The effect of riser scale on the hydrodynamics and particle residence-time distribution

Transcription

1 IFSA 2011, Industrial Fluidization South Africa: Edited by A. Luckos & P. den Hoed Johannesburg: Southern African Institute of Mining and Metallurgy, 2011 The effect of riser scale on the hydrodynaics and particle residence-tie distribution Z. Liu, M. Takafuji and T. Suda IHI Corporation, Japan Keywords: scale up, fluidized bed, riser, hydrodynaics, height of riser, pressure drop, inventory, residence tie distribution Abstract The effect of riser scale on the hydrodynaics of riser was nuerically studied in this paper. A coercial software package, BARRACUDA, was eployed to calculate the behaviour of the riser. First, experients have been done to evaluate the perforance of BARRACUDA. The coparison of the experient and calculation showed that the calculations by BARRACUDA were reasonable. Secondly, nuerical calculations of risers have been carried out for three scales of risers (riser heights, H, were 7.5, 15 and 30, and other geoetrical structures were ratio to the height). Particle circulation rate G s, superficial gas velocity U 0, and bed aterial were specified in this study. It was found that the distribution of volue fraction of particle was different for different scale riser. The velocity field of particle and gas were also changed when the scale was changed. It was found that the inventory was proportional to the square root of riser height. The pressure profile along altitude direction was alost the sae if the pressure was noralized by the weight of inventory. For low volue fractions of particle, it was found that the ean residence tie was proportional to H INTRODUCTION A gasifier of coal and bioass TIGAR has been developed by IHI Copany. This is a circulating fluidized bed with dual reactor, one is the riser cobustor and the second is bubbling bed gasifier (Figure 1) [1]. The solid fuel is thrown into the bubbling bed gasifier, where part of fuel is gasified. The reaining char goes to riser and is cobusted for the heat. The heated bed aterial goes to the gasifier to supply the heat for the endotheric gasification reaction. Previous studies showed scale effect on the hydrodynaic behaviour in the bubbling bed gasifier. [2] In this paper, we showed the effect of riser scale on the hydrodynaic behaviour of the riser using nuerical calculation. A large riser can be scaled down with siilarity, if a specified bed aterial and specified superficial velocity are selected. [3] Qi [4] proposed an epirical siilarity paraeter for fully developed zone of riser. There are two kinds of approach to study hydrodynaics of riser by nuerical calculation. One is Euler-Lagrange approach, [5] in which fluid is described by Euler approach and particles are described by Lagrange approach. Another is Euler-Euler approach, [6] which take particles as continuous phase and both of fluid and particles are described by the Euler approach. In this paper, we use BARRACUDA s ultiphase particle in a cell approach (Euler-Lagrange Hybrid) [7] to calculate the behaviour of hydrodynaics of riser. A paper in the proceedings of a conference on fluidization Noveber 2011, Johannesburg, South Africa

2 258 Figure 1. Diagra of TIGAR (Twin IHI Gasifier) Firstly, the results calculated by BARRACUDA were validated with experients. It is found that the results of BARRACUDA were reasonable. Secondly, BARRACUDA was applied to siulate the risers of different scale, (heights are 7., 15 and 30 ). After which, the hydrodynaics of riser with different scales were copared by ters of diensionless values. The distribution of fraction of particle, velocities of fluid and particles were studied in addition to the pressure drop, inventory and residence tie distribution of particle. The relationship between the riser scale and hydrodynaics of riser was explored. Experients VALIDATION OF NUMERICAL SIMULATION Experients have been done to validate the nuerical calculation. The experiental apparatuses are scheatically shown in Figure 1. The height of riser was 15 and inner diaeter of riser was The bed aterial is the silica sand. The diaeter distribution of sand is shown in Figure 2. Three test cases were listed Table 1. The pressures at various elevations were easured by pressure sensors. The particle circulation rate was easured by a sensor in the cyclone dipleg. The sensor easures the particle circulation rate by easuring the ipact of falling particles on a detection plate. The calibration has been ade for the sensor. The results of pressure drop are shown in Figure 3.

3 259 Figure 2. Diaeter distribution of bed aterial Table 1. Conditions of experients HeightofriserH Dia eterofriserd Bed aterial - Density ofbed aterial kg/ 3 unit Case 1 Case 2 case silica sand 2610 Bulk density ofbed aterial kg/ Flow rate ofpri ary airq p 3N/h Particle circulation rate Gs kg/(2 s) Nuerical calculations The sae conditions were set to the nuerical calculations with the experients, i.e., the sae geoetries, the sae velocities of air flow, the sae particle circulation rates. The calculations were transient. When the ass flow rate at the exit was unchanged and equal to that of inlet, the riser nuerical results were consider to be at steady state. Analysis was based on tieaveraged nuerical results over a period of 70 s. The calculating results of pressure drop are show in Figure 3. Figure 3. Coparisons of experients and calculations on the pressure drop

4 260 Coparisons of nuerical siulation and experient The results of nuerical calculation and experients were copared to validate the BARRACUDA. In the Figure 3, The coparisons were ade for the pressure drop of three cases of different particle circulation rate G s (G s = 26.6 kg/ 2 s, G s = 18.8kg/ 2 s and G s = 10.1 kg/ 2 s). Here, pressure drop is defined as P P 0, where P 0 is the pressure at exit of riser. It was found that the calculations of BARRACUDA were reasonable. THE EFFECT OF SCALE OF RISER ON THE HYDRODYNAMICS Conditions of calculation In order to investigate the effect of riser scale, three cases were studied. The conditions of three cases were siilar, because it was expected to find correlations with scale under siple condition. The calculation conditions are listed in Table 2. The geoetries of riser were in proportion to the height of riser. The particle circulation rate G s and superficial velocity U 0 were kept the sae. The bed aterial for the calculations was based on the properties of the silica sand. Table 2. The conditions for calculation unit CaseA CaseB CaseC Height ofriserh InnerDia eterofriserd H/D - Bed aterial - Density ofbed aterial kg/ 3 Velocity ofpriary airu0 /s Particle circulation rate Gs kg/(2 s) The following was the results deduced fro calculation. Distribution of volue fraction of particle 41.7 silica sand

5 261 D=0.18 D=0.72 D=0.36 D=0.18 [-] D=0.36 [-] H= 7.5 H= 15 H= 30 H=7.5 H=15 H=30 (a)the whole riser H= H=15 H=30 (b)the botto of riser Figure 4. The volue fraction of particle of different scale Figure 4(a) shows the distribution of εp the volue fraction of particle, for whole risers. It was found that the εp distributions were different, especially in the botto of riser. In order to observe clearly, the botto parts were enlarged in Figure 4(b). It was found that the particles concentrated on the inlet side for large scale (H = 30 ), but the particles concentrated on the opposite of the inlet for sall scale (H = 7.5 ). If the scales of riser changed, the distributions changed uch. So it was not expected to find siilarity even the riser was scaled up with siilar geoetry and siilar flow conditions. Velocity of particles The velocities of particles are shown in Figure 5. The vertical coponent of particle velocity Vz is shown in Figure 5(a). It was found that the upward velocity of particle in the core zone was larger for the larger riser. This was because that wall effect had little effect on the larger riser. Figure 5(b) shows the vector of particle velocity at botto. The ain flow patterns are arked with dash lines. It was found that the flow pattern was uch different for different scale. The large riser was ore affected by inlet, because the ass flow rate (kg/s) was large for large riser. We realized again that siilarity did not exist for these risers.

6 262 D=0.18 D=0.72 D=0.36 D=0.18 D= H=30 H=15 H=7.5 H=15 H=30 (Vertical velocity of particle of whole riser 2.88 VP [/s] VP,Z [/s] H=7.5 D=0.36 H=7.5 H=15 H=30 (b)vector of particle velocity at the botto of riser Figure 5. The velocity of particle of different scale Inventory and pressure profile We failed to find the exact siilarity for these risers, but we still intended to find soe correlations between riser scale and values concentrated in industry. Using these correlations, we expected to ake prediction approxiately for industrious design. In this section, we tried to find how to estiate the pressure drop. Industries pay a lot of concern on the pressure drop, when we design a riser and select the blower for it. It is iportant that we should predict the pressure drop when the scale of riser is changed. Is it correct that the pressure drop is caused by the weight of bed aterial? How to estiate the inventory weight of bed aterial in riser when the particle circulation rate is given? The pressure drop is related to the inventory for low loading riser where the agnitude of the particle acceleration is negligible. The inventory height H0is defined as the heap height that all the bed aterial in the riser is heaped in the botto with zero superficial velocity. In the Figure 6 shows the calculation results of inventory height H0. It was found that the H0/H was approxiately expressed as equation (1). Of course, the coefficient will be changed if the particle circulation rate Gs or other conditions are changed, but the fact will not change uch that inventory height H0 is in proportion to the square root of height of riser H. (1)

7 263 Figure 6. Inventory height H 0 Pressure [Pa] P/(gH 0 ) [-] Diensionless elevation Z/H [-] (a) Pressure Drop Profile Diensionless elevation Z/H [-] (b) Diensionless Pressure Drop Figure 7. Pressure drop Figure 7(a) shows the calculation results of pressure at each elevation for three riser configurations. Z is elevation of riser. Figure 7(b) shows the diensionless pressure, which is noralized by equation (2). Where ρ b is the bulk density of sand. The followings were found fro Figure 7: a) The three lines becae one line approxiately. It eant that the diensionless pressure drop profile did not change with the riser scale. b) The total diensionless pressure drop was close to 1in these cases, which eant that the pressure was ainly caused by the weight of bed aterial and the effect of acceleration of particle was negligible for these cases. Residence tie distribution (RTD) When the heat and ass transfer and reaction of particles are considered, it is very iportant to predict the residence tie of particle. Because Lagrange approach was used in this study, the entry tie and exit tie of particles could be counted, so the residence tie distribution (RTD) of particles could be gotten. Figure 8 shows the results of RTD. It was found that the residence tie is longer for higher riser and the peak of RTD curve is lower for higher riser. (2)

8 264 Figure 8. Residence tie distribution (RTD) There are two ways to estiate the ean residence tie in the riser [8]. One way is fro pressure drop across the riser, P Riser, by calculating the riser volue occupied by particles divided by the voluetric flow of particles in the riser. (3) where A is area of cross section of the riser. is the average volue fraction of particle. Since the acceleration pressure drop was neglected in this study, the pressure drop across the riser can be written as where g is gravity acceleration. Equation (3) can be rewritten as follows by substituting Equation (4) into it: (4) (5) Another way to estiate the ean tie is calculation of breakthrough tie of particle t_ th, by assuption of plug flow. = (6) where is superficial velocity and is terinal velocity of particle.

9 265 Figure 9. Mean residence tie (7) The ean residence tie calculated by BARRACUDA siulation t is shown in Figure 9. As references, t,p and t_ th are also shown in Figure 9. The correlations between the ean residence tie and riser height were expressed as equations (7). It was found that the ean residence tie was close to t,p, and t becae closer to t,p for larger scale riser, which eans that the t,p could be a good prediction of ean residence tie for large scale riser. It was found that t was uch larger than t _th, which eant that the particles in the riser did not break-through the riser straightly, but went upwards and downwards. The total path of particle was longer than the riser height. Referring Equation (7), ean residence tie t was proportional to H 0.706, instead of proportional to H. CONCLUSIONS The effects of riser scale on hydrodynaics were investigated in this paper. The followings were concluded: 1. Nuerical calculation of riser by software BARRACUDA was reasonable. 2. Even if the geoetry of riser was siilar and the super velocity and particle circulation rate were fixed, the distribution of volue fraction of particle, velocity field of particles and gas were different for different scale risers. The hydrodynaics was not of siilarity under these conditions. 3. Even though, soe correlations between the hydrodynaics and riser scale were found. The pressure drop profile was alost the sae if it was noralized by the weight of inventory. The inventory height was proportional to the square root of riser height. The ean residence tie was proportional to. Equation (5) was a good estiation of residence tie for large scale riser. REFERENCES 1. T. Suda, M. Takafuji, Y. Matsuzawa, T. Fujiori Gasification of lignite coal using dual circulating fluidized bed gasifier, International Syposiu on Gasification and Application (ISGA2008).

10 Z. Liu, M. Narukawa, M. Takafuji, T. Suda Prediction and validation of effect of bed length on RTD of coal in a bubbling fluidized bed, Fluidization XIII, pp M. Horio, M. Takada, M. Ishida, N. Tanaka The siilarity rule of fluidization and its application to solid ixing and circulation control, Fluidization, Engineering Foundation, pp X. Qi, J. Zhu, W. Huang Hydrodynaic siilarity in circulating fluidized bed risers, Cheical Engineering Science, 63: H. Zhou, G. Flaant, D. Gauthier, J. Lu Lagrangian approach for siulating the gas-particle flow structure in a circulating fluidized bed riser, International Journal of Multiphase Flow 28: R. Andreux, G. Petit, M. Heati, O. Sionin Hydrodynaic and solid residence tie distribution in a circulating fluidized bed: Experiental and 3d coputational study, Cheical Engineering and Processing 47: D.M. Sinder, P.J. O Rourke, M.J. Andrews Sedient flow in inclined vessels calculated using ultiphase particle-in cell odel for dense particle flow, Int. J. Multiphase Flow 24: A.T. Harrison, J.F. Davidson, R.B. Thorpe Particle residence tie distributions in circulating fluidized beds, Cheical Engineering Science 58: