Committee:
     PRESIDENT: GIARDINA, GIORGIA
     SECRETARI: HURLIMANN ZIEGLER, MARCEL
     VOCAL: ROSSETTO, RUDY
Thesis abstract: Interferometry with Synthetic Aperture Radar (InSAR) is a remote sensing technique to measure displacements of the observed surface, either the ground or any structure on it. Although the systematic acquisition of SAR data from satellites started in the 1990s, InSAR is a technique still in development. The continuous improvement of the successive SAR satellite constellations in terms of reliability, revisit time, spatial resolution and data accessibility, and the processing techniques of SAR data in the estimation of surface displacements is improving the performance of InSAR. Thus, this doctoral thesis aims to explore the new opportunities that InSAR offers today. In particular, the objectives are, first, to explore novel contributions of InSAR in geological engineering, more specifically, in land monitoring and the hydrogeological characterisation of the ground; and, second, to do so while covering both qualitative and quantitative applications at both regional and local scales. These goals are addressed through four questions. The first question aims to investigate whether InSAR’s spatial resolution and precision are high enough today to monitor ground deformation at such a local scale and in such a changing environment as a construction site. Applied to the ground deformation induced by the dewatering of civil engineering works, InSAR provides a much more complete view and understanding of the observed phenomena than more common alternative techniques such as levelling. However, levelling usually provides more precise measurements. Thus, a combination of both techniques will improve and make the monitoring of civil engineering works more reliable. The second question is whether, because of its spatial data density and its precision, InSAR is suitable for the numerical modelling of extensive ground deformation phenomena. The results achieved thanks to InSAR data in the hydromechanical numerical modelling of regional and long-term ground subsidence related to the overexploitation of confined aquifers prove to be significantly better than the results obtained with other data sources. InSAR data can, first, cover the whole extent of the aquifer system and identify several different deformation phenomena and, second, improve the design of the numerical models and the calibration of the ground parameters, thus achieving a significantly better reproduction and understanding of the observed ground subsidence that enable the design of more adequate and effective policies for land planning and the management of (ground)water resources. The third question turns to a novel InSAR technique: coherence change detection (CCD), a method to detect and map, under certain conditions, changes in the observed surface. In particular, the question is whether InSAR coherence can indeed remotely detect torrential sediment transport events in arid environments. The conclusion is that InSAR coherence is not only able to detect the occurrence of such events but shows a greater detection capacity than the meteorological data, being much more operative and economical than field campaigns. Finally, the fourth question is whether InSAR coherence can remotely map these events, aiming to evaluate the uncertainty that other factors that may also affect InSAR coherence introduce in the CCD maps. The comparison of several mapping methods with different sensitivity to each component of the coherence changes show that CCD methods are able to locate the areas affected by such events, although rugged areas require a thorough analysis and the precise delimitation of the affected areas still presents some uncertainty. All in all, despite the weaknesses that are still to be resolved, InSAR already offers valuable contributions to geological engineering.

Committee:
     PRESIDENT: CUETO-FELGUEROSO LANDEIRA, LUIS
     SECRETARI: LEDESMA VILLALBA, ALBERTO
     VOCAL: YOSHIOKA, KEITA
Thesis abstract: Induced seismicity due to hydraulic stimulation and circulation is a hurdle for the development of Enhanced Geothermal Systems (EGS). The monitored seismicity has occasionally reached magnitudes large enough to be felt by local populations, and in several EGS cases the largest magnitude earthquakes occur few hours to several months after the cessation of the injection operations. This thesis aims to i) deepen the understanding of the processes that trigger such induced seismicity, ii) improve forecasting methodologies and iii) explore strategies to mitigate its occurrence. The focus is especially placed on post-injection seismicity. The well-documented case of Basel EGS, Switzerland, is adopted as verification example.Firstly, the identification of the triggering mechanisms at Basel is performed making use of a coupled hydro-mechanical model with increasing level of complexity. A simple fault-crossed elastic domain is compared with a homogeneous elastic domain to highlight the importance of structural heterogeneities on the direct and indirect (poroelastic) effects of pressure diffusion as triggering mechanisms of seismicity. Additionally, another hydro-mechanical model is implemented by a discrete faulting network, based on seismic interpretations. The model simulates the plastic reactivations of faults, and the effects of pore pressure diffusion, poroelastic stressing, shear-slip stress transfer and slip weakening are distinguished. Simulation results show that the faults located in the vicinity of the injection well fail during injection, mainly triggered by pore pressure build-up. At the stop of injection, poroelastic stress relaxation leads to the immediate rupture of faults that were stabilized during the injection. Stress redistribution is also a prominent triggering mechanisms of post-injection reactivation of distant faults. Slip-induced friction weakening on certain faults amplifies the potential of these faults to reactivate multiple time during injection and after its stop.The development of a forecasting methodology for induced seismicity is the second major achievement of this thesis. The hydro-mechanical model previously implemented is combined with a seismicity rate model, the Gutenberg-Richter law and the Epidemic Type Aftershock Sequence model to quantify the seismicity magnitude, of both mainshocks and aftershocks, associated to the stress changes due to fluid injection. The application of this hybrid methodology on the discrete-faulted hydro-mechanical model of Basel EGS gives the opportunity to explore the effects of different strategies of injection on the enhancement of fault permeability and induced seismicity rate. A constant injection protocol followed by a progressive decrease of injection rate mitigates post-injection seismicity, while enhancing the permeability of the faults in the domain. The thesis also addresses the discussion of the final cessation of injection, comparing the results of shut-in and bleed-off of the well. Results suggest that shutting-in the well can mitigate better early post-injection seismicity than bleeding-off the well, but pore pressure diffusion can destabilize critically-oriented faults in the reservoir for a longer period of time.By providing enhanced understanding, new methodologies and practical solutions, this thesis represents a substantial step-forward in the mitigation and control of induced seismicity, and in particular of post-injection seismicity, in EGS.

Committee:
     PRESIDENT: SEVILLA CÁRDENAS, RUBÉN
     SECRETARI: FERNANDEZ MENDEZ, SONIA
     VOCAL: PEROTTO, SIMONA
Thesis abstract: Incompressible flow simulations are crucial for numerous scientific and engineering applications, from automotive aerodynamics to biomedical device design. Traditional finite volume (FV) methods, such as the cell-centered finite volume (CCFV) approach used by OpenFOAM, face significant challenges related to mesh quality, particularly non-orthogonality and skewness. These issues often result in accuracy loss and lead to a complex and time-consuming mesh generation procedure, especially for complex geometries. This thesis addresses these issues by developing and implementing a face-centered finite volume (FCFV) solver for viscous laminar incompressible flows within the OpenFOAM framework.The FCFV method can be seen as a hybridisable discontinuous Galerkin (HDG) method using the lowest order approximation. It achieves first-order convergence for velocity, pressure, and velocity gradient without requiring reconstruction, thus significantly reducing mesh sensitivity. Additionally, the face-centered finite volume method circumvents the Ladyzhenskaya-Babuška-Brezzi (LBB) condition, enabling monolithic solution strategies that solve for velocity and pressure simultaneously or staggered approaches based on algebraic splitting that avoid the introduction of non-physical boundary conditions, unlike standard OpenFOAM solvers.The primary contribution of this thesis is the seamless integration of an efficient FCFV solver into OpenFOAM, offering a robust alternative for cases dealing with complex geometries and/or distorted meshes. This development includes both steady-state and transient formulations and introduces a novel hybrid pressure FCFV formulation to enhance accuracy for higher Reynolds number simulations.Extensive benchmarking and validation against well-known test problems, always compared with standard OpenFOAM solvers, demonstrate the robustness and accuracy of the FCFV solvers, particularly on meshes with high non-orthogonality and skewness. The results indicate that the FCFV method always maintains optimal first-order convergence for all variables and outperforms the standard CCFV solvers currently available within OpenFOAM in cases with distorted meshes. The developed FCFV solvers offer a valuable alternative for OpenFOAM users dealing with complex geometries, enabling the use of unstructured and stretched meshes without sacrificing accuracy or stability.

Committee:
     PRESIDENT NO PRESENCIAL: HASSAN, OUBAY
     SECRETARI: HEUZÉ, THOMAS
     VOCAL NO PRESENCIAL: BARLOW, ANDREW JOHN
Thesis abstract: In the realm of Computer-Aided Engineering applied to fast solid dynamics, the intricate mechanical behaviours exhibited by materials when subjected to strong dynamic forces, high speed impacts and complex interactions is modelled efficiently and with high fidelity. Employed in diverse fields such as aerospace, automotive, defence and more, the principal interest is to simulate and comprehend the responses of solids, providing insights into stress propagation and deformation patterns. However, the pursuit of such ambitious goals faces inherent limitations: the accurate representation of material behaviours is an ongoing challenge, and the intricate interplay between simulation accuracy and computational efficiency demands thoughtful insights. More specifically, the chosen kinematics paradigm and the discretisation of the continuum often restrict numerical frameworks in the array of problems they can simulate. Simulations in fast solid dynamics may feature locking, numerical instabilities, checker-boarding, or other difficulties related to the nonlinear nature of the equations at stake.In the objective to address the aforementioned shortcomings, this thesis will build on the set of equations by developing a new mixed formulation based on first-order hyperbolic equations and written with the Arbitrary Lagrangian-Eulerian viewpoint. That approach, used here to describe solid bodies, aims at circumventing bottlenecks of Lagrangian and Eulerian methods by distinguishing the behaviour of the mesh from the evolution of the continuum. Moreover, an intuitive aspect of the ALE method is to have the mesh move in such a way that large distortions are reduced (numerical advantage) while maintaining a representation of the material's physical behaviour that is either comparable or surpassing the results obtained with classical methodologies (better simulation outcome). A key feature will be the presentation of an ALE computational framework where the mesh motion relies on a conservation law, and the geometry is therefore automatically updated without the need of an ad hoc procedure.An acoustic Riemann solver based on upwinding stabilisation, as well as a linear gradient reconstruction, will be used to counteract instabilities brought by the Vertex-Centred Finite Volume Method employed in the framework, and to enhance the overall accuracy. The nonlinear hardening laws will be solved using a Newton-Raphson algorithm. The new framework introduced in this work will be implemented from scratch on the open-source platform OpenFOAM, a tool of choice in industrial and academic environments. The time integration will be tackled by the multi-stage Total Variation Diminishing Runge-Kutta method. Eventually, the robustness and accuracy of the novel computational framework will be examined through a series of challenging numerical examples involving complex body deformations, as well as plastic and thermal considerations.

Committee:
     PRESIDENT: SOLÉ, AURELIA
     SECRETARI: DE MEDINA IGLESIAS, VICENTE CÉSAR
     VOCAL: LA ROCCA, MICHELE
Thesis abstract: This study offers a comprehensive understanding of the hydrological dynamics of the Magdalena River (MR) basin, located in Colombia. Multiple elements that affect streamflow were analyzed, such as climate-forcing drivers and human-induced ones, to understand the complex interactions that shape the region's hydrology.Firstly, the influence of many factors on the flow of water downstream of the Magdalena River was studied. The research identified El Niño southern oscillation (ENSO) episodes as crucial climate-forcing drivers and human-induced modifications such as reservoir evaporation. The complex nature of streamflow changes over time was highlighted by showing the variations in the average, volume, and maximum streamflow, as well as oscillations in evaporation and minimum streamflow, especially during positive ENSO episodes. These findings offer important insights into the changing hydrological regime of the MR basin, emphasizing the complex combination of elements that influence its flow patterns throughout time.Moreover, the study explored the hydrological connection between the MR and the Ciénaga Grande de Santa Marta (CGSM) wetland, revealing the interdependence of these two ecosystems. For this, the study explores the vulnerability of downstream habitats, especially wetlands, to changes in streamflow inputs by taking a broad approach that views the entire wetland as a unified unit. It identified crucial threshold ranges where the inflow from the Magdalena River to the CGSM becomes uncertain. This analysis highlights the urgent need to understand the interactions between water flow and wetland ecosystems and their significant impact. Furthermore, the present research utilizes Long Short-Term Memory (LSTM) neural network models to predict streamflow changes at the Calamar gauging station. The goal was to improve the precision of streamflow predictions by combining data from several gauging stations and reservoir evaporation records. Finally, this study can help enhance the comprehension of the intricate hydrological processes in the MR basin, revealing the interconnected effects of climate fluctuations, human actions, and ecosystem dynamics. In this context, this research sets the foundation for creating well-informed water resource management strategies in Colombia that protect wetland ecosystems' ecological health and adaptability in the face of ongoing environmental changes.

Committee:
     PRESIDENT: MENENDEZ, MONICA
     SECRETARI: RIBAS VILA, IMMACULADA
     VOCAL: SIMONI, MICHELE
Thesis abstract: As the improvement of last-mile logistics operations becomes a more and more pressing challenge to make our cities more sustainable, both from an economic and environmental perspective, a wide sample of new technologies and innovative distribution schemes are currently being piloted by logistics companies and studied within the research community. Beyond the necessary shift in the powertrain technologies, from internal combustion engine vehicles to electric ones, more efficient delivery strategies and fleet typologies are needed. This is especially flagrant in the particular use case of the courier, express, parcel market, which has surged in the last few years and is the source of many inefficiencies in our neighborhoods.In this context, two-echelon networks appear as potential solutions. In these schemes, the journey of freight is divided into two distinct parts, first from the carrier’s distribution center to a network of micro-hubs and then to its final destination. Nevertheless, even if there is a reduction of the total travelled kilometers and greenhouse gas emissions, two-echelon deliveries using e-cargobikes, which is the most common model at the moment, may result in more operation costs than conventional deliveries with vans because of more complex logistics processes and higher infrastructure costs. In addition, e-cargobikes have a lower capacity, which affects their operational efficiency since more delivery routes are needed, leading to an increase of the vehicle fleet.To make two-echelon delivery schemes economically sustainable, which will eventually lead to large-scale deployments, autonomous vehicles (AVs), both ground ones and air delivery drones, seem to be promising solutions with the potential to reduce the delivery operation costs and, at the same time, the emissions of greenhouse gases. Nevertheless, the exact benefits of AVs, jointly with other technologies, such as intelligent parcel lockers, still need to be assessed at a large scale and from a holistic point of view, beyond purely operational considerations.As a consequence, in this doctoral thesis, we build a complete assessment framework for AV technologies in the context of city logistics, from planning to operations, which combines the continuous approximation methodology and a discrete optimization procedure to efficiently tackle the capacitated multi-trip vehicle routing problem with time windows and battery capacity constraints.More specifically, distinct parsimonious logistics models including the usage of air drones, ground robots and parcel lockers are built considering the characteristics of these different fleets. Moreover, a thorough life-cycle analysis is performed to quantify the environmental impact of these technologies. These different logistics models are then assessed at a large European scale and integrated with discrete optimization methods to give a holistic vision of AV operations for last-mile logistics.This doctoral thesis will help practitioners, researchers and city representatives understand the main challenges at stake with autonomous technologies from a logistics perspective, apart from the social and regulatory aspects, which are not considered. The obtained results are encouraging since the combination of a two-echelon network of delivery micro-hubs with ground autonomous delivery robots has the potential to reduce city logistics costs by around 20% in adequate conditions. Finally, externalities could also be reduced but a very particular focus should be set on the reduction of the greenhouse gas emissions from the autonomous vehicle production, to avoid observing an increase of the global warming potential on a life-cycle basis.

Committee:
     PRESIDENT: GONZALEZ-AREGALL, MARTA
     SECRETARI: MARTÍNEZ DÍAZ, MARGARITA
     VOCAL: CHEN, GANG
Thesis abstract: In this thesis, challenges and opportunities of Chinese ports and shipping is investigated from the multi-faced perspectives, i.e., the challenges between ports in the background of the Greater Bay Area (GBA), the challenges of new trade routes to traditional shipping pattern, the opportunities of traditional shipping in the dominant era of land-based transportation and the challenges of prosperous shipping to the port traffic.Hong Kong, which is one of the world’s largest shipping centres and once the leading port in the GBA, has experienced a continuous decline in recent years and has been successively overtaken by two other GBA ports (i.e. Shenzhen and Guangzhou). Chapter 2 explores the concentration, inequality, and competition of the GBA multi-port system during the 1972–2020 period. Results indicate that: 1) Hong Kong has witnessed an entire K-wave evolution, and 2) the evolution of the GBA multi-port system is a result of the asynchronous development stages of different GBA ports; 3) Missing the opportunity period of 2000–2008, the transfer of capital from Hong Kong and the Chinese “opening up” policy are the direct, the underlying and the root causes behind this evolution; 4) A tri-hub multi-port system has gradually taken shape in the GBA finally, in which Hong Kong is the second runner-up after Shenzhen and Guangzhou. The prospect of the Hong Kong Port would not be optimistic, and a negative impact of Guangzhou on Hong Kong could be expected in the near future. The newly opened Trans-Asian Railway offers opportunities for trade logistics between China and ASEAN (Association of Southeast Asian Nations). Chapter 3 selected six factors including cost, time, transport volume, reliability, safety, and environmental friendliness to investigate the impact on logistics for trade between China and ASEAN. It explores the most ideal solution for the Sino-Thai logistics route by using the AHP-TOPSIS approach and ranks the current five Sino-Thai logistics routes. The results indicate that the Sea-land transport (0.787) is still the best option, the Kunming-Bangkok Roadway (0.452) came in second, followed by the newly opened Trans-Asian Railway (0.431); The Nanning-Bangkok Roadway (0.373) ranks second to last, and the river-land transport service (0.249) is the last resort. In this sense, the newly opened Trans-Asian Railway would influence the current pattern of Sino-Thai logistics to some extent. Efficient inter-island transportation is crucial for the development of an island’s economy and traffic accessibility. Chapter 4 investigates the travel mode choice of passengers between two different scenarios (i.e. short sea shipping (SSS) and road transport) and also examines the factors influencing the travel mode choice with an anonymous questionnaire based on the state preference survey. Logit-based models (i.e. Multinomial Logit Model, Random Parameter Multinomial Logit Model, and Finite Mixture model) are used to estimate the determinants of mode choice. Sensitivity analysis is proceeded to determine the impact of adjusting variables on mode choice. Findings suggest that passengers who choose SSS pay more attention to the convenience of the journey, while those who choose road transport are more concerned about the travel cost. Ships’ routeing was first adopted to reduce the risk of ship collisions and promote marine traffic efficiency. In chapter 5, a quantitative assessment method based on the Gini coefficient is proposed to evaluate the ship collision risk in ships’ routeing waters. This method is applied to Ningbo Zhoushan Port (NZP), the Gini coefficient of the course over ground (COG) for each leg is calculated with the historical Automatic Identification System (AIS) data, and then the Hierarchical Clustering Algorithm (HCA) is applied to classify the 28 legs of the ships’ routeing waters in NZP into 5 risk levels. The results show that: PA 4 and 7 are high-risk areas; PA 1 and 2 are medium-high-risk areas.

Committee:
     PRESIDENT: CREMONESI, MASSIMILIANO
     SECRETARI: BONET CARBONELL, JAVIER
     VOCAL: WU, WEI
Thesis abstract: Granular materials, ranging from soils and food grains to construction aggregates and pharmaceutical powders, are prevalent in both natural and industrial contexts. Their complex physical behavior, which can exhibit both solid- and fluid-like properties depending on several external and internal factors, arises from their discrete composition, characterized by intricate particle interactions. This complexity is further amplified by thermal effects, which explains the current gaps in achieving a comprehensive understanding of the thermomechanics of granular matter. This thesis explores the thermomechanical behavior of granular media from different perspectives, including numerical and experimental analyses. In particular, two key thermal effects are addressed throughout the research, namely, heat generation from mechanical energy dissipation and thermal expansion in confined systems. In the physical experiments, the heat generation phenomenon is studied for granular mixing operations using two widely-used industrial devices: a rotating drum and a bladed mixer. In both setups, the temperature rise of granular flows is measured under varying operational conditions. Regarding numerical investigations, two approaches are developed. An efficient Discrete Element Method (DEM) framework is used to investigate heat generation mechanisms and provide insights into the thermomechanical behavior of granular flows. Additionally, a novel multiscale numerical framework is developed, coupling continuous and discrete methods through a machine-learning-based surrogate model, to efficiently simulate thermal expansion in confined granular media. In both strategies, the underlying objective is to reduce the high computational cost of simulating granular media, allowing for large-scale analyses in terms of both time and space.

Committee:
     PRESIDENT: HASSAN, OUBAY
     SECRETARI: NOGUEIRA GAREA, XESÚS ANTÓN
     VOCAL: CARDIFF, PHILIP
Thesis abstract: The fast and accurate simulation of laminar and turbulent incompressible flows is crucial in science and engineering. Low-order numerical strategies, particularly finite-volume (FV) methods based on cell-centred and vertex-centred approaches, remain essential in CFD, especially in industrial settings, due to their favourable trade-off between computational cost and accuracy. However, these methods face challenges such as stabilising convective-dominated flows, handling velocity-pressure coupling, and managing numerical flux reconstruction on distorted and stretched grids. The face-centred finite volume (FCFV) method emerges as an alternative to standard FV methods. Derived from a mixed formulation of the discontinuous Galerkin method, FCFV avoids flux reconstruction at cell faces, making its accuracy and convergence nearly insensitive to grid quality. It also satisfies the Ladyzhenskaya-Babuška-Brezzi (LBB) condition without special treatment for velocity-pressure coupling. This work showcases the FCFV method for simulating laminar and turbulent incompressible flows for the first time. The formulation is based on the Reynolds-averaged Navier-Stokes (RANS) equations coupled with the negative Spalart-Allmaras (SA) model. Three new convective stabilisations inspired by Riemann solvers are proposed, along with monolithic and staggered solution strategies for the RANS-SA system and two relaxation strategies for pseudo-time marching. A new hybrid pressure FCFV formulation is also introduced to improve the FCFV accuracy in laminar flow simulations by enriching the pressure space through relaxed compressibility conditions. Both FCFV and hybrid pressure FCFV achieve first-order convergence of velocity, velocity gradient tensor, and pressure, accurately predicting engineering quantities such as drag and lift on structured and unstructured meshes. By avoiding gradient reconstruction, these methods are less sensitive to mesh quality, even on highly distorted grids. Numerical benchmarks for laminar and turbulent, steady and transient cases assess the performance, accuracy, and robustness of the proposed methodologies. Implemented in Fortran 90, these methods lay the foundation for the integration of FCFV techniques within the CFD community. The work concludes with a detailed discussion of the Fortran 90 FCFV code implementation.

Committee:
     PRESIDENT NO PRESENCIAL: MASSING, ANDRÉ
     SECRETARI: BAIGES AZNAR, JOAN
     VOCAL: COLOMÉS GENÉ, JOSEP ORIOL
Thesis abstract: This thesis covers the development of large-scale numerical methods for the simulations of partial differential equations on arbitrarily complex geometries. The target application of this thesis is the structural simulation of buildings and civil infrastructures, in which lightweight and aesthetical demands usually increase the complexity of their geometries. In these applications, empirical experimentation is often not feasible during the design loop. Thus, we rely on numerical simulations to predict the performance of these shapes under realistic loads, e.g., wind loads that increase the complexity with the sophistication of the geometry. Current simulation tools are based on unstructured body-fitted meshes. The generation of unstructured meshes is time-consuming and involves human intervention. Furthermore, body-fitted methods cannot efficiently exploit modern high performance computing resources. The main goal of this thesis is to design novel simulation tools for rapid, accurate, and automated solutions of partial differential equations on geometries described by computer-aided design. Thus, we aim for a framework that combines (1) an automated pipeline from computer-aided design to finite element analysis, (2) a novel space-time formulation for moving geometries, and (3) a scalable implementation for high performance computing resources. Our developments are accessible through open-source software within the Gridap ecosystem and FEMPAR packages (written in Julia and Fortran, respectively).The contributions of this thesis increase the functionality of state-of-the-art unfitted (or immersed or embedded) finite element methods. These methods utilize structured background meshes to solve partial differential equations on complex domains. In the literature, these domains are implicitly represented by level sets. To address this limitation, (1) we developed a robust algorithm that solves problems on domains described by linear boundary representations. This algorithm is based on robust polyhedra clipping algorithms. We have tested the algorithm against all the analysis-ready geometries (STL files) in the Thingi10k dataset (almost 5,000). We have extended this algorithm to high-order boundary representations. In this extension, we utilize Bernstein-Bézier basis and multi-variate root-finding algorithms. We validate the resulting method with analytical benchmarks and real-world geometries from computer-aided design files. Then, (2) we formulated an unfitted space-time finite element framework for moving explicit geometries. In this formulation, we utilized space-only meshes, circumventing the need for 4D geometrical algorithms. In turn, we developed a transfer method for evaluating the initial values at each time slab. The results matched with analytical analysis and external numerical experiments. Furthermore, we have demonstrated the applicability of fluid problems on rotating complex (2D and 3D) geometries.Finally, (3) we proposed the acceleration of the methods of this thesis through highly scalable algorithms. These algorithms tackle the bottlenecks of parallelization of the intersection algorithms. We have demonstrated the scalability of these algorithms over one billion cells and 12,000 cores. Furthermore, we can combine these algorithms with adaptive mesh refinement techniques to reduce the computational cost further. These tools provide the means to significantly accelerate the design-to-simulation pipeline while increasing the fidelity of the results.

Committee:
     PRESIDENT: LEHMKUHL BARBA, ORIOL
     SECRETARI: MARTINEZ FRUTOS, JESUS
     VOCAL NO PRESENCIAL: RICCHIUTO, MARIO
Thesis abstract: This thesis develops a framework to perform shape optimization under uncertainties for a body under the action of aerodynamic forces. The solution of the flow is performed with finite elements using the full potential equation with an embedded approach, where the object of study is defined implicitly with a level set function. The optimization problem is solved by combining different software packages to perform the solution of the flow, advance in the optimization loop and perform uncertainty quantification. The first contribution of the thesis is the development of a full embedded approach for the solution of the full potential equation. Due to the inviscid hypothesis of potential solvers, these require the definition of a gap in the computational mesh in order to generate lift, known as the wake. Based on previous works where the wake is defined implicitly with an embedded approach, this work also considers the geometry as an embedded body. Mesh refinement and numerical terms are employed to improve the definition of the geometry in the mesh and ensure the definition of the Kutta condition. The solver is validated for two and three dimensions for subsonic and transonic flows with different reference data. Another contribution of the thesis is the development of the adjoint analysis for the subsonic full potential equation with embedded geometries in two dimensions. Each coordinate of the object of study is considered a design parameter in the adjoint analysis, where the effect of the level set function is considered. The sensitivities of the objective function with respect to the design parameters are validated by comparing them to the sensitivities obtained by using a finite differences approach. A shape optimization problem where the lift coefficient is maximized with geometrical constraints is solved as an example of application of the adjoint sensitivities. The embedded shape optimization problem is extended to consider uncertainties in the inlet condition. The optimization problem is reformulated by choosing a risk measure, the Conditional Value-at-risk, which is minimized. The adjoint sensitivities are adapted for the stochastic case, considering the selected risk measure. The estimation of the risk measure is performed thanks to an external uncertainty quantification library, by applying a novel approach which uses Monte Carlo methods to estimate the Conditional Value-at-risk. The stochastic case is solved in a distributed environment, where each optimization step deploys a Monte Carlo hierarchy to estimate the objective function and its gradients.

Committee:
     PRESIDENT: AURICCHIO, FERDINANDO
     SECRETARI: CAICEDO SILVA, MANUEL ALEJANDRO
     VOCAL: DOMINGO-ESPIN, MIQUEL
Thesis abstract: This thesis aims towards the understanding and optimization of Additive Manufacturing (AM) components through the application of accurate computational multiscale simulations. The research is guided by three primary objectives: (1) the development of a multiscale orthotropic material model tailored for Fused Filament Fabrication (FFF) components, (2) the formulation of an optimization strategy to enhance the mechanical performance of FFF parts, and (3) the comprehensive characterization and modeling of their failure mechanisms.To address the first objective, a printing pattern-based orthotropic material model in the framework of multiscale analysis is constructed. This model accounts for the intricate interactions occurring at both filament and component scales in FFF, offering a nuanced representation of the material's behavior. By bridging these scales, the model ensures a comprehensive understanding of the mechanical response of FFF parts, enabling accurate predictions of their performance and failure modes.The second objective focuses on the optimization of the mechanical performance of FFF components. Leveraging the developed multiscale material model, an optimization strategy based on a novel statistics-based algorithm and an orthotropic failure criterion is formulated. The computational domain generation strategy is also tackled, analyzing different approaches and taking advantage of an Adaptive Mesh Refinement (AMR) technique to reduce the computational cost of the simulations. The developed methodology is fully embedded into the AM workflow.The final objective involves the characterization and modeling of the different failure mechanisms present in FFF parts. The different failure modes exhibited by FFF components are identified for each printing pattern present in the component, and a Mechanism-Based (MB) damage criterion is developed to model their stiffness degradation. In addition, an MB cracking model that accounts for the orthotropic brittleness of FFF parts is presented and validated against experimental tests.The central motivation driving this thesis is to alleviate the dependency on costly experimental procedures for characterizing and/or predicting the mechanical behavior of FFF components by performing high-precision and inexpensive multiscale simulations. The outcomes of this study aim to improve the current Design for Additive Manufacturing (DfAM) guidelines.

Committee:
     PRESIDENT NO PRESENCIAL: KARPYN, ZULEIMA
     SECRETARI: VAUNAT, JEAN
     VOCAL: ASENSIO SÁNCHEZ, LAURA
Thesis abstract: Geological Carbon Storage (GCS) holds immense promise for mitigating atmospheric CO2 emissions, making the sealing capacity of caprocks pivotal as storage scales up. Despite excellent sealing capacities observed in intact rock samples in laboratory experiments, the presence of discontinuities like bedding planes, fractures, and faults in field-scale applications poses monitoring challenges. This thesis aims to develop a methodology for characterizing potential low-carbon geo-energy sites across various spatiotemporal scales, spanning from millimeters and nanoseconds to kilometers and decades.To achieve this, the thesis employs a multi-scale approach, starting with large-scale investigations and progressing to in-situ underground rock laboratories and laboratory investigations. The initial phase explores four analytical solutions of pore pressure diffusion with periodic sources to enable real-time interpretation of in-situ data. Comparisons with numerical solutions, incorporating reservoir deformation due to pressure waves, reveal consistent results when identical assumptions are maintained. Despite distinct wave propagation patterns depending on the application (energy storage, liquid injection into shale, and enhanced geothermal systems stimulation), errors remain below 3% for effective mean stress variations across all cases. This approach allows for rapid reactions to unexpected events, crucial in real-time decision-making.The subsequent phase involves plane-strain coupled hydro-mechanical simulations of periodic CO2 injection for 20 years, simulating the CO2LPIE experiment at Mont Terri Underground Rock Laboratory. Results highlight the anisotropic behavior of the material, influencing pore pressure changes and stress variations along bedding planes. Capillary effects in nanoscale pores prevent CO2 penetration into Opalinus Clay, causing it to dissolve into the resident brine. Pore pressure oscillations imposed at the injection well attenuate within tens of centimeters, emphasizing the need for closely located monitoring boreholes to observe the periodic signal.Finally, a 3D numerical model simulates water injection into a fractured Opalinus Clay specimen, accounting for fracture geometry with stress-dependent aperture changes. The model reveals a significant nine-order magnitude span in fracture permeability, impacting fluid flow within the rock specimen. Achieving the best fit with experimental results involves incorporating a natural fracture normal stiffness ranging from 18.7 MPa/mm at lower effective mean stresses to 187.2 MPa/mm at higher confinements. This underscores the critical importance of defining hydro-mechanical parameters of fractures under realistic stress conditions, with implications for secure underground storage.In summary, this thesis provides a comprehensive methodology for characterizing low-carbon geo-energy sites at multiple scales. Analytical solutions offer real-time insights, coupled hydro-mechanical simulations illuminate material anisotropy, and 3D numerical models reveal fracture behavior. The findings contribute to advancing the understanding and implementation of Geological Carbon Storage, offering crucial insights for environmentally sustainable energy practices.

Committee:
     PRESIDENT: BALAIRON PÉREZ, LUIS
     SECRETARI: SANCHEZ JUNY, MARTI
     VOCAL: BERMÚDEZ PITA, MARÍA
Thesis abstract: Hazard classification for dams and off-stream reservoirs, which entails identifying potential damages in the event of structural failure, is a crucial tool for implementing local-level risk reduction plans. Consequently, national administrations have developed guidelines including suggested methodologies and tools.The modification of the Spanish Regulation of the Public Hydraulic Domain (Royal Decree 849/1986, of April 11), carried out through Royal Decree 9/2008, of January 11, obliges owners of off-stream reservoirs with a height of 5 meters or a capacity higher than 100,000 m3, whether public or private, to develop a classification study based on the potential risk of their failure (Articles 356 and 367). This represented a significant paradigm shift, under the former regulation, such a study was only required for dams exceeding 15 m in height or those with a height between 10 and 15 m and a capacity exceeding 100,000 m3.The procedure for this classification is time and resource-consuming, and in the specific case of owners of off-stream reservoirs, they may not have these assets. Therefore, this research proposes a simplified methodology to classify off-stream reservoirs, utilizing a surrogate Machine Learning (ML) model that is simpler and has a lower computational cost than conventional approaches. Additionally, the influence of two sources of uncertainties on hazard classification is analysed. This research is based on the generation of synthetic data. A specialized tool in Iber was developed to generate massive 2D hydraulic models of synthetic off-stream reservoir failures, which make all the processes of construction, calculation and extraction of results automatic.The first analysis was focused on the effect of selecting a breach parametric model on the hydraulic variables, the potential damages, and the hazard classification of the structures. Three common parametric models were compared, using a set of synthetic cases and a real off-stream reservoir. The results highlighted that the choice of the model has significant effects. Notably, the erodibility of the material exerts a high influence, surpassing that of the failure mode. The use of an inappropriate model or a lack of information regarding dike material can lead to overly conservative or underestimated results, consequently affecting hazard classification.The ML model constructed for the simplified methodology was a Random Forest classifier capable of identifying potential damages at any point in the vicinity of an off-stream reservoir. This ML model was trained using synthetic data, offering an estimation of potential damages by considering the physical characteristics of the structure, the surrounding terrain, and vulnerable areas. During a real case application, the simplified methodology achieved an accuracy rate of 91%. The simplified methodology allows owners and administration to obtain a pre-classification without the need to make a 2D hydraulic model, which saves time and money.Furthermore, an interface called ACROPOLIS was developed, integrating the ML model. Users can apply the simplified methodology through ACROPOLIS, which guides them step by step, providing the overall classification of the off-stream reservoir based on Spanish regulations.Finally, the analysis of uncertainties related to breach formation and the location of the breaking points in reservoirs was integrated. This involved comparing the current deterministic approach for hazard classification and a newly proposed fourth-step probabilistic approach that accounts for uncertainties in constructing hazard maps. The study revealed variations in classifications between scenarios, as different breaking points and breach formations generate diverse classifications that can affect emergency plans. Additionally, the proposed visualization can be used for various purposes, including tracking the evolution of categorization over time due to land use changes.