Combustion Model Development Based On Flamelet Generated Manifolds And Application On Partially Premixed Combustion

The Gasoline partially premixed combustion(PPC)is a new combustion mode with great potential.To understand the potential combustion mechanism of this combustion mode different from traditional diesel and gasoline engines,researchers have carried out optical diagnostic experiments and direct numerical simulations and found that multiple combustion modes such as spontaneous auto-ignition and flame propagation coexist and profoundly affect engine combustion.However,there is still lack of understanding of combustion mechanism behind such coexistence of spontaneous auto-ignition and flame propagation,which may be ilucidated by high-precision large eddy numerical simulation.There are mainly two major challenges:On the one hand,there is a lack of a suitable combustion model.It is necessary to consider turbulence-chemistry interaction(TCI)and to capture the complex combustion process with coexistence of multiple combustion modes.It also should afford to the extremely high computational load of large eddy simulation of engines.The second is the lack of quantitative analysis approaches to distinguish multiple combustion modes in the combustion process.This paper develops a high-efficiency and high-precision numerical simulation platform for gasoline PPC engines based on the open source CFD code OpenFOAM.First,a flamelet generated manifolds(FGM)combustion model is developed and successfully applied to traditional turbulent spray combustion.And FGM has been further developed to make it more accurately predict the key characteristics of spray combustion.The FGM model is also developed and tested based on turbulent premixed flame.Finally,the FGM model is developed based on the gasoline PPC engine experiment and applied to the large eddy numerical simulation of the gasoline PPC optical engine.The chemical explosive mode analysis(CEMA)method of quantitative analysis is used to distinguish the multiple combustion modes.First,the FGM model is developed and verified in turbulent spray combustion.The FGM coupled with the presumed probability density function(pPDF)is verified by the spray combustion experiment of n-dodecane constant volume bomb based on ECN(engine combustion network),Spray A.An approach dealing with enthalpy loss is proposed to consider the temperature drop and reaction inhibition caused by evaporation.The RANS simulation of Spray A is carried out using double β-PDF for the mixture fraction and progress variable.In addition,the FGM coupled with transport probability density function(tPDF)method is developed,combined with the famous Euler stochastic field(ESF)combustion model,to obtain the FGM+ESF model.And FGM+ESF is used to simulate Spray A with RANS and LES.It is found that in RANS,the number of stochastic fields has a great influence,not only affecting the flame liftoff length,but also affecting the flame structure.Too few stochastic fields will cause the flame to be asymmetric,and the position of the OH peak will deviate greatly from the experiment.In LES,the macroscopic combustion characteristics are not affected by the number of stochastic fields.Compared with traditional FGM,FGM+ESF can predict the spontaneous ignition in the fuel rich zone.The FGM model is developed and verified in a turbulent premixed flame.Combining FGM and artificial thickened flame(ATF),the FGM+ATF model is obtained.Based on the public experimental data of the swirl burner(SwB)of Cambridge University,a large eddy simulation study is carried out,and it is found that the two coupled models can achieve better results than the traditional FGM,and can accurately capture the turbulent premixed flame structure.To apply FGM to high-dimensional control variables,a variety of techniques are used to solve the memory issue of FGM.The first is to combine the artificial neural network method,couple OpenFOAM and PyTorch,train the neural network to fit the flamelet database,and replace the look-up table in the CFD calculation,which greatly reduces the memory occupation.The second is to use the MPI shared memory technology to solve the common problem of loading the flamelet database in parallel computation.An FGM model suitable for gasoline PPC engines is developed and verified by actual engine experiments.Based on the optical engine experiment,the large eddy simulation of the gasoline PPC engine is carried out,combined with the CEMA method to diagnose the result,distinguished a variety of combustion modes,and analyzed its influence on heat release.The CEMA of the large eddy simulation results shows that the factors that contribute the most to the chemical explosion mode at the time of hightemperature auto-ignition are temperature and H2O2,and the maj or elementary reaction to CEM are H+O2(?)+OH,C3H5+HO2(?)OH+C2H3+CH2O and C2H4+OH(?)C2H3+H2O,and H2O2(?)OH+OH before the high-temperature auto-ignition.According to the results of CEMA:In the early stage of the two-stage heat release process,the spontaneous ignition mode plays an absolute dominant role;the flame propagation mode appears twice,once after the low-temperature heat release,which belongs to cool flame propagation;the second time is in the late stage of hightemperature heat release,it belongs to high-temperature flame propagation,which contribute a relatively large proportion of the total heat release rate.The trend of combustion mode affecting heat release rate is also found:the increase of heat release rate is always accompanied by an increase in spontaneous ignition mode;the decrease of heat release rate is always accompanied by an increase in flame propagation mode.