
Filippo GattiCentraleSupélec | ECP · Laboratoire de Mécanique de Paris Saclay UMR 9026
Filippo Gatti
PhD
Enseignant-Chercheur
About
84
Publications
13,396
Reads
How we measure 'reads'
A 'read' is counted each time someone views a publication summary (such as the title, abstract, and list of authors), clicks on a figure, or views or downloads the full-text. Learn more
527
Citations
Introduction
Computational Earthquake Engineering & AI
Additional affiliations
Education
September 2014 - September 2017
September 2014 - September 2017
September 2011 - April 2014
Publications
Publications (84)
The paper presents a numerical approach for the construction of fragility curves for shallow metro tunnels considering the soil-structure-interaction (SSI) and the aging effects due to corrosion in the lining behaviour. In particular, the proposed methodology is presented and applied on a representative soil-tunnel system. Tunnel dynamic response i...
This paper presents a novel deep-learning pipeline to segment large railway datasets with minimal manual annotation, notoriously time consuming. The pipeline adapts DINOv2 for labeling point clouds, with tailored self-distillation pre-training and fine-tuning. The adopted transformer architecture successfully generalizes to multiple railway dataset...
Le lundi 1er janvier 2024, la péninsule de Noto, au Japon, a été frappée par un séisme de magnitude MJ7,6 (Mw 7,5-7,6). L'épicentre se trouve à l'extrémité Nord-Est de la péninsule. Ce séisme a été ressenti avec une intensité JMA égale ou supérieure à six (6-, 6+ et 7) sur l'ensemble de la péninsule. Le séisme de Noto a provoqué 245 morts et 320 bl...
With the advent of increased computational power and advanced numerical methods, the simulation of turbulent fields has become central to many contemporary disciplines. Traditional solvers, while delivering high-quality predictions, still struggle in accurately providing fast estimations of flows due to the complexity and inherently chaotic nature...
In this study, we propose a class-conditioned Generative Adversarial Autoencoder (cGAAE) to improve the realism of simulated ultrasonic imaging techniques, in particular the Multi-modal Total Focusing Method (M-TFM), based on the availability of both simulated and experimental TFM images. In particular, this work studied the case of the inspection...
Seismic hazard analyses in the area of a nuclear installation must accountfor a large number of uncertainties, including limited geological know-ledge. Combining the accuracy of physics-based simulations with theexpressivity of deep neural networks can help to quantify the influence ofcrustal geological uncertainties on surface ground motion. This...
The ever-improving performances of physics-based simulations and the rapid developments of deep learning are offering new perspectives to study earthquake-induced ground motion. Due to the large amount of data required to train deep neural networks, applications have so far been limited to recorded data or two-dimensional (2D) simulations. To bridg...
Numerical simulation of source-to-structure wave propagation consists of a challenging and demanding task in terms of computational capabilities and resources. Some difficulties related to the multi-scale character of the numerical problem can be alleviated using a spectral element (SEM) and finite element (FEM) weak coupling, such as the one descr...
Cette contribution vise à montrer les résultats et performances du simulateur de propagation d'onde 3D à haute fidélité SEM3D (déposé [1, 2] et open source) sur GPU et sur cloud. SEM3D est principalement employé pour simuler des tremblements de terre dans le domaine solide-fluide, des failles actives à la surface libre, à travers des régions urbain...
Numerical simulations are essential tools to evaluate the solution of the wave equation in complex settings, such as three-dimensional (3D) domains with heterogeneous properties. However, their application is limited by high computational costs and existing surrogate models lack the flexibility of numerical solvers. This work introduces the Multipl...
Numerical simulations are computationally demanding in three-dimensional (3D) settings but they are often required to accurately represent physical phenomena. Neural operators have emerged as powerful surrogate models to alleviate the computational costs of simulations. However, neural operators applications in 3D remain sparse, mainly due to the d...
The ever-improving performances of physics-based simulations and the rapid developments of deep learning are offering new perspectives to study earthquake-induced ground motion. Due to the large amount of data required to train deep neural networks, applications have so far been limited to recorded data or two-dimensional simulations. To bridge the...
Seismic hazard analyses in the area of a nuclear installation must account for a large number of uncertainties, including limited geological knowledge. It is known that some geological features can create site-effects that considerably amplify ground motion. Combining the accuracy of physics-based simulations with the expressivity of deep neural ne...
This chapter is meant to provide a basic yet solid understanding of artificial neural networks, to a heterogeneous readership. In particular, the chapter presents three major types of neural networks, namely: feed-forward multi
layer perceptrons (MLP ), convolutional neural networks (CN N ) and recurrent
neural networks (RNN). It was presented at t...
This chapter is a follow-up of the basic introduction provided in the chapter
Artificial Neural Networks: layer architectures, optimizers and automatic differentiation, which prepares the readership to the following theoretical insights. It was presented at the 2023 ALERT Geomaterials doctoral school, in Aussois, in September 2023.
In particular, t...
In this paper, the transient dynamic response of shear type multi-storey buildings subjected to earthquake ground motion is generated via adversarial learning technique under different damage conditions, starting from the relevant undamaged responses. A Representation Generative Adversarial Network (RepGAN) is trained on a database of synthetic acc...
In the context of the nondestructive testing and evaluation research community, global sensitivity analysis (GSA) methods are widespread tools for quantifying the sensitivity of measurements concerning the variation of inputs over the whole design space. For parameter ranking purposes, GSA methods have been historically preferred by NDT&E scholars...
This paper presents a deep-learning surrogate model tailored for a fast generation of realistic ultrasonic images in the Multi-modal Total Focusing Method (M-TFM) framework. The method employs both physics- and data- driven data-sets. To this end, we propose a Conditional U-Net (cU-Net) to perform a controlled generative process of high-resolution...
With the recent rise of neural operators, scientific machine learning offers new solutions to quantify uncertainties associated with high-fidelity numerical simulations. Traditional neural networks, such as Convolutional Neural Networks (CNN) or Physics-Informed Neural Networks (PINN), are restricted to the prediction of solutions in a predefined c...
The 2019 Le Teil earthquake is an illustrative example of a moderate (M W 4.9) yet damaging event, occurring at shallow depth (≈1 km) in a region with little to no geophysical data available. Therefore, using a high-fidelity wave propagation code, we performed numerical simulations of the Le Teil earthquake in a highly uncertain framework, investig...
This paper presents the application of a coupling strategy between the Spectral (SEM) and Finite Element (FEM) Methods to solve the soil-structure interaction (SSI) problem. The SEM-FEM coupling benefits from the mesh refinement capabilities of the FEM in modeling the structure and the near soil, with the realism of the SEM for regional scale earth...
In this paper, Physics-Based numerical Simulation (PBS) is employed to render a broadband (0-10 Hz) realization of the near-field seismic response of the experimental nuclear site of Cadarache, located nearby the active Middle Durance fault (South-Eastern France). The sensitivity of the earthquake numerical model to geological features is investiga...
In this paper, we present SEM3D: a 3D high-fidelity numerical earthquake simulator that is tailored to predict the seismic wave field of complex earthquake scenarios from the fault to the epicenter site. SEM3D solves the wave-propagation problem by means of the spectral element method (SEM). The presented demonstrative test case was a blind MW6.0 e...
We present a novel approach for utilizing big data to estimate economic impacts of disasters by performing end-to-end simulations of disasters and economy at 1:1 scale. In order to realize such fine-grained economic simulations, we developed a high-performance computing extension for Agent Based Economic Models (ABEMs) and seamlessly integrated it...
Realistic physics-based 3D earthquake simulation for source-to-structure wave propagation consists of a powerful numerical tool for seismic response prediction of critical structures submitted to high safety standards. Structural response considering soil–structure interaction (SSI) is usually estimated by Finite Element Method (FEM) approach, as i...
This study investigates the sensitivity of physics-based earthquake prediction to the choice of the regional geological model. The generally poor information on the mechanical properties of the Earth's crust induces large uncertainty margins on the synthetic seismic response. Therefore, uncertainty quantification must be associated with high-fideli...
In this paper, the seismic response analysis of the Kashiwazaki-Kariwa nuclear power plant (KKNPP, Japan) is presented. The basic earthquake scenarios correspond to a Mw4.4 aftershock that followed the Mw6.6 2007 Chuetsu Earthquake, which largely damaged the KKNPP site. A high-fidelity source-to-site earthquake simulator is exploited to generate ph...
Civil structures and infrastructures such as buildings, bridges, tunnels and dams play a crucial role in our society. Their safety and health are threatened by different factors: aging, progressive accumulation of damage and alteration of working and environmental conditions with respect to the design ones, are just a few examples. Rebuilding these...
Civil structures, infrastructures and lifelines are constantly threatened by natural hazards and climate change. Structural Health Monitoring (SHM) has therefore become an active field of research in view of online structural damage detection and long term maintenance planning. In this work, we propose a new SHM approach leveraging a deep Generativ...
The prediction of the seismic response of critical structures is highly sensitive to many
aspects, among which the earthquake source and the geological setting are prominent. The related uncertainty issues must be taken into account in seismic risk mitigation studies, for example through the evaluation of several realizations of a future earthquake...
This study investigates the sensitivity of physics-based earthquake prediction to the choice of the regional geological model. The generally poor information on the mechanical properties of the Earth's crust induces large uncertainty margins on the synthetic seismic response. Therefore, uncertainty quantification must be associated with high-fideli...
The prediction of the seismic response of critical structures is highly sensitive to many aspects, among which the earthquake source and the geological setting
are prominent. The related uncertainty issues must be taken into account in seismic
risk mitigation studies, for example through the evaluation of several realizations of
a future earthquake...
This paper investigates the effect of regional geological structures on the reliability of synthetic earthquake prediction. The lack of information related to the Earth’s crust’s mechanical proper-
ties may render misleading synthetic predictions with large margins of uncertainty. This shortcoming invalidates the advantages of high-fidelity numeric...
Source-to-site earthquake simulation (4.95 billions DOFs) of the Cadarache region (South Eastern France), for a Mw6.0 strong ground motion scenario. Code: SEM3D (Ver 2017.04 Registered at French Agency for Protection of Programs (Dépôt APP), 2017). Simulation performed @Occigen (CINES allocation 2020-A0080410444) on 2000 MPI processes. Contact: Dav...
SEM3D (Ver 2017.04 Registered at French Agency for Protection of Programs (Dépôt APP), 2017) is Spectral Element Method code to solve elasto-acoustics wave propagation problems. It is jointly developed by MSSMat Laboratory of CentraleSupélec, CNRS, CEA and Insitut de Physique du Globe de Paris. SEM3D is capable of simulating broad-band non-linear s...
The main activities of the last months were intended to the wave-propagation code SEM3D [1],based on Spectral Element Method, for the analysis of the seismic response of the Cadarache nuclear site and of the European experimental site of Argostoli (to conclude the activities started with allocations A4-A7). Those activities lead to the publication...
This paper presents a new strategy to blend the outcome of physics-based numerical simulations with massive but poorly-labelled experimental databases such as in-situ data routinely recorded for monitoring purposes. The proposed approach relies on
a set of adversarial learning techniques with a twofold purpose: (1) finding two reduced-dimensional n...
This paper presents a new strategy to blend the outcome of physics-based numerical simulations with massive but poorly-labeled experimental databases such as in-situ data routinely recorded for monitoring purposes. The proposed approach relies on a set of adversarial learning techniques with a twofold purpose: (1) finding two reduced-dimensional no...
This paper investigates the combined effects of natural geological heterogeneities of the Earth’s crust and the highly fractured transition zone across fault discontinuities. The numerical exercise is performed by means of the Spectral Element Method. The heterogeneous domains constituting (1) the crustal model and (2) the thin layer corresponding...
Tectonic and seismic activities induce crustal rock fracturing in the immediate surroundings of buried fault discontinuities. The heterogeneous nature of this 3-D scatter distribution leaves a high-frequency footprint on the recorded wave-field at surface. This study represents an attempt to clarify the complex relationship between the propagated w...
Snapshot of earthquake simulation at Argostoli basin (Greece)
Maximum accuracy : 10 Hz (at softest layer)
20h wall-time (for 20 sec earthquake duration) on 4000 MPI cores @ Occigen (CINES) supercomputer (France)
The scope of this paper is to give an insight into the advantages of a new, all-embracing, modeling approach of a strong ground motion scenario, by carrying out a source-to-structure analysis at regional scale, accounting explicitly for the uncertainties related to the databases and the models. To this end, a suitable case-study is represented by t...
Despite the vast installation of dense arrays in certain regions of the world, there is still a lack of recordings in
the proximity of the fault. Thus, physics based simulations of ground motion are needed to fill this gap. When
the fault is near, it is necessary to consider the finite effects of the source rupture. Variability of slip velocity
fun...
The influence of the regional 3-D geology on the synthetic earthquake ground motion prediction is herein assessed, studying the seismic response of Kashiwazaki-Kariwa nuclear power plant (KKNPP), during the 2007 Niigata seismic sequence (west Japan), so to explore the potentiality and practicability of the innovative computational tools available....
This work aims to study numerically the effect of both topography and bathymetry on ground motion prediction at a regional scale. For this purpose, a 3D model was developed using a numerical code (i.e. SEM3D) that permit to carry out such studies. This code is based on the spectral element method allowing solving of wavespropagation problems in thr...
The influence of the regional 3-D geology on the synthetic earthquake ground motion prediction is herein assessed, studying the seismic response of Kashiwazaki-Kariwa nuclear power plant (KKNPP), during the 2007 Niigata seismic sequence (west Japan), so to explore the potentiality and practicability of the innovative computational tools available....
The aim of this paper is to clarify the extent of near-
source effects and non-linear soil response that took place at
the Japanese nuclear site of Kashiwazaki-Kariwa (KK),
during the MW6.6 Niigata-Ken Chūetsu-Oki earthquake
(July, 16th 2007). The interest of this case study stems from
the relative small source-to-site distance and shallow
hypocent...
In this paper, a novel strategy to generate broad-band earthquake ground motions from the results of 3D physics-based numerical simulations (PBS) is presented. Physics-based simulated ground motions embody a rigorous seismic wave propagation model (i.e., including source-, path- and site- effects), which is however reliable only in the long period...
In this study, a numerical investigation is performed on a realistic source-to-site earthquake scenario, with the aim to assess the role of complex three-dimensional (3-D) geological structures on the predicted wave-field. With this respect, the paper pointedly targets the seismic response of nuclear power plants in near-field conditions and the ve...
The effect of soft sediments on the dispersion of the earthquake ground motion is numerically investigated herein, to assess the effects of (i) the 3-D geological interfaces and (ii) of the spatial fluctuations of the mechanical properties. A source-to-site computational model is built-up, for a complex applicative case, configured as a 3-D soft ba...
The BBS2S Platform : a holistic philosophy for seismic wave propagation
L’ambition de ce travail est la prédiction du champ d’onde incident réalistique, induit par des mouvement forts de sol, aux sites d’importance stratégique, comme des centrales nucléaires. À cette fin, un plateforme multi-outil est développé et exploité pour simuler les aspects différents d’un phénomène complexe et multi-échelle comme un tremblement...
The physics-based numerical simulation is a reliable tool to build-up source-to-site strong ground motion earthquake scenarios.
This study presents a preliminary calibration of a large-scale seismological model of
the M W 6.6 Niigata-Ken Chuetsu-Oki (Japan) earthquake of July, 16 th 2007. The strong ground
motion affected a wide area in the surrounding of the Kashiwazaki-Kariwa Nuclear Power Plant
(KKNPP). Due to the relative small source-to-site distance and shallow hypoce...
The paper presents a numerical approach for the construction of seismic fragility curves for shallow metro tunnels considering the soil-structure-interaction (SSI) and the aging effects due to corrosion of the lining reinforcement. The tunnel response under ground shaking is evaluated through 2D non-linear dynamic analyses, for increasing levels of...
The physics-based numerical simulation is a reliable tool to build-up source-to-site strong ground motion earthquake scenarios. In this paper, we enhance those forward simulations by including (1) the hysteretic non-linear behaviour of geomaterials and (2) the heterogeneous nature of soils. Scattering-induced damping, ground motion spatial incohere...
L'objective de cette étude est l'analyse des effets de site induits au site nucléaire de Kashiwazaki-Kariwa (Japon), par le tremblement de terre Niigata-ken Chuetsu-Oki de 2007. La base de données accélérométrique disponible concerne le séisme principal et les cinq premières répliques. Parmi ces signaux, ceux qui proviennent des deux installations...
Preliminary findings in the seismic site response of the Kashiwazaki-Kariwa Nuclear Power Plant during the 2007 Mw6.6 Niigata Chuetsu Oki earthquake
In this paper, a simplified one-dimensional site response was evaluated for the nuclear site of Kashiwazaki - Kariwa (the largest power plant in the world), in Japan, during the Mw6.6 Niigata-Ken Chuetsu-Oki earthquake (July, 16th 2007). The available database covers partially the Main Shock (MS) and the first five aftershocks. The attention is foc...
Soil-structure interaction phenomena are significantly affecting the seismic behavior and vulnerability of tunnels in soft soils. The paper presents numerical fragility curves that are computed from dynamic time history analyses of representative circular tunnels-soil systems. The mechanical properties of the tunnel lining, the characteristics of t...