PARTICLE MIXING IN GAS-SOLID BUBBLING BED K. SAVOLAINEN AND R. KARVINEN Department of Energy and Process Engineering Tampere University of Technology, Finland Kai.savolainen@tut.fi, Tel. +358 40 8490882 ABSTRACT The goal of the study is to increase the understanding of a mixing process in a bubbling bed. Mixing of solid particles in a 1360x420-mm room-temperature gas-solid bubbling fluidized bed was studied. Air at room temperature was used with plastic particles of 3.3 mm diameter and the particle density was equal to 1300 kg/m3. The minimum fluidization velocity of the particles is 0.85 m/s and the actual fluidization velocity used in the experiments was from 1 to 2.5 m/s. White and black particles were used and the particle concentration was measured in the bed by taking samples from the bed at different times. White and black particles in each sample were separated, counted and weighted. The diffusivity coefficient of particles in the bed was obtained from an analytical equation, which governs the concentration as a function of time. Keywords: Particle mixing, Gas-solid fluidization, Bubbling bed, Cold model 1. INTRODUCTION Mixing of particles is an important factor in fluidized bed combustors. The horizontal mixing of fuel particles affects the performance of the beds. The mixing has a great influence on the distribution of heat release from the fuel. In a bubbling fluidized bed the particle movement is caused by rising bubbles. Bubbles transport particles upwards in their wakes and particles flow downwards in the places, where there are no bubbles. An important factor in the operation of a fluidized bed combustor is how the fuel particles are mixed with the bed particles. In bubbling beds the particle concentration is high, which makes the horizontal mixing of coal or any fuel difficult. According to the literature the horizontal convective mixing of particles occurs mainly at the bed surface, where bubbles burst (Mostoufi et al. 2001). In the regions, where velocity gradients are high, also the diffusive mixing contribute to the process. The vertical mixing is many times larger than the lateral one, and contribute more effectively to the bed mixing (Kunii & Levenspiel, 1977). 2. REVIEW OF MIXING STUDIES 2.1. Radioactive tracking technique Mostoufi et al. (2000) studied the diffusivity of solid particles in a 152-mm gas-solid fluidized bed with 38-μm sand and 70-μm FCC particles. The column was made of a plexi glass-pipe with 1500 mm height. The gas was air at room temperature and at atmospheric pressure. Superficial gas velocities were from 0.5 to 2.8 m/s for sand, and from 0.44 to 0.9 m/s for FCC, respectively. The initial height of the bed was 0.22 m for all the experiments. Movement of a tracer was observed by radioactive particle tracking technique. The tracers were made of a mixture of gold powder and epoxy resin. Tracer sizes of 420, 500 and 600μm were used. They were being activated in a nuclear reactor to each experiment. The produced isotope 198Au, emitting γ-rays, can be found by detectors. In each experiment, a single tracer was placed into the bed to move freely among the bed particles. The movement of the tracer was monitored for 5 hours, during which about 820 000 points were acquired. Stein et al. (2000) observed particle motion in a 3-D bubbling fluidized bed at atmospheric pressure. Positron emission particle tracking (PEPT) was used to observe particle trajectory and solids velocity. Two acrylic columns were employed: a small one with a 70mm diameter and the height of 430 mm, and a large one with a 141 mm diameter and the height of 600 mm. Distributor plates were made of a metal containing 80-130 holes with 0.5-1.55 mm diameter. During the experiment, the column was placed between two γ-ray detectors. Resin beads with 0.55-0.75 mm in diameter and a density of 1100 kg/m3 were used. The minimum fluidization velocity of these particles is 0.11 m/s. The tracers used were resin beads (660-μm) selected from the bulk. An ion exchange technique was employed in order to produce 18F (half life=110 min.), and tracer particles were activated by ion exchange with the radioactive water produced. 2.2. Thermal tracking technique Glicksman et al. (2008) used a thermal tracking method in order to study mixing charasteristics in a ¼-scale model (0.85x2.1 m) of a pressurized bubbling fluidized bed combustor. The model was at room temperature and pressure. Particles cooled by liquid nitrogen, were injected into the bed in the same way as pulverized coal is injected. An array of thermistors was mounted inside the bed. Thermistors were used to trace the path of the cooled particles when they are mixed with the other bed particles. It was assumed that heat transfer to the air was negligible during the short time period detected (10 s.). After a short period of time the bed returns to its initial temperature and another test can be performed. Multiple runs at the same test conditions were made. For each thermistor, the temperature results at each time were averaged. 2.3. Wall effect Liu et al. (2007) studied the wall effect in micro fluidized beds (MFB). They measured the minimum fluidization and minimum bubbling velocities of silica sand particles in air-blown micro beds. Three different particle sizes were used: 96.4, 242.1 and 460.6-μm. The particle density is 2600 kg/m3 in any case. Beds of 120 mm high and with inner diameters of less than 32 mm were studied. Static bed heights varied from 20 to 50 mm. The fluidization characteristics in MFBs should be different from those in the ordinary-size fluidized beds due to strong wall effect. Experiments were carried out in three cylindrical quarts-glass columns with a height of 120mm and inner diameters of 12, 20 and 32 mm, respectively. A sintered plate of 5 mm in thickness made from 150-μm silica sand particles was used as a gas distributor. 2.4. Phosphorescent tracking technique Pallarès and Johnsson (2006) applied a phosphorescent particle tracer to study the mixing of fuel particles in fluidized beds. The technique is based on capturing the phosphorescence of a single particle by a video recording with a digital image analysis in a 2-dimensional fluidized bed with a transparent front wall. The experimental rig used is a cold riser with a cross section of 0.02x0.4 m and a height of 2.15 m, having a perforated plate as an air distributor. The front side of the riser is made of transparent Perspex. Since the solids mixing in the bottom region was the focus of the work, only the first 0.85 m of the riser was video recorded. A recording time of 20 min per run was chosen, because it was noticed, that recordings longer than 15 min provided no additional information. Glass beads with the size and the density similar to those of sand particles typically used in fluidized bed boilers were used as a bed material: dp=330 μm and ρp=2600 kg/m3. Particles belong to the group B in the Geldard classification and have a minimum fluidization velocity umf=0.12 m/s and a terminal velocity of ut=1.76 m/s under ambient conditions. Several fluidization regimes were included in the runs: bubbling, turbulent, circulating and pneumatic. The height of the dense bottom bed varied from 0 to 0.69 m. Tracer particles used were cylindrical capsules made of transparent plastic filled with a self-phosphorescent solution. The reference particle had a large size and a lower density simulating a fuel particle in a fluidized bed boiler. In order to study the influence of the size and the density, two types of tracer particles were used: one type smaller in size and the other with a higher density similar to that of the bed solids. The method of evaluation is based on the tracking a single particle and it was assumed that interactions between fuel particles do not influence their mixing process. 2.5. Possibilities of numerical modeling Numerical modeling of the fluid flow has increased rapidly as a result of computer development. The movement of a single bubble in a bed has been solved though it requires plenty of computational capacity and there are problems to fix boundary conditions and to take into account the interaction between particles. The modeling of very small scale units can be performed, but the numerical calculation of a real bubbling bed is not possible in the near future (Huang, 2007). 3. MIXING EXPERIMENTS IN PLASTIC PARTICLE MODEL The experimental arrangement is shown in Figure 1. The test rig consists of an air blower (not shown in the figure), an air chamber, a grate and a reactor with transparent walls. Air at room temperature was blown into the air chamber (a) below the grate (b). A thick sieve was placed above the grate in order to prevent particles falling down into the air chamber. Air volume flow rate was measured with an orifice. The air flows into the reactor through the grate with 30 holes with diameters equal to 102 mm. The grate geometry, i.e., the number and size of grate holes, was chosen according to test results without the grate showing the structure of the free bubbling bed. In a bubbling bed, in which the air was blown through the sieve, the locations of bubbles are chancing arbitrarily. When a grate with fixed holes was used, bubble locations were permanent. In the bed, plastic white and black particles with diameters equal to 3.32 and 3.86 mm were used. Particle densities were 1300 and 950 kg/m3, respectively. The bed height was 125 mm. The minimum fluidization velocity of the bed particles in the study is umf=0.85 m/s and the actual fluidization velocity used in the experiments varied from 1 to 2.5 m/s. Particle properties in the study are shown in Table 1. Table 1. Particle properties. Particle color Particle diameter Ø [mm] Particle volume [mm3] Particle density [kgm-3] Particle bulk density [kgm-3] Particle mass [g] White 3,32 19.2 1300 900 0.025 Black 3.86 30.1 950 630 0.029 The study is based on the collection of particles from the bed at different times. Particles with two different colors were used. White particles were used as the bed particles and black particles were tracer particles. At the beginning of an experiment black particles were located in the center of the reactor (Figure 1c). The width of the black particle region was 100 mm. When the air flow from the blower was fed into the reactor by opening the valve of the blower, the mixing started. After a fixed period of time, the valve was closed and the Reactor Bubbling bed 420 Thick sieve hole Ø 102 mm 10 columns 3 rows a bubble 145 Air tube from blower 1360 mm Air chamber Grate hole size and distribution (a part of grate) (a) (b) y x 125 Black particles White particles White and black particles mixed (c) (d) Sample gap Pipe (plastic) y z Particle distribution (y-z-plane) (e) Piston shaft (aluminum) Handle (f) Figure 1. Test rig: bubbling bed (a), grate hole distribution (b), experiment initial condition (c), mixed condition of particles (d), vertical and horizontal mixing condition (e), sample probe (f). bubbling was finished. Particle samples consisting of a number of black and white particles were taken from the bed at different locations with a sample probe, which is shown in Figure 1f. The white and black particles in each sample were separated, counted and weighted. The volume fraction of black tracer particles was calculated. The concentration profile of black particles in the bed is governed by a type of equation similar to the heat conduction equation, namely ∂C ∂ 2C =a 2 ∂t ∂x (1) in which C is the particle concentration and a the diffusivity. The solution of the problem in Figure 1c can be found by using the Laplace transform technique (see Appendix 1) 2 ⎞ ⎛ x C = C0 e (βx + aβ t )erfc⎜⎜ + β at ⎟⎟ ⎠ ⎝ 2 at (2) β is the coefficient depending on the thickness of the black zone in Figure 1(c). The diffusivity coefficient a can be found from Eq. (2), when the concentration C is known on the basis of measurements in a fixed location and at the known instant of time. 4. RESULTS Particle samples taken from one run with the sample probe are shown in Figure 2. The samples were taken from the bed at the location y=60 mm and z=0. The vertical and horizontal distribution of particles is shown in Figure 1(e). The Figure shows that the mixing in the y direction and in the z-direction is homogenous, because the tracer particles are uniformly distributed. Samples were taken at five different x-locations, namely x=0, 150, 250, 350 and 450 mm. The result of one sample is shown in Table 2. Fluidization velocity used was equal to 1.7 m/s. The diffusion coefficient appears to be larger when t=10 s and x=250 mm. The bubble dimension in the bed is db= 100 mm. It is obvious that the first bubbles, when bubbling begins, throw black particles from the centerline of the bed. These particles drop on the bed surface at this location. When the bubbling continues, the coefficient approaches the same value, which is valid everywhere in the bed. t2= 21 s t1= 10 s X=0 mm 150 250 350 450 Figure 2. Particle samples taken from bubbling bed with probe. Table 2. Typical particle sample. Sample probing time t [s] uf=1.7 ms-1 Sample probing location x [mm] Sample mass m [g] Black particle volume fraction [m3/m3] Diffusion coefficient a [m2/s] 10 10 10 10 10 0 150 250 350 450 47.5 53.6 65.1 76.9 69.2 0.216 0.119 0.0810 0.0206 0.0049 0.00172 0.00183 0.00293 0.00153 0.00156 21 21 21 21 21 0 150 250 350 450 34.6 68.2 53.3 69.0 67.1 0.156 0.109 0.0966 0.0552 0.0285 0.00168 0.00189 0.00212 0.00194 0.00176 5. CONCLUSIONS The particle mixing in a bubbling fluidized bed cold-model was measured by taking tracer particle samples from the bed. In the measurements concentration changes of tracer particles were recorded in fixed locations. The experimental arrangement was such that one-dimensional assumption can be made. The mixing is governed by the equation similar to the heat conduction, and its solution can be found analytically. The effective diffusion coefficient of tracer particles can be obtained by solving the inverse diffusion problem. It was observed that the mixing in a horizontal direction is very uniform when the tracer and bed particles have almost the same size and density. In that case the effective diffusion coefficient of lateral mixing is also constant. References GLICKSMAN, L., CARR, E., NOYMEN, P. (2008). Particle injection and mixing experiments in a onequarter scale model bubbling fluidized bed. Powder Technology 180 (2008) 284-288. HUANG, C., WANG, Y., WEI, F. (2007). Solids mixing behavior in a nano-agglomerate fluidized bed. Powder Technology 179 (2007) 229-236. HUANG, J. (2007). Analysis of industrial granular flow applications by using advanced collision models. Ph. D. Thesis, Lappeenranta University of Technology. KUNII, D. & LEVENSPIEL, O. (1977). Fluidisation engineering. New York: Robert E. Krieger. LIU, X., XU, G., GAO, S. (2007). Micro fluidized beds: Wall effect and operability. Chem. Eng. Journal (2007). Article in press; received 16 November 2006; accepted 24 April 2007. MOSTOUFI, N., CHAOUKI, J. (2001). Local solid mixing in gas-solid fluidized beds. Powder Technology 114 (2001) 23-31. PALLARÈS, D., JOHNSSON, F. (2006). A novel technique for particle tracking in cold 2-dimensional fluidized beds simulating fuel dispersion. Chem. Eng. Science 61 (2006) 2710-2720. SCHLICHTHAERLE, P., WERTHER, J. (2001). Solids mixing in the bottom zone of a circulating fluidized bed. Powder Technology 120 (2001) 21-33. STEIN, M., DING, Y. L., SEVILLE, J. P. K., PARKER, D. J. (2000). Solids motion in bubbling gas fluidized beds. Chem. Eng. Science 55 (2000) 5291-5300. TAUZ, H. (1971). Wärmeleitung und temperaturausgleich. Verlag Chemie GmbH Weinheim. WEI, X., SHENG, H., TIAN, W. (2006). Charasterizing dispersion by image analysis in ICFB. Int. J. of Heat and Mass Transfer 49 (2006) 3338-3342. WINAYA, I., N., S., SHIMIZU, T., YAMADA, D. (2007). A new method to evaluate horizontal solid dispersion in a bubbling fluidized bed. Powder Technology 178 (2007) 173-178. APPENDIX 1 Mixing in a semi-infinite domain is shown in a figure below. Initial black particle concentration in the white domain is C(x, t=0)=0. Black particles ∞ C(0,t) C(x, t) 0 White particles C(∞,t)=0 ∞ x δ The governing equation of mixing is ∂C ∂ 2C =D 2 ∂t ∂x (1) Diffusivity, boundary and initial conditions are δ ∂C (0, t ) ∂C (0, t ) =D ∂t ∂x (2) C(∞, t) = 0, C(0, t=∞) = 0 By using Laplace transformation: L[f] = d 2C s − C=0 dx 2 k ∫ ∞ 0 fe − st dt = f , equation (1) and (2) give (3) sδ C (0, s ) − δ = D d C (0, s ) dx (4) The solution of which is C= 1 e − qx , sδ + Dq q= s D (5) The inverse transformation of (5) is 2 ⎛ x ⎞ C = C 0 e (βx + Dβ t )erfc⎜⎜ + β Dt ⎟⎟ ⎝ 2 Dt ⎠ (6)