Magnetic resonance studies of a gas-solids fluidised bed: Jet-jet and jet-wall interactions

Particuology 8(2010)617-622Contents lists available at Science DirectPARIICLOLOGYparticuologyELSEVIERjournalhomepagewww.elsevier.com/locate/particMagnetic resonance studies of a gas-solids fluidised bed Jet-jet andjet-wall interactionsMeenal Pore, Daniel Holland Thusara C Chandrasekera, Christoph R muller bAndrew ]. Sederman, John S Dennis ,* Lynn F Gladden, John F. DavidsonDepartment of Mechanical and Process Engineggstrasse 3, 8092 Zurich, SwitzerlARTICLE IN FOABSTRACTMagnetic resonance imaging(MRI)gave images of air jets from orifices in the distributor plate of a bedof poppy seeds. Attention focused on two features10July2010(1)The interaction between nearby vertical jets from two, three or four orifices:uidised bed(2)Wall effects, where one or more orifices created vertical jets near the vertical wall of the cylindercontaining the particle bed.Distributor designThe results show that nearby jets are mutually attracted likewise a jet near a wall bends out of thevertical, towards the wall. For multiple adjacent jets, the jet lengths show dependence on orifice layoutthe lengths are in reasonable agreement with published measurements, by other methods, for single jets.The MRi gives three-dimensional images of the single jets and of multiple jets, separate or mergingo 2010 Chinese Society of Particuology and Institute of Process Engineering, Chinese Academy ofSciences. Published by Elsevier B V. All rights reserved1. Introductiontheir velocities, in both single-and two-phase granular systems,whereas, e.g. ECT is limited to measurements of the void fraction1. 1. Magnetic resonance imaging and fluidisationand the velocity of voids, i.e. the rise velocity of bubbles and slugsAdvances in MRI now allow imaging of particulate systems toDespite the widespread use of fluidised beds for processing, high spatial and temporal resolution. The technique is limited byvarious fluidisation phenomena are still poorly understood mak- the need to have particles composed of materials containing MR-ing design and scale-up difficult. Fluidised systems are optically sensitive nuclei ( typically H)which are reasonably mobile at aopaque so that it is difficult to observe the behaviour of particles molecular level, by the bore size of the magnet and by the needand gas within the bulk of the bed. Observations on 2D beds with to eliminate metals from experimental apparatus being studiedtransparent walls give indications, but, owing to wall effects, these Muller et al. (2008)have described how relevant measurements ofmay not be representative of the hydrodynamics in a 3D bed. Only fluidisation phenomena can be made. For example, time-averageda few experimental techniques exist to provide non-intrusive mea- 3D intensity maps, constructed from spatial maps of signal inten-surements in opaque granular systems, the most important being sity, and hence particle concentration, can be used to calculate localpositron emission particle tracking(Stein, Martin, Seville, McNeil, values of the voidage and to re-construct stable structures, such a7), electrical capacitance tomography(du, Aristo, jets. Time-averaged maps of particle velocity in 3D can be obtainedFan, 2006), X-ray attenuation( rowe Yocono, 1976 )and mag- with spatial resolution within a plane of typically -1 mm x 1 mmtic resonance imaging( holland et al., 2007: Sederman, Gladden, and axial resolution of -0 2 mm, although the acquisition time isMantle, 2007)2.5 h To image temporally-varying structures, such as bubblinlet he advantage of magnetic resonance(MR)(e. g. Muller et al, jets(Grace Lim, 1987 ), ultra-fast techniques have been devel-08)is that it can measure the distribution of solids (voidage ) and oped(Muller et al., 2008)giving, typically, a temporal resolutionof 25 ms and spatial resolution of 1.7 mm x 4 mm. Finally, measurements of palCorresponding author. Tel. +44 1223 334787: fax: +44 1223 334796of material is中国煤化工E-mail address: jsd3@cam ac uk(S Dennis).degree of gas-seHCNMHGanalogous to tracer1674-2001/s-see front matter o 2010 Chinese Society of Particuology and Institute of Process Engineering, Chinese Academy of Sciences. Published by Elsevier B V. All rights reserveddoi:10.1016/ 1. partic.2010.07.015M. Pore et al / Particuology 8(2010) 617-622Results from various studies on jetting beds have mainly been summarised as correlations between a scaled jet length(e.g. Li/do, whereifice diameter(m)Li is jet length and do is the diameter of the orifice)and eitherparticle diameter(m)the orifice velocity Uo, or the superficial gas velocity, Us. Rees etDjet diameter(m)al. (2006)studied the effect of the pitch between holes on lengthgravitational acceleration(m /s)of jet. The lengths were found to increase with decreasing pitch,distance between centre of orifice and wall(m)suggesting that jets tend to stabilise each other This agrees withitical distance between the centre of orifice andthe correlations of Blake et al. (1990), showing that longer jets areformed in beds with multi-orifice distributors compared to thosewith just a single orificeLi jet length(m)Early work on jetting was conducted in 2D beds as these allowedQ volumetric gas flowrate through an orifice(m/s)easy, visual observations. However, the correlations from thesevolumetric flowrate through an orifice at minimumfluidisation(m/s)studies consistently predicted the jet lengths to be larger than theUs superficial gas velocity(m/s)values measured in 3d beds using optical probes, pressure sen-sor methods, etc ( Wen et al., 1981). This suggests that a jet is alsominimum fluidisation superficial velocity(m/s)tabilised by a wall and that a longer permanent void is able toorifice jet) velocity(m/s)exist before bubble formation occurs Kimura. horiuchi, Watanabedynamic viscosity(ns/m)Matsukata, and Kojima(1995)reported that bubbles formed fromjet half anglea single, centrally-positioned jet from a nozzle 3 or 6 mm in diam-Pffluid density(kg/m)article density( kg/m)eter in a 3D cylindrical bed, 50 mm I.D., moved horizontally oncethey separated from the tip of the jet and ascended near the wallIt is likely that a jet near the wall region might have been stabilisedas a permanent void over a large distance before breaking uexperiments, such as conducted in positron emission particle track- bubblesing(pept).To summarise, understanding the effect of jet-jet and jet-walinteractions on jet structure is important in optimising distributor1. 2. Imaging of gas jets in fluidised bedsdesigns. This paper investigates the effect of jet-jet and jet-walinteractions in a gas-solid fluidised bed containing either geldartThis paper is concerned with the use of Mr to study the forma- Group B or Group d particles using MRL. The measurements of jettion and evolution of jets of gas injected into a gas-solid fluidised length and volume were used to examine the spatial stability ofbed such as those occurring at the holes in a multi-hole distribjets in the vicinity of (i)other jets and (ii)a walltor plate. This is important in the design of fluidised bed reactorsfor three reasons. Firstly, in industrial-scale reactors, much of thegas-solids contacting may occur in the region near the distribu2. Experimentaltor where the bubbles are still small. Higher in the bed, the smallbubbles coalesce forming large fast-moving bubbles leading to gas2.1. Fluidised bed apparatusbypassing and poor contacting. The design of the distributor influences the size and shape of the bubbles formed and is therefore aThe bed was contained in a Perspex column, shown schematimajor factor in determining gas-solids mixing in gas-solid bedslly in Fig. 1(a). The plenum chambers below each distributor hadSecondly, the velocity of the gas through the orifices in multi-hole a volume of 200 cm to dampen fluctuations in gas pressure frondistributors can be an order of magnitude higher than the super- compressed air supply. Three types of distributor were usedficial velocity of gas in the bed, leading to the formation of jets inwhich the particles are travelling at a high velocity relative to the 1) To study jet-jet interaction, four distributors with differentrest of the bed. this can result in the erosion of walls or bed interarrangements of orifices, each 2 mm in diameter, were used, asnals, such as cooling tubes, within the region affected by the jetsshown in Fig. 1(b). The distributors had, variously, two holesFinally, dead zones(regions of stationary particles) can be found(M2), three holes in a linear pitch(M3a), three holes on a tri-on distributors between the orifices(Rees et aL., 2006). Materialangular pitch(M3b)and four holes on a square pitch(M4)in these zones is poorly fluidised and can result in the sintering ofTwo different geometries of three-hole distributor were used tosolids and fouling of the distributor when exothermic reactions arestudy the effect of orifice arrangement on jet-jet interactionsbeing undertakenThe orifice spacing was chosen to be large enough to ensurePrevious investigations on the behaviour of jets in fluidised bedsthat wall effects were negligible, but far enough apart to formive used pressure signal analysis( Vaccaro, Musmarra, Petreccaindividual jets at the orifice which could be imaged using MRI997), optical probes(Wen, Deole, Chen, 1981; Blake, Webb, 2) To study jet-wall interactions, a distributor with four holes atsunderland, 1990 ), triboelectric probes (ohn, Reischl, devorvarying distances from the wall was used with an orifice diame1980: Berruti, Dawe Briens, 2009)and thermal tracer methter of2 mm, as shown in Fig. 1(b)(W1). The holes were arrangedods(berruti et al., 2009; McMillan et al., 2005; Zhu, Wang, Fan,to minimise any interaction between the jets themselves. The2000). These methods are intrusive since the probes distort the maximum orifice-wall separation was 10 mm, selected to bejet flow, but they are easy to use in large-scale, opaque systemsgreater than the maximum radius(0.5 D; =8 mm)expected forX-radiography(Cleaver, Ghadir, Tuponogov, Yates, Cheesman, a jet at the flow rates used, so that the jet would not, in princi-1995: Rowe, MacGillivray, Cheesman, 1979)has also been usedolealler+ nrifirnwall distance wasto image jetting. More recently, x-ray computed tomography has4 mm. the mini中国煤化工1ng the distributorbeen used to produce stacked 2D images of a fluidised bed (Mudde,plate during its2010). Rees et al. (2006)and Muller et al. (2009 )used mri to image 3)A distributor(S1CNMHGas used to producethe region just above a perforated-plate distributor and were able a single jet with negligible wall effects. The results from jet-jetto image zones of defluidised material, as well as the jet structures.and jet-wall distributors were compared with those from theM. Pore et al. Particuology 8(2010)617-622r f coilbparticlescentre of r.f.mmM3aM3b3 mm60 mmam12.7mm6nylon tailpieceand gas inlet24 mm4 mm 8Fig 1.(a) Experimental set up of the jetting bed and (b) layout of the distributors used (orifice diameter do=2 mm; S1: single orifice, M2: 2 holes, M3a: 3 holes in a lineartch, M3b: 3 holes in an equilateral triangular pitch, M4: 4 holes in a square pitch and w1: wall distributor with wall-orifice separation distances of 4, 6, 8 and 10 mm).single orifice jet to quantify the stabilising influence of interac- 3. Resultstionsmaging theTo image the solid phase using MR, the particles must containa high concentration of mobile protons. Poppy seeds were usedThe typical greyscale slices from the 3D results in Fig. 2(a) showin this study because of their high content of oil Seeds of diameter jets formed at a wall distributor (W1)with wall-orifice separation0.5 mm (Group b)and 1. 2 mm(Group d), with measured minimum distances of 4 mm(Fig. 2(a), left)and 8 mm( Fig. 2(a), right). Fig. 2(b)fluidisation velocities at 298 K, Umf, of 0.13 m/s and 0.30 m/s, and is the binary-gated image of Fig. 2(a)from which geometrical proparticle densities of 1060 kg/m3 and-960 kg/m, respectively, erties are measured. Fig. 2(c)shows a schematic representation ofwere used. The height of the bed at incipient fluidisation was con- the geometry of an axisymmetric jet; however, the jet angle, e, andstant at 115 mm for the Group B seeds and 85 mm for the Group jet width, Dj, are impossible to define in asymmetric jets formedD seeds. The bed was left to stabilise for 300 s after any change in when jet-jet or jet-wall interactions occur, as seen in Fig. 2(b)as flowrate, before measurements were made. The bed was oper- (left ) The halo' seen around the jets is an MR artefact caused byted at atmospheric pressure and the particles were fluidised with the short repetition time between adjacent slices and by particleMulleret al (2009)reported that the start-up method used affected excite( etween excited slices; it is not seen when a single slice isair supplied at 1 bar(gauge), metered to the bed via a rotameter.the geometry of the jets formed To improve the reproducibility ofHere, the term jet is used for a pIt void formed abovethe results, the bed was filled with particles whilst air was flowing an orifice, as defined by Rowe et al. (1979 ) Two types of structurethrough the bed at a rate corresponding to minimum fluidisation, of permanent jets have been reported in the literature: (i)a conicalQmf, and the measurements at the highest velocity were taken first void with an ellipsoidal or hemispherical top and (ii)a stable jetand the gas flow rate was gradually decreased2.2. Magnetic resonance imagingc)DA Bruker DMX200 spectrometer was used to image the particles,mploying a proton('H)frequency of 199. 7 MHz with an activelyshielded gradient system, capable of producing a maximum gra-ent strength of 0. 139T/m a birdcage radiofrequency(r f ) coilinternal diameter 63 mmsed to excite and detect 1h nucleE)pulse sequence(muller et al., 2009)55 mmas used to produce time-averaged 3D images of the distributorgion. A repetition time, TR, of 730 ms and an echo time, Te, of2.51 ms were used, with four averages per slice. 2D slice imag(x-y plane) were acquired at 128 x 128 pixels with a field of view of55 mm x 55 mm, giving an in-plane spatial resolution of 0.43 mmIn the axial direction 41 slices, each 1 mm thick, were acquiredgiving an axial spatial resolution of 1 mm. The images producedwere time-averaged over 300 s The temporal stability of the jets中国煤化工was studied using an ultra-fast imaging sequence(muller et al2009). Ultra-fast two-dimensional axial and transverse slicesFig. 2.(a)z-y multiCN MH GGroup d particlesall jet distributor (w1)with 7-4 mm(left)and 1-8 mm(right),()binaryimaged with a temporal resolution of 25 ms and a spatial resolution image from(a)(grey region indicates the position of the distributor) and (c)theof 1.7 mm in the radial direction and 4.4 mm in the axial direction. geometrical properties of a jetM. Pore et al / Particuology 8(2010) 617-622s5 mmA mmU=11.0 cm/sU=12. 7 cm/sU=15.3cm/sFig 3. Greyscale intensity x-z slices(a-c)and 3D reconstructions of jets(d-f)from a 3-hole linear(M3a) distributor with Group B particlesstem with an upper region where bubbles break away periodically istic void due to the jet was no longer present. The value of each jet(Grace Lim, 1987; Tsukada Horio, 1990). From Fig. 2(a)it can be volume was calculated directly from the 3D resulseen that the first structure was observed in this study. At low ratesFig. 4 shows how the measured length of jets increased withof gas flow, axisymmetric jets formed above each orifice. As volu- increasing orifice velocity Uo, in line with the results of othersmetric gas flow rate, Q, approached that for minimum fluidisation, (Blake et al., 1990: Merry 1975: Muller et al., 2009). As the orificeQmf, the jets began to interact, with the tops of the jets curving gas velocity is increased above Uo=15 m/s for Group B seeds andtowards each other. Above Qmf the jets merged and gas passed Uo-45 m/s for Group D seeds, multiorifice distributors produce jetsthrough the centre of the bed creating a dilute core. This is seen in that are longer than those from a single orifice distributors. At3(b)and (c)(Us=12.7 cm/s and 15.3 cm/s with Umf=13 cm/s). gas velocities, at which the jets do not interact, all the distributorsFor the three-and four-holed distributors(M3a, M3b and M4)the produce the same length of jet for a given value of Uo. It can alsojets all merged at the top. For the two-holed distributor (m2 )one be seen in Fig 4 that smaller lengths of jet were produced usingjet formed a spout which broke through the bed and the other jet Group d seeds than Group B, in agreement with previous work byFilla, Massimilla, and Vaccaro(1983)and Musmarra(2000)Fig 3(a)-(c)shows, for the M3a distributor, greyscale intenFig 5 shows that the volume of a jet is dependent on the orificeity x-z slices from 3D intensity maps, whilst Fig 3(d)-(f)shows velocity but not on the number of holes in the distributor3D reconstructions of the binary gated results, derived fromFig 6 shows the effect of orifice-wall distance on jet length forFig 3(a)-(C), for increasing superficial velocity, Us, of gas. At the the wall distributor(W1)compared with a jet lengths for a sinlowest velocity, the middle jet is taller than the outer jets. The 3d gle orifice distributor(S1)and correlations from literature. The jetreconstruction (Fig 3(d)) shows that the three jets are still dis-tinct at this flow rate. At Us-12.7 cm/s, the two outer jets becomelonger and the 3D reconstruction( Fig 3(e))shows the tops of thejets bending towards each other. At Us=15.3 cm/s, the jets havemerged at the top( fig 3(c))and the top of the middle jet is movedoff the central axis(Fig 3(f)302M3b (Group D3. 2. Quantitative analysisJet dimensions, shown in Fig. 2(c), were calculated from binary-gated, multi-Slice, spin-echo images, such as those derived frombhe(9为0(Oro甲BBake(1990)(Grop D)Fig 3(a)-(c). Since tall jets were observed sometimes to rotateabout their nominal vertical axes higher up in the bed, measur中国煤化工ing jet heights from axial slices was inaccurate as the top of the jetCNMHGmight have moved out of the plane of the slice. Instead, the value ofFig 4. Mean jet length(Li)versus orifice velocity for multi-orifice distributors com-the jet height was determined by examining successive transverse pared with jet length correlations from Blake et al. (1990) for the distributorsslices in the x-y plane and noting the point at which the character- M2-M4 and S1 shown in Fig. 1(b)M. Pore et al. Particuology 8(2010)617-6221400Fig. 5. Mean volume of one jet as a function of orifice velocity for Group B particles Fig. 7. Scherfor the distributors M2-M4 anshown in Fig. 1(bdiagram of a wall jet with(a)Us Umf, (b)Us- Umf and (c)Us>for /15 m/s(Us approaching Umf)with the Group b al. (1983 )also reported lower jet lengths with increasing particleparticles, it was found that the tops of the jets bent towards each diameter, attributing it to increased momentum transfer betweenother, indicating that jet-jet interactions then occur. Indeed, for the gas and particles for larger particles, leading to a faster dissipaUo>15 m/s, jet length is affected by the number of holes and their tion of the gas momentum, and hence shorter jetsgeometric arrangement, with more holes giving longer jet lengthsdicating that jet-jet interactions stabilise jets5. ConclusionsIt was found that the volume of a single jet was not affected byV凵中国煤化工the number of orifices. However, the length of jets formed at multiIn this studyorifice distributors was found to increase as the number of orifices fluidisation systCNMHGts in visually opaqueporal resolution, giv-increased. Therefore longer, narrower jets will be formed from a ing quantitative and qualitative information on jets in a 3d beddistributor with more orifices for a given value of UMaps of solids concentration obtained in this study have been usedM. Pore et al / Particuology 8(2010) 617-622to investigate the effect of jet-jet and jet-wall interactions on jet Grace, J.R.& Lim, C.]. (1987). Permanent jet formation in beds of particulate solidsstability.The Canadian Journal of Chemical Engineering, 65(1), 160-162Jet-jet interactions were found to stabilise jets in beds of GroupHolland, D ]. Muller, C.R., Davidson, J. F, Dennis, J.S., Gladden, L.E.urst.A N.et al. (2007). Time-of-flight variant to image mixing of granularB and group d particles, with minimum fluidisation velocities, Umf,of Magnetic Resonance, 187(2),199-204of 0. 13 m/s and 0.30 m/s, respectively Jet-jet interactions occurred John, WReischl, G.& Devor, w(1980). Charge transferat orifice velocities, Uo>15 m/s(i.e Us approaching or greater than Kimura, T. Horiuchi, K. Watanabe, T. Matsukata, M. Kojima, T. (1995). Experimaps For Uo>15 m/ s length of jet increased with increasingUmt) and were observed quantitatively using 2D and 3D intensmental study of gas and particle behavior in the grid zone of a jettingbed cold model Powder Technology, 82(2). 135-143.reported in previous studies, and also with increasing number ofMcMillan, ], Zhou, D, Ariyapadi, S, Briens, C, Berutti, F, Chan, E(2005). Characteruid spray droplets and partiorifices. The length of jet for multi-orifice distributors was greaterd Industrial and Engineering Chemistry Research, 44(14), 49than that for a jet from a single orifice, showing that jet-jet inter- Merry, J (1975). Penetration of vertical jets into fluidised beds. AlChE Journal, 21actions stabilise jets. It was also found that the number of orificedid not affect jet volume. For the same distributor and Uo, there isMudde, R F (2010). Time-resolved X-ray tomography of a fluidised bed. Powdermore gas leakage into the particulate continuous phase for GroupMuller, C.R., Holland, D J. Davidson, J. F, Dennis. .S, Gladden. L F, Hayhurst, A.D particles than for Group bN, et al. (2009). Geometrical and hydrodynamical study of gas jets in packedJets near the wall of the bed are stabilised by the wall and cannd fluidised beds using magnetic resonance. The Canadian Journal of Chemicalring.87(4).517-52m.83ause spouting at the bed depths used in this study. It was found Muller, C.R. Holland, D J Sederman. A] ,Mantle, M D, Gladden, LF,&Davidson,that jets from an orifice at I-4 mm from the wall were twice as) Magnetic resonance imaging of fllong as jets from a single orifice, located away from a wall. For the Musmarra, D(2000). Influence of particlebed and conditions used in this study the stabilising effect of thelength in gas fluidised beds. Industrial and Engineering Chemistry Research, 39(7).wall extended crit=6 mm into the bed. jets with I> crit were notstabilised by the wall and were of the same length as a single jet atRees, A C, Davidson, J F, Dennis, J S, Fennell, P. S, Gladden, L F, Hayhurst, A N, etthe centre of the beda gas-fluidised bed, as imaged using magnetic resonance Chemical engineerinScience,61(18),6002-6015AcknowledgmentRowe, P, MacGillivray, H,& Cheesman, D(1979). Gas discharge from an orificeinto a gas fluidised bed. Transactions of the institution of chemical engineers194-199This work was funded by the Engineering and Physical Sciences Rowe, P N,& Yocono, C (1976). The bubbling behaviors of fine powders whenResearch Council (Grant number EP /F041772/1)Sederman, A J, Gladden, L F, Mantle, M. D (2007). Application of magnetic resoReferences8(1),23-38.Stein, M, Martin, T W. Seville, ]. P, K, McNeil, P. A,& Parker, D ] (1997). PositronBerruti, F, Dawe, M.,& Briens. C (2009). Studys-solid fluidised bed. Powder Technology,到2如2如mntions In). Chaouki, F. Arachi, M. P. Dudukovic(Eds , Nonasidenitoring of multiphase flows(pp. 161-184). Amsterdam: Elsevier.Blake, T, Webb, H,& Sunderland, P (1990). The nondimensionalization of equa- Tsukada, M,& Horio. M.(1990). Gas motion and bubble formation at the distributortions describing fluidisation with application to the correlation of jet penetrationof a fluidised bed. Powder Technology, 63(1), 69-74Cleaver,J.AS, Ghadiri, M. Tuponogov, V G. Yates, J G,&Cheesman, D J(1995). Vaccaro, S, Musmarra, D,& Petrecca, M (1997). Evaluation of the jet penetrationgineerIngDu, B, Waristo, W.& Fan, L. S(2006). Flow dynamics of gas solid fluidised bedsWen, C,, Deole, N,& Chen, L(1981). A study of jets in a three-dimensional gasFilla, M, Massimilla, L,& vaccaro,S (1983 ). Gas jets in fluidised beds: The influenced bed. Powder Technology, 31(2). 175-184particle size, shape and densitand solids entrainment. InternationalZhu, C, Wang, x, Fan, L -S(2000). Effect of solicJournal of Multiphase Flow, 9(3)liquid jets in gas-solid flows. Powder Technology, 11中国煤化工CNMHG