Dissertations/Thesis

Clique aqui para acessar os arquivos diretamente da Biblioteca Digital de Teses e Dissertações da UnB

2024
Dissertations
1
  • Sarah Gomes da Silva Paes da Costa
  • Information theory applied to numerical simulations of tokamak plasmas in a low-to-high confinement transition.

  • Advisor : RODRIGO ANDRES MIRANDA CERDA
  • COMMITTEE MEMBERS :
  • RODRIGO ANDRES MIRANDA CERDA
  • GABRIELA CUNHA POSSA
  • RONNI GERALDO GOMES DE AMORIM
  • IBERE LUIZ CALDAS
  • Data: Jan 5, 2024


  • Show Abstract
  • Turbulence exerts significant influence over radial transport in the edge region of tokamak plasmas, and is a critical factor in magnetic confinement for fusion experiments. Despite its substantial impact, our understanding of turbulence in this context remains limited. Coherent structures are fundamental in the realm of turbulent transport within fusion plasmas. Entropy and complexity, derived from information theory, serves as a valuable tool to quantify the level of order or disorder in turbulent plasmas. Notably, these coherent structures contribute to the observation of low spectral entropy values in data obtained from space plasmas and numerical simulations of magnetohydrodynamic turbulence.

    In this analysis, we focus on two-dimensional numerical simulations of the modified Hasegawa-Wakatani equations, which provide a simplified nonlinear model for electrostatic resistive drift-wave turbulence in plasmas. We construct a bifurcation diagram illustrating the transition from a turbulent regime to one dominated by zonal flows, effectively suppressing turbulence. The degree of spatial order or disorder during this transition is obtained by computing the Jensen-Shannon complexity-entropy index of the velocity, derived from the electrostatic potential. This index uses the normalized power of shearlet coefficients as a probability distribution. Our findings reveal that the turbulent regime exhibits a higher degree of entropy and a lower degree of complexity, contrasting with the regime dominated by zonal flows characterized by lower entropy values and a higher degree of complexity. These results hold the potential to advance our understanding of nonlinear processes within drift-wave turbulence in fusion plasmas.

2
  • Simelia dos Santos
  • Assessment of Multiaxial and Classical Methods in Fatigue Life Prediction: Studies on SAE 4130 and AL 2024-T3 Alloys

  • Advisor : JORGE LUIZ DE ALMEIDA FERREIRA
  • COMMITTEE MEMBERS :
  • JORGE LUIZ DE ALMEIDA FERREIRA
  • DANIEL MONTEIRO ROSA
  • EDER LIMA DE ALBUQUERQUE
  • MARCUS VINICIUS COSTA SA
  • Data: Mar 12, 2024


  • Show Abstract
  • The present work aims to compare fatigue life prediction under the influence of mean stress for SAE 4130 steel and Al 2024-T3 aluminum alloys, using both multiaxial and classical models, including critical plane models. For this purpose, the routines of the method were implemented using MATLAB software directly, for determining the critical plane, including in-phase and outof-phase loading, scanning how normal and shear stresses vary on the plane, defining and storing the maximum stress evolution on each cutting plane, and calculating life predictions. The method consists of predicting fatigue life and failure plane orientations through maximum normal and shear stress amplitudes. To evaluate the method, critical planes of multiaxial models were estimated, followed by the calibration of Modified Wöhler Curve constants, and calibrations of traditional Walker and Kwofie models. Life predictions were made for SAE 4130 steel and Al 2024-T3 alloy considering experimental tests available in the literature for different loading ratios under tensile stress with different stress ratios. The validation of life prediction models, considering both multiaxial and uniaxial models, was carried out through analytical comparison between estimated and experimental lives. The results were evaluated through the performance of the error index and a confidence band of  3 lives, allowing an analysis of the proximity between predicted and experimental lives. Analyzing the results obtained from life predictions based on the presented results, there does not seem to be a single fatigue life prediction model that is superior for both SAE 4130 steel and Al 2024-T3 aluminum alloy. However, it is observed that the Walker and Kwofie models stand out as good options, offering reasonably satisfactory predictions for both materials

2023
Dissertations
1
  • Larissa Saboia da Rocha
  •  Evaluation of occupational noise exposure and hearing loss in urban bus drivers in the Federal District

  • Advisor : MARIA ALZIRA DE ARAUJO NUNES
  • COMMITTEE MEMBERS :
  • MARIA ALZIRA DE ARAUJO NUNES
  • MARCUS VINICIUS GIRAO DE MORAIS
  • VANESSA RESENDE NOGUEIRA CRUVINEL
  • ERASMO FELIPE VERGARA MIRANDA
  • Data: Mar 27, 2023


  • Show Abstract
  • Urban bus drivers are exposed to noise during the work. When at high levels, occupational exposure to noise may be associated with hearing tinnitus, communication difficulties and hearing loss. The aim of the present study was to measure the levels of exposure to which urban bus drivers in the Federal District are subjected during their working hours, as well as to investigate the prevalence of noise induced hearing loss (NIHL).The sample consisted of 500 public transport drivers with a mean age of 46 years and working time predominantly over 15 years. Anamnesis was applied with questions about life and work habits, sound pressure levels were measured using a CESVA model SC101 sound level meter, and data from the last sequential audiometry performed by the workers was collected. Difficulty communicating inside vehicles was reported by 20.40% of professional drivers and hearing buzzing noises at the end of working hours by 28.40%. There were 39.80% of drivers with tests suggestive of NIHL, with 60.80% of bilateral cases and 39.20% unilateral. 72.86% of the compromised audiometries belonged to individuals with more than 15 years of experience in the profession and there was a progression in the number of exams with NIHL suggestion the greater the age of the driver. The sound pressure levels were not higher than the exposure limit and the level of action provided for in Occupational Hygiene Standard 01.

2
  • AMANDA ARYDA SILVA RODRIGUES DE SOUSA
  • Structural Health Monitoring of Beams Using Machine Learning Techniques


  • Advisor : MARCELA RODRIGUES MACHADO
  • COMMITTEE MEMBERS :
  • AMERICO BARBOSA DA CUNHA JUNIOR
  • FRANCISCO EVANGELISTA JUNIOR
  • MARCELA RODRIGUES MACHADO
  • MARCUS VINICIUS GIRAO DE MORAIS
  • Data: Mar 27, 2023


  • Show Abstract
  • Structural damage can cause changes in a system's local flexibility, resulting in unwanted displacements and vibrations. Statistical analysis of dynamic structural features enables us to differentiate between the current structural condition and predict its lifespan for short or extended periods. Two damage indexes are used to build a dataset from the beam's natural frequency and frequency response function under both undamaged and damaged conditions. As a crack can directly impact system vibration, the DI can detect damage and measure its severity. In this work, we use machine learning (ML) algorithms, such as k-nearest neighbors (kNN), Support Vector Machine (SVM), Decision Tree (DT), Random Forest (RF), and Naïve Bayes (NB), to monitor the structural integrity of a cantilevered beam reinforced with mass. We discuss the challenges associated with using, performing, and implementing each method. Numerical results demonstrate that the ML algorithm using DI calculated from FRF (FRAC DI) better classifies the dataset than the DI natural frequency. We also showcase the performance of the five ML techniques in classifying the beam's integrity indication. The ML algorithms' accuracy is considered to be fast, precise, and efficient in detecting and estimating cracks in beam structures for numerical simulation data. However, when it comes to experimental data, the algorithms are less accurate but with a good level of robustness.

3
  • DAVI MATIAS DUTRA DA SILVA
  • COMPARATIVE STUDY OF OPTIMIZATION ALGORITHMS FOR PARAMETER IDENTIFICATION OF THE SIGMOID MODEL IN MR DAMPERS

  • Advisor : SUZANA MOREIRA AVILA
  • COMMITTEE MEMBERS :
  • SUZANA MOREIRA AVILA
  • POLLIANA CANDIDA OLIVEIRA MARTINS
  • ANDRÉ MURILO DE ALMEIDA PINTO
  • ALDEMIR APARECIDO CAVALLINI JUNIOR
  • Data: Jun 23, 2023


  • Show Abstract
  • This study focuses on the parameter identification of a commercial magnetorheological (MR) damper. The nonlinear behavior of the MR damper was modeled using the numerically parameterized sigmoid model proposed by Wang, which utilized experimental dynamic behavior of a commercial MR damper and applies a method to fit symmetric and asymmetric sigmoid functions using experimental data. Two optimization methods, namely the Nelder-Mead simplex search method (fminsearch) and the differential evolution (DE), were proposed as minimization algorithms. The performance of the optimization methods was compared. The dependency of frequency excitation, piston displacement, current applied in the coil and the operating temperature of the MR damper were also evaluated. The validation of the model parameter was achieved by comparing experimental results with predicted values. The results show that the proposed methodology is effective in identifying the parameter of the MR damper and can be used to improve the performance of the suspension system.

4
  • José Leandro Cardoso Rivera Vila
  • DISCRETE WAVELET BASED DAMAGE LOCALIZATION ON BEAM-LIKE STRUCTURES

  • Advisor : JHON NERO VAZ GOULART
  • COMMITTEE MEMBERS :
  • JHON NERO VAZ GOULART
  • MARIA ALZIRA DE ARAUJO NUNES
  • SUZANA MOREIRA AVILA
  • GUILHERME FERREIRA GOMES
  • Data: Oct 31, 2023


  • Show Abstract
  • This research investigates damage detection using Discrete Wavelet Transform (DWT) as method of analysis. In addition, the influence of boundary conditions, such as the Wavelet mother type and decomposition level, and many others, were also investigated regarding the efficacy of the method. Static and dynamic regimes were considered. The objective is to define, through the wavelet coefficients, a method able to detect, localize and quantify damage. For Static analysis, Discrete Wavelet Transform (DWT) data treatment is made onto displacement data obtained in beams submitted to transverse static loads applied at selected points over its length. Modal analysis is performed onto the first five levels of vibration modes of the structures. Three beam supports are investigated: simply supported, cantilever, and double-clamped. For each one, diverse parameters of influence were analyzed. As the localization of probable region of the damage and the quantification of this damage are the main goal, understanding how damage localization is impacted by the boundary conditions of each case is crucial to ensure the reliability of the method. Additionally, the influence of a baseline function for comparison, decomposition level, wavelet mother type, damage severity and the polynomial order were also studied. Damage was simulated through a local reduction of the elasticity module of one or more elements. The method then treats the displacement or modal data and reveals the probable region with damage with good precision. Two statistical damage indicators were developed: d1 - index for damage localization; d2 -index for damage quantification. Results show good potential for the proposed DWT-based for both damage localization and quantification indexes. The approach has shown great concordance between many indexes of damage localization, but at the same time a limitation of the method for damage near the edges of a beam. Also, for damage quantification, must be noticed that the damage severity prediction curves were found to be very similar, regardless the type of its supports.

5
  • Jalusa Maria da Silva Ferrari
  • Técnica Numérica Aplicada à Identificação de Danos em Estruturas Parafusadas Usando Respostas Dinâmicas no Domínio do Tempo

  • Advisor : JHON NERO VAZ GOULART
  • COMMITTEE MEMBERS :
  • JHON NERO VAZ GOULART
  • MARIA ALZIRA DE ARAUJO NUNES
  • SUZANA MOREIRA AVILA
  • LIÉRCIO ANDRÉ ISOLDI
  • Data: Nov 10, 2023


  • Show Abstract
  • In several engineering fields, the need to identify and monitor damage in complex and difficult-to-access structures still presents a major challenge for existing non-destructive monitoring techniques. Methods based on Structural Health Monitoring (SHM) demonstrate the potential to identify and monitor damage to structures through an on-board system for immediate comparison with an existing database. In this sense, the present work proposes a numerical study using the SHM damage prediction method with the solution of the inverse problem. The study structure was a fixed beam with elastic boundary conditions modeled using a low-order fidelity model. Using finite elements, a bolted joint with 4 NAS6208-16 bolts was modeled in Ansys Parametric Design Language (APDL) using the COMBIN14 element from 6 stiffnesses, where 3 stiffnesses are torsional and 3 are flexural. A convergence analysis was performed, resulting in the evaluation of the modeling of the problem with elastic conditions and discretization of 20 elements using quadratic interpolation functions between the nodes. Beam damage was defined as a loss of tightening torque on the bolts ranging from 5\% to 50%, resulting in a loss of stiffness in the beam connection. An R-index capable of evaluating results based on acceleration in the vertical direction and normalized with the maximum acceleration signal collected was proposed. Transient simulations using modal superposition were performed to evaluate the robustness of the R-index. As a first result, it was observed that the selected R-index is robust about changes in the force application position and in the number and distribution of nodes for data reading. It was possible to approximate the results through a single quadratic curve, based on the averages obtained from the R-index. 95% confidence limit curves were drawn from the average values, enabling an attempt to solve the inverse problem. With the addition of noise of 1% to 5% in the vertical acceleration data obtained, the detection of torque loss at low levels was impaired. Despite this, through the analysis of the confidence interval, the method was still capable of satisfactorily identifying torque losses above 25%, especially when the use of the average of the values obtained from the sensors was considered to calculate the R-index.

6
  • Ana Luiza Piragibe Freire
  • ROLE OF LAGRANGIAN VORTICES AS MATERIAL TRANSPORT BARRIERS IN A MODEL OF TOKAMAK PLASMAS

  • Advisor : RODRIGO ANDRES MIRANDA CERDA
  • COMMITTEE MEMBERS :
  • RODRIGO ANDRES MIRANDA CERDA
  • POLLIANA CANDIDA OLIVEIRA MARTINS
  • ERICO LUIZ REMPEL
  • JUAN PABLO DE LIMA COSTA SALAZAR
  • Data: Dec 18, 2023


  • Show Abstract
  • In fusion plasma, numerical simulations are commonly employed to investigate the confinement properties of plasma in the bulk region of tokamaks. The modified Hasegawa-Wakatani (MHW) equations are used to model the behavior of plasma, which enables us to understand the radial transport in two-dimensional numerical simulations of electrostatic resistive drift-wave turbulence. By utilizing the MHW equations, we have gained insights into the low-to-high confinement (L-H) transitions that occur spontaneously in the plasma when it moves from a low confinement stage, characterized by turbulent flow, to a turbulence-suppressed regime known as zonal flow. To investigate these transitions, we vary the value of a control parameter α, which is related to adiabaticity, in numerical simulations, and observe the transition between the two regimes. This simplified model of L-H transitions can provide valuable information for tokamaks. The chaotic mixing properties of the flow are characterized by means of Lagrangian coherent structures (LCS). First, we compute the finite -time Lyapunov exponent (FTLE) of the calculated velocity field derived from the electrostatic potential to better characterize the chaotic mixing of the turbulent and zonal flow regimes. Then, we compare the statistics of the chaotic mixing of the two regimes using probability distribution functions (PDFs). Lastly, we identify the Lagrangian vortices using geodesic theory to further our understanding of the chaotic mixing of the two regimes, by also implementing the calculation of the kinetic energy for the vortices and total domain. These results can contribute to the understanding of turbulent transport processes in magnetic confinement fusion plasmas.

2022
Dissertations
1
  • Pedro Henrique de Queiroz Rocha
  • VIBRATION CONTROL OF A WIND TOWER USING HYBRID MASS DAMPER

  • Advisor : SUZANA MOREIRA AVILA
  • COMMITTEE MEMBERS :
  • CARLA TATIANA MOTA ANFLOR
  • MARCUS VINICIUS GIRAO DE MORAIS
  • SUZANA MOREIRA AVILA
  • ZENÓN JOSÉ GUZMÁN NÚÑEZ DEL PRADO
  • Data: Jul 22, 2022


  • Show Abstract
  • The increase in energy consumption leads to a great expansion of the world energy generation with various energy sources, among them renewable sources are increasingly highlighted. Wind energy is one of the sources that has been gaining more and more spotlight in the hall of generators, which consequently brings more studies on the structural reliability of wind towers that can cause undesirable levels of vibrations that can compromise the safety and reliability of the structure. This work proposes the application of a structural control system, in a way of an hybrid mass damper (HMD), installed in a wind tower, subjected to seismic and wind loads. Two models are analyzed: considering or not the flapwise effect on the blades. Furthermore, it is proposed to compare the system response for different situations: system without control, passive control, active control and hybrid controllers (Instantaneous Optimal Control (IOC), Proportional Integral Derivative (PID) and Classic Optimum (LQR)). Numerical-computational simulations were performed using the software MAPLESOFT, MATLAB with toolbox Simulink. The numerical results obtained were analyzed and compared with each other, and it was observed that the IOC was the better in reducing the dynamic response of the tower. Although the results of this work are based on simplified models with preliminary results, they may become a basis for more advanced studies in sophisticated models and full-scale applications.

SIGAA | Secretaria de Tecnologia da Informação - STI - (61) 3107-0102 | Copyright © 2006-2024 - UFRN - app09_Prod.sigaa03