/10589/183123