Powder Technology 276 (2015) 134–143

Contents lists available at ScienceDirect

Powder Technology

j ourna l homepage: www.e lsev ie r .com/ locate /powtec

Pressure drop and pressure fluctuations in spouted beds with binarymixtures of particles

Wei Du a,b, Lifeng Zhang c, Bo Zhang b, Shuhui Bao b, Jian Xu b, Weisheng Wei b,⁎a State Key Laboratory of Heavy Oil Processing, China University of Petroleum, Beijing 102249, PR Chinab The Key Laboratory of Catalysis, China National Petroleum Corp., China University of Petroleum, Beijing 102249, PR Chinac Department of Chemical and Biological Engineering, University of Saskatchewan, Saskatoon, Canada

⁎ Corresponding author. Fax: +86 10 89734979.E-mail address: weiws@cup.edu.cn (W. Wei).

http://dx.doi.org/10.1016/j.powtec.2015.02.0160032-5910/© 2015 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 10 November 2014Received in revised form 4 January 2015Accepted 8 February 2015Available online 17 February 2015

Keywords:Spouted bedFlow regime transitionPressure drop fluctuationStatistic characteristicsPower spectral analysisBinary mixtures

When handling fine particles in spouted beds, addition of coarse particles has shown improved spouting stabilitythan single particle systems. However, segregation may still occur because of insufficient mixing in binary mix-tures, which will adversely influence the process performance. Therefore, in this study, analysis of pressure dropand its fluctuation signals were for the first time used to understand mechanisms of flow regime transitions inspouted beds with binary mixtures. The results showed that the typical varying sequence of pressure drop canbe observed for spouted bed with binary mixtures and the peak pressure drop is related to the mixing degreeof particles, which is mainly influenced by the inter-particle forces between fine particles and their counterpartcoarse ones. The statistic characteristics of pressure drop time series, i.e., average value, standard deviation andprobability distributions, were found to vary for different flow regimes. Therefore, they could be used for thecharacterization of these flow regimes. The spouting stability of binary mixtures can be reflected by powerspectrum analysis; the influences of particles size and density difference on spouting stability were discussedthrough power spectral analysis.

© 2015 Elsevier B.V. All rights reserved.

1. Introduction

Spouted beds, as an alternative to fluidized beds for handling coarseparticles larger than 1 mm in diameter (typically Group D particlesaccording to the Geldart classification) have been widely employed invarious physical operations such as drying, coating and granulation[1]. Spouted beds also possess some unique structural and flow charac-teristics of great potential applications as chemical reactors [2–10].However, themain factor impeding their wider use as chemical reactorsis the limited interfacial area because of use of relatively large particles,leading to lower conversions [11]. In particular, they are not suitable forbeing used in a mass transfer limited process where only the externalcatalyst surface is effective [5]. Therefore, operating the spouted bedwith relatively smaller particles (such as Group B particles accordingto the Geldart classification) is considered to be a remedy to increase in-terfacial areas and enhance conversions while the desirable spoutingcharacteristics remain intact.

Spouted beds operated with Group B particles have larger gas solidcontact areas, thus leading to increased conversions. In the literature,it has been reported that the spouting of Group B particles is significant-ly different from that of GroupD particles and a stable spouting can onlybe achieved under strict conditions [12–15]. It has been shown that

adding coarse particles can effectively improve the stability of spoutedbeds with fine particles [16–19]. In spouted beds, as noted by Huilinet al. [20], the solid–solid drag force is caused by particle collisions.The internal friction for mono-sized particles is considered to be relatedto the granular temperature, which takes both the particle velocity fluc-tuations and particle collisions into account according to a kinetic theo-ry of granular flow (KTGF) [21]. When adding coarse particles, thisstress can be greatly decreased as can be explained by KTGF. However,excessive addition of coarse particles to the spouted bed is no longerbeneficial and particle segregation is observed, thereby giving rise to adecrease in the spouting stability. Therefore, an advanced understand-ing of spouting formation, mixing behavior and flow regime transitionmechanisms is still lacking. In the literature, a few diagnostic toolshave been employed to analyze flow regime transitions. In general,they can be classified into three categories: direct measurement (visualobservation or advanced instrumentation such as PIV, LDV etc.), probemeasurement (pressure probes or optical fiber probes, etc.) and X-raymeasurement. Among those tools, pressure measurement is the mostcommonly adopted one due to its robustness, ease in use, and economicadvantages.

The analysis of pressure fluctuations, mainly related to motionswithin the bed, has beenwidely used for decades for identifyingflow re-gimes in fluidized beds. Various analysis methods have been describedin detail in published comprehensive reviews [22–25]. In general,there exist three methods for analyzing pressure signals, which are

135W. Du et al. / Powder Technology 276 (2015) 134–143

time domain method, frequency domain method, and state spacemethod. The time domain method is typically the first step in the dataanalysis, where either the standard deviation or the average absolutedeviation is often used to identify a flow regime change [26–30]. Thestatistical analysis in time domain is the simplest and the mostcommonly employed; it is also very fast and easily applicable. Themost commonly usedmethod in time domain is to study the amplitudeof signals, expressed as a standard deviation (viz., square root of second-order statistical moment). The change in amplitude with operatingconditions has been of interest to many fluidization researchers foridentification of transitions between regimes. For instance, for a circu-lating fluidized bed, the gas velocity corresponding to the peak of thevariation is typically defined as the onset of the transition to turbulentregimes, while that corresponding to the point where the variationlevels off is defined as the onset of the turbulent regime [31]. However,as pointed out by Dhodapkar and Klinzing [32], the amplitude ofpressure fluctuations alone is not sufficient to elucidate the compositionof the fluctuating signals and thus spectral analysis via Fast FourierTransform (FFT) has been applied on time series of pressure data influidized beds (e.g. [26,33]). Frequency domain analysis includes the es-timation of power spectral density functions that contain informationregarding the frequency distribution in the pressure time series.Power spectral density functions are obtained via Fourier transforma-tion of signals. Analysis of frequency distribution has been widely ap-plied in time series analysis of fluidized beds for the characterizationof flow regimes (e.g. [34,35]) and for verification of scale-up relation-ships for fluidized beds (e.g. [36]). Dhodapkar and Klinzing [37]concluded that the nature of the static wall pressure fluctuations in flu-idized beds depends on the particle size, particle density, bed height,column diameter, location of the pressure taps and the gas velocity.Frequency analyses of pressure fluctuations on both a conventionalspouted bed [38] and a slot-rectangular spouted bed [39] have demon-strated promising results on flow regime identification. The disadvan-tage of power spectral density function is that deciding which peak inthe power spectrum is treated as a dominant frequency sometimescan be subjective [40]. Therefore, a combination of different analysismethods is needed to gain a better view of hydrodynamic behavior influidized beds. Therefore, the presentwork utilized both statistical anal-ysis and frequency domain analysis to identify flow regime transitionsof spouted beds of binary mixtures.

Fig. 1. Schematic diagram of the experimental apparatus. 1. Compressor; 2. Pressure regulator; 3150 mm; 8. Pressure taps; 9. pressure transducer; 10. A/D converter; and 11. PC.

More recently, the pressure fluctuation analysis also has been usedto characterize the flow behavior in spouted beds [41–45]. However,the above analysis was predominantly carried out with pressure fluctu-ation signals collected from spouted bedwithmono-sized particles. Thepressure fluctuations obtained from a binary system could be quite dif-ferent. In the literature, this difference was noted in a fluidized bed byChen et al. [46] due to the presence of particle mixing and segregationin binary particle systems. However, to the best of our knowledge,such an analysis for spouted beds with binary mixtures is not reportedin the literature. Recognition and characterization of flow regimes arecritical for designing and operating spouted beds, in particular, whenoperated with binary mixtures. In view of this knowledge gap, the ob-jectives of this study were to investigate pressure drop in a spoutedbed with binary mixtures and to identify the flow regime transitionand particle mixing/segregation by means of pressure signal analysis.

2. Experimental setup

A schematic diagramof the experimental set-up is shown in Fig. 1. Inthis work, a plexi-glass spouted bed, with 80mm in diameter, 4–10mmof nozzle diameters and 60° of conical base angle was adopted. The ex-periments were carried out at ambient conditions. The gas flowrate wascontrolled by a pressure regulator and measured by several flowmeterswith different measuring ranges (1.6–16 m3/h and 6–60 m3/h). Afterthe particles were charged into the spouted bed, the air flowrate wasadjusted for different bed heights to achieve flow regime transitions.The pressure drops and pressure fluctuations were measured by pres-sure transducers (Omega, PX164-010D5V) installed by an interval of16.7 mm along bed wall and the pressure data were recorded by a PCafter A/D conversion. The bottom pressure trap was installed on thewall just above the nozzle. Pressure fluctuations were collected atthree bed levels, i.e., total bed, lower section and upper section of bed,through plastic tubes installed on the column wall. The particles usedwere composed by Al2O3 particles and silica gel particles with a volumeratio of 4:1.

The properties of the particles used are presented in Table 1, includ-ing narrowly-distributed silica gel, Al2O3 and glass beads. The densitiesand voidages at loosely packing state of the particles were measured bya water displacement method for glass beads and wax was used forsilica gel and Al2O3 particles. The volume ratio refers to the bulk volume

. Buffer tank; 4. Gate valve; 5.Mass flow controller; 6. Spouted bed 80mm; 7. Spouted bed

Table 1Properties of experimental materials.

Particles Notation dp,mm

ρp, kg · m-3 ρb, kg · m-3 Geldartgroups

Fine particles 1# Al2O3 A1 0.2 1449 1058 B2# Al2O3 A2 0.39 1498 1062 B3# Al2O3 A3 0.79 1551 1074 B

Coarse particles 1# silica gel S1 1.05 1253 868 D2# silica gel S2 1.75 1410 965 DGlass beads GB 1.76 2797 1528 D

Fig. 2. The relationship between U and ΔPs in single sized particle systems.

Fig. 3. Effect of particle diameter on ΔPs.

136 W. Du et al. / Powder Technology 276 (2015) 134–143

ratio. The particle diameter was measured by a statistical averagedmethod with counting more than 600 particles. In this study, Al2O3

particles with diameters of 0.20 mm, 0.38 mm and 0.79 mm (Group Bparticles), were used as fine particles. Coarse particles (Group Dparticles) were silica gel particles of 1.05 mm and 1.75 mm and glassbeads of 1.76 mm.

Pressure drop fluctuation analysis was conducted as follows:

1. Average value:

ΔP ¼ 1n

Xni¼1

ΔPi ð1Þ

2. Standard deviation:

σ ¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi1

n−1

Xni¼1

ΔPi−ΔPð Þ2vuut ð2Þ

3. Average absolute deviation:

δ ¼ 1n

Xni¼1

ΔPi−ΔP�� �� ð3Þ

4. Probability density function:

f xð ÞΔx ¼ limT→∞

1T

XΔTð Þ ð4Þ

5. Power spectral density function:

SXX ωð Þ ¼Z∞

−∞

ϕXX τð Þe−iωτdτ ð5Þ

where,ΔP is the averaged pressure drop, P is the pressure drop, σ is thestandard deviation, δ is the average absolute deviation, fx is the domi-nant frequency, SXX is power spectral density and ΦXX is the signal.

3. Pressure drop of binary particle spouting

3.1. Pressure drops for spouted beds of mono-sized particles

In order to establish a baseline, the whole bed pressure drop for dif-ferent mono-sized particle systems was firstly measured. The packingheights were 160 mm for all beds. The results are shown in Fig. 2. Itcan be seen that the trend of the pressure drop against superficial gasvelocity is consistent with classical spouted bed results reported byMathur and Epstein [11] and Epstein and Grace [47]. In general, thepressure drop increases with increasing superficial gas velocity (staticbed and internal cavity) until it reaches a maximum value, followedby a decrease (inner spouting) with further increasing superficial gasvelocity. A sudden decrease in the pressure drop is observed due tothat enough solids have been displaced from the center core, indicating

that a steady spouting is established. Afterwards, the pressure dropremains constant when the gas flow rate further increases.

Fig. 2 also shows that the peak pressure drop and the initial spoutingpoint differ among different particles investigated. For glass beads, thepeak and stable spouting pressure drop are 1.61 kPa and 0.8 kPa, respec-tively, which are the highest among these particles under investigation.The peak pressure drop for A2 particle is lower than that for A3 particle.But the stable spouting pressure drop for A2 particle is higher than thatfor A3 particles. This can be explained by that the spouting of A2 parti-cle is less stable and more bubbles are generated in the bed. FromFig. 2, it can be concluded that for the peak pressure drop, the follow-ing order holds, GB N A3 N S1 N A2 N S2 while for stable spouting pres-sure drop, the order is GB N A2 N A3 N S1 N S2. As discussed by Bi [48],the pressure drop is directly related to ρbgH ((ρs − ρf)(1 − ε)gH),where ρb is particle bulk density, H is packing bed height, ρf is gasdensity, and ε is voidage. Therefore, the whole bed pressure drop ismainly determined by the particle density.

3.2. Pressure drop for spouted beds with binary mixtures

The effect of the particle diameter on the pressure drop was investi-gated in binary mixtures composed by A3 particles with two coarseparticles (S1 and S2). The mixing ratio was set at A3:S1 (or S2) = 2:1and the packed bed height was 160 mm for both cases.

Fig. 3 presents that the trend of the bed pressure drop against super-ficial gas velocity in a spouted bed of binary mixtures is very similar to

137W. Du et al. / Powder Technology 276 (2015) 134–143

that observed in a mono-sized particle spouted bed. It should be notedthat the peak pressure drop for a mixture of A3 and S2 particles occursat a higher gas velocity compared to that for amixture of A3 and S1 par-ticles. This observation is due to that the mixing degree for A3 and S2particles is relatively low and small bubbles are very likely generatedwhen the inner cavity appears close to the bed surface. The experimen-tal studies show that the segregation would happen in binary mixturesof particles with a certain range of density difference and mixing ratio,operated even when the bed was fully spouted. In this study, theonset of spouting can be reflected by pressure drop curves, for example,in Fig. 3, when the mixture has a smaller density difference (A3 mixedwith S1), the pressure drop curve has a clearly identified maximum,which indicates nearly no segregation occurred. However, when theparticles has a larger difference (A3 mixed with S2), a flat peak is ob-served in the pressure drop curve, which interprets that the pressuredrop first reaches to the frontier of flat peakwhen the velocity increasedto a velocity that light particle become fluidized. After that, because ofparticle segregation, the pressure drop did not change much until thevelocity reached a velocity that heavy particle become fluidized(as seen from the figure, the velocity is still far from that for the heavyparticle become spouted). With further increasing the superficial gasvelocity, the bedwas fully spouted andpressure drop decreased quickly.In a binary particle system, there exist three inter-phase forces, that is,gas–primary solid drag force, gas–secondary solid drag force, andsolid–solid force. In principle, the drag force between solid and gasphase can be estimated according to the Ergun [49] equation(for αg b 0.8) and the Wen–Yu [50] equation (for αg N 0.8) given by:

βErgun ¼ 150α2s μg

αgd2p

þ 1:75αsρg

dpv−uj j;αg b 0:8 ð6Þ

βWen−Yu ¼ 34CD

αsρg

dpv−uj jα−2:65

g ;αg ≥ 0:8: ð7Þ

From the above equations, it is clearly shown that the drag force dif-fers among particles due to their different particle diameters, particlesto move upward at different velocities. As a result, more bubbles tendto be formed between the two solid phases and particle segregationoccurs. These bubbles will penetrate into the packing bed, loweringthe peak pressure drop, evidenced by that 1.04 kPa is observed for theA3/S2 mixture compared to 1.1 kPa for that of the A3 and S1 mixture.Moreover, the spouting pressure drop for the A3 and S2 mixture(0.57 kPa) is seen to be lower than that for the A3 and S1 mixture(0.59 kPa).

Fig. 4. Effect of particle density on ΔPs.

The effect of the particle density on the pressure drop was investi-gated by mixing two coarse particles (S2 and GB) into A3 particles at afixed volume ratio of 1:4. The packed bed height remained at 160 mm.As shown in Fig. 4, the peak pressure drop for the A3/GB mixture is1.4 kPa at a gas superficial velocity of 0.32 m/s, which is higher than1 kPa at a gas superficial velocity of 0.26 m/s for the A3/S2 mixture.The pressure drop for A3/GB and A3/S2 mixtures at stable spouting re-gime is 0.71 kPa and 0.58 kPa, respectively. As discussed earlier, thelarger ρb of the A3/GB mixture leads to a higher pressure drop. As thediameters of mixing particles are similar to each other, the discrepancyof the drag forces is mainly caused by the difference in the particledensity. Therefore, the peak pressure drop for a system with a higherparticle density occurs at a higher gas superficial velocity.

The peak pressures of different systems with varying mixing ratiosof coarse particles are shown in Fig. 5. It's seen that for A2 and GBsystem, the peak pressure drop is 0.82 kPa at the GB content of 20%,and then increases rapidly with increasing the GB content to 1.2 kPaat the GB content of 67%. As noted by Bi [48], almost all correlationsshowed that peak pressure drops are in a linear relationship withρbgH ((ρs − ρf)(1 − ε)gH) for spouted beds of mono-sized particles.Therefore, it can be expected that in a spouted bed of binary mixtures,the peak pressure drop will increase with increasing themixture densi-ty. Since GB particles have the highest particle density among thosecoarse particles investigated, increasing its content could lift thepressure drop. However, such an increase is not in a linear manner, in-dicating that the drag force between fine particles and coarse particlesin a binary mixture bed differs from that for a spouted bed only withcoarse particles.

Fig. 5 also shows that the peak pressure drop decreases slightly inthe bed with binary mixtures of A2 particles with S1 and S2 particlesdue to a lower density of silica particles. In the experiments, it was ob-served that the mixing degree of A2 and S1 system is the highestamong all mixtures while the pressure fluctuation of this binary systemis the lowest. This observation is similar with that reported by Sau et al.(2008) that in a binary mixture fluidized bed, the peak pressure drop isdetermined by the mixing degree as well as the particle properties andthe mixing ratio. The better the mixing degree is, the more stable thebed is.

The nozzle diameter is known to have an important impact on thepeak pressure drop in spouted beds [48]. Fig. 6 shows a trend of thepeak pressure drop against the nozzle diameter. In this work, fournozzle diameters employed were 4 mm, 6 mm, 7.6 mm and 10 mm.The data reveal that the peak pressure drop decreases with increasingthe nozzle diameter. For the A2 and GB mixture, the peak pressuredecreases from 1.3 kPa at di = 4 mm to 0.97 kPa at di = 10 mm. It'sobserved that the spouting became unstable and the maximum

Fig. 5. Effect of coarse particle mixing ratio on ΔPm.

Fig. 6. Effect of nozzle diameter on ΔPm.

138 W. Du et al. / Powder Technology 276 (2015) 134–143

spouting bed height decreased significantly when the nozzle diameterapproached 10 mm. For a spouted bed, gas is easier to penetrate intothe annulus with a larger nozzle due to a resultant larger spout diame-ter, therefore leading to particles in this region being fluidized and con-sequently, decreasing the total pressure drop of the bed. For the spouted

Fig. 7. The whole bed pressure signals in different regimes. (For Dc= 80 mm, θ=60°, di = 7.6b. static bed (u = 0.09 m/s); c. spouting (u = 0.22 m/s); and d. slugging (u = 0.39 m/s).

bed with A2 and GB mixture, because of the segregation occurringduring the spouting and spreading to annulus region, some portion ofgas also pass through the viodage between segregated solid phases,thereby the bed pressure drop decreases more rapidly than those forother mixture spouted beds.

4. Pressure drop fluctuations in binary spouted beds

4.1. Pressure fluctuations and statistical analysis

In a spouted bed of single sized particle systems, a typical flowregime transition map shows that below the maximum spoutable bedheight, the bed changes from static bed regime to stable spoutingregime, and then to slugging regime with increasing the gas velocity[47]. In a spouted bed of mixed particles, similar flow patterns wereobserved. Representative pressure fluctuations for three typical flowregimes, static bed, stable spouting and slugging, are shown in Fig. 7.Relatively smooth pressure signals are found for both static bed and sta-ble spouting bed, with larger fluctuations observed for the latter flowpattern. In contrast, the pressure signal becomes irregular and highlyfluctuating in slugging flow. Pressure fluctuations in the spouted bedfor a binary particle system are similar to those results for a single par-ticle size system [45]. The bed pressure drop signals are more irregularand fluctuating in spouting and slugging flow regimes. The difference

mm, P= 0.1 MPa, H = 0.13 m, VA2:VS1= 4:1). a. Comparison of different flow patterns;

Fig. 8. Probability distribution of pressure fluctuations in different regimes. 1. For upper cylindrical section; 2. For lower cylindrical section; and 3. For total bed. (For Dc= 80mm, θ=60°,di = 7.6 mm, P = 0.1 MPa, H = 0.13 m, VA2:VS1 = 4:1).

139W. Du et al. / Powder Technology 276 (2015) 134–143

is considered to be related to non-uniformmixing of the particles in thespouted bed with binary mixture.

In the statistical analysis, the probability distribution of pressuresignal is a direct measure of deviations between time series pressuredata and their averaged value. Fig. 8 shows the probability distributionsof pressure signals for different flow regimes presented in Fig. 7. Itappears that there are considerable differences among the probabilitydistributions for the regimes observed. In static bed regime, the pressurehas the narrowest distribution, with the lowest deviation (±10%) ofpressure fluctuations from the averaged value. In the stable spoutingregime, the probability distribution becomes wider, which is mainlyconcentrated in the region of ±40% around the averaged value. In theslugging regime, the widest probability distribution range, ±60%, isobserved.

Table 2Average value and standard deviation of pressure fluctuations for three different flowregimes. 1-For upper cylindrical section; 2-for lower cylindrical section; and 3-for the totalbed.

Static bed Stable spouting Slugging

Average value Curve 1 16.8 12.7 15.4Curve 2 161 125 144

(Pa) Curve 3 557 514 526Standard deviation Curve 1 0.87 1.49 3.31

Curve 2 12.6 24.9 46.2(Pa) Curve 3 49.8 81.9 203

Standard deviations of the pressure fluctuations have been wide-ly used to identify a regime change in gas–solid fluidized beds[26–30]. Table 2 shows the averaged values and standard deviationsof differential pressure fluctuations collected at different bedsections in a spouted bed. It is seen in this table that at the samebed section, the differential pressure drop at static bed regime hasthe highest average value while the lowest value is observed atstable spouting regime. However, the standard deviation generallyincreases at the air flow and rate is increased from a static bedthrough spouting regime to slugging. The sharp increase in thestandard deviation from spouting to slugging is due to unstableflow states of gas and solids within the bed, though the averagepressure drops for the two regimes are similar.

Table 2 also indicates that there are significant differences in thepressure drops and their standard deviations at different bed sections.In the upper section of the bed, the lowest pressure drop and its stan-dard deviation are found because the flow is in the fountain regionwhere the gas and solid flows show minimal impacts on the pressuredrop. In the lower section of the bed, the pressure drop and its standarddeviation are also low as theflow is in the annulus regionwhere gas andsolids are in regular motion. However, near to the nozzle, the solids arecirculated back into the spout region by the highest gas velocity. The in-duced intensemotions lead to the highest pressure drop and the largeststandard deviation. Thus, the total bed pressure dropfluctuations can beadequately utilized to reflect flow characteristics of spouted beds ofmixed particle systems.

140 W. Du et al. / Powder Technology 276 (2015) 134–143

From the above discussions, it can be seen that the statistic charac-teristics of time series pressure drop, that is, average value, standarddeviation and probability distributions, are quite different for differentflow regimes in spouted beds with binary mixtures. Thus, statisticalanalysis of pressure fluctuations, in particular, the total bed pressuresignal, can reflect dynamics of a spouted bed of binary mixtures andrecognize prevailing flow patterns therein. But the pressure fluctuationfor a binary system is more irregular than that of a single sized particlesystem which is consistent with the current results.

4.2. Frequency domain analysis

4.2.1. Effect of particle diameter of binary mixturesAs noted previously, frequency domain analysis transforms the

information from the time domain to the frequency domain, whichincludes estimation of spectral density or spectral amplitude in thepressure time series. After analysis, dominant frequencies are usuallyidentified and correlated to various underlying physical phenomena.In the literature, spectral analysis has been employed to characterizedifferent flow regimes in spouted beds [43–45,51]. However, all previ-ous studies were conducted with the pressure time series measuredfor single sized particles in spouted beds. In a binary mixture, morecomplex dynamic behaviors are expected due to inherent interactionsamong particles with different particle sizes and densities. In experi-mental observation, it was found that with introducing a secondarycoarse particle into a primary fine particle system, the stable spoutingrange could be considerably widened. The spouting stability of this bi-nary mixture is further investigated by power spectrum analysis. Fig. 9illustrates effect of particle diameters of binary mixtures on amplitudesdetermined by Fast Fourier Transform (FFT). The pressure time serieswere measured in binary mixtures composed of A2 particles and twolarge particles (S1 and S2 particles) at two different volume ratios(fine/coarse particles = 2:1 and 1:1). In Fig. 9a, at a superficial gas

Fig. 9. The spectrum analysis of mixtures of different diameters.

velocity of 0.22 m/s, stable spouting was attained for both A2/S1 andA2/S2 mixtures at the same volume ratio of 2:1. A dominant frequencywith amplitude of 12 is found at about 6 Hz for the A2/S2mixturewhilethere is no distinct dominant frequency observed for the A2/S1mixture.Instead, a broader distribution of frequencies, ranging from 5 Hz to15 Hz, is observed, which is presumably associated with low frequencygas turbulence. This also suggests that the spouting stability of theA2/S1mixture is lower than that of the A2/S2 mixture. When raising thecoarse particle content to a ratio of 1:1, the spouting was obtained at asuperficial gas velocity of 0.28 m/s. Fig. 9b illustrates that a dominantfrequency is found at about 10 Hz for the A2/S2 mixture, and the signalis more distinct than that of a mixture at the ratio of 2:1. However, thedominant frequency is still not clearly shown for the A2/S1 mixtureand again the frequencies are widely distributed. From the aboveanalysis, it can be concluded that larger particles show more profoundimpact on improving spouting stability of fine particles.

4.2.2. Effect of mixture particle densityTo investigate the effect of particle density, experiments were per-

formed in binary mixtures of A2 particles with two coarse particles(S2 particles and glass beads) at three volume ratios of A2/coarseparticles = 4:1, 2:1 and 1:1. Fig. 10a shows the power spectrum ofdifferent mixtures at a volume ratio of 4:1. It can be seen in this figurethat the dominant frequencies cannot be found and frequencies arewidely distributed for both mixtures, implying that the spouting is notstable and low frequency gas turbulence dominates dynamics in thebed. However, the A2/GB mixture shows relatively stable spouting asnarrower distributions of frequencies, ranging from 5Hz to 15 Hz, is ob-served, compared to a range of 4 Hz–18 Hz for the A2/S2 mixture. Withincreasing the coarse particle content to 2:1 A2/coarse particles ratio, adominant frequency appears at about 6 Hz for the A2/S2 mixture asshown in Fig. 10b. Similarly, the dominant frequency for the A2/GBmix-ture is clearly shown at 10 Hz, but two interference frequencies appear

Fig. 10. The spectrum analysis of systems with different density.

141W. Du et al. / Powder Technology 276 (2015) 134–143

at 6 Hz and 15 Hz, corresponding to the dominant frequencies in singlesizedA2 andGB systems, respectively. The results reveal that themixingdegree for mixtures with larger density difference such as the A2/GBsystem becomes poor and particle separation occurs in the spoutedbed. Further evidence can be found from Fig. 10c for 1:1 mixing ratiomixtures. It is clearly shown that a dominant frequency of 10 Hz is ob-served for the A2/S2 mixture and interference frequencies are greatlydampened. However, for the A2/GB system, the two enhanced interfer-ence frequencies appear at 6 Hz and 15 Hz, indicate that the particleseparation is remarkable in the mixtures. In addition, the amplitude ofdominant frequency decreases from 30 to 20 as compared with thatshown in Fig. 10b. Therefore, the density difference in the mixturecannot be too big even their particle sizes are similar for purposes ofimproving spouting stability.

Despite that the spouting stability can be improved in the binary sys-tem, the separation of particlesmay occur due to too large differences inparticle densities and particle sizes. The poor mixing will lead to unde-sired flow patterns. In order to further understand the effect of mixingdegree on the spouting, spectrum analysis was conducted with thepressure signals collected in binary mixtures with A2 particles mixedwith three coarse particles, S1, S2, and GB. As the density and the sizeof GB particles are greatly larger than those of the A2 particles, particlesegregation is notable and the stratified spouting and bubbling are ob-served as shown in Fig. 11a. The dominant frequency in this mixture isfound to be at about 10 Hz, whereas two interference frequencies ap-pear at 6 Hz and 15 Hz, indicating the three particle circulations in thesystem, i.e., the separated GB and A2 particles and the uniformmixture.In the A2/S2 mixture, the system shows a better spouting as illustratedin Fig. 11b with the dominant frequency more clearly shown at 10 Hz.However, there still exists an interference frequency at 6 Hz, indicatingthat some A2 particles are separated from the mixture in the spoutingprocess. Since the A3 and S1 particles have closest particle sizes anddensities, the particles are well mixed without segregation and the

Fig. 11. Effect of mixing degree on spectrum analysis.

mixture has the best spouting stability. Fig. 11c shows that thedominant frequency is 14 Hz, and the narrowest frequency distributionis obtained. The results are in close agreementwith Chao et al., [52] thatthe fine particle content is of crucial important for stable fluidization ofbinary particle system.

4.2.3. Effect of mixing particle ratioAs stated earlier, the spouting stability for the single A2 particle sys-

tem is poor due to its smaller size. However, adding S2 particles with asimilar particle density to A2 particles, the spouting range and stabilitycan be greatly improved. Thisfinding can be further verified by the pres-sure fluctuation analysis. Fig. 12 shows the relationships of the averagepressure drops and the standard deviations against the coarse particlecontent. All measurements were taken at minimum spouting velocities.In general, the average pressure drop for the single sized A2 particle(625 Pa) decreases to 410 Pa for a single sized S2 particle spouted bedwhen the coarse particle content increases from 0 to 1. A slight increasein the standard deviation from that for the single sized A2 particle in thespouted bed is attributed to that the lower superficial gas velocity is re-quired for spouting a single sized A2 system. As the content of the coarseparticle increases, it starts to increase and then drops after the volumeratio reaches 4:1. This decrease in the standard deviations indicatesthat the particle motion in the spouted bed becomes more regular anda more stable spouting is attained.

The results of spectral analysis of the total bed pressure fluctuationsare shown in Fig. 13. As the content of coarse particles increases, the fre-quencydistribution becomesnarrower, as illustrated in Fig. 13a towardsFig. 13e. A dominant peak at around 10 Hz appears in Fig. 13e for a mix-ture with more coarse particles. When the binary mixture containsmore fine particles (A2), the spouting is not stable and the frequencydistribution is broad with no dominant frequency found. In addition,the multiple peaks observed in Fig. 13a and 13b could be related tothe existence of several periodical circulating components in the bed.As the peak frequency turns to be more distinct, the amplitude alsoincreases. For instance, it increases from 7 in Fig. 13c to 24 in Fig. 13eat 10 Hz. Compared to denser coarse particles such as glass beads,lighter coarse particles (density similar with that of fine particles) aremore favorably utilized to improve spouting stability for fine particles,as evidenced bymore distinct dominant frequencies with higher ampli-tudes observed in the pressure time series.

5. Conclusions

In this work, the pressure drop and fluctuation signals were mea-sured in spouted beds of binary mixtures. Both statistical and frequency

Fig. 12. Effect of mixing ratio on average value and standard deviation.

Fig. 13. Effect of mixing ratio on amplitude spectrum analysis.

142 W. Du et al. / Powder Technology 276 (2015) 134–143

domain analysis were employed to identify flow regime transitions inthe spouted bed. The following main conclusions can be drawn:

1. The trend of the pressure drop against superficial gas velocities formixed particles in spouted beds is similar with that for single sizedparticle systems. However, the peak pressure drop is related to themixing degree of the mixture, which is mainly determined by theinter-particle forces. The mixture density and mixing ratios bothshow an impact on the pressure drop fluctuations. The stressbetween coarse and fine particles in a binary particle system is great-ly less than the internal friction for a single sized particle system,leading to a better spouting stability in the binary system.

2. The pressure drop signals at static bed and stable spouting bed areregular but highly fluctuating in slugging regime. More irregularand fluctuated signals are found for binary particle mixtures as com-paredwith that for single sized particle systems at the same spoutingregime. The statistic characteristics of pressure drop time series(average value, standard deviation and probability distributions)are different for different flow regimes. Therefore, it can be used forthe reorganization of these flow regimes and their transitions.

3. The spouting stability of a binary system was evaluated by a powerspectrum analysis. Narrower frequency distribution has been ob-served for well mixed systems while no dominant frequency isshown for spouted beds at unstable spouting conditions.

4. Our results suggest that spouting stability in spouted beds with fineparticles can be improved by mixed with coarse particles. However,the density difference in the mixture cannot be too large even if theparticle sizes of the two particles are quite similar. Addition of lighter(similar density with that of fine particles) but larger (in diameter)coarse particles is more effective in improving spouting stability offine particles in spouted beds.

Nomenclatures

b1 − b6 constants in pressure drop equation regressed by

non-linear analysis

–

CD

drag coefficient Dc column diameter mm dp particle diameter mm

dp

average particle diameter mm

dpl

diameter of large particles mm dps diameter of small particles mm di orifice diameter mm fm dominant frequency Hz f(x) probability density function g gravitational constant m/s2

H

packing bed height mm ΔPm maximum bed pressure drop Pa ΔPs spouting pressure drop Pa ?P averaged pressure drop Pa SXX power spectral density W/Hz U superficial gas velocity m/s V bulk volume m3

VA1

bulk volume of 1# Al2O3 m3

VA2

bulk volume of 2# Al2O3 m3

VA3

bulk volume of 3# Al2O3 m3

Vs1

bulk volume of 1 ∗ silica gel particles m3

Vs2

bulk volume of 1 ∗ silica gel particles m3

VGB

bulk volume of glass beads m3

xl

the mass content of the large particles % xs the mass content of the small particles %

Greek letters

αs solid volume fraction αg gas volume fraction β fluid–particle interaction coefficient Kg/m3s μ gas viscosity Pa.s ρb particle bulk density kg/m3

ρg,ρf

gas density kg/m3

ρp

particle density kg/m3

xs

mixing particle density kg/m3

Ρpl

density of the large particles kg/m3

Ρps

density of the small particles kg/m3

τ

delay time of reconstruction s Φ particle sphericity ΦXX autocorrelative function σ standard deviation δ average absolute deviation ε voidage ω power spectral function

Subscripts

c column –

Ergun

Ergun equation Wen–Yu Wen–Yu equation g gas s solid f fluid –

l

large –

m

maximum –

ms

minimum spouting –

P

particle –

Pl

large particle –

Ps

small particle –

i

inlet –

Acknowledgments

Authors should thank the National Natural Science Foundation ofChina under Grant No. 21076230 and No. 21176256 and the ScienceFoundation of China University of Petroleum, Beijing (No. KYJJ2012-03-01) for funding this project.

References

[1] K.B.Mathur, P.E. Gishler, A technique for contacting gases with coarse solid particles,AICHE J. 1 (1) (1955) 157–164.

[2] K.G. Marnasidou, S.S. Voutetakis, G.J. Tjatjopoulos, I.A. Vasalos, Catalytic partialoxidation of methane to synthesis gas in a pilot-plant-scale spouted-bed reactor,Chem. Eng. Sci. 54 (15) (1999) 3691–3699.

143W. Du et al. / Powder Technology 276 (2015) 134–143

[3] L. Mleczko, K.J. Marschall, Performance of an internally circulating fluidized-bedreactor for the catalytic oxidative coupling of methane, Can. J. Chem. Eng. 75 (3)(1997) 610–619.

[4] M. Olazar, R. Aguado, J. Bilbao, A. Barona, Pyrolysis of sawdust in a conical spouted-bed reactor with a HZSM-5 catalyst, AICHE J. 46 (5) (2000) 1025–1033.

[5] L.D. Schmidt, M. Huff, S.S. Bharadwaj, Catalytic partial oxidation reactions andreactors, Chem. Eng. Sci. 49 (24) (1994) 3981–3994.

[6] B.H. Song, A.P. Watkinson, Three-stage well-mixed reactor model for a pressurizedcoal gasifier, Can. J. Chem. Eng. 78 (1) (2000) 143–155.

[7] T.A. Sue-A-Quan, G. Cheng, A.P. Watkinson, Coal gasification in a pressurizedspouted bed, Fuel 74 (2) (1995) 159–164.

[8] Y. Uemura, M. Azeura, Y. Ohzuno, Y. Hatate, Products from flash pyrolysis of plastics,Proceedings of the International Conference on Solid Waste Technology andManagement, Widener Univ. School Eng., Chester, PA, USA, 1999, pp. 91–96.

[9] Y. Uemura, H. Ohe, Y. Ohzuno, Y. Hatate, Carbon and hydrogen from hydrocarbonpyrolysis, Proceedings of the International Conference on Solid WasteTechnology and Management, Widener Univ. School Eng., Chester, PA, USA, 1999,pp. 622–627.

[10] H.B. Vuthaluru, D.K. Zhang, Effect of Ca- and Mg- bearing minerals on particle ag-glomeration defluidization during fluidized-bed combustion of a South Australianlignite, Fuel Process. Technol. 69 (1) (2001) 13–27.

[11] K.B. Mathur, N. Epstein, Spouted Bed, Academic Press, New York, 1974.[12] A. Çeçen, H. Littman, M.H. Morgan III, Flow regime diagrams and the stability of

spouting of fine glass spheres with air, in: J.F. Large, C. Laguerie (Eds.), FluidizationVIII, Engineering Foundation, NY, 1996, pp. 207–216.

[13] P. Chandnani, Gas Spouting of Fine ParticlesM. A. Sc. Thesis University of BritishColumbia, Vancouver, British Columbia, Canada, 1984.

[14] J.R. Grace, Contacting models and behavior classification of gas–solid and other two-phase suspensions, Can. J. Chem. Eng. 64 (6) (1986) 353–363.

[15] S.N. Wang, J. Xu, W.S. Wei, G. Shi, X.J. Bao, H.T. Bi, Gas spouting hydrodynamics offine particles, Can. J. Chem. Eng. 78 (1) (2000) 156–160.

[16] H. Huang, G. Hu, F. Wang, Experimental study on particles mixing in an annularspouted bed, Energy Convers. Manag. 49 (2) (2008) 257–266.

[17] M. Olazar, M.J. San Jose, F.J. Penas, A.T. Aguayo, J. Bilbao, Stability and hydrodynamicsof conical spouted beds with binary mixtures, Ind. Eng. Chem. Res. 32 (11) (1993)2826–2834.

[18] Y. Zhang, B. Jin, W. Zhong, Experiment on particle mixing in flat-bottom spout–fluidbed, Chem. Eng. Prog. 48 (2009) 126–134.

[19] B. Ren, Y. Shao, W. Zhong, B. Jin, Z. Yuan, Investigation of mixing behaviors in aspouted bedwith different density particles using discrete elementmethod, PowderTechnol. 222 (2012) 85–94.

[20] H. Lu, Y. He, W. Liu, D. Jianmin, D. Gidaspow, J. Bouillard, Computer simulations ofgas–solid flow in spouted beds using kinetic-frictional stress model of granularflow, Chem. Eng. Sci. 59 (4) (2004) 865–878.

[21] C.K.K. Lun, S.B. Savage, D.J. Jeffrey, Kinetic theories for granular flow: inelastic parti-cles in Couette flow and slightly inelastic particles in general flow field, J. FluidMech. 140 (1984) 223–256.

[22] H.T. Bi, A critical review of the complex pressure fluctuation phenomenon in gas–solids fluidized beds, Chem. Eng. Sci. 62 (13) (2007) 3473–3493.

[23] S. Sasic, B. Leckner, F. Johnsson, Characterization of fluid dynamics of fluidized bedsby analysis of pressure fluctuations, Prog. Energy Combust. Sci. 33 (5) (2007)453–496.

[24] J.R. van Ommen, R.F. Mudde, Measuring the gas–solids distribution in fluidizedbeds — a review, Int. J. Chem. React. Eng. 6 (1) (2008) 1542–6580 (ISSN (Online)).

[25] J.R. van Ommen, S. Sasic, J. van der Schaaf, S. Gheorghiu, F. Johnsson, M.O. Coppens,Time-series analysis of pressure fluctuations in gas–solid fluidized beds— a review,Int. J. Multiphase Flow 37 (5) (2010) 403–428.

[26] C.A.S. Felipe, S.C.S. Rocha, Prediction of minimum fluidization velocity of gas–solidfluidized beds by pressure fluctuation measurements — analysis of the standarddeviation methodology, Powder Technol. 174 (3) (2007) 104–113.

[27] M.R. Parise, O.P. Taranto, P.R.G. Kurka, L.B. Benetti, Detection of the minimum gasvelocity region using Gaussian spectral pressure distribution in a gas–solid fluidizedbed, Powder Technol. 182 (3) (2008) 453–458.

[28] C. Sobrino, S. Sánchez-Delgado, N. García-Hernando, M. de Vega, Standard deviationof absolute and differential pressure fluctuations in fluidized beds of group Bparticles, Chem. Eng. Res. Des. 86 (11) (2008) 1236–1242.

[29] T.Y. Yang, L.P. Leu, Study of transition velocities from bubbling to turbulent fluidiza-tion by statistic and wavelet multi-resolution analysis on absolute pressurefluctuations, Chem. Eng. Sci. 63 (7) (2008) 1950–1970.

[30] W.H. Zhang, X.G. Li, Origin of pressure fluctuations in an internal-loop airlift reactorand its application inflow regimedetection, Chem. Eng. Sci. 64 (5) (2009) 1009–1018.

[31] H.T. Bi, L. Fan, Existence of turbulent regime in gas–solid fluidization, AICHE J. 38 (2)(1992) 297–301.

[32] S.V. Dhodapkar, G.E. Klinzing, Pressure fluctuations in pneumatic conveyingsystems, Powder Technol. 74 (2) (1993) 179–195.

[33] O. Trnka, V. Veselý,M. Hartman, Z. Beran, Identification of the state of a fluidized bedby pressure fluctuation, AICHE J. 46 (3) (2000) 509–514.

[34] H.T. Bi, J.R. Grace, Effects of measurements on transition velocities used to demar-cate the onset of turbulent fluidization, Chem. Eng. J. 57 (3) (1995) 261–271.

[35] J.C. Schouten, M.L.M. Stappen, C.M. van den Bleek, Scale-up of chaotic fluidized bedhydrodynamics, Chem. Eng. Sci. 51 (10) (1996) 1991–2000.

[36] M.T. Nicastro, L.R. Glicksman, Experimental verification of scaling relationships forfluidized beds, Chem. Eng. Sci. 39 (9) (1982) 1373–1384.

[37] S.V. Dhodapkar, G.E. Klinzing, Pressure fluctuation analysis for a fluidized bed, AIChESymposium Series 89 (296) (1993) 170–183.

[38] M. Wang, J. Xu, W. Wei, X. Bao, X. Bi, Pressure Fluctuations in Gas–Solids SpoutedBeds, Fluidization X, Engineering Conferences International, Brooklyn, NY, USA,2001. 149–156.

[39] L.A.P. Freitas, O.M. Dogan, C.J. Lim, J.R. Grace, D. Bai, Identification of flow regimes inslot-rectangular spouted beds using pressure fluctuations, Can. J. Chem. Eng. 82(2004) 60–73.

[40] F. Johnsson, R.C. Zijerveld, J.C. Schouten, C.M. van den Bleek, B. Leckner, Characteri-zation of fluidization regimes by time-series analysis of pressure fluctuations, Int. J.Multiphase Flow 26 (4) (2000) 663–715.

[41] S. Devahastin, A.S. Mujumdar, Some hydrodynamic and mixing characteristics of apulsed spouted bed dryer, Powder Technol. 117 (3) (2001) 189–197.

[42] G. Hu, Y. Li, X. Gong, Spoutable bed height and pressure fluctuation of a novel annu-lar spouted bed with V-shaped deflector, Powder Technol. 185 (2) (2008) 152–163.

[43] N.E.C. Lopes, V.A.S. Moris, O.P. Taranto, Analysis of spouted bed pressure fluctuationsduring particle coating, Chem. Eng. Prog. 48 (6) (2009) 1129–1134.

[44] M. Tao, B. Jin, W. Zhong, Y. Yang, Analysis of pressure fluctuation signals in theunderfeed circulating spouted bed, Chem. Eng. Process. 49 (4) (2010) 340–347.

[45] J. Xu, X.J. Bao, W.S. Wei, G. Shi, S. Shen, H.T. Bi, J.R. Grace, C.J. Lim, Statistical andfrequency analysis of pressure fluctuations in spouted beds, Powder Technol. 140(1–2) (2004) 141–154.

[46] Y.G. Chen, Z.P. Tian, Z.Q. Miao, Analysis of the pressure fluctuations in binary solidscirculating fluidized bed, Energy Convers. Manag. 47 (5) (2006) 611–623.

[47] N. Epstein, J.R. Grace, Spouted and Spout–Fluid Beds, Cambridge University Press,Cambridge, 2010.

[48] H.T. Bi, Chapter 2, initiation of spouting, in: N. Epstein, J.R. Grace (Eds.), Spouted andSpout–Fluid Beds: Fundamentals and Applications, Cambridge University Press,2010, pp. 17–28.

[49] S. Ergun, Fluid flow through packed columns, Chem. Eng. Prog. 48 (2) (1952) 89–94.[50] C.Y. Wen, Y.H. Yu, Mechanics of fluidization, Chem. Eng. Prog. Symp. Ser. 62 (2)

(1966) 100–111.[51] E. Piskova, L. Mörl, Characterization of spouted bed regimes using pressure

fluctuation signals, Chem. Eng. Sci. 63 (9) (2008) 2307–2316.[52] Z.X. Chao, Y.F. Wang, J.P. Jakobsen, M. Fernandino, H.A. Jakobsen, Investigation of the

particle–particle drag in a dense binary fluidized bed, Powder Technol. 224 (2012)311–322.