Projeto da geometria do divergente de tubeira de motor-foguete para empuxo máximo usando o método de otimização de forma
Resumo
Resumo: A otimização da geometria da seção divergente de tubeiras de motores-foguete é investigada no presente trabalho. A necessidade de geometrias otimizadas para tubeiras de motores-foguete se justifica pela economia do consumo de combustível que pode ser alcançada com o design de melhores sistemas de propulsão, possibilitando a redução da massa de propelente no veículo, a qual representa aproximadamente 80% da massa dos veículos aeroespaciais. Neste trabalho, buscam-se geometrias do divergente que possibilitem a maximização do desempenho da tubeira, avaliado quantitativamente pelo coeficiente de empuxo gerado pela tubeira. Dois casos foram considerados: raio de saída fixo e raio de saída variável. A metodologia de otimização empregada utiliza o método de otimização de forma, o qual usa o algoritmo de otimização de Evolução Diferencial da biblioteca Pagmo, selecionado após a avaliação e comparação do desempenho de diferentes algoritmos de otimização estocásticos, avaliados no problema em questão. No processo de otimização, o otimizador sugere a coordenada radial de pontos distribuídos na parede da seção divergente da tubeira. Otimizações foram realizadas com um, dois, três e quatro pontos de otimização com o intuito de avaliar o efeito do número de pontos de otimização nas geometrias otimizadas. As coordenadas axiais desses pontos de otimização foram definidas em otimizações em malhas grossas considerando variáveis tanto as coordenadas axiais quanto as radiais. Também foi avaliado o efeito do refino da malha nas geometrias ótimas. O escoamento em cada geometria proposta durante as otimizações foi resolvido numericamente com o código SU2, sendo o escoamento modelado matematicamente pelo modelo invíscido, representado pelas equações de Euler, as quais foram resolvidas pelo Método dos Volumes Finitos. O problema foi abordado de maneira bidimensional axissimétrica. Os resultados indicam que o aumento no número de pontos de otimização não produz aumento significativo no coeficiente de empuxo, sendo que esse incremento é mais expressivo na variação de um para dois pontos de otimização. Quanto ao efeito do refino da malha nas geometrias ótimas, verificou-se que com o refino da malha as coordenadas radiais dos pontos de otimização apresentam menores diferenças e tendem a um valor fixo. A maior diferença relativa nas coordenadas radiais entre as duas malhas mais refinadas utilizadas foi de 0,66%. Aumento de desempenho no coeficiente de empuxo de até 1,37% em relação ao perfil base foi obtido nas geometrias otimizadas. Ainda, obtiveram-se geometrias de tubeiras com coeficientes de empuxo superior a uma tubeira com o perfil do divergente gerado pelo método de Rao, com diferenças relativas de até 0,188%. Os coeficientes de empuxo dos perfis otimizados e dos perfis utilizados para comparação foram apresentados com estimativas de incerteza/erro numérico calculadas com o estimador GCI, o estimador Convergente e o estimador baseado na Multiextrapolação de Richardson. Abstract: The optimization of the divergent section's geometry of rocket engines' nozzles is investigated in the present work. The need for optimized geometries for rocket engines' nozzles is justified by fuel consumption savings that can be achieved with the design of better propulsion systems, enabling the reduction of propellant mass in the vehicle, which represents approximately 80% of the mass of aerospace vehicles. In this work, it is sought geometries of the divergent that allow the maximization of the nozzle's performance, evaluated quantitatively by the thrust coefficient generated by the nozzle. Two cases were considered: fixed exit radius and variable exit radius. The optimization methodology employed uses the shape optimization method, which uses the Differential Evolution optimization algorithm of the Pagmo library, selected after the evaluation and comparison of the performance of different stochastic optimization algorithms, evaluated in the problem in question. In the optimization process, the optimizer suggests the radial coordinate of points distributed on the wall of the divergent section of the nozzle. Optimizations were performed with one, two, three and four optimization points in order to assess the effect of the number of optimization points on the optimized geometries. The axial coordinates of these optimization points were defined in optimizations in coarse meshes considering both the axial and radial coordinates as variables. The effect of mesh refinement on optimal geometries was also assessed. The flow in each geometry proposed during the optimizations was numerically solved with the SU2 code, and the flow was mathematically modeled by the inviscid model, represented by the Euler's equations, which were solved by the Finite Volume Method. The problem was approached in a two-dimensional axisymmetric way. The results indicate that the increase in the number of optimization points does not produce a significant increase in the thrust coefficient, and this increment is more expressive in the variation from one to two optimization points. Regarding the effect of the mesh refinement on the optimal geometries, it was found that with the mesh refinement, the radial coordinates of the optimization points show smaller differences and tend to a fixed value. The largest relative difference in radial coordinates between the two most refined meshes used was 0.66%. Performance increase on the thrust coefficient of up to 1.37% in relation to the base profile was obtained in the optimized geometries. Furthermore, it was obtained nozzle geometries with higher thrust coefficients than a nozzle with the divergent profile generated by Rao's method, with relative differences of up to 0.188%. The thrust coefficients of the optimized profiles and the ones used for comparison were presented with estimates of numerical uncertainty/error calculated with the GCI estimator, the Convergent estimator and the estimator based on the Repeated Richardson Extrapolation.
Collections
- Dissertações [129]