UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Solids motion and mixing in high-density circulating fluidized beds Kirbaş, Görkem 2004

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata

Download

Media
831-ubc_2004-931404.pdf [ 11.46MB ]
Metadata
JSON: 831-1.0058657.json
JSON-LD: 831-1.0058657-ld.json
RDF/XML (Pretty): 831-1.0058657-rdf.xml
RDF/JSON: 831-1.0058657-rdf.json
Turtle: 831-1.0058657-turtle.txt
N-Triples: 831-1.0058657-rdf-ntriples.txt
Original Record: 831-1.0058657-source.json
Full Text
831-1.0058657-fulltext.txt
Citation
831-1.0058657.ris

Full Text

SOLIDS MOTION AND MIXING IN HIGH-DENSITY CIRCULATING FLUIDIZED BEDS by G O R K E M K I R B A § B.Sc, Middle East Technical University, 1998 M . S c , Middle East Technical University, 1999 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE F A C U L T Y OF G R A D U A T E STUDIES Department of Chemical and Biological Engineering We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH C O L U M B I A August 2004 © Gorkem Kirbas, 2004 ABSTRACT This study was directed to the hydrodynamics of vertical risers, such as those used in catalytic cracking and other chemical reactors. The work was carried out in risers of 0.2 m and 0.076 m diameters with FCC particles of mean diameter 70 um, primarily using optical fibre probes and solid tracer response measurements. The flow characteristics of high-density gas-solids systems were investigated in detail by measuring the local transient particle velocity, solids concentration and solids fluxes at different radial locations and axial positions in the 0.2 m diameter riser operating over a wide range of net solids circulation fluxes (120 kg/m2s < G s < 350 kg/m2s) and superficial gas velocities (5 m/s < U g < 8 m/s). These data were used to generate radial and axial profiles of solids hold-up, particle velocity and solids flux and to show how the profiles were influenced by the operating conditions. The flow behaviour in the riser was observed to be a function of height as well as the operating conditions. With a change in operating conditions, the local particle velocity and solid fluxes in the relatively dilute core region of the riser changed more significantly than those in the wall region. The opposite trend was observed with respect to the local solid hold-up distributions. The effect of riser diameter on the local solids flow structure in high-density circulating fluidized bed systems was investigated by comparing the findings of this study with those reported by previous researchers in 76 mm and 104 mm diameter CFB risers. The increase in riser diameter in this study was observed to result in lower cross-sectional average solids holdup at the same superficial gas velocity and net circulation flux under high-density conditions. The local particle velocity in the central region was higher for the larger diameter riser, while the difference in the annulus was insignificant. The time-mean solids fluxes were found to be nearly equal in the central top section of the riser. The difference between the two sets of measurements increased closer to the wall. In the top section of the riser, in the wall region, the time-mean solids fluxes were larger for the smaller diameter riser. The opposite trend was observed for the bottom section of the riser in the wall region. In order to investigate the flow structure in the wall region, an annular downflow layer thickness 5 (= 1- rc/R) was introduced, where rc was defined as the radial position where the time mean net solids flow switches from being upwards to downwards. The extent of ii correlation between particle velocity and solids hold-up fluctuations was examined by investigating their covariance. The increase in covariance under high-density conditions indicated the importance of utilizing flux measurements, rather than particle velocity measurements, in the detection of flow direction in the wall region. When solids circulation flux was increased together with the superficial gas velocity, annular downflow was observed to decrease. Solids mixing and motion were investigated in the 0.076 m diameter high-density circulating fluidized bed riser utilizing phosphorescent-coated FCC particles as the tracer. Special tracer coating, injection and detection techniques were developed for this part of the work. An axial dispersion model was utilized to determine axial solids dispersion coefficients, and the results are interpreted with the help of hydrodynamic data obtained in the same column. This study confirms that axial solids dispersion decreases when the dense suspension upflow regime is reached, i.e. when net downward flow of particles disappears at the wall. iii TABLE OF CONTENTS ABSTRACT ii TABLE OF CONTENTS iv LIST OF TABLES vii LIST OF FIGURES viii ACKNOWLEDGEMENT xv CHAPTER 1 INTRODUCTION 1 1.1 Research objective and principal tasks 4 1.2 Organization of thesis 4 CHAPTER 2 EXPERIMENTAL SET-UP 6 2.1 Fluid coker cold model (with riser) 6 2.2 Dual-loop high-density circulating fluidized bed (HDCFB) 10 2.3 Pressure measurements and apparent solids hold-up calculations 19 CHAPTER 3 OPTICAL PROBE MEASUREMENT SYSTEM 22 3.1 Particle velocity measurements 26 3.2 Solids concentration measurements 30 3.2.1 Elimination criteria 42 3.2.2 Cross-sectional averaging of the radial profiles of particle 46 velocity, solids hold-up and flux CHAPTER 4 SOLIDS HOLD-UPS AND MOTION IN HIGH-DENSITY CFB RISERS 48 iv 4.1 Local solids hold-up profiles 48 4.2 Local particle velocity profiles 67 4.3 Local solids flux profiles 79 4.4 Solids motion in the wall region 88 CHAPTER 5 SOLIDS MIXING STUDIES IN HDCFB UNIT 105 5.1 Introduction 105 5.2 Measurement system 114 5.2.1 Tracer particles 114 5.2.2 Tracer activation and injection system 118 5.2.3 Tracer detection system 124 5.3 Data treatment procedure 128 5.4 Analysis of data 131 5.5 Modelling 145 5.5.1 Axial dispersion model 145 5.5.1.1 Initial and boundary conditions 146 5.5.1.2 Solution of Eq. (5.5) 148 5.5.1.3 Measurement of V p 150 5.6 Results and discussion of the model 151 CHAPTER 6 CONCLUSIONS 160 6.1 Summary of major conclusions 160 6.2 Recommendations for future work 162 NOMENCLATURE 163 v REFERENCES 166 APPENDIX A - Procedure of preparation of the tracer particles used in solids 183 mixing studies APPENDIX B - Properties of the light sources used in solids mixing studies 185 vi LIST OF TABLES Table 2.1 Size distribution of FCC particles from sieve analysis 10 Table 3.1 Effect of cut-off and variance interval constant, a0 values on the 45 percentage of eliminated data and on average percentage deviations Table 4.1 Properties of high-density risers and their operating conditions used in 53 comparisons Table 5.1 Summary of experimental studies on solids mixing in fluidized beds 106 Table 5.2 Optical properties of phosphorescent pigment particles 115 Table 5.3 Properties of tracer particles used in solids mixing experiments 116 Table A. 1 Proportions of the ingredients used in tracer particle preparation 183 Table B.l The properties of the light sources used in the preliminary tests 185 and solids mixing experiments (provided by the suppliers) vii LIST OF FIGURES Figure 2.1 Schematic of semi-cylindrical fluid coker cold model 7 Figure 2.2 Schematic showing the locations of the pressure ports on the semi- 9 cylindrical coker and its riser Figure 2.3 Schematic diagram of high density CFB unit 11 Figure 2.4 Change of standard deviation of superficial gas velocity at 13 G s = 250 kg/m2s Figure 2.5 Change of standard deviation of overall solids circulation fluxes 14 at U g = 6 m/s Figure 2.6 Pressure profile in FfDCFB unit [from Liu, 2001] 16 Figure 2.7 Schematic of bottom part of the FIDCFB unit with "sore thumb" 17 (patent pending) installed Figure 2.8 Comparison of pressure difference between Pi, measured at 18 bottom of 2nd downcomer, and P2, measured at exit of U-valve, before and after installation of "sore thumb" (Ug = 6 m/s) Figure 2.9 Apparent (solid symbols) and actual (open symbols) solids hold-up 21 profiles at U = 8 m/s for G s = 181 and 425 kg/m2 as determined by Issangya (1998) Figure 3.1 Details of 3-fibre optical probe 23 Figure 3.2 Schematic of blind zone elimination with the glass window and the 24 measurement region at the tip of the optical probe Figure 3.3 Optical fibre probe particle velocity/concentration measurement 25 system Figure 3.4 Calibration of effective separation distance, L e by rotating disk 27 method Figure 3.5 Algorithm of the main computer program hydrodynamics.m 28 Figure 3.6 Algorithm of the calib.m program 29 Figure 3.7 FCC+coke mixture calibration system 34 viii Figure 3.8 Calibration curves obtained by using FCC+coke mixture in slowly 35 moving packed bed calibration system Figure 3.9 Comparison between solid hold-up recorded by optical fiber probe 36 (Channel 2) and integrated across column cross-section and apparent solid hold-up values obtained from pressure gradients, i.e. Eq. (3.6) Figure 3.10 Change of calibration curve by using different power n values 37 Figure 3.11 Parity plot of solids fraction from pressure drop (Eq. (3.6)) and 38 optical fibre probe measurements for different values of the index n = 0.77, 1.00, 1.50, 2.50 Figure 3.12 Parity plot of solids fraction from pressure drop Eq. (3.6) and 39 optical fibre probe measurements using the calibration equation obtained by (Eq. (3.5)) Cui et al. (2001) Figure 3.13 Calibration equation iteration program algorithm 40 Figure 3.14 Final response curve for calibration of optical probe for solid 41 concentration Figure 3.15 Parity plots obtained by using cut-off value = 0.5 in elimination 43 criteria Figure 3.16 Concentric rings and their dimensions used in the calculation of 47 cross-sectional averaging values Figure 4.1 Axial profiles of apparent solids hold-up for various solids 50 circulation fluxes at U g = 6 m/s Figure 4.2 Axial profiles of apparent solids hold-up for G s = 250 kg/m2s and 51 various superficial gas velocities Figure 4.3 Effect of solids circulation flux on apparent solids hold-up for 52 various heights along the riser of the fluid coker cold model at U g = 6 m/s Figure 4.4 Comparison of apparent solids hold-ups measured in the 0.203 m 55 riser of this study with those measured by Issangya (1998) and Liu (2001) in 0.076 m diameter riser of the dual-loop FfDCFB unit [Ug = 6 m/s] Figure 4.5 Comparison of apparent solids hold-ups measured in this study 56 (Driser = 0.203 m) with those measured by Liu (2001) (Dnser = 0.076 m) and Malcus (2000) (DriSer = 0.14 m) ix Figure 4.6 Reproducibility of local time-mean solids hold-up measurements 59 performed by 3-fibre optical probe at z = 0.76 m and z = 4.42 m for G s = 250 kg/m2s and U g = 6 m/s Figure 4.7 Comparison of axial solids hold-up profiles measured by pressure 60 transducers (open symbols) and by optical fibre probe (closed symbols): (a) U g = 6 m/s, G s = 250 kg/m2s, (b) U g = 6 m/s, G s = 330 kg/m2s, (c) U g = 5 m/s, G s = 250 kg/m2s, (d) U g = 8 m/s, G s = 250 kg/m2s Figure 4.8 Radial profiles of time-mean solid hold-up and their time-traces 61 obtained using optical fibre probe under (a) high-density condition: G s = 252 kg/m2s, U g = 6 m/s, z = 0.76 m; (b) low-density condition: G s = 255 ks/m2s, U g = 6 m/s, z = 4.42 m Figure 4.9 Radial profiles of local time-mean solids hold-up at different axial 63 locations for U g = 6 m/s and various solids circulation fluxes Figure 4.10 Radial profiles of local time-mean solids hold-up at different axial 64 locations for U g = 6 m/s and G s = 330 kg/m2s Figure 4.11 Radial profiles of local time-mean solids hold-up at different axial 65 locations for G s = 250 kg/m2s and various superficial gas velocities Figure 4.12 Local solid hold-up traces with corresponding probability 66 distribution plots for six radial positions at z = 0.76 m with U g = 6 m/s and G s = 341 kg/m2s. (Sampling rate = 280 kHz , sampling time = 30 s) Figure 4.13 Radial profiles of standard deviation of local solids hold-up at 68 different axial locations for U g = 6 m/s and various solids circulation fluxes Figure 4.14 Comparison of radial profiles of time-average and concentration- 70 weighted time-mean particle velocities at z = 0.76m for U g = 6 m/s and G s = 341 kg/m2s Figure 4.15 Reproducibility of local concentration-weighted time-mean particle 72 velocity measurements performed by 3-fibre optical probe at z = 0. 76 m and z = 4.42 m for at G s = 250 kg/m2s and U g = 6 m/s Figure 4.16 Local concentration-weighted time-mean radial particle velocity 73 profiles at different axial locations for U g = 6 m/s and various solids circulation fluxes x Figure 4.17 Local concentration-averaged time-mean radial particle velocity 74 profiles at different axial locations for G s = 250 kg/m 2s and various superficial gas velocities Figure 4.18 Local concentration-averaged time-mean radial particle velocity at 75 different axial locations for U g = 6 m/s and G s = 330 kg/m 2s Figure 4.19 Comparison of local time-mean particle velocity profiles obtained 77 in this study (D r i s e r = 200 mm) and in the FfDCFB riser of L i u (2001) (Dnser = 76 mm) Figure 4.20 Local particle velocity traces and corresponding probability 78 distribution plots at six radial positions for z = 0.76 m, U g = 6 m/s G s = 341 kg/m 2s. (Sampling rate = 280 kHz , sampling time = 30 s) Figure 4.21 Reproducibility of local time-mean solids flux measurements 81 performed by utilizing 3-fibre optical probe at z = 0.76 m and z = 4.42 m for at G s = 250 kg /ms and U g = 6 m/s Figure 4.22 Radial profiles of local time-mean solids flux at different heights 82 for U g = 6 m/s and various net solids circulation fluxes Figure 4.23 Radial profiles of local time-mean solids fluxes at different 83 heights for U g = 6 m/s and G s = 330 kg/m 2s Figure 4.24 Radial profiles of local time-mean solids flux at different heights 85 for G s = 250 kg/m 2s and different superficial gas velocities Figure 4.25 Comparison of radial profiles of local time-mean solids flux 86 obtained in this study (Dnser = 200 mm) with those determined in the H D C F B riser by L i u (2001) (D r i s e r = 76 mm) Figure 4.26 Local solids flux traces with corresponding probability distribution 87 plots at six radial positions at z = 0.76 m for U g = 6 m/s and G s = 341 kg/m 2s. (Sampling rate = 280 kHz , sampling time = 30 s) Figure 4.27 Comparison of 8v P and 8Gs at different heights for U g =6 m/s and 90 three different net solids circulation fluxes Figure 4.28 Effect of operating conditions on radial profiles of covariance term, 92 i.e. V p 8 s for U g = 6 m/s and two values of solids circulation fluxes Figure 4.29 Radial profiles of time-mean solids circulation fluxes with 94 corresponding boundaries of annular wall layer and axial apparent solids hold-up profile for U g = 6 m/s and G s = 125 kg/m 2s x i Figure 4.30 Radial profiles of time-mean solids circulation fluxes with 95 corresponding boundaries of annular wall layer and axial apparent solids hold-up profile for U g = 6 m/s and G s = 330 kg/m2s Figure 4.31 Change of annular wall layer thickness with (a) solids circulation 97 flux for U g = 6m/s (b) superficial gas velocity for G s = 250 kg/m2s Figure 4.32 Variation of (a) apparent solids hold-up, (b) time-mean solids flux 98 at r/R = 1, and (c) annular wall layer thickness, 5Gs, with G s at U g = 6 m/s and z = 0.76 m Figure 4.33 Variation of (a) apparent solids hold-up, (b) time-mean solids flux 99 at r/R = 1, and (c) annular wall layer thickness, 5Gs, with G s at U g = 6 m/s and z = 1.27 m Figure 4.34 Variation of (a) apparent solids hold-up, (b) time-mean solids flux 100 at r/R = 1, and (c) annular wall layer thickness, 5G s, with G s at U g = 6 m/s and z = 3.1 m Figure 4.35 Adoption of the flow regime map developed by Kim et al. (2004) 102 to the data collected in this study Figure 4.36 Comparison of annular layer wall thicknesses obtained in this study 103 (Driser = 200 mm) and in HDCFB riser by Liu (2001) (D r i s e r = 76 mm) for U g = 6 m/s and similar net solids fluxes Figure 5.1 SEM photograph of tracer particles 116 Figure 5.2 SEM photographs of (a) phosphorescent pigment particles 117 (b) FCC particles Figure 5.3 Schematic diagram of the high-density circulating fluidized bed 119 unit with the injection system and the detection ports (Numbered from 1 to 4). For details of the activation-injection system see Figures 5.4 and 5.5 Figure 5.4 Tracer activation-injection system (see Figure 5.3 for its location 120 relative to the HDCFB system) Figure 5.5 Configuration of UV light bulbs fastened on the aluminium 121 reflective plate which was rolled into a cylinder around the quartz injector glass tube (see also Figure 5.4) Figure 5.6 Effect of light source and exposure time on the intensity of emitted 122 light recorded right after the light was extinguished Figure 5.7 Tracer pulse input recorded at point 1 in Figure 5.3 123 xii Figure 5.8 Tracer decay curve calibration system 125 Figure 5.9 Decay calibration curve averaged from 20 determinations 126 Figure 5.10 Experimental set-up used to test the change of intensity of the 127 emitted light with the concentration of tracer in FCC + tracer mixtures Figure 5.11 Effect of concentration of tracer particles in tracer + FCC mixtures 127 on the measured emitted light intensity Figure 5.12 Comparison of raw data and data after 5 pt-FFT smoothing process 129 (Signal recorded in HDCFB riser at z = 5.48 m and r/R = 0.0 operating at G s = 450 kg/m2s, U g = 6m/s) Figure 5.13 Comparison of data after baseline adjustment and data after decay 130 curve correction (Signal recorded in FfDCFB riser at z = 5.48 m and r/R = 0.0 operating at G s = 450 kg/m2s, U g = 6m/s) Figure 5.14 Comparison of data after decay curve correction and after 30 pt-FFT 130 smoothing process (Signal recorded in HDCFB riser at z = 5.48 m and r/R = 0.0 operating at G s = 450 kg/m2s, U g = 6m/s) Figure 5.15 Comparison of two replicate measurements and their average 131 (Signal recorded in HDCFB riser at z = 5.48 m and r/R = 0.0 operating at G s = 450 kg/m2s, U g = 6m/s) Figure 5.16 Effect of solids circulation rate on the concentration profiles at 133 different net axial locations for U g = 4m/s Figure 5.17 Effect of solids circulation rate on the concentration profiles at 134 different axial locations and at r/R = 0 for U g = 6 m/s Figure 5.18 Effect of solids circulation rate on the concentration profiles at 135 different axial locations and at r/R = 0 for U g = 8 m/s Figure 5.19 Axial profiles of apparent solids hold-up for various solids 136 circulation fluxes and at U g = 4, 6 and 8 m/s Figure 5.20 Radial profiles of local time-mean solids flux measured by Liu 138 (2001) at z = 4.2 m for various net solids circulation fluxes: (a) U g = 4 m/s; (b) 6 m/s; (c) 8m/s Figure 5.21 Effect of superficial gas velocity on the concentration traces 139 at different axial locations and at r/R = 0 for G s = 250kg/m2s xin Figure 5.22 Effect of superficial gas velocity on the concentration traces 140 at different axial locations at r/R = 0 for G s = 350kg/m2s Figure 5.23 Effect of superficial gas velocity on the concentration traces 141 at different axial locations and at r/R = 0 for G s = 450kg/m2s Figure 5.24 Tracer concentration traces at three different radial positions and 143 axial locations for U g = 6m/s and G s = 110 kg/m s Figure 5.25 Tracer concentration traces at three different radial positions and 144 axial locations for U g = 6 m/s and G s = 450 kg/m2s Figure 5.26 The grid distribution together with the assigned particle velocities 147 and dispersion coefficients used in the solution of Eq. (5.9) Figure 5.27 Comparison of experimental concentration values with those from 152 axial dispersion model at z = 1.55 m and z = 5.48 m for G s = 450kg/m2s, U g = 6 m/s Figure 5.28 Comparison of experimental concentration values with those from 153 axial dispersion model at z = 1.55 m and z = 5.48 m for G s = 450kg/m2s, U g = 8 m/s Figure 5.29 Effect of superficial gas velocity on axial solids dispersion 154 coefficients for G s = 110, 250, 350 and 450 kg/m2s Figure 5.30 Effect of solids circulation rate on axial solids dispersion 155 coefficients for U g = 4, 6 and 8 m/s Figure 5.31 Flow regime map 157 Figure 5.32 Comparison of axial solids dispersion coefficients obtained in this 159 study with the axial gas dispersion coefficients obtained by Liu (2001) in the same unit (The Liu values are those based on the simple one-dimensional gas dispersion model) xiv X V ACKNOWLEDGEMENTS I would like to extend my sincere gratitude to all of those who gave their support and encouragement during the completion of this thesis. First and foremost, I would like to express my sincerest thanks to my supervisors, Dr. J.R. Grace, Dr. C.J. Lim and Dr. X.T. Bi for their invaluable and patient guidance and their continued encouragement throughout my studies. I am especially indebted to Dr. J.R. Grace for his unwavering efforts in accommodating all of the administrative demands that were placed upon him. This work was made possible by the financial support of the Higher Education Council of Turkey and NSERC. I sincerely thank both of these organisations for making this work possible. Dr. Sung Wong Kim was instrumental in turning the normally tedious and frustrating act of running experiments into something memorable. I extend my gratitude to Dr. Xuqi Song whose expertise helped to bring the very challenging solids mixing experiments to fruition. It was a blessing to be surrounded by the wonderful members of the fluidization research group (past and present), all of whom contributed to the success of this project in their own unique way. I appreciate all of the technical support provided by the support staff and faculty of the department. Special thanks goes to Peter, Graham, Qi, Horace, Alex, Amber, Helsa, Lori and Darcy. xvi I acknowledge Dr. Bruce Bowen, Dr. Chad Bennington and Dr. Jinzhong Liu for their valuable guidance and insightful discussions. How do I begin to thank Naoko? Colleague, mentor, friend, and virtually family, she was a constant source of inspiration, motivation, hope and love, who never doubted or stopped believing in me. Aliye, Emre and Rubab were tireless cheerleaders who supported me through the peaks and valleys of this work. Even from afar, Cagri and Pelin kept me going. Monica, Poupak, Ana and Abba never failed to provide a smile when I needed it. And all of my friends at St. John's College who added colour to my life. Thank you all. I cannot thank Stephen enough. His extraordinary patience, support and understanding were critical to the completion of this work. Thank you for standing by me during the ups and downs of this roller coaster ride, and for always putting a smile on my face even during my darkest hours. To my parents and my brother Onur, thank you for helping me to keep things in perspective, and for supporting me unconditionally even when I seemed to be in over my head. xvii Chapter I Introduction 1 CHAPTER 1 INTRODUCTION Circulating Fluidized Beds (CFB) have been investigated extensively by researchers due to their practical applications in catalytic and gas-solid reactions. Industrial processes that utilise CFB technology can be divided into gas-solids reaction processes, gas-phase catalytic reaction processes and physical processes like drying. The first group, which includes processes such as combustion of various fuels (e.g. coal) and alumina calcination, usually have low reaction rates and therefore do not necessarily require high gas velocities or high solids circulation rates. However, catalytic gas-phase reactions, like fluid catalytic cracking of petroleum, Fischer-Tropsch synthesis, and partial oxidation of butane, require a relatively high gas velocity in the riser to minimize gas backmixing, and give limited catalyst deactivation. Their higher reaction rates also call for higher gas velocities because of the desired short contact time between the gas and solids. Extensive studies have been conducted on CFB systems operating at relatively low suspension densities (rarely exceeding 3% solids by volume in the fully developed region), low gas velocities (< 8 m/s) and modest solids circulation rates (G s < 100 kg/m2s), conditions typical of CFB reactors with gas-solid reaction processes. However, the hydrodynamics of CFB reactors used for gas-phase catalytic reaction processes with high solids hold-ups have received significant attention only in the last decade. Several research groups, those at PSRI (Karri and Knowlton, 1998, 1999), Erlangen University (Grassier and Wirth, 1999), University of Saskatchewan (Malcus et al., 2002), and the University of Western Ontario (Parsinnen and Zhu, 2001), have built pilot-scale high-density CFB units in their laboratories and conducted numerous experiments to delineate and understand the flow structure in this type of reactors. At relatively low circulation fluxes, flow of a gas-solid suspension in a riser is characterized by a core-annulus flow pattern where a rapidly rising, dilute suspension in the core is surrounded by a slowly falling dense suspension in the region adjacent to the walls. Investigators (e.g. Rhodes and Laussmann, 1992) have also reported that there is a dense region at the bottom of the riser, leading smoothly to a lean region above. In these two Chapter I Introduction 2 regions, flow properties like particle velocity, solids concentration, solids flux distributions, and gas and solids mixing, have been found to differ substantially. These flow properties were observed during experiments carried out at relatively low solids fluxes (Gs < 200 kg/m2s) and overall volumetric solids concentrations less than roughly 10% (Yerushalmi and Cankurt, 1979; Li and Kwauk, 1980; Werther, 1994; Zhu and Bi, 1995). Although catalytic gas phase reactors operate at high solids circulation fluxes (Gs > 500 kg/m2s), high velocities (~ 8 m/s or more) and high solids hold-ups (~ 7-25%), most academic researchers have only been able to achieve high-density conditions at the bottom of the risers due to the limitations of the experimental units. Bi and Zhu (1993) pointed out some factors which have prevented study of more dense systems. From a pressure balance simulation around a CFB loop, they proposed a means of achieving high density conditions in the riser, involving a second riser to lift the solids from the first riser to a higher level, facilitating a taller downcomer and enabling higher solids circulation fluxes, while imposing a small penalty in terms of blower capacity. Based on this concept, Issangya (1998) designed, constructed and operated a dual-loop high density CFB unit (FfDCFB) at the University of British Columbia. This unit could achieve solids hold-ups as high as 15 to 25% by volume averaged over the entire riser, and FCC net circulation fluxes of 500 kg/m2s and beyond for superficial air velocities up to 8 m/s, well beyond the capabilities of most other research units. The most striking observed feature was that at high-density conditions, there was no net downward flow of particles along the walls of the riser. This meant that the core-annulus flow structure with downflow at the wall observed in previous studies was no longer applicable for high-density systems. After extensive investigation, they concluded that the flow regime in their high-density system differed significantly from the previously recognized flow regimes, including fast fluidization. They named this flow regime 'Dense Suspension Upflow (DSU)' (Grace et al., 1999) and defined it as 'a flow regime where there is net upflow of solids across the entire riser, strong interactions between particles, gas velocity/solids flux conditions beyond any of the types of choking, and overall volumetric solids concentration of about 0.07 to 0.25, with little axial variation.' The physical configuration associated with this mode of operation leading to this flow regime was called a 'high-density circulating fluidized bed'. Later, it was found that a Chapter 1 Introduction 3 transition from dense suspension upflow regime to fast fluidization regime could occur part way up a circulating fluidized bed riser (Liu, 2001; Kim et al., 2004). In order to achieve better understanding of the characteristics of this flow regime, further investigations were carried out by Liu et al. (1999, 2002, 2003a, 2003b) on the same unit. Local particle velocity, solids concentration and solids fluxes were measured, while gas residence time distributions (RTD) were investigated by comparing RTD curves obtained by stimulus-response experiments with predictions from a radially non-uniform dispersion model in which the radial profiles of voidage and gas velocity were taken into account. Axial gas dispersion coefficients were interpreted with the help of the hydrodynamic results obtained in the same column. This study confirmed that the axial gas dispersion decreased when dense suspension upflow conditions were reached, i.e. when the net flow of particles at the wall changed from upwards to downwards. Several other research groups, e.g. those at PSRI (Karri and Knowlton, 1998, 1999), Erlangen University (Grassier and Wirth, 1999), University of Saskatchewan (Malcus, 2000; Malcus and Pugsley, 2001; Malcus et al., 2002) and the University of Western Ontario (Parsinnen and Zhu, 2001), have also studied HDCFB systems with broadly similar results. Although the earlier work has been very helpful, there are still many unresolved questions. For example, the effect of the riser diameter on the hydrodynamics of the flow regime have not been investigated sufficiently. To fully understand this flow regime, the solids mixing behaviour should also be determined. In catalytic cracking reactors, catalyst particles are subject to deactivation due to aging, sintering or deposition of coke on the catalyst surface (Fogler, 1992). Since catalyst activity has a crucial effect on the rate of mass transfer and affects the selectivity and conversion in the bed, it is important to know the residence time distribution of the particles and their mixing mechanisms. The movement of solids also influences heat transfer between the suspension and the riser wall or internal heat transfer surfaces, as well as erosion rates of surfaces. Consequently, reliable knowledge of solids motion and the mechanism of solids mixing would be helpful in design. However, few data have been reported on solids mixing (Wei et al. 1994, 1995; Harris, 2002) in CFB reactors, and no results have been reported for high-density CFB conditions. This is an area requiring investigation. Chapter I Introduction 4 /./ Research objective and principal tasks Although an increasing number of studies have been performed, there are still gaps in information. Therefore, it was the principal objective of the research presented in this thesis to fill in the missing parts. To achieve this, the main tasks undertaken were as follows: Simultaneous measurement of local transient particle velocity, solids concentration and solids flux distributions in a 0.2 m diameter riser of a high-density circulating fluidized bed using the 3-fibre optical probe developed by Liu (2002) (see also Liu et al., 2003a). - Detailed experimental investigation of the flow pattern in the region adjacent to the walls of the riser of a high-density circulating fluidized bed. Study of the effect of riser diameter on the local solids structure in high-density circulating fluidized bed systems. - Development of a simple and accurate experimental technique with fast response to investigate solids motion and mixing in the dense suspension upflow regime. Measurement of transient local solid tracer concentration distributions to obtain information on solids mixing and dispersion in high-density risers. 1.2 Organization of thesis Chapter 1 presents a brief introduction to the current work and a brief overview of high-density circulating fluidized bed systems. Detailed literature reviews for each specific topic are distributed among later chapters. Chapter 2 summarizes the detailed design of the two experimental CFB units used in this work, together with general measurements such as superficial gas velocity, solids circulation flux and pressure profiles. Chapter 3 provides information on the 3-fibre optical probe used for simultaneous measurement of local transient particle velocity, solids concentration and solids flux. A detailed discussion on the calibration of optical fibre probes is also included. Chapter I Introduction 5 Chapter 4 presents the experimental results of measurements conducted using the 3-fibre optical probe. The flow pattern in the riser of 0.2 m diameter of the fluid coker cold model under wide range operating conditions is studied in detail. The effects of riser diameter on the local solids structure of high-density circulating fluidized bed systems are also investigated. Finally, the flow pattern in the region adjacent to the walls of the riser is explored. Chapter 5 begins by describing the measurement system developed for solids mixing studies. Radial and axial profiles of tracer concentration measured over a wide range of high density/high flux operating conditions are reported next. An axial dispersion model is then proposed and used to determine solids axial dispersion coefficients. The conclusions of this work, together with some recommendations for future studies on FfDCFB systems, appear in Chapter 6. Chapter 2 Experimental Set-up 6 CHAPTER 2 EXPERIMENTAL SET-UP Two experimental set-ups were used in this study: the semi-cylindrical fluid coker cold model sponsored by Syncrude Canada (see Knapper et al., 2002) and the dual-loop high-density circulating fluidized bed (HDCFB) (see Issangya, 1998). Descriptions of both apparatuses and the measurement methods for basic hydrodynamic properties such as overall solids circulation flux and superficial gas velocity are presented in this chapter. 2.1 Fluid coker cold model (with riser) The first set of experiments were performed in the cylindrical riser of the semi-cylindrical fluid coker. The unit designed, constructed and commissioned for a study of conditions relevant to Syncrude's commercial fluid cokers is shown in Figure 2.1. The unit consists of a plexiglas half-column reactor section, a plexiglas full-column riser culminating in a 45° converging nozzle impinging on a baffle separator, 6 primary cyclones in parallel, one secondary cyclone and two baghouses. The riser is 0.20 m in diameter and 5.90 m high. The details of the experimental unit can be found in Knapper (2001). The particles are stored in the main reactor section. They are fluidized by numerous nozzles located in the reactor and stripping sections. The particles flow downwards from this section through a standpipe to a U-bend. The overall solids circulation flux is controlled by a pinch valve located at the bottom of the standpipe. After passing through the U-bend, solids are carried through the riser, and then pass through a venturi constriction to enter the impingement box where many particles are disengaged by impingement on a baffle. Most of the remaining entrained particles are separated by the six parallel primary cyclones located at the exit of the impingement box. They are then returned to the dense bed region of the semi-cylindrical main reactor through return diplegs. A secondary cyclone and two bag filter houses in parallel capture any particles remaining in the exiting air stream. Chapter 2 Experimental Set-up 7 Figure 2.1 Schematic of semi-cylindrical fluid coker cold model Chapter 2 Experimental Set-up 8 An orifice meter measures the gas flowrate to the riser. The pressure drop across the orifice meter and the flow rates of the aeration gases are used to calculate the superficial gas velocity, U g . The overall solids circulation flux through the riser, G s , is calibrated by monitoring the pressure drop across the venturi constriction at the top of the riser, while simultaneously measuring the solids mass flux in the standpipe using a fibre optical velocimeter probe to determine the solids void fraction and velocity (Knapper et al., 2002) The whole unit is transparent and well instrumented to allow for visual observation and a range of experimental measurements. To measure pressure profiles and apparent cross-sectional average voidages along the riser, differential and absolute pressure transducers are used. The pressure taps are mounted flush with the wall of the column and are covered with 38 urn mesh stainless steel screens glued over the ports to prevent solids from entering the pressure-sensing lines. The signals from the pressure transducers (Omega, PX140) are fed to a computer via a data acquisition board using a software written by C. J. Lim using Visual Basic® programme. The sampling is at a frequency of 0.10 Hz. The locations of the pressure ports are depicted in Figure 2.2. A reconditioned mobile diesel-driven Sulair Model 8V-927A compressor supplies air with a maximum capacity of 0.66 m3/s. Supplementary air from the building compressor provides an additional 0.06 m3/s. The particles used throughout this project were spent fluid catalytic cracking catalyst (FCC) particles of mean diameter 70 am and density 1700 kg/m3 (reported by the supplier). They were donated by Chevron Texaco Corp. in Burnaby, British Columbia. Their size distribution is given in Table 2.1. The FCC particles have a minimum fluidization velocity, Umf, in air at 20°C and 1 bar pressure of 0.0032 m/s and a loosely packed bed voidage of 0.51. One disadvantage of these particles is that they undergo considerable electrostatic charging, especially in plexiglas units. In order to alleviate these electrostatic effects, when catalyst is loaded to the unit, it is mixed with approximately 0.5 weight % of Larostat 519 powder, a quartenary salt of ammonium, supplied by PPG Industries Inc. After each baghouse cleaning, together with FCC particles, additional amount of Larostat is again loaded to the unit. In addition, the unit is grounded at several points. Figure 2.2 Schematic showing the locations of the pressure ports on the semi-cylindrical coker and its riser Chapter 2 Experimental Set-up Table 2.1 Size distribution of FCC particles from sieve analysis Mesh size (um) Mass fraction (%) 210 - 250 1.4 150-210 4.5 125 -150 5.8 105 - 125 11.9 75 -105 33.1 45-75 37.8 0-45 5.5 2.2 Dual-loop high-density circulating fluidized bed (HDCFB) The solids mixing experiments were carried out in the dual-loop high-density circulating fluidized bed unit designed and constructed by Issangya (1998) and modified by Liu (2001). A schematic of this equipment is shown in Figure 2.3. The details of the unit can be found in Issangya (1998) and Liu (2001). The HDCFB unit consists of two plexiglas risers, two downcomers, a curved-plate impingement separator, cyclones and an air filter baghouse. The main riser is 76 mm in diameter and 5.60 m high. From this riser, the suspension flows through a curved plate installed in a 0.91 m x 0.46 m x 0.61 m box by a 0.40 m long vertical piece of pipe with a 0.03 m nozzle at its end inclined at 30° from the vertical. This impingement-type separator has a low pressure drop and reasonably high separation efficiency and is able to handle high-density suspensions before further separation in a cyclone. The 0.10 m diameter cyclone is used as a secondary separator after which solids are sent to a temporary storage drum through a recycle feed line. From the impingement-type separator, the solids travel into the 0.31 m in diameter, 3.66 m long downcomer which is maintained at minimum fluidization conditions. Solids from the downcomer are then fed via a Chapter 2 Experimental Set-up +~ Ven Figure 2.3 Schematic diagram of high density CFB unit 1. 1st riser (main riser), 2. Impingement separator, 3. Storage tank, 4. 2n d riser, 5. Downcomer, 6. Baghouse, 7. Gate valve, 8. Cyclones, 9. Butterfly valve, 10. Orificemeter, 11. Rotameter, 12. Root's blower, 13. Sore thumb, 14. Recycle feed lines, 15. Pressurized drum P- Absolute pressure transducer port, AP - Pressure transducer port Chapter 2 Experimental Set-up 12 pinch valve (EVR series open flame type) into the second riser through a J-valve equipped with secondary air ports and small aeration lines. The second riser of diameter 0.10 m and height 9.14 m lifts the solids from the first downcomer to a higher level, facilitating a taller second downcomer. This design gives sufficient solids inventory and pressure head to provide a wide range of solids circulation rates. A smooth exit bend then guides the suspension into the primary cyclone, which does not have a conical base. This type of cyclone was selected by Issangya (1998) due to its ability to handle large fluxes of solids, which might have choked the conical exit of a conventional cyclone. While some of the solids are sent to the temporary storage drum after passing through a secondary cyclone, air from both secondary cyclones passes through a filter baghouse before being discharged to the atmosphere. Finally, solids fall into the 0.31 m diameter, 7.62 m tall second downcomer which is also maintained at minimum fluidization conditions. The circulation of suspension in the HDCFB unit is completed with its flow through a pinch valve, then a U-valve from the second downcomer back to the bottom of the first riser. The air supply to the unit consists of a shared building compressor rated at a capacity of 0.03 Nm3/s at 200 kPag (64 SCFM at 30 psig), a blower (4A-Universal RAI Rotary) of capacity of 0.12 Nm3/s at 68 kPag (250 SCFM at 10 psig) and a diesel-driven Root's blower (Mitsubishi model 4D3) of capacity 0.08 Nm3/s at 40 kPag (172 SCFM at 5.6 psig). Two orifice meters are used to measure the gas flow rates in both risers. The other lesser gas flow rates, such as aeration gas in the downcomers, non-mechanical J-valves and gate valves, are measured by rotameters and included with the overall flow in the riser when calculating the superficial gas velocity (Ug). It was easier to maintain U g constant than the solids circulation flux (Gs). The average of the recorded superficial gas velocity as well as its standard deviation within the duration of measurement was reported for each operating condition. An example of change of standard deviation of superficial gas velocity measurements for G s = 250 kg/m s is presented in Figure 2.4. Chapter 2 Experimental Set-up 13 CO 03 o •a 1 1 -*-> CO 1 2 3 4 5 6 7 8 9 Superficial gas velocity, U g [m/s] Figure 2.4 Change of standard deviation of superficial gas velocity at G s = 250 kg/m2s The overall solids circulation flux, G s , is measured by collection of solids over limited times on a porous butterfly valve. The butterfly valve is installed in the upper part of the second downcomer. It consists of two semi-circular drilled plates covered with 3 mm diameter holes on a triangular pitch which rotate downward on their axes. During measurements, the downflowing solids are trapped by rapidly rotating the two halves upward to the horizontal position. This butterfly valve method for the measurement of overall solids circulation flux is not an ideal technique since it interferes with the system, as it can cause a drop of the solids level in the downcomers. However, Issangya (1998) reported that the total solids head in the downcomer remains constant throughout this procedure, except for a small pressure drop due to the valve itself. During the measurements the solids circulation fluxes were measured several times in order to make sure that the operating conditions had not changed. For each operating condition, the average of these several measurements and their standard deviations were calculated. In Figure 2.5, the change of standard deviation with the overall solids circulation flux at a fixed superficial gas velocity (U g = 6 m/s) is presented. Chapter 2 Experimental Set-up 14 100 200 300 400 500 Overall solids circulation flux, G s [kg/m s] 600 Figure 2.5 Change of standard deviation of overall solids circulation fluxes at Ug = 6 m/s It was found that it became more challenging to maintain steady state operation as the solids circulation flux increased. This explains the increase of standard deviation with increasing G s . Note that G s values are reported to three significant figures throughout this thesis, but values are subject to experimental uncertainty. At a 90% confidence level, the uncertainty can be estimated as ± sx2.92/V3 , i.e., ± 1.69s where s is the standard deviation from Figure 2.5, 2.92 is the Student t value for 90% confidence interval with 3 measurements (2 degrees of freedom). To measure pressure profiles and apparent cross-sectional average voidages along the riser, differential pressure transducers are installed across 0.30 m intervals along the main riser, while absolute pressure transducers are installed at ten different locations in the unit. The sampled signals are saved to a computer by means of data-acquisition software written by C. J. Lim using the Visual Basic® programme. The sampling was performed at 0.10 Hz. Chapter 2 Experimental Set-up 15 The FCC particles employed in the first set of experiments were also utilized in this system. As to the fluid coker system, electrostatic charges were alleviated by proper earthing and addition of Larostat 519. During this study, two modifications were installed in the dual loop FfDCFB system: (a) "Sore Thumb" in second downcomer: Liu (2001) presented a pressure balance measurements around the FfDCFB column and prepared the pressure diagram given in Figure 2.6. As shown in this figure, the largest pressure drop in the system occurs between the bottom of the second downcomer and the entrance of the first riser. Although he was able to reduce this pressure drop by replacing the original non-mechanical L-valve (see Issangya, 1998) by a U-valve, there was still a significant pressure drop caused by the shape of the connection of the downcomer to the pinch valve. As might be suspected from Figure 2.3, the particles cannot leave the downcomer uniformly because of the diameter difference between the downcomer and the pipe connection to the pinch valve (Ddowncomer = 0.31 m to Du-vaive = 0.076 m). This configuration was believed to hinder the smoothness of transfer of suspension to the riser, limiting the overall solids circulation rate and causing high pressure drops. Therefore, in this study, a special standpipe entrance, referred to as "sore thumb", was installed in the second downcomer, as illustrated in Figure 2.7. This sore thumb has a 240 mm tall bottom part consisting of a regular screen having 8 mm x 13 mm slots. It has approximately 70% open area. Holes of diameter 10 mm were drilled on the 102 mm tall conical top part to give 30% open area. This geometry is based on a successful geometry tested during the Syncrude-sponsored fluid coker work. The pressure difference between Pi (Figure 2.7), measured at the bottom of the downcomer, and P2, measured at the exit of the U-valve, were determined before and after installation of the sore thumb. As illustrated in Figure 2.8, the pressure drop between these two points decreased with the modification, especially at high solids circulation fluxes. The flow of particles from the pinch valve to the U-valve was also visually monitored to be smoother and more stable. Chapter 2 Experimental Set-up 16 Figure 2.6 Pressure profile in HDCFB unit [from Liu, 2001] (a) U g = 5 m/s, G s = 470 kg/m2s, U g 2 = 6 m/s; (b) U g = 7 m/s, G s = 608 kg/m2s, U g 2 = 7 m/s. (c) schematic of the HDCFB unit showing the positions of the pressure ports (labelled from A to H). Chapter 2 Experimental Set-up 17 Downcomer All dimensions are in mm A I R Figure 2.7 Schematic of bottom part of the FfDCFB unit with "sore thumb" (patent pending) installed Chapter 2 Experimental Set-up 18 1 45 40 35 30 25 i 20 15 10 5 i 0 ° Without sore thumb A With sore thumb 100 200 300 A A A A A o o AAAA A A A 400 500 Overall solids circulation flux (kg/m s) 600 Figure 2.8 Comparison of pressure difference between Pi, measured at bottom of 2nd downcomer, and P2, measured at exit of U-valve, before and after installation of "sore thumb" (Ug = 6 m/s) (b) Temporary storage drum: The accumulation of FCC particles in the baghouse is a function of the percentage of fines, the solids circulation flux and the superficial gas velocity. When the HDCFB unit is operated under high-density and high-velocity conditions, the baghouse fills up within a two to four hours of operation, causing an increase in the pressure at the entrance of the cyclone located at the top of the first riser. Since this makes the control of the superficial gas velocity difficult and causes instability in the riser flow, the operation has to be interrupted to clean the baghouse. In order to reduce the number of these interruptions and tedious cleanings, two 76 mm in diameter ducts were connected to the bottom parts of the secondary cyclones located at the top of the first and second risers (see Figure 2.3). The ducts were then connected to a pressurized drum via a ball valve. The 0.5 m in diameter, 0.5 m long drum has a volume of approximately 0.06 m3. The fullness of the drum was checked by tapping on the drum and shaking the ducts. Once the drum was full, the ball valve was closed and by Chapter 2 Experimental Set-up 19 pressurizing the drum, the particles were loaded back to the system from a port located on the U-valve (see Figure 2.3 and 2.7). After each loading, a few minutes were needed for the system to return to its previous steady state before proceeding with the measurements. When analysing the recorded hydrodynamic data, any data recorded during these particle loading time periods were excluded. 2.3 Pressure measurements and apparent solids hold-up calculations Determination of pressure profiles and apparent cross-sectional average voidages along risers necessitates the measurement of differential and absolute pressures. The average differential and absolute pressures and the standard deviation of the differential pressure are calculated from instantaneous differential and absolute pressures, respectively, i.e., AP 1 N = - E Az N ~ fAP^ Az (2.1) 1 N °"AP = 1 N N i = l 7APN AP" i A z (2.2) (2.3) where N is the number of data points collected during the sampling period. The vertical variation of cross-sectional average solids hold-up in circulating fluidized bed risers is usually inferred from the gradient of absolute pressure profiles or from direct measurement of differential pressures across equal intervals along the riser. The pressure drop per unit length is then equated to the weight of the solids and fluid per unit area, assuming that the combined effects of gas-wall friction, solids-wall friction and acceleration/deceleration are negligible. Therefore, the apparent voidage, £ a p p a r e n t , or solids hold-up, es a p p a r e n t , is usually estimated from Chapter 2 Experimental Set-up 20 ^ - - |pp(l-e)+pge]g (2.4) - Ppesg ( w i t h p p » p g ) This method of estimating solids hold-up is subject to significant error when wall friction or solids acceleration cannot be neglected compared with the static head of the solids (Yerushalmi and Avidan, 1985). The method of Louge and Chang (1990), based on integration of the momentum equation for one-dimensional accelerating suspension flow, was modified by adding terms for gas and solids-to-wall friction by Issangya (1998). This technique was used by Issangya to correct for the effects of particle acceleration and friction. As shown in Figure 2.9, deviations between the apparent and corrected solids hold-ups were always less than 20%. In a separate test performed by Issangya (1998), the deviation between apparent solids hold-ups and mean values obtained by integrating the local voidage data, measured with an optical fiber probe, was found to be about -6% to +10%. These deviations were small enough that the findings were not affected significantly by obtaining the solids hold-ups from equation (2.4). Therefore, solids hold-ups presented in this study are apparent solids hold-ups, es a p p a r e n t , calculated from: A P e, _ Az (2.5) 's, apparent P p g Chapter 2 Experimental Set-up i • — i — • i • oo I o c • 1 oo \ \ EH G s [kg/m*s] - O - O 181 - • - • 425 V o l 1 0.0 0.1 0.2 0.3 0.4 0.5 Solids holduD. Cl - R) Figure 2.9 Apparent (solid symbols) and actual (open symbols) solids hold-up profiles at U = 8 m/s for G s = 181 and 425 kg/m2s as determined by Issangya (1998) Chapter 3 Optical Probe Measurement System 22 CHAPTER 3 OPTICAL PROBE MEASUREMENT SYSTEM Several experimental methods, e.g. photographic and video (Issangya et al., 1999) capacitance probes (Grace et al., 1999), pitot tubes, x- or y-ray transmission, laser-Doppler anemometry, isokinetic or non-isokinetic solids suction probes (Contractor et al., 1994; Knowlton, 1995) and momentum probes (van Zoonen, 1962; Kim et al., 2004a, 2004b), have been used to measure particle velocity, solids concentrations and solids fluxes in circulating fluidized bed systems. Many researchers have preferred optical fibre probes because of their simplicity, high accuracy and relatively low cost. In addition, because of their compact size, optical probe measurements are localized, with minimum disturbance to the flow dynamics (Zhang et al., 1998). Although these measurement systems have been used widely, few of these techniques offer the possibility of determining local instantaneous particle velocities and solids concentrations simultaneously. Therefore, in this study, the 3-fibre optical probe measurement system developed by Liu (2001) (see also Liu et al., 2003a) was employed to obtain simultaneous measurements of local instantaneous solids volume concentrations, particle velocities and solids fluxes. Description of the measurement system together with detailed information on its calibration are presented in this chapter. The probe and the signal conditioning components of the measurement system utilized in this study were fabricated by the Institute of Process Engineering in Beijing, China. The system, labelled "PV-4A", includes an optical fibre probe, a light-voltage signal converter, signal pre-conditioning circuits and a high-speed data acquisition card whose sampling speed can be as high as 2 MHz. As shown in Figure 3.1, the 2 mm thick probe tip consists of three aligned quartz fibres, each of diameter 0.26 mm, in contact with each other. The central fibre projects light into the multi-phase suspension, while the other two fibres receive back-scattered light from moving particles. Chapter 3 Optical Probe Measurement System 23 D = 0.26 mm 2 mm Figure 3.1 Details of 3-fibre optical probe Chapter 3 Optical Probe Measurement System 24 One factor that needs attention with this fibre configuration is the blind zone which either falls outside the region illuminated by the emission fibre or cannot be seen by the detection fibres, as shown in Figure 3.2. In order to prevent measurements from being biased due to particles moving in the blind zone, a glass cover was placed over the probe tip [Liu (2001), Cui et al. (2001), Liu et al. (2003a)]. As suggested by Liu, a thickness of about 0.5 mm was enough to remedy this problem. Measurement region Blind zone ! Light Light Light out in out Figure 3.2 Schematic of blind zone elimination with the glass window and the measurement region at the tip of the optical probe The particle velocity/concentration measurement system using the optical fibre probe is depicted in Figure 3.3. The signals collected by the two receiving fibres are sent to two separate photomultipliers. The signals are then saved to a computer by the high-speed data acquisition software written by Liu (2001) using C programming language under the DOS operating system, allowing data to be acquired at speeds up to 300 kHz for each channel. This is followed by off-line cross-correlation of the signals and conversion of the amplitude to local concentrations. Probe j Chapter 3 Optical Probe Measurement System 25 Chapter 3 Optical Probe Measurement System 26 3.1 Particle velocity measurements Light reflected by particles passes through two receiving fibres and generates two sets of signals. If the flow structure does not change between these two receiving fibres and the particles move in the same direction, the two signals are identical, but separated by a time delay TAB- This time delay can be deduced by calculating the cross-correlation of the two signals, i.e. from 0 <T> = j J l A (t)IB (t + T)dt (3.1) or the cross-correlation coefficient function, 4 \ / b ( T ) - Z A ( Q IB(t) ( 3 2 ) a. cr •A 'R where IA(t) and Ig(t) are the time-average signal intensities at time t for fibres A and B, while O j A and 0"Jb are the variances of the signals generated by two receiving fibres, respectively. TAB is the time delay at which the cross-correlation function or correlation coefficient function reaches a maximum value, corresponding to the average time of passage of flow structures between the two receiving fibres during integration time T. The particle velocity can then be determined from V p = ^ (3.3) T A B where L e is the effective separation distance between the two receiving fibres. In the present work, L e was determined by attaching one or more FCC particles to a rotating disk, recording the speed of rotation of the disk and then calculating L e by using the cross-correlation method. As shown in Figure 3.4, the effective separation distance of the probe was found to be equal to 0.31 mm. Chapter 3 Optical Probe Measurement System -1 27 0.40 0.35 0.30 0.25 0.20 0.15 0.10 0.05 0.00 MeanLe = 0.31 mm Standard deviation = 0.03 mm 0.2 0.3 0.4 0.5 0.6 0.7 0.8 Velocity of the rotating disk (m/s) 0.9 1.0 Figure 3.4 Calibration of effective separation distance, L e by rotating disk method Cross-correlation analysis was performed using MATLAB® software according to the algorithms portrayed in Figures 3.5 and 3.6. The raw data collected by two receiving fibres (channels 1 and 2) were first divided into N groups. Following the approach of Liu (2001), each group was assigned 4096 recorded binary voltage values. As discussed by Ellis (2003), dividing the raw data into N groups of 4096 points each was optimal from the viewpoint of ability to capture variations in particle velocity and computation time. Each velocity point was obtained by cross-correlating the data in each group over an integration period of T, the time for each channel to load each group, i.e., T~ 15 ms for a sampling frequency of 280 kHz. The corresponding solids volume fraction is obtained by averaging over the same period of time. Chapter 3 Optical Probe Measurement System 28 C START ) Obtain the raw data Divide them into groups of N data points Calculate arithmetic mean of each group Perform cross-correlation for each group . Calculate maximum correlation coefficient, P I A I B Calculate particle velocity Call calib.m program for utilization of elimination criteria and calculation of solids hold-up and fluxes Call aver age.m program for calculation of time mean values Tabulate radial distribution of velocity, solids hold-up and flux values for each operating condition and axial location Call crossav.m program to calculate of cross-sectional average values of particle velocity, solids hold-up and fluxes PRINT ( STOP ) Figure 3 .5 Algorithm of the main computer program hydrodynamics.m Chapter 3 Optical Probe Measurement System 29 © Load the calibration equation I Convert voltage signals to concentration values Calculate solid flux values using the particle velocity values together with the corresponding solids hold-up values Elimination criteria No •| Reject data No Reject data No Reject data ( R E T U R N ) Figure 3.6 Algorithm of the calib.m program Chapter 3 Optical Probe Measurement System 30 3.2 Solids concentration measurements Since the particle diameter is much smaller than the fibre diameter of this optical probe, light is reflected by many particles in the measurement volume at any instance, enabling the probe to detect also the instantaneous local solids concentration. As discussed by Liu (2001), the measuring volume at the tip of the probe is dependent on the local solids concentration, decreasing as the solids concentration increases. Therefore, careful calibration is vital for accurate measurement of solids concentration. Previous researchers have used a number of different calibration techniques and have pointed out that it is difficult to find an accurate and reproducible calibration method for heterogeneous suspensions. Most calibrations have been performed in liquid-solid systems because of their capability to provide stable high-concentration solid suspensions. In a water-fluidized bed, Hartge et al. (1986a, 1986b) found a linear relationship between voltage due to light reflection and solids concentration for solid hold-ups up to 50% with quartz particles of mean diameter 58 urn. They then assumed that linearity will still hold in gas-solid systems and generated the calibration equation constants from the concentration measurements obtained by y-ray absorption or pressure transducer measurements. Werther et al. (1993), Schlichthaerle and Werther (1999) and Nieuwland et al. (1996) utilized the same technique with a calibration equation of the form AV = kens (3.4) where AV was the difference between the output signal measured in a two-phase system and that obtained in the same fluid without any solid particles present. Rensner et al. (1991) showed that the index n depends only on particulate phase properties, whereas the coefficient k depends on the type of fluid and the characteristics of the electronic and optical components of the probe system. Because n was independent of the fluid type, it was possible to determine it from calibration in water-solid mixtures. The factor k, which should be determined in an air-solids mixture, was then obtained subsequently by measuring the output signal in a packed bed of particles with an accurately known solids volume fraction. However, in view of the differences in the optical properties (in particular, the refractive index) between gases and Chapter 3 Optical Probe Measurement System 31 liquids, the validity of such a calibration in gaseous suspensions is questionable and should be examined carefully. Zhou et al. (1994) and Liu (2001) verified that there is a linear relationship between the voidage and the output of the optical fibre probe obtained in a liquid-solid system and a gas-solid system by using a dropping/trapping technique. Since the calibration obtained in a gas-solid system was also linear, they only measured the two extreme conditions (an empty column and a packed bed), and then connected these two points with a straight line to obtain their calibration curves. Many other researchers (e.g. Herbert, 1994; Issangya, 1998; Matsuno, 1992; Lin et al., 2001; Ellis, 2003) have utilized dropping/trapping techniques to calibrate their optical fibre probes and have obtained non-linear calibration curves. One difficulty encountered during dropping/trapping tests has been a decrease in stability of the calibration unit at high solid concentrations. Zhang et al. (1998) significantly improved the dropping/trapping technique by using a specifically designed gas-solid downer system with controlled gas-solids flow. The solids concentration was directly measured using a pair of sling shot valves. An iteration procedure was then employed to obtain the true calibration function f, which was used to convert the instantaneous voltage signals into solids concentrations, which were then time-averaged. Cui et al. (2001) employed a series of cubic-shaped solid samples of FCC set in polystyrene for calibration. These mixtures were used to simulate the gas-solid system, in which the transparent polystyrene replaced the air. The calibration was expressed empirically by the relationship 1-e 0.4u ' ' ( 1 5 ) V - V where u= — with V = voltage, V 0 = voltage at empty column, and V m f = voltage at V mf ~ V o minimum fluidization. Few researchers have investigated the calibration of solids concentration theoretically. Reh and Li (1990) were among the first to develop a model for optical fibre probes. Rensner and Werther (1991) presented a model for signal generation which described probe operating Chapter 3 Optical Probe Measurement System 32 conditions and measuring volumes. Lischer and Louge (1992) simulated the calibration for homogenous, dense, random suspensions of smooth, monodisperse, transparent dielectric spheres by using a ray-tracing Monte Carlo algorithm that predicts systematic uncertainties of the sensor's output, the extent of the measurement volume and the effects of changing optical properties. Amos et al. (1996) included the influence of particle size on the probe response functions in their model. Bergougnoux et al. (1999) and Bellino et al. (2001) improved their Monte Carlo simulation by incorporating geometric and optical properties of the fibres, scattering properties of the particles and the solids volume fraction of the suspension. For a parallel-fibre reflective probe, Liu et al. (2003a) developed a simple mechanistic model which explained the existence of the "blind zone", and successfully predicted the reflective and calibration results for probes of different sizes. Their model also showed the influence of parameters such as single fibre diameter and the distance between adjacent fibres on the performance of optical probes used to measure particle concentrations. Many researchers have preferred to perform their calibrations inside the fluidized beds in order to obtain calibration curves reflecting the required scale and flow. Hartge et al. (1986a, 1986b) determined the equation constants of the linear relationship they found for a liquid-solid calibration system by integrating the probe signal measured by traversing the probe across the entire diameter of the bed. Lischer and Louge (1992) calibrated their fibre optic probe against a quantitative capacitance probe and pointed out that the calibration experiments should be carried out with a particle-size distribution identical to that of the suspension under study. Zhang et al. (1991) calibrated their probe by traversing it through a circulating fluidized bed and comparing the results with average solids volume fractions inferred from pressure drop measurements calculated from A P / A z (~ ^ ^s,apparent — ~7 ~ ^ ' ' P p § They employed a polynomial fit to represent the relationship between bed voidage and the probe output signal S: Chapter 3 Optical Probe Measurement System 33 e = C0+C1S+ C 2 S 2 + C 3 S 3 +...+ C kS k (3.7) They carried out measurements at (N+l) different operating conditions to produce (N+l) equations in order to find the unknown constants C 0 . . .C N . Similarly, Lin et al. (2001) used three fluidized beds of different solids fraction and obtained a relationship between the density output signal, N and the solids fraction which was well fitted by a Boltzmann function l z _ 0.749-0.750 l + exp[4.858(N-l.l)] In the present study, three calibration systems were explored. First, the dropping/trapping technique developed by Issangya (1998) and later employed also by Liu (2001) and Ellis (2003) in the UBC group was utilized. However, this method produced considerable scatter, probably mainly due to imperfect closing of the slide valve, fluctuation of the downflow of solids and electrostatic charging, causing particles to adhere to the tip of the probe. Therefore, a second method, the FCC+coke mixture calibration technique was attempted. FCC (Prccbuik = 850 kg/m3) and coke particles (pCOke,buik = 600 kg/m3) were both sieved using screens of aperture sizes 90 and 150 um to provide a mean particle size of 120 um. FCC and coke particles had similar bulk densities. Different concentration mixtures were prepared by combining known masses of FCC and coke particles. Since coke particles are black and therefore absorb most visible light, it was assumed that they behave as voids, while only FCC particles reflect light. The particles are loaded into the tube from the top when the valve is closed (see Figure 3.7). Then the valve is opened very slightly to allow the particles to travel downwards slowly as a loosely packed moving bed. Once a steady fall of solids was attained, the optical probe signals were recorded at 100 Hz for 10 s. For each condition 10 measurements were obtained. By changing the concentration of FCC in FCC+coke mixture, different solids concentrations were simulated. Chapter 3 Optical Probe Measurement System 34 Velocity/voidage measurement system Ch 1 Ch 2 0 Figure 3.7 FCC+coke mixture calibration system Before each experiment, the optical probe system had to be turned on for at least half an hour for the system to stabilize. Afterwards, the probe was inserted into a "black box", supplied with the system to set the lower voltage value (V0). The black box was a tube with an inner coating of black paint to absorb all visible light emitted from the probe, thus simulating the solid suspension condition with es = 0. Then the probe was inserted into the FCC+coke mixture calibration unit with the tube filled with only FCC particles. This gave the upper voltage of the system (Vmf). The corresponding packed bed FCC voidage is then calculated from £mf = 1 " — (3-9) PP The V 0 and V m f values were recorded before and after the calibration tests and also for each experimental run. Calibration curves obtained using the FCC+coke mixture system are given in Figure 3.8. Chapter 3 Optical Probe Measurement System 35 1.0-0.8-0.6 0.4 0.2 H 0.0 1 • 1 • 1 Channel 1 1 •—^ ,o -/ /o 0/ -0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0 ( V - V ) / ( V - V ) Figure 3.8 Calibration curves obtained by using FCC+coke mixture in slowly moving packed bed calibration system Since there are two receiving fibres, two sets of concentration data were recorded by the probe system. Following the approach of Liu (2001), the signal recorded by Channel 2 was selected for the calculation of concentration. The calibration equations were first assumed to be of power equation form, i.e., l - £ 1 - 8 m f ' v-v. v V - V (3.10) Best fit values of n were found to be 0.77 for Channel 1 and 0.69 for Channel 2. In order to check the accuracy of the solid hold-up values calculated using the above calibration equation, their cross-sectional average values were compared with the cross-sectional apparent solid hold-up values determined from the measured pressure drops (Eq. (3.6)). For each operating condition and axial location, solid hold-ups were measured at 6 different radial locations (r/R = 0.0, 0.25, 0.50, 0.75, 0.88, 0.94) using the optical fibre probe. Apparent solid hold-ups were measured around the optical probe measurement point using pressure transducers installed flush with the wall. Chapter 3 Optical Probe Measurement System 36 As shown in Figure 3.9, there was considerable scatter in the data, with the majority of the solids hold-up values measured by the optical fibre probe system being higher than the apparent solids hold-up values (from Eq. (3.6)). This discrepancy may be due to the inability of the calibration system to represent the flow dynamics of the actual system. Hartge et al. (1986a, 1986b) and Zhang et al. (1991) recommended that the calibration be performed in the medium of interest, reflecting the required scale and flow dynamics. Therefore, it is decided to calibrate the optical fibre probe against the apparent solid hold-ups measured by the pressure transducers (Eq. (3.6)). First, it was assumed that the calibration equation can be represented by a power equation form with a different power index, n. 0.00 0.05 0.10 0.15 0.20 Apparent solid hold-up Figure 3.9 Comparison between solid hold-up recorded by the optical fiber probe (Channel 2) and integrated across column cross-section and apparent solid hold-up values obtained from pressure gradients, i.e. Eq. (3.6). Chapter 3 Optical Probe Measurement System 37 Calibration curves for different n values are plotted in Figure 3.10. As can be seen in the parity plots of solids fraction from pressure drop and optical fibre probe measurements in Figure 3.11, the solid hold-up decreases, while the power n of the fitted power equation increases. As n decreases, the gap between the solid hold-ups from pressure transducer and probe measurement is reduced for low concentrations, but augmented for high concentrations. From these curves, it can be concluded that a power law type curve (Eq. (3.10)) is not a good choice for the calibration equation. 0.0 0.2 0.4 0.6 0.8 1.0 ( V - V o ) / ( V m f - V o ) Figure 3.10 Change of calibration curve by using different power n values Chapter 3 Optical Probe Measurement System 3 8 0.00 0.05 0.10 0.15 0.20 Apparent solid hold-up 0.00 0.05 0.10 0.15 0.20 Apparent solid hold-up 0.00 0.05 0.10 0.15 0.20 Apparent solid hold-up 0.00 0.05 0.10 0.15 0.20 Apparent solid hold-up Figure 3.11 Parity plot of solids fraction from pressure drop (Eq. (3.6)) and optical fibre probe measurements for different values of the index n = 0.77,1.0, 1.5, 2.5 Chapter 3 Optical Probe Measurement System 39 For comparison purposes, the calibration equation proposed by Cui et al. (2001) for different types of optical fibre probes was also examined. As demonstrated in the parity plot in Figure 3.12, there is good agreement only at low solids concentrations. Therefore, this calibration equation was also discarded. Note that the geometry of Cui's probe differs significantly from that used in this project. Apparent solid hold-up Figure 3.12 Parity plot of solids fraction from pressure drop Eq. (3.6) and optical fibre probe measurements using the calibration equation (Eq. (3.5)) obtained by Cui et al. (2001) Among many types of equations tested, a calibration equation of the form of the Boltzmann equation gave the best overall results. The computer programs written in MATLAB® and following the algorithms summarized in Figures 3.6 and 3.13 were employed in determination of the calibration equation. For each calibration equation tested, the solid hold-ups were compared with the apparent values. If they differed by more than 35%, a new calibration equation with different equation constants was tried. During this operation, it was not difficult to find a calibration equation which gave less than 20% error in the solids hold-up. However, another point needs careful attention. Many different radial distributions of solid hold-up can give the same cross-sectional average. This causes no problem if one is only dealing with cross-sectional averages. Chapter 3 Optical Probe Measurement System 4 0 C START ) Obtain the raw data Divide them into groups of N data points Calculate arithmetic mean of each group Perform cross-correlation for each group T Calculate maximum correlation coefficient P I A I B Calculate particle velocity Call calib.m program for utilization of elimination criteria and calculation of solid hold-up and fluxes Call average.m program for calculation of time mean values T Tabulate radial distribution of velocity, solid hold-up and flux values for each operating condition and axial location Call crossav.m program to calculate of cross-sectional average values of particle velocity, solids hold-up and fluxes Compare solid hold-ups with those measured by pressure transducer (Eq. (3.6)) Compare solid flux with overall solids circulation flux measured via the return loop No P R I N T f ( S T O P ) Figure 3.13 Calibration equation iteration program algorithm Chapter 3 Optical Probe Measurement System 41 However, if the radial distributions are needed for analysis, an additional criterion is required to check the cross-sectional averages and obtain the correct radial distributions. Since the optical fibre probe used in this study can measure solids hold-up and circulation flux simultaneously, comparing the cross-sectional average net solids flux with the overall solids circulation flux determined with the quick-closing valve in the return line provides an alternative means of checking the accuracy of the calibration equation. It was found that when the best calibration equation, selected based only on the cross-sectional average solids hold-up, was employed, agreement between the cross-sectional average solids flux and the overall solids circulation flux was very poor. Other researchers have not been able to measure solids hold-up and solids flux simultaneously using the same optical probe. Therefore, they never encountered or confronted this issue. On the other hand, in this study, both solids hold-up and flux values were examined carefully. To best satisfy both the solids hold-up and flux criteria, the calibration curve given in Figure 3.14 was selected. 0.0 0.2 0.4 0.6 0.8 1.0 (v-v o)/(vm f-v o) Figure 3.14 Final response curve for calibration of optical probe for solid concentration Chapter 3 Optical Probe Measurement System 42 Parity plots of solid hold-up and flux determined via this calibration curve are given in Figure 3.15. Average absolute percentage deviations were calculated by means of 100 N '•s.m i=l ( £ s , m - P r o b e ) - (apparent es) apparent 8 s(3.11) 100^ (^s.m-pwbe^ ~ (Overall solids circulation flux) Overall solids circulation flux (3.12) There are two additional points that need attention with respect to the algorithms given in Figure 3.5, 3.6 and 3.13: selection of elimination criteria, - cross-sectional averaging technique. 3.2.1 Elimination criteria In circulating fluidized beds, particles may reverse directions. In addition, a flow structure travelling non-vertically past one fibre may not be detected by the second fibre during the integration time, causing the cross-correlation coefficients to be low or indeterminate over the measurement period. Data which are uncorrectable or poorly correlated due to such phenomena need to be eliminated. Previous researchers have employed different elimination criteria. In this study, the elimination criteria of Liu (2001) were followed, with data being eliminated if the calculated velocity: falls in the range of ± | V m j J where | V m J = 2*Le/T since the maximum time delay cannot exceed T/2, differs by more than " a a " (called the variance interval constant) standard deviations from the average, - has a cross correlation coefficient less than a cross-correlation coefficient cut-off value. Chapter 3 Optical Probe Measurement System -30% 0.00 0.05 0.10 0.15 0.20 0.25 0 100 200 300 400 G s , m . probe Figure 3.15 Parity plots obtained by using cut-off value = 0.5 in elimination criteria (a) Comparison of solid hold-ups recorded by the optical fiber probe with apparent solid hold-ups (b) Comparison of solid fluxes recorded by the optical fibre probe and overall solids circulation flux Chapter 3 Optical Probe Measurement System 44 The first criterion is introduced because of the limitations of the measurement system and the cross-correlation analysis. The lowest measurable absolute velocity is related to the integration time, T. The maximum time delay cannot exceed T/2, with the particle velocity corresponding to this time delay being I VmJ = 2Le/T where L e is the effective distance between the two receiving fibres in the probe. For example, for T = 30 ms and L e = 0.31 mm, the lowest measurable absolute velocity is 0.02 m/s. Therefore, data with apparent magnitudes of velocity in the range of ± I VmJ were eliminated. Very few particles are expected to have velocities of magnitude less than this in CFB systems. Hence any bias introduced by the inability of the probe to measure such small velocities is expected to be very small. For the last two elimination criteria, a literature survey was done. Werther et al. (1996) and Ellis (2003) eliminated data with cross-correlation coefficients < 0.6. This criterion caused 20-40% of their data to be discarded. Militzer et al. (1992) eliminated data if the correlation coefficient was less than 0.5 and also if the calculated velocities differed by more than 2 standard deviations from the average. Liu (2001) applied a similar criterion with the correlation coefficient required to be greater than 0.5 and individual calculated velocities to differ by no more than five standard deviations from the average. In this study, different cut-off and variance interval constant, aCT values were tested. The particle velocities with cross-correlation coefficient larger than the cut-off value and in the range -aa * ov < V p < a^ * were accepted while the rest were eliminated. Note that since particle velocities and solid hold-ups were measured simultaneously, eliminating a particle velocity also led to the elimination of the corresponding solids hold-up and flux values. Therefore, the above elimination criteria affect not only the particle velocities, but also the solids hold-ups and fluxes. Hence, the effects of the elimination criteria on the accuracy of the solid hold-up and flux in addition to particle velocity were also evaluated by comparing them with apparent solid hold-up and overall solids circulation flux. As shown in Table 3.1, as the cut-off value increases or aa decreases, the number of eliminated data increases. In practice, the number of data should be large enough to allow suitable averaging over a sufficiently large ensemble of flow structure elements. Chapter 3 Optical Probe Measurement System 45 Table 3.1 Effect of cut-off and variance interval constant, a G values on the percentage of eliminated data and on average percentage deviations Cases Cut-off value % eliminated data 85 fcs,m (Eq. 3.11) 8 G (Eq. 3.12) A 3 0.5 20 26 24 B 3 0.6 21 25 25 C 3 0.7 23 26 22 D 4 0.5 17 28 22 E 4 0.6 19 26 21 V ; . , 0 - 7 . :v.21i//. ' • . • : 2 0 1/.-, G 5 0.5 16 31 22 H 5 0.6 18 28 24 I 5 0.7 20 26 22 It can be inferred from this table that Case F gives the least average absolute percentage deviations 8 F and 877 among other cases without significantly increasing the amount of s,m us,m eliminated data. These targets were satisfied in this study by setting aCT and the cut off values to be equal to 4 and 0.7, respectively. Chapter 3 Optical Probe Measurement System 46 3.2.2 Cross-sectional averaging of the radial profiles of particle velocity, solids hold-up and flux Before calculating the cross-sectional average solids concentration and flux, their time-mean values were calculated from ^4^'(t)dt4§e-i=i?fe» (3-i3) °>fJl G' ( t ) d t =¥t G' J (3J4> The calculation of the time-mean particle velocity is discussed in Chapter 4. Following the approach of Liu (2001), cross-sectional average values were first obtained by the integration of the fourth order polynomial curve fit of local time-mean values over the column cross-section, i.e. = j p X 2^rdr=/ o 12^( Pd( P (3.15) G i i n = | ) 1 2 G i ( p d i p (3.16) where cp = r/R. Since more than 60% of the data collected at r/R = 1.0 were eliminated after employing the elimination criteria, instead of using its measured value, its value was extrapolated using the data collected at other radial locations. In this study, the cross-sectional average values were estimated by dividing the bed cross-section into 6 concentric rings as shown in Figure 3.16. Chapter 3 Optical Probe Measurement System 47 El 1 1 £ 2 1 1 ! 63 I x I £4 ^ A i 8 5 ! 1 j Be A 1 r/R 9 r/R r/R r/R T r/R r/R 0.0 / 0.25 0.50 / 0.75 0.88 0.94 0.13 0.38 0.63 0.81 0.91 0.97 1.00 Figure 3.16 Concentric rings and their dimensions used in the calculation of cross-sectional averaging values The weighted sum in each ring with respect to its surface area was calculated using: e s m = (0.132 - 0 2 ) £ , +(0.382-0.132)e2 +(0.632 - 0 . 3 8 2 ) £ 3 +(0.812 - 0 . 6 3 2 ) £ 4 + (0.912 - 0 . 8 1 2 ) £ 5 +(0.972 - 0 . 9 1 2 ) £ 6 +(12 - 0 . 9 7 2 ) £ 7 = 0.02 E,+0.13 E2+0.250 E3+0.27 E4+0.16 E5+0.12 E6+0.06 E 7 (3.17) In Eq. (3.17), £7, which corresponds to the value at the wall (r/R = 1.0), was obtained by a "piecewise cubic Hermite" extrapolation using MATLAB® software. Similarly, the cross-sectional average solids fluxes can be calculated employing Eq. (3.17), together with their radial distributions. Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 48 CHAPTER 4 SOLIDS HOLD-UPS AND MOTION IN HIGH-DENSITY CFB RISERS In the work presented in this Chapter, dense suspension upflow regime characteristics were explored by simultaneously measuring particle velocity, solids concentration and solids flux under different operating conditions and comparing the results with measurements previously conducted (Liu, 2001) utilizing the same measurement method in a HDCFB riser of smaller diameter. In sections 4.1, 4.2 and 4.3, the local solids concentration, particle velocity and solids flux profiles are analysed, respectively. In section 4.4, the motion of solids at the wall region is considered in some detail. 4.1 Local solids hold-up profiles Solids hold-up (es) is one of the key parameters in characterizing flow in circulating fluidized beds. A number of experimental investigations have demonstrated that the axial variation of cross-sectional average solids hold-up depends on many factors such as operating conditions, particle properties, solids inventory, as well as the experimental configuration [Li et al., 1988; Bai et al, 1992; Issangya et al., 1997; Liu 2001]. In these studies, a typical CFB riser was suggested to contain a dilute region towards the top and a dense region near the bottom with an inflection point between, giving an overall sigmoidal profile. Li and Kwauk (1980) reported that increasing the superficial gas velocity (Ug) at a constant net solids circulation flux (Gs) reduced the height of the dense region, whereas an increase in net solids flux at a fixed superficial gas velocity increased the solids hold-up and caused the dense region to expand. This was also monitored by Bai et al. (1992). The height of the dense region depends on the combination of U g and G s . According to the combination of U g and G s selected, the height of the bottom dense section may eventually reach the top of the riser, Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 49 leading to wholly dense phase operation or the dense suspension upflow regime reported by Issangya et al. (1997, 2000) and Liu (2001). Ouyang et al. (1993) and Issangya (1998) reported that once this dense suspension operation is reached, the solids hold-up profiles in this dense region are not very sensitive to either U g or G s . Although there have been a number of studies on identification of the flow characteristics of this dense suspension upflow regime, the effect of diameter of the riser on the hydrodynamics of the flow regime has not been investigated sufficiently. It is the objective of this section to investigate the solids hold-up profiles in the 0.20 m diameter riser of the fluid coker cold model compared to the findings obtained by Liu (2001) and Issangya (1998) in the 76 mm diameter riser of the HDCFB unit and by Malcus (2000) in a 104 mm diameter riser. In this study, the vertical variations of cross-sectional average solids hold-up were first inferred from direct measurements of differential pressures over equal intervals along the riser of the fluid coker cold model (see Figure 2.1). As explained in Chapter 2, apparent solids hold-ups were calculated from AP/Az 8 — (4.1) s, apparent v ' Figure 4.1 shows the axial apparent solids hold-up profiles in the riser as a function of the solids circulation flux at a superficial gas velocity of 6 m/s. As depicted in this figure, at low G s (Gs = 125 kg/m2s), a relatively dilute and uniform distribution occupied the whole riser, except for the bottom section, which was no doubt affected by particle acceleration, as well as by the solids entry configuration. With increasing solids net flux (125 < G s < 330 kg/m s), esapparent increased, and the shape of the axial distribution profiles changed to an S-shape, with a dense region at the bottom and a relatively dilute region towards the top. Figure 4.2 shows the effect of the superficial gas velocity on the apparent solids hold-up profiles. The lines shown were obtained by sigmoidal curve fitting using Origin® software. For high superficial gas velocities (Ug > 7 m/s), the apparent solids hold-up profiles did not change, except at the bottom of the riser. However, 8S a p p a r e n t increased sharply in the riser as U g decreased from 6 to 5 m/s, except near the top (z = 4.42 m). Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 0.00 0.05 0.10 0.15 0.20 0.25 Apparent solids hold-up [ - ] Figure 4.1 Axial profiles of apparent solids hold-up for various solids circulation fluxes at U p = 6 m/s Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 51 5 H • U g = 5 m/s A Ug = 6 m/s • U g = 7 m/s O U g = 8 m/s 0 -I 1 1 1 1 — 1 0.00 0.05 0.10 0.15 0.20 Apparent solids hold-up [-] Figure 4.2 Axial profiles of apparent solids hold-up for G s = 250 kg/m2s and various superficial gas velocities Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 52 Similar trends have been observed in several previous studies of high-density riser flows (Malcus et al., 2002; Karri and Knowlton, 2002; Manyele et al., 2002). The effect of solids circulation flux on solids hold-up at several heights along the riser is shown in Figure 4.3 for a superficial gas velocity of 6 m/s. 0.25 |w 3 O tt] 3 'o tt) -4—> c a a cu cu < 0.20 0.15 0.10 0.05 0.00 o • • Height [ m ] 0.76 1.67 3.10 4.42 Height interval 0.64 - 0.88 1.55 - 1.79 2.98 - 3.22 4.22 - 4.62 400 Solids circulation flux, G [ kg/m s ] Figure 4.3 Effect of solids circulation flux on apparent solids hold-up for various heights along the riser of the fluid coker cold model at U g = 6 m/s Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 53 A dense region formed in the lower section of the riser, with a steep increase in E S A P P A R E N T as G s increased. When G s exceeded 200 kg/m2s, \ apparent a t t n e lowest measurement level (0.76 m) appeared to approach a constant value of about 0.21; e s a p p a r e n t showed no further increase with increasing G s indicating that this location was in the high-density condition or dense suspension upflow regime (Issangya et al., 1999). With a further increase of G s above -200 kg/m2s, the high-density condition was also reached at the next measurement level (z = 1.67 m). However, the two locations nearest to the top (z = 3.10 and 4.42 m) were lean, without a sharp increase of e S ) a p p a r ent f ° r m e given G s range. These results indicate that the flow behaviour and regime transitions are functions of height in the riser (see Kim et al., 2004a). In order to investigate the diameter effect, the axial solids hold-up profiles obtained in this study are compared with those measured by Issangya (1998) and Liu (2001) in the 76 mm diameter riser of the HDCFB unit (Figure 2.3) and by Malcus (2000) in a 140 mm diameter riser. The properties of these experimental units and that of Malcus (2000) are summarized in Table 4.1. Table 4.1 Properties of high-density risers and their operating conditions used in comparisons Reference Size of the riser Operating conditions Particle properties Diameter (m) Height (m) u g (m/s) G S (kg/m2s) Type P P . (kg/m3) dp (um) Issangya (1998) 0.076 6.1* 5-8 38 - 336 FCC 1600 67 Liu (2001) 0.076 5.6* 5-8 11-553 FCC 1600 67 Malcus (2000) 0.140 7.0 4.7 150 -264 FCC 1740 89 This study 0.203 5.9 5-8 22 - 345 FCC 1700 70 * Height changes reflect modifications in geometry to facilitate increased flux and improved operation. Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 54 In order to prevent the height of the unit affecting the comparison, the axial measurement locations were normalized by dividing them by the total height of the riser under examination. As shown in Figure 4.4, although the measurements of Liu (2001) and Issangya (1998) were performed in the same experimental unit, the apparent solids hold-ups measured by Liu were significantly lower than those of Issangya in the entrance section due to Liu's modification of the bottom of the riser to include a U-bend and a venturi. Since both dual-loop HDCFB units and the fluid coker cold model unit had similar riser heights and employed the same FCC particles, differences between the data were mainly due to the differences in riser diameter and entrance section. The apparent solid hold-ups in this study were always lower than those measured in 76 mm diameter unit. The effect of riser diameter was relatively small when the apparent solid hold-up was less than 0.05 in the top-dilute section of the unit, or for the bottom dense section (i.e. for eS ) a pparent - 0-2)-In order to examine the effect of diameter at the dense section in more detail, attempts were made to raise the solids circulation flux further, but due to the limitations of the available blower and the design of the fluid coker unit, steady state operation was impossible for solids circulation fluxes beyond about 350 kg/m2s. Therefore, the superficial gas velocity was reduced to 5 m/s to attain denser conditions. In Figure 4.5, the data collected by Malcus et al. (2002) in a 0.14 m diameter high-density riser are also included for comparison. Details of this unit are provided in Table 4.1. The lines shown in the figure were obtained by sigmoidal curve fitting using Origin® software. The above observations that the diameter of the riser does not have a large impact on the very dense and very dilute sections of the riser and that the solid hold-up decreases with increasing diameter of the riser are supported by Figure 4.5. However, contrary to these findings, Yan et al. (2004) measured axial and radial profiles of solids hold-up with a reflective-type fibre optic concentration-probe and pressure transducers and observed that the solids hold-up was higher for the larger bed diameter. The experiments were conducted in a twin-riser CFB system having two 10-m long risers of diameter 76 mm and 203 mm which had identical configurations, except for their diameters. It is believed that the difference between the findings of Yan et al. (2004) and this study might have been caused by the differences in the entrance and exit configurations of the risers under consideration in the present study. Since Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 55 ^ 0.6 J3 SP N O 0.0 0.1 0.2 0.3 0.4 Apparent solids hold-up [-] - | 1 1 r 0.0 0.1 0.2 0.3 0.4 Apparent solids hold-up [-] —•— Liu (2001) : G s = 125kg/m2s Issangya (1998): G s = 151 kg/m2s — • • — This study : G s = 125 kg/m2s — • — Liu (2001) : G s = 228 kg/m2s •O - Issangya (1998): G s = 205 kg/m2s — ••— This study : G s = 202 kg/m2s 0.0 0.1 0.2 0.3 0.4 Apparent solids hold-up [-] • Liu (2001) : G s = 299 kg/m2s This study : G s = 290 kg/m2s Issangya (1998): G s = 311 kg/m2s Issangya (1998): G s = 336 kg/m2s - This study: G s = 327 kg/m2s Figure 4.4 Comparison of apparent solids hold-ups measured in the 0.203 m riser of this study with those measured by Issangya (1998) and Liu (2001) in 0.076 m diameter riser of the dual-loop HDCFB unit [ U g = 6 m/s] Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers o Liu (2001): G s = 286 kg/m2s, U g = 5.0 m/s, D . * nser = 0.076 m, H = 5.6 m • Malcus (2000) = 264 kg/m2s, U g = 4.7 m/s, D . ' nser = 0.140 m,H = 7.0 m A This study : G s = 250 kg/m2s, U g = 5.0 m/s, D . ' nser = 0.203 m, H =  5.9m 0.00 0.05 0.10 0.15 0.20 Apparent solids hold-up [-] 0.25 Figure 4.5 Comparison of apparent solids hold-ups measured in this study (Driser = 0.203 m) with those measured by Liu (2001) ( D r i s e r = 0.076 m) and Malcus (2000) ( D r i s e r = 0.140 m) Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 57 direct comparison between this study and that of Yan et al. (2004) was only possible for one relatively dilute operating condition (Gs ~ 100 kg/m2s and U g ~ 6 m/s), it is not clear whether or not the findings would differ also over a broad range of experimental conditions. As observed in Figures 4.1 - 4.5, it was not possible to have a dense suspension covering the whole riser under any operating conditions in this study. However, at solids circulation fluxes around 300 kg/m2s in the smaller column (see Figure 4.4), Issangya reported that a dense suspension persisted over the entire riser height. Liu's axial solids hold-up profile was also more uniform throughout the riser than those observed in this study. The different profiles likely arose because the earlier work adopted a dual-loop CFB system with a standpipe whose height was larger than that of the riser and with a sufficient solids inventory that a high pressure-drop and elevated solids hold-up were imposed over the entire riser. Different exit geometries may also have contributed to different axial profiles of solids hold-up. Other researchers have been also only able to achieve dense regions in lower sections of their risers due to limitations of their experimental units. Therefore, they analysed dense suspensions only in the bottom dense sections of their risers. Bai et al. (1996), based on their measurements with a momentum probe, suggested that the solids in the dense bottom region are in a violent turbulent state, as suggested by earlier researchers (Schnitzlein and Weinstein, 1988; Bolton and Davidson, 1988). While Sun et al. (1999) observed core-annulus behaviour in the bottom zone of their riser, other researchers have considered the flow in the bottom region to be cluster-like, as indicated by a high intermittency index (Brereton and Grace, 1993) or as bubbling fluidized beds (Win et al., 1994). The variety of opinions about the state of fluidization in the dense bottom region of a CFB riser shows that there is a need for more detailed study. In addition to hydrodynamic properties, researchers have also disagreed with respect to determination of the height of this dense section of the riser. Svensson et al. (1993) defined the height of this zone, based on their pressure transducer measurements, as the height where the pressure profile began to deviate from a straight line. Schlichthaerle and Werther (1999) measured solids concentration by y-ray absorption and an optical fibre probe and suggested the height of the bottom zone to be the height where the solids concentration started to decrease. As shown in the previous figures, it is very difficult to distinguish the bottom zone Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 58 from the rest of the riser by just looking at the axial solids hold-up profiles and applying the above definitions. Therefore, the apparent solid hold-ups measured by the pressure transducers were next compared with the cross-sectional mean values, obtained by integrating the local time-mean solids hold-ups, measured by the 3-fiber optical probe presented in Chapter 3, over the riser cross-section, i.e. Eq. (3.17). Measurements were obtained at five axial levels (z = 0.76, 1.27, 1.67, 3.10 and 4.42 m) and seven radial locations (r/R = 0.00, 0.25, 0.50, 0.75, 0.88, 0.94, 1.00). At each measurement location, three replicate measurements were performed. The averages of these measurements were taken and reported in this Chapter. In order to show the reproducibility of these measurements, the standard deviations were calculated and used in the calculation of error bars as shown in Figure 4.6. — st _, The error bars in this figure corresponds to 90% confidence interval and cover es ± —^J-Vn where s is the estimate of standard deviation based on the n measurements and tn_i is the Student t. For 3 measurements (degree of freedom = n-1 = 2) and 90% confidence interval, tn_i is equal to 2.92. As can be seen in this figure, it was possible to perform reproducible measurements using the 3-fibre optical probe. The variance in the replicate measurements increased as solid hold-up increased. In both axial locations, the variance in measurements increased towards the wall region. In Figure 4.7, the cross-sectional average of solids hold-ups measured by the optical fibre probe are compared with those measured by pressure transducers. Agreement is reasonably good for most of the riser, excluding the inlet section. This helps to justify the use of the optical fiber probe to measure solids concentrations, given that the pressure gradient method has been validated (Arena et al., 1989) and adopted by many authors, although it leads to some errors in the lower region where neglecting the particle acceleration is not fully justified. Radial profiles of time-mean solid hold-ups together with their time-traces obtained using the optical fibre probe under typical high and low density conditions are shown in Figures 4.8 (a) and (b), respectively. In both cases, denser structures with higher fluctuations were observed moving outwards from the axis to the riser wall. At the centre of the riser, fluctuations were very weak in both cases. In the wall region, on the other hand, there was a significant difference in the solids hold-up values, the high-density condition having higher solids hold-up than the low-density condition. Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 59 3 O C/3 -a s 1 S H 0.00 0.0 0.2 0.4 0.6 0.8 Radial location, r/R [-] 1.0 3 O B i B 0.0 0.2 0.4 0.6 Radial location, r/R [-] 0.8 1.0 Figure 4.6 Reproducibility of local time-mean solids hold-up measurements performed by 3-fibre optical probe at z = 0.76 m and z = 4.42 m for G = 250 kg/m2s and U = 6 m/s Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 60 n 1 1 r 0.00 0.05 0.10 0.15 0.20 0.25 Cross-sectional average solids hold-up SP 'S a (c) 0 • o* • o 0 0.00 0.05 0.10 0.15 0.20 0.25 Cross-sectional average solids hold-up SP '5 a (b) • o • o 0 1 1 1 1 r 0.00 0.05 0.10 0.15 0.20 0.25 Cross-sectional average solids hold-up J3 SP '53 X n 1 1 r 0.00 0.05 0.10 0.15 0.20 0.25 Cross-sectional average solids hold-up Figure 4.7 Comparison of axial solids hold-up profiles measured by pressure transducers (open symbols) and by optical fibre probe (closed symbols): (a) U g = 6 m/s, G s = 250 kg/m2s, (b) U g = 6 m/s, G s = 330 kg/m2s, (c) U g = 5 m/s, G s = 250 kg/m2s, (d) U g = 8 m/s, G s = 250 kg/m2s Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 0.5 1 0.4 0.5 -f 0.4 0.3 0.2 0.1 0.0 1 2 3 4 5 6 7 8 9 10 (a) 1.0 r/R [-] 0.5 -\ 0.4 -\ 0.3 -\ 0.2 0.1 0.0 0.5 0.4 ! 0.3 0.2 0.1 0.0 0.5 0.4 0.3 0.2 0.1 0.0 0.5 0.4 0.3 0.2 0.1 0.0 1 2 3 4 5 6 7 8 9 10 Time [s] 1 2 3 4 5 6 7 8 9 10 Time [s] 0 ^1 2 3 4 5 6 7 8 9 10 Time [s] 0.0 —i— 0.2 0.4 —i— 0.8 (b) 1.0 r/R[-] Figure 4.8 Radial profiles of time-mean solid hold-up and their time-traces obtained using optical fibre probe under (a) high-density condition: G s = 252 kg/m2s, U g = 6 m/s, z = 0.76 m; (b) low-density condition: G s = 255 kg/m2s, U g = 6 m/s, z = 4.42 m Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 62 Figure 4.9 plots radial profiles of local time-mean radial solids hold-up at different levels for U g = 6 m/s and various solids circulation fluxes. Corresponding axial apparent solid hold-up profiles appear in Figure 4.1. The lines shown were obtained using the spline curve method supplied by SigmaPlot 7.0 software. It is seen that for all cases, a relatively dilute core of solids hold-up less than 0.1 co-existed with a dense wall region where the solids hold-up was as high as 0.5 in the bottom section of the riser. In the top section, on the other hand, the solids hold-up was relatively low and uniform across the entire cross-section. At z = 0.76 m and z = 1.27 m, there was little change of radial profile with G s for G s > 300 kg/m2s. These findings are consistent with those reported by Issangya et al. (2000) and Liu (2001) for the dual-loop FfDCFB column (see Chapter 2) with FCC particles based on an optical fibre probe. The axial change of solids hold-up profiles within the riser for a high-density condition (Ug = 6 m/s and G s = 330 kg/m2s) can be seen more clearly in Figure 4.10. As shown in the figure, a flatter solids hold-up profile was found to exist in the top section, changing to a steeper radial profile in the lower sections of the riser. For 0.76 m < z < 3.1 m, the profiles did not change significantly, except in the wall region. The effect of superficial gas velocity on the radial time-mean solids hold-up distributions at different axial locations is presented in Figure 4.11 for U g = 5, 6, 7 and 8 m/s and G s = 250 kg/m2s. Corresponding axial profiles of apparent solids hold-up appear in Figure 4.2. As U g decreased, the time-mean solids hold-ups increased at the wall of the riser, while the central region remained almost unchanged. It must be noted that in Figures 4.9 and 4.11, it was impossible to make direct comparisons for some operating conditions because there were missing data due to problems encountered during some of the measurements. In particular, on some very dry days, electrostatic charging inside the unit affected the optical fibre probe (causing deposition on the tip) such that the collected data had to be discarded after utilizing the elimination criteria explained in Chapter 3. In order to gain better understanding of the local flow behaviour, especially in dense suspension locations, it is important to examine the instantaneous behaviour in addition to the time-average behaviour. Figure 4.12 shows 20 s traces describing the local variation of instantaneous solid hold-up at six radial positions at z = 0.76 m for U g = 6 m/s and G s = 341 kg/m2s. Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 0.5 0.4 [ 0.3 0.2 0.1 0.0* 0.5 0.4 0.3 0.2 0.1 0.0 £ z = 4.42 m z = 3.10 m G s = 204 k g / m s G s = 255 k g / m 2 s G s = 123 kg / rn s G s = 203 k g / m 2 s } s = 244 k g / m s Go = 340 kg/rri s 3 o rt rt o el <4 G s = 198 k g / m s G s = 240 k g / m 2 s G s = 331 k g / m 2 s G s = 127 k g / m s G s = 201 k g / m 2 s G s = 246 k g / m 2 s G s = 327 k g / m 2 s G s = 344 k g / m 2 s 0.5 0.4 0.3 0.2 0.1 0.0 z = 0.76 m G s = 124 k g / m s G s = 206 k g / m 2 s G s = 252 k g / m 2 s G s = 288 k g / m 2 s G s = 341 k g / m 2 s 0.0 0.2 0.4 0.6 0.8 1.0 Radial location, r/R [-] Figure 4.9 Radial profiles of local time-mean solids hold-up at different axial locations for U e = 6 m/s and various solids circulation fluxes Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 64 0.5 A T 0.4 0.3 0.2 0.1 A 0.0 0.0 A -z = 0.76 m z = 3.10m - B - z = 4.42 m 0 -A .41 0.2 0.4 0.6 Radial location, r/R [-] 0.8 1.0 Figure 4.10 Radial profiles of local time-mean solids hold-up at different axial locations for TL = 6 m/s and G s = 330 kg/m2s Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 65 0.5 [ 0.0 0.2 0.4 0.6 0.8 1.0 Radial location, r/R [-] Figure 4.11 Radial profiles of local time-mean solids hold-up at different axial locations for G s = 250 kg/m2s and various superficial gas velocities Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers -cs I—H o O a -i-> cd 4-1 C/> •i— i 13 o o 0.5 0.4 0.3 0.2 0.1 0.0 0.5 0.4 0.3 0.2 0.1 0.0 0 10 15 10 15 20 1 1 II 1 20 O 0.0 0.1 0.5 0.4 0.3 0.2 0.1 0.0 0.0 0.5 0.4 0.3 0.2 0.1 0.0 0.0 0.5 0.4 0.3 0.2 0.1 0.0 0.5 0.4 0.3 0.2 0.1 0.0 0.2 0.3 0.4 0.5 r/R = 0.88 o.i 0.2 0.3 0.4 0.5 0.0 0.1 r/R = 0.50 o.i 0.2 0.3 0.4 0.5 r/R = 0.25 0.0 0.1 0.2 0.3 0.4 0.5 r/R = 0.00 0.0 0.1 0.2 0.3 0.4 0.5 Solid hold-up Figure 4.12 Local solid hold-up traces with corresponding probability distribution plots for six radial positions at z = 0.76 m with U g = 6 m/s and G s = 341 kg/m s (Sampling rate = 280 kHz , sampling time = 30 s) Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 67 The cross-sectional time-mean solids hold-up was measured, using the optical fibre probe, to be 0.155, a typical high-density value. A single low solid hold-up peak and very little fluctuation were found at the centre, indicating that the flow was relatively dilute (solids hold-up < 0.03). Moving outwards, the suspension became denser, the fluctuations started to increase and the solids hold-up was distributed bimodally (one high and one low concentration peak). Near the wall, the low solids hold-up peak was no longer observed and was replaced by a high concentration peak at £ s = 0.5 s esmf. Figure 4.13 plots radial profiles of standard deviation of local solids hold-up at different axial locations for U g = 6 m/s and various solids circulation fluxes. Corresponding time-mean solids hold-up profiles appear in Figure 4.9. As shown in Figure 4.13, in the top relatively dilute section of the riser, the solids hold-up fluctuations were relatively small, and there was little increase in the standard deviation close to the wall. In the bottom, dense section of the riser, on the other hand, the standard deviation reached a maximum between the axis and the wall. As the solids circulation flux increased, the peaks moved slightly towards the axis. These peaks were reported to be a distinctive feature of high-density CFB risers (Issangya, 1998 and Liu, 2001). Issangya (1998) suggested that the position of the maximum standard deviation might be used to identify a boundary between core and annulus regions. 4.2 Local particle velocity profiles Few studies have been conducted on the radial distribution of particle velocity in the dense section of CFB risers or in high-density CFB units operated at solids circulation fluxes beyond 200 kg/m2s due to the limitations of experimental set-ups and measurement systems, van Zoonen (1962) was among the first to determine net particle velocity profiles. He employed a Pitot tube in a 0.05 m diameter, 10 m high riser. He reported the particle velocities at the wall to be always positive. Utilizing a more accurate measurement system, Nieuwland et al. (1996) measured parabolic radial profiles of local particle velocities and positive particle velocities at the wall using an optical fibre probe in a CFB riser operating at high superficial gas velocities (7.5 m/s < U g < 15 m/s) and high solids fluxes (100 kg/m s < G s < 400 kg/m2s). They suggested that the shape of the radial particle velocity profiles has a Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers z = 4.42 m 3 o </> T3 O M-i O c o -4—» > T3 T3 00 G s = 204 kg/m s G s = 255 kg/m2s - • - G s = 123 kg/m s - O — G s = 203 kg/m2s -•— G s = 244kg/m2s - • - G s = 340 kg/m2s G s = 198 kg/iri s G s = 240 kg/rn s G s = 331 kg/m2s G s = 127 kg/m^s G s = 201 kg/m2s G s = 246 kg/m2s G s = 327 kg/m2s G s = 344 kg/m2s 0.20 0.15 0.10 0.05 0.00 O z = 0.76 m 0.0 1.0 G s = 124 kg/m s G s = 206 kg/m2s G s = 252 kg/m2s } s = 288 kg/rn s G s = 341 kg/m s 0.2 0.4 0.6 0.8 Radial location, r/R [-] Figure 4.13 Radial profiles of standard deviation of local solids hold-up at different axial locations for U e = 6 m/s and various solids circulation fluxes Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 69 weak dependence on U g and G s . In the lower part of their riser (h/H = 0.28), there was no net downward particle flow in the wall region. On the other hand, Wei et al. (1998) detected downward net particle velocities near the wall of a 0.186 m diameter riser operated at high-density conditions using an improved optical fibre laser Doppler velocimeter. After reviewing various reported methods for measuring local particle velocities in gas-solid suspensions, Zhu et al. (2001) suggested a compact five-fibre optical probe system. Liu (2001) drew attention to the need to investigate the effect of suspension density on particle velocities. Using a 3-fibre optical probe, he was able to simultaneously measure particle velocities, solids concentrations and fluxes in a riser of 0.076 m diameter, a component of the dual-loop HDCFB unit. In the present study, local instantaneous particle velocities were measured in a 0.20 m diameter riser of the fluid coker cold model with the aid of the optical fibre system developed by Liu (2001). Details of the particle velocity measurement and the calculation procedure appear in Chapter 3. Since the effect of riser diameter on particle velocity profiles has not been investigated previously, it was the purpose of this study to compare the findings of this work with those of Liu (2001). Before proceeding with the analyses, the procedure employed to determine time-mean particle velocities has to be clarified. Qian and Li (1994) drew attention to the miscalculation of the time-mean particle velocities. Many researchers calculated the time-mean particle velocities by Since particle velocity is dependent on particle concentration, instead of directly averaging it over time, it should be weighted with respect to particle concentration. Qian and Li (1994) defined the mean particle velocity as: (4.2) (4.3) where esis the time-mean solids hold-up and G sis the time-mean solids flux, defined respectively as (4.4) Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 70 Substituting G s into Eq. (4.3), \ = ^= J I V P . . ( t ) e , . ( t ) d t * (Vp,i(t))dt (4.6) Equation (4.6) indicates that time-average particle velocity has to involve weighting with respect to particle concentration. Therefore, in this study, the time-mean particle velocities were calculated from V ^^HKiW^Wldt (4.7) In Figure 4.14, radial profiles of time-averaged and concentration-weighted time-averaged particle velocities are compared for z = 0.76 m, U g = 6 m/s and G s = 341 kg/m2s. 0.0 0.2 0.4 0.6 0.8 1.0 Radial location, r/R [-] Figure 4.14 Comparison of radial profiles of time-average and concentration-weighted time-mean particle velocities at z = 0.76 m for U g = 6 m/s and G s = 341 kg/m2s Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 71 The concentration-weighted time-average particle velocities are seen to be smaller than the particle velocities averaged directly over time at most radial locations. The particle velocities were measured at five axial levels (z = 0.76, 1.27, 1.67, 3.10 and 4.42 m) and seven radial locations (r/R = 0.00, 0.25, 0.50, 0.75, 0.88, 0.94, 1.00). At each measurement location, three replicate measurements were performed. The averages of these measurements were taken and reported in this Chapter. In order to show the reproducibility of these measurements, the standard deviations were calculated and used in the calculation of error bars as shown in Figure 4.15. The error bars in this figure corresponds to 90% st confidence interval and cover V p w ± —j^- where s is the estimate of standard deviation Vn based on the n measurements and tn_i is the Student t. For 3 measurements (degree of freedom = n-1 = 2) and 90% confidence interval, tn.i is equal to 2.92. As can be seen in this figure, it was possible to perform reproducible measurements utilizing the 3-fibre optical probe. Figure 4.16 plots the local concentration-weighted time-mean particle velocity profiles at different axial locations for U g = 6 m/s and various solids circulation fluxes. Corresponding local time-mean solids hold-up profiles appear in Figure 4.9. The particle velocity reached a maximum at the centre and gradually decreased towards the wall. For most cases, an increase in solids circulation flux led to an increased local particle velocity in the central region. The increase in particle velocity with increasing G s was more apparent in the top, relatively dilute section, whereas towards the bottom of the riser, the radial profiles were not very sensitive to G s for the range covered. As seen in the figure, in the top section of the riser, negative velocities occurred near the wall, although their magnitudes were very small (< 1 m/s). As one moves from the top section of the riser to the bottom section, for the same operating conditions, the radial location where the particle velocity becomes negative moves outwards towards the wall. A similar trend was observed when the solids circulation flux was increased at a fixed axial location. The effect of superficial gas velocity on the radial concentration-weighted time-mean particle velocity distributions at different heights for U g = 5, 6, 7 and 8 m/s and G s = 250 kg/m2s is presented in Figure 4.17. Corresponding local time-mean solids hold-up profiles Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers i i 13 > & c '+-» 1-1 -»-» e <u o c o O Radial location, r/R [-] 0.0 0.2 0.4 0.6 0.8 Radial location, r/R [-] Figure 4.15 Reproducibility of local concentration-weighted time-mean particle velocity measurements performed by 3-fibre optical probe at z = 0.76 and z = 4.42 m for G = 250 kg/m2s and U = 6 m/s Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 73 z = 4.42 m 13 > 13 +-» u c 8 CD • 1—I o • i-H •4—» h <D O fl o U z = 1.67 m z= 1.27 m G s = 204 kg/rn s G s = 255 kg/m2s G s = 123 kg/rn s G s = 203 kg/m2s G s = 244 kg/m2s G s = 340 kg/m2s G s = 198 kg/rn s G s = 240 kg/m2s G s = 331 kg/m2s G s = 127 kg/m^s G s = 201 kg/m2s G s = 246 kg/m2s G s - 327 kg/m2s G s = 344 kg/m2s 0.0 z = 0.76 m - • - G s = 124 kg/n/s - O — G s = 206 kg/m2 s -•— G s = 252 kg/m2s - V — G s = 288 kg/m2s - D — G s = 341 kg/m2s Figure 4 0.2 0.4 0.6 0.8 Radial location, r/R [-] 16 Local concentration-weighted time-mean radial particle velocity profiles at different axial locations for U g = 6 m/s and various solids circulation fluxes Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 74 z = 4.42 m £ 16 > o o > O H C s i •a o • CD o e o U 0.0 z = 1.67 m z= 1.27 m z = 0.76 m ug = u = U s u„ u„ 5 m/s 6 m/s 7 m/s U8 = U s = U s = U = 5 m/s 6 m/s 7 m/s 8 m/s : 6 m/s : 7 m/s : 8 m/s U s = U c = U = 5 m/s 6 m/s 7 m/s 8 m/s : 6 m/s ; 7 m/s = 8 m/s 1.0 0.2 0.4 0.6 0.8 Radial location, r/R [-] Figure 4.17 Local concentration-averaged time-mean radial particle velocity profiles at different axial locations for G s = 250 kg/m2s and various superficial gas velocities Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 75 appear in Figure 4.11. As shown in Figure 4.17, as U g increased, the time-mean particle velocities increased in the core of the riser, while the wall region remained almost unchanged. As shown more clearly in Figure 4.18, the region with negative particle velocity values near the walls of the riser diminishes as one moves from the top towards the bottom of the riser at the same operating condition. The particle velocity profile also started flattening in the wall region at z = 1.27 m, suggesting increased uniformity of flow under dense suspension conditions. Similar trends were observed by Liu (2001) (see also Liu et al., 2002) in the dual-loop HDCFB unit as explained in Chapter 2. Figure 4.18 Local concentration-averaged time-mean radial particle velocity at different axial locations for U g = 6 m/s and G s = 330 kg/m s Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 76 In order to investigate the effect of diameter of the riser on the radial particle velocity profiles, the distributions obtained in this study were compared with those obtained by Liu (2001) in the HDCFB riser. Since Liu (2001) reported V p rather than V p w , V p values are compared in Figure 4.19. At each normalized axial location, z/H, the time-mean particle velocities measured in this study are greater than those obtained by Liu. This difference is consistent with the lower axial solids hold-ups obtained in this study compared to those in Liu's study under similar operating conditions (see Figure 4.4). In the wall region, on the other hand, similar profiles were observed in the two studies for the bottom part of the riser. Towards the top of the riser, negative particle velocities were encountered only in this study. The behaviour of solids in the wall region is discussed in more detail in Section 4.4 below. Fluctuations in the instantaneous local particle velocity reflect the turbulence of the local suspension flow. The standard deviation of particle velocities is related to the turbulent intensity and is a key parameter in modelling the suspension flow. Figure 4.20 shows 20 s traces describing the local variation of instantaneous particle velocity at six radial positions at z = 0.76 m for U g = 6 m/s and G s = 341 kg/m2s. The corresponding probability distributions are also plotted. As shown in Figure 4.20, in the dilute central region of the riser, the particle velocity was high and always positive. The fluctuations in this region had high amplitudes, although the solids hold-up fluctuations were of low amplitudes in this region (Figure 4.12). The corresponding probability distributions of particle velocity were relatively broad, while the corresponding solids hold-up probability distributions were very narrow. It can be inferred from these figures that in the dilute central region, a homogenous, relatively dilute suspension exists, where solids are flowing upwards at relatively high velocities. Towards the wall of the riser, the suspension became denser and the fluctuations in particle velocity decreased, while the fluctuations in solids hold-up started to increase. As the probability distributions of particle velocity became narrower, negative particle velocity values were encountered. Although a big fraction of particles had negative velocities at r/R = 0.94, as discussed by Bi et al. (1996) and Kirbas et al. (2002), this does not necessarily indicate that the net flow direction was downwards at that location because the velocity and solids hold-up are correlated. This point is discussed further in section 4.4 below. Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 77 |> > 1 c 03 1) to B -#— Liu (2001): G = 236 kg/rn s, z/H = 0.75 • O - This study: G s= 255 kg/m2s, z/H = 0.75 —•— Liu (2001): G = 214kg/m2s, z/H = 0.64 • O - This study: G s= 244 kg/m2s, z/H = 0.53 — L i u (2001): G = 202 kg/m2s, z/H = 0.39 • O - This study: G s= 240 kg/m2s, z/H = 0.28 15 H 12 9 6 3 0 -3 _ o.... "-a... — — 1 1 1 1 0.0 0.2 0.4 0.6 Radial location, r/R [-] 0.8 - • - Liu (2001): G = 228 kg/m2s, z/H = 0.14 • O - This study: G s= 252 kg/m2s, z m = 0.13 1.0 Figure 4.19 Comparison of local time-mean particle velocity profiles obtained in this study (Dnser = 203 mm) and in the HDCFB riser of Liu (2001) (D r i s er = 76 mm) Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 78 d. > > Ji -a 4—* .SP "53 & c _o a a u o c o o <*> o CO c S3 25 20 15 10 5 0 -5 25 20 15 10 10 15 20 10 15 20 •8 -8 -5 0 0.10 0.08 0.06 0.04 0.02 0.00 10 15 20 25 10 15 20 25 r/R = 0 . 2 5 -5 10 15 20 25 Time [s] Particle velocity, V p w [m/s] Figure 4.20 Local particle velocity traces and corresponding probability distribution plots at six radial positions for z = 0.76 m, U g = 6 m/s and G s = 341 kg/m2s (Sampling rate = 280 kHz , sampling time = 30 s) Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 79 4.3 Local solids flux profiles In order to have a complete understanding of the hydrodynamic properties of a flow regime, local solids circulation fluxes should also be measured and analysed. In most early studies, researchers utilized non-isokinetic sampling probes to measure radial profiles of solids flux (Monceaux et al., 1986; Bader et al., 1988; Rhodes et al., 1988). These studies were conducted in low-density CFB units, and the authors detected core-annulus flow structure with the solids moving downward at the wall. In the last decade, sampling probes have been employed to measure local solids fluxes in high-density circulating fluidised beds. Wei et al. (1997) measured local solids fluxes in a CFB riser operating at 18 kg/m2s < G s < 235 kg/m2s and 1.8 m/s < U g < 10.5 m/s. They observed three types of radial profiles: Parabolic/core-annular profile characterized by rapidly rising particles in a dilute core surrounded by slowly falling particles in a denser region adjacent to the riser wall; Radially uniform profile characterized by relatively uniform radial profiles of solids flux, with no downward flux at the riser wall; U-shaped profile characterized by higher solid flux at the wall than at the axis. Using an extraction probe, Karri and Knowlton (1998, 1999) measured solids flux profiles in a riser of 0.2 m diameter and observed that the time-mean fluxes in the wall region were negative at low gas velocities (3 to 5 m/s) and at all net solids circulation fluxes studied (49 to 580 kg/m2s). For these conditions, the radial solids flux profiles were found to be parabolic, with the highest flux occurring in the centre of the riser. At higher gas velocities and solids circulation fluxes, they observed the solids flow at the wall to be upwards. In addition, at high gas velocities the flux profiles were inverted relative to those at lower gas velocities. At high gas velocities and medium or high solids circulation fluxes, the local solids flux at the wall was greater than in the centre of the riser. Similar results were reported by Malcus (2000). Issangya (1998) measured profiles of local time-mean solids flux in the 76 mm diameter riser of the dual-loop HDCFB unit (see Figure 2.3) utilizing an "inverted-U" suction tube to Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 80 measure the upwards solids flux and an inclined tube to collect the downflowing solids by gravity. The net solids fluxes obtained by difference showed the profile to be roughly parabolic and indicated the net flow to be, on average, upwards both in the central region and adjacent to the wall. They also observed that the solids flux increased with solids circulation flux and was only slightly affected by the superficial gas velocity. Similar trends were reported by Liu (2001) who measured particle velocities, solids fluxes and solids concentrations simultaneously using a 3-fibre optical probe in the same HDCFB riser. In this section of the thesis, local solids fluxes measured employing the measurement system developed by Liu (2001) (see also Liu et al., 2003a and 2003b) are reported and compared with those obtained by Liu in the riser of smaller diameter. The local time-mean solids fluxes were measured at five axial levels (z = 0.76, 1.27, 1.67, 3.10 and 4.42 m) and seven radial locations (r/R = 0.00, 0.25, 0.50, 0.75, 0.88, 0.94, 1.00). At each measurement location, three replicate measurements were performed. The averages of these measurements are reported in this Chapter. In order to show the reproducibility of these measurements, the standard deviations were calculated and used in the calculation of error bars as shown in Figure 4.21. The error bars in this figure corresponds to 90% confidence — s t interval and cover G s ± —j^- where s is the estimate of standard deviation based on the n Vn measurements and tn.i is the Student t. For 3 measurements (degree of freedom = n-1 = 2) and 90% confidence interval, tn-i is equal to 2.92. As can be seen in this figure, reproducible results were obtained from three replicate measurements. The cross-sectional average net solids flux values were also compared with the overall solids circulation flux determined with the quick-closing valve in the return line. As discussed in greater detail in Chapter 3, the average absolute percentage deviation (Eq. (3.12)) was calculated to be 20%. Radial profiles of local solids flux at different heights for U g = 6 m/s and various solids circulation fluxes are depicted in Figure 4.22. Corresponding profiles of local time-mean solids hold-up and particle velocity appear in Figures 4.9 and 4.16, respectively. The time-mean solids fluxes were highest at the centre of the riser, decreasing towards the wall. Increasing G s caused the local solids flux to increase in the dilute core, especially in the top part of the riser. Closer to the bottom of the riser, on the other hand, profiles were not affected Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers C N f X 3 co "3 CO C K i (U s I e 500 400 J 300 0.0 0.2 0.4 0.6 0.8 1.0 Radial location, r/R [-] CN B "5b x 3 CO T3 C B i a 500 400 300 200 100 0.0 0.2 0.4 0.6 0.8 Radial location, r/R [-] Figure 4.21 Reproducibility of local time-mean solids flux measurements performed by utilizing 3-fibre optical probe at z = 0.76 m and z = 4.42 m for G q = 250 kg/m2s and U = 6 m/s Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 600 400 200 0 -200 600 400 200 0 -200 z = 4.42 m O v \j 1 1 1 1 1 z = 3.10m G s = 204 kg/rri s G = 255 kg/m2s G s = 123 kg/rn s G s = 203 kg/m2s G s = 244 kg/m2s G s = 340 kg/m2s z = 1.67 m G s = 198 kg/m s G s = 240 kg/m2s G s=331 kg/m2s G = 127 kg/rn s S D G s = 201 kg/m2s G s = 246 kg/m2s G s = 327 kg/m2s G s = 344 kg/m2s z = 0.76 m G s = 124 kg/m2s G s = 206 kg/m2s G s = 252 kg/m2s G s = 288 kg/m2s G =341 kg/m2s -200 0.0 0.2 0.4 0.6 0.8 1.0 Radial location, r/R [-] Figure 4.22 Radial profiles of local time-mean solids flux at different heights for IL = 6 m/s and various net solids circulation fluxes Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 83 significantly by G s over the range of measurement. This finding is consistent with the results of Malcus (2000). At G s = 341 kg/m2s and z = 0.76 m, a totally different profile is observed. Near the wall, the magnitude of the solids flux first increased, reached a maximum, then decreased. This is likely to have been caused by flow development in the wall region. This type of profile can be seen more clearly in Figure 4.23 where axial change of radial profiles of solids flux are plotted for U g = 6 m/s and G s = 330 kg/m2s at different levels in the riser. As can be seen in Figure 4.22, the time-mean solids fluxes in the top dilute section of the riser were measured to be negative at the wall indicating downward flow of particles at r/R = 1. Moving towards the dense bottom section, the solids fluxes were positive except for G s < 200 kg/m2s. Figure 4.23 Radial profiles of local time-mean solids fluxes at different heights for U g = 6 m/s and G s ~ 330 kg/m2s Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 84 The effect of superficial gas velocity on the radial profiles of time-mean solids flux at different heights is presented in Figure 4.24 for U g = 5, 6, 7 and 8 m/s and G s = 250 kg/m2s . Corresponding profiles of local time-mean solids hold-up and particle velocity appear in Figures 4.11 and 4.17, respectively. As shown in Figure 4.24, the superficial gas velocity had little impact on the profiles, except at z = 1.27 m and U g = 5 m/s. For this very dense condition, the radial profile was rather flat, with quite high upward solids fluxes at the wall. Liu (2001) in FfDCFB riser operating at U g = 8 m/s and G s = 542 kg/m2s and Karri and Knowlton (1999) in their 0.305 m diameter CFB riser operating at U g = 12.2 m/s and G s = 586 kg/m2s observed similar behaviour. It can be concluded from these findings that as the solids circulation flux increases, a point is reached where the time-mean flow in the riser is all upwards and relatively uniform. Since homogeneity affects the axial dispersion in CFB reactors, this is a highly desirable type of flow. In order to investigate the effect of riser diameter on the radial profiles of solids flux, the distributions obtained in this study are compared with those reported by Liu (2001) in the FfDCFB riser in Figure 4.25 for G s = 250 kg/m2s and U g = 6 m/s. The time-mean solids fluxes were found to be nearly equal in the central top section of the riser. The difference between the two sets of measurements increased closer to the wall. In the top section of the riser, in the wall region, the time-mean solids fluxes were larger for the smaller diameter riser. The opposite trend was observed for the bottom section of the riser in the wall region. The measurement system employed in this study enables local instantaneous flux measurements. The 20 s traces in Figure 4.26 show the local variation of instantaneous solids fluxes at six radial positions at z = 0.76 m for U g = 6 m/s and G s = 341 kg/m2s. The corresponding probability distributions are also plotted. As shown in this figure, the amplitudes of the fluctuations increased from the centre to the wall. The probability distributions demonstrate that negative solids fluxes occurred toward the wall of the riser. At r/R = 0.94, the local time-mean solids flux was measured to be negative. These solids flux profiles are utilized in the analysis of the flow in the wall region in the next section. Radial location, r/R Figure 4.24 Radial profiles of local time-mean solids flux at different heights for G s = 250 kg/m2s and different superficial gas velocities Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 86 700 Liu (2001): G = 236 kg/m s, z/H = 0.75 •O • • This study: G s = 255 kg/m2s, z/H = 0.75 - • - Liu (2001): G = 214 kg/m's, z /H = 0.64 - O • • This study: G s = 244 kg/m2s, z /H = 0.53 Liu (2001): G = 202 kg/m2s, z/H = 0.39 • • • O " This study: G = 240 kg/m2s, z /H = 0.28 - • - Liu (2001): G = 228 kg/m2s, z/H = 0.14 • • O • This study: G = 252 kg/m2s, z/H = 0.13 0.0 0.2 0.4 0.6 0.8 1.0 Radial location, r/R [-] Figure 4.25 Comparison of radial profiles of local time-mean solids flux obtained in this study ( D n s e r = 203 mm) with those determined in the HDCFB riser by Liu (2001) (D r i s er = 76 mm) [U g = 6 m/s] Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 87 X 00 T3 • i—I oo O <D C e 00 C cd o 10 15 20 2000 1000 -1000 -2000 2000 1000 0 -1000 -2000 2000 1000 0 -1000 -2000 2000 1000 0 -1000 -2000 10 15 20 10 15 20 a o S-H 10 15 20 5 10 15 20 Time [s] -2000 -1000 1000 2000 Solids flux [kg/m s] Figure 4.26 Local solids flux traces with corresponding probability distribution plots at six radial positions at z = 0.76 m for U g = 6 m/s and G s = 341 kg/m s (Sampling rate = 280 kHz , sampling time = 30 s) Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 88 4.4 Solids motion in the wall region In sections 4.1, 4.2, and 4.3, the particle velocities, solids concentrations and fluxes measured in the 0.2 m diameter riser of the coker cold model were analysed to explore the characteristics of the dense suspension upflow (DSU) regime, which has been reported to have distinctive features and trends distinguishing it from fast fluidization regime, such as time-mean upflow of solids at the wall, reduced axial mixing and a closer approach to plug flow (Kari and Knowlton, 1998, 1999; Issangya, 1998; Liu, 2001). However, as pointed out in section 4.1, due to the design and limitations of the units, it is not always possible for a dense suspension to persist over the entire riser height. When this is the case, it becomes very important to define operating conditions as well as the axial location at which transition to the DSU regime occurs. During these determinations, researchers have focused on the behaviour of particles and their flow direction in the wall region (Kari and Knowlton, 1998, 1999; Issangya, 1998; Liu, 2001). For this purpose, they defined a parameter called annular downflow layer thickness, 8 (=1- rc/R), where rc is the radial coordinate of the position where the time-mean net solids flow switches from being upwards to downwards. In the literature, 8 has been defined mainly in four different ways: a) 8vP = normalized distance to the wall from the point of zero particle velocity (Yang and Gautam,1995; Zhou et al., 1995; Parssinen and Zhu, 2001); b) SQS = normalized distance to the wall from the point of zero solids flux (Karri and Knowlton, 1998, 1999, 2002; Issangya et al., 2000; Liu, 2001; Kirbas et al. 2002); c) Smomentum = normalized distance to the wall from the point where the particle momentum flux passes through zero (Bai et al., 1995; Kim et al., 2004); d) 8rj = normalized distance to the wall from the point where the maximum standard deviation of local voidage fluctuations occurs (Issangya et al., 2000). As discussed by Bi et al. (1996), the annular downflow layer thickness differs according to which measured quantity or variable is taken. Issangya et al. (2000) also observed that 8 a and 8GS differ significantly, with 8 G s always being larger than 8C T. Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 89 In the comparisons given in literature, 8 values were measured using different devices and the measurements were not performed simultaneously. In order to be able to perform a complete comparison of 8 values, it would be best to measure them simultaneously using the same measurement system. Since the 3-fibre optical probe employed in this study enables simultaneous measurement of particle velocity, solids concentration and solids f lux, it was the first objective of this section to compare the two most widely used 8 values, Sv p and 8Gs, to clarify their relationship. In Figure 4.27, 8v p and 8G s are compared at several axial locations for U g = 6 m/s and three different solids circulation fluxes. In al l previous studies reporting 8v p , the authors did not specify whether they used the concentration-weighted particle velocities, V , (Eq. 4.7) or mean values obtained by directly averaging over time, V p (Eq. 4.2). Since most researchers did not acknowledge the difference between V and V p , it is assumed here that they employed V p measurements to obtain 8yP. Therefore, in Figure 4.27, V p measurements are used, instead of V , to determine 8vP. In all cases, 8y P is larger than 8G s at the bottom dense section of the riser, while the two values approach each other towards the top dilute section of the riser. According to B i et al. (1996), the difference between 8 values is due to the strong correlation between local instantaneous particle velocities and solids hold-ups. In order to understand the extent of interaction between particle velocities and solids hold-ups, a detailed analysis is pursued in this study. B i et al. (1996), based on theoretical analysis, pointed out that, due to interaction between local particle velocity fluctuations and solids concentration fluctuations, the time-mean particle flux and velocity do not fall to zero at the same radial position. A s a result, Sv P is not equal to 8Gs- Since in Figure 4.27 it was observed that 8Gs was approximately equal to 8v P in the dilute section of the riser, from the explanation of B i et al. (1996), it can be inferred that the correlations between velocity and solids hold-up fluctuations were less significant at those heights. Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 90 G s = 125 kg/m2s G s = 250 kg/m2 s G s = 340 kg/m2s 0,00 0.05 0.10 0.15 0.00 0.05 0.10 0.15 0.00 0.05 0.10 0.15 Annulur layer thickness Annulur layer thickness Annulur layer thickness Figure 4.27 Comparison of 5V p and 5Gs at different heights for U g = 6 m/s and three different net solids circulation fluxes Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 91 This can be proven with the help of the definitions of instantaneous particle velocity and solids concentration, in the forms e . = e + E ' (4.8) ^ S , l s s v ' V • = V + V (4.9) where the primes denote fluctuating components and the bars denote time-averages. Similarly, the instantaneous solids flux can be written as: G § l l =P pe l t lV I , 1 = P p { ( £ s + 0 ( V p + Vp')} (4.10) Combining equations 4.8 - 4.10 and time-averaging, while recognizing that e's = 0 and V' p -0, leads to G s = P p £ s v p = p p ( v p £ s + v ; e ' s ) (4.1D The second term in the brackets indicates the covariance of particle velocity and concentration and it shows the extent of interaction between the two. Unless the fluctuations in solid hold-up and particle velocity are independent, the covariance term cannot be neglected (Zhu and Soo, 1992). Therefore, it is also incorrect to assume that the product of local time-mean particle velocity and time-mean concentration gives the local time-mean solids flux, ie.: G ^ P p V ^ ( 4 1 2 ) For this reason, attempting to determine the third of the three time-mean quantities (G s , V p , E s ) from the other two will give erroneous results. In Figure 4.28, the effect of operating conditions on the value of the covariance term is examined for different axial locations with U g = 6 m/s and G s = 202 kg/m2s as well as 330 kg/m2s. Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers O c ca > o U 0.4 0.6 Radial location, r/R [-] 0.1 0.0 u > o u -0.1 H -0.2 -0.3 0.0 G s = 330 kg/mzs z = 0.76 m V o z= 1.27 m T z = 1.67 m v — z = 3.10m 0.2 0.4 0.6 Radial location, r/R [-] 0.8 1.0 Figure 4.28 Effect of operating conditions on radial profiles of covariance term, i.e., Vpe's for U g = 6 m/s and two values of solids circulation fluxes Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 93 It can be observed that in all cases the covariance was negative, i.e., the particle velocity and concentration were negatively correlated. The absolute value of the covariance increased as one moves toward the bottom dense section of the riser. In the top section of the riser, on the other hand, the covariance was very small or zero. For the same operating condition and height, the absolute value of the covariance was almost zero for r/R < 0.5 and increased steeply in the wall region. For dilute conditions, 8 V p and 8G s are very close to each other because the covariance is very small. Similarly, for dense conditions, the difference between 8Gs and 8vp can be explained by high covariance values, especially in the wall region where the annular layer thicknesses were determined. Since solids fluxes reflect fluctuations in both particle velocities and solids concentrations, it is believed that the flow direction in the wall region can be best characterized by solids flux measurements. Therefore, in the rest of this section, SG s is used for further analysis of the thickness of the wall region. Axial variations of 8Gs inside the riser for G s = 125 kg/m2s and 340 kg/m2s are depicted in Figures 4.29 and 4.30, respectively. In order to identify the flow characteristics in the riser under these operating conditions, the radial distributions of time-mean solids circulation fluxes and axial apparent solids hold-up profdes are also included. In the axial 8G s profiles, the annular layer thickness is assumed to be zero at both the inlet and exit of the riser. In these figures, all the important parameters needed to characterize the flow in the wall region are combined to give a complete picture of flow in the riser. At G s = 125 kg/m2s, the time-mean solids flux was downward near the wall at all measurement levels, with the thickness of the annular downflow layer passing through a maximum with increasing height. With increasing G s at fixed U g , as shown in Figure 4.30, the magnitude of the time-mean solids fluxes at each measurement location increased due to increasing solids hold-up. For the operating conditions investigated, no net downflow region was monitored below z = 1.27 m, indicating that the DSU regime was reached at the bottom of the riser. Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 94 6.0 CN I X 3 CS S o 3 p G 0) s z = 3.10m z= 1.27 m z = 0.76 m 0.0 0,2 0.4 0.6 0.8 1.0 Radial location, r/R Fast fluidization regime \ Flow transition__j surface ! DSU regime o v i — i — i — T -1.0 -0.8 -0.6 -0.4 -0.2 0.0 Radial location, r/R 5.0 4.0 3.0 2.0 1.0 0.0 T 0.00 0.05 0.10 0.15 Apparent solids hold-up [-] Figure 4.29 Radial profiles of time-mean solids circulation fluxes with corresponding boundaries of annular wall layer and axial apparent solids hold-up profile for U g = 6 m/s and G s = 125 kg/m2s Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 6 i r 6 CN I M^ X 3 53 C O 3 -150 z = 3.10 m z = 1.67 m z = 1.27 m z = 0.76 m 0.0 0.2 0.4 0.6 0.8 1.0 Radial location, r/R A A A Fast fluidization regime Dense suspension 0 upflow regime A 0 < r n — i — i — r -1.0 -0.8 -0.6 -0.4 -0.2 0.0 Radial location, r/R 0 ~ i — i — r 0.00 0.08 0.16 0.24 Apparent solids hold-up [-] Figure 4.30 Radial profiles of time-mean solids circulation fluxes with corresponding boundaries of annular wall layer and axial apparent solids hold-up profile for U p = 6 m/s and G s = 3 3 0 kg/m2s Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 96 As shown in Figures 4 . 29 and 4 .30 , the regime transition surface migrated upwards towards the riser exit when the solids circulation flux increased from G s = 125 to 3 4 0 kg/m2s. Issangya (1998) reported that the dense suspension persisted over the entire riser height as G s was increased further. However, this could not be monitored in this study due to the limitations of the experimental unit. The effects of solids circulation flux and superficial gas velocity on axial variation of annular wall layer thickness are plotted in Figure 4 .31 (a) and (b), respectively. Forz< 1.27 m: 8GS decreased when G s increased from 125 kg/m2s to 2 5 0 kg/m2s, while 8 G S was zero for both 2 5 0 kg/m2s and 3 3 0 kg/m2s; 8QS was zero for all superficial gas velocities; for 1.27 m<z< 4 .42 m: As G s increased from 125 kg/m2s to 2 5 0 kg/m2s, 8Q s started to increase, while an increase in G s from 2 5 0 to 3 3 0 kg/m2s caused 8 G S to decrease, Increasing superficial gas velocity ( U G ) , caused SG s to increase up to z = 4 .42 m. Figures 4 .32 , 4 .33 and 4 . 34 each plot change of apparent solids hold-up, local time-mean solids flux at r/R = 1 and annular wall layer thickness with overall superficial gas velocity at three heights all for U G = 6 m/s. In these figures, the arrows were located at the positions where the following was observed: (a) levelling off of apparent solids hold-up, (b) zero crossing of time-mean solids flux measured at r/R = 1, (c) disappearance of annular wall layer thickness. As it was discussed in Kim et al. (2004), the arrows indicate the transition from the fast fluidization flow regime to the dense suspension upflow regime based on each of these three quantities. The locations of these arrows for all three parameters correspond to approximately same solids circulation flux value in Figures 4 . 3 2 and 4 .33 . In Figure 4 .34, on the other hand, all three criteria indicated that the transition to dense suspension upflow regime had not been reached at this height (z = 3.1 m) for the highest solids circulation flux (Gs = 3 5 0 kg/m2s) investigated. Kim et al. (2004) utilized these three criteria in developing the regime map to predict the boundary between fast fluidization to dense suspension upflow regimes. Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers Figure 4.31 Change of annular wall layer thickness with (a) solids circulation flux for U g = 6 m/s (b) superficial gas velocity for G s = 250 kg/m2s Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 98 Figure 4.32 Variation of (a) Apparent solids hold-up, (b) time-mean solids flux at r/R =1, and (c) annular wall layer thickness, 5 r , with G at U = 6 m/s and z = 0.76 m Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 99 0.12 ? 0.10 rt CO T3 0.08 0.06 o C/3 I 0.04 < 0.02 T 3 2 -S £ a ^ I -4—> o 0.00 300 200 100 0 -100 0.15 t/3 .a 1 o.io rt •— 1 g o c c < 0.05 0.00 a) z = 1.27 m — i i i i i b) • i i i i i l - c) i 1—*"T"" — i I V "— 0 50 100 150 200 250 300 350 Solids circulation flux, G, [kg/m2s] Figure 4.33 Variation of (a) Apparent solids hold-up, (b) time-mean solids flux at r/R =1, and (c) annular wall layer thickness, 8Gs, with G s at U = 6 m/s and z = 1.27 m Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 100 Figure 4 . 3 4 Variation of (a) Apparent solids hold-up, (b) time-mean solids flux at r/R =1, and (c) annular wall layer thickness, 5„ , with G at U = 6 m/s and z = 3.1 m Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 101 In Figure 4.35, this flow regime map was adopted in examining the type of flow regime the unit was operated under while the measurements in this study were conducted. The boundary lines shown in Figure 4.35 correspond to the following equations developed by Kim et al. (2004): Boundary between fast fluidization and dilute pneumatic transport: (Upr/e) = 1 ( ^ \ -0.90 159^  G Ar" 0 2 0 (4-1 3) Ppesvt J Boundary between dense suspension upflow and fast fluidization: 0.59 Ar" 0 2 0 (4-1 4) Ppesv, Eqs. (4.13) and (4.14) covered the range of variables: 0.10 < (G/pp£,v,) < 61, 7.5 < [(t//e)/vj < 490, 4.7 <Ar< 97. As can be seen in Figure 4.35, the flow regime map does a very good job of distinguishing between the fast fluidization regime and dense suspension upflow regime. During the development of this flow regime map, it was assumed that the solids flow at the wall would be upwards in the pneumatic conveying region. However, in this study, the data points falling within this region showed downward flow at the wall. Further investigation is required around the boundary between fast fluidization and the pneumatic conveying regions. The annular wall layer thicknesses were finally compared in Figure 4.36 with those obtained by Liu (2001) employing the same 3-fibre optical probe in the 0.076 m diameter riser of the HDCFB unit. Corresponding local time-mean solids flux profiles appear in Figure 4.25. For this operating condition, the dimensionless annular wall layer thickness decreased with an increase in riser diameter in the bottom dense section of the riser. In the top section, the opposite behaviour was detected. Comparing three risers with diameters 66, 97 and 150 mm operating at relatively dilute condition (Ug = 1-4 m/s, G s = 20-80 kg/m2s), Bai et al. (1995) observed the same trend. Chapter 4 Solids Hold-ups and Motion in High-Density CFB 102 • Downflow at the wall Boundary between fast fluidization and dense suspension upflow regi Onset of dilute pneumatic conveying o Upflow at the wall Figure 4.35 Adoption of the flow regime map developed by Kim et al. (2004) to the data collected in this study Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 103 1.0 0.8 A ± 0.6 X ! S? X ! N O 0.4 H 0.2 O O O • Liu (2001): Gs = 220 kg/rn s O This study: Gs = 250 kg/m2s 0.0 - 1 - 1 1 r 0.0 0.1 0.2 0.3 0.4 0.5 Figure 4.36 Comparison of annular layer wall thicknesses obtained in this study (Dnser = 200 mm) and in HDCFB riser by Liu (2001) (DriSer = 76 mm) for U p = 6 m/s and similar net solids fluxes Chapter 4 Solids Hold-ups and Motion in High-Density CFB Risers 104 Based on the experimental axial profile of solids concentration, the riser can be vertically divided into two main zones, a dense zone at the lower section with no core-annulus structure and a dilute zone in the upper section subject to a clear core-annular flow structure. Axially, the flow appears to develop from a more radially homogeneous turbulent pattern to a core-annulus structure with increasing height. Turbulence and interparticle collisions appear to be two likely mechanisms that diminish the wall downflow region, while cluster formation and lack of lateral solids transfer are possible mechanisms leading to downflow. Chapter 5 Solids Mixing Studies in HDCFB Unit 105 CHAPTER 5 SOLIDS MIXING STUDIES IN HDCFB UNIT 5.1 Introduction Various techniques have been used to study the movement and mixing of solid particles. Among these, tracer experiments and residence time distributions are often most useful to reaction engineers. They provide a diagnostic tool for the detection of maldistributions and flow patterns inside a reactor. They are useful in measuring the parameters required for simplified flow models and to provide input and guidelines for creating and testing such models. Finally, understanding the residence time distribution greatly simplifies the solution of many problems in reactor design, including the choice of reactor configuration, optimization and scale-up. Numerous experimental techniques have been used to study solids mixing in fluidized beds. Potentially, any method capable of differentiating tracer particles via a characteristic that does not alter its overall flow behaviour is suitable. However, one must be careful since the choice of tracer particles, the method of introducing and detecting the tracer, and inappropriate boundary conditions can all cause experimental error. A tracer for solids mixing experiments should ideally be: hazard-free; inexpensive; - applicable to equipment of different scales; - usable repeatedly, without having to be separated from the bed material; able to provide a fast and accurate response; applicable over a wide range of concentrations; able to measure local tracer concentrations without perturbing the system; inert and non-interfering. Table 5.1 provides a compilation of relevant published work on solids mixing in fluidized beds. oo co a o CD 3 CD 2! > 2! > 4^ I 4^ 2 > n o P •z > Q_ ON oo O •a n _r o o EL T3 co _ o 2. eg O ^-CD % P CV 2 > 1 > c_ 3 3 m cf a. p CT> S o 8? to Os O c p a N CO P 3 . <=L tO tO U l U l Os Os 4^ Os 0 0 4^ U l U l o o oo c_ n o 3 > S ft © O b to I o o oo o P CTQ Ct © O O ft^ CD 3 CO CD _j' CTQ P 00 =s £ ft' o 3" p S" CTQ 3 CD on _ _ c o CD ^ — J I— o o o CO P a. CD p to to U l 4^ T j n n 3 N CD CL > 2 > SO fe p oo — 5 ^ c g H C3 B. CL p U l u> U l o b 0 0 I > T l N T j o o to cp O CD w co 3" 4^ U l o tO CD O " © P to r-> c CD l 8 ° X 0 0 U l to os bo O c P N CO p 3 a. 2 > 4^ SO CT P 3 CD •3 V, o P Tl CD a o ? 3 CD > 4^ VO U l o 00 c_ n o U l o o U l o o p ;> so cT so <-t © CD o U l b 4^ U l I U l * ^ 00 p 3 to 4^ - J O o Os O P p CfQ CC CTQ •_. CD c a. Os ft 0 0 3" 00 to - J O O Os > c ft-cr o 00 b- a-N ' CD CL f P H CD ^5. "D 3 •o H cs cfT. 3." 3. TJ ct p o (D p 3 3 era 3' g & N' CD CL cr CD CL CO n a-re u, 0 0 o s Co s C o 5 o Kondukov et al. (1964) Kojima et al. (1989) Helmrich et al. (1985) Harris et al. (2002b, 2000d, 2003) Du and Wei (2001) Du et al. (1999) Chesonis et al. (1991) Chesonis et al. (1990) Author 0.17 0.05 0.16 0.14 0.14 0.14 0.10 0.10 D(m) Size of bed o Ln b \ tO Ln bo 10.4 10.4 ON 4^ ON X g fluidized (m) 0.84-1.80 0.50-2.15 0.05 1 ON 2.3-9.0 2.20-7.84 u> Ln 3.5-4.5 a era • 1 0.009 1.1-26.9 5-60 1 ON O N O i 3 P Alumina silicate FCC X n o Phosphorescent particles FCC Alumina particles containing phosphor Alumina Alumina Type Particles N/A 1000 N/A 4060 1545 1710 3460 3460 3 U l Particles 2650-2900 ON O -J to O o -0 4^ LT\ 4^ Ni h-* to o Particles Alumina silicate tagged with Co60 FCC particles tagged with fluorescent dye era Crq sa ro D-°- 5' z a - i O £ U ) Phosphorescent particles Phosphor coated particles Alumina particles containing phosphor Petroleum coke CaCl2  impregnated on alumina Type Tracer particles N/A 1000 N/A 4060 760-10000 1710 2000 N/A l - o 3 ' Tracer particles 2888 ON O N/A to O o 15-1600 Ln 4^ ON 00 to a. It Tracer particles H 65 cr 5T 3 era c N O" CL c r CL o o 3 «—• 5' c CD CL c& C o o &• Co 3 OO Co On S' 8 o Muramoto et al. (1985) Morooka et al. (1983) Milne and Berruti (1991) May (1959) Martin et al. (1992) Lin et al. (1999) Lin et al. (1985) Author 0.12 0.07, 0.12 0.05 0.05-1.52 0.19 4.0 x 6.2 0.14 D(m) Size of bed to Ln tO Ln b ~o bo 11.7 22.8 N/A H(m) fluidized (m) 2.39, 5.25, 10.30 0.50-2.50 5.60 0.24 4.20-14.2 4.05-2.61 0.30 - 0.90 a era I 0-20 to N/A OJ in to £ <-n i u> 1 4^  Cn FCC Low alumina FCC Sand FCC FCC Coal and straw Glass beads Type Particles 900-1000 950-1100 2470 N/A 1560 N/A 420-600 ^ - D 3 * Particles 60-65 ' ON o o ON U\ o o , ON to N/A 2500 Particles FCC particles tagged with fluorescent dye FCC particles tagged with fluorescent dye Gallium-68 labelled wood and sand Wood particles Large wood particles FCC tagged by Iodine-132 and Tellerium 3 a. ui - J n 00 Manganese-56 Scandium-46 Type Tracer particles 900-1000 950-1100 2470 450 N/A N/A 3000 2500 3 ' Ul Tracer particles 60-65 ON O 106 370 1700-2360 N/A N/A to o o 420-600 C L • a B Tracer particles H 65 nT in 3 CTQ 5" c & N' n> C L cr ft C L cn o o 3 5' c ct C L TO C o o Co 3' C o *-* R a. Co O 00 Smolders and Baeyens (2000) Sitnai(1981) Schlichthaerle and Werther (2001) Roques et al. (1991) Rhodes et al. (1991) Ran et al. (2001) Patience et al. (1991) Author 0.10 1.20 x 1.20 0.30 x 1.00 0.10, 0.20 0.15-0.31 0.19 0.08 D(m) Size of bed ON u i 1.3-1.4 00 U l b ON to VO b U l b H(m) fluidized Km) 2.8-8.2 0.60-0.91 0.50 - 2.5 2.8-5.0 U l 1 U l U l 4^  1 ON Lo I ' CO N.—y 4^  i - J t to o i 00 VO 5-80 o I o o o to U l 1 Ov ON 3 P CO Sand Silica sand Quartz sand Phosphorescent particles FRF5 powder Alumina particles containing phosphor Sand Type Parti N/A 2580 N/A N/A 2456 1710 2630 EL* icles VO O o o o U l o o r -j o VO LLZ NaCI Ferromagnetic powder 00 c_ n o to Phosphorescent particles NaCI Alumina particles containing phosphor Si-28 converted to Al-28 Type Tracer 2200 3870 1530 N/A 2160 1710 2630 3 ' UJ particles h—* U l o ON - J o 7700 53-90 - j NO LLZ CI-TS H 65 5" U i 00 c o Ct> a. g 3 B g. ft" co O 3 co g_ 51 >< 5' Crq 5' S3 C 51 N' ft o. c r ft O -n o 3 5' c ct Wei et al. (1998) Wei et al. (1996) Wei et al. (1995) Wei et al. (1994) Viitanen (1993) van Zoonen (1962) Author 0.14 0.14 0.14 0.14 1.00 0.05 D(m) Size of fluidized bed (m) 10.4 15.0 12.0 Os 39.0 10.0 X ? Size of fluidized bed (m) 2.20 - 7.84 2.6 - 8.0 2.67 - 7.84 2.3 - 9.0 b 1.5 - 12.0 a e r a I CO 1 Os o 8-80 1 o 5-60 4>-OO 00 40-1000 t o CO Alumina particles containing phosphor Alumina particles containing phosphor Alumina particles containing phosphor Alumina particles containing phosphor FCC FCC Type Particles 1710 1710 1710 1710 2400 1600 3 ' Particles U l 4^ U l 4^ U l 4i. U l 4*. —j o h - to U l o o , I" Particles Alumina particles containing phosphor Alumina particles containing phosphor Alumina particles containing phosphor Alumina particles containing phosphor FCC labeled with FCC tagged with 5 wt% NH4CI Type Tracer particles 1710 1710 1710 1710 2400 N/A 3 ' Tracer particles U l 4*. U l U l 4^ U l 4^ -j o N/A Cu P 3, H cr 5" 3 CTQ 3" C S* N* o> a . <y CT> D-n o 3 3* C CH C -u, O Co 3 OQ U 3 s TO Co 3 ' s Zheng et al. (1992) Westphalen and Glicksman (1995) Westphalen and Glicksman (1995) Author 0.09 0.16 x 0.16 0.20 D(m) Size of bed 10.0 0 0 b X g fluidized On) 1.75-2.62 b 3.0-5.0 I CO VO I 4^  O 0-30 3 P to CO FCC FCC Quartz sand Type Particles VO o 1400 2350 /—s 3 ' Particles 4^  —] ui Ln 0 0 o Particles FCC tagged with NaCI Hot FCC particles Hot sand particles Type Tracer particles N/A 1400 2350 I T , 3"° Tracer particles N/A Ln Ln 0 0 o I H to ST 3 CTQ 3 ' c St N ' n a-cr cc CL n o 3 5" 3 cn CL •8 u , Co o &-On s OQ s On a' s Chapter 5 Solids Mixing Studies in HDCFB Unit 112 It is seen that there have been many studies, with a wide variety of tracer particles and experimental conditions. May (1959) was among the first to measure the axial dispersion coefficient of solids by employing radioactive tracers and to investigate the effect of column diameter on solids mixing. Following a similar approach, a number of researchers (e.g. Ambler et al., 1990; Milne and Berruti, 1991; Patience et al., 1991; Martin et al., 1992) have investigated the mechanisms of solids movement using different radioactive tracers in bubbling, turbulent and circulating fluidized bed (CFB) units. By using tracer particles containing Manganese56 in a commercial scale CFB boiler, Lin et al. (1999) highlighted the applicability of this technique in large-scale hot units. Recently, some researchers (e.g. Weinell et al., 1992, 1993; Garncarek et al., 1994; Mostoufi and Chaouki, 1999, 2001; Stein et al., 2000) have utilized single particle tracer techniques in fluidized beds. They were able to follow a single radioactive particle and construct a trajectory of its movement. Although the radioactive tracer and single particle-tracking methods have the advantage of being non-intrusive, they are often impractical because of safety concerns and permitting issues. Another disadvantage is that radioactive tracers can be used only for short, transient experiments after which the activity must be allowed to decay to negligible levels before proceeding with other experiments. In addition, in order to be detectable, the particle to be followed must not be smaller than ~ 500 pm in diameter. Due to safety issues, many researchers prefer to use other tracers like ferromagnetic or solid CO2 particles, or solids tagged with salt or organic solvents. Sitnai (1981) conducted residence time measurements using ferromagnetic tracers in an atmospheric fluidized bed, while Avidan and Yerushalmi (1985) utilized the same technique in bubbling, turbulent and slugging beds. Rhodes et al. (1991) employed sodium chloride as tracer, detected by dissolving samples in water and then measuring their electrical conductivity. Using a similar tracer, Zheng et al. (1992) and Smolders and Baeyens (2000) investigated core-annulus flow phenomena in fast fluidized beds. Chesonis et al. (1990) prepared tracer particles by impregnating porous alumina particles with CaCl2 and then used atomic absorption to determine tracer solid concentrations in samples of solids withdrawn from the bed. In a later experiment, Chesonis et al. (1991) measured solids mixing using petroleum coke tracer particles. Although the tracers mentioned above have been used extensively, they are tedious to prepare and can be used only for a limited number of experiments after which the tracer Chapter 5 Solids Mixing Studies in HDCFB Unit 113 material has to be separated from the bed material or the entire bed material has to be replaced. Therefore, they provide few data, often insufficient for full analysis. Another disadvantage is that they do not have a fast response and cannot measure transient local concentrations. As discussed by Wei and Zhu (1996) and Harris (2002), this hinders the accuracy of the residence time distribution concentration curves, especially for brief residence times and limited extents of mixing. Several research groups have employed fluorescent and phosphorescent tracer particles in order to eliminate these drawbacks. Morooka et al. (1983) and Kojima et al. (1989) used FCC particles dyed with fluorescent pigments. These particles emitted strong visible light upon excitation by ultraviolet (UV) light. The UV light was transmitted to the system by a multi-fibre optical probe, and visible light emitted by the particles was in turn transmitted through the same probe and detected by photomultipliers equipped with filters. In addition to examining axial solids dispersion in CFB units, Baba et al. (1984), Morooka et al. (1989) and Tayebi et al. (1999) were also able to measure local particle velocities with the same measurement system. Brewster and Seader (1980), Roques et al. (1993) and Wei et al (1994, 1995, 1996, 1998) preferred phosphorescent tracer particles. These have afterglow times longer than those of fluorescent particles. The emitted light of the phosphorescent tracer particles decays over time intervals of up to a few minutes. Based on this technique, they were able to measure solids tracer concentration vs. time at different radial and axial positions within circulating fluidized beds. Recently, Harris et al. (2001, 2002a, 2002b, 2003) improved the measurement of residence time distributions based on phosphorescent tracer particles by carefully defining the experimental boundary conditions. Closed boundary conditions were obtained by using an annular feeder fluidized bed at the entrance and an inlet jet mixer at the exit boundary. This review of experimental methods is presented to provide insight into the advantages and disadvantages of the techniques used in solids mixing studies. Among them, the phosphorescent tracer technique appears to be the most appropriate technique for high-density, high-flux systems. Therefore, in this study, FCC particles coated with phosphorescent material were selected as tracer particles for solids mixing studies because of the following advantages: Chapter 5 Solids Mixing Studies in HDCFB Unit 114 - The tracer particles are almost identical to the bed material in all but optical properties. - There is no tracer accumulation in the bed since, within a few minutes after addition, the light emission of the tracer particles decays to a point where they are not detectable, making these particles indistinguishable from the rest of the bed material. - The tracer particles can be detected by a light detector online in a rapid, simple and accurate manner. The measurement system has a fast response allowing accurate instantaneous concentration measurements at different radial and axial locations. In addition to being safer, this technique is less expensive than radioactive tracer methods. 5.2 Measurement system Local transient tracer concentrations and particle residence time distributions were measured by injecting pulses of phosphorescent pigment-coated FCC tracer particles, activated by high intensity UV light at the return leg of the high-density circulating fluidized bed unit described in Chapter 2. The solids mixing measurement system had three main components: - tracer particles, tracer activation and injection system, - tracer detection system. In the next sections, each of these components is discussed in detail. 5.2.1 Tracer particles The tracer particles were prepared by coating FCC particles with a phosphorescent pigment, B.C.W. Phosphor-G21, purchased from Beijing Chemical Works. The phosphorescent pigment is composed of ZnS:Cu particles of mean diameter 10-15 pm and density 4100 kg/m3. They emit visible radiation following exposure to light. Detectable emission can last up to several minutes, contrary to fluorescent particles which only radiate Chapter 5 Solids Mixing Studies in HDCFB Unit 115 while exposed to light. The optical properties of the phosphorescent pigment particles are tabulated in Table 5.2. Since the density and diameter of the phosphorescent particles differ significantly from those of the FCC particles, phosphorescent particles would very likely behave hydrodynamically differently when interspersed with FCC particles in a fluidized bed. Therefore, phosphorescent pigment particles cannot be used as separate tracer particles. Instead, it was decided to coat FCC particles with phosphorescent pigment to provide the tracer particles. Table 5.2 Optical properties of phosphorescent pigment particles Particles Phosphorescent pigment Emission colour Yellow-green Best activation wavelength 365 nm Emission wavelength 500-580 nm with a peak at 530 nm The tracer particles were prepared by affixing ZnS:Cu to FCC particles with a Polyvinyl Alcohol - water solution. The procedure is explained in detail in Appendix A. The key properties of the tracer particles are summarized in Table 5.3. For comparison purposes, properties of the phosphorescent pigment and FCC particles are also included in Table 5.3. Although the particle properties of the phosphorescent pigment and tracer particles are very different, their optical properties remained the same during their preparation. Scanning electron microscope (SEM) photographs of the tracer particles phosphorescent pigment and FCC particles in the size range 75 pm < dp < 106 pm appear in Figures 5.1 and 5.2(a), 5.2(b), respectively. In Figure 5.1, the small particles attached to each other and to bigger particles are phosphorescent pigment particles, whereas the larger underlying particles are FCC. As can be seen, even though the tracer particles passed through vigorous grinding during their preparation, they continued to adhere to the surface, aided by the powerful binder. Consequently, only a negligible fraction, if any, of the phosphorescent particles detached from the FCC particles during the tracer studies in the fluidized bed. Chapter 5 Solids Mixing Studies in HDCFB Unit Table 5.3 Properties of tracer particles used in solids mixing experiments 116 Particles Phosphorescent pigment FCC particles Tracer particles (Phosphorescent coated FCC particles) Particle density (kg/m3) 4100 1700 N/A Bulk density (kg/m3) 2706 850 844 Mean particle size, dp (pm) 10-15* 70** 72 given by the supplier Sauter mean diameter measured by sieve analysis Figure 5.1 SEM photograph of tracer particles Figure 5.2 SEM photographs of (a) phosphorescent pigment particles (b) FCC particles Chapter 5 Solids Mixing Studies in HDCFB Unit 118 5.2.2 Tracer activation and injection system The tracer particles were activated at the centre of the return leg (U-bend) of the HDCFB unit (Figure 5.3). The injection system is illustrated separately in Figure 5.4. The tracer particles were loaded into the injector via a ball valve from the top of the conical shaped expanded glass section connected to the 33 mm inner, 38 mm outer diameter and 1 m long quartz glass tube. Once the particles had been added, they were fluidized using the building air. The flow rate of the air was set, using a rotameter, to 5 x IO"4 m3/s, sufficient to fluidize the tracer particles vigorously and to provide a dilute suspension throughout the quartz glass tube. The expanded top section minimized the loss of particles to the bag filter. While the particles were being fluidized, they were photo-activated by four 25 Watt (0.015 m diameter, 0.330 m long) and four 40 Watt (0.02 m diameter, 0.62 m long) UV light bulbs. These bulbs were fastened on a rectangular thin aluminium-reflective plate as shown in Figure 5.5. Once the bulbs were fastened, the rectangular plate was then rolled into a cylinder surrounding the quartz glass tube of the injector by attaching the hooks at its ends (see Figure 5.5). In addition to holding the UV bulbs, this aluminium cylinder protected the environment from UV light exposure. It also augmented the light by keeping reflected light inside the injector system. In order to select the best light source to yield a sufficient number of excited tracer particles that can be detected in high-density systems, three different light sources, 500 Watt-Halogen worklight, 13 Watt 2-pin fluorescent lamp and 50 Watt spherical-shaped hand-held UV light, were investigated before the UV light bulbs were chosen, as shown in Figure 5.5. The properties of these light sources are given in Table B.l of Appendix B. The tests were performed by shining light on a plate of tracer particles and recording the intensity of light immediately after the light source was turned off. The light sources were fastened at the same location in each case, and the effect of the light source and the time of exposure of the particles to it were examined. The fluorescent lamp was found to be insufficient to excite the tracer particles. Figure 5.3 Schematic diagram of the high-density circulating fluidized bed unit with the injection system and the detection ports (Numbered from 1 to 4). For details of the activation-injection system see Figures 5.4 and 5.5 Chapter 5 Solids Mixing Studies in HDCFB Unit 120 Tracer input Rotameters Pressure gage Aluminium reflective cylinder (Figure 5.5) HEI Solenoid valve *• To bag filter 0.25 m 38 mm *. 280 mm T 1 m Quartz glass tube Ball valve Optical fibre ^ \ \ probe Detection box LZZ1 • o o Tracer injected to the riser To computer Figure 5.4 Tracer activation-injection system (see Figure 5.3 for its location relative to the HDCFB system) Chapter 5 Solids Mixing Studies in HDCFB Unit 121 0.88 m N\r#N-° Figure 5 .5 Configuration of U V light bulbs fastened on the aluminium reflective plate which was rolled into a cylinder around the quartz injector glass tube (See also Figure 5.4) As shown in Figure 5.6, the intensity of emitted light when the particles were excited by U V light was higher than that excited by visible light. Since the tracer particles can be excited in 2 ps (Harris, 2002b), the effect of exposure time was not a significant factor. Although the spherical hand-held U V light could be used in preliminary tests, a thin, long U V bulb was found to be more suitable for the tracer injection system. Therefore, in this study, the U V light bulbs shown in Figure 5.5 were deployed to activate the tracer particles. With this injection system, the tracer can be introduced as a pulse or as a step input. The pulse input produces a readily interpretable result and maximizes the potential for resolving the detailed structure. A step input, on the other hand, generates the system F-curve, which can be differentiated to yield the E-curve. However, differentiation of experimental data amplifies noise arising from sampling. Curve-fitting of the F-curve before integration can also lead to poorer estimates of the parameters. This arises because detailed features are less distinctive on the F-curve than on the E-curve. Therefore, in this study, the tracer particles were injected as a pulse input to the system. Chapter 5 Solids Mixing Studies in HDCFB Unit 122 4 > L__l 3 13 2 §> 1 0 A A A A o O O ° o o O Regent 500 Watt Halogen Worklight A Hand-held UV Light 0 20 40 60 Light exposure time [s] 80 100 Figure 5.6 Effect of light source and exposure time on the intensity of emitted light recorded right after the light was extinguished Since the pressure of the air in the injector tube and the mass of injected tracer particles affect the pulse characteristics, they needed to be selected carefully. Minimizing the mass of injected particles and increasing the pressure inside the injector would produce a sharp pulse with sufficient power to inject all of the particles in a brief period. However, the mass of particles injected must also be enough for tracer particles to be detected when they are mixed with FCC particles inside the HDCFB riser. Tests were performed with different masses of tracer particles, and the emitted light intensity from the activated tracer particles was recorded inside the riser. By trial and error, it was concluded that injection of 100 g of tracer particles was sufficient for them to be detected distinctly inside the riser. The pressure inside the injector was selected to be 16 psig (110 kPa), high enough to inject the tracer as a pulse input, but low enough not to break the quartz injector glass. Building air was used to pressurize the injector. The air entering with the tracer was calculated to increase the air flow inside the riser of the HDCFB unit by only 2-3% during the brief injection period of approximately 1-2 s. The pressure in the injector tube and the mass of tracer injected were kept constant in each experiment in order to have reproducible pulse inputs to the system. Different injector Chapter 5 Solids Mixing Studies in HDCFB Unit 123 fluidization times were also examined, and 5 minutes was selected as a time sufficient for the particles to be mixed and uniformly activated by the light source. In summary, 100 g of tracer particles were fluidized for 5 minutes inside the injector system while being activated by UV light. The injector (Figure 5.4) was then pressurized to 16 psig and the tracer was injected as a pulse input at the U-bend of the HDCFB unit (Figure 5.3). As demonstrated in Figure 5.7, this procedure with the selected specifications of the injector gave reasonably sharp input pulses. This injection system enabled for reproducible input pulses to the system. Time [s] Figure 5.7 Tracer pulse input recorded at point 1 in Figure 5.3 Chapter 5 Solids Mixing Studies in HDCFB Unit 124 5.2.3 Tracer detection system The intensity of the light emitted by the activated tracer particles was measured by a detection system purchased from the Institute of Chemical Metallurgy, Beijing, China. This system consists of a 12 mm diameter, 0.68 m long single-fibre optical probe and a control box where a photomultiplier and an A/D converter are located. During the solids mixing experiments, four detection systems were utilized simultaneously. The locations of these detectors on the HDCFB unit are indicated by circled numbers in Figure 5.3. The collected signals from each detection system were saved on a computer by means of data acquisition software written by C. J. Lim in the Visual Basic® programme. The sampling by each optical fibre probe was performed at 100 Hz for periods of 2 minutes. The experiments were conducted in the dark at night, with the HDCFB riser column covered by thick black curtains to ensure that there was no external light permeating through the walls to cause erroneous results. The strength of the light emitted by the tracer particles decays with time. Therefore, the recorded signals need to be corrected for this decay. The decay calibration curve was determined by utilizing the same injection system, with its outlet attached to a small black box where the optical probe was located (Figure 5.8). By maintaining the same injection specifications and procedures as described in section 5.2.2 above, the tracer particles were injected into the black box where they immediately accumulated around the probe and lost contact with the light. Therefore, they immediately started decaying. Consequently, the recorded signal gave the decay calibration curve. Based on 20 separate determinations, which were then averaged, the decay curve given in Figure 5.9 was obtained. Many different equation types were tested in attempt to fit the data (including exponential decay). The following calibration equation was found to give the best fit to these results: Ldecay(t) = Io(t+D' a 6 7 (5.1) with t having units of s. Here, Ijecay is the intensity of light emitted by the tracer particles and Io is the initial intensity of the emitted light. IQ is a function of the properties of the tracer particles and the activation light source only (Shionoya and Yen, 1999). Chapter 5 Solids Mixing Studies in HDCFB Unit 125 1 m Aluminium reflective cylinder (Figure 5.5) HEI 280 mm Quartz glass tube Solenoid valve Ball valve Detection box Black box Optical fibre probe a o o To computer Figure 5.8 Tracer decay curve calibration system Chapter 5 Solids Mixing Studies in HDCFB Unit 126 3.0-j 2.5 -> 2.0 -Time [s] Figure 5.9 Decay calibration curve averaged from 20 determinations The calibration equation presented by Wei et al. (1995) was of the form Idecay(t) = Iot-109 (5.2) Although this equation fitted their experimental decay curve data very well, it was not able to predict the intensity of the emitted light at t = 0 which should be IQ, rather than infinity. Therefore, in this study, Idecay was selected to be proportional to (t + l) n rather than tn. The change of intensity of the emitted light with the concentration of tracer in FCC + tracer mixtures was also examined during preliminary tests. Using the set-up presented in Figure 5.10, known masses of FCC particles were poured into a beaker and stirred by a magnetic stirrer. The optical fibre probe was dipped into the beaker. The Regent 500 Watt Halogen worklight was used to excite the tracer particles. Since the probe was immersed below the particle surface in the beaker, the only light it could detect was that emitted from the tracer particles. The intensity of the emitted light was recorded by the detection system and averaged over a sampling time of 100 s. By adding more tracer particles to the beaker and detecting the emitted light intensity, the relationship between the tracer concentration in FCC + tracer mixtures and light intensity was obtained, as illustrated in Figure 5.11. Chapter 5 Solids Mixing Studies in HDCFB Unit 127 500 Watt Halogen Worklight Optical fibre probe rH70mm I X 70 mm Stirrer o Detection box L" • o o To computer Figure 5.10 Experimental set-up used to test the change of intensity of the emitted light with the concentration of tracer in FCC + tracer mixtures Figure 5.11 Effect of concentration of tracer particles in tracer + FCC mixtures on the measured emitted light intensity Chapter 5 Solids Mixing Studies in HDCFB Unit 128 As shown in Figure 5.11, the relationship between the tracer concentration and light intensity can be assumed to be linear. By taking into consideration this linear relationship, together with the decay of the emitted light with time, the concentration of tracer in the FfDCFB unit at the position of the probe is calculated utilizing the measured optical fibre signal from: C(t) = I(t) ^ I(t) _ I(t)(t + 1) 0 6 7 C 0 Idecay(t) I 0 (t + ! ) - « 6 7 IG where I(t) is the intensity of the measured signal and C(t) is the local concentration of tracer in the HDCFB unit. 5.3 Data treatment procedure In order to obtain a complete understanding of local transient tracer concentrations, recorded signals were analysed according to the following procedure: 1) The measured voltage vs. time signal was loaded into the computer. 2) The time at which the injection of the tracer was first detected at the location labelled number 1 in Figure 5.3 was set to t = 0. 3) The measured signal was smoothed by the 5-point Fast Fourier Transform (FFT) filter provided by Origin® software to filter out background noise. This filtering technique was selected because of its accuracy and wide application in electrical signal analysis [Origin® software manual, 1999]. FFT filter smoothing was accomplished by removing Fourier components with frequencies higher than the cut-off frequency: (5.4) * cut-off n A.t where n is the number of data points specified, and At is the time between adjacent data points. Larger values of n result in lower cut-off frequencies and thus a greater degree of smoothing. The raw data are compared with the filtered data in Figure 5.12. Chapter 5 Solids Mixing Studies in HDCFB Unit 129 3.5 3.0 2.5 Raw data 5 pt-FFT smoothed signal > 0.0 0 1 3 4 5 6 Time [s] Figure 5.12 Comparison of raw data and data after 5 pt-FFT smoothing process 4) The signal baseline was adjusted to zero. 5) The signal was corrected for the decay of tracer emissions using the decay calibration equation (see Figure 5.13). 6) The signal was smoothed further utilizing a 30-point Fast Fourier Transform (FFT) filter using Origin® software to clarify features like peaks, determine peak width and simplify residence time calculation (see Figure 5.14) 7) For each set of conditions, two-to-four replicate measurements were performed. These measurements were then averaged to give a single result. This procedure damps out variations in the riser flow and their effect on the concentration profiles, while also helping to minimize injection errors (see Figure 5.15).The signals at each axial location were compared with each other, as well as with those recorded under different operating conditions of the HDCFB unit, to provide an understanding of solids mixing inside the system. (Signal recorded in HDCFB riser at z = 5.48 m and r/R = 0.0 operating at G s = 450 kg/m2s and U g = 6 m/s) Chapter 5 Solids Mixing Studies in HDCFB Unit 0 1 2 3 4 5 6 Time [s] Figure 5.13 Comparison of data after baseline adjustment and data after decay curve correction (Signal recorded in HDCFB riser at z = 5.48 m and r/R = 0.0 operating at G s = 450 kg/m2s and U g = 6 m/s) Decay-corrected signal 30 pt-FFT smoothed signal 0 1 2 3 4 5 6 Time [s] Figure 5.14 Comparison of data after decay curve correction and after 30 pt-FFT smoothing process (Signal recorded in HDCFB riser at z = 5.48 m and r/R = 0.0 operating at G s = 450 kg/m2s and U g = 6 m/s) Chapter 5 Solids Mixing Studies in HDCFB Unit 131 3.5 0 1 2 3 4 5 6 Time [s] Figure 5.15 Comparison of two replicate measurements and their average (Signal recorded in HDCFB riser at z = 5.48 m and r/R = 0.0 operating at G s = 450 kg/m2s and U g = 6 m/s) 5.4 Analysis of data In this study, local transient tracer concentrations were measured by injecting pulses of phosphorescent pigment-coated FCC tracer particles, activated by high-intensity U V light at the return leg of the dual-loop HDCFB unit described in Chapter 2. The intensity of the light emitted by the activated tracer particles was measured by a detection system. During the experiments, four detectors were utilized simultaneously. The locations of these detectors on the HDCFB unit are indicated by the circled numbers in Figure 5.3. At each axial location, the concentration of tracer was also measured at three radial positions (r/R = 0.0, 0.5, 0.9) by traversing the probe from the wall to the centre of the riser between injections. For each sampling location two-to-four replicate measurements were performed. The sampling by each detector was conducted at 100 Hz for periods of 2 minutes. According to the operating condition the HDCFB unit was operating, it would take 4-12 min for particles to travel through the dual loop of the HDCFB unit and come back to the injection position. Therefore, Chapter 5 Solids Mixing Studies in HDCFB Unit 132 as it was also observed in the raw signal collected during the experiments, none of the injected tracer particles stayed active when they were looping back to the system. The HDCFB unit was operated at three different superficial gas velocities (Ug = 4, 6, 8 m/s) and various solids circulation fluxes (Gs = 100, 250, 350, 450 kg /m2s). The signals collected at different radial and axial locations were compared with each other, as well as with those recorded under different operating conditions, to provide a broad understanding of solids mixing inside the HDCFB riser. The experiments were conducted in the dark at night, with the HDCFB riser column covered by thick black curtains to ensure that there was no external light permeating through the walls to cause erroneous results. To our knowledge, radial and axial tracer concentrations have not been measured in such a wide range of operating conditions. Measurements conducted at high-density conditions are especially scarce. Therefore, it was the objective of this study to explore solids mixing in HDCFB unit operating under different flow regimes, especially in the DSU regime. The effects of solids circulation rate on the concentration profiles measured at r/R = 0 for Ug = 4, 6 and 8 m/s are depicted in Figures 5.16, 5.17 and 5.18, respectively. Apparent solids hold-ups, measured simultaneously utilizing pressure transducers and calculated employing Eq. (2.5), are plotted in Figure 5.19. It was impossible to measure the tracer concentration at U g = 4 m/s and Gs= 450 kg/m2s since the gas velocity was insufficient to transport the solids steadily. Consequently, it was impossible to have steady state operation for solids fluxes beyond 350 kg/m2s at U g = 4 m/s. There was an electrical problem in the detector located at z = 1.55 m during the experiments when the unit was operating under U g = 6 m/s and G s = 250 kg/m2s. Therefore, that set of data is excluded in Figure 5.17. In all cases, the tracer was detected shortly after its injection (< 1 s) as part of the upward flow of particles in the core region without transfer to the wall region. This core flow of tracer was detected as the first peak. According to the operating condition, this peak was followed by a second smaller peak or by a long tail due to backmixing. \ Chapter 5 Solids Mixing Studies in HDCFB Unit 6 -5 -4 -3 -2 -1 -0 z = 5.48 m r/R = 0 G s =110 kg/nTs G s = 250 kg/nTs G s = 350 kg/nTs O u u z = 1.55 ni r/R = 0 14 z = 0.62 m r/R = 0 Figure 5.16 Effect of solids circulation rate on the concentration profiles at different axial locations and at r/R = 0 for U = 4 m/s Chapter 5 Solids Mixing Studies in HDCFB Unit 134 z = 5.48 ni r/R = 0 G s = 110kg/m : s ( J = 250 kg/m :s G s = 350 kg/nTs G = 450 kg/nrs Time [s] Figure 5.17 Effect of solids circulation rate on the concentration profiles at different axial locations and at r/R = 0 for TJ„ = 6 m/s Chapter 5 Solids Mixing Studies in HDCFB Unit Figure 5.18 Effect of solids circulation rate on the concentration profdes at different axial locations and at r/R = 0 for U , = 8 m/s Chapter 5 Solids Mixing Studies in HDCFB Unit 136 si 00 G s = 110 kg/m s G s = 250 kg/m2s G s = 350 kg/m2s i — i — r 0.00 0.05 0.10 0.15 0.20 0.25 Apparent solids hold-up [-] 110 kg/m2s - O - G s = 250 kg/m2s -V- G s = 350 kg/m2s - ^ G s = 450 kg/m2s i—i—i—r 0.00 0.05 0.10 0.15 0.20 0.25 Apparent solids hold-up [-] 110kg/m2s - O - G s = 250 kg/m2s -f- G s = 350 kg/m2s 450 kg/m2s i — i — i — r 0.00 0.05 0.10 0.15 0.20 0.25 Apparent solids hold-up [-] Figure 5.19 Axial profiles of apparent solids hold-up for various solids circulation fluxes and at U = 4, 6 and 8 m/s Chapter 5 Solids Mixing Studies in HDCFB Unit 137 As presented in Figure 5.16, at 4 m/s, tracer concentration profiles all had very long tails at each axial location and solids circulation flux, which implies high axial dispersion in the riser. This behaviour can be explained with the help of the local time-mean solids circulation flux distributions measured by Liu (2001) in the same unit at different heights and operating conditions using a 3-fibre optical probe. As can be seen in Figure 5.20(a), Liu observed negative net fluxes at the wall at U g = 4 m/s, indicating downward net particle flow along the wall. It is believed that the non-uniformity of flow caused by the downflow of particles could be responsible from the long tails of tracer concentration traces. In Figure 5.17, at 6 m/s, multi-peak and wide concentration profiles were only seen when the solids circulation flux was 100 kg/m2s. As G s increased to 250 kg/m2s, the first peak narrowed, and it was followed by one small peak, rather than multiple peaks observed at 100 kg/m2s. The second peak diminished with further increase in G s . At 8 m/s, all concentration distributions, except that measured at 250 kg/m s, had narrow peaks with relatively low dispersion. At G s = 250 kg/m2s, on the other hand, a significant second peak was observed. As can be seen in Figure 5.20(c), the thickest annular downflow layer was observed at this operating condition. It is believed that recirculation caused by the core-annulus flow structure corresponding to this operating condition was responsible for the second peak. Figures 5.21, 5.22 and 5.23 plot the effect of superficial gas velocity on the tracer concentration profiles at different levels and at r/R = 0 for G s = 250 kg/m2s, 350 kg/m2s and 450 kg/m2s, respectively. At G s = 250 kg/m2s, as the superficial gas velocity increased, the peaks in the concentration profiles narrowed. A more significant change was observed when U g was increased from 4 to 6 m/s. As U g increased, the rise time, i.e. the time at which the first peak was detected, decreased. This indicates an increase in particle velocity with increasing superficial gas velocity. This was also observed in particle velocity measurements conducted in the 0.20 m diameter fluid coker riser presented in section 4.2. Similar trends were observed at G s = 350 kg/m2s, as shown in Figure 5.22. At G s = 450 kg/m2s, the profiles had very narrow peaks with no significant tail. Net upward flow for these operating conditions attests to the structure of these concentration profiles. Chapter 5 Solids Mixing Studies in HDCFB Unit (a) Gs (kg/m s) - • - 2 3 6 355 417 (b) (c) Figure 5.20 Radial profiles of local time-mean solids flux measured by Liu (2001) at z = 4.2 m for various net solids circulation fluxes: (a) U g = 4 m/s; (b) 6 m/s; (c) 8 m/s Chapter 5 Solids Mixing Studies in HDCFB Unit 139 Figure 5 .21 Effect of superficial gas velocity on the concentration traces at different axial locations and at r/R = 0 for G = 250 kg/m2s 8 ° Chapter 5 Solids Mixing Studies in HDCFB Unit 140 6 - z = 5.48 in U„ = 4 m/s c r/R = 0 U =6 m/s 5 - j. U =8 m/s 4 -3 -2 -1 -0 - i i i i i i 0 2 4 6 8 10 12 14 z = 1.55 in r/R = 0 0 2 4 6 8 10 12 14 Time [s] Figure 5.22 Effect of superficial gas velocity on the concentration traces at different axial locations at r/R = 0 for G s = 350 kg/m2s Chapter 5 Solids Mixing Studies in HDCFB Unit Figure 5.23 Effect of superficial gas velocity on the concentration traces at different axial locations and at r/R = 0 for G s = 450 kg/m2s Chapter 5 Solids Mixing Studies in HDCFB Unit 142 These comparisons together with the hydrodynamic data collected at the same unit by Liu (2001) prove that at high superficial gas velocity and solids circulation flux, the HDCFB unit operated in the dense suspension upflow regime which has distinctive features and trends distinguishing it from the fast fluidization regime, in particular time-mean upflow of solids at the wall, reduced axial mixing and a closer approach to plug flow (Karri and Knowlton, 1998, 1999; Issangya, 1998; Liu, 2001). In this study, radial profiles of local transient tracer concentrations were also measured at three radial positions (r/R = 0.0, 0.5, 0.9) at each axial location for each operating condition. Figures 5.24 and 5.25 plot tracer concentration traces measured at 3 different radial locations and axial locations for G s = 110 kg/m2s, U g = 6 m/s, and for G s = 450 kg/m2s, U g = 6 m/s, respectively. These two conditions were selected to further investigate the difference between a dense and a relatively dilute condition. In all cases, the rise time, i.e. the time at which the first peak was detected, was larger at r/R = 0.9 than other radial positions. Rhodes et al. (1991) also reported that particles in the region near the riser wall had greater residence times compared to those in the centre of the riser and suggested that these results were in accordance with fluid dynamic observations in this flow regime, thereby indicating the existence of a core-annulus structure with higher solids velocities in the core than in the annulus. In each case, the dispersion of the concentration traces increased closer to the wall. At G s = 450 kg/m2s and z = 0.62 m, however, the concentration profiles showed very similar behaviour. It is believed that this indicates the existence of dense suspension upflow regime in this region where there is reduced axial mixing and closer approach to plug flow. Although the above discussion provides important and detailed qualitative information on solids mixing inside the riser at different locations and under various operating conditions, quantitative comparison is also required. Therefore, in the next section, an axial dispersion model is proposed and employed for further analysis of solids mixing in the riser of the HDCFB unit. Chapter 5 Solids Mixing Studies in HDCFB Unit 0 2 4 6 8 10 12 14 z = 1.55 m 0 2 4 6 8 10 12 14 8 H 0 2 4 6 8 10 12 14 Time [s] Figure 5.24 Tracer concentration traces at three different radial positions and axial locations for Ua = 6 m/s and G s = 110 kg/m2s Chapter 5 Solids Mixing Studies in HDCFB Unit z = 5.48 m r/R = 0.9 r/R = 0.5 r/R = 0.0 14 Time [s] Figure 5.25 Tracer concentration traces at three different radial positions and axial locations for U C T = 6 m/s and G s = 450 kg/m"s Chapter 5 Solids Mixing Studies in HDCFB Unit 145 5.5 Modelling A wide variety of models have been used to describe solids mixing in fluidized beds. The two most common types are dispersion models and models based on core-annular flow structure. Berruti and Kalogerakis (1989), Ambler et al. (1990), Berruti et al. (1995), and Smolders and Baeyens (2000) successfully employed models based on a core-annular flow structure. Their dispersion models were based on a hydrodynamic model defining the two phase flow characteristics of the riser. Mixing was modelled in terms of particle interchange between the core and annulus. However, the majority of researchers (van Zoonen, 1962; Avidan and Yerushalmi, 1985; Kojima et al., 1989; Patience et al., 1991; Rhodes et al., 1991; Bai et al., 1992; Zheng et al., 1992; Viitanen, 1993; Wei et al, 1994, 1995, 1996, 1998; Harris, 2003) employed dispersion models. In most of the studies, the extent of solids mixing was presented in terms of a single dispersion coefficient (axial dispersion coefficient). In order to be able to compare the results of this study with those in the literature, an axial dispersion model was employed in this study. Although it is a highly simplified method, a dispersion model may be a reasonable choice given that both the gas and solids do not deviate greatly from plug flow.. 5.5.1 Axial dispersion model The axial-dispersion model can be written as ^ = D > f £ - v ^ (55) where C is the concentration of tracer particles at time t. This equation assumes convective transport of solids with a one-dimensional velocity V p in the upward direction, superimposed by axial dispersion characterized by a constant axial dispersion coefficient, D a x (Werther and Hirschberg, 1997). In order to be able to solve Eq. (5.5), the particle velocity, axial dispersion coefficient, and initial and boundary conditions should be specified. Chapter 5 Solids Mixing Studies in HDCFB Unit 146 5.5.1.1 Initial and boundary conditions Specification of initial and boundary conditions depends upon how the tracer particles are introduced and measured within the bed. There are two kinds of boundary conditions, i.e. open and closed. In closed boundary conditions, fluid enters and leaves solely by plug flow, and thus without re-entry. Varying velocities, back diffusion, swirls and eddies are not permitted at the boundary. An open boundary condition, on the other hand, presumes that the same flow distribution and mixing behaviour occur at the boundary as elsewhere in the column. For a perfect pulse injection, analytical solution of Eq. 5.5 is possible for both open and closed boundary conditions (Levenspiel, 1996). However, since in this study, the tracer was injected at the return leg of the high-density circulating fluidized bed (see Figure 5.3), in order to obtain the dispersion only in the riser, the dispersion in the return leg of the riser has to be eliminated from the overall dispersion. As shown in Figure 5.3, at the connection of the return leg, air entrance and the venturi section at the bottom of the riser, there will be a highly mixed zone, which for many operating conditions behaves very differently from plug flow. Therefore, this section has to be either modelled separately or subtracted from the overall dispersion calculated by the model. Since tracer was also detected at z = 0.62 m in this study, those measurements can be utilized to provide the upstream boundary condition to avoid this problem. As discussed in Chapter 4, in CFB risers, there are usually two axial mixing regions: a dense region at the riser base and a relatively dilute region in the upper part of the riser. Since the flow patterns in these regions are different, the dispersion coefficients representing each zone can also differ significantly from each other. Therefore, for solids mixing modelling of these systems, dividing the riser into two sections and determining the axial dispersion coefficients for each section separately would give more accurate results. Since, in this study, the tracer concentrations were also measured at z = 1.55 m, these measurements can be employed as a boundary. Although z = 1.55 m does not directly correspond to the hydrodynamic height of the bottom dense section since this height changes according to operating conditions, it is still useful in dividing the riser into two sections. With the help of Figure 5.26, the boundary conditions for each zone can be written as: Chapter 5 Solids Mixing Studies in HDCFB Unit 147 Section 2 Section 1 j = 394 V P.5 j =212 44 V. P.4 j = 108 . Vp,3 j = l i = 94 V P.2 i = 47 V p.i - 1 x i =  _-z = 5.48 m D ax,2 i_z = 1.55 m z = 1.55 m D ax,l z = 0.62 m Figure 5.26 The grid distribution together with the assigned particle velocities and dispersion coefficients used in the solution of Eq. (5.5) Chapter 5 Solids Mixing Studies in HDCFB Unit 148 Section 1: B.C.I: @ z = 0.62m, C = C 2 = 0.62m ( 5 - 6 ) B.C. 2: @ z= 1.55 m, C = C z = 1 . 5 5 m ( 5 - 7 ) Section 2: B.C.I: @ z= 1.55m, C = C Z = i.55m ( 5 - 8 ) B.C. 2: @ z = 5.48 m, C = C z = 5.48m ( 5 - 9 ) The following initial condition was utilized in solution of Eq. (5.5) for each zone: LC. : @ t<0, C = 0 (5.10) 5.5.1.2 Solution ofEq. (5.5) Since analytical solution of Eq. (5.5) with the given initial and boundary conditions was impossible, it was numerically solved using Matlab Solver. For this purpose, the partial differential equation presented in Eq. (5.5) was first transformed into an ordinary differential equation (ODE). This transformation was carried out by utilizing a 2-point upwind differencing scheme for the spatial derivatives 3C/3z and d2C/dz2 for each section as: SecionJ: ^ = P. , f C " ' 2 C J + C » 1 - V p \ ^ - \ (5.11) dt ax'' 1 Az 2 j p 1 Az J Section2- — = D \?2L2Sl±S*±\ - V B - ! ^ L L (5.12) ti n z. ^ u a X i 2 p A z r where i = 2, N-l, j = 2, M-land N, M and Az are the total number of grid points and grid spacing, respectively. The boundary conditions can also be re-written as Section 1: B.C.I: @ i = l , C = C 2 = 0.62m ( 5 - 1 3 ) B.C.2: @ i = N, C N = C z = 1.55m ( 5 - 1 4 ) Chapter 5 Solids Mixing Studies in HDCFB Unit 149 Section 2: B.C.I: @ j = l , C = C z = 1 . 5 5 m (5-15) B.C.2: @ j = M, C M =CZ = 5 .48m (5.16) It was assumed that the same flow distribution and mixing behaviour occur at i = N and N-l and j = M and M- l , respectively. The two dispersion coefficients, D a X ) i and D a X ) 2 , were then obtained utilizing the least square regression method by matching the calculated concentration values at i = N-l and j = M-l with measured concentration values at i = N and j = M , respectively. The effect of value of the grid spacing, Az, on the accuracy of the solution was examined, and it was concluded that the difference between the measured and calculated concentration increased with decreasing Az. However, for Az < 0.01, no significant improvement was observed with a further decrease in Az. Therefore, in this study it was chosen to be 0.01. The spatial grid can be seen more clearly in Figure 5.26. Eqs. (5.11) and (5.12) were written for all interior grid points, which led to N and M ordinary differential equations in section 1 and section 2, respectively. Starting from the initial condition for concentration at all grid points, together with the required boundary conditions, the right-hand side of Eqs. (5.11) and (5.12) were calculated and the resulting ODEs were sent to an ODE solver. The values of concentrations at the next time step were made available by the ODE solver. This procedure continued until a specified time, T. In this study, since the sampling was continued for 2 minutes, T was chosen as 2 min. RKF45 (Runge-Kutta-Fehlberg) provided in Matlab solver was utilized as the ODE solver. The 45 indicates that a fourth-order Runge-Kutta method is embedded in a fifth-order Runge-Kutta method. Essentially, a solution to the ODEs is computed with the fourth-order Runge-Kutta algorithm and also with the fifth-order Runge-Kutta algorithm. The two solutions are then compared to estimate the error in the numerical solution. If the estimated error exceeds the user-specified error tolerance, the solution is repeated with a smaller time step, to improve the accuracy of integration over time. Chapter 5 Solids Mixing Studies in HDCFB Unit 150 5.5.1.3 Measurement of Vp The average particle velocity can be calculated from: V P , m = (5.17) Pp ^s, apparent where G s is the overall solids circulation flux and es,apparent is the apparent solids hold-up inferred from direct measurement of differential pressures. Five pressure transducers across equal intervals along the riser were used to measure the apparent solids hold-up. Therefore, five different particle velocities were calculated along the riser. As discussed in Chapter 4, Eq. (5.17) is not strictly correct since variations in Gs,j and eSii are likely to be correlated. However, radial profiles of these quantities were not available at the heights of interest, so we were forced to use this equation. Its effect on the dispersion coefficients is expected to be small. In accordance with the schematic in Figure 5.26, the following ODEs established for the two sections of the riser are as follows: Section 1: between i = 2 - 46 between i = 47 - 93 d t - 0 - 1 ! I p - 2 r ^ z j (5-19) Section 2: between j = 2 -107 dC f C i + 1 - 2 0 = + C n ] f C i - C , . ] between j = 108 - 211 dt "'21 A z 2 I ' • ' j Az J P ' Chapter 5 Solids Mixing Studies in HDCFB Unit 151 between j = 212 - 393 = D dC [ C j + 1 -2Cj + C H C J - C H (5.22) dt " a x ' 2 A z 2 P ' 5 | Az These ODEs were then solved simultaneously by following the procedure explained in section 5.5.1.2. 5.6 Results and discussion of the model Experimental concentrations are compared with those obtained from the axial dispersion model at z = 1.55 m and z = 5.48 m for G s = 450 kg/m 2s, U g = 6 m/s and G s = 450 kg/m 2s, U g = 8 m/s in Figures 5.27 and 5.28, respectively. Since the model does not take into account the radial change of tracer concentration, only concentrations measured at the centre of the riser (r/R = 0) were used in the fitting and comparisons. As seen in Figures 5.27 and 5.28, the model predicts the shape of the distribution and the arrival times of the two characteristic peaks accurately. The accuracy, especially at z = 5.48 m, increased as the superficial gas velocity increased. A s discussed in section 5.4, the dispersion in the tracer concentration distribution decreased as the superficial gas velocity increased. A s dispersion decreases, the axial dispersion model fits the experimental data more accurately. The change of axial dispersion coefficients, D a x i and D a X 2 , with superficial gas velocity and solids circulation flux is presented in Figures 5.29 and 5.30, respectively. For all cases except at G s = 110 kg/m 2s at the bottom section of the riser, as the superficial gas velocity increased, dispersion also increased. The impact was more significant in the top section of the riser. Similar trends were observed when overall solids circulation flux was increased at U g = 8 m/s. However, at U g = 4 and 6 m/s, different trends were observed. It is believed that the dispersion coefficient can be significantly different according to the flow regime existing in the section under examination. In order to examine the type of flow regime the unit was operated under while these measurements were conducted, the flow regime map developed by K i m et al. (2004) was adopted. Since the cross-sectional average solids hold-up was employed Chapter 5 Solids Mixing Studies in HDCFB Unit 0 2 4 6 8 10 12 14 Time [s] 0 2 4 6 8 10 12 14 Time [s] Figure 5.27 Comparison of experimental concentration values with those from axial dispersion model at z = 1. 55 m and z = 5.48 m for G = 450 kg/m2s, U = 6 m/s Chapter 5 Solids Mixing Studies in HDCFB Unit 153 0 2 4 6 8 10 12 14 Time [s] 0 2 4 6 8 10 12 14 Time [s] Figure 5.28 Comparison of experimental concentration values with those from axial dispersion model at z = 1. 55 m and z = 5.48 m for G, = 450 kg/m2s, U = 8 m/s Chapter 5 Solids Mixing Studies in HDCFB Unit 154 Between z = 0.62 m - z = 1.55 m Between z = 1.55 m - z = 5.48 m CD o o c •— o-1/1 CM 8 • 6 • 4 2 • 0 8 • 6 4 2 • 0 8 6 4 2 0 G =250 kg/m s G =350kg/m2s -i 1 1 r-G =450 kg/nTs G = HOkg/m2 s 8 • 6 4 2 0 8 6 4 2 0 8 6 4 2 0 G = 250 kg/m2s G =450 kg/m2s 6 7 8 9 3 4 5 Superficial gas velocity, U (m/s) -r-7 G s = 350 kg/m2s o o o Figure 5.29 Effect of superficial gas velocity on axial solids dispersion coefficients for G s = 110, 250, 350 and 450 kg/m2s Chapter 5 Solids Mixing Studies in HDCFB Unit 155 Between z = 0.62 m - z = 1.55 m Between z = 1.55 m - z = 5.48 m 8 400 400 100 200 300 400 100 200 300 400 Solids circulation flux, G (kg/m s) Figure 5.30 Effect of solids circulation rate on axial solids dispersion coefficients for U g = 4, 6 and 8 m/s Chapter 5 Solids Mixing Studies in HDCFB Unit 156 in developing this regime map, the flow regime in the bottom and top sections of the riser can be deduced separately. The details of this flow regime map are given in Chapter 4. As can be seen in Figure 5.31, most of the measurements were performed while the unit was operated under the fast fluidization regime. According to this flow regime map, the top section of the riser (z = 1.55 m - 5.48 m) was always operated under fast fluidization regime. At the bottom dense section of the riser, on the other hand, according to the operating condition, both fast fluidization and dense suspension upflow regime were observed. As can be seen in Figure 5.31, all data corresponding to DSU regime were located in the close proximity of the FF/DSU regime transition boundary. Therefore, most of the data collected at the bottom section of the riser fell into the FF/DSU transition region. Although it is expected that the axial solids dispersion coefficient at the DSU regime would decrease because of the disappearance of the net downward flow of particles at the wall, the trend observed in Figures 5.29 and 5.30 seems to suggest that fully developed dense suspension upflow regime was not reached in this study. The increase of dispersion coefficient with increase in overall solids circulation flux is also believed to indicate that just increasing solids circulation flux at a fixed U g is insufficient to achieve or maintain dense suspension upflow regime in a system. Both U„ and G s should be increased in order to be able to reach the DSU regime. In the studies presented in Table 5.1, it was reported that solids axial dispersion coefficients were in the range of 0.01 to 10 m2/s. However, these measurements were mostly conducted in units operated at low superficial gas velocities and modest solid circulation fluxes. The higher axial dispersion coefficients observed in this study, compared to those in the aforementioned studies, are believed to be due to the difference in operating regimes. Therefore, direct comparison of this study's results with those in the literature is not possible. However the axial solid dispersion coefficients can be compared with the axial gas dispersion coefficients obtained in the same unit by Liu (2001) (see also Liu et al., 1999). Note that the gas dispersion values from Liu were obtained based on his simple one-dimensional gas dispersion model without consideration of radial profiles. As can be seen in Figure 5.32, gas and solids dispersion coefficients were in the same order of magnitude and showed similar trends with the change in superficial gas velocity. However, there is still a need for further studies in order to be able to compare them more extensively. Chapter 5 Solids Mixing Studies in HDCFB Unit 157 • Dense Suspension Upflow regime Boundary between fast fluidization and dense suspension upflow regime O Fast Fluidization regime 25 -i 1 Figure 5.31 Flow regime map Chapter 5 Solids Mixing Studies in HDCFB Unit 158 Ideally, radial dispersion, in addition to axial dispersion, needs to be taken into account while modelling the solids mixing in CFB systems. Wei et al. (1995) proposed a two-dimensional axially dispersed model described by at a 2 c DJ dc} dz2 + • 3r V r ac a z (5.23) With this model, they were able to successfully determine the radial dispersion coefficient, Dr, in addition to the axial dispersion coefficient. An important recommendation of this thesis for the future (see Chapter 6) is to build a similar model and to determine radial, as well as axial, dispersion coefficients. Chapter 5 Solids Mixing Studies in HDCFB Unit 159 Gs ~ 250 kg/m s 7 6 < r t l 8 o S a 2 * i a < o O O g "e T3 8 7 6 5 4 3 2 1 0 5 6 7 I L (m/s) o Gas dispersion coefficient • Solids dispersion coefficient Gs ~ 350 kg/m s —r-4 U g (m/s) Figure 5.32 Comparison of axial solids dispersion coefficients obtained in this study with the axial gas dispersion coefficients obtained by L iu (2001) in the same unit (The L iu values are those based on the simple one-dimensional gas dispersion model) Chapter 6 Conclusions 160 CHAPTER 6 CONCLUSIONS 6.1 Summary of major conclusions With the principal objective of obtaining a complete overall picture of flow structure in high-density CFB systems, extensive measurements involving axial apparent solids hold-up profiles, local transient particle velocity, solids concentration and flux distributions, annular wall layer thicknesses, local solids tracer concentrations and axial solids dispersion coefficients were conducted, and the following key conclusions have been drawn. - The local transient particle velocity, solids concentration and flux distributions measured utilizing the 3-fibre optical probe developed by Liu (2001) suggested that the flow behaviour and flow regime transitions in the 0.2 m diameter riser of the fluid coker cold model are strong functions of height, as well as operating conditions. - Detailed examination of the different calibration techniques for the optical fibre probe suggested that the calibration curves should be adapted to the actual unit after being obtained in other simplified systems, to assure that they meet constraints such as the integrated flux matching the total flux obtained by other techniques and the cross-sectional average solids hold-up matching that based on pressure profiles. With a change in the operating condition, the local particle velocity and solid flux in the relatively dilute core region of the riser changed more significantly than those at the wall region. The opposite trend was observed with respect to the local solid hold-up distribution. - The increase in riser diameter from 0.076 to 0.2 m in this study was observed to result in lower cross-sectional average solids holdup at the same superficial gas velocity and net circulation flux under high-density conditions. The local particle velocity in the central region was higher for the larger diameter riser, while the difference in the annulus was insignificant. The time-mean solids fluxes were found to be nearly equal in the central top section of the riser. The difference between the two sets of measurements increased closer to the wall. In the top section of the riser, in the wall region, the time-mean solids fluxes Chapter 6 Conclusions 161 were larger for the smaller diameter riser. The opposite trend was observed for the bottom section of the riser in the wall region. The extent of correlation between particle velocity and solids hold-up fluctuations was examined by investigating their covariance. The increase in covariance for high solids hold-up conditions suggested that the flux measurements, rather than particle velocity measurements, should be utilized in the detection of the thickness of the wall layer, i.e. 8. Under dilute conditions, the difference between the alternative methods of deriving 5 is small. The thickness of the wall layer approaches zero both as dilute pneumatic conveying and DSU are approached. Accordingly, 5 may increase or decrease as U g or G s is increased while the other one is maintained constant. When they are both increased (beyond 8 m/s and 450 kg/m2s for the conditions of this work), the thickness of the downflow layer diminishes. Solids mixing and motion were investigated in a high-density circulating fluidized bed by utilizing phosphorescent-coated FCC particles. An axial dispersion model was utilized to determine solids axial dispersion coefficients, and the results were interpreted with the help of hydrodynamic results obtained in the same column. The solids dispersion coefficient was found to range from 1 to 7 m2/s, in the same order as the gas dispersion coefficient obtained by Liu from the same unit. The axial solids dispersion coefficients did not show a clear decrease trend with increasing the solids flux at fixed gas velocities based on current measurements covering mostly the fast fluidization regime and the transition boundary region between fast fluidization and dense suspension upflow regime, indicating that a fully developed dense suspension upflow regime was not established in the riser for the operating conditions covered in this study. Chapter 6 Conclusions 162 6.2 Recommendations for future work Based on the experience gained in this study, the following recommendations for future extensions of the work are suggested. - To develop a two-dimensional axially and radially dispersed plug flow model, with axial and radial dispersion coefficients derived by utilizing the radial and axial solids concentrations measured in the high-density CFB unit. - To further study solids mixing under the fully developed dense suspension upflow regime. To study all the phenomena and properties determined in this study in risers of diameter > 0.2 m. To investigate the effects of system pressure and temperature on high-density riser hydrodynamics. - To study the hydrodynamics for other particle properties, e.g. different density, different particle size distributions, Geldart group B particles. - To provide CFD and other advanced model predictions to compare with the experimental results produced in this work. (The results, combined with those of Liu, 2001, provide a unique and detailed database for high-density risers.) Nomenclature 163 NOMENCLATURE Variable interval constant -C Tracer concentration arbitrary Co Initial tracer concentration arbitrary D a x Axial solids dispersion coefficient m2/s D r Radial solids dispersion coefficient m2/s Driser Diameter of the riser m dp Mean particle diameter um Fcut-off Cut-off frequency -g Gravitational acceleration m/s2 G s Overall net solids circulation flux kg/m2s Gs,i Instantaneous solids flux kg/m2s G, Time-mean solids flux kg/m2s a Gsm , G s r n Cross-sectional average time-mean solids flux kg/m2s h Axial location m H Height of the riser m I Intensity of the measured signal V Idecay Intensity of light emitted by tracer particles V Time-average signal intensity for fibre A V Time-average signal intensity for fibre B V Lo Initial intensity of emitted light V L e Effective separation distance between fibres mm r Radial co-ordinate m R Radius of the riser m t Time s T Integration period ms Ug Superficial gas velocity m/s Minimum fluidization velocity m/s V Voltage V Voltage at minimum fluidization V Vo Voltage at empty column V v P Particle velocity m/s V p, w Concentration weighted time-mean particle velocity m/s Nomenclature Greek symbols AP Pressure drop Pa Az Vertical separation between two ports m 5 Annular wall layer thickness -8GS Annular wall layer thickness based on solids flux -m^omentum Annular wall layer thickness based on momentum -8vP Annular wall layer thickness based on particle velocity -e Voidage s^, apparent Apparent solids hold-up -£mf Bed voidage at minimum fluidization -es Solids hold-up -Cross-sectional average time-mean solids hold-up -£ s,mf Solids hold-up at minimum fluidization -Pbulk Bulk density kg/m3 Pg Gas density kg/m3 Cross-correlation coefficient function -Pp Particle density kg/m3 Cross-correlation coefficient -Variance of signal generated by fibre A -Variance of signal generated by fibre B -\. Average absolute percentage deviation in solids hold-up -8o„m Average absolute percentage deviation in solids flux -X Time shift s TAB Transit time between fibers A and B s Nomenclature Abbreviations AID Analog/digital CFB Circulating fluidized bed DSU Dense suspension upflow FF Fast fluidization FCC Fluid cracking catalyst FFT Fast Fourier Transform FfDCFB High-density circulating fluidized bed RTD Residence time distribution SEM Scanning electron microscope UV Ultraviolet Superscripts ' Fluctuating component Time-mean 1 Instantaneous values References 166 REFERENCES Ambler, P.A., Milne, B.J., Berruti, F., Scott, D.S., Residence Time Distribution of Solids in a Circulating Fluidized Bed: Experimental and Modelling Studies, Chemical Engineering Science, 4 5 , 2179-2186 (1990). Amos, G., Rhodes, M.J., Benkreira, FL, Calculation of Optic Fibres Calibration Curves for the Measurement of Solids Volume Fractions in Multiphase Flows, Powder Technology, 88, 107-121 (1996). Arena U., Cammarota, A., Marzocchella, A., Massimila, L., Solids Flow Structures in a Two-Dimensional Riser of a Circulating Fluidized Bed, Journal of Chemical Engineering of Japan, 22, 236-241 (1989). Aue-Klett, C , Al-Shawabkeh, A., Werther, J., Gas and Solids Mixing in the Bottom Zone of a Circulating Fluidized Bed, Circulating Fluidized Bed Technology VII, Eds. J.R. Grace, J. Zhu and H. de Lasa, CSChE, Ottawa, pp. 153-160 (2002). Avidan, A. and Yerushalmi, J., Solids Mixing in an Expanded Top Fluid Bed, AIChE J., 31, 835-841 (1985). Baba, T., Nakajima, M., Morooka, S., Matsuyama, Ff., A New Measuring System for Flow Patterns of Solid Particles in Gas-Solid Fluidized Bed, Journal of Chemical Engineering of Japan, 17, 275-279 (1984). Bader, R., Findlay, J., Knowlton, T.M., Gas/Solid Flow Patterns in a 30.5 cm-Diameter Circulating Fluidized Bed, Circulating Fluidized Bed Technology II, Eds. P. Basu and J.F. Large, Pergamon Press, Compiegne, France, pp. 123-137 (1988). References 167 Bai, D. and Kato, K., Quantitative Estimation of Solids Holdups at Dense and Dilute Regions of Circulating Fluidized Beds, Powder Technology, 101, 183-190 (1999). Bai, D.-R., Jin, Y., Yu, Z.-Q., Zhu, J.-X., The Axial Distribution of the Cross-Sectionally Averaged Voidage in Fast Fluidized Beds, Powder Technology, 71, 51-58 (1992). Bai, D., Shibuya, E., Masuda, Y., Nakagawa, N., Kato, K., Flow Structure in a Fast Fluidized Bed, Chemical Engineering Science, 51, 957-966 (1996). Bai, D., Shibuya, E., Masuda, Y., Nishio, K., Nakagawa, N., Kato, K., Distinction Between Upward and Downward Flows in Circulating Fluidized Beds, Powder Technology, 84, 75-81 (1995). Bai, D., Yi, J., Jin, Y., Yu, Z., Residence Time Distributions of Gas and Solids in a Circulating Fluidized Bed, Fluidization VII, Eds. O. E. Potter, D. J. Nicklin, Engineering Foundation, pp. 195-202 (1992). Bellgardt, D. and Werther, J., A Novel Method for the Investigation of Particle Mixing in Gas-Solid Systems, Powder Technology, 48,173-180 (1986). Bellino, I., Bergougnoux, L., Misguich-Ripault, J., Firpo, J.-L., Particle Size Effects on an Optical Fibre Sensor Response, Powder Technology, 115, 68-74 (200f). Bergougnoux, L., Firpo, J.-L., Misguich-Ripault, J., Optical Fibre Sensor for Measuring High Particle Concentrations, Powder Technology, 105, 413-417 (1999). Berruti, F., Chaouki J., Godfroy L., Pugsley R.S., Patience G.S., Hydrodynamics of Circulating Fluidized Beds Risers: A Review, Canadian Journal of Chemical Engineering, 73, 579-602 (1995). References 168 Berruti F. and Kalogerakis, N., Modelling the Internal Flow Structure of Circulating Fluidized beds, Canadian Journal of Chemical Engineering, 6 7 , 1010-1014 (1989). Bi, Ff.T. and Grace, J.R., Flow Regime Diagrams for Gas-Solid Fluidization and Upward Transport, International Journal of Multiphase Flow, 2 1 , 1229-1236 (1995). Bi, H.T., Zhou, J., Qin, S.Z., Grace, J.R., Annular Wall Layer Thickness in Circulating Fluidized Bed Risers, Canadian Journal of Chemical Engineering, 7 4 , 811-814 (1996). Bi, Ff.T. and Zhu, J., Static Instability Analysis of Circulating Fluidized Beds and Concept of High-density Risers, AlChe J., 3 9 (8), t272-1280 (1993). Bi, J., Yang, G., Kojima, T., Lateral Mixing of Coarse Particles in Fluidized Beds of Fine Particles, Chem. Eng. Res. And Design, 7 3 , 162-167 (1995). Bolton, L.W. and Davidson, J.F., Recirculation of Particles in Fast Fluidized Risers, Circulating Fluidized Bed Technology II, Eds. P. Basu and J.F. Large, Pergamon Press, pp. 139-146 Compiegne, France (1988). Brereton, C M . and Grace, J.R., Microstructural Aspects of the Behaviour of Circulating Fluidized Beds, Chemical Engineering Science, 48, 2565-2572 (1993). Brewster, B.S. and Seader, J.D., Nonradioactive Tagging Method of Measuring Particle Velocity in Pneumatic Transport, AIChEJ., 2 6 , 325-327 (1980). Chesonis, D.C., Klinzing, G.E., Shah, Y.T., Dassori, C.G., Hydrodynamics and Mixing of Solids in a Recirculating Fluidized Bed, Industrial & Engineering Chemistry Research, 2 9 , 1785-1792 (1990). References 169 Chesonis, D.C., Klinzing, G.E., Shah, Y.T., Dassori, C.G., Solids Mixing in a Recirculation Fluidized Bed, Circulating Fluidized Bed Technology III, Eds. P. Basu, M. Horio, M. Ffasatami, New York: Pergamon Press, pp. 587-592 (1991). Cui, FL, Mostoufi, N., Chaouki, J., Comparison of Measurement Techniques of Local Particle Concentration for Gas-Solid Fluidization, Fluidization X, Eds. M. Kwauk, J. Li and W.C. Yang, Engineering Foundation, New York, pp. 779-786 (2001). Du, B. and Wei, F., Lateral Solids Mixing Behaviour of Different Particles in a FCC Riser, Fluidization X, Eds. M. Kwauk, J. Li, W.-C. Yang, United Engineering Foundation Inc., New York, pp. 197-204 (2001). Du, B., Wei, F., Wang, Z., Effect of Particle Size on Solids Mixing in a FCC Riser, Circulating Fluidized Bed VI, Dechema, Frankfurt, Ed. J. Werther, pp. 405-410 (1999). Ellis, N., Hydrodynamics of Gas-Solid Turbulent Fluidized Beds, Ph.D Thesis, University of British Columbia, Vancouver, Canada (2003). Fogler H.S., Elements of Chemical Reaction Engineering, 2n d Edition, Prentice-Hall International Inc., New Jersey, 1992. Garncarek, Z., Przybylski, L., M. Botterill, J.S., Bridgwater, F., Broadbent, C.J., A Measure of the Degree of Inhomogenity in a Distribution and Its Application in Characterizing the Particle Circulation in a Fluidized Bed, Powder Technology, 80, 221-225 (1994). Grace, J.R., Reflections on Turbulent Fluidization and Dense Suspension Upflow, Powder Technology, 113, 242-248 (2000). Grace, J.R., Issangya, A.S., Bai, D., Bi, H.T., Zhu, J., Situating the High-Density Circulating Fluidized Bed, AIChEJ., 45, 2108-2116 (1999). References 170 Grassier, T. and Wirth, K.-E., Radial and Axial Profiles of Solids Concentration in a High-loaded Riser Reactor, Ed. J. Werther, Circulating Fluidized Bed Technology VI, DECHEMA, Germany, 65-70 (1999). Harris, A.T., Residence Time Distributions in Circulating Fluidized Beds, Ph.D. Thesis, University of Cambridge, UK (2002a). Harris, A.T., Davidson, J.F., Thorpe, R.B., A Novel Method for Measuring the Residence Time Distribution in Short Time Scale Particulate Systems, Chemical Engineering Journal, 89, 127-142 (2002b). Harris, A.T., Davidson, J.F., Thorpe, R.B., Particle Residence Time Distributions in Circulating Fluidized Beds, Chemical Engineering Science, 58, 2181-2202 (2003). Harris, A.T., Thorpe, R., Davidson, J., Residence Time Distribution in Circulating Fluidized Beds- Dimensional Analysis and Correlation, Fluidization X, Eds. M. Kwauk, J. Li, W.-C. Yang, United Engineering Foundation Inc., New York, 221-228 (2001). Harris, A.T., Thorpe, R.B., Davidson, J.F., Characterization of the Annular Film Thickness in Circulating Fluidised-Bed Riser, Chemical Engineering Science, 57, 2579-2587 (2002c). Harris, A., Thorpe, R., Davidson, J., The Measurement of Particle Residence Time Distributions in Circulating Fluidized Beds, Circulating Fluidized Bed Technology VII, Eds. J.R. Grace, J. Zhu and H. de Lasa, CSChE, Ottawa, 145-152 (2002d). Hartge, E.-U., Li, Y., Werther, J., Analysis of the Local Structure of the Two Phase Flow in a Fast Fluidized Bed, Circulating Fluidized Bed Technology, Ed. P. Basu, Pergamon Press, Toronto, pp. 153-160 (1986a). References 171 Hartge, E.-U., Li, Y., Werther, J., Flow Structures in Fast Fluidized Beds, Fluidization V, Eds. K. Ostergaard and A. Sorensen, Fluidization V, Engineering Foundation, New York, pp. 345-352 (1986b). Heimlich, H., Schugerl, K., Janssen, K., Decomposition of NaHC03 in Laboratory and Bench Scale Circulating Fluidized Bed Reactors, Circulating Fluidized Bed Technology, Ed. P. Basu, Toronto, New York, Pergamon Press, pp. 161-166 (1985). Herbert, P.M., Gauthier, T.A., Briens, C.L., Bergougnou, M.A., Application of Fibre Optic Reflection Probes to the Measurement of Local Particle Velocity and Concentration in Gas-Solid Flow, Powder Technology, 80, 243-252 (1994). Issangya, A.S., Flow Dynamics in High Density Circulating Fluidized Beds, Ph.D. Thesis, University of British Columbia, Vancouver, Canada (1998). Issangya, A.S., Bai, D., Bi, H.T., Lim, K.S., Zhu, J., Grace, J.R., Axial Solids Holdup Profiles in a High-Density Circulating Fluidized Bed Riser, Circulating Fluidized Bed Technology V, Eds. M. Kwauk and J. Li, Science Press, Beijing, pp. 60-65 (1997). Issangya, A.S., Bai, D., Bi, H.T., Lim, K.S., Zhu, J., Grace, J.R., Suspension Densities in a High-Density Circulating Fluidized Bed Riser, Chemical Engineering Science, 54, 5451-5460 (1999). Issangya, A.S., Grace, J.R., Bai, D., Zhu, J., Further Measurements of Flow Dynamics in a High-Density Circulating Fluidized Bed Riser, Powder Technology, 111, 104-113 (2000). Karri, S.B.R. and Knowlton, T.M., Flow Direction and Size Segregation of Annulus Solids in a Riser, Fluidization IX, Eds. L.S. Fan and T. M. Knowlton, Engineering Foundation, New York, pp. 189-196 (1998). References 172 Karri, S.B.R. and Knowlton, T.M., A Comparison of Annulus Solids Flow Direction and Radial Solids Mass Flux Profiles at Low and High Mass Fluxes in a Riser, Ed. J. Werther, Circulating Fluidized Bed Technology VI, Germany, Dechema e.V., pp. 71-76 (1999). Karri, S.B.R. and Knowlton, T.M., Wall Solids Upflow and Downflow Regimes in Risers for Group A Solids, Circulating Fluidized Bed Technology VII, Eds. J.R. Grace, J. Zhu and H. de Lasa, CSChE, Ottawa, pp. 310-316 (2002). Kim, S.W., Kirbas, G., Bi, H.T., Lim, C.J., Grace, J.R., Flow Structure and Thickness of Annular Downflow Layer in a Circulating Fluidized Bed Riser, Powder Technology, 142, 48-58 (2004a). Kim, S.W., Kirbas, G., Bi, H.T., Lim, C.J., Grace, J.R., Flow Behavior and Regime Transition in a High-Density Circulating Fluidized Bed Riser, Chemical Engineering Science (in press) (2004b). Kirbas, G., Lim, C.J., Grace, J.R., Bi, H.T., Solids Motion near the Wall of High Density Circulating Fluidized Beds, Proceedings of World Congress of Particle Technology, CD-ROM, Sydney, Australia (2002). Knapper, B., Experimental Studies on the Hydrodynamics of Fluid Bed Cokers, Master's Thesis, University of Saskatchewan, Saskatoon, Canada (2001). Knapper, B., Berruti, F., Grace, J.R, Bi, H.T., Lim, C.J., Hydrodynamic Characterization of Fluid Bed Cokers, in Circulating Fluidized Bed Technology VII, Eds. J.R. Grace, J. Zhu and H. de Lasa, CSChE, Ottawa, 263-270 (2002). Kojima, T., Ishihara, K.-L, Guilin, Y., Furusawa, T., Measurement of Solids Behaviour in a Fast Fluidized Bed, Journal of Chemical Engineering of Japan, 22, 341-346 (1989). References 173 Kondukov, N.B., Kornilaev, A.N., Skachko, I.M., Akhromenkov, A.A., Kruglov, A.S., An Investigation of the Parameters of Moving Particles in a Fluidized Bed by a Radioisotopic Method, International Chemical Engineering, 4, 43-47 (1964). Levenspiel, O., The Chemical Reactor Omnibook, 5th Ed., OSU Book Stores Inc., Oregon, 1996. Li, J., Tung, Y., Kwauk, M., Axial Voidage Profiles of Fast Fluidized Beds in Different Operating Regimes, Circulating Fluidized Bed Technology II, Eds. P. Basu and J.F. Large, Pergamon Press, Compiegne, France, pp. 193-203 (1988). Li, Y.C. and Kwauk, M., The Dynamics Of Fast Fluidization, Fluidization, Eds. J.R.Grace, and J. M. Matsen, Plenum Press, New York, pp. 537- 544 (1980). Lin, J.S., Chen, M.M., Chao, B.T., A Novel Radioactive Particle Tracking Facility for Measurement of Solids Motion in Gas Fluidized Beds, AIChE J., 31, 465-472 (1985). Lin, Q., Wei, F., Jin, Y., Transient Density Signal Analysis and Two-Phase Micro-Structure Flow in Gas-Solids Fluidization, Chemical Engineering Science, 56, 2179-2189 (2001). Lin, W., Weinell, C.E., Hansen, P.F.B., Dam-Johansen, K., Hydrodynamics of a Commercial Scale CFB Boiler-Study with Radioactive Tracer Particles, Chemical Engineering Science, 54, 5495-5506 (1999). Lin, Y.-C. and Leu, L.-P., Voidage Profiles in Magnetically Fluidized Beds, Powder Technology, 120, 199-215 (2001). Lischer, D.J. and Louge, M.Y., Optical Fibre Measurements of Particle Concentration in Dense Suspensions: Calibration and Simulation, Applied Optics, 31, 5106-5113 (1992). References 174 Liu, J . , Particle and Gas Dynamics of High Density Circulating Fluidized Beds, Ph.D. Thesis, University of British Columbia, Vancouver, Canada (2001). Liu, J . , Grace, J.R., Bi, H.T., Radial Distribution of Local Particle Velocity in a High-Density Circulating Fluidized Bed Riser, in Circulating Fluidized Bed Technology VII, Eds. J.R. Grace, J. Zhu and H. de Lasa, CSChE, Ottawa, pp. 341-348 (2002). Liu, J. , Grace, J.R., Bi, H.T., Novel Multifunctional Optical Fiber Probe: I. Development and Validation, AIChEJ., 49 , 1405-1420 (2003a). Liu, J . , Grace, J.R., Bi, H.T., Novel Multifunctional Optical Fiber Probe: n. High Density CFB Measurements, AIChEJ., 49 , 1421-1432 (2003b). Liu, J. , Grace, J.R., Bi H.T., Morikowa, H., Zhu, J. , Gas Dispersion in Fast Fluidization and Dense Suspension Upflow, Chem. Eng. Science, 54 , 5441-5449 (1999). Louge, M.Y. and Chang, H., Pressure and Voidage Gradients in Vertical Gas-Solid Risers, Powder Technology, 60, 197-201 (1990). Malcus, S., Experimental and Modelling Study of a High Suspension Density CFB Riser, Ph.D. Thesis, The University of New Brunswick, Canada (2000). Malcus, S., Cruz, E., Rowe, C , Pugsley, T.S., Radial Mass Flux Profiles in a High-Suspension Density Circulating Fluidized Bed, Powder Technology, 125, 5-9 (2002). Malcus, S. and Pugsley, T.S., Investigation of the Axial Variation of the CFB Riser Hydrodynamics by Means of Electrical Capacitance Tomography (ECT), Fluidization X, Eds. M. Kwauk, J. Li and W.C. Yang, Engineering Foundation, New York, pp. 763-770 (2001). References 175 Manyele, S.V., Khayat, R.E., Zhu, J.-X., Investigation of the Dynamics of a High-Flux CFB Riser using Chaos Analysis of Pressure Fluctuations, I Chemical Engineering & Technology, 25, 801-810 (2002). Martin, M.P., Turlier, P., Bernard, J.R., Wild, G., Gas and Solid Behavior in Cracking Circulating Fluidized Beds, Powder Technology, 70, 249-258 (1992). Matsuno, Y., Yamaguchi, H., Oka, T., Kage, H., Higashitani, K., The Use of Optic Fiber Probes for the Measurement of Dilute Particle Concentrations: Calibration and Application to Gas-fluidized Bed Carryover, Powder Technology, 36, 215-221 (1983). May, W.G., Fluidized-Bed Reactor Studies, Chemical Engineering Progress, 55, 49-56 (1959). Militzer, J., Hebb, J.P., Jollimore, G., Shakourzadeh, K., Solid Particle Velocity Measurements, Fluidization VII, Eds. O. E. Potter, D. J. Nicklin, 762-769, Engineering Foundation, New York, (1992). Milne, B.J. and Berruti, F., Modelling the Mixing of Solids in circulating Fluidized Beds, Circulating Fluidized Bed Technology III, Eds. P. Basu, M. Horio, M. Hasatami, New York: Pergamon Press, pp. 575-580 (1991). Monceaux, L., Azzi, M., Molodtsof, Y., Large, J.F., Particle Mass Flux Profiles and Flow Regimes Characterization in a Pilot Scale Fast Fluidized Bed Unit, in Fluidization V, Eds. K. Ostergaard and A. Sorenson, Engineering Foundation, New York, 337-344 (1986) Morooka, S., Kago, T., Kato, Y., Flow Pattern of Solids Particles in Freeboard of Fluidized Bed, Fluidization IV, Eds. D. Kunii and R. Toei, pp. 291-298 (1983). References 176 Morooka, S., Kusakabe, K., Ohnishi, N., Gujima, F., Matsuyama, FL, Measurement of Local Fines Movement in a Fluidized Bed of Coarse Particles by a Fluorescent Tracer Technique, Powder Technology, 58, 271-277 (1989). Mostoufi, N. and Chaouki, J., Critical Study of the Axial Movement of Solids in Gas-Solid Fluidized Beds, Proc. ofAIChE 1999 Annual Meeting, 56-64 (1999). Mostoufi, N. and Chaouki, J., Local Solid Mixing in Gas-Solid Fluidized Beds, Powder Technology, 114, 23-31 (2001). Muramoto, T., Nakajima, M., Morooka, S., Matsuyama, FL, Measurement of Flow of Solid Particles in Gas Fluidized Bed, Chemical Engineering Communications, 35, 193-201 (1985). Nieuwland, J.J., Meijer, R., Kuipers, J.A.M., van Swaaij, W.P.M., Measurements of Solids Concentration and Axial Solids Velocity in Gas-Solid Two-Phase Flows, Powder Technology, 87, 127-139 (1996). Origin, Data Analysis and Technical Graphics Software Version 6, User Manual, Microcal Software Inc. (1999). Ouyang, S., Lin, J., Potter, O.E., Ozone Decomposition in a 0.254m Diameter Circulating Fluidized Bed Reactor, Powder Technology, 74, 73-78 (1993). Parssinen, J.H. and Zhu, J.-X., Particle Velocity and Flow Development in a Long and High-Flux Circulating Fluidized Bed Riser, Chemical Engineering Science, 56, 5295-5303 (2001). Patience, G.S., Chaouki, J., Kennedy, G., Solids Residence Time Distribution in CFB Reactors, Circulating Fluidized Bed Technology 111, Eds. P. Basu, M. Horio, M. Hasatami, New York, Pergamon Press, pp. 599-604 (1991). References 111 Qian, G. and Li, J., Particle-Velocity Measurement in CFB with an Integrated Probe, Circulating Fluidized Bed Technology IV, Ed. Avidan, A. A., American Institute of Chemical Engineers, New York, USA, 274-278 (1994). Ran, X., Wei, F., Wang, Z., Jin, Y., Lateral Solids Dispersion in a High-Density Riser with Swirling Air Flow, Powder Technology, 121, 123-130 (2001). Reh, L. and Li, J., Measurement of Voidage in Fluidized Beds by Optical Probes, Circulating Fluidized Bed Technology III, Eds. P. Basu, M. Horio, M. Hasatami, New York: Pergamon Press, pp. 163-170(1991). Rensner, D. and Werther, J., Modelling and Application of a Fiber Optical Measuring System for Higher Concentrated Multiphase Flows, Laser Anemometry and Application, Fourth International Conference on Laser Anemometry, Advances and Applications, Eds. A. Dybbs, B. Ghorashi, Cleveland, Ohio, August 5-9, 1991, 2, ASME, New York, pp. 753-761 (1991). Rhodes, M.J. and Laussmann, P., Characterizing Non-uniformities in Gas-particle Flow in the Riser of a Circulating Fluidized Bed, Powder Technology, 72, 277-284 (1992) Rhodes, M.J., Laussmann, P., Villain, F., Geldart, D., Measurement of Radial and Axial Solids Flux Variations in the Riser of a Circulating Fluidized Bed, Circulating Fluidized Bed Technology II, Eds. P. Basu and J.F. Large, Pergamon Press, Compiegne, France, 155-164 (1988). Rhodes, M.J., Sollaart, M., Wang, X.S., Flow Structure in a Fast Fluid Bed, Powder Technology, 99, 194-200 (1998). Rhodes, M.J., Zhou, S., Hirama, T., Cheng, H., Effects of Operating Conditions on Longitudinal Solids Mixing in a Circulating Fluidized Bed Riser, AIChE J., 37, 1450-1458 (1991). References 178 Roques, Y., Gauthief, T., Pontier, R., Briens, C.L., Bergougnou, M.A., Residence Time Distributions of Solids in a Gas-Solids Downflow Transported Reactor, Circulating Fluidized Bed Technology IV, Ed. A. Avidan, AIChE, New York, pp. 555-559 (1993). Schlichthaerle, P. and Werther, J., Axial Pressure Profiles and Solids Concentration Distributions in the CFB Bottom Zone, Chemical Engineering Science, 54, 5485-5493 (1999). Schlichthaerle, P. and Werther, J., Solids Mixing in the Bottom Zone of a Circulating Fluidized Bed, Powder Technology, 1 2 0 , 21-33 (2001). Schnitzlein, M.G. and Weinstein, Ff., Flow Characterization In High-Velocity Fluidized-Beds Using Pressure-Fluctuations, Chemical Engineering Science, 43, 2605-2614 (1988). Shionoya, S. and Yen, W.M., Phosphor Handbook, CRC Press, USA (1999). Sitnai, O., Solids Mixing in a Fluidized Bed with Horizontal Tubes, Industrial & Engineering Chemistry Process Design and Development, 2 0 , 533-538 (1981). Smolders, K. and Baeyens, J., Overall Solids Movement and Solids Residence Time Distribution in a CFB-Riser, Chemical Engineering Science, 55, 4101-4116, (2000). Stein, M., Ding, Y.L., Seville, J.P.K., Parker, D.J., Solids Motion in Bubbling Gas Fluidized Beds, Chemical Engineering Science, 55, 5291-5300 (2000). Sun, G., Chao, Z., Fan, Y., Shi, M., Hydrodynamic Behaviour in the Bottom Region of a Cold FCC Riser, Circulating Fluidized Bed Technology VI, Ed. J. Werther, 179-184, DECHEMA e.V., Germany (1999). Svensson, A., Johnsson, F., Leckner, B., Fluid-dynamics of the Bottom Bed of Circulating Fluidized Bed Boilers, Ed. L. Rubow, Proc. 12th Int. Conf. Fluidized Bed Combustion, ASME, New York, 887-897 (1993) References 179 Tayebi, D., Svendsen, H. F., Grislingas, A., Mejdell, T., Johannessen, K., Dynamics of Fluidized Bed Reactors. Development and Application of a New Multi-fiber Optical Probe, Chemical Engineering Science, 54, 2113-2122 (1999). van Zoonen, D., Measurements of Diffusional Phenomena and Velocity Profiles in a Vertical Riser, in Proc. Conf. Interaction Between Fluids & Particles. Instn. Chem. Engrs., London, pp. 64-71 (1962). Viitanen, P.I., Tracer Studies on a Riser Reactor of a Fluidized Catalyst Cracking Plant, Industrial and Engineering Chemistry Research, 32, 577-583 (1993). Wei, F., Cheng, Y., Jin, Y., Yu, Z., Axial and Lateral Dispersion of Fine Particles in a Binary-Solid Riser, Canadian Journal of Chemical Engineering, 76, 19-26 (1998). Wei, F., Jin, Y., Yu, Z., Chen, W., Mori, S., Lateral and Axial Mixing of the Dispersed Particles in CFB, Journal of Chemical Engineering of Japan, 28, 506-510 (1995). Wei, F., Lin, FL, Cheng, Y., Wang, Z., Jin, Y., Profiles of Particle Velocity and Solids Fraction in a High-Density Riser, Powder Technology, 100, 183-189 (1998). Wei, F., Lu, F., Jin, Y., Yu, Z., Mass Flux Profiles in a High Density Circulating Fluidized Bed, Powder Technology, 91, 189-195 (1997). Wei, F., Wang, Z., Jin, Y., Yu, Z. and Chen, W., Dispersion of Lateral and Axial Solids in a Cocurrent Downflow Circulating Fluidized Bed, Powder Technology, 81, 25-30 (1994). Wei, F. and Zhu, J.-X., Effect of Flow Direction on Axial Solid Dispersion in Gas-Solids Cocurrent Upflow and Downflow Systems, Chemical Engineering Journal, 64, 345-352 (1996). References 180 Weinell, C.E., Dam-Johansen, K., Johnson, J.E., Gluchowski, Z., Lade, K., Single Particle Behaviour in Circulating Fluidized Beds, Fluidization VII, Eds. O. E. Potter, D. J. Nicklin, Engineering Foundation, New York, pp. 295-304 (1992). Weinell, C.E., Dam-Johansen, K., Johnsson, J.E., Single Particle Velocities and Residence Times in Circulating Fluidized Beds, Circulating Fluidized bed Technology IV, Ed. A. Avidan, AIChE, New York, pp. 571-576 (1993). Werther, J., Fluid Mechanics of Large-scale CFB Units, in Circulating Fluidized Bed Technology IV, Ed. A.A. Avidan, 1-14, (1994) Werther, J., Hage, B., Rudnick, C , A Comparison of Laser Doppler and Single-Fibre Reflection Probes for the Measurement of the Velocity of Solids in a Gas-Solid Circulating Fluidized Bed, Chemical Engineering and Processing, 35, 381-391 (1996). Werther, J., Hartge, E.-U., Rensner, D., Measurement Techniques for Gas-Solid Fluidized-Bed Reactors, International Chemical Engineering, 33, 18-27 (1993). Werther, J. and Hirschberg, B., Chapter 4: Solids Motion and Mixing, Circulating Fluidized Beds, Eds. J.R. Grace, A.A. Avidan, T.M. Knowlton, London: Chapman and Hall, pp. 119-148 (1997). Westphalen, D. and Glicksman, L., Lateral Solid Mixing Measurements in Circulating Fluidized Beds, Powder Technology, 82, 153-167 (1995). Win, K.K., Nowak, W., Matsuda, H., Hasatani, M., Kruse, M., Werther, J., Radial Gas Mixing in the Bottom Part of a Multi-Solid Fluidized-Bed, Journal of Chemical Engineering of Japan, 27, 696-698 (1994). References 181 Yan A.-J., Zhu H.-Y., Zhu J., Comparison of FCC Particle Flow in a Twin-Riser System, Fluidization XI, Eds. U. Arena, R. Chirone, M. Miccio, P. Salatino, Engineering Conferences International, Brooklyn (2004). Yang, H. and Gautam, M., Experimental Study on the Interface of Core-Annulus Flow in the Riser of a Circulating Fluidized Bed, Powder Technology, 85, 57-64 (1995). Yerushalmi, J. and Avidan, A., High Velocity Fluidization, in Fluidization, 2nd Ed., Eds. J. F. Davidson, R. Clift, D. Harrison, Academic Press, London, 225-291 (1985). Yerushalmi, J. and Cankurt, N.T., Further Studies of the Regimes of Fluidization, Powder Technology, 24, 187-205 (1979). Zhang, H., Johnston, P.M., Zhu, J.-X., de Lasa, H.I., Bergougnou, M.A., A Novel Calibration Procedure for a Fibre Optic Solids Concentration Probe, Powder Technology, 100, 260-272 (1998). Zhang, W., Tung, Y., Johnsson, F., Radial Voidage Profiles in Fast Fluidized Beds of Different Diameters, Chemical Engineering Science, 46, 3045-3052 (1991). Zheng, C , Tung, Y., Li, H., Kwauk, M., Characteristics of Fast Fluidized Beds with Internals, Fluidization VII, Eds. O. E. Potter, D. J. Nicklin, Engineering Foundation, pp. 275-283 (1992). Zhou, J., Grace, J.R., Lim, C.J., Brereton, C.M.H., Particle Velocity Profiles in a Circulating Fluidized Bed Riser of Square Cross-Section, Chemical Engineering Science, 50, 237-244 (1995). Zhou, J., Grace, J.R., Qin, S., Brereton, C.M.H., Lim, C.J., Zhu, J., Voidage Profiles in a Circulating Fluidized Bed of Square Cross-Section, Chemical Engineering Science, 49, 3217-3226 (1994). References 182 Zhu, C. and Soo, S.L., Statistical Properties of Unsteady Gas-Solid Suspensions with Strong Particle-Particle Interactions in Horizontal Pipe Flows, Powder Technology, 73, 51-60 (1992). Zhu, J.-X and Bi, H.T., Distinctions Between Low-density and High-density Circulating Fluidized Beds, Canadian Journal of Chemical Engineering, 73, 644-649, (1995). Zhu, J.-X., Li, G.-Z., Qin, S.-Z., Li, F.-Y., Zhang, H., Yang, Y.-L., Direct Measurements of Particle Velocities in Gas-Solid Suspension Flow Using a Novel Five-Fiber Optical Probe, Powder Technology, 115, 184-192 (2001). Appendix A 183 APPENDIX A Procedure of the Preparation of the Tracer Particles used in Solids Mixing Studies The tracer particles in this study were prepared by coating FCC particles with a phosphorescent pigment. Since polyvinyl alcohol (PVA) has been used as a binder in several previous phosphorescent pigment coating applications (Shionoya and Yen, 1999), it was used to affix the phosphorescent pigment to FCC particles in this study. In order to obtain a liquid binder, the powder PVA was mixed with water. According to the product information of PVA (Sigma-Aldrich, Product #: 363 146), it is soluble in water at 50 g per litre around 100°C. Following a suggestion in the manual of the supplier of the phosphorescence light detection system, the amount of pigment used to coat FCC particles was selected to be 10 wt.% of the FCC that would be coated. Different proportions of PVA + water and FCC + phosphorescent particles were tested to obtain a dense and moist, uniform cake-like-texture when mixed. After trial and error, the ingredients were mixed in the proportions given in Table A. l . Table A . l Proportions of the ingredients used in tracer particle preparation Ingredient wt.% FCC 52.5 Phosphorescent pigment 5.3 PVA 2.2 Water 40 Appendix A 184 The following procedure was employed to prepare the tracer particles: (1) Heat the water in a beaker to 100°C. (2) Add PVA while vigorously mixing and continue mixing until the PVA powder dissolves in the water. (3) Cool down the mixture to room temperature before adding FCC and phosphorescent pigment particles, since light emission capacity of phosphorescent materials have been reported to decrease with temperature (Shionoya and Yen, 1999). (4) Pour PVA + water mixture into Hobart mixer. (5) Add FCC and phosphorescent particle, and slowly stir the mixture until it has a cake-like-texture. (6) Pour the mixtures into rectangular pans and leave them in an oven at 40°C overnight. (7) Remove the pans from the oven once they are dry and become one-piece solid structures. (8) Crush and sieve them into similar size distribution as the FCC particles (Table 2.2) using the facilities of the Coal and Mineral Processing Laboratory at the University of British Columbia. Appendix B 185 APPENDIX B Properties of the Light Sources Used in Solids Mixing Studies Table B.l The properties of the light sources used in the preliminary tests and solids mixing experiments {provided by the suppliers) Type Regent 500 Watt Halogen Worklight Philips 2-pin Compact Fluorescent lamp (PL-S 2) Lectro Science Ltd. HSUV-365 Hand-held UV Light UV Light Bulbs used in solids mixing experiments Power, Watts 500 13 50 25 and 40 Light emission wavelength (nm) 400 - 700 400 - 700 365 254 Light intensity (lumens) 10500 688 900 N/A N/A : Not available 

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

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"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.831.1-0058657/manifest

Comment

Related Items