A Lumped Kinetic Model for Low- and Medium-Temperature Coal Tar Hydrocracking Pr

Chen Mang; Yuan Ying; Zhao Jiamin

(1. SINOPEC Fuel Oil Sales Co., Ltd., Beijing 100029; 2.Guangdong University of Petrochemical Technology, College of Chemical Engineering, Maoming 525000;3. SINOPEC Research Institute of Petroleum Processing, Beijing 100083)

Abstract: A five-lump kinetic model for hydrocracking of coal tar is proposed to simulate the product distribution on the basis of the experimental data of coal tar in a laboratory fixed-bed down flow reactor with a Ni/Mo catalyst at a temperature of 360―380°C. The kinetic model includes 5 lumps, viz.: unreacted coal tar, diesel, gasoline, gas, and coke. The product distribution predicted and measured were compared and analyzed, indicating that the kinetic model in this research is suitable for the hydrocracking process of coal tar. Meanwhile, the hydrocracking mechanism of coal tar was further analyzed on the basis of the kinetic model. During the entire hydrocracking process of coal tar, the diesel fraction served as an intermediate component of the reaction.

Key words: coal tar; hydrocracking; kinetics; lumped model

1 Introduction

Coal tar, a by-product of coal pyrolysis process, contains a large number of aliphatic, aromatic, alicyclic, and heterocyclic compounds[1]. The output of coal tar is very large in China (about 15 million tons in 2018) due to the large number of coal resources[2-3]. Thus, there is economic interests for using coal tar to produce conventional liquid fuels (e.g., gasoline and diesel)[4-5]. Since coal tar contains a lot of heterocyclic compounds (e.g., S, N), it currently should comply with the strict environmental standards. Therefore,in recent years, the hydrotreatment of coal tar is emerging as a notable trend for the conversion of low-value coal tar into a valuable clean fuel (e.g., gasoline and diesel)[6-9].

Coal tar hydrotreatment process is a complex reaction process involving a series of reactions including hydrodesulfurization, hydrodenitrification, hydrodemetallization,macromolecular cracking, cyclization, isomerization, and saturation of olefins and partial aromatics, accompanied with complex physical processes such as momentum transfer, energy transfer, and mass transfer[10-13]. Hence,an in-depth investigation of the reaction mechanisms for the hydrotreatment of coal tar is imperative in guiding the comprehensive utilization of coal tar. The reaction kinetics is one of the most critical approaches for interpretation of reaction mechanisms, as well as clarification and correlation between reaction steps to provide important guidance for operation optimization[14-17]. However, due to the extremely complicated composition of coal tar, it is extremely difficult to establish a detailed description of the reaction kinetics of each single component. A lumped kinetic method, widely used in petroleum system[14,18-20],may possibly accomplish the above goal for investigation of coal tar. However, only a few studies of the lumping kinetic model application for coal tar have been reported.Chang Na, et al.[21]established a three-lumped model for the high-temperature hydrocracking of coal tar and examined the hydrocracking behavior of high-temperature coal tar in supercritical xylene. Dai Fei, et al.[22]established an eight-lump kinetic model for describing coal tar hydrogenation and emphatically discussed the effect of operating conditions such as space velocity, hydrogen/oil ratio, temperature, initial hydrogen pressure, and other reaction conditions. Recently, Dai Fei, et al.[23-24]further developed a new kinetic model of coal tar hydrogenation process via the carbon number, which can provide a good prediction for reaction mechanisms. However, this method is so complex that the repeatability is difficult for different coal tar. Therefore, it is essential to develop a kinetic model, which is based on the yet acceptable lumping data,for catalyst screening and basic process studies.

Hence, a laboratory fixed-bed down flow reactor with a Ni/Mo catalyst operating at a temperature of 360―380 °C was employed for developing a five-lump kinetic model for coal tar hydrocracking. Kinetic parameters are estimated by the least-squares criterion and the nonlinear regression program based on the Marquardt algorithm. The product yield is predicted and compared with the experimental value. The present study also focuses on the reaction mechanism during the coal tar hydrocracking process.

2 Experimental

2.1 The experimental device

All experiments were carried out in a fixed-bed highpressure experimental device (Figure 1). The core of the reaction unit is an isothermal reactor (with an inner diameter and a total length equating to 2.54 cm and 143 cm, respectively) equipped with a centrally located thermocouple. The reactor temperature was maintained at the desired level by using three heat preservation zones.

2.2 Hydrogenation experiment

The feedstock for the experiment belongs to mediumtemperature coal tar samples. Its properties are listed in Table 1. Commercial Ni/Mo catalysts (with a specific surface area of 210 m2/g, a pore volume of 0.42 cm3/g, an average pore diameter of 8.2 nm, a metal loading of 10% of MoO3and 3% of NiO) were used for all experiments. Firstly, the catalyst was pre-sulfided by a standard procedure using straight-run gas oil containing 3% of dimethyl disulfide (DMDS) before injecting the feed at a hydrogen pressure of 4.0 MPa, a LHSV of 1.2 h-1and a reaction temperature of 340 °C for 16 h. After sulfidation, the conditions for each run were adjusted to the desired operating temperature (360―400 °C), pressure(6―12 MPa), hydrogen flow rate (1 400―1 800 cm3/h) and liquid hourly space velocity (LHSV of 0.4―1.2 h-1). After a reaction stabilization period of 2 h, the product samples were collected in the course of each run. The product was divided into gas, liquid, and coke, wherein the liquid product was distilled under atmospheric pressure to cut into gasoline and diesel fractions. The residue remaining after distillation was treated as the residual tar. On the basis of material balance,the collected gases were then immediately analyzed by a re finery gas analyzer (Varian CP-3800).

Table 1 Properties of coal tar sample

Figure 1 Schematic of the experimental device

3 Results and Discussion

3.1 Effect of LHSV on product distribution

Typically, products obtained from the hydrocracking of coal tar were mainly divided into gas, diesel, gasoline,and coke. Figure 2 shows the variation of the products yield obtained from hydrocracking of coal tar as a function of LHSV. Clearly, with a decreasing LHSV,the conversion of coal tar gradually decreased, and the diesel yield first increased and then slightly decreased,while the yield of gasoline and gas gradually increased,implying that high-boiling-point molecules are converted into lighter molecules. A majority of the researchers have reported that the thermal reaction of hydrocarbons is in agreement with the first-order reaction, and the secondary reaction of the products mainly occurred in the heavy fraction[19,25-27]. Thus, the variation of diesel yield indicated that the secondary reaction of diesel products is relatively obvious, while the secondary reaction of gasoline components is relatively insignificant (Figure 2),the results can match up to other pilot-plant studies[7].

An extremely small effect of LHSV on product yield,when LHSV was kept at below 0.5 h-1, was observed,indicating that the hydrocracking reaction almost reached its equilibrium conversion. This result also implied that,under moderate hydrocracking conditions, the severity of the reaction in the industrial system is not allowed to exceed that used in our experiments. If the severity is too high, the liquid yield would decrease, and the gas yield would also increase, and consequently the coke deposited on the catalyst will also reach a high level, which would stop the commercial plant operation, leading to an eventual reduction of the economic bene fits.

Another observation from the Figure 2 is that the gasoline yield gradually increased, and a reaction equilibrium was still not reached during the hydrocracking reaction.Two explanations for this phenomenon were provided:(1) The reaction for the conversion of coal tar to gasoline components did not attain an equilibrium; or (2) Gasoline components were generated from catalytic decomposition of diesel components. It is impossible to determine which mechanism predominates by the experimental results.However, it may be beneficial to accomplishing this goal via the calculation of the kinetic parameters of each reaction path, which will be explained in the next section.

Figure 2 The product distribution in coal tar hydrocracking process verus LHSV

3.2 Kinetic model

Figure 3 shows the proposed kinetic model, which includes 5 lumps (i.e., unconverted coal tar, diesel,gasoline, gas, and coke, respectively) and 10 kinetic parameters. Upon considering that the reactions between the five lumps are complicated, the following assumptions are made in the actual calculation process for the convenience of calculation:

(1) All reactions are first-order reactions.

(2) The reaction between the components is irreversible;that is, only the former component can generate the latter component, but the latter cannot generate the former.

(3) The reaction is assumed to be controlled by the reaction kinetics; the effect of diffusion is ignored; and the deactivation of catalyst is not considered.

Based on the above hypothesis, the basic equation of the lumped model for the hydrogenation of coal tar was established as follows:

The kinetic model was incorporated into the model for an isothermal plug flow reactor. Based on previous experience, the axial dispersion and external and internal gradients were ignored. Based on the sum of the squared errors between the experimental and the calculated product components, the objective function was minimized to calculate the optimal set of kinetic parameters. The objective function was solved by using the least-squares criterion and the nonlinear regression procedure was based on the Marquardt algorithm.

Figure 3 Kinetic reaction model in coal tar hydrocracking process

Table 2 summarizes the values of kinetic parameters and the activation energy of each reaction. The first finding of the results shown in Table 2 con firmed the aforementioned portion in the last paragraph of the previous section:There are two main sources of gasoline components,i.e., the hydrocracking of coal tar and the diesel fraction,respectively. Both k8and k9are zero, indicating that the gasoline component does not react further. Hence, the gasoline component increases throughout the reaction.

Table 2 Kinetic parameters and activation energy

It can be seen from Table 2 that the highest kinetic parameter k1was obtained for the formation of the diesel fraction, also indicating that diesel was the main product obtained from the hydrogenation of coal tar, and the diesel fraction would further react (with k5and k6being not equal to zero) as the middle fraction for the entire hydrogenation process.

The selectivity for the hydrocracking of coal tar slightly varied at different temperatures. For example, at 360 °C,gas and coke would not be formed from diesel, because k6and k7values were zero. Conversely, these values of parameters were not equal to zero at 370 °C and 380 °C.As some values of the kinetic parameters were zero,not all activation energies were estimated. Figure 4 shows the Arrhenius plot of all kinetic constants. The correlation coefficient R2for the logarithm of the reaction rate constant lnk and the reciprocal of the reaction temperature 1/T was greater than 0.95. In addition,Table 2 summarizes the activation energy values for some of the reactions, which were within the range reported in previous studies.

Figure 4 Variation of dynamic constant as a function of temperature

The obtained kinetic parameters were substituted into the reaction rate equation, and the reaction rate differential equation was solved by the fourth-order Runge-Kutta method. The total yields of lumps at different times were obtained. The predicted values for the lumped models were compared with the measured values (Figures 5 and 6). Results revealed that the predicted values for the lumped models were similar to the measured values,indicating that the proposed kinetic model could fit in with the hydrocracking of coal tar.

Figure 5 Comparison of the prediction data for the diesel yield

Figure 6 Comparison of the prediction data for the gasoline yield

4 Conclusions

The experiments for hydrocracking of coal tar were carried out in a laboratory micro-reactor. A five-lump parallelsequence reaction model applicable to the hydrocracking of coal tar was established. The model parameters were estimated by the least-squares method. Statistical test results obtained from the predicted values by the model and the experimental values revealed a good agreement.Moreover, the hydrogenation mechanism of coal tar was further analyzed on the basis of the kinetic model,indicating that the diesel fraction served as an intermediate component during the entire hydrogenation of coal tar.

Acknowledgment: The authors gratefully acknowledge the financial support by the State Key Project of SINOPEC(ST18009-3-17).