(570p) A Model for the Prediction of Coke Deposition During Thermal Cracking of Ethane/Propane Mixtures

INTRODUCTION Thermal cracking of light hydrocarbons is the main route for the production of important raw materials for the chemical industry, such as ethylene and propylene. The most used technology for ethylene production involves injection of a mixture of hydrocarbons, preferably ethane, into a coil located in a furnace with multiple burners that provide the required energy for the highly endothermical cracking reactions. Associated with cracking there is undesirable coke deposition on the walls of the coil. Coke deposits build with reactor operation time and increase up to a point in which pressure drop along the reactor and reduction on heat transfer across the reactor's wall are so high that the furnace needs to be stopped for decoking. The operational time before decoking is of the order of 20 to 90 days, depending on process conditions and load. To decrease coke deposition, steam is added to the hydrocarbon mixture at a ratio (known as dilution factor) that typically varies between 0.2 to 1 kg steam/kg hydrocarbons [1]. As the use of heavier hydrocarbon fractions is expected to increase as light oil reserves are depleted, a model that can predict the effect of changes in hydrocarbon composition on the time before decoking as well as how operational parameters, such as pressure and temperature, affect the rate of coking deposition, is something desirable. Prior to simulation of coke formation, it is necessary to model the process of hydrocarbon cracking. In the reviewed literature, the seminal work of Sundaram and Froment [2-4] is recognized as one of the first studies on this area. These authors analyzed the pyrolysis of ethane, propane, isobutane and n-butane as well as their mixtures. They proposed homogeneous [2, 3] and radical mechanisms for these processes [4]. After Sundaram and Froment's work, Ranzi and collaborators [5-7] carried out various studies in which they modeled the conversion of different hydrocarbon mixtures during the cracking processes. These studies finally lead to the SPYRO code, currently used in the hydrocarbon industry, to predict cracking. This code is currently licensed by Pyrotec, a divison of Technip [8]. Froment and coworkers studied coke deposition as well. In 1990, Froment published a detailed review [9] of the state of the art on coke deposition where he summarized the current knowledge of coke formation during cracking of ethane, propane and naphtha. In that review, Froment addressed some of his previous work on coke deposition for ethane [10] and propane [11] cracking. Future understanding of the coke-formation process was obtained by Albright et al. [12-14] and more recently by Wauters and Marín [15]. Despite all these studies, there are very limited kinetic expressions available in the open literature for the prediction of coke deposition during cracking of ethane and propane. In fact in the refereed literature, only Sundaram and Froment [10, 11] explicitly report kinetic expressions for coke formation during ethane and propane cracking. Other studies, such as those of Ranzi et al. [5-7] and Albright et al. [12-15] show a very good description of the mechanism of coke formation but detailed kinetic expressions are missing. This paper describe a model, based on the aforementioned model by Sundaram and Froment, for coke deposition during the cracking of ethane and propane mixtures. Predictions by the model are analyzed and compared to industrial data. The paper describes the main limitations of the model and proposes strategies to improve the model in order to obtain more scientifically sound predictions. MODEL To be consistent with the selection of Froment's kinetics on coke deposition, we selected the mechanism described by Sundaram and Froment for cracking of ethane and propane mixtures and shown in [1] and [2]. These gas-phase kinetics were combined with the coke formation reactions shown in Table 1 where R 1 and R 3 represent the conversion of ethane and propane, respectively, to products that can lead to coke formation. R 2 and R 4 are the actual coke-formation reactions. When the models by Sundaram and Froment [10, 11] for coke formation are coupled with mechanisms that predict ethane and propane cracking, R 2 and R 4 are used with the concentration of C4+ (that represents hydrocarbons with four or more carbons) and propylene to simulate coke deposition. RESULTS In order to evaluate the model, data from an industrial cracking unit was compared to model predictions. The input to the reactor was a mixture of ethane/ethylene (99.5/0.5 molar %). In the model, the reactor geometry and the temperature and pressure boundary conditions were those measured by the furnace operators. Table 2 compares model results with industrial data. The ratio of predictions and measurements shows a reasonable agreement for the main products: ethane, ethylene, propane and propylene. Predictions for methane, propane, butane, butene and acetylene are much more inaccurate probably because the kinetic constants were calculated in experiments aimed at the prediction of the most abundant compounds. Figure 1 presents the predicted reduction in coil diameter along the reactor after a 700-h run. Results are for mixtures of pure ethane and propane with dilution factors of 0.5 and 0.4 kg steam/kg hydrocarbon respectively. According to the model, coke deposition starts at 20 m from the reactor entrance because the homogeneous model for ethane cracking predicts negligible C4+ formation before that length. Contrary, for propane, there is coke formation at short distances in the reactor as C3H6 production starts almost from the start of the process. The other parameter that controls coke deposition is temperature. At low reactor lengths temperature may be too low to start the coke-formation process. Figure 2 shows the predicted reduction in tube diameter while increasing the percentage of propane in the mixture. In this case, we used the cracking of ethane and propane model to predict the composition of the gaseous products and the sum of the coke deposition models for ethane and propane to predict the diameter reduction as suggested by Froment et al. [10, 11]. Figure 2 shows that diameter reduction increase from 8% with ethane to 60% with propane. Interestingly, for 100% ethane and propane in Figure 1 and for the same reactor conditions, the predicted diameter reduction was 14% and 18% respectively. Summing up the rates of coke deposition for ethane and propane leads to a reduction on the predicted coke deposition for ethane and an increase in the value for propane. The inconsistency on coke deposition predictions for the 100% ethane and 100% propane cases depending on whether the rates are summed or not is clearly an undesirable result. Nevertheless, it allows a first order estimate on the trend that addition of propane to ethane can have on coke deposition. References 1. Froment, G.P., Van de Steene, B.O., Van Damme, P.S., Narayanan, S., and Goossens, A.G., Thermal cracking of ethaneandethane-propane mixtures. Ind. Eng. Chem. Proc. Des. Dev., 1976. 15(4): p. 495-504. 2. Sundaram, K.M. and Froment, G.F., Modeling of thermal cracking kinetics i: Thermal cracking of ethane, propane and their mixtures. Chem. Eng. Sci.,1977. 32(6): p. 601-608. 3. Sundaram, K.M. and Froment, G.F.,Modeling of thermal cracking kinetics--ii : Cracking of isobutane, of n-butane and of mixtures ethane-propane-n-butane. Chem. Eng. Sci., 1977. 32(6): p. 609-617. 4. Sundaram, K.M. and Froment, G.F., Modeling of thermal cracking kinetics. 3. Radical mechanisms for the pyrolysis of simple paraffins, olefins, and their mixtures. Ind. Eng. Chem. Fund., 1978. 17(3): p. 174-182. 5. Dente, M., Ranzi, E., and Goossens, A.G., Detailed prediction of olefin yields from hydrocarbon pyrolysis through a fundamental simulation model (spyro). Comp. Chem. Eng., 1979. 3(1-4): p. 61-75. 6. Dente, M., Pierucci, S., Ranzi, E., and Bussani, G., New improvements in modeling kinetic schemes for hydrocarbons pyrolysis reactors. Chem. Eng. Sci., 1992. 47(9-11): p. 2629-2634. 7. Ranzi, E., Sogaro, A.,Gaffuri, P.,Pennati, G., and Faravelli, T., A wide range modeling study of methane oxidation. Comb. Sci. Techn., 1994. 96(4): p. 279 - 325. 8. Tecnip. Accessed 2009 [cited; Available from: http://www.technip.com/english/index.html. 9. Froment, G.F., Coke formation in the thermal cracking of hydrocarbons. Rev. Chem. Eng., 1990. 6(4): p. 294-328. 10. Sundaram, K.M., Damme, P.S.V., and Froment, G.F., Coke deposition in the thermal cracking of ethane. AIChE J, 1981. 27(6): p. 946- 951. 11. Sundaram, K.M. and Froment,G.F.,Kinetics of coke deposition in the thermal cracking of propane. Chem. Eng. Sci., 1979. 34(5): p. 635-644. 12. Albright, L.F., Comments on kinetic modeling of coke formation during steam cracking. Ind. Eng. Chem. Res., 2002. 41(24): p. 6210-6212. 13. Albright, L.F. and Marek, J.C., Analysis of coke produced in ethylene furnaces: Insights on process improvements. Ind. Eng. Chem. Res., 1988. 27(5): p. 751-755. 14. Albright, L. and Marek, J.,Coke formation during pyrolysis: Roles of residence time, reactor geometry, and time of operation. Ind. Eng. Chem. Res., 1988. 27(5): p. 743-751. 15. Wauters, S. and Marin, G.B., Kinetic modeling of coke formation during steam cracking. Ind. Eng. Chem. Res., 2002. 41(10): p. 2379-2391.