|UFDC Home||myUFDC Home | Help|
1 PHYSICAL AND CFD MODELS OF PM SEPARATION AND SCOUR IN HYDRODYNAMIC UNIT OPERATION S By HWAN CHUL CHO A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS F OR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2012
2 201 2 Hwan Chul Cho
3 To my parents who always stand behind me, supporting me, and believing there is nothing that I cannot achieve and to everyone who has encouraged and su pported me to achieve a Ph.D.
4 ACKNOWLEDGMENTS First and foremost, I express my deep appreciation to my advisor, Dr. John J. Sansalone, who gave me the best chance to work in this great academic area. He has consistently guided, encouraged and supporte d me throughout the journey to my Ph.D. program. His patient elucidation, enlightening ideas and precious comments have contributed a lot to my understanding of this research area and shaping my concept of scientist and engineer. It was and will be great f ortune and enormous inspiration for me in my life. I also extend my sincere appreciation to the distinguish professors on my committee: Dr. Ben Koopman Dr. James Heaney and Dr. Jennifer Curtis I am also very grateful to my committee members for their he lpful advice on the dissertation work. I express my thanks to my colleagues: Dr. Jong Yeop Kim, Dr. Natalie Magill Winberry, Dr. Gaoxiang Ying, Dr. Christian Berretta Dr. Joshua Dickenson, Dr. Ruben Kertesz and Dr. Tingting Wu who shared me with their kn owledge and helpful discussion. My appreciation also extends to my colleagues including, Mr. Karl Seltzer Ms. Christina Herr Joiner Mr. Adam Marquez Ms. Valarie Thorsen, Ms. Aniela Burant, Mr. Gregory Brenner, Ms. Sowmya Sankaran, Mr. Saurabh Raje Ms. Giuse p pina Garofalo Dr Young min Cho, D r. Se jin Youn, Dr Myung heui Woo and Mr. Rascal Cho for their valuable assistance and help Their friendships have been one of my important accomplishments in the past five years. I would like to express my specia l and warmest thanks to my best friend, Dr Subbu Srikanth Pathapat i, for not only his enormous help, but also his sincere friendship
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ ............... 4 LIST OF TABLES ................................ ................................ ................................ ........................... 8 LIST OF FIGURES ................................ ................................ ................................ ......................... 9 LIST OF ABBREVIATIONS ................................ ................................ ................................ ........ 11 ABSTRACT ................................ ................................ ................................ ................................ ... 16 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .................. 18 2 PHYSICAL MODELING OF PARTICULATE MATTER WASHOUT FROM A HYDRODYNAMIC SEPARATOR ................................ ................................ ....................... 22 Overview ................................ ................................ ................................ ................................ 22 Methodology ................................ ................................ ................................ ........................... 24 Physical Model Configuration ................................ ................................ ......................... 24 Flow Velocity Measurement ................................ ................................ ........................... 25 Pre deposited PM ................................ ................................ ................................ ............ 25 R TD Test ................................ ................................ ................................ .................. 26 Scour Thresholds ................................ ................................ ................................ ...... 28 Results ................................ ................................ ................................ ................................ ..... 29 In S itu V elocity P rofiles ................................ ................................ ................................ .. 29 Washout PM Granulometry ................................ ................................ ..................... 30 Residence Time Distributions (RTDs) ................................ ................................ ..... 31 Densimetric Froude Number ................................ ................................ .................... 32 Time Rate of Washout ................................ ................................ .............................. 33 Initiation of Scour ................................ ................................ ................................ ............ 34 Summary ................................ ................................ ................................ ................................ 38 3 PHYSICAL AND CFD MODELING OF PM SEPARATION AND SCOUR IN HYDRODYNAMIC SEPARATORS ................................ ................................ .................... 53 Overview ................................ ................................ ................................ ................................ 53 Objectives ................................ ................................ ................................ ............................... 56 Methodology ................................ ................................ ................................ ........................... 56 Physical Clarification and Re suspension Function Modeling ................................ ....... 56 CFD Modeling ................................ ................................ ................................ ................. 59 CFD Governing Equations ................................ ................................ .............................. 59 Particulate Phase Modeling ................................ ................................ ............................. 61 Re suspension and Washout Modeling ................................ ................................ ........... 63
6 Numerical Procedure ................................ ................................ ................................ ....... 63 Volumetr ic Efficiency Calculation ................................ ................................ .................. 63 Results ................................ ................................ ................................ ................................ ..... 64 Comparison Physical and CFD Modeling ................................ ................................ ....... 64 P hysical Modeling of Separation and washout Function ................................ ................ 64 PSD Result ................................ ................................ ................................ ....................... 66 PM Dynamics ................................ ................................ ................................ .................. 68 Fluid Velocity Magnitude ................................ ................................ ................................ 69 Probability of PM Separation and Washout ................................ ................................ .... 70 Summary ................................ ................................ ................................ ................................ 71 4 STEPWISE STEADY CFD MODELING OF UNSTEADY FLOW AND PM LOADING TO UNIT OPERATIONS ................................ ................................ ................... 82 Overview ................................ ................................ ................................ ................................ 82 Objectives ................................ ................................ ................................ ............................... 84 Methodology ................................ ................................ ................................ ........................... 84 Watershed and Three Hydrodynamic Separator Configurations ................................ ..... 84 Physical Modeling Methodology ................................ ................................ .................... 85 CFD Modeling Methodology ................................ ................................ .......................... 86 Particulate Phase Modeling ................................ ................................ ............................. 89 Modeling of Static Screen and Cartridge ................................ ................................ ........ 90 CFD Parameters ................................ ................................ ................................ ............... 91 Stepwise Step Modeling Removal and PM Separation ................................ ................... 91 Result ................................ ................................ ................................ ................................ ...... 93 Event Hydrology Indices ................................ ................................ ................................ 93 Probabilities of PM Separation by the BHS, SHS, and VCF ................................ .......... 95 Stepwise Steps Comparison to Measured Data ................................ ............................... 96 Summary ................................ ................................ ................................ ................................ 98 5 REMOVAL AND PARTITIONING OF NITROGEN AND PHOSPHORUS OF NUTRIENTS IN HYDRODYNAMIC SEPARATOR ON URBAN RAINFALL RUNOFF PARTICULATE MATTER GENERATED FROM IMPERVIOUS SURFACE CARPARK ................................ ................................ ................................ ........ 111 Overview ................................ ................................ ................................ ............................... 111 Objectives ................................ ................................ ................................ ............................. 112 Methodology ................................ ................................ ................................ ......................... 11 3 Catchment ................................ ................................ ................................ ...................... 113 Data Acquisition, Management, and Sampling ................................ ............................. 113 PM Separation ................................ ................................ ................................ ............... 114 Water Chemistry Analysis ................................ ................................ ............................. 115 Nitrogen and Phosphorus Analysis ................................ ................................ ............... 115 Partitioning Indices f or Nitrogen and Phosphorus ................................ ........................ 116 Hydrologic and Loading Parameters ................................ ................................ ............. 117 Analysis of Recovered Sediment Deposit from Hydrodynamic Sepa rator ................... 117 Results ................................ ................................ ................................ ................................ ... 117
7 Event Hydrology ................................ ................................ ................................ ........... 117 Overall Treatment Efficie ncy of BHS as a Function of Hydrology .............................. 118 PM fraction and PM based N and P fraction masses distribution ................................ 119 Event based Nitrogen a nd Phosphorus Loadings ................................ .......................... 119 Nutrients Removal Efficiency as a function of Hydrology ................................ ........... 120 Nutrients Partitioning ................................ ................................ ................................ .... 121 Nutrient from the Recovered Sediment Deposit ................................ ........................... 123 Summary ................................ ................................ ................................ ............................... 125 6 CONCLUSION ................................ ................................ ................................ ..................... 139 APPENDIX A CHAPTER 3. PHYSICAL AND CFD MODELING OF PM SEPARATION AND SCOUR IN HYDRODYNAMIC SEPARATORS ................................ ............................... 142 B CHAPTER 4. STEPWISE STEADY CFD MODELING OF UNSTEADY FLOW AND PM LOADING TO UNIT OPERATIONS ................................ ................................ ........... 146 LIST OF REFERENCES ................................ ................................ ................................ ............. 159 BIOGRAPHICAL SKETCH ................................ ................................ ................................ ....... 169
8 LIST OF TABLES Table page 2 1 Medianwashout rate and effluent mass load as a function of flow rates. SM is sandy silt in the Unified Soil Classification System (USC S). SM I is a SCS 75, SM II is a SCS 106, and SM III is NJDEP gradation ................................ ................................ ......... 40 2 2 d 10 d 50 d 90 for effluent SM I, SM II, and SM III as a function of flow rates ................... 41 2 3 d 10 d 50 d 90 for pre deposited PM ................................ ................................ ..................... 42 2 4 The summary of RTD tests as a function of flow rate.Qd is hydraulic design flow rate for baffled HS. Flow beyond 1 00% Qd over flows inlet weir and is not treated ....... 43 3 1 Summary of measured and modeled separation and washout function result with RPD. ................................ ................................ ................................ ................................ ... 73 4 1 Hydrologic indices across storm events for BHS, SHS, and VCF ................................ 101 4 2 CFD model comparisons to measured data across storm events for BHS, SHS, and VCF ................................ ................................ ................................ ................................ 102 4 3 CFD parameters for BHS, SHS, and VCF. ................................ ................................ ...... 103 5 1 Hydrologic characterization of the 10 rainfall runoff events monitored between May 24, 2010 and August 21, 2010 in Gainesville, FL ................................ ........................... 126 5 2 (TN), total dissolved phosphorus (TDP), and total phosphorus (TP). ............................. 127 5 3 Summary of event mean value and range of variation of the dissolved fraction (f d ) and partition coefficient ( K d ) of nitrogen and phosphorus for influent and effluent runoff. ................................ ................................ ................................ ............................... 128
9 LIST OF FIGURES Figure page 2 1 A plan view schematic of the baffled HS testing facility across sectional profile of the HS and PSDs for pre deposited PM. ................................ ................................ ........... 44 2 2 Velocity as a function of ADV height at ( A ) location in baffled HS and the mean flow velocity in the baffled HS as a function of flow rate and SOR.. ............................... 45 2 3 Effluent PSDs for the range of flow rates as a function of flow rate. Gamma p arameters as a function of surface overflow rate. Design SOR = 464.5 L/min m 2 ........ 46 2 4 T he relationships between turbidity and SORand PND as a function of SOR. T he relationships between turbidi ty and effluent PM and PND.. ................................ ............. 47 2 5 Median washout rate (g/min) and t i linear relationship in plot A and B as a function of SOR has a R 2 ........................ 48 2 6 Effluent PM as a function of densimetric Froude number, flow rates, and SOR.R 2 is 0.96 for SM I, 0.96 for SM II, and 0.97 for SM III as a function of Froude number. ....... 49 2 7 Effluent mass load range of flow rates as a function of time normalized to maximum duration. Washoff model parameters as a function of surface overflow rate. ................... 50 2 8 Measured and modeled washout rate in g/minas a function of time normalized to maximum duration and volume normalized to 7.6 turnover volume. ................................ 51 2 9 r SM I, SM II, and SM III as a function of particle Reynolds number. ( u is shear velocity, and v is kinematic viscosity of water). ............................... 52 3 1 Modeled PSD plots of treated and washout PM from BHS, VHS, and SHS. ................... 74 3 2 VHS, and SHS units. ................................ ................................ ................................ .......... 75 3 3 Effluent PM trajectories inside the BHS, VHS, and SHS for particle with diameters of 25 p ) is 2.65 g/cm 3 ............ 76 3 4 Washed out PM trajectories inside the BHS, VHS, and SHS for particle with p ) is 2.65 g/cm 3 .................... 77 3 5 Fluid velocity magnitude as a function of flow rates in BHS, VHS, and SHS. ................. 78 3 6 Pr obability of PM separation and washout by the BHS, VHS, and SHS. ......................... 79 3 7 Gamma parameters as a function of flow rates for BHS, VHS, and SHS (Q d for BHS is 9.1 L/s, Q d for VHS is 79.3 L/s, and Q d for SHS is 31.2 L/s). ................................ ....... 80
10 3 8 Volumetric efficiency (VE) as a function of flow rates in BHS, VHS, and SHS. ............. 81 4 1 Plot A is a plan view schematic of a BHS testing watershed in Gainesville, FL. Plot B is a plan view schematic of SHS and MFS testing watershed in Baton Rouge, LA ... 104 4 2 Isometric views of the geometries of unit operations ................................ ..................... 105 4 3 Probability of PM separation by unit operations and represent shape factor and scaling factor ................................ ................................ ................................ ................... 106 4 4 A ) is a CDF for the range of rainfall runoff flow rate (L/s) in BHS. B ) is flow rates (L/s) and effluent PM mass (g) as a function of elapsed time in BHS .. .......................... 107 4 5 A ) is a CDF for the range of rainfall runoff flow rate (L/s) in SHS. B ) is flow rates (L/s) and effluent PM mass (g) as a function of elapsed time in SHS .. ........................... 108 4 6 A ) is a CDF for the range of rainfall runoff flow rate (L/s) in VCF. B ) is flow rates (L/s) and effluent PM mass (g) as a function of elapsed time in VCF .. .......................... 109 4 7 Mean and variation of the stepwise steady model absolute RPD for BHS, SHS, and VCF. The lower right quartile box plot is the variation of absolute RPDs. .................... 110 5 1 Profile section of 1.21 m diameter BHS deployed for physical modeling loaded by urban source area catchment. ................................ ................................ ........................... 129 5 2 PM fraction and PM based N and P fraction masses distribution within ea ch monitored rainfall runoff event. ................................ ................................ ....................... 130 5 3 Separation for TN, and TP in different fractions as a functi on of PM fractions. Range bars represent standard deviation. ................................ ................................ .................... 131 5 4 Phosphorus mass concentration distri butions for each PM fractions. ............................ 132 5 5 Nitrogen mass concentration distributions for each PM fractions. ................................ .. 133 5 6 f d values and eq uilibrium coefficient, K d values of nitrogen and phosphorus in influent and effluent. ................................ ................................ ................................ ....... 134 5 7 Granulometric equilibrium distribution of ammonium nitrogen, nitrate nitrogen, TKN, phosphate and T P ................................ ................................ ................................ .. 135 5 8 The f d of influent and effluent TN (TP) as a function of cumu lative treated rainfall runoff volume. ................................ ................................ ................................ .................. 136 5 9 The cumulati ve gamma distribution parameters ( for shape factor and for scaling factor) for event based normalized particle size distributions (PSD). .. ........................... 137 5 10 Cumulative influent and effluent m ass of PM, phosphorus (P), and nitrogen (N) through the entire monitoring campaign for BHS in Gainesville, FL. ............................ 138
11 LIST OF ABBREVIATION S ADM Axial Dispersion Model ADV Acoustic Doppler Ve locimetry AOCM P Monodispersed AL Ox Coated Granular Media BHS Baffled H ydrodynamic S eparator BMP Best Management Practices CDF Cumulative Density Function CFD Computational Fluid Dynamics COD Chemical Oxygen Demand CSO Combined Sewer Overflows CV Co ntrol Volume D.O Dissolved Oxygen DPM Discrete P article M odeling EMC Event Mean Concentration FVM Finite V olume M ethod HS Hydrodynamic S eparator ICP MS Inductively Coupled Plasma Mass Spectrometry IPRT Initial Pavement Residence Time MBE Mass bala nce E rror MS4 Municipal Separate Storm Sewer System N Nitrogen NJDEP New Jersey Department of Environment Protection P Phosphorus PAH Polycyclic Aromatic Hydrocarbon
12 PDH Previous Dry Hours PLC Programmable Logic Controller PM Particulate Matter PSD Particle Size Distribution QA Quality A ssurance QC Quality C ontrol RANS Reynolds A veraged Navier Stokes RCF Radial Cartridge Filter RPD Relative P ercent D ifference RTD Residence Time Distribution SHS Screened H ydrodynamic S eparator SIMPLE Semi I mpl icit M ethod for P ressure L inked E quation SOR Surface O verflow R ate SM Non C ohesive S andy S ilt SSC Susupended Sediment Concentration SSE Sum of Squared Errors SWMM Storm W ater M anagement M odel TDN Total Dissolved Nitrogen TDP Total Dissolved Phosphoru s TDS Total Dissolved Solid TISM Tanks in Series model TKN Total Kjehldahl Nitrogen TN Total Nitrogen
13 TP Total Phosphorus TSS Total Suspended Solid UOPs Unit Operations and Processes USCS Unified Soil Classification System VCF Volumetric Clarifyi ng Filtration VE Volumetric Efficiency VHS Vortex H ydrodynamic S eparator d rain Rainfall Depth i rain max Maximum Rainfall Intensity n inf Number of Influent Samples n eff Number of Effluent Samples Q d Hydraulic Design Flow Rate Q med Median Flow Rate Q p Peak Flow Rate t i The time at which tracer initially appears t p The time at which peak concentration is observed t rain Rainfall Runoff Duration t t Theoretical residence time t 50 The time at which 50 % of tracer had passed through the reactor t 90 The time at which 90 % of tracer had passed through the reactor MDI Morrill Dispersion Index 1 / MDI Volumetric Efficiency t i /t t Index of short circuiting
14 t p /t t Index of modal retention time t 50 /t t Index of average retention time (t 50 )/(t 50 t p ) Haz 2 The variance t Mean detention time based on discrete time step measurements, T C i Concentration at i th measurement, ML 3 i Time increment about C i T 2 Variance based on discrete time measurements, T 2 d Diameter of the soil parti cle s Mass density of the soil d Diameter of the soil particle g Acceleration due to gravity Angle of friction of the soil S S Specific gravity of soil The shear velocity Shields parameter V runoff Vol ume of Runoff C d Dissolved fraction concentration C p Particulate bound fraction concentration C s Particulate bound mass (mg/g of dry particulate mass) f d Dissolved fraction f p Particulate bound fraction K d Partitioning coefficient
15 M d Dissolved mass M p Particulate bound mass M S Normalized Cumulative Mass Loading for PM M TN Normalized Cumulative Mass Loading for Total Nitrogen M TP Normalized Cumulative Mass Loading for Total Phosphorus
16 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy PHYSICAL AND CFD MODELS OF PM SEPARATION AND SCOUR IN HYDRODYNAMIC UNIT OPERATION S By Hwan Chul Cho May 201 2 Chair: John J. Sans alone Major: Environmental Engineering Sciences A hydrodynamic separator (HS) is commonly used as a preliminary unit operation for separation of particulate matter (PM) and PM associated constituents transported by urban rainfall runoff. Advantages of H S units are passivity, small treatment footprint, ease of retrofitting into existing sewer or treatment system infrastructure, efficacy for neutrally buoyant substances and detritus, low head loss, and capacity for hydraulic bypass beyond a given flow rate Although the small footprint of an HS is advantageous for integration into sewer (storm or combined) or drainage systems, it also concentrates flow energy. In many HS units where PM sludge is not isolated or in units not maintained (cleaned) frequently, washout of previously separated PM sludge can result in short periods of net export of PM. The purpose of this study was to increase understanding of the hydrodynamic and clarification response of best management practices ( BMPs ) for urban rainfall runoff management F our types of unit operations were investigated by means of a coupled experimental and numerical approach. Additionally t his study investigates PM washout from three HS units as a function of steady flow rates and particle size distributions (PSDs), using a computational fluid dynamics (CFD) assessment CFD is a branch of fluid mechanics that uses numerical methods to integrate the Navier Stokes equations
17 to solve fluid flow problems. Four types of full scale HS units were modeled in 3D using FLUENT v 6.0. A finite volume method (FVM) was applied to discretize the governing equations into the physical space directly. Modeling in 3 D is less susceptible to the complications from the lack of geometric symmetry, comp lex static screen geometry, vortex flow and gravitational forces on the motion of particles in unit operations Post processing the CFD predictions provided insight into the mechanistic behavior of the HS by means of three dimensional hydraulic profiles, p article trajectories and pressure distributions. A stepwise steady flow model effectively predicts the monitored storm events data in UOPs. This study examines the inter and the intra event nitrogen and phosphorus removal as a function of particle size, hydrology and partitioning for an urban carpark treated by unit operations with significant biogenic loadings.
18 CHAPTER 1 INTRODUCTION Impervious surface area and urban runoff can impair receiving waters. Many sources of PM, nutrients (N, and P), metals, and anthropogenic chemicals are present and exposed to rainfall runoff in urban areas (Lee and Bang 2000). PM is a significant vehicle in the transport of these constituents by urban runoff (Sansalone 2002). PM delivered by rainfall runoff varies temporal ly within a storm event and across storm events and can vary spatially within the same watershed (Sansalone 2002). Hydrodynamic separators ( HS ) are commonly used urban unit operations to remove oil and inorganic materials including PM in urban stormwater r unoff and c ombined s ewer o verflows (CSOs) (Brombach 1987, Brombach et al. 1993, Pisano et al. 1994, USEPA 1999, Andoh and Saul 2003). In North America there are over 50,000 HS operating HS units. The advantages of HS are that they are passive devices, of ten have a small footprint, and can be easily retrofitted to existing infrastructure. However, many of these systems are left unattended for long stretches of time (e.g., a minimum range of 1 to 47 days between rainfall events in Gainesville, FL ) An estim ation of long term performance of detention basins has been derived by storm water management model (SWMM) simulation (Nix et al. 1988). Minimizing scour from a HS was recently added to the essential elements of b est m anagement p ractices (BMPs) design (EPA 200 4 ). Therefore, management of scour from HS is a challenge that need to be addressed. Sediment transport and re suspension has been widely studied by many researchers however, most of research has focused on sediment re suspension and transport mechani sms for open channel flow (Cellino and Lemmin 2004, Gargett et al. 2004, Orlins, and Gulliver, 2003) not in a HS. T here is a need to quantify scour in a baffled HS (BHS) to provide designers with an understanding of scour as function of particle diameter, and flow rates. In this study, a series of
19 scour tests was performed on BHS with two different sediment pre loaded conditions with particle size distributions ( PSDs ) defined prior to testing across six flow rates representing 25 % (0.91 L/s) to 125 %(11.31 L/s) of the maximum h ydraulic o perating f low r ate (Q) or design flow rate (Q d ) of the unit test ed The tests are conducted under steady flow regimes, to better distinguish the effects of different pre loaded PM and flow rates. Residence time distribution ( RTD ) testing is conducted to characterize the flow mixing behavior in a baffled HS. The experimental RTD testing is performed with pulse input method. The flow rate (Q), and the geometry of a HS influences the flow mixing behavior of a HS. A primary compo nent of this study is a comparison of multi phase physical modeling to CFD modeling of HS CFD approaches are increasingly utilized to model particle laden flows ( Curtis et al 2004 ; van Wachem et al 2003) CFD can predict fluid flow, mass transfer, chemi cal reactions, and related phenomena by solving governing fluid equations using numerical methods. CFD modeling has been used for describing the behavior of rainfall runoff unit operations and processes ( UOPs ) (Pathapati and Sansalone 2009). In urban stor mwater, CFD has enhanced the modeling of PM separation for transient flows (Sansalone and Pathapati 2009; Garofalo and Sansalone 2011) and heterodisperse particle size distributions (Dickenson and Sansalone 2009) as well as re entrainment of PM by scouring mechanisms (Pathapati and Sansalone 2012). However, a 3 D numerical model to simulate the transient hydrodynamics with variable PSDs needs longer computational time than a model simulating steady state flow (Pathapati and Sansalone 2011). T he unsteady com putational time for the hydrodynamic separator (HS) is approximately 30.5 days using a workstation equipped with dual quad core 2.6 GHz processors and 32 GB of random access memory (Pathapati and Sansalone 2011). Stepwise step flow modeling can be used t o reduce computational time.
20 Stepwise steady CFD modeling utilizes a series of discretized steady flow steps of rainfall runoff events. Identifying numbers of stepwise flow steps is necessary to ensure efficient use of computational time. Excessive phosph orus loading to natural waters has been known for decades to accelerate eutrophication and lead to decline in water quality of rivers, lakes and oceans (Weon et al. 2002). In urban rainfall runoff, phosphorus partitions from PM into dissolved phases. Rainf all runoff characteristics including water chemistry, unit residence time, and unsteady loading, dictate the partitioning of phosphorus and nitrogen. Additionally oxidation reduction, mixing, particulate composition and particulate bound concentration gr adients can affect the extent of partitioning of phosphorus and metals (Sansalone and Buchberger 1997). PM based phosphorus distributes across the PSDs (Ma et al. 2010). Particulate bound phosphorus is distri buted across sediment (> P hosphorus predominantly found in urban rainfall runoff is associated with particles greater than 75 m (Sansalone et al. 1998). Physical unit operations typically only target the particulate bound fraction of phosphorus (Barrett et al. 1998, Bartone and Uchrin 1999, Comings et al. 2000, Dechesne et al. 2005). For this reason, the phosphorus control in physical unit operations may be ineffective without knowledge of the partitioning. Phosphorus partitioning is important not only to understand phosphorus fate and transport, but also to select appropriate unit operation in the watershed. Elevated nitrogen is ubiquitous in urban watersheds (Hopkinson and Vallino 1995). Nitrate nitrogen leaching from biogenic material is a source of contamination of surface and groundwater (Broschat 1995, Ku and Hershey1997). Total nitrogen in runoff is associated with dissolved and particulate phases as well as biogenic material. Nutrie nt uptake is greatly influenced by environmental variables such as water availability and temperature (Marschner
21 1995). Characterization of nitrogen in urban runoff is necessary not only for improving treatment strategies for nitrogen reduction, and but al so for identifying viable treatment. This dissertation focuses on the examination of scour, PM separation mechanisms, partitioning of nitrogen and phosphorus, and behavior of a B HS to separate PM, nitrogen and phosphorus. The performance of a B HS is evalu ated and studied by controlled and uncontrolled physical models Models are subject ed to steady conditions with regula ted PSD gradation and actual rainfall runoff treatment under stepwise steady step conditions as well as CFD modeling.
22 CHAPTER 2 PHYSICA L MODELING OF PARTICULATE MATTER WASHOUT FROM A HYDRODYNAMIC SEPARATOR 1 Overview Particulate matter (PM) serves as a vehicle for the transport of chemicals, acting as a substrate to which chemicals reversibly partition (Berretta and Sansalone 2011a; Brezon ik 2002; Dean 2005). Beyond being a substrate for chemicals, PM also impacts the turbidity and oxygen demand of receiving waters (Sansalone 2002, Shinya et al. 2000). Screened hydrodynamic separator (HS) units are commonly used as preliminary unit operatio ns for publicly owned treatment works (POTWs), followed by primary clarification (sedimentation) and filtration. For baffled HS units, PM is separated by settling, although the original design intent was primarily for oil and grease separation (OGS). Histo rically, screened forms of HS units have been applied to preliminary and high rate treatment of combined sewer overflows (CSOs) (Brombach et al., 1987; Brombach et al.; 1993; Pisano et al., 1994; USEPA 1999; Andoh et al., 2003). Advantages of HS units are passivity, small treatment footprint, ease of retrofitting into existing sewer or treatment system infrastructure, efficacy for neutrally buoyant substances and detritus, low head loss, and capacity for hydraulic bypass beyond a given flow rate. Nonetheles s, commensurate with these advantages are disadvantages. For instance, while a small footprint provides economy and integration into standard drainage structures, the limited spatial footprint and volume results in concentration of hydraulic energy, short residence times and limited PM separation whether for Type I (discrete) or II (flocculent) settling. One challenge associated with HS units is the need for frequent maintenance (cleaning) so that, as an HS collects PM, the unit neither functions 1 Re printed with permission from Cho, H., and Sansalone, J.J., (2012), Physical modeling of particulate matter washout from a hydrodynamic separator, Journal of Environmental Engineering
23 as a sourc e of PM and chemicals through washout, nor generates anaerobic conditions in the stored PM sludge (USEPA 2002, Sansalone et al. 2010). There has been a long history of scour research, with a specific interest in bridge foundation scour under the influence of flood flows, the study of which dates back over a half century (Shields 1936; Vanoni 1946). More recently, there have been questions regarding the applicability of approaches such as Shields scour diagram across different flows regimes (Buffington 1999) Although HS units are subject to open channel flow, they are small footprint units subject to highly unsteady flows, high velocities and complex intra unit hydrodynamics. Because these units are infrequently (on the order of once a year) maintained (Sans alone and Pathapati 2009) high fluid velocities at the PM sludge interface generate scour and washout. Hence, there is a need to characterize these phenomena. In this study, PM washout is specifically examined through controlled physical modeling supported by PM, flow and velocity monitoring. Velocity monitoring uses Acoustic Doppler Velocimetry (ADV) to measure fluid velocities and vectors (Kraus et al. 1994; Lohrman et al.1994;Voulgaris et al. 1998). In addition, residence time distribution (RTD) curves a re used to characterize the hydrodynamic behavior of systems operating at steady flow (Fernandez Semprere 1994). The RTD methodology is well documented and therefore not reproduced herein (Levenspiel 1962; Froment and Bischoff 1979; Smith 1981; Denbigh and Turner 1984; Fogler 1992; Westerterp et al. 1993). In this study, physical model load response experiments are performed on a full scale baffled HS subject to the following conditions: (1) pre deposited PM of differing PSDs and (2) differing steady flow r ates representing 10 % (0.91 L/s) to 125 % (11.31 L/s) of the maximum design flow rate (Q d ). For the HS unit, washout is hypothesized to be a function of the PSD of the pre loaded PM and of the flow rate (or SOR) at a given PM deposit (sludge) depth. In or der
24 to test this hypothesis, data, including effluent PM and PSDs, velocity profiles in the HS and RTDs, are determined at each flow rate and for each batch of pre loaded PM. It is hypothesized that washout of the pre loaded PM can be explained using these data to illustrate effluent PM load generated from previously settled PM deposited in the HS. It is further hypothesized that effluent PM concentration is a function of the densimetric Froude number and SOR for each PSD. These results are contrasted with the open Methodology Physical Model Configuration A schematic of the HS component of the physical model system is presented in Figure 2 1 along with the HS profile view. While the HS unit was initially develop ed as a commercial/industrial OGS unit, (hence the inverted horizontal baffle design), for over a decade the baffled HS has primarily served as a small urban watershed settling unit for PM (USEPA 1999). In this controlled physical model study the source in fluent for each model run is potable water (PM < 0.1 mg/L as suspended sediment concentration (SSC) turbidity <0.3 NTU and 20 C 2C) stored in two 45 m 3 tanks. The flow delivery system consists of two centrifugal pumps (19 L/s 3 HP and 75.6 L/s 10 HP ). The flow monitoring system consists of dual monitoring meters in parallel, an electromagnetic meter for flows from 1 to 272 L/s and a volumetric flow meter for flow from 0.1 L/s to 10 L/s. The diameter of the baffled HS is 1.22 m. The distance from the bottom of the HS lower chamber to the invert of the HS outlet is 1.52 m (H max ). The surface area of the unit is 1.17 m 2 At the design flow rate, one unit volume (one turnover volume) is approximately 1780 L. The inlet and outlet diameters are 0.2 m each. The design flow rate (Q d ) is 9.05 L/s, which is observed to represents 90% of the hydraulic capacity of the unit (defined as the flow rate at which bypass over the overflow weir begins).
25 Flow Velocity Measurement All ADV based velocity measurements are ta ken within the HS. The velocity measurements are monitored using a 10 MHz ADV (velocity range: 1 mm/s 2.5 m/s) (Sontek/YSI). The velocity determination is based on the Doppler shift principle, which is implemented with a bistatic (focal point) acoustic D oppler system and consists of a transmitter and three receivers (Voulgaris 1998). The ADV is a multi function (sound emitter, sound receiver, and signal conditioning electronic module) water velocity monitoring device for precise, in situ measurement of ve locity. Fluid velocities are measured at 10 locations in the HS. In a horizontal plane at the water PM bed interface, (0.17H max ; H max = 1.52 m), there are seven measurement points. In addition, velocity measurements are made at the center of the unit, lo cation ( A ) at 3 different heights (0.34, 0.50, and 0.67H max ). A schematic of the velocity measurement locations is shown in Figure 2 1 A Fluid velocity is measured at each flow rate during the washout and RTD runs. Pre deposited PM Three hetero disperse and non cohesive sandy silts (SM) are utilized in three series of separate but equivalent testing conditions to examine HS washout. The Unified Soil Classification System (USCS) is utilized for textural classification of each gradation of PM (Coduto 1999) The mass based PSDs of the pre loaded PM are measured utilizing wet dispersion laser diffraction based on the principle of Mie small angle light scattering and diffraction due to the presence of particles between the laser emitter and detector (Finlayson Pitts and Pitts 2000). Diffraction patterns are dependent on particle sizes, with larger scattering angles associated with smaller size particles. Although the siliceous silts are sub rounded, Mie theory applies not only to spherical but also non spherica l particles (Jonasz 1991). Scattering patterns are utilized to determine the % volume of particle sizes of the PSD through an optical method in
26 which the detectors and windows are integral parts of the measurement zone. The particle analyzer used is a Malv three PM gradations are designated as SM I, SM II and SM III, with a d 50m 67 Si lica. As part of the testing matrix summarized in Table 2 1, PM washout testing is performed at 0.91 (10), 2.26 (25), 4.78 (50), 6.79 (75), 9.05 (100), and 11.31 L/s (125% of Q d ). Each washout run is conducted by first pluviating PM across the entire surfa ce area (1.17 m 2 ) of the HS bottom chamber to a level depth of 0.17H max The HS is then filled with water at a flow rate of less than 0.5 L/s to avoid any flow scouring of the PM bed. Since HS units store runoff between runoff events, runs are conducted wi th the HS unit filled with water to the outflow invert. Flow is diverted around the HS until a steady state flow is achieved. Once flow is directed into the HS, sampling is initiated and discrete 1L samples are manually taken in duplicate at equal volume s ampling intervals, calculated based on flow rate and time. Run duration is chosen such that approximately 7 volumetric turnovers are achieved (each HS volume is 1.78 m 3 ). Ten duplicate samples are collected for PSD and PM analysis (as SSC). Studies have de monstrated that SSC is a representative gravimetric index for hetero disperse PSDs, which include sand size or coarser PM (Gray et al. 2000). SSC analysis is carried out by filtering the entire volume of each replicate R TD Test The RTD quantifies the hydrodynamic characteristics of a unit operation at a given flow rate (Fogler 1992). In this study, the friction velocity at the water PM interface (0.17H max ) generated at a given flow rate is indexed by the SOR of the HS unit. The RTD of the baffled HS is determined by using NaCl as non reactive tracer and measuring conductivity as a function of
27 flow rate and time. A concentrated NaCl solution is injected as a pulse input. The RTD analysis time is approximately 3 4 times the theoretical mean residence time (Nauman and Buffham 1983). Effluent conductivity is monitored with a calibrated conductivity probe (600 OMS O) manufactured by YSI Inc. Each tracer test is validated by a mass balance check with the error tolerance set at 5% by mass. The inert tracer ( NaCl : 200 g/500 mL) is prepared in 1 L polypropylene (PP) bottle, and injected in the influent drop box as a single pulse. The tracer concentration is meas ured at the effluent pipe for each fully developed flow rate from 2.26 to 11.31 L/s. The conductivity probe is fully submerged at the outfall of the effluent pipe. Conductivity measurements are taken at 1 second intervals. The mean residence time is define d as follows ( Hazen 1904; Tchobanoglous et al. 2003): ( 2 1) The RTD function E(t) is related to the concentration, C(t) while the cumulative RTD curve is designated as the F curve. E(t) and F(t) are expressed as follows. ( 2 2) ( 2 3) The following RTD indices are reported in this study: t t ,, the theoretical residence time, (defined as t t = V/Q, where V is the volume of the HS unit, and Q is the flow through the system); the Morrill dispe rsion index (MDI := t 90 / t 10 where tx is the time at which x% of the tracer had eluted the mean detention time based on (t 50 )/(t 50 t p ) (Levenspiel 1972; Letter man 1999), an index of short circuiting used for overflow models (Magill and Sansalone 2010); and t p the time
28 at which peak concentration is observed. The values of t x (x=10, 50, 90, etc) and t p are obtained from measured tracer data. The mass balance err or with respect to the mass of tracer injected was 5 %. The variance of tracer concentration as a function of time is approximated as follows. ( 2 4) In this expression, 2 c is the variance and is based on discrete time measurements (T 2 ); t i is the time at which tracer initially appears. Scour Thresholds The applicability of the Shields approach to scour thresholds in the HS unit and commensurate PM washout is tested in th is study. Scour of non cohesive PM at the water PM deposit interface is induced when the effective vertical and lateral confining stresses acting on PM at this interface are less than the shear stress generated by the flow at this interface (Annandale 2006 ). Common parameters to estimate PM transport capacity include the critical c ) needed to generate the incipient motion of a particle is expressed as follows (Beheshti and Ataie As htiani 2008). ( 2 5 ) In this expression, d is particle diameter, s is mass density of PM or soil, g is acceleration due to gravity, and is the internal angle of friction of the PM or soil. Peck (1974) determi ned a range of 29 < < 41 for very loose to very densely packed non cohesive sediment. The critical shear stress cannot be predicted directly from a Shields curve and requires an iterative procedure. (Beheshti and Ataie Ashtiani 2008) Shear stress can be defined based on the shear velocity. ( 2 6 )
29 is the shear velocity. The Shields parameter, w hich is widely used to predict the initiation of particle motion, is not measured directly but rather derived from shear stress and c the shear stress is expressed in dimensionless form as the Shields parameter. ( 2 7 ) In this expression, S s s s is the sediment viscosity [ML 1 T 1 ]; 1 T 1 ]); and g is the acceleration due to gravity [L/T 2 ]. Scour occurs 0 is l Results In S itu V elocity P rofiles Tables 2 1 and 2 2 summarize the physical model conditions and the results of 18 washout experiments performed on the baffled HS with SM I, SM II and SM III pre loaded PM deposits. As illustrated in Figure 2 2, the flow velocities in the HS range from 0.7 to 44.6 mm/s for all locations, depths and flow rates. Spatial symmetry does not consistently translate into velocity profile symmetry. For example, when comparing symmetric locations ( D ) and ( E ) or ( F ) and ( G ) in Figure 2 1 A the velocity profiles are not symmetric. However, there is an approximately linear relationship between flow rate and mean velocity at each spatial location and at each depth along the central axis of the HS. Maximum velocities consisten tly occur near the effluent drop pipe which is located between locations ( B ) and ( C ) in Figure 2 1. Given the requirements of continuity and the smaller 102 mm diameter of the effluent drop pipe, higher velocities are expected due to the pressure increase at the effluent drop pipe. Based on the spatial distribution of these velocities with a minimum mean velocity occurring at ( A ), all further velocity measurements are performed with the ADV fixed at location ( A ) for each depth.
30 Washout PM Granulometry The g ranulometry of eluted PM is represented as mass based PSDs, particle number density (PND), turbidity and PM measured as SSC. Washout response of the HS is examined as a function of SOR and pre deposited PM gradation based on measured effluent PSDs. The PSD results in Figure 2 3 indicate that PM washout is predominately fine PM for all gradations of pre deposited PM. Upon converting the PSD into a mass distribution, the effluent mass based d 50m r anged from 0.8 to 1.0 for SM III for the range of flow rates tested in Figure 2 2. Higher flow rates generate consistently higher values of d 50m for all gradations. In Figures 2 3 A C and E the mass based P SDs are plotted. These graphs take the form of a two parameter gamma distribution. The probability density function of a gamma distribution is given by the following expression. ( 2 8 ) factor. Below, G(x) is the cumulative gamma distribution and x represents particle diameter. ( 2 9 ) representatio n of the shape (hetero dispersivity or decreasing uniformity of the size) of the PSD is parameterized in Figures 2 3 B D and F for each eluted PSD. Figures 2 3 B D and F illustr which constant
31 pre deposited PM gradation.. The shape parame ter remains relatively constant across the range of SORs in particular for the coarser hetero disperse PM, SM III. In contrast, the scaling parameters for the eluted PSDs generated from the much finer and nominally more uniform SM I and II remain essential than for the eluted PM generated from SM III. As seen in Figures 2 4 A and C the phenomenon of turbidity, which is primarily generated by fine PM, displays linear trends with SOR. The se trends are unique for each pre deposited gradation. Figure 2 4 also illustrates the linear relationship between effluent PM as a gravimetric concentration and turbidity. The results in Figures 2 4 B and D do not illustrate distinctly different linear rel ationships for SM I and II due to the primary influence of eluted fine PM on concentration and turbidity. In contrast, while the eluted PND and turbidity are also linearly related for SM III as shown in Figure 2 4 B and D this coarser and hetero disperse gr adation produces an elution with lower PND and turbidity. Residence Time Distributions (RTDs) Table 2 3 illustrates the RTD indices of the HS as a function of flow rates. At lower flow rates the low volumetric efficiency values indicate that the entire HS volume is not utilized. For a given HS unit size the flow rate (or SOR) is seen to be a primary factor in determining washout behavior of the HS unit. SOR is calculated as follows. ( 2 10 ) In this expression Q (L 3 /T) is flow rate an d A (L 2 ) is the surface area. Ideal discrete settling theory (Tchobonaglous et al 2003) suggests that if a particle has a settling velocity greater than the SOR, then the particle is separated as in a clarification basin. PM separation is calculated as fo llows.
32 ( 2 11 ) The first term on the right of this expression represents the mass fraction of particles with settling velocity ( u p ) greater than SOR and the second term represents the mass fraction of particles with settling velocity ( u p ) less than Q/A SOR was used as an index which allows a direct comparison to Type I (discrete) settling velocities. Measured residence times are lower than theoretical residence times for flow rates lower than the design flow rate (9.05 L/s), and high er than theoretical residence times for flow rates higher than the design flow rate. At lower flow rates, volumetric zones of the HS are not 2 3 provides 2 / t t 2 and volumetric efficiency based on MDI, t 50 / t t and the modal retention time index t p / t t Measured and theoretical residence times did not differ significantly at the design flow rate of 9.05 L/s, suggesting that at the d esign flow these zones are mobilized. Figure 2 5 illustrates the median washout rate and short circuiting index as a function of SOR. Median washout rates increase linearly with increasing SOR for each of the PM gradations. Figure 2 5 A illustrates that S M I, the finest gradation, has the highest median washout rate of the three gradations. The index of short circuiting, t i /t t also increases linearly with increasing SOR, similarly to the behavior of the median washout rate. With increasing SOR there is hi gher washout rate from the HS as the volume at the PM water interface is increasingly mobilized. Densimetric Froude Number Previous research (Aguirre Pe et al. 2003) has suggested using a modified form of the Froude number, the densimetric Froude number, to relate flow characteristics to scour (in this case washout). The densimetric Froude number as a function of PM diameter is useful as an
33 independent variable in predicting the washout, as shown in Figure 2 6. This Froude number relates flow velocity to t he PSD granulometric and gravimetric characteristics as follows: ( 2 12 ) In this expression, s is the specific gravity, d 50m is from the effluent PM and V is the flow velocity in the baffled HS. The flow velocity, V, in the baffled H S is measured using ADV. To evaluate sedimentation for a particle subject to Type I settling, SOR as a function of PM size is a commonly used design index parameter for wastewater and stormwater detention/retention basins. SOR has units of velocity (L/T). The upward overflow velocity is compared with the nominal downward velocity of each particle size to determine sedimentation. Time Rate of Washout As the flow ranged from 10 to 125% of Q d the median SSC eluted ranged from 8.2 mg/L to 20.6 mg/L for SM I, f rom 7.3 mg/L to 13.2 mg/L for SM II, and from 2.0 mg/L to 10.4 mg/L for SM III. For SM I at a flow of 0.1Q d the eluted SSC increased by 13% as compared to SM II and by 310% as compared to SM III. The effluent d 50 values summarized in Table 2 2 illustrate that the eluted PM was dominated by PM in the suspended range. Figure 2 7 shows that mass washout decreases exponentially with time. For each gradation, the net eluted mass approaches a low asymptotic equilibrium value after approximately 0.7(t/t max ), whic h equates to approximately 9500 L or 5.3 turnover volumes for this unit. The initial exponential decline lasting up to 0.4(t/t max ) is likely a result of scouring the pre deposited non cohesive PM interface where the effective confining stress is essentiall y zero and there is negligible interlocking between particles. As a saturated non cohesive bed, the primary source of shear strength is frictional and the surface layer does not benefit from a confining stress. There is a clear dependence of the initial wa shout magnitude on the corresponding magnitude of the influent flow rate. However,
34 irrespective of the flow rate and PM gradation, a first order exponential washout model similar to that employed by Sartor and Boyd (1974), Alley (1981), and Alley and Smith (1981) for washoff of PM from pavement by runoff is observed. Analogous to these washoff models where PM availability is not limiting, the evolution of washout can be expressed as follows. ( 2 13 ) In this expression, t is the efflue nt sampling time (min); M is the mass load (g); m 0 is M for the initial washout rate (g/min), and k is inverse washout time scale (1/min). The mass and rate of washout are modeled by this equation. Eluted mass rates as a function of volume are shown in Fig ure 2 8 for each flow rate. Initiation of Scour The resolution of the ADV utilized in this study ranged from 1 mm/s to 2.5 m/s. Fluid velocities were measured at a level within 5 mm of the sediment surface, providing an approximation of the velocity at the water PM interface that induced shear. To account for fluctuations of velocities at each measurement location all velocities were obtained as a mean of seven equally distributed locations across the area of the HS at a given level as shown in Figure 2 1. In addition, at each location, velocities in the x, y and z direction were recorded in triplicate at 1 second intervals, for each steady flow rate, for run times ranging from a minimum of 20 minutes to a maximum of 100 minutes. The open volume encompassed by the three small legs of the ADV is approximately 3000 mm 3 The impact of the ADV legs and cable on the flow field were assumed to be negligible as the HS volume is much larger (1.78 m 3 ) Shear stresses are determined using measured velocities adjacent t o the water PM interface The results are shown in Figure 2 9 for each deposited PM gradation. SM I, SM II and SM III illustrate similar trends for shear stress and critical shear stress at the water PM interface.
35 Based on physically measured velocities, P SDs and flow rates, the PM bed interface shear stress does not provide a clear relationship with the observed washout rates. Currently, there is no shear stress criterion for HS units. The dimensionless shear stress in the HS at each flow rate is significa ntly lower than the Shields criterion, suggesting that there should not be scouring. However, even with low shear stresses measured washout occurs. Using the Shields criterion the washout rate cannot be inferred directly from the calculated shear stress at the PM water interface of the baffled HS. Given that the baffled HS is a circular Type I settling tank, a (1996) is used to calculate scour velocity in settling tanks and it is also tested against the measured results. ( 2 14 ) In this expression, V sc is the scour velocity; s is the specific gravity of the PM deposit; g is the acceleration due to gravity; d is the scoured PM size; Q is the flow ra te; B is equivalent width of the settling tank, and k ranges from 0.5 and 0.8 depending upon the specific geometry of the configuration (Ingersoll and McKee 1956). Resulting scour velocity ranged from 0.003 to 0.01 mm/s, across the SOR range. These veloci ties are sufficiently smaller than measured velocities, indicating that washout should not be generated; a result not supported by the observed washout. The results of this study have several practical implications. One implication relates to the freque ncy of maintenance to minimize washout. For instance, for the baffled HS it is shown that washout occurs when separated PM reaches a given depth. This depth can be calibrated for the HS unit with catchment PM loading data and annual rainfall runoff volume either as
36 measured data or from continuous simulation models such as the Stormwater Management Model (SWMM). While HS units are all predominately Type I settling units, there are a number of internal configurations for HS units. For example, in contrast to the horizontally baffled HS unit of this study, a screened HS consisting of a screened and volute chamber is only capable of size fraction of SM III. The screened HS design directs flow energy in a downward spiral towards the sump that collects these PM deposits, allowing for mobilization and washout of PM (Pathapati passing PM as coarse as gravel size. In contrast, results from this study illustrate that the design of the baffled HS unit volumetrically isolates the deposited PM. There is a tradeoff between hydraulic capacity (flow rate or SOR), dissipation of flow energy and the propensity for washout. The horizontally baffled HS unit provides dissipation of incoming flow energy with depth and, correspondingly, creat es an isolation zone for deposited PM. In comparison, in the screened HS flow energy is directed circumferentially downward and into the vertically screened area to the sump containing deposited PM, before exiting upward and outward through the screen, the volute chamber, and finally from the unit. This study illustrates that washout is a function of the PSD of deposited PM and also of SOR. While deposited PSDs and SOR can be controlled to a degree by HS unit sizing and design, results of this study suggest that isolation of the deposited PM is critical. Even with isolation above the deposited PM, many studies have demonstrated that water chemistry can change significantly in unit operations that store submerged or wet PM deposits between runoff events. For example, sumps with stored runoff and PM can go anaerobic within two days; far shorter than the mean time between runoff events for nearly all climatic regions of the USA
37 (Sansalone et al. 2010). Additionally, PM sludge deposits in wet sumps are habitat fo r microbial growth and a source for chemical leaching from PM (Ying and Sansalone 2008). Frequent cleaning is not practiced for de centralized unit operations (manufactured or non proprietary). Yet the tradeoff is that unmaintained unit operations can be u nintended sources of PM, pathogens and chemicals thereby degrading their intended role as temporary sinks between maintenance points. While there are environmental benefits from more frequent maintenance, stakeholders can also benefit through generation of load credit programs to offset maintenance costs. States such as Florida are providing nutrient load credits for quantifiable and documented maintenance practices of unit operations and drainage appurtenances (Berretta et al. 2011b). If consideration is g iven to on line versus off line installation of HS units, from these results, the baffled HS unit provided isolation of deposited PM. With isolation, effluent indices of the baffled HS for each PM gradation tested were nominal but demonstrable at the desig n flow rate with gravimetric concentrations of 9 to 18 mg/L, effluent turbidity ranged from 10 to 40 NTUs and PND was in the range of 10 9 #/L. While these results can allow on line application with maintenance of the baffled HS, other HS configurations are more prone to washout. By comparison, at the design flow rate of the screened HS described above, the PM washout concentration for the coarsest PM gradation (SM III) is 51 mg/L. Therefore on line versus offline applications require washout evaluations as conducted in this study. Additionally, while the PSDs used are reproducible certification gradations, these are inorganic (siliceous) particles and the aqueous matrix is a reproducible potable water matrix. This provides a reproducible and precise testing metric for washout comparisons but may not accurately reproduce field conditions of varying runoff chemistry, a mixture of biogenic and anthropogenic PM, microbial growth and anaerobic conditions in unit operation sumps and sludge zones.
38 Summary Hydrodynam ic separator (HS) units are commonly deployed in small developed watersheds to provide stormwater PM separation. A baffled HS unit, one common HS type, was analyzed for washout of pre deposited PM as a function of surface overflow rates (SOR) indexed as fl ow rates from 10 to 125% of the HS design flow. Washout was also examined for three pre deposited hetero III at < o obtain a hydrodynamic does not consistently translate to spatially similar velocities at the water PM interface. Results from RTD analysis with an inert tracer demonstrat ed that the mean and theoretical residence times did not differ significantly at the design flow rate. Only as the flow approached design flow rate was the volume of the HS mobilized towards the depth of the water PM interface. As a function of SOR, the me dian rate of washout ranged from 0.4 to 13.3 g/minute for SM I, from 0.3 to 4.9 g/minute for SM II, and from 0.2 to 3.1 g/minute for SM III and were statistically significantly different. A densimetric Froude number, relating flow velocity to gravimetric and granulometric PM indices of the washout PM, reproduced modeled PM washout as a mass concentration for all PSDs across the SOR range. During washout the finer PM at or near the water PM interface subject to negligible effective confining stress was pref erentially mobilized and eluted. The unmobilized coarser PM fraction of the PSD functioned to confine lower PM and attenuate continued PM elution. For each PM gradation there was an exponential decline in PM washout on a gravimetric basis as a function of washout time. In contrast to gravimetric based washout, eluted turbidity and particle number density (PND) primarily influenced by finer suspended PM, both displayed a linearly increasing washout trend as a function of SOR and PSD. Washout, as
39 measured by turbidity and PND, differed markedly between the coarsest (SM III) and the finer PSDs (SM I and II). Irrespective of the basis used, washout indices are a function of SOR and PSD for a given depth of PM deposits. Using the Shields criterion, an open chan nel approach to washout, negligible washout was predicted, which did not replicate the measured washout from the HS unit. The scour velocity method for settling tanks of Swamee and Tyagi (1996) also did not reproduce the measured washout results of this st udy. Results suggest the investigation of models capable of coupling hydrodynamics and PSD/PND such as computational fluid dynamics (CFD).
40 Table 2 1 Medianwashout rate and effluent mass load as a function of flow rates. SM is sandy silt in the Unified S oil Classification System (USCS). SM I is a SCS 75, SM II is a SCS 106, and SM III is NJDEP gradation Texture Classification of PM Target flow rate (L/s) Operating flow rate (L/s) Surfaceover flow rate (L/min m 2 ) Volume of flow (L) Time duration of run (mi n) Median washout rate (g/min) Median SSC [mg/L] Effluent mass load (g) SM I (Sandy silt < 75 m) = 0.8, 0.91 0.90 46.7 12902 250.0 0.43 8.2 298.5 2.26 2.46 126.3 14205 100.0 1.06 9.8 386.5 4.78 4.80 245.3 14394 50.0 2.81 11.0 485.4 6.79 7.00 348.5 13861 33.3 4.89 13.9 644.0 9.05 9.30 464.5 13936 25.0 7.33 17.8 742.5 11.31 11.60 58 0.5 13931 20.0 10.28 20.6 897.6 SM II (Sandy silt < 100m) = 0.7, 0.91 0.90 46.7 12955 250.0 0.28 7.3 167.6 2.26 2.42 126.3 14538 100.0 0.81 7.7 247.7 4.78 4.73 245.3 14205 50.0 1.64 8.5 286.0 6.79 6.98 348.5 13824 33.3 2.35 9 .4 431.6 9.05 9.37 464.5 14072 25.0 3.11 11.0 559.3 11.31 11.65 580.5 13992 20.0 4.88 13.2 692.2 SM III (Sandy silt < 1000 m) = 0.6, 0.91 0.90 46.7 13142 250.0 0.18 2.0 168.7 2.26 2.45 126.3 14214 100.0 0.52 2.5 226.2 4.78 4.79 245.3 14316 50.0 1.89 5.4 263.7 6.79 6.88 348.5 13842 33.3 2.60 7.5 387.5 9.05 9.21 464.5 13358 25.0 2.80 8.8 428.6 11.31 11.61 580.5 13607 20.0 3.11 10.4 536.6
41 Table 2 2 d 10 d 50 d 90 for effluent SM I, SM II, and SM III as a function of f low rates Effluent PM Operating flow rate (L/s) SM I (Sandy silt < 75 m) SM II (Sandy silt < 100m) SM III (Sandy silt < 1000 m) d 10 (m) d 50 (m) d 90 (m) d 10 (m) d 50 (m) d 90 (m) d 10 (m) d 50 (m) d 90 (m) 0.91 0.6 0.8 1.6 1.7 1.2 9.3 6.7 5.0 3 4.7 2.26 0.7 0.8 2.0 1.9 2.0 15.6 7.9 5.9 51.7 4.78 0.8 0.8 2.0 2.2 2.3 16.4 8.3 6.2 54.6 6.79 0.8 0.9 2.2 2.4 2.3 17.3 9.0 8.6 60.4 9.05 0.8 1.0 2.2 2.5 2.5 17.7 10.2 10.6 65.4 11.31 0.8 1.0 2.3 2.5 3.1 17.3 10.6 15.0 74.6
42 Table 2 3 d 10 d 50 d 9 0 for pre deposited PM Pre deposited bed PM Pre deposited PM depth (m) SM I (Sandy silt < 75 m) SM II (Sandy silt < 100m) SM III (Sandy silt < 1000 m) d 10 (m) d 50 (m) d 90 (m) d 10 (m) d 50 (m) d 90 (m) d 10 (m) d 50 (m) d 90 (m) 0.25 1.6 15.0 70.3 1.8 22.0 75.2 7.2 67.0 335.9
43 Table 2 4 The summary of RTD tests as a function of flow rate.Qd is hydraulic design flow rate for baffled HS. Flow beyond 100% Qd over flows inlet weir and is not treated RTD statistics Q (L/s) 2.26 (L/s) 4.78 (L/s) 6.79 (L/s) 9.05 (L/s) 11.31 (L/s) % of Q d 25% 50% 75% 100% 125% 786.1 372.3 261.9 196.5 157.2 t mean (s) 749.1 337.3 222.9 197.0 182.1 t i (s) 45.0 34.0 32.0 28.0 26.0 t p (s) 68.0 52.0 53.0 52.0 46.0 t 10 (s) 81.0 69.0 61.0 49.0 45.0 t 50 (s) 495.0 209.0 159.0 137.0 125.0 t 90 (s) 1587.0 687.0 521.0 407.0 323.0 MDI 19.6 10.0 8.5 8.3 7.2 1/MDI 0.05 0.10 0.12 0.12 0.14 t i /t t 0.06 0.09 0.12 0.14 0.17 t p /t t 0.09 0.14 0.20 0.26 0.29 2 /t t 2 1.36 0.27 0.31 0.27 0.09 t 50 /t t 0.63 0.56 0.61 0.70 0.80 Hazen's N 1.16 1.33 1.50 1.61 1.58
44 Figure 2 1 Plot A) is a plan view schematic of the baffled HS testing facility and the velocity meters placement in the baffled HS is shown. Plot B) illustrates across sectional profile of the HS through the centerline of the unit. Plot C ) PSDs for pre deposited PM. A) Plan view of HS testing syste m B) Section view of Baffled HS Pre deposited PM C) PSDs for pre deposited PM
45 Figure 2 2 Velocity as a function of ADV height at ( A ) location in baffled HS and the mean flow velocity in the baffled HS as a function of flow rate and SOR. (H max = 1.52 m; the distan ce from the bottom of the unit to the invert of the outlet is 1.52m).
46 Figure 2 3 Effluent PSDs for the range of flow rates as a function of flow rate ( A SM I, C SM II, E SM III).Gamma parameters as a function o f surface overflow rate ( B SM I, D SM II, F SM III). Design SOR = 464.5 L/min m 2
47 Figure 2 4 For each non cohesive PSD (SM I at < 75 m, SM II at < 100 m, and SM III at < 1000 m), plots A and C illustrate the relationships between turbidity and surface overflow rate (SOR)and PND (particle number density) as a function of SOR, respectively. For these PSDs, plots B and D illustrate the relationships between turbidity and effluent PM and PND respectively. All R 2 values exceed 0.94 and p = 0.05
48 Figure 2 5 Median washout rate (g/min) and t i linear relationship in plot A and B as a function of SOR has a R 2
49 Figure 2 6 Effluent PM as a function of densimetric Froude number, flow rates, and SOR.R 2 is 0.96 for SM I, 0.96 for SM II, and 0.97 for SM III as a function of Froude number. R 2 is 0.98 for SM I, 0.97 for SM II, and 0.98 for SM III as a function of surface overflow rate
50 Figure 2 7 Effluent mass load range of flow rates as a function of time normalized to m aximum duration ( A SM I, C SM II, E SM III) Washoff model parameters as a function of surface overflow rate ( B SM I, D SM II, F SM III)
51 Figure 2 8 Measured and modeled washout rate in g/minas a functio n of time normalized to maximum duration and volume normalized to 7.6 turnover volume
52 Figure 2 9 number. ( u is shear velocity, and v is kinematic viscosity of water)
53 CHAPTER 3 PHYSICAL AND CFD MOD ELING OF PM SEPARATI ON AND SCOUR IN HYDRODYNAMIC SEPARAT ORS Overview Rainfall runoff (stormwater) transports a mixture of hetero disperse particulate matter (PM) and chemicals that part ition to and from PM (Sheng et al 2008; Lee and Bang 2000). Once entrained in stormwater, PM separation is challenging in an urban environment due to uncontrolled variable of flow rates, spatially distributed loadings and land area or infrastructure const raints to support stormwater control. Therefore, small footprint devices such as hydrodynamic separators (HS) units (U.S. EPA 1999; Rushton 2004) have been deployed specifically as preliminary unit operations for PM where larger scale retention basins tha t provide hydrologic and PM control are not needed (Rushton 2004). A unique feature of HS units is that the particle trajectory can potentially approach that of much larger basins (Field and ling their PM separation is fairly recent whether as a function of steady flow (Sansalone and Pathapati 2009) or for storm events with varying flow rates and hetero disperse particle size distributions (PSD) (Sansalone and Pathapati 2009; Kim and Sansalone 2008). While unit operation models can reproduce PM fate with physically based methods such as combining SOR and PM phenomena, intra event data are critical to couple unsteady flow rates with PSDs, PM granulometry, and partitioning (Sansalone et al. 2010 ). Therefore many unit operations models remain as steady flow approaches. On the other hand, models such as the Stormwater Management Model (SWMM) were designed for the complexity of fully unsteady flow in complex urban systems (Huber and Dickenson 1988 ), with a focus on modeling how the urban interface modified rainfall runoff relationships.
54 In recent years, computational fluid dynamics (CFD) has been applied to environmental engineering. For example, CFD has been used to model the behavior of a screen ed HS under steady flow and constant PSDs (Faram and Harwood 2003). CFD modeling can provide detailed information regarding the unit operation load response, including velocity profiles, particle tracks, pressure gradients, species transport, density and thermal gradients and turbulence (Versteeg and Malalasekera 1995). Similarly regulatory agencies (State of Pennsylvania 2003) require HS units to be physically modeled at steady flow rates and constant PSDs. Over the last several decades studies of PM se paration by HS units have been examined vortex feature of HS units give longer magnitude of particle trajectory than in traditional unit operations. Fenner et al. (1 997, 1998) have suggested that no single dimensionless group can be used in describing, and scaling the PM separation performance. Li et al. (1999) in their study of a partial exfiltration system determined that 2 D numerical model has been applied to simu late variably saturated flow. U.S. EPA (1999) report of HS concluded that HS mostly rely on gravitational force as well as centrifugal forces in order to separate PM. Andoh et al. (2003) studied a hydrodynamic vortex separator (HDVS) and found that CFD si mulation could be applied for the assessment of the efficiency of a HDVS intended for PM separation. Luyckx and Berlamont (2004) considered a vortex separator and their modeling indicated that a vortex separator is based on settling velocity of PM. Cates et al. (2009) monitored 26 storm events with HS unit and overall TSS removal was 59%, while turbidity results show a median EMC reduction of 57%. A baffled HS (BHS), which functions primarily as a sedimentation device, has been tested to assess the perfo rmance of PM removal and a function has been developed linking PM removal to Pclet number ( Pe ) (Wilson et al. 2009).
55 More recently, studies of HS units have examined PM washout (commonly as scour testing) through physical and numerical models. From a mo nitoring campaign that encompassed seven rainfall runoff events, Yu and Stopinski (2001) reported that effluent concentrations exceeded influent PM concentrations for a VHS. Ruston (2004) carried out a monitoring campaign for a SHS loaded by a 54 hectare surface parking area and demonstrated that 51% of the events produced higher effluent than influent PM concentrations as total suspended solid (TSS). Kim et al. (2007) tested a SHS for PM washout (scour testing) and documented that the pre deposited PM he ight in the sump had the most dominant impact on the degree of scour during the duration of each run. A screened HS (SHS), which combines vortex separation, screening and sedimentation, has been tested for PM separation and washout using a pre deposited h rates (Pathapati and Sansalone 2012). Washout from a BHS has been physically modeled as a function of SOR and PSDs for a fixed depth of PM deposits (Cho and Sansalone 2 012). Results indicated that a densimetric Froude number reproduced modeled PM washout as a mass concentration. HS units have been tested physically (Jianghua et al. 2009; Wilson et al. 2009) as well as numerically using CFD (Andoh et al. 2003) to quanti fy PM separation. Cellino and Lemmin (2004) demonstrated that the burst cycle plays an important role in PM suspension mechanics in open channel flow. Gargett et al. (2004) investigated PM transport and re suspension in shallow seas. Research related to PM transport and re suspension has predominantly focused on open channel flow such as clear water and shallow seas, however, modeling of PM washout (re suspension) has not been done much for unit operations in urban watershed area. In this study the
56 complexi ty of PM separation and washout in HS units can be quantified by a physically validated CFD modeling approach Objectives This study applies the methodology to three mechanistically unique HS, each with a unique geometry. The three HS are modeled so as to have the same SOR. This provides basis for comparison that is independent of the geometry. In this study, physical model of PM separation and washout experiments are performed on a full scale BHS, VHS, and SHS subject to the following conditions: (1) PM in jection and pre deposited PM of PSDs known a priori across a range of SOR and (2) differing steady flow regimes representing 10 to 125 % of the design flow rate (Q d ) for each HS units. The physical models are then compared with CFD modeling. This study the n develops a separation function for each HS as a function of particle diameter which can be utilized to predict washout under different pre deposited PM conditions, flow rates and PSDs Methodology Physical Clarification and Re suspension Function Modelin g This study was performed with three different types of commercial HS units: BHS, VHS, and SHS. The hydraulic design flow rate (Q d ) for BHS, VHS, and SHS are 9.1 L/s, 79.3 L/s, and 31.2 L/s, respectively. The Q d for the 3 types of HS unit are provided by the manufacturer. The BHS has a diameter of 1.2 m, a sedimentation chamber height of 1.5 m, surface area of 1.2 m 2 and a volume of 1.8 m 3 The BHS is made for specifically oil and grease and floatables separation and consists of a sedimentation chamber, d rop tee inlet and outlet pipes, and an overflow weir. PM separation is dominated by gravitational separation. Ranges of SOR are from 47.3 to 590.9 L/min m 2 which is between 0.10 and 1.25 of the BHS Q d The VHS has a rectangular chamber, a width of 3.1 m, a depth of 1.2 m, a height of 2.1 m, surface area of 3.7 m 2
57 and volume of a 7.0 m 3 The VHS consists of a swirl chamber and two baffles. PM separation is predominantly due to vortex and gravitational separation. Ranges of SOR are from 128.0 to 1280.2 L/mi n m 2 or, between 0.10 and 1.00 of the VHS Q d The SHS has a diameter of 2.1 m, a screened chamber height of 1.7 m, surface area of 3.6 m 2 and a volume of 6.1 m 3 The SHS consists of static Vortex and gravitational separation are the dominate PM separation mechanisms despite the static screen installed in the unit. Ranges of SOR are from 52.3 to 653.5 L/min m 2 or, between 0.10 and 1.25 of the SHS Q d The influent and pre deposited PM consists of hetero dispersed and non cohesive sandy silt (SM). The mass based PSDs of influent and pre deposited PM are shown in Figure 3 1. The d 50m of SM is 67 Based on Unified Soil Classification System (USCS) standard, this gradation is classified as of each run is chosen for a volumetric turnover of approximately 7.6 (BHS volume is 13.7 m 3 VHS volume is 53.2 m 3 and SHS volume is 46.4 m 3 ). BHS and SHS are analyzed under six different flow rates ranging from 0.10 to 1.25 Q d VHS is analyzed under five different flow rates (0.1 Q d 1.00 Q d ) for separation and washout function For the separation function, the PM injection tank had to be cleaned thoroughly using potable water with a hose and brush such that no PM was present in the slurry tank prior to the run. The pipes leading to the BHS, and SHS units were flushed with potable water to remove any PM or biogenic material in the unit. The PM injection tank was filled with 180 liters of clean water into which the prepared SM gradation was added. The pro grammable logic control (PLC) was set with the target flow rate and the PM injection pump was set with the steady injection rate
58 in Hz. The data logger (CR3000) was compiled to start logging the flow rate every second. Ten duplicate 1 L samples were collec ted for PSD, and PM analysis, (as suspended solid concentration (SSC)). Previous studies have demonstrated that SSC is a representative gravimetric index for hetero disperse gradations that include sand size or coarser PM (Gray et al 2000). SSC analysis is carried out by filtering the entire sample volume of each replicate through a nominal 1.0 m filter (ASTM 2007; Ying and Sansalone 2008). Separation function is measured by SSC removal (%), which is computed as a percentage (by mass) of particles captured by the HS ( M HS ) relative to the particles ( M INF ) added into the system ( 3 1) A mass balance analysis is also conducted after every event to ensure mass conservation based on influent, effluent and recovered mass in HS. A criterion is set by requiring the magnitude of the mass balance error (MBE) to be equal to 10 by dry mass. ( 3 2) Washout function tests are performed at 5 and 6 different flow rates for VHS (0.10 Q d 1.00 Q d ) and BHS (0.10 Q d 1.25 Q d ). The SHS was performed w ith 2 different flow rates (1.00 Q d and 1.25 Q d ). Each re suspension run is conducted by first pluviating the entire surface area of the bottom chamber of the HS with PM to a pre deposited PM. Since HS units store runoff between runoff events, each run is conducted with an HS unit filled with water above an undisturbed pre deposited PM bed. Flow is diverted around the HS until a steady state flow is achieved. Once re suspension flows are directed into the HS, sampling is initiated and discrete1 L samples are manually taken in duplicate at equal sampling intervals, calculated based on flow rate and time. PSDs and SSC are analyzed for washout function. Washout rate is measured by
59 washout mass load (g), which has been computed as a ratio of mass of particles washed out by the HS ( W HS ) to the duration time ( T ) of the system ( 3 3) The intensity of washout is expressed as washout rate (g/min) while the magnitude of washout can be evaluated by effluent mass load (or effluent c oncentration in this specific case because total influent volume was set as a constant per each run) CFD Modeling The Navier Stokes equations can define any single phase fluid flow, but are non linear partial differential equations. It is difficult to sol ve them directly, so we need numerical methods. CFD is a branch of fluid mechanics that uses numerical methods to integrate the Navier Stokes equations in order to solve fluid flow problems. Three types of full scale HS units are modeled in 3D using FLUENT v 6.0. A finite volume method (FVM) is applied to discretize the governing equations into the physical space directly. Modeling in 3 D is less susceptible to the complications which arise from the lack of geometric symmetry, complex static screen geometry vortex flow and gravitational forces on the motion of particles in HS. A k model is suited for separation and washout function (Morin et al 2008; Liang et al 2005; Pathapati and Sansalone 2011). Two equation Reynolds Averaged Navier Stokes (RANS) models are applied to swirling multiphase flows in the HS (Pathapati and Sansalo ne 2009; Garofalo and Sansalone 2011). The standard k model has been applied to turbulent flow model in HS successfully (Pathapati and Sansalone 2009; Garofalo and Sansalone 2011) CFD Governing Equations The governing equations are derived for incompres sible flow. The conservation of mass and momentum are determined using the RANS equations as following
60 ( 3 4) The momentum equations are as follows: x momentum: ( 3 5 ) y momentum: ( 3 6 ) z momentum: ( 3 7 ) In these equation, is fluid density, u, v, and w are Reynolds averaged fluid velocities, g is sum of body forces, and p is the Reynolds averaged pressure. The momentum continuity equation for the x, y and z directions can be obtained by assigning values of u corresponding ly, where u is combining of the x, y, and z velocity vector component, since the hydrodynamics of HS vary as a function of x, y, and z spatial coordinates. The 3 D Navier Stokes equations for a Newtonian fluid are determined by 3 D velocity vector componen ts Turbulence modeling is widely applied the two equation k model. k and equations allow one to determine the turbulent velocity and length scales independently. The transport equations of the standard k model are expressed by the following equations For k and : ( 3 8 ) ( 3 9 )
61 In this expression, is the generation of due to the mean velocity gradients; is the generation of due to buoyancy; are constants; are turbulent Prandtl numbers for and ; are user defined source terms. The values of C C C k used in this model are 1.44, 1.92, 0.09, 1.0 and 1.3 respectively (Launder and Spalding 1974) that eddy viscosity () is a non physical quantity, and is expressed by the following equation ( 3 10) In this expression, is turbulent kinetic energy per unit mass, [L 2 T 2 ] and is the rate of dissipation of turbulent kinetic energy per un it mass, [L 2 T 3 ]. As an assumption of the standard k is taken to be isotropic,. From the standard k velocity profiles, kinetic energy, eddy energy and eddy diffusivity, are genera ted. The standard k through the B HS Particulate Phase Modeling For the both separation and washout function studies, the Euler Lagrangian approach is applied to mod el the particle behavior in the HS. This approach is valid for dilute multiphase flows when PM volume fraction is less than 10% (Elgobashi 1991). A Lagrangian discrete particle model (DPM) is applied to track particles. Due to the extremely dilute nature o f the flow, the DPM assumes there are no particle particle interactions. Particle trajectories are calculated by
62 integrating the particle force balance equation. The Lagrangian DPM is derived from force ttling (Pathapati and Sansalone 2009) ( 3 11) ( 3 12) ( 3 13) ( 3 14) where, u p is a particle velocity, u is fluid velocity; p is pa rticle density, is fluid density; d p is a particle diameter; is viscosity; a 1 a 2 and a 3 are empirical constants that apply to smooth spherical particles as a function of the Reynolds number (Morsi and Alexander 1972); and R e p is a particle Reynolds n umber. The PSD is divided into 18 classes of particles based on a standard sieve. The particle diameter is constant within same class. Particles are tracked for each steady flow rate. The particles that become trapped in the HS are considered to have been removed through HS. The particle removal efficiency is calculated by the following equation ( 3 15) In this expression, is the number of particles that remain in the baffled HS, and N I i s the number of particles injected at the inlet
63 Re suspe nsion and Washout Modeling The method for modeling re suspension from a flat PM bed is performed. An Eulerian Lagrangian approach is used for modeling. First, turbulence modeling is used to obtain a steady state flow field. Following this, a series of pl ane surfaces is created in the PM deposit region of each HS. The interval between these plane surfaces is equal to at least one particle diameter for each particle size tracked. Particles are then placed (velocity = 0, to simulate particles at rest) at gri d points across these surfaces. With mesh size of ranging from 1.5 cm to 2.0 cm, each layer has 37802, 9586, and 45472 points available for tracking particles for BHS, VHS, and SHS units, respectively. Particles are then tracked for tracking lengths obtain ed by a simulated tracer study unbiased tracking length that can be used to predict PM washout. The particles that exit the HS through the outlet are considere d to be re suspended Numerical Procedure The geometries are spatially discretized into 3.2, 5.5, and 2.1 million units for the BHS, VHS, and SHS. As mentioned in previous sections, FVM and a second order upwind scheme are applied in this study. In additio n, the Semi Implicit Method for Pressure Linked Equation (SIMPLE) algorithm (Patankar 1980) is applied.. Through convergence of the numerical process, a numerical method can meet pre designated standards of consistency and stability. Convergence is achieve d when the error residuals fall below the preset convergence criteria (10 4 ) Volumetric Efficiency Calculation CFD models for residence time distribution (RTD) and complete hydrodynamic profiles of HS, including the spatial frequency distribution of velocities. The following RTD indices are reported in this study: t t the theoretical residence time, (defined as t t = V/Q, where V is the volume of the
64 HS unit, and Q is the flow through the system); the volumetric efficiency (VE) (VE := t 10 /t 90 where t x is the time at which x% of the tracer had eluted from the HS) (Levenspiel 1972; Letterman 1999) Results Comparison Physical and CFD Modeling Results of separation and w ashout function for the three HS units are compared with results from experimentally validated CFD models, as shown in shown in Table 3 1 and Figure 3 2 as a function of flow rates. The absolute relative percent difference (RPD) is used to evaluate CFD mod el results with respect to the full scale physical model. Absolute RPD is calculated by the following equation. ( 3 16) Results indicate that the CFD model predictions of PM removal and re suspension rate reproduce the measured dat a with an absolute RPD less than 10% across inflow rate (0.10 Q d to 1.25 Q d ) P hysical Modeling of Separation and washout Function Results for separation and washout VHS, and SHS units as a function of flow rates ranging from 10 to 100% of Q d for each unit. BHS and SHS were also tested at 1.25 Q d The same results for separation and washout function are summarized in Table 3 1, along with SOR. For separation function, PM removal for the tested gradation ranged from 52.3 to 77.6%, at 0.10 to 1.25 for the BHS Q d and for the SHS it ranged from 42.3 to 70.0%, at 0.10 to 1.25 Q d On the other hand, the CFD results showed P M removal for BHS ranging from 51.7 to 74.6%, with RPD less than 5.1% for all flow rates (up to 9.1 L/s in BHS). CFD results for SHS show PM removal ranging from 42.0 to 69.9%, with RPD
65 less than 8.6% (up to 31.2 L/s in SHS). Physical modeling of separatio n function was not conducted for the VHS. However, the CFD results indicate that PM removal has a decreasing trend as flow rate is increasing, ranging from 42.8 to 62.3%, under the specified operating range of flow rate. PM removal varies slightly for the three HS units in the SOR range of 47.3 to 1280.2 L/min m 2 which indicates that the PM removal significantly depends on SOR In order to describe this phenomenon, the fundamental separation mechanisms utilized by BHS, VHS, and SHS are identified. The influ ent flow is directed into the lower chamber of system by the head created at the weir and orifice configuration in BHS. This system incorporates gravitational settling. The cylindrical design of the BHS is to avoid turbulent eddies and dead zones during hi gh flow rates which might re suspension the settled PM from the bottom of the system. The influent flow in VHS is directed to the swirl chamber where it forms, a vortex. Vortex makes coarser PMs to settle down in the swirl chamber. Then, flow goes into sec ond settling chamber containing baffles. The internal hydraulic geometry of SHS is designed such that the entire influent flow enters the volute area by passing through the screen. Therefore, the overall particle separation in a SHS is accomplished by two serial UOPs. This conceptual process flow model also requires that the particle gradation entering the volute area be directly influenced by the separation performance in the screen area Figure 3 2 illustrate s the variation of mean effluent SSC and washou t rate for HS units as a function of flow rates. Both the mean washout mass load and washout rate show a generally linear increase with increasing flow rate for all units. There was higher washout and a wider range in PM washout from the VHS, and SHS than BHS ranging from 1.1 to 25.1 g/min (from 47.3 L/min m 2 to 590.9 L/min m 2 in BHS), from 20.8 to 964.9 g/min (from 128.0 L/min m 2 to 1280.2 L/min m 2 in VHS), and from 2227.3 to 3046.0 g/min (from 52.3 L/min m 2 to 653.5
66 L/min m 2 in BHS), at 100% of pre depo sited PM capacity. In contrast, the CFD results showed washout rates for BHS ranging from 1.0 to 25.5 g/min, with RPD less than 9.1% for all flow rates (up to 590.9 L/min m 2 ). CFD result for VHS indicates that washout rate has an increasing trend while flo w rate increases, ranging from 19.4 to 977.0 g/min (up to 1280.2 L/min m 2 ). Physical modeling for washout function was not conducted at low flow rates for the SHS, however, CFD results for SHS showed PM removal ranged from 263.3 to 3299.9 g/min, with RPD l ess than 8.3% (up to 653.5 L/min m 2 ) under the specified SOR. The washout rates vary slightly for the three units in the SOR range of 47.3 to 1280.2 L/min m 2 which means that the PM removal significantly depends on SOR. Also, comparing the types of HS unit s, it is clear that the geometry of HS has a significant impact on washout, especially in regards to VHS and SHS. SHS has more than 9 times higher washout rate at similar SORs. Overall, BHS had lower washout than VHS, and SHS across operating flow rates. O ne reason for this is that in BHS unit, hydraulic energy is dissipated as the flow hits the drop tee inlet pipe, thereby creating a more quiescent environment. On the other hand, the VHS, and SHS units have no energy dissipation and also have the added was hout due the vortex in the swirl chamber PSD Result A comparison was made among the performance of the HS units (BHS, VHS, and SHS) to evaluate the impact of changing the flow rate on PSD. Figure 3 1 illustrates the variations in the effluent and washed out d 50m across flow rates. CFD results show that the absolute RPDs for each flow rate in all the HS units are less than 10%. Results clearly demonstrate that the effluent d 50m becomes coarser with increasing flow rate for all 3 HS units. The d 50m increase d linearly d d d in SHS for separation function
67 Compared to changes in the d 50m for effluent PSD, the particle size variability is more pronounced as shown in Figure 3 1 which illustrates the d 50m for PSD gradations. Figure 3 1 illustrate s PSDs in the effluent at constant flows from 10 to 125% of design flow rate with SM gradations. As hypothesized, the HS unit discharg ed finer PSDs when the effluent flow had lower flow rates. Except for a few discrepancies, the results indicate that the effluent PSD became increasingly coarser with increasing flow rate. The general trends of PSDs in Figure 3 1 for VHS and SHS did not va ry as much as BHS across the range of flow rates. Effluent PSDs were more significantly influenced by the geometry of the HS Washout PSD data obtained from each experimental run is compared to investigate the performance of each unit as function of flow rates. Results demonstrate that washout PSDs from the BHS consist predominantly of fine particles, indicating that coarse particles are not washed out from the BHS. Washout PSDs are finer at low flow rates than at higher flow rates due to the corresponding ly smaller stream power available to suspend particles. This translates into a relatively stable washout rate at these flow rates, which agrees with the overall system washout rate previously discussed. It appears as if the mildly increasing washout rates possibly vary linearly over the range of flow rates studied. The minimum particle diameter not washout from d A larger fraction of coarse st with the VHS and SHS in Figure 3 1 Plot ( D ) in Figure 3 1 illustrates PSDs in the washout at constant flows from 10% to 100 % of design flow rate at 100% of PM capacity for the SM gradation in VHS. The results from the VHS show that a larger gradation of particles was eluted from the unit with increasing flow rates. Washout PSDs were closer to the SM gradation with increasing flow rates. The minimum d Washout
68 PSDs from SHS ar e illustrated on plot ( F ) in Figure 3 1 from 100 to 125% of Q d at 100% PM capacity for the SM gradation in SHS. The PSD results from the SHS were in between those of BHS and VHS, with an increasing sizes eluted at increasing flow rates. The minimum particl e d Therefore, the washout particle gradations in VHS and SHS were coarser than that in the BHS PM Dynamics The particle trajectories were calculated by a Lagrangian DPM for 3 different HS units for separation and washout function, across a range inflow rates. Figure 3 3 and 3 4 compare trajectories of treated and re suspended discrete particles of selected diameters for three different hydrodynamic separators: BHS, VHS, and SHS. For sep aration function, ( A ), ( B ) and ( C ) in Figure 3 3 washout function in Figure 3 4, ( A ), ( B ) and ( C respectively. Figure 3 3 illustrates the dynamics of a specific PM size for separati on function in HS dependence of particle separation on particle size is clear. It can be observed for the coarse end of the size spectrum that particles are influen ced predominantly by gravitational forces from all three HS units, whereas for the fine end of the size spectrum the suspended particle behavior is largely a function of inertial hydrodynamic forces As illustrated in plot ( A ), and ( B ) of Figure 3 4, 10, a settleable) are significantly washed out from the HS units. However, in plot ( C ) of Figure 3 4, there is no movement of pre deposited PM detected. The BHS showed significantly less PM re suspension than did the VHS or SHS. Comp ared to BHS and SHS, VHS has significant re
69 suspension and washout, because VHS has about 2 times higher Q d than SHS, and 7 times higher Q d than BHS Fluid Velocity Magnitude In order to understand the hydrodynamics within the three HS units, CFD results w ere examined in the form of graphical representations of fluid velocities in the HS by means of fluid pathlines, and fluid velocity frequency distributions. Figure 3 5 provides fluid velocity magnitude frequency distributions in the BHS, inner swirl and o uter chamber of the VHS, and the sump and volute chamber of SHS. As shown in Figure 3 5, the BHS has the lowest fluid mean velocities Figure 3 5 ( A ) depicts mean fluid velocities in the BHS. As may be seen, low magnitude velocities necessary for quiescent conditions and discrete settling dominate this distribution Figure 3 5 can be utilized to determine the possible separation mechanisms in the three HS units Figure 3 5 ( B ) and ( C ) depict the higher magnitude of velocities which result in the swirl chamb er and outer chamber of the VHS, and SHS, respectively. This is due to the presence of a swirling forced vortex in the chamber. Separation within the swirl chamber of the VHS tends to rely upon inertial separation due to the presence of vortics in a small swirl chamber with high SOR (128.0 to 1280.2 L/min m 2 ). It is noted that while the swirl chamber diameter in the VHS is identical to that of the BHS, the Q d of VHS is 2.7 times higher than that of BHS. The correspondingly higher SOR results in higher washo ut rates in the VHS as compared to the BHS. The same comparison gives different results when compared to the SHS, which has a similar range of SOR. The fluid velocity profile shows that SHS has higher velocities in the unit. The inlet configuration and s edimentation chamber configuration also affect the hydraulic behavior of the HS
70 Probability of PM Separation and Washout Modeled PM separation and washout probabilities as a function of PM diameter and flow rate are illustrated in Figure 3 6 via 3 D graph s. Plots ( A ), ( C ) and ( E ) show separation function for BHS, VHS, and SHS. BHS has the most efficient PM removal as a function of flow rates. BHS separates PM diameters larger than 25 m at 0.10 Q d and larger than 150 m at 1.25 Q d VHS separates PM diamet er larger than 150 m at 0.10 Q d and larger than 500 m at 1.00 Q d SHS separates PM diameter larger than 106 m at 0.10 Q d and larger than 250 m at 1.25 Q d The washout function of modeled PM washout probabilities are shown in plots ( B ), ( D ) and ( F ) in Figure 3 6. As shown in plot ( B ), BHS has significantly lower PM washout probability compared to VHS, and SHS. PM diameters larger than 25 m in BHS do not re suspension at all at the highest flow rate, 1.25 Q d In contrast, PM diameters around 300 m w ere re suspended from VHS at 1.00 Q d and PM diameters of 250 m were re suspended from the SHS as well. A gamma distribution function is used to model PM separation and washout. ( 3 17) factor. f(x) is the cumulative gamma distribution. These parameters are shown in Figure 3 7 as a function of flow rates for each eluted PSD. Conceptually, the shape fa ctor may be thought of as uniformity of the eluted PSD as compared to the heterogeneity of the influent PM and pre deposited PM gradation. The shape factor values are decreasing as the flow rate is increased while the scaling factor values are increasing a cross the flow rates. It is observed that a more hetero disperse PSD is eluted from the VHS, with the effluent PM approaching to influent PM and pre deposited PM gradations as the flow rate increases
71 Summary This study examined the PM separation capacity of two types of HS units across a range of influent loading conditions, with the SOR ranging from 47.3 to 590.9 L/min m 2 for BHS, and from 52.3 to 653.5 L/min m 2 for SHS. Measured effluent PM removal ranged from 52.3 to 77.6%, for BHS, and SOR ranged from 47.3 to 590.9 L/min m 2 PM removal for SHS ranged from 42.3 to 70.0% and SOR ranged from 52.3 to 653.5 L/min m 2 Results indicate that SOR has a significant influence on PM removal in HS units. Comparing among effluent PSDs from the three HS units, PSD fro m VHS is consistently coarser across flow rates than in the other two HS units. Secondly, physical modeling of washout from three types of HS units was performed with same SM gradation of 100% pre deposited PM. SOR ranges were same condition as separation function. Washout rates ranged from 1.1 to 25.1 g/min from BHS; from 20.8 to 964.9 g/min from VHS; and from 2227.3 to 3046.0 g/min from SHS. BHS has significantly lower washout rate than other two HS units. The interesting thing is that SHS has the highest washout rate for the same SOR, which ranged from 128.0 to 590.9 L/min m 2 The BHS has significantly lower washout rates than other two HS units. Even though SHS has lower Q d than VHS, the SHS has a significantly higher washout rate. The reason that SHS ha s such a high washout rate is that the geometry of the SHS directs, inflow from the screened area to the pre deposited PM, causing PM re suspension. Three different HS units are successfully modeled in terms of PM behavior with CFD, using FVM, a standard k particles. CFD models are validated for PM concentration, mass and PSDs with less than 10% RPD. Lagrangian particle trajectory results show that VHS has coarsest eluted and washout PM, as wel l as, the highest washout rates. The vortexing inner chamber results in a higher rate of re
72 suspension of finer PM in the SHS. A CFD based probability function is developed for each HS for particle elution as function of flow rate and diameter. Such probab ility functions, combined with any available physical modeling data can provide a reliable method of predicting PM yield from a HS, thereby reducing error in subsequent operations in treatment trains. The volumetric efficiency of the three HS units is illu strated in Figure 3 8. It is noted that the BHS behaves differently from the SHS and VHS. The primary hydraulically distinguishing aspect of the BHS is the absence of a turbulent vortexing region. Volumetric efficiency, while commonly used as an index to d etermine the deviation of a given flow regime from plug flow, can misrepresent the dynamics of particles in a HS. Clearly, for a similar volumetric efficiency, the VHS exhibits more washout than the SHS. In consideration of these results, the use of CFD to quantify treatment and washout from HS units is validated and holds great promise for future studies
73 Table 3 1 Summary of measured and modeled separation and washout function result with RPD. Type of HS Surface overflow rate (L/min m 2 ) Separation funct ion %) RPD (%) Washout function g) RPD (%) Measured Modeled Measured Modeled BHS VHS SHS
74 Figure 3 1 Modeled PSD plots of treated and washout PM from BHS, VHS, and SHS.
75 Figure 3 2 VHS, and SHS units.
76 Figure 3 3 Effluent PM trajectories inside the BHS, VHS, and SHS for particle with diameters p ) is 2.65 g/cm 3
77 Figure 3 4 Washed out PM trajectories inside the BHS, VHS, and SHS for parti cle with p ) is 2.65 g/cm 3
78 Figure 3 5 Fluid velocity magnitude as a function of flow rates in BHS, VHS, and SHS.
79 Figure 3 6. Probability of PM separation and washout by the BHS, VHS, and SHS.
80 Figure 3 7 Gamma parameters as a function of flow rates for BHS, VHS, and SHS (Q d for BHS is 9.1 L/s, Q d for VHS is 79.3 L/s, and Q d for SHS is 31.2 L/s).
81 Figure 3 8 Volumetric efficiency (VE) as a function of flow rates in BHS, VHS, and SHS.
82 CHAPTER 4 STEPWISE STEADY CFD MODELING OF UNSTEADY FLOW AND PM LOADING TO UNIT OPERATIONS Overview Rainfall runoff transports nutrien ts, particulate matter (PM), and organic materials that affect the water volume and quality of a water body (Pathapati and Sansalone 2009; Kim and Sansalone 2008; Wang et al. 2003; Lee and Bang 2000). Designing treatment systems or unit operations (UOP) in urban areas is further challenged by unsteady hydrologic loads, complexity of PM and chemical constituents (Liu et al. 2008). Traditional treatment options such as detention/retention basins are often difficult to implement in urban areas, due to lack of available land area. Typically, the performance of UOPs has been assessed by physical modeling. Wilson et al. (2009) assessed PM removal in baffled hydrodynamic separator (BHS) by pilot scale testing. Hunt et al. (2008) tested bioretention pollutant remov al and peak flow mitigation. Physical modeling, while valuable for global performance assessment, is not easily amenable to retrofitting and iterative design; physical modeling at various scales for UOPs is often limited by available funds and infrastruct ure. In recent years, a coupled physical and numerical modeling approach has proved effective in describing the separation mechanisms of UOPs. Computational fluid dynamics (CFD) has been applied for both controlled steady flows and particle size distributi ons and fully transient rainfall runoff events. Lee et al. (2010) physically tested a vortex hydrodynamic separator (VHS) under steady flow conditions and applied CFD modeling to predict PM removal for particle size and flow rates. Dickenson and Sansalone (2009) examined PM discretization requirements with a CFD model for selected levels of granulometric size hetero dispersivity. Pathapati and Sansalone (2009) report that a CFD model was able to accurately reproduce physical model data for PM separation for a screened HS for steady flow
83 rates. This was then extended to unsteady flows and PM loadings (Sansalone and Pathapati 2009) comparing storm event based captured and effluent PM from a monitored empty bed SHS to unsteady CFD model predictions of PM fat e. Garofalo and Sansalone (2011) demonstrated that CFD model accuracy for simulating elution of hetero disperse PM under transient hydraulic loadings is dependent on time resolution of the flow field, spatial discretization of the computational domain, and the PSD size discretization CFD can predict fluid flow, mass transfer, chemical reactions, and related phenomena by solving governing fluid equations using numerical methods. In urban stormwater, CFD has enhanced the modeling of PM separation for transie nt flows (Sansalone and Pathapati 2009; Garofalo and Sansalone 2011) and heterodisperse particle size distributions (Dickenson and Sansalone 2009) as well as re entrainment of PM by washout mechanisms (Pathapati and Sansalone 2012). However, a 3 D numerica l model needs longer computational time to simulate the transient hydrodynamics with variable PSDs than steady state flow (Pathapati and Sansalone 2011). For example, the unsteady computational time for the hydrodynamic separator (HS) is approximately 30. 5 days using a workstation equipped with dual quad core 2.6 GHz processors and 32 GB of random access memory (Pathapati and Sansalone 2011). To minimize computational time, a stepwise steady flow model is suggested as a possible solution. Stepwise steady CFD modeling is a series of discretized steady flow steps of rainfall runoff events. Identifying numbers of flow steps needed for economical computational time is crucial. In this manuscript, the PM separation is characterized for stepwise steady flow, fo r three commonly used UOPs : BHS, SHS and a volumetric clarifying filter (VCF) unit The main purpose of this study is to demonstrate modeling of unsteady flow utilizing a series of discretized Stepwise steady flow rainfall runoff model.
84 Objectives There p rimary objective are addressed in this manuscript. The first objective is to characterize the PM separation by three HS units across 4 discrete storm events from an urban impervious surface area. The second objective is to develop a CFD model for stepwise step flow rates and influent PSDs for three UOPs The third objective is to compare the results of the numerical model to paired experimental results and to quantify the differences in the two approaches. Methodology Watershed and Three Hydrodynamic Separ ator Configurations The first area of interest is the University of Florida Reitz Union paved surface parking facility catchment in Gainesville, FL. The schematic plan view of catchment with BHS is shown in Figure 4 1( A ) Considering the minimal slope valu es characterizing the monitored area, particularly the paved surface parking facility area (approximate E W 3% and N S 1.5% slope), and the contributing area has the potential to be influenced by rainfall intensity and wind direction. Depending on the sto rm event the watershed area is approximately 500 m 2 The second area of interest is directly adjacent to Interstate 10 and City Park Lake in urban Baton Rouge, Louisiana. Rainfall runoff directed to SHS and VCF system. A schematic plan view of the catchme nt with SHS and VCF is shown in Figure 4 1( A ) The drainage system was designed to intercept the lateral pavement sheet flow from the concrete paved watershed. The watershed area is approximately 1088 m 2 Isometric views of three different UOPs are shown in Figure 4 2: BHS, SHS, and VCF. The design hydraulic operating volumetric flow rates (Q d ) are 9.1 L/s for BHS, 15.9 L/s for SHS, and 5.7 L/s for VCF based on the manufactures specifications. The BHS consists of a bypass weir and drop tee for directing f low into the sedimentation chamber. The SHS consists of static
85 cross flow cylindrical screen (2.4 mm screen opening diameter). The VCF consists of five gravity driven radial cartridge filters (RCF) in a vault. Each RCF contains approximately 49.8 kg of mon o disperse d Al Ox coated granular media (AOCM) p with a median diameter of 3.5 mm 0.8 mm and specific gravity of 2.35 0.01. The total porosity of each RCF is 0.71 0.041. The five cartridges with filled (AOCM) P are housed in a 1.17 m by 2 12 m detentio n vault structure. Detailed description of these three UOPs can be found elsewhere (Cho and Sansalone 2012; Pathapati and Sansalone 2011; Sansalone et al. 2009). These UOPs operate primarily based on gravitational PM separation, however, the SHS also perfo rms size separation screening and VCF performs filtration through (AOCM) P cartridge. Physical Modeling Methodology M onitoring station design was driven by monitoring procedures related to the physical processes investigated; in particular representative mo nitoring that requires manual sampling. Rainfall runoff PM is monitored across the inlet and outlet of the treatment unit during rainfall events. Two different watershed areas were investigated: carpark loading in Gainesville, FL, and highway loading in Ba ton Rouge, LA. Table 4 1 shows hydrologic indices across storm events for BHS, SHS, and VCF. D etailed results regarding the watershed, hydrology, pollutant loads and water chemistry for SHS and VCF are available elsewhere ( Sansalone et al. 2009: Kim and Sa nsalone 2008 ) Four events for the each UOPs were monitored and modeled. Table 4 1 and 4 2 show the hydrologic indices, influent/effluent PM data and PSD data. 10 to 18 samples were manually taken from influent and effluent drop boxes depending on duration and rainfall intensity of each event. The flow rate at the time of sampling, and throughout the storm duration, was recorded automatically by the flow meter The mean numbers of sample sets are 13, 13 and 11 for the VCF, PC and HS respectively. After col lecting all the rainfall runoff samples from the events, the samples were taken
86 back to the laboratory for immediate analyses. The efficiency of the system was assessed using laboratory analyses for s uspended sediment concentration (SSC) (ASTM 1999), PSD, and mass balance. PSD was determined using a laser diffraction particle analyzer (Malvern Instruments: Hydro 2000G) in a batch mode analysis. SSC analysis was performed to quantify particle concentration for each effluent sample collected from each run and calculate the effluent mass load for each storm event. Laboratory analyses were conducted for the replicate influent and effluent samples consisting of the individual replicate (A or B) samples. Analyses are conducted within a few hours of the run comple tion to maintain the same conditions (temperature) and minimize flocculation. In order to define unsteadiness of storm events, a unsteadiness parameter ( ) is defined in this study. The calculation of is as follo ws: ( 4 1 ) In this expression dV represents d erivative of inflow volume [L 3 ] ; d t represents d erivative of time [T] ; and Q med represents m edian flow rate of storm event [L 3 /T] CFD Modeling Methodology CFD is a branch of fluid mechanics that uses numerical methods to in tegrate the Navier Stokes equations in order to solve fluid flow problems. Three types of full scale UOPs are modeled in 3 D using FLUENT v 6.0. A finite volume method (FVM) is applied to discretize the governing equations into the physical space directly. Modeling in 3 D is less susceptible to the complications which arise from the lack of geometric symmetry, complex static screen geometry, vortex flow and gravitational forces on the motion of particles in UOP
87 The governing equations are derived for inco mpressible flow. The conservation of mass and momentum are determined using the RANS equations as following ( 4 2 ) The momentum equations are as follows: x momentum: ( 4 3 ) y momentum: ( 4 4 ) z momentum: ( 4 5 ) In these equation, is fluid density, u, v, and w are Reynolds averaged fluid velocities, g is sum of body forces, and p is the Reynolds averaged pressure. The momentum continuity equation for the x, y a nd z directions can be obtained by assigning values of u correspondingly, where u is combining of the x, y, and z velocity vector component, since the hydrodynamics of HS vary as a function of x, y, and z spatial coordinates. The 3 D Navier Stokes equation s for a Newtonian fluid are determined by 3 D velocity vector components. A k suspension function (Morin et al 2008; Liang et al 2005; Pathapati and Sansalone 2011). Two equation Reynolds Averaged Navier Stokes (RANS) models are applied to swirling multiphase flows in the HS (Pathapati and Sansalone 2009; Garofalo and Sansalone 2011). The standard k model has been applied to turbulent flow model in HS successfully (Pathapati and Sansalone 2009; Garofalo and Sansalone 2011).
88 Turbulence modeling is widely applied using the two equation k model. k and equations allow one to determine the turbulent velocity and length scales independently. The transport equations of the standard k model are expressed by the following equations. For k and : ( 4 6 ) ( 4 7 ) In this expression, is the generation of due to the mean velocity gradients; are constants; are turbulent Prandtl numbers for and The values of C C C k used in this model are 1.44, 1.92, 0.09, 1.0 and 1.3 respectively (Launder and Spalding that eddy viscosity () is a non physical quantity, and is expressed by the following equation. ( 4 8 ) In this expression, is turbulent kinetic energy per unit mass, [L 2 T 2 ] and is the rate of dissipation of turbulent kinetic energy per unit mass, [L 2 T 3 ]. Per standard application of k assumed to be isotropic. I nformation regarding flow field and turbulent field, such as velocity profiles, kinetic energy, eddy energy and eddy diffusivity, are generated from standard k model equations The standard k standar d wall functions is an effective approach to modeling flow through the B HS.
89 Particulate Phase Modeling T he Euler Lagrangian approach is applied to model the particle behavior in the UOPs This approach is valid for dilute multiphase flows when PM volume fr action is less than 10% (Elgobashi 1991). A Lagrangian discrete particle model (DPM) is applied to track particles. Due to the extremely dilute nature of the flow, the DPM assumes there are no particle particle interactions. Particle trajectories are calcu lated by integrating the particle force balance equation. settling (Pathapati and Sansalone 2009). ( 4 9 ) ( 4 10 ) ( 4 11 ) ( 4 12 ) where, u p is a particle velocity, u is fluid velocity; p is particle density, is fluid density; d p is a particle diameter; is viscosity; a 1 a 2 and a 3 are empirical constants that apply to smooth spherical particles as a function of the Reynolds number (Morsi and Alexander 1972); and R e p is a particle Reynolds number. The PSD is divided into 22 classes of particles based on a standard sieve. The particle di ameter is constant within same class. Particles are tracked for each steady flow rate. The particles that become trapped in the HS are considered to have been removed through HS. The particle removal efficiency is calculated by the following equation.
90 ( 4 1 3 ) In this expression, is the number of particles that remain in the baffled HS, and is the number of particles injected at the inlet. Modeling of Static Screen and Cartridge The static screen and cartridge filter are modeled as a porous pe rforated plate with the addition of a momentum source term to the standard fluid flow equations ( Pathapati and Sansalone 2009 ) The source term is composed of two parts: a viscous loss term and an inertial loss term. For simple homogeneous media and surfic ial filter the sink term equation is as follows: ( 4 1 4 ) In this expression, S i is the source term for the i th momentum equation, is permeability, C 2 i is the inertial resistance factor, v i is the velocity in the i th mo mentum equation, and v i is the velocity in a computational cell. The momentum sink contributes to the pressure gradient in the porous computational cell, creating a pressure drop that is proportional to the fluid velocity in the cell. Flow in porous media has traditionally been modeled analytically using comparison s to pipe/conduit flow and specifying analogous parameters such as the hydraulic diameter and roughness coefficient. Laminar flow (Re < 10) through porous media has been successfully modeled by ap plying Darcian type equations. Models such as Blake Plummer and Carman Kozeny equations were developed to account for transitional flow regimes. These models were extended by Ergun (1952) to account for turbulent flow. The Ergun equation for packed beds ap plies to flow regimes from laminar to turbulent and is expressed by the following equation.
91 ( 4 1 5 ) d m bed, v is the superficial velocity through the packed bed and is fluid density. The permeability ( ) and the inertial resistance coefficient ( C 2 ) can be expressed as follows. ( 4 1 6 ) ( 4 1 7 ) CFD Parameters CFD parameters for three UOPs are shown in Table 4 3. The v olume fraction of secondary phase (particle) is less than 0.1% in Table 4 3 and can be applied to a Lagrangian particle tracking approach. Second order tetrahedrons element type is utilized to discretizing the computational domain (Qi and Lin 200 6). The s tandard k was applied in this study. A s econdary order upwind scheme was utilized to solve for flow parameters (Barth and Jespersen 1989; Pathapati and Sansalone 2009). The operating pressure was 14.7 psi (atmospheric). DPM particle density is 2.65 g/cm 3 and particle diameters range from 1 micron to 9750 micron. DPM iterative convergence limit was set at 10 3 and applied to continuity, momentum, turbulent kin etic energy, and dissipation rate (Ranade 2002). Stepwise Step Modeling Removal and PM Separation The number of stepwise steady steps is determined based on creating a cumulative
92 distribution function (CDF) of flow rates entering the UOPs. The first step was to develop a CDF for flow for each storm event. Following this step, the influent samples and effluent samples are plotted on the CDF. This provides an idea of the number of discretization steps needed. The number of stepwise steady steps required to a ccurately represent the transient PM clarification varied predominantly as a function of the fluctuation of influent flow rates monitored by correlating volumes, peak flow rates, elapsed time and sampling time. Influent PSD was input into the model at t he appropriate flow rate for the stepwise steady model. The influent particles were tracked through the UOP after tracking for a given tracking the UOP and ite ratively increasing tracking length until no more tracer particles would exit the system. Typically, this was done with a 1% particle balance error. Since rainfall runoff flow is discretized with CDF throughout entire storm event duration, PM separation can be calculated at the each stepwise steady flow rate. PM separation at the i th stepwise PM(i) is calculated as follows. ( 4 1 8 ) N i is the number of particles injected in the influent and N UOP is the number of particles that incomplete in the unit operation after the specified tracking length. Overall PM separation, PM is calculated as follows. ( 4 1 9 ) In this expression, q(0) and q(t) are the influent flow rat es, n(t 1 ) and n(t n ) are influent samples at time t 1 and t n respectively. M Inf is the influent PM mass.
93 Result Event Hydrology Indices Event based hydrological indices including PDH, d rain t r unoff as well as V in and Q p and Q med ,, were monitored and record ed for a total of 4 storm events for each UOPs as shown in Table 4 1. Observed events varied in duration from 31 to 415 min, total rainfall runoff volume ranged from 594 to 48306 L, median flow rate ranged from 0.1 to 2.4 L/s, peak flow rate ranged from 0. 6 to 25.3 L/s, and rainfall depths ranged from 2.5 to 71.4 mm. The watersheds had the dominant an thropogenic activity of traffic resulting from the highway and parking lot. The values of for each hydrographs are summarized in Table 4 1. Unsteadiness parameter equation was modified from Garofalo and S ansalone (2011). The normalized value of time with respect to the peak time was not applied to this study due to the difference of peak time among all storm events. Based on measured data of Pathapati and Sansalone (2009) as well as Garofalo and Sansalone (2011), the of actual hydrographs from a watershed are typically ranging from 0 to 6; considered quasi steady, 6 to 14; considered unsteady, and 14 or greater; considered highly unsteady. The values of for each hydrographs are summarized in Table 4 1 The values of for BHS ranged from 2.99 to 15.73 These values indicate that the storm events for B HS have from quasi steady to highly unsteady influent flows. The values of for SHS ranged from 1.42 to 8.24 These values indicate that the storm events for S HS have from quasi steady to unsteady influent flows. The values of for VCF ranged from 3.59 to 17.46 The quasi steady value ranged from 0 to 6 is comparable to 3.59 and 4.25 for the physical model hy draulic and PM loading (Garofalo and Sansalone 2011) Also, unsteadiness of two storm events for VCF, which ranged from 11.89 to 17.46, are considered unsteady to highly unsteady
94 (4 20 ) parameters were estimated by minimizing the sum of squared errors (SSE), resulting in maximizing the coefficient of determination between the measured a nd modeled data. Below, F (x) is the cumulative gamma distribution and x represents flow rate in Table 4 1 (4 21 ) Influent and effluent flow rates are heterodisperse throughout all storm events for the three UOPs BHS and SHS have s ame influent and effluent flow rates. Table 4 2 summarizes the influent PM mass of each storm event, modeled effluent PM mass comparison with measured effluent PM mass, the median influent and effluent PSDs across runoff events monitored for the BHS, SHS, and VCF as a mass based cumulative PSD. C umulative PSDs were examined and results were described by an optimized cumulative gamma distribution function summarized in Table 4 2. The runoff median PM diameter (d 50m to Results indicated that BHS has same shape and scaling factor from both influent and effluent. The influent and effluent flow rate s are equal in the BHS since it does not have a screen or cartridge in the unit. The drop tee section split s the inflow to the sediment chamber providing dissipation of turbulence and flow velocity. e screen area did not result in change in a difference in infl uent and effluent flow rate s T he dominant mechanism of PM separation of SHS is gravitational al though the inertial separation occurs at low flow rates.
95 VCF consists of a sedimentation chamber with five RCF. The RCF results in dissipation of turbulent energy and flow velo city (Pathapati 2009). Th e RCF generates 5 mm median head loss (Sansalone et al. 2009) resulting in a lag between inflow and outflow from the VCF unit. Gamma parameter values are different between inflow and outflow. The mechanisms of PM separation for VCF are gravitational separation and filtration. Filtration with (AOACM) P increases PM removal. Probabilities of PM Separation by the BHS, SHS, and VCF Modeled stepwise PM separation as a function of PM diameter and flow rate is illustrated in Figure 4 4 via 3 D surface plots. The upper limit flow rate was chosen as the maximum monitored flow rate among UOPs The maximum flow rate was approximately 25 L/s from April 26, 2009 storm event. BHS and VCF separated all PM larger than 300 m across the entire range of flow rates up to 25 L/s. BHS displays a fairly consistent probability of PM separation throughout the entire flow rate range (0 to 25 L/s). One advantage of the BHS The prevention of separated PM from being re suspended due to the internal bypass and d rop tee configuration. In contrast, the SHS tends to separate coarser particles ( > 800 m ) across the entire range of flow rates up to 25 L/s At the highest flow rate (25 L/s), all three UOPs separated PM larger than 2000 m Table 4 2 summarizes PM sep arated by each UOPs modeled with a cumulative gamma distribution function. In this section, the gamma probability density function represents PM separation as a function of flow rate, Q, and particle diameter, symbolized as x in equation 4 1 7 Also, F(x) is the cumulative gamma distribution in the equation. These parameters are shown in Figure 4 3 as a function of flow rates for each eluted PSD. Conceptually, the shape factor may be thought of as uniformity of the eluted PSD as compared to the heterogeneity of the influent PM and pre
96 deposited PM gradation. These parameter trends illustrate that BHS and VCF are able to separate a large range of PM throughout entire flow rate ranges. In contrast, SHS depends on coarser PM removed according to widely ranged Stepwise Steps Comparison to Measured Data The s tepwise steady flow CFD model was performed with each discretized steady flow level. The discretized steady flow steps were determined using a cumulative density function (CDF) across flow rates during the unsteady storm events, illustrated in Figure 4 4 to 4 6 The CDF s show the frequency of flow rates entering the UOPs Results of stepwise steady flow modeling for the three UOPs are compared with results from experimentally validated CFD models, as shown in shown in Figure 4 4 to 4 6 The CFD model results of PM mass are compared to measured data throughout 4 storm events for the three UOPs in Table 4 2. T here is a significant difference in PM removal among the three UOPs especially between SHS and the tw o other UOPs (BHS and VCF). BHS has 64.7 to 95.1% PM removal; SHS has 38.3 to 62.0% PM removal; and VCF has 84.0 to 97.1% PM removal. Results shows that PM removal is a significantly less for the SHS while VCF has the highest PM removal The absolute rela tive percent difference (RPD) is used to evaluate the CFD model results with respect to the full scale physical model. RPDs are shown in each Figure for the three UOPs Absolute RPD is calculated by the following equation. ( 4 22 ) Results from the lowest flow rate to the highest flow rates indicate that the stepwise steady CFD model predictions of PM removal reproduce the measured data with an absolute RPD less than 10% across 12 total events The separated PM mass and RPD between measured and modeled PM separated in the three UOPs is provided for each event in Table 4 2
9 7 Hydrographs are shown in Figure 4 4 to 4 6 and stepwise steady models to monitored effluent PM mass cumulatively are shown utilizing rainfall runoff PSDs. In ad ditional analys e s the discretization steps for the stepwise steady flow model are further increased by utilizing flow rates and event duration time between monitoring points of PSDs. The mean number of steps is is 35 for BHS 38 for SHS and 38 VCF. Accord ing to numbers of stepwise steady steps, BHS had mild storms than other two UOPs since the more steps needed depending on fluctuation of runoff flow rates. The solid and dashed lines represent effluent PM mass predicted by the stepwise steady and monitored PM mass respectively All three figures are representative of monitored storm event PM loads. These results illustrate that the stepwise steady model for the three UOPs is representative of monitored PM across each events. The error at each step is calc ulated relative to the monitored PM separation for each UOP. Figure 4 7 depicts overall stepwise steady CFD modeling error as a function of integrating over an increasing number of stepwise steps predictions for PM separation. The RPDs for each UOP has a s teep decreasing trend in error with increasing discretization of flow rates. Due to the inherent effect of the filters on dissipating turbulent energy, the effect of changing flow rates on the PM separation response is lower for the VCF compared to the BH S and SHS. This is evidenced by the lower RPD for the VCF compared to the BHS and SHS. Stepwise step CFD modeling for three UOPs reproduce measured PM within 10% RPD ranges as well as significantly lower computational time. According to previous studies (P athapati and Sansalone 2011), the computational times of the unsteady flow CFD simulations for HS is a factor of approximately 16 times greater than for the stepwise steady CFD model. S tepwise steady flow modeling was used to successfully
98 reproduce unstead y flow results and PM loadings in this study S tepwise steady steps modeling will therefore be utilized for unsteady HS modeling with less computational times in the future. Summary Unsteady flow events and PM separations with three UOPs in urban source ar ea water sheds were examined, providing variable volumetric control. In this study, four source area rainfall runoff events are monitored and treated real time by the three UOPs The BHS unit is installed in a carpark area in Gainesville, FL, and the other two UOPs are installed in a high way area in Baton Rouge, LA. The main source constituent from both watershed areas is primar il y influenced by traffic. The main parameters in the CFD model that can be "tuned" are standard Morsi and Alexander drag coeffici ents and the wall functions in the k epsilon turbulence model. Stepwise steady modeling was able to successfully reproduce experimental data s utilizing standard values for the drag coefficients and wall functions. No calibration or "tuning" was required. The main purpose of this study was to model unsteady flow PM separation across a rainfall runoff event by utilizing CFD model predictions at discrete steady flow rates. Stepwise steady CFD modeling was conducted without any tunable parameters. The only thi ng that can be changed as CFD parameters was the standard Morsi and Alexander drag coefficient; however, this drag coefficient was not tuned. The standard Morsi and Alexander drag coefficient was given in Table SI 1 (Morsi and Alexander (1972). Stepwise st eady modeling showed that if PM separation data as a function of particle diameter is available at steady flow rates, this can be integrated at a fine degree of discretization to predict PM separation for unsteady rainfall runoff events. Once a steady flow CFD database is established across a wide range of flow rates, this can be used to predict fully transient rainfall
99 runoff events. This results in faster modeling time for predicting unsteady behavior, compared to transient modeling for an entire storm. C FD modeling of PM separation is based on the assumption of discrete particle settling. In order to check if this assumption holds true across different flow configurations, validation of models with experimental data is advised. Physical modeling of UOPs w ith careful QA/QC is required to be carried at at least representative flow rates, with the required PSDs. This study accomplished that each stepwise steady flow results compared to monitored data of PM separation with RPD less than 10% for three UOPs St epwise steady steps are determined by based on CDF. The n umber of stepwise steps depends on volume, storm duration time, and strength of storm events. Comparing PM mass from each three UOPs the modeled data consistently match with the monitored data altho ugh runoff flow rates are randomly distributed through storm duration time. Results indicate that the stepwise steady flow model effectively predicts the monitored storm event data in UOPs. Fully unsteady CFD modeling can provide more accurate results tha n other modeling results; however, a stepwise steady flow CFD model gives less than 10% error for monitored unsteady storm events with more than 35 flow steps All three devices show a steep decrease in error with increasing steady steps in Figure 4 7 The error is hypothesized to theoretically reach zero as the number of steady flow steps approaches infinity. Overall we see that the VCF has the lowest range of error. In general, the effect of unsteady flow on the VCF is mitigated by the presence of filters which act like a resistance, and reduce the impact of smaller fluctuations in flow rates. This study concludes that a stepwise steady approach is able to successfully model PM separation, across 3 mechanistically different UOPs, and is validated with r epresentative monitored data. The stepwise steady step model reproduced PM separation with significantly
100 more efficient computational time than a fully unsteady CFD model. This stepwise steady CFD modeling approach can reduce simulation time approximately 14 18 times when compared to a fully transient analysis. In consideration of these results, the use of stepwise steady CFD modeling to quantify PM separation from UOP s is validated with unsteady hydrologic indices and infers economic with reducing computa tional time as well as hold ing a great promise for future implication
101 Table 4 1 Hydrologic indices across storm events for BHS, SHS, and VCF Type of UO Storm event V in (L) PDH (hr) d rain (mm) Q med (L/s) Q peak (L/s) t runoff (min) Gamma distribution parameters Influent hydrograph Effluent hydrograph BHS (2010) 29 June 7504 192 20.6 1.5 17.6 15.73 40 0.74 3.26 0.74 3.26 11 July 2362 261 11.7 0.3 5.2 9.47 42 0.34 23.37 0.34 23.37 28 July 8316 36 19.1 1.2 16.0 7.87 50 0.36 7.82 0.36 7.82 14 August 594 28 2.5 0.3 2.5 2.99 31 0.61 0.32 0.61 0.32 SHS (2004 June 30, 2005) 14 March 28464 204 26.2 0.7 6.4 2.66 415 0.74 1.46 0.74 1.46 20 August 17687 26 17.3 0.3 17.5 7.82 60 0.22 18.56 0.22 18.56 14 October 1672 84 2.6 0.1 0.6 1.42 201 0.40 0.41 0.40 0.41 30 June 5856 143 19 .1 0.3 9.4 8.24 57 0.46 7.23 0.46 7.23 VCF (2006) 21 April 2927 927 4.1 0.1 13.4 17.46 49 0.12 3.16 0.87 0.26 29 April 48306 84 71.4 2.4 25.3 11.89 177 0.69 6.45 0.88 4.35 04 July 2779 352 3.8 0.2 6.2 3.59 68 0.31 2.40 0.46 1.42 05 Jul y 3838 25 5.6 0.2 7.9 4.25 94 0.25 2.77 0.21 3.26 V in PDH, d rain Q med Q peak t urnoff represent volume of rainfall runoff, previous dry hours, event duration, median flow rate, peak flow rate, rainfall runoff duration, respectively.
102 Table 4 2 CFD mod el comparisons to measured data across storm events for BHS, SHS, and VCF Type of UO Storm event Influent PM Effluent PM d 50m Gamm distribution parameters Influent PM Effluent PM Measured (g) Measured (g) Modeled (g) Influent Effluent BHS (2010) 29 June 6997.6 1285.1 1218.8 72 49 1.18 78.17 1.26 50.89 11 July 1001.6 195.4 202.2 113 36 0.90 196.99 1.13 45.17 28 July 2404.5 848.8 769.9 110 48 0.90 193.13 0.99 98.09 14 August 195.1 9.6 10.3 182 43 0.85 336 .20 0.79 99.73 SHS (2004 June 30, 2005) 14 March 4949.1 2539.3 2441.9 43 20 0.70 137.02 1.42 18.41 20 August 10591.0 4022.4 3774.7 300 52 0.44 2740.89 1.06 69.92 14 October 541.3 266.5 246.1 45 13 0.63 124.07 1.88 8.83 30 June 4043.8 2 494.5 2328.8 69 39 0.73 212.51 1.28 39.94 VCF (20 06 ) 21 April 4161.7 119.7 123.8 15 5 0.70 34.93 1.26 5.27 29 April 10466.2 1678.7 1622.6 99 24 0.48 452.09 0.65 71.68 04 July 831.8 100.8 91.6 26 21 0.67 63.14 1.23 27.34 05 July 1065.1 13 8.7 148.6 18 18 0.68 56.14 1.51 16.64
103 Table 4 3. CFD parameters for BHS, SHS, and VCF. CFD parameters BHS SHS VCF Mesh size of geometry 3.2e+06 2.1 e+06 5.2e+06 Element type Second order tetrahedrons Turbulence m odel Standard Wall fu nctions Realizable Solution method SIMPLE, segregated Momentum Second order upwind Turbulent kinetic energy Second order upwind Turbulent dissipation rate Second order upwind Operating pressure (kPa) 101.35 Operating temperature (K) 288.16 Volumetric design flow rate (L/s) 9.1 15.9 5.7 Volume fraction of secondary phase (%) 0.01 0.07 0.01 0.05 0.01 0.03 DPM particle density (g/cm 3 ) 2.65 DPM particle diameter (m) 1 9750 D PM drag coefficients Morsi and Alexand er DPM boundary conditions Walls polynomial Outlets Conver gence limits Continuity 10 3 Momentum 10 3 Turbulent kinetic energy 10 3 Dissipation rate 10 3 SIMPLE represents s emi implicit method for pressure linked equ ation (Baffled hydrodynamic separator BHS, screened hydrodynamic separator SHS, volumetric clarify filtration VCF)
104 Figure 4 1 Plot A is a plan view schematic of a BHS testing watershed in Gainesville, FL. Plot B is a plan view schematic of SHS and MFS testing watershed in Baton Rouge, LA
105 Figure 4 2 Isometric views of the geometries of A ) baffled hydrodynamic separator (BHS), B ) screened hydrodynamic separator (SHS), and C ) volumetric clarifying filtration system (VCF)
106 Figure 4 3 Probability of PM separation by the baffled hydrodynamic separator ( BHS ) screened hydrodynamic separator ( SHS ) and volumetric clarify filtration ( VCF ) and represent shape factor and scaling factor
107 Figure 4 4 Plot A ) is a cumulative distribution function (CDF) for the range of rainfall runoff flow rate (L/s) in baffled hydrodynamic separator ( BHS ) Plot B ) is flow rates (L/s) and effluent PM mass (g) as a function of elapsed time in BHS (Number of flow steps = 37 ) V in Q med Q peak and RPD represent unsteadiness parameter, median flow rate, peak flow rate, and relative percentage difference.
108 Figure 4 5 Plot A ) i s a cumulative distribution function (CDF) for the range of rainfall runoff flow rate (L/s) in screened hydrodynamic separator ( SHS ) Plot B ) is flow rates (L/s) and effluent PM mass (g) as a function of elapsed time in SHS (Number of flow steps = 40 ) V in Q med Q peak and RPD represent unsteadiness parameter, median flow rate, peak flow rate, and relative percentage difference.
109 Figure 4 6. Plot A ) is a cumulative distribution function (CDF) f or the range of rainfall runoff flow rate (L/s) in volumetric clarify filtration ( VCF ) Plot B ) is flow rates (L/s) and effluent PM mass (g) as a function of elapsed time in VCF (Number of flow steps = 49 ) V in Q med Q peak and RPD represent unsteadiness parameter, median flow rate, peak flow rate, and relative percentage difference.
110 Figure 4 7. Mean and variation of the stepwise steady model absolute relative percentage difference (RPD) with increasing number of monitoring points for BHS, SHS, and VCF. The lower right quartile box plot is the variation of absolute RPDs across the number of steps.
111 CHAPTER 5 REMOVAL AND PARTITIONING OF NITROGEN AND PHOSPHORUS OF NUTRIENTS IN HYDRODYNAMIC SEPA RATOR ON URBAN RAINFALL RUNOFF PARTICULATE MATTER GENERATED FROM IMPERVIOUS SURFACE CARPARK Overview Stormwater runoff can be a significant source of Nitrogen (N) and Phosphorus (P) leading to eutrophication (Field and Sullivan, 2002; Heaney et al., 1999; Lee and Bang, 2000; Novotny and Witte, 1997). Sources of N (P) include non point sources in non urban areas such as biogenic materials and fertilizers (Turner et al., 1999; Zhang and Jrgense 2004) as well as urban pavements (Wang et al. 2003). Partitioni ng controls the transport and cyclic dynamics of P in land/water ecosystems and is key to understanding impacts on aquatic ecosystems (Brezonik and Stadelmann 2002). The export of N in stormwater runoff poses several threats to environmental and human heal th and consumes a large share of public resources (Parker et al.2000, NRC 2001). Nitrogen loading is primarily a function of processes that affect concentrations rather than of the geometry of the catchment which conveys precipitation into runoff (Lewis an d Grimm 2007). In addition to the partitioning of total nitrogen (TN) and total phosphorus (TP) into the runoff. Partitioning, mobility and fate of P are dependent on PM size ranges (Shinya et al. 2003; Zhou et al. 2005; Berretta and Sansalone 2011). Several N and P treatment methods were investigated by previous studies, such as infiltration and detention basins (Bartone and Uchrin 1999; Dechesne et al. 2005); constructed wetland systems (Gervin and Brix 2001; Seo et al. 2005); s edimentation tank systems (Stenstrom et al. 2002); vegetative controls (Barrett et al. 1998); filtration with floating media filters (Visvanathan et al. 1996); and urban wet detention ponds (Wu et al. 1996; Comings et al. 2000; Wang et al. 2004).
112 Control of PM is challenging in an urban environment, due to complex hydrology, and constraints of land availability and infrastructure. Many of treatment methods are based on volumetric control and therefore may not be applicable in situations where land availabi lity is limited. In such situations, hydrodynamic separators (HS) are commonly used to treat runoff (Kim and Sansalone 2008). However, while HSs can be effective in removing particulate bound N (P), they face the problems of re uptake of N (P) from capture d PM deposited during previous storm events. Additionally, because HS are not designed to capture the entire runoff volume of a storm, determining efficiency is dependent on understanding the transport and partitioning of N (P) during an individual storm e vent. Understanding efficiency and P and N distributions as a function of particle size is necessary for appropriate sizing, retrofitting and optimizing of HS type units. In this study, separation and partitioning of particulate matter (PM) bound nitrogen (N) and phosphorus (P) in a baffled hydrodynamic separator (BHS) located in an urban car park with an impervious surface are investigated during 10 storm events. The results are examined as a function of hydrology. A wide gradation of captured PM from BHS was analyzed after 10 storm events were completed. Objectives The first objective is to determine the distribution of N and P across suspended, settleable and sediment fractions in rainfall runoff for 10 discrete storm events from an urban carpark with a n impervious surface. The second objective is to examine equilibrium partitioning of N and P between dissolved and particulate phases in rainfall runoff. The last objective is to investigate the N and P separation in BHS as a function of particle size dist ribution (PSD) and treated volume
113 Methodology Catchment The experimental site is illustrated in Figure 5 1. The site is located at an urban impervious surface area (Parking lot in University of Florida). Depending on the intensity of the storm event, the catchment area ranges around ~500 m 2 The catchment area slope is approximately 3% E W and 1.5% N S. There are vegetated islands between parking spaces in the watershed which can contribute biogenic materials to impervious surface area. The overall experi mental setup consists of the following components: a system pipe to deliver the rainfall runoff from the catchment to a BHS (10.1 m), a Parshall flume equipped with an ultrasonic sensor for measuring flow rates, a drop box for influent sampling, and a gate valve to divert flow to BHS. This make it possible to monitor the performance of each device for a given storm or series of storms. The installation is shown in Figure 5 1 Data Acquisition, Management, and Sampling The rainfall depth was collected by t ipping bucket rain gauges manufactured by Texas Electronics Inc. (0.254 mm bucket capacity). Flow measurement from the watershed was monitored with 25.4 mm Parshall flume. A 30 kHz ultrasonic sensor (model Shuttle Level Transmitter, MJK Inc.) connected to a Campbell Scientific CR 1000 (Campbell Scientific, Inc.) is used for flow depth monitoring in the Parshall flume and for real time data logging. Both influent and effluent samples from each event were sealed within five separate Nalgene polypropylene sc rew closure bottles. Three of the bottles were 1 L bottles and were used for the particle size distribution (PSD) analysis, PM separation and analysis, nutrient (N and P) analysis and metal analysis. The other two bottles were 0.5 L and were both used for the probe analysis and some of the PM analysis. All of the bottles were taken in succession of one another and thus considered identical in composition. Influent samples were taken manually by
114 the grab method following the parshall flume and before the run off reached the dropbox. Effluent samples were also taken manually by the grab method immediately following the effluent pipe located on the side of the BHS unit PM Separation The 1.1 L samples were used to recover a sizeable quantity of sediment particle s by on the sieve for each sample were carefully collected using a spatula and a stainless steel pick and then placed in a clean, labeled and tared Petri dish. T he 1.0 L samples are also used for (No. 200) sieve. The actual volume of the nominal 1 L runoff sample was first measured (since approximately 100 mL were utili zed) using a graduated cylinder. The 1.0 L of solution that was passed through the sieve was recovered and placed into an Imhoff Cone without the loss of any additional PM. Each Imhoff Cone with a nominal 1 L of stormwater was set aside for a quiescent set tling of 60 minutes. The settleable PM settled to the bottom of the Imhoff cone and these particles were carefully recovered from the Imhoff Cones by slowly decanting the supernatant from the cone and obtaining the particles from the bottom of the cone in clean, labeled and tared Petri dishes. Replicate samples were obtained for each sample; hence all particle fractions were recovered from the replicate (A and B) runoff samples. Part of the supernatant recovered from each Imhoff Cone (almost 50 mL) was then passed through a fractionation column using an air pump. The filtrate was directly used to determine concentrations of Ions (50 mL) like Nitrate (NO 3 ), Phosphate (PO 4 3 ) and dissolved COD. All these constituents were measured by spectrophotometer (Hach DR/5000). The remaining supernatant for each sample replicate was then used to determine the suspended PM fraction. 100 mL from the well mixed supernatant was used to separate the
115 ane filters were then dried in the oven at 105 C. The filters were prepared in advance and tare weighed before use. The difference in weight gives the mass of the PM. A sub sample of the supernatant was used to calculate the suspended particulate bound to tal nitrogen (TN) (Suspended TN). As for the measurement of metals, the Inductively Coupled Plasma Mass Spectrometry (ICP MS) was used Water Chemistry Analysis Water quality parameters, such as pH, salinity, dissolved oxygen (D.O), redox, conductivity, total dissolved solids (TDS), and alkalinity were measured on rainfall runoff samples within 24 hrs. Alkalinity was measured in triplicate using Method 2320B (APHA 1998). Conductivity and TDS were measured with a conductivity meter in replicate. The prob es are standardized with a 3 point curve prior to analyses. A pH meter with a 3 point calibration was used to measure the pH of the samples in replicate. The samples were filtered through a pre weighed 1.2 m glass fiber filter and dried to determine the s uspended solid concentration (SSC) Nitrogen and Phosphorus Analysis Samples were fractionated into dissolved and particulate phases. The particulate bound phosphorus and nitrogen analyses are performed using the method from Standard Methods for Water and Wastewater (APHA et al. 1998). After acid digestion of particles and the filtrate, the concentration values of total dissolved and particulate bound phosphorus and nitrogen were obtained by Cadmium Reduction Method # 8039. To determine the concentration of nitrate and phosphate, a spectrophotometer (Hach DR/5000) is used. The quantity of suspended PM bound N is obtained by subtracting the total dissolved nitrogen (TDN) fraction from the supernatant N fraction. In particular a sample from the
116 supernatant o btained after the removal of the sediment and settleable particles from the 1.0 L sample solutions is acid digested. Similarly a sample from the filtrate is also acid digested as mentioned. The TN from both the supernatant and filtrate samples is obtained and the difference of these values gives the TN bound to suspended particles Partitioning Indices for Nitrogen and Phosphorus The TN and Total Phosphorus (TP) are the sums of the dissolved fraction concentration and particulate bound fraction concentratio n of N (P), respectively. Therefore TN and TP can be expressed with dissolved and particulate bound fraction, as in the following equations: ( 5 1) ( 5 2) In this expression, M d is dissolved mass; and M p is the particulate bound mass. If f d > 0.5, N or P is mainly in dissolved form, otherwise N (P) is predominantly in particulate bound form (Sansalone and Buchberger 1997). The partitioning coefficients, K d is defined as the ratio of the equilibriu m concentration of a dissolved fraction mass with respect to particulate bound fraction mass. The equation is as follows. ( 5 3) In this expression K d is the equilibrium partitioning coefficient between particulate bound ma ss and dissolved mass (L/Kg) while Cs is the particulate bound N (P) mass (mg/g of dry particulate mass). The partitioning coefficient can be used to evaluate the distribution between dissolved and particulate bound N (P)
117 Hydrologic and Loading Parameters Hydrologic and transport parameters were measured for each rainfall event, and are shown in Table 5 1. The parameters include previous dry hours (PDH), event duration (t rain ), rainfall depth (d rain ), maximum rainfall intensity (i rain max ), initial pavemen t residence time (IPRT), runoff volume (V runoff ), maximum flow rate (Q p ), median flow rate (Q med ), runoff coefficient (C), and number of samples (n) Analysis of Recovered Sediment Deposit from Hydrodynamic Separator Sieve analysis is used to determine th e particle size distribution as required or gradation of an aggregate. Analysis followed ASTM D422 63 with additional sieve sizes (ASTM 1993; Sansalone et al. 1998). After being air dried at a constant temperature of 40C and having their dry proper weight s measured PM, dried PM samples are disaggregated and sieved through a set of graded mechanic sieves. The aggregates are placed in the top of the sieve stack and covered with a lid. The sieves are properly secured in the mechanical shaker and then the sha ker is turned on for five minutes. The materials retained on each of the sieves are weighed, including the weight retained on the pan, and the results are recorded. The Fineness Modulus for each PM Sample is then computed. Sieve analysis follows the standa rd procedure ASTM D422 (ASTM 1998). Dry PM separated on each of the stainless steel sieves is weighed and stored separately in round clear sample bottles. A 95 to 98% recovery of PM is required for sieve analysis Results Event Hydrology Event based hydr ological indices including PDH, d rain i rain max as well as V runoff and Q p and Q med (both influent and effluent), were monitored and recorded for a total of 10 storm events occurring between May 24 th 2010 and August 21 th 2010 as shown in Table 5 1. Mon itored storm events during the field test program varied in duration from 25 minutes on August 24 th
118 2010 to 60 minutes on July 31 st 2010. The storm events ranged from 1.8 mm on May 24 th 2010 to 23.6 mm of rainfall depth on July 31 st 2010. The PDH was f rom 15 hours to 261 hours in between storm events. The volume of event runoff was from 79 L to 2386 L. Resulting peak flow rates ranged from 1.2 L/s on May 24 th 2010 to 17.6 L/s on June 29 th 2010, while the median flow rates ranged from 0.03 to 1.52 L/s as shown in Table 5 1. The IPRT referred to in Table 5 1 is the time required for rainfall to satisfy certain conditions including pavement surface wetting and depression storage filling, as well as airborne re suspension, and atmospheric evaporation. In c omparing all 10 storm events, IPRT results varied from 0.4 minutes to 7.3 minutes as shown in Table 5 1. In addition, IPRT is controlled by a combination of PDH and abstractions during each storm (Sansalone et al. 2005). The site hydrology is described for each of the 10 events by runoff hydrographs plotted with f d and f p in Figures 5 6 to 5 7. The hydrographs were observed to respond quickly to fluctuations in rainfall intensity Overall Treatment Efficiency of BHS as a Function of Hydrolog y A total of ten storm events, encompassing a wide range of flow rates, were routed through the BHS between May 24 th 2010 and August 21 st 2010 and with total volume of approximately 32407 L of rainfall runoff from the experimental watershed located in the University of Florida. Each of the storms was unique in regards to their natural and anthropogenic pollutant loadings. All ten storms had wide variations of d rain (1.8 mm 23.6 mm), period of PDH (0.4 min. 7.3 min.), V runoff (871 L 9031 L), Q p (1.2 L/s 16.0 L/s), i rain max (7.6 mm/hr 137.2 mm/hr), and t rain (25 min. 60 min.). These results are summarized in Table 5 1, along with PDH, t rain d rain i rain max IPRT, V runoff Q p Q med number of influent/effluent samples (n inf /n eff ), sampling coverage, and percen t of hydraulic design utilized at Q p Removal efficiency for the 9 TARP qualified storm events ranged from 47% to 98%. The most significant factor in determining the amount captured by the BHS unit was the peak flow intensity of
119 influent heading into the u nit, which has a hydraulic design capacity (Q d ) of 9.05 L/s. In each high intensity storm, the level of total suspended solid (TSS), SSC, TN and TP removal dropped significantly, resulting in a range of 46% to 98% removal. The negative values are shown be cause of re suspension in the unit. Results for PM separation, and TN, TP and SSC removal efficiencies measured for the BHS unit with ten different runoff events are summarized in from Table 5 1 PM fraction and PM based N and P fraction masses distributio n A summary of the relative fractions of the suspended, settleable, and sediment PM for each event are shown in Figure 5 results in rainfall runoff, ranging from 76.0 to 99.5% with a mean of 89.0%. The settleable PM ranges from 0.03 to 12.9% with a mean of 6.3%, while the suspended PM fraction ranged from 0.02 to 17.7% with a mean of 4.7% in rainfall runoff. The fractions of N (P) in rainfall runoff are compared to the fraction of PM mass. The results are plotted in Figure 5 2 on an event basis for each PM fraction (suspended, settleable, sediment, TSS and SSC). A 1:1 l ine of equal separation behavior is illustrated as a reference. Results indicate that there is less than a 1:1 ratio between N (P) associated with settleable PM and suspended PM with a few events deviating from this trend. However, N (P) fractions for sedi ment PM have a larger than 1:1 ratio. Event based Nitrogen and Phosphorus Loadings Each storm event was analyzed for N (P) loadings across PM fractions. The concentrations of total dissolved and total N (P) are presented in Table 5 2. The concentrations o f N (P) varied across 10 storm events. TDN of rainfall ll runoff ranged
120 experiment was higher than that found in previous studies w ithin the local watershed (Rushton 2001; Passeport and Hunt 2009). For instance, TN loadings in the urban areas were previously found to be 1.63, 0.556, and 0.548 mg/L (Rushton 2001; Passeport and Hunt 2009). TN concentrations were higher than in previous studies because this research was performed during the summer, during which time more biogenic materials from vegetated islands in the watershed area are generated. TDP of rainfall h a median 1983; Brown et al. 2003), TP loadings in urban areas were found to be 0.21, 0.29, 0.33 and 0.30 mg/L. This specific carpark transports higher magnitudes o f event mean concentration (EMC) than observed in previous research. Approximately 700 vehicles per day pass by this watershed area (Berretta and Sansalone 2011) which also has a significant load of biogenic materials from vegetated areas in the parking lo t. Leaves falling from trees and grass cuttings, also contribute to the biogenic loads on the pavement and these are eventually conveyed to the BHS unit with the runoff. Nutrients Removal Efficiency as a function of Hydrology Based upon the measurements described above, frequency distributions of N (P) fractions were examined for both influent and effluent. The measured N (P) of total dissolved, suspended,
121 settleable, sediment, and total fraction concentrations are presented in Table 5 2 as EMC values. Me dian, mean and standard deviation values are also presented. The percentage of TP separation varied from 42% to 97% for storms that had a peak flow below the design flow capacity of the unit and not including the first storm which had no effluent. As a co mparison, the percentage of TP separation in storms that had a peak flow above the design flow capacity of the unit varied from 22% to 60%. Similar results were observed for TN. The percentage of TN separation in storms that had a peak flow below the desig n flow capacity of the unit and not including the first storm varied from 12% to 84%. In comparison, the percentage of TN separation in storms that had a peak flow above the design flow capacity of the unit varied from 7% to 59%. It is clear that the unit does perform better, when the peak flow does not exceed the hydraulic design capacity. The BHS unit separation behavior for PM bound N (P) fractions is compared to separation of PM in Figure 5 3 on an event basis for each PM fraction (suspended, settleabl e, and sediment). A 1:1 line of equal separation behavior is p resented as a reference. Results indicate there is less than a 1:1 relationship between N (P) separation associated with settleable PM, frequency distributions for N (P ) are well described by a log The medians of each frequency distribution for N (P) were significantly reduced between influent and ef fluent as illustrated in Figure 5 4 and 5 5. Nutrients Partitioning In rainfall runoff nutrients such as N (P) are partitioned into dissolved and particulate bound fractions. The partitioning of N (P) in urban rainfall runoff influenced by primary IPRT, rainfall pH, oxidation reduction potential (ORP), conductivity and the quantity of soli ds present in rainfall runoff (Sansalone et al., 1997). The dissolved fraction (f d ) and equilibrium partitioning
122 distribution (as K d L/Kg) between dissolved and particulate N (P) phases were examined in both th e influent and effluent. Figure 5 6 summarize s the trends in f d of TN and TP and the magnitude of TDN and TDP concentrations as a function of hydrology for all events. The trend of variation in f d from the catc hment area are the main factor causing the variation of f d through all the events. The role of suspended, settleable, and sediment PM based N (P) were examined with respect to f d and K d Table 5 3 shows that event mean f d values for N ranged from 0.54 to 0 .85, indicating that the majority of N was associated with the dissolved gradation in the rainfall runoff. Event mean f d values for P ranged from 0.08 to 0.42 indicating that the majority of P was associated with the particulate gradation in the rainfall r unoff. Results are illustrated in Figure 5 6, note that all frequency distribution are modeled as log gravimetric index based on all PM in the sample. Partitioning was examined based on suspended, settleable, and sediment PM fractions of N (P). Initi al TDP concentrations are greater than concentrations at the end of event. The results indicated that the BHS unit was effective in separation of the suspended, settleable, and sediment PM fractions of N (P). The initial TDP has higher concentrations as co mpared to the end of the event. The results showed that the effluent from the unit has a statistically significant increase in the dissolved fraction. Additionally, f d varied by over an order of magnitude between events (which is potentially related to pre vious dry hours). K d values in Table 5 3 are also shown in order of decreasing median value. P exhibits much higher K d values than N. K d values ranged from 180.5 L/Kg to 17649.7 L/Kg for N, while values ranged from 8408.2 L/Kg to 37625.2 L/Kg for P. K d va lues are statistically significantly higher for effluent as compared to influent for all the PM fractions for N (P).
123 Nutrient from the Recovered Sediment Deposit Figure 5 7 illustrates the distribution of dry granulometric mass for a given particle diamete r on an incremental and cumulative basis for each PM fraction from less than 25 m to coarser than 9500 m. This analysis was performed to evaluate the gradation of N (P) mass as a functi on of particle diameter. Figure 5 7 illustrates a general trend for d ifferent N (P) species for NO 3 N, NH 3 N, total kjeldahl nitrogen (TKN), PO 4 3 P, and TP, nutrients mass is predominantly associated with the coarse fraction of PM (coarser than 63 m). The cumulative NO 3 N, NH 3 N, and TKN recovered mass was 32 mg, 679 mg, and 45086 mg respectively. The cumulative PO 4 3 P, and TP recovered mass was 1886 mg, and 24695 mg. The mass recovery for each of NO 3 N, NH 3 N, and TKN significantly increased for particle sizes greater than 75 m. The percentage of NO 3 N, NH 3 N, and TKN b etween 75 m and 2000 m was 82%, 88%, and 85%, respectively. For PO 4 3 P, and TP, the percentages between 75 m and 2000 m were 92% and 93%. Most of the nutrients in recovered PM deposits are from the sediment PM fraction. According to data from recovere d deposited mass, NH 3 N is a less abundant species of dissolved nitrogen in runoff; the ratio of NO 3 N to NH 3 N was about 22:1, which means nitrite concentration can be neglected. NH 3 N includes both ammonium and ammonia in the equilibrium aqueous phase. Based on the neutral pH of 7 for rainfall runoff in catchment, the major form of NH 3 N is expected to be ammonium (NH 4 + ). Figure 5 8 summarized the trend in the f d of TN (TP) as a function of cumulative treated rainfall runoff volume. The variation trend of the f d does not follow the changing of treated volume exactly due to the hydrologic complexity. The mean f d values of influent and effluent TN are 0.57 and 0.68, and the mean f d values of influent and effluent TP are 0.28 and 0.36. The mean f d values of influent and effluent TP are significantly lower than TN. This result indicates that the predominance of TP is associated with the PMs.
124 The influent and effluent PSDs were modeled as gamma distributions on an event basis. Influent and effluent PSD differ ences were shown with the probability density function of a gamma distribution as follows: ( 5 4) nimizing the sum of squared errors (SSE), resulting in maximizing the coefficient of determination between the measured and modeled data. Below, F(x) is the cumulative gamma distribution and x represents flow rate in Figure 5 7 and 5 9. ( 5 5) Figure 5 9 shows that the PMs are heterodisperse through all 10 storm events. Cumulative PSDs were examined and results were described by an optimized cumulative gamma distribution function summarized in Figure 5 9. The runoff median PM diameter (d 50m ) ranged within 3%. Interestingly, Influent P mass trend was similar to influent PM mass as a function of cumulative treated volume in BHS. This result indicates th at P mass is associated with PM mass. Figure 5 10 illustrates that cumulative total mass separations for PM, TP and TN were 79.0, 60.2 and 39.4% from BHS. TP mass separation follows PM mass separation; however, TN has significantly lower mass separation ef ficiency. This result indicates P is associated with PM than N.
125 Summary For the given BHS configuration, operating under the loading conditions of the 10 rainfall runoff events considered in this study, sediment PM bound N (P) concentrations ranged from 76 .0 to 99.5%, settleable PM bound N (P) concentrations ranged from 0.03 to 12.9%, and suspended PM bound N (P) concentrations ranged from 0.02 to 17.7% in rainfall runoff. The coarser fraction of PM generated the highest N (P) concentrations and mass, beca use most of N (P) is associated with PM (> 75 m). Sediment PM (> 75 m) represents a significant source area inventory and requires frequent maintenance and management for in situ unit operations to ensure proper function. The PM bound N separation effi ciency for rainfall intensities between 0.07 inches and The event mean PM bound N separation efficiency ranged from 17 to 100%, with an arithmetic mean of 56%, based 81%. Compared to PM separation, the events mean PM bound N (P) separation efficiencies were lower throughout all storm events. The cumulative total mass separation of PM, TP and TN are 79.0 60.2 and 39.4% from BHS. PM and TP separation efficiencies are significantly higher than TN, which indicate that P is predominantly associated with PM than N. Results indicate that the partitioning of N in BHS units is one of the reasons for lower mass separation efficiencies of N. HS type devices, while reasonably effective for removing nutrients that associate with coarser PM, are ineffective at targeting other fractions. The effects of re suspension of removed sediment should not be underestimated
126 Table 5 1 Hydrologic characterization of the 10 rainfall runoff events monitored between May 24, 2010 and August 21, 2010 in Gainesville, FL Event date (2010) PDH (hr) t rain (min) d rain (mm) i rain max (mm/hr) IPRT (min) V runoff (L) Q p (L/s) Q med (L/s) C ( ) n ( ) 24 May 17 36 1.8 7.6 7.3 871 1.2 0.18 0.98 12 04 June 164 42 11.2 106.7 1.2 2952 13.2 0.32 0.53 20 17 June 26 36 7.1 30.5 1.5 1946 6.1 0.76 0.55 20 29 June 192 40 20.6 106.7 0.4 5704 17.6 1.52 0.34 20 11 July 261 42 11.7 76.2 5.6 2377 5.2 0.3 2 0.41 20 28 July 36 50 19.1 137.2 1.0 8316 16.0 1.20 0.75 20 31 July 15 60 23.6 76.2 5.2 9031 13.2 0.60 0.76 20 13 August 112 25 2.8 55.9 2.0 314 3.2 0.06 0.23 20 14 August 28 31 2.5 22.9 1.6 594 2.5 0.25 0.47 20 21 August 83 31 2.8 45.7 2.1 299 1.5 0.03 0.21 20 Median 60 38 9.2 66.1 1.8 2162 5.7 0.32 0.50 20 Mean 93 39 10.3 66.6 2.8 3240 8.0 0.52 0.52 19 SD 87 10 8.3 41.7 2.3 3295 6.4 0.50 0.25 2.5 PDH, t rain d rain i rain max IPRT, V runoff Q p Q med C and n represent previous dry h ours, event duration, rainfall depth, maximum rainfall intensity, initial pavement residence time, runoff volume, maximum flow rate, median flow rate, runoff coefficient and number of samples, respectively.
127 Table 5 2 dissolved nitrogen (TDN), total nitrogen (TN), total dissolved phosphorus (TDP), and total phosphorus (TP). Event Date (2010) TDN TN TDP TP EMC i EMC e EMC i EMC e EMC i EMC e EMC i EMC e (%) (%) (%) L] (%) 24 May 1688 N/A 100 2709 N/A 100 854 N/A 100 3186 N/A 100 04 June 1188 1098 8 1384 1316 5 209 257 23 2367 1487 37 17 June 1454 1547 6 3859 1897 51 123 77 37 5807 610 90 29 June 1787 491 73 3840 1581 59 275 221 20 4564 1807 60 11 July 2 104 1901 10 2709 2383 12 424 425 0 1793 1046 42 28 July 939 656 30 2005 1599 20 238 228 4 2379 1366 43 31 July 240 177 26 668 528 21 240 177 26 668 527 17 13 Aug 1159 1072 8 7542 1239 84 331 451 36 23640 654 98 14 Aug 572 715 25 2502 978 61 349 469 35 1869 866 54 21 Aug 1674 1104 34 3708 1995 46 437 415 5 2538 1085 57 Mean 1281 973 26 3093 1502 46 348 302 10 4881 1050 60 Median 1321 1072 18 2709 1581 49 303 257 5 2459 1046 56 SD. 578 532 37 1884 561 31 203 141 40 6751 436 28
128 Tab le 5 3 Summary of event mean value and range of variation of the dissolved fraction (f d ) and partition coefficient ( K d ) of nitrogen and phosphorus for influent and effluent runoff. Event date (2010) Nitrogen Phosphorus f d K d f d K d Influent Effluent In fluent Effluent Influent Effluent Influent Effluent 24 May 0.85 N/A 1476.5 N/A 0.42 N/A 21290.9 N/A 04 June 0.66 0.92 180.5 873.1 0.08 0.28 8453.6 36181.5 17 June 0.42 0.79 1065.3 5793.9 0.06 0.20 20187.8 154042.6 29 June 0.54 0.38 1606.2 21301.1 0.11 0.18 14247.5 64775.9 11 July 0.79 0.87 957.2 3093.3 0.37 0.42 8408.2 42781.4 28 July 0.47 0.66 3288.1 9490.3 0.15 0.27 20769.4 51380.6 31 July 0.44 0.40 11232.5 38769.2 0.40 0.34 11023.3 68619.6 13 August 0.66 0.83 3575.5 19817.1 0.27 0.66 29570.7 4746 2.9 14 August 0.27 0.67 17649.7 38361.7 0.21 0.52 23770.1 78332.8 21 August 0.58 0.58 7654.8 91307.3 0.26 0.35 37625.2 206288.3 Median 0.56 0.67 4868.6 25423.0 0.24 0.34 19534.7 83318.4 Mean 0.57 0.68 2447.2 19817.1 0.23 0.36 20478.6 64775.9 SD 0.18 0.20 5673.3 28449.9 0.13 0.15 9416.0 57964.4
129 Figure 5 1 Profile section of 1.21 m diameter BHS deployed for physical modeling loaded by urban source area catchment.
130 Figure 5 2 PM fraction and PM ba sed N and P fraction masses distribution within each monitored rainfall runoff event. Each symbol represents a rainfall runoff event. Range bars represent standard deviation
131 Figure 5 3 Separation for TN, and TP in different fractions as a function of PM fractions. Range bars represent standard deviation
132 Figure 5 4 Phosphorus mass concentration distributions for each PM fractions. All data are modeled as log normal distributi ons (p < 0.05). Influent and effluent distributions are statistically significantly different (p < 0.05) for each PM fractions.
133 Figure 5 5 Nitrogen mass concentration distributions for each PM fractions. All data are modeled as log normal distributions (p < 0.05). Influent and effluent distributions are statistically significantly different (p < 0.05) for each PM fractions.
134 Figure 5 6 f d values and equilibrium coefficient, K d va lues of nitrogen and phosphorus in influent and effluent. There are statistically significant difference between infludent f d and K d value and effluent f d and K d value (p <0.05).
135 Figure 5 7 Granulometric equilibrium distribution of ammonium nitrogen, nitrate nitrogen, TKN, phosphate and T P
136 Figure 5 8. The f d of influent and effluent TN (TP) as a function of cumulative treated rainfall runoff volume. A f d of influent TN, B f d of effluent TN, C f d of influent TP, D f d of effluent TP. f d50 represents the mean dissolved fraction.
137 Figure 5 9. The cumulative gamma distribution parameters ( for shape factor and for scaling factor) for event based normalized particle size distributions (PSD). Each point is representative of an event..
138 Figure 5 10. Cumulative influent and effluent mass of PM, phosphoru s (P), and nitrogen (N) through the entire monitoring campaign for baffled hydrodynamic separator (BHS) in Gainesville, FL.
139 CHAPTER 6 CONCLUSION This dissertation focused on a coupled experimental and numerical approach to characterize particulate matter (PM) separation and scour by stormwater unit operations and processes (UOPs) for steady and transient hydrologic, hydraulic and pollutant loadings. F our UOPs were modeled physically and numerically inclcuding a baffled hydrodynamic separator (BHS) a vor tex hydrodynamic separator (VHS) a screened hydrodynamic separator (SHS) and a volumetric clarifying filtration system (VCF) T his study examine d the inter and intra event N and P removal as a function of particle size, hydrology and partitioning for an urban carpark treated by a BHS with significant biogenic loadings. A commonly used BHS was analyzed for washout of pre deposited PM as a function of surface overflow rates indexed as flow rates from 10 to 125% of the HS design flow. The CFD model was validated with experimental data across a range of flow rates and particle size distributions, PSDs. Furthermore, a SHS and a VHS were physically and numerically compared to a BHS unit. Three different HS units were successfully modeled with CFD to access PM behavior, using finite volume method ( FVM ) a standard k a Lagrangian discrete phase model ( DPM ) to track particles. CFD models were validated for PM concentration, mass and PSDs with less than 10% RPD. Lagrangian particle trajectory results show that VHS has coarsest eluted and washout PM, as well as, the highest washout rates. The vortexing inner chamber results in a higher rate of re suspension of finer PM in the screened hydrodynamic separator A CFD based probability function was developed for each hydrody namic separator ( HS ) for particle elution as function of flow rate and diameter. Such probability functions, combined with available physical modeling data can provide a reliable method of predicting PM yield from a HS
140 PM separation by a BHS SHS and VCF system for transient hydraulic and particulate loads observed in a real time rainfall runoff event was modeled with stepwise steady flow CFD model by the application of the standard k model. Four discrete rainfall runoff events for each UOPs were modeled T he modeled results agreed with the measured data (Absolute RPD <10%). This study successfully applied stepwise steady step CFD modeling of three HS units as validat ed by matching the physically observed PM mass data. Instead of computing a fully unsteady CFD model, the stepwise steady step model provides an efficient method of simultation Th e stepwise steady CFD modeling reduces the required computing time by more than 16 times. Even with complex unstea dy flow in urban area can be representative with stepwise steady simulation. A calibrated/validated CFD based iterative approach to design of unit operation s has the potential to provide reduced prototyping costs with improved performance, as a result of c arefully designed experimental matrices, focused on PM control requirements for effluent discharges. A CFD approach to modeling the PM removal characteristics and PM washout of UOPs is a state of the art approach to reducing the uncertainty that results fr om assuming ideal conditions, thus providing a more effective method for pollution control. Urban rainfall can be a significant source of N and P, both in particulate bound and dissolved forms. Results indicate that partitioning in rainfall runoff resulted predominantly in N the highest concentrations of N (P), but the highest mass is also associated with the coarser PM from 5 to 100 % from the BHS with a median 49%, and PM bound P removal ranged from 17 to 100 % from the BHS with a median 56%. Results indicate that there is generally less than a 1:1 relation ship between the removal of N (P)
141 and associated removal of set a few events deviating from this trend. This study concludes that while particulate P and N constitute a large portion of the total removal by the BHS, the long term effects of re uptake and re partitioning need to be studied in addition to particulate scour, as part of a maintenance program.
142 APPENDIX A CHAPTER 3. PHYSICAL AND CFD MODELING OF PM SEPARATION AND SC OUR IN HYDRODYNAMIC SEPARAT ORS Figure A 1. Schematic view and dimensions of baffled hydrodynamic separator (BHS) The effective volume of the unit indicates the volume occupied of water without any influent flow Baffled hydrodynamic separator (BHS) Dimensions Unit diameter 1.22 m Unit height 1.52 m Effective volume of unit 1.78 m 3 Bypass baffle height 0.23 m Diameter of influent pipe 0.30 m Diameter of effluent pipe 0.30 m Length of influent droplet pipe 0.43 m Length of effluent droplet pipe 0.41 m Overall unit surface area 1.17 m 2
143 Figure A 2. Schematic view an d dimensions of vortex hydrodynamic separator (VHS) The effective volume of the unit indicates the volume occupied of water without any influent flow Vortex hydrodynamic separator (VHS) Dimensions Inner vortex chamber diameter 2.13 m Unit height 2.13 m Unit depth 1.22 m Unit width 1.74 m Effective volume of unit 7.00 m 3 Diameter of effluent pipe 0.30 m Diameter of effluent pipe 0.30 m Overall unit surface area 3.70 m 2
144 Figure A 3. Schematic view and dimensions of screened hydrody namic separator (SHS) The effective volume of the unit indicates the volume occupied by a static column of water without any influent flow Screened hydrodynamic separator (SHS) Dimensions Unit diameter 2.13 m Unit height 1.68 m Effective volume of uni t 2.96 m 3 Volume of cylindrical sump 0.15 m 3 Diameter of sump 0.64 m Diameter of effluent pipe 0.25 m Diameter of screen 0.64 m Aperture opening Screened area 1.27 m 2 Overall unit surface area 3.58 m 2
145 Table A 1. Morsi and A lexander values as function of Reynolds number are reported below. ( Morsi and Alexander 1972) Re a 1 a 2 a 3 <0.1 24.0 0 0 0.1 < Re < 1 22.73 0.0903 3.69 1 < Re < 10 29.1667 3.8889 1.222 10 < Re < 100 46.5 116.67 0.6167 100 < Re < 1000 98.33 2 778 0.3644 1000 < Re < 5000 148.62 4.75 10 4 0.357 5000 < Re < 10,000 490.546 57.87 10 4 0.46 10,000 < Re < 50,000 1662.5 5.4167 10 6 0.5191 Figure A 4. Reynolds number for three hydrodynamic separators (HS) as a function of flow rate
146 APPENDIX B CHAPTER 4. STEPWISE STEADY CFD MODELING OF UNSTEADY FLOW AND PM LOADING TO UNIT OPER ATIONS Figure B 1. Schematic view and dimensions of baffled hydrodynamic separator (BHS) The effective volume of the unit indicat es the volume occupied of water without any influent or effluent flow Baffled hydrodynamic separator (BHS) Dimensions Unit diameter 1.22 m Unit height 1.52 m Effective volume of unit 1.78 m 3 Bypass baffle height 0.23 m Diameter of influent pipe 0. 30 m Diameter of effluent pipe 0.30 m Length of influent droplet pipe 0.43 m Length of effluent droplet pipe 0.41 m Overall unit surface area 1.17 m 2
147 Figure B 2. Schematic view and dimensions of screened hydrodynamic separator (SHS) The effective volume of the unit indicates the volume occupied by a static column of water without any influent or effluent flow Baffled hydrodynamic separator (BHS) Dimensions Unit outer diameter 0.89 m Unit inner diameter 0. 50 m Unit height 1.23 m E ffective volume of unit 0.65 m 3 Diameter of influent pipe 0.15 m Diameter of effluent pipe 0.20 m Distance from top to screened area 0.33 m Aperture opening Screened area 0.52 m 2 Overall unit surface area 0.62 m 2
148 Figure B 3. Sche matic view and dimensions of volumetric clarifying filter (VCF) The effective volume of the unit indicates the volume occupied of water without any influent or effluent flow
149 AOCM represents Aluminum oxide coated media Volumetric clarifying filter (VCF) Dimensions Cartridge outer diameter 0.46 m Cartridge inner diameter 0.08 m Cartridge height 0.53 m Cartridge media size (d 50 ) 3.56 mm Cartridge media specific gravity 2.35 g/cm 3 Cartridge media specific surface area 0.94 m 2 /g Cartridge media porosity 36.7% Cartridge media dry bulk density 0.68 g/cm 3 Filter media AOCM Unit height (influent) 1.69 m Unit height (effluent) 1.87 m Unit depth 1.17 m Unit width 2.12 m Effective volume of unit 1.92 m 3 Diameter of in fluent pipe 0.15 m Diameter of effluent pipe 0.15 m Bottom t o V notch weir on baffle 0.39 m Overall unit surface area 2.25 m 2
150 Table B 1. Morsi and Alexander values as function of Reynolds number are reported below. ( Morsi and Alexander 1972) Re a 1 a 2 a 3 <0.1 24.0 0 0 0.1 < Re < 1 22.73 0.0903 3.69 1 < Re < 10 29.1667 3.8889 1.222 10 < Re < 100 46.5 116.67 0.6167 100 < Re < 1000 98.33 2778 0.3644 1000 < Re < 5000 148.62 4.75 10 4 0.357 5000 < Re < 10,000 490.546 57.87 10 4 0.46 10,000 < Re < 50,000 1662.5 5.4167 10 6 0.5191 Figure B 4. Reynolds number for three hydrodynamic separators (HS) as a function of flow rate
151 Table B 2 The injection particulate matter (PM) size Injection PM size 2000 850 600 425 300 250 180 150 106 75 63 53 45 38 25 10 7 5 3 1 Table B 3 Particle injection time (min), flow rate (L/s), influent and effluent PM concentration, median particle diameter (d 50m ), shape factor, and scaling factor of s torm 1 (29 June 20 10 ) BHS Injection time (min) Flow rate (L/s) PM inf [mg/L] PM eff [mg/L] d 50m 0.0 0.00 0.00 0.00 0.0 0.00 0.00 2.3 2.21 2444. 3 133. 2 50.5 0.70 117.88 4.1 5.67 1488. 5 538. 2 78.6 0.70 201.67 5.6 1.59 863. 9 548. 6 78.1 1.09 91.57 7.3 4.97 756. 6 510. 8 108.5 0.87 199.71 9.6 4.15 428.9 198. 9 74.6 1.37 67.53 12.6 2.28 487. 9 20 1 0 62.3 0.98 80.08 15.6 2.42 396. 9 63.2 71.8 1.32 67.72 20.6 1.24 522. 8 36.7 62.8 1.09 75.85 26.6 0.48 136.3 7 5.0 85.2 1.57 61.26 35.6 0.07 4929. 9 22. 4 68.8 1.24 73.97
152 Table B 4 Particle injection time (min), flow rate (L/s), influent and effluent PM concentration, median particle diameter (d 50m ), shape factor, and scaling factor of s torm 2 (11 July 20 10 ) BHS Injection time (min) Flow rate (L/s) PM inf [mg/L] PM eff [mg/L] d 50m 0.0 0.00 0.0 0.0 0.0 0.00 0.00 6.0 0.16 1760.1 18.2 124.8 0.57 493.56 8.0 0.46 445. 9 6. 6 109.8 0.84 19 2.01 10.0 0.35 195.7 8.0 125.6 0.83 220.96 12.0 0.43 1001. 6 14. 5 109.1 1.10 138.28 15.0 3.21 223.2 57. 9 41.0 0.95 63.93 18.3 1.54 49. 5 66.9 61.3 1.39 55.97 22.5 2.94 231. 5 65. 5 145.9 1.24 152.58 25.5 2.76 996.8 143. 7 150.8 1.70 105.02 28.3 0.55 391. 4 101.1 191.2 1.68 127.81 34.3 0.02 1573. 6 29. 5 190.3 0.66 496.34 Table B 5 Particle injection time (min), flow rate (L/s), influent and effluent PM concentration, median particle diameter (d 50m ), shape factor, and scaling factor of s torm 3 (29 June 20 10 ) BHS Injection time (min) Flow rate (L/s) PM inf [mg/L] PM eff [mg/L] d 50m 0.0 0.00 0.0 0.0 0.0 0.00 0.00 4.4 2.47 8088. 6 52. 1 187.4 0.67 484.48 5.9 2.14 782. 8 1 7.0 125.6 0.57 4 98.67 7.3 2.83 419.7 192.8 86.4 0.81 155.24 10.7 3.82 207. 9 107.0 125.6 0.83 220.96 17.2 0.04 3137. 0 39. 1 109.2 1.09 139.65 20.3 2.20 567.9 62.0 41.5 0.95 64.27 23.3 3.74 54 9.0 64.4 61.2 1.40 55.25 28.3 13.63 260. 5 163. 8 146.9 1.22 157.01 38.5 0.97 78.5 41.9 150.9 1.71 103.87 48.7 0.98 20. 4 12. 7 191.5 1.67 128.35
153 Table B 6 Particle injection time (min), flow rate (L/s), influent and effluent PM concentration, median particle diameter (d 50m ), shape factor, and scaling factor of s torm 4 (11 July 20 10 ) BHS Injection time (min) Flow rate (L/s) PM inf [mg/L] PM eff [mg/L] d 50m 0.0 0.00 0.0 0.0 0.0 0.00 0.00 8.8 0.23 977. 8 8. 4 249.0 1.88 165.55 10.2 0.46 267. 1 8. 1 170.6 0.88 314 .32 11.9 2.30 739.7 15. 8 183.8 0.83 371.15 12.7 2.14 240. 1 25. 7 171.8 0.82 357.48 13.6 0.38 108.1 25.2 129.7 0.76 321.84 14.6 0.46 700.4 29.1 428.9 1.18 474.71 15.5 0.38 115.9 43.2 406.4 1.00 541.59 16.7 0.38 56.7 63.9 182.1 0.74 398.36 22.0 0.35 33 3 18.1 85.4 0.56 324.19 27.1 0.06 13. 8 16. 8 50.4 0.83 97.48 Table B 7 Particle injection time (min), flow rate (L/s), influent and effluent PM concentration, median particle diameter (d 50m ), shape factor, and scaling factor of s torm 1 (13 March 2004) S HS Injection time (min) Flow rate (L/s) PM inf [mg/L] PM eff [mg/L] d 50m 0.0 0.00 0.0 0.0 0.0 0.00 0.00 7.0 0.0 1 230. 9 0.4 45.2 0.77 35.36 22.0 0.05 10.6 1.5 28.4 0.56 132.35 101.0 0.02 1421. 8 6.4 52.2 0.85 20.95 105.0 0.25 863. 8 1.1 60.3 0.79 15.67 109.0 0.76 62. 6 0.2 43.9 0.81 24.72 113.0 1.16 164. 8 8.1 52.8 0.82 22.43 117.0 2.73 40 2.0 21.6 49.0 0.84 16.46 121.0 2.19 0. 1 6. 6 45.0 0.85 11.39 125.0 1.93 36.2 6. 2 41.5 0. 86 17.95 129.0 1.99 3.7 2.7 56.4 0.83 14.13 133.0 0.89 12. 7 6. 6 43.6 0.87 18.78 148.0 0.41 51. 2 0.5 55.4 0.95 10.21 163.0 0.55 34.6 0. 2 46.1 0.88 13.64 208.0 1.07 1.5 0.1 51 .4 0.8 5 1 8 34 253.0 0.88 12.6 0.1 4 4.6 0. 7 7 1 9 63 283.0 1.55 4.2 0.1 5 1.2 0. 88 15 46 393.0 1.81 17.1 0.1 38 .1 0. 79 22 3 4
154 Table B 8 Particle injection time (min), flow rate (L/s), influent and effluent PM concentration, median particle diameter (d 50m ), shape factor, and scaling factor of s torm 2 (20 August 2004) SHS Injection time (min) Flow rate (L/s) PM inf [mg/L] PM eff [mg/L] d 50m 0.0 0.00 0. 0 0 0.0 0.0 0.00 0.00 10.0 0.0 1 4259.82 334.20 255.9 1.19 289.68 12.0 1.13 3859.64 866.17 232.0 0.98 341.91 14.0 3.71 374.87 307.82 188.8 1.45 169.73 17.0 5.08 27 98.83 215.85 195.0 1.70 142.08 21.0 15.80 375.64 111.78 199.8 1.66 145.46 26.0 15.85 445.04 115.89 211.5 1.05 273.43 29.0 7.18 206.69 105.61 541.2 3.98 152.60 31.0 3.71 161.64 50.82 735.3 4.15 194.53 33.0 1.61 108.08 60.53 547.4 3.37 187.00 34.0 1.01 138.77 51.59 551.0 3.71 168.52 36.0 0.74 66.95 43.42 2 88 7 1.22 188 29 38.0 0.46 63.99 48.66 2 64 2 1.4 2 1 65 37 40.0 0.29 65.60 45.63 2 56 7 1. 38 1 55 91 43.0 0.19 92.46 34.02 297 3 1. 45 1 52 50 46.0 0.12 127.12 334.20 285 .0 1. 21 132 76 Table B 9 P article injection time (min), flow rate (L/s), influent and effluent PM concentration, median particle diameter (d 50m ), shape factor, and scaling factor of s torm 3 (14 October 2004) SHS Injection time (min) Flow rate (L/s) PM inf [mg/L] PM eff [mg /L] d 50m 0.0 0.00 0.0 0.0 0.0 0.00 0.00 31.0 0.01 230. 9 0 .0 94.4 0.90 174.45 63.0 0.04 618.5 379.5 13.6 0.97 21.33 69.0 0.31 849.2 347.9 38.5 0.58 156.25 73.0 0.43 604. 3 303.5 28.5 0.74 71.45 79.0 0.42 458.4 28 3.0 116.2 1.02 172.27 86.0 0.4 6 384. 1 257.7 133.4 0.71 350.89 106.0 0.43 332. 5 232. 2 64.7 0.78 116.59 122.0 0.27 256.8 209.6 115.4 2.04 62.12 132.0 0.11 200.2 198.1 85.3 1.54 63.79 140.0 0.06 197. 2 200. 4 22.4 0.72 58.61 155.0 0.09 23 7.0 210. 6 90.7 0.71 152.71 165.0 0.08 236.0 218 .5 80.3 0.56 175.31 176.0 0.03 192. 8 217. 1 60.2 0.66 1 33 25 197.0 0.01 19 4.0 217. 1 54.3 0.54 1 16 82
155 Table B 1 0 Particle injection time (min), flow rate (L/s), influent and effluent PM concentration, median particle diameter (d 50m ), shape factor, and scaling factor of s torm 4 ( 30 June 200 5 ) SHS Injection time (min) Flow rate (L/s) PM inf [mg/L] PM eff [mg/L] d 50m 0.0 0.00 0.0 0.0 0.0 0.00 0.00 9.0 0.01 3075.7 0.0 133.3 0.92 226.78 11.0 0.57 1753.3 462.9 268.8 0.76 561.33 12.0 2.18 1016.7 424.3 102.1 0.54 443.80 14.0 4.03 681.9 394.4 77.7 0.57 319.00 15.0 6.81 405.8 323.3 40.0 0.59 171.56 17.0 11.06 265.0 207.7 38.9 0.56 190.10 19.0 13.21 166.5 136.4 44.6 0.53 220.46 21.0 13.53 153.5 120.2 106.0 0.53 410.79 2 3.0 12.75 142.8 94.5 45.7 0.55 207.91 29.0 10.22 104.6 81.9 83.4 0.66 278.95 33.0 2.50 74.9 69.7 39.8 0.5 8 1 81 34 36.0 0.59 118.1 62.2 42.9 0.5 3 1 75 11 42.0 0.54 332.8 95.8 45.1 0.5 7 1 70 60 48.0 1.50 301.3 114.2 48.2 0.5 8 1 80 12 58.0 2.30 105.4 103.1 46.3 0.5 8 1 90 47 68.0 1.97 113.0 99.9 37.1 0.56 1 82 15 Table B 1 1 Particle injection time (min), flow rate (L/s), influent and effluent PM concentration, median particle diameter (d 50m ), shape factor, and scaling factor of s torm 1 (21 April 2006) VCF Injection time (min) Flow rate (L/s) PM inf [mg/L] PM eff [mg/L] d 50m 0.0 0.00 0.0 0.0 0.0 0.00 0.00 1.0 1.72 5805.5 158.9 33.1 0.57 132.80 3.0 3.64 2695.4 130.9 24.5 0.66 78.85 6.0 4.30 931.9 122.7 18.3 0.66 54.28 10.0 0.43 767.6 116.1 11.5 0.67 34.13 13.0 0.25 603.0 108.9 12.1 0.68 33.75 16.0 0.10 254.9 109.1 7.5 0.69 21.07 19.0 0.04 183.3 108.6 7.3 0.65 23.32 23.0 0.01 174.1 100.5 35.7 1.13 41.27 27.0 0.02 247.3 118.0 25.1 0.71 52.74 32.0 0.02 227.4 112.6 5.8 0.67 18.49 3 7.0 0.01 161.0 126.2 6.2 0.95 11.97 42.0 0.02 183.1 106.4 33.7 0.98 47.23 49.0 0.01 183.0 117.7 16.3 0.60 43.08
156 Table B 1 2 Particle injection time (min), flow rate (L/s), influent and effluent PM concentration, median particle diameter (d 50m ), shape factor, and scaling factor of s torm 1 (29 April 2006) VCF Injection time (min) Flow rate (L/s) PM inf [mg/L] PM eff [mg/L] d 50m 0.0 0.00 0.0 0.0 0.0 0.00 0.00 5.0 0.20 5036.4 58.8 11.5 0.70 34.37 9.0 8.74 769.4 100.6 8.7 0.51 60. 66 13.0 6.38 325.3 77.8 19.4 0.84 41.10 17.0 3.14 272.7 49.4 123.5 0.91 207.60 21.0 1.66 256.1 41.2 49.0 0.47 373.64 25.0 1.28 192.1 44.6 315.6 24.55 12.89 31.0 1.69 150.7 39.6 13.7 0.74 37.23 37.0 1.78 124.2 38.3 248.2 1.54 164.52 43.0 0.94 135.3 2 6.9 20.1 0.48 169.53 49.0 0.96 152.5 30.8 299.9 15.44 19.40 60.0 0.60 165.3 26.0 24.0 0.50 179.53 69.0 0.38 112.7 20.0 297.4 9.80 30.01 76.0 0.39 99.9 17.0 31.8 0.49 242.11 87.0 0.24 77.1 23.6 97.0 0.51 386.80 104.0 0.21 60.1 25.7 47.6 0.49 332.80 1 34.0 0.54 58.3 19.0 309.7 19.99 15.51 143.0 0.58 42.4 19.4 13.5 0.87 25.39 159.0 0.17 37.9 19.8 75.9 0.94 100.69 Table B 1 3 Particle injection time (min), flow rate (L/s), influent and effluent PM concentration, median particle diameter (d 50m ), s hape factor, and scaling factor of s torm 1 ( 04 July 2006 ) VCF Injection time (min) Flow rate (L/s) PM inf [mg/L] PM eff [mg/L] d 50m 0.0 0.00 0.0 0.0 0.0 0.00 0.00 6.0 0.06 2236.3 40.5 40.6 0.64 139.16 8.0 0.80 631.6 46.4 76.4 0.59 250.61 10.0 0.94 253.4 34.2 22.5 0.77 55.68 12.0 0.60 245.0 45.4 129.9 0.58 332.81 13.0 0.26 211.7 43.4 19.9 0.78 45.24 15.0 0.21 328.2 39.9 40.4 0.60 174.43 18.0 0.16 166.2 45.6 17.7 1.03 23.35 20.0 0.08 171.2 41.1 23.2 2.93 9.41 26.0 0.10 477.0 42.0 14.9 0.74 34.67 29.0 0.14 154.1 48.7 47.3 0.91 93.35 31.0 0.07 111.0 39.6 16.0 0.82 30.03 34.0 0.05 87.6 39.9 27.8 0.61 183.36 37.0 0.02 87.0 32.8 13.2 0.82 23.57
157 Table B 1 4 Particle injection time (min), flow rate (L/s), influent and effluent PM concen tration, median particle diameter (d 50m ), shape factor, and scaling factor of s torm 1 (04 July 2006) VCF Injection time (min) Flow rate (L/s) PM inf [mg/L] PM eff [mg/L] d 50m 0.0 0.00 0.0 0.0 0.0 0.00 0.00 6.0 0.05 219.6 23.1 15.2 0.76 36.34 9.0 0.19 79.3 21.6 27.1 0.53 149.70 12.0 0.12 45.9 26.0 9.3 0.83 19.54 15.0 0.06 74.4 30.5 25.7 0.85 45.93 19.0 0.08 49.8 32.7 7.8 0.70 23.77 22.0 0.04 61.9 30.3 56.6 0.67 183.79 26.0 0.05 182.8 38.9 10.7 0.69 32.52 29.0 0.41 237.8 45.0 103.3 0.59 315.56 32.0 0.59 159.7 37.6 8.3 0.79 18.83 36.0 0.23 150.7 31.7 28.8 1.87 19.19 40.0 0.23 60.6 35.0 7.0 0.79 16.56 48.0 0.14 35.4 25.7 25.0 0.93 33.17 58.0 0.02 29.8 26.8 13.3 0.72 35.95 65.0 0.09 12.9 12.5 64.2 0.79 121.15 Table B 1 5 C omputing time for stepwise steady CFD modeling and fully transient flow as a function of number of steady steps # of steady steps Stepwise steps computing time (hrs.) Fully transient flow computing time (hrs.) 10 1.5 1 72 12 20 3.0 1 72 12 30 4.5 1 72 12 40 6.0 1 72 12 50 7.5 1 72 12
158 Table B 15. Computing time for stepwise steady CFD modeling and fully transient flow as a function of number of steady steps UOP Storm event BHS (2010) 29 June 15.73 11 July 9.47 28 July 7.87 14 August 2.99 SHS (2004 2005) 14 March 2.66 20 August 7.82 14 October 1.42 30 June 8.24 V CF (2006) 21 April 17.46 29 April 3.59 04 July 11.89 05 July 4.25
159 LIST OF REFERENCES Aguirre Pe J, Olivero M.L., and Moncada A.T. (2 003) Particle densimetric Froude number for estimating sediment transport. Journal of Hydraulic Engineering. 129 (6) pp 428 437. Alley W.M. (1981) Estimation of impervious area washoff parameter, Water Resources Research, 17(4), pp 1161 1166. Alley W.M. and Smith P.E. (1981) Estimation of accumulation parameters for urban runoff quality modeling, Water Resources Research 17(6), pp 1657 1664. Andoh, R.Y.G., and Saul, A.J. (2003) The use of hydrodynamic vortex separators and screening systems to improve water quality. Water Sci ence and Technol ogy 47(4), pp 175 183. Annandale, G.W. (2006) Scour Technology, 1st Ed., McGraw Hill Professional, New York. APHA American Public Health Association, Eation, A.D., Clesceri, L.S., Greenberg, A.E. (Eds.), (1998) Standard m ethods for the e xamination of w ater and w astewater, 20 th ed. Washington, DC. ASTM (American Society for Testing and Materials), (1993) Standard practice for dry preparation of soil samples for particle size analysis and determination of soil co nstants. In: Annual Book of Standards, Designation, D 421 85, 04.08, Philadelphia, 8 9. ASTM, (1998). Standard t est m ethod for p article s ize a nalysis of s oils, D 422 63.ASTM International, West Conshohocken, Pennsylvania. ASTM (1999) Standard test method for determining sediment concentration in water samples, /Annual book of standards, Designation: D /3977 97, 04.08, Philadelphia, pp 395 400. ASTM (2006) Standard practice for classification of soils for engineering purposes (Unified Soil Classification System). ASTM standard D2487. ASTM International, West Conshohocken, P A. ASTM, (2007) Standard t est m ethod for d etermining s ediment c oncentration in w ater s amples: American Society of Testing and Materials, D 3977 97, 11.02, 389 394, doi:10.1520/D3977 97R 07 Barrett, M.E., Walsh, P.M., Malina, J.F. Jr. and Charbeneau, R.J. (1998) Performance of Vegetative Controls for Treating Highway Runoff, J ournal of Environ mental Eng ineering 124(11), pp 1121 1128. Barth, T.J., and Jespersen, D. (1989) The design and application of upwind schemes on unstructured meshes, Technical Report AIAA 89 0366, AIAA 27th Aerospace Sciences Meeting, Reno, Nev Bartone, D.M. and Uchrin, C.G. (1999) Comparison of pollutant removal efficiency for two residential storm water basins. Journal of Environmental Engineering 125(7), pp 674 677.
160 Beheshti, A.A. and Ataie Ashtiani, B. (2008) Analysis of threshold and incipient conditions for sediment movement, Costal Engineering 55(5), pp 423 430 Berretta, C., and Sansalone, J. J. (2011) H ydrologic transport and partitioning of phosphorus fractions. Journal of Hydrology 403 (1 2), pp 25 36. Berretta, C. and Sansalone, J.J. (2011) Speciation and transport of phosphorus in source area rainfall runoff, Water Air and Soil Pollution 222(1 4) pp 351 365 Berretta, C., Rage, S., and Sansalone, J.J. (2011) Quantifying nutrient loads associated with urban particulate matter through source control and maintenance practices, Report to Florida Stormwater Association, Tallahassee, FL. Brezonik, P.L. and Stadelmann, T.H. (2002). Analysis and predictive models of stormwater runoff volumes, loads, and pollutants concentrations from watersheds in the Twin Cities metropolitan area, Minnesota, USA. Water Resources 36, pp 1743 1757. Brombach, H. (1987) L iquid solid separation at vortex storm overflows, paper presented at 4th International Conference on Urban Storm Drainage, Lausanne, Switzerland, pp 103 108. Brombach, H., Xanthopoulos, C. Hahn, H.H., and Pisano, W.C. (1993) Experience with vortex separators for combined sewer overflow control, Water Science and Technology. 27(5 6), pp 93 104. Broschat, T.K. (1995) Nitrate, phosphate, and potassium leaching from container grown plants fertilized by several methods HortScience 30, pp 74 77 Brown, T., Schueler, T., Wright, T., Winer, R., and Zielinkski, J. (2003) Maryland Chesapeake and A tlant ic coastal bays critical area 10% rule guidance manual. Center for Watershed Protection, Ellicott City, MD Buffington, J.M. (1999) The legend of A.F.Shields, Journal of Hydraulic Engineering 125(4), pp 376 387 Cates, E.L., Westphal, M.J., Cox, J.H., Calabria, J. and Patch, S.C. (2009) Field evaluation of a proprietary storm water treatment system: Removal efficiency and relationships to peak flow, season, and dry time, Journal of Environmental Engineering 135(7), pp 511 517, http://dx.doi.org/10.1061/(ASCE)0733 9372(2009)135:7(511) Cellino, M ., and Lemmin, U (2004) Influence of coherent flow structures on the dynamics of suspended sediment transport in open channel flow, Journal of Hydraulic Engineering 130 ( 11 ) 1077 1088, doi:10.101/(ASCE)0733 9429(2004)130:11(1077) Cho, H C., and Sansalone, J.J. (2012) Physical Modeling of Particulate Matter Washout from a Hydrodynamic Separator, Journal of Environmental Enginee ring doi: http://dx.doi.org/10.1061/(ASCE)EE.1943 7870.0000556
161 Coduto, D.P. (1999) Geotechnical Engineering: Principles and Practices Prentice Hall, Upper Saddle River, NJ Co mings, K.J., Booth, D.B. and Horner, R.R. (2000) Storm water removal by two wet ponds in Bellevue, Washington. Journal of Environmental Engineering 126(4), pp 321 330. Curtis, J.S., and van Wachem, B. (2004) Modeling particle laden flows: A research out look AIChE J ournal 50 (11) pp 2638 2645. Dean, C., Sansalone, J., Cartledge, F., and Pardue, J. (2005) Influence of hydrology on stormwater metal element speciation at the upper end of the urban watershed, Journal of Environmental Engineering 131(4), p p 632 642. Dechesne, M., Barraud, S., and Bardin, J.P. (2005) Experimental assessment of stormwater infiltration basin evolution. Journal of Environmental Engineering 131, pp 1090 1098. Denbigh, K. and Turner, J. (1984) Chemical Reactor Theory, An Intro duction. Cambridge University Press, Cambridge Dickenson J.A., and Sansalone J.J. (2009) Discrete Phase Model Representation of Particulate Matter (PM) for Simulating PM Separation by Hydrodynamic Unit Operations Environ mental Sci ence and Technol ogy 4 3(21), pp 8220 8226 DOI:10.1021/es901527r Elgobashi, S.E. (1991) Particle laden turbulent flow: Direct simulation and closure models, Applied Scientific Research 48 ( 3 4 ) pp 301 314, doi:10.1007/BF02008202 Ergun, S. (1952) Fluid flow through packed col umn, Chemical Engineering Progress 48(2), pp 89 94 Faram, M.G. and Harwood, R. (2003) A method for the numerical assessment of sediment interceptors, Water Science Technology 47 ( 4 ) pp 167 174. Fenner, R.A. and Tyack, J.N. (1997) Scaling laws for hydro dynamic vortex separators, Journal of Environmental Engineering 123(10), pp 1019 1026. Fenner, R.A. and Tyack, J.N. (1997) Scaling laws for hydrodynamic vortex separators, Journal of Environmental Engineering 123(10), pp 1019 1026. Fernandez Sempere, J. Font Montesinos, R., and Espejo Alcaraz, O. (1995) Residence t ime d istribution for u nsteady s tate s ystems, Chemical Engineering Science 50(2), pp 223 230 Ferziger, J.J, and Peric, M. (2002) Computational methods for fluid dynamics, 3rd Ed., Springer V erlag, Berlin, Germany. Swirl technology: Enhancement of design, evaluation, and application, Journal of Environmental Engineering 122 ( 8 ) pp 741 748, doi:10.1061/(ASCE)0733 9372(1996)122:8(741)
162 Field, R. and Sullivan, D. (2002) Management of wet weather flow in the urban watershed. InWet weather Flow in the Urban Watershed: Technology and Management. Finlayson Pitts, B.J., and Pitts, J.N. (2000) Chemistry of the Upper and Lower Atmosphere Theory, Experiments and Applications 1st Ed., Academic Press, CA, USA, pp 365 368 Fogler, H. S. (1992) Elements of Chemical Reaction Engineering, 2 nd Ed., Prentice Hall, Englewood Cliffs, NJ Froment, G.F. and Bischoff, K.B. (1990) Chemical Reactor Analysis and Design, 2 nd Ed., Wiley, New York Gargett, A., Wells, J., Tejada Martinez, A.E., and Grosch, C.E. (2004) Langmuir s upercells: A m echanism for s ediment r esuspension and t ransport in Shallow Seas, Science 306 ( 5703 ) pp 1925 1928, doi:10 .1126/science.1100849. Garofalo, G., and Sansalone, J.J. (2011) Transient elution of particulate matter from hydrodynamic unit operations as a function of computational parameters and runoff hydrograph unsteadiness, Chemical Engineering Journal doi:10.10 16/j.cej.2011.09.086. Gervin, L., and Brix, H. (2001). Removal of nutrients from combined sewer overflows and lake water in a vertical flow constructed wetland system. Water Science and Technology 44(11 12), pp 171 176. Gray, J.R., Glysson, G.D., Turcios, L.M., Schwarz, G.E. (2000) Comparability of s uspended s ediment c oncentration and t otal s uspended s olids d ata, U.S Dep ar t ment of the Interior, USGS Information Services Reston, Virginia. Hazen, A ( 1904 ) On sedimentation, Transaction, ASCE, 53, pp 45 71 Heaney, J.P., Wright, L., Sample, D. (1999) Water Environment Research 71(2), pp 241 250 Hopkinson, C.S. and Vallino, J.J. (1995) The relationships among man s activities in watersheds and estuaries a model of runoff effects on patterns of estuarine community metabolism. Estuaries and Coasts 18(4), pp 598 621. Huber, W.C., and Dickinson, R.E. (1988) manual. US Environmental Protection Agency, Athens, Georgia. R eport No. EPA 600/3 88 001a, pp 569 Hunt, W.F., Jarrett, A.R., Smith, J.T., and Sharkey, L.J. (2006) Evaluating bioretention hydrology and nutrient removal at three field sites in North Carolina, Journal of Irrigation and Drainage Engineering 132(6), pp 600 608 Hunt, W.F., Smith, J.T., Jadlocki, S.J., Hathaway, J.M., and Eubanks, P.R. (2008) Pollutant removal and peak flow mitigation by a bioretention cell in urban Charlotte, N.C., Journal of Environmental Engineering, 134(5), DOI: 10.1061/(ASCE)0733 9 372(2008)134:5(403)
163 Ingersoll, A.C. and McKee, J.B. (1956) Fundamental concepts of rectangular settling tanks, Transactions of the American Society of Civil Engineers, 102, pp 463 543. Jianghua, Y., Qitao, Y., and Kim, Y. (2009) Performance analysis of a hydrodynamic separator for treating particulate pollutants in highway rainfall runoff, Environmental Engineering Research 14(4), pp 262 269 Jing, S.R., Lin, Y.F., Lee, D.Y., and Wang, T.W. (2001) Nutrient removal from polluted river water by using const ructed wetlands, Bioresource Technology 76(2), pp 131 135 Jonasz, M. (1991) Size, shape, composition and structure of microparticles from light scattering, Principle, methods, and application of particle size analysis (Ed. By J.P.M. Syvitski), pp 143 16 2, Cambridge University Press, New York. Kayhanian, M., Suverkropp, C., Ruby, A., and Tsay, K. (2007) Characterization and prediction of highway runoff constitudent event mean concentration, Journal of Environmental Management 85(2), pp 279 295 Kim, J Y ., and Sansalone, J.J. (2008) Hydrodynamic separator of particulate matter transported by source area runoff Journal of Environmental Engineering 134(11), pp 912 922 Kim, J Y, and Sansalone, J.J. (2008) Event based size distributions of particulate m atter transported during urban rainfall runoff events, Water Research 42(10 11), pp 2756 2768, http://dx.doi.org/10.1016/j.watres.2008.02.005 Kim, J Y., Pathapati, S., Liu, B., an d Sansalone, J.J. (2007) Treatment and maintenance of Stormwater hydrodynamic separator: a case study, Proceedings of the 9th Biennial Conference on Stormwater Research and Watershed Management, Orlando, FL. Kraus, N.C., Lohrman, A., Cabrera, R. (1994) N ew Acoustic Meter for measuring 3D Laboratory Flows, Journal of Hydraulic Engineering 120 (3), pp 406 412. Ku, C.S.M., and Hershey, D.R. (1997) Growth response, nutrient leaching and mass balance for potted poinsettia. I. Nitrogen, J ournal of Am erican So c iety for Hort icultural Sci ence 122, pp 452 458 Launder, B.E., and Spalding, D.E. (1974) The numerical computation of turbulent flows, Computer Methods in Applied Mechanics and Engineering 3 ( 2 ) pp 269 289, doi:10.1016/0045 7825(74)90029 2 Lee, J.H., and Bang, K.W. (2000) Characterisation of urban stormwater runoff. Water Research 34(6), pp 1773 1780 Lee, J.H, Bang, L.W., Choi, C.S. and Lim, H.S. (2010) CFD modeling of flow fi eld and particle tracking in a hydrodynamic Stormwater separator, Water Science and Technology 62(10), pp 2381 2388, doi:10.2166/wst.2010.480 Letterman, R.D. (1999) Water Quality and Treatment, 5 th Ed., McGraw Hill, New York.
164 Levenspiel, O. (1962) Chemi cal Reaction Engineering 3 rd Ed., Wiley, New Jersey. Lewis, D.B., and Grimm, N.B. (2007) Hierarchical regulation of nitrogen export from urban catchments: interactions of storms and landscapes Ecological applications : a publication of the Ecological Soc iety of America, 17, pp 2347 2364. Li, Y., Buchberger, S.G. and Sansalone J.J. (1999) Variably saturated flow in a storm water partial exfiltration trench, Journal of Environmental Engineering 125(6), pp 556 565 Liang, D., Cheng, L., and Li, G. (2005) N umerical modeling flow and scour below a pipeline in currents.Part II. Scour simulation, Coastal Engineering 52 ( 1 ) pp 43 62, doi:10.1016/j.coastaleng.2004.09.001 Liu, B., Be rretta, C., Ying, G., and Sansalone, J.J. (2008) Control of highway stormwater during event and interevent retention in best management practices, Journal of the Transportation Research Board 2120, 115 122, Transportation Research Board of the National A cademies, Washington, D.C. Lohrmann, A., Cabrera, R., and Kraus, N.C. (1994) Acoustic Doppler velocimeter (ADV) for laboratory use. Proc. Conf. on Fundamentals and Advancements in Hydraulic Measurements and Experimentation, Buffalo, NY, American Society o f Civil Engineers, pp 351 365. Luyckx, G., and Berlamont, J. (2004) Removal efficiency of swirl/vortex separators, Urban Water Journal 1(3), pp 251 260, DOI:10.1080/15730620410001731991 Ma, J, Ying, G, and Sansalone, J.J. (2010) Transport and distributi on of particulate matter phosphorus fractions in rainfall runoff from roadway source areas, J ournal of Environ mental Eng ineering 136(11), pp 1197 1205 Marchner, H. (1995) Mineral nutrition of higher plants 2nd ed. Academic Press, New York. Morin, A., Fi gue, J.F., Schaffner, J., and Steinhardt, J. (2008) 11th International Conference on Urban Drainage, Edinburgh, Scotland, UK Morsi, S.A., and Alexander, A.J. (1972) An investigation of particle trajectories in two phase flow systems, Journal of Fluid Mec hanics 55 ( 02 ) pp 193 208, Cambridge University Press, doi:10.1017/S0022112072001806 Nauman, E. B., Buffham, B. A. (1983) Mixing in continuous flow systems, Wiley, New York. Novotny, V., and Witte, J.W. (1997) Ascertaining aquatic ecological risks of ur ban stormwater discharges Water Research 31(10), pp 2573 2585. NRC [National Research Council] (2001) Assessing the TMDL approach to water quality management National Academy Press, Washington, D.C., USA.
165 Orlins, J.J. and Gulliver, J.S. (2003) Turbule nce quantification and sediment re suspension in an oscillating grid chamber, Experiments in Fluids 34 ( 6 ) pp 662 677, doi:10.1007/s00348 003 0595 z Parker, J.T.C., Fossum, K.D., and Ingersoll, T.L. (2000) Chemical characteristics of urban stormwater sed iments and implications for environmental management, Maricopa County, Arizona. Environmental Management 26 pp 99 115 Passeport, E., and Hunt, W.F. (2009). Asphalt parking lot runoff nutrient characterization for eight sites in North Carolina, USA. Journal of Hydrologic Engineering 14(4), pp 352 361 Patankar, S. (1980) Numerical Heat Transfer and Fluid Flow 1 st Edition, Hemisphere, Washington, D.C. Pathapati, S., and Sansalone, J.J. (201 2 ) Modeling particulate matter re suspension and washout from urban drainage hydrodynamic separators, Journal of Environmental Engineering 138(1), 90 100, doi:10.1061/(ASCE)EE.1943 7870.0000427 Pathapati, S., and Sansalone, J.J. (2011) Can a stepwise steady flow computational fluid dynamics model reproduce un steady particulate matter separation for common unit operations?, Environmental Sciences and Technology 45 ( 13 ) pp 5605 5613, doi : 10.1021/es103584c. Pathapati, S., and Sansalone, J.J. (2009) CFD modeling of a storm water hydrodynamic separator, Journal o f Environmental Engineering 135 ( 4 ) 191 202, doi:10.1061/ASCE 0733 9372 2009 135:4 191 Peck, R.B., Hanson, W.E., and Thornburn, T.H. (1974) Foundation e ngineering, Wiley, New York. Pisano, W.C., and Brombach, H. (1994) Operational experience with vortex solids separators for combined sewer overflow (CSO) c ontrol, Water Science and Technology 29( 1 2), pp 383 391 Qi, D., and Lin, W. (2006) T Grid: A new grid environment, Proc., 1 st Int. Multi Symposiums on Computer and Computational Sciences Conference ( pp 611 618. Ranade, V.V. (2002) Academic Press, London, U.K. Rushton, B.T. (2001) Low impact parking lot design reduces runoff and pollutant loads, Journal of Water Resources Planning and Management 127(3), pp 172 179 Rushton, B. (2004) Broadway outfall stormwater retrofit incorporating a CDS unit, paper presented at Critical Transitions in Water and Environmental Resources Management, Proceeding of the Wo rld Water and Environmental Resources Congress, Salt Lake City, Utah, USA.
166 Sansalone, J.J. and Buchberger, S.G. (1997) Partitioning and first flush of metals in urban roadway storm water Journal of Environmental Engineering 123(2), pp 134 143. Sansalon e, J. J. Koran, J. M. Smithson, J. A. and Buchberger, S. G. (1998) Physical characteristics of highway solids transported during rainfall. Journal of Environmental Engineering 124(5), 427 440. doi:10.1061/(ASCE)0733 9372(1998)124:5(427). Sansalone, J.J. ( 2002) Physical and chemical nature of stormwater pollutants, in Wet weather flow in the urban watershed: Technology and Management, pp 43 66 Field, R. and Sullivan, D. eds., New Jersey, USA. Sansalone, J.J. and Pathapati, S. (2009) Particle dynamics in a hydrodynamic separator subject to transient rainfall runoff, Water Resources Research 45, W09408, doi:10.1029/2008WR007661 Sansalone, J.J., Liu, B., and Kim, J Y (2009) Volumetric clarifying filtration of urban source area rainfall runoff, Journal of Environmental Engineering 135(8), pp 609 620, http://dx.doi.org/10.1061/(ASCE)EE.1943 7870.0000044 Sansalone J. J. Liu B., and Ying G (2010) Volumetric filtration of rainfal l runoff: (II) Event based and inter event nutrient fate J ournal of Environmental Engineering 136(12) pp 1331 1340 Sansalone, J.J., Ying, G., and Lin, H. (2010) Distribution of metals for particulate matter transported in source area rainfall runoff, Journal of Environmental Engineering 136(2), http://dx.doi.org/10.1061/(ASCE)EE.1943 7870.0000139 Sartor J. D, Boyd G. B, Agardy F. J (1974) Water pollution aspects of street surface contaminants, Water Pollution Control Federation 46(3), pp 458 467 Seo, D.C., Cho, J.S., Lee, H.J., and Heo, J.S. (2005) Phosphorus r etention c apacity of f ilter m edia for e stimating the l ongevity of c onstructed w etland. Water Research 39, pp 24 45 2457. Sheng, Y., Ying, G., and Sansalone, J.J. (2008) Differentiation of transport for particulate and dissolved water chemistry load indices in urban source area watersheds, Journal of Hydrology 361 ( 1 2 ) 144 158, doi:10.1016/j.hydrol.2008.07.039 Shi elds, A. F. (1936) Anwendung der Aehnlichkeitsmechanik und der Turbulenzforschung auf die Geschlebebewegung, Mitteilung der PreussischenVersuchsenstaltfrWasserbau und Schiffbau W. P. Ott and J. C. van Uchelen, translators. Heft 26, Berlin Shinya M., Tsu chinaga T., Kitano M., Yamada Y., and Ishikawa M. (2000) Characterization of heavy metals and polycyclic aromatic hydrocarbons in urban highway runoff, Water Science and Technology 42( 7 8), pp 201 208 Shinya, M., Tsuruho, K., Ishikawa, M. (2003) Evaluat ion of factors influencing diffusion of pollutant loads in urban highway runoff. Water Science Technology 47(7 8), pp 227 232.
167 Smith, J.M. (1981) Chemical Engineering Kinetics, 3 rd Ed. McGraw Hill, New York State of Pennsylvania (2003) The Technology Acceptance Reciprocity Partnership Protocol for http://www.dep.state.p a.us/dep/deputate/pollprev/techservices/tarp/pdffiles/Tier2protocol.p df Stenstrom, R.S., Clausen, J.C. and Askew, D.R. (2002). Treatment of p arking l ot s tormwater u sing a s torm t reat s ystem. Environmental Science & Technology 36(20), pp 4441 4446. Swamee, P. K., Tyagi, A. (1996) Design of Class I sedimentation tank, Journal of Environmental Eng ineering, 122(1), pp 71 73 Tchobanoglous G., Burton F. L., and Stensel H. D. (2003) Wastewater Engineering: Treatment, Disposal, and Reuse., 4 th Ed., McGraw Hill, New York Turner, R.K., Georgiou, S., Gren, I.M., Wul F., Scott, B., Soderqvist, T., Bateman, I.J., Folke, C., Langaas, S., Zylicz, T., Maler, K.G., and Markowska, A. (1999) Managing nutrient tion study. Ecological Economics 30, pp 333 352. U.S. Environmental Protection Agency, (1983) Results of the n ationwide u rban r unoff p rogram. f inal r eport. No. NTIS PB84 185552. Washington, DC U.S. Environmental Protection Agency, (1993) Research and Development, Washington, DC. U.S. Environmental Protection Agency (1999) Stormwater technology fact sheet h ydrodynamic vortex separators, Rep. EPA 832 F 99 017, Washington, D.C. U.S. Environmental Prote ction Agency (2002) National management measures to control nonpoint source pollution from urban areas draft. EPA 8421 B 02 003 Valloulls, I.A., and List, E.J. (1984) Numerical simulation of a sedimentation basin. 1. Model development, Environmental Scie nce and Technology, 18, pp 242 247. Valloulls, I.A., and List, E.J. (1984) Numerical simulation of a sedimentation basin. 2. Design application, Environmental Science and Technology, 18, pp 253 257. Van Wachem, B.G.M., and Almstedt, A.E. (2003) Methods f or multiphasecomputational fluid dynamics, J ournal of Chem ical Eng ineering 96, pp 81 98. Vanoni, V.A. (1946) Transportation of suspended sediment by water, Transportation of American Society of Civil Engineering 111, pp 67 133. Versteeg, H.K., and Malal asekera, W. (1995) An introduction to computational fluid dynamics: The finite volume method approach 1 st edition, Prentice hall, London.
168 Voulgaris, G. and Trowbridge, J.H. (1998) Evaluation of Acoustic Doppler Velocimeter (ADV) for Turbulence Measureme nts, Journal of Atmospheric Oceanic Technology 15, pp 272 289 Wang, G.T., Chen, S., Barber, M.E. and Yonge, D.R. (2004) Modeling f low and p ollutant r emoval of w et d etention p ond t reating s tormwater r unoff. Journal of Environmental Engineering 130(11), p p 1315 1321. Wang, H., Hondzo, M., Xu, C., Poole, V., and Spacie, A. (2003) Dissolved oxygen dynamics of streams draining an urbanized and an agricultural catchment. Ecological Modelling 160, pp 145 161. Weon, S Y., Lee, C W., Lee, S I., Koopman, B. (20 02) Nitrite inhibition of aerobic growth of Acinetobacter sp. Water Research ., 36, pp 4471 4476 Westerterp, K.R., Van Swaaij, W.P.M. and Beenackers, A.A.C.M. (1993) Chemical Reactor Design and Operation Wiley, New York Wilson, M.A., Mohseni, O., Gullive r, J.S., Hozalski, R.M., and Stefan, H.G. (2009) Assessment of hydrodynamic separators for stormwater treatment, Journal of Hydraulic Engineering 135 ( 5 ) 383 392, doi : http://dx.doi.org/10.1061/(ASCE) HY.1943 7900.0000023 Wu, J.S., Holman, R.E. and Dorney, J.R. (1996) Systematic e valuation of p ollution r emoval by u rban w et d etention p onds. Journal of Environmental Engineering 122(11), pp 983 988. Visvanathan, C., Werellagama, D.R.I.B., and Aim, R.B. ( 1996) Surface water pretreatment using floating media filter. Journal of Environmental Engineering 122, pp 25 33. Ying, G. and Sansalone, J.J. (2008) Partitioning and granulometric distribution of metal leachate from urban traffic dry deposition particu late matter subject to acidic rainfall and runoff retention, Water Research 42(15), pp 4146 4162 Ying, G. and Sansalone, J.J. (2008) Granulometric relationships for urban source area runoff as a function of hydrologic event classification and sedimentati on, Water Air and Soil Pollution 193(1 4), pp 229 246 Ying, G., and Sansalone, J.J. (2008) Differentiation of transport for particulate and dissolved water chemistry load indices in rainfall runoff from urban source area watersheds, Journal of Hydrology 361 ( 1 2 ) 144 158, doi:10.1016/j.jhydrol.2008.07.039 Yu, S.L., and Stopinski, M.D. (2001) Testing of ultra urban stormwater best management practices, VTRC 01 R7, Virginia tran sportation research council, Charlottesville, Va. Zhang, J., and Jrgensen, S.E. (2005) Modeling of point and non point nutrient loadings from a watershed, Environmental Modeling and Software 20(5), pp 561 574 Zhou, A., Tang, H., Wang, D. (2005) Phospho rus adsorption on natural sediments: modeling and effects of pH and sediment composition. Water Research 39, pp 1245 1254.
169 BIOGRAPHICAL SKETCH Hwan Chul Cho was born in Daegu South Korea and he came to the United States in Fall 200 7 after receiving his Bachelor of Engineering degree in Department of E nvironmental E ngineering at Ajou University, South Korea He receive d his Ph.D. in environmental engineering from University of Florida in April 20 1 2 His doctoral research was focused on physical and compu tational fluid dynamics ( CFD ) models of PM separation and scour in hydrodynamic unit operation s He completed his research under the guidance of Dr. John J. Sansalone in the Department of Environmental Engineering and Sciences. Hwan Chul pursues life to th e fullest, loves his family deeply, and enjoys deep and intimate relationships.