Flow phenomena in the exit zone of a circulating fluidized bed

Flow phenomena in the exit zone of a circulating fluidized bedUlrike Lackermeier,Joachim Werther *Technische Uni v ersitat Hamburg-Harburg,Arbeitsbereich Verfahrenstechnik 1,21071Hamburg,Germany Recei v ed 19November 2001;recei v ed in re v ised form 23January 2002;accepted 23January 2002AbstractAn experimental in v estigation of the gas and solids flow in a pilot-scale circulating fluidized bed (CFB)cold model with two different abrupt exit configurations (L-shape and extended top)has been carried out.Measurements of axial pressure profiles,high-speed v ideo images of the flow phenomena at the wall as well as local optical probe measurements inside the exit zone are presented.Contrary to published results obtained in bench-scale CFB risers the axial profiles of the apparent solids v olume concentration obtained by pressure measurements showed no indication of an increased solids hold-up in the v icinity of the exit,which confirms the conclusion by Pugsley et al.(Can.J.Chem.Eng.75(1997)1001)that this is a scale effect.The local probe measurements showed the well-known core Áannulus flow structure pre v ailing until the riser top.In the v icinity of the exit this flow structure is superimposed by a strong horizontal v elocity component directed to the exit duct.In comparison to the con v entional L-shaped abrupt exit the extended top does not increase the solids in v entory in the riser.#2002Elsevier Science B.V.All rights reserved.Keywords:Core Áannulus flow structure;Circulating fluidized bed;Abrupt exit configurations1.IntroductionIn gas Ásolid reactions,bustion,but also for heterogeneously catalysed gas-phase reactions the solids hold-up and the residence time distribution of the particles in the reactor are known to be important for the performance of the respecti v e processes.Numerous experimental in v estigations of circulating fluidized bed (CFB)systems published in the literature ha v e shown that different exit geometries (i.e.smooth or abrupt exits)can ha v e a significant influence on the hydro-dynamics within circulating fluidized bed risers,which result in changes of the v ertical distribution of the solids concentration and thus of the solids residence time distribution.Especially for abrupt exit configurations a significant increase of the apparent solids concentration near the exit has been reported for laboratory and bench-scale CFB units with riser diameters below 200mm.To the present authors’knowledge this effect has not been obser v ed in industrial CFB units.Moredetailed in v estigations are therefore desirable for larger riser diameters.In general,two exit geometries can be distinguished:a smooth exit,if the top exit cur v ature is relati v ely gentle and an abrupt exit,where the exit is at a right angle to the mean (v ertical)direction of flow in the riser.In the case of an abrupt exit configuration,the top of the riser is supposed to act more or less like a gas Ásolid separator,as stated by Grace [1].Many of the particles reaching the top due to their inertia may miss the exit,strike the top wall of the v essel and flow back down near the outer wall of the v essel.If the top exit cur v ature is relati v ely gentle on the other hand,in case of a smooth exit,the particles follow the gas streamlines closely enough that most of them lea v e through the exit [1].Some hydrodynamic studies support this v iew.For example,Brereton and Grace [2]in v estigated the influence of abrupt and smooth exit configurations of a bench-scale CFB facility (riser height 9.3m,inside diameter 152mm).Using an abrupt exit an increase in the apparent suspension density was deduced from the measured axial pressure profile,which was attributed to the inertial separation of particles from the gas stream,leading to an increased solids hold-up for a gi v en net circulation rate.A smooth exit,on the contrary,led to a*Corresponding author.Tel.:'49-40-42878-3039;fax:'49-40-42878-2678.E-mail address:werther@tu-harburg.de (J.Werther).Chemical Engineering and Processing 41(2002)771Á783www.else v /locate/cep0255-2701/02/$-see front matter #2002Elsevier Science B.V.All rights reserved.PII:S 0255-2701(02)00008-9radically different shape of the axial pressure profile at the top from which a continuous decrease of the apparent suspension density with height was deduced. Since this effect will reduce solids back-mixing and promote a uniform residence time of particles,as it is desirable in riser cracking or fast pyrolysis reactors,a smooth exit appears to be optimal for such applications. On the other hand,if,as in combustion,it is desirable to increase the total residence time of solids in the reactor and to promote suspension-to-wall heat transfer,an abrupt exit appears to be a better choice[2].Zhou et al.[3Á5]in v estigated v oidage and particle v elocity profiles in a square cross-sectional CFB riser (146)146mm2cross-section,riser height9.14m)with an abrupt exit configuration.According to their fibre optical probe measurements the riser exit caused on the one hand an asymmetric lateral distribution of v oidage near the top of the riser such that the time-a v eraged v oidage on the exit side was lower than on the opposite side.Because of the exit,gas and particles reaching the top of the riser had a horizontal component towards the exit.On the other hand the lateral profiles of particle v elocities were asymmetric,too.The maximum upward v elocity was shifted to the exit side and the wall layer seemed to be thicker on the opposite side.Similar to the results of Brereton and Grace[2]are those obtained by Zheng and Zhang[6,7],who in v esti-gated the influence of smooth and abrupt exit config-urations in a bench-scale CFB unit(riser height5.25m, 0.102m ID).They pointed out that the exit configura-tion could affect the hydrodynamic characteristics of the gasÁsolid two phase flow in a CFB combustor not only in the immediate v icinity of the exit but also for a considerable distance down the column.When using an abrupt exit geometry the solid particles separated from the flow collided with the roof of the abrupt exit and caused an increase of the bulk density in the top region. If a blinded-T geometry was used,a ca v ity was formed between cyclone le v el and the projected roof.Particles coming from the separated flow entered the ca v ity and were thrown up into the free space and underwent an energy con v ersion from kinetic energy to gra v itational potential energy.With increasing altitude of the parti-cles in the ca v ity,the particle v elocity decreased.If the projected roof was not high enough,the decelerated and concentrated flow of particles would collide with the projected roof and this caused the bulk density to increase further.Therefore a riser exit with suitably chosen geometrical design parameters,including the height of projected roof and the contracted exit area, could enhance the internal solids recycling,hence pro-long the residence time of fine char particles and would enhance heat transfer between wall and the gasÁsolid flow in a CFB combustor.Narrowing the bed cross-section near the exit zone Zheng and Jin[8]also showed that this is a simple and effecti v e way to enhance the internal circulation and reduce the external circulation of bed material simultaneously.Their test results showed a peak v alue of the internal circulation of bed material in CFB with a narrowed bed cross-section near the exit zone.After inserting an obstacle into the duct of an abrupt exit Glicksman et al.[9]obtained also an increase of the solids concentration at the top of the riser.Jin et al.[10]and Bai et al.[11]measured the axial distribution of the solids v olume concentration in a plexiglass CFB cold model(riser height11m,140mm ID)using a fibre optical probe as well as the axial pressure profile.If an abrupt exit with a guiding baffle was used,no back-mixing was obser v ed.The axial distribution of the solids concentration was not influ-enced by the exit configuration.A dilute phase at the top and a dense phase at the bottom of the riser were obser v ed.When using a blind-T configuration instead a significant increase of the solids v olume concentration was obser v ed at the top of the riser.They di v ided the exit section into three regions:a turning-back region, where the solids hit the top of the riser,accumulate and change their flow direction;a region of momentum exchange of solid,where a v iolent momentum exchange happens between solids,which turn downward,and those carried upward by the gas and a traditional fast fluidized region,where the influence of the turning-back particles v anished and the solids mo v ed upward.Jin et al.[10]also showed that the length of the region of momentum exchange was mainly affected by the super-ficial gas v elocity.At low gas v elocities,the influence of the upward flowing gas and solids on the particles turning back was small.When the superficial gas v elocity increased,the v elocity of the upward flowing particles increased also,and the momentum exchange between the upward and the downward falling particles became stronger.This led to a decrease of the length of the momentum-exchange region.Pugsley et al.[12]in v estigated the relati v e influence of smooth and abrupt exit configurations on the axial pressure drop profiles in two circulating fluidized bed risers(riser height6m,0.1m ID and riser height12m, 0.2m ID).The axial profile of the pressure gradient shifted to significantly higher v alues when the riser was equipped with an abrupt exit.To predict the axial pressure drop profile they de v eloped a simple empirical model that corresponded well with the experimental data.It was shown that the influence exerted on the axial pressure profile by the abrupt exit configuration was dependent upon the riser diameter and the particle characteristics.They suggested a mechanism in which particles that were reflected by the abrupt exit may reflux along the riser wall to the base of the column if the riser diameter and the particle terminal v elocity were large enough.If the riser diameter was smaller,the refluxing particle layer could attain only a certainckermeier,J.Werther/Chemical Engineering and Processing41(2002)771Á783 772maximum thickness before the particles were re-en-trained upward by the shearing action of the gas phase. Johnsson et al.[13]studied the effect of the exit geometry on the flow pattern in the riser of a cold1/9 scale model of the Chalmers12MW CFB boiler and compared different exit configurations or rather differ-ent ducts between the riser and the cyclone at different fluidization v elocities.They showed that differences between net solids fluxes during operation with different exits were in some cases of one order of magnitude.A decrease in the net solids flux,which means an increase in the internal back-mixing,could be obtained by inserting obstacles in the upper part of the riser.An increase in the net solids flux was achie v ed by narrowing the duct from the riser to the cyclone inlet.Howe v er, there were almost no differences in the solids flux between an abrupt exit,an extended exit and an enhanced extension.Glicksman et al.[14]reported that particles might also pile up in the horizontal duct connecting the riser exit to the primary cyclone.This phenomenon was obser v ed when a plexiglass duct was employed in their study. With sufficient accumulation of particles,some particles slipped back periodically from the horizontal duct into the riser where they mo v ed downward along the wall. The solids hold-up in a CFB riser is dependent on the solid in v entory in the system as well as on the superficial gas v elocity and,as shown by Schnitzlein and Weinstein [15],can be influenced by the system ing an abrupt and a smooth exit configuration in their0.1m diameter riser they showed that higher externally circulating solids fluxes could be obtained with a smooth exit configuration.Wu and Alliston[16]in v estigated the influence of the fluidizing air on the external solids circulation in a plexiglass pilot-scale CFB cold model(riser height10.7 m,610mm ID,with and without extended top and abrupt exit)and obser v ed the flow phenomena from outside.From their v isual obser v ations they concluded that,using the extended riser top,the solids concentra-tion was greater in that part of the riser where the cyclone inlet was located.Clusters of particles dashed into the extended section and some e v en touched the top of the riser.While mo v ing downward,the clusters gained greater momentum,which allowed them to pass through the suction zone of the cyclone inlet and to continue to fall into the riser column.Without the extended section,the particle mo v ement was different. Here most of the area was co v ered with particles that hit the roof and bounced back to the riser because of their impact on the ing different fluidizing v elocities they showed that an increase of the superficial gas v elocity led to an increase of the externally circulating mass flux.Assuming the same amount of bed material and the same superficial gas v elocity both exit config-urations led to different external solids circulation ing an extended section,the solids circulation rate was lower.They concluded that an extended top section caused more entrained bed material to remain in the riser.It should be noted,howe v er,that all the conclu-sions were based on v isual obser v ation of the riser from outside which enables a judgement of the fluid me-chanics in the immediate wall region only.Van der Meer et al.[17,18]in v estigated the influence of the exit design on the flow pattern in a square cross-section riser(0.14)0.14m cross-section,riser height5.1 m).With regard to the solids fluxes,a larger downflow was found at the wall opposite to the exit and in the corners.The maximum upward solids flux was located near the riser centreline,but was shifted towards the exit side.A reflux ratio was suggested which was defined as the ratio of the solids downward flux in the riser and the external solids circulation rate.The exit design had a large effect on the solids reflux ratio leading to v alues in the range of0.11Á2.7.For example,using abrupt exit geometries the reflux ratio increased with increasing height of the extended section.In summary,the literature indicates a marked influ-ence of the exit design on the local CFB fluid mechanics. Howe v er,many conclusions ha v e been drawn from bench-scale in v estigations.Since scale-up is a general problem in fluidized-bed technology it was decided at the beginning of the present work to in v estigate the exit effects on riser hydrodynamics on a somewhat larger scale with a v ariety of measurement methods including pressure measurements,local probe measurements and high-speed v ideo obser v ation from the outside.2.ExperimentalThe experiments were carried out in a circulating fluidized bed cold model with a cross-section of 0.3)1.0m2and a height of8.5m(cf.Fig.1).The fluidizing air is pro v ided by a200kW rootes blower and distributed by20bubble caps at the base of the riser. The solids are separated by a two-stage cyclone system and returned into the riser v ia a siphon at a height of1 m with an angle of458.The externally circulating mass flux is measured by a weighting section in the down-comer pipe.For the obser v ation of flow phenomena from outside se v eral obser v ation windows are installed. To measure the pressure profile along the riser,the riser is equipped with16pressure transducers whereas at the down-comer pipe fi v e pressure transducers are located. The bed material used was quartz sand with a surface mean diameter d p of140m m(terminal v elocity u t01m/ s,minimum fluidization v elocity u mf00.3m/s,both with air at ambient conditions).The facility can be equipped with two different v ersions of an abrupt exit:A so-called L-shaped configuration(as shown in Fig.1)and a T-shapedckermeier,J.Werther/Chemical Engineering and Processing41(2002)771Á783773(blind-T,extended top)configuration.A detailed sketch of the riser top is shown in Fig.2.If an L-shaped exit configuration is used the o v erall height of the riser is 8.5m as mentioned ing a T-shaped configuration an additional section of 0.5m height is added on the top of the riser so that an o v erall riser height of 9.0m is obtained.The transition to the cyclone starts at a heightof 7.57m.It has a width of 0.16m and a height of 0.86m.Local solids v olume concentrations and v elocities were measured by means of a two-channel fibre-optical probe,as shown in Fig.3.The probe as well as the measurement principle ha v e been described in detail by Rensner and Werther [19].Light emitted by a laser diode is sent into the measuring v olume through an optical fibre and is reflected according to the amount of particles in front of the tip.The back-scattered light is detected by a photo diode,which con v erts the signal into an output v oltage.From this output v oltage a local instantaneous solids v olume concentration can be cal-culated using a calibration function [19].ThetwoFig.1.Sketch of the circulating fluidized bedunit.Fig.3.Operating principle of a fibre optical probe and a sketch of the two-channel probe head (dimensions in mm,U 0v oltage).ckermeier,J.Werther /Chemical Engineering and Processing 41(2002)771Á783774sensing surfaces are arranged v ertically abo v e each other at a distance of 2.3mm(see Fig.3).Using cross-correlation of the two signals the v ertical v elocity component of the mo v ing gasÁsolid structure can be determined.In addition,the flow in the exit region was obser v ed from outside v ia a high-speed v ideo camera.The camera was the Kodak Ektapro Motion Analyser1012(EM 1012)with an integrated image intensifier.The v ideo camera in combination with the image intensifier allows frame rates up to6000frames per second(fps),exposure times down to10m s and a picture resolution of 192)239pixels with256grey scale le v els.A tele-lens (f0180mm)was chosen for the obser v ation.For the direct illumination a con v entional photo lamp was used that was positioned at an angle of458to the camera axis to be able to capture structured images due to light-and-shadow effects.To apply image processing methods the frame work of KHOROS(de v eloped by Khoral Research Inc.)was used.KHOROS includes se v eral standard algorithms of signal and image processing where specific user-defined algorithms can be implemented easily.For some algo-rithms MATLAB(MathWorks Inc.)has been used.A detailed description of the applied image processing may be found in Ref.[20].3.Results and discussion3.1.Pressure measurementsAs has been shown by the numerous in v estigations cited abo v e abrupt exit configurations can lead to a significant increase of the apparent solids concentration calculated from the axial pressure profile near the top of the riser.Therefore,as a first approach to in v estigate the fluid mechanics in the exit zone of the CFB riser,the axial profile of the apparent solids concentration was ing the measured axial pressure profile the apparent solids v olume concentration c v,D p has been calculated fromCv;D p 0D pD yg rs(1)where g is the acceleration due to gra v ity and r s is the solids density.In Fig.4the v ertical profiles of the apparent solids v olume concentration,c v,D p,are shown for both exit configurations at different superficial gas v elocities and rates of external solids ing a constant solids in v entory of300kg in the system the superficial gas v elocity was increased from2up to4m/s. At the bottom of the riser high apparent solids v olume concentrations are obser v ed,which rapidly decrease with increasing height.Since a constant solids in v entory is used,the height of the bottom zone decreases with increasing gas v ing either the L-shaped configuration or the T-shaped configuration no increase of the solids concentration at all is obser v ed in the upper part of the riser or near the exit.These results are in good agreement with measurements for instance ob-tained in a1:9scale model of the12MWth CFB combustor at Chalmers[13],at the CFB combustor at Chalmers itself[21]and at the600MWth CFB boiler in Gardanne[22].It might therefore be concluded that the increase of the solids concentration near the exit,which has been reported by other authors,is a small-scale effect,which is typical of bench-scale CFB risers as has been indicated already by Pugsley et al.[12].In addition to the total riser pressure drop the externally circulating solids mass flux G s based on the riser’s cross-sectional area has been measured.Gi v en a constant solids in v entory in the CFB system,the riser in v entory and the external circulation rate G s adjust themsel v es for a preselected fluidizing v elocity u.The dependence of the externally circulating solids mass flux on the pressure drop D p R o v er the riser for different superficial gas v elocities is shown in Fig. 5.With increasing superficial gas v elocity the riser pressure drop decreases and the externally circulating solids mass flux G s increases for both exit configurations. Comparing the results with regard to the different exit configurations,it follows that with the L-shaped con-figuration slightly higher externally circulating rates are obtained for a gi v en solids hold-up,and gas v elocity, respecti v ely.On the other hand,if we look for a gi v en gas v elocity and gi v en G s v alue at the resulting riser pressure drop it seems that a T-shaped exit would lead to a lower riser pressure drop and thus to a lower solids in v entory in the riser.Howe v er,before taking such conclusions one should bear in mind that the accuracy of the G s measurement is fairly low and will probably not be better than910%.Gi v en this inaccuracy it becomes a bit doubtful whether the differences between the two exit configurations are significant.3.2.Analysis of high-speed v ideo imagesFor a detailed in v estigation of the flow phenomena in the wall region of the exit zone high-speed v ideo images were recorded.The high temporal resolution of the v ideo system re v ealed a complex flow structure in the wall layer.The entire two-dimensional flow pattern with different degrees of local turbulence was e v aluated with the systematic methods of digital image analysis.Fig.6depicts an example of the results of the digitally analysed two-dimensional solids v elocity field obtained from two succeeding images(frame rate:1000fps) pro v ided by a two-dimensional correlation technique. The image was subdi v ided into10)12subregions,on which the correlation procedure was applied.For eachckermeier,J.Werther/Chemical Engineering and Processing41(2002)771Á783775subregion a characteristic v elocity v alue is e v aluated,which is drawn in Fig.6as a v elocity v ector (arrow)fixed in the centre of the corresponding subregion.Although the photo shows a v ery irregular flow structure the v elocity field has quite a regular appear-ance.In Fig.7the time-a v eraged v alues of the horizontal and v ertical solids v elocities obtained from the image analysis at different locations on or close to the side wall (z 00.3m,cf.Fig.8)are shown as a function of the distance x from the side wall for the L-shaped exit configuration.For each of the time-a v eraged v elocities a time-inter v al of 1.683s (which corresponds to nine recording periods of 0.182s duration)was taken into account.It can be seen that the horizontal v elocities start with about zero at the left hand wall which isopposite to the exit to the cyclone and increase linearly until v alues between 1.0and 1.5m/s near the entrance of the cyclone.Note that the exit duct leading into the cyclone is between y 07.57and 8.43m.This means that the position relati v e to the entrance of the cyclone is important with respect to the horizontal v elocity.The shorter the distance to the cyclone,the higher the horizontal v elocity.For the different heights within the exit region nearly no influence on the profiles of the horizontal v elocity can be seen.In contrast to the horizontal v elocity,the profiles of the v ertical v elocity components in Fig.7show no influence of the horizontal position x for both exit configurations.Considering the dependence on the v ertical position a slight influence can be seen.The solids flow structures near the top of the riser ha v e downward v elocities between 0.7and 1m/s.With increasing distance from the top the downward v elo-cities increase to v alues between 1.5and 2m/s.This obser v ation may be explained by the fact that at the top of the riser the solids hitting the top wall are returned into the downflow direction.The v elocity at the wall near the top is therefore lower than further down in the riser where the increase of the downward v elocity,with increasing distance from the top,can be explained by the acceleration due to gra v ity and also by the increasing thickness of the wall layer [23].In Fig.8the frequency distributions of the two v elocity components are plotted as grey v alues (the darker the grey v alue,the higher the frequency).Data from 3)3calculation positions,that means a cross-section of 2.66)2.66cm,o v er a time inter v al of 1.683s ha v e been taken into account (10368v elocity v alues).The chosen positions are located in the edges of the obser v ation window.Similar to the results shown before it can be seen,that the downward v ertical velocityFig.4.Profiles of the apparent solids v olume concentrations c v ,D p deduced from the axial pressure profile at different operating conditions for both exitconfigurations.Fig.5.Externally circulating solids mass flux G s and riser pressure drop D p R depending on the superficial gas v elocity for both exit configurations (open symbols:L-shaped exit,closed symbols:T-shaped exit;total solids in v entory 300kg).ckermeier,J.Werther /Chemical Engineering and Processing 41(2002)771Á783776increases with increasing distance from the top wall,whereas the horizontal v elocity increases with decreas-ing distance to the cyclone inlet duct.3.3.Probe measurementsIn order to obtain more detailed information about the flow phenomena inside the exit zone fibre optical probe measurements were carried out at different v ertical and horizontal positions inside the riser top.The different measurement locations are shown in ing the two-channel fibre optical probe,both solids v olume concentrations and v elocities were measured.In Fig.9profiles of the local time-a v eraged solids v olume concentrations along tra v erses perpendicular to the front wall (i.e.in the z -direction,cf.Fig.2)at different distances x from the left side wall (which is the wall opposite to the cyclone inlet)are shown for two different heights inside the T-shaped exit configuration.At a height of 7.9m.i.e.near the of centre the cyclone inlet duct low solids v olume concentrations between 1.5and 2v ol.%are obser v ed in the middle of the riser (z 0150mm),which remain more or less constant in the core region.With increasing proximity to the front wall (z 0300m)the concentration increases to v alues between 2and 2.5v ol.%.This is the same flow pattern *low solids v olume concentrations in the core and higher solids v olume concentrations in the wall region *that pre v ails throughout the whole upper dilute zone.Considering the distance x from the side wall nearly no influence of the v icinity of the exit duct to the cyclone can be obser v ed.E v en the tra v erse measured at x 00.9m,i.e.0.1m before the exit duct,shows the same characteristic core Áannular structure.With increasing height abo v e the gas distributor the solids concentration in the core as well as in the wall region decreases.Inside the extended top abo v e the transition to the cyclone,i.e.at a height of 8.75m,lowFig.6.Two-dimensional flow patterns and instantaneous v elocities in the exit wall region (L-shaped exit configuration,u 03m/s,G s 026.5kg/m 2s,window size 20)16cm 2).Fig.7.E v aluation of high-speed v ideo images:time-a v eraged hor-izontal (v x )and v ertical v elocities (v y )at different locations (L-shaped exit configuration,u 03m/s,G s 026.5kg/m 2s,the exit duct to the cyclone is at x 01m,it starts at y 07.57m and ends at y 08.43m).ckermeier,J.Werther /Chemical Engineering and Processing 41(2002)771Á783777。