In this study,the flow field structure inside a scramjet combustor is numerically simulated using the flamelet/progress variable model.Slope injection is considered,with fuel mixing enhanced by means of a streamwise v...In this study,the flow field structure inside a scramjet combustor is numerically simulated using the flamelet/progress variable model.Slope injection is considered,with fuel mixing enhanced by means of a streamwise vortex.The flow field structure and combustion characteristics are analyzed under different conditions.Attention is also paid to the identification of the mechanisms that keep combustion stable and support enhanced mixing.The overall performances of the combustion chamber are discussed.展开更多
An OpenFOAM based turbulence combustion solver with flamelet generated manifolds (FGMs) is presented in this paper. A series of flamelets, representative for turbulent flames, are calculated first by a one-dimensional...An OpenFOAM based turbulence combustion solver with flamelet generated manifolds (FGMs) is presented in this paper. A series of flamelets, representative for turbulent flames, are calculated first by a one-dimensional (1D) detailed chemistry solver with the consideration of both transport and stretch/curvature contributions. The flame structure is then parameterized as a function of multiple reaction control variables. A manifold, which collects the 1D flame properties, is built from the 1D flame solutions. The control variables of the mixture fraction and the progress variable are solved from the corresponding transport equations. During the calculation, the scalar variables, e.g., temperature and species concentration, are retrieved from the manifolds by interpolation. A transport equation for NO is solved to improve its prediction accuracy. To verify the ability to deal with the enthalpy loss effect, the temperature retrieved directly from the manifolds is compared with the temperature solved from a transport equation of absolute enthalpy. The resulting FGM-computational fluid dynamics (CFD) coupled code has three significant features, i.e., accurate NO prediction, the ability to treat the heat loss effect and the adoption at the turbulence level, and high quality prediction within practical industrial configurations. The proposed method is validated against the Sandia flame D, and good agreement with the experimental data is obtained.展开更多
An automated method to optimize the definition of the progress variables in the flamelet-based dimension reduction is proposed. The performance of these optimized progress variables in coupling the flamelets and flow ...An automated method to optimize the definition of the progress variables in the flamelet-based dimension reduction is proposed. The performance of these optimized progress variables in coupling the flamelets and flow solver is presented. In the proposed method, the progress variables are defined according to the first two principal components （PCs） from the principal component analysis （PCA） or kernel-density-weighted PCA （KEDPCA） of a set of flamelets. These flamelets can then be mapped to these new progress variables instead of the mixture fraction/conventional progress variables. Thus, a new chemistry look-up table is constructed. A priori validation of these optimized progress variables and the new chemistry table is implemented in a CH4/N2/air lift-off flame. The reconstruction of the lift-off flame shows that the optimized progress variables perform better than the conventional ones, especially in the high temperature area. The coefficient determinations （R2 statistics） show that the KEDPCA performs slightly better than the PCA except for some minor species. The main advantage of the KEDPCA is that it is less sensitive to the database. Meanwhile, the criteria for the optimization are proposed and discussed. The constraint that the progress variables should monotonically evolve from fresh gas to burnt gas is analyzed in detail.展开更多
To improve the NO modelling in turbulent flames,the flamelet/progress variable(FPV)model is extended by introducing NO mass fraction into the progress variable and incorporating an additional NO transport equation.Two...To improve the NO modelling in turbulent flames,the flamelet/progress variable(FPV)model is extended by introducing NO mass fraction into the progress variable and incorporating an additional NO transport equation.Two sets of flamelet databases are tabulated with progress variables based on major species and NO mass fraction,respectively.The former is used for the acquisition of the main thermochemical variables,while the latter is employed for NO modelling.Moreover,an additional transport equation is solved to obtain the NO mass fraction,with the source term corrected using the scale similarity method.Model assessments are first conducted on laminar counterflow diffusion flames to identify lookup-related errors and assess the suitability of progress variable definitions.The results show that the progress variables based on major species and NO could correctly describe the main thermochemical quantities and NO-related variables,respectively.Subsequently,the model is applied to the large eddy simulation(LES)of Sandia flames.The results indicate that the extended FPV model improves the NO prediction,with a mean error for NO prediction at 55%,significantly lower than those of existing FPV models(130%and 385%).The LES with the extended FPV model quantitatively captures NO suppression in the mid-range of Reynolds numbers from 22400(Flame D)to 33600(Flame E),but underestimates the NO suppression at higher Reynolds numbers from 33600 to 44800(Flame F).This underprediction is primarily attributed to the underestimation of local extinction levels in flames with high Reynolds numbers.展开更多
In order to develop a turbulent diffusion combustion model for supersonic flow, the physical argument of the extension of the flamelet model to supersonic flow was presented, and the flow field of a hydrogen/air diffu...In order to develop a turbulent diffusion combustion model for supersonic flow, the physical argument of the extension of the flamelet model to supersonic flow was presented, and the flow field of a hydrogen/air diffusion combustion generated by axisymmetric supersonic jets was numerically simulated by employing the flamelet model. Using the experimental data, value of the model coefficient of scalar dissipation in the flamelet model was revised specifically for supersonic flow. The computational results of the modified flamelet model were compared with the experimental results, and it was indicated that the precision of the modified flamelet model was satisfying. Based on the numerical results and flamelet theory, the influence mechanisms of turbulence fluctuation on the average state equation and chemical reaction rate were studied for the first time. It was found that the fluctuation correlation of species mass fractions and temperature has little effect on the averaged gas state equation; the temperature fluctuation decreases the product of H2O, but its effect is small; the fluctuation of species mass fractions increases the product of H2O in the region close to oxidizer while decreases the product of H2O in other regions; the fluctuation correlation of species mass fractions and temperature largely decreases the product of H2O.展开更多
Being physically intuitionistic and computationally efficient, the flamelet model has obtained increasing attention and becomes popular in numerical simulations of supersonic combustion. However, the flamelet model wa...Being physically intuitionistic and computationally efficient, the flamelet model has obtained increasing attention and becomes popular in numerical simulations of supersonic combustion. However, the flamelet model was firstly built for simulations of the low-speed flows. Therefore it is still unclear whether the assumption of the fiamelet model is reasonable in supersonic combustion. This paper tries to identify the existence of the flamelet mode in supersonic combustion. Firstly, the interaction of the turbulent fluctuation and the flame is discussed, based on which the combustion regime is distinguished. Then the charac- teristic length scale of the turbulent flows and the combustion inside the scramjet combustor are calculated and compared, which are used to identify whether the assumption of the flamelet model is reasonable in supersonic flows. The results show that for premixed combustion, due to the low fluctuation velocity in the recirculafion zone and the shear layer, the assumption of the flamelet model is established for all the flight Mach numbers, while for non-premixed combustion the assumption is also established for most of the flight Mach numbers except for very high Mach number under which the slow reaction mode dominates the combustion. In order to quantitatively examine the combustion mode, numerical calculations are performed to simulate the strut-injection supersonic combustion which has been experimentally investigated in German Aerospace Center. The results show that the supersonic combustion occurs in the fully-developed turbulent regions and the assumption of the flamelet model is established in the whole flow field.展开更多
In the present work,artificial neural networks(ANN)technique combined with flamelet generated manifolds(FGM)is proposed to mitigate the memory issue of FGM models.A set of ANN models is firstly trained using a 68-spec...In the present work,artificial neural networks(ANN)technique combined with flamelet generated manifolds(FGM)is proposed to mitigate the memory issue of FGM models.A set of ANN models is firstly trained using a 68-species mass fractions in mixture fraction-progress variable space.The ANN prediction accuracy is examined in large eddy simulation(LES)and Reynolds averaged Navier-Stokes(RANS)simulations of spray combustion.It is shown that the present ANN models can properly replicate the FGM table for most of the species mass fractions.The network models with relative error less than 5%are considered in RANS and LES to simulate the Engine Combustion Network(ECN)Spray H flames.Validation of the method is firstly conducted in the framework of RANS.Both non-reacting and reacting cases show the present method predicts very well the trend of spray and combustion process under different ambient temperatures.The results show that FGM-ANN can replicate the ignition delay time(IDT)and lift-off length(LOL)precisely as the conventional FGM method,and the results agree very well with the experiments.With the help of ANN,it is possible to achieve high efficiency and accuracy,with a significantly reduced memory requirement of the FGM models.LES with FGM-ANN is then applied to explore the detailed spray combustion process.Chemical explosive mode analysis(CEMA)approach is used to identify the local combustion modes.It is found that before the spray flame is developed to the steady-state,the high CH_(2)O zone is always associated with ignition mode.However,high CH_(2)O zone together with high OH zone is dominated by the burned mode after the steady-state.The lift-off position is dominated mainly by the diffusion mode.展开更多
The Reynolds analogy concept has been used in almost all turbulent reacting flow RANS(Reynoldsaveraged Navier–Stokes)simulations,where the turbulence scalar transfers in flow fields are calculated based on the modele...The Reynolds analogy concept has been used in almost all turbulent reacting flow RANS(Reynoldsaveraged Navier–Stokes)simulations,where the turbulence scalar transfers in flow fields are calculated based on the modeled turbulence momentum transfer.This concept,applied to a lean premixed combustion system,was assessed in this paper in terms of exit temperature distribution.Because of the isotropic assumption involved in this analogy,the prediction in some flow condition,such as jet cross flow mixing,would be inaccurate.In this study,using Flamelet Generated Manifold as reaction model,some of the numerical results,obtained from an annular combustor configuration with the turbulent Schmidt number varying from 0.85 to 0.2,were presented and compared with a benchmark atmospheric test results.It was found that the Schmidt numberσt in mean mass fraction f transport equation had significant effect on dilution air mixing process.The mixing between dilution air and reaction products from the primary zone obviously improved asσt decreased on the combustor exit surface.Meanwhile,the sensitivity ofσt in three turbulence models including Realizable k-ε,SST(Shear Stress Transport)and RSM(Reynolds Stress Model)has been compared as well.Since the calculation method of eddy viscosity was different within these three models,RSM was proved to be less sensitive than another two models and can guarantee the best prediction of mixing process condition.On the other hand,the results of dilution air mixing were almost independent of Schmidt number Sct in progress variable c transport equation.This study suggested that for accurate prediction of combustor exit temperature distribution in steady state reacting flow simulation,the turbulent Schmidt number in steady state simulation should be modified to cater to dilution air mixing process.展开更多
Flamelet Generated Manifold(FGM)is an example of a chemistry tabulation or a flamelet method that is under attention because of its accuracy and speed in predicting combustion characteristics.However,the main problem ...Flamelet Generated Manifold(FGM)is an example of a chemistry tabulation or a flamelet method that is under attention because of its accuracy and speed in predicting combustion characteristics.However,the main problem in applying the model is a large amount of memory required.One way to solve this problem is to apply machine learning(ML)to replace the stored tabulated data.Four different machine learning methods,including two Artificial Neural Networks(ANNs),a Random Forest(RF),and a Gradient Boosted Trees(GBT),are trained,validated,and compared in terms of various performance measures.The progress variable source term and transport properties are replaced with the ML models.Particular attention was paid to the progress variable source term due to its high gradient and wide range of its value in the control variables space.Data preprocessing is shown to play an essential role in improving the performance of the models.Two ensemble models,namely RF and GBT,exhibit high training efficiency and acceptable accuracy.On the other hand,the ANN models have lower training errors and take longer to train.The four models are then combined with a one-dimensional combustion code to simulate a counterflow non-premixed diffusion flame in engine-relevant conditions.The predictions of the ML-FGM models are compared with detailed chemical simulations and the original FGM model for key combustion properties and representative species profiles.展开更多
Owing to the increasing worldwide demand for natural gas,the development of a large submerged combustion vaporizer is required.Its burner is equipped with a water spray nozzle to reduce nitrogen oxides,and a practi-ca...Owing to the increasing worldwide demand for natural gas,the development of a large submerged combustion vaporizer is required.Its burner is equipped with a water spray nozzle to reduce nitrogen oxides,and a practi-cal simulation method is required for the optimal design.The non-adiabatic flamelet approach can predict the combustion emissions and is useful for reducing simulation costs.However,as the number of control variables increases,the database requires larger memory and cannot be dealt with by general computers.In this study,an artificial neural network(ANN)model based on a five-dimensional flamelet database,which includes the effects of heat loss and vapor concentration by sprayed water evaporation,is developed.Furthermore,large eddy sim-ulations(LESs)for turbulent combustion fields with and without water spray are conducted employing flamelet generated manifold(FGM)approach with this ANN model,and the validity is investigated.For comparison,a lab-scale burner equipped with a water spray nozzle is manufactured,and combustion experiments with and without water spray are conducted.The results show that CO,NO,temperature,and reaction rate of progress variable predicted by the present ANN model are in good agreement with those of a five-dimensional flamelet database.In the condition without water spray,the flame behavior predicted by the LES employing the FGM/ANN ap-proach is in good agreement with that employing the conventional FGM approach,while indicating much lower memory,although there appeared some quantitative discrepancies in the temperature against the experiment probably partially because of the insufficiency of the FGM approach for the present complex flame structure.In the condition with water spray,the LES employing the FGM/ANN approach is able to capture the effect of the water spray on the flame behavior in the experiment,such that the water spray decreases the temperature,which causes the decrease in NO but increase in CO.展开更多
In turbulent combustion simulations, the flow structure at the unresolved scale level needs to be reasonably modeled. Following the idea of turbulent flamelet equation for the non-premixed flame case, which was derive...In turbulent combustion simulations, the flow structure at the unresolved scale level needs to be reasonably modeled. Following the idea of turbulent flamelet equation for the non-premixed flame case, which was derived based on the filtered governing equations(L. Wang, Combust. Flame 175, 259(2017)), the scalar dissipation term for tabulation can be directly computed from the resolved flowing quantities, instead of solving species transport equations. Therefore, the challenging source term closure for the scalar dissipation or any assumed probability density functions can be avoided;meanwhile the chemical sources are closed by scaling relations. The general principles are discussed in the context of large eddy simulation with case validation. The new model predictions of the bluff-body flame show sufficiently improved results, compared with these from the classic progress-variable approach.展开更多
The effects of heat loss on the structure of laminar flamelets,which are the constitutive elements of turbulent flames under the most common operating conditions,are investigated for typical aeronautical gas-turbine o...The effects of heat loss on the structure of laminar flamelets,which are the constitutive elements of turbulent flames under the most common operating conditions,are investigated for typical aeronautical gas-turbine operating conditions at take-off.The magnitude of heat loss is quantified via the "enthalpy defect" measured with respect to an adiabatic flame.A procedure to generate laminar flamelets with an assigned enthalpy defect at the boundaries is devised and applied to nonpremixed propane/air flames,as propane reproduces the essential features of higher hydrocarbon combustion.It is found,contrary to commonly held beliefs,that the enthalpy defect has a significant effect on the concentration not only of minor species,but also of main reaction products.Such effects are found in general to be more pronounced for fuel-rich conditions.An impact is anticipated on the formation rate of nitric oxides.The effects of scalar dissipation rate are also discussed.展开更多
基金This work was supported by the National Natural Science Foundation of China(No.12002193)the Shandong Provincial Natural Science Foundation,China(No.ZR2019QA018).
文摘In this study,the flow field structure inside a scramjet combustor is numerically simulated using the flamelet/progress variable model.Slope injection is considered,with fuel mixing enhanced by means of a streamwise vortex.The flow field structure and combustion characteristics are analyzed under different conditions.Attention is also paid to the identification of the mechanisms that keep combustion stable and support enhanced mixing.The overall performances of the combustion chamber are discussed.
基金Project supported by the Strategic Priority Research Program of the Chinese Academy of Sciences(No.XDA 21060102)Guangdong Provincial Key Laboratory of New and Renewable Energy Research and Development of China(No.y809jh1001)
文摘An OpenFOAM based turbulence combustion solver with flamelet generated manifolds (FGMs) is presented in this paper. A series of flamelets, representative for turbulent flames, are calculated first by a one-dimensional (1D) detailed chemistry solver with the consideration of both transport and stretch/curvature contributions. The flame structure is then parameterized as a function of multiple reaction control variables. A manifold, which collects the 1D flame properties, is built from the 1D flame solutions. The control variables of the mixture fraction and the progress variable are solved from the corresponding transport equations. During the calculation, the scalar variables, e.g., temperature and species concentration, are retrieved from the manifolds by interpolation. A transport equation for NO is solved to improve its prediction accuracy. To verify the ability to deal with the enthalpy loss effect, the temperature retrieved directly from the manifolds is compared with the temperature solved from a transport equation of absolute enthalpy. The resulting FGM-computational fluid dynamics (CFD) coupled code has three significant features, i.e., accurate NO prediction, the ability to treat the heat loss effect and the adoption at the turbulence level, and high quality prediction within practical industrial configurations. The proposed method is validated against the Sandia flame D, and good agreement with the experimental data is obtained.
基金Project supported by the National Natural Science Foundation of China(Nos.50936005,51576182,and 11172296)
文摘An automated method to optimize the definition of the progress variables in the flamelet-based dimension reduction is proposed. The performance of these optimized progress variables in coupling the flamelets and flow solver is presented. In the proposed method, the progress variables are defined according to the first two principal components （PCs） from the principal component analysis （PCA） or kernel-density-weighted PCA （KEDPCA） of a set of flamelets. These flamelets can then be mapped to these new progress variables instead of the mixture fraction/conventional progress variables. Thus, a new chemistry look-up table is constructed. A priori validation of these optimized progress variables and the new chemistry table is implemented in a CH4/N2/air lift-off flame. The reconstruction of the lift-off flame shows that the optimized progress variables perform better than the conventional ones, especially in the high temperature area. The coefficient determinations （R2 statistics） show that the KEDPCA performs slightly better than the PCA except for some minor species. The main advantage of the KEDPCA is that it is less sensitive to the database. Meanwhile, the criteria for the optimization are proposed and discussed. The constraint that the progress variables should monotonically evolve from fresh gas to burnt gas is analyzed in detail.
基金supported by the National Natural Science Foundation of China(51906075)the National Key Research and Development Program of China(2019YFE0100100).
文摘To improve the NO modelling in turbulent flames,the flamelet/progress variable(FPV)model is extended by introducing NO mass fraction into the progress variable and incorporating an additional NO transport equation.Two sets of flamelet databases are tabulated with progress variables based on major species and NO mass fraction,respectively.The former is used for the acquisition of the main thermochemical variables,while the latter is employed for NO modelling.Moreover,an additional transport equation is solved to obtain the NO mass fraction,with the source term corrected using the scale similarity method.Model assessments are first conducted on laminar counterflow diffusion flames to identify lookup-related errors and assess the suitability of progress variable definitions.The results show that the progress variables based on major species and NO could correctly describe the main thermochemical quantities and NO-related variables,respectively.Subsequently,the model is applied to the large eddy simulation(LES)of Sandia flames.The results indicate that the extended FPV model improves the NO prediction,with a mean error for NO prediction at 55%,significantly lower than those of existing FPV models(130%and 385%).The LES with the extended FPV model quantitatively captures NO suppression in the mid-range of Reynolds numbers from 22400(Flame D)to 33600(Flame E),but underestimates the NO suppression at higher Reynolds numbers from 33600 to 44800(Flame F).This underprediction is primarily attributed to the underestimation of local extinction levels in flames with high Reynolds numbers.
文摘In order to develop a turbulent diffusion combustion model for supersonic flow, the physical argument of the extension of the flamelet model to supersonic flow was presented, and the flow field of a hydrogen/air diffusion combustion generated by axisymmetric supersonic jets was numerically simulated by employing the flamelet model. Using the experimental data, value of the model coefficient of scalar dissipation in the flamelet model was revised specifically for supersonic flow. The computational results of the modified flamelet model were compared with the experimental results, and it was indicated that the precision of the modified flamelet model was satisfying. Based on the numerical results and flamelet theory, the influence mechanisms of turbulence fluctuation on the average state equation and chemical reaction rate were studied for the first time. It was found that the fluctuation correlation of species mass fractions and temperature has little effect on the averaged gas state equation; the temperature fluctuation decreases the product of H2O, but its effect is small; the fluctuation of species mass fractions increases the product of H2O in the region close to oxidizer while decreases the product of H2O in other regions; the fluctuation correlation of species mass fractions and temperature largely decreases the product of H2O.
基金supported by the National Natural Science Foundation of China (Grant No. 50906098)
文摘Being physically intuitionistic and computationally efficient, the flamelet model has obtained increasing attention and becomes popular in numerical simulations of supersonic combustion. However, the flamelet model was firstly built for simulations of the low-speed flows. Therefore it is still unclear whether the assumption of the fiamelet model is reasonable in supersonic combustion. This paper tries to identify the existence of the flamelet mode in supersonic combustion. Firstly, the interaction of the turbulent fluctuation and the flame is discussed, based on which the combustion regime is distinguished. Then the charac- teristic length scale of the turbulent flows and the combustion inside the scramjet combustor are calculated and compared, which are used to identify whether the assumption of the flamelet model is reasonable in supersonic flows. The results show that for premixed combustion, due to the low fluctuation velocity in the recirculafion zone and the shear layer, the assumption of the flamelet model is established for all the flight Mach numbers, while for non-premixed combustion the assumption is also established for most of the flight Mach numbers except for very high Mach number under which the slow reaction mode dominates the combustion. In order to quantitatively examine the combustion mode, numerical calculations are performed to simulate the strut-injection supersonic combustion which has been experimentally investigated in German Aerospace Center. The results show that the supersonic combustion occurs in the fully-developed turbulent regions and the assumption of the flamelet model is established in the whole flow field.
文摘In the present work,artificial neural networks(ANN)technique combined with flamelet generated manifolds(FGM)is proposed to mitigate the memory issue of FGM models.A set of ANN models is firstly trained using a 68-species mass fractions in mixture fraction-progress variable space.The ANN prediction accuracy is examined in large eddy simulation(LES)and Reynolds averaged Navier-Stokes(RANS)simulations of spray combustion.It is shown that the present ANN models can properly replicate the FGM table for most of the species mass fractions.The network models with relative error less than 5%are considered in RANS and LES to simulate the Engine Combustion Network(ECN)Spray H flames.Validation of the method is firstly conducted in the framework of RANS.Both non-reacting and reacting cases show the present method predicts very well the trend of spray and combustion process under different ambient temperatures.The results show that FGM-ANN can replicate the ignition delay time(IDT)and lift-off length(LOL)precisely as the conventional FGM method,and the results agree very well with the experiments.With the help of ANN,it is possible to achieve high efficiency and accuracy,with a significantly reduced memory requirement of the FGM models.LES with FGM-ANN is then applied to explore the detailed spray combustion process.Chemical explosive mode analysis(CEMA)approach is used to identify the local combustion modes.It is found that before the spray flame is developed to the steady-state,the high CH_(2)O zone is always associated with ignition mode.However,high CH_(2)O zone together with high OH zone is dominated by the burned mode after the steady-state.The lift-off position is dominated mainly by the diffusion mode.
基金funded by the Key Programs of Chinese Academy of Sciences (Project No. ZDRW-CN-2017-2)
文摘The Reynolds analogy concept has been used in almost all turbulent reacting flow RANS(Reynoldsaveraged Navier–Stokes)simulations,where the turbulence scalar transfers in flow fields are calculated based on the modeled turbulence momentum transfer.This concept,applied to a lean premixed combustion system,was assessed in this paper in terms of exit temperature distribution.Because of the isotropic assumption involved in this analogy,the prediction in some flow condition,such as jet cross flow mixing,would be inaccurate.In this study,using Flamelet Generated Manifold as reaction model,some of the numerical results,obtained from an annular combustor configuration with the turbulent Schmidt number varying from 0.85 to 0.2,were presented and compared with a benchmark atmospheric test results.It was found that the Schmidt numberσt in mean mass fraction f transport equation had significant effect on dilution air mixing process.The mixing between dilution air and reaction products from the primary zone obviously improved asσt decreased on the combustor exit surface.Meanwhile,the sensitivity ofσt in three turbulence models including Realizable k-ε,SST(Shear Stress Transport)and RSM(Reynolds Stress Model)has been compared as well.Since the calculation method of eddy viscosity was different within these three models,RSM was proved to be less sensitive than another two models and can guarantee the best prediction of mixing process condition.On the other hand,the results of dilution air mixing were almost independent of Schmidt number Sct in progress variable c transport equation.This study suggested that for accurate prediction of combustor exit temperature distribution in steady state reacting flow simulation,the turbulent Schmidt number in steady state simulation should be modified to cater to dilution air mixing process.
基金This work was funded by the Netherlands Organisation for Scientific Research(NWO,project number 14927).
文摘Flamelet Generated Manifold(FGM)is an example of a chemistry tabulation or a flamelet method that is under attention because of its accuracy and speed in predicting combustion characteristics.However,the main problem in applying the model is a large amount of memory required.One way to solve this problem is to apply machine learning(ML)to replace the stored tabulated data.Four different machine learning methods,including two Artificial Neural Networks(ANNs),a Random Forest(RF),and a Gradient Boosted Trees(GBT),are trained,validated,and compared in terms of various performance measures.The progress variable source term and transport properties are replaced with the ML models.Particular attention was paid to the progress variable source term due to its high gradient and wide range of its value in the control variables space.Data preprocessing is shown to play an essential role in improving the performance of the models.Two ensemble models,namely RF and GBT,exhibit high training efficiency and acceptable accuracy.On the other hand,the ANN models have lower training errors and take longer to train.The four models are then combined with a one-dimensional combustion code to simulate a counterflow non-premixed diffusion flame in engine-relevant conditions.The predictions of the ML-FGM models are compared with detailed chemical simulations and the original FGM model for key combustion properties and representative species profiles.
基金The temperature measurements and PIA were supported by Prof.M.Nishioka of University of Tsukuba and Prof.K.Nishino of Yokohama National University,respectively.This work was partially supported by MEXT as"Program for Promoting Researches on the Supercomputer Fu-gaku"(Digital Twins of Real World’s Clean Energy Systems with Inte-grated Utilization of Super-simulation and AI).
文摘Owing to the increasing worldwide demand for natural gas,the development of a large submerged combustion vaporizer is required.Its burner is equipped with a water spray nozzle to reduce nitrogen oxides,and a practi-cal simulation method is required for the optimal design.The non-adiabatic flamelet approach can predict the combustion emissions and is useful for reducing simulation costs.However,as the number of control variables increases,the database requires larger memory and cannot be dealt with by general computers.In this study,an artificial neural network(ANN)model based on a five-dimensional flamelet database,which includes the effects of heat loss and vapor concentration by sprayed water evaporation,is developed.Furthermore,large eddy sim-ulations(LESs)for turbulent combustion fields with and without water spray are conducted employing flamelet generated manifold(FGM)approach with this ANN model,and the validity is investigated.For comparison,a lab-scale burner equipped with a water spray nozzle is manufactured,and combustion experiments with and without water spray are conducted.The results show that CO,NO,temperature,and reaction rate of progress variable predicted by the present ANN model are in good agreement with those of a five-dimensional flamelet database.In the condition without water spray,the flame behavior predicted by the LES employing the FGM/ANN ap-proach is in good agreement with that employing the conventional FGM approach,while indicating much lower memory,although there appeared some quantitative discrepancies in the temperature against the experiment probably partially because of the insufficiency of the FGM approach for the present complex flame structure.In the condition with water spray,the LES employing the FGM/ANN approach is able to capture the effect of the water spray on the flame behavior in the experiment,such that the water spray decreases the temperature,which causes the decrease in NO but increase in CO.
基金the National Natural Science Foundation of China(Grant No.11572330)the support from the Engineering Research Center of Gas Turbine and Civil Aero Engine,Ministry of Education of China+1 种基金the support from International Clean Energy Talent Program by China Scholarship Council(Grant No.201904100044)Open founding of National Key Laboratory of Science and Technology on Aero-Engine Aero-Thermodynamics of China(Grant No.6142702180307).
文摘In turbulent combustion simulations, the flow structure at the unresolved scale level needs to be reasonably modeled. Following the idea of turbulent flamelet equation for the non-premixed flame case, which was derived based on the filtered governing equations(L. Wang, Combust. Flame 175, 259(2017)), the scalar dissipation term for tabulation can be directly computed from the resolved flowing quantities, instead of solving species transport equations. Therefore, the challenging source term closure for the scalar dissipation or any assumed probability density functions can be avoided;meanwhile the chemical sources are closed by scaling relations. The general principles are discussed in the context of large eddy simulation with case validation. The new model predictions of the bluff-body flame show sufficiently improved results, compared with these from the classic progress-variable approach.
文摘The effects of heat loss on the structure of laminar flamelets,which are the constitutive elements of turbulent flames under the most common operating conditions,are investigated for typical aeronautical gas-turbine operating conditions at take-off.The magnitude of heat loss is quantified via the "enthalpy defect" measured with respect to an adiabatic flame.A procedure to generate laminar flamelets with an assigned enthalpy defect at the boundaries is devised and applied to nonpremixed propane/air flames,as propane reproduces the essential features of higher hydrocarbon combustion.It is found,contrary to commonly held beliefs,that the enthalpy defect has a significant effect on the concentration not only of minor species,but also of main reaction products.Such effects are found in general to be more pronounced for fuel-rich conditions.An impact is anticipated on the formation rate of nitric oxides.The effects of scalar dissipation rate are also discussed.
