This is an author-deposited version published in: Eprints ID: 6168

Transcription

1 Open Archive Toulouse Archive Ouverte (OATAO) OATAO is an open access repository that collects the work o Toulouse researchers and makes it reely available over the web where possible. This is an author-deposited version published in: Eprints ID: 6168 To link to this article: DOI:1.116/J.POWTEC URL: To cite this version: Santos, Edgar Tavaresdos and Hemati, Mehrdji and Andreux, Régis and Ferschneider, Gilles (211) Gas and solid behaviours during deluidisation o Geldart-A particles. Powder Technology, vol. 211 (n 1). pp ISSN Any correspondence concerning this service should be sent to the repository administrator: sta-oatao@listes.di.inp-toulouse.r

2 Gas and solid behaviours during deluidisation o Geldart-A particles E.T. Santos a,b,, M. Hémati a, R. Andreux b, G. Ferschneider b a Laboratoire de Génie Chimique de Toulouse UMR 553, INP-ENSIACET, 4 Allée Emile Monso, BP 44362, 313 Toulouse, Cedex 4, France b IFP Energies nouvelles, BP3, 6936 Solaize, France a b s t r a c t Bed collapsing experiments were carried out in a cold-air transparent column 192 mm in diameter and 2 m high. Typical Fluid Catalytic Cracking (FCC) catalyst with a mean particle size o 76 μm and a density o 14 kg/m 3 was used. Both single and double-drainage protocols were tested. The local pressure drop and bed surace collapse height were acquired throughout the bed settling. Typical results were ound regarding dense phase voidage o a luidised bed and the bed surace collapse velocity. In addition, bubble raction was calculated based on the collapse curve. Experimental results showed that windbox eect is signiicantly reduced compared to previous works since the volume o air within the windbox was reduced. The comparison o single/double-drainage protocols revealed a new period in the deluidisation o Geldart-A particles concerning gas compressibility. Through the temporal analysis o local pressure drop, the progress o the solid sedimentation ront rom bottom to top was determined, analysed and modelled. 1. Introduction The downward convey o gas solid dense mixtures in standpipes is a major issue in the Fluid Catalytic Cracking (FCC) process, which is traditionally the major conversion unit o heavy liquid oils to lighter and more valuable gas products in a reinery. The dense transport can be best achieved whilst maintaining the particles in a moderately aerated state with interstitial gas. The aerated gas-particle suspension behaves as a luid, thus the pressure along the vertical pipe increases. The static pressure head is directly related to the apparent density o the suspension and the height o the pipe. Standpipe lows are certainly the most illustrative case o this wide amily o two-phase gas solid dense downward lows. Many authors have experimentally studied these lows, underlying that the powders luidity is a key point in processes involving gas solid streams [1-3]. Due to the pressure build-up that compresses the interstitial gas, the luidised behaviour o the downward low cannot be maintained unless additional gas is injected through aeration taps positioned along the standpipe. However, the injected gas low rates can cause instabilities: either local gas-particle low deaeration when the injected gas low rates are too low, or appearing o stagnant gas voids when the injected gas low rates are too high [1]. These two phenomena are respectively described as the local deluidisation and the local over-aeration. Based on the analysis o the gas/solid slip velocity, deduced rom local pressure Corresponding author at: Laboratoire de Génie Chimique de Toulouse UMR 553, INP-ENSIACET, 4 Allée Emile Monso, BP 44362, 313 Toulouse, Cedex 4, France. Tel.: ; ax: address: edgar.tavaresdossantos@ensiacet.r (E.T. Santos). gradient measurements, Leung and Wilson [4] showed that standpipe lows can be described in dierent regimes such as moving bubbling luidised bed, a moving homogeneous luidised bed, or a moving packed bed that rubs against the internal wall o the standpipe. As a matter o act, the gas solid low can eventually switch rom one regime to the other rom the top to the bottom o the standpipe, depending on the solid mass lux, the distance between each aeration tap, and the aeration gas low rates [5, 6]. Thus, the behaviour o the moving gas-particle suspension under successive over- and under-aeration is a key parameter or the control and the stability o the low. The knowledge o the characteristic times o the deaeration, the voidage o the suspension, the bubble velocity and void raction in case o heterogeneous luidisation, is critical. However, because deluidisation and reluidisation are hard to investigate in the real condition standpipe low, the study in a static luidised bed is a straightorward manner to observe, describe, understand, and model the driving basic mechanisms. One o the most standard techniques is the single-drainage bed collapse technique, used to determine the voidage and gas velocity o the dense phase and bubble ractions in gas solid luidised beds o ine powders [7-14]. It consists o luidising a given powder, cutting-o the luidisation gas supply, and then measuring the collapsing bed height and the pressure drop generated by the exiting interstitial gas across the bed. It has been shown that the eect o the residual gas exiting rom the windbox through the bed cannot be neglected, and improvements o the bed collapse technique have been proposed. They consist either in modiying the equations or the voidage and gas velocity o the dense phase [1] or in perorming the dual-drainage bed collapse technique [15]. The previous studies ocused on the eects o the (i) physical properties o the powders (diameter, density, percentage o ines),

3 (ii) the bed height, (iii) the distributor (porous or perorated plate), (iv) windbox volume, (v) operating pressure, (vi) fluidising velocity beore defluidisation. It was shown that the dynamic deaeration behaviour (voidage o the emulsion phase, bubble raction, deaeration time) strongly depends on the initial state o the bed; i.e. homogeneously or heterogeneously fluidised. Moreover, deaeration occurs with two propagation ront lines: a downward at the surace o the bed corresponding to the exiting o gas phase, and an upward at the distributor relative to the compaction o the solid phase which was never studied systematically. This study ocuses on the deaeration behaviour o the FCC powder in a static fluidised bed under operating conditions representative o the continuous standpipe flows. The aim o this work is to describe, quantiy, and model, the eects o the driven parameters. Both singleand double-drainage techniques are used: the ormer models the standpipe flow when local deaeration occurs with only upwards interstitial gas leakage; the latter when it occurs with upwards and downwards interstitial gas leakage. o the dense phase velocity o a fluidised bed. However, one should consider the existence o other proposed methods to calculate ud rom the voidage/velocity curve [9] or using a Richardson Zaki equation type: ud = ut εnd, with ut the particle terminal velocity value and n an exponent characteristic to each powder [9] Gas behaviour during defluidisation Most bed collapse experiments have been carried out using a single drainage method in which residual gas in the bed and windbox is drained through the bed. In this case, the eect o the residual gas flow rom the windbox through the bed cannot be neglected in the bed collapse measurements by the single drainage method [15]. To assess this contribution, Tung and Kwauk [1] corrected the collapsing bed level velocity, uc, with the gas leakage velocity rom the windbox, uw, using the ollowing expression: uw =!W A " + ΔPd Hw Pa tc ð2þ 1.1. State o the art o the bed collapse technique results A typical defluidisation collapse curve is shown in Fig. 1. The fluidised bed collapse is composed o three zones: bubble escape (zone 1); dense phase sedimentation (zone 2) and consolidation (zone 3). In most reported cases, when the air is suddenly cut-o, the bed surace initially collapses as bubbles rise up to the bed surace. The bed height then decreases at a constant rate as the particles sedimentation takes place until the bed level stabilisation. Two parameters can be determined rom the defluidisation curve analyses (Fig. 1) [9]: the height o the dense phase, HD, o the fluidised bed is obtained by extrapolating to zero time the sedimentation slow settling rate. The voidage o the dense phase, εd, is determined by means o a mass balance: εd = 1 mp ρp A HD ð1þ where mp represents the mass o particles in the bed, ρp the particles density and A the cross-sectional area o the column; the dense phase gas velocity, ud, is given by the slope o the collapse curve, uc [13]. Tung and Kwauk [1], in analogy with the sedimentation theory, showed that the deaeration velocity is the same as that where W represents the weight o the powder in the bed, A the crosssectional area o the column, Hw the height o the windbox, ΔPd the pressure drop through the distributor, Pa the atmospheric pressure and tc the time instant at the end o zone 2 (Fig. 1). The double drainage method was also studied [15, 16]. Rowe et al. [17] reported that the exact values o the voidage and the gas velocity o the dense phase can be obtained only i the pressure o the windbox is reduced to zero as soon as possible as the fluidising gas is shut-o which was not the case o most previous works in the literature [15]. Park et al. [15] evaluated the eect o the residual gas volume in the windbox and proposed an optimum dual-drainage method Solid behaviour during defluidisation Visual observations o a 2D transparent fluidised moke-up allowed Geldart and Wong [13] to conclude the existence o a 2-ront solid displacement during defluidisation: one corresponding to bed surace collapse and another corresponding to solid upward sedimentation (Fig. 1). The upward sedimentation ront has not been the subject o many studies and inormation on this phenomenon is still missing. Regarding its velocity, the development o the Engelhard test provided the first data [18]. The test consists in measuring the required time or the pressure measured 3.48 cm above the distributor to start decreasing ater the air supply is cut-o. The test presents some limitations bubble escape H dense phase sedimentation consolidation HD LB slope: uc zone I ε1, u1 LInt zone II ε2 udis t tb t t t tb t t Fig. 1. Typical stages and analysis o Geldart-A FCC catalyst defluidisation curve: schematic o the 2-ront solid displacement during defluidisation [13].

4 Table 1 FCC catalyst physical properties. Powder Cracking catalyst vent 14 Particle mean diameter, d p (μm) 76 Apparent density, ρ p (kg/m 3 ) 14 Bulk density, ρ b (kg/m 3 ) 818 Tapped density, ρ tap (kg/m 3 ) 936 Repose angle ( ) recipient due to its intrusive character and the poor estimation o the sedimentation ront using only one point. Also, when the measured pressure is zero one part o the bed above the captor is already in a ixed bed state. To overcome this actor, Sharma et al. [18] have deined the deluidisation time (transition rom luidised to ixed bed), rom the pressure drop temporal data, as the irst observed slope changing. air compressor vent 1.4. Objectives Even though the bed collapse test has been the subject o numerous studies, some additional knowledge and characterisation can be obtained when analysing the temporal local pressure drop signal. This new inormation can be used to transpose this static phenomenon to the dynamics o gas solid dense downward lows. Moreover, the experimental results were successully modelled using the interpretation o Cherntongchai and Brandani [19] or the bed collapse experiment. The aim o this work is, thereore, to provide new relevant understanding o the deluidisation o Geldart-A particles, such as the FCC catalyst, to later transpose to the complexity o the downward gas solid dense phase lows. 2. Experimental 2.1. Materials Typical FCC particles were used. Physical properties are given in Table 1. The catalyst has a size distribution between 1 and 2 μm. The mean particle diameter is 76 μm and its density o 14 kg/m 3 thereore belonging to the group A o the Geldart classiication [2]. Fixed bed Homogeneous luidization Transition Heterogeneous luidization Fig. 3. Drawing o the experimental apparatus. Legend: 1 valve; 2 pressure regulator; 3 valve (single-drainage method); 4 rotameter; 5 manometer; 6 valve; 7 three-way valve (double-drainage method); 8 windbox; 9 distributor; 1 luidised bed; 11 local pressure transducers; 12 total pressure transducer; 13 recorder; 14 cyclone. Fluidisation properties (u m ; u mb ; ε m ; ε mb ) were determined experimentally in a classic luidised bed using dierent bed weights (7 kg; 14 kg and 21 kg). The minimum luidising velocity was determined using the Richardson method [21]. The luidisation regimes were identiied plotting the bed mean voidage against the gas velocity (Fig. 2). The typical regimes were identiied [9]: ixed bed: u bu m ; homogeneous luidisation: u m bu bu mb ; transition: u mb bu b1.5 u mb ; heterogeneous luidisation: u mb bu Experimental apparatus Bed collapse experiments were carried out in a cold-air transparent Plexigas column 192 mm in diameter and 2 m high. A cyclone is placed at the top to collect the entrained solids as shown in Fig. 3. Since the ratio o the column diameter to the mean particle diameter y = x ε (-) P dis (Pa) Uplow Downlow u m u mb u (mm/s) U (m/s) Fig. 2. FCC catalyst luidisation curve. Fig. 4. Experimental characterisation o the distributor.

5 Bed height, m U 1=.5 m/s, ε 1 = δ b exp t, s Fig. 5. Model validation with Cherntongchai and Brandani [19] results. ( ) this work. δ b exp is larger than 25, the possible wall eects can be neglected in the present experiments. The pressure drop across the porous distributor is ound to be linear relatively to the gas velocity and is o 1559 Pa when luidising at 1.2 cm/s (Fig. 4). The gas phase, air rom an oil-ree compressor, is ed to the bed through a pressure regulator, a calibrated rotameter that provides a luidising velocity, u, between 1 mm/s and 8 mm/s, a conical shape windbox (D w =19 cm; H w =19 cm) and a porous distributor. The relative humidity o the air is 3%. H (m) u - u mb (m/s) Fig. 7. Eect o luidising velocity, u u mb, on the raction o bubbles in the bed. Single drainage; double drainage. ( ) H =.3 m; ( ) H =.597 m; ( ) H =.885 m; ( ) H =.915 m. A ruler was placed on the wall o the column to measure the expanded and collapsing bed heights. Local and global pressure drop was measured using pressure transducers. The pressure data was acquired with a sampling requency,, o 1 Hz Procedure Initially a known weight o solids (7.2 kg; kg; kg and 21.5 kg) was loaded into the column providing our dierent static bed heights: 3 mm; 597 mm; 885 mm and 915 mm. The powder was then luidised by the compressed air at the desired supericial velocity (u mb bu b8 u mb ). When the system reached a steady-state (pressure luctuations irrespective o time) the air supply was suddenly cut-o. Single and double drainage protocols were then applied. To evaluate the single drainage method, the air supply was cut-o at the rotameter (Fig. 3) thus orcing the residual air to be purged through the bed o particles. Double drainage method was studied by cutting o the air at the.2 +2 % % H (m) δ b exp δ b calc Fig. 6. Evolution o the bed height with time during deluidisation, single (illed)/double u drainage method comparison. H =885 mm. (, ) u =47.4 mm/s, u mb = 7:29; u (, ) u =6.3 mm/s, u mb = :97. Fig. 8. Comparison between theoretical and experimental raction o bubbles in the bed. Single (illed)/double drainage. ( ) H =.3 m; ( ) H =.597 m; ( )H =.885 m; ( ) H =.915 m.

6 εd (-) entrance o the windbox through a three-way valve (Fig. 3) placing both column bottom and top at atmospheric pressure. The bed height, recorded with a digital camera at 25 ps, and pressure drop (=1 Hz) were acquired simultaneously during the deluidisation phenomena. 3. Modelling The experimental results were modelled based on the model proposed by Cherntonghai and Brandani [19]. The main assumptions and equations o the model detailed in [19] are presented hereunder: 1. One-dimensional system; 2. In the bed sections, the luid is incompressible; 3. Negligible inertial eects; 4. tn, u dis u m ; 5. The top luidised section remains at a constant void raction ε 1 corresponding to a gas velocity u 1 (Fig. 1); 6. The gas in the windbox is compressible and is considered ideal. From the solid and gas-phase mass balance the ollowing equations were deduced: dl B dt dl Int dt = u dis u 1 ð3þ = 1 ε 1 dl B ε 2 ε 1 dt where L B is the height o the collapsing bed, L Int the height o zone II, u dis the supericial velocity o the gas lowing through the distributor, ε 1 and ε 2 the porosities o zones I and II, respectively (Fig. 1). The problem is closed with an equation or the distributor velocity: dp w dt u mb 2 u mb = P w V w u dis A u (mm/s) Fig. 9. Eect o luidising velocity, u u mb, on the dense phase voidage. Single drainage; double drainage. ( ) H =.3 m; ( ) H =.597 m; ( ) H =.885 m; ( ) H =.915 m. where P w is the absolute pressure in the windbox, V w the windbox volume and A the cross-sectional area. The distributor velocity, u dis is calculated based on the pressure drop across the distributor, ΔP d (Fig. 4).The ΔP d is determined evaluating the momentum balance: ΔP d = ðp w P atm Þ ρ g L B ΔP 1 ΔP 2 ð6þ εd (-) ð4þ ð5þ where ΔP 1 corresponds to the pressure drop in zone I, luidparticle system at equilibrium, ΔP 2 the zone II ixed bed pressure drop unction o u dis, determined with experimental results or Ergun modiied law as presented in [19]. The bubble escape stage was introduced perorming a total mass balance over the control volume and assuming that the bubbling zone properties are similar to the stationary bubbling luidised bed, giving the ollowing equation: dl Tot dt = u dis u : ð7þ The developed model was tested using the parameters provided in [19]. Fig. 5 shows, as expected, that the present model predicts the same behaviour. To be noted that only the single drainage experiments were modelled since the discharge valve was not characterised. 4. Results and discussion 4.1. Bed collapse curve The collapsing behaviour previously reported by Geldart and Wong [13] was observed: typical single drainage bed collapse curves composed o bubble escape, particle sedimentation and bed consolidation periods were obtained when the initial gas velocities were above the minimum bubbling state (Fig. 6.a). When the luidising velocity was at the minimum bubbling state, only constant rate sedimentation and consolidation periods were ound (Fig. 6.b). Double drainage method showed a decrease o the total surace settling time,, or the two initial bed structures studied: homogeneous and heterogeneous. This results in a aster bed surace collapse P i / P P i / P Gas compressibility τ comp. τ 2 H τ P Fig. 1. Gas compressibility eect. Double drainage; single drainage. H =915 mm; u =6.3 mm/s (H-H )/H (H-H )/H

7 1 ε1=.462 ε2=.422 (H=.918m) u1=.32m/s.96 LB(m) rate, as well as a smaller bed height, thus evacuating more interstitial gas within the bed. Also, the bed surace collapse is rather parabolic than linear in the case o a homogeneous fluidised bed (Fig. 6) Fluidised bed initial condition eect.92.8 t4.6 LInt(m) t3 t2.4 t ε1= ε2=.418 (H=.911m) u1=.29m/s Heterogeneous fluidised bed From the zone 1 o the collapse curve (Fig. 6.a), the raction o bubbles in the bed was determined using the ollowing equation:.95 LB(m) Homogeneous fluidised bed Concerning the homogeneous fluidised bed, the abrupt collapse o the bed is not observed: the dense phase voidage, εd, is equal to the bed voidage, ε. Moreover, the voidage decreases at a constant rate with time consequently resulting in a airly constant collapsing bed surace velocity, uc, with a mean value o 3.5 mm/s using the single drainage method. This value is in agreement with previously reported values by Geldart and Radtke [1] and Geldart and Wong [13] or powders with similar physical properties. The use o the double drainage method resulted in a 3% increase o the collapsing velocity, uc, due to the escape o gas through the distributor..92 δb = Hi Htr Hi.8 ð8þ t3 1 LInt(m) Pi (Pa) t2 t Fig. 12. Model fit o single drainage experiments: u = 6.3 mm/s; u = 47.4 mm/s. Points: experimental; Lines: model Pi (Pa).4 where Hi represents the mean bed height beore defluidisation and Htr the transition height rom bubble escape to dense phase sedimentation. According to Fig. 7, increasing the fluidising velocity leads to an increase o the raction o bubbles in the bed. This was observed or all o the tested bed heights. Moreover, or a given fluidising velocity, increasing the bed height leads to a decrease o the bubble raction in the bed (Fig. 7). These results can be explained by the influence o the bed height on bubble diameter and velocity. Indeed, the increase o bed height will lead to an augmentation o bubble size and velocity which will cause a decrease in the raction o bubbles as shown by the ollowing expressions: 4 4 δb = u umb Ub ð9þ 3 1 where Ub is the bubble velocity given by 2 1 qffiffiffiffiffiffiffiffi Ub = u um + :711 gdb t1 t2 t3 13 ð1þ 15 t4 Fig. 11. Local pressure drop temporal evolution during defluidisation. Double drainage; single drainage. ΔP1: mm; ΔP2: mm; ΔP3: mm; ΔP4: mm. The mean bubble diameter, db, can be estimated, according to Mori and Wen [21], through the expression: % & H db = dm ðdm db Þ exp :3 i 2D ð11þ

8 ε 1 =.462 ε 2 =.422 (H =.918m) u 1 =.32m/s Table 2 Comparison o the solid sedimentation ront velocity. Drainage H / (mm/s) u s, (mm/s) Error (%) Single Double P w (Pa) Fig. 13. Evolution o the absolute pressure in the distributor during deluidisation or single drainage and u =6.3 mm/s: ( ) experimental; ( ) model. where D represents the column diameter, d m, the maximal theoretical bubble diameter, which is given by h i :4 d m = 1:64 A u u mb ð12þ Typical results were ound concerning the luidising velocity inluence on the dense phase voidage [9, 11, 13]. For instance, the results ound in this work are identical to those obtained by Barreto et al. [11] when studying the luidising velocity eect on the dense phase voidage by means o X-ray absorption and bed collapse technique or similar powders. Fig. 9 shows that the dense phase voidage: decreases rapidly or luidising velocities between u mb and 1.5 u mb ; decreases with the augmentation o the bed height. The bubble size dependence o gas luidising velocity and bed height explains these results. In act, the increase o the bubble size leads to a compression o the interstitial gas by the bubbles, leading to a less aerated and more compacted suspension. Empirical correlations are proposed or the two evacuation methods in study: single drainage: ε D = :38u ;35 H ;51 ;6 mm=s u 5 mm=s; 2 H = D 6 ð14þ and d b, the bubble initial diameter, by double drainage: d b = : :4: u u mb ð13þ ε D = :36u ;46 H ;32 ; 6 mm=s u 5 mm=s;2 H = D 6 ð15þ From the previous set o equations, the theoretical raction o bubbles was calculated. Fig. 8 shows that predictions are not accurate or experiments using single drainage method as well as or high luidising velocities. The discrepancies in the results come rom: the act that correlations proposed by Mori and Wen [22] are adapted to Geldart-B particles; the uncertainty in measuring the bed height in zone 1. Indeed, it has already been reported the diiculty in analysing images o heterogeneous luidised beds [9] Regarding zone 2, dense phase sedimentation, the collapsing velocity decreases with the augmentation o the luidising velocity and bed height. Again, the luidising velocity and the bed height are key parameters in the aeration state o the dense phase o a luidised bed. Moreover, the increase o luidisation velocity and bed height and the subsequent increase o bubble rising velocity leads to the ormation o a less aerated dense phase presenting higher resistance to the evacuation o the interstitial gas. From the experimental data the ollowing correlations were proposed: single drainage: u c = :12u ;18 H ;23 ; 6 mm=s u 5 mm=s;2 H = D 6 ð16þ (s) H (m) Fig. 14. Eect o the initial bed height, H, on the surace settling time,. u =6.3 mm/s. ( ) Single drainage; ( ) Double drainage. double drainage: u c = :14 u ;21 H ;19 ; 6 mm=s u 5 mm=s;2 H = D 6: ð17þ Notice that the exponent o the bed height parameter is similar to the one proposed by Abrahamsen and Geldart [9]: u c H.244 (18) Gas compressibility eect A new trend, never reported in the literature, is observed when plotting simultaneously the normalised total bed pressure drop and the normalised collapse bed level during the deluidisation by a single

9 drainage method which is not observed when using the double drainage method. Indeed, Fig. 1 shows the existence o a period during which the pressure drop across the bed is positive whilst the bed o particles is completely settled. Thereore, a gas stream passes through the bed during this period whilst deluidisation (sedimentation o solid particles) is complete. This can be explained by the expansion o the interstitial gas in the bed once the deluidisation is complete. In this case, the gas compression is a non-negligible phenomenon, despite the small dimensions o the bed. A new characteristic time, τ comp, is deined between the complete bed settling ( ) and the zero pressure drop (τ P ) times (Fig. 1). This time is related to the gas compressibility and characterises the decompression o the interstitial gas in the ixed bed. The slope changing in the global pressure drop curve corresponds to the complete settling o the bed o particles (Fig. 1). In addition, the gas velocity in the dense phase cannot be estimated by the slope o the collapsing bed level [7, 8, 13, 23] since the surace collapse curve is also aected by the gas compressibility during deluidisation Solid behaviour during deluidisation Local pressure drop measurements during the bed deluidisation showed that a ast deluidisation o the middle o the bed irst occurs when using a double-drainage protocol (Fig. 11.a). Also, negative local pressure drop values are observed in the bottom o the bed hence air escapes through the windbox. The gas escaping through the windbox eliminates the compressibility period observed when using a singledrainage method (Fig. 11.b). Local pressure drop data o single drainage experiments show a transition point (Fig. 11.b), as previously reported (Fig. 1). This transition point takes place when the bed settling is complete. Transposing this observation, the local pressure drop data underlines the existence o a ront o solids displaced rom the bottom to the top o the column reaching the surace when the bed settling is complete. This phenomenon, already described by Geldart and Wong [13] using a black-lighted two-dimensional bed deluidisation experiments, was never properly quantiied. The temporal analysis o local pressure drop can determine the progress o the sedimentation ront o the solid particles in the deluidisation. It was considered that the moment when the oscillations o the local pressure drop begin is associated with the complete settling o particles in the bed, as shown in Fig Modelling Fig. 12.a compares the experimental and simulated results or a homogeneous luidised bed using the single drainage method. The model satisactorily predicts the two characteristic ronts o deluidisation using the initial and inal voidage values or zones I (ε 1 ) and II (ε 2 ). The velocity in the luidised zone (u 1 ) is similar to the minimum luidising velocity (.3 m/s). Moreover, the pressure in the windbox is suitably described until the beginning o the compressible stage (Fig. 13). The heterogeneous luidised bed is also correctly described by the model (Fig. 12.b). The voidage o the ixed bed zone (ε 2 ) is calculated by a mass balance with the height o the settled bed and the values o u 1 and ε 1 correspond to the minimum luidising conditions. The results corroborate the previous indings that: the heterogeneous structure leads to a compression o the interstitial gas by the exiting o bubbles, leading to a less aerated and more compacted suspension. The local pressure drop analysis and the sedimentation ront issued rom the simulation results are in agreement. Thereore, the local pressure drop signal analysis may be a powerul tool or phenomena identiication in standpipe dysunction Bed height inluence on homogeneous luidised beds Fig. 14 shows that, in the presence o a homogeneous luidised bed, the surace collapse settling time increases linearly with the static bed height. This indicates the constant settling rate o the bed surace throughout deluidisation. Moreover, this linear evolution o with the static bed height is observed or the two drainage methods: single and double. Double drainage method removes the eect o the gas contained in the windbox. Then, i with the single drainage method the same linear evolution with bed height is observed, it can be concluded that the same phenomena are present. Thus, the inluence o the gas contained in the windbox can be neglected in the present deluidisation experiments. This is related to the reduction in the windbox volume to the luidised bed air volume ratio by using a conical shaped windbox, 48% with H =3 mm and 16% with H =915 mm when compared to previous works where it represented between 6% and 33% [15, 23]. Furthermore, the time required or the complete settling o the surace is equal to the time required or the sedimentation ront to go rom to H. Thus, the velocity o the ascending sedimentation ront can be calculated by: u s; = H (19). The results are presented in Table 2. The values determined by the two methods are concordant (temporal analysis o local pressure drop data and using Eq. (19)). Finally, the characteristic sedimentation rate o a Geldart-A particles homogeneous luidised bed can be calculated by simple laboratory experiments through the determination o the total surace collapse time or dierent static bed heights. Moreover, this velocity can be used to estimate the maximal solid lux in a non-aerated standpipe. In such lows the descending velocity o the particles should be higher than the solid sedimentation ascending ront. 5. Conclusion The aim o the present work was to provide new relevant inormation on the deluidisation o Geldart-A particles to later transpose to downward gas solid dense lows. Experimental observations showed the typical collapsing behaviour. The indings regarding the dense phase voidage o a luidised bed and the bed surace collapse velocity are in agreement with the two-phase theory and with previous authors reports or luidising velocities higher than the minimum bubbling velocity. Moreover, the raction o bubbles was calculated based on the collapse curve. Results are also in agreement with two phase theory and the use o ast camera image analysis to this end may lead to improvements on the determination o bubble properties. The existence o a gas compressibility period was described through the simultaneous analysis o both surace collapsing level and pressure drop measurements. Moreover, the deinition o the gas velocity in a dense phase luidised bed, reported to be given by the slope o the surace collapse curve, is not accurate. Thus, to characterise this parameter, a new undamental test must be ound or the validation o numerical simulation must be used. Through the temporal analysis o local pressure drop the progress o the solid sedimentation ront rom bottom to top was determined. The velocity o the ascending sedimentation solid ront is ten times superior to the collapse o the bed surace. The experimental results were itted using the model proposed by [19]. The bed height collapse is correctly described and the upward sedimentation ront is similar to the one determined through the local pressure drop analysis. The local pressure drop signal analysis may provide new inormation on standpipe lows dysunction phenomena identiication. List o symbols A cross-sectional area o column, m 2 d b bubble mean diameter, m

10 d b bubble initial diameter, m d m maximal theoretical bubble diameter, m d p particle mean diameter, μm D column diameter, m acquisition requency, Hz g gravity, m/s 2 H(t) collapsing bed height at time, t H static bed height, m H D dense phase bed height, m H i initial bed height, m H R air humidity, % H tr height at deluidised ree bubble bed, m H w windbox height, m L B height o the collapsing bed, m L Int height o zone II, m L Tot height o the bubbling luidised bed, m m p mass o particles, kg n exponent in Richardson Zaki equation, ( ) P a atmospheric pressure, Pa P w windbox absolute pressure, Pa t time, s t c required time to start consolidation, s U b bubble velocity, mm/s u 1 emulsion phase supericial velocity in zone I, mm/s u c collapsing bed surace velocity, mm/s u D dense phase gas velocity, mm/s u dis supericial velocity o gas lowing through the distributor, mm/s u luidising velocity, mm/s u mb minimum bubbling velocity, mm/s u m minimum luidising velocity, mm/s u s, homogeneous bed sedimentation solid ront velocity, mm/s V w windbox volume, m 3 W weight o powder in the bed, N Greek symbols δ raction o bubbles in the bed, ( ) ΔP pressure drop, Pa ΔPZ average pressure drop during 6 s, Pa ΔP 1 pressure drop in zone I, Pa ΔP 2 pressure drop in zone II, Pa ΔP d total pressure drop across the distributor, Pa ΔP i pressure drop at time instant i, Pa ε 1 voidage in zone I, ( ) ε 2 voidage ion zone II, ( ) ε(t) bed voidage at any time t, ( ) ε D dense phase voidage, ( ) bed voidage at minimum bubbling velocity, ( ) ε mb ε m bed voidage at minimum luidising velocity, ( ) ρ b bulk density, kg/m 3 ρ gas density, kg/m 3 ρ p particle density, kg/m 3 ρ tap tapped density, kg/m 3 τ comp compressibility time, s bed height settling time, s zero pressure drop time, s τ P Reerences [1] D. Geldart, A.L. Radtke, The eect o particle properties on the behaviour o equilibrium cracking catalyst in standpipe, Powder Technol. 47 (1986) [2] S. Bodin, C. Briens, M.A. Bergougnou, T. Patureaux, Standpipe low modeling, experimental validation and design recommendations, Powder Technol. 124 (22) [3] Y.M. Chen, Recent advances in FCC technology, Powder Technol. 163 (26) 2 8. [4] L.S. Leung, L.A. Wilson, Downlow o solids in standpipes, Powder Technol. 7 (1973) [5] E.T. Santos, R. Andreux, M. Hemati, G. Ferschneider, Inluenceoaerationtaps spacing in standpipe lows, IFSA Industrial Fluidization South Arica Proceedings, 28, pp [6] E.T. Santos, Etude expérimentale et numérique du soutirage des particules d'un lit luidisé. Application au cas industrial du FCC, PhD Thesis, Université de Toulouse, France, 21. [7] K. Rietema, Proc. Int. Symp. Fluidization, Eindhoven, 1967, 154. [8] A.A.H. Drinkenburg, K. Rietema, Gas transer rom bubbles in a luidized bed to the dense phase II. Experiments, Chem. Eng. Sci. 28 (1973) [9] A.R. Abrahamsen, D. Geldart, Behaviour o gas-luidized beds o ine powders Part II. Voidage o the dense phase in bubbling beds, Powder Technol. 26 (198) [1] Y. Tung, M. Kwauk, China-Japan Symposium on Fluidization 49 (1982) 155. [11] G.F. Barreto, J.G. Yates, P.N. Rowe, The measurement o emulsion phase voidage in gas luidized beds o ine powders, Chem. Eng. Sci. 38 (1983) [12] G.F. Barreto, J.G. Yates, P.N. Rowe, The eect o pressure on the low o gas in luidized beds o ine particles, Chem. Eng. Sci. 38 (1983) [13] D. Geldart, C.Y. Wong, Fluidization o powders showing degrees o cohesiveness II. Experiments on rates o de-aeration, Chem. Eng. Sci. 4 (1985) [14] G.F. Barreto, G.D. Mazza, J.G. Yates, The signiicance o bed collapse experiments in the characterization o luidized beds o ine powders, Chem. Eng. Sci. 43 (1988) [15] J.J. Park, J.H. Park, I.S. Chang, S.D. Kim, C.S. Choi, A new bed-collapsing technique or measuring the dense phase properties o gas-luidized beds, Powder Technol. 66 (1991) [16] P.G. Muritt, P.L. Bransby, De-aeration o powders in hoppers, Powder Technol. 27 (198) [17] P.N. Rowe, D.J. Chessmann, J. Yong, Y. Zhiquing, Proceedings o World Congress III Chemical Engineering (1986). [18] D.S. Sharma, T. Pugsley, R. Delatour, Three-dimensional CFD model o the deareation rate o FCC particles, AIChE Journal 52 (26) [19] P. Cherntongchai, S. Brandani, A model or the interpretation o the bed collapse experiment, Powder Technol. 151 (25) [2] D. Geldart, Types o gas luidisation, Powder Technol. 7 (1973) [21] J.F. Richardson, Fluidization, Academic Press, New York, [22] S. Mori, C.Y. Wen, Flow estimation o bubble diameter in gaseous luidised beds, AIChE Journal 21 (1975) [23] D. Geldart, C.Y. Wong, Fluidization o powders showing degrees o cohesiveness I. Bed expansion, Chem. Eng. Sci. 39 (1984)