Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Computer-aided study of Vacuum-UV/UV process for removal of organic pollutants from water supplies Bagheri, Mehdi 2015

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

Item Metadata

Download

Media
24-ubc_2016_may_bagheri_mehdi.pdf [ 2.56MB ]
Metadata
JSON: 24-1.0228655.json
JSON-LD: 24-1.0228655-ld.json
RDF/XML (Pretty): 24-1.0228655-rdf.xml
RDF/JSON: 24-1.0228655-rdf.json
Turtle: 24-1.0228655-turtle.txt
N-Triples: 24-1.0228655-rdf-ntriples.txt
Original Record: 24-1.0228655-source.json
Full Text
24-1.0228655-fulltext.txt
Citation
24-1.0228655.ris

Full Text

   COMPUTER-AIDED STUDY OF VACUUM-UV/UV PROCESS FOR REMOVAL OF ORGANIC POLLUTANTS FROM WATER SUPPLIES  by  Mehdi Bagheri  B.A.Sc., University of Tehran, 2006 M.A.Sc., University of Tehran, 2008  A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF  DOCTOR OF PHILOSOPHY in The Faculty of Graduate and Postdoctoral Studies  (Chemical and Biological Engineering) THE UNIVERSITY OF BRITISH COLUMBIA  (Vancouver) April 2016  © Mehdi Bagheri, 2016   ii  Abstract Water authorities are increasingly worried about the occurrence of organic micropollutants (e.g., algal toxins, endocrine disrupting compounds, pesticides, industrial chemicals, taste and odor compounds) in water supplies. Removal of organic micropollutants (OMPs) from water is cost-prohibitive, particularly for small and remote communities. Vacuum-UV/UV process, an incipient catalyst/chemical-free advanced oxidation process (AOP), is potentially a cost-effective solution for removal of harmful micropollutants from water.  The main objective of this thesis was to investigate the feasibility of VUV/UV process for the removal of OMPs using a comprehensive computational fluid dynamics (CFD) analyzes. The developed model involved simultaneous resolution of the local transfer equations of momentum, mass, and radiative energy (for UV and VUV radiations), along with a complex kinetic scheme with more than 50 reactions. Given the importance of 185 nm and 254 nm emissions for the accurate modeling of the VUV/UV process, a new experimental method for measuring VUV and UV emissions of the mercury lamps was proposed. To assess the CFD model, VUV-induced degradation of model pollutants (atrazine, p-CBA) in ultrapure water samples was investigated under laminar flow conditions utilizing an axisymmetric laboratory-scale reactor. Afterwards, using an asymmetrical pilot-scale VUV/UV reactor, experimental validation of the CFD model was conducted for simulating the degradation of model pollutants (atrazine, 1,4-dioxane) in synthetic and natural contaminated waters under turbulent flow regime. Comparison of the modeling and experimental data indicated that the developed CFD model was able to predict successfully the degradation rate of target pollutants in the analyzed reactors. In addition, the proposed model showed to predict well the impact of the flow rates, and water matrix (NOM and alkalinity) on target pollutants degradation with less than 3 % average absolute relative deviation  iii  (AARD%). Relying on the insights gained by CFD analysis (e.g., knowing the critical role of pollutant mass transfer on the AOP performance of VUV systems), an improved VUV/UV process was developed through retrofitting baffles within the reactor. When compared the pollutant degradation and energy consumption of VUV/UV and H2O2/UV processes, superior performance of the improved VUV/UV process was observed.      iv  Preface The work performed in this thesis included the resulting submission of several manuscripts to peer-reviewed journals in the field of environmental and water technology, as well as a few conference presentations. Chapter 2 introduced a complete computational fluid dynamic (CFD) model of VUV/UV photoreactors for water treatment applications; a version of chapter 2 was published as:  Bagheri, M., Mohseni, M., 2014. Computational fluid dynamics (CFD) modeling of VUV/UV photoreactors for water treatment. Chem. Eng. J. 256, 51–60.   Chapter 3 investigated impact of different design parameters of annular VUV/UV photoreactors to enhance their degradation performance of organic pollutants; a version of chapter 3 was published as:  Bagheri, M., Mohseni, M., 2015. A study of enhanced performance of VUV/UV process for the degradation of micropollutants from contaminated water. J. Hazard. Mater. 294, 1–8.   Chapter 4, with the experimental assistance of a summer student Anaïs Redonnet, from the Ecole Nationale Supérieure de Chimie de Rennes, France, investigated the impact of hydrodynamics on pollutant degradation and energy efficiency of H2O2/UV and VUV/UV processes; a version of chapter 4 has been submitted as:  Bagheri, M., Mohseni, M., 2015. Impact of hydrodynamics on pollutant degradation and energy efficiency of VUV/UV and H2O2/UV oxidation processes. J. Environ. Manage.164, 114–120.   Chapter 5, with the experimental assistance of an analytical lab technician Rob North, evaluated a three-dimensional CFD model for the performance evaluation of a pilot-scale VUV/UV system with an asymmetrical geometry under a wide range of operational parameters: these are under preparation for submission as:  v   Bagheri, M., Mohseni, M., 2015. Removal of organic micropollutants from surface water using Vacuum-UV oxidation process: a pilot-scale study, in preparation.   Bagheri, M., Mohseni, M., 2015. Pilot-scale study of 185 nm advanced oxidation process for remediation of 1,4-dioxane contaminated waters, in preparation.  List of conference presentations:  Bagheri M., Imoberdorf G., Mohseni M. “Modeling of Vacuum-UV photoreactors using computational fluid dynamics” 62nd Canadian Chemical Engineering Conference, CSChE 2012, October 14-17, Vancouver, BC, 2012.   Bagheri M., Imoberdorf G., Mohseni M. “Hydrodynamics of UV photoreactors for drinking water treatments” 15th Canadian National Conference and 6th Policy Forum on Drinking, CWWA 2012, October 21-24, Kelowna, BC, 2012.   Bagheri M., Imoberdorf G.E., Mohseni M. “A CFD-based Study of the degradation of micropollutants induced by Vacuum-UV radiation” Proceedings of IUVA & IOA World Congress, Las Vegas, Nevada, September 22-26, 2013.   Bagheri M., Mohseni M. “Removal of contaminants of emerging concern from drinking water using a new generation of UV-based AOP” AWWA Water Quality Technology Conference (WQTC), November 16-20, New Orleans, La., 2014.   Bagheri M., Mohseni M. “Sustainable & cost-effective removal of organic pollutants & pathogenic organisms from surface water using VUV/UV process: A pilot study” RES'EAU Annual Meeting Conference, May 27-29, Kelowna, BC, 2015.   Bagheri M., Mohseni M. “CFD simulation of degradation of organic pollutants from contaminated surface water using Vacuum-UV oxidation process” 65th Canadian Chemical Engineering Conference, October 4-7, Calgary, AB, 2015.    vi  Table of contents Abstract .......................................................................................................................................... ii Preface ........................................................................................................................................... iv Table of contents .......................................................................................................................... vi List of tables.................................................................................................................................. ix List of figures ..................................................................................................................................x List of abbreviations ....................................................................................................................xv Acknowledgements ................................................................................................................... xvii Dedication ................................................................................................................................. xviii Chapter 1: Introduction ................................................................................................................1 1.1 Organic micropollutants in water supplies ..................................................................... 1 1.1.1  Removal of organic micropollutants from water supplies ..................................................................... 1 1.1.2  Advanced oxidation processes (AOPs) ................................................................................................. 3 1.1.2.1  H2O2/UV oxidation process ......................................................................................................... 4 1.2 Research objectives and chapters .................................................................................... 6 1.3 Literature review ............................................................................................................. 8 1.3.1  Principles of Vacuum-UV process ........................................................................................................ 9 1.3.2  VUV process in water treatment applications ..................................................................................... 11 1.3.2.1  Modeling of the VUV/UV process ............................................................................................ 15 1.3.2.2  Measurement of the 185 nm emission power of VUV lamps .................................................... 18 1.4 Knowledge gaps in VUV/UV research ......................................................................... 19 Chapter 2: Computational fluid dynamics (CFD) modeling of VUV/UV photoreactors for water treatment ............................................................................................................................23 2.1 Chapter introduction ..................................................................................................... 23 2.2 CFD model development .............................................................................................. 25 2.2.1  Hydrodynamic and mass transfer model ............................................................................................. 25 2.2.2  Fluence model ..................................................................................................................................... 26 2.2.3  Kinetic model ...................................................................................................................................... 27 2.2.4  Geometrical models and mesh structure .............................................................................................. 29 2.2.5  Boundary conditions ............................................................................................................................ 29 2.2.6  Physical properties .............................................................................................................................. 30 2.2.7  Numerical solution method and strategy ............................................................................................. 30 2.3 Experimental setup and procedure ................................................................................ 31  vii  2.3.1  Micropollutant degradation for model evaluation ............................................................................... 31 2.3.2  Determination of lamp power output ................................................................................................... 32 2.3.3  Analytical methods .............................................................................................................................. 34 2.4 Results and discussion .................................................................................................. 34 2.4.1  Lamp power output ............................................................................................................................. 34 2.4.2  Model evaluation ................................................................................................................................. 36 2.4.3  UV and VUV fluence rate distribution ................................................................................................ 39 2.4.4  Role of different degradation pathways ............................................................................................... 40 2.4.5  Species distribution ............................................................................................................................. 42 2.4.5.1 •OH radical distribution ............................................................................................................. 42 2.4.5.2  Hydrogen peroxide .................................................................................................................... 44 2.4.5.3  Degradation by-products ........................................................................................................... 45 2.5 Chapter conclusions ...................................................................................................... 46 Chapter 3: A study of enhanced performance of VUV/UV process for the degradation of micropollutants from contaminated water ................................................................................47 3.1 Chapter introduction ..................................................................................................... 47 3.2 Materials and methods .................................................................................................. 49 3.2.1  Experimental setup and procedure ...................................................................................................... 49 3.2.2  Analytical methods .............................................................................................................................. 50 3.3 Model development ...................................................................................................... 50 3.4 Computer-aided design exploration .............................................................................. 53 3.5 Results and discussion .................................................................................................. 56 3.5.1  Experimental evaluation of the CFD model ........................................................................................ 56 3.5.2  Effect of reactor hydraulic diameter .................................................................................................... 58 3.5.3  Effect of reactor illuminated surface area ........................................................................................... 61 3.5.4  Effect of simultaneous change in hydraulic diameter and reactor illuminated surface area ................ 62 3.5.5  Effect of retrofitting baffles ................................................................................................................. 63 3.5.6  Model evaluation of baffle-aided design ............................................................................................. 64 3.5.7  Pressure drop and energy use considerations ...................................................................................... 65 3.6 Chapter conclusions ...................................................................................................... 67 Chapter 4: Impact of hydrodynamics on pollutant degradation and energy efficiency of VUV/UV and H2O2/UV oxidation processes ..............................................................................69 4.1 Chapter introduction ..................................................................................................... 69 4.2 Materials and methods .................................................................................................. 71 4.2.1  Experimental setup and procedure ...................................................................................................... 71  viii  4.2.2  Analytical methods .............................................................................................................................. 72 4.3 Model development ...................................................................................................... 73 4.4 Results and discussion .................................................................................................. 75 4.4.1  CFD model validation and analysis ..................................................................................................... 75 4.4.2  Impact of flow characteristics ............................................................................................................. 79 4.4.3  Electrical energy-per-order analysis .................................................................................................... 82 4.4.4  Improved VUV/UV photoreactor design ............................................................................................ 85 4.5 Chapter conclusions ...................................................................................................... 87 Chapter 5: Pilot-scale remediation of surface water co-contaminated by 1,4-dioxane and atrazine using VUV/UV process: A CFD-based study .............................................................89 5.1 Chapter introduction ..................................................................................................... 89 5.2 Materials and methods .................................................................................................. 91 5.2.1  Experimental setup and procedure ...................................................................................................... 91 5.2.2  Analytical methods .............................................................................................................................. 93 5.3 Model development ...................................................................................................... 94 5.4 Results and discussion .................................................................................................. 98 5.4.1  CFD model validation and analysis ..................................................................................................... 98 5.4.1.1  UV and VUV fluence rate distributions .................................................................................. 102 5.4.1.2  Role of different degradation pathways ................................................................................... 104 5.4.1.3  Local concentration of target micropollutants ......................................................................... 105 5.4.2  Effect of the flow rates on micropollutant degradation ..................................................................... 107 5.4.3  Effect of the water matrix (NOM and alkalinity) .............................................................................. 111 5.4.4  Model evaluation with raw water samples ........................................................................................ 113 5.5 Chapter conclusions .................................................................................................... 116 Chapter 6: Conclusions and recommendations for future research .....................................118 6.1 Summary of results ..................................................................................................... 118 6.2 Future work ................................................................................................................. 121 Bibliography ...............................................................................................................................128 Appendix A - User defined numerical codes ............................................................................. 139 Appendix B - Experimental error and analysis .......................................................................... 145   ix  List of Tables Table 1-1 Recent experimental investigations involving VUV AOPs for water treatment .......... 13 Table 2-1 List of the reactions taking place in a VUV/UV photo-reactor for degradation of p-CBA....................................................................................................................................................... 28 Table 2-2 Kinetic rate equations for the p-CBA main degradation pathways .............................. 41 Table 3-1 List of the reactions taking place in a VUV/UV photoreactor for degradation of atrazine....................................................................................................................................................... 52 Table 3-2 List and values of the explored design parameters for the annular VUV/UV photoreactor....................................................................................................................................................... 55 Table 3-3 Calculated pressure drops within the inlet and outlet of reactor and the total energy-per-order (EEOt) for different design modification scenarios ............................................................. 67 Table 4-1 Kinetic sub-model utilized in VUV/UV and H2O2/UV advanced oxidation of p-CBA74 Table 5-1 Natural water properties (Capilano Reservoir, BC) ..................................................... 93 Table 5-2 List of the reactions taking place for degradation of 1,4-dioxane and atrazine in the 185 nm oxidation process .................................................................................................................... 96       x  List of Figures Figure 1-1 Scheme of the main degradation pathways for removal of OMPs during H2O2/UV process............................................................................................................................................. 4 Figure 1-2 Number of UV/H2O2 worldwide installations for water treatment and the predicted next installations in next decade ............................................................................................................. 5 Figure 1-3 Scheme of the main degradation pathways for removal of OMPs during VUV process ....................................................................................................................................................... 10 Figure 1-4 Scheme of by-products formation during subsequent •OH oxidation of organic pollutants ....................................................................................................................................... 10 Figure 2-1 Reactor geometry utilized for 2D model development with dimensions .................... 29 Figure 2-2 Schematic of the VUV experimental set-up ................................................................ 32 Figure 2-3 Schematic of the experimental apparatus for the 185 nm and 254 nm lamp power measurement ................................................................................................................................. 33 Figure 2-4 (a) 254 nm, (b) 185 nm output of the VUV lamp versus the amalgam temperature .. 35 Figure 2-5  Comparison of experimental data obtained by the VUV/UV experimental set-up and CFD model .................................................................................................................................... 37 Figure 2-6 Parts per billion (ppb) p-CBA concentrations in the VUV/UV photoreactor at different flow rates ....................................................................................................................................... 38 Figure 2-7 (a) 254 nm, (b) 185 nm incident radiation profile along the central axis of the VUV/UV photoreactor .................................................................................................................................. 39 Figure 2-8 Volume-weighted average kinetic rate for the main degradation pathways versus flow rates ............................................................................................................................................... 41 Figure 2-9 Concentration contours of OH radical within the VUV/UV photoreactor (M) .......... 43 Figure 2-10 OH radical dose distribution within the VUV/UV photoreactor .............................. 44 Figure 3-1 Parameterized VUV/UV photoreactor geometry utilized for this study (a: sleeve inner diameter, b: sleeve outer diameter, c: air gap, d: reactor inner diameter, e: optical thickness) .... 54 Figure 3-2 Comparison of experimental data obtained by the VUV/UV experimental set-up and CFD model .................................................................................................................................... 56 Figure 3-3 Parts per billion (ppb) atrazine concentrations with the corresponding ⦁OH (M) profiles at different flow rates .................................................................................................................... 57  xi  Figure 3-4 Removal ratio of atrazine (C/C0) and energy-per-order (EEO) values for different design modification scenarios at 1, 3.5, 6.5 L/min ....................................................................... 59 Figure 3-5 Median, mean, maximum, and P (10-90) values for the residence time, UV, VUV, and ⦁OH dose distribution of different design modification scenarios ............................................... 60 Figure 3-6 Comparison of experimental data obtained by the VUV/UV set-up using retrofitting baffles and the CFD model ........................................................................................................... 65 Figure 4-1 Schematic of the VUV/UV and H2O2/UV prototype photoreactor ............................. 75 Figure 4-2 Comparison of the p-CBA removal efficacy for the UV, VUV/UV, 2 and 5-ppm H2O2/UV processes using a conventional design photoreactor prototype .................................... 77 Figure 4-3 Parts per billion (ppb) p-CBA concentrations with the corresponding •OH concentrations (mol/L) in the conventional design photoreactor ................................................. 78 Figure 4-4 Comparison of the p-CBA removal efficacy for UV, VUV/UV, 2 and 5-ppm H2O2/UV processes using an improved prototype photoreactor via incorporation of central baffles .......... 81 Figure 4-5 Parts per billion (ppb) p-CBA concentration profiles within the improved design through introduction of central baffles ......................................................................................... 83 Figure 4-6 Electrical energy-per order (EEO) analysis of the conventional photoreactor and improved design (central baffles) ................................................................................................. 85 Figure 4-7 Comparison of the p-CBA concentration (ppb) contours within the improved VUV/UV photoreactor using central and combinatorial baffles ................................................................... 87 Figure 5-1 Schematic of utilized VUV pilot set-up for experiments ............................................ 92 Figure 5-2 Generated mesh for the pilot VUV system ................................................................. 94 Figure 5-3 Comparison of CFD predictions and experimental data for degradation of 100 ppb (a) atrazine (b) dioxane in distilled water and 2 ppm SRNOM samples ............................................ 99 Figure 5-4 Comparison of CFD predictions and experimental data for •OH scavenging effect of 100 ppb (a) dioxane on degradation of 100 ppb atrazine (b) atrazine on degradation of 100 ppb dioxane ........................................................................................................................................ 101 Figure 5-5 (a) VUV 185 nm, (b) UV 254 nm incident radiation profile along the VUV/UV pilot system ......................................................................................................................................... 103 Figure 5-6 •OH concentrations (mol/L) contour within the VUV pilot system at flow rate of 6 L/min ........................................................................................................................................... 104 Figure 5-7 Parts per billion atrazine and dioxane concentrations within the VUV pilot system at 6 L/min ........................................................................................................................................... 106 Figure 5-8 Parts per billion atrazine and dioxane concentrations within the VUV pilot system at 30 L/min ...................................................................................................................................... 107  xii  Figure 5-9 Flow streamlines colored by dioxane concentrations at 6 and 30 L/min flow rates . 108 Figure 5-10 •OH dose (a, b), residence time (c, d), UV dose (e, f), and VUV dose (g, h) distributions within the pilot system at 6 L/min (in red) and 30 L/min (in blue) flow rates ........................... 109 Figure 5-11 Model predictions and experimental data for the influence of (a) SRNOM (b) bicarbonate alkalinity, on atrazine and dioxane degradation ...................................................... 112 Figure 5-12 (a) Comparison of experimental data and model predictions and (b) EEO analysis for atrazine and dioxane degradation in CR water samples ............................................................. 115  Figure 6-1 In-situ generation of ozone gas and its integration to the VUV reactors...................124     xiii  List of symbols a Sleeve inner diameter (cm) b Sleeve outer diameter (cm) c Air gap distance (mm) e Optical thickness (mm) d Reactor inner diameter (cm) l Light path length (cm) mi Mass fraction of species i ݊ Refractive index ki Reaction rate constant for reaction i ݇ Wavelength bands Ci Inlet outlet concentrations of target pollutant (mol.L-1) Co Outlet concentrations of target pollutant (mol.L-1) Di Reactor inner diameter (cm) Do Reactor outer diameter (cm) Dm Molecular diffusivity (m2s-1) ܨ Fraction of radiant energy emitted in the interval of λ1 to λ2 at temperature T I Incident radiation (W/m2) ܫሺݎԦ, ݏԦሻ Total fluence rate in each direction sሬԦ at position rሬԦ  ݏ	ሬሬԦ Propagation direction ݎ	ሬሬԦ Position vector Ji Diffusive flux of species i (mol.s-1) K Equilibrium constant  L Reactor length (cm) N Total number of species P Pressure (Pa) Pt Total pressure drop (Pa) Q Volumetric flow rate of water (m3/h) Ri Net reaction rate per unit volume (mol.s-1) ܶ Temperature (ºC)  xiv  U Unit tensor V Velocity (m/s) V Reactor volume (L) WL Input electrical energy of UV lamp (kW) WO Equivalent electrical energy associated with the peroxide purchasing cost (kW)Wp Input electrical energy of pump (kW)  Scattering coefficient S-B Stefan-Boltzmann constant Φ Quantum yield ɛ Molar extinction coefficient (M-1cm-1) p Pump electrical efficiency  Density (kg/m3) ߮ Phase function for the in-scattering of photons λ Wavelength (nm) μ Molecular viscosity (Pa.s) τ Viscous stress tensor ߙ Spectral absorption coefficient (m-1) ݄νଵ଼ହ	௡௠ 185 nm VUV photons  ݄νଶହସ	௡௠ 254 nm  UV photons      xv  List of abbreviations AARD Average Absolute Relative Deviation AC Activated Carbon AOP Advanced Oxidation Process AWWA American Water Works Association CECs Contaminants of Emerging Concern  CFD Computational Fluid Dynamics CR Capilano Reservoir   2,4-D 2,4-Dichlorophenoxyacetic Acid DBPs Disinfection By Products  DO Discrete Ordinate  DOC Dissolved Organic Carbon DW Distilled Water EEO Electrical Energy-per-Order  EPA Environmental Protection Agency EU European Union GC Gas Chromatography HPLC High-Performance Liquid Chromatography IUPAC International Union of Pure and Applied Chemistry IUVA International Ultraviolet Association LC/MS Liquid Chromatography-Mass Spectrometry  LPM Low-Pressure Mercury MF Microfiltration  MCL Maximum Contaminant Level NDMA N-Nitrosodimethylamine NF Nanofiltartion NOM Natural Organic Matter OMPs Organic Micropollutants  p-CBA Para-Chlorobenzoic Acid PIV Particle Image Velocimetry  xvi  ppm parts-per-million ppb parts-per-billion RO Reverse Osmosis Re Reynolds Number  RTE Radiative Transfer Equation RSE Relative Standard Error  RSM Reynolds Stress Models SE Standard Error  SRNOM Suwannee River Natural Organic Matter  SUVA Specific Ultraviolet Absorbance  TOC Total Organic Carbon 2D Two-Dimensional UF Ultrafiltration  UV Ultraviolet UV-abs-254 Absorbance of Ultraviolet Radiation at 254 nm  UVT Ultraviolet Transmittance  VUV Vacuum-UV  k- k-epsilon    xvii  Acknowledgements To my supervisor, Professor Madjid Mohseni, thank-you for your guidance and friendship. To my Ph.D. committee, Dr. Michael Sasgas from Trojan Technologies and Dr. Fariborz Taghipour for reviewing my research proposal and providing valuable feedback to strengthen it. I am also thankful for their review of the dissertation and subsequent examinations. To my wife Marjan and my son Kian Armani, the source of my strength, inspiration and happiness. To my family, thank-you for their endless love, thrust, and continuous support. To the students I have mentored and supervised over the years. Thank-you for your patience. To my friends Rob North and Gustavo Imoberdorf for teaching me few analytical techniques in addition to valuable assistance in the VUV pilot system related experiments.  To my labmates (in the AOP lab) in particular Ramin Rezaei and Mehdi Bazri for their assistance in experimental work. To the CHBE machine shop. For building my experimental apparatus with special attention. To UBC, Natural Sciences and Engineering Research Council of Canada (NSERC), RES'EAU-WaterNET for their financial support on my work. Thank-you for your trust.    xviii  Dedication   To my caring wife, Marjan, and inspiring son, Kian Armani             “A word to the wise is enough”                      Plautus  1  Chapter 1: Introduction  1.1 Organic micropollutants in water supplies The presence of organic micropollutants (OMPs) in water supplies is of great environmental and public health concern. OMPs consist of a vast and expanding array of natural (e.g., mycotoxins and phytoestrogens) as well as anthropogenic substances such as chemical solvents, pesticides, herbicides, pharmaceuticals, personal care products and so on. Although such contaminants are being detected typically at low concentrations (e.g., a few ng L-1 to several µg L-1), the concern is primarily due to their potential to cause chronic health problems. Studies have shown their negative impacts on different strata of living organisms such as plants, sea-life, and animals (Ericson et al., 2010; Oaks et al., 2004). For example, OMPs such as atrazine (a widely used herbicide) and diclofenac (a commonly used pharmaceutical) have shown alarming effects on fish, frogs and vultures, respectively (Oaks et al., 2004; Richardson and Ternes, 2014). Further, replicated exposure to 1,4-dioxane (a widely used chemical solvent) in drinking water resulted in liver and nasal cavity tumors in rats, liver carcinomas and adenomas in mice, and liver and gall bladder tumors in guinea pigs (Stickney et al., 2003). Natural OMPs such as microcystins (originated by blue-green algae) have shown chronic liver injury from oral exposure as well as sever poisoning effect in both human and animals (Jochimsen et al., 1998). As a result, several regulatory agencies (e.g., U.S. EPA, EU directive) have taken measures to identify the problems associated with OMPs and have included several OMPs in their directive frameworks (Richardson and Ternes, 2011, 2014). OMPs can enter the water supply from various sources. This includes discharge of untreated and treated wastewater, seepage from landfills, septic systems, and sewer lines, and runoff from animal wastes and agricultural activities. Natural micropollutants such as mycotoxins are introduced to natural water bodies due to the release of toxins from blue-green algae (cyanobacteria). Additional  2  natural micropollutants are biologically produced taste and odor compounds (Watson, 2004), which are not primarily a toxicological problem but are of great aesthetic concern. Owing to the high bioaccumulation potential and easy transport in the environment, organic micropollutants often find their ways into drinking water supplies such as ground and surface waters. Thus, OMPs are recognized increasingly as potential threats to drinking water quality. 1.1.1  Removal of organic micropollutants from water supplies Removal of organic micropollutants depends on physicochemical properties of micropollutants and the treatment conditions. However, conventional water/wastewater treatment processes have not shown to be effective to control a wide range of OMPs. For instance, using coagulation–flocculation in jar tests with Ohio River water, atrazine removal of 4–11% was observed (Miltner, 1988). In another instance, less than 28% removal efficiency of pharmaceuticals and hormones such as diclofenac and estriol was attained during the sedimentation stage (Behera et al., 2011). Similarly, conventional biological processes were observed ineffective for degrading a broad range of antibiotics (Verlicchi et al., 2012). Likewise, majority of pesticides (e.g., triazines) and chemical solvents (e.g., dioxanes) are toxic to microorganisms and quite resistant to microbial degradation (Adams et al., 1994; Wackett et al., 2002). As for filtration processes, micropollutants removal have not shown to be effective during microfiltration (MF) and ultrafiltration (UF), as the membrane pore sizes are much larger than the molecular sizes of micropollutants. Meanwhile, nanofiltartion (NF) showed still somewhat permeable to some relatively small micropollutants (Steinle-Darling et al., 2010). In this context, alternative water treatment methods have gained popularity because of increasingly stringent environmental regulations and concerns regarding OMPs removal. In particular, technologies such as reverse osmosis (RO), activated carbon (AC) and advanced oxidation processes (AOPs) have shown promises for the removal of a wide range of OMPs from water sources. Among  3  those, AOPs are particularly advantageous since they are capable of completely or partially oxidizing the target pollutant, rather than transferring it into other phases that is occurring in RO and AC. However, implementation of AOPs can be cost-prohibitive, particularly in small to medium size water treatment facilities due to the high operational and maintenance costs (Sudhakaran et al., 2013). This necessitates the growing need for identifying robust water treatment processes at a lower cost and with less energy demand yet minimized use of chemicals to promote environmental sustainability.  1.1.2  Advanced oxidation processes (AOPs) Advanced oxidation processes (AOPs) are based on the in-situ generation of hydroxyl radicals (•OH), which reacts non-selectively with electron-rich sites of organic contaminant at rates often approaching the diffusion-controlled limit (109-1010 M-1s-1) (Linden and Mohseni, 2014). The effectiveness of AOPs for removal of a wide range of OMPs from water sources have been demonstrated extensively (Jović et al., 2013; Matilainen and Sillanpä, 2010; Sanches et al., 2010; Shu et al., 2013; Yang et al., 2013; Zoschke et al., 2012). Following the production of hydroxyl radicals, the organic contaminants undergo several sequential oxidation reactions resulting in mineralization of organics present in the water, however, often the aim is partial oxidation of pollutants into innocuous products. In general, most AOPs requires addition of a catalyst and or exogenous chemical oxidants. Not only addition of auxiliary does imposes operational complexities, but also it contributes significant to process costs. Among all various AOPs, UV-initiated ones have proven to be more efficient in terms of •OH generation (Audenaert et al., 2011), and of them H2O2/UV is the one which is currently commercialized and widely studied (Audenaert et al., 2011; Cicek et al., 2007).  4  1.1.2.1 H2O2/UV oxidation process Since the early 1900s, the UV radiation, in particular the germicidal region between 200-300 nm, has been used as an effective drinking water disinfection. Despite the effectiveness of UV radiation for inactivation of microorganisms, it has not shown adequate for the effective removal of a wide range of OMPs from water supplies (Yang et al., 2013). However, UV radiation in conjunction with hydrogen peroxide (H2O2/UV) has proven capable of degrading a wide range of OMPs from water sources. In principle, two main degradation mechanisms contributes to the removal of OMPs during the H2O2/UV process: UV direct-photolysis and OH-radical oxidation (shown in Figure 1-1).  The extent to which UV-photolysis contributes to the degradation of target pollutant is dependent on the optical properties of organic pollutant. While some compounds (e.g., NDMA) are well degraded by direct-photolysis, OH-radical oxidation is the predominant removal mechanism for others (e.g., 1,4-dioxane). Nonetheless, there are also many OMPs (e.g., atrazine) which are degraded by a combination of direct-photolysis and OH-radical oxidation during the H2O2/UV oxidation process.  Figure 1-1 Scheme of the main degradation pathways for removal of OMPs during H2O2/UV process The first H2O2/UV process for drinking water treatment applied at full-scale was commissioned in 1998 in Salt Lake City, USA (Sarathy and Mohseni, 2006). Since then, the number of large-scale H2O2/UV installations increased steadily with an average slope of 1.5 per year (Figure 1-2). The growing interest in the H2O2/UV process for water production is driven mainly by the effectiveness of the process to remove both chemical and microbial contaminants in small footprint. Additionally, H2O2 → 2 •OH UV radiation    (200-280nm) OMPs OMP 5  unlike ozonation, disinfection by products (DBPs) such as bromate are not formed during H2O2/UV process.  Despite several large scale applications of H2O2/UV process, there are still several issues that negatively impact its worldwide operation in water facilities (Giri et al., 2011; Oppenländer, 2003). Of the crucial importance is the very low UV light absorption coefficient of H2O2 that significantly hinders its conversion to hydroxyl radicals (Giri et al., 2011; Glaze et al., 1987; Holt et al., 1948). This can translate into high peroxide demand, with only a very small amount of it being converted to ⦁OH and the rest remaining unreacted (Keen et al., 2013). Since the surplus H2O2 can negatively impact the health of living bodies (Buchanan et al., 2006; Oppenländer, 2003; White, 2010), post-treatment and monitoring of excess peroxide is required, adding to the cost and complexity of this process. The demanding need for quenching the residual peroxide imposes further costs and complexity to the H2O2/UV process, specifically for applications in small to medium size treatment facilities.  Figure 1-2 Number of H2O2/UV worldwide installations for water treatment and the predicted next installations in next decade (Malley, 2010)  024681012141618201996 1998 2000 2002 2004 2006 2008 2010 2012Number of  UV/H2O2installationsYear0204060801001201995 2005 2015 2025 6  1.2 Research objectives and chapters The structure of this thesis was written as a logically flowing narrative, with each chapter of the thesis separated into a clear and concise research topic. Each chapter was built on the knowledge accumulated from the previous, which led to development of a comprehensive computational fluid dynamics (CFD) model for an in-depth analysis of VUV/UV process for the removal of organic micropollutants from water. The developed model was then validated rigorously under a broad range of operational variables using bench scale and pilot scale VUV/UV photoreactors. This thesis looked at four separate tasks, each with specific objectives: i.  Chapter 1 (Introduction and literature review)   a. Provided background information on the problem of OMPs in water supplies and technologies for the removal of OMPs.  Particular emphasis was placed on H2O2/UV and VUV/UV oxidation processes, in order to underline the significance and novelty of this thesis.  ii. Chapter 2 (Development of a 2D CFD model for a lab-scale VUV/UV reactor with synthetic water samples)   a. Demonstrated the capability of VUV/UV process for continuous flow degradation of a model pollutant (p-CBA) spiked in MilliQ water using a simplified lab-scale axisymmetric reactor. b. Developed a comprehensive two-dimensional CFD model for in-depth analysis VUV/UV process, which validated experimentally using results from a lab-scale VUV/UV photoreactor under 0.5-6.5 L/min flow.  c. Proposed a protocol for measurement of the VUV 185 nm and UV 254 nm emissions from ozone-generating low-pressure mercury amalgam lamps under a wide range of lamp temperature.    7  iii. Chapter 3 (Study of an enhanced VUV/UV process) a. Investigated impact of the main design parameters of annular photoreactors on the AOP performance and energy-efficiency of the VUV/UV process for removal of a model OMP, atrazine. b. Analyzed UV, VUV and •OH dose distributions of VUV/UV reactors for different simulation-driven design scenarios resulting in an improved VUV/UV reactor design through retrofitting central baffles.   c. Verified experimentally the 2D CFD model for removal of a model pollutant (atrazine) from MilliQ water using a prototype baffle-aided VUV/UV reactor. d. Proposed an enhanced VUV/UV process for removal of OMPs from water.  iv. Chapter 4 (Comparative study of VUV/UV and H2O2/UV oxidation processes) a. Studied and compared the degradation and energy-efficiency of VUV/UV AOP to that of UV, 2-ppm and 5-ppm H2O2/UV processes for the removal of a target pollutant (p-CBA) from MilliQ water under 0.5-6.5 L/min flow rate. b. Investigated the impact of hydrodynamics (flow characteristics) on degradation and energy-efficiency of H2O2/UV and VUV/UV processes using MilliQ water and methanol as a surrogate •OH scavenger. c. Evaluated the extent of methanol concentration with the equivalent •OH scavenging impact of 2-ppm natural organic matter (NOM) on energy efficiency of the VUV/UV and H2O2/UV AOPs using the electrical energy-per-order (EEO) analysis. d. Validated experimentally prediction capability of the proposed CFD model to study the AOP performance of VUV/UV and H2O2/UV processes in the presence of a surrogate •OH scavenger (methanol).   8  v. Chapter 5 (Development of a 3D CFD model for a pilot-scale VUV/UV system using real water samples) a. Developed and validated experimentally a 3D variant of the initially proposed CFD model for in-depth performance analysis of a pilot-scale VUV/UV system with a non-symmetrical geometry (e.g., composed of two sequentially installed VUV reactors) under 6-30 L/min flow rate. b. Conducted feasibility study of the pilot-scale system for remediation of waters co-contaminated by different micropollutants (e.g., atrazine and 1,4-dioxane) using CFD simulation. c. Analyzed •OH scavenging impact of the main water constituents (e.g., NOM and alkalinity) at different concentrations on the degradation efficiency of the VUV/UV pilot system using CFD and experimental results. d. Evaluated experimentally prediction performance of the proposed CFD model to investigate the remediation of natural surface water (Capilano reservoir, BC, Canada) contaminated by OMPs. e. Compared degradation and energy efficiency of the pilot system for parallel and sequential configurations of the two VUV reactors. vi. Chapter 6 (Conclusions and future work) a. Provided a brief summery and conclusion of this thesis as well as a couple of proposed research topics, which can potentially leading to advancement of the VUV technology in water treatment.   The knowledge generated by this thesis is likely the most comprehensive modeling study of the VUV/UV process under a wide range of operational parameters for water treatment applications to date. In addition, this thesis is the first investigation, which presented feasibility study of VUV/UV  9  process under continuous flow operation using laboratory and pilot-scale photoreactors. Positive impact is expected for small and remote communities in Canada and worldwide, considering the simple and cost-effective operation of the VUV/UV process as a chemical-free technology offering an extra barrier against microorganisms, and lowering the health risks associated to organic pollutants in a small footprint.  The following section provides a general review of the available literature on the principles of VUV process and modeling studies related to VUV/UV and H2O2/UV processes. The literature review is meant to be an introduction to the VUV oxidation process literature, and not specific to the research chapters. A more specific review directly relevant to each individual chapter’s research objectives is provided as part of that chapter. 1.3  Literature review 1.3.1 Principles of Vacuum-UV process  Vacuum-UV (VUV) radiation, relying on high energy VUV photons (λ < 200 nm) for the generation of •OH without utilizing any auxiliary oxidant (e.g., peroxide), is a viable alternative for removal of a broad range of organic micropollutants (Imoberdorf and Mohseni, 2011c). The term “Vacuum-UV” has been recommended for radiation within the region of 100-200 nm by IUPAC (International Union of Pure and Applied Chemistry) (Nic et al., 2005). The VUV radiation is capable of generating hydroxyl radical from homolysis (equation, 1-1) and ionization (equation, 1-2) of water molecules (Gonzalez et al., 2004):  ܪଶܱ ൅ ݄ν௏௎௏ 	 	→	ܪܱ• ൅ ܪ• (1-1)ܪଶܱ ൅ ݄ν௏௎௏ 	 	→	ܪܱ• ൅ ݁ି௔௤ ൅ ܪା (1-2) 10  Degradation of target OMPs proceeds through •OH-oxidation pathway as well as direct VUV photolysis (Figure 1-3). Degradation mechanism involving OH radicals covers a manifold of parallel radical reaction (Gonzalez et al., 2004) resulting in generation of intermediates and by-products, eventually leading to complete mineralization. Figure 1-4 displays a summarized pathway leading by the consecutive •OH-oxidation of organic compounds.  Figure 1-3 Scheme of the main degradation pathways for removal of OMPs during VUV process  Low-pressure mercury (LPM) lamps, emitting mostly monochromatic radiation at 254 nm, are widely used for water treatment applications (e.g., disinfection, H2O2/UV). Some modifications of the quartz envelop allows transmission of some 185 nm VUV radiation along with the dominant 254 nm UV emission. Since the 185 nm (< 200 nm) photons are capable of ozone generation from oxygen, these lamps are often referred to ozone generating or VUV lamps (Light sources, 2013). VUV lamps are readily available with an identical purchasing price to widely used UV lamps. Furthermore, both UV and VUV lamps are compatible to operate with a matching electrical ballast resulting in identical electrical power consumptions. Given that VUV (ozone-generation low-pressure mercury) lamps emit both VUV 185 nm and UV 254 nm radiations, the process involved is referred to as the VUV/UV process.   Figure 1-4 Scheme of by-products formation during subsequent •OH oxidation of organic pollutants  •OH + OrganicAlcohols, Ketones, Aldehydes, Carboxylic AcidsAcetic, Formic, Oxalic AcidsCO2 + H2OH2O → •OH VUV radiation    (100-200nm) OMPOMP 11  VUV/UV technology is a chemical-free solution to remove a wide range of microorganisms and micropollutants from water with low capital and maintenance costs (Imoberdorf and Mohseni, 2011c). Compared to H2O2/UV AOP, VUV/UV process offers all the advantages associated with the H2O2/UV process, plus the additional benefit of no chemical addition, handling, and storage that in turn can translate into a simpler and greener process with potentially lower operating cost. In addition, the VUV/UV process would require neither post-treatment nor monitoring for peroxide, making it much simpler in terms of operation. All these technical and operational advantages render the VUV/UV process a suitable candidate specifically for applications in small, rural and remote community water systems for the growing demand of simultaneous pathogen deactivation and micropollutants removal.  1.3.2 VUV process in water treatment applications VUV radiation was found highly efficient for the oxidative degradation of organic compounds dissolved or dispersed in aqueous and in gaseous phases (Schuchmann et al., 1978). Commonly, two lines of excitation are used: the Xe2*-emission at 172 nm and the Hg-emission at 185 nm. The use of Xe2*-emission in water treatment is limited due to the very high absorption coefficient of water at 172 nm. The 185 nm VUV radiation (known as VUV/UV process) is frequently used for the production of ultrapure water, and for the treatment waste gases originating from hoods (canteens, chemical and biological laboratories), offices and hospital rooms (Robl et al., 2012). In recent years, the growing concern over the frequent occurrence of organic micropollutants has stimulated many water researcher to study the potential application of VUV radiation for removal of OMPs in water. The first complete study which utilized 172 nm VUV radiation for effective degradation of a wide range of organic micropollutants in ultrapure water was performed by Oppenländer and Gliese (2000). Later, kinetic of pollutants degradation in MilliQ water during VUV process was thoroughly  12  investigated by Gonzalez et al. (2004). Next, several investigators compared the removal efficacy of VUV process for degrading OMPs when compared to various UV-initiated AOPs. For instance, Han et al. (2004) demonstrated 3-6 times greater removal efficacy of p-CBA (p-chlorobenzoic acid) in the VUV/UV AOP when compared to UV and TiO2/UV processes. Later, Kim and Tanaka (2009) studied removal of 30 different micropollutants (pharmaceuticals and personal care products) spiked into pure water using the VUV/UV process. Their results indicated superior removal performance of the VUV/UV AOP when compared to UV process. Zoschke et al. (2012) evaluated the treatment cost of the VUV/UV and H2O2/UV AOPs for remediation of taste and odor compounds (e.g., geosmin and 2-methyl isoborneol) in terms of the electrical energy-per-order (EEO) analysis. The authors underlined the cost competitiveness of VUV/UV over H2O2/UV process with up to 50% lower EEO for small batch-recirculating reactors. Recently, Arany et al. (2013) reported a comparative study of VUV (172 nm), VUV/UV (185 nm + 254 nm) and UV (254 nm) processes for the degradation of an anti-inflammatory drug (Naproxen) in a recirculating batch reactor. Among the investigated processes, VUV/UV AOP showed the greatest removal for the target pollutant. Given the effectiveness of VUV radiation for removal of OMPs in ultrapure samples, Imoberdorf and Mohseni (2011c) studied degradation of a target herbicide (2,4-dichlorophenoxyacetic acid) in raw water samples using a lab-scale recirculating VUV/UV reactor. The authors indicated significant reduction in degradation of the target pollutant due to the presences of NOM and alkalinity in the water matrix. They also showed that the degradation rate of the target compound was independent of its initial concentration. Similarly, significant influence of main water constituents (e.g., DOC, carbonate/bi-carbonate) on the VUV-induced decomposition of 1,4-dioxane in raw water samples was reaffirmed by Matsushita et al. (2015) using a lab-scale recirculating reactor. Table 1-1 briefly reviews most recent and experimental studies, which utilized VUV radiation to degrading OMPs or NOM in synthetic and natural water samples.   13  Table 1-1 Recent experimental investigations involving VUV AOPs for water treatment Reference Target compounds Reactor size Lamp type Process type Parameters monitored Oppenländer & Gliese (2000) 20 different alcohols & phenols Di=1.6 cm, L=25cm Xe2* excilamp (172nm) Semi-batch TOC change over time for deionized water samples Al-Momani et al. (2002) 10 different textile dyes l= 5.5 mm, L= 85 cm  Do= 1.5 cm LPML (185 + 254 nm) Batch Effect of VUV/UV treatment on BOD & COD change on textile waste-water samples using conductivity & spectrophotometry Thomson et al. (2004) NOM V= 0.9 lit, l= 1.94 cm LPML (185 + 254 nm) Semi-batch Removal of NOM in surface water samples using DOC & spectrophotometry Buchanan et al. (2005) NOM V= 0.9 lit, l= 1.94 cm LPML (185 + 254 nm) Semi-batch NOM fractionation in surface water samples using DOC & HPSEC Dobrović et al. (2007)  NOM Di= 2.5 cm, Do= 4 cm LPML (185 + 254 nm) Semi-batch NOM removal for natural lake water using spectrophotometry Buchanan et al. (2006) NOM V= 0.9 lit, l= 1.94 cm LPML (185 + 254 nm) Batch By-products & nitrite formation followed by VUV/UV treatment for surface water samples using TOC & spectrophotometry Alapi & Dombi (2007) Phenol L=34 cm, Do=3 cm LPML (185 + 254 nm) Semi-batch Methanol (scavenger) & contaminant initial concentration effect on the removal ratio of MilliQ water samples using HPLC, LCMS & spectrophotometry Quici et al. (2008)  Citric and gallic acids L= 25 cm, Do=3 cm  Xe2* excilamp (172nm) Semi-batch Effect of pH & dissolved oxygen on distilled water samples using TOC & HPLC analysis Tasaki et al. (2009) Methyl orange Do=1.5cm,  l= 1.7 cm LPML (185 + 254 nm) Semi-batch Effect of pH, N2 & O2 micro-bubble purging for de-ionized water samples using TOC & spectrophotometry Kutschera et al. (2009) Geosmin & 2-methyl isoborneol (2-MIB) V= 11.5 lit LPML (185 + 254 nm) Semi-batch Effect of NOM, alkalinity & pH on [⦁OH]ss concentration (pCBA as probe compound) using GC-MS for  ultrapure & surface water samples. Kim & Tanaka (2009) 30 different kinds of PPCPs Do=30 cm, L=109 cm LPML (185 + 254 nm) Batch Degradation rate of contaminants using HPLC, LC/MS/MS analysis for pure water samples Afzal et al. (2010) Anatoxin-a V=0.003 lit cuvette Medium pressure UV lamp Batch Effect of pH, temperature & initial contaminant concentration on natural and synthesized water samples using HPLC analyze Ratpukdi et al. (2010) NOM V=16 lit, Do= 30 cm    (4 lamps) LPML (185 + 254 nm) Batch Effect of pH & ozone addition on  surface water samples using DOC, HPSEC & spectrophotometry Imoberdorf & Mohseni (2011a) NOM V= 0.085 lit,  Do=2.5cm Di=1.5cm LPML (185 + 254 nm) Semi-batch Effect of pH, recycling flow & alkalinity on surface water samples using TOC, HPSEC & spectrophotometry Imoberdorf & Mohseni (2011b) Formic Acid V= 0.085,  lit Do=2.5cm Di=1.5cm LPML (185 + 254 nm) Batch & semi-batch Degradation rate of the contaminate on MilliQ water samples using spectrophotometry  14  Reference Target compounds Reactor size Lamp type Process type Parameters monitored Imoberdorf & Mohseni (2011c) 2,4-Dichlorophenoxyacetic acid V= 0.085, Do=2.5cm, Di=1.5cm LPML (185 + 254 nm) Batch & semi-batch Effect of pH, alkalinity & initial pollutant concentration on ultrapure  & different surface water samples using HPLC & TOC Mouamfon et al. (2011)  Sulfamethoxazole Di= 8 cm, l= 3cm  LPML (185 + 254 nm) Semi-batch Effect of pH, nitrate, humic acid & alkalinity on MilliQ water & sewage effluent samples using HPLC analysis Ma et al. (2011) Azo dyes V=2 lit LPML (185 + 254 nm) Semi-batch Effect of pH, VUV intensity, dye & H2O2 initial concentration along with TiO2 loading does on deionized water samples using HPLC & spectrophotometry Robl et al.(2012) Methanol V=0.24 lit Xe2* excilamp (172nm) Semi-batch Monitoring pH, DOC & H2O2 concentration change over the course of VUV reactions for MilliQ water samples Zoschke et al. (2012) Geosmin & 2-methyl isoborneol (2-MIB) V=0.1, 1.3, & 3.1; l= 0.7, 2.2, & 3.9 cm LPML (185 + 254 nm) Semi-batch Removal ratio in ultrapure, tap and surface water samples using HPLC & spectrophotometry  for three different reactor sizes Gu et al. (2012) 1,1,1-trichloroethane V=1 lit, Do=7 cm, L=25cm LPML (185 + 254 nm) Batch Effect of pH, chlorine, alkalinity, humic acid & persulfate on degradation of target OMP in MilliQ water samples using HPLC, TOC & spectrophotometry Azrague et al. (2012) 2,4 dihydroxibenzoic acid (2,4-D) V= 350 cm3, L=25 D Do=3 cm Xe2* excilamp (172nm) Semi-batch Effect of oxygen and temperature on degradation of 2,4-D in double distilled water samples using LC-MS and HPLC-UV Huang et al. (2013) Methylene blue (MB) V=0.4 lit, Do=10 cm LPML (185 + 254 nm) Batch Effect of TiO2 dosage on MB mineralization of  synthetic waste-water samples using LC/MS, TOC & spectrophotometry Arany et al. (2013) Naproxen L=13 cm, Do=5 cm Di=4 cm Xe2* excilamp (172nm) & LPML (185 + 254 nm) Semi-batch Effect of dissolved O2 & initial contaminant concentration on MilliQ water samples using HPLC, TOC & spectrophotometry Moussavi et al.  (2014) Organophosphate pesticides L=40 cm,  Di=25 cm LPML (185 + 254 nm) Batch Effect of pH, initial concentration of pollutant, effect of different anions on degradation rate of the target OMPs in distilled water samples using HPLC Wang et al. (2014) p-CBA 4-chlorophenol 4-nitrophenol L=33 cm;  Di=5.5 cm LPML (185 + 254 nm) Semi-batch Effect of air bubbles & TiO2 coating on removal ratio of the target compounds in pure water samples using HPLC Matsushita et al. (2015) 1,4-dioxane V=7.5 lit, Do=3.3 cm Di=2.4 cm LPML (185 + 254 nm) Semi-batch Effect of pH, different concentration of H2O2 and the use of TiO2 on decomposition rate of 1,4-dioxane in surface and ground water samples using GC/MS Light path length = l; Reactor Volume= V; Reactor inner diameter = Di; Reactor outer diameter = Do, Reactor length =L; Natural organic matter= NOM; low-pressure mercury lamp = LPML; Liquid chromatography-electrospray tandem mass= LC/MS/MS; Liquid chromatography–mass spectrometry= LC/MS; Total organic carbon: TOC. 15  As presented in Table 1-1, all previous experimental efforts are restricted to batch and/or semi-batch operational conditions, which highlight the need for the demonstration of this promising technology under continuous operation for practical applications.  1.3.2.1 Modeling of the VUV/UV process Despite the many experimental-based studies involving VUV-induced degradation of micropollutants, there are very limited modeling attempts that investigate removal of OMPs by VUV radiation. The first complete kinetic modeling of the VUV/UV process was conducted by Imoberdorf and Mohseni (2011b, 2012) to simulate degradation of formic acid in MilliQ water samples using a completely mixed batch reactor. The proposed model was composed of 40 reactions and showed to predict well the concentration of formic acid. Later, the authors (Imoberdorf and Mohseni, 2012) expanded the applicability of their model by incorporating the •OH scavenging impact of NOM and alkalinity to predict degradation of 2,4-D, and obtained good agreement between observed and predicted data. The authors indicated the presence of large dead zone (dark) areas that significantly limit the efficacy of the VUV/UV process. Crapulli et al. (2014) have recently presented a comprehensive kinetic model, predicting hydrogen peroxide profiles during the VUV 185 nm and 172 nm VUV processes. They reported very good consistency between the modeling and experimental results. In addition, the model showed capable to predict well the VUV-induced degradation of formic acid. Besides, their modeling results confirmed that short-lived hydroxyl radicals generated from 172 nm photolysis of water were present at a radial distance of around 230-390 µm, depending on the specific operating conditions.  Despite the presence of complete kinetic models for the VUV-induced degradation of organic pollutants, lack of proper models that account for the hydrodynamics (flow characteristics) on degradation efficacy of the VUV/UV process is among the key impediments hindering the their  16  practical applications. Complete modeling of the VUV/UV process requires simultaneous resolution of hydrodynamics (turbulence), fluence rate at both 185 nm and 254 nm wavelengths, and a comprehensive reaction scheme taking place within VUV/UV photoreactors (Bagheri et al., 2013).  Computational Fluid Dynamics (CFD) is an established and effective numerical approach for modeling complex processes, and has been used extensively in recent years for the design and optimization of UV disinfection and oxidation reactors (Alpert et al., 2010; Duran et al., 2011; Elyasi and Taghipour, 2010; Santoro et al., 2010; Taghipour and Sozzi, 2005). More specifically, CFD has been used the evaluation of different hydrodynamics (turbulence) sub-models for simulation of annular UV photoreactors (Alpert et al., 2010; Duran et al., 2011; Elyasi and Taghipour, 2010; Santoro et al., 2010; Taghipour and Sozzi, 2005). Sozzi and Taghipour (2006a) experimentally evaluated the CFD-simulated velocity profiles of three different turbulence models (e.g., S k–ɛ, R k–ɛ, RSM) versus those obtained from particle image velocimetry (PIV). For the investigated annular reactors (concentric and normal inlets) and under the operating conditions (Re ~10000), both R k–ɛ and RMS turbulence models displayed the best overall match to the experimental PIV measurements. Similarly, Duran et al. (2009) analyzed the CFD-computed mass transfer rates for four different turbulence models (e.g., S k–ɛ, R k–ɛ, RSM, AKN) with the experimental results in commercial-type annular reactors under transient and turbulent flow regimes (3000 < Re < 11000). Although they observed similar CFD results from all the utilized models for a kinetic-controlled condition, when the reactor was exposed to some degree of mass transfer limitation, AKN and RSM exhibited a slightly better performance. Duran et al. (2009) also highlighted the critical role of near-wall models coupling to S k–ɛ, R k–ɛ, and RSM turbulence equations for more reliable CFD predictions.  17  In terms of application and validity of CFD modeling for the UV irradiance distribution prediction, Duran et al. (2010b) demonstrated a very good agreement between CFD-simulated fluence rates and experimental results obtained from the chemical actinometry for annular flow through UV reactor. The authors also highlighted the importance of lamp power output and the effect of temperature on the accurate modeling of the radiation field within UV reactors. Similarly, Ho (2009) showed good consistency between the CFD-based radiation results obtained from discrete ordinate (DO) modeling and actinometry experiments for different locations of a batch annular UV reactor.  Apart from the individual hydrodynamic and irradiation sub-models assessment, the overall validation of CFD simulation for UV photoreactors was performed by the analysis of inlet and outlet concentration of target compounds. Using this approach, Alpert et al. (2010) performed the CFD study of a pilot-scale H2O2/UV AOP for degradation of an organic contaminant indicator (methylene blue). The authors highlighted the essential needs for accounting the role of water matrix constituents, which scavenge the •OH radical, in the simulation. Using a sensitivity analysis of the model input parameters, the authors found strong impact of the second reaction rate constant of the target contaminant with •OH radical on their model predictions. The results showed larger model deviation from pilot plant results by increasing the flow rate.  Recently, Santoro et al. (2010) utilized a two-dimensional CFD simulation for the study of two different target pollutants in H2O2/UV process under laminar and turbulent flow regime, as well as a wide range of UV doses. It was observed that CFD modeling estimated the degradation rate with good accuracy when compared to the experimental data. They also reported a better AOP performance for cross flow (e.g., lamp orientation perpendicular to flow direction) reactor compared to parallel configuration.   18  Accordingly, a few other researchers demonstrated the validity of CFD simulations for the performance evaluation of H2O2/UV process by simultaneous consideration of hydrodynamic, fluence (254 nm) and kinetic rate sub-models as well as the characterization of the background water. Despite the tremendous efforts made by prior studies (Alpert et al., 2010; Duran et al., 2011; Elyasi and Taghipour, 2010; Santoro et al., 2010; Wols et al., 2015), none of the earlier investigations have differentiated the role of  254 nm direct photolysis and the •OH-oxidation pathways in net removal of the target pollutant with their proposed model. This is even much more critical for the VUV/UV process in which three different degradation pathways contribute to the removal of target pollutant (e.g., 185 nm direct photolysis, 254 nm direct photolysis, and •OH oxidation). Furthermore, to best of our knowledge no CFD-based study of the VUV/UV or VUV process that incorporates the flow characteristics has been reported in the open literature. Not only does the presence of rigorously validated CFD model contribute to better understanding of the VUV photoreactors, but also it can be utilized as a cost-effective benchmark to facilitate their design, optimization and scale-up procedure.  1.3.2.2 Measurement of the 185 nm emission power of VUV lamps Knowing the lamp emissions at 185 nm VUV and 254 nm UV under different operating temperatures is the first step for developing an accurate prediction of photoreactors performance. Recently, Sasges et al. (2007) proposed a standard method for quantifying the 254 nm output of low pressure mercury lamps from a single irradiance measurement. Using the Lambertian emitter assumption for the UV lamp and Keitz model, they obtained accurate data for the lamp power output from a single irradiance measurement compared to 10-degree intervals data taken from the goniometric method. In addition, the authors recommended placing the UV detector normal to the lamp center at a distance greater than arc length. Based on this work, IUVA “International  19  Ultraviolet Association” has announced a standard protocol for the measurement of 254 nm lamp power output (Lawal et al., 2008). According to the IUVA protocol, the lamp and detector distance was selected at one meter.  Despite the presence of IUVA protocol for measuring the UV emission of lamps, no report has been released on the temperature dependency of UV lamp power output in the open literature. The only attempt is restricted to the study of Duran et al. (2010b) who plotted the normalized UV output versus the water temperature using chemical actinometry. It was observed that as the water temperature increases the UV lamp output increases until it reaches to a maximum, followed by a decaying trend. Similarly, to date no attempt for quantification of the 185 nm lamp power output has been reported in the open literature. In addition, studies on understanding the impact of lamp temperature on the UV and VUV lamp power output of the LPMA lamps are missing in the literature. 1.4 Knowledge gaps in VUV/UV research Several key points and knowledge gaps were identified after thorough review of the existing scientific literature on the VUV oxidation process. This thesis aims to address the first five research gaps as possible, in hopes of providing a better understanding of the science and application of VUV oxidation process for water treatment. Lack of a simulation tool  Despite some similarities between the H2O2/UV and VUV/UV AOPs, the CFD simulation of the VUV/UV AOP requires a different modeling approach due to the simultaneous propagation of UV and VUV photons that initiate a distinct governing kinetic scheme for the VUV/UV process. In this sense, more than 50 reactions (e.g., photochemical, equilibrium and radical reactions) have been reported for the VUV photolysis of water; only a few of those have been considered by the  20  H2O2/UV CFD studies (Gonzalez et al., 2004; Imoberdorf and Mohseni, 2012). Not only would a complete CFD simulation tool offer a better understanding of the process, but also it provides a cost-effective tool for the design and optimization of photoreactors (Wols, 2012), facilitating practical applications of the VUV/UV process. Further, once the model is validated experimentally for both laboratory and pilot-scale, it can be effectively applied for the comparative studies between the H2O2/UV and VUV/UV processes to explore the impact of reactor design, oxidant dose and water matrix on the AOPs performance. Large dark zone area in VUV photoreactors  It is known that high molar absorptivity of water at lower wavelengths (e.g., VUV) results in a steep gradient of VUV radiation and •OH radical in the proximity of quartz sleeve (Barrett and Mansell, 1960; Weeks et al., 1963). The absence of VUV radiation and consequently •OH radical in farther areas of the sleeve will generate large dark zone areas within the reactor which limit the performance VUV/UV process (Imoberdorf and Mohseni, 2011b). Hence, the presence of any mixing and circulation zones in proximity of the sleeve may lead to a greater efficacy of this process by improving the mass transfer and migration of pollutants toward the reacting zones. Even though mixing appears as an important parameter influencing the performance of diffusion-controlled systems such as VUV oxidation process, no theoretical or experimental study has explored this parameter. Continuous mode operation for a wide range of performance variables  Although the VUV/UV process has shown to be very effective in small bench-scale reactors, there are no reports on the effectiveness of this process under continuous operation that is the case in practical applications. More importantly, it would be valuable to understand the impact of water matrix constituents on the efficacy of continuously operating VUV/UV reactors. Given the wide- 21  scale application of H2O2/UV process, comparative studies which investigates the energy consumption and cost competitiveness of VUV/UV and H2O2/UV needs assessment under continuous flow operation is lacking among the literature. Similarly, comparing the extent at which reactor hydrodynamics (flow characteristics) affect the removal and energy efficiency of VUV/UV and H2O2/UV oxidation process can shed light on the design of improved reactors with less energy consumption. Unknown lamp power output  Recent studies show the promise of low pressure mercury amalgam (LPMA) lamps for both disinfection and advanced oxidation applications (IJpelaar et al., 2010; Oppenländer, 2003). However, there is no report in the open literature, which relates the lamp UV power to the lamp temperature. This need is even greater when it comes to 185 nm lamp power output for which no experimental data has been released in the open literature.  Lack of pilot scale study of the VUV/UV process for water treatment Pilot testing of a lab scale technology is a key step toward its practical implementations. To the best of our knowledge, no pilot scale study of the VUV process for water treatment has been reported in the open literature.  Impact of water temperature on the performance of VUV/UV photoreactors As discussed earlier, the UV lamp power output is a function of lamp temperature (Duran et al., 2010b; Wols et al., 2015). Thereby, it is recognized that water temperature could affect the lamp temperature and subsequently the UV output (Wols et al., 2015). This is even more significant for the VUV/UV process, where water absorption coefficient at 185 nm is strongly related to water temperature (e.g., up to 5% change for each 1°C temperature change) (Barrett and Mansell, 1960; Price et al., 1960; Weeks et al., 1963). That is, lower water temperature results in higher penetration  22  of the VUV photons in water. Indeed, very few studies have studied the VUV water absorption coefficient and its relationship with water temperature. Lack of VUV quantum yield data There is a lack of experimental information on the VUV quantum yield for a wide range of organic and inorganic water constituents. Currently, there is no standard method available for the experimental determination of this parameter. Impact of water turbidity on VUV absorption coefficient of water There is a lack of understanding on the impact of water turbidity on the VUV absorption coefficient of water. This is of paramount important for application of VUV technology for removal of OMPs from waters with different turbidity.    23  Chapter 2: Computational fluid dynamics (CFD) modeling of VUV/UV photoreactors for water treatment  2.1 Chapter introduction The use of photoreactors in water treatment applications has increased substantially over the past few years (Shannon et al., 2008). They are employed commonly in ultraviolet (UV) disinfection and UV-based advanced oxidation processes (AOPs). A particularly emerging and promising UV AOP is the VUV/UV process, which involves the use of ozone-generating mercury lamps that emit 185 nm Vacuum-UV (VUV) and 254 nm UV radiations. Compared to the other widely studied AOPs (e.g., O3/UV, H2O2/UV, TiO2/UV), the VUV/UV process offers the advantage of no chemical or catalyst addition, which in turn translates into simpler, and more sustainable operation along with potentially lower operating cost.  Many researchers have already reported the effectiveness of small scale batch VUV/UV photoreactors for the removal of a wide range of recalcitrant micropollutants (e.g., pesticides, herbicides, taste and odor compounds, pharmaceutical and personal care products) from water sources (Imoberdorf and Mohseni, 2012; Kutschera et al., 2009; Yuan et al., 2009; Zoschke et al., 2012). However, despite the many advantages of this technology and the promising results obtained through lab scale studies, there are still some factors that stymie the wide scale application of the VUV/UV photoreactors for water treatment. Lack of proper model and simulation tool for predicting and analyzing the performance of VUV/UV photoreactors is among the key factors hindering their practical implementation. Complete modeling of the VUV/UV process involves the simultaneous resolution of the local transfer equations of momentum, mass, and radiative energy (for UV and VUV radiations), along with a complex kinetic scheme with more than 40  24  reactions (Bagheri et al., 2013). Computational Fluid Dynamics (CFD) is an established and effective tool for modeling complex processes, and has been used extensively in recent years for the design, optimization and scale-up of UV disinfection and oxidation photoreactors (Alpert et al., 2010; Duran et al., 2010b; Duran et al., 2010a; Elyasi and Taghipour, 2010; Santoro et al., 2010; Taghipour and Sozzi, 2005; Wols, 2012). All studies indicated the importance of using a comprehensive kinetic scheme along with a detailed radiation model including the reflection/refraction/absorption of photons in the UV photoreactor.  However, none of the previous H2O2/UV CFD studies (Alpert et al., 2010; Elyasi and Taghipour, 2010; Santoro et al., 2010) differentiated the role of 254 nm direct-photolysis and the •OH oxidation pathways in net removal of the target pollutant. More importantly, despite some similarities between the H2O2/UV and VUV/UV processes, the mechanism by which hydroxyl radicals are generated is distinct for each (Gonzalez et al., 2004; Imoberdorf and Mohseni, 2011b, 2012). Although production of •OH in VUV/UV systems relies on the photolysis of water at 185 nm, hydrogen peroxide photolysis at 254 nm is the predominant mechanism for the generation of •OH in the H2O2/UV AOP (Gonzalez et al., 2004; Imoberdorf and Mohseni, 2011b, 2012). While photolysis of water forms •OH, H•, H+ and e-aq, the UV photolysis of hydrogen peroxide merely produces •OH, which introduces a different radical reaction scheme for each. While in H2O2/UV process, direct-photolysis of the target contaminants occurs dominantly at 254 nm wavelength, both 185 nm and 254 nm emissions contribute in the contaminants’ direct-degradation in VUV/UV process. All above underline the need for a modified modeling strategy for the VUV/UV AOPs.  The primary objective of this chapter was to develop a comprehensive CFD simulation tool for an in-depth analysis of the VUV/UV process applied to water treatment. The proposed computational model integrates different sub-models such as hydrodynamics, species mass transport, chemical  25  reaction kinetics, and irradiance distribution within the reactor. Furthermore, the impact of water temperature on its 185 nm VUV absorption coefficient was incorporated within the model. The radiation field within the reactor was modeled using the non-gray discrete ordinate (DO) sub-model, which allows for independent and simultaneous propagation of 185 nm VUV and 254 nm UV photons. In doing so, knowledge of UV and VUV outputs from the lamp at working (experimental) temperature was required and experimentally obtained. The developed model was evaluated experimentally using a bench scale flow-through prototype VUV/UV photoreactor, treating a model pollutant, para-chlorobenzoic acid (p-CBA). Utilizing the model, different degradation pathways within VUV/UV photoreactors were discussed. The results from this chapter will provide the water professionals with a cost-effective tool for the design and optimization of VUV/UV photoreactors, leading to the practical application of VUV/UV process.  2.2 CFD model development  2.2.1 Hydrodynamic and mass transfer model Assuming that the fluid (water mixture) is Newtonian, incompressible, isothermal, with constant physical properties and under laminar steady state flow, the hydrodynamic and species transport equations were as follows (Bird et al., 2007; Ranade, 2002): Mass conservation equation: ׏ ∙ ሺܸሻ ൌ 0  (2-1)Momentum conservation equation: ׏ ∙ ሺߩܸܸሻ ൌ െ׏ܲ െ ׏ ∙ ߬                            (2-2)where the stress tensor is: τ ൌ μሺ׏ܸ ൅ ׏்ܸሻ െ ଶଷ ߤ׏ ∙ ܸܷ  (2-3)Species conservation equation:  26  ׏ ∙ ሺߩܸ݉௜ሻ ൌ െ׏ ∙ ܬ௜ ൅ ܴ௜	,			݅ ൌ 1, 2, … ,ܰ െ 1 (2-4)where the diffusive flux of species i is estimated using Fick’s first law of diffusion: 	ܬ௜ ൌ െ	ܦ௠׏ሺߩ	݉௜ሻ (2-5)In Eqs. (1) – (5),   was density, V is velocity, P is pressure, τ is viscous stress tensor, μ is molecular viscosity, U is unit tensor, mi is mass fraction of species i, Ji is diffusive flux of species i, N is the total number of species, Ri is the source rate of species i (net reaction rate per unit volume) and Dm is the molecular diffusivity of species i in the mixture. It should be noted that the role of gravity was found negligible because of the short photoreactor length (< 30 cm). 2.2.2 Fluence model Calculation of the fluence rate field was carried out by solving non-gray discrete ordinate (DO) model which considers the integration of radiative transfer equation (RTE) over each wavelength interval . By doing so, the total fluence rate ܫሺݎԦ, ݏԦሻ in each direction ݏ	ሬሬԦ	at position ݎ	ሬሬԦ was calculated by the sum over the ݇	wavelength bands:   ܫሺݎ,ሬሬԦ ݏ	ሬሬԦሻ ൌ෍ܫఒೖሺݎ,ሬሬԦ ݏ	ሬሬԦሻ௞∆ߣ௞ (2-6)For the ozone-generating lamps, predominantly emitting at 185 nm VUV and 254 nm UV, ݇ was equal to 2. Moreover, for each wavelength band, ܫఒሺݎ,ሬሬԦ ݏ	ሬሬԦሻ is calculated by solving the RTE: ߘ ∙ ሺܫఒೖሺݎԦ, ݏԦሻ, ݏԦሻ ൅ ሺߙఒ൅σ௦ሻܫఒೖሺݎ,ሬሬԦ ݏ	ሬሬԦሻ ൌ ܨܽఒ݊ଶߪௌି஻ ܶସߨ ൅σ௦4ߨන ܫఒೖሺ ݎ,ሬሬԦ ݏሬሬԦ´ሻ߮ሺ	ݎ,ሬሬԦ	ݏ	ሬሬԦ´ሻ݀ସగ଴ߗ´ (2-7)where ܫ is the photon radiance, n is the refractive index, S-B is the Stefan-Boltzmann constant (5.67×10-8 Wm-2 K-4), ݎ	ሬሬԦ is the position vector, ݏ	ሬሬԦ is the propagation direction vector, ߙ is the spectral absorption coefficient,  is the scattering coefficient, ߮ is the phase function for the in-scattering of photons, and ߗ´ is the solid angle of the scattering direction vector ݏ	ሬሬԦ´. Additionally, ܨ is the fraction of radiant energy emitted in the wavelength interval of λ1 to λ2 at temperature ܶ  27  in a medium of refractive index	݊. A surface emission model was used to represent the lamp emission. Furthermore, the reflection/refraction/absorption of photons within the air gap (separating the lamp and the quartz sleeve region) and on the quartz sleeve was included in the radiation modeling. More details about the lamp emission model and its corresponding CFD-based modeling approach can be found elsewhere (Ho, 2009; Duran et al., 2010b). 2.2.3 Kinetic model The detailed kinetic model for the VUV systems has been previously verified by several studies using small batch-scale fully mixed photoreactors (Gonzalez et al., 2004; Imoberdorf and Mohseni, 2011b, 2012). There are three types of volumetric reactions happening in a VUV/UV photoreactor (e.g., equilibrium, photochemical, and radical reactions), as presented in Table 2-1 (Imoberdorf and Mohseni, 2012; Ross et al., 1994). Accordingly, the main degradation pathways of a model contaminant in the presence of VUV and UV radiations are 185 nm direct-photolysis, 254 nm direct-photolysis, and •OH radical chain reactions. Herein, to partly investigate the •OH scavenging effect of the degradation intermediates, the first generation oxidation by-products of the model contaminant were integrated in reaction scheme with kinetic constant rates similar to that of the main target contaminant (Table 2-1) (Gonzalez et. al., 2004; Ross et al., 1994; Wols and Hofman-Caris, 2012). The kinetic model was incorporated into the CFD using numerical subroutines (user-defined functions) to prescribe the net rate of generation shown in equation (2-4) via Ri for each chemical species. As for reaction simulation, based on the fully developed laminar flow regime in the reactor, the “laminar finite rate” model that ignores the effect of turbulent fluctuations on reaction rates was selected. However, since the laminar finite rate model utilizes the average species concentration in each cell, it is of paramount importance to ensure finer mesh grids for the areas with high concentration gradients .  28  Table 2-1 List of the reactions taking place in a VUV/UV photo-reactor for degradation of p-CBA No. Equilibrium reactions Constant 1 ܱଷ•	ି ൅ ܪା ↔ ܪܱଷ• pKa1 = 8.2 2 ܪଶ0ଶ ൅ ܱܪି 	↔ ܪଶ0 ൅ ܪܱଶି pKa2 = 11.6 3 ܪܱ• ൅ ܪܱି 	↔	ܪଶܱ ൅ ܱ•ି pKa3 = 11.9 4 ܪܱଶ• 	↔ ܪା ൅ ܱଶ•ି pKa4 = 4.8 5 ܱܪି ൅ ܪା 	↔	ܪଶܱ Kw = 10-14 Photochemical reactions 6 ܪଶܱ ൅ ݄νଵ଼ହ	௡௠ 	 	→	ܪܱ• ൅ ܪ•   Φ6 = 0.33  mole.ein-1 7 ܪଶܱ ൅ ݄νଵ଼ହ	௡௠ 	 	→	ܪା ൅ ݁ି௔௤ ൅ ܪܱ•   Φ7 = 0.045 mole.ein-1 8 ܪଶܱଶ ൅ ݄νଵ଼ହ	௡௠ 	 	→	2	ܪܱ•   Φ8 = 0.5  mole.ein-1 9 ܪଶܱଶ ൅ ݄νଶହସ	௡௠ 	 	→	2	ܪܱ•   Φ9 = 0.5  mole.ein-1 Contaminant reactions 10 ݌ܥܤܣ ൅ ݄νଶହସ	௡௠ 	 	→ ܥ଻ܪହܥ݈ܱଶ െ 254 ሺ∗ ݌ܥܤܣ254ሻ   Φ10 = 0.013  mole.ein-1 11 ݌ܥܤܣ ൅ ݄νଵ଼ହ	௡௠ 	 	→ ܥ଻ܪହܥ݈ܱଶ െ 185 ሺ∗ ݌ܥܤܣ185ሻ *Φ11 = 0.013  mole.ein-1 12 ܥ଻ܪହܥ݈ܱଶሺ݌ܥܤܣሻ ൅ ܪܱ• → ∗ ݌ܥܤܣ െ ܱܪ   k12 = 5.0×109   M-1.s-1 13 ݌ܥܤܣ254 ൅ ܪܱ• 	 	→	∗ ݌ܥܤܣ െ ܱܪ *k13 = 1 ×109  M-1.s-1 14 ݌ܥܤܣ185 ൅ ܪܱ• 	 	→	∗ ݌ܥܤܣ െ ܱܪ *k14 = 1 ×109  M-1.s-1 15 ݌ܥܤܣ െ ܱܪ ൅ ܪܱ• 	 	→	∗ ݌ܥܤܣ െ ܱܪܱܪ *k15 = 1 ×109   M-1.s-1 Other involving reactions 16 ܪܱ• ൅ ܪଶ 	 	→	ܪଶܱ ൅ ܪ•  k16 = 3.9×107    M-1.s-1 17 ܪܱ• ൅ ܪܱ• 	 	→	ܪଶܱଶ k17 = 4.2×109     M-1.s-1 18 ܪܱ• ൅ ܪ• 	 	→	ܪଶܱ k18 = 7×109      M-1.s-1 19 ܪܱ• ൅ ݁ି௔௤ 	 	→	ܪܱି k19 = 3×1010     M-1.s-1 20 ܪܱ• ൅ ܪܱଶ• 	 	→	ܪଶܱ ൅	ܱଶ k20 = 6.6×109   M-1.s-1 21 ܪܱ• ൅ ܱଶ•ି 	 	→	ܱଶ ൅ ܱܪି k21 = 1.1×1010  M-1.s-1 22 ܪܱ• ൅ ܪܱଶି 	 	→	ܱଶ•ି	 ൅ ܪଶܱ k22 = 6.8×109   M-1.s-1 23 ܪܱ• ൅ ܪଶܱଶ 	 	→	ܪܱଶ• ൅ ܪଶܱ k23 = 2.7×107   M-1.s-1 24 ܪܱଷ• 	→	ܪܱ• ൅	ܱଶ k24= 1.1×105    s-1 25 ܱଷ• ൅ ܪା 	 	→	ܪܱ• ൅	ܱଶ k25 = 9×1010     M-1.s-1 26 ݁ି௔௤ ൅ ܪଶܱଶ 	 	→	ܱܪି ൅ ܪܱ• k26 = 1.3×1010   M-1.s-1 27 ܪଶܱଶ ൅ ܪ• 	 	→	ܪܱ•ି ൅ ܪଶܱ k27 = 5×107      M-1.s-1 28 ܪܱଶ• ൅	ܪ• 	 	→ 	ܪଶܱଶ k28 = 2×1010     M-1.s-1 29 ܪܱଶ• ൅	ܱଶ•ି ൅ ܪଶܱ	 	→	ܪଶܱଶ ൅ ܱଶ ൅ ܪܱି k29 = 9.7×107     M-1.s-1 30 ܪܱଶ• ൅ ܪܱଶ• 	 	→	ܪଶܱଶ ൅ ܱଶ k30 = 8.3×105   M-1.s-1 31 ݁ି௔௤ ൅ ܱଶ 	 	→	ܱଶ•ି k31= 1.8×1010   M-1.s-1 32 ݁ି௔௤ ൅ ܪଶܱ	 	→	ܱܪି ൅ ܪ• k32 = 1×103        s-1 33 ݁ି௔௤ ൅ ܪ• ൅ 	ܪଶܱ	 	→	ܱܪି ൅ ܪଶ k33 = 3.4×1010   M-1 .s-1 34 ݁ି௔௤ ൅	݁ି௔௤ ൅	2	ܪଶܱ	 → 2 ܱܪି ൅ ܪଶ k34 = 6×10 9      M-1 .s-1 35 ݁ି௔௤ ൅ ܪା 	 	→ ܪ• k35 = 2.8×1010  M-1 .s-1 36 ݁ି௔௤ ൅ ܪܱଶି 	 	→	ܱܪି ൅ ܱ•ି k36 = 3.5×109   M-1 .s-1 37 ܪܱି ൅ ܪ• 	→	݁ି௔௤ ൅	ܪଶܱ k37 = 2.5×107   M-1 .s-1 38 ܱଶ ൅ ܱ•ି 	 	→	ܱଷ•ି k38 = 3.8×109   M-1 .s-1 39 ܱଶ•ି ൅ ܪ• 	→	ܪܱଶି k39 = 2.7×107   M-1 .s-1 40 ܱଶ ൅ ܪ• 	 	→	ܪܱଶ• k40 = 1.2×1010  M-1 .s-1 41 ܱ•ି ൅	ܪܱଶି 	 	→	ܱܪି ൅	ܱଶ•ି k41 = 4×108      M-1 .s-1 42 ܪ• ൅ ܪ• 	 	→	ܪଶ k42 = 5.5×109   M-1 .s-1  Starred reactants/products are hypothetical and are only defined for model development.  Starred values are estimated.    29  2.2.4 Geometrical models and mesh structure For the purpose of simplicity and because of radial symmetry, an axisymmetric two-dimensional geometry was considered (Figure 2-1). Additionally, only half of the shown 2D geometry was used for the CFD computation due to the symmetry across the middle horizontal plane. A boundary-layer mesh was setup on the walls, particularly on the sleeve surface where high concentration gradients occur. ANSYS® 14.5 simulation packages were used to build and discretize the problem domain via 15418 structured hexahedral cells. Using the velocity, radiation, and contaminant species concentration values at several locations of the reactor, the utilized grids were verified to provide mesh-independent results. The solution must be grid-independent.     Figure 2-1 Reactor geometry utilized for 2D model development with dimensions 2.2.5 Boundary conditions The inlet concentrations of p-CBA, as model contaminant, and dissolved oxygen were set at 0.30 and 9.2 ppm, respectively. At the inlet, the mass flow rate boundary condition with normal direction to the boundary was set. A fully developed flow condition was applied for the outlet region. A no-slip boundary condition was imposed on the walls. In addition, zero diffusive flux of species was specified on the walls. As per radiation field boundary conditions, the lamp plasma was defined as a zero-thickness wall, semi-transparent and fully specular (Duran et al., 2010b). The lamp end caps were semi-transparent, fully diffused absorbing walls. The suprasil quartz sleeve was set as a 1.5 mm-thick, semi-transparent, fully specular wall. While water and air were 7 cm 2.5 cm 27.8 cm 2.8 cm 1.5 cm Air gap Water Lamp Quartz sleeve Symmetry plane 30  defined as a fluid continuum, all the remaining walls were specified as zero-thickness, opaque, fully diffused, non-reflective walls (Duran et al., 2010b).  2.2.6 Physical properties Because of very low concentrations of model contaminant utilized in this study (~0.3 mg/L), constant values were defined for the viscosity (µ=1×10-3 Pa.s) as well as for the refractive index of water (1.458 and 1.375 for 185 nm and 254 nm wavelengths, respectively). For the suprasil quartz sleeve, the refractive index was assigned at 1.575 and 1.505 for 185 and 254 nm wavelengths, respectively (Daimon and Masumura, 2007). To calculate the mixture diffusion coefficient of the solution, dilute approximation theory which accounts for each species diffusion in pure water was utilized (Cussler, 2009). Moreover, to account for the dependency of 185 nm water absorption coefficient on temperature (+0.05 cm-1/ºC) (Barrett and Mansell, 1960), a linear function which corresponds to 1.4 cm-1 at 25ºC (Barrett and Mansell, 1960; Halmann, 1966) was coupled to the radiative transfer equation. Assuming a constant temperature for the air trapped between the sleeve and water, decadic absorption coefficient of oxygen was set to 0.166 atm-1       cm-1 (Creasey et al., 2000). Molar extinction coefficient of p-CBA (pH~7) at 185 nm and 254 nm were measured to be 21380 M-1cm-1 and 2450 M-1cm-1, respectively. 2.2.7 Numerical solution method and strategy ANSYS® 14.5 Fluent was employed to read the mesh and perform the CFD computations. The segregated steady-state solver was used to solve the governing equations. Second order upwind discretization scheme was applied except for pressure for which the standard scheme was selected. The SIMPLE algorithm was chosen for the pressure-velocity coupling. An angular discretization of 256 divisions was utilized for solving the RTE. This number was found to be sufficient to avoid the appearance of the “ray effect” (Duran et al., 2010b); and to overcome control angle overhang,  31  4×4 pixelation was used . Convergence of the numerical solution was assured by monitoring the scaled residuals to a criterion of at least 10-6 for the continuity, momentum, concentration and radiation variables. Additionally, the variation of velocity magnitude, model contaminant concentration, and radiation fluence rate at several points of the computational domain were used as indicators of convergence. The total execution time of each transient simulation was about 4 h on a 3.50 GHz AMD Phenom™ II X4 970 processor with 8 GB RAM. With consideration of the facts that in this problem velocity and radiation fields did not interact, the CFD model was solved in two steps. First, the RTE equation was solved for obtaining the radiation field across the computational domain. Second, keeping the converged radiation field solutions “frozen”, the mass, momentum conservations coupled with the species equations were solved for computing the velocity and concentration profiles within the system. This solving strategy saves computation time and brings stability to the solution (Duran et al., 2009).  2.3 Experimental setup and procedure 2.3.1 Micropollutant degradation for model evaluation  To evaluate experimentally the proposed CFD model, a flow through continuously operating VUV/UV photoreactor was used (Figure 2-2). The photoreactor was made of Plexiglas, with an annular configuration, operated with a 42 W low pressure mercury amalgam ozone-generating lamp (GPHVA357T5VH, Light Sources Inc.), longitudinally placed at the axial center of the reactor. The internal and external diameters of the annular photoreactor and synthetic quartz sleeve (Heraeus Quartz America LLC, SUPRASIL310) were 7 cm and 2.8 cm, respectively, giving a hydraulic diameter of 4.2 cm for treating the water. The presence of a large distance between the reactor inlet and lamp (e.g., ~ 40 cm) assured the assumption of fully developed flow in the reactor. The lamp was equipped with K type thermocouple wire (OMEGA®, US) for the temperature  32  monitoring at the amalgam position. In addition to monitoring the amalgam location temperature (T1), four additional K type thermocouple sensors were employed to monitor the temperatures of the lamp envelope surface (T2), trapped air inside the sleeve (T3), quartz sleeve surface (T4), and water at 0.5 cm distance from the reactor body (T5).  Figure 2-2 Schematic of the VUV experimental set-up 2.3.2 Determination of lamp power output The 185 nm and 254 nm outputs of the VUV lamp were measured using a research radiometer (IL 1700, SED185 gold cathode detector; SED240 sensor, and NS254 filter), calibrated by International Light Technologies before the experiments. In doing so, a custom-made wooden box, painted black inside, with outer dimensions of 118 cm×38 cm×28 cm and total inner volume of 95 L was fabricated (Figure 2-3). The detector was placed and sealed at the center of one of the walls, 100 cm away from the lamp axis (M. Sasges, A. Pol, A. Voronov, J. Robinson, 2007). The Data analysis Quartz sleeve InletOutlet TMVUV Lamp FM 33  radiometer reading (radiation intensity) was translated to lamp power using the mathematical model based on Keitz equation (Sasges et al., 2012). Since oxygen absorbs the 185 nm radiation and initiates a series of radical reactions through ozone formation, the apparatus was purged continuously with high-purity nitrogen gas over the course of experiments. Furthermore, two battery-operated fans (O2 Cool®) provided circulation and prevented the formation of dead zones. Ozone indicator badges (Epak Electronics Ltd., UK) were used at the outlet to ensure no detectable concentration of ozone was accumulated inside the box. Moreover, purging nitrogen through the sleeve allowed studying the temperature dependency of the lamp outputs. Three K type thermocouples were utilized for the temperature monitoring of amalgam position (T), lamp envelope surface (T"), and sleeve surface (T"). In all the experiments, the lamp was allowed to run for approximately 30 min to ensure its stability before collecting any data (Lawal et al., 2008). It is worth mentioning that in order to account for the net emission power; the reflection of the lights from the walls was measured and subtracted from the calculated values.   Figure 2-3 Schematic of the experimental apparatus for the 185 nm and 254 nm lamp power measurement  TM Inlet Inlet Outlet  34  2.3.3 Analytical methods The concentration of p-CBA was directly analyzed with HPLC (Dionex UltiMate™ 3000, US). Separation was accomplished using a reverse phase Nova Pak C18, 4 m 3.9150 mm separator column. A sample volume of 20 L with a flow rate of 0.45 mL/min (55% acetonitrile/45% water/1% phosphoric acid) was injected to the system and the detection wavelength was set at 235 nm. As for pH adjustment, phosphate buffer solution, a mixture of monosodium phosphate and disodium phosphate (Sigma-Aldrich, Canada), was employed for buffered water samples. The concentration of hydrogen peroxide was determined via UV spectrophotometry utilizing the ܫଷି  method (Klassen et al., 1994). Molar extinction coefficient of p-CBA solutions at 254 nm was measured with a UV–vis spectrophotometer (Cary 100), using 1 cm thickness quartz cell. A VUV-UV–vis spectrophotometer (Cary 4000 Varian) with 0.1 cm suprasil quartz cell was utilized for determining molar extinction coefficient at 185 nm. In this case, the spectrophotometer container was required continuous purging with high purity nitrogen gas.   2.4 Results and discussion 2.4.1 Lamp power output Figure 2-4 displays the 254 nm and 185 nm emissions of the VUV lamp, respectively, versus the lamp amalgam temperature obtained from the wooden box apparatus (Figure 2-3). Of all monitored temperature locations, the amalgam temperature showed greater impact on the lamp emission output. In this case, while the lamp amalgam temperature varied between 52 ºC to 95 ºC, the corresponding lamp surface temperature was altered approximately from 90 ºC to 125 ºC, which represented a narrower temperature range. To determine the effect of light reflection from sidewalls (Figure 2-3), experiments were conducted, which showed less than 4% and 7% reflection  35  share for 185 nm and 254 nm, respectively. This was done by placing a black painted screen (e.g., 5cm×32cm) in the middle of the wooden box to block the directly emitted radiations from the lamp and only allowed the reflected radiations to reach the detector. According to Figure 2-4a, the 254 nm lamp power increased with the increase in the amalgam temperature until it reached a maximum, followed by a decaying trend. Similar observation was reported by Duran et al. (2010b) who plotted the 254 nm lamp output versus the water temperature. Meanwhile, the 185 nm power of the VUV lamp showed an ascending trend within the overall studied temperature range (Figure 2-4b). Such trend is justifiable with the existence of an optimal mercury vapor pressure (e.g., 0.01 mbar) in which the maximum mercury resonance radiation within UV lamps occurs (Schalk et al., 2006).   8.08.59.09.510.010.511.011.512.050 55 60 65 70 75 80 85 90 95 100254 nm lamp output power (W)Amalgam temperature (°C)  VUV lamp (a) (Error bars indicate 1% standard deviations)   36   Figure 2-4 (a) 254 nm, (b) 185 nm output of the VUV lamp versus the amalgam temperature The operating amalgam temperature of the VUV lamp during the degradation experiments (Figure 2-2) was approximately 65±1.0 ºC depending upon the flow rates of water (e.g., 0.5-6.7 L/min). Based on the results of Figure 2-4 and operating amalgam temperature of the experiments, the 185 nm and 254 nm lamp power outputs were calculated as 0.45±0.05 W and 10.5±0.2 W, respectively. These values were used as inputs for the CFD modeling of the VUV/UV photoreactor.  2.4.2 Model evaluation Figure 2-5 presents the CFD model prediction along with the experimental data for the degradation of 300 ppb p-CBA in MilliQ water under a range of flow rates. It should be noted that a sensitivity analysis was also conducted to ensure that the model predictions were insensitive to small variations within the reported accuracy of the kinetic rate constants. Based on the utilized experimental apparatus, the Reynolds number was in the range of 100 (~ 0.5 L/min) to 1500 (~ 6.7 0.200.250.300.350.400.450.500.550.6050 55 60 65 70 75 80 85185 nm lamp power output  (W)Amalgam temperature (°C)  VUV lamp (b) (Error bars indicate 3% standard deviations)  37  L/min) indicating a fully laminar flow regime in the photoreactor. As shown in Figure 2-5, it is evident that the utilized VUV/UV photoreactor was able to provide a nearly complete degradation of the p-CBA at very low flow rates (<0.4 L/min). Increasing the flow rate to moderate and/or higher values, however, led to lower p-CBA degradation.  Overall, the CFD model predicted the experimental data very well, with some deviations observed at lower flow rates. The greater discrepancy in the model prediction for p-CBA removal at low flow rates may attributed to higher scavenging contribution by p-CBA degradation by-products. Although the model accounted for •OH scavenging impact of the first-generation by-products of p-CBA, multiple unknown or undefined intermediates and by-products may arise from an increasing concentration of first-generation degradation products at lower flow rates.  Figure 2-5  Comparison of experimental data obtained by the VUV/UV experimental set-up and CFD model 0.00.10.20.30.40.50.60.70 1 2 3 4 5 6 7C/C0Flow rate (L/min) Experimental data CFD model(Error bars represent the standard deviation from the mean of three independent experiments)  38  In order to gain better understanding of the performance efficacy of the VUV/UV process at different flow rates, the local concentration contours of p-CBA along the photoreactor were analyzed (Figure 2-6). The higher the flow rates, the smaller the dose of 185 nm radiation and consequently lower •OH concentration. As depicted in Figure 2-6, increasing the flow rate rapidly reduced the active reactor volume to a thin-film layer of water passing by the proximity of quartz sleeve. In this sense, the concentration of the contaminants at the reactor outlet may assumed by mixing of two separate steams: a highly purified stream (passing through the reacting zone) and a poorly purified stream (by passing the reacting zone). Therefore, mass diffusion of the target contaminant to the reacting zone is the key step that controls the degradation rate of the target pollutant in the VUV/UV photoreactor.  Figure 2-6 Parts per billion (ppb) p-CBA concentrations in the VUV/UV photoreactor at different flow rates 300285 270 255 240 225 210 195 180 165 150 135 120 105 90 75 60 45 30 15 0 0.5 L/min, Re =102 1.5 L/min, Re =319  3.4 L/min, Re =745 6.7 L/min, Re =1457  Flow direction 2.1cm  39  2.4.3 UV and VUV fluence rate distribution Figure 2-7 demonstrates the 254 nm and 185 nm fluence rate distributions along a perpendicular axis bisecting the longitudinal axis of the VUV/UV photoreactors, respectively. As shown in Figure 2-7a, the 254 nm incident radiation decreased with increasing radial distance from the lamp because of absorption in water (UVT=98%, resulting in negligible effect) and the increasing circumferential area. In contrast, attenuation of radiation in pure water was strong at 185 nm, leading to significant attenuation of radiation within a small water layer path (Figure 2-7b). The more significant decay of the 185 nm radiation was mainly due to the very high absorption coefficient of water at this wavelength (1.4 cm-1 at 25 °C) (Barrett and Mansell, 1960; Halmann, 1966) which resulted in a large dark zone area in the VUV/UV photoreactor. Therefore, the 185 nm direct-photolysis of the target contaminant occurred predominantly in the area with less than 1 cm distance from the quartz sleeve.   02004006008001000120014005 10 15 20 25 30 35UV incident radation (W/m2 )Distance (mm)254 nm incident radiation profileWater sectionAir sectionSleeve wall(a) 40   Figure 2-7 (a) 254 nm, (b) 185 nm incident radiation profile along the central axis of the VUV/UV photoreactor 2.4.4 Role of different degradation pathways To compare the role of each degradation pathway in the net degradation of p-CBA, the kinetic rate equations of the main degradation pathways are presented in Table 2-2. As presented, the 185 nm and 254 nm direct-photolysis were greatly dependent on the intrinsic optical property of the target pollutant as well as the fluence rate values. As such, the product of quantum yield and molar extinction coefficient can well describe the susceptibility of the target pollutant to direct-photolysis. For example, the quantum yield and molar extinction coefficient of p-CBA at 254 nm are 0.013 (mol. ein-1) and 2370 (M-1cm-1) (Rosenfeldt and Linden, 2007); whereas the corresponding values for Atrazine (a commonly investigated micropollutant) are 0.05 (mol. ein-1) and 3860 (M-1cm-1) (Nick et al., 1992), respectively. In this case, Atrazine is about 6.3 times more susceptible to 254 nm direct-photolysis compared to p-CBA.  01020304050605 10 15 20 25 30 35VUV incident radation (W/m2)Distance (mm)185 nm incident radiation profileWater sectionAir sectionSleeve wall(b) 41  Table 2-2 Kinetic rate equations for the p-CBA main degradation pathways Degradation pathways Kinetic rate equation ࡯ૠࡴ૞࡯࢒ࡻ૛ሺ࢖࡯࡮࡭ሻ ൅ ࡴࡻ⦁	→	࢖࡯࡮࡭ െ ࡻࡴ ݎ௣஼஻஺ ሺுை.ሻ ൌ ݇ ൈ ܥுை•	 ൈ 	ܥ௣஼஻஺	 ࢖࡯࡮࡭ ൅ ܐૅ૛૞૝	࢔࢓ →	∗ ࢖࡯࡮࡭૛૞૝ ݎ௣஼஻஺ ሺଶହସ ௡௠ሻ ൌ ߶ଶହସ ൈ ߝଶହସ	 ൈ ܫ	ଶହସ	 ൈ ܥ௣஼஻஺  ࢖࡯࡮࡭ ൅ ܐૅ૚ૡ૞	࢔࢓ →	∗ ࢖࡯࡮࡭૚ૡ૞ ݎ௣஼஻஺ ሺଵ଼ହ ௡௠ሻ ൌ ߶ଵ଼ହ ൈ ߝଵ଼ହ	 ൈ ܫଵ଼ହ	 ൈ ܥ௣஼஻஺  * Φ: Quantum yield; ɛ: molar extinction coefficient; I: Incident radiation. Figure 2-8 shows volume-weighted average kinetic rates of the p-CBA main degradation pathways within the VUV/UV photoreactor. The •OH degradation (oxidation) rate of p-CBA was approximately 8 to 15 times larger than that associated with 254 nm direct-photolysis and about  800 to 1200 times greater than the rate of 185 nm direct-photolysis depending on the flow rates. Comparing the presented values revealed •OH-oxidation as the predominant (controlling) mechanism contributing in the p-CBA degradation regardless of flow rate values.   Figure 2-8 Volume-weighted average kinetic rate for the main degradation pathways versus flow rates 020000400000.51.43.46.7720375810002100290033009000200003400049000 185 nm  photolysis 254 nm photolysis•OH Oxidation10-12 (M-1.s) 42  Through conducting two different “What-if” scenarios for zero UV and zero VUV emissions from the lamp, it was determined that the sole emission of the VUV radiation acted similarly, particularly at higher flow rates, to that obtained by the combined UV and VUV emissions in terms of the p-CBA removal efficacy. As for the sole emission of UV photons, the highest removal took place at low flow rates (e.g., the higher UV dose) where the initial concentration of the p-CBA dropped by 36% and 21% at 0.5 and 0.9 L/min, respectively.  2.4.5 Species distribution 2.4.5.1 •OH radical distribution Given the key role of •OH concentration, it was important to investigate the profile and distribution of •OH radical within the VUV/UV photoreactors. Figure 2-9 shows the •OH radical concentration contours along the reactor for the lowest and highest studied range of flow rates. As displayed, the •OH is densely concentrated in a short distance from the sleeve, varying from 1.5×10-9 to        7.5×10-11 M. The longer the retention time (lower flow rates), the higher the concentration of •OH from direct-photolysis of water molecules. The volume-weighted average •OH concentrations were computed to be 8.84×10-11 and 4.07×10-11 M for 0.5 and 6.7 L/min, respectively. Whilst considerable •OH concentration was still present at 6.7 L/min flow rate, significant concentration of the initial p-CBA (e.g., 38%) remained unreacted. This underscored the determining role of mass transfer on the AOP performance of the VUV/UV photoreactor.  To assess the contaminant exposure to hydroxyl radicals, (Sozzi and Taghipour, 2006b), •OH dose distributions for the 0.5 and 6.7 L/min flow rates were computed using a Lagrangian particle tracking approach (Equation 2-8). Doing so, uniformly distributed particles were released at the reactor inlet (presenting the target contaminant), and the received •OH dose was integrated along the particle trajectories. The delivered dose at each point was calculated by multiplying the average  43  local •OH concentration by the period Δt that the particle was exposed to that certain local concentration of •OH radical (Figure 2-10).  Figure 2-9 Concentration contours of ⦁OH radical within the VUV/UV photoreactor (M) ⦁OH	dose	delivery ൌ න ሾ	⦁OH	ሿ	݀ݐ௧଴ (2-8)As illustrated in Figure 2-10, an order of magnitude lower •OH dose was received at higher flow rate (e.g., 6.7 L/min) where contaminants predominantly passed quickly through regions of lower •OH concentration. This further explains the lower removal efficacy obtained at higher flow rates. The •OH dose distributions suggest that an enhancement in the removal efficacy of the process will most likely result from improved reactor hydrodynamics where greater number of particles pass through the area in proximity of the quartz sleeve (e.g., rich in ⦁OH concentration).  0.5 L/min, Re = 102 6.7 L/min, Re = 1457 1.5e-9              1.3e-9                 1e-9                 8.3e-10                6e-10               3.8e-10                      7.5e-112.1 cm  44    Figure 2-10 OH radical dose distribution within the VUV/UV photoreactor 2.4.5.2 Hydrogen peroxide  H2O2 formed in the VUV/UV reactor primarily through the recombination reaction of OH radicals. At the same time, H2O2 consumed by the reaction with •OH (acting as scavenger). Further, 185 nm and 254 nm direct-photolysis of H2O2 promoted formation of •OH radical. Comparing the average rates for •OH formation and consumption on the involving hydrogen peroxide reactions for the studied case, the •OH consumption rate of H2O2 exceeded its 01020304050607080-2.7E-10 2.7E-10 4.0E-08 8.0E-08 1.2E-07 1.6E-07 2.0E-07 2.4E-07 2.8E-07 3.2E-07 3.6E-07 4.0E-07Frequency %•OH dose delivery (M.s)(a) Flow rate 0.5 L/min0102030405060708090100-1.2E-11 1.2E-11 1.9E-09 3.9E-09 5.8E-09 7.7E-09 9.6E-09 1.2E-08 1.3E-08 1.5E-08 1.7E-08 1.9E-08Frequency %•OH dose delivery (M.s)(b) Flow rate 6.7 L/min 45  •OH formation. Model predictions for the outlet concentration of H2O2 were between 48.6 and 12.8 ppb for 0.5 and 6.7 L/min flow rate, respectively. Using ܫଷି  method, the experimentally measured H2O2 concentrations at the outlet were in agreement with those of CFD predictions; less than 90 ± 50 ppb H2O2 was detected at the lowest flow rate.  2.4.5.3 Degradation by-products  Experimental identification/quantification of all the degradation by-products and intermediates of the target contaminant formed through the VUV/UV process is a formidable task beyond the scope of this work. Based on earlier studies, it is evident that the first-generation byproducts undergo •OH-oxidation with reaction rate constants similar to that of the main target contaminant (e.g. 1×109 M-1.s-1 (Wols and Hofman-Caris, 2012). Although the proposed CFD model was not validated experimentally for an accurate prediction of degradation by-products, it can be used as an approximate tool for obtaining valuable insights on the first-generation degradation products (Benitez et al., 2013). This can shed light on the relative importance of the degradation products of an intended target contaminant through VUV/UV induced treatment. Utilizing the developed model, the outlet estimated concentration of p-CBA by-product obtained from the 185 nm direct-photolysis was less than 1 ppb over the entire studied flow rate. Similarly, the outlet concentration of p-CBA by-product from 254 nm direct-photolysis varied between 22 and 7 ppb for 0.5 and 6.7 L/min, respectively. These results indicate little concern over the concentration of byproducts from the 185 nm direct-photolysis pathway.     46  2.5 Chapter conclusions In this chapter, a CFD model for simulating VUV/UV photoreactors was developed and evaluated against experimental data. The performance of the model was assessed in terms of its capability for predicting the VUV/UV induced degradation rate of p-CBA, as a model contaminant. Furthermore, to the 185 nm VUV and 254 nm UV lamp emissions from an ozone-generating lamp were experimentally measured at different lamp temperature and used as inputs for model development. The CFD model showed to predict successfully the degradation rate of p-CBA while compared to that of experimental results. The performed analysis confirmed that mass transfer in proximity of the quartz sleeve limits strongly degradation rates of organic contaminants in VUV/UV photoreactors. Comparing the direct-photolysis of the model contaminant at 254 nm and 185 nm vs. •OH radical based degradation of the contaminant revealed the later one as the most dominant degradation pathway in the VUV/UV photoreactor. The •OH dose distribution's analysis underlined that an enhancement in the AOP performance of the VUV/UV process will most likely result from an improved reactor. Therefore, this chapter highlights the demanding need for investigating the impact of reactor hydrodynamics (e.g., integration of retrofitting baffles) on the removal efficacy of VUV/UV process.     47  Chapter 3: A study of enhanced performance of VUV/UV process for the degradation of micropollutants from contaminated water   3.1 Chapter introduction Contaminants of emerging concern (CECs) are increasingly recognized as potential threats to drinking water quality (Halden, 2010; Richardson and Ternes, 2011). Ultraviolet (UV) based advanced oxidation processes (AOPs) have, over the past decade, demonstrated great promise for the treatment of CECs in water. Vacuum-UV (VUV)/UV process is an incipient competitive UV AOP that eliminates the need for an auxiliary chemical oxidant, e.g. hydrogen peroxide, resulting in a greener and simpler process with great potential for commercial applications (Bagheri and Mohseni, 2014; Bagheri et al., 2013). The VUV/UV technology relies on the synergistic emissions of 185 nm VUV and 254 nm germicidal UV radiations emitted from ozone-generating lamps. The efficiency and mechanisms of the VUV/UV process for the removal of a wide range of CECs have been studied by several researchers (Arany et al., 2013; Duca et al., 2013; Imoberdorf and Mohseni, 2012; Kim and Tanaka, 2009; Zoschke et al., 2014). For example, Imoberdorf and Mohseni (2012) presented a complete kinetic model for the description of VUV/UV induced degradation of a model herbicide (2,4-dichlorophenoxyacetic acid) in absence and presence of the main •OH scavengers (e.g., NOM, alkalinity). For both conditions, the authors demonstrated a good agreement between the model predictions and experimental results obtained from a mixed batch VUV/UV reactor. Recently, Arany et al. (2013) reported a comparative study of VUV (172 nm), VUV/UV (185 nm + 254 nm) and UV (254 nm) processes for the degradation of an anti-inflammatory drug (Naproxen) in a recirculating batch reactor. Among the investigated processes, VUV/UV AOP showed the greatest removal ratio for the target contaminant. Despite the many  48  advantages of this technology and the promising results obtained through lab-scale batch experiments, there are still some factors that hinder the wide-scale application of the VUV/UV photoreactors for water treatment. Lack of studies demonstrating the removal efficacy of the VUV/UV process under continuous flow reactors are among the key factors hindering their practical implementation. This is of paramount importance since the very high absorption coefficient of water for 185 nm radiation limits the oxidation power of the process to a thin film optical path (e.g., less than 1 cm) (Barrett and Mansell, 1960; Price et al., 1960). Furthermore, the extent to which background water constituents limit the AOP performance of VUV/UV systems needs analysis under continuous flow operation. Meanwhile, the many design factors associated with continuously operating flow-through VUV/UV photoreactors make experimental investigation of the process very time and resource-intensive.  An effective approach to overcoming pertinent design issues is the application of computational fluid dynamics (CFD) modeling. In the past, CFD simulation has been extensively applied for design exploration studies in UV disinfection and UV AOPs (Crapulli et al., 2014; Duran et al., 2010b; Duran et al., 2011; Elyasi and Taghipour, 2010; Santoro et al., 2010; Sozzi and Taghipour, 2006b; Wols et al., 2012). For example, Wols et al. (2012) employed CFD simulation for the design exploration study of multi-lamp UV photoreactors for the removal of a model herbicide (atrazine). Several design aspects (e.g., placing mirrors, determining the lamp size and placing reactors in series) were considered and the results underlined the importance of hydrodynamic characteristics of the UV photoreactors in their treatment efficacy. To date, no similar computer-aided design study that investigates some of the main design aspects of the VUV/UV photoreactors has been reported.  49  The primary objective of this chapter was to perform a computer-aided design exploration of flow-through, continuously operating, VUV/UV photoreactors for the effective remediation of CECs from water sources. This included sensitivity analysis of the main design parameters of annular VUV/UV reactors (hydraulic diameter, illuminated surface area, and incorporation of central baffles) through application of CFD simulation. Among the explored design scenarios, an optimal scenario was selected based on the electrical energy-per-order (EEO) analysis and was evaluated experimentally for the treatment of atrazine, as a model contaminant. The approach and results presented here will provide the water professionals with a cost-effective methodology for the analysis and design-exploration of the flow-through VUV/UV photoreactors, leading to the practical application of the VUV/UV process. 3.2 Materials and methods 3.2.1 Experimental setup and procedure Experimental tests were conducted in a prototype annular VUV/UV photoreactor operating under continuous flow through conditions. The reactor was made of Plexiglas, with an annular configuration, equipped with a 42 W low pressure mercury amalgam ozone-generating lamp (GPHVA357T5VH, Light Sources Inc.), longitudinally placed at the axial center of the reactor. The internal and external diameters of the annular photoreactor and synthetic quartz sleeve (Heraeus Quartz America LLC, SUPRASIL310) were 7 cm and 2.8 cm, respectively, yielding a radial gap of 2.1 cm corresponding to a hydraulic diameter of 4.2 cm for treating the water. The presence of approximately 10 hydraulic diameters between the reactor inlet and lamp (e.g., ~ 40 cm) ensured that the flow was fully developed before the irradiated zone. Meanwhile the operating temperatures at the amalgam position, lamp envelope surface and water at 0.5 cm distance from  50  the reactor body were monitored using K type thermocouples (OMEGA®, US). This photoreactor configuration was utilized as the base case for the design exploration studies.  3.2.2 Analytical methods The concentration of atrazine was directly analyzed with HPLC (Dionex UltiMate™ 3000, US). Separation was accomplished using a reverse phase Nova Pak C18, 4 m 3.9150 mm separator column. A sample volume of 10 L with a flow rate of 0.8 mL/min (50% methanol/50% water at pH 2.5 using phosphoric acid) was injected to the system and the detection wavelength was set at 222 nm. Using a TOC Analyzer (Shimadzu TOC-VCPH) with the non-purgeable organic carbon (NPOC) method, concentrations of methanol and its oxidation by-products were quantified as an aggregate and reported as mgcarbon per L. For pH adjustment, a phosphate buffer solution (a mixture of monosodium phosphate and disodium phosphate (Sigma-Aldrich, Canada) was employed. A VUV-UV–vis spectrophotometer (Cary 4000 Varian) with 0.1 cm suprasil quartz cell was utilized for determining molar extinction coefficient at 185 nm. In this case, the spectrophotometer was purged continuously with high purity nitrogen gas (Bagheri et al., 2013).  3.3 Model development CFD modeling of the VUV/UV AOP requires simultaneous resolution of hydrodynamics, fluence rate at both 185 nm and 254 nm wavelengths, and a comprehensive kinetic scheme (Bagheri and Mohseni, 2014). Herein, for the purpose of simplicity and because of radial symmetry, only half of the shown 2-dimensional geometry was used for the CFD computation (Figure 3-1). ANSYS® 14.5 simulation packages were used to build and discretize the problem domain via 15418 structured hexahedral cells. The utilized grids were verified to provide mesh-independent results. ANSYS® Fluent 14.5  was employed to read the mesh and perform the CFD computations. As for the hydrodynamic sub-models, the laminar hydrodynamic equation was utilized under laminar  51  flow conditions (e.g., Re < 2100), whereas the Realizable k-epsilon (k-) and Reynolds stress model (RSM) turbulence sub-model was utilized and compared under turbulent flow conditions (Rauen et al., 2008).  The system of kinetic rate equations describing the production of hydroxyl radical from water photolysis and the corresponding degradation of atrazine is shown in (Table 3-1). Accordingly, the main degradation pathways of the model contaminant in the presence of VUV and UV radiations are; 185 nm direct photolysis, 254 nm direct photolysis, and •OH chain reactions. Herein, to partly investigate the •OH scavenging effect of the degradation intermediates, the first generation by-products of the model contaminant were included in the reaction scheme shown in Table 3-1. Also, it is known that the presence of natural organic matter (NOM), ubiquitous in water sources, greatly affects the efficiency of the VUV/UV process (Imoberdorf and Mohseni, 2011a, 2011b, 2012). This is mainly attributed to NOM’s affinity to react with •OH (scavenging effect) and reduce its availability to react with the target pollutant. Integration of the •OH scavenging impact of NOM is a complicated task due to its complex nature and variable structural and functional characteristics. Hence, in this work methanol, a common surrogate •OH scavenger, was utilized to simulate the •OH scavenging impact of NOM. Based on earlier investigations(Gonzalez et al., 2004), methanol undergoes widely studied sequential oxidation kinetic scheme which can be summarized by the three main reactions shown in Table 3-1. Considering the second order reaction rate constant of Suwannee River NOM (a widely studied NOM) and •OH is equal to 1.56×104 (mgC/L)-1s-1 (Westerhoff et al., 2007), the equivalent methanol concentration were defined.    52  Table 3-1 List of the reactions taking place in a VUV/UV photoreactor for degradation of atrazine No. Equilibrium reactions Constant 1 ܱଷ•	ି ൅ ܪା ↔ ܪܱଷ• pKa1 = 8.2 2 ܪଶ0ଶ ൅ ܱܪି 	↔ ܪଶ0 ൅ ܪܱଶି pKa2 = 11.6 3 ܪܱ• ൅ ܪܱି 	↔	ܪଶܱ ൅ ܱ•ି pKa3 = 11.9 4 ܪܱଶ• 	↔ 	ܪା ൅ ܱଶ•ି pKa4 = 4.8 5 ܱܪି ൅ ܪା 	↔	ܪଶܱ Kw = 10-14 Photochemical reactions 6 ܪଶܱ ൅ ݄νଵ଼ହ	௡௠ 	→	ܪܱ• ൅ ܪ•   Φ6 = 0.33  mole.ein-1 7 ܪଶܱ ൅ ݄νଵ଼ହ	௡௠ 	→	ܪା ൅ ݁ି௔௤ ൅ ܪܱ•   Φ7 = 0.045 mole.ein-1 8 ܪଶܱଶ ൅ ݄νଵ଼ହ	௡௠ 	→	2	ܪܱ•   Φ8 = 0.5  mole.ein-1 9 ܪଶܱଶ ൅ ݄νଶହସ	௡௠ 	→	2	ܪܱ•   Φ9 = 0.5  mole.ein-1 Contaminant reactions 10 ܥ଼ܪଵସܥ݈ ହܰ ൅ ݄νଶହସ	௡௠ 	→ ܥ଼ܪଵସܥ݈ ହܰ െ 254 ሺ∗ ܣܴܶ254ሻ    Φ10 = 0.05  mole.ein-1 11 ܥ଼ܪଵସܥ݈ ହܰ ൅ ݄νଵ଼ହ	௡௠ 	→ ܥ଼ܪଵସܥ݈ ହܰ െ 185 ሺ∗ ܣܴܶ185ሻ *Φ11 = 0.05  mole.ein-1 12 ܥ଼ܪଵସܥ݈ ହܰ	ሺܣݐݎܽݖ݅݊݁ሻ ൅ ܪܱ• → ∗ ܣܴܶ െ ܱܪ   k12 = 2.4×109   M-1.s-1 13 ܣܴܶ254 ൅ ܪܱ• 	 	→	∗ ܣܴܶ െ ܱܪ *k13 = 2.4×109  M-1.s-114 ܣܴܶ185 ൅ ܪܱ• 	 	→	∗ ܣܴܶ െ ܱܪ *k14 = 2.4×109  M-1.s-115 ܣܴܶ െ ܱܪ ൅ ܪܱ• 	 	→	∗ ܣܴܶ െ ܱܪܱܪ *k15 = 2.4×109   M-1.s-1Other involving reactions 16 ܪܱ• ൅ ܪଶ 	 	→	ܪଶܱ ൅ ܪ• k16 = 3.9×107    M-1.s-117 ܪܱ• ൅ ܪܱ• 	 	→	ܪଶܱଶ k17 = 4.2×109     M-1.s-118 ܪܱ• ൅ ܪ• 	 	→	ܪଶܱ k18 = 7×109      M-1.s-119 ܪܱ• ൅ ݁ି௔௤ 	 	→	ܪܱି k19 = 3×1010     M-1.s-1 20 ܪܱ• ൅ ܪܱଶ• 	 	→	ܪଶܱ ൅	ܱଶ k20 = 6.6×109   M-1.s-121 ܪܱ• ൅ ܱଶ•ି 	 	→	ܱଶ ൅ ܱܪି k21 = 1.1×1010  M-1.s-122 ܪܱ• ൅ ܪܱଶି 	 	→	ܱଶ•ି	 ൅ ܪଶܱ k22 = 6.8×109   M-1.s-123 ܪܱ• ൅ ܪଶܱଶ 	 	→	ܪܱଶ• ൅ ܪଶܱ k23 = 2.7×107   M-1.s-124 ܪܱଷ• 	→	ܪܱ• ൅	ܱଶ k24= 1.1×105    s-1 25 ܱଷ• ൅ ܪା 	 	→	ܪܱ• ൅	ܱଶ k25 = 9×1010     M-1.s-126 ݁ି௔௤ ൅ ܪଶܱଶ 	 	→	ܱܪି ൅ ܪܱ• k26 = 1.3×1010   M-1.s-1 27 ܪଶܱଶ ൅ ܪ• 	 	→	ܪܱ•ି ൅ ܪଶܱ k27 = 5×107      M-1.s-128 ܪܱଶ• ൅	ܪ• 	 	→ 	ܪଶܱଶ k28 = 2×1010     M-1.s-129 ܪܱଶ• ൅	ܱଶ•ି ൅ ܪଶܱ	 	→	ܪଶܱଶ ൅ ܱଶ ൅ ܪܱି k29 = 9.7×107     M-1.s-130 ܪܱଶ• ൅ ܪܱଶ• 	 	→	ܪଶܱଶ ൅ ܱଶ k30 = 8.3×105   M-1.s-131 ݁ି௔௤ ൅ ܱଶ 	 	→	ܱଶ•ି k31= 1.8×1010   M-1.s-1 32 ݁ି௔௤ ൅ ܪଶܱ	 	→	ܱܪି ൅ ܪ• k32 = 1×103        s-1 33 ݁ି௔௤ ൅ ܪ• ൅ 	ܪଶܱ	 	→	ܱܪି ൅ ܪଶ k33 = 3.4×1010   M-1 .s-1 34 ݁ି௔௤ ൅	݁ି௔௤ ൅	2	ܪଶܱ	 → 2 ܱܪି ൅ ܪଶ k34 = 6×10 9      M-1 .s-1 35 ݁ି௔௤ ൅ ܪା 	 	→ ܪ• k35 = 2.8×1010  M-1 .s-1 36 ݁ି௔௤ ൅ ܪܱଶି 	 	→	ܱܪି ൅ ܱ•ି k36 = 3.5×109   M-1 .s-1 37 ܪܱି ൅ ܪ• 	→	݁ି௔௤ ൅	ܪଶܱ k37 = 2.5×107   M-1 .s-1 38 ܱଶ ൅ ܱ•ି 	→	ܱଷ•ି k38 = 3.8×109   M-1 .s-139 ܱଶ•ି ൅ ܪ• 	→	ܪܱଶି k39 = 2.7×107   M-1 .s-140 ܱଶ ൅ ܪ• 	 	→	ܪܱଶ• k40 = 1.2×1010  M-1 .s-141 ܱ•ି ൅	ܪܱଶି 	 	→	ܱܪି ൅	ܱଶ•ି k41 = 4×108      M-1 .s-142 ܪ• ൅ ܪ• 	 	→	ܪଶ k42 = 5.5×109   M-1 .s-1Methanol scavenging reactions 43 ܥܪଷܱܪ ൅ ܪܱ• →∗ ܥܪଶܱ k43 = 9.8×108   M-1.s-144 ∗ ܥܪଶܱ ൅ ܪܱ• →∗ ܥܪܱܱܪ k44 = 1×109     M-1.s-1 45 ∗ ܥܪܱܱܪ ൅ ܪܱ• → 	݌ݎ݋݀ݑܿݐݏ k45 = 3×108     M-1.s-1  Starred reactants/products are hypothetical and are only defined for model development.  Methanol oxidation is a simplified scheme to main •OH reactions.  53  The calculations of the fluence rate field for 185 nm and 254 nm photons were performed by solving the radiative transport equation (RTE) using non-gray discrete ordinate (DO) sub-model (Bagheri and Mohseni, 2014). Furthermore, the reflection, refraction, and absorption of photons within the air gap (separating the lamp and the quartz sleeve region) and on the quartz sleeve were incorporated within the radiation sub-model. Moreover, to account for the dependency of 185 nm water absorption coefficient on temperature (+0.05 cm-1/ºC), a linear function which corresponds to 1.4 cm-1 at 25ºC (Barrett and Mansell, 1960; Halmann, 1966) was coupled to the radiative transfer equation (Bagheri and Mohseni, 2014; Bagheri et al., 2013). Measured UV254 transmittance of the samples was found to be 98% and this value was used in the simulations. More details about the utilized CFD sub-models, boundary conditions and numerical parameters can be found elsewhere (Bagheri and Mohseni, 2014). The total execution time of each simulation was about 4 h on a 3.50 GHz AMD Phenom™ II X4 970 processor with 8 GB RAM. 3.4 Computer-aided design exploration  In addition to the base case VUV/UV photoreactor geometry (described earlier), seven different design scenarios were investigated numerically and compared in terms of their AOP performance. The objective of this study was to establish the design parameters that improve the design of the VUV/UV photoreactors for the removal of a model CEC (atrazine). The investigated design parameters were: the reactor hydraulic diameter (case I and case II), the illuminated surface area (case II and case IV), combination of reactor hydraulic diameter and illuminated surface area (case V and case VI), and finally the incorporation of seven equally spaced central baffles within the base photoreactor geometry (case VII). Cases I to VI were accomplished by varying the reactor shell and quartz sleeve diameter. Case VII was investigated by retrofitting annular baffles within the photoreactor. The air gap between the lamp envelope and the quartz sleeve was constrained to  54  the range of 5-25 mm, to maintain lamp temperature in a suitable range, based on the lamp manufacturer’s recommendation (Light sources, 2013). Furthermore, the minimum hydraulic diameter constraint for the photoreactor (10 mm) was selected based on the 185 nm optical path for MilliQ water at approximately 21°C (Bagheri and Mohseni, 2014). Table 3-2 lists the values of the design parameters for the seven experiments along with the base geometry, which was utilized for the CFD model validation. For further illustration, the explored design factors were presented schematically using a two-dimensional parameterized geometry (Figure 3-1).  Figure 3-1 Parameterized VUV/UV photoreactor geometry utilized for this study (a: sleeve inner diameter, b: sleeve outer diameter, c: air gap, d: reactor inner diameter, e: optical thickness) The reactor design scenarios were assessed in terms of their atrazine removal and lamp electrical energy-per-order (Bolton et al.) (EEO, kWh/m3/order) (Eq. 3-1).   ܧܧܱ ൌ ௅ܹܳ ൈ ܮ݋݃ሺܥ௜ ܥ௙ሻ⁄  (3-1)where WL is the input electrical energy of the lamp (kW), Q is the water volumetric flow rate (m3/h), Ci and Cf are concentration of the target contaminant, atrazine, at the inlet and outlet (final), respectively. Although using EEO as a descriptive parameter to evaluate electrical energy efficiency of AOP reactors may not an ideal approach, it is the most widely applied methodology d27.8 cm b  c Air gap Water Lamp Sleeve Symmetry  plane e a  55  in the literature for assessment of the electrical energy and removal performance of AOP (Afzal et al., 2010; Bolton et al., 2001; Zoschke et al. 2012, 2014). In addition, since changes to the reactor hydrodynamics due to fluid flow rate will contribute to the electrical cost of the system, the required electrical cost of pumping should be incorporated into the calculation of EEO.  In this work, the EEO values for each scenario were compared at three different flow rates (e.g., 1, 3.5, 6.5 L/min) for the MilliQ water with 300 ppb atrazine. It should be noted that the EEO values do not account for electrical energy required for pumping of fluid (water) within the reactor. Accordingly, the calculated pressure drops (P) across inlet and outlet of the explored design configurations were calculated and reported based on the horizontal installation of reactors.  Table 3-2 List and values of the explored design parameters for the annular VUV/UV photoreactor Design scenarios sleeve inner diameter (a) sleeve outer diameter (b) air gap (c) reactor inner dimater (d) Illuminated surface area (cm2) hydraulic diameter (e) Base case 25 mm 28 mm 5 mm 70 mm 244.5 21 mm I 25 mm 28 mm 5 mm 49 mm 244.5 10.5 mm II 25 mm 28 mm 5 mm 112 mm 244.5 42 mm III 45 mm 48 mm 15 mm 90 mm 419.2 21 mm IV 65 mm 68 mm 25 mm 110 mm 593.9 21 mm V 35 mm 38 mm 10 mm 70 mm 331.9 16 mm VI 45 mm 48 mm 15 mm 70 mm 419.2 11 mm VII Retrofitting seven equally-spaced central baffles within the base case (Baffles’ dimensions: inner diameter = 38 mm, outer diameter =65 mm,  thickness =6 mm)   56  3.5 Results and discussion 3.5.1 Experimental evaluation of the CFD model Figure 3-2 presents the CFD model predictions along with the experimental data for the degradation of 300 ppb atrazine in MilliQ water plus methanol at concentration with •OH scavenging impact equivalent to 2 ppm NOM. It should be noted that a sensitivity analysis was also conducted to ensure that the model predictions were insensitive to small variations within the reported accuracy of the kinetic rate constants.   Figure 3-2 Comparison of experimental data obtained by the VUV/UV experimental set-up and CFD model for base case  0.00.10.20.30.40.50.60.70.80.90 1 2 3 4 5 6 7C/C0Flow rate (L/min) Exp. atrazineCFD model Exp. (2 ppm NOM + atrazine)(Error bars represent the standard deviation from the mean of three independent experiments)  57  Based on the utilized experimental apparatus, the Reynolds number was in the range of 100 (~ 0.5 L/min) to 1500 (~ 6.5 L/min) which corresponds to fully laminar flow regime in the photoreactor. As shown in Figure 3-2, in the absence of 2 ppm NOM, a nearly complete degradation of atrazine took place at very low flow rates (~ 0.5 L/min). However, the presence of 2 ppm NOM resulted in a pronounced reduction in the degradation rate of atrazine, specifically at moderate and higher flow rates. As can be seen from the Figure 3-2, the CFD model predicted the experimental data very well for the entire dataset.            Figure 3-3 Parts per billion (ppb) atrazine concentrations with the corresponding ⦁OH (M) profiles at different flow rates Based on the average kinetic rates for the main degradation pathways, the •OH degradation (oxidation) rate of atrazine is approximately 3 to 7 times larger than that associated with 254 nm direct-photolysis, and about 80 to 140 times greater than the rate of 185 nm direct-photolysis, 1.5e-11              1.1e-9                  6.1e-11                1.5e-9 2.1 cm atrazine 6.5 L/min atrazine 1 L/min 2.1 cm atrazine + 2ppm NOM 6.5 L/min 0              60             120            180             240          300atrazine + 2ppm NOM 6.5 L/min 1 L/min   / i   / i   58  depending on the flow rates. In order to gain better understanding of the performance efficacy of the VUV/UV process at different flow rates, the local concentration contours of atrazine and •OH in the photoreactor were analyzed (Figure 3-3). As depicted in Figure 3-3, the dose of 185 nm radiation, and consequently, the •OH concentration, is inversely proportional to the flow rate. Even though increasing the flow rate or presence of NOM rapidly reduces •OH to a thinner film layer, there is still a concentrated layer of OH radical near the sleeve, which could be utilized for the conversion of the unreacted atrazine in the sleeve vicinity. Therefore, mass transfer of the target contaminant to the reacting zone is a key variable that may control the pollutant degradation rate in the VUV/UV photoreactor. This suggests that optimized hydrodynamic design could have substantial impact on the VUV/UV photoreactor performance. 3.5.2 Effect of reactor hydraulic diameter The atrazine removal for different water layers at three flow rates of 1, 3.5, 6.5 L/min may be seen in Figure 3-4. The data labels on each bar chart present the EEO values (kWh/m3/order). While the base case has a hydraulic diameter of 21 mm, case I and case II had half and twice the hydraulic diameters, respectively. As shown in Figure 3-4(a), substitution of the reactor shell with the one of smaller-diameter (case I) resulted in approximately 36%, 21%, and 17% lower EEO values at 1, 3.5, and 6.5 L/min, respectively. On the other hand, the EEO improved for case II by 0.7%, 6.5%, and 9% at 1, 3.5, and 6.5 L/min, respectively. In order to gain an in-depth insight toward each design scenario, a Lagrangian particle tracking approach was used to analyze the contaminant exposure to hydroxyl radicals, UV 254 nm and VUV 185 nm photons (•OH, UV and VUV dose distributions) (Bagheri and Mohseni, 2014; Santoro et al., 2010).  In the Lagrangian model, uniformly distributed particles were released at  59  the reactor inlet (presenting the target contaminant) where the velocity distribution was also uniform, and the received •OH, UV and VUV doses were integrated along the particle trajectories. Figure 3-4 Removal ratio of atrazine (C/C0) and energy-per-order (EEO) values for different design modification scenarios at 1, 3.5, 6.5 L/min The delivered dose at each point was calculated by multiplying the average local value of a property of interest (e.g., •OH concentration, UV or VUV radiation intensities) by the time period t for which the particle was exposed to that value. Afterwards, the maximum, minimum, median and P(10-90) values were calculated and reported for each property, and a frequency distribution created (Figure 3-5). The P(10-90) indicates the range of data within which 10 to 90 percent of the 0.450.410.370.710.520.440.720.580.500.00.10.20.30.40.50.61.0 3.5 6.5 Case 1   Base  Case 20.680.510.430.710.520.440.660.500.420.00.10.20.30.40.50.61.0 3.5 6.5 Case 3   Base  Case 40.590.470.400.710.520.440.460.420.380.00.10.20.30.40.50.61.0 3.5 6.5 Case 5   Base  Case 60.940.840.821.191.000.920.160.280.260.710.520.440.00.10.20.30.40.50.60.70.81.00 3.50 6.50  Case 7 (atrazine + 2pm NOM)  Base (atrazine + 2pm NOM)  Case 7 (atrazine)  Base (atrazine)(a) (b)(c) (d)C/C0 Flow rate (L/min) Flow rate (L/min) Flow rate (L/min) Flow rate (L/min) C/C0 C/C0 C/C0  60  values of interest lie. Graphs presented in Figure 3-5 only represent the data for 3.5 L/min because of similar results obtained at different flow rates.   Figure 3-5 Median, mean, maximum, and P (10-90) values for the residence time, UV, VUV, and ⦁OH dose distribution of different design modification scenarios Comparing the P(10-90) of the base design case to that of the cases I and II, it is evident that case I, with the thinner water layer, benefits from the most uniform •OH, UV and VUV dose distributions. With the •OH oxidation being the most relevant degradation mechanism in this  61  study, case I has higher median and minimum delivered •OH dose than the base case and case II. In contrast, case II, with the larger water layer, and consequently larger volume and longer residence time appears to receive greater UV dose compared to case the base system and case I. Generally, considering the high UV transmittance (UVT) of 98% for the samples, increasing the hydraulic diameter will greatly favor the UV degradation pathway, especially at lower flow rates. In other words, when water UVT is very high (i.e., 98%), UV radiation intensity is maintained for a larger optical path and therefore the illuminated volume is increased. Besides, increasing the optical path of the reactor results in a larger cross-sectional area for the flow, which can translate into the greater water retention time. Both factors, increased retention time and larger illuminated volume, amplify the role of 254 nm UV direct degradation of atrazine in case II. It should be noted that as the UV molar absorptivity of water increases, a less pronounced UV photolysis of the contaminant is expected due to the faster attenuation of 254 nm radiation and the shorter optical path in the photoreactor. 3.5.3 Effect of reactor illuminated surface area  Exploring the impact of the illuminated surface area was required to conduct at constant reactor hydraulic diameter, which necessitates the simultaneous replacement of reactor shell and quartz sleeve. Utilizing a constant hydraulic diameter of 21 mm, the impact of increasing the illuminated surface area was explored in case III and case IV, which had respectively about 70% and 140% larger illuminated surface area compared to the base case design. The larger illuminated surface area resulted in only a minor improvement in atrazine removal and the EEO values for cases III and V (Figure 3-4b). While a larger illuminated surface area increased the reacting volume of the reactor with respect to •OH, since the lamp output was fixed, the radiation per unit area was reduced by a corresponding amount. In addition, oxygen in the larger air gap between the lamp  62  and sleeve absorbed additional 185 nm radiation, resulting in a decrease in OH radical generation. Compared to the base system, the transmitted 185 nm VUV radiation from the sleeve wall to water was calculated to be lower by 10% and 18% for cases III and IV, respectively. This was mainly attributed to the increasing circumferential area and the greater absorption of 185 nm photons by oxygen in the volume between the lamp and the sleeve (absorption coefficient: 0.166 atm-1cm-1) (Creasey et al., 2000). Knowing the longer residence time distribution and the smaller maximum UV and VUV intensities in case IV, a more uniformed UV, VUV and •OH dose distribution was attained when compared to the base case and case III. 3.5.4 Effect of simultaneous change in hydraulic diameter and reactor illuminated surface area  The combined effect of hydraulic diameter and illuminated surface area was investigated by increasing the sleeve diameter by 35% and 70% while keeping the outer reactor diameter fixed (cases V and VI). In this way, 24% and 48% smaller hydraulic diameter and 35% and 70% greater illuminated surface area were obtained respectively for cases V and VI compared to the base case. As shown in Figure 3-4 (c), the EEO values for case V were lower than that of the base case by 16%, 11%, and 9% at flow rates of 1, 3.5, and 6.5 L/min, respectively. Greater improvements of 35%, 24%, and 14% were calculated at 1, 3.5, and 6.5 L/min, respectively, for case VI (Figure 3-4c). Even though case VI benefited from the more uniform •OH dose distribution with a larger median value (compared to case V and the base case), it suffers from the lower minimum and median UV dose delivery. Clearly, the reactor hydraulic diameter is a relevant design factor that significantly determines the removal efficacy of the VUV/UV process. On the contrary, the reactor illuminated surface area had much less influence on the removal efficacy of the VUV/UV process when compared to hydraulic diameter. Furthermore, the extent of removal efficacy improvement  63  obtained from a smaller hydraulic diameter and a larger illuminated surface area at low to medium flow rates was more significant than the higher flow rates. This is mainly attributed to the lower concentration of the •OH formed at higher flow rates, thereby diminishing the •OH-oxidation pathway. 3.5.5 Effect of retrofitting baffles Incorporating baffles can enhance mass transfer within the photoreactor and may thereby improve the removal efficacy of the process. This may be particularly significant for VUV reactors due to the short penetration depth of 185 nm radiation. To assess the role of baffles in the photoreactor performance, seven equally spaced annular baffles were considered (case VII). Compared to the base case configuration, case VII led to significant improvement of the EEO values by 77%, 46%, and 40% for flow rates 1, 3.5, 6.5 L/min, respectively. This is the greatest enhancement obtained among all the investigated design modification cases (Figure 3-4d). As demonstrated in Figure 3-5, not only did case VII benefit from the most evenly distributed •OH, UV and VUV doses, but also larger minimum and median values were attained. Approximately 50% greater median •OH dose was observed through the baffle-aided design compared to that obtained in case I and or case VI. Baffles effectively expedited the transport of atrazine from the bulk flow to the high [•OH] region near the sleeve, thereby enhancing the degradation rate of the contaminant.  To investigate the performance and energy efficacy of the baffle-reactor (case VII) in the presence of an •OH scavenger (i.e., NOM), the same analysis was performed by incorporating methanol at concentrations equivalent to scavenging capability of 2 ppm NOM. As shown in Figure 3-4d, the presence of 2 ppm NOM increased the EEO (decreased efficiency) significantly by 69%, 72%, and 107%, compared to that of MilliQ, for the three flow rates of 1, 3.5, and 6.5 L/min, respectively. The improvement in performance achieved by adding baffles was reduced by the  64  addition of the radical scavenging methanol. Compared with the base case geometry with 2 ppm NOM, incorporation of baffles improved the EEO by 22%, 16%, and 11% at 1, 3.5, and 6.5 L/min flow rate, respectively.  3.5.6 Model evaluation of baffle-aided design Laboratory experiments were conducted to evaluate the accuracy of the simulations for the case with the added baffles. Figure 3-6 shows the CFD model predictions along with the experimental data for the degradation of 300 ppb atrazine using the VUV/UV photoreactor with and without baffles (i.e., base case and case VII designs). The CFD model predictions agreed well with the experimental data on atrazine degradation in MilliQ water and water containing 2 ppm NOM equivalent •OH scavenger. The average absolute relative deviation (AARD) was 3.1% for the experimental data and CFD predicted values for atrazine removal ratio through application of the Realizable k-epsilon (k-) turbulence sub-model. Although incorporation of the Reynolds Stress Model (RSM) slightly improved the CFD model performance with an AARD of 1.9%, it doubled the computational time from approximately 4 h to 8 h. Since most of the predicted values from the RSM turbulence sub-model are between the experimental values and the k- simulated results, for the purpose of simplicity only CFD predicted values with k-  sub-model and experimental values were plotted in Figure 3-6. Similar observations were found by incorporating methanol at concentrations equivalent to scavenging capability of 2 ppm NOM. However, the CFD predicted atrazine removal ratios showed an increased AARD of 4.2% and 3.8% with k- turbulence and RSM sub-models, respectively. The greater deviation of CFD model predictions in the case of methanol addition was attributed primarily to greater experimental uncertainties due to the quantification of methanol at low concentrations.   65   Figure 3-6 Comparison of experimental data obtained by the VUV/UV set-up using retrofitting baffles and the CFD model  Apart from the model validation purposes, Figure 3-6 reveals the extent of enhancement in the EEO values resulting from retrofitting baffles, showing more significant enhancement at low to mid flow rates (e.g., < 4.5 L/min). Consistent with the model predictions, experimental data confirmed lesser improvement in atrazine removal when •OH scavenger was present in the solution. This is mainly attributed to the amplified •OH scavenging impact of NOM as a result of an improved mixing provided through integration of baffles within the VUV/UV photoreactor. 3.5.7 Pressure drop and energy use considerations The calculated pressure drop between the photoreactor inlet and outlet for various design scenarios was reported in Table 3-3. As expected, pressure drop was inversely proportional to the hydraulic 0.00.10.20.30.40.50.60.70.80 1 2 3 4 5 6 7C/C0Flow rate (L/min)  Exp. with baffles  (atrazine)  Exp. with baffles (2 ppm NOM + atrazine)  Exp. (atrazine)  Exp. (2 ppm NOM + atrazine)  CFD model (Realizable k-epsilon )(Error bars represent the standard deviation from the mean of three independent experiments)  66  diameter. Conversely, increasing the illuminated surface area at constant hydraulic diameter reduced the head loss within the reactor. However, the photoreactor pressure drop was more sensitive to the hydraulic diameter alteration than the illuminated surface area. Since pumping losses represent an additional required energy input to the system, the pumping electrical input of the design scenarios should be incorporated into the analysis. Using equations (3-2) and (3-3), the total electrical energy-per-order (EEOt) analysis was performed and reported in Table 3-3.  ௣ܹ	ሾkWሿ ൌ ܳ	ሾ݉ଷ ݄⁄ ሿ ൈ ߂ ௧ܲሾ݌ܽ	ሿߝ௣ ൈ 3600 ൈ 1000  (3-2)ܧܧ ௧ܱ	ሾkWh/݉ଷ/orderሿ ൌ ௉ܹ ൅ ௅ܹܳ	ሾ݉ଷ ݄⁄ ሿ ൈ ܮ݋݃ሺܥ௜ ܥ௙ሻ⁄  (3-3)where Wp is the input electrical energy of the pump (kW), Q is the water volumetric flow rate (m3/h), p is the pump electrical efficiency (0.75), Ci and Cf are concentrations of the target contaminant, atrazine, in the influent and effluent stream, respectively. It should be noted that Pt is total pressure drop of the system including a net static head of 1 m between suction and discharge parts in addition to the pressure drop within the horizontally installed VUV/UV reactor. Although the explored reactor design scenarios behaved differently in terms of pressure drop, the reported values of <100 Pascal can be considered negligible in terms of electrical power requirement towards the total reported EEO. Based on Table 3, the required pump power was much smaller than the lamp power. This is an indication of greater electrical energy demand of the UV lamp as compared to that of pumping. In addition, despite the increased head loss associated with the baffles-aided design scenario, this design provided the most energy-efficiency of the VUV/UV treatment process based on the calculated EEOt. Meanwhile, a less pronounced improvement in the energy-efficiency of the contaminant remediation was observed in the presence of 2 ppm NOM  67  equivalent •OH radical scavenger. This is mainly because incorporation of baffles facilitates the reaction of •OH scavengers as well as the contaminant.  Table 3-3 Calculated pressure drops within the inlet and outlet of reactor and the total energy-per-order (EEOt) for different design modification scenarios Design scenarios Pressure drop [Pa] EEOt [kWh/m3/order] 1 L/min 3.5 L/min 6.5 L/min 1 L/min 3.5 L/min 6.5 L/min Base case 0.08 0.35 0.81 0.714 0.529 0.455 I 0.74 3.09 6.81 0.452 0.422 0.382 II 0.01 0.04 0.08 0.724 0.591 0.517 III 0.15 0.65 1.43 0.684 0.519 0.443 IV 0.41 1.63 3.46 0.668 0.511 0.434 V 0.05 0.23 0.51 0.594 0.476 0.415 VI 0.04 0.17 0.36 0.460 0.428 0.396 VII 2.2 19.1 58.6 0.165 0.286 0.273  3.6 Chapter conclusions This chapter was to investigate the treatment feasibility and cost-effectiveness of VUV/UV process for effective degradation of emerging contaminants under continuous flow conditions. Through development of a complete descriptive CFD model, it was found that the AOP performance of the VUV/UV process is strongly correlated to the flow characteristics of VUV/UV photoreactors. Meanwhile, assessment of a wide range of reactor design parameters (e.g., hydrodynamic diameter, illuminated surface area and the incorporation of baffles) was performed in terms of the generated •OH concentration profile, UV 254 nm and VUV 185 nm delivered dose distributions. Thereby, an improved VUV/UV process with up to 72% electrical energy input saving for the  68  effective degradation of target contaminant was accomplished through incorporation of retrofitting central baffles. The results from this study will provide the water professionals with a cost-effective approach for the analysis and design exploration of the flow-through photoreactors, leading to the wide-scale application of the VUV/UV process.    69  Chapter 4: Impact of hydrodynamics on pollutant degradation and energy efficiency of VUV/UV and H2O2/UV oxidation processes  4.1 Chapter introduction Ultraviolet (UV) based advanced oxidation processes (AOPs), and in particular H2O2/UV AOP, have demonstrated great promise for the removal of organic micropollutants in water. Despite large-scale applications, H2O2/UV AOP suffers from the high cost of hydrogen peroxide including acquisition, transportation and storage (Gunten, 2003). More importantly, since a significant portion of utilized H2O2 will remain unreacted within the system, regular monitoring and post treatment of the residual peroxide are essential (Zoschke et al., 2012). The need for quenching the residual peroxide imposes further costs and complexity to the H2O2/UV process, specifically for applications in small- to medium-sized water treatment facilities. Vacuum-UV/UV AOP is an alternative competitive technology that eliminates the need for an auxiliary chemical oxidant, e.g., hydrogen peroxide, resulting in a simpler and more environmentally sustainable process with greater economic potential for commercial applications. The effectiveness of VUV/UV process for the removal of a wide range of contaminants under small batch recirculating reactors has been studied extensively (Arany et al., 2013; Crapulli et al., 2014; Imoberdorf and Mohseni, 2011, 2012; Kim and Tanaka, 2009; Zoschke et al., 2014). The experimental comparison of the H2O2/UV and VUV/UV AOPs in small batch recirculating reactors indicates superior performance of the VUV/UV process. Mouamfon et al. (2011) reported up to 20% greater performance for the VUV/UV process when compared with the widely applied H2O2/UV AOP for degrading a target pharmaceutical, sulfamethoxazole, in a lab-scale batch reactor. Similarly, Zoschke et al. (2012) evaluated the treatment cost of the VUV/UV and  70  H2O2/UV AOPs for remediation of taste and odor compounds (e.g., geosmin and 2-methyl isoborneol) in terms of the electrical energy-per-order (EEO). The authors underlined the cost competitiveness of VUV/UV over H2O2/UV process with up to 50% lower EEO for small batch recirculating reactors.  Despite the promising results obtained through lab-scale batch experiments, the practical implementation of the VUV/UV AOP within the water industry has not been fulfilled. The lack of experimental and modeling studies demonstrating the treatment capability of the VUV/UV systems under continuous flow operations is among the key impediments hindering their large-scale implementation. Furthermore, the impact of hydroxyl radical (•OH) scavengers on the performance of the VUV/UV AOP needs to be evaluated under continuous-flow operation. This is of paramount importance, since the very high absorption coefficient of water for 185 nm VUV radiation limits the oxidation power of the process to a thin film optical path (Barrett and Mansell, 1960; Price et al., 1960) (e.g., less than 1 cm). Finally, there is a significant need for investigating the impact of flow characteristics (reactor hydrodynamics) on the competiveness of the VUV/UV systems, when compared with the more established H2O2/UV process. Hydrodynamic studies have often been conducted through incorporation of baffles (Duran et al., 2011) or static mixers (Oppenländer et al., 2005), modification of the reactor geometry (Santoro et al., 2010) or injection of gas bubbles into the reactor (Azimi et al., 2014). Among these, the implementation of baffles has received greater attention due to its technical feasibility, operational practicality and cost-effectiveness. The main objective of this chapter was to utilize computational fluid dynamics (CFD) modeling for an in-depth comparative study of the VUV/UV and H2O2/UV AOPs for the removal of para-chlorobenzoic acid as a target model pollutant in a conventional continuously operating  71  photoreactor. A core aspect of this chapter included exploring the role of flow characteristics (hydrodynamics) in the competiveness of VUV/UV systems when compared with the H2O2/UV process. Furthermore, the extent of •OH scavenging of natural organic matter (NOM) and its influence on the energy-efficiency of the VUV/UV and H2O2/UV AOPs were assessed in terms of EEO. To date, no similar study that employs a mechanistic model for an in-depth comparative study of the VUV/UV and H2O2/UV AOPs under continuous-flow conditions has been reported.  4.2 Materials and methods 4.2.1 Experimental setup and procedure Experimental tests were conducted in a prototype annular photoreactor operating under continuous flow conditions (Figure 4-1). The reactor was made of Plexiglas, with an annular configuration, equipped with a 42 W low-pressure mercury amalgam lamp, longitudinally placed at the axial center of the reactor. The internal and external diameters of the annular photoreactor and synthetic quartz sleeve (Heraeus Quartz America LLC, SUPRASIL310) were 7 cm and 2.8 cm, respectively, yielding a radial gap of 2.1 cm corresponding to a hydraulic diameter of 4.2 cm for treating the water. The presence of approximately 10 hydraulic diameters between the reactor inlet and lamp (e.g., ~ 40 cm) ensured that the flow was fully developed before the irradiated zone. Meanwhile, the operating temperatures at the amalgam position, lamp envelope surface and water at 0.5 cm distance from the reactor body were monitored using K-type thermocouples (OMEGA®, US). An ozone-generating lamp (GPHVA357T5VH, Light Sources Inc.) was utilized for experiments involving VUV/UV AOP, whereas an ozone-free lamp (GPHVA357T5L, Light Sources Inc.) with identical dimension and power requirement was employed for UV and or H2O2/UV processes. The operating amalgam temperature of the utilized lamps during the degradation experiments were approximately 65 ± 2.0 °C depending upon the flow rates of water. Knowing the amalgam  72  temperature, 185 nm VUV and 254 nm UV irradiances of the lamps were calculated using a radiometer (Bagheri and Mohseni, 2014). Details of the utilized radiometry technique for the 185 nm and 254 nm lamp power measurements can be found elsewhere (Bagheri and Mohseni, 2014; Sasges et al., 2012).  Figure 4-1 Schematic of the VUV/UV and H2O2/UV prototype photoreactor 4.2.2 Analytical methods The concentration of p-CBA was directly analyzed with HPLC (Dionex UltiMate™ 3000, US). Separation was accomplished using a reverse phase Nova Pak C18, 4 m 3.9150 mm separator column. A sample volume of 20 L with a flow rate of 0.45 mL/min (55% acetonitrile/44% water/1% phosphoric acid) was injected to the system and the detection wavelength was set at 235 nm. For pH adjustment, a phosphate buffer solution (a mixture of monosodium phosphate and  73  disodium phosphate (Sigma-Aldrich, Canada) was employed. Concentration of hydrogen peroxide was determined via UV spectrophotometry utilizing the ܫଷି  method (Klassen et al., 1994). Hydrogen peroxide was purchased as 35 wt. % solution (Sigma-Aldrich, Canada) in water. Using a TOC Analyzer (Shimadzu TOC-VCPH) with the non-purgeable organic carbon (NPOC) method, concentrations of methanol and its oxidation by-products were quantified as an aggregate and reported as mg carbon per L.  4.3 Model development CFD simulation of the VUV/UV process necessitates simultaneous resolution of hydrodynamics, fluence rate at both 185 nm and 254 nm wavelengths and a comprehensive kinetic scheme (Bagheri and Mohseni, 2014; Bagheri et al., 2013). It should be noted that since the kinetic scheme and radiation system governing H2O2/UV process are already part of VUV/UV multiphysics, the developed model could be utilized for both VUV/UV and H2O2/UV AOPs. ANSYS® 14.5 simulation platform was employed to build and discretize the simplified 2D axisymmetric reactor geometry via 15418 structured hexahedral cells. As well, ANSYS® Fluent 14.5 was employed to read the exported mesh and perform the CFD computations. As for the hydrodynamics sub-models, the laminar hydrodynamics equation was utilized under laminar flow conditions (e.g., Re < 2100), whereas the Realizable k-epsilon (k-) sub-model was utilized and compared under turbulent flow conditions (Duran et al., 2011; Elyasi and Taghipour, 2010; Santoro et al., 2010). The utilized kinetic rate equations for the degradation of the target pollutant, p-CBA, is described in Table 4-1. To simulate •OH scavenging impact of NOM, methanol was utilized as a surrogate •OH scavenger. Based on earlier investigations (Gonzalez et al., 2004), methanol undergoes a widely studied sequential oxidation kinetic scheme which can be well summarized via three main reactions (see Table 4-1).   74  Table 4-1 Kinetic sub-model utilized in VUV/UV and H2O2/UV advanced oxidation of p-CBA No. Equilibrium reactions Constant 1 ܱଷ•	ି ൅ ܪା ↔ ܪܱଷ• pKa1 = 8.2 2 ܪଶ0ଶ ൅ ܱܪି 	↔ ܪଶ0 ൅ ܪܱଶି pKa2 = 11.6 3 ܪܱ• ൅ ܪܱି 	↔	ܪଶܱ ൅ ܱ•ି pKa3 = 11.9 4 ܪܱଶ• 	↔ 	ܪା ൅ ܱଶ•ି pKa4 = 4.8 5 ܱܪି ൅ ܪା 	↔	ܪଶܱ Kw = 10-14 Photochemical reactions 6 ܪଶܱ ൅ ݄νଵ଼ହ	௡௠ 	→	ܪܱ• ൅ ܪ•   Φ6 = 0.33  mole.ein-1 7 ܪଶܱ ൅ ݄νଵ଼ହ	௡௠ 	→	ܪା ൅ ݁ି௔௤ ൅ ܪܱ•   Φ7 = 0.045 mole.ein-1 8 ܪଶܱଶ ൅ ݄νଵ଼ହ	௡௠ 	→	2	ܪܱ•   Φ8 = 0.5  mole.ein-1 9 ܪଶܱଶ ൅ ݄νଶହସ	௡௠ 	→	2	ܪܱ•   Φ9 = 0.5  mole.ein-1 Contaminant reactions 10 ݌ܥܤܣ ൅ ݄νଶହସ	௡௠ 	→ ܥ଻ܪହܥ݈ܱଶ െ 254 ሺ∗ ݌ܥܤܣ254ሻ  Φ10 = 0.013  mole.ein-1 11 ݌ܥܤܣ ൅ ݄νଵ଼ହ	௡௠ 	→ ܥ଻ܪହܥ݈ܱଶ െ 185 ሺ∗ ݌ܥܤܣ185ሻ *Φ11 = 0.013  mole.ein-1 12 ܥ଻ܪହܥ݈ܱଶሺ݌ܥܤܣሻ ൅ ܪܱ• → ∗ ݌ܥܤܣ െ ܱܪ   k12 = 5.0×109   M-1.s-1 13 ݌ܥܤܣ254 ൅ ܪܱ• 	 	→	∗ ݌ܥܤܣ െ ܱܪ *k13 = 1.0×109  M-1.s-1 14 ݌ܥܤܣ185 ൅ ܪܱ• 	 	→	∗ ݌ܥܤܣ െ ܱܪ *k14 = 1.0×109  M-1.s-1 15 ݌ܥܤܣ െ ܱܪ ൅ ܪܱ• 	 	→	∗ ݌ܥܤܣ െ ܱܪܱܪ *k15 = 1.0×109   M-1.s-1 Other involving reactions 16 ܪܱ• ൅ ܪଶ 	 	→	ܪଶܱ ൅ ܪ• k16 = 3.9×107    M-1.s-117 ܪܱ• ൅ ܪܱ• 	 	→	ܪଶܱଶ k17 = 4.2×109     M-1.s-1 18 ܪܱ• ൅ ܪ• 	 	→	ܪଶܱ k18 = 7.0×109   M-1.s-1 19 ܪܱ• ൅ ݁ି௔௤ 	 	→	ܪܱି k19 = 3.0×1010  M-1.s-1 20 ܪܱ• ൅ ܪܱଶ• 	 	→	ܪଶܱ ൅	ܱଶ k20 = 6.6×109   M-1.s-1 21 ܪܱ• ൅ ܱଶ•ି 	 	→	ܱଶ ൅ ܱܪି k21 = 1.1×1010  M-1.s-1 22 ܪܱ• ൅ ܪܱଶି 	 	→	ܱଶ•ି	 ൅ ܪଶܱ k22 = 6.8×109   M-1.s-1 23 ܪܱ• ൅ ܪଶܱଶ 	 	→	ܪܱଶ• ൅ ܪଶܱ k23 = 2.7×107   M-1.s-1 24 ܪܱଷ• 	→	ܪܱ• ൅	ܱଶ k24= 1.1×105    s-1 25 ܱଷ• ൅ ܪା 	 	→	ܪܱ• ൅	ܱଶ k25 = 9×1010     M-1.s-1 26 ݁ି௔௤ ൅ ܪଶܱଶ 	 	→	ܱܪି ൅ ܪܱ• k26 = 1.3×1010   M-1.s-1 27 ܪଶܱଶ ൅ ܪ• 	 	→	ܪܱ•ି ൅ ܪଶܱ k27 = 5.0×107   M-1.s-1 28 ܪܱଶ• ൅	ܪ• 	 	→ 	ܪଶܱଶ k28 = 2.0×1010  M-1.s-1 29 ܪܱଶ• ൅	ܱଶ•ି ൅ ܪଶܱ	 	→	ܪଶܱଶ ൅ ܱଶ ൅ ܪܱି k29 = 9.7×107     M-1.s-1 30 ܪܱଶ• ൅ ܪܱଶ• 	 	→	ܪଶܱଶ ൅ ܱଶ k30 = 8.3×105   M-1.s-1 31 ݁ି௔௤ ൅ ܱଶ 	 	→	ܱଶ•ି k31= 1.8×1010   M-1.s-1 32 ݁ି௔௤ ൅ ܪଶܱ	 	→	ܱܪି ൅ ܪ• k32 = 1.0×103    s-1 33 ݁ି௔௤ ൅ ܪ• ൅ 	ܪଶܱ	 	→	ܱܪି ൅ ܪଶ k33 = 3.4×1010  M-1 .s-134 ݁ି௔௤ ൅	݁ି௔௤ ൅	2	ܪଶܱ	 	→ 2 ܱܪି ൅ ܪଶ k34 = 6.0×10 9  M-1 .s-135 ݁ି௔௤ ൅ ܪା 	 	→ ܪ• k35 = 2.8×1010  M-1 .s-136 ݁ି௔௤ ൅ ܪܱଶି 	 	→	ܱܪି ൅ ܱ•ି k36 = 3.5×109   M-1 .s-137 ܪܱି ൅ ܪ• 	→	݁ି௔௤ ൅	ܪଶܱ k37 = 2.5×107   M-1 .s-1 38 ܱଶ ൅ ܱ•ି 	 	→	ܱଷ•ି k38 = 3.8×109   M-1 .s-139 ܱଶ•ି ൅ ܪ• 	→	ܪܱଶି k39 = 2.7×107   M-1 .s-140 ܱଶ ൅ ܪ• 	→	ܪܱଶ• k40 = 1.2×1010  M-1 .s-141 ܱ•ି ൅	ܪܱଶି 	 	→	ܱܪି ൅	ܱଶ•ି k41 = 4.0×108   M-1 .s-142 ܪ• ൅ ܪ• 	 	→	ܪଶ k42 = 5.5×109   M-1 .s-1Methanol scavenging reactions 43 ܥܪଷܱܪ ൅ ܪܱ• →∗ ܥܪଶܱ k43 = 9.8×108   M-1.s-1 44 ∗ ܥܪଶܱ ൅ ܪܱ• →∗ ܥܪܱܱܪ k44 = 1.0×109   M-1.s-1 45 ∗ ܥܪܱܱܪ ൅ ܪܱ• → 	݌ݎ݋݀ݑܿݐݏ k45 = 3.0×108   M-1.s-1  Starred reactants/products are hypothetical and are only defined for model development.  Methanol oxidation is a simplified scheme to main •OH reactions.  75  Based on the •OH reaction rate constant of Suwannee River NOM of 1.56×104 (mgC/L)-1s-1 (Westerhoff et al., 2007), the concentration of methanol corresponding to  •OH scavenging of 2-ppm NOM was calculated. Fluence rate fields for 185 nm and 254 nm photons were computed by solving the radiative transport equation (RTE) using a non-gray discrete ordinate (DO) sub-model. Furthermore, the reflection, refraction and absorption of photons within the air gap (separating the lamp and the quartz sleeve region) and on the quartz sleeve were incorporated within the radiation sub-model (Duran et al., 2010). Moreover, to account for the dependency of 185 nm water absorption coefficient on temperature (+0.05 cm-1/ºC), a linear function which corresponds to 1.4 cm-1 at 25 ºC was coupled with the radiative transfer equation (Bagheri and Mohseni, 2014). Measured UV254 transmittance of the samples was found to be 98%, and this value was used in the simulations. The 185 nm intensity inputs of the VUV/UV (ozone-generating) and UV (ozone-free) lamps, imported from the radiometer results, were 0.450 (± 0.030) W and 0.005 (± 0.001) W, respectively. Similarly, the 254 nm intensity of both lamps were set as 10.5 (± 0.2) was at the system’s operating temperature. More details about the utilized CFD sub-models, boundary condition, solution method and strategy can be found elsewhere (Bagheri and Mohseni, 2014; Bagheri et al., 2013). The total execution time of each simulation was about 4 h on a 3.50 GHz AMD Phenom™ II X4 970 processor with 8 GB RAM. 4.4 Results and discussion 4.4.1 CFD model validation and analysis  Based on the utilized experimental apparatus, the Reynolds number was in the range of 100 (~ 0.5 L/min) to 1500 (~ 6.5 L/min), corresponding to fully laminar and uniform flow regime in the photoreactor. Figure 4-2 (a) displays CFD model predictions and experimental data for the degradation of 300 ppb p-CBA in the annular photoreactor operating at flow rates of 0.5-6.5 L/min.  76  Similarly, Figure 4-2 (b) shows the •OH scavenging impact of 2-ppm NOM equivalent on the removal of 300 ppb p-CBA for different UV-based processes (e.g., UV, VUV/UV, 2-ppm and 5-ppm H2O2/UV). As shown in Figure 4-2, the measured and predicted values for the degradation of p-CBA were in good agreement over the entire dataset. It should be noted that experimental degradation data represent the mean of three independent replicates at any given flow rate. Further, the predicted degradation rates did not change with variations of the model input variables (e.g., kinetic rate constants) within the reported accuracy range (Bagheri and Mohseni, 2014).   Based on Figure 4-2 (a), p-CBA was not effectively degraded with direct UV 254 nm photolysis.  On the other hand, nearly complete removal of p-CBA was attained by VUV/UV process at the very low flow rate of 0.5 L/min. In addition, much greater p-CBA removal rates were obtained with the H2O2/UV AOP when compared with the VUV/UV process. Nonetheless, differences between the performances of the H2O2/UV and VUV/UV AOPs decreased with increasing the flow rates. Also, presence of a surrogate •OH scavenger resulted in nearly similar p-CBA removals for the VUV/UV and 2-ppm H2O2/UV processes, at flow rates greater than 3.5 L/min (Figure 4-2 b). Considering the •OH scavenging impact of methanol as an NOM surrogate, it is evident that the H2O2/UV process with 2- and 5 ppm peroxide concentrations was affected to a greater extent than the VUV/UV process.  The average rates of •OH-initiated oxidation and the 185 nm and 254 nm direct photolysis were calculated to determine their respective contributions in the overall removal of p-CBA in the VUV/UV system. It was observed that the •OH-initiated oxidation rate of p-CBA was approximately 8 to 15 times greater than the rate of 254 nm direct photolysis, and 800 to1200 times greater than the rate of direct 185 nm photolysis, depending upon flow rates.   77    Figure 4-2 Comparison of the p-CBA removal efficacy for the UV, VUV/UV, 2 and 5-ppm H2O2/UV processes using a conventional design photoreactor prototype 0.00.10.20.30.40.50.60.70.80.91.00 1 2 3 4 5 6 7C/C0Flow rate (L/min)   UV   VUV/UV   2ppm H2O2/UV   5ppm H2O2/UV0.00.10.20.30.40.50.60.70.80.91.00 1 2 3 4 5 6 7C/C0Flow rate (L/min)  UV   VUV/UV   2ppm H2O2/UV   5ppm H2O2/UV(a) Without NOM (b) 2 ppm NOM (CFD model predictions are shown with plus sign)  78  This indicated that •OH-oxidation is the predominant (controlling) mechanism contributing to p-CBA degradation regardless of flow rate. Knowing the critical role of •OH-oxidation degradation in each of the UV, VUV/UV 2- and 5-ppm H2O2/UV processes, the local concentration contours of p-CBA and •OH within the utilized prototype photoreactor were analyzed (Figure 4-3). Although the presented contours in Figure 4-3 are restricted to the flow rate of 3.5 L/min and ultrapure water (e.g., in the absence of a surrogate •OH scavenger), similar patterns were observed at different flow rates and in the presence of methanol as the surrogate •OH scavenger.   Figure 4-3 Parts per billion (ppb) p-CBA concentrations with the corresponding •OH concentrations (mol/L) in the conventional design photoreactor UV VUV/UV  2ppm H2O2/UV UV VUV/UV  2ppm H2O2/UV 5ppm H2O2/UV 3.5 L/min 5ppm H2O2/UV 3.5 L/min 3.5 L/min 3.5 L/min 2.1 cm 0            45             90           135          180            225             300 0          1.8          3.6           5.4           7.2            9           10      12 × 10-11 79  Given the very limited contribution of •OH, the UV process was only able to reduce p-CBA concentration from the inlet concentration of 300 ppb to 278 ppb. The VUV/UV process, with the greater contribution of OH radical, reduced p-CBA concentration by nearly 50%. Meanwhile, 2- and 5-ppm H2O2/UV processes showed the most significant removal of p-CBA, bringing its concentration to 98 ppb and 32 ppb, respectively. This is primarily attributed to the more uniform •OH concentration over the entire photoreactor volume obtained by the H2O2/UV AOPs, as opposed to the formation of very concentrated layer of •OH in the proximity of the quartz sleeve by the VUV/UV process (see Figure 4-3). With the steep gradient of •OH, and subsequent formation of large dead zone areas in the VUV/UV photoreactor, mass transfer limitation greatly affected performance of the VUV/UV system when compared with the H2O2/UV processes. Therefore, improved mixing that stimulates the migration of target pollutants into the reacting zone (•OH concentrated area) may potentially enhance the performance of VUV/UV systems. 4.4.2 Impact of flow characteristics To explore the impact of reactor hydrodynamics (flow characteristics) on the performance of UV, VUV/UV and 2 and 5-ppm H2O2/UV processes, central baffles were retrofitted within the prototype photoreactor (shown in Figure 4-1). Baffles were made of Teflon with 6 mm thickness, and inner and outer diameters of 38 mm and 65 mm, respectively. The 6 mm baffles were selected to minimize the shadowing effect of radiation while maintaining mechanical strength against a wide range of flow rates. The selection of seven baffles with 5 cm spacing was to assure formation of the vortexes (circulation zones) over the entire illuminated volume of the photoreactor. This was done through CFD visualization of the flow patterns, showing velocity vectors within the baffle-aided photoreactor.  80  Figure 4-4 compares experimental degradation of p-CBA along with the CFD predicted values using improved hydrodynamics achieved by retrofitting central baffles. The consistency of experimental and CFD-predicted data demonstrates the extended capability of the proposed CFD model through incorporation of a turbulence sub-model (e.g., Realizable k- with the enhanced near-wall treatment equation) for application under improved reactor hydrodynamics. As depicted in Figure 4-4 (a), introduction of central baffles within the prototype photoreactor resulted in superior degradation of the VUV/UV process over the 2-ppm H2O2/UV AOP. In addition, the •OH scavenging of 2-ppm NOM reduced the degradation efficiency of the VUV/UV and H2O2/UV AOPs. However, a greater extent of reduction was observed for the 2-ppm H2O2/UV when compared with the VUV/UV AOP.  This is an indication of the greater •OH scavenging impact of NOM on p-CBA removal efficiency of 2-ppm H2O2/UV process (Figure 4-4 b). The impact of turbulence and mixing on degradation of the target pollutant in the UV, VUV/UV and 2- and 5-ppm H2O2/UV processes can be observed through the p-CBA concentration profiles shown in Figure 4-5. Since OH radical is a very short-lived species with greater concentration in the vicinity of the quartz sleeve, mixing and circulation zones promoted through implementation of baffles can significantly stimulate •OH oxidation of the target pollutant (p-CBA). As illustrated in Figure 4-5, incorporation of central baffles within the prototype reactor led to about 65% improvement in p-CBA removal by the VUV/UV process compared with the base case of conventional reactor configuration (e.g., 150 ppb to 53 ppb). The significant enhancement attained by the modified VUV/UV prototype is primarily due to the improved migration of p-CBA to the •OH concentered area in the proximity of quartz sleeve, resulting in amplified p-CBA oxidation.   81    Figure 4-4 Comparison of the p-CBA removal efficacy for UV, VUV/UV, 2 and 5-ppm H2O2/UV processes using an improved prototype photoreactor via incorporation of central baffles shown with plus sign) 0.00.10.20.30.40.50.60.70.80.91.00 1 2 3 4 5 6 7C/C0Flow rate (L/min)  UV   VUV/UV   2ppm H2O2/UV   5ppm H2O2/UV0.00.10.20.30.40.50.60.70.80.91.00 1 2 3 4 5 6 7C/C0Flow rate (L/min)  UV   VUV/UV   2ppm H2O2/UV   5ppm H2O2/UV(CFD model predictions are(a) Without NOM (b) 2 ppm NOM  82  For the direct UV 254 nm photolysis, incorporating central baffles resulted in slightly lower p-CBA removal efficacy compared with that of the base design (outlet concentration decreased from 278 ppb to 280 ppb). This is attributed mainly to shadowing effect of the baffles, which block part of the emitted UV radiation, thereby lessening the UV direct photolysis of p-CBA pathway. Not only did the attenuation of UV radiation decrease the p-CBA direct photolysis, but also it led to lower formation of •OH from the UV photolysis of hydrogen peroxide. Nonetheless, 6% (98 ppb to 92 ppb) and 20% (32 ppb and 25 ppb) greater removal performance of the 2- and 5-ppm H2O2/UV processes were obtained in the presence of central baffles. Thus, the degradation enchantment attained through improved mixing and acceleration of •OH oxidation pathway outweigh the reduction in direct UV photolysis of pollutant caused by baffles shadowing. More importantly, improved mixing caused by the presence of baffles led to the greater enhancement of p-CBA removal in the H2O2/UV process with higher hydrogen peroxide concentration. 4.4.3 Electrical energy-per-order analysis  The treatment cost competiveness of the UV, VUV/UV, 2 and 5-ppm H2O2/UV processes were assessed in terms of electrical energy-per-order (EEO, kWh/m3/order) (Equation 4-1), based on 300 p-CBA inlet concentrations. ܧܧܱ ൌ ௅ܹ ൅ ௉ܹ ൅ܹ݋ܳ ൈ ܮ݋݃ሺܥ௜ ܥ௙ሻ⁄  (4-1)௣ܹ	ሾkWሿ ൌ ܳ	ሾ݉ଷ ݄⁄ ሿ ൈ ߂ ௧ܲሾ݌ܽ	ሿߝ௣ ൈ 3600 ൈ 1000  (4-2)WL is the input electrical energy of the lamp (kW), WO is the purchasing cost of the utilized oxidant (e.g., hydrogen peroxide) interchanged into an equivalent electrical energy (kW), Q is the volumetric flow rate of water (m3/h) and Ci and Cf are the inlet and outlet concentrations of the target pollutant (p-CBA), respectively. Assuming the pump’s electrical efficiency () is 0.75, the  83  input electrical energy of the pump (Wp) was calculated by Eq. (2). It should be noted that Pt represents the total pressure drop of the system including a net static head of 1 m between suction and discharge parts, in addition to the pressure drop within the vertically installed VUV/UV reactor.   Figure 4-5 Parts per billion (ppb) p-CBA concentration profiles within the improved design through introduction of central baffles Comparing the lamp and pump energy inputs for the investigated baffle-aided photoreactor, the required pumping power was much smaller (less than 60 times) than the lamp power. On the other hand, the equivalent electrical energy associated with the peroxide purchasing cost (WO) was significantly larger (more than six times) than that of the lamp. The peroxide cost conversion to an VUV/UV  UV 2ppm H2O2/UV 5ppm H2O2/UV 0                 45               90               135             180              225              270       300 53 ppb 280 ppb 92 ppb 25 ppb 3.5 L/min   3.5 L/min   3.5 L/min   3.5 L/min    84  equivalent electrical energy basis was performed assuming 0.15 $/kWh for electrical cost and 0.07 $/g peroxide for H2O2 purchasing (Chemical Market Reporter, 2013). It should be noted that the EEO analysis did not include the cost associated with quenching of the residual peroxide for each of the 2- and 5-ppm H2O2/UV processes.  The EEO values at three different flow rates (e.g., 1, 3.5 and 6.5 L/min) with and without the equivalent •OH scavenging impact of 2-ppm NOM surrogate are illustrated in Fig. 6. As shown in Figure 4-6 (a), in the absence of a surrogate •OH scavenger, the treatment cost of the VUV/UV process is minimal at different flow rates when compared with the UV, 2- and 5-ppm H2O2/UV processes. It is evident that peroxide cost affected significantly the overall treatment cost of the H2O2/UV processes. In addition, incorporation of central baffles reduced the energy requirement of the VUV/UV AOP by approximately 55%. However, energy savings obtained through introduction of central baffles in the 2- and 5-ppm H2O2/UV processes was only about 6%. Furthermore, the •OH scavenging impact of 2-ppm NOM raised significantly the EEO values for both VUV/UV and H2O2/UV AOPs. In the presence of a 2-ppm NOM surrogate (see Figure 4-6 a and Figure 4-6 c), the required energy of the H2O2/UV process increased more than that of the VUV/UV AOP.  This indicates the greater impact of NOM on increasing the EEO of the H2O2/UV compared with that of the VUV/UV process.   Overall, the extent of energy savings attained by means of the improved reactor hydrodynamics (e.g., incorporating central baffle) was less significant in the presence of a surrogate •OH scavenger equivalent to 2-ppm NOM. This is mainly attributed to the amplified •OH scavenging impact of NOM as the result of improved mixing provided through the integration of baffles within the photoreactor. However, as illustrated in Figure 4-6 (d), the VUV/UV process showed significant superiority in terms of p-CBA treatment cost over the UV, 2- and 5-ppm H2O2/UV processes.  85   Figure 4-6 Electrical energy-per order (EEO) analysis of the conventional photoreactor and improved design (central baffles)   4.4.4 Improved VUV/UV photoreactor design Given the significant impact of reactor hydrodynamics on the removal efficacy of the VUV/UV process, CFD simulation was utilized to evaluate a wide range of baffle configurations in terms of 0123456781.0 3.5 6.5EEO (kWh/m3/order)Flow rate (L/min) VUV/UV UV  2ppm H2O2/UV  5ppm H2O2/UV0123456781.0 3.5 6.5EEO (kWh/m3/order)Flow rate (L/min) VUV/UV  UV  2ppm H2O2/UV  5ppm H2O2/UV048121620241.0 3.5 6.5EEO (kWh/m3/order)Flow rate (L/min)048121620241.0 3.5 6.5EEO (kWh/m3/order)Flow rate (L/min)Without NOM (b) Central baffle-aided design 2 ppm NOM 2 ppm NOM  (a)  Conventional design  Without NOM (d) Central baffle-aided design (c)  Conventional design   86  their p-CBA degradation rate. Baffles with identical dimensions were retrofitted within the base reactor attached to the reactor-wall, quartz-sleeve or in between the reactor wall and quartz sleeve (central configuration). Analyzing the CFD-simulated results showed much greater p-CBA removal efficacy for the central baffle configuration. This was mainly because the quartz sleeve attached baffles blocked approximately 20% of the valuable VUV photons that could effectively contribute to the generation of •OH. Meanwhile, the sole installation of wall-attached baffles or central baffles led to fluid short-circuiting or channeling, which negatively affected the p-CBA removal efficacy of the VUV/UV process. Nonetheless, limited flow channeling with a more effective mixing pattern was accomplished through a combinatorial design consisting of reactor-wall-attached and central baffles in a sequential arrangement. Figure 4-7 compares the extent of p-CBA removal enhancement obtained through implementation of the central baffles when compared with the combined design approach. Using a Lagrangian particle tracking approach (Bagheri and Mohseni, 2014; Sozzi and Taghipour, 2006) in the CFD platform, p-CBA exposure time to hydroxyl radicals was quantified for each baffle design scenarios. The combined wall-attached and central baffle aided design showed a more uniformed •OH dose distribution with approximately 15% greater average value when compared with the central baffle design. Using this improved photoreactor design, the p-CBA removal performance of the VUV/UV system surpassed that of the 5-ppm H2O2/UV process, underscoring the significance of hydrodynamics in the VUV-induced degradation of target pollutants.   87   Figure 4-7 Comparison of the p-CBA concentration (ppb) contours within the improved VUV/UV photoreactor using central and combinatorial baffles 4.5 Chapter conclusions This chapter employed a complete numerical modeling tool for better understanding of degradation mechanisms and removal efficiency of VUV/UV process when compared with the widely studied H2O2/UV AOP under continuous flow condition. Of particular interest was the comparison of the energy cost competiveness of VUV/UV AOP over the H2O2/UV process. While hydroxyl radical (•OH) oxidation was the dominant degradation pathway in both processes, under uniform flow regime H2O2/UV with 2- and 5 ppm H2O2 provided greater degradation efficiency of the target pollutant than VUV/UV process. However, the introduction of mixing and circulation zones to the VUV/UV system resulted in its superior performance compared with the 2-5 ppm H2O2/UV AOP. Comparing the •OH concentration profile within the prototype photoreactor, VUV/UV AOP showed a significantly larger concentration of •OH with a sharper gradient than that of the H2O2/UV process. In addition, EEO analysis suggested that incorporation of the Central Combinatorial 0                45               90              135            180           225            270      300 53 ppb 22 ppb 3.5 L/min  3.5 L/min   88  circulation zones resulted in up to 50% reduction in the overall energy cost of VUV/UV AOP, while less than 5% reduction was observed for the 5-ppm H2O2/UV process. The extent of energy saving attained by means of improved reactor hydrodynamics (e.g., incorporation of a central baffle) was less significant in the presence of a surrogate •OH scavenger equivalent to 2-ppm NOM. Overall, removal performance of the VUV/UV was strongly correlated to the hydrodynamics of the system. This important consideration requires special attention when Vacuum-UV AOPs are engineered using annular photoreactors.      89  Chapter 5: Pilot-scale remediation of surface water co-contaminated by 1,4-dioxane and atrazine using a VUV/UV process: A CFD-based study  5.1 Chapter introduction The presence of organic micropollutants (OMPs) in water supplies is of great environmental and public health concern. OMPs consist of a vast and expanding array of natural (e.g., mycotoxins and phytoestrogens) as well as anthropogenic substances such as chemical solvents, pesticides, herbicides, pharmaceuticals, personal care products and so on. Although such contaminants are being detected typically at low concentrations (e.g., a few ng L-1 to several µg L-1), the concern is primarily due to their potential to cause chronic health problems. OMPs can enter the water supply from various sources and conventional treatment processes have not shown to be effective for the elimination of micropollutants from water supplies (Ivančev-Tumbas, 2014; Luo et al., 2014; Stackelberg et al., 2007).  Over the past decades, treatment technologies such as advanced oxidation processes (AOPs), reverse osmosis (RO), and adsorption by activated carbon (AC) have been considered as alternatives for the removal of a wide range of OMPs. Particularly, AOPs are advantageous since they are capable of completely or partially oxidizing the target pollutant, rather than transferring it into other phases that would be the case in RO and AC (Sudhakaran et al., 2013). Among various AOPs, combination of ultraviolet radiation and hydrogen peroxide (H2O2/UV) has been demonstrated very promising, thereby widely-employed for the removal of OMPs at large-scale treatment facilities (Kruithof et al., 2007). Despite wide applications, H2O2/UV process suffers from the high cost of hydrogen peroxide including purchasing, transportation and storage (Gunten, 2003). More importantly, the need for regular monitoring and post treatment of the surplus  90  peroxide impose further costs and complexity to the H2O2/UV process, specifically for applications in small to medium-sized water treatment facilities with stringent operational and financial constraints. Alternatively, Vacuum-UV (VUV) involving 185 nm irradiation is an incipient AOP technology that eliminates the demanding need for an auxiliary chemical oxidant, e.g., hydrogen peroxide, resulting in a greener and simpler process with greater economic interests for commercial applications. In the past, the efficiency and mechanisms of the VUV process for the removal of a wide range of OMPs (e.g., pesticides, herbicides, taste and order compounds, pharmaceutical and personal care products) under batch laboratory scale photoreactors have been studied extensively (Duca et al., 2013; Imoberdorf and Mohseni, 2011b; Zoschke et al., 2014). For example, Imoberdorf and Mohseni (2012) presented a complete kinetic model for the description of VUV induced degradation of a model herbicide (2,4-dichlorophenoxyacetic acid) in absence and presence of the main •OH scavengers (e.g., NOM, alkalinity). For both conditions, the authors demonstrated a good agreement between the model predictions and experimental results obtained from a mixed batch VUV/UV reactor. Recently, Bagheri and Mohseni (2014) developed a two-dimensional computational fluid dynamics (CFD) model to investigate thoroughly the of continuous-flow degradation of OMPs in ultrapure (MilliQ) water samples. Using a simplified lab-scale VUV photoreactor with axisymmetric geometry, the model showed to predict well experimental degradation of the target OMPs for synthetic water samples at 0.5-6 L/min. Despite the promising results obtained in lab-scale experiments, lack of studies demonstrating the removal efficacy of the pilot scale VUV photoreactors with real water samples are among the key factors hindering their practical implementation. Furthermore, the extent to which background water constituents limit the AOP performance of the VUV systems requires assessment under continuous flow  91  operation. In this case, the presence of models validated rigorously under a wide range of flow conditions and water compositions can significantly facilitate development of VUV systems for effective remediation OMPs from water sources.   The primary objective of this chapter was to utilize the results of our previous chapters and build a comprehensive three-dimensional CFD model for analyzing degradation of OMPs in natural surface water samples using a pilot-scale VUV photoreactor operating at 6-30 L/min flow. Experimental validation of the proposed model was fulfilled through monitoring of degradation of two different model OMPs (a widely used herbicide, atrazine, and a commonly-used chemical solvent, 1,4-dioxane, simply refer as dioxane) in a pilot-scale VUV system consist of two photoreactors in series. Further, impact of operational parameters such as flow rate and main water matrix constitutes (e.g., natural organic matter and alkalinity) on the removal efficacy of the model micropollutants was predicted and evaluated experimentally. The presented model offers a cost-effective benchmark for the design, optimization and scale-up of VUV systems, leading to their practical application in water treatment trains to facilitate removal of OMPs from aquatic environments.  5.2 Materials and methods 5.2.1 Experimental setup and procedure Experimental work involved operating a pilot-scale VUV system comprising of two photoreactors installed in series (shown in Figure 5-1). Each reactor was made of stainless steel, with an annular configuration, and was equipped with a 140 W low-pressure mercury amalgam ozone-generating lamp (Strahler NAQ 170/90 XL, Heraeus Noblelight), emitting 185 nm and 254 nm UV, with 19 mm diameter and illuminated length of 800 mm. The internal and external diameters of the annular  92  photoreactor and synthetic quartz sleeve (Heraeus Quartz America LLC, SUPRASIL310) were 5.5 cm and 2.8 cm, respectively, yielding a radial gap of 1.35 cm corresponding to a hydraulic diameter of 2.7 cm for treating the water. Two perforated baffles were installed with 50 cm pitch. The entrance baffles were placed at 5 cm distance from the inlet cones.             1. Contaminated water 4. Ozone-generating low pressure mercury amalgam 2. Perforated baffle  lamp (185 nm + 254 nm) 3. Quartz sleeve (Suprasil 310) 5. Treated effluent Figure 5-1 Schematic of utilized VUV pilot set-up for experiments Natural water samples were collected from the Capilano Reservoir (CR), in British Columbia, Canada. The main properties and characteristics of the natural water used in this study were presented in Table 1. The unfiltered samples were stored at 4°C prior to use. Synthetic waters were prepared using distilled water (DW) supplemented with NOM isolate from Suwannee River 1 2 5 3 4  93  (1R101N), purchased from the International Humic Substances Society. The synthetic water samples were buffered at pH 7 using a mixture of monosodium phosphate and disodium phosphate. Further, synthetic water samples were purged with pure oxygen to ensure an oxygen content close to saturation level at 21.5 °C. Inlet water temperature was set 21.5 (± 0.1) °C. Table 5-1 Natural water properties (Capilano Reservoir, BC) DOC0 [mg/L] 1.8 UV 254,0 [cm-1] 0.066 SUVA0 3.6 pH0 6.5 Conductivity [µS/cm] 11 Chloride [mg/L] 0.6 Dissolved oxygen [mg/L] 7.8 Alkalinity [mg/L CaCO3] 2.6  5.2.2 Analytical methods The concentration of 1,4-dioxane (dioxane) was analyzed with a gas chromatograph equipped with mass spectrometer (GC/MS, Saturn 2200, Varian). The instrument was equipped with CombiPal auto sampler. Atrazine concentration was quantified by high-performance liquid chromatography (Dionex UltiMate™ 3000, US) with the detection wavelength set at 222 nm. Concentrations of NOM and its oxidation by-products were quantified by dissolved organic matter (DOC) as mgcarbon per L, using a TOC Analyzer (Shimadzu TOC-VCPH). The concentration of hydrogen peroxide was determined via UV spectrophotometry utilizing the ܫଷି  method (Klassen et al., 1994). For pH adjustment, a phosphate buffer solution (a mixture of monosodium phosphate and disodium phosphate (Sigma-Aldrich, Canada) was employed. It is worth mentioning that error bars  94  represented in the current study are the standard deviation from the mean of three independent experiments with three samples injections. It is worth mentioning that error bars represented in the current study are the standard deviation from the mean of three independent experiments. 5.3 Model development The CFD model utilized in this investigation was built on our previous study that simulated the VUV-induced degradation of micropollutants in a symmetrical lab-scale photoreactor with ultrapure water samples (Bagheri and Mohseni, 2014). This section aimed to expand capability of the initial proposed CFD model for application in VUV photoreactors with a more complicated configuration (e.g., asymmetrical geometry with retrofitting baffles) and with natural water samples.   Figure 5-2 Generated mesh for the pilot VUV system  95  CFD simulation of VUV process necessitates simultaneous resolution of hydrodynamics, fluence rate at both 185 nm and 254 nm wavelengths, and a comprehensive kinetic scheme. In doing so, ANSYS® 15 simulation platform was employed to build and discretize the pilot plant geometry via 357390 cells with a dominant fraction of structured hexahedral cells (e.g., > 95%). Grid dependency analysis was performed to ensure less than 1% change in the concentration of target pollutant in the outlet, by refining the grid sizes. Besides, ANSYS® Fluent 15 was employed to read the exported mesh and perform the CFD computations. Figure 5-2 displays a schematic of the discretized geometry of the pilot system using ANSYS® mesh generator. As for the hydrodynamics sub-model, Realizable k-epsilon (k-) equation couples with enhanced wall treatment modification was utilized under 4700 < Re < 23500 (Duran et al., 2009). The utilized kinetic rate equations for the degradation of the target pollutants, atrazine and 1,4-dioxane, along with NOM •OH scavenging, and alkalinity related reactions are described in Table 5-2 (Bagheri and Mohseni, 2014, 2015; Gonzalez et al., 2004; Westerhoff et al., 2007). Fluence rate fields for 185 nm and 254 nm photons were computed by solving the radiative transport equation (RTE) using non-gray discrete ordinate (DO) sub-model. Furthermore, the reflection, refraction, and absorption of photons within the air gap (separating the lamp and the quartz sleeve region) and on the quartz sleeve were incorporated within the radiation sub-model. Moreover, water temperature within the reactor was defined 23 ºC,  using arithmetic mean temperature between of the inlet and outlet, corresponding to 185 nm water absorption coefficient of 1.2 cm-1 (Barrett and Mansell, 1960). Measured UV254 transmittance (UVT) of the samples was 95% and 0.86% for the synthetics and raw water samples (CR), which utilized in the simulations. More details about the utilized CFD sub-models, boundary condition, solution method and strategy can be found elsewhere (Bagheri and Mohseni, 2014, 2015; Bagheri et al., 2013). The total execution time of  96  each simulation was about 9 h on a 3.50 GHz AMD Phenom™ II X4 970 processor with 8 GB RAM. The 185 nm and 254 nm intensity inputs of the VUV lamp (ozone generating), were 2.9 ± 0.2 W and 46 ± 1 W, respectively. The reported intensities were based on the radiometry results (Bagheri and Mohseni, 2014; Sasges et al., 2012) obtained from a 70W VUV lamp (Strahler NIQ 60/35 XL, Heraeus Noblelight) with the identical technology to those lamps used for this study. This was mainly because experimental measurement of 185 nm output of VUV lamps requires continuous purging of radiometry apparatus with pure nitrogen gas (to assure elimination of oxygen gas, which absorbs 185 nm photons). Thereby, conducting the radiometry tests by a VUV lamp with a shorter length will bring more stability over the experimental variables along with significant reduction in nitrogen consumption of the system (Bagheri and Mohseni, 2014).  Table 5-2 List of the reactions taking place for degradation of 1,4-dioxane and atrazine in the 185 nm oxidation process No. Equilibrium reactions Constant 1 ܱଷ•	ି ൅ ܪା ↔ ܪܱଷ• pKa1 = 8.2 2 ܪଶܱଶ ൅ ܱܪି ↔ ܪଶܱ ൅ ܪܱଶି pKa2 = 11.6 3 ܪܱ• ൅ ܪܱି ↔	ܪଶܱ ൅ ܱ•ି pKa3 = 11.9 4 ܪܱଶ• 	↔ 	ܪା ൅ ܱଶ•ି pKa4 = 4.8 5 ܱܪି ൅ ܪା ↔	ܪଶܱ Kw = 10-14 6 ܪଶܥܱଷ ↔	ܪା ൅ ܪܥܱଷି  pKa6 = 6.3 7 ܪܥܱଷି ↔ ܪା ൅ ܥܱଷଶି pKa7 = 10.3             Photochemical reactions 8 ܪଶܱ ൅ ݄νଵ଼ହ	௡௠ 	 	→	ܪܱ• ൅ ܪ•   Φ8 = 0.33  mole.ein-1 9 ܪଶܱ ൅ ݄νଵ଼ହ	௡௠ 	 	→	ܪା ൅ ݁ି௔௤ ൅ ܪܱ•   Φ9 = 0.045 mole.ein-1 10 ܪଶܱଶ ൅ ݄νଵ଼ହ	௡௠ 	 	→	2	ܪܱ•   Φ10 = 0.5  mole.ein-1 11 ܪଶܱଶ ൅ ݄νଶହସ	௡௠ 	 	→	2	ܪܱ•   Φ11 = 0.5  mole.ein-1             Contaminant reactions 12 ܥସܪ଼ܱଶ	ሺ1,4 െ ܦ݅݋ܽݔ݊݁ሻ ൅ ܪܱ• → ∗ ܦ݅݋ݔ െ ܱܪ   k12 = 2.8×109    M-1.s-1 13 ܦ݅݋ݔ െ ܱܪ ൅ ܪܱ• 	 	→	∗ ܦ݅݋ݔ െ ܱܪܱܪ *k13 = 1×109        M-1.s-1 14 ܥ଼ܪଵସܥ݈ ହܰ	ሺܣݐݎܽݖ݅݊݁ሻ ൅ ܪܱ• → ∗ ܣܴܶ െ ܱܪ   k14 = 2.4×109   M-1.s-1 15 ܥ଼ܪଵସܥ݈ ହܰ ൅ ݄νଶହସ	௡௠ 	 	→ ܥ଼ܪଵସܥ݈ ହܰ െ 254 ሺ∗ ܣܴܶ254ሻ  Φ15 = 0.05  mole.ein-116 ܥ଼ܪଵସܥ݈ ହܰ ൅ ݄νଵ଼ହ	௡௠ 	 	→ ܥ଼ܪଵସܥ݈ ହܰ െ 185 ሺ∗ ܣܴܶ185ሻ *Φ16 = 0.05  mole.ein-117 ܣܴܶ െ ܱܪ ൅ ܪܱ• 	 	→	∗ ܣܴܶ െ ܱܪܱܪ *k17 = 2.4×109   M-1.s-1  97  No. Contaminant reactions Constant 18 ܣܴܶ185 ൅ ܪܱ• 	 	→	∗ ܣܴܶ െ ܱܪ *k18 = 2.4×109   M-1.s-1 19 ܣܴܶ254 ൅ ܪܱ• 	 	→	∗ ܣܴܶ െ ܱܪ *k19 = 2.4×109   M-1.s-1              Other involving reactions 20 ܪܱ• ൅ ܪଶ 	 	→	ܪଶܱ ൅ ܪ•   k20 = 3.9×107   M-1.s-1 21 ܪܱ• ൅ ܪܱ• 	 	→	ܪଶܱଶ   k21= 4.2×109     M-1.s-1 22 ܪܱ• ൅ ܪ• 	 	→	ܪଶܱ   k22 =7.0×109   M-1.s-1 23 ܪܱ• ൅ ݁ି௔௤ 	 	→	ܪܱି k23 = 3.0×1010  M-1.s-1 24 ܪܱ• ൅ ܪܱଶ• 	 	→	ܪଶܱ ൅	ܱଶ k24 = 6.6×109   M-1.s-1 25 ܪܱ• ൅ ܱଶ•ି 	 	→	ܱଶ ൅ ܱܪି k25 = 1.1×1010  M-1.s-1 26 ܪܱ• ൅ ܪܱଶି 	 	→	ܱଶ•ି	 ൅ ܪଶܱ k26 = 6.8×109   M-1.s-1 27 ܪܱ• ൅ ܪଶܱଶ 	 	→	ܪܱଶ• ൅ ܪଶܱ k27 = 2.7×107   M-1.s-1 28 ܪܱଷ• 	→	ܪܱ• ൅	ܱଶ k28 = 1.1×105    s-1 29 ܱଷ• ൅ ܪା 	 	→	ܪܱ• ൅	ܱଶ k29 = 9.0×1010  M-1.s-1 30 ݁ି௔௤ ൅ ܪଶܱଶ 	 	→	ܱܪି ൅ ܪܱ• k30 = 1.3×1010   M-1.s-1 31 ܪଶܱଶ ൅ ܪ• 	 	→	ܪܱ•ି ൅ ܪଶܱ k31 = 5.0×107   M-1.s-1 32 ܪܱଶ• ൅	ܪ• 	 	→ 	ܪଶܱଶ k32 = 2.0×1010  M-1.s-1 33 ܪܱଶ• ൅	ܱଶ•ି ൅ ܪଶܱ	 	→	ܪଶܱଶ ൅ ܱଶ ൅ ܪܱି k33 = 9.7×107     M-1.s-1 34 ܪܱଶ• ൅ ܪܱଶ• 	 	→	ܪଶܱଶ ൅ ܱଶ k34 = 8.3×105   M-1.s-1 35 ݁ି௔௤ ൅ ܱଶ 	 	→	ܱଶ•ି k35 = 1.8×1010  M-1.s-1 36 ݁ି௔௤ ൅ ܪଶܱ	 	→	ܱܪି ൅ ܪ• k36 = 1.0×103    s-1 37 ݁ି௔௤ ൅ ܪ• ൅ 	ܪଶܱ	 	→	ܱܪି ൅ ܪଶ k37 = 3.4×1010  M-1.s-1 38 ݁ି௔௤ ൅	݁ି௔௤ ൅	2	ܪଶܱ	 → 2 ܱܪି ൅ ܪଶ k38 = 6.0×10 9  M-1.s-1 39 ݁ି௔௤ ൅ ܪା 	 	→ ܪ• k39 = 2.8×1010  M-1.s-1 40 ݁ି௔௤ ൅ ܪܱଶି 	 	→	ܱܪି ൅ ܱ•ି k40 = 3.5×109   M-1.s-1 41 ܪܱି ൅ ܪ• 	→	݁ି௔௤ ൅	ܪଶܱ k41 = 2.5×107    M-1.s-1 42 ܱଶ ൅ ܱ•ି 	→	ܱଷ•ି k42 = 3.8×109   M-1.s-1 43 ܱଶ•ି ൅ ܪ• 	→	ܪܱଶି k43 = 2.7×107   M-1.s-1 44 ܱଶ ൅ ܪ• 	 	→	ܪܱଶ• k44 = 1.2×1010  M-1.s-1 45 ܱ•ି ൅	ܪܱଶି 	 	→	ܱܪି ൅	ܱଶ•ି k45 = 4.0×108   M-1.s-1 46 ܪ• ൅ ܪ• 	 	→	ܪଶ k46 = 5.5×109   M-1.s-1 47 ܱܰܯ ൅ܪܱ• →∗ ܱܰܯതതതതതതത		 k47 = 1.4×108   M-1.s-148 ∗ ܱܰܯതതതതതതത ൅ ܪܱ• →∗ ܱܰܯധധധധധധധ k48 = 1.4×108   M-1.s-149 ܥܱଷଶି ൅ ܪܱ• 	→ 	ܥܱଷ•ି ൅ ܱܪି k49 = 4.0×108   M-1.s-1 50 ܪܥܱଷି ൅ ܪܱ• 	→ 	ܥܱଷ•ି ൅ ܪଶܱ k50 = 8.5×106   M-1.s-1 51 ܥܱଷ•ି ൅ ܪܱ• 	→∗ ܪܥ ସܱି  k51 = 3.0×109  M-1.s-1 52 ܥܱଷ•ି ൅ ܪଶܱଶ 	→ 	ܪܥܱଷ•ି ൅ ܪܱଶ• k52= 4.3×105  M-1.s-1 53 ܥܱଷ•ି ൅ ܪܱଶି 	→ 	ܥܱଷଶି ൅ ܪܱଶ• k53 = 5.0×107  M-1.s-1 54 ܥܱଷ•ି ൅ ܱଶ•ି	 	→ 	ܥܱଷଶି ൅ ܱଶ k54 = 7.0×108  M-1.s-1 55 ܥܱଷ•ି ൅ ܥܱଷ•ି 	→	ܥଶܱ଺ଶି k55 = 1.3×107  M-1.s-1  Starred reactants/products are hypothetical and are only defined for model development  Starred values are estimated.     98  5.4 Results and discussion 5.4.1 CFD model validation and analysis  Figure 5-3 (a) displays the CFD model predictions along with the experimental data for the degradation of 100 ppb atrazine in synthetic water samples, including distilled water (DW) and 2 ppm SRNOM, for the pilot-scale VUV apparatus operating at 6-30 L/min flow rates. Similarly, Figure 5-3 (b) presents comparison of experimental and predicted data for degradation of 100 ppb 1,4-dioxane (dioxane) in synthetic water samples. It should be noted that experimental degradation data represent the mean of three independent replicates at any given flow rate. Further, the predicted degradation rates did not change with variations of the model input variables (e.g., kinetic rate constants) within the reported accuracy range (Bagheri and Mohseni, 2014). The sign “R1” represents the first reactor outlet and “R1+R2” indicates the overall performance of the pilot VUV system attained by both reactors operating in series. Based on the utilized experimental apparatus, the Reynolds numbers were in the range of 4700 (6 L/min) to 23500 (30 L/min) which corresponds fully turbulent flow regime in the photoreactors.  As can be seen from Figure 5-3, the measured and predicted values for the degradation of atrazine and dioxane were in good agreement over the entire dataset. The relative discrepancy in the predicated degradation rate at low flow rates (e.g., 6 L/min) may arise from an increasing concentration of unknown/undefined intermediates and degradation by-products at lower flow rates. A greater extent of deviation of atrazine predictions was consistent with our observation in the simplified lab-scale VUV photoreactor for the low flow rates when a great fraction of the inlet atrazine concentration underwent degradation (Bagheri and Mohseni, 2015).   99    Figure 5-3 Comparison of CFD predictions and experimental data for degradation of 100 ppb (a) atrazine (b) dioxane in distilled water and 2 ppm SRNOM samples 0.00.10.20.30.40.50.60.70.80 6 12 18 24 30C/C0Flow rate (L/min)  Exp. Atrazine R1  Exp. Atrazine R1+R2  Pre. CFD model0.00.10.20.30.40.50.60.70.80 6 12 18 24 30C/C0Flow rate (L/min)(b)  Exp. 1,4-Dioxane R1  Exp. 1,4-Dioxane R1+R2  Pre. CFD model(a) 100  In other side, model deviations in predicted degradation ratios of atrazine and dioxane at higher flow rates (e.g., 30 L/min) may be mainly attributed to the complexities associated with the turbulent fluctuations at higher Reynolds values (Alpert et al., 2010; Wols, 2012). This observation was consistent to the past studies that reported an underestimation of predicted degradation rates at high turbulent conditions specifically at pilot-scale UV-AOP reactors (Alpert et al., 2010).   While near complete removal of the individual target pollutants from contaminated water at very low flow rates of 6 L/min, significant reductions in the removal efficacy of the system was evident with increasing the flow rates that requires further investigation. Comparing the degradation of atrazine and dioxane (shown in Figure 5-3), a greater removal of atrazine was obtained when compared to dioxane at different flow rates. This is more noticeable at lower (e.g., 6 L/min) flow rates than higher flow rates (e.g., 30 L/min). Given the higher reactivity of dioxane (kOH = 2.8×109 M-1.s-1) than atrazine (kOH = 2.4×109 M-1.s-1) to react with OH radical, a greater extent of dioxane removal is anticipated via •OH-oxidation pathway. On the contrary, while UV-direct photolysis partly contributes to the •OH-oxidation of atrazine, it is an irrelevant removal mechanism for dioxane. Thus, combination of UV-direct photolysis and •OH-oxidation favored the degradation of atrazine which resulted in its greater removal when compared to dioxane. To investigate •OH scavenging effect of a competing organic micropollutants on degradation of a pollutant of interest, simultaneous degradation of atrazine and dioxane were compared to their individual degradation. Figure 5-4 (a) displays the CFD model predictions and experimental data for degradation of atrazine in synthetic water (DW+2 ppm NOM) co-contaminated by an equivalent concentration of dioxane (100 ppb) using the pilot VUV system. Similarly, Figure 5-4 (b) shows the predicted and experimental data for degradation of dioxane in synthetic water co-contaminated by an equivalent concentration of atrazine (100 ppb).   101   Figure 5-4 Comparison of CFD predictions and experimental data for •OH scavenging effect of 100 ppb (a) dioxane on degradation of 100 ppb atrazine (b) atrazine on degradation of 100 ppb dioxane  0.00.10.20.30.40.50.60.70.80 6 12 18 24 30C/C0Flow rate (L/min)  Exp. Atrazine R1 ( + Diox)  Exp. Atrazine R1+R2 ( + Diox)  Exp. Atrazine R1  Exp. Atrazine R1+R2  Pre. CFD model0.00.10.20.30.40.50.60.70.80 6 12 18 24 30C/C0Flow rate (L/min)  Exp. 1,4-Dioxane R1 (+ Atr)  Exp. 1,4-Dioxane R1+R2 (+ Atr)  Exp. 1,4-Dioxane R1  Exp. 1,4-Dioxane R1+R2  Pre. CFD model(a) (b)  102  As shown, CFD model predicts showed to be in good agreement with the experimental data over the entire range of flow rates. It is worth mentioning that apart from the accurate prediction of the target pollutants degradation in the utilized VUV system, the proposed CFD model predicted well the removal of up to about 10-20% DOC (from the initial 2-ppm concentration) depending upon the flow rates. Similarly, the maximum hydrogen peroxide (H2O2) concentration in the outlet of the pilot system, operating at 6 L/min, was 0.6 (±0.1) ppm that was consistent with the model predictions. Comparing the pilot system performance for waters co-contaminated by dioxane and atrazine (Figure 5-4) to that of their single treatment (Figure 5-3), the removal efficacy of system was dropped because of the competition of each pollutant for reacting with OH radicals. While this reduction in the removal efficacy of atrazine was about 5-15%, dioxane degradation was reduced approximately by 5-25%, depending on the flow rates. As anticipated, the simultaneous presence of both pollutants influenced the degradation of dioxane more than atrazine because part of the atrazine degradation triggered by UV-direct photolysis will not be influenced by the presence of dioxane in water. However, atrazine competes with dioxane for reaction with •OH-oxidation which is the only relevant degradation mechanism for dioxane removal. Overall, effective removal of both pollutants (e.g., about 90%) was observed at low flow rate of 6 L/min. 5.4.1.1 UV and VUV fluence rate distributions Figure 5-6 (a,b) displays the VUV 185 nm and UV 254 nm fluence rate distributions within the pilot system, respectively. As shown, 185 nm and 254 nm incident radiations decreased with increasing radial distance from the lamp because of absorption in water and the increasing circumferential area. As displayed in Figure 5-6 (a), significant fractions of VUV photoreactors   103    Figure 5-5 (a) VUV 185 nm, (b) UV 254 nm incident radiation profile along the VUV/UV pilot system  26 23 20 16 13 10 7 3 0 W/m2528 469 411 352 293 235 176 117 59 W/m213.5 mm 13.5 mm (b) (a)  104  are dark zones with respect to 185 nm radiation due to the very high absorption coefficient of water at this wavelength (1.4 cm-1 at 25 °C) (Barrett and Mansell, 1960; Halmann, 1966). Nonetheless, insignificant attenuation of 254 nm radiation (Figure 5-6 b) was anticipated because of the high UV transmittance of synthetic water samples (e.g., UVT=95%). 5.4.1.2 Role of different degradation pathways Knowing the predominant role of •OH-oxidation in the degradation of dioxane (Martijn et al., 2010), impact of different degradation pathways on the overall removal of atrazine was investigates. Based on the average kinetic rates, •OH-degradation rate of atrazine was approximately 3 times larger than that associated with 254 nm direct-photolysis, and about 50 times greater than the rate of 185 nm direct-photolysis.      Figure 5-6 •OH concentrations (mol/L) contour within the VUV pilot system at flow rate of 6 L/min  ×10-10  0        0.3        0.6        0.8      1.1       1.4     1.7      2     2.2       2.5        13.5 mmM   105  The greater contribution of •OH-oxidation pathway in comparison to 185 nm and 254 nm direct-photolysis in the VUV process were also concluded in our earlier investigation using a lab-scale photoreactor at different initial concentration of atrazine (Bagheri and Mohseni, 2015). Although, the more pronounced role of •OH oxidation pathway observed in the lab-scale photoreactor (e.g., more than eight times) was primarily due to the absence of NOM in background water composition that magnified •OH based degradation of the target pollutant. Given the key role of •OH concentration, Figure 5-5 displays the local concentration contours of •OH in the utilized pilot system for 6 L/min flow. Hydroxyl radical is densely concentrated in the proximity of the quartz sleeves, protecting the lamps, with a maximum of 2.5×10-10 M. In addition, it was demonstrated in Figure 5-5 that nearly 10 times reduction in the •OH concentration occurs within 7 mm distance from the quartz sleeves. This underlines mass transfer of the target contaminant to the reacting zone as a key parameter, which controls the pollutant degradation rate in the VUV photoreactors. 5.4.1.3 Local concentration of target micropollutants To gain better understanding of the removal performance of the VUV pilot system, local concentration contours of the target pollutants were investigated for the central 2D plane along the horizontal axis. As shown in Figure 5-6 (a,b), at 6 L/min the pilot system degraded the target pollutants, from 100 ppb (e.g., inlet stream) to 7 ppb atrazine and 10 ppb dioxane in the outlet flow. Moreover, outlet concentration of atrazine and dioxane from the first reactor was 26 ppb and 34 ppb, respectively. Nonetheless, a greater extent of atrazine degradation was accomplished in proximity of the quartz sleeve. Based on Figure 5-6, sudden shifts in colors, and consequently effective degradation of the target pollutants took place where perturbing flows created by the inlet sections and baffles were evident. In particular, the inlet flow impinges on the inner tube, from the front for the L-shape configuration, and then hitting the perforated baffles causing the split of the  106  flow and the creation of flow separation, recirculation, and reattachment zones, which significantly improves pollutants mass transfer to the reacting zone. Since OH radical is a very short-lived species with greater concentration in the vicinity of quartz sleeve, mixing and circulation zones can significantly stimulate the •OH oxidation of the target pollutants.   Figure 5-7 Parts per billion atrazine and dioxane concentrations within the VUV pilot system at 6 L/min On the contrary, at 30 L/min flow the atrazine and dioxane degradation efficacy of the system was only about 45% and 38%, respectively (shown in Figure 5-8). Thus, operation of the pilot system at higher flow rates (e.g., 30 L/min) affected significantly the AOP performance of the VUV reactors. However, in both 6 and 30 L/min flows, the second VUV photoreactor showed less 7  10  34  26  atrazine1,4-dioxaneppb 0         10         20          30          40          50         60          70          80         90       100 (a) (b)  107  atrazine removal when compared to the first reactor. This is mainly due to the lower atrazine concentration in the second reactor inlet that will reduce atrazine degradation rate (k•OH×Catr).  Figure 5-8 Parts per billion atrazine and dioxane concentrations within the VUV pilot system at 30 L/min  5.4.2 Effect of the flow rates on micropollutant degradation  To understand further the impact of flow rate on the removal efficacy of the pilot system, flow streamlines colored by the dioxane local concentration were shown at 6 and 30 L/min (Figure 5-9). Streamlines are series of curves that are tangent to the velocity vectors of the fluid flow and are widely employed for the better understanding of the velocity and pressure distribution in a given system. As can be seen in Figure 5-9, by altering the flow of the pilot system the flow characteristics (hydrodynamics) of the VUV photoreactors was affected largely. Based on Figure ppb 0         10         20          30          40          50         60          70          80         90       100 62  1,4-dioxane55  atrazine78 76  (a) (b)  108  5-9 (b), the pilot VUV system suffers from lack of mixing, circulation zones particularly at higher flow rates that channeling and short-circuiting were intensified. Thereby, with increasing the flow the turbulence created by change of direction and expansion then transported by convection and pollutant mass transfer decreases as the flow redevelops downstream of the baffles. This underlines the critical role of hydrodynamics of VUV photoreactors on their AOP performance.   Figure 5-9 Flow streamlines colored by dioxane concentrations at 6 and 30 L/min flow rates  ppb 0         10         20          30          40          50         60          70         80         90       100 10 ppb  62 ppb  (a) (b)  109  It is worth mentioning that changing the flow rate not only affected flow characteristics (hydrodynamic) of the pilot system, but also will leave significant impact on the residence (retention) time of the VUV reactors. The longer the retention time (lower the flow rates), the higher the concentration of •OH from direct-photolysis of water molecules. To gain an in-depth insight toward the impact of flow rates on degradation efficacy of the pilot system, a Lagrangian particle tracking approach was used to analyze the contaminant exposure to hydroxyl radicals, as well as UV 254 nm and VUV 185 nm photons (•OH, UV and VUV dose distributions) (Santoro et al., 2010; Sozzi and Taghipour, 2006b). In the Lagrangian model, uniformly distributed particles were released at the reactor inlet (presenting the target pollutants) where the velocity distribution was also uniform, and the received •OH, UV and VUV doses were integrated along the particle trajectories using the random walk model. The delivered dose at each point was calculated by multiplying the average local value of a property of interest (e.g., •OH concentration, UV or VUV radiation intensities) by the time period t for which the particle was exposed to that value. Afterwards, frequency distributions for each property of interests were plotted at 6 and 30 L/min flow rates (Figure 5-10).  0 5 10 15 20 25 301.83.24.66.07.48.710.111.512.914.315.617.0•OH dose delivery (M.s)(a)×10 -100 5 10 15 20 25 300.41.11.92.73.54.35.15.96.67.48.29.0•OH dose delivery (M.s)(b)×10 -10% %  110  Figure 5-10 •OH dose (a, b), residence time (c, d), UV dose (e, f), and VUV dose (g, h) distributions within the pilot system at 6 L/min (in red) and 30 L/min (in blue) flow rates As shown in Figures 5-10 (a, b), approximately 2.5 times greater •OH doses were obtained from 6 L/min flow than from 30 L/min. Meanwhile, the retention time of the system at 30 L/min varied between 7.6 and 14.3 sec where 90% of the particles stayed less than 10 sec in the system. Whereas 0 5 10 15 20 25 30 35 4018.920.021.022.123.224.325.326.427.528.629.630.7Residence time distribution (sec) (c)0 5 10 15 20 25 30 35 407.68.28.89.410.010.611.211.812.413.013.714.3Residence time distribution (sec) (d)0 5 10 15 20 25248280312344377409441473505537569601UV dose distribution (mJ/cm2 )(e)0 5 10 15 20 257393114134155175196217237258278299UV dose distribution (mJ/cm2 )(f)0 5 10 15 20 25 301.93.75.67.49.311.113.014.816.718.520.422.2VUV dose distribution (mJ/cm2 )(g)0 5 10 15 20 25 300.31.42.63.74.85.97.08.19.210.311.412.5VUV dose distribution (mJ/cm2 )(h)% % % % % %  111  in the case of 6 L/min flow the maximum retention time was 18.9-30.7 sec, in which 90% of the particles stayed less than 27.2 sec, indicating a narrow time distribution and optimum flow operation of the pilot system (Figure 5-10 c, d). Thus, it is important to investigate the residence time and •OH dose distributions within the VUV systems while validating and improving their performance. Assuming a UV dose requirement of 186 mJ/cm2 for the 4-log reduction in of persistent waterborne viruses (Hulsey et al., 2004), the utilized pilot VUV photoreactor which emits at both 185 nm and 254 nm, at 6 L/min flow provided a minimum UV dose that is 30% greater than the mentioned requirement (Figure 5-10 e, f). This indicates the added advantage of simultaneous water disinfection and effective micropollutant removal in a compact footprint through the utilized. 5.4.3 Effect of the water matrix (NOM and alkalinity)   It is known that NOM and alkalinity, ubiquitous in water sources, greatly affect the performance of the VUV process (Afzal et al., 2010; Imoberdorf and Mohseni, 2011a). Figures 5-11 (a), investigates the •OH scavenging impact of SRNOM concentration (no alkalinity) on degradation of the target pollutants at 10 L/min flow through the VUV pilot system. As shown, by doubling the SRNOM from 1 to 2 ppm, and from 2 to 4 ppm at 10 L/min, atrazine removal efficacy of the system reduced by 10% and 8%, respectively. However, the same variations in SRNOM concentrations resulted in a pronounced influence on the degradation of dioxane with 12% and 17% reductions. The less impact of NOM on degradation of atrazine degradation is mainly because changing NOM concentrations did not altered significantly the atrazine removal by direct UV-photolysis. In addition, given the 17% greater reaction affinity of dioxane for •OH-oxidation, it is expected that the higher NOM concentrations leave a larger influence on the dioxane degradation more than atrazine.  112     Figure 5-11 Model predictions and experimental data for the influence of (a) SRNOM (b) bicarbonate alkalinity, on atrazine and dioxane degradation 01020304050607080901001 2 4Pollutant degradation %Suwannee River NOM (mg/L) Atrazine (R1+R2)  1, 4-Dioxane (R1+R2) Atrazine (R1)  1, 4-Dioxane (R1) CFD model01020304050607080901000 25 50 100Pollutant degradation %Alkalinity as CaCO3 (mg/L) Atrazine (R1+R2)  1, 4-Dioxane (R1+R2) Atrazine (R1)  1, 4-Dioxane (R1) CFD model(a) (b)  113  Figure 5-11 (b) displays the target pollutants degradation in distilled water samples with 2 ppm SRNOM and different alkalinity concentrations (bicarbonate). About 6% and 7% reduction in the atrazine degradation efficacy was obtained by doubling the alkalinity concentration from 25 to 50, and 50 to 100 as mg/L CaCO3. However, this resulted in 10% and 15% drop in the dioxane removal. As can be seen in Figures 5-7 (a, b), the proposed CFD model was capable to predict well the experimental variation in pollutants degradation at different SRNOM and alkalinity concentrations. Overall, predicted degradation rates showed more sensitivity to variations of NOM concentrations than that of alkalinity. Besides, it is worth mentioning that sequential installation of two VUV photoreactors expanded significantly flexibility of the pilot system to accommodate variations in the background water constituents. This can be very important where pollutants and NOM concentrations are subjected to wide range of seasonal visibilities. 5.4.4 Model evaluation with raw water samples To assess the feasibility of VUV technology for natural water samples, degradation of atrazine and dioxane in surface water samples collected from Capilano Reservoir (CR), in British Columbia were studied. In addition, it is important to verify experimentally the capability of CFD model for predicting the degradation of target pollutants in real water samples. Figure 5-12 (a) compares CFD predicted and experimental degradation rates of 100 ppb of the target pollutants in CR water samples obtained by the VUV pilot system. Compared to synthetic water samples, UV transmittance (UVT) of raw water samples was about 7% less (e.g., UVT = 89%) which resulted in reduced UV-photolysis of atrazine (especially at lower flow rates). However, no evident change in the dioxane degradation was observed because UV-photolysis was irrelevant removal mechanism for dioxane.  114  The predicted degradation rates of target pollutants were in good agreement with the experimental results. However, when compared to synthetic waters CFD model showed to partly underestimate the target pollutants degradation specifically dioxane. This could be mainly due to different characteristics of NOM of CR water samples which resulted in less •OH reactivity of CR NOM (less k•OH) compared to the SRNOM.  In order to assess the cost-effectiveness of the utilized VUV pilot system for the remediation of surface water (CR) co-contaminated by atrazine and dioxane, electrical energy-per-order (EEO, kWh/m3/order) analysis was performed for the natural water samples spiked with 100 ppb of the target pollutants using Eq. (5-1).  ܧܧܱ	ሾkWh/݉ଷ/orderሿ ൌ ௅ܹܳ	ሾ݉ଷ ݄⁄ ሿ ൈ ܮ݋݃ሺܥ௜ ܥ௙ሻ⁄  (5-1)where WL is the input electrical energy of the lamp (kW), Q is the water volumetric flow rate (m3/h), Ci and Cf are concentrations of atrazine, in the inlet and effluent stream, respectively. Figure 5- 12 (b) shows the EEO values for the first VUV photoreactor as well as the overall system of two VUV photoreactors installed in series. As shown in Figure 5-8b, EEO values over the entire investigated flow rates were below 0.82 kWh/m3/order for both target pollutants. The maximum EEO of 0.81 kWh/m3/order belonged to 6 L/min flow with approximately 90% (1-log) removal of both dioxane and atrazine. Overall, increasing the flow rates decreases the EEO values, it drops significantly atrazine degradation efficacy of the pilot system. Comparing energy requirement of the VUV process for the first reactor to the performance of the two reactors installed in series, the later demonstrated superior energy-effectiveness particularly for the removal of dioxane.  115    Figure 5-12 (a) Comparison of experimental data and model predictions and (b) EEO analysis for atrazine and dioxane degradation in CR water samples 01020304050607080901006 12 18 24Pollutant degradation %Flow rate (L/min) Atrazine (R1+R2)  1, 4-Dioxane (R1+R2) Atrazine (R1)  1, 4-Dioxane (R1) CFD model0.00.10.20.30.40.50.60.70.80.91.06 12 18 24EEO (kWh/m3/order)Flow rate (L/min) Atrazine (R1)  1, 4-Dioxane (R1)  Atrazine (R1+R2)  1, 4-Dioxane (R1+R2)(a) (b)  116  Based on Figure 5-8b, assuming two parallel VUV reactors each operating at 6 L/min flow, EEO value was 0.81 kWh/m3/order. In other side, operation of two sequential installed reactors operating at 12 L/min flow resulted in EEO of 0.72 kWh/m3/order. Comparing the energy-effectiveness of VUV process for parallel and sequential installations of the two reactors, 12% electrical energy saving was obtained with sequential configuration the reactors. It is worth mentioning that comparison of the above scenarios were based on dioxane energy requirement due to the greater EEOs for dioxane removal than atrazine. 5.5 Chapter conclusions This work was to investigate the treatment feasibility a pilot-scale VUV process for effective remediation of synthetic and natural waters co-contaminated by dioxane and atrazine, model micropollutants, using CFD simulation. Thereby, approximately 90% (1-log) reduction of the target pollutants was attained in both synthetic and natural water samples under 6 L/min (Re ~ 4700) indicating the potential of VUV technology as a chemical-free and straightforward solution of micropollutants removal. However, significant reductions in the atrazine and dioxane degradation was observed with increasing the flow rates. In addition, the AOP performance of the VUV process showed strong correlation with flow characteristics (hydrodynamics) of the system as well as •OH scavenging constituents of the background water, particularly NOM concentration. Despite the greater tendency of dioxane for •OH-oxidation, overall removal performance of the VUV system for atrazine was more than dioxane degradation due to the contribution of UV-direct photolysis in atrazine degradation. Whilst, the utilized pilot system was composed of two sequentially installed VUV reactors, it was demonstrated that sequential operation of the reactors led to 12% energy saving of the VUV process when compared to the parallel installations of reactors. Further, sequential installation of VUV photoreactors in the proposed treatment system  117  led to greater flexibility of the system concerning seasonal variations in background water constitutions The developed CFD model showed to predict well degradation of the target pollutants, reduction of DOC, and H2O2 outlet concentration obtained via the pilot VUV system.    118  Chapter 6: Conclusions and recommendations for future research  6.1 Conclusions  This thesis was to develop a comprehensive mechanistic computational fluid dynamics (CFD) model for in-depth analysis and prediction of micropollutants removal by VUV oxidation process. The proposed CFD model involved simultaneous resolution of the local transfer equations of momentum, mass, and radiative energy for UV 254 nm and VUV 185 nm radiations, along with a complex kinetic scheme with more than 40 reactions. The original aspects of the thesis have been the development of a comprehensive CFD simulation tool for analyzing and validating the performance of laboratory or pilot-scale VUV reactors; an improved VUV process with comparable pollutant removal efficacy and energy costs to that of the widely used H2O2/UV AOP; and a new experimental approach for measuring the 185 nm and 254 nm emissions from VUV lamps. Some of the key conclusions and specific outcomes of the study are as follows:  Using an axisymmetric lab-scale VUV/UV reactor, a two-dimensional variant of the CFD model was evaluated experimentally for the degradation of p-CBA, a model pollutant, in (MilliQ) synthetic water samples under laminar flow regime. The model prediceted well the experimental degradation of p-CBA over the entire range of the investigated flow rates (e.g., 0.5-6.5 L/min). Utilizing the developed model, the estimated outlet concentration of p-CBA by-product obtained from the 185 nm direct-photolysis was less than 1 ppb indicating little concern over the concentration of byproducts from the 185 nm direct-photolysis pathway. In addition, it was shown that mass transfer in proximity of the quartz sleeve limits strongly the degradation rates of pollutant in VUV photoreactors.   119   Knowing the great importance of the lamp emissions for accurate modeling of the VUV/UV process, an experimental approach was proposed the emissions of an ozone-generating lamp at VUV 185 nm and UV 254 nm was measured at a wide range of lamp temperature.  Relying on the insights gained by CFD analysis (e.g., the key role of flow characteristics on the AOP performance of VUV reactors), seven different seniors were defined to explore impact of the main design parameters of VUV reactors on their removal efficacy. Different design scenarios were assessed in terms of VUV, UV, •OH dose distributions as well as electrical energy consumption. Thereby, an improved VUV/UV process with up to 72% electrical energy input saving for the effective degradation of a target pollutant (atrazine) was accomplished through incorporation of central baffles within the reactor geometry. The developed CFD model was validated experimentally for the removal of atrazine from MilliQ samples in a baffled lab-scale VUV reactor with turbulent flow regime. It was also shown that by using the RSM turbulence sub-model instead of Realizable k-ɛ equation, the relative error of model predictions decreased at the cost of a substantial increase in computation time (e.g., 4 h to 8 h). In addition, the capability of the model to predict the influence of a surrogate •OH scavenger, methanol, on degradation efficacy of the VUV process was verified experimentally.   Knowing the wide-scale application of H2O2/UV AOP for effective removal of a broad range of organic pollutants, competitiveness of the VUV/UV and H2O2/UV processes were compared for the removal of p-CBA in MilliQ water samples using a lab-scale reactor under laminar and turbulent flow regime. Whilst hydroxyl radical (•OH) oxidation was the dominant degradation pathway in both processes, H2O2/UV with 2 and 5 ppm H2O2  120  provided greater contaminant degradation efficiency than the VUV/UV process under laminar flow regime. Nonetheless, introduction of mixing and circulation zones to the VUV process resulted in its superior performance compared to the H2O2/UV AOP. Comparing the •OH concentration profile within the prototype photoreactor, VUV AOP showed significantly larger concentration of •OH with a sharper gradient than that of the H2O2/UV process. Based on the electrical energy-per-order (EEO) analysis, incorporation of circulation zones resulted in up to 50% improvement in the economy of VUV AOP while the corresponding enhancement in the 5-ppm H2O2/UV system was less than 5%. Furthermore, the extent to which methanol, a probe •OH scavenger, affected the removal efficacy of both VUV/UV and H2O2/UV oxidation processes were investigated. Thus, it was identified that methanol showed a greater •OH scavenging impact on the AOP performance of H2O2/UV than VUV/UV process. Consequently, using a combinatorial baffle design approach, greater removal performance of the VUV/UV process than that of H2O2/UV with 5-ppm peroxide was accomplished. In addition to accurate prediction of target pollutant degradation by the VUV/UV process, the proposed CFD model was capable to predict well the AOP performance of H2O2/UV AOP with different peroxide concentrations.  Given limited application of the two-dimensional CFD model for simulating the axisymmetric reactors, a three-dimensional version of the initially built model was adopted and validated experimentally using a pilot-scale VUV system operating at 6-30 L/min corresponding to Reynolds numbers of 4700-23500. The pilot system was comprised of two annular VUV reactors installed in series, each equipped to two perforated baffles. Using the 3D CFD model, degradation of 100 ppb atrazine and 100 ppb 1,4-dioxane (a  121  commonly used chemical solvent) in synthetic and surface water samples was investigated. Thereby, approximately 90% (1-log) reduction of the target pollutants was attained in both synthetic and natural water samples under 6 L/min (Re ~ 4700) indicating the potential of VUV technology as a chemical-free and straightforward solution for micropollutant removal. In addition, the predicted degradation rates of target pollutants were in good agreement with the experimental results in both synthetic and raw water samples.  Furthermore, to expand the applicability of the developed model, the kinetic sub-model was extended (e.g., more than 50 reactions) to incorporate the •OH scavenging effect of SRNOM and bicarbonate alkalinity on the removal efficacy of the pilot VUV/UV process. As a result, the 3D CFD model showed to predict well the degradation of target pollutant with different concentrations of SRNOM and alkalinity in distilled water samples. Apart from the accurate prediction of the target pollutants degradation in the utilized VUV system, the proposed CFD model predicted well the removal of up to about 10-20% DOC (from the initial 2-ppm concentration) depending upon the flow rates. Similarly, the maximum hydrogen peroxide (H2O2) concentration in the outlet of the pilot system, operating at 6 L/min, was 0.6 (±0.1) ppm that was consistent with model predictions. Whilst the utilized pilot system was composed of two sequentially installed VUV reactors, it was demonstrated that sequential operation of the reactors led to 12% energy saving of the VUV process as opposed to the parallel installations of reactors. 6.2 Future research  The future of VUV/UV process for water purification still requires elucidation of several key components. A list of future research topics that can contribute in furthering the scientific advancement of VUV process are listed briefly as follows:  122  6.2.1 Design improvement of VUV/UV reactors The importance of mixing and circulation zones on improving the pollutant degradation and energy consumption of the VUV process was shown by this thesis. In this sense, the greater the mixing, the higher the removal efficacy of the VUV oxidation process. Therefore, optimization of hydrodynamics of the VUV reactors plays a crucial role on their cost effectiveness. Using the 2D CFD model, this thesis compared a couple of different baffled reactors in terms of their AOP performance. However, it would be worthwhile to conduct a comprehensive design exploration study to provide an optimal design of baffled annular VUV reactors by means of the developed 3D CFD simulation in this thesis. In addition to using baffles, improving the hydrodynamics of VUV reactors can be attained via design modifications in the inlet and outlet segments, and/or through incorporation of multiple inlet and outlet streams. In this case, using the developed 3D CFD model by this thesis can significantly reduce the cost and time of experimental efforts leading to practical applications of VUV technology in water treatment. 6.2.2 Feasibility study of the multi-lamps VUV/UV reactor Currently, multi-lamps UV systems are widely used for water disinfection and H2O2/UV AOP applications. Given significant differences between H2O2/UV AOP and VUV process, it would be very important to investigate feasibility of reactors with multiple VUV lamps for effective degradation of organic pollutants. In addition, since configuration of the lamps can significantly affect hydrodynamics of the system, conducting design exploration studies that compares pollutant degradation and energy efficacy of multi-lamps reactors will be valuable in furthering the VUV treatment of contaminated waters, particularly at large-scale.  123  6.2.3 Impact of water temperature and turbidity on the AOP performance of the VUV process It is known that water temperature can significantly affect the 185 nm absorption coefficient of water, thereby influencing the removal efficacy and energy consumption of the VUV oxidation process. Thus, it is very beneficial to explore experimentally the impact of water temperature on the AOP performance of VUV process. In addition, there is a lack of knowledge on the correlation of water turbidity and 185 nm absorption coefficient of water. Similarly, the potential impact of water turbidity on pollutant removal and energy efficiency of VUV oxidation process requires elucidation. 6.2.4 Evaluating the potential of ozone formation by VUV/UV process One way to generate turbulence and mixing in the VUV process is purging air or other gasses (e.g., O2, N2) into the photoreactor. Through formation and rising of bubbles in VUV reactors significant turbulence could be generated resulting in improved mass transfer of target pollutants into the reacting zone near quartz sleeve, thereby enhancing the removal efficacy of the VUV process. The larger the bubbles, the greater motions created, and consequently bubbles will induce more mixing and turbulence within the VUV reactors. Thus, feasibility of dispersing (purging) air bubbles into the reactor for the enhanced removal of micropollutants by VUV process is worth investigating further. Particularly, effect of bubble size and distribution on pollutant degradation efficacy of the VUV reactors needs to be evaluated. In addition, the practicality of in-situ ozone generation from airflow passing by the VUV lamp envelope, and then its injected by diffusers into the VUV reactors requires assessment. In this sense, Figure 6-1 (a) displays a simplified schematic of the VUV reactor with internal ozone generation and purging system presented by Zoschke et al. (2012). Although the authors showed the improved removal efficacy of the VUV system purged  124  by internal ozone, they did not differentiate the critical role of mixing and turbulence caused by the air and ozone purging. As a result, investigating the potential impact of the purge ozone and air gas calls out further investigations. Similarly, a wide range of different design scenario for in-situ generation of ozone can be investigated. Figure 6-1 (b) shows a for more effective way to generate ozone by an excimer 172 nm VUV lamp, followed injection into the VUV reactor.            Figure 6-1 In-situ generation of ozone gas and its integration to the VUV reactors 6.2.5 Investigating the long-term performance of VUV lamps with natural and synthetic quartz envelops  The utilized VUV lamps in this thesis were low-pressure mercury amalgam lamps with synthetic quartz envelopes. Given the greater transparency of natural quartz for 185 nm VUV photons, it would be valuable to investigate feasibility of VUV lamps with natural quartz envelopes for improved emission of 185 nm radiation. Considering the greater potential of natural quartz for solarization-induced     AirHg VUV lampOzone  Air (N2)  Treated water    AirHg VUV lamp Ozone  Air (N2)  Treated water  VUV excimer lamp (b)  (a)   125  damage by VUV radiation, long-term monitoring of the 185 nm emissions from VUV lamps is critical research topic worth further investigation. 6.2.6 Exploring effect of inorganic ions on the AOP performance of VUV/UV process As shown in this current thesis, NOM and alkalinity can significantly slow down degradation of target pollutants. However, the influence of inorganic ions (e.g., calcium, potassium, sodium, iron, and sulfate) was not investigated. Thus, it is important to investigate the potential impact of inorganic ions in the removal efficacy of the VUV process under continuous flow operation. This is of high importance for removal of micropollutants from contaminated groundwater with high iron and manganese contents. 6.2.7 Impact of the VUV/UV process on different fractions of NOM  NOM is recognized as a precursor of disinfection by-products and a potential cause of membrane fouling. It can also deteriorate water biostability and increase bacterial regrowth and biofilm formation potential within the distribution systems. Since, NOM consists generally of hydrophobic, hydrophilic, and neutral fractions of varying chemical characteristics (e.g., UV254 absorbance) within a wide range of molecular weights, it is important to investigate the impact of VUV/UV treatment process on different NOM fractions. 6.2.8 Investigating nitrite formation in the VUV/UV treatment of drinking water The formation of nitrite ions becomes important in the wavelength range from 200 to 220 nm or when the emission at 185 nm is applied for water treatment. The US EPA has set a maximum contaminant level (MCL) for nitrite at 1 mg/L (measured as nitrogen). In the European drinking water standards, the nitrite limit is defined as being lower than 0.1 mg/L. Under 185 nm irradiation emission of the low-pressure VUV lamps, the photolysis of nitrate leads to the formation of nitrite  126  and oxygen (equation 6-1) (Zoschke et al., 2014). Thus, it will be valuable to investigate the potential of nitrite formation from the VUV/UV treatment of contaminated waters with high nitrate content. In particular, analysis of nitrite needs to be conducted at the lowest investigated flow rates to assure the generated nitrite is below the regulatory limit. ܱܰଷି ൅ ݄ݒ	ଵ଼ହ	௡௠ → ܱܰଶି ൅ ଵଶ	ܱଶ (6-1)6.2.9 Comparative investigation of Cl2/UV and Cl2/VUV/UV oxidation processes for removal of micropollutants  In recent years, chlorine/UV advanced oxidation process has received attention in water research communities. Thereby, it would be valuable to investigate whether introduction of VUV 185 nm radiation to the Cl2/UV process will have a synergetic effect for the removal of micropollutants. In this case, expanding the applicability of the proposed CFD model to incorporate the chlorine related reactions can significantly help the in-depth analysis of Cl2/VUV/UV oxidation process for the removal of micropollutants.  6.2.10 Study of the VUV/UV process for effective removal of micropollutants from wastewater  H2O2/UV AOP has demonstrated effectiveness for removal of micropollutants from treated domestic secondary wastewater (effluent); however, the degradation efficacy of this process can be negatively impacted by low UV transmittance of wastewater (James et al. 2014). In this case, feasibility of VUV process for the treatment of wastewater with low UV transmittance will require further investigation. In addition, it would be very valuable to validate experimentally the proposed 3-D CFD model of this thesis for simulating degradation of micropollutants in wastewater by means of VUV/UV process. Consequently, investigating cost analysis of VUV/UV and H2O2/UV  127  oxidation processes for the removal of contaminants of emerging concerns (e.g., pharmaceuticals and personal care products) from wastewater effluent would be valuable to conduct.  128  Bibliography Adams C. D., Scanlan P. A., Secrist N. D. Oxidation and biodegradability enhancement of 1,4-dioxane using hydrogen peroxide and ozone. Environmental Science and Technology 1994; 28(11):1812–8.  Afzal A., Oppenländer T, Bolton J. R., El-Din M. G. Anatoxin-a degradation by advanced oxidation processes: Vacuum-UV at 172 nm, photolysis using medium pressure UV and UV/H2O2. Water Research 2010; 44(1):278–86.   Alapi T., Dombi A. Comparative study of the UV and UV/VUV-induced photolysis of phenol in aqueous solution. Journal of Photochemistry and Photobiology A: Chemistry 2007; 188(2): 409–18.   Al-Momani F, Touraud E, Degorce-Dumas JR, Roussy J, Thomas O. Biodegradability enhancement of textile dyes and textile wastewater by VUV photolysis. Journal of Photochemistry and Photobiology A: Chemistry 2002; 153(1):191–7.   Alpert S. M., Knappe D. R. U., Ducoste J. J. Modeling the UV/hydrogen peroxide advanced oxidation process using computational fluid dynamics. Water Research 2010; 44(6):1797–808.  ANSYS® Academic Research, Release 14.5, Help System, ANSYS FLUENT Tutorial Guide, ANSYS, Inc. 2013.  Arany E., Szabó R. K., Apáti L., Alapi T., Ilisz I., Mazellier P. Degradation of naproxen by UV, VUV photolysis and their combination. Journal of Hazardous Materials 2013; 262:151–7.   Audenaert W., Vermeersch Y., Van H., Stijn W.H., Dejans P., Dumoulin A, Nopens I. Application of a mechanistic UV/hydrogen peroxide model at full-scale: sensitivity analysis, calibration and performance evaluation. Chemical Engineering Journal 2011; 171(1):113–26.   Azimi Y., Allen D. G., Farnood R. R. Enhancing disinfection by advanced oxidation under UV irradiation in polyphosphate-containing wastewater flocs. Water Research 2014; 54(0):179–87.   Azrague K., Pradines V., Bonnefille E., Claparols C., Maurette M., Benoit-Marquié F. Degradation of 2,4-dihydroxibenzoic acid by vacuum UV process in aqueous solution: kinetic, identification of intermediates and reaction pathway. Journal of Hazardous Materials 2012; 237-238:71–8.   129  Bagheri M., Imoberdorf G. E., Mohseni M. A CFD-based Study of the degradation of micropollutants induced by Vacuum-UV radiation: Proceedings International Congress on Ultraviolet Technologies (IUVA), Las Vegas, NV, September 22-26; 2013; 2013.  Bagheri M, Mohseni M. Computational fluid dynamics (CFD) modeling of VUV/UV photoreactors for water treatment. Chemical Engineering Journal 2014; 256:51–60.  Bagheri M., Mohseni M. A study of enhanced performance of VUV/UV process for the degradation of micropollutants from contaminated water. Journal of Hazardous Materials 2015; 294(0):1–8.  Barrett J., Mansell A. L. Ultra-violet absorption spectra of the molecules H2O, HDO and D2O. Nature 1960; 187:138.  Behera S. K., Kim H. W., Oh J., Park H. Occurrence and removal of antibiotics, hormones and several other pharmaceuticals in wastewater treatment plants of the largest industrial city of Korea. Science of the Total Environment 2011; 409(20):4351–60.  Benitez F. J., Acero J. L, Real F.J., Roldan G., Rodriguez E. Photolysis of model emerging contaminants in ultra-pure water: kinetics, by-products formation and degradation pathways. Water Research 2013; 47(2):870–80.  Bird R. B., Stewart W. E., Lightfoot E. N. Transport Phenomena: Wiley; 2007.  Bolton J. R., Bircger, K. T. G., Tolman, W. C. A. Figure-of merit for the technical development and application of advanced oxidation technologies for both electricand solar-derived systems. Pure and Applied Chemistry 2001; 73:627–37.  Buchanan W., Roddick F., Porter N. Formation of hazardous by-products resulting from the irradiation of natural organic matter: comparison between UV and VUV irradiation. Chemosphere 2006; 63(7):1130–41.  Buchanan W., Roddick F., Porter N., Drikas M. Fractionation of UV and VUV pretreated natural organic matter from drinking water. Environmental Science and Technology 2005; 39(12):4647–54.  Ho. C. K. Evaluation of reflection and refraction in simulations of ultraviolet disinfection using the discrete ordinates radiation model. Water Science and Technology 2009; 59(12):2421–8.    130  Carballa M., Omil F., Lema J. M., Llompart M., García-Jares C., Rodríguez I. Behavior of pharmaceuticals, cosmetics and hormones in a sewage treatment plant. Water Research 2004; 38(12):2918–26.  Cicek N., Londry K., Oleszkiewicz J. A., Wong D., Lee Y. Removal of selected natural and synthetic estrogenic compounds in a Canadian full-scale municipal wastewater treatment plant. Water Environment Research 2007; 79(7):795–800.   Clara M., Gans O., Windhofer G., Krenn U., Hartl W., Braun K. Occurrence of polycyclic musks in wastewater and receiving water bodies and fate during wastewater treatment. Chemosphere 2011; 82(8):1116–23.   Crapulli F., Santoro D., Sasges M. R., Ray A. K. Mechanistic modeling of vacuum UV advanced oxidation process in an annular photoreactor. Water Research 2014; 64:209–25.  Creasey D. J., Heard D. E., Lee J. D. Absorption cross-section measurements of water vapour and oxygen at 185 nm. Geophysical Research Letters 2000; 27(11):1651–4.  Cussler E. L. Diffusion: Mass Transfer in Fluid Systems: Cambridge University Press; 2009.  Daimon M., Masumura A. Measurement of the refractive index of distilled water from the near-infrared region to the ultraviolet region. Applied Optics 2007; 46(18):3811–20.  Dobrović S., Juretić H., Ružinski N. Photodegradation of natural organic matter in water with UV irradiation at 185 and 254 nm: importance of hydrodynamic conditions on the decomposition rate. Separation Science and Technology 2007; 42(7):1421–32.  Duca C., Imoberdorf G., Mohseni M. Novel collimated beam setup to study the kinetics of VUV‐induced reactions. Photochemistry and photobiology 2013; 90(1):238–40.  Duran J. E., Mohseni M., Taghipour F. Modeling of annular reactors with surface reaction using computational fluid dynamics (CFD). Chemical Engineering Science 2010a; 65(3):1201–11.   Duran J. E, Mohseni M., Taghipour F. Computational fluid dynamics modeling of immobilized photocatalytic reactors for water treatment. AIChE Journal 2011; 57(7):1860–72.  Duran J. E., Taghipour F., Mohseni M. CFD modeling of mass transfer in annular reactors. International Journal of Heat and Mass Transfer 2009; 52(23):5390–401.    131  Duran J. E., Taghipour F., Mohseni M. Irradiance modeling in annular photoreactors using the finite-volume method. Journal of Photochemistry and Photobiology A: Chemistry 2010b; 215(1):81–9.  Elyasi S., Taghipour F. Simulation of UV photoreactor for degradation of chemical contaminants: model development and evaluation. Environmental Science and Technology 2010; 44(6):2056–63.  Ericson H., Thorsén G., Kumblad L. Physiological effects of diclofenac, ibuprofen and propranolol on Baltic Sea blue mussels. Aquatic Toxicology 2010; 99(2):223–31.  Giri R. R., Ozaki H., Takayanagi Y., Taniguchi S., Takanami R. Efficacy of ultraviolet radiation and hydrogen peroxide oxidation to eliminate large number of pharmaceutical compounds in mixed solution. International Journal of Environmental Science and Technology 2011; 8(1):19–30.   Glaze W. H., Kang J., Chapin D. H. The chemistry of water treatment processes involving ozone, hydrogen peroxide and ultraviolet radiation. Ozone: Science and Engineering 1987; 9(4):335–52.  Gonzalez M. G., Oliveros E., Wörner M., Braun A. M. Vacuum-ultraviolet photolysis of aqueous reaction systems. Journal of Photochemistry and Photobiology C: Photochemistry Reviews 2004; 5(3):225–46.   Gu X., Lu S., Qiu Z., Sui Q., Banks C. J., Imai T. Photodegradation performance of 1, 1, 1-trichloroethane in aqueous solution: In the presence and absence of persulfate. Chemical Engineering Journal 2013; 215:29–35.  Gunten U. V. Ozonation of drinking water: Part II. Disinfection and by-product formation in presence of bromide, iodide or chlorine. Water Research 2003; 37(7):1469–87.  Halden R. U., editor. Contaminants of Emerging Concern in the Environment: Ecological and Human Health Considerations. Washington, DC: American Chemical Society; 2010.  Halmann M. Temperature dependence of absorption of liquid water in the far-ultraviolet region. Journal of Physical Chemistry 1966; 70:580–81.  Han W., Zhang P., Zhu W., Yin J., Li L. Photocatalysis of p-chlorobenzoic acid in aqueous solution under irradiation of 254 nm and 185 nm UV light. Water Research 2004; 38(19):4197–203.   132  Huang H., Leung D. Y. C., Kwong P. C. W., Xiong J., Zhang L. Enhanced photocatalytic degradation of methylene blue under vacuum ultraviolet irradiation. Catalysis Today 2013; 201:189–94.   IJpelaar G. F., Harmsen D. J., Beerendonk E. F., Leerdam, V., Robin C., Metz D. H., Knol A. H. Comparison of low pressure and medium pressure UV Lamps for UV/H2O2 treatment of natural waters containing micropollutants. Ozone: Science and Engineering 2010; 32(5):329–37.  Imoberdorf G., Mohseni M. Degradation of natural organic matter in surface water using vacuum-UV irradiation. Journal of Hazardous Materials 2011a; 186(1):240–6.   Imoberdorf G., Mohseni M. Modeling and experimental evaluation of Vacuum-UV photoreactors for water treatment. Chemical Engineering Science 2011b; 66(6):1159–67.   Imoberdorf G., Mohseni M. Kinetic study and modeling of the Vacuum-UV photoinduced degradation of 2,4-D. Chemical Engineering Journal 2012;187:114–22.   Imoberdorf G. E., Mohseni M. Experimental study of the degradation of 2, 4-D induced by Vacuum-UV radiation. Water Science and Technology 2011c; 63(7):1427.   Ivančev-Tumbas I. The fate and importance of organics in drinking water treatment: a review. Environmental Science and Pollution Research 2014; 21(20):11794–810.   James C. P., Germain E., Judd S. Micropollutant removal by advanced oxidation of micro-filtered secondary effluent for water reuse. Separation and Purification Technology. 2014; 127:77–83   Jochimsen E. M., Carmichael W., W., An J. S., Cardo D. M., Cookson S. T., Holmes C. E. Liver failure and death after exposure to microcystins at a hemodialysis center in Brazil. New England Journal of Medicine 1998; 338(13):873–8.  Jović M., Manojlović D., Stanković D., Dojčinović B., Obradović B., Gašić U. Degradation of triketone herbicides, mesotrione and sulcotrione, using advanced oxidation processes. Journal of Hazardous Materials 2013; 260:1092–9.   Keen O. S., Dotson A. D., Linden K. G. Evaluation of hydrogen peroxide chemical quenching agents following an advanced oxidation process. Journal of Environmental Engineering 2013; 139(1):137–40.  Kim I., Tanaka H. Photodegradation characteristics of PPCPs in water with UV treatment. Environment international 2009; 35(5):793–802.   133   Klassen N. V., Marchington D., McGowan H. C. H2O2 determination by the I3- method and by KMnO4 titration. Analytical Chemistry. 1994; 66(18):2921–5.  Kruithof J. C., Kamp P. C., Martijn B. J. UV/H2O2 Treatment: a practical solution for organic contaminant control and primary disinfection. Ozone: Science and Engineering 2007; 29(4):273–80.  Kutschera K., Börnick H., Worch E. Photoinitiated oxidation of geosmin and 2-methylisoborneol by irradiation with 254nm and 185nm UV light. Water Research 2009; 43(8):2224–32.  Lawal O., Dussert B., Howarth C., Platzer K., Sasges M., Muller J. Proposed method for measurement of the output of monochromatic (254 nm) low pressure UV lamps. IUVA News 2008; 10:14–7.  Light sources online product manual. "Low Pressure Mercury Lamps" product manual Available online at: www.light-sources.com/germicidal-uvc-lamps/products/low-pressure-mercury-lamps, 2013.  Linden K. G., Mohseni M. Advanced Oxidation Processes: Applications in Drinking Water Treatment. In: Comprehensive Water Quality and Purification: Elsevier; 2014. p. 148–172.  Luo Y., Guo W., Ngo H. H., Nghiem L. D., Hai F. I., Zhang J. A review on the occurrence of micropollutants in the aquatic environment and their fate and removal during wastewater treatment. Science of the Total Environment 2014; 473-474:619–41.  Ma C, Hong G, Chen H, Hang N, Shen Y. Photooxidation contribution study on the decomposition of azo dyes in aqueous solutions by VUV-Based AOPs. International Journal of Photoenergy 2011; 10:1–8.  Malley J. P. UV in water treatment: issues for the next decade. IUVA News 2010(12):18–25.  Martijn B. J., Fuller A. L., Malley J. P., Kruithof J. C. Impact of IX-UF pretreatment on the feasibility of UV/H2O2 treatment for degradation of NDMA and 1,4-Dioxane. Ozone: Science and Engineering 2010; 32(6):383–90.  Matilainen A., Sillanpä M. Removal of natural organic matter from drinking water by advanced oxidation processes. Chemosphere 2010; 80(4):351–65.    134  Matsushita T., Hirai S., Ishikawa T., Matsui Y., Shirasaki N. Decomposition of 1,4-dioxane by vacuum ultraviolet irradiation: Study of economic feasibility and by-product formation. Process Safety and Environmental Protection 2015; 94:528–41.  Miltner R. J. Treatment of seasonal pesticides in surface waters. Cincinnati, Ohio: Water Engineering Research Laboratory, Office of Research and Development, U.S. Environmental Protection Agency; 1988.  Mouamfon M., Li W., Lu S., Chen N., Qiu Z., Lin K. photodegradation of sulfamethoxazole applying uv-and vuv-based processes. Water, Air, and Soil Pollution 2011; 218(1-4):265–74.   Moussavi G., Hossaini H., Jafari S. J., Farokhi M. Comparing the efficacy of UVC, UVC/ZnO and VUV processes for oxidation of organophosphate pesticides in water. Journal of Photochemistry and Photobiology A: Chemistry 2014; 290:86–93.  Nic M., Jirat J., Kosata B., Hovorka L., Znamenacek J. IUPAC gold book. Research Triangle Park, NC: International Union of Pure and Applied Chemistry; 2005.  Nick K., Schöler H. F., Mark G., Söylemez T., Akhlaq M. S., Schuchmann H. P. Degradation of some triazine herbicides by UV radiation such as used in the UV disinfection of drinking water. Aqua-Journal of Water Supply: Research and Technology 1992;41(2):82–7.   Oaks JL, Gilbert M, Virani MZ, Watson RT, Meteyer CU, Rideout BA et al. Diclofenac residues as the cause of vulture population decline in Pakistan. Nature 2004;427(6975):630–3.  Oppenländer T. Photochemical purification of water and air. Weinheim, Cambridge, Wiley-VCH; 2003.  Oppenländer T., Gliese S. Mineralization of organic micropollutants (homologous alcohols and phenols) in water by Vacuum-UV oxidation (H2O-VUV) with an incoherent xenon-excimer lamp at 172 nm. Chemosphere 2000; 40(1):15–21.   Oppenländer T., Walddörfer C., Burgbacher J., Kiermeier M., Lachner K., Weinschrott H. Improved vacuum-UV (VUV)-initiated photomineralization of organic compounds in water with a xenon excimer flow-through photoreactor (lamp, 172 nm) containing an axially centered ceramic oxygenator. Chemosphere 2005; 60(3):302–9.   Price W. C., Harris P. V., Beaven G. H., Johnson E. A. Ultra-violet absorption spectra of the molecules H2O, HDO and D2O. Nature 1960; 188:45–6.   135  Quici N., Litter M. I., Braun A. M., Oliveros E. Vacuum-UV-photolysis of aqueous solutions of citric and gallic acids. Journal of Photochemistry and Photobiology A: Chemistry 2008; 197(2):306–12.   Ranade V. V. Computational flow modeling for chemical reactor engineering: Academic Press; 2002.  Ratpukdi T, Siripattanakul S, Khan E. Mineralization and biodegradability enhancement of natural organic matter by ozone–VUV in comparison with ozone, VUV, ozone–UV, and UV: Effects of pH and ozone dose. Water Research 2010; 44(11):3531–43.   Rauen W. B., Lin B., Falconer R. A., Teixeira E. C. CFD and experimental model studies for water disinfection tanks with low Reynolds number flows. Chemical Engineering Journal 2008; 137(3):550–60.   Richardson S. D., Ternes T. A. Water Analysis: emerging contaminants and current issues. Analytical Chemistry 2011; 83(12):4614–48.  Richardson S. D., Ternes T. A. Water analysis: emerging contaminants and current issues. Analytical Chemistry 2014; 86(6):2813–48.  Robl S., Wörner M., Maier D., Braun A. M. Formation of hydrogen peroxide by VUV-photolysis of water and aqueous solutions with methanol. Photochemical and Photobiological Sciences 2012; 11(6):1041–50.   Rosenfeldt E. J., Linden K.G. The ROH, UV concept to characterize and the model UV/H2O2 process in natural waters. Environmental Science and Technology 2007;41(7):2548–53.  Ross A. B., Mallard W. G., Holman W. P., Buxton G. V., Huie R. E., Neta P., editors. NDRL-NIST Solution kinetic database - Ver. 2: NIST Reference data, Gaithersburg, MD; 1994.  Sanches S, Barreto Crespo, Maria T, Pereira VJ. Drinking water treatment of priority pesticides using low-pressure UV photolysis and advanced oxidation processes. Water Research 2010; 44(6):1809–18.   Santoro D., Raisee M., Moghaddami M., Ducoste J., Sasges M., Liberti L. Modeling hydroxyl radical distribution and trialkyl phosphates oxidation in UV−H2O2 photoreactors using computational fluid dynamics. Environmental Science and Technology 2010; 44(16):6233–41.  Sarathy S. R., Mohseni M. An overview of UV-based advanced oxidation processes IUVA News 2006; 7:16–27.  136   Sasges M., Robinson J., Daynouri F. Ultraviolet lamp output measurement: a concise derivation of the Keitz equation. Ozone: Science and Engineering 2012; 34(4):306–9.  Sasges M., Pol A., Voronov, A., Robinson, J. A standard method for quantifying the output of UV lamps, Proceedings International Congress on Ultraviolet Technologies (IUVA), Los Angeles, CA, August 26-29; 2007.  Schalk S., Adam V., Arnold E., Brieden K., Voronov A., Witzke H. D. UV-lamps for disinfection and advanced oxidation–Lamp types, technologies and applications. IUVA News 2006; 8(1):32–7.   Schuchmann H., Ritter A., Sonntag C. V. Photolysis at λ 185 nm of hexamethyldisiloxane in the liquid phase. Journal of Organometallic Chemistry 1978; 148(3):213–23.  Shannon M. A., Bohn P. W., Elimelech M., Georgiadis J. G., Mariñas B. J., Mayes A. M. Science and technology for water purification in the coming decades. Nature 2008;452(7185):301–10.  Shu Z., Bolton J. R., Belosevic M., Din, E. l., Gamal M. Photodegradation of emerging micropollutants using the medium-pressure UV/H2O2 advanced oxidation process. Water Research 2013; 47(8):2881–89.  Sozzi D. A., Taghipour F. Computational and experimental study of annular photo-reactor hydrodynamics. International Journal of Heat and Fluid Flow 2006a; 27(6):1043–53.  Sozzi D. A., Taghipour F. UV reactor performance modeling by Eulerian and Lagrangian methods. Environmental science and technology 2006b; 40(5):1609–15.  Stackelberg P. E., Gibs J., Furlong E. T., Meyer M. T., Zaugg S. D., Lippincott R. L. Efficiency of conventional drinking-water-treatment processes in removal of pharmaceuticals and other organic compounds. Sci. Total Environ. 2007; 377(2-3):255–72.  Steinle-Darling E., Litwiller E., Reinhard M. Effects of sorption on the rejection of trace organic contaminants during nanofiltration. Environmental Science and Technology 2010; 44(7):2592–8.  Stickney J. A., Sager S. L., Clarkson J. R., Smith L. A., Locey B. J., Bock M. J. An updated evaluation of the carcinogenic potential of 1,4-dioxane. Regulatory Toxicology and Pharmacology 2003; 38(2):183–95.   137  Sudhakaran S., Lattemann S., Amy G. L. Appropriate drinking water treatment processes for organic micropollutants removal based on experimental and model studies - a multi-criteria analysis study. Science of the Total Environment 2013; 442:478–88.  Taghipour F., Sozzi A. Modeling and design of ultraviolet reactors for disinfection by-product precursor removal. Desalination 2005; 176(1):71–80.   Tasaki T., Wada T., Fujimoto K., Kai S., Ohe K., Oshima T. Degradation of methyl orange using short-wavelength UV irradiation with oxygen microbubbles. Journal of Hazardous Materials 2009; 162(2):1103–10.   Thomson J., Roddick F., Drikas M. Vacuum ultraviolet irradiation for natural organic matter removal. Journal of Water Supply: Research and Technology - Aqua 2004; 53:193–206.   Verlicchi P., Al Aukidy M., Zambello E. Occurrence of pharmaceutical compounds in urban wastewater: removal, mass load and environmental risk after a secondary treatment-a review. Science of the Total Environment 2012; 429:123–55.  Wackett L., Sadowsky M., Martinez B., Shapir N. Biodegradation of atrazine and related s-triazine compounds: from enzymes to field studies. Applied Microbiology and Biotechnology 2002; 58(1):39–45.  Wang J., Yang C., Wang C., Han W., Zhu W. Photolytic and photocatalytic degradation of micro pollutants in a tubular reactor and the reaction kinetic models. Separation and Purification Technology 2014; 122:105–11.  Watson S. Aquatic taste and odor: a primary signal of drinking-water integrity. Journal of Toxicology and Environmental Health, Part A 2004; 67(20):1779–95.  Weeks J. L., Meaburn G., Gordon S. Absorption coefficients of liquid water and aqueous solutions in the far ultraviolet. Radiation Research 1963; 19(3):559–67.  Westerhoff P., Mezyk S. P., Cooper W. J., Minakata D. Electron pulse radiolysis determination of hydroxyl radical rate constants with Suwannee River fulvic acid and other dissolved organic matter isolates. Environmental Science and Technology 2007; 41(13):4640–6.  White GC. White's handbook of chlorination and alternative disinfectants. Fifth ed. Hoboken N.J: Wiley; 2010.  Wols B. Computational fluid dynamics in drinking water treatment. London, New York: IWA Publishing; 2012.  138   Wols B. A., Harmsen D. J. H., Wanders-Dijk J., Beerendonk E. F., Hofman-Caris C. H. M. Degradation of pharmaceuticals in UV (LP)/H2O2 reactors simulated by means of kinetic modeling and computational fluid dynamics (CFD). Water Research 2015; 75:11–24.  Wols B. A., Hofman-Caris C. H. Review of photochemical reaction constants of organic micropollutants required for UV advanced oxidation processes in water. Water Research 2012; 46(9):2815–27.   Yang W., Zhou H., Cicek N. Treatment of Organic Micropollutants in Water and Wastewater by UV-based Processes: A Literature Review. Critical Reviews in Environmental Science and Technology 2013; 44(13):1443–76.   Yuan F., Hu C., Hu X., Qu J., Yang M. Degradation of selected pharmaceuticals in aqueous solution with UV and UV/H2O2. Water Research 2009; 43(6):1766–74.   Zoschke K., Börnick H., Worch E. Vacuum-UV radiation at 185 nm in water treatment-a review. Water Research 2014; 52:131–45.  Zoschke K., Dietrich N., Börnick H., Worch E. UV-based advanced oxidation processes for the treatment of odour compounds: Efficiency and by-product formation. Water Research 2012; 46(16):5365–73.   139  Appendices  -  User defined numerical codes (C++) A.1 Net reaction rate calculation  /***************************************************************************************/  This code is a simplified sample code written in C++ programing language for calculation of net reaction rates involved 44 reactions.  Mehdi Bagheri, 2015 PhD Thesis University of British Columbia, Vancouver, British Columbia, Canada     /***************************************************************************************/  #include "udf.h"   DEFINE_NET_REACTION_RATE(net_rxn,c,t,particle,pressure,temp,yi,rr,jac) {  /* Defining the reaction rate constants */  real k1=2.5e7,k2=2.8e10,k3=3.9e7,k4=1.8e10,k5=1.3e10,k6=9.4e7; real k7=3.8e9,k8=4.2e9,k9=7e9,k10=3.4e10,k11=3e10,k12=5.5e9; real k13=5.2e10,k14=3.3e2,k15=9e10,k16=5e8,k17=5.7e4,k18=1.1e5; real k19=1.2e10,k20=8e5,k21=5e10,k22=8.3e5,k23=9.7e7; real k24=1.44e11,k25=2.6e‐5,k26=1.1e10,k27=6.6e9,k28=1e3,k29=1.3e10; real k30=6e9,k31=3.5e9,k32=2.7e7,k33=6.8e9,k34=5e7,k35=2e10,k36=2.7e7; real k37=4e8,k38=5e9,k39=5e9,k40=1e9,k41=5e9,k42=9.7e8,k43=10e8,k44=2.95e8;   int i=0; Material *m = THREAD_MATERIAL(t); Material *sp_k;  int k;   nspe = MIXTURE_NSPECIES(m); real mw[nspe];  mixture_species_loop(m, sp_k, k) {   real mw[k]= MATERIAL_PROP(sp_k,PROP_mwi); }  140   /* YI is mass fraction,  C_R(c,t) is the density of the mixture */  for(i;i<nspe;i++)  { C[i]=C_R(c,t)*yi[i]/mw[i]; }  /* i=0=OH‐, i=1=O2‐, i=2=O‐, i=3=O‐, i=4=O2‐, i=5=O3‐, i=6=HO2‐; i=7=HO3, i=8=OH, i=9=H, i=10=H+,  i=11=e‐, i=12=O2, i=13=pCBA‐OH‐OH, i=14=pCBA185‐OH,  i=15=pCBA254‐OH,  i=16=HCOOH, i=17=pCBA, i=18=pCBA‐OH, i=19=pCBA185, i=20=pCBA254, i=21=CH3OH, i=22=CH2O, i=23=H2O2, i=24=HCO2, i=25=H2O, */       rr[0]=‐k1*C[0]*C[9]‐k5*C[0]*C[8]+k6*C[2]+k10*C[9]*C[11]+k11*C[8]*C[11]‐k16*C[23]*C[0]+k17*C[25]*C[6]+k23*C[5]*C[1]‐  k24*C[0]*C[10]+k25*C[25]+k26*C[8]*C[1]+k28*C[11]+k29*C[11]*C[23]+2*k30*C[11]*C[11]+k31* C[11]*C[6]+k37*C[2]*C[6];      rr[1]=k4*C[11]*C[12]+k20*C[5]‐k21*C[1]*C[10]‐k23*C[1]*C[5]‐k26*C[1]*C[8]+k33*C[6]*C[8]‐k36*C[1]*C[9]+k37*C[2]*C[6];  rr[2]=k5*C[0]*C[8]‐k6*C[2]‐k7*C[2]*C[12]+k31*C[11]*C[6]‐k37*C[2]*C[6];    rr[3]=k7*C[12]*C[2]‐k13*C[3]*C[10]+k14*C[7]‐k15*C[3]*C[10];  rr[4]=‐k3*C[8]*C[4]+k10*C[9]*C[11]+2*k12*C[9]*C[9]+k30*C[11]*C[11];  rr[5]=k19*C[9]*C[12]‐k20*C[5]+k21*C[10]*C[1]‐2*k22*C[5]*C[5]‐k23*C[2]*C[5]+k32*C[23]*C[8]‐k35*C[5]*C[9];  rr[6]=k16*C[23]*C[0]‐k17*C[6]*C[25]‐k31*C[6]*C[11]‐k33*C[6]*C[8]+k36*C[1]*C[9]‐k37*C[6]*C[2];  rr[7]=k13*C[3]*C[10]‐k14*C[7]‐k18*C[7];  rr[8]=‐k3*C[8]*C[4]‐k5*C[8]*C[0]+k6*C[2]‐2*k8*C[8]*C[8]‐k9*C[8]*C[9]‐k11*C[11]*C[8]+k15*C[3]*C[10]‐k26*C[8]*C[1]‐k27*C[8]*C[5] +k29*C[11]*C[23]‐k32*C[23]*C[8]‐k33*C[8]*C[6]+k34*C[9]*C[23]‐k38*C[17]*C[8]‐k39*C[18]*C[8]‐k40*C[19]*C[8]‐k41*C[20]*C[8]‐k42*C[21]*C[8]‐k43*C[8]*C[22]‐k44*C[8]*C[24]+(0.001*0.33*2.303*167*Qrad185)+(0.001*0.045*2.303*167*Qrad185) +(2*(0.001*0.5*2.303*C[23]*(29700*Qrad185)))+(2*(0.001*0.5*2.303*C[23]*1960*Qrad254);  rr[9]=‐k1*C[0]*C[9]+k2*C[11]*C[10]+k3*C[8]*C[4]‐k9*C[8]*C[9]‐k10*C[11]*C[9]‐2*k12*C[9]*C[9]‐k19*C[12]*C[9]+k28*C[11] ‐k34*C[23]*C[9]‐k35*C[5]*C[9]‐k36*C[9]*C[1]+(0.001*0.33*2.303*167*Qrad185);  141         rr[10]=‐k2*C[10]*C[11]‐k13*C[10]*C[3]+k14*C[7]‐k15*C[10]*C[3]+k20*C[5]‐k21*C[1]*C[10]‐k24*C[10]*C[0]+k25*C[25]+(0.001*0.045*2.303*167*Qrad185);  rr[11]=k1*C[9]*C[0]‐k2*C[10]*C[11]‐k10*C[11]*C[9]‐k11*C[11]*C[8]‐k28*C[11]‐k29*C[11]*C[23]‐2*k30*C[11]*C[11]‐k31*C[11]*C[6]‐k1*C[11]*C[0] +(0.001*0.045*2.303*167*Qrad185;  rr[12]=‐k4*C[12]*C[11]‐k7*C[2]*C[12]+k15*C[10]*C[3]+k18*C[7]‐k19*C[12]*C[9]+k22*C[5]*C[5]+k23*C[5]*C[1]+k26*C[8]*C[1]+k27*C[8]*C[5];  rr[13]=k38*C[8]*C[18];  rr[14]=k40*C[8]*C[19];;  rr[15]=k41*C[8]*C[20];  rr[16]=k43*C[8]*C[22]‐k44*C[8]*C[16];  rr[17]=‐k38*C[8]*C[17]‐(0.001*0.013*2.303*258100*C[17]*Qrad254)‐(0.001*0.013*2.303*258100*C[17]*Qrad185);  rr[18]=k38*C[8]*C[17];  rr[19]=0.001*0.013*2.303*258100*C[17]*Qrad254;  rr[20]=0.001*0.013*2.303*258100*C[17]*Qrad185;  rr[21]=k42*C[8]*C[21];  rr[22]=k42*C[8]*C[21]‐k43*C[8]*C[22];  rr[23]=k8*C[8]*C[8]‐k16*C[23]*C[0]+k17*C[25]*C[6]+k22*C[5]*C[5]+k23*C[5]*C[1]‐k29*C[11]*C[23]‐k32*C[23]*C[8]‐k34*C[9]*C[23]‐(0.001*0.5*2.303*C[23]*29700*Qrad185) ‐(0.001*0.5*2.303*C[23]*1960*Qrad254);  rr[24]=k44*C[8]*C[16];  rr[25]=k1*C[9]*C[0]+k3*C[8]*C[4]+k5*C[8]*C[0]‐k6*C[2]+k9*C[8]*C[9]+k10*C[11]*C[9]+k16*C[23]*C[0]‐k17*C[25]*C[6]‐k23*C[1]*C[5]+k24*C0]*C[10]‐k25*C[25]+k27*C[5]*C[5] ‐k28*C[11]‐2*k30*C[11]*C[11]+k32*C[23]*C[8]+k33*C[8]*C[6]+k34*C[23]*C[9]‐(0.001*0.33*2.303*167*Qrad185)‐(0.001*0.045*2.303*167*Qrad185); }  /* for components in the equilibrium equations CO32m, HO2m, HO2rad  CO32m = (pow (10,‐10.3)*HCO3m/pow (10,‐pH));  142  HO2m = (pow (10,‐11.6)*H2O2/pow (10,‐pH)); HO2rad = (pow (10,‐pH)*O2radm/pow (10,‐4.8));  /*     143  A.2 •OH dose delivery calculation  /***************************************************************************************/  This UDF code is for computing the •OH dosage along a particle trajectory           Mehdi Bagheri, 2015 PhD Thesis University of British Columbia, Vancouver, British Columbia, Canada     /***************************************************************************************/  #include "udf.h" #include "dpm.h" #include "prop.h"  static real oh_0;   /* oh_0 is the initial molar concentration of OH species */  static real x0, y00;  DEFINE_DPM_SCALAR_UPDATE(oh_dosage, cell, thread, initialize, p) {   cphase_state_t *c = &(p‐>cphase);   if (initialize)     {       p‐>user[0] = 0.;       oh_0 = P_PROD_SPECIES_INDEX(p);       x0=p‐>state.pos[0];       y00=p‐>state.pos[1];      }    else     {       p‐>user[0] += P_DT(p) * .5 * (oh_0 +C8);       oh_0 = C8;     } }  DEFINE_DPM_OUTPUT(oh_output, header, fp, p, thread, plane) {  char name[100];   if (header)   {     if (NNULLP(thread))      par_fprintf_head(fp,"(%s %d)\n",thread‐>head‐>dpm_summary.sort_file_name,11);     else      par_fprintf_head(fp,"(%s %d)\n",plane‐>sort_file_name,11);  144                       par_fprintf_head(fp,"(%10s %10s  %10s  %10s  %10s  %10s  %10s"                   " %10s  %10s  %10s %10s %s)\n",                     "X0","Y0",                     "X","Y","U","V","diameter","T","mass‐flow",                     "time","oh‐Dosage","name");         } else {  sprintf(name,"%s:%d",p‐>injection‐>name,p‐>part_id); #if PARALLEL       par_fprintf(fp,                "%d %d ((%10.6g  %10.6g  %10.6g  %10.6g  %10.6g  %10.6g  "                "%10.6g  %10.6g  %10.6g  %10.6g  %10.6g) %s)\n",                p‐>injection‐>try_id,p‐>part_id,                x0,y00,                p‐>state.pos[0], p‐>state.pos[1],                p‐>state.V[0], p‐>state.V[1],                p‐>state.diam, p‐>state.temp, p‐>flow_rate, p‐>state.time,                p‐>user[0], name); #else       par_fprintf(fp,                "((%10.6g  %10.6g  %10.6g  %10.6g  %10.6g  %10.6g  "                "%10.6g  %10.6g  %10.6g  %10.6g %10.6g) %s)\n",                x0,y00,z0,                p‐>state.pos[0], p‐>state.pos[1],                p‐>state.V[0], p‐>state.V[1],                p‐>state.diam, p‐>state.temp, p‐>flow_rate, p‐>state.time,                p‐>user[0], name); #endif  }  }     145  -  Experimental error and analysis In each chapter, experimental results are either described in graphical form, or written form (e.g., 0.12 ± 0.02 mg/L), where error bars represent the standard deviation. Significant figures in all experimental results indicate the precision of experimental data. A population is all values of a particular entity. Standard deviation (s) and rejection of outliers were based on the following calculations, ܵܽ݉݌݈݁	݉݁ܽ݊ ൌ ̅ݔ ൌ 1݊෍ݔ௜௡௜ୀଵ (B-1)ݏ ൌ ඨ∑ ሺݔ௜ െ ̅ݔሻଶ௡௜ୀଵ݊ െ 1  (B-2)where n is the number of replicates or observations made, x is the ith value, and ̅ݔ is the mean of the data set. Using T-test (t), the confidence interval form as given below േݐ ൌ ሺ̅ݔ െ ߤሻ√݊ݏ  (B-3)where µ is the population mean. The t value is a constant that depends on the risk we wish to take and the number of degrees of freedom (n-1) on which the standard deviation, s, is based. The constant can be obtained from the Student’s t table. It should be noted that the value of t is calculated based on 95% confidence interval, which states that there is a probability of 95% of the mean falling in between. Similarly, Q-test with 95% confidence was used for removal of outliers from the collected experimental data. To compare statistically CFD model predictions versus experimental values, absolute relative deviation (ARD) and average absolute relative deviation (AARD) were reported based on the following equations.      146  ܣܴܦ ൌ ∑ หݕ௜,௣௥௘ௗ௜௖௧௘ௗ െ ݕ௜,௘௫௣௘௥௜௠௘௡௧௔௟ห௡௜ୀଵ݊  (B-4)ܣܣܴܦ ൌቆ∑ หݕ௜,௣௥௘ௗ௜௖௧௘ௗ െ ݕ௜,௘௫௣௘௥௜௠௘௡௧௔௟ห௡௜ୀଵ ݕ௜,௘௫௣௘௥௜௠௘௡௧௔௟ ቇ݊  (B-5)where y, can be degrade rate, or reactor conversion at certain flow rates and n is number of investigated flow rates.    

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.24.1-0228655/manifest

Comment

Related Items