UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Cyclone scale-up and radial gas concentration profiles Engman, Randy W. 1990-12-31

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata


UBC_1991_A7 E53.pdf [ 6.33MB ]
JSON: 1.0058775.json
JSON-LD: 1.0058775+ld.json
RDF/XML (Pretty): 1.0058775.xml
RDF/JSON: 1.0058775+rdf.json
Turtle: 1.0058775+rdf-turtle.txt
N-Triples: 1.0058775+rdf-ntriples.txt
Original Record: 1.0058775 +original-record.json
Full Text

Full Text

CYCLONE SCALE-UP AND RADIAL GAS CONCENTRATION PROFI BY RANDY W. ENGMAN B.A.Sc., The University of Calgary, 1986 A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF APPLIED SCIENCE in THE FACULTY OF GRADUATE STUDIES . Department of Chemical Engineering We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA September 1990 ©Randy W. Engman, 1990 \ In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department The University of British Columbia Vancouver, Canada DE-6 (2/88) ABSTRACT A two part study was undertaken to explain the performance of cyclones operated in circulating fluidized bed combustion (CFBC) systems. In the first part, collection efficiency tests were performed on a one-ninth scale polyacrylic cyclone model of the industrial scale cyclone at the 22 MWe CFBC facility at Chatham, New Brunswick. Emphasis was placed on scale-up .considerations, loading effects, inlet geometry effects, and flow visualization trials. Experiments were performed at room temperature with inlet velocities between 3.7 and '5.5 m/s, solids loading between 0.05 and 7.5 mass solids/mass air with two different solids systems. There was disappointing agreement between the results from the Chatham unit, scaled according to Stokes Number scaling, and the findings obtained from the cold model unit. There was a minimum in the particle collection efficiency for particles of diameter 2.5 to 3.0 um, apparently associated with agglomeration effects in the cyclone. Particle collection efficiency was found to increase with increased particle loading for the conditions studied. Changes in the inlet geometry gave inconclusive results. The experimental results were limited by problems associated with feeding and recycling the fines solids system used. — i i i — In the second part radial gas concentration profiles of a secondary cyclone serving the UBC pilot scale Circulating Fluidized Bed Combustor were performed at temperatures of about 870 <>C. Concentrations of O2 , CO2 , NOx , CH4 , CO and SO2 were measured. An increase in [CO], and to a lesser extent [CO2], was measured near the cyclone wall. There appeared to be little radial variation in the concentration of other species. Further work is required to allow the cold model to operate continuously, with particles which can be fed more freely, and to obtain radial gas concentration profiles within the primary cyclone of the UBC CFBC system. -i v-TABLE OF CONTENTS Page ABSTRACT i i LIST OF TABLES vi LIST OF FIGURES viACKNOWLEDGMENTS xINTRODUCTION 1 PART I COLLECTION EFFICIENCY TESTS 4 1.1 BACKGROUND AND THEORY 5 1.1.1 Introduction1.1.2 Dimensional Analysis and Physical Similarity 6 1.1.3 Fluid - Particle Separation Cyclonic Separators 10 1.1.4 Previous Scaling Work 15 1.1.5 Loading Effect on Cyclone Collection Efficiency 22 1.1.6 Summary 31.2 EXPERIMENTAL APPARATUS AND PROCEDURE 33 1.2.1 Introduction 31.2.2 Model Cyclone Apparatus 33 1.2.3 Particulate Solids 45 1.2.4 Data Acquisition and Analysis 45 1.2.5 Error Sensitivity 50 1.2.6 Chatham Cyclone Data 3 1.3 RESULTS and DISCUSSION 59 1.3.1 Scaling Consideration 64 1.3.2 Loading Effect 71.3.3 Inlet Modifications 77 1.3.4 Flow Visualization 83 1.4 CONCLUSIONS AND RECOMMENDATIONS 85 - v -PART II HOT CYCLONE TESTS 87 2.1 INTRODUCTION 8 2.2 THEORY 82.3 APPARATUS AND DATA ACQUISITION 95 2.4 RESULTS AND DISCUSSION 101 2.5 CONCLUSIONS AND RECOMMENDATIONS 110 Nomenclature 112 References 6 Appendix 120 -v i -List of Tables Page Table 1.1 List of important parameters 7 Table 1.2 Chatham operating conditions for 58 April 17, 1990. References as indicated. Table 1.3 Experimental data. Note that cyclone 62 configuration information can be found in Figure 1.10. (V.F. = vortex finder posi t ion). Table 1.4 Cyclone Operating conditions. 70 Table 1.5 Particle loading data 74 Table 1.6 Inlet modification tests 80 Table 2.1 Analytical instrument description 98 Table 2.2 UBC CFBC operating conditions 101 -vi i-LIST OF FIGURES FIGURE 1.1a FIGURE 1.1b Collection Efficiency vs. (NRBP)(NsT)0.5 Conditions: D = 50 mm, temperatures between 20 and 693 °C, pressures between 140 and 2500 kPa, inlet velocities between 0.18 and 5.2 m/s, dust loadings between 0.04 and 9.56 g/m3• (11) Collection efficiency curves with "fish hook" shape for primary and secondary cyclones. Conditions: D = 1.2 m, temperatures between 640 and 900°C, inlet velocities between 16.3 and 27.4 m/s, dust loadings between 1.5 and 140 g/m3.(12) page 18 19 FIGURE 1.2 a, Gas and particle flow across imaginary cylinder. (21) FIGURE 1.2 b. Experimental grade efficiency as a function of (S)°*8 ( dimensionless particle diameter ) for industrial sized cyclones in air at room temperature (21). 21 23 FIGURE 1.3 FIGURE 1.4 Figure 1.5 Loading effect on cyclone collection efficiency. 25 Cyclone diameter up to 3.65 m diameter. Eo is zero load (less than 1 grain/ft3) loading, EL is higher loading efficiency. (18) Predicted separation efficiency of fine 26 particles swept out 23 of the gas by large particles due to agglomeration (24). Conditions as stated. Leith and Licht correlations with and 31 without loading and saltation corrections. Conditions: D = 0.05 and 0.91 m, temperatures between 20 and 850 °C, inlet velocities between 1.8 and 46 m/s, dust loadings from 0.43 and 4450 g/m3.(23) -vi i i-FIGURE 1.6 Figure 1.7 Figure 1.8 Figure 1.9 Figure 1.10 UBC apparatus schematic through B38. used in runs Bl Figure 1.11 Figure 1.12 FIGURE 1.13 Figure 1.14 Figure 1.15 Figure 1.16 Figure 1.17a Figure 1.17b Figure 1.17c Figure 1.18 Plot of blower speed vs flow/pressure drop. (Pumps and Power Inc.). Diagram of UBC solids hopper used in runs Bl through B35 Diagram of UBC model reactor top and model cyclone. Entrance geometry configurations for inlet modification tests. Shaded areas show inserts. All dimensions in mm. See section 1.3.3 for details. Size distribution of feed solids. ( 22 um size ) Photograph of FCC test solids ( 22 micron diameter ). Collection efficiency error estimates for equations 1-29, 1-30, 1-31, and 1-32. Schematic of Chatham CFB Boiler (27) Photograph of fines from Chatham fluid bed heat exchanger, sampled on April 17, 1990. Catch hopper mass vs time indicating steady state solids feed rate. Run B4. Catch particle size distribution for run B4 Particle size analysis by Elzone analysis machine. Catch particle size distribution for run BIO Particle size analysis by image analyzer. Collection efficiency for runs BIO and B4. Run B4 particle size distributions determined by Elzone particle analysis instrument. Run BIO particle size distribution by image analysis methods. Conditions as stated in Table 1.3. a.) Particle size distributions from Chatham fluid bed heat exchangers sampled April 17, 1990. b.) Collection efficiency curve derived from particle size distributions. E = (Clt/(Clt + Lit)) 34 36 37 39 40 46 47 52 54 56 61 63 68 - i x-Figure 1.19 Collection efficiency curves for run B4, the Chatham cyclone and the Chatham cyclone shifted according to Stokes Law scaling. Scaling conditions as per Table 1.4. 69 Figure 1.20 Figure 1.21 Figure 1.22 UBC and Chatham data plotted accordim to Abrahamson and Allen correlations. a. ) UBC data (run B4 conditions) b. ) Chatham data. See Table 1.4 operating conditions. for 71 UBC Run B4 and Chatham data compared with Parker et al. data. Conditions as stated in Table 1.4 Loading effect on collection efficiency ( UBC data ). Vi = 5.0 m/s, T = 21 <>C, P = 1 atm. 72 75 Figure 1.23 Figure 2.1 Figure 2.2 FIGURE 2.3 FIGURE 2.4 FIGURE 2.5 Comparison of collection efficiencies for different inlet configurations. See Table 1.3 for conditions. ( solids loading corrected ). 81 Predicted particle trajectories in a vertical plane within a Stairmand type Low loading conditions. (25) Mean particle trajectories for particles of diameter 1 to 10 microns. Mean particle trajectories, 3 micron. Random particle trajectory of 2 microns particle in turbulent flow, 90 cyclone, a. b. c. Predicted gas flow patterns in a Stairmand type cyclone. Low loading conditions (25). Predicted combined axial and radial velocity vector diagram in a Stairmand type cyclone. Low loading conditions(25) UBC CFBC schematic (29). Scale drawing of secondary cyclone of UBC CFBC system. 92 93 94 97 FIGURE 2.6 Gas sampling system serving UBC CFBC system. 99 -x-FIGURE 2. 7 Gas concentration profiles for run 17. 103 FIGURE 2. 8 Gas concentration prof i1es for run 18. 104 FIGURE 2. 9 Gas concentrat ion prof i1es for run 5. 105 Figure 2. 10 Gas concentrat ion prof i1es for run 6. 106 Figure 2. 11 Gas concentrat ion profiles for run 10. 107 FIGURE Al Figure A2 Figure A3 Figure A4 Figure A5 Figure A6 Figure A7 APPENDIX FIGURES 120 Shake down test summary. All runs performed 121 before run Bl. Pressure drop vs air flow for orifice plate. 122 Data logging program serving UBC model cyclone 123 apparatus. Schematic of attempted recycle system schematic 124 showing a high solids loading feed and measurement vessels, multiclone and bag filter arrangements. Particle size distributions for runs B4 and B10 125 Run B4 particle size analysis by Elzone analysis instrument. Run B10 particle size analysis by image analysis methods. UBC CFBC solid fuel analysis for run B17 (29). 126 Mass balance, as performed on a per channel basis 127 for run B10. Image analysis particle size distributions. Fines loss (below 15 microns) attributed to filter inefficiency. Figure Figure A8 A9 Temperature data for Part I experiments. Data summary for Parts I and II. 128 129 -x i -ACKNOWLEDGMENTS I would like to acknowledge the immense help of my advisors Dr. John R. Grace, Dr. C. J. Lim, and Dr. Clive M.H. Brereton. Their time, energy, and personal commitment was central to the completion of this work. As well the staff of the Department of Chemical Engineering, The Pulp and Paper Center and other employees of the University of British Columbia provided colorful support throughout the progress of this work. The financial support of Energy, Mines and Resources, and The New Brunswick Electric Power Commission is greatly acknowledged. Finally I would like to thank my parents without whom this thesis would not have been completed. i " 1 " INTRODUCTION Cyclonic separators, generally simply called cyclones, have been used in gas cleaning operations for well over a century largely because they offer good particle collection efficiencies under extreme and varying conditions and are simple to design, fabricate, and operate. A good overview of various designs, applications can be found in reference 33. Although they have been replaced by more efficient devices in many pollution control applications, cyclones have been the subject of renewed interest for high temperature solids-gas separation in combined power cycles and within Circulating Fluidized Bed Combustor (CFBC) applications. It was because of particle collection efficiency problems encountered in one such application (1), the Chatham Circulating Fluidized Bed Demonstration Project, a 22 MWe CFBC electric generation facility in New Brunswick, that the present study was undertaken. In spite of the simplicity of cyclonic separators, it is still not possible to predict from fundamental principles particle collection efficiencies for all geometries and operating conditions. A great deal of work has been done to develop equations and models to predict cyclone performance for certain standard geometries ( Eg. Stairmand and Lappel designs) under conditions of low solids loading and low temperature. However, little work has been done to validate these equations and models for extreme conditions of temperature and particle loading. Also, little work has been done to verify the models - 2 -for cyclones of large industrial scale. Thus there is a need to establish scaling criteria to predict the operation of high temperature, high loading, large cyclones from the operation of lab scale models. This thesis is divided into two parts. Part A deals with collection efficiency tests and considerations of scale-up. Part B considers measured combustion gas concentration profiles within a high temperature cyclone serving a pilot scale CFBC system and was intended to provide data for further research in the CFB field. The objectives of the two parts are as follows: Part I. Collection efficiency and scale-up studies. Firstly, to examine scale-up considerations by comparing the performance of a cold model cyclone to a large industrial high temperature cyclone. Secondly, to demonstrate the effect of solids loading on solids capture efficiency by means of laboratory experiments on a cold model cyclone. Thirdly to examine the effect of inlet modifications on capture efficiency. Lastly to perform visualization of particle flows within a cyclone. Part II. Gas concentration profiles within a secondary cyclone of a CFBC. To measure and report combustion gas concentration profiles within a secondary cyclone of a pilot scale CFBC facility. - 3 -For each of parts A and B, a discussion of previous related work and theory is followed by a brief description of the apparatus and experimental procedure. Experimental findings are then presented and discussed and conclusions are drawn. - 4 -PART I COLLECTION EFFICIENCY TESTS i 1.1 BACKGROUND AND THEORY 1.1.1 Introduction Cyclones are examples of inertial separating devices. The particle collection efficiency, that is the mass ratio of particles caught to those fed within a given size range(6), is often expressed by a collection or grade efficiency curve. Ideally the designer can predict cyclone performance, and thus the collection efficiency curve for all sizes of particles from first principles, or lacking that, from accurate empirical relationships. Unfortunately this is not always possible, as the gas and particle behaviour is not well enough understood to predict particle trajectories under all circumstances. Nor is it always possible to rely on empirical relationships, as they are design-specific and offer good accuracy only when standard designs are considered. See reference 6 for a comparison of published data and empirical relationships. Thus there is a need for cold modeling and for valid scaling criteria (13). The work described in this thesis is of benefit for the consideration of large, high temperature and non-standard cyclones with tangential inlets. - 6 -1.1.2 DIMENSIONAL ANALYSIS AND PHYSICAL SIMILARITY. Applying the principles of dimensional analysis to the problem of cyclone particle collection performance requires a complete list of the physical quantities controlling the fate of particles within a cyclone. These parameters can be divided into four groups: (i) those describing the cyclone itself, (ii) the operating parameters, (iii) properties of the particles being separated, and (iv) properties of the gas which carries the solids. Table 1.1 lists the most important variables. - 7 -Table 1.1 List of impor Cyclone dimensions: Body diameter Inlet depth Inlet width Outlet diameter Outlet length Cylinder length Overall height Bottom diameter Operating parameters: Inlet velocity Vi Gas split ratio Qc Loading Ratio LP Relative Ae/A* acceleration . (centrifugal/gravitational) -Particulate characteristics: Particle diameter dp Particle density fp Shape factor (T> Gas characteristics: Gas density j>% Gas viscosity Uc tant parameters D a b Do Lo Lc j Ho Db -8-The seventeen variables listed above do not provide a complete list, as other variables may be important. For example surface roughness may play a role and the smooth polyacrylic surface of the scale model is not similar to the refractory lining of the high temperature Chatham cyclone. This is neglected. If, however, the dimensional analysis is limited to one geometric configuration or cyclone design, only one of the first eight need be considered since geometric similarity applies between the cold model and the Chatham cyclone being modeled. With ten individual variables and three fundamental dimensions (mass, length, time), there are seven independent dimensionless groups requiring separate consideration. The analysis can be simplified further if one assumes that the .split ratio, i.e. the fraction of gas leaving by way of the underflow, is negligible and that there is similarity of shape factor for the solids systems considered. In addition if the relative acceleration is large in both cases ( i.e. Ac/Ae > 10) this factor can be neglected as well. With these simplifications, four dimensionless groups are required. The most Common independent groupings of the remaining variables listed above are: Flow Reynolds number NREf = PeDVi/uK (1-1) Density ratio Np = PP/p« (1-2) Stokes number N8t = dp2ViPp/(Dult) (1-3) Loading ratio Lp mass solids flow mass gas flow (1-4) In addition to maintaining geometric similarity it is desirable to have kinematic and dynamic similarity. Kinematic similarity is similarity in motion, which implies that the paths that representative particles follow are geometrically similar and are travelled in a consistent, scaled period of time (20). Thus particle accelerations must also be similar (20). Dynamic similarity involves similarity of forces. In order that the two systems under comparison be dynamically similar, the magnitude of forces at each point must also be similar (20). It is commonly impossible to satisfy all requirements simultaneously. Thus ones of lesser importance are often sacrificed in order to assure similarity of others. In cyclones operating at high flow Reynolds numbers (i.e. high inlet velocities), centrifugal, as opposed to gravitational, are thought to dominate the motion of smaller particles. In situations where the viscous and inertial forces are most significant the flow Reynolds number may be used to compare experimental observations, provided the conditions of geometric similarity are met (20). However, it would be incorrect to assume that, after the onset of turbulent flow in cyclones, collection efficiency is independent of Reynolds number. In fact the contrary was found, i.e. collection efficiency is found to vary with the helical turbulent intensity (34). Thus the Stokes number, combined with geometric and Reynolds number similarity, may be used in the scaling process. - 10 -1.1.3 FLUID - PARTICLE SEPARATION CYCLONIC SEPARATORS Before reviewing previous works on the subject of scale up of cyclone performance, it is useful to develop an equation describing particle capture, in order to demonstrate the relevance of the key parameters. This simplistic description makes some questionable assumptions and is only intended to introduce the effects of various parameters on collection ef f i c i ency. Consider the fate of a particle of diameter dcrit that is caught with 50% efficiency in a cyclone of diameter D with an tangential entrance way having width b and height a. We make the following assumptions (5): 1. Particles move independently of one another. 2. The drag on the particle can be described by Stokes law regime expression. . 3. Buoyancy and gravity effects are negligible. 4. The tangential velocity of the particle is constant and equal to the inlet velocity. 5. Secondary effects such as re-entrainment from the walls, eddy currents, etc. are negligible. 6. Laminar flow conditions. 7. Relaxation time is negligible. 8. Particles separate at constant velocity 9. Once at the wall particles have negligible chance of reentrainment. - 11 -10. Particles must reach the wall by moving across a gas stream, which retains its shape after entering the cyclone. Upon entering a cyclone particles are acted on by a centrifugal force equal to Fc = mVi2/R where: m = particle mass Vi= tangential particle velocity R = radial coordinate of the particle. Fc = ttdorit3pPVi2/6R centrifugal force (1-5) This force accelerates the particle towards the wall. Opposing this motion is the drag force: FD = (Co AP ps Vp2)/2 (1-6) where Vp is the particle's radial velocity component. The drag coefficient can be obtained by neglecting inertial terms in the Navier - Stokes equation for a rigid sphere in an unbounded fluid (5) i.e. CD = drag coefficient = 24/NRep = 24ps / (pg Vp dcrit) (1-7) * In this simplified explanation, the particle must travel a radial distance b during a residence time defined by the time needed to travel a circular path of distance nDN in order to be - 12 -collected. If not it will enter a zone near the bottom of the cyclone where the radial gas velocity is much higher than in the separation zone ( See Figure 2.3 ). The particle motion is thus: IT D N • Substituting for Co, AP = rr*dc r i t2 /4, the drag force is seen to be: FD = 3j»K dc r i t Vi b/DN Stokes law resistance (1-8) where: b = entrance width N = Number of revolutions Equating expressions 1-5 and 1-8 allows prediction of particle cut diameter as follows: Fc = FD (l-8a) HtUri t3f>PVi 2/6R = 3^8 dc r i t Vi b/DN (l-8bdent2 = 18u,tbR/(tIJ>pDNVi ) (l-8c) letting R = D/2 - b/2 such that b = D - 2R der it2 = 18uK R(D - 2R)/(WJ>PDNVi ) (l-8d) dent2 = 18ue R(l - 2R/D)/(KpPNVi ) (l-8eor der it = 3f 2u,R ( 1 - 2R/D )]o.s (i_9) [ ttpP Vi N ] i - 13 -This equation has generally the right form as cyclone experiments have shown (6) that collection efficiency improves with : 1. Increasing particle density. 2. Increasing cyclone vortex speed (i.e. Vi ) 3. Decreasing cyclone diameter. However as one would expect, such simplified descriptions have limitations and do not describe performance for all conditions. The effect that each parameter has been found to have on collection efficiency will now be considered. Particle loading is considered separately in section 1.1.4. Inlet velocity Collection efficiency has been found experimentally to increase with gas inlet velocity in several studies - 14 -(7,8,9). Increasing inlet velocity increases tangential velocity thereby improving separation. These improvements are not without limit, however, as secondary effects such as particle re-entrainment and eddy currents work to offset collection efficiency at higher Vi . The length of the gas vortex which exists below the vortex finder is a function of inlet velocity and cyclone geometry . The vortex end is experimentally defined as the position where the vortex core seeks the wall (16). Increasing Vi lengthens the vortex until the end interferes with the wall causing re-entrainment of the particles already caught, thus causing a reduction in collection efficiency.,The observed peak in curves of efficiency vs Vi have been successfully delayed with the addition of an inverted cone, located axially above the solids exit (14, 15). The inverted cone is thought to anchor the vortex end and prevent it from reentraining particles at the wall. See drawing on Table 1.1 for an example of this cone. Particle density As stated above the centrifugal force is proportional to particle mass and hence to particle density. Theoretically collection efficiency is proportional to pp0*3. However, experimentally this is not confirmed (10). Gas Viscosity and Density. - 15 -Particle motion is resisted by the viscous drag. For low particle Reynolds numbers, ( 10"5 < p8dpVP/uK < 0.1 ) ( where VP = particle velocity relative to the gas), viscous effects dominate and Stokes law can be used to predict individual particle behaviour in a gas. This is of particular importance in the discussion of scaling the performance of large cyclones at high temperature by small cold models. Particles separated in hot cyclones do so at lower particle Reynolds numbers (considering similar size, density and velocity) than those in cold models because of increased gas viscosity and reduced gas dens i ty. 1.1.4. PREVIOUS CYCLONE SCALING WORK One of the first works on the subject of scaling cyclone performance was presented by Stairmand (2) in 1951. The following method for predicting performance of geometrically similar cyclones operating under varying conditions was proposed. To find the size of the dust caught with the same efficiency as the test dust, multiply the test dust size by: /density of the test dust (1-10) / density of the new dust /test flow (1-11) J new flow 7 new viscosity (1-12) test viscosity /diameter of the new model (1-13) \l diameter of the test model -16-These relations combined from the Stokes number Nst and imply that collection efficiency is a function of N8t only for a given geometry. Stairmand was careful to limit this scaling procedure to density differences within the ranges 1000 - 4000 kg/m3. For cyclone diameter, the method is not recommended "without some experimental confirmation". Indeed mention is made of less than expected performance with cyclones of diameter greater than 1.22 m diameter.(2) Other attempts have been made to correlate capture efficiency with dimensionless groups and thus provide a scaling basis. Parker et al. (11) performed experiments on a 50 mm diameter cyclone at a relatively low inlet velocities ( 5 m/s) at temperatures up to 700 °C, and pressures up to 25 atm. Data were plotted on a log log plot of dpa vs (Nst)(NRE)°•3» the terms were defined as follows: dp. = dp(C,pP)o.5 (g/cm3)o.s (1-14) Nst = C'dPK* pp VI/(9U*DH) (1-15) NRE = Pz D Vi/ u, (1-16) where C = Cunningham correction factor 'PP = particle density g/cm3 DH = 2ab/(a+b) (1-17) Note that the particle* diameter in the Stokes number doso refers to the mass median diameter of the feed aerosol used in his study and that DH is the hydraulic diameter of the cyclone inlet . For the small cyclones under consideration the - 17 -resultant plot was nearly linear. See Figure 1.1a. Also noted in this study was a minimum collection efficiency phenomenon^ occurring generally between 2 and 4 microns. This was attributed to the breaking up of agglomerated particles during sample preparation for particle analysis. By examining the performance of two 1.2 m dia. "Stairmand high efficiency" cyclones used in series within a pressurized fluidized bed combustor, Wheeldon et al.( 12) were able to plot grade efficiency curves for both the primary and secondary cyclones. The tests were done under conditions of high loading ( 50 g/m3), temperatures of 640 to 910 °C and pressures of 6 and 12 bar. The authors compared their data, obtained in a relatively large cyclone, to the Stairmand data (2) obtained in small ( 31.5 mm dia.) cyclones. The comparison of Stokes numbers for particles caught with 50% efficiency is: FBC primary N„t5o = 4 to 10 x 10"5 FBC secondary Nat5o = 8.6 to 19.6 x 10~5 Stairmand N8t5o = 9.4 x 10-5 The author attributed the deviations to particle loading effects and concluded that standard expressions are sufficient to describe cyclone performance under elevated pressures and temperatures. Also observed in this study was a "fish hook" shaped collection efficiency curves which shows improved collection for smaller particles. (See Figure 1.1b). This effect was attributed to particle agglomeration within the cyclones of smaller particles on to much larger ones. Analysis 90% COLLECTION DIAMETER, ^jmA - 8T -PAATICIC SIZC, akront. FIGURE 1.1b Collection efficiency curves with iish nook" shape for primary and secondary cyclones(12). Conditions: D - 1.2 m, temperatures between 640 and 900°C, inlet velocities between 16.3 and 27.4 m/s, dust loadings between 1.5 and 140 g/m3. - 20 -procedures of the particles caused some disassociation and suggested higher collection efficiencies for the fines. This effect was also noted by Parker et al. In an attempt to explain performance from a wide variety of cyclone designs operated under industrial conditions ( 160 < D < 1600 mm, 20 < temperature < 950 °C ) Abrahamson and Allen (21) plotted capture efficiency vs a dimensionless particle diameter, S°•3, defined as : S° • 5 = [Vrp/Vr(C ]o.5 (1-18) where Vrp = Radial particle velocity Vr* = Radial gas velocity These last two quantities are evaluated at a radial position R = Do/2 just below the vortex finder. (See Figure 1.2 q) VTK is assumed to be constant over an imaginary cylinder, of radius R», extending from the lip of the vortex finder to a point on the cone having a diameter equal to that of the vortex finder. Vrp is calculated for the experimental data using the Abraham drag coefficient expression (16) which accounts for non - Stokesian fluid-particle behavior up to a particle Reynolds number of 6000 and gives Vrp without repetitive calculations. - Zl -- 22 -The procedure followed was: 1. Calculate G, a • = [p*Ux2dp3pp ]/(R,Ua2 ) (1-19) where Ux = tangential gas velocity at R«, assumed to be constant and equal to the inlet velocity. 2. Calculate N mot i on): p (Reynolds number for radial particle Nrorp = 20.5[(Ga°•5/9.61 + 1)0•3 - 1]2 (1-20) 3. Calculate Vrg from NRBTP. Particles which have an equal chance of being caught or lost • ( i.e. dp 5o ) should on average move towards the gas exit as often as to the wall and thus the ratio Vrp/VrK should have a value equal to one. The plot shows better agreement at the dp 5o position than at other points along the grade efficiency curve ( see Figure 1.2 »b), a factor the authors attribute to differences in cyclone geometry which give different particle re-entrainment characteristics. The analysis is said to be limited to low concentrations of dust ( less than 10 g/m3), and to return flow designs with rectangular slot entries. 1.1.5 Loading effects on cyclone collection efficiency Improved particle collection efficiency for cyclones operating at high dust loadings has been noted by several workers. (2,3,6,7,11,13,). Perhaps the most comprehensive description of this effect is detailed in an efficiency vs. loading plot published by the American Petroleum Institute - 23 -"j I I I I " • 1 400 • 2 300 3 1600 O 3 31S a L 160 >< f \ S 160 w 5S Ed o 6 280 u 7 370 fa Es. Ed - « 8 1S0 V 6 800 COLLECT! • T 9 190 COLLECT! - T 11 203 4 O y °£ * ^/V" cyclone 0. <^ a« no. flMn ^ a4 * 18 270 - \ * / • 13 630 \ * <* • * "15 152. « o  23 305 + o * * . OfOps: l-* 4 V a 22 49S 4. * • > I , r I t • I I I • I I I I I I I I 1 10 vr FIGURE 1.2 b. Experimental grade efficiency as a function of (S)0** ( dimensionless particle diameter ) for industrial sized cyclones in air at room temperature (21). - 24 -(API)(17), reproduced in Figure 1.3 (13). A clear increase in capture efficiency with loading is noted. The influence of loading is seen to be greater for cyclones operated at lower " zero load Eo" efficiencies than for those with very efficient Eo. This effect is attributed to a scrubbing effect of large particles on finer particulate and has been accounted for in several ways. An excellent discussion of the mechanisms considered to be responsible is given by Mothes and Loffler (24). They point out that in spite of the observed reduction in particulate tangential velocity ( and thus separating force ) collection efficiency increases with increasing loading. They attribute the effect to agglomeration mechanisms. Figure 1.4 plots, the expected collection efficiency for small particles (1 to 4 jim) as separated by larger particles ( 15 p.m ). Briefly the "scrubbing effect" is a process whereby smaller particles are forced towards the wall because of the motion of larger particles which are separating out at a higher velocities. Either through direct impact with the larger particle (and the formation of a larger separating mass) or by entrainment in the flow field behind the larger particle, the smaller particle moves toward the cyclone wall with an increased velocity. LOADING, GRAINS OF SOLIDS / CU. FT. OF GAS FIGURE 1.3 Loading effect on cyclone collection efficiency. Cyclone diameter up to 3.65 m diameter. E0 is zero load (less than 1 grain/ft3) loading, EL is higher loading efficiency. (18) particle size x/pm FIGURE 1.4 Predicted separation efficiency of fine particles swept out of the gas by large particles due agglomeration <24). T«(x> — predicted colection efficiency r« — cyclone radius, ss restitution coefficient. ri vortex finder radius b_ inlet width h. inlet depth V„ = inlet velocity Xo = mass median particle diameter — particle density Cn solids loading g/m3 - 27 -An empirical expression proposed by Ogawa (22) for the effect of dust loading in conventional cyclones has the form (modified to give efficiency in %): bi = 0.032 dimensionless ki = -.0157 dimensionless Co = dust loading ( g/m3) for a 90 mm diameter cyclone operated with flyash with between 14 and 16 m/s. The negative value for ki accounts for a decrease in efficiency as Co increases noted in experimental data of this author, contrary to the effect noted by other workers. For cyclones with an axial inlet geometry, collection efficiency was found to increase with loading. Another empirical relationship was proposed by the API(17) in 1955 of the form: Here the subscript 'o' refers to low loading conditions arbitrarily taken as 1 gr/ft3 ( 2.3 g/m3). Another correlation from the API (18) was of the form: efficiency(^) = 100 - bi[exp(-ki Co)1100 (1-21) where efficiency = 100- ( 100- e0)[c«/cft]°•2 (1-22) P(E) = P(E0) + A log L (1-23) - 28 -where A is the slope of the nearly linear curves relating P(e) and L and was empirically fitted by a polynomial as follows: A = 0.67 - 2.11*E0 + 5.63*Eo2 - 4.00*Eo3 (l-24a) and L the dust loading in expressed grains/ft3. P(e) and P(E0) are the probabilities associated with the collection efficiencies at high and low loading respectively, i.e. the fraction passing the cyclone at a given loading. These probabilities can be approximated by the log of [(l-x)/x] as: (Catch probability) = [1 - pass probability] = [1 - P(x)] = [ 1 - log[(l-x)/x] ] such that [ 1 - log[(l-E)/E] ] = [ 1 - log[(l-Eo)/E0] ] + Alog( L ) log[(l-E)/E] = log[(1-Eo)/Eo] - log LA = log[ (1-Eo )/EoL* ] [(1-E)/E] = [(l-Eo)/EoLA] [1/E - 1] = [(1-Eo )/EoL* ] E = Eo LA / [ Eo LA + 1 - Eo ] such that ln(l-E) = ln[(l-E0)/(l + E0[LA-1])] (l-24b) (l-24c) (l-24d) (l-24e) (l-24f) (l-24g) (l-24h) (1-241) (l-25a) - 29 -Mansin and Koch(23) developed a correlation that accounted for the change in efficiency by modifying the effective viscosity of the gas as: Ut.pp = jiK[l + 0.091ogL + .02(logL)2]. (l-25b) (note log base 10 in equation l-25b) The model took the form: ln[ (1-E) (1+Eo [LA - 1])°'5 ] = \ -2.3SFACTOR[NST(n+1)AiG/(LFAC2DC2) ] < o . 4 l / ( n • l ) ) (1-26) where: E = collection efficiency Eo = low loading efficiency as determined by: ln(l - E) = 2[[NsT(n+l)/(Ai/2Dc2)G]<°-s/<n+i>> (1-27) L = L loading gr/f3 SFACTOR = [ (Vi /V8 )/2.5]° • *i for V-i /V8 < 2.5 SFACTOR = [(Vi/Vs)/2.5]-o•51 otherwise Vs = particle saltation velocity Nst = Stokes number Ai = Inlet area G = cyclone configuration parameter. LFAC = 1 + 0.091ogioL + .02( logio)2 Do = Cyclone diameter n = Vortex exponent This is actually the Leith modifications for particle and Licht sal tat ion (12) model with arid loading effects. 30 -Figure 1.5 compares the Leith and Licht equation with and without these modifications. - 3x -Fraction passing - data a.) Leith/Licht model without corrections. Legend o Ernst » Knowiton 0 Ogawa - Parker 4 Exxon ? Curt Fraction passing - data b.) Leith/Licht model with loading and saltation corrections. Figure 1.5 Leith and Licht correlations with and without loading and saltation corrections. Conditions: D = 0.05 and 0.91 m, temperatures between 20 and 850 °C, inlet velocities between 1.8 and 46 m/s, dust loadings from 0.43 and 4450 g/m3. (2$) - 32 -1.1.6. Summary When describing fluid particle separation processes in cyclones it has been found necessary to include other parameters besides the Stokes number. Particle loading and secondary effects such as particle re-entrainment play an important role and should be taken into consideration. An increase in collection efficiency with increased particle loading has been reported by several workers. - 33 -1.2 Apparatus and Experimental Procedure 1.2.1 Introduction Several experimental arrangements were tested in the course of the experimental program before the set-up was finalized. Results from these preliminary "shake down" experiments are not reported in the body of this thesis but do appear in Figure Al of the appendix and are discussed in section 1.3.1. What follows is a description of the equipment, particulate solids used, and the procedure for data acquisition and analysis. 1.2.2 Model Cyclone Apparatus Figure 1.6 shows a schematic of the apparatus used for the majority of tests performed in Part I of this thesis project. The apparatus was comprised of four separate systems: 1. Blower, piping, orifice flow metering system 2. Solids feed system 3. Polyacrylic cyclone 4. Filter system. Each is described below. For all experiments, compressed air was provided by a positive displacement Rootestype blower, powered by a 37 kW ( 50 hp) diesel engine. The air flow could be varied over a wide range ( 0 - 1100 scfm or 0 - 0.52 m3/s) by controlling the SEE APPENDIX FOR DETAILED DIMENSIONS BYPASS DAMPER RDDTS TYPE BLDWER SOLIDS FEED HOPPER ORIFICE PLATE & WATER MANOMETER PLEXIGLASS MODEL CYCLONE SIMULATED REACTOR TOP 0.61 m \\\ ,\ \\\\ .•AD CELLS IT 150 MM DIA PLASTIC v PIPE \ CATCH HOPPER FILTER BAG. .•AD SCALE FIGURE 1.6 UBC apparatus schematic used in runs Bl through B38. - 35 -engine speed and a bypass damper. The bypass damper was shut for all experiments to ensure constant flow. The flow vs. blower speed curves are shown in Figure 1.7. A 150 mm (nominal) dia. plastic pipe connected the blower to the feed hopper while in between, located after more than 4 m of straight pipe run, was the flow measurement orifice plate. This brass plate, designed and installed according to ASME standards ( 28 ), had a 101.5 mm (4 in.) i.d. orifice. The orifice pressure drop was measured by means of a water manometer, and because the reading was usually unstable, due to fluctuations in pressure in the simulated reactor top, the best accuracy attainable was +\- 5 mm. The absolute pressure at the orifice plate was observed to fluctuate with solids loading and was in the neighborhood of 30 to 60 cm (12 to 24 inches) water column pressure. The flow vs. orifice plate pressure drop curve appears in Appendix I ( Figure A2 ) . Solids to be fed to the system were contained in a 3.05 m tall by 0.36 m square hopper equipped with a manually operated 0.254 m dia. (aperture opening) cone valve (see Figure 1.8.). Located at the bottom of the hopper was a 0.36 m square windbox, constructed with a 30% free area distributor plate and lined with a 3 mm thick layer of commercial grade bleached kraft paper (softwood) to ensure even distribution of air. Additional fluidization air was provided by a 6 mm diameter pipe entering opposite to the solids valve. Fluidization air flow rates were measured by means of a rotameter. During the runs the hopper was pressurized, ( by means of a 32 mm dia. 36 - . SUTORBILf 7MF SERIES r BLOWER Typical ffti-foniiance - Pressure Mode Standard InlGt Conditions 900 1000 J ion i::fiO 1300 1400 ' 1500 OPERATING SPEED - RPM 1600 1700 REV. 1 7-15-85 1600 Mi-/ Fic Figure 1.7 Plot of blower speed vs flow/pressure drop. ( Pumps and Power Inc.) 3.05-k- 0,36 -H SDLIDS FEED VALVE WIND BOX <SEE DETAIL) TRANSPARENT PLEXIGLASS FACE SUPPLEMENTARY FLUIDIZATION AIR: CONE VALVE HAND CRANK 150 MM DIA SOLIDS DICHARGE PIPE WINHBDX DETAIL 0.35 DISTRIBUTER1 PLATE < 30 7. FREE AREA ) 1 50 MM T Figure 1.8 Diagram of UBC solids hopper used in runs Bl through B35. - 38 -pipe connected upstream of the orifice plate), to a pressure slightly above that of the transport air to allow steadier and more rapidsolids feeding. The hopper was supported by three load cells ( 1000 kg capacity each, 9 mV output at rated capacity) which allowed mass flow rate to be determined for the duration of the experimental runs. Hopper mass information was logged on an XT computer at 1 s intervals. Error is estimated to be +/- 1 kg, largely due to friction between the hopper and the pneumatic transport pipe. A flexible connection was used at this junction to minimize this friction. The air transport pipe turned twice between the solids input point and the cyclone, once through 45°, then again through 90°. In order to model the Chatham cyclone more effectively the polyacrylic cyclone was preceded by a polyacrylic entrance section, and a steel enclosure geometrically similar to the top of a circulating fluidized bed riser section. This included a slanting riser ceiling piece, inclined at 10° towards the cyclone as shown in Figure 1.9. A 50 mm wide rubber strip sealed this reactor top to the cyclone entrance way and was installed taut, thereby minimizing flow interference. The polyacrylic entrance-section was built as a scale model of the original configuration of the entrance-way. It was 445 mm high by 140 mm wide at the entrance point to the cyclone. Within the entrance-way were installed wooden inserts, designed to model proposed geometric changes being considered for the Chatham cyclone. Details of the entrance geometries studied are given in Figure 1.10. FIGURE 1.9 DIAGRAM DF MDDEL REACTDR TOP AND MDDEL CYCLONE < ALL DIMENSIONS IN MM ) 150 r 395 JL_ 60 M .ajsa:*:'»:ftsa*v-l CONFIGURATION CI BASE CASE SEE FIGURE 1.1 FOR DIMENSIONS CONFIGURATION C2 30 i CONFIGURATION C3 CONFIGURATION C3 WITH LOVER ED VORTEX FINUER Fzgure 1.10 Entrance geometry configurations for inlet modification tests. Shaded areas show inserts All dimensxons in mm. See section 1.3.3 for detaiis - 41 -The model cyclone itself was a one-ninth (0.11) scale replica of the Chatham CFBC cyclone existing at Chatham New Brunswick. The model cyclone was constructed from 6 mm thick clear polyacrylic and dimensions were held to a tolerance of +/- 3mm. The scale was decided upon by assuming that the cyclone Stokes number was roughly similar i.e. NsT UBC = NST CHATHAM Operating conditions are presented in the results section, in Table 1.4. It can be immediately noticed that the Chatham cyclone design chosen is non-standard. Compared to standard designs it is squat, having a height/diameter ratio of Ho/D = 2.8, while standard designs have Ho/D = 3.7 to 4.0. In addition the vortex finder or gas exit duct does not extend below the floor of the entrance way, which raises the possibility of gas short circuiting the separation zone. Extending the vortex finder below the entrance floor by D/10 is not uncommon in standard designs (5). Also peculiar to this design is the inclusion of an annular zone, concentric with the cyclone itself but positioned above the entrance ceiling but below the cyclone ceiling. It is believed that this squat design was chosen to meet dimensional requirements, but it unclear why the concentric annular region near the top was included. The model provided for two types of vortex finders, the first being made of steel with provision for an additional section, while the second made of polyacrylic, could be retracted from the cyclone body, thereby simulating other - 42 -configurations. Gas exited the cyclone to a short 300 mm dia. steel duct which then turned 90°, traversed approximately 600 mm and finally turned down 90° to a bag filter. For runs Bl to B10, performed under high loading conditions it was necessary to use a large sock shaped bag filter to handle the higher solids flows. This bag filter had the dimensions of 0.78 m dia. by 4 m long and was made of 100% cotton, 452 g/m2 weight. The bag filter was securely attached to the exit duct, thus preventing solids losses. For subsequent runs, performed under low loading runs, a high efficiency (99 % collection efficiency for 5 micron diameter particles, 4 m2 cloth area) ventilation bag filter was used because it offered higher collection efficiency and could handle the lower solids loadings. A 1.7 m long, 100 mm dia. flexible hose connected the solids exit of the cyclone to a storage hopper. The hopper was located below and to the side of the cyclone, thus reducing the possibility of solids re-entrainment from the hopper back to the cyclone, a problem experienced by Stairmand (2). The solids hopper itself rested on a load scale, allowing for measurement of the solids caught. The scale was calibrated to +/- 10 grams, and values were logged on an XT computer at 1 s intervals. The data logging program appears in Figure A3 of the Appendi x. In an effort to simulate the high loading conditions in the cyclone at Chatham a solids recycle system was designed and built. Figure A4 in the appendix shows a schematic of the - 43 -system. Briefly the setup envisioned recycling separately, in two similar systems, the solids caught and those passing the polyacrylic model cyclone. Not only were the systems to recycle the solids, they were also intended to measure the solids flow to provide loading and collection efficiency data. Each solids recycle system included a fluidized seal equipped with dual distributor plates, and a solid flow measurement vessel equipped with a porous measurement swing plate, cone valve for solids flow control, and a distributor plate to distribute fluidization air. The solids measurement vessel was intended to measure solids.flow by means of a porous swing plate positioned within the measurement vessel (first proposed by Turner(40)). This porous paper lined, 30 % free area punched hole plate could be manually rotated to block solids falling within the measurement vessel. Once the plate was rotated closed, the falling solids were to collect on the lined plate and be measured. Measurement was to be performed by observing the pressure differential across the plate and deposited solids and thus infer a solids flow rate. The pressure differential was to arise due to the upward flow of fluidizing air originating from the distributor plate located at the bottom of the measurement vessel, below the solids exit. While the larger solids recycle system was to receive solids directly from the polyacrylic cyclone model, the second smaller recycle system was to handle fines that passed the model. These fines were to be caught with a multiclone, - 44 -containing 120 small plastic cyclones, each 50 mm in diameter. These high efficiency Stairmand type cyclones were to separate the fines from gas stream, and deposit them in a steel hopper with sides inclined at 45°. From there the fines were to have fallen into the fluid seal, continued into the fines measurement vessel and finally recycled back into the feed stream. The cleaned gas stream continued on to a bag filter, positioned just downstream of the multiclone. Fines flow measurement was to occur in a similar fashion as in the large recycle system. Upon commissioning it was found that this system did not work for two reasons. Firstly the multiclone was unable to capture all of the fines and let an unacceptable amount of solids pass, overloading the filter. Attempts to increase the multiclone efficiency by increasing the inlet velocity to each small cyclone were unsuccessful. Secondly it was found that the solids caught by the multiclone were so cohesive that they remained in the multiclone hopper. Other elements of the system such as the fluidized seals and the porous measurement swing plates remain untested. While it was considered possible to rectify these problems with another multiclone design and different, solids system, it was not considered to be a practical means to achieve the objectives of this study, given the resources (i.e. time constraint) at hand. - 45 -1.2.3 Particulate solids Fluid cracking catalyst fines, obtained from tertiary cyclones serving a fluid catalytic cracking unit at the Chevron Canada Ltd. oil refinery were used in the experiments reported in this thesis. The equivalent volume sphere diameter, as determined using an Elzone particle analysis instrument, was approximately 22 Jim. The size distribution of the feed material appears in Figure 1.11. A photograph of the test dust appears in Figure 1.12. The bulk density was determined with the solids in a loose form and was found to be 770 kg/m3. Assuming a voidage of 0.5 the particle density was estimated to be 1540 kg/m3. 1.2.4 Data acquisition and analysis Each test of the cold model cyclone was performed according to the following procedure: 1. As an initialization procedure, the system was cleaned out by rapping each component part while scouring the system with air. The mass of the feed hopper, catch hopper( with lid connection off), and filter bag were next recorded. The wet and dry bulb temperature were then noted. (See Appendix Figure A8 for temperatures) The plastic cyclone was wrapped with aluminum foil to reduce electrostatic effects, and the feed hopper pressur i zat i on line was jthen connected. Mass fraction ( % ) \ i a CA N> S u B f* (0 "1 p> 3 cr e (0 »* N 0 (D P CO o SIZE 10 20 30 40 50 60 70 60 90 100 6.78)* 7.32> -» 7.9D-8.54> •fl 9,22)-2 9.96) 110.75)-Zll.6l> (X 12.53>-•-13.53) » 14.61)-a15.77> E17.03)-P 18.3y> | 19.86)-™5:.44> ?23.15)-"> 24.99> 26.99)-29.14> •31.46) 33.97) 36.66X 39.60) co 42.76) 46.17> 49.85) 53.83) 58.12) —* - -*. * —« -» -* — o o 0 * . ... ft —* * * X -* I 10 20 30 40 50 60 70 80 ?0 100 Figure 1.12 Photograph of'FCC test solids ( 22 um mean size') - 48 -With the fluidization air turned on to the wind box, the feed hopper was charged by scooping solids from a barrel and feeding, by means of a funnel, through the 50 mm dia. feed port at the top of the hopper. The receiving storage hopper was sealed. The blower was connected and started. Then the air flow was adjusted via. engine speed. The data logging program was started on the XT computer. The fluidization air was increased until solids were seen to be bubbling and circulating within the hopper, (fluidizing air flow was approximately 100 lpm) The solids feed valve was rapidly opened and the hopper rapped as solids were rapidly fed. Figure 1.8 shows the valve arrangement. The solids valve was quickly shut after the desired amount of solids had been fed. - 49 -9. With the blower still running, the system ( pipes, model cyclone, exit pipes ) was violently rapped to dislodge solids adhering to inner surfaces. 10. The blower was stopped, and the filter bag mass, receiving hopper mass, and feed hopper mass were recorded. Solids samples were taken from the filter bag and receiving hopper for analysis of their particle size distributions. Deviations or unusual occurrences from the above procedure are reported in Table 1.3. In order to derive the collection efficiency curves needed for this study, both the particle size distribution and masses of the feed, catch and loss particles needed to be obtained. As previously mentioned, the total catch and feed masses were determined from the load cells and load scale. The mass of particulate passing to the model cyclone was measured by determining the difference in the filter mass before and after each run. An Elzone particle analysis instrument (model 286XY), interfaced with an AT computer, was used to characterize the particle size distribution. The instrument was calibrated prior to analysis with particles of mean diameter of 5 jim and 20 pm, to correspond approximately to the expected mean sizes of the cyclone loss and catch respectively. ASTM standard CG90 -71 T for particle size distribution analysis by electronic counting - 50 -was followed. A Leitz Tas Plus Image Analysis System was used to confirm the particle size distributions. The particle collection efficiency curve for each run was obtained by comparing the mass caught to that fed for each channel ( i.e. for each size interval ) as follows: Ei = ( C ci )/( C + L )fi (1-28) where Ei = collection efficiency for channel i. C = total mass caught L = total mass lost ( via. filter bag measurement) ci = fraction by mass for channel i of catch, fi = fraction by mass by channel of feed 1.2.5 Error Sensitivity As in any experimental program, the error expected in measurements must be significantly less than the observed quantities in order for the results to be meaningful. The experiments described in this thesis were set up with this in mind. For example, in order that there be sufficient resolution between experiments, the solids chosen had to be fine enough to allow a significant amount to pass the cyclone. - 51 -As well, the manner in which the efficiency was calculated was important. Given the feed, catch, and passing masses from a particular test, there are four ways to calculate the primary dependent variable collection efficiency. They are: a.) E = (catch mass) (1-29) (feed mass) b.) E = (catch mass) (1-30) (catch mass + loss mass) c. ) E = (feed mass - loss mass) (1-31) (feed mass) d. ) E = (feed mass - loss mass) (1-32) (catch mass + loss mass) In Figure 1.13 a comparison is made of the collection efficiency errors resulting from hypothetical experimental errors for equations 1-29 to 1-32. The graphs were prepared while considering the case with a gross collection efficiency of 95%. From these graphs it can be seen that equations 1-30 and 1-31 are less susceptible to experimental error than 1-29 or 1-32. The error in each variable depends on solids caught up in the system, attrition or agglomeration (if any) and measurement error. The efficiency of the bag filter influences Efficiency error ( % ) Efficiency error ( % ) i s I n c O 0 I Cfl •1 n B S3 w ft M> ft _ ft i S1 © s> o G> | N © e 0 © 9 0 o / V \ * / / f f / / / _ 1 + B ' O PS Ul _ » w z * 7 H, ^ w 2 0 i r* u S U s u 5 S 0 0 0 9 6 s s s s s e 1 ' ^ 1 \ \ \ * \ \ \ ii rt n n rt 0 a 0' e> ?' fi fi fi fi \ \ \ \ \ \ \ \ Efficiency error ( % ) Efficiency error ( % ) i o w CO ps n O fi I CO •J o n ft e i ui s 01 S 6 0 SI S © s S W * ri y r * o' ?' • 0 ? r. I' 0 »: • * •; \ K »: ii »: f 9 ^ s s O 3 G> © o to S S o B O © - zs - 53 -loss measurement error. As previously stated the scale accuracies were: feed scale +/- 1 kg catch scale +/- 0.01 kg loss scale +/- 0.001 kg The measurements suffered from the problem of solids lodging in the system. The feed measurement was particularly susceptible to error. This is because the solids fed to the system had more opportunity to be caught up in the piping and entrance way, thus reducing the actual amount reaching the cyclone chamber itself. The catch mass measurement fared much better as the passage to the storage hopper could be easily cleared of any material that was retained. Measurement of solids not captured by the cyclone was hindered not so much by the scale accuracy, but rather by the inevitable loss of particles due to filter inefficiency. This rendered the loss particle size distribution inaccurate and unreliable. A mass balance, performed on a per channel basis failed to close for particle smaller than 15 um dia as is shown in Figure A7 Thus collection efficiency calculations were based on the catch and feed particle size distributions as per equation 1-28. 1.2.6 Chatham cyclone data Figure 1.14 shows a schematic of the Chatham CFB Boiler, a complete description of the installation can be found in igure 1.14 Schematic of Chatham CFB Boiler(32) - 55 -reference 1. Briefly the Chatham CFB boiler consists of a 23.8 m high fluidized bed furnace which discharges solids and combustion gases to a 5.6 m diameter cyclone. Solids are separated from the combustion gases in the cyclone and are returned to the bottom of the furnace. A portion of these solids pass through the Fluid Bed Heat Exchanger (FBHE). The catch sample analyzed in this thesis was taken from the FBHE. A photograph of the Chatham cyclone fines appears in Figure 1.15. The loss particle size distribution was established by analyzing particulate samples from the bag house with the Elzone particle analyzer. The catch sample was sieved ( ASTM Standard Test Method for Sieve or Screen Analysis of Fine and Coarse Aggregates: C-136-76 ) to give a rough particle size distribution. The fines (i.e. material passing a #70 mesh screen) were analyzed further with the Elzone particle analyzer. Equation 1-30 was used to calculate gross collection efficiency data. Solids flux in the reactor was measured at several levels (32) and found to be of the order of 20 kg/m2s. This is said to be similar to values, as yet unpublished, found by isokinetic sampling trials performed by others in the cyclone inlet (35). The solids loading, calculated assuming 1600 m3/s gas flow at STP, and a reactor riser area of 16 m2 was estimated to be 10 kg solids/ kg gas. The runs typically lasted for weeks. Thus it is assumed that steady state conditions prevailed. It is assumed that the Figure 1.15 Photograph of fines from Chatham fluid bed heat exchanger, sampled on April 17, 1990. - 57 -size distributions did not significantly change during transport from the cyclone to the point of sampling or during transport from New Brunswick to Vancouver. This assumption could not be verified because at the time of writing it was not possible to obtain samples directly from the base of the cyclone itself. Data concerning the operation conditions or other entrance configurations will be available pending completion of work at the University of New Brunswick (35). All samples were taken on April 17, 1990, during which time the cyclone had entrance configuration C3 (see Figure 1.10). The operating conditions are summarized in Table 1.2. - 58 -TABLE 1.2 CHATHAM OPERATING CONDITIONS FOR APRIL 17, 1990 COMMENTS REFERENCE TEMPERATURE GAS FLOW GAS VISCOSITY GAS PRESSURE PRESSURE DROP SOLIDS FLUX GROSS COLLECTION EFFICIENCY 850 C 1600 m3/min. 0.000018 kg/ms TOP OF FURNACE (32) STP (3237.5 cm HjO 12.25 cm H2O 20 kg/mJ s (36) @ CYCLONE ENTRANCE (32) (32) (see note above) (32) 99.2 % (1) While the method for the cyclone efficiency calculation is not stated, it is believed that the catch mass flow is calculated from an energy balance across the fluid bed heat exchanger and thus is not y_ery accurate. The bulk particle density for Chatham solids was measured to be 1600 kg/m3 and assuming a voidage of 0.4 the particle density was calculated to be 2650 kg/m3. - 59 -1.3. RESULTS INTRODUCTION Part I of this study examines the collection efficiencies of two geometrically similar cyclones operating at different temperatures, with different solids, and with different inlet velocities. It also examines the performance of the smaller cyclone under different inlet velocities and with various inlet geometries. This variation in inlet geometries was performed in order to provide some initial guidance on similar proposed changes in the Chatham cyclone. Significant attempts were made to match the loading conditions of the Chatham unit but this was not completely possible. Problems associated with feeding solids, collection of the fines not captured by the cyclone, and solids analysis limited the accuracy of the results. The polyacrylic cyclone was sized so as to not exceed the capacity of the mobile blower while maintaining a cyclone inlet velocity between 5 and 10 m/s, and yet be as large as possible. Less consideration was paid to the required solids loading which proved to be very difficult to meet. Initial efforts focused on a system capable of recycling all the solids ( both those caught by the cyclone and those lost, see Figure A4 in the appendix ). These attempts were thwarted by the inability of a 'homebuilt' multiclone, consisting of 120 cyclones, each of 50 mm diameter, to capture all the fines. The particles caught by the multiclones was so- cohesive that it failed to fall out of the multiclone catch hopper, adhering instead to the hopper walls, which were inclined at 45°, without aggressive rapping. The failure to achieve complete recycle - 60 -meant that the solids could only be fed through once, as a short batch operation. Typically each run involved feeding for a period of only a few minutes. Figure 1.16 plots catch hopper mass vs time for run B4, giving an indication of the variation in the feed rate. Typically there was little variation in the feed rate* The particles were chosen as a compromise under a set of conflicting requirements. It had to be fine enough to allow significant losses ( and thus experimental resolution between different conditions ) and yet not so fine that the fraction passing could not be handled. The first three runs were performed with FCC solids with a mean diameter of 60 microns and resulted in 99% and 98.3% capture efficiencies respectively, too high for the resolution required. Subsequent runs used FCC solids with a mean diameter of 22 microns and resulted in acceptable lower efficiencies. Unfortunately, these fines were more cohesive, and proved to be more difficult to feed and analyze. Experimental results are summarized in Table 1.3. Experiments Bl to B3 were shakedown runs using the larger solids ( 60 micron mean ). Runs B4 to Bll attempted to achieve high solids loading rates, while the remaining experiments focused on solids loading effects and the influence of inlet modifications on collection efficiency. Preliminary "shakedown" tests, performed before run Bl, used solids with mean particle diameters of 11 jum and 50 nm, at inlet velocities between 3.6 and 5.6 m/s. Inlet geometry - 61 -128 20 % 100 200 300 400 500 time (s) Figure 1.16 Catch hopper mass vs time indicating steady state solids feed rate. Run B4. RUN 1 DATE PURPOSE :ONPIGIIRATION see Figure .10) SOLIDS AVERAGE DIAMETER LOADING MASS SOLIDS) HASS AIR) FLON •3/s INLET VELOCITY •/s UASSES EFF. * COUUENTS h CAUGHT k LOSS ei 15/5/90 SHAKE DOWN CI, HIGH V.F. 60 UH FCC 2.60 0.19 3.67 n/a "/» n/a FEED PROBLEMS B2 22/5/90 SHAKE DOWN C3, HIGH V.F. 60 UU FCC 0.63 0.19 3.67 79.6 0.80 99 BAD HASS BALANCE B3 23/5/90 SHAKE DOWN CJ, HIGH V.F. 60 UU FCC 3.14 0.19 3.67 SO 0.84 98 feed lass caught in pipes IM 24/5/90 scaling CJ, HIGH V.F. 22 UU FCC 1.25 0.19 3.67 83 1.96 98 •ass balance closes to 2.7H> B5 24/5/90 scaling CJ, HIGH V.F. 22 UU FCC n/a 0.19 3.67 n/a »/« n/i feed probleis B6 20/6/90 scaling C3, HIGH V.F. 22 UU FCC 1.96 0.26 5.04 85.15 3.69 96 GOOD RUN B7 20/5/90 scaling C3, HIGH V.F. 22 UH FCC n/a 0.26 5.04 n/a n/a n/a 99 feed probleis B8 02/7/90 scaling C3, HIGH V.F. 22 UU FCC 7.17 0.28 5.50 101.78 1.13 GOOD RUN B9 10/7/90 scaling C3, HIGH V.F. 22 UU FCC 7.49 0.26 5.04 91.50 1.44 98 GOOD RUN BIO 10/7/90 scaling CJ, HIGH V.F. 22 UH FCC 5.38 0.19 3.67 126.99 0.61 99.5 solids vahe stuck open Bit 12/7/90 scaling C3, HIGH V.F. 22 UH FCC n/a 0.26 5.04 n/a n/* n/a 91 feed probleis Bl 2 18/1/90 LOADING CJ, HIGH V.F. 22 UH FCC 0.23 0.26 5.04 3.9481 0.3697 GOOD RUN Bll 18/7/90 LOADING CJ, HIGH V.F. 22 UH FCC 0.12 0.26 5.04 1.9692 0.2972 87 GOOD RUN BH 18/7/90 LOADING CJ, HIGH V.F. 22 UH FCC 0.13 0.26 5.04 4.3351 0.4936 90 GOOD RUN BIS 18/7/90 LOADING CJ, HIGH V.F. 22 UU FCC 0.15 0.26 5.04 2.6652 0.14S 89 GOOD RUN BI6 18/7/90 LOADING CJ, HIGH V.F. 22 UU FCC 0.44 0.26 5.04 3.636 0.4035 90 GOOD RUN BI7 18/7/90 LOADING CJ, HIGH V.F. 22 UH FCC 0.18 0.26 5.04 4.8347 0.497S 91 GOOD RUN BIS 18/7/90 LOADING CJ, HIGH V.F. 22 UU FCC 1.30 0.26 5.04 9.513 0.4339 96 GOOD RUN BI9 18/7/90 LOADING CI, HIGH V.F. 22 UH FCC 0.048 0.26 5.04 2.7358 0.5217 84 GOOD RUN B20 18/7/90 4/8/90 LOADING CJ, HIGH V.F. 22 UU FCC 0.66 0.26 5.04 2.8295 0.2171 93 GOOD RUN 821 GEOV. CHNG. C2, HIGH V.F. 22 UH FCC 0.36 0.24 4.58 14.5138 0.7697 95 GOOD RUN B22 4/8/90 GEOH. CHNG. C2, HIGH V.F. 22 UH FCC 0.26 0.24 4.58 8.6637 0.7046 92 GOOD RUN B21 4/8/90 GEOH. CHNG. C2, HIGH V.F. 22 UH FCC 0.15 0.24 4.58 7.4569 0.(722 92 GOOD RUN B2 4 4/8/90 GEOU. CHNG. CJ, HIGH V.F. 22 UU FCC 0.23 0.24 4.58 7.1836 0.4793 94 GOOD RUN B25 4/8/90 GEOH, CHNG. CJ, HIGH V.F. 22 UH FCC 0.23 0.24 4.58 7.5607 0.603 93 GOOD RUN B26 4/8/90 GEOU. CHNG. CI, HIGH V.F. 22 UU FCC 0.20 0.24 5.41 6.8837 0.35 95 GOOD RUN B27 4/8/90 GEOU. CHNG. CI, HIGH V.F. 22 UH FCC 0.20 0.24 5.41 5.5855 0.294 95 GOOD RUN B28 4/8/90 GEOU. CHNG. CI, HIGH V.F. 22 UU FCC 0.24 0.24 5.41 4.3072 0.2865 94 GOOD RUN B29 4/8/90 GEOU. CHNG. CJ, LOW V.F. 22 UH FCC 0.36 0.24 4.58 6.0028 0.2812 96 GOOD RUN D30 4/8/90 GEOU. CHNG. CJ, LOW V.F. 22 UU FCC 0.34 0.24 4.58 5.6125 0.304 95 GOOD RUN 611 4/8/90 29/08/90 GEOU. CHNG. CJ, LOW V.F. 22 UH FCC 0.11 0.24 4.58 2.8368 0.229 93 GOOD RUN B32 GEOH. CHNG. CJ, LOW V.F. 22 UU FCC 0.17 0.24 4.58 6.2787 0.3067 95 GOOD RUN B33 29/09/90 GEOU. CHNG. CI, HIGH V.F. 22 UH FCC 0.17 0.24 3.81 4.3102 0.3403 93 GOOD RUN B34 29/08/90 29/08/90 GEOU. CHNG. CI, HIGH V.F. 22 UU FCC 0.096 0.24 3.81 4.2827 0.3315 93 GOOD RUN B3S GEOU. CHNG. CJ, HIGH V.F. 22 UH FCC 0.097 0.24 4.58 3.9135 0.5S07 88 GOOD RUN TABLE 1.3 EXPERIMENTAL RESULTS - 63 -configurations CI and C3 were used with vortex finder length as a variable. Particle loading rates were below 0.05 kg solids/kg air. Collection efficiencies were found to vary between 41 and 96 %. A summary of this test appears in Figure Al of the Appendix. Runs B4 and BIO were chosen for comparison in the scaling study because of particle analysis and data logging problems (i.e. program failed to record data ) in the other runs. Specifically it was found that the particle catch distribution was greater than the particle feed distribution for particle diameters less than 15 microns. It is thought that the feed sample, which was taken before run B4, was finer than that actually fed in subsequent runs. This is attributed to the practice of recycling spent solids from one run to the subsequent run. This was necessary as all the available solids had to be fed in each run in order to achieve the desired loading levels. Unfortunately feed samples were not taken for each run as it was assumed that collection efficiency could be calculated from the catch and loss distributions alone. Note that fresh, unrecycled feed solids were used in runs B4 and BIO. - 64 -1.3.1 Scaling Considerations Catch particle size distributions for runs B4 and BIO appear in Figure 1.17a and Figure 1.17b respectively. Figure 1.17c shows the collection efficiency curve for run B4 as determined from particle size distributions obtained with the Elzone particle analysis instrument( model 286 XY) and the collection efficiency curve for run BIO as determined by the Tas Plus Image analyzer ( model LSI -11 ). The first method reports the equivalent volume sphere diameter, while the second leads to an equivalent projected-area circle diameter (37). Figure A5 of the appendix details the particle size information. Note that in Figure 1.17c the masstfof particles greater than 20 um in diameter have been combined into one channel with mean diameter of 31 um. A similar adjustment was made to the run BIO collection efficiency curve with particles larger than 15 jam diameter being combined into one channel of 31 jum mean diameter. This grouping of the higher end channels was performed because of the erratic nature of the collection efficiency curve above 20 microns. It can be seen from Figure 1.17c that the cyclone was 50% efficient for particles of 10 um diameter. A similar performance is noted in the efficiency curve for run BIO which shows a dP5o of 10 um. This is surprisingly efficient performance for a 0.61 m diameter, non standard design cyclone operating with an inlet velocity of only 3.6 m/s. 0.0 05 + + + + + + + + i cn + + + + + + + + + + + t+-*-+i.++ 4+ + t + + + 4 ^ t » 0 35 70 particle diameter ( microns ) Figure 1.17a Catch particle size distribution for run B4 Particle size analysis by Elzone analysis •>i;i (>li i m». 0.1 O 0.05 n O'-' —i— 35 70 particle diameter ( microns ) Figure 1.17b Catch particle size distribution for run BIO Particle size analysis by image analyzer. 100 90 80 70 60 50 40 30 20 10 0 • m • : 0 • • • . " 43-Via G la 0." • • -a »— i 1 i 03 4* O 10 20 30 40 50 Particle diameter ( microns ) 60 70 - RUN B4 • RUN BIO Figure 1.17c Collection efficiency curves for runs B4 and BIO'. Run B4 particle size distributions determined by Elzone particle analysis instrument. Run BIO particle size distribution by image analysis methods. Conditions as stated in Table 1.3. - 65 -There is a minimum in the collection efficiency at about 2.5 um diameter in the run BIO efficiency curve, with a small increase in efficiency for smaller particles. There are three probable reasons for this. The first is that because of the cohesive nature of the solids, significant amounts of fines may have adhered to larger particles in the catch. During analysis these fines agglomerated with larger particles may have broke free and been counted as individual particles. This would have increased the mass fraction of fines reported in the catch. The second possible reason arises from the problems of dispersing the dust in the feed gas stream. Particles fed to the cyclone came from the solids feed hopper and were maintained in a bubbling fluidized bed. It is possible that the smaller particles started out in an agglomerated state, were fed into the gas stream in this state and finally separated as large agglomerates. It may be necessary to provide a means of dispearsing the particulate while in the feed stream in order that the fines could be truly dispersed. The third possible reason for the surprisingly high measured fines collection efficiency is that the fines may have been swept out of the gas stream by other larger particles. This effect has been reported by others (12, 24). Mothes and Loffler (24) discuss in detail the phenomenon of improved collection efficiency for small particles separated in cyclones with high particle concentrations. They state that particle separation mechanisms other than separation in the vortex must have a major effect on particle separation and - 66 -develop a model to describe this effect. Briefly the calculation of fine particle collection due to agglomeration involves three steps: 1. The initial deposition efficiency of on larger particles settling towards calculated. 2. The gas volume cleaned by the larger particles traveling towards the wall is determined. 3. The decrease in fine particle concentration caused by the cleaning effects of the larger particles is estimated. The model predicted that the separation efficiency of small particles in cyclones is a function of scrubbing particle size, small particle size, dust concentration, flow conditions and material properties. As an example they considered the case of 15 um diameter particles scrubbing out particles sized below 6 um diameter. The results are presented in Figure 1.4 and show a peak in the collection efficiency curves for particles sized between 2 to 3 um diameter. A similar effect is noted in Figure 1.17c and occurs in the same range of particle diameters. Considering the Chatham cyclone size distribution in Figure 1.18(a) the loss distribution was clearly separated from the catch particle size distribution with little overlap. fine particles the wall is - 67 -Comparison of the two distributions resulted in a typical collection efficiency curve with a dpso value of 41 microns as can be seen in Figure 1.18(b). This is in the range reported in reference 1, that being 30 to 45 microns. Performance comparison The collection efficiency curves for run B4 and the Chatham cyclone are plotted in Figure 1.19. In order to verify Stokes scaling, as interpreted by Stairmand, the performance of the Chatham cyclone was shifted to see if the performance would be close to the performance of the UBC cyclone. That is: N8t of shifted dpso = Nst of Chatham dpso dps 0 n 8 w*ppUBCVlUBC = dps Oold2 Pp CHATHAMVl CHATHAM DuBC UVBC DcHATHAM UCHATHAM dps On •« 1 ~ dps 0 o 1 d * pP C H A T H AM V1 CHATHAM Du B C JHQBC J>pUBC Vi UBCDCH ATH AM UCHATHAM Table 1.4 states the assumed scaling conditions. While the curves are brought closer together, a significant discrepancy is still seen. It is clear that the cold model with Stokes law scaling predicts too optimistically the performance of the Chatham cyclone. - 68 -c o « w a S A. 20 40 60 80 100 120 140 Particle diameter (microns) • Catch + loss 160 180 >> o e v w G 0 V o u 100 90 80 70 60 50 10 0 • • • fl m m % ' 1 B. 20 40 60 80 100 120 Particle diameter (microns) 140 160 180 Figure 1.18 a.) Particle size distributions from Chatham fluid bed heat exchangers sampled April 17, 1990. b.) Collection efficiency curve derived from particle size distributions. E = (Cli/(Cli + Llx)) - 69 ->-o z UJ M o M a. u. LU 2 O U LU _J _l o o 100 50?-• • • tc J UBC. . RUN B4 SHIFTED CHATHAM DATA*^.' i ' • CHATHAM DATA paw—Jlm»H» 10 20 30 40 50 60- 70 PARTICLE DIAMETER (pm) Figure 1.19 Collection efficiency curves for run B4, the Chatham cyclone and the Chatham cyclone shifted according to Stokes Law scaling. Scaling conditions as per Table 1.4. - 70 -TABLE 1.4. CYCLONE OPERATING CONDITIONS Chatham UBC (Run B4) CYCLONE DIAMETER 5.6 0.61 m TEMPERATURE 850 21 «C GAS DENSITY 0.326 1.20 kg/m3 GAS VISCOSITY 0.000045 0.000018 kg/m/s PARTICLE DENSITY 2650. 1540. kg/m3 AIR FLOW (@STP) 1600 11.2 m3/min. INLET VELOCITY 20.4 3.66 m/s STOKES NUMBER 0.020 0.0029 REYNOLDS NUMBER 830 000 149 000 LOADING RATIO 8.8 1.4 kg/kg Note that the Stokes Number is defined as: NST so = dp5 0,ppVi/(18 u D) and the Flow Reynolds number is defined as: Nae = p«ViD//i In Figure 1.20 experimental data are plotted against the particle size dependent dimensionless number S°•5 proposed by Abrahamson and Allen (21) to correlate efficiency data from large, high temperature cyclones to those operating at room temperature. This approach was first discussed in section 1.1.4. Data from the Chatham cyclone appears to closely follow the trend shown in Figure 1.2b with the dpso value falling close to 1. However data from Run B4 are less comparable, indicating a much greater collection efficiency than would be expected* The Parker et al... study compared small cyclones operating at extreme temperatures and found that the efficiency data could be plotted against (NRB)(NST)°•5. A plot of the UBC - 71 ->• u z LU M U M li lt. UJ 100 80 20 0 a \ / • « a • 1 H- 1 I 1 I 0 1.0 SO .5 2.0 A. >-O z LU M o UL LU LU 100 80 60 40 20 0 MgW"» HUH* ** a • II a a * 1 B 1.0 So.s 2.0 Figure 1.20 UBC and Chatham data plotted according to Abrahamson and Allen correlations. a. ) UBC data (run B4 conditions). b. ) Chatham data. See Table 1.4 for operating conditions - 72a -and Chatham data on the same graph appears in Figure 1.21. Both the UBC and Chatham data fall above the small cyclone Parker et al. data. This was also found to be the case for other larger cyclones compared in their study. The authors suggest that cyclone diameter must play an important role (11). The major discrepancy between the UBC and Chatham collection efficiency curves, and the fact that the model's performance was much superior to other cold models compared in the Abrahamson study leads one to suspect the experimental data. It is suspected that,in the UBC tests, particles separated in an agglomerated form, separating as large masses rather than independent entities. Sample preparation, which required sonicating the sample in a liquid media, may well have resulted in disassociation of particles which were then counted individually. Indeed a certain degree of agglomeration is evident in the photographs of the test dust. A means of determining the level of agglomeration within the cyclone would be needed to confirm this. Particle agglomeration may have been influenced by two factors. 1. Electrostatic forces, induced by the motion of the particles in the polyacrylic cyclone, may have caused particles to agglomerate. Repeating the experiments in a Steel cyclone or with the addition of an anti-static compound may reduce these effects. 2. Poor dispersion of particles being fed to the cyclone may have resulted in the particles entering the cyclone in an agglomerated state. Changing the experimental apparatus to b ' 1 ""'I i i ) i 11111 ~i—i i i 11111 1—i i i M 11 j 1—ii i mi) oo UBC^ 6 °o°(b A CHATHAM • • • 3 O AAT. OATA A EXXON DATA KMOWLTOM DATA ci i—i i i i ml i—i i i i ml i i i i i ml i i i i i iiii i i i i 11nl 2 S 4 • 1 234 «t254*12 347*12J46' 2 2 4 « 4 T 10 to 10 D 10 10 (N j (Nj" DMENSJONLESS Figure 1.21 UBC Run B4 and Chatham data compared with Parker et al. data. Conditions as stated in Table 1.4 - 73 -include a section where particles are smashed against a baffle or disassociated by sonic action may resolve this problem. - 74 -1.3.2 Loading effect Runs B9 and B12 through B20 were performed for the particle loading study and the results are plotted in Figure 1.22 and the data is presented in Tabel 1.5. A clear increase in collection efficiency is noted as particle loading is increased. Table 1.5 Particle Loading Data run # loading collection effficiency MASS SOLIDS MASS AIR % B12 0.235 91 B13 0.124 87 B14 0.134 90 B15 0.147 89 B16 0.437 90 B17 0.376 1 B18 1.303 96 B19 0.049 84 B20 0.658 93 B9 7.49 8 These reults are consistant with that reported in other works (18) in two ways in that efficiency increases with loading and no limit or maximum is found. Attempts to increase the loading to levels greater than 7.5 kg solids/ kg air were not successful due to solids feeding limitations. The loading effect is commonly attributed to the action of particle-particle collisions and agglomeration mechanisms which allow the solids to settle out in large clusters and strands rather than individual particles (23,24). Confirmation of this 100 98 96 94 92 90| 88 86 84 0.01 0.1 Loading ratio (kg solids/ kg air ) 1.0 10 Figure 1.22 Loading effect on collection efficiency ( UBC data ). Vi = 5.0 rn/s, T = 21 °C, P = 1 atm - 76 -theory is not possible from this experimental work, but it may be possible to verify this if a method of measuring agglomeration levels in the separation zone as a function of inlet loading can be found. Unfortunately it was not possible to achieve the very low loadings ( 1 gr/ft3, 2.3 g/m3 ) reported in the API study (see Figure 1.3) because of problems with the large solids feeding cone valve. An improved experiment would include a means of feeding solids on the 1 to 10 gr/ft3 (2.3 to 23 g/m3) range. A smaller cone valve, one tenth the size of the 0.25 m diameter valve would likely suffice. - 77 -1.3.3 Inlet Modifications Table 1.6 summarizes the experiments devoted to inlet modifications. In order to compare the collection efficiencies for the different inlet configurations the efficiencies listed here have been modified to account for loading effects. The modification procedure was as follows: All runs are compared on a basis of the loading found in run B24 i.e. 0.232 (mass solids/ mass air). This run was chosen because its loading value fell in the middle of all the other loadings encountered in the inlet modification tests. A best fit line was drawn through the loading data found in section 1.3.2 and the equation of this line established as: E( L ) = (0.0301)ln(L) + 0.9097 where E( L ) = collection efficiency at loading L L = Loading (mass solids/mass air) Taking the derivative one obtains: dE( L ) = 0.0301 dL L - 78 -Corrected efficiency values (ECorr.) were obtained by according to: Eoorr . = Egiper. + 0.0301(Lexper . - LB 24 )/Lexper. where Eexper. = efficiency determined at other loading Lesper. = loading in test During the run the inlet geometry and vortex finder length were altered, resulting in different inlet velocities and particle paths within the cyclone. The runs have been placed in order of increasing inlet velocity, with the final three runs differing from the rest in that the vortex finder was elongated 212 mm in this situation. The data is plotted in Figure 1.25. Inlet geometry details are shown in Figure 1.10 and briefly described below. Configuration CI. The base case situation with no inlet inserts or vortex finder changes have been made. See Figure 1.9 for dimensions. Configuration C2. The inlet floor has been raised 50 mm above the original floor and the ceiling lowered 30 mm for a short section near the reactor. The rear wall has been straightened out reducing the inlet width to 110 mm from 150mm. - 79 -Configuration C3. As per configuration C2 but with the inlet width enlarged to 135 mm. Configuration C4. As per configuration C3 but with the vortex finder lowered 212 mm. - 80 -TABLE 1.6 Inlet Modification Tests RUN CONFIGURATION COLLECTION EFFICIENCY B26 cl 96 B27 lB28 cl 4 B33 l 9B34 cl 7 B24 c3 94 B25 \ c3 3 B35 c3 2 B21 c2 94 B22 2 2 B23 c2 3 B29 c3 & low vortex finder 94 B30 -c3 & low vortex finder 94 B31 c3 & low vortex finder 96 B32 3 & low vortex finder 96 It would be expected that collection efficiency would increase as the entrance inlet area was decreased, that is as the configuration was changed from Cl to C3 to C2. However, this was not evident in the data and the base case with the lowest inlet velocity performed better than either C2 or C3. Configuration C3 coupled with a lower vortex finder offered better collection efficiency than C2 or C3, alone but still not significantly different than the base case. For the same geometric configuration it was not possible to reproduce the same collection efficiency as is seen in Figure 1.25. Only in runs B26, B27, B28 and B33, for configuration Cl, were similar collection efficiencies noted. This variation is attributed to differences in the experimental Collection Efficiency ( % ) (Q c -< OJ o a o o H. o 3 -h 3 CL -h TI (D 0) -( (t> 3 ft r+ 0 CO -1 to o 3 3 O ^ -h <t> W ft 0 O H-H. O a 3 a> CO -h O H« ft 0 C 0 0) -< CL B> H- ft o o -h 3 H. (Q O 3 H o w o o -1 -1 <I> o ft 0) (P </> 3 (6 O (D H' <p H W a 0) cr -h W H- O o o z ft c » H~ w 0 z 0 W o 0 0 - o V r-- 82 -conditions between these runs, specifically the variations in the loading rate. While it seems clear that collection efficiency is greater than 90% for all runs, it is not possible, based on the data collected to date, to declare a "winner" among those tried. These puzzling results suggest that a different experimental approach is necessary. An improved experimental program would result in a steadier feed rate, perhaps by way of a mechanical feeder as it was difficult to repeat loading levels with the present solids valve feeding arrangement. As well, a solids system less prone to agglomeration would allow efficiency curves, which span a full range of efficiencies, to be determined. Examination of these improved efficiency curves may offer better answers. - 83 -1.3.4 Flow visualization Separate runs of the cyclone apparatus were performed for the purpose of flow visualization. Collection efficiency was not evaluated during these runs. These runs were performed at low loading in order that flows in the interior could be observed. During the trials the following observations were made for all geometries: Recirculation zone below roof: It could be seen that a distinct annular region exists, concentric with the cyclone itself but situated above the roof of the entrance and below the roof of the cyclone. Suspended solids either enter this zone or fall from it to the separation region below. Any solids that found their way into this zone are eventually reentrained in the incoming flow, but spend considerable time in the region. Reentrainment into incoming flow: Solids that manage to reach the outer wall must drop below the bottom of the entrance if they are not to be reentrained again by the incoming flow. The Stokes number scaling - 84 -approach is unlikely to apply to particle-particle and particle-wall interactions. Their effect may cause performance to be somewhat unrepresentative of performance in the full scale cyclone. It can be argued that those particles whose misfortune it was to reentrained must then be separated once again with particles in the incoming flow, and thus may be representative of matters at Chatham. For the base case, where no inlet inserts were used, a recirculation zone could be seen positioned in the entrance way near the top of the model combustor. This recirculation was not evident with the inserts in place as in Configuration C2 or C3. There was no great visible difference in the particle paths for the existing and modified vortex finder configuration. This may be due to the difficulty in seeing flow patterns in the dilute region close to the vortex finder. More advanced techniques for flow visualization (e.g.. laser Doppler techniques) would be required to study this in detail. The video tape sent to Energy, Mines, and Resources shows the . observations discussed above. - 85 -1.4 Conclusions and Recommendations. The main features of part I of this study are as follows: 1. A one-ninth scale polyacrylic model of the non standard design industrial cyclone operated at the 22 MWe CFBC facility at Chatham, New Brunswick has been constructed and operated at room temperature. Particle collection performance has been tested under various solids loadings and inlet geometries. 2. A solids recycle system, utilizing a multiclone to capture fines and a recycle system to provide continuous operation was built, but did not work well because of solids capture problems and the cohesive nature of the fine solids. As a result, all of the results presented are for a batch system. 3. At inlet velocities varied between 3.7 and 5.5 m/s, solids loadings between 0.05 and 7.5 (kg solids/kg air), for FCC solids having a mean diameter of 22 jim, cyclone collection efficiencies remained above 90%. 4. There was disappointing agreement between the results from the Chatham unit, scaled according to Stokes Number scaling, and the findings obtained from the cold model unit. 5. The grade efficiency curve showed a minimum efficiency for fine particles, 2.5 to 3.0 um in diameter. This is likely due to agglomeration effects as a mass balance performed on a per channel (i.e. -86-size interval) basis did not close for particles smaller than 15 um diameter. 6. Increasing particle loading led to an increase in collection efficiency. 7. Inconclusive results were found when the inlet configuration was changed while using 22 um mean particle diameter FCC solids. 8. Flow visualization trials were performed with the polyacrylic cyclone. Reentrainment into the incoming flow of captured solids skipping along the wall was observed. A videotape of these observation was prepared and sent to Energy, Mines and Resources in Ottawa. It is recommended that, in order to verify the scale-up criteria, the experiments be repeated with a different solids system less prone to agglomeration. Successful collection efficiency studies have been performed using flyash at lower loadings (11, 12) suggesting that flyash might be a suitable material. - 87 -PART II HOT CYCLONE TESTS - 88 -2.1 INTRODUCTION Circulating Fluidized Bed Combustors (CFBC) rely on inertial separation devices to separate combustion gases from entrained solid particles and to return those solids to the reactor. Cyclones are usually chosen to perform this high temperature, high loading and sometimes high pressure separation because they offer reasonably good particle collection efficiency and are easy to design, operate and maintain. When used in such circumstances, some combustion inevitably occurs within the cyclone. In order to better understand the combustion processes, within the cyclone, radial combustion gas concentration profiles were measured within a secondary cyclone serving a pilot scale CFBC system operated at the University of British Columbia. The profiles presented in this thesis were all obtained with Highvale coal, a low sulphur coal from Alberta as a fuel. For details of properties of the coal see See Figure A6 of the appendix. In this section a brief review of gas and solids flows within a cyclone is given, followed by a description of the apparatus and a presentation of the measured combustion gas combustion profiles. The profiles are then discussed and conclusions presented. 2.2 Theory Combustion processes occurring within cyclones are dependent on the nature of the solids, the composition of - 89 -the gas as well as the operation mode of the combustion system. Gas and solids flow patterns within cyclones are functions of several variables including cyclone dimensions and geometry, gas flow rates, and particle loading. When used in a CFBC system, the solids are a combination of unreacted fuel particles, inert solids (Eg. sand and ash) and sorbent material (if a sorbent is used). The combustion gas composition depends on fuel type, combustor configuration, mode of operation, operating temperature and other parameters. In order to understand the gas concentration profiles, it is first necessary to briefly describe the gas and solids flow patterns within a reverse return cyclone and to present the overall combustion equations which characterize CFBC systems. Flow patterns Gas and solids flow patterns within a cyclone are intimately related. Solids flow patterns have been observed in a cold model by the author (see section 1.3.4) and documented by several other workers (2, 25). The radial, tangential, and axial components of the solids velocity vary with position and solids loading (15). Predictions of low loading particle mean trajectories appear in Figure 2.1 (25) and show the predicted axial and radial position of particles of various sizes. - 90 -Figure 2.1 Predicted particle trajectories in a vertical plane within a Stairmand type cyclone. Low loading conditions. (25) a. Mean particle trajectories for particles of diameter 1 to 10 microns. b. Mean particle trajectories, 3 micron. c. Random particle trajectory of 2 microns particle in turbulent flow. - 91 -In Figure 2.2 the axial gas velocity is seen to be a function of radius, with downward motion occurring in the outer regions, while within the core the flow reverses and travels up with increased velocity towards the vortex finder. The tangential velocity has been found to increase with decreasing radius, reaching a maximum in the central core region below the vortex finder. Increased particulate loading has been found to reduce the tangential component of gas velocity (24). Figure 2.3 shows the combined radial and axial velocity vector field. Particle trajectories Particle motion within these complex flow patterns depends on gas and particle characteristics, and on solids loading (15). As solids loading increases, particle-particle collisions become more frequent, forming larger clusters and strands which assist particle collection. Once the particles have reached the wall they may be re-entrained should some disturbance occur, forcing the solids back into the gas flow. Disturbances such as large bouncing particles, interior wall surface imperfections and interference of the gas vortex with the cyclone wall have been discussed by various authors (2,24). At the wall a dense layer of solids forms, with particles travelling around and down the cyclone body, eventually concentrating into a distinctive dense strand which "snakes" its way down to the solids exit at the base. - 92 -Figure 2.2 Predicted radial and axial gas flow patterns in a Stairmand type cyclone. Low loading conditions (25). - 93 -::;iwi ii tnitti j« .iMiitti .Ijliltti ,ninth ittttittt 111 111 11111 Itltl ii jtllti., ttti Figure 2.3 Predicted combined axial and radial velocity'vector diagram in a Stairmand type cyclone. Low loading conditions(25) - 94 -Summary of combustion equations. While it is not within the scope or intent of this thesis to discuss heterogeneous combustion mechanisms for coal combustion within a CFBC, it is useful to describe the primary reactions affecting the gases measured within the cyclone. The gases C02 , CO, CH*, SO2 , and NO, are all formed during coal combustion. CO2 is primarily formed by the two reactions (26): C* + Oj -> CO2 CO + 1/2 02 -> C02 . CO can be formed as follows: C* + CO2 -> 2CO C + HiO -> CO + H2 CO2 -> CO + 1/2 O2 C* represents the burning char. CH4 can be formed from: C* + 2 H2 -> CHJ , or released as volatiles are evolved. SO2 can be considered to form via: S + O2 -> SO2 . while NOS is formed by reactions of the type 1/2 N2 + x/2 O2 -> NO, . At temperatures of interest in FBC processes, it is fuel nitrogen, rather than nitrogen present in the air, which is predominantly responsible for NO* formation (38). - 95 -Because most combustion within the CFBC system occurs before the secondary cyclone, much of each gas measured originates before the cyclone. A lesser amount is formed within the cyclone contributing to the measured values. 2.3 Apparatus and Data Acquisition The apparatus used in this high temperature study is the pilot scale fluidized bed combustor at the University of British Columbia in its modified configuration (30). A schematic is shown in Figure 2.4. In brief, the set-up includes a refractory lined reactor (152mm square in -cross-section by 7.3m tall), a primary and secondary cyclone with provision for solids recirculation via an L-valve and jet educator respectively, hoppers for feeding the fuel and sorbent, and primary and secondary air injection. The system is well instrumented. Comprehensive descriptions can be found elsewhere (27, 29). The secondary cyclone was chosen over the primary cyclone for these experiments because of the probe plugging problems associated with gas sampling in regions of high solids density (e.g.. at the cyclone wall). The insulated stainless steel secondary cyclone is situated after the primary cyclone and thus receives the reactor flue gases and particles not captured in the primary cyclone. A scale drawing of the secondary cyclone appears in Figure 2.5. This 0.2 m i.d. cyclone was modified to allow the insertion of a gas sampling probe through the wall at several levels. The probe itself was connected via cooling - 96 -secondary air water liquid fuel Q natural gas primary air Simplified schematic diagram of circulating fluidized bed combustion facility 1. Reactor; 2. Windox; 3. Primary cyclone; 4. Secondary cyclone; 5. Re cycle hopper; 6. Standpipe; 7. Eductor; 8. Secondary air preheater; 9. Flue gas coolers; 10. Baghouse; 11. Induced draught fan; 12. Fuel hopper; 13. Sorbent hopper; 14. Rotary values; 15. Secondary air ports; 16. Membrane wall; 17. Pneumatic feed line; 18. External burner; 19. Ventilation; 20. Calorimetric section FIGURE 2.^ CFBC Schematic (27) - 97 -PORTS LDCATED BELLTW THIS POSITION (PDRTS SHOWN 90° CV IN ELEVATION BELOW) EXIT INLET VIDTH = 50 MM 650 SDLIDS EXIT"H FIGURE 2.5 Scale drawing of secondary cyclone of UBC CFBC system. - 98 -and filtration stages to a gas sampling train. Figure 2.6 shows a schematic of this apparatus. The gas sampling train leads to five analyzers for the measurement of CO2, CO, CH4, SO2, N0X, and O2 gases. A brief summary of key features of these instruments is presented in Table 2.1. A more detailed review is found in reference 34. TABLE 2.1: DESCRIPTION OF ANALYTICAL INSTRUMENTS GAS MAKE/ MODEL 02 C02 FUJI 732 CO HORIBA PMA 200 OXYGEN ANALYZER FUJI 732 CH4 FUJI 730 S02 HORIBA PIR 2000 NOX MONITOR LABS INC. MODEL 8840 RANGE OPERATION (ACCURACY) PRINCIPAL (% full scale) 0 - 25 % ( 1% ) 0 - 20 % ( 1 % ) PARAMAGNETIC TYPE NDIR 0 - 1000 PPM NDIR ( 1 % ) 0-0.5 % ( 1 % ) 0-1000 PPM ( 1 % ) 0-500 PPM ( 1 % ) NDIR NDIR RESPONSE t ime 20s 5 s 5 s 5 s 5 s CHEMILUMINESCENCE 3 min Each of these instruments was calibrated using standard gases prior to each run. The procedure followed to obtain data was as follows: STAINLESS STEEL PROBE: 6HM DIA. Nn SD CH-4 CO / 2. °2 r PUMF i HEAT EXCHANGER SECONDARY CYCLDNE FRDM DTHER SAMPLING PLTINTS GAS ANALYSIS METERS FIGURE 2.6 Gas sampling system serving UBC CFBC system. -100 -1. With the pilot plant operating as nearly as possible under steady state conditions and gas analyzers prepared according to manufacturers' instructions, calibrated, the sample probe was inserted to the desired radial position and gas sampling commenced. 2. After a period of 4 minutes the gas transport lines were assumed to be purged and the gas analyzers reading steady state values. 3. Gas concentration values were recorded. 4. A flue gas [02] measurement was obtained from a separate gas meter operating on a separate sampling line connected downstream of the cyclone. The sampled gas stream was filtered and cooled. - 101 -5. The sample probe and filter were purged with compressed air, then moved to the next radial position. A random order was followed in probe positioning in order to avoid systematic variations. 2.4 Results and Discussion The CFBC operating conditions under which the gas concentration profiles were obtained are outlined in Table 2.2 for each of the five runs where data were obtained. Table 2.2 UBC CFBC operating conditions RUN # RUN AIR RATIO RUN SUPERFICIAL TEMPERATURE 2nd/prim. VELOCITY IN RISER ( C ) (M/S) 17 870 1 6 18 870 1 6 5 886 2 6 6 870 2 6 10 870 0.5 7 The secondary air ports, as shown in Figure 2.4, supplied between 50 to 100 % of the primary flow and were located 0.9 m above the primary air distributor. The riser velocity is - 102 -the superficial velocity in the upper part of the column above the secondary air ports. The gas concentration profiles for runs 17, 18, 5, 6, and 10 appear in Figures 2.7, 2.8, 2.9, 2.10, and 2.11, respectively. For each gas the actual measured concentration is plotted against the non-dimensional radius r/R, where R refers to the cylinder or core radius at that level. For all of the measurements it was possible to maintain sufficient sample gas flow to allow easy measurement, with the exception of readings taken right at the wall. For these measurements gas flow was possible for 10 minutes at most before blockage occurred, necessitating probe and filter purging. This is understandable since a denser region of solids exists at the wall than in the interior of the cyclone. While it was possible to obtain measurements in the vicinity of the wall the values may be affected by excessive solids in the probe and filter portions of the sampling system itself. Char caught in the probe could continue to burn, thereby increasing the CO and C02 values while, reducing the measured O2 values. This effect is assumed to be negligible, however, as the gases and particulate were quickly cooled after extraction in the cooling section of the probe and thus were much less reactive. Typically the solids caught by the secondary cyclone had a mean diameter between 40 to 58 pm (29). Trends for the various components measured were as follows: - 103 -25 20 CCCLI vs r/R e ° e $ a? ^ 15 } CM o 10 < - 5 + 0 0.5 r/R [ Oj 1 vs r/R [ NOx ] vs r/R ( CO 1 vs r/R 180 -2" 904 O o o O O C o 0.5 r/R f S02 ] vs r/R [ CH* ] vs r/R FIGURE E.7 Gas Concentration Profiles for run 17. port # 4 - 104 -f CO J vs r/R [ COj 1 vs r/R a* 05f 33 U — 0 o u [ SOs 1 vs r/R S a a ~ 0.5f O 00 250 a 200 a a 150 4-O 55 * 100 50f FIGURE 3.8 Gas Concentration Profiles for run 18. port # 4 Qi u o -ia o •dd | 'QM | Pi -o * ( 'ID I tA o 8 •dd I 'OM I "'III | "OH | Pi t. X I MO I % I MI3 ! in d add | to J •dd ( «OD I •dd J l0 \ •dd | in | -_*_fO ei add ( IQ | •dd f IQ | o ft. tA •dd I IQ3 J O M O add I IQD J -i O o •dd ( OD 1 •dd ( 03 1 •dd J 03 | add ( 03 | •dd | 03 | - 106 -[CO 1 vs r/R [ CO* ] vs r/R 3? 20-15 10 u 5 O 5 r/R [ SOt } vs r/R 200 - 10Dt o w 50 + o o 0.5 r/R [ CHU ] vs r/R Figure S.10 Gas Concentration Profiles for run 6. port # 5 [ GO ] vs r/R S a a — 30 o u * " 10I COr 1 vs r/R 20T " o o o O r Qt 1 vs r/R 05 . r/R [ SOz 1 v? r/R -* 7' ~ 6 « 5 o o «4 o 2i-- 1 200 B a a o o — 100} o 0 0.5 r/R to r/R [ CH4 1 vs r/R i o 5 2 + —' 1 as o o o o 9 ,05 1 r/R Figure 2.11 Gas Concentration Profiles for run 10. port # 4 - 108 -NO, : No radial trend for NOx is apparent from the traverses performed in the five experiments. Values in the 200 to 225 ppm range were typical, with the exception of the traverse performed at port # 8 near the bottom of the cyclone. The discrepancy at this level is attributed to difficulties in maintaining flow at this port due to blocking by particles. SOj : Values between 0 and 140 ppm [SO2] were observed within the cyclone, but with no evidence of a clear trend. In Figure 2.10 the lowest SO2 concentrations were often observed near the wall, probably because of the denser solids region existing near the wall. While no sorbents were added during the runs, there may be sorption by calcium or other elements in the ash accounting for the reduced values. CH4 : During most tests the gas analyses were unable to detect measurable [CH4]. Only in runs 10 and 18 there were detectable values, these being of the order of 0.002 to 0.004 percent. No radial trend is evident. CO2 : Values between 16.1 and 19.3 percent were measured within the cyclone and a slight radial trend is observed in - 109 -three of the traverses. Combusting char particles in the denser solids region near the wall are likely responsible for the observed small increase near the wall. CO: As with the [CO2], [CO] tends to be somewhat greater at the wall than in the interior in five of the radial traverses. Four traverses show a doubling of the [CO], again indicating some char combustion inside the secondary cyclone. Os : No clear trends in the [ O2 ] were seen in the profiles indicating that the extent of combustion occurring within the cyclone is rather small, at least for this fuel. Comparing the gas residence times of the cyclone and the riser it is clear that the cylone volume is less than that of the riser, approximately 60 % of the riser volume. As well because there is a greater chance for gas short circuiting in the cyclone than in the reactor the reduced extent of combustion is not surprising. Gas sampling for each run lasted less than three hours and during that time the combustor was maintained as closely as possible at steady state conditions. The primary indication of steady operation was the flue gas 02 concentration which was held between 3.5 and 6 percent for all readings, and 3.5 to 4.5 percent for more than 90% of - 110 -the time. Due to the nature of the experiments some scatter in the measured values was inevitable and clouded any subtle trends that may have existed. As well, the flow in a cyclone is very well mixed and any regional combustion activity would be difficult to detect. For example if combustion was preferred at a given radius, one would only expect to find increased combustion products concentrations if the flow was sufficiently segregated. This is not typically the case in cyclones as the flow is more turbulent and well mixed than segregated. An improved experiment would see continuous monitoring at several radial points simultaneously (rather than sequentially), with the focus being placed on the region near the wall. 2.5 Conclusions and Recommendations The main features of part II of this work are as follows: 1. Radial gas concentration profiles within a secondary cyclone serving a high temperature Circulating Fluidized Bed Combustor operating with a low sulphur coal as fuel have been measured and presented. 2. Radial gradients in the gas concentration profile were negligible for [NOx], [SOj], [CH41, and [Oj]. - Ill -3. Near the wall [CO] levels increased, as did [CO2], suggesting increased char combustion in this zone. It is recommended that attempts be made to determine similar radial gas concentration profiles in the primary cyclone of the UBC pilot scale CFB combustor. This would require a probe capable of sampling gases in the much denser solids regions of the primary cyclone. It is also recommended that simultaneous monitoring at several radial points be performed as opposed to a sequential sampling. This would reduce the uncertainty imposed on the radial profiles due to the time varying nature of the species concentration within the gas flow (i.e. non-steady state operation). Profiles should be attained with a less reactive coal (e.g. anthracite) and a high sulphur coal (e.g. Minto ). Cyclone reactions may play a more important role for these fuels. -112-NOMENCLATURE a Inlet depth ( m ) A Coefficient ( - ) ^Ac Centrifugal acceleration ( m/s2) AB Gravitational acceleration ( m/s2) Ai Inlet area ( m2 ) AP Particle Area (m2) b Inlet width ( m ) be Inlet width ( m ) bi Coefficient ( m ) ci Fraction by mass for channel i of catch. ( - ) cm Solids loading (g/m3) Co " Zero load " dust concentration ( grains/ft3, g/m3 ) c« Dust concentration ( grains/ft3 ) C Total mass caught ( kg ) C Cunningham correction factor ( - ) C Dust concentration ( grains/ft3, g/m3 ) CD Drag coefficient ( - ) dcrit Critical particle diameter ( m ) dp Particle' diameter ( um ) dPa Aerodynamic diameter ( g /cm3 )o.5 dpC Feed aerosol mass medial diameter ( cm ) dpso Particle diameter caught with 50% efficiency (mi crons) D Body diameter ( m ) - 113 -Db • Bottom diameter ( m ) Dc Cyclone diameter ( m ) DH Hydraulic diameter of cyclone inlet ( m ) Do Outlet diameter ( m ) Dz Imaginary cylinder diameter ( m ) e0 Zero load collection efficiency ( % ) E Collection efficiency ( % ) Eoorr. Loading corrected collection efficiency ( % ) Ee xper . Experimental collection efficiency ( % ) Ei Collection efficiency for channel i. ( % ) Eo Collection efficiency at low loading ( % ) fi Feed mass for channel i ( kg ) Fo Centrifugal force ( N ) FD Particle drag force ( N ) G Cyclone configuration parameter. ( - ) Ga Galileo number ( - ) he Inlet depth ( m ) Ho Overall height ( m ) ki Coefficient ( - ) kpi Restitution coefficient ( - ) li Fraction by mass for channel i of loss. ( kg ) L Experimental solids loading ( kg/kg solids/air ) LB 24 Particle loading in run B24 ( kg/kg solids/air ) Ley Cylinder length ( m ) Lni Total mass passing cyclone ( kg ) Lo Outlet length ( m ) LP Particle loading ratio ( kg/kg solids/air ) LFAC Apparent gas viscosity coefficient ( - ) m Particle mass ( kg ) -114-n Vortex exponent ( - ) N Particle path revolutions ( - ) NR E t Flow Reynolds number ( - ) NR e p Particle Reynolds number ( - ) NR grp Radial particle Reynolds number ( - ) NBt Stokes number ( - ) Ns t 5 0 Stokes number for cut diameter ( - ) P(e) ^ Probability associated with high loading E ( - ) P(Eo) Probability associated with low loading E ( - ) Q Gas flow ( m3 /s ) Qc Gas split ratio ( - ) cyclone radius ( m ) ri Vortex finder radius ( m ) R Radial co-ordinate ( m ) R. Radius of imaginary cylinder ( m ) S° • 3 Dimensionless particle diameter ( m ) SFACTOR Saltation factor ( - ) TA(X) Predicted collection efficiency ( - ) V, Inlet Velocity ( m/s ) Vi Inlet velocity ( m/s ) vP Radial particle velocity component ( m/s ) V,p Radial particle velocity ( m/s ) Vre Radial gas velocity ( m/s ) Vtp Tangential particle velocity ( m/s ) Vfl Particle saltation velocity ( m/s ) -115-Greek symbols p Gas density ( kg/m3) pP Particle density ( kg/m3) XG Mass median particle diameter ( m ) x Shape factor ( - ) uK Gas viscosity ( kg/ms) utt;p Effective gas viscosity ( kg/ms ) - 116 -References 1. Sorbent optimization for FBC technology, Chatham Circulating Fluidized Bed Demonstration project, Energy Mines, and Resources report (May 1989). (j£>. Stairmand, C.J., "The design and performance of cyclone separators.," Trans. Instn. Chem. Engrs. (Vol. 29, 1951): pp. 357-383. (Ttj. Beeckmans, J. M. , Aerosol Measurement; Lundgren, D. , Ed.; University at Florida Press, Gainsville, FL, 1979. 4. Calvert, S., Englund, H., Handbook of Air Pollution  Technology, New York: John Wiley and Sons, 1984, pp. 320-329. 5. Strauss, W. ,„ Industrial gas Cleaning 2nd edition., New York: Pergamon Press, 1975, pp. 238-276. (6\ Leith, D., Mehta, J., "Cyclone performance and Design," Atmospheric Environment (1973 Vol. 7): pp. 527-549. '/fT). Gauthier, T.A. , Briens, C.L., Galtier, P., "Uniflow ^ cyclone efficiency study," Dept. of Chemical and Biochemical Engineering, University of Western Ontario. 8. Ter Linden, A. J., Tonindustrie-Zei tung, 2 2(i i i) 49 (1953). 9. Mukhopadhyay, S.N., Chowdhury, K.C. Roy, "The collecting efficiency of a cyclone separator," , Brit. Chem. Eng. (April 1970): Vol. 15, No. 4. pp. 529 10. Whitby, K.T., Lundgren, D.A. , Jordan, R.C, J. Air.  Pollution Control Association 11,(1961): pp. 503-504. - 117 -V \1_1/. Parker, R. , Jain, R., and Calvert,. "Particle collection in cyclones at a high temperature and pressure," Env. Sci. Tech. (1981): 15 (4) pp. 451-458. 12. Wheeldon, J. M., Burnard, J.K. "Performance of Cyclones in the Off-Gas Path of a Pressurized Fluidized Bed Combustor," Filtration and Separation (May June 1987): pp. 178-187. (13) Dietz, P.W., "Collection efficiency of cyclone 1/ separators," AIChE Journal. (1981), (Vol. 27, No. 6) 14. Gloger, J., Hanke, S., Energietechnik (1971): 21(4) 158-164. 15. Mothes, H. , Ldffler, F., "Motion and Separation of particle in a cyclone," Chem. Engg. Process. (1984) 18 pp. 323-331. 16. Abraham, F.F., Physics Fluids (1970); 13,- pp.. 2194-5. (17.' " Cyclone Dust Collectors," API publication, A.P.I. New York (1955). 18,1 American Petroleum Institute (API) "Manual on Disposal of Refinery Wastes," API publication 931, Chp. 11 (1975). 19/. Sproull, W.T., "Effect of Dust Concentration Upon the Gas Flow Capacity of a cyclonic Collector," J. Air Po11.  Control Assn. 16 (8), pp. 439-441 (1966). 20. Massey, B.S., Mechanics of fluids, 4th ed. New York: Van Nostrand Reinhold Company, 1979 pp. 222 -227. 21. Abrahamson, J., Allen, R.W.K., "The efficiency of conventional reverse return cyclones at high temperatures," ChemE SYMPOSIUM Series No. 99. (1988) . 22. Ogawa, A., " Theoretical approach with Markov Process Model to Seperation Processes of Cyclone dust collector depending on feed dust concentration," J. Col 1.  Engineer ing, Nihon Univ., A-26, (March 1985). Note: Referenced in 41. - 118 -23. Masin, J.G., Kock, W.H., "Cyclone Efficiency and Pressure Drop Correlations in Oil Shale Retorts," Environmental Progress Vol 5, No. 2, (May 1986), pp. 116 -122. 24]. Mothes, M. , LOffler, F. , "Motion and deposition of particle in cyclones," Ger. Chem. Eng., 8 (1985,) pp. 223 -233. 25. Boysan, F., Ayers, W.H., Swithenbank, J., "A Fundamental Mathematical Modelling Approach to Cyclone Design," Trans. IChem Engg. Vol. 60, 1982. 26. Laurendeau, N. M., "Heterogeneous Kinetics of coal char gasification and combustion," Prog Energy Combust. Sci, Vol.4, pp. 221-270 (1978). 27. Grace, J.R. and Lim C.J., " Circulating Fluidized Bed Combustion of Coal, Woodwastes and Pitch," Final report prepared for Energy, Mines and Resources Canada, under contract 24ST.23440-6-9007 (1987). 28. A.S.H.R.A.E., "Flow measurements, instruments and apparatus," ASME Power Test Codes. Part 5 of Ch. 4. (1959). 29. J. R. Grace, C.M.H. Brereton, C.J. Lim, R. Legros, R.C. Senior, R.L. Wu, J.R. Muir, R. Engman "Circulating Fluidized Bed Combustion of Western Canadian Fuels," Final report prepared for Energy Mines and Resources Canada, under contract 52 SS.23440-7-9136 August 1989. 30. R. Legros, C.M.H. Brereton, C.J. Lim, H. Li, J.R. Grace and E.J. Anthony . "Combustion characteristics of different fuels in a Pilot Scale Circulating Fluidized Bed Combustor," Proc. Conference on Fluidized Bed  Combust ion. 661-666 (1989) 31. Wu, L. R., PhD. thesis, "Heat Transfer in Circulating Fluidized Beds" University of British Columbia, Department of Chemical Engineering (1989). 32. M. F. Couturier, B. Doucette and S. Poolpal, " A Study of Gas concentration, Solids Loading and Temperature profiles within the Chatham CFB Combustor," a report to ENERGY MINES AND RESOURCES CANADA, (Nov. 1989). 33. Stern, A.C, gen. ed. , Air Pollution 3rd edition, New York: Academic Press, 1977, chapter 3, pp. 98-136. - 119 -34. Razgaitis, R., Guenther, D.A., "Separation Efficiency of a Cyclone Separator with a Turbulence-Suppressing Rotating Insert," Transactions of the ASME Vol. 103, (July 1981). 35. M.F. Couturier, Professor,Department of Chemical Engineering University of New Brunswick, Personnel commun i cat i on. 36. Sucec, J., Heat Transfer, WM.C.Brown Publishers: Dubuque, Iowa (1985), p. All 37. Seville, J., et al., (1984) Particle Characterization Vol. 1, no.l Weinheim, Federal republic of Germany (July 1984) 38. J. Zhao, J. R. Grace, C.J. Lim, C.M.H. Brereton, R. Legros, and E.J. Anthony "NOx emissions in a pilot scale circulating fluidized bed combustor," in Proceedings of EPRI/EPA 1989 Joint Symposium on Stationary Combustion NOx Control. 39. Leith,D., W. Licht, "The Collection Efficiency of Cyclone Type Particle Collectors-A New Theoretical Approach," Air Pollution and It's Control, AIChE Symposium Series 126,68, 1972. 40. Turner 41. Patterson, P.A., PhD. Thesis, "High Temperature Cyclones", Department of Chemical Engineering, McGill University, Montreal (1989). - 120 -APPENDIX PRELIMINARY TESTS PART I CONDITIONS SOLIDS: 11 um mean dia. flyash. Inlet velocity: ^.4-5 m/s Configuration: CI ( base case ) variable vortex finder length. V.F. position L/D CFULLY ADJUSTABLE V.f.> L/H collection efficiency o.is 0.11 50.0 0,59 0.54 73.9 0.81 i 0.74 i ! 96.3 PART II CONDITIONS SOLIDS: 50 um mean dia. flyash. Inlet velocity: 5.^ m/s Configuration: 1. C3 without extended vortex finder. S. variable vortex finder length. V.F. posliior: |L/D L/H collection efficiency EXISTING 0.21 0.23 41 V. PROPOSED O.o3 0.57 47 y. '.nodl'FlcQ "tions FIGURE Al Shake down test summary. All runs performed before run Bl. ORFICE FLOW 0 4 8 12 16 20 24 28 PRESSURE DROP (cm water manometer) FIGURE AS Orifice flow curve. M3/s vs pressure drop. - 123 -Figure A3 Data logging program serving UBC model cyclone apparatus. REM PROGRAM FOR INSTANTANEOUS SAMPLING OF LOAD CELLS OPEN "cyco.QUT" FOR OUTPUT AS #2 OPEN aCOrfl:9600,N,8,l,CStDS" FOR RANDOM AS tl INPUT "run nueber?", ni PRINT IE. "run nuaber = ",'nt 3 TIKE$ = "00:00:00" 5 ta« = TIME*. 7 seconds = VAL(KID$(ts$, 1, 2)) * 3600 + VAL(NID$(tB$, 4, 2)1 t 60 • VSL<HID$(ts$, ?, 2)> 9 IF (seconds - hh) = 1 THEN 20 20 PRINT "logging data" hh = seconds PRINT II, "2"; INPUT #1, bt PRINT tl, "3"; INPUT tl, c* PRINT 12, VAL(MID$(b$, 5, 6)), VfiL(MID$(c«, 5, 61) PRINT VAL(MID$(b$, 5, 61), VAL(MID$(c$, 5, 6)1 GOTO 5 30 PLEXIGLAS MODEL CYCLONE / MULTICLONE FLUIDIZED SEAL WIND BOX -4-— TO BAGHOUSE PLATE VALVE LARGE MEASUREMENT VESSEL PLATE VALVE SMALL MEASUREMENT VESSEL 3 FROM BLOWER Figure A4 Schematic of attempted recycle system schematic showing a high solids loading feed and measurement vessels, multiclone and bag filter arrangements. -125-PAGE 12S INSERTED FOR PAGE NUMBER CONTINUITY - 126 -Figure A6 UBC CFBC solid fuel analysis for run B17 (29). Proximate and Ultimate Analyses of Solid Fuel Proximate Analyses (as received) Highvale Coal Volatile Matter 30.5 Fixed Carbon 42.1 Ash 12.2 Moisture 15.2 Ultimate Analyses (dry basis) Carbon 62.4 Hydrogen 3.6 Nitrogen 0.8 Sulphur 0.2 Oxygen (by difference) 18.7 Ash 14.3 Higher Heating Value (MJ/kg) 24.0 - 127 -+ U -1 Gl 9 8 7 6 5 4-3 2 -t + '0 "3 lb lb 2b 2^ 3b 3» 4Q 45 particle dLi-aw^ t<ju»m> Figure A7 Mass balance, as performed on a per channel basis for run BIO. Image analysis particle size distributions. Fines loss (below 15 microns) attributed to filter inefficiency. - 128 -Figure A8 Temperature data for Part I experiments. RUN # Dry bulb temperature Wet bulb temperature »C »C Bl 12 10 B2 15 11 B3 12 8 B4 11 7 B5 11 7 B6 16 12 B7 16 12 B8 16 12 B9 19 15 BIO 19 15 Bll 22 18 B12 18 15 B13 18 15 B14 18 15 B15 18 15 B16 18 15 B17 18 15 B18 18 15 B19* 18 15 B20 18 15 B21 22 19 B22 22 19 B23 22 19 B24 22 19 B25 22 19 B26 22 19 B27 22 19 B28 22 19 B29 22 19 B30 22 19 B31 22 19 B32 15 11 B33 15 11 B34 15 11 B35 15 11 - 129 -FIGURE A9 COLLECTION EFFICIENCY DATA &0N B4 RON B10 sixe efficiency sixe croDS % •icrons 1.7 0 0.74 7.9 0 2.23 8.11 35 3.71 8.32 32 5.2 8.53 39 6.68 8.75 35 8.16 8.98 32 9.65 9.21 44 11.13 9.45 46 12.62 9.(9 48 14.1 9.94 49 15.59 10.2 54 to 10.46 54 46.76 10.73 65 11.01 62 11.29 69 11.58 11 19 75 11. od 12.19 80 12.5 78 12.8! 86 13.IS 88 13.49 89 13.84 90 14.2 92 14.57 94 14.94 95 15.53 93 15.72 96 16.13 99 16.54 99 16.97 99 17.41 100 17.86 98 18.32 98 18.79 100 19.28 99 19.77 100 20.28 98 20.81 to 100 efficiency 100 S.4 - 130 -FIGURE A9 DATA SUMMARY FOR PARTS ( AND (I (CONTINUED) CHATHAM COLLECTION EFFICIENCY DATA DATA size MICRONS 7.32 7.51 7.70 7.90 8.11 8.32 8.53 8.75 8.98 9.2L 9.45 9.69 9.94 10.20 10.46 10.73 11.01 11.29 11.58 11.88 12.19 12.50 12.82 13.16 13.49 13.84 14.20 14.57 14.94 15.33 15.72 16.13 16.54 16.97 17.41 17.86 18.32 18.79 19.28 19.77 20.28 20.81 21.34 21.89 22.46 23.04 23.63 24.24 24.87 25.51 26.17 26.84 27.54 28.25 28.97 29.72 30.49 31.28 32.08 32.91 33.76 34.63 35.52 36.44 37.38 38.35 39.33 40.35 LOSS FRACTION 5.97E-05 5.97E-05 9.42E-05 6.59E-05 9.89E-05 9.89E-05 1.04E-04 1.08E-04 1.44E-04 1.51E-04 1.518-04 1.96E-04 2.45E-04 2.45E-04 2.97E-04 3.52E-04 3.96E-04 5.01B-04 5.65E-04 7.30E-04 7.79E-04 9.55E-04 1.12E-03 1.30E-03 1.65E-03 2.03E-03 2.21E-03 2.73E-03 3.31E-03 3.86E-03 4.70E-03 5.348-03 6.82R-03 7.94E-03 9.61E-03 1.16E-02 1.34E-02 1.61E-02 1.88E-02 2.19E-02 2.58E-02 2.85E-02 3.23E-02 3.69E-02 3.95E-02 4.26E-02 4.62E-02 4.73E-02 4.72E-02 4.92E-02 4.78E-02 4.718-02 4.74E-02 4.54E-02 4.34E-02 3.698-02 3.39E-02 2.69E-02 2.25E-02 1 CATCH FRACTION 0.008+00 0.008+00 0.008+00 0.00E+00 0.00E+00 O.OOE+00 0.00E+00 O.OOE+00 0.1 0.1 O.OOE+00 5.55E-06 0.00E+00 5.77E-06 O.OOE+00 6.21E-06 O.OOE+00 O.OOE+00 O.OOE+00 O.OOE+00 6.88E-06 O.OOE+00 O.OOE+00 O.OOE+00 O.OOE+00 7.99E-06 O.OOE+00 O.OOE+00 O.OOE+00 O.OOE+00 9.10E-06 O.OOE+00 O.OOE+00 9.99E-06 1.27E-06 O.OOE+00 O.OOE+00 1.09E-05 67E-02 . 228-02 8.94E-03 7.80E-03 6.05E-03 5.33E-03 3.38E-03 4.00E-03 1.46E-06 O.OOE+00 O.OOE+00 1.57E-06 0.00E+00 1.63E-06 25E-06 0.001+00 ..76E-06 1.65E-05 69E-06 93B-06 [.046-05 3.97E-06 4.13E-06 1.34E-05 6.53E-06 1.39E-05 t.148-05 1:718-05 1.208-05 2.28E-05 2.618-05 2.93E-05 3.29E-05 6.93E-0S 6.578-05 EFFICIENCY 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.22 0.00 0.19 0.78 0.00 0.17 1.58 0.34 0.19 1.00 0.38 0.42 1.40 0.81 1.85 1.92 3.37 3.20 7.95 11.84 14.7205 19.9943 37.4109 47.1316 CHATHAM COLLECTION EFFICIENCY DATA DATA siie MICRONS 41.39 42.46 43.55 44.68 45.83 47.01 48.23 49.47 50.75 52.05 53.40 54.78 56.19 57.64 59.13 60.65 62.22 63.82 65.47 67.16 68.89 70.67 72.49 74.36 76.28 78.25 80.26 82.34 84.46 86.64 88.87 91.17 93.52 95.93 98.41 100.95 103.55 106.22 108.96 111.77 114.66 117.62 120.65 123.76 126.96 130.23 133.59 137.04 140.57 144.20 147.92 151.74 155.65 159.67 163.79 168.01 LOSS FRACTION 3.10E-03 3.198-03 2.22E-03 2.13E-03 1.628-03 1.858-03 9.588-04 9.73E-04 1.418-03 1.028-03 4.248-04 6.508-04 1.338-03 1.148-03 1.178-03 7.168-04 9.86E-04 1.008-03 1.04E-03 2.398-03 5.438-04 1.688-03 1.148-03 0.008+00 1.518-03 1.558-03 1.89E-03 6.538-04 1.66E-03 2.408-03 1.058-03 1.088-03 0.008+00 2.278-03 1.56E-03 0.1 0.008+00 O.OOE+00 0.008+00 O.OOE+00 0.008+00 O.OOE+00 0.008+00 0.1 0.1 0.008+00 0.008+00 O.OOE+00 O.OOE+00 O.OOE+OO O.OOE+00 O.OOE+00 O.OOE+00 O.OOE+00 0.008+00 CATCH FRACTION 8.50E-05 1.148-04 1.07E-04 1.128-04 9.92E-05 9.768-05 1.318-04 2.148-04 2.298-04 2.958-04 3.678-04 4.128-04 5.108-04 6.558-04 8.238-04 9.928-04 1.288-03 1.638-03 2.128-03 2.828-03 3.398-03 4.538-03 6.188-03 7.788-03 9.238-03 1.078-02 1.278-02 1.398-02 1.57E-02 1.678-02 1.68E-02 1.938-02 1.838-02 1.908-02 1.76E-02 1.538-02 1.50E-02 2.068-02 2.528-02 3.10E-02 3.12E-02 3.61E-02 4.43E-02 4.97E-02 5.09E-02 4.60E-02 4.668-02 4.76E-02 5.638-02 6.14E-02 6.248-02 6.36E-02 4.278-02 3.38E-02 7.408-03 8.33E-03 EFFICIENCY % 49.40 62.84 60.67 69.80 68.17 73.40 76.48 91.12 91.53 90.59 94.29 97.81 97.30 95.77 97.08 97.50 98.79 98.79 98.98 99.21 98.49 99.74 99.41 99.68 100.00 99.69 99.74 99.70 99.91 99.78 99.69 99.88 99.87 100.00 99.72 99.78 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 100.00 - 131 r GAS CONCENTRATION DATA FOR ION 17 READING r/1 0! CO CO! S02 CB4 HOX % ppy % ppy % PPM 0.13 4 41 20 107 0 209 O.SO 3 48 21 297 0 214 0.S7 4 47 20 305 0 219 0.83 4 46 20 297 0 214 1.00 3 16S 20 132 0 209 O.SO 4 41 20 109 0 211 0.83 4 41 20 205 - 0 215 0.33 4 4! 20 130 o. 214 0.67 4 41 20 203 0 215 1.00 4 134 20 66 0 215 O.SO 4 31 20 S3 0 216 0.83 4 37 20 155 0 216 1.00 3 88 !l 176 0 203 GAS CONCENTRATION DATA FOR RON 18 reading t r/R 02 CO CO! S02 CB4 NOX % Dpi % ppa t ppi 0 4 36 20 0 0 196 0 4 34 19 0 0 207 0 5 41 19 0 0 201 1 4 46 20 0 0 207 1 3 54 21 0 0 201 I 2 134 22 0 0 190 Radial Gas Concentration profiles RON 5 PORT 2 r/R 02 CO C02 S02 CR4 NOX 1.0 4.1 35 18.4 50 0 198 0.8 4.1 20 18.4 116 0.002 195 0.7 4 20 18.4 113 0.002 195 0.5 4 19 18.5 114 0.002 190 0.3 3.9 20 18.5 112 0 190 0.2 3.9 22 18.4 104 0 185 0.0 3.9 24 18.4 101 0 180 0.0 3.9 26 18.3 90 0 180 0.3 4.3 21 18.4 112 0 18S 0.7 4.1 20 18.6 120 0 200 0.2 4.2 !l 18.5 116 0 200 1.0 4.3 39 18.3 40 0 200 0.5 4.3 !l 18.5 101 0 189 0.0 4.3 21 18.3 111 0 201 1.0 4.2 SO 18.5 40 0 20S Radial Gas Concentration profiles, Run 5 Port 4 r/R 02 CO C02 S02 cm NOX 0.4 4.7 25 17.8 78 0 201 0.6 4.5 33 18 89 0 204 0.8 4.2 34 18.1 41 0 202 0.9 4.4 40 17.8 34 0 201 0.9 4.3 42 18.1 10 0 200 1.0 4.3 51 17.8 10 0 200 Figure A9 Data summary for Parts I and II. Note: All gas concentration data in ppm except values for 02, CH4, and COz . - 132 -Radial Gas Concentration profiles, Ran 5 Port 5 02 CO CO! SO! CH4 NOX 5.7 24 17.1 40 0 1!0 4.6 25 17.9 48 0 195 4.5 22 17.8 77 0 201 4 19 17.9 87 0 202 4.5 18 18.1 89 0 203 4.6 17 18.1 . 90 0 203 4.5 19 18.1 90 0 204 4.7 20 18.1 73 0 205 >S CONCENTRATION DATA RON 5, PORT 6 til 02 CO C02 S02 CH4 NOX 0.33 3.58 35.76 19.78 0.00 0.00 195.59 0.50 3.91 33.53 19.34 0.00 0.00 206.76 0.67 4.58 41.35 19.22 0.00 0.00 201.18 0.83 4.25 45.82 20.01 0.00 0.00 206.76 1.00 2.91 53.65 20.90 0.00 0.00 101.18 0.50 2.12 134.12 21.57 0.00 0.00 190.00 GAS CONCENTRATION DATA RON 5, PORT 8 I til 02 CO CO! SO! CH4 NOX 1 0.46 4.8 30 17.7 90 0 170 2 0.67 4.3 24 17.9 96 0 !0! 3 0.83 4.5 !7 17.8 88 0 203 4 1.00 4.6 59 17.5 !! 0 145 ( RON t 6 RADIAL GAS CONCENTRATION DATA r/R O! CO CO! S02 CU4 NOX 1.00 4.10 63.00 17.30 0.00 0.00 175.00 0.83 4.!0 39.00 17.00 0.00 0.00 225.00 0.67 4.20 32.00 16.90 0.00 0.00 220.00 0.33 4.40 31.00 17.20 0.00 0.00 225.00 0.17 5.20 29.00 16.40 0.00 0.00 230.00 0.00 4.00 40.00 17.60 0.00 0.00 230.00 0.00 2.80 48.00 18.40 0.00 0.00 230.00 1.00 NA 78.00 18.40 0.00 0.00 220.00 GAS CONCENTRATION DATA FOR RON 6 RON t t/l 02 CO C02 S02 CB4 2 1.00 4.6 23 16.8 50 0.004 6 0.29 5 23 16.4 120 0.003 8 0.50 4.6 18 16.9 140 0.003 10 0.79 5.5 16 16.1 135 0.003 12 0.13 5.1 22 16.4 90 0.003 14 1.00 3.5 13 17.8 140 0.003 16 0.29 4.2S 16 17.3 140 0.003 Figure A9 Data summary for Par ts I and Note: All gas concentration data in 1 for 02, CH* , and C02 • ppm except values 


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics

Country Views Downloads
China 29 8
United States 27 8
Turkey 12 0
Latvia 4 0
Colombia 4 0
Russia 4 0
Malaysia 1 0
Poland 1 0
United Kingdom 1 0
City Views Downloads
Unknown 24 0
Beijing 23 0
Ashburn 11 0
Mountain View 6 1
Shenzhen 6 8
San Francisco 5 0
Saint Petersburg 4 0
Wilmington 2 0
Cheyenne 1 0
Shah Alam 1 0

{[{ mDataHeader[type] }]} {[{ month[type] }]} {[{ tData[type] }]}
Download Stats



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items