Comparative study of gas–solids flow patterns inside novel multi-regime riser and conventional riser

Chemical Engineering Journal 215–216 (2013) 188–201

Contents lists available at SciVerse ScienceDirect

Chemical Engineering Journal journal homepage: www.elsevier.com/locate/cej

Comparative study of gas–solids flow patterns inside novel multi-regime riser and conventional riser Xiaolin Zhu, Chaohe Yang, Chunyi Li ⇑, Yibin Liu, Lu Wang, Teng Li, Qiang Geng State Key Laboratory of Heavy Oil Processing, China University of Petroleum, Qingdao 266580, PR China

h i g h l i g h t s " Systematic investigation of the flow patterns in a novel multi-regime riser. " Detailed description of the local flow structure in the diameter-enlarged section. " Comprehensive analysis of the transient solids concentration signals. " Comparison of key flow features between the novel riser and the conventional riser.

a r t i c l e

i n f o

Article history: Received 29 June 2012 Received in revised form 28 September 2012 Accepted 1 October 2012 Available online 10 November 2012 Keywords: Multi-regime Circulating fluidized bed Diameter-enlarged section Annular air distributor Flow structure Gas–solids interaction

a b s t r a c t A novel CFB riser integrated with an enlarged bottom section was presented in this paper, with the aim to increase overall solids concentration, improve local flow structure and intensify gas–solids contact in the bottom region. Detailed measurements of the flow patterns indicated a multi-regime flow was achieved in this novel riser with a dense-phase bottom region and a dilute upper region. Compared with nonuniform radial profiles of solids concentration and particle velocity presented at the high-density bottom region of a conventional riser, high cross-sectional averaged solids concentration with limited radial gradient was observed in the diameter-enlarged section of this novel riser. Local flow structures in the bottom regions of both risers were also depicted. Better gas–solids contact over the conventional riser was confirmed by the analysis of transient signals and probability distribution of solids concentration. Finally, key flow features of the both risers were summarized. Ó 2012 Elsevier B.V. All rights reserved.

1. Introduction Circulating fluidized beds (CFBs) usually exhibit high gas–solids throughput with the advantages of continuous operation and independent control of gas–solids circulation rate, have been widely applied in many commercial processes, such as fluidized catalytic cracking, F–T synthesis, alumina roasting, coke combustion and gasification [1]. With the advancement of modern industry, the risers that generally serve as reactors for these processes have also undergone constant modifications to meet higher technological standard. Based on comprehensive researches conducted on various CFBs, extremely non-uniform gas–solids suspension flow characterized by core-annulus flow pattern and solids back-mixing near the wall is often observed in the conventional risers [2–4]. Such nonuniform flow structure coupled with low solids concentration ⇑ Corresponding author. Tel.: +86 532 86981862; fax: +86 532 86981718. E-mail address: [email protected] (C. Li). 1385-8947/$ - see front matter Ó 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.cej.2012.10.067

results in segregation of gas–solids phases and reduced gas–solids contact efficiency. Solids back-mixing is greatly reduced under high-flux [5] and high-density conditions [6,7]. The concept of high-density circulating fluidized bed (HDCFB) riser was first proposed by Bi and Zhu [8] and realized in a dual-loop CFB system. Grace et al. [9] defined HDCFB risers as operations under Gs > 200 kg/m2 s and es > 0:1 through the entire riser, and ‘‘dense-suspension upflow’’ (DSU) regime was further proposed to represent the hydrodynamics in these high-density risers. However, obvious radial gradient of solids concentration with local es increasing from less than 0.06 in the central region to about 0.44 near the wall still exists in the aforementioned studies conducted on HDCFB risers, which could still lead to gas bypassing and low gas–solids contacting quality. Similarly, although the radial profiles of solids concentration and particle velocity in CFB downers are more uniform than those in the risers, and virtually no solids back-mixing near the wall, the overall solids concentration in the CFB downers is far less than the risers as a result of much higher particle velocity generated by

X. Zhu et al. / Chemical Engineering Journal 215–216 (2013) 188–201

189

Nomenclature Gs Gs,Local Gs h L r R t Ug Umf V

solids flow rate (kg/m2 s) local solids flux (kg/m2 s) calculated solids flux (kg/m2 s) riser height (m) static bed height in storage vessel (m) radial coordinate (m) column radius (m) time interval (s) superficial gas velocity (m/s) minimum fluidization velocity (m/s) bulk volume (m3)

the accelerating effect of gravity [10,11], and this further restricts CFB downers’ application in processes demanding high solidsto-gas ratio and intensive gas–solids contact. To improve the hydrodynamics and intensify the interaction between gas–solids phases, various new riser structures have been designed, such as the riser with a diameter-enlarged section [12–15] or even coupled with draft tube generating internal solids circulation [16,17], circulating-turbulent fluidized bed (C-TFB) integrated with the advantages of both circulating and turbulent fluidized beds [18,19], and the tapered riser possessing high operating flexibility [20,21]. Due to simple application and unit upgrade, risers equipped with top/bottom-enlarged section have been extensively applied to FCC processes for different processing purposes. A top-enlarged riser is employed in the Maximizing IsoParaffins (MIPs) process for promoting secondary reactions that will simultaneously reduce the olefin content in gasoline fraction and maximize the production of iso-paraffins [22,23]. In contrast, the Two-Stage Riser Catalytic Cracking for Maximizing Propylene yield (TMP) process adopts a bottom-enlarged riser to facilitate the conversion of light olefins reprocessed [12,24]. Only a few researches on gas–solids macro-flow structure in the diameterenlarged riser have been reported, such as Wang’s study [13] and Lu’s simulation work [14] of a top-enlarged riser, and Gan’s investigation [12,17] into a bottom-enlarged riser. While, the analysis of transient solids concentration signals which comprehensively illustrates the interaction between gas–solids two phases has not been covered yet. In this paper, a novel CFB riser integrated with an enlarged bottom section is proposed, with the aim to increase overall solids concentration, improve local flow structure and intensify gas– solids contact in the bottom region without disturbing the riser’s upper flow behavior. This novel riser is designed to satisfy processes demanding for various hydrodynamic and reaction environments by a combined feeding scheme. The reactants with low reactivity and adsorption capacity should be preferentially fed into the bottom diameter-enlarged section to enhance the conversion and avoid competitive adsorption [25,26], while more reactive and adsorptive reactants could be fed into the upper region. Certainly, the feeding sequence can also be adjusted according to different technological requirements. Systematic researches have been conducted to investigate the macro-flow structure and characterize the gas–solids contacting behavior in this novel riser. In general, a dense phase bottom section (operated in bubbling or turbulent fluidization depending on Ug) with better gas–solids contact and a relatively dilute upper region (operated in fast fluidization or pneumatic transportation, also depending on Ug) similar to conventional risers are formed in this novel riser. Therefore, it can be termed as a ‘‘multi-regime riser’’.

Vp

particle velocity (m/s)

Greek symbols qb bulk density (kg/m3) qp particle density (kg/m3) es solids volume concentration es,mf solids volume concentration at minimum fluidization es cross-sectional averaged solids volume concentration rs standard deviation of solids concentration fluctuations s measurement time interval (s)

This study compared the gas–solids flow patterns of the novel multi-regime riser with a conventional riser. Based on local solids concentration and particle velocity measured by an optical fiber probe, macro-flow structure of the bottom dense region was depicted. Moreover, the transient solids concentration signals were also analyzed to evaluate the gas–solids interaction. Finally, the key flow features of both investigated risers were summarized.

2. Experimental 2.1. Experimental apparatus The research was conducted in a circulating fluidized bed system as shown in Fig. 1. The CFB system is made of Plexiglass, and consists of a riser (total height of 10.6 m) with diameterenlarged bottom section, a storage vessel with solids inventory kept at about 650 kg, a measuring vessel for determining the solids flow rate Gs, a cyclone and a bag filter for separating and collecting the fine particles. Further details of the experimental apparatus are listed in Table 1. After passing through the riser, the solids were separated with gas in the cyclone and descended to the storage vessel. The solids were then transferred back into the riser, while the fine particles that escaped from the cyclone were collected in the bag filter. The whole riser could be divided into three different sections: (1) the pre-lifting section (0.0–0.8 m) with a bottom multi-hole air distributor (about 15% open area); (2) the diameter-enlarged section (0.8–2.6 m), where an annular pipe air distributor (Fig. 1b) with upward nozzles was fixed at the bottom; and (3) conveying section (2.6–10.6 m) with a consistent diameter throughout its height, where four nozzles were fixed symmetrically at the bottom. For the tests of the conventional riser, diameter-enlarged section was substituted by a section with diameter equal to pre-lifting section and conveying section. Moreover, four nozzles were also fixed symmetrically at its bottom. The maximum air flow rate of the root blower was 660 Nm3/h, capable of supplying sufficient air for the investigated operating conditions. Air (0.18 MPa) from the root blower was introduced into the riser (operated at ambient conditions) at three different places: (1) through the bottom air distributor at the pre-lifting section; (2) through the annular pipe air distributor at the diameter-enlarged section; and (3) through the four symmetrically fixed nozzles at the conveying section. The detail sizes of the three different air distributors are listed in Table 2. In this paper, the volume flow rate ratio of three-way gas injection was 0.39:1.04:1.00, and the ratio was kept constant when changing the total superficial gas velocity Ug (respect to the conveying section). The specific air injection flow rates of the three air distributors under different

190

X. Zhu et al. / Chemical Engineering Journal 215–216 (2013) 188–201

Fig. 1. Schematic diagram of the circulating fluidized bed system.

superficial gas velocities are listed in Table 3, which are in accordance with all the tests. In addition, a small amount of air was also

introduced into the storage vessel for maintaining minimum fluidization condition.

191

X. Zhu et al. / Chemical Engineering Journal 215–216 (2013) 188–201 Table 1 Detail sizes of the circulating fluidized bed system. Item

Height (m)

Diameter (m)

Multi-regime riser Pre-lifting section Diameter-enlarged section Conveying section Conventional riser Storage vessel Measuring vessel Cyclone

10.6 0.8 1.8 8.0 10.6 6.0 2.5 1.0

0.1–0.2 0.1 0.2 0.1 0.1 0.5 0.25 0.5

Table 2 Detail sizes of the three different air distributors. Bottom multi-hole air distributor Pore number Pore diameter, mm Open area,% Distance between two pores, mm

245 2.5 15.3 5

Middle annular pipe air distributora Annulus diameter, mm Nozzle number Nozzle inner diameter, mm Nozzle direction

160 20 4 Upward

Upper feed nozzles Nozzle number Nozzle inner diameter, mm Nozzle direction

4 20 30° from vertical

a For the tests of the conventional riser, the middle annular pipe air distributor was substituted by four symmetrically fixed nozzles, of which detail dimensions were identical to the upper feed nozzles.

Table 3 Specific air injection flow rates of the three air distributors under different superficial gas velocities. Superficial gas velocity, m/s (respect to conveying section)

Air volume flow rate, Nm3/h Bottom multihole air distributor

Middle annular pipe air distributora

Upper feed nozzles

4 6 8 10 12 14

18.2 27.3 36.4 45.5 54.6 63.6

48.4 72.7 96.9 121.1 145.3 169.6

46.4 69.6 92.8 116.0 139.2 162.4

a For the tests of the conventional riser, the middle annular pipe air distributor was substituted by four symmetrically fixed nozzles, however, the specific air injection flow rates were kept the same.

tion velocity Umf calculated from Leva’s correlation [1] is 0.0021 m/s, and the particle size distribution is listed in Table 4. After passing through an inclined pipe (60° from horizontal), on which a flapper valve was installed to regulate the solids flow rate, the solids entered the pre-lifting section at a height of 0.2 m above the bottom air distributor, and then were accelerated by air at ambient conditions. 2.2. Measurement methods Reflective-type optical fiber probes are commonly used for measuring local solids holdup in fluidized beds [27], with considerable advantages including inhibition of the possible strong interferences on the overall flow structure and almost free of interference of temperature, humidity, electromagnetic fields and electrostatics. Specifically the PV-6 optical fiber probe, developed by the Institute of Process Engineering, Chinese Academy of Sciences, Beijing, China, was used to simultaneously measure transient solids volume concentration and particle velocity in multiregime riser at fifteen axial positions as shown in Fig. 1a (h = 0.53, 1.10, 1.31, 1.55, 1.74, 1.94, 2.14, 3.35, 4.20, 5.35, 6.18, 7.26, 8.09, 9.01 and 9.76 m), and eleven radial positions (r/R = 0.0, 0.16, 0.38, 0.5, 0.59, 0.67, 0.74, 0.81, 0.87, 0.92 and 0.98) for each axial level. In the case of conventional riser, there were twelve axial measuring positions (h = 0.53, 1.31, 1.74, 2.14, 3.35, 4.20, 5.35, 6.18, 7.26, 8.09, 9.01 and 9.76 m) with the same eleven radial positions. The probe was 4 mm in diameter and consisted of two subprobes with an effective tip area of 1 mm  1 mm for each. The gap between two sub-probes was 1.67 mm, and each one was composed of 25 lm-diameter light-emitting and receiving quartz fibers arranged in alternating array. Also, 0.2 mm glass film covers were placed over the two sub-probe’s tips to prevent solids from occupying the probe’s blind zone. During the tests, the light reflected by the particles was transmitted to two photo-multipliers through the fibers, and then converted into voltage signals and transferred to a PC for further processing. Calibration is required since the relationship between voltage signals and solids volume concentration is not linear. A calibration process with the same particles used in this research was conducted as the method proposed by Zhang et al. [28]. Finally, the voltage signals were converted to solids volume concentrations by the calibration equation. Theoretically, the cross-sectional averaged solids concentration es should be calculated as Eq. (1). In this study, the radial measurement positions excluding the center point were obtained by dividing the column cross-section into ten portions with equal area and determining the mid-points of these portions. Therefore, the es could be calculated by averaging local es at ten radial positions (excluding center point).

es ¼

Table 4 Size distribution of the FCC particles. Particle size (lm)

Volume fraction (%)

0–40 40–63 63–80 80–100 100–126 126–159 159–200

1.0 13.9 21.0 26.0 21.2 12.1 4.8

FCC catalysts with mean diameter of 90 lm and particle density of 1500 kg/m3 were used in this research. The minimum fluidiza-

Z

1

pR

2

R

2pr es dr ¼ 0

Z 0

1

r r  2es d R R

ð1Þ

In principle, particles moving vertically align one above the other, passed through two sub-probes and generated similar signals with shift in time. Cross-correlating the two signals in series gave the time delay, and further yielded the velocity of moving particles. To derive the general solids flow trend, the velocities of upflow and downflow particles were determined according to the forward and backward correlation of the signals, respectively. And the net particle velocity was calculated by subtracting the negative velocity from the positive velocity. To increase the reliability of the measurement and calculation, the sampling time s was set at about 15 s with sampling frequency of 50 kHz for pre-lifting and diameter-enlarged section, and 100 kHz for conveying section. At least five repeated measurements were carried out.

192

X. Zhu et al. / Chemical Engineering Journal 215–216 (2013) 188–201

Table 5 Comparison of Gs determined by measuring vessel and Gs calculated in conveying section (Ug = 12 m/s). Item

Conventional riser

Multi-regime riser

Gs determined by measuring vessel, kg/m2 s Gs calculated at h = 6.14 m, kg/m2 s Relative error, %

301

299

326

317

8.3

6.0

A switch valve was used to collect circulating particles of volume V in the measuring vessel at a certain time interval t for determining the solids flow rate Gs (respect to the conveying section) as

Gs ¼

qb V pR2 t

ð2Þ

For a more comprehensive understanding of the flow patterns in both risers, local solids flux Gs,Local was calculated from the corresponding local time mean solids concentration es and net particle velocity Vp as Eq. (3). Furthermore, the overall mean solids flux Gs calculated from Eq. (4) was integrated across the diameter-enlarged cross-section for multi-regime riser and the riser cross-section for conventional riser.

Gs;Local ¼ qp es V p Gs ¼

Z

1 2

pR

R

2prGs;Local dr ¼ 2

o

ð3Þ Z o

1

r r  R R

qp es V p d

ð4Þ

In stable operation, both Gs determined by the measuring vessel and Gs calculated in the conveying section have been listed in Table 5, the relative error is less than 10%, demonstrating the measured solids concentration and particle velocity are relatively reliable. Experiments under different operating conditions (Ug = 8 and 12 m/s, Gs = 200 and 300 kg/m2 s) were conducted to characterize the hydrodynamic behavior of this novel multi-regime riser, especially in the diameter-enlarged section. Moreover, comparative study on the conventional riser was also carried out. During each experiment, the air flow rate was first set for the three air distributors, and the storage vessel was brought to minimum fluidization. The solids flow control valve was then opened allowing the solids to flow into the riser. After the system reached steady state, the optical fiber probe was inserted into the riser column at different elevations, and artificially moved from r/R = 0 to 0.98 to measure the local solids concentration and particle velocity. Once the measurement was finished, the switch valve was activated for a certain time interval to determine the solids flow rate Gs. 3. Results and discussion 3.1. Gas–solids flow patterns 3.1.1. Solids circulation rate in the riser Under certain superficial gas velocity Ug, the maximum solids circulation rate Gs realizable in a CFB riser is strongly influenced by the geometry and the total solids inventory [8,29]. With fullyopening of the solids flow control valve, variations of solids circulation rate Gs with superficial gas velocity Ug in both investigated risers under different static bed heights L in the storage vessel are plotted in Fig. 2. For comparison, the solids saturation carrying capacity calculated from the equation of Bai and Kato [30] is also plotted in Fig. 2. As a result of significant gain in saturation carrying capacity of gas phase, the Gs first increased rapidly with increasing Ug in both

Fig. 2. Variation of solids circulation rate with superficial gas velocity.

investigated risers, and the Gs deviated little from the solids saturation carrying capacity. However, once Ug exceeded 10 m/s, the increase of Gs slowed down, and the difference between the Gs and the solids saturation carrying capacity became significant, indicating that the increase in solids entrainment rate overtook the solids feed rate and the Gs was restricted by the solids feeding system. Another reason for the Gs realizable being much less than the solids saturation carrying capacity might be the combined gas feeding scheme, which further restricted the effective carrying capacity of gas phase. To obtain a solids circulation rate up to 300 kg/m2 s, the superficial gas velocity should be higher than 8 m/s. Therefore, the superficial gas velocities of 8 m/s and 12 m/s were chosen for the following tests. For higher available pressure head, the Gs achieved higher values with increasing solids inventory in both risers investigated. To maintain the high-flux operation, the static bed height L in storage vessel was kept at approximately 3.6 m during the following study. In addition, under the same solids inventory, the Gs realizable in the conventional riser was constantly higher than the multi-regime riser, and this might be attributed to higher pressure loss in the diameter-enlarged section, possibly indicating a higher solids concentration in this novel riser. 3.1.2. Axial solids concentration distribution Thorough researches on solids distribution are essential for better understanding of flow patterns in circulating fluidized beds. Axial profile of solids concentration may be linear, exponent, S-type or C-type depending on the diameter of the riser, operating conditions, solids properties, secondary air injections and outlet structures [31–34]. Based on the assumption of symmetric solids distribution, cross-sectional averaged solids concentration was calculated by averaging local es at ten radial positions (excluding center point) in the present research, and the axial profiles of cross-sectional averaged solids concentration s in multi-regime riser and conventional riser under different operating conditions are displayed in Fig. 3. Different flow regimes have been established in this novel riser. A dense gas–solids suspension with solids volume concentration ranging from 0.28 to 0.40 is discovered in diameter-enlarged and pre-lifting section as well as a typical pneumatic transport flow regime in conveying section where solids volume concentration is less than 0.11. In the conveying section, the solids concentration gradually decreased with bed height, and then remained steady demonstrating a fully developed gas–solids flow. For air injection through four symmetrically fixed nozzles, the solids concentration at the bottom of conveying section was much lower than that of

X. Zhu et al. / Chemical Engineering Journal 215–216 (2013) 188–201

the upper region. A slight increase of solids concentration was also found in the riser outlet due to its restraint effect. In both the diameter-enlarged section of the multi-regime riser and the section of the conventional riser at the same height, the solids concentration decreased gradually with bed height. But for the secondary air injection through the middle distributor, a low density bottom was observed in both investigated risers. In the case of the diameter-enlarged section of this novel riser, the solids concentration was much higher than that of the conventional riser in which the cross-sectional averaged solids concentration was constantly below 0.3. After analyzing the radial solids distributions integrated in Fig. 3 and presented in Fig. 4, it was easy to conclude that the main difference of solids distribution lies within the center of diameter-enlarged section. However, there was no significant difference for the axial and radial profiles of solids distribution in pre-lifting and conveying section. As a result, the following discussion is mainly focused on the gas–solids flow patterns in the diameter-enlarged section and the conventional riser at identical bed height. One of the key advantages of this novel riser over any existing risers was the relatively high solids density at the bottom region coupled with uniform radial solids distribution, which could demonstrate certain superiority in processes demanding high solids-togas ratio, vigorous gas–solids contact and uniform heat transfer.

193

3.1.3. Radial solids concentration distribution and corresponding standard deviation The effects of operating conditions on radial solids distributions in the diameter-enlarged section of the multi-regime riser (h = 1.31, 1.74 and 2.14 m) were also examined. As displayed in Fig. 5, under all tested operating conditions, the solids concentration increased monotonically from 0.1 to 0.3 at the center to approximately 0.43 near the wall. Due to higher solids holdup with extensive fluctuations in the central region, the radial profiles of solids concentration in the diameter-enlarged section was more uniform than the bottom dense region of the conventional riser. Yan and Zhu [31,35] have systematically studied the effect of riser diameter on the solids concentration distribution under fast fluidization regime. All these results indicated that the solids concentration was higher and the radial solids concentration profile was much steeper for the larger-diameter riser. The same effect of riser diameter on solids distribution in riser bottom region has been reported by Zhu and Zhu [36]. The simulation work conducted by Lu et al. [37] also demonstrated that the solids concentration at the wall increased with increasing riser diameter. In our novel multi-regime riser, due to the lower superficial gas velocity in the diameter-enlarged section, a dense phase bottom region operated in bubbling or turbulent fluidization was established. Precisely because of the different flow regimes [38–40], the radial

Fig. 3. Axial and radial profiles of solids concentration in both investigated risers (s = 15 s, Umf = 0.0021 m/s).

194

X. Zhu et al. / Chemical Engineering Journal 215–216 (2013) 188–201

Fig. 4. Radial profiles of solids concentration in the conveying section of both investigated risers (s = 15 s, Umf = 0.0021 m/s).

Fig. 5. Influence of operating conditions on radial solids distributions and standard deviation in the diameter-enlarged section of the multi-regime riser (s = 15 s, Umf = 0.0021 m/s).

X. Zhu et al. / Chemical Engineering Journal 215–216 (2013) 188–201

solids concentration distribution was more uniform in the diameter-enlarged section of this novel riser. Evidently, higher superficial gas velocity led to a steeper radial solids distribution, but solids flow rate had no significant influence on radial uniformity. With the bed height increasing from 1.31 m to 2.14 m, the solids concentration decreased to some extent at the center, but changed little near the wall, demonstrating that the solids concentration was mostly saturated in this region. Radial profiles of standard deviation of solids concentration in corresponding areas were also plotted in Fig. 5. Compared with the conventional riser, of which the maximum standard deviation occurred at middle regions where radial position r/R = 0.6–0.8 [38,39], radial profiles of standard deviation were different in this novel riser. Standard deviation indicating gas–solids fluctuation remained severe in the central region, but decreased continuously towards the wall. However, for the higher superficial gas velocity 12 m/s, the extent of fluctuation decreased to some extent in the central region for axial position of 2.14 m. Furthermore, the variation trend of standard deviation fitted well with that of solids concentration. The higher the local solids concentration was, the less vigorous the fluctuations were. This is probably related to the high local solids concentration that took up the space for gas–solids fluctuations [39]. Detailed correlation between the solids concentration and the corresponding standard deviation in both risers will be further discussed. Relatively higher cross-sectional averaged solids concentrations with limited radial gradient in the diameter-enlarged section could contribute to the better gas–solids interactions in this novel riser. Coupled with more intensive gas–solids and inter-particle interactions, this novel riser would further lead to a much better contacting and mass transfer quality. 3.1.4. Radial net particle velocity distribution Monitoring the particle velocity is an important approach to determine the local solids flow structure, and the residence time of particles in fluidized beds also depends heavily on the local particle velocity. Fig. 6 displays the radial profiles of net particle velocity in the diameter-enlarged section of the multi-regime riser and the conventional riser at same bed heights. The maximum net particle velocity at the bed axis of both risers shared a common trend that it decreased monotonically with radial position moving outwards towards the wall. Moreover, the net particle velocity increased along the axial height, especially in the central region, indicating the flow development of gas–solids suspension. Obvious differences in particle velocity distributions still could be found despite the similarities mentioned above. Caused by the much lower superficial gas velocity and radial gradient of solids distribution in the diameter-enlarged section, the radial profiles of particle velocity in this novel riser were more uniform than that in the conventional riser on the same bed heights. Consequently, the gas phase velocity was also better distributed at radial position. On the contrary, the non-uniform radial profiles of particle velocity in the conventional riser would result in wide residence-time distributions of particles and gas, which were unfavorable for chemical reactions demanding high selectivity [40]. With the consideration of net upflow of particles and overall solids concentration exceeding 0.2, it could be concluded that dense suspension upflow (DSU) [9] regime was achieved in the high-density bottom region of the conventional riser. Compared with the conventional riser in which the particles near the wall mainly flowing upwards, the net velocities of certain particles in the wall region of the multi-regime riser were below zero, indicating more severe solids back-mixing in this novel riser. This mainly occurred at the bottom of the diameter-enlarged section, and became less obvious with the uprising fluid along bed height. Such solids back-mixing phenomenon could be attributed to the poorly

195

distributed air through the annular pipe distributor, which generated remarkable influence on the local solids flow pattern and needed further improvement. Local particle velocity is directly related to local solids concentration, and there is no direct correlation with the axial and radial positions, indicating an inherent link between each other [5,35]. Fig. 7 plots the particle velocities against the corresponding solids concentrations in the bottom dense region of both investigated risers. As displayed in Fig. 7, the local particle velocity decreased gradually with increasing solids volume concentration. Such tendency was probably related to lower effective drag resulted from more intensive solids aggregation at higher local solids concentration. The correlation works of solids concentration and particle velocity in the dense bases of both investigated risers were also carried out. For the conventional riser:

V p ¼ 1:73 lnðes Þ  1:30

ð5Þ

For the multi-regime riser:

V p ¼ 0:71 lnðes Þ  0:55

ð6Þ

The distinguish between these correlations might be resulted from different flow regimes in the bottom dense regions of both investigated risers. Compared with the DSU regime formed in the bottom region of the conventional riser, the turbulent gas–solids flow in the diameter-enlarged section of multi-regime riser greatly inhibited the uprising particle movements, and the formation of turbulent flow could be further confirmed by the analysis of transient solids concentration signals. Moreover, the particle velocity in the multi-regime riser was constantly less than that of the conventional riser in spite of the particle velocity deviated slightly between the investigated risers at dense conditions, and this might also be attributed to the same reason. 3.1.5. Local net solids flux Local net solids flux Gs,Local that calculated from the local solids concentration and particle velocity was divided by over mean solids flux Gs to obtain dimensionless local net solids flux. Radial profiles of dimensionless local net solids flux in the diameter-enlarged section of multi-regime riser and the conventional riser at the same bed heights are plotted in Fig. 8. For all three axial positions in the bottom dense region of the conventional riser investigated, a maximum local net solids flux was observed at a middle region where r/R is approximately 0.7. There was no significant variation in the radial net solids flux profile along the bed height. Furthermore, the local net solids fluxes at all radial positions were positive, suggesting that the high solids concentration and circulation rate in the conventional riser significantly suppressed the downwards movement of the particles near the wall. Compared with the conventional riser, the solids flow at the bottom of diameter-enlarged section in this novel riser was more uniform, where the maximum local net solids flux occurred at the bed axis. One possible explanation could be that the particles coming from the pre-lifting section possessed certain momentum and mainly passed the diameter-enlarged section through the central region. With the flow developing along the bed height, the local solids concentration decreased gradually, especially in the central region as discussed above. And this phenomenon further resulted in a non-uniform profile of local net solids flux similar to the conventional riser in the upper region of diameter-enlarged section. In general, relatively uniform profiles of local net solids flux with maximum occurred at the bed axis or the middle region were observed in the investigated risers. Due to different operating conditions and riser geometries, U-shape profile only for dilute

196

X. Zhu et al. / Chemical Engineering Journal 215–216 (2013) 188–201

Fig. 6. Comparison of radial net particle velocity distributions at three different axial positions under the same operating condition (Ug = 12 m/s, Gs = 300 kg/m2 s; s = 15 s, Umf = 0.0021 m/s).

conditions [41] and ‘‘hook-shaped’’ profile in the riser bottom reported by Malcus et al. [42] were not formed here. In addition, the negative local net solids flux near the wall combined with the former discussion of local particle velocity further verified the solids back-mixing in the wall region and the necessity of improving annular air distributor. If not, the slow particles updating rate and resulted long residence time could lead to serious coking problem and turn up in the industrial application of this novel riser.

Fig. 7. Variation of particle velocity with solids concentration at bottom dense regions of both investigated risers under the same operating condition (Ug = 12 m/s, Gs = 300 kg/m2 s; s = 15 s, Umf = 0.0021 m/s).

3.1.6. Local flow structure Based on the discussions above, both solids distributions and flow structures in the diameter-enlarged section of multi-regime riser (except for bottom and upper transition region, i.e., the gray zones in Fig. 9, which needed further investigation) and the conventional riser with corresponding heights were obtained. As illustrated in Fig. 9, both of the investigated risers possessed a dilute base for the secondary air injection and the solids concentration decreased gradually with the uprising fluid along the axial direction. However, significant difference in solids distribution and flow structure still existed in the investigated risers.

Fig. 8. Comparison of dimensionless local net solids fluxes at three different axial positions under the same operating condition (Ug = 12 m/s, Gs = 300 kg/m2 s; s = 15 s, Umf = 0.0021 m/s).

X. Zhu et al. / Chemical Engineering Journal 215–216 (2013) 188–201

Extremely non-uniform radial flow with a dilute central region and highly concentrated wall region has been established in the conventional riser. Despite of the high particle velocity in the central region of the conventional riser, the solids concentration was extremely low for the segregation of gas–solids phases, and this resulted in a relatively low local net solids flux at the axis of riser column. Moreover, despite the solids concentration was elevated near the wall, the particle velocity was close to zero as a result of the wall-effect, therefore, the local net solids flux was also very low in this region. Consequently, radial net solids flux distribution with a middle peak was formed in the conventional riser. Compared with the bottom dense region of the conventional riser, solids phase in the novel design was much denser, especially in the central region. As a result of lower superficial gas velocity Ug and less severe segregation of gas–solids phases in the diameterenlarged section, the radial solids distribution was also more uniform in the novel riser. Furthermore, solids back-mixing which became less severe along the bed height was observed in the wall region of diameter-enlarged section. This phenomenon was confirmed by the analysis of both local particle velocity and net solids flux, indicating further improvement of the annular air distributor was required. Despite some solids back-mixing occurred in the novel riser, the radial profiles of local net solids flux at the base of the diameter-enlarged section were still more uniform than that in the conventional riser because of the smooth transition structure connecting pre-lifting section and diameter-enlarged section, which would further facilitate gas–solids contact and mass/heat transfer. 3.2. Micro-flow behaviors 3.2.1. Characteristics of transient solids concentration signals Besides the overall solids distribution and flow structure, transient solids concentration signals are also worth detailed investigation for illustrating the interaction between gas and solid phases,

Fig. 9. Schematic diagrams of solids distribution and flow structure in both investigated risers.

197

which remarkably affects the gas–solids contact quality in fluidized beds [43,44]. Transient signals of solids concentration at three radial positions (r/R = 0.00, 0.59, 0.87) in the diameter-enlarged section of multi-regime riser and the conventional riser on the same bed height (h = 1.74 m) are plotted in Fig. 10. Higher solids concentrations with steeper corresponding fluctuations were observed in this novel riser, suggesting different flow regimes in these fluidized systems. In the conventional riser, although there were some high value peaks illustrating the formation of clusters, the solids concentrations at the bed axis (r/R = 0.00) and middle region (r/R = 0.59) were quite low, which indicated that the dilute gas–solids suspension dominated these regions. Moving outwards towards the wall (r/R = 0.87), more clusters were formed and the fluctuations of transient solids concentration signals became less intense and finally began to die down. Based on the discussion above, it could be concluded that the gas phase tended to pass through the central region of the conventional riser, while the particles mainly concentrated near the wall. Such non-uniform radial flow structure resulted in severe gas by-passing, which further inhibited the mixing and contacting between the gas–solids phases. In contrast, the gas–solids contacting behavior in this novel riser was greatly improved, which was contributed by the high solids concentration in the central region accompanied with vigorous fluctuations. Moreover, although the fluctuations became less intensive towards the wall because of denser suspension, they were still more frequent than the conventional riser. Such high amplitude and frequency gas–solids fluctuation similar to turbulent flow regime [38,43] would lead to stronger inter-particle collisions, which further suppressed the separation of gas phase from solids phase. Thus, gas by-passing was reduced in this novel riser and the gas–solids contacting quality was improved, providing one more advantage over the conventional riser. By means of fast Fourier transform (FFT) of the transient solids concentration signals displayed in Fig. 10, the corresponding amplitude spectra demonstrating the periodicity of the signal fluctuations are obtained, and the results are plotted in Fig. 11. Compared with the bubbling regime characterized by the amplitudes distributed over a small range of frequency with dominant frequency about 2 Hz for group A particles [38,45], a broad amplitude spectrum with dominant frequency less than 1 Hz similar to the turbulent flow regime [46,47] was observed in the diameterenlarged section of the multi-regime riser. Combined with the relatively higher amplitudes than the conventional riser operated in the DSU regime, it could be inferred that the gas–solids flow in the diameter-enlarged section was probably dominated by the turbulent regime. As for the bottom dense region of the conventional riser, the amplitudes were distributed over a wide frequency range and no clear dominant frequency was observed, suggesting a less degree of periodic particle motion than the novel riser. In addition, the overall amplitude of fluctuation signals in the conventional riser achieved its maximum value some distance from the wall (the amplitude spectrum at wall is not given), being different from the multi-regime riser where the maximum overall amplitude was obtained in the bed axis (consistent with the radial standard deviation displayed in Fig. 5). Fig. 12 further demonstrates the difference between gas–solids suspensions in the bottom dense regions of both investigated risers by plotting the standard deviation against the corresponding solids concentration. It is apparent from Fig. 12 that the fluctuation was closely related to the suspension density in both risers investigated. At low suspension density, the standard deviation increased with rising solids concentration, but did not increase infinitely. For more space being taken up by the solids phase, the gas–solids fluctuations became less intensive as the solids concentration approached es,mf. The maximum fluctuations were obtained at

198

X. Zhu et al. / Chemical Engineering Journal 215–216 (2013) 188–201

Fig. 10. Comparison of transient solids concentration signals in both risers at the same axial position (h = 1.74 m) and under the same operating condition (Ug = 12 m/s, Gs = 300 kg/m2 s; s = 15 s, Umf = 0.0021 m/s).

es  0.27 in both risers, close to the observation of Issangya et al. [6]. Quadratic correlation works of the dense bottom regions of both investigated risers were also conducted. For the conventional riser:

rs ¼ 1:99es ðes;mf  es Þ

ð7Þ

For the multi-regime riser:

rs ¼ 2:54es ðes;mf  es Þ

ð8Þ

Although a certain degree of scatter was observed for solids concentration above 0.2, these equations still fitted well with the data in both risers. In addition, for a certain solids concentration, the fluctuations were more intensive in the multi-regime riser than the conventional riser, further confirming that different flow regimes dominated the gas–solids flow in separated risers. Such vigorous gas–solids fluctuations in this novel riser would further result in improved gas–solids contacting quality. Abundant information of gas–solids flow behaviors was included in the transient signals of solids concentration. To get a deeper insight into the flow pattern, further statistical analysis of the transient signals was carried out. 3.2.2. Comparison in probability density distribution Probability density distribution (PDD) analysis is an effective statistical method to investigate the phase structures and the interaction between phases. This method was also conducted in the present research. Typical PDD curves of three radial positions

(r/R = 0.00, 0.59, 0.87) in the diameter-enlarged section of multiregime riser and the conventional riser on the same bed height (h = 1.74 m) are illustrated in Fig. 13. The peak with lower solids concentration in PDD curves stood for the dilute phase, while the peak with relatively higher solids concentration represented the dense phase [36,44]. In general, single-peak curves illustrated that the gas–solids flow was dominated by only one density distribution profile, and bi-peak curves indicated the coexistence of dilute and dense phase. In the bed axis (r/R = 0.00) of the conventional riser, there was only one peak with low solids concentration in the PDD curves, and this narrow range of solids concentration distribution represented dilute gas–solids suspension mainly composed of gas phase. On the other hand, in the central region of the multi-regime riser, two major peaks were observed, including one representing the dilute phase and another standing for the dense phase. Such bi-peak PDD curve indicated that a large amount of particles dispersed into the dilute phase and the dense phase was formed even in the central region of this novel riser. Although bi-peak PDD curves were generated in both middle regions (r/R = 0.59) of the investigated risers, the solids concentration distribution of the novel riser was wider and flatter with more particles distributed in the dense phase. Furthermore, moving outward towards the wall (r/R = 0.87), the PDD curve shifted to peaks with high solids concentration for the conventional riser, illustrating that the flow was dominated by dense phase. With regard to the multi-regime riser, the high solids concentration peaks near the wall was much

X. Zhu et al. / Chemical Engineering Journal 215–216 (2013) 188–201

199

Fig. 11. Comparison of amplitude spectra of the transient solids concentration signals in both risers at the same axial position (h = 1.74 m) and under the same operating condition (Ug = 12 m/s, Gs = 300 kg/m2 s; s = 15 s, Umf = 0.0021 m/s).

phases were observed in the multi-regime riser, which would further improve the gas–solids contacting quality. 3.3. Summary of key flow features

Fig. 12. Standard deviation as a function of solids concentration at bottom dense regions of both investigated risers under different operating conditions (Ug = 8 m/s, Gs = 200 kg/m2 s and Ug = 12 m/s, Gs = 300 kg/m2 s; s = 15 s, Umf = 0.0021 m/s).

lower and broader, and possessed a long tail towards the left, which indicated that the dense phase in the wall region contained more gas than the conventional riser. For all the radial positions investigated, wide distributions of solids concentration resulted from a better mixing of gas–solids

Key flow features of the diameter-enlarged section of the multiregime riser, together with the bottom dense region of the conventional riser, are summarized in Table 6. In both investigated risers, additional air was injected into the bottom of conveying section via four symmetrically fixed nozzles, which lifted the particles upwards and entrained particles out of the riser as quickly as possible. Therefore, stable solids fluidization and circulation could be achieved, even with relatively low superficial gas velocity (1–2 m/s) in the diameter-enlarged section. Although further work is still required to restrain the solids back-mixing in this novel riser, certain improvements over the conventional riser for integrating the advantages of turbulent flow regime have been achieved, such as high solids concentration, uniform flow structure, and vigorous gas–solids contacting behavior. And all of these aforementioned features indicate a promising industrial application of this novel riser reactor. 4. Conclusions A novel multi-regime riser characterized by a dense-phase bottom region and relatively dilute upper region was proposed in this work. Hydrodynamic distinctions between the gas–solids suspensions

200

X. Zhu et al. / Chemical Engineering Journal 215–216 (2013) 188–201

Fig. 13. Comparison of probability density distributions of transient solids concentration signals in both risers at the same axial position (h = 1.74 m) and under different operating conditions (Ug = 8 m/s, Gs = 200 kg/m2 s and Ug = 12 m/s, Gs = 300 kg/m2 s; s = 15 s, Umf = 0.0021 m/s).

Table 6 Key flow features of both investigated risers. Position

Bottom dense region of conventional riser

Diameter-enlarged section of novel multi-regime riser

Typical superficial gas velocity (m/s) Maximum solids circulation rate Overall solids concentration (%)

4–8

1–2

Higher Gs for lower pressure loss

Lower Gs for higher pressure loss

20–30

30–40

Non-uniform axial flow structure with high solids concentration

Particle velocity at wall

Non-uniform axial flow structure with relatively low solids concentration Non-uniform radial profiles of solids concentration and particle velocity Positive net particle velocity near the wall

Local solids flux

Middle net solids flux peak

Micro-flow behaviors Gas–solids contact Particle motion Phase structure Flow regime

Segregation of gas–solids phases Relatively chaotic motion Mainly dominated by dilute phase Dense suspension upflow

Macro-flow structures Axial uniformity Radial uniformity

in the conventional riser and this novel riser were comprehensively studied. Results found no significant difference in the dilute upper region and most of the work was concentrated on the different flow patterns in the dense base.

Both solids concentration and particle velocity varying smoothly towards the wall Negative net particle velocity near the wall indicative of solids backmixing Center net solids flux peak gradually varying to middle net solids flux peak Vigorous gas–solids contacting behavior Relatively periodic motion Mainly dominated by dense phase Turbulent regime

Based on detailed measurements of local solids concentration and particle velocity, together with the calculated local net solids flux, general gas–solids flow structures in the diameter-enlarged section of multi-regime riser and the bottom dense region of

X. Zhu et al. / Chemical Engineering Journal 215–216 (2013) 188–201

conventional riser were obtained. Compared with the conventional riser, high cross-sectional averaged solids concentrations with uniform radial distributions in this novel riser might lead to better gas–solids contact and mass/heat transfer. Although some solids back-mixing still occurred in this novel riser, relatively uniform radial profiles of local net solids flux were achieved at the base of diameter-enlarged section, which were different from the middle net solids flux peaks observed in the conventional riser. Besides the more intensive gas–solids contact in this novel riser confirmed by the analysis of transient solids concentration signals, PDD analysis also demonstrated a better gas–solids mixing behavior over the conventional riser, which was supposed to reduce gas bypassing and further improve the gas–solids contacting quality. In general, compared with the dense suspension upflow formed in the conventional riser, remarkable superiorities were achieved in this novel riser for integrating turbulent flow in the diameter-enlarged section. This novel multi-regime riser shows promise for gas–solids two phase processes demanding high solids-to-gas ratio, intense gas– solids contact and excellent mass/heat transfer. In the meantime, further improvement in the annular air distributor is still needed before its industrial application. Acknowledgement The authors are grateful to the National 973 Program of China (No. 2012CB215006) for the financial support. References [1] Y. Jin, J. Zhu, Z. Wang, Z. Yu, Fluidization Engineering Principles, first ed., Tsinghua Univ. Pub., Beijing, 2001. [2] W. Zhang, Y. Tung, F. Johnsson, Radial voidage profiles in fast fluidized beds of different diameters, Chem. Eng. Sci. 46 (1991) 3045–3052. [3] D. Bai, E. Shibuya, Y. Masuda, K. Nishio, N. Nakagawa, K. Kato, Distinction between upward and downward flows in circulating fluidized beds, Powder Technol. 84 (1995) 75–81. [4] T.S. Pugsley, F. Berruti, A predictive hydrodynamic model for circulating fluidized bed risers, Powder Technol. 89 (1996) 57–69. [5] J. Pärssinen, J.X. Zhu, Particle velocity and flow development in a long and high-flux circulating fluidized bed riser, Chem. Eng. Sci. 56 (2001) 5295–5303. [6] A.S. Issangya, J.R. Grace, D. Bai, J. Zhu, Further measurements of flow dynamics in a high-density circulating fluidized bed riser, Powder Technol. 111 (2000) 104–113. [7] F. Wei, H. Lin, Y. Cheng, Z. Wang, Y. Jin, Profiles of particle velocity and solids fraction in a high-density riser, Powder Technol. 100 (1998) 183–189. [8] H. Bi, J. Zhu, Static instability analysis of circulating fluidized beds and concept of high-density risers, AIChE J. 39 (1993) 1272–1280. [9] J. Grace, A. Issangya, D. Bai, H. Bi, J. Zhu, Situating the high-density circulating fluidized bed, AIChE J. 45 (1999) 2108–2116. [10] H. Zhang, W.X. Huang, J.X. Zhu, Gas–solids flow behavior: CFB riser vs. downer, AIChE J. 47 (2001) 2000–2011. [11] R. Deng, H. Liu, F. Wei, Y. Jin, Axial flow structure at the varying superficial gas velocity in a downer reactor, Chem. Eng. J. 99 (2004) 5–14. [12] J. Gan, H. Zhao, A.S. Berrouk, C. Yang, H. Shan, Numerical simulation of hydrodynamics and cracking reactions in the feed mixing zone of a multiregime gas–solid riser reactor, Ind. Eng. Chem. Res. 50 (2011) 11511– 11520. [13] X. Wang, S. Gao, Y. Xu, J. Zhang, Gas–solids flow patterns in a novel dual-loop FCC riser, Powder Technol. 152 (2005) 90–99. [14] B. Lu, W. Wang, J. Li, X. Wang, S. Gao, W. Lu, Y. Xu, J. Long, Multi-scale CFD simulation of gas–solid flow in MIP reactors with a structure-dependent drag model, Chem. Eng. Sci. 62 (2007) 5487–5494. [15] D. Wang, C. Lu, C. Yan, Effect of static bed height in the upper fluidized bed on flow behavior in the lower riser section of a coupled reactor, Particuology 7 (2009) 19–25. [16] C. Yan, C. Lu, Y. Liu, R. Cao, M. Shi, Hydrodynamics in airlift loop section of petroleum coke combustor, Powder Technol. 192 (2009) 143–151.

201

[17] J. Gan, C. Yang, C. Li, H. Zhao, Y. Liu, X. Luo, Gas–solid flow patterns in a novel multi-regime riser, Chem. Eng. J. 178 (2011) 297–305. [18] H. Zhu, J. Zhu, Gas–solids flow structures in a novel circulating-turbulent fluidized bed, AIChE J. 54 (2008) 1213–1223. [19] J. Zhu, Circulating turbulent fluidization – A new fluidization regime or just a transitional phenomenon, Particuology (2010). [20] B. Chalermsinsuwan, P. Kuchonthara, P. Piumsomboon, Effect of circulating fluidized bed reactor riser geometries on chemical reaction rates by using CFD simulations, Chem. Eng. Process. 48 (2009) 165–177. [21] B. Chalermsinsuwan, P. Kuchonthara, P. Piumsomboon, CFD modeling of tapered circulating fluidized bed reactor risers: hydrodynamic descriptions and chemical reaction responses, Chem. Eng. Process. 49 (2010) 1144–1160. [22] Y. Xu, J. Zhang, J. Long, A modified FCC process MIP for maximizing isoparaffins in cracked naphtha, Pet. Process. Petrochem. 32 (2001) 1–5. [23] N. Li, An application of MIP process in FCC unit, Pet. Process. Petrochem. 41 (2010) 27–29. [24] C. Li, C. Yang, H. Shan, Maximizing propylene yield by two-stage riser catalytic cracking of heavy oil, Ind. Eng. Chem. Res. 46 (2007) 4914–4920. [25] A. Corma, F. Melo, L. Sauvanaud, F. Ortega, Light cracked naphtha processing: controlling chemistry for maximum propylene production, Catal. Today 107 (2005) 699–706. [26] J. Verstraete, V. Coupard, C. Thomazeau, P. Etienne, Study of direct and indirect naphtha recycling to a resid FCC unit for maximum propylene production, Catal. Today 106 (2005) 62–71. [27] P. Herbert, T. Gauthier, C. Briens, M. Bergougnou, Application of fiber optic reflection probes to the measurement of local particle velocity and concentration in gas–solid flow, Powder Technol. 80 (1994) 243–252. [28] H. Zhang, P. Johnston, J.X. Zhu, H. De Lasa, M. Bergougnou, A novel calibration procedure for a fiber optic solids concentration probe, Powder Technol. 100 (1998) 260–272. [29] X. Liu, X. Cui, G. Sun, T. Suda, M. Narukawa, Y. Liu, G. Xu, Buildup of high solids flux conveying flow by coupling a moving bed to the riser bottom, AIChE J. 55 (2009) 2477–2481. [30] D. Bai, K. Kato, Saturation carrying capacity of gas and flow regimes in CFB, J. Chem. Eng. Jpn. 28 (1995) 179–185. [31] A. Yan, J. Zhu, Scale-up effect of riser reactors (1): axial and radial solids concentration distribution and flow development, Ind. Eng. Chem. Res. 43 (2004) 5810–5819. [32] X.B. Qi, W.X. Huang, J. Zhu, Comparative study of flow structure in circulating fluidized bed risers with FCC and sand particles, Chem. Eng. Technol. 31 (2008) 542–553. [33] A. Marzocchella, U. Arena, Hydrodynamics of a circulating fluidized bed operated with different secondary air injection devices, Powder Technol. 87 (1996) 185–191. [34] S.K. Gupta, F. Berruti, Evaluation of the gas–solid suspension density in CFB risers with exit effects, Powder Technol. 108 (2000) 21–31. [35] A. Yan, J. Zhu, Scale-up effect of riser reactors: particle velocity and flow development, AIChE J. 51 (2005) 2956–2964. [36] H. Zhu, J. Zhu, Characterization of fluidization behavior in the bottom region of CFB risers, Chem. Eng. J. 141 (2008) 169–179. [37] H. Lu, G. Dimitri, B. Jacques, W. Liu, Hydrodynamic simulation of gas–solid flow in a riser using kinetic theory of granular flow, Chem. Eng. J. 95 (2003) 1– 13. [38] D. Bai, A. Issangya, J. Grace, Characteristics of gas-fluidized beds in different flow regimes, Ind. Eng. Chem. Res. 38 (1999) 803–811. [39] H. Zhu, J. Zhu, Comparative study of flow structures in a circulating-turbulent fluidized bed, Chem. Eng. Sci. 63 (2008) 2920–2927. [40] X. Qi, H. Zhu, J. Zhu, Demarcation of a new circulating turbulent fluidization regime, AIChE J. 55 (2009) 594–611. [41] F. Wei, F. Lu, Y. Jin, Z. Yu, Mass flux profiles in a high density circulating fluidized bed, Powder Technol. 91 (1997) 189–195. [42] S. Malcus, E. Cruz, C. Rowe, T. Pugsley, Radial solid mass flux profiles in a highsuspension density circulating fluidized bed, Powder Technol. 125 (2002) 5–9. [43] Q. Lin, F. Wei, Y. Jin, Transient density signal analysis and two-phase microstructure flow in gas–solids fluidization, Chem. Eng. Sci. 56 (2001) 2179–2189. [44] H. Cui, N. Mostoufi, J. Chaouki, Characterization of dynamic gas–solid distribution in fluidized beds, Chem. Eng. J. 79 (2000) 133–143. [45] D. Bai, J.R. Grace, J.X. Zhu, Characterization of gas fluidized beds of group C, A and B particles based on pressure fluctuations, Can. J. Chem. Eng. 77 (1999) 319–324. [46] A.S. Issangya, Flow Dynamics in High Density Circulating Fluidized Beds, PhD Thesis, University of British Columbia, Vancouver, Canada, 1998. [47] V. Kashkin, V. Lakhmostov, I. Zolotarskii, A. Noskov, J. Zhou, Studies on the onset velocity of turbulent fluidization for alpha-alumina particles, Chem. Eng. J. 91 (2003) 215–218.