Large Eddy Simulations Of Turbulrnt Gas Combustion Using Detailed Tabulated Chemistry

Combustion,one of major ways of energy conversion,has an essential place in nowadays energy structures.For practical combustors,such as gas turbine,aero-engine,internal combustion engine,industrial furnace,etc.,turbulent combustion has always been encountered.As the name suggests,turbulent combustion contains two complex physical and chemical phenomena,one is the multi-scale turbulent flow,and the other is the multi-scale reacting process.And it determines the work efficiency and pollutant emissions of combustion system to a great extent,even for whether the system works securely and stably or not.Numerical simulations have significant meanings to optimized design burner and rationally organize combustion.Large eddy simulation(LES)is a popular way in turbulent combustion simulations,in which the large-scale structures are solved exactly along with the rests modeled.However,there are two issues required special handing in LES for turbulent stratified flames:one is the way to capture thin stratified flames under LES grid precision;the other is the way to couple detailed chemical kinetics mechanism efficiently.To solve the two above problems,sub-grid scale(SGS)combustion models are required.Therefore,the SGS combustion models are an important factor that directly affects the quality of large eddy simulation for turbulent combustion.In this paper,large eddy simulations of several typical turbulent combustion processes have been performed,including turbulent premixed/stratified flames,turbulent non-premixed flames and turbulent MILD combustion,aiming at developing SGS combustion models,analyzing turbulent combustion characteristics as well as seeking the major influence factors for pollutant emissions,the main work and conclusions are as follow,(1)LESs have been performed to investigate Cambridge/Sandia non-swirling premixed and stratified flames,SwB1 and SwB5 respectively,using algebraic flame surface density(FSD)model coupled with flamelet generated manifolds(FGM)method.In this coupled SGS combustion model(FSD-FGM),the FSD approach is used to describe combustion/turbulence interactions and the FGM method is employed to determine major scalars.The statistical results show that the FSD-FGM model is able to capture the vector and scalar changes of the investigated premixed and stratified flames.Judged from the probability distributions of equivalence ration and orientation angle of stratified flame SwB5,the local stratified flame belongs to premixed mode at upstream and back-supported mode at downstream.(2)LESs of turbulent premixed piloted flame-turbulent Bunsen flame F3 have been conducted using artificial thickened flame(ATF)model with FGM method.Based on the analysis of laminar flame structures,two self-contained flame sensors are used to track the diffusion and reaction processes with different spatial scales in the flame front,respectively.In addition,avoiding equilibrium hypothesis between turbulence and reaction,the dynamic wrinkling factor model is also performed to represent the flame area changes caused by turbulence and tested.The statistical results confirm that the static coupled SGS combustion model shows a marvelous ability in predicting the distributions of the temperature and velocity,while the distributions of the species need to be improved.The dynamic model has certain advantages in predicting the distributions of temperature in the flame front at downstream and the distributions of radial velocity at upstream with an acceptable increasing CPU time,which indicates that the dynamic model can preserve the premixed flame propagation characteristics better.(3)LESs of Cambridge/Sandia turbulent stratified flame series have been performed.The moderate stratified(SwB6)and highly stratified(SwB10)flames under moderate swirling conditions are investigated using two different SGS combustion models,namely FSD-FGM model as well as DTF-FGM model respectively.The statistical results show that the DTF-FGM model is more superior along with the peak values and locations of mass fraction of intermediate species CO captured correctly.Analyzed from the axial distributions of local swirl number,an increasing local swirl number is observed in cold case as a longer and stronger recirculation zone.Whereas because of a relative shorter and weaker recirculation zone as well as the flow field radial dilation,the local swirl number in two stratified flames decreases along the axial direction.Judged from the distributions of probability density functions of strain rate component,the compression is dominating as for the effects of turbulence on flame front in the two stratified flames.(4)LESs of Sandia turbulent non-premixed flame D and E have been carried out using FGM method coupled with three kinds of presumed probability density function(PPDF),the latter is included to consider the interactions between turbulence and flame(TFIs).Besides,an additional transport equation of NO with different modeling strategies is also solved to capture its emissions.The statistical results indicate that the TFIs can be described reasonably by all three PPDF,the difference between them is mainly reflected in the predicted distributions of NO.Moreover,the ATF model could also be used to simulate NO formation,but it requires developing more advanced wrinkling model and flame sensor for further improvement.Seen from the scatter distributions,the regions of high temperature and high mass fraction of NO of Flame D and E are both mainly located around ?=1.0 and rich side.Compared with Flame D,the local extinction is more obvious in Flame E,meanwhile the emission of NO is lower.The mass fraction of NO of both flames decreases rapidly along the scalar dissipation rate of mixture fraction firstly,and then keeps nearly unchanged.Also,its peak values are mainly in the region where the scalar dissipation rate of mixture fraction is relative small.(5)LESs of turbulent moderate or intense low-oxygen dilution(MILD)combustion of Adelaide Jet in Hot Coflow(AJHC)have been performed.Two typical detailed tabulated chemistry method,namely perfectly stirred reactor(PSR)model and closed homogeneous batch reactor(AI)model,are compared.The TFIs are also modeled by three kinds of PPDF respectively.The statistical results show that the performances of two tabulation methods are similar.Moreover,the optimal PDF of different scalars is not the same.Proven by Budget analysis,the flame base of HM3 is controlled by auto-ignition,while that of HM1 is the result of both mixing and reaction.The distributions of NO within the flame core area are similar for both flames.The mass fraction of NO of HM1 is smaller than that of HM3 with almost one order of magnitude,which demonstrates the reduction of NO emission can be achieved by reducing oxygen concentration.Judged from the Pearson correlation coefficients between different scalars and the mass fraction of NO,the mixture fraction has a comparable important effect on NO formation in HM3 flame with temperature,while the correlation between mixture fraction and NO formation is much lower in HM1 flame.