Understanding how swirl flows can be applied to processes in the oil and gas industry and how problems might hinder them, are the focus of this thesis. Three application areas were identified: wet gas metering, liquid loading in gas wells and erosion at pipe bends due to sand transport. For all three areas, Computational Fluid Dynamics (CFD) simulations were performed. Where available, experimental data were used to validate the CFD results. As a part of this project, a new test loop was conceived for the investigation of sand erosion in pipes. The results obtained from CFD simulations of two-phase (air-water) flow through a pipe with a swirl-inducing device show that generating swirl flow leads to separation of the phases and creates distinct flow patterns within the pipe. This effect can be used in each of the three application areas of interest. For the wet gas metering application, a chart was generated, which suggests the location of maximum liquid deposition downstream of the swirling device used in the ANUMET meter. This will allow taking pressure and phase fraction measurements (from which the liquid flow rate can be determined) where they are most representative of the flow pattern assumed for the ANUMET calculation algorithms. For the liquid loading application, which was taken as an upscaling of the dimensions investigated for the wet gas metering application, the main focus was on the liquid hold-up. This parameter is defined as the ratio of the flowing area occupied by liquid to the total area. Results obtained with CFD simulations showed that as the water rate increases, the liquid hold-up increases, implying a more effective liquid removal. Thus, it was concluded that the introduction of a swirler can help unload liquid from a gas well, although no investigation was carried out on the persistance of the swirl motion downstream of the device. For the third and final application, the erosion at pipe bends due to sand transport, the main focus was to check the erosion rate on the pipe wall with and without the introduction of a swirler. The erosion rate was predicted by CFD simulations. The flow that was investigated consisted of a liquid phase with solid particles suspended in it. The CFD results showed a significant reduction in erosion rate at the pipe walls when the swirler was introduced, which could translate into an extended working life for the pipe. An extensive literature review performed on this topic, complemented by the CFD simulations, showed the need for a dedicated multiphase test loop for the investigation of sand erosion in horizontal pipes and at bends. The design of a facility of this type is included in this thesis. The results obtained with this work are very encouraging and provide a broad perspective of applications of swirl flows and CFD for the oil and gas industry.

O projeto de pesquisa propôs avaliar a metodologia de Simulação de Escalas Adaptativas (SAS) para descrever escoamentos turbulentos e não-reativos utilizando estudos de casos característicos, amplamente documentados, os quais possuem comportamentos do escoamento distintos presentes em diferentes regiões de uma câmara de combustão. O primeiro estudo de caso foi a análise do escoamento sobre um degrau, em que foi avaliada a capacidade do modelo Simulação de Escalas Adaptativas, frente aos modelos de Navier-Stokes com Média de Reynolds (RANS) e Simulação de Grandes Escalas (LES) e aos dados experimentais, em prever a distribuição de pressão, ponto de recolamento e de perfis de velocidade ao longo do domínio após a separação. Pode-se notar que o modelo SAS apresentou resultados praticamente idênticos aos resultados obtidos pelo modelo RANS com relação à distribuição de pressão e a posição ponto de recolamento. Porém, os perfis de velocidade apresentaram algumas discrepâncias com relação aos perfis de velocidade dos modelos RANS e LES e dos resultados experimentais. Um segundo estudo de caso foi a análise do escoamento através de um turbilhonador, em que a capacidade do modelo SAS foi avaliada, comparando seus resultados com os resultados do modelo de Navier-Stokes Não-Estacionárias com Média de Reynolds (URANS) e com os dados experimentais, em prever perfis de velocidade em regiões de recirculação presentes neste estudo de caso. Pode-se observar que ambos os modelos conseguiram prever as principais estruturas de recirculação do escoamento, porém, os perfis de velocidade apresentaram significativas discrepâncias com relação aos dados experimentais. Em seguida, foram feitas comparações entre os modelos SAS e URANS com relação à previsão da precessão central de vórtice e de estruturas de vórtices, das quais foi observado que o modelo SAS apresenta uma maior capacidade para prever estas estruturas em relação ao modelo URANS. / The research project aimed to evaluate the Scale-Adaptive Simulation (SAS) methodology to describe turbulent and non-reactive flows using characteristic, widely documented, case studies, which have distinct flow behaviors present in different regions of a chamber of combustion. The first case study was the analysis of a flow over a backward-facing step, from which the Scale-Adaptive Simulation (SAS) model capacity was evaluated, compared to the Reynolds Averaged Navier-Stokes (RANS) and Large-Eddy Simulation (LES) models and experimental data, in order to predict the pressure distribution, reattachment point and velocity profiles throughout the domain after separation. It can be noticed that the SAS model presented results almost identical to the results obtained by the RANS model in relation to the pressure distribution and reattachment position. However, the velocity profiles presented some discrepancies in respect to RANS and LES velocity profiles and the experimental results. A second case study was the analysis of the flow through a swirler, from which the capacity of the SAS model was evaluated, comparing its results to the results of the Unsteady Reynolds Averaged Navier-Stokes (URANS) model and with the experimental data, to predict velocity profiles in recirculation regions present in this case study. It can be observed that both models were able to predict the main recirculation structures of the flow, however, the velocity profiles presented significant discrepancies in relation to the experimental data. Then, comparisons were made between the SAS and URANS models in respect to the prediction of vortex precession vortex core and vortex structures, from which it was observed that the SAS model presents a greater capacity to predict these structures in relation to the URANS model.