Gas Turbine Combustor Simulation With Various Turbulent Combustion Models

Abstract

Along with the development of computing technology, large-eddy simulation turns to be a useful tool for practical study. For fast estimation, the front line researchers still use the Reynolds-averaged Navier-Stokes (RANS) method nowadays. RANS still is the major tool for gas turbine chamber (GTC) designers, but there is not a universal method in RANS GTC spray combustion simulation at present especially for the two-phase turbulent combustion. Usually there are two main steps in two-phase combustion: the liquid fuel evaporation and the gas mixture combustion. Thus, three widely used turbulent combustion models: the Eddy-Break-Up and Arrhenius model (EBU), Laminar Flame-let Model (LFM) and Eddy-Dissipation-Concept (EDC) turbulent combustion models are firstly tested against a methane-air turbulent gas jet flame (Flame D) measured by Sandia Lab and next a two-phase turbulent swirl spray combustion in a complex GTC. The predictions of the LFM model are the best in jet flame simulation to show its ability in gas combustion prediction. The comparison between the simulation results and the experimental results showed that LFM model could properly consider the interaction between turbulence and chemistry in the gas combustion in most regions; EBU model overestimated the turbulent effect in most regions; though EDC model takes the chemistry effect into account, the turbulence seems be overestimated too. The simulated GTC performed well in experiments especially when the fuel-air mixture equivalence ratio (MER) in its main-reaction-zone (MRZ) is 0.7, so the three combustion models are all applied in this case, with the same 90° spray angel, same material properties and the same discrete ordinates (DO) radiation model. In LFM prediction, the high temperature regions are distributed around the margin of the circumfluence zone and the downstream regions after MRZ, which does not agree with the test observation. The LFM model deals well with the gas combustion, so the reason for this poor performance must be of kerosene evaporation. LFM model is a fast-chemistry model, but the kerosene reaction rate is not very fast and the evaporation makes the global reaction slower. Furthermore the mixture fraction is a conservation scalar in FLM model but it is changed by the kerosene evaporation especially in the MRZ where the kerosene was mainly vaporized. Generally, the EBU and EDC results are better: the high temperature regions are mostly in MRZ when MER is 0.7. The EDC model also has good predictions of different MERs in MRZ. When MER is 1.3, the unburned kerosene continue reaction after primary-air-holes; when MER is 0.3, there is nearly no kerosene there. Additionally, effects of the spray angle, material property are studied.