<%BANNER%>

The Effects of Electrostatic Charge on Particle Flows and Deposition

Permanent Link: http://ufdc.ufl.edu/UFE0044894/00001

Material Information

Title: The Effects of Electrostatic Charge on Particle Flows and Deposition
Physical Description: 1 online resource (203 p.)
Language: english
Creator: Bunchatheeravate, Pongpumin Poom
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2012

Subjects

Subjects / Keywords: deposition -- electrostatic -- fluorescence -- particle
Chemical Engineering -- Dissertations, Academic -- UF
Genre: Chemical Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Particulate matters have a wide range of application industry. Particle processes ranges from mixing to transportation. Two parameters that remain relatively unknown but occur commonly in all processed are electrostatic charge and particle morphologies. Understanding of these two parameter can led improve efficiency of the processes. Particle charging can occur during transport and causes deposited. The work involve here investigate spherical/non-spherical particle charging and deposition in pipe through two different application: pneumatic conveying and pulmonary deposition. A simple, yet essential, process is transporting particles in a pipe from one location to another. This process is also call pneumatic conveyer. When particles are suspended in the air this is call diluted pneumatic conveying system. During transport, particle gains electrostatic charge upon contact with the pipe wall. A method was devise to predict the particle charge in pneumatic conveying system. The first step is done by performing an experiment measuring particle charge after a fix pipe length with varying initial particle charge. This experiment will determine charging parameters for the specific particle-pipe system. The next step is to perform computational fluid dynamics (CFD) simulations to determine collision frequency occur in the experiment. Now it is possible to accurately predict particle charging in any system, with just one more CFD simulation. This method accurately predicts particle charge for any system as long as the particle-pipe system is the same as the experiment. The particle charge and morphology can led to different deposition seen by uncharged spherical particle. These two parameters can lead to an improvement in the field of pulmonary deposition. Uncharged non-spherical particles were sent into a triple bifurcating airways model. The local depositions were determined using fluorescence imaging. The results indicate that low momentum particle deposition were a combination between inertial impaction and sedimentation, whereas high momentum particle deposited through inertial impaction and sedimentation. Charge particles were then generated using electrospray. The results indicate that the particles with bipolar charge distribution, resulting in zero net charge, gain overall improvement in all airways equally. When particles with unipolar charge distribution enter the airway the space charging effect dominated the particle deposition.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Pongpumin Poom Bunchatheeravate.
Thesis: Thesis (Ph.D.)--University of Florida, 2012.
Local: Adviser: Curtis, Jennifer S.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2012
System ID: UFE0044894:00001

Permanent Link: http://ufdc.ufl.edu/UFE0044894/00001

Material Information

Title: The Effects of Electrostatic Charge on Particle Flows and Deposition
Physical Description: 1 online resource (203 p.)
Language: english
Creator: Bunchatheeravate, Pongpumin Poom
Publisher: University of Florida
Place of Publication: Gainesville, Fla.
Publication Date: 2012

Subjects

Subjects / Keywords: deposition -- electrostatic -- fluorescence -- particle
Chemical Engineering -- Dissertations, Academic -- UF
Genre: Chemical Engineering thesis, Ph.D.
bibliography   ( marcgt )
theses   ( marcgt )
government publication (state, provincial, terriorial, dependent)   ( marcgt )
born-digital   ( sobekcm )
Electronic Thesis or Dissertation

Notes

Abstract: Particulate matters have a wide range of application industry. Particle processes ranges from mixing to transportation. Two parameters that remain relatively unknown but occur commonly in all processed are electrostatic charge and particle morphologies. Understanding of these two parameter can led improve efficiency of the processes. Particle charging can occur during transport and causes deposited. The work involve here investigate spherical/non-spherical particle charging and deposition in pipe through two different application: pneumatic conveying and pulmonary deposition. A simple, yet essential, process is transporting particles in a pipe from one location to another. This process is also call pneumatic conveyer. When particles are suspended in the air this is call diluted pneumatic conveying system. During transport, particle gains electrostatic charge upon contact with the pipe wall. A method was devise to predict the particle charge in pneumatic conveying system. The first step is done by performing an experiment measuring particle charge after a fix pipe length with varying initial particle charge. This experiment will determine charging parameters for the specific particle-pipe system. The next step is to perform computational fluid dynamics (CFD) simulations to determine collision frequency occur in the experiment. Now it is possible to accurately predict particle charging in any system, with just one more CFD simulation. This method accurately predicts particle charge for any system as long as the particle-pipe system is the same as the experiment. The particle charge and morphology can led to different deposition seen by uncharged spherical particle. These two parameters can lead to an improvement in the field of pulmonary deposition. Uncharged non-spherical particles were sent into a triple bifurcating airways model. The local depositions were determined using fluorescence imaging. The results indicate that low momentum particle deposition were a combination between inertial impaction and sedimentation, whereas high momentum particle deposited through inertial impaction and sedimentation. Charge particles were then generated using electrospray. The results indicate that the particles with bipolar charge distribution, resulting in zero net charge, gain overall improvement in all airways equally. When particles with unipolar charge distribution enter the airway the space charging effect dominated the particle deposition.
General Note: In the series University of Florida Digital Collections.
General Note: Includes vita.
Bibliography: Includes bibliographical references.
Source of Description: Description based on online resource; title from PDF title page.
Source of Description: This bibliographic record is available under the Creative Commons CC0 public domain dedication. The University of Florida Libraries, as creator of this bibliographic record, has waived all rights to it worldwide under copyright law, including all related and neighboring rights, to the extent allowed by law.
Statement of Responsibility: by Pongpumin Poom Bunchatheeravate.
Thesis: Thesis (Ph.D.)--University of Florida, 2012.
Local: Adviser: Curtis, Jennifer S.

Record Information

Source Institution: UFRGP
Rights Management: Applicable rights reserved.
Classification: lcc - LD1780 2012
System ID: UFE0044894:00001


This item has the following downloads:


Full Text

PAGE 1

1 THE EFFECT S OF ELECTRO STATIC CHARGE ON PARTICLE FLOWS AND DEPOSITION By PONGPUMIN POOM BUNCHATHEERAVATE A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2012

PAGE 2

2 2012 Pongpumin Poom Bunchatheeravate

PAGE 3

3 To my family and friends for their support

PAGE 4

4 ACKNOWLEDGMENTS I thank my advisor, Dr. Jennifer Curtis for her constant guidance, supervision, and encouragement throughout the year. Under her guidance I had gain new prospective on my research. Thank you to Dr. Scott Brown for all his input and help over the year. Thos e input were significant in my progress. Thank you to Dr. Shuji Matsusaka for letting me into his group at the Kyoto University in Japan. During my stay in Japan I lad gain many viewpoints that would not had been possible elsewhere. Thank you to Dr. Jan Ma rijnissen and Dr. Caner Yurteri for their constant guidance of electrospraying technique. And lastly thank you to all my friends, co worker, and family members who had supported me. And lastly the work presented in this manuscript is made possible through the funding from NSF PIRE, project number NSF OISE 0968313, Grant in Aid for Scientific Research (B) (No. 23360341) from the Japan Society for the Promotion of

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 8 LIST OF FIGURES ................................ ................................ ................................ ........ 10 ABSTRACT ................................ ................................ ................................ ................... 14 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 16 Particle Charging During Transport ................................ ................................ ........ 17 Particle Deposition to Pulmonary Airways ................................ .............................. 19 2 PREDICTION OF PARTICLE CHARGING IN A DILUTE PNEUMATIC CONVEYING SYSTEM ................................ ................................ ........................... 20 Background ................................ ................................ ................................ ............. 20 Theory ................................ ................................ ................................ ..................... 23 Charge Transfer Model for Metal Metal Systems ................................ ............. 23 Charge Transfer Model for Metal Insulator and Insulator Insulator Systems .... 25 Charge Transfer Model for Repeated Impacts ................................ ................. 26 Experiments ................................ ................................ ................................ ............ 27 Experimental Setup ................................ ................................ .......................... 27 Short Pipe Experimentation with Varying Initial Charge ................................ ... 29 Validation Experiments with Longer Pipes ................................ ....................... 30 Particles and Pipes ................................ ................................ ........................... 30 Result s and Discussion ................................ ................................ ........................... 30 Varying Initial Charge in the Particle Charger ................................ ................... 30 Short Pipe Experimentation with Varying Initial Charge ................................ ... 31 Predicting Particle Charging using Stair Stepping Procedure .......................... 32 Predicting Particle Charging using the Semi Th eoretical Equations ................. 32 Technique Validation ................................ ................................ ........................ 34 Summary ................................ ................................ ................................ ................ 34 3 CHARGE PREDICTION USIN G COMUTATIONAL FLUIDS DYNAMICS (CFD) SIMULATION ................................ ................................ ................................ .......... 46 Background ................................ ................................ ................................ ............. 46 Model Description ................................ ................................ ................................ ... 47 Governing Equa tions ................................ ................................ ........................ 47 Drag Models ................................ ................................ ................................ ..... 47

PAGE 6

6 Gas Phase Stress ................................ ................................ ............................ 48 Solid Stress ................................ ................................ ................................ ...... 49 Turbulence Interaction ................................ ................................ ...................... 50 Computer Simulation ................................ ................................ ........................ 51 Determining Collision Number ................................ ................................ .......... 51 Experiments ................................ ................................ ................................ ............ 53 Simulation ................................ ................................ ................................ ............... 53 Results and Discussion ................................ ................................ ........................... 54 Particle Wall Collision ................................ ................................ ....................... 54 Determining ................................ ................................ ............................ 56 Comparisons to Experimental Results ................................ ............................. 56 Expanding the Prediction Model ................................ ................................ ............. 57 Scale Up the Pip e Line ................................ ................................ ..................... 57 Monitoring the Environmental Condition ................................ ........................... 58 Scaling the Particle Size ................................ ................................ ................... 59 Accounting for the Complex Flow Geometry ................................ .................... 60 Summary ................................ ................................ ................................ ................ 61 4 DEPOSITION OF CHARGED PARTICLES IN BIFURCATING AIRWAY ............... 69 Background ................................ ................................ ................................ ............. 69 Electrospray Technique ................................ ................................ .................... 71 Production of Charged, Spherical Particles ................................ ............... 71 Production of Charged, Non Spherical Particles ................................ ........ 73 Near Infrared (NIR) Spectroscopy ................................ ................................ .... 74 Experiments ................................ ................................ ................................ ............ 74 Bifurcating Model ................................ ................................ .............................. 74 Generation of Spherical Particles ................................ ................................ ..... 7 5 Generation of Non Spherical Particles ................................ ............................. 76 Experimental Setup ................................ ................................ .......................... 77 Characte rizing Neutrally Charged Particles ................................ ...................... 79 Particle Deposition ................................ ................................ ........................... 80 NIR Imaging ................................ ................................ ............................... 80 Results ................................ ................................ ................................ .................... 82 ................................ ................................ ....... 82 Deposition of Positively Charged Spheres ................................ ....................... 83 Deposition of Non Spherical Particles ................................ .............................. 84 Concentration of the Deposited Particles ................................ ......................... 85 Summary ................................ ................................ ................................ ................ 86 5 DEPOSITION OF NON SPHERICAL PARTICLES IN BIFURCATING AIRWAY .. 106 Background ................................ ................................ ................................ ........... 106 Deposition Efficienc y ................................ ................................ ...................... 108 Deposition Mechanisms ................................ ................................ ................. 109 Particle Characterization ................................ ................................ ................ 111 Bounda ry Layer ................................ ................................ .............................. 112

PAGE 7

7 Experiments ................................ ................................ ................................ .......... 112 Particle Characterization ................................ ................................ ................ 112 Airway Model ................................ ................................ ................................ .. 113 Particle Depositio n ................................ ................................ ......................... 113 Quantifying the Deposition ................................ ................................ ............. 114 Results ................................ ................................ ................................ .................. 115 Particle Characteristics ................................ ................................ ................... 115 Partic le Deposition ................................ ................................ ......................... 116 Deposition Model ................................ ................................ ............................ 117 Model Validation ................................ ................................ ............................. 120 Summary ................................ ................................ ................................ .............. 121 6 CONCLUSION AND RECOMMENDATION ................................ .......................... 137 Particle Charging During Transport ................................ ................................ ...... 137 Particle Deposition in Pulmonary Airways ................................ ............................. 138 Use of NIR Fluorescence ................................ ................................ ............... 138 Effect of Electrostatic charge on Particle Deposition ................................ ...... 139 Deposition of Charged Spheres ................................ ................................ ..... 139 Deposition of Charged, Rod Shaped Particles ................................ ............... 139 Deposition of Charged Hollow, Balloon Shaped Particles .............................. 139 Future Experiments ................................ ................................ .................. 140 Deposition o f Non Spherical Particles ................................ ............................ 140 Future applications to simulation ................................ ................................ .... 141 A P PENDIX A EFFECT OF V EX ON THE CONTACT POTENTIAL ................................ .............. 143 B RAW DATA FOR PARTICLE CHARGING EXPERIMENT ................................ ... 146 C PARAMETRIC STUDY ................................ ................................ ......................... 156 D ADDITIONAL COMPARISON FOR CHARGE MODEL VALIDATION .................. 169 E ADDITIONAL IVIS IMAGES FOR DIFFERENT PARTICLE SHAPE .................... 187 LIST OF REFERENCES ................................ ................................ ............................. 195 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 203

PAGE 8

8 LIST OF TABLES Table page 2 1 Characteristics of particles used in the experiment ................................ ............ 36 2 2 Characteristic of pipes used in the experiment ................................ ................... 36 2 3 Characteristic values for each particle in 1 m pipe ................................ ............. 37 3 1 Term used in the turbulence and solid stress model ................................ .......... 62 3 2 Inputs and outputs for the simulation ................................ ................................ .. 63 3 3 Air flow in each type of pipe at each pressure setting ................................ ......... 63 3 4 Characteristics of particles used in the experiment ................................ ............ 64 3 5 Characteristic of pipes used in the experiment ................................ ................... 64 3 6 Characteristic values for each particle in 1 m pipe ................................ ............. 64 4 1 Dimension of the bifurcating airway model ................................ ......................... 88 4 2 Electrospray solution properties ................................ ................................ ......... 88 4 3 Mass percent deposition of particles ................................ ................................ .. 88 4 4 The average deposition concentration of particles ................................ ............. 89 5 1 Dimension of the bifurcating airway model ................................ ....................... 122 5 2 Flow parameter in the bifurcating airway mode l ................................ ............... 122 5 3 Characterization for each type of particle ................................ ......................... 122 5 4 The percentage of deposition in each generation of the airway model ............. 123 5 5 Deposition parameter for each particle in each airway ................................ ..... 124 5 6 Contribution of deposition from each mechanism ................................ ............. 125 B 1 Raw data for borosilicate particle copper pipe system ................................ ..... 146 B 2 Raw data for borosilicate particle stainless steel pipe system .......................... 146 B 3 Raw data for borosilicate particle natural glass pipe system ............................ 147 B 4 Raw data for borosilicate particle pyrex pipe system ................................ ........ 147

PAGE 9

9 B 5 Raw data for soda lime particle copper pipe system ................................ ........ 148 B 6 Raw data for soda lime particle stainless steel pipe system ............................. 148 B 7 Raw data for soda lime particle natural glass pipe system ............................... 149 B 8 Raw data for soda lime particle pyrex pipe system ................................ .......... 149 B 9 Raw data for PMMA particle copper pipe system ................................ ............. 150 B 10 Raw data for PMMA particle stainless steel pipe system ................................ 150 B 11 Raw data for PMMA particle natural glass pipe system ................................ ... 151 B 12 Raw data for PMMA particle pyrex pipe system ................................ ............... 151 B 13 Raw data for crushed soda lime particle copper pipe system .......................... 152 B 14 Raw data for crushed soda lime particle copper pipe system .......................... 152 B 15 Raw data for crushed soda l ime particle copper pipe system .......................... 153 B 16 Raw data for crushed soda lime particle copper pipe system .......................... 153 B 17 Raw data for JSC 1A particle copper pipe system ................................ ........... 154 B 18 Raw data for JSC 1A particle Stainless steel pipe system ............................... 154 B 19 Raw data for JSC 1A particle natural glass pipe system ................................ .. 155 B 20 Raw data for JSC 1A particle pyrex pipe system ................................ ............. 155 C 1 Inputs and outputs for the simulation ................................ ................................ 158

PAGE 10

10 LIST OF FIGURES Figure page 2 1 Schematic of experimental units ................................ ................................ ......... 38 2 2 Experimental setup f or measuring particle charge ................................ .............. 39 2 3 Specific charge of particles leaving the particle charger ................................ ..... 40 2 4 q m ( L =1) vs q m,0 for borosilicate particles in 1m natural glass pipe ...................... 41 2 5 q m ( L =1) vs q m,0 for borosilicate particles in 1m copper pipe ............................... 42 2 6 Predicting particle charge using stair step method ................................ ............. 43 2 7 Predicting charging profile for borosilicate particles ................................ ............ 44 2 8 Comparison of the experimental data with pre dicted charging profile ................ 45 3 1 Comparison of single phase prediction to benchmark data. ............................... 65 3 2 Collision number for soda lime particle in copper pipe ................................ ....... 66 3 3 Collision number along the pipe for soda lime particle in copper pipe ................ 6 6 3 4 Predicted charging profile for soda lime particles in copper pipe ........................ 67 3 5 Predicted charging profile for JSC 1A particles in natural glass pipe ................. 68 4 1 Schematic of electrospray ................................ ................................ .................. 90 4 2 Schematic of the bifurcating airway model ................................ ......................... 91 4 3 Properties of spherical particles ................................ ................................ .......... 92 4 4 PLGA particles generated in various solvents ................................ .................... 93 4 5 Microscopic images of rod shaped particles ................................ ....................... 94 4 6 Microscopic images of hollow, balloon shaped particles ................................ .... 95 4 7 Experimental setup ................................ ................................ ............................. 96 4 8 Experimental setup using a parallel plate capacitor ................................ ............ 97 4 9 Voltage build up rates in the Faraday cup at different plate voltages ................. 98 4 10 Emission spectra from solid PLGA film embedded with NIR dye ........................ 99

PAGE 11

11 4 11 Compilation of IVIS Lumina images showing no photons in the airway model 100 4 12 Example of in dividual regions of interest for generation 1 airway ..................... 100 4 13 ................................ ........................... 101 4 14 ................................ ........ 102 4 15 IVIS Lumina images of positively charged sphere ................................ ............ 103 4 16 Evolution of deposition patterns for positively charged sphere ......................... 104 4 17 Deposition of rod shaped particles. ................................ ................................ .. 105 5 1 Shape and characteristic dimensions of different types of particles ................. 126 5 2 Schematic of the bifurcating airway model ................................ ....................... 127 5 3 Schematic of the inverse cyclone seeder setup ................................ ................ 128 5 4 Particle viewed under optical microscope ................................ ......................... 129 5 5 IVIS Lumina images of glass spheres ................................ .............................. 131 5 6 IVIS Lumina images of glass flakes 1 ................................ ............................... 132 5 7 Plot of the against Stokes number ................................ ............................... 133 5 8 Deposition efficiency of the secondary mechanism ................................ .......... 134 5 9 vs. ................................ ................................ ................................ 136 A 1 Different cases of contact charge transfer in applied electric field .................... 145 C 1 The effect of mass loading ................................ ................................ ............... 159 C 2 The effect of pipe radius ................................ ................................ ................... 160 C 3 The effect of mean gas velocity ................................ ................................ ........ 161 C 4 The effect of maximum solid volume fraction ................................ .................... 162 C 5 The effect of particle diameter ................................ ................................ .......... 163 C 6 The effect of particle density ................................ ................................ ............. 164 C 7 The effect of particle sphericity ................................ ................................ ......... 165 C 8 The effect of particle particle coefficient of restitution ................................ ....... 166

PAGE 12

12 C 9 The effect of particle wall coefficient of restitution ................................ ............ 167 C 10 The effect of wall specularity coefficient ................................ ........................... 168 D 1 Comparison for borosilicate particles and copper pipe ................................ ..... 169 D 2 Comparison for borosilicate particles and stainless steel pipe .......................... 170 D 3 Comparison for borosilicate particles and natural glass pipe ............................ 171 D 4 Comparison for borosilicate particles and Pyrex pipe ................................ ....... 172 D 5 Comparison for soda lime particles and stainless steel pipe ............................ 173 D 6 Comparison for soda lime particles and natural glass pipe .............................. 174 D 7 Comparison for soda lime particles and Pyrex pipe ................................ .......... 175 D 8 Comparison for PMMA particles and copper pipe ................................ ............ 176 D 9 Comparison for PMMA particles and stainless steel pipe ................................ 177 D 10 Comparison for PMMA particles and natural glass pipe ................................ ... 178 D 11 Comparison for PMMA particles and Pyrex pipe ................................ .............. 179 D 12 Comparison for crushed soda lime particles and copper pipe .......................... 180 D 13 Comparison for crushed soda lime particles and stainless steel pipe ............... 181 D 14 Comparison for crushed soda lime particles and natural glass pipe ................. 182 D 15 Comparison for crushed soda lime particles and Pyrex pipe ............................ 183 D 16 Comparison for JSC 1A particles and copper pipe ................................ ........... 184 D 17 Comparison for JSC 1A particles and stainless steel pipe ............................... 185 D 18 Comparison for JSC 1A particles and Pyrex pipe ................................ ............. 186 E 1 IVIS Lumina images of hollow glass particles ................................ ................... 187 E 2 IVIS Lumina images of glass rod particles ................................ ........................ 188 E 3 IVIS Lumina images of glass flake 1 particles ................................ .................. 189 E 4 IVIS Lumina images of glass flake 2 particles ................................ .................. 190 E 5 IVIS Lumina images of crushed glass particles ................................ ................ 191

PAGE 13

13 E 6 IVIS Lumina images of silver flake particles ................................ ..................... 192 E 7 IVIS Lumina images of Aluminum Oxide particles ................................ ............ 193 E 8 IVIS Lumina images of JSC 1A particles ................................ .......................... 194

PAGE 14

14 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy THE EFFECTS OF ELECTROSTATIC CHARGE ON PARTICLE FLOWS AND DEPOSITION By Pongpumi n Poom Bunchatheeravate December 2012 Chair: Jennifer S. Curtis Major: Chemical Engineering Particulate matters have a wide range of applications in industry. Particle processes range from mixing to transportation. Electrostatic charge and particle morp hologies are two parameters that remain relatively unknown, yet they are very important in all processes. The understanding of these two parameters can improve the efficiency of particle processes and their applications. Particle charging occurs during tra nsport; charged particles deposit more readily than neutral particles. The non spherical shape of the particles can cause differences in flow and charging characteristics. This investigation seeks to expand the knowledge in charging and deposition of spher ical and non spherical particles through two different applications: pneumatic conveying and pulmonary deposition. Transporting particles in a pipe from one location to another is a simple yet essential process; this process is called pneumatic conveying. Dilute pneumatic conveying is a specific case when the particle concentration in the flow is very low. During transport, particles gain electrostatic charge upon contact with the pipe wall. A method coupling small scale experimentation and computer simulat ion was devised to

PAGE 15

15 method it was possible to accurately predict particle charging for both spherical and non spherical particles in any given pipe length, under different air flow rates and particle mass flow rates. The particle charge and morphology can lead to different depositions as seen by uncharged spherical particles. These two parameters can lead to an improvement in the field of pulmonary deposition. A technique using near infrared fluorescence was devised to visualize and quantify particle deposition in a complex airway. Particles of various shapes and charges were sent into a bifurcating airway model. Deposition of the charged particles depended upon the charge dist ribution of all the particles. When uncharged particles were used, the deposition indicated that low momentum particle deposition was a combination between inertial impaction and diffusion; whereas high momentum particles deposited through inertial impacti on and sedimentation.

PAGE 16

16 CHAPTER 1 INTRODUCTION Particulate matters are essential in many processes in industry. Examples of these processes include transporting particles in a production line, using catalytic particles in fluidized beds to increase the production yield, filtration of dust to minimize contamination, and aerosolizing medicinal particles for use in inhalation therapy. The knowledge of ea ch individual process is well developed. However, there is a characteristic of particulate matter that still remains relatively unknown in all processes. The effect of electrostatic charges on particles in many of these processes still remains elusive. In particle suspension, particles gain electrostatic charge when they come into contact with other surfaces. The most appropriate process to study particle charging is during particle transport in pipe flow. The study of how charges are generated in a pipelin e system can lead to the understanding of particle charging during contact. Once particles become charged, the electrostatic force draws particles toward the wall where they deposit. The best scenario to understand the effect on the enhancement of depositi on is in a series of branching airways. With these goals in mind, two sets of experiments were devised in order to: Measure particle charging in diluted pneumatic conveying systems Accurately predict particle charging in diluted pneumatic conveying systems Gauge the use of near infrared fluorescence to quantify particle deposition Determine the effect of particle charging on particle deposition Determine the effect of particle morphology on particle deposition These experiments were performed using both sp herical and non spherical particles to determine the influence of particle shape.

PAGE 17

17 The work presented in the following chapters seeks to investigate the particle charging and the deposition of charged particles in the pipeline. Initially, we have particles flowing through a pipeline. As particles flow in the pipeline, they can gain or lose electrostatic charges. The changes in electrostatic charge on the particles are investigated in Chapter 2 and Chapter 3. Electrostatic charges on the particles cause the p articles to be more readily deposited on the surface of the pipe. A bifurcating airway model is used to understand the particle deposition in different geometries in a pipeline. Each bifurcating airway contains a section of straight pipe (representing a st raight pipeline) and a branching section, where the pipe bifurcates into two airways (representing a split or bend in the pipeline). In addition, bifurcating airways have direct application in pulmonary drug delivery; the airways contain a structure simila r to the human lung. The work in the next chapter includes developing a technique to visualize particle deposition in each region of the airway (Chapter 4). To understand particle deposition in the bifurcating airways, two different sets of experiments wer e performed. The first set of experiments investigated the effect of electrostatic charge on particle deposition (Chapter 4); the second set of experiments investigated the effect of particle morphology on particle deposition (Chapter 5). The work in Chapt er 4 and Chapter 5 is used to isolate the effects of particle charge and particle shape. Particle Charging During Transport Particle transport (pneumatic conveying) is an essential component in many operations. Pneumatic conveying transports bulk particulate materials from one point to another with minimum cost. When the solid volume is very low compared to the gas volume (less than 1%), particles are suspended in the air during transport; this operating regime is known as dilute phase transport (R hodes 2008) Airflow is sent into

PAGE 18

18 the pipe at a high velocity to move the particle suspension. Understanding particle charging in pneumatic conveying can lead to a multitude of applications in many industries. During particle transport, particles acquire electrostatic charge upon contact with the pipe wall. Neglecting electrostatic charge buildup on the particles will result in many serious issues. Some examples include: dust explosions or explosion hazards that are caused by spark discharges from excessiv ely charged particles (Cartwright et al. 1985; Eckhoff 2011; Glor 1985). Charged particles can also accumulate in pipelines causing blockages and flow interferences (Forward et al. 2009). In pharmaceutical applications, charged particles can agglomerate an d segregate, leading to uneven drug formulation and dosages (Peart et al. 2003; Supuk et al. 2011). Charged particles generated in medical inhalers can lead to drastically different deposition patterns in the lungs (Leong et al. 2009). On the other hand, t here are benefits to having charged particles during transport. Particle charge is key to an online, non intrusive monitoring system for flow information, including but not limited to mass loading, particle distribution and solid velocity (Matsusaka and Ma suda 2006; Matsusaka et al. 2008a; Deng et al. 2011; Gajewski 1996; 2006). Regardless of application, there is a need to measure and control particle charge in pneumatic conveying systems. We presented here in Chapter 2 and 3, a predictive method coupling a small scale experiment with computer simulation to predict charge on both spherical and non spherical particles. Experiments were first performed to determine the charging properties and the charge of the particles was then predicted using CFD simulatio n.

PAGE 19

19 Particle Deposition to Pulmonary Airways Charged particles tend to deposit on surfaces. A straight pipe is less than ideal for understanding particle deposition. However, branching airways provide a perfect geometry to study particle deposition. Unlike a straight pipe airway, branching airways have multiple sections that use different deposition mechanisms. A direct application for deposition in a series of branching airways is pulmonary drug delivery. The human lungs contain multiple branching airways. Particle deposition in pulmonary airways has great importance in pulmonary medication. The lungs provide a large area for drug delivery. Particles deposited in the lungs can degrade and release drugs into the problematic area; if particles penetrate deep e nough and deposit in the alveoli, the drugs can enter into the bloodstream. Understanding how charged particles and non spherical particles deposit in the branching airways can greatly enhance pulmonary medication. For this experiment we first investigate techniques using near infrared (NIR) fluorescence to visualize and quantify particle deposition. Using the NIR fluorescence, we investigated the effect of charged particles (spherical and non spherical), as well as uncharged (spherical and non spherical) p articles, in a series of bifurcating airways.

PAGE 20

20 CHAPTER 2 PREDICTION OF PARTIC LE CHARGING IN A DIL UTE PNEUMATIC CONVEY ING SYSTEM Background Particle charging is a common phenomenon that occurs naturally and frequently. Any process that uses dry powde rs will involve charge d particles. Dilute phase pneumatic conveying, in which particles are suspended in gas and transported in a pipeline, is such a process found in many particle processing and manufacturing industries. As particles are being transported they acquire electr ostat ic charge upon contact with the pipe wall. Neglecting charges in design of particle handling processes may result in many serious issues. Examples include dust explosions or explosion hazards that are cause d by spark discharge fro m overly charged particles (Cartwright et al. 1985; Eckhoff 2011; Glor 1985). Charged particles can also accumulate in pipelines causing blockage and flow interference (Forward et al. 2009) In pharmaceutical applications, charged particles can agglomerate a nd segregate, leading to changes in drug formulation and dosages (Peart et al. 2003; Supuk et al. 2011) Charged particles generated in medical inhalers can led to drastically different deposition patterns in the lung (Leong et al. 2009) On the other hand, particle charging can be beneficial; it is possible to use the measured particle charge to create an online, non intrusive monitoring system for flow information including but not limited to mass loading, particle distribution and solid velocity (Matsusaka and Masuda 2006; Matsusaka et al. 2008a; Deng et al. 2011; Gajewski 1996; 2006) Different methods have been used to study the effects of electrostatic charge, using both computer simulations and experimen tation. Lim et al. (2012) used large ed dy simulation and the discrete element method (DEM) to investigate the effects of

PAGE 21

21 electrostatic charge on flowing granular materials. They found that charged particles exhibit a ring like distribution during pipe flow with more particles found closer towar d the pipe wall. They also investigated the effect on the drag force and the electric field generated due to the electrostatic force. Watano et al. (2003) performed DEM simulations as well as experiments to investigate electrostatic charging when a pile of PMMA particles was blown through a horizontal pipe. They found that the number of wall collisions and the vertical component of the impact velocity are important parameters for charging. However, most of the research on particle charging has been experim ental, focused on determining the overall dependency of electrostatic charging on specific parameters. Masuda et al. (1972) coated a section of a pipeline with a polymer film and used a galvanometer to measure the current transfer upon impact. The number o f particle contacts with the wall was then related to the charge transfer. They found that the current transfer between particles and the pipe surface depends on the contact frequency, contact area, contact time, and particle size. Smeltzer et al. (1982) m easured the charge buildup on glass beads flowing through plexi glass pipes. Nickel wires were inserted into the plexi glass pipe, and the charge was measured when particles came into contact with the wire. They found that the charge was linearly related t o the collision frequency, which was measured in the various pipe sections. They also determined that impact charging was the primary mechanism for charge transfer in dilute pneumatic conveying. Yao et al. (2006) used electrical capacitance tomography, tog ether with particle imaging velocimetry, to measure the time required for a flowing particle system to reach electrostatic equilibrium in a cyclic pipe flow. They

PAGE 22

22 found that a longer time is required for a more complex geometry, and electrostatic equilibri um is reached faster in a horizontal pipe than a vertical pipe. Rather than measuring the charge transfer between particles and pipe surfaces, it is also possible to directly measure the charge on particles using a Faraday cup, also known as a Faraday cag e. In this method, when particles enter the Faraday cup, an electric current is sent to the electrometer. The current reading measured from the electrometer can then be directly related to the amount of charge possessed by the particles. Kanazawa et al. (1 995) used an electrostatic voltmeter to measure the charge on the pipe surface and a Faraday cup to measure the particle charge in a flowing particle system. They found a bipolar distribution of charge for the irregularly shaped particles, but a single pol ar distribution of charge for spherical particles. Nieh and Nguyen (1988) also used a Faraday cup to determine the effect of humidity, particle velocity, and particle diameter on particle charging behavior. They found that humidity has a significant effect on particle charging, and particles are neutralized at high humidity. Smaller particles were also found to have a higher charge to mass ratio. Only a handful of previous studies have investigated the charging behavior for a single particle impact (impact charging) and how the charging can be controlled. Matsusaka et al. (2000) perform ed experiments using a 31 mm rubber sphere impacting on a steel plate and came to the conclusion that the charge transferred in a single impact depends on the initial charge o f the material the impact velocity and the contact area. Matsuyama and Yamamoto (1995) perform ed experiments with 3 mm polymer particles and a metal plate; they found that the impact charge is linearly dependent on the initial charge of the particle. Ther e is also a charge beyond which a particle will no

PAGE 23

23 longer gain or lose charge upon further impact; this is known as the equilibrium charge and is independent of the impact conditions. Watanabe et al. (2007) studied 500 600 m drug particles impacting a stai nless steel plate; they found that the equilibrium charge is related to the contact potential difference. Matsusaka et al. (2002) also observed this equilibrium charging in a flowing particle system after a certain pipe length Matsusaka et al. (2007) then confirmed in a flowing particle system that the amount of charge transfer is a function of the initial charge and the equilibrium charge. They observed that the amount of charge on the particles varies exponentially as a function of pipe length, and the c harge transfer continues until particles reach equilibrium charge in the system With this charging relationship, they were able to control the particle charge in pipe flow by changing the wall material with which the particles come in contact. In an atte mpt to control particle charge, Matsusaka et al. (2008b; 2011) have designed and built particle chargers. As particles flow through a conical or cylindrical chamber, the centrifugal force direct s the particle s toward the wall where they impact and become c harge d The particle chargers work on the principle of contact electrification; a particle can gain or lose charge depending on the state of the material with which it comes in contact. The important parameters for the particle charger are the electrostatic properties between the particles and the wall surface and the applied electric field (see Appendix A for the effect of the electric field on the charge transfer ). Theory Charge T ransfer M odel for M etal M etal S ystems I n metal metal contact electrification, electrons are transfer red between the surfaces. As a result, one material become s positively charge d while the other become s

PAGE 24

24 negatively charge d The charge transfer between the two metals after one contact withou t the influence of external electric field, is given by: (2 1) where is the contact capacitance between the two metals and is the contact potential difference that is directly dependent on the work function of the two metals. The w ork function is related to the amount of energy that is required to strip electron s from the surface; this value is strongly influence d by the surface state (e.g., the amount of valance electrons on the surface). The contact potential difference between the tw o metals is given by (2 2) where is the work function for each metal, and is the elementary charge of an electron. Electrons move from the metal with the lower work function to the metal with the higher work function. Contact electrification for two metals is fairly straightforward ; however, when an insulator is involve d, the theory become s more complex. Although controversy exists on the mechanism of the charge transfer for non metal contact (e.g. electron transfer, ion transf er, or chemical reaction at the contact site), the concepts in Equation 2 2 can still be appl ied for a n insulator metal or insulator insulator systems The work function of an insulator cannot be directly measure d. The surface state (e.g. roughness and geometry), the environment (e.g. air or vacuum), and other physical conditions directly n measured at a specific condition, is used in place of the metal work functi on (Castle 1997)

PAGE 25

25 Charge T ransfer M odel for M etal I nsulator and I nsulator I nsulator S ystems The typical charge transfer model that is applied for any two materials in contact is known as the condenser model (CM). The condenser model describes the contact between the two materials in terms of a capacitor. The contact surfaces are considered to be the two plates in the capacitor, and the charge is transferred due to the potential difference of the two plates or materials. The total potential difference assoc iated with the condenser model when particles are in contact can be expressed as: (2 3) where is calculated from Equation 2 1 and based on the effective work functions, and are the potential differences due to the image charg e and the space charge, respectively, and is the potential difference caused by external factors, e.g. an imposed electric field (Matsusaka 2011; Matsusaka et al. 2010). Image charge occurs when the electric field from a point charge induces a charge o n its surrounding. Space charge occurs when multiple point charges act together to form a cloud of charge. The terms and are both negative as both inhibit the charge transfer. The charge transfer in the condenser model can then be depicted as: ( 2 4) where is the charging efficiency and is the capacitance between the two materials. In the case of metal metal contact, is equal to one, is equal to and In addition, if then Equation 2 4 reduces to Equation 2 1 which describes the charge transfer for metal metal contacts in the absence of an external electric field.

PAGE 26

26 Charge Transfer Model for R epeated I mpacts To obtain the charge of a particle q as a function of the number of contacts n with a wall, a continuous quantity is used: ( 2 5) Equation 2 5 is the continuous form of the single contact ( n =1) equation depicted in Equation 2 4. During contact, and are proportional to ; they are generally defined as: ( 2 6) ( 2 7) where and are constants associated with the effect of the image charge and space charge respectively. The terms and are not a function of and are treated as constants. In addition to these contributions to charge transfer, Itakura et al. ( 1996) found that electrostatic leakage from the particle electric field to its surroundings, after the particles are no longer in contact, plays a significant role in the final charge of the particle. The charge leakage is caused by the redistribution of c harge on the particle surface. This effect is directly proportional to This term is given as ( 2 8) where is a constant associated with the effect of charge leakage The total charge transfer can then be represented by ( 2 9)

PAGE 27

27 where constant includes the effects of the image charge, space charge, and charge relaxation, and includes the effects of the work functions and the external electric fields. Equation 2 9 can be solved as a first order, non homogeneous differential equation with the initial conditions and resulting in ( 2 10) Equation 2 10 can be further generalized into the following form: ( 2 11) where is the initial charge of the particle, is the equilibrium charge of the particle when tends to infinity, and is a constant specific to the particle wall system. When applying this model to particle flow in a pipe, it can be assumed that the frequency of the particle wall impacts per unit pipe length is constant, and that the number of impacts is proportional to the pipe length Therefore, Equation 2 11 can be rewritten as (2 12 ) in Equation 2 12 is analogous to ; this term is known as the characteristic length. Finally, dividing Equation 2 12 through by the particle mass and replacing in Equation 2 12 by the charge to mass ratio i.e. the specific charge, yields the following equation which describes the par ticle charge as a function of pipe length: (2 13) E xperiments Experimental Setup The experimental system is composed of four sections including a feeder, a particle charger, a pipe, and a Faraday cup electrometer as shown in Figure 2 1. Figure

PAGE 28

28 2 1 A gives a detailed diagram of the feeder section, which consists of a vibratory feeder and a particle ejector. The vibratory feeder was made by placing a long stem funnel on top of a stainless steel plate, and piezo electric vibrator s were connected to the funnel and plate. The particle mass flow rate was controlled by the electric signal sent from a feeder controller (VST 01 Control system, IMP Co. Ltd.). Particle mass flow rate in all experiments is approximately 0.05 mg/s. Particle s leaving the feeder were drawn, by vacuum, into a particle ejector. The particle ejector consists of two sections upstream where the particles enter and downstream where the particles mix with the air and exit. A nozzle opening was used to separate the two sections and functioned to direct the compressed air toward the exit; in doing so, the compressed air created suction in the upstream section. The air flow rate was controlled using a pressure controller; during the experiment, the flow rate was kept a t 15 m/s. To prevent electrostatic disturbance, all components in the feeder section were grounded. After leaving the ejector, particles were sent to the particle charger in order to control the particle charge. The particle charger used was the same one d eveloped in with a conical chamber is shown in Figure 2 1 B Particles enter from the top and exit from the bottom of the charger; the broken line illustrates the particle f low path in the particle charger. The conical shape allows for maximum particle wall contact as the particles flow through the charger. Within the particle charger, there is a cone shaped electrode at the center. The inner wall of the charger, i.e. inner e lectrode, was designed in such a way that the cavity was 0.01 m wide. The outer wall of the charger, i.e. the outer electrode, was grounded during operation. Both electrodes were made using

PAGE 29

29 stainless steel. A high voltage source (Matsusada Precission Inc.) was connected to the inner electrode to generate an electric field between the inner and outer walls of the cavity. As particles come into contact with the outer electrode, the particles can gain or lose electrons as explained in Appendix A. The final cha rge of the particles leaving the charger directly depends upon the charge supplied to the inner electrode. Upon exiting the particle charger (or pipe), the particle charge was measured using a Faraday cup electrometer. As shown in Figure 2 1 C the Faraday cup was connected to a vacuum system in order to collect particles into a thimble filter. The charge reading from the inner conductive surface was sent to the electrometer (Takeda Riken Co. Ltd.). The mass of the particles collected was determined using an analytical balance (least count of 0.001g) The specific charge was obtained from the total charge and mass information. Short Pipe Experimentation with Varying Initial Charge The schematics of the setups for these sets of experiments are shown in Figures 2 2 A and 2 2 B The arrows on the left in Figure 2 2 represent particles exiting the particle ejector shown in Figure 2 1A In Figures 2 2 A and 2 2 B a 0.3m polyurethane tube connects the particle ejector to the particle charger. A 0.07 m glass tube was al so used to direct the charged particles out of the particle charger. The Faraday cup was placed at the exit of the 0.07 m glass tube to measure as seen in Figure 2 2 A The Faraday cup was also placed after the 1m pipe to measure seen in Figure 2 2 B Exit charge data for the particles were collected for each particle pipe material combination by varying the initial charge applied to the particles using the particle charger.

PAGE 30

30 Validation Experiments with Longer Pipes In order to validate the proposed technique, the exit charge data for the particles were collected at different lengths of pipe. The experimental setups are shown in Figure 2 2 C and 2 2 D In these setups, a 0.1 m brass pipe is placed after the particle ejector. The Faraday cup was positioned as seen in Figure 2 2 C to measure for these validation experiments. The setup in Figure 2 2 D was employed to collect particle charging data after 1 2 3 and 4 m pipe length. The air flow rate in the pipes was measured after the ejecto r and after 1 2 3 and 4 m pipe to ensure that there were no changes in the flow rate with each additional pipe length. Particles and Pipes A total of five different types of particles and four different types of pipe wall materials were tested. A sum mary of particles and pipes with their specifications are found in Table 2 1 and 2 2 respectively. Due to the extensive amount of data, only the results of borosilicate particles flowing in a natural glass pipe and in a copper pipe will be discussed as a r epresentative example. The rest of the data are included in appendix B. Result s and Discussion Varying Initial Charge in the Particle Charger The initial charge of the particles entering the 1m pipe was adjusted using the particle charger (Figure 2 2 A) Th e initial charge was measured at different voltages applied to the inner electrode of the charger. Figure 2 3 shows the specific charge of the borosilicate particles as a function of electric field strength in the particle charger. The polarity of the part icle charge could be changed from negative to positive. However, there is a limit to the particle charging with higher applied electric field because of

PAGE 31

31 electrical discharge in the particle charger. Applying voltage higher than this limit causes a decrease in particle charge; this is demonstrated by the maximum in the curve shown in Figure 2 3. Short Pipe Experimentation with Varying Initial Charge Particle charge measurements were made using the setup in Figures 2 2 A and 2 2 B The initial charge of the pa rticles is denoted and the particle charge collected after flowing in the 1m pipe is denoted The particle charge measurements were repeated three times. A relationship between and for borosilicate particles in the natural glass pipe is shown in Figure 2 4. The versus curve exhibits a linear relationship, which agrees with Equation 2 9 in the theory on the repeated impacts model. Matsuyama and Yamamoto (1995) observed sim ilar behavior in their single particle impact experiments. Figure 2 4 also includes a broken line representing The experimental data for versus are higher than the broken line; this result indicates that the particles obtained a posi tive charge from the wall by contact electrification. In other words, the electrons on the particles were transferred to the wall. Figure 2 5 shows the experimental results of versus for borosilicate particles in a copper pipe. The versus curve again exhibits a linear relationship. However, in this case, the experimental line intersects with a line representing at a value of charge was within the range in initial particle char ges tested, and the value of the equilibrium charge is

PAGE 32

32 negative) charge, if borosilicate particles flow through an infinitely long pipe, the particles will consistently reach this equilibrium charge value. Predicting P article C harging using S tair S tepping P rocedure A technique to predict particle charge in any given length of pipe is proposed in Figure 2 6. The solid curve, i.e. versus line, is the same as that in Figure 2 4, which is the result for the borosilicate particles traveling through the natural glass pipe. First, an initial charge on the line is found. This point is labeled as From this point a vertical line to the versus line is drawn. This verti cal line represents the final charge on the particles after traveling through 1m of the natural glass pipe with initial charge A horizontal line is then drawn to the line in order to set a new initial charge If this stair stepping proces s is repeated, the final charge of the particles at any given pipe length can be determined. The versus line is not parallel to the line; therefore, at a certain point the two lines will intersect. This intersection point represents the value of the particle charge when it has reached its equilibrium state, i.e. the particles will no longer gain or lose charge, regardless of continued interactions with the pipe surface. Predicting P article C harging using the S emi T heoretical E quation s The experimental measurements in the short 1 m pipe can be used to determine the two constants and in the semi theoretical Equation 2 13, which describes the particle charge as a function of pipe length. Those two parameters and are easily found using the relationship in Figure 2 4. In Figure 2 4, the specific charge of particles traveling thro ugh a pipe can be expressed as the following linear equation: (2 14 )

PAGE 33

33 where and are constants, The equilibrium charge can be calculated by finding the intersection point between Equation 2 14 and line Mathematically th is equilibrium charge is given by (2 15 ) To order to determine a comparison of Equation 2 14 to Equation 2 13 results in: (2 16) or (2 17 ) The resulting constants and for each type of particle in each pipe wall material are summarized in Table 2 3 The values of are 11.3 m and 3.9 m for the borosilicate natural glass and the borosilicate copper systems, respectively. In general, as the value of increases, the slo wer the charge transfers between the particles and the wall. Based on the values of in Table 2 3 the insulator metal system exhibits higher charge transfer than the insulator insulator system. Figure 2 7 A shows the predicted charging profiles for boro silicate particles in a natural glass pipe using Equation 2 13 with and determined from the short pipe experimentation. Figure 2 7 B shows the predicted charging profiles for borosilicate particles in a copper pipe. The multiple lines in both figures show the predictions for different initial charges The particle charge approaches the equilibrium value more quickly (i.e. shorter pipe lengths) for the borosilicate copper pipe (insulator metal) system than for the borosilicate glass (insulator insulator) pipe system. The charging

PAGE 34

34 profiles are symmetrical with respect to the equilibrium for the same absolute value of T ech nique Validation To validate the technique for predicting particle charging in any given pipe length, particle charging data were collected using different lengths of pipe as shown Figures 2 2 C and 2 2 D Again, the particle charge measurements were repeate d three times. These data were compared with the predicted charging profiles as given by equation 13 with values for and based on the short pipe experimentation. For all the particle pipe wall material combinations tested, the comparisons between p redicted and measured charging profiles are in good agreement. Figure 2 8 A compares the predicted charging profile to the experimental data for borosilicate particles flowing in the natural glass pipe. Figure 2 8 B shows the comparisons between predicted an d measured charging profiles for borosilicate particles in the copper pipe. Two additional representative comparisons are shown in Figures 2 8 C and 2 8 D Figure 2 8 C presents the comparisons for PMMA spherical particles in a stainless steel pipe, and Figur e 2 8 D for non spherical crushed soda lime glass particles in a Pyrex pipe. The small deviations are likely caused by variations in the surface state of pipes. Summary A predictive technique for particle charging behavior in pipes of any given length has b een developed and tested for a range of particle characteristics and pipe wall materials. This predictive technique is based on experimentation in a short pipe section with varying initial particle charge. The predictive technique is applicable to any meta l metal, metal insulator and insulator insulator particle wall system with two limitations.

PAGE 35

35 First, particle charging must be via the impact charging mechanism. For example, this predictive technique would not be applicable to dense phase pneumatic conveyin g where particles are in sustained contact with the wall. Second, the particle and pipe wall materials and the operating conditions (e.g. air and solids flow rate) in which and are determined from the short pipe experimentation must be the same as the case in which particle charging is being predicted in longer length pipes. Given these two limitations, this predictive technique should work well for a full range of particle characteristics (including initial charge) and pipe wall materials

PAGE 36

36 Table 2 1. Characteristics of particles used in the experiment Type of particle Shape Mass median diameter Mass mean Standard Borosilicate Spherical 49 48 3 Soda lime Spherical 79 78 8 PMMA Spherical 67 67 16 Crush soda lime Non spherical 66 66 16 JSC 1A (30 90 Non spherical 55 54 16 Table 2 2 Characteristic of pipes used in the experiment Material of pipe Inner diameter (m) Copper 0.0070 Stainless steel 0.0070 Natural glass 0.0068 Pyrex 0.0068

PAGE 37

37 Table 2 3 Characteristic values for each particle in 1 m pipe Material of particle Material of pipe c ( ) d L 0 (m) q m ( ) Borosilicate Copper 0.770.20 13.42.0 3.90.3 594 Stainless steel 0.260.29 83.82.9 0.70.7 11311 Natural glass 0.910.47 60.22.4 11.30.5 71119 Pyrex 0.950.40 24.13.2 19.10.4 47314 Soda lime Copper 0.730.23 4.72.1 3.10.4 173 Stainless steel 0.240.22 42.36.5 0.70.6 567 Natural glass 0.910.20 19.52.4 10.30.3 2127 Pyrex 0.920.29 14.73.9 11.40.4 1758 PMMA Copper 0.560.56 339.813.8 1.70.7 76627 Stainless steel 0.490.64 736.814.8 1.40.7 145143 Natural glass 0.590.53 204.88.9 1.90.6 50421 Pyrex 0.700.53 111.49.2 2.80.4 36813 Crush soda Copper 0.800.33 3.11.2 4.40.4 153 lime Stainless steel 0.590.28 2.50.7 1.90.4 62 Natural glass 0.910.35 15.91.4 10.70.4 1798 Pyrex 0.960.29 12.51.0 22.20.4 2859 JSC 1A Copper 0.760.27 25.41.7 3.60.4 1056 (30 Stainless steel 0.390.33 27.31.0 1.10.6 456 sieve fraction Natural glass 0.880.18 30.81.9 8.10.3 2657 Pyrex 0.950.32 10.30.4 19.90.4 2118

PAGE 38

38 A B C Figure 2 1 Schematic of experimental units A ) P article feeder B ) P article charger C ) F araday cup electrometer All units are in mm

PAGE 39

39 A B C D Figure 2 2. Ex perimental setup for measuring particle charge. A ) A fter leaving the particle charger B ) A f ter 1 m pipe length C ) A fter the ejector, D ) A fter different pipe length

PAGE 40

40 Figure 2 3. Specific charge of borosilicate particles leaving the particle charger as a function of applied electrica l field in the particle charger

PAGE 41

41 Figure 2 4. q m ( L =1) vs q m,0 for borosilicate particles in 1m natural glass pipe

PAGE 42

42 Figure 2 5. q m ( L =1) vs q m,0 for borosilicate particles in 1m copper pipe

PAGE 43

43 Figure 2 6. Predicting particle charge using stair step method

PAGE 44

44 A B Figure 2 7. Predicting charging profile for borosilicate particles A ) N atural glass pipe B ) C opper pipe

PAGE 45

4 5 A B C D Figure 2 8. Comparison of the experimental data with predicted charging profile A ) B orosilicate particles in natural glass pipe B ) B orosilicate particles in copper pipe C ) PMMA particles in stainless steel pipe D ) Crush s oda lime particle in Pyrex pipe

PAGE 46

46 CHAPTER 3 CHARGE PREDICTION US ING COMUTATIONAL FLU IDS DYNAMICS (CFD) SIMULATION Background Using the previously determined parameter ( and ) in Chapter 2, it is possible to determine the charging using computer simulation. Previously only a few attempts were made to create a predictive model for particle charging. Tanoue et al. (2001) used coupled CFD DEM method to predict 2D particle traject ory. Turbulent gas phase was model using Navier stokes equations then solve s the particle motion using force balances. The charge transfer at each collision was derived theoretically. They had found that the number of collision in pipe is a function of S to kes number. Cangialosi et al. (2006) use d direct simulation Monte Carlo (DSMC) simulations to statistically determine the charge on particle and pipe at any time. In DSMC uses Boltzmann equation and Knudson number to determine the flow for fluid and partic les. The DSMC simulation was able to predict particle charge in straight pipe; prediction for cu rve pipe was also possible with fitting parameter. These two types of simulation provide detail result but require high computational cost. In this work the co ntinuum CFD model was employed to determine the collision frequency in the pipe. Unlike DEM and DSMC simulation, the continuum model is performed with much lower computational cost, minimizing the computational time. The advantage makes the continuum model an attractive model for use in industry. With the parameters determine in the experiment and the collision frequency determined by CFD simulation, it is now possible to accurately predict the particle charge of the same particle pipe system under any oper ating condition (e.g. pipe length, pipe diameter, flow velocity, and particle loading).

PAGE 47

47 Model Description Governing Equations A Eularian model is used for modeling the diluted gas solid particle system in a pipe. The governing equations follow the volume a verage equations of Anderson and Jackson (1967) The continuity equations are given as: (3 1 ) (3 2 ) Where is time, is the solid volume fraction (Note that is the gas volume fraction), is the mean gas velocity, and is the mean solid velocity. Equation 3 1 is the gas phase continuity equation and Equation 3 2 is the solid phase continuity equation. The momentum equations are: (3 3 ) (3 4 ) is pressure, is the gas density, is the solid density, and is the acceleration due to gravity, is the gas phase stress, is the solid phase stress, and is the drag force. Equation 3 3 is the gas momentum equation and Equation 3 4 is the solid momentum equation. Closure models are required to solve the continuity and momentum equations. The closure models will determine the drag force the gas phase stress and solid stress Drag M odels The drag models represent the momentum trans fer between the gas phase and the solid phase. The drag force is define as

PAGE 48

48 ( 3 5) The drag coefficient are determined semi theoretically. The drag coefficient follows the model by Haider and Levenspiel (1989). This model uses sphericity to mod el drag for both spherical and non spherical particle. Sphericity refers to how spherical a particle is. It is express as the ratio of the surface area of sphere with the same volume as the particle to the surface area of the particle. The Haider and Leven spiel is given as (3 6) using (3 7) (3 8) (3 9) (3 10) (3 11) Gas P hase S tress The turbulent model employed is the standard k balances the convective and diffusive energy transport with the generation and dissipation of turbulent kinetic energy. The k (3 12) (3 13)

PAGE 49

49 The two phase k model (Equation 3 14 and 3 15) is similar to the single phase model. An additional term is the interaction term caus e by the solid phase. The addition of particle can result in enhancement or damping of the turbulent kinetic energy. The gas phase stress is calculate using (3 14) The gas phase stress is made up of two parts the stress cause by the intrinsic viscosity of the fluid and the stress cause by turbulent is determine by using equation from Rao et al. (2011). (3 15) and the turbulent stre ss contribution is (3 16) is the maximum solid volume fraction. The constants are summarized in Table 3 1. Solid S tress In the case of diluted pipe flow, the solid stress is separate out into the kinetic solid stress and the collisional solid stress The frictional solid stress is excluded f rom the model for diluted flow. The equation for solid stress is (3 17) The kinetic solid stress contains a damping function to limit this stress from approaching infinity. Both the kin etic and collisional solid stress are cause by the motion of particles ; both stresses are depend on the particle fluctuating velocity The magnitude of particle fluctuating velocity is call granular temperature (3 18)

PAGE 50

50 The granular temperatur e balance is (3 19) where is the granular conductivity, is the dissipation due to collision, and is the granular temperature modulation. Additional equations are required to close the granular temperature equation. The equations for the solid stress follow the stress model theoretically derived by Lun et al. (1984). The solid stress equations are as follow: (3 20) (3 21) (3 22) (3 23) (3 24) The definition of and are listed in Table 3 1. Turbulence Interaction Two interaction terms are seen in the model: in the k model equations (Equation 3 12 and 3 13), and in the granular temperature equation (Equation 3 19). The interaction terms represent the energy transfer in the fluctuating component between the t wo phases. Louge et al. (1991) simplified the interaction terms into (3 25) (3 26)

PAGE 51

51 is known as the cross correlation model. Many authors proposed different expression for The model used w ill follow the cross correlation model derived by Sinclair and Mallo (1998) and validated by Rao et al. (2011). (3 27) Computer Simulation Computational simulations of two phase pipe flow were performed using Equation 3 1 to 3 27 The equations are discretized using the finite volume based method describe in Patankar and Spadling (1967). No slip boundary conditions are used for the gas phase. The solid phase boundary conditions follow Johnson and Jackson (1987). Boundary condition for k and follow Bolio et al. (1995). The inputs and outputs for the code are listed in Table 2 2. The mass loading is the mass ratio of particle flow rate to gas flow rate. The specularity coefficient is semi empirical constant arising from momentu m loss during wall collision ; this term represents the surface roughness of the wall. Determining C ollision Number The collision frequency can be calculated using the particle fluctuating velocity and the inter grain spacing The particle fluctuati ng velocity is determined from the granular temperature using Eq uation 3 18 In this case of axisymmetric fully developed flow, the particle fluctuating velocity is the random movement of the particle in the direction, (i.e. particle velocity toward o r away from the pipe wall). The inter grain spacing is calculate using (3 28)

PAGE 52

52 The inter grain spacing is the average distance from particle surface to other surfaces. Beyond this length particle will come into contact with another particle or the wall. The local collision frequency of the solid phase at any radial point in the pipe is ( 3 29 ) The unit of is collision per time. The charge transfers will occur during particle wall contacts. The particle wall collision is Eq uation 3 29 calculated at the wall. The total particle wall collision number for all the particles flowing through the pipe adjusted for particle mass is found by multiplying the particle wall collision by the particle residence time determined from the superfi cial velocity ( 3 30 ) where is the velocity of the particles. Using from Equation 3 30 it is possible to determine particle charging in a system using ( 3 31 ) Equation 3 31 is another form of Equation 2 11, this equation determine the specific charge for multiple particle system based on the number of collision. Instead of looking at the particle wall collision number for a single particle Equation 3 31 uses the total particle wall collision number for all particle s The term in Eq uation 3 31 is determined by relating the term, determined from the experiment in Chapter 2, to the determined for that specific experimental condition. Another collision parameter that is useful to know is the particle wall collision number for a single particle along the pipe length This is estimate using the

PAGE 53

53 particle collision in a 1 m pipe increment. At any given time the total number of collision in a 1m pipe and the number of particle in a 1 m pipe are ( 3 32 ) (3 3 3 ) Where is the number of particle in per unit mass and is the average particle mass flow rate. The number of collision in a 1m pipe (i.e collision per length) is (3 3 4 ) The term is useful for compari son of the particle collision in the pipe in each system; this term i s independent of particle flow rate as well as pipe length. In an experimental setting, Matsusaka et al. (2002) found to be around one collision per 77 Experiments The values for the charging parameters are taken from Table 2 3. Experiments collecting the specifi c charge of particle at different pipe lengths and flow rates were performed using the experimental setup describe in Chapter 2 (Figure 2 2 C and 2 2 D ). In these experiment four different pressure setting were used (refer to as setting 1 to 4). The flow rat e and velocity for each pressure setting are listed in Table 3 3. The particle flow rates for these experiments are between 0.2 to 1.0 mg/s. Simulation CFD simulations were performed to match the experimental conditions that were used in the experiment (bo th the short pipe experiment, and the validation experiment). The inputs were measure for each experiment. The properties of the particles and the

PAGE 54

54 pipe are listed in Table 3 4 and 3 5; unlike in Chapter 2, the particle diameters used are the count mean sur face area diameter. The specularity coefficient for each pipe was determined semi empirically to be 0.02 for glass pipe and 0.08 for metal pipes (Bolio et al.1995). The gas properties were taken at standard condition. Finally the coefficients of restitutio n used were taken from reported values in the work by various authors (Brauer 1980; Gondret et al. 2002; Sommerfield and Huber 1999). The coefficients of restitution are between 0.9 and 1. Grid size of 200 points in the radial passes the grid sensitivity t est. The model was validated using the single phase simulations for Re = 7,000 (Eggel et al. 1994), Re = 21,800 (Schildknecht et al. 1979), and Re = 50,000 (Laufer 1953). The velocity profile and the turbulent kinetic energy profile match that of the valid ation data. An example is shown for Re = 21,800 in Figure 3 1. The two phase simulations were validated by Rao et al. (2011). Results and Discussion Particle Wall Collision CFD simulations were used to determine the particle wall collision in the pipe. The outputs for CFD simulation were used along with Equation 3 28 through 3 34 to determine the two collision number and for all the experiment performed were calculated to be between 0.1 to 1 collisions per meter, depending on the condition. The collision numbers are on the same order of magnitude as those observed in experiment by Matsusaka et al. (2002) The validating experiments varied the operating condition (i.e. air velocity and particle flow rate). CFD simulations were performed to check the effect of each operating parameters on the collision number. The simulations were performed under a

PAGE 55

55 typical co ndition for the soda lime particles in copper pipes; performing simulations for other particles and pipes also yields similar trends. Figure 3 2 shows the collision number of the particle, traveling along a straight pipe. Each line in Figure 3 2 repres ents the gas velocity used in the validation experiment. In Figure 3 2, the particle mass flow rates were kept constant at 0.58 m/s. The simulations show that as flow velocity increases the number of collisions should increase as well. This result is expec ted. In general increasing the flow velocity increases the number of particle wall collision in the pipe. Higher flow velocity increases the turbulent in the pipe. The higher the turbulent level in the flow will increase the likelihood of particle coming i n contact with the wall. This result was found true when simulation were performed for other velocity as well. From this relationship, more charge transfer will be seen for experiments operating at higher velocity. Another parameter that was varied in the experiments is the mass flow rate, In the simulation the changing the mass flow rate changes the mass loading term (see Table 3 2). The collision numbers along the pipe length were determine d from the simulation at different mass loading this is shown in Figure 3 3. The relationship in Figure 3 3 indicates that as the particles flow rates entering the pipe increases, the fewer the collisions will occur for an individual particle. This would be possible when the addition of particles hindered the particl e near the centerline from reaching the wall. These centerline particles will not collide with wall thus lower the average collision number. From this relationship, it is expected that less charge transfer will be observe when the mass flow rate increases. However the particle feeder used in the experiment

PAGE 56

56 does not have the capability to precisely control the particle feed rate to test this effect. A parametric study of each parameter is presented in Appendix C. Determining CFD simulation was done for each individual experiment ran in the short pipe experiments ( Figure 2 2 A and 2 2 B ) to determined the collision number Under the assumption that the number of collision in the pipe is constant it is possible to convert (use in Equation 2 13) into (use in Equation 3 31 ). was determined using ( 3 35 ) in Equation 3 35 is calculated using Equation 3 3 0 for the experimental condition specific to the experiment that was used to determined ; and L is 1 meter Equation 3 35 directly converts the characteristic length into the characteristic collision number, for each particle pipe system is also listed in Table 3 6. With and Equation 3 31 prediction of particle charge is possible but the operating conditi on must remain the same as the short pipe experiment as the number of collision changes with the operating condition. But with and Equation 3 31 we expect to be able to predict particle charging for the same particle pipe system but with different o perating condition. Comparisons to E xperimental R esults Validation experiments were performed for each particle pipe system at each pressure setting for pipe lengths of up to 4 m (see Fig ure 2 2 C and 2 2D ) A simulation was performed to match each experime ntal condition The collision number determined under each condition was used with and to predict the particle charge. The predictive results are compared to the experimental results. Example result s are shown

PAGE 57

57 for soda lime particle s in copper p ipe (Fig ure 3 4 ) and JSC 1A in natural glass pipe (Fig ure 3 5 ). In both Fig ures 3 4 and 3 5 the points are the experimental measurement and the solid lines are the prediction made using CFD simulations. Each experiment was performed at different velocity and particle mass flow rate ; these are displayed for each figure. The collision number along the pipe length is also label in each figure. In both Fig ures 3 4 and 3 5 the general trend from the experiment is that as the flow velocity increases the am ount of charge transfer to/from particle also increases, this is in agreement with the trend predicted in Figure 3 2 In both figures, the predictions were able to capture the effect of impact charging. In Fig ure 3 4 the predicted results were very simil ar for each simulation; the soda lime copper pipe system reaches equilibrium. The equilibrium charge was reach in the prediction but the experiment result did not reach equilibrium. This deviation is likely to due to experimental error and the environmenta l condition (i.e. humidity and temperature) and inhomogeneity of the pipe material causing some differences in This deviation did not cause significant changes in the prediction before reaching the equilibrium charge. The JSC 1A natural glass pipe pr edictions in Fig ure 3 5 also show very good match with the experimental results when the particles had not reach the equilibrium charge. Similar results were seen for other particle pipe system. The result s for other particle pipe system are in Appendix D. Expanding the Prediction Model Scale Up the Pipe Line The experiments were performed at a constant pipe size. In process design it is given that the pipe size will be scaled up in the actual processes; thus it is important to

PAGE 58

58 validate that this predictiv e method will also be valid when the pipe size changes. In the current method, charge is transferred as a function of particle wall collision. When changing the pipe size, not much is expected to change in the way particles impact the wall. The experiment and the method described in this work are expected to be valid when increasing or decreasing the pipe diameter. However it is necessary to test this hypothesis. In the future, experiments and simulations should be performed with different pipe diameters to determine whether it is possible to predicted particle charging for different pipe diameters. Doing so will validate the predictive technique for scaling up the pipeline. Monitoring the Environmental Condition While performing the experiments it was foun d that the environmental condition, namely the humidity and temperature, strongly affected the charge measured on the particle. We had speculated that both temperature and humidity mainly affect the equilibrium charge of the particle and the wall, but more testing is needed. In the future, this experiment should be performed in a control chamber. The humidity and the temperature should be accurately and easily controlled. The experiments measuring the equilibrium charge and the characteristic length should be performed under a range of relative humidities and temperatures. From this data it should be possible to create an empirical model that modifies the equilibrium charge and characteristic length between the two materials as a function of the relative hu midity and temperature. (3 36) (3 37)

PAGE 59

59 RH is the relative humidity, T is the temperature, and the superscript indicates the measurement made at a reference condition possible at room temperature and dry air. We expected t hat these general functions and will be independent of the material properties, but that remains to be determined. Experiments should also be performed to check the validity of this model after including the environmental effect. This validation will t ake into account the day to day variation as well as the location. Scaling the Particle Size The current predictive method is only valid for the same particle size as the experiment. In some cases, it is important to change the particle size in the system. To take into account the changes in the particle diameter, it is important to know two different parameters: the surface of contact and the duration of contact. Next it is important to determine whether the charge transfer between the surfaces is limited by the surface area in contact, the duration of contact, or both. If the charge transfer occurs very slowly, a greater surface contact will not make a significant difference on the final charge on the particle. Thus the longer the contact, the longer the c harge transfer. And in the other case, if the charge transfer occurs very quickly (as soon as the contact occurs) the contact surface will reach a charge equilibrium with the wall. The contact area between the two materials will control how much charge is transferred. The most likely scenario will be that the charge transfer is dependent on both the contact area and contact time. A model is needed to describe the charge transfer for a collision. Once the charge transfer model is made, it will be possible to convert the number of collisions used in the predictive equation to a new term including contact area and contact time.

PAGE 60

60 Accounting for the Complex Flow Geometry Ema et al. (2003) found that the charge transfer between the particles and the surfaces depend s on the angle of impact. Charge transfers at certain angles were found to be as much as five time greater than direct collisions at 90. Ema et al. indicated that this effect is due to the rolling and slipping motion of the particles. The highest charging efficiency occurs at a critical angle when particles switch from rolling to slipping. In the current technique charge prediction is only valid for a straight pipe. In a straight pipe the angle of impact was found to be consistently around 80; this effec t can be neglected. However in a more complex pipeline (i.e curved, bent, and split) the impact angle will change. In order to take into account the different angles of impact, first the critical angle in which slipping and rolling occurs should be determi ned. Theoretical models can be made to determine this based on the roughness of the two materials. Then it would be possible to incorporate this relation as (3 38) where is the charging efficiency of the impact angle. Using this, we can perform simulations with different flow geometries to determine particle wall collision as well as the average impact angle. With Equation 3 38 each section of the geometry needs to be mod eled separately depending on the average impact angle. It may be possible that the effect of the impact angle will tie into determining the particle wall contact area and contact time. If what Ema et al. hypothesized is true, then the contact area will ch ange depending on the rolling and slipping motion of the particle. But it is also important to take into account that when the particle is slipping and rolling, the charge transfer will also be different due to the friction created by the particle slip.

PAGE 61

61 Su mmary Particle charge in straight pipe was predicted using the particle pipe parameters determined from an experiment along with the total number of particle wall collisions determined with CFD simulation. For the experiments, particles were sent into a st raight pipe, and the particle charge leaving the pipe was measured. By varying the initial charge it was possible to determine the particle wall equilibrium charge and a charge transfer parameter. With this information a CFD simulation can be perform to de termine particle wall collision number for a single particle under any operating condition (gas velocity and particle flow rate). This prediction method worked well for both spherical particles and non spherical particles. Determining of collision frequenc y is not limited to use of CFD simulation. CFD DEM is also encouraged but will result is much higher simulation time. Excellent predictions were observed for different types of particle flowing through straight pipes for a range of gas velocities and parti cle mass flow rates.

PAGE 62

62 Table 3 1 Term used in the turbulence and solid stress model

PAGE 63

63 Table 3 2. Inputs and outputs for the simulation Parameter Description Input mass loading ( ) pipe radius (L) gas viscosity (ML 1 T 1 ) gas density (ML 3 ) mean gas velocity (LT 1 ) maximum solid volume fraction ( ) particle diameter (L) particle density (ML 3 ) sphericity ( ) particle particle coefficient of restitution ( ) particle wall coefficient of restitution ( ) specularity coefficient ( ) Output solid volume fraction ( ) gas velocity in the z direction (LT 1 ) particle velocity z direction (LT 1 ) granular temperature (L 2 T 2 ) turbulent kinetic energy (L 2 T 2 ) turbulent dissipation (L 2 T 3 ) Table 3 3. Air flow in each type of pipe at each pressure setting Pipe Pressure setting Air flow rate (m 3 /s) Velocity (m/s) Reynolds Number Metal pipe (copper & Stainless steel) 1 0.000257 6.7 3120 2 0.000482 12.5 5860 3 0.000805 20.9 9760 4 0.001290 33.5 15600 Glass pipe (natural glass and Pyrex) 1 0.000285 7.9 3560 2 0.000534 14.7 6670 3 0.000891 24.5 11100 4 0.001420 39.3 17800

PAGE 64

64 Table 3 4. Characteristics of particles used in the experiment Material of particle Density (kg/m 3 ) Diameter Sphericity Maximum solid volume fraction Borosilicate 2220 48 1.00 0.65 Soda lime 2580 75 1.00 0.65 PMMA 1380 50 1.00 0.65 Crush soda lime 2670 43 0.76 0.47 JSC 1A 3230 27 0.86 0.53 Table 3 5. Characteristic of pipes used in the experiment Material of pipe Inner diameter (m) Specularity coefficient Copper 0.0070 0.008 Stainless steel 0.0070 0.008 Natural glass 0.0068 0.002 Pyrex 0.0068 0.002 Table 3 6 Characteristic values for each particle in 1 m pipe Material of particle Material of pipe ( ) (m) (#) Borosilicate Copper 594 3.90.3 67.4 5.8 Stainless steel 11311 0.70.7 12.812.9 Natural glass 71119 11.30.5 115.1 4.8 Pyrex 47314 19.10.4 98.4 2.2 Soda lime Copper 173 3.10.4 23.02.8 Stainless steel 567 0.70.6 5.1 4.6 Natural glass 2127 10.30.3 44.81.3 Pyrex 1758 11.40.4 49.31.6 PMMA Copper 76627 1.70.7 14.95.8 Stainless steel 145143 1.40.7 12.36.6 Natural glass 50421 1.90.6 11.33.7 Pyrex 36813 2.80.4 16.42.7 Crush soda Copper 153 4.40.4 7.77.4 lime Stainless steel 62 1.90.4 32.87.9 Natural glass 1798 10.70.4 176.56.8 Pyrex 2859 22.20.4 365.66.0 JSC 1A Copper 1056 3.60.4 130.314.1 (30 Stainless steel 456 1.10.6 38.321.1 sieve fraction Natural glass 2657 8.10.3 179.06.5 Pyrex 2118 19.90.4 439.48.4

PAGE 65

65 A B Figure 3 1. Comparison of single phase prediction to benchmark data by Schildnecht et al. (1979) A) Velocity profile. B) Turbulent production profile.

PAGE 66

66 Figure 3 2. C ollision number for soda lime particle in copper pipe at different velocities, the mass flow rate is constant at 0.58 mg/s Figure 3 3. Collision number along the pipe for soda lime particle in copper pipe as a function of mass flow rate and velocity

PAGE 67

67 A B C D Figure 3 4. P redicted charging profile for soda lime particles in copper pipe for different air velocity and mass flow rates

PAGE 68

68 A B C D Figure 3 5. P redicted charging profile for JSC 1A particles in natural glass pipe for different air velocity and mass flow rates

PAGE 69

69 CHAPTER 4 DEPOSITION OF CHARGED PARTICLES IN BIFURCA TING AIRWAY Background The charge on a particle is affected by its properties, e.g. size and composition, as well as by the method of charging. The presence of charge in a particulate mixture can dramatically influence its flow behavio r, and understanding this effect has the potential to improve many particulate processes, including those found in medicine (Bailey et al., 1998; Peart et al., 2003). For example, knowledge of the behavior of charged particles can lead to more efficient pa rticle drug delivery systems using inhalers. Researchers have found that aerosols generated from inhalers contained some amount of charge on the particles (Kwok et al., 2005; Byron et al., 1997) Ali et al. (2009) measured the amount of charged particles g enerated from a commercial metered dose inhaler (MDI) and a dry powder inhaler (DPI) and found that up to 40% of the generated particles contained charges. To date, researchers have performed experimental deposition studies with uncharged particles in airw ays; the common experimental technique that is employed is the single breath regional deposition (SBRD) technique (Brand et al. 2005; Ferin et al., 1983; Yu and Finlay, 2005). SBRD compares inhaled and exhaled particles to determine the amount of deposited particles but cannot provide local deposition pattern information. For local deposition patterns, visualization techniques such gamma camera, single photon computed tomography (SPECT), and fluorescence are commonly used. Although gamma camera and SPECT c an be used in vivo, a clear deposition image requires a long acquisition time and a high dose of radiation ( Biddiscombe et al., 1993; Brown et al., 2002; Eberl et al., 2001; Dunbar et al., 2002 ;

PAGE 70

70 Dolovich and Labiris, 2004). Fluorescene techniques do not h ave these drawbacks, but they are limited to airway models and measurements are rather tedious. For example, Oldham et al. (2000) deposited fluorescence tagged polystyrene particles into a silicone airway model of the lung and then the airway was cut in h alf and deposited particles were counted using a fluorescence microscope. Similarly, Kim and Fischer (1999) deposited urine tagged oleic acid particles in glass airway models and then local deposition was determined by dissolving, in water, the particles i n each section. Both groups found that the majority of uncharged particles deposited near the regions where airway bifurcations occurred, with very little deposition occurring in the straight conduit sections of the airway, and even less in the inlet airwa y. In comparison to uncharged particles, there is a paucity of research on charged particle deposition in airways. Melandri et al. (1983) and Prodi and Mularoni (1985) used the SBRD technique with human volunteers and found that charged particles deposit t wo times more than their uncharged counterpart. Chan et al. (1978) tagged Fe 2 O 3 particles with a technetium 99m radiotracer, then aerosolized and charged the particles using a spinning disc aerosol generator, to determine particle deposition in a rubber ai rway model of the human lung. The charging of the particles occurred by triboelectrification between the particles and the spinning disc, and the deposition patterns were determined using a gamma counting system. Chan et al. also observed an increase in pa rticle deposition for charged particles, with the enhancement in deposition greatest for particles less than 5 m and occurring in the middle section of the trachea. Chan et al. concluded that the increase in deposition was due to the image

PAGE 71

71 force the att raction of a charged particle toward the wall due to the difference in electrical charge. Complementary computational fluid dynamics (CFD) simulations and experiments of charged particle deposition in a single bifurcating glass airway model were recently performed by Leong et al. (2009) The particles were charged using corona charging, and an excessive amount of particles were deposited so that they could be viewed with the naked eye. They observed that uncharged particles deposit mainly in the bifurcating regions, which was attributed to inertial impaction and gravity, while the charged partic les deposited on both the straight airway section and the bifurcating region. Their CFD simulations showed that the electrostatic force significantly increases deposition when particles are in the near wall region, and that the overall deposition efficienc y for charged particles was four times greater than uncharged particles. In this work, a relatively new, simple and non intrusive technique near infrared (NIR) fluorescence was used to determine local particle deposition in a model airway. The electros pray technique was employed to vary the charge on the particle, as well as to consistently generate particles with varying morphologies. With this novel combination of techniques, the effects of particle charge and morphology are explored, and local deposi tion patterns can be easily detected and measured. Electrospra y Technique Production of C harged, S pherical P articles Electro hydro dynamic atomization (electrospray) is capable of producing a constant stream of monodispersed particles from a liquid solutio n. A general schematic of an electrospray system is given in Figure 4 1 The spraying solution is pumped

PAGE 72

72 through a small (mm) nozzle, and a high voltage is applied to the nozzle with a ground counter electrode positioned below the nozzle. At the tip of the spraying nozzle, the electrostatic force binds the spraying solution into a cone shape, known as the Taylor cone. The spraying solution then atomizes into droplets at the tip of the Taylor cone. Electrospray generates particles with a very narrow size dis tribution; these particles also contain an electrostatic charge, which can be controlled using corona discharge. The particles that are generated by electrospray depend on the liquid solution properties; namely, the polymer concentration, density, surface tension, viscosity, and conductivity of the spraying solution. Hartman (1998) formulated a scaling law for electrospray, which defines the droplets generated from electrospray based on the physical properties of the solution. The scaling law is based upon two different spraying modes the varicose break up mode and the whipping mode. In the varicose break up mode, the Taylor cone is stable; monodispersed particles are generated in this mode. In the whipping mode, the Taylor cone is no longer stable but is in constant movement; the liquid stream leaving the cone is in the form of a whip. The particles generated in the whipping mode are polydispersed. The scaling law requires knowledge of the jet current (I) at the spraying cone. For a Newtonian fluid with a flat radial velocity profile in the spraying cone, the jet current is calculated by ( 4 1) where is the solution surface tension, is the solution conductivity, and is the solution volumetric feed rate (Hartman, 1998). The scaling law approximates the droplet size generated in each spray mode as:

PAGE 73

73 (4 2) (4 3) where is the permeability in a vacuum and is the density of the spraying solution. For a given solution and feed rate, the smaller of the two droplet sizes calculated for and from equations 2 and 3 indicates the stable break up of the droplet and gives the spraying mode. This smaller droplet size is The final solid parti cle formed after the solvent evaporates can then be calculated from (4 4) where is the mass fraction of the polymer solute, is the diameter of the solid particle after the solvent evaporates, and is the density of the polymer solute. Additional details associated with the derivation of this scaling law can be found in Yurteri et al. (2010). Production of C harged, N on S pherical P articles The production of non spherical particles using electrospray is relatively new a nd the techniques not so well developed. Liu et al. (2008) have produced cup shaped and polygonal shaped particles and Bhaskar et al. (2010) have produced rod shaped and disc shaped particles. Both studies link the formation of non spherical particles to t he solvent properties, polymer concentration, and solution feed rate. Almeria et al. (2010) explained the formation of non spherical particles due the solute entanglement network and breakup of the droplets.

PAGE 74

74 Near Infrared (NIR) Spectroscopy NIR spectroscopy uses NIR fluorophores to emit light with wavelength close to infrared light. NIR fluorophores are excited by light with a wavelength that is slightly longer than red light. NIR spectroscopy is a safer imaging alternative to radiotracers an d overcomes auto fluorescence issues associated with common green fluorophore. Previously, the FDA had approved only indocyanine green (ICG) fluorophore for use in humans. Recently, more NIR fluorophores have become commercially available for use in vivo. The nature and availability of NIR fluorophores makes NIR spectroscopy a safe and relatively inexpensive alternative for analyzing particle deposition (Frangioni, 2003; Hilderbrand and Weissleder, 2010) In vivo NIR usage has been well developed in areas s uch as the brain, heart, eye, and other vital organs (Frangioni, 2003) NIR techniques have also been used to detect cancer cells in the lungs. Troyan et al. (2009) found that NIR fluorescence was just as effective for in vivo imaging as the conventional r adiotracer. Experiments Bifurcating Model The bifurcating airway model in this study is based on the first four airways of (Weibel, 1963). h the trachea, which is labeled as generation 0 (G0). The G0 airway branches into two identical daughter airways, both called generation 1 (G1). This branching sequence continues until G23. The angle between the two daughter airways is always 70 degrees. The dimensions of each generation are listed in Figure 4 1 The final airways, G3, were modified with barbed

PAGE 75

75 fittings for additional use with a pump and filter; these airways are label as the connecter sections The original Weibel model did not specify th e azimuthal angle. However, Howatson Tawhai et al. (2000) indicated that the branching plane of the lung is, on average, at a 90 angle with respect to the preceding airway. Although further investigations by CT scans found that the angle less than 90 degre es (Nowak et al., 2003; Zhang and Kleinstreuer, 2002) the model system in this work was generated using 90 degrees to reduce the geometric complexity. The bifurcating airway model was glass blown using Corning 7740 Pyrex glass. Figure 4 2 presents a schem atic of the bifurcating airway model. Generation of Spherical Particles Particles were poly(lactic co glycolic acid) (PLGA) polymer dyed with NIR iodide (Sigma Aldrich). The spraying solution was 0.75% by wt. PLGA polymer in a 4:1 acetone dichloromethane solution. Addition of the NIR dye to the solution increased its electrical conductivity. To keep the particle size and the fluorescence properties constant for all solutions, the NIR dye was added so that the spraying solutions had a constant electrical conductivity of 80 Figure 4 2 lists the properties of the solution used in the spherical particle experiments. The scaling law predicts that electrospraying with 1.1 mL/hour (3.0510 10 m 3 /s) solution feed rate will result in a varicose break Figure 4 3 A shows a sample of the spherical particles generated; Figure 4 3 B presents the particle size d istribution, which is quite narrow. The average surface area diameter is 2.18 m with a geometric standard deviation of 1.26. The small peak

PAGE 76

76 between 1 and 2 microns may be due to the production of secondary particles. Particle size variations between experiments were minimal, varying less than 8%, and modifications in the particle charge did not affect the particle size. Generation of Non Sph erical Particles Rod shaped particles were generated using PLGA in an acetone ethyl acetate solution. Figure 4 4 documents the effect of composition of the solvent mixture on the particle morphology. Guided by these results, the rod shaped particles used i n the experiments ( Figure 4 5 ) were generated with a PLGA solution at a 0.75% by weight concentration in 6:4 acetone ethyl acetate solution fed at 0.90 mL/hr (2.5010 10 m 3 /s). On average, the rod shaped particles were 1.3 m in diameter with an aspect r atio of 3.3. Although the rod shaped particles were not as monodispersed as the spherical particles, there were minimal day to day variations in the generated particle mix. Hollow shell particles were produced by spraying a polyethylene glycol (PEG) solut ion with baking soda added following Marijnissen et al. 2010. Applying heat to the sprayed particles decomposes the baking soda embedded in the particle. The gas in the particle then expands and creates a hollow shell particle similar to a balloon. Unfort baking soda was eliminated and an alternate procedure was developed. A new solution of 4% by wt. of PEG (MW = 2000) in a 40:1 ethanol carbonated water solution was fed at 0.5 mL/h our (1.3910 10 m 3 /s) and generated the hollow particles shown in Figure 4 6 In this case, the predicted particle size ( E quation 4 4), before particle expansion, is 0.52 m. Since the hollow particles require more time to evaporate, a 30cm long metal pi pe was added to the experimental setup (shown in Figure 4 7 ) to accommodate the

PAGE 77

77 slower evaporation rate of carbonated water used for particle generation. Figure 4 6 details PEG particles collected at different locations downstream from the spraying nozzle. These images show a marked increase in particle size as the solvent is evaporated and carbonated water emerges, causing particle to swell. Experimental Setup The electrospray chamber is oriented vertically and consists of five separate elements: the spra ying nozzle, the spraying ring, the discharge needles, the metallic cone, and a particle charge measuring system ( Figure 4 7 ). Solution is fed through the spraying nozzle using an automatic syringe dispenser (NE 500, New ERA Pump System Inc.). High voltage s are applied to both the nozzle and the ring. Electrospraying occurs at the tip of the spraying nozzle with the spraying ring acting as a counter electrode. The potential difference between the nozzle and the ring atomizes the solution into droplets. An a pplied positive voltage generates a positively charged particle. The charge of the generated particles can be modified using the discharge needles. For example, an applied negative voltage is used to neutralize a positively charged particle. Depending on the level of applied voltage, the particle can be fully or partially neutralized. The spraying nozzle, ring, and the discharge needle were encased in a cylindrical tube made of borosilicate II glass ( L = 0.305 m, ID = 0.102 m). The particles then flow into a metal cone, 0.254 m in length, tapered to gradually reduce the flow diameter from the glass tube to the opening of the bifurcating airway model. The gradual reduction in diameter of the cone minimizes turbulence effects, and the cone can also be used to filter out particles. For example, if positive voltage is applied to the cone, the cone will attract and collect negatively charged particles at the wall before they exit the cone.

PAGE 78

78 The average particle charge is monitored in real time using a charge meas urement system composed of a ring probe and an electrometer (6514 system electrometer, Keithley Instrument Inc.) following the charge measurement systems of Vercoulen (1995) and Gajewski (1985; 1997). When a charged particle moves toward the ring probe, th e particle induces a current in the ring, which is sent to the electrometer. The current measurement is directly related to the charge on the particle. For a stream of particles with a constant flux, the current reading remains constant over time. The ring probe is placed outside of the airway and does not come into contact with particles. The ring probe is encased in a grounded shield to prevent interference from any external electrical fields. The charge reading from the ring probe is not the true charge of the particles as the signals are affected by the material properties of the ring probe and shielding. Hence, calibration with a Faraday cup is used to determine the true charge of the particles. After passing through the charge measurement system, the particles then flow through Tygon tubing and enter the airway model. During the experiment, the generated particles are carried downward towards the airway model using compressed air entering from the top of the chamber. A sheath of air flows near the wal l to keep particles from depositing there. Total airflow in the chamber (compressed air and sheath flow) was 28 LPM (0.000467 m 3 /s), controlled using rotometers. Since the electrospray chamber is 0.610 m in length, it takes about 5 seconds for the particle s to flow from the spraying nozzle to the airway model, ensuring sufficient time for the solvent to evaporate to form the solid particle. The constant particle flux (~ 0.1 mg/min) is measured separately by collecting particles in a filter

PAGE 79

79 placed after the ring. The relative humidity for these experiments was approximately 74%. Four high voltage DC power supplies (HCN 35 20000 and HCL 35 20000, Fug; Series 230, Berton Associates Inc.) are connected to the spraying nozzle, spraying ring, discharge needle, an d metallic cone. To generate positively charged particles, positive voltage is applied to the spraying nozzle, spraying ring, and metallic cone while low negative voltage is applied to the discharge needles. To generate negatively charged particles, the po larity used to generate the positively charged particles is reversed. Neutral particles can be made by applying the same voltage as either the positively or negatively charged particles but with higher voltage at the discharge needle and grounded metallic cone. Characterizing Neutrally Charged Particles In the case of zero charge reading from the charge measurement system, it was unclear whether the particle mix contained both positive and negative charges, which cancel out for the overall neutral charge o r whether individual particles were without any change. An additional experiment was used to determine the particle charging. A parallel plate capacitor and a faraday cup were positioned after the electrospray chamber, as shown in Figure 4 8 A voltage was applied to one plate while the other plate was grounded. Changing the voltage on the parallel plate capacitor filters out the particles of different charges, resulting in changes in the measured charge for particles collected in the faraday cup over a one minute time interval. Figure 4 9 presents the magnitude of the charge collected from such a test for positive, neutral and negative readings from the charge measurement systems. Positively charged particles ( Figure 4 9 A) show a positive gain in voltage w hen positive

PAGE 80

80 voltage is applied to the plate, and the opposite is true for negatively charged particles ( Figure 4 9 C ). This confirms that a positive reading from the charge measurement system yielded a mix of all positive particles and vice versa for the n egative particles. Figure 4 9 B for the neutral particles, exhibits a charge collected that is minimally positive but not detectable in the charge measurement system. However, with the neutral particles, the reading from the faraday cup also exhibited in stances when the voltage reading decreased. The decrease in voltage indicates the presence of negatively charged particles. Hence, the particles are not completely neutral, but rather consist of a mix of positively and negatively charged particles with (al charge. Particle Deposition Particles were sprayed until steady state conditions were achieved in the chamber. At steady state, the particle flux, particle size, and particle charge exiting the spraying chamber were constant. The priming period necessary to achieve steady state took up to four hours. Once the spraying chamber is at steady state, the airway model is attached and particles are collected for a four hour period. This collection period could be reduced by using an improved emis sion filter, better matching the emission wavelength (as discussed in the next section) or by using another dye. NIR Imaging The IVIS Lumina medical imaging system (Caliper Life Science) was used to visualize the particle deposition. The IVIS Lumina is capable of generating excitation light at a specific range of wavelengths and collecting the resulting emission light at a diffe rent range of wavelengths. The optimum excitation and emission wavelengths for the specific NIR dye in the PLGA polymer were determined using the Nanolog

PAGE 81

81 Spectrofluorometer (Horiba Scientific). Samples of thin polymer film, made by evaporating the spraying solution on a glass slide, were used to obtain these wavelengths giving the highest fluorescence signals. The optimum excitation and emission wavelength are 690 nm and 780 nm, respectively, as shown in Figure 4 10 Based on the optimal excitation and emi ssion wavelengths, the best available ICG background excitation and emission filters (665 695nm and 810 875nm, respectively) were selected to correspond to these wavelengths. Operating conditions for the IVIS Lumina were also optimized for the highest si gnal to noise ratio. The best signal to noise ratio was found to occur at 20 seconds integration time and the largest aperture. Black tape was used to reduce reflection and to block out the signal to prevent counting photons from different airway section s. In order to calibrate the IVIS signal, known masses of polymers, with a constant concentration of NIR dye (as used in the experiments), were placed into the IVIS Lumina along with the airway model. The standards relate the photon count to the polymer m ass. The relation between photon count and polymer mass in a single ROI was consistently found to be: (4 5) where is the mass of the polymer in grams. The photon count in the case of zero polymer mass was that giv en off by the background in the IVIS Lumina chamber. Since lighting influences the photon counts, the airway model and the standards were placed in the same location and orientation for all deposition experiments. The photon count variation between two i mages of a single experiment was less than 5%. The connector sections at the end of airway model were not included in the deposition analysis as the

PAGE 82

82 amount of particles deposited in that section varied greatly day to day due to the sharp reduction in airw ay diameter (see Figure 4 2 ). Results For each deposition experiment, two IVIS Lumina images were taken. The first was an ordinary photograph; the second was an image showing the fluorescence signal. By overlaying the two images, the resulting picture pre sents the particle deposition in specific locations of the airways. Figures 4 11 through 4 17 are generated in this manner, and all are displayed with the same color scale for the photon count. Since the airway model could not be imaged in a single pictu re, multiple images were taken for each experiment. Figure 4 11 gives the images associated with the various sections in the airway model for the case in which no particles were deposited and no fluorescence signal is detected. Each section was divided u p into region of interests (ROIs) ( Figure 4 12 ) of constant area to determine the photon counts in that section ( Figure 4 12 ). Figure 4 13 shows a representative deposition pattern with the IVIS Lumina after four hours of deposition testing. In the image, yellow/orange/red color indicates where particles have deposited, with the yellow regions indicating the highest particle deposition. Fig ure 4 3 shows the percentage of the total mass deposition from airway 0 to bifurcation 2. The average and standard deviations values are based on three experiments performed on different days. Information regarding deposition dynamics was also obtained at three different times during the deposition process; these images are shown in Figure 4 14 The total photon counts are displayed in the upper right corner of each image.

PAGE 83

83 the airway as more particles were deposited. This deposition pattern is different than what is expected for uncharged spheres. For uncharged spheres, the bulk of the deposition occurs at the airway bifurcations due to particle inertia and impaction (Oldh am et al. (2000); Kim and Fisher (1999)) However, in Figures 4 13 and 4 14, there is deposition in the bifurcating region, as well as the conduit airway. This difference in deposition is likely due to the small positive charge the particles hold. This hypothesis is based on the fact that the airway models of Oldham and Kim and Fischer are smaller than the one used here, Hence, it would be expected that the deposition in the conduit airway would be less in the present case, rather than greater, unless t here were some other factor or mechanism, such as particle charge, present. Deposition of Positively Charged Spheres Positively charged spheres (4.64 x 10 19 C/particle) were used in this part of the study. The deposition of positively charged spheres prov ed to be more troublesome charged spheres deposited on the wall near the region of particle generation; this drastically reduced the particle flux entering the airway. A ground wire was wrapped around the glass sectio n of the spraying chamber to reduce any patch charges on the glass surface. Humidity is another issue that directly affects deposition for positively charged spheres. At high humidity (>80% RH), no detectable particle flux entered the model. To obtain a strong measurable signal, the deposition experiments were carried out under a relative humidity below 50% and the resulting flux was consistently 0.13 mg/minute. Although the use of grounding wire and

PAGE 84

84 should also increase under these conditions and the deposition patterns should remain the same. Figure 4 15 presents the deposition of positively charged spheres after four hours of testing; Figure 4 3 gives the deposition mass percentages of the positively charged spheres. Positively charged spheres deposit much more strongly than the deposition occurs in airway 2 in comparison to the observation is likely due to space charge effects in which like charged particles in the particle cloud repel each other, pushing particles towards the wall much more quickly Figure 4 16 documents the dynamics of the deposition of positively charged spheres at three different times during the particle deposition process. Positively charged spheres do not exhibit the same steady build up of particles in each section as charged spheres initially deposit near the entrance (airway 0) of the model. The deposition in airway 0 continues until a saturation in particle deposition is reached, and then deposition begins in the next region downst ream. Positively charged particles only deposit at clean surfaces of the airway due to particle particle repulsion from previously deposited particles, and this gives rise to the particle deposition downstream. Deposition of Non Spherical Particles For n on the particles. The deposition images for the rod shaped particles are displayed in Figure 4 17 A ; the mass percent depositions in each section are given in Figure 4 3 In comparison shaped particles

PAGE 85

85 deposition was found in the upper section of the airway model. In addition, it was visually observed that the depositions of particles in each straight conduit section of the a irway were in patches as depicted in the sketch shown in Figure 4 17 B When a rod shaped particle deposits on the wall, a portion of the rod shaped particle will likely protrude, acting as a seed such that electrostatic effects pull other rod shaped parti cles to the seed particle to deposit. As particles continue to pile onto the seed particles, this gives rise to the formation of the deposition patches. Hollow, balloon shaped particles were also delivered into the airway at a mass flow rate of approximate ly 0.02 g/min, which is lower than the spherical particles. Nevertheless, the flow rate in terms of particle number was twice as high as the spherical particles. Knowing the mass flow rate, the size of the solid particle (0.52 m) before expansion, and t he solid particle density, the flow rate in terms of particle number is calculated. Even with this significant increase in flow rate in terms of particle number, negligible amounts of the hollow, balloon shaped particles were collected in the airway after 8 hours of deposition. It is expected that the low density, hollow particles behave in a similar fashion to submicron particles in which very little deposition occurs in the airways. However, when hollow, balloon shaped particles penetrate deep into the airways, a greater amount of deposition is expected due to interception. Concentration of the Deposited Particles charged spheres, and rod shaped particles (data given in Figure 4 3 ) are normali zed with the surface area in each of the deposition sections to obtain particle concentration information in the airway model. The surface area of each section is estimated using a 3D auto CAD model of the airway. Figure 4 4 presents these particle concen tration values, as well as

PAGE 86

86 the total surface area, in each section. For all cases, by far, the highest particle shaped particles, deposition concentration increases as the airway size reduces. For positively charged spheres, the deposition concentration in the bifurcating section is also much higher than the branching section, but the particle concentrations comparing airways 0 and 1, as well as comparing bifurcations 0 and 1, are similar, indicating a saturation for particle deposition in those regions. Summary Charged particles embedded with NIR dye were generated by electrospray and deposited in a glass airway model. The deposition of charged particles was visualized with an IVIS Lumina system. The NIR visualization technique proved to be an easy to use, non intrusive method to quickly quantify particle deposition. The NIR technique also has great potential to provide a safe and inexpensive way to visualize in vivo particl e deposition. Further investigation using this technique is warranted for example, matching excitation and emission wavelengths to minimize the deposition time. Charged particles were found to deposit differently from uncharged particles. Charged partic les deposit in both the conduit airways as well as the bifurcation region, as opposed to just the region near the bifurcation. How charged spheres deposit is also a function of the distribution of charges within the particle cloud. Spheres with zero net ch arge show a gradual build up of particles on all airway surfaces over time. Spheres with only positively charged particles initial deposit preferentially in the upper section of od shaped particles, deposition occurs in patches along the airways. Hollow, balloon

PAGE 87

87 shaped particles are conveyed through all four branches of the airway without depositing.

PAGE 88

88 Table 4 1. Dimension of the bifurcating airway model Airway Number of Airway D (m) L (m) G0 1 0.0180 0.1200 G1 2 0.0122 0.0476 G2 4 0.0083 0.0190 G3 8 0.0056 0.0076 Table 4 2. Electrospray solution properties Solution for spheres Solution for hollow balloon shaped particles Polymer PLGA PEG Weight concentration 0.75% by wt. 4.00% by wt. 3.0510 10 2.7810 10 Table 4 3. charged, and rod shaped particles Positively c harged particles Rod shaped particles Airway 0 29.60% 4.60% 47.29% 5.58% 36.79% 5.38% Bifurcation 0 7.40% 0.65% 14.78% 4.48% 12.72% 2.05% Airway 1 18.54% 1.59% 20.94% 1.49% 13.52% 2.61% Bifurcation 1 16.03% 2.42% 9.34% 1.81% 12.67% 3.65% Airway 2 10.14% 0.78% 3.64% 0.62% 5.61% 1.70% Bifurcation 2 18.29% 2.59% 4.01% 0.52% 18.69% 1.76%

PAGE 89

89 Table 4 4. Total surface area in each airway section and the average deposition charged particles, and rod shaped particles Total surface area (mm 2 ) (g/m 2 ) Positively c harged particles (g/m 2 ) Rod shaped particle s (g/m 2 ) Airway 0 6130 48 77 60 Bifurcation 0 503 147 294 253 Airway 1 3011 61 69 45 Bifurcation 1 436 367 214 290 Airway 2 1129 76 27 42 Bifurcation 2 396 465 102 476

PAGE 90

90 Figure 4 1. S chematic of electrospray

PAGE 91

91 Figure 4 2. Schematic of the bifurcating airway model

PAGE 92

92 A B Figure 4 3. Properties of s pherical particles A ) M icroscopic image B ) P article size distribution

PAGE 93

93 A B C D E Figure 4 4 PLGA particles generated in various solvents A ) A cetone, B ) 67:33 acetone ethyl acetate, C ) 50:50 acetone ethyl acetate, D ) 33:67 acetone ethyl acetate, and E ) E thyl acetate Sprayed with 2.7810 10 m 3 /s spray of 1% by wt. PLGA

PAGE 94

94 Figure 4 5 Microscopic images of rod shaped particles

PAGE 95

95 A B C Figure 4 6. Microscopic images of hollow, balloon shaped particles collected at different location. A ) The spraying nozzle. B ) T he discharge needles (6 cm downstream from the spraying nozzle) C ) T he charge measurement system (690 cm downstream from the spraying nozzle)

PAGE 96

96 Figure 4 7. Experimental setup

PAGE 97

97 Figure 4 8. Experimental setup using a parallel plate capacitor to characterize the charge of the particle mixtures

PAGE 98

98 A B C Figure 4 9. Voltage build up rates in the Faraday cup at different plate voltages A ) P ositively charged particles B N C ) N egatively charged particles

PAGE 99

99 Figure 4 10. Emission spectra from solid PLGA film embedded with NIR dye for different excitation wavelengths

PAGE 100

100 Figure 4 11. Compilation of IVIS Lumina images showing no photons in the airway mode l Figure 4 12. Example of individual regions of interest for generation 1 airway

PAGE 101

101 Figure 4 1 3 IVIS Lumina images of sphere

PAGE 102

102 Figure 4 1 4 sphere

PAGE 103

103 Figure 4 1 5 IVIS Lumina images of positively charged sphere

PAGE 104

104 Figure 4 1 6 Evolution of deposition patterns for positively charged sphere

PAGE 105

105 A B Figure 4 1 7 Deposition of rod shaped particles. A) Compilation of IVIS Lumina images showing rod shaped particle deposition patterns B) Schematic of the observed deposition patches for the rod shaped particles

PAGE 106

106 CHAPTER 5 DEPOSITION OF NON SPHERICAL PARTICLES IN BIFURCA TING AIRWAY Background The human respiratory system provides a vast area for quick and non invasive drug delivery with minimal side effects (Marianecci et al. 2011). The large surface area of the lungs makes pulmonary medication an effective path for drug delivery. The first usage of pulmonary medication was in 1554 BC, for treatment of asthma (Sanders 2011) Since then dr ug delivery tools have been refined into the contemporary medical inhaler. Drugs and particulate matter are loaded into an inhaler, and then aerosolized into the lung. The devices and the drug formulations for pulmonary drug delivery are well developed; wh ile relatively little effort went into engineering the particles. It is well accepted that particles between 1 10 um in size have the highest deposition efficiency in the human lung. Many researchers studied the overall particle deposition using ideal mono disperse spherical particles (Pavia et al. 1977; Schlesinger et al. 1977; Emmett et al. 1982) A small number of researchers studied the benefits of using controlled non spherical particles for inhalation. They found that those fibers with certain aspect ra tios and particles of certain shapes aerosolized from the dry powder inhaler much better than others (Kaialy et al. 2011). Once they enter the lung, particles with higher smoothness and aspect ratios can travel deeper into the airway (Zeng et al. 2000). Fi nally after deposition, alveolar macrophages interact with the particle differently depending on the particle morphology (Champion et al. 2007). The importance of particle morphology extends beyond the benefits of particle flow and deposition; drug release from particles is affected as well. Drug releases from microfilament polymers

PAGE 107

107 were ten times longer than those from spherical particles (Geng et al. 2007). Zero order drug release is possible using hemispherical particles (Hsieh et al. 1983) The majority of the investigations of non spherical particles were focused on the fibers. The investigations were a direct response to the danger posed by asbestos. diameter. The fi ber orientation during flow depends on its aspect ratio. There are two possible flow extremes. Particles can align with the flow along the streamline. This orientation reduces the drag force on the particle. Particles can also align perpendicular to the st reamline. This orientation increases the drag force on the particle and decreases the particle velocity. Fibers with an aspect ratio between 1 and 5 have no preferential flow orientation; while fibers, with aspect ratio greater than 5, align themselves alo ng the streamline (Timbrell 1965). With the exception of fiber, few particle shapes were investigated. Hu et al. (2008) generated micron size budesonide particles. The shapes under investigation were ellipsoidal, elliptic flake, and rectangular flake. Par ticles were sent into a throat airway, then into a multi stage stage impinger. They found that flake particles outperform ellipsoidal particles with more flake particles depositing in the latter stages of the impinger. Hassan et al. (2009) studied sphere, pollen shaped, plate, cubic, and rod shaped particles. Pollen particles are spherical particles with non uniform surface morphology, e.g. spiked surface. Particles were aerosolized using an inhaler and deposited into a cascade impactor at a flow rate of 30 LPM. They found that pollen shaped particles improved aerosolization from inhaler and deposition in the lung.

PAGE 108

108 In addition to shape, particle density also plays an important role in particle deposition. Large porous particles had become a powerful medical delivery tool. Large porous particles allow for higher drug loading but they have the flow characteristic of diameter, have a respirable fraction of 50% whereas solid particles with a diameter of Respirable fraction is related to how much particles remain in the lung after inhaled They also found that the amount of porous ude higher than non porous particles. In human clinical trial, the commercialized PulmoSphere particles (porous sphere) can double the deposition in the lung (Hirst et al. 2002). For non spherical particles, Su et al. (2009) found that low momentum fibers penetrated deeper into the lung, whereas high momentum fibers deposited more in the upper airways. The investigation in this chapter attempts to look at the effects of particle shape on the local particle deposition. Ten different types of particles with a erodynamic diameters between 3 sphere, rod, flake, and random shape. The local depositions were visualized and quantified using near infrared (NIR) fluorescence. An empirical model was made relating t he deposition efficiency in each airway. Deposition Efficiency Deposition efficiency characterizes the particle deposition in the lung. Deposition efficiency is defined as the ratio of the mass of particles deposited in the airway to the mass of par ticles entered into the airway Deposition

PAGE 109

109 efficiency can be calculated for an individual airway (local deposition efficiency) or as a whole (total deposition efficiency). Deposition efficiency can be calculate using: (5 1) where the sub script n denotes the airway of interest. Equation 5 1 is valid for individual airways or for series of airways. Deposition Mechanisms In any airway there are three main deposition mechanisms: inertial impaction, sedimentation, and diffusion (Hinds 1982) Inertial impaction occurs when a particle is unable to follow the fluid flow path. When the airflow changes direction, particles come into contact with the surrounding wall and deposit. Inertial impaction is classified using the Stokes number Stokes to the characteristic dimension of the obstacle. The Stokes number for particle flowing in a bifurcating airway is (5 2) where is the particle density, is the particle diameter, is the average fluid velocity in the airway, is the fluid viscosity, and is the airway diameter. Many researchers related particle deposition to the Stokes number. Gawronski and Szewczyk (1986) and Chan and Schreck (1980) both found a logarithmic rel ationship between Stokes number and the overall deposition efficiency in the tracheobronchial tree. Sedimentation is the deposition due to the influence of gravity. The important forces in sedimentation are gravity, drag, and buoyancy. Using the force bala nce the terminal settling velocity is calculated to be

PAGE 110

110 (5 3) where is the fluid density, and is the acceleration due to gravity. For flow in a pipe, the deposition due to sedimentation in the absence of other deposition mechanisms can be related to the sedimentation parameter : (5 4) where is the length of airway. The thi rd mechanism, diffusion, occurs by random bombardment of air molecules on the particle. This mechanism is known to be important for smaller particles. The can be calculated using the Stokes Einstein equation: (5 5) where is the Boltzmann constant, and is the temperature. Similar to the sedimentation parameter, particle deposition caused by diffusion in a pipe flow can be related to the diffusion parameter : (5 6) where is the air flow rate. I n pipe flow, both diffusion and sedimentation parameters determine the particle wall loss (Davies 1973). There are two other deposition mechanisms: interception and electrostatic. Interception is relevant for particles with large aspect ratios such as fibe r. Interception occurs when the particle shape or particle size prevents the particle from passing through the airway. Electrostatic deposition is cause by the influence of electrostatic charges on the particle. The effects of both mechanisms are expected to be minimal in this study.

PAGE 111

111 Particle Characterization In this study, particles are characterized into three general shapes: sphere, rod, and flake. Important parameters are the particle surface area and particle volume A schematic of each particle shape is shown in Figure 5 1. Spherical particles (Figure 4 1 A) are assumed to be perfectly spherical; thus the equations for surface area and volume of a sphere can be used (5 7) (5 8) Rod particles (Figure 5 1 B) are particles with a high aspect ratio the equations for surface area and volume of the cylinder can be used (5 9) (5 10) Flake particles (Figure 5 1 C ) are characterized with the projected surface area perimeter and particle thickness Flake particles do not have a fixed projected area; the projected area differs from one particle to another. The equations for surface area and volume are: (5 11) (5 12) Lastly, there are particles that do not have a definable shape. Each partic le in the sample is different from another. These particles possess morphologies that are a combination of rod, tetrahedron, block, and cube. These random shapes are

PAGE 112

112 generalized to a flake particle in order to approximate the particle surface area and volu me. To account for the non spherical factor of the particle sphericity is used. Sphericity sphere. Sphericity is defined as the ratio of the surface area of a sphere with the same vo lume as the particle to the actual surface area of the particle: (5 13) Sphericity values range between zero and one, with one being completely spherical. Boundary Layer A boundary layer is a layer of flow near the wall where the flow is dominated by the viscosity. Within the boundary layer, the drag force becomes important and the fluid velocity profile approaches its minimum, as the fluid gets closer to the wall. Particl e diffusion becomes more significant within the boundary layer than outside the layer. The boundary layer in the pipe will be approximated as the Blasius boundary layer for a flat plate ; the boundary layer thickness is given as given as (5 14) The boundary layer thickness is normalized by the length of the airway Experiments Particle Characterization Ten different types of particlse were used for this experiment: solid glass spheres (Mosci), hollow glass spheres (Dantec Dynamic), glass rods (Nippon Electronic Glass co.), two types of glass flakes (Glassflake Ltd.), talc flakes (Barretts Minerals inc.), silver

PAGE 113

113 flakes (Inframat Advance Material LLC.), Aluminum oxide powders (AIRBRASIVE), crushed glass powders, and fine fra ctions of JSC 1A lunar simulants. Images of each particle type were taken under an optical microscope to determine particle shape and sizes. The images taken from the optical microscope were processed using ImageJ software. Samples up to 2500 particles we re used to obtain representative characteristics for each type of particle. The particle aerodynamic diameter was determined using the TSI PSD 3603 Aerosizer (TSI). Particle densities were obtained using the Ultrapyc 1000 gas pycnometer (Quantachrome co rp.). Airway Model The airway model used is the experiment is described in Chapter 4. For this set of experiment, rather than considering the airway branch section and bifurcatin g section, this experiment looked at the deposition in each airway generation (see Figure 5 2). The dimension of each generation and the surface airway in each generation are listed in Table 5 1. Particle Deposition from this point on the dye will be r efered to as NIR dye. Each type of particle was mixed into a solution of ethanol and NIR dye. The ethanol solvent dried off, leaving the particles coated with the NIR dye. Particles were placed in an inverse cyclone seeder. A small amount of Larostat antis tatic powder (BASF) was mixed in with the particles to minimize the electrostatic charge on the particles. Figure 5 3 shows the diagram of the inverse cyclone seeder and the experimental setup. In the cyclone seeder, air entered from the side and entrained particles from the particle bed at the bottom of the cyclone

PAGE 114

114 seeder. Entrained particles exited the cyclone seeder from the top and flowed into the airway model. In all experiments, the air flow rates were kept constant at the rate of 30 LPM (0.0005 m3/s) The flow parameters (air flow rate, average fluid velocity, residence time Reynolds number, and boundary layer thickness) for each generation are listed in Table 5 2. The average particle mass flow rate was determined at the point where the particle entered the airway model. A filter was placed where the airway model is in Figure 5 3. The mass of the particles collected is determined for a set period of time. The mass flow rate was different for each type of particle. During experiment, particles were sent into the airway model until enough particles were deposited such that strong signals were seen in the fluorescence image. Some particle types (e.g. JSC 1A) required longer time because weaker signals were emitted from the NIR dye. The deposited airwa y model was taken to the imaging device directly after depositio n. A total of three experiments were performed for each type of particle. Quantifying the Deposition The same method as described in Chapter 4 was used to obtain images from the IVIS Lumina. A fter deposition, the airway model was placed into the IVIS Lumina and the photon counts were determined for each airway. Multiple images were taken at different angles to capture the entire geometry of the airway model. Sample powders of known mass were pl aced into the IVIS Lumina with each picture to standardize the images together. The particle photon counts (total photon count minus the background) from the each airway were used to determine the percentage of particle deposited in each section.

PAGE 115

115 Particles deposited in the airway model were washed off using ethanol alcohol. The washed particles were dried and weighted to determine the mass of the deposited particles. The total mass of deposited particles was multiplied by the percentage of the particle phot on counts in each airway to determine the mass of particles deposited in individual airways. The total mass of particles entered into the airway was determined using the average particle mass flow rate and the duration time of the deposition. The depositio n efficiency in each airway was calculated using Equation 5 1. Results Particle Characteristics A representative image for each type of particle taken from the optical microscope is in Figure 5 4. Particle shape was classified into sphere, rod, flake, and random shape. The projected area diameters were calculated from the images. The surface area and volume for each type of particle were calculated using Equation 5 7 through 5 12, the thickness for flake particles were given by the manufactures. The surface area and volume are used to calculate the sphericity. The projected area diameter, aerodyna mic diameter, particle density, and sphericity for each type of particle are listed in Table 5 3. Particle aerodynamic diameter links particle flow characteristic to a sphere of a given density with the aerodynamic diameter; whereas the projected area diam eter indicates the actual size of the particle seen through the microscope. Comparing the two diameters can give information on how the particle size and how sphere

PAGE 116

116 Particle Deposition An image of the clean airway model taken with the IVIS Lumina is shown in Figure 4 11 In Figure 4 11 no photon signals were detected from the airway model. As a comparison, representative images for glass spheres and glass flakes 1 a re shown in Figure 5 5 and 5 6 respectively. In each image, the higher color temperature indicates higher photon counts; red being the lowest and yellow being the highest. All three images are on the same color temperature scale. Figure 5 5 is the depositi on of the glass sphere in the airway model. As expected, the majority of the depositions occur at the bifurcating section, with very little deposition in the straight tube section of the airway. This result is similar to the finding by Kim and F ischer (199 9). Figure 5 6 shows the deposition of glass flake 1. The glass flake 1 deposited differently than the glass sphere. Instead of just depositing in the bifurcating section, glass flake 1 also deposited in the straight tube airway. Talc flakes also show depo sition at the bifurcating section of the airway. Particle photon counts were calculated from each image. The percent deposited in each generation of the airway was calculated and listed in Table 5 4. Table 5 4 also shows the average weight of the particle s that entered and deposited in the airway model. Based on the values in Table 5 4, the deposition efficiencies were calculated using Equation 5 1. The aerodynamic diameters (Table 5 3) were used along with the flow information (Table 5 2) to calculate the parameters important to each deposition mechanism (Equation 5 2,5 4, and 5 6). The deposition efficiency, stokes number, sedimentation parameter, and diffusion parameter in each airway for each type of particle are listed in Table 5 5.

PAGE 117

117 Deposition M odel S ix different types of particles were chosen to create an empirical model for particle deposition. Glass spheres, hollow glass spheres, glass rods, glass flakes 1, silver flakes and aluminum oxide particles were used. The empirical model proposed in this ch apter uses an adjusted deposition efficiency : (5 15) where is the surface area of the airway, and is the surface area for airway G0 is the deposition efficiency in airway n is calculate by normalizing with the particle sphericity and the ratio of the surface area of the airway to the surface area of the largest airway. Normalizing the deposition efficiencies with the surface area will allow comparison of the deposition efficiency from one airway generation to the next. F rom all the IVIS Lumina results, every type of particle exhibits higher deposition at the bifurcating section of the airway. It can be concluded that inertial impaction is the main deposition mechanism for particles with aerodynamic diameter between 1 m depositing in a bifurcating airway. Figure 5 7 is a plot of the adjusted deposition efficiency as a function of Stokes number. In Figure 5 7 there appear to be two distinct relationships: one for flakes and low density particles, and the other for high density particles. Fitting the adjusted deposition efficiency with the Stokes number gives: (5 16) is the adjusted deposition efficiency arising from the inertial impaction, this term is plotted with a dotted line in Figure 5 7 In ertial impaction alone was not enough to explain all the particle deposition. Two other mechanisms, sedimentation and diffusion, were included to resolve the additional deposition. Inertial impaction is the primary

PAGE 118

118 deposition mechanism. Sedimentation and d iffusion are the secondary deposition mechanisms. Modifications are made to the sedimentation and diffusion parameter, the adjusted parameters are: (5 17) (5 18) where (5 19) the term is the adjustment for the influence of th e shape and density. The calculations of each parameter were done using the aerodynamic diameter as the particle diameter, this assumes that the particles are spherical. Adjustment is needed to correct for the non spherical aspect of the shape. Sphericity is used to adjust for the non spherical nature of the particle. The lower the sphericity, the higher the surface area the particle has. Higher surface area led to higher drag that hindered the particle flow. Higher surface area also increases the contact a rea for diffusivity. A particle with low sphericity (e.g. flakes) will have a lower terminal velocity than a sphere and will also have a higher diffusivity. The specific gravity term is used to limit the effect of the sphericity. Some particles, such as si lver flakes or aluminum oxide, have low sphericity but high density. These particles depositions appear similar to the deposition of the solid sphere, where very little deposited on the straight section of the airway. Based on this observation, the effect of sphericity is negated by the density of the particle. A specific gravity term is used, at higher density, the term approaches 1; and at lower density the specific gravity enhances the effect of the diffusion. The adjusted diffusion term also includes

PAGE 119

119 the dimensionless boundary layer term The diffusion mechanism is important in the boundary layer, thus this term is included in the calculation of the diffusion parameter. Figures 5 8 A and 5 8 B show the plot of versus and versus respectively. The trends in Figure 5 8 A and 5 8 B are similar to that in Figure 5 7 power law fitting was also used in for these two mechanisms; these result in: (5 20) (5 21) and are plotte d as dotted lines in figure 5 8 A and 5 8 B This model will assume that the sedimentation and diffusion are competing mechanisms. The two secondary mechanisms are assumed to be mutually exclusive and only one will occur depending on the particle property. T he specific momentum of the particle is used to determine which secondary mechanism will take place. (5 22) where is a shape factor defined by (5 23) The shape factor represents the ratio of aerodynamic diameter to the rati o of the surface area diameter. In general if the surface area diameter is less than the aerodynamic diameter then is greater than one, this means the actual particle has less surface area than the aerodynamic diameter. Lower surface area means less drag and a higher particle velocity. The opposite is true as well. The specific momentum is used to relate the effectivenes s of Brownian diffusion. Particles with low felt greater

PAGE 120

120 influence from the air molecules during collision; whereas high particles are whether sedimentation or dif fusion will occur, the total deposition efficiency can be calculated using: (5 24) is the predicted deposition efficiency predicted. Model Validation The model is checked with all ten types of particles used in the experiments. Table 5 6 lists the adjusted deposition efficiencies from each experiment and the adjusted deposition efficiency calculated using Equation 5 24 The plot in Figure 5 9 compares with The dotted line in Figure 5 9 represents the relationship. The fitting match well for both the six particle types used to create the model, as well as the other four particle types. The RMS deviation between a nd is 0.24. The airway used in these experiments is based on generation 0 to generation 3 would be useful for further experiments. Airways with deeper generation wil l lead to a more realistic result that is much closer to the deposition in the human pulmonary airway. However with a more complex airway, it will become harder to determine the local deposition efficiency. We would then need a more complex system to deter mine the deposition efficiency. The recommended system would be something akin to SPECT where the images are taken in slices and are put together to create a 3D model.

PAGE 121

121 NIR fluorescence was proven to be successful, but unless a machine similar to the Xenoge n IVIS is available to take 3D images, using commercially available imagining systems will be more efficient. Summary This chapter presented an empirical model relating the deposition efficiency particle characteristic. The deposition efficiency in a singl e airway was broken down in to the primary deposition mechanism, inertial impaction, and a secondary mechanism (either sedimentation or diffusion). The two secondary mechanisms are mutually exclusive and chosen by the specific momentum of the particle. Thi s model is expected to work for any non spherical particles with 1 10 m aerodynamic diameters, given that the representative morphology the particles can be categorized into a specific shape. This model may fail for the cases when particles travel deeper into the airway or when particle density is too low. The empirical model is expected to fail in these cases because sedimentation and diffusion will be the dominating deposition mechanisms instead of inertial impaction.

PAGE 122

122 Table 5 1. Dimension of the bifurc ating airway model Airway Number of Airway D (m) L (m) S (m 2 ) G0 1 0.0180 0.1200 0.006632 G1 2 0.0122 0.0476 0.001723 G2 4 0.0083 0.0190 0.000431 G3 8 0.0056 0.0076 0.000127 Table 5 2. Flow parameter in the bifurcating airway model for the average airflow rate of 30 LPM Airway Q V G0 0.0005 1.965 0.061 2358 0.00479 G1 0.00025 2.139 0.022 1739 0.00289 G2 0.000125 2.310 0.008 1278 0.00176 Table 5 3. Characterization for each type of particle Particle Particle shape Glass spheres Sphere 2.892 8.470 2326.7 1 Hollow glass spheres Sphere 5.715 8.360 1038.4 1 Glass rods Rod 10.858 10.898 2610.1 0.762 Glass flakes 1 Flake 12.551 8.124 2491.6 0.393 Glass flakes 2 Random 8.694 10.810 2472.0 0.783 Crush glass Random 5.368 7.989 2556.6 0.796 Talc flakes Flake 5.624 3.419 2865.3 0.477 Silver flakes Flake 8.070 6.096 9552.8 0.26 Aluminum oxide Random 5.624 5.549 4580.2 0.411 JSC 1A, fine fraction Random 3.257 6.522 3200 0.628

PAGE 123

123 Table 5 4. The percentage of deposition in each generation of the airway model, and the particle weight deposited in the model Glass sphere Hollow glass sphere Glass rod Glass flake 1 Glass flake 2 G0 (%) 6.70 2.11 10.13 4.15 8.90 1.15 29.90 1.70 9.13 1.71 G1 (%) 21.99 6.88 20.92 2.64 28.38 2.49 17.12 1.15 30.09 3.97 G2 (%) 28.48 4.00 29.64 3.38 16.54 1.34 16.78 0.06 25.22 2.91 Connector (%) 42.82 6.55 39.31 6.16 46.19 4.30 36.20 0.55 35.55 2.56 Wt deposited (g) 0.01309 0.00217 0.01351 0.00500 0.00838 0.00094 0.01137 0.00229 0.01949 0.01019 Wt entered (g) 0.18473 0.04307 0.05833 0.05465 0.02544 0.00308 0.02637 0.00776 0.09965 0.03846 Crush glass Talc flake silver flake Aluminum Oxide JSC 1A fine fraction G0 (%) 15.72 1.86 12.88 5.75 6.23 1.49 1.16 0.08 4.75 0.85 G1 (%) 24.47 3.56 17.66 5.23 21.70 2.41 21.38 1.90 16.31 2.40 G2 (%) 18.86 2.72 26.97 3.31 18.58 2.20 19.19 1.56 22.90 1.07 Connector (%) 40.96 3.92 42.28 6.16 53.49 5.72 58.27 1.52 56.04 2.65 Wt deposited (g) 0.01765 0.00712 0.00277 0.00141 0.0036 0.00159 0.00328 0.00131 0.03859 0.00890 Wt entered (g) 0.05950 0.00675 0.01620 0.00329 0.01494 0.00909 0.03473 0.01818 0.15102 0.05284

PAGE 124

124 Table 5 5. Deposition parameter for each particle in each airway Airway Glass spheres G0 0.48 0.0562 0.0128 5.32 G1 1.57 0.0903 0.0069 4.22 G2 2.06 0.1434 0.0038 3.37 Hollow glass spheres G0 2.35 0.0245 0.0056 5.39 G1 4.96 0.0393 0.0030 4.28 G2 7.40 0.0623 0.0016 3.42 Glass rods G0 1.72 0.1044 0.0239 4.14 G1 5.66 0.1677 0.0128 3.28 G2 5.45 0.2663 0.0070 2.62 Glass flakes 1 G0 2.93 0.0554 0.0127 5.55 G1 9.62 0.0890 0.0068 4.40 G2 6.21 0.1413 0.0037 3.52 Glass flakes 2 G0 12.89 0.0973 0.0222 4.17 G1 8.48 0.1563 0.0120 3.31 G2 9.07 0.2482 0.0065 2.64 Crush glass G0 0.11 0.0550 0.0126 5.65 G1 2.02 0.0883 0.0068 4.48 G2 1.85 0.1402 0.0037 3.58 Talc flakes G0 4.66 0.0113 0.0026 1.32 G1 7.61 0.0181 0.0014 1.05 G2 6.35 0.0288 0.0008 8.35 Silver flakes G0 1.50 0.1196 0.0273 7.40 G1 5.30 0.1921 0.0147 5.87 G2 4.79 0.3050 0.0080 4.69 Aluminum oxide G0 2.20 0.0475 0.0109 8.13 G1 3.08 0.0763 0.0058 6.45 G2 4.86 0.1212 0.0032 5.15 JSC 1A, fine fraction G0 1.21 0.0459 0.0105 6.92 G1 4.22 0.0736 0.0056 5.49 G2 6.19 0.1169 0.0031 4.38

PAGE 125

125 Table 5 6. Contribution of deposition from each mechanism Airway Glass spheres G0 0.53 0.38 0.91 0.48 G1 3.00 3.72 6.72 6.02 G2 16.44 35.12 51.56 31.73 Hollow glass spheres G0 0.02 22.74 22.77 2.35 G1 0.14 45.48 45.62 19.10 G2 0.77 93.29 94.06 113.96 Glass rods G0 5.12 0.09 5.21 2.26 G1 29.26 0.86 30.11 28.62 G2 160.24 8.09 168.33 110.22 Glass flakes 1 G0 0.50 48.17 49.21 7.46 G1 2.84 97.41 100.25 94.35 G2 15.57 199.81 215.38 243.67 Glass flakes 2 G0 3.95 0.13 4.08 16.46 G1 22.57 1.29 23.86 41.66 G2 123.61 12.18 135.79 178.49 Crush glass G0 0.48 0.82 1.30 0.14 G1 2.76 8.00 10.76 9.79 G2 15.13 75.51 90.64 35.88 Talc flakes G0 0.00 47.14 47.14 9.77 G1 0.01 94.28 94.29 61.36 G2 0.04 193.39 193.44 204.87 Silver flakes G0 8.44 0.02 8.45 5.77 G1 48.16 0.17 48.34 78.59 G2 263.80 1.63 265.43 284.33 Aluminum oxide G0 0.28 2.17 2.45 5.36 G1 1.62 21.26 22.88 28.90 G2 8.85 200.72 209.56 182.29 JSC 1A, fine fraction G0 0.25 1.70 1.95 1.93 G1 1.42 16.64 18.05 25.85 G2 7.77 157.05 164.82 151.71

PAGE 126

126 A B C Figure 5 1. Shape and characteristic dimensions of different types of particles. A) S phere B) Rod. C) F lake particle

PAGE 127

127 Figure 5 2. Schematic of the bifurcating airway model

PAGE 128

128 Figure 5 3. Schematic of the inverse cyclone seeder setup

PAGE 129

129 A B C D E F Figure 5 4. Particle viewed under optical microscope A ) G lass spheres B ) H ollow glass spheres C ) G lass rods D ) G lass flakes 1 E ) G lass flakes 2 F ) C rush ed glass G ) T alc flakes H ) S ilver flakes I ) Aluminum oxide J ) Fine JSC 1A

PAGE 130

130 G H I J Figure 5 4. Continued

PAGE 131

131 Figure 5 5 IVIS Lumina images of glass spheres

PAGE 132

132 Figure 5 6 IVIS Lumina images of glass flakes 1

PAGE 133

133 Figure 5 7 Plot of the against Stokes number. The dotted line indicates the relation for the adjusted deposition efficiency of the inertial impaction mechanism

PAGE 134

134 A Figure 5 8 D eposition efficiency of the secondary mechanism A) S edimentation, B) D iffusion The dotted line indicates the calculat ed relationship for the mechanism

PAGE 135

135 B Figure 5 8 Continued

PAGE 136

136 Figure 5 9 vs. The dotted line is the relationship

PAGE 137

137 CHAPTER 6 CONCLUSION AND RECOM MENDATION The works presented deal greatly with the transport of particles in the airways. Particles become charged and deposit along the flow geometry. The important contributions are: Developed a standardized method for measuring particle wall charging characteristics ( and ). Demonstrated the prediction of particle charge for a pipe flow system using experiment coupling with CFD simulations. Validated the use of NIR fluorescence in the visualization and quantification of particle deposition Showed that particle deposition is dependent not only on how much c harge is on the particles but also on the distribution of charge Acquired data for the local deposition of non spherical particles in a series of branching airways. Particle C harging D uring T ransport The works detailed in Chapters 2 and 3 demonstrated that it was possible to predict the particle charge in various conditions based on information obtained from a small scale experiment. An experiment was performed to determine what happened to the particle charge after traveling through a single 1 m pipe. Then a CFD simulation was performed, matching the experimental condition, to determine the particle wall collision. With this set of information it is possible to predict particle charge in the same particle pipe system at different operating conditions. This method is valid for both spherical and non spherical particles. There are many ways to improve this model. The future continuation of this work should focus on expanding the model as mentioned in detail in Chapter 3. In the future, the model should take in to account: Changes in particle size

PAGE 138

138 Changes in pipe Changes in environmental condition (humidity and temperature) Complex flow geometry Lastly, the characteristic parameters as well as the equilibrium charge should be related to the particle and wall ma terial parameters. This would be difficult because the charge transfer theory of insulators is not well developed. But if these parameters could be calculated from material properties, an experiment would no longer be needed. Particle charge could be deter mined directly using simulation and all these effects could be incorporated into a commercial CFD code to determine particle charge in any particulate system. Particle Deposition in Pulmonary Airways Use of NIR Fluorescence The works described in Chapters 4 and 5 demonstrated that it is possible to both visualize and quantify particle deposition in a complex airway geometry using near infrared fluorescence. The NIR images looked clear and the photon count in the images was used to determine the mass of the particle deposited in the airway model. The only issue we encountered was the long deposition time needed in order to get a strong signal coming from the particle. However the lack of signal was attributed to the poor matching of the dye and the camera fil ter set. In future studies, it is very important to better coordinate the dye and the filter set. Ultimately, the use of NIR fluorescence is aimed toward in vivo clinical trials. The next step in testing this is to perform experiments with animals. Similar experiments with charged particles and non spherical particles should be performed in rats. By pe rforming the experiment with animals, it will be possible to determine the feasibility of particle imaging using NIR fluorescence. During this test it would be good to do a

PAGE 139

139 comparative study with other common imaging techniques such as the gamma camera. Th e comparison will tell how the NIR fluorescence stacks up against other techniques currently used. Effect of Electrostatic charge on Particle Deposition Deposition of C harged S phere s The deposition of charged particles shows that electrostatic charge on th e particles strongly influences particle deposition. Charged particles deposited much more than uncharged particles. Uncharged spheres, found in prior studies and Chapter 5, are found to deposit mainly in the area where the airway branching occurs; whereas charged particles deposited everywhere. The deposition of charged spheres also depends upon the charge on the spheres. For particle clouds of positively charged particles, deposition occurs sequentially in each airway particles pass through. This is hypot hesized to be caused by particle particle repulsion. Particles with zero net charge are found to deposit in each airway evenly over time. Deposition of Charged Rod Shaped Particles Rod shaped particles were generated using an electrospray. The depositions of rod shaped particles were similar to those of spherical particles because the aspect ratios of the rod shaped particles were fairly low. Rod shaped particles deposited in patches; this is thought to be caused by the deposited rods extending from the w all surface and becoming seed particles for additional particles to deposit. Deposition of Charged Hollow, Balloon Shaped Particles The hollow, balloon shaped particles were generated using an electrospray. This newly introduced technique using carbonated solution was successful in producing

PAGE 140

140 model without depositing. In the future we recommend that we document the rate of particle growth over time. The same technique fo r collecting particles at different distances from the generation would best document the effect because cameras do not have the magnification to capture the particle size. Future E xperiments In the future, more experiments should focus on: A comprehensive study using different levels of charge on the particles Varying the aspect ratio of rod shaped particles with electrostatic charge Varying the charge distribution on the particle surface Sending hollow, balloon shaped particles into smaller airways Depos ition with other types of non spherical particles These experiments will help us gain a better understanding of the influence of electrostatic charge on particle deposition. Deposition of Non Spherical P articles Particles between 1 in size were successfully deposited and quantified using NIR fluorescence. The result showed that particle depositions can be explained using a combination of two mechanisms. Particle deposition depends on the particle specific momentum. Particles with hi gh momentum deposited through particle impacting on the branching section of the airway; particle movement was also caused by gravity. Particles with low momentum also deposit by impacting at the branching section of the airway but instead of additional de position caused by gravity, the low momentum particles deposit through Brownian movement toward the wall surface. Particle shapes were also important; the shape factors used to describe particle shapes were important in determining the particle deposition.

PAGE 141

141 The empirical model presented in Chapter 5 is developed from six different types of particles and tested with ten different particle types. Additional work is needed to further test and improve the model; these additional tests include: Experiments with m onodispersed particles Experiments using rods with varying aspect ratio Experiments using flakes of varying surface area Experiments with additional particle types Experiments with smaller airways Another possible experiment is to deepen our understanding of particle morphology. Parameters such as particle roughness will surely affect particle flow and deposition. Suggestions for this would be to start with spherical particles and rod shaped particles with surface roughness. Roughness has been found to affect particle flow leaving the inhaler; it should also affect deposition as well. Future application s to simulation The overall goal of investigating particle shape and charge is to relate particle deposition to computational simulation. Due to the various particle particle interactions seen in this work, we recommend the use of CFD DEM simulation. In the CFD DEM we should be able to determine the steady state fluid flow information for the airway model so that we can then simulate the particle particle interaction in the flow system. Electrostatically charged particles will involve implementing interactions caused by electrostatic charge. This should include both particle particle interaction as well as particle wall interaction. The resultant f orce on the particle will be a function of charge of the two materials as well as the distance between them. Non spherical particles should also be examined to determine the effect of particle shape. And lastly for simulations of non spherical, charged par ticles, we expect that the distribution of electrostatic charge

PAGE 142

142 on these particles will depend on the particle shape (such as charge accumulating toward the edges); this information should be taken into account for large particles. CFD DEM simulation will help to identify the flow effect of the electrostatic charge on the particle and the effect of non sphericity of the particle.

PAGE 143

143 APPENDIX A EFFECT OF V EX ON THE CONTACT POTEN TIAL The external electric field has a signi ficant influence on contact electrification. The term in Eq u ation 2 3 arises from the electric fields in the surrounding environment, including electric fields in nature (e.g. the earth magnetic field) and man made electric fields (e.g. electric field from power line). A schematic of the electric potential of a particle in contact with a surface (Figure A.1) can help understand the effect of In Figure A.1, the energy levels of the particle and wall are presented on the left and right hand sides, re spectively. The Fermi level is a hypothetical energy level for an electron inside the particle. The vacuum level is the energy level such that the potential energy of the electron is zero. The difference between the Fermi level and the vacuum level is equal to the work function of the particle. Consider Case 1 in Figure A.1 where no ext is higher while the potential difference due to image charge and space charge remain constant. The final voltage potential of the particle is determined by the wo rk function and the final potential of the wall consists of the wall work function and the contribution from the image charge and the space charge. The charge transfer is equal to the potential difference of the two surfaces, During contact, electrons w ill transfer from lower voltage potential to higher voltage potential, as the potential is zero at the vacuum level. In Case 1, electrons will flow from the wall to the particle, resulting in negatively charged particles. Cases 2, 3, and 4 show what occurs when the external potential is applied to the wall surface. In Case 2, a small amount of potential is applied to the wall; the final potential of the wall is now closer to the potential of the particle. The result of such a contact is that smaller charge transfer occurs during contact, so that the

PAGE 144

144 resultant particle in Case 2 will not gain as many electrons as the resultant particle from Case 1. In Case 3, the potential of the wall is equal to the potential of the particle, and no charge transfer occurs be tween the particle and the wall. In Case 4, a strong electric field is applied to the wall such that the potential of the wall is higher than the work function of the particle. In this case, the vacuum level is shown for the wall as more energy is required to remove electrons from the surface of the wall than from the becomes positively charged. The examples shown in Cases 2, 3, and 4 suggest that electric field ca n directly influence how particles are charged. The particle charger used in this study builds upon this concept. In the particle charger, the electric field is intentionally applied to the wall surface to artificially modify the potential difference betwe en the particle and the wall. The charge of the impacted particle is directly related to the applied electric field.

PAGE 145

145 Figure A 1. Different cases of contact charge transfer in applied electric field

PAGE 146

146 APPENDIX B RAW DATA FOR PARTICL E CHARGING EX PERIMENT Table B 1. R aw data for b orosilicate particle copper pipe system Electric field 0 274 234 100 198 169 200 152 132 300 117 105 400 91 58 500 34 33 600 10 22 700 14 4 800 41 11 900 50 16 1000 26 Table B 2. R aw data for b orosilicate particle stainless steel pipe system Electric field 0 274 153 100 198 139 200 152 127 300 117 112 400 91 98 500 34 96 600 10 83 700 14 79 800 41 73 900 50 76

PAGE 147

147 Table B 3. R aw data for b orosilicate particle natural glass pipe system Electric field 0 274 210 100 198 115 200 152 75 300 117 47 400 91 9 500 34 26 600 10 50 700 14 72 800 41 88 900 50 98 Table B 4. R aw data for b orosilicate particle pyrex pipe system Electric field 0 286 260 50 239 195 100 196 161 150 171 135 200 146 114 300 105 84 400 82 41 500 50 17 600 17 13 700 15 31 800 34 56 900 46 60

PAGE 148

148 Table B 5. R aw data for s oda lime particle copper pipe system Electric field 0 240 174 100 195 152 200 156 119 300 137 106 400 109 83 500 95 73 600 80 67 700 75 56 800 61 50 900 53 41 1100 48 Table B 6. R aw data for s oda lime particle stainless steel pipe system Electric field 0 240 99 100 195 92 200 156 82 300 137 72 400 109 69 500 95 66 600 80 61 700 75 58 800 61 59 900 53 56

PAGE 149

149 Table B 7. R aw data for s oda lime particle natural glass pipe system Electric field 0 240 194 100 195 162 200 156 125 300 137 101 400 109 86 500 95 67 600 80 58 700 75 44 800 61 34 900 53 26 Table B 8. R aw data for s oda lime particle pyrex pipe system Electric field 0 240 202 100 195 169 200 156 133 300 137 104 400 109 86 500 95 72 600 80 62 700 75 50 800 61 40 900 53 37

PAGE 150

150 Table B 9. R aw data for PMMA particle copper pipe system Electric field 0 2249 1477 100 967 1124 200 766 870 300 579 724 400 509 648 500 537 589 600 407 533 700 352 463 800 326 432 900 211 378 Table B 10. R aw data for PMMA particle stainless steel pipe system Electric field 0 2249 1749 100 967 1326 200 766 1211 300 579 1163 400 509 1056 500 537 1008 600 407 959 700 352 846 800 326 762 900 211 686

PAGE 151

151 Table B 11. R aw data for PMMA particle natural glass pipe system Electric field 0 2249 1477 100 967 912 200 766 682 300 579 620 400 509 546 500 537 493 600 407 455 700 352 391 800 326 311 900 211 261 Table B 12. R aw data for PMMA particle pyrex pipe system Electric field 0 2249 1638 100 967 857 200 766 701 300 579 530 400 509 479 500 537 474 600 407 394 700 352 337 800 326 314 900 211 203

PAGE 152

152 Table B 13. R aw data for c rushed soda lime particle copper pipe system Electric field 0 65 45 100 21 22 150 19 11 200 9 2 300 13 11 400 27 26 500 44 36 600 51 47 700 59 51 800 61 47 900 60 55 1000 61 1100 56 Table B 14. R aw data for c rushed soda lime particle copper pipe system Electric field 0 65 41 100 21 17 150 19 11 200 9 5 300 13 5 400 27 13 500 44 22 600 51 26 700 59 32 800 61 34 900 60 33

PAGE 153

153 Table B 15. R aw data for c rushed soda lime particle copper pipe system Electric field 0 65 39 100 21 13 150 19 1 200 9 8 300 13 27 400 27 41 500 44 54 600 51 67 700 59 70 800 61 70 900 60 72 Table B 16. R aw data for c rushed soda lime particle copper pipe system Electric field 0 65 44 100 21 15 150 19 8 200 9 6 300 13 24 400 27 41 500 44 54 600 51 63 700 59 68 800 61 70 900 60 73

PAGE 154

154 Table B 17. R aw data for JSC 1A particle copper pipe system Electric field 0 189 166 100 93 99 200 31 50 300 14 15 400 44 3 500 61 22 600 69 29 700 70 31 800 63 900 48 1000 38 Table B 18. R aw data for JSC 1A particle Stainless steel pipe system Electric field 0 189 103 100 93 60 200 31 38 300 14 22 400 44 13 500 61 4 600 69 0 700 70 1

PAGE 155

155 Table B 19. R aw data for JSC 1A particle natural glass pipe system Electric field 0 189 131 100 93 58 200 31 2 300 14 40 400 44 66 500 61 86 600 69 93 700 70 98 Table B 20. R aw data for JSC 1A particle pyrex pipe system Electric field 0 189 165 100 93 85 200 31 20 300 14 24 400 44 50 500 61 66 600 69 80 700 70 81

PAGE 156

156 APPENDIX C PARAMETRIC STUDY A parametric study was performed to determine the effect of each input on the collision number per unit length, n/L. Each input to the CFD simulation was change independent of other input. The effect of each input on n/L is summarized in Table C 1. The inp uts that strongly affect the particle wall collision are: the mass loading the particle diameter the pipe radius the particle wall coefficient of restitution and the specularity coefficient The coefficient of restitution and the specularity coefficient are functions of particle pipe system, thus these parameter is changes with different material uses. The important parameters for determining the collision number in the CFD simulation are mass loading, particle diameter, and pipe r adius. Unlike what is shown in Chapter 3, the velocity did not affect the collision number per unit length. In the parametric study increasing the air velocity also increase the amount of particles in the system (mass loading is constant); thus the Effect seen in Chapter 2 is included in the mass loading term. Ema et al (2003) indicated that the charging efficiency for each contact is depended upon the impact angle Particle impacting at a certain angle can led up to five times as much charge transfer. The parametric study also looked at the impact angle to determine if this is an issue. The impact angle is found by (C 1) In the CFD model the impact velocity is the same as the fluctuating velocity. The impact angle was determined for each par ameter used in the parametric study. The results are also in Table C 1. Impact angle is insensitive to the operating condition, it

PAGE 157

157 change with the pipe radius, the particle wall coefficient of restitution, and the specularity coefficient. The coefficient o r restitution and speculartity coefficient depend on the particle pipe system. Changing the pipe radius changes the impact angle by about 3. This is small enough not to cause a significant change in the charge transfer for a straight pipe. The effects of each parameter are documented in Figure C 1 to C 10.

PAGE 158

158 Table C 1. Inputs and outputs for the simulation Input Effect on Effect on impact angle with no effect with with no effect no effect no effect no effect with no effect no effect no effect no effect no effect no effect no effect with with with with

PAGE 159

159 A B Figure C 1. The effect of mass loading on A ) collision frequency per pipe length for a single particle, B ) particle wall impact angle

PAGE 160

160 A B Figure C 2. The effect of pipe radius on A ) collision frequency per pipe length for a single particle, B ) particle wall impact angle

PAGE 161

161 A B Figure C 3. The effect of mean gas velocity on A ) collision frequency per pipe length for a single particle, B ) particle wall impact angle

PAGE 162

162 A B Figure C 4. The effect of maximum solid volume fraction on A ) collision frequency per pipe length for a single pa rticle, B ) particle wall impact angle

PAGE 163

163 A B Figure C 5. The effect of particle diameter on A ) collision frequency per pipe length for a single particle, B ) particle wall impact angle

PAGE 164

164 A B Figure C 6 The effect of particle density on A ) collisio n frequency per pipe length for a single particle, B ) particle wall impact angle

PAGE 165

165 A B Figure C 7. The effect of particle sphericity on A ) collision frequency per pipe length for a single particle, B ) particle wall impact angle

PAGE 166

166 A B Figure C 8. The effect of particle particle coefficient of restitution on A ) collision frequency per pipe length for a single particle, B ) particle wall impact angle

PAGE 167

167 A B Figure C 9. The effect of particle wall coefficient of restitution on A ) collision fr equency per pipe length for a single particle, B ) particle wall impact angle

PAGE 168

168 A B Figure C 10. The effect of wall specularity coefficient on A ) collision frequency per pipe length for a single particle, B ) particle wall impact angle

PAGE 169

169 APPENDIX D ADDITIONAL COMPARISO N FOR CHARGE MODEL V ALIDATION A B C D Figure D 1. Comparison for borosilicate particles and copper pipe for different air velocity and mass flow rate

PAGE 170

170 A B C D Figure D 2. Comparison for borosilicate particles and stainless ste el pipe for different air velocity and mass flow rate

PAGE 171

171 A B C D Figure D 3. Comparison for borosilicate particles and natural glass pipe for different air velocity and mass flow rate

PAGE 172

172 A B C D Figure D 4. Comparison for borosilicate particles and P yrex pipe for different air velocity and mass flow rate

PAGE 173

173 A B C D Figure D 5. Comparison for soda lime particles and stainless steel pipe for different air velocity and mass flow rate

PAGE 174

174 A B C D Figure D 6. Comparison for soda lime particles and natural glass pipe for different air velocity and mass flow rate

PAGE 175

175 A B C D Figure D 7. Comparison for soda lime particles and Pyrex pipe for different air velocity and mass flow rate

PAGE 176

176 A B C D Figure D 8. Comparison for PMMA particles and copper pi pe for different air velocity and mass flow rate

PAGE 177

177 A B C D Figure D 9. Comparison for PMMA particles and stainless steel pipe for different air velocity and mass flow rate

PAGE 178

178 A B C D Figure D 10. Comparison for PMMA particles and natural glass pipe for different air velocity and mass flow rate

PAGE 179

179 A B C D Figure D 11. Comparison for PMMA particles and Pyrex pipe for different air velocity and mass flow rate

PAGE 180

180 A B C D Figure D 12. Comparison for crushed soda lime particles and copper pipe for different air velocity and mass flow rate

PAGE 181

181 A B C D Figure D 13. Comparison for crushed soda lime particles and stainless steel pipe for different air velocity and mass flow rate

PAGE 182

182 A B C D Figure D 14. Comparison for crushed soda lime particles and natural glass pipe for different air velocity and mass flow rate

PAGE 183

183 A B C D Figure D 15. Comparison for crushed soda lime particles and Pyrex pipe for different air velocity and mass flow rate

PAGE 184

184 A B C D Figure D 16. Comparison for JSC 1A particles and copper pipe for different air velocity and mass flow rate

PAGE 185

185 A B C D Figure D 17. Comparison for JSC 1A particles and stainless steel pipe for different air velocity and mass flow rate

PAGE 186

186 A B C D Figure D 18. Comparison for JSC 1A particles and Pyrex pipe for different air velocity and mass flow rate

PAGE 187

187 APPENDIX E ADDITIONAL IVIS IMAG ES FOR DIFFERENT PAR TICLE SHAPE Figure E 1. IVIS Lumina images of hollow glass particles

PAGE 188

188 Figure E 2. IVIS Lumina images of glass rod particles

PAGE 189

189 Figure E 3. IVIS Lumina images of glass flake 1 particles

PAGE 190

190 Figure E 4. IVIS Lumina images of glass flake 2 particles

PAGE 191

191 Figure E 5. IVIS Lumina images of crushed glass particles

PAGE 192

192 Figure E 6. IVIS Lumina images of silver flake particles

PAGE 193

193 Figure E 7. IVIS Lumina images of Aluminum Oxide particles

PAGE 194

194 Figure E 8. IVIS Lumina images of JSC 1A particles

PAGE 195

195 LIST OF REFERENCES Ali, M., Mazumder, M. K. and Martonen, T. B. (2009). Measurements of Electrodynamic Effect on the Deposition of MDI and DPI Aerosols in a Replica Cast of Human Oral Pharyngeal Laryngeal Airways, J Aerosol Med Pulm D 22:35 44. Almeria, B., Deng, W., Fahm y, T. M. and Gomez, A. (2010). Controlling the Morphology of Electrospray Generated PLGA Microparticles for Drug Delivery, J Colloid Interface Sci 343:125 133. Anderson, T. B. and Jackson, R. (1967). A Fluid Machanical Description of Fluidized Bed, Ind Eng Chem Fundam 6:527 237. Bailey, A. G., Hashish, A. H. and Williams, T. J. (1998). Drug Delivery by Inhalation of Charged Particles, J Electrostat 44:3 10. Bhaskar, S., Pollock, K. M., Yoshida, M. and Lahann, J. (2010). Towards Designer Microparticl es: Simutaneous Control of Anisotropy, Shape, and Size, Small 6:404 411. Bolio, E., Yasuna, J. and Sinclair, J. (1995). Diluted Turbulent Gas Solid Flow in Riser With Particle Particle Interaction, AIChE J 41:1375 1388. Brand, P., Meyer, T., Haussermann, S., Schulte, M., Scheuch, G., Bernhard, T., Sommerauer, B., Weber, N. and Griese, M. (2005). Optimum Peripheral Drug Deposition in Patients With Cystic Fibrosis, J Aerosol Med 18:45 54. Brauer, H. (1980). Report on Inv estigations on Particle Movement in Straight Horizontal Tubes, Particle/Wall Collision and Erosion of Tubes and Tube Bends, J Powder Bulk Solids Technol 4:3 12. Byron, P. R., Peart, J. and Staniforth, J. N. (1997). Aerosol Electrostatics I: Properties of Fine Powders Before and After Aerosolization by Dry Powder Inhalers, Pharm Res 14:698 705. Cangialosi, F., Liberti, L., Notarnicola, M. and Stencel, J. (2006). Monte Carlo Simulation of Pneumatic Tribocharging in Two Phase Flow for High Inertia Paricles, Powder Technol 165:39 51. Cartwright, P., Singh, S., Bailey, A. G. and Rose, L. J. (1985). Electrostatic Charging Characteristics of Polyethylene Powder During Pneumatic Conveying, IEEE T Ind Appl 1A 21:541 546. Castle, G. S. P. (1997). Contact Charging Between Insulators, J Electrostat 40&41:13 20.

PAGE 196

196 Champion, J. A., Katare, Y. K. and Mitrag otri, S. (2007). Particle Shape: A New Design Parameter for Micro and Nanoscale Drug Delivery Carriers, J Control Release 121:3 9. Chan, T. L., Lippmann, M., Cohen, V. R. and Schlesinger, R. B. (1978). Effect of Electrostatic Charges on Particle Depositi on in a Hollow Cast of the Human Larynx Tracheobronchial Tree, J Aerosol Sci 9:463 468. Chan, T. L., Schreck, R. M. and Lippmann, M. (1980). Effects of the Laryngeal Jet on Particle Deposition in the Human Trachea and Upper Bronchial Airways, J Aerosol Sci 11:447 459. Davies, C. N. (1973). Diffusion and Sedimentation of Aerosol Particles from Poiseuille Flow in Pipes, J Aerosol Sci 4:117 128. Deng, X., Chen, D. and Yang, W. (2011). Study on Electrodynamic Sensor of Multi Modality System for Multiphase Flow Measurement, Rev Sci Instrum 82:124701. Eckhoff, R. K. (2011). Are Enhanced Dust Explosion Hazards to be Forseen in Production, Processing and Handeling of Powders Consisting of Nano Size Particles, in Nanosafe 2010: International conference on sa fe production and use of nano materials J. Phys: Conf. Ser. 304. Edwards, D. A., Hanes, J., Caponetti, G., Hrkach, J., Ben Jebria, A., Eskew, M. L., Mintzes, J., Deaver, D., Lotan, N. and Langer, R. (1997). Large Porous Particles for Pulmonary Drug Delive ry, Science 276:1868 1871. Eggels, J. G. M., Unger, F., Weise, M. H., Adrian, R. J., Friedrich, R. and Nieuwstadt, F. T. M. (1994). Fully Developed Turbulent Pipe Flow: A Comparison Between Direct Numerical Simulation and Experiment, J Fluid Mech 268:17 2 209. Ema, A., Yasuda, D., Tanoue, K. and Masuda, H. (2003). Tribo Charge and Rebound Characteristics of Particles on Inclined or Rotating Metal Target, Powder Technol 135 136:2 13. Emmett, P., C,, Aitken, R. J. and Hannan, W. J. (1982). Measurements of the Total and Regional Deposition of Inhaled Particles in the Human Respiratory Tract, J Aerosol Sci 13:549 560. Farkas, A. and Balashazy, I. (2007). Simulation of the Effect of Local Obstructions and Blockage on Airflow and Aerosol Deposition in Centralhuman Airways, J Aerosol Sci 38:865 884. Ferin, J., Mercer, T. T. and Leach, L. J. (1983). The Effect of Aerosol Charge on the Deposition and Clearance of TiO 2 Particles in Rats, Environ Res 31:148 151.

PAGE 197

197 Forward, K. M., Lacks, D. J. and Sankaran, R. M. (2009). Triboelectric Charging of Granular Insulator Mixtures Due Solely To Particle Particle Interactions, Ind En g Chem Res 48:2309 2314. Gajewski, J. B. and Szaynok, A. (1981). Charge Measurement of Dust Particles in Motion, J Electrostat 10:229 234. Gajewski, J. B. (1996). Electrosatic, Inductive Ring Probe Bandwidth, Meas Sci Technol 7:1766 1775. Gajewski, J. B. (1997). Electric Charge Measurement in Pneumatic Installations, J Electrostat 40&41:231 236. Gajewski, J. B. (2006). Non Contact Electrostatic Flow Probes for Measuring the Flow Rate and Charge in the Two Phase Gas Solid Flows, Chem Eng Sci 61: 2262 2270. Gawronski, R. and Szewczyk, K. W. (1986). Inertial Deposition of Particles in the Human Branching Airways, J Aerosol Sci 17:795 801. Geng, Y., Dalhaimer, P., Cai, S., Tsai, R., Tewari, M., Minko, T. and Discher, D. E. (2007). Shape Effects of Filaments Versus Spherical Particles in Flow and Drug Delivery, Nat Nanotechnol 2:249 255. Glor, M. (1985). Hazards Due To Electrostatic Charging of Powders, J Electrostat 16:175 191. Gondret, P., Lance, M. and Petit, L. (2002). Bouncing Motion of Sph erical Particles in Fluids, Phys Fluids 14:643 652. Haider, A. and Levenspiel, O. (1989). Drag Coefficient and Terminal Velocity of Sphere and Non Spherical Particles, Powder Technol 58:63 70. Hartman, R. P. A. (1998). Electrohydrodynamic Atomization in the Cone Jet Mode. From Physical Modeling to Powder Production, in Applied Sciences Delf University of Technology, Delft, The Netherlands. Hassan, M. S. and Lau, R. W. M. (2009). Effect of Particle Shape on Dry Particle Inhalation: Study of Flowability, Aerosolization, and Deposition Propeties, AAOS PharmSciTech 10:1252 1262. Hinds, W. C. (1982). Aerosol technology: Properties, behavior, and measurement of airborne particles 2nd eds. John Wiley & Sons, New York. Hirst, P. H., Pitcairn, G. R., Weers, J. G., Tarara, T. E., Clark, A. R., Dellamary, L. A., Hall, G., Shorr, J. and Newman, S. P. (2002). In Vivo Lung Deposition of Hollow Porous Part icles from a Pressurized Metered Dose Inhaler, Pharm Res 19:258 264.

PAGE 198

198 Howatson Tawhai, M., Pullan, A. J. and Hunter, P. J. (2000). Generation of An Anatomically Based Three Dimensional Model of the Conducting Airways, Ann Biomed Eng 28:793 802. Hsieh, D. S. T., Rhine, W. D. and Langer, R. (1983). Zero Order Controlled Release Polymer Matrices for Micro and Macromolecules, J Pharm Sci 72:17 22. Hu, T. T., Zhao, H., Jiang, L. C., Le, Y., Chen, J. F. and Yun, J. (2008). Engineering Pharmaceutical Fine Particles of Budesonide for Dry Powder Inhalation (DPI), Ind Eng Chem Res 47:9623 9627. Itakura, T., Masuda, H. and Ohtsuka, C. (1996). The Contact Potential Difference of Powder and the Tribo Charge, J Electrostat 38:213 226. Johnson, P. and Jackson R. (1987). Frictional Collisional Constitutive Relations for Granular Materials with Applications to Plane Shearing, J Fluid Mech 176:67 93. Kaialy, W., Alhalaweh, A., Velaga, S. and Nokhodchi, A. (2011). Effect of Carrier Particle Shape on Dry Powder Inhaler Performance, Int J Pharm 421:12 23. Kanazawa, S., Ohkibo, T., Nomoto, Y. and Adachi, T. (1995). Electrification of a Pipe Wall During Powder Transport, J Electrostat 35:47 54. Kim, C. S. and Fischer, D. M. (1999). Deposition Characteristics of Aerosol Particles in Sequentially Bifurcating Airway Models, Aerosol Sci Technol 31:198 220. Kwok, P. C. L., Glover, W. and Chan, H. (2005). Electrostatic Charge Characteristics of Aeros ols Produced from Metered Dose Inhalers, J Pharm Sci 94:2789 2799. Labiris, N. R. and Dolovich, M. B. (2003). Pulmonary Drug Delivery. Part I: Physiological Factors Affecting Therapeutic Effectiveness of Aerosolized Medications, Br J Clin Pharmacol 56:588 599. LaMarche, C. Q., Curtis, J. S. and Metzger, P. T. (2011). Permeability of JSC 1A: A Lunar Soil Simulant, Icarus 212:383 389. Laufer, J. (1954). The Structure of Turbulence in Fully Developed Pipeflow, in NACA Report Leong, F. Y., Smith, K. A., Wang, C., Matsusaka, S. and Hua, J. (2009). Electrostatic Effects on Inertial Particle Transport in Bifurcated Tubes, AIChE J 55:1390 1401. Lim, E. W. C., Yao, J. and Zhao, Y. (2012). Pneumatic Transport of Granular Material with Electrostatic Effects, A IChE J 58:1040 1059.

PAGE 199

199 Liu, J., Rasheed, A., Dong, H., Carr, W. W., Dadmun, M. D. and Kumar, S. (2008). Electrospun Micro and Nanostrctured Polymer Particles, Macromol Chem Phys 209:2390 2398. Longest, P. W., Vinchurkar, S. and Martonen, T. (2006). Tran sport and Deposition of Respiratory Aerosols in Models of Childhood Asthma, J Asrosol Sci 37:1234 1257. Louge, M., Mastrorakos, E. and Jenkins, J. (1991). The Role of Particle Collision in Pneumatic Transport, J Fluid Mech 231:345 359. Lun, C., Savage S., Jeffery, D. and Chepumiy, N. (1984). Kinetic Theory of Granular Flow: In Elastic Particles in Coutte Flow and Slightly Inelastic Particle in General Flow Field, J Fluid Mech 140:223 256. Marianecci, C., Di Marzio, L., Rinaldi, F., Carafa, M. and Al haique, F. (2011). Pulmonary Delivery: Inovative Approaches and Perspectives, J Biomater Nanobiotech 2:567 575. Marijnissen, J. C. M., Yurteri, C. U., Erven, J. and Ciach, T. (2010). Medicine Nanoparticle Production by EHDA, in Nanoparticles in Medicine and Environment, Inhalation and Health Effects J. C. M. Marijnissen and L. Gradon, eds., Springer Science, 39 58. Masuda, H., Komatsu, T. and Iinoya, K. (1972). The Static Electrification of Particles in Gas Solid Pipe Flow, AIChE J 22:558 564. Matsusaka, S., Ghadiri, M. and Masuda, H. (2000). Electrification of an Elastic Sphere by Repeated Impacts on a Metal Plate, J Phys D: Appl Phys 33:2311 2319. Matsusaka, S., Umemoto, H., Nishitani, M. and Masuda, H. (2002). Electrostatic Charge Distrib ution of Particles in Gas Solids Pipe Flow, J Electrostat 55:81 96. Matsusaka, S. and Masuda, H. (2006). Simultaneous Measurement of Mass Flow Rate and Charge To Mass Ratio of Particles in Gas Solids Pipe Flow, Chem Eng Sci 61:2254 2261. Matsusaka, S. Oki, M. and Masuda, H. (2007). Control of Electrostatic Charge on Particles by Impact Charging, Adv Powder Technol 19:229 244. Matsusaka, S., Fukuda, H., Sakura, Y., Masuda, H. and Ghadiri, M. (2008a). Analysis of Pulsating Electric Signals Generated in Gas Solids Pipe Flow, Chem Eng Sci 63:1353 1360. Matsusaka, S., Ando, K. and Tanaka, Y. (2008b). Development of Electrostatic Charge Controller for Particles in Gases Using Centrifugal Contact, J Soc Powder Technol 45:380 386.

PAGE 200

200 Matsusaka, S., Maruyama, H., Matsuyama, T. and Ghadiri, M. (2010). Triboelectric Charging of Powders: A Review, Chem Eng Sci 65:5781 5807. Matsusaka, S. (2011). Control of particle tribocharging, KONA 29:27 38. Matsuyama, T. and Yamamoto, H. (1995). Characterizing the Electrostatic Charging of Polymer Particle by Impact Charging Experiments, Adv Powder Technol 6:211 220. Matsuyama, T., Ogu, M., Yamam oto, H., Marijnissen, J. C. M. and Scarlett, B. (2003). Impact Charging Experiments with Single Particles of Hundred Micrometre Size, Powder Technol 135 136:14 22. Maynard, A. D. and Kuempel, E. D. (2005). Airborne Nanostructured Particles and Occupationa l Health, J Nanopart Res 7:587 614. Melandri, C., Tarroni, G., Prodi, V., De Zaiacomo, T., Formignani, M. and Lombardi, C. C. (1983). Deposition of Charged Particles in the Human Airways, J Aerosol Sci 14:657 669. Nieh, S. and Nguyen, T. (1988). Effects of Humidity, Conveying Velocity, and Particle Size on Electrotatic Charges of Glass Beads in a Gasous Suspension Flow, J Electrostat 21:99 114. Oldham, M. J. (2000). Computational Fluid Dynamic Predictions and Expe rimental Results for Particle Deposition in an Airway Model, Aerosol Sci Technol 32:61 71. Patankar, S. and Spadling, D. (1967). Heat and Mass Transfer in Boundary Laryer Morgan Grampain, London. Pavia, D., Thomson, M. L., Clarke, S. W. and Shannon, H. S. (1977). Effect of Lung Function and Mode of Inhalation on Penetration of Aerosol into the Human Lung, Thorax 32:194 197. Peart, J., Kulphaisal, P. and Orban, J. C. (2003). Relevance of Electrostatics in Respiratory Drug Delivery, Business Briefing: Pha rmagenerics :84 87. Prodi, V. and Mularoni, A. (1985). Electrostatic Lung Deposition Experiments with Humans and Animal, Ann Occup Hyg 29:229 240. Rao, A., Curtis, J., S., Hancock, B. C. and Wassgres, C. (2011). Numerical Simulation of Dilute Turbulent G as Particle Flow with Turbulence Modulation, AIChE J Rhodes, M., ed. (2008). Introduction to particle technology John Wiley & Sons Inc., Hoboken, NJ.

PAGE 201

201 Sanders, M. (2011). Pulmonary drug delivery: An historical overview, in Controlled pulmonary drug delive ry H. D. C. Smyth and A. J. Hickey, eds., Springer, New York, NY, 51 73. Schlesinger, R. B., Bohning, D. E., Chan, T. L. and Lippmann, M. (1977). Particle Deposition in a Hollow Cast of the Human Tracheobronchial Tree, J Aerosol Sci 8:429 445. Sinclair, J. and Mallo, T. (1998). Describing Particle Turbulence Interaction in a Two Fluid Modeling Framework, in ASME Fluid Engineering Division Summer Meeting ASME Press, New Yorks, 7 14. Smeltzer, E. E., Weaver, M. L. and Klinzing, G. E. (1982). Indi vidual Electrostatic Particle Interaction in Pneumatic Transport, Powder Technol 33:31 42. Sommerfield, M. and Huber, N. (1999). Experimental Analysis and Modeling of Particle Wall Collisions, Int J Multiphas Flow 25:1457 1489. Su, W. C. and Cheng, Y. C. (2009). Deposition of Man Made Fibers in Human Respiratory Airway Casts, J Aerosol Sci 40:270 284. Supuk, E., Hassanpour, A., Ahmadian, H., Ghadiri, M. and Matsuyama, T. (2011). Tribo Electrification and Associated Segregation of Pharmaceutical Bulk Powders, KONA 29:208 223. Tanoue, K., Tanaka, H., Kitano, H. and Masuda, H. (2001). Numerical Simulation of Tribo Electrification of Particles in a Gas Solids Two Phase Flow, Powder Technol 118:121 129. Timbrell, V. (1965). The Inhalation of Fibrous Dusts Ann NY Acad Sci 132:255 273. Vercoulen, P. H. W. (1995). Electrostatic Processing of Particles. A Tool in Particle Technology, in Applied Sciences Delf University of Technology, Delft, The Netherlands. Watanabe, H., Ghadiri, M., Matsuyama, T., Ding, Y. L., Pitt, K. G., Maruyama, H., Matsusaka, S. and Masuda, H. (2007). Triboelectrification of Pharmaceutical Powders by Particle Impact, Int J Pharm 334:149 155. Watano, S., Saito, S. and Suzuki, T. (2003). Numerical Simulation of Electrostatic Charge in Powder Pneumatic Conveying Process, Powder Technol 135 136:112 117. Weibel, E. R. (1963). Morphometry o f t he Human Lung Springer, Berlin. Yao, J., Zhang, Y., Wang, C. H. and Liang, Y. C. (2006). On the Electrostatic Equilibrium of Granular Flow in Pneumatic Conveying System, AIChE J 52:3775 3793.

PAGE 202

202 Zeng, X. M., Martin, G. P., Marriott, C. and Pritchard, J. (2000). The Influence of Carrier Mor phology on Drug Delivery by Dry Powder Inhalers, Int J Pharm 200:93 106. Zhang, Z. and Kleinstreuer, C. (2001). Effect of Particle Inlet Distributions on Deposition in a Triple Bifurcation Lung Airway Model, J Aerosol Med 14:13 29.

PAGE 203

203 BIOGRAPHICAL SKETCH Pongpumin Poom Bunchatheervate was born in Bangkok, Thailand in 1998. He later move o Kentucky, USA. It was in Kentucky that received his bachelor degree in c hemical e ngineering (2006) and his MBA (2007) at the University of Kentucky. In 2007, he joi ned the University of Florida to pursue his PhD in c hemical e ngineering.