APRÈN. Portal de la producció docent del professorat de la UPC

To predict the fatigue life of metals, three common tests can be used with different fluctuating loading conditions, including axial, rotating bending, and pure bending. However, experimental research shows a discrepancy between these tests, in which the shortest fatigue life corresponds to the axial test, while the pure bending has the longest one, and the rotating bending is in between. This thesis has been dedicated to numerically evaluating this discrepancy between the result of fatigue tests and analyzing the effect of load type on the failure of metals subjected to high-cycle fatigue conditions. The constitutive law used for the analysis of high cycle fatigue is based on isotropic damage modeling, in which the deterioration of the material is described using two parameters, fatigue reduction factor and damage internal variable. Through numerical analysis, various finite element models are generated, including cylindrical, hourglass, and dog-bone specimens. After monotonic and fatigue characterization of the utilized material, the three mentioned fatigue tests are replicated and simulated in Kratos Multiphyiscs, considering their specific loading condition. For each fatigue test, stress distribution, damage propagation, and the evolution of fatigue reduction factor are expressed. The results obtained from the numerical studies are coherent and consistent with the experimental observations. Furthermore, we demonstrate that initiation of damage is independent of the loading condition, contrary to its propagation stage. The crack propagation stage is a relatively small part of the axial fatigue life. In contrast, in the pure bending and the rotating bending tests, the crack propagation stage was longer, respectively. Therefore, this project indicates how the statement of the loading condition and the stress distribution can affect the damage propagation stage and the deterioration of the material through fatigue tests.

El Secuestro Geológico del Carbono (GCS) ha surgido como una estrategia para mitigar las emisiones de dióxido de carbono (CO2) y combatir el calentamiento global. La industria de petróleo y gas ha demostrado que las formaciones rocosas pueden almacenar gases, y cuenta con experiencia en la modelación de reservorios. El atrapamiento físico, en el que el CO2 queda confinado dentro de la estructura porosa de la roca, es fundamental en garantizar la contención a largo plazo. La modelización precisa de inyección de CO2 es esencial para mejorar la GCS. DRAC, un código de Elementos Finitos (FE), modela problemas geomecánicos, transporte, y reacciones químicas. DRAC también simula discontinuidades dentro de medios porosos utilizando elementos junta de espesor nulo con doble nodo. En el contexto de este trabajo, DRAC ha sido mejorado para modelar fluidos compresibles, específicamente CO2, utilizando la ecuación de estado (EOS) de Peng-Robinson, seleccionada por su precisión en el comportamiento del CO2 en las fases gaseosa y supercrítica. La EOS de Peng-Robinson y otras propiedades de transporte del CO2 se compararon con base de datos existentes para garantizar su precisión, y se integró a DRAC. El código fue verificado con soluciones analíticas de estado estacionario y transitorias. Las pruebas iniciales se realizaron en problemas desacoplados con condiciones isotérmicas e isobáricas, en medios porosos y elementos junta. Además, se verificó un problema acoplado gas-mecánico. Después, casos de aplicación evaluaron la inyección de CO2 a distintas presiones en medios porosos fracturados. Las simulaciones revelaron el efecto de enfriamiento Joule-Thomson debido a caídas de presión. Los resultados de DRAC se compararon con los de otros códigos establecidos, mostrando concordancia, con pequeñas discrepancias atribuidas a diferencias de las formulaciones. En general, la exitosa implementación del flujo de CO2 en DRAC representa un avance en sus capacidades para la investigación y la aplicación del GCS

The spinning process faces a significant challenge in determining the parameters for machining new real physical prototypes, often relying on trial-and-error method to find these new process parameters. However, the excessive labor and material costs make this manufacturing process very expensive and the time the time required to produce the final parts becomes impractical. Furthermore, due to parameter errors, workpiece failures are very common and we don't know what caused the problem. Therefore, developing a finite element model could significantly aid in identifying optimal parameters and would improve the entire machining process. With this in mind, this master's thesis aims to develop, validate and optimize a numerical model to facilitate the determination of these new process parameters. This study presents a comparison between experimental analysis and numerical simulation of the spinning process, specifically aimed at validating and optimizing the numerical model

The study of Magnetohydrodynamics (MHD) for plasma systems circulating in tokamaks is crucial for improving and developing more efficient and functional nuclear fusion reactors, advancing in the research of a new clean and sustainable source of energy. Nonetheless, MHD simulations must be provided with an initial configuration of the plasma, frequently based on the equilibrium state. The Grad-Shafranov equation models the equilibrium balancing the plasma pressure and the magnetic confinement in a nuclear reactor for an axisymmetrical plasma system, yielding as a result the poloidal magnetic flux ψ field used notably to visualise the shape of the magnetically confined plasma crosssection. Following guidelines from [1] and [2], EQUILI has been developed as a new independent and functional module inside the high performance computing (HPC) multiphysics Finite Elements (FE) code ALYA [3]. EQUILI solves for a given tokamak geometry the Grad-Shafranov equation using an iterative CutFEM solver. CutFEM is part of a branch of FE methods characterised by an unfitted computational mesh, where geometries and domains are embedded and interfaces are parametrized using level-set functions. This particular method is adapted for problems where interfaces are affected by large deformations and resizing, thus making it well-suited to address magnetically confined plasma equilibrium problems. While the tokamak's confining magnets' currents and positions can be individually adjusted to accommodate a variety of plasma pressure and current profiles in terms of plasma positioning and shaping, the current carried by the plasma depends directly on its cross-section shape, which at the same time is affected by the plasma current self-induced magnetic field. Due to this coupling, the problem needs to be solved using an iterative solver and having the plasma shape not fixed and free (free-boundary problem) to evolve towards the equilibrium configuration, while being constraint by its own circulating current. Therefore, the flexible and deformable nature of the plasma cross-section demands in fact the implementation of a FE method that can deal with this plasticity and must be able to easily track such changes in the plasma/vacuum interface geometry. The module was validated against several ITER equilibrium formulations and a rigorous sensibility test was performed, considering variations in the poloidal field coils' currents conforming the axisymmetrical nuclear fusion reactor.

Artificial microswimmers show great potential in various fields, performing complex tasks at the microscale. Accurate modeling of these swimmers is crucial to understanding their behavior, including self-propulsion and interaction with their fluid environment. At this scale, the principle of low Reynolds number hydrodynamics is typically used to study microswimmers, as inertial forces are negligible compared to viscous forces. Consequently, Stokes hypothesis is fundamental in their mathematical modeling. However, research indicates that complex geometries, flow paths, and the influence of channel walls can lead to higher drag forces than those predicted by Stokes hypothesis. In this study, asymptotic approximations of the Navier-Stokes equations are introduced to incorporate nonlinear inertial terms. This approach linearizes the Navier-Stokes problem into a series of Stokes problems, allowing examination of the influence of nonlinearity on the total drag force acting on micro-swimmers. A comprehensive study is conducted using the open-source Stokes solver HDGLab for the analysis. Two benchmark problems are solved to validate the algorithms implemented in HDGLab. The total drag force acting on micro-swimmers in low Reynolds number flows is analyzed through two different problem sets. Initially, a single-sphere microswimmer configuration is studied, and the drag forces are calculated. In the second case, a three-sphere microswimmer with two different stroke instances is exposed. A critical analysis of the drag force for each approximation is presented, illustrating the variation of drag force for low Reynolds numbers.

Shells are generally defined as structures with fixed shapes that deform under applied loads. In common engineering contexts, a shell is a load-bearing element, that typically does not undergo considerable geometrical changes. Nevertheless, in some applications such as soft robotics, it is desired to engineer structural elements that morph. A morphing structure could either generate movement by the existence of mechanisms that are considered instabilities, or a power input into the system such as activity. Active nematics are a class of liquid crystals capable of performing mechanical power by transducing chemical energy. This work attempts to study an active nematic surface, whose activity is fueled by (elongated) living cells. Our bio-hybrid system is analogous to an active nematic surface as it reshapes upon actuation. Motivated by the experiments where elongated cells are patterned nematically onto a thin sheet, we study the reshaping of the shell intending to understand the relation between the input patterns of nematic cell orientation and the output shape. For this purpose, we combine the appropriate shell theories with active nematic models in a computational framework and derive a shell model with active tension along a nematic texture.

Modelling and simulating accretion growth by characterizing the prestrain as a function of time

The use of continuum damage models for fiber-reinforced plastic materials within the finite element method presents a mesh dependence problem called localization, which is associated with the strain-softening constitutive behaviour of the material. For an accurate prediction of matrix cracks, one possible solution is to align the mesh with the local fiber direction. However, this demands a considerable meshing effort by the user, in particular when dealing with layups and complex geometries. To address this problem, this thesis proposes a method that uses non-aligned meshes, effectively mitigating spurious matrix cracking. The method consists of artificially block the damage initiation and propagation of matrix cracks that are not aligned with the fiber direction, ensuring physically-based crack growth. In order to do that, is important to know the information of the neighbour elements, where matrix damage appears, but also to apply a proper criterion to prevent the spurious damage. The efficacy of the proposed method is validated through comparison with experimental data and the conventional damage model through different open hole tests up to failure. The results demonstrate the improvement and accuracy of the proposed method in predicting matrix damage patterns and its potential to enhance the reliability of virtual testing of composite structures.

Additive manufacturing has opened unexplored possibilities in the fabrication of elastic structures not manufacturable via traditional moulding and machining processes. The goal of this project is to explore the benefits of introducing machine learning (ML) techniques into existing topology optimisation pipelines to improve their efficiency and accuracy. Special emphasis will be devoted to assess the capability of ML algorithms to reduce the computational burden due to the mesh generation and adaptation steps in topology optimisation.

Conjugate heat transfer (CHT) problems play a crucial role in modelling and simulation of cooling phenomena in thermal management systems, e.g., for battery electric vehicles. CHT involves the solution of a system of coupled partial differential equations, encompassing the incompressible Navier-Stokes equations for the fluid and the Fourier equation for the thermal phenomena. The project will focus on the development and implementation of a numerical simulation tool for CHT problems to be integrated in a topology optimisation pipeline. ​

This master's thesis explores the implementation of convective schemes in incompressible flow modeling, emphasizing their impact on energy conservation, momentum, and angular momentum. Specifically the skew-symmetric term and the EMAC term recently proposed. The study begins with a literature review on incompressible fluid simulation and introduces the variational multiscale method, applied specifically to stabilize formulations in incompressible fluid flow. Then the study of conservation properties for the convective terms and finally some numerical tests.

In the field of Computational Fluid Dynamics (CFD), a significant amount of computational power and human effort is dedicated to create the mesh used for solving problems related to viscous fluid flows. When dealing with the solution of compressible Navier-Stokes equations, it is essential to have accurate meshes that can capture intricate features, such as shock waves, turbulent flow, and boundary layers. The brute-force approach of uniformly refining the mesh is often too computationally expensive. Instead, the mesh generation often relays on the expertise of the engineers to manually define the spacing function of the mesh using techniques like background meshes and sources, resulting in a highly time consuming process. In addition, the solutions of these type of problems are extremely sensible to small changes in flow conditions and/or geometry.

In this work Model Order Reduction (MOR) techniques are applied for the simulation of side weirs. The main goal is to generate a surrogate model to predict the water discharge depending on the geometry (crest and length) of the weir and the inflow rate. The work has been done in collaboration with Canal de Isabel II, the responsible company for the management of the urban drainage infrastructure in the Madrid region. A side weir is a complex infrastructure that takes part of the sanitation system of a city that discharges the load of some sewer pipes by redirecting part of the flow to a secondary system. This is automatically activated when the primary combined sewer system is overloaded, as in case of meteorological adverse situations. The surrogate is built based on a training set of 161 finite element flow simulations, processed using a singular value decomposition to generate a basis. Several interpolation techniques are tested to construct the surrogate, namely Inverse Distance Weighting (IDW), natural, linear, nearest and radial basis. The IDW technique is found to be the more accurate for this application, providing velocity distribution errors under 2%. Results are delivered to Canal de Isabel II in a friendly interface format, a MATLAB App, that allows the end-user to run cheap and fast new ROM simulations along the cross-section of interest and obtain the velocity distribution, the air-water interface level and its coordinates.

Nuclear fusion, the process that could potentially bring clean, abundant, and virtually limitless power generation on Earth, has been a subject of intense research and exploration for decades. Among the scientific difficulties are breeding blankets, which play a crucial role in harnessing the energy generated during the fusion process. This study focuses on the analysis and exploration of magnetohydrodynamic (MHD) flows within the context of nuclear fusion breeding blankets. The thesis is structured into two sections: the first section is dedicated to the theory behind MHD, while the second section focuses on numerical simulations using developed MHD formulations. The numerical code employed in this thesis predominantly utilizes the phi-formulation, aligning with common practices in research applications. OpenFOAM and Foam-Extend are the primary tools used to simulate the MHD phenomena, with cases run for Hunt and Shercliff case setups. Furthermore, the cases are extended to higher Hartmann numbers, demonstrating excellent agreement with the analytical solutions for Hartmann numbers up to Ha=2500. Additionally, a small application to a U-bend case, representing scenarios commonly found in breeding blankets, is also presented.

Since 1980s, Additive Manufacturing (AM) has been used by many manufacturers to create 3D printed products. All the recent developments in AM and its ease of creating complex geometries, made AM become the preferred option for engineering applications, biomedical components, etc. There are lots of different methods and different materials that have been developed for AM. This thesis is focused on Fused Filament Fabrication (FFF). In FFF, a thermoplastic material is extruded from a heated nozzle to form a layer- by-layer structure. FFF can be an advantageous method because of material, cost, and time optimization. On the other hand, the component created by FFF can show different material properties than the original raw material. The final part can show anisotropic behavior due to the layer-by-layer deposition although the raw material is fully isotropic. In addition, the inner zone of a FFF part is conformed by air gaps in order to save material and printing time. Because of the anisotropic behavior and the complex geometry of FFF components, their mechanical characterization becomes complex. To obtain the material properties of a computational homogenization technique is performed. Different patterns and different in-fill densities are analyzed in order to provide a database for the material properties of this zone. A damage analysis is made to a FFF component with the homogenized properties for the different considered patterns and densities.

The research presented in this study investigates the performance of two different approaches, namely the staggered approach and the coupled approach, in simulating growth glioblastoma cells in microfluidic devices. The main objective is to compare the accuracy and computational efficiency of these approaches in various cases. The objective of this research is to enhance the efficiency and versatility of the coupled approach by introducing modifications that decouple highly coupled nonlinear equations in order to develop a staggered approach. The study focuses on identifying the appropriate scenarios and conditions where each approach can be effectively implemented. To achieve these goals, we begin by analyzing the coupled approach and identifying its limitations in terms of computational efficiency and flexibility. By decoupling the highly coupled equations, we aim to alleviate these limitations and improve the overall performance of analysis, specifically in terms of computational time. Through extensive experimentation and analysis, we establish the suitable situations and conditions where each approach excels.

Compositional reservoir simulation run times scale with the square of the number of components used to characterize the fluid. Therefore, having usable and practical reservoir models requires that we minimize the number of components without sacrificing prediction accuracy. In this work we: 1. Validate a novel approach that automates the compositional lumping by using clustering techniques that allow the number of components to be controlled 2. Discuss its implementation into a simulator pre-processor to allows quick evaluation of the impact on production profiles and run-times of reducing the number of components in actual simulation runs 3. Explore alternative approaches to compressing the compositional data in the simulator cells to allow component convection to be carried out in compressed format using: (a) Singular Value Decomposition (SVD) (b) Non-negative matrix factorization (NNMF) (c) Auto-encoders (AE)

Transverse stresses are a frequent cause of failure in Fiber-Reinforced Polymers. Numerical prediction of the onset and evolution of this phenomenon still requires significant computational resources due to the use of highly refined meshes and complex numerical models. This work, on the basis of previous promising research by other authors on the topic, continues developing an out-of-plane stress recovery technique applied on linear shell elements based on First-Order Shear Deformation Theory. In comparison with preceding work, this formulation is able to recover transverse stresses of full integration shell elements by means of L2-projection. Furthermore, it has been detected that recovered values in areas with variable curvature are highly sensitive to transverse shear stiffness. In this line, an iterative technique is proposed to address the lack of accuracy of transverse shear stiffness computation. Several numerical tests are performed to validate the proposed implementations, and the results reaffirm the capacity of the method to predict out-of-plane failure as a post-processing tool, increasing its applicability with respect to previous works.