https://repositorio.ufu.br/handle/123456789/5146 2026-08-02T03:07:14Z https://repositorio.ufu.br/handle/123456789/49241 Title: Experimental study on interlayer temperatures in HSLA steel wire arc additive manufacturing aiming for self-tempering Abstract: Directed Energy Additive Manufacturing (GMA-DED) of hardened, high-strength, low-alloy (HSLA) steel components can produce hard, potentially brittle deposits that require subsequent heat treatment to achieve the desired properties. In this thesis, a thermal management strategy for HSLA thin walls manufactured by GMA-DED was developed and validated, in which progressively increasing interlayer (IT) temperatures were actively controlled to keep the entire thin wall at the same temperature and intensify an in situ heat treatment (self-tempering), preserving the wall geometry and avoiding productivity losses due to excessive waiting times between layers. The basic criterion of analysis was to maintain always the same wall width, keeping the wire feed speed constant and adjusting the deposition speed, even at the expense of variations in the heat source's energy per unit length of the deposited layers. Using the hardened ER90S-B3 consumable, the shielding gas for thin-wall deposition under short-circuit transfer was initially selected by combining operational stability metrics with geometry-based criteria, resulting in the choice of the Ar + 8% CO₂ mixture. This gas provided the selected parameters with the highest operational stability and the most consistent wall quality, and was adopted as a reference condition for investigating the influence of IT on the microstructural and geometric evolution of the wall. The Near-Immersed Active Radiation Heating (NIARH) approach was developed, which allowed control of the IT up to 550 °C during thin-walled deposition while maintaining stable weld pool dynamics and short-circuit transfer. An original pyrometry-based system was designed and evaluated to monitor the IT and the cooling rates of the layers. A graphical methodology was developed to estimate the dilution between layers and to predict the architecture of heat treatments applied sequentially to the deposited layers during multiple thermal cycles. The results showed that the thermal history was mainly governed by the IT, counterbalancing the effects of preheating with those of lower deposition energy per unit of layer length (the cooling rates were independent of the IT, although the permanence time of each layer in the tempering temperature range increased). Porosity was measured using Archimedes' principle and showed no significant effect of the TI. However, the harshness showed clear dependence on IT: it remained relatively high and stable at low ITs and decreased as IT increased. The microstructural observations of the walls indicated the predominant presence of bainite, except in the condition with the highest TI, which showed a tendency toward increased carbide formation. Overall, the results demonstrated the feasibility of promoting in situ self-tempering through high, controlled interlayer temperatures in thin walls produced by GMA-DED, offering a practical way to adjust properties while maintaining the process's geometric integrity and stability. 2026-03-31T00:00:00Z https://repositorio.ufu.br/handle/123456789/49169 Title: Comportamento de revestimentos de ferramentas de corte no fresamento do Inconel 718 Abstract: This study aimed to investigate the influence of different coatings applied to solid carbide end mills in the end milling of Inconel 718, a nickel-based superalloy characterized by high mechanical strength and low machinability. Three commercial coatings (ALCRONA EVO® (AlCrN), TISAFLEX® (AlTiN/TiSiXN), and LATUMA® (AlTiN)) were evaluated, in addition to an uncoated condition, under different combinations of cutting speed (50 and 80 m/min) and feed per tooth (0.035 and 0.070 mm/tooth). The experimental tests included the analysis of tool life, surface roughness, cutting temperature, cutting forces, and chip morphology. Tool life was determined based on a maximum flank wear criterion of 0.3 mm and expressed in terms of volume of material removed. Surface roughness was evaluated using the Ra parameter, temperature was measured by infrared thermography under dry conditions, and cutting forces were monitored using a piezoelectric rotating dynamometer. Chip morphology was qualitatively analyzed using optical microscopy. The results showed that tool performance was significantly influenced by the coating, with TISAFLEX® exhibiting the best performance and longest tool life, while the uncoated tool showed the poorest results. Feed per tooth was identified as the most influential parameter, leading to reduced tool life and increased surface roughness (Ra) and cutting forces. Temperature measurements presented relatively low values, attributed to the short cutting length, although an increasing trend was observed along the cutting pass. The chips exhibited similar geometry across the evaluated conditions, being more influenced by the cutting parameters than by the coating. Overall, the results demonstrate that the proper combination of coating and cutting parameters is essential to improve the milling of Inconel 718, contributing to enhance the tool life and surface quality. 2026-07-21T00:00:00Z https://repositorio.ufu.br/handle/123456789/49158 Title: Aplicação de redes neurais convolucionais a imagens de modelos térmicos para a estimativa de tamanho e posição de inclusões que simulam tumores mamários Abstract: Breast cancer is the most prevalent and deadly neoplasm among women worldwide, highlighting the urgent need for early detection strategies that are accurate, safe, and accessible. Mammography, while the gold standard, has limitations, including repeated radiation exposure, reduced sensitivity in dense breasts, and reliance on expert interpretation, while ultrasound and magnetic resonance imaging increase cost and complexity. Infrared thermography emerges as a non-invasive, low-cost, and radiation-free alternative, but visual inspection alone lacks the sensitivity to detect deep or small tumors. In this context, machine learning methods have shown promise in extracting subtle thermal patterns beyond human perception. This study implements and trains convolutional neural networks in Python using the Tensorflow library to learn the position and diameter of a spherical inclusion in thermal models, using steady-state temperature maps generated through finite element simulations in Ansys Mechanical. The methodology is applied to two thermal models: i) parallelepiped: simulating an experiment performed on a silicon parallelepiped with inclusions of spheres heated by thermal resistors, the neural network trained with 270 matrices (27x39) achieved an accuracy of 99.10% and a sensitivity of 98.10%, and was able to correctly predict the position and diameter of the inclusion when applied to an experimental thermogram. ii) anatomical model: simulating metabolically active tumor tissue immersed in anatomical breast geometry representing healthy breast tissue, the neural network trained with 965 matrices (50x50) achieved an accuracy of 97.63% and a sensitivity of 95.62%. These results demonstrate that a neural network can reliably learn geometric and physical characteristics related to inclusions (which simulate the presence of tumors) from low-resolution thermal data, indicating the feasibility of using artificial intelligence applied to thermal images for tumor detection. 2026-02-26T00:00:00Z https://repositorio.ufu.br/handle/123456789/49122 Title: Revisão e proposta de estratégia de controle de movimento para um gerador linear a ímã permanente movido a pistão livre para regime permanente Abstract: This work investigates motion control in a free-piston linear generator (FPLG), with the objective of regulating electromagnetic energy extraction in order to ensure stable operation within mechanical limits and adequate performance in electrical power generation. For this purpose, the system is modeled as a combustion-excited mass–spring oscillator in which the electromagnetic force is represented as an equivalent damping proportional to the piston velocity, considering an electrical architecture composed of passive three-phase rectification followed by a DC–DC boost converter. Three control strategies are developed and evaluated under the same modeling and excitation conditions: (i) proportional–integral control with peak displacement detection, (ii) control based on the mechanical energy of the cycle, and (iii) nonlinear model predictive control (NMPC). The comparison is carried out in terms of piston stroke, oscillatory regime stability, energy balance, and DC power delivered to the load. The results show that PI control with peak detection achieves good steady-state performance, but with higher latency under disturbances; the mechanical-energy-based control regulates energy extraction throughout the oscillation cycle, contributing to stroke containment and robustness to combustion variations; and NMPC enables explicit handling of operational constraints, at the cost of increased command modulation and lower uniformity of the electrical power in the considered architecture. It is concluded that the strategies present different trade-offs between mechanical stability, robustness, and energy utilization, providing a comparative basis for controller design in FPLGs. 2026-06-16T00:00:00Z