Catalyst Deactivation in Adiabatic Prereforming - American Chemical

Chapter 13

Catalyst Deactivation in Adiabatic Prereforming: Experimental Methods and Models for Prediction of Performance Downloaded by UNIV OF MISSOURI COLUMBIA on September 4, 2014 | http://pubs.acs.org Publication Date: June 6, 1996 | doi: 10.1021/bk-1996-0634.ch013

Thomas S. Christensen and Jens Rostrup-Nielsen Haldor Topsøe A/S, Nymøllevej 55, DK-2800 Lyngby, Denmark

Adiabatic prereforming has become an established process step in the production of syngas. The main deactivation phenomena for the prereforming catalyst are sulphur poisoning and carbon formation. Carbon formation proceeds either as carbon whiskers at high temperature or as gum at low temperature. Trouble free operation is ensured by selecting operating conditions and catalyst type. Experimental methods are used to characterize the deactivation phenomena. Prereforming catalysts have been investigated by use of Temperature Programmed Reaction (TPR) showing the presence of different carbon morphologies. Application of the pseudo-adiabatic reactor principle has ensured reliable adiabatic bench-scale tests providing data for industrial design. Methods based on intrinsic activity measurement have been developed to reveal the impact of the deactivation phenomena. Sulphur adsorption isotherms and mathematical models describing the complex interaction between pore diffusion, poisoning rate and reaction kinetics have been established and are used for design and performance prediction.

Adiabatic prereforming is a low-temperature steam reforming process (7,2) which is used in modern syngas production to reduce energy consumption and investment (3). The prereforming process was earlier used in the manufacture of methane-rich gases for town gas and SNG (substitute natural gas) (2,4). Today prereforming is an integrated part of the plants producing hydrogen, ammonia, methanol, and CO/H mixtures. In the prereformer, all higher hydrocarbons are converted over a nickel catalyst into methane, carbon oxides and hydrogen. The methane reforming and shift equilibria are established at the reactor exit at the adiabatic equilibrium temperature: 2

C H + nH 0 CO + H 0 CO + 3 H n

m

2

2

2

-+ nCO + (n + /2m)H CO2 + H2 ^ CH4 + H 0 1

2

2

(-AH° 8

C-*

-> carbon

Whisker formation (10)

r C„H -* -> -"CH CH -# gum Gum formation (11) Assuming that the process conditions have been selected to avoid whisker carbon formation (r = 0), the rate of gum formation can be expressed by equation 12: P

,,

X

2

2

c

rp = r - r A

(12)

H

The activation energy for adsorption of hydrocarbons on nickel is about 40 kJ/mole (16) and the activation energy for hydrocracking is in the range 160200 kJ/mole (77). This means that below the temperature, T , at which r > r , i.e. r > 0, gum formation takes place. To ensure trouble free operation the prereformer must operate within the temperature window of T < T < T . P

P

A

H

P

c

Models for Performance Prediction Prediction of the deactivation rate can be made by empirical as well as by computational methods. The models can be used at two stages: • In the design phase; for proper sizing of the reactors and for determination of the necessary catalyst volume for the required operating period. • In the plant operation; for observation of the actual deactivation progress and determination of the optimal time for catalyst change out. Empirical Methods. The graphical deactivation plot is a very useful empirical method for prediction of the catalyst performance and for estimation of catalyst lifetime (18,19). The deactivation plot shows the length of the reactionfrontas a function of time. This illustrates the movement of the temperature profile caused by the progressive deactivation of the catalyst. The method is illustrated in Figure 3. The temperature increase over the catalyst bed is calculated as AT = T^t - T^t and a certain percentage hereof, e.g. 90% (AT90) is calculated. The axial distance in the

In Deactivation and Testing of Hydrocarbon-Processing Catalysts; O'Connor, P., et al.; ACS Symposium Series; American Chemical Society: Washington, DC, 1996.

Downloaded by UNIV OF MISSOURI COLUMBIA on September 4, 2014 | http://pubs.acs.org Publication Date: June 6, 1996 | doi: 10.1021/bk-1996-0634.ch013

13. CHRISTENSEN & ROSTRUP-NIELSEN

Adiabatic Prereforming

191

catalyst bed corresponding to this temperature increase, usually called Z90, is then readfromthe temperature profile. The deactivation plot is made by depicting Z90 as the length of the reaction front versus time-on-stream. The inverse slope of the deactivation plot is called the resistance number, R, defined as kg hydrocarbon feed required to deactivate one gram of catalyst. At slow deactivation a large resistance number is obtained and a small resistance number at fast deactivation. The impact of the process parameters on the rate of gum formation has been reported by Moseley et al (19) in terms of the resistance number R. A large resistance number was obtained with high steam-to-carbon feed ratio, high hydrogen partial pressure and with high H/C ratio of the hydrocarbon feed, whereas particularly low temperature caused a decrease in R. A high content of aromatic compounds was also found to decrease R and this can be explained by the aromatic hydrocarbons having high r and low r . A

H

Computational Methods. Mathematical models describing the interaction between pore diffusion, poisoning and catalyst activity are essential for reactor design and for performance prediction. Mathematical models combining reaction kinetic and catalyst poisoning models have been developed and built into the computer programs, REFRAD and SPOIS (1,20). The models have proven useful in describing observed performance of adiabatic prereformers in bench-scale operation as well as in industrial operation (1,6). Experimental Methods for Characterization of Catalyst Deactivation Chemical Analyses. A spent and deactivated catalyst can be characterized by chemical analysis of carbon and poisons (S, Si, K, Na) deposited on the catalyst. The average sulphur coverage, 0 , can be calculated when the total sulphur chemisorption capacity of the catalyst has been determined. The sulphur chemisorption capacity is proportional to the nickel surface area and methods for determining the sulphur capacity by chemisorption of H S have been developed by Rostrup-Nielsen (2). Due to the pore diffusion restriction of the reactions, the sulphur is chemisorbed in the outer shell of the catalyst pellet according to equation 5. The average sulphur coverage of the pellet, 0 , is remarkably lower than the equilibrium coverage, 0, until the chemisorptionfronthas moved to the centre of the pellet. The radial distribution of the sulphur and other poisons can be measured by a scanning electron microprobe. A radial profile of the sulphur coverage in a catalyst pellet takenfromthe top of a prereformer catalyst bed illustrating the shell poisoning nature is shown in Figure 4. For further characterization of the deactivation phenomena, advanced laboratory testing methods such as TPR (Temperature Programmed Reaction) and intrinsic activity measurements are used. av

2

av

TPR (Temperature Programmed Reaction). Characterization of the carbon types formed on catalysts during operation can be made by use of TPR. Tests are performed by heating the reactor linearlyfromambient to 1000°C while flowing pure hydrogen through the catalyst bed (21). The carbon is reacting with the hydrogen and methane is formed as the most predominant compound and is detected as a function of temperature. Carbon morphologies found on prereforming catalysts have been investigated in TPR studies (22). Prereforming catalysts, Ni supported by MgO, that have exhibited various deactivation phenomena during operation in industrial and laboratory adiabatic reactors, have been selected for the investigation. The catalysts had been in operation with feed mixtures of naphtha, steam and hydrogen with a steam-to-carbon ratio in the range 1.5-3.0 and in a temperature range of 400-530°C. The peaks observed on the TPR curves for the prereforming catalysts are characterized by their


192

DEACTIVATION AND TESTING OF HYDROCARBON-PROCESSING CATALYSTS

Length of Reaction Front, z

Temperature ( ° C )


T

< b

9 0

exlt 100%

T ~O 9

75%

1\ •inlet /

j [

50% Z

90

25%

a

6

Bed Distance (m)

12

18

Time On-stream (months)

Figure 3. Graphical deactivation plot for performance prediction. a) Estimation of length of reactionfront,Z90fromtemperature profile b) Deactivation plot

0

0.2

0.4

0.6

0.8

1

Relative Radial Distance from Center

Figure 4. Measured radial sulphur distribution in a prereforming catalyst pellet from top of a bed after 1130 hours of operation with naphtha feed at H 0/C = 1.5 and inlet temperature = 450°C. 2


24

13.

CHRISTENSEN & ROSTRUP-NIELSEN


193

peak temperature; Table II. McCarty et al (21) have by model studies characterized the different carbon morphologies found after exposure of nickel catalysts to carbon monoxide and to ethylene. The peaks listed in Table II have been assigned a carbon morphology according to the nomenclature used by McCarty et al (21). The carbon morphology found reacting at high temperature, 630-720°C, has been assumed related to the platelet/pyrolytic e-carbon described by McCarty et al (21). In total six different classes of carbon morphologies have been observed for the prereforming catalysts exposed to various feeds and operating conditions (22). Downloaded by UNIV OF MISSOURI COLUMBIA on September 4, 2014 | http://pubs.acs.org Publication Date: June 6, 1996 | doi: 10.1021/bk-1996-0634.ch013

Table II

Carbon Morphologies on Adiabatic Prereforming Catalysts found by H -TPR. Heating Rate = 6°C/min 2

Class

Peak Temperature (°Q 190-205 260-300 420-530 570-600 630-720 950-1000

1 2 3 4 5 6

Type

Carbon Morphology N o m e n c l a t u r e a c c o r d i n g t o M c C a r t y et al (21)

a' a &

5 e G

Chemisorbed carbon Chemisorbed carbon Gum / 3 carbon film Filamentous carbon Platelet / Pyrolytic carbon Graphitized carbon

The individual catalyst sample typically shows the presence of 2-4 peaks and Figure 5 shows a TPR diagram for a catalyst that has been in operation for 6 months in an adiabatic prereformer with a naphtha feedstock at relative low temperature. Three peaks have been identified for this catalyst and they are ascribed to chemisorbed carbon (a-type) at 300°C, gum (P-type) at 480°C and the platelet/pyrolytic carbon (e-type) at 650 °C. In addition some graphitized carbon is also observed as a weak peak at high temperature around 950°C. The normal unpoisoned prereforming catalyst contains only a-type and e-type. For prereformer operation it is of special interest to study whether carbon is formed as gum (P-carbon) or as whiskers (8-carbon) dependent on the feedstock, operating conditions and catalyst. Filamentous whisker carbon has only been seen by electron microscopy examinations (TEM) on catalysts with 8-type carbon. Catalysts, that judged on activity measurements are deactivated by gum formation, all have a TPR peak in the range of the P-carbon. The gum peak is often a very broad peak. The broad nature of the gum peak could be caused by variations in the hydrogen content of the polymerization deposits and some transformation with time could also change the peak temperature. McCarty et al (21) found that the peak temperature of the pcarbon increased with increasing coverage. The temperature intervals for the TPR peaks of the prereforming catalysts are for most of the types the same as found in the studies by McCarty et al (21) when differences in the TPR heating rate are considered. The variations seen can most likely be explained by the differences in hydrocarbon feed and exposure conditions. Catalyst support can also effect the peak temperature, especially for the adsorbed surface carbon types (a- and P-carbon). McCarty et al (21) reported large peaks of Ni-carbide (Ni C, y-carbon) when the catalyst was exposed to ethylene at the very low temperature of 300°C. This type of carbon is not expected to be present at the prereforming conditions as it is not stable at temperatures above 330°C, due to decomposition (27), nor has it been identified by X-ray diffraction examinations of prereforming catalysts. 3

Intrinsic Activity Measurements. Measurement of the intrinsic reaction rate is a standard test to characterize prereforming catalysts. The catalyst is exposed to steam



194


reforming of ethane at isothermal conditions in a gradient-less plug flow microreactor (20). The activity measurements can be used for screening tests in development of new catalysts and for measurement of the degree of deactivation. In the catalyst bed of the adiabatic prereformer, the steam reforming reactions are controlled by pore diffusion, so application of intrinsic activity measurements must be done with great caution. The effectiveness factor for reforming of higher hydrocarbons is below 10-15% in most parts of the prereformer. Catalyst samples exposed to prereforming conditions in bench-scale tests have been investigated for their catalyst activity. The samples have been collected from various positions in the catalyst beds and after various operating periods ranging from 50 to 1200 hours. Catalysts with a broad range of sulphur coverages have in this way been obtained. The standard test involves measurement of the intrinsic reaction rate for ethane steam reforming at 500°C and a feed ratio of H 0/C = 4.0 applying crushed down catalyst of 0.3-0.5 mm sievefraction.In Figure 6 the reaction rate is plotted versus (1-0 ). Activities have been normalized with the activity of unpoisoned catalysts. Sulphur poisoning of a steam reforming catalyst takes place as a shell poisoning and the sulphur coverage in the shell is typically 80-90% at prereforrning conditions, as Figure 4 illustrates. The intrinsic activity throughout the pellet is given by equation 6 and the activity in the sulphur poisoned shell is therefore very low. When the activity of such a prereforming catalyst is measured on crushed-down catalysts, it can be shown that the measured intrinsic activity ideally is a straight line versus (l-6 ), as shown on Figure 6. The catalysts are divided into two groups by this activity test method; normal sulphur poisoned catalysts and catalysts deactivated by gum in combination with sulphur poisoning. Some of the catalysts had been exposed to gum formation due to special operating conditions and feedstocks. For these catalysts a larger decrease in the catalyst activity is observed than can be ascribed to sulphur poisoning alone. The reaction rate is still proportional to the sulphur poisoning, but equation 6 is no longer valid. Figure 6 shows that the additional gum deactivation factor on the intrinsic reaction rate is in the range 2-5. The TPR examinations of prereforming catalysts showed that the gum deposits (Pcarbon) are relatively reactive. New methods have been developed combining a standard activity measurement with temperature programmed treatment of the catalyst in hydrogen or a hydrogen-containing gas. These regeneration tests can be used for verification of deactivation type; sulphur or gum. Sulphur is very slowly regenerable in H , whereas gum can be gasified at moderate temperatures. In Figure 7 the regenerative effect of the treatment with pure hydrogen is shown dependent on the regeneration temperature rangingfrom450 to 650°C. All activity measurements were made for steam reforming of methane at 500 °C after 3 hours of treatment at the regeneration temperature. It is seen that a sulphur poisoned catalyst is not regenerable. According to equation 4 sulphur poisoning is reversible, but in practise it is a very slow process at low temperature. Both examples of gum formation shown in Figure 7 are regenerable but it takes a longer time to restore the catalyst activity for a gum poisoned catalyst that has been exposed to long-term operation. It is most likely that the nature of the gum changes with time as the TPR analyses showed that the amount of P-carbon on the 100-hour old gum poisoned catalyst was 50-60% of the amount found on the 6-month old catalyst. 2

8V

av

2

Bench-scale Testing as a Method for Investigation of Catalyst Deactivation Testing in bench-scale equipment is a valuable experimental method for investigation of the impact of deactivation on the catalyst performance when full-size catalyst pellets, industrial mass velocity and pressure are used. Bench-scale testing is often




195


150

Temperature (°C)

Figure 5. TPR (Temperature Programmed Reaction) of an adiabatic prereforming catalyst after 6 months of operation with naphtha feedstock. Heating rate: 6°C per min.

o o

1.2

O O lO

1

>

0.8

I

0.6

Ec

0.4

>

0.2

o < ' /> c

0}

a:

0

• sulphur • gum and sulphur

•

0.2

0.4 1- 0

0.6

0.8

a v

Figure 6. Intrinsic reaction rate in dependence of sulphur poisoning for a) normal catalysts (•), b) gum deactivated catalysts (A). The dotted line represents the ideal relation for the average intrinsic activity of a normal shell poisoned catalyst, 0 = 0.9. Steam reforming of ethane, H 0/C = 4.0, P = 1 bar. Activities normalized with activity of unpoisoned catalyst. 2



196


used to test new process concepts, new catalysts as well as for catalyst performance on special feedstocks (20). Testing of adiabatic processes with large temperature gradients over the catalyst bed in bench-scale units is often misleading due to the heat loss, however, application of pseudo-adiabatic reactors with controlled heat loss compensation ensures reliable adiabatic testing (75). The compensation heat is controlled so the temperature drop over the insulation layer between the reactor and the heater elements is zero, as illustrated in Figure 8 for one heater zone. During catalyst deactivation the reactor wall temperature and the heater temperature will adjust to the level of the catalyst. In this way, the heat-input balances the heat loss. The reactor is equipped with a number of separate heater zones along the catalyst bed to rninimize deviations from adiabatic operation (75). Investigation of deactivation by gum formation and estimation of resistance numbers as a function of operating conditions and feedstock type are examples of bench-scale tests in which an insufficient heat loss compensation may spoil the results. The prereformer performance for a given, heavy naphtha feed (IBP/FBP = 80/205 °C) was investigated in a bench-scale pseudo-adiabatic prereformer at a low inlet temperature of400°C and at a H 0/C-ratio of 2.0. The test showed an extreme case of deactivation by gum formation illustrating what happens when T < T . The temperature profiles are shown in Figure 9. A satisfactory conversion was observed just after start-up; the exit gas contained no higher hydrocarbons and the temperature profile corresponded to the expected conversion of the higher hydrocarbons. Nevertheless, already within thefirst24 hours of operation a fast deactivation took place so the temperature profile had moved far down in the catalyst bed and after only 72 run hours a breakthrough of higher hydrocarbons was observed. The resistance number was only a few per cent of the resistance numbers obtained at higher temperatures for the same feedstock. The test showed that for this given naphtha feedstock the low inlet temperature of 400°C was outside the allowed window of operation. It also shows the importance of having detailed knowledge of the feedstock properties. When gum formation proceeds, the minimum temperature in the catalyst bed decreases with time. This could be explained by a shift in the reaction mechanism so more endothermic reaction steps are prevailing. The decrease in the bed temperature speeds up the deactivation by gum formation. This aspect of gum formation is also seen on the temperature profiles in Figure 9. Calculations with a heterogenous reactor model have shown that the decreasing minimum catalyst bed temperature could also be explained by a change of the effectiveness factors for the reactions. The radial poisoning profiles in the catalyst pellets influence the complex interaction between pore diffusion and reaction rates and this results in a shift in the overall balance between endothermic and exothermic reactions. Another example of the capability of bench-scale testing is a performance test converting a heavy, straight-run naphtha (IBP/FBP = 90/185°C) with 18 wt% aromatic compounds at a very low H 0/C-ratio of 1.5 in an adiabatic prereformer for production of a CO-rich gas (6). The test was carried out in a pseudo-adiabatic reactor and lasted 1130 hours (1V& months). The test was carried out at higher space velocity compared with industrial reactors and thus it simulates the upper part of the prereformer. The displacement of the temperature profile is seen in Figure 10 and it shows that a satisfactory conversion of the hydrocarbons was obtained throughout the test. The graphical deactivation plot has been used for evaluation of the performance, as illustrated in Figure 11. The deactivation is fastest during the first approx. 200 hours whereupon it stabilizes at a lower level. This behaviour is explained by the initial sulphur poisoning of the outer surface of the catalyst pellets having the highest impact on the activity. The overall resistance number has been 2

P

2





197

Figure 7. Regenerative effect of H treatment. Method combining activity measurements for steam reforming of methane (H 0/C = 4.0, P = 1 bar, 500°C) with temperature programmed treatment in H . Activities normalized with activity of unpoisoned catalyst. 2

2

2

Catalyst bed 1 Reactor wall

iMJlOMJl

lnsula,i

°

n

Heater

uuiMjiiin Figure 8. Principle lay-out of the pseudo-adiabatic reactor (one heater zone)



198


540

^> 5 2 0 O o

© 500 2

Catalyst Deactivation in Adiabatic Prereforming - American Chemical

Recommend Documents