In real gas turbines, multiple nozzles are used instead of a single-nozzle; therefore, interactions between flames are inevitable. In this study, the effects of flame-flame interaction on the emission characteristics and lean blowout limit were analysed in a CH4-fueled single- and dual-nozzle combustor. OH* chemiluminescence imaging showed that a flame-interacting region, where the two flames from the nozzles were merged, was present in the dual-nozzle combustor, unlike the single-nozzle combustor. Flow-field measurements using particle image velocimetry confirmed that a faster velocity region was formed at the flame merging region, thereby hindering flame stabilisation. In addition, we compared the emission indices of NOx and CO between the two combustors. The emission indices of CO were not significantly different; however, a distinct effect of flame-flame interaction was indicated in NOx. To understand the effect of flame-flame interaction on NOx emissions, we measured temperature distribution using a multi-point thermocouple. Results showed that a wider high-temperature region was formed in the dual-nozzle combustor compared to the single-nozzle combustor; this was attributable to the high OH* chemiluminescence intensity in the flame-interacting region. Furthermore, it was confirmed that the size of this interacting region caused the deformation of the temperature distribution in the combustor, which can induce a difference in the increase ratio of NOx emission between high and low equivalence ratio ranges. In conclusion, we confirmed that flame-flame interaction significantly affected temperature distribution in the downstream of the flame, and the change in temperature distribution contributed primarily to the varying concentration of the emission gas.

The Lean Direct Injection (LDI) combustor is one of the low-emissions combustors with great potential in aero-engine applications, especially those with high overall pressure ratio. A preliminary design tool providing basic combustor sizing information and qualitative assessment of performance and emission characteristics of the LDI combustor within a short period of time will be of great value to designers. In this research, the methodology of preliminary aerodynamic design for a second-generation LDI (LDI-2) combustor was explored. A computer code was developed based on this method covering the design of air distribution, combustor sizing, diffuser, dilution holes and swirlers. The NASA correlations for NOx emissions are also embedded in the program in order to estimate the NOx production of the designed LDI combustor. A case study was carried out through the design of an LDI-2 combustor named as CULDI2015 and the comparison with an existing rich-burn, quick-quench, lean-burn combustor operating at identical conditions. It is discovered that the LDI combustor could potentially achieve a reduction in liner length and NOx emissions by 18% and 67%, respectively. A sensitivity study on parameters such as equivalence ratio, dome and passage velocity and fuel staging is performed to investigate the effect of design uncertainties on both preliminary design results and NOx production. A summary on the variation of design parameters and their impact is presented. The developed tool is proved to be valuable to preliminarily evaluate the LDI combustor performance and NOx emission at the early design stage.

The semi-analytical prediction of pollutants emissions from gas turbines in the conceptual design phase is addressed in this paper. The necessity of this work arose from an urgent need for a comprehensive model that can quickly provide data in the conceptual design phase. Based on the available inputs data in the initial phases of the design process, a chemical reactor network (CRN) is defined to model the combustion with a detailed chemistry. In this way, three different chemical mechanisms are studied for Jet-A aviation fuel. Furthermore, the droplet evaporation for liquid fuel and the non-uniformity in fuel-air mixture are modelled. The results of a developed augmented modelling tool are compared with the pollutants data of two annular engine's combustors. The CRN results have good agreement with the actual engine test rig emissions output. In conclusion, the augmented CRN has shown to be efficient in predicting engine emissions with a very short executing time (few seconds) using a small CPU requirement such as a personal computer.