Xu Hou,Bochong Chen,Zhenzhou Ma,Jintao Zhang,Yuanhang Ning,Donghe Zhang,Liu Zhao,Enxian Yuan,Tingting Cui

1 School of Chemical Engineering,Changchun University of Technology,Changchun 130012,China

2 Advanced Institute of Materials Science,Changchun University of Technology,Changchun 130012,China

3 School of Chemistry and Chemical Engineering,Yangzhou University,Yangzhou 225009,China

4 Department of Chemistry,Tsinghua University,Beijing 100084,China

Keywords:Empirical model Normal-alkane Cyclo-alkane Pyrolysis Light olefins

ABSTRACT Due to the complexity of feedstock,it is challenging to build a general model for light olefins production.This work was intended to simulate the formation of ethylene,propene and 1,3-butadiene in alkanes pyrolysis by referring the effects of normal/cyclo-structures.First,the pyrolysis of n-pentane, n-hexane,n-heptane, n-octane, n-nonane, n-decane,cyclohexane,methylcyclohexane, n-hexane and cyclohexane mixtures,and n-heptane and methylcyclohexane mixtures were carried out at 650-800°C,and a particular attention was paid to the measurement of ethylene,propene and 1,3-butadiene.Then,pseudo-first order kinetics was taken to characterize the pyrolysis process,and the effects of feedstock composition were studied.It was found that chain length and cyclo-alkane content can be qualitatively and quantitively represented by carbon atom number and pseudo-cyclohexane content,which made a significant difference on light olefins formation.Furthermore,the inverse proportional/quadratic function,linear function and exponential function were proposed to simulate the effects of chain length,cycloalkane content and reaction temperature on light olefins formation,respectively.Although the obtained empirical model well reproduced feedstock conversion,ethylene yield and propene yield in normal/cycloalkanes pyrolysis,it exhibited limitations in simulating 1,3-butadiene formation.Finally,the accuracy and flexibility of the present model was validated by predicting light olefins formation in the pyrolysis of multiple hydrocarbon mixtures.The prediction data well agreed with the experiment data for feedstock conversion,ethylene yield and propene yield,and overall characterized the changing trend of 1,3-butadiene yield along with reaction temperature,indicating that the present model could basically reflect light olefins production in the pyrolysis process even for complex feedstock.

1.Introduction

Entered into 21st century,the continuous progress of living standards puts a growing demand for light olefins,especially ethylene,propene and 1,3-butadiene [1,2].Hydrocarbons pyrolysis is one traditional and leading source for light olefins,which is carried out at 750-900 °C.The detailed information about hydrocarbons behavior at high temperatures helps to improve light olefins production[3-8].It is well known that hydrocarbons pyrolysis follows the free radical mechanism.Feedstock composition and operating condition make a significant difference on the pyrolysis activity and product distribution [9,10].Due to the complexity of feedstock composition and operating condition,the accurate details about hydrocarbons pyrolysis process remain challenging.Normal/cyclo-alkanes are the important and representative compositions of petrochemical hydrocarbons,and always selected as model reactants in the researches about the pyrolysis process.The related modeling studies are considered as the fundamental and necessary ways to reveal the nature of the pyrolysis process to produce light olefins.

According to the theoretical calculation,Zhouet al.[11]developed a mechanistic model to predict ethylene and propene formation inn-pentane pyrolysis at 650-875 °C.Pantet al.[12]developed an empirical model including a primary reaction and 24 secondary reactions forn-heptane pyrolysis,and it well simulated ethylene and propene formation at 680-750 °C.Zámostny′et al.[13] developed a mechanistic model forn-heptane pyrolysis using the automatic reaction network generator.The simulation results overall agreed with the experimental data for ethylene,propene and butadienes.Zeppieriet al.[14] developed a mechanistic model forn-decane pyrolysis by extending and improvingn-heptane model.Although the simulation results overall agreed with the experiment results,limitations were observed with regard to propene and 1,3-butadiene.Jiaet al.[15] developed a mechanistic model including 842 reactions of 164 species for supercriticalndecane pyrolysis,and it satisfactorily simulated the formation of ethylene and propene at 500-670 °C.Zhuet al.[16] developed an empirical model including one global reaction of 18 species for supercriticaln-decane pyrolysis with the conversion less than 13%.The deviation between the prediction and experiment data forn-decane at the reactor outlet was within 3.9%.Zhanget al.[17] developed an empirical model for supercriticaln-dodecane pyrolysis at 450-730°C.The relative errors between the prediction and experiment data for conversion were within 10%,and the formation of ethylene and propene was well reproduced.Zenget al.[18]developed studied a mechanistic model forn-dodecane pyrolysis at 477-1157°C under various pressures.It captured the characteristics of ethylene,propene and 1,3-butadiene formation in both their and literature work aboutn-dodecane pyrolysis.Wanget al.[19] developed a mechanistic model including 557 reactions of 148 species for cyclohexane pyrolysis.It well reproduced the formation of ethylene,propene and 1,3-butadiene at 677-1247 °C.Khandavilliet al.[20] developed a mechanistic model including 806 reactions of 241 species for cyclohexane pyrolysis at 640-800 °C.The calculation data well agreed with the experiment data for ethylene,propene and 1,3-butadiene.Compared with five popular models,this model was more accurate and transferable for the literature work about cyclohexane pyrolysis.Wanget al.[21,22] developed a mechanistic model including 1570 reactions of 249 species for methylcyclohexane combustion,and the simulation results well agreed with the experiment data for ethylene,propene and 1,3-butadiene.

As the research moved along,the pyrolysis of more and more alkanes was studied by the mechanistic or empirical model.A general and simple modeling methodology was extremely expected for the pyrolysis process to produced light olefins.Renet al.[23]proposed a step-by-step construction methodology via the automatic reaction network generator to establish naphtha pyrolysis model.It started with a single hydrocarbon and then merged all the considered reaction networks.The final model including 82,130 reactions of 1947 species well reproduced the formation of ethylene and propene inn-decane pyrolysis at 510-850 °C.Moreover,it predicted the yields of ethylene and propene in naphtha steam cracking at 716-828 °C,and the deviations were within acceptable ranges.Although the model was huge and complicated,it encouraged the researchers by confirming the feasibility of a general model for hydrocarbons pyrolysis.The insights into the effects of feedstock are crucial to develop and simplify the mathematical model for the pyrolysis process.Zámostny′et al.[24]studied the effects of molecular structures on hydrocarbons pyrolysis at 810 °C.It was found that pyrolysis products exhibited general trends with normal,branched,cyclic,aromatic and unsaturated structures.Thus,it may be possible to simplify the mathematical model by incorporating the function descriptor of molecular structures.

This work was focused at the modeling studies about light olefins formation in normal/cyclo-alkanes pyrolysis.The pyrolysis tests were carried out in a micro fixed-bed reactor at 650-800 °C,and the cracked gas was analyzed by an online gas chromatograph.n-Pentane,n-hexane,n-heptane,n-octane,n-nonane,n-decane,cyclohexane and methylcyclohexane were chosen as model reactants;binary mixtures of normal-alkane and cycloalkane (n-hexane and cyclohexane mixtures,andn-heptane and methylcyclohexane mixtures)were prepared and chosen as model reactants.Pseudo-first order kinetics was taken to characterize the pyrolysis process,and an empirical model was developed to simulate ethylene,propene and 1,3-butadiene formation.Furthermore,the accuracy and flexibility of the present model was validated by predicting light olefins formation in the pyrolysis of multiple alkane mixtures.This work may provide important information about hydrocarbons behaviors at high temperatures,which was beneficial to develop the general and simple modeling methodology for the pyrolysis process to produce light olefins.

2.Experimental

2.1.Reactants

Fig.1.Schematic diagram of the micro fixed-bed test bench.

n-Pentane (n-C5),n-hexane (n-C6),n-heptane (n-C7),n-octane(n-C8),n-nonane (n-C9),n-decane (n-C10),cyclohexane (c-C6) and methylcyclohexane (mc-C6) were purchased from Rhawn Reagent Company (Shanghai,China).The purities were higher than 99%(mass),and were used without further purification.Six kinds ofn-hexane and cyclohexane mixtures,five kinds ofn-heptane and methylcyclohexane mixtures,and four kinds of multiple hydrocarbon mixtures were prepared by blending pure alkanes.High purity nitrogen (N2,＞99.999%(vol)) was used as carrier gas.

2.2.Experiment apparatus

The schematic diagram of the micro fixed-bed test bench was presented in Fig.1.Reactant and N2were delivered by the syringe pump(Legato 100,KD Scientific)and mass flowmeter,respectively.They were preheated and uniformly mixed in the preheater,and then entered the reactor tube,which was represented by a 304 stainless steel tube (inner diameter:4 mm;outer diameter:6 mm;heating length:300 mm).A two-zone radiation furnace was employed for heating and controlling the temperature of reactor tube.The cracked gas went through the thermostatic heater,and was injected into an online gas chromatograph (7890B,Agilent).This gas chromatograph was equipped with two separated channels and flame ionization detectors:HP-AL/S capillary column(30 m × 0.53 mm × 15 μm) was used to analyze C1-C5 species;HP-PONA capillary column (50 m × 0.2 mm × 0.5 μm) was used to analyze C6-C10 species.

2.3.Experiment procedure

The pyrolysis experiments were carried out on the micro fixedbed test bench under atmospheric pressure.As a typical case,N2flow was set as 100 ml·min-1;the preheater,radiation furnace and thermostatic heater were set as 300 °C,300 °C and 200 °C,respectively.At the same time,the gas chromatograph was turned on.This condition was maintained for 1 h to clean the reaction system,and stabilize the analytical instrument.Then,reactant was introduced at 1.35 g·h-1,and the reactor tube was heated to and maintained at the target temperature.When it was 30 min after reaching the target temperature,the cracked gas was sampled and analyzed by the online gas chromatograph.Feedstock conversion and product distribution were calculated based on the chromatogram of cracked gas.Feedstock conversion (X,%),mass selectivity to producti(Si,%(mass))and mass yield of producti(Yi,%(mass)) were defined and calculated as Eqs.(1)-(3),respectively.

Fig.3.The parameters for the inverse proportional and quadratic functions.

Fig.4.Comparison between the experiment and calculation data for normal-alkanes pyrolysis.(a) Feedstock conversion.(b) Ethylene yield.(c) Propene yield.(d) 1,3-Butadiene yield.

whereAiwas the calibrated peak area for producti,andARwas the calibrated peak area for reactant in the chromatogram of cracked gas.

3.Results and Discussion

In order to simulate light olefins production,the present work was focused on the effects of chain length on normal-alkanes pyrolysis and the effects of cyclo-alkane content on the pyrolysis of normal-alkane and cyclo-alkane mixtures.A particular attention was paid to the analysis of ethylene,propene,and 1,3-butadiene formation,and pseudo-first order kinetics was taken to characterize the pyrolysis process.According to the experiment results,an empirical model including several composite functions was proposed,and continuously improved as the research moved along.The accuracy and flexibility of the present model were validated by predicting light olefins formation in the pyrolysis of multiple hydrocarbon mixtures.

3.1.Normal-alkanes pyrolysis

n-Pentane,n-hexane,n-heptane,n-octane,n-nonane andn-decane were employed to explore the effects of chain length on normal-alkanes pyrolysis.The chain length was represented by the number of carbon atom,and in the range of 5-10.The kinetics analysis was carried out to simulate the effects of chain length on normal-alkanes decomposition and light olefins formation.According to the high-temperature and low-pressure conditions,thepseudo-first order kinetics was taken to characterize the pyrolysis process,and the decomposition rate of feedstock (RD) was expressed as Eq.(4).

The rate constant of feedstock decomposition (kD) was expressed as Eqs.(5) and (6)

The formation rate of producti(RFi) was expressed as Eq.(7).

The rate constant of productiformation (kFi) was expressed as Eqs.(8) and (9).

whereCwas the reactant concentration,twas the residence time,andC0was the initial reactant concentration.

According to the experiment data,the rate constants of feedstock decomposition and product formation were calculated and presented as a function of chain length in Fig.2.Rate constants exhibited general trends with chain length at different reaction temperatures.An increase of chain length first promoted then hardly changed the rate constants of feedstock decomposition and ethylene formation,while first decreased then increased the rate constants of propene and 1,3-butadiene formation.Zenget al.[18] studied the effects of chain length on the reactivity and product distribution in the pyrolysis ofn-decane,n-dodecane andn-tetradecane at 627-1077 °C.It was found that an increase of chain length promoted the pyrolysis reactivity and the formation of ethylene and 1,3-butadiene,which was consistent with the experiment results in the present work.It was deduced that the chain length of normal-alkanes was one important factor to pyrolysis activity and light olefins formation,which can be qualitatively and quantitively characterized by the number of carbon atom.

Fig.2.Rate constants of (a) feedstock decomposition,(b) ethylene formation,(c) propene formation,and (d) 1,3-butadiene formation in normal-alkanes pyrolysis.

Based on the above results,the effects of chain length on feedstock decomposition and ethylene formation were simulated by the inverse proportional function (Eqs.(10) and (11)),and those on propene and 1,3-butadiene formation were simulated by the quadratic function (Eqs.(12) and (13)).

wherenwas the chain length and represented by the number of carbon atom.

Fig.5.Prediction data by the present model and experiment data for (a) Mixture I,(b) Mixture II and (c) Mixture III pyrolysis.

As shown in Fig.2,the proposed functions well fitted with the experiment data.The proposed functions included ten parameters at each reaction temperature,and they were summarized in Table S1 (see Supplementary Material).Inspired by Arrhenius equation,the effects of reaction temperature on the parameters in the inverse proportional and quadratic functions were simulated by the exponential function (Eq.(14)).

Table 1Compositions of the prepared multiple hydrocarbon mixtures

whereTwas the reaction temperature.

As shown in Fig.3,the exponential function well fitted with the parameters for the inverse proportional and quadratic functions.The parameters for the exponential function were summarized in Table S2.

According to the achieved equations and parameters,an empirical model was developed to simulate the effects of reaction temperature and chain length on light olefins formation in normalalkanes pyrolysis,and it was expressed as Eqs.(15)-(18).

The experiment data for feedstock conversion,ethylene yield,propene yield and 1,3-butadiene yield were reproduced by this model.As shown in Fig.4,the experiment data was plotted against the calculation data to make the deviation clear.The solid-45°line(y=x) represented the ideal fitting (deviation=0),and the dash lines indicated the boundaries of±2.5%errors for feedstock conversion (Fig.4a) and ±1.5%(mass) errors for product yield (Fig.4b-d).The data points evenly dispersed in or around the error threshold,indicating the good agreement between the calculation and experiment data.

Fig.6.Rate constants of (a) feedstock decomposition,(b) ethylene formation,(c) propene formation,and (d) 1,3-butadiene formation in the pyrolysis of n-hexane and cyclohexane mixtures.

Fig.7.Rate constants of (a) feedstock decomposition,(b) ethylene formation,(c) propene formation,and (d) 1,3-butadiene formation in the pyrolysis of n-heptane and methylcyclohexane mixtures.

The accuracy and flexibility of this model was further validated by simulating light olefins formation in the pyrolysis of normalalkane mixtures.As shown in Table 1,three kinds of normalalkane mixtures(Mixture I,II and III)were prepared and employed as model reactants.As shown in Fig.5,the prediction data well agreed with the experiment data,confirming the good accuracy and flexibility of the present model for simulating light olefins production in normal-alkanes pyrolysis even for normal-alkane mixtures.

3.2.Pyrolysis of normal-alkane and cyclo-alkane mixtures

n-Hexane and cyclohexane mixtures as well asn-heptane and methylcyclohexane mixtures were employed to explore the effects of cyclo-alkane content on the pyrolysis of normal-alkane and cyclo-alkane mixtures.The rate constants of feedstock decomposition () and product formation () in the pyrolysis of normalalkane and cyclo-alkane mixtures were calculated by Eqs.(4)-(9).As shown in Figs.6 and 7,the rates of feedstock decomposition,ethylene formation and propene formation decreased while those for 1,3-butadiene formation increased with the elevated cycloalkane content at all reaction temperatures.That was the presence of cyclo-alkanes inhibited the formation of ethylene and propene while promoted the formation of 1,3-butadiene.Shenet al.[25]studied the pyrolysis ofn-octane and ethylcyclohexane at 600-900 °C.It was found that ethylcyclohexane exhibited a lower conversion and yield of ethylene and propene,while a higher yield of 1,3-butadiene compared ton-octane.It was consistent with the experiment results in the present work.It was deduced that cycloalkane content was also one important factor to pyrolysis activity and light olefins formation.

Effects of cyclo-alkane including cyclohexane and methylcyclohexane were represented by the factor of ring content,i.e.the content of pseudo-cyclohexane in alkane mixtures.The relative rates were obtained by comparing the rate constants of mixtures pyrolysis(and)to those of normal-alkane pyrolysis(kDandkFi).As shown in Fig.8,the relative rates exhibited general trends along with ring content at all reaction temperatures.An increase of ring content reduced the relative rates of feedstock decomposition,ethylene formation and propene formation,while promoted that for 1,3-butadiene formation.Based on this phenomenon,the effects of ring content on the relative rate were simulated by the linear function (Eqs.(19)-(22)).

wherexwas the ring content and represented by pseudocyclohexane content in the mixtures.

Fig.8.Relative rates of (a) feedstock decomposition,(b) ethylene formation,(c) propene formation,and (d) 1,3-butadiene formation in the pyrolysis of normal-alkane and cyclo-alkane mixtures.

As shown in Fig.8,the linear functions reproduced the overall evolution of the relative rates along with ring content in the mixtures.

Fig.9.The parameters for the linear functions.

The linear function included four parameters at each reaction temperature,and they were summarized in Table S1.Also,the effects of reaction temperature on the parameters were simulated by the exponential function(Eq.(14)).As shown in Fig.9,the exponential function well fitted with the parameters for the linear function.The parameters for the exponential function were summarized in Table S2.

The empirical model was improved to simultaneously simulate the effects of reaction temperature,chain length and feedstock composition on light olefins production in hydrocarbons pyrolysis,and it was expressed as Eqs.(23)-(26).

Fig.10.Comparison between the experiment and calculation data for the pyrolysis of normal-alkane and cyclo-alkane mixtures.(a)Feedstock conversions.(b)Ethylene yield.(c) Propene yield.(d) 1,3-Butadiene yield.

The experiment data for the pyrolysis of normal-alkane and cyclo-alkane mixtures was reproduced by the present model.As shown in Fig.10a-c,the data points evenly dispersed in or around the error threshold,indicating the good agreement between the calculation and experiment data for feedstock conversion,ethylene yield and propene yield.As shown in Fig.10d,certain data points significantly deviated from the error threshold,indicating the deviations between the calculation and experiment data for 1,3-butadiene yield.

Fig.11.Prediction data by the present model and experiment data for Mixture IV pyrolysis.

The accuracy and flexibility of the present model was further validated by simulating light olefins production in the pyrolysis of multiple hydrocarbon mixtures.As shown in Table 1,Mixture IV including four kinds of normal-alkanes and two kinds of cyclo-alkanes were prepared,and employed as model reactants.As shown in Fig.11,the prediction data well agreed with the experiment data,confirming the good accuracy for simulating feedstock conversion,ethylene yield and propene yield.Although the present model had defects in calculating 1,3-butadiene yield,it was capable of predicting the general trend of 1,3-butadiene yield along with reaction temperature.Thus,it was addressed that the present model including four composite functions could basically reflect light olefins production in hydrocarbons pyrolysis even for complex composition.

4.Conclusions

Briefly,the pseudo-first order kinetics was taken to characterize the pyrolysis process.The inverse proportional/quadratic function,linear function,and exponential function were proposed to simulate effects of chain length,cycloalkane content,and reaction temperature.Although the empirical model including four composite functions well reproduced feedstock conversion,ethylene yield and propene yield in normal/cyclo-alkane pyrolysis,it exhibited limitations in simulating 1,3-butadiene formation.In addition,the present model successfully predicted light olefins production in the pyrolysis of multiple hydrocarbon mixtures.The prediction data well agreed with the experiment data for feedstock conversion,ethylene yield and propene yield,and overall characterized the changing trend of 1,3-butadiene yield along with reaction temperature,indicating that the present model could basically reflect light olefins production in hydrocarbons pyrolysis even for complex composition.Further refinement of the functions and parameters may lead to higher prediction accuracy of light olefins production and wider application in the pyrolysis process.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgements

The authors gratefully acknowledge for the financial support from the National Natural Science Foundation of China(21908010)and Jilin Provincial Department of science and technology (20200201095JC).

Supplementary Material

Supplementary data to this article can be found online at https://doi.org/10.1016/j.cjche.2021.03.037.