Acta Polytechnica Vol. 52 No. 3/2012

Experimental Investigation of Radial Gas Dispersion Coecients

in a Fluidized Bed

Jir Stefanica1, Jan Hrdlicka1

1CTU in Prague, Faculty of Mechanical Engineering, Department of Energy Engineering, Technicka 4,166 07 Praha, Czech Republic

Correspondence to: jiri.stefanica@seznam.cz

Abstract

In a uidized bed boiler, the combustion eciency, the NOX formation rate, ue gas desulphurization and uidized bed

heat transfer are all ruled by the gas distribution. In this investigation, the tracer gas method is used for evaluating the

radial gas dispersion coecient. CO2 is used as a tracer gas, and the experiment is carried out in a bubbling uidized

bed cold model. Ceramic balls are used as the bed material. The eect of gas velocity, radial position and bed height is

investigated.

Keywords: uidized bed, gas mixing, radial dispersion.

1 Introduction

Fluidized bed combustion technology is an ecientand ecological way of combusting low quality fuels.The fuel is combusted in a bed of inert material thatis brought to a uidized state by passing through air.In a uidized state, the bed particles are in force equi-librium, resulting in uid-like behaviour of the bed.Extensive gas and solid mixing inside the uidizedbed provides a large active surface for all chemicalreactions, and also for heat transfer. As a result, thecombustion temperature can be lower than for othertypes of combustion, while high combustion eciencyis achieved. A temperature range of 800950 C isusually used, providing ideal conditions for in-situdesulphurisation and reduction of NOX . This tem-perature is also necessary to prevent agglomerationof the bed material.

A detailed uidized bed behaviour description isneeded for the design and control of a uidized bedboiler. There are many quantities for characterizinga uidized bed. One of the most important charac-teristics is the extent of mixing, because it rules theintensity of all mass and heat transport processes.Gas and solid dispersion coecients in axial and ra-dial direction are used to describe the extent of mix-ing in the bed. This paper will present an exper-imental evaluation of the gas dispersion coecientin radial direction Dgr for a bed of ceramic balls.Ceramic balls are not a typical inert bed material;quartz sand and coal ash are more widely used. Ce-ramic balls have recently come under considerationfor use in combustion of biomass and waste derivedfuels [1].

The most widely used method for dispersion co-ecient evaluation is the tracer gas method. Thismethod uses a suitable tracer gas, which is injectedin a uidized bed at one point. The concentration ofthe tracer gas in the primary uidization gas is thenmeasured at other points of the bed. Gas dispersioncoecients can be determined using an appropriatemodel. The tracer gas experiment methodology ispresented in detail in [2]. The step response tracergas method is suitable for evaluating the axial dis-persion coecient Dga. Some authors have used thetracer gas method with steady state to determine alsoDgr, the radial dispersion coecient [35]. Consider-ing the similarities of the two types of experiments,all tracer gas related studies have been taken intoaccount. Some authors (e.g. [5,7]) consider axial dis-persion to be less important than radial dispersion.In this work the steady state tracer gas method wasused to determine the gas dispersion coecients inradial direction Dgr.

2 Theory

The radial gas dispersion coecient has been evalu-ated by the steady state tracer gas method. Varioustracer gases have been described in the literature. Weselected CO2 for our experiment. After setting thetracer gas ow rate, it was necessary to wait for thetracer gas concentration to stabilize and achieve asteady state. The tracer gas concentration was thenmeasured at the investigated height, at points withdierent radial positions, see the concentration pro-les for heights 520 cm and uidization velocity of0.44 m/s in Figure 1.

97

Acta Polytechnica Vol. 52 No. 3/2012

Figure 1: Radial dispersion proles for ceramic balls,0.44 m/s

The dispersed plug ow model from equation (1)can be used for the evaluation. Assuming dispersedplug ow, the concentration proles should agreewith equation (2), where x is the radial position andxm is the radial position where the tracer gas concen-tration reaches half of the maximum value. Accord-ing to [5], the solution for suciently distant bound-aries is given by equation (3), and Dgr can be calcu-lated by equation (4).

uCgh

= Dgr

[1

x

x

(xCgh

)]+ dga

2Cgh2

(1)

CgCa

=

(1

2

)( x2x2m

)(2)

CaC

=uR2

HDgr(3)

Dgr =u R C

H Cg (4)

The results can also be seen in terms of dimen-sionless numbers as the dependence of the Pecletnumber from equation (5) and the Reynolds parti-cle number from equation (6).

Pe =L uD

(5)

Rep =Dp u g

(6)

3 Experimental

The experiment was carried out in a bubbling u-idized bed cold model, see Figure 2. Consideringthe ambient temperature of the investigated bed, themodel was made out of a 220 mm plexiglass cylinderto enable visual observation of the experiment. Thebody of the model was supported by a steel tripodstand 1. The uidization air was supplied by an airfan in the lower section 2. Tracer gas was injectedby pipe 3 in the middle of the perforated plate dis-tributor 5 (open area of 2.51 %) that was supportingthe uidized bed. CO2 was used as a tracer gas, for

safety reasons and also as a readily available detec-tion method. In the wind-box section 4 between theuidization air inlet and the distributor plate, theair ow became steady. The tracer gas concentrationwas measured through the access points 6.

Figure 2: Experimental apparatus

The tracer gas concentration was measured us-ing an ASEKO Air CmHn non-dispersive infrared gasanalyser.

A bed of ceramic balls was used for the Dgr in-vestigation. The material properties are presented inTable 1. The dependence of Dgr on mean uidizationvelocity and bed height was investigated.

Table 1: Material properties

Ceramic balls

[kg/m3] 800

[] 0.34

[] 0.95

Dp [mm] 3.11

umf [m/s] 0.49

During the experiment, the tracer gas ow ratewas kept below 1.5 % of the primary uidizationair ow rate. Bed heights of 5, 10, 15 and 20 cmwere measured. Fluidization velocities of 0.44 m/s,0.58 m/s, 0.73 m/s and 0.88 m/s were used, repre-senting (1 2) umf . The velocities were chosen inaccordance with the ow rate measurement capabili-ties.

98

Acta Polytechnica Vol. 52 No. 3/2012

4 Results and discussion

The Dgr values that were obtained are shown in Fig-ures 36. The values varied between 0.264.68 cm2/s,which is in agreement with the ndings of other au-thors. The measurements showed the same trend forall conditions. Regardless of bed height or uidiza-tion velocity, two peak values were observed, markingthe regions where the radial gas mixing is most in-tensive. The rst peak is the region in centre of thebed, and the second peak region is near the wall. Itcan also be observed that, while the extent of mix-ing decreases with increasing uidization velocity inthe centre of the bed, the situation is reversed nearthe wall. With increasing gas velocity the wall re-gion gains in signicance, due to the enhanced gasand solid back mixing that takes place in the regionnear the walls.

It can also be observed that radial gas mixing de-creases with increasing bed height. In the course ofvarying the bed height, the most signicant dierencewas observed when the bed height was increased from5 cm to 10 cm. The drop in the extent of mixing wasless signicant for further increases in bed height.

The results in terms of dimensionless numbers areshown in Figure 7. The expected linear increase of Pewith Rep was observed. For higher beds, the depen-dence on Rep increased. For lower bed heights, onlya slight increase or even stagnation was observed.

Figure 3: Dgr for ceramic balls, 1 umf

Figure 4: Dgr for ceramic balls, 1.33 umf

Figure 5: Dgr for ceramic balls, 1.66 umf

Figure 6: Dgr for ceramic balls, 2 umf

Figure 7: Dimensionless parameters

5 Conclusion

This paper has described how the experimental appa-ratus for radial gas dispersion coecient evaluationwas designed and how a steady state tracer gas ex-periment was conducted. Ceramic balls were usedas the bed material. The eects of uidization gasvelocity and bed height were investigated.

For the conditions that were used, the radial gasdispersion coecient values were determined to be0.264.68 cm2/s. This is in agreement with previousexperiments for a bubbling bed, where Dgr values of15 cm2/s were mostly found.

The highest value on each level was found eitherin the centre of the bed or at the wall. With in-creasing velocity, the peak near the wall was found

99

Acta Polytechnica Vol. 52 No. 3/2012

to increase, while the peak in the bed centre was de-creased. The radial gas dispersion coecients werefound to decrease with bed height. Linear depen-dence of Pe on Rep was found. For higher beds, thedependence on Rep increased. For lower bed heights,the Pe value could be considered to be independentfrom Rep taking into account the measurement un-certainty.

Legend

[] void fraction

[] sphericity

[Pa s] dynamical viscosity [kg/m3] density of the bed material

g [kg/m3] uidization gas density

ca [vol. %] maximal concentration of thetracer gas for a given heightmeasured at the centre of thebed

cg [vol. %] concentration of tracer gas ata given point

c [vol. %] concentration of tracer gasafter perfect mixing with theuidization gas

D [m] bed diameter

Dga [m2/s, cm2/s] gas dispersion coecient in

axial direction

Dgr [m2/s, cm2/s] gas dispersion coecient in

radial direction

Dp [m] particle diameter

L [m] characteristic dimension

x [m] radial distance from thevertical axis of the bed

xm [m] value of x whereCgCa

=1

2

u [m/s] mean gas velocity

umf [m/s] gas velocity for minimaluidization

Pe [] Peclet number

Rep [] Reynolds number for particles

References

[1] Hrdlicka, J., Obsil, M., Hrdlicka, F.: Field Testof Waste Wood Combustion Using Two DierentBed Materials, In Proceedings of the 16th SCEJSymposium on Fluidization and Particle Process-ing. Tokyo : The Society of Chemical Engineers,Japan, 2010, p. 4043.

[2] Kunii, D.: LEVENSPIEL Fluidization Engineer-ing, Butterworth Heinemann, 1991.

[3] Namkung, W., Kim, S. D.: Radial Gas Mixing ina Circulating Fluidized Bed, Powder Technology,113, 2000, p. 2329.

[4] Sterneus, J., Johnsson, F., Leckner, B.: Charac-teristics of Gas Mixing in a Circulating FluidisedBed, Powder Technology, 126, 2002, p. 2841.

[5] Chyang, C.-S., Han, Y.-L., Chien, C.-H.: Gas Dis-persion in a Rectangular Bubbling Fluidized Bed,Journal of the Taiwan Institute of Chemical En-gineers, 41, 2010, p. 195202.

[6] Sane, S. U. et al.: An experimental and mod-elling investigation of gas mixing in bubbling u-idized beds. Chemical engineering science, 51,1996, p. 1 1331147.

[7] Atimtay, A., Cakaloz, T.: An Investigation onGas Mixing in a Fluidized Bed, Powder Technol-ogy, 20, 1978, p. 17.

100