Citation
Performance Monitoring and Particulate Matter CFD Modeling of an Infrastructure-Constrained "Bio-Detention" System Loaded by Pavement Rainfall Runoff in Northcentral Florida

Material Information

Title:
Performance Monitoring and Particulate Matter CFD Modeling of an Infrastructure-Constrained "Bio-Detention" System Loaded by Pavement Rainfall Runoff in Northcentral Florida
Creator:
Brenner, Gregory A
Publisher:
University of Florida
Publication Date:
Language:
English

Thesis/Dissertation Information

Degree:
Master's ( M.E.)
Degree Grantor:
University of Florida
Degree Disciplines:
Environmental Engineering Sciences
Committee Chair:
SANSALONE,JOHN JOSEPH
Committee Co-Chair:
DELFINO,JOSEPH J
Committee Members:
HEANEY,JAMES
Graduation Date:
12/13/2013

Subjects

Subjects / Keywords:
Average linear density ( jstor )
Charge flow devices ( jstor )
Diameters ( jstor )
Flow velocity ( jstor )
Maintenance costs ( jstor )
Nitrogen ( jstor )
Particulate materials ( jstor )
Phosphorus ( jstor )
Rain ( jstor )
Surface runoff ( jstor )
bmp
cfd
nutrients
pm
stormwater
City of Gainesville ( local )

Notes

General Note:
Control strategies for nutrients and particulate matter (PM) in runoff from the built environment have received significant attention over the last several decades.  The impetus for such strategies has been load and concentration reduction in response to regulatory drivers such as total maximum daily loads (TMDL) across the USA and more recently, numeric nutrient criteria (NNC) in Florida.  Therefore, best management practices (BMPs) are increasingly promoted.  For example, bio-detention is intended to provide hydrologic control and sequester constituents such as PM, nutrients and metals. One object of this study is the performance monitoring of a bio-detention filter system with respect to particulate matter (PM), nitrogen (N), phosphorus (P), various water chemistry measurements, and runoff detention and retention for 12 rainfall-runoff events.  Also, an economic and performance based analysis of this system is compared with other unit operations from previous studies in similar urban environs.  The other objective is to develop a computational fluid dynamics (CFD) model to correctly predict the fate and transport of influent PM through the bio-detention filter system.  It is hypothesized the model could correctly predict the amount of mass separated by the system and also the particle size distribution (PSD) of the escaped influent PM.  The model is to be validated with all 12 measured rainfall-runoff events. The bio-detention system in this study has a volume of 2.9 m3 and is loaded directly by runoff from an asphalt paved roadway in north central Florida.  Across the monitoring campaign, on a mass basis, the system sequestered 69% of total PM and 57 % of suspended PM.  The mass-based mean PM particle size decreased from 100 µm in the influent to 21 µm in the effluent.  For the entire campaign, on a mass basis, the system sequestered 32% of TN and 41% of TP.  The system sequestered 5% of dissolved nitrogen (DN) although effluent DN increased relative to the influent in six events.  Similarly, the system sequestered 20% of dissolved phosphorus (DP) but effluent DP increased relative to the influent in five events. This study assessed the ability of  Euler-Lagrangain CFD modeling to correctly predict the fate and transport of influent PM mass based on step-wise steady state flow rates and PSD in a bio-detention filtration system.  Validated with the 12 rainfall-runoff events, the model predicted influent PM mass separation with a mean of 5 % absolute relative percent difference (ARPD), with half of the events overestimating and half underestimating the separation of influent PM.  Along with mass separated, the escaped PSDs of PM were predicted with a mean of 7% normalized root mean squared error (NRMSE).

Record Information

Source Institution:
UFRGP
Rights Management:
Copyright Brenner, Gregory A. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Embargo Date:
12/31/2015

Downloads

This item has the following downloads:


Full Text

PAGE 1

1 PERFORMANCE MONITORING AND PARTICULATE MATTER CFD MODELING OF AN INFRASTRUCTURE PAVEMENT RAINFALL RUNOFF IN NORTHCENTRAL FLORIDA By GREGORY ALLEN BRENNER A THESIS PRESENTED TO T HE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF ENGINEERING UNIVERSITY OF FLORIDA 2013

PAGE 2

2 2013 Gregory A llen B renner

PAGE 3

3 To my family

PAGE 4

4 ACKNOWLED G MENTS I would like to thank Dr. John J. Sansalone for mentoring me throughout my graduate research. I would also like to thank Dr. Joseph J. Delfino and Dr. James P. Heaney for their support and service as members of my committee; Kelsi Tiempe Valerie Thorsen, Jordan Se gal, Rachel Aponte for their assistance on fieldwork and laboratory analysis ; and Lynn Dirk for helping improve my technical writing

PAGE 5

5 TABLE OF CONTENTS page ACKNOWLED G MENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGURES ................................ ................................ ................................ .......... 9 LIST OF ABBREVIATIONS ................................ ................................ ........................... 11 ABSTRA CT ................................ ................................ ................................ ................... 14 CHAPTER 1 GLOBAL INTRODUCTION ................................ ................................ ..................... 16 2 PHYSICAL MODELING OF NUTREINT AND PARTICULATE MATTER SEQUESTRATION IN AN INFRASTRUCTURE C RUNOFF ................................ ... 18 Introduction ................................ ................................ ................................ ............. 18 Objectives ................................ ................................ ................................ ............... 20 Methodology ................................ ................................ ................................ ........... 21 Source Area and Bio detention Filter System ................................ ................... 21 Hydrologic Measurement ................................ ................................ .................. 22 Influent and Effluent Sampling ................................ ................................ .......... 23 Event Mean Concentration ................................ ................................ ............... 23 Water Chemistry ................................ ................................ ............................... 24 Particulate Matter ................................ ................................ ............................. 24 Phosphorus and Nitrogen ................................ ................................ ................. 26 Results and Discussi on ................................ ................................ ........................... 27 Hydrology ................................ ................................ ................................ ......... 27 Particulate Matter ................................ ................................ ............................. 27 Phosphorus ................................ ................................ ................................ ...... 28 Nitrogen ................................ ................................ ................................ ............ 29 Water Chemistry ................................ ................................ ............................... 29 Economics ................................ ................................ ................................ ........ 30 Conclusions ................................ ................................ ................................ ............ 31 3 COMPUTATIONAL FLUID DYNAMICS MODELING OF PARTICULATE MATTER IN URBAN RAINFALL RUNOFF TREATMNET THROUGH A BIO DETENTION FILTER SYSTEM ................................ ................................ .............. 44 Introduction ................................ ................................ ................................ ............. 44 Objectives ................................ ................................ ................................ ............... 45 Methodology ................................ ................................ ................................ ........... 46

PAGE 6

6 Source Area and Bio detention Filter System ................................ ................... 46 Hydrologic Measurement ................................ ................................ .................. 47 Particulate Matter ................................ ................................ ............................. 47 Event Mean Concentration ................................ ................................ ............... 48 PM Leaching ................................ ................................ ................................ .... 48 CFD ................................ ................................ ................................ .................. 49 Results and Discussion ................................ ................................ ........................... 54 Bio detention Results ................................ ................................ ....................... 54 CFD Calibration ................................ ................................ ................................ 54 CFD Results ................................ ................................ ................................ ..... 54 Conclusion ................................ ................................ ................................ .............. 55 4 GLOBAL SUMMARY AND CONCLUSIONS ................................ .......................... 66 APPENDIX ADDITIONAL SYSTEM MONITORING INFORMATION ........................ 68 Methodology ................................ ................................ ................................ ........... 68 Hydrology ................................ ................................ ................................ ......... 68 Filter Bypass ................................ ................................ ................................ ..... 68 EMC/EMV ................................ ................................ ................................ ......... 69 Water Chemistry ................................ ................................ ............................... 70 Particulate Matter ................................ ................................ ............................. 70 Phosphorus and Nitrogen ................................ ................................ ................. 71 Unit Operation Economic Comparison ................................ ............................. 71 Filter Media ................................ ................................ ................................ ....... 73 Event Hydrology Summaries ................................ ................................ .................. 73 LIST OF REFERENCES ................................ ................................ ............................. 102 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 107

PAGE 7

7 LIST OF TABLES Table page 2 1 Hydrologic data for all 12 monitored rainfall runoff events during the campaign with initial pavement residence time (IPRT), previous dry hours (PDH), and influent and effluent system flow rate characteristics.. ..................... 33 2 2 Influent a nd effluent particle size distributions (PSD) for all 12 monitored rainfall runoff events during the campai gn as event mean values (EMV) ......... 34 2 3 Event PM and nutrient fraction masses and perce nt removals (PR) for all 12 monitored rainfall mass t hroughout monitoring campaign. ................................ ............................. 35 2 4 Water Chemistry Measurements for 11 monitor ed rainfall runoff events during the campaign as event mean values (EMV) or event mean concentrations (EMC).. ................................ ................................ ....................... 36 3 1 Event based particulate matter (PM) mass percent separation (PS) for all 12 monitored rainfall runoff events. PS is defined as the change in influent to effluent PM mass divided by the influent mass. ................................ .................. 57 3 2 Influent and escaped influent particle size distributions (PS D) for all 12 monitored rainfall runoff events during the campaign. The d 10 is the diameter in which 10 % of the total mass has diameter less than this size.. ..................... 58 3 3 This table summarizes the relative percent difference (RPD) and normalized root mean squared error (NRMSE) between measured and modeled results ... 59 A 1 Filter bypass results. Measured hydrographs were transformed into watershed hydrographs and then the runoff volume was compared with rainfall volume to arrive at runoff coefficients (C) ................................ ................ 78 A 2 Unit operation (UO) PM and nutrient performance based on UO monito ring results and their economi cs (with a 25 year design life) ................................ .... 79 A 3 Monitoring results and economic analysis for the screened hydrodynamic separator (SHS) system. ................................ ................................ ................... 80 A 4 Monitoring results and economic analysis for the volumetric clarify ing filter (VCF) system ................................ ................................ ................................ ...... 81 A 5 Monitoring results and economic analysis for the baf fle hydrodynamic separation (BHS) system. Annual runoff volume is based on the median volumetric rainfall runoff coefficient of 0.56 (Sansalone and Cho 2011). ......... 82

PAGE 8

8 A 6 Monitoring results and economic analysis for the radial cartridge filter (RCF) system. Annual runoff volume is based on the median volumetric rainfall runoff coefficient of 0.56 (Sansalone and Berre tta 2009). ................................ 83 A 7 Monitoring results and economic analysis for the Jellyfish (JF4) system. Annual runoff volume is based on the median volumetric rainfall runoff coeffic ient of 0.56 (Sansalone 2011). ................................ ................................ 84 A 8 Monitoring results and economic analysis for the bio detention system. ............ 85 A 9 Influent and effluent particle size distributions (PSD) for all 12 monitored rainfall runoff events during the camp aign.. ................................ ........................ 86 A 10 Event based particulate matter (PM) fractions as event mean concentration (EMC) and value (EMV) for all 12 rainfall runoff events during the campaign.. ................................ ................................ ................................ .... 87 A 11 Event based particulate matter (PM) mass percent separation (PS) for all 12 monitored rainfall runoff events. PS is defined as the change in influent to effluent PM mass divided by the influent mass. ................................ .................. 88 A 12 Event based phosphorous (P) fractions as event mean concentration (EMC) for all 12 rainfall runo ff events during the campaign. ................................ .......... 89 A 13 Event based phosphorus (P) mass percent separation (PS) for all 12 mon itored rainfall runoff events. ................................ ................................ ......... 90 A 14 Event based phosphorous (N) fractions as event mean concentration (EMC) for all 1 2 rainfall runo ff events during the campaign. ................................ ......... 91 A 15 Event based nitrogen (N) mass percent separation (PS) for all 12 monitored rainfall runoff events. ................................ ................................ .......................... 92

PAGE 9

9 L IST OF FIGURES Figure page 2 1 The contributing drainage area schematic for the eastside filter of NE 19 th Terrace between NE 8 th and 10 th Avenue in Gainesville, Fl. ............................... 37 2 2 Side view of the bio detention filter system and flow monitoring system. The soil media occupies 2.85 m 3 (101 ft 3 ). ................................ ................................ 38 2 3 Probability den sity functions (pdf) of hydrologic flow rates, PM fractions, total dissolved solids (TDS) and pH throughout the events for the entire monitoring campaign. ................................ ................................ ......................... 39 2 4 Probability density functions ( pdf) of phosphorus and nitrogen fractions. ........... 40 2 5 Cumulative influent and effluent particulate matter (PM) fractions throughout the Gainesville, Fl. bio detention filter monitoring campaign by elapsed intra event time. ................................ ................................ ................................ .......... 41 2 6 Cumulative influent and effluent phosphorus (P) fractions throughout the Gainesville, Fl. bio detention filter monitoring campaign by elapsed intra event time ................................ ................................ ................................ ........... 42 2 7 Cumulative influent and effluent phosphorus (N) fractions throughout the Gainesville, Fl. bio detention filter monitoring campaign by elapsed intra event time. ................................ ................................ ................................ .......... 43 3 1 Eastside bio detention filter system aerial view schematic. Arrows show runoff flow from curb to PVC channel, parshall flume, and into system. ............. 60 3 2 An example of a system effluent PSD, with (A) untreated PM discharged from filter, and (B) PM deposition recovered from 100 mm filter effluent PVC pipe. B section was disregarded in modeling. ................................ .................... 61 3 3 Model inputs of media particle size (A) and influent PM particle size distribution (PSD) (B). ................................ ................................ ......................... 62 3 4 Model bio detention filter geometry with 442,000 computational cells (9.4 mL/cell) ................................ ................................ ................................ ............... 63 3 5 Measured and CFD modeled effluent PSDs for 2012 rainfall runoff events. ...... 64 3 6 Measured and CFD modeled effluent PSDs for 20 12 rainfall runoff events ....... 65 A 1 The theoretical and measured calibration curves for the effluent 25.4 mm (1 in) Parshall flume. ................................ ................................ ............................... 93

PAGE 10

10 A 2 Influent and effluent filter hydrographs and hyetographs for January 18 through June 7, 2013 events. ................................ ................................ ............. 94 A 3 Influent and effluent filter hydrographs and hyetographs for June 14 through Au gust 6, 2013 events. ................................ ................................ ....................... 9 5 A 4 Filter runoff bypass. The correlation between flows received by the filter (Q F ) to the catchment area flow (Q C ) are represented in two parts as a linear function up to 11 L/min Q C and then followed by a power function. .................... 96 A 5 Phosphorus reagent calibration curve. Hach Permachem PhosVer 3 Phosphate Reagent was used in accordance to Standard Methods 4500 P B acid hydrolysis. ................................ ................................ ................................ ... 97 A 6 Nitrogen 2714045 / 2672145 was used in accordance to Standard Methods4500 N C for samples grea ter than 25 mg/L N. ................................ .... 98 A 7 2671745 / 2672145 was used in accordance to Standard Methods4500 N C for samples less than 25 mg /L N. ................................ ................................ ....... 99 A 8 Probability density functions (pdf) of influent and effluent dissolved oxygen (DO) throughout the events for the entire monitoring campaign. ...................... 100 A 9 Probability density functions (pdf) of influent and effluent oxidation reduction potential (Redox) throughout the events for the entire monitoring campaig . 101

PAGE 11

11 LIST OF ABBREVIATIONS ADT average daily traffic ARPD absolute relative percent difference flow weighted average concentration (M/L 3 ) C 1 ,C 2 empirical constants in the standard k C 2 inertial resistance factor (m 1 ) c(t) time variable dissolved or particulate bound concentration (M/L 3 ) C D_i d rag coefficient CFD computation fluid dynamics d 10 particle diameter at which 10 % of partic le gradation mass is finer (m) d 50 particle diameter at which 50 % of particle gradation mass is finer (m) d 90 particle diameter at which 90 % of particle gradation mass is finer (m) DI deionized water DN discretization number D p particle diameter (m) DPM discrete phase model EMC event mean concentration ET evapotranspiration F Di buoyancy/gravitational force per unit mass of particle g i sum of the body forces in the th direction (m s 2 ) H water stage (in) k turbulent kinetic energy per unit mass (m 2 s 2 ) K 1 K 2 K 3 empirical constants as function of particle Re i LID low impact development

PAGE 12

12 M total constituent mass (M) NRMSE normalized root mean squared error PM particulate matter PSD particle size distribution PVC polyvinylchloride p Reynolds average pressure (kg m 2 ) p d particle diameter (m) flow rate (L/s) time variable flow (L 3 /T) RANS Reynolds average Navior Stokes Re i Reynolds number for a pa rticle ROW right of way S mean strain rate (m s 1 ) S i source term of the th momentum equation t time tr total event time TS time step (min) TSS total suspended solids u i Reynolds average velocity in the th direction (m s 1 ) u j Reynolds average velocity in the th direction (m s 1 ) i j Reynolds stresses (m 2 s 2 ) v superficial velocity (m s 1 ) V total event volume (L 3 ) v p_i particle velocity in the th direction (m s 1 ) x i i th direction vector (m)

PAGE 13

13 x j j th direction vector (m) permeability (m 2 ) turbulent energy dissipation viscosity (m 2 s 2 ) porosity viscosity (kg m 1 s 1 ) fluid viscosity (m 2 s 1 ) T eddy viscosity (m 2 s 1 ) fluid density (kg m 3 ) d particle density (kg m 3 ) k Prandl number relating eddy diffusion of k to the momentum eddy viscosity Prandl number relating eddy diffusio

PAGE 14

14 Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Engineering PERFORMANCE MONITORING AND PARTICUL ATE MATTER CFD MODELING OF AN INFRASTRUCTURE PAVEMENT RAINFALL RUNOFF IN NORTHCENTRAL FLORIDA By Gregory Allen Brenner December 2013 Chair: John J. Sansalone Major: Environmental Engineering Sciences Control strategies for nutrients and particulate matter (PM) in runoff from the built environment have received significant attention over the last several decades. The impetus for such strategies has been load and concentration reduction in response to regulato ry drivers such as total maximum daily loads (TMDL) across the USA and more recently, numeric nutrient criteria (NNC) in Florida. Therefore, best management practices (BMPs) are increasingly promoted. For example bio detention is intended to provide hyd rologic control and sequester constituents such as PM, nutrients and metals. One object of this study is the performance monitoring of a bio detention filter system with respect to particulate matter (PM), nitrogen (N), phosphorus (P), various water chemis try measurements, and runoff detention and retention for 12 rainfall runoff events. Also, an economic and performance based analysis of this system is compared with other unit operations from previous studies in similar urban environs. The other objecti ve is to develop a computational fluid dynamics (CFD) model to correctly predict the fate and transport of influent PM through the bio detention filter system. It is hypothesized the model could correctly predict the amount of mass

PAGE 15

15 separated by the system and also the particle size distribution (PSD) of the escaped influent PM. The model is to be validated with all 12 measured rainfall runoff events. The bio detention system in this study has a volume of 2.9 m 3 and is loaded directly by runoff from an asp halt paved roadway in north central Florida. Across the monitoring campaign, on a mass b asis, the system sequestered 69% of total PM and 57 % of suspended PM. The mass based mean PM particle size decreased from 100 m in the influent to 21 m in the effl uent. For the entire campaign, on a mass basis, the system sequestered 32% of TN an d 41% of TP. T he system sequestered 5% of dissolved nitrogen (DN) although effluent DN increased relative to the influent in six events. Similarly, the system sequestered 20% of dissolved phosphorus (DP) but effluent DP increased relative to the influent in five events. T his study assessed the ability of Euler Lagrangain CFD modeling to correctly predict the fate and transport of influent PM mass based on step wise steady state flow rates and PSD in a bio detention filtration system. Validated with the 12 rainfall runoff events, the model predicted influent PM mass separation with a mean of 5 % absolute relative percent difference (ARPD), with half of the events overestim ating and half underestimating the separation of influent PM. Along with mass separated, the escaped PSDs of PM were predicted with a mean of 7% normalized root mean squared error (NRMSE).

PAGE 16

16 CHAPTER 1 GLOBAL INTRODUCTION Urbanization engenders many issue s related to rainfall runoff with impervious surfaces being particularly prevalent reaching up to 80% in the center of urban cities in the United States (Imhoff et al. 2010). Not only do these built environs generate significant loads of particulate matt er (PM) but the imperviousness increases both peak flow rates and volume of runoff by eliminating of infiltration and decreasing evapotranspiration (Department of Environmental Resources 1999). These alterations transport entrained PM on an order of magni tude greater than occurs in a natural environment (Viessman and Lewis 2002) which can significantly degrade the quality of receiving water bodies (Shammaa 2002). Along with increased turbidity, PM provides protective havens for microbes (Dickenson and S ansalone 2009) and adsorbs various pollutants such as heavy metals and nutrients, including nitrogen (N) and phosphorus (P) (Sansalone and Ying 2008; Guo et al. 2004). A growing solution to PM loads in rainfall runoff is low impact development (LID). LID is a decentralized treatment method where rainfall runoff is treated at the source as opposed to traditional methods of one large, regional treatment operation. An LID unit operation (UO) blends with the natural soundings and the treatment mimics na tural hydrology and treatment by using processes such as infiltration (U.S. EPA 2013). Bio detention filtration is an LID solution to decreasing constituent loads in rainfall runoff and consists of plant and soil media confined within a pre cast infrastru cture and, hence, is sub media and the plant life contributes to pollutant uptake. Currently available bio detention filtration UOs incorporate evapotranspiration (ET), soil i nfiltration, filtration,

PAGE 17

17 adsorption, bio transformation, storage and volatilization. Therefore, bio detention filtration is a potential solution to increased nutrients and PM generated and transported in the built environment. With growing regulatory dema nds, such as total maximum daily loads (TMDL), forecasting UO performance to sizing future UOs or to predict treatment capabilities of an existing system in response to a changing environment is crucial. A powerful tool for such forecasting is computation fluid dynamics (CFD). CFD uses numerical methods for modeling physical fluid systems and allows for analysis of more variables than the traditional overflow rate theory; CFD enables modeling of turbulent flows and flow through permeable media. CFD also enables particle tracking and separation within the fluid with respect to gravimetric forces and momentum. In this study we document the monitoring of a bio detention system and compare its performance with other UOs. These results will then be used to calibrate and validate a CFD model of the bio detention system to predict the fate and transport of influent PM.

PAGE 18

18 CHAPTER 2 PHYSICAL MODELING OF NUTREINT AND PARTICULATE MATTER SEQUESTRATION IN AN INFRASTRUCTURE SYSTEM LOADED B Y RAINFALL RUNOFF Introduction Urbanization engenders many issues related to rainfall runoff, but impervious surfaces are particularly prevalent with 297 m 2 of imperviousness per person in the United States (Elvidge et al. 2007). The center of urban citi es can reach up to 80% imperviousness in the US with these numbers growing with population and development increase (Imhoff et al. 2010). Impervious surfaces such as roads, sidewalks, driveways, rooftops, and compacted/densifed soils alter many components of the natural hydrologic cycle by increasing both volume and flow rate of runoff, eliminating infiltration and decreasing evapotranspiration (ET) (Department of Environmental Resources 1999). These hydraulic modifications increase conveyance of runoff, decrease lag times, and increase leaching and transport of soluble and particulate constituents. A hydrographic comparison between a pre developed area and a built environ with the same drainage area shows three main differences: greater peak discharge, increased runoff volume, and decreased time to peak or lag time. Further, it has been observed that urbanization increases the rate of peak discharge more rapidly than the increased volume generated (Viessman and Lewis 2002). These differences from the natural hydrologic cycle contribute to significant depletion of surficial groundwater supply, to increased flooding, and to deterioration of surface waters. Imperviousness directly quantifies the degree of hydrologic modification and indirectly quantifie s delivery of anthropogenic constituent mass generated/transported in the built environs.

PAGE 19

19 Indeed, these impervious surfaces and infrastructure generate and transport significant loads of dissolved, colloidal and suspended particulate matter (PM) in a com plex heterogeneous mixture that includes metal elements and organic and inorganic compounds (U.S. EPA 2013b). Since constructed impervious surfaces and drainage systems are designed to provide more rapid conveyance than in natural and porous systems, tra nsport of entrained PM can be an order of magnitude greater than particulate transport in a natural environment (Viessman and Lewis 1996), which can significantly degrade the quality of surface waters (Shammaa 2002). Along with the increase in turbidity, PM provides protective havens for microbes (Dickenson and Sansalone 2009) and adsorbs various pollutants such as heavy metals and nutrients, including nitrogen (N) and phosphorus (P) (Sansalone and Buchberger 1997). Rainfall runoff levels of the metal e lements Zn, Cu, Cd, Pb, Cr, Ni and Hg as well as nutrients in built environs are significantly greater than ambient background levels and, for many land uses, often exceed surface water discharge (Gnecco et al. 2006). Significant amounts of N can be trac ed to natural and anthropogenic sources, including atmospheric deposition, detritus, and fertilization. P sources are much the same as N but with less significant atmospheric deposition. Both nutrients are particularly problematic with eutrophication, as most freshwater systems are phosphorus limited and marine are nitrogen limited (Correll 1998 ; Oviatt et al. 1995). The increase in the limiting nutrient in the water body promotes algal blooms, which not only diminishes aesthetical value but also dissol ved oxygen (D.O.), leading to hypoxia, a serious detriment to the ecosystem (Kofinas and Kioussis 2003).

PAGE 20

20 A growing solution to pollutant loads in rainfall runoff has been low impact development (LID). LID is a decentralized treatment method where rainfa ll runoff is treated at the source as opposed to traditional methods of one large, regional treatment operation. The unit operations blend with the natural soundings and the treatment, mimicking natural hydrology and treatment, uses processes such as infi ltration (U.S. EPA 2013a). Bio detention filtration is an LID solution to pollutant loads in rainfall runoff. This unit operation consists of plant and soil media confined within a pre cast infrastructure and, hence, is sub infiltrates through the media and the plant life contributes to pollutant uptake. Currently available bio detention filtration unit operations incorporate evapotranspiration (ET), soil infiltration, filtration, adsorption, bio t ransformation, storage and volatilization. Although with the incorporation of plant life, there is the potential for macropore formation due to the roots, which increase infiltration rate and short circuit runoff (Lucas 2010). A notable aspect of this the system, which serves as a temporary storage that attenuates peak flow rate and increases lag time to eliminate flooding in the urban setting. Bio detention filtration has an aesthetic a dvantage of being concealed inside the gutter with top soil and plant life being the only evidence of its presence. Therefore, bio detention filtration is a potential solution for nutrients and PM generated and transported in the built environment and hyd rologic restoration. Objectives The objects of this study are to monitor the performance of a bio detention filter system loaded by a paved roadway in north central Florida. The performance with respect to PM, nitrogen, phosphorus, pH, DO, oxidation reduc tion potential (Redox),

PAGE 21

21 Conductivity, total dissolved solids (TDS) and runoff detention and retention will be monitored for 12 rainfall runoff events with rainfall depth greater than 0.25 mm (0.1 in). The constituents will be observed through total mass a nd event mean concentrations/values (EMC/EMV) to arrive at percent separation (PS) and change in concentration/value when applicable. Also, the economics and performance of this system will be compared with hydrodynamic separators, a volumetric clarifying filter, a radial cartridge filter, and street sweeping from previous studies in similar urban environs. Methodology Source Area and Bio detention Filter System The system monitored is located on a right of way (ROW) of a small urban watershed in north cen tral Florida (NE 19th Terrace between NE 8 th and 10 th Avenue in Gainesville). The system was loaded by rainfall runoff from one paved lane (including a bike lane) and curb/gutter of a two lane asphalt paved street running north south (NE 19 th Terrace) whi ch connects a residential neighborhood to a retail development area that receives an average daily traffic (ADT) load of approximately 4,000 vehicles per day. The back to back of curb width of the street is 9.8 m (32 ft). The drainage area for the monito red system is approximately 365 m 2 (3,800 ft 2 ), depending on rainfall and wind intensity and direction. Between the back of the pavement curb and the ROW is a 1.5 m (5 ft) wide concrete sidewalk separated from the curb by 2.0 m (6.5 ft) of grass area. Th e areas beyond the ROW are single family residential housing on the west side and vegetated wooded area on the east side (Figure 2 1 ). On the eastside of the pavement centerline, the system was loaded by runoff from the northbound lane that drains by gravi ty to the south towards the monitored

PAGE 22

22 system. This runoff, as sheet flow, drained diagonally to the southeast towards the eastside curb and gutter and continued as concentrated gutter flow to the curb opening that led to the inlet of the monitored system. From the curb opening, a semi circular polyvinylchloride (PVC) channel with 2% slope transported the concentrated flow to a 50.8 mm (2 in) Parshall flume, where depth of flow was monitored. Stormwater flowed out of the Parshall flume and cascaded direct ly into a drop box that allowed manual sampling of full cross sectional influent. The outflow from the drop box discharged into the inlet section of the system. Once inside the system, the influent runoff flowed down through a vertical rectangular 1.8 x 2.4 m (6 x 8 ft) soil column that was constrained in the precast concrete walls and floor. Effluent was discharged as concentrated flow through the precast structure into a 102 mm (4 in) PVC pipe followed by an effluent 25.4 mm (1 in) Parshall Flume for fl ow measurement and sampling. From the Parshall Flume; effluent flowed by gravity into a storm sewer. The conveyance system and filter is illustrated in Figure 2 2 Hydrologic Measurement Rainfall depth was measured using a calibrated tipping bucket with a resolution of 0.25 mm (0.01 in) equipped with a data logger to record data during each rainfall event. The rainfall intensity was found by dividing the rainfall depth by the duration of time for that depth. The tipping bucket was placed and leveled in a horizontal direction and was situated approximately 6.1 m (20 ft) south of the system on the sidewalk to avoid tree canopy influences and streetlight interference with rainfalls. The influent flow rate was measured using a 50.8 mm (2 in) Parshall flume calibrated in a previous study (Sansalone et al. 2009). The stage discharge relationship for the Parshall flume is give n by the following calibration E quation 2 1

PAGE 23

23 ( 2 1) For this stage d ischarge relationship, using the units as measured and calibrated, Q represents the flume discharge in units of L/s and H is the water stage in units of inches. The effluent flow rate was measured using a 25.4 mm (1 in) calibrated Parshall flume. The stag e discharge relationship for the Parshall flume is give n by the following calibration E quation 2 2 ( 2 2) For this stage discharge relationship, using the units as measured and calibrated, Q repr esents the flume discharge in units of L/s and H is the water stage in units of inches. Influent and Effluent Sampling Replicate samples were taken throughout each rainfall runoff event and were collected in 1 L polypropylene wide mouth bottles. Influent and effluent samples were collected from the free fall effluent of each Parshall flume. For the 12 events, the number of influent and effluent samples ranged from 6 to 11 samples depending on duration of rainfall runoff event. For each sampling set at a given sampling time, 6 replicate bottles were coll ected for PSDs, PM and nutrient fractions. These samples were composited based on flow volume to generate event mean concentrations (EMCs) and event based influent and effluent mass loads. Event Mean Conce ntration Due to varying concentrations throughout rainfall runoff events, a single index know as an EMC defined by Sansalone and Buchberger (1997) was used to characterize event concentrations of the constituents. EMCs represent the runoff

PAGE 24

24 volume based av erage of the constituent and is defined by the total pollutant load (mass) divided by the total event runoff volume as follows. ( 2 3) where M is the total mass of constituent over entire event duration (M); V is the total volume of flow over entire event duration (L 3 ); is the flow weighted average concentration for entire event (M/L 3 ); tr is the event duration; c(t) is the time variable dissolved or par ticulate bound concentration (M/L 3 ); q(t) is the time variable flow, (L 3 /T); and t is the time (T). Some constituents are not expressed as a concentration, but rather a value, such as pH. Event mean values (EMV) are defined as above with value instead of concentration. Water Chemistry All the water chemistry parameters were measured upon arrival to the laboratory after the rainfall runoff event and conducted with calibrated electrodes. pH, D.O., oxidation reduction potential, and conductivity and total di ssolved solids (TDS) were measured in accordance with Standard Methods 4500 H + B, 4500 O G, 2580, and 2510, respectively (American Public Health Association 1995). Particulate Matter Replicate 1 L samples were analyzed for PSD using laser diffraction and Mie light scattering theory with a Malvern Mastersizer 2000 Hydro 2000G (Malvern 2013). Volumetric PSDs were converted to % finer by mass with the use of their total PM values. Incremental percent volume for each particle size was multiplied by the tota l PM value and then by the total volume of sample to arrive at % mass. PSD results

PAGE 25

25 were also simplified as d 10 d 50 and d 90 values based on a cumulative mass distribution for the PSD. In this study PM was measured as three fractions of suspended PM (< 25 m), accounts for the entire gradation of PM. To ensure all PM mass in a sample was accounted for, a 1 L bottle was used for PM separation as opposed to taking a sub aliquot o f the sample. All PM was dried in an aluminum weighing dish at 105 o C for 24 h. From the total PM, separate fractions of PM can be identified by mechanism or size fractionation depending on reporting or mechanistic requirements. For example, sediment P M was separated out and by passing a precisely measured, nominal 1 L sample through a 75 m sieve (No. 200 Fisher Scientific U.S.A. Standard Testing Sieve) (Sansalone and Kim 2008). Settleable PM can be collected following Standard Methods 2540 F (Americ an Public Health Association 1995). After a sample is passed through the 75 m sieve, the sample is allowed to settle in a 1 L Imhoff Cone for 60 minutes and the settled PM mass at the bottom of the Imhoff Cone represents for the settleable PM. This sett leable PM is typically initially measured as mL of PM per volume (nominally 1 L). This result can then be converted to a gravimetric concentration of settleable PM as mg/L through dry mass of settled material, specific gravity, and known sample volume (Sa nsalone and Kim 2008). Suspended PM was determined from the settled (at 60 minutes) supernatant (Standard Methods 2540 D) (American Public Health Association 1995). The supernatant in the Imhoff Cone was passed through a 0.45 m glass fiber filter usin g a vacuum filtration apparatus to collect

PAGE 26

26 the suspended PM mass on the filter. Therefore, total PM can be further fractionated into sediment, settleable and suspended PM fractions for further examination. In this study, the volatile fraction was quanti fied in accordance with Standard Methods 2540 E (American Public Health Association 1995) by placing dried, measured PM in a muffle furnace at 550 o C for 60 minutes to allow for the volatile fraction to be converted to the gas phase and released. The vol atile fraction is found by measuring the mass lost in the furnace relative to the dry mass initially collected on the filter. Phosphorus and Nitrogen PM mass collected from suspended, settleable and sediment PM was recovered, prepared and analyzed for P an d N. The stormwater supernatant from the Imhoff Cone, generated as filtrate that had passed through the filter during the suspended PM was collected for dissolved P and dissolved N analysis. The summation of all nutrient PM fractions accounted for the to tal concentration of each nutrient (Sansalone and Kim 2008 ; and Berretta and Sansalone 2012). PM associated and dissolved P fractions were analyzed through spectrophotometry. PM associated P was first digested according to Standard Methods 4500 P B Aci d Hydrolysis (American Public Health Association 1995) to convert all phosphorus into orthophosphates as required to be able to react with the reagent for spectrophotometry. Dissolved P was analyzed as soluble reactive phosphorus (SRP), identified as tot al dissolved P (DP). PM associated and dissolved nitrogen content was determined through persulfate digestion and spectrophotometry in accordance with Standard Methods 4500 N C (American Public Health Association 1995).

PAGE 27

27 Results and Discussion Hydrology F or all 12 rainfall runoff events the hydrologic data recorded is presented with hyetographs and influent and effluent hydrographs which display total rainfall depth, rainfall intensity, and flow rates as a function of time. Only storms with a rainfall dep th of 2.5 mm (0.10 in) or greater were sampled. Each rainfall runoff event produced a total rainfall depth that exceeded 3.0 mm (0.12 in), met the required 2.5 mm of rainfall depth, and produced sufficient influent and effluent total volume for system mon itoring. The system was designed to be on line and therefore intervening and unmonitored events between monitored events continued to load the system. The duration of the monitored events ranged from 68 to 1,034 min and influent event volume treated range d from 207 to 13,882 L. The system only minimally attenuated the flows with a mean influent flow rate of 12.5 L/min and mean effluent flow rate of 11.6 L/min. The probability density function (pdf) of influent and effluent flows are shown in Figure 2 3. The soil based system served as a slight storage of runoff as it increased soil moisture ranging from 14 to 152 L per event. Event based hydrology summaries are shown in Table 2 yetograph a re shown in the Appendix in Figures A 2 and A 3. Particulate Matter For all 12 rainfall runoff events, the PSDs were summarized with the % finer by mass d 10 d 50, and d 90 values and modeled to a gamma distribution with parameters shown in Table 2 2 and A 9 The system decrease d d 50 values from a mean of 100 to 21 m. As expected, the higher peak flow rates had the tendency to increase the size

PAGE 28

28 of influent PSD, but there was no observed influence of influent PSD or flow rate on effluent PSD. For all 12 rainfall runoff events, t otal mass and EMCs for all PM fractions were measured and summarized in Table 2 3 and shown in detail in Table A 10 and A 1 1 and 78%, respectively. With the volatile s eparation being greater than total PM, infers a positive attribute that the system collects OM and releases mineral PM that would help reduce biochemical oxygen demand (BOD). Figure 2 3 displays the probability density function (pdf) of the varying PM EMC s throughout all the events and Figure 2 5 displays the cumulative PM mass as a function of treated rainfall runoff. Phosphorus For all 12 rainfall runoff events DP, PM fraction associated P, and TP was measured and summarized in Table 2 3 and shown in det ail in Tables A 12 and A 13 as were 20, 40 and 41%, respectively. The DP EMC increased from the system in 5 events which also had the longest duration. Either, the longer the system was saturated, the more likely it would release existing P into the passing runoff, or transfer influent PM associated P into the dissolved phase. Because of the system ability to store a portion of the runoff, the DP mass was only increased by the system in one event with a negative PS of 8%. Figure 2 4 displays the pdf of the varying P concentrations throughout all the events and Figure 2 6 displays the cumulative P mass as a function of treated rainfall runoff.

PAGE 29

29 Nitrogen For all 12 rainfall runo ff events DN, PM fraction associated N, and TN was measured and summarized in Table 2 3 and shown in detail in Tables A 14 and A 15 respectively. The DN EMC increased from the system in 6 events, and again due to the storage, DN mass was increased in 4 events. No correlations of increased nitrogen to other parameters were observed. The suspended PM associated N EMC even increased in one event, but all events displayed positive separation of suspended PM associated N mass. Figure 2 4 displays the pdf of the varying N concentrations throughout all the events and Figure 2 7 displays the cumulative N mass with treated rainfall runoff. Water Chemistry For 11 of the rainfall runoff events water chemistry for pH, D.O., redox, conductivity, and TDS were monitored and are show in Table 2 4. The pH EMV decreased in value for all events measured but two with a mean influent EMV of 7.5 and effluent of 7.2. No observable trend in redox al teration was found with a volume based mean influent EMV of 313 mV and effluent 342 mV. As with the relationship between redox and D.O., no system alteration was observed. The influent volume based mean EMC of D.O. was 6.2 mg/L and effluent was 5.5 mg/L. Although, since redox overall increased, D.O. should have increased as well. It should be noted that redox includes all oxidizing and reducing agents not just D.O, and since DN levels were shown to increase from the system, it could be inferred that the system will decreases D.O. but increase NO 3 more for an overall increase in redox. The EMCs of conductivity and TDS increased from influent to effluent in all events measured. The volume based mean

PAGE 30

30 of conductivity increased from 46 to 64 S/cm and TDS in creased from 31 to 44 mg/L. The pdf of all the water chemistry parameters are shown in Figure 2 3, A 8 and A 9. Economics A treatment performance based analysis and economic evaluation for this system is compared to runoff unit operations from previous st udies on similar types of source areas either in Florida or Louisiana and commonly utilized for source area treatment. The systems tested with equal or more extensive monitoring campaigns were an SHS (screened hydrodynamic separator, HS providing PM based separation) in Baton Rouge (BTR), a VF (volumetric filter providing detention, sedimentation and adsorptive filtration, AF), a BHS (baffled hydrodynamic separator, HS providing PM based separation) in Gainesville (GNV), a URF (upflow radial filter providi ng AF) in GNV, a VCF (vertical cartridge filter providing PM based separation) in GNV and the bio detention system of this study providing detention, sedimentation and AF. The economics of each of these systems, in terms of $/kg treated and $/m 3 of influ ent runoff treated are compared and then compared to the economics of street sweeping as a maintenance practice to recover PM and nutrients from a multi year study that the City of Gainesville participated with 13 other MS4s from around Florida. The cost of each UO is based on capital and construction costs with on a 25 year design life irrespective of maintenance costs. While the VF in BTR did provide nominal peak flow and volume attenuation none of the UO can be considered to provide water quantity cont rol and certainly not hydrologic restoration benefits; these units are potential load reduction units with regular maintenance required. The system of this study is constrained in a concrete structure and as such only provided nominal peak flow and volume attenuation; hence the term bio detention. While load reduction is the focus of this

PAGE 31

31 stu dy and UO comparison of Table A 2 recognition that urban hydrology drives urban water chemistry and loads is critical when considering the treatment viability, econo mics and sustainability of any UO or maintenance practice. While street sweeping does not provide hydrologic restoration, the practice removes PM and constituent loads (metals, transport by rainfall runoff. As summarized in Table A 8 the system of this study (infrastructure constrained bio detention) treated the smallest drainage source area and was loaded by the lowest medium influent flow rate, while providing the largest U O volume and surface area; and therefore the lowest surface loading rate (SLR) of all UO compared in this study. The PS (based on the influent mass) was 2 nd lowest for PM and TN and in the lowest in % separation of all UO for TP Economically, the system provided a n above average $/kg and $/m 3 cost for PM, TP and TN. If the economics of the system are compared to the current maintenance practice of street sweeping, for which the City of Gainesville participated, the $/kg cost of the constrained bio deten tion system is on order of magnitude higher for PM TP and TN. It is noted that while the typical Florida BMP (a retention basin) provides much needed hydrologic control, the $/kg cost as a treatment system are an order of magnitude higher than street swe eping. Conclusions The performance of a constrained bio detention system was monitored from January to August 2012 over 12 rainfall events resulting in a total of 240 mm of rainfall runoff from one newer asphalt paved lane of a moderately sized college tow n in north central Florida that serves as a connector between a large retail store and a local residential neighborhood. The drainage area to the system was approximately 365 ft 2

PAGE 32

32 depending on local meteorological parameters of the rainfall runoff event. F or the entire monitoring campaign the system separated a mean of 69% of total PM and 57% of suspended PM, and reduced PSD from a mean of 100 to 21 m. The mean separation by the system for TN and TP was 33% and 41%, respectively. The largest area of conc ern with the bio detention system is the dissolved phase, questioning the bioaccumulation potential of the system. Six events showed an increase in DN concentration; five events showed an increase in DP concentration, in particular the longer duration eve nts; and all events showed an increase in TDS and conductivity. An analysis of various unit operations shows the bio detention filtration with above average economic performance. Although, Street sweeping has proven to be the most suitable solution to con stituent loads in stormwater. Unfortunately it is not coupled with important hydrologic restoration, but it does display that addressing the constituents before they become suspended in rainfall runoff is the more feasible and most efficient approach.

PAGE 33

33 T able 2 1. Hydrologic data for all 12 monitored rainfall runoff events during the campaign with initial pavement residence time (IPRT), previous dry hours (PDH), and influent and effluent system flow rate characteristics. Statistical values are derived from volume weighted event mean values (EMV) Event Date (2012) Rainfall Depth IPRT PDH Influent Effluent Volume Mean Median Peak Volume Mean Median Peak Flow Flow Flow Flow Flow Flow (mm) (min) (hrs) (L) (L/min) (L/min) (L/min) (L) (L/min) (L/ min) (L/min) 18 Jan 3.81 34 1 287 1.4 1.9 3.1 198 1.0 1.0 2.7 22 Feb 3.56 19 149 208 2.0 2.1 4.2 176 1.6 1.4 3.8 21 Apr 3.56 5 26 256 1.7 1.1 5.4 210 1.4 0.9 4.0 16 May 4.32 44 26 345 2.8 2.1 8.2 293 2.6 2.4 8.0 29 May 18.54 5 12 2267 8.9 6.1 44.7 225 7 8.9 7.9 28.2 7 Jun 14.99 27 39 2071 8.8 6.0 86.5 1984 8.2 8.0 39.7 14 Jun 19.30 39 94 2273 11.7 7.8 88.3 2121 10.5 8.5 29.9 24 Jun 144.78 45 69 13882 14.2 10.9 49.8 13800 13.5 12.7 45.6 10 Jul 11.43 3 117 1099 21.1 3.1 100.9 984 17.6 3.6 84.3 13 Jul 8.38 5 26 495 11.3 3.7 87.5 481 7.6 1.8 52.7 1 Aug 3.30 5 42 230 4.3 4.2 19.2 192 2.7 0.8 11.4 6 Aug 3.56 8 24 264 5.1 3.7 18.4 211 4.0 2.8 11.7 Mean 90.78 20 52 8820 12.5 8.7 56.7 8739 11.6 10.0 41.3 Median 144.78 45 69 13882 14.2 10.9 49.8 13800 13. 5 12.7 45.6 Std. Dev. 20.12 16 44 6035 3.8 3.0 21.9 6032 3.5 3.7 14.5

PAGE 34

34 Table 2 2. Influent and effluent particle size distributions (PSD) for all 12 monitored rainfall runoff events during the campaign as event mean values (EMV). The d 50 is the diamet er in which 50% of the total mass has diameter less than this size. Measured PSDs are represented with modeled gamma distributions values are derived from volume weighted EMVs Event Date (2012) Influent P SD Effluent PSD d 50 Gamma Distribution d 50 Gamma Distribution (m) (m) 18 Jan 32.1 0.60 148.61 6.0 0.80 14.90 22 Feb 71.8 0.59 250.09 11.6 0.48 62.16 21 Apr 104.6 0.60 513.69 52.1 0.77 141.49 16 May 152.3 0.56 456.47 22.3 0.72 53.31 29 M ay 151.1 1.17 228.88 32.7 0.98 86.52 7 Jun 47.1 0.59 206.59 33.8 0.82 40.99 14 Jun 95.6 0.68 230.15 33.5 1.68 24.18 24 Jun 101.9 0.84 200.95 14.8 0.87 43.71 10 Jul 82.2 0.78 353.02 30.4 1.01 43.76 13 Jul 143.3 0.82 253.57 19.5 1.75 13.25 1 Aug 77.5 0 .67 167.46 14.8 0.81 71.27 6 Aug 20.0 0.61 186.34 20.0 0.93 31.10 Mean 99.7 0.82 221.50 21.2 0.97 46.20 Median 101.9 0.84 200.95 14.8 0.87 43.71 Std. Dev. 27.1 0.15 54.31 9.1 0.27 19.02

PAGE 35

35 Table 2 3. Event PM and nutrient fraction masses and percent r emovals (PR) for all 12 monitored rainfall mass throughout monitoring campaign. One event for dissolved phosphorus and four events for dissolved nitrogen have negative PRs indicating a net increase in constituent mass as the runoff passes through the system Event Date (2012) 18 Jan. 22 Feb. 21 Apr. 16 May. 29 May. 7 Jun. 14 Jun. 24 Jun. 10 Jul. 13 Jul. 1 Aug. 6 Aug. Total PM Suspended M i (g) 11.4 8.8 12.5 8.7 70.6 62.9 56.1 543.5 60.2 24.6 14.5 9. 2 873.7 M e (g) 5.8 2.4 2.6 6.7 44.6 24.7 31.8 213.3 19.2 20.8 4.5 4.2 376.6 PR (%) 48.7 72.6 78.9 22.3 36.8 60.7 43.2 60.8 68.2 15.5 69.0 54.8 56.9 Total M i (g) 15.5 11.3 29.4 33.2 240.4 289.0 166.6 924.7 323.8 132.4 67.7 12.0 2234.2 M e (g) 6.8 3.1 5.1 13.1 70.7 48.0 42.0 424.0 45.0 27.3 7.2 4.7 692.0 PR (%) 56.3 72.8 82.8 60.6 70.6 83.4 74.8 54.1 86.1 79.4 89.4 61.0 69.0 Volatile M i (g) 9.2 6.1 12.6 16.3 175.1 185.4 78.3 499.9 201.0 71.7 26.7 7.8 1290.1 M e (g) 3.8 1.9 2.8 7.6 18.4 26.2 7 .0 196.3 11.7 2.1 9.9 2.4 290.2 PR (%) 59.2 68.0 77.8 53.4 89.5 85.8 91.0 60.7 94.2 97.1 62.9 69.3 77.5 Phosphorus Dissolved M i (mg) 174.0 172.2 203.8 155.3 417.9 321.6 659.4 3707.5 1326.2 213.5 94.9 62.8 7446.4 M e (mg) 165.1 117.8 141.5 61.8 214.3 261.1 218.7 4013.3 559.0 155.1 62.6 36.5 5970.3 PR (%) 5.1 31.6 30.6 60.2 48.7 18.8 66.8 8.2 57.8 27.4 34.0 41.8 19.8 Suspended M i (mg) 155.4 122.9 133.7 140.5 478.1 667.8 1037.8 3858.4 267.5 420.2 131.4 70.2 7413.7 M e (mg) 88.9 7 1.3 60.4 71.2 222.5 217.3 704.3 2602.9 229.3 120.0 29.9 51.7 4418.0 PR (%) 42.8 42.0 54.8 49.3 53.5 67.5 32.1 32.5 14.3 71.4 77.2 26.4 40.4 Total M i (mg) 338.6 325.5 418.5 356.7 1583.2 1829.5 1989.1 8423.8 2200.6 949.4 390.3 141.8 18805.1 M e (mg) 2 56.2 193.0 216.0 142.2 502.9 695.6 935.1 6813.3 868.5 373.6 100.8 89.6 11097.2 PR (%) 24.3 40.7 48.4 60.1 68.2 62.0 53.0 19.1 60.5 60.6 74.2 36.8 41.0 Nitrogen Dissolved M i (mg) 72.8 185.5 276.5 291.9 509.9 603.5 604.4 5288.0 1782.2 27 8.8 216.3 156.6 10109.7 M e (mg) 67.4 195.2 260.5 146.6 431.7 720.9 1034.3 4811.6 1487.8 349.7 132.2 95.6 9638.0 PR (%) 7.5 5.2 5.8 49.8 15.3 19.5 71.1 9.0 16.5 25.4 38.9 39.0 4.7 Suspended M i (mg) 59.2 59.5 207.1 107.2 502.4 1072.4 1260.9 6143. 5 2165.8 107.8 501.4 95.6 12187.3 M e (mg) 23.7 58.0 151.4 69.7 449.6 476.5 795.3 4815.2 1377.1 86.9 252.7 43.3 8556.3 PR (%) 60.0 2.5 26.9 35.0 10.5 55.6 36.9 21.6 36.4 19.4 49.6 54.7 29.8 Total M i (mg) 159.1 262.7 520.8 552.5 1614.0 3083.5 2477.0 12376.0 5279.0 875.0 1350.6 272.9 28550.1 M e (mg) 96.7 255.5 421.2 247.2 953.2 1283.8 1859.5 10253.3 2996.3 501.4 649.0 143.4 19517.1 PR (%) 39.2 2.7 19.1 55.2 40.9 58.4 24.9 17.2 43.2 42.7 52.0 47.5 31.6

PAGE 36

36 Table 2 4. Water Chemistry Measurements f or 11 monitored rainfall runoff events during the campaign as event mean values (EMV) or event mean concentrations (EMC). Statistical values are derived from volume weighted EMVs or EMCs Event Date (2012) pH Redox (mV) D.O. [mg/L] Conductivity (S/cm) TDS [mg/L] EMV i EMV e EMV i EMV e EMC i EMC e EMC i EMC e EMC i EMC e 18 Jan 7.58 7.41 415.06 423.01 8.22 8.25 73.76 94.85 49.42 94.85 22 Feb 7.60 7.60 389.82 441.40 7.51 8 .82 86.81 117.97 58.16 79.04 21 Apr 7.69 7.36 359.80 375.19 6.95 7.43 109.90 126.10 73.64 84.49 16 May 7.34 6.94 465.45 534.27 6.45 5.65 58.90 79.86 39.46 53.51 29 May 7.52 7.30 176.37 177.63 5.59 5.49 99.93 144.52 66.95 96.83 7 Jun 7.70 7.14 286.40 44 6.04 5.86 4.88 53.53 69.29 35.87 46.43 24 Jun 7.50 7.12 317.84 336.20 6.33 5.34 31.84 47.09 21.33 31.55 10 Jul 7.23 7.48 336.98 337.67 5.35 5.35 46.81 58.60 31.36 39.26 13 Jul 6.72 6.60 474.42 460.92 7.65 8.43 61.06 71.25 40.91 47.74 1 Aug 7.88 7.41 41 4.90 438.16 5.33 5.03 55.73 78.69 37.34 52.72 6 Aug 6.76 6.55 507.18 482.03 6.90 7.45 68.09 90.87 45.62 60.88 Mean 7.49 7.15 312.91 341.91 6.23 5.51 45.80 64.42 30.68 43.59 Median 7.50 7.12 317.84 336.20 6.33 5.34 31.84 47.09 21.33 31.55 Std. Dev. 0.18 0.16 62.47 73.61 0.49 0.75 23.09 31.30 15.47 21.68

PAGE 37

37 Figure 2 1. The contributing drainage area schematic for the eastside filter of NE 19 th Terrace between NE 8 th and 10 th Avenue in Gainesville, Fl. Runoff flows southeast in the northbound lane towa rds the system inlet. Area extends from centerline of road to back of curb, 5 m (16 ft) for an approximate drainage area of 365 m 2 (3,800 ft 2 )

PAGE 38

38 Figure 2 2. Side view of the bio detention filter system and flow monitoring syst em. The soil media occupies 2.85 m 3 (101 ft 3 )

PAGE 39

39 Figure 2 3. Probability density functions (pdf) of influent and effluent hydrologic flow rates, PM fractions of total suspended and total PM, total dissolved solids (TD S) and pH throughout the events for the entire monitoring campaign. The pdf model curves were generated using a 3 parameter, peak, lognormal regression. The mean is , and the

PAGE 40

40 Figure 2 4. Probability density functions (pdf) of influent and effluent dissolved phosphorus (DP), suspended PM associated phosphorus (SP), and total phosphorus (TP); and dissolved nitrogen (DN), suspended PM associated nitrogen (SN), and total nitrogen (TN) fractions throughout the events for the entire monitoring campaign. The pdf model curves were generated using a 3 parameter, peak, lognormal regression. The mean is , and the standard

PAGE 41

41 Figure 2 5. Cumulative influent and effluent particulate matter (PM) fractions throughout the Gainesville, Fl. bio detention filter monitoring campaign by elapsed intra event time

PAGE 42

42 Figure 2 6. Cumulative influent and effluent phosphorus (P) fractions throughout the Gainesville, Fl. bio detention filter monitoring campaign by elapsed intra event time

PAGE 43

43 Figure 2 7. Cumulative i nfluent and effluent phosphorus (N) fractions throughout the Gainesville, Fl. bio detention filter monitoring campaign by elapsed intra event time

PAGE 44

44 CHAPTER 3 COMPUTATIONAL FLUID DYNAMICS MODELING OF PARTICULATE MATTER IN URBAN RAIN FALL RUNOFF TREATMNET T HROUGH A BIO DETENTION FILTER SYSTEM Introduction Urbanization engenders many issues related to rainfall runoff with impervious surfaces being particularly prevalent reaching up to 80% in the center of urban cities in the United States (Imhoff et al. 2010 ). Not only do these built environs generate significant loads of particulate matter (PM) but the imperviousness increases both peak flow rates and volume of runoff by eliminating of infiltration and decreasing evapotranspiration (Department of Environmen tal Resources 1999). These alterations transport entrained PM on an order of magnitude greater than occurs in a natural environment (Viessman and Lewis 2002) which can significantly degrade the quality of receiving water bodies (Shammaa 2002). Along wi th increased turbidity, PM provides protective havens for microbes (Dickenson and Sansalone 2009) and adsorbs various pollutants such as heavy metals and nutrients, including nitrogen (N) and phosphorus (P) (Sansalone and Ying 2008 ; Guo et al. 2004). A growing solution to PM loads in rainfall runoff is low impact development (LID). LID is a decentralized treatment method where rainfall runoff is treated at the source as opposed to traditional methods of one large, regional treatment operation. An LI D unit operation (UO) blends with the natural soundings and the treatment mimics natural hydrology and treatment by using processes such as infiltration (U.S. EPA 2013). Bio detention filtration is an LID solution to decreasing PM loads in rainfall runof f, consists of plant and soil media confined within a pre cast infrastructure and, hence, is sub

PAGE 45

45 the plant life contributes to pollutant uptake. Currently available bio detention filtration UOs incorporate evapotranspiration (ET), soil infiltration, filtration, adsorption, bio transformation, storage and volatilization. Therefore, bio detention filtration is a potential solution to increased nutrients and PM generate d and transported in the built environment. With growing regulatory demands, such as total maximum daily loads (TMDL), forecasting UO performance to sizing future UOs or to predict treatment capabilities of an existing system in response to a changing envi ronment is crucial. A powerful tool for such forecasting is computation fluid dynamics (CFD). CFD uses numerical methods for modeling physical fluid systems and allows for analysis of more variables than the traditional overflow rate theory; CFD enables modeling of turbulent flows and flow through permeable media. CFD also enables particle tracking and separation within the fluid with respect to gravimetric forces and momentum. This study use s CFD modeling of PM separation by a bio detention filtration system that incorporates filtration and settling as the mechanism of separation. Traditional observations of PM have only consisted of a few categories, including total suspended solids (TSS) and settleable PM. In this study we observe PM in depth in the form of particle size distributions (PSD) as percent finer by mass. This allows for the entire particle size gradation and quantities to be observed. Objectives The objectives of this CFD model study were to correctly predict the fate and transport of influent PM through the bio detention filter system. We hypothesized the model could correctly represent the amount of mass separated by the system with respect to influent loadings and also, with respect to the PM escaped from the system,

PAGE 46

46 correctly match the PSD to the measured, escaped PM PSD. The model is to be validated with all of the 12 measured rainfall runoff events from January to August 2012. Methodology Source Area and Bio detention Filter System The system used to validate the model was locat ed on a right of way (ROW) of a small urban watershed in north central Florida (NE 19th Terrace between NE 8 th and 10 th Avenue in Gainesville). The system was loaded by rainfall runoff from one paved lane (including a bike lane) and curb/gutter of a two l ane asphalt paved street running north south (NE 19 th Terrace) which connects a residential neighborhood to a retail development area that receives an average daily traffic (ADT) load of approximately 4,000 vehicles per day. The back to back of curb width of the street is 9.8 m (32 ft). The drainage area for the monitored system is approximately 365 m 2 (3,800 ft 2 ), depending on rainfall and wind intensity and direction. Between the back of the pavement curb and the ROW is a 1.5 m (5 ft) wide concrete sid ewalk separated from the curb by 2.0 m (6.5 ft) of grass area. The areas beyond the ROW are single family residential housing on the west side and vegetated wooded area on the east side (Figure 2 1). On the eastside of the pavement centerline, the system wa s loaded by runoff from the northbound lane that drains by gravity to the south towards the monitored system. This runoff, as sheet flow, drained diagonally to the southeast towards the eastside curb and gutter and continued as concentrated gutter flow to the curb opening that led to the inlet of the monitored system. From the curb opening, a semi circular polyvinylchloride (PVC) channel with 2% slope transported the concentrated flow to a 50.8 mm (2 in) Parshall flume, where depth of flow was monitored. Stormwater flowed

PAGE 47

47 out of the Parshall flume and cascaded directly into a drop box that allowed manual sampling of full cross sectional influent. The outflow from the drop box discharged into the inlet section of the system. Once inside the system, the i nfluent runoff flowed down through a vertical rectangular 1.8 x 2.4 m (6 x 8 ft) soil column that was constrained in the precast concrete walls and floor. Effluent was discharged as concentrated flow through the precast structure into a 102 mm (4 in) PVC p ipe and into an effluent structure for sampling; this structure was located in a precast access structure. From the precast access structure; effluent flowed by gravity into a storm sewer. The conveyance system and filter is illustrated in Figure 2 2. Hydr ologic Measurement Rainfall depth and intensity was measured at the field site using a calibrated tipping bucket equipped with a data logger to record data during each rainfall event. Only events totaling in 2.54 mm (0.1 in) of rainfall depth were analyze d to ensure sufficient runoff and influent and effluent flows. Influent and effluent flows were measured with calibrated parshall flumes and data logger equipped ultrasound sensors to record water stage in the flumes. Particulate Matter Replicate 1 L sampl es were analyzed for PSD using laser diffraction and Mie light scattering theory with a Malvern Mastersizer 2000 Hydro 2000G (Malvern 2013). Volumetric PSDs were converted to % finer by mass with the use of their Total PM values. Incremental percent vol ume for each particle size was multiplied by the Total PM concentration and then by the total volume of sample to arrive at % mass. PSD results were also simplified as d 10 d 50 and d 90 values based on a cumulative mass distribution for the PSD. Where 1 0, 50, and 90 % of the particle mass is finer than that

PAGE 48

48 size, respectively. Also PSDs were modeled with gamma distributions to express the In this study PM was also measured as three fractions of suspended PM (< 25 accounts for the entire gradation of PM. Event Mean Concentration Due to varying concentrations throughout rainfall runoff events, a singl e index know as an event mean concentration (EMC) defined by Sansalone and Buchberger (1997) was used to characterize event concentrations of the constituents. EMCs represent the runoff volume based average of the constituent and is defined by the total p ollutant load (mass) divided by the total event runoff volume as follows. ( 3 1) where M is the total mass of constituent over entire event duration (M); V is the total volume of flow over entire event duration (L 3 ); is the flow weighted average concentration for entire event (M/L 3 ); tr is the event duration; c(t) is the time variable dissolved or particulate bound concentration (M/L 3 ); q(t) is the time variable flow (L 3 /T); and t is the time (T). Certain constituents are expressed as a value rather than concentration, such as PSDs. These were measured as event mean value (EMV) PM Lea ching Due to exposure of the system to the elements, a significant amount of PM mass collected in the effluent of the system is not escaped influent PM, but rather leached PM from the 20 ft effluent pipe that lies in a sewer system. A mock event was set u p by

PAGE 49

49 loading the system with ~900 L (420 L being the median event runoff volume) of deionized water (DI) at the inlet of the filter to ensure all mass collected in the effluent was leached PM. Throughout the mock event, effluent samples were collected to give EMCs of leached PM. The PM EMC of leached sediment was 7.84 mg/L, which was greater than the average measured event EMC of 7.80. Therefore, all sediment PM was assumed to have been leached PM and not associated with influent PM and not included in t he model. For events with a higher measured settleable PM EMC than the leached EMC, particularly events with higher peak flow, the leach EMC was subtracted from the event EMCs before settleable mass was calculated. Events with lower measured settleable E MC were not included in the model and all settleable PM assumed to have been leached. The leached suspended PM was consistent throughout the event with sample standard deviation of 3.44 and an EMC of 9.50 mg/L. The leached EMC concentration was subtracte d from event EMCs and was not included in the model. Measured effluent PSDs were bimodal, indicating the collected mass had two sources. The larger particle diameter peaks were cropped off, which coincided with the leached PM having a particle diameter i n the settleable range and below (< 75 m). An example of a measured system effluent PSD is shown in Figure 3 2 with the escaped influent PM section and leached coagulated PM section. CFD CFD modeling of the bio detention system was undergone with ANSYS F luent 13.0. Reynolds Average Navier Stokes (RANS) mass conservation and momentum equations for incompressible flow serve as the governing equations and are as shown in Equations 3 2 and 3 3

PAGE 50

50 motion with t he assumption that the stresses are a summation of viscosity and pressure. With steady state flows as used in the model, the mean change in fluid mo mentum becomes zero leaving Equation s 3 2 and 3 3 ( 3 2) ( 3 3) i and x j are the i th and j th directi on vector for 3D modeling, u i and u j are the Reynolds averaged velocity in the i th and j th direction, p is the Reynolds averaged pressure, and g i is the sum of body forces in the th direction. The turbulent flow and stresses, or Reynolds stresses, wer e modeled using the semiempirical k 3 4 through 3 6. ( 3 4) ( 3 5) With, ( 3 6) k = 1.2, and C 2 being the turbulent energy dissipation rate, S being the mean strain rate, v T being the eddy viscosity, v being the fluid viscosity, as defined before, and representing the Reynolds stresses.

PAGE 51

51 The filter media was modeled as a simple, homogeneous media porous zone. The porous zone was modeled with a sourc e term as a momentum sink applied to the standard fluid flow equations. The source term applies the pressure drop gradient created by the media that is proportional to the velocity of the fluid in the computational cell. The source term consists of two p arts; a viscous loss term and an inertial loss term. The source term and rela ted equations can be viewed in E quations 3 7 3 9. ( 3 7) 2 ) are defined below. ( 3 8) ( 3 9) Where D p 0 .30 for porosity was found by adding measured volumes of water to a measured volume of dried media. Due to the heterogeneous nature of the media particles, the media particle diameter (D p ) was another variable used to calibrate the model. A PSD of the me dia was found using Fisher Scientific U.S.A. Standard Testing Sieves. A d 50 was measured around 1800 m. The median particle size was found to be around 106 m. Assuming a particle density of 2.65 g/cm 3 coinciding with silica sand, the number of particl es in a size range was found by dividing the mass by the mass per one

PAGE 52

52 spherical particle. This gave a range of 106 to 1800 m particle diameters for calibration. The filter media PSD is shown in Figure 3 3 The Euler Lagrangain approach was used to model the particle phase. This approach includes a Lagrangian discrete particle model (DPM) within an Euler control volume reference frame. It assumes no particle particle interactions and is derived from le settling as shown in E quations 3 10 through 3 13. The particle acceleration is equal to the summation of the particle drag force and buoyancy/gravitational force and is tracked as the particle moves in the flow field. ( 3 10) ( 3 11) ( 3 12) ( 3 13) p is the particle density, d p is the particle diameter, v p_i is the particle velocity in the i th direction, v i is the localized fluid velocity in the i th direction, and is the dynamic viscosity. The particle tracking length, or the length in space the particles were tracked before deemed trapped or escaped, was set a 70 m, nearly three times the length of the system to ensure no false trapped particles. The geometry was designed in AutoCAD 3D 2010 from as built drawings and transferred to ANSYS Workbench for meshing. The geometry included the transverse

PAGE 53

53 flow d istributor, the filter and effluent pipe. The mesh ed geometry is shown in Figure 3 4 and consists of 442,000 computation cells or 9.4 mL/cell. One of the inputs into the model is the particle diameter. The PSD is broken down and represented by select spe cific particle sizes. The discretization number (DN) represents the resolution, or what quantity of particle sizes are used for representation. It has been shown that the model efficiency increase with higher DN at the cost of computation time (Garofalo and Sansalone 2011). Studies with CFD and stormwater treatment unit operations have found a DN of 16 to be sufficient (Garofalo and Sansalone 2011 ; Dickenson and Sansalone 2009). Due to this study utilizing a steady state model with significantly less computation demand, a safe discretization number (DN) of 21 different sized particles were tracked in the model with diameters from 0.5 to 1200 m. This encompassed PM from the lower range of negligible particle mass to high end of particles with consist ent 100 % removal. The measured influent mass based PSDs were used to create particle size mass increments to be inputted. The other input into the model is the hydrology. Steady state flow runs were calculated and interpolated across the event hydrogra phs. A post processed time step (TS) of one minute average flow rates were matched with PM size percent removal to arrive at an EMV percent removal of each PM size which was then applied to influent total mass and PSDs. Post processing included user defi ned programing with batch script files. This transferred flow rates and their partnered PM percent removes to effluent PM mass and PSD. Therefore new events with different hydrology and influent PM could be examined quickly without the use of time intensi ve CFD simulations.

PAGE 54

54 Results and Discussion Bio detention Results The Bio detention filter system, for the 12 measured rainfall runoff events, received peak flow rates up to 101 L/min and mean influent flows of 12.5 L/min. For all the events, the system r anged from 16 to 79 % with a mean of 57 % percent separation (PS) of suspended PM and ranged from 53 to 86 % with a mean 68 % of total PM. Event total PM mass and PS are sh own in Table 3 1 The system lowered influent PM mean particle size from a d 50 of 100 m to 7 m. PSD results are summarized in T able 3 2 CFD Calibration Model calibration was accomplished with varying the media particle diameter, porosity, and particle density and comparing CFD results to measured values. The model error was minimiz ed with a media particle diameter of 300 m for permeability, which falls inside the measured range. The initial porosity of 0.30 proved to be accurate in the model. Previous modeling studies dealing with rainfall runoff have used a particle density of 2 .65 g/cm 3 which coincides with silica sand (Pathapati and Sansalone 2009; Dickenson and Sansalone 2009). The accuracy of the model in the present study was maximized with a particle density of 2.40 g/cm 3 This value is validated due to collected efflu ent mass being, on average, 42 % volatile. This suggests a significant fraction of PM is organic matter which typically has a density of 0.9 to 1.3 g/cm 3 (Pilatti et al. 2006) and therefore less dense than silica sand. CFD Results The escaped PM PSD is a n import aspect of unit operations either for post treatment or its effect on the receiving system. The measured effluent PSD had a mean

PAGE 55

55 d 50 of 7 m and the modeled effluent PSD had a mean d 50 of 5 m. The accuracy of the model with respect to effluent PSD s was measured with normalized root mean squared error ( NRMSE) which is shown below in E quation 3 14. ( 3 14) The root mean squared error (RMSE) is normalized by the rang e of % finer by mass values, 0 to 100 %. The model accuracy for escaped PSD ranged from 3 to 10 % NRMSE with a mean of 7 % for all 12 measured events. The measured modeled PSD comparisons can be seen in Figures 3 5 and 3 6 The PM mass separated by the system was matched for measured and modeled for all 12 rainfall runoff events. These values were compared using an absolute relative percent diffe rence (RPD), which is shown in E quation 3 15 below. ( 3 15) The absolute RPD for the events ranged from 0 to 12 % and a mean of 5 %. The model had no tendency to overestimate or underestimate the mass removed with six events for both. Only one event was grea ter than 10 %, and this event had the lowest peak flow rate and the third lowest total PM EMC. All the measured model comparison in formation can be seen in Table 3 3 Conclusion This study developed a CFD model that correctly predicted influent PM mass and PSD separation through filtration and settling in a bio detention filter system loaded by rainfall runoff from an asphalt paved roadway. The model was validated with 12 north central Florida rainfall runoff events with peak flow rates ranging from 3 to 1 01 L/min and influent PSD d 50 s ranging from 20 to 152 m. RANS and k

PAGE 56

56 equations were used to model the turbulent fluid phase and the media was modeled as a simple homogeneous media by applying a fluid momentum sink with a source term. The p article phase was tracked with a Lagrangian DPM at a DN of 21. Varying steady state flow rate simulations interpolated across a hydrograph with a TS of 1 minute proved sufficient, opposed to the more computationally demanding unsteady simulations. For 11 of 12 rainfall runoff events, total influent mass separated fell below 10% ARPD, with one event at 12% and a mean of 5%. For particle sizes escaped, effluent PSD matched with an average NRMSE of 7 %. The model, however, did not account for leached PM i n the system effluent. Future event forecasting would require a two stage process: first the model should predict the influent PM mass separated, and then the leached PM mass (from the mock event) should be to be added to derive the total system effluent This study shows the effectiveness of CFD modeling of infiltration and settling UOs to meet the growing demands of regulations, whether the need is to correctly size and type a new UO or to forecast existing UO performance in response to ever changing e nvironmental conditions.

PAGE 57

57 Table 3 1. Event based particulate matter (PM) mass percent separation (PS) for all 12 monitored rainfall runoff events. PS is defined as the change in influent to ed by the entire cumulative mass throughout the monitoring campaign Event Based PM Mass Percent Separation Event Date (2012) 18 Jan. 22 Feb. 21 Apr. 16 May. 29 May. 7 Jun. 14 Jun. 24 Jun. 10 Jul. 13 Jul. 1 Aug. 6 Aug. Total Suspended M i (g) 11.4 8.8 12.5 8.7 70.6 62.9 56.1 543.5 60.2 24.6 14.5 9.2 873.7 M e (g) 5.8 2.4 2.6 6.7 44.6 24.7 31.8 213.3 19.2 20.8 4.5 4.2 376.6 PS (%) 48.7 72.6 78.9 22.3 36.8 60.7 43.2 60.8 68.2 15.5 69.0 54.8 56.9 Settleable M i (g) 1.6 1.0 1.5 5.6 42.1 28.3 15.2 120.6 20.4 3.7 3.9 2.2 243.8 M e (g) 0.5 0.3 0.4 2.4 15.5 4.8 1.6 87.6 6.2 1.8 0.7 0.2 121.7 PS (%) 67.8 70.2 74.8 57.4 63.2 83.1 89.7 27.4 69.4 50.2 81.5 89.3 50.1 Sediment M i (g) 2.6 1.5 15.4 19.0 127.7 197.8 95.3 260.6 243.2 104.2 49.4 0.6 1116.7 M e (g) 0.5 0.4 2.0 4.0 10.6 18.5 8.6 123.1 19.6 4.6 2.0 0.3 193.7 PS (%) 82.5 75.4 86.8 79.0 91.7 90.7 91.0 52.8 91.9 95.5 96.0 53.0 82.7 Total M i (mg) 15.5 11.3 29.4 33.2 240.4 289.0 166.6 924.7 323.8 132.4 67.7 12.0 2234.2 M e (mg) 6.8 3.1 5.1 13.1 70.7 48.0 42 .0 424.0 45.0 27.3 7.2 4.7 692.0 PS (%) 56.3 72.8 82.8 60.6 70.6 83.4 74.8 54.1 86.1 79.4 89.4 61.0 69.0

PAGE 58

58 Table 3 2. Influent and escaped influent particle size distributions (PSD) for all 12 monitored rainfall runoff events during the campaign. The d 10 is the diameter in which 10 % of the total mass has diameter less than this size. The d 50 is the diameter in which 50 % of the total mass has diameter less than this size. The d 90 is the diameter in which 90 % of the total mass has diameter less than t his size. Values are listed as volume based event averaged values as event mean values (EMV). Statistical values are derived from volume weighted event mean values EMV Event Date (2012) Influent PSD (m) Effluent PSD (m) d 10 d 50 d 90 d 10 d 50 d 90 18 Jan 3.9 32.1 313.5 0.8 3.9 12.5 22 Feb 37.5 71.8 511.9 0.9 2.7 11.0 21 Apr 11.7 104.6 473.4 1.7 5.3 11.2 16 May 8.1 152.3 912.8 1.5 4.9 11.7 29 May 11.4 151.1 654.6 1.2 3.9 26.6 7 Jun 6.6 47.1 468.4 1.8 4.2 11.8 14 Jun 10.0 95.6 56 5.5 3.5 6.0 16.1 24 Jun 9.6 101.9 652.2 2.0 6.7 33.7 10 Jul 8.9 82.2 586.1 3.4 10.3 21.3 13 Jul 18.5 143.3 452.3 3.8 11.8 24.2 1 Aug 7.8 77.5 287.4 3.0 6.6 14.0 6 Aug 5.5 20.0 102.9 2.6 6.6 14.7 Mean 9.8 99.7 607.5 2.1 7.1 18.7 Median 9.6 101.9 652. 2 2.0 6.7 33.7 Std. Dev. 3.2 27.1 102.0 0.7 1.7 8.6

PAGE 59

59 Table 3 3. This table summarizes the relative percent difference (RPD) and normalized root mean squared error (NRMSE) between measured and modeled results al ong with influent and escaped influent PSD gamma parameters. Measured PSDs are represented with modeled gamma Event Date (2012) Mass Separated RPD (%) Effluent PSD NRMSE (%) Event based gamma distribution parameters for the measured PSDs Influent Escaped Influent 18 Jan. 12 8.6 0.60 148.61 1.32 3.98 22 Feb. 2 2.9 0.59 250.09 1.36 2.98 21 Apr. 2 5.9 0.60 513.69 2.19 2.77 16 May. 6 8.3 0.56 456.47 1.98 2.98 29 May. 5 6.6 1.17 228.88 1.02 7.27 7 Jun. 9 7.7 0.59 206.59 2.75 1.78 14 Jun. 3 7.5 0.68 230.15 2.69 2.93 24 Jun. 0 5.4 0.84 200.95 1.20 9.12 10 Jul. 2 6.1 0.78 353.02 2.25 5.22 13 Jul. 10 9.4 0.82 253.57 2.16 6.36 1 Aug. 5 8.1 0.67 167.46 2.95 2.62 6 Aug. 7 10.2 0.61 186.34 2.63 2.92 Mean 5 7.2 0.71 266.32 2.04 4.24 Range 11 7.3 0.61 365.08 1.93 7.34

PAGE 60

60 Figure 3 1. Eastside bio detention filter system aerial view schematic. Arrows show runoff flow from curb to PVC channel, parshall flume, and into system. Effluent follows out t he underground PVC pipe through the parshall flume and exists into the storm sewer

PAGE 61

61 Figure 3 2. An example of a system effluent PSD, with (A) untreated PM discharged from filter, and (B) PM deposition recovered from 10 0 mm filter effluent PVC pipe. B section was disregarded in modeling

PAGE 62

62 Figure 3 3. Model inputs of media particle size (A) and influent PM particle size distribution (PSD) (B). The black circles represent measured v alues with the grey line representing the gamma distribution model of the media PSD. The black lines represent the influent PM PSDs for all 12 events

PAGE 63

63 Figure 3 4. Model bio detention filter geometry with 442,000 computational cells (9.4 mL/cell)

PAGE 64

64 Figure 3 5. Measured and CFD modeled effluent PSDs for 2012 rainfall runoff events

PAGE 65

65 Figure 3 6. Measured and CFD modeled effluent PSDs for 2012 rainfall runoff events

PAGE 66

66 CHA PT ER 4 G LOBAL SUMMARY AND CONCLUSIONS This study evaluates a bio detention filter system through physical and CFD modeling. The system was loaded by rainfall runoff from an asphalt paved roadway in north central Florida. 12 different events were monitored with rainfall depth greater than 2.5 mm. These events were analyzed for PM, N, P, various water chemistry parameters, and runoff detention and retention. This data was used to compare the performance of this system to other unit operations and validate and c alibrate a CFD model of the system. For the entire monitoring campaign the system separated a mean of 69% of total PM and 57% of suspended PM, and reduced PSD from a mean of 100 to 21 m. The mean separation by the system for TN and TP was 33% and 41%, r espectively. The largest area of concern with the bio detention system is the dissolved phase, questioning the bioaccumulation potential of the system. Six events showed an increase in DN concentration; five events showed an increase in DP concentration, in particular the longer duration events; and all events showed an increase in TDS and conductivity. An analysis of various unit operations shows the bio detention filtration with above average economic performance. Although, Street sweeping has proven to be the most suitable solution to constituent loads in stormwater. Unfortunately it is not coupled with important hydrologic restoration, but it does display that addressing the constituents before they become suspended in rainfall runoff is the more fe asible and most efficient approach.

PAGE 67

67 This CFD model developed correctly predicted influent PM mass and PSD separation through filtration and settling in a bio detention filter system The model was validated with 12 rainfall runoff events with peak flow r ates ranging from 3 to 101 L/min and influent PSD d 50 s ranging from 20 to 152 m. RANS and k equations were used to model the turbulent fluid phase and the media was modeled as a simple homogeneous media by applying a fluid momentum sink with a source term. The particle phase was tracked wit h a Lagrangian DPM at a DN of 21. Varying steady state flow rate simulations interpolated across a hydrograph with a TS of 1 minute proved sufficient, opposed to the more computationally demanding unsteady simulations. For 11 of 12 rainfall runoff event s, total influent mass separated fell below 10% ARPD, with one event at 12% and a mean of 5%. For particle sizes escaped, effluent PSD matched with an average NRMSE of 7 %. The model, however, did not account for leached PM in the system effluent. Futur e event forecasting would require a two stage process: first the model should predict the influent PM mass separated, and then the leached PM mass (from the mock event) should be to be added to derive the total system effluent. This study shows the effec tiveness of CFD modeling of infiltration and settling UOs to meet the growing demands of regulations, whether the need is to correctly size and type a new UO or to forecast existing UO performance in response to ever changing environmental conditions.

PAGE 68

68 AP PENDIX ADDITIONAL SYSTEM MONITORING INFORMATION Methodology Hydrology Rainfall depth was measured using a calibrated Texas Electronics, Inc. Tipping Bucket Model TR 525USW (Texas Electronics, Inc. 2007) with a resolution of 0.25 mm (0.01 in). An Onset HOBO Event data logger (Onset 2013) was used to record data during each rainfall event. The data includes a timestamp for every 0.25 mm of rainfall. The rainfall intensity was found by dividing the 0.25 mm by the duration of time for that depth. Water s tage in the Parhsall flume was measured with a Shuttle MJK ultrasonic sensor (30 kHz) (CAN AM 2011) and recorded with a Campbell Scientific, Inc. CR1000 data logger (Campbell Scientific 2013). The water stage was recorded every second throughout the eve nt, and for post processing, the data was compressed to minute averages. The effluent parshall flume was calibrated by taking various stage measurements with paired timed volumetric measurements. The calibrat ion curve is shown in Figure A 1. All 12 even t hyetographs and influent and effluent hydrographs can be seen in Figures A 2 and A 3. Filter Bypass Only a portion of the catchment area flow was received by the filter. The remaining flow bypassed the filter inlet and proceeded to the downhill gutter. This produces tremendous error in runoff coefficients (C) to the extent that rain gage and parshall flume measurements were questioned. To validate the rainfall depth and event volume a bypass experiment was set up. The catchment was loaded with varyin g flow

PAGE 69

69 rates to determine how much flow was bypassed at each catchment area flow rate. Timed volumetric measurements were taken at the flow source and at the inlet to the filter. A relationship was established to transfer filter hydrographs to catchment area hydrographs to match total volume with rainfall depth and arrive at event Cs. For faster flow rates a power function matched the measured data correctly. But with power functions, the smaller values produce a higher ratio than the maximum feasible f ilter flow/catchment flow (Q F /Q C ) of 1:1. Or, the filter flow rate would translate to a smaller watershed flow rate. This was resolved with a hybrid calibration curve starting with a linear relationship followed by the power relationship. The linear rela tionship modeled the measured with an R 2 value of 0.86, and the power function modeled the measured with a normalized root mean squared error (NRMSE) of 0.5 %, with the RMSE normalized to the range of flows. The Q F /Q C relationship can be seen in Figu re A 4 and Table A 1. EMC/EMV Each concentration and value at the sampling times were measured from both duplicate samples taken. The average between those concentrations and values were used in the EMC/EMV calculation. The samples represented the entire flo w from either the start of runoff or the end of previous sample time to the end of that sample time. The mass each sample accounts for was found by multiplying the concentration by the volume they represented. The total of each samples mass accounts for the event total mass. The event total mass divided by the event total volume equals the EMC. EMVs ume and value is

PAGE 70

70 divided by the total volume to arrive at the EMV. PSD EMVs were found as a Water Chemistry All electrodes were deployed through a Thermo Scientific Orion 5 Star Meter (Thermo Scientific 2013c). pH was measured using an Orion 9157BN Triode 3 in 1 pH/ATC Electrode (Thermo Scientific 2013b). Dissolved Oxygen (D.O.) was measured using an Orion 08301MD Dissolved Oxygen Probe (Thermo Scientific 2013d). The oxidation red uction potential (redox) was measured using a Fisher Scientific Accumet 13 620 81 redox Electrode (Fisher Scientific 2013). Conductivity and Total Dissolved Solids (TDS) were measured with an Orion 013010MD Conductivity Cell (Thermo Scientific 2013e). P articulate Matter Whatman Glass Microfibre Filter GF/F 1822 047 (Whatman 2009) was used for suspended PM collection and measurement. The filters were prepared by passing DI water through the filter and dried; this was continued until constant mass was a chieved. The increase in mass of the filter from the sample accounts for PM mass. All PSDs were modeled to gamma distributions to express the particle size distribution (PSD) shape and size with two parameters. Gamma distributions are statistical contin uous probability distributions that can relate to PSDs. PSDs were modeled by varying the gamma parameters until the sum of the squared error at each parameter is expr

PAGE 71

71 Phosphorus and Nitrogen The P quantity was analyzed through spectrophotometry using Hach Permachem PhosVer 3 Phosphate Reagent (Hach 2012c) and a Hach DR 5000 UV Vis Spectrophotometer (Hach 2012f) at a wavelength of 880 nm. Suitable calibrat ion curves with R 2 values greater than 0.99 were generated using Hach Phosphate Standard Solution (50 mg/L) (Hach 2012b) diluted to incremental phosphate concentrations with deionized (DI) water treated with a Thermo Barnstead Nanopure Diamond Water Syste m (Thermo Scientific 2013a). Calibration curves are shown in Figu re A 5. The nitrogen persulfate digestion and spectrophotometry was performed with Hach Test 2012d) for samples greater than 25 mg /L N and Total Nitrogen 2671745 / 2672145 (Hach 2012e) for samples less than 25 mg/L N and a Hach DR 5000 UV Vis Spectrophotometer (Hach 2012f) at a wavelength of 410 nm. This procedure is in accordance with Standard Methods 4500 N C (American Public He alth Association 1995). Suitable calibration curves with R 2 values greater than 0.98 were generated using Hach Nitrogen, Ammonia Standard Solution (10 and 100 mg/L) (Hach 2012a) diluted to incremental nitrogen concentrations with DI water treated with a Thermo Barnstead Nanopure Diamond Water System (Thermo Scientific 2013a). Calibration curves are shown in Figure A 6 and A 7. Unit Operation Economic Comparison First the annual cost for each unit operation (UO) system was calculated with its total lif e time cost, a payback period of 25 years and interest rate of 3% (McMahon 2013; Moneychimp 2013). The total cost of each UO for its designed life time includes

PAGE 72

72 total labor, total material and total maintenance costs. This is shown in Equation A 1 where A j is the unit annual cost, P j is the total cost of labor and materials, O j is the annual maintenance cost, j represents PM, TP, or TN, i is the interest rate, and N is the payback period. ( A 1) Second, the annual runoff volume and total mass for PM, TP, and TN for each UO was calculated. The annual runoff volume was based on the annual precipitation, surface area and rainfall runoff coefficients in the studied source area. The calculation is shown in Equation A 2 where V t is the annual treated runoff volume, i is the annual precipitation, A s is the surface area of the source area, and C v is the rainfall runoff coefficient. The total mass of PM, TP and TN is based on the annua l runoff volume and the annual mean concentrations. The calculation is shown in Equation A 3 where M j is annual treated mass for constituent j and [C] j is the annual mean concentration for constituent j. ( A 2) ( A 3) Third, the unit cost of PM, TP, and TN by weight for each UO was calculated. The unit cost of constituents are based on the annual cost and mass of each constituent. The calculation is shown in Equation A 4 where G w,j i s the cost per kg of constituent j. ( A 4) Fourth, the unit cost by volume treated for each UO was calculated. The unit cost s are based on the annual cost of each UO and the total annual treated volume.

PAGE 73

73 The calculation is shown in Equation A 5 where G v,j is the cost per treated volume of constituent j ( A 5 ) Filter Media The filter media PSD was found through mechanical sieving with U.S.A Standard Testing Sieves using dri ed original media that had not treated rainfall runoff. The measured data and the modeled gamma distribution are seen in Figure A 3 Event Hydrology Summaries 18 January 2012 The 18 January event was a low intensity rainfall with 3.81 mm (0.15 in) of dep th that produced 287 L (76 gal) of runoff loading the system. The peak rainfall intensity was 77 mm/hr (3.0 in/hr) and the peak runoff was 3.1 L/min (0.8 gal/min). At the cessation of runoff to the system the runoff storage was 89 L (24 gal) which repres ented a moisture content increase of 2.2 % for the entire system volume. The number of previous dry hours (pdh) since the last rainfall event was 1.25 hours. 22 February 2012 The 22 February event had 3.56 mm (0.14 in) of rainfall depth that produced 208 L (54 gal) of runoff loading the system. The peak rainfall intensity was 193 mm/hr (7.62 in/hr) and the peak runoff was 4.2 L/min (1.1 gal/min). At the cessation of runoff to the system the runoff storage was 32 L (8 gal) which represented a moisture co ntent increase of 0.8 % for the entire system volume. The number of pdh since the last rainfall event was the greatest for this event at 149 hours. 21 April 2012

PAGE 74

74 The 21 April event had 3.56 mm (0.14 in) of rainfall depth that produced 256 L (68 gal) of r unoff loading the system. The peak rainfall intensity was 387 mm/hr (15.24 in/hr) and the peak runoff was 5.4 L/min (1.4 gal/min). At the cessation of runoff to the system the runoff storage was 47 L (12 gal) which represented a moisture content increase of 1.1 % for the entire system volume. The number of pdh since the last rainfall event was 26 hours. 16 May 2012 The 16 May event had 4.32 mm (0.17 in) of rainfall depth that produced 345 L (91 gal) of runoff loading the system. The peak rainfall inten sity was 387 mm/hr (15.24 in/hr) and the peak runoff was 8.2 L/min (2.2 gal/min). At the cessation of runoff to the system the runoff storage was 52 L (14 gal) which represented a moisture content increase of 1.3 % for the entire system volume. The numbe r of pdh since the last rainfall event was 26 hours. 29 May 2012 The 29 May event was part of tropical storm Beryl with 18.54 mm (0.73 in) of rainfall depth that produced 2268 L (599 gal) of runoff loading the system. The peak rainfall intensity was 1,54 8 mm/hr (60.96 in/hr) and the peak runoff was 44.7 L/min (11.8 gal/min). At the cessation of runoff to the system the runoff storage was 10 L (3 gal) which represented a moisture content increase of 0.3 % for the entire system volume. The number of pdh s ince the last rainfall event was 12 hours. 7 June 2012 The 7 June event had 14.99 mm (0.59 in) of rainfall depth that produced 2071 L (547 gal) of runoff loading the system. The peak rainfall intensity was 1,935 mm/hr

PAGE 75

75 (76.20 in/hr) and the peak runoff wa s 86.5 L/min (22.9 gal/min). At the cessation of runoff to the system the runoff storage was 87 L (23 gal) which represented a moisture content increase of 2.1 % for the entire system volume. The number of pdh since the last rainfall event was 39 hours. 14 June 2012 The 14 June event had 19.30 mm (0.76 in) of rainfall producing 2273 L (601 gal) of runoff loading the system. The peak intensity was 1,548 mm/hr (60.96 in/hr) and the peak runoff was 88.3 L/min (23.3 gal/min). At the cessation of runoff to the system the runoff storage was 152 L (40 gal) which represented a moisture content increase of 3.7 % for the entire system volume. The number of pdh since the last rainfall event was 94 hours. 24 June 2012 The 24 June event was part of tropical storm Debby with 144.78 mm (5.70 in) of rainfall depth that produced 13,883 L (3,667 gal) of runoff loading the system. This was the largest of all the events with a duration of 1,035 minutes. The peak rainfall intensity was 1,935 mm/hr (76.20 in/hr) and the p eak runoff was 49.8 L/min (13.2 gal/min). At the cessation of runoff to the system the runoff storage was 82 L (22 gal) which represented a moisture content increase of 2.0 % for the entire system volume. The number of pdh since the last rainfall event w as 69 hours. 10 July 2012 The 10 July event had 11.43 mm (0.45 in) of rainfall depth that produced 1,099 L (290 gal) of runoff loading the system. The peak rainfall intensity was 2,323 mm/hr (91.44 in/hr) and generated the greatest flowrate with the peak runoff at 100.9L/min

PAGE 76

76 (26.7 gal/min). At the cessation of runoff to the system the runoff storage was 115 L (30 gal) which represented a moisture content increase of 2.8 % for the entire system volume. The number of pdh since the last rainfall event was 117 hours. 13 July 2012 The 13 July event had 8.38 mm (0.33 in) of rainfall producing 495 L (131 gal) of runoff loading the system. The peak rainfall intensity was 2,323 mm/hr (91.44 in/hr) and the peak runoff was 87.5 L/min (23.1 gal/min). At the cessa tion of runoff to the system the runoff storage was 14 L (4 gal) which represented a moisture content increase of 0.3 % for the entire system volume. The number of pdh since the last rainfall event was 26 hours. 1 August 2012 The 1 August event had 3.30 mm (0.13 in) of rainfall depth that produced 230 L (61 gal) of runoff loading the system. The peak rainfall intensity was 774 mm/hr (30.48 in/hr) and the peak runoff was 19.2 L/min (5.0 gal/min). At the cessation of runoff to the system the runoff storag e was 38 (10 gal) which represented a moisture content increase of 0.9 % for the entire system volume. The number of pdh since the last rainfall event was 42 hours. 6 August 2012 The 6 August event had 3.56 mm (0.14 in) of rainfall depth that produced 26 4 L (70 gal) of runoff loading the system. The peak rainfall intensity was 1,161 mm/hr (45.72 in/hr) and the peak runoff was 18.4 L/min (4.9 gal/min). At the cessation of runoff to the system the runoff storage was 52 L (14 gal) which represented a moist ure

PAGE 77

77 content increase of 1.3 % for the entire system volume. The number of pdh since the last rainfall event was 24 hours.

PAGE 78

78 Table A 1. Filter bypass results. Measured hydrographs were transformed into watershed hydrographs and then the runoff volume was compared with rainfall volume to arrive at runoff coefficients (C) Event Date (2012) Rainfall Depth Received System Volume Catchment Runoff Volume Rainfall Volume C (mm) (L) (L) (L) 18 Jan 3.81 287 1082 1341 0.79 22 Feb 3.56 208 924 1252 0.71 21 Apr 3.56 256 1006 1252 0.78 16 May 4.32 345 1101 1520 0.70 29 May 18.54 2267 5465 6528 0.84 7 Jun 14.99 2071 5045 5276 0.95 14 Jun 19.30 2273 5638 6796 0.83 24 Jun 144.78 13882 41805 50972 0.81 10 Jul 11.43 1099 3892 4024 0.95 13 Jul 8.38 495 1880 2951 0.62 1 Aug 3.30 230 936 1163 0.76 6 Aug 3.56 264 973 1252 0.74 Mean 90.80 8820 26304 31961 0.82 Median 144.78 13882 41805 50972 0.81 Std. Dev. 20.12 6035 21145 22198 0.09

PAGE 79

79 Table A 2. Unit operation (UO) PM and nutrient performance based on UO monitoring results and their economics (with a 25 year design life), as price per mass constituent removed, compared to street sweeping recovery economics. SLR is the surface loading rate of the UO. Street sweeping proves to be the most cost effect ive for all three constituent monitored Monitoring campaign information Parameter and description SHS in BTR VCF in BTR BHS in GNV RCF in GNV JF4 in GNV Biodetention in GNV Florida based street sweeping Influent volume treated (m 3 ) 118.9 171.8 45.1 6 1 16.3 253.4 Based on results of 14 Florida MS4s for street sweeping cost, sampling (> 100 samples), analysis (PM, TP, TN). Street sweepings were recovered before runoff transport Unit operation (UO) volume (m 3 ) 0.45 2.18 1.78 0.08 2.3 4.08 UO surface a rea (m 2 ) 0.64 2.48 1.17 0.16 1.17 4.46 Drainage source area (m 2 ) 1088 1088 500 500 500 365 Flow rate (L/s) Site Median 2.5 1.8 3.1 1.1 0.9 0.2 Range 0.1 7.9 0.1 5.1 0.2 11.3 0.1 3.6 0.1 5.5 0.02 0.2 Median SLR (L/min m 2 ) 126.6 31.5 123.1 187.5 25.6 2.7 PM mass Influent 38.2 kg 29.0 kg 13.6 kg 10.0 kg 45.6 kg 34.9 kg Effluent 17.6 kg 3.0 kg 2.9 kg 0.8 kg 0.8 kg 11.2 kg % removal 53.9 90.8 81.1 92.1 98.2 67.9 TP mass Influent 116.5 g 13.2 g 72.5 g 18.0 g 205.8 g 2 95.7 g Effluent 78.5 g 4.8 g 23.7 g 4.1 g 74.7 g 220.6 g % removal 32.6 60.1 56 81.6 59 25.3 TN mass Influent 150.6 g 10.9 g 51.2 g 70.6 g 163.6 g 433.0 g Effluent 107.5 g 6.9 g 29.0 g 40.5 g 8.6 g 316.7 g % removal 28.6 42.8 46 13.9 55 26.9 Constituent and cost category Economics PM removed $/kg of PM 94.0 96.5 180.9 85.6 56.0 75.6 0.2 TP removed $/kg of TP 50939.4 298806.6 39665.9 56628.7 19145.5 23863.2 567 TN removed $/kg of TN 44911.8 627493.9 87193.5 26150.8 16193.4 15409.5 364

PAGE 80

80 Table A 3. Monitoring results and economic analysis for the screened hydrodynamic separator (SHS) system. Annual runoff volume is based on the median volumetric rainfall runoff coefficient of 0.506 (Kim and Sansalone 2009; Ma et al. 2010). Labor + Mat erial is the total cost for the unit and the installation. Percent separation is represented by PS Parameter System and Source Area Volume treated by SHS (m3) 118.9 30 year mean annual rainfall, I (mm) 1602.2 Source area (paved), A s (m 2 ) 1088.0 Annual runoff volume 1 V t (m 3 ) 882.1 PM concentration, [C] j [mg/L] Median 303.1 Range 179.2 863.4 Event based flow rate, Q (L/s) Median 2.5 Range 0.1 7.9 Constituent: PM TP TN Total mass on an annual basis, M j Infl uent 38.2 kg 116.5 g 150.6 g Effluent 17.6 kg 78.5 g 107.5 g PS (%) 53.9 32.6 28.6 Economic Analysis Labor + Material, P j ($) 20,000 30,000 Interest rate, i (%) 3 Payback period, N (years) 25 Annual maintenance cost, O j ($) 500 Unit annual cost A j ($) 1935.7 Cost per m 3 G v,j ($) 2.2 Constituent: PM TP TN Cost per Kg, Gw,j ($) 94.0 50939.4 44911.8

PAGE 81

81 Table A 4. Monitoring results and economic analysis for the volumetric clarifying filter (VCF) system. Annual runoff volume is based on the median volumetric rainfall runoff coefficient of 0.506 (Liu et al. 2010). Labor + Material is the total cost for the unit and the installation. Percent separation is represented by PS Parameter System and Source Area Volume treated by VCF (m3) 171.8 30 year mean annual rainfall, I (mm) 1602.2 Source area (paved), A s (m 2 ) 1088.0 Annual runoff volume 1 V t (m 3 ) 882.1 PM concentration, [C] j [mg/L] Median 299.5 Range 147.8 1421.8 Event based flow rate, Q (L/s) Medi an 1.8 Range 0.1 5.1 Constituent: PM TP TN Total mass on an annual basis, M j Influent 29.0 kg 13.2 g 10.9 g Effluent 3.0 kg 4.8 g 6.9 g PS (%) 90.8 60.1 42.8 Economic Analysis Labor + Material, P j ($) 30,000 40,000 Interest rate, i (%) 3 Pa yback period, N (years) 25 Annual maintenance cost, O j ($) 500 Unit annual cost, A j ($) 2510.0 Cost per m 3 G v,j ($) 2.8 Constituent: PM TP TN Cost per Kg, Gw,j ($) 96.5 298806.6 627493.9

PAGE 82

82 Table A 5. Monitoring results and economic analysis for th e baffle hydrodynamic separation (BHS) system. Annual runoff volume is based on the median volumetric rainfall runoff coefficient of 0.56 (Sansalone and Cho 2011). Labor + Material is the total cost for the unit and the installation. Percent separation i s represented by PS Parameter System and Source Area Volume treated by BHS (m3) 45.1 30 year mean annual rainfall, I (mm) 1228.3 Source area (paved), A s (m 2 ) 500.0 Annual runoff volume 1 V t (m 3 ) 343.9 PM concentration, [ C] j [mg/L] Median 478.6 Range 92.6 2282.0 Event based flow rate, Q (L/s) Median 3.1 Range 0.2 11.3 Constituent: PM TP TN Total mass on an annual basis, M j Influent 13.6 kg 72.5 g 51.2 g Effluent 2.9 kg 23.7 g 29.0 g PS (%) 81.1 56 46 Economic Analysis Labor + Material, P j ($) 20,000 30,000 Interest rate, i (%) 3 Payback period, N (years) 25 Annual maintenance cost, O j ($) 500 Unit annual cost, A j ($) 1935.7 Cost per m 3 G v,j ($) 5.6 Constituent: PM TP TN Cost per Kg, Gw,j ($) 180.9 39665.9 87193.5

PAGE 83

83 Table A 6. Monitoring results and economic analysis for the radial cartridge filter (RCF) system. Annual runoff volume is based on the median volumetric rainfall runoff coefficient of 0.56 (Sansalone and Berretta 2009). Labor + Material is the total cost for the unit and the installation. Percent separation is represented by PS Parameter System and Source Area Volume treated by RCF (m3) 6.0 30 year mean annual rainfall, I (mm) 1228.3 Source area (paved), A s (m 2 ) 500.0 Annu al runoff volume 1 V t (m 3 ) 343.9 PM concentration, [C] j [mg/L] Median 369.9 Range 25.9 1914.0 Event based flow rate, Q (L/s) Median 1.1 Range 0.1 3.6 Constituent: PM TP TN Total mass on an annual basis, M j In fluent 10.0 kg 18.0 g 70.6 g Effluent 0.8 kg 4.1 g 40.5 g PS (%) 92.1 81.6 13.9 Economic Analysis Labor + Material, P j ($) 4,000 6,000 Interest rate, i (%) 3 Payback period, N (years) 25 Annual maintenance cost, O j ($) 500 Unit annual cost, A j ($) 787.1 Cost per m 3 G v,j ($) 2.3 Constituent: PM TP TN Cost per Kg, Gw,j ($) 85.6 56628.7 26150.8

PAGE 84

84 Table A 7. Monitoring results and economic analysis for the Jellyfish (JF4) system. Annual runoff volume is based on the median volumetric rainfal l runoff coefficient of 0.56 (Sansalone 2011). Labor + Material is the total cost for the unit and the installation. Percent separation is represented by PS Parameter System and Source Area Volume treated by JF4 (m3) 116.3 30 year mean annual rainfall, I (mm) 1228.3 Source area (paved), A s (m 2 ) 500.0 Annual runoff volume 1 V t (m 3 ) 343.9 PM concentration, [C] j [mg/L] Median 444.5 Range 78.2 1401.7 Event based flow rate, Q (L/s) Median 0.9 Range 0.1 5.5 Con stituent: PM TP TN Total mass on an annual basis, M j Influent 45.6 kg 205.8 g 163.6 g Effluent 0.8 kg 74.7 g 8.6 g PS (%) 98.2 59 55 Economic Analysis Labor + Material, P j ($) 30,000 40,000 Interest rate, i (%) 3 Payback period, N (years) 25 An nual maintenance cost, O j ($) 500 Unit annual cost, A j ($) 2510.0 Cost per m 3 G v,j ($) 7.3 Constituent: PM TP TN Cost per Kg, Gw,j ($) 56.0 19145.5 16193.4

PAGE 85

85 Table A 8. Monitoring results and economic analysis for the bio detention system Paramete r System and Source Area Volume treated by bio detention (m3) 253.4 30 year mean annual rainfall, I (mm) 1228.3 Source area (paved), A s (m 2 ) 365.0 Annual runoff volume 1 V t (m 3 ) 363.1 PM concentration, [C] j [mg/L] Median 0 Range 0 Event based flow rate, Q (L/s) Median 0.2 Range 0.02 0.2 Constituent: PM TP TN Total mass on an annual basis, M j Influent 34.9 kg 295.7 g 433.0 g Effluent 11.2 kg 220.6 g 316.7 g PS (%) 67.9 25.3 26.9 Economic Analysis Labor + Material, P j ($) 20,000 25,000 Interest rate, i (%) 3 Payback period, N (years) 25 Annual maintenance cost, O j ($) 500 Unit annual cost, A j ($) 1792.1 Cost per m 3 G v,j ($) 4.9 Constituent: PM TP TN Cost per Kg, Gw,j ($) 75.6 23863.2 15409.5

PAGE 86

86 Table A 9. Influent and effluent particle size distributions (PSD) for all 12 monitored rainfall runoff events during the campaign. The d 10 is the diameter in which 10 % of the total mass has diameter less than this size. The d 50 is the diameter in whi ch 50 % of the total mass has diameter less than this size. The d 90 is the diameter in which 90 % of the total mass has diameter less than this size. Values are listed as volume based event averaged values as event mean values (EMV). Statistical values a re derived from volume weighted event mean values EMV Event Date (2012) Influent PSD (m) Effluent PSD (m) d 10 d 50 d 90 d 10 d 50 d 90 18 Jan 3.9 32.1 313.5 1.1 6.0 111.0 22 Feb 37.5 71.8 511.9 1.0 11.6 56.1 21 Apr 11.7 104.6 473.4 4 .2 52.1 145.9 16 May 8.1 152.3 912.8 2.9 22.3 110.9 29 May 11.4 151.1 654.6 4.0 32.7 118.7 7 Jun 6.6 47.1 468.4 3.5 33.8 99.6 14 Jun 10.0 95.6 565.5 9.9 33.5 86.3 24 Jun 9.6 101.9 652.2 4.3 14.8 88.2 10 Jul 8.9 82.2 586.1 5.3 30.4 118.0 13 Jul 18.5 143.3 452.3 5.1 19.5 43.9 1 Aug 7.8 77.5 287.4 4.3 14.8 88.2 6 Aug 5.5 20.0 102.9 4.4 20.0 102.9 Mean 9.8 99.7 607.5 4.7 21.2 93.5 Median 9.6 101.9 652.2 4.3 14.8 88.2 Std. Dev. 3.2 27.1 102.0 1.8 9.1 14.7

PAGE 87

87 Table A 10. Event based particulate matt er (PM) fractions as event mean concentration (EMC) and value (EMV) for all 12 rainfall runoff events during the campaign. The change in concentration/value from influent to effluent is constituent concentration/value as the runoff passes through the system Event Based PM Fractions Event Date (2012) 18 Jan. 22 Feb. 21 Ap r. 16 May. 29 May. 7 Jun. 14 Jun. 24 Jun. 10 Jul. 13 Jul. 1 Aug. 6 Aug. Mean Median Std. Dev. Suspended EMC i [mg/L] 40.9 42.5 48.6 25.1 31.1 30.4 24.7 39.2 54.8 49.7 62.8 34.8 37.8 39.2 7.3 EMC e [mg/L] 29.5 13.7 12.6 23.0 19.8 12.5 15.0 15.5 19.5 43.3 2 3.4 19.6 16.9 15.5 4.9 [mg/L] 11.4 28.8 36.0 2.1 11.4 17.9 9.7 23.7 35.3 6.5 39.5 15.2 20.9 23.7 7.4 Settleable EMC i [mg/L] 5.6 5.0 6.0 16.1 18.6 13.7 6.7 8.7 18.5 7.4 16.9 8.3 10.5 8.7 3.9 EMC e [mg/L] 2.6 1.8 1.9 4.7 6.9 2.4 0.7 6.3 6.3 3.8 3.8 1.1 5.3 6. 3 2.1 [mg/L] 3.1 3.2 4.2 11.5 11.7 11.2 6.0 2.3 12.2 3.6 13.2 7.2 5.2 2.3 4.0 Sediment EMC i [mg/L] 9.3 7.2 60.2 55.2 56.3 95.5 41.9 18.8 221.2 210.4 214.7 2.4 47.7 18.8 54.7 EMC e [mg/L] 2.3 2.1 9.7 10.2 4.7 10.3 4.0 8.9 19.9 9.6 10 .4 1.4 8.6 8.9 3.4 [mg/L] 7.0 5.1 50.4 45.0 51.6 85.3 37.9 9.9 201.3 200.7 204.3 1.0 39.1 9.9 52.8 Total EMC i [mg/L] 55.8 54.6 114.8 96.5 106.0 139.6 73.3 66.6 294.5 267.5 294.4 45.5 96.0 66.6 60.4 EMC e [mg/L] 34.4 17.5 24.2 37.9 3 1.3 25.2 19.8 30.7 45.7 56.7 37.6 22.2 30.8 30.7 6.8 [mg/L] 21.5 37.1 90.6 58.6 74.7 114.4 53.5 35.9 248.8 210.8 256.9 23.4 65.2 35.9 56.6 Volatile EMV i (%) 59.3 53.6 42.8 49.2 72.9 64.1 47.0 54.1 62.1 54.1 39.4 64.7 57.1 54.1 6.9 EMV e (%) 49.8 63.0 55.2 58.2 26.1 54.7 28.6 46.3 26.1 5.2 31.4 50.9 42.5 46.3 10.9 (%) 9.5 9.4 12.3 9.0 46.8 9.4 18.4 7.8 36.0 48.9 8.0 13.7 14.6 7.8 14.1

PAGE 88

88 Table A 11. Event based particulate matter (PM) mass percent separation (PS) for all 12 monitored rainfall runoff events. PS is defined as the change in influent to cumulative mass throughout the monitoring campaign Event Based PM Mass Percent Separ ation Event Date (2012) 18 Jan. 22 Feb. 21 Apr. 16 May. 29 May. 7 Jun. 14 Jun. 24 Jun. 10 Jul. 13 Jul. 1 Aug. 6 Aug. Total Suspended M i (g) 11.4 8.8 12.5 8.7 70.6 62.9 56.1 543.5 60.2 24.6 14.5 9.2 873.7 M e (g) 5.8 2.4 2.6 6.7 44.6 24.7 31.8 213.3 19.2 20.8 4.5 4.2 376.6 PS (%) 48.7 72.6 78.9 22.3 36.8 60.7 43.2 60.8 68.2 15.5 69.0 54.8 56.9 Settleable M i (g) 1.6 1.0 1.5 5.6 42.1 28.3 15.2 120.6 20.4 3.7 3.9 2.2 243.8 M e (g) 0.5 0.3 0.4 2.4 15.5 4.8 1.6 87.6 6.2 1.8 0.7 0.2 121.7 PS (%) 67.8 70.2 74.8 57.4 63.2 83.1 89.7 27.4 69.4 50.2 81.5 89.3 50.1 Sediment M i (g) 2.6 1.5 15.4 19.0 127.7 197.8 95.3 260.6 243.2 104.2 49.4 0.6 1116.7 M e (g) 0.5 0.4 2.0 4.0 10.6 18.5 8.6 123.1 19.6 4.6 2.0 0.3 193.7 PS (%) 82.5 75.4 86.8 79.0 91.7 90.7 91.0 52 .8 91.9 95.5 96.0 53.0 82.7 Total M i (mg) 15.5 11.3 29.4 33.2 240.4 289.0 166.6 924.7 323.8 132.4 67.7 12.0 2234.2 M e (mg) 6.8 3.1 5.1 13.1 70.7 48.0 42.0 424.0 45.0 27.3 7.2 4.7 692.0 PS (%) 56.3 72.8 82.8 60.6 70.6 83.4 74.8 54.1 86.1 79.4 89.4 61.0 69.0 Volatile M i (mg) 9.2 6.1 12.6 16.3 175.1 185.4 78.3 499.9 201.0 71.7 26.7 7.8 1290.1 M e (mg) 3.8 1.9 2.8 7.6 18.4 26.2 7.0 196.3 11.7 2.1 9.9 2.4 290.2 PS (%) 59.2 68.0 77.8 53.4 89.5 85.8 91.0 60.7 94.2 97.1 62.9 69.3 77.5

PAGE 89

89 Table A 12. Eve nt based phosphorous (P) fractions as event mean concentration (EMC) for all 12 rainfall runoff events during the campaign. Negative values indicated a decrease in constituent c oncentration/value as [C] indicating a net increase in P concentration from the system. Statistical values are derived from volume weighted event mean values EMC Event Based Pho sphorus Fraction Event Date (2012) 18 Jan. 22 Feb. 21 Apr. 16 May. 29 May. 7 Jun. 14 Jun. 24 Jun. 10 Jul. 13 Jul. 1 Aug. 6 Aug. Mean Median Std. Dev. Dissolved EMC i [g/L] 606.2 828.9 795.2 450.6 184.3 155.3 290.1 267.1 1206.2 431.2 412.8 238.2 321.0 267 .1 221.4 EMC e [g/L] 834.9 668.8 744.9 221.8 199.0 246.8 296.5 327.4 573.2 325.9 350.9 178.6 331.1 327.4 107.4 [g/L] 228.7 160.1 50.2 228.7 14.7 91.5 6.4 60.3 633.1 105.3 61.9 59.6 10.1 60.3 153.4 Suspended EMC i [g/L] 541.3 591.7 521.8 407.7 210.8 322.5 456.5 277.9 243.3 848.7 571.1 266.5 320.0 277.9 2456.6 EMC e [g/L] 449.4 404.9 318.0 255.5 206. 6 205.4 448.8 212.3 235.1 252.3 167.7 252.6 245.9 212.3 75.4 [g/L] 91.9 186.7 203.8 152.2 4.2 117.1 7.7 65.6 8.2 596.4 403.3 13.8 74.0 65.6 90.1 Settleable EMC i [g/L] 17.1 106.7 120.1 65.8 103.7 79.8 28.4 19.0 75.9 36.3 62.6 28.6 3 9.6 19.0 31.4 EMC e [g/L] 7.6 9.8 38.0 18.1 40.2 15.1 2.7 10.4 22.7 13.9 14.2 3.7 14.0 10.4 9.9 [g/L] 9.4 96.9 82.1 47.7 63.5 64.7 25.6 8.6 53.2 22.4 48.4 24.9 25.5 8.6 23.8 Sediment EMC i [g/L] 14.9 39.2 195.8 110.8 199.3 325.8 1 00.1 42.8 476.1 601.5 650.6 4.7 129.2 42.8 147.6 EMC e [g/L] 3.6 12.5 36.5 15.2 21.1 102.6 5.0 5.7 41.1 193.3 31.7 2.9 22.2 5.7 37.5 [g/L] 11.3 26.8 159.3 95.6 178.3 223.2 95.1 37.1 435.0 408.2 619.0 1.8 107.0 37.1 120.0 Total EMC i [g/L] 1179.4 1566.6 1632.9 1034.9 698.2 883.5 875.0 606.8 2001.6 1917.6 1697.1 537.9 814.4 606.8 351.9 EMC e [g/L] 1295.5 1096.1 1137.5 510.6 466.9 569.9 753.0 555.7 872.0 785.4 564.5 437.8 607.5 555.7 166.9 [g/L] 116.1 470.5 495.4 524.3 231.3 313.6 122.0 51.0 1129.6 1132.3 1132.6 100.1 206.9 51.0 253.6

PAGE 90

90 Table A 13. Event based phosphorus (P) mass percent separation (PS) for all 12 monitored rainfall runoff events. PS is defined as the change in influent to effluent P mass d ivided by the influent mass. Negative PS indicates a net cumulative mass throughout the monitoring campaign. One event had a net increase in dissolved P mass from the system Event Bas ed Phosphorus Mass Percent Separation Event Date (2012) 18 Jan. 22 Feb. 21 Apr. 16 May. 29 May. 7 Jun. 14 Jun. 24 Jun. 10 Jul. 13 Jul. 1 Aug. 6 Aug. Total Dissolved M i (mg) 174.0 172.2 203.8 155.3 417.9 321.6 659.4 3707.5 1326.2 213.5 94.9 62.8 7446.4 M e (mg) 165.1 117.8 141.5 61.8 214.3 261.1 218.7 4013.3 559.0 155.1 62.6 36.5 5970.3 PS (%) 5.1 31.6 30.6 60.2 48.7 18.8 66.8 8.2 57.8 27.4 34.0 41.8 19.8 Suspended M i (mg) 155.4 122.9 133.7 140.5 478.1 667.8 1037.8 3858.4 267.5 420.2 131.4 70.2 7413.7 M e (mg) 88.9 71.3 60.4 71.2 222.5 217.3 704.3 2602.9 229.3 120.0 29.9 51.7 4418.0 PS (%) 42.8 42.0 54.8 49.3 53.5 67.5 32.1 32.5 14.3 71.4 77.2 26.4 40.4 Settleable M i (mg) 4.9 22.2 30.8 22.7 235.1 165.3 64.4 263.4 83.4 17.9 14.4 7.5 924.7 M e (mg) 1.5 1.7 7.2 5.0 43.3 108.6 4.2 127.0 40.1 6.6 2.5 0.8 347.9 PS (%) 69.2 92.2 76.6 77.8 81.6 34.3 93.4 51.8 52.0 63.3 82.4 90.0 62.4 Sediment M i (mg) 4.3 8.2 50.2 38.2 452.0 674.7 227.4 594.4 523.5 297.8 149.6 1.2 3020.3 M e (mg) 0.7 2.2 6.9 4.2 22.7 10 8.6 7.8 70.2 40.1 92.0 5.6 0.6 361.1 PS (%) 83.4 73.1 86.2 88.9 95.0 83.9 96.6 88.2 92.3 69.1 96.2 52.0 88.0 Total M i (mg) 338.6 325.5 418.5 356.7 1583.2 1829.5 1989.1 8423.8 2200.6 949.4 390.3 141.8 18805.1 M e (mg) 256.2 193.0 216.0 142.2 502.9 695.6 935.1 6813.3 868.5 373.6 100.8 89.6 11097.2 PS (%) 24.3 40.7 48.4 60.1 68.2 62.0 53.0 19.1 60.5 60.6 74.2 36.8 41.0

PAGE 91

91 Table A 14. Event based phosphorous (N) fractions as event mean concentration (EMC) for all 12 rainfall runoff events during the cam paign. Negative values indicated a decrease in constituent concentration/value as the runoff passes through the system. Six events had a positive dissolved and one event had a concentration from the system. Statistical values are derived from volume weighted event mean values EMC Event Based Nitrogen Fraction Event Date (2012) 18 Jan. 22 Feb. 21 Apr. 16 May. 29 May. 7 Jun. 14 Jun. 24 Jun. 10 Jul. 13 Jul. 1 Aug. 6 Aug. Mean Median Std. Dev. Dissolved EMC i [g/L] 253.7 892.9 1078.6 847.1 224.9 291.4 265.9 380.9 1621.0 563.1 940.4 625.4 439.3 380.9 298.0 EMC e [g/L] 340.7 1108.5 1242.2 500.8 191.2 363.4 487.6 348.7 1511.3 72 6.9 689.8 467.5 438.4 348.7 280.4 [g/L] 87.0 215.5 163.6 346.3 33.6 72.0 221.7 32.2 109.7 163.7 250.6 157.9 0.8 32.2 102.4 Suspended EMC i [g/L] 206.3 286.5 808.0 311.1 221.6 517.9 554.7 442.5 1969.9 217.8 2179.9 381.6 525.2 442.5 378.1 EMC e [g/L] 119.8 329.6 722.0 238.2 199.2 240.2 374.9 348.9 1398.9 180.6 1318.3 211.6 373.6 348.9 240.5 [g/L] 86.5 43.1 86.0 72.9 22.4 277.7 179.8 93.6 571.0 37.2 861.6 170.0 151.7 93.6 217.9 Settleable EMC i [g/L] 28.9 21.7 47.8 123.7 100.8 89.1 40.5 20.3 92.2 35.5 1 30.6 61.6 42.7 20.3 32.1 EMC e [g/L] 12.9 0.4 10.0 79.8 11.8 19.4 2.7 31.5 26.4 23.1 153.2 9.2 26.6 31.5 17.6 [g/L] 16.0 21.3 37.8 43.9 89.0 69.7 37.8 11.1 65.7 12.4 22.7 52.4 16.1 11.1 39.5 Sediment EMC i [g/L] 65.1 63.1 97.6 321.1 164.5 590.7 228.6 47.7 1118.5 951.0 2621.2 21.2 224.7 47.7 369.5 EMC e [g/L] 15.7 12.4 33.9 25.6 20.0 24.1 11.4 13.9 107.0 111.6 153.2 12.6 23.5 13.9 27.0 [g/L] 49.5 50.7 63.7 295.5 144.5 566.6 217.2 33.8 1011.5 839.4 2468.0 8.6 201 .1 33.8 345.4 Total EMC i [g/L] 554.0 1264.2 2032.0 1603.0 711.8 1489.1 1089.7 891.5 4801.6 1767.4 5872.1 1089.8 1192.4 891.5 847.9 EMC e [g/L] 489.1 1450.8 2008.1 844.4 422.3 647.2 876.5 743.0 3043.8 1042.2 2314.5 700.9 872.1 743.0 539.2 [g/L ] 64.9 186.6 23.9 758.6 289.5 841.9 213.2 148.5 1757.9 725.2 3557.5 388.9 320.0 148.5 393.4

PAGE 92

92 Table A 15. Event based nitrogen (N) mass percent separation (PS) for all 12 monitored rainfall runoff events. PS is defined as the change in inf luent to effluent N mass divided by the influent mass. Negative PS indicates a net cumulative mass throughout the monitoring campaign. Four events had a net increase in dissolved N mas s from the system Event Based Nitrogen Mass Percent Separation Event Date (2012) 18 Jan. 22 Feb. 21 Apr. 16 May. 29 May. 7 Jun. 14 Jun. 24 Jun. 10 Jul. 13 Jul. 1 Aug. 6 Aug. Total Dissolved M i (mg) 72.8 185.5 276.5 291.9 509.9 603.5 604.4 5288.0 1782.2 278.8 216.3 156.6 10109.7 M e (mg) 67.4 195.2 260.5 146.6 431.7 720.9 1034.3 4811.6 1487.8 349.7 132.2 95.6 9638.0 PS (%) 7.5 5.2 5.8 49.8 15.3 19.5 71.1 9.0 16.5 25.4 38.9 39.0 4.7 Suspended M i (mg) 59.2 59.5 207.1 107.2 502.4 1072.4 1260.9 6143.5 2165.8 107.8 501.4 95.6 12187.3 M e (mg) 23.7 58.0 151.4 69.7 449.6 476.5 795.3 4815.2 1377.1 86.9 252.7 43.3 8556.3 PS (%) 60.0 2.5 26.9 35.0 10.5 55.6 36.9 21.6 36.4 19.4 49.6 54.7 29.8 Settleable M i (mg) 8.3 4.5 12.3 42.6 228.6 184.5 92.0 282.2 101 .4 17.6 30.0 15.4 1004.0 M e (mg) 2.6 0.1 2.1 23.4 26.7 38.5 5.7 434.0 26.0 11.1 5.7 1.9 575.8 PS (%) 69.2 98.6 82.8 45.2 88.3 79.1 93.8 53.8 74.3 36.7 81.0 87.8 42.6 Sediment M i (mg) 18.7 13.1 25.0 110.7 373.0 1223.2 519.7 662.3 1229.7 470.8 602.9 5. 3 5249.1 M e (mg) 3.1 2.2 7.1 7.5 45.2 47.9 24.1 192.5 105.4 53.7 258.4 2.6 747.0 PS (%) 83.4 83.4 71.6 93.2 87.9 96.1 95.4 70.9 91.4 88.6 57.1 51.5 85.8 Total M i (mg) 159.1 262.7 520.8 552.5 1614.0 3083.5 2477.0 12376.0 5279.0 875.0 1350.6 272.9 28550 .1 M e (mg) 96.7 255.5 421.2 247.2 953.2 1283.8 1859.5 10253.3 2996.3 501.4 649.0 143.4 19517.1 PS (%) 39.2 2.7 19.1 55.2 40.9 58.4 24.9 17.2 43.2 42.7 52.0 47.5 31.6

PAGE 93

93 Figure A 1. The theoretical and measured ca libration curves for the effluent 25.4 mm (1 in) Parshall flume

PAGE 94

94 Figure A 2. Influent and effluent filter hydrographs and hyetographs for January 18 through June 7, 2013 events

PAGE 95

95 Figure A 3. Influent and effluent filter hydrographs and hyetogr aphs for June 14 through August 6, 2013 events

PAGE 96

96 Figure A 4. Filter runoff bypass. The correlation between flows received by the filter (Q F ) to the catchment area flow (Q C ) are represented in two parts as a linear function up to 11 L/min Q C and then followed by a power function

PAGE 97

97 Figure A 5. Phosphorus reagent calibration curve. Hach Permachem PhosVer 3 Phosphate Reagent was used in accordance to Standard Methods 4500 P B acid hydrolysis

PAGE 98

98 Figure A 6. Nitrogen 2714045 / 2672145 was used in accordance to Standard Methods4500 N C for samples greater than 25 mg/L N

PAGE 99

99 Figure A 7. 2671745 / 2672145 was used in accordance to Standard Methods4500 N C for samples less than 25 mg/L N

PAGE 100

100 Figure A 8. Probability density functions (pdf) of influent and effluent dissolved oxygen (DO) throughout the events for the entire monitoring campaign. The pdf model curves were generated using a 2 parameter lognormal distribution model The mean is , and th

PAGE 101

101 Figure A 9. Probability density functions (pdf) of influent and effluent oxidation reduction potential (Redox) throughout the events for the entire monitoring campaign. The pdf model cu rves were generated using a 2 parameter lo gnormal distribution model

PAGE 102

102 LIST OF REFERENCES adsorptive filter subject to inra a nd inter J. Environment Management, 103, 83 94. J. Hydrology 403, 25 36. Correll, D. L. (1998). The Role of Phosphorus in the Eutrophication of Receiving Waters: A Review. J. Environ. Qual. 27(2), 261 266. Impact Development Hydrologic < http://water.epa.gov/polwaste/green/upload/lid _hydr.pdf > (Jul. 11, 2013). particulate matter PM for simulating PM separation by hydrodynamic unit Environ. Sci. Technol. 43(21), 8220 8226. Eaton, A. D., Clesceri, L. S., and Greenberg, A. E. (1995). Standard Methods for the Examination of Water and Wastewater 19th ed. American Public Health Association, Washington, DC. Elvidge, C. D., Tuttle, B. T., Sutton, P. C., Baugh, K. E., Howard, A. T., Milesi, C., Sensors 7, 1962 1979. < http://www.fishersci.com/ecomm/ servlet/fsproductdetail?storeId=10652&product Id=1643037&catalogId=29104&matchedCatNo=13620115&fromSearch=1&searc hKey=electrodes||electrode||orp&highlightProductsItemsFlag=Y&endecaSearchQ uery=%23store%3DScientific%23nav%3D0%23rpp%3D25%23offSet%3D0%23k eyWord %3Dorp%2Belectrode&xrefPartType=From&savings=0.0&xrefEvent=135 9396712108_4&searchType=PROD > (Jan. 28, 2013). hydrodynamic unit operations as a function of computationa l parameters and Chem. Eng. J . 175, 150 159. Stormwater orth American Surface Water Quality Conference, Denver, CO. Guo, L., Zhang, J. Z., and Gu Global Biochemical Cycles 18, 1 12.

PAGE 103

103 Ammonia http://www.hach.com/nitrogen ammonia standard solution 10 mg l as nh3 n 500 m l/product?id=7640208932&callback=qs > (Jan. 28, 2013). < http://www.hach.com/phosphate standard solution 50 mg l as po4 nist 500 ml/product?id=7640199704&callback=qs > (Jan. 28, 2013). < http://www.hach.com/phosver 3 phosphate reagent powder pillows 10 ml pk 100/product?id=764019604 3 > (Jan. 28, 2013). http://www.hach.com/total nitrogen reagent set hr tnt/product?id=7640209860&callback=qs > (Jan. 28, 2013). trogen Reagent Set, http://www.hach.c om/total nitrogen reagent set lr tnt/product?id=7640209858&callback=qs > (Jan. 28, 2013). 00 UV Vis Spectrophotometer). < http://www.hach.com/dr 5000 uv vis spectrophotometer/product?id=7640447361&callback=qs > (Jan. 28, 2013). Imhoff, Remote Sensing of Environment 114, 504 513. Kim, J. of Particulate Matter J. Environ. Eng. 134(11), 912 922. Kofinas, P., and Kioussis, D. R. (2003). Reactive phosphorous removal from aquaculture and poultry productions systems using polymeric hydrogels. Environ. Sc i. Technol. 37(2), 423 427. Event J. Environ. Eng. 136(12), 1321 1330. ration Detention Urban Retrofits with Journal of Hydrologic Engineering, 15, 486 498.

PAGE 104

104 Matter Phosphorus Fractions in Rain fall J. Environ. Eng. 136(11), 1197 1205. Malvern. (2013). Mastersizer 2000. < http://www.malvern.com/labeng/products/mastersizer /ms2000/mastersizer2000.htm > (Jan. 28, 2013). rcent for Jan 2000 InflationData.com < http://inflationdata.com/inflation/Inflation_Rate/CurrentInflation.asp > (Jul. 30, 2013). < http://www.moneychimp.com/articles/econ/inflation_calculator.htm > (Jul. 28, 2013). Onset HOBO Data L 002 http://www.onsetcomp. com/products/data loggers/h07 002 04 > (Jan. 28, 2013). ecosystem level experiment on nutrient l imitation in temperate coastal marine Mar. Ecol. Prog. Ser., 116, 171 179. Pathapati, S. and Pressure Drop in a Storm J. Environ. Eng. 135(2), 77 8 5. between Soil Particle Density and Organic Matter to Mollisols of Santa Fe World Congress of Soil Science Philadelphia, PA. < http://crops.confe x.com/crops/wc2006/techprogram/P18352.HTM > (Jul. 19, 2013). 79. a Runoff Using In 277. with Urban Particulate Matter (PM) and Biogenic/Litter Recovery thr ough Current Association, Gainesville, FL, 1 69.

PAGE 105

105 J. Environ. Eng. 123 (2), 134 143. Partitioning and granulometric distribution of metal leachate from urban traffic dry deposition particulate matter subject to acidic rainfall and runoff retention Water Research, 42(15), 4146 4162. Rainfall 108. Sansalone, J. J., Liu, B., and Kim, J. J. Environ. Eng. 135, 609 620. Sansalone, J. J. and Kim, J. Transport of Particulate Matter Fractions in J. Environmental Quality 37(5), 1883 1893. Shammaa Y., Zhu, D. A., Gy r k, L. L., and Can. J. Civ. Eng. 29, 316 324. 525USW Rainfall http://www.texaselectronics.com/de tail_tr525usw.htm > (Jan. 28, 2013). Thermo Scientific. ( http://www.thermoscientific.com/ ecomm/servlet/productsdetail_11152___11953117_ 1 > (Jan. 28, 2013). in 1 pH/ATC P < http://www.thermoscientific .com/ecomm/servlet/productsdetail?productId=12784406&groupType=PRODUC T&searchType=0&storeId=11152&from=search > (Jan. 28, 2013). Star pH/RDO/Conductivity Portable Multiparameter Meter. < http://www.thermoscientific.com/ecomm/servlet/productsdetail_11152___11960 610_ 1 > (Jan. 28, 2013). Thermo Scientific. < http://www.thermoscientific.com/ecomm/servlet/productsdetail?productId=1196 3330&groupType=PRODUCT&searchType=0&storeId=11152&from=search > (Jan. 28, 2013).

PAGE 106

106 Electrod < http://www. thermoscientific.com/ecomm/servlet/productsdetail?productId=11961323&groupT ype=PRODUCT&searchType=0&storeId=11152&from=search > (Jan. 28, 2013). ource Area Runoff Using In Sciences. < http://water.epa.gov/polwaste/green/ > (Jul. 19, 2013). U.S. EPA. (201 < http://w ater.epa.gov/polwaste/nps/urban.cfm > (Jul. 25, 2013). Viessman, W., and L. Lewis. (2002). Introduction to Hydrology 5 th ed., Prentice Hall, New Jersey. < http://www.whatman.com/GlassMicrofiberBinderFree. a spx#1822 047 > (Jan. 28, 2013).

PAGE 107

107 BIOGRAPHICAL SKETCH Greg Brenner was born and bre d in Huntsville, Alabama, and his passion is whole life has been snow and water skiing ice hockey and swimming; and he wish es to dedicate his career to th e protection of our natural waters