The Fast Fluidized Bed

0 = overhead vapor. Q = heat transfer rate. Si = steam rate. T = temperature. U = overall heat transfer coefficient. W = accumulation (level) in an ef...
4 downloads 0 Views 843KB Size
x = liquid product mole fraction y = vapor mole fraction z = feed mole fraction

Subscripts v = vapor 1 = liquid f = feed 1 = component1 2 = component2 = perturbation of indicated variable (from steady state) A

Evaporator Example A = heat transfer area B = bottomsflowrate C = mole fraction F = feed rate f = controller function h = liquid enthalpy H = vapor enthalpy 0 = overheadvapor Q = heat transfer rate si = steam rate T = temperature U = overall heat transfer coefficient W = accumulation (level) in an effect Subscripts 1 = first effect 2 = second effect f = feed stream 01 = overhead vapor from first effect

Si = first effect steam feed A = perturbation of indicated variable (from steady state) Superscripts MC = primary feedback controller MM = feed forward uncoupling controller MK = feed forward disturbance controller Literature Cited Andre, H., Ritter, R . A,, Can. J. Chem. Eng., 46, 259 (1968). Cheng, Y. C., Ward, T. J., lnd. Eng. Chem., Process Des. Dev., 14, 193 (1975). Greenfield, G. G., Ward, T. J., h d . Eng. Chem., Fundam., 6, 564, 571 (1967). Hutchinson. J. F., McAvoy, T. J., lnd. Eng. Chem., Process Des. Dev., 12, 226 (1973). Kavanagh, R. J., Trans. AI€€, 76 [Part 111, 95 (1957). Kleinpeter, J. A,, P h D Dissertation, Tulane University, New Orleans, La., 1968. Kieinpeter, J. A,. Weaver, R. E. C., AlChEJ., 17 (3), 513 (1971). Liu, S. L., lnd. Eng. Chem., Process Des. Dev., 6, 460 (1967). Neweii, R. B., Fisher, D. G., "Optimal, Multivariable Computer Control of a Pilot Plant Evaporator", IFACiiFlP 3rd International Conference on Digital Computer Applied to Process Control, Helsinki, Finland, 1971. Newell, R. B., Fisher, D. G., "Implementation of Optimal, Multivariable Setpoint Changes on a Pilot Plant Evaporator", 2nd iFAC Symposium on Multivariable Control Systems, Duesseldorf, Germany, 197 1. Palmenberg, R. E., M.S. Thesis, Clarkson College, Potsdam, N.Y., 1973. Palmenberg. R. E., Ward, T. J., "Regulator Control of a Double-Effect Evaporator", 21st Canadian Chem. Eng. Conf.. Montreal, Canada, 1971. Rich, S. E., Law, V. J., Weaver, R. E. C., "Model Characteristics and the Control of Nonlinear Multivariable Processes", Proceedings, 1974 Joint Automatic Control Conference, 319 AIChE. N.Y., 1974. Ritter, R. A,, Andre, H., Can. J. Chem. Eng., 48, 696 (1970). Wilson, R. G., Fisher, D. G., Seborg, D. E., AlChEJ., 20 (6),1131 (1974).

Receiued for reuieu September 9, 1974 Accepted September 19, 1975

The Fast Fluidized Bed Joseph Yerushalmi,' Davld H. Turner, and Arthur

M. Squires

Deparrment of Chemical Engineering, City College, City University of New York, New York, New York 1003 1

Fast fluidization is a technique for bringing gas at high velocity into intimate contact with a fine solid in an entrained dense suspension characterized by extreme turbulence and extensive refluxing of dense packets and strands of particles. The technique is primarily oriented toward gas-solid reactor applications where it offers several important advantages over the conventional, low-velocity bubbling fluidized bed. This paper delineates the regime of fast fluidization; it records observation of the fast fluidized bed in equipment of transparent walls; and it presents data on the fluidization characteristics of a cracking catalyst in a 3-in. round fast bed. Solid concentrations approaching 25% of the bed volume were typically achieved at gas velocities around 10-15 ft/sec. Corresponding slip velocities were an order of magnitude greater than the free-fall velocity of the largest particle in the test solid.

Introduction Fast fluidization is a technique for bringing a high velocity gas into intimate contact with a fine solid in what is essentially an entrained, dense suspension. The technique is primarily oriented toward gas-solid reactor applications, catalytic and noncatalytic. The solid in the fast fluidized bed may typically occupy up to 25% of the bed volume and is in a state of extreme turbulence marked by extensive refluxing of dense strands and packets of particles. Notwithstanding some commercial experience, the phenomenon of fast fluidization remains virtually unexplored. Fast fluidization offers several important advantages, to be described presently, over contacting of gas and solid in a

conventional, low-velocity bubbling fluidized bed; we have accordingly undertaken its investigation. In this paper we describe the phenomenon of fast fluidization, noting its historical roots and its commercial use to date. We record observations of the fast fluidized bed in a transparent two-dimensional rig, including observations derived from movies filmed at high speed, and we present data on the fluidization characteristics of a cracking catalyst in a round, 3-in. i.d. fast fluidized bed. The closing discussion draws distinctions between the regime of fast fluidization and other gas-solid transport regimes and sets forth the qualities and advantages of the fast fluidized bed for reactor applications. Ind. Eng. Chem., Process Des. Dev., Vol. 15, No. 1, 1976

47

Background T h e Bubbling and Turbulent Regimes of Fluidization. Figure 1 depicts a phase diagram typifying the pressure gradient across a bed of fine powder, solids belonging to group A of Geldart’s classification (Geldart, 1973), as a function of the velocity of a gas flowing upward through the bed. The gradient first increases sharply due to the rise of pressure across the fixed bed of solid. At minimum fluidP ization velocity, the powder begins to expand in particulate fluidization and the pressure gradient becomes essentially equal to the fluidized density of the bed. Beyond the minimum bubbling velocity, bubbles appear and grow as the velocity is raised further. Over the years, “fluidization” has become virtually synonymous with operation in the bubbling regime-corresponding typically to gas velocities around 1 ftlsec. An underlying function of fluidization has often been to afford contact between a gas and a large inventory of solid surface per unit bed volume. Little attention was paid to opportunities of achieving the same end at gas velocities beyond the bubbling regime. An important paper by Lanneau (1960) provided strong evidence for such opportunities. Lanneau studied the fluidization characteristics of a fine powder in a bed 15 ft deep contained in a column 3 in. in inside diameter. Experiments were conducted at two pressure levels, 10 and 60 psig, and at gas velocities to about 5 ft/sec. The character of the bed at a given gas velocity was manifest in recordings obtained from small-point capacitance probes inside the column. Lanneau noted that as the velocity was raised beyond 1 ft/sec, the heterogeneous, two-phase character of the bubbling regime gave way to a condition of increasing homogeneity. At velocities in the range of 3 to 5 ft/sec, the tracings from the probes indicated that a condition of “almost a uniform or ‘particulate’ fluidization was approached”. Lanneau provided an argument that a velocity around 1 ft/sec, where the two-phase, heterogeneous structure of the bubbling bed is most pronounced, is the worst velocity to use from the standpoint of efficiency of contact between gas and solid. Lanneau provided a rationale for what the practical man already knew (Braca and Fried, 1956), that contacting is better a t higher velocity. It is now of course appreciated that in a bed operating in the bubbling regime: (a) a considerable portion of the gas residing in the rising bubbles often bypasses the solid in the bed; and (b) backmixing of gas is appreciable. This cuts conversions, and sometimes promotes undesirable secondary reactions. In addition: (a) experience has taught that bubbling fluidized beds are not easy to scale up; (b) the low fluidizing gas velocities characteristic of the bubbling regime translate to low processing capacities per unit cross sectional area of the bed; and (c) solid mixing is not sufficiently vigorous to allow processing of a solid that does not flow freely or a solid that tends to agglomerate. Because Lanneau worked in steel equipment, it remained for Kehoe and Davidson (1971) to describe the transition from bubbling to what they called the “turbulent regime” of fluidization, and it remained for Massimilla (1973) to provide laboratory research demonstrating the higher contacting efficiency in a turbulent fluidized bed. What Kehoe and Davidson witnessed, as the velocity of the gas flowing through a narrow bed of fine powder was raised, was a breakdown of a slugging regime into “a state of continuous coalescence-virtually a channelling state with tongues of fluid darting in zig-zag fashion through the bed”. The point of breakdown of the slugging regime was not sharp. For several powders with particle sizes below 90 j ~ the , transitions to turbulence were judged to occur be48

Ind. Eng. Chem., Process Des. Dev., Vol. 15, No. 1 , 1976

-I--

FWlOlZED OENSlW IS A UARIUO FUNCTION OF %LID RATE

FWIOIZEO OENSITY NEARLY IUMPLNDWI OF U K l O RATE

..

LOG PRESSURE GRAOIENT

W16HER

m’y

PARTICULATE EXPANSION

LOG VELOCITY

’~

EQuILI~Iu~~LIo CONVEYIw6 RATE AT 3ATURATION“

Figure 1. Fluidization phase diagram for a fine powder, showing different regimes of fluidization.

tween 1 and 2 ft/sec. In Lanneau’s experiments, in which the solid ranged in size from 40 to 100 p , the transition occurred around 3 ft/sec. Lanneau as well as Kehoe and Davidson worked with beds of small diameter. To our knowledge, corresponding experiments conducted in larger fluidized beds have not been reported. Commercial beds of fine powders often operate at velocities around 2 to 3 ft/sec, and descriptions of such beds give rise to the impression that some of them might very well be operating in the turbulent regime. Regenerators in modern fluid crackers typically operate a t 2.5 to 3 ft/sec; considerable carryover occurs, and the bed surface is not well defined, there being a continuous gradation in density from the bottom of the bed to the cyclones (Saxton and Worley, 1970). We believe, at least insofar as fine solids are concerned, that the turbulent regime is not peculiar to vessels of small size, but is an intrinsic regime of fluidization. In what follows we shall so presume. A bubbling fluidized bed displays a distinct upper surface level; in a turbulent bed, the upper surface may be regarded as present but is of course considerably more diffused owing to the large carryover attending operation at higher fluidizing velocities. The carryover from a fluidized bed, operating either in the bubbling or in the turbulent range, depends upon the fluidizing gas velocity and the distance above the bed at which carryover is measured. If this distance is beyond the Transport Disengaging Height, then the carryover is constant, as if the gas were “saturated” with solid. Referring back to Figure 1, as the fluidizing gas velocity spans the turbulent regime, if the experimenter maintains a definite upper bed level at a fixed elevation in the bed, a point would be reached where there is a sharp drop in the bed density over a narrow velocity range, as seen at the right for the lower curve in Figure 1. It will be appreciated that to preserve a bed level, the experimenter must return solid at precisely the same rate that material is carried over from the bed. That is to say, the solid return must equal the carryover a t saturation, as indicated in the figure. Fast Fluidization. If, on the other hand, the experimenter causes solid to flow into the bottom of the fluidized bed at a rate beyond the saturation carryover, the upper bed level will move out of the top of the vessel. The effect of solid rate upon bed density is slight at gas velocities below those associated with the aforementioned sharp drop in density. Addition of solid to a bubbling or a turbulent bed at a rate beyond the saturation carryover will simply cause the vessel to fill up with solid at a slight increase in density. At higher gas velocities, however, the fluidized density becomes a strong function of the solid rate into the bottom of the bed. Lewis and Gilliland (1950) clearly recognized as early as 1940 what we have dubbed the fast fluidized bed condition:

A

dh\/

Table I. Properties of the Catalyst0 -

Particle

diameter, p

Wt %

0-40 40-80

14 56

80-100 100-130 >130

19 9

2

uparticle density = 55 lb/ft3; settled density = 31.9 lb/ft3; volume-surface average diameter = 60 p.

7TO BAG FILTER

1

LOG PRESSURE GRADIENT

ER RATE ~~

3-INCH FAST BED

yp-p

TEST SYSTEM

-

LOG VELOCITY Figure 2. Fluidization phase diagram for a fine powder, showing schematic diagrams of equipment suitable for use in the bubbling, turbulent, and fast fluidization regimes.

“If one will operate at a gas velocity sufficient to blow all or substantially all of the solid material out of the reactor in a relatively short time, provided no fresh solid material be introduced during this time, but will feed into the reactor simulaneously solid material at a sufficiently high rate, one can maintain in the reactor a high concentration of solid granules approaching that of the ‘liquid state’ [of the bubbling fluidized bed] . . . and yet be blowing the solid particles out of the top of the reactor a t a corresponding rate.” Lewis and Gilliland reported a fast fluidized density as high as 10 lb/ft3 for a pulverized clay catalyst (settled density, 35 lb/ft3) a t a superficial fluidizing gas velocity of about 8 ft/sec. Those who have practiced pneumatic transport of solid are likely to be familiar with instances where powders were conveyed under essentially fast fluidized conditions, but examples of use of fast fluidized bed reactors are relatively rare. Early catalytic crackers, operating in the so-called upflow mode (Kunii and Levenspiel, 1969),represented an attempt to utilize the fast fluidized bed, but several mechanical difficulties brought about the use of bubbling fluidization, which Lewis and Gilliland also pioneered, in the cracking process. Although the Synthol reactor at Sasolburg, South Africa, for Fischer-Tropsch synthesis may be a fast bed, it remained for Lurgi Chemie und Huttentechnik GmbH of Frankfurt, West Germany, to appreciate the broad commercial potential of the fast fluidized bed and to realize this potential in two successful commercial processes (Reh, 1971). In the first process, developed jointly by Lurgi and Vereinigte Aluminum Werke A.G., aluminum hydroxide is calcined to provide cell-grade alumina. One of the values of the fast bed can be appreciated from the fact that the first VAW-Lurgi unit has an inside diameter of 12 f t 5 in. for 560 tons/day, whereas an earlier conventional fluidized bed furnace was 22 ft in inside diameter and had a capacity of only 280 tons/day. Lurgi is also marketing the fast bed for the absorption of hydrogen fluoride from effluent gases from Hall cells producing aluminum. Lurgi’s design for the aluminum hydroxide calciner is shown schematically a t the upper right in Figure 2. In contrast, Figure 2 shows equipment for bubbling and turbulent beds having approximate-

COMPANION BED

FLUIDIZING G A 5

LIFTGAS

Figure 3. Schematic of the 3-in. fast bed system.

ly the same gas-treating capacity. The fast bed installation is characterized by a large external cyclone and a standpipe of large diameter for circulation of the solid a t the rate required to maintain the fast bed condition.

Experimental Section Our stand for study of fast fluidization includes a twodimensional fast bed (2 X 20 in. in cross section and 23 ft tall) built in Plexiglas and used primarily for visual observations, and a 3-in. i.d. round fast bed, 24 f t in heigh. A 6-in. fast bed is being commissioned. The beds are blown with air from a compressor that delivers 1200 CFM a t 10 psig. The present program explores the fluidization characteristics of several solids of different sizes, size-distributions, and densities. Here we report observations and data on a fluid cracking catalyst. Properties of the solid are given in Table I. The 3-in. fast bed system is shown schematically in Figure 3. On the left is a 1-ft i.d. “companion”, low-velocity bubbling bed that serves for storage of the solid and control of solid circulation. Solid flows by gravity from the bottom of the companion bed into a well-aerated 3-in. i.d. U-tube that extends without contraction or expansion to the bottom of the 3-in. fast bed seen on the right. Solid rate can be controlled by a butterfly valve installed in the 3-in. tube just below the bottom of the 1-ft companion bed. Having traversed the fast bed, the solid is returned via cyclones to the companion bed. The 3-in. fast bed is constructed in alternating glass and aluminum sections to permit observaInd. Eng. Chem., Process Des. Dev., Vol. 15,No. 1, 1976

49

1

I

I

I

1

I

I

I

I

1

I 46 l

i

A? L0S/FT3

4

e

4

i

‘t

I

I

I

I

10

PO

30

40

I

50

SOLID RATE; LOS/FTZ-SEC

I

I

I

1

10

90

¶O

40

1

50

SOLID RATE; LBS/FT*-SEC

Figure 4. Pressure gradient (across a section extending from 7.5 to 18.5 f t above the bottom of the 3-in. fast bed) vs. solid rate at different fluidizing gas velocities.

Figure 5. Pressure gradient (across a section extending from 2 to 7.5 ft above the bottom of the %in. fast bed) vs. solid rate at different fluidizing gas velocities.

tion of portions of the bed and easy access to other portions. Solid circulation rates are measured with the aid of a sintered-plate butterfly valve installed in the middle of the 1-ft companion bed as follows. At a given moment, the butterfly valve is closed, and the rate of descent of bed level just below the valve is timed. Solid returning to the companion bed through the cyclone diplegs forms a fluidized bed on top of the butterfly valve, which accordingly acts as a distributor for this upper bed. It should be appreciated that, apart from some additional pressure introduced by the porous butterfly valve, the solid head in the companion bed remains constant throughout this procedure. Pressure readings are taken at many locations around the solid circulation loop, and particularly along the fast bed a t elevations of 2, 7.5, 13, 18.5, and 24 ft from its bottom.

continuous. We surmise that the strands and ribbons of the dense phase become linked in a reticulated net of rapidly circulating material that includes many vortices resembling tiny tornadoes. The impression is that gas-solid interaction is on a fine scale. The fast bed appears to the eye to afford intimate contact between a gas at high velocity and a large inventory of solid surface per unit volume of bed. There is no substantial change in the appearance of the fast bed over the 23 ft height of our equipment. In the remainder of this section, we present data obtained in the 3-in. fast fluidized bed. Figure 4 shows the pressure gradient across a middle section of the fast bed (extending from an elevation of 7.5 to 18.5 ft from the bottom of the bed) as a function of superficial gas velocity and solid rate. Solid inventories in the column were not measured, but at the range of conditions shown in Figure 4,the pressure gradient can be taken tentatively to reflect the fluidized density in the section of the bed in question. (We are mindful of the need to explore discrepancies between the measured pressure gradient and the fluidized density, and we plan experiments toward this end.) Figure 4 thus attests to the high solid loadings that can be achieved in the fast fluidized bed. In the middle of the column, solid densities are as high as 13.5 lb/ft3, corresponding to a solid concentration of about 25% by volume. Pressure gradients are normally higher at the bottom section of the bed, and, within the range of gas and solid rates used in our experiments, are lower toward the top of the bed. Figure 5 depicts pressure gradients measured at the bottom of the bed. Figure 6 shows pressure-gradient profiles along the height of bed for three fluidizing gas velocities and several solid rates. Slip velocities are high in the fast fluidized bed. Those corresponding to the data of Figure 4 are shown in Figure 7. It should be appreciated that the theoretical free-fall velocity of the median particle in our test solid (60 p in diameter) is merely 0.25 ft/sec, and that of the largest particle (130 p ) is about 1 fthec. Yet, Figure 7 records slip velocities approaching 10 ft/sec. The large slip velocities in the fast bed are a measure of the high degree of backmixing of solid. More precisely, the high slip velocities arise from the characteristic structure of the solid in the fast bed. A dense packet of particles would naturally have an effective freefall velocity considerably larger than that of a single particle in isolation. If the packet is sufficiently large, it cannot be sustained by the rising gas, it will fall back, and will subsequently undergo disintegration in one manner or another. Presence of glass sections in the %in. column permitted

Results We have viewed the operation of the two-dimensional fast bed by means of high-speed photography. The air velocity was held fixed at 12 ft/sec, and the solid throughput was varied. At low solid throughputs, the solid is conveyed upward in dilute-phase transport, but contrary to the impression created by many discussions of this subject, fine particles are not streaming upward discretely. Even a t very low solid loadings (0.1 lb/ft3) some segregation is apparent. Relatively denser clouds of particles go up surrounded by a more dilute environment. At higher solid loadings, though still within the dilute phase regime, solid segregation becomes more pronounced. Particles throng into vertical streamers which move upward surrounded by a faster moving leaner phase. Some of the streamer or strands weave a bit or even halt momentarily. At a solid loading around 2 lb/ft3, solid backmixing gradually comes into play, and the fast bed is established. The fast bed can be regarded as essentially a dense suspension marked by vigorous and intensive backmixing of solid. At loadings between 3 and 5 lb/ft3, the solid at any moment appears distributed in two phases. Dense strands and ribbons rise and fall and drift from side to side a t high speeds, while the bulk of the column is occupied by particles moving rapidly upward in a more dilute environment. Solid interchange between the two phases appears rapid and extensive; dense strands of solid break apart, some gradually and some in an explosive fashion, as new strands form. At loadings beyond 5 lb/ft3, observation of the details of the structure of the fast bed becomes difficult but suggests that each of the two phases present becomes on the whole 50

Ind. Eng. Chem., Process Des. Dev., Vol. 15, No. 1, 1976

1

GAS VELOCITY. 8.0 F T E K SOLID RATE;

SOLID RATE; ,LBS/FT’-SEC,

15-’ LBSIFT*- SEC,



15& IO-

=\

- I5

37. I 0

50 L I D RATE; ,LBS/FT~-SEC, ,LW FT %EC, Y

43.5 b

\+

+=”I 1

0

.

2

1

A

IA --

A \

4.0

\I:::

vox>:

I

“1

V -\

7.I

*\ 1

I

- IO

D -\

16.3~

5-

I

1

GAS VELOCITY: 14.8 FT/SEC

GAS VELOCITY. 12.5FT/SEC

-

1 I

=-¤

I

I

I

I

_

1

1

10.2 \

?;-- - A

I

I

I

I

Figure 6. Pressure gradient profiles in the 3-in. fast fluidized bed. PI

SLIP VELOCITY;

FT/SEC

./--_.--

c

., .......................... .......’............................. v-c----- *- --. -10.3

8-

-

6k

t + 4-

IGAS VELOCITY;FTISEC

_________._--G15 VELMIITY: --.-14,8Fl/SEC

t IO

,A?

--____ -----__

-

12.5

..IC-----

.;‘

0.0

i

58

1

2 -

0

I I5

oLT 10

AP.

I 30

40

50

S O L I D RATE; L 8 S / F T 2 - S E C

Figure 7. Slip velocities corresponding to the data shown in Figure 4.

us to observe the transition from slugging to turbulent fluidization. What we saw resembles very closely the description provided by Kehoe and Davidson (1971). Further, there does not appear, a t least so far as the eye can tell, any essential difference between the structure of the circulating solid and gas masses in turbulent or fast fluidization. Although observation of bubbling (slugging, in fact) fluidization in the 3-in. column is of course easy, the setup does not allow precise control of solid circulation a t sufficiently low rates to permit observation of strictly batch fluidization, with preservation of a definite upper bed level, a t gas velocities that traverse the range from bubbling through the turbulent regime and into fast fluidization. We have conducted experiments in the bubbling and turbulent regimes a t relatively low solid circulation rates, however, to provide pressure-gradient profiles shown in Figure 8 for gas velocities below and near the transition from turbulent to fast regimes. Combining these data with results for fast bed operation, we can provide a tentative phase diagram showing bubbling, turbulent, and fast regimes for fluid cracking catalyst (Figure 9). Pressure gradients were derived from measurements across the bottom section of the bed. The velocity marking the boundary between the turbulent and the fast regimes falls around 5-6 ft/sec. Transitions from bubbling (slugging, in fact) to turbulent fluidization are judged to have occurred between 2 and 3 fthec. Our 6-in. fast bed setup is better equipped to achieve good control of solid flow at low rates, and will provide an improved version of Figure 9. Discussion We have attempted to inculcate the view that the regime of fast fluidization represents a natural extension of bub-

LOS/FTb

L

00

4.n

21.18

IS.7s

10.21

HEIGHT FROM THE BOTTOM OF THE FAST BED; F T Figure 8. Pressure-gradient profiles a t gas velocities in the buhbling and turbulent range.

I

I

AP.

AL’

LBWFT’

I I

I

L

1 3

1

1

1

1

1

I

I

I

4 5 c 18910 eo GAS VELOCITY; FT/SEC

1

I

.

Y)

40

50

Figure 9. Phase diagram showing fluidization regimes for the fluid cracking catalyst. Ind. Eng. Chem., Process Des. Dev., Vol. 15, No. 1, 1976

51

bling and the turbulent fluidization regimes. In the bubbling fluidized bed, gas is contacted with solid at high concentration. The gas velocity is normally held down to avoid excessive carryover. In the turbulent bed, solid concentrations are somewhat lower, but gas throughputs are higher. The design of a turbulent bed, however, must allow for considerable carryover. In the fast fluidized bed, gas velocities are sufficiently high to effect total entrainment of the particles in a short time, but solid concentrations are still fairly high provided the material is recirculated a t a suitable rate. As Figure 4 illustrates, for a given gas velocity, the solid concentration in the bed increases with solid rate, and in principle it might be possible to achieve solid concentrations comparable to those in the bubbling bed. In practice, however, very high solid rates might not be feasible, and solid densities in the fast bed would generally be smaller (though not necessarily by a large factor) than in a bubbling bed. The lower density of the fast bed need not mean a lower effectioe inventory, in view of the better gas-solid contacting of the fast regime. Moreover, the solid inventory of the fast bed may be varied over a range of values, by increasing the height of the bed, without the deterioration in gas-solid contacting efficiency commonly experienced in the case of deeper bubbling beds. For some applications, the ability of an operator to vary a fast bed’s solid inventory, by changes in solid recirculation rate, may be advantageous: for example, to vary catalyst inventory to compensate for loss of catalyst activity where a degree of conversion must be closely held, or to vary carbon inventory in a fast bed gasifying carbon by steam or carbon dioxide in order to obtain a handle on bed temperature. In bubbling, turbulent, and fast beds, gas is brought into contact with a large inventory of solid surface per unit bed volume. In certain cases it might be desired to contact a gas with only a small amount of solid surface, as in the case of gas reactions in the presence of a very active catalyst, or it might be necessary to minimize the duration of contact between the gas and solid. In such cases, operation in a dilute-phase transport reactor proves usually advantageous. In such reactors, gas velocities are typically very high (50 ft/sec, say) and the solid loading is correspondingly low, often below 1 lb/ft3. Temperature is nonuniform if a large heat effect is present. It is also important to appreciate the distinction between fast fluidization and the riser reactor used in modern petroleum catalytic cracking art (Saxton and Worley, 1970). Gas velocity in the riser increases with height, but typically ranges in excess of 35 ft/sec in the upper portions of the riser. Solid rates through commercial risers range from 100 to as much as 300 lb/ft2-sec and solid loadings are in the neighborhood of 10 lb/ft3 (Matsen, 1975). But the demixing of the gas and catalyst in the riser reactor is gross in scale by comparison with the fine-scale demixing of the fast bed condition. A probe traversing the fast bed would pass every inch or so from a lean void into a strand of solid or vice versa. In contrast, density contours across the riser reactor reveal extreme segregation; high density zones, most likely representing solid flowing downward, appear along the walls, while the core remains dilute (Saxton and Worley, 1970). The riser reactor, unlike fast fluidization, does not appear as a natural extension of bubbling and turbulent fluidization. It will be interesting to explore the way in which the gross demixing of the riser reactor evolves from fast fluidization upon increase in gas velocity. So far, we have considered only the fluidization of fine solids, namely, aeratable solids that fall into group A of Geldart’s classification. The question naturally arises, can coarser particles be maintained in fast fluidization? We do not know the answer. We can think of no reason that would 52

Ind. Eng. Chem., Process Des. Dev., Vol. 15, No. 1, 1976

bar fast fluidization of coarser material in principle, but experience suggests that circulation of a coarse solid a t high rates around a loop might prove difficult to achieve. Coarse material transported pneumatically in vertical lines chokes when the solid concentration exceeds a certain value, and the suspension collapses into slug flow. In contrast, fine solids can be maintained in dispersed, dense flow without any gross instabilities (Yousfi and Gau, 1974). It should, however, be remarked that many of the impressions regarding choking in vertical pneumatic transport have arisen from experiments that were conducted in equipment of small size. Experiments in equipment of larger size, especially with coarse solids, are expensive, but are needed to answer the question at hand. To close this paper, we discuss the advantages of the fast fluidized bed.

Advantages of the Fast Fluidized Bed. The fast bed enjoys all the qualities that brought the bubbling fluidized bed into widespread use. We may note the following. (1)The temperature is uniform throughout the fast bed. This is a consequence of the high degree of solid mixing in the fast bed. Lurgi records a uniform temperature throughout its 60-ft high fast bed calciner. Schmidt (1973) published temperature profiles obtained in a Lurgi pilot scale calciner. Though bed densities were relatively low (about 4 lb/ft3), temperature throughout the fast bed was essentially constant. (2) The fast bed is capable of bringing a cold solid or gas feed almost instantaneously to the bed temperaure. (3) Heat transfer rates to walls and immersed surfaces in the fast bed are comparable to those for a bubbling fluidized bed. This has been recently demonstrated in trials conducted at Battelle Memorial Institute (Kiang et al., 1975). However, in addition, the fast bed holds several advantages over the bubbling fluidized bed. (4) High processing capacities. Vessel diameter is a far more important cost factor than vessel height, especially in applications at elevated pressure. In light of this fact, the capacity advantage of the fast bed is obvious from a glance at Figure 2. ( 5 ) Our impression is that the fast fluidized bed affords excellent contact between gas and solid. Lanneau’s argument (1960) and Massimilla’s data (1973) support this impression, as do observations by Wainwright and Hoffman (1974) of excellent contacting for the oxidation of orthoxylene in what was probably a fast fluidized bed. Lurgi (Reh, 1972) also reports that absorption of hydrogen fluoride from Hall cells showed that the fast bed produced an effluent containing on the order of one-tenth of the H F found in effluent from a conventional bubbling fluidized bed. (6) Despite the backmixing of solid in the fast bed, we believe that the high gas velocities preclude appreciable backmixing of gas, so that the operation might approach a plugflow condition. (7) The fast fluidized bed ought to handle cohesive solids that might otherwise be difficult to fluidize in the bubbling bed. (8) The fast bed might prove easier to scale up than a bubbling fluidized bed. It may be noted that Lurgi’s scaleup of its calcining plant was extremely smooth. Scale-up was made in two steps: A 5-in. i.d. unit was followed by a pilot 3 f t in inside diameter and 26 f t tall. Both the pilot and the first commercial plant achieved design operation within a few weeks of start up. At any rate, because of the fast bed’s capacity, it might often not be imperative to scale the equipment to uncomfortably large dimensions.

Acknowledgment Work on fast fluidization is supported by Grant AER72-03426 A No. 4 from the RANN Program ("Research Applied to National Needs") of the National Science Foundation. Valuable discussions with L. Reh are gratefully acknowledged. M. J. Gluckman supervised the installation of the compressor and assisted generously with advice. The City College Chemical Engineering Shop under J. Bodnaruk constructed the experimental setup. A. E. McIver assisted in the installation of the equipment. Literature Cited Braca, R. M.,Fried, A. A,, "Operation of Fluidization Processes," in "Fluidization," D. F. Othrner. Ed., Reinhold. New York, N.Y., 1956. Geldart, D.. Powder Techno/., 7 , 285 (1973). Kiang, K. D., Liu, K. T., Nack, H.. Oxley, J. H., "Heat Transfer in Fast Bed", paper presented at the International Conference on Fluidization, Asilornar. Calif., June 1975.

Kehoe, P. W. K., Davidson, J. F.. Inst. Chern. Eng. (London) Syrnp. Ser., 33, 97 (1971). Kunii, D., Levenspiel. O., "Fluidization Engineering," Chapter 2, Wiley, New York, N.Y., 1969. Lanneau, K. P., Trans. Inst. Chern. fng., 38, 125 (1960). Lewis, W. K., Gilliland. E. R.. U.S. Patent No. 2,498,088 (Feb 21, 1950). Massirnilla, L., AIChE Syrnp. Ser., 69, No. 128. 11 (1973). Matsen, J. M.. "Some Characteristics of Large Fluid Solids Circulation Systerns", paper presented at the International Conference on Fluidization, Asilornar, Calif., June 1975. Reh, L., Chern. €ng. Prog., 67(2), 58 (1971). Reh, L., personal communication, 1972. Saxton, A. L., Worley. A. C., OiIGasJ., 68(20), 82 (1970). Schmidt, H. W., "Combustion in a Circulating Fluid Bed", in Proceedings of third International Conference on Fluidized-Bed Combustion, Environmental Protection Technology Series, EPA-65012-73-053, p 11-1-1, Dec 1973. Wainwright, M. S.,Hoffman, T. W., Adv. Chern. Ser., 133, 669 (1974). Yousfi, Y., Gau, G., Chern. Eng. Sci.. 29, 1939 (1974).

Received for reuiew September 17;1974 Accepted July 21,1975

Degradation and Isomerization Reactions Occurring during Alkylation of lsobutane with Light Olefins Bharat Doshi and Lyle F. Albright* Purdue University, West Lafayette, Indiana 47907

Trimethylpentanes and dimethylhexanes degrade and isomerize when contacted with sulfuric acid at conditions similar to those used in commercial alkylation reactors. The products obtained include isobutane, all isoparaffins formed during the alkylation of isobutane with light (C,-C,) olefins, and hydrocarbons (or conjunct polymers) that dissolve in the acid phase. lsobutane also slowly reacts in the presence of sulfuric acid to form the same isoparaffins. Kinetic equations have been developed for each C8 isoparaffin and isobutane. Degradation reactions are apparently important relative to acid consumption, and suggestions to minimize them are presented. Important features of the alkylation mechanism have also been clarified.

Trimethylpentanes, whose research octane numbers vary from 100 to almost 110, degrade and isomerize when contacted with concentrated sulfuric acid. Reactions of 2,2,4and 2,3,4-trimethylpentanes (2,2,4-TMP and 2,3,4-TMP) were investigated a t 25OC by Hofmann (1964) and by Kramer (1967a), respectively. About 19% of 2,2,4-TMP reacted during a 10-min run producing isobutane, light ends (C, to C7 hydrocarbons), other trimethylpentanes (TMP's), dimethylhexanes (DMH's), and heavy ends (C, and higher hydrocarbons). The isoparaffins formed were the identical ones produced during alkylation. Kramer found that 2,3,4T M P degraded a t a faster rate than 2,2,4-TMP. He also found that increasing the water content of the acid from 1.5 to 11.1%caused lower rates of degradation. The degradation products from both 2,2,4- and 2,3,4T M P often had similar compositions implying that a common intermediate, probably a trimethylpentyl cation (TMP+), was formed. Sulfuric acid is a sufficiently good oxidizing agent (Kramer, 1967b) so that the ion intermediates will be formed from a trimethylpentane (TMP) as follows TMP

+ 4H,SO,

TMP*

--+

+ 2H,O+ +

3HS01- + SO2(1)

Although these investigations give some indications of the reactions that probably occur during alkylation, the conditions that have been investigated are relatively differ-

ent from those employed in commercial alkylation reactors in at least three respects. (a) 25OC is significantly higher than 10 to 15OC usually employed commercially. (b) The sulfuric acids employed as acid feedstocks contained 1.5 to 11 wt % water but no dissolved hydrocarbons. The acids employed in commercial units always contain some (up to several percent) dissolved hydrocarbons which when separated from the acid (Miron and Lee, 1963) are often called conjunct polymers. The amount of both dissolved hydrocarbons and water in the sulfuric acid have been found to be of considerable importance during alkylation in affecting the quality of the alkylate produced (Mosby and Albright, 1966; Li et al., 1970; and Albright et al., 1972). Furthermore, there is reason to believe that dissolved hydrocarbons were formed in the investigations of both Hofmann and Kramer since the hydrogen-to-carbon ratios in the degradation products reported were higher than that of the feed TMP. These dissolved hydrocarbons that have a lower hydrogen-to-carbon ratio would explain this observation. (c) In commercial units, the alkylate is mixed with isobutane that is used in large excesses. Isobutane may have an effect on the degration reactions. More information is needed relative to degradation and isomerization reactions for at least three reasons. First, degradation and consequent decrease in the quality of the alkylate may be of significance in a t least some commercial Ind. Eng. Chem., Process Des. Dev., Vol. 15, No. 1, 1976

53