A New Kinetic Model for Residue Hydrotreating Based on Chemical Reaction Section

Zhang Kui; Dai Lishun; Nie Hong; Shao Zhicai; Liu Tao;Deng Zhonghuo; Han Wei; Jia Yanzi

(SINOPEC Research Institute of Petroleum Processing, Beijing 100083)

Abstract: New definition of virtual molecular-group components about CH and CH2 is proposed in the paper. The new reaction, hydrogenation of CH to CH2 (HDCH), can be used to represent the change of C and H in RHT. By using the lumping approach, the kinetic models about HDCH, HDS, HDN, HDCCR, HDM, HDNi and HDV reactions were established. They are key components of the new kinetic model for RHT based on reaction sections. During the construction of kinetic model, goodness-to- fit (R2) between test data and model data was set as the target value, and the model parameters were calculated accurately. In veri fication test, the predicted content of H, S, N, CCR, M, Ni and V can match well with the test data.

Key words: residue hydrotreating; unsaturated hydrocarbons; kinetics; model; lumping

1 Introduction

Residue is the heaviest distillate in crude oil, and most impurities (such as metal, sul fide and nitrides) in crude oil are enriched in the residue. Residue hydrotreating (RHT)is an important technology for upgrading residue. Most impurities can be removed through the RHT process, and then the hydrotreated residue can be cracked into more valuable products by FCC (fluid catalytic cracking) or other processes[1,2].

During the RHT process, the reactions mainly include HDS, HDN, HDCCR, and HDM (such as HDNi and HDV). These reactions have different reaction characteristics. As about 80% of N atoms in residue are contained in resins and asphaltenes[3], the saturation of N-containing unsaturated heterocyclics is necessary before the C-N hydrogenolysis, resulting in low HDN reactivity[4]. Compared to the N compounds, the S atom can be more easily removed by C-S hydrogenolysis after the saturation of S-containing unsaturated heterocyclics[5], since the C-S bond has a lower bond energy. Meanwhile, some S compounds can be removed by HDS reaction through direct C-S hydrogenolysis. So HDS generally has a high reactivity. The HDCCR and HDM reactions can reduce the content of Conradson carbon residue (CCR) and metals, respectively. But the produced coke and metals (such as Ni and V) can deposit on the catalyst surface, easily leading to the rapid deactivation and nonrenewable characteristic of RHT catalyst[6-7]. Besides, the composite catalysts,comprised of the HDM and HDS catalysts, are used in RHT. The catalysts are chosen on the basis of activity,selectivity and stability. For the same reaction, different catalysts display the various reactivity. Thus, research on the reaction kinetic model of composite catalysts is necessary and essential for RHT.

Recently, many researches have been working on RHT kinetic models[8-15]. During the modeling process, the lumping approach[16]is extensively used. For example,Callejas[17-20], Calderón[21], and Asaee[22]developed the 4-,5- and 6-lump cracking models for RHT, respectively.Actually, there are two types of kinetic models, including the models[23-30]based on overall chemical reaction sections and the models[31-32]based on each chemical reaction section. Compared with the former model, when the composition of composite catalysts changes, the latter one can better predict the catalytic performance.

Marafi[32]considers that the performance of the overall hydrotreating process relating to various reactions, such as HDS, HDN, HDCCR, and HDM, is clearly linked to the catalyst in different reactors.

In the present work, two types of commercial catalysts,HDM and HDS catalysts, developed by the Sinopec Research Institute of Petroleum Processing (RIPP) were loaded separately in two fixed-bed reactors. The firstreaction and the second-reaction sections form and fix the HDM and the HDS catalyst beds, respectively.Through RHT tests, a new kinetic model for RHT based on reaction sections was established. Meanwhile, a new reaction type “HDCH” was employed and will be described later in the paper.

2 Experimental

2.1 Test preparation

Residue A was used as the feed, with its main properties listed in Table 1. As shown in Table 1, the contents of Ni and V are 22.7 μg/g and 80.4 μg/g, respectively. The S content of 4.632% in the residue is by 20 times more than the N content of 0.23%.

The RHT tests were carried out in a pilot-scale plant shown in Figure 1. Along the stream flow direction, HDM and HDS catalysts were loaded successively into the two fixed-bed reactors, constituting section I and section II.The products and the sample between two reactors were collected and analyzed during RHT tests. The two used catalysts are commercial catalysts developed by RIPP.The properties of these catalysts are listed in Table 2.Besides the HDM/HDS catalysts, a small amount of guard catalyst was also loaded above the HDM catalyst in the first reactor. The guard catalyst has a high Fe/Ca tolerance capability for a relatively long time.

After the hydrotreating catalyst sul fidation, the RHT tests were carried out under conditions covering: a reaction temperature (T) ranging from 365 °C to 405 °C in R1 and R2; a reaction pressure (p) ranging from 13.0 MPa to 16.0 MPa; a liquid hourly space velocity (LHSV) ranging from 0.15 h-1to 0.40 h-1; and a H2-to-oil (H2/oil) volume ratio of 700:1. One part of RHT tests was applied to construct the new kinetic model, and the other part of RHT tests was used as the veri fication test.

Table 1 Main properties of residue A

Figure 1 The pilot-scale plant for RHT tests

2.2 Sample analysis

Density of residue and its hydrotreated oil was analyzed by a DMA 4500M density meter (Anton Paar Company),and a CAV 2100 fully-automatic digital kinetic viscometer(Cannon Instrument Company) was used to analyze the kinematic viscosity. C and H content can be obtained by a Vario ELCube elemental analyzer (Elementar, Inc.) for detecting the amount of produced gases after the sample combustion. A Lab-X3500SCL X-ray fluorescence spectrometer (Oxford Instruments, Inc.) was applied to determine the S content based on the strength of the S-Kα characteristic spectral line with 2.3 keV. The N content was analyzed by a multi EA5000 elemental analyzer(Analytik Jena AG) through detecting the light emitted from the decay of excited-state NO2. A MCRT-160 micro carbon residue tester (Aicor by PAC, Inc.) was used to analyze CCR content. Ni and V content were analyzed on Optima 7300DV inductively coupled plasma-optical emission spectrometer (PerkinElmer company). The sum of Ni and V content is considered as the metal (M)content.

Table 2 Main properties of the catalysts used in RHT tests

3 Kinetic Model

3.1 Chemical reactions and model equations

The elements in residue and its hydrotreated oil are mainly C, H, S, and N, and the sum of C and H content is usually more than 95%. In the paper, the content of hydrocarbons were normalized after removing S and N,and is considered as the sum of C and H. Meanwhile, the hydrocarbons contained in residue and its hydrotreated oil are re-divided and abstractly defined as consisting of CH and CH2groups which are considered as pseudocomponents. The CH pseudo-component represents the unsaturated hydrocarbons with a C/H atom ratio of 1:1,and the CH2pseudo-component represents the saturated hydrocarbons with a C/H atom ratio of 1:2. According to the de finition, the weight content of CH and CH2can be calculated by the C/H atom ratio or the content of C and H. For example, when the C/H atom ratio is 1.6, which is represented by the symbol of CH1.6, and the molar fraction of CH and CH2is 0.6 and 0.4 respectively. Thus,the weight content of CH and CH2can be calculated and the formulas are derived as follows:

After hydrotreating, the unsaturated hydrocarbons can be saturated with H2through reactions, such as hydrogenation reaction of asphaltenes (HDAs), HDCCR,etc. According to the new definition of CH and CH2,the hydrogenation of CH to CH2(HDCH) can represent the hydrogenation of unsaturated hydrocarbon. Beside HDCH reaction, other reactions of RHT mainly include HDS, HDN, HDCCR, HDM, HDNi, HDV, and so on.During RHT, all reactions occur simultaneously in two reaction sections, section I and section II. Figure 2 shows the reaction pathways of main reactions that occur in section I and section II. As shown in Figure 2, the S-and N- compounds can be hydrogenated by H2to form H2S and NH3, separately. The CCR can be transferred to the hydrocarbons with a high H/C ratio (H-CCR) by HDCCR reaction. The metal-compound (R-M), such as Ni-compound (R-Ni) and V-compound (R-V), can react with H2and H2S to form metal sul fides (M-S), including Ni sul fide (Ni-S) and V sul fide (V-S).

Figure 2 The reaction pathways of main reactions in RHT

The simplified chemical reaction equations in RHT are listed as follows.

Based on the chemical reactions, the differential equation of HDCH, HDS, HDN, HDCCR, HDM, HDNi, and HDV reactions in the same form is established and listed as follows.

After integrating the differential equations, the equations are listed as follows.

3.2 Process for solving the model parameters

On the basis of model equations (10) — (12) and test data, the model parameters (such as n, α, A0, E) can be solved. The processes of solving model parameters are the same for the two reaction sections. Figure 3 shows the process chart of calculating model parameters. As shown in Figure 3, the reaction order (n) can be calculated from the test data under various LHSV. The p-reaction index(α) is obtained from the test data under various p. The A0and E are derived from the test data under various T.For all calculations, the goodness-to- fit (R2) is considered as the target value for model convergence. The closer the R2is near to 1, the more accurate the calculation is.Table 3 lists the obtained values of model parameters for the two reaction sections. The HDM catalyst has lower hydrogenation activity than the HDS catalyst due to its relatively low active metal amount. Thus, the HDCH and HDN activation energy of HDM catalyst is relatively higher than that of HDS catalyst, as shown in Table 3. The S compounds with high reactivity firstly react on HDM catalyst instead of HDS catalyst, and the HDS activation energy of HDM catalyst is lower than that of HDS catalyst. With the same reason as the HDS reaction, the HDCCR, HDM, HDNi, and HDV activation energy of HDM and HDS catalysts shows the similar performance to the HDS reaction. Meanwhile, the pore size of HDM catalyst is larger than HDS catalyst, and the diffusion resistance of compounds entering the channel of HDM catalyst is smaller than that of HDS catalyst. This is another reason showing that why the HDS, HDCCR,HDM, HDNi and HDV activation energy of HDM catalyst are smaller than HDS catalyst.

Figure 3 Calculation process chart of model parameters

3.3 Accuracy of the kinetic model

Before applying the kinetic model, it is necessary to make sure whether the established kinetic model can express the performance of HDCH, HDS, HDN, HDCCR, HDM,HDNi, and HDV reactions. The calculated content of CH, S, N, CCR, M, Ni, and V in hydrotreated residue can be obtained on basis of the above kinetic model.By comparing the calculated value with the measured value, the established kinetic model can be verified clearly. Figures 4―10 show the calculated conversions vs. the measured conversions for HDCH, HDS, HDN,HDCCR, HDM, HDNi, and HDV reactions, separately.The conversion was calculated on the basis of the equation: “Conversion = (Content in residue - Content in hydrotreated residue) /Content in residue”. As shown in Figures 4―10, the calculated conversions match well with the measured conversions. Thus, the established kinetic model has a high accuracy.

Table 3 Model parameters of the two reaction sections, section I and section II.

Figure 4 The calculated HDCH conversion vs. the measured HDCH conversion

Figure 5 The calculated HDS conversion vs. the measured HDS conversion

Figure 6 The calculated HDN conversion vs. the measured HDN conversion

3.4 Prediction using the kinetic model

The established kinetic model can be considered as a base case to predict the CH, S, N, CCR, M, Ni, and V content in the product. Figure 11a shows the predicted HDCH conversion versus the measured value. The two data match well with each other. Based on the CH value, the H content in hydrotreated residue can be obtained. Figure 11b shows the predicted H content versus the measured data. The predicted H content can match well with the measured data, so the HDCH reaction kinetic model can be firstly and efficiently employed to monitor the H change in RHT.

Figure 7 The calculated HDCCR conversion vs. the measured HDCCR conversion

Figure 8 The calculated HDM conversion vs. the measured HDM conversion

Figure 9 The calculated HDNi conversion vs. the measured HDNi conversion

Figure 10 The calculated HDV conversion vs. the measured HDV conversion

After transferring the content to conversion, the predicted conversions versus the measured conversions for HDS,HDN, HDCCR, HDM, HDNi, and HDV reactions are shown in Figure 12a―12f, separately. As shown in Figure 12a, the deviation of HDS conversion between the predicted value and the measured value is very small, so the HDS kinetic model is very accurate. And the deviation of HDN, HDCCR, HDM, HDNi, and HDV conversions between the predicted value and the measured value is a little bit bigger as shown in Figure 12b-12f. So the HDN,HDCCR, HDM, HDNi, and HDV kinetic models also can be applied to predict the content in product, which is slightly less accurate than the HDS kinetic model.

Figure 11 (a) The predicted HDCH conversion vs. the measured HDCH conversion,(b) the predicted H content vs. the measured H content

Figure 12 The predicted conversions of HDS, HDN, HDCCR, HDM, HDNi, and HDV vs. their measured conversions(a) HDS; (b) HDN; (c) HDCCR; (d) HDM; (e) HDNi; (f) HDV

4 Conclusions

Modeling the kinetics of RHT is essential for predicting the properties of the hydrotreated oil. The work is done in the paper, and the conclusions are described as follows.

(1) Hydrocarbons contained in residue and its hydrotreated oil can be re-divided into two virtual molecular-group components, CH and CH2. The kinetic model of HDCH reaction can be firstly and efficiently employed to monitor the H change in RHT.

(2) On basis of the test data, the model parameters can be calculated with goodness-of-fit (R2) as the target value.The calculated data fit well with the measured data, so the obtained kinetic model about HDCH, HDS, HDN,HDCCR, HDM, HDNi, and HDV reactions has high accuracy.

(3) The kinetic model can be considered as the base case to predict the H, S, N, CCR, M, Ni, and V content in hydrotreated oil. All deviations between the predicted value and the measured value are small. The deviation of HDS conversion between the predicted value and the measured value is the smallest among the reactions in RHT.

Nomenclature

A0——Pre-exponential factor;

C——The mass fraction of CH, S, N, CCR, M, Ni, and V in hydrotreated residue, %;

Cin——The mass fraction of CH, S, N, CCR, M, Ni, and V in residue, %;

Cout——The mass fraction of CH, S, N, CCR, M, Ni, and V in middle sample or hydrotreated residue, %;

E——Activation energy, kJ/mol;

k——Apparent rate constant of chemical reaction, and k = A0e-E/RT;

LHSV——Liquid hourly space velocity, h-1;

n——Reaction order;

p0——A unit of pressure, p0=1.0 MPa;

p——Partial pressure of hydrogen, MPa;

R——Avogadro’s number, R = 8.314 J/(mol·K);

R2——Goodness-to- fit, R2= 0—1;

T——Reaction temperature, K;

t——Residence time (t = 1/LHSV), h;

wC——The mass fraction of C in residue or hydrotreated residue, %;

wCH——The mass fraction of CH in residue or hydrotreated residue, %;

wCH2——The mass fraction of CH2in residue or hydrotreated residue, %;

wH——The mass fraction of H in residue or hydrotreated residue, %;

α——Reaction index of pressure.

Acknowledgments:This work was financially supported by the SINOPEC Research Program (Grant KL20009 and 118015-2).