Study on co-cracking performance ofdifferent hydrocarbon mixture in a steam pyrolysis furnace☆

Benfeng Yuan ,Jinlong Li*,Wenli Du ,Feng Qian ,*

1 Key Laboratory ofAdvanced Controland Optimization for ChemicalProcesses,Ministry ofEducation,East China University ofScience and Technology,Shanghai200237,China

2 Schoolof Information Science and Technology,East China University ofScience and Technology,Shanghai200237,China

1.Introduction

Ethylene and propylene are important raw materials of the petrochemicalindustries for other chemicalproducts and mainly produced by the cracking of petroleum distillates.The cracking is a process where the different hydrocarbons are transformed into light hydrocarbons such as paraf fin,ole fin and so on in a steam pyrolysis furnace(thereafter called furnace).Atpresent,there are severaldifferentcracking routines applied in the industries[1],such as catalystcracking[2,3],hydrocracking and steam cracking.

The steam cracking has a wide range ofapplications in the chemical industries,and itis also the mostenergy-consuming process in chemical industries.It is found that the radiation section ofa furnace consumes approximately 65%oftotalprocess energy[4].In a furnace,a gaseous or liquid hydrocarbon feedstock like naphtha,LPG,ethane and propane is thermally cracked with steaminstead ofoxygen.Atpresent,the naphtha is a widely used feedstock and its consumption continues to increase in China.Nowadays,it is one of the most important feedstocks of the industrialfurnace[5].

A steam pyrolysis furnace is mainly composed of two sections:a convection section and a radiation section.Some investigations have been performed in the convection section[6–9],however,more attention has been paid to the radiation section in the past few decades,in which the mixtures ofhydrocarbon feedstocks are thermally cracked into ole fins,aromatics,methane hydrogen and so on.In the cracking process,the productyield distribution is strongly affected by the hydrocarbon partialpressure,the coiloutlet temperature,and the resident time,especially the type offeedings.High ole fin yields tend to high temperature,low partialpressure,and high residenttimes[10].To predict the effect of these parameters on the ole fin yields,a good reaction model is essential,and the modeling methods can be divided into three kinds:empiricalmodel,molecular reaction modeland free radical reaction model.A detailmodelbased on the individualmolecular reactions had been outlined by Van Damme etal.[11].Van Geem etal.[12]carried out an investigation of the cracking simulation ofethane by a 1-Dmodeland a 2-Dreactormodel,and obtained the effectof the temperature pro files on yields in steam cracking.Marco etal.[13]established a kinetic modeling study ofethane cracking for optimalethylene yield.These models based on free radialreactions just apply to the chemical experiments in the laboratory without taking into consideration the realapplication problems in ethylene plants.The molecular reaction modelwas proposed by kumar etal.[14],in which one primary reaction modelwith a first-order primary step and 21 secondary reactions were included.The Kumar modelhas been widely used in later researches[15–17].For instance,Hu et al.[18,19]carried out a coupled simulation ofa steamcrackerwith Kumarmodelto study the flow,combustion and heat transfer process in the furnace and obtained product yields and coke rates.

Simulation,control and optimization of ethylene cracking furnace can be carried out based on these reaction models.Masoumi etal.[20]developed a static and a dynamic cracking modelto predict the optimal temperature pro files along the reactor and study the performances of temperature control loop for different controller parameters and disturbances.Mehdi etal.[21]carried outa dynamic optimization of a cracking reaction in a tube reactor by a 1-dimensionalpseudo-dynamic modeland improved operating pro fit by 13.1%.Shu et al.[22]proposed a distributed parameter model for a tube reactor in an ethylene cracking furnace based on 3-Dcombustion monitoring,and predicted the distributions of the heat flux and the tube temperature.

In the past years,more attention has been paid to the cracking of single feedstock like naphtha,ethane and propane.As the crude becomes heavier and heavier allover the world,the costof the cracking of hydrocarbon feedstocks for ethylene and propylene increases gradually,especially in China whose demand for ethylene and propylene is high and increases gradually.Thus,the selection and optimum allocation of raw materials have received more attention,and the co-cracking of different feedstocks becomes more and more common.Co-cracking is a process where severaldifferentfeedstocks are cracked together.Many researchers think that thermalcracking is a chemical reaction based on free radical mechanism,and thus an easily cracking and high-selectivity hydrocarbon can promote decomposition of other dif ficulty cracking components in the cocracking process[23].

In the present work,the simulations of co-cracking of ethane and propane,and naphtha and LPG mixtures are performed with Coilsim1D[24,25]in tube reactors of two different furnaces:one is an industrial USC-Mfurnace and the other USC-U furnace.This paper aims to study the performance of the co-cracking ofethane and propane,and naphtha and LPGmixtures in the tube reactor ofa furnace,and the effects of the mixing ratio oftwo differentfeedstocks,the coiloutlettemperature and the coil outlet pressure on the product distribution are explicitly discussed.

2.Mathematical Models and Physical Properties

Colisim1D[24,25]is developed by Laboratory for ChemicalTechnology in Ghent University.And itis a commercialsteam cracking software developed to simulate the steam cracking of the hydrocarbons in a tube reactor including the simulations of the tube reactor and the run length of industrialsteam cracking tubes.The simulation is performed based on a free-radical reaction mechanism and a 1-dimension reactor model.The software package SimCo is used for feedstock reconstruction using a neuralnetwork method and a Shannon entropy optimization method.

2.1.Reactor model

The steam cracking is a non-isothermal,non-adiabatic and nonisobaric process.A 1-D reactor modelhas been used in this work ignoring the effectofgradients in the radialdirection.In a 1-Dreactormodel,a set of steady state continuity equations for feed components are solved simultaneously with the energy equation and the momentum equation.These equations have been listed in Table 1.

To perform a simulation of the cracking modelthe reactor conditions should be needed.In Coilsim1D,four methods have been speci fied:the temperature and pressure pro file,the heat flux pro file[26],the external walltemperature pro file and the outletconditions.In each way the coil inlet temperature,coilinlet pressure and the dilution ofsimulation are allrequired.

2.2.Feedstock model

The detailfeedstock components are needed for a free-radicalreaction model,butnotfor a molecular reaction model.During the past years,severalanalyticaltechniques to generate a detailed molecular composition have been developed and improved such as gas chromatography(GC),gas chromatography–mass spectrometry(GC–MS)and high performance liquid chromatography(HPLC)[24].However these techniques are error-prone and time-consuming for chemicalindustries.In general,only severalglobalcharacteristics are measured on line called the commercialindices such as the speci fic density,the PIONA weight fractions and a set of ASTM boiling points.Thus,many methods have been proposed to getthe information of detailed compositions from the commercial indices[24,27,28].

In this work,the simulation package SimCo is used to generate the detailed compositions based on a number ofcommercialindices.This soft contains two methods for the molecular reconstruction.One is the Shannon Entropy method which is the main method for allkinds of feedstock,and the other is an arti ficial neutralnetwork(ANN)which is only used for naphtha fractions[29].ANN can only be used when the naphtha is indeed within range and allnecessary commercial indices are available from the input file.The inputs of this neutral network are the initial,50%and finalboiling point of the ASTMD86 boiling point curve,the globalPIONA weight fractions and the speci fic density.

2.3.Physicalproperties

Usually,the steam cracking is always carried out at relatively low pressures and high temperatures.Hence,the process gas can be considered as an idealgas mixture in the tube reactor,and thus the density of the gas mixture can be calculated by the idealgas law:

The viscosity of the mixture is given by:

In this equation,μiis the viscosity of the pure component i and is de fined as[30]:where Pciand Tciare the criticalpressure and the criticaltemperature of the component i,and Tris the ratio of the gas temperature and the criticaltemperature.In Eq.(5),the interaction parameter Φijis calculated by

Table 1 Balance equations

Table 2 Comparison between the modeling results and the experimentaldata(naphtha)

Table 3 Comparison between the modeling results and the experimentaldata(LPG)

Table 4 Commercialindices of naphtha

Table 5 Feedstock compositions of LPG

Table 6 Reactor conditions

Fig.1.Effect ofmass fraction(C2H6)on CH4,C2H4,C3H6.

Fig.2.Effect of mass fraction(C2H6)on the totalyields of P&E and high value-added products.

Fig.3.Effect ofmass fraction(LPG)on CH4,C2H4,C3H6.

Fig.4.Effect of mass fraction(LPG)on the total yields of P&E and high value-added products.

The thermal conductivity is calculated in the same way as the viscosity:

The heat capacity of the mixture is computed based on a mass fraction average of the heat capacities of pure species as

3.Model Validation

Fig.5.COT effect on ethylene and propylene yields(C2/C3).

To verify the reliability of the theoreticalmodel,some simulations of naphtha cracking and LPG cracking are performed with Coilsim1D,and different convergence conditions of COT and P/E are speci fied.Model validation is carried outby comparing the modeling results with the experimentaldata obtained from the petrochemicalindustries.

Table 2 compares the modeling results and the experimentaldata of naphtha cracking.As can be seen,the modeling results are all a bit smaller than the experimentaldata expect for the yield ofhigh valueadded products with the convergence condition of P/E,and the results with the convergence condition of P/E are closer to the experimental data than that with the convergence condition of COT.The average relative errors of the product yields are allbelow 7%.The comparison between the modeling results and the experimentaldata of LPG cracking is shown in Table 3.As shown,the average relative errors ofproduct yields are smallexcept for the error ofpropylene yield,and the results with the convergence condition of COT are closer to the experimental data.

In general,the modeling results are in good agreement with the experimentaldata and can re flect the product yields in the realsteam pyrolysis furnace.

4.Model Setting

4.1.Reactor geometry

The simulations of U-shape and W-shape tube reactors are performed in this work.Some parameters of tubes should be speci fied like tube layer,tube number,length of the coil,tube internaldiameter and wallthickness.

4.2.Feedstock setting

Ethane and propane,and naphtha and LPG mixtures are chosen as the feedstocks for the USC-Mfurnace and the USC-U furnace.The detailed information ofnaphtha and LPG is presented in Tables 4 and 5.

4.3.Reactor conditions

The simulation model provides a set of ordinary(non-linear)differentialequations which need to be solved.As mentioned before,the reactor conditions determine the solution of these equations.In this work the outlet conditions are speci fied as cracking severity indices such as the propylene/ethylene ratio,the methane/propylene ratio,a key component conversion,or a fixed yield of ethylene or methane.

Fig.6.COT effect on the yield of P&E and high value-added products(C2/C3).

Calculations are performed with the shooting method[31]which can calculate the inlet pressure and the heat flux pro file corresponding to the speci fied cracking severity indices.For an industrialfurnace,a pressure related to severity index(either COP or the ethylene to ethane yield ratio)and one of the temperature related severity index(propylene to ethylene yield ratio or COT)are available at the reactor outlet,so they can be set as boundary conditions in a simulation for an industrialfurnace,then the inletpressure and temperature become a settable variable to get the desired outlet speci fication.

The outlet conditions of the single tube reactor consist ofmass flow rate,hydrocarbon/steam,the coil inlet temperature and pressure(a settable value),the coiloutlet pressure and temperature.The detailed values are provided in Table 6.

5.Results and Discussion

The co-cracking ofethane and propane,and LPG and naphtha mixtures in the USC-M and USC-U furnaces was studied,respectively.The effects of the mixing ratio,the coiloutlet temperature(COT)and the coiloutletpressure(COP)were discussed on the yield distributions of methane,ethylene and propylene,and some indices were also taken into consideration,e.g.the summarized yields of ethylene and propylene,the totalyields of the high value-added products(H2,C2H4,C3H6,C4H6,C6H6)which determined the pro fitability of the industrialfurnace.

5.1.Effectofmixing ratio oftwo differentfeedstocks

Fig.7.COT effect on ethylene and propylene yields(LPG/NAP).

Fig.8.COT effect on the yield of P&E and high value-added products(LPG/NAP).

In the cracking process,ethane tends to be transformed to ethylene and propane tends to be transformed to propylene,and this corresponds to Fig.1.It can be seen that ethylene yield increases with increasing proportion of ethane,and the growth rate tends to be gentle.Methane yield and propylene yield decrease,and propylene yield loss becomes slower gradually.Fig.2 indicates the summarized yields of ethylene and propylene and the total yields of high value-added products.It can be seen that a minimal yield exists and the maximum point can be obtained when the mass fraction of ethane reaches 1.0.Furthermore,the two curves change gently,and the gap between minimum and maximum is within 3%.Thus,the cocracking ofethane and propane mixtures has a negative effect on the pro fitability of the industrialfurnace.

企業人力資源信息化建設的過程中，最重要的就是選擇一套科學有效的軟件系統，目前，國內大多數企業實施人力資源信息化的過程中都借助軟件開發商進行系統開發工作。但是國內現有的軟件開發商只能算是單純的軟件開發商，并不是人力資源管理信息化軟件的開發商，很少有開發商能去研究人力資源相關知識或者雇傭專業人力資源人才參與軟件開發，能真正去研究如何系統的將人力資源管理工作和信息化有機結合起來，真正為企業人力資源管理工作服務的就更少了。

The productyield pro files with the mass fraction ofLPG are shown in Figs.3 and 4.Considering the actual production process,the mass fraction of LPG is de fined between 0 and 0.5.As can be seen in Fig.3,the ethylene and methane yields increase with increasing proportion of LPG,while the propylene yield decreases.This can be explained by the easy cracking ofLPG.LPGis lighter than naphtha and can be cracked easily to generate more smallmolecules which can promote the cracking ofnaphtha to obtain a higher ethylene yield.The gap between the minimum and maximum of ethylene yield can even reach 7%.Fig.4 illustrates that the summarized yields ofethylene and propylene and the totalyields of the high value-added products increase gradually,especially for ethylene and propylene,and the totalyields increase by 5%.Consequently,LPG can have a positive effecton the cracking ofnaphtha and improve the pro fitability of the industrialfurnace.

5.2.Effect of the coiloutlet temperature

Fig.9.COP effect on the yield ofethylene and propylene yields(C2/C3).

Fig.10.COP effect on the yield of P&E and high value-added products(C2/C3).

According to the industrialdata,the coiloutlet temperature varies from 830 °C to 870 °C for the co-cracking of ethane and propane mixtures.Figs.5 and 6 show the product yield pro files with the coiloutlet temperature.In Fig.5,the ethylene yields increase with increasing coil outlet temperature for different mass fractions of ethane.As the steam cracking is an endothermic reaction,a higher temperature can promote the cracking reaction and lead to a higher ethylene selectivity.However,the propylene yields decrease when the massfraction ofethane is below 0.8.Moreover,the downtrend becomes more and more obvious with increasing mass fraction ofpropane.This can be explained by a high propylene concentration.When the mass fraction ofpropane is high,more propylene can be generated and a high temperature can also accelerate the decomposition of propylene.However,the propylene yields increase slowly when the mass fraction of ethane exceeds 0.8 since less propane generates less propylene,leading to a low propylene concentration.Also,a low propylene concentration willreduce the decomposition ofpropylene.

From Fig.6 the optimum summarized yields ofethane and propane exist within the experimentaltemperature when the mass fraction of ethane is within 0.2.The smaller mass fraction of ethane,the lower coiloutlet temperature of the optimum yields.The same goes for high value-added products but the two indices always increase when the proportion of ethane exceeds 0.2.

Figs.7 and 8 give an overview of the product yield pro file with the coiloutlet temperature for the co-cracking of LPG and naphtha mixtures.As the same as the co-cracking of ethane and propane mixtures,the ethylene yields increase and the propylene yields decrease with increasing temperature for different mass fractions of LPG.In addition,the optimum yields can be obtained within the experimentaltemperature for the summarized yields ofethylene and propylene and the total yields of high value-added products.It can also be seen that the coil outlet temperature at the point of the optimum yields rises with increasing mass fraction of LPG.

Fig.11.COP effect on ethylene and propylene yields(LPG/NAP).

Fig.12.COP effect on the yield of P&E and high value-added products(LPG/NAP).

5.3.Effectof the coiloutlet pressure

In Fig.9,ethylene yield increases firstly and then decreases with increasing coiloutpressure when the massfraction ofethane is small,and an optimum yield exists.However,ethylene yield decreases when the mass fraction ofethane exceeds 0.5,and the change is small.For propylene,the yield decreases with the coiloutlettemperature when the mass fraction of ethane is small,and the yield increases with the coiloutlet temperature when the mass fraction ofethane becomes larger.Overall,the variation trend is slow and the coiloutlet pressure has a little in fluence on ethylene and propylene yields.

In Fig.10,the summarized yieldsofethylene and propylene decrease with the coiloutlet pressure for different mass fractions of ethane because ofa rising number of molecules in the cracking reaction.Corresponding to Fig.2,the summarized yields of ethylene and propylene decrease firstly and then increase with increasing mass fraction of ethane.For high value-added products,the totalyields decrease when the mass fraction ofethane is zero or exceeds 0.8,and when the mass fraction of ethane is between 0 and 0.8,an optimum value can be obtained.

Figs.11 and 12 illustrate the product yield pro file with the coilout pressure for the co-cracking of LPG and naphtha mixtures.As can been seen,the yields alldecrease with the coiloutlet pressure increasing for ethylene,propylene,P&E and high value-added products,and a high pressure can restrain the cracking reaction However,the change of ethylene yield is small,while the yields of other products decrease dramatically.

5.4.Interaction effectofmixing ratio and coiloutlet temperature

Figs.13 and 14 give an overview of the interaction effect of the mixing ratio and the coiloutlet temperature on the yields of P&E and high value-added products.As can be seen in Fig.13,the biggest pro fitability is obtained for the co-cracking of ethane and propane mixtures when COT is at point of 870°C and the mass fraction of ethane is 1.From Fig.14,an optimum pro fitability exists within the experimental temperature when the mass fraction of LPG is the highest.Thus,the co-cracking ofethane and propane mixtures has a negative effecton the pro fitability of the steam pyrolysis furnace,however,LPG can improve the pro fitability of the cracking ofnaphtha.

Fig.13.Effect of mass fraction(C2H6)and COT on the yields of P&E,high value-added products(C2/C3).

Fig.14.Effect of mass fraction(LPG)and COT on the yields of P&E,high value-added products(LPG/NAP).

5.5.Interaction effectofmass ratio and coiloutletpressure

The interaction effectof the mixing ratio and the coiloutletpressure on the yields of P&E and high value-added products are shown in Figs.15 and 16.As can be seen,a high coil outlet pressure leads to a low pro fitability.For the co-cracking ofethane and propane mixtures,the mimum yield exists with increasing mass fraction ofethane corresponding to Fig.2.Consequently,the results can be summarized that the co-cracking of ethane and propane mixtures has a negative effect on the pro fitability of the industrialfurnace,however,LPG can promote the pro fitability of the cracking ofnaphtha.

6.Conclusions

·The co-cracking of ethane and propane mixtures has a negative effect on the pro fitability of the industrial furnace.On the contrary,LPG can promote the generation ofethylene and decrease the propylene yield.In general,LPG has a positive effect on the pro fitability of the industrial furnace.

·In the co-cracking process of ethane and propane mixtures,the ethylene yield increases with increasing coiloutlet temperature,while the propylene yield decreases when the mass fraction ofethane is below 0.8 and the propylene yields increase slowly when the mass fraction of ethane exceeds 0.8.The optimum summarized yields of ethane and propane can be obtained within the experimental temperature when the mass fraction of ethane is within 0.2.The smaller mass fraction ofethane,the lower coiloutlet temperature of the optimum yields,and the same goes for high valueadded products.For the co-cracking of LPG and naphtha mixtures,ethylene yield increases and propylene yield decreases with increasing coiloutlet temperature,and an optimum yield exists within the experimentaltemperature for P&E and high value-added products.

Fig.15.Effect ofmass fraction(C2H6)and COP on the yields of P&E,high value-added products(C2/C3).

Fig.16.Effect of mass fraction(LPG)and COP on the yields of P&E,high value-added products(LPG/NAP).

·The most product yields decrease lightly with increasing coil outlet pressure in the co-cracking process ofethane and propane mixtures,and an optimum value exists for high value-added products when the mass fraction ofpropane is big enough.For the co-cracking of LPG and naphtha mixtures,the product yields alldecrease with increasing coil outlet pressure,and the effect of the coiloutlet pressure is apparent.

Nomenclature

[1]K.Barazandeh,O.Dehghani,M.Hamidi,E.Aryafard,M.R.Rahimpour,Investigation of coiloutlet temperature effect on the performance of naphtha cracking furnace,Chem.Eng.Res.Des.94(2014)307–316.

[2]N.Rahimi,R.Karimzadeh,Catalytic cracking of hydrocarbons over modi fied ZSM-5 zeolites to prouce light ole fins:a review,Appl.Catal.A Gen.398(1–2)(2011)1–17.

[3]J.Knoll,U.Singh,J.Nicolich,R.Gonzalez,M.Ziebarth,C.Fougret,S.Brandt,Unit cell volume as a measure ofdealumination of ZSM-5 in fluid catalytic cracking catalyst,Ind.Eng.Chem.Res.53(42)(2014)16270–16274.

[4]T.Ren,M.Patel,K.Blok,Ole fins from convection and heavy feedstocks:energy use in steam cracking and alternative processes,Energy 31(4)(2006)425–451.

[5]J.J.Gao,J.Du,Optimizing cracked stock,enhancing ethylene yield,Pet.Petrochem.7(2010)31–34.

[6]S.Ghorai,K.D.P.Nigam,CFD modeling of flow pro files and interfacialphenomena in two-phase flow in pipes,Chem.Eng.Process.45(1)(2006)55–65.

[7]S.C.K.De Schepper,G.J.Heynderickx,G.B.Marin,CFD modeling ofallgas–liquid and vapor–liquid flow regimes predicted by the Baker chart,Chem.Eng.J.138(1–3)(2008)349–357.

[8]S.C.K.De Schepper,G.J.Heynderickx,G.B.Marin,Modeling the evaporation of a hydrocarbon feedstock in the convection section of a steam cracker,Comput.Chem.Eng.33(1)(2009)122–132.

[9]S.C.K.De Schepper,G.J.Heynderickx,G.B.Marin,Coupled simulation of the flue gas and process gas side of a steam cracker convection,AIChE J.55(11)(2009)2773–2787.

[10]B.Basu,D.Kunzru,Catalyst pyrolysis of naphtha,Ind.Eng.Chem.Res.31(1992)146–155.

[11]P.S.Van Damme,G.F.Froment,W.B.Balthasar,Scaling up of naphtha cracking coils,Ind.Eng.Chem.Process.Des.Dev.20(2)(1981)366–376.

[12]K.M.Van Geem,G.J.Heynderickx,G.B.Marin,Effect ofradialtemperature pro files on yields in steam cracking,AIChE J.50(2004)173–183.

[13]M.W.M.van Goethem,S.Barendregt,J.Grievink,P.J.T.Verheijen,M.Dente,E.Ranzi,A kinetic modeling study of ethane cracking for optimalethylene yield,Chem.Eng.Res.Des.91(6)(2013)1106–1110.

[14]P.Kumar,D.Kunzru,Modeling ofnaphtha pyrolysis,Ind.Eng.Chem.Process.Des.Dev.24(3)(1985)774–782.

[15]Z.Geng,Y.Cui,L.Xia,Q.Zhu,G.Xiangbai,Compromising adjustmentsolution ofprimary reaction coef ficients in ethylene cracking furnace modeling,Chem.Eng.Sci.80(2012)16–29.

[16]S.Seifzadeh Haghighi,M.R.Rahimpour,S.Raeissi,O.Dehghani,Investigation ofethylene production in naphtha thermalcracking plant in presence of steam and carbon dioxide,Chem.Eng.J.228(15)(2013)1158–1167.

[17]N.Zhang,T.Qiu,B.Chen,CFD simulation of propane cracking tube detailed radical kinetic mechanism,Chin.J.Chem.Eng.21(12)(2013)1319–1331.

[18]G.Hu,H.Wang,F.Qian,Numericalsimulation on flow,combustion and heattransfer of ethylene cracking furnaces,Chem.Eng.Sci.66(8)(2011)1600–1611.

[19]G.Hu,H.Wang,F.Qian,Y.Zhang,J.Li,Comprehensive CFD simulation of product yields and coking rates for a floor-and wall-fired naphtha cracking furnace,Ind.Eng.Chem.Res.50(24)(2011)13672–13685.

[20]M.E.Masoumi,S.M.Sadrameli,J.Tow fighi,A.Niaei,Simulation,optimization and controlof a thermalcracking furnace,Energy 31(4)(2006)516–527.

[21]M.Berreni,M.Wang,Modelling and dynamic optimization of thermal cracking of propane for ethylene manufacturing,Comput.Chem.Eng.35(12)(2011)2876–2885.

[22]S.Zheng,X.Zhang,C.Qi,H.Zhou,Modeling ofheat transfer and pyrolysis reactions in ethylene cracking furnace based on 3-D combustion monitoring,Int.J.Therm.Sci.94(2015)28–36.

[23]L.Li,Co-pyrolysis and separate pyrolysis of hydrocarbon,Chem.Eng.He Nan 1(1990)15–16.

[24]K.M.Van Geem,D.Hudebine,M.F.Reynier,F.Wahl,J.J.Verstraete,G.B.Marin,Molecular reconstruction of naphtha steam cracking feestocks based on commercialindices,Comput.Chem.Eng.31(9)(2007)1020–1034.

[25]K.M.Van Geem,R.Zajdlik,M.F.Reynier,G.B.Marin,Dimensionalanalysis for scaling up and down steam cracking coils,Chem.Eng.J.134(2007)3–10.

[26]G.Hu,H.Wang,F.Qian,K.M.Van Geem,C.M.Schietekat,G.B.Marin,Couple simulation ofan industrialnaphtha cracking furnace equipped with long-flame and radiation burners,Comput.Chem.Eng.38(5)(2012)24–34.

[27]M.T.Klein,G.Hou,R.J.Bertolacini,L.K.Broadbelt,A.Kumar,Molecular modeling in heavy hydrocarbon conversions,Fuel86(12–13)(2007)2023.

[28]J.J.Verstraete,N.Revellin,H.Dulot,D.Hudebine,Molecular reconstruction of vacuum gasoils,Abstr.Pap.Am.Chem.Soc.227(2004)(010-FUEL).

[29]S.P.Pyl,K.M.Van Geem,M.F.Reyniers,G.B.Marin,Molecular reconstruction of complex hydrocarbon mixtures:an application of principal component analysis,AIChE J.56(12)(2010)3174–3188.

[30]R.C.Reid,J.M.Prausnitz,B.R.Poling,Properties of Gases and Liquids,McGraw-Hill,1979.

[31]D.B.Meade,B.S.Haran,R.E.White,The shooting technique for solution oftwo-point boundary value problems,Mapletech 3(1996)85–93.