Spoke 5 "Environment & Natural Disasters" of the National Centre in High Performance Computing (HPC), Big Data and Quantum Computing aims to develop methodologies for modeling natural physical processes capable of causing disasters that lead to loss of life and socio-economic damage. Among the natural processes that cause disasters are earthquakes, which, as they propagate through soil deposits, undergo surface amplification phenomena capable of causing damage to structures and infrastructures. Consequently, predicting amplification effects becomes of fundamental importance for reducing seismic risk in highly exposed areas. In this context, the large-scale numerical modeling of seismic wave propagation requires the use of complex numerical models and the adoption of parallel computing techniques through HPC systems. The thesis proposes an operational methodology for 1D, 2D, and 3D linear and nonlinear dynamic FEM modeling, applied to the seismic response of an ideal natural slope. The simulations took into account the dynamic solid-fluid interaction through the implementation of the u-p formulation and were carried out using the Plaxis and OpenSees computational codes. The latter allows the resolution of FEM models in parallel through the use of high-performance computing infrastructures. The cyclic response of the soils was described by adopting visco-elastic and visco-elasto-plastic models with Rayleigh damping. In particular, the Pressure Independent Multi-Yield model with isotropic and kinematic hardening was adopted in the OpenSees code, while the Hardening Soil model with small strain stiffness was used in Plaxis. The comparison between the 1D and 2D results allowed validating the OpenSees predictions against those of a finite element software well-established in the scientific literature and, more generally, identifying the advantages and limitations of adopting different numerical tools for evaluating the seismic response of a natural slope. Finally, the processing of the maximum surface accelerations obtained from the 3D analysis allowed defining local seismic hazard maps for the entire study area.
Lo Spoke 5 “Environment & Natural Disasters” del Centro Nazionale in High Performance Computing (HPC), Big Data and Quantum Computing si pone come obiettivo lo sviluppo di metodologie per la modellazione di processi fisici naturali capaci di provocare disastri tali da indurre perdite di vite umane e danni socio-economici. Tra i processi naturali fonte di disastro rientrano i terremoti, che, nel propagarsi all’interno di depositi di terreno, subiscono fenomeni di amplificazione in superficie in grado di provocare danni alle strutture e alle infrastrutture. Di conseguenza, la previsione degli effetti di amplificazione diviene di fondamentale importanza per la riduzione del rischio sismico su aree fortemente esposte. In tale contesto, la modellazione numerica della propagazione sismica su larga scala richiede l’uso di modelli numerici complessi e l’adozione di tecniche di calcolo parallelo attraverso sistemi HPC. La tesi propone una metodologia operativa di modellazione FEM dinamica lineare e non lineare 1D, 2D e 3D, applicata alla risposta sismica di un pendio naturale ideale. Le simulazioni hanno tenuto conto dell’interazione dinamica solido-fluido attraverso l’implementazione della formulazione u-p e sono state svolte utilizzando i codici di calcolo Plaxis e OpenSees. Quest’ultimo permette la risoluzione di modelli FEM in parallelo attraverso l’uso di infrastrutture di calcolo ad alte prestazioni. La risposta ciclica dei terreni è stata descritta adottando modelli visco-elastici e visco-elasto-plastici con smorzamento di Rayleigh. In particolare, nel codice OpenSees è stato adottato il modello Pressure Independent Multi Yield con incrudimento isotropo e cinematico, mentre in Plaxis si è utilizzato l’Hardening Soil model with small strain stiffness. Il confronto tra i risultati 1D e 2D ha permesso di validare le previsioni di OpenSees rispetto a quelle di un software agli elementi finiti consolidato nella letteratura scientifica e, più in generale, di identificare i vantaggi ed i limiti dell'adozione di diversi strumenti numerici per la valutazione della risposta sismica di un pendio naturale. Infine, l’elaborazione delle accelerazioni massime di superficie ottenute dall’analisi 3D ha consentito di definire mappe di pericolosità sismica locale per l’intera area oggetto di studio.
Analisi di risposta sismica locale su area vasta usando risorse HPC
CAVALLO, GIANLUCA
2026
Abstract
Spoke 5 "Environment & Natural Disasters" of the National Centre in High Performance Computing (HPC), Big Data and Quantum Computing aims to develop methodologies for modeling natural physical processes capable of causing disasters that lead to loss of life and socio-economic damage. Among the natural processes that cause disasters are earthquakes, which, as they propagate through soil deposits, undergo surface amplification phenomena capable of causing damage to structures and infrastructures. Consequently, predicting amplification effects becomes of fundamental importance for reducing seismic risk in highly exposed areas. In this context, the large-scale numerical modeling of seismic wave propagation requires the use of complex numerical models and the adoption of parallel computing techniques through HPC systems. The thesis proposes an operational methodology for 1D, 2D, and 3D linear and nonlinear dynamic FEM modeling, applied to the seismic response of an ideal natural slope. The simulations took into account the dynamic solid-fluid interaction through the implementation of the u-p formulation and were carried out using the Plaxis and OpenSees computational codes. The latter allows the resolution of FEM models in parallel through the use of high-performance computing infrastructures. The cyclic response of the soils was described by adopting visco-elastic and visco-elasto-plastic models with Rayleigh damping. In particular, the Pressure Independent Multi-Yield model with isotropic and kinematic hardening was adopted in the OpenSees code, while the Hardening Soil model with small strain stiffness was used in Plaxis. The comparison between the 1D and 2D results allowed validating the OpenSees predictions against those of a finite element software well-established in the scientific literature and, more generally, identifying the advantages and limitations of adopting different numerical tools for evaluating the seismic response of a natural slope. Finally, the processing of the maximum surface accelerations obtained from the 3D analysis allowed defining local seismic hazard maps for the entire study area.| File | Dimensione | Formato | |
|---|---|---|---|
|
XXXVIII - Gianluca Cavallo.pdf
accesso aperto
Licenza:
Tutti i diritti riservati
Dimensione
52.13 MB
Formato
Adobe PDF
|
52.13 MB | Adobe PDF | Visualizza/Apri |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14242/359008
URN:NBN:IT:POLIBA-359008