Developoment of solid oxide fuel cell (SOFC) ejector flow channel by numerical methods

The ever-increasing costs and environmental effects of fossil energy resources increase the importance of efficiency in energy systems. Solid oxide fuel cells (SOFC) under development enable micro-combined heat and power generation (mCHP) in residential buildings, thus, distributed electricity production with the highest efficiency. Recirculation of the anode exhaust gas not only increases the efficiency of the SOFCs whose dimensions are limited due to their high costs but also provides water vapor and heat energy to the fuel reformer at the SOFC inlet. Ejectors are preferred for recirculation of the high temperature (~1000oC) anode exhaust gas for low-power SOFCs (Pel < 5 kWe). The ejector's recirculation rate (RR), which directly affects the performance of the SOFC, depends on the operating conditions and the compressible flow structures varying with the geometric dimensions of cross-sections along the main flow channel. In the presented thesis, the main flow channel of the ejector that recirculates hot anode exhaust gas of a SOFC generating 2 kWe of electricity is designed. To this end, the effects of suction chamber constriction angle (α2), suction chamber length (Ls), mixing chamber diameter (Dm), mixing chamber length (Lm), diffuser expansion angle (α3), diffuser length (Ld) and the length of the divergent section of the primary nozzle (Lnd) on the RR were investigated by computational fluid dynamics (CFD) and Taguchi experimental design method. First of all, the geometry of the critical cross-sections such as throat diameter (Dt) of the convergent-divergent primary nozzle was determined with the 1-D thermo-hydrodynamic compressible flow models suggested for ejectors in the literature. Then, three levels were selected for the ranges recommended in the open literature for each of the other geometric dimensions of the main flow channel of the ejector that could not be determined directly by calculations. In the design of the ejector models used in the first stage of CFD experiments, the levels of the five selected flow channel geometric parameters were matched according to the L27(35) orthogonal array suggested by Taguchi. CFD analyses were performed with SolidWorks Flow Simulation software. RR calculated with the results of CFD analyses were evaluated according to the "the-larger the-better" signal to noise (S/N) ratio proposed by Taguchi. The optimum levels and percentace severity for the considered five geometric parameters in the first stage of the analyzes are determined respectively as α3= 9° (32.09%), Ls= 8,1 mm (23.54%), α2= 15° (16.97%), Lm= 24.4 mm (11.23%) and Ld= 48.7 mm (0,1%). These results are consistent with the literature. In the second stage analyses for three different mixing chamber diameters (Dm/Dt= 6, 9 ve 12), CFD analyses were performed on 27 ejector models prepared for each mixing chamber diameter (81 different ejector models in total). In these analyses, the closest results to the targeted RR of 7.2 for SOFC were obtained with the ejector models prepared for Dm/Dt= 9. The optimum levels and percentage severity for the considered six geometric parameters in the second stage of the analysis are determined respectively as α3= 7° (35.01%), Lm= 27,3mm (25.85%), Lnd= 0,58mm (12.75%), α2= 25° (7.65%), Ld= 54,5mm (7.13%) and Ls= 3,0mm (1,67%). These results are consistent with the literature. The most important geometric design parameters of the ejector flow channel in terms of RR are the diameter (Dm) and length (Lm) of the mixing chamber, as expected. As the diameter of the mixing chamber increases, the required length becomes shorter and the optimal ratios of these two parameters were determined as Dm/Dt= 9 and Lm/Dm= 5-7. Diffuser expansion angle (α3) is also one of the important parameters and the determined optimal angle is 8° in accordance with the literature. The effect of the suction chamber's constriction angle (α2) on the RR is less than the parameters mentioned so far. This angle should be greater than 20° according to the obtained results. As the diameter of the mixing chamber increases, the effects of the suction chamber and primary nozzle divergent section lengths on the GBO increase. The diffuser length is the ejector flow channel parameter with the lowest effect on the RR. Therefore, by making the diffuser as short as possible, the overall length of the ejector can be shortened as well. As a general conclusion, it has been shown that main flow channel design of SOFC-specific anode exhaust gas recirculation ejector can be developed by numerical methods (Taguchi experimental design and CFD analysis).