prediction of industrial trickle-bed operation from pilot-plant data

Faculty of Technology and Metallurgy, University of Belgrade, Karnegijeva 4, P.O. Box 494, 11001 Beograd,. Yugoslavia. Volker W. Meyn, Dieter K. Sever...
0 downloads 0 Views 734KB Size
Ind. Eng. Chem. Res. 1991,30, 2059-2065

2069

KINETICS AND CATALYSIS Hydrotreating of Used Oil: Prediction of Industrial Trickle-Bed Operation from Pilot-Plant Data Dejan

U.Skala,* Marko D. Saban, a n d Aleksandar M. Orlovid

Faculty of Technology and Metallurgy, University of Belgrade, Karnegijeva 4, P.O. Box 494, 11001 Beograd, Yugoslavia

Volker

W.Meyn, Dieter K.Severin, and I r a d j G.-H. Rahimian

German Institute for Petroleum Research, 3392 Clausthal-Zellerfeld, Germany

MomEilo V . Marjanovi6 Refinery 'Beograd", PanEevaEki put 83, 11001 Beograd, Yugoslavia

Used oil hydrotreating was investigated in a pilot trickle-bed reactor (TBR) a t 27&350 "C, 5-7 m a , and 1.1-4.6 liquid hourly space velocity (LHSV) and with different hydrogen/oil ratios using a commercial Co-Mo/A1203 catalyst. Hydrodesulfurization (HDS), hydrodeoxygenation (HDO), and metals removal were investigated by using a modified power-law kinetic model with a power term for LHSV. It was found that the HDS and HDO reactions can be described by pseudo-first-order kinetics. The removal of metals was found to be primarily due to the physical process of deposition on the catalyst bed. With the use of the kinetic data from a pilot plant, the simulation of an industrial TBR was performed. Simulated HDS and HDO, removal of metals, and prediction of catalyst deactivation agreed well with the industrial data for three charges of catalyst. Introduction The increased usage of various additives in motor oils in the past reached a point where the old process of clay/acid refining was not sufficiently effective for the successful regeneration of used oil. Most conventional regeneration processes are currently based on several technologically different steps (Short et al., 1987). The first step is usually the separation of water, low hydrocarbons, and other low-boiling components. The KTI (Kinetics Technology International) process then uses special wiped-film evaporation to separate the unsaturated and some of the chlorinated and sulfur components from the heavy impurities (Berry, 1979). The Snamprogetti and IFP (Institut FranGais du Petrole) processes in this stage of refining employ propane solvent extraction (Audibert, 1978; Short et al., 1987). All the above processes use hydrotreating as a very efficient final step for the stabilization of a partially refined used oil. As a result of hydrotreating, almost complete removal of traces of metals, carbonyl and sulfur compounds, as well as olefin saturation in the partially refined used oil is expected. Hydrotreating is carried out industrially in a trickle-bed reactor (TBR) at 270-350 OC, 4-6 MPa, and 1-2 LHSV over Co-Mo/A1203 catalyst. The main problem in the hydrotreating step, particularly in the IFP process, which is applied at the Refinery 'Beograd"-Naftagas, is rapid catalyst deactivation. The average catalyst consumption is reported to be 2-3 kg/ton of refined oil (MarjanoviE, 1990). Therfore, one of the main points of the research work was to identify the reason(s) for rapid catalyst deactivation during used oil hydrotreating. The rates of

*Towhom correspondence should be addressed.

Table I. Properties of Used Oil (Average Values) density, kg/m3 900 acid number 0.82 Conradson number 0.25 ash content, % 0.04 sulfur content, % 0.70 4.8 x 10-4 viscosity of used oil, Pa s 20.0 x 10-8 surface tension, N/m*

HDS, HDO, and metals removal were determined in a pilot TBR. These data were then used for the simulation of an industrial process. The simulated HDS, HDO, removal of metals, and catalyst deactivation were compared with three charges of catalyst from an industrial reactor. Experimental Section The pilot plant used for the hydrotreating studies was described previously in detail by Skala et al. (1988). It consists of a TBR 60 cm long with a 2.0-cm inner diameter, packed with 1.5-mm trilobe extrudates of a Co-Mo/A120s catalyst (Shell 447t). The catalyst was diluted with inert particles in the ratio 1:2.3 v/v. The hydrotreating reactions were studied under steady-state operation at 5,6, and 7 MPa, temperatures of 270,300,325, and 350 "C,hydrogen/oil ratios of 100-900 cmN3/cms,and 1.1-4.6 LHSV's. Some properties of the feedstock for hydrotreating are shown in Table I. The determination of the conversion degrees of sulfur, carbonyls (denoted as oxygen), and metals in the feed oil was used for the calculation of the HDS, HDO, and hydrodemetalization (HDM) rates. The following analytical methods were used elemental C, H, and S analyses (Heraus analyzer); FTIR (Nikolet FT 60 SX) for carbonyl compound content; IATROSCAN (a combination of column and gas chromatography) for rapid

0888-5885/91/2630-2059$02.50/0 0 1991 American Chemical Society

2060 Ind. Eng. Chem. Res., Vol. 30, No. 9,1991 350°, LHSV: 1,3 h

LHSV, h-l

:I I

300°, CD=1,3 @ 2.1 CARBONYL

-- i LO

-'

@=2,9

X

SULFUR

40 20

50

70

60

P , bar

Figure 3. Conversion of S, M, C, and A versus pressure (350 "C, LHSV = 1.3 h-l). a w

>

40

70 b a r ,

LHSV= 1.28 h - l

z

20

20 I

I

1

280

320

300

340 t,

60

50

70

P,bar

Figure 1. Muence of pressure and LHSV on the conversion degree of sulfur (S) and carbonyl (C)compounds, metals (M), and asphaltenes (A) at 300 "C. 100J ,

O C

Figure 4. Conversion of S, C, M, and A versus temperature at 70 bar and LHSV = 1.28 h-l. 350° I

70 bar (-4,

50 bar (-4

-

LHSV: 1,l h - l , 35O0/60 bar

100

200

300 400 H2 /oil,

500

mi/m3

Figure 2. Conversion of S, C, M, and A versus H2/oil ratio (350"C, 60 bar, and LHSV = 1.1 h-l). analysea of the polar compounds and metals, and elemental

Pb, Fe, Zn, Ca, and P analyses (Perkin Elmer ICP/6OOO). Results and Discussion Pilot-Plant Data. The conversion degree was calculated as the ratio of the absolute decrease of the corresponding and the initially measured values in the feed (partially refined used oil): (1) Xi = ((@i,o - @ i ) / @ i , J where cPi denotes the content of sulfur (S),carbonyls (C), metals (MI, and asphaltenes (A). The results are shown in Figures 1-4. It can be seen from the figures that the HDS reaction is the slowest ope. All the data indicate that the efficient removal of sulfur compounds (>60% HDS) ensures a significant conversion of C, M, and A. The conversion of C, M, and A was almost independent of LHSV (Figure 1)and pressure (Figure 3). A smaller hydrogen to oil ratio at the reactor inlet negatively influences the HDS (Figure 2), but is found to have no effect on the HDM and conversion of heavy compounds (asphaltenes). The conversion of all the compound classes increases with increasing temperature (Figure 4). Above 300 OC and 5 MPa the conversion of M, A, and C is very efficient. The results in Figures 1-4 suggest that there is a difference in the HDS mechanism and the mechanisms of M, A, and C

1,s

2

3

4

5

6

LHSV. h-'

Figure 5. Analysis of sulfur and aaphaltene conversion using the hold-up model.

removal. The removal of M and A is likely to be primarily by physical deposition in the catalyst bed. The removal of C seem to follow the transient mechanism between S and M. Monitoring of the HDS process can be used as the main parameter for evaluating the efficiency of the hydrotreating process. Kinetic Analysis. It is often observed in pilot plants that catalyst effectiveness depends on the liquid flow rate. Henry and Gilbert (1973) have shown that such an effect can be explained by taking the total liquid holdup into account. Using the correlation of Satterfield et al. (1969) for liquid holdup, Henry and Gilbert derived the following equation for a first-order reaction in an isothermal TBR -In (1 - Xi) 0: L,1/9(LHSV)-2/9d62/3v1/9 (2) The above equation suggests that the first-order power-law kinetic model for HDS should include the power term for LHSV to compensate for the deviations from plug flow in a pilot reactor. This is shown in Figure 5 for the conversion of S and A and in Figure 6 for the conversion of C. In the case of HDS the slope of -In (1 - X i ) versus LHSV in a log-log diagram was found to be -0.77 (Figure 5),which is reasonably close to the theoretically predicted value

Ind. Eng. Chem. Res., Vol. 30, No. 9, 1991 2061 Table 11. Kinetic Parameterr of Ueed Oil Hydrogenation (HDS,HDO) reference k, h-l kn, mf/(mSclt h)

E, kJ/mol

0.33-1.0" 0.37-1.0b 0.3&0.Mb 2.16 5.2-7.5 0.27

Mohmmed and Hankish (1985). 300-340 "C Mann et al. (1988), 350 O C present study, 270-350 O C 'Second-order reaction, k, h-' (mas%)-'.

* 1.5-order reaction, k, h-'

430 1 x 107

T

I\\

CO -Mo v Ni-Mo Ni-Mo

87,l kJImol (Monn,19881

'"1 \ ;\, ' 1 1,2

bar

HDS, gas oil HDO, used oil HDS, used oil

(mas%P6.

CARBONYL

7 -

remarks HDS, residue HDS, gas oil HDS, distillate of used oil

80-120 60-100 81.5 71.2 62.9 87.1 21 83

Shah and Paraskos (1975)

0.4

x

0

(Mohammed Hankish.1985, 62.9 kJImol) (present E = 82,9 kJImol work)

-O,:[

-0.8

1,s

3

2

4

5

6

-1.6

LHSV, h - l

Figure 6. Analysis of carbonyl compound conversion using the hold-up model.

-I2

1,6

1,7

1,8

1.9 1IT x

First order

(hold-up modell

Figure 8. Arrhenius plot for HDS of wed oil and vacuum gas oil.

1.2

-

08

7

C

'

0.4

0 .2

.4

.6

8 LHSV-"

Figure 7. Determination of kinetic parameters for HDS of used (partially refined) oil.

(-0.67). A much smaller effect of the LHSV on the conversions of A and C was determined in Figures 5 and 6: -0.25 and -0.32, respectively. These results may indicate the different mechanisms by which A and C are removed as compared with HDS (Shah, 1979). The results of HDS and HDO kinetic analyses (preexponential factor ko and the apparent energy of activation E) were compared with the literature data on the hydrotreating of used oils in Table 11. The HDS and HDO reactions in the used oil are defined by the following equations,which are valid for the case of the initial catalyst activity (arbitrarily equal to 11,i.e., not more than 30-40 h of catalyst bed on stream: - ~ D S =

1 X 10' exp(-lOOOO/T)CS

-rHm = 0.43 X

(m3,fl/(m3,t h)) (mas% S) (3)

loa exp(-2500/T)Co (m30il/(m3clth)) (mas% 0)(4)

As the data from the pilot plant showed that the metals removal was unaffected by temperature between 300 and 350 O C , the rate of metals deposition on the catalyst bed was defined by the equation -rHDM

Id K - l

4.17C~ (m30il/(m3cl,h)) (mas% M) (5)

The HDS and HDO reactions are exothermic,while metals removal by deposition is assumed to have no significant heat effect. The total heat effect of the HDS reaction has been estimated to be about 2200 kJ/kg of S, and of the HDO reaction about 2000 kJ/kg of 0. Catalyst Deactivation. Selected samples of the deactivated catalyst from an industrial reactor have been analyzed and the results published by JovanoviE et al. (1989). Several assumptions were used for checking the main source of catalyst deactivation, based primarily on the quality of the used oil feedstock (such as Conradson number and ash content). The results can be summarized in the following (JovanoviE et al., 1989): (a) The difference in the Conradson number of the used and hydrotreated oil correlates with the coke deposited on the catalyst surface and in the catalyst bed. (b) The differences in the ash and metals content of the used and hydrotreated oil correlate with the amount of metals deposited on the catalyst surface. (c) Both deposits (coke and metals) are responsible for the catalyst deactivation. The coke deposit is found to affect the catalyst bed porosity. An empirical relation between the change in the bed porosity and the decline in catalyst activity has been found. JovanoviE et al. (1989) have shown that about 1200 kg of intraparticle deposits can be expected and about 200 kg of interparticle coke in an industrial reactor packed with about 4300 kg of hydrotreating catalyst (reactor diameter 1m and height 9 m). An intraparticle deposit is formed, according to the above study, primarily from the coke, but the deposited metals were identified as important catalyst poisons. Under industrial conditions the average coke conversion in the used oil was 65%, and according to elemental analyses 28.4% of the converted coke was deposited inside the catalyst particles.

Modeling of an Industrial Trickle-Bed Reactor The model is based on three main reactions (HDS, HDO, and HDM), each of them being pseudo-first-order and with defined kinetic constants derived from the ex-

2062 Ind. Eng. Chem. Res., Vol. 30, No. 9, 1991

periments performed in a laboratory trickle-bed reactor (eq 3-5). The maea balances for sulfur,carbonyl, and metal compounds are Cs,& dXs/& = (-rms)aVrPoii(l - €0)

(6)

CO,& dXo/dz = (-rHDO)aVrPoil(l - €0)

(7)

- €0)

(8)

CM,& dXM/dz = (-rm&VrPoiI(l

The energy balance for adiabatic operation of the trickle-bed reactor is determined by dT/& = [Vr(l - tO)/(LCp~+ GCN)] X [(-rHDS)Poil(-mHDS) + (-rHM))Poil(-mHDO)l (9) These four differential equations were the basis of a simplified mathematical model of an industrial TBR that enabled calculation of the sulfur, carbonyl, and metal conversion, as well as the temperature change along the length of the reactor. Because of the very fast catalyst deactivation process (JovanoviE et al., 1989; MarjanoviE, 1990),the inlet temperature of the feed must be increased continuously with time in order to maintain the desirable and required conversion in an industrial reactor. The experimental results from the pilot plant indicated that hydrodesulfurization was the slowest step determining the rate of the whole hydrogenation process. Therefore, the criterion for a sulfur conversion of 60% (Xs = 0.6) is the set parameter that must be satisfied, which means that it dictates the necessary rate of increasing feed temperature. The catalyst activity a is dependent upon both coke and metal deposition on the catalyst surface. The catalyst activity decline was calculated with an equation originally proposed by Shah et al. (1976): a = (1- 1.58Cd,,06)0~6

(10)

The concentrations of coke and metal deposits on the catalyst surface was calculated on the basis of results published by Jovanovii! et al. (1989): Cdep = [o.185cmke,0 + Cwh,OlLt/W

- 1.58(to - ~t)0.s]0*6

(12)

which enables the calculation of the decrease in the bed porosity with the duration of the hydrogenation process: tt

=

€0

- [(l - ~'"')/1.58]~

Table IV. Comparative Results of Simulation and Experimental Data industrial data simulated dataa product product Shell 477-1.5 time on yield, time on yield, mm trilobe stream, h tons stream, h tons 1st charge 650 1400 650 1340 2ndcharge 648 1350 600 1328 3rdcharge 550 1050 800 1568 "The criterion for determining the total time of catalyst wage was conversion degree of sulfur compounds greater than 6046,and maximum allowable temperature in the catalyst bed 360 "C.

'\0 0

100

200

KM

400

500

600

700

Mx,

t(h)

Figure 9. Change of catalyst activity during used oil hydrogenation (simulation) for all three charges of catalyst (according to Shah et al. (1976)).

(11)

Since the catalyst activity decline is dependent on the concentration of deposits, it is obvious that catalyst deactivation could also be related to the decrease in bed porosity. The change in bed porosity changes the hydrodynamic conditions in the reactor, causing an increase in the pressure drop through the catalyst bed. Therefore, it was assumed that the change in catalyst bed porosity could be related under some special circumstances (equal density of catalyst and deposit in the bed) to the catalyst activity by an equation similar in form to the one proposed by Shah et al. (1976) = [l

Table 111. Properties of Catalyst Charge and Average Values of the Operating Conditions in an Indurtrial Plant catalyst type Shell 4774 (trilobe) 150 mass flow rate of hydrogen, kg/h mass flow rate of used oil, kg/h 2200 1 reactor diameter, m catalyst bed length divided in two parts 9 with other internal part, m initial catalyst porosity, ?% 40

(13)

Evaluation of the bed porosity according to eq 13 makes it possible to calculate the pressure drop using equations from the literature. An equation proposed by Midoux et al. (1976) and Charpentier and Favier (1975) for foaming hydrocarbons gave the best results in predicting the initial pressure drop in the catalyst bed. That was the starting point for estimating the good sides of the equations proposed in the literature because the initial bed porosity was a known parameter. That is, all other equations for the

t(h)

Figure 10. Conversion of sulfur organic compounds for all three charges of catalyst (simulation).

pressure drop calculation in a two-phase flow through a packed bed gave too low a value for the initial pressure drop. Smulation of the hydrogenation process in an industrial reactor for three charges of catalyst was based on the experimentally derived reaction rates for HDS, HDO, and HDM and the above presented discussion and assumption related to the change of catalyst activity (eq 10 and 11). The gas and liquid flow rates, as well as the inlet concentrations of sulfur, metal (ash), coke (Conradson), and carbonyl compounds were taken from a real industrial plant (average values are given in Table III). The obtained results are shown in Table IV and in Figures 9-14. The data presented in Table IV, which show the total time of catalyst usage in a hydrogenation process, as well as the total yield of product for the first and the second charge of catalyst, indicate good agreement between the

Ind. Eng. Chem. Res., Vol. 30,No. 9,1991 2063 cetolyst charge, 2

;-----,

/

simulat ion

kj

I

'II

1

g 08

experimental data

04

'1

I 1

0

20

10

30

50

40

70

60

80 90 t/tt&"'

xx)

Figure 11. Dependence of the relative increase of average temperature in the reactor on the reduced time of catalyst life. The curves present three Werent chargea of catalyst and Tam"s curve (Tamm et al., 1981) for the hydrotreatment of atmospheric residue.

I

.-m

c

4-

0

1001 0

87-

Kx)

200

300

400

500

600

L

030

-

79L

I

0

200

xx)

300

500

400

600 t(h1

I

800 t(h1

700

-

Figure 13. (a) Conversion of metal compounds for the second charge of catalyst: simulation and processing plant data. (b) Conversion of metal compounds (simulation) for three charges of caMyst (1, 2, and 3, respectively, Table 11). (c) Conversion of metal compounds for three charges of catalyst at the front end of the catalyst bed (after 10% of total catalyst volume). b

Chorpent ier's equation

experimental

12

I-

u)

\

'

1

'

'

'

'

1

1

1

1

1

1

0 " 3 0 0 b M ) 5 0 0 6 0 0

t(h)

" ' ' " ' ' 1 ' 1 1 0

Figure 12. Conversion of carbonyl compounds: (a) Comparison of data obtained by simulation (1) and data from a processing plant for the first charge of catalyst (2). (b) Comparison of data obtained by simulation (1)and data from a processing plant for the second charge of catalyst (2). (c) Comparison of data obtained by simulation (1) and data from a processing plant (2) for the third charge of catalyst.

Figure 14. Dependence of pressure drop versue time. Simulated data (curve) and data from a real induetrial reactor (pointa) for the first charge of catalyst.

calculated and experimental data. However, in the case of the third catalyst charge, the simulation data are about 50% larger than those observed during the hydrogenation of used oil. Figures9-14 generally represent the results of simulation which are in some cases also compared with the corresponding values determined during the hydrogenation of used oil. The change of catalyst activity is shown in Figure 9 for all three catalyst charges. Simulation curves were calculated using Shah's correlation (Shah et al., 1976), which was originally developed for the case of catalyst

deactivation during the HDM process. In the present work the total deposit on the catalyst surface was calculated by eq 11,taking into account the concentration of coke and metal compounds in the feedstock. The analysis of sulfur compound conversion with time of catalyst usage is demonstrated in Figure 10 for three catalyst charges. It is obvious that the sulfur conversion is always above 6090,which has also been the main task in the postulated boundary values for determining the necessary inlet and average temperature in the catalyst bed. That is, as shown by the results obtained in the

0

100

200

300

400

500

600

0

t(h)

2064 Ind. Eng. Chem. Res., Vol. 30, No. 9, 1991

pilot-plant study, HDS is the limiting step that also determines the overall reaction rate of hydrogenation of the used oil. That practically indicates the faster rates of HDO and HDM as well as the reaction of aromatic and olefinic compound saturation. Thus, during the simulation of sulfur compound conversion a temperature increase at the reactor inlet of 5 K was applied when the conversion degree dropped below 60%. This criterion was accepted as a “real” one usually performed in industrial operation. Unfortunately, the data on sulfur compound conversion on an industrial scale were not collected completely. Some analyses performed during the usage of the first catalyst charge indicate that HDS was always about 60% conversion. Figure 11 shows the relationship between the average temperature increase in the catalyst bed and reduced time. The reduced time is determined as the ratio of time and total time of catalyst usage in industrial operation. The quantity designated by Q on the ordinate axis is defined as Q = (Tav- To)/(T,,- To), where Tmaxpresents the maximal average temperatuie in a catalyst bed during the hydrogenation process with the specified catalyst charge. Curves 1,2, and 3 are given for the three different charges of catalyst that were investigated in the present work. They are the result of the calculation of the necessary average temperature in the catalyst bed for obtaining the desired conversion of sulfur compounds of 60%. In the same figure, the fourth curve represents the data given by T a ” et al. (1981) after the analysis of the hydrogenation process of atmospheric residues. Figure 12 indicates the carbonyl compound conversion. Besides the simulated data there are also those that were determined by analysis of the product composition from an industrial plant. Obviously, very good agreement was obtained in the case of predicting the carbonyl compound conversion with the real data from the process plant. The conversion of metal compounds, Le., their removal from partially refined used oil, is shown in Figure 13a. The values of the simulated and experimentally determined data were compared for the second charge. The agreement is quite good except at the end of the catalyst life where a much larger quantity of metals was detected in the hydrogenation product as compared to the quantity predicted by the simulation. Figure 13b indicates that the conversion of metal compounds for all the investigated catalyst charges is related to the time of catalyst usage. From these data one can conclude that practically complete removal of the metal compounds could be achieved during the first 300 h of hydrogenation of the used oil. Figure 1% shows some additional information that could be used for determining the change of catalyst activity in the bed. That is, the performed analysis, on the basis of the determined mathematical model, shows how the conversion of metal compounds decreases in the first 10% of the catalyst volume with time. Such data could be used for better understanding of the catalyst deactivation process, because the catalyst activity generally depends on the coke and metal deposited on the catalyst surface (Shah et al., 1976; JovanoviE et al., 1989). Deactivation of the catalyst is more pronounced at the inlet of the catalyst bed, where the largest concentration of metals and asphaltenes (coke) is deposited. Figure 14 shows the result of the calculation of the pressure drop through the catalyst bed using literature data (Charpentier and Favier, 1975) compared with the values measured at an industrial plant (first charge): ULG = L~UA€LG/(UL + UG) (14) where

ASLG = SLG - ( u L P o i l + U G P /~f t ~ ) SLc = L[1 + (l/X’)

+ (6.55/X’0.48)]2/e, x’= [L/G(L,AZ’G/p& + i)]0.5

and APG is the pressure drop of gas phase which flows throughout the bed of catalyst particles, calculated by using an equation similar to Ergun’s (Charpentier and Favier, 1975; Midoux et al., 1976). It shows a typical c w e characterized by a practically constant value of the pressure drop during the long time of catalyst life followed by a rapid increase, more or less exponential, at the end of catalyst activity, caused primarily by coke plugging of the catalyst bed. Conclusion The presented results of the hydrogenation kinetics of used oil determined in a trickle-bed pilot plant were successfully used for the mathematical modeling of industrial reactor operation, and particularly for the determination of sulfur, metal, and carbonyl compound conversion. The assumption that the main reactions in the hydrogenation of used oil are HDS, HDM, and HDO is supported by the results of simulation of catalyst life in a process plant and by determination of the yield of product for three different charges of catalyst. The results of the mathematical modeling of an industrial plant is also consistent with the assumption that coke deposition and the adsorption of metal compounds on the catalyst surface cause rapid catalyst deactivation. The change in catalyst activity is related quite well to the amount of the total deposit on the catalyst surface using Shah’s correlation. This equation, in modified form, was also used with good results for the prediction of the pressure drop through the catalyst bed. It links the catalyst activity with the porosity of the catalyst bed using Charpentier’s equation for the pressure drop for foaming hydrocarbons. Acknowledgment This work was done under the Agreement of Scientific Educational Cultural and Technical Cooperation between the Socialist Federal Republic of Yugoslavia and the Federal Republic of Germany. Financial support by the KFA International Bureau D5170 Julich on the German side as well as the research Fund of the Region Belgrade and the Jugopetrol-Refinery“Beograd” on the Yugoslavian side is gratefully acknowledged. Nomenclature a = catalyst activity Cab,O = initial concentration of metals determined as ash, mas%, (kg/kgoil) x 100 Cmke,O = initial concentration in used oil, mas%, (kg/kgoil)X C,, ~~

100 = total deposit of coke and metals at catalyst surface, w%at

Cu .. = total concentration of metals (ash) in used oil, mas%, (kg/kg0J X 100 CM,o = initial total concentrationof metals (ash) in used oil, mas%, (kg/kgoil) x 100 Co = concentration of carbonyl compounds in used oil, mas%, (kg/kgoil) X 100 C O ,= ~ initial concentration of carbonyl compounds in used oil, mas%, (kg/kgoil)X 100 C p L = specific heat capacity of oil, J/(kg K) C = specific heat capacity of gas, J/(kg K) concentration of sulfur compounds in used oil, mas%,

dG=

(kg/kg0iJ X 100

Ind. Eng. Chem. Res., Vol. 30, No. 9, 1991 2065 Cs,o= initial concentration of sulfur compounds in used oil, mas%, (kg/kgog) X 1Qo g = acceleration of gravitational force, m/sz d = equivalent diameter of catalyst particles, m mass flow rate of gas, kg/h AH- = enthalpy of hydrodeoxygenation reaction, J/kg of 0 AHws = enthalpy of hydrodesulfurization reaction, J/kg of S kWM = reaction rate constant of HDM reactions, m30fl/(m3,t h) km = reaction rate constant of HDO reactions, m30g/(m3,t h) kmS = reaction rate constant of HDS reactions, m30fl/(m3, h) L = liquid mass flow rate of oil, kg/h LHSV = liquid hourly space velocity, m30u/(m3cath) L, = length of reactor, m Q = reduced temperature, (Tav- To)/T,, - To) -rmM = reaction rate of hydrodemetalation, kg of M m30il/ ( b o i l m3at h) -rm = reaction rate of hydrodeoxygenation, kg of 0 m30i1/(kgoi~m3at h) -rm = reaction rate of hydrodesulfuration, kg of S m30il/(kgd m3pt h) t = time, h T = temperature, K t , = total time of catalyst usage (Figure 111, h T,, = average temperature in the catalyst bed, K T,, = maximal value of temperature in the catalyst bed, K To = initial temperature, K UG = superficial gas velocity, m/s UL= superficial liquid velocity, m/s V, = volume of reactor, m3 W = mass of catalyst, kg X = conversion degree z = dimensionless axial distance in the reactor

HDO = hydrodeoxygenation HDM = hydrodemetalation C, 0 = carbonyls S = sulfur compounds M = metals Registry No. Co, 7440-48-4; Mo,

Greek Symbols

Shah, Y. T.; Mhaskar, R. D.; Paraskos, J. A. Optimum quench location for a hydrodesulphurization with the time varying catalyst activity. Znd. Eng. Chem. Process Des. Deu. 1976, 15, 400-406. Short, H.; Hunter, D.; Parkinson, G. Three rerefining for waste lube oils. Chem. Eng. 1987, July 20, 21-23. Skala, D. U.; Saban, M. D.; JovanoviE, J. A.; Mayn, V. W.; Rahimian, I. G.-H. Pilot plant design for the process study of hydrodenitrogenation and hydrogen consumption. Ind. Eng. Chem. Res. 1988,27, 1186-1193. Tamm, P. W.; Harnsberger, H. F.; Bridge, A. G. Effects of feed metals on catalyst aging in hydroprocessing residuum. Zng. Eng. Chem. Process Des. Deu. 1981,20, 262-273.

d=

= initial catalyst bed porosity = catalyst bed porosity at time t @i = content of sulfur, metals, asphaltenes, carbonyls, etc., kg/kgoi,l -.-= initial content of sulfur, metals, asphaltenes, carbonyls, e t ~ .kg/kgoil , v = kinematic viscosity, m2/s pOg = density of used oil, kg/m30u tu: = correction factor given in eq 14, kg/(m2 a) Atu: = correction factor given in eq 14, kg/(m2 s) to tt

Abbreviations

HDS = hydrodesulfurization

7439-98-1; C, 1440-44-0.

Literature Cited Audibert, F. Huiles usagees schemes IFP de reraffinage. Rev. Znst. Fr. Pet. 1978,33,935-946. Berry, R. Rerefining waste oil. Chem. Eng. 1979, April 23,104-106. Charpentier, J.-C.; Favier, M. Some liquid hold-up experimental data in tricklebed reactors for foaming and nonfoaming hydrocarbons. AZChE J. 1975,21,1213-1218. Henry, C. H.; Gilbert, J. B. Scale up of pilot plant data for catalytic hydroprocessing. Ind. Eng. Chem. Process Des. Dev. 1973, 12, 328-334.

JovanoviE, N. N.; StankoviE, M. V.; MarjanoviE, M. V.; Skala, D. U. Catalyst deactivation in hydrogenation of used motor oils. J. Serb. Chem. Soc. 1989,54, 145-154. Mann, R. S.; Sambi, I. S.; Khulbe, K. C. Hydrofiiing of heavy gas-oil on zeolite-alumina supported nickel-molybdenum catalyst. Znd. Eng. Chem. Res. 1988,27,178&1792. MarjanoviE, M. V. Personal communication, 1990. Midoux, N.; Favier, M.; Charpentier, J.4. Flow pattern, pressure loss and liquid hold-up data in gas-liquid downflow packed beds with foaming and nonfoaming hydrocarbons. J. Chem. Eng. Jpn. 1976,9,350-356.

Mohammed, A.-H. A.-K.; Hankish, K. Interpretation of hydrogenation kinetics of spent oil distillate from UV spectroscopy. Fuel 1985,64,921-924.

Satterfield, C. N.; Pelossof, A. A.; Sherwood, T. K. Mass transfer limitations in a trickle-bed reactor. AZChE J. 1969,15,226234. Shah, Y. T. Gas-Liquid-Solid Reactor Design; McGraw-Hill New York, 1979; Chapter IV. Shah, Y. T.; Paraskos, J. A. Criteria for axial dispersion effects in adiabatic trickle-bed hydroprocessing reactors. Chem. Eng. Sci. 1975,31, 1169-1176.

Received for review May 14, 1991 Accepted May 28, 1991