Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Developmemt of UV photoreactor models for water treatment Elyasi, Siamak 2009

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

Item Metadata

Download

Media
24-ubc_2009_fall_siamak_elyasi.pdf [ 5.71MB ]
Metadata
JSON: 24-1.0058653.json
JSON-LD: 24-1.0058653-ld.json
RDF/XML (Pretty): 24-1.0058653-rdf.xml
RDF/JSON: 24-1.0058653-rdf.json
Turtle: 24-1.0058653-turtle.txt
N-Triples: 24-1.0058653-rdf-ntriples.txt
Original Record: 24-1.0058653-source.json
Full Text
24-1.0058653-fulltext.txt
Citation
24-1.0058653.ris

Full Text

DEVELOPMENT OF UV PHOTOREACTOR MODELS FOR WATER TREATMENT by SIAMAK ELYASI B.Sc., Sharif University of Technology, Iran, 1985 M. Sc., Tarbiat Modares University, Iran, 1997 M. Sc., Chalmers University of Technology, Sweden, 2003  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES (CHEMICAL AND BIOLOGICAL ENGINEERING)  THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) April 2009 © Siamak Elyasi, 2009  Abstract Ultraviolet (UV) reactors are promising for the future of drinking water and wastewater treatment technology. Models simulating the performance of UV reactors enhance our understanding of the fundamental principles governing the operation of these units. When modeling the performance of UV reactors, governing equations for all related phenomena are derived and solved. The theoretical models and experimental approaches for evaluating the results of these models are comprehensively reviewed and presented in this research. The thesis presents a step-by-step methodology for solving the governing equations of UV reactors. This research presents a general method that integrates the Fresnell, Snell, and Beer-Lambert laws for modeling the radiant distribution in a medium. The model uses the boundary conditions to realistically simulate the fluence/irradiance rate around the radiant source, in particular, in the zone closest to the radiant source. Different low-pressure UV lamps were tested under different operating conditions using photodiodes and a radiometer. The experimentally measured irradiance rate is in excellent agreement with the results of the simulation. Conventionally, the performance of a UV reactor is evaluated using the concentration of photoreactive chemicals at the outlet vs. the inlet. This research presents a novel method for measuring the concentration distribution of a photoreactive chemical inside a photoreactor using a modified planar laser-induced fluorescence method. The fluence distribution was measured for a pilot scale photoreactor under different operating conditions. The visualized result of the fluence distribution revealed significant information about the local/overall performance of the photoreactor. This method is a  ii  powerful diagnostic tool for the determination of the local performance inside a UV reactor, as well as for the evaluation of models simulating UV reactor behavior. A computational fluid dynamic (CFD) model was developed in order to simulate UV photoreactors in the Eulerian framework for microbial disinfection and chemical removal using a UV-based hydroxyl radical initiated oxidation process. The integrated CFD model of UV photoreactor performance was successfully validated by comparisons with experimental results. This verified procedure can be applied to the simulation and optimization of UV photoreactors with different geometries and operating conditions.  iii  Table of Contents Abstract  .....................................................................................................................ii  Table of Contents .........................................................................................................iv List of Tables ................................................................................................................ ix List of Figures................................................................................................................ x Nomenclature ............................................................................................................. xvi Dedication ................................................................................................................... xx Acknowledgements..................................................................................................... xxi Co-authorship Statement .......................................................................................... xxii Chapter 1. UV Photoreactor Principles........................................................................ 1 1.1 Introduction .......................................................................................................... 1 1.2 Modeling .............................................................................................................. 4 1.2.1 Hydrodynamics................................................................................................ 5 1.2.2 Optical Modeling of Radiation ......................................................................... 6 1.2.3 Kinetics ........................................................................................................... 8 1.2.3.1 Kinetics of Microorganism Inactivation................................................... 8 1.2.3.2 Kinetics of Chemical Contaminant Decomposition.................................. 9 1.3 Model Evaluation ............................................................................................... 12 1.3.1 Evaluation of Hydrodynamic Model .............................................................. 12 1.3.2 Evaluation of Radiation Model....................................................................... 13 1.3.3 Evaluation of the Integrated Model of Reactor Performance........................... 13 1.3.3.1 Evaluation of UV Reactor for Microbial Disinfection............................ 13 1.3.3.2 Evaluation of UV Reactor for Removing Chemical Contaminants ......... 14 1.4 Thesis Objectives................................................................................................ 15 1.4.1 Hydrodynamics.............................................................................................. 15 1.4.2 Radiation Field .............................................................................................. 15 1.4.3 Kinetics ......................................................................................................... 16 1.4.4 Integrated Model............................................................................................ 16 1.5 Significance ........................................................................................................ 17 1.6 References .......................................................................................................... 18  iv  Chapter 2. General Method of Simulating Radiation Fields Using Measured Boundary Values ...................................................................................... 25 2.1 Introduction ........................................................................................................ 25 2.2 Radiation Modeling ............................................................................................ 27 2.2.1 One-Dimensional (1D) Source ....................................................................... 32 2.2.2 Two-Dimensional (2D) Source ...................................................................... 32 2.2.3 Three-Dimensional (3D) Source..................................................................... 33 2.3 Model Evaluation ............................................................................................... 34 2.3.1 Experimental Setup........................................................................................ 34 2.3.2 Results ........................................................................................................... 36 2.4 Conclusions ........................................................................................................ 45 2.5 References .......................................................................................................... 47 Chapter 3. Simulation of UV Photoreactor for Water Disinfection in Eulerian Framework ................................................................................ 50 3.1 Introduction ........................................................................................................ 50 3.2 Modeling Procedure............................................................................................ 52 3.2.1 Fluid Flow Modeling ..................................................................................... 52 3.2.2 Volumetric Inactivation Rate Modeling.......................................................... 54 3.2.3 Radiation Modeling ....................................................................................... 57 3.2.4 Efficiency of a UV Photoreactor .................................................................... 58 3.2.5 Computational Strategy for an Integrated UV Photoreactor Performance ....... 61 3.3 CFD Model Setup............................................................................................... 62 3.4 Experimental Procedure...................................................................................... 62 3.4.1 MS2 Fluence–Response and Bioassay............................................................ 63 3.4.2 UV Photoreactor Bioassay ............................................................................. 63 3.4.3 UV Photoreactor Performance........................................................................ 63 3.4.4 UV Photoreactor Design Optimization ........................................................... 66 3.5 Conclusion.......................................................................................................... 68 3.6 References .......................................................................................................... 70 Chapter 4. Fluence Distribution and Performance Evaluation of UV Reactor Using Optical Diagnostic Techniques ...................................................... 73  v  4.1 Introduction ........................................................................................................ 73 4.2 Principles of the Measurement Techniques ......................................................... 75 4.2.1 Planar Laser-Induced Fluorescence (PLIF)..................................................... 75 4.2.2 Chemical Kinetics.......................................................................................... 77 4.3 Experimental ...................................................................................................... 78 4.3.1 Materials and Chemicals ................................................................................ 78 4.3.2 Reaction Rate Measurement under Controlled Conditions in Collimated-Beam Photoreactor...................................................................... 79 4.3.3 Concentration Measurements by PLIF in Flow-Through Photoreactor ........... 82 4.3.3.1 Flow-Through Photoreactor................................................................... 82 4.3.3.2 PLIF Tests............................................................................................. 84 i. PLIF Setup................................................................................................. 84 ii. Calibration Procedure ............................................................................... 85 iii. Concentration Measurement..................................................................... 86 4.4 Results and Discussions ...................................................................................... 87 4.4.1 Optical Absorbance of Rhodamine WT Solution............................................ 87 4.4.2 Direct Oxidation of Rhodamine WT............................................................... 88 4.4.3 Photolysis of Rhodamine WT ........................................................................ 89 4.4.4 Photo-Initiated Oxidation of Rhodamine WT with Hydrogen Peroxide and UV Radiation.......................................................................................... 89 4.4.5 Concentration Profile in the Photoreactor....................................................... 90 4.5 Source of Errors and Uncertainties...................................................................... 95 4.6 Conclusion.......................................................................................................... 97 4.7 References .......................................................................................................... 99 Chapter 5. Simulation of UV Photoreactor for Degradation of Chemical Contaminants: Model Development and Evaluation ............................ 102 5.1 Introduction ...................................................................................................... 102 5.2 Theory .............................................................................................................. 104 5.2.1 Mass and Momentum Conservation (Hydrodynamics) ................................. 104 5.2.2 Radiant Energy Conservation....................................................................... 105 5.2.3 Species Mass Conservation .......................................................................... 106  vi  5.3 Experimental Methods ...................................................................................... 108 5.3.1 Material ....................................................................................................... 108 5.3.2 Concentration Measurement......................................................................... 108 5.3.3 Flow-Through Pilot-Scale Photoreactor ....................................................... 109 5.3.3.1 Velocity Measurement ......................................................................... 109 5.3.3.2 Radiant Energy Measurement.............................................................. 111 5.3.3.3 Photoreaction Rate Measurement ........................................................ 112 5.3.3.4 Concentration Profile of RhWT Measurement (PLIF).......................... 113 5.4 CFD Model Setup............................................................................................. 115 5.5 Results and Discussion ..................................................................................... 116 5.5.1 Evaluation of the Hydrodynamic Model....................................................... 116 5.5.1.1 Velocity Field Measurement (PIV)...................................................... 116 5.5.1.2 Simulation of Hydrodynamics ............................................................. 118 5.5.2 Evaluation of the Radiation Model............................................................... 121 5.5.3 Kinetic Model Determination....................................................................... 123 5.5.4 Evaluation of the Integrated Model .............................................................. 124 5.6 Source of Uncertainty and Errors ....................................................................... 127 5.6.1 Uncertainty in Velocity Field Measurements................................................. 128 5.6.2 Uncertainty in the Concentration Field Measurements .................................. 130 5.7 Conclusions ...................................................................................................... 131 5.8 References ........................................................................................................ 133 Chapter 6. Conclusions and Recommendations....................................................... 136 6.1 Conclusions of the Research ............................................................................. 136 6.2 Significance of the Research ............................................................................. 138 6.2.1 Theoretical Achievements............................................................................ 138 6.2.1.1 Radiation Modeling............................................................................. 139 6.2.1.2 Microbial Disinfection Modeling with the Eulerian Approach ............. 139 6.2.2 Experimental Works Accomplishments........................................................ 139 6.2.2.1 Oxidation Rate of Rhodamine WT using a UV-based AOP ................. 139 6.2.2.2 Concentration Profile Measurement..................................................... 140 6.3 Limitations ....................................................................................................... 140  vii  6.3.1 Theoretical Part – Radiation......................................................................... 140 6.3.2 Experimental Part ........................................................................................ 141 6.3.2.1 Hydrodynamics ................................................................................... 141 6.3.2.2 Reaction Rate of Oxidation of RhWT .................................................. 141 6.4 Recommendations for Future Work .................................................................. 141 6.4.1 Theoretical Work ......................................................................................... 142 6.4.1.1 Hydrodynamic..................................................................................... 142 6.4.1.2 Radiation............................................................................................. 142 6.4.2 Experimental Work...................................................................................... 143 6.4.2.1 Hydrodynamic..................................................................................... 143 6.4.2.2 Reaction Rate ...................................................................................... 143 6.4.2.3 Planar Laser-Induced Fluorescent (PLIF) ............................................ 144 6.5 References ........................................................................................................ 146 Appendix 1. C Codes for Modeling Radiation Distribution in the Photoreactor ........................................................................................... 148 Appendix 2. Reaction Rate Data of RhWT and H2O2 under UV Radiation ........... 152 Appendix 3. Radiant Power Measured and Simulated Values ................................ 155 Appendix 4. Technical Limitations and Solutions in Detail..................................... 161 1 Limitations .......................................................................................................... 161 1.1 Theoretical Part – Radiation............................................................................ 161 1.2 Experimental .................................................................................................. 162 1.2.1 Hydrodynamics ...................................................................................... 162 1.2.2 Concentration Profile in the UV Photoreactor (PLIF Method) ................ 163 2 Recommendations for Improving Experimental Setups........................................ 164 Appendix 5. Analytical Solution of Plug Flow Photoreactor ................................... 165  viii  List of Tables 1.1. The primary reactions in the H2O2/UV-based AOPs system.................................... 10 3.1. Performance evaluation of the two industrial UV photoreactors from biodosimetry experiments and simulation studies. Log reductions were calculated based on outlet/inlet concentrations of MS2 in photoreactors. For simulation results, maximum and minimum values correspond to the sleeve efficiencies of 100% and 90%, respectively. For experimental results, the values represent deviation with 95% confidence intervals of the average log reduction ................................................................................................................ 65 Appx 5.1. 1D model predicted RhWT Concentration at one specific location for different operating conditions (0.006±0.002, 0.015±0.002, and 0.020±0.002 kg/s, corresponding to velocity ratio 2 and 3) ...................................................... 167  ix  List of Figures 1.1. Radiant energy balance for a control volume. ........................................................... 6 2.1. Conservation of radiant energy. .............................................................................. 28 2.2. Pathway of a ray from source to active area through the UV lamp body (1), the sleeve (2), the reflector (3), and the active area sensor (4)....................................... 30 2.3. Position of reflector glasses under a UV lamp in a quartz sleeve. The apparatus consists of: two window glasses (1), a UV photodiode extension arm (2), a UV sensor (3), a UV lamp with 200 W input power (4), a quartz sleeve (5), and the position of the hole in the sleeve (6). ......................................................... 35 2.4. The supporting mechanism of the UV lamp and sensor consists of: a vertical rail (1), a horizontal rail (2), a linear stage for the lamp (3), a linear stage in z direction (4), a linear stage for the sensor (5), the prism holder (6), the UV sensor extension arm (7), the UV photodiode (8), and the UV lamp (9). ................. 36 2.5. Experimental values for the power of the larger lamp (symbols) in air vs. calculated values (continuous lines) using the 1D model. ....................................... 39 2.6. Experimental values for the power of large lamp (symbols) in air vs. calculated values (continuous lines) using the 3D model. ....................................... 39 2.7. Experimental values for the power of the larger lamp (symbols) in air vs. calculated values (continuous lines) using a 3D model, assuming a cylindrical emitter with a 7 mm diameter. ................................................................................ 40 2.8. Experimental values for the radiant power of the larger lamp (symbols) in a quartz sleeve with no air flowing inside the sleeve vs. the 1D model-predicted values (continuous lines) and the 3D model, cylindrical emitter with 7 mm diameter, values (dashed lines). .............................................................................. 41  x  2.9. Experimental values for the radiant power of the larger lamp (symbols) in a quartz sleeve with air flowing inside the sleeve (0.001 m3/s) vs. the 1D modelpredicted values (continuous lines) and the 3D model, cylindrical emitter with 7 mm diameter, values (dashed line)....................................................................... 42 2.10. Effect of reflection on radiation distribution of the larger UV lamp in a quartz sleeve with air flowing inside the sleeve (1 L/s). Experimental value for the radiation power (symbols) vs. the 1D model-predicted values (continuous lines) and the 3D model, cylindrical emitter with 7 mm diameter, values (dashed lines). ........................................................................................................ 43 2.11. Irradiance rare distribution around a high-output UV lamp in stagnant ambient air; measurements (symbols), 1D modeling predictions (continuous lines), and 3D modeling, cylindrical emitter with 3.25 mm diameter, predictions (dashed lines). ...................................................................................... 44 2.12. Simulation results for partial radiation inside the plasma of a high-output UV lamp in the air; measurements (symbols) and partial 3D modeling, cylindrical emitter with 3.25 mm diameter, predictions (continuous lines). .............................. 45 3.1. Log reduction of MS2 at various UV fluences (dose–response curve). .................... 57 3.2. Profiles of total received fluence (dose) inside S12Q and S8Q UV photoreactors.......................................................................................................... 66 3.3. Profiles of received fluence (dose) in cross sections inside side inlet port UV photoreactors. Type A and B are flat-end sleeve and hemispherical-end sleeve, respectively. ........................................................................................................... 68 4.1. Collimated-beam photoreactor setup consisting of: polished aluminum reflector (1), mid-size UV lamps (2), collimator (3), quartz window (4), jacketed reactor body (5), variable speed stirrer (6), thermometer (7), UV sensors (8), amplifier (9), voltmeter (10), sampling ports (11), and UV spectrophotometer with circulating pump (12)........................................................ 80  xi  4.2. The flow-through UV photoreactor used for PLIF measurements (all dimensions are expressed in cm). ........................................................................... 82 4.3. Schematic of pilot-scale photoreactor consisting of: product reservoir (1), feed reservoir (2), stirrer (3), centrifugal pump (4), flow/pressure/temperature meter (5), photoreactor (6), laser source (7), digital camera (8), PLIF control unit (9), data acquisition system (10), and online spectrophotometer (11). .............. 83 4.4. Schematic view of the PLIF apparatus for measuring concentrations of the photoreactive solution. The setup consists of: a pulsed laser at 532 nm (1); a digital camera with a high-pass filter (2), a digital camera with a band-pass filter (3), a photoreactor (4), a beam splitter (5), a diffusive reflector (6), reemitted light at 588 nm (7), and laser sheets (8, 9).................................................. 85 4.5. Concentration profile of rhodamine WT (ppb) in the UV reactor at steady state condition without correction for laser energy variation. Color bar (linear scale) indicates concentrations from 130 (red) to 80 ppb (blue) and arrows show the flow directions. Illustrations 1-3 are for mass flow rates of 0.006, 0.015, and 0.020 kg/s, respectively. ........................................................................................ 90 4.6. Concentration profile of rhodamine WT (ppb) in the UV reactor with correction for the laser energy variation at steady state condition. Color bar (linear scale) indicates concentrations from 130 (red) to 80 ppb (blue), and arrows show the flow directions. Illustrations 1-3 are for mass flow rates of 0.006, 0.015, and 0.020 kg/s, respectively. ............................................................. 91 4.7. Received UV fluence profile of rhodamine WT (ppb) in the UV reactor under steady-state conditions. Color bar (linear scale) indicates fluence, from 60 (red) to 5 J/m2 (blue) and arrows show the flow directions. Illustrations 1-3 are for mass flow rates of 0.006, 0.015, and 0.020 kg/s, respectively. ........................... 92 4.8. Measured velocity profile inside the reactor using PIV (first half of the reactor)................................................................................................................... 93  xii  4.9. Uncertainty in RhWT concentration profile measurement (ppb) with a 95% confidence interval in the UV reactor. Color bar (linear scale) indicates concentration uncertainty from ±30 (red) to 0 ppb (blue), and arrows show the flow directions. Cases 1-3 are for mass flow rates of 0.006, 0.015, and 0.020 kg/s, respectively.................................................................................................... 97 5.1. The photoreactor used for PLIF measurements (all dimensions are expressed in cm)................................................................................................................... 109 5.2. Schematic view of pilot-scale photoreactor, which consists of: product reservoir (1), feed reservoir (2), stirrer (3), centrifugal pump (4), flow/pressure/temperature meter (5), photoreactor (6), laser source (7), digital camera (8), PIV/PLIF control unit (9), data acquisition system (10), and online spectrophotometer (11). ............................................................................. 110 5.3. Reflector glasses over the UV lamp in the quartz sleeve. The apparatus consists of: two window glasses (1), a UV photodiode extension arm (2), a UV sensor (3), a UV lamp (4), a quartz sleeve (5), and the position of the hole in the sleeve (6).................................................................................................... 112 5.4. Schematic diagram of bench-scale collimated-beam UV photoreactor: parabolic reflector (1), UV lamp (2), collimator (3), double jacket rector (4), stirrer (5), and thermometer (6). ........................................................................... 113 5.5. PIV results at mid cross-section of the reactor for two different mass flow rates: 0.005 (A) and 0.014 (B) kg/s. Due to the light scattering and reflection, unrealistic velocity vectors were generated from some parts of the reactor. These have been removed (blank areas)................................................................ 117 5.6. Velocity (x-component) profile and contour of turning points at different sections of the reactor for two different flow rates. Vertical lines indicate different x-positions in cm. ................................................................................... 118  xiii  5.7. Simulated velocity profile (CFD results) at mid cross-section of reactor for two different flow rates: 0.005 (A) and 0.014 (B) kg/s.......................................... 119 5.8. Velocity difference (vectors), VPIV -VSIM, and x-components of velocity reactor. Vertical lines indicate x-position and curves show zero axial velocity difference (VxPIV-VxSIM). The color bar (linear scale) indicates the velocity difference (vectors) from 0.015 (red) and 0.03 (red) to 0 m/s (blue) for cases A and B, respectively. The arrows show the flow directions. The velocity measurements near the surfaces are removed (white regions) because of the high uncertainty in the measurement due to the reflections. .................................. 120 5.9. Irradiance rate at different distances indicated by  ,  ,  , , and  for  distances 1, 10, 20, 30, and 50 mm from the surface of the lamp, respectively. Experimental values for the radiation power (symbols) vs. the 1D modelpredicted values (continuous lines) and the 3D model, cylindrical emitter with 7 mm diameter, values (dashed lines). .................................................................. 122 5.10. Fluence rate (W/m2) profile at the mid-cross-section of the photoreactor. ........... 123 5.11. Concentration profile of rhodamine WT (ppb) in the UV reactor with the laser energy at a steady-state condition for two different mass flow rates: 0.006 (A) and 0.015 (B) kg/s. .............................................................................. 125 5.12. Concentration profile of rhodamine WT in the UV photoreactor calculated by the integrated model for two different mass flow rates: 0.006 (A) and 0.015 (B) kg/s. Concentrations less than 80 ppb are shown as white............................... 126 5.13. Concentration difference (CPLIF - CSimulation) in ppb for two different mass flow rates: 0.006 (A) and 0.015 (B) kg/s. Concentration differences higher than 30 ppb are shown as white. ........................................................................... 126 5.14. Uncertainty in the RhWT concentration profile measurement (ppb) with a 95% confidence interval in the UV reactor for two different mass flow rates: 0.006 (A) and 0.015 (B) kg/s. The color bar (linear scale) indicates the  xiv  concentration uncertainty from ±30 (red) to 0 ppb (blue), and the arrows show the flow directions................................................................................................ 127 5.15. Standard deviation of velocity (m/s) profiles for two different mass flow rates, 0.005 (A) and 0.014 (B) kg/s. Arrows show the direction of flow. The calculated average measured velocity using PIV (all measured value divided by the number of data) for cases A and B are 0.023 and 0.053 m/s, respectively. ......................................................................................................... 129 5.17. Velocity difference contour (m/s) for two different mass flow rates, 0.005 (A) and 0.014 (B) kg/s. Arrows show flow direction............................................. 130 5.17. Standard error of concentration profile (ppb) for two different mass flow rates, 0.006 (A) and 0.015 (B) kg/s. Arrows show flow direction.......................... 131 Appx 5.1. Plug flow UV reactor. ................................................................................. 166  xv  Nomenclature A  Area (m2)  ABSλ Absorbance at wavelength λ Alamp  Surface area of UV lamp (m2)  C  Concentration of microorganism (kg/m3)  C0  Initial concentration of microorganism (kg/m3)  CRhWT Concentration of rhodamine WT (ppb) CRhWT,0 Initial Concentration of rhodamine WT (ppb) CV1  Concentration of rhodamine WT (ppb) at velocity V1  CV2  Concentration of rhodamine WT (ppb) at velocity V2  c  Concentration (M)  ci  Concentration of ith component (M)  Dm,i  Molecular diffusion coefficient (m2/s)  Ec  Captured Energy (J)  Eλ  Volumetric absorbed energy of spectrum with “λ” wavelength (J/m3)  Ei,λ  Volumetric absorbed energy of spectrum with “λ” wavelength by ith component (J/m3)  F  Volumetric external forces (N/m3)  G  Fluence rate (W/m2)  Gabs  Absorbed fluence rate (W/m2)  Gvol  Volume-weighted average fluence rate (W/m2)  Gλ  Fluence rate of spectrum with “λ” wavelength (W/m2)  g  Gravitational acceleration (m/s2)  H  Local absorbed fluence (dose) by microorganisms or chemicals (J/m2)  Hmax  Maximum theoretical absorbed fluence (J/m2)  Hreal  Measured absorbed fluence (J/m2)  h  Planck’s constant (J.s)  Iλ  Radiant intensity of spectrum with “λ” wavelength (W/sr)  I  Average radiant intensity of entire spectrum (W/sr)  I0  Average radiant intensity of entire spectrum at source of radiation (W/sr)  Ispec  Intensity passing through spectrophotometer cell (W/sr)  xvi  Ispec ,0 Intensity entering into cell (W/sr) Ir  Irradiance rate (W/m2)  jeff,i  Effective molecular diffusive rate of species i (kg/s.m3)  jλ  Intensity diffusive flux for spectrum wavelength λ (W/sr.m2)  K  Received local radiant energy by microorganism per unit volume (J/m3)  Ksetup Setup slope (sr.s/M) Koffset Setup intercept (J) k  Absorption coefficient of the medium for the whole spectrum (m-1)  kave  Average absorption coefficient of the medium for whole spectrum (1/m)  ki  Absorption coefficient of medium, ith (1/m)  klocal  Local absorption coefficient of medium (1/m)  kturb  Turbulent kinetic energy (m2/s2)  kλ  Absorption coefficient of medium for spectrum wavelength λ (1/m)  k′  Inactivation rate (m2/J)  L  Distance or length (m)  Lave  Average path of radiation through the element (m)  Leff  Effective distance between source and medium position (m)  Li  Pathway length of ray through medium, ith (m)  Lphoto Length of photo-reactive region in photoreactor (m) lcell  Light path length of spectrophotometer cell (cm)  lc,p  Distance between sensor and emitter (m)  llamp  Length of cylindrical lamp (m)  ls,p  Distance of investigation point from the light source (m)  m  Number of different optical media  N  Number of microorganism per unit volume (MPN/m3)  N0  Initial number of microorganism in unit volume (MPN/m3)  Np  Avogadro’s constant  N  Number of moles  P  Power (W)  Pi,λ  Volumetric absorbed power of spectrum with “λ” wavelength by ith component (W/m3)  xvii  Ppoint  Power at studied point (W)  PV  Volumetric emitted power (W/m3)  Px  Power per unit of length (W/m)  Px,α  Power of each element on the surface source at location x and angle α per unit area (W/m2)  Px,α,r  Power of each volumetric source element per unit of volume (W/m3)  Pλ  Volumetric absorbed power of spectrum with “λ” wavelength (W/m3)  p  Pressure (Pa)  Q  Volumetric flow rate (m3/s)  Qλ  Energy of one mole of photon at a specific wavelength “λ” (einstein)  ri  Volumetric rate of Photolysis for ith component (M/s)  rlamp  Radius of cylindrical lamp (m)  rreactor Internal radius of annular UV photoreactor (m) rsleeve External radius of UV lamp sleeve Si  Source term of species i (kg/s.m3)  s  Position vector (m)  T  Transmittance  Ti  Fraction of transmitted portion of ray from one medium to another  Tc  Fraction of light that reaches the sensor  Ts  Fraction of light passing through solution  t  Time (s)  u  Velocity vector (m/s)  V  Photo-reactive volume of UV photoreactor (m3)  Ve  Volume of element (m3)  v  Velocity of medium (m/s)  Vx  x component of velocity vector (m/s)  Vy  y component of velocity vector (m/s)  vc  Speed of light (m/s)  x0  Initial x coordinate of lamp ends (m)  xf  Final x coordinate of lamp ends (m)  xviii  xi  Mass fraction of species i  x, y , z Cartesian coordinates (m) xp,yp,zp Coordinates of investigation point in medium Z  Susceptibility constant for microorganism (m2/J)  α  Temporary angles (radians)  ε  Turbulent dissipation rate (m2/s3)  εi,λ  Molar absorption coefficient of spectrum with “λ” wavelength by ith component (m2/kmole)  ελe  Molar absorption coefficient of solution at wavelength λe nm (m2/kmole)  ελs  Molar absorption coefficient of solution at wavelength λs nm (m2/kmole)  ηcon  UV photoreactor efficiency based on concentration log reduction  ηDose  UV photoreactor efficiency based on delivered fluence  Θi  Quantum yield of ith component  Θc  Quantum efficiency of sensor  Θs  Quantum efficiency of florescent chemical  θ  Angle (radians)  θLn  Angle between the normal vector of the studied area and the incident ray (radians)  λ  Wavelength of radiation beam (m)  ρ  Density (kg/m3)  σT,i  Turbulent Schmidt number for component i  σλ  Scattering coefficient of media for the spectrum wavelength λ (1/m)  τ  Viscous stress tensor (Pa)  νT  Turbulent momentum diffusivity (m2/s)  Ψ  Phase function  Ω  Solid angle (steradians)  Ω′  Solid angle (steradians)  xix  Dedication I dedicate this thesis to my wife, Mahnaz, who did more than her share around the house as I focused and worked on my research project. Without her support in providing such a gentle atmosphere, I would still be trying to resolve the first stage of this project.  xx  Acknowledgements This research project would not have been possible without the support of many people. The author wishes to express his gratitude to his supervisor, Dr. Fariborz Taghipour, for his effective support and guidance. Deepest gratitude is also due to the Ph.D. thesis committee, Drs. Sheldon I. Green, Chad P.J. Bennington, and John R. Grace, for their constructive guidance and comments on the quality of this research. Special thanks also to Drs. Bruce D. Bowen and Majid Mohseni for their kind technical and lab equipment support. The author would also like to convey profound gratitude to the Killam Trust, for awarding the Izaak Walton Killam Memorial Predoctoral Fellowship, the Natural Sciences and Engineering Research Council of Canada (NSERC), for providing financial support, and R-Can Environmental Inc. for providing all valuable experimental data supporting chapter 3 of the thesis. The author wishes to express his love and gratitude to his beloved family for their understanding and endless love throughout the duration of his studies.  xxi  Co-authorship Statement The results of this research are presented in four different manuscripts which correspond to chapters two to five. The manuscripts’ authors are Siamak Elyasi, and Fariborz Taghipour. Dr. F. Taghipour is my research supervisor at the University of British Columbia. The literature review, topics research idea development, experimental design, experimental setups design/building/assembling, performing experiments and data collection, data analysis, and report preparation were done extensively by S. Elyasi, under supervision of Dr. F. Taghipour who financially supported all research activities. Finally, I, S. Elyasi, prepared final draft for each manuscript after careful revision and approval of my research supervisor, Dr. F. Taghipour.  xxii  Chapter 1. UV Photoreactor Principles 1.1 Introduction Chlorination of drinking water is the most effective method for disinfecting and removing almost all pathogens (excluding protozoa); many countries have used this method for many decades. However, the disinfectants themselves can react with naturally occurring materials in water to form unintended organic and inorganic disinfection byproducts (DBPs) which may pose health risks. Concentrations and types of DBPs vary based on the source of raw water. In 1979, the United States Environmental Protection Agency (US EPA) set the maximum limit for the total trihalomethanes (TTHM) content in drinking water at no more than 100 µg/L. In 1997, the US EPA lowered the regulation to 80 µg/L [1]. In Canada, this limit has been set at 100 µg/L since 2006 [2]. Morris et al. [3] have shown a clear and significant association between TTHM and neoplastic disease, as well as bladder and colon cancer. TTHM is one of 280 DBPs that have been found in drinking water, and few of these have been tested for their potential toxicity to humans. For example, the DPB, N-nitrosodiummethylamine (NDMA) can be produced through a reaction involving monochloramine [4] and is classified as carcinogenic by the US EPA. A risk assessment has shown a 10-6 (1 person in 1 million studied sample) lifetime risk level of cancer from exposure to NDMA at 0.7 ng/L [5]. In Canada, there is no maximum contaminant level specified for provincial or federal drinking water. A major challenge for water suppliers is how to balance the risks from microbial pathogens and disinfection byproducts.  1  Ozonation is a possible alternative to chlorination, as it has no harmful byproducts (in the absence of bromide), but it is not as effective as chlorination. Another substitute for drinking water disinfection is UV irradiation that uses high-energy electromagnetic wave, in the range of 250-260 nm. UV technology provides an effective disinfection method for both water and wastewater treatment. There are several advantages to this technology, such as [6]: - UV can inactivate (disinfect) most viruses, spores, and cysts; - UV disinfection is a physical process using electrical power. No reactive chemicals are involved that might cause the elimination, handling, or generation of hazardous, toxic, and corrosive chemicals; - There is no toxic residue that might endanger aquatic or human life; - UV disinfection is user-friendly for operators; - UV disinfection equipment requires less space than other methods. In contrast, the major drawbacks of UV technology are: - Insufficient radiant energy in the reactor may not effectively eliminate microorganisms; - Fouling and scaling are major issues affecting the treatment of hard water and the iron ion content in hard water can dramatically decrease the performance of UV reactors. Periodic cleaning of UV lamp sleeves can control fouling; - Performance of UV reactors is highly affected by the optical properties of water. Any type of turbidity drastically reduces the efficiency of the system; - Capital cost of UV technology is higher than chlorination technology;  2  - Exposure to dry lamps can produce deleterious health effects (total exposure to UV should be less than 6 mW/cm2 per day). In addition to disinfection of microorganisms, UV can also remove persistent chemical contaminants in water when used in combination with an oxidant, such as hydrogen peroxide. This is called a UV-based advanced oxidation process (AOP). Combining UV radiation and an oxidant chemical, like hydrogen peroxide, produces hydroxyl radicals which are highly reactive. The redox potentials of the most commonly used oxidizers—fluorine, hydroxyl radicals, atomic oxygen, ozone, hydrogen peroxide, permanganate, and chlorine dioxide—are 3.03, 2.80, 2.42, 2.07, 1.77, 1.67, and 1.5 volts, respectively [7]. This implies that hydroxyl radicals have a very high reaction rate constant and the capability to oxidize major persistent chemicals [8]. AOP can be employed to optically clear wastewater containing contamination levels less than 1000 ppm [9] or COD levels in wastewater less than 5 g/L [10]. The overall performance of a UV reactor for the inactivation of microorganisms or removal of contaminants is dramatically affected by the operating parameters within a photoreactor. Considering all the parameters in a system and their complex interactions brings us to the need for at least a semi-mechanistic model. A numerical solution of the mathematical model using Computational Fluid Dynamics (CFD) would enhance our knowledge of the principles involved and better equip us to find the means of improving the performance of UV reactors. In considering the microbial disinfection of water, two strategies can be employed to develop a useful model. In the first approach, the Lagrange method, the medium (water) is treated as a continuous phase and the microorganisms as a discrete phase.  3  Applying drag force with a random walk algorithm tracks the particles (microorganisms) from inlet to outlet. On each track, the radiant energy received is calculated and interpreted to find the total disinfection rate using averaging methods [11-14]. If the microorganisms are treated as a reactive species, the mass conservation of the species can be applied to resolve the concentration of microorganisms or any other chemical pollutant at any position in the photoreactor. In this approach, the Eulerian method, the microorganisms are treated as a continuous phase in the same manner as any other reacting species. Accordingly, the fluence (dose) received by the microorganism should be adapted for a continuous phase [12,15,16]. A very common evaluation method of UV reactor performance models is to compare the predicted and experimental results of the concentration or number of active microorganisms at the outlet stream for known inlet values [11-16]. However, this method is not a comprehensive evaluation technique. It compares only the inlet/outlet concentrations and does not reveal any discrepancies between the model and the experiments concerning the concentration profiles inside the reactor. The present research tries to resolve the limitations of the current validation methods by developing a new, non-intrusive means of measuring the concentration profile inside a UV reactor. In addition, for a complete validation, each component of the integrated model of reactor performance, hydrodynamics (velocity field), radiation, and species concentration are evaluated separately.  1.2 Modeling Future development and improvements to existing UV-AOP systems require an understanding of the characteristics of the system, including the hydrodynamics of the  4  UV-reactors, the optical characteristics of the entire domain, and the kinetics of the reactions that inactivate microorganisms or oxidize chemical contaminants, in addition to the manner in which these characteristics interact with one another. 1.2.1 Hydrodynamics The velocity field in a system can be simulated by solving the equations of conservation of mass and momentum. The general form of single-phase conservation of momentum is expressed as: ∂ (ρu ) + ∇ ⋅ (ρuu ) = ∇p − ∇ ⋅ τ + ρg + F ∂t  1.1  where ρ, u, p, τ, g, and F are medium density, velocity vector, pressure, viscous stress tensor, gravitational acceleration, and external body force, respectively. A three-dimensional, time-dependent numerical solution of Equation 1.1 (Direct Numerical Simulation, DNS) is only applicable for low Reynolds number flow regimes due to the extensive computational resources required. At higher Reynolds number, acceptable engineering approaches to solving Equation 1.1 are statistical methods or classical approaches, including solving the Reynolds Average Navier-Stokes (RANS) form of the equation [17]. In the RANS approach, the Reynolds stress tensors are semiempirically correlated using algebraic [18], one-equation [19], two-equation (such as standard k- ε [20], Re-Normalisation Group (RNG) k- ε [21], realizable k- ε [22], or standard k-ω [23]) and multiple equation of turbulence models (such as the Reynolds Stress Model (RSM) [24-26]).  5  1.2.2 Optical Modeling of Radiation UV radiation plays a major role in photoreactors. The distribution of radiant energy is the main operating parameter governing the local kinetic rate of photoreactions. A mathematical representation of the radiant energy field can be obtained by solving the Radiation Transfer Equation (RTE). The photon balance equation (or RTE), as shown in Equation 1.2 for a control volume (Figure 1.1), can be analytically solved for simple geometries. For more complex geometries, many studies have used the finite volume numerical method to solve the RTE for simple reactors [27-30]. Although the finite volume model yields a general solution, it has some limitations, such as spatial and directional discretization errors, as investigated by Raithby [27]. emission from other zones  incoming scattered radiation from Ωʹ to Ω Radiation direction Ω  outgoing radiation  incoming radiation outgoing scattered radiation  emission to the other zones  Figure 1.1. Radiant energy balance for a control volume. For each directional unit vector (Ω) at a specific wavelength (λ), the Radiation Transfer Equation (RTE) can be obtained from the following photon balance [28]: (outgoing intensity - incoming intensity) + absorbed intensity + outgoing scattered = (incoming emission – outgoing emission to other parts) + incoming scattered ∇[Ω I λ (s , Ω, t )] + k λ (s , Ω , t )I λ (s , Ω , t ) + σ λ (s, Ω, t )I λ (s, Ω, t ) = jλ (s, t ) +  1 σ λ (s, t )∫ Ψ (Ω ′Ω )I λ (s, Ω ′, t )dΩ ′ Ω′= 4π 4π  1.2  6  where s, Ω, t, Iλ(s,Ω,t), kλ, σλ, jλ, and Ψ(ΩʹΩ) are position vector, directional unit vector, time, radiant intensity for specific solid angle Ω and specific location s, absorption coefficient of the medium, scattering coefficient of the particulates in the medium, radiation emission, and phase function, respectively. In the absence of emission (operating at low temperature) and scattering (clear solution without particulates and bubbles), Equation 1.2 can be simplified to Equation 1.3 which it is easily solvable by point source summation and ray tracing methods. dI (s, Ω ) + k ( s, Ω ) I (s, Ω ) = 0 ds  1.3   L  I (L, Ω ) = I 0 (Ω ) exp − ∫ k ( s, Ω ) ds   0   1.4  where I(L,Ω), I0, and k are the radiant intensity at distance L from the source for an average spectrum and a specific solid angle Ω, radiant intensity at the source, and the local absorption coefficient, respectively. Equations 1.3 and 1.4 are the Beer-Lambert equation for steady-state conditions in the differential and integral forms, respectively. Using the definition of radiant intensity, the expression for the received fluence rate (G) for a specific solid angle (Ω) can be derived as:  L  1 G (L, Ω ) = 2 I 0 (Ω ) exp − ∫ k ( s, Ω ) ds  L  0   1.5  Equation 1.5 is the key formula for deriving a mathematical model of the radiation distribution for any type of radiant point source. Radiation modeling is explained in detail in Chapter 2.  7  1.2.3 Kinetics  For UV microbial disinfection, the inactivation rate of microorganisms is a dimerization process that occurs on the double strand of the microorganism’s DNA. It is typically related to the total amount of radiant energy received by the microorganism. In a UV-based advanced oxidation process, the reaction rate of organic material decomposition is directly related to the generation of hydroxyl radicals. The following sections review these two processes separately. 1.2.3.1 Kinetics of Microorganism Inactivation  Germicidal ultraviolet radiation (at 253.7 nm) provides enough energy to initiate a reaction between two thymine molecules in the DNA of microorganisms [30]. The UV light causes the dimerization of adjacent thymine molecules in the DNA, which results in the inactivation of microorganisms, thereby rendering them harmless [30]. The rate of inactivation of microorganisms is related to the received fluence (dose). For some microorganisms, the inactivation rate by UV radiation can be approximated using firstorder kinetics [32]:  N (t ) = exp (− GZt ) N0  1.6  where N, N0, G, and Z, are the number of active microorganism after time t and at initial time in a unit volume, received fluence rate, and susceptibility constant for the microorganism, respectively. Noakes et al. [33] found that first-order kinetics is an appropriate mathematical model for simulating the inactivation rate of airborne microorganisms. However, for many waterborne microorganisms, this model may not be applicable. Other models may be employed as presented by other researchers [e.g. 34]. All these models are valid under 8  specified conditions. No single model can predict the broad range of change of optical transmittance, particle content, hardness, and chemical concentrations. Using data obtained from the experimentally measured UV response curve for microorganisms under real conditions is the best way to model the disinfection rate. 1.2.3.2 Kinetics of Chemical Contaminant Decomposition  The hydroxyl radical is a highly reactive intermediate that oxidizes broad ranges of organic and inorganic contaminants in water [7]. The hydroxyl radical is a key component in Advanced Oxidation Processes (AOPs) and can be produced from oxidant chemicals such as ozone, hydrogen peroxide, and oxygen in the presence of UV radiation. UV-based AOPs are industrially well established (and installed at over 200 sites), primarily in North America [35], and have several advantages, such as a high rate of deterioration of persistent organic materials (compared to oxidants alone) and the production of harmless byproducts. Three primary reactions take place for degrading an organic material (RH) in an UV-hydrogen peroxide-based AOP [36]: Direct oxidation: RH + H2O2  Oxidized Product + H2O  Photolysis:  RH*(intermediates)  Products  1.8  Intermediates  Products  1.9  RH (aq) + hν  Hydroxyl radical RH + OH˚ oxidation:  1.7  The kinetic mechanism of hydroxyl radical production from hydrogen peroxide in aqueous solutions under exposure to UV radiation is summarized in Table 1.1.  9  Table 1.1. The primary reactions in the H2O2/UV-based AOPs system No. Reaction Photolysis of hydrogen peroxide: 1H2O2 + hν 2OH˚ 2HO2 + hν OH˚ + ½ O2 Equilibrium and dissociation in water: 3H2O2 H+ + HO24H2O H+ + OHPropagation stage: 5H2O2 + OH˚ HO2˚ +H2O 6HO2- + OH˚ HO2˚+ OH˚ 7HO2 H+ + ˚O28H+ + ˚O2HO2˚ ˚ 9H2O2 + HO2 O2+OH˚+ H2O 10H2O2 + ˚O2O2+OH˚+ OHTermination stage: 11OH˚ + OH˚ H2O2 12OH˚ + ˚O2OH- + O2 13OH˚ + HO2˚ H2O + O2 14HO2˚+ ˚O2HO2- + O2 ˚ ˚ 15HO2 + HO2 H2O2 + O2  Rate Constant  Reference  Θ 1 = 0.5 Θ 2 = 0.5  37 37  pK3 = 11.6 pK4 = 14  38  k5 = 2.7 * 107 M-1 s-1 k6 = 7.5 * 109 M-1 s-1 k7 = 3.2 * 105 s-1 k8 = 2 * 1010 M-1 s-1 k9 = 3 M-1 s-1 k10 = 0.13 M-1 s-1  39 40 41 38 42 43  k11 = 5 * 109 k12 = 1010 k13 = 6.6*109 k14 = 9.7 * 107 k15 = 8.3 * 105  40 40 44 45 46  M-1 s-1 M-1 s-1 M-1 s-1 M-1 s-1 M-1 s-1  To derive a mathematical model of UV photolysis reactions in an oxidant solution (Equation 1.9 and reactions 1 and 2 in Table 1.1), the local rate of absorbed radiant energy in the medium should be calculated. Considering the light absorbance correlation for a medium, UV radiation can be absorbed by specific chemicals of the medium according to:  Ei , λ =  ε i , λ ci n  ∑ε  j ,λ  Eλ  1.10  cj  j =1  where Ei,λ, Eλ, εi,λ, ci, and n are the volumetric rate of energy absorbance for wavelength λ by the ith component, the rate of local volumetric absorbed energy, the molar extinction coefficient of the ith component for wavelength λ, the concentration of the ith component, and the number of light absorbing chemical components in the medium, respectively.  10  Considering the Beer-Lambert law, the local volumetric rate of absorbance of radiant power can be derived from an energy balance around a control volume in the medium (an infinitesimal element) as: n  G ( s ) A (1 − exp( −k local Lave ) )  Pλ ( s ) = lim  λ = k local Gλ ( s ) = 2.303Gλ ( s )∑ ε i ,λ ci  Lave→ 0 Ve i =1    1.11  where Gλ and the fluence rate is expressed by: 4π  Gλ ( s ) = ∫ I λ (s, Ω ) dΩ 0  1.12  and Ve, klocal, and Lave (=Ve/A) are the volume of the element, local absorption coefficient in the medium, and the average path length of radiation through the element, respectively. The amount of energy of one mole of a photon can be calculated using Planck’s formula as: Qλ =  h N p vc  1.13 λ where Qλ, h, Np, vc, and λ are the energy of one mole of a photon at a specific wavelength, Planck’s constant, Avogadro’s constant, speed of light, and wavelength of the radiation beam, respectively. The number of moles of the ith component removed through the photolysis reaction is defined as the quantum yield (Θi). Considering Planck’s law (to calculate the energy content of the photons) and quantum yield, the rate of the photolysis reaction is: ri =  Pi ,λ ( s )Θ i Qλ  = 19.25 λε i,λ ci Θi G (s )  1.14  The aforementioned method was used for reaction rate determination of the rhodamine WT/hydrogen peroxide system as a pollutant indicator. A more detailed explanation is given in Chapter 4.  11  1.3 Model Evaluation UV reactors are sophisticated items of equipment; as a consequence, analytical solutions of the governing equations for them do not exist. In addition, the complicated behavior of the UV lamp, the turbulent flow through the reactor, and the effect of the optical properties of the fluid on the disinfection or chemical removal rate can cause unexpected results. As a result, each governing equation and the integrated system of equations (that yield the concentration profile inside the reactor) should be evaluated against experimental data to realistically simulate UV reactors for industrial applications. 1.3.1 Evaluation of Hydrodynamic Model  The Particle Image Velocimetry (PIV) technique and image processing for hydrodynamic tests is a very powerful tool for revealing the flow velocity. In the PIV method, fluid motion is made visible by adding small tracer particles. The scattered light of an almost-perfectly parallel (collimated) laser beam from the particles reveals the positions of the tracer particles at two instants in time; from which their displacements can be calculated. By processing the two captured positions from multiple particles, it is possible to infer the flow velocity field. The PIV technique has been extensively used in velocity studies, in particular for water-jets [e.g. 47,48]. Sozzi [49] used this technique to verify CFD results for an industrial-scale annular UV photoreactor. His investigation showed that the realizable k- ε model acceptably predicted the flow field inside an annular photoreactor with side inlet/outlet ports [49]. This technique was also used to determine the velocity field for this research. A more detailed explanation is given in Chapter 5.  12  1.3.2 Evaluation of Radiation Model  Considering the real characteristics of a UV lamp, the rays pass through the lamp body and lamp sleeve until they reach a point in the medium. This is a complicated phenomenon because of absorbance by as well as reflection and refraction from/through the lamp’s sleeve and body [50]. These complex optical phenomena need to be evaluated by measuring the UV irradiance rate produced by the UV lamp when inserted into a sleeve. Several researchers employed a radiometer to monitor and control the radiation field [51-62]. A research radiometer [63] with a UVC-selective SiC-based UV sensor, was used for conducting experimental measurements. The sensor has a solar blind filter that measures the irradiance rate within wavelengths of 230-285 nm. The relative irradiance rate with respect to one specific point in the medium was used to evaluate the radiation model. This method was applied in order to reduce the effect of measurement uncertainties on the model evaluation. Experimental measurement of a radiation field using a radiometer is presented in Chapter 2. 1.3.3 Evaluation of the Integrated Model of Reactor Performance  The UV reactor model validation reported in the literature is limited to measuring the inlet/outlet concentrations [11-16]. In this research, for the first time, a new method has been developed for measuring the performance of UV reactors for removing chemical contaminants using concentration profile measurements throughout the entire UV reactor. 1.3.3.1 Evaluation of UV Reactor for Microbial Disinfection  To evaluate the mathematical models of UV reactors for microbial disinfection, the MS2 bacteriophage was used as a model microorganism to determine the fluence  13  delivered by two industrial UV-photoreactors under various operating conditions (Chapter 3). An MS2 bioassay was conducted at the photoreactors’ inlets and outlets in order to determine their inactivation under various conditions. The developed integrated model of the UV photoreactor for microbial inactivation was successfully evaluated for two industrial UV photoreactors. The log-scaled MS2 concentration ratio of the reactor outlet to inlet (log reduction) was used to compare the bioassay with the simulation result. Considering the uncertainty in the measurement of the total MS2 count, which is a characteristic of the bioassay, the simulation predictions were in good agreement with the experimental results (Chapter 3). 1.3.3.2 Evaluation of UV Reactor for Removing Chemical Contaminants  Conventionally, the performance of a UV reactor is evaluated using the ratio of concentration of photoreactive chemicals at the outlet vs. the inlet. However, this method does not provide any information about local performance inside the reactor domain. This research presents a novel method for measuring the distribution of concentration inside a photoreactor using a modified planar laser-induced fluorescent method. Rhodamine WT (RhWT) is selected as the chemical contaminant. The fluorescent characteristic of this chemical (re-emitting light upon excitation) is a function of rhodamine WT concentration. A green pulsed laser sheet (at 533 nm) excites the fluorescent chemical, and it re-emits light at different wavelength (at 588 nm). The re-emitted light can be captured using a digital camera and interpreted as concentration. This method can reveal the concentration profile of RhWT even in a reactive solution under UV radiation with an oxidant (photo-initiated oxidation). Rhodamine WT has a moderate reaction rate with hydroxyl radicals, but it does not show any reaction with oxidants (e.g., hydrogen peroxide) and during photolysis (UV and water). As a result, it is a suitable candidate for 14  revealing valuable information about the local/overall performance of UV reactors. A detailed explanation of the method is presented in Chapter 4.  1.4 Thesis Objectives The overall objectives of this research are to study the phenomena affecting UV photoreactor performance including hydrodynamics, radiant energy, and kinetics, and to develop two semi-mechanistic integrated models to account for all the aforementioned parameters when simulating the behavior of UV microbial disinfection and UV-initiated oxidation (AOP) reactors. The developed models are evaluated against experimental data. The specific objectives of this research are: 1.4.1 Hydrodynamics  The hydrodynamic characteristics of a pilot-scale UV photoreactor are studied using Particle Image Velocimetry (PIV) and, in parallel, a CFD model is developed to predict the velocity field. This study is common to both the microbial disinfection and organic decontamination UV photoreactors (Chapters 3 and 5). 1.4.2 Radiation Field  The mathematical models of UV radiant energy distribution are developed by taking all the optical parameters such as medium absorbance into account. These models are evaluated for different low-pressure UV lamps at different operating conditions using a research radiometer. Comparing the simulation result with the measured radiant field reveals the most accurate model as interpreted by the characteristics of the lamps (Chapter 2).  15  1.4.3 Kinetics  The kinetics of inactivation of MS2 (as a model microorganism) and UV-initiated oxidation of rhodamine WT (as a model organic contaminant) are studied separately as follows: - The inactivation rate of bacteriophage MS2 in the presence of UV radiation as a standard model of a microbiological pollutant is considered in Chapter 3; - The reaction/photoreaction rate of rhodamine WT (RhWT) as a model chemical contaminant is determined using hydrogen peroxide and UV radiation. RhWT, with florescent characteristics, is used to measure the concentration profile through the photoreactor (Chapter 4). 1.4.4 Integrated Model  Taking into account the hydrodynamics, radiant energy, and kinetics, two integrated CFD models are developed for simulating microbial disinfection and organic contaminant UV-initiated oxidation. - UV reactor performance for microbial disinfection is modeled, and the model predicted results are evaluated against bioassay results at the inlet and outlet of a commercial UV photoreactor (Chapter 3); - UV reactor performance for a UV photo-initiated oxidation process is simulated using the photo-kinetic rate of rhodamine WT. The model is evaluated not only by using the concentrations at the inlet and outlet, but also by employing Planar Laser-Induced Fluorescence (PLIF) to determine the concentration distribution through one the vertical cross-section of the photoreactor. This method is a new approach for evaluating the integrated model of UV photoreactors. The method is  16  explained in detail in Chapter 4, and its application to evaluate UV photo-initiated oxidation reactors is presented Chapter 5.  1.5 Significance Considering public health and environmental issues, UV reactors are a promising new technology for the microbial disinfection of drinking water and persistent chemical removal from industrial water/wastewater systems. The future employment of this technology will depend to a great degree, on its technical performance to a great degree. A reliable model that includes all the relevant phenomena is needed to address this issue properly and to enhance our fundamental knowledge of UV reactor operations. Due to the complexity of this system, all the elements of the model should be evaluated both separately and together. This research attempts to cover all of the general issues that arise in the modeling and model evaluation of UV reactors. In addition, a new experimental technique has been developed to measure the performance of UV reactors using fluorescent chemicals, and a general Eulerian method is presented for modeling UV reactors.  17  1.6 References 1- U.S. Environmental Protection Agency (EPA), 1998. National Primary Drinking  Water Regulations: Disinfections and Disinfection Byproducts. Vol. 63, No. 241, 69389-69476. 2- Environmental and Workplace Health, Guideline for Canadian Drinking Water Quality  Summary  Table,  http://www.hc-sc.gc.ca/ewh-semt/pubs/water-eau/sum_guide-  res_recom/chemical-chimiques_e.html#4, visited at April 23, 2008. 3- Morris, R.D., Audet, A.M., Angelillo, I.F., Chalmers, T.C., and Mosteller, F., 1992.  Chlorination, Chlorination By-Product and Cancer: A Meta Analysis. American Journal of Public Health, 82, 7, 955-963. 4- Choi, J., and Valentine, R.L., 2002. Formation of N-Nitrosodimethylamine (NDMA)  from Reaction of Monochloramine: A New Disinfection By-Product. Water Research, 38, 817-824. 5- US EPA., 1997. N-nitrosodimethylamine CASRN 62-75-9, Integrated Risk  Information Service (IRIS) Substance File. 6- Connor, K., Martin, C. O., and Jensen, J. N., 2004. Evaluation of Ultraviolet (UV)  Radiation Disinfection Technologies for Waste Water Treatment Plant Effluent. New York State Energy and Development Authority NYSERDA, Report 04-07. 7- Munter, R., Preis, S., Kallas, J., Trapido, M. and Veressinina, Y, 2001. Advanced  Oxidation Processes (AOPs): Water Treatment Technology for the Twenty-First Century. Kemia-Kemi, 28, 354-362. 8- Von Gunten, U., 2003. Ozonation of Drinking Water: Part I. Oxidation Kinetics and  Product Formation. Water Research, 37, 1443-1467. 9- Kidman, R.B., and Tsuji, K. S., 1992. Preliminary Costs Comparison of Advanced  Oxidation Processes. Los Alamos National Laboratory, LA-12221-MS.  18  10- Andreozzi, R., Caprio, V., Insola, A., and Marotta, R., 1999. Advanced Oxidation  Processes (AOP) for Water Purification and Recovery. Catalysis Today, 53, 51-59. 11- Munoz, A., Craik, S., and Kresta, K., 2007. Computational Fluid Dynamics for  Predicting Performance of Ultraviolet Disinfection -Sensitivity to Particle Tracking Outputs. Journal of Environmental Engineering Science, 6, 285-301. 12- Ducoste, J.J., Liu, D., and Linden, K., 2005. Alternative Approaches to Modeling  Fluence Distribution and Microbial Inactivation in Ultraviolet Reactors: Lagrangian versus Eulerian. Journal of Environmental Engineering, 10, 1393-1403. 13- Lawryshyn, Y.A., and Cairns, B., 2003. UV Disinfection of Water: The Need for UV  Reactor Validation. Water Science and Technology: Water Supply, 3, 293-300. 14- Chiu, K., Lyn, D. A., Savoye, P., and Blatchley, E. R., 1999. Integrated UV  Disinfection Model Based on Particle Tracking. Journal of Environmental Engineering, 125, 7-16. 15- Templeton, M.R., Hofmann, R., Andrews, R.C., and Whitby G.E., 2006.  Biodosimetry Testing of a Simplified Computational Model for the UV Disinfection of Wastewater. Journal of Environmental Engineering Science, 5, 29-36. 16- Lyn, D. A., Chiu, K., and Blatchley, E. R., 1999. Numerical Modeling of Flow and  Disinfection in UV Disinfection Channels. Journal of Environmental Engineering, 125, 17-26. 17- Ranade, V.V., 2002. Computational Flow Modelling for Chemical Reactor  Engineering. Academic Press. First edition. London. England. 18- Prandtl, L., 1925. Bericht Uber Untersuchungen Zur Ausgebildeten Turbulenz. Z.  Agnew. Math. Mech., 5, 136. 19- Spalart, P., and Allmaras, S., 1992. A One-Equation Turbulence Model for  Aerodynamic Flows. Technical Report AIAA-92-0439, American Institute of Aeronautics and Astronautics.  19  20- Launder, B. E., and Spalding, D. B., 1972. Lectures in Mathematical Models of  Turbulence. Academic Press, London, England. 21- Yakhot, V., and Orszag, S. A., 1986. Renormalization Group Analysis of Turbulence,  I. Basic Theory. J. Sci. Comput., 1, 1. 22- Shih, T. H., Liou, W. W., Shabbir, A., Yang. Z., and Zhu, J., 1995. A New k- ε Eddy-  Viscosity Model for High Reynolds Number Turbulent Flows - Model Development and Validation. Computers Fluids, 24, 227-238. 23- Wilcox, D. C., 1998. Turbulence Modeling for CFD. DCW Industries, Inc., La  Canada, California. 24- Launder, B. E., 1989. Second-Moment Closure: Present... and Future?. Inter. J. Heat  Fluid Flow, 10, 282-300. 25- Gibson, M. M., and Launder, B. E., 1978. Ground Effects on Pressure Fluctuations in  the Atmospheric Boundary Layer. J. Fluid Mech., 86, 491-511. 26- Launder, B. E., Reece, G. J., and Rodi, W., 1975. Progress in the Development of a  Reynolds-Stress Turbulence Closure. J. Fluid Mech., 68, 537-566. 27- Raithby G. D., 1999. Evaluation of Discretization Errors in Finite-Volume Radiant  Heat Transfer Predication. Num. Heat Transfer, 36, 241-264. 28- Romero, R. L., Alfano, O. M., and Cassano, A. E., 1997. Cylindrical Photocatalytic  Reactors. Radiation Absorption and Scattering Effects Produced by Suspended Fine Particles in an Annular Space. Ind. Eng. Chem. Res., 36, 3094-3109. 29- Cassano, A. E., Martin, C. A., Brandi, R. J., and Alfano, O. M., 1995. Photoreactor  Analysis and Design: Fundamentals and Application. Ind. Eng. Chem. Res., 34, 21552201. 30- Sommer, R., Pribil, W., Appelt, S., Gehringer, P., Eschweiler, H., Leth, H., Cabaj, A.,  and Haider, T., 2001. Inactivation of bacteriophages in water by means of non-  20  ionizing (uv-253.7 nm) and ionizing (gamma) radiation: a comparative approach. Water Research, 35, 3109-3116. 31- Chui, E. H., and Raithby, G. D., 1993. Computation of Radiant Heat Transfer on a  Non-Orthogonal Mesh Using the Finite-Volume Method. Num. Heat Transfer B., 23, 269-288. 32- Bolton, J. R., and Stefan, M. I., 2002. Fundamental Photochemical Approach to the  Concepts of Fluence (UV dose) and Electrical Energy Efficiency in Photochemical Degradation Reactions. Research on. Chemical Intermediates, 28, 857–870. 33- Noakesa, C. J., Fletchera, L. A., Beggsa, C. B., Sleigha, P. A., and Kerr, K. G., 2004.  Development of a Numerical Model to Simulate the Biological Inactivation Airborne Microorganisms in the Presence of Ultraviolet. Journal of Aerosol Science, 35, 489– 507. 34- Sommer, R., Pribil, W., Appelt, S., Gehringer, P., Eschweiler, H., Leth, H., Cabaj, A.,  and Haider, T., 2001. Inactivation of Bacteriphages in Water by Means of NonIonizing (UV-253.7 nm) and Ionizing (Gamma) Radiation: A Comparative Approach. Water Research, 35, 3109-3116. 35- Stefan, M. I., Hoy, A. R., and Bolton, J. R., 1996. Kinetics and Mechanism of the  Degradation and Mineralization of Acetone in Dilute Aqueous Solution Sensitivity by the UV Photolysis of Hydrogen Peroxide. Environ. Sci. Technol., 30, 2382-2390. 36- Oppenlander, T., 2003. Photochemical Purification of Water and Air, Wiley-VCH. 37- Crittenden, J. C., Hu, S., Hand, D. W., and Green, S.A., 1999. A Kinetic Model for  H2O2/UV Process in a Completely Mixed Batch Reactor. Wat. Res., 33, 2315–2328. 38- Perry, R. H., Green, D. W., and Maloney, J. D., 1981. Chemical Engineering’s  Handbook 5th edition, Mcgraw-Hill, NY 39- Buxton, G. V., Greenstock, C. L., Helman, W. P., and Ross, A. B., 1988. Critical  Review of Rate Constant for Reactions of Hydrated Dlectrons, Hydrogen Atoms and  21  Hydroxyl Radicals (AOH/AO-) in Aqueous Solution. J. Phys. Chem. Ref. Data, 17, 513-886. 40- Beltran, F. J., Rivas, J., Alvarez, P. M., Alonso, M. A., and Acedo, B., 1999. A  Kinetic Model for Advanced Oxidation Processes of Aromatic Hydrocarbons in Water: Application to Phenanthrene and Nitrobenzene. Ind. Eng. Chem. Res., 38, 4189–4199. 41- Rivas, F. J., Beltran, F. J., Carbajo, M., and Gimeno, O., 2003. Homogeneous  Catalyzed Ozone Decomposition in the Presence of Co(II). Ozone Sci. Eng., 5, 261– 271. 42- Koppenol, W. H., Butler, J., and Van Leeuwen, J. W. L., 1978. The Haber-Weiss  Cycle. Photochem. Photobiol. J., 28, 655-660. 43- Weinstein, J., Benon, H. J., and Bielski, H. J., 1979. Kinetics of Interaction OH HO2  and ˚O2- Radicals with Hydrogen Peroxide in Aqueous Solution Peroxide. The HaberWeiss Reaction. J. Am. Chem. Soc., 101, 58-62. 44- Schested, K., Rasmussen, O. L., and Fricke, H., 1968. Rate Constant of OH with  HO2, O2- and H2O2+ from Hydrogen Peroxide Formation in Pulse-Irradiated Oxygenated Water. J. Phys. Chem., 72, 626-631. 45- Peyton, G.A., and Glaze, W. H., 1988. Destruction of Pollutants in Water with Ozone  in Combination with Ultraviolet Radiation. 3. Photolysis of Aqueous Ozone. Envi. Sci. Tech., 2, 761-767. 46- Bielski, H. J., Benon, H. J., Cabelli, D. E., Ravindra, L. A., and Alberta, A. B., 1985.  Reactivity of Perhydroxyl/Superoxide Radicals in Aqueous Solution. J. Phys. Chem. Ref. Data, 14, 1041-1100. 47- Carlomagno, G. M., 2005. Colours in a Complex Fluid Flow. Optics and Laser  Techno., 38, 230-242.  22  48- Park, W. G., Yun, H. S., Chun, H. H., and Kim, M. C., 2005. Numerical Flow  Simulation of Flush Type Intake Duct of Waterjet, Ocean Engineering, 32, 17-18. 49- Sozzi, A., 2005. CFD and PIV Investigation of UV Reactor Hydrodynamics.  University of British Columbia, Chemical and Biological Engineering Department, Master Thesis. 50- Bolton, J. R., 2000. Calculation of Ultraviolet Fluence Rate Distributions in an  Annular Reactor: Significance of Refraction and Reflection. Wat. Res., 34, 33153324. 51- Mamane-Gravetz, H., Linden, K. G., Cabaj, A., and Sommer, R., 2005. Spectral  Sensitivity of Bacillus Subtilis Spores and MS2 Coliphage for Validation Testing of Ultraviolet Reactors for Water Disinfection. Environ. Sci. Technol., 39, 7845-7852. 52- Martino, M., Liss, P. S., and Plane, J. M. C., 2005. The Photolysis of Dihalomethanes  in Surface Seawater, Environ. Sci. Technol., 39, 7097-7101. 53- Shen, C., Fang, S., Bergstrom, D. E., and Blatchley, E. R., 2005. (E)-5-[2-  (Methoxycarbonyl)ethenyl] cytidine as a Chemical Actinometer for Germicidal UV Radiation, Environ. Sci. Technol., 39, 3826-3832. 54- Sundararajan, C., and Falvey, D. E., 2005. Photorelease of Carboxylic Acids, Amino  Acids, and Phosphates from N-Alkylpicolinium Esters Using Photosensitization by High Wavelength Laser Dyes. J. Am. Chem. Soc. (Communication), 127, 8000-8001. 55- Rosenfeldt, E. J., and Linden, K. G., 2004. Degradation of Endocrine Disrupting  Chemicals Bisphenol A, Ethinyl Estradiol, and Estradiol during UV Photolysis and Advanced Oxidation Processes. Environ. Sci. Technol., 38, 5476-5483. 56- Romero, R. L., Alfano, O. M., and Cassano, A. E., 2003. Radiation Field in an  Annular, Slurry Photocatalytic Reactor. 2. Model and Experiments. Ind. Eng. Chem. Res., 42, 2479-2488.  23  57- Sharpless, C. M., Page, M. A., and Linden, K. G., 2003. Impact of Hydrogen  Peroxide on Nitrite Formation During UV Disinfection. Water research, 37, 47304736. 58- Bertilsson, S., and Widenfalk, A., 2002. Photochemical Degradation of PAHs in  Freshwaters and Their Impact on Bacterial Growth – Influence of Water Chemistry. Hydrobiologia, 469, 23–32. 59- Cramer, N. B, Scott, J. P., and Bowman, C. N., 2002. Photopolymerizations of Thiol-  Ene Polymers without Photoinitiators. Macromolecules, 35, 5361-5365. 60- Magnuson, M. L., Kelty, C. A., Sharpless, C. M., Linden, K. G., Fromme, W., Metz,  D. H., and Kashinkunti, R., 2002. Effect of UV Irradiation on Organic Matter Extracted from Treated Ohio River Water Studied through the Use of Electrospray Mass Spectrometry. Environ. Sci. Technol., 36, 5252-5260. 61- Puma, G. L., Yue, P. L., 2001. A Novel Fountain Photocatalytic Reactor: Model  Development and Experimental Validation. Chemical Engineering Science, 56, 27332744. 62- Sharpless, C. M., and Linden, K. G., 2001. UV Photolysis of Nitrate: Effects of  Natural Organic Matter and Dissolved Inorganic Carbon and Implications for UV Water Disinfection. Environ. Sci. Technol., 35, 2949-2955. 63- International Light Inc., 2005. IL1700 datasheet.  http://intl-light.com/product/meter/doc/IL1700.pdf  24  Chapter 2. General Method of Simulating Radiation Fields Using Measured Boundary Values 2.1 Introduction Natural UV radiant energy promotes the production of radicals (e.g. hydroxyl and oxide radicals) that can cause photodegradation, photosynthesis, or photolysis in gases, liquids, or on photocatalytic surfaces without the addition of any other chemicals. The application of UV photoreactors to drinking water is a well-established technology, but other applications of this clean technology are becoming popular. For example, there is growing interest in using UV photoreactors for food preservation [1]. Mahmoud et al. [2] verified the use of UV radiation for the sterilization of cheese whey, while Kucuk et al. [3] successfully pasteurized apple cider using a thin-film UV reactor. In addition, a number of researchers have demonstrated the possibility of converting methane to methanol using UV radiation under the mild conditions of near ambient temperature and pressure [e.g., 4-7]. The modeling of UV photoreactors requires simultaneous solution of the mass and momentum conservation equations, along with the radiation field equation, all of which affect the rate of photochemical reactions. The governing equations of a photoreactive system can be combined and solved simultaneously by applying Computational Fluid Dynamics (CFD). In order to achieve a better understanding of photoreactors for design and optimization purposes, it is essential to have a reliable model for the radiant energy distribution that is capable of simulating the behavior of any UV source. A mathematical representation of the radiant energy can be obtained by A version of this chapter will be submitted for publication. Elyasi, S., Taghipour, F., 2009. General Method for Calculating Radiation Fields Using Measured Boundary Values. 25  solving the Radiation Transfer Equation (RTE). The photon balance equation and the RTE can both be solved analytically for simple geometries. For more complicated geometries, several researchers have used the finite volume numerical method [8-11]. Romero et al. [11] presented a detailed description of radiation field modeling for an annular photocatalytic reactor adapted from the general transport theory and, particularly, from neutron transport applications. Although the finite volume model yields a general solution to the RTE, it has some limitations, such as spatial and directional discretional errors investigated by Raithby [8]. The discretization of the control volume and solid angle should be carefully selected in order to minimize those errors. In addition, this method needs considerable computational resources, particularly when the number of solid angle discretization is increased. These limitations make the utilization of this method problematic, although it has been employed to simulate the radiant field in photoreactors [12]. This work attempts to present a simple, general method of modeling radiant energy distribution for any type of radiant source through consideration of the medium properties and the UV source geometry. It simplifies the RTE to the Beer-Lambert formula, which is easily solvable for all rays from any complex UV source using numerical methods. This method considers refraction/reflection through/from the body of the UV lamp and sleeve of the UV source. In addition, it resolves the reflections from other sources, such as the reactor body. Using an iterative numerical technique, this method can also handle the scattering effect of particulates and optical variations in the medium.  26  2.2 Radiation Modeling Before presenting the general radiation model, certain standard optical radiation terms are provided below: Solid angle, Ω, is defined as the surface area of a section of a sphere divided by r2,  where r is the radius of the sphere. The solid angle has units of steradians and the maximum solid angle is 4π steradians. Radiant power, P, is the rate of radiant energy emitted in all directions by a light  source. In general, radiant power includes all spectral waves emitted by the source. For UV photochemistry or disinfection, P is usually specified only for the effective spectral wave range. Radiant intensity, I, is the total radiant power emitted by a source in a given  direction per solid angle Ω. Note that, in a non-absorbing medium, the radiant intensity does not decay with distance. For a sphere, I = P/4π; for a flat plane, I = P/2π.  Although radiant energy is naturally a scalar parameter, emitted radiation is described by its spectral wavelength range and directional distribution. Radiation transfer is defined as the transmission of electro magnetic radiation through space under the influence of the medium. The chemicals and particles in the medium can absorb, scatter, or re-emit the incident rays. Figure 2.1 shows a photon balance within a control volume of the medium for a specific direction.  27  emission incoming scattered radiation from other from Ω′to Ω zones incoming radiation  Radiation direction Ω  outgoing radiation  outgoing scattered emission to radiation the other zones Figure 2.1. Conservation of radiant energy. For each directional unit vector (Ω) at a specific wavelength (λ), the Radiation Transfer Equation (RTE) can be obtained from the following photon balance equation: (outgoing intensity – incoming intensity) + absorbed intensity = (incoming emission – outgoing emission) + (incoming scattered – outgoing scattered) ∇[Ω I λ (s , Ω, t )] + k λ (s , Ω , t )I λ (s , Ω , t ) + σ λ (s, Ω, t )I λ (s, Ω, t ) = jλ (s, t ) +  1 σ λ (s, t )∫ Ψ (Ω ′Ω )I λ (s, Ω ′, t )dΩ ′ Ω′= 4π 4π  2.1 where s, Ω, t, Iλ(s,Ω,t), kλ, σλ, jλ, and Ψ(Ωʹ Ω) are position vector, directional unit vector, time, radiant intensity for specific solid angle Ω and specific direction s, absorption coefficient of the medium, scattering coefficient of the particulates in the medium, radiation emission, and phase function, respectively. The photon energy equation generally assumes that scattering is multiple, but independent. In many engineering applications, including drinking water treatment, the medium is transparent without solid particulates, or scattering is non-elastic and can be integrated into the absorption coefficient. Thus, the in/out scattering to/from the control volume is negligible. In the majority of photolysis and photocatalytic processes, which are normally carried out at near ambient temperatures, internal emissions can also be neglected. In addition, most such processes operate at steady state. Thus, the above RTE 28  can be simplified for the average spectrum of the light for a specific solid angle. The simplified equation is: dI (s, Ω ) + k ( s , Ω ) I (s, Ω ) = 0 ds  2.2  This is referred to as the Beer-Lambert equation for steady-state conditions. Written in integral form, Equation 2.2 becomes: L I (Ω ) = I 0 (Ω ) exp − ∫ k (s , Ω ) ds   0   2.3  where I(Ω) is the radiant intensity at distance L from the source for an average spectrum wavelength, and I0(Ω) is the radiant intensity at the source. Irradiation is normally measured in terms of the irradiance rate or power flux 2  (W/m ) for each wavelength band [13]. The irradiance rate yields the amount of total energy per unit area, per unit time, from all directions. Using the definition of intensity, the following equation for the irradiance rate can be derived for an infinitesimal circular area, A, centered at the measurement location in the medium: Ir (Ω ) =  I (Ω ) Ω A  2.4  where Ir(Ω) is the irradiance rate for one specific direction. If the area is shrunk to a point, the solid angle, Ω, calculated based on its definition, is equal to  A cos(θ ) , where L is the distance between the source of radiation L2  and the center of the area, and θ is the angle between the ray and the normal vector of the area. Thus, Equation 2.4 can be simplified as: Ir (Ω ) =    1 ( ) I Ω exp − k ( s , Ω ) ds   cos(θ ) 0 ∫ L2  L   2.5  29  Considering refraction and reflection of the rays through and from different media, Equation 2.5 can be generalized as: Ir (Ω ) =   n  m I ( Ω ) cos ( θ ) exp − k ( s , Ω ) ds  ∏ Ti ∑ ∫ i Ln 2 0  i =1 overLi  i =1  n   ∑ Li   i =1  1  2.6  where Li, θLn, ki, and Ti, are the path length of the ray through the ith medium, the angle between the normal vector of the studied area and the incident ray the absorption coefficient of the ith medium, and the fraction of the transmitted portion of the ray from one medium to another, respectively. Removing the cosine of the angle term from Equation 2.6 yields the fluence rate, which is a crucial parameter in photoreaction rate correlations. Figure 2.2 represents the traveling pathway of a ray:  4 L6 L3  L2 L1  T1  T2  L4 T3  L5  θL6  T5  T4 3 2  1 Figure 2.2. Pathway of a ray from source to active area through the UV lamp body (1), the sleeve (2), the reflector (3), and the active area sensor (4).  Considering the refraction index of each medium, Fresnel, and Snell’s laws can be applied to calculate the refracted ray and the fraction of the ray that passes through the interface between two different media. If the reflector (3 in Figure 2.2) is a transparent medium (e.g., the body of UV lamps in the medium), Fresnel’s law can be applied to that  30  surface too. Applying this procedure, called ray tracing, all UV sources yields the total irradiance rate on the studied surface. Integration of the irradiance rate of all elements on the surface (A) yields the total incident radiant power (P). This value is read using a radiometer and a UV sensor, or it changes the concentration per unit time of photosensitive chemicals in actinometry methods. If the surface of the sensor is covered by an optical filter, the effect of the filter should be included in the T term of Equation 2.6. The total power can therefore be calculated using:         n m    2π    1 P = ∫  ∫ I ( Ω ) θ k ( s Ω ) ds T cos exp − ,   dΩ dA ∑ ∏ L i i 0 2 ∫ n n   i =1 overLi  i =1   Area  0       ∑ Li         i =1   ( )  2.7  Equation 2.7 is the key formula for deriving a mathematical model of the radiation distribution for any type of radiant source. Although all terms in Equation 2.7 are well formulated using the optical properties of the medium, no general correlation for intensity at the source is available. Considering some simple assumptions, a mathematical model can be derived for the intensity at the source, I0. These assumptions cannot be applied to all types of UV sources under all operating conditions. This is because the simplified assumptions typically presume that the UV radiant energy from the lamp is distributed uniformly. However, phenomena, such as the cooling effect of the medium on the operation or aging of the lamp, make the assumption of uniform radiant energy distribution not always practical. Three general forms for the UV source distribution can be assumed:  31  2.2.1 One-Dimensional (1D) Source  The UV source is assumed to be a line-diffusive emitter. Emission from each infinitesimal source on the line is uniform in all directions. If the 1D UV source (the xaxis) is located on the lamp axis, Equation 2.7 can be modified to:      n  m  1  1 = Px cos(θ Ln ) exp  − ∑ ∫ k i (s, Ω )ds  ∏ Ti  dx  2 4π x∫0   n   i =1 overLi  i =1    ∑ Li     i =1   xf  Ppoint  2.8  where Ppoint, Px, x0, and xf are the power at the studied point, the power per unit length of each element on the source at location x, the coordinates of the first source element, and the last source element, respectively. If Px is assumed to be uniform along the plasma arc, with no refraction and no reflection from the UV lamp body, the model is that presented by Bolton [14]. If the effect of the sleeve is not taken into consideration either, then one obtains the LSSE model presented by Jacob et al. [15-18]. 2.2.2 Two-Dimensional (2D) Source  This model assumes that the surface of the UV source (cylindrical UV lamp) behaves as a diffusive radiant energy emitter. As a result, refraction and reflection through and from the surface of the UV lamp are ignored. The general form of the equation is:          n  m   rlamp  π −α max  1 = Px ,α cos(θ Ln ) exp − ∑ ∫ k i (s , Ω )ds  ∏ Ti  dα dx  ∫  2 n 4π x∫0  α max    i =1 overLi  i =1       ∑ Li      i =1     xf  Ppoint  2.9  32  where α max    rlamp = arcsin  n   ∑ Li  i =1         2.10  and Px,α and rlamp are the power of each element on the surface source at location x and angle α per unit area, and lamp outer radius, respectively. It is important to note that Equation 2.9 is written using cylindrical coordinates where the x-axis is located on the lamp axis. If Px,α is assumed to be constant and the refraction/reflection terms are canceled, Equation 2.10 is the form presented by Zolner & Williams [19,20]. If the surface is assumed to be a Lambertian emitter, a cosine angle between the exiting ray and the normal vector of the surface should be added to the integral. 2.2.3 Three-Dimensional (3D) Source  This model is a more general form and assumes that all volume elements in the plasma of the UV lamp (cylindrical UV lamp) behave as diffusive emitters. The general form of the equation is:  Ppoint =  rlamp 4π  rlamp  ∫ 0             x f 2π     n  m    1 Px ,α , r cos θ Ln exp  − ∑ ∫ k i (s, Ω )ds  ∏ Ti dα dx dr  ∫ ∫ 2  x0  0   n   i =1 overLi  i =1        ∑ Li            i =1   ( )  2.11 where Px,α,r is the power per unit volume of each volumetric source element. Assuming a uniform power for each element and ignoring the refractions and reflections, Equation 2.11 is in the form presented by Irazoqui et al. [21,22].  33  2.3 Model Evaluation In order to derive a mathematical model for radiant energy distribution in a medium, different assumptions of 1D, 2D, and 3D sources are applied to determine the intensity in Equation 2.2. Selecting the best model requires evaluation of all these assumptions using experimental measurements, as demonstrated in the following sections. 2.3.1 Experimental Setup  Two high-output, low-pressure UV lamps from Emperor Aquatics Inc., Pottstown, PA, were used as UV sources with the following specifications: electrical power inputs 200 W and 50 W, outer diameters 0.0193 m and 0.015 m, arc lengths 1.07 m and 0.38 m (GIA1120T6L, 20050), respectively. A radiometer (IL1700) from International Light Technologies, Peabody, MA, and a solar blind UV sensor (SIC01MC) from Roithner LaserTechnik, Vienna, Austria, were used to measure power at different points from the UV sources under different conditions. The small size of the UV sensor active SiC chip, at 0.0005 m by 0.0005 m, can be approximated as a point. As a result, integration over the surface of the active sensor area is ignored. In addition, a sharp-cut filter at the top, and an aperture below, contract the spectrum responsiveness and field-of-view of the sensor to 235-280 nm and 45 degrees (approx.), respectively. The active sensor area is located 0.0054 m from the top of the filter. The effect of a quartz sleeve was investigated for the larger lamp (200 W) using a quartz tube with an inner/outer diameter of 0.0222/0.0252 m and a length of 1.3 m with an opening (diameter 0.006 m) at the dead end. Cooling air inside the sleeve was circulated using a vacuum pump to cool the lamp along its axis. For generating reflection  34  effects, two window glasses, 0.0103 m apart, were attached below the sleeve. The radiation distribution was measured at different locations with air blowing inside the sleeve (Figure 2.3). At each end, the UV lamps were fixed on two linear stages horizontally and two rails vertically. The sensor was located on a prism holder and the two linear stages on a rail. This mechanism allowed the lamp axis to be level with the ground and for adjustment of the horizontal position of the lamps. Also, the sensor direction (view angle) and position (x, y, z direction) could be adjusted with respect to the axis of the lamp. During all of the tests, the sensor faced the axis of the lamp vertically. With 10 degrees of freedom, the position of the sensor at different locations with respect to the position of the lamp was measured with a precision of 0.001 m. Figure 2.4 shows the arrangement for measuring the radiation field below the larger lamp. Thick arrows in Figure 2.4 show the direction of translation or rotation of the different components. 6  5 4 3 2  42 mm  1 10.3 mm  Figure 2.3. Position of reflector glasses under a UV lamp in a quartz sleeve. The apparatus consists of: two window glasses (1), a UV photodiode extension arm (2), a UV sensor (3), a UV lamp with 200 W input power (4), a quartz sleeve (5), and the position of the hole in the sleeve (6).  35  3  9  8 1  7  6 5 4  2  Figure 2.4. The supporting mechanism of the UV lamp and sensor consists of: a vertical rail (1), a horizontal rail (2), a linear stage for the lamp (3), a linear stage in z direction (4), a linear stage for the sensor (5), the prism holder (6), the UV sensor extension arm (7), the UV photodiode (8), and the UV lamp (9). 2.3.2 Results  The mid-size high-output UV lamp (50 W) was tested in stagnant air with no further changes, but the larger UV lamp (200 W) was tested under different operating conditions; in stagnant air (without the sleeve), in the sleeve with no blowing air, and in the sleeve with cooling air at 0.0013±0.0006 m3/s. The areas under the lamps were scanned. At each location of the sensor, 15 measurements were taken, and the average was reported. For all measurements, the standard deviations were less than 1% of the calculated averages for each point. In order to ensure the consistency of the UV lamps’ radiant energy output, some points were tested twice. For modeling, the following assumptions were made:  36  - The entire sensor geometry was modeled. Due to the low view from the angle of the sensor (47 degrees), no reflection from the surface of the filter was considered. The normal vector of the sensor surface was perpendicular to the axis of the lamp for all measurements; - The measurement performed at 1 mm (closest practical distance from the top of the sensor on the filter) from the body of the lamp (or body of sleeve) and the recorded data were employed to calculate the power density (px, px,α, and px,α,r) with the unit of radiometer output (arbitrary unit) at the source (boundary condition) for Equations 2.8, 2.9, and 2.11. Calculating the radiant energy for the boundary conditions (BCs), the BC values were assumed to be the same as those of the measurements (for the same x-location) multiply by a correction factor (constant). The correction factor was calculated simulating the measurement results at 1 mm. Applying the correction factor, the distribution of power density along the plasma arc length (x) was calculated using linear correlation for discrete points in which the measurement were performed used for predicting power density at different distances from the plasma arc. For the 2D and 3D models, it was assumed that the radial distribution of energy was constant (px,α= px, and px,α,r= px). These assumptions can be corrected if measurements are performed around the lamp with the proper setup and without changing the position of the lamp; - For refraction and reflection calculations, Fresnel, and Snell’s laws were used with no modification. In reality, rays are infinitely bouncing back and forth (reflection) in the medium between two different media, such as the quartz body of the UV  37  lamp, but the first reflection from the interface of the two media is the most important one. For all calculations, only the first reflection was considered for each ray, and the second and other reflections were assumed to be zero. For air, window glass, quartz, and mercury vapor, refraction indices of 1, 1.52, 1.52 [23], and 1 (assumption), as well as UV transmittances of 100, 0, 89.6, and 100% were known, respectively (UV transmittance of mercury vapor was assumed 100%). Overall, the comparison between the simulation and the measurements indicated that applying refraction terms improved the simulation results considerably and that reflection has no significant impact on the results. The following figures show the results for different conditions and different lamps. For all figures, the continuous lines represent simulation results, and they are marked from 1 mm to 50 mm to show the distance from the surface of the sensor filter (5.4 mm above the active area of the sensor) to the surface of the lamp or surface of the sleeve. In addition, the experimental values are shown using symbols to represent discrete point measurements. Figures 2.5 and 2.6 show the measured and simulated/calculated power values, with arbitrary numbers, for the larger lamp in the air, considering the 1D and 3D models, respectively, with refraction through the lamp body. Using the aforementioned procedure, the calculated values from the 2D model yielded higher values at farther distances comparing to closest one (1 mm far from the lamp surface). This discrepancy occurred because of the simplifying assumption for setting up the model. It was assumed that the backside of the lamp surface has no detected emission because of the shadowing effect of the front side (UV detector faced the front face). Correcting this assumption, shadowing effect of lamp should be measured and modeled (out of the scope of this work). Calculated power at 10 and 20 mm were  38  higher than those measured at 1 mm distance from the lamp surface. If the calibration factor (was not measured) is multiplied by the power value, the real value of radiant  Measured/Calculated Power (a.u.)  irradiance rate will be obtained in units of W/m2.  3.0E-06 1mm  2.5E-06 2.0E-06  10mm  1.5E-06  20mm 30mm  1.0E-06  50mm  5.0E-07 0.0E+00 -5  5  15  25  35  45  55  Horizontal distance of sensor from tip of the lamp filament (cm)  Figure 2.5. Experimental values for the power of the larger lamp (symbols) in air vs.  Measured/Calculated Power (a.u.)  calculated values (continuous lines) using the 1D model.  3.0E-06 1mm  2.5E-06 2.0E-06  10mm  1.5E-06  20mm 30mm  1.0E-06  50mm  5.0E-07 0.0E+00 -5  5  15  25  35  45  55  Horizontal distance of sensor from tip of the lamp filament (cm)  Figure 2.6. Experimental values for the power of large lamp (symbols) in air vs. calculated values (continuous lines) using the 3D model.  39  If the measured values at 1 mm are taken as the base for calculating the intensity at the source (I0 in Equation 2.7) and, consequently, the radiant energy at further distances from the lamp, it seems that the 1D model predicted values lower than those measured, while the 3D model predicted values higher than measured. It is possible that, when considering the UV lamp as a 3D source of emission, not all elements in the plasma of the lamp are emitting equally. Several studies were performed considering different distributions of the radiation from the plasma; e.g., a Gaussian distribution with different widths, and a uniform volumetric distribution with a smaller diameter than the inner diameter of the lamp. For this specific lamp (larger lamp), considering the plasma as a cylinder with a diameter of 7 mm having a uniform volumetric distribution yielded the  Measured/Calculated Power (a.u.)  best results (Figure 2.7).  3.0E-06 1mm  2.5E-06 2.0E-06  10mm  1.5E-06  20mm 30mm  1.0E-06  50mm  5.0E-07 0.0E+00 -5  5  15  25  35  45  55  Horizontal distance of sensor from tip of the lamp filament (cm) Figure 2.7. Experimental values for the power of the larger lamp (symbols) in air vs. calculated values (continuous lines) using a 3D model, assuming a cylindrical emitter with a 7 mm diameter.  In a UV reactor, UV lamps typically operate inside a quartz lamp sleeve. Therefore, the radiation distribution of a UV lamp with a sleeve was studied  40  experimentally and numerically at different conditions. The setup was tested under two different operating conditions: with no air flowing in the gap between the lamp and the sleeve and with an airflow rate of 0.001 m3/sec at 22±1ºC. The air blowing into the sleeve maintains an ambient temperature at the sleeve surface in a UV reactor, which is a crucial parameter for studying isothermal photoreaction rates. While maintaining an ambient temperature on the sleeve surface, the airflow changes the UV lamp operating temperature and consequently changes the performance of the plasma inside the lamp. This causes a non-uniform radiation distribution along the lamp length. The results are presented for the 3D model (dashed line), assuming 7 mm cylindrical emitter, and the 1D model (continuous line) in Figure 2.8 without air flowing and, in Figure 2.9, with a cooling air flow The figures clearly demonstrate the effect of a cooling medium on the radiation distribution. At both operating conditions, the 1D model better predicts the  Measured/Calculated Power (a.u.)  distribution of radiant energy. Air flow direction 2.5E-06 1mm  2.0E-06  10mm  1.5E-06  20mm  1.0E-06  30mm 50mm  5.0E-07 0.0E+00 -5  5  15  25  35  45  55  Horizontal distance of sensor from tip of the lamp filament (cm) Figure 2.8. Experimental values for the radiant power of the larger lamp (symbols) in a quartz sleeve with no air flowing inside the sleeve vs. the 1D model-predicted values (continuous lines) and the 3D model, cylindrical emitter with 7 mm diameter, values (dashed lines).  41  Measured/Calculated Power (a.u.)  1.5E-06  1mm  10mm  1.0E-06  20mm 30mm  5.0E-07  50mm  0.0E+00 -5  5  15  25  35  45  Horizontal distance of sensor from tip of the lamp filament (cm) Figure 2.9. Experimental values for the radiant power of the larger lamp (symbols) in a quartz sleeve with air flowing inside the sleeve (0.001 m3/s) vs. the 1D modelpredicted values (continuous lines) and the 3D model, cylindrical emitter with 7 mm diameter, values (dashed line). Considering real operation, reflections from other objects (e.g., other UV lamps or the body of the UV reactor, etc.) affect the total output radiant power at each location inside the UV reactor domain. To test this phenomenon, the UV lamp in its sleeve was placed on two parallel window glasses (Figure 2.3). Figure 2.10 shows that both models (1D and 3D) predicts the results accurately for areas close to the lamp, but deviation is observed far from the lamp because of greater reflection. It may be concluded that window glasses have a greater reflection than predicted by Fresnel’s law using the refractive index of window glass. Likely, the micro-layer at the surface of window glass has a different refractive index in comparison to the bulk medium. The optical properties of the surface highly depend on the crystal structure of the silica molecules. Direct measurement of these parameters is expected to improve the modeling results.  42  Measured/Calculated Power (a.u.)  1.5E-06 1mm  1.0E-06 10mm 20mm 30mm  5.0E-07  50mm  0.0E+00 -5  5  15  25  35  45  Horizontal distance of sensor from tip of the lamp filament (cm) Figure 2.10. Effect of reflection on radiation distribution of the larger UV lamp in a quartz sleeve with air flowing inside the sleeve (1 L/s). Experimental value for the radiation power (symbols) vs. the 1D model-predicted values (continuous lines) and the 3D model, cylindrical emitter with 7 mm diameter, values (dashed lines).  Passing air over a UV lamp reduces the skin temperature of UV lamp and, hence, the plasma temperature (mercury vapor inside of the low pressure lamp) is reduced consequently. The Optical transmittance and excitation/emission characteristic of mercury vapor is highly affected by the operating temperature, which impact absorption and generation of UV radiation, respectively. As a result, UV lamp has lower UV emission at lower temperature. Finally, the high-output medium size UV lamp (50 W) was studied and three modeling assumptions (1D, 2D, and 3D) were tested in comparison to the experimental results. The 2D model poorly predicted the irradiance rate distribution, likely, because the back side of the UV lamp was not considered in the model (shadowing effect) and surface was assumed as uniform emitter instead of non-uniform, e.g., Lambertian (cosine emitter). Taking the measured values at 1.5 mm as the basis for the model boundary  43  conditions, the 1D model calculated radiance value lower than those measured, while the 3D model calculated higher than measured values (Figure 2.11). This was in agreement with the trend observed for the larger lamp (Figures 2.5 and 2.6). The deviation was corrected by considering a portion of the volume of plasma as a diffusive radiator. In this case, a coaxial cylinder with a radius of 3.25 mm yielded the most accurate prediction (Figure 2.12). These results clearly suggest that the spatially uniform assumption is invalid as the boundary condition. The first curve (1.5 mm from the surface of the lamp) with a diamond symbol (  ) shows the non-uniform distribution of output power along  the axis of the lamp. Irradiance rate contours can likely be considered horizontal far from  Measured/Calculated Power (a.u.)  the lamp (i.e., a uniform distribution of radiant power), but not close to the lamp.  1.7E-06  1.5mm  1.2E-06 12mm  7.0E-07 39mm  2.0E-07 -5  5  15  25  35  45  Horizontal distance of the sensor from the tip of the lamp filament (cm) Figure 2.11. Irradiance rare distribution around a high-output UV lamp in stagnant ambient air; measurements (symbols), 1D modeling predictions (continuous lines), and 3D modeling, cylindrical emitter with 3.25 mm diameter, predictions (dashed lines).  44  Measured/Calculated Power (a.u.)  1.7E-06  1.5mm  1.2E-06 12mm  7.0E-07 39mm  2.0E-07 -5  5  15  25  35  45  Horizontal distance of sensor from tip of the lamp filament (cm) Figure 2.12. Simulation results for partial radiation inside the plasma of a high-output UV lamp in the air; measurements (symbols) and partial 3D modeling, cylindrical emitter with 3.25 mm diameter, predictions (continuous lines).  2.4 Conclusions This study shows that the modeling of radiant power distribution depends highly on the boundary conditions (BC), especially near the surface of a lamp. The power density BC is a function of several parameters related to manufacturing, operation, and aging. Assuming a uniform distribution of radiant energy over the entire length of a UV lamp may not represent, its true characteristics, especially in the region close to the surface of the lamp. The BC values for power density should be measured experimentally for a specific lamp at specific operating conditions before they are integrated into a radiation model. This approach can improve the modeling considerably and produce more reliable results, especially near the lamp surface where more vigorous photoreaction occurs.  45  It seems that there is no universal method for modeling all UV lamp powers at the radiation source (lamp body) using a 1D, 2D, or 3D model. Accurate results can be obtained from the 3D model by assuming that only a portion of the core of the plasma behaves as a diffusive emitter. The size of the core (diameter) can be estimated by fitting model prediction to power measurements taken at different distances from the surface of the lamp. However, for engineering applications with minimal computation resources, a 1D source seems to predict the radiant power distribution with reasonable accuracy. For modeling the radiant field inside a UV reactor, reflections from the bodies of the other lamps and the reactor itself can have an impact on radiation distribution and should be taken into consideration in order to improve the simulation. However, reflection from the interface of two media with two different refractive indices (Fresnel’s Law) (i.e., from the plasma into the lamp quartz body) has little impact on the modeling results. The most important phenomenon that should be considered in the modeling of UV radiation from a lamp is the refraction effect (Snell’s Law). Integration of Snell’s Law into the model was found to improve the results significantly.  46  2.5 References 1- Beltran, J. A., Barbosa-Cánovas, G. V., 2004. Advantages and Limitations on  Processing Foods by UV Light. Food Sci. Tech. Int., 10, 137-147. 2- Mahmoud, N. S., Ghaly, A. E., 2004. On-Line Sterlization of Cheese Whey Using  Ultraviolet Radiation. Biotechnol. Prog. 20, 550-560. 3- Kucuk, S, Arastoopour, H, Koutchma, T., 2003. Modeling of UV Dose Distribution in  a Thin-Film UV Reactor for Processing of Apple Cider. J. Food Eng., 65,125-136. 4- Noceti, R. P., Taylor, C. E., D'Este, J. R., 1997. Photocatalytic Conversion of  Methane. Catalysis Today, 33,199-204. 5- Taylor, C. E., Noceti, R. P., 2000. New Developments in the Photocatalytic  Conversion of Methane to Methanol, Catalysis Today, 55, 259-267. 6- Gondal, M. A., Hameed, A., Suwaiyan, A., 2003. Photo-Catalytic Conversion of  Methane into Methanol Using Visible Laser. Applied Catalysis A:General, 243,165174. 7- Gondal, M. A., Hameed, A., Yamani, Z. H., Arfaj, A., 2004. Photocatalytic  Transformation of Methane into Methanol Under UV Laser Irradiation over WO3, TiO2 and NiO Catalysts. Chemical Physical Letters, 392, 377. 8- Raithby, G. D., 1999. Evaluation of Discretization Errors in Finite-Volume Radiant  Heat Transfer Predication. Num. Heat Transfer, 36, 241-264. 9- Chui, E. H., Raithby, G. D., 1993. Computation of Radiant Heat Transfer on a  Nonorthogonal Mesh Using the Finite-Volume Method. Num. Heat Transfer B., 23, 269-288. 10- Cassano, A. E., Martin, C. A., Brandi, R. J., Alfano, O. M., 1995. Photoreactor  Analysis and Design: Fundamentals and Application. Ind. Eng. Chem. Res., 34, 21552201.  47  11- Romero, R. L., Alfano, O. M., Cassano, A. E., 1997. Cylindrical Photocatalytic  Reactors. Radiation Absorption and Scattering Effects Produced by Suspended Fine Particles in an Annular Space. Ind. Eng. Chem. Res., 36, 3094-3109. 12- Pareek, V. K., Cox, S., Adesina, A. A., 2003. Light Intensity Distribution in  Photocatalytic  Reactors  Using  Finite  Volume  Method.  Melbourne:  Third  International Conference on CFD in the Minerals and Process Industries. 229-234. 13- Bolton, J. R., 1999. Ultraviolet Application Handbook (1st edition). Bolton  Photoscience, Inc. 14- Bolton, J. R., 2000. Calculation of Ultraviolet Fluence Rate Distributions in an  Annular Reactor: Significance of Refraction and Reflection. Wat. Res., 34, 33153324. 15- Jacob, S. M., Dranoff, J. S., 1969. Light Intensity Profiles in an Elliptical  Photoreactor. AIChE J. 15, 141-144. 16- Jacob, S. M., Dranoff, J. S., 1970. Light Intensity Profiles in a Perfectly Mixed  Photoreactor. AIChE J., 16, 359-363. 17- Jacob, S. M., Dranoff, J. S., 1969. Design and Analysis of Perfectly Mixed  Photochemical Reactors. Radial Scale-Up of perfectly mixed photochemical Reactors. Chem. Eng. Prog. Symp. Ser., 64, 54-63. 18- Jacob, S. M., Dranoff, J. S., 1966. Radial Scale-Up of Prefectly Mixed Photochemical  Reactors. Chem. Eng. Prog. Symp. Ser., 62, 47-55. 19- Williams, J. A., 1978. The Radial Light Intensity Profile in Cylindrical Photoreactors.  AIChE J., 24, 335-337 20- Zolner, W. J., Williams, J. A., 1971. Three Dimensional Light Intensity Distribution  Model for an Elliptical Photoreactor. AIChE J., 17, 502-503.  48  21- Irazoqui, H. A., Cera, J., Cassano, A. E., 1976. The Radiation Field for the Point and  Line Source Approximation and the Three-Dimensional Source Models: Application to Photoreactors. Chem. Eng. J., 11,27-37. 22- Irazoqui, H. A., Cera, J., Cassano, A. E., 1973. Radiation Profiles in an Empty  Annular Photoreactor with a Source of Finite Spatial Dimensions. AIChE J., 19, 460467. 23- Wikipedia, Visited on January 2009  http//en.wikipedia.org/wiki/List_of_indices_of_refraction  49  Chapter 3. Simulation of UV Photoreactor for Water Disinfection in Eulerian Framework 3.1 Introduction UV technology is a reliable and cost-effective solution to ensure the safety of drinking water. Demand for this technology as one of the best candidates for replacing conventional chlorine disinfection units is growing quickly. Several companies around the world are developing different types of UV photoreactors for water disinfection units. In parallel, researchers are attempting to find the best techniques for the simulation of UV photoreactors. Darby et al. [1] derived a mathematical model for inactivation rate based on experimental work. However, their work did not take the hydrodynamics of the reactor into consideration. Chiu et al. [2] presented an integrated method which took into account the reactor hydrodynamics. In that model, the velocity field was measured by laser Doppler velocimetry, making the modeling very time-consuming to perform. In order to derive an appropriate model for UV photoreactors, the radiation field and chemical reaction or microorganism inactivation rate should be integrated with the reactor velocity field. During the last decade, remarkable developments in computational resources have enabled accurate predictions of the velocity field for modeling chemical reactors by computational fluid dynamics (CFD). Bass [3], Do-Quang et al. [4], Lawryshyn and Lu [5] , and Lyn et al. [6] have presented models of UV photoreactor hydrodynamics based on CFD. Two main approaches can be taken to simulate UV reactors by CFD: the Eulerian–Lagrangian and the Eulerian frameworks. In the Eulerian–Lagrangian approach, microorganisms are considered as a dispersed phase, and the motion of dispersed phase A version of this chapter has been published. Elyasi, S., Taghipour, F., 2005. Simulation of UV Photoreactor for Water Disinfection in Eulerian Framework. 50 Chemical Engineering Science, 61, 14, 4741-4749.  particles are explicitly simulated by solving the equations of motion for each particle in the Lagrangian framework. The inactivation rate is calculated based on the radiation field (fluence rate) and the residence time for each interval on the particle tracks (trajectories). Inactivation of microorganisms is separately calculated on each trajectory and an ensemble average is reported as the total inactivation. Unluturk et al. [7], Lawryshyn and Cairns [8] , and Wright and Lawryshyn [9] used the Eulerian–Lagrangian approaches to simulate UV photoreactors for microbial water disinfection. In the Eulerian approach for a single flow regime, all the species components are considered as part of the continuous phase, and the source term in the species mass conservation equation is accounted for a volumetric reaction rate for the reactant on product species. The Eulerian method has been employed by a number of researchers for simulating chemical reactors. Kamimura et al. [10] used this method to simulate an ozone/UV photoreactor for removing organic contaminants. To the authors’ knowledge, the Eulerian method has not been applied to modeling photoreactors for microorganism inactivation, likely due to the lack of a defining equation for the volumetric inactivation rate  of  microorganisms.  Using  the  volumetric  inactivation  rate  definition,  microorganisms can be treated as chemical species where their local concentration in each control volume (CV) in the medium is a function of their fluxes through the CV and the rate of disinfection of microorganisms in the CV. The main focus of this method is to solve the microorganism conservation equation of the microorganisms for each CV throughout the entire medium instead of considering their trajectory in the medium, which is the approach of the Lagrangian method.  51  The Eulerian–Lagrangian method can cover all the domains inside the reactor, if the trajectories of enough particles are taken into account. In contrast, the Eulerian approach accounts for the entire reactor domain in the calculation, regardless of the particle (microorganism) concentration, by solving the mass conservation equation of the microorganisms in the reactor domain. In this research, the Eulerian approach was applied to model UV disinfection reactors by deriving the volumetric inactivation rate as a time-independent function in a steady-state simulation. Microorganisms were treated as a reactive component without molecular diffusion. The volumetric rate of inactivation was derived as a source term in the mass conservation equation for each species (e.g. active or inactivated microorganisms). A general method was developed to derive the time-independent volumetric inactivation rate function using bioassay data from the UV fluence–response curve. The model for UV photoreactor performance simulation was evaluated against experimental biodosimetry results from industrial UV photoreactors.  3.2 Modeling Procedure 3.2.1 Fluid Flow Modeling  Application of the laws of conservation of mass and momentum yields a basic set of equations governing the motion of the fluid, which are used to calculate velocity and pressure fields. In the steady state conditions with no source term, the equations for conservation of mass and momentum are expressed as ∇⋅v = 0  3.1  ∇ ⋅ (ρvv ) = ∇p − ∇ ⋅ τ + ρg + F  3.2  52  where ρ, v, p, τ, g, and F are density of the medium, velocity vector, pressure, viscous stress tensor, gravitational acceleration, and external body force vector, respectively. Using the classical approach for solving Equation 3.2, a turbulence model such as the realizable k–ε model can be selected for turbulence modeling [11]. The relaxation time of microorganisms (10-7 s) with 1 µm diameter and 1030 kg/m3 density in water is much less than the Kolmogrov time (>10-4 ) in the inactivation zone to UV photoreactors. Therefore, the particles can be considered as having the same velocity as that of the fluid [12] and the mass conservation equation of species i at steady state takes the following general form [13]: ∇ ⋅ (ρxi v ) = −∇. jeff ,i + Si  3.3  where xi, Si, and jeff,i are mass fraction, the volumetric rate of production or consumption of species i in the medium, and the effective diffusion flux for component i, respectively. For a UV disinfection reactor, the volumetric rate of consumption is defined as the volumetric inactivation rate of microorganisms that is turbulence independent. Considering the effect of turbulent dispersion as a function of turbulent Schmidt number, the effective diffusion flux of species i is often defined as: ν  j eff ,i =  T + Dm ,i ∇(ρx i ) 3.4  σ T ,i  where νT, σT,i, and Dm,i are the turbulent momentum diffusivity, species turbulent Schmidt number, and molecular diffusion coefficient for component I, respectively. Since the size of microorganism is typically much larger than that of a water molecule, the molecular diffusion rate of microorganisms in the medium is negligible. The effect of turbulent dispersion was considered by selecting the species (microorganism) turbulent Schmidt number of 0.7. This value for the turbulent Schmidt number was quoted by Baldyga and  53  Bourne [14] for a round jet (inlet to a reactor). With negligible molecular diffusion at steady state conditions, Equation 3.3 can be simplified to ∇ ⋅ ( ρxi v ) +  νT 0.7  ∇ 2 (ρxi ) = Si  3.5  For the simulation of a UV photoreactor, since the inactivation of microorganisms does not change the physical properties of the medium, the continuity equation (Equation 3.1) and the momentum conservation equation (Equation 3.2) can be solved together without considering the mass conservation equation of any species. As a result, after obtaining the velocity field, the species mass conservation equation can be solved separately. This method enables changes in the radiation field and correspondingly recalculations of the microorganism concentration without resolving the momentum conservation equation for each trial, thereby reducing the computation time dramatically. 3.2.2 Volumetric Inactivation Rate Modeling  The germicidal effects are directly related to the total fluence (dose) of UV radiant energy absorbed by a microorganism. The total UV fluence is the integral of the local UV fluence rate received by the microorganisms during the exposure time t0≤t≤tf, i.e., tf  H = ∫ G (s, λ )dt t0  3.6  where H, G(s,λ), and t are total absorbed fluence, local fluence rate for specific wavelength λ whose location is addressed by the position vector s, and the time of exposure from initial t0 to final time tf, respectively. The rate of inactivation by UV radiation for some microorganisms can be approximated as first-order [15]. Noakesa et al. [16] showed that first-order kinetics is an  54  appropriate mathematical model for simulating the inactivation rate of airborne microorganisms. However, for many waterborne microorganisms, this model may not be applicable and other models may be employed as presented by other researchers [17]. In general, the mathematical model for inactivation rate can be derived using any type of correlation, such as a polynomial curve fit, which is a general method for representing experimental inactivation data. This method reveals that the microorganism concentration is a log function of total absorbed UV fluence, based on the fluence–response curve of bioassay data. The general form of the log reduction equation can be written as log  N0 C = log 0 = f (H ) N C  3.7  where N0, N, C0, C, and f are the initial and instantaneous number of microorganisms per unit volume, the initial and instantaneous concentration of microorganisms, and a general function of total absorbed fluence (H) that may be a polynomial. The rate of inactivation based on concentration can be derived by differentiating Equation 3.7 with respect to time to obtain  1 dC df ( H ) df ( H ) dH =− =− ln(10)C dt dt dH dt  3.8  If the total fluence (Equation 3.6) is substituted in Equation 3.8, a general formula for the volumetric reaction rate of microorganisms as a function of total absorbed fluence can be derived as dC df ( H ) = − ln(10)G ( s, λ )C dt dH  3.9  where the fluence (H), could be expressed as the inverse function (g) of the concentration ratio (from bioassay data), based on Equation 3.7, i.e.,  55  C   H = g  log 0  C    3.10  If the total absorbed fluence from Equation 3.10 is substituted into Equation 3.9, the volumetric rate of reaction will only be a function of local microorganism concentration and fluence rate. In order to show the application of the presented method, one example is demonstrated below using MS2 as a model microorganism. Figure 3.1 shows the bioassay data of MS2 inactivation by UV irradiance. The fluence–response curve obtained for MS2 in this figure can be represented by the polynomial equation: N  C log 0  = log 0  N  C 3.11    = 9.0 × 10 −10 H 3 − 3.0 × 10 −6 H 2 + 6.2 × 10 −3 H   Accordingly, the total absorbed fluence, which is the inverse function of Equation 3.11 (using polynomial curve fitting), can be expressed as C  C  H = 22.3 log 2  0  + 149.6 log 0  C  C   3.12  56  Figure 3.1. Log reduction of MS2 at various UV fluences (dose–response curve). This equation can be employed to reveal the total received fluence through the UV reactor for MS2. The volumetric inactivation rate for this specific microorganism using Equations 3.9 and 3.11 is expressed by dC = − ln(10) G ( s ) C 27 × 10 −10 H 2 − 6 × 10 −6 H + 6.2 × 10 −3 3.13 dt where H can be substituted by Equation 3.12. By this method, the volumetric inactivation  (  )  rate is only a function of concentration and local fluence rate. This inactivation rate can be directly used as the source term in Equation 3.5. 3.2.3 Radiation Modeling  A well-designed UV system optimizes the reactor configuration and lamp spacing to ensure the effective delivery of UV radiant energy to the medium over the entire domain. Therefore, it is crucial to accurately model the UV radiation distribution inside photoreactors. A radiation model based on the emission of UV radiation from the surface of a cylindrical lamp [18] is used in order to simulate radiation field in the reactors. In this model, the UV fluence rate at each part of the reactor, G(s), is calculated from  57  2 exp − k ave (x − x p ) − 2rlamp sin (θ ) y 2p + z 2p + rlamp + y 2p + z 2p  2 (x − x p )2 − 2rlamp sin (θ ) y 2p + z 2p + rlamp + y 2p + z 2p 2  P G (s) = 2 4π llamp  xf  π −α max  ∫ ∫α x0  max    dθ dx  3.14  α max   r lamp = arcsin  2  y p + z 2p        3.15  where s, P, kave, xp, yp, and zp are the position vector, the total effective (e.g. germicidal) radiant power of the lamp, the average absorption coefficient of the medium, and the coordinates of the investigation point in the medium, respectively. x0, xf, rlamp, and llamp are the initial and final x coordinates (axial) of the lamp ends, outside diameter, and length of the lamp (llamp = xf - x0), respectively. 3.2.4 Efficiency of a UV Photoreactor  The performance of a UV photoreactor is influenced by the radiation distribution and velocity profile. The reduction of live microorganisms is a suitable indicator for comparison of different reactor configurations. In order to determine the performance deviation of a real reactor from an ideal one, the maximum mathematical log reduction of microorganisms should be derived. The maximum inactivation happens in a UV photoreactor if all the microorganisms in all cross-sections of the reactor receive an equal UV fluence. This occurs either if there is no radial mixing, but the axial velocity is proportional to the UV fluence rate at each reactor cross-section (resulting in the same total fluence for each line stream), or if the flow is an ideal plug flow with perfect radial mixing (resulting in a uniform concentration profile of microorganism at each crosssection). Both scenarios result in a uniform species concentration at the reactor outlet.  58  Considering the first scenario, the instantaneous fluence, dH, that is constant across the normal cross-sectional plane, can be derived for a small length interval, dx, in the direction of the line stream as dH = G ( s )  dx v  3.16  Since the received fluence is to be constant for all streamlines, the velocity profile on the normal cross-sectional plane as a function of fluence rate is given as v=  G( s) K  3.17  where K represents local absorbed fluence by microorganisms per unit length of the photoreactor (or received local radiant energy by microorganism per unit volume) and is constant for an ideal reactor. Equation 3.17 is the velocity profile required for the case of maximum efficiency with no radial mixing. This equation can be used to determine the velocity profile required to attain the maximum UV photoreactor performance. The integration of velocity over a cross-sectional plane is equal to the volumetric flow rate: Q = ∫ vdA = ∫  G( s) 1 dA = ∫ G ( s)dA K K  3.18  where Q and dA are the volumetric flow rate (known) and cross-sectional area element, respectively. Theoretically, the total maximum absorbed fluence (Hmax) for an ideal UV photoreactor can be obtained by integrating Equation 3.16 over the reactor length. If the parameters v and K are substituted from Equations 3.17 and 3.18, respectively, the maximum theoretical fluence delivered by a photoreactor can be expressed as  59  H max =   1 Lphoto  dx = Gvol V G ( s ) dA ∫ ∫ 0   Q Q    3.19  where Lphoto, V, and Gvol are the length and volume of the photo-reactive region of the reactor, and the volume-weighted average fluence rate (calculated based on the radiation model), respectively. The efficiency of the UV photoreactor, ηcon, can be defined by dividing the actual achievable log reduction (or delivered fluence) by the maximum theoretical log reduction (or maximum fluence):  η con   N   log  N 0  real =  N   log  N 0  max  or η Dose =  3.20  H real H max  3.21  For a specific UV photoreactor, the maximum fluence (Hmax) can be calculated for the given geometry, operating conditions, and medium specification to compare the results with the fluence calculated from experimental data (Hreal) to obtain the photoreactor efficiency (ηDose). To calculate the maximum fluence (Equation 3.19) based on the surface emission radiation model (Equation 3.14), the fluence rate should be integrated over the entire reactor domain (for calculating Gvol) as:  H max  xf P = ∫ 2π llampQ x 0  ∫  rreactor rlamp    xf  ∫x  0   π −α max  ∫α  max  2 2  exp − kave (x − x p ) − 2rlamp r sin (θ ) + rlamp + r 2    rdθdx drdx  p 2 (x − x p )2 − 2rlampr sin (θ ) + rlamp + r2    3.22 where x0 and xf (llamp=xf-x0) are the initial and final x-coordinates of UV lamp ends.  60  3.2.5 Computational Strategy for an Integrated UV Photoreactor Performance  In this work, the commercial CFD software Fluent (6.2.16) was used to solve the governing equations of mass, momentum, and species mass conservation. In addition, two external macros were developed and integrated into the Fluent software in order to calculate the source terms of the species mass conservation equations (Equation 3.5) and the radiation field (Equation 3.14). The following general procedure explains the steps taken to develop the CFD model of a UV photoreactor for microorganism inactivation. - The computational domain of the reactor is discretized into a finite number of cells. The number of cells is selected such that solution of the governing equations becomes grid independent; - The fluence rate is calculated for each cell center using Equation 3.14 and is kept in a corresponding address in the memory as user-defined memory for the next steps; - The mass and momentum conservation equations are solved as a segregated system of equations with implicit formulation to determine the flow field in the reactor; - After obtaining the velocity field, the mass conservation equation of live microorganisms (species mass conservation equation) is solved separately based on the calculated velocity field. The volumetric reaction rate from Equation 3.9 including the UV fluence rate from Equation 3.14 (stored from step 2) is used as the source term. The density of the species is assumed equal to the density of water. Otherwise the physical properties of the medium change and the calculated velocity field is not applicable. 61  The ratio of microorganism concentration at the reactor outlet to the inlet yields the performance of the system. The ratio of local concentration to inlet concentration can also be employed in Equation 3.10 to obtain the total absorbed fluence by the microorganisms throughout the UV photoreactor.  3.3 CFD Model Setup The performance of two industrial UV photoreactors was simulated using the aforementioned computational strategy. The UV photoreactors were modeled by discretizing the physical domain into approximately 960,000 structured hexahedral cells. A statistical analysis showed that increasing the number of cells to 1,660,000 has little impact on the total velocity field. The coefficient of determination (or R-square) of total velocity for 2,000,000 points in the computational domain calculated with 960,000 and 1,660,000 cells was 0.97, implying that the smaller number of cells (960,000) was adequate to achieve a mesh-independent solution. Water physical properties were considered to specify the coefficients of the governing equations. A no-slip boundary condition with enhanced wall function was selected for all internal surfaces. Considering  the  requirements  for  conservativeness,  boundedness,  and  transportiveness (the directionality of influence), the QUICK differencing scheme was selected [19]. The STANDARD and SIMPLEC schemes were selected for pressure and pressure–velocity coupling, respectively. The relative errors between two successive iterations were less than 10-5 for each scaled residual.  3.4 Experimental Procedure To evaluate the models of UV reactor performance, the MS2 phage was used as a model microorganism to determine the fluence delivered by two industrial UV-  62  photoreactors under various operating conditions. The results were compared with those predicted by the CFD model. 3.4.1 MS2 Fluence–Response and Bioassay  MS2 was prepared in accordance with USEPA Method 1602. A quasi-collimated beam apparatus [15] was used to determine the UV fluence–response curve for MS2 in the feed water (Figure 3.1). A MS2 bioassay was conducted at the photoreactor’s inlet and outlet in order to determine their inactivation under various conditions. 3.4.2 UV Photoreactor Bioassay  The experimental study of UV photoreactor performance using MS2 inactivation was conducted using two industrial UV photoreactors manufactured by R-Can Environmental Inc.; model S12Q with 0.94 m length, 0.089 m diameter, and 0.022 m inlet/outlet port size, and S8Q with 0.9 m length, 0.065 m diameter, and 0.022 m inlet/outlet port size. The reactors S12Q and S8Q included UVC lamps of 0.0225 m diameter with output powers of 39 and 37 W, respectively, with 35% germicidal efficiency. The S12Q and S8Q UV reactors are specifically designed to treat microbiologically contaminated ground or surface water. The mass flow rate of water through the reactors was kept constant at 0.505 kg/s for the S8Q and 0.757 kg/s for the S12Q UV photoreactor. However, different water qualities (different UV transmittance) were used to study various disinfection conditions. 3.4.3 UV Photoreactor Performance  Both models of UV photoreactors (S12Q and S8Q) were simulated by the computational strategy explained in the modeling section. Figure 3.1 shows the laboratory data of MS2 response to UV exposure in drinking water. Equation 3.11 was  63  derived using a third-order polynomial equation fitted through all the data presented in Figure 3.1. This equation shows the concentration change of MS2 as a function of total received UV fluence. The inverse function of Equation 3.11 was derived, as presented by Equation 3.12, which reveals the received UV fluence as a function of MS2 concentration. Equations 3.11 and 3.12 were applied to derive the source term (Equation 3.13) in the mass conservation equation of MS2. The transmittance of the UV lamp sleeve usually varies from 90% to 100%. In order to take this into consideration and to show the sensitivity of the results to the lamp sleeve efficiency, the UV photoreactor performance was modeled by changing the UV transmission of the sleeve from 90% to 100%. The simulations were conducted at the given flow rate with the conditions described in the experimental section. The predicted log reductions of MS2 as an indication of the performance of the S12Q and S8Q reactors are compared to the experimental results in Table 3.1. The simulation results include the variation of MS2 log reduction for different UV transmissions of the lamp sleeve at the given operating conditions (flow rate and medium transmission). If the uncertainty in the measurement of the total MS2 count, which is the characteristic of bioassay, is taken into account, the simulation predictions are in good agreement with the experimental results. This indicates the reliability of the proposed Eulerian method for the simulation and design of UV photoreactor for microbial inactivation.  64  Table 3.1. Performance evaluation of the two industrial UV photoreactors from biodosimetry experiments and simulation studies. Log reductions were calculated based on outlet/inlet concentrations of MS2 in photoreactors. For simulation results, maximum and minimum values correspond to the sleeve efficiencies of 100% and 90%, respectively. For experimental results, the values represent deviation with 95% confidence intervals of the average log reduction Different Model Medium transmission Log reduction Simulation results Log reduction Experimental results  Model S8Q  Model S12Q  85%  98%  85%  98%  2.45±0.13  3.44±0.17  1.97±0.08  3.30±0.16  2.14±0.12  3.46±0.36  1.91±0.42  3.50±0.50  Figure 3.3 shows the received UV fluence contours of longitudinal cross sections of S12Q and S8Q reactors for a 98% transparent medium. The contours of received fluence are asymmetrical and non-uniform, indicating potential for reactor design improvement. The theoretical efficiencies of the S8Q and S12Q reactors for MS2 inactivation, calculated by Equation 3.20 (or Equation 3.21 for fluence), based on the possible maximum fluence calculated by Equation 3.19 are 84% (78% for fluence) and 72% (63% for fluence), respectively. The efficiencies of less than 100% indicate the potential for improving the performance of these photoreactor systems. Although the calculated irradiance (Section 2.3.2) using 2D model (considering the explained procedure in Chapter 2) showed higher values vs. measurements, the predicted values of UV photoreactor performance (CFD model) is in good agreement with bioassay results. It is likely because for model set up, the efficiency of the lamp was assumed 30% (a generally accepted value for 2D radiation model). The UV output  65  efficiency might be higher than 30%, and, hence, considering this low value for lamp output is compensated using 2D model (2D predicted higher values).  Figure 3.2. Profiles of total received fluence (dose) inside S12Q and S8Q UV photoreactors. 3.4.4 UV Photoreactor Design Optimization  The model for simulating the UV photoreactor performance in the Eulerian framework can calculate UV received fluence profile in the domain of any UV photoreactor. This capability provides valuable information for improving the design of UV photoreactors to achieve better performance. For any given operating conditions, the best performance for a UV microbial disinfection system will be attained if there is a uniform received fluence (dose) profile in the entire domain of the reactor. As shown in Figure 3.2, the received UV fluence profile  66  for the S12Q reactor is considerably higher above the lamp because the velocity is lower in this region. If the received fluence profile becomes more uniform in the upper and lower parts of the photoreactor, the performance of the system will be considerably improved. Taking this concept into consideration, two new configurations for S12Q reactor were simulated by changing the position of the inlet (type A, Figure 3.3), and modifying the shape of the lamp sleeve tip (type B, Figure 3.3). The log reduction of MS2 increased from 3.49 to 4.33 and 4.72 for the reactors type A and B, respectively. For a 98%/cm transparent medium (kave=2 m-1), the theoretical efficiencies of the type A and B reactors based on Equations 3.21 and 3.22 are 90% and 98%, respectively. These efficiencies will be reduced with decreasing UV transmittance (increasing kave) of the medium. The side inlet configuration provides a proper velocity profile around the UV lamp. The velocity profile closely follows the value of the UV fluence rate (higher velocity at higher fluence rate areas) according to Equation 3.17, resulting in a more uniform delivered fluence (a function of fluence rate and velocity). Therefore, the performance of the UV photoreactors (in particular type B) is considerably improved. If Equation 3.17 is taken into account for producing appropriate velocity distribution during the reactor design, UV reactors could be designed with efficiencies close to those of ideal photoreactors.  67  Figure 3.3. Profiles of received fluence (dose) in cross sections inside side inlet port UV photoreactors. Type A and B are flat-end sleeve and hemispherical-end sleeve, respectively.  3.5 Conclusion A general procedure for UV photoreactor simulation in the Eulerian framework was introduced in order to solve the governing equations of flow motion, radiation distribution, and volumetric microorganism inactivation rate. The agreement of the simulation results with the experimental bioassay data for the two industrial UV reactors verified the model reliability. Using the proposed uncoupled concept, the reactor velocity field can be calculated independent of microorganism inactivation, if the reactor configuration and main physical specifications of the medium remain unchanged. As a result, a sensitivity analysis for studying the effect of fluence rate distribution and species concentration on the photoreactor performance can be performed separately without solving the velocity field for each study, resulting in a considerable reduction of computational time.  68  The Eulerian model has some advantages over the Lagrangian approach. For example, the received fluence (dose) profile and the local volumetric rate of inactivation, which are the appropriate indicators for improving the photoreactor design and determining the performance of UV photoreactors, can be demonstrated. The model can be applied to UV photoreactor simulation and optimization, resulting in a more efficient design of UV reactors.  69  3.6 References 1- Darby, J., Heath, M., Jacangelo, J., Loge, F., Swaim, P., and Tchobanoglous, G., 1995.  Comparison of UV Irradiation to Chlorination: Guidance for Achieving Optimal UV Performance. Water Environment Research Foundation.Alexandria, Virginia. 2- Chiu, K., Lyn D. A., Savoye, P., and Blatchley, E. R., 1999. Integrated UV  Disinfection Model Based on Particle Tracking. Journal of Environmental Engineering, ASCE, 125, 459-466. 3- Bass, M. M., 1996. Latest Advances in UV Disinfection Hydrodynamic Simulation  and Relation to Practical Experiences. Proceeding AQUATECH, Amsterdam. 4- Do-Quang, Z., Djebbar, R., Blatchley, E. R., and Lain, J. M., 1997. Computational  Fluid Dynamics Modeling of Ultra-Violet Disinfection Reactor Performance: Optimization of Flow in Vertical Lamp Open Channel. ASCE-CSCE Environmental Engineering Conference, Edmonton. 5- Lawryshyn, Y. A., and Lu, D., 1999. UV Reactor Design, It‘s More Than Putting a  Lamp in a Pipe. Journal of WCPM, 41, 106-109. 6- Lyn, D. A., Chiu, K., and Baltchley, E. R., 1999. Numerical Modeling of Flow and  Disinfection in UV Disinfection Channels. Journal of Environmental Engineering ASCE, 125, 17-23. 7- Unluturk, S. K., Arastoopour, H., and Koutchma, T., 2004. Modelling of UV Dose  Distribution in a Thin-film UV Reactor for Processing of Apple Cider. Journal of Food Engineering, 65, 125-136. 8- Lawryshyn, Y. A., and Cairns, B., 2003. UV Disinfection of Water: The Need for UV  Reactor Validation. Water Science Technology, 3, 293-300. 9- Wright, H. B., Lawryshyn, A., 2000. An Assessment of the Bioassay Concept for UV  Reactor Validation. Water Environment Federation on: Disinfection of Wastes in the New Millennium, New Orleans.  70  10- Kamimura, M., Furukawa, S., and Hirotsuji, J., 2002. Development of a Simulator for  Ozone/UV Reactor Based on CFD Analysis. Water Science Technology, 46, 13-19. 11- Shih, T. H., Liou, W. W., Shabbir, A., Yang, Z., and Zhu, J., 1995. A New k-ε Eddy-  Viscosity Model for High Reynolds Number Turbulent Flows - Model Development and Validation. Computers Fluids, 24, 227-238. 12- Baldyga, J., and Orciuch, W., 2001. Barium Sulphate Precipitation in a Pipe-an  Experimental Study and CFD Modelling. Chemical Engineering Science, 56, 24352444. 13- Ranade, V. V., 2002. Computational Flow Modeling for Chemical Reactor  Engineering. Academic Press, 1 st edition. pp. 135. 14- Baldyga, J., Bourne, J.R., 1999. Turbulent Mixing and Chemical Reactions.Wiley,  New York, p. 330. 15- Bolton, J. R., and Stefan, M. I., 2002. Fundamental Photochemical Approach to the  Concepts of Fluence (UV dose) and Electrical Energy Efficiency in Photochemical Degradation Reactions. Research on. Chemical Intermediates, 28, 857–870. 16- Noakesa, C. J., Fletchera, L. A., Beggsa, C. B., Sleigha, P. A., and Kerr, K. G., 2004.  Development of a Numerical Model to Simulate the Biological Inactivation Airborne Microorganisms in the Presence of Ultraviolet. Journal of Aerosol Science, 35, 489– 507. 17- Sommer, R., Pribil, W., Appelt, S., Gehringer, P., Eschweiler, H., Leth, H., Cabaj, A.,  and Haider, T., 2001. Inactivation of Bacteriphages in Water by Means of NonIonizing (UV-253.7 nm) and Ionizing (Gamma) Radiation: A Comparative Approach. Water Research, 35, 3109-3116. 18- Elyasi, S., and Taghipour, F., 2005. Simulation of a UV Photoreactor in the Eulerian  Framework Governing by Complex Deactivation Rate of Microorganisms. Third International Congress on Ultraviolet Technologies, Whistler, Canada.  71  19- Versteeg, H. K., Malalasekera, W., 1995. An Introduction to Computational Fluid  Dynamics The Finite Volume Method. Pearson Eduacation Limited, 1st edition , pp. 125-134.  72  Chapter 4. Fluence Distribution and Performance Evaluation of UV Reactor Using Optical Diagnostic Techniques 4.1 Introduction In 1955, Switzerland and Australia started using UV-disinfecting municipal drinking water stations. By 1985, the number of such stations had increased to 500 and 600, respectively, in these two countries [1]. UV-based photolysis and photo-initiated oxidation have great potential for the inactivation of microorganisms and degradation of a wide range of contaminants in water. The effect of UV on microorganisms is due to a photochemical reaction caused by the absorption of radiant energy, which results in photochemical damage to their DNA and, hence, their inactivation. UV technology is recognized as one of the best available for water disinfection. Many major cities in North America including New York City, with its 6.7 m3/s water consumption, are installing UV reactors for water treatment. The electrical power consumption of installed UV facilities are 30-180 kJ/m3 [2]. Any improvement in the performance of such a system would lead to considerable environmental and economic gains. In addition to disinfection of microorganisms, UV can also be applied to the degradation of persistent chemical contaminants in water when used in combination with an oxidant, such as hydrogen peroxide. The mechanism for the photolysis of hydrogen peroxide is the splitting of the molecule into two hydroxyl radicals, which are highly reactive components. The redox potentials of hydroxyl radicals, after fluorine, are the highest among other oxidizers, such as atomic oxygen and chlorine dioxide [3]. This implies that hydroxyl radicals will oxidize most organic persistent contaminants [4]. UVhydrogen peroxide can be used for treating clear water/wastewater that contains A version of this chapter will be submitted for publication. Elyasi, S., Taghipour, F., (2009). Fluence Distribution and Performance Evaluation of UV Reactor Using Optical 73 Diagnostic Techniques  contamination levels of less than 1000 ppm [5] or COD levels less than 5 g/L [6]. The process is applicable to small- to medium-sized industrial units for the treatment of contaminated water and for the production of highly pure water for the pharmaceutical and microelectronic industries [7]. The conventional method of evaluating the performance of a UV reactor is to compare the concentrations of pollutants or active microorganisms at the outlet of the reactor with those at the inlet [8-12]. There has been considerable effort put into improving this technique by introducing various chemicals and new methods. Bohrerova et al. [13] developed a method for calculating the received UV fluence through a photoreactor using photochemically active fluorescent microspheres. This technique was able to calculate the fluence rate distribution in the UV reactor using the concentrations at the outlet. Although this method can provide valuable information on fluence distribution inside a UV reactor, the reactor itself is treated as a black box and the fluence distribution through the UV reactor is not revealed directly. In this research, for the first time (to the author’s knowledge), a method is presented for mapping the fluence distribution in the entire UV reactor by adapting and modifying conventional planar laser-induced fluorescence (PLIF). Rhodamine WT (RhWT), a fluorescent chemical, is resistant to photolysis, but is oxidized hydroxyl radicals [14, 15]. If the RhWT solution is excited with a green laser at 532 nm, the irradiance of the re-emitted light at 588 nm from the solution is linearly proportional to the concentration of RhWT. This phenomenon can be used to trace UV photoreactive solutions (RhWT and hydrogen peroxide) in a UV reactor in order to determine the received fluence by the solution. This qualitative concentration diagnostic method is a  74  very powerful tool for the non-intrusive visualization of the concentration and the UV fluence profile inside a UV reactor. This method can reveal the existence of problematic zones in the photoreactor at various design and operating conditions. Therefore, mapping the profile of concentration and fluence in the reactor using this method can have a significant impact on understanding the reactor behavior. This can lead to alternative UV reactor designs with improved performance. The information can also be used for comprehensive evaluation of models simulating the performance of UV reactors (e.g., CFD models).  4.2 Principles of the Measurement Techniques 4.2.1 Planar Laser-Induced Fluorescence (PLIF)  Fluorescence occurs in some chemicals because photons absorbed by the chemical trigger the emission of more photons with longer wavelengths. For a clear solution (no particulates), the re-emitted radiant energy (or power) can be calculated using the quantum efficiency, Θs, denoting the ratio of the total energy emitted by a fluorescent chemical per quantum of absorbed energy. Combining quantum efficiency and the Beer-Lambert law, the re-emitted power is: PV ( x, y, z ) =  2.303Ts ls, p  2  n  ls , p    ε λs , f c f Θ s I 0 exp − ∫  2.303∑ ε λs ,i ci  dl   0    i =1   4.1  where PV, ls,p, Ts, ελs, c, I0, n, λs, l, x, y, and z are volumetric emitted power, distance of the investigation point from the light source, portion of light that passes through the solution, molar extinction coefficient (at wavelength λs nm), concentration of all chemicals (index i stands for the ith component and f for the fluorescent chemical), the intensity of light at its source, the number of chemicals, wavelength of photons leaving  75  the light source, length on the beam pathway, and coordinates of the studied point in the solution, respectively. The fraction of light that reaches the point in the solution (Ts) depends on the number of interfaces that the rays pass through and the characteristics of the interfaces as reflectors. If the emission of each element in the fluorescent solution is assumed to be diffusive, the energy captured by an array of sensors (CCD of a digital camera) or a single sensor is:  lc , p  n  Tc   ∑ ε c dl  Ec (xc , y c , z c ) = Θ P t ∆ V ∆ A cos θ exp c V xc , yc , zc c, p  ∫0  i =1 λe ,i i   4πl c2, p    4.2  where Ec, Tc, ελe, lc,p, Θc, t, ∆V, ∆A, θc,p, xc, yc, and zc are energy captured by the sensor (the value of each pixel on the CCD of camera), the fraction of the emitted light that reaches the sensor, the molar extinction coefficient (at wavelength λe nm), the distance between the sensor and emitter, the quantum efficiency of the sensor or CCD of the camera, the exposure (or integration) time of the sensor, the glowing volume element in the solution, the active area of the sensor or pixel area on the CCD of a camera, the angle between the normal vector of the sensor area and the incident ray from the emitter, and the coordinates of the sensor (pixel), respectively. Due to the terms involving the absorbance of photons by the medium, Equation 4.2 shows high non-linearity between the concentration of fluorescent chemical in the solution and the recorded value on the pixels of the CCD of a camera. Equation 4.2 can be simplified if the following criteria are respected:  76  - Chemical concentrations are relatively low, in which case, the absorption of photons (from the source and re-emitted from the fluorescent chemical) can be ignored. - The position of the laser source, the studied solution, and the camera are fixed during the experiment. As a result, all constants can be integrated into one factor as a setup coefficient. - The integration time of the camera is longer than the lifetime of the fluorescent chemical for a pulse laser source. As a result, the simplified equation for each point in the solution and corresponding point on the CCD of a digital camera is: E c ( xc , y c , z c ) = K setup ( xc , y c , z c )I 0 c f ( x, y, z ) + K offset ( xc , y c , z c )  4.3  where Ksetup are Koffset are the setup and offset coefficients (slope and intercept) for each pixel on the CCD of a camera. The first part on the right side of the equation is similar to the equation presented by Guilbault [16]. The setup coefficient (Ksetup) takes all geometrical, optical and operational parameters into consideration. The offset coefficient (Koffset) is the produced analog current by the CCD of a digital camera when I0 is equal to zero (dark current), in order to account for any background optical noise. The term I0 should not be omitted because the beam emerging from the laser is non-uniform due to heterogeneities in the lasing medium. 4.2.2 Chemical Kinetics  Hydroxyl radicals have great potential for decomposing a wide range of organic and inorganic contaminants in water [17]. The hydroxyl radicals can be produced from  77  oxidant chemicals, such as ozone, hydrogen peroxide, and oxygen in the presence of UV radiation, e.g., 2OH˚  H2O2 + hν  4.4  Hydroxyl radical oxidation, direct oxidation, and photolysis are the three primary reactions that take place during the degradation of organic materials in a UV-hydrogen peroxide system [18]: Direct Oxidation: RH (organics) + H2O2  Oxidized Product + H2O  4.5  Photolysis:  RH*  Product  4.6  Product  4.7  RH (aq) + hν  Hydroxyl radical RH + OH˚ oxidation  Intermediates  The rate of reaction of an organic material (in this case, the fluorescent chemical) is the algebraic summation of all three of these reactions. The limiting reaction is the production of the hydroxyl radicals, which is a linear function of the UV radiation distribution in the photoreactor.  4.3 Experimental 4.3.1 Materials and Chemicals  The fluorescent chemical was rhodamine WT, abbreviated as RhWT, with CAS Registry number 37299-86-8 and the index name xanthylium, 9-(2,4-dicarboxyphenyl)3,6-bis(diethlamino)-, chloride, disodium salt in a 20 wt% aqueous solution. Rhodamine WT is a bright fluorescent red dye originally developed for water tracing applications. Potassium iodide, potassium iodate, and borax were used to calibrate the UV radiant energy passing through the bench-scale photoreactor. Hydrogen peroxide was selected as an oxidant and was measured using potassium iodide, sodium hydroxide, and ammonium molybdate tetrahydrate. All chemicals were obtained from Fisher Scientific Inc., 78  excluding rhodamine WT and ammonium molybdate tetrahydrate, obtained from Turner Designs Inc. and Acros Organics, respectively. 4.3.2 Reaction Rate Measurement under Controlled Conditions in Collimated-Beam Photoreactor  The Chemical and photochemical reactions of rhodamine WT and hydrogen peroxide were measured using a customized collimated beam photoreactor. The photoreactor (Figure 4.1) was comprised of two mid-size UV lamps, a UV radiation reflector, a UV beam collimator, a double-jacketed reactor with a volume of 250 ml and a round quartz (0.003 mm thickness) at the bottom, a variable speed stirrer at the top of the reactor, and two UV sensors (SIC01M-C from Roithner LaserTechnik) below the quartz window connected to an amplifier (Multifunctional 2-Channel Amplifier Board from Roithner LaserTechnik). The voltage signal from the amplifier was recorded using a multi-meter (MultiPro 530 from Extech Instruments) and a computer. The photoreactor sample port (bottom) was connected to a flow-through cuvette of a UV-Vis spectrometer (Cary 100) via its circulating pump. The sampling solution was returned to the reactor via the top sampling port. This configuration allowed online measurement of the optical density at different wavelengths during the reaction. Figure 4.1 shows the configuration of the setup.  79  6  7  12  5  11 Cooling water  4  9 8 10  3 2 b 1  Figure 4.1. Collimated-beam photoreactor setup consisting of: polished aluminum reflector (1), mid-size UV lamps (2), collimator (3), quartz window (4), jacketed reactor body (5), variable speed stirrer (6), thermometer (7), UV sensors (8), amplifier (9), voltmeter (10), sampling ports (11), and UV spectrophotometer with circulating pump (12).  The concentration of rhodamine WT (RhWT) was measured using its absorbance at 555.5 nm by the UV-Vis spectrophotometer. To investigate the effect of hydrogen peroxide on the absorbance of RhWT, different solutions of hydrogen peroxide and RhWT at various concentrations were tested. The concentration of hydrogen peroxide was measured using the method presented by Klassen et al. [19]. Hydrogen peroxide can -  oxidize an iodide solution in alkaline pH to generate I3 ions. The absorbance at 352 nm reveals the concentration of this ion using an UV-Vis spectrometer considering the molar extinction coefficient of the ion. UV radiant energy that passed through the photoreactor was measured using the iodide-iodate actinometer method presented by Rahn et al. [20]. The iodide/iodate solution was fed into the photoreactor after the UV lamp has been turned on and stabilized after 30 minutes. The online UV-Vis spectrophotometer was used to 80  continuously measure the absorbance of the solution at 352 nm, which is related to the concentration of I3- or absorbed UV radiant energy in the reactor. At the same time, the signals (voltage) produced by the UV sensors below the quartz window were recorded as reference points for the period of measurement. The absorbance (concentration) results yielded the amount of radiant energy entering into the photoreactor, considering the quantum yield of the iodide-iodate actinometer. The direct oxidation rate of rhodamine WT was measured by mixing a solution of RhWT and hydrogen peroxide with concentrations of 791±4 ppb and 9.72±0.08 ppm, respectively, in the collimated-beam photoreactor for a two-hour period, while the UV lamp was not operating. The photolysis rate of rhodamine WT was evaluated in the photoreactor while the UV lamp was operating in a stabilized mode (as indicated by a constant voltage signal from the sensors). The photoreactor was charged with the RhWT solution (805±4 and 600±3 ppb), and during the process of photolysis, its concentration was measured for a period of two hours. Because the rate of photolysis is a linear function of the concentration of the chemicals, the concentration of RhWT was selected at a high level to enhance the photolysis rate. The photo-initiated oxidation rate of RhWT with hydrogen peroxide and UV radiation was determined using RhWT solutions with a concentration of 126±1 ppb and hydrogen peroxide with a concentration of 10.14±0.05 ppm in the collimated-beam photoreactor. The concentration of RhWT was measured as a function of time using the online spectrophotometer, and the concentration of hydrogen peroxide at the end of -  reaction (9.77±0.05 ppm) was determined using the I3 method.  81  4.3.3 Concentration Measurements by PLIF in Flow-Through Photoreactor 4.3.3.1 Flow-Through Photoreactor  Considering the PLIF criteria [21], a flow-through UV photoreactor was carefully designed in order to increase the accuracy of the measurements by minimizing or eliminating possible optical noise. The height and width of the photoreactor were minimized to reduce optical absorbance. The reactor body was built of glass to be UV resistant and to eliminate the staining effect of RhWT on the reactor surface. A highperformance, high-output, low-pressure UV lamp (200 W, arc length 1.07 m from Emperor Aquatics Inc.) installed inside a quartz tube below the reactor was used as the radiation source. The UV reactor configuration and its dimensions are provided in Figure 4.2. Outlet  2.5  To Vacuum Inlet 4.5  UV Lamp Quartz Sleeve Air in  Figure 4.2. The flow-through UV photoreactor used for PLIF measurements (all dimensions are expressed in cm).  82  A schematic diagram of the entire system consisting of the pump, piping, instrumentation, and storage tanks, is shown in Figure 4.3.  7 3  11 6  FTP  5 1  8  2  Drain 4  10  9  Figure 4.3. Schematic of pilot-scale photoreactor consisting of: product reservoir (1), feed reservoir (2), stirrer (3), centrifugal pump (4), flow/pressure/temperature meter (5), photoreactor (6), laser source (7), digital camera (8), PLIF control unit (9), data acquisition system (10), and online spectrophotometer (11).  The UV lamp was operated for 30 to 45 minutes to allow stabilization. During this period, the pump (4) circulated distilled water at 20±1ºC from tank 1 through the photoreactor. The photoreactive solution—hydrogen peroxide (10.30±0.05 ppm) and rhodamine WT (126 ppb) at 21.4±0.3ºC—was filled into tank 2. Once the UV lamp stabilized, distilled water was pumped out to the drain system and the photoreactive solution was fed into the reactor. Flow rate and temperature were recorded using a calibrated flow/temperature meter (FluidVision 4000 from Proteus Industries). The concentration of RhWT (at the outlet of the photoreactor) was measured by an online spectrophotometer. The temperature of the solution was kept at 20±1ºC, and cooling air  83  was passed through the sleeve of the photoreactor at 1.33±0.16 L/s. This procedure guaranteed isothermal photoreaction in the reactor. Returned liquid was collected into tank 1 for final treatment, the deterioration of the RhWT, before sending it to the drain. 4.3.3.2 PLIF Tests i. PLIF Setup  A conventional PLIF system consist of a single camera, a laser source, a synchronizer unit, and an image acquisition/processing system. Typically, a pulse laser having a uniform beam energy is used. For a fixed laser energy (e.g., 10 mJ), a pulse-topulse laser beam energy variation of about ±7% has been reported [21]. In our experiments, similar energy variations were measured. Measuring the energy value of the laser beam (I0 in Equation 4.3) improves the accuracy for calculating the local concentration of the fluorescent species. For this research, a second camera was integrated into the conventional system to achieve a local laser energy content measurement by mapping the laser energy at each pulse. The beam from the laser source at 532 nm (Nd:YAG laser, Solo III from NewWave Research) was divided in two using a beam splitter. One part of the beam acted as a light sheet after passing through the reactive solution in the photoreactor, and this caused the light to be re-emitted at 588 nm. The re-emitted light was captured using a 12-bit digital camera (C8484 from Hamamatsu Cooperation) equipped with a high-pass filter (>550 nm). The other part of the beam was reflected from a barium sulphate-coated glass and captured by a second camera with a band-pass filter (532±10 nm). Synchronization, image acquisition, and image processing were performed by a FlowMap system hub from Dantec Dynamics. Figure 4.4 shows the configuration of the PLIF setup.  84  1 6 5  9  3 8  y 7  x z  4  2  Figure 4.4. Schematic view of the PLIF apparatus for measuring concentrations of the photoreactive solution. The setup consists of: a pulsed laser at 532 nm (1); a digital camera with a high-pass filter (2), a digital camera with a band-pass filter (3), a photoreactor (4), a beam splitter (5), a diffusive reflector (6), reemitted light at 588 nm (7), and laser sheets (8, 9). ii. Calibration Procedure  During the calibration and measurement, the coordinates of the cameras, laser source, and reactor were unchanged, and the temperature of the solutions remained at 20±1 ºC. Hydrogen peroxide solutions (9.7±0.6 ppm) with different rhodamine WT concentrations (10 to 130 ppb) were passed through the reactor, and 400 images were captured for each concentration using both cameras while the laser operated. The value of each pixel (Ec in Equation 4.3) at location (xc, yc, zc) was subtracted from the value of each pixel while the laser was not operating (to correct for any background light) and normalized by dividing by the total value of the total reflected light (I0) captured by the  85  second camera (to correct for any variation in the pulsed bean energy). Finally, the average of 400 normalized images for different concentrations of RhWT were used to calculate Ksetup and Koffset, the slope and intercept, for each pixel on the image. The results of the calibration calculation were two-dimensional arrays of Ksetup and Koffset for each point in the reactor excited by the laser sheet. The presence of hydrogen peroxide in the calibration procedure is necessary to achieve accurate results, as the fluorescent characteristic of RhWT heavily depends on the oxidant content of the solution [22]. Because the photoreactive solution is bleached with a high-power laser pulse, the solution should flow through the reactor and be replaced with fresh solution for each run. Rhodamine WT has a fluorescent effect under UV radiation (254 nm), which creates a background (optical noise) in the final measurement. To quantify and make corrections for this background noise, the UV reactor ran (with the UV lamp on) with different concentrations of RhWT without hydrogen peroxide, and 400 images were captured in the absence of the laser pulse. The average values of the images were used to calculate the glowing background image (offset) for the final concentrations in the UV reactor. In order to reconstruct the concentration profile throughout the entire reactor, the length of the reactor was divided into three zones: the inlet, middle, and outlet zones. Based on the length of the reactor and the field of view of the camera, adequate overlap was considered between two adjacent zones. iii. Concentration Measurement  A stock solution of hydrogen peroxide (10.30±0.05 ppm) and RhWT (126 ppb) at 21.4±0.3ºC was fed into the reactor once the UV lamp was stabilized. Three different 86  mass flow rates (0.006±0.002, 0.015±0.002, and 0.020±0.002 kg/s, which correspond to mean axial velocities of 0.06±0.02, 0.15±0.02, 0.20±0.02 m/s, respectively, at the inlet of UV photoreactor) were tested. The flow rates are reported with 95% confidence intervals. For each operating condition, the inlet, middle, and outlet zones of the reactor were studied, and 400 images were captured for each section. The average of the energy captured (value) in the images after subtracting the glowing effect of the background (offset) was used for the concentration calculation. This procedure was applied to all three zones covering the entire length of the reactor.  4.4 Results and Discussions 4.4.1 Optical Absorbance of Rhodamine WT Solution  The correlations of the absorbance vs. the concentration of the rhodamine solution for hydrogen peroxide concentrations less than 10 ppm and for three different wavelengths were measured to be:  I spec   = −(9.85 ± 0.05) ×10 −5 C RhWT lcell ABS 532 = log 10  I   spec, 0   4.8  ABS555.5 = −(2.14 ± 0.01) × 10 −4 C RhWT lcell  4.9  ABS 588 = −(1.32 ± 0.02 ) × 10 −5 C RhWT lcell  4.10  where ABS, Ispec, Ispec ,0, CRhWT, and lcell are absorbance at a specified wavelength, intensity leaving the spectrophotometer cell, intensity entering into the cell, concentration of rhodamine WT solution in ppb, and the light path length of the cell in cm, respectively. Consequently, the molar extinction coefficient (ε) of rhodamine WT at wavelengths 532, 555.5, and 588 nm were calculated as 109000±550, 237000±1110, and 14900±220 M1  cm-1, respectively.  87  Considering the wavelength of the laser source (532 nm), 11% of the laser energy content is absorbed by the medium for a 4 cm (reactor height) layer of rhodamine WT solution at 126 ppb. This value points to considerable changes in the radiation intensity, which should be taken into consideration by performing separate calibrations for each point in the solution, in order to find a calibration map (2D) instead of a calibration curve (1D). Using this technique, optical absorbance is integrated into the calibration map with some limitations. The concentration gradient along the laser beam, from top to bottom, should not change significantly (e.g., less than 35 ppb to achieve an error of less than 3%). This limitation should be considered in the case of laminar flow, specifically, because no vigorous mixing occurs in the reaction zone in the direction of the laser beam or for high-output UV lamps. In this research, the concentration change over the length of the laser beam did not exceed 35 ppb. If the concentration gradient is high (at the end of the reactor) or the light pathway is long, the absorbance terms in Equations 4.2 and 4.3 should be calculated to reduce the error corresponding to high optical density. For re-emitted rays at 588 nm from the 126 ppb solution of RhWT, the thickness of the reactor should be controlled (e.g. less than 8 cm to achieve less than 3% drops based on absorbance). For this research, the 1 cm thickness of the reactor did not have any significant impact on the absorbance at 588 nm at the maximum concentration. 4.4.2 Direct Oxidation of Rhodamine WT  The rate of change of the RhWT concentration with an excess of hydrogen peroxide and no UV radiant was calculated to be: d  C RhWT    = −(8.53 ± 1.05) × 10 -7  dt  C RhWT,0   4.11  88  where t and CRhWT,0 are time (s) and initial concentration of RhWT, respectively. It is important to note that, after 3.25 hours, the concentration of RhWT dropped by 1%. This means that the stock solution can only be prepared and kept for less than 1 hour without significantly changing the concentration of RhWT, and this reaction can be ignored due to the low residence time of the solution inside the reactor (of the order of seconds). 4.4.3 Photolysis of Rhodamine WT  RhWT is a persistent chemical under UV exposure; the concentration of RhWT did not change after receiving 1000 J/m2 of UV irradiance. Thus, the photolysis of RhWT can be ignored. 4.4.4 Photo-Initiated Oxidation of Rhodamine WT with Hydrogen Peroxide and UV Radiation  The photo-initiated oxidation rate of RhWT with an excess of hydrogen peroxide after receiving UV irradiance or a UV fluence, H (J/m2), was calculated to be:  C ln  RhWT  C RhWT ,0    = (− 4.76 ± 0.09 ) × 10 −3 H    4.12  The reaction of RhWT with hydroxyl radicals produces other intermediates and by-products. If the concentration of by-products exceeds a certain limit (65% conversion of RhWT), their reactions become dominant, and this correlation would be invalid. As a result, the linear relationship between the received fluence and the log of concentration ratios is only valid for concentrations of reacted RhWT higher than 40 ppb. Equation 4.12 can be used to calculate the received UV irradiance or fluence at all locations inside the reactor.  89  4.4.5 Concentration Profile in the Photoreactor  Figures 4.5 and 4.6 show the measured concentration profiles through the reactor in different sections, which are stitched together. Due to the technical limitations, such as difficulties in positioning the cameras at the exact position used for the calibration procedure (±5 mm), as well as the flow rate fluctuations during the experiments (±0.002 kg/s), the averaged images from the three different sections of the reactor did not overlap perfectly. Also, the values very close to the reactor walls were masked because they did not represent the true concentrations as a consequence of reflections from the white glue joints in this region of the reactor. For all images, a horizontal median filter with rank 40 was applied to remove the shadowing effects of the laser sheet on the images. Inlet zone  Middle zone  130 ppb  Outlet zone  1  2  3 80 ppb Figure 4.5. Concentration profile of rhodamine WT (ppb) in the UV reactor at steady  state condition without correction for laser energy variation. Color bar (linear scale) indicates concentrations from 130 (red) to 80 ppb (blue) and arrows show the flow directions. Illustrations 1-3 are for mass flow rates of 0.006, 0.015, and 0.020 kg/s, respectively.  90  In Figure 4.5, the concentration is calculated according to the conventional PLIF method without taking into account the laser energy content (i.e., I0 is omitted). At higher flow rates, higher concentrations of RhWT should extend toward the outlet of reactor; however, this cannot be concluded from the images. This discrepancy was resolved when the laser energy was taken into consideration, as presented in Figure 4.6. 130 ppb  1  2  3 80 ppb  Figure 4.6. Concentration profile of rhodamine WT (ppb) in the UV reactor with correction for the laser energy variation at steady state condition. Color bar (linear scale) indicates concentrations from 130 (red) to 80 ppb (blue), and arrows show the flow directions. Illustrations 1-3 are for mass flow rates of 0.006, 0.015, and 0.020 kg/s, respectively.  The “Received Fluence Profile” throughout the reactor is a very useful piece of information. The fluence distribution reveals the local and global performance of a UV reactor, taking all the important phenomena, such as hydrodynamics, UV radiant distribution and kinetics, into consideration. The fluence profile points directly to the most efficient and deficient zones in the reactor with respect to the reactor performance.  91  The fluence distribution was calculated using Equation 4.12 for different flow rates in the pilot scale UV reactor as presented in Figure 4.7. 60 J/m2  1  2  3 5 J/m2  Figure 4.7. Received UV fluence profile of rhodamine WT (ppb) in the UV reactor under steady-state conditions. Color bar (linear scale) indicates fluence, from 60 (red) to 5 J/m2 (blue) and arrows show the flow directions. Illustrations 1-3 are for mass flow rates of 0.006, 0.015, and 0.020 kg/s, respectively.  As shown in Figure 4.7, at the inlet of the reactor, case 1, the received fluence is relatively high. The fluence near the inlet is highest at the lowest flow rate, as can be seen in case 1, Figure 4.7. This is confirmed by considering the profile of concentration in Figure 4.6, which shows relatively low concentration (e.g., treated flow) at the reactor inlet, particularly at lower flow rates. Mapping the velocity profile in the reactor revealed the reason for the high fluence at the inlet. Particle Image Velocimetry (PIV) was applied to measure the velocity profile, using the same system described for the PLIF experiments. Figure 4.8 shows the measured velocity profile (vector) inside the reactor for a 0.014±0.002 kg/s flow rate.  92  Figure 4.8. Measured velocity profile inside the reactor using PIV (first half of the reactor).  It is obvious from Figure 4.8 that part of the flow re-circulates (due to a pressure difference) in the upper part of reactor toward the inlet. This recalculation brings back a portion of the reacted chemicals (bleached) from the middle of the reactor. At lower flow rates (e.g., case 1), due to the low velocities at the upper left-hand corner of the reactor, the residence time is very high (i.e., the zone is nearly stagnant), and this section receives a higher fluence because of the longer exposure time. At relatively higher flow rates (e.g., cases 2 and 3), the re-circulating velocities in this section are increased, and the chemicals in this zone are mixed with fresh fluid from the inlet. In the other words, the nearly stagnant zone disappears at higher flow rates causing a more uniform concentration. The back-flow circulation causes the diminishing of the reactant (RhWT) vertical concentration gradient at the entrance of the reactor up to approximately 20 cm. This effect is more obvious for higher flow rates because of a more vigorous mixing. Once the recirculation effect is reduced at about 20 cm from reactor inlet, the concentration gradient is more distinguishable with lower concentrations closer to the UV lamp, where the photoreaction rate is higher. Figure 4.7, case 1 shows a higher receiving fluence at the inlet adjacent to the UV lamp because of a low circulating stream at that  93  location. Figures 4.7, cases 2 and 3, clearly shows the mixing effect and residence time of the fresh fluid near the inlet region. The other important phenomenon is the fluence distribution in the layer close to the lamp. Due to the lack of turbulence in the reactor, turbulent diffusion is omitted, and convective mass transfer dominates, even for higher flow rates. This information is very useful in analyzing the performance of photoreactors and technical optimization. The concentration trend in the photoreactor for different flow rates can be interpreted if the photoreactor is assumed to be as an ideal plug flow reactor. A simple correlation can be derived in order to estimate the concentration of RhWT close to the outlet zone of the photoreactor, where the flow is nearly developed (details in Appendix 5). The concentration for any specific location over the reactor length for two different flow rates (average velocities) as a function of inlet velocity and concentration (CRhWT,0 =126 ppb) is  CV 2   C RhWT , 0     CV 1 =   C RhWT , 0     V1      V 2      4.13  where CV1 and CV2 are the concentrations of RhWT at one specific location for two different average velocities V1 and V2, respectively. Using Equation 4.13, the calculated concentrations at 35 cm from inlet, 22 cm above the lamp are 100, 114, 117 ppb for flow rates of 0.006, 0.015, and 0.020 kg/s, respectively. The calculated values are comparable with the PLIF measured concentrations of 108±10, 111±10, and 115±10 ppb at the same position for the three different flow rates, respectively. This point was selected for comparison, since the effect  94  of flow circulation to some degree is eliminated beyond 20 cm and the uncertainty associated with PLIF measurements is relatively low. The PLIF images in the very last section of the reactor do not highlight the concentration differences at various flow rates. This could have been the result of the uncertainty associated with the measurements which are particularly important in this region of the reactor (see Section 4.5).  4.5 Source of Errors and Uncertainties The procedure presented here for measuring reactive flow using a modified PLIF is a valuable tool for evaluating the performance of UV photoreactors. However, some potential sources of error and uncertainty should be considered when interpreting the data. Some of these are instrumental errors include the ability of a digital camera to effectively capture sequential images, a possible negative effect of temperature on the circuit board of the camera, potential instability of the UV lamp emission (leading to variations in excitation of fluorescent chemical), and imperfect matching of coordinates between the calibration and measured concentration images. Using a digital camera and correcting the coordinates of the images by software considerably reduced these potential sources of errors, but did not eliminate them. Precise quantification of the instrumental errors is impossible. However, the major sources of error and uncertainty are evaluated and addressed below. The major source of error in the concentration calculation procedure, based on calibration maps, occurs where the concentration gradient is high. These high concentration gradients occur near the ends of the reactor, where the RhWT concentration is very low, close to the UV lamp surface, and high, far from the lamp, due  95  to reaction or recycling of material at the top of the reactor (near the inlet). The maximum error occurs when the concentration at the top is higher (due to a high absorption of light at the reactor exit) than at the bottom (see Section 4.4.1). Therefore, all measurements in the outlet sections are expected to have a relatively high degree of uncertainty for all cases, and this explains the similar concentration profiles found for all the cases in Figure 4.6, at the ends of the reactor. This error could be minimized or eliminated if the equation for the calibration map (Equation 4.3) could be derived without the simplifying assumptions concerning the optical density. The second major source of uncertainty in the calculations is the linear curve fitting for each pixel of the images. Considering the error bars for each measurement, it can be seen that there is no unique linear function between the captured emitted light (Ec) and the RhWT concentration (Cf). In other words, any of the lines located in a band (between the high and low values of the data considering error bars) could be used in the concentration calculation. The upper and lower bound can be calculated from Equation 4.3 as: Ec = (K setup ± ∆K setup )I 0 c f + (K offset ± ∆K offset )  4.13  where ∆Ksetup and ∆Koffset are used to calculate the upper/lower values for the slope and intercept. Using Equation 4.13, the uncertainty (upper/lower value) in the measurement of RhWT concentration was calculated and is shown in Figure 4.9.  96  ±30 ppb  1  2  3 0 ppb  Figure 4.9. Uncertainty in RhWT concentration profile measurement (ppb) with a 95% confidence interval in the UV reactor. Color bar (linear scale) indicates concentration uncertainty from ±30 (red) to 0 ppb (blue), and arrows show the flow directions. Cases 1-3 are for mass flow rates of 0.006, 0.015, and 0.020 kg/s, respectively.  This figure clearly shows that, at the inlet and outlet sections, the degree of uncertainty in calculation is higher than in the central zones because of the highly reflective vertical walls in these zones (which are not considered in the calculation procedure), and the high concentration gradients of RhWT.  4.6 Conclusion The growing use of UV reactors brings with it the need to improve their performance and develop new measuring techniques for performance evaluation. A modified Planar Laser-Induced Fluorescent (PLIF) method was developed for mapping the concentration and fluence profiles throughout a UV reactor. It was observed that  97  integrating the laser energy into the conventional PLIF and using a 2D calibration map instead of a typical 1D-calibration curve increases the accuracy of the measurements. The method described here provided UV fluence profiles at each cross-section of a photoreactor under various operating conditions and allowed the radial and longitude dispersion of photoreactive chemical to be estimated. The fluence distribution reveals valuable information about the performance of each reactor zone, assisting technical optimization of the reactor. Although this technique is a powerful visual diagnostic tool, technical limitations can cause significant errors. As described, if there is a considerable absorption of the laser light or re-emitted light in the system, optical density should be considered in the equations used for calibration and concentration measurements. Most industrial UV reactors operate in the turbulent regime, and the radial concentration gradients are expected to be lower than 5 to 7 ppb/cm, resulting in negligible radiation absorption. Therefore, this method should be applicable without further modification.  98  4.7 References 1- Kruithof, J. C., and Van Der Leer., R. C., 1990. Practical Eperiences with UV-  Disinfection in The Netherlands. Proceedings of the American Water Works Association Seminar on Emerging Technologies in Practice, Annual Conference of the American Water Works Association, Cincinnati, OH, June 17 – 21. 2- Office of Water, 2006. Disinfection Guidance Manual for the Final Long Term  Enhanced Surface Water Treatment Rule, EPA 815-R-06-007. 3- Munter, R., Preis, S., Kallas, J., Trapido, M. and Veressinina, Y, 2001. Advanced  Oxidation Processes (AOPs): Water Treatment Technology for the Twenty-First Century. Kemia-Kemi, 28, 354-362. 4- Von Gunten, U., 2003. Ozonation of Drinking Water: Part I. Oxidation Kinetics and  Product Formation. Water Research, 37, 1443-1467. 5- Kidman, R.B., and Tsuji, K. S., 1992. Preliminary Costs Comparison of Advanced  Oxidation Processes. Los Alamos National Laboratory, LA-12221-MS. 6- Andreozzi, R., Caprio, V., Insola, A., and Marotta, R., 1999. Advanced Oxidation  Processes (AOP) for Water Purification and Recovery. Catalysis Today, 53, 51-59. 7- Legrini, O., Oliveros, E., and Braun, A. M., 1993. Photochemical Processes for Water  Treatment. Chemical Reviews, 93, 671-698. 8- Entrala1, E, Garin, Y. J. F., Meneceur, P, Hayat, M., Scherpereel, G., Savin, C, Feliers,  C., and Derouin, F., 2007. Pilot-Scale Evaluation of UV Reactors Against In-Vitro Infectivity of Cryptosporidium Parvum Oocysts. FEMS Immunology & Medical Microbiology, 51, 555–561. 9- Cotton, C., and Passantino, L., 2005. Regulations in the United States: Requirements  and Guidance for Ultraviolet Disinfection of Drinking Water, J. Environ. Eng. Sci., 4, S57–S63.  99  10- Lawryshyn, Y. A., and Cairns, B., 2003. UV Disinfection of Water: the Need for UV  Reactor Validation, Water Science and Technology: Water Supply, 3, 293–300. 11- Scheible, O. K., and Weber, E. T., 2003. Verification Test Plan for the Suntec  Environmental UV Disinfection for Secondary Effluent Applications Version 3, US Environmental Protection Agency Edison, NJ and HydroQual, Inc. Mahwah, NJ. 12- Petri, B., Cairns, W., Gowman, L., and Mao, T., 2005. Safeguarding Public and  Environmental Health: What Are the Necessary Requirements of UV Reactor Validation Protocols?. Journal of Water and Environment Technology, 3,85-92. 13- Bohrerova, Z., Bohrer, G., Mohanraj, S.M., Ducoste, J., and Linden K.G., 2005,  Experimental Measurements of Fluence Distribution in a UV Reactor Using Fluorescent Microspheres, Environ. Sci. Technol., 39, 8925-8930. 14- Sen, S., Tsai, K., Gillis, P., Larkins, R., Spradling, R., and Melton, L. A., 1999.  Evaluation of Micro-Mixing Models in Simulating Liquid Phase Turbulent Reacting Flow. AIChE Annual Meeting, Dallas, Texas., 31 October–5 November, paper 309i. 15- Guilbault, G.G., 1990, Practical Fluorescence. New York: Marcel Dekker. 16- Karasso, P. S., and Mungal, M. G., 1997. PLIF Measurement in Aqueous Flows  Using the Nd:YAG laser. Experiments In Fluid 23, 382-387. 17- Gunten U. V., 2003. Review, Ozonation of Drinking Water: Part I. Oxidation  Kinetics and Product Formation. Water Research, 37, 1443-1467. 18- Oppenlander, T., 2003. Photochemical Purification of Water and Air, Wiley-VCH. 19- Klassen, N. V., Marchington, D., and McGowan, C. E., 1994. H2O2 Determination by  I3- Method and by KMnO4 Titration. Analytical Chemistry, 66, 2921-2925. 20- Rahn, R. O., Stefan, M. I., Bolton, J. R., and Goren, E., 2003. Quantum Yield of the  Iodide-Iodate Chemical Actinometer: Dependence on Wavelength and concentration. Photochemistry and Photobiology, 78, 146-152.  100  21- Melton, L. A., and Lipp, C. W., 2003. Criteria for Quantitative PLIF Experiments  Using High-Power Lasers. Experiments In Fluid 35, 310-316. 22- Becker R.S., 1969. Theory and Interpretation of Fluorescence and Phosphorescence,  Wiley Interscience, 1st edition.  101  Chapter 5. Simulation of UV Photoreactor for Degradation of Chemical Contaminants: Model Development and Evaluation 5.1 Introduction UV technology, in combination with an oxidant such as hydrogen peroxide, provides an effective method of water and wastewater treatment. This photo-initiated oxidation process (also referred to as the UV-based advanced oxidation process, AOP) can be applied when treating optically clear water or wastewater that contains contaminants with concentrations of less than 1000 ppm [1]. In UV-based AOP, hydroxyl radicals which are generated through the reaction of UV radiation with hydrogen peroxide, play a major role in oxidizing persistent chemicals. The rate constant of the reaction of hydroxyl radicals with persistent chemicals is typically very high (107 to 109 M-1s-1) [2]; however, the overall rate of reaction is low due to the necessarily low concentration of hydroxyl radicals. A mathematical model simulating the performance of UV reactors can contribute to our understanding of UV technology for water treatment, allowing us to obtain the full benefit of UV photoreactors. The modeling of a UV photoreactor is a challenging area of research because of the multi-physics nature of this type of reactor. The model should allow for the effects of fluid movement and mixing (hydrodynamics), the distribution of radiant energy (radiation field), the rate of deterioration of the chemicals (photochemical/chemical kinetics), and the interaction of these phenomena with one another. The UV reactor performance model (semi-mechanistic approach) consists of a system of partial differential equations governing various phenomena in the UV reactor that should be solved simultaneously. Due to the complexity of the reactor geometry and interaction of A version of this chapter will be submitted for publication. Elyasi, S., Taghipour, F., (2008). Simulation of UV Photoreactor for Degradation of Chemical Contaminants: 102 Model Development and Evaluation  many phenomena, analytical solutions do not exist, and a numerical solution is the only practical approach, using relevant techniques such as Computational Fluid Dynamics (CFD). The modeling results for reactor performance should be validated against experimental data (hydrodynamics, radiation, and species concentration) to ensure their applicability to reactor design and optimization. A number of researchers have developed models of photoreactor performance [e.g., 3-7]; however, they have used conventional methods for model evaluation. These methods compare the modeling predictions with the experimental results for the overall reactor performance, which is determined by comparing the ratio of the outlet to the inlet concentration of the reactor. This method does not reveal any discrepancies that might exist between the model predictions and the experimental values for velocity, radiant energy, and concentration of species inside the photoreactors. Some other researchers [8, 9] have compared the hydrodynamics from the reactor performance model with experimental velocity data inside the photoreactor obtained using Particle Image Velocimetery (PIV). But, in these cases, the CFD model predictions have been typically validated using PIV experimental data for only a few strategic locations inside the photoreactors, instead of throughout the entire reactor domain. In this research, a general methodology is presented for developing an integrated CFD model of UV reactor performance and evaluating all parts of the model separately. The results of each part of the integrated model (hydrodynamic, radiation, species concentration) are evaluated separately against experimental values in a UV photoreactor  103  in order to develop a more practical and reliable integrated model. The approach can be summarized as follows: - The simulated velocity field is compared with PIV measurements at every point, and velocity vector profiles for the entire reactor cross-section are presented; - The radiation field is modeled and compared with measured values, which are recorded using a photodiode sensor located at discrete positions throughout for the entire domain of the reactor; - The photoreaction rate of a chemical used as a model contaminant for the UVadvanced oxidation process is measured in a bench-scale photoreactor under controlled conditions to obtain a photoreaction kinetic model; - An integrated model of hydrodynamics, radiation, and conservation of a chemical species is developed to determine the concentration profile throughout the reactor. The integrated model is evaluated using concentration data obtained with a modified Planar Laser-Induced Fluorescence (PLIF) method. This powerful technique reveals the concentration profile throughout the entire domain of the reactor and can be useful to validate the full model. The strategy presented can be employed for any type of UV reactor in order to obtain a reliable integrated model for performance simulation.  5.2 Theory 5.2.1 Mass and Momentum Conservation (Hydrodynamics)  The velocity field can be obtained by solving the equations of mass and momentum conservation. The general forms of the conservation of mass and momentum (Navier-Stokes equation) are:  104  ∂ (ρ ) + ∇ ⋅ (ρu ) = 0 ∂t ∂ (ρu ) + ∇ ⋅ (ρuu ) = ∇p − ∇ ⋅ τ + ρg + F ∂t  5.1  5.2  where t, ρ, u, p, τ, g, and F are time, medium density, velocity vector, pressure, viscous stress tensor, gravitational acceleration, and external body force, respectively. An analytical solution of the system of partial non-linear differential equations is not available for complex geometrics, so these must be solved numerically. The threedimensional, time-dependent numerical solution of Equation 5.2 (Direct Numerical Simulation, DNS) is only applicable for a very small computational domain and a laminar flow regime due to the extensive computational resources required. For the turbulent regime, an acceptable engineering approach to solve Equation 5.2 is the use of statistical methods or the classical approach that solves the Reynolds Averaged Navier-Stokes (RANS) form of the equations [10]. In the RANS approach, the Reynolds stress tensors are semi-empirically correlated using algebraic [11], one-equation [12], two-equation (e.g., standard k- ε [13], RNG k- ε [14], realizable k- ε [15], or standard k-ω [16]), or multiple equation models of turbulence such as the Reynolds Stress Model (RSM) [1719]. 5.2.2 Radiant Energy Conservation  The simplified form of the Radiation Transfer Equation (RTE) [20] is the BeerLambert law that is applicable in many UV reactors, due to a lack of scattering (no significant concentration of particulates) and emission (relatively low temperature) throughout the medium. The differential form of the Beer-Lambert law for steady-state conditions is:  105  dI (s, Ω ) + k ( s, Ω ) I (s , Ω ) = 0 ds  5.3  where I and k are the intensity and the absorption coefficient for a specific solid angle (direction) Ω and position vector s. For a photoreactor, changes in intensity depend not only on the medium absorption coefficient but also on the refraction/reflection through/from different materials in the medium, such as the quartz body of the UV lamp, the quartz protector of the UV lamp (sleeve), and the body of the reactor. These refraction/reflection phenomena should be integrated into Equation 5.2. The photoreaction rate is a function of the radiant power obtained by integrating the intensity over the entire solid angle (4π). Considering all the above-mentioned effects, the radiant power per unit area, or fluence rate, at a point (described in Chapter 2) is:     4π  n  m   1 G = ∫ I (Ω ) exp − ∑ ∫ k i (s, Ω )ds  ∏ Ti  dΩ 2 0 n   i =1 overLi  i =1  0   ∑ Li     i =1    5.4  where Li, I0, ki, and Ti, are the path length of a ray through the medium ith, intensity at source, the absorption coefficient of medium i, and the fraction of the ray transmitted from one medium to another, respectively. Fluence rate is the key parameter in the photoreaction rate correlation. 5.2.3 Species Mass Conservation  For each individual chemical (species m) in the computational domain, the mass conservation equation is: ∂ (xm ρ ) + ∇ ⋅ (xm ρu ) = −∇ ⋅ jeff + S m ∂t  5.5  106  where x, jeff, and S are mass fraction, effective diffusive flux, and source/sink term of species m, respectively. The diffusive term can be derived using Fick’s law with the molecular diffusion coefficient of the species in the medium and a turbulent Schmidt number of 0.7 [21]. The source/sink term depends on the nature of the reaction occurring in the domain of the reactor. Adding an oxidant (e.g., hydrogen peroxide) to the UV-based photoreactor produces hydroxyl radicals, very strong oxidants that can oxidize many chemical contaminants. Three potential primary reactions occur in parallel for the degradation of a chemical contaminant: 1) direct oxidation of chemical contaminant with oxidant; 2) photolysis by UV radiation; and 3) reaction with hydroxyl radicals [22]. The value of the sink term in the species conservation equation for the reactant is the sum of these three reaction rates. For many chemicals, the rates of reactions are available [23]; however, for certain chemicals, the rates need to be measured. In addition, if a chemical of interest (contaminant) exists in the presence of other impurities (which is the case in many practical applications), the rates may be affected and, hence, should be measured experimentally. The reaction rates can be determined using a bench-scale collimatedbeam UV photoreactor under controlled conditions (i.e., well-mixed with a known radiation field). It is important to note that the photoreaction rate is a function of the radiant power (fluence rate) distribution in the medium. Therefore, the fluence rate should be simulated prior to photoreaction rate modeling. In addition to the fluence rate, which appears in the source term of the species mass conservation equation, the velocity also appears in the convection term of this equation. Therefore, the velocity field needs to be simulated. As a result, for developing an integrated model of reactor performances,  107  which reveals the concentration profile of species, it is essential to also develop models of hydrodynamics and radiation distribution.  5.3 Experimental Methods 5.3.1 Material  A fluorescent chemical, Rhodamine WT, abbreviated as RhWT, was selected as the chemical candidate for photoreactor modeling and model evaluation. RhWT, with the CAS registry number 37299-86-8 and the index name Xanthylium, 9-(2,4dicarboxyphenyl)-3,6-bis(diethlamino)-, chloride, disodium salt in a 20 wt% aqueous solution was supplied by Turner Designs Inc. Potassium iodide, potassium iodate, and borax were used to calibrate the UV radiant energy passing through a bench-scale photoreactor (used for kinetic measurements). Hydrogen peroxide was used as an oxidant and was measured using potassium iodide, sodium hydroxide, and ammonium molybdate tetrahydrate. Ammonium molybdate tetrahydrate was supplied by Acros Organics, and all of the other chemicals were supplied by Fisher Scientific Inc. 5.3.2 Concentration Measurement  The absorbance of different weight percentages of RhWT (10 ppb to 10 ppm) at different hydrogen peroxide concentrations was measured at 555.5 nm (maximum absorbance) using a Cary 100 UV-Vis spectrophotometer to obtain its extinction coefficient. The concentration of hydrogen peroxide in the solution was measured using the method presented by Klassen et al. [24]. Hydrogen peroxide can oxidize iodide solutions under alkaline pH conditions and generates I3- ion. The absorbance of the solution at 352 nm was used to back-calculate the concentration of hydrogen peroxide.  108  5.3.3 Flow-Through Pilot-Scale Photoreactor  A flow-through pilot-scale photoreactor was tested under different flow conditions. The dimensions of the reactor were selected to meet the criteria for Planar Laser-Induced Fluorescence that were recommended by Menton and Lipp [25] in order to minimize the optical absorbance. The body of the reactor was built from glass to prevent its transparency (being affected) by either UV radiation or RhWT adsorption over the experiment period. At the bottom of the reactor, a quartz sleeve was attached to hold a large UV lamp (200W, arc length 1.07 m, from Emperor Aquatics Inc.). A hole at the end of the sleeve with air suction at the other end maintained an ambient skin temperature on the surface of the sleeve and maintained isothermal conditions. Figure 5.1 shows the dimensions and configuration of the pilot-scale reactor. Outlet  2.1 50.0  Inlet  4.5  1.0 To Vacuum Air in  1.0  Quartz Sleeve (Φ2.5)  UV Lamp  Figure 5.1. The photoreactor used for PLIF measurements (all dimensions are expressed in cm).  5.3.3.1 Velocity Measurement  Particle image velocimetry was applied to measure the velocity field. The flowthrough photoreactor was installed in a piping network consisting of pumps, piping, instrumentation, a storage tank, and a Particle Image Velocimetry (PIV) setup. Figure 5.2 109  shows a schematic diagram of the PIV setup. The same setup was used to measure the concentration profile by Planar Induced-Laser Fluorescence (PLIF), as explained in Section 5.3.3.4.  3 11  2  1  5  7  6 8  Drain  4  Drain 10  9  Figure 5.2. Schematic view of pilot-scale photoreactor, which consists of: product reservoir (1), feed reservoir (2), stirrer (3), centrifugal pump (4), flow/pressure/temperature meter (5), photoreactor (6), laser source (7), digital camera (8), PIV/PLIF control unit (9), data acquisition system (10), and online spectrophotometer (11).  The PIV setup employed was the “FlowMap 2D” system from Dantec Dynamics which consists of: software for capturing and processing images; a 12-bit digital camera (“HiSense MKII” from Hamamatsu Photonic K.K.) equipped with a narrow-band filter at 532 nm; and a laser pulse source at 532 nm (“Laser Solo III-15Hz” from NewWave Research Company). PIV measurement criteria [26, 27] were taken into consideration in order to minimize errors during the experiments. Seeding the fine particles (polyamide with diameters of 10-20 µm) causes the reflection of light produced by the pulsed laser  110  source. During the very short lifetime of the pulse (5-10 ns), the displacement of the particle is almost zero and it can be concluded that the scattered light represents the location of the particles captured by the digital camera. After a short period of time, a second image is captured using a second laser pulse. Comparison of the two images (cross correlation) using a proper length scale reveals the length and direction of the velocity vector in the plane of the laser sheet. To reduce the noise, a Gaussian window function with a coefficient of 1 was used. Because window functions do not use the information near the edges of an interrogation area, a 25% overlap (4 pixels) of the interrogation window of 16-by-16 pixels was considered in the calculations. In addition, to broaden the narrow correlation peaks and remove the effect of the neighboring points in the correlation plane, a low-pass Gaussian filter with a coefficient of 1.5 was applied to the frequency domain of the Fourier transform calculation [26]. Using the aforementioned criteria, a 0.25 m length of the reactor (from reactor inlet) was studied using PIV. This zone was divided into three sections. For each section, 400 double images were captured at two different mass flow rates of 0.005±0.002 and 0.014±0.002 kg/s. The values are reported with 95% confidence intervals. The 400 images were processed to reveal the average velocity field. Finally, the velocity fields from the three sections were stitched together to represent the velocity profile inside the reactor (0.25 m length of the reactor). 5.3.3.2 Radiant Energy Measurement  A radiometer (IL1700, from International Light Technologies) and a solar blind UV sensor (SIC01M-C, from Roithner Laser Technik) were used to measure the irradiance rate at different points inside the reactor (where filled with air). The top side of the reactor was removed, and the sensor was inserted into the flow pathway of the reactor 111  as shown in Figure 5.3. The mid-plane inside the reactor was scanned, changing the position of the sensor inside the reactor.  1  10.3 mm  2 3 42 mm 4 5 6 Figure 5.3. Reflector glasses over the UV lamp in the quartz sleeve. The apparatus consists of: two window glasses (1), a UV photodiode extension arm (2), a UV sensor (3), a UV lamp (4), a quartz sleeve (5), and the position of the hole in the sleeve (6). 5.3.3.3 Photoreaction Rate Measurement  The photoreaction rate of RhWT (as a model contaminant) and hydrogen peroxide (as an oxidant) in the presence of UV radiant energy was measured in a customized UV collimated-beam photoreactor under controlled conditions (explained in Chapter 4). The reactor consisted of: two mid-size UV lamps on an aluminum reflector below a beam collimator; a double-jacketed reactor of 250 mL, with a round quartz window (3 mm thickness) at the bottom; and a variable speed stirrer at the top. The photoreactor sample port (bottom) was connected to the flow-through cuvette of a UV-Vis spectrometer (Cary 100) via the circulating pump. Prior to measuring the photoreaction rate of the RhWT, the actinometry solution of iodide/iodate, and the online spectrometer were used to  112  accurately determine the UV radiant power passing through the collimated-beam photoreactor [28]. The concentration of RhWT solution was 126±1 ppb and the hydrogen peroxide concentration was 10.14±0.05 ppm. These were measured as a function of absorbed radiation to find the photoreaction rate. The rate of direct oxidation and photolysis of RhWT without hydrogen peroxide were also measured (as explained in detail in Chapter 4). 5 6 4  Circulating Cooling water  Sampling ports  3  1  2  Figure 5.4. Schematic diagram of bench-scale collimated-beam UV photoreactor: parabolic reflector (1), UV lamp (2), collimator (3), double jacket rector (4), stirrer (5), and thermometer (6).  5.3.3.4 Concentration Profile of RhWT Measurement (PLIF)  Fluorescent chemicals (e.g., RhWT) can re-emit light at different wavelengths (e.g., 580 nm) than the at exciting wavelengths (e.g., 532 nm). The intensity of the reemitted light depends directly on the concentration of the fluorescent chemical. This behavior allows the measurement of the fluorescent chemical concentration profile using the same setup as for PIV. The technique is referred to as Planar Induced-Laser  113  Fluorescence (PLIF). Replacing the narrow-band filter of the camera in the PIV setup with a high-pass filter (>550 nm) can convert the PIV setup to a PLIF setup. Due to the nature of pulsed lasers, the beam energy is not spatially uniform. In addition, the laser energy varies over time in consecutive pulses (±7% as measured in our experiments). To increase the accuracy of the measurements by considering these variations, the laser beam was split into two parts using a beam splitter. The energy of one beam was measured using another 12-bit digital camera with a narrow-band (532 nm) filter from a reflective surface as the reference. The other beam excited the fluorescent solution through the reactor and this was captured using the second camera with a high-pass filter (>550 nm). This technique (described in detail in Chapter 4) provides a way of accounting for the pulse-to-pulse variation in laser energy and, hence, increases the accuracy of the concentration measurements. For the PLIF calibration process, different solutions of RhWT (10-130 ppb) with a constant hydrogen peroxide concentration (9.65±0.56 ppm) were pumped through the reactor, and 400 images were captured for each concentration using both cameras. The value of each pixel on the resulting image, after subtracting the dark base (no laser source), was divided by the total value of reflected light (captured by the second camera) to allow normalization based on the reference beam. Finally, the normalized value for each pixel was plotted in relation to the concentration of RhWT to determine a linear correlation for that pixel. This resulted in a two-dimensional (2D) calibration map (see Section 4.3.3.2.ii). A stock solution of hydrogen peroxide (10.30±0.05 ppm) and RhWT (126±1 ppb) at 21.4±0.3ºC was fed into the reactor while the UV lamp was stabilized. Two different  114  flow rates of 0.006±0.002 and 0.015±0.002 kg/s were tested. For each operating condition, the inlet, middle, and outlet of the reactor were studied, and 400 images were captured for each section. The average of the normalized images, after subtracting the glowing background image, was used to calculate the RhWT concentration profiles inside the flow-through UV photoreactor. This procedure was applied for three zones through the entire length of the reactor (the details of this procedure can be found in Chapter 4).  5.4 CFD Model Setup In modeling the velocity inside the UV photoreactor for various flow rates, different viscous models—laminar, turbulent (one-equation, Spalart-Allmaras, Realizable k- ε, and Standard k-ω for transitional flow) and mixed (different zones, different models, based on velocity)—were tested. Due to the low concentration of chemical contaminants, the physical properties of water were used to specify the coefficients of the governing equations. A no-slip boundary condition with enhanced wall function was selected for internal surfaces. The numerical solution was applied to solve the equation by discretizing the computational domain from 46,000 to 560,000 structured hexahedral cells. In addition, different discretization methods, first, second, and QUICK, were tested to check the sensitivity of the solutions. The STANDARD and SIMPLE schemes were selected for pressure and pressure velocity coupling, respectively. Fluent 6.2.18 software was employed to solve the conservation of mass and momentum equations numerically. The coefficient of determination (or R-square) of the total velocity in the computational domain, calculated with 136,000 and 300,000 cells was greater than 0.975 (in all cases), implying that 136,000 cells was adequate to achieve a mesh-independent solution. The  115  results of the simulations were compared using the experimental data from PIV measurements. Both second- and first-order discretization methods were used to solve the species conservation equation. It was observed that there was no significant differences in the predictions of the two methods. As a result, for all simulations, first-order discretization was used. The sink term in the equation was measured in the bench-scale photoreactor, described in the next section. In addition, the turbulent Schmidt number was fixed at 0.7 to calculate the turbulent diffusion coefficient for the transitional and turbulent flow regimes. The molecular diffusion coefficient of RhWT reported by Shilton at ambient temperature (3.6×10-9 m2/s) was applied [29].  5.5 Results and Discussion 5.5.1 Evaluation of the Hydrodynamic Model 5.5.1.1 Velocity Field Measurement (PIV)  The value of R2 (the square of the correlation coefficient between two observed data values) for the average of 400 and 350 images for each of the two flow rates was close to 1, suggesting that 400 images were statistically sufficient for averaging. The results of the PIV measurement for the two flow rates for the entrance region of the reactor are shown in Figure 5.5. For each image, the length of the arrows is proportional to the velocity magnitude. The flow pattern within the UV reactor begins at the inlet jet and passes over the hemisphere of the sleeve in the inlet section where it diverts to the main section of the reactor. Due to the high gradient of momentum in different directions of the jet stream, a back-circulation of the fluid at the inlet zone is produced causing recycle of fluid near the  116  top of the reactor.  Higher inlet velocities (with higher gradients) result in greater  vorticity and flow rate of the recycled flow and, consequently, reduce the path height of the main fluid at the bottom of the reactor (close to sleeve). Figure 5.6 shows the contours of the turning points (on the vertical lines where velocity is equal to zero). This clearly shows the difference in the height of the high velocity zone near the bottom of the reactor for the two different flow rates.  25 cm Inlet Zone  Inlet  Inlet  Main Zone  A  B  Velocity (m/s) 0.05 0.04 0.03 0.02 0.01 0.00 0.15 0.12 0.09 0.06 0.03 0.00  Figure 5.5. PIV results at mid cross-section of the reactor for two different mass flow rates: 0.005 (A) and 0.014 (B) kg/s. Due to the light scattering and reflection, unrealistic velocity vectors were generated from some parts of the reactor. These have been removed (blank areas).  117  0.0  5.5  7.0  9.0  12.0  16.0  20.0  x  Scale Velocity A: 0.018 m/s B: 0.080 m/s  Inlet  Negative Positive Velocity Velocity  Figure 5.6. Velocity (x-component) profile and contour of turning points at different sections of the reactor for two different flow rates. Vertical lines indicate different x-positions in cm. 5.5.1.2 Simulation of Hydrodynamics  The average Reynolds numbers corresponding to the flow rates over the entire cross-section at the round inlet of the photoreactor (Figure 5.1) are 530±230 and 1480±280 for cases A and B, respectively. Due to variations in the velocity at each crosssection, the local Reynolds number differs on any vertical line (from top to bottom) within the reactor, but the maximum Reynolds number cannot exceed the inlet average Reynolds number. As a result, in case A, the flow regime is laminar throughout the entire reactor length; in case B, however, the area close to the bottom of the reactor (near the sleeve) is in the transition regime, whereas the area far from the sleeve is laminar. The aforementioned observations were modeled using different viscous models (laminar, transitional, and turbulent) with different numerical approaches. The CFD results from laminar and standard k-ω transitional flow simulations using the first-order discretization method yielded the best fit of the PIV data for cases A and B, respectively (Figure 5.7). Overall, the flow patterns were well predicted by the model.  118  A  Inlet  B  Inlet  Velocity (m/s) 0.05 0.04 0.03 0.02 0.01 0.00 0.15 0.12 0.09 0.06 0.03 0.00  Figure 5.7. Simulated velocity profile (CFD results) at mid cross-section of reactor for two different flow rates: 0.005 (A) and 0.014 (B) kg/s.  In order to show the deviation of the simulation results from the experimental measurements (PIV), the velocity vectors of the simulation results were subtracted from the PIV measurement for each corresponding point for the two different cases (Figure 5.8). Considering the scale of the images, the error corresponding to the PIV measurements, and the instrumental errors, there is a good correspondence for both cases, although with better agreement for case B. In the inlet zone, none of the models could predict the high circulation flow accurately. In addition, there is a narrow band below the contour of the turning points for which both models predicted a smaller velocity field than was found experimentally. The overall degree of agreement can be evaluated using the coefficient of determination (R2) between the PIV velocity vectors and those of simulation, which is expressed as: R2 = 1−  ∑ (Vx − Vx ∑ (Vx − Vx PIV  PIV  SIM  PIV , Ave  ) 2 + (Vy PIV − Vy SIM ) 2 ) 2 + (Vy PIV − Vy PIV , Ave ) 2  5.5  119  where Vx and Vy are the x and y components of the velocity vector, respectively, and indices PIV, SIM, and Ave represent the PIV measurements, simulation results, and the average of the measured values (PIV) throughout the entire domain, respectively. The Rsquares for cases A and B are 0.86 and 0.92, respectively (a perfect match would result in R2 = 1.0). Better statistical agreement was found between the simulation results and experimental values for case B. It is expected that the simulation and experimental concentration profiles, which are highly affected by the velocity field, will show better agreement in case B, as well.  Inlet  A, Scale  = 0.015 m/s 0.015 m/s m/s  0.0 0  Inlet  B, Scale  = 0.03 m/s 0.03 m/s m/s  0.0 0  Figure 5.8.  Velocity difference (vectors), VPIV -VSIM, and x-components of velocity  reactor. Vertical lines indicate x-position and curves show zero axial velocity difference (VxPIV-VxSIM). The color bar (linear scale) indicates the velocity difference (vectors) from 0.015 (red) and 0.03 (red) to 0 m/s (blue) for cases A and B, respectively. The arrows show the flow directions. The velocity measurements near the surfaces are removed (white regions) because of the high uncertainty in the measurement due to the reflections.  120  5.5.2 Evaluation of the Radiation Model  Considering the UV lamp as a line or volume source, the UV radiant emission from the source can be modeled using measured boundary conditions (see Chapter 2). It is important to note that the UV lamp does not behave as a uniform emission source because the moving air (used in this experiment as a coolant in the gap of the sleeve and lamp to maintain ambient temperature on the outer sleeve surface) changes the temperature gradient along the side of the lamp plasma dramatically. This causes a nonuniform emission that is lower closer to the lamp tip, where air enters the gap. As a result, radiant power should be measured close to the surface of the lamp/sleeve to calculate the intensity of the UV lamp at the source (boundary conditions). In addition, for a more realistic model, reflection from the body of the reactor should be integrated into the model. A comparison of the experimental and modeling results for the irradiance rate inside the air-filled reactor (absorption coefficient is equal to zero) is shown in Figure 5.9.  121  Measured/Calculated Power (a.u.)  1.5E-06 1mm  1.0E-06 10mm 20mm 30mm  5.0E-07  50mm  0.0E+00 -5  5  15  25  35  45  Horizontal distance of sensor from tip of lamp filament (cm) Figure 5.9. Irradiance rate at different distances indicated by  ,  ,  , , and  for  distances 1, 10, 20, 30, and 50 mm from the surface of the lamp, respectively. Experimental values for the radiation power (symbols) vs. the 1D modelpredicted values (continuous lines) and the 3D model, cylindrical emitter with 7 mm diameter, values (dashed lines).  Modeling the radiation (considering refraction and reflection from the reactor walls), with the radiant source as a volumetric or linear emitter provided a reasonable agreement with the experimental results. However, the results from the linear source model were more satisfactory over the various distances from the lamp. As a result, the line source (1D) model was selected for modeling the radiation field inside the reactor, taking into account the absorption coefficient of RhWT and hydrogen peroxide (4.48 m-1) , the reflection/refraction from/through the sleeve and the lamp quartz body, as well as the reflection from the side walls of the reactor. The simulated fluence rate profiles inside the photoreactor are shown in Figure 5.10. In the areas about 3 cm above the lamp sleeve surface, the fluence rate is six times less than in the area close to the surface of the sleeve.  122  Outlet  Inlet  0.0  6.8  13.5  20.3  27.0  33.8 (W/m2)  Figure 5.10. Fluence rate (W/m2) profile at the mid-cross-section of the photoreactor. 5.5.3 Kinetic Model Determination  The total rate of reaction of the rhodamine WT with hydrogen peroxide in the presence of UV is the summation of the three primary rates of direct oxidation, direct photolysis, and reaction with hydroxyl radicals. The rate of direct oxidation and direct photolysis were found to be insignificant (see Chapter 4, Section 4.3.2). The total photoinitiated oxidation rate of RhWT and hydrogen peroxide after receiving H J/m2 UV fluence was measured (using the bench scale collimated-beam batch photoreactor discussed in Sections 4.3.2 and 4.4 in detail) to be: C  ln  RhWT  = (− 4.757 ± 0.092) × 10 −3 H  130   5.6  where CRhWT is the concentration of RhWT after receiving H (J/m2) fluence. This correlation was developed for the concentration of RhWT of less than 130 ppb (initial concentration of RhWT) and hydrogen peroxide at 10 ppm. Considering the volume of the bench-scale collimated-beam photoreactor and the agitation effect of the impeller, the photoreactor is a well-mixed batch reactor, and the rate of reaction for a batch reactor can be written as: dC RhWT = (− 4.757 ± 0.092 ) × 10 −3 Gabs C RhWT dt  5.7  where Gabs is the absorbed fluence rate.  123  Considering the temperature of reaction rate tests during the experimental measurements (21±1 ˚C), Equation 5.7 is valid only for ambient temperature. Passing cooling air over UV lamp in the PLIF measurements maintained temperature of reaction zones at 21±1 ˚C. During all tests, the skin temperature of sleeve while UV lamp was operated, did not exceed than 24 ˚C, and, hence, the isothermal assumption using Equation 5.7 is valid. 5.5.4 Evaluation of the Integrated Model  The concentration profile of RhWT is the result of the interaction of hydrodynamics, fluence rate, and reaction rate. Therefore, measuring the concentration profile throughout the entire reactor is the best indicator for the evaluation of the photoreactor model performance. As far as the author is aware, no experimental measurement of the concentration profile in a photoreactor has been reported in the open literature. Figure 5.11 shows the measured concentration of the RhWT profile through the mid-cross-section of the photoreactor. The concentration profile through the entire length of the reactor was studied by investigating three sections of the reactor from inlet to outlet separately. The three images (averaged from 400 images for each zone) were combined and stitched together. Due to the technical limitations in keeping the setup perfectly consistent while studying different zones, it was not possible to obtain a perfect match in the areas where two adjacent images overlapped. The values close to the boundaries of the reactor walls were masked because they did not represent the true concentrations (as a consequence of reflections from the white glue joints in these regions of the reactor). For all images, a horizontal median filter of rank 40 was applied to remove the shadowing effects of the laser sheet in the images.  124  130 ppb  A 1  B 2  80 ppb  Figure 5.11. Concentration profile of rhodamine WT (ppb) in the UV reactor with the laser energy at a steady-state condition for two different mass flow rates: 0.006 (A) and 0.015 (B) kg/s.  Radiation measurement in the inlet zone (-5 to 1 cm from the tip of the lamp) showed a very low level of radiant energy (almost zero). This implies that the rate of conversion of RhWT was very low. In other words, higher concentrations of RhWT are expected in this zone, in contrast to what was observed experimentally (Figure 5.11). This can be explained by considering the flow pattern in the reactor (Figure 5.5), which plays a major role in controlling the concentration. Low velocity (flow) of fluid at the reactor inlet zone increases the residence time of chemicals and consequently enhances the conversion consequently. In addition, the recycled flow of low concentration fluid from downstream to upstream dilutes the concentration of the inlet stream. As a result, lower concentrations levels of RhWT are expected at the inlet zone. In the area closer to the lamp (sleeve) where the fluence rate is at its maximum level, a minimum concentration of RhWT should be observed. These features are clearly demonstrated in Figure 5.11.  125  The integrated model of reactor performance simulated the concentration profile in the UV reactor by solving the governing equations of mass, momentum, radiant energy, and species conservation. The results are shown in Figure 5.12.  Figure 5.12. Concentration profile of rhodamine WT in the UV photoreactor calculated by the integrated model for two different mass flow rates: 0.006 (A) and 0.015 (B) kg/s. Concentrations less than 80 ppb are shown as white.  The difference between the modeling and experimental results of RhWT concentration for the two flow rates (cases A and B) are shown in Figure 5.13.  30 ppb  A  B Figure 5.13. Concentration difference (CPLIF  -5 ppb - CSimulation) in ppb for two different mass  flow rates: 0.006 (A) and 0.015 (B) kg/s. Concentration differences higher than 30 ppb are shown as white.  The reliability of the simulation results (Figure 5.13) should be compared with the uncertainty associated with the calibration map for calculating the concentration contours in the PLIF procedure (Section 4.5) as shown in Figure 5.14.  126  ±30 ppb  A  B 0 ppb Figure 5.14. Uncertainty in the RhWT concentration profile measurement (ppb) with a  95% confidence interval in the UV reactor for two different mass flow rates: 0.006 (A) and 0.015 (B) kg/s. The color bar (linear scale) indicates the concentration uncertainty from ±30 (red) to 0 ppb (blue), and the arrows show the flow directions.  Overall, there is close agreement between the modeling and experimental results throughout the entire cross-section of the reactor. The differences between the model predictions and the experimental results are in the range of measurement uncertainty. The agreement is better for case B, likely due to a better prediction of the velocity profile. For both cases, in a small zone at the end of the reactor, adjacent to the UV lamp, the difference in concentration shows a notable deviation. In that zone, the significant vertical concentration gradient causes an error in the PLIF measurement (as explained in Chapter 4). Overall, the disagreement between the results can be attributed to the uncertainty associated with experiments, such as instrumental error and error of the PLIF method, as well as the errors associated with numerical methods and model parameters.  5.6 Source of Uncertainty and Errors There are several errors associated with the optical diagnostic measurement tools, PIV and PLIF. Some of these are related to the digital camera, such as potential errors 127  due to the effects of photon noise, thermal noise, on-chip electronic noise, amplifier noise, quantization noise, and shading. Using a scientific camera helped to reduce all of these noise distortions but could not eliminate them. Further sources of error for the PLIF measurements arises from surrounding lights, re-emission rays, the scattering of rays by bubbles and particles, reflection from walls, light absorption by medium, the calibration curve (map), and different positions of the camera during calibration procedures and concentration measurements. These optical errors and concentration uncertainties associated with the calibration map are explained in detail in Section 4.5. For the PIV measurements, the major sources of errors are in-plane and out-of plane motions, and the scattering of rays by bubbles and particles. The errors were minimized as far as possible based on the reported criteria [25, 26]. Simplifying assumptions for calculating velocity and concentrations under steady-state conditions as sources of uncertainty are discussed in the following sections. 5.6.1 Uncertainty in Velocity Field Measurements  In experimental work, it is extremely difficult to attain ideal steady-state conditions for low velocity phenomena because of the nature of the mechanical components of the system. Any vibrations in the experimental equipment, fluctuations in electricity, swinging of flexible joints, etc., create variations in the operating parameters. The velocity (unlike the concentration) is particularly affected by the aforementioned fluctuations at low velocity conditions. The common practical engineering approach, (especially in turbulence studies, where fluctuations in velocity are an inherent element of the system) is to capture the velocity field over a long period and then use the average value of the parameter as representative of steady-state conditions. The standard  128  deviation of the velocity is the best indication of how far velocity is from a steady-state condition. Figure 5.15 shows the standard deviation of velocity for the two cases studied, A and B. 0.025 m/s  A  B  0.00 0.05 m/s  0.00  Figure 5.15. Standard deviation of velocity (m/s) profiles for two different mass flow rates, 0.005 (A) and 0.014 (B) kg/s. Arrows show the direction of flow. The calculated average measured velocity using PIV (all measured value divided by the number of data) for cases A and B are 0.023 and 0.053 m/s, respectively.  The areas with high values of standard deviation (dark red color) happen because of errors in measurement (e.g. in/out particles into the captured image for PIV) and fluctuation of the flow rates (± 0.002 kg/s) for all cases. In order to evaluate the steadystate assumption, standard deviation should be considered with average velocity (Figure 5.5) for each point. Normalized average standard deviations (average standard deviation divided by average velocity) are 0.46 and 0.27 for flow rates 0.006± 0.002 and 0.015± 0.002 kg/s, respectively. The sources of deviation form steady state are believed to be the  129  limitations associated with the experimental measurements. As a result, the steady state assumption was considered for solving the mass and momentum conservation equation. Considering standard deviations and other sources of uncertainty in the measurement, the absolute velocity difference, as measured by PIV minus that calculated by simulation ( VPIV − VSIM ) as shown in Figure 5.17, is located within the range of standard deviation (Figure 5.15), and again shows good agreement between the simulation results and the PIV measurements. 0.05 m/s  A  B  0.00 0.1 m/s  0.00  Figure 5.17. Velocity difference contour (m/s) for two different mass flow rates, 0.005 (A) and 0.014 (B) kg/s. Arrows show flow direction.  5.6.2 Uncertainty in the Concentration Field Measurements  The effects of the optical density of the medium on the errors produced and on uncertainty in the concentration calculations (calibration map) are discussed in detail in Section 4.5. Due to the nature of the equipment, it was not possible to eliminate the fluctuations in flow. The standard deviation for concentration was calculated, and is shown in Figure 5.17. The standard error (standard deviation divided by the square root  130  of number of images, where the number of images is 400 for each case) is less than 1 ppb for all areas in the two cases. This very low standard error, considering the concentration measurement uncertainties (Figure 5.14), proves the statistical reliability of the averaging procedure for calculating the concentration profile at steady state. 0.6 ppb  A  B 0 ppb Figure 5.17. Standard error of concentration profile (ppb) for two different mass flow rates, 0.006 (A) and 0.015 (B) kg/s. Arrows show flow direction.  5.7 Conclusions Conventional methods for evaluating photoreactor models typically rely exclusively on concentration measurements at the reactor inlet and outlet. These methods cannot show the discrepancies between the model predictions and real values inside the photoreactor, nor can they reveal the causes of the deviation. The methods utilized in this research can be applied to evaluate photoreactor models, as well as models of similar systems. This approach evaluates the accountability of each component of the integrated model as well as the results of the integrated model for the entire computational domain. Considering this method, each component of the model can be adjusted properly according to the operational parameters of the system under study, to predict the performance of the system with minimum uncertainty. As a result, the discrepancy for  131  each component is revealed, and the model can be applied to a wide range of operating conditions. This approach reduces uncertainty in the model setup and provides a solution for individual phenomena. Consequently, it decreases the bias of the final integrated model solution. Considering the uncertainties in the measurements of velocity, fluence rate, and concentration, overall, the favorable agreement between the experimental data and the simulated results for each governing equation (momentum, mass, and radiant energy), and the integrated system of equations (species mass conservation) verifies the reliability of the presented methodology.  132  5.8 References 1- Kidman, R. B., and Tsuji, K. S. (LA-12221-MS) Preliminary Costs Comparison of  Advanced Oxidation Processes, Los Alamos National Laboratory Unofficial Copy, SJB 1 -21-91. 2- Gunten U. V., 2003, Ozonation of drinking water: Part I. Oxidation Kinetics and  Product Formation, Water Research, 37, 1443–1467. 3- Castrillo, S. R. V., and Lasa H. I., 2007. Performance Evaluation of Photocatalytic  Reactors for Air Purification Using Computational Fluid Dynamics (CFD), Industrial & engineering chemistry research, 46, 18, 5867-5880. 4- Munoz, A., Craik, S., and Kresta, S., 2007. Computational Fluid Dynamics for  Predicting Performance of Ultraviolet Disinfection Sensitivity to Particle Tracking Inputs, Journal of Environmental Engineering Science, 6, 285–301. 5- Ducoste1, J. J., Liu, D., and Linden, K., 2005. Alternative Approaches to Modeling  Fluence Distribution and Microbial Inactivation in Ultraviolet Reactors: Lagrangian versus Eulerian, Journal of Environmental Engineering, ASCE 131, 10, 1393–1403. 6- Kamimura, M., Furukawa, S., and Hirotsuji, J., 2002. Development of a Simulator for  Ozone/UV Reactor Based on CFD Analysis, Water Science and Technology, 46, 11– 12, pp 13–19, IWA Publishing. 7- Janex, M. L., Savoye, P., Do-Quang, Z., Blatchley E., and Laine J. M., 1998, Impact of  Water Quality and Reactor Hydrodynamics on Wastewater Disinfection by UV, Use of CFD Modeling for Performance Optimization, Water Science Technology, 38, 6, 71-78. 8- Sozzi, A., and Taghipour, F., 2007. The Importance of Hydrodynamics in UV  Advanced Oxidation Reactors, Water Science & Technology, 55, 12, 53–58.  133  9- Liu, D,Wu, C, Linden, K, and Ducoste, J, 2007. Numerical Simulation of UV  Disinfection Reactors: Evaluation of Alternative Turbulence Models. Applied Mathematical Modeling, 31, 1753–1769. 10- Ranade, V. V., 2002. Computational Flow Modeling for Chemical Reactor  Engineering. Academic Press. First edition. London. England. 11- Prandtl, L., 1925. Z.Angew. Math. Mech., 5,136. 12- Spalart, P., and Allmaras, S., 1992. A One-Equation Turbulence Model for  Aerodynamic Flows. Technical Report AIAA-92-0439, American Institute of Aeronautics and Astronautics. 13- Launder, B. E., and Spalding, D. B., 1972. Lectures in Mathematical Models of  Turbulence. Academic Press, London, England. 14- Yakhot, V., and Orszag, S. A., 1986. Renormalization Group Analysis of Turbulence,  I. Basic Theory, J. Sci. Comput. 1, 1. 15- Shih, T. H., Liou, W. W., Shabbir, A., Yang. Z., and Zhu, J., 1995. A New k- ε Eddy-  Viscosity Model for High Reynolds Number Turbulent Flows - Model Development and Validation. Computers Fluids 24, 227-238. 16- Wilcox, D. C., 1998. Turbulence Modeling for CFD. DCW Industries, Inc., La  Canada, California. 17- Launder, B. E., 1989. Second-Moment Closure: Present... and Future?. Inter. J. Heat  Fluid Flow 10, 282-300. 18- Gibson, M. M., and Launder, B. E., 1978. Ground Effects on Pressure fluctuations in  the Atmospheric Boundary Layer. J. Fluid Mech. 86, 491-511. 19- Launder, B. E., Reece, G. J., and Rodi, W., 1975. Progress in the Development of a  Reynolds-Stress Turbulence Closure. J. Fluid Mech. 68, 537-566.  134  20- Romero R. L., Alfano O. M., and Cassano A. E., 1997. Cylindrical Photocatalytic  Reactors. Radiation Absorption and Scattering Effects Produced by Suspended Fine Particles in an Annular Space. Ind. Eng. Chem. Res., 36, 3094-3109. 21- Elyasi, S., Taghipour, F., 2005. Simulation of UV Photoreactor for Water  Disinfection in Eulerian Framework. Chemical Engineering Science, 61, 14, 47414749. 22- Oppenlander, T., 2003. Photochemical Purification of Water and Air, Wiley-VCH. 23- National Institue of Standard and Technology, Chemical Kinetics Database,  http://kinetics.nist.gov/kinetics/index.jsp, retrieved May 2008. 24- Klassen, N. V., Marchington, D., and McGowan, C. E., 1994. H2O2 Determination by  I3- Method and by KMnO4 Titration. Analytical Chemistry, 66, 2921-2925. 25- Melton, L. A., and Lipp, C. W., 2003. Criteria for Quantitative PLIF Experiments  Using High-Power Lasers. Experiments In Fluid 35, 310-316. 26- FlowMap Manual, 2004, Particle Image Velocimetry Instrumentation, Installation &  User’s guide, Dantec Dynamics. 27- Willert, C. E., and Gharib, M., 1991. Digital Particle Image Velocimetry.  Experiments in Fluids, 10, 181-193. 28- Rahn, R. O., Stefan, M. I., Bolton, J. R., and Goren, E., 2003. Quantum Yield of the  Iodide-Iodate Chemical Actinometer: Dependence on Wavelength and Concentration. Photochemistry and Photobiology, 78, 146-152. 29- Shilton, A., 2001. Studies into the Hydraulics of Waste Stabilisation Ponds. Doctor of  Philosophy, Massey University, New Zealand.  135  Chapter 6. Conclusions and Recommendations 6.1 Conclusions of the Research This study demonstrates a systematic simulation/modeling of a UV photoreactor for microbial disinfection or organic contaminant degradation using a UV-based advanced oxidation process. A semi-mechanistic approach is taken for the model development, using a numerical method to solve a system of governing partial and ordinary differential equations. The governing equations can be categorized into four types: mass; momentum; radiant energy; and mass of species conservation equations. There are some empirical or semi-empirical terms in the governing equations, (e.g., boundary conditions for the UV lamp and UV source emitter). As a result, the output of each equation should be evaluated against experimental data in order to determine the accuracy of the model, as well as its solution for the geometry under study and the operating conditions. The aforementioned methodology was used for the modeling and model evaluation to present a reliable and practical approach for the simulation of UV photoreactors. This approach evaluates the accountability of each component of the integrated model, as well as the results of the integrated model. A new method developed for comparing the results of the velocity field for the entire computational domain involved using the CFD model and those measurements using PIV. It clearly revealed the deviation between the simulated data and the experimental results throughout the entire domain of the studied zone. The results of particle image velocimetry (PIV) for the laminar flow (0.005 kg/s) showed, overall, a 86% (R2 equal to 0.86) match between the model and the experimental data. For transitional flow (0.014 kg/s), the k-ω model with transient assumption showed better  136  agreement with R2 equal to 0.92, compared to the other turbulence models. Considering the instrumental bias and numerical error, it can be concluded that the semi-mechanistic model was capable of predicting the velocity field with a relatively high accuracy. This subject is explained in Chapter 5. Considering the refraction and reflection phenomena, a general formula was derived for radiant energy distribution.  The formula was solved assuming various  geometries for the radiation source (e.g., 1D, 2D, and 3D) using a non-uniform assumption for the radiant source as the boundary condition (UV lamp), unlike the conventional approach. The modeling result for each type of radiation source was evaluated against experimental data. The experimental data were generated from the direct measurement of the UV radiant field for a specific UV lamp using a photodiode and radiometer. Considering the experimental assessment, it was proved that the line (1D) and volumetric (3D) emission for a low-pressure UV lamp produced acceptable results, if reflection from sidewalls and the refraction through each medium were taken into account (Chapter 2). As a result, it seems that: 1) there is no universal model for modeling all type of UV lamps; 2) considering refraction improves the model prediction; and 3) applying boundary conditions based on measurements produced more realistic results, especially near the lamp surface. The final step for modeling a UV reactor is solving the species mass conservation equation. This equation includes volumetric rate of consumption/production, molecular diffusion, and mass convection of species. The volumetric rate is a direct function of radiant energy distribution and concentration of species. The rates can be measured in a lab-scale photoreactor under controlled conditions. A novel powerful non-intrusive visual  137  diagnostic tool, the modified planar laser-induced fluorescence method, was developed in order to measure the concentration profile through a cross-section of the reactor (Chapter 3). Using a 2D calibration method and employing a second camera improved the accuracy of the method.  The results from the model and experimental measurement  showed that the accuracy of the prediction of the concentration is highly dependent on the accuracy of the simulated results for the velocity field. Overall, the simulation of the concentration profile predicted acceptable results. It can be concluded that the mechanistic/semi-mechanistic Eulerian approach described here for modeling the performance of an industrial UV photoreactor is a practical and acceptable methodology of evaluating the UV reactor performance simulation for microbial disinfection and chemical contaminant removal (UV-initiated oxidation), if parameters and assumptions are accurately chosen considering the governing criteria [e.g. 1-5].  6.2 Significance of the Research The outcome of this research enhances our understanding of the performance of commercial UV photoreactors for the disinfection of pathogenic microorganisms or the oxidation of persistent chemicals using photo-initiated oxidation. The systematic approach presented for modeling the UV reactor can also be applied for modeling similar systems. The outcome can be categorized into theoretical and experimental parts. 6.2.1 Theoretical Achievements  The main objective of this research focused on deriving and using a general model for the simulation of UV reactors. This general approach enables us to model and  138  simulate the performance of commercial UV reactors for a wide variety of applications if the limitation of each governing equation is taken into consideration. 6.2.1.1 Radiation Modeling  A general formula is derived and presented in Chapter 2. The formula can model the distribution of radiant energy for each wavelength throughout the reactor domain, considering the optical properties of the media. The formula is directly applicable for modeling the radiant energy distribution in a clear medium, without any particles. This approach can be expanded to a system with particles, if each particle is considered as a radiation source using an iterative numerical process. 6.2.1.2 Microbial Disinfection Modeling with the Eulerian Approach  In the Eulerian approach of modeling, any type of chemical or microorganism can be treated as a chemical species and the received fluence can be calculated for each control volume inside the reactor domain. This method can be used if there is an acceptable mathematical method for deriving the local volumetric inactivation rate. In this research, this method is derived and presented for microbial disinfection modeling. 6.2.2 Experimental Works Accomplishments 6.2.2.1 Oxidation Rate of Rhodamine WT using a UV-based AOP  Rhodamine WT (RhWT) was selected as a model organic contaminant, due to its unique chemical characteristics to measure the performance of UV photoreactors. A customized bench-scale UV photoreactor was built in order to measure: 1) the direct oxidation of RhWT in the presence of hydrogen peroxide; 2) the photolysis of RhWT; and 3) the oxidation of RhWT with the hydroxyl radical. A correlation is presented for a low concentration of oxidant (hydrogen peroxide) and RhWT. The presented method is a  139  general approach that can be used to measure the rate of removal of any chemical in the water stream considering real operational conditions in a UV-based AOP. 6.2.2.2 Concentration Profile Measurement  The typical method for evaluating the performance of any type of reactor is measuring the conversion rate of reactants based on the concentration of species at the outlet/inlet of the reactor. However, this method shows the overall performance of the reactor without providing any information about the local performance inside the reactor. In this research, a novel method has been developed that enables us to measure the concentration profile (RhWT in this case) inside a UV photoreactor. The result of the measurement directly shows the local concentration at each point inside the UV reactor. The concentration profile reveals the local deficiency, the local high/low volumetric reaction rate, and the local effect of hydrodynamics on the performance of the UV reactor.  6.3 Limitations The procedures presented in the theoretical and experimental parts of this thesis can be used if the technical limitations are taken into consideration. The limitations are classified as follows. 6.3.1 Theoretical Part – Radiation  A number of simplifying assumptions were considered in the radiation distribution correlation. Among these simplifications, the constant absorption coefficient assumption is only valid for a low concentration of chemicals and non-reactive media, or if the variation in the concentration is low (received low fluence and low conversion  140  efficiency) in the reactor. In an industrial photoreactor, this assumption is not valid and the absorption coefficient should be considered on each ray track. If a medium contains particles with semi-elastic reflection (i.e., scattering is dominant), the simplified equations are not valid for modeling the propagation of radiant energy. Considering this situation, each element in the medium behaves like a UV source, and conservation of radiant energy should be solved separately for each source using an iterative numerical method. 6.3.2 Experimental Part 6.3.2.1 Hydrodynamics  The most convenient way to compare the CFD simulation prediction with the experimental results from PIV is at a steady-state condition. Equipment and setup should be prepared in such a way that steady-state conditions are easily achievable. Considering the technical limitations of the equipment, this goal was not fully met, and statistical study was performed (averaging). 6.3.2.2 Reaction Rate of Oxidation of RhWT  The measured and presented correlation of the reaction rate of RhWT in the photo-initiated oxidation is valid for low concentrations of RhWT (less than 130 ppb) and at fixed concentrations of hydrogen peroxide (10 ppm). The correlation is not valid for reaction rates beyond these criteria.  6.4 Recommendations for Future Work Some of the techniques presented in this research have been developed for the first time. There are many opportunities for improving them in order to present mature  141  tools for research in related industries. From a theoretical and experimental point of view, these opportunities are: 6.4.1 Theoretical Work 6.4.1.1 Hydrodynamic  Due to technical limitations, all of the tests were carried out at low flow rates (laminar and transitional). Industrial UV reactors operate at a higher velocity, particularly in the turbulence regime. It was observed that the k-ω model has the potential to predict the velocity profiles, in particular where an inlet behaves like a jet. It is recommended that the k-ω model be evaluated at higher Reynolds numbers. 6.4.1.2 Radiation  The suggested formula for modeling radiant distribution has great potential to be applied to model any type of radiant source in different media. In addition, this method uses minimal computational resources, as compared to other radiation models, such as the Discrete Ordinate (DO) model. Using accurate optical parameters considerably increases the level of confidence in the predictive capability of the model. The following procedure is suggested to provide a method accurately predicting the radiation distribution: - The proposed radiation model should be expanded for modeling radiation distribution in a medium containing particles, considering the effect of absorption and scattering of the light by particles. Particles behave as sources of UV radiant energy in the medium. As a result, the model should be solved for each control volume in the medium containing particles. This evaluation will make the model widely applicable for any type of UV photoreactors including photo-catalytic type;  142  - The shadowing effect of the lamps (sleeves) should be studied and modeled; - The amount of reflection from glass surfaces (quartz and window glass) or reactor body should be measured experimentally and inserted into the formula; - The simplification for the constant absorption coefficient should be removed and a quick numerical algorithm should be developed to calculate the average absorption coefficient on the track of each ray. This improvement will make the formula and the method widely applicable for industrial UV reactors working with clear media. 6.4.2 Experimental Work 6.4.2.1 Hydrodynamic  PIV velocity measurement should be performed under precise controlled conditions in order to provide accurate data for evaluating the modeling results. The setup should be able to control all the physical properties of the fluid accurately. As a result, a designated accurate instrument and equipment should be utilized, particularly for hydrodynamic tests. 6.4.2.2 Reaction Rate  There is a considerable potential for RhWT to be used for evaluating the performance of commercial UV reactors. A correlation that shows the rate of oxidation (deterioration) of RhWT in photo-initiated oxidation for a wide range (UV-based AOP) of RhWT and hydrogen peroxide concentrations would help scientists and engineers use this method. The correlation should cover a wide range of concentrations. Due to high radiant energy in the sleeve, normally the temperature of the skin of a sleeve is higher than the surrounding environment. Therefore, the isothermal oxidation  143  rate of RhWT near the lamp is not valid. In order to solve this problem, the derived correlation should be able to predict the rate of reaction for a wider range of operations. The skin temperature of the sleeve (larger lamp) in stagnant air is 55°C. The flow of fluid over the sleeve improves the heat transfer coefficient considerably; the acceptable range is 20-45°C. 6.4.2.3 Planar Laser-Induced Fluorescent (PLIF)  The presented modified method for PLIF is the starting point for improving this scientific diagnostic tool for studying flow patterns and the UV-based oxidation rates. There are several ways to improve the method, as follows: - A considerable limitation of the method is the dependency on the position of the object, camera and laser source. In order to resolve these limitations, all simplifying assumptions for deriving the equation (Equation 4.2) should be ignored. In other words, the derived equation (Equation 4.2) should be solved numerically considering all parameters in the terms of Kset up and Koffset (setup and offset coefficient). As a result, the method will be a function of distances and camera and laser operating parameters; - The effect of the absorption of laser and re-emitted light by an excited medium is an important factor that limits the dimensions of the object under study (reactor). To resolve this issue, the optical density of the medium should not be simplified (not a constant) in Equations 4.1, 4.2. Using the ray-tracing method, the same methodology applied in the radiation modeling (Chapter 2) can resolve this limitation. An effective algorithm will numerically solve the equations without simplifying the terms of the absorption coefficient. This method can solve the change in concentration alongside the laser rays from the laser source; 144  - The ray tracing method should also be applied to the rays leaving the exited medium (at 588 nm), but the chemical concentration through the layer through which re-emitted light is traveling is unknown (unknown absorption coefficient). Technically, this problem can be solved if this space (pathway of re-emitted light) can be scanned from the lowest distance between the camera and the laser sheet to the highest distance. The result from the first laser sheet (lowest distance) ignoring absorption at 588 nm (because of the low pathway), can be used to calculate the unknown concentration profile on this laser sheet. The results from the first profile can be used sequentially to calculate the next laser sheets, apart from the camera. This method requires a high-speed scanner. The scanning speed depends on the velocity profile.  145  6.5 References 1- Munoz, A., Craik, S., and Kresta, K., 2007. Computational Fluid Dynamics for  Predicting Performance of Ultraviolet Disinfection -Sensitivity to Particle Tracking Outputs. Journal of Environmental Engineering Science, 6, 285-301. 2- Elyasi, S., and Taghipour, F., 2005. Simulation of a UV Photoreactor in the Eulerian  Framework Governing by Complex Deactivation Rate of Microorganisms. Third International Congress on Ultraviolet Technologies, Whistler, Canada. 3- Connor, K., Martin, C. O., and Jensen, J. N., 2004. Evaluation of Ultraviolet (UV) Radiation Disinfection Technologies for Waste Water Treatment Plant Effluent. New York State Energy and Development Authority NYSERDA, Report 04-07. 4- Munter, R., Preis, S., Kallas, J., Trapido, M. and Veressinina, Y, 2001. Advanced Oxidation Processes (AOPs): Water Treatment Technology for the Twenty-First Century. Kemia-Kemi, 28, 354-362. 5- Andreozzi, R., Caprio, V., Insola, A., and Marotta, R., 1999. Advanced Oxidation Processes (AOP) for Water Purification and Recovery. Catalysis Today, 53, 51-59.  146  Appendices  147  Appendix 1. C Codes for Modeling Radiation Distribution in the Photoreactor For solving the radiation distribution energy problem in the photoreactor used, Equation 2.8 should be solved numerically. The following code, written in C language, shows the algorithm employed. The codes are compiled and embedded into the Fluent software for calculating the radiation distribution for each computation cell in the reactor. The result is kept in individual memory space used to calculate the reaction rate of RhWT. #include "udf.h" DEFINE_ON_DEMAND(CAL_RAD_1) { float var1, factor1,k_medium, interval; int i,j,k,g; double dx; double lamp1,lamp2; double xpp,ypp,xp1,xp2,s1,s2,s3,s4,s5,l0,l1,l2,l3,l4,ABS,sum, t1, t2,TT, r2, xtmp, zr, xs, power,x1, ss1, ss2; double xinter[23],yinter1[23],yinter2[23]; Domain *d; real xp[ND_ND]; Thread *t; cell_t c; d=Get_Domain (1); lamp1=0.0755; lamp2=0.4989; //k_medium=0.01; // absorption coefecient of medium base 10 1/m xinter[1]=0.0755; xinter[2]=0.0855; xinter[3]=0.0955; xinter[4]=0.1155; xinter[5]=0.1355; xinter[6]=0.1555; xinter[7]=0.1755; xinter[8]=0.1955; xinter[9]=0.2155; xinter[10]=0.2355; xinter[11]=0.2555; xinter[12]=0.2755; xinter[13]=0.2955; xinter[14]=0.3155; xinter[15]=0.3355; xinter[16]=0.3555; xinter[17]=0.3755; xinter[18]=0.3955; xinter[19]=0.4155; xinter[20]=0.4355; xinter[21]=0.4555; xinter[22]=1; yinter1[1]=2.9543E-05; yinter1[2]=1.4149E-04; yinter1[3]=1.3567E-04; yinter1[4]=1.8700E-04; yinter1[5]=2.1280E-04; yinter1[6]=2.1733E-04; yinter1[7]=3.1742E-04; yinter1[8]=2.7685E-04; yinter1[9]=2.9020E-04; yinter1[10]=3.1131E-04; yinter1[11]=3.0023E-04; yinter1[12]=3.3008E-04; yinter1[13]=3.3876E-04; yinter1[14]=3.5395E-04; yinter1[15]=3.5516E-04; yinter1[16]=3.2764E-04; yinter1[17]=5.4830E-04; yinter1[18]=4.2983E-04; yinter1[19]=4.7438E-04; yinter1[20]=4.9250E-04; yinter1[21]=4.9185E-04; yinter1[22]=4.9185E-04;  148  yinter2[1]=3.2679E-05; yinter2[2]=1.4848E-04; yinter2[3]=1.3355E-04; yinter2[4]=1.7931E-04; yinter2[5]=2.0068E-04; yinter2[6]=2.0256E-04; yinter2[7]=2.9364E-04; yinter2[8]=2.5521E-04; yinter2[9]=2.6753E-04; yinter2[10]=2.8778E-04; yinter2[11]=2.7865E-04; yinter2[12]=3.0751E-04; yinter2[13]=3.1624E-04; yinter2[14]=3.3019E-04; yinter2[15]=3.3003E-04; yinter2[16]=3.0231E-04; yinter2[17]=5.0112E-04; yinter2[18]=3.8863E-04; yinter2[19]=4.2452E-04; yinter2[20]=4.3746E-04; yinter2[21]=4.3614E-04; yinter2[22]=4.3614E-04; interval=1000; //check point for type of calculation dx=(lamp2-lamp1)/interval; var1=3; //check point for type of calculation factor1=4600; //check point for type of calculation k_medium=4.48;// absorption coefecient of medium base 10 1/m thread_loop_c(t,d) // strt over thread { begin_c_loop(c,t) // start over each cells { C_CENTROID(xp,c,t); // recall the coordinate of study point C_UDMI(c,t,0)=0; ypp=sqrt(xp[1]*xp[1]+xp[2]*xp[2]); if ((ypp>=0.012)&&(ypp<=0.0551)) { sum=0; for (j=1;j<=3;j++) //j=1 reflection from right, j=2 direct ray, j=3 reflection from left { if (j==1) ypp=sqrt(xp[1]*xp[1]+(-0.0102-xp[2])*(-0.0102-xp[2])); // transformation to new coordinate, this one converts from 3D to 2D if (j==2) ypp=sqrt(xp[1]*xp[1]+xp[2]*xp[2]); if (j==3) ypp=sqrt(xp[1]*xp[1]+(0.0102-xp[2])*(0.0102-xp[2])); for (i=1;i<=interval;i++) { xs=lamp1-dx/2+i*dx; // calculating power base on the situation of sleeve if (var1==1.0) // clean sleeve wit low air flow rate { power=0.2700672*xs*xs*xs*xs*xs-.4730317*xs*xs*xs*xs+0.3212876*xs*xs*xs0.1032293*xs*xs+0.0161922*xs-0.0007281; // R2=0.9913 } if (var1==2.0) // clean cleeve with high air flow rate { power=0.40689*xs*xs*xs*xs*xs-0.6460746*xs*xs*xs*xs+0.3987614*xs*xs*xs0.1178414*xs*xs+0.0171349*xs-0.0007409; //R2=0.9918 } if (var1==3.0) // dirty sleeve with low air flow rate { for (g=2;g<=22;g++) { if ((xs>=xinter[g-1])&&(xs<=xinter[g])) power=(yinter1[g]-yinter1[g-1])/(xinter[g]-xinter[g-1])*(xs-xinter[g1])+yinter1[g-1]; } }  149  if (var1==4.0) // dirty sleeve with high air flow rate { for (g=2;g<=22;g++) { if ((xs>=xinter[g-1])&&(xs<=xinter[g])) power=(yinter2[g]-yinter2[g-1])/(xinter[g]-xinter[g-1])*(xs-xinter[g1])+yinter2[g-1]; } } power=power*factor1; xpp=xp[0]; // first guess for finding refractions xp1=xp[0]+1; while (fabs(xp1-xp[0])>1e-7) //finding imaginary position of study point to meet refraction { x1=(xs-xpp)/(-ypp)*(0.0085-ypp)+xpp; //first incident point inner of the lamp s1=(x1-xs)/sqrt((x1-xs)*(x1-xs)+0.0085*0.0085); s2=0.6579*s1;s5=0.7502*s1; xp1=x1+(ypp-0.0126)*s5/sqrt(1-s5*s5)+0.00245*s2/sqrt(1s2*s2)+0.00165*s1/sqrt(1-s1*s1); xpp=xpp+0.001; x1=(xs-xpp)/(-ypp)*(0.0085-ypp)+xpp; //first incident point inner of the lamp s1=(x1-xs)/sqrt((x1-xs)*(x1-xs)+0.0085*0.0085); s2=0.6579*s1;s5=0.7502*s1; xp2=x1+(ypp-0.0126)*s5/sqrt(1-s5*s5)+0.00245*s2/sqrt(1s2*s2)+0.00165*s1/sqrt(1-s1*s1); xpp=xpp-0.001*(xp[0]-xp2)/(xp1-xp2); } x1=(xs-xpp)/(-ypp)*(0.0085-ypp)+xpp; //first incident point inner of the lamp if ((xp1-(0.0126-ypp)/ypp*(xs-xp1))<0.498) // left wall blocks the ray { // findind transmitance coeffecient fro all walls t1=((sqrt(1-s1*s1)-1.52*sqrt(1-s2*s2))/(sqrt(1-s1*s1)+1.52*sqrt(1s2*s2))); t1=t1*t1; t2=((sqrt(1-s2*s2)-1.52*sqrt(1-s1*s1))/(sqrt(1-s2*s2)+1.52*sqrt(1s1*s1))); t2=t2*t2; t1=t1+t2; t1=1-t1/2; TT=t1*t1*t1; t1=(1.52*sqrt(1-s5*s5)-1.333*(sqrt(1-s2*s2)))/(1.52*sqrt(1s5*s5)+1.333*sqrt(1-s2*s2)); t2=((1.52*sqrt(1-s2*s2)-1.333*(sqrt(1-s5*s5)))/(1.52*sqrt(1s2*s2)+1.333*sqrt(1-s5*s5))); t2=t2*t2; t1=t1+t2; t1=1-t1/2; TT=TT*t1; if (j==1 || j==3) //fresnel law for reflection from wall { ss1=xp[1]/ypp; ss2=0.88867*ss1;  150  t1=((1.333*sqrt(1-ss1*ss1)-1.5*sqrt(1-ss2*ss2))/(1.333*sqrt(1ss1*ss1)+1.5*sqrt(1-ss2*ss2))); t1=t1*t1; t2=((1.333*sqrt(1-ss2*ss2)-1.5*sqrt(1-ss1*ss1))/(1.333*sqrt(1ss2*ss2)+1.5*sqrt(1-ss1*ss1))); t2=t2*t2; t1=t1+t2; t1=1-t1/2; TT=TT*t1; } // finding all lenghts l0=sqrt((xs-x1)*(xs-x1)+0.0085*0.0085); l1=0.001/sqrt(1-s2*s2); l2=0.00165/sqrt(1-s1*s1); l3=0.00145/sqrt(1-s2*s2); l4=sqrt((ypp-0.0126)*(ypp-0.0126)+(xp[0](x1+l1*s2+l2*s1+l3*s2))*(xp[0]-(x1+l1*s2+l2*s1+l3*s2))); ABS=pow(10,(-47.6*l1-47.6*l3-l4*k_medium)); ABS=ABS*TT; r2=(l0+l1+l2+l3+l4); r2=r2*r2; sum=power*TT*ABS/r2*dx+sum; //intensity excluding 4pi lamp lenght } } } sum=sum/4/3.141593; C_UDMI(c,t,0)=sum; } } end_c_loop(c,t) } }  151  Appendix 2. Reaction Rate Data of RhWT and H2O2 under UV Radiation The following data were recorded for deriving the oxidation rate of rhodamine WT in a hydrogen peroxide solution and a constant UV radiant field. Table 1. Absorbance of the solution at different wavelengths and different times during the UV-based AOP of RhWT  Elapsed Time s 0 28 64 99 134 170 205 240 291 316 352 387 422 457 492 528 563 604 639 674 709 745 780 816 851 892 927 980 1015 1050 1086  Run I Photodiode Reading Volts 0.288 0.289 0.292 0.289 0.288 0.289 0.289 0.289 0.289 0.288 0.287 0.287 0.29 0.286 0.285 0.288 0.284 0.287 0.288 0.288 0.286 0.288 0.284 0.289 0.289 0.289 0.286 0.288 0.289 0.288 0.289  Absorbance (1/cm) @254 @555.5 nm nm 0.047 0.024 0.047 0.024 0.047 0.0237 0.0471 0.0231 0.0471 0.0228 0.0472 0.0224 0.0472 0.0222 0.0472 0.0215 0.0473 0.0217 0.0473 0.021 0.0473 0.0205 0.0472 0.0203 0.0472 0.0199 0.0472 0.0194 0.0472 0.0191 0.0472 0.0187 0.0472 0.0183 0.0471 0.0185 0.0471 0.0178 0.0471 0.0175 0.047 0.0172 0.047 0.0169 0.047 0.0166 0.0469 0.0164 0.0469 0.016 0.0468 0.0158 0.0465 0.0155 0.0461 0.0134 0.0458 0.0133 0.0455 0.0132 0.0453 0.0129  Elapsed Time s 0 36 71 122 147 182 218 253 288 323 358 394 435 470 505 540 575 611 646 681 722 757 793 828 864 899 934 969 1020 1045 1081  Run II Photodiode Reading Volts 0.294 0.294 0.299 0.297 0.284 0.284 0.287 0.284 0.289 0.285 0.286 0.285 0.287 0.276 0.287 0.280 0.279 0.281 0.285 0.284 0.286 0.285 0.288 0.282 0.284 0.294 0.284 0.285 0.282 0.277 0.283  Absorbance (1/cm) @254 @555.5 nm nm 0.0446 0.0221 0.0447 0.0211 0.0447 0.0208 0.0448 0.0196 0.0448 0.0192 0.0447 0.0187 0.0446 0.0184 0.0445 0.0182 0.0445 0.0178 0.0444 0.0176 0.0443 0.0172 0.0443 0.0171 0.0442 0.0166 0.0443 0.0162 0.0443 0.0159 0.0444 0.0156 0.0444 0.0154 0.0444 0.0150 0.0445 0.0148 0.0445 0.0147 0.0445 0.0142 0.0444 0.0141 0.0443 0.0138 0.0442 0.0136 0.0441 0.0133 0.0440 0.0131 0.0439 0.0128 0.0438 0.0126 0.0437 0.0122 0.0437 0.0121 0.0437 0.0117  152  1121 1156 1197 1232 1268 1303 1338 1374 1409 1444 1495 1520 1556 1591 1626 1662 1697 1732 1783 1826 1862 1896 1932 1967 2002 2037 2088 2113 2149 2184 2219 2255 2290 2325 2361 2402 2437 2473 2508 2543 2578 2614 2649 2700 2724 2760 2796  0.288 0.288 0.286 0.285 0.289 0.288 0.287 0.288 0.287 0.287 0.287 0.288 0.288 0.29 0.291 0.287 0.289 0.29 0.289 0.289 0.288 0.287 0.288 0.288 0.285 0.289 0.287 0.286 0.288 0.289 0.29 0.288 0.287 0.288 0.289 0.287 0.288 0.286 0.285 0.289 0.286 0.287 0.288 0.289 0.287 0.286 0.287  0.045 0.0447 0.0445 0.0444 0.0444 0.0443 0.0442 0.0442 0.0441 0.044 0.044 0.044 0.044 0.044 0.044 0.044 0.044 0.044 0.044 0.044 0.0439 0.0438 0.0438 0.0437 0.0437 0.0436 0.0435 0.0435 0.0436 0.0436 0.0436 0.0436 0.0436 0.0436 0.0436 0.0436 0.0436 0.0436 0.0436 0.0436 0.0436 0.0436 0.0436 0.0436 0.0435 0.0434 0.0432  0.0125 0.0122 0.0122 0.0119 0.0116 0.0115 0.0112 0.011 0.0109 0.0107 0.0111 0.0103 0.0099 0.0099 0.0095 0.0095 0.0092 0.0092 0.009 0.0087 0.0086 0.0086 0.0084 0.0083 0.0081 0.008 0.0078 0.0076 0.0077 0.0075 0.0073 0.0071 0.0072 0.0075 0.0068 0.0067 0.0067 0.0066 0.0063 0.0065 0.0063 0.0062 0.0062 0.0059 0.0059 0.0057 0.0055  1116 1151 1187 1222 1257 1292 1334 1369 1404 1439 1474 1510 1545 1580 1621 1657 1692 1727 1762 1798 1833 1868 1919 1945 1980 2015 2050 2085 2121 2156 2192 2233 2268 2304 2339 2375 2410 2445 2481 2522 2557 2592 2628 2663 2698 2733 2768  0.285 0.287 0.283 0.277 0.285 0.284 0.292 0.290 0.296 0.284 0.287 0.287 0.287 0.290 0.290 0.291 0.297 0.291 0.288 0.290 0.284 0.291 0.290 0.293 0.295 0.284 0.294 0.284 0.283 0.294 0.293 0.293 0.290 0.290 0.298 0.294 0.285 0.288 0.300 0.286 0.289 0.293 0.290 0.282 0.292 0.291 0.287  0.0437 0.0437 0.0437 0.0437 0.0437 0.0437 0.0437 0.0437 0.0437 0.0437 0.0437 0.0437 0.0437 0.0437 0.0437 0.0437 0.0437 0.0438 0.0438 0.0438 0.0438 0.0438 0.0438 0.0439 0.0439 0.0440 0.0440 0.0440 0.0441 0.0441 0.0441 0.0441 0.0441 0.0440 0.0439 0.0439 0.0438 0.0437 0.0437 0.0436 0.0436 0.0437 0.0437 0.0437 0.0437 0.0437 0.0437  0.0115 0.0113 0.0112 0.0110 0.0108 0.0106 0.0104 0.0102 0.0100 0.0097 0.0097 0.0095 0.0091 0.0090 0.0089 0.0087 0.0086 0.0085 0.0082 0.0080 0.0079 0.0080 0.0077 0.0076 0.0073 0.0073 0.0072 0.0072 0.0070 0.0070 0.0068 0.0067 0.0065 0.0064 0.0062 0.0063 0.0062 0.0059 0.0060 0.0058 0.0060 0.0057 0.0064 0.0055 0.0055 0.0053 0.0054  153  2831 2866 2902 2937 2988  0.29 0.286 0.289 0.288 0.288  0.0431 0.043 0.0429 0.0427 0.0427  0.0056 0.005 0.005 0.0049 0.0047  2819 2845 2879 2915 2950  0.286 0.293 0.289 0.289 0.291  0.0438 0.0437 0.0437 0.0437 0.0437  0.0052 0.0052 0.0050 0.0049 0.0049  The initial and final values of temperature, concentration of hydrogen peroxide, volume of solution, and height of the solution are shown in the table. Table 2. Initial and final values of the operating parameters. Run I Temperature  Concentration of H2O2  Total Volume of  Height of Solution  (ppm)  Solution (ml)  (cm)  (º C) Initial  Final  21±1  21±1  Initial  Final  Initial  Final  Initial  Final  10.14±0.05 9.77±0.05 250.0±0.5 248.6±0.5 6.25±0.01 6.22±0.01 Run II  Temperature  Concentration of H2O2  Total Volume of  Height of Solution  (ppm)  Solution (ml)  (cm)  (º C) Initial  Final  Initial  21±1  21±1  9.79±0.05  Final  Initial  Final  Initial  9.27±0.05 250.0±0.5 247.7±0.5 6.25±0.01  Final 6.2±0.01  154  Appendix 3. Radiant Power Measured and Simulated Values The following tables show the measured radiant power for different UV lamps at different operating conditions, including the simulated results of the 1D and 3D models. Table 1. Measured and simulated radiant power for different locations around a lowpressure UV lamp (large one).  Coor dinat e  Measure d Signal  -2 0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5 6 7 8 9 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42  3.89E-08 4.56E-07 1.50E-06 2.07E-06 2.40E-06 2.61E-06 2.63E-06 2.62E-06 2.61E-06 2.61E-06 2.59E-06 2.59E-06 2.60E-06 2.65E-06 2.65E-06 2.63E-06 2.64E-06 2.64E-06 2.68E-06 2.73E-06 2.78E-06 2.79E-06 2.78E-06 2.80E-06 2.82E-06 2.83E-06 2.85E-06 2.79E-06 2.79E-06 2.78E-06 2.81E-06 2.83E-06 2.86E-06  1mm from surface Simulated Signal LSSE  VSSE  1.56E-07 6.02E-07 1.42E-06 2.09E-06 2.41E-06 2.59E-06 2.64E-06 2.62E-06 2.62E-06 2.60E-06 2.59E-06 2.59E-06 2.60E-06 2.65E-06 2.65E-06 2.63E-06 2.64E-06 2.64E-06 2.68E-06 2.73E-06 2.78E-06 2.79E-06 2.78E-06 2.80E-06 2.82E-06 2.83E-06 2.85E-06 2.79E-06 2.79E-06 2.78E-06 2.81E-06 2.83E-06 2.86E-06  4.19E-08 6.95E-07 1.40E-06 2.06E-06 2.42E-06 2.59E-06 2.64E-06 2.63E-06 2.62E-06 2.60E-06 2.59E-06 2.59E-06 2.61E-06 2.65E-06 2.65E-06 2.63E-06 2.64E-06 2.64E-06 2.68E-06 2.73E-06 2.78E-06 2.79E-06 2.78E-06 2.80E-06 2.82E-06 2.83E-06 2.85E-06 2.79E-06 2.79E-06 2.78E-06 2.81E-06 2.83E-06 2.86E-06  VSSE r=3.5 1.67E-07 6.25E-07 1.42E-06 2.09E-06 2.43E-06 2.60E-06 2.65E-06 2.63E-06 2.63E-06 2.61E-06 2.60E-06 2.60E-06 2.61E-06 2.66E-06 2.66E-06 2.64E-06 2.65E-06 2.65E-06 2.69E-06 2.74E-06 2.79E-06 2.80E-06 2.79E-06 2.81E-06 2.83E-06 2.84E-06 2.86E-06 2.80E-06 2.80E-06 2.79E-06 2.82E-06 2.84E-06 2.87E-06  6 mm Simulated Signal  Coo rdin ate  Measure d Signal  LSSE  VSSE  -2 0 1 2 3 4 5 6 7 8 9 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 42 40 36 32 28  8.12E-08 4.80E-07 1.67E-06 2.07E-06 2.14E-06 2.14E-06 2.13E-06 2.12E-06 2.12E-06 2.15E-06 2.14E-06 2.15E-06 2.14E-06 2.16E-06 2.18E-06 2.20E-06 2.21E-06 2.23E-06 2.22E-06 2.20E-06 2.21E-06 2.26E-06 2.26E-06 2.26E-06 2.23E-06 2.23E-06 2.27E-06 2.27E-06 2.26E-06 2.23E-06 2.22E-06 2.22E-06 2.21E-06  1.55E-07 5.43E-07 1.57E-06 1.99E-06 2.05E-06 2.03E-06 2.02E-06 2.03E-06 2.06E-06 2.07E-06 2.05E-06 2.06E-06 2.06E-06 2.09E-06 2.13E-06 2.17E-06 2.18E-06 2.17E-06 2.18E-06 2.20E-06 2.20E-06 2.22E-06 2.17E-06 2.17E-06 2.17E-06 2.19E-06 2.21E-06 2.23E-06 2.23E-06 2.21E-06 2.17E-06 2.17E-06 2.20E-06  4.60E-08 7.41E-07 1.88E-06 2.43E-06 2.50E-06 2.47E-06 2.46E-06 2.48E-06 2.51E-06 2.52E-06 2.50E-06 2.51E-06 2.51E-06 2.55E-06 2.60E-06 2.64E-06 2.65E-06 2.65E-06 2.66E-06 2.68E-06 2.69E-06 2.71E-06 2.65E-06 2.65E-06 2.64E-06 2.67E-06 2.69E-06 2.72E-06 2.72E-06 2.69E-06 2.64E-06 2.65E-06 2.69E-06  VSSE r=3.5 1.70E-07 5.82E-07 1.63E-06 2.07E-06 2.13E-06 2.11E-06 2.10E-06 2.12E-06 2.15E-06 2.15E-06 2.14E-06 2.14E-06 2.14E-06 2.18E-06 2.22E-06 2.25E-06 2.27E-06 2.26E-06 2.27E-06 2.29E-06 2.30E-06 2.31E-06 2.26E-06 2.26E-06 2.26E-06 2.28E-06 2.30E-06 2.32E-06 2.32E-06 2.30E-06 2.26E-06 2.26E-06 2.30E-06  155  Table 2. Measured and simulated radiant power for different locations around lowpressure larger UV lamps in a quartz sleeve.  Coordin ate 2 3 4 5 6 7 8 9 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 50 50 40 40 30 20  1mm from surface Simulated Signal Measured Signal LSSE VSSE 3.37E-08 5.09E-08 5.24E-08 2.03E-07 1.77E-07 1.73E-07 3.44E-07 5.01E-07 5.17E-07 1.72E-06 1.67E-06 1.66E-06 2.18E-06 2.20E-06 2.19E-06 2.25E-06 2.26E-06 2.25E-06 2.20E-06 2.21E-06 2.21E-06 2.21E-06 2.21E-06 2.21E-06 2.20E-06 2.21E-06 2.20E-06 2.24E-06 2.24E-06 2.24E-06 2.24E-06 2.25E-06 2.24E-06 2.23E-06 2.23E-06 2.23E-06 2.21E-06 2.22E-06 2.21E-06 2.24E-06 2.24E-06 2.24E-06 2.23E-06 2.24E-06 2.23E-06 2.23E-06 2.24E-06 2.23E-06 2.23E-06 2.23E-06 2.23E-06 2.20E-06 2.21E-06 2.20E-06 2.21E-06 2.21E-06 2.21E-06 2.19E-06 2.20E-06 2.19E-06 2.19E-06 2.20E-06 2.19E-06 2.18E-06 2.18E-06 2.18E-06 2.16E-06 2.17E-06 2.16E-06 2.15E-06 2.15E-06 2.15E-06 2.13E-06 2.14E-06 2.13E-06 2.13E-06 2.13E-06 2.13E-06 2.12E-06 2.12E-06 2.12E-06 2.10E-06 2.10E-06 2.10E-06 2.10E-06 2.10E-06 2.10E-06 2.10E-06 2.15E-06 2.15E-06 2.14E-06 2.15E-06 2.15E-06 2.20E-06 2.21E-06 2.21E-06 2.25E-06 2.24E-06 2.24E-06  Coord inate 2 4 6 8 10 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 44 44 46 50 50 40 34 26 20 12 8 4  10 mm from surface Simulated Signal Measured Signal LSSE VSSE 4.93E-08 5.07E-08 6.71E-08 3.79E-07 4.85E-07 6.46E-07 1.51E-06 1.57E-06 2.01E-06 1.57E-06 1.62E-06 2.08E-06 1.57E-06 1.61E-06 2.07E-06 1.57E-06 1.61E-06 2.07E-06 1.58E-06 1.63E-06 2.10E-06 1.58E-06 1.63E-06 2.10E-06 1.58E-06 1.63E-06 2.09E-06 1.57E-06 1.62E-06 2.08E-06 1.59E-06 1.63E-06 2.10E-06 1.59E-06 1.63E-06 2.09E-06 1.58E-06 1.63E-06 2.09E-06 1.58E-06 1.62E-06 2.09E-06 1.57E-06 1.61E-06 2.07E-06 1.57E-06 1.61E-06 2.07E-06 1.55E-06 1.60E-06 2.06E-06 1.56E-06 1.60E-06 2.05E-06 1.56E-06 1.59E-06 2.04E-06 1.55E-06 1.58E-06 2.03E-06 1.54E-06 1.57E-06 2.02E-06 1.53E-06 1.55E-06 2.00E-06 1.53E-06 1.55E-06 2.00E-06 1.52E-06 1.55E-06 1.99E-06 1.50E-06 1.53E-06 1.97E-06 1.50E-06 1.53E-06 1.97E-06 1.53E-06 1.57E-06 2.02E-06 1.55E-06 1.60E-06 2.05E-06 1.58E-06 1.62E-06 2.09E-06 1.59E-06 1.63E-06 2.10E-06 1.61E-06 1.63E-06 2.10E-06 1.60E-06 1.62E-06 2.08E-06 3.92E-07 4.85E-07 6.46E-07  156  Table 3. Measured and simulated radiant power for different locations around lowpressure larger UV lamps in a quartz sleeve with air blowing inside the sleeve.  Coordinate 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 50 40 30 20  1mm from surface Simulated Signal Measured Signal LSSE VSSE 5.22E-09 2.46E-08 3.35E-07 4.45E-07 5.25E-07 5.90E-07 6.47E-07 7.01E-07 7.41E-07 7.93E-07 8.30E-07 8.65E-07 8.95E-07 9.15E-07 9.50E-07 1.00E-06 1.10E-06 1.17E-06 1.23E-06 1.29E-06 1.34E-06 1.40E-06 1.44E-06 1.47E-06 1.50E-06 1.50E-06 1.28E-06 9.56E-07 8.03E-07  4.68E-09 4.09E-08 3.33E-07 4.47E-07 5.26E-07 5.91E-07 6.49E-07 7.02E-07 7.43E-07 7.95E-07 8.32E-07 8.67E-07 8.96E-07 9.17E-07 9.52E-07 1.00E-06 1.10E-06 1.17E-06 1.23E-06 1.29E-06 1.34E-06 1.40E-06 1.44E-06 1.47E-06 1.51E-06 1.51E-06 1.29E-06 9.52E-07 7.95E-07  4.62E-09 4.24E-08 3.32E-07 4.46E-07 5.25E-07 5.90E-07 6.47E-07 7.00E-07 7.42E-07 7.93E-07 8.30E-07 8.65E-07 8.94E-07 9.15E-07 9.50E-07 1.00E-06 1.10E-06 1.17E-06 1.23E-06 1.29E-06 1.34E-06 1.40E-06 1.44E-06 1.47E-06 1.50E-06 1.50E-06 1.29E-06 9.50E-07 7.93E-07  Coordinate 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 50 50 40 30 20 12 6  10 mm from surface Simulated Signal Measured Signal LSSE VSSE 8.11E-09 3.67E-08 2.41E-07 3.31E-07 3.86E-07 4.35E-07 4.75E-07 5.13E-07 5.45E-07 5.80E-07 6.10E-07 6.36E-07 6.56E-07 6.74E-07 6.96E-07 7.35E-07 8.08E-07 8.63E-07 9.08E-07 9.51E-07 9.87E-07 1.02E-06 1.10E-06 1.10E-06 9.51E-07 6.97E-07 5.84E-07 4.40E-07 2.48E-07  3.71E-09 4.23E-08 2.33E-07 3.25E-07 3.82E-07 4.30E-07 4.72E-07 5.10E-07 5.42E-07 5.77E-07 6.05E-07 6.31E-07 6.52E-07 6.68E-07 6.93E-07 7.33E-07 8.00E-07 8.53E-07 8.98E-07 9.40E-07 9.78E-07 1.02E-06 1.10E-06 1.10E-06 9.40E-07 6.93E-07 5.77E-07 4.30E-07 2.33E-07  4.75E-09 5.72E-08 2.98E-07 4.18E-07 4.92E-07 5.53E-07 6.07E-07 6.56E-07 6.97E-07 7.43E-07 7.79E-07 8.11E-07 8.38E-07 8.59E-07 8.91E-07 9.43E-07 1.03E-06 1.10E-06 1.16E-06 1.21E-06 1.26E-06 1.31E-06 1.41E-06 1.41E-06 1.21E-06 8.91E-07 7.43E-07 5.53E-07 2.98E-07  157  Table 4. Measured and simulated radiant power for different locations around lowpressure high-output UV lamps in air.  Coordinate  Measured Signal  0 0.5 1 1.5 2 2.5 3 3.5 4 5 6 7 8 9 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 38 36 34 32 30  1.22E-06 1.61E-06 1.58E-06 1.62E-06 1.64E-06 1.63E-06 1.61E-06 1.61E-06 1.61E-06 1.60E-06 1.61E-06 1.62E-06 1.61E-06 1.61E-06 1.62E-06 1.63E-06 1.65E-06 1.66E-06 1.67E-06 1.68E-06 1.70E-06 1.71E-06 1.73E-06 1.75E-06 1.77E-06 1.79E-06 1.80E-06 1.84E-06 1.28E-06 1.27E-06 1.82E-06 1.80E-06 1.80E-06 1.77E-06  1.5mm from surface Simulated Signal LSSE 1.22E-06 1.56E-06 1.62E-06 1.61E-06 1.64E-06 1.63E-06 1.62E-06 1.61E-06 1.61E-06 1.60E-06 1.61E-06 1.62E-06 1.61E-06 1.61E-06 1.62E-06 1.63E-06 1.65E-06 1.66E-06 1.67E-06 1.68E-06 1.69E-06 1.71E-06 1.73E-06 1.75E-06 1.77E-06 1.79E-06 1.80E-06 1.83E-06 1.28E-06 1.28E-06 1.83E-06 1.80E-06 1.79E-06 1.77E-06  VSSE 1.22E-06 1.55E-06 1.63E-06 1.61E-06 1.64E-06 1.63E-06 1.62E-06 1.61E-06 1.61E-06 1.60E-06 1.61E-06 1.62E-06 1.61E-06 1.61E-06 1.62E-06 1.63E-06 1.65E-06 1.66E-06 1.67E-06 1.68E-06 1.69E-06 1.71E-06 1.73E-06 1.75E-06 1.77E-06 1.79E-06 1.80E-06 1.83E-06 1.28E-06 1.28E-06 1.83E-06 1.80E-06 1.79E-06 1.77E-06  VSSE r=3.25 1.22E-06 1.55E-06 1.63E-06 1.61E-06 1.64E-06 1.63E-06 1.62E-06 1.61E-06 1.61E-06 1.60E-06 1.61E-06 1.62E-06 1.61E-06 1.61E-06 1.62E-06 1.63E-06 1.65E-06 1.66E-06 1.67E-06 1.68E-06 1.69E-06 1.71E-06 1.73E-06 1.75E-06 1.77E-06 1.79E-06 1.80E-06 1.83E-06 1.28E-06 1.28E-06 1.83E-06 1.80E-06 1.79E-06 1.77E-06  158  Table 4. Measured and simulated radiant power for different location around lowpressure high output UV lamp in air (continued)  Coordinate 0 1 2 3 4 5 6 7 8 9 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 36 34 32 28 24 20 16 12 8  Measured Signal 6.91E-07 9.80E-07 9.90E-07 9.97E-07 1.00E-06 1.01E-06 1.02E-06 1.02E-06 1.02E-06 1.03E-06 1.04E-06 1.05E-06 1.06E-06 1.06E-06 1.08E-06 1.09E-06 1.10E-06 1.11E-06 1.13E-06 1.14E-06 1.16E-06 1.17E-06 1.17E-06 1.19E-06 7.99E-07 1.20E-06 1.18E-06 1.18E-06 1.15E-06 1.12E-06 1.10E-06 1.08E-06 1.07E-06 1.05E-06  12 mm from surface Simulated Signal LSSE 7.50E-07 9.47E-07 9.64E-07 9.65E-07 9.57E-07 9.56E-07 9.60E-07 9.62E-07 9.60E-07 9.61E-07 9.64E-07 9.72E-07 9.80E-07 9.87E-07 9.95E-07 1.00E-06 1.01E-06 1.02E-06 1.03E-06 1.04E-06 1.05E-06 1.07E-06 1.08E-06 1.08E-06 7.62E-07 1.08E-06 1.08E-06 1.07E-06 1.04E-06 1.02E-06 1.00E-06 9.87E-07 9.72E-07 9.60E-07  VSSE 9.51E-07 1.17E-06 1.20E-06 1.20E-06 1.19E-06 1.19E-06 1.19E-06 1.19E-06 1.19E-06 1.19E-06 1.20E-06 1.21E-06 1.22E-06 1.22E-06 1.23E-06 1.24E-06 1.25E-06 1.26E-06 1.28E-06 1.29E-06 1.31E-06 1.32E-06 1.34E-06 1.33E-06 9.44E-07 1.33E-06 1.34E-06 1.32E-06 1.29E-06 1.26E-06 1.24E-06 1.22E-06 1.21E-06 1.19E-06  VSSE r=3.25 8.14E-07 1.01E-06 1.03E-06 1.03E-06 1.03E-06 1.03E-06 1.03E-06 1.03E-06 1.03E-06 1.03E-06 1.03E-06 1.04E-06 1.05E-06 1.06E-06 1.07E-06 1.07E-06 1.08E-06 1.09E-06 1.10E-06 1.12E-06 1.13E-06 1.14E-06 1.16E-06 1.15E-06 8.17E-07 1.15E-06 1.16E-06 1.14E-06 1.12E-06 1.09E-06 1.07E-06 1.06E-06 1.04E-06 1.03E-06  159  Table 4. Measured and simulated radiant power for different location around lowpressure high output UV lamp in air (continued)  Coordinate  Measured Signal  0 1 2 3 4 5 6 7 8 9 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 36 32 28 24 20 16 12 8 4  3.63E-07 4.94E-07 5.70E-07 5.83E-07 5.84E-07 5.84E-07 5.83E-07 5.83E-07 5.82E-07 5.82E-07 5.82E-07 5.83E-07 5.83E-07 5.84E-07 5.83E-07 5.82E-07 5.81E-07 5.81E-07 5.82E-07 5.81E-07 5.80E-07 5.78E-07 5.75E-07 5.44E-07 3.28E-07 5.48E-07 5.83E-07 5.86E-07 5.87E-07 5.89E-07 5.91E-07 5.93E-07 5.93E-07 5.92E-07  39 mm from surface Simulated Signal LSSE 7.65E-07 4.26E-07 4.58E-07 4.67E-07 4.67E-07 4.66E-07 4.66E-07 4.66E-07 4.67E-07 4.68E-07 4.69E-07 4.72E-07 4.76E-07 4.80E-07 4.83E-07 4.87E-07 4.90E-07 4.95E-07 5.00E-07 5.05E-07 5.12E-07 5.17E-07 5.26E-07 4.95E-07 3.70E-07 4.95E-07 5.17E-07 5.05E-07 4.95E-07 4.87E-07 4.80E-07 4.72E-07 4.67E-07 4.67E-07  VSSE  VSSE r=3.25  5.90E-07 6.19E-07 6.35E-07 6.35E-07 6.33E-07 6.33E-07 6.34E-07 6.35E-07 6.35E-07 6.37E-07 6.42E-07 6.47E-07 6.52E-07 6.56E-07 6.61E-07 6.66E-07 6.72E-07 6.79E-07 6.87E-07 6.95E-07 7.02E-07 7.15E-07 6.71E-07 5.03E-07 6.71E-07 7.02E-07 6.87E-07 6.72E-07 6.61E-07 6.52E-07 6.42E-07 6.35E-07 6.35E-07  4.80E-07 5.09E-07 5.20E-07 5.21E-07 5.19E-07 5.19E-07 5.20E-07 5.20E-07 5.21E-07 5.22E-07 5.26E-07 5.31E-07 5.35E-07 5.38E-07 5.42E-07 5.47E-07 5.51E-07 5.57E-07 5.63E-07 5.70E-07 5.76E-07 5.86E-07 5.51E-07 4.12E-07 5.51E-07 5.76E-07 5.63E-07 5.51E-07 5.42E-07 5.35E-07 5.26E-07 5.20E-07 5.21E-07  160  Appendix 4. Technical Limitations and Solutions in Detail 1 Limitations The following technical limitations cause bias during data acquisition. These technical limitations can be removed or reduced to improve the existing system and reduce data acquisition errors. The limitations are classified as follows. 1.1 Theoretical Part – Radiation  The terms in the derived formula contain several optical parameters. The most important parameters are the optical density or optical absorption coefficient, index of refraction, and reflection coefficient of surfaces at specific wavelengths. In order to accurately simulate the distribution of radiant energy, it is crucial to measure those parameters with a high level of accuracy. For pure material (such as distilled water) at very low concentrations, the proposed data from the published tables for pure chemicals are usable; however, for higher concentrations of contaminants, these data are no longer valid. Concerning reflection from a transparent surface, two parameters should be measured: 1- Reflection due to a difference in refractive index that is a function of incident angle, and 2- Reflection due to the optical properties of the surface (angle independent). Reflection is important, especially when a number of lamps have been inserted into the media. This work can address reflection due to index of refraction (first issue), but reflection due to optical properties of the surface (second issue) should be measured directly with proper instruments. The radiation distribution correlation was simplified somewhat. Among these simplifications, the constant absorption coefficient assumption is only valid for low  161  concentrations of chemicals and non-reactive media, or if the variation in concentration is low (low received fluence and low conversion efficiency) in the reactor. In an industrial photoreactor, this assumption is unlikely to be valid, and the absorption coefficient should be considered on each ray track. If a medium contains particles with semi elastic reflection (i.e., scattering is dominant), the simplified equations are not valid for modeling the propagation of radiant energy. Considering this situation, each element in the medium behaves like a UV source, and these equations should be solved separately for each source. 1.2 Experimental 1.2.1 Hydrodynamics  The most convenient way to compare the CFD simulation predictions with the experimental results from PIV is at a steady state condition with fully developed flow at the inlet/outlet. In order to generate such a condition for an experimental set up, several criteria should be taken into consideration. The most important ones are: - The inlet and outlet pipe should be long enough that flow is fully developed at the inlet/outlet of the reactor (50 times the ID). - No free (non-fixed) flexible tubes should be use to connect the inlet and outlet of the reactor to the piping network. The flexible tube creates a disturbance in the system due to its movement. - A precise pump (no pulsation) is required to maintain flow at one set point. - An accurate pulse dumper is required after the pump to eliminate all the fluctuation produced at the outlet of the pump. - A precise control system should maintain the flow rate at a specific set point.  162  - A temperature control system is crucial for maintaining the temperature of circulating water at a specific point. The control system keeps physical parameters (density and viscosity) constant while measurements are being performed. Because of the length of the reactor, the size of the equipment, and the dimensions of the lab space in which the tests were performed, criterion 1 was not fully met. Item 2 could not technically be met because, during the tests (PIV and PLIF), it was necessary to incline the reactor so that it could be drained fully or to vent entrapped air bubbles. Due to the limitations of experimental equipment used here, items 3 to 6 could not be achieved. As a result, a perfect steady state condition was not fully met, and a statistical study was performed (averaging). For the low flow rate, laminar flow, the control of flow with no fluctuation was not achieved, but at higher flow rates the fluctuation of flow was considerably lower at low flow rates. 1.2.2 Concentration Profile in the UV Photoreactor (PLIF Method)  The modified planar laser, inducing fluorescence to measure the concentration contour in a cross-section of a UV reactor, is a very powerful diagnostic tool. This method should be employed with extreme caution. The limitations of this method are: - The UV reactor needs to be transparent and the body of the reactor should be far enough from the UV source if it is made of transparent polymer. The UV radiation changes the color of the polymer and creates considerable error in the captured image of the excited fluorescent chemical. - The method can be used with no further modifications if the variation in the concentration of RhWT is low along the direction of the laser rays. For high power UV lamps, the change in concentration should be considered and an integral form of the equation should be applied. 163  - The position of all optical instruments strongly affects the captured image. All equipment should be fixed in one position. The simplified method presented is valid only for a fixed position of the equipment. Calibration and measurements should be performed for one position. - The method is highly affected by any source of light. The measurement should be performed in a dark room with high absorption of the surfaces, e.g. the walls of the room.  2 Recommendations for Improving Experimental Setups Velocity measurement is a very sensitive laboratory test. The test should be performed under precise controlled conditions in order to evaluate the results of modeling. The set up should be able to control all the physical properties of the fluid accurately. As a result, the following modifications are required to create a professional hydraulic test station for sophisticated velocity measurement. - The tanks should be equipped with coils and circulating cold/warm water. The cooling/warming system maintains the temperature of the circulating medium at one set-point. The feedback signal of the temperature controller would be connected to the existing digital temperature recorder. - The existing pump should be replaced with an accurate centrifugal pump with a flow controller. This added mechanism would run the fluid through the system at a precise flow. The set-point of the flow controller (PID) would be set according to experimental test requirements. The feedback signal of the flow controller would be connected to the existing flow meter.  164  Appendix 5. Analytical Solution of Plug Flow Photoreactor Generally, there is no linear relationship between velocity and reactant concentrations in a reactor. However, a simple correlation can be derived with some assumption. In order to derive a correlation between the velocity and the conversion of reactant (rhodamine WT) in a UV reactor, the following assumptions are considered: 1- The fluence rate of UV lamp is constant for all cases and velocity dose not change UV radiant power at source, UV lamp is operated as velocity independent, 2- UV absorption is a function of chemicals concentrations in the medium. Using UVVis spectrometer, the UV absorption measurement showed that the concentration of rhodamine WT at 125 ppb (and below) has low UV absorbance in comparison to hydrogen peroxide (10 ppm). In addition, concentration measurement showed that change in hydrogen peroxide concentration is negligible due to the low concentration of RhWT during the reaction period (0.5 ppm drop in hydrogen peroxide concentration for completion of reaction). As a result, it can be assumed that the received UV fluence (rate) for all flow rates are the same and dose not change with time (change of RhWT concentration) and velocity. 3- Velocity is constant in each cross section (simplifying assumption). Figure 1 presents a tubular, 1D reactor, with any cross section shape.  165  Area of cross section = A  x Inlet at V m/s And concentration C0  dx  Figure Appx 5.1. Plug flow UV reactor. Considering one-dimensional modeling, the mass conservation of chemical species at x (m) from the inlet over a control volume with thickness dx is: Molar flow rate at Inlet – Molar flow rate at Outlet = Molar rate of dissipation C xVA − C x + dxVA = Adx  or C xV − C x + dxV = dx  dC x dt  dC x dt  1 2  3  where Cx, Cx+dx, V, A, t are concentration of raw material at x and x+ dx from inlet, medium velocity, area cross section of reactor, and time, respectively. The rate of photoreaction can be derived from thesis Equation 4.12 ( C ln   C0    = (− 4.757 ± 0.092) × 10 −3 H where C0 is 125 ppb and H is received fluence).    The differential form of Equation 4.12 is: 1 dC dH dH = (− 4.757 ± 0.092 ) × 10 −3 =a = aG C dt dt dt  4  where a and G are constant rate and received fluence rate, respectively. Combining Equations 3 and 4 reveals:  166  V  dC = aGC dx  5  According to the aforementioned assumptions, the integral form of Equation 5 (G, C, and V are length and concentration independent) is:  C  aG ln   = x  C0  V  6  for any specific location (x) at two different velocities one can derive:  C  V C  C ln  V 1  = 2 ln  V 2  or  V 2  C 0  V1  C 0   C 0   V1       CV 1   V 2   =   C   0   7  where CV1 and CV2 are concentration of reactant (RhWT) at one specific location for two different velocities V1 and V2, respectively. Substituting the initial concentration (C0=125 ppb) and using the velocity ratio for different operating conditions (1, 2, and 3) the concentration for different scenario can be estimated as shown in Table 1.  Table Appx 5.1. 1D model predicted RhWT Concentration at one specific location for different operating conditions (0.006±0.002, 0.015±0.002, and 0.020±0.002 kg/s, corresponding to velocity ratio 2 and 3) Distance from reactor inlet Concentration for base case (0.006 kg/s, ratio 1) in ppb Concentration for double velocity in ppb Concentration for triple velocity in ppb  x1  x2  x3  98  74  55  111  96  83  115  105  95  Using Equation 7, the calculated concentrations at one point at the outlet zone of the photoreactor are 55, 83, and 95 ppb for three different (Table 1, last column),  167  respectively. The calculated results are in good harmony with experimental measurements 61±5, 81±3, and 94±4 ppb for three velocities, using an online UV spectrophotometer at the reactor outlet. Although there is a close agreement between the experimentally measured and calculated results, the 1D model is not applicable for the reactor entrance because of the back flow from reactor mid-section of the reactor (the recycling stream) that is not considered for deriving the 1D model. The 1D model is more suitable for concentration prediction once the flow is developed (close to the reactor outlet).  168  

Cite

Citation Scheme:

        

Citations by CSL (citeproc-js)

Usage Statistics

Share

Embed

Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                        
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            src="{[{embed.src}]}"
                            data-item="{[{embed.item}]}"
                            data-collection="{[{embed.collection}]}"
                            data-metadata="{[{embed.showMetadata}]}"
                            data-width="{[{embed.width}]}"
                            async >
                            </script>
                            </div>
                        
                    
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:
http://iiif.library.ubc.ca/presentation/dsp.24.1-0058653/manifest

Comment

Related Items