چكيده به لاتين
In the present dissertation, optimization and sensitivity analysis of an aviation model gas turbine has been done. The aim of this study is to simultaneously reduce NOx pollutant and improve the combustor pattern factor. For analyzing liquid fuel spray, including distribution, breakup, and evaporation of the droplets, Eulerian-Lagrangian method is used. To investigate the characteristics of the reactive spray flow, Reynolds Average Navier-Stokes (RANS) approach (Realizable k-ɛ model), discrete ordinates radiation heat transfer model (DO), and laminar flamelet combustion model with C10H22 chemical mechanism are used. NOx pollutant is modelled as a post-processing with a finite rate model. Discrete ordinates radiation heat transfer model is capable in a wide range of optical thicknesses and is frequently used in combustion modeling. Furthermore, by using sensitivity analysis, susceptibility of the output parameters to the input parameters are studied. In other words, in this method, the influence of the input parametrs on the output parametrs is investigated by making sensible changes in the input parameters. The input parameters include diameter, angle, location, and temperature of the stabilizer jets. By optimization of the results, the best member of the data set, which simultaneously minimizes NOx pollutant and pattern factor, is selected. The numerical data is produced by design of experiments (DOE) and full factorial method. The obtained results from sensitivity analysis show that temperature is the most influential variable on the output parameters. In the next step, the obtained results are trained by neural network and then these trained results are used in genetic algorithm. Moreover, optimization results show a reduction of 5.14% in pattern factor and 38.7% in NOx pollutant.
