You are on page 1of 9

Powder Technology 155 (2005) 17 – 25

www.elsevier.com/locate/powtec

Identification and characteristics of different flow regimes in a


circulating fluidized bed
Esmail R. Monazamb, Lawrence J. Shadlea,*, Joseph S. Meia, James Spenikb
a
National Energy Technology Laboratory, U. S. Department of Energy, 3610 Collins Ferry Rd., Morgantown, West Virginia 26507-0880, United States
b
REM Engineering Services, PLLC, 3537 Collins Ferry Rd., Morgantown, West Virginia 26505, United States

Received 26 July 2004; received in revised form 11 March 2005; accepted 15 March 2005
Available online 21 June 2005

Abstract

A series of experiments was conducted in a 0.3-m diameter circulating fluidized bed (CFB) cold model to evaluate the operating flow
regimes and their transitions. A single unambiguous experimental method was developed to identify the transitions between CFB operating
regimes. Experiments were conducted at riser gas velocities ranging from dense phase turbulent, through fast fluidization (S-shape riser
pressure profile), and up to dilute-phase flow regimes. A transient method was applied to a low density, Geldart Type B, cork bed material.
Two distinct transition velocities were found by analyzing the time required to empty out all solids from the riser of the CFB after cutting off
solids flow. The lowest transition velocity marked the transition between the dense-phase turbulent and the fast fluidization flow regimes,
while a higher or second transition represented the transition between the fast fluidization and the dilute-phase flow regimes. Based on the
experimental results, the axial pressures and its fluctuations along the riser exhibited markedly distinct profiles in each of the three different
operating flow regime regions as defined by these two transport velocities.
D 2005 Elsevier B.V. All rights reserved.

Keywords: Transport velocity; Circulating fluidized bed; Flow regimes; Gas solids flows; Transient method; Fast fluidization

1. Introduction the gas velocity approaches the terminal velocity of a single


particle, U t, turbulent fluidization occurs which is accom-
Circulating fluidized beds (CFB) processes have widely panied by high particle elutriation rates. When the recircu-
been utilized in many different applications including lation of solids to the bottom of the riser becomes necessary
combustion, gasification, catalytic cracking, aluminum to keep a given inventory of solids in the bed, fluidized beds
oxide calcinations, etc. Understanding of the flow regimes operate in the circulating fluidized bed (CFB) regimes.
in CFB risers is important because different flow regimes Numerous studies on CFB riser regimes are available in
provide vastly different gas – solid mixing, and thus different the literature e. g.,[2 – 11]. There is a lot of controversy in
chemical reaction rates [1]. Therefore, choosing the correct the literature on the actual events occurring in different flow
gas velocity at which to transport the solids is the key to regimes and the boundary between these regimes [12]. A
successful design, scale-up and operation of CFB systems. qualitative fluidization map was initially proposed by
A classical bubbling fluidized bed regime is formed Yerushami et al. [13] and, later refined by Yerulshalmi
immediately above the gas distributor when the gas velocity and Cankurt [2]. According to Yerulshalmi and Cankurt,
exceeds minimum fluidization velocity, U mf, in conventional transport velocity is defined as the velocity at which it is
gas –solid fluidization of Geldart’s Group B particles. When possible to carry all of the solids fed into the riser out again,
and thus it is impossible to maintain a fluidized bed without
continuous recycle of solids back into the fluid bed. This is
* Corresponding author. Tel.: +1 304 285 4647; fax: +1 304 285 4403. the critical gas velocity defining the transition between
E-mail address: lshadl@netl.doe.gov (L.J. Shadle). turbulent and fast fluidization flow regimes. Li and Kwauk
0032-5910/$ - see front matter D 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.powtec.2005.03.019
18 E.R. Monazam et al. / Powder Technology 155 (2005) 17 – 25

[3] developed a regime map by plotting the voidage against developed for saturated carrying capacity [15]; the basic
superficial gas velocity. The transition from low velocity to difference is that the latter uses a riser emptying transient
high velocity fluidization was poorly characterized. Grace after halting solids flow rather than bed filling transient. The
[4] proposed a unified flow regime diagram that shows the time dependent data gathered using the bed emptying
operating ranges of conventional fluidized bed, spouted method have been recently reported to reproduce the
beds, circulating fluidized beds and transport systems. steady-state, axial, solids flow profiles and to provide a
However, the transition of different flow regimes was not measure of the effects of solids flux and gas velocity on
clearly delineated. Bi and Grace [5] presented a flow regime flow development along the riser [16]. Both techniques can
map for conventional fluidized bed and dilute phase be used to find the relationship between gas velocity and
transport. Their flow regime map was lacking of transitions solids flux for the onset of fast fluidization (S-shaped axial
to dense-phase transport. Mori et al. [6] defined the transport solids fraction profile), but neither identifies the range or
velocity as the velocity above which no interface exists upper and lower bounds, in terms of gas velocity, for the fast
between a dense bed and an upper dilute freeboard at any fluidized regime. The purpose of this study was to identify
solids circulation rate. the upper and lower bounds for the fast fluidization regime.
Therefore, there is lack of good understanding in the
literature regarding the gas –solid flow and the actual events
occurring in different flow regimes. In this study, a new 2. Experiment
method is presented to determine the limiting velocities that
demarcate the turbulent, fast fluidized, and homogeneous or Experiments were conducted in a flanged steel riser with
dilute flow regimes. The method employed was an 1.22-m acrylic section of 30.48 cm ID and 15.45-m height.
adaptation of Perales method [14] based on measuring the The test unit configuration is described by Monazam et al.
time required to empty all solids from the riser of CFB after [15] and shown in Fig. 1. The main fluidizing air was fed
cutting solids flow under different gas velocities. In the past through a perforated plate into the riser and the reported gas
an operating regime map for a new bed material required velocity was corrected for temperature and pressure as
collecting performance parameters, such as apparent solids measured at the base of the riser. Twenty incremental
holdup, for a large number of steady state conditions differential pressures were measured across the length of the
varying the gas velocity and solids flux. Recently several riser using transmitters calibrated within 0.1% of full-scale
more rapid transient methods have been developed [11,14– or about 2 Pa/m. The sum of the incremental pressure drops
16]. Perales’ method [14] was developed to define the was routinely verified against an overall riser pressure drop
transition between the turbulent regime and fast fluidization.
Different researchers have applied this method, but have
produced some conflicting results. It will be shown in this
study that this technique will also identify a similar φ=0.3 m
transition between fast fluidization and the homogeneous
or dilute regime. It was also recognized that the transition
between fast fluidized and dilute regimes depended upon
more than the gas velocity. Even at very low gas velocities,
if the solids flux is less than the elutriation rate, the riser will
remain very dilute. Bi et al. [17] conducted a comprehensive
literature review classifying choking into three separate
categories depending on the reason for the flow regime
change: Types A, B, and C. Type C is the classical choking
H=15.3 m

when the gas velocity is insufficient to transport the solids


being fed into the riser. This marks the transition between φ=0.25 m
turbulent and fast fluidization regimes. Type C choking
corresponds to Yerulshami’s transport velocity when the
solids flow rate exceeds elutriation rates.
Type B choking is a result of exceeding the head pressure
limit of the gas supply or blower or exceeding solids feed
rate restrictions in the recycle leg [17]. Type A choking is
the transition between dilute and fast fluidization as a result
exceeding the capacity for a given gas flow rate to carry
solids, otherwise known as the saturated carrying capacity.
A method was reported to measure Type A choking by Bai
et al. [11] using the pressure transient at the base of the riser
during solids filling. This is similar to the transient method Fig. 1. Schematic of the NETL cold-flow Circulating Fluid Bed unit.
E.R. Monazam et al. / Powder Technology 155 (2005) 17 – 25 19

Table 1 the solids inventory was carried out of the riser, the pressure
Bed material properties drop across the riser decreased. As expected, an increase in
Cork characteristics: the gas velocity decreased the time required to empty the
qs kg/m3 189 bed. These measurements are useful for determining the
qb kg/m3 95 different flow regimes and the transition between each
eb 0.45
regime for the corresponding gas velocity.
d p50 Am 1170
d sv Am 812 The transport velocity defined by Yerushalmi and
Ut m/s 0.86 Cankurt [2] is an important parameter for the design and
U mf m/s 0.17 operation of a circulating fluidized bed riser since it
e mf 0.50 determines the minimum gas velocity required to bring the
/ 0.69
riser from the turbulent flow regime to a stable fast
fluidization flow regime. In the literature, there are several
measurement that was calibrated within 0.45 Pa/m. This techniques available to determine this transport velocity
riser was connected to a cyclone and standpipe (0.25-m ID) [12]. In this study, the method employed was to measure the
through a solids loop-seal of 0.2-m ID and 1.52-m high. All solids emptying transient for the riser in a circulating
experiments were carried out at ambient temperature. The fluidized bed. This method was developed to determine the
air’s relative humidity was maintained at 20% to minimize transport velocities, and in turn, the different flow regimes
effects of static charge building up on the solids. and their boundaries [14,22]. It was chosen because of its
In this study the bed material used was cork. The material reliability and simplicity. This method measures the time
properties are presented in Table 1. This is a natural wood required to empty out all of the bed material after the solids
product with irregular size and shape that was classified as circulation rate was abruptly stopped while the riser was
Geldart Type B particulate using the method described by operating at a steady state condition at a given gas velocity.
Goossens [18]. The size distribution was normally distrib- This was accomplished by diverting the flows being
uted with particles ranging in size from 500 to 1500 Am as introduced into the base of the standpipe to the atmosphere.
reported earlier [19]. Cork is a light material which When conducting similar tests with denser bed materials
maintains similar gas : solid density ratio compared to that produced sizable pressure gain in the standpipe, it may
pressurized coal gasification systems operating at 1255 K also be necessary to extract compressed air trapped within
and 30 atmospheres. the standpipe bed using a vacuum pump. After the shutoff of
Sun and Grace [20] demonstrated how differential the solids flow, the pressure gradient across the entire riser
pressure fluctuations vary with changes in fluidization was recorded, at a sampling rate of 1 Hz, as a function of
regimes. Differential pressure fluctuations were measured time. The pressure gradients are shown in Fig. 2 as a
using high speed pressure drop measurements at two riser function of time after operating at various steady-state solid
axial positions or elevations. The measurements were made flow rates. As the solids inventory was carried out of the
with short impulse lines over 2-foot intervals in the riser for riser, the pressure drop across the riser decreased.
180 s and recorded at 25 Hz. These data were analyzed by
calculating the normalized amplitude of the pressure
fluctuations using the following equation [20]: 3. Results and discussion
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
N ;; 2 Transient methods have been published to estimate solids
~ DPi  DP =N distribution and solids fluxes [16], Type A and Type C
i¼1
FP ¼ ; ð1Þ
DP
;; 4
Where DP i is the instantaneous pressure differential, DP is 2
Ug(m/s),Gs(kg/m s)
3.5
average pressure differential, and N is the number of data
3 1.83,0.55
points.
2.15,1.18
∆Priser (kPa)

Mass circulation rate was continuously recorded by 2.5


measuring the rotational speed of a twisted spiral vane 2.46,1.93
2
located in the packed region of the standpipe bed [21]. For 1.5
2.82,3.12
the bed empty time tests, while the riser of CFB was 3.36,3.95
1
operated at a steady state condition for a given gas velocity, 3.96,4.90
the solids circulation rate was abruptly stopped. This was 0.5
accomplished by diverting the flows being introduced into 0
the base of the standpipe to the atmosphere. After this solids 0 200 400 600 800 1000
flow cutoff the pressure gradient across the entire riser was Time (s)
recorded as a function of time. These pressure drops were Fig. 2. Decay profiles for overall riser pressure drop for cork after halting
recorded as a function of time at a sampling rate of 1 Hz. As solids flow while maintaining different gas velocities.
20 E.R. Monazam et al. / Powder Technology 155 (2005) 17 – 25

1000 quadratic coefficient. Upon further inspection, the high


velocity data were split into two separate linear regions. In
Emptying Time (s)

800
each region a quadratic model produced a quadratic term
I
600 which was not statistically significant at the 95% CL. The
transition velocity between the medium velocity and highest
400
velocity regions (U g å 3.35 m/s) was found to coincide with
200
II tip of the parabola in the DPM sU g plot developed by
III
Yerulshalmi and Cankurt [2]. Thus, the intermediate regime,
0 Region II, between 2.2 and 3.35 m/s was identified as the
0.0 1.0 2.0 3.0 4.0 5.0 fast fluidized regime, while the high velocity regime above
Ug (m/s) 3.35 m/s was characterized as belonging to the homogenous
or dilute regime, Region III.
Fig. 3. Time required for all the particles to be entrained out of the riser as a
function of gas velocities. The emptying time was relatively independent of the
riser bed inventory. Analysis of several tests in which the
choking [17], as well as the saturation carrying capacity circulation rate was varied for a given gas velocity exhibited
[15]. These methods have been tested favorably against the same emptying time within 2– 3 s.
steady state measurements. These methods only require the The first data set included all the data obtained at gas
assumption that the time rate of change of the riser pressure velocity lower than 2.2 m/s. The second set of data included
drop is equal to the time rate of change of solids flux during all tests with riser gas velocities higher than 2.2 m/s and
cutoff transient. However, this application does not even lower than 3.35 m/s. The third set of data represented all the
require that assumption, since the measurement employed data having riser gas velocities higher than 3.35 m/s. These
here and by Perales [14] was of the total duration to empty three regions are identified in Fig. 3 as Region I, which
all solids from the riser. In this case the assumption is that covers all low gas velocities, Region II, which covers all
the pressure differential across the riser will record an intermediate gas velocities, and Region III, which covers all
elevated value as long as appreciable solids reside in the high gas velocities. Riser gas velocity, U g = 2.2 m/s,
riser and that the value will drop to some zero or baseline signified the initiation of particle entrainment and the onset
value when all the solids are emptied from the riser. In other of particle transport. This velocity is referred to as the
words, the pressure differential across the riser can be taken transport velocity as defined by Yerulshalmi and Cankurt
as a relative measurement; the duration is recorded for the [2], and Bi and Grace [5] refer to this as Type A, or classical
pressure drop to go from an initial steady-state value to a choking. In this paper, we refer to this as the lower transport
constant, empty, value at some point in time after all solids velocity or 1st transition velocity.
flow into the riser was halted. This 1st transition velocity or lower transport velocity
At low gas velocities, there was no significant particle obtained for cork was compared with the predicted values
entrainment so that the time required to empty the bed was based on the available correlations given in Table 2. The
very long. As the gas velocities increased, there was a sharp discrepancy was not surprising since the particle density of
increase in the particle carryover and, hence the time cork (189 g/cm3) was at least three times lower than the
required to empty the bed decreased. Fig. 3 is a plot of lowest particle density used in published literature. The
the solid emptying time against the different riser gas second transition velocity occurred at gas velocity of about
velocities. From this figure three distinctive linear data sets 3.3 m/s. This second transition velocity was considered to
can be clearly identified. Two transition gas velocities can be the upper transport velocity. Below this gas velocity, gas
be identified to separate these three sets of data. Regression reaches its carrying capacity to fully suspend every particle
analysis of the data provided statistical verification of three in the riser and, hence, solids begin to accumulate at the
linear regions. The initial separation of linear regions as lower region of the riser. This critical riser gas velocity is
identified by Perales [14] produced two data sets differ-
entiated by gas velocity (U g å 2.2 m/s). Each of these
subsets was tested separately using a simple quadratic Table 2
regression model to determine if there was any further Comparison of correlations for Yerushalmi and Cankurt’s transport velocity
curvature. In the low velocity data subset the quadratic term Reference Correlation U tr, m/s
was not significant at the 95% confidence limit (CL), and a Bi and Grace [5] Re t = 1.53 Ar0.50 1.90
simple linear model fit the data well. Data points that were Smolders and Baeyen [12] Re t = 1.75 Ar0.468 1.64
Perales et al. [14] Re t = 1.41 Ar0.483 1.51
fit well with this linear model represented a homogenous
Adanez et al. [22] Re t = 2.078 Ar0.463 1.87
subset of low velocity data in the dense flow turbulent Bi and Fan [7] Re t = 2.28 Ar0.419 1.39
regime or Region I. Lee and Kim [25] Re t = 2.916 Ar0.345 1.00
A linear regression model of the high velocity subset Mori et al. [6] Re t = 1.46 Ar0.56 3.09
exhibited definite lack of fit. A quadratic model for the high Chehbouni et al. [26] Re t = 0.169 Ar 0.545 (D / d p) 0.3
1.74
This work 2.2
velocity subset was found to have a statistically significant
E.R. Monazam et al. / Powder Technology 155 (2005) 17 – 25 21

18 2.5
16 2 entry
0.9 kg/m s 2
14 2.3 exit
12 5.6 1.5
Height (m)

Fp
9.5
10 1
11.3
8
6 0.5

4 0
2 0 5 10 15
Gs (kg/m2-s)
0
0 0.1 0.2 0.3 0.4 Fig. 6. Variation of F p as a Function of G s (U g = 1.83 and 2.21 m/s).
Solids fraction (1-ε)
Fig. 4. Solids fraction profiles along the height of the riser for cork using bed with solid fraction of 0.3 at the bottom region of the
different solid flux for low gas velocity I (U g = 1.83 m/s). riser. The solid fractions, then, decreased gradually to
between 0.01 –0.15 dependent upon the solid circulation
also defined by Bi and Grace [5] as the Type A or accu- rate. These axial solid fraction profiles, hence, provided the
mulate choking velocity. evidence that the riser was operated in the turbulent flow
In the present study, a series of steady-state experiments regime.
was conducted to characterize the internal flow behavior of The variation of the average solids holdup at the lower
a riser across a wide range of flow regimes. A series of (entrance) and the upper (exit) region of the riser with the
experiments was statistical designed using a fully duplicated solids flux for a given gas velocity (U g = 1.83) is shown in
6 by 6 factorial test with the riser velocity and mass Fig. 5. The solids holdup did not vary for different solid flux
circulation rate as the independent parameters. The riser gas at the lower region; however, it increased linearly with
velocity was chosen to cover the range of velocity from increasing solids flux at the upper region of the riser. This
below the 1st transition velocity to above the 2nd transition increase in solids holdup at the exit may be attributed to the
velocity. interaction between the fine and coarse solids in the
Both the measured and computed axial solids fraction freeboard. The same qualitative behavior was reported to
profiles below the first transition velocity (< 2.2 m/s) are be due to momentum interchange by collisions between the
plotted in Fig. 4 at various circulation rates. The computed fast-moving fine particles and the slower-moving larger
axial fraction profiles in the riser were obtained from the particles [24].
corresponding differential pressure drop assuming negli- In addition to the axial solids profile and the dynamic
gible contributions due to acceleration and wall friction solids decay rate, the normalized differential pressure
[23]. These axial solids fraction profiles in the riser were fluctuations, F p, of each regime was considered. The value
then calculated based on the following expression [23]: of F p was the highest in Region I, the turbulent regime,
approaching a value of about 1.5 at higher solids fluxes. The
dDP
¼ qs ð1  eÞg: ð2Þ F p appeared to increase with solids flux and then decrease to
dz a low value (Fig. 6). A statistical analysis of the 23 data
Both the experimental and the computed axial solid points recorded in Region I indicated that there was a
fraction profiles in the riser exhibited a comparatively dense marginal dependence on axial position where the pressure
fluctuations were recorded. Only the interactions between
0.50 axial position and solids flux were found to be significant at
the 95% confidence limit (Table 3). This does not reflect a
0.40 (1-ε)entry
Solid fraction (1– ε )

Table 3
0.30 The value of F p in Turbulent regime I U g(1.8, 2.2m/s), G s(0.35, 2.2 kg/
m2-s), position (4.26, 10.35 m)
0.20 Source DF Sum of squares F Ratio Prob > F
Position 1 0.53226 3.3233 0.088
0.10 (1-ε)exit Ug 1 0.12102 0.7556 0.398
Gs 1 0.13465 0.8407 0.374
0.00 Position*U g 1 0.07622 0.4759 0.501
0 2 4 6 8 10 12 Position*G s 1 0.76436 4.7724 0.045
U g*G s 1 0.02836 0.1771 0.680
Gs (kg/m2 s)
Position*U g*G s 1 0.03601 0.2249 0.642
Fig. 5. Variation of solids fractions at the entrance and exit of the riser as a Model 7 1.56687 1.3976 0.276
function of solid flux for low gas velocity I (U g = 1.83 m/s). Error 15 2.40241
22 E.R. Monazam et al. / Powder Technology 155 (2005) 17 – 25

18 2.5
16 2 entry
0.93 kg/m s 2
14 2.4 exit
12 1.5
Height (m)

5.6

Fp
10 9.4
1
8
6 0.5
4
0
2 0 5 10 15
0 Gs (kg/m2-s)
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
Fig. 9. Variation of F p as a Function of G s (U g = 3.23 m/s).
Solids fraction (1-ε)
Fig. 7. Solids fraction profiles along the height of the riser for cork using The average solids holdup at both the lower and upper
different solid flux for intermediate gas velocity II (U g = 3.23 m/s). region of the riser is plotted in Fig. 8 against the solids flux
at a riser gas velocity of 3.23 m/s. The average solids holdup
strong dependence on either axial position or solids flux. at the lower region of the riser increased linearly with the
Thus, the F p did not vary much within this regime as a increasing solids flux to the point where the onset of fast
function of operating parameters or position along the riser. fluidization began and remained constant thereafter. The
The characteristic feature in Region I was dominated by the average solids holdup at the upper region of the riser
high solids holdup and high F p. In this region the solids increased linearly with increasing solids flux to the point
fraction exponentially decayed along the height of the riser where it approached the same solids holdup as in the lower
as the solids were accelerated from the inlet to the outlet region. This is the point where the onset of fully dense-
(Fig. 4). phase flow regime begins.
The axial solids holdup in the riser is shown in Fig. 7 for Region II represents the lowest gas velocities that require
operations between the two transition velocities, i.e. in solids circulation to maintain a fluid bed. The F p dropped by
Region II. The solids fraction profiles provided evidence 50% when going from Region I to Region II, the fast
that the riser progressed from a dilute-phase flow to fast fluidized regime. It decreased dramatically as the gas
fluidization, and then, transitioned to a dense-phase flow velocity increased. The value of F p was nearly constant as
regimes with increasing solids flux at a given gas velocity the solids flux increased within this regime and it was
(U g = 3.23 m/s). These profiles were characteristic of these mostly independent of an axial position in the riser (Fig. 9).
respective flow regimes. At low solids fluxes the constant Within Region II, F p was not affected by the gas velocity at
low solid fraction in the riser was typical for a fully dilute- the 95% confidence limit (Table 4). This is interesting since
phase flow regime. As the solids flux increased up to 5.6 kg/ the solids flux is, for the most part, limited by the saturation
m2-s the riser axial profile became S-shaped as is typical of carrying capacity [15], which in turn is a function of the gas
a riser operated in the fast fluidization flow regime with a velocity. The significance of the solids flux and axial
dense bottom and dilute-phase at the top of the riser. At position was due to the fact that the F p was slightly higher at
higher solids fluxes the constant and higher solid fraction low solids fluxes near the exit, but for the high solids fluxes
profile was characteristic of a riser in a dense-phase flow it was higher near the entrance. This was generally
regime. consistent with the trend for solids holdup in these various
locations with changes in solids flux. These differences
0.20 were reproducible though rather small, and when plotted
along with the rest of the data taken at these velocities were
(1-ε)entry found to change only slightly with either the solids flux or
Solid fraction (1– ε )

0.15 axial position. The higher F p observed at the exit position at


lowest solids flux was a result of the pressure fluctuations
(1-ε)exit
0.10

Table 4
0.05 The value of F p in Fast Fluid Bed regime II U g(2.7, 3.23 m/s), G s(2.16,
5.18 kg/m2-s), Position (4.26, 10.35 m)
Source DF Sum of Squares F Ratio Prob > F
0.00
0 2 4 6 8 10 12 Position 1 0.0002531 8.978 0.0401
Ug 1 0.0000732 2.596 0.1824
Gs (kg/m2 s)
Gs 1 0.0011472 40.688 0.0031
Fig. 8. Variation of solids fractions at the entrance and exit of the riser as a Model 3 0.0014735 17.421 0.0092
Error 4 0.0001128
function of solid flux for intermediate gas velocity II (U g = 3.23 m/s).
E.R. Monazam et al. / Powder Technology 155 (2005) 17 – 25 23

18 Table 5
16 2
The value of F p in Core Annular region III U g (4.05, 5.23 m/s), G s(5.48,
0.93 kg/m s 15.24 kg/m2-s), position (4.26, 10.35 m)
14 2.4
Source DF Sum of Squares F Ratio Prob > F
12 5.6
Height (m)

9.4 Position 1 0.3024 170.38 <.0001


10 Ug 1 0.0010 0.54 0.4673
11.3
8 Gs 1 0.0037 2.06 0.1572
Position*U g 1 0.0326 18.37 <.0001
6
Position*G s 1 0.1110 62.54 <.0001
4 U g*G s 1 0.0565 31.84 <.0001
2 Position*U g*G s 1 0.0009 0.49 0.4867
Model 7 0.5080 40.89 <.0001
0
Error 48 0.0852
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
Solids fraction (1-ε)
because the riser at this gas velocity has no limiting
Fig. 10. Solids fraction profiles along the height of the riser for cork using
different solid flux for high gas velocity III (U g = 3.87 m/s).
saturation carrying capacity.
The flow in Region III can be generally characterized as
remaining constant while the average value approached homogeneous. There was relatively little change in the F p
zero. The result was an exaggerated value for the F p at these when compared to the changes in this same parameter for
very dilute conditions. Thus, we must conclude that in this Regions I and II. The analysis of variance for a general
regime the flow characteristics appeared rather uniform and linear model of F p data taken in this regime demonstrated
independent of solids fraction, gas velocity, and axial no significant dependence on the main factors, circulation
position. rate nor riser gas velocity (Table 5); however, the axial
The axial solids fraction profiles of the riser at gas position was significant. The F p near the top of the riser was
velocity beyond the 2nd transition velocity, U g¨3.35 m/s, larger than that at the lower position (Fig. 12). On the other
(i.e. the riser operated in Region III) are shown in Fig. 10. hand, every one of the two-way interactions were quite
Examination of these solid fraction profiles demonstrates significant in this analysis of variance. This includes the
that the riser was operated in very low solid fraction interaction between position by gas velocity, position by
conditions. These solids fraction profiles with low solid solids flux, and gas velocity by solids flux. While this is not
holdup clearly suggested that the riser was operating in a a strong dependence because of the lack of any significance
dilute-phase flow regime for all the solids flux examined. for the main factors, the F p at a given axial position was
Furthermore, the so-called ‘‘C’’-shape profile was obtained somewhat dependent upon the value of the solids flux and
in this dilute-phase flow regime. These ‘‘C’’-shape profiles the value of the gas velocity. We believe that this was a
were formed due to the reflux of solid particles in the abrupt result of the higher solids fraction near the top of the riser
T-exit at high gas velocities. The variation of the average due to reflux of solids from the blind tee outlet which
solids holdup at both the lower region and the upper region appeared to be a function of the gas velocity and the solids
of the riser are shown in Fig. 11 with the solids flux for the flux. At higher fluxes the solids fraction became higher near
gas velocity of 3.87 m/s. The solids holdup at the lower and the entrance and exit (Fig. 8) resulting in a more pronounced
upper region of the riser increased linearly with the C-shape. Similarly at lower gas velocities the solids fraction
increasing solids flux. The solids holdup increased with becomes higher at the entrance and exit. The analysis of
solid flux in both the riser’s lower and upper regions, variance of the F p has captured these interactions.
Knowing the transition velocities provided the practical
0.25 operating limits to the saturation carrying capacity. Relation-

0.20 2.5
Solid fraction (1– ε )

entry
0.15 2
exit
1.5
0.10
Fp

(1-ε)exit 1
(1-ε)entry
0.05
0.5
0.00
0 2 4 6 8 10 12 0
Gs (kg/m2 s) 0 2 4 6 8 10 12
Gs (kg/m2-s)
Fig. 11. Variation of solids fractions at the entrance and exit of the riser as a
function of solid flux for high gas velocity III (U g = 3.87 m/s). Fig. 12. Variation of F p as a Function of G s (U g = 3.81 and 3.88 m/s).
24 E.R. Monazam et al. / Powder Technology 155 (2005) 17 – 25

6 measured from steady state operations over a range of riser


5
gas velocities provided an objective means to identify the
two transition velocities. The lowest transition velocity
4 marked the transition between the dense-phase turbulent
Gs (kg/s-m2)

and the fast fluidization flow regimes, while a higher or


3 second transition represented the transition between the fast
2
fluidization and the dilute-phase flow regimes.
These transition velocities determine the upper and
1 lower bounds, gas velocities, for the fast fluidization
Ut1 Ut2 regime. For a given bed material, operating at velocities
0 lower than U t1 will be in the turbulent flow regime. This is
2 3 4 5
Ug/Ut
the same as operating below Bi and Grace’s Type C
choking [5]. Operating a riser at velocities above U t2 will
Fig. 13. The saturated carrying capacity for cork showing the limiting lower not produce an S-shape axial pressure gradient profile in
and upper transport velocities, U t1 and U t2, respectively. the riser and operations will be in the dilute flow regime
(core-annulus and homogeneous flow). A riser operating
ships in the literature are reported unbounded [5,15,19]. It is between these transition velocities is in the fast fluidization
reasonable to expect that the saturation carrying capacity regime and the axial solids holdup depends upon the solids
drops to zero at some velocity approaching the transport flux and system inventory: above the saturated carrying
velocity or the 1st transition velocity. The upper limit to capacity (SCC) the riser bed is fully dense, below the SCC
saturation carrying capacity expressions was the transition the riser is dilute, and at the SCC the riser is dense at the
where the dense bed and the dilute became indistinguish- base and dilute above (S-shape). In other words, the state
able. In other words, above the 2nd transition the gas can of the bed in this fast fluidized regime depends upon
carry as much solids as can be fed into the riser. The whether the riser is operated above or below Type A
experimental data on saturation carrying capacity reported choking conditions [5]. Based on the experimental results,
by Shadle et al. [19] are presented in Fig. 13 for this same the axial pressures and the F p along the riser exhibited
cork bed material with the upper and lower bounds clearly markedly distinct profiles in each of the three different
demarcated. Circulation rates below the line are dilute operating flow regimes regions as defined by these two
transport, on or near the line are S-shaped or fast fluidized, transport velocities.
and above this line are in dense transport. Gas velocities
below the lower transport velocity are in a turbulent or
bubbling regime, while flows above the second transition Acknowledgments
velocity are in core-annulus or dilute flow regimes.
The authors acknowledge the Department of Energy for
funding the research through the Fossil Energy’s Integrated
4. Summary Gasification Combined Cycle program. Operations of the
CFB unit were made possible with support from Jim
There is some confusion in the literature on the range of DeVault, Alain Lui, and Todd Worstell.
velocities over which the different fluidization flow
regimes exist, the characteristics of each regime, and on
the existence of different transition velocities. A transient
method is presented here which readily allows one to References
identify operational features and critical transition veloc-
[1] W.C. Yang, AIChE J. 30 (1984) 1025.
ities. A single unambiguous experimental method was [2] J. Yerushalmi, N.T. Cankurt, Powder Technol. 24 (1979) 187.
developed to identify the two transitions between CFB [3] Y. Li, M. Kwauk, in: J.R. Grace, J.M. Matsen (Eds.), Fluidization,
operating regimes. Experiments were conducted at riser gas Plenum, New York, 1980, p. 537.
velocities ranging from dense phase turbulent, through fast [4] J.R. Grace, Can. J. Chem. Eng. 64 (1986) 353.
fluidization (S-shape axial pressure gradient profile in the [5] H.T. Bi, J.R. Grace, Int. J. Multiph. Flow 21 (1995) 1229.
[6] S. Mori, K. Kato, E. Kobayashi, D. Liu, M. Hasatani, M. Hattori, T.
riser), and up to dilute-phase flow regimes. Perales’ method Hirama, H. Takauchi, AIChE Symp. Ser. 9 (88) (1992) 17.
[14] was applied to identify Yerulshalmi’s transport [7] H.T. Bi, L.-S. Fan, AIChE Annual Meeting, Los Angeles, 1991
velocity [2] for a low density, Geldart Type B, cork bed November, p. 17.
material; measuring the time required to empty out all [8] C.S. Teo, L.S. Leung, in: N.P. Cheremisinoff (Ed.), Encyclopedia of
solids from the riser of the CFB after cutting off the solids Fluid Mechanics. Solids and Gas – Solids Flows, vol. 4, Gulf Publish-
ing, Houston, 1986, pp. 611 – 663.
flow. However, when operating over a wide range of gas [9] S.B.R. Karri, T.M. Knowlton, in: P. Basu, M. Horior, M. Hasatani
velocities, not one but, two distinct transition velocities (Eds.), Proceedings of the Third International Conference on Circulat-
were found. Analysis of variance on the emptying times ing Fluidized Beds, Pergamon Press, New York, 1991, p. 67.
E.R. Monazam et al. / Powder Technology 155 (2005) 17 – 25 25

[10] R.C. Zijerveld, F. Johnson, A. Marzocchella, J.C. Schouten, Powder [19] L.J. Shadle, E.R. Monazam, J.S. Mei, in: J.R. Grace, J. Zhu, H. de
Technol. 95 (1998) 185. Lasa (Eds.), Circulating Fluidized Bed Technology VII Ottawa,
[11] D. Bai, A.S. Issangya, J.R. Grace, Powder Technol. 97 (1998) 59. Canada, 2002, p. 255.
[12] K. Smolders, J. Baeyens, Powder Technol. 119 (2001) 269. [20] G. Sun, J.R. Grace, AIChE J. 38 (1992) 716 – 722.
[13] J. Yerushalmi, N.T. Cankurt, D. Geldart, B. Liss, AIChE Symp. Ser. [21] J.C. Ludlow, L. Lawson, L. Shadle, M. Syamlal, in: J.R. Grace, J. Zhu,
74 (1978) 1. H. de Lasa (Eds.), Circulating Fluidized Bed Technology VII, Ottawa,
[14] J.F. Perales, T. Coll, M.F. Llop, L. Puigjaner, J. Arnaldos, J. Cassal, Canada, 2002, p. 513.
in: P. Basu, M. Horio, M. Hasatani (Eds.), Circulating Fluidized Bed [22] J. Adanez, L.F. de Diego, P. Gayan, Powder Technol. 77 (1993) 61.
Technology III, Pergamon, Oxford, 1990, p. 73. [23] M. Louge, H. Chang, Powder Technol. 60 (1990) 197.
[15] E.R. Monazam, L.J. Shadle, L.O. Lawson, Powder Technol. 121 [24] D. Geldart, D.J. Pope, Powder Technol. 34 (1983) 95.
(2001) 205. [25] G.S. Lee, S.D. Kim, Powder Technol. 62 (1990) 207.
[16] E.R. Monazam, L.J. Shadle, Powder Technol. 139 (2004) 89 – 97. [26] A. Chehbouni, J. Chaouki, C. Guy, D. Klvana, Can. J. Chem. Eng. 73
[17] H.T. Bi, J.R. Grace, J.X. Zhu, Multiph. Flow 19 (1993) 1007 – 1092. (1995) 41.
[18] W.R.A. Goossens, Powder Technol. 98 (1998) 48.

You might also like