CFD simulations of quenching process for partial oxidation of methane:Comparison of jet-in-cross- flow and impinging flow configurations☆

Xinyu Yu,Tianwen Chen,Qi Zhang,Tiefeng Wang*

Beijing Key Laboratory of Green Reaction Engineering and Technology,Department of Chemical Engineering,Tsinghua University,Beijing 100084,China

1.Introduction

Acetylene is used as the feedstock of many industrial processes because of its highly active chemical properties[1].The commonly used methods for acetylene production include the most widely applied non-catalytic partial oxidation(POX)process[1-6],calcium carbide process[7],arc discharge process[8]and pyrolysis process[9,10].The exploitation of shale gas has significantly decreased the price of natural gas and makes the natural gas-based acetylene industry more competitive[11].Thus,it could be expected that the POX process of methane would maintain the main method of acetylene production in the future.

The POX process is quite complex,because it involves fast mixing of feedstocks and millisecond reactions.The product gas mixture has a high temperature of 1800 K,and must be quenched within milliseconds when the acetylene yield reaches its maximum to prohibit further acetylene consumption reactions.The quenching technology is the key to the POX process and several quenching methods have been proposed,including water quenching[12,13],oil quenching[12,14,15]and indirect cooling[16]processes.The water quenching process,which is widely used in the industrial POX process,uses water as the quenching medium and the temperature of the gas mixture drops rapidly from 1800 K to 350 K.Although a high yield of acetylene is achieved,the terminal temperature of 350 K is too low to recover the vast heat.The oil quenching process is an improvement of the water quenching process.It uses aromatic oil with a high boiling point as the quenching medium and the terminal temperature is increased to 470-520 K[12],but results in a complicated composition of the product gas.In the indirect cooling process,the heat is removed with a heat exchanger and the terminal temperature is about 570 K.However,this quenching process has a time scale of 0.3-0.5 s,which is too long for terminating the acetylene consumption reactions of milliseconds.It is therefore suitable for POX to produce syngas but not for acetylene production.

In our previous works,a new quenching technology for the POX process has been proposed,in which the cold pyrolysis gas is used as the quenching medium[17,18].The pyrolysis gas quenching process is shown in Fig.1.The key to the pyrolysis gas quenching process is the rapid mixing of hot reacting gas and quenching stream.The temperature of the gas mixture is controlled by the mole flow ratio of the quenching stream to the mainstream.For flow ratio of2.4,the final temperature of the mixture is about 850 K.The gas mixture is subsequently cooled by a heat exchanger to recover the heat for preheating the feedstock or steam production.The cooled gas mixture is divided into two streams:one is used as the quenching stream and the other flows to the subsequent acetylene separation unit.This quenching process has a good heat recovery efficiency and little effect on the subsequent acetylene separation process.

The fast mixing between the hot product gas of 1800 K and cold pyrolysis gas of 400 K decreases the temperature of the gas mixture to 700-850 K in a few milliseconds.This quenching process is essentially similar to the transfer line exchanger(TLE)used in steam-cracking of hydrocarbons for ethylene production,in which a heat exchange efficiency of 60%-80%can be achieved.Assuming a heat exchange efficiency of 80%,the heat recovered in the quenching process can produce 8-10 tons of medium-pressure steam per ton of acetylene.This can greatly increase the competitiveness of the POX industry.

Despite the advantages of the gas quenching process,many works are needed to quantitatively analyze its feasibility and performance.For example,the appropriate temperature needs to be explored to ensure low enough acetylene consumption.Because the pyrolysis gas quenching process involves turbulent mixing and millisecond reactions,experimental studies are very difficult.In this case,the computational fluid dynamics(CFD)simulation is efficient to predict the fields of velocity,temperature and species concentration and quantitatively analyze the quenching efficiency.

The complicated millisecond reactions make the fast chemistry simplification not applicable for the gas quenching process,and a finite-rate chemistry model should be used.There are three finite-rate chemistry models available for simulations of turbulent reactions in FLUENT,i.e.the eddy dissipation model,the eddy-dissipation concept(EDC)model and the Probability Density Function(PDF)approach[19].In the eddy dissipation model,the reaction rates are controlled by the turbulence;therefore it is not suitable for the multi-step chemical mechanism situation.The EDC model,although more computationally efficient,is found to predict wrong results with default model parameters in our previous research on POX combustion and it lacks scientific basis to adjust those model parameters.The PDF approach is more accurate to describe a system with finite-rate chemistry because it closes the reaction terms without hypothesis.The Reynolds-averaged Navier-Stokes(RANS)model coupled with PDF model has been successfully used to simulate the non-premixed and premixed jet flames[20,21]and the POX process of methane in our previous work[22],thus it is used in the present work.

The present work aims to simulate the pyrolysis gas quenching process.Two important flow configurations,namely the jet-in-cross- flow(JICF)[23-27]and impinging flow[28-31]are systemically compared.Firstly,the mixedness,residence time distribution and temperature decreasing rate in a non-reactive quenching process are studied to compare the mixing performance.Then the reactive quenching process is simulated by using CFD coupled with detailed reaction mechanism.The loss of acetylene during the quenching is calculated to quantitatively describe the performance of pyrolysis gas quenching process.Furthermore,the consumption path of acetylene is analyzed,and the effect of temperature on acetylene consumption is studied to guide further optimization of the quenching process.

2.Computational Specifications

2.1.Model description

The steady-state 3D simulations are performed by ANSYS FLUENT.The dimensionless governing equations of continuity and momentum for steady-state flow in Cartesian coordinates are as follows:

Compared with more advanced models such as Reynolds Stress Model(RSM),Large Eddy Simulation(LES)and Direct Numerical Simulation(DNS),thek-ε turbulence model is more suitable for simulating an industrial-size apparatus because it has a relatively low computational cost.The realizablek-ε model has been successfully applied to simulate the industrial apparatus,such as the combustion reactor[32],burner orifice[33],bakery pilot oven[34],furnace[35]and POX reactor[22].For the study of JICF[36]and impinging flow[37,38],the realizablek-ε model also shows good predictions;therefore the realizablek-ε model is used in this work.The Reynolds stress term in Eq.(2)is expressed by the Boussinesq approximation as[39]:where μtis the turbulent viscosity,kis the turbulent kinetic energy,and δijis the Kronecker delta.

In the realizablek-ε model,conservation equations for the kinetic energykand the dissipation rate of turbulent kinetic energy ε are expressed as:

whereare respectively the volume fractions of A and B after the mixing is complete.When A and B are completely mixed,η=1;when A and B are completely separated,η=0.

In the literature,the intensity of segregation(IOS)[26,41-43]defined by Danckwerts[44]has been widely used to quantify the mixing quality.The definition of IOS is:

whereare the average volume fractions of species A and B,respectively,and σfAis the standard deviation offA.According to this definition,a smallerIsstands for a better mixing,andIs=0 indicates a complete mixing.

The intensity of segregation merely reflects the overall mixing efficiency.In comparison,the local mixedness can describe the spatial distribution of the mixing performance.The overall mixing is described by the cross-sectional averaged mixednesscalculated as:

whereAis the cross-section area andaiand ηiare respectively the crosssection area and local mixedness of celli.Using the FLUENT software,the local mixedness for each cell is calculated with the User Defined Functions(UDFs),and the cross-sectional averaged mixedness is calculated by the area-weighted average command.

The performance of the mixing apparatuses for quenching in the POX process is further studied by CFD simulations coupled with detailed reaction mechanism.The composition PDF transport model is solved by the Lagrangian particle-based Monte Carlo method.The EMST(Euclidean Minimum Spanning Tree)model[45]with default valueCφ=2.0 is used as the particle mixing model.The detailed reaction mechanism used in this research is Leeds 1.5[46],which includes 37 species and 175 reactions.Our previous work shows that this mechanism has a good accuracy for the methane POX process[1].

2.2.Flow geometry

Two kinds of flow geometry are investigated,i.e.the JICF configuration and the impinging flow configuration.For the JICF configuration,the length and diameter of the main pipe are respectively set as 0.8m and 0.4 m according to the POX industrial reactor.Four jet pipes are introduced 0.2 m away from the inlet of the mainstream,and they are evenly distributed around the main pipe,as shown in Fig.2.

The diameter of the jet pipe is 0.094 m.For the impinging flow configuration,the diameter of the main pipe is also 0.4 m.The total length of the main pipe is 0.5 m,i.e.the distance between the two opposing inlets of hot mainstream and cold quenching stream is 0.5 m.The outlet of the impinging flow apparatus is the lateral surface of a cylinder with a diameter of 0.8 m and a height of 0.1 m at the middle axial position of the main pipe,as shown in Fig.2.

2.3.Boundary conditions

2.3.1.Boundary conditions for non-reactive mixing simulations

In the non-reactive mixing simulations,the temperature and the compositions of the mainstream and quenching stream are the same as those in the reactive mixing,as shown in Table 1.The fixed mole flow ratio(quenching stream to mainstream)of 2.4 is used.The mainstream velocity is set as 100 m·s-1according to the gas velocity in the industrial POX process.Using the data shown in Table 2,the velocity of the quenching stream is calculated to be 240 m·s-1for the JICF and 53.3 m·s-1for the impinging flow.The Reynolds numbers of the mainstream and quenching stream are both higher than 105,therefore the realizablek-εturbulence model is suitable.An initial turbulent intensity of 10%is used for the inlets of the mainstream and quenching stream.The detailed settings of the boundary conditions are listed in Table 2.

2.3.2.Boundary conditions for reactive mixing simulations

The boundary conditions for the reactive mixing simulations are the same as those for non-reactive simulations.For the gas quenching process,the hot product gas should be quenched when the concentration of acetylene reaches its maximum.Since the cold quenching stream is introduced 0.2 m away from the mainstream inlet,the temperature and composition at the mainstream inlet are set as the values obtained 0.2 m upstream of the position with the maximum acetylene concentration.These data are taken from the simulation results of the methane POX process in our previous work,in which a good agreement is obtained between the CFD simulations and the industrial data[22].The quenching stream is assumed to have the same composition of major species as the product gas for convenient analysis of acetylene loss during the quenching process.However,the amount of radicals in the quenching stream is negligible because of the low temperature(300-400 K).The temperature and composition profiles of the mainstream inlet and quenching stream inlet are listed in Table 1.

2.4.Numerical settings

The body- fitted grid generated by ICEM is used.Four mesh sizes,i.e.4 mm,6 mm,8 mm and 10 mm,are tested to verify the mesh independence for a case of JICF with a mainstream velocity of 100 m·s-1and quenching stream velocity of 240 m·s-1.The mesh independence test shows that the 6 mm mesh is fine enough to give accurate profiles of temperature and concentration;therefore it is used for the subsequent simulations.

In both the non-reactive and reactive mixing simulations,the steady-state segregated solver with implicit formulation is used.Considering that the realizablek-ε model is primarily valid for the flow in a region far from walls,the standard wall functions are used in the near-wall region.A second-order upwind discretization scheme is used for all governing equations,and the PRESTO!Algorithm is used for pressure discretization.The SIMPLE algorithm is used for the pressure-velocity coupling with the default values for the relaxation factors.

In the simulations of non-reactive mixing,the Species Transport Model is used to simulate the mixing of the mainstream and quenching stream.Our test has shown that the non-reactive mixings of the mainstream and quenching stream predicted by the Species Transport Model and PDF Model with default model parameters are almost the same.Thus the Species Transport Model is used in the non-reactive mixing because it is much more computationally efficient.The Discrete Phase Model(DPM)is used to track the massless particles injected into the mainstream to analyze the movement of fluid elements at different positions.For each simulation,about 1000 massless particles are injected and the residence time distribution of the mainstream is obtained by counting the residence time of the particles.The number of iterations for the continuous phase is set to 5 per DPM iteration and the maximum number of tracking steps is set to 2000 to ensure all the particles flow out of the computational domain.

In the simulations of reactive mixing,thein situadaptive tabulation(ISAT)algorithm developed by Pope[47]is used to accelerate the calculations of chemistry source term.The ISAT error tolerance εtolis important for computational accuracy[48].Four different ISAT error tolerances,i.e.10-3,10-4,10-5and 10-6,are tested.The simulated axial profiles of the acetylene concentration in the JICF are shown in Fig.3(a).The simulation results become independent of εtolwhen εtol≤ 10-5.Therefore εtol=10-5is used for the subsequent simulations of reactive mixing.In the steady-state composition PDF transport simulations,the statistical errors are also affected by the number of particles per cell(Np)and the number of iterations in pseudo-time average(Ni).The effect ofNpis checked by setting it to 20,40 and 60 withεtol=10-5.The simulated axial profiles of acetylene concentration in the JICF configuration are shown in Fig.3(b).The increase ofNpslightly reduces the fluctuation of the acetylene concentration;thereforeNp=60 is used for the following simulations.The valueNiis set according to Fluent User's Guide[40],in whichNiis initially set to the default value of 50,and then increases by 0.2 per iteration after the steady-state solution is obtained.

Fig.2.Schematic of the JICF and impinging flow configurations.

3.Results and Discussion

3.1.Model validations

Before the simulations of the quenching process,the model has been validated for both the non-reacting mixing and the POX process of methane.Since the accurate prediction of the flow field is critical for the accurate prediction of quenching process,the RANS equations with realizablek-ε model is firstly validated by experimental data obtained in a labscale JICF mixing apparatus.The JICF mixing apparatus has a main pipe of 0.1 m in diameter,and four perpendicular jets of 0.05 m in diameter.Air is used as the working fluid and methane of 500×10-6is added to the mainstream for the mixing measurement.The concentration of methane is measured by gas chromatography with flame ionization detector.Fig.4 compares the measured and simulated methane concentrations at different axial and radial positions.The good agreement between the simulated and experimental results con firms that the model used in the present work is reliable to predict the mixing.

Table 1 Temperature and concentration of major species for the reactive mixing simulations

Because the experimental data of pyrolysis gas quenching process is unavailable,the reliability of CFD model coupled with detailed reaction mechanism is validated by the industrial data of the BASF POX process of methane.The computational domain and boundary conditions used in our research are the same as those in the literature[22].The computational domain is a three-nozzle structure,which is a simplification of the industrial POX reactor.The main nozzle has a diameter of 25 mm and the pilot oxygen nozzle has a diameter of 3 mm,both of which are consistent with the industrial POX reactor.In the simulations,the mixture of methane and oxygen preheated to 873 K are jetted into the reaction chamber through the three nozzles with a velocity of 100 m·s-1.The mole ratio of oxygen to methane is fixed at 0.55,which is used in the industrial POX process.The auxiliary oxygen withambient temperature is jetted into the reaction chamber through the pilot oxygen nozzle with a velocity of 27 m·s-1.The data shown in Fig.5 is based on the area-weighted averages of the species concentration at the exit of the industrial reactor,which is 400 mm away from the burner block.The Leeds 1.5 mechanism and other commonly used mechanisms are compared.The Leeds 1.5 mechanism gives the best overall predictions of the major species,especially the most concerned acetylene concentration;therefore it is used in the following studies.

Table 2 Boundary conditions for non-reactive mixing simulations

Fig.3.Effect of(a)ISAT error tolerance and(b)number of particles per cell on axial profiles of acetylene mass fraction.

Fig.4.Comparison of measured and simulated profiles of the methane mole fraction.

Fig.5.Comparison of simulated and industrial data of the BASF POX process.

3.2.Non-reactive mixing

3.2.1.Distribution of mixedness

Fig.6 shows the pseudo-color image of the distribution of the local mixedness in the central plane.In the JICF configuration,the quenching jets initially mix with the mainstream at the windward side.However,the quenching jets penetrate so deep that they mix with each other and cause a low local mixedness region in the center.The mixing region at the leeward side gradually expands to the nearwall region and the center of the main pipe.The local mixedness in the near-wall region is also low;indicating the mixing in this region is poor.The reason is that the over-penetration causes a high pressure region in the main pipe center.As a result,some of the mainstream flows around the jets,resulting in a poor mixing effect in the near-wall region.In other words,the over-cooled region in the center leads to an under-cooled region near the wall.The overpenetration could be avoided by decreasing the jets velocity.However,with a fixed JICF configuration,the decreasing of jets velocity leads to a higher terminal temperature.

Compared with the JICF configuration,the impinging flow configuration does not have the problems of over-cooling and under-cooling.As shown in Fig.6,the quenching stream and the mainstream mix with each other more uniformly.The cross-sectional averaged mixedness η at the outlet is 0.984 for the impinging flow and 0.914 for JICF,indicating that the impinging flow configuration has a better mixing.However,it is not strict to say that the impinging flow has a better performance than the JICF solely based on the local mixedness,because it only compares the uniformity of the mixing.Other important results,such as the residence time distribution,temperature decreasing rate and acetylene loss,are further studied in the following sections to comprehensively compare the JICF and impinging flow configurations.

As shown in Fig.7(a),in the impinging flow configuration,the opposing jets impinge towards the mainstream because the inlet momentum of the mainstream is lower than that of the quenching stream.Correspondingly,the turbulent kinetic energy on the mainstream side is higher than that on the quenching stream side.As shown in Fig.7(b),the high turbulent kinetic energy zone is on the mainstream side.Due to the strong eddy mixing and high shear stress rate,the mixing in this region is faster.As a result,the mixing quality is higher on the mainstream side than on the quenching stream side in the outlet channel,as shown in Fig.6(b).

3.2.2.Residence time distribution of the mainstream

Because the acetylene consumption reactions strongly depend on the temperature,the ideal quenching should decrease the mainstream temperature very quickly and uniformly.Accordingly,a narrow residence time distribution(RTD)is needed.The RTD of the mainstream is obtained by analyzing the residence time of the massless tracing particles injected at the mainstream inlet.The dimensionless residence time is defined as θ=twhereis the average residence time of all tracing particles.The RTD in terms of θ for both the JICF and impinging flow configurations are shown in Fig.8.It can be seen that the RTD of the impinging flow is narrow and has a single peak,while that of the JICF is much wider and has multiple peaks.Normally,a better uniformity of the mainstream RTD ensures a uniform temperature decrease of the mainstream.In this sense,the impinging flow configuration is better than the JICF configuration.

Fig.6.Profiles of local mixedness η in(a)JICF and(b)impinging flow configurations.

Fig.7.(a)Velocity distribution and(b)Turbulent kinetic energy distribution for the impinging flow configuration(M is mainstream and Q is quenching stream).

3.2.3.Temperature decreasing rate of the mainstream

The temperature decreasing rate of the mainstream is very important for the quenching process because a quick temperature decrease can effectively reduce the acetylene consumption.The temperature variation of the mainstream during the quenching process has also been analyzed from the massless tracing particles.The results of the JICF and impinging flow configurations are compared in Fig.9.Fig.9(a)-(e)and(h)-(i)respectively shows the temperature variations of the mainstream injected at lines ofY=0 andY=X,with an interval of 0.04 m.

Fig.8.Dimensionless residence time distribution of the mainstream.

In general,the JICF has a faster temperature decrease than the impinging flow.However,the JICF configuration has two disadvantages.Firstly,the over-cooled region in the center of the main pipe can occur by over-penetration of the jets.The over-cooled mainstream firstly drops to a low temperature and then gradually increases to the terminal temperature,as shown in Fig.9(a)-(c).Meanwhile,some of the mainstream in the JICF configuration is not well cooled to the terminal temperature,as shown in Fig.9(d)and(e).Secondly,the temperature at the outlet in the JICF configuration has a larger variance than that in the impinging configuration.The stream with a higher temperature has a higher loss of acetylene;therefore a larger variance of temperature at the outlet will lead to a higher loss of acetylene.

3.3.Reactive mixing

3.3.1.Profiles of temperature and acetylene concentration

As discussed above,the impinging flow configuration has a more uniform mixing but has a smaller temperature decreasing rate than the JICF configuration.The combined effect of these two factors on the yield of acetylene is further studied in the reactive quenching flows.

Fig.10 shows the temperature variations of the mainstream injected at 9 different positions.It can be seen that the temperature decreasing rate is still generally larger in the JICF than in the impinging flow.The temperature decreasing rates are very similar for the non-reactive mixing and reactive mixing,indicating that the fast chemical reactions have no significant effect on mixing.

Fig.9.Temperature decreasing rates of the mainstream in non-reactive mixing.

Fig.10.Temperature decreasing rates of the mainstream in reactive mixing.

Fig.11 shows the variation of acetylene mass fraction.It can be seen that the variations can be divided into three stages,i.e.the fast increasing stage,the slow increasing stage and the decreasing stage.The fast increasing stage corresponds to the first platform of temperature in Fig.10,in which the hot mainstream maintains pyrolysis reactions and therefore acetylene keeps generating.Since the reaction time of the pyrolysis reactions in the JICF is longer than that in the impinging flow,the pyrolysis reactions go more thoroughly;thus the acetylene concentration at the end of the fast increasing stage in the JICF is higher than that in the impinging flow.The slow increasing stage is formed due to the mixing between the mainstream and the quenching stream,corresponding to the fast decrease of temperature in Fig.10.Since the quenching stream has a slightly higher concentration of acetylene than the mainstream,the mixing leads to an increase in the acetylene concentration of the main stream.The decreasing stage of acetylene follows closely to the end of quenching.More specifically,in those overcooled cases such as Fig.11(a)-(c)and(f)for the JICF,the acetylene concentration begins to decrease at the lowest temperature.While in those cases without over-cooling,the acetylene concentration begins to decrease when the temperature reaches another platform.This indicates that acetylene consumption reactions still occur,although in a slow rate,when the mixing is complete.Therefore,the gas mixture should be further quickly cooled in a heat exchanger to completely stop the acetylene consumption and recover the heat.

3.3.2.Acetylene loss in the quenching process

The amount of acetylene loss is critical for evaluation of the quenching process.The acetylene loss in the quenching process is closely related to the temperature profile.The axial and radial temperature profiles in both the JICF and impinging flow configurations are compared,as shown in Fig.12.In the JICF,the temperature profile becomes more uniform with increasing axial position.In the impinging flow,the mixing mainly occurs at the impinging interface of the mainstream and the quenching stream,and the temperature becomes more uniform in the outlet pipe.Complete mixing is achieved atz=0.6 m in the JICF andr=0.25 m in the impinging flow.

The acetylene loss is compared for the JICF and impinging flows at positions where the cross-sectional averaged mixedness=0.9.The plane corresponding to=0.9 isz=0.59 m in the JICF configuration andr=0.24 m in the impinging flow configuration.The acetylene loss ratio is calculated from the flow rates of acetylene at the mainstream inlet,quenching stream inlet and plane chosen for analysis.The calculation results are shown in Table 3.It can be seen that the acetylene loss ratio is much lower in the impinging flow than in the JICF,suggesting that the impinging configuration is better for the pyrolysis gas quenching process.

3.3.3.Acetylene loss in the heat exchange process

As the mixing between the mainstream and quenching stream is completed in the quenching process,the gas mixture in the subsequent heat exchanger has uniform composition.Therefore,the reactions in the heat exchanger can be simulated with CHEMKIN to reduce the computation cost.According to the steam cracking process for ethylene production,the quenching boiler can cool the hot gas from 1150 K to 700 K in 0.05-0.1 s.Therefore the cooling of the gas mixture containing acetylene from 850 K to 400 K by a heat exchanger in less than 0.1 s is feasible.As the gasmixture is uniform,the closed homogenous batch reactor(0-dimensional)is used for the simulation.The pressure in the reactor is held at 101.325 kPa.The initial temperature and composition are the same as those of the outlet of the quenching apparatus.The solver maximum step time is set as 2×10-5s.By adjusting the heat loss rate of the gas mixture,the quenching from 850 K to 400 K in 0.1 s and 0.05 s are realized.

Fig.11.Profiles of mass fraction of acetylene in JICF and impinging flow configurations.

Fig.12.Pseudo-color image of temperature in JICF and impinging flow configurations.

Table 3 Acetylene loss ratio in the quenching process

The variation of the acetylene concentration in the heat exchanger is shown as a function of temperature during the cooling process in Fig.13 with cooling time of 0.05 s and 0.1 s.The initial conditions of these simulations are the same as the area-weighted averages at η=0.9 in the JICF configuration.It can be seen that the acetylene consumption reactions mainly occur above 800 K.When the temperature drops below 650 K,the rate of acetylene consumption is negligible.By shortening the cooling time,the loss of acetylene is significantly decreased.The acetylene loss ratio in the heat exchanger is 1.31%for a cooling time of 0.1 s,and decreases to 1.04%for 0.05 s.

3.3.4.Consumption path of acetylene

Fig.13.Acetylene concentration profiles at different temperatures in the heat exchange process.

The consumption path of acetylene is studied to guide further optimization of the quenching process.The sum of the mass fractions of C2H2,CH4,H2,C2H4,C4H2,H2O,CO,CO2and O2at the outlet in the impinging flow is higher than 99.5%,therefore only these major species are considered.Fig.14 shows their concentration profiles of the mainstream injected at(X=0.08,Y=0)in the impinging flow configuration.It can be seen that the decreasing stage of acetylene concentration corresponds to an increase in the ethylene concentration and a slight decrease in the hydrogen concentration.Note that the slight decrease in hydrogen mass fraction means a significant consumption of hydrogen because of its light molecule weight.The concentrations of other major species remain stable in the acetylene consumption stage,suggesting that they are irrelevant to the acetylene consumption reactions.At 4 ms,which is obviously in the decreasing stage of acetylene,the mass fractions of C2H2,H2and C2H4are respectively 10.800%,6.105%and 0.489%.At the outlet(5.24 ms),the mass fractions ofC2H2,H2and C2H4are respectively 10.669%,6.096%and 0.627%.Hence the changes of mass fractions of C2H2,H2and C2H4are respectively-0.131%,-0.009%and 0.138%,which are consistent with the stoichiometry of C2H2+H2=C2H4.These results show that acetylene hydrogenation reaction is the dominating reaction in the decreasing stage of acetylene concentration.Acetylene is also consumed by its hydrogenation to ethylene in the temperature decreasing stage,the superficial acetylene concentration increases because the mainstream mixes with the quenching stream with a higher acetylene concentration.

4.Conclusions

The mixing and quenching performance of the JICF configuration with four jet inlets and the impinging flow configuration have been studied by CFD simulations and detailed reaction mechanism.The distribution of mixedness,residence time distribution,temperature decreasing rate,and loss ratio of acetylene are compared.The results and analysis lead to the following conclusions:

(1)The impinging flow has a better mixing than the JICF.The impinging flow does not have the over-penetration or under penetration problem of JICF.The more uniform mixing in the impinging flow realizes smaller temperature variance at the outlet.In addition,the residence time is more uniform in the impinging flow than in the JICF.

(2)The quenching simulations by the RANS-PDF model coupled with Leeds 1.5 mechanism show that the acetylene loss ratio is 2.89%in the JICF and 1.45%in the impinging flow,indicating that the impinging configuration is better for the pyrolysis quenching process.

(3)Acetylene is mainly consumed by its hydrogenation to ethylene in the quenching process.Acetylene consumption reactions mainly occur above 800 K,and become negligible below 650 K.

Fig.14.Concentration profiles of major species in impinging flow configurations.

(4)The loss ratio of acetylene in the subsequent heat exchange process is 1.31%for a cooling time of 0.1 s and decreases to 1.04%for a cooling time of 0.05 s.The overall acetylene loss in the whole quenching and heat recovery process is below 2.5%if the impinging flow configuration is used.

[1]Y.Liu,T.Wang,Q.Li,D.Wang,A study of acetylene production by methane flaming in a partial oxidation reactor,Chin.J.Chem.Eng.19(3)(2011)424-433.

[2]G.H.Bixler,C.W.Coberly,Wulff process acetylene,Ind.Eng.Chem.45(12)(1953)2596-2606.

[3]J.Elliott,H.H.McClure,Production of Acetylene by the Partial Oxidation of Hydrocarbons,U.S.Pat.,2945074(1958).

[4]C.H.O.Berg,Process and Apparatus for Acetylene Production,U.S.Pat.,2679540(1950).

[5]M.Russ,D.Grossschmidt,P.Renze,M.Vicari,H.Neuhauser,K.R.Ehrhardt,C.Weichert,Process and Apparatus for Preparing Acetylene and Synthesis Gas,U.S.Pat.,8506924(2013).

[6]Z.Wang,D.Zheng,Energy analysis and retro fitting of natural gas-based acetylene process,Chin.J.Chem.Eng.16(5)(2008)812-818.

[7]G.K.Earl,Calcium Carbide/Water Acetylene Gas Generator,U.S.Pat.,4444159(1982).

[8]S.Yao,E.Suzuki,N.Meng,A.Nakayama,A high-efficiency reactor for the pulsed plasma conversion of methane,Plasma Chem.Plasma Process.22(2)(2001)225-237.

[9]M.Holmen,O.A.Rokstad,A.Solbakken,High-temperature pyrolysis of hydrocarbons:1.Methane to acetylene,Ind.Eng.Chem.Process.Des.Dev.15(3)(1976)439-444.

[10]J.Happel,L.Kramer,Acetylene and hydrogen from the pyrolysis of methane,Ind.Eng.Chem.59(1)(1967)39-50.

[11]S.Paltsev,H.D.Jacoby,J.M.Reilly,Q.J.Ejaz,J.Morris,F.O'Sullivan,S.Rausch,N.Winchester,O.Kragha,The future of U.S.natural gas production,use,and trade,Energy Policy39(9)(2011)5309-5321.

[12]M.Bachtler,R.R.Schnur,P.Passler,O.Scheidsteger,W.Kastenhuber,G.Schlindwein,R.Konig,Process for the Production of Acetylene and Synthesis Gas,U.S.Pat.,5824834(1995).

[13]L.Erwin,A.Werner,S.F.Wilhelm,Production of Acetylene by Partial Oxidation of Hydrocarbons and Apparatus Therefor,U.S.Pat.,2966533(1960).

[14]D.Willi,F.Otto,K.Erich,M.Ferdinand,T.Walter,Production of Acetylene,U.S.Pat.,3242225(1966).

[15]J.G.Russell,Oil Quench Process for Partial Oxidation of Hydrocarbon Gases,U.S.Pat.,3022148(1958).

[16]D.Stapf,P.Passler,M.Bachtler,O.Scheidsteger,B.Bartenbach,Preparation of Acetylene and Synthesis Gas,U.S.Pat.,6365792(2002).

[17]T.Wang,Q.Li,D.Wang,Method and Device for Quenching and Heat Recovery in Partial Oxidation Process of Natural Gas,CN Pat.,CN102329189A(2012).(in Chinese).

[18]Q.Li,Non-catalytic Partial Oxidation of Methane to Produce Acetylene and Synthesis Gas(Ph.D.Thesis)Tsinghua Univ.,China,2013(in Chinese).

[19]S.B.Pope,PDF methods for turbulent reactive flows,Prog.Energy Combust.Sci.11(2)(1985)119-192.

[20]R.F.Cao,S.B.Pope,The influence of chemical mechanisms on PDF calculations of nonpremixed piloted jet flames,Combust.Flame143(4)(2005)450-470.

[21]D.H.Rowinski,S.B.Pope,PDF calculations of piloted premixed jet flames,Combust.Theor.Model.15(2)(2011)245-266.

[22]Q.Zhang,Y.Liu,T.Chen,X.Yu,J.Wang,T.Wang,Simulations of the industrial reactor of methane partial oxidation by CFD coupled with detailed chemistry at industrial operating conditions,Chem.Eng.Sci.142(2016)126-136.

[23]M.T.Kandakure,V.C.Patkar,A.W.Patwardhan,J.A.Patwardhan,Mixing with jets in cross- flow,Ind.Eng.Chem.Res.48(14)(2009)6820-6829.

[24]F.Bazdidi-Tehrani,A.Shahmir,A.Haghparast-Kashani,Numerical analysis of a single row of coolant jets injected into a heated cross flow,J.Comput.Appl.Math.168(1)(2004)53-63.

[25]P.Arora,A.K.Saha,Three-dimensional numericalstudy of flow and species transport in an elevated jet in cross flow,Int.J.Heat Mass Transf.54(1)(2011)92-105.

[26]H.Urson,M.F.Lightstone,M.J.Thomson,A numerical study of jets in a reacting cross flow,Numer.Heat Transfer Part A40(26)(2001)689-714.

[27]F.C.C.Galezzo,G.Donnert,P.Habisreuther,N.Zarzalis,R.J.Valdes,W.Krebs,Measurement and simulations of turbulent mixing in a jet in cross flow,J.Eng.Gas Turbines Power133(6)(2011),061504-1.

[28]M.Sohrabi,B.Zareikar,Modeling of the residence time distribution and application of the continuous two impinging streams reactor in liquid-liquid reactions,Chem.Eng.Technol.28(1)(2005)61-66.

[29]M.Sohrabi,A.M.Jamshidi,Studies on the behaviour and application of the continuous two impinging streams reactors in gas-liquid reactions,J.Chem.Technol.Biotechnol.69(4)(1997)415-420.

[30]B.K.Johnson,R.K.Prud'Homme,Chemical processing and micromixing in confined impinging jets,AIChE J49(9)(2003)2264-2282.

[31]F.Schwertfirm,J.Gradl,H.C.Schwarzer,W.Peukert,M.Manhart,The low Reynolds number turbulent flow and mixing in a confined impinging jet reactor,Int.J.Heat Fluid Flow28(6)(2007)1429-1442.

[32]B.Imteyaz,M.A.Habib,M.A.Nemitallah,A.Jamal,Investigation of liquid ethanol evaporation and combustion in air and oxygen environments inside a 25 kW vertical reactor,Proc.Inst.Mech.Eng.Part A229(6)(2015)647-661.

[33]L.C.B.S.Reis,J.A.Carvalho,M.A.R.Nascimento,L.O.Rodrigues,F.L.G.Dias,P.M.Sobrinho,Numerical modeling of flow through an industrial burner orifice,Appl.Therm.Eng.67(2014)201-213.

[34]M.Boulet,B.Marcos,M.Dostie,C.Moresoli,CFD modeling of heat transfer and flow field in a bakery pilot oven,J.Food Eng.97(3)(2010)393-402.

[35]G.Besagni,R.Mereu,F.Inzoli,CFD study of ejector flow behavior in a blast furnace gas galvanizing plant,J.Therm.Sci.24(1)(2015)58-66.

[36]Z.W.Li,W.X.Huai,Z.D.Qian,Study on the flow field and concentration characteristics of the multiple tandem jets in cross flow,Sci.China Technol.Sci.55(10)(2012)2778-2788.

[37]E.V.Kartaev,V.A.Emel'Kin,M.G.Ktalkherman,V.I.Kuz'Min,S.M.Aul'Chenko,S.P.Vashenko,Analysis of mixing of impinging radial jets with cross flow in the regime of counter flow jet formation,Chem.Eng.Sci.119(6)(2014)1-9.

[38]E.V.Kartaev,V.A.Emelkin,M.G.Ktalkherman,S.M.Aulchenko,S.P.Vashenko,V.I.Kuzmin,Formation of counter flow jet resulting from impingement of multiple jets radially injected in a cross flow,Exp.Thermal Fluid Sci.68(2015)310-321.

[39]D.C.Wilcox,Turbulence Modeling for CFD,DCW Industries Inc.,La Canada,California,1993.

[40]Fluent,Fluent 6.3 User's Guide,Fluent Inc.,Lebanon,New Hampshire,2006.

[41]D.Liscinsky,B.True,J.D.Holdeman,Experimental Investigation of Cross flow Jet Mixing in a Rectangular Duct,National Aeronautics and Space Administration,1993 93-2037.

[42]J.T.Kroll,W.A.Sowa,G.S.Samuelsen,Optimization of orifice geometry for cross flow mixing in a cylindrical duct,J.Propuls.Power16(6)(2000)929-938.

[43]R.P.Fuller,P.K.Wu,A.S.Nejad,J.Schetz,Comparison of physical and aerodynamic ramps as fuel injectors in supersonic flow,J.Propuls.Power14(2)(1958)135-145.

[44]P.V.Danckwerts,The definition and measurement of some characteristics of mixtures,Appl.Sci.Res.3(4)(1952)279-296.

[45]S.Subramanian,S.B.Pope,A mixing model for turbulent reactive flows based on Euclidean minimum spanning trees,Combust.Flame115(4)(1998)487-514.

[46]K.J.Hughes,T.Turányi,A.R.Clague,M.J.Pilling,Development and testing of a comprehensive chemical mechanism for the oxidation of methane,In.J.Chem.Kinet.33(9)(2001)513-538.

[47]S.B.Pope,Computationally efficient implementation of combustion chemistry using in situ adaptive tabulation,Combust.Theor.Model.1(1)(1997)41-63.

[48]A.R.Masri,R.Cao,S.B.Pope,G.M.Goldin,PDF calculations of turbulent lifted flames of H2/N2 fuel issuing into a vitiated co- flow,Combust.Theor.Model.8(1)(2004)1-22.