M.G.Ktalkherman *,I.G.Namyatov

1 Khristianovich Institute of Theoretical and Applied Mechanics,Russian Academy of Science,Siberian Branch,Novosibirsk 630090,Russia

2 Voevodsky Institute of Chemical Kinetics and Combustion,Russian Academy of Science,Siberian Branch,Novosibirsk 630090,Russia

Keywords:Lique fied petroleum gas Pyrolysis Ole fins Fast-mixing reactor

ABSTRACT Currently,thermal decomposition ofhydrocarbons for the production ofbasic petrochemicals(ethylene,propylene)is carried out in steam-cracking processes.Aside from the conventional method,under consideration are alternative ways purposed for process intensification.In the context of these activities,the method of hightemperature pyrolysis of hydrocarbons in a heat-carrier flow is studied,which differs from previous ones and is based on the ability of an ultra-short time of feedstock/heat-carrier mixing.This enables to study the pyrolysis process at high temperature(up to 1500 K)at the reactor inlet.A set of model experiments is conducted on the lab scale facility.Lique fied petroleum gas(LPG)and naphtha are used as a feedstock.The detailed data are obtained on temperature and product distributions within a wide range of the residence time.A theoretical model based on the detailed kinetics ofthe process is developed,too.The effectofgoverning parameters on the pyrolysis process is analyzed by the results of the simulation and experiments.In particular,the optimal temperature is detected which corresponds to the maximum ethylene yield.Product yields in our experiments are compared with the similar ones in the conventionalpyrolysis method.In both cases(LPGand naphtha),ethylene selectivity in the fast-mixing reactor is substantially higher than in current technology.

1.Introduction

Thermal decomposition of hydrocarbons for the production of basic petrochemicals(ethylene,propylene)is carried out in steam-cracking process.Liquid(naphtha,gas oil)and gaseous(ethane,propane,butane)hydrocarbons are the feedstock for pyrolysis.The feedstock is heated in the convection section of the furnace.It mixes with the superheated steam and then enters a fired tubular reactor(radiant coil),where the endothermic decomposition of feedstock occurs within the controlled residence time,resulting in the formation of light ole fins and co-products.In order to prevent the loss of most valuable products during secondary reactions,the residence time in the reactor is limited.Normally,it lies within the range of 0.1–0.5 s.Then the mixture enters the transfer-line exchanger from which the cooled mixture passes through compression,separation and gas separation units.

The efficiency of pyrolysis process mainly depends on energy and feedstock consumptions.The typical values of specific energy consumptions and typical product yields can be found in literatures[1,2]and[3–5],respectively.In spite of some difference in the general technological pattern of the process,the performance indicators of principal pyrolysis unit(ethylene and propylene yields)are different slightly[3–5].Actually,the capability of process intensification in the framework of the conventional method of pyrolysis has been almost getting to the end.Accordingly,analysis of alternative methods is of interest.Such investigations have been performed for a long time and various alternative methods are considered[6].Catalytic pyrolysis and pyrolysis of feedstock in a high-temperature heat carrier flow seem to be the most promising.The first one was recently implemented in a large-scale reactor[7]and realized in a demonstration plant[8].The ethylene yield in both cases was remained almost the same as in furnace pyrolysis and the propylene yield was increased significantly.

The other way to intensify the pyrolysis process with the main purpose to increase the ethylene yield is related with the increased temperature in the reaction zone,which is impossible in the conventional method because of restricted heat resistance of coil tubes.This limitation can be overcome if the heat is supplied not through the tube wall,but is directly transferred via mixing with the highenthalpy carrier gas with the heat storage sufficient to realize the pyrolysis at high temperatures.All experiments in the heat-carrier flow show that ethylene yields higher than in the traditional method.The vast reference list on hydrocarbons pyrolysis in the heat carrier flow can be found in literatures[9,10].

The key problem in this method is the process of feedstock mixing with the heat carrier.Taking into account that the feedstock residence time in the reactor must be much shorter than in the traditional method,the mixing time must be minimal in order to reduce the duration of feedstock residence in the high-temperature area of the mixing region because it results in uncontrolled reactions prior to the reactor inlet.In the experiments indicated[9,10],various patterns of feedstock and heat carrier mixing are described.Those of them involving the principle of spatial separation of mixing and reaction areas are of most interest.For example,the mixing partially[11]or completely[12,13]occurred in a supersonic flow,the temperature of which is below the reaction starttemperature.Atthe end ofmixing ofthe feedstock and heatcarrier,the flow in shock waves transforms to subsonic one.Flow deceleration results in the rapid increase of the flow temperature up to the assigned value and then the pyrolysis begins.

A method of feedstock pyrolysis in a high-temperature heat carrier flow has been proposed[14].The method differs from the above mentioned ones and is based on the possibility of ultra-short time of mixing of the feedstock and heat carrier.A set of experiments in a gas-dynamic facility[15]enables to determine the optimum geometry of the mixer operating in the radial jets collision with a cross flow mode.The ultrashorttime ofthe feedstock/heatcarriermixing permits to study the process ofhydrocarbon pyrolysis atthe reactor inlettemperature above the maximum values achievable in the conventional method.In turn,as the process temperature rises,the ethylene concentration in pyrolysis products increases,too.

The block flow diagram of the considered process is shown in Fig.1.The main parts of the pyrolysis unit are combustion chamber,mixer,reactor and transfer line exchanger.The heat carrier is generated in the combustion chamber into which the fuel,oxygen(in stoichiometric proportion)and overheated steamare supplied.The temperature at the outlet from the combustion chamber is controlled by the steam flow rate.Gas jets ofthe pyrolyzed feedstock are injected into the heatcarrier flow normally to the reactor axis.The feedstock can be a preliminary evaporated liquid or a gas,and if it is necessary to improve the mixing quality,itis injected together with the steam.The geometry ofthe jetinjection unit must provide fast and qualitative mixing of the feedstock and heat carrier,and the geometry of the reactor channel should correspond to the optimum mixture residence time,at which the maximum yield of the valuable products is reached.Similar to the conventional method of furnace pyrolysis,the pyro–gas–steam mixture is cooled quickly in order to avoid the loss of the principal products.The method of residual heat utilization is the same as in the conventional method of pyrolysis.The gas separation scheme remains the same,too.

2.Mixer for the Pyrolysis Reactor

Special attention is paid on the problem of mixing.This problem is a key one for the proposed pyrolysis method.A series of tests were performed preliminary in a model gas-dynamic facility.Under study was the mixing process of the radial-jet system with cross flow in a cylindrical channel.The geometrical and gas-dynamic parameters were varied within wide ranges.

Schematic of the device and experimental results can be found in literature[15].Both flows,the main and injected ones,were inert.The injected gas entered through nozzles normally to the main flow.The mixing quality was inferred by temperature profiles on the mixeroutlet.The main flow(air)was heated,whereas the air injected through the mixer nozzles was cold.The temperature distribution over the cross section was measured in 140 points at the mixer outlet.

The measurements enabled to evaluate the value f′/feq,

where i is the enthalpy,G is the flow rate,R is the channelradius,r is the current radius,subscript m corresponds to the main flow parameters,index j corresponds to the injected gas parameters and index eq represents the parameters of the completely mixed flow.

The value f in the case of the inert flow represents the concentration of the injected substance,parameter f′/feqcharacterizes the mixing quality(concentration field uniformity),and feqis the injected substance concentration at the complete mixing.

The value f′/feqdepends on the parameters of the main flow and injected gas as well as on the geometry parameters of the mixer,such as relative nozzle diameter d/D and nozzle spacing S/d,and relative channel length L/D(D is the mixer diameter,d is the nozzle diameter).The data obtained for the specific mixer geometry were processed as a relation

where h/D is the jet penetration in the unconfined cross flow；the value h is determined by the empiric ratio

where n is number of the jet nozzles.

It is evident that the value h/D does not correspond to the jet penetration into the confined flow and does not have any physical sense at h/D≥0.5.It is assumed,however,that it would be possible to reduce the quantity of governing parameters with the use of the“jet penetration parameter”h/D,which depends on the relative hole diameter and momentum ratio,and that the mixing quality could be predicted for the other operating conditions with the use of obtained dependencies.

Full description of the experimental data can be found in literature[15].Fig.2 shows the concentration nonuniformity of the flow on the outlet of the mixing chamber(its length L/D=1)versus the“jet penetration parameter”for one studied variantofthe mixer.The data ofFig.2 demonstrate the high mixing quality at h/D≥0.5,i.e.when the mixer operates in the mode of interaction of colliding(in the near-axis region)jets and gas cross flow.This variant of the mixer was selected for the experimental pyrolysis reactor.

Fig.1.Block flow diagram.

Fig.2.The nonuniformity of the injected gas concentration at mixer length L/D=1.

3.Experimental

3.1.Model facility

To study the process of hydrocarbon pyrolysis in the fast-mixing reactor,a model experimental device was manufactured.Its design represented the main peculiarities of the proposed process scheme.Two sets of experiments were performed using LPG and naphtha as the feedstock.The schematic of the facility in the configuration shown in Fig.3 corresponds to the experiments with naphtha.In contrary to the experiments with LPG,a feedstock evaporation unit is added,whereas the basic units such as a burner,combustion chamber,mixer and reactor remained unvaried.

The reactor consists of two sections with the diameters of 40 and 80 mm,and lengths of 1.15 m and 1.52 m,respectively.The combustion chamber was water-cooled(to protect it from high-temperature combustion products),and the reactor walls were covered with a 10 mm layer of mullite wool.Every part of the facility was made of stainless steel alike Steel 321.Thermocouples and water-cooled samplers were located along the reactor.

The evaporation unit included a cylindrical vessel into which naphtha was charged before the experiments,electrical nitrogen heater and evaporator in which naphtha was sprayed in a hot nitrogen flow.Nitrogen pressed out naphtha from a bottle.The heater presented a stainless-steel coil with the length of 4.5 m,external tube diameter of 8 mm,and wallthickness of1 mm.The coilwas overallthermally isolated.It was heated by the current passing in it.The gaseous nitrogen–naphtha mixture went from the evaporator into the receiver connected to the reactor mixer with 8 individual tubes.To provide the uniform distribution of the flow over injecting nozzles,a disc plate with a critical size orifice was installed in each line.

To choose the mixing chamber geometry,as was mentioned above,the results of the analysis of the quality of the mixing of the jets in the cylindrical channel were used[15].The mixer version taken as basic was the one where the sufficiently high uniformity of the resulting flow was reached within a short distance(Fig.2),namely within the relative length L/D=1 where L and D are the mixer length and channel diameter,respectively.Thus,the reactor mixer geometry was similar to the one studied in the gas-dynamic device,i.e.,the quantity of injection nozzles n and ratio of nozzle diameters and mixer channel(d/D)was similar in both cases.Nozzle diameters in the LPG-and naphtha experiments were 0.75 and 1.0 mm,respectively.

The arguments governing the mixer dimension choice were the following.Reduction ofthe channeldiameter D for geometrically similar mixers atthe same parameters ofthe main and injected flows enables to reduce the mixing time,since the absolute length of the mixing area decreases(L/D being constant)and the mean velocity in the mixer channel increases.

The second factor must be taken into account also.Namely,when the mixer operates in the pyrolysis device,the momentum ratio of the jet to the main flow(J)must be such to provide the jets colliding on the channel axis(h/D＞0.5).As was shown in literature[15],this condition is necessary for every studied mixers,when it is needed to reduce the mixing length.

Fig.3.Experimental fast-mixing reactor.

Thus,it was assumed that for geometrically similar mixers,with the same value of the “jet penetration parameter”and the same distance L/D,the flow nonuniformity would also be the same.In order to verify the validity of such an approach to the mixer geometry choice,the flow uniformity at the outlet of the mixer with the diameter D=15 mm connected to the combustion chamber was studied.Notice that the duct diameter in the gas dynamic model was 40 mm.The measurement technique was the same as in literature[15].The temperature of combustion products at the mixer inlet was roughly similar to the main experiments and nitrogen was used as an injected gas.The measurements were carried out in the cross section L/D=1 from the injection point.The parameters of the main and injected flows corresponded to the “jet penetration parameter”h/D=0.45,which coincided with its value in the experiments with hydrocarbon pyrolysis.

The results of these tests are presented in Fig.2.As is seen,as the geometrical and dynamic similarity of the processes in the mixers(one was analyzed in the gas-dynamic device,the other in the pyrolysis facility)is respected,the values of the parameter f′/feq,which characterize the flow nonuniformity,are in satisfactory agreement with each other.

By convention assuming that the mixing region length is L/D=1,it is possible to evaluate the mixing time by using the mean-velocity value in the mixing channel.In the conditions ofthese tests,this value is about 0.05 ms.The rate of feedstock heating depends on the ratio of the feedstock temperature difference at the mixer inlet and outlet to the mixing time；it is about 107K·s-1.

3.2.Measurement technique and experimental conditions

The hydrogen–air combustion products(with small hydrogen excess over stoichiometry)were used as the heat carrier.Hydrogen excess was needed to prevent possible incomplete utilization of the air oxygen.During the experiment,the flow rates ofnaphtha,nitrogen,air,and hydrogen were measured as well as the pressure in the combustion chamber and mixer receiver,cooling water flow rate and its heating in the cooling annulus,the temperature of the naphtha–nitrogen mixture and temperature distribution along the reactor axis.The temperature was detected by K-type thermocouples(junction diameter 0.7 mm),regarding the correction foremission.The gassampleswere collected into 150 mlsyringes.The sampler presents a copper tube of 6 mm in diameter with a stainless-steel capillary of 1 mm in diameter.The 90-degree angled capillary nose was entered into the tube mouth and flush-sealed on the wall.The capillary intake was oriented toward the flow.The tube of sampler passed through the reactor walls.Pyrolysis product quenching occurred directly in the capillary cooled by the water flowing in the tube.

Samples from each station were collected with two syringes and the mixture composition was analyzed with two gas chromatographs:Crystallux-4000 M(TCD detector)and Ajilent 6850(FID detector).Aside from hydrocarbons,the content of hydrogen,carbon oxide,and nitrogen in the mixture was detected.The total amount of the detected hydrocarbons reached 42.

Fig.4 shows the temperature history to illustrate the sequence of actions in experiments with naphtha.First,nitrogen flow rate is adjusted,and then the electric heater is switched on and the walls of the heater–evaporator–receiver system are heated.A dashed line in Fig.4 shows the temperature in the receiver.At the moment t=6 min,the combustion chamber is initiated,and the heating of the combustion chamber,mixer and reactor is started.Solid lines in Fig.4 mean the indications of the thermocouples located on the reactor axis.The top line corresponds to the thermocouple at 45 mm from the injection cross section,the bottom one—at 1875 mm.At the moment t=18 min(an arrow)the naphtha supply is started.Three minutes later the sampling is started and all operating parameters are almost unchanged within the sampling process.Then the supply of hydrogen and naphtha is stopped,the current source is off and the device is cooled by air and nitrogen for a certain period of time.

Fig.4.Temperature temporal history in experiment.Thermocouple positions/cm:1—3,2—4.8,3—7.5,4—22,5—42,6—72,7—147；dotted line—temperature of injected naphtha–nitrogen mixture.

3.3.Feedstock

In the LPG-tests,the mixture of such composition was used:С3Н8—71.3%,n-C4H10—4.1%,i-C4H10—4.2%,C2H6—19.7%,C3H6—0.3%and C4H8—0.4%.The used naphtha was characterized by the following parameters:density of 714.1 kg·m-3(15°С),initial boiling point of 35°С and final boiling point of 167°С.110 components were detected in the naphtha composition.The group composition was:normal hydrocarbons,34.031%；aromatics,4.834%；iso+cycles,54.812%；fractions С1–С4,6.127%and ole fins,0.196%.

3.4.Experimental conditions

Pyrolysis of LPG in the fast-mixing reactor was studied within the temperature range at the reactor inlet T0=1300–1520 K.The value of T0was detected by the thermal balance on the assumption of the inert mixing of the feedstock and heat carrier,and flow uniformity at the mixer outlet.Hydrogen combustion efficiency determined by the gas-analysis was above 99%.When determining T0,the quantity of the heat supplied in the cooling jacket of the combustion chamber and measured in each test,was taken into account.The feedstock/heat carrier ratio was varied within the range of 3.6–13.3.

The main parameters featuring the operation conditions in the experiments with naphtha are presented in Table 1.

Table 1 Operating conditions

4.Modeling

In addition to the experimental study,a simplified quasi-onedimensional model was developed to describe the hydrocarbon pyrolysis inside the fast-mixing reactor.The simulation was started at the distance of x/D=1 from the jet injection point where the flow was assumed to be uniform,and the process of feedstock/heat carrier mixing was so quick that feedstock decomposition in the mixing region could be ignored,that is T0=Tinert.Further temperature variations followed the measurements.

A kinetic model[16]included 689 reactions involving 155 species was used for the modeling of the LPG pyrolysis.This model involved hydrocarbon molecules and radicals with carbon atom number up to 5 and some high-molecular hydrocarbons including aromatics,such as benzene and naphthalene.Reactions with oxygen-containing species were ignored.The H2O molecule was treated solely as an inert one in three-body reactions.Reactions with carbon were also ignored.The model was tested by the results of the study of ethane pyrolysis in a shock-wave reactor[13].The simulations by this model agree well with the distributions of ethylene and ethane concentrations over the reactor length[14].However,as the concentration of other hydrocarbons in the pyro–gas was small during the ethane pyrolysis,it was impossible to define reliably the correctness of the model for the description of the reaction kinetics for the mixtures of hydrocarbons used in this work.

The simulations involved a kinetic model including 764 reactions with 197 species was used in case of naphtha pyrolysis.The kinetic model was combined from three parts.The model of butane pyrolysis[16]approved before[17]was used as a base.Additive reactions included saturated and unsaturated hydrocarbons with carbon atoms quantity up to 11[18]and cyclic hydrocarbons—cyclohexane,methylcyclohexane[19],etc.The species containing oxygen and respective reactions were excluded.In order to simplify the scheme,paraffins with the atom number of 7 and above were joined in groups.For example,the group “n-C7H16”contained all normal and isomers of non-cyclic hydrocarbons with the carbon atoms number 7.Non-aromatic cyclic hydrocarbons were divided into two groups— the“cyclohexane”and the“methylcyclohexane”group.The “cyclohexane group”contained the cyclic hydrocarbons without associated methyl,ethyl and other admixtures,while the“methylcyclohexane group”included all the other cycles.Respectively,in the initial conditions the mass concentrations of the particles in one group were summed.

The naphtha composition used in simulation is presented in Table 2.

Table 2 Reduced naphtha composition

5.Results and Discussion

5.1.Experiments with LPG

The results of the run where the ethylene yield was the highest was considered.Characteristic temperatures of the process were:combustion product adiabatic temperature Tad=2400 K(p=0.12 MPa),flow temperature at the mixer inlet T0=1750 K and inert mixing temperature Tinert=1400 K.The ratio of heat carrier/feedstock mass flow was 13.3.Characteristic mixing time tmix=0.05 ms.

The distribution of the mass concentrations of main pyrolysis products and temperature profile along the reactor axis is presented in Fig.5.The upper x axis shows the distance from the section located at 15 mm from the jets injection point(L/D=1),the lower one represents the residence time correlating to a certain x-coordinate value.The temperature distribution from the plane L/D=1 to the first experimental point is approximated by a straight line.Note that the temperature distribution along the reactor axis in these experimental conditions is effected by the relative small flow nonuniformity at the mixer outlet,as well as by the thermal boundary layer on the walls.But,since the major processes in the reacting flow occur in the reactor inlet region,the heat flux to the wall resulting in the temperature reduction on the axis promotes the reaction retardation in the area of decreased temperatures.The main variation in the pyrolysis product composition is observed within the temperature range of 1400–1200 K within ～20 ms,and then the dependences Ci(t)are noticeably flattened.Fig.5 shows fast conversion of the feedstock in the high-temperature region.Even at the distance of x=0.09 m(t=1.09 ms),the mass concentration of propane is 0.36 and in the last sampling point(x=2.8 m,t=126 ms)it decreases to 0.058.In accordance to the feedstock rate expense,the ethylene concentration reaches the value of 0.473 at the distance x=2.8 m.Moreover,in the first sampling point(x=0.09 m),the ethylene mass fraction is 0.318.Propylene concentration changes weakly over the reactor length—from 0.086 in the first sampling point to 0.073 in the last one.Hence,the ratio of С3Н6/С2Н4at the reactor outlet is 0.15.

More detailed information on the product composition is presented in Fig.6,where the product yields are compared to the similar values in the furnace pyrolysis.The data for SC-1 furnace[4]were selected for comparison because this furnace featured the highest ethylene selectivity.It is evident that the ethylene yield in the fast-mixing reactor is much higher and the propylene yield is lower than in the conventional method of pyrolysis.It is also noted the lower methane concentration and higher acetylene content in our experiments as compared to literature[4].These differences result from both the almost 200 K-increase of the maximum process temperature and the temperature profile character reversal.The temperature rises downstream in the furnace-pyrolysis method,whereas the temperature decreases as the residence time increases in the reactor in the case of pyrolysis in the heat carrier flow.The process kinetics should be changed.

5.2.Experiments with naphtha

Three experiments were carried out.The airand hydrogen flow rates were almost permanent and the temperature of the flow after mixing calculated on the assumption of inert feedstock/heat carrier mixing(Tinert)was varied due to the variation of injected naphtha/nitrogen mass flow rates.In each case,the injected naphtha temperature was higher than the final boiling temperature.

Figs.7–9 illustrates the temperature change and main pyrolysis product distribution along the reactor axis.The data of the experiment 2 are considered in detail.In this test,the flow rates of hydrogen,air,nitrogen and naphtha are 0.18 g·s-1,10.25 g·s-1,3.0 g·s-1and 1.85 g·s-1,respectively.The volume fraction of naphtha in the heat carrier/feedstock composition is 0.03.The rest of the experimental parameters are presented in Table 1.The fraction of the components presented in Fig.7 is approximately 95%of the total pyrolysis products,and total amount of the detected species exceeded 30.Fig.7 presents quite a complete image of the product content.

Fig.5.Temperature and product yields vs.residence time and distance from reactor inlet.

Fig.6.Comparison ofproductyields in the fast-mixing reactorand steam cracking process.

Fig.7.Temperature and productyields vs.residence time and distance from reactorinletat T inert=1320 K.

Fig.8.Temperature and product yields vs.residence time at T inert=1240 K.

The data in Fig.7 show the extremely high reaction rate at the initial stage of naphtha thermal decomposition.Because of the successful location of the thermocouples and samplers,the process is analyzed in details.The first thermocouple is located at 15 mm from the injection cross section(x=0)and the first sampler is within 24 mm from this cross section(x=9 mm).The temperature in the first point(1340 K)is close to the design temperature Tinert=1320 K,calculated on the assumption of the inert mixing.In the further points(x=30 and 48 mm),however,the temperature varies insignificantly(1300 and 1335 K,respectively)instead of the expected drop.It begins to decrease only upon this point(1304 K at x=78 mm,t=0.33 ms).Such a temperature behavior can be resulted from the incomplete mixing within the range L/D=1.The temperature on the axis here evidently differs from the average value and on a certain distance L/D＞1,the endothermic decomposition process passes in the conditions of weakly varying temperature due to the in flow of heat to the axis area from neighboring areas with the higher temperature.When analyzing the dependence T(t),it can be assumed that the mixing time tmix≈0.05 ms is insufficient for the complete feedstock/heat carrier mixing.Moreover,since the inert mixing condition in this experiment(T0=1750 K,Tmix=1320 K)does not realize completely,the reactions apparently begin in the high-temperature areas of the mixing region.In particular,this is indicated by the fact that in the first sampling point corresponding to t≈0.02 ms,the ethylene concentration reaches 35%,whereas the propylene concentration is maximum(15%).As the value Tinertis reduced to 1240 K(Fig.8),the effect of above-mentioned factors is weakened and the ethylene concentration is 25%in the first sampling point,but the maximum values in both cases differ insignificantly.The issue of temperature history in fluence on the product yield will be considered below during the analysis of the simulation results.

As was mentioned before,the distinctive feature of the concentration profiles in Fig.7 is the fast growth of the concentrations at the initial stage of the process.Then,at t≥25 ms,the dependencies Ci(t)become smooth.Almost permanent level of the concentrations within big distances from the inlet is maintained by the relative low flow temperature(T＜1100 K).Fig.7 does not reflect any in fluence of the secondary reactions which would result in the ethylene concentration reduction.

Fig.9.Temperature and product yields vs.residence time at T inert=1400 K.

Fig.10 illustrates the temperature in fluence on the composition of pyrolysis products.The gas analysis data are presented in the positions of maximum ethylene concentration.In various tests,these distances represent the residence time t=24–34 ms.High ethylene yield is recorded in all experiments.As the temperature grows,the maximum ethylene concentration in the pyrolysis products rises to 49.4%at Tinert=1400 K.The propylene yield decreases along the reactor(as was mentioned above,the maximum propylene yield is observed near the reactor inlet).The butadiene yield also decreases with increasing temperature.Low level of high-molecule compounds C5+should also be noted.But this is typical only for the late stage of naphtha decomposition.In the early stage,the content of fraction C5+in the pyrolysis products is much higher.Above 20 components of fraction C5+with the total concentration of 40,10,and 5%are identified in the first sampling point at Tinert=1240,1320 and 1400 K,respectively.As the residence time increases,fraction С5+decreases significantly.

Fig.10.Effect of temperature on the product yields.

6.Parametric Analysis of the Hydrocarbon Pyrolysis in the Fastmixing Reactor

As was mentioned above,the experimental conditions in the model device(Fig.3)do not completely correlate with the proposed process scheme(Fig.1).First,the heat carrier composition is different.in real conditions.It is the products of combustion of a stoichiometric hydrocarbon–oxygen mixture diluted with the superheated steam(i.e.CO2+H2O),whereas in the experiments the products of combustion of the stoichiometric hydrogen–air mixture(N2+H2O)are used.Second,in real conditions the reactor walls must be thermally insulated in order to increase the energy efficiency ofthe process.Thermalinsulation of the experimental reactor is incomplete and the effect of the heat transfer to the wall affected the temperature distribution on the reactor axis,especially far from the reactor inlet.The effectof the first factor lies in the fact that at the same temperature(owing to heat capacity difference),the enthalpy of the real heat carrier will be higher.And although the heat carrier itself does not participate in the reactions,the temperature distribution along the reactor will not be identical,which should cause the variation of concentration profiles.

To analyze the effectofthe above-mentioned factors on the pyrolysis process and to predict the process characteristics in real conditions,a series ofsimulations were performed with the above-described theoreticalmodel.The modelreliability was verified by means ofcomparison of the simulation resultsand experiment.The simulationswere performed for the conditions of each test and their results are presented in Figs.5 and 7–9.As is seen,the simulations results describe quite well the behavior of the concentration profiles of the major components of the reacting mixture even near the reaction beginning area.It is especially valid for ethylene that the experimental profiles of its concentrations are in good agreement with the simulation within the wide temperature range.For the other components,their total concentration is 10%–15%and the simulation results are in worse agreement with the experiment[Fig.7(b)].A number of factors in fluence the correctness of the used model,such as the chosen kinetic scheme and kinetic constants,hydrocarbon composition admitted for the simulations,and also the assumption on the complete and inert mixing of the feedstock and heat carrier.

6.1.Propane feed

The simulation results for the conditions of propane pyrolysis reactor are first considered.Propane is chosen as the feedstock,because this component is the main one in the composition of the LPG utilized in our experiments.It permits to evaluate the effect of the difference between the model heat carrier and the real one on the pyrolysis process.Initial propane temperature is taken to be equal to 300 K.The heat carrier is the product of combustion of the stoichiometric methane–oxygen mixture diluted with the superheated steam.The temperature of methane and oxygen is 300 K.The steam temperature is taken to be 500 K.Steam flow rate supplied into the combustion chamber depended on the condition of T0=1900 K at the mixer inlet.As discussed earlier,the feedstock/heat carrier mixing was assumed to be instant.Propane flow rate is correlated to the values,the inert-mixing temperatures Tinert=1200,1300,1400,and 1500 K,the ratio of heatcarrier and feedstock flow rates is g=1.7,2.3,3.4,and 4.4,respectively.

It should be noted that according to the experimental results from the model device,the value of T0chosen for the simulations is likely to be the maximum for LPG pyrolysis since further increase of T0at the assigned value of Tinertdoes not lead to the growth of ole fins yield.Hence,the mixture temperature at the mixer inlet governs the steam/combustion product ratio and Tinertvalue dictates the heat carrier/feedstock ratio.The chosen range of Tinertcorrelates to the one studied in the model device.The geometry applied in the experiments enables to realize the inert-alike mixing conditions.With the values of T0and Tinertchosen for the simulations,the operating conditions of the mixer are favorable for the fast qualitative mixing since the“jet penetration parameter”is higher than thatin testwith the maximumyield ofole fins(T0=1750 K,Tinert=1400 K,g=13.1).According to our experiments[14],with the constant mixer geometry,the growth of the“jet penetration parameter”h/d,which is caused in this case by the rising ratio ofthe flow rates(momentum)of the injected and main jets,results in the improvement of mixing quality.The mixing time of～0.05 ms in our tests is enough to realize such conditions in real case.For this reason it is possible to ignore the effect of the reactions in the mixing region on the pyrolysis process(to a first approximation).Thus,the mixer geometry applied in our experiments should provide the implementation of the assumption of the inert mixing and flow uniformity at the reactor inlet adopted in the calculations.

The effect of the temperature Tinerton the pyrolysis process is first considered.In these simulations,the pressure in the reactor is assumed to be equal to 0.1 MPa.Fig.11 presents the reacting flow temperature versus the residence time at various temperatures at the reactor inlet.In each simulation version,fast temperature drop is noted.Later,the rate of temperature variation decelerates and the pyrolysis process continues in almost isothermic conditions at high values of t.

Figs.12–14 illustrates the in fluence of the initial temperature on the yield ofthe mostvaluable pyrolysis products ofethylene and propylene.In addition,the concentration profiles of propane and ethane are also shown.Figs.12–14 also present the time variation of the sum concentrations of ethylene and propylene,as well as the same sum with due regard to the С2Н6and С3Н8recycle(propane and ethane from product yields are assumed to be added to the feedstock).In this case the ethane efficiency is assumed to be the same as the propane efficiency,i.e.(С2Н4+ С3Н6)rec=(C2Н6+ С3Н6)/(1- С2Н6-C3H8).

Fig.11.Temperature profiles in the real heat carrier at T inert=1300,1400,and 1500 K.

Fig.12.Simulation of the propane pyrolysis at T inert=1300 K.

Fig.13 shows not only the simulation results but also the values of the totalconcentration ofС2Н4andС3Н6,including the data with recycle of С2Н6and С3Н6,obtained in the experiments with the model heat carrier(products of combustion of the hydrogen–air mixture).In this case,the effect of the heat carrier composition on the total concentration of light ole fins is relatively small.

Fig.13.Simulation of the propane pyrolysis at T inert=1400 K.

Fig.14.Simulation of the propane pyrolysis at T inert=1500 K.

Similar to the experiments with the model heat carrier,the simulations predict drastic rise of ethylene and propylene concentrations at the initialstage ofthe pyrolysis process and the maximumС3Н6concentration is reached much earlier than the maximum С2Н4concentration.Initial flow temperature highly effects the time of achievement of the ethylene concentration maximum from≈700 ms at Tinert=1300 K to～5 ms at Tinert=1500 K.The time of achievement of the maximum concentrations of sum of ole fins varies within smaller range from～1 ms at Tinert=1500 K to～120 ms at Tinert=1300 K.It is also noted the reduced contribution of the recycle flow in the total concentration of light ole fins as the residence time in the reactor increases since the propane conversion rises fast,and ethane concentration is small and slightly changes in time.

The data of Fig.15 generalize the simulation results presented in Figs.12–14 and demonstrate the in fluence of the initial temperature on the main indexes of the pyrolysis process.Because of the different types of time dependencies of concentrations of С2Н4,С3Н6,С2Н6and С3Н8,the maximums of С2Н4and С2Н4+ С3Н6do not coincide atdifferent temperatures.Total concentration of ole fins rises almost linearly together with the temperature.The change of the considered parameters within the studied temperature range is relatively low.The data of Fig.15 permit to find the optimum value of the initial temperature Tinert≈1400 K,which correlates to the maximum concentration of sum of ole fins(0.577).With regard to the ethane and propane recycle,this value rises up to 0.641.

Fig.15.Effect of temperature on the ole fin yields.

Aside from the temperature,the effect of pressure on the propane pyrolysis is also studied.The simulations are performed as the pressure in the reactor is 0.1,0.5,and 1.0 MPa.The initial temperature is taken to be equal to 1300 K.Fig.16 shows the varying concentrations of ethane and propane and their sum versus the pressure within the wide range of the residence time in the reactor.As is seen,the ethylene concentration more depends on the pressure,which increases as the residence time in the reactor rises.But in the range of t values where the total concentration of ole fins approaches its maximum,the difference in data at p=0.1 and 1.0 MPa is small.The data of Fig.17,which show the varying concentrations of С2Н4,С2Н4+ С3Н6,and sum of ole fins with regard to the recycle(in the maximum point of the sum of С2Н4+ С3Н6),demonstrate the weak in fluence of the pressure on the total ole fin concentration.

Fig.16.Comparison of ole fin yields at the pressure p=0.1 MPa(solid lines)and p=1 MPa(dashed lines).

Fig.17.Effect of pressure on the ole fin yields at T inert=1300 K.

6.2.Naphtha feed

The results of the similarinvestigation for the case of naphtha pyrolysis in the fast-mixing reactor are now considered.In addition to the experimental data in the model device,the simulation results for the real reactor operating conditions correlating to the proposed process scheme are utilized.The simulation parameters involved the possibility ofimplementation ofthe assumption on the inert-type ofthe feedstock/heat-carrier mixing,which is well confirmed by our experiments.The simulation is carried out for three temperatures of the feedstock/heat carrier mixture at T0=1320,1400 and 1500 K.The first two values are the same as in tests 2 and 3.In these simulation variants,the heat carrier temperature before the mixer is the same as in the experiments(Table 1),and for the case T0=1500 K,the value of2100 K is adopted.It is assumed that naphtha is supplied in the mixer as a vapor at 500 K.These temperatures correspond to the feedstock/heat carrier ratio g=2.3,3.17 and 2.54.The rest simulation conditions are the same as in the LPG case.

The effect of the heat carrier composition on the pyrolysis process is first analyzed.To do this,the results of the experiment 2(the model heat carrier Н2О +N2)are compared with the simulations for the real heat carrier(H2O+CO2).In this case,for condition correctness,the injected flow in the simulations presents a naphtha/steam mixture at 500 K in order to provide similar value g=7.1 both in the experiment and simulation.Hence,the difference is in the heat carrier composition alone.The results in Fig.18 show that in this case the effect of the heat carrier composition on the concentration profiles is weak.It is evident that the significant dilution of the feedstock by the heat carrier promotes this fact.

Fig.18.Comparison of the naphtha pyrolysis simulation with real heat carrier(lines)and experiment with model heat carrier(symbols):T inert=1320 K,g=2.3.

In the other simulation cases at lower g,the difference between the simulation and experimental data is higher.It results from the difference of not only heat carrier enthalpy(at the same temperature),but also ofg values.Common in fluence ofthese factors on the pyrolysis process imposes different effects on the ethylene concentration profile behavior.

Above facts are illustrated in Figs.19 and 20.In both cases,the simulation predicts dramatic temperature drop in the early stage of the process and then the temperature ofthe reacting flow is almostconstant.In turn,ethylene concentration rises significantly in the inlet section.At T0=1320 K and g=2.3,the temperature is relatively low in the “isothermal”section(1100 K)and the ethylene concentration reaches its maximum(46.8%)only at t=460 ms.In the experiments at the same initial temperature and g=7.3,the temperature level in the initial section is higher than in the simulation,and the maximum С2Н4is reached much faster at t≈ 34 ms,and its value(47.6%)agrees well with the simulation.In the case of T0=1400 K and g=3.17,the growth of both parameters as compared to the previous version causes the increase of the flow temperature in the “isothermal”section up to 1165 K.As a result,the ethylene maximum is reached noticeably faster and the value itself is in good agreement with the experimental finding(Fig.19).

As expected,further increase of the initial temperature up to 1500 K again intensifies the pyrolysis.The parameters of this simulation are chosen in such a way to increase the temperature of the “isothermal”section of the temperature profile even more.In this case it is equal to 1220 K.The maximum ethylene concentration corresponds the residence time in the reactor t≈15 ms(Fig.20).

The data presented in Figs.21 and 22 summarize the results of the numerical simulation of the naphtha pyrolysis process in the fast-mixing reactor(in the framework of adopted simplifying assumptions).Fig.21 illustrates the temperature in fluence on the yield of ethylene,ethylene/propylene sum,and the same sum with regard to ethane recycle.Here,the insignificant contribution of the recycle in the total yield of ole fins and weak temperature effect on the ethylene yield are noted.

Fig.19.Comparison of the naphtha pyrolysis simulation with real heat carrier(lines)and experiment with model heat carrier(symbols):T inert=1400 K,g=3.17.

The data in Fig.22 demonstrate the total composition of the pyrolysis products at the moment of achievement of the maximal ole fin concentration in tmax=246,24.6,and 4.6 ms at T=1320,1400,and 1500 K,respectively.Note the two main peculiarities of Fig.22 in respect to the ole fins.All simulation variants feature high ethylene yield and as the temperature increases,the propylene yield also rises.The latter is related with the time tmaxreduction as the temperature rises.The shorter the residence time,as is indicated in the simulations and experiments,the higher the propylene concentration in the pyrolysis products.

Finally,the compositions ofthe naphtha pyrolysis products in the fastmixing reactor are compared with the similar data for the furnace pyrolysis.For comparison,the data of furnaces SRT(company ABB LUMMUS GLOBAL)representing the full-range naphtha are taken.Fig.23 shows the data of the experiments with the model heat carrier at T0=1400 K,the results of pyrolysis simulation in the heat carrier flow at the same temperature,and also the respective data of SRT furnaces[3].As compared to the conventionalmethod,the significantincrease ofthe ethylene yield is the peculiarity ofthe pyrolysisprocessin the heatcarrier flow.The main reason is the higher process temperature and respective reduction of the residence time,and also reversing temperature variation,i.e.,the temperature in the soaker rises downstream,while it decreases in the fast-mixing reactor with changed kinetic processes.

Fig.20.Comparison of the naphtha pyrolysis simulation with real heat carrier(lines)and experiment with model heat carrier(symbols):T inert=1500 K,g=2.54.

Fig.21.Effect of temperature on the ole fin yields(real heat carrier).

Fig.22.Effect of temperature on the all product yields(real heat carrier).

Fig.23.Comparison of the naphtha pyrolysis in fast mixing reactor with furnace method.

7.Conclusions

This article presents the results of simulation and experimental investigation ofthe pyrolysis ofLPGand naphtha in a high-temperatureflow at the ultra-short time of feedstock/heat carrier mixing.Under study is the in fluence of the heat carrier composition,pressure,and residence time on the composition ofthe pyrolysisproducts.The experimentsperformed in the model facility demonstrate high ethylene selectivity.The results of the simulations tested by the experimentaldata also predicthigh efficiency ofthe pyrolysis in the fast-mixing reactorforrealconditions ofpractical implementation.Among other things,relative growth of the ethylene yield relative to the furnace pyrolysis ofpropane is 1.2,whereas fornaphtha pyrolysis itreaches value about1.5.In such a case the sumofethylene and propylene yield is also higher.The concept of the fast-mixing reactor adopted in our experiments can also be applied in various other cracking processes with short residence time.

Nomenclature

C mass concentration,%

D mixer channel diameter,mm

d mixer orifice diameter,mm

g ratio of heat carrier and feedstock flow rates

h jet penetration depth,mm

L mixer channel length,mm

P pressure,MPa

S mixer orifice spacing,mm

T temperature,K

t time,ms

x distance from injecting plane,m

Subscripts

ad adiabatic process

i a mixture component

inert corresponds to the inert conditions of feedstock with heat carrier mixing

r chemical reaction

rec recycling

0 corresponds to the conditions at the mixer inlet