Mohsin Ali; Liao Zuwei Yao Yang; Sun Jingyuan; Jiang Binbo;Wang Jingdai Yang Yongrong

(1. Zhejiang Provincial Key Laboratory of Advanced Chemical Engineering Manufacture Technology,College of Chemical and Biological Engineering, Zhejiang University, Hangzhou 310027;2. State Key Laboratory of Chemical Engineering, College of Chemical and Biological Engineering, Zhejiang University, Hangzhou 310027)

Abstract: Ethane steam cracking process in an industrial reactor was investigated. An one-demsional (1D) steady-state model was developed firstly by using an improved molecular reaction scheme and was then simulated in Aspen Plus.A comparison of model results with industrial data and previously reported results showed that the model can predict the process kinetics more accurately. In addition, the validated model was used to study the effects of different process variables, including coil outlet temperature (COT), steam-to-ethane ratio and residence time on ethane conversion, ethylene selectivity, products yields, and coking rate. Finally, steady-state optimization was conducted to the operation of industrial reactor. The COT and steam-to-ethane ratio were taken as decision variables to maximize the annual operational pro fit.

Key words: ethane steam cracking, tubular reactor, Aspen Plus, molecular reaction scheme, process simulation, process optimization

1 Introduction

Ethylene is one of the abundantly used chemicals in petrochemical industry and is considered as the base component in many chemical processes because of its purity and low cost. Ethylene is commonly used to produce plastics, lubricants, fibers, and starting material in other applications. In the late 20thcentury, ethylene production had increased dramati cally from 29 billion lb to 180 billion lb[1]. Currently, the leading method of ethylene production is steam cracking of a different range of hydrocarbons, among which the commonly used feedstocks include naphtha, ethane, propane, butane,and gas oil[2-6]. Recently, the growth of shale gas has gained attention, as it contains a high ratio of ethane by speci fication which caused the signi ficant decrease in the price of ethane. Therefore, a large amount of cheap ethane produced by shale gas has resulted in the cheaper ethylene production by ethane steam cracking[7-8]. Furthermore, the ethane steam cracking is considered as highly selective towards ethylene production as only a small amount of other light ole fins are coproduced.

1.1 Literature review

Although various techniques have been applied to improve the system efficiency of the hydrocarbon processing units[9-12], the kinetic reaction scheme is the basic component of modeling and simulation as it describes the actual ethane cracking mechanism taking place in the reactor. The accuracy of predicted product distribution by simulation and modeling is highly dependent on reaction scheme used. While many modeling and simulation studies in this field are aimed to predict the distribution of products under a wide range of process conditions.Sundaram and Froment[13]successfully simulated a steam cracker to investigate the ethane cracking process by using molecular reaction mechanism. Belohlav, et al.[14]conducted a detailed study on modeling of ethane cracking plant, and the cracking model involved a set of radical and molecular reactions, with the results validated with industrial data under diverse reaction conditions. By assuming a detailed radical mechanism, Tarafder, et al.[15]conducted a study on modelling and multi-objective optimization of an industrial ethane reactor and a wide range of operating conditions were depicted for process optimization. Gujarathi, et al.[16]performed a simulation study on ethane cracking and predicted the results of products concentrations, temperature and pressure profiles were compared with industrial data. Ranjan,et al.[17]investigated the ethane cracking kinetics in an industrial reactor using the improved molecular reaction scheme and the effects of different reaction variables on product composition were studied. Yancheshmeh, et al.[18]modelled an ethane cracking reactor and investigated the effect of steam and CO2addition as diluents on coking rate and on hydrogen and ethylene production. Caballero,et al.[19]carried out simulation and optimization study of ethane cracking to predict product distribution, and the comparison of predicted results with industrial data showed that the used method was a good alternative to deal with optimization problems. Barza, et al.[20]conducted mathematical modeling of ethane cracking by using molecular reaction scheme with coke formation approach and proposed that the adopted modeling approach was more capable of predicting coke formation and product composition. In all above studies, different radical and molecular reaction schemes have been used and validated with industrial data. Ethane cracking actually proceeds by radical mechanism and it can provide a better understanding of reactions mechanism,and furthermore the radical reactions are more promising in describing the coke formation inside the reactor coils[13].Ethylene is believed to be the main precursor of coking by some secondary reactions[21]. However, for computational simplicity, molecular reaction schemes have been mostly used for modeling and simulation of ethane cracking.Sundaram and Froment[13]firstly proposed a molecular scheme which has been widely used in many previous studies but this molecular reaction scheme fails to consider some high molecular weight components of products such as C4H8and C6H6. Later on, a detailed molecular reaction scheme was proposed by Ranjan. et al.[17]with a total of 11 reactions and 10 components which seemed to be a better explanation of ethane cracking kinetics.

1.2 Motivation and scope of this study

Ranjan, et al.[17]proposed a detailed molecular reaction scheme with 11 reactions and conducted a simulation study of ethane cracking in a plug flow reactor (PFR) by using Aspen plus as the simulation software. But this study did not address the coke formation which took place inside the reactor.Moreover, the simulation results of product distribution still had margin to improve. In this study, a more detailed reaction scheme with 15 reactions has been proposed with the addition of 4 additional reactions to address the shortcomings of previous study. The highlighted contributions of this current study are: (1) development of more comprehensive ethane cracking model in Aspen Plus by using improved molecular reaction scheme which can predict the product distribution more precisely and also take into account the coke formation;(2) evaluation of the improved model with the effects of cracking severity factors such as COT, steam-to-ethane ratio and ethane flow rate or residence time on conversion of ethane, ethylene selectivity, products yields and coking rate;and (3) the operating pro fit of the process has been optimized by using COT and steam-to-ethane ratio as decision process variables in Aspen Plus. The remaining parts of the article are organized as follows: Section 2 presents model development,while sensitivity analysis of cracking reactor and process optimization are illustrated in Sections 3 and 4.

2 Model Development

2.1 Assumptions

To simplify the ethane cracking model, the following assumptions have been taken:

1) Only vapor and gas phases are considered inside PFR:there is no liquid.

2) Hydrodynamic parameters, and thermal entrance region effects are neglected. Moreover, bending effects of coils are not considered and the whole set of radiant coils is assumed as a single PFR.

3) All cracking reactions are assumed to start at the entrance of PFR and stop at the exit.

4) Water involved reactions are not considered.

5) Coke thickness is assumed to be constant within each hour and only coke buildup on the inner sides of coils is considered (heat and mass accumulation inside reactors are neglected).

6) Molecular reaction scheme is adopted and only molecular components are considered in final product.

2.2 Plug flow reactor (PFR) and operating conditions

Ethane cracking reaction regime is a set of various homogenous reactions taking place along the length of the radiant coils and this process can be well explained by using PFR model. Therefore, the built-in PFR model in Aspen Plus was used for process simulation. The current simulation is based on process conditions and reactor configurations of a real industrial reactor described by Ranjan, et al.[17]. The specifications of reactor included a total of 48 tubes with a tube length of 10.5 m and an inner tube diameter of 0.085 m. Preheated feed containing a mixture of 100% pure ethane and diluent steam entered into PFR at 670 °C and 2.2 bar.For base case simulation, the mass flow rate of ethane was taken as 30 t/h with a steam-to-ethane ratio of 0.3 (kg/kg).Ranjan, et al.[17]had tested different temperature profiles from literature for process simulation[1516], because industrial temperature pro file was unavailable. The results showed that the isothermal temperature pro file had given better results of product composition than two other nonlinear temperature profiles, therefore a steady-state model was developed with isothermal temperature profile. Furthermore, the coil outlet temperature (COT) is always considered as the most important reference variable to ensure the product composition and it can be manipulated accordingly by adjusting the heat flux provided by the furnace. For ease of comparison, the current study adopted the same isothermal condition along with the improved molecular reaction scheme.

2.3 Reaction scheme

Ethane cracking produces a different kind of components in product stream, which mainly includes H2, CH4, C2H2,C2H4, C2H6, C3H6, C3H8, C4H6, C4H8, and C6H6. The reaction scheme proposed by Sundaram and Froment[13]did not include the higher molecular weight components.Ranjan, et al.[17]proposed a detailed reaction scheme containing 11 reactions. This scheme had 5 original reactions proposed by Sundaram and Froment[13]with the addition of 6 other reactions. However, the reaction scheme proposed by Ranjan, et al.[17]did not account for the formation of coke inside the reactor and also had a margin of improvement to insure the better prediction of product composition. Hence, an improved molecular reaction scheme has been proposed in this work with the addition of 4 additional reactions in previous reaction scheme of Ranjan, et al[17]. The details of selected molecular reaction scheme are illustrated in Table 1, in which reactions (12) and (13) represent the formation of additional amount of methane and ethylene and reactions(14) and (15) explain the coke formation.

Table 1 Improved molecular reaction scheme used in this study for cracking of ethane

2.4 Base case simulation and results

The PFR model was selected in Aspen Plus for base case simulation. Simulation was carried out by using improved molecular reaction scheme with the following employed process conditions: ethane mass flow rate = 30 t/h, steam mass flow rate = 10 t/h, and isothermal condition (850 °C).In Table 2, the results of current base case simulation and results reported by Ranjan, et al.[17]are compared with the industrial data. The improved reaction scheme can predict the product composition more accurately than the previously reported reaction scheme. Reactions (12)and (13), in the improved reaction scheme has obviously increased the amount of methane along with more ethane conversion in final product composition.

Table 2 Comparison of simulations results with industrial data

3 Sensitivity Analysis of Cracking Reactor

Sensitivity analysis of cracking reactor was performed with the base case model and the improved reaction scheme was applied to investigate the effects of some key process variables. In industrial cracking process, the decision variables which can directly affect the ethane reactor performance are normally COT, the steam-toethane ratio, and the ethane flow rate. The value of each variable was varied in certain range while all others parameters were kept constant to analyze the impacts of changes on some calculated quantities which can directly re flect the reactor performance. These quantities included ethane conversion, ethylene selectivity, yields of product components, and coking rate. Coking rate prediction method proposed by Sundaram and Froment[22]was used to calculate the coking rate which can be descried as:

whereRcis coking rate [kg/ (m2s)],Acis pre-exponential factor for coking reactions [kg m3/ (kmol m2s)],Ecis activation energy for coking reactions (kJ/kmol), andCis the concentration (kmol/m3) of those components which are considered as coking precursors. The accumulative coking rate was calculated by using Equation (3) separately for both coking reactions (ethylene and propylene). Previously, different research studies explained the role of ethylene and propylene as coke precursors in cracking reactions but ethylene is more effective than propylene[21,23-24]. Therefore, coking due to ethylene and propylene was considered in this study to calculate the overall coking rate.

3.1 Effect of COT

COT is one of the most important variables which plays a significant role in predicting conversion, product composition, and coking rate. The effect of COT was investigated by conducting simulations in a certain range of reactor temperature ranging from 650 °C to 1 000 °C,while all other parameters were kept constant, same as in the base case. The sensitivity analysis showed that the increase in COT resulted in increase of ethane conversion and decrease in ethylene selectivity as presented in Figure 1. It can be observed in Figure 2 that the ethylene yield, which was promoted by high conversion and high selectivity, increased with a rising temperature as ethylene selectivity decreased at lower rate while ethane conversion increased sharply after 700 °C. The ethylene and hydrogen yields attained a maximum at 900 °C, with the observed yields reaching 63% and 4.7%, respectively.However, above 900 °C the yields of ethylene and hydrogen seemed to decrease and only methane yield increased, even though ethane conversion showed an increasing trend. This happened because some side reactions started at elevated temperature which involved ethylene and ethane as reactants and eventually could produce more methane and other by-products[25].

Figure 1 Effect of COT on ethane conversion and ethylene selectivityEthane conversion;Ethylene selectivity

Figure 2 Effect of COT on products yieldsMethane;Hydrogen;Ethylene

The coking rate should be increased with rising temperature according to the coking rate Equation (3).In Figure 3, the effect of temperature on coking rate has been studied, indicating that the coking rate increased dramatically with increased temperature. Especially at above 900 °C, the coking rate value seems to be very high. When the coking rate is higher, the coke build-up at inner side of coils will be faster, therefore the interval between two consecutive decoking operations will be lesser. Normally, the reactor decoking procedure needs 24—48 hours so this interruption causes a negative impact on overall production. Hence, it can be concluded from the results that when the reactor temperature reaches above 900 °C, the corresponding ethane conversion and coking rate values are not conducive to operating economics.

3.2 Effect of steam-to-ethane ratio

Figure 3 Effect of COT on coking rate

Ethane cracking into ethylene and hydrogen is considered as reversible reaction as shown in the reaction scheme.Therefore, steam is used as the diluent to reduce the partial pressure of ethane and to keep the reaction favorable toward ethylene production. Hence we assumed steam to be inert so its presence will not produce any byproduct. To analyze the role of steam-to-ethane ratio on reactor performance, simulation studies were conducted by changing the steam-to-ethane ratio in a specified range from 0 to 1 at a constant COT of 850 °C. It can be seen from Figure 4 that as the steam-to-ethane ratio increased, the ethane conversion decreased along with a slight increase in the selectivity of ethylene. The effect of steam-to-ethane ratio on yields of products is illustrated in Figure 5, it can be noticed that the yield of ethylene reached a maximum in a steam-to-ethane ratio range of 0.3 to 0.4 (kg/kg). Further increase in the steam-to-ethane ratio reduced the cracking performance of reactor because of drastic decrease in ethane conversion. On the other hand, the impact of steam-to-ethane ratio on coking rate is presented in Figure 6. As expected, the increase in steamto-ethane ratio had decreased the coking rate, and the coking rate value reduced from 0.0092 to 0.0042 [kg/(m2h)]as the steam-to-ethane ratio increased from 0 to 1 (kg/kg).Therefore, it can be concluded from above observations that it is better to operate the reactor at a lowest possible steam dilution with an acceptable coking rate.

3.3 Effect of ethane flow rate or residence time

Figure 4 Effect of steam-to-ethane ratio on ethane conversion and ethylene selectivityEthane conversion;Ethylene selectivity

Figure 5 Effect of steam-to-ethane ratio on products yieldsEthylene;Methane;Hydrogen

Figure 6 Effect of steam-to-ethane ratio on coking rate

Feedstock flow rate has a deep impact on residence time and thus on product composition. Therefore, to study the effect of ethane flow rate, the reactor length was changed in a range of 1 m to 15 m instead of changing the ethane flow rate as variation of both ethane flow rate or reactor length will only change the residence time of reactor.The motive of reactor length selection as a modifying variable was just to make the simulation comparatively simpler with less process manipulations. The reactor was operated at a fixed COT of 850 °C with a volumetric flow rate of 9.5 m3/s and a fixed inner coil diameter of 0.085 m,while the equivalent residence time was varied in a range of 0.02 s to 0.42 s. In Figure 7, a sharp increasing trend in ethane conversion and a slight decrease in ethylene selectivity were observed along the length of reactor and it might occur possibly due to increase in residence time.The effect of reactor length or residence time on products yields is presented in Figure 8, and it can be noticed that there is a very promising variation in ethylene yield at a low residence time of up to 0.3 s. But when the reactor length was above 11 m, at higher residence time there was no significant increase observed in ethylene or hydrogen yields, except for a slight increase in methane yield because of ethane conversion. However, the high residence time may not be desirable because it can initiate the secondary reactions which would ultimately decrease the yield of desired products that is why the residence time is usually preferred in a range of 0.1 s to 0.3 s[15,26]. The coking rate increased along the length of reactor as shown in Figure 9 and this trend was obvious,as the coking rate was primarily dependent on ethylene concentration and residence time. Therefore, coking rate is directly related to the desired yield of ethylene. Hence,running a reactor at a possibly shorter residence time is economically favorable with acceptable coking rate and maximum desired products yields.

Figure 7 Effect of reactor length on ethane conversion and ethylene selectivityEthane conversion;Ethylene selectivity

Figure 8 Effect of reactor length on products yields—Ethylene;MethaneHydrogen

Figure 9 Effect of reactor length on coking rate

4 Process optimization

In a cracking furnace, the best profitable operating conditions will be very complicated as it will involve the yields of all products, the operating temperature, the steam dilution and coking rate, etc. The simulated results in the previous sections have suggested that the major operating variables are likely to be COT and steamto-ethane ratio. Therefore, these two process variables were considered for optimization study and the range of constraints were set as shown below:

4.1 Methodology

In formulating the optimization case study, the following assumptions have been taken: (1) The separation costs of final product stream have not included; (2) Only ethylene is assumed to be sold; (3) The effect of coke thickness on heat transfer rate along the reactor tube is not considered.Annual operating pro fit was used as an objective function to maximize. The objective function was developed in this study based on previous work[27]and it was explained as the income from ethylene minus the different costs.These costs included: the ethane cost which was fed as feedstock, the steam cost, the radiant heat cost and the decoking cost. Steam price factor, decoking time required per cycle, coke density and annual production time were taken from previous study[27]. Decoking cost of a cracking furnace depends on furnace speci fications and operating conditions, therefore an approximate cost of decoking has been taken from the literature[28]. The spot prices of ethylene and ethane in September 2019 have been considered[29]. The shutdown of reactor was considered to happen when the coke thickness reached 0.006 m inside the coils. The objective function was calculated on an annual basis as follows:

whereF1denotes the ethylene flow rate;F0stands for the ethane flow rate;Fsrepresents the steam flow rate; andndis the decoking times per year. The meaning and values of all other parameters used in equations (4) to (6) are summarized in Table 3. Moreover, the coking frequency was calculated as follows:

wheretedenotes the time length between two decoking operations, andRcstands for the coking rate.

Table 3 Parameters for optimization

4.2 Optimization results and discussions

Steady state optimization was conducted in Aspen Plus by using COT and steam-to-ethane ratio as manipulated process variables. The results of optimization study are summarized in Table 4. Simulation results of base case are compared with the results obtained after optimization study, it can be observed that after optimization the annual operating pro fit has been improved. The coil outlet temperature increased from 850 °C to 880 °C, which has increased the yields of all products, and especially the ethylene yield increased from 52% to 63%. This factor has a major contribution to higher annual profit.On the other side, the rise in operating temperature also increased the coking rate therefore the expenses of decoking during reactor shutdown become higher. After optimization, the steam-to-ethane ratio also increased from 0.3 to 0.59 to deal with higher coking rate. However,the production time between two consecutive decoking operations reduced from 55.89 to 28.14 days. The overall operation has improved and the annual profit increased from $936,414 at the base case operating conditions to$14,097,200 at the optimal operating conditions.

Table 4 Optimization results of case study.

5 Conclusions

Modeling, simulation and optimization of steam cracking of ethane in a plug flow reactor has been carried out. A steady-state one dimensional model with improved molecular reaction scheme was used and validated with the industrial results. The simulation results were compared with industrial data and as well as with literature results, it was concluded that the results of current work were able to match better with industrial data and seemed to be a promising representation of overall kinetics of ethane cracking. Furthermore, a sensitivity analysis of reactor was performed using the validated model with improved reaction scheme.The impact of different process parameters including COT, steam-to-ethane ratio, and ethane flow rate on ethane conversion, ethylene selectivity, products yields, and coking rate was investigated. Finally,the annual operating profit was maximized in Aspen Plus using COT and steam-to-ethane ratio taken as decision variables. The results indicate that steadystate optimization had improved the annual operating profit from $936,414 to $14,097,200, when the COT and steam-to-ethane ratio were set at 880°C and 0.59 kg/kg, respectively. Therefore, the results of this work can potentially have a positive impact on the industrial economics. Hence, all these features make this study very appropriate and can provide a signi ficant guidance to process engineers in ethylene industry.

Acknowledgement: The financial support provided by the Project of National Natural Science Foundation of China(21822809 & 21978256), the Fundamental Research Funds for the Central Universities, and the Foundation of State Key Laboratory of High-efficiency Utilization of Coal and Green Chemical Engineering (Grant No.2018-K23) are gratefully acknowledged.