Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Probing the electrostatics and hydrodynamics in gas-solid fluidized beds He, Chuan 2015

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

Item Metadata

Download

Media
24-ubc_2015_may_he_chuan.pdf [ 13.07MB ]
Metadata
JSON: 24-1.0135688.json
JSON-LD: 24-1.0135688-ld.json
RDF/XML (Pretty): 24-1.0135688-rdf.xml
RDF/JSON: 24-1.0135688-rdf.json
Turtle: 24-1.0135688-turtle.txt
N-Triples: 24-1.0135688-rdf-ntriples.txt
Original Record: 24-1.0135688-source.json
Full Text
24-1.0135688-fulltext.txt
Citation
24-1.0135688.ris

Full Text

PROBING THE ELECTROSTATICS AND HYDRODYNAMICS IN GAS-SOLID FLUIDIZED BEDS  by Chuan He  M.Sc., Nanjing University of Science and Technology, 2010 B.Eng., Nanjing University of Science and Technology, 2008  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE AND POSTDOCTORAL STUDIES (Chemical and Biological Engineering)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver)  March 2015  © Chuan He, 2015ii Abstract Novel electrostatic dual-tip probes, combined with suitable signal analysis methods, were developed for in-situ measurement and monitoring of particle charge density levels and bubble properties inside gas-fluidized beds. The probes were calibrated in several particulate flow devices: ejector-funnel, motor-pulley, vertical tube and vibration tray setups, as well as a two-dimensional fluidized bed. The effects of particle charge density, solid flux, particle velocity and angle of impact on the transferred current received by the probe from charged particles were quantified. For dual-tip (two-material) probes, substantial differences were observed in the signals from the two tips made of different materials, arising mainly from charge transfer and depending on the hydrodynamics and charge density inside the bed.   The probes were deployed with glass beads and polyethylene particles for both single bubble injection and freely bubbling experiments in two- and three-dimensional fluidization columns of different scales. Statistical and Fast Fourier Transform analysis showed that current signals were strongly affected by the local hydrodynamics in the fluidized bed. The amplitudes of current signal peaks, peak frequencies, as well as mean and standard deviations of the current increased with increasing superficial gas velocity. Local particle charge density and bubble behaviour were estimated by a signal processing procedure with decoupling methods. The probes were tested in steady state experiments, as well as in dynamic experiments by abruptly changing the superficial gas velocity or adding antistatic agent. Both particle charge density and bubble rise velocity obtained from the probes were of the same order of magnitude and followed similar trends as those directly measured by a Faraday cup and video images, respectively. The electrostatic probe signal was found to not always be consistent with the charge polarity and charge density on the particles. The probe signals and particles charge densities may have different polarity and relative magnitudes for different operating conditions and particle properties: density, mean size and size range, dielectric constant, sphericity, roughness and hydrophobicity. Particles with narrow size distribution and larger mean size generated higher charge densities. The novel probe has potential for in-situ monitoring electrostatic charges and hydrodynamic behaviour in gas-solid fluidized beds. iii Preface At the time of writing this thesis, three research papers have been published, one in press and one under reviewed. The author carried out all stages of the research described in the published and submitted papers and also prepared manuscripts. The co-authors supervised the work, provided feedback and insight throughout this process, and edited the manuscripts.  He C., Bi X. T., Grace J. R. Simultaneous measurements of particle charge density and bubble properties in gas-solid fluidized beds by dual-tip electrostatic probes. Chemical Engineering Science, 2015, 123, 11-21.  He C., Bi, X. T., Grace J. R. Contact electrification of a novel dual-material probe with charged particulate flow in an ejector-funnel setup. Powder Technology, 2014, 253, 1-9.  He C., Bi X. T., Grace J. R. A novel dual-material probe for in situ measurement of particle charge densities in gas-solid fluidized beds. Particuology, in press.   He C., Bi X. T., Grace J. R. Decoupling electrostatic signals from gas-solid bubbling fluidized beds. under review.  He C., Bi X. T., Grace J. R. Dual-material probe measurement of electrostatic charges and hydrodynamics in gas-solid fluidized beds. Proceedings of 11th International Conference on Fluidized bed Technology, Beijing, China, May 2014. Above papers correspond to Chapters 2-4 of this thesis. A manuscript corresponds to Chapter 5, He C., Bi X. T., Grace J. R. “In-situ monitoring electrostatics and hydrodynamics in gas-solid bubbling fluidized beds” will be submitted to a journal for publication.   The author has also given several presentations at international, national and local conferences and scientific/technical forums:  He C., Bi X. T., Grace J. R. Dual-material probe measurement of electrostatic charges and hydrodynamics in gas-solid fluidized beds. 11th International Conference on Fluidized bed Technology, Beijing, China, May 14-17, 2014. (Keynote Presentation)  He C., Bi X. T., Grace J. R. Decoupling electrostatics signals from gas-solid bubbling fluidized beds. International workshop on Fluid-Particle Systems, Vancouver, Canada, June 26-27th, 2014.  iv  He C., Bi X. T., Grace J. R. Analysis of electrostatic signals from gas-solid fluidized bed. 62nd Canadian Chemical Engineering Conference, Vancouver, BC, Canada, October 14-17, 2012.  The authors also submitted one patent application titled “Method and apparatus for monitoring electrostatics in fluidized beds” to University-Industry Liaison Office (UILO) at UBC.  v Table of Contents Abstract .................................................................................................................................... ii Preface ..................................................................................................................................... iii Table of Contents .................................................................................................................... v List of Tables .......................................................................................................................... ix List of Figures ......................................................................................................................... xi List of Symbols and Abbreviations .................................................................................. xxiii Acknowledgements ........................................................................................................... xxvii Chapter 1 Introduction........................................................................................................... 1 1.1 Electrostatics in gas-solid fluidized beds .........................................................................1 1.2 Problem definition ...........................................................................................................1 1.3 Thesis objectives ..............................................................................................................2 1.4 Thesis outline ...................................................................................................................3 1.5 Interaction between electrostatics and hydrodynamics in gas-solid fluidized beds ........4 1.6 Electrostatic charge measurement techniques .................................................................5 1.6.1 Faraday cup and particle trajectory tracking methods ............................................. 5 1.6.2 Probes ....................................................................................................................... 8 1.7 Particle charging models ................................................................................................14 1.7.1 Models for pneumatic transport ............................................................................. 14 1.7.2 Models for fluidized beds ...................................................................................... 14 1.8 Electrostatic signal analysis ...........................................................................................15 1.8.1 Electrostatic signals from pneumatic transportation .............................................. 15 1.8.2 Electrostatic signals from fluidized beds ............................................................... 17 Chapter 2 Probes Design, Fabrication and Calibration .................................................... 23 2.1 Introduction ....................................................................................................................23 2.2 Design and fabrication ...................................................................................................23 2.2.1 Design .................................................................................................................... 23 2.2.2 Fabrication ............................................................................................................. 27 2.3 Calibration of probes outside fluidized bed ...................................................................28 2.3.1 Ejector-funnel setup ............................................................................................... 28 vi 2.3.2 Motor-pulley setup ................................................................................................. 32 2.3.3 Vertical tube setup and vibration tray setup........................................................... 33 2.4 Calibration of probes inside fluidized bed .....................................................................35 2.5 Results and discussion ...................................................................................................37 2.5.1 Ejector-funnel setup ............................................................................................... 37 2.5.2 Motor-pulley setup ................................................................................................. 47 2.5.3 Vertical tube and vibration tray setups .................................................................. 49 2.5.4 Two-dimensional fluidized bed ............................................................................. 51 2.5.5 Probes calibrations with particles of different properties ...................................... 59 2.6 Summary ........................................................................................................................63 Chapter 3 Measurements in Gas-Solid Fluidized Beds ..................................................... 65 3.1 Introduction ....................................................................................................................65 3.2 Experimental equipment and methods ...........................................................................65 3.2.1 Two-dimensional fluidization column ................................................................... 65 3.2.2 Three-dimensional fluidization column of ID 0.10 m ........................................... 65 3.2.3 Three-dimensional fluidization column of ID 0.30 m ........................................... 66 3.3 Results and discussion ...................................................................................................68 3.3.1 Results from two-dimensional fluidization column ............................................... 68 3.3.2 Results from three-dimensional fluidization column of ID 0.10 m ....................... 77 3.3.3 Results from three-dimensional fluidization column of ID 0.30 m ....................... 82 3.4 Summary ........................................................................................................................87 Chapter 4 Decoupling Electrostatic Signals from Gas-Solid Fluidized Beds .................. 89 4.1 Introduction ....................................................................................................................89 4.2 Signal analysis procedure ..............................................................................................89 4.2.1 Peak (bubble) detection .......................................................................................... 89 4.2.2 Peak (bubble) selection .......................................................................................... 91 4.3 Decoupling methods ......................................................................................................93 4.3.1 Charge transfer and induction model ..................................................................... 93 4.3.2 Dual-tip (two-material) probe ................................................................................ 94 4.3.3 Dual-tip (one-material) probe ................................................................................ 96 4.4 Results and discussion ...................................................................................................98 vii 4.4.1 Signal processing ................................................................................................... 98 4.4.2 Results from two-dimensional fluidization column ............................................. 105 4.4.3 Results from three-dimensional fluidization column of ID 0.10 m ..................... 117 4.4.4 Results from three-dimensional fluidization column of ID 0.30 m ..................... 118 4.4.5 Comparison of two types of dual-tip probes ........................................................ 123 4.4.6 Signals from conventional single-tip probes in pressurised column with original PE-II .............................................................................................................................. 124 4.4.7 Differences between polyethylene and glass beads ............................................. 126 4.5 Summary ......................................................................................................................127 Chapter 5 In-situ Monitoring in Gas-Solid Fluidized Beds ............................................ 129 5.1 Introduction ..................................................................................................................129 5.2 Experimental equipment and methods .........................................................................130 5.3 Results and discussion .................................................................................................131 5.3.1 Effect of sampling parameters ............................................................................. 131 5.3.2 Changing Ug in two-dimensional bed of 500-600 m GB .................................. 133 5.3.3 Adding antistatic agent to two-dimensional bed of 500-600 m GB .................. 136 5.3.4 Changing Ug in two-dimensional bed of 100-1500 m PE ................................. 138 5.3.5 Changing Ug in three-dimensional bed of 100-1500 m PE ............................... 140 5.4 Summary ......................................................................................................................143 Chapter 6 Overall Conclusions and Recommendations for Future Work .................... 144 6.1 Conclusions ..................................................................................................................144 6.1.1 Probes fabrication and calibration........................................................................ 144 6.1.2 Measurements in gas-solid fluidized beds of glass beads and polyethylene ....... 145 6.1.3 Signal analysis and decoupling ............................................................................ 146 6.1.4 In-situ monitoring in gas-solid fluidized beds ..................................................... 147 6.2 Recommendations for future work ..............................................................................148 References ............................................................................................................................ 149 Appendix A Photographs of experimental components .................................................. 162 A.1 Fabricated probes ........................................................................................................162 A.2 Ejector-funnel setup ....................................................................................................163 A.3 Motor-pulley and vibration tray setups .......................................................................164 viii A.4 Two-dimensional fluidization column ........................................................................165 A.5 Three-dimensional Plexiglas fluidizaton columns ......................................................166 A.6 Pressurized three-dimensional stainless steel fluidization column .............................167 Appendix B Experimental details ...................................................................................... 168 B.1 Experimental data from calibration in two-dimensional fluidization column ............168 B.2 X-ray Photoelectron Spectrometers characterization of Ni and TiN tips ...................173 B.3 Probe response time ....................................................................................................174 B.4 Sensitivity test in 0.10 m ID three-dimensional fluidization column .........................175 B.5 Troubleshooting signal drift of custom-made probe ...................................................178 B.6 Determination of minimum fluidization velocity .......................................................179 B.7 Particle size distributions ............................................................................................183 B.8 Synchronization of camera and probe .........................................................................185 B.9 Polarity of cumulative charge signals from probe ......................................................187 B.10 Further results from pressurized three-dimensional fluidization column .................191 B.10.1 Sampling parameters and signal conditioning................................................... 192 B.10.2 Statistical and time-frequency analysis ............................................................. 198 B.10.3 Conclusions ....................................................................................................... 201 Appendix C Details of modeling and signal processing ................................................... 202 C.1 Simulation results from a charge transfer and induction model .................................202 C.2 Improvement of charge transfer model .......................................................................206 C.3 MATLAB codes for charge transfer and induction model as well as signal processing and decoupling ...................................................................................................................211  ix List of Tables Table 1.1 Charge density on particles by different powder operations [3, 29]. ........................ 5 Table 1.2 Summary of collision probes previously used in fluidized beds. ........................... 12 Table 1.3 Summary of charge density measurement techniques used in gas-solid fluidized beds. ..................................................................................................................... 13 Table 2.1 Summary of tested probes in this work. ................................................................. 27 Table 2.2 Electrical properties and hardness of different materials used in the experiments [74, 106-115]. ...................................................................................................... 30 Table 2.3 Directly measured charge density (qm), bubble size (DB) and rise velocity (UB) and current peaks (Ipeak) from single bubble injection experiments. .......................... 53 Table 2.4 Directly measured charge density (qm), bubble rise velocity (UB) and current peaks (Ipeak) from freely bubbling experiments with synchronization (see Section 2.4). ............................................................................................................................. 57 Table 2.5 Fitted parameters in Eq. (2.13) for different probe tips and particle size distributions of PE powders. ................................................................................ 62 Table 4.1 Decoupled charge densities and bubble rise velocities based on different decoupling methods for one segment of signals (~1 s duration) for two-dimensional bubbling fluidized bed at different superficial gas velocities, Ug. 104 Table 4.2 Estimated charge densities and bubble rise velocities from 2t1m probe with different configurations by different decoupling methods at different Ug in freely bubbling two-dimensional fluidized bed of 500-600 m GB. ........................... 113 Table 4.3 Average charge densities (qm) measured by Faraday cup (FC) and decoupled from 2t2ma probe, and bubble rise velocities (UB) decoupled from probe at different Ug in 0.10 m ID three-dimensional bubbling bed of 500-600 m GB. ............. 117 Table 4.4 Decoupled qm and UB from 2t1m probes with different configurations and tips and Faraday cup (FC) at different Ug in freely bubbling 0.30 m ID three-dimensional fluidized bed of original PE-I particles. ............................................................. 121 Table 4.5 Mean bubble rise velocity (UB) decoupled from the 1t1m probe in pressurized stainless steel three-dimensional column with original PE-II particles at different operating conditions. .......................................................................................... 125 x Table B.1 Directly measured charge density (qm), bubble rise velocity (UB) and current peaks (Ipeak) from synchronization experiments (see Section 2.4) with 2t2mb probe and 106-212 m GB. ................................................................................................ 168 Table B.2 Directly measured charge density (qm), bubble rise velocity (UB) and current peaks (Ipeak) from synchronization experiments (see Section 2.4) for different cases corresponding to Table 2.5. (Fluidizing gas air: T=222oC, RH=62 %) ........ 169 Table B.3 Ratio of mean currents from Ni and TiN of 2t2ma probe when Ug was changed in 0.10 m ID freely bubbling fluidized bed. (fs= 100 Hz, Fluidizing gas air: T=222oC, RH=102%.) ................................................................................... 177 Table B.4 Faraday cup measurements by scooping particles (static bed height: 0.30 m, Ug= 0.30 m/s). ........................................................................................................... 188 Table B.5 Faraday cup measurements of particles discharged through sampling port within 6 s and 25 s (static bed height: 0.30 m, sampling port: 0.22 m above distributor, Ug= 0.30 m/s). .................................................................................................... 188 Table B.6 Results from Faraday cup measurements of particles discharged through sampling port (Static bed height: 0.30 m, sampling port: 0.22 m above distributor). ....... 189 Table B.7 Minimum fluidization velocities of polyethylene particles (PE-II) at different pressures in pressurized three-dimensional column. ......................................... 191 Table B.8 Sampling parameters for each run in pressurized three-dimensional column. .... 192 Table B.9 Summary of literature on sampling conditions and denoising of fluidized bed electrostatic signals. ........................................................................................... 193 Table B.10 Values used in de-noising for different 1t1m probes and sampling frequencies. ........................................................................................................................... 199 Table C.1 Input parameters used in simulation. ................................................................... 203 Table C.2 Simulated minimum and maximum peaks and their relative differences in Figures C.2 and C.3. ....................................................................................................... 204 xi List of Figures Figure 1.1 Schematic of a Faraday cup (adapted from Cross [30]). ......................................... 6 Figure 1.2 Faraday cup column with quick-opening distributor for measuring particle charging (adapted from Sowinski et al. [32]). ....................................................... 7 Figure 1.3 Experimental setup for particle trajectory tracking (1) fluidized bed (2) air-plate capacitor (3) high tension source (4) “pneumatic gun” (5) valves (6) motion-picture camera (adapted from Wolny [35]). .......................................................... 8 Figure 1.4 Capacitance probe (adapted from Guardiola et al. [36]). ........................................ 9 Figure 1.5 (a) Induction probe (adapted from Chen et al. [22]) and (b) Copper ring probe (adapted from Armour-Chélu et al. [38]). ........................................................... 10 Figure 1.6 Schematics of (a) industrial and (b) laboratory collision probes (adapted from Park et al. [45] and Chen et al. [11]). ........................................................................... 11 Figure 1.7 Sample real-time reactor electrostatic probe data (Marino [90]). ......................... 17 Figure 2.1 Schematic of dual-tip (two-material) probes: (a) design a; (b) design b. .............. 24 Figure 2.2 Schematic of dual-tip (one-material) probe with different configuration and tip shape: F configuration (two tips protruding); Γ configuration (upper tip protruding, lower tip retracted); L configuration (upper tip retracted, lower tip protruding); flat and rod shape tips. ..................................................................... 26 Figure 2.3 Schematic of experimental apparatus for ejector setup: (a) overall setup; (b) elbow and pipe combinations. (PVC: polyvinyl chloride; SS: stainless steel) (not to scale) .................................................................................................................... 29 Figure 2.4 Schematic of technique for measuring average particle velocity at different angles of inclination of probe. ........................................................................................ 32 Figure 2.5 Experimental setup of motor-pulley system and different positions of 2t2mb probe tip (not to scale). .................................................................................................. 33 Figure 2.6 Schematic of experimental apparatus for vertical tube setup (not to scale). ......... 34 Figure 2.7 Schematic of experimental apparatus for vibration tray setup (not to scale). ....... 34 Figure 2.8 Schematic of two-dimensional column, bubble injection system and charge measurement system (not to scale). ..................................................................... 35 xii Figure 2.9 Charge carried by particles after passing through ejector setup shown in Figure 2.3b with different pipe materials; Average gas velocity =2.79 m/s (at outlet of vertical pipe with average pipe inner diameter is 25 mm). .................................. 38 Figure 2.10 Charge carried by particles after passing through ejector setup with different gas velocities for PVC+PVC elbow and pipe combination (see Figure 2.3b). .......... 39 Figure 2.11 Measured charge densities on particles in dilute phase with different pipe materials and particle velocities. .......................................................................... 40 Figure 2.12 Typical cumulative charge signals measured by 2t2ma probe in dilute phase flow with different pipe materials; average particle velocity Vp=4.42m/s; collision angle θ=0 degree. ................................................................................................. 41 Figure 2.13 Characteristics of Ni and TiN tips at different particle velocities in dilute phase flow, collision angle θ=0 degree; four pipe combinations as shown in Figure 2.14b. ................................................................................................................... 42 Figure 2.15 Comparison of fitted and experimental data for (a) dilute phase flow ( > 0.99), and (b) dense phase flow (0.87   0.72)........................................................... 44 Figure 2.16 Relationship between –Itran/Ws and collision angle θ at different particle velocities for two tips in dilute phase flow: (a) Ni and (b) TiN. ......................... 45 Figure 2.17 Relationship between –Itran/Ws and collision angle θ for two tips at Vp=4.42 m/s in dilute phase flow; pipe material combination: Al+Al. .................................... 46 Figure 2.18 Cumulative charge signal measured by 2t2mb probe (facing vertically downward) when negatively charged ball passed by in motor-pulley setup. ...... 47 Figure 2.19 Cumulative charge signal measured by covered tips of 2t2mb probe in two-dimensional fluidized bed: (a) single bubble; (b) free bubbling, Ug =0.30 m/s. . 48 Figure 2.20 Ratio (Cmin,TiN / Cmin,Ni) from 2t2mb probe (see Table 2.1) (a) in motor-pulley setup; (b) in two-dimensional fluidized bed. ....................................................... 49 Figure 2.21 Cumulative charge signals from 2t2ma probe at different locations in vertical tube setup, shown in Figure 2.6. .......................................................................... 49 Figure 2.22 Comparison of cumulative charge signals from 2t2ma probe with both tips uncovered and one tip (Ni or TiN) covered in vertical tube setup (probe at bottom of tube with probe tip facing upward). ................................................................ 50 Figure 2.23 Cumulative charge signal from 2t2ma probe in vibration tray setup. ................. 51 xiii Figure 2.24 Charge (raw signal), current (derivative of charge) signal from 2t2mb probe and synchronized snapshots for single bubble injection in two-dimensional fluidized bed. (Back pressure for bubble injection: 345 kPa; solenoid valve opening time: 0.2 s; ~2E−4 m3 pressurized air injected; centre of probe location: 0.17 m above bubble injector; bubble injector: 0.04 m above distributor) ................................ 52 Figure 2.25 Current signals from 2t1mfF probe for single bubble injection in two-dimensional fluidized bed containing glass beads (Pressure in bubble injector: 345 kPa; solenoid valve opening time: 0.2 s; ~2E−4 m3 pressurized air injected; centre of probe port 0.17 m above bubble injection port; bubble injector: 0.04 m above distributor). ................................................................................................ 54 Figure 2.26 Current signals from 2t1m probe with different configurations for multiple bubble injections in two-dimensional fluidized bed containing glass beads (Pressure in bubble injector: 345 kPa; solenoid valve opening time: 0.2 s and closing time: 3 s; centre of probe port 0.17 m above bubble injection port; bubble injector: 0.04 m above distributor). ..................................................................... 55 Figure 2.27 Charge density decay curve after cutting off gas flow following free-bubbling with compressed air (RH=10%) and nitrogen (RH=2%). ................................... 56 Figure 2.28 Comparison of empirical correlation (Eqs. (2.13-2.15)) and experimental data for two probe materials. ............................................................................................. 59 Figure 2.29 Contour plots of current peaks from Ni tip of 2t2mb probe as a function of bubble rise velocity and particle charge density from calibration experiments for 106-212 m and 500-600 m GB........................................................................ 60 Figure 2.30 Synchronized 2t1mrΓ probe (inserted 0.42 m above the distributor) and differential pressure signals across the bed in two-dimensional fluidization column of 710-850 m PE at Ug= Umf+0.04 m/s. (Fluidizing gas air: T=23.2oC, RH=4.2 %) ........................................................................................................... 62 Figure 2.31 Comparison of fitted data (from Eq. (2.13)) and experimental data for different probe tips and particle size ranges of PE powders. ............................................. 63 Figure 3.1 Schematic of 0.10 m ID three-dimensional column and charge measurement system (not to scale). ........................................................................................... 66 xiv Figure 3.2 Schematic of 0.30 m ID three-dimensional column and charge measurement system (not to scale). ........................................................................................... 67 Figure 3.3 Cumulative charge (a) and current (b) signals from 2t2mb probe in freely bubbling two-dimensional fluidized bed of 500-600 m GB with static bed height of 0.30 m at Ug=0.50 m/s over duration of 5 s. (Fluidizing gas air: T= ~22oC, RH= ~10 %) ......................................................................................................................... 69 Figure 3.4 Mean and standard deviation of current as a function of superficial gas velocity and particle charge density in freely bubbling two-dimensional fluidized bed with a static bed height of 0.30 m. ............................................................................... 70 Figure 3.5 Comparison of max min( ) / 2I I  and min max( ) / 2I I  for both materials at different superficial gas velocities in two-dimensional fluidized bed with static bed height of 0.30 m. ............................................................................................................. 71 Figure 3.6 Cumulative charges and current signals from 2t2mb probe at different superficial gas velocity in two-dimensional freely bubbling bed of 106-212 m GB. (Fluidizing gas air: T= ~22oC, RH= ~10 %) ....................................................... 72 Figure 3.7 Mean currents as a function of superficial gas velocity and particle charge density for 106- 212 m GB. ........................................................................................... 73 Figure 3.8 Cumulative charge and current signals from 2t1mfF probe in freely bubbling two-dimensional fluidized bed of 500-600 m GB at different superficial gas velocities. (Fluidizing gas air: T=~23oC, RH=~3 %) .......................................... 74 Figure 3.9 Cumulative charge and current signals from 2t1m probe with different configurations in freely bubbling two-dimensional fluidized bed of 500-600 m GB at excess superficial gas velocities, Ug−Umf =0.02 and 0.14 m/s. (Fluidizing gas air: T=~23oC, RH=~3 %) .............................................................................. 75 Figure 3.10 Cumulative charge and current signals from 2t1mfF probe in freely bubbling two-dimensional fluidized bed of 710-850 m PE at different Ug. (Fluidizing gas air: T=~23oC, RH=~7 %) .................................................................................... 76 Figure 3.11 Cumulative charge signals from 2t2ma probe horizontally inserted to centre and near the wall in 0.10 m ID freely bubbling fluidized bed (Ug=0.29 m/s). (Fluidizing gas air: T= ~22oC, RH= ~10 %) ....................................................... 77 xv Figure 3.12 Mean and standard deviation of current signals from 2t2ma probe as a function of probe insertion distance. .................................................................................. 78 Figure 3.13 Comparison of cumulative charge signals from 2t2ma probe inserted: (a) horizontally and (b) vertically into 0.10 m ID freely bubbling fluidized bed (Ug=0.33 m/s). (Fluidizing gas air: T= ~22oC, RH= ~10 %) .............................. 79 Figure 3.14 Typical current signals from: (a) homogeneous particulate flow with particles dropped from funnel; (b) 0.10 m ID freely bubbling fluidized bed. .................... 79 Figure 3.15 Current signals from 2t2ma probe inserted vertically into 0.10 m ID freely bubbling fluidized bed at different superficial gas velocities. (Fluidizing gas air: T= ~22oC, RH= ~10 %) ....................................................................................... 80 Figure 3.16 (a) Mean, (b) standard deviation and (c) normalized standard deviation of current signals from 2t2ma probe as function of superficial gas velocity in 0.10 m ID freely bubbling fluidized bed (probe inserted vertically). (Experiments for Ug=0.22, 0.25, 0.28 and 0.31 m/s were conducted on a different day as experiments for Ug=0.33, 0.36, 0.39 and 0.44 m/s) ............................................. 81 Figure 3.17 FFT of current signals from 2t2ma probe in 0.10 m ID freely bubbling fluidized bed at different superficial gas velocities. ........................................................... 82 Figure 3.18 Current signal from 2t2mb probe in 0.30 m ID freely bubbling fluidized bed of 420-590 m GB at different superficial gas velocities. (Fluidizing gas air: T=405 oC, RH=154 %) .................................................................................... 83 Figure 3.19 Cumulative charge and current signals from 2t1m probes at different Ug in freely bubbling 0.30 m ID three-dimensional fluidized bed of original PE-I particles. (Fluidizing gas air: T=405 oC, RH=154 %) .................................................... 84 Figure 3.20 Cumulative charge and current signals from 2t1mfΓ probe (insertion 0.25 m above distributor) at different Ug in 0.30 m ID fluidized bed of original PE-I particles. (Fluidizing gas air: T=405 oC, RH=154 %) ..................................... 85 Figure 3.21 Current signals from 2t1mfF probes at different locations in 0.30 m ID fluidized bed of original PE. (probe 1: z1=0.25 m, probe2: z2=0.70 m and probe 3: z3=1.00 m, Ug-Umf=0.24 m/s) (Fluidizing gas air: T=405 oC, RH=154 %) ................. 86 Figure 3.22 Standard deviations of current signals from 2t1m probes and absolute pressure signals versus superficial gas velocity. ................................................................ 87 xvi Figure 4.1 Flowchart of peak detection and selection algorithm adopted for signal processing. ............................................................................................................................. 90 Figure 4.2 Theoretical charge and current signals received from collision probe when a single bubble passed by. ................................................................................................. 94 Figure 4.3 Principle of 2t1m probe (F configuration) to determine bubble rise velocity and size. ...................................................................................................................... 96 Figure 4.4 (a) Raw data, (b) peak-detected data and (c) peak-selected data, generated from 2t2mb probe in two-dimensional bubbling fluidized bed of glass beads. ........... 99 Figure 4.5 Effects of different parameters on number of selected bubbles from a 40 s period of time-series signals (70 bubbles in total) in bubble selection algorithm of signal processing for 2t2mb probe in two-dimensional fluidized bed. ........................ 101 Figure 4.6 Comparisons of charge density from 2t2mb probe and Faraday cup measurements at different Ug in freely bubbling two-dimensional fluidized bed: (a) Static bed height 0.30 m with probe 0.22 m above distributor; (b) Static bed height 0.50 m with probe 0.42 m above distributor. ................................................................. 106 Figure 4.7 Comparisons of bubble rise velocity from 2t2mb probe and video measurements under different Ug in freely bubbling two-dimensional fluidized bed: (a) Static bed height 0.30 m with probe 0.22 m above distributor; (b) Static bed height 0.50 m with probe 0.42 m above distributor. ............................................................. 107 Figure 4.8 Comparison of decoupled (a) charge densities; (b) bubble rise velocities from 2t2mb probe by different decoupling methods in two-dimensional bubbling bed of 500-600 m GB. ............................................................................................ 108 Figure 4.9 Comparison of decoupled (a) particle charge densities; (b) bubble rise velocities from 2t2mb probe by different decoupling methods in two-dimensional bubbling bed of 106-212 m GB. ..................................................................................... 109 Figure 4.10 Decoupled bubble size and bubble rise velocities from 2t1mfF probe by two tips peak-times method at different Ug in freely bubbling two-dimensional fluidized bed of 500-600 m GB. ..................................................................................... 111 Figure 4.11 Comparison of charge densities and bubble rise velocities from 2t1mfF probe by different decoupling methods with Faraday cup and video measurements at xvii different Ug in freely bubbling two-dimensional fluidized bed of 500-600 m GB. ..................................................................................................................... 112 Figure 4.12 Decoupled bubble size and bubble rise velocity from 2t1mfF probe by two current peaks tips method at different Ug in freely bubbling two-dimensional fluidized bed of 710-850 m PE. ....................................................................... 114 Figure 4.13 Comparison of charge densities and bubble rise velocities from 2t1mfF probe by different decoupling methods with Faraday cup and video measurements at different Ug in freely bubbling two-dimensional fluidized bed of 710-850 m PE. ........................................................................................................................... 115 Figure 4.14 Comparison of charge densities and bubble rise velocities from 2t1m probes of different configurations by different decoupling methods, Faraday cup and video measurements in freely bubbling two-dimensional fluidized bed of 710-850 m PE at Ug−Umf = 0.09 m/s. .................................................................................. 116 Figure 4.15 Comparison of decoupled charge densities from 2t2mb probe by different decoupling methods and Faraday cup in 0.10 m ID three-dimensional bubbling bed of 500-600 m GB. ..................................................................................... 118 Figure 4.16 Comparison of decoupled (a) charge densities; (b) bubble rise velocities from 2t2mb probe by T2-Ni, T2-TiN and TT2 methods in 0.30 m ID three-dimensional bubbling bed of 420-590 m GB. ................................................. 119 Figure 4.17 Comparison of charge densities and bubble rise velocities from 2t1mfF probe by different decoupling methods with Faraday cup measurement at different Ug in 0.30 m ID three-dimensional fluidized bed of original PE-I particles. ............. 120 Figure 4.18 Decoupled charge densities from 2t1mfF probe and Faraday cup at different Ug in 0.30 m ID three-dimensional fluidized bed of original PE-I particles (probe: insertion 0.75 m above distributor, static bed height=0.88 m). ......................... 122 Figure 4.19 Decoupled charge densities from 2t1mfF probes at different axial locations and Ug in 0.30 m ID three-dimensional fluidized bed of original PE-I particles (probe 1: z1=0.25 m, probe 2: z2=0.70 m and probe 3: z3=1.00 m, static bed height=0.88 m). ...................................................................................................................... 123 Figure 4.20 Comparison of average relative errors with results from direct measurements and probes in two-dimensional bubbling bed of 500-600 m GB. .......................... 124 xviii Figure 4.21 Comparison of relative frequency distributions of decoupled bubble rise velocities from 1t1m probes located at different heights in pressurized column of original PE-II particles (Ug= Umf +0.10, absolute pressure in freeboard: 276 kPa). ........................................................................................................................... 126 Figure 5.1 Effect of sampling frequency on estimated charge density from 2t1mfF probe by (a) two tips peak-times method and (b) one-tip time-difference method when Ug-Umf changed abruptly from 0.02 to 0.11 m/s. .................................................... 132 Figure 5.2 Effect of sampling time interval on estimated charge density from 2t1mfF probe by: (a) two tips peak-times method; and (b) one-tip time-difference method when Ug-Umf abruptly changed from 0.02 to 0.11 m/s. ............................................... 133 Figure 5.3 Current signals from 2t1mfF probe in freely bubbling two-dimensional fluidized bed of 500-600 m GB when Ug-Umf abruptly changed from 0.02 to 0.11 m/s. (Fluidizing gas air: T=232oC, RH=22 %.) .................................................... 134 Figure 5.4 Comparison of charge densities and bubble rise velocities from 2t1mfF probe by different decoupling methods with Faraday cup and video measurements in freely bubbling two-dimensional fluidized bed of 500-600 m GB when Ug-Umf changed from 0.02 to 0.11 m/s. (fs= 200 Hz, ts= 10 min, Fluidizing gas air: T=232oC, RH=22 % ) .................................................................................... 134 Figure 5.5 Comparison of charge densities and bubble rise velocities from 2t2mb probe by different decoupling methods with Faraday cup and video measurements in freely bubbling two-dimensional fluidized bed of 500-600 m GB when Ug-Umf abruptly changed from 0.01 to 0.07 m/s. (fs= 200 Hz, ts= 10 min, Fluidizing gas air: T=232oC, RH=52 % ) ............................................................................. 135 Figure 5.6 Current signals from 2t1mfF probe in freely bubbling two-dimensional fluidized bed of 500-600 m GB at Ug-Umf=0.05 m/s after 0.1 wt% Larostat 519 was added. (Fluidizing gas air: T=221oC, RH=21%.) .......................................... 137 Figure 5.7 Comparison of charge densities from 2t1mfF probe by different methods with Faraday cup, and decoupled bubble rise velocities in freely bubbling two-dimensional fluidized bed of 500-600 m GB at Ug-Umf=0.05 m/s after 0.1 wt% Larostat 519 was added. (Fluidizing gas air: T=221oC, RH=21%.) ............. 138 xix Figure 5.8 Cumulative charge and current signals from 2t1mfΓ probe in freely bubbling two-dimensional fluidized bed of original PE-I particles when Ug-Umf changed abruptly from (a) 0.04 to 0.18 m/s; (b) 0.18 to 0.04 m/s. (Fluidizing gas air: T=241oC, RH=61 %) ..................................................................................... 139 Figure 5.9 Comparison of charge densities and bubble rise velocities from 2t1mfΓ probe with Faraday cup and video measurements in freely bubbling two-dimensional fluidized bed of original PE-I particles when Ug-Umf was increased abruptly from 0.04 to 0.18 m/s, then decreased abruptly from 0.18 to 0.04 m/s. ..................... 140 Figure 5.10 Current signals from 2t1mfΓ probe in freely bubbling 0.30 m ID three-dimensional fluidized bed of original PE-I particles when Ug-Umf changed from (a) 0.04 to 0.19 m/s; (b) 0.19 to 0.04 m/s. (Fluidizing gas air: T=405 oC, RH=154 %) ...................................................................................................... 141 Figure 5.11 (a) Comparison of charge densities from 2t1mfΓ probe with Faraday cup measurement in freely bubbling 0.30 m ID three-dimensional fluidized bed with original PE-I particles when Ug-Umf was increased abruptly from 0.04 to 0.19 m/s, and then decreased abruptly from 0.19 to 0.04 m/s . (b) Corresponding derived bubble rise velocities. ........................................................................... 141 Figure 5.12 Current signals from 2t1mrΓ probe in freely bubbling 0.30 m ID three-dimensional fluidized bed of original PE-I particles when Ug-Umf was changed from (a) 0.18 to 0.05 m/s; (b) 0.05 to 0.20 m/s. (Fluidizing gas air: T=405 oC, RH=154 %) ...................................................................................................... 142 Figure 5.13 (a) Comparison of charge densities from 2t1mrΓ probe with Faraday cup measurement in freely bubbling 0.30 m ID three-dimensional fluidized bed with original PE-I particles when Ug-Umf changed abruptly from 0.18 to 0.05 m/s, then from 0.05 to 0.20 m/s. (b) Corresponding derived bubble rise velocities. ........ 143 Figure A.1 Photographs of 3D printed parts with metal tips and fabricated probes. ............ 162 Figure A.2 Photographs of ejector-funnel setup. .................................................................. 163 Figure A.3 Photographs of (a) motor-pulley and (b) vibration tray setups. ......................... 164 Figure A.4 Photographs of two-dimensional fluidization column (Its width is 0.307 m). ... 165 Figure A.5 Photographs of 0.10 m ID and 0.30 m ID three-dimensional Plexiglas fluidization columns. ............................................................................................................. 166 xx Figure A.6 Photographs of 0.15 m ID pressurized three-dimensional stainless steel fluidization column. ........................................................................................... 167 Figure B.1 XPS of Ni and TiN tips before and after experiments. ....................................... 174 Figure B.2 Schematic of setup for testing probe response time. .......................................... 174 Figure B.3 Changes in mean currents, ratio and difference of mean currents from Ni and TiN of 2t2ma probe in 0.10 m ID freely bubbling fluidized bed when Ug changed continuously from 0.22 to 0.36 m/s. (fs= 100 Hz, ts= 5 min, Fluidizing gas air: T=222oC, RH=102 % ) .................................................................................. 176 Figure B.4 Experimental minimum fluidization velocity (Umf) curve for original PE-I (100-1500 m) in two-dimensional fludization column. ........................................... 179 Figure B.5 Experimental minimum fluidization velocity (Umf) curve for sieved PE (710-850 m) in two-dimensional fludization column. .................................................... 179 Figure B.6 Experimental minimum fluidization velocity (Umf) curve for GB (106-212 m) in two-dimensional fludization column. ................................................................ 180 Figure B.7 Experimental minimum fluidization velocity (Umf) curve for GB (500-600 m) in two-dimensional fludization column. ................................................................ 180 Figure B.8 Experimental minimum fluidization velocity (Umf) curve for original PE-II (wide size distribution) in pressurized three-dimensional fluidization column. (absolute pressure in freeboard: 138 kPa) ......................................................................... 181 Figure B.9 Experimental minimum fluidization velocity (Umf) curve for original PE-II (wide size distribution) in pressurized three-dimensional fluidization column. (absolute pressure in freeboard: 276 kPa) ......................................................................... 181 Figure B.10 Experimental minimum fluidization velocity (Umf) curve for original PE-II (wide size distribution) in pressurized three-dimensional fluidization column. (absolute pressure in freeboard: 414 kPa) ......................................................................... 182 Figure B.11 Particle size distribution of GB (500-600 m). ................................................ 183 Figure B.12 Particle size distribution of GB (106-212 m). ................................................ 183 Figure B.13 Particle size distribution of GB (420-590 m). ................................................ 184 Figure B.14 Particle size distribution of original PE-I (100-1500 m). ............................... 184 Figure B.15 Particle size distribution of sieved PE (710-850 m). ...................................... 184 xxi Figure B.16 Example of synchronized frame with 2t2mb probe signal from freely bubbling two-dimensional fluidized column at Ug=0.39 m/s. .......................................... 185 Figure B.17 Labview VI for simultaneous start of analog input and output. ....................... 186 Figure B.18 Cumulative charge signals from 2t2mb probe in freely bubbling two-dimensional fluidized column with glass beads at different Ug (Static bed height: 0.30 m, probe: 0.22 m above distributor, Exps 1 and 5 correspond to Table B.6). ........................................................................................................................... 190 Figure B.19 Locations of 1t1m probes in pressurized three-dimensional fluidization column. ........................................................................................................................... 192 Figure B.20 Cumulative charge (a) and current (b) signals from 1t1m probes in pressurized three-dimensional column (Ug=Umf+0.1 m/s, absolute pressure in freeboard region: 276 kPa, fs=25 Hz, ts=600 s). ................................................................ 194 Figure B.21 Mean current signals from 1t1m probes in pressurized three-dimensional column (fs=200 and 300 Hz, Ug= Umf+0.1 m/s). ............................................................ 194 Figure B.22 FFT power spectra of raw current signals and noise from 1t1m probe (probe B) in pressurized three-dimensional column for different operating conditions (fs= 300 Hz). ............................................................................................................. 195 Figure B.23 FFT power spectra of raw signal, noise and denoised signals by different methods from 1t1m probe (Probe B) in pressurized three-dimensional column (Ug=Umf+0.1 m/s, absolute pressure in freeboard: 276 kPa and fs=300 Hz). .... 196 Figure B.24 Synchronized pressure (a) and current signal (1t1m probe, Probe A) (b) of 5 s period for different sampling frequencies at Ug=Umf+0.1 m/s and absolute freeboard pressure of 276 kPa. .......................................................................... 197 Figure B.25 Pressure and current signals from different 1t1m probes for different sampling frequencies at Ug=Umf+0.1 m/s and absolute freeboard pressure of 276 kPa.... 198 Figure B.26 (a) Mean and (b) standard deviation of denoised current signals from different 1t1m probes and sampling frequencies at Ug=Umf+0.1 m/s and absolute freeboard pressure of 276 kPa. ........................................................................................... 200 Figure B.27 FFT spectra of bed pressure (P) (left) and current (1t1m probe, Probe C) (right) signals under different operating conditions for sampling frequency of 200 Hz. ........................................................................................................................... 201 xxii Figure C.1 Representation of specific charge distribution due to two additive components. ........................................................................................................................... 202 Figure C.2 Simulated induced charges and currents for different DB (50, 80 and 100 mm) when qm0= -1C/kg. ........................................................................................... 204 Figure C.3 Simulated induced charges and currents with different qm0 (-1, -2 and -5C/kg) when DB= 100 mm............................................................................................. 204 Figure C.4 Simulated transfer currents from Ni and TiN tips of 2t2m probe caused by bubble passing and decoupled charge density and average particle velocity. ............... 205 Figure C.5 Schematic of model improvement. ..................................................................... 210 xxiii List of Symbols and Abbreviations ai fitted constant in Eq. (2.5), dimensionless A area of parallel plate, m2 Ap probe tip surface area, m2 bi fitted constant in Eq. (2.5), C s2/kg m2 ci fitted constant in Eq. (2.5), C/kg C capacitance between contact bodies (probe and particles), F Cmin minimum charge, C ds average particle diameter, m D electric flux density, C/m2 DB bubble size/diameter, m DP ball probe diameter, m E electric field due to movement of cloud of particles, V/m f particle collision frequency, s-1 fs sampling frequency, s-1 H cross-correlation integration time period, s I total current, A Ī average current, A kb constant in Eq. (C.10), F-1 kc constant in Eq. (C.6), F-1 ke constant in Eq. (C.9), F-1 kr constant in Eq. (C.15), F-1 L vertical distance between probe center and tip of bubble injector, m mp mass of particles, kg n number of collisions, dimensionless n0 characteristic number of particle electrification, dimensionless q charge on particles, C qe equilibrium charge, C qi initial charge, C qm charge density or specific charge on particles, C/kg qm0 initial charge density or specific charge, C/kg q∞ charge on particle after collision, C Q squared distance between two attractors in Eq. (1.1) xxiv Q  estimator of Q in Eq. (1.1) Qs surface charge density on particle, C/m2 R cross-correlation function of two time-series signals RB radius of bubble, m RP radius of probe, m RH relative humidity S estimator for the normalized squared distance between two attractors, defined in Eq. (1.1) t time, s ts sampling time interval, s T temperature, oC UB average bubble rise velocity, m/s Ug superficial gas velocity, m/s Umf minimum fluidization velocity, m/s V total potential difference, V Vp average particle velocity, m/s Vnp normal component of average particle velocity, m/s Vtp tangential component of average particle velocity, m/s Vb non-contact potential difference caused by space charge effect, V Vc contact potential difference based on surface work functions, V VC variance in Eq. (1.1) Ve non-contact potential difference caused by image effect, V Ws mass flow rate of particles striking probe surface, kg/s z0 critical gap between contact bodies, m  Subscripts B bubble sd standard deviation  tran transfer ind induction max maximum min minimum  xxv Greek letters i fitted parameter in Eq. (2.13), kg/m i fitted parameter in Eq. (2.13), C s2/kg m2 i fitted parameter in Eq. (2.13), C/kg q transferred charge at time interval t, C q0 impact charge at zero initial charge, C t time interval during contact, s x moving distance of particles during time interval t, m z distance between upper and lower tips, m  time difference between maximum and minimum peaks from one tip, s B time for single bubble to pass probe, s  local voidage, dimensionless mf voidage at minimum fluidization, dimensionless θ collision angle, degrees θ’ collision angle where current starts to decrease with increasing angle, degrees  maximum tolerable time difference between current peaks from two materials, s  permittivity of medium, F/m o permittivity of vacuum or air, 8.854×10-12 F/m r relative permittivity or dielectric constant, i.e./o, dimensionless p particle density, kg/m3  coefficient defining threshold value used in peak detection, dimensionless P effective work function of particles, eV M work function of metal, eV tip ratio of mass of particles striking probe to mass of dropped particles, dimensionless  lower boundary of ratio of current peaks from two materials, dimensionless  upper boundary of ratio of current peaks from two materials, dimensionless  Abbreviation Al aluminum  FC Faraday cup GB glass beads Larostat 519 one type of antistatic additives xxvi Ni nickel PE polyethylene particles PVC polyvinyl chloride SD standard deviation SS stainless steel  TiN titanium nitride T1 transfer-induction method  T2 time difference method  TT1 two transfer currents method TT2 two current peaks method 2t2m dual-tip (two-material) probe 2t2ma dual-tip (two-material) probe (design a) 2t2mb dual-tip (two-material) probe (design b) 2t1m dual-tip (one-material) probe 2t1mfF dual-tip (one-material) probe (flat tips, F configuration) 2t1mf dual-tip (one-material) probe (flat tips, Γ configuration) 2t1mfL dual-tip (one-material) probe (flat tips, L configuration) 2t1mrΓ dual-tip (one-material) probe (rod tips, Γ configuration) 2t1mfrΓ dual-tip (one-material) probe (flat and rod tips, Γ configuration) 1t1m one-tip (one-material) probe (conventional probe)  xxvii Acknowledgements First and foremost, I would like to express my sincere gratitude to my supervisors, Prof. Xiaotao Bi and Prof. John Grace, for their excellent guidance and continuous support in all my research efforts. Thanks Dr. Bi for giving me the opportunity to be part of renowned Fluidization Research Group; and thanks Dr. Grace for enlarging my vision of science. I am very treasured and honored to have such successful role models in my life.   I would like to thank my research committee members, Prof. Jim Lim and Prof. Jane Z. Wang, for their valuable suggestions, support and care throughout my PhD project. They always provide coherent answers to my endless questions.  I would like to acknowledge NOVA Chemicals Corporation and NSERC for providing financial support to this research project. Many thanks to Dr. Bob Quaiattini and others from NOVA Chemicals for reviewing and approval my publications, as well as helpful discussions.   I am grateful to fellow professors, graduate students, visiting scholars in the group and Department of Chemical and Biological Engineering. Special thanks to Prof. James Feng for letting me use the high speed camera; Prof. Norman Epstein, Prof. Naoko Ellis, Prof. Lifeng Zhang, Prof. Cai Liang, Dr. Turki A. Al-Smari, Dr. Farzaneh Jalalinejad, Dr. Zhiwei Chen, Dr. Yulong Ding and Mr. Hafiz Rahman for their continuous help, useful discussion and encouragement during my studies. I would also like to thank the staff in workshop, store and administration of the department for their professional support.   I wish to appreciate Dr. Ted Knowlton and Dr. Reddy Karri from PSRI, Prof. Allisa Park from Columbia University and Prof. Poupak Mehrani from University of Ottawa, for their helpful discussions and advices.  Last but not least, I am sincerely grateful to my parents for their endless support and encouragement throughout my years of education. 1 Chapter 1 Introduction 1.1 Electrostatics in gas-solid fluidized beds Fluidization occurs when solid particles are transformed into a fluid-like state by being suspended in a gas [1]. Gas-solid fluidized bed reactors have advantages of rapid solids mixing, temperature uniformity, excellent heat transfer properties, relatively low pressure drops, ability to continuously add/remove particles and capacity to add liquids. Therefore they have been widely used in numerous industrial applications such as drying, coating, combustion, gasification, metallurgical ore roasting, fluid catalytic cracking, and acrylonitrile and polyethylene production.   Electrostatic phenomena in particle handling related processes are complex and unpredictable [2]. They were first reported in 1950s. The electrostatic charges in gas-solid fluidized bed reactors and gas-solid transport lines result from a balance between charge generation and dissipation, and can significantly affect the operation. They may be generated by different contact terms: particle-particle, particle-gas, and particle-wall. Charges can cause the particles to adhere to the reactor wall, then fuse and melt together on the reactor wall to form “sheets”. Significant reactor wall sheeting in polyolefin reactors can cause plugging of the reactor product discharge system or loss of fluidization. These consequences typically require reactor shutdown to have the sheets removed, with significant negative economic impact each time both in lost production and in maintenance costs to clean the reactor. Thus there are significant economic incentives to prevent the formation of wall sheets [3]. Moreover, accumulation of electrostatic charges can cause hazardous electrical discharges, leading to sparks, fires or even explosions.   1.2 Problem definition To avoid the adverse effects of electrostatics outlined above, effective measurement techniques are needed to monitor electrostatic charges in fluidized beds. Understanding the cause of sheeting with metallocene catalysts in fluidized bed polyethylene reactors has for 2 many years been hampered by the lack of suitable instrumentation. Most sheeting incidents with these catalysts have occurred with little or no advanced warning by previously used process instruments, including conventional static probes [4-8]. Moreover, the development of high-activity catalysts combined with advanced reaction technologies has created a growing need to monitor electrostatic behaviour. This requires the development of reliable and accurate measurement techniques which can provide transient local particle charge density, as well as charge density distribution across a spectrum of particle sizes [9].  There are many different types of measurement tools to measure electrostatic charges, but none of these is standardized. Collision-type electrostatic probes [10, 11] are most common in industry. However, it is difficult to interpret voltage or current signals acquired by these probes. The current or voltage signal received from a collision probe not only reflects the particle charge density in the bed, but it is also a function of local dynamic properties, such as bubble size and rise velocity. Thus, the current or voltage from a collision probe is also influenced by changes in local hydrodynamics. The commonly-used collision probe cannot differentiate signal changes caused by charge density changes from those due to hydrodynamic changes. Charge density and hydrodynamic information are both embedded in the time-varying charge signals measured by electrostatic probes in fluidized bed reactors. To determine the charge density using collision probes, it is essential to decouple hydrodynamic information from charge variations.   An ideal probe would decouple the hydrodynamics from the charge density. Until now, electrostatic charge signals registered by collision probes have been poorly understood. Proper interpretation of these signals could help to understand the relationship between electrostatic charge and the hydrodynamics of fluidized beds.   1.3 Thesis objectives The intention of this thesis is to provide a reliable tool to monitor in-situ particle charge density levels in reactors to assist in the prevention of charge buildup in the reactor, associated wall sheeting and possible accidents. Also this tool could help to measure the 3 dynamic electrostatic charging properties of bulk powders, assisting in the development of particle coating, mixing and transportation, pharmaceutical powders, their delivery systems and other industrial powder handing processes [12-14].  The objective of this study is to develop a model-based signal analysis tool in conjunction with experimental confirmation for monitoring electrostatic charges and hydrodynamics behaviour in gas-solid fluidized beds. The central goal is to in-situ measure particle charge density levels and bubble properties inside fluidized beds by a novel electrostatic probe combined with suitable signal analysis methods. Other objectives include:  To improve the interpretation and understanding of charge signals registered by electrostatic probes;  To monitor the electrostatics and hydrodynamics changes in gas-solid fluidized bed;  To examine the effects of particle properties and column scale on particle charging and signal analysis;  To improve understanding of electrostatic charge buildup and dissipation processes in gas-solid fluidized beds.  1.4 Thesis outline Chapter 1 introduces the background and objectives of this study, then reviews relevant literature on electrostatics and hydrodynamics in fluidized beds, electrostatic measurement techniques, particle charging models and electrostatic signal analysis in pneumatic transportation and the fluidized beds. Chapter 2 describes the design/fabrication and calibration of novel electrostatic probes. The results from measurements of fabricated probes in different fluidized beds are presented in Chapter 3. Chapter 4 focuses on proposing several signal analysis and decoupling methods and provides results from analyzing the probe signals. The ability of the probes for in-situ monitoring of charge and hydrodynamic changes in fluidized beds is reported in Chapter 5. Chapter 6 presents overall conclusions and recommendations for future work.  4 1.5 Interaction between electrostatics and hydrodynamics in gas-solid fluidized beds Electrostatic charges have been found to influence fluidized bed hydrodynamics, including bubble size and shape, particle velocity and fines entrainment in different fluidization equipment [15-18]. In circulating fluidized beds, electrification of the particles has been found to affect gas and solid velocities. Modeling of a riser [19] revealed that an electrical field can force particles towards the outer wall, so that electrostatic charges can promote the well-known tendency for most particles to reside in an outer annular layer in the circulating fluidized bed risers. Electrical charges in the bed can affect bubble size, but contradictory results have been reported in the literature. Some have indicated an increase in bubble size [15], whereas others [16] reported a decrease in bubble size due to increasing bed charge. Some researchers tried to find the charge distribution around the bubbles in a two-dimensional fluidized bed by interpreting collision probe and induction probe signals [20-24]. Their results suggested that the charge density outside the bubble gradually decreases toward the bubble-dense phase interface, with a nearly zero charge density inside the bubble. Jalalinejad et al. [25] adapted the Computational Fluid Dynamics Two-Fluid-Model in MFIX (an open-source code originated by the U.S. Department of Energy) to investigate the effect of electrostatics on hydrodynamics. This model predicted that electrostatic charges can cause a single bubble to elongate and rise more quickly.  Changes of hydrodynamics in gas-solid fluidized beds could cause a change in electrostatics signals registered by probes because of the change in the size and rise velocity of bubbles, as well as contacts among particles, and between the particles and reactor wall. Effects of several factors (e.g. pressure, superficial gas velocity, bubble size, initial bed height and distributor) on bed electrification have been investigated [16, 26, 27]. To determine the charge density and the hydrodynamics using static probes, one needs to separate hydrodynamic information from charge signals.  5 1.6 Electrostatic charge measurement techniques The fundamental electrical quantities of measurements are electrostatic charge, current (charge transfer rate), and voltage (electric potential difference). Electrical signal data are usually expressed as charge density (C/m3, C/kg) or current density (A/m3, A/kg) [28]. Charge density on particles resulting from different powder operations are summarized in Table 1.1. With few exceptions, polymeric powders are poor conductors (with a volume resistivity of 108-16 Ω m). For industrial grade polymeric powders, the level of charge produced per unit mass of powder depends more on the amount of work done on the powder during handling/processing than on the chemical composition of the material [29]. Fluidization results in the highest charge density, likely because of the greater surface area over which charges can be generated. Table 1.1 Charge density on particles by different powder operations [3, 29]. Operation Charge density (C/kg) Sieving 10-5~10-3 Pouring 10-3~10-1 Scroll feed transfer 10-2~1 Grinding 10-1~1 Micronizing 10-1~10-2 Pneumatic conveying 1~102 Fluidization 10~102  The polarity and charge density on each particle provide crucial information on the degree of particle charging and the magnitude of electrostatic forces acting on individual particles [9]. Various techniques have been used to measure particle charge density in gas-solid flow systems. There are two main techniques to measure the electrostatic charges− Faraday cup (direct method) and electrostatic probes (indirect method).   1.6.1 Faraday cup and particle trajectory tracking methods A Faraday cup is a double-wall vessel of any suitable shape. The outer wall is grounded and forms an electrical screen preventing external charges from affecting the measurements. The 6 inner wall is connected to an electrometer which measures the charge by detecting the voltage built-up across a known capacitance, as shown in Figure 1.1. When a charged object enters the inner cup, an equal and opposite charge is induced on the wall of the inner cup. This charge is stored on the capacitor in the electrometer and measured [30].    Figure 1.1 Schematic of a Faraday cup (adapted from Cross [30]).  Based on this principle, Mehrani et al. [31] developed a Faraday cup fluidization column with a copper column (diameter 0.1 m, height 2.1 m) inside a surrounding copper column (diameter 0.2 m, height 1.7 m) as the outer cup. The outer column was grounded to eliminate external electrical interference. The fluidization column was insulated from other parts of the inner and outer columns by Teflon cylinders and a Teflon distributor plate, and connected to an electrometer to measure the charges induced on the wall. Recently, an improvement was made so that bed material and fines can be discharged directly into two separate Faraday cups at the top and bottom of the bed, as shown in Figure 1.2 [32-34]. In this way, the average charge density in the bed and fines could be measured after each fluidization test. The Faraday cup measures the charge density of particles withdrawn (sampled) from the fluidized bed, so it is unable to reveal the local transient charge density distribution inside the bed. As an off-line measurement tool, it is unable to monitor industrial reactors in-situ. Moreover, charge generation or dissipation during particle sampling may affect the measurement accuracy. The Faraday cup is an open system susceptible to variations in environmental factors such as atmospheric air humidity, electromagnetic noise and sample discharging from the fluidized bed, and this renders it unreliable [14].  7 In particle trajectory tracking, single particles are ejected from the bed into a homogenous electric field by means of a local pneumatic impulse, as shown in Figure 1.3. Trajectories of ejected particles are captured by a camera, and particle charge densities are then determined by analyzing the trajectories of ejected particles subject to the electric and gravitational fields [35].   Figure 1.2 Faraday cup column with quick-opening distributor for measuring particle charging (adapted from Sowinski et al. [32]).       8   Figure 1.3 Experimental setup for particle trajectory tracking (1) fluidized bed (2) air-plate capacitor (3) high tension source (4) “pneumatic gun” (5) valves (6) motion-picture camera (adapted from Wolny [35]).  1.6.2 Probes Electrostatic charge buildup inside fluidized beds has been measured by electrostatic probes of three major types: Capacitance Probes, Induction Probes and Collision Probes. Unlike the Faraday cup, which is a static measurement tool, electrostatic probes signals contain dynamic information on particle charging and hydrodynamics inside the bed.   1.6.2.1 Capacitance probes  Guardiola et al. [36, 37] used a capacitance probe to measure the degree of electrification in fluidized beds, as shown in Figure 1.4. In their technique, the probe and distributor were considered to be parallel metallic plates, while the bed acted as a dielectric medium. The probe-to-distributor voltage drop was measured. This method averages the charge over most of the bed cross-section.       9   Figure 1.4 Capacitance probe (adapted from Guardiola et al. [36]).  1.6.2.2 Induction probes The fundamental principle of an induction probe is that a real charge induces an image of itself on a conducting surface. There are two main types. One involves a ball or bar probe with a shielded head, as shown in Figure 1.5(a) [22]. Another, often used in pneumatic conveying systems, is a ring probe, as shown in Figure 1.5(b) [38]. These non-contacting probes have the advantage of not disturbing the flow since they are not directly exposed to the fluidized material. However, their disadvantage is that they are not directly exposed to the fluidized material. Particle-wall interactions, rather than particle-particle interactions, determine the output. Therefore induction probes are unsuitable for providing local information on non-homogeneous flow systems. 10  (a)  (b) Figure 1.5 (a) Induction probe (adapted from Chen et al. [22]) and (b) Copper ring probe (adapted from Armour-Chélu et al. [38]).   1.6.2.3 Collision probes Collision probes are the most commonly used to measure electrostatics in the industry. A unit is shown in Figure 1.6(a). Made of highly conductive materials, they are often used to measure the charge or current induced and transferred to the probe tip by charged particles. Ciborowski and Wlodarski [39] developed an electrode made of platinum wire (0.5 mm diameter), ending in a “smallish” (~5 mm diameter estimated based on a photograph) ball, tethered inside the fluidized bed by a silk thread and connected to an electrometer to measure 1 cm diameter copper probe Teflon cylinder with copper shield to charge amplifier coaxial cable grounded shield on column wall Plexiglas column wall           emulsion phase    Plexiglas column wall             bubble phase 11 the electrical potential in a fluidization column of 0.06 m diameter and 0.555 m height. Fujino et al. [40] adopted a similar approach by inserting into the fluidized bed a spherical brass terminal of 6.0 mm diameter, tethered by a nylon thread and connected to an electrometer. A grounded brass distributor plate served as the reference electrode. Park et al. [41] and Chen et al. [11] mounted collision ball probes (as shown in Figure 1.6(b)) to measure charges induced and transferred by particles surrounding rising bubbles in a two-dimensional fluidized bed. Moughrabiah et al. [42, 43] and Wang et al. [44] installed collision probes at various heights along their columns to measure the charge distribution inside fluidized beds. The collision ball probes used by previous researchers and their features are compared and summarized in Table 1.2.   (a)   (b) Figure 1.6 Schematics of (a) industrial and (b) laboratory collision probes (adapted from Park et al. [45] and Chen et al. [11]).   12 Table 1.2 Summary of collision probes previously used in fluidized beds. Reference Structure & materials Probe tip Layers (from outside to inside) Connect to electrometer (electrode) Ciborowski and Wlodarski [39] N/A (Mounted with the help of silk thread) Platinum wire Platinum ball (0.5 mm D) Fujino et al. [40]  N/A Nylon thread Brass ball (6 mm D) Tardos and Pfeffer [10] N/A N/A N/A Park & Chen et al. [21, 41] Brass tube (6.7 mm O.D.)→Glass sleeve (5.8 mm O.D.) Alumel Stainless steel ball (3.2 mm D) Moughrabiah et al. [42] Brass tube (12.7 mm O.D.)→Polyethylene tube (8.75 mm O.D.) →Ceramic tube (5.5 mm O.D.) →Teflon tube Copper wire Stainless steel ball (5.3 mm D) Wang et al. [44] Teflon sleeve (6.8 mm D) →Brass sleeve (5.2 mm) →Teflon sleeve (3.8 mm) Brass Copper bar (1 mm)  Table 1.3 lists the measurement techniques used to determine particle charge density in laboratory gas-solid fluidized beds. Collision probes have rarely been employed for this purpose because the current or voltage signals received from them not only reflect the particle charge density in the bed, but also local dynamic properties, such as bubble size and velocity.  13 Table 1.3 Summary of charge density measurement techniques used in gas-solid fluidized beds. Researchers Flow regime in the bed Particles Measurement technique Signal and interpretation Tardos and Pfeffer [10] Bubbling Porcelain Faraday cup qm=q/mp Fasso et al. [46] Bubbling Glass beads Faraday cup qm=q/mp Wolny and Opaliǹski [47] Bubbling Polystyrene Faraday cup qm=q/mp Wolny and Kaźmierczak [35] Bubbling Polystyrene Tracking particle trajectory Single particle trajectory, numerical analysis of trajectory of charged particle in electrical and gravitational field Napier [48] Bubbling Glass beads, sugar Field meter qm=q/mp Jiang [49] Fast Fluidization FCC, Polyethylene Faraday cup qm=q/mp Tucholski and Colver [50] Fast Fluidization Glass beads Faraday cup qm=q/mp Mountain et al. [51] Bubbling Polymers Faraday cup qm=q/mp Ali et al. [52]; Zhao et al. [53] Bubbling Polyamide Faraday cup qm=q/mp Revel et al. [54] Slugging Polyethylene Faraday cup qm=q/mp Murtomaa et al. [55] Bubbling Glass beads, lactose, cellulose Faraday cup qm=q/mp Chen et al. [11] Single bubbles Glass beads Collision Probe Charge, fitting Cmin to a charge induction model with known bubble size Mehrani  et al. [56]; Omar et al. [57] Bubbling Glass beads, Polyethylene Faraday cup column qm=q/mp Chen et al. [23, 24] Bubbling Glass beads Multi induction probes Induced charge, signal reconstruction to obtain charge density around bubble Sowinski et al. [33, 34] Bubbling and slugging Polyethylene Faraday cup column qm=q/mp 14 1.7 Particle charging models Although the above measurement techniques provide information about electrostatics in fluidized beds, what these signals really mean is largely unknown. Several groups have attempted to develop models to link the charges on particles in pneumatic transport lines and gas-solid fluidized beds to signals registered by different types of probes.  1.7.1 Models for pneumatic transport Dilute flow in pneumatic transport pipelines is relatively simple because it is a dilute phase flow, and the measured charge or current is mainly caused by collisions between the particles and the wall. Zhu and Soo [58] modified a previous ball probe theory to apply to a dense gas-solid suspension, to estimate the electric current passing through the ball probe due to collisions between the probe and particles in pneumatic transport lines by analogy between the mechanisms of electric charge transfer and heat transfer. Cole et al. [59] proposed that charge transfer depends on the total potential difference V, which is made up of three parts: Vc based on the surface work functions, Ve caused by the image effect and Vb caused by the space charge effect. Based on this, Matsusaka et al [60, 61] proposed a theory of wall charging in gas-solid pipe flow, in which the current flow to the pipe wall caused by collisions between particles and the outer wall depends on the mass flow rate and charge density. They also introduced a way to measure the mass flow rate and charge density of particles simultaneously by two short pipes made of different materials of substantially different work function.  1.7.2 Models for fluidized beds The situation in the fluidized bed is more complex than in pneumatic transport pipelines. Park et al. [20] introduced a model to illustrate the linkage between signals from ball probes and specific charge density of bed particles. The total charge consists of two parts: induced charge and transferred charge. Charge induction occurs when a charged object repels electrons from, or attracts electrons to, the surface of a second object. This creates a charged region in the second object being charged. Charge transfer occurs to generate a contact 15 potential difference between contacting surfaces when different materials come into contact with each other. The model considered the situation where a single gas bubble surrounded by charged particles passed a probe. The charges on particles induce charges onto the probe, and charges on particles are also transferred to the probe due to collisions between the particles and probe. The model has some limitations: There is uncertainty in estimating the dielectric constant; frictional charging between the probe and particles is neglected; the charge density on the surface of the bubble is assumed to be uniform, but it is not uniform in practice; particles inside the bubble which affect the output when the probe is inside the bubble are neglected.  Chen et al. [21] modified this model by considering the charge distribution around the bubble and the background charge density, subject to charge build-up on particles remote from the bubble; they also allowed for the effect of bubble wakes and drift. The model predictions for bubbles passing the probe were in good agreement with experimental results.  1.8 Electrostatic signal analysis Signals from fluidized beds such as absolute or differential pressure signals [62-66], voidage [67, 68], vibration [69] and acoustic signals [70-72], have been widely analyzed in the literature. Unlike those signals which are affected by only one physical factor, such as pressure signals, electrostatic signals contain both hydrodynamics and charge information. This section first reviews electrostatic signals from pneumatic conveying systems, followed by the analysis techniques for electrostatic signals from different fluidization systems in the literature.   1.8.1 Electrostatic signals from pneumatic transportation  In pneumatic conveying, charges are mainly generated by collisions between particles and pipe inner walls. Signals from two separate electrostatic rings are often cross-correlated to measure the solid flow rate, concentration and particle velocity [73-76]. Matsusaka et al. [77] analyzed the charge signals from a detection metal pipe. The signals varied over a wide range, from positive to negative, and contained two components, one induced by the 16 transport of a cloud of charged particles and the other transferred by repeated impacts of particles with the inner wall. The average velocity of the cloud flowing in the pipe can be obtained from the time interval between peak signals measured with a detection pipe. The Hilbert–Huang transform (HHT) or a neutral network has been used to analyze electrostatic fluctuations detected by a ring-shape electrostatic sensor, revealing information on the gas-solid flow regime and hydrodynamics in pneumatic transport [78-81]. The HHT of the electrostatic fluctuation signal can show how the instantaneous frequencies and amplitudes of the electrostatic fluctuation vary with time, thus revealing its non-linear and non-stationary intrinsic nature. The energy distribution of intrinsic mode functions (IMFs) with different scales shifts from coarse scale (low frequency oscillations) to fine scale (high frequency oscillations) with increasing superficial gas velocity and decreasing particle concentration.   Zhang et al. [82] developed an integrated electrostatic and capacitance technique for measuring the volumetric concentration of coal and biomass, obtained from the outputs of electrostatic and capacitance sensors by a dual regression method in a three phase flow in a pneumatic conveying pipeline. Hussain et al. [83] developed a ring-shaped induction sensor for measuring charge and polarity of charged particles. Bunchatheeravate et al. [84] developed a method to predict particle charging in a straight pipe by a calibration equation and parameters. First, experiments on particle charging in a 1 m long pipe section were performed to obtain a model equation with two parameters related to the properties of the particle and pipe wall materials. Then, with equation and parameters, the particle charging in a straight pipe of any given length can be predicted. However the prediction of particle charging by this method has two limitations: first, particle charging must be via an impact charging mechanism; second, the particle, pipe wall materials and the operating conditions in the prediction must be the same as those in which the parameters and equation were determined.  Overall, the charge signals from the pneumatic conveying system have been given much attention, whereas there has been little analysis of the electrostatic signals from fluidization systems. Due to the complexity of flow in fluidized beds, it is difficult to interpret voltage or current signals measured by collision-type electrostatic probes inside fluidized beds. 17 1.8.2 Electrostatic signals from fluidized beds This section reviews work on analysis of electrostatic signals by statistical, time-frequency, non-stationary analysis and signal reconstruction methods. Some studies have also shown that electrostatic signals can provide useful information on local fluidization characteristics [85], moisture content [86, 87], proportion of fines in the bed [88] and bed level [44].  1.8.2.1 Statistical analysis Mean and standard deviation are the most commonly used characteristics for characterizing electrostatic levels in fluidized beds. In industrial applications, signals measured by electrostatic probe (see the example shown in Figure 1.7) are processed by means of averaging, with periods of averaging from 10 milliseconds to 10 hours. The signal may also be treated to provide a root mean square (RMS), a standard deviation of the signal, an absolute value of the signal, or an average absolute value of the signal [6, 8, 89].     Figure 1.7 Sample real-time reactor electrostatic probe data (Marino [90]).  The simplest method analyzing electrostatic signals is to plot a sequence of data points of the measured signal. Boland et al. [91, 92] observed that the voltage signals from a shielded ball probe are similar in trend to pressure signals in the bed. The mean of current/charge generally represents the net transferred part, while the standard deviation of current/charge portrays the net induction part [93]. The experimental set-up of Liu at al. [94] consisted of a pressurized 18 gas-solid fluidized column and a charge measurement system. Seven collision probes were deployed at different radial and axial positions. Experiments were then carried out at different operating pressures and superficial gas velocities. The pressure drop across the bubbling bed was also measured. The charge behaviour of various probes was investigated by comparing the fluctuations and polarity of charge signals. The influence of superficial gas velocity on the charge signals was studied by comparing standard deviations of the pressure drop and cumulative charge signals. When the superficial gas velocity increased, the standard deviations of both pressure drop and cumulative charges increased, and the characteristic frequency of cumulative charges changed, consistent with the pressure drop signals.   Tiyapiboonchaiya et al. [27] investigated the average current measured by several copper strips in a three-dimensional Plexiglas column when fluidizing polypropylene particles. It was found that the average current reached a maximum value in the lower part of the bed and decreased with increasing height, because of more collisions among particles, between solids and walls, and greater friction between the solid and gas at a higher inlet gas velocity.   Park and Fan [45] studied electrostatic phenomena of high density polyethylene (HDPE) particles in a gas-liquid-solid fluidized bed, with liquid as the continuous phase. The results showed that the mean voltage measured by a commercial electrostatic probe did not change significantly with varying superficial gas velocity, whereas the standard deviation of the voltage output increased as the superficial gas velocity increased. As the superficial liquid velocity increased, the magnitude of both the mean and standard deviation of the voltage output increased.   Cheng et al. [95, 96] investigated electrostatic phenomena of sand particles in a combined triple-bed circulating fluidized bed, which was composed of a riser, a downer, a bubbling fluidized bed (BFB), a multi-tube solid distributor for the downer, a gas-solid separator and a gas-sealing bed, used to increase solids transfer from the BFB to the riser. The induced currents were measured by insulated copper rings outside the walls in the upper region of the riser, lower region of the downer and gas-sealing bed. With increasing superficial velocity in the gas-sealing bed, the average induced currents first increased and then approached a 19 constant, consistent with the variation of solid flux and average solids holdup. With increasing superficial gas velocity in the riser, the average induced current in the riser increased. On the other hand, the average induced current decreased in the downer with increasing downward superficial air velocity.   Bi et al. [93] asserted that the standard deviations of current signals from a collision ball probe reflect changes in local hydrodynamics, as well as in particle charge density. Normalized by average values, the dimensionless standard deviation could be a good indicator of local hydrodynamics.  Application of higher-order moments, such as skewness and kurtosis of electrostatic signals has not been reported in the literature.  1.8.2.2 Time-frequency analysis Time-frequency analysis is typically carried out with the aid of the Fourier transform [64]. Yao et al. [16] plotted the power spectrum and power density distribution of signals from a collision ball probe and differential pressure fluctuations for passage of a single bubble. They observed a high degree of similarity between the two signals, confirming that the signals from the ball probe were influenced by the local hydrodynamics. Liu et al. [94] compared the power spectral distribution of charge signals and pressure drop across the bed. It was found that a probe 0.54 m above the distributor experienced a transition of characteristic frequency from 0 to ~1 Hz, similar to pressure drop fluctuations. The characteristic frequency of the charges from this probe varied in a manner consistent with that of the pressure drop when the superficial gas velocity changed. Cheng et al. [95] analyzed the power spectra of induced current, showing that some frequency peaks from the riser were almost independent of the superficial gas velocity, indicating that theses peaks might be caused by electrostatics and/or noise in the signal and not affected by the flow structure. The dominant frequency in the downer was fixed at 0.017 Hz, which may be because of the low frequency variation of solids mass flux and solids holdup, as well as the co-current flow pattern in the downer.   20 1.8.2.3 Non-stationary analysis Many systems are chaotic and time series signals can be analyzed in phase space, with the long-term evolution of the system forming a topological structure called an attractor [97]. The attractor of a system answers the question “where is the system if one waits long enough”; it is a fingerprint of the system, reflecting its hydrodynamic state [98]. Demirbas et al. [99] analyzed the charge signals from an induction probe in a 2D fluidized bed by standard deviation and attractor comparison methods. The standard deviation demonstrates the degree of local charge separation, represented by the difference between the minimum and maximum of the data points (as measured by an electrometer) at a specific location. The attractor comparison method is based on the idea that the state of a fluidized bed at a given time can be determined by projecting all variables governing the system in a multidimensional space. The attractor comparison method statistically compares the attractors re-constructed from time series (for a selected reference period and evaluation period) and calculates a value, the S-value, defined as:  CQS=V (Q)  (1.1) where Q is the estimator of Q, Q is the squared distance between two attractors, and VC is variance. For attractors generated by the same mechanism, S has an expectation of zero and a standard deviation of one. An S-value > 3 indicates with more than 95% confidence that two attractors were generated by different mechanisms [100]. The authors compared the S-values of pressure and charge fluctuations while varying the relative humidity and found that the S-values of pressure and charge fluctuations were insensitive to RH in the first 150 min, and began to differ after the RH level deceased to less than 27%. There is no report on characterizing electrostatic signals in terms of other non-linear characteristics such as K entropy, Lyapunov exponent and correlation dimension.  Yu et al. [101] analyzed electrostatic potential signals from fluidized beds by Hurst analysis of wavelet decomposed signals. The micro-scale signal had only one Hurst exponent < 0.5, whereas the meso-scale signals had one Hurst exponent > 0.5 and another < 0.5, and the macro-scale signals had only one Hurst exponent > 0.5. These three scales are believed to 21 reflect the action of particles, bubbles and macroscopic dynamics. For bubbling fluidization, the energy percentage of the meso-scale component is the highest, confirming that the electrostatic signals mainly reflect the motion of bubbles. The energy percentages of the three scales were sensitive to transitions in flow patterns.   1.8.2.4 Signal reconstruction  Electrical capacitance tomography (ECT) imaging is based on reconstruction of permittivity distribution from the measured capacitance. Electrostatic tomography (EST) involves static charge imaging of charged particles [102, 103]. It has been found that strong electrostatics can affect the ECT and could result in failure of data acquisition [104]. Zhou et al. [105] proposed a dual-mode sensor technique which uses the permittivity distribution obtained from ECT to provide a priori knowledge for the inversion of EST.  In order to determine the charge distribution around a single rising bubble in a two-dimensional fluidized bed, Chen et al. [23] reconstructed the charge signals measured by four induction probes with the assumption that the bubble is symmetrical and charge around the bubble remains the same as it rises. The emulsion phase far from the bubble was found to be charged negatively, and charge density decreased toward the bubble-dense phase interface, nearly reaching zero inside the bubble.  Chen et al. [11] tried to deduce particle charge density from charge signals recorded by a ball probe during single bubble injection experiments. Bubble size was estimated from the injected bubble volume, and charge density was estimated by inserting bubble diameter into an equation which represents the induced charge received by the probe when the injected bubble passes.   These options, from statistical and time-frequency to non-stationary and signal reconstruction, again show that the signals are influenced by local hydrodynamics. The signals contain useful information on charge levels in fluidized beds coupled with hydrodynamics. However, these methods were not able to derive the particle charge density and hydrodynamic property from the cumulative charge signals. Statistical and time-22 frequency analyses are simple, but they cannot decouple particle charge density from probe signals. Non-stationary analysis and signal reconstruction methods may need complicated signal processing procedures and additional instrumentation. Hence, new model-based signal analysis methods are needed to extract the particle charge density and bubble properties. 23 Chapter 2 Probes Design, Fabrication and Calibration 2.1 Introduction An electrostatic probe with two tips/sensors is proposed to decouple particle charge density and hydrodynamics. Two dual-tip probes were designed and constructed: one uses signals from two detectors of different materials with significantly different work functions to extract the charge density on particles and the bubble rise velocity by solving two equations, one for each probe. The other consists of two vertically-aligned tips of the same material. In this case, bubble properties were first obtained by analyzing the signals, and then used in estimating the particle charge density.   2.2 Design and fabrication 2.2.1 Design 2.2.1.1 Dual-tip (two-material) probe A dual-tip (two-material) probe was designed and fabricated for the measurement of charge densities in bubbling fluidized beds. Unlike the two short pipes proposed by Matsusaka et al. [60], which are not suitable for local and online measurements of charge density in fluidized beds, the probes which are often used in fluidized beds can supply local and real-time information.   A probe containing two materials of significantly different work functions was developed. Titanium nitride (TiN) and nickel (Ni) were selected as the probe materials for their large difference in work functions (TiN: 2.9 eV and Ni: 5.0-5.4 eV, respectively, as shown in Table 2.2), relatively high hardness, low cost and widespread availability.  24 Teflon rod, 19 mm ODSurface tips6 mm by 6 mmCoaxial cable, 3.2 mm ODTo charge amplifier To charge amplifier Conductive copper tape as shield 1mm thick Ni23 m thick TiN film coated on 1mm thick stainless steel (a) To charge amplifier To hargeamplifier Coaxial Cables, 3.2 mm ODConductive copper tape as shieldTeflon rod, 19 mm ODSurface tips, 6 mm by 6 mm178 mm8.3 mm15.24 mm (b) Figure 2.1 Schematic of dual-tip (two-material) probes: (a) design a; (b) design b.  Figure 2.1(a) illustrates one schematic design of the probe: In this case, the probe tip contains two metal pieces, each 6×6 mm in cross-section and 1 mm thick, attached to coaxial cables. If the probe tip size is too large, it may disturb local flow and be insensitive to the particle and bubble dynamics; if it is too small, the signals it registers from surrounding charged particles may be too weak compared to background noise. The probe tip size was selected based on initial tests which showed a reasonable response from the electrometer used in this study. A Teflon rod of 19 mm outer diameter prevented charge leakage and maintained a 25 high resistance to the ground. A copper tape was rounded outside the Teflon rod as electrical shield and grounded during measurements to reduce background electrical noise from the column walls.   Results presented below in Section 2.5.1.5 for tests when continuous streams of particles were directed onto the surface of the probes showed that the largest differences between currents transferred to/from the two probe materials occurred when the charged particles struck the probe at right angles. Therefore, a second probe was designed to allow charged particles to directly strike downward-facing metal tips, as shown in Figure 2.1(b).   2.2.1.2 Dual-tip (one-material) probe Figure 2.2 shows a schematic of a dual-tip (one-material) probe with different configurations. Each tip is a square nickel piece (for consistence with the dual-tip (two-material) probe), 6×6 mm in cross-section and 1 mm thick, connected to a coaxial cable. The probes are suspended vertically in the bed to allow charged particles to directly strike their downward-facing metal surfaces. The probe was fabricated so that each tip can be retracted, with the results that the probe can be configured into a collision-collision probe with both tips protruding, “F” configuration, or a collision-induction probe, with only one tip exposed, “Γ” and “L” configurations. Also as shown in Figure 2.2, two types of probe tips were tested, one made of nickel, a 6×6 mm flat surface of thickness 1 mm; the other was made of stainless steel with a rod shape, 4 mm in diameter and 5 mm long.  This probe was designed with the two tips aligned one above the other; other possible configurations include side-by-side tips, one inner cylinder tip and the other outer annulus tip, and even probes with more than two tips. Those probes could be configured as collision probes or collision-induction probes. These probes can also be changed into induction-induction probes with both tips hidden. Because the calibration equation was obtained from a collision probe, the induction probe case was not tested in this work. Also the induction probe is unsuitable to provide local information on non-homogeneous flow systems (see Section 1.6.2.2).  26                Figure 2.2 Schematic of dual-tip (one-material) probe with different configuration and tip shape: F configuration (two tips protruding); Γ configuration (upper tip protruding, lower tip retracted); L configuration (upper tip retracted, lower tip protruding); flat and rod shape tips.   11 mm 5 mmTwo rod tips: 4 mm OD Upper tip, 6 mm by 6 mm square pieceLower tip, 6 mm by 6 mm square pieceCoaxial cables, 3.15 mm ODTo charge amplifiersMetal shielding, 19 mm OD11 mm between two tips15 mm 100 mm 15 mm Retractable sensing tips (flat) 27 The probes used in this work are summarized in Table 2.1.  Table 2.1 Summary of tested probes in this work. Designation Description Diagram 2t2m  Dual-tip (two-material) probe   2t2ma (design a) Figure 2.1(a)  2t2mb (design b) Figure 2.1(b) 2t1m  Dual-tip (one-material) probe   2t1mfF (flat tips, F configuration) Figure 2.2  2t1mfΓ (flat tips, Γ configuration) Figure 2.2  2t1mfL (flat tips, L configuration) Figure 2.2  2t1mrΓ (rod tips, Γ configuration) Figure 2.2  2t1mfrΓ (flat & rod tips, Γ configuration) Figure 2.2 1t1m  One-tip (one-material) probe (conventional probe) Similar to  Figure 1.6(b)* *See Moughrabiah et al. [42] for more precise diagram. Both the 2t2m and 2t1m probes use two periods of signals provided by two sensing tips to decouple the signal. These two types of dual-tip probes are compared in Section 4.4.5.  2.2.2 Fabrication The probes were fabricated by the following steps:  1. Probe head and tips were drawn in Solidworks software and printed by a 3D printer (Objet30) using a proprietary material (VeroWhite). The printed parts were non-conductive and maintained high resistance to the ground, while a shield reduced background electrical noise from charges buildup on the column walls. 2. Metal sheets (25×25 mm in cross-section) were cut, filed into small pieces (6×6 mm in cross-section), and glued to coaxial wires by conductive epoxy. Inappropriate gluing may cause signal drift.  3. The coaxial wires were connected to Female BNC connectors by soldering. 4. The electrical insulation and performance of the probe were tested. The metal mesh of the two coaxial cables should not contact the tip. This prevents charge leakage and eliminates interference between the two current signals. In addition, the electrical wires should be as short as possible to minimize noise. Initial Tests showed that the charge measured by the probe with no contact with particles was close to the baseline (e.g. 0.001 V), two orders of magnitude less than when the probe was struck by charged particles (e.g. 0.1 V). 28 Resolving signal drift from a fabricated probe is illustrated by a case study in Appendix B.5. Photographs of the fabricated probes appear in Appendix A.1.   2.3 Calibration of probes outside fluidized bed The 2t2m probe is intended for use in a fluidized bed where complex two-phase flow, including bubbles and a dense phase are present. Thus the probe was tested and calibrated for both dilute phase and dense phase gas-solid flows in an ejector-funnel setup, which is able to generate charge densities on the particles with different magnitude and polarity. The probe was also tested by dropping charged particulate flows generated from a vertical tube and vibration tray setup, to make sure that there was a difference in the charge/current signals from the probe. The origin of such difference in the signals from the probe was also investigated in a motor-pulley setup, where the probe only experienced induced charge. Details are provided in the following paragraphs.  2.3.1 Ejector-funnel setup A 2t2ma probe (see Table 2.1) was calibrated by using an ejector-funnel setup shown in Figure 2.3a and Appendix A.2. For dilute phase experiments, the Plexiglas funnel was removed. Glass beads were fed through a glass funnel and evenly dispersed into an ejector (RAV375H, AIR-VAC), then passed through a 90 elbow-type fitting and a straight vertical pipe. Four combinations of polyvinyl chloride (PVC, 25 mm long and 24 mm ID), stainless steel (SS, 22 mm long and 26 mm inner diameter) and aluminum (Al, 23 mm long and 25 mm inner diameter) were selected for the 90 elbow and pipe, as shown in Figure 2.3b. For dense phase experiments, the Plexiglas funnel collected the charged particles from the ejector, and then dropped them onto the probe surface.   In this case, glass beads belonging to Geldart Group B were used with a volume weighted mean diameter of 624 m and size distribution of 360-830 m, determined by a Malvern Mastersizer 2000. According to Cross [30], electron energies in an insulator are a function of 29 electron position in energy distribution, surface impurities and local atomic structure, as well as the chemical properties of the material. Therefore, the work function of an insulator must be determined experimentally. The work function, dielectric constant and resistivity of silicon dioxide (the major component of glass) and different types of glass are provided in Table 2.2. Before each test, the glass beads were washed with ethanol and water, then dried overnight to eliminate impurities and dust. All these tests were conducted at room temperature of 19~23C; with environmental relative humidity of 27~52%. ElectrometerElectrostatic collision probeFaraday cupLoad particlesEjectorData acquisition systemComputerAir flow meterC p ess airPlexiglass funnelPart b140 mm 270 mm 90 mm Electrometer a. overall setup PVCPVCPVCSSSSSSAluminumAluminumPVC+PVC PVC+SS SS+SSradius: ~ 60 mm  b. pipe combinations Figure 2.3 Schematic of experimental apparatus for ejector setup: (a) overall setup; (b) elbow and pipe combinations. (PVC: polyvinyl chloride; SS: stainless steel) (not to scale) 30 Table 2.2 Electrical properties and hardness of different materials used in the experiments [74, 106-115]. Materials  Property Glass Titanium nitride (TiN) Nickel (Ni) Stainless steel (SS) Polyvinyl chloride (PVC) Aluminum (Al) Work function, eV silicon dioxide: 5.0 2.9  5.04-5.35 [107] 4.96-5.03 [108] 4.4 4.85 [109] 5.13 [110] 4.06~4.26 [107] 3.38~4.08[108] Hardness, kg/mm2 560 2300 1340 N/A N/A N/A Dielectric constant silicon dioxide: 3.78  iron-sealing glass: 8.41  N/A N/A N/A 3.2  N/A Conductivity, S/m or Volume resistivity, .cm iron-sealing glass :1E10 .cm  soda-borosilicate: 7E7 .cm  silicon dioxide: >1E19 .cm 3~7E7 S/m 1.43E7 S/m [107] 6.9E-6 .cm [108] 1.45E6 S/m [111] 90E-6 .cm [108] 1E14 .cm  3.5E7 S/m [112] 2.62E-6 .cm [108]  31 2.3.1.1 Charge density The total net charge, q (C), on particles after they pass through the ejector-funnel setup, was measured by an electrically insulated Faraday cup, connected to an electrometer (Keithley Model 6514). This Faraday cup was composed of an inner copper vessel of 150 mm inner diameter and an outer copper cup of 200 mm inner diameter, insulated from each other by Teflon blocks, with the outer cup grounded. The charge density, qm (C/kg), of particles was obtained by dividing the net charge by the mass of particles collected by the Faraday cup, =mpqq m  (2.1)  2.3.1.2 Mass flow rate  The particle mass flow rates onto the probe tip surface were measured by a sampling probe, with dimensions identical to that of the probe, but without metal plates, so that particles enter the rectangular openings, are collected and then weighed. This allows tip, the mass fraction of particles striking the probe, to be calculated as ,,= p collectedtipp droppedmm (2.2)   2.3.1.3 Transferred current  The probe was placed 50 mm below the pipe orifice/Plexiglas funnel and aligned with the pipe/funnel. The charge signals from the two metal tips were amplified by two electrometers (Kistler model 5010B) and logged into a computer by a data acquisition card (PCIe-6321, National Instrument) and Labview software, with a sampling frequency of 100 Hz. For each run, the current transferred from particles to the probe was calculated by least-square linear fitting of the cumulative charge vs time curve, with the transferred current corresponding to the slope.  32 2.3.1.4 Particle velocity Figure 2.4 shows the technique used to measure the average particle velocity (Vp) in dilute phase flow. A high-speed video camera (MS70K, Mega Speed), operating at 4000 frames per second, obtained images of particles striking the probe. The particle velocity was then calculated by dividing distances travelled by the time interval. Twenty particles were analyzed in each case. The particle velocity in dense phase flow was varied by changing the vertical distance between the funnel orifice and the probe surface, since the particle velocities decrease with decreasing acceleration distance [116]. The angle of collision (θ) (see Figure 2.4) was also varied. High- pe dvideo c meraθθUnpUtp5 cm Figure 2.4 Schematic of technique for measuring average particle velocity at different angles of inclination of probe.  2.3.2 Motor-pulley setup A motor-pulley system was built to check the induced charges on the 2t2m probe, as shown in Figure 2.5 and Appendix A.3. A DC compact gear motor (12 V, max 50 rpm) drives a plastic ball (0.38 m diameter) vertically at a constant velocity by rotating a pulley and a string. The ball can be moved upward or downward by changing the rotational direction of the motor. In each experiment, the plastic ball was charged negatively by rubbing with hair before being moved upward, then downward at 0.1 m/s. A 2t2mb probe (see Table 2.1) was placed at the middle of the total height in such a way that it would not hinder the ball’s movement (no contact with the ball), with different positions as shown in Figure 2.5, to avoid differences in the relative horizontal positions of the probe and two tips. Because the ball is 33 much larger than the probe tip, induced charges received by the probe may be affected if the charged ball passes the probe asymmetrically. This can be verified by orienting the probe both upward and downward if there is any asymmetry. MotorPulleyCharged ballElectrostatic probe0.65 mVertically downwardVertically upwardHorizontally downwardHorizontally upwardElectrometerProbe tip: Ni Probe tip: TiN Figure 2.5 Experimental setup of motor-pulley system and different positions of 2t2mb probe tip (not to scale).  2.3.3 Vertical tube setup and vibration tray setup Figure 2.6 shows a schematic of the vertical tube setup. Glass beads were loaded into a funnel with a flow control valve. Then a steady continuous flow of particles was allowed to drop through a vertical Plexiglas tube (0.05 m inner diameter) and collide with the probe. Particles gained charge through colliding with the tube inner surface. A 2t2ma probe (see Table 2.1) was used to measure the charges on the particles. The probe was placed at two locations in order to change the contact angle between the probe tip and the particulate flow: at the bottom of the tube with the probe tip surface facing upward and at the side of the tube with the probe tip surface vertical. The charge density on the particles was measured by a Faraday cup located at the bottom of the tube.  34 ElectrometerElectrostatic probeFaraday cupLoad particlesGas-solid dispersed flow pipeData acquisition systemComputerFlow control funnel Electrostatic probeElectrometer Figure 2.6 Schematic of experimental apparatus for vertical tube setup (not to scale).  Figure 2.7 shows a schematic of the vibration tray. Glass beads were continuously loaded on to a vibrated tray (ERIEZ 15A) and gain charges via contacts between the particles and metal tray. After the charged particles have collided with the probe, their charge densities were measured by the Faraday cup.  ElectrometerElectrostatic collision probeFaraday cupLoad particlesVibration feederData acquisition systemComputerElectrometer Figure 2.7 Schematic of experimental apparatus for vibration tray setup (not to scale). 35 2.4 Calibration of probes inside fluidized bed Figure 2.8 shows a schematic of the two-dimensional Plexiglas fluidization column with inside dimensions 0.307 m wide, 22 mm thick and 1.24 m high (a picture is shown in Appendix A.4). The distributor contained seven evenly-spaced holes of 3 mm diameter. Three ports were installed on the back wall of the column to accommodate the electrostatic probe, in addition to a sampling port on the front wall and a bubble injection port. The sampling port (6.35 mm in diameter, inclined downward at 30 to the vertical) and the electrostatic probe were located on opposite faces of the column at the same height and horizontal position, in order to provide corresponding localized measurements.  Faraday cupAir in Air inWindboxElectrostatic probea) Front view b) Side view0.17 m1.24 m0.307 m0.022 mSampling port0.012 mPAir inSolenoid valveBubble injectorBubble injectionElectrostatic probe Figure 2.8 Schematic of two-dimensional column, bubble injection system and charge measurement system (not to scale).  Extra dry air was supplied to the bubble injector, where both compressed air and nitrogen, controlled by separate flow meters, were used as the fluidizing gas for the main column. The particles were glass beads, sieved to a size range of 500-600 m, washed with water and ethanol, and dried in an oven overnight to eliminate possible impurities, dust and moisture. The bed was fluidized at different superficial gas velocities, with static bed heights of 0.30 36 and 0.50 m, and with the probe 0.22 and 0.42 m above the distributor, respectively. The average volumetric fraction of glass beads in the static bed was measured to be 0.63. All experiments were conducted at room temperature, 19~23C. The relative humidity and temperature of both the gases and the environment were monitored by a humidity and temperature indicator (VAISALA HM141).  Charges on the particles inside the column were measured by both a Faraday cup (direct method) and a fabricated collision probe (indirect method). The Faraday cup results provided in-bed particle charge densities, which were used for probe calibration. Particles emerging from the sampling port passed through a sampling tube and dropped into the Faraday cup. The total lengths of the sampling port and the tube were designed to be as short as possible (~0.15 m). At the beginning of each sampling process, a plug at the end of the tube was removed and particles resting inside the sampling port and tube, as well as a small volume of bed particles, were first discharged and discarded until charge equilibrium was established in the sampling system. Particles from the bed were then continuously dropped into the Faraday cup, after which the plug was replaced to stop the sampling.   For bubble injection experiments, gas was injected via a solenoid valve into the column from a stainless steel cylinder of volume 1000 ml. The size of the injected bubbles was varied by adjusting the pressure in the cylinder and the opening time of the solenoid valve by means of a Labview program. A high-speed video camera (MS70K, Mega Speed), operating at 500 frames per second, recorded the bubble movement during single bubble injection and freely bubbling experiments. The electrostatic signal transmitted by the probe was synchronized with the frames recorded by the camera. Details of synchronization are illustrated in Appendix B.8. The probe was also sampled at 500 Hz in synchronization experiments. Based on the results in Sections B.10.1 and 5.3.1, for the dual-tip (two-material) and conventional probes, a sampling frequency of at least 100 Hz was required; and for the dual-tip (one-material) probe, a sampling frequency of 500 Hz was chosen. A sampling time of 5-10 min, which is higher than 2 minutes required for properly capturing the dynamic behaviour of the fluidized bed, was chosen for all the tested probes. Bubble rise velocities were determined from recorded video images using Image J software. 37  2.5 Results and discussion 2.5.1 Ejector-funnel setup 2.5.1.1 Electrostatic charging of particles in ejector-funnel setup In the ejector-funnel setup, particles were charged by particle-particle and particle-wall collisions. The effects of gas velocity and pipe materials on the charge density of particles were examined.  2.5.1.2 Effect of pipe materials Different elbow fittings and pipe materials (see Figure 2.3b) were utilized to change the charge density on the particles at the same mass flow rate or average particle velocity. Figure 2.9 shows that the particles gained positive charges when they interacted with PVC+PVC and PVC+SS pipes, and negative charges upon striking the SS+SS and Al+Al pipes. The polarity of charges on the particles depends on the work functions of the particles and contact materials. Of the four cases, PVC+PVC gave the highest positive cumulative charge on particles, while Al+Al provided the greatest negative cumulative charge. The 90 elbow pipe fitting mainly determined the final particle charge magnitude and polarity, because most charges were generated by collisions with the elbow. The cumulative charge curves in Figure 2.9 show an almost linear relationship between transferred charge and time, indicating uniform injection and flow of particles into the pipe.  In order to obtain reproducible results, all elbow fittings and pipes were grounded after each run, and the pipe was flushed with air to remove fine dust and dissipate charges on the pipe. For each run, the initial charges were maintained in the range of 0~2nC for the particles and < 1 nC for the pipe to minimize the possible influence of initial charges on the particles and the pipe wall. This approach was effective, as reflected by the almost constant slopes of the cumulative charge curves over the entire dropping periods of ~50 s in each case, as shown in Figure 2.9.  38  Figure 2.9 Charge carried by particles after passing through ejector setup shown in Figure 2.3b with different pipe materials; Average gas velocity =2.79 m/s (at outlet of vertical pipe with average pipe inner diameter is 25 mm).  2.5.1.3 Effect of gas velocity Figure 2.10 shows that the cumulative charges increased with increasing gas velocity for the test with PVC+PVC. Similar trends were observed for PVC+SS, SS+SS and Al+Al pipe wall materials combinations. The abrupt change in slope for a gas velocity of 3.26 m/s was caused by an adjustment of the gas flow meter (increase in valve opening to required flow rate) in the middle of the test.  The above results indicate that the particle charge density could be altered by varying the gas velocity or by changing the pipe wall material. Watanabe et al. [117] developed an impact charging test rig for single particles acquiring charge during collision with a metal piece. The relationship between impact charge q and initial charge qi was represented by a linear equation:  0 1- ieqq q q       (2.3) where the characteristic charge q0 is the transferred charge for particles with zero initial charge, and increases with increasing impact/collision velocity. The equilibrium charge qe is 39 the charge needed to overcome the work function difference of two objects when there is no net charge transferred. It is independent of impact/collision velocity, but depends on the material properties. Our results are consistent with Eq.(2.3).  Figure 2.10 Charge carried by particles after passing through ejector setup with different gas velocities for PVC+PVC elbow and pipe combination (see Figure 2.3b).  2.5.1.4 Characteristics of particle charging with the probe Figure 2.11 shows the charge density on particles in the dilute phase due to interactions with the four different pipe material combinations at different particle velocities, determined by the Faraday cup. Charge densities on the particles were reproducible in each case. Each reported value is the average of three measurements, with the error bars representing   one standard deviation from these values. As shown in Figure 2.10, the charge density typically increased with increasing gas velocity due to frictional charging. However, it did not change very much for SS+SS, compared with the other three wall material combinations. This discrepancy probably arose from differences in work functions and surface conditions (such as moisture content and roughness). 40 PVC+PVC PVC+SS SS+SS Al+Al-12-8-40481216Charge density, C/kg VP=4.42 m/s VP=5.67 m/s VP=6.39 m/s Figure 2.11 Measured charge densities on particles in dilute phase with different pipe materials and particle velocities.  The electrostatic charge signals measured by the probe are plotted in Figure 2.12, similar to the cumulative charge curve on particles measured by the Faraday cup. The charge, measured by the probe when charged particles contacted the probe surface, increased as particles continued to strike the probe surface over the dropping test period. The difference between the final and initial charges was the total charge transferred over the dropping test period, whereas the slope is a measure of the current transferred from the probe to the electrometer, regarded as the current transferred from the charged particles to the probe. Figure 2.12 shows that the transferred current varied with different charge densities on the particles. Similar results were found for other gas velocities. Figure 2.12 also shows that the transferred current measured by the Ni tip was larger than that measured by the TiN tip for the experimental conditions investigated. Eq. (2.3) predicts that the charge transferred during contact (q) depends not only on the equilibrium charge (qe, which is related to the work function difference between particles and metal surface), but also on the initial charge on particles (qi) and the characteristic charge (q0, which is related to particle contact velocity, effective contact area, surface condition and other material properties). The equilibrium charge transfer (qe) is expected to be higher for the TiN tip than the Ni tip based on the work functions listed in Table 2.2 because the work function difference between TiN and glass is larger than that between Ni and glass, if one assumes that the work function of the glass beads used in the experiment is the same as that of pure SiO2. However, the magnitude of the 41 characteristic charge (q0) is strongly influenced by the collision velocity between particles and the metal surface. As a result, the relative magnitude of transferred currents between the metal and the colliding particles for two different metals may vary (as shown in Figure 2.12) and even cross over when the particle velocity or flow rate changes [60].  The electric current generated from continuous collisions is given by stranpWI qm   (2.4) This current is mainly transferred current, with the negative sign accounting for its direction.   Figure 2.12 Typical cumulative charge signals measured by 2t2ma probe in dilute phase flow with different pipe materials; average particle velocity Vp=4.42m/s; collision angle θ=0 degree.  42 -10 -5 0 5 10 150246810121416 Ni, Vp=6.39 m/s TiN, Vp=6.39 m/s Ni, Vp=4.42 m/s TiN, Vp=4.42 m/s linear fitItran / Ws, C/kgCharge density, C/kg   Figure 2.13 Characteristics of Ni and TiN tips at different particle velocities in dilute phase flow, collision angle θ=0 degree; four pipe combinations as shown in Figure 2.14b.  Figure 2.13 plots the ratios of transferred current to mass flow rate at the probe tip against charge density for both probe tips. Error bars represent mean  one standard deviation of at least three repeated measurements. Linear relationships between Itran/Ws and qm for the two materials were obtained by least-squares fitting at a constant particle velocity. However, this linear equation varies with particle collision velocity: the higher the average particle velocity, the larger the current transferred to the probe. Therefore, the effect of particle velocity should be taken into consideration when correlating the transferred current, as in earlier correlations of John et al. [118] and Zhu and Soo [58]. The transferred current correlation proposed by Matsusaka et al. [60] does not include the particle velocity, whereas a similar equation developed by industrial vendors [90, 119] suggested a proportionality to the square of particle velocity. The following form of equation was able to correlate the transferred current from charged particles to probe tip with the particle charge density and collision velocity:  , 2tran ii m i p isI a q bV cW    (2.5)  Least-squares regression of all experimental data was conducted for the two materials exposed to dilute and dense phase flow, respectively. For dense phase flow (0.87   0.72), fitting resulted in 43 -6 2 2 -71 1 15.44  -1.32 ( / )  -3.67 ( / )a b E C s kg m c E C kg    , , (Ni) (2.6) -7 2 2 -82 2 22.99  -7.79 ( / )  -6.13 ( / )a b E C s kg m c E C kg    , , (TiN) (2.7) For dilute phase flow( > 0.99), -8 2 2 -61 1 10.481,  8.84 ( / )  2.80 ( / )a b E C s kg m c E C kg    , (Ni) (2.8) -8 2 2 -62 2 20.346,  5.49 ( / )  1.53 ( / )a b E C s kg m c E C kg    , (TiN) (2.9) Voidage within other range of values were not tested in this study. Obviously fitted constants ai, bi and ci are related to the voidage, as shown in Eq. (C.28) of Appendix C.2, ai are smaller as  increases. bi are more important for dense phase (Eqs. (2.6) and (2.7)) than dilute phase (Eqs. (2.8) and (2.9)). It is better that the probe could be calibrated in gas-solid flows with different voidage. The total net transferred charges include both triboelectrification and transfer of pre-charges. Therefore, the ai, bi and ci constants are also functions of the properties of the probe materials and particles such as the dielectric constant of particles, work function difference between the probe tip material and the bed particles, probe tip size and particle size and shape (Eq. (C.28) in Appendix C.2). Changes in temperature and pressure are assumed not to significantly affect physical properties other than the work function of the probe tip and particles, which may also be affected by surface and environmental conditions. The experimental and calculated values from Eq. (2.5) with the above a, b and c values are compared in a parity plot in Figure 2.15. Coefficient of determination R2= 0.98 and 0.96 were found for Ni and TiN in the dilute phase flow, whereas R2 was 0.94 and 0.95 for dense phase flow. Based on Eq. (2.5), the transferred current is a function of three variables: particle charge density (qm), solid flow rate (Ws) and particle contact velocity (Vp). Ws is related to Vp by (1 )s p p PW V A    (2.10)  Since the transferred current is now a function of qm and Ws, with the two current signals from the probe, the charge density in the fluidized bed can be estimated by solving Eqs. (2.5) and (2.10) together.  44  Figure 2.15 Comparison of fitted and experimental data for (a) dilute phase flow ( > 0.99), and (b) dense phase flow (0.87   0.72).  2.5.1.5 Effect of collision angle on current transfer from particles to probe The effect of collision angle was investigated by inclining the probe surface to the solids flow, as shown in Figure 2.3b. Figure 2.16 shows the relationship between –Itran/Ws and collision angle θ for the two materials at different particle velocities in dilute phase for the Al+Al pipe material. Three runs were performed for each condition, with error bars again representing   one standard deviation from average values (three samples). For both materials, a higher normal component of collision velocity resulted in larger current transferred, consistent with the results in Section 2.5.1.3. –Itran/Ws for both materials first increased (from negative to positive), then decreased, as the collision angle increased from 0 to 90, with a maximum at θ’= 60. Changing the collision angle alters the effective or projected contact area between the probe tip and charged particles. According to a rolling-slipping model [120], charged particles start to roll on the probe tip as they collide with the probe tip. For the contact angle smaller than a critical value, θ  θ’, increasing collision angle increases the effective contact area because of an increase in rotation of particles on the probe tip surface; for θ > θ’, the slipping (or sliding) effect increases gradually with angle, and the rolling speed of particles decreases, leading to a decrease in effective contact area. Figure 2.16 also shows that the polarity of the transferred current changes from positive to negative as the collision angle increases.  45 0 15 30 45 60 75 90-3-2-101234Angle, deg-Itran/ Ws, C/kgVp=4.42 m/sVp=5.67 m/sVp=6.39 m/s(a) Ni  0 15 30 45 60 75 90-10123 Vp=4.42 m/s Vp=5.67 m/s Vp=6.39 m/s(b) TiN-Itran / Ws, C/kg  Angle, deg Figure 2.16 Relationship between –Itran/Ws and collision angle θ at different particle velocities for two tips in dilute phase flow: (a) Ni and (b) TiN.  Tanoue et al. [121] also found that the polarity of transferred current changes from positive to negative as impact angle increases when negatively charged glass beads continously struck a rotating aluminum target for 10 min. They suggested that a contaminated layer of glass beads on the metal surface changed the effective contact area, leading to the polarity changes through a reversal of effective contact potential difference. John et al. [118] observed that total transferred charge is influenced by the polarity of charges on the particles when the charged aerosol particles collide with the metal surface. They explained this effect by assuming a p-n junction [122] at the particle-probe contact with both the probe surface and particles treated as semiconductor materials. For positive triboelectric charge, the particle surface material is p-type, and the probe surface material is n-type. When neutral particles contact the probe, electrons flow from the probe to the particles and positive charges are transferred from the particles to the probe until the contact potential is equilibrated. If the particles carry a negative charge, this produces a negative bias voltage across the junction, impeding charge transfer, and the effects are reversed.   In our case, the glass beads were pre-charged negatively by the ejector-funnel setup (Al+Al), and the triboelectric charge to be generated from particle-wall collision was expected to be positive based on Figures 2.11 and 2.12. The total net transferred charges include both triboelectrification and transfer of pre-charges. When the collision angle increased, the 46 normal component of particle velocity decreased, leading to less triboelectrification (positive charge decreases). At the same time, the increase of collision angle extended the contact time during which the more charges were transferred from charged particles by electrical conduction (negative charge increases). So the net transferred charge changed from positive to negative as the collision angle increased. The sign and magnitude of the measured total transferred current are determined by the combined effects of reduced triboelectrification from normal collision and increased transfer of charges from charged particles to the probe tips.   Figure 2.17 shows the difference between transferred currents for the two probe tip materials with changing collision angle θ at constant particle velocity. The maximum difference was found at θ=0, and decreased with increasing collision angle. Increasing the collision angle changed both the effective contact area and the normal component of collision velocity. When the collision angle increased, the transfer of charges on particles increased due to the increased contact time. Since the surface conductivities of the two metal tips were very similar (see Table 2.2), the charges transferred from particles to the two materials were similar. Meanwhile, as the normal component of particle velocity decreased, the triboelectrification, which is a strong function of the work functions of the two contact materials, was reduced, leading to a gradual decrease in the difference of total transferred currents from particles to the two probe materials. 0 15 30 45 60 75 90-2.5-2.0-1.5-1.0-0.50.00.51.01.5 -Itran / Ws, C/kg Ni TiNAngle, deg   Figure 2.17 Relationship between –Itran/Ws and collision angle θ for two tips at Vp=4.42 m/s in dilute phase flow; pipe material combination: Al+Al. 47  2.5.2 Motor-pulley setup To elucidate the difference in signals received by the two materials of the 2t2m probe when a charged object passes, a series of experiments was conducted in the motor-pulley setup and the two-dimensional fluidized bed to examine the performance of the two tips of different materials. In the motor-pulley calibration system, when the charged ball moved upward toward the probe, induced charges were detected by the probe, as shown in Figure 2.18. When the charged ball approached the probe, the induced charge measured by the probe negatively increased and reached a maximum magnitude; as the ball moved away from the probe, the magnitude of the induced charge decreased and almost returned to its original value. This was repeated four times, resulting in four peaks in the signal trace. The two probe tips of different materials showed almost the same minimum induced charges and signal traces. 0 10 20 30 40 50 60-35-30-25-20-15-10-50510TiNNiCumulative charge, pCTime,s TiN NiCmin Figure 2.18 Cumulative charge signal measured by 2t2mb probe (facing vertically downward) when negatively charged ball passed by in motor-pulley setup.  To further check the difference in charge induction for the two materials, tests were carried out in the two-dimensional fluidized bed. To prevent direct contact with particles, the probe tip was covered by, in order of application, Teflon PTFE tapes (for pipe thread and electrical insulation, respectively), tape and black electrical insulation tape. The particles were next fluidized at different superficial gas velocities to alter the charge levels on the particles. Figure 2.19 shows cumulative charge signals from the probe during single bubble 48 injection and free bubbling. Negligible differences are observed in the traces and amplitudes for signals from the two materials for these two cases. For single bubble injection, the charge signal first decreased due to the negatively charged bed particles which have a relatively higher charge density than was the case in Figure 2.24, and then showed a peak when the bubble passed. In free bubbling, the charge signals from the two tips did not show a substantial increase/decrease within 200 s, indicating that no significant transfer of charges was measured by the probe.  0 2 4 6 8 10-3.0-2.5-2.0-1.5-1.0-0.5Cumulative charge, nCTime, s Ni TiNsingle bubbleqm= -4.14 C/kg(a)NiTiN0 50 100 150 200 250-3-2-101234Cumulative charge, nCTime, s Ni TiNfree bubblingqm= -2.93 C/kg(b) Figure 2.19 Cumulative charge signal measured by covered tips of 2t2mb probe in two-dimensional fluidized bed: (a) single bubble; (b) free bubbling, Ug =0.30 m/s.  The difference between the two tips was quantified by the ratio of their minimum charges, i.e., Cmin,TiN / Cmin, Ni. Figure 2.20 (a) shows this ratio for the probe at different locations relative to the ball in the motor-pulley experiment. The ratio ranged from 0.89 to 1.12, with an average of 0.99. In Figure 2.20 (b), the ratio maintained an average value of 0.92 for various particle charge densities in the two-dimensional fluidized bed, confirming that the ratio of induced charges from the two materials varies very little with changing particle charge density.   From both the motor-pulley experiments and the fluidization experiments, we can conclude that the difference in charge/current signals for the two materials of the 2t2m probe arose mainly from charge transfer during particle-probe collisions. The small difference in induction charges may result from small differences in the probe tip sizes, as well as slight asymmetry of trajectory as charged objects passed the probe. 49  0.00.20.40.60.81.01.21.4Horizontally  upwardHorizontally downward Vertically   upwardCmin,TiN / Cmin,Ni   Vertically downward(a)-9 -8 -7 -6 -5 -4 -3 -2.60.81.01.2Cmin,TiN / Cmin,Ni  Charge density, C/kg(b) Figure 2.20 Ratio (Cmin,TiN / Cmin,Ni) from 2t2mb probe (see Table 2.1) (a) in motor-pulley setup; (b) in two-dimensional fluidized bed.  2.5.3 Vertical tube and vibration tray setups Figure 2.21 shows cumulative charges from the 2t2ma probe (see Table 2.1) at two locations in the vertical tube setup.    Figure 2.21 Cumulative charge signals from 2t2ma probe at different locations in vertical tube setup, shown in Figure 2.6.  Ni TiN Ni TiN 50 For both locations, there were substantial differences between the signals from the two tips. The charge signals were higher in magnitude when the probe was located at the bottom of the tube, with its tip facing upward than when at the side with the tip oriented vertically. Also the relative magnitudes of signals from the two tips changed between these two cases, similar to the previous results in the ejector-funnel setup (Section 2.5.1). The probe was also rotated to change the relative positions of the two tips, in order to investigate whether differences could be caused by non-uniform particulate flow. Both cases showed that substantial differences existed between the signals from the two tips regardless of their relative positions.  The probe was also checked with one tip covered by Teflon tape and placed at the bottom of the tube. Charge signals from the probe with one tip covered appear in Figure 2.22. In both cases, the signals were compared between both tip uncovered and one tip covered. The slopes of signals from the covered tip decreased because of reduced charge transfer. Differences continued to exist in the signals from the two tips.  0 20 40 60 80 100 120 1400.000.050.100.150.20Ni1 and TiN1: qm= -4.57C/kg; Ni2 and TiN2: qm= -5.28C/kg Cover TiN tip Ni2 uncoveredNi1 uncoveredTiN1 uncoveredTiN2 coveredCumulative charge, CTime, s0 20 40 60 80 100 120 1400.000.050.100.150.20Ni1 and TiN1: qm= -2.44C/kg; Ni2 and TiN2: qm= -4.75C/kgCover i tip  TiN2 uncoveredNi1 uncoveredNi2 coveredCumulative charge, CTime, sTiN1 uncovered Figure 2.22 Comparison of cumulative charge signals from 2t2ma probe with both tips uncovered and one tip (Ni or TiN) covered in vertical tube setup (probe at bottom of tube with probe tip facing upward).  Cumulative charge signals from the probe in the vibration setup are plotted in Figure 2.23. When charged particles started to drop onto the probe surface, cumulative charges for both tips increased, reaching constant levels after dropping, with substantial differences between the two signals. The slopes of the charge signals versus time represent average currents transferred from the charged particles during the dropping. This confirms that there was a substantial difference between the signals from the two tips.  51 0 20 40 60 80 1000.000.020.040.060.080.10end of droppingqm= -0.39 C/kgVibration speed at 7Cumulative charge, CTime, s Ni TiNstart droppingNiTiN Figure 2.23 Cumulative charge signal from 2t2ma probe in vibration tray setup.  2.5.4 Two-dimensional fluidized bed The 2t2m probe and 2t1m probe (see Table 2.1) were next deployed and calibrated in a thin “two-dimensional” fluidization column to assist with the interpretation and confirmation of results for single bubble injection experiments.   2.5.4.1 Single bubble injection experiments Figure 2.24 shows typical charge and corresponding current signals measured by the 2t2mb probe (see Table 2.1) during single bubble injection with a background superficial gas velocity of Umf.. Gas was injected via a solenoid valve into the column from a stainless steel cylinder. The time-varying signal from the bubbling bed differs greatly from that determined previously for homogeneous continuous particle flow, which resulted in a nearly constant transfer current. In the fluidized bed, the current signal was strongly affected by local hydrodynamics, with fluctuations caused by passage of bubbles. Also the magnitude of particle charge density was much greater in the fluidized bed.  52 3.0 3.2 3.4 3.6 3.8 4.0 4.2 4.4 4.6 4.8 5.0-3-2-1013.0 3.2 3.4 3.6 3.8 4.0 4.2 4.4 4.6 4.8 5.0-4-202Charge, 10-9 Cnose(b) wake(d)Imax(e)Current, 10-8 ATime, sNiTiNTiNNiImincentre(c)Injection(a)     Figure 2.24 Charge (raw signal), current (derivative of charge) signal from 2t2mb probe and synchronized snapshots for single bubble injection in two-dimensional fluidized bed. (Back pressure for bubble injection: 345 kPa; solenoid valve opening time: 0.2 s; ~2E−4 m3 pressurized air injected; centre of probe location: 0.17 m above bubble injector; bubble injector: 0.04 m above distributor)  The two peak values of current signals, Imax and Imin, received by the electrostatic probe as each bubble passed, are directly related to the charge density of the particles surrounding the bubble, as well as to the bubble size, shape and rise velocity. Comparison of the synchronized video frames with current signals indicates that the first minimum corresponded to the moment of bubble injection, whereas the maximum peak, Imax, was associated with the bubble nose reaching the probe, and the second minimum peak, Imin, corresponded to the arrival of the bubble wake. Imin arises from a combination of transferred 3.578 s (a) 3.648 s (b) 3.806 s (c) 3.930 s (d) 3.990 s (e) 53 and induced currents, and is therefore a function of particle charge density and bubble properties. The difference in magnitudes of the minima corresponding to the two materials is due to different transferred currents. The charge density in the bed was changed by fluidizing particles at different Ug, then reducing Ug to Umf and injecting a single bubble. Particle charge density was measured by discharging bed particles through the sampling port into the Faraday cup, with the results given in Table 2.3. It is seen that the magnitudes of Ipeak (minimum peaks in this case because relative larger magnitude than maximum peaks) increased as the size of injected bubbles increased, while the particle charge density remained more or less constant, consistent with results reported previously [11, 41]. The magnitude of Ipeak also increased as the particle charge density increased for a given bubble size.  Table 2.3 Directly measured charge density (qm), bubble size (DB) and rise velocity (UB) and current peaks (Ipeak) from single bubble injection experiments. qm (C/kg) UB (m/s) DB (m) Ipeak,Ni (A) Ipeak,TiN (A) -8.02E-7 0.40 0.052 -1.29E-08 -1.21E-08 -8.70E-7 0.37 0.06 -1.76E-08 -1.44E-08 -1.04E-6 0.44 0.077 -2.06E-08 -1.76E-08 -9.00E-7 0.41 0.062 -2.27E-08 -1.61E-08 -9.66E-7 0.47 0.088 -2.80E-08 -2.45E-08 -1.77E-6 0.53 0.112 -3.43E-08 -2.94E-08 -1.55E-6 0.48 0.091 -3.60E-08 -3.09E-08 -7.74E-7 0.51 0.101 -3.61E-08 -3.11E-08 -1.05E-6 0.54 0.113 -3.77E-08 -2.95E-08 -1.68E-6 0.58 0.131 -4.42E-08 -3.29E-08 -1.05E-6 0.55 0.119 -4.89E-08 -4.09E-08 -2.14E-6 0.47 0.087 -5.96E-08 -2.95E-08  Figure 2.25 shows current signals measured by 2t1mfF probe (see Table 2.1) following single bubble injection into the bed of glass beads, with a background superficial gas velocity of Umf. Comparison of the synchronized video frames with current signals from both tips shows an initial minimum of no interest corresponding to bubble injection, whereas the maximum peaks, Imax,2 from the lower tip and Imax,1 from the upper tip, are associated with the bubble nose reaching the probe, and the minimum peaks, Imin,2 and Imin,1, correspond to 54 arrivals of the bubble wake. The times corresponding to these peaks, tmax,1 and tmin,1 from the upper tip, tmax,2 and tmin,2 from the lower tip, were in order tmax,2< tmax,1< tmin,2< tmin,1, indicating that the injected bubble passed the two tips in succession.  -1.0-0.50.00.5-1.0-0.50.00.5lower tipCurrent, E-7 Aupper tip1.510.5tmin,2tmax,2tmin,1Current, E-7 ATime, stmax,10 Figure 2.25 Current signals from 2t1mfF probe for single bubble injection in two-dimensional fluidized bed containing glass beads (Pressure in bubble injector: 345 kPa; solenoid valve opening time: 0.2 s; ~2E−4 m3 pressurized air injected; centre of probe port 0.17 m above bubble injection port; bubble injector: 0.04 m above distributor).  Current signals from the 2t1m probe with different configurations in multiple bubble injection experiments are shown in Figure 2.26. Clear time delays appeared between the peaks from both the F and Γ configurations, whereas the signal from the L configuration failed to show clear time delays between the peaks from the two tips. For F and Γ, one tip registers signals as a collision probe, while the other generates a reference signal caused only by induction. Comparison of the F and Γ signals shows that the peak amplitudes from the upper tip were larger for Γ, because of the lack of obstruction from the lower tip, while the amplitudes and shapes of peaks from the lower tip were similar for F and Γ. Comparison of the F and L signals reveals that the signals from the lower tip did not differ much, while the upper tip signal had much smaller peak amplitudes, and the shape of signals changed as well. This may be because of interference with the induced charge received by the upper tip (retracted) from the lower tip. The shape of current signals from each protruding tip was not affected by the presence of the other tip, and there was little change in peak amplitudes. The signals from the retracted tip were greatly affected by its position relative to the protruding tip.  55 -0.6-0.4-0.20.00.20.4-0.6-0.4-0.20.00.20.4-0.6-0.4-0.20.00.20.4(L)()lower tip upper tiplower tipupper tip(F) Current, 1E-7 Alower tipupper tip100Time, s (F: both tips protruding, : only upper tip protruding, L: only lower tip protruding as shown in Figure 2.2.)  Figure 2.26 Current signals from 2t1m probe with different configurations for multiple bubble injections in two-dimensional fluidized bed containing glass beads (Pressure in bubble injector: 345 kPa; solenoid valve opening time: 0.2 s and closing time: 3 s; centre of probe port 0.17 m above bubble injection port; bubble injector: 0.04 m above distributor).  2.5.4.2 Freely bubbling experiments with synchronization  In the freely bubbling experiments, the electrostatic signals transmitted by the 2t2mb probe were synchronized with photographic frames recorded by the camera (see Section 2.4). The charge density was varied by changing the superficial gas velocity, and measured by releasing particles into the Faraday cup during freely bubbling. The charge density remained the same when the bed was continuously fluidized at constant Ug, but decreased when the superficial gas velocity was reduced to Umf. Figure 2.27 shows the bed charge density measured from the sampling port by the Faraday cup after free bubbling when fluidizing with compressed air and with nitrogen. A higher charge density was generated with nitrogen as the fluidizing gas, presumably due to the lower humidity of the nitrogen. The charge density decreased for both cases after the fluidizing velocity was lowered to its minimum fluidization value. The charge density for using nitrogen showed a slight increase at 10 minutes, likely in error caused by the Faraday cup measurements, although it did not affect the overall 56 decreasing trend of charge density with time. When particles were fluidized, they acquired charges, mainly due to particle-particle and particle-wall collisions, whereas at minimum fluidization, collisions nearly disappeared and the charge generation rate decreased.  0 2 4 6 8 100.51.01.52.02.53.03.54.0-qm, in-bed charge density, C/kgTime, min Compressed air N2  Figure 2.27 Charge density decay curve after cutting off gas flow following free-bubbling with compressed air (RH=10%) and nitrogen (RH=2%).  In these experiments, single bubbles passing the probe in vertical alignment were selected from recorded videos for analysis to obtain bubble rise velocity, and corresponding current peaks were selected from the synchronized 2t2mb probe signals. Results from freely bubbling experiments with synchronization appear in Table 2.4. The magnitudes of Ipeak from both materials increased as the particle charge density increased, and a similar pattern was found when the bubble rise velocity increased. Comparison of Tables 2.3 and 2.4 reveals that particle charge densities were higher during freely bubbling than after single bubble injection, in agreement with results in Figure 2.27. The current peaks were also larger in free bubbling than for single bubble injection.       57 Table 2.4 Directly measured charge density (qm), bubble rise velocity (UB) and current peaks (Ipeak) from freely bubbling experiments with synchronization (see Section 2.4). Ug-Umf (m/s) qm (C/kg) UB (m/s) Ipeak,Ni (A) Ipeak,TiN (A) 0.12 -4.82E-6 1.05 -1.02E-6 -6.91E-7 0.12 -4.61E-6 1.05 -1.03E-6 -5.76E-7 0.12 -5.35E-6 0.91 -8.39E-7 -5.26E-7 0.12 -5.35E-6 1.11 -7.73E-7 -4.28E-7 0.12 -5.35E-6 0.91 -8.88E-7 -5.10E-7 0.09 -4.25E-6 0.77 -6.17E-7 -5.02E-7 0.09 -4.25E-6 0.77 -5.76E-7 -2.22E-7 0.09 -4.27E-6 0.91 -5.10E-7 -3.37E-7 0.09 -4.51E-6 0.91 -5.02E-7 -3.95E-7 0.09 -4.51E-6 0.71 -6.82E-7 -5.02E-7 0.06 -3.97E-6 0.67 -2.96E-7 -1.81E-7 0.06 -3.97E-6 0.80 -5.18E-7 -4.03E-7 0.06 -3.97E-6 0.67 -4.19E-7 -4.03E-7 0.06 -4.23E-6 0.77 -3.95E-7 -3.78E-7 0.06 -4.42E-6 0.83 -7.24E-7 -4.85E-7 0.06 -4.42E-6 0.83 -7.89E-7 -3.95E-7 0.02 -2.02E-6 0.59 -2.88E-7 -2.30E-7 0.02 -2.02E-6 0.59 -9.87E-8 -7.40E-8 0.02 -2.02E-6 0.56 -9.87E-8 -7.40E-8 0.02 -2.02E-6 0.40 -1.40E-7 -9.04E-8 0.02 -2.71E-6 0.63 -1.40E-7 -8.22E-8 0.02 -2.71E-6 0.43 -1.56E-7 -1.07E-7 0.02 -3.01E-6 0.53 -2.38E-7 -1.97E-7 0.02 -3.01E-6 0.50 -2.22E-7 -1.81E-7 0.02 -3.01E-6 0.43 -2.80E-7 -2.47E-7  2.5.4.3 Probe equations The transferred current from charged particles to the probe tip in a particulate flow system can be related to charge density and particle velocity by Eq. (2.5) with ai, bi and ci being fitted constants. The solid flow rate at the probe tip, Ws, can be expressed by Eq. (2.10). With 58 the particle velocity in the bubble nose and wake regions approximated by the bubble rise velocity and the voidage near the bubble by mf, the voidage at minimum fluidization, Eq. (2.5) can be re-arranged into the following form which represents the transferred current received by the probe when the bubble nose/wake contacts the probe.  2, (1 ) ( )tran i p mf B p i m i B iI U A a q bU c     (2.11)  Induced current equation was obtained from fitting the simulation results based on a simple charge induction model [21] (see Appendix C.1): 1.1, = -0.17ind i m BI q U  (2.12)  The total current arises from a combination of transferred and induced currents. Based on Eqs. (2.11) and (2.12), an equation of the following form  2, (1 ) ( )peak i i m B p mf B p i B iI q U U A U        (2.13) is proposed to represent the current peaks from the two materials. Eq. (2.13) was fitted to measured data (Ipeak,, qm and UB) for bubbles passing the probe from both single bubble injection (Table 2.3) and freely bubbling experiments with synchronization (Table 2.4), with different particle charge densities (qm) and bubble rise velocities (UB). For the present case with p= 2500 kg/m3, mf=0.37 for the dense phase and Ap= 3.6E−5 m2 (cross-sectioned area of probe tip surface), 2 21 1 10.21( / ),  4.59 -7( / ), 2.95 -6( / )kg m E C s kg m E C kg       (Ni) (2.14) 2 22 2 20.15( / ),  2.90 -6( / ), 1.21 -6( / )kg m E C s kg m E C kg       (TiN) (2.15)  The experimental and calculated values from Eq. (2.13) with the above ,  and  values are compared in a parity plot in Figure 2.28. Coefficients of determination were R2= 0.90 and 0.85 for Ni and TiN, respectively. Note that more experimental data could help in the fitting, thereby improving the accuracy of the measurements. The fitted constants are likely related to the properties of the probe materials and particles (see Section 2.5.1.4). Ipeak,1 and Ipeak,2 represent the current peaks from the two materials when the bubble nose/wake contacted the probe.  59 -1.2 -1.0 -0.8 -0.6 -0.4 -0.2 0.0-1.2-1.0-0.8-0.6-0.4-0.20.0 Ipeak,1 Ipeak,2Ipeak (Fitted value), AIpeak (Experimental value), A   Figure 2.28 Comparison of empirical correlation (Eqs. (2.13-2.15)) and experimental data for two probe materials.  For the dual-tip (one-material) probe, assuming that the effect of probe dimension on the fitted values can be ignored, the same empirical correlation was used in analyzing signals from the probe with the same particles.   2.5.5 Probes calibrations with particles of different properties  2.5.5.1 Glass beads with a size range of 106-212 m Following a similar procedure as above, the 2t2mb probe (see Table 2.1) was calibrated with 106-212 m GB. Single bubbles passing the probe in vertical alignment were selected from video recordings for analysis of bubble rise velocity, and corresponding current peaks were selected from the synchronized probe signals. Many data were discarded because of irregular bubble shapes or movement, such as coalescence, splitting or non-vertical passages. Maximum current peaks were picked from the signals at low Ug, while minimum peaks were selected from the signals at high Ug, because of the relative magnitudes at each Ug as shown in Figure 3.6. Parameters in Eq. (2.13) were fitted from measured Ipeak, qm and UB data (see Table B.1 in Appendix B.1) from the synchronization experiments with p= 2500 kg/m3, mf=0.37 for the dense phase and Ap= 3.6E−5 m2 for the present case. Coefficient of determination R2= 0.80 and 0.79 were found for Ni and TiN in this case, respectively.  60 2 21 1 10.23( / ),  1.10 -6( / ), 6.61 -6( / )kg m E C s kg m E C kg       (Ni) (2.16) 2 22 2 20.32( / ),  5.05 -7( / ), 1.04 -5( / )kg m E C s kg m E C kg       (TiN) (2.17) where i, i and i are fitted constants in Eq. (2.13). Comparing parameters in Eqs. (2.14-15) and (2.16-17), most parameters became larger for small particles; and 1 was larger than 2 for large particles, but the opposite was found for small particles, consistent with Figures 3.6 and 3.7.  -5.0 -4.5 -4.0 -3.5 -3.0 -2.5 -2.0 -1.5 -1.00.40.50.60.70.80.91.01.1Current, A500-600 mCharge density in the bed, C/kgBubble rise velocity, m/s-1.045E-06-9.177E-07-7.903E-07-6.630E-07-5.356E-07-4.083E-07-2.809E-07-1.536E-07-2.625E-08- .6 -2.4 -2.2 -2.0 -1.8 -1.6 -1.4 -1.20.40.50.60.70.80.91.0Current, A106-212 mCharge density in the bed, C/kgBubble rise velocity, m/s-1.360E-07-1.026E-07-6.925E-08-3. 88E-08-2.500E-093.087 - 86.425 - 89.762 -1.310E-07 Figure 2.29 Contour plots of current peaks from Ni tip of 2t2mb probe as a function of bubble rise velocity and particle charge density from calibration experiments for 106-212 m and 500-600 m GB.  Figure 2.29 shows the relationships between measured data (Ipeak, UB and qm) for glass beads with different mean size. For both cases, Ipeak increased with increasing qm when UB was 61 constant. However, a different trend of UB and Ipeak relationship was found in the data from particles with different sizes. When qm was constant, signal amplitudes from particles with larger mean size (500-600 m) show a larger variation when increasing UB than that from the particles with smaller mean size (106-212 m). This suggests hydrodynamics have different effect on magnitudes of electrostatic signals for particles of different sizes.   2.5.5.2 Polyethylene (PE) particles with narrow and wide size ranges Electrostatic charge generation is a well-known issue in polymerization reactors, leading to serious drawbacks, as discussed in Chapter 1. Polyethylene has particle properties which differ from those of glass beads in density, sphericity, roughness, dielectric constant etc. Due to these differences, the generation of electrostatic charge in fluidized bed may differ, resulting in different signals measured by electrostatic probes. Therefore, it is important to test the applicability of the proposed probes in fluidized beds of polyethylene particles.   Polyethylene resin powders used in this work were provided by NOVA Chemicals Corporation directly from their commercial fluidized bed reactors. Both original and sieved polyethylene particles were tested. The original ones had a relatively wide size range of 100-1500 m, whereas the sieved ones had a relatively narrow size range of 710−850 m. These size distributions were obtained by a Malvern Mastersizer 2000. (See Appendix B.7 for PSD graphs.) The minimum fluidization velocities were obtained from the intersection of two straight lines of pressure drop versus superficial gas velocities curves. (See Appendix B.6 for Umf curves.)  Based on a similar procedure as above, calibration experiments with polyethylene particles of different size ranges were conducted with different particle charge density (qm) and bubble rise velocity (UB) in the freely bubbling bed. Current signals from the probe and differential pressure signals across the bed are compared in Figure 2.30. These indicate a similar bubble frequency of 1-2 Hz, again suggesting that the electrostatic signals and bed hydrodynamics are related.  62 2 3 4 5 6 7 8 9 10 11 12-2.5-2.0-1.5-1.0-0.50.00.51.01.52.02 3 4 5 6 7 8 9 10 11 122.82.93.03.1Current, 1E-8 ATime, s upper lowerPressure signal, VTime, s Figure 2.30 Synchronized 2t1mrΓ probe (inserted 0.42 m above the distributor) and differential pressure signals across the bed in two-dimensional fluidization column of 710-850 m PE at Ug= Umf+0.04 m/s. (Fluidizing gas air: T=23.2oC, RH=4.2 %)  2t1mfΓ and 2t1mrΓ probes (see Table 2.1) were tested. Eq. (2.13) was fitted to measured (Ipeak,, qm and UB) data (see Table B.2 in Appendix B.1) with p=918 kg/m3, mf=0.55 for the dense phase. Fitted constants are shown in Table 2.5, with R2 larger than 0.80 for polyethylene particles and the probe tips material tested in this work. i, i and i are fitted constants related to the properties of the probe materials and particles (see Section 2.5.1.4). The experimental and values obtained from Eq. (2.13) with the fitted ,  and  values are compared in a parity plot in Figure 2.31. Similar to Figure 2.28, it is likely that more experimental data could improve the coefficient of determination and measurement accuracy. The properties of probe tip material, e.g. work function, electrical conductivity, tip size and shape, affect current signals and the calibration equation. Table 2.5 Fitted parameters in Eq. (2.13) for different probe tips and particle size distributions of PE powders. Case i, kg/m i, s2/kg·m2 i, C/kg R2 Particle size range of PE, m Tip shape Tip size (cross-sectional area) Tip material (a) 2.74E-3 4.45E-8 2.64E-7 0.82 710-850 flat 3.6E-5 m2 Nickel (b) 6.06E-3 1.80E-7 2.18E-8 0.80 100-1500 flat 3.6E-5 m2 Nickel (c) 1.25E-2 2.49E-6 7.98E-7 0.89 710-850 rod 6.9E-5 m2 Steel (d) 2.16E-2 -1.27E-6 2.84E-6 0.85 100-1500 rod 6.9E-5 m2 Steel  63 -25 -20 -15 -10 -5 0-25-20-15-10-50 wide PSD narrow PSDIpeak (fitted value), nAIpeak (experimental value), nAflat nickel tip  -70 -60 -50 -40 -30 -20 -1-70-60-50-40-30-20-10  wide PSD narrow PSDIpeak (fitted value), nAIpeak (experimental value), nArod steel tip  Figure 2.31 Comparison of fitted data (from Eq. (2.13)) and experimental data for different probe tips and particle size ranges of PE powders.  2.6 Summary Two dual-tip probe designs were fabricated: a dual-tip (two-material) probe, consisting of side-by-side tips made of materials of significantly different work functions, and a dual-tip (one-material) probe, having two vertically aligned tips of the same material. With retractable tips, the dual-tip (one-material) probe can be configured into a collision-collision probe with both tips protruding (“F” configuration), or a collision-induction probe, with only one tip exposed (“Γ” and “L” configurations).  The dual-tip (two-material) probe was calibrated in an ejector-funnel experimental setup, with the charge density on the particles changed by varying the pipe material and gas velocities. Several major factors that affect the transferred charge were studied, including the charge density, solid flux, particle velocity and angle of impact. The results were fitted to a semi-empirical equation (Eq. (2.5)) to quantify the effects of these parameters on the transferred current received by the probe from charged particles. Two tips, made of two materials (Ni and TiN) whose work functions differ, show different transferred currents when struck by charged particles, with the difference depending on the particle charge density, impact/collision velocity and contact angle. The largest difference between transferred 64 currents from the two materials was obtained when the particles struck the probe surface at right angles.   Substantial differences were observed in charge and current signals from the tips made of the same two materials in both vertical tube and vibration tray setups, as well as in both single-bubble and freely bubbling experiments in the two-dimensional fluidized bed. From both motor-pulley experiments and fluidization experiments, the difference in charge/current signals for the two probe materials arose mainly from charge transfer. The probes were calibrated with glass beads and polyethylene particles of different sizes in both single bubble injection and freely bubbling modes in the two-dimensional fluidized bed. The electrostatic signal transmitted by the probe was synchronized with the frames recorded by the camera. Eq. (2.13) was fitted to measured data (Ipeak, qm and UB) for bubbles passing the probe with different particle charge density (qm) and bubble rise velocity (UB) values. Several factors were found to affect the fitted parameters, in particular the dielectric constant of the particles, work function difference between the probe tip material and the bed particles, probe tip size and shape, and particle density and size distribution. For a dual-tip (one-material) probe, injected bubbles passed the two tips in succession and clear time delays appeared between the peaks from both the F and Γ configurations. The shapes of the current signals from each protruding tip were not significantly affected by the presence of the other tip, and there was little change in peak amplitudes. However, the signals from the retracted tip were greatly affected by its position relative to the protruding tip.  65 Chapter 3 Measurements in Gas-Solid Fluidized Beds 3.1 Introduction In this Chapter, the probes developed in Chapter 2 are employed to measure electrostatic charges in a two-dimensional fluidization column and two three-dimensional columns of inner diameters of 0.10 m and 0.30 m, with glass beads and polyethylene particles with different size ranges as bed materials.   3.2 Experimental equipment and methods 3.2.1 Two-dimensional fluidization column The two-dimensional fluidization column had an inside width of 0.307 m, a thickness of 22 mm and a height of 1.24 m, as shown in Figure 2.8. A sampling port and a probe port were located on opposite faces of the column at the same height (0.42 m above the distributor) and horizontal position (mid-point), in order to provide localized measurements. Particles from the sampling port were discharged into a Faraday cup to measure in-bed charge density directly. A high-speed video camera recorded bubble movement during experiments, with bubble rise velocities determined from the recorded video images. Details of the experimental facilities are provided in Section 2.4.   3.2.2 Three-dimensional fluidization column of ID 0.10 m Figure 3.1 shows a schematic diagram of a 0.10 m ID three-dimensional fluidization column and charge measurement system. The cylindrical column was made of Plexiglas with an inner diameter of 0.10 m and a height of 1.0 m. A double perforated-plate distributor was installed at the bottom of the column, 0.17 m above the gas entrance. The top and bottom distributor plates both had 31 aligned orifices of diameters 2.0 mm (above) and 3.0 mm (below), with a fine metal screen (~38 m) between the two plates. Compressed air was introduced into the column, controlled by a panel-mount air flow meter. Electrostatic probes were inserted in two ways: either vertically from the top of the column into the dense phase 66 of the bed at the axis 0.11 m above the distributor, or horizontally from the side 0.10 m above the distributor. A sampling port (6.35 mm in diameter, inclined downward at 30 to the vertical) was installed on the column wall at the same height of the probe port in the dense phase region of the bed. Particles from the sampling port were discharged into a Faraday cup to measure the in-bed charge density directly. The static bed height was maintained at 0.20 m throughout the experiments. Details of the charge measurements can be found in Sections 2.3 and 2.4. Air flow meterVentCompressed airData acquisition systemComputerElectrometerSampling portFaraday cupElectrostatic  probe ElectrometerDesiccant dryer Figure 3.1 Schematic of 0.10 m ID three-dimensional column and charge measurement system (not to scale).  3.2.3 Three-dimensional fluidization column of ID 0.30 m  Free bubbling experiments were also performed in a three-dimensional fluidization column of inner diameter 0.30 m and height 4.5 m [67], as shown schematically in Figure 3.2 and Appendix A.5. A double perforated plate distributor containing top and the bottom plates, both having 98 holes, with diameters 5.6 mm and 64 mm, respectively, with a fine metal screen mesh (~38 m) sandwiched between the two plates to prevent particles from falling into the windbox. Fluidizing air was supplied by a blower with a maximum capacity of 425 67 Nm3/h at 69 kPa. The air flow rate was controlled by a by-pass line located close to the blower, and calculated from the pressure drop across an orifice plate. A sampling port (12.7 mm in diameter, inclined downward at 30 to vertical) was located 0.25 m above the distributor. Particles from the sampling port were discharged into a Faraday cup to measure in-bed charge densities directly.       VentData acquisition systemElectrometersFaraday cupElectrometerComputerBlowerBypass lineOrifice plate CycloneBag house filterCycloneStatic bed heightExpanded bed heightDistributor platesGas inlet0.25 m0.70 m1 mProbe 1 Probe 2 Probe 3  Figure 3.2 Schematic of 0.30 m ID three-dimensional column and charge measurement system (not to scale).  In some of the experiments, three 2t1m probes (see Table 2.1) were inserted at different heights along the axis of the column. Three probes were installed at the same time below the surface of the expanded bed. With increasing the superficial gas velocity, some fine particles were entrained from the top of the column and collected by two external cyclones. It was observed that the inside of the return leg was coated by polyethylene particles during the experiments, due to electrostatics.   68 Both relative humidity and temperature of the gases were monitored during all experiments at the exit of fluidization columns. Particles were washed with water and ethanol, and then dried overnight in an oven to eliminate dust and moisture. Both narrow and wide size ranges particles were tested. Each bed was fluidized at a certain superficial gas velocity for at least an hour (1.5 h for the 0.30 m ID column) to let the system reach steady state before measurements. All measurements were then performed three times at each superficial gas velocity. The order of the superficial gas velocities was chosen randomly. The probe signals were not synchronized with photographic images in these experiments.  3.3 Results and discussion 3.3.1 Results from two-dimensional fluidization column The following experiments were conducted on different days from the synchronization experiments used for probe calibration. Therefore the data used to estimate qm and UB were not the same as those used to obtain the calibration equation.  3.3.1.1 Dual-tip (two-material) probe with 500-600 m GB  A 2t2mb probe, as shown in Figure 2.1(b), was used. Figure 3.3 shows 5 s traces of charge signals and corresponding current signals from the probe for a superficial gas velocity of 0.50 m/s. The maximum and minimum peaks (Imax and Imin) were defined, for a series of bubbles rising and passing during freely bubbling. The variation in peak amplitudes is due to variations in bubble properties and trajectories in the freely bubbling flow regime. The magnitudes of peaks were normally larger for Ni than for TiN. This seems to be contradictory, given that the work function difference between TiN (2.9 eV) and glass beads (SiO2: 5.0 eV) is larger than that between Ni (5.0-5.4 eV) and glass beads (see Table 2.2). One might expect a higher work function difference between contacted objects to result in a larger transferred charge. However, the total transferred charge during contact depends not only on the work function between the particles and metal surface, but also on the initial charge on particles, contact velocity, effective contact area, surface conditions and other material properties [61, 118]. The total net transferred currents between the metal and the 69 colliding particles for two different metals may vary and even change direction when such factors as particle charge density, particle velocity or flow rate change [60]. Moreover, work functions of dielectric particles and metals can be significantly affected by surface and environmental conditions (e.g., roughness, geometry, contamination, oxide layer and surface chemistry) [84, 123]. Similar results were found in Sections 2.5.1.4 and 2.5.3.  By counting the number of peaks within a certain period of time, the void frequency could be estimated. For example, the estimated frequencies were ~1 Hz at Ug=0.30 m/s and 3~4 Hz at Ug=0.50 m/s. For all superficial gas velocities tested, there were measurable differences between the peak amplitudes for the Ni and TiN tips.  -0.40-0.36-0.32-0.28-0.24-0.20-0.16-0.12-0.08Cumulative charge, CTime,sTiNNi0 1 2 3 4 5(a)-0.6-0.5-0.4-0.3-0.2-0.10.00.1-0.6-0.5-0.4-0.3-0.2-0.10.00.1ImaxCurrent, ACurrent, A Ni0 1 2 3 4 5IminTime, s TiN(b) Figure 3.3 Cumulative charge (a) and current (b) signals from 2t2mb probe in freely bubbling two-dimensional fluidized bed of 500-600 m GB with static bed height of 0.30 m at Ug=0.50 m/s over duration of 5 s. (Fluidizing gas air: T= ~22oC, RH= ~10 %)  As noted above, as a bubble surrounded by charged particles passed the collision probe, two current peaks appeared with opposite signs, corresponding to the moments when the bubble nose and wake reached the probe. Both peaks involve summations of transferred and induced currents. The transferred currents are assumed to be the same when the bubble nose and wake contact the probe. The induced currents arise by charged particles passing the probe, but not necessarily contacting the probe tip. If, for example, a bubble with negatively charged particles passes the probe, a negative induced charge is registered by the conductive probe tip. After passage of the bubble, the induced charge returned to zero. A positive induced current was registered at the maximum peak, followed by a negative induced current of equal 70 magnitude at the minimum peak. Because of near front/back symmetry for rising bubbles, the magnitudes of the induced currents should be similar, as indicated by the simulation results in Appendix C.1. Therefore the average of the maximum and minimum peak values cancels out the induced charges, as demonstrated by a simple induction model: max tran indI I I  (3.1) min tran indI I I   (3.2) so that max min( ) / 2tranI I I  (3.3) and max min( ) / 2indI I I   (3.4)  -9 -8 -7 -6 -5 -4 -3 -2 -1-35-30-25-20-15-10-5051015202530Ug=0.6 m/sUg=0.5 m/sUg=0.4 m/sCurrent, nACharge density in the bed, C/kg Mean, Ni Mean, TiN SD divided by 10, Ni SD divided by 10, TiNUg=0.3 m/s Figure 3.4 Mean and standard deviation of current as a function of superficial gas velocity and particle charge density in freely bubbling two-dimensional fluidized bed with a static bed height of 0.30 m.   The mean and standard deviation (SD) of current are plotted in Figure 3.4 as a function of charge density and superficial gas velocity. Both the mean and SD increased as the in-bed charge density and superficial gas velocity increased, with the difference between the two materials increasing in a similar manner. This again suggests that the charge/current signals measured by an electrostatic probe in a fluidized bed are affected by both particle charge density and bed hydrodynamics. The charge inside the fluidized bed is closely related to the charge density on the particles inside the bed, as well as to the bed hydrodynamics. Generally, the mean current represents the net transferred current [93]. Although single 71 bubble experiments showed that bubbles may split after contacting the probe, the average transferred current, tranI , and average induced current, indI , could still be related via Eqs. (3.3-3.4) to the average of the maximum and minimum peaks heights, maxI and minI .  Figure 3.5 compares max min( ) / 2I I  and max min( ) / 2I I  from the two materials. It shows larger difference in the former than the latter for the two materials, suggesting that the difference in current signals from the two materials mainly results from the transferred parts, whereas the induced currents from the two materials are likely to be the same; this is consistent with the results in Section 2.5.2. Comparing mean currents in Figure 3.4 and max min( ) / 2I I  values in Figure 3.5, the former is much smaller in magnitude than the latter, with the difference between the two materials (Ni and TiN) much larger in the latter case than in the former. The mean current is low because of the weak signals (magnitude close to zero) associated with the passage of the emulsion phase far from the bubble and possible electrical noise. Figure 3.5 also shows that the magnitude of transfer currents from both materials increased as the superficial gas velocity and charge density increased, with the difference between the two signals increasing in a similar manner. Peak values are therefore used in the signal analysis in Chapter 4.  -0.6-0.5-0.4-0.3-0.2-0.10.0 (Imax+Imin)/2, Ni (Imax+Imin)/2, TiN (Imin-Imax)/2, Ni (Imin-Imax)/2, TiNCurrent, A0.400.30 0.50 0.60Superficial gas velocity, m/s Figure 3.5 Comparison of max min( ) / 2I I  and min max( ) / 2I I  for both materials at different superficial gas velocities in two-dimensional fluidized bed with static bed height of 0.30 m.  72 3.3.1.2 Dual-tip (two-material) probe with 106-212 m GB Charge and current signals from the 2t2mb probe (see Table 2.1) with fluidizing 106-212 m GB are illustrated in Figure 3.6. 0 50 100 150 200 250 3000.00.20.40.60.81.00 50 100 150 200 250 3000.00.20.40.60.81.00 50 100 150 200 250 3000.00.20.40.60.81.00 50 100 150 200 250 300-1.0-0.8-0.6-0.4-0.20.0TiNTiNTiNTiNNiNiNiUg=0.14 m/sUg=0.11 m/sUg=0.09 m/sUg=0.07 m/s Ni TiN Ni TiNCumulative charge, CTime, s Ni TiNNiCumulative charge, CTime, s Ni TiNCumulative charge, CTime, sCumulative charge, CTime, s -20-100120-20-1001020-20-1001020-20-1001020 Current, 1E-8 A Ni  TiN  Ug=0.14 m/sUg=0.11 m/sUg=0.09 m/sUg=0.07 m/s10Time, s0 Figure 3.6 Cumulative charges and current signals from 2t2mb probe at different superficial gas velocity in two-dimensional freely bubbling bed of 106-212 m GB. (Fluidizing gas air: T= ~22oC, RH= ~10 %)  73 The charge signals from both tips first decreased, then increased with polarity changing from positive to negative, with increasing superficial gas velocity (Ug), indicating that the transferred current changed from positive to negative as Ug increased. The difference between the slopes of the charge signals from the two tips first decreased then increased with Ug increased. For 106-212 m GB, in the current signals from the probe, positive peaks had higher amplitudes than negative peaks at Ug=0.07 m/s, whereas the negative peaks were more dominant at Ug=0.14 m/s. For 500-600 m GB, the minimum peaks were always dominant for the tested range of Ug, and the charge signals were negative with the difference between the slopes of charge signals from the two tips increasing as Ug increased. Polarity change results were similar to those in Sections 2.5.1.5, 3.3.1.3 and as reported by others [2, 118, 121, 124, 125]. In this case, the polarity change may reflect the combined effect of tribocharging and transfer of charges from charged particles to the probe tips. The probe signal polarity is not necessarily the same as the polarity of the particles and the probe signal may not be used as an indication of the charge polarity on particles. It is necessary to extract the particle charge density from the probe signal. More results are shown in Appendix B.9. -2.8 -2.6 -2.4 -2.2 -2.0 -1.8 -1.6 -1.4 -1.2-1.0-0.8-0.6-0.4-0.20.00.20.40.60.81.0Ug=0.14 m/sUg=0.11 m/sUg=0.09 m/sMean current, 1E-8 ACharge density in the bed, C/kg Ni TiNUg=0.07 m/s Figure 3.7 Mean currents as a function of superficial gas velocity and particle charge density for 106- 212 m GB.  The mean currents are plotted in Figure 3.7 as a function of charge density (qm) and superficial gas velocity (Ug). The mean currents first decreased, then switched from positive to negative and increased in magnitude as Ug increased. The difference between the signals 74 from the two materials changed in a similar manner, consistent with the results in Figure 3.6. The relative magnitude of the net current transferred between the metal and contacting particles for the two metals may vary and even reverse with variation of such factors as particle properties, charge density, particle velocity and gas flow rate, as shown in previous results (see Sections 2.5.1.4, 2.5.3 and 3.3.1.1) and as reported by Matsusaka and Masuda [60].  3.3.1.3 Dual-tip (one-material) probe with 500-600 m GB 2t1m probes, as shown in Table 2.1, were employed. The effect of superficial gas velocity on the 2t1mfF probe signals is shown in Figure 3.8. The slopes of cumulative charge signals increased for both tips, indicating increased charge transfer as Ug increased. The fluctuations from each tip also increased as Ug increased, because of the increase in bubble size and frequency. The fluctuations from the lower tip were larger than from the upper one, due to direct exposure to bubbles. The polarity of the cumulative charge signals from the lower tip changed as Ug varied, consistent with previous results and literatures. This may be because of the combined effect of triboelectrification and transfer of charges or contamination of the probe tip surfaces. The amplitudes of current signal peaks from both tips increased as Ug increased. The bubble frequency, found by counting peaks, also increased with increasing superficial gas velocity. 0 20 40 60 80 100-1.2-1.0-0.8-0.6-0.4-0.20.00.20.4Ug-Umf = 0.14 m/sUg-Umf=0.08 m/s upper tip lower tipCumulative charge, C Time, sUg-Umf = 0.02 m/s-101-100-10010Current, 1E-7 A upper tip  lower tip10Ug-Umf=0.14 m/sUg-Umf=0.08 m/sUg-Umf=0.02 m/sTime, s0 Figure 3.8 Cumulative charge and current signals from 2t1mfF probe in freely bubbling two-dimensional fluidized bed of 500-600 m GB at different superficial gas velocities. (Fluidizing gas air: T=~23oC, RH=~3 %)  75 The 2t1m probe with a different configuration (see Table 2.1) was also tested in the bed at different superficial gas velocities. The effects of different configurations on the cumulative charge and current signals from this probe are shown in Figure 3.9. Slopes of cumulative charge signals from the upper tip increased (positively at Ug-Umf=0.02 m/s and negatively at Ug-Umf=0.14 m/s) when the probe configuration changed in the order L, F, Γ. The current peak amplitudes increased as well. The slopes of cumulative charge signals and current peak amplitudes from the lower tip increased when the probe configuration changed in an order of of Γ, F, L. At Ug-Umf=0.14 m/s, the cumulative charge signal from the lower tip increased positively, possibly due to the interference of charges on the column wall.  0 50 100 150 200 250 300-0.050.000.050.100.150.200.250.300.35Ug-Umf=0.02 m/s(F)(L)( )( )(L)(F)Cumulative charge, C Time, s upper tip lower tip      20 40 60 80 100-1.0-0.8-0.6-0.4-0.20.00.20.40.6Ug-Umf=0.14 m/s(L)( ) upper tip lower tip(L)( )Cumulative charge, C i , (F) -4-202-4-202-4-202(L)( )(F) upper tip  lower tipUg-Umf=0.02 m/sCurrent, 1E-7 A100Time, s  -10-505-10-505-10-505(L)( )(F) upper tip  lower tipCurrent, 1E-7 AUg-Umf=0.14 m/s100Time, s (F: both tips protruding,  : only upper tip protruding, L: only lower tip protruding as shown in Figure 2.2)  Figure 3.9 Cumulative charge and current signals from 2t1m probe with different configurations in freely bubbling two-dimensional fluidized bed of 500-600 m GB at excess superficial gas velocities, Ug−Umf =0.02 and 0.14 m/s. (Fluidizing gas air: T=~23oC, RH=~3 %)  76 3.3.1.4 Dual-tip (one-material) probe with 710-850 m PE Fouling/sheeting (0.20-0.25 m above static bed height) was observed on the column wall after fluidization. Since the relative magnitudes of signals from the two materials of the 2t2m probe may change for different operating conditions and particle properties, causing difficulty in signal processing (see Sections 2.5.1, 3.3.1 and 4.4.2.2), the 2t1m probe was employed for the polyethylene tests. Figure 3.10 shows cumulative charge and current signals from 2t1mfF probe in freely bubbling experiments at different superficial gas velocities. The slopes of cumulative charge signals, amplitudes of current peaks from both tips and bubble frequencies, all vs. time, increased as Ug increased. The cumulative charge signals from the lower tip showed little variation with changing superficial gas velocity, whereas in Figure 3.19 the slope of charge signals from the lower tip increased significantly as increasing Ug. This may because a possible thin layer of particles was coated on the lower tip surface at low Ug, but vanished by flushing effect as increasing Ug.  -0.5-0.4-0.3-0.2-0.10.00.1-0.4-0.20.00.20.4-0.5-0.4-0.3-0.2-0.10.00.1-0.4-0.20.00.20.40 20 40 60 80 100 120-0.5-0.4-0.3-0.2-0.10.00.1-0.4-0.20.00.20.4Current, 1E-7ACumulative charge, E-7CUg-Umf=0.04 m/sUg-Umf=0.09 m/sUg-Umf=0.09 m/sUg-Umf=0.04 m/sUg-Umf=0.01 m/sUg-Umf=0.01 m/sTime, s upper tip   lower tipTime, s0 10 Figure 3.10 Cumulative charge and current signals from 2t1mfF probe in freely bubbling two-dimensional fluidized bed of 710-850 m PE at different Ug. (Fluidizing gas air: T=~23oC, RH=~7 %) 77  3.3.2 Results from three-dimensional fluidization column of ID 0.10 m 3.3.2.1 Effects of probe inserted distance and direction Glass beads of size range of 500-600 m were tested next. A 2t2ma probe (see Table 2.1) was inserted horizontally into the bed. The probe tip was moved from the centre to near the wall. Data were recorded 20 min after changing the probe position each time to achieve steady state. Typical cumulative charge signals from the probe are shown in Figure 3.11. The slopes of signals at the centre were larger than near the wall and the slope for TiN was larger than for Ni in both cases.  0 50 100 150 200 250 300-0.14-0.12-0.10-0.08-0.06-0.04-0.020.000.02Ni, near wallTiN, near wallNi, centreCumulative charge, CTime, sTiN, centre Figure 3.11 Cumulative charge signals from 2t2ma probe horizontally inserted to centre and near the wall in 0.10 m ID freely bubbling fluidized bed (Ug=0.29 m/s). (Fluidizing gas air: T= ~22oC, RH= ~10 %)  Figure 3.12 shows that the magnitudes of the mean and SD of current for both materials decreased when the probe was moved from the centre of the column (probe insertion distance: 0.05 m) to near the wall (insertion distance: 0.005 m). When the probe was moved to the centre, the magnitude of both the mean and SD of currents for both materials increased. The results were repeatable. Due to the wall effect, hydrodynamics are weaker near the wall than in the centre of the column, resulting in lower frequency and intensity of particle-particle and particle-probe surface interactions. Differences existed in the mean and 78 SD of currents from the two materials, regardless of their relative positions. This difference was altered when the hydrodynamics and/or charge density inside the bed changed.   Cross-correlation and coherence function were also calculated to analyze current signals from the two materials. Results show that with almost no time delay between these two signals, the two signals corresponded well to each other at 0~10 Hz, but became unrelated at frequency > 50 Hz. The coherence function increased with increasing superficial gas velocity. -0.04 -0.02 0.00 0.02 0.04 0.06-0.5-0.4-0.3-0.2-0.10.0(move back to centre) (near wall) (centre)Mean current, 1E-9 AProbe insertion distance, m Ni-Mean TiN-mean2345678 SD-Ni SD-TiNSD of current, 1E-9 A Figure 3.12 Mean and standard deviation of current signals from 2t2ma probe as a function of probe insertion distance.  Figure 3.13 shows the cumulative charges from the probe inserted at different angles into the column. The relative magnitude of transferred currents from the two materials differed in the two cases, indicating that the contact angle (probe insertion direction) between the particles and probe surface affects the magnitude of transferred current. Similar results are reported in Sections 2.5.1.4, 2.5.3, and 3.3.1. The slopes of the charges from the two materials and the difference between signals were larger when the probe was inserted vertically than that when it was inserted horizontally. The probe surface faced downward so that charged particles could directly strike the metal tips. The transferred current was larger from TiN than from Ni when the probe was inserted horizontally, consistent with Figure 3.11. Therefore, the probe was inserted vertically in the next of the experiments. 79 0 50 100 150 200 250 300-0.35-0.30-0.25-0.20-0.15-0.10-0.050.000.05Cumulative charge, CTime, s(a) TiNNi 0 10 15 20 25 30 35 40-1.0-0.8-0.6-0.4- .20.0(b)Cumulative charge, Ci e, sNiTiN Figure 3.13 Comparison of cumulative charge signals from 2t2ma probe inserted: (a) horizontally and (b) vertically into 0.10 m ID freely bubbling fluidized bed (Ug=0.33 m/s). (Fluidizing gas air: T= ~22oC, RH= ~10 %)  3.3.2.2 Comparison of current signals Figure 3.14 shows a typical time series of current signals measured by the 2t2ma probe (see Table 2.1) from a homogeneous particulate flow and a bubbling fluidized bed. -0.4-0.20.0-50(b)(a)20 Ni TiNCurrent, 1e-7A Ni TiN2Time, s0 Figure 3.14 Typical current signals from: (a) homogeneous particulate flow with particles dropped from funnel; (b) 0.10 m ID freely bubbling fluidized bed.  In Figure 3.14(a), the particles were dropped continuously at a steady flow rate directly from a Plexiglas funnel, with an average particle velocity (Vp) of 0.86 m/s; the probe was placed 50 mm below the funnel exit and aligned with it, with the probe surfaces facing upward, 80 perpendicular to the particulate flow. The homogeneous continuous flow resulted in a constant transferred current, with small variations, and the current signal was smaller in magnitude than that in the fluidized bed because of a smaller charge density on the particles. In Figure 3.14(b), the particles were fluidized with a superficial gas velocity (Ug) of 0.36 m/s. The current signal was larger in magnitude in the bubbling fluidized bed than that in the homogenous dropping flow because of greater charges on the particles resulting from fluidization. The current signal was also strongly affected by local hydrodynamics, with fluctuations in the current signal caused by bubble movement.   Raw signals are presented at different superficial gas velocities (Ug) in Figure 3.15. Collisions of negatively charged particles with the probe resulted in a negative current. Because of the dimensions of the probe and the way the probe was inserted in the bed, the flat sensor pieces received little induced current, with most of the signals being contributed by the current transferred from collisions of charged particles. The peak amplitudes from both materials, as well as peak frequencies, increased with increasing Ug. The signals also showed a difference between the peak amplitudes from the two materials.   -20-100-20-100-20-100-20-100Current, 1E-7 AUg=0.36m/sUg=0.33m/s Ni   TiNUg=0.44m/sUg=0.39m/sTime,s 0 51 2 3 4 Figure 3.15 Current signals from 2t2ma probe inserted vertically into 0.10 m ID freely bubbling fluidized bed at different superficial gas velocities. (Fluidizing gas air: T= ~22oC, RH= ~10 %)  81 3.3.2.3 Statistical and FFT analysis Figure 3.16 shows the mean, standard deviation and normalized standard deviation of current signals from the probe in the freely bubbling fluidized bed. Both the mean and standard deviation of current signals increased as Ug increased, and the difference between the signals from the two materials also increased, consistent with the results in Figure 3.15. Also by measuring qm from sampling particles out of the bed, it suggested that the mean currents from the two materials and the difference between these two currents increased as qm increased. The difference between the two signals, which depends on the work function difference of the two probe-tip materials, varied as a result of changing the charge density and/or hydrodynamics. The work function difference is only expected to affect the transfer part, not the induction part, as indicated in Section 2.5.2. In this case, a 2t2ma probe was inserted vertically into the dense phase of the bed that the probe measured mainly charge transfer, so both mean and standard deviation of current signals from this novel probe include charge transfer information, and are influenced by the difference in material properties, as shown in Figure 3.16 (a) and (b). The normalized standard deviation cancels out the charge density and mainly reflects local hydrodynamics [93]. Figure 3.16(c) shows that the normalized standard deviations (standard deviation/mean) of the two materials increased as Ug increased. Figure 3.16 (a) Mean, (b) standard deviation and (c) normalized standard deviation of current signals from 2t2ma probe as function of superficial gas velocity in 0.10 m ID freely bubbling fluidized bed (probe inserted vertically). (Experiments for Ug=0.22, 0.25, 0.28 and 0.31 m/s were conducted on a different day as experiments for Ug=0.33, 0.36, 0.39 and 0.44 m/s)  Figure 3.17 shows the power spectrum of current signals, obtained from Fast Fourier Transformation (FFT) analysis. For both materials, the amplitude increased with increasing superficial gas velocity, confirming that current signals are closely related to hydrodynamics. 0.20 0.25 0.30 0.35 0.40 0.45-2.5-2.0-1.5-1.0-0.50.0TiNUg, m/sMean current, 1E-7 ANi(a)0.20 0.25 0.30 0.35 0.40 0.450.00.51.01.52.02.53.03.5TiNNiUg, m/sStandard deviaton, 1E-7 A(b) 0.20 0.25 0.30 0.35 0.40 0.45-10-8-6-4-2TiNUg, m/sNormalized standard deviation Ni(c) 82 The amplitudes are always larger for Ni than for TiN, confirming that the difference in these two signals is significant. Both Ni and TiN show characteristic frequencies of 0~10 Hz at different gas velocities. 0 10 20 30 40 5005101520250 10 20 30 40 5005101520250 10 20 30 40 50012243648600 10 20 30 40 50012243648600 10 20 30 40 50012243648600 10 20 30 40 50012243648600 10 20 30 40 5002652781041300 10 20 30 40 500265278104130Amplitude, nAAmplitude, nAAmplitude, nAAmplitude, nA0.33m/s0.36m/s0.39m/s0.44m/s Frequency, HzNi TiNFrequency, Hz Figure 3.17 FFT of current signals from 2t2ma probe in 0.10 m ID freely bubbling fluidized bed at different superficial gas velocities.  3.3.3 Results from three-dimensional fluidization column of ID 0.30 m 3.3.3.1 Dual-tip (two-material) probe with 420-590 m GB Glass beads with a size range of 420-590 m (Geldart group B, 30/40 mesh, Potters) were used in these experiments. A 2t2mb probe (see Table 2.1) was employed and inserted 0.25 m above the distributor to the centre to measure the charge inside the bed. The static bed height was 0.40 m. Figure 3.18 shows the current signals from the probe at different superficial gas velocities. The peak amplitudes and frequencies from both materials increased with increasing Ug. The peak amplitudes from the two materials also differed. Comparison of FFT probe signals with and without the air flow showed that the amplitude of noise was much smaller, so that the effect of noise could be ignored.  83 -4-202-4-202-4-202Ug=0.37m/sUg=0.32m/sUg=0.27m/s Ni  TiN0 10Current, 1E-7 ATime, s Figure 3.18 Current signal from 2t2mb probe in 0.30 m ID freely bubbling fluidized bed of 420-590 m GB at different superficial gas velocities. (Fluidizing gas air: T=405 oC, RH=154 %)  3.3.3.2 Dual-tip (one-material) probe with original PE-I (100-1500 m) Since the relative magnitudes of signals from the two materials of the 2t2m probe may change for different operating conditions and particle properties (e.g. mean size and size distribution), causing difficulty in signal processing (see Sections 2.5.1, 3.3.1 and 4.4.2.2), the 2t2m probe may be unsuitable for particles of wide size distribution. 2t1m probes with different configurations and tip shapes (see Table 2.1) were next investigated with original polyethylene particles of relatively wide size distribution as the bed material. The static bed height was maintained at 0.40 m, with the probe inserted 0.25 m above the distributor to the axis to measure the charge inside the bed. Typical cumulative charge and current signals from probes at different Ug are shown in Figure 3.19. The slopes of cumulative charge signals vs time, amplitudes of current peaks from both tips and bubble frequencies all increased as Ug increased, consistent with our results from the two-dimensional fluidized bed (Section 3.3.1).  84 0 50 100 150 200 250 300-3.5-3.0-2.5-2.0-1.5-1.0-0.50.0Ug- Umf =0.14 m/s  upper tip   lower tipUg- Umf=0.10 m/s Cumulative charge, 1E-8CTime, sUg- Umf=0.19 m/s -101-202-303 upper tip   lower tipCurrent, 1E-8 AUg- Umf=0.19 m/sUg- Umf=0.14 m/s Ug- Umf=0.10 m/s Time, s0 10 (2t1mfF probe) 0 50 100 150 200 250 300-7-6-5-4321Ug- Umf=0.06 & 0.11 m/s upper tip   lower tipUg- Umf=0.19 m/sCumulative charge, 1E-8CTime, s-101-303-505 upper tip   lower tipUg- Umf=0.06 m/sUg- Umf=0.11 m/sUg- Umf=0.19m/sCurrent, 1E-8 ATime, s0 10 (2t1mrΓ probe) Figure 3.19 Cumulative charge and current signals from 2t1m probes at different Ug in freely bubbling 0.30 m ID three-dimensional fluidized bed of original PE-I particles. (Fluidizing gas air: T=405 oC, RH=154 %)  It is of interest also to test multiple probes at different axial heights and higher superficial gas velocities, in the turbulent fluidization flow regime. The static bed height was maintained at 0.88 m. Figure 3.20 shows charge and current signals from a 2t1mfΓ probe (see Table 2.1, insertion 0.25 m above distributor) at different Ug. Slopes of charge signals from both upper tip (protruded) and lower tip (retracted), as well as the current peak frequency first increased then decreased as increasing Ug. Current peak amplitudes increased with increasing Ug. Note that the probe signal is related to charge density level and local hydrodynamics inside the bed. In the bubbling regime, where bubble coalescence was dominant, the bubble size and frequency increased with increasing Ug, so that more charge was generated by more and 85 stronger collisions between particles. Thus the slope of charge signal and the current peak amplitudes increased. In the turbulent fluidization flow regime, the charge induction decreased as large bubbles were replaced by small and transient voids, causing the slope of the charge signals to decrease. Because of decreased separation distance between leading-trailing voids and increased average distance between particles in the turbulent fluidization flow regime, current peak frequency and time gap between groups of peaks increased. The current peak amplitude, which is more related to the charge density and Ug, also increased.  0 10 20 30 40 50-1.2-1.0-0.8-0.6-0.4-0.20.0Ug-Umf= 0.73 m/s, upper tipUg-Umf= 0.73 m/s, lower tipUg-Umf= 0.31 m/s, upper tipUg-Umf= 0.31 & 0.57 m/s, lower tipUg-Umf= 0.57 m/s, upper tipUg-Umf= 0.15 m/s, lower tipUg-Umf= 0.15 m/s, upper tipCumulative charge, CTime, sProbe 1-10-50510-20020-40-2002040-40-2002040Ug-Umf= 0.73 m/sUg-Umf= 0.57 m/sUg-Umf= 0.31 m/s upper tip  lower tip (Probe 1)Ug-Umf= 0.15 m/s0 10Current, 1E-8 ATime, s Figure 3.20 Cumulative charge and current signals from 2t1mfΓ probe (insertion 0.25 m above distributor) at different Ug in 0.30 m ID fluidized bed of original PE-I particles. (Fluidizing gas air: T=405 oC, RH=154 %)  Current signals from 2t1mfF probes at different height (Probe 1, 2, 3 as shown in Figure 3.2) are plotted in Figure 3.21. Current peak frequency decreased from Probes 1 to 3 as the bubbles coalesced and grew and the current peak frequency from Probe 1 was larger than that from Probes 2 and 3 because of more collisions near the distributor. Current peak amplitudes for Probes 1 and 3 were slightly larger than that from Probe 2, indicating there might be a small charge density distribution along the column height, although further validation is needed.  86 -40-2002040-20-1001020-20-1001020 upper tip  lower tip0 10Current, 1E-8 AProbe 3Probe 2Time, sProbe 1 Figure 3.21 Current signals from 2t1mfF probes at different locations in 0.30 m ID fluidized bed of original PE. (probe 1: z1=0.25 m, probe2: z2=0.70 m and probe 3: z3=1.00 m, Ug-Umf=0.24 m/s) (Fluidizing gas air: T=405 oC, RH=154 %)  Standard deviations (SD) of current signals from the upper tips of probes and absolute pressure signals at different Ug are shown in Figure 3.22. The transition velocity, Uc, determined from the maximum standard deviation of current signals, tended to decrease with increasing distance above the distributor. This transition velocity (~0.5-0.6 m/s) is different from the values determined from pressure fluctuations (~0.9 m/s) and from the correlation based on absolute pressure measurement (0.84 m/s) [126, 127]. Similar results were found for corresponding superficial velocity of maximum heat transfer coefficient and Uc from standard deviation of pressure fluctuations [128]. The electrostatic signals contain both charge and hydrodynamics information whereas the pressure signals only reflect hydrodynamics inside the bed. The SD from Probe 1 was larger than that from Probe 2, possibly due to the higher charge density near the distributor than the middle of the bed, similar to results in Figure 3.21. The standard deviation of electrostatic signals was related to charge induction caused by passing voids. In the bubbling regime, bubble coalescence was dominant so that charge induction increased as the bubble size increased; in the turbulent fluidization flow regime, charge induction decreased as large bubbles were replaced by small and transit voids.   87 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.90.00.51.01.52.02.53.03.5Standard deviation of current, ASuperficial gas velocity, m/s probe 1, z1=0.25 m probe 2, z2=0.70 m 0. .3 0.4 0.5 0.6 0.7 0.8 0.9 1.00.10.20.30.40.50.60.70.8Standard deviation of absolute pressure, kPaSupercial gas velocity, m/s Figure 3.22 Standard deviations of current signals from 2t1m probes and absolute pressure signals versus superficial gas velocity.  3.4 Summary The probes developed in Chapter 2 determined electrostatic charges on particles in fluidization columns of different shapes and scales: a two-dimensional column and two three-dimensional columns of inner diameters of 0.10 m and 0.30 m.  Based on experiments in these three columns, there were substantial differences between the signals (mean and standard deviation of currents and current peak amplitudes) for the Ni and TiN tips of dual-tip (two-material) probe at all superficial gas velocities tested. The differences were altered when the hydrodynamics and/or charge density inside the bed changed. These differences were independent of the relative position of the two materials. Current signals were strongly affected by local hydrodynamics in a bubbling fluidized bed. The sign of the cumulative charge signals does not necessarily represent the polarity of charge on the particles. The amplitudes of current signal peaks, as well as peak frequencies, increased with increasing the superficial gas velocity.   In the two-dimensional fluidization column, for a dual-tip (two-material) probe, both the mean and standard deviation increased as the in-bed charge density and superficial gas velocity increased, with the difference between the two materials increasing in a similar manner. The average transferred current, tranI , and average induced current, indI , can be 88 related via Eqs. (3.3-3.4) to the average maximum and minimum peaks heights, maxI and minI .   For a dual-tip (one-material) probe, the slopes of cumulative charge signals with time for both tips first increased then decreased as Ug increased. Fluctuations from the lower tip were larger than from the upper one, due to direct exposure to bubbles. For probes with different configurations, the slopes of cumulative charge signals and current peak amplitudes were greater for protruding tips than for retracted tips. Signals from the probes at different axial locations were different. A transition velocity determined from maximum standard deviation of the probe signals was different from the Uc obtained from pressure fluctuations and correlation.   Experiments in the 0.10 m ID column showed that the mean and SD of currents from two materials of the dual-tip (two-material) probe were larger at the centre than near the wall. Transferred currents from the two materials, as well as differences between them, were larger when the probe was inserted vertically rather than horizontally. Statistical analysis showed that the mean and standard deviations of current from both probe tips increased with increasing superficial gas velocity, suggesting that both transfer and induction charges increased. The normalized standard deviation (standard deviation/mean) of the two materials increased as superficial gas velocity increased, and the difference between them was nearly unchanged as the superficial gas velocity increased. From FFT analysis, both materials revealed increased amplitude with increasing superficial gas velocity, confirming that the current signals are strongly influenced by hydrodynamics and, at the same time, there are substantial differences between the two signals.   89 Chapter 4 Decoupling Electrostatic Signals from Gas-Solid Fluidized Beds 4.1 Introduction Electrostatic signals registered by collision probes contain useful information, but are poorly understood. As reviewed in Chapter 1, several studies have analyzed the signals and found that electrostatics and hydrodynamics in fluidized beds are related. However, none of the earlier studies was able to decouple the signals to obtain charge density. The charge density carried on fluidized particles can assist on-line process monitoring and benefit understanding electrostatic phenomena. This chapter introduces signal selection criteria, processing procedures and decoupling methods for the probes tested in Chapters 2 and 3. Signals from these probes are analyzed, and the results from each method are compared with direct measurements in both two- and three-dimensional freely bubbling fluidized beds.   4.2 Signal analysis procedure  Charge/current signals measured by the custom-made probes were processed by first selecting appropriate peaks, followed by calculations using those peaks. Raw electrostatic signals contain noise picked up from the environment and exhibit baseline drift, usually due to weak conductive isolation of the probe. Therefore proper signal conditioning may be needed to pre-treat the signals. Two approaches are helpful to avoid or remove noise from signals: better grounding and shielding during signal acquisition, and filtering the recorded data [83].  4.2.1 Peak (bubble) detection Electrostatic signals are related to bubble and particle movement inside fluidized beds. Selecting the appropriate peaks (bubbles) is important to eliminate irregular signals linked to bubble splitting, coalescence, rising obliquely or missing one of the two sensors. Bubble 90 movement is related to peaks in electrostatic signals. To select bubbles, a selection algorithm was developed in Matlab as described in Figure 4.1. Codes are provided in Appendix C.3.  Signal conditioning(de-noise and/or baseline correction)Electrostatic current signalCalculate statistical values(mean and standard deviations) Peakselection stepFind maximum and minimum peaks Minimum peak among minima, maximum peak among maxima Selected pairs of peaks from two tips 1 & 2Peakdetection stepMaximum tolerable time difference between peaks from tips 1 & 2 ≤  Bmin ≤ time difference between maximum and minimum peaks ≤ Bmax  Ratio of peak amplitudes from tips 1 & 2 YesYesNoEliminatedNoEliminatedNoEliminatedPeak amplitudes > threshold valuesYesNoEliminatedSequence of maximum and minimum peaks corresponding times from tips 1 & 2 Ratio of time lag between maximum peaks to that between minimum peaks from tips 1 & 2Dual-tip (two-material) probe Dual-tip (one-material) probe YesNoNoYes  Figure 4.1 Flowchart of peak detection and selection algorithm adopted for signal processing.  91 After signal conditioning (de-noise and/or baseline correction), average current (Ī) and standard deviation (Isd) can be calculated from the raw signal from the probe. A sudden increase or decrease in the current signal indicates the passage of a bubble. Minimum and maximum peaks are identified wherever there is clear evidence of bubble passage, starting with a distinct peak from a bubble nose and completed by another from the wake. The magnitudes of the minimum and maximum peaks (Imin and Imax) from both tips are also compared with two threshold values, based on the average and standard deviation of the raw signals. Current signals which do not cross the threshold may be due to signal changes not directly linked to bubble passage, or to irregular bubbles reaching the probe. We impose the criteria: max,1 1 ,1( ) sdI i I I       min,1 1 ,1( ) sdI i I I   (4.1) max,2 2 ,2( ) sdI i I I     min,2 2 ,2( ) sdI i I I   (4.2) where I  and sdI  are the mean and standard deviations of the raw signal, and subscripts 1 and 2 denote two tips.  Extremes among the minima and maxima are identified to ensure that only one minimum peak and one maximum peak exist within each signal segment, representing a single bubble. The time difference between adjacent maximum and minimum peaks from a tip represents the time for a bubble to pass the probe, and is thus related to bubble size. We require max max,1 min,1 min( ) ( )B Bt j t j      (4.3) max max,2 min,2 min( ) ( )B Bt j t j      (4.4) where Bmax and Bmin represent the times for maximum and minimum allowable bubbles to pass the probe. These criteria (Eqs. (4.1-4)) define the peaks caused by passing bubbles.   4.2.2 Peak (bubble) selection Signals from a second tip (electrode) of the probe were then used together with the first tip signals to select appropriate bubbles. The criteria depend on how the two tips are configured.  92 4.2.2.1 Dual-tip (two-material) probe For a dual-tip (two-material) probe with two side by side tips, the current peaks, from the two different materials should appear at the same time to ensure that the bubble contacted the two tips simultaneously, so that:  ,1 ,2( ) ( )peak peakt j t j   (4.5) where  is the maximum tolerable time difference between current peaks from the two materials, and subscripts 1 and 2 denote the two tips.   Also, the ratio of the current peaks from two tips is a function of the difference between the two materials. A smaller ratio means a larger difference between currents from two materials. The amplitude from TiN and Ni tips differs and needs to be tested for each case. For example, if the amplitudes of current peaks from TiN were smaller than that from Ni,  2,1( )( )peakpeakI jI j   (4.6) where  and  are the chosen lower and upper boundaries of the ratio.   4.2.2.2 Dual-tip (one-material) probe For a dual-tip (one-material) probe with two tips vertically aligned, signals were selected if they satisfied the following criteria. The maximum and minimum peaks corresponding times (tmax,1, tmin,1, tmax,2, tmin,2) should follow the sequence tmax,2< tmax,1< tmin,2< tmin,1, where subscripts 1 and 2 denote the upper and lower tips, separated by a distance much smaller than the bubble size. A ratio of time lag between maximum peaks from the two tips to the time shift between minimum peaks from the two tips, is required to be within the interval of 0.8−1.2, allowing a reasonable range of deviations for bubbles passing the two tips vertically, based on criteria used in signal analysis of optical fiber probes [129]. Hence we require max,2 max,1 min,2 min,1t t t t   max,1 max,2min,1 min,20.8 1.2t tt t   (4.7) All pairs of peaks that satisfied these criteria are selected and used for calculation. 93 4.3 Decoupling methods Charge density, as well as bubble rise velocity, can be calculated by different decoupling methods using amplitudes and corresponding times of selected pairs of peaks.  4.3.1 Charge transfer and induction model  The charge transfer and induction model developed by Park et al. [20] and Chen et al. [21] simulated the transferred and induced charges received by a collision probe due to a single bubble passing in a fluidized bed. The following assumptions were made:  a. The total charge/current received by the probe consists of two components: charge induction and charge transfer. For highly charged particles, charge separation is much smaller than charge transfer and is neglected. b. Each bubble is spherical, with its diameter remaining constant as the bubble rises at a constant velocity after injection.  c. Only one bubble contacts the probe at a time. d. No particles are inside the bubble, and charges are negligible inside bubbles. e. The charge distribution is obtained by superposition of two simple components: a uniformly charged bed with specific charge qm and a charged spherical ball with uniform charge density -qm. f. The velocities of particles in the nose and wake regions are related to the bubble rise velocity.  As a bubble with surrounding charged particles passes a collision probe, an image charge is induced. For example, if a bubble surrounded by negatively charged particles passes the probe, a negative charge is registered by the conductive probe tip, as shown in Figure 4.2. The negative charge reaches a minimum (Cmin) when the probe is inside the bubble. As shown in Section 3.3.1, Itran and Iind, which represent the transferred and induced currents when the bubble nose/wake makes contact with the probe, can be derived from the maximum and minimum peaks via Eqs. (3.3) and (3.4). It is assumed that the charge density of particles surrounding bubbles is the same, regardless of the bubble size, shape and movement at a particular location inside the bed.  94 Electrostatic probeBubblenoseBubble wakeBubblecentreQttICharge transferImaxIminCmin Figure 4.2 Theoretical charge and current signals received from collision probe when a single bubble passed by.  4.3.2 Dual-tip (two-material) probe 4.3.2.1 Transfer-induction method (Solving for UB and qm together) The transfer and induction currents are both functions of particle charge density and hydrodynamic properties in the fluidized bed, such as bubble diameter and rise velocity. They can be decoupled by solving the simultaneous equations for the transfer and induction parts: ( , ( ));tran m B BI f q U D      ( , ( ))ind m B BI g q U D  (4.8)  For 2t2m probe, the transfer current equation (Eq. (2.5)) for two materials can be obtained by calibrating the probe in a charged particulate flow system (Section 2.5.1). The solid flow rate (Ws) is related to average particle velocity (Vp) by Eq. (2.10) with the particle velocity in the bubble nose and wake regions approximated by the bubble rise velocity and the voidage near 95 the bubble by mf, the voidage at minimum fluidization. The induced current could be related to charge density and bubble rise velocity by Eq. (2.12).  4.3.2.2 Two transfer currents/current peaks methods (Solving for UB and qm together) Different signals, generated from the two tips, could be used as well if calibration equations for the two tips are known, so that qm and UB can be estimated by solving the two equations for the tips together. For a dual-tip (two-material) probe, the following two methods could be used based on the transfer and total current for the two materials.   (a) Two transfer currents method: The two materials produce different transfer currents. Therefore, the signal can be decoupled by simultaneous solving two transfer equations.  (b) Two current peaks method: The total currents from the two materials differ. The total current is a sum of the transfer and induction parts:  ( , ( ))total tran ind m B BI I I h q U D   (4.9) Application of Eq. (4.9) to each of the two materials allows the current signal to be decoupled. Equations of this form corresponding to the current peaks for two materials were obtained (Eq. (2.13)) by calibrating the probe with different particle charge density (qm) and bubble rise velocity (UB) for single bubble injection and free bubbling experiments (Section 2.5.4).  4.3.2.3 Time-difference method (Solving for UB first, then qm) The time interval between adjacent maximum and minimum peaks in the current signal represents the time for a single bubble to pass the probe. Since the probe is much smaller than the bubble, the distance through which the bubble moves is related to the bubble diameter.  BBDU   (4.10)  The bubble diameter is related to its rise velocity for isolated bubbles [130, 131] by 96 0.7110.597B g mf BB g mf BU U U gDU U U gD     three-dimensional fluidized bed (4.11) two-dimensional fluidized bed From Eqs. (4.10) and (4.11), the bubble rise velocity UB and diameter DB can be estimated first, after which qm can be solved by inserting UB into the transfer, induction or total current relationships, Eqs. (4.8) or (4.9).  4.3.3 Dual-tip (one-material) probe Instead of using empirical correlation of Eq. (4.11), one could measure bubble rise velocity, then particle charge density could be estimated by inserting that rise velocity into Eq. (4.9). Figure 4.3 shows the probe with two tips separated vertically by a known distance, z. When a bubble passes, each tip registers a current signal, with maximum and minimum peaks corresponding to arrival of the bubble nose and wake. The time lag between the peaks from the two tips, t, can be used to determine the bubble rise velocity, and the bubble size can be estimated from the time difference  between the times corresponding to arrival of the bubble nose and wake from either tip, as shown in Eq. (4.10). Note that the measured bubble size is the pierced chord length of the bubble. 22 mm Itmax,1tmin,1 tmax,2tmin,2Ibubble nosebubble nosebubble wake11 mm upper tiplower tip11 mm  Figure 4.3 Principle of 2t1m probe (F configuration) to determine bubble rise velocity and size.  BzU t   (4.12) 97 4.3.3.1 Two-tips cross-correlation (time-average) method  The time lag, t, can be obtained by cross-correlation or Eq. (4.13). The cross-correlation function of time-series signals from two tips is: 1 2 1 201( ) lim ( ) ( )HI I TR t I t I t t dtH    (4.13) where t is the time shift for the cross-correlation (coefficient) function to reach a maximum, indicating the time delay where two signals are best aligned, corresponding to the average time for bubble passage between the two tips. For a known t, UB and qm can then be obtained from Eqs. (4.12) and (4.9).  4.3.3.2 Two-tips peak-times (dynamic) method  The other way to obtain the time lag t is based on the time difference between peaks from the upper and lower tips. Selected pairs of peaks representing bubbles were used to calculate the time lag, t, and time difference, , by max,1 max,2 min,1 min,2( ) ( )2t t t tt      (4.14) min,1 max,1 min,2 max,2( ) ( )2t t t t      (4.15) Then UB and DB were estimated by Eqs. (4.12) and (4.10), respectively. While qm was calculated from Eq. (4.9) based on the current peaks from each tip.   4.3.3.3 One-tip time-difference (dynamic) method  Similar to the dual-tip (two-material) probe, signals from one tip could also be used to estimate bubble rise velocities and particle charge densities, as explained in Section 4.3.2.3.   The two tips cross-correlation method gives time-average from signals recorded over a period of time, while the two tips peak-times and one-tip time-difference methods provide dynamic values for specific times. For a probe with F configuration (both tips protruding), qm from the lower tip is preferred because bubbles reaching the lower tip are affected less by the 98 probe, whereas for only one tip protruding, results are applied from that tip. Error bars were generated by applying the Student t distribution to find 90% confidence intervals.  4.4 Results and discussion 4.4.1 Signal processing 4.4.1.1 Effects of bubble selection criteria Figure 4.4 shows raw data, peak-detected data and peak-selected data from the 2t2mb probe (see Table 2.1) after applying the bubble selection criteria to a same 20 s period of time signals from the two-dimensional bubbling bed. The algorithm picks peaks (bubbles) based on the chosen criteria. The Peak-detected data show detected peaks after applying Eqs. (4.1-4) and the peak-selected data represent selected peaks after applying Eqs. (4.5) and (4.6).   99              Figure 4.4 (a) Raw data, (b) peak-detected data and (c) peak-selected data, generated from 2t2mb probe in two-dimensional bubbling fluidized bed of glass beads.  To choose the proper criteria and improve the accuracy of the bubble selection algorithm, as well as calculation results, the effect of each parameter on selected bubble numbers was investigated while keeping other parameters constant, as shown in Figure 4.5. A 40 s period of time-series signal with a total of 70 bubbles from the probe was adopted for analysis.  100 The total current is the sum of the transfer and induction components; the mean current represents the net transferred current, and the standard deviation represents the net induction current. The  coefficient (see Eqs. (4.1) and (4.2)) affects the threshold that defines the amplitude of a peak. The number of bubbles selected decreased with increasing , decreasing sharply beyond  =1. Thus,  was chosen to be 1 for subsequent analyses. The maximum tolerable time difference between current peaks from the two materials (tips),  (see Eq. (4.5)), levels off beyond 0.01 s. This seems to be a sensitive value for . Also  was smaller than 0.01 based on comparison of the probe signal with synchronized video images. The upper and lower boundaries of ratio of current peaks from the two materials (tips),  and  (see Eq. (4.6)) are the most sensitive factors. The number of selected bubbles increases when the upper boundary  is increased, and becomes stable beyond  =1. On the other hand, the number of selected bubbles decreases when the lower boundary  increases. A larger ratio of current peaks from the two materials (tips), means a larger difference between the signals from the two materials (tips), causing more bubbles to be selected. However, an extremely large or small ratio implies a non-vertical bubble trajectory, for example acceptance of bubbles moving in such a way that most particles only collide with one tip. Such cases were observed in the videos. Differences in current signals from the two tips mainly arose from the transfer components, so  was chosen as 1 based on the average ratio of induction parts.  was taken as 0.4 based on analyzing the signals from the fluidized bed and dropping tests of charged particles (Sections 2.5.4 and 2.5.1). Time for maximum and minimum allowable bubbles to pass the probe, Bmax and Bmin, are related to minimum and maximum allowable bubble sizes (DBmin and DBmax) by Eqs. (4.3) and (4.4). The number of selected bubbles slightly decreased with increasing DBmin and increased with increasing DBmax and became stable after DBmax= 0.2 m. Too small a bubble size may cause probe signals too weak to be undetected, and the maximum bubble size needs to be smaller than the column width (0.3 m). Also from visual observations of recorded videos, 0.01 and 0.2 m were chosen as the DBmin and DBmax for current analysis.    101                             Figure 4.5 Effects of different parameters on number of selected bubbles from a 40 s period of time-series signals (70 bubbles in total) in bubble selection algorithm of signal processing for 2t2mb probe in two-dimensional fluidized bed.  0 10 20 30 40 500510152025303540Selected number of bubblesMinimum allowable bubble size (DB,min), mm     0.0 0.1 0.2 0.3 0.4 0.50510152025303540Selected number of bubblesMaximum allowable bubble size (DB,max), m 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.00510152025303540Selected number of bubblesLower boundary of the ratio of two current peaks () 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3051015202530Selected number of bubblesUpper boundary of the ratio of two current peaks () - .01 0.00 0.01 0.02 0.03 0.040510152025303540Selected number of bubblesMaximum tolerable time difference between current peaks (), s 1 2051035Selected number of bubblesCoeffici nt ) 102 4.4.1.2 Effect of decoupling methods  Different decoupling methods identified in Section 4.3.2 were applied to the current signal from 2t2mb probe (see Table 2.1) in the two-dimensional fluidized column. Each tip (Ni or TiN) of the probe can be regarded as a 1t1m probe. Table 4.1 provides decoupled qm and UB obtained by different methods for a signal segment of ~1 s duration (represents one bubble) from the probe in the bubbling two-dimensional bed with different superficial gas velocities. Results calculated by the transfer-induction method (T1) from each tip were represented as T1-Ni and T1-TiN, respectively. In the time-difference method (T2), UB was first obtained from the time interval between adjacent peaks, and then inserted into calibration equations to calculate qm. Three equations, representing induced, transfer and peak currents, were used to decouple the signal for each material. In total six qm results were obtained based on six equations for the two materials (tips), referred to as T2-Ni-ind, T2-TiN-ind, T2-Ni-tran, T2-TiN-tran, T2-Ni-peak and T2-TiN-peak. For two tips, the two transfer currents method (TT1) and two current peaks method (TT2) were applied to decouple the signal.  The particle charge densities measured by the Faraday cup (FC) serve as a reference. qm decoupled by the T2-Ni-ind, T2-TiN-ind, T2-Ni-peak, T2-TiN-peak and TT2 methods have the same order of magnitudes and similar trends as those measured by a Faraday cup. UB decoupled by T2 and TT2 methods also have the same order of magnitudes and trends as the results from video analysis. UB decoupled by method TT2 was larger than that decoupled by the T2 method, and a similar trend was found in bubble nose velocities (UB,nose) and bubble wake velocities (UB,wake) as velocities derived from video analysis.   The T2-Ni-peak, T2-TiN-peak and TT2 methods were selected for further evaluation because these methods do not require separation of transfer and induction components in the signals. The signals from the probes represent total charge or current. To separate these two components, one needs to identify each segment of signal caused by every bubble movement; however, peaks are often affected by bubble interaction and splitting [132]. The model assumes that bubbles have a constant velocity when contacting the probe; however videos show that the interaction between bubble and the probe may cause the wake to rise more quickly than the nose, leading to inaccurate results when seeking to separate the transfer and 103 induction parts of the signal. Hence, if one wants to separate the transfer and induction parts, a combined collision and induction probe may be a better option. The T2 method first estimates UB when decoupling the signal, while the TT2 method uses additional information provided by the probe to deconvolute the signal. The decoupled results require accurate calibration equations and proper signal processing to eliminate unwanted bubbles. Two current peaks method (TT2) likely gave closer charge density results compared with those from Faraday cup measurement. Time-difference method (T2) showed less scatter results because of using bubble rise velocity correlation.  104 Table 4.1 Decoupled charge densities and bubble rise velocities based on different decoupling methods for one segment of signals (~1 s duration) for two-dimensional bubbling fluidized bed at different superficial gas velocities, Ug.  Decoupled charge densities, -qm, C/kg Ug, m/s T1-Ni T1-TiN T2-Ni-ind T2-TiN-ind T2-Ni-tran T2-TiN-tran T2-Ni-peak T2-TiN-peak TT1 TT2 FC 0.29 0.46 0.41 1.46 1.09 0.33 0.26 2.36 2.02 -0.13 1.82 2.02 0.33 0.77 0.82 2.06 1.31 0.85 0.93 3.41 2.93 0.26 2.62 3.97 0.36 1.12 1.23 2.69 1.88 1.13 1.30 4.30 4.11 0.39 3.88 4.27 0.39 1.86 8.29 4.58 3.09 2.51 3.19 7.24 7.00 0.65 6.43 4.82  Decoupled bubble rise velocities, UB, m/s Ug, m/s T1-Ni T1-TiN T2-Ni-ind T2-TiN-ind T2-Ni-tran T2-TiN-tran T2-Ni-peak T2-TiN-peak TT1 TT2 Video UB,nose UB,wake 0.29 1.24 1.06 0.43 0.43 0.43 0.43 0.43 0.43 N/A* 0.70 0.40 0.40 0.33 1.34 0.84 0.55 0.55 0.55 0.55 0.55 0.55 1.81 0.81 0.63 0.67 0.36 1.56 1.04 0.71 0.71 0.71 0.71 0.71 0.71 2.22 0.81 0.63 0.91 0.39 1.73 0.31 0.76 0.76 0.76 0.76 0.76 0.76 3.32 0.88 0.59 1.05 *Complex number when solving equations. T1: Transfer-induction method (Section 4.3.2.1) TT1: Two transfer currents method (Section 4.3.2.2(a)) TT2: Two current peaks method (Section 4.3.2.2(b)) T2: Time-difference method (Section 4.3.2.3) FC: Faraday cup ind: Induction tran: Transfer   105 4.4.2 Results from two-dimensional fluidization column 4.4.2.1 Dual-tip (two-material) probe with 500-600 m GB Gauss’ law may be used to give an upper limit on the particle charge that can accumulate before ionization of the surrounding fluid occurs [3]. Charge accumulation on the surface of a dielectric particle is related to the electric field at the particle surface. For a spherical particle in a mono-ionized field, a saturation charge is reached due to a DC corona discharge when the attractive field due to the field distortion equals the repulsive field due to the charge on the particle. By equating the force due to the distorted field and the force of repulsion from the charged particle, the maximum surface charge density on the particle, Qs (C/m2), is given [11, 30, 133] by  0sQ E p  (4.16) where 3 / ( 2),r rp     and r is the relative permittivity of the particle. E is the electrical field around the particle. For an air breakdown potential of around 3000 kV/m and 0, the permittivity of vacuum, of 8.85E−12 F/m, the maximum charge density in our experiments is estimated to be 1.9~2.3E−4 C/kg, with a relative permittivity of 3 [11], a particle density of 2500 kg/m3 and a particle size range of 500-600 m.  Initially, the two transfer current method (Section 4.3.2.2(a)) was applied with the calibration equations (Eqs. (2.5)-(2.7)) for transfer currents developed from gas-solid homogeneous flow. In addition, min max( ) / 2I I  from the two materials were assumed to correspond to transfer currents to estimate the charge density. However, the resulting values were an order of magnitude smaller than the charge densities measured by the Faraday cup. This suggests that the calibrations developed in homogeneous gas-solid flow cannot be applied directly to the freely bubbling fluidized bed because of the distributions of particle collision velocity and solid concentration associated with bubble movement.   As an alternative, the two current peaks method (Section 4.3.2.2(b)) was used to estimate the charge density and bubble rise velocity in freely bubbling beds with Eqs. (2.13)-(2.15), based on current peaks and measured charge density and bubble rise velocity for single bubbles. 106 The data used for calculation were from freely bubbling experiments without synchronization. Current peaks from the two materials were chosen from time-series data from the probe by applying the selection criteria and parameters described above (=1, =0.01 s, DBmin=10 mm, DBmax=200 mm, =1 and =0.4). For each pair of selected current peaks corresponding to passage of a suitable bubble, the charge density and bubble rise velocity were then estimated by solving Eq. (2.13) for the two materials simultaneously, with i, i and i  from Eqs. (2.14) and (2.15). At least 20 bubbles were selected for each superficial gas velocity and analyzed to obtain an average charge density and bubble rise velocity. Error bars were generated by applying the Student t distribution to find 95% confidence intervals. Bubbles are selected in a given size range, rising vertically aligned with the probe, and not involved in coalescence or splitting. Charge densities from the Faraday cup are averages of at least three determinations. 0123456789(a)-qm, charge density in the bed, C/kgSuperficial gas velocity, m/s Faraday cup probe0.30 0.40 0.50 0.600123450.29 0.390.360.33-qm, charge density in the bed, C/kgSuperficial gas velocity, m/s Faraday cup probe(b)  Figure 4.6 Comparisons of charge density from 2t2mb probe and Faraday cup measurements at different Ug in freely bubbling two-dimensional fluidized bed: (a) Static bed height 0.30 m with probe 0.22 m above distributor; (b) Static bed height 0.50 m with probe 0.42 m above distributor.   Figure 4.6 compares qm derived from Eq. (2.13) with those measured by the Faraday cup based on particles withdrawn through the sampling port. qm from both the Faraday cup and probe are of order 1E−6 C/kg for two static bed heights, similar to values reported in the literature [40, 53] and much less than the maximum qm estimated by Eq. (4.16). As expected, the magnitude of the charge density increases with increasing Ug for both direct and indirect measurements. Differences in charge densities from direct (Faraday cup) and indirect (probe) 107 measurements for Ug= 0.6 m/s are larger for case a (lower static bed height) than for case b. Note that the calibration equations were obtained in situations where bubbles were passing the probe one by one. However, in case a, with the probe closer to the distributor, videos showed that bubbles coalesced frequently in the lower section of the bed, likely making bubbles rise less vertically. The same factor led to a lower proportion of bubbles meeting all the discrimination criteria in case a than in case b.  0.00.20.40.60.81.01.21.4 Decoupled bubble rise velocity, m/s by analyzing video by probeSuperficial gas velocity, m/s0.30 0.40 0.50 0.60(a)0.00.20.40.60.81.01.2.29(b)Decoupled bubble rise velocity, m/sSuperfici l gas velocity, m/s by analyzing video by probe.33 .36 .39  Figure 4.7 Comparisons of bubble rise velocity from 2t2mb probe and video measurements under different Ug in freely bubbling two-dimensional fluidized bed: (a) Static bed height 0.30 m with probe 0.22 m above distributor; (b) Static bed height 0.50 m with probe 0.42 m above distributor.   Figure 4.7 compares the estimated UB from decoupling electrostatic charge signals with those obtained by analyzing videos of the two-dimensional bed at different Ug. The calculated UB from the probe signals are similar in magnitude to those from the video images. Both methods indicate that, as expected, UB increased with increasing Ug. The relative magnitudes of UB obtained from the probe and video analysis are closer than the charge density results in Figure 4.6, with most error bars overlapping.   The charge density (qm) and bubble rise velocity (UB) were also estimated based the time-difference method (see Section 4.3.2.3). As shown in Figure 4.8, both calculated qm and UB varied in a consistent manner, with the same order of magnitude as those from direct measurements (Faraday cup for qm and video analysis for UB). In Figure 4.8(b), decoupled UB from the time-difference method are closer to bubble nose velocity (UB,nose) from video analysis and deviate less than those calculated by the two current peaks method which are 108 nearer to the (UB,wake) bubble wake velocity, especially at higher superficial gas velocities. Note that Eqs. (2.14) and (2.15) were obtained based on UB,wake and the interaction of bubble and the probe may result in an unequal bubble rise velocities at the bubble nose and wake. Also it was found that UB by the time-difference method were almost the same when derived from the Ni or the TiN tip at each superficial gas velocity.  0123456(a) by Faraday cup by probe use T2-Ni by probe use T2-TiN by probe use TT2-qm, charge density in the bed, C/kgSuperficial gas velocity, m/s0.29 0.33 0.36 0.39.00.20.40.60.81.01.2 UB, wake by analyzing video UB, nose by analyzing video by probe use T2-Ni by probe use T2-TiN by probe use TT2UB, bubble rise velocity, m/sSuperficial gas velocity, m/s0.390.360.330.29(b)  Figure 4.8 Comparison of decoupled (a) charge densities; (b) bubble rise velocities from 2t2mb probe by different decoupling methods in two-dimensional bubbling bed of 500-600 m GB.   In general, the time-difference method could be used to analyze signals from the conventional collision probe of single electrode. A second electrode, however, is needed to select appropriate signals or bubbles before the decoupling method could be applied, based on the empirical relationship between the bubble rise velocity and bubble size. The two current peaks method is based on the difference between the signals from two tips of different materials.   4.4.2.2 Dual-tip (two-material) probe with 106-212 m GB The criteria with =1, =0.01 s, =2.5 and =1, DBmin=0.01 m and DBmax=0.2 m were next applied to data for this case. The upper and lower ratios of current peaks from the two materials were changed because of the variance in relative magnitudes of Ni and TiN tips for 106-212 m GB. The charge density (qm) and bubble rise velocity (UB) are estimated by the previously developed decoupling methods (see Section 4.3.2).   109 Figure 4.9a compares the average charge densities from the probe with those measured by the Faraday cup, based on particles withdrawn through the sampling port. Comparison of Figures 4.8a and 4.9a indicated that the charge densities from the probe and Faraday cup are of the same order of magnitude and vary in a similar manner with changing Ug. In Figure 4.9b, UB decoupled from the two current peaks method (TT2) for 106-212 m GB failed to predict the trend of UB with changing Ug, but the predicted UB values from the time-difference method are closer to those obtained by video analysis. This figure also shows that UB from the latter method are smaller than from the former, consistent with results in Figure 4.8b, because the bubble nose slowed down as the probe was approached, while the calibration equations for the two current peaks method were based on the wake velocity.  0.00.51.01.52.02.53.0(a)-qm, charge density in the bed, C/kg by Faraday cup by probe use T2-Ni by probe use T2-TiN by probe use TT20.07 0.09 0.11 0.14Superficial gas velocity, m/s 0.00.20.40.60.81.01.21.4(b)Decoupled bubble rise velocity, m/s0. . 0.1 0.14Su rfi i l l city, m/s UB, wake by analyzing video images UB, nose by analyzing video images by probe  Ni by probe  2-TiN by probe use TT2  Figure 4.9 Comparison of decoupled (a) particle charge densities; (b) bubble rise velocities from 2t2mb probe by different decoupling methods in two-dimensional bubbling bed of 106-212 m GB.   For particles with narrow size distribution but different mean size, probe signals and charge density results were compared. 500-600 m GB generated more charge density than 106-212 m GB (see Figures 4.8 and 4.9) and the charge signals and current peak amplitudes from the probe were larger for particles with larger mean size (see Figures 3.3 and 3.6). This is because the larger particle size generated larger momentum during particle-particle and particle-wall collisions [92].   110 For the 2t2m probe, the relative magnitude of signals from the two tips changes with operating conditions, e.g. particle velocity, flow rate and collision angle, as shown in Sections 2.5.1 and 2.5.3, and may also change with particle size from results presented in Section 3.3.1 and this section. Since the relative magnitudes of signals from the two materials are used to screen data in signal processing, particles of different properties (e.g. mean size and size distribution) may cause difficulty in signal processing. This suggests a 2t2m probe may be unsuitable for particles of a wide size distribution.   4.4.2.3 Dual-tip (one-material) probe with 500-600 m GB Probe signals were treated with criteria outlined in Section 4.2: coefficient  was set to 0.5 for the maximum peaks and 1 for the minimum peaks, because of lower amplitudes of maximum peaks in signals from this work, minimum and maximum allowable bubble sizes (DBmin and DBmax) were chosen as 0.01 m and 0.2 m, consistent with 2t2m probe. At each superficial gas velocity, about 4-15% pairs of peaks satisfied these criteria for glass beads and 5-22% for polyethylene particles. These pairs of peaks remained after applying the above criteria were then analyzed by applying the decoupling methods and probe calibration equation to extract charge density and bubble properties. In view of the random movement of bubbles in the fluidized bed, it is reasonable to see that only a small percentage of bubbles passed the local probe location vertically without splitting/coalescing.   Figure 4.10 shows the bubble rise velocity and diameter for bubbles analyzed by two tips peak-times. Although the bubble rise velocities and sizes covered broad ranges, the overall trend for individual bubbles followed Eq. (4.11), and the mean values were close to predictions from Eq. (4.11). As expected, both the mean bubble rise velocity and bubble size increased as Ug increased. The bubble pierced chord length changed during passing the probe because of obstruction by the tips. When ratios of time difference () for lower and upper tips were calculated for the selected bubbles at different Ug, the results were in the range of 0.88 to 1.2, with an average of 1.03 and a standard deviation of 0.07. This indicates that the bubble size (pierced chord length) was indeed influenced by the intrusive probe tips (F 111 configuration) and altered within a reasonable range when passing the two vertical tips based on the chosen criteria.  0.00 0.05 0.10 0.15 0.20 0.250.00.20.40.60.81.01.21.40.00 0.05 0.10 0.15 0.20 0.250.00.20.40.60.81.01.21.40.00 0.05 0.10 0.15 0.20 0.250.00.20.40.60.81.01.21.4selected bubbles from exp oneselected bubbles from exp two Eq. (4.12)UB, m/sUB, m/sUB, m/sUg-Umf=0.14 m/sUg-Umf=0.08 m/sDB, mUg-Umf=0.02 m/s  (*Experiments one and two were conducted on different days)  Figure 4.10 Decoupled bubble size and bubble rise velocities from 2t1mfF probe by two tips peak-times method at different Ug in freely bubbling two-dimensional fluidized bed of 500-600 m GB.  Signal analysis results from the 2t1mfF probe (see Table 2.1) are shown in Figure 4.11, with UB derived from all three methods compared with those derived from video images. UB from the probe were similar in magnitude to those from the video images. All results indicate that UB increased with increasing superficial gas velocity. The one-tip time-difference method showed less scatter because of the use of bubble rise velocity correlation (Eq. (4.11)). Also UB by the one-tip time-difference method was similar when derived from either the upper or lower probe tip. Average relative absolute discrepancies between the results from the direct (video analysis) and indirect measurements were 13% for the two tips cross-correlation method, 12% for the two tips peak-times method and 18% for the one-tip time-difference method. The two tips cross-correlation and two tips peak-times methods likely resulted in smaller relative absolute discrepancies due to the use of directly measured UB values.  112 Charge density, qm, from the 2t1mfF probe and the Faraday cup for different Ug are compared in Figure 4.11. Results from all three methods are similar in magnitude and consistent with those measured by the Faraday cup. The average relative absolute discrepancies between the probe and Faraday cup measurement were 27% for the two tips cross-correlation method, 29% for the two tips peak-times method and 30% for the one-tip time-difference method.  024681012(Exp 1)0.140.080.02  Faraday cup  Two tips cross-correlation Two tips peak-times  One tip time-difference-qm, charge density in the bed, C/kgUg-Umf, m/s0.00.20.40.60.81.01.21.4(Exp 1)UB, bubble rise velocity, m/s from video  Two tips cross-correlation Two tips peak-times  One tip time-difference0.02 0.08Ug-Umf, m/s0.14 012345678(Exp 2) Faraday cup  Two tips cross-correlation Two tips peak-times  One tip time-difference-qm, charge density in the bed, C/kg0.02 0.08Ug-Umf, m/s0.140.00.20.40.60.81.01.21.4from video Two s cross-correlation Two tips peak-times One tip time-differenceUB, bubble rise velocity, m/s0.02 0.08g-Uf, m/s0.14(Exp 2) (*Experiments one and two were conducted on different days) Figure 4.11 Comparison of charge densities and bubble rise velocities from 2t1mfF probe by different decoupling methods with Faraday cup and video measurements at different Ug in freely bubbling two-dimensional fluidized bed of 500-600 m GB.   Results from the probe with different configurations (shown in Figure 2.2) are compared in Table 4.2. The Γ probe (only upper tip protruding) was intended to provide smaller errors 113 than the F probe (two tips protruding). Decoupled UB from the F probe by the two tips peak-times method seems to be smaller than from the other two methods, possibly because the two tips were not well aligned. The one-tip time-difference methods produced UB values closer to direct values from video recordings than the other two methods.  Table 4.2 Estimated charge densities and bubble rise velocities from 2t1m probe with different configurations by different decoupling methods at different Ug in freely bubbling two-dimensional fluidized bed of 500-600 m GB.  Two tips cross-correlation method (Section 4.3.3.1)  F: both tips protruding : only upper tip protruding Ug−Umf, m/s Faraday cup Charge densities, −qm, C/kg 0.02 3.440.18 1.66 1.03 0.08 4.760.27 3.61 3.78 0.14 4.930.74 6.36 6.79 Ug−Umf, m/s Video image Bubble rise velocities, UB, m/s 0.02 0.550.08 0.53 0.59 0.08 0.700.07 0.70 0.62 0.14 1.060.12 0.80 0.69 Two tips peak-times method (Section 4.3.3.2)  F: both tips protruding : only upper tip protruding Ug−Umf, m/s Faraday cup Charge densities, −qm, C/kg 0.02 3.440.18 3.200.63 2.400.60 0.08 4.760.27 6.270.93 5.581.08 0.14 4.930.74 9.412.10 6.071.09 Ug−Umf, m/s Video image Bubble rise velocities, UB, m/s 0.02 0.550.08 0.450.22 0.390.22 0.08 0.700.07 0.400.08 0.250.05 0.14 1.060.12 0.480.08 0.630.09 One-tip time-difference method (Section 4.3.3.3)  F: both tips protruding : only upper tip protruding Ug−Umf, m/s Faraday cup Charge densities, −qm, C/kg 0.02 3.440.18 1.850.41 1.320.13 0.08 4.760.27 3.230.38 2.240.25 0.14 4.930.74 5.591.26 4.360.48 Ug−Umf, m/s Video image Bubble rise velocities, UB, m/s 0.02 0.550.08 0.550.10 0.520.04 0.08 0.700.07 0.720.03 0.740.04 0.14 1.060.12 0.780.03 0.750.03  114 4.4.2.4 Dual-tip (one-material) probe with 710-850 m PE Bubble rise velocities and sizes derived from the two tips peak-times method are plotted in Figure 4.12. The overall trend of individual bubbles is consistent with Eq. (4.11), but the mean values are smaller than predicted by this equation. When ratios of the time difference () for the lower and upper tips were calculated for the selected bubbles at different Ug, the results were in the range of 0.85 to 1.2 with an average of 1.0 and a standard deviation of 0.07. Again the selected bubble pierced chord length varied within a reasonable range while passing the probe.  0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.200.00.20.40.60.81.01.20.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.200.00.20.40.60.81.01.20.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 0.200.00.20.40.60.81.01.2 Selected bubbles Eq. (4.11)UB, m/sUB, m/sUB, m/sDB, mUg-Umf=0.01 m/sUg-Umf=0.04m/sUg-Umf=0.09m/s Figure 4.12 Decoupled bubble size and bubble rise velocity from 2t1mfF probe by two current peaks tips method at different Ug in freely bubbling two-dimensional fluidized bed of 710-850 m PE.  From Figure 4.12, UB estimated from the two tips peak-times and two tips cross-correlation methods were smaller than when derived from video recordings. UB from the probe are similar in magnitude and changed in a consistent manner with those from the video analysis as Ug varied. The one-tip time-difference method agreed more closely with the video-measured UB than the other two methods, with a relative absolute deviation of 15%. The relative differences between decoupled average UB from the upper and lower tips were less 115 than 1.2% for different Ug, indicating that the decoupled UB was similar when derived from either the upper or lower probe tip.  Based on Eq. (4.16), the maximum possible charge density in our experiments for polyethylene particles is expected to have been 3.3~3.9E−4 C/kg, for a particle relative permittivity of 2.3, particle density of 918 kg/m3 and particle size range of 710−850 m [11, 43]. As shown in Figure 4.13, qm estimated from both methods was well below this maximum charge density and increased with increasing Ug, consistent with the Faraday cup data. Results from the one-tip time-difference method showed smaller error bars and were closer to the results from direct measurement, with a relative absolute error of 45%. This error was ~15% larger than the relative absolute error from the glass beads experiments, likely because of the empirical correlation used in calculating UB, and error propagation when estimating qm by Eq. (2.13).   0510152025 Faraday cup  Two tips cross-correlation Two tips peak-times  One tip time-difference-qm, charge density in the bed, C/kg0.01 0.04 0.09Ug-Umf, m/s0.00.10.20.30.40.50.60.70.8 f om video Two tips cross-correlation Two tips peak-times  One tip time-diff renceUB, bubble rise velocity, m/s.01 0.04Ug-Umf, m/s0.09 Figure 4.13 Comparison of charge densities and bubble rise velocities from 2t1mfF probe by different decoupling methods with Faraday cup and video measurements at different Ug in freely bubbling two-dimensional fluidized bed of 710-850 m PE.   The probe with different configurations was also tested in the freely bubbling fluidized bed of polyethylene particles with the retractable tip protruding, and then retracted. Comparison of signals from cases F, Γ and L, revealed that cumulative charge signals from each tip decreased when it was retracted and was highest when it was the sole protruding tip (upper tip in Γ, lower tip in L), consistent with the glass bead results. Figure 4.14 shows decoupled 116 results from the probe with different configurations. At Ug−Umf = 0.09 m/s, qm and UB estimated from the Γ probe have relative absolute discrepancies of 10% and 16%, respectively, which were closer to the direct measurements than for cases F and L. The one-tip time-difference method again resulted in smaller error bars than for the other two methods. The Γ probe tended to give the most accurate charge density results.  051015202530Two tips peak-times One tip time-difference F probe  probe L probe Faraday cup-qm, charge density in the bed, C/kgTwo tips cross-correlation0.00.20.40.60.81.0  F probe  probe L probe from videoUB, bubble rise velocities, m/sTwo tips cross-correlationTwo tips peak-times One tip time-difference Figure 4.14 Comparison of charge densities and bubble rise velocities from 2t1m probes of different configurations by different decoupling methods, Faraday cup and video measurements in freely bubbling two-dimensional fluidized bed of 710-850 m PE at Ug−Umf = 0.09 m/s.   The discrepancies between results from the probes and direct measurements may arise from the calibration equation (with more experimental data needed to improve the coefficient of determination, R2), signal processing criteria, and errors from direct measurements of qm and UB from the Faraday cup and video records. Most of the results have errors of 10-20% with a few falling between 20 and 30%. Considering the uncertainty and unpredictability of electrostatics, a maximum error of ±30% is regarded as acceptable. The Faraday cup is an open system susceptible to variations in environmental factors such as atmospheric air humidity, electromagnetic noise and additional charging/discharging during sample discharging from the fluidized bed, and this renders it an unreliable tool [14]. Although uncertainty and errors exist in Faraday cup measurement, it is the only available method to measure charge density on dielectric powders. Therefore, the results from the Faraday cup serve as a reference. The ability of the indirect collision probe signal decoupling method to estimate the particle charge density in fluidized beds is encouraging. The probe developed in 117 this study can thus serve as an online tool for monitoring changes in particle charge density, rather than providing accurate charge density values.  4.4.3 Results from three-dimensional fluidization column of ID 0.10 m A 2t2ma probe (see Table 2.1) was inserted vertically from the top of the three-dimensional column. Time-series signals from the probe were examined to select pairs of minimum peaks from the two materials, corresponding to passage of vertically-traveling bubbles, for calculation purpose. The following criteria were applied for data screening (=0.5, =0.03 s, =1 and =0.4, see Section 4.2). Because the probe was vertically inserted in the bed and bubbles were greatly disturbed so that the criteria for bubble size (DBmin and DBmax) in Eqs. (4.3) and (4.4) were not applied here. Coefficient  (in Eqs. (4.1) and (4.2)) and maximum tolerable time difference  (in Eq. (4.5)) were adjusted to have more than 20 data points selected at each velocity. The charge density (qm) and bubble rise velocity (UB) are estimated by solving Eq. (2.13) for the two materials simultaneously, using i, i and i values from Eqs. (2.14) and (2.15).   Table 4.3 compares the charge densities (average values) from Eq. (2.13) with those measured by the Faraday cup, based on particles withdrawn through the sampling port. The charge densities from the Faraday cup and probe methods are of the same order of magnitude and vary in a similar manner with changing superficial gas velocity. The decoupled bubble rise velocities are within 0-1 m/s and increasing, as expected, with increasing superficial gas velocities.  Table 4.3 Average charge densities (qm) measured by Faraday cup (FC) and decoupled from 2t2ma probe, and bubble rise velocities (UB) decoupled from probe at different Ug in 0.10 m ID three-dimensional bubbling bed of 500-600 m GB. Ug, m/s 0.22 0.25 0.28 0.33 0.36 0.39 0.44 -qm, C/kg FC 2.43 2.78 3.81 4.66 5.13 5.68 5.76 Probe 1.11 1.24 1.41 2.21 4.32 4.83 5.71 UB, m/s 0.60 0.61 0.66 0.79 1.25 0.89 1.10  118 A 2t2mb probe (see Table 2.1) probe was inserted 0.10 m above the distributor in the centre of the bed. Current signals at different superficial gas velocities were treated with the same criteria for data screening (=1, =0.01 s, DBmax=0.2 m and DBmin=0.01 m, =1 and =0.4, see Section 4.2) as in Section 4.4.2. The particle charge densities (qm) were estimated by the various analysis methods (see Section 4.3.2) and compared with those measured by Faraday cup in Figure 4.15. The charge densities from the probe are again of the same order of magnitude and vary in a consistent manner with those from the Faraday cup.  0.00.20.40.60.81.01.21.41.60.330.28-qm, charge density in the bed, C/kgSuperficial gas velocity, m/s Faraday cup by probe use T2-Ni by probe use T2-TiN by probe use TT20.25  Figure 4.15 Comparison of decoupled charge densities from 2t2mb probe by different decoupling methods and Faraday cup in 0.10 m ID three-dimensional bubbling bed of 500-600 m GB.  4.4.4 Results from three-dimensional fluidization column of ID 0.30 m 4.4.4.1 Dual-tip (two-material) probe with 420-590 m GB Comparison of the power spectra of raw and noise signals estimated from static (no air flow) experiments indicated that the effect of noise was negligible. Dynamic signals from the 2t2mb probe were screened using the same criteria parameters (=1, =0.01 s, DBmin=0.01 mm, DBmax=0.2 m, =1 and =0.4, see Section 4.2) as in Sections 4.4.2 and 4.4.3. Glass beads with a size range of 420-590 m were used in these experiments.  Figure 4.16 compares qm calculated from the probe by different methods with those measured by the Faraday cup based on particles withdrawn through the sampling port in the 119 three-dimensional column. These two sets of results follow a similar trend as the superficial gas velocity changes. Comparison of Figures 4.6a and 4.16a show that the charge densities in the three-dimensional column tended to be smaller than those from the two-dimensional column. This might be related to the higher moisture content and temperature of the air supplied to the three-dimensional column. The calculated UB, as shown in Figure 4.16b, increased with increasing superficial gas velocity.  01234(a)0.370.320.27-qm, charge density in the bed, C/kgSuperficial gas velocity, m/s by Faraday cup by probe use T2-Ni by probe use T2-TiN by probe use TT2 0.00.20.40.60.81.0(b)UB, bubble rise velocity, m/sSuperficial gas velocity, m/s by probe use T2-Ni by probe use T2-TiN by probe use TT20.27 0.32 0.37 Figure 4.16 Comparison of decoupled (a) charge densities; (b) bubble rise velocities from 2t2mb probe by T2-Ni, T2-TiN and TT2 methods in 0.30 m ID three-dimensional bubbling bed of 420-590 m GB.   4.4.4.2 Dual-tip (one-material) probe with original PE-I (100-1500 m) Dynamic signals from the 2t1m probes were screened using the same criteria parameters as in Section 4.4.2. Figure 4.17 compares qm calculated from the probe by different methods (see Section 4.3.3) with those measured by the Faraday cup based on particles withdrawn through the sampling port (located 0.25 m above the distributor) in the three-dimensional column. Results from these two methods followed a similar trend as the superficial gas velocity changed. The calculated UB, shown in Figure 4.17, increased with increasing superficial gas velocity. The effect of intrusive probe tips on the derived bubble size was also studied. Ratios of time differences () for the lower and upper tips calculated for the selected bubbles at different Ug fell into the range of 0.86 to 1.2, with an average of 1.0 and a standard deviation of 0.06. Again this indicates that the pierced chord lengths of selected bubbles varied within a limited range while passing the probe. The relative differences 120 between the decoupled average UB from the upper and lower tips were less than 1.3% for different Ug, indicating that the decoupled UB values were similar whenever derived from the upper or the lower probe tip. 0.00.51.01.52.02.53.03.54.00.190.14-qm, charge density in the bed, C/kgUg-Umf , m/s Faraday cup Two tips cross-correlation  Two tips peak-times One tip time-difference 0.10.00.1.20.30.40.50.60.70.8UB, bubble rise velocity, m/sUg-Umf , m/s Two tips cross-correlation Two tips peak-times One tip time-difference 0.10 0.140.19 Figure 4.17 Comparison of charge densities and bubble rise velocities from 2t1mfF probe by different decoupling methods with Faraday cup measurement at different Ug in 0.30 m ID three-dimensional fluidized bed of original PE-I particles.  Several 2t1m probes with different configuration and tips (see Table 2.1) were also tested. Signals from these probes were analyzed by the one-tip time-difference method. As shown in Table 4.4, calculated UB increased as Ug increased. qm values were calculated by inserting UB into different calibration equations. Charge densities from the probes and Faraday cup followed a similar trend as the superficial gas velocity changed. For the Γ probe, the horizontal distance between the protruding and retracted tips is important. If this distance is too large, signals from two tips may not be well correlated.  Table 4.4 and Figures 4.13a and 4.17a show that the charge densities in the three-dimensional column tended to be smaller than those from the two-dimensional column when fluidizing PE with wide and narrow size distributions in these two columns. Also the variations of charge density at different Ug in the three-dimensional column were smaller than that in the two-dimensional column. There are two possible reasons, one is the higher moisture content and temperature of the air supplied to the three-dimensional column and the greater wall contact area in the two-dimensional column. The other reason may be the effect of particle size distribution. Particles with a wider size distribution generated a smaller 121 charge density because of the possible bi-polar charging created by small and large particles [56, 134-136]. Most commonly, smaller particles tend to be charged negatively and larger particles positively [53, 137, 138], but opposite results have also been reported [43, 56]. Mono-size particles result in the same polarity of charge on particles because of less charge generation due to particle-particle contacts relative to particle-wall contacts. So that the average charge density was larger for particles with a narrow size distribution.   Table 4.4 Decoupled qm and UB from 2t1m probes with different configurations and tips and Faraday cup (FC) at different Ug in freely bubbling 0.30 m ID three-dimensional fluidized bed of original PE-I particles.  Ug-Umf, m/s qm by probe, -C/kg qm by FC, C/kg UB by probe, m/s (a) 2t1mfF probe (see Table 2.1) 0.10 1.020.05 1.570.25 0.560.02 0.14 1.350.06 1.810.19 0.650.02 0.19 2.310.11 2.700.61 0.740.03 (b)  2t1mfΓ  probe (see Table 2.1) 0.09 1.180.09 1.350.21 0.540.02 0.13 1.220.15 1.610.32 0.640.03 0.19 3.400.37 2.020.42 0.830.04 (c)  2t1mrΓ probe (see Table 2.1) 0.06 3.670.08 1.240.24 0.490.03 0.11 3.680.13 1.550.17 0.650.03 0.19 3.700.22 2.070.38 0.830.05 (d)  2t1mfrΓ probe (see Table 2.1) 0.04 1.560.15 1.300.52 0.510.03 0.11 2.020.18 1.550.31 0.690.05 0.19 3.690.54 1.940.60 0.800.05  Signal from the 2t1mfF probe (insertion 0.75 m above distributor, static bed height=0.88 m) was analyzed by the one-tip time-difference method (see Section 4.3.3.3). Figure 4.18 shows results from both the probe and Faraday cup as Ug increased up to 0.9 m/s. A similar trend was observed for the two measurements, with the charge density increasing with increasing Ug, then becoming more or less stable with increasing Ug. Beyond Ug=0.6 m/s, signal analysis could not pick up proper peaks caused by passing bubbles, because of complex flow in the turbulent fluidization flow regime in which it is hard to identify single passing bubbles: bubble coalescence is replaced by splitting and strong interaction between leading-trailing 122 voids. Average void velocity could still be obtained by using cross-correlation function of the signals from the lower and upper tips. However the calibration equation was obtained under the two-phase flow condition where probe signals corresponded to single bubbles passing aligned with the probes. So the calibration equation could not be applied. It is suggested that the probe should be calibrated under similar conditions as the actual operation.  0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.91234567 Probe Faraday cup-qm, in-bed charge density, C/kgUg, m/s Figure 4.18 Decoupled charge densities from 2t1mfF probe and Faraday cup at different Ug in 0.30 m ID three-dimensional fluidized bed of original PE-I particles (probe: insertion 0.75 m above distributor, static bed height=0.88 m).   Current signals from 2t1mfF probes at different axial heights were analyzed and decoupled charge densities are shown in Figure 4.19. qm near the distributor (Probe 1) and bed surface (Probe 3) are larger than that in the middle of the bed (Probe 2), possibly due to more collisions between particle-particle and particle-probe tip surface near the distributor, as well as highly charged fine particles near the bed surface during fluidization. The qm near the distributor and in the middle of the bed increased as increasing Ug, while qm near the top surface showed large variation, maybe due to unstable phenomena near the bed surface. Further validation of charge density distribution inside the bed by Faraday cup measurement is needed.  123 0.25 0.30 0.35 0.40 0.45 0.500510152025 Probe 3 Probe 2 Probe 1-qm, in-bed charge density, C/kgUg, m/s Figure 4.19 Decoupled charge densities from 2t1mfF probes at different axial locations and Ug in 0.30 m ID three-dimensional fluidized bed of original PE-I particles (probe 1: z1=0.25 m, probe 2: z2=0.70 m and probe 3: z3=1.00 m, static bed height=0.88 m).   4.4.5 Comparison of two types of dual-tip probes The dual-tip (two-material) probe (2t2m) and dual-tip (one-material) probe (2t1m) were compared based on probe tip, signal analysis and decoupled results. Both probes have two tips (electrodes). The dissimilarity is that the 2t2m probe tips are made of two different materials, placed side by side, whereas the 2t1m probe tips are made of the same material, with one tip above the other. Nickel and Titanium nitride were selected as probe tip materials in both cases. The Ni and TiN tips before and after experiments were characterized by X-ray Photoelectron Spectrometers (XPS) with results shown in Appendix B.2. XPS spectra for both tips before and after experiments were similar with metallic nickel and TiN identified in both cases, and that oxidation of both tips occurred within 8 nm of the surface (measurement depth of XPS), a very thin layer compared to the total thickness of 1 mm.   Both probes are able to estimate UB and qm aided by signal analysis and decoupling methods. The signal processing criteria are different, as well as the decoupling methods. For 2t2m probe, signal analysis is based on two equations, representing current peak values from the two different materials and depends on the properties of the tip materials and particles. From Sections 2.5.1, 3.3.1 and 4.4.2.2, the relative magnitudes of signals from these two tips may change for different operating conditions and particle properties, causing difficulty in 124 calibrating the probe and decoupling the signals when particle properties, e.g. mean size and size distribution, change with time. The advantage of the 2t1m probe is that they can determine UB directly from the cross-correlation/time lag between electrostatic signals from the two probe tips.   Discrepancies between direct measurements (Faraday cup and video analysis) and two types of probes existed for the two-dimensional bubbling bed containing GB with a size range of 500-600 m. For each type of probe, the average values from each decoupling method are compared. As shown in Figure 4.20, the average relative error of qm from the 2t2m probe is lower than for the 2t1m probe, and the relative error of UB from the former is larger than from the latter. For the 2t1m probe, UB is measured directly from signals; for the 2t2m probe, on the other hand, UB is calculated based on a calibration equation. A measurement error of the order of  30% should be expected when using these probes. This suggests that the probe may be unable to provide highly-accurate values, but is able to provide trends and monitor changes with time for different operating conditions.  051015202530Average relative error, % dual-tip (two-material) probe dual-tip (one-material) probeqmUB Figure 4.20 Comparison of average relative errors with results from direct measurements and probes in two-dimensional bubbling bed of 500-600 m GB.  4.4.6 Signals from conventional single-tip probes in pressurised column with original PE-II Experiments were also conducted in a pressurized gas-solid fluidization column of inner diameter 0.15 m and height of its straight section 2.0 m (as shown in Appendix A.6). The 125 column and charge measurement system were described by Moughrabiah et al. [42] and Liu et al. [94]. Several conventional single-tip electrostatic (1t1m) probes made by Moughrabiah et al. [42] were inserted at different positions in the pressurized stainless steel three-dimensional column to evaluate electrostatic charging behaviour of polyethylene particles. The major difference of this probe to our other probes is the number of tips, with the 1t1m probe having only a single tip.  The current signals from the 1t1m probe were analyzed by the time-difference method (Section 4.3). Although these signals were only screened by applying peak detection criteria (Eqs. (4.1-4.4)), the decoupled results could still be used to evaluate trends. The effects of superficial gas velocity and pressure are shown in Table 4.5. When the bed pressure was maintained constant, the mean UB decoupled from the probe (probe B, located in the dense phase, as shown in Figure B.19) was within 0-1 m/s and increased as expected with increasing Ug. For constant Ug, the mean UB decoupled from the probe first increased then decreased as the absolute pressure increased. The initial increase in UB maybe because bubble flow is increasingly concentrated toward the centre of the column than at atmospheric pressure [139]. The subsequent decrease in UB may be due to smaller bubbles leading to smoother fluidization [140].  Table 4.5 Mean bubble rise velocity (UB) decoupled from the 1t1m probe in pressurized stainless steel three-dimensional column with original PE-II particles at different operating conditions.  absolute pressure in freeboard, 138 kPa (Constant) superficial gas velocity (Ug), m/s Umf+0.05 Umf+0.10 Umf+0.15 mean UB, m/s 0.406 0.486 0.608  Ug= Umf +0.10 (Constant) absolute pressure in freeboard, kPa 138 276 414 mean UB, m/s 0.486 0.527 0.504  The current signals from probes located at different heights were also analyzed by the time-difference method. As shown in Figure B.19, probe A was located in the lower part of the bed near the distributor, probe B in the middle part of the bed dense phase and probe C in the upper part of the bed. All three probes were inserted radially to just reach the axis of the column. As shown in Figure 4.21, the mean UB decoupled from each probe increased slightly 126 with height, indicating increasing UB along the column as the bubbles coalesced and grew. The one electrode probe could be used to calculate qm and UB, but the results were widely spread. More results on the effects of sampling parameters and signal conditioning, as well as statistical and time-frequency analysis of electrostatic signals from the conventional single-tip probes in the pressurized column can be found in Appendix B.10.   0.2 0.3 0.4 0.5 0.6 0.70.00.10.20.30.4 Probe C0.2 0.3 0.4 0.5 0.6 0.70.00.10.20.30.4average UB=0.47 m/saverage UB=0.46 m/sRelative FrequencyProbe B0.2 0.3 0.4 0.5 0.6 0.70.00.10.20.30.4average UB=0.44 m/sBubble rise velocity, m/sProbe A Figure 4.21 Comparison of relative frequency distributions of decoupled bubble rise velocities from 1t1m probes located at different heights in pressurized column of original PE-II particles (Ug= Umf +0.10, absolute pressure in freeboard: 276 kPa).  4.4.7 Differences between polyethylene and glass beads Although generation of electrostatic charges depends on several factors, such as relative humidity and temperature of the fluidizing gas, results from the glass beads (GB) and polyethylene particles (PE) can be compared when obtained under similar operating conditions of excess superficial gas velocity, temperature and relative humidity. Based on Figures 3.8 and 3.10, current peak amplitudes registered by the probe were larger for the glass beads than for the polyethylene particles. This may be caused by the difference in particle density (GB 2500 kg/m3 vs. PE 918 kg/m3) and electrical conductivity. Glass beads, which are denser and thus have a higher momentum at the same particle velocity, as well as having a higher electrical conductivity, tend to transfer more charges when they collide with 127 the probe. The current signals for PE particles also show many sub-peaks around the major ones, as shown in Figure 3.10 and previous experimental results [43].   On the other hand, based on Faraday cup measurements, qm for PE was higher than for GB at the same excess gas velocity, Ug−Umf = 0.09 m/s (see Figure 4.11 and 4.13). This indicates that higher charge density was generated on the PE particles than on the GB in the fluidized bed. This is also supported by the strong electrostatic effects observed in gas-phase polymerization reactors. The higher charge density on PE particles is related to the difference in dielectric constant and hydrophobicity of the two types of particles. PE has a lower dielectric constant (2.3 F/m) [141] than GB (5-10 F/m) [142]. The PE used in this work is regarded as hydrophobic [143, 144], and hence not as susceptible to formation of a water film on the particle surface as hydrophilic materials, such as GB [145]. So charge dissipation from the PE particles was less than from the GB, affecting the net accumulation of charges. The results also indicate that the relative amplitudes of current or cumulative charge signals received from electrostatic probes are not always consistent with the relative amplitude of particle charge density for different particles or for operation at different conditions. Therefore, it is imperative to extract particle charge densities from signals received by electrostatic probes via the signal decoupling methods developed and demonstrated in this study.  4.5 Summary Local particle charge density and bubble behaviour were estimated by collision probes of dual-tips based on the dynamic charge/current signals and calibrated equations. A signal processing procedure combined with a bubble selection algorithm is used to pre-treat raw signals from proposed probes. Decoupling based on a charge transfer and induction model extracted bubble properties and particle charge densities. Signals from the probes were processed from both two- and three-dimensional freely bubbling fluidization columns of different scale containing glass beads and polyethylene particles. Steady state experiments showed that both particle charge density and bubble rise velocity obtained from the probes were of the same order of magnitude and followed a similar trend with those directly 128 measured from Faraday cup and video images, respectively, as the superficial gas velocity changed.  For dual-tip (two-material) probe, time-difference method and two current peaks method were selected because of promising results in extracting qm, compared with values directly measured by a Faraday cup. Decoupled UB by time-difference method were closer to bubble nose velocity from video analysis and deviated less than ones calculated by a two current peaks method, which were closer to bubble wake velocity. Relative magnitudes of signals from the two materials may change for different operating conditions and particle properties, causing difficulty in calibrating the probe and in decoupling the signals. As a result, the 2t2m probe may be unsuitable for different particles, e.g. those of wide size distribution.   For dual-tip (one-material) probes, bubble rise velocity and bubble size for bubbles analyzed by two tips peak-times followed Eq. (4.11). Both the mean bubble rise velocity and bubble size increased as Ug increased. Results from one-tip time-difference method showed less scatter because of the use of bubble rise velocity correlation (Eq. (4.11)) and were closer to the results from direct measurement. Based on results from both two- and three-dimensional fluidized beds, the Γ probe with only the upper tip protruding and the F probe with two tips protruding provided closer agreement with the charge density results from Faraday cup measurements. The intrusiveness of the probe tips (F configuration) was investigated, and results show that bubble size varied within a reasonable range when passing the two vertical tips.  Differences in the signals and charge densities from glass beads and polyethylene particles are due to differences in particle properties such as density, dielectric constant, sphericity, roughness and hydrophobicity. Particles with the narrow size distribution and larger mean size generated a higher charge density. Charge signals and current peak amplitudes from the probe were larger for particles with a larger mean size. Analyzing signals from conventional single-tip probes showed widely scattered results.   129 Chapter 5 In-situ Monitoring in Gas-Solid Fluidized Beds 5.1 Introduction Electrostatic charge generation from particle-particle and particle-wall contacts in gas-solid fluidized beds is virtually inevitable for most systems. Charged particles are known to cause problems including particle agglomeration and reactor wall fouling. Several methods have been reported for mitigating bed electrification and reactor fouling, such as: humidification, antistatic agents, or increasing the superficial gas velocity to clean the distributor plate [6, 8, 146, 147]. However, a reliable charge measurement tool capable of monitoring the charges is still missing. Commonly-used electrostatic probes cannot detect changes of in-bed particle charge density because cumulative charge signals from probes are affected by changes in both electrostatics and hydrodynamics. Also there are currently no techniques to measure charging of bulk powders and no standardized methods to measure electrostatics of pharmaceutical powders [14].  A suitable charge density measurement tool could identify potential risks earlier and prevent hazards due to electrostatic discharges. It could also monitor the effectiveness of different strategies intended to reduce the charge density to provide safer operation for handing dielectric powders. Ideally signal analysis should be neither time-consuming nor complicated, while being suitable for on-line analysis of large units.   In previous chapters, the novel probes investigated gave reasonable results in steady state experiments in response to charge density and hydrodynamics changes. However, it is necessary to test the probes’ responses to continuous changes in the bed for the purpose of online monitoring.   In this chapter, the charge density level and hydrodynamics inside a two-dimensional fluidized column and a 0.30 m ID three-dimensional fluidization column were modified by abruptly changing the superficial gas velocity and by adding an antistatic agent while the 130 probes continuously monitored electrostatic charges. Signals from the probes are analyzed, and results are compared with those from a Faraday cup and video image analysis.   5.2 Experimental equipment and methods As shown in Figure 2.8, the two-dimensional fluidization column has an inside width of 0.307 m, a thickness of 22 mm and a height of 1.24 m. The Plexiglas three-dimensional fluidization column is shown in Figure 3.2, with an inner diameter of 0.30 m and a height of 4.5 m. Details of the experimental facilities are provided in Sections 2.4 and 3.2. Glass beads with a narrow size range of 500−600 m and polyethylene particles with a wide size range (100-1500 m) were used as the bed materials. Because the 2t2m probe may be unsuitable for particles of wide size distribution (see Sections 2.5.1, 3.3.1 and 4.4.2.2), 2t1m probes with different configurations (see Table 2.1) were used in the tests covered in this chapter.   The static bed height was maintained at 0.50 m and the probe was located 0.42 m above the distributor in the two-dimensional fluidization column. In the three-dimensional column, the static bed height was maintained at 0.40 m and the probe was 0.25 m above the distributor. Both beds were operated in the free bubbling mode at a constant superficial gas velocity for at least an hour to reach steady state. Then measurements were started to continuously monitor the charge density and hydrodynamics for about an hour. Then superficial gas velocity was abruptly changed or antistatic agent was quickly added into the bed, with monitoring continuing for a further hour.   Time-series signals from the probes were analyzed by the procedure illustrated in Section 4.2. After selecting appropriate pairs of peaks, different methods (explained in Section 4.3) were applied to decouple the signals. For the 2t1m probe configured with two tips protruding (F configuration), current peaks from the lower tip were used to calculate qm because of less interference than the upper tip. The results presented in this chapter are time-averaged values, with error bars corresponding to 90% confidence intervals.  131 5.3 Results and discussion  5.3.1 Effect of sampling parameters For monitoring purpose, the probe response time needs to be faster than the time over which major changes occur in hydrodynamics and charge density. Simple tests, as described in Appendix B.3, suggested that the probe response time was of order 1 millisecond or less. In data processing of online monitoring, appropriate choices of the sampling time interval and measurement frequency are important. High sampling frequency (fs) and long time interval (ts) increase data volume, whereas low sampling frequency or short duration causes loss of information. Therefore, the effects of sampling parameters on decoupling results were first investigated. Raw signals from the 2t1mfF probe with a sampling frequency, fs of 1000 Hz were first de-sampled into 500 Hz, 200 Hz and 100 Hz, for the same duration of 10 min, and then processed and analyzed by the criteria and methods outlined in Sections 4.2 and 4.3.   As shown in Figure 5.1, as fs increased, the decoupled particle charge densities (qm) from the probe by both two tips peak-times and one-tip time-difference methods increased and were closer to the Faraday cup results. However, the qm results when fs=1000 Hz were not stable and did not show as clear a change when Ug-Umf changed abruptly from 0.02 to 0.11 m/s. Probe signals show that peak amplitude and corresponding time vary with sampling frequency, so that selected pairs of peaks were different, leading to different decoupled results at different sampling frequencies. Sampling at higher frequencies can pick up higher frequency signals of little or no interest which may cause unstable and inaccurate results. On the other hand, if fs is too low, the probe cannot detect bubble rise velocity (UB) within certain a distance interval. A sampling frequency of 200-500 Hz is recommended for the 2t1m probe.  132 0 20 40 60 80 100 120012345678910change Ug(a)-qm, charge density in the bed, C/kgTime, min fs= 1000 Hz fs= 500 Hz fs= 200 Hz fs= 100 Hz Faraday cup 0 20 40 60 80 100 120012345678(b)change Ug  -qm, charge density in the bed, C/kgTime, min fs= 1000 Hz fs= 500 Hz fs= 200 Hz fs= 100 Hz Faraday cup Figure 5.1 Effect of sampling frequency on estimated charge density from 2t1mfF probe by (a) two tips peak-times method and (b) one-tip time-difference method when Ug-Umf changed abruptly from 0.02 to 0.11 m/s.   The influence of sampling time interval for monitoring the electrostatics response of the fluidized bed subject to a stepwise change is shown in Figure 5.2. Continuous time-series signals from the probe with a constant sampling frequency of 200 Hz were divided into reduced periods with different time intervals (20, 10, 5 and 2.5 min). As ts, the sampling time interval, decreased from 20 min to 2.5 min, qm from the probe decoupled by different methods responded more quickly to changes in Ug. However, the error from each method also increased as the number of selected peaks decreased. The sampling time for steady state (characteristic time of the system) has been investigated in Section B.10.1 and results showed that a sampling interval at least 2 min was required. In this case, for a purpose of monitoring the change in charge density inside the bed subject to a stepwise change in Ug, a shorter sampling time interval will be preferred to capture the dynamic change. However, considering increased errors associated with decreased sampling time interval, a sampling interval of 5-10 min was chosen for the monitoring purpose in this case study.   133 0 20 40 60 80 100 120 140 160 180012345678change Ug  -qm, charge density in the bed, C/kgTime, min ts= 20 min ts= 10 min ts= 5 min ts= 2.5 min Faraday cup(a) 0 20 40 60 80 100 120 140 160 180012345678change Ug  -qm, charge density in the bed, C/kgTime, min ts= 20 min ts= 10 min ts= 5 min ts= 2.5 min Faraday cup(b) Figure 5.2 Effect of sampling time interval on estimated charge density from 2t1mfF probe by: (a) two tips peak-times method; and (b) one-tip time-difference method when Ug-Umf abruptly changed from 0.02 to 0.11 m/s.   5.3.2 Changing Ug in two-dimensional bed of 500-600 m GB Dual-tip (one-material) probe Electrostatic charges inside the bed have been found to be affected by fluidized bed hydrodynamics because of changes in bubble behaviour and contacts among particles and between particle and the reactor wall. Effects of changing several factors (e.g. pressure, superficial gas velocity, bubble size, initial bed height and distributor) on bed electrification have been investigated in previous work [16, 26, 27, 41, 42, 94]. However, continuous changes of particle charge density and hydrodynamics have not been properly monitored and evaluated.  134 3300 3350 3400 3450 3500 3550 3600-40-30-20-10010Current, 1E-7 ATime, s upper tip lower tipchange Ug Figure 5.3 Current signals from 2t1mfF probe in freely bubbling two-dimensional fluidized bed of 500-600 m GB when Ug-Umf abruptly changed from 0.02 to 0.11 m/s. (Fluidizing gas air: T=232oC, RH=22 %.)  The current signals are plotted versus time in Figure 5.3. The current signals responded quickly and substantially to the change of Ug. Current signals from both tips increased in magnitude when Ug changed, with peak amplitudes from the lower tip again larger than from the upper tip. As shown in Figure 5.4, the predicted charge density and bubble rise velocity from the probe were of similar orders of magnitude with those obtained from Faraday cup measurement and analysis of video frames, respectively, and able to respond appropriately when the superficial gas velocity changed.  0 20 40 60 80 100 120 140 160 18001234567-qm, charge density in the bed, C/kgchange Ug two tips cross-correlation two tips peak-times one tip time-difference Faraday cupTime, min 0 20 40 60 80 100 120 140 160 1800.00.20.40.60.81.0 two tips cross-correlation two tips peak-times one tip time-difference from video analysischange UgUB, average bubble rise velocity, m/sTime, min Figure 5.4 Comparison of charge densities and bubble rise velocities from 2t1mfF probe by different decoupling methods with Faraday cup and video measurements in freely bubbling two-dimensional fluidized bed of 500-600 m GB when Ug-Umf changed from 0.02 to 0.11 m/s. (fs= 200 Hz, ts= 10 min, Fluidizing gas air: T=232oC, RH=22 % ) 135  Dual-tip (two-material) probe A 2t2mb probe (see Table 2.1) was used to monitor the changes when the superficial gas velocity was raised abruptly from Ug-Umf= 0.01 to 0.07 m/s. Results are shown in Figure 5.5. The resulting decoupled qm and UB changed consistently with results from direct measurements as Ug changed. qm from two current peaks method (Section 4.3.2.2) were closer to the Faraday cup measurement results after Ug was abruptly changed. Also qm data from Faraday cup measurements were smaller in magnitudes and varied less after abruptly increasing Ug, compared to Figure 5.4. This may be due to higher relative humidity and less of a step change in Ug-Umf (0.06 m/s rather than 0.09 m/s) in Figure 5.5 than in Figure 5.4, and the charge density level inside the bed may be unstable because of not enough fluidization time (Ug was changed just at 60 min, and might need a longer time).   0 20 40 60 80 100 120012345678change Ug-qm, charge density in the bed, C/kgTime, min Faraday cup by probe use TT2 by probe use T2-Ni by probe use T2-TiN 0 20 40 60 80 100 1200.00.20.40.60.81.01.2chan e UgUB, average bubble rise velocity, m/sTi e, min from video analysis by probe use TT2 by probe use T2-Ni by probe use T2-TiN Figure 5.5 Comparison of charge densities and bubble rise velocities from 2t2mb probe by different decoupling methods with Faraday cup and video measurements in freely bubbling two-dimensional fluidized bed of 500-600 m GB when Ug-Umf abruptly changed from 0.01 to 0.07 m/s. (fs= 200 Hz, ts= 10 min, Fluidizing gas air: T=232oC, RH=52 % )  More experiments were conducted by using a 2t2ma probe in the 0.10 m ID three-dimensional fluidization column (see Figure 3.1) with glass beads. Two sets of experiments were conducted. In the first experiment, the bed was operated at different Ug for extended time intervals (1 h, 3 min or 20-30 s). Then Ug was abruptly changed with probe measurements taken over certain sampling intervals (3 min or 20-30 s). The results indicated 136 that the probe has a short response time (~1 ms) and is sensitive to changes in both hydrodynamics and charge density in the bed. In the second experiment, Ug was changed stepwise to several different values with about one hour between each shift. Results suggested that after changing Ug abruptly each time, mean currents from both Ni and TiN tips and in-bed charge density first increased, and then became stable as time increased. Both the mean currents and absolute difference between the mean currents from the two tips increased with increasing Ug. Details are shown in Appendix B.4.   5.3.3 Adding antistatic agent to two-dimensional bed of 500-600 m GB Since changing Ug changes both the charge density and hydrodynamics inside the bed, it is necessary to test the probe’s ability to respond to charge density change alone. Wolny and Opalinski [47] concluded that the addition of fines neutralizes electric charges in the bed, independent of the electrical properties of the fines. The fines were said to change the contact conditions and transfer charges between particles in the bed, resulting in neutralization of the whole bed. Yao et al. [16] found that both the mean and standard deviation of mean current signals from a ball probe fell to insignificant levels after adding 0.1 wt% Larostat 519 powder to a freely bubbling bed of polyethylene resin particles. Larostat 519 (provided by BASF Chemical Company) is an antistatic agent which reduces static charge build-up during particle handling and processing. Its major component is quaternary ammonium salts, a non-polymeric cationic compound which is considered to be a hydrophilic additive [148]. It has a density of 520 kg/m3 and a size range of 6-20 m. The Larostat particles are non-spherical, with uneven surfaces. Previous work [16, 41] has focused on examining the effect of fines addition on recorded current and charge buildup in the bed, but it is also necessary to quantify the time-variation of particle charge density and hydrodynamics following sudden addition of fines to the bed.  137 50 60 70 80 90 100-20-15-10-50550 60 70 80 90 100-20-15-10-505 Figure 5.6 Current signals from 2t1mfF probe in freely bubbling two-dimensional fluidized bed of 500-600 m GB at Ug-Umf=0.05 m/s after 0.1 wt% Larostat 519 was added. (Fluidizing gas air: T=221oC, RH=21%.)  The bed was operated in the bubbling flow regime at Ug-Umf=0.05 m/s overnight to reach steady state. Then 0.1 wt% Larostat 519 was added by pouring quickly from the top of the column. Figure 5.6 shows the current signals from the probe before and after adding the antistatic agent. Signals from both probe tips decreased by one order of magnitude as a result of adding the anti-static fines. Signals from after the adding were close to the baseline, reaching the minimum limit which the electrometer could measure. The current signal significantly decreased with polarity changed from negative to positive. From visual observation, wall fouling was significantly reduced and disappeared within a minute of adding the antistatic agent. As shown in Figure 5.7, qm from the Faraday cup showed a significant drop when Larostat 519 was added. The addition of the antistatic agent increases the surface conductivity by increasing the water adsorption at low relative humidity. Probe signals before 70 min showed detectable amplitudes and therefore were chosen to analyze. Before the adding, decoupled charge densities were in same order of magnitude and close to Faraday cup results; after the adding, charge density predicted by the probe decreased consistently with Faraday cup measurement. Decoupled bubble rise velocities from the probe 138 were more or less the same (~0.6 m/s) before and after the addition over the limited period of time analysed.    147 Results indicated that both particle charge density and bubble rise velocity obtained from the probes were of the same order of magnitude and followed similar trends as those directly 

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.24.1-0135688/manifest

Comment

Related Items