Effects of Hydrogen Partial Pressure on ... - ACS Publications

Jan 14, 2010 - ... in a micro trickle-bed reactor, using a commercial NiMo/γ-Al2O3 catalyst. ... No clear trend was observed for the effect of H2 pp ...
1 downloads 0 Views 3MB Size
Energy Fuels 2010, 24, 772–784 Published on Web 01/14/2010

: DOI:10.1021/ef9010115

Effects of Hydrogen Partial Pressure on Hydrotreating of Heavy Gas Oil Derived from Oil-Sands Bitumen: Experimental and Kinetics M. Mapiour,† V. Sundaramurthy,†,‡ A. K. Dalai,*,† and J. Adjaye§ † Catalysis and Chemical Reaction Engineering Laboratories, Department of Chemical Engineering, University of Saskatchewan, Saskatoon, Saskatchewan S7N 5A9, Canada and §Syncrude Edmonton Research Centre, Edmonton, Alberta T6N 1H4, Canada. ‡ Present address: Bioprocessing Unit, Agriculture, Biotechnology and Food Division, Saskatchewan Research Council, Saskatoon, Saskatchewan S7N 2X8, Canada.

Received September 10, 2009. Revised Manuscript Received December 17, 2009

The effect of hydrogen partial pressure (H2 pp) on hydrotreating conversions, feed vaporization, H2 dissolution, and H2 consumption was studied in a micro trickle-bed reactor, using a commercial NiMo/ γ-Al2O3 catalyst. Heavy gas oil (HGO) from Athabasca bitumen was used as feed. The H2 pp level was set inside the reactor by means of manipulating other operating variables, namely, H2 purity, pressure, gas/oil ratio, liquid hourly space velocity (LHSV), and temperature. Their ranges were as follows: 75-100 vol % (with the rest methane), 7-11 MPa, 400-1200, 0.65-2 h-1, and 360-400 °C, respectively. HYSYS was used to determine the inlet and outlet H2 pp. The results show that hydrodenitrogenation (HDN) and hydrodearomatization (HDA) are significantly more affected by H2 pp than hydrodesulfurization (HDS), with HDN being the most affected. Moreover, it was observed that H2 dissolution and H2 consumption increase with increasing H2 pp. No clear trend was observed for the effect of H2 pp on feed vaporization. Kinetic studies of HDS, HDN, and HDA were performed using the power law model, multi-parameter model, and Langmuir-Hinshelwood-type (L-H) model, and the prediction abilities of the resultant models were tested. It was determined that, while the multi-parameter model yielded better prediction, the L-H model had an advantage in that it took a lesser number of experimental data to determine its parameters. The prediction ability of the power law was not tested because it excludes many operating variables.

pp on hydrotreating conversions. It has been qualitatively reported that pressure, H2 purity, and gas/oil ratio have greater influences on H2 pp in comparison to temperature and LHSV.6 Donald et al.4 also argued that, for an application such as hydrotreating, there is a great difference between the inlet and outlet H2 pp. Furthermore, the authors claimed that outlet H2 pp is more important because (i) “outlet conditions reflect the catalyst’s last chance to cause feedstock change”, (ii) “outlet conditions more nearly approximate average conditions throughout the catalyst bed”, and “it is the conservative approach to analyzing a given situation”. Independent variables (temperature, pressure, gas/oil ratio, H2 purity, and LHSV) affect H2 pp by influencing factors such as liquid feed vaporization, H2 dissolution, and H2 consumption. Amounts of feed vaporization and dissolved H2 can be estimated or calculated using vapor/liquid equilibrium.4,7,8 The authenticity of the use of vapor/liquid equilibrium calculations to determine feed vaporization, dissolved H2, and H2 pp has been verified by many authors.9-11 When the

1. Introduction The principal operating variables in hydrotreating are temperature, hydrogen partial pressure (H2 pp), and liquid space velocity (LHSV).1-3 Donald et al.4 argued that H2 pp is often a misunderstood variable for two reasons: (1) its relatively important effect on hydroprocessing conversions and (2) the appropriate mathematical approach required for its determination. The authors have adequately discussed the latter; however, there are limited reports on the former. H2 pp is usually assumed to be equal to the system pressure. While in some cases this approach may be a reasonable approximation, it is not entirely a sound approach. As stated by Dalton’s law,5 H2 pp is a product of the vapor-phase mole fraction. In hydrotreating applications, the vapor-phase mole fraction itself is a function of the temperature, pressure, gas/oil ratio, H2 purity, and LHSV.2-4 Thus, unlike temperature, pressure, gas/oil ratio, H2 purity, and LHSV, H2 pp is not an independent variable. It is therefore important to observe and even quantify the effects of the independent variables (temperature, pressure, gas/oil ratio, H2 purity, and LHSV) on H2 pp before focusing on the effect of H2

(6) Antos, G. J.; Aitani, A. M. Catalytic Naphtha Reforming; CRC Press: Boca Raton, FL, 2004. (7) Alvarez, A.; Ancheyta, J. Appl. Catal., A 2008, 351, 148. (8) Mu~ noz, J. A. D.; Alvarez, A.; Ancheyta, J.; Rodrı´ guez, M. A.; Marroquı´ n, G. Catal. Today 2005, 109, 214. (9) Ramanujam, S.; Leipziger, S.; Weil, S. A. Vapor-liquid equilibrium for a hydrogen/simulated coal-derived liquid system. Ind. Eng. Chem. Process Des. Dev. 1985, 24, 364–368. (10) Wilson, G. M.; Johnston, R. H.; Hwang, S.-C.; Tsonopoulos, C. Volatility of coal liquids at high temperatures and pressures. Ind. Eng. Chem. Process Des. Dev. 1981, 20, 94–104. (11) Lal, D.; Otto, F. D.; Mather, A. E. Solubility of hydrogen in Athabasca bitumen. Fuel 1999, 78, 1437–1441.

*To whom correspondence should be addressed. Telephone: þ1-306966-4771. Fax: þ1-306-966-4777. E-mail: [email protected]. (1) Gary, J. H.; Handwerk, G. E.; Kaiser, M. J. Petroleum Refining: Technology and Economics, 5th ed.; CRC Press: Boca Raton, FL, 2007. (2) Speight, J. G. The Desulfurization of Heavy Oils and Residua; Marcel Dekker, Inc.: New York, 1981. (3) Speight, J. G. The Desulfurization of Heavy Oils and Residua; Marcel Dekker, Inc.: New York, 2000. (4) McCulloch, C. D.; Roeder, R. A. Hydrocarbon Process. 1976, 55 (2), 81. (5) Fogler, H. S. Elements of Chemical Reaction Engineering, 3rd ed.; Prentice Hall Professional Technical Reference (PTR): New York, 1999. r 2010 American Chemical Society

772

pubs.acs.org/EF

Energy Fuels 2010, 24, 772–784

: DOI:10.1021/ef9010115

Mapiour et al.

experimental results were compared to those determined using thermodynamic models, e.g., the Peng-Robenson equation, the authors concluded that the two set of results were in good agreement. For instance, Lal et al.11 measured H2 in Athabasca bitumen using a batch autoclave at a temperature range of 50-300 °C and H2 pp up to 24.8 MPa. The authors concluded that thermodynamic equations, namely, the PengRobinson equation (original and modified), the SoaveRedlich-Kwong equation, and the Grayson-Streed method, yielded accurate results. The total H2 consumption is determined by estimating the chemical H2 consumption, i.e., H2 consumed during hydrotreating and hydrocracking of the feed. The chemical H2 consumption must then be added to dissolved H2, providing a small allowance for leaks, etc.4,12 Generally, little hydrocracking takes place at normal hydrotreating conditions4 and, therefore, can be assumed negligible for calculation purposes. The first objective of this work is to study the effect of the independent variables (temperature, pressure, gas/oil ratio, H2 purity, and LHSV) on feed vaporization, H2 dissolution, H2 consumption, and inlet and outlet H2 pp. The second objective of this work is to observe the effect of the H2 pp on hydrodesulfurization (HDS), hydrodenitrogenation (HDN), and hydrodearomatization (HDA) of heavy gas oil (HGO). The final objective of this work is to develop kinetic expressions for HDS, HDN, and HDA.

(GC)-simulated distillation (model CP3800, Varian, Palo Alto, CA) following the standard procedure ASTM D2887. To determine the inlet H2 pp, the feed boiling range data were fed into HYSYS along with the gas compositions and flow rate information. To determine the outlet H2 pp, the boiling range data of the product were fed into HYSYS along with the gaseous compositions and flow rate information. Notice that, at the outlet conditions, the gaseous compositions must account for the produced H2S and NH3 and the decrease in H2 because of H2 consumption. H2S and NH3 are produced as a result of HDS and HDN, respectively. HYSYS can be set such that it reports vapor/liquid equilibrium as mole fractions. Inlet or outlet H2 pp is then calculated by multiplying the H2 mole fraction by the system pressure. 2.3. Determination of Total H2 Consumption. As mentioned in the Introduction, the total H2 consumption is a summation of chemical H2 consumption and dissolved H2, assuming that any mechanical H2 loss and hydrocracking are negligible. An equation that may aid total H2 consumption calculation is4,15 total H2 consumption ¼ chemical H2 consumption þ dissolved H2 ðscf=bblÞ

ð1Þ 4,7,8

Dissolved H2 is determined from vapor/liquid equilibrium ½ðCA Þf - ðCA Þp   densityfeed chemical H2 consumption ¼ 100  2  12  379 þ H0 H2 S þ H0 NH3 þ HH2 S þ HNH3

2. Experimental Section H0 H2 S ≈ HH2 S ¼

2.1. Hydrotreating of HGO. A trickle-bed reactor was used in this work. The reactor inner diameter and length were 10 and 240 mm, respectively. The reactor was loaded with 5 g of commercial NiMo/γ-Al2O3 catalyst, with an average diameter of 1.5 mm, and was diluted with SiC to improve the hydrodynamics of the system and help achieve a near-plug-flow behavior.4 The catalyst bed was heated using an electric furnace, and sulfidation was carried out using butanethiol solution. Next, the experiments proceeded as designed, and samples were collected and analyzed for sulfur, nitrogen, and aromatic conversions. The analyses were accomplished by employing the combustion/fluorescence technique (ASTM D5463) for sulfur, the combustion/chemiluminescence technique (ASTM D4629) for nitrogen, and gated-decoupled 13C nuclear magnetic resonance (NMR) for aromatic contents. The HGO feedstock, from oil-sands bitumen supplied by Syncrude Canada Ltd., was analyzed and found to contain about 4.1 wt % S, 0.32 wt % N, and 31.5 wt % aromatics. Detailed experimental information is provided elsewhere.13 2.2. Determination of H2 pp Using HYSYS. Two of the thermodynamic models that are suitable for vapor/liquid equilibrium calculations for hydrotreating applications are the Grayson-Streed and Chao-Seader eqautions.14 The ChaoSeader equation is good for temperatures below 530 K (257 °C), whereas the Grayson-Streed equation (an extension of the Chao-Seader equation) can be used for higher temperatures and pressures as high as 4700 K (4427 °C) and 200 bar (20 MPa). The Grayson-Streed equation is therefore preferred4 and, as such, was used in this study. Both models are available in Aspen HYSYS 2006. Boiling range distributions of the feed and the liquid products were determined using gas chromatography

H0 NH3 ≈ HNH3 ¼

½ðSÞf - ðSÞp   densityfeed 100  32

ð1aÞ  379

½ðNÞf - ðNÞp   densityfeed  379 100  14

ð1bÞ ð1cÞ

where CA, S, and N are aromatic carbon, sulfur, and nitrogen contents (wt %), respectively, subscripts f and P are feed and products, respectively, H0 H2S and H0 NH3 are the amount of H2 necessary to form a hydrocarbon during HDS and HDN (scf/bbl), respectively, HH2S and HNH3 are the H2 content of H2S and NH3 in the product gas (scf/bbl), respectively, 379 is the number of standard cubic feet in a mole of an ideal gas (scf/ mol), and densityfeed is 346 lb/bbl. The unit for H2 consumption is standard cubic feet per barrel (scf/bbl) (Note that the denominators 100  2  12, 100  32, and 100  14 in eqs 1a, 1b, and 1c, respectively, are explained as follows: The factors 12, 32, and 14 are the molecular weights of carbon, sulfur, and nitrogen, respectively. The 100 factor in the denominator is C, S, and N in weight percent. The factor of 2 in eq 1a is because the aromaticity is measured between two carbon atoms).

3. Results and Discussion In this section, the effects of the independent variables (pressure, temperature, LHSV, gas/oil, and H2 purity) on feed vaporization, hydrogen dissolution, hydrogen consumption, and inlet and outlet H2 pp were studied. Moreover, the correlations between inlet and outlet H2 pp and hydrotreating conversions were studied. As mentioned in the Introduction, H2 pp is significantly more affected by pressure, gas/oil ratio, and H2 purity than temperature and LHSV.6 Therefore, these three important variables were used in the central composite inscribed method (using Expert design 6.0.1) to design the experiments. Their ranges were as follows: pressure was

(12) Lee, S. Encyclopedia of Chemical Processing; CRC Press: Boca Raton, FL, 2005; Vol. 1. (13) Mapiour, M.; Sundaramurthy, V.; Dalai, A. K.; Adjaye, J. Effect of hydrogen purity on hydroprocessing of heavy gas oil derived from oilsands bitumen. Energy and Fuels 2009, 23, 2129–2135. (14) Sinnott, R. K.; Coulson, J. M.; Richardson, J. F. Coulson and Richardson’s Chemical Engineering, 4th ed.; Butterworth-Heinemann: Woburn, MA, 2005.

(15) Hisamitsu, T.; Shite, Y.; Maruyama, F.; Yamane, M.; Satomi, Y.; Ozaki, H. Bull. Jpn. Pet. Inst. 1976, 18 (2), 146.

773

Energy Fuels 2010, 24, 772–784

: DOI:10.1021/ef9010115

Mapiour et al. Table 1. R2 Statistics for the Models

7-11 MPa; H2 purity was 75-100 vol %; and gas/oil ratio was 400-1200 (with the rest methane). In our pervious work, it was found that methane was inert toward commercial Ni-Mo/γ-alumina under these experimental conditions.13 The temperature and LHSV were kept constant at 380 °C and 1 h-1, respectively. In a separate set of experiments, the effects of the temperature and LHSV on feed vaporization, hydrogen dissolution, hydrogen consumption, and H2 pp were studied. The temperature and LHSV ranges were 360-400 °C and 0.65-2 h-1, respectively. H2 purity, pressure, and gas/oil ratio were kept constant at 100%, 9 MPa, and 800, respectively. LHSV was kept constant at 1 h-1 when the effect of the temperature was studied. The temperature was kept constant at 380 °C when the effect of the LHSV was studied. 3.1. Effect of the Pressure, H2 Purity, and Gas/Oil Ratio on Feed Vaporization, H2 Dissolution, H2 Consumption, and H2 pp. The regression analysis of experimental data generated the following equations:

ð2Þ

outlet H2 pp ¼ 36:589 - 0:339  purity - 4:965  pressure - 2:561  10 - 3  gas=oil - 4:653  10 - 6  gas=oil2 þ 0:0509  purity  pressure þ 1:326  10 - 3  pressure  gas=oil ð3Þ vaporized feed ¼ - 0:51728 - 9:53149  10 - 3  purity þ 0:14977  pressure þ 3:88345  10 - 3  gas=oil - 3:09359  10 - 4  pressure  gas=oil

R2

adjusted R2

predicted R2

vaporized feed dissolved H2 H2 consumption inlet H2 pp outlet H2 pp

0.8793 0.9995 0.9220 0.9994 0.9513

0.8471 0.9992 0.9074 0.9993 0.9288

0.7311 0.9987 0.9074 0.9988 0.8002

significance of factors means that insignificant factors or interactions must be excluded from the model.16 Significance of the factors or the interactions are evaluated using the p value (probability value). When a p value of a factor or an interaction is greater than 0.05, it is certain at a 95% confidence level that that factor or interaction is insignificant and can therefore be excluded from the final mathematical model. The reduced models are presented in eqs 2-6. R2, a value that always falls between 0 and 1, is the relative predictive power of a model.16 The closer to 1 the R2, the better the model represents the experimental observations. However, note that by simply incorporating more factors or interactions R2 may be increased, while the predictive power of the model is not improved. Because of this shortcoming of R2, the use of adjusted R2 is advised. Adjusted R2 is a modification of R2, but unlike R2, it only increases when the newly included factor(s) or interaction(s) are significant.17 Another quantity is predicted R2. While R2 indicates how well the model fits the experimental data at hand, predicted R2 indicates how well the model predicts responses for new observations. The R2, adjusted R2, and predicted R2 values of the factors and interactions of the developed correlations are summarized in Table 1. To test the predictive ability of the generatered correlations, three experiments (at conditions that were different than those of the expreimental design used to generate the data for the correlations development) were conducted. In these three experiments, pressure, temperature, LHSV, and gas/oil ratio were kept constant at 9 MPa, 380 °C, 1 h-1, and 800, respectively, while H2 purity was varied as follows: 50, 80, and 90 vol % (with the rest methane). At these conditions, quantities such as inlet H2 pp, outlet H2 pp, vaporized feed, dissolved H2, and H2 consumption were experimentally determined and compared to those predicted by the correlations, as shown in Table 2. The maximum percentage differences for inlet and outlet H2 pp were 2.2 and 15.3%, respectively. However, this relatively high percentage difference, 15.3%, was observed at the extrapolated experimental condition, i.e., at 50% H2 purity. When the comparisons were performed solely within the range of conditions used to develop the correlations, the maximum percentage differences for inlet and outlet H2 pp were 0.4 and 0.9%, respectively. Therefore, these correlations are only valid within the experimental ranges, and extrapolation is not adequate. A normalized parity plot that compares experimental values to predicted values for inlet H2 pp, outlet H2 pp, vaporized feed, dissolved H2, and H2 consumption is presented in Figure 1. Panels a and b of Figure 2 are three-dimensional plots of the effects of pressure, H2 purity, and gas/oil ratio on inlet H2 pp. This figure shows that inlet H2 pp increases with increasing pressure and H2 purity but is not affected by the gas/oil ratio. The effect of the pressure and H2 purity are explained

inlet H2 pp ¼ 0:149 - 2:401  10 - 3  purity þ 7:289  10 - 3  pressure þ 9:889  10 - 3  purity  pressure

model

ð4Þ

dissolved H2 ¼ - 23:95817 þ 0:67529  purity - 3:56483  pressure þ 2:15964  10 - 3  gas=oil - 3:94802  10 - 3  purity2 þ 0:23328  pressure2 þ 0:11314  purity  pressure - 1:69706  10 - 4  purity  gas=oil ð5Þ H2 consumption ¼ þ 336:13279 þ 5:67696  purity þ 33:99691  pressure þ 0:084567  gas=oil ð6Þ Two statistical tests (test of significance of factors and R2 test) were used to evaluate how well the experimantal data were represented by the correlations. Use of the test of

(16) Lazic, Z. R. Design of Experiments in Chemical Engineering, 1st ed.; Wiley-VCH Verlag GmbH: Weinheim, Germany, 2004. (17) Montgomery, D. C. Design and Analysis of Experiments, 4th ed.; John Wiley and Sons: New York, 1997.

774

Energy Fuels 2010, 24, 772–784

: DOI:10.1021/ef9010115

Mapiour et al.

Table 2. Comparison between the Predicted and Observed Valuesa inlet H2 pp (MPa)

outlet H2 pp (MPa)

vaporized feed (g/h)

dissolved H2 (scf/bbl)

H2 consumption (scf/bbl)

purity (vol %)

a

b

a

b

a

b

a

b

a

b

50 80 90

4.4 7.1 8.0

4.5 7.1 8.0

2.4 6.0 7.1

2.8 5.9 7.1

1.24 0.95 0.85

1.08 0.95 0.90

32.6 63.9 72.8

26.8 56.0 67.8

992 1196 1228

993 1164 1220

a

a, predicted by models; b, observed experimentally.

Figure 1. Normalized parity plot for inlet H2 pp, outlet H2 pp, vaporized feed, dissolved H2, and H2 consumption. The values are normalized to the maximum reading in each category.

Figure 3. Surface response of the effect of the pressure, H2 purity, and gas/oil ratio on outlet H2 pp. The temperature and LHSV were constant at 380 °C and 1 h-1, respectively.

Panels a and b of Figure 3 show the effect of the pressure, H2 purity, and gas/oil ratio on outlet H2 pp. Outlet H2 pp increases with increasing pressure, H2 purity, and gas/oil ratio. However, the enhancing effect of the gas/oil ratio tends to plateau at higher values (approximately 800 and above). To fully explain the effects of the pressure, H2 purity, and gas/oil ratio on inlet and outlet H2 pp one must first study their effects on the factors that influence inlet and outlet H2 pp, namely, feed vaporization, H2 dissolution, and H2 consumption. Figures 4-6 depict the effects of the pressure, H2 purity, and gas/oil ratio on feed vaporization, H2 dissolution, and H2 consumption, respectively. In panels a and b of Figures 4 and 5, it can be observed that increasing the pressure and H2 purity and decreasing the gas/oil ratio result in increases in H2 dissolution and decreases in feed vaporization. The effect of the pressure on feed vaporization can be explained by Le Chatelier’s principle. As the pressure increases, the equilibrium counters this change by converting more gas into liquid because liquid takes less space.16 The effect of the pressure on

Figure 2. Surface response of the effect of the pressure, H2 purity, and gas/oil ratio on inlet H2 pp. The temperature and LHSV were constant at 380 °C and 1 h-1, respectively.

by Dalton’s law. The gas/oil ratio does not affect the inlet H2 pp because, as the amount of treat gas increases, the amount of feed vaporization increases as well; thus, vapor composition stays the same.4 775

Energy Fuels 2010, 24, 772–784

: DOI:10.1021/ef9010115

Mapiour et al.

Figure 5. Surface response of the effect of the pressure, H2 purity, and gas/oil ratio on dissolved H2. Figure 4. Surface response of the effect of the pressure, H2 purity, and gas/oil ratio on vaporized feed. The total liquid flow is 5 g/h.

liquid volume for H2 to dissolve. Consequently, the amount of the dissolved H2 decreases. Similar results were observed by McCulloch et al.4 Increasing the H2 purity means that methane is replaced with H2. Wilson et al.,20 in an attempt to examine volatility of coal liquids, determined the interactions of CH4 and H2 with coal liquids. The authors found that the binary interaction constants for CH4/coal liquids and H2/coal liquids were 0.08 and 0.25, respectively. It is therefore reasonable to expect that the interaction between H2 and other heavy hydrocarbons, such as HGO, would be higher than that of CH4. On the basis of this assumption, H2/HGO binary mixing is expected to exhibit more negative deviation from Raoult’s law; i.e., the molecules in the binary mixture have a lower escaping tendency. Hence, a lower feed vaporization was observed as the H2 purity was increased. Increasing the H2 purity also led to an increase in H2 pp. As explained by Henry’s law, increases in the partial pressure of a gas lead to increases in its dissolution. Hence, increases in a H2 pp, caused by increasing the H2 purity, led to increases in H2 dissolution. Panels a and b of Figure 6 show that H2 consumption increases with increasing pressure, H2 purity, and gas/oil ratio. In general, increasing pressure, H2 purity, and gas/oil ratio enhance hydrotreating conversions, leading to higher H2 consumption.

H2 dissolution can be explained by Henry’s law. This law states that the concentration of dissolved gas is directly proportional to its partial pressure at a constant temperature. Therefore, if the pressure is increased, causing an increase in the partial pressure of the gas, the amount of the dissolved gas increases.18 Hence, an increase in H2 dissolution is observed. The effect of the decreasing gas/oil ratio on feed vaporization can be explained in terms of the mass-transfer driving force. At a constant oil flow rate, the gas/oil ratio is decreased by a decreasing gas flow rate. As the gas flow rate is decreased, less vaporization takes place because of the decrease in the mass-transfer driving force.19 Observed results of the effect of the gas/oil ratio on H2 dissolution is counterintuitive; one would expect that, as the gas/oil ratio increases, more H2 would be dissolved. Increasing the gas/oil ratio increases H2 pp and, according to Henry’s law, forces additional H2 dissolution. However, the results suggest that increasing the gas/oil ratio leads to decreases in H2 dissolution. The reason for this is that both feed vaporization and H2 dissolution occur simultaneously and, as the gas/oil ratio is increased, more liquid feed is vaporized, leaving a smaller (18) Tro, N. J. Introductory Chemistry Essentials; Pearson/Prentice Hall: New York, 2009. (19) Wankat, P. C. Separation Process Engineering; Prentice Hall: New York, 2007.

(20) Wilson, G. M.; Johnston, R. H.; Hwang, S.-C.; Tsonopoulos, C. Ind. Eng. Chem. Process Des. Dev. 1981, 20 (1), 94.

776

Energy Fuels 2010, 24, 772–784

: DOI:10.1021/ef9010115

Mapiour et al. Table 3. Results of the Effects of the Temperature and LHSV on Feed Vaporization, H2 Dissolution, H2 Consumption, and Inlet and Outlet H2 pp H2 consumption (scf/bbl)

inlet H2 pp (MPa)

outlet H2 pp (MPa)

48 81 129 173

1358 1352 1249 1170

8.9 8.9 8.9 8.9

8.2 8.2 8.3 8.3

81 81 81 77 74

1234 1286 1352 1319 1285

8.9 8.9 8.9 8.8 8.8

8.4 8.3 8.2 8.1 8.1

vaporized dissolved feed at H2 at outlet (scf/bbl) outlet (g/h) LHSV (h-1) 0.65 0.73 1 0.84 1.5 0.91 2 1.00 temperature (°C) 360 0.52 370 0.67 380 0.84 390 1.15 400 1.40

Figure 6. Surface response of the effect of the pressure, H2 purity, and gas/oil ratio on outlet H2 consumption.

Figure 7. Effect of the temperature on HDA. Pressure, LHSV, gas/ oil ratio, and H2 purity were 9 MPa, 1 h-1, 800, and 100%, respectively.

3.2. Effect of the Temperature and LHSV on Feed Vaporization, H2 Dissolution, H2 Consumption, and H2 pp. The results of the effects of the temperature and LHSV on feed vaporization, hydrogen dissolution, hydrogen consumption, and inlet and outlet H2 pp are given in Table 3. The results show that increasing the temperature leads to increases in feed vaporization and very slight decreases in H2 dissolution. Increasing the temperature causes increases in the kinetic energy of the species, leading to increases in feed vaporization. Also, the higher the temperature, the more a gas expands and the harder it is for a gas to dissolve in a liquid. As a result, a decrease in H2 dissolution was observed.21 Increasing the LHSV leads to increases in feed vaporization and H2 dissolution. An increase in the LHSV corresponds to an increase in the feed rate (liquid flow rate). According to Raoult’s law, as the mole fraction of a component in a solution increases, its partial pressure does as well and, consequently, its escaping tendency increases. Therefore, as more liquid feed is introduced into the reactor, more of it evaporates. Also, when the liquid flow rate is increased, there is more liquid volume for H2 to dissolve. Hence, increases in H2 dissolution were observed. H2 consumption decreases with increasing the LHSV. Increasing the LHSV results in decreases in hydrotreating conversions because the residence time is reduced. As a result, there is a decrease in H2 consumption. H2 consumption passes through a maximum with respect to the tempera-

ture. The reason for this is that, shown in Figure 7, HDA passes through a maximum as the temperature is gradually increased from 360 to 400 °C. Inlet and outlet H2 pp do not vary significantly with changes in the temperature or LHSV (see Table 3). 3.3. Effect of the H2 pp on H2 Consumption, Dissolved H2, and Feed Vaporization. It is important to look at the above factors because of their influences on H2 pp.2,3 Hence, an attempt was made to correlate H2 pp and these factors. The results are shown in Figure 8. From Figure 8, it is evident that both hydrogen consumption and hydrogen dissolution increase with increasing H2 pp. The reason for this is that increasing H2 pp generally improves hydrotreating activities, thus, increasing H2 consumption. As explained by Henry’s law, increases in the partial pressure of a gas result in increases in its dissolution. Hence, increasing the H2 pp brings about increases in H2 dissolution. No clear correlation between the H2 pp and feed vaporization was observed. 3.4. Effect of the H2 pp on Hydrotreating Conversions. Figures 9 and 10 show that HDN and HDA were significantly more affected by H2 pp than HDS. This fact is often explained in terms of the HDS mechanism versus the HDN mechanism.22,23 Hydrogenation of a N-containing ring occurs prior to C-N bond scission. Thus, the HDN rate can be affected by the equilibrium of N-ring hydrogenation

(21) Martin, A.; Bustamante, P. General principle of solubility and distribution. In Physical Pharmacy: Physical Chemical Principles in the Pharmaceutical Sciences, 4th ed.; Lippincott, Williams, and Wilkins: Philadelphia, PA, 1993; pp 212-236.

(22) Kabe, T.; Ishihara, A.; Qian, W. Hydrodesulfurization and Hydrodenitrogenation; Kodanacha Ltd.: Tokyo, Japan, 1999. (23) Fang, X. Shiyou Xuebao 1999, 15 (5), 6 (in Chinese).

777

Energy Fuels 2010, 24, 772–784

: DOI:10.1021/ef9010115

Mapiour et al. Table 4. Correlations between the H2 pp and Hydrotreating Conversionsa equations

R2

HDS HDN HDA

for Inlet H2 pp y = 0.2932x þ 92.694 y = 0.231x2 þ 2.675x þ 31.954 y = -0.2506x2 þ 6.5619x þ 16.192

0.0773 0.8018 0.7727

HDS HDN HDA

for Outlet H2 pp y = 0.2417x þ 93.374 y = 0.5883x2 - 1.9673x þ 53.48 y = -0.1759x2 þ 4.9757x þ 26.778

0.062 0.8414 0.8935

activity

a

Figure 8. Correlations between the H2 pp and H2 consumption and the H2 pp and dissolved H2.

y, hydrotreating conversion; x, H2 pp.

hydrogenation rate. Consequently, an increase in HDA conversion is observed as the H2 pp is raised. Simple correlations that relate hydrotreating conversions to inlet and outlet H2 pp were developed, and the results are given in Table 4. It can be seen in this table that the R2 for HDS is very small because, within the H2 pp range of study, HDS is not strongly affected by H2 pp for reasons discussed earlier in this section. To test the predictive ability of the generated correlations, three experiments were conducted at conditions that were not part of the experimental design used to generate the correlations and their conversions were compared to those predicted by the correlations (see Table 5). In these three experiments, pressure, temperature, LHSV, and gas/oil ratio were constant at 9 MPa, 380 °C, 1 h-1, and 800, respectively, while H2 purity was varied as follows: 50, 80, and 90 vol % (with the rest methane). It was determined that the maxmium precentage differences for HDN and HDA using inlet H2 pp correlations were 3 and 4%, respectively. They were 4 and 7%, resepectively, when outlet H2 pp correlations were used. The percentage differences for HDS were not determined because the R2 was too small. 3.5. Effect of H2S on Hydrotreating Conversion. H2S is generated during hydrotreating as a product of HDS reactions. Most studies reported that H2S inhibits hydrotreating activities,24-29 yet it is required to maintain the active chemical state of the catalyst.29 H2S inhibition is caused as H2S competes against organosulfur and organonitrogen for the same active sites on the catalyst. H2S generated inside a hydrotreater can have an equilibrium value as high as 5 mol % in the recycle gas.30 This concentration of H2S not only inhibits hydrotreating activities but also reduces H2 pp. Therefore, in practice, H2S is removed in the amine unit. Unfortunately, some H2S remains in the recycle stream and is fed into the hydrotreater30 along with other difficult to remove impurities, such as methane.31 In our previous work,13 it was determined that the only effect induced by the presence of methane was the decrease of H2 pp inside the reactor, which in turn led to decreases in

Figure 9. Effect of the inlet H2 pp on HDS, HDN, and HDA.

Figure 10. Effect of the outlet H2 pp on HDS, HDN, and HDA.

because N-ring hydrogenation occurs before nitrogen removal (hydrogenolysis). HDS does not always require hydrogenation; it can proceed via two possible mechanisms: (1) ring hydrogenation followed by hydrogenolysis or (2) direct hydrogenolysis. To understand the difference in the HDS mechanism versus that of the HDN mechanism, the bond energies of CdS, C-S, CdN, and C-N must be compared. The bond energies of CdS and C-S are the same, 536 kJ/ mol, and the bond energies of CdN and C-N are 615 and 389 kJ/mol, respectively.22 It is therefore energetically favorable to hydrogenate CdN to C-N before C-N bond scission, whereas for CdS and C-S, there is no particular preference.22 The mechanism for HDA is hydrogenation, and an increase in the H2 pp results in an enhancement of the

(24) Herbert, J.; Santes, V.; Cortez, M. T.; Zarate, R.; Dı´ az, L. Catal. Today 2005, 107-108, 559. (25) Girgis, M. J.; Gates, B. C. Ind. Eng. Chem. Res. 1991, 30, 2021. (26) Hanlon, R. T. Energy Fuels 1987, 1, 424. (27) Sie, S. T. Fuel Process. Technol. 1999, 61, 149. (28) Ancheyta-Juarez, J.; Aguilar-Rodrı´ guez, E.; Salazar-Sotelo, D.; Betancourt-Rivera, G.; Quiroz-Sosa, G. Stud. Surf. Sci. Catal. 1999, 127, 347. (29) Bej, K.; Dalai, A.; Adjaye, J. Energy Fuels 2001, 15, 377. (30) Gruia, A. Handbook of Petroleum Processing; Springer: New York: 2006; Chapter 8. (31) Turner, J.; Reisdorf, M. Hydrocarbon Process. 2004, March, 61–70.

778

Energy Fuels 2010, 24, 772–784

: DOI:10.1021/ef9010115

Mapiour et al.

Table 5. Comparison between the Predicted and Observed Values for the Correlations between the H2 pp and Hydrotreating Conversions determined from the models (%) determined experimentally (%)

using inlet H2 pp

using outlet H2 pp

purity (vol %)

HDS

HDN

HDA

HDS

HDN

HDA

HDS

HDN

HDA

50 80 90

92 95 96

38 64 71

38 48 49

94 95 95

46 63 68

41 50 53

94 95 95

52 62 69

42 53 56

Two of the commonly used correlations are the power law and Langmuir-Hinshelwood-type (L-H) models.29,34-37 Another effective but seldom used model is the multi-parameter kinetic model.38 In the following subsections, the kinetic studies for HDS, HDN, and HDA are discussed using these three correlations. 3.6.1. Power Law Model. This model is often preferred because of its simplicity.34 In comparison to the L-H or multi-parameter-type model, the power law model has fewer parameters that must be determined. Kinetic parameters that can be determined using this model are the apparent rate constant and reaction order

Table 6. Effect of Butanethiol Added to Feed on Hydrotreating Conversions

conversion (%) HDS HDN HAD

100 vol % H2 purity

80 vol % H2 purity

added butanethiol (wt %)

added butanethiol (wt %)

0

1

3

0

1

3

96.6 76.1 54.3

95.4 69.0 46.7

94.3 67.2 45.5

95.1 65.2 48.2

90.6 52.2 43.6

90.4 51.3 44.9

hydrotreating conversions. In this work, an effort was made to determine what takes place when both methane and high concentrations of H2S are present in the reactor. Because of the serious health hazards associated with direct handling of H2S, high concentrations of H2S were generated inside the reactor by adding different concentrations of buthanthiol to the HGO feed.29 Buthanthiol decomposes under the chosen reaction conditions into 1-butene and H2S.32 Two sets of experiments were conducted: one set with no methane in the gaseous stream (100% H2 purity) and another with 20 vol % methane and 80 vol % H2. In both sets, experiments were carried out at different concentrations of butanethiol, 0, 1, and 3 wt %, in the HGO feed (additions of 1 and 3 wt % butanethiol in the feed correspond to increases in H2S partial pressure in the reactor by 33 and 113 kPa, respectively; this is based on the assumption that 1 mol of butanethiol produces 1 mol of H2S under experimental conditions). The temperature, pressure, gas/oil ratio, and LHSV were kept constant at 380 °C, 9 MPa, 800, and 1 h-1, respectively. The results of the effect of H2S on hydrotreating conversions are presented in Table 6. The results show that all HDS, HDN, and HDA conversions decrease as the concentration of H2S is increased by adding butanethiol to the HGO feed. A notable point is that, as the butanethiol concentration was increased from 0 to 1 wt %, there were considerable decreases in HDS, HDN, and HDA conversions. However, as the butanethiol concentration was increased from 1 to 3 wt %, no major changes in HDS, HDN, and HDA conversions were observed. This may be interesting with regard to H2S removal from the recycled gas. As this finding suggests, there appears to be an optimal amount of H2S that has to be removed, beyond which no significant beneficial effects on the hydrotreating conversions are realized. 3.6. Kinetic Studies for HDS, HDN, and HDA. The effects of variables, such as temperature, pressure, LHSV, and gas/ oil ratio, on the hydrotreating performance of a catalyst can be predicted by a suitable kinetic expression or model.33,34

dC ¼ - kC n dt

ð7Þ

where C is the sulfur, nitrogen, or aromatics content, t is the residence, k is the apparent rate constant, and n is the reaction order. Its solutions are Cf - Cp ¼

ki LHSV

for n ¼ 0

  Cp ki ¼ for n ¼ 1 Cf LHSV   1 1 ki for n 6¼ 0; 1 - n - 1 ¼ ðn - 1Þ n-1 LHSV Cp Cf ln

ð7aÞ ð7bÞ ð7cÞ

where n is the reaction order, ki is the apparent rate constant for species “i”, Cp is the concentration of the product (wt %), Cf is the concentration of the reactant (wt %), and LHSV is the liquid hourly space velocity. The activation energies can then be determined using the Arrhenius equation, eq 8 ki ðTÞ ¼ ko e - E=RT

ð8Þ

where ko is the Arrhenius constant, E is the activation energy (kJ/mol), R is the gas constant (kJ mol-1 K-1), and T is the temperature (K). 3.6.2. Power Law Analysis of HDS, HDN, and HDA. The power law model has been used in many studies of kinetics modeling of HDS and HDN. However, literature information on the kinetic studies of the HDA of real feed, such as petroleum and synthetic middle distillate, are very scarce, possibly because of the complexity of the reactions.35 The reaction orders for the HDS, HDN, and HDA were determined using the power law model, and the results are (35) Owusu-Boakye, A. Two-stage aromatics hydrogenation of bitumen-derived light gas oil. Master’s Thesis, University of Saskatchewan, Saskatoon, Saskatchewan, Canada, 2005. (36) Botchwey, C.; Dalai, A.; Adjaye, J. Energy Fuels 2003, 17, 1372. (37) Botchwey, C.; Dalai, A.; Adjaye, J. Int. J. Chem. React. Eng. 2006, 4, No. A20. (38) Ai-jun, D.; Xu, C.; Lin, S.; Chung, K. H. J. Chem. Eng. Chin. Univ. 2005, 19 (5), 762.

(32) Horie, O.; Nishino, J.; Amano, A. Int. J. Chem. Kinet. 1978, 10, 1043. (33) Knudsen, K. G.; Cooper, B. H.; Topsoe, H. Appl. Catal., A 1999, 189, 205. (34) Ferdous, D.; Dalai, A. K.; Adjaye, J. Ind. Eng. Chem. Res. 2006, 45, 544.

779

Energy Fuels 2010, 24, 772–784

: DOI:10.1021/ef9010115

Mapiour et al.

Table 7. Results of Activation Energies and Reaction Orders Using Power Law, L-H, and Multi-parameter Kinetic Modelsa activation energy (kJ/mol)

reaction order

reactions

power law

L-H

multi-parameter

power law

multi-parameter

HDS HDN HDA (360-380 °C) HDA (380-400 °C)

101 79 30 -18

99 69 62 -9

119 112 34

2 1.5 1.5 1.5

2.68 2.02 pseudo-first-order

a

The assumption for L-H is that HDS, HDN, and HDA are pseudo-first-order.

summarized in Table 7. The values of reaction orders were determined from the best fit of the experimental data. Difference values of n, thus different forms of the solutions of eq 7, were tested, and the ones that yielded the highest R2 values were considered the appropriate reaction orders. In the experimental conditions chosen, LHSV ranged between 0.65 and 2 h-1, while the temperature, pressure, gas/oil ratio, and H2 purity were constant at 380 °C, 9 MPa, 800, and 100%, respectively. The reaction orders of HDS, HDN, and HDA were determined to be 2, 1.5, and 1.5 respectively. Arrhenius plots for HDS, HDN, and HDA (not shown) were generated using experimental conditions where the temperature ranged between 360 and 400 °C, while the LHSV, pressure, gas/oil ratio, and H2 purity were constant at 1 h-1, 9 MPa, 800, and 100 vol %, respectively. R2 for Arrhenius plots range between 0.97 and 0.99. The Arrhenius plots show that HDS and HDN reactions are irreversible under these experimental conditions used in this study. It is well-known that, under industry conditions [temperature at 340-425 °C and pressure at 55-170 atm (5.617.2 MPa)], both HDS and HDN are irreversible.25 Also, the Arrhenius plot shows that HDA is a reversible process under the considered experimental conditions. The HDA apparent reaction rate increases with the temperature until 380 °C, after which it starts to decrease. According to the literature, maximum HDA is achieved between 370 and 385 °C.1 The activation energies for HDS, HDN, and HDA were calculated, and the results are summarized in Table 7. The activation energies for HDS and HDN were 79 and 101 kJ/ mol, respectively. Because of reversibility of HDA, two activation energies were calculated for the temperature ranges of 360-380 and 380-400 °C. In the 360-380 °C range, the value was 30 kJ/mol, and in the 380-400 °C range, the value was -18 kJ/mol. The explanation for this phenomenon is that increasing the temperature has two competing effects on HDA: (1) increased reaction rates and (2) lower equilibrium conversions.25 Thus, at lower temperatures, HDA is kinetically controlled, while at higher temperatures, it is equilibrium-controlled. Consequently, in practice, a balance must be struck between using lower temperatures to achieve maximum reduction of aromatic content and using higher temperatures to give high reaction rates and a minimum amount of catalyst charge per barrel of feed.1 The effect of H2 pp on HDS, HDN, and HDA kinetics were also observed. In the experiments, H2 purity was varied between 50 and 100 vol % (with the rest methane), while the temperature, pressure, gas/oil ratio, and LHSV were kept constant at 380 °C, 9 MPa, 800, and 1 h-1, respectively. The data were analyzed using the power law model, and the results are presented in Figure 11. The results show that HDN is more sensitive to H2 pp than HDS and HDA. This is

Figure 11. Rate constant as a function of the (a) inlet H2 pp and (b) outlet H2 pp.

explained by the differences in mechanisms of the HDS, HDN, and HDA, as discussed in section 3.3. Similar results were observed by Fang;23 however, the study did not address HDA. 3.6.3. L-H Model. The use of the L-H model for kinetic modeling of real industrial feed is very complicated because of the many coefficients that must be determined as well as the difficulty in their determination.37 However, it is thought to be a better approach than the other two correlations because it accounts for the inhibition caused by H2S and other species under hydrotreatment.37 Two simpler versions of this model, eqs 9 and 10, were used in this study. Equation 9 was used to describe HDS and HDN kinetics,34,37 while eq 10 was used to describe HDA kinetics.35 Equation 10 assumes that H2 does not inhibit hydrogenation. In all three processes (HDS, HDN, and HDA), the reaction orders were assumed to be pseudofirst-order. - ri ¼

ki Ki KH2 PH2 Ci 1 þ Ki Ci þ KH2 PH2 þ KH2 S PH2 S

ð9Þ

kf Ki KH2 PH2 Ci 1 þ Ki Ci þ KH2 S PH2 S

ð10Þ

- ri ¼

Equations 9 and 10a were solved using Maple V software, which yielded the following solutions. 780

Energy Fuels 2010, 24, 772–784

: DOI:10.1021/ef9010115

For eq 9, the solution is

Mapiour et al.

0

0

! 11 Cio Ki þ lnðCio Þ þ lnðCio ÞKH2 PH2 þ lnðCio ÞKH2 S PH2 S KH2 PH2 Ki ki C B B B tþ CC B C KH2 PH2 Ki ki B CC B CC BKi expB B CC B 1 þ KH2 PH2 þ KH2 S PH2 S @ AC B B C B C C ð1 þ KH2 PH2 þ KH2 S PH2 S ÞLambertWB B C 1 þ KH2 PH2 þ KH2 S PH2 S B C B C B C B C B C B C @ A

Ci ðtÞ ¼

Ki

ð9aÞ where 3 8 125 5 54 6 x - x þ ð0Þ7 LambertWðxÞ ¼ x - x2 þ x3 - x4 þ 2 3 4 5 and

! 11 Cio Ki þ lnðCio Þ þ lnðCio ÞKH2 PH2 þ lnðCio ÞKH2 S PH2 S KH2 PH2 Ki ki C B B B tþ CC C B KH2 PH2 Ki ki B CC B CC BKi expB B CC B 1 þ KH2 PH2 þ KH2 S PH2 S @ AC B C B C B C x ¼B C B 1 þ KH2 PH2 þ KH2 S PH2 S C B C B C B C B C B C B A @ 0

ð9bÞ

0

For eq 10, the solution is

ð9cÞ

! 11 Cio Ki þ lnðCio Þ þ lnðCio ÞKH2 S PH2 S KH2 PH2 Ki ki C B B B tþ CC C B KH2 PH2 Ki ki B CC B CC BKi expB B CC B 1 þ KH2 S PH2 S @ AC B C B C B C B ð1 þ KH2 S PH2 S ÞLambertWB C 1 þ K P C B H2 S H2 S C B C B C B C B C B A @ 0

0

Ci ðtÞ ¼

Ki

ð10aÞ

where 3 8 125 5 54 6 x - x þ ð0Þ7 LambertWðxÞ ¼ x - x2 þ x3 - x4 þ 2 3 4 5

ð10bÞ

and ! 11 Cio Ki þ lnðCio Þ þ lnðCio ÞKH2 S PH2 S KH2 PH2 Ki ki C B B B tþ CC C B KH2 PH2 Ki ki B CC B CC BKi expB B CC B 1 þ KH2 S PH2 S @ AC B C B C B C x ¼B C B 1 þ KH2 S PH2 S C B C B C B C B C B C B A @ 0

0

781

ð10cÞ

Energy Fuels 2010, 24, 772–784

: DOI:10.1021/ef9010115

Mapiour et al.

An Excel solver was used to solve eqs 9a and 10a. The apparent rate constants and adsorption equilibrium constants were obtained by the means of a trial-and-error method.34 The partial pressures of H2 and H2S were obtained from the HYSYS analysis. R2 for L-H predictions was greater than 0.99. 3.6.4. L-H Analysis of HDS, HDN, and HDA. The data for this analysis were generated using experimental conditions, where the temperature ranged from 360 to 400 °C, while the LHSV, pressure, gas/oil ratio, and H2 purity were constant at 1 h-1, 9 MPa, 800, and 100 vol %, respectively. Equations 9a and 10a were used to determine apparent rate constants and adsorption equilibrium constants for HDS, HDN, and HDA. The results are given in Table 8. All of the adsorption constants showed a decreasing trend with an increasing temperature, implying that HDS, HDN, and HDA are all exothermic reactions.5 The decrease of the H2S adsorption constant with the temperature means that H2S inhibition on hydrotreating decreases with an increasing temperature.34 Results of apparent rate constants show that HDA increases with the temperature at temperatures below 380 °C and decreases at temperatures above 380 °C for reasons explained in section 3.2. Activation energies were also calculated from Arrhenius plots (not shown). The

activation energies for HDS and HDN were 99 and 69, respectively. Because of the reversibility of HDA, two activation energies were calculated within the temperature ranges of 360-380 and 380-400 °C. In the 360-380 °C range, the value was 62 kJ/mol, and in the 380-400 °C range, the value was -9 kJ/mol. R2 for the Arrhenius plots ranged between 0.98 and 0.99. 3.6.5. Multi-parameter Model. The multi-parameter model is a better model than the overly simplified power law model because it includes more process variables and, thus, the effects of more variables on hydrotreating activities can be observed. The multi-parameter model is shown below, eq 11.38 dC ¼ ki PH m C n ðG=OÞq ð11Þ dt Equation 11 solution is ln

where Cf and Cp are the concentrations of nitrogen, sulfur, or aromatics in the feed and product, respectively, ko is the preexponential factor, s is E/R, where E is the activation energy and R is the gas constant, n is the reaction order, m, q, and c are the empirical regression factors, PH is the H2 pp (in this work, outlet H2 pp), G/O is the gas/oil ratio, and LHSV is the liquid space velocity. 3.6.6. Multi-parameter Model Analysis of HDS, HDN, and HDA. The data for this analysis were generated under experimental conditions, where the temperature, pressure, gas/oil ratio, LHSV, and H2 purity ranged between 360 and 400 °C, 7 and 11 MPa, 400 and 1200, 0.65 and 1 h-1, and 75 and 100 vol % (with the rest methane), respectively. The data were analyzed using the nonlinear regression model in Polymath software. The parameters for HDS, HDN, and HDA are shown in Table 9. The activation energies and reaction orders of HDS, HDN, and HDA were 119 kJ/mol and 2.68, 112 kJ/mol and 2.02, 34 kJ/mol and 1 (pseudo-first-order), respectively. The parameters for HDA were determined for

temperature (°C) 370

380

390

400

ks (h-1) Ks (MPa) KH2 (MPa) KH2S (MPa)

1.73 8.93 1.81 125.99

HDS 2.20 7.30 1.80 113.99

2.90 5.74 1.74 101.99

3.86 4.26 1.61 91.00

5.29 3.18 1.55 79.90

kn (h-1) Kn(MPa) KH2 (MPa) KH2S (MPa)

2.08 2.15 1.81 125.99

HDN 2.64 2.09 1.80 113.99

3.15 1.97 1.74 101.99

3.68 1.78 1.61 91.00

4.66 1.67 1.55 79.90

ka (h-1) Ka (MPa) KH2 (MPa) KH2S (MPa)

4.01 4.50 2.86 119.00

HDA 4.92 3.79 2.54 108.99

5.74 3.16 2.45 101.98

5.61 2.91 2.40 88.49

5.47 2.65 2.34 74.99

Table 11. Rate Constants and Adsorption Constants of HDA Using eq 10

Table 9. Multi-parameter Model Parameters for HDS, HDN, and HDA parameter ko s m q c n

ð11aÞ

" # 1 1 1 ko eð - s=TÞ PH m ðG=OÞq - n-1 ¼ ; n > 1 ð11bÞ n 1 n - 1 Cp Cf LHSVC

Table 8. Summary of the Rate Constants and Adsorption Constants Determined Using the L-H Model

360

Cf ko eð - s=TÞ PH m ðG=OÞq ¼ ;n ¼ 1 Cp LHSVC

temperature (°C)

HDS

HDN

HDA

3.53  1010 1.42  104 0.98 -0.31 2.72 2.68

8.03  108 1.35  104 1.82 -0.22 2.04 2.02

1.63  102 4.08  103 0.47 -0.01 0.24 pseudo-first-order

360 ka (h-1) Ka (h-1) KH2 (MPa) KH2S (MPa)

3.70 0.85 2.54 124.99

370 HDA 4.45 0.76 2.37 113.99

380

390

400

5.18 0.60 2.21 102.06

5.14 0.58 2.15 80.83

5.05 0.55 2.09 76.99

Table 10. Comparison on the Predictive Power of Multi-parameter versus L-H Models determined from the models (%) determined experimentally (%)

multi-parameter

L-H

purity (vol %)

HDS

HDN

HDA

HDS

HDN

HDA

HDS

HDN

HADa

HDAb

50 80 90

92 95 96

38 64 71

38 48 49

92 95 95

45 67 72

38 50 53

73 93 95

30 63 70

100 93 83

29 46 50

a

Using eq 10. b Using eq 9.

782

: DOI:10.1021/ef9010115

783

-17

30 34 58

85

74 80 105 92 93.5 79 112 69

72 132 94

1.5 pseudo-first-order

1.5 pseudo-first-order pseudo-first-order 1.5 2 2 1.5 pseudo-first-order 1.5 2.02 pseudo-first-order

1

Yui and Dodge40 Ai-jun et al.38 Owusu-Boakye et al.41 Ferdous et al.33 Bej et al.39,45 Mann et al.42 Yui and Sanford43 Botchwey et al.44 present work present work present work

286-541 214-559 170-439 185-576 210-655 HGO 196-515 210-600 258-592 259-592 260-592

power law multi-parameter L-H L-H power law power law power law L-H power law multi-parameter L-H

1.5 1.5 pseudo-first-order 1 1.5 1.5 1 pseudo-first-order 2 2.68 pseudo-first-order

1 1.6

pseudo-first-order

151 141 55 87 28 87 138 114.2 101 119 99

HDA 380 °C HDA 380 °C

the temperature below 380 °C. As the temperature increased above 380 °C, the HDA conversion decreased as the hydrogenation reversed. R2 for HDS, HDN, and HDA were 0.76, 0.92, and 0.90, respectively. 3.6.7. Comparison of the Prediction Power of Different Kinetic Models. A comparison of the activation energies and reaction orders obtained using different kinetic models is presented in Table 7. Most studies often report how well the predicted data agree with the experimental data, which is used to generate the model(s); i.e., they report R2. A better approach is to test the ability of the developed model(s) to predict new observations or data that are not used in the generation of the model(s), i.e., to test their predicted R2. Hence, three experiments were conducted, in which the pressure, temperature, LHSV, and gas/oil ratio were kept constant at 9 MPa, 380 °C, 1 h-1, and 800, respectively. Only H2 purity was varied as follows: 50, 80, and 90 vol % (with the rest methane). None of these conditions were used in the development of the kinetic models. Also, note that the experimental condition at 50% H2 purity is an extrapulated condition; i.e., it falls outside the range of the conditions originally used to develop the models. The power law model could not be used because of its exclusion of many of the process variables. The comparison between the multi-parameter and L-H models is presented in Table 7. The multiparameter model was reasonbly accurate at predicting values for HDS, HDN, and HDA conversions. The L-H model was reasonbly accurate at predicting values for HDS and HDN conversions; however, it could not predict the extrapulated condition well. Moreover, HDA predicted results using eq 11a, which is a version of the L-H model, were not logical because they suggested that HDA conversion increases with decreasing H2 purity. Thus, the assumption that H2 does not inhibit HDA was discarded, and eq 10a was used instead. The results of the predictions are shown in Table 10. With the assuption that H2 does indeed inhibit HDA, better agreement between the predicted and experimental data were obtained. The results of the apparent rate constant and equilibrium adsorption constants of HDA using eq 10a are given in Table 11. The activation energy was calculated for each of the two temperature ranges: 360-380 and 380-400 °C. In the 360-380 °C range, the value was 58 kJ/mol, and in the 380-400 °C range, the value was -5 kJ/mol. R2 for Arrhenius plots ranged between 0.92 and 0.99. The advantage of the multi-parameter model is that it results in better predicted values even for extrapolated conditions, while the advantage of the L-H model is that a smaller amount of experimental data is needed to determine its parameters. A comparison between the activation energies and reaction orders determined in this work and those found in the literature is summarized in Table 12. It can be seen in this table that activation energies and reaction orders determined in this work are in reasonable agreement with those reported in the literature. Discrepancies in the activation energy values can be attributed to changes in the (assumed) reaction

-5

Mapiour et al.

reaction order

Energy Fuels 2010, 24, 772–784

Energy Fuels 2010, 24, 772–784

: DOI:10.1021/ef9010115

Mapiour et al.

mechanism or interference of physical phenomenon, such as diffusion.34

HDS and HDA. (5) Correlations between outlet H2 pp and hydrotreating conversions had higher R2 than those of inlet H2 pp. This may suggest that it is better to use outlet H2 pp for design applications. (6) Increasing H2 pp results in increases in hydrogen consumption and dissolution; however, no clear correlation was obtained with regard to feed vaporization. (7) The multi-parameter model gave better predictions of hydrotreating conversions than the L-H model.

4. Conclusions Within the range of the experimental conditions considered in this study, the following conclusions were made: (1) Increasing pressure and H2 purity lead to increases in inlet H2 pp. The gas/oil ratio does not have a significant effect on inlet H2 pp. (2) An increasing pressure, gas/oil ratio, and H2 purity lead to increases in outlet H2 pp. The effects of the pressure and H2 purity are more significant than that of the gas/oil ratio. (3) The temperature and LHSV do not have significant effects on inlet or outlet H2 pp. (4) HDS, HDN, and HDA increase with increasing H2 pp. Within the range of the conditions studied, HDN is more affected by H2 pp than

Acknowledgment. Financial support from Syncrude Canada Ltd. and Natural Sciences and Engineering Research Council of Canada (NSERC) Industrial Postgraduate Scholarship (IPS) is acknowledged. Also acknowledged are members of Catalysis and Chemical Reaction Engineering Laboratories (Department of Chemical Engineering, University of Saskatchewan) and Miss Cristina Weir.

784