https://hdl.handle.net/ripcmb/30126 Fri, 03 Apr 2026 16:00:08 GMT 2026-04-03T16:00:08Z https://hdl.handle.net/ripcmb/847056 Título: Intensification of CO2 removal from low partial pressure streams coupling gas permeation and membrane contactors Autor(es): Mendes, Felipe Brandão de Souza Abstract: CO2 accumulation in the atmosphere of Itinerant Confined Spaces (ICS) is critical for crew safety because it can affect their cognition if CO2 concentration exceeds certain limits. The development of a national process for CO2 removal enables technological independence, and the major challenge is to remove CO2 at low partial pressures. Process simulations revealed that the coupling of gas permeation and the membrane contactor enhanced CO2 removal, which was reflected in a smaller membrane area. The simulation also provided operational curves correlating the membrane transport properties (permeance and selectivity) with the membrane area, which are important for guiding gas permeation membrane development due to the identified lack of commercial membranes. Amine-functionalized microgels, exhibiting 100 GPU for CO2 and 50 GPU for CO2/N2 selectivity, were fabricated using a spray coater. The microgels were continuously produced in a laminar flow reactor, allowing for the scale-up of the synthesis. The equipment volume was conservatively estimated at 11.6 m³ when correlating the properties of the best-synthesized membrane with the simulation results. This thesis contributes to the development of a CO2 removal process for ICS. Descrição: O acúmulo de CO2 na atmosfera de Espaços Confinados Itinerantes (ECI) é crítico para a segurança da tripulação, pois pode afetar sua cognição se o CO2 ultrapassar certos limites. O desenvolvimento de um processo nacional de remoção de CO2 possibilita a independência tecnológica, e o grande desafio a ser enfrentado é a remoção de CO2 a baixa pressão parcial. Simulações de processo revelaram que o acoplamento de permeação de gás e contactor com membrana intensificou a remoção de CO2, o que foi refletido em uma diminuição na área de membrana. A simulação também forneceu curvas operacionais correlacionando as propriedades de transporte de membrana (permeância e seletividade) com a área da membrana, que são importantes para orientar o desenvolvimento de membranas de permeação de gás, visto que a falta de membranas comerciais foi identificada. Membranas contendo microgéis funcionalizados com aminas foram fabricadas usando a técnica de spray-coating. A melhor membrana apresentou 100 GPU de CO2 e 50 de seletividade CO2/N2. Os microgéis foram produzidos continuamente em um reator de fluxo laminar, permitindo o escalonamento da síntese. O volume do equipamento foi conservadoramente estimado em 11,6 m³ quando se correlacionaram as propriedades da melhor membrana sintetizada com os resultados da simulação. Esta tese contribui para o desenvolvimento de um processo de remoção de CO2 para ECI. Tipo: doctoralThesis Mon, 01 Jan 2024 00:00:00 GMT https://hdl.handle.net/ripcmb/847056 2024-01-01T00:00:00Z https://hdl.handle.net/ripcmb/846568 Título: Arquiteturas de hardware para redes neurais convolucionais visando compromisso entre custo e desempenho Autor(es): França, Alexandre Bazyl Zacarias de Abstract: Convolutional neural networks, especially when embedded in devices with low power consumption requirements, present a significant challenge: balancing the typically high precision expected of neural networks, response time, and the demands on computational, logical, and memory resources. This work presents three hardware architectures for convolutional neural networks with high degree of parallelism and component reuse implemented in a programmable device. The first design, which is termed memoryful architecture, uses as much memory as necessary to store the input data and intermediate results. The second design, which is termed memoryless architecture, defines and explores a specific input sequencing pattern to completely avoid the use of RAM. The third design, which is termed cache memory-based architecture, is an intermediate solution, where the standardization of the input sequence is also explored but an auxiliary memory is used to store some intermediate results and, consequently, improve processing time. We compare the three designs in terms of power, area and processing time. Allowing memory usage increases overall hardware cost but reduces processing time. At the other end, completely eliminating memory usage increases operation parallelism, but compromises processing time. A trade-off between memory usage and processing performance is achieved in the cache memory-based architecture that optimizes processing time but at a cost in terms of hardware resources. Descrição: As redes neurais convolucionais, especialmente as embarcadas em dispositivos com requisitos de baixo consumo de energia, apresentam um desafio significativo: equilibrar a alta precisão esperada das redes neurais, o tempo de resposta e as demandas de recursos computacionais, lógicos e de memória. Este trabalho apresenta três arquiteturas de hardware para redes neurais convolucionais com alto grau de paralelismo e reutilização de componentes implementadas em um dispositivo programável. O primeiro projeto, denominado arquitetura com memória, utiliza a quantidade de memória necessária para armazenar os dados de entrada e os resultados intermediários. O segundo projeto, denominado arquitetura sem memória, define e explora um padrão específico de sequenciamento da entrada para evitar o uso de memória RAM. O terceiro projeto, denominado arquitetura com memória cache, é uma solução intermediária, onde a padronização da sequência de entrada também é explorada mas uma memória auxiliar é utilizada para armazenar alguns resultados intermediários e, consequentemente, melhorar o tempo de processamento. Comparamos as três arquiteturas em termos de potência, área e tempo de processamento. Permitir o uso de memória aumenta o custo geral de hardware, mas reduz o tempo de processamento. Na outra extremidade, dispensar completamente o uso de memória aumenta o nível de paralelismo mas compromete o tempo de processamento. O balanceamento entre uso de memória e desempenho é alcançado na arquitetura com memória cache que otimiza o tempo de processamento mas com custo em termos de recursos de hardware. Tipo: doctoralThesis Sun, 01 Jan 2023 00:00:00 GMT https://hdl.handle.net/ripcmb/846568 2023-01-01T00:00:00Z https://hdl.handle.net/ripcmb/846441 Título: Study of hydrogen risk in a marine nuclear reactor during a loss of coolant accident Autor(es): Vilela, Guilherme Trindade Descrição: Marine nuclear reactors, especially when powering submarines, take most of the benefits which this technology can provide. Due to their high power density, small fuel storage volume required, and the long or even dismissed interval of refueling, marine reactors provide good applicability of nuclear technology. However, the cladding of the reactor fuel elements employs materials that can generate hydrogen posing combustion risks and threats to the containment integrity. Space is limited in a marine reactor containment, and a small amount of hydrogen released can become a potential combustion source. In this study, a computational simulation using the commercial multi-purpose code ANSYS Fluent was performed to provide local distributions of temperature, pressure, hydrogen, and steam concentrations in a reactor containment. These parameters were used for the assessment of hydrogen combustion risk during an accidental scenario. The utilization of a multipurpose code presented the inconvenience of the absence of in-built phase change models for condensation modeling. The condensation phenomenon was modeled in the code through the implementation of external subroutines to introduce steam mass sinks and water mass sources on the domain cells. The magnitude of these sources was calculated based on empirical condensation correlations. During the progression of an accident, condensation plays an important role in limiting pressure increase and increasing heat removal in the containment structures. Moreover, condensation also affects the hydrogen combustion risk as it reduces the volume fraction of steam in the containment atmosphere and allows the expansion of the flammable hydrogen cloud. Hydrogen combustion risk and the possibility of slow deflagration, flame acceleration, and deflagration to detonation transition were assessed from the utilization of the Sigma and Lambda criteria. A theoretical zero-dimensional model has been proposed for the validation of the simulation results, due to the absence of experimental data. Tipo: masterThesis Sun, 01 Jan 2023 00:00:00 GMT https://hdl.handle.net/ripcmb/846441 2023-01-01T00:00:00Z https://hdl.handle.net/ripcmb/846416 Título: Simulations of the scaled reactor cavity cooling system experimental facility with flownex Autor(es): Sena, Marcos Salomão de Descrição: The scaled Water-cooled Reactor Cavity Cooling System (WRCCS) experimental facility reproduces a passive safety feature to be implemented in Generation IV nuclear reactors. It keeps the reactor cavity and other internal structures in operational conditions by removing heat leakage from the reactor pressure vessel. The present work used Flownex to model the facility and predict the experimental thermal-hydraulic behavior. Three representative steady-state cases defined by the bulk volumetric flow rate were simulated (Re = 2,409, Re = 2,490, and Re = 11,524). Flownex predictions of the cavity outlet temperature, risers’ temperature profile and flow rate split were compared with the experimental data and previous RELAP simulations. The comparisons are in reasonable agreement with previous studies, demonstrating the ability of Flownex to simulate the RCCS behavior. For the low Re cases, temperature and flow split across the risers are evenly distributed. Conversely, there’s an asymmetry trend in both temperature and flow distributions for the high Re case. Additionally, a sensitivity analysis was performed using the Re = 2,409 case to assess the impact on the system’s temperature and flow due to power reduction transferred to the risers. The results showed very good adherence to the RELAP studies. Finally, a loss of secondary coolant scenario was conducted. Although Flownex employs a two-phase homogeneous mixture model, the global behavior of the average system’s parameters reasonably agrees with predictions by previous studies in RELAP. Tipo: masterThesis Sun, 01 Jan 2023 00:00:00 GMT https://hdl.handle.net/ripcmb/846416 2023-01-01T00:00:00Z