Dai Meng; Li Shicai; Li Yang; Xu Dahai; Ding He; Chen Guang; Guo Rong

(SINOPEC Dalian Research Institute of Petroleum and Petrochemicals, Dalian 116045)

Abstract: In this work, the influence of reaction conditions on hydrocarbon distribution and product quality was tested to obtain the rule of variation with different residence time. A semi-experience and semi-mechanism dynamics model was established and modified. Based on the simulation results, the hydrogenation reactor is divided into five regions: the upper two regions are mainly saturated with most of tri-cyclic and bi-cyclic aromatics, and 90% of nitrides are converted, while all simple sulfide and some 4,6-DMDBT compounds undergo transformation reactions. In the lower three regions, the reaction rate of HDS and HDN slows down, while the rate of monocyclic aromatic saturation increases. The HDS, HDN and HDA rates of the optimized catalyst grading scheme are significantly increased.

Key words: diesel, HDS reactor, reaction type, kinetic model, simulation

1 Introduction

With an increasingly strict environmental laws and regulations, the quality of diesel products has been improving. In order to further improve the efficiency of diesel catalytic hydrogenation reaction, the reaction environment and reaction types must be fully understood.The catalyst grading technology[1-5]is the most flexible and adaptable as compared with the development of new catalysts or new reactors, and can also save investment in new installations. The reaction area in the reactor is generally divided into two parts. The upper reaction temperature and H2S concentration are quite low, and the hydrogen partial pressure is relatively high. As the feedstock flows downwards, the hydrogenation reaction gives off heat and generates H2S, resulting in a corresponding increase in temperature at the lower part of the reactor, in which the partial pressure of hydrogen decreases. According to the different reaction environment, for high pressure reactors, the Ni-Mo type hydrogenation catalyst is usually loaded on the upper part of the reactor[6-8]for removal of simple sulfides, nitrides and aromatics. The Co-Mo catalyst is loaded on the lower part of the reactor to realize a deep hydrodesulfurization rate at high temperature. However, this classification is not detailed enough to guide the further development of catalyst grading system under the new requirements for improved quality and efficiency.

At present, in the hydrogenation reactor, the chemical reaction law is mainly investigated by reaction kinetics. In the process of deep hydrodesulfurization of diesel fuel oil,the removal of sulfide mainly follows three reaction routes,viz.: hydrodesulfurization (HDS), direct desulfurization(DDS), and alkyl transfer desulfurization[9]. The simple structure of sulfide, thiol, thiophene, and benzothiophene(BT) on the surface of the catalyst mainly follows the DDS route. However, due to the limitation of steric hindrance, complex macromolecular sulfides, such as 4,6-dimethyldibenzothiophene (4,6-DMDBT), mainly follow the route of HDS and alkyl transfer.

For Ni-Mo-Co catalyst, Zhu Zelin[10]concluded that the kinetics of hydrodesulfurization of thiophene could be described as a second-order model, while the activation energy calculated was 76.21 kJ/mol, and the frequency factor was 3.67×108. For Co-Mo catalyst,based on the kinetics of thiophene and benzothiophene hydrodesulfurization, Hui Yang[11]proposed that the rate control steps involved only one active site, and there was a competitive adsorption with H2S. Xu Yongqiang[12]studied the hydrodesulfurization reaction of 4,6-DMDBT on Co-Mo catalyst, and suggested that the HDS reaction route is obviously superior to DDS.Yu Hang[13]established a three-lump kinetic model for hydrodesulfurization of diesel fractions of Fushun shale oil, by dividing the sulfides into three lumps according to the reaction rate, and then Yu Hang proposed that pressure effect on the refractory sulfide was greater than that of the easily removable ones.

As for HDN of aromatic aza-cyclic compounds, the azacyclic compounds must be hydrogenated first before the removal of nitrogen atoms[14]. Before the C-N bond of aniline compounds is broken, aromatics saturation is also required. Xu Zhengli[15]studied the reaction law of HDN of gas oil, and proposed that the main factors affecting the denitrification reaction were temperature and pressure, as well as the H2S concentration in circulating hydrogen, and the nitride type.

The hydrogenation reaction of aromatic compounds is reversible and cannot be completely transformed under typical conditions[16]. According to the hydrogenation mechanism of aromatics, Jiang Hongbo[17]divided the compounds contained in diesel fraction into three lumps according to aromatic rings, and established a lumped reaction kinetic model, which considered the influence of adsorption competition.

In the test, sulfides were divided into four lumps according to their reaction mechanisms. Then tests were carried out at changing temperature to investigate the influence of liquid hourly space velocity (LHSV) on impurity content and hydrocarbon composition of the product. According to test results, the dynamic models of HDS, HDN and HDA were established and rationalized,and the parameters were modified. Finally, by simulating the change of S, N, aromatics and H2S content in different areas of the reactor, the corresponding reaction types were deduced, and the reactor was divided into five parts. The traditional catalyst gradation system only uses two different types of catalysts. The volume ratio is usually estimated based on experience and cannot be adjusted flexibly according to the properties of feedstock and product requirements. This work takes into account the interactions between different substances, with the upper part being two and the lower part being three,which could better guide different types of catalysts in the reactor to develop their corresponding characteristics and form synergistic effects. Compared with the Ni-Mo catalyst only, the optimized system can further reduce the saturation of mono-cyclic aromatics, and decrease the hydrogen consumption. Compared with the use of Co-Mo catalyst alone, the optimized catalyst system has a higher denitrification activity and can further promote the ultradeep desulfurization reaction. Compared with the original catalyst system (Ni-Mo and Co-Mo catalyst grading), the optimized system can alleviate the operating conditions,which is equivalent to improving the reaction activity of the catalyst system under the same conditions.

2 Process Tests

2.1 Raw materials and catalyst properties

The straight run diesel was used as raw material. Catalyst A (Ni-Mo type), catalyst B (Mo-Co type), and bulk catalyst C (W-Mo-Ni type), which have been used in the industry, were selected to carry out the process test and optimization test. The properties of raw materials and catalysts are shown in Table 1 and Table 2, respectively.

Table 1 Main properties of straight run diesel

In different reaction areas, the reaction environment was studied by analyzing the product quality at different liquid hourly space velocity (LHSV). The H2S content and hydrogen partial pressure were obtained by simulation.

Table 2 Properties of catalysts

Flow diagram of the HDS reactor system is shown in Figure 1. Process tests were conducted under conditions covering a system pressure of 6.4 MPa, a reaction temperature of 350 °C and 360 °C，and an oil/hydrogen volume ratio of 400. Catalyst A is loaded onto the upper part of the reactor, and catalyst B is loaded onto the lower part at a ratio of 1:1. Optimizing tests were performed under conditions covering a system pressure of 6.4 MPa,a reaction temperature of 350 °C, an oil/hydrogen volume ratio of 400, and a constant LHSV. The reactor was loaded with catalysts C, A, and B from the top to the bottom, and the ratio of these three catalysts was 2:1:2.

2.2 Establishment of 4-lumps of HDS reaction

The structure of sulfur compounds in diesel fractions is complex and the reaction rates are different, so it is difficult to fit them dynamically according to a certain reaction order. In this experiment, for the first time the sulfur compounds with different reaction mechanisms[18-19]were divided into four lumps, and the kinetic equation of sulfur was established using a semi-mechanism and semiempirical model.

According to the structure type, reaction rate and route[20-21], the sulfides are divided into 4 lumps: S1 are sulfides including mercaptans, thioethers and thiophenes,which can be completely removed at deep HDS level.S2 are benzothiophenes, which can be removed by direct desulfurization (the DDS route) because of their rapid reaction rate. S3 are sulfides of dibenzothiophenes without substituents in the benzene ring position 4 or 6. HHDBT generated in the HYD route has a structure similar to the benzothiophenes and therefore belongs to S2. S4 are dibenzothiophenes with substituents in the ring positions 4 and 6 and they have three reaction routes, viz.: DDS, HYD and alkyl transfer. The reaction process of DDS and HYD routes are similar to that of S3. As there are no substituents occurring in the ring positions 4 and 6 after alkyl transfer,the reaction product of which can be regarded as S3. The reaction network of four-lumped sulfides is shown in Figure 2, whereKiis the reaction rate constant.

2.3 Establishment of kinetic model

2.3.1 Kinetic model of HDS

Figure 1 Flow diagram of HDS reactor systems

Figure 2 Schematic diagram of hydrodesulfurizationreaction network

In the same lump, the reaction mechanism of sulfides is similar and the reaction rate is of the same order of magnitude. Therefore, it is assumed that all reactions in the network conform to the first-order irreversible reactions. According to the reaction network in Figure 2,the differential equations of various sulfides are as follows(Equations 1 and 2):

whereCS1—CS4is the concentration of S1—S4, μg/g,respectively;tis reaction time, h;kis a reaction rate coefficient;Eis reaction activation energy, J/mol;PH2is hydrogen partial pressure, MPa;Tis reaction temperature,K;a,b,c,dis coefficient of equation, respectively;CN,CH2Sis the nitride and H2S concentration(μg/g), respectively;CAis concentration of aromatics, %.

During the reaction, H2S is produced by hydrogenation of sulfides, which will affect the hydrogenation reaction through competitive adsorption[22]. As mentioned above,the differential equation of the formation rate of inorganic sulfur is as follows (Equation 3):

Relevant studies show that H2S mainly affects the DDS route of reaction involving DBT[23], but has little influence on the HYD route. As noted above, the influence on the reaction conforms to the following rule:

whereBis the apparent adsorption equilibrium constant of H2S.

Nitrogen content has a much larger influence on reaction rate of HYD route than DDS[24]. It is assumed that the nitrogen content affects only the reaction rate coefficientsK5andK6(HYD route) of S3 and S4 lumps, and complies with the following rules:

whereDis the apparent adsorption equilibrium constant of nitrides.

Polycyclic aromatics also have an impact on competitive adsorption via the HYD route of HDS, which conforms to the following law:

2.3.2 Kinetic model of HDN

In this paper, the effect of temperature, hydrogen concentration, and bi-cyclic aromatics content[25]on the HDN reaction conforms to then-order reaction kinetics. The kinetic model of HDN reaction is as follows (Equation 4):

2.3.3 Kinetic model of aromatics saturation

Aromatics can be divided into three groups, viz.: tricyclic aromatics, bi-cyclic aromatics, and mono-cyclic aromatics. There is a reversible reaction of hydrogenation of bi-cyclic aromatics to mono-cyclic aromatics.

Assuming that all reactions conform to the first-order kinetic model, the kinetic equations of aromatics saturation are as follows (Equation 5):

whereCNa,CPis concentration of naphthenes and paraffins, %.

2.4 Solving parameters of kinetic model

The Levenberg-Marquardt (L-M) algorithm was chosen to fit the equation, which is an improvement of the Newton method. The minimization of objective function is called nonlinear least squares problem, and the L-M algorithm has become a standard for solving this problem. After the initial parameter values of linear regression are obtained,the L-M algorithm is used to carry out regression, and the test data of diesel fuel oil under different operating conditions are correlated. After desulfurization of diesel fuel oil, the optimal objective functionZis the sum of residual squares of test value and the calculated value of sulfur content (Equation 6):

The equations were calculated by using the software MATLAB. The initial values of each parameter were given, and the calculated results were compared with the experimental results in order to adjust the parameters and optimize the minimum parameter setting of the equation.The results of parameter optimization are shown in Tables 3—5.

Table 3 Optimization results of parameters of HDS kinetic model

Table 4 Optimization results of parameters of HDN kinetic model

Table 5 Optimization results of parameters of aromatics saturation kinetic model

3 Results and Discussion

3.1 Reaction results and comparison with predicted results

The inversion of the volume hourly space velocity can be regarded as the residence time of reaction.With the extension of residence time, the properties of hydrogenated products have changed, reflecting the different reaction types in the reactor. Figure 3 and Figure 4 show the sulfur and nitrogen contents of hydrogenated products at 350 °C and 360 °C, respectively, which are compared with the predicted results that were calculated on the basis of established kinetic models.

Judging from the results of experiments obtained at 350°C, it can be seen that the S and N content decreases gradually with the downward flow of raw materials. The S content of refined oil could be less than 10 μg/g, and the N content could be less than 1 μg/g under conditions covering a system pressure of 6.4 MPa and a liquid hourly space velocity of 1.0 h-1, when the diesel fuel oil was processed in the presence of the Mo-Ni type and the Mo-Co type catalysts. From the point of view of the reactor area, the material would have to flow to the bottom to reach the sulfur content standard at 350 °C. Predictions and experimental data show the same trend.Judging from the results of experiments obtained at 360 °C,it can be seen that the S content of refined oil could be less than 10 μg/g, and the N content could be less than 1 μg/g under conditions covering a system pressure of 6.4 MPa and a liquid hourly space velocity of 1.5 h-1, when the diesel fuel oil was processed in the presence of the Mo-Ni type and the Mo-Co type catalysts. This fact denotes that the sulfur content can meet the standard in the middle and lower parts of the reactor. The predicted results are consistent with the experimental results.

Figure 3 Comparison of predicted and experimental results of S and N content at 350 °C

Figure 4 Comparison of predicted and experimental results of S and N content at 360 °C

Figure 5 and Figure 6 compare the predicted and experimental results of hydrocarbons obtained at 350 °C and 360 °C.

Figure 5 Comparison of predicted and experimental results of hydrocarbons at 350 °C

Figure 6 Comparison of predicted and experimental results of hydrocarbons content obtained at 360 °C

It can be seen from Figure 5 and Figure 6 that the content of paraffinic hydrocarbons and naphthenic hydrocarbons increased slightly. At a reaction temperature of 360 °C,paraffins and naphthenes are produced more rapidly than that obtained from aromatics at 350 °C. In the lower part of the reactor, the content of naphthenes increases rapidly and should be formed through the conversion of monocyclic aromatics. The predicted results also show the same trend as the experimental data.

In Figure 7 and Figure 8, the predicted and experimental results for aromatics content were compared between two different reaction temperatures, namely 350 °C and 360 °C.

Figure 7 Comparison of predicted and experimental results of aromatics content obtained at 350 °C

Figure 8 Comparison of predicted and experimental results of aromatics content obtained at 360 °C

As shown in Figure 7 and Figure 8, the mono-cyclic aromatics content at first increased and then decreased as the reaction progressed, while the bi-cyclic aromatics content decreased gradually, and the tri-cyclic aromatics content decreased significantly at the beginning of the reaction. At 350 °C, the mono-cyclic aromatics content peaked as the feedstock was introduced to the middle of the reactor, while the mono-cyclic aromatics peaked in the upper part of the reactor at 360 °C. The increase of mono-cyclic aromatics content in the upper part of the reactor is attributed to the higher conversion rate of polycyclic aromatics in comparison to that of mono-cyclic aromatics. However, in the lower part of the reactor, the decrease in mono-cyclic aromatics content is ascribed to a large conversion to cycloalkanes, the increase of which is consistent with that shown in Figures 5 and 6. The simulation results are consistent with the experimental results.

According to the analysis results, the sulfur and nitrogen content was in logarithmic function with the reaction residence time, and the paraffins and naphthenes content followed the same rule. With the prolongation of reaction residence time, the content of bi-cyclic and tri-cyclic aromatics decreased rapidly, while the mono-cyclic aromatics content increased at first and then decreased.

3.2 Simulation calculation and deduction of reactor environment

By comparison, it can be seen that predicted results of the established kinetic models are basically consistent with the experimental results. The simulation accuracyR2for sulfur content (< 20 μg/g) is about 0.7739.

The kinetic model (Equation 2) was used to predict the H2S content at different reaction residence time. The simulation results are shown in Figure 9. It can be seen that the H2S content increased rapidly with the progress of HDS reaction and stayed high. At 360 °C，the H2S content increased more quickly than that obtained at 350 °C.

The whole reactor is divided into five reaction regions according to the calculation results of the kinetic models.The reaction environment and main chemical reactions in each area are shown in Figure 10. It can be seen that the content of polycyclic aromatics and nitrogen has a great influence on the removal of sulfides in the upper part of the reactor, especially on the sulfides that have steric hindrance and have to be processed through the HYD route. DBT with no substituents can be removed at the 3rdbed of the reactor, while the mono-cyclic aromatics and DBT with substituents enter into reaction in the middle of the reactor. The content of impurities and aromatics in each reaction area is shown in Figure 11.

In the first two reaction regions, hydrogenation reactions that have a high reaction rate are carried out, including the saturation of most tri-cyclic and bi-cyclic aromatics, and the conversion of simple-structure sulfides and 90% of nitrides. In the last three reaction regions, the reaction rate of HDS and HDN slows down, while the saturation rate of mono-cyclic aromatics increases. It is speculated that in the early stage of hydrogenation reaction, conversion rate of the refractory sulfur compounds was mainly affected by polycyclic aromatics and nitrides. While in the later stage, it was greatly affected by the inhibition of hydrogen sulfide.

Figure 9 Variation of H2S content with reaction residence time at different temperature

Figure 10 Main chemical reactions occurring in each region

Figure 11 Impurity and aromatics content in each region

3.3 Results of optimization test

Depending on the type of reactions mentioned above, the catalyst grading system was optimized. Catalyst C (a bulk catalyst with high hydrogenation activity) was loaded in the first two regions of the reactor to increase the reaction rate of polycyclic aromatic hydrocarbons and nitrides,and the influence on HDS reaction could be decreased accordingly. Catalyst A (Ni-Mo type) and Catalyst B (Co-Mo type) were loaded according to the grading system in the last three reaction regions to increase the reaction rate of HYD and DDS of sulfides. Optimized results of S and N content are shown in Figure 12.

Figure 12 Predicted and experimental results of S and N content after optimization

As shown in Figure 12, the optimized grading system has a much faster reaction rate of sulfides removal. At the bottom of the reactor, the S content is less than 0.1 μg/g.The N content is 1 μg/g in the middle of the reactor, while it is 360 μg/g before optimization. Optimized results of the paraffins, naphthenes and aromatics contents are shown in Figure 13 and Figure 14.

Figure 13 and Figure 14 show that more paraffins are produced with the optimized catalysts grading system,which can promote the further transformation of naphthenes. The content of mono-cyclic aromatics is also lower than that obtained from the pre-optimized system,indicating that the optimized one has stronger activity of hydrogenation and naphthenes conversion.

Figure 13 Predicted and experimental results of paraffins and naphthenes content after optimization

Figure 14 Predicted and experimental results of aromatics content after optimization

The reaction activity and chemical hydrogen consumption of catalysts A, B, and C are shown in Table 6, and they are compared with 2 kinds of grading systems. The evaluation result is obtained under conditions covering a reaction temperature of 350 °C, a system pressure of 6.4 MPa，an oil/hydrogen ratio of 400, and a gas hourly space velocity of 3 h-1. The straight run diesel was used as the raw material. The volume ratio of A/B in grading system is 1:1, while the volume ratio of C/A/B in grading system is 2:1:2.

Table 6 Reaction activity of Catalysts A, B and C

As shown in Table 6, the catalyst C has excellent HDS,HDN, and aromatics saturation activity. At the same time,the chemical hydrogen consumption of catalyst C is much high than the other two catalysts. The C+A+B grading system has advantages in combining both the catalyst activity and the chemical hydrogen consumption.

4 Conclusions

1) The test results show that the sulfur and nitrogen content was in logarithmic function with the reaction time, while the paraffins and naphthenes content followed the same rule. With the prolongation of reaction residence time, the content of bi-cyclic and tri-cyclic aromatics decreased rapidly, while the mono-cyclic aromatics content increased at first and then decreased.

2) Based on the influence of nitrogen compounds and polycyclic aromatic hydrocarbons on HDS reactions,and the difference of sulfides structure and rate of HDS reactions, the semi-empirical and semi-mechanism kinetic model established thereby has good adaptability and accuracy, which can be used for studies on reactions occurring in different reaction regions of the reactor.

3) Based on simulation results, in the first two reaction regions, the hydrogenation reactions that had a high reaction rate would take place, which included the saturation of most tri-cyclic and bi-cyclic aromatics, and the conversion of simple-structure sulfides and 90% of nitrides. In the last three reaction regions, the reaction rate of HDS and HDN slowed down, while the saturation rate of monocyclic aromatics was increased. It is speculated that in the early stage of hydrogenation reaction, the conversion rate of the refractory sulfur compounds was mainly affected by polycyclic aromatics and nitrides. However, in the later stage it was greatly affected by the inhibition of hydrogen sulfide.4) Calculation results of the kinetic model indicate that the first two reaction regions have a fast reaction rate coupled with most complex reaction types. Catalysts with high hydrogenation activity can be considered to be loaded in the upper part of the reactor to increase the reaction speed of polyaromatic hydrocarbons and nitrides, and to quickly reduce the influence on hydrodesulfurization. The last three reaction regions loaded with Ni-Mo catalyst (with high HYD activity) and Co-Mo catalyst (with high DDS activity) according to the grading system can increase the reaction speed of direct desulfurization of sulfides.

Acknowledgment:This study was financially supported by the National 13th Five-Year Plan Project of China (No.YS2017YFGX010180).