|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 TOXICITY AND CLARIFICATION OF ANTHROPOGENIC CONSTITUENTS TRANSPORTED BY URBAN RAINFALL-RUNOFF By BO LIU A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2007
2 2007 Bo Liu
3 DEDICATION To my late grandmother Junrong Yang, my late father Zhiyong Liu, my uncle, Zhijun Liu, my aunt, Ying Yu and my sister Miao Liu who are the benchmark to which I strive and who always stand behind me, supporting me and believing there is noth ing that I cannot achieve.
4 ACKNOWLEDGMENTS First and foremost, I express my cordially gratitude and deep appreciation to my advisor, Dr. John J. Sansalone, who gave me the best chance to work in this great academic area. He has consistently guided, encouraged and supported me throughout the journey to my Ph.D. program. His patient elucidation, enlightening ideas and pr ecious comments have co ntributed a lot to my understanding of this research ar ea and shaping my concept of scie ntist and engineer. It was and will be great fortune and enormous inspiration for me in my life. I also extend my sincere appreciation to th e distinguish professors on my committee: Dr. Mark T. Brown, Dr. Jean-Claude J. Bonzongo and Dr. Michael D. Dukes. I am also very grateful to my committee members for thei r helpful advice on the dissertation work. I express my thanks to my colleagues: Mr. Subbu-Srikanth Pathapati, Dr. Jong-Yeop Kim, Dr. Jia Ma, Dr. Gaoxiang Ying, Dr Tianpeng Guo and Dr. Xuheng Kuang, who shared me with their knowledge and helpful disc ussion. My appreciation also extends to my colleagues including, Ms. Natalie Magill, Mr. Robert W. Rooney, Mr. Sa urabh N. Raje, Mr. Sundeep Gulati, Mr. Ruben A. Keztesz a nd Ms. Tingting Wu for their valu able assistance and help. Their friendships have been one of my important accomplishments in the past four years.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS...............................................................................................................4 LIST OF TABLES................................................................................................................. ..........9 LIST OF FIGURES................................................................................................................ .......11 LIST OF ABBREVIATIONS........................................................................................................17 ABSTRACT....................................................................................................................... ............25 CHAPTER 1 GLOBAL INTRODUCTION.................................................................................................27 2 ACUTE TOXICITY OF PARTICULA TE MATTER IN URBAN RAINFALLRUNOFF ON FATHEAD MINNOWS AND CHANNEL CATFISH..................................36 Introduction................................................................................................................... ..........36 Objectives..................................................................................................................... ..........38 Background..................................................................................................................... ........39 Physical and Chemical Characteri stics of PM in Urban Runoff.....................................39 Toxicity Mechanism of PM.............................................................................................40 Experimental Factors in Toxicity Test............................................................................41 Methodology.................................................................................................................... .......42 Sample Collection...........................................................................................................42 Organism Holding...........................................................................................................43 Water Quality Analysis...................................................................................................44 Acute Lethal Toxicity Test..............................................................................................45 Particle Size Distribution (PSD) and Gill Function........................................................45 Oxygen Consumption Test..............................................................................................46 Results........................................................................................................................ .............47 Event Hydrology and Sampling......................................................................................47 Water Quality and Exposure Conditions.........................................................................48 Acute Lethal Toxicity Test for Channel Catfish.............................................................50 Acute Sub-lethal Test for Fathead Minnows...................................................................51 Fish Gill Function Test....................................................................................................52 Oxygen Consumption Test..............................................................................................52 Discussion..................................................................................................................... ..........53 Conclusions.................................................................................................................... .........55 3 THE BIOACCUMULATION AN D JOINT ACTIONS OF METALS IN URBAN RAINFALL-RUNOFF ON FATHEAD MINNOWS AND GREEN ALGAE.....................72 Introduction................................................................................................................... ..........72
6 Objectives..................................................................................................................... ..........76 Methodology.................................................................................................................... .......76 Sample Collection and Preparation.................................................................................76 Experimental Animal (minnows) Holding......................................................................77 Bioaccumulation Test......................................................................................................77 One-compartment Bioconcentration and Depuration Model..........................................78 Green Algae Stock Culture..............................................................................................79 EC50 Determination.........................................................................................................80 Metal Joint Action Evaluation.........................................................................................81 Results and Discussion......................................................................................................... ..82 Metals from Urban Rainfall-runoff.................................................................................82 Bioaccumulation Mechanism and Metal Speciation.......................................................83 Metal Bioaccumulation...................................................................................................84 Metal Joint Toxicity........................................................................................................87 Conclusions.................................................................................................................... .........89 4 STORMWATER ADSORPTIVE-FILTRAT ION TESTING OF A RADIAL FLOW CARTRIDGE SYSTEM.......................................................................................................110 Introduction................................................................................................................... ........110 Objectives..................................................................................................................... ........111 Methodology.................................................................................................................... .....112 Testing Configuration and Description.........................................................................112 Radial Flow Filtration System and Engineered Media..................................................114 Data Acquisition and Management...............................................................................115 Calibration Procedures for Flow and Head Data...........................................................115 PM Filtration Experiment..............................................................................................117 Phosphorus Adsorption and Adsorp tion-filtration Experiments...................................119 Evaluation Metrics.........................................................................................................120 Results and Discussion.........................................................................................................121 SSC Removal Efficiencies............................................................................................121 Particle Size Distribution...............................................................................................123 Head Loss as a Function of Flow Rate..........................................................................124 Phosphorus Concentrati on Change Using (AOCM)P....................................................125 Head Loss for Phosphorus Run Using (AOCM)P.........................................................127 Phosphorus Removal Using (AOCM)P.........................................................................127 Phosphorus and PM Rem oval Using Perlite.................................................................128 Conclusions.................................................................................................................... .......129 5 RESPONSE OF A SOURCE AREA WATERSHED AND VOLUMETRIC CLARIFYING FILTER TO RAINFA LL-RUNOFF AND ANTHROPOGENIC LOADINGS....................................................................................................................... ...146 Introduction................................................................................................................... ........146 Objectives..................................................................................................................... ........147 Methodology.................................................................................................................... .....148 Site and VCF System Configuration.............................................................................148
7 Data Acquisition and Management...............................................................................149 Gamma Distribution of H ydrograph and Hyetograph...................................................151 Residence Time and Contact time.................................................................................151 Runoff Coefficients.......................................................................................................152 Hydrological VCF System Model as a Function of Time.............................................152 Detention Water Chemistry...........................................................................................153 Results and Discussion.........................................................................................................154 Event Hydrology Summary...........................................................................................154 Time of Concentration and Runoff Coefficient.............................................................155 Dry period, Hydrograph and Hyetograph......................................................................156 Hydrological Response of VCF System........................................................................158 Surface Loading Rate, Residence Time and Contact Time...........................................159 Detention Test...............................................................................................................161 Speciation of Nitrogen and Phosphorus........................................................................162 Conclusions.................................................................................................................... .......162 6 TESTING OF VOLUMETRIC CLARIFYI NG FILTER SYSTEM FOR URBAN RAINFALL-RUNOFF TREATMENT ON PARTICULATE MATTER REMOVAL.......179 Introduction................................................................................................................... ........179 Objectives..................................................................................................................... ........182 Methodology.................................................................................................................... .....182 Volumetric Clarifying Filter Syst em Structure and Configuration...............................182 Rainfall and Hydrological Data Collection...................................................................183 Experiment Sampling Design........................................................................................184 PM fractions, SSC and Turbidity..................................................................................185 Particle Size Distribution...............................................................................................186 Particle Size Indices......................................................................................................187 Evaluation Metrics.........................................................................................................188 Quality Assurance and Quality Control (QA/QC)........................................................188 Results and Discussion.........................................................................................................189 Storm Events Description and Mass Fractions..............................................................189 Experimental Sampling.................................................................................................190 Mass Concentrations of PM Fractions..........................................................................192 Removal Efficiencies and Mass Fractions....................................................................193 SSC and Particle Size Distribution................................................................................194 Turbidity and SSC.........................................................................................................196 Recovered Solids Gradation and Particle Characteristics.............................................197 Conclusion..................................................................................................................... .......198 7 VOLUMETRIC CLARIFYING FILTRATION FOR PHOSPHORUS IN URBAN RAINFALL-RUNOFF.........................................................................................................212 Introduction................................................................................................................... ........212 Objectives..................................................................................................................... ........215 Previous Work.................................................................................................................. ....216 Bench Scale Study.........................................................................................................216
8 Radial Flow Cartridge Test...........................................................................................216 Methodology.................................................................................................................... .....216 Experimental Site and System Configuration...............................................................216 Experiment Setup and Sampling...................................................................................218 Particulate Fractions......................................................................................................218 Phase Fractionation.......................................................................................................219 Analysis of Phosphorus.................................................................................................220 Event Mean Concentrations and Removal Efficiencies................................................221 Categorical Analysis......................................................................................................222 Results and Discussion.........................................................................................................222 Events Summary............................................................................................................222 TP and TDP Concentration...........................................................................................223 TP and TDP Removal....................................................................................................225 TP and TDP Cumulative Mass......................................................................................226 Removal of Phosphorus Associated with PM...............................................................227 Fractions of Phosphorus Associated with PM...............................................................228 Phosphorus Specific Capacity (PSC)............................................................................229 Conclusion..................................................................................................................... .......230 8 GLOBAL CONCLUSIONS.................................................................................................243 LIST OF REFERENCES.............................................................................................................250 BIOGRAPHICAL SKETCH.......................................................................................................263
9 LIST OF TABLES Table page 2-1 Exposure condition and water quality analysis results during 96 hours for four storm events......57 3-1 Dissolved and particulate fractions of meta l species in urban rainfall-runoff and their corresponding toxic rules gi ven by LDEQ and OEPA ....................................................91 3-2 Summary of water quality (Mean, Min, Max) for metal bioaccumulation test on fathead minnows ..............................................................................................................92 3-3 Summary of kinetics parameters for the one-compartment bioaccumulation model ku and kd represent the rate of uptake a nd depuration respectively R2 indicate the goodness-of -fit for the model ..........................................................................................93 3-4 Summary of the depuration model kinetics parameters after 9 days bioaccumulation test R2 indicate the goodness-of -fit for the model ........................................................94 4-1 Summary of particle size di stribution characteristics, gr anulometric parameters and physical properties for Sil-co-Sil 106 util ized in the single radial flow cartridge filtration test .............................................................................................................. .....131 4-2 Experimental matrix for single radial fl ow cartridge filtration, adsorption and adsorption-filtration runs by using (AOCM)P under various operati ng flow capacity ..132 4-3 Summary of SSC results for single radial flow cartridge filtration run under various operating flow rates and 75 Bed Volume with (AOCM)P The target influent concentration is 200 mg/L Sil-co-Sil 106 as SSC ..........................................................133 4-4 Summary of particles mass results for single radial flow cartridge filtration run at various operating flow rates and 75 Bed Volume with (AOCM)P The target influent concentration is 200 mg/L Sil-co-Sil 106 as SSC ..........................................................134 4-5 Summary of Goodness-of-Fit (GOF) of gamma distribution for (AOCM)P filtration runs...135 4-6 Summary of test parameters including phosphorus removal for phosphorus runs with (AOCM)P at various target flow rates The initial total phosphorus target concentration is 1 00 mg/L as PO4 3-P .........................................................................136 4-7 Non uniform adsorption of phosphorus across si ngle radial flow cartridge vertically for phosphorus runs with and without 200 mg /L Sil-co-Sil 106 lo aded under various operating flow rates with (AOCM)P ...............................................................................137 4-8 Summary of the total phosphorus and SSC removal by using perlite under different influent loading and flow conditions .............................................................................138
10 5-1 Hydrological indices for 19 storm events treated by the VCF system for a 1088 m2 watershed in Baton Rouge, LA ......................................................................................165 5-2 Hydrological indices for 19 storm events treated by the VCF system on a 1088 m2 watershed in Baton Rouge, LA ......................................................................................166 5-3 Gamma model parameters of hydrograph a nd hyetograph, RT and CT for 19 storm events treated by VCF system ........................................................................................167 6-1 Summary of mass of each particulate matte r (PM) fraction (suspended, settleable and sediment) and SSC for 19 storm events .........................................................................200 6-2 Summary of statistical charac teristics of particle size distribution for the particulate matter (PM) recovered from sedimentation ba y and top, middle and bottom layers of cartridges ................................................................................................................... .....201 7-1 Summary of cartridge and media information utilized in Volumetric Clarifying Filter system...232 7-2 Summary of hydrological indices for 19 storm events treated by Volumetric Clarifying Filter system on a small watershed in Baton Rouge, LA .............................233 7-3 Phosphorus specific capacity (PSC) fr equency and odds ratio for suspended, settleable and sediment fractions ...................................................................................234
11 LIST OF FIGURES Figure page 2-1 Exposure condition and water quality analysis results during 96 hours for four storm events..58 2-2 Total suspended solid (TSS), turbidity and rainfall-runoff flow rate change with elapsed time for two shorter rainfall-runoff events. The arrows show three sampling points: the first flush, middle event flush and end event...................................................59 2-3 Turbidity values and Cu, Pb, Cd and Mn change with during 96-hour non-renewal static exposure for storm event April 06, 2004. Triangle dot was turbidity values. Solid circle and open circle were partic ulate and dissolved metals respectively..............60 2-4 Turbidity values and Cu, Pb, Cd and Mn change during 96-hour non-renewal static exposure for storm event October 24, 2004. Tr iangle dot was turbidity values. Solid circle and open circle were particul ate and dissolved metals respectively........................61 2-5 Turbidity Response curve for Cumulative mort ality of channel catf ish postlarva as a function of exposure time (left side) and th e trend based on probit values (right side) BTR first flush and middle event flush..............................................................................62 2-6 Response curve for Cumulative mortality of channel catfish postlarva as a function of exposure time (left side) and the trend base d on probit values (right side) for storm event April 06, 2004..........................................................................................................63 2-7 Response curve for Cumulative survival rate of juvenile fathead minnow as a function of exposure time (triangular dot). Solid curves represent probit values for storm event 24 October 2004. NOET represen ts the no observed effect time..................64 2-8 Response curve for Cumulative survival rate of juvenile fathead minnow as a function of exposure time (triangular dot). Solid curves represent probit values for storm event 24 April 2004. NOET repres ents the no observed effect time.......................65 2-9 Response curve for Cumulative survival rate of juvenile fathead minnow as a function of exposure time (triangular dot). Solid curves represent probit values for storm event 22 April 2005. NOET represents the No Observed Effect Time...................66 2-10 Total volume concentration and cumulative number distribution of particles across the size gradation from 1 to 250 m for storm water runoff before 60 minutes quiescent settling............................................................................................................. ...67 2-11 Total volume concentration and cumulati ve number of partic les across the size gradation from 1-250 m for storm water runoff after 60 minutes quiescent settling.......68
12 2-12 The log linear relationship between the total length and total weight of juvenile fathead minnow (upper plot). The log linear relationship between oxygen consumption rate based on unit weight of fish exposed and total weight of test fish........69 2-13 The dissolved oxygen concentration change with time for juvenile fathead minnows exposed in solution with solid con centration at 300mg/L for 24 hours.............................70 2-14 The amount of oxygen consumption based on unit weight of fish change with time after 6 hours exposure to settled and uns ettled storm water runoff samples for October 24, 2004 event......................................................................................................71 3-1 Isobole of metal A and B projected EC50 valu es for toxicity joint action assessment. The solid line represents the theoreti cal values for additive effect....................................95 3-2 Diagrammatic view of the relationship betw een dose and concentration at the target site under different conditions of expo sure frequency and elimination rate......................96 3-3 Distribution of inorganic metal species as a function of pH for DI matrix and stormwater matrix..............................................................................................................97 3-4 Uptake, accumulation and concentration cha nge of Cu for juvenile fat head minnows as function of exposure time in a 2-L aerated batch test chamber. The dash lines are 95% CI of modeled curves.................................................................................................98 3-5 Uptake, accumulation and concentration cha nge of Zn for juvenile fat head minnows as function of exposure time in a 2-L aerated batch test chamber. The dash lines are 95% CI of modeled curves.................................................................................................99 3-6 Uptake, accumulation and concentration cha nge of Pb for juvenile fat head minnows as function of exposure time in a 2-L aerated batch test chamber. The dash lines are 95% CI of modeled curves...............................................................................................100 3-7 Uptake, accumulation and concentration cha nge of Cd for juvenile fat head minnows as function of exposure time in a 2-L aerated batch test chamber. The dash lines are 95% CI of modeled curves...............................................................................................101 3-8 Metal bioaccumulation for juvenile fat h ead minnows as function of exposure time based on the one-compartment model.............................................................................102 3-9 Influence of metal Mass Ratio (MR) on expos ure time in a 9-day for Cu, Zn, Pb and Cd on fathead minnows. Meu indicates the uptake of metal in mass, Mew indicates the amount of metal mass remaining in the aqueous solution.........................................103 3-10 The EC50 (median effective concentration) values of Cu on the growth rate of green algae. The hatched areas indicate the 95% CI of EMC values for dissolved and particulate fractions of Cu for June 30, 2005 event.........................................................104
13 3-11 The EC50 (median effective concentration) values of Cd on the growth rate of green alage. The hatched areas indicate the 95% CI of EMC values for dissolved and particulate fractions of Cd for June 30, 2005 event.........................................................105 3-12 The EC50 (median effective concentration) values of Zn on the growth rate of green algae. The hatched areas indicate the 95% CI of EMC values for dissolved and particulate fractions of Zn for June 30, 2005 event.........................................................106 3-13. The projected Cu and Cd fractions of EC50 values determined from the EC50 value of Cu and Cd mixture based on the EC50 ratio (90:150 = 1:1.7) was plot on Cu axis and Cd axis and indicate the synergistic action......................................................................107 3-14 The projected Cu and Zn fractions of EC50 values determined from the EC50 value of Cu and Zn mixture based on the EC50 ratio (90 : 250 = 1 : 2.8) was plot on Cu axis and Zn axis and indicate the synergistic action...............................................................108 3-15 The projected Cd and Zn fractions of EC50 values determined from the EC50 value of Cd and Zn mixture based on the EC50 ratio (150 : 250 = 1 : 1.7) was plot on Cd axis and Zn axis and indicate the synergistic action...............................................................109 4-1 Section view of the single radial flow car tridge and design info rmation of cartridge and media. The volume of the cartridge indicates the volume occupied by media (AOCM)P filled................................................................................................................139 4-2 Particle gradations of infl uent with the targ et concentration of 200 mg/L Sil-co-Sil 106 and effluent under 7 operating flow rates ranged from 0.19 to1.51 L/s for filtration runs over 75 Bed Volume (B.V.) with (AOCM)P.............................................140 4-3 Change of the head loss with bed volume (1 B.V. = 54.1 L) for filtration runs with (AOCM)P at 0.38, 0.57, 0.76, 1.14, 1.32 and 1.51 L/s as operating flow rate. The linear lines showed the be st fit of the trend.....................................................................141 4-4 Filter head loss change over flow rate, macro porosity and media diameter based on the calculation of Ergun equation....................................................................................142 4-5 Total phosphorus (TP) concentration and pH change for influent and effluent over bed volumes (1 B.V. = 54.1 L) for phosphorus adsorption runs with (AOCM)P at selected flow rates (0.19, 0.38, 0.76 and 1.14 L/s)..........................................................143 4-6 Total phosphorus, total dissolved phosphorus concentrations and pH change of influent and effluent over Bed Volume with (AOCM)P..................................................144 4-7 Change of the head loss over Bed Volume (1 B.V. = 54.1L) for phosphorus adsorption runs with (AOCM)P at 0.19, 0.38, 0.76 and 1.14 L/s flow. Quadratic polynomial curve fit the relationship be tween head loss and flow rate...........................145
14 5-1 Plan view of experimental site and Volu metric Clarifying Filt er (VCF) system, CB represents catch basin and 304.8 in the e quation represents unit conversion (mm to ft) factor..................................................................................................................... ......168 5-2 Location of the pressure transducers instal led in the Volumetric Clarifying Filter (VCF) system. d50, k and m represent the median size, hydraulic conductivity and macro porosity of the media (AOCM)P used in this study..............................................169 5-3 Methodology for determining average pave ment residence time (APRT), initial pavement residence time (IPRT), even t residence time and contact time.......................170 5-4 Incremental volumetric runoff coefficients plots showing tangential approach to maximum runoff coefficient value as func tion of elapsed time for both high volume and low volume storm events..........................................................................................171 5-5 Historical rainfall and dry period data for Baton Rouge site and representative events among 19 storm events monitored based on flow and volume statistics.........................172 5-6 Hyetographs, hydrographs and head loss as a function of normalized elapsed time for four events which represent low, medium and high flow events and event with fully drained condition trea ted by VCF system........................................................................173 5-7 The relationship between flow rate and h ead loss for a volumetric clarifying filter system for 19 storm events..............................................................................................174 5-8 Surface loading rate, residence time and contact time change as a function of normalized elapsed time of the four storm events which represent low, medium and high flow events and event with fully drained condition trea ted by VCF system...........175 5-9 Model comparison between inflow routing and outflow routing for incoming flow rate to volumetric filter cartridges....................................................................................176 5-10 Redox potential, pH, nitrogen and phosphorus species change as a function of detention residence time..................................................................................................177 5-11 Species distribution of nitrogen and phosphor us as a function of pH for rainfall, runoff at time 0 hr (time for end of the 09 August event runoff), 48 hr and 96 hr (48 and 96 hours retention af ter 09 August event).................................................................178 6-1 Plan view of site and configuration of Vo lumetric Clarifying Filter (VCF) system.......202 6-2 Suspended, settleable and sediment fractions of influent and e ffluent based on mass for 19 storm events...........................................................................................................203 6-3 Event based comparison between composite auto sample and composite manual sample. Thick line with slope = 1 repr esents therotical values.....................................................204
15 6-4 The coupled relationship between flow rate influent and effluent suspended solid mass concentration and mass removal (as an index of VCF system performance) by clarification in a small wate rshed for 19 storm events....................................................205 6-5 The coupled relationship between flow rate influent and effluent settleable solid mass concentration and mass removal (as an index of VCF system performance) by clarification in a small wate rshed for 19 storm events....................................................206 6-6 The coupled relationship between flow rate influent and effluent sediment solid mass concentration and mass removal (as an index of VCF system performance) by clarification in a small wate rshed for 19 storm events....................................................207 6-7 The coupled relationship between flow rate PSD and influent and effluent SSC by clarification in a small wate rshed for 19 storm events....................................................208 6-8 Log-normal distributions of influent and effluent SSC concentrations for total 19 storm events................................................................................................................... ..209 6-9 Relationship among turbidity (NTU), Suspended solid concentration (TSS) and settleable and sediment solid s concentration (SSC-TSS)................................................210 6-10 Particle size distribution (m ass gradation) for solids recovered from sedimentation bay and volumetric clarifying filters. Three layers (bottom, middle and top) were gently washed and solids were recovered separately. The solid lines are cumulative gamma distribution fitting curves....................................................................................211 7-1 Plan view of experimental site and Volu metric Clarifying Filt er (VCF) system............235 7-2 The coupled relationship between flow rate, influent and effluent TP concentration and mass removal for VCF in a small watershed on event basis for 2006 events...........236 7-3 The coupled relationship between flow rate, influent and effluent TDP concentration and mass removal for VCF in a small watershed on event basis for 2006 events...........237 7-4 Log-normal distributions of influent and effluent TP and TDP concentrations for total 19 storm events........................................................................................................238 7-5 Mass delivery of TP and TDP in in fluent and effluent runoff volume on accumulative basis for eight urban rainfall-runoff events (from April 21 to July 04, 2006) treated by VCF system..........................................................................................239 7-6 Mass delivery of TP and TDP in in fluent and effluent runoff volume on accumulative basis for eight urban rainfall-runoff events (from July 05 to August 09, 2006) treated by VCF system..........................................................................................240 7-7 The relationship between PM removal a nd particulate phosphorus removal for suspended, settleable and sediment fractions on event basis...........................................241
16 7-8 Particulate phosphorus fractions associated with suspended, settleable and sediment PM for influent and effluent ba sed on mass for 19 storm events....................................242
17 LIST OF ABBREVIATIONS A Area of hydrograph, hyetograph Ae Effective area of system vault Aw Area of the watershed, m2 AD Additive effect ADP Antecedent dry period ADT Average daily traffic AN Antagonistic effect ANCOVA Analysis of covariance ANOVA Analysis of covariance (AOCM)P Aluminum oxide coated media as pumice APHA American Public Health Association APRT Average pavement residence time, min ASTM American Society for Testing and Materials a/v Grain surface area per unit of grain volume B Notch width, m BCF Bioconcentration factor BLM Biological ligand model BMP Best management practice BTC Batch test chamber B.V. Bed volume C Runoff coefficient Ce Effective coefficient of discharge Cd Coefficient discharge
18 Css Residual concentration in body at steady state CCC Criteria Continuous Concentration CI Confidence interval CMC Criteria Maximum Concentration CMLR Cumulative mass loading rate, mg/s CSO Combined sewer overflow CT Contact time, min CTR California Toxic Rule d50 Median size in diameter, ( m) d50m Median size in diameter on mass basis Deff Duration of effluent, min gd Geometric mean diameter Dinf Duration of influent, min Drain Duration of rainfall, min DI Deionized water DIN Dissolved inorganic nitrogen, [mg/L] DM Drainage management DO Dissolved oxygen DOtx Dissolved oxygen concen tration at time x, [mg/L] DOt0 Dissolved oxygen concen tration at time 0, [mg/L] DP Dry period E Effluent flow as a function of elapsed time EC50 Median effective concentration, [ g/L]
19 EMC Event mean concentration, [mg/L] EMF Event mean flow, (L/s) E(x) Expected value of variable x fd Dissolved fraction g acceleration of gravity, 9.81 m/s2 GOF Goodness-of-fit H, h Head, mm Hmax Maximum head loss, mm H50 Mean head loss, mm HDPE High-density polyethylene HF High flow HS Hydrodynamic separator i Hyetograph I Influent flow as a function of elapsed time Imax Maximum rainfall intensity, mm/hr Imin Minimum rainfall intensity, mm/hr Iavg Average rainfall intensity, mm/hr In Rainfall intensity of the n th interval, cm/s ICP-MS Inductive coupled plasma mass spectrometer IPRT Initial pavement residence time j jth hydrograph, hyetograph k kth interval of hydrograph or hyetograph k0 zero-order coefficient
20 k1 first-order coefficient k2 Dimensionless constant ku Uptake rate constant kd First order rate c onstant of depuration K Dimensional factor GK Kurtosis LF Low flow LT50 Median lethal time Mo Oxygen consumption rate M0 Constituent mass on the surface at the beginning of event, g MBE Mass balance error MC Metal concentration, [ g/L] Me Metal MF Medium flow MR Mass ratio MS4 Municipal Separate Storm Sewer System MT Cumulative constituent mass delivered, g n Sample size N Density of algae cells, #/ml NH3-N Ammonia nitrogen NOAA The National Oceanic an d Atmospheric Administration NOET No observed effective time NPDES National pollutant disc harge elimination system
21 NTU Nephelometric turbidity units O Oxygen consumed per gram organism OR Odds ratio p probability value P Rainfall depth, mm Ps Rainfall intensity, mm/hr PAH Ploycyclic aromatic hydrocarbon PCC Portland cement concrete PDD Previous dry days PDH Previous dry hours, hours PM Particulate matter PND Particle number distribution PP Particulate phosphorus PR Percent removal PSC Phosphorus specific capacity, mg/g PSD Particle size distribution PVC Permanent virtual circuit PSD Particle size distribution Q Flow rate, (L/s) Qeff-max Maximum effluent flow rate, L/s Qeff-avg Average effluent flow rate, L/s Qinf-max Maximum influent flow rate, L/s Qinf-avg Average influent flow rate, L/s
22 Qn Volumetric flow rate of the n th interval, cm3/s Qs Storm runoff intensity, mm QA/QC Quality assurance/quality control R2 Coefficient of determination RPD Relative percentage difference, % RPM Round per minute RT Residence time, min RT Residence time (detention time in the VCF system) s Standard deviation S1 Storage in the system vault, L S2 Storage in the cartridge, L SAS Statistical analysis software S.D. Standard deviation SedP Phosphorus bound to sediment particles SetP Phosphorus bound to settleable particles kS Skewness SLR Surface loading rate, L/min-m2 SMC Site mean concentration, [ g/L] SSA Specific surface area SSC Suspended sediment concentration, [mg/L] SSE Sum of squared error SusP Phosphorus bound to suspended particles S(x) Standard deviation of variable x
23 SY Synergistic effect t Depuration time, or elapsed time, min t1 Incubation times at the begi nning of exponential growth phase t2 Incubation times at the end of exponential growth phase tc Time of concentration, min tD Duration of time TDP Total dissolved phosphorus, [mg/L] tn Time at the nth interval TN Total nitrogen, [mg/L] TP Total phosphorus, [mg/L] tr Rainfall duration, min ts Runoff duration, min TSS Total suspended solid, [mg/L] TVC Total volume concentration, L/L u Power index UOPs Unit operation and processes V Superficial velocity, (L/s) Vin, VINF Volume of influent, L VEFF Volume of effluent, L Vs Volume of samples, L VS Volume of storage, L Vo+s Volume of outflow and storage, L VBE Volume balance error, %
24 VCF Volumetric clarifying filter vds Vehicle during storm W Weight of organism WQV Water quality volume, L Z Head, ft Algae growth rate or Fluid viscosity, (N-s/m2) Diameter, (mm) 50 Median contact time, (min) Fluid density, (g/cm3) m Bed porosity or void fraction Distance between centroids V-Notch angle Scale factor of gamma model Shape factor of gamma model Diameter, mm 50 Median value, corresponding to 50% of the cumulative frequency g Standard deviation of geometric mean I Inclusive graphic standard deviation Mean removal efficiency
25 Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy TOXICITY AND CLARIFICATION OF ANTHROPOGENIC CONSTITUENTS TRANSPORTED BY URBAN RAINFALL-RUNOFF By Bo Liu December 2007 Chair: John J. Sansalone Major: Environmental Engineering Sciences Urban rainfall-runoff has been recognized as a specific source of contamination potentially impacting on receiv ing aquatic ecosystems. The chemical composition has been reported for anthropogenic constituents transporte d in a complex heterogeneous mixture that includes particulate matter in a wi de gradation (from colloidal to gravel size), metals, nutrients, etc. (Buckler and Granato 1999; Marsalek 1999). In addition to that, the hydrologic, physical, and chemical phenomena associated with rainfall -runoff are deterministically complex, spatially diffuse, highly unsteady, and the arrival of su ch events and associated loadings have deterministic and stochastic components. This study examined four rainfall-runoff events captured from an in strumented municipal separate storm sewer system (MS4) watershed in Baton Rouge, LA. Untreated runoff has higher accumulative mortalities of test organisms than runoff treated through only physical filtration, indicating the role of particulates on toxicity. The median lethal time (LT50) and no observed effective (NOET) values indica te that higher particulate loads associated with mass-limited events were most toxic, and fo r mass-limited events that toxicity decreased across the event. Oxygen consumption results indicated as total part iculate concentrations increase, the suspended particles were more toxic than settleable and sedi ment particles. Results also indicated that Cu,
26 Pb and Cd exhibited a significant bioaccumulation in metals after nine days exposure and the observed toxic effects for metal binary mixtur es among Cu, Zn, Cd were all synergistic. To address the pollutants in relation to the environmental and ecological issues such as acute and chronic toxicities, many in situ treatment strategies ( unit operations and processes, UOPs) have been developed. A volumetric clarifying filter (VCF) system was evaluated in this study with respect to chemical processes such as phosphorus adsorption and precipitation onto high surface area of aluminum oxide coated pumice (AOCM)P and unit operation such as filtration for particulate-bound pollutants removal. This system was operated to provide some level of volumetric control depending on volumet ric sizing and outflow design but in practice does not provide hydrologic restoration. Based on field data collected from nineteen storm events monitored, results indicate that the r echargeable VCF system was capable of an 89.4% reduction of the incoming suspended sediment concentration (SSC). Total phosphorus (TP) concentrations for effluent were reduced to less than 0.1 mg/L. The dissolved phosphorus (TDP) fraction removal varied but was in the range of 51% to 91% depending on the influent concentration, flow rate, and VCF system mana gement. In addition, the VCF management for the system after each storm influenced the pollu tant partitioning and redox change. Generally, this volumetric clarifying filter system proved to be a feasible water chemistry and particulate treatment system for SSC, TP and TDP but not a hydrologic restoration system.
27 CHAPTER 1 GLOBAL INTRODUCTION According to the U.S. Environmental Prot ection Agency (2000), st ormwater runoff from highways and other roadways is recognized as a specific source of contamination potentially impacting receiving water ecosystems. The chem ical composition reported for water quality constituents in a complex heterogeneous mixture includes a wide gradati on of particulate matter (PM), metals, nutrients, petroleum-related orga nic compounds (including polycyclic aromatic hydrocarbons, or PAHs) and herbicides applied to vegetation edging highway corridors (Buckler and Granato 1999; Marsalek 1999). Data illustrate that loads and con centrations of these anthropogenic constituents are significantly above ambient background levels and, for urban land uses, can exceed surface water discharge criteria a nd toxicity thresholds on either an event or long-term basis (Horner et al. 1994; Makepeace et al. 1995; Morrison et al. 1993; Waller et al. 1995). The hydrologic, physical, and chemical phenom ena associated with urban rainfall-runoff are deterministically complex, spatially diffu se and highly unsteady. Although the relationship between water quality and quantity behavior can be deterministic to a significant extent, quality and quantity parameters can also vary by orders of magnitude acro ss a single event (Sansalone et al. 1998). Therefore, control and tr eatment of the anthropogenic cons tituents transported in urban rainfall-runoff become significantly more challenging when periodic or stochastic loadings from anthropogenic activities are combined with the stochastic arrival of rainfall-runoff events (Sansalone et al. 1998). Rainfall-runoff can mobilize and transport a wide gradation of particulate matter ranging in size from smaller than 1 m (colloidal) to gr eater than 10,000 m (gravel) in source area watersheds, which receive significant contri butions from anthr opogenic activities and infrastructure, particularly transportation and maintenance regimes (Sansalone et al. 1998).
28 Abraded pavement and vehicular tires were repo rted to respectively c onstitute 40-50% and 2030% of the total particulate matter generated in urban rainfall-runoff (K obriger and Geinopolos 1984). PM is a potential concern not only because of the environmental and ecological issues related to solids, but also because many contaminan ts bind to the surfaces of these particles and are then transported into and through aquatic environments (S ansalone 2002). From a water quality perspective, PM with reactive sites and large surface-to-volume ratios can mediate partitioning and transport of chemical species while also serving as reservoirs for many reactive constituents (Sansalone et al 1998). Even though the dissolved loads of heavy metals are significant, most of the pollutant burden appear s to be transported by solids (Lee et al. 1997; Marsalek 1999). Therefore, it is also importa nt to understand that pa rticulate delivery and granulometry which play an important role in partitioning and separa tion of PM (Stumm and Morgan 1996; Sansalone et al. 1998). While the chemical species partition and distribute across the entire PM size gradation, the suspended PM is potentially more m obile and acutely bioavailable than the comparatively more coarse sedi ment fractions that settle more easily (Cristina and Sansalone 2003). Particulate matte r entrained as suspended or be d load material in rainfall runoff may settle in receiving waters and interact with natural sediment in stream deposition and re-suspension cycles and it can also impact biota and deplete dissolv ed oxygen while reducing conveyance capacities and increasing dredging fr equencies (Breault and Granato 2000; James 1999). In addition, the bioavailability and toxicity of a contaminan t is altered when it binds to sediment particles. Thus, when these solid cont aminants remain in the water column, they are toxic to fish and other aquatic life. They can still cause dama ge even when they settle out (Gjessing et al. 1984; Morris on et al 1990; Makepeace 1995).
29 Previous research suggests that metal elemen ts can be leached after contact with acidic rainfall, and their partitioning and speciation are also highly dynamic and driven by hydrology and coupled water chemistry (Morrison et al. 199 0; Sansalone and Buchberger 1997). The levels of metal elements in urban stormwater, principa lly zinc (Zn), copper (Cu), cadmium (Cd), lead (Pb), arsenic (As), chromium (Cr), and nickel (Ni) are reported to be significantly above background levels; and for many urban and transpor tation land uses, they often exceed surface water discharge criteria on an event basis for both dissolved and particulate-bound phases (Foster and Charlesworth 1996; Sansalone and Buchberg er 1997; Sansalone et al. 1998; Sansalone and Tribouillard 1999). In urban environments, meta l species are generate d primarily from the leaching of construction materials and the abrasion of vehicular parts; and they are considered to contaminate not only surface water, but also the ground water near highways and urban areas (Makepeace 1995; Barrett 1993). Unlike organic pollutants, metal species fr om these anthropogenic sources are not degraded in the environment, so they constitute an important class of acute and chronic toxic contaminants which potentially affect receivi ng waters, aquatic life, and ultimately the food chain (Mishra et al. 2004). It s hould be noted that some metal species are essential to living organisms in trace amounts (for example Cu a nd Zn) and they usually have a narrow, optimal concentration range for growth and reproduction; but both excess and shortage of these metal species can be detrimental to organisms (Pelgrom et al. 1994), and when present in unusually high concentrations they can become toxi c to aquatic organisms (Wepener et al 2001). Cadmium can act as an algal nu trient under conditions of Zn limitation (Lee and Morel 1995). Other metals such as Pb have no known biologi cal function (Seymore 1994). Certain of these non-essential trace metals are often found in urban rainfall runoff and become the major
30 contaminants of aquatic environments (Munger et al. 1999) which have toxic effects on aquatic organisms (Witeska et al. 1995). Phosphorus has been long recognized as the limiting nutrient for eutrophication, and as the pollutant of primary concern for the ecologi cal heath of fresh waters (Correll 1998). Eutrophication commonly plagues la custrine systems due to elevated concentrations of introduced phosphorus which accumulates in thei r stagnant surface waters (Wendt and Corey 1980; Sharply 1999; Welch 1992). A significant lo ad of phosphorus can be transported in stormwater runoff, including many types of nonpoint rainfall-runoff di scharges from urban areas (EPA 1993). Several origins of phosphorus in stormwater are known today, such as discharges from automobile exhaust and other combustion processes, bu rning organic material releases phosphorus into the atmosphere and mineral liberated from pavement surface during weathering (Strecker 1994). In many cases, phos phorus is introduced into the aquatic environment in a number of different chemical fo rms, and has been generally described as being present in the aqueous and solid phases and may change according to a dynamic equilibrium (Compton et al. 2000). Besides a simple partitio ning concept between diss olved and particulate phases, phosphorus also distributes across the pa rticulate size gradat ion in rainfall-runoff (Sansalone et al. 1998; Sansalone and Kim 2007). To address those environmental and ecologi cal issues in relation to the pollutants transported in urban rainfall-runoff, a variety of in situ treatment strategies (unit operations and processes, UOPs) have been developed over the last several decades to control the quality and quantity of stormwater in urban areas, such as settling basins and filtra tion (Schueler 1987; Li et al. 1999). The success or failure of rainfall-r unoff UOPs depends largely upon our understanding and application of hydrologic processes and the pr evalence of control methods that are integral
31 components of best management practices (BMPs) (Sansalone 2005). Furthe rmore, water quality standards have often tended to be event based and deterministic, which fails to recognize the stochastic nature of interact ions between hydrology and the physic al and biochemical processes which might affect the water quality in both stor mwater treatment facilities and urban receiving waters (Wong et al. 2002). Thus, the knowledge of treatment viability as a function of the hydrologic, physical, and chemical ch aracteristics of rainfall-runoff loadings is critical to the success of a new generation of control strategi es, BMPs, and sustainable urban development. The hydrologic concepts of interest with respec t to the design of BM Ps are closely related to the design objectives of the BMPs which can be peak discharge control, volume control, water quality management, pollutant removal, groundwate r recharge or a combin ation of two or more of these objectives (Clar et al. 2004). Historically, many of the BM P structural techniques were designed solely to avoid flooding, such as dete ntion and retention ponds and they therefore rarely incorporate features that promote water quality and polluta nts treatment. Since the passage of the National Pollutant Discharge Elimin ation System Storm Water Phase I permitting regulations in the 1980s, and particularly the NPDES Storm Water Phase II permitting regulations in 2003, there has been a large number of suggested structural BMPs developed to manage the quantity and more importantly to improve the quality of rainfall-runoff in the most cost-effective manner. Traditionally, settling basins have been the mo st prevalent unit operation since such basins can provide hydrologic control and PM separatio n. While such basins function to provide hydrologic and PM control, studies have reported that it may be difficult fo r basins to achieve a target effluent suspended PM concentrations due to re-suspension of PM, and short-circuiting or scour due to lack of maintenance or uncontrolled in flow rates, unless such basins have sufficient
32 surface area and volume (Henderson and Bromag e 1988; Johnson and Chen 2006; Dumas and Bergheim 2001; Piedrahita 2003). The increasing cost of urban land, in part, has led to the consideration of small unit opera tions such as hydrodynamic separa tors (HS). A wide range of hydrodynamic separators have been developed as preliminary unit operations to separate gross solids and debris from stormwater flows a nd combined sewer overflows with regular maintenance (Brombach et al. 1993; Pisano and Brombach 1994; Andoh and Saul 2003). However, the HS is designed for separation of gross solids (Wong et al 1997; Allison et al. 1998; Walker et al. 1999) and not for finer suspended particles. In addition, such HS systems do not provide any form of hydrologic restoration. Recently, there is increasing in terest in stormwater filt ration, thereby promoting the removal of particles, particulat e-bound constituents and dissolved chemical species that primary unit operations such as basins otherwise discharg e as primary effluent. Media for such systems function to provide suspended PM separati on, pH control, adsorption, and/or surface complexation in order to provide effective mass tran sfer and control of chemical species (Liu et al. 2005; Liu et al. 2005; Sansalone and Buchberg er 1997; Revitt et al. 1990). In general, filtration technologies are typically accomplished by a replaceable fi lter media and they are more capable of removing finer particles than screening. As would be expected, filtration performance was associated with the effluent water quality (SSC, TSS, turbidity, PSD etc.), water production (unit filter run volume), and head-loss devel opment (Clark 2000). In addition, filtration performance also depends on many factors such as the surface loading rate, residence time, and the physicochemical characteristics of the media (t ype, size, porosity, and surface charge) (Liu et al. 2001; Liu et al. 2004; Teng and Sansalone 20 04). Media selection should be based on the environmental factors and cost of the project.
33 Field and laboratory studies ha ve been undertaken by several researchers to determine the phosphorus removal by using engineered media filtra tion systems. Han et al. (2003) showed that the soluble iron species deposited on juniper fi ber acted as a phosphorus adsorbent with sorption capacity about 2.2 mg/g. Erickson et al. (2007) fo und the steel-enhanced sa nd filtration retained between 25 and 99% of dissolved phosphorus rem oval. In addition, pumice has been tested and used in various environmental applications mainly as an adsorbent, filtration media, biofilm or catalyst support. It exhibited a hi gh potential for use as a filter be d material for turbidity removal under rapid filtration conditions (Farizoglu et al. 2003). Furthermore, pumice may provide more surface area for oxides coating, which may enhan ce the adsorptive removal of phosphorus (Kitis et al. 2007). A number of previ ous studies have indicated that activated alumina is able to remove phosphorus efficiently th rough adsorption and ion excha nge (Hano et al. 1997; Seida and Nakano 2002; Zhao and Sengupta 1998). Therefore, an engineered, designed media filtration system would be able to remove chemical cons tituents, including phosphor us, and targeting both dissolved and particulate phases from urban ra infall-runoff while avoidi ng the high construction and operation costs of a stormwater treatment facility or mechanic al treatment process (Erickson et al. 2007). It should be also noted that the removal of ch emical constituents such as metals, nitrogen and phosphorus transported in runoff also varied under different pH and redox conditions due to speciation and partitioning. Aerobic and an aerobic conditions are the most common circumstances occurring in a natural environmen t or in rainfall-runoff collection systems and treatment operation/process units. A singular or series of manageme nt structures and associated processes that were designed to drain surface water runoff, such as detention basins, were studied to understand their potential to a lleviate adverse effects (such as water quality deterioration) in
34 conventional drainage systems in urban envir onments (Butler and Davies 2000). The retention time generally varies between the seasons due to less or more frequent rainfall events, since seasonality is an important characteristic of a hydroperiod. A longer retention time was found to be correlated positively with a higher biodegr adation rate (Butler and Davies 2000; Scholz 2004). Urban stormwater runoff has the potential to change the pH and redox levels along with residence time in the treatment unit, which renders many toxins available in the storage pool so that they may have an immediate effect on receiving water systems, both in situ and potentially downstream (Cooke 1991). However, properly mana ging stormwater will a void problems with adverse impacts, in addition to erosion and flooding (WoodwardClyde Consultants 1991; Newton 1989; Stockdale 1991). This study focused on examining toxicity of PM and metals transported in urban rainfallrunoff and evaluating the hydrological response s and pollutant removal efficiency of a volumetric clarifying filter (VCF ) system. The performance of a series of single radial flow cartridges with filled engineered media was st udied by field treatment experiment under steady conditions and the VCF system was studied thro ugh the treatment of 19 actual rainfall-runoff events under unsteady conditions and detention mana gement. There are six major chapters in this dissertation. Chapter 2 examines the lethal and su blethal toxicity of part iculate matter (PM) in urban rainfall-runoff on early life stage of channel catfish and j uvenile stage fathead minnows. Chapter 3 investigates the bioaccumulation of me tals by fathead minnows and the joint actions of metals on growth inhibition of green algae. Chapter 4 examines the adsorption-filtration performance of a radial flow cartridge system with filled engineered media (AOCM)P and control media (perlite) under vari ous operating flow rates. Chapter 5 investigates the response of a source area watershed and volumetric clarifyi ng filter system to urban rainfall-runoff and
35 anthropogenic loadings. Chapter 6 examines the event-based treatment performance of the VCF system on PM transported by rainfall-runo ff from a small urban watershed (1088 m2) in terms of filtration and sedimentation. Chapter 7 examines the performance of phosphorus removal by the VCF system, including the concentration and mass reduction of the pa rticulate phosphorus by filtration and sedimentation and dissol ved phosphorus by chemical adsorption.
36 CHAPTER 2 ACUTE TOXICITY OF PARTICULATE MA TTER IN URBAN RAINFALL-RUNOFF ON FATHEAD MINNOWS AND CHANNEL CATFISH Introduction Stormwater runoff from urban land uses and cover conditions, particularly impervious pavement surfaces, has been recognized as a s ource of contamination im pacting receiving water ecosystems (USEPA 1984). The chemical compos ition has been reported for water quality constituents such as particulate matter (PM) metals, nutrients, petr oleum-related organic compounds (including polycyclic aromatic hydr ocarbons, PAHs) and herbicides applied for urban land uses dominated by pavement (Buckl er and Granato 1999; Marsalek 1999). PM is potentially a concern not only because of the envi ronmental and ecological issues related to PM itself, but also since many contaminants can be associated with these particles and transported through the aquatic environment (S ansalone 2002). Even though dissolved loads of metals are significant and most immediately bio-available (Sansalone and Buchberger 1997), a significant fraction of the pollutant load is also transported by PM (L ee et al. 1997; Marsalek 1999). Anthropogenic activities and infrastructure, pa rticularly transporta tion, construction and maintenance regimes, are significant contri butory sources of PM ranging from sub-micron particles to gravel-size materi al (Sansalone et al. 1998; Muschack 1990). For example, abraded pavement and vehicular tires were reported to constitute by mass 40-50% and 20-30% of total PM generated (Kobriger and Geinopolos 1984; Sa nsalone et al. 1998). From a water quality perspective, fine suspended particles are of specific importance due to their combination of mobility and high potential adsorptive capacity for natural macromolecules, heavy metals, nutrients and organic contaminants despite the fact that suspended particle s can represent about a third of the total partic ulate in source area runoff (Ran et al. 2000; Means and Wijayaratne 1982). In contrast, coarse particles transported by urba n rainfall-runoff are gene rally considered as a
37 significant contributor to PM mass and hence have a relative ly lower surface-to-volume ratio and specific surface area (SSA) for active constituents adsorption. However, the potential of their capacity for pollutants delivery and partitioning is not negligible. Therefore, in addition to contributing turbidity and particle mass load ing, PM in urban runo ff generally acts as a significant reservoir on which toxic constituents may be absorbed, transported, and released into receiving waters (McKenzie and Irwin 1983; Gj essing et al. 1984; Mo rrison et al 1990; Makepeace 1995). The transport of these constituents including PM from urban infrastructure and direct discharge to natural water systems can represent both an acute loading on an event basis and a chronic loading over durations of prolonged e xposure (Snyder, 1995). When these particulate contaminants remain in the water column, they are potentially more t oxic to fish and other aquatic life. Even when the sediment and settleab le fractions of the particulate gradation settle out in a receiving water system, these fractions can cause impairment to the aquatic system and represent a long-term source of anthropogenic load subjected to variable redox, pH and hydrodynamic conditions. Improper construction practi ces for infrastructure can generate high levels of turbidity that can impair karst springs resulting in large trout die-offs due to clogged fish gills (Werner 1983). For example, construction of an interstate in West Virginia resulted in large quantities of disturbed si lt and clayey particulates wash ing into a system of caverns, causing a large trout kill due to silt build-up in thei r gills (Garton 1977). On the other hand, PM entrained as bed load material in runoff may settle in receiving waters and interact with natural sediment in the deposition and re-suspension cycl es of streams, thereby impacting biota and depleting dissolved oxygen while also reduci ng conveyance capacities and increasing dredging frequencies (Breault and Granato 2000; James 1999) In addition, binding to sediment particles
38 alters the bioavailability and toxicity of contaminants. Alt hough the toxicity of dissolved contaminants has been examined, the toxicity of the same contaminant associated with PM and the toxicity of simply PM is poorly understood. A primary difficulty in examining the relative toxicity of dissolved versus particle-bound contam inants has been the inability to manipulate the phase concentration separately. Some studies have inferred the relative toxi city of particle-bound contaminants from indirect evidence (Langston 1984). Generally, adverse effects on aquatic ecosystem s may result from toxicant exposure that directly causes the death of an organism (acute effects) or pr oduces sub-lethal effects on the organisms ability to develop, grow, and reproduce in the ecosystem (chronic effects). In order to identify and control toxic inputs to aquatic systems, toxicity tests are performed and pollutant levels are extrapolated from these laboratory-derived data to field conditio ns that can be deemed safe for aquatic ecosystems (Anderson and DA pollonia 1978). Although there have been many studies on the toxic effects of pollutants in the dissolved phase, the action of PM has received less attention, especially for anthropoge nic PM transported in storm water runoff. This study utilizes two economically and ecologically important species, freshwater fathead minnows ( Pimephales promelas ) and channel catfish ( Ictalurus punctatus ), to address toxicity of PM in storm water. Information of this study is criti cal for effective treatabil ity for toxicity reduction, control and regulatory frameworks for storm wate r discharges and the receiving environment. Objectives This study examined toxicity of PM transporte d in source area pavement sheet flow from Interstate-10 over City Park Lake in Baton R ouge, Louisiana. The overall goal of this study was to examine the acute lethal and s ub-lethal toxic effect of PM in direct discharges, especially suspended and settleable PM in the urban rainfall-runoff. There ar e five objectives in this study. The first objective was to evaluate the lethal toxi city of this source area runoff on early life stage
39 organisms as a function of elapsed time and hydrograph. The second objective was to examine the lethal toxicity of the runoff in relation to PM size classes on early life stage organisms. The third objective was to examine the lethal toxic ity of runoff on juvenile stage organisms. The fourth objective was to determine the size class or gradation of the PM that contributed to fish gill function failure. The final objective was to determine the oxygen consumption rate of fish when exposed to suspended, settleable and sediment classes of PM. Background Physical and Chemical Charact eristics of PM in Urban Runoff PM transported by the urban rainfall-runoff is he tero-disperse. The smallest particles, those with one dimension less than 1 m, are usually called colloidal, and those that are larger than this limit are said to be gravitoidal (Gustafsson and Gschwend 1997). The operational definition of dissolved and particulate impur ities is frequently established by a 0.45 m pore-size membrane filter, but colloidal partic les can be smaller than this dimension. The fractions greater than 1 m are nominally designa ted as suspended (< 25 m), settleable (25 75 m) and sediment (> 75 m) fractions according to their size ranges and, more importantly, the settling mechanisms. The effect of gravity on the tr ansport of colloidal PM tends to be negligible compared with Brownian diffusion and compared with suspended (<25 m), settleable (25-75 m) and sediment (> 75 m) PM. Colloidal and suspended PM have significantly more surface area per unit mass. There is considerable evidence that partitioning between PM and aqueous phase has a major effect on the occurrence, tr ansport, fate and biological eff ects of natural and anthropogenic chemicals in aquatic systems (Allan 1986; Yan et al. 2000). Whether in pa vement runoff, urban storm water or any aqueous system, there is a temporal partitioning between heavy metals in solution and PM whether these solids are in susp ension (TSS) or are settleab le solids that may be
40 part of a fixed or mobile bed load. This dynamic process includes specific mass transfer mechanisms of sorption, ion exchange and surfac e complexation with both organic and inorganic sites on the solid matter. Add itionally, these partitio ning reactions are generally non-linearly reversible between the solid phase and the so luble phase concentrations (Sansalone 2002). Toxicity Mechanism of PM Adsorbed chemicals may enter an organism through the general body surface and respiratory surfaces (e.g., gills) as they gradua lly dissociate from PM to the water in immediate contact with these areas. Chemicals taken into the body across the lining of the gastrointestinal tract come from ingested particulate matter (including ingested sediments and suspended particles) (Heath 1987; Sorensen 1991). On the other hand, chemicals absorbed through dermal or oral exposure are subjected to first-pass elimination in the kidney or liver and then the gills. In contrast, chemicals absorbed across the gills en ter the systemic circulation directly with no possibility of first-pass elimination and asso ciated mitigation (Rand G.M. 1995; chapter 1: Aquatic Toxicology). The branchial epithelium of a fish gill is finely divided to provide a large surface area, and it is highly permeable to f acilitate gas exchange, thereby entraining the physiological penalty the metabolic cost of compensating ionic regulation. Many transport channels, exchangers, and enzymes in the gill epithelium are negatively charged, to which specific positively charged metal ions bind. Monovalent metals (e.g., Cu+, Ag+) can disrupt Na+ (and Cl-) uptake, and divalent metals (e.g., Cd2+, Zn2+) can disrupt Ca2+ uptake (Playle et al. 1989). According to the Biotic Ligand Model (B LM), there exists a competition for metal complexation between environmental ligands and bi otic ligands. PM has the capacity to interact physico-chemically (surface complexation, ligan d exchange, hydrophobic association) with a range of inorganic and organic contaminants including metals (Ran 2000). Additionally, fish gills are delicate and easily damaged by abrasive silt particles. As sediment begins to accumulate
41 in the filaments, fish open and close their gill s excessively to expunge the silt. If irritation continues, mucus is produced to protect the gi ll surface, which may impede the circulation of water over gills and interfere with fish resp iration (Berg 1982). Furt hermore, filter-feeding species produce mucus to trap particles on the gill rakers or on the roof of the mouth and then transport both to the esophagus. Experimental Factors in Toxicity Test According to Woltering (1984) and Viljoen (199 9), sensitivity of organisms to pollutants varies among assorted species, popu lations, and life stages. Species difference in susceptibility to chemicals may be due to differences in acces sibility. In addition, rates and patterns of metabolism and excretion can substantially affect susceptibility. It is im portant to choose the most suitable test organism or species (Roux 1990), to ensure reliable, relevant, and ecologically significant or meaningful results. Species of commercial, recreational, or ecological value that hold an important position in the food chain, lead ing to man or other important species, should be considered. Fish have been the most popular choice as test organisms because they are presumably the best understood organisms in the aquatic environment (Buikema et al 1982). In the assessment of toxicity the most signi ficant factors related to exposure are the kind, duration, frequency of exposure, and the concen tration of the chemical. Acute exposures to chemicals that are rapidly absorbed in hours to days generally produce immediate effects, but they may also produce delayed effects. Chronic exposure to chemicals may induce effects that develop slowly (weeks, months, or years). For instance, fine part icles that remain in suspension are considered to produce acute exposures when they are at a high enough concentration to be potentially trapped by gill rakers and produce gill trauma (Servizi and Martens 1987), gill flaring (Berg 1982; Berg and Northcote 1985), increas ed coughing frequency (S ervizi and Martens 1992) and even death (Servizi and Martens 1987). Sockeye exposed to volcanic ash by
42 Newcomb and Flagg (1983) experienced greate r mortality rates at lower concentrations, indicating that the combination of slightly larger, more angular particles in volcanic ash may cause higher mortality rates. Once they coagulat e and settle down to the sediment, the impacts tend to be chronic, especially for the bent hic invertebrate community. Chemicals bound to sediment may partition into the water column a nd then accumulate in fish tissues via direct exposure or through bioaccumulation across linked trophic levels. Increased water temperature can increase the so lubility of many substances, influence the chemical form, and govern the amount of oxygen dissolved in the water. Such changes can interact with direct deleterious effects of elevated temperatur e. The metabolic rate may double for every 10oC rise in temperature. The buffering capacity of water is related to pH and alkalinity and the highest level of protec tion for freshwater organisms is in the 6.5-8.5 pH range. Both temperature and pH are also the key factor for metal and phosphorus partitioning between the aqueous and solid phases, which is directly asso ciated with toxicity. In addition, it is well known that hardness can markedly affect toxicity, espe cially that of heavy metals. Higher levels of calcium in the fish tissues make the cell membranes in the gills less permeable, so that less metal can enter fish (Rand and Petrocelli 1985). Methodology Sample Collection All rainfall runoff samples were collected dire ctly from the bridge deck outfall using manual sampling at an experimental site loca ted on an elevated sect ion of Interstate-10 traversing City Park Lake in Baton Rouge, Louisi ana. The Baton Rouge Interstate-10 site (above City Park lake) is a 1,088-m2 complete-paved watershed com posed of two identical Portland cement concrete catchments with an average da ily traffic volume of 70,40 0 vehicles. The runoff
43 volume was collected from the elevated bridge catchment and drained through a vertical 5-m length of a 20.3-cm pipe into an instrumented sampling facility directly below the watershed. Discrete time-based samples were collected manually in dupl icate with 4-L wide mouth polypropylene bottles. Samples were composited into three sampling intervals determined based on the duration of the event and intervals designa ted as beginning, middle and end of the runoff duration. The beginning of the event is designated as the first flush ir respective of whether the transport was mass-limited or flow-limited (San salone and Cristina 2004). Samples taken during the middle one hour of the event are designed as a middle event flush (one hour flush) and at the end of the event as the end event flush (e nd flush). Flow rates were measured by using a Parshall flume and 70kHz ultrasonic sensor. Organism Holding Larval channel catfish ( Ictalurus punctatus ) younger than 20 days hatched from the same batch were obtained from the Aquaculture Rese arch Station, Agricultur e Center at Louisiana State University. They were used for an early life stage acute lethal test in a storm event on April 06, 2004. Juvenile stage fathead minnows ( Pimephales promelas ) with weight and length at 0.44 0.12 g and 3.48 0.32 cm (Mean SD) respectively were obtained from Dixie Fish Farm in Liberty, Mississippi. A st andard requirement for acute fish test s is that the length of the largest fish should not be more than twice that of the sma llest fish in the same test. Fish were held in a 200 L polypropylene tank containi ng de-chlorinated Baton R ouge city tap water. Water temperature varied between 19 to 22oC. All experiments were perf ormed at the temperature to which the fish had been acclimated for at least 48 hours. Following arrival, fish were fed daily, but were starved at least 24 hours prior to their introduction to the experimental conditions which ensured they were in a post-absorptive state. Fi sh that appeared to be stressed from visual observation were discarded a nd not used in the test.
44 Water Quality Analysis All samples were transported to the laborat ory immediately following the event for regular water quality analysis which included pH, disso lved oxygen (DO), temperature, conductivity, total suspended solid (TSS), turbidity, and part icle size distribution ( PSD) analyses. A HACH spectrophotometer was used to mon itor nitrite and toxic ammonia (NH3-N) concentrations since the ranges of these parameters were more of interest rather than the exact single values throughout the toxicity tests. Dissolved oxygen concentration wa s measured by using Orion 810 A plus oxygen meter. Turbidity was measured usi ng a Hach turbidimeter. The analytical method to determine total suspended so lids concentration (TSS) is Sta ndard Method 2540 D (defined as Residue, Non-Filterable). TSS was measured fro m the supernatant of an Imhoff cone; a wellmixed l-liter sample was placed in the Imhoff co ne and allowed to settle for 60 minutes. The sediment fraction (> 75 m) was separated by wet-sieving before being placed in the Imhoff Cone. The Imhoff Cone supernatan t was filtered through a nominal 1 m glass fiber filter and the residue retained on the filter was dried until a constant weight was achieved at 105C. Particle size distribution was car ried out with a laser-based pa rticle analysis instrument. The methodology uses laser light diffraction to measure the amount and patterns of light scattered by particles. Metal analysis was conducted with an inductive coupled plasma mass spectrometer (ICP-MS) (Elan 6000, Perkin-Elmer Sc ience). For dissolved metals analysis, the sample was filtered with a 0.45 micron filter, an d the filtrate was preserved with acid and analyzed by the method listed above. For part icle bound metals analysis, water samples containing particles were digested in nitric, perchloric acid to di ssolve all the metals. Particulates remaining in the digestate were removed by filtra tion. Soil standards were used as control to examine the QA/QA for particle metal analysis.
45 Acute Lethal Toxicity Test From each of the well-mixed composite sample s representing the first flush, one hour flush and end flush, a 4-L aliquot was filtered through a TSS filter. Each of the filtered samples was separated into two 2-L sub-sample s which were further filtered (0.45 m) and then placed in clean, labeled, pre-rinsed glass beakers. The same procedure was carried out for the non-filtered samples. The test fish were exposed in groups of 10 animals in 2-L batch test chambers (BTC) containing runoff test solutions or control tap water. Two replicat es were used for each treatment and control (20 animals per treatment or cont rol group). All tests were maintained under specified test conditions (static non-renewal exposure ; room temperature; photoperiod 16 h light and 8 h dark) and aeration was provided to ensu re that the concentra tions of dissolved oxygen and test substances did not fall below acceptable le vels. On the other hand, water quality analysis was conducted every 24 hours to ensure that the concentrations of metabolic products did not exceed acceptable levels. Fish we re transferred by using a soft, fine mesh nylon screen. The test initiated when the test organisms were first pl aced in BTCs containing treatment solutions or control water. Mortalities were recorded one hour afte r test initiation, and hourly or every five hours thereafter depending on the fish behavior. Fish were considered to be dead when they did not respond to gentle prodding. Dead fish were removed immediately to prevent fouling of the test solution. Particle Size Distribution (PSD) and Gill Function Composite urban rainfall-runoff samples were separated into two treatment groups after representative collection: untreat ed samples with full gradation of PM fractions contained and treated samples with settleable and sediment fractions excluded by 60 minutes of quiescent settling. Test fish were exposed to those runoff solutions and a correspon ding control represented
46 the runoff solutions without fish exposure. Both treatment and controls were properly mixed by using a sample rotator at the speed of 30 rounds per minute (RPM) during the test to achieve continuous suspension of particle s in the solution for fish expos ure in an appropriate manner. PSD analysis was conducted at 0 hour, 3 hours, 6 hours and 12 hours after test initiation to examine the total volume concentrations (TVC, L/L) as well as particle number distribution (PND). Oxygen Consumption Test Selected gradation of particles had been collected from previ ous storms at the site from which urban rainfall-runoff was collected. These particles were differentiated into sediments coarse particles (>75 m), settlable solids (25-75 m), and suspended solids (<25 m) and these were pre-soaked with de-chlorinated tap wa ter overnight before each test. The target concentration of total solid was at 300 mg/L acco rding to site mean concentration (Sansalone 2005). The test fish (juvenile stage fathead minnow s) were introduced into 500-ml culture flasks filled with test medium. Light mineral oil was then added into the system to form an oil layer in order to isolate oxygen from the waters surface. In addition, a stirrer system was applied to keep the particles suspended in th e test solution without causing stress for fish. An Orion 810A+ oxygen meter was used to record the dissolved o xygen concentration at specified time interval until the fish reached mortality. The oxygen consum ption of the fish remaining in the culture flask after each sampling interval was measured and corrected after all the fish body weights were known at the end of each test. The oxygen cons umption rate for test fish can be calculated according to the following equation 2-1: ) ( ) (00t t W V DO DO Mx t t Ox (2-1)
47 Where MO : Oxygen consumption rate (mg-g-1-hr-1); DOtx represents dissolved oxygen concentration in the water at time x, [mg/L]; DOt0 represents dissolved oxygen concentration in the water at the beginning, [m g/L]; W represents total body we ight of fish used (g); tx represents time x after initiation of test (hr) and t0 represents the initiation time of test (hr). Regression analysis was used to model the relationship between the oxygen consumed per gram of fish (O) and the time (t) used for th e oxygen consumption as shown in equation 2-2 and equation 2-3: ) ln( ) ln( t b a O (2-2) bt a O (2-3) Where a and b are the intercept and slope of the log-linear function. The oxygen consumption rate (MO) was determined in equation 2-4 by the differentiation of time variable: 1 b a Ot e b M dt dO (2-4) Regression analysis and pairwise t-test was used in a statisti cal analysis to evaluate and compare the difference of the impact, 0.05 was th e critical value in al l statistical analyses. Results Event Hydrology and Sampling For this study, rainfall-runoff volumes were co llected from four sepa rate storm events and the hydrographs for each event were shown in Figure 2-1 and Figure 2-2. The 06 April 2004 and 24 April 2004 storm events were grouped and compared as high intensity, high runoff volume events. In contrast, the 24 October 2004 and 22 April 2005 events were grouped as low intensity, low runoff volume events. First-flush, middle even t flush and end event flush were captured and the role of hydrologic transport on total suspended solid (TSS) and turbidity were examined across each event. In general, there exists a good, concurre nt relationship between TSS and
48 turbidity. For the high intensity ev ents, both TSS concentration and turbidity exhib it a strong first flush phenomenon and an exponential decay as a f unction of elapsed time o ccurred in spite of flow pattern. A similar trend was found in the 22 April 2005 event (low intensity event), though to a lesser extent. However, a typical proportio nate delivery of concen tration to hydrograph occurred in the 24 October 2004 event. It is impor tant to note here that concentrations are of interest in this study because an adverse response is usually el icited by exposure concentration and duration at certain degree. Generally, first flush presented the peak value of TSS and turbidity for all events with the October 24, 2004 event as an exception. Furthermore, differences were observed between the middle event and end event flush on TSS concen tration and turbidity but these were not as significant in comparis on to first flush events with times or higher concentrations by orders of ma gnitude. For the acute lethal tests on early life stage channel catfish, runoff volumes from the 06 April 2004 event were utilized and three other events were used for lethal tests on juvenile stage fathead mi nnows due to the lesser sensitivity they have in general. Water Quality and Exposure Conditions Water quality are summarized in Table 21 including temperature, pH, ammonia (NH3-N), nitrite (NO2 --N) and dissolved oxygen (DO) for all treatm ents and controls during 96 hours static non-renewal test. The pH ranged from 7.0 to 8.5 and ambient temperature changed from 19oC to 22oC. Ammonia and nitrite levels were below 0.05 mg/L and 0.4 mg/L in most tests respectively. Dissolved oxygen concentrations we re all above 5 mg/L which was essential to provide a healthy exposure condition. Figure 2-3 and Figure 2-4 illustrate the change of turbidity and metal partitioning over exposure time for the 06 April 2004 and 24 Octobe r 2004 events, respectively. For clarity of description, separate plots were created for Cu, Pb, Cd, Mn, and turbidity. The bottom plot of
49 Figure 2-3 and 2-4 indicated that turbidity valu es dramatically decreas ed in 24 hours after the test was initiated and remained relatively consta nt throughout the test af ter 48 hours of exposure. This trend suggested that the flocculation and se ttling process occurred during the exposure and thus particle concentrations and distributi on changed with exposure time although a gentle agitation was provided for aeration. Therefore, test organisms were most likely to be exposed to the highest concentration of par ticles including colloidal, suspended, settleable and sediment particles at the beginning, and followed by lowe r concentration of suspended and colloidal materials that remained in the runoff solution and which was commonl y seen in the field. In this case, coarser particles, especially sediment PM are not considered to have significant impact on test organisms due to the much shorter exposur e time. However, the potential chronic effect contributed by PM will become a long term issu e since toxic constituents bounded to them can be released gradually for years through partitioning. Dynamic partitionings of four selected meta l elements (Cu, Pb, Cd and Zn) over exposure time are presented in Figures 2-3 and Figure 24. Results show that orders of magnitude difference was observed between dissolved and particulate-bound metal concentrations for Pb for both events. It is interesting to note that me tals associated with PM did not follow the pattern of the turbidity change. For example, while pa rticulate-bound Cu, Pb and Zn decreased from 116.6 to 30.6 g/L, from 68.2 to 1.3 g/L and from 696.0 to 237.1 g/L in 24 hours of exposure after the first flush collected from the 06 Apr il 2004 event, respectivel y, particulate-bound Cd was found to maintain a relatively constant le vel at the concentration of approximately 10 g/L. Generally, three mechanisms explain the decr ease of particulate-bound metal concentrations: partitioning, flocculation and settling, and ab sorption and ingestion by test organisms. In addition, these dynamic changes also varied across each individual event monitored depending
50 on the dissolved metal concentration and particle-t o-particle interaction as well as particle size distribution. For the 24 October event, the ch ange of particulate-bound metal concentrations tended to be less significant. Two explanations can be made rega rding this phenomenon. First, a selective range of particles ingested or trap ped by test fish varied by species and more importantly by life stages. Second, mass ratios of particulate-bound metal to particles generally decreased in orders of magnitude from colloidal, suspended to settleable and sediment fractions. In other words, relatively coarse r particles removed by juvenile stage fish are more responsible for PM reduction in terms of mass concentration but less responsible for metals associated with them. Acute Lethal Toxicity Test for Channel Catfish The time-response curves shown in Figure 2-5 and Figure 2-6 illustrated that the cumulative mortality and probit values of mortal ity of post-larvae channel catfish change as a function of exposure time. The left plot in Fi gure 2 showed that th e cumulative mortality reached 100% in 10 hours for the first flush and the untreated (unfiltered) flush had higher cumulative mortalities than the corresponding filtered flush during exposure, indicating that the untreated flush containing PM tended to be more toxic. The probits values of cumulative mortalities are calculated based on a log scal e and the zero values represented the 50% cumulative mortalities. The medi an lethal exposure time (LT50) values were 1.7 hours and 4.9 hours for the unfiltered and filtered first flus h, 18 hours and 30 hours for the unfiltered and filtered middle flush, 24 hours and 41 hours for the unfiltered and end flush, and 13 hours and 43 hours for unfiltered and filtered event compos ite samples as shown in the figures. Generally, the LT50 values are smallest for the first flush and increase from first flush to end event flush which indicates that first flush is most toxic and the toxi city decreases over the event. When examining the data from event composite samples, it was noted that moderate
51 toxicity responses were observed and the LT50 values were in the range of those for discrete flush samples collected from the Baton Rouge site, indicating the possible average toxicity. In addition, results showed that all the untreate d samples are more toxic than treated ones as indicated by LT50 as a critical index us ed in this study. Acute Sub-lethal Test for Fathead Minnows Survival rate, values of probit and no obs erved effect time (NOET) for fathead minnows when exposed to urban rainfall-runoff collected from the 24 October 2004, 24 April 2004 and 22 April 2005 events are summarized in Figure 2-7, Figure 2-8 and Figure 2-9, respectively. The effect referred to in these tests is death. Larg e variations in toxic re sponses were observed among the three events studied, partly because the detection of toxic responses may be affected by the type of events monitored (high or low in tensity, mass-limited or flow-limited) as well as the antecedent dry period (ADP) prior to the ev ent. However, for all events the cumulative mortalities in 96 hours are le ss than 50% and no 96hr LT50 can be derived in this case as a consequence. A quick perusal of the survival curve indicate s that the most toxic responses were found in the first flush for three events (April 22, 2004; October 24, 2 004; and April 22, 2005), with average NOET values of 12.1, 5.2 and 5.2 hour s for untreated flush samples, and 48, 48 and 22.3 hour for the filtered flush samples, respective ly. It was further noted that untreated samples were likely to be more toxic than treated sample s for all events as indicated by both the survival curves and NOET values. As shown in Figure 2-7, the NOETs for treated samples were greater than 48 hours while those for untreated sample s were less than 12 hours for the 24 October 2004 event. Similar phenomena were observed for two other storm events. In addition, significant reductions in toxic responses were observed for the April 22 2004 event during the first one hour of the event, but change was hardly obs erved between the one hour flush and end flush
52 according to the survival data. Furthermore, survival rates were around 100% at 96hours and more than 85% in 7 days for both the 1-hour flus h and end flush samples, while they were less than 90% in 96 hours and 70% in 7 days for first flush samples. In general, these data indicated some toxicity reduction, although these reductions were relatively small. Fish Gill Function Test Figure 2-10 showed the particle size distribution (PSD) and incremental particle number distribution (PND) across th e range from 1 to 250 m for runoff samples before 60 minutes of quiescent settling. At the beginning of the test there was no significant difference between the samples with fish loaded as treatment group and t hose without fish loaded as control group. After 3 hours, the total volume concentra tions (TVC) of particles were lower in the treatment group as compared to that in control group. The TVC valu es were generally increased in 6 hours after the initiation of the test. The difference tended to be more significant after 12 hours, especially for the fine particles ranging from 5 m to 30 m. On the other hand, PND values also showed a significant difference for particles less than 10 m. In contrast, Figure 2-11 illustrate the PSD and PND change for samples afte r 60 minutes of quiescent settling. There was a slight difference between the treatment and control groups accordi ng to the PSD results. The PND results showed that the only significant difference exhibited for fine suspended pa rticles with size range less than 5 m after 6 hours of initiation of test. In general, PM with size range less than 25 m was highly associated with gill function failure potentially. Oxygen Consumption Test Comparison of total weight of fish with corre sponding total length of fish were fit to a first-order exponential model as shown in Figure 2-12. In addi tion, a significant correlation was established between fish oxygen consumption rate and total weight of fish in differen size rangesunder laboratory conditions. DO concentrat ion changed over time after 24 hours exposure
53 for the test solution containing PM (total solids, TS = 300 mg/L) with th ree different fractions: suspended, settleable and sediment particles. The slope of dissolved oxygen concentration to exposure time when exposed to suspended PM is si gnificantly greater than that when exposed to settleable and sediment PM, as well as the cont rol. Additionally, the oxygen consumption rates based on unit weight test fish were compared by using analysis of covariance (ANCOVA) as well as pair-wise test. All the R2-values indicate a goodness of f it for the linear relationship between oxygen consumption and time (Figure 2-14). Results indicated that there is no significant difference between sediment PM treatme nt and control which s uggest that particles larger than 75 m had no or little acute toxic effect or influence on the gill function. Discussion Urban stormwater can exhibit acute toxicity depending on the source, storm characteristics, timing during the storm, and overa ll drainage design. Sansalone a nd Buchberger (1997) reported that flow rate and duration of storm event co ntrolled the yield and size of transported PM. Generally, fine particles contribute most to the le thal and sub-lethal toxicity due to their larger specific surface area which allows more contaminants, including metals, to bind to them. In addition, fine solids could suspend in the wate r column and increase the exposure time for an organism. The larger particles ar e most considered to contribute to chronic effects after they settled down. It is generally accepted that larger particles (>75 um) play a minor role in the determination of acute toxicity. In most cases, fi rst flush samples were considered to be most toxic during the storm event because the accumulati on of contaminants and high concentration of road salts, solids and metals (Rokosh et al 1997) all contribute to increased first-flush toxicity. Sampling efforts focusing on the first flush ar e likely to produce more toxic responses, but volumes and durations of such flows are relatively small (Marsalek et al. 1999).
54 In general, immature or young neonatal organisms appear to be more sensitive to chemical agents than are adult organisms, since the diffe rences in degree of deve lopment of detoxication mechanisms may be involved in age-dependent t oxicity effects. Embryos may be less sensitive than adults because, at particular stages, they may have protective or impermeable membranes. Early life stages tend to be more susceptible to sediment due to their limited movement and tolerance (Servizi and Martens 1990). Even re latively low amounts of deposited sediment can limit inter-gravel water ex change, reduce interstitial dissolved oxygen, and effectively smother developing eggs and alevins (Scrivener and Brownlee 1989). The possible routes for toxicant entry in fish include direct uptak e across the gills, uptake by the sk in, and ingestion of particles. Pollutant bounded to particles could cause second ary exposure. For the juvenile and adult fish, ingestion is not likely as an expos ure pathway for fine particles. Bu t the epithelial ti ssues in gills are able to trap the suspended particles and they are not acutely toxic to juvenile and adult stage fish. The untreated samples that contained suspen ded and settleable solid s could facilitate the toxicity, especially for the first flush. Dissolved metals also contribute a lot because of their high bioavailability. There was a reasonable linear relationship be tween the oxygen consumption rate and the total weight of test fish. It might be due to the greater gill surface area per amount of body mass for smaller individuals of a sp ecies (Rani 1998). Fish exposed to the fine solid solutions and urban rainfall-runoff exhibited an increase in oxy gen consumption that was possibly associated with any observed variation in gill function. Fine particles (<25 um) at sufficiently high concentrations could be trapped by gill filament s and would result in mucus precipitation on the gills surface. The resulting increased metabolic energy demand will be reflected in the increased ventilation (such as gas exchange) of the fish (Grobler et al. 1989). An increase in oxygen
55 consumption rate in urban runoff tests might be associated to not onl y particles, but also dissolved fractions. Fish gills as a respiratory orga n also play an important role in the excretion and elimination process. Part of the dissolved he avy metals uptake by fish could be transferred by biotransformation and removed from the body by excretion. Thus, more oxygen and energy will be needed in this case. No difference of oxygen consumption rate between settled and unsettled runoff samples suggested that settleable and sediment particles was insufficient to significantly impact gill function. The information in this study will be helpful in selecting better BMPs for controlling the quality of stormwat er runoff, unit operation and process design on toxicity reduction. Conclusions Untreated urban stormwater can exhibit acu te and chronic toxicity, depending on the source loadings, receiving water conditions, aquatic species, granulometry and aqueous chemistry. Results of this study demonstrated that the finely suspe nded PM are the largest contributors to both lethal and sublethal toxicity, due to their ab ility to stay suspended in the water column and their interference with gill function. In addition, the association of contaminants with these particles, and the fact th at they remain entrained, also provides a bioavailability route for ingestion by the organism given the increas ed exposure time. With respect to acute toxicity, the larger particles (>75 m) play a minor role. In most cases, the first flush was most acutely toxic during the storm even t because of the washoff of accumulated contaminants and high concentration of solids an d metals all. Sampling efforts focusing on the first flush are likely to demons trate higher toxic responses, but volumes and durations of such flows are relatively small (Marsalek et al., 1999 ), and are generally not known a-priori.
56 Untreated runoff (as opposed to filtered runo ff) containing suspended, settleable and sediment PM can result in acute and chronic toxi city; in addition to dissolved contaminants such as metals which are readily bioavailable. A dynamic partitioning process happened continuously between dissolved and particulate phases and it was very difficult to separate them or consider each of them alone. Untreated runoff has highe r cumulative mortalities than runoff treated through only physical filtration; indicating the role of particulates on toxicity, since the sand or perlite filter material di d not reduce the soluble cont aminant concentrations. LT50 and NOET values indicate that higher particulate loads asso ciated with mass-limited events were most toxic, and for mass-limited events that toxicity decreased across the event. Soluble and particulatebound metals had a toxicity impact on early-life stage aquatic sp ecies since these species are potentially exposed to particulate and particul ate bounded chemicals through gill and ingestion. Fine particles with size range less than 25 m are easily trapped by gill tissue. Oxygen consumption results indicate that as total par ticulate concentrations increase, the suspended particles are more toxic than settleable and sedi ment particles. No diffe rence in aquatic species oxygen consumption rate under settleable and se diment loads suggested that settleable and sediment particles was insufficient to significantly impact gill function.
57 Table 2-1. Exposure condition and water qualit y analysis results during 96 hours for four storm events Event Exposure time (hr) Temp. (oC) pH NH3-N [mg/L] NO2-N [mg/L] DO [mg/L] 0 20.7 0.507.90 0.35 0 0 6.90 0.4 0 24 22.2 0.377.28 0.44 0.010 0.0100.20 0.056.63 0.9 7 48 21.1 0.177.53 0.21 0.022 0.0150.33 0.0 6 7.00 1.05 72 22.8 0.147.48 0.13 0.035 0.0060.33 0.0 6 6.97 0.4 0 April 06, 2004 96 22.5 0.207.48 0.12 0.037 0.0060.33 0.0 6 7.17 0.7 6 0 21.3 0.587.90 0.35 0 0 7.47 0.6 7 24 21.3 0.587.28 0.44 0.027 0.0150.50 0.0 6 6.53 0.75 48 21.0 0.007.530.21 0.037 0.0150.46 0.057.13 0.71 72 22.7 0.587.48 0.13 0.040 0.0170.43 0.11 7.07 0.4 2 April 24, 2004 96 22.3 0.587.48 0.12 0.050 0.0000.46 0.157.60 0.46 0 21.3 0.106.71 0.04 007.03 0.55 24 20.3 0.056.80 0.11 0.020 0.0100.33 0.156.87 0.25 48 20.3 0.107.05 0.03 0.040 0.0100.36 0.056.80 1.14 72 22.2 0.137.07 0.02 0.047 0.0060.36 0.066.63 0.47October 24, 2004 96 20.1 0.147.02 0.02 0.053 0.0060.36 0.067.03 0.31 0 20.1 0.107.01 0.01 007.30 0.46 24 20.8 0.086.92 0.01 0.023 0.0060.43 0.116.63 0.60 48 21.0 0.136.90 0.04 0.037 0.0150.36 0.066.43 0.57 72 21.2 0.106.86 0.04 0.040 0.0000.36 0.057.63 0.45April 22, 2005 96 20.5 0.186.94 0.12 0.057 0.0100.33 0.0 6 7.43 0.5 9 Notes: Data are presented as: Mean Concentr ation 95% confidence interval. Control group had a survival rate of 95%
58 Elapsed time (min) 050100150200250 Flow rate (L/s) 0 1 2 3 4 5 TSS [mg/L] 0 30 60 90 120 150 Turbidity (NTU) 0 100 200 300 400 500 Flow rate TSS Turbidity 06 April 2004 Sample point Elapsed time (min) 04080120160200 Flow rate (L/s) 0 1 2 3 4 5 TSS [mg/L] 0 100 200 300 400 500 600 0 1000 2000 3000 4000 5000 6000 Flow rate TSS Turbidity 24 April 2004 Sample point Figure 2-1. Exposure condition and water qualit y analysis results during 96 hours for four storm events.
59 Elapsed time (min) 051015202530 Flow rate (L/s) 0 1 2 3 4 TSS [mg/L] 100 150 200 250 300 Turbidit y ( NTU ) 0 200 400 600 800 Flow rate TSS Turbidity 24 October 2004 Sample point Elapsed time (min) 010203040 Flow rate (L/s) 0.0 0.2 0.4 0.6 0.8 TSS [mg/L] 100 150 200 250 300 Turbidity (NTU) 150 200 250 300 350 Flow rate TSS Turbidity 22 April 2005 Sample point Figure 2-2. Total suspended so lid (TSS), turbidity and rainfall -runoff flow rate change with elapsed time for two shorter rainfall-runoff events. The arrows show three sampling points: the first flush, middl e event flush and end event.
60 First flush Cu [ g/L] 0.1 1 10 100 1000 Middle flush End flush 0.1 1 10 100 1000 024487296 Turbidity (NTU) 10 100 1000 024487296 024487296 10 100 1000 0.1 1 10 100 1000 0.1 1 10 100 1000 Cu [ g/L] Pb [ g/L] Pb [ g/L] 0.1 1 10 100 1000 0.1 1 10 100 1000 0.1 1 10 100 1000 0.1 1 10 100 1000 Cd [ g/L] Cd [ g/L] Zn [ g/L] Zn [ g/L] Turbidity (NTU)Exposure time (hour) Figure 2-3. Turbidity values and Cu, Pb, Cd and Mn change with during 96-hour non-renewal static exposure for storm event April 06, 2004. Triangle dot was turbidity values. Solid circle and open circle were partic ulate and dissolved metals respectively.
61 First flush Cu [ g/L] 0.1 1 10 100 1000 Middle flush End flush 0.1 1 10 100 1000 0 48 96 144 192 Turbidity (NTU) 10 100 1000 0 48 96 144 192 0 48 96 144 192 10 100 1000 0.1 1 10 100 1000 0.1 1 10 100 1000 Cu [ g/L] Pb [ g/L] Pb [ g/L] 0.1 1 10 100 1000 0.1 1 10 100 1000 0.1 1 10 100 1000 0.1 1 10 100 1000 Cd [ g/L] Cd [ g/L] Zn [ g/L] Zn [ g/L] Turbidity (NTU)Exposure time (hour) Figure 2-4. Turbidity values and Cu, Pb, Cd a nd Mn change during 96-hour non-renewal static exposure for storm event October 24, 2004. Tr iangle dot was turbidity values. Solid circle and open circle were particulate and dissolved metals respectively.
62 Cumulative mortality (%) 0 20 40 60 80 100 Exposure time (hour) 110 mortality untreated filtered Exposure time (hour) 110 probits (log(M/S)) -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 probits untreated filtered LT501.74.9 Exposure time (hour) 110100 Cumulative mortality (%) 0 20 40 60 80 100 mortality untreated filtered Exposure time (hour) 110100 probits (log(M/S)) -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 probits untreated filtered LT5024 41 Figure 2-5. Turbidity Response curve for Cumulative mortality of channel catfish postlarva as a function of exposure time (left side) a nd the trend based on probit values (right side) BTR first flush and middle event flush.
63 Exposure time (hour) 110100 Cumulative mortality (%) 0 20 40 60 80 100 mortality untreated filtered Exposure time (hour) 110100 Probits (log(M/S)) -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 probits untreated filtered LT5018 30 Exposure time (hour) 110100 Cumulative mortality (%) 0 20 40 60 80 100 mortality untreated filtered Exposure time (hour) 110100 probits (log(M/S)) -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 probits untreated filtered LT501343 Figure 2-6. Response curve for Cumulative mortalit y of channel catfish po stlarva as a function of exposure time (left side) and the trend based on probit values (right side) for storm event April 06, 2004.
64 0.5 0.6 0.7 0.8 0.9 1.0 50 60 70 80 90 100 Probits 0.5 0.6 0.7 0.8 0.9 1.0 0 24 48 72 96 120 144 168 192 Survival rate (%) 50 60 70 80 90 100 0 24 48 72 96 120 144 168 192 0.5 0.6 0.7 0.8 0.9 1.0 50 60 70 80 90 100 Exposure time (hour) Filtered first flush Filtered middle flush Filtered end flush Untreated first flush Untreated middle flush Untreated end flush NOET NOET NOET NOET NOET NOET Figure 2-7. Response curve for Cumulative surv ival rate of juvenile fathead minnow as a function of exposure time (triangular dot). Solid curves represent probit values for storm event 24 October 2004. NOET repres ents the no observed effect time.
65 0.5 0.6 0.7 0.8 0.9 1.0 50 60 70 80 90 100 Probits 0.5 0.6 0.7 0.8 0.9 1.0 Exposure time (hour) 0 24 48 72 96 120 144 168 192 Survival rate (%) 50 60 70 80 90 100 0 24 48 72 96 120 144 168 192 0.5 0.6 0.7 0.8 0.9 1.0 50 60 70 80 90 100 Filtered First flush Filtered 1-hour flush Filtered End flush Untreated First flush Untreated 1-hour flush Untreated End flush NOET NOET NOET NOET NOET NOET Figure 2-8. Response curve for Cumulative surv ival rate of juvenile fathead minnow as a function of exposure time (triangular dot). Solid curves represent probit values for storm event 24 April 2004. NOET repres ents the no observed effect time.
66 0.5 0.6 0.7 0.8 0.9 1.0 50 60 70 80 90 100 Probits 0.5 0.6 0.7 0.8 0.9 1.0 Exposure time (hour) 0 24 48 72 96 120 Survival rate (%) 50 60 70 80 90 100 0 24 48 72 96 120 0.5 0.6 0.7 0.8 0.9 1.0 50 60 70 80 90 100 Filtered First flush Filtered middle flush Filtered End flush Untreated First flush Untreated middle flush Untreated End flush NOET NOET NOET NOET NOET NOET Figure 2-9. Response curve for Cumulative surv ival rate of juvenile fathead minnow as a function of exposure time (triangular dot). Solid curves represent probit values for storm event 22 April 2005. NOET represen ts the No Observed Effect Time.
67 Cumulative PND (e +8) 0 25 50 75 100 W/O fish W fish 0 hour TVC [uL/L] 0 10 20 30 40 W/O fish W fish 3 hour Particle diameter ( m) 1 10 100 Cumulative PND (e +8) 0 10 20 30 40 W/O fish W fish 6 hour Particle diameter ( m) 1 10 100 TVC [uL/L] 0 25 50 75 100 W/O fish W fish 12 hour Figure 2-10. Total volume concentration and cumula tive number distribution of particles across the size gradation from 1 to 250 m for storm water runoff before 60 minutes quiescent settling.
68 Cumulative PND (e +8) 0 10 20 30 40 50 W/O fish W fish 0 hour T60 (settled) TVC [uL/L] 0 3 6 9 12 15 W/O fish W fish 3 hour T60 (settled) Particle diameter ( m) 1 10 100 Cumulative PND (e +8) 0 10 20 30 40 50 W/O fish W fish 6 hour T60 (settled) Particle diameter ( m) 1 10 100 TVC [uL/L] 0 3 6 9 12 15 W/O fish W fish 12 hour T60 (settled) Figure 2-11. Total volume concentration and cu mulative number of pa rticles across the size gradation from 1-250 m for storm water runoff after 60 minutes quiescent settling.
69 LN(Total Length in cm) 1.01.11.21.31.41.5 LN (total weight in gram) -1.6 -1.4 -1.2 -1.0 -0.8 -0.6 -0.4 TW = 0.0162(TL) 2.6289 R 2 =0.89 n =12 Total weight (g) 0.30.40.50.188.8.131.52 V o2 (mghr -1 g -1 ) 0.3 0.4 0.5 0.6 V O2 = 0.6161 (TW) -0.1386 TW = 0.438 (0.115) g VO2 = 0.37 (0.016) mg-hr -1 g -1 R 2 = 0.64 TL 3.475 0.32 cm TW 0.438 0.12 g Mean S.D. Figure 2-12. The log linear relati onship between the total length and total weight of juvenile fathead minnow (upper plot). The log linear relationship between oxygen consumption rate based on unit weight of fish exposed and total weight of test fish.
70 Exposure time (hour) 01020304050 DO concentration [mg/L] 0 2 4 6 8 lethal point Control, 0mg/L < 25 m, 300 mg/L 25-75 m, 300 mg/L >75 m, 300 mg/L Figure 2-13. The dissolved oxygen concentrati on change with time fo r juvenile fathead minnows exposed in solution with solid concentration at 300mg/L for 24 hours.
71 Exposure time (hour) 05101520 Oxygen consumption (mg/g) 0 2 4 6 8 Least square means (Tukey) i/j T0 T60 control T0 0.0704 <.0001 T60 0.0704 <.0001 control <.0001 <.0001 T0 control T60 ANCOVA Figure 2-14. The amount of oxygen consumption base d on unit weight of fish change with time after 6 hours exposure to settled and uns ettled storm water runoff samples for October 24, 2004 event.
72 CHAPTER 3 THE BIOACCUMULATION AND JOINT ACTIONS OF METALS IN URBAN RAINFALLRUNOFF ON FATHEAD MIN NOWS AND GREEN ALGAE Introduction Urban storm water runoff is recognized as a cause of water quality degradation in many studies with anthropogenic constituent concentrat ions which may exceed toxicity thresholds (Horner et al. 1994; Makepeace et al. 1995; Morrison et al. 1993; Waller et al. 1995). The levels of zinc (Zn), copper (Cu), cadmium (Cd), lead (P b), arsenic (As), cromium (Cr), and nickel (Ni) in urban rainfall-runoff are found to be signi ficantly above background levels; and for many urban and transportation land uses, they often exceed surface water discharge criteria on an event basis for both dissolved and particulate-bound phas es (Foster and Charle sworth 1996; Sansalone and Buchberger 1997; Sansalone et al. 1998; Sansalone and Tribouillard 1999). In urban environments, infrastructure and vehicular compon ents with metals are ubiquitous. Metals in runoff, surface water and eventually groundwater can be generated from processes such as leaching, abrasion and oxidation of these materials (Makepeace 1995; Barrett 1993). Furthermore, unlike other constituents such orga nic compounds, metal speci es are not potentially degraded in the environment, constituting an impor tant class of acute and chronic toxicants with the potential to impact receiving waters, aquatic life, and ultimately the food web (Mishra et al. 2004). Among the metals detected in storm water runo ff, Cu, Zn, Pb and Cd are detected most frequently (US EPA 1983). Some metal species, su ch as Cu and Zn, are essential to aquatic organisms in trace amounts and they usually ha ve a narrow optimal concentration range for growth and reproduction, and both excess and sh ortage can be detrimental to organisms (Pelgrom et al 1994) and become toxic with unusually high concentrations (Wepener et al.
73 2001). Cadmium was found to act as an algal nutr ient under conditions of Zn limitation (Lee and Morel 1995). Other metals such as Pb have no known biological function (Seymore 1994). Certain of these non-essential trace metals are often found in the urban rainfall runoff and become the major contaminants of aquatic envi ronments (Munger et al. 1999) that have toxic effect on aquatic organisms (Witeska et al. 1995). Once storm water runoff is discharged to rece iving water systems, aquatic organisms are exposed to metals directly thr ough their contact with water and sediment, or indirectly through ingestion of vegetation, invertebrates or other smaller fish (Kime et al. 1996). In addition, many metal species in organic and inorganic forms have a tendency to accumulate in organisms to levels that lead to stress and potentially to mortality (James et al. 1998). Therefore, it is important to examine the bioaccumulation of metals in a system in orde r to assess the possible impact upon an organisms health either by exposu re to a pollutant or by primary or secondary consumption at different trophic levels, and ev en to assess possible impacts upon human health (e.g. types of fish consumed) (Kotze et al. 1999). It is commonly accepted that the metal speciation is crucial to evaluate its potential fate, bioavailability and toxicity; above and beyond the evaluation of dissolved and particulate fractions of a metal. MINTEQA2 is availa ble to calculate the chemical activity and concentrations of free metal ions and distribution of metal comp lexes (Allison et al. 1998). The high mobility of Cu and Cd can be attributed to complexation by ligands in solution, and of Cr to the presence of Cr(VI) species (Mason et al. 1999). Among the metals considered in some studies, Cu(II) has the highest affinity to organi c ligands, whereas Zn(II) is usually only weakly complexed. Complexing ligands typically increas e the concentration of dissolved metal and therefore the metal mobility (Davis et al. 1993; Lee et al. 1994).
74 Studies have demonstrated that metal uptake by aquatic organisms is a two-phased process, initially involving rapid adsorption or surface bi nding, followed by a slower transport into the cell structure. Transport of metals into the intracellular structure may be aided by either diffusion of the metal ion across the cell me mbrane or by active transport by a carrier protein (Brezonik et al. 1991; Wepener et al. 2001). Even though metal elimination routes are more abundant than uptake routes in some cases, metal accumulation occurs more rapidly than removal, probably due to the presence of metal binding proteins (metallo thionein) in the tissues of aquatic organisms (Kargin and Cogun 1999). The onset of toxicity can occur at any total body concentration, if the uptake rate changes such that it exceeds the comb ined rates of excretion and detoxification for a time period sufficient for the concentration of metabolically available metal to exceed a threshold. A number of researchers have determined th at the concentration of a substance in the organism (expressed as body concentr ation, critical internal con centration, tissu e residue, tissue concentration or body burden) was a better predictor of e ffect than water concentration, sediment concentration, or equilibri um partitioning (Niimi and Kissoon 1994; Ledin 2000). Few contaminated discharges release onl y one metal to aquatic ecosystems. The Environmental Protection Agency (1991) assumes that chemical toxicants act in an additive way, as opposed to being antagonistic (les s toxicity than predicted) or s ynergistic (greater toxicity than predicted). In general, joint toxicity often occurs among chem icals with a similar mode of action while chemicals with dissimilar modes of acti on (e.g., zinc and diazinon ) show antagonistic, little, or no interaction, such that the toxicity of a binary mixture shows toxi city less than that of the most toxic component (Howell 1985; He rbes and Beauchamp 1977; Schultz and Allison 1979; Deneer et al. 1988). These studies suggest that toxicity resulting from stressor mixtures is challenging to predict simply base d on additivity or chemical type However, in cases where one
75 toxicant of the mixture dominates, ignoring toxic interactions within th e mixture adds little uncertainty to the overall assessment. Metals provide an example. In practice, many metal mixtures are dominated by a speci fic metal element so that assessing the potency of the mixture on the basis of the most potent component is ofte n effective. A number of studies have shown that interactions of chemical mixtures can cha nge from antagonistic to synergistic based on the life stage of the organisms, concentrations or leve ls of the contaminants, or duration of exposures (Spehar and Fiandt 1986; Munawar et al. 1987). Howe ver, little of these previous studies have been devoted to the metals generated from urban rainfall-runoff. Therefore, an autecological study of metal bio accumulation and joint t oxicity is important for urban runoff impact bioassays, providing informa tion that cannot be accurately determined or extrapolated from other assessment components. Ch emical analyses, without biological analyses, typically underestimate the seve rity of the problems, in many cases (Lenat and Eagleson 1981; Lenat et al. 1981). Toxicity testing with the proper c ontrols will allow for sensitive and meaningful assessments of ecosystem quality and will identify stressor magnitude frequency, and duration. The juvenile fathead minnows Pimephales promelas and green algae Selenastrum capricornutum were used as model species in this study. We were particularly interested in whether the metal accumulation in fish coul d be described by one-compartment, uptakedepuration models and whether th is modeling was useful in term s of describing the accumulation process and understanding the chronic effects of metals resulted from urban rainfall-runoff. Additionally, we hypothesized the jo int toxicity of binary mixture of metals on the growth inhibition of green algae were additive. This st udy may help in efforts to evaluate the metal toxicity in urban rainfall-runoff in an empirical manner.
76 Objectives The overall goal of this study was to invest igate the chronic adverse effect of metal bioaccumulation on juvenile stage fathead minnows and the joint toxicity of selected binary metal mixtures on green algae. There were four objectives in this study. The first objective was to characterize the potential contri bution of metals to toxicity as well as their speciation in urban rainfall-runoff. The second objective was to ex amine metal bioaccumulation and depuration on fathead minnows Pimephales promelas as well as a one-compartme nt bio-concentration model evaluation on toxicity. The third objective was to examine the potential toxicity of urban runoff through determination of the medi an effective concentration (EC50) of metals for green algae Selenastrum capricornutum The last objective was to evaluate the joint toxic action of metal species in runoff on green algae Selenastrum capricornutum Methodology Sample Collection and Preparation Two experimental sets of samples were used in this study. In the first set, samples were experimental solutions prepared by diluting the re ference metal stock solution (Fisher Scientific Inc.) to the target concentrati ons based on the maximum event m ean concentrations (EMCs) of previous rainfall runoff collected at the Baton Rouge experimental site (Sansalone 2005). The initial target nominal concentrations are 600 g/L, 1,200 g/L, 300 g/L and 100 g/L for Cu, Zn, Pb and Cd, respectively. Reference concentrat ed stock solutions for Cu, Zn, Pb and Cd were freshly prepared in deionized (DI) water. Another set of samples were filtrate of urban rainfall runoff directly collected from the Baton Rouge e xperimental site. The filtrate was prepared by passing the rainfall runoff through a 0.45 m membrane filter. The experimental water temperature was maintained at around 25oC throughout the study. All test containers were soaked in 5% nitric acid for at least 24 hours, followed by rinsing at least three times with DI
77 water prior to use. MINTEQA2 was employed to calculate and compare the speciation distributions of Cu, Zn, Pb and Cd between the DI fresh water matrix and urban rainfall-runoff matrix. Experimental Animal (minnows) Holding Juvenile fathead minnows (0.39 0.12 g, mean S.D.) collected from Dixie fish farm in Liberty, Mississippi were acclimated in a 40-L st ock tank filled with standard fresh water based on DI matrix required by EP A (APHA 1992; ASTM 1992; U.S. EPA 1991) for 96 hours under a constant aeration condition prior to experiment ation, according to fish culturing and bioassay protocol by Standard Methods for the examinati on of water and wastewat er (APHA et al. 1992). During the acclimation period, fish were fed on alte rnate days with a commercial tropical flake at 3% of body weight and maintained on a photop eriod of 16 hours of light and 8 hours of darkness. Water was changed 8 hours after each feeding to remove excess food and metabolic waste. Less than 3% mortalities occurred durin g the preliminary acclimation period. Dead and abnormal individuals due to stre ss were immediately removed. Bioaccumulation Test After the preliminary acclimation, groups of 12 juvenile fathead minnows were randomly assigned to three 2-L batch test chambers (BTC) as replicates and exposed to DI-matrix based test solutions for Cu, Pb, Zn and Cd treatments at target nominal concen trations under static nonrenewal conditions for 9 days. DI-matrix standard freshwater controls were established as the treatments above simultaneously. Fish were not fed during the tests. Following 9 days of exposure, water samples were collected daily and total metal concentrations in the test medium were tested for metal uptake and bioaccumulation by test fish. At the onset and the end of each experiment, exposure and control waters were anal yzed daily to ensure that the water quality parameters were within safe levels for temper ature, pH, ammonia and nitrite concentrations.
78 Fish were checked at least three times daily for mortality and dead fish were removed and stored in a freezer as soon as they were observed and recorded. The fish in individual BTCs were sacrificed by freezing after 9 days of exposure. For metal analyses, the frozen fish samples were first thawed to room temperature, dried at 50oC overnight, weighed and then digested in 70% HNO3 (approximately 3 mL acid per g dry weight of tissue). After diges tion, all the samples were diluted to the concentrations appropriate for inductively coupled plasma mass spectrometry (ICP-MS, Perkin-Elmer Elan 6000) to determine the Cd, Cu, Zn and Pb concentrations. For quality assurance and quality cont rol, blanks and standard refere nce materials were treated the same way as the fish samples and used for these analyses to ensure the analytical results were within the expected range. The results of the accumulation experiments we re reported as concentration of metal per weight of fish ( g Me / g dry weight of fi sh). Another set of experiments was conducted to study the depuration rate by transferring surviving fish into clean DI matrix fresh water after exposure from each set of metal treatments. All glassware used in the metal analysis was soaked in 10% nitric acid for 24 hours and rinsed three times with DI water prior to use. The uptake of total Cu, Pb, Cd and Zn were first calculate d as a dry weight mass ratio (MR, g/g), defined as the ratio of the metal mass uptake per the unit weight test fish ( g/g) to the metal mass remaining in the aqueous solution per the un it weight test fish ( g/g). The uptake rate constant ( ku, Lkg 1 h 1) was calculated directly as the slope of the curve between metal uptake and exposure time. In contrast, depuration rate kd refers the metal loss from body per unit of time. One-compartment Bioconcentration and Depuration Model The uptake of metal from water is considered to be directly proportional to the exposure concentration (Cw) and the rate of depuration is directly proportional to the concentration in the organism (C). Steady state occurs when the rate of uptake balances the rate of loss, so that:
79 ss d w uC k C k dt dC 0 (3-1) d u w ssk k C C BCF (3-2) t k C BCF Cd wexp 1 (3-3) Where in equation 3-1, ku and kd are uptake rate constant (m g/gh) and first order rate constant for depuration, respectively. Css is the residual concentration in body at steady state. The steady-state bioconcentration fact or is defined as BCF above in equation 3-2 and equation 3-3. The time required to approach steady state, t, is determined by the depuration rate constant. The depuration of metals from the test organi sms transferred to uncont aminated water after a certain time of exposure to the contaminated water can be desc ribed as the equations below: t k C Log C Logd 0 (3-4) bt a C Log (3-5) Where t is the depuration time and a is the l og value of the residue concentration at the start of the depuration test. Green Algae Stock Culture Green algae Selenastrum capricornutum (UTEX 1648) cultures were obtained from Culture Collection of Algae, University of Texas (UTEX). To ensure that the stock cultures of this algae was grown axenically, a laminar flow cabinet was sterilized using UV light for at least 30 minutes prior to use. All of the following steps involved in tr ansferring algae were conducted in this environment. 2.0 mL of the starter al gal culture was aseptically transferred, using a disposable sterile pipette, into a 250 ml Erlenm eyer flask containing 50 mL liquid ASTM growth medium (ASTM 1995). The algal cultures were incubated at 24 2 oC under continuous cool white fluorescent illumination with an intensity of 200 mol m-2-s-1 PAR at the surface of the
80 flask. The flasks with the algae were placed on a continuous shaker at 100 rpm. Three to five days are required to reach the exponential gr owth phase which was the phase used for the toxicity tests described below. To ensure a re gular supply of exponentia lly growing algal cells, cultures need be renewed twice a week by aseptica lly transferring 2.0 mL of a stock algal culture between 3 and 5 days post-inoculation, to a 250 mL Erlenmeyer flask co ntaining 50 mL of fresh liquid growth medium. EC50 Determination The acute toxicity of indivi dual metal on green algae growth was first determined and hence the toxic index ( EC50) was used for the toxicity tests with a binary mixture of metals. In this study, a 96 hour acute toxi city test was carried out using Cu, Cd and Zn. Lead was considered as dominant in particulate-bound form and partitioning from particulate phase to dissolved phase was not significant in 96 hours, as determined from a previous study, thus it was not demonstrated in this test. Generally, a series of nine concentrations pe r test (including controls ) were prepared while each test concentration consisted of three replicates in which alga l cells were stocked in a 20-ml clean, autoclaved glass tube fo r each replicate. A preliminary or screening test was conducted first so that the test concentrations for the definitive test that followed could be refined into a more narrow or reasonable range. Growth can be measured by microscopi c cell counts using the hemacytate counting plate. On the other hand, density or biomass of algae measured by using spectrophotometer in terms of absorbance values fo r algae suspensions gave the same results as the cell count approach. The av erage specific growth rate ( ) for exponentially growing cultures can be calculated as the following equation 3-6:
81 1 21 2ln ln t t N Nt t (3-6) where t1 and t2 are incubation times taken at the beginning and end of the exponential growth phase and N was the density of algae cells Alternatively the average specific growth rate may be derived from the slope of the regression li ne in a plot of ln N versus exposure time. In this study, normalized growth rate was defined as the specific growth rate in the treatment sample divided by that in the control sample. In other words, normalized growth rates for control groups are always 1.0. The EC50 represented the metal concentr ation which caused 50% growth inhibition can be interpolated from th e resulting graph or probit analysis. Metal Joint Action Evaluation The binary metal mixture (Cu-Zn, Cu-Cd and Cd -Zn) were prepared in the same way as described and used in the single metal tests a bove, but the concentrati on ratios of the mixture were based on the ratios of EC50 values determined for those two metals from previous tests. For example, EC50 values determined for metal A and B were 100 /L and 200 /L respectively indicated that a series of concentr ations of A-B mixtures will have a same ratio of 1 : 2 for A : B. The joint action of metals was determin ed using the isobole method based on the assumption that the mixture toxicity of A and B can be estimated by dividing the concentration of each toxicant in the mixture w ith the concentration of a single applied toxicant that yields the same effect as the mixture. As shown in Figure 3-1, an isobole is constructed from different dose-response curves describi ng the two metal elements (A and B) both singly and in combination at isoeffective concentrations. A stra ight line (additivity line) connecting a specific effect (EC50) for each toxicant singly represents the comb inations of A and B when the toxicants elicit an additive effect. In this study, 95% conf idence intervals ( 95% CI) were applied to the fitted isobole in order to determine whether they overlap the additiv ity line statistically. Then, the
82 EC50 value of the metal mixture determined from the test above was partitioned into an EC50 fraction attributed to metal A (projected EC50-A) and another EC50 fraction to metal B (projected EC50-A), based on the ratio in the metal mixture. The projected EC50 plotted on the appropriate isobole for metal A and B fell within the 95% CI bounda ries and they were considered to have an additive joint toxicity ef fect (AD); the projected EC50 below the lower boundary of 95% CI were considered to have a synergistic effect (SY) wh ich indicated a greater effect than AD (A + B > A-B); and projected EC50 above the higher boundary of 95% CI were considered to have an antagonistic joint effect (AN) wh ich indicated that a lesser effect than AD occurred (A + B < AB). Results and Discussion Metals from Urban Rainfall-runoff A comparison between the metal concentrations (dissolved, particulate and total) of the current study in Baton Rouge at the I-10 site and those of previous studies conducted in Cincinnati at an I-75 site are pr esented in Table 3-1, together with both acute and chronic toxic rules in surface water quality standards given by the Louisiana Department of Environmental Quality (LDEQ) and water quality criteria gi ven by Ohio Environmental Protection Agency (OEPA). It should be noted here th at the rules were adjusted by site mean concentrations (SMC) of hardness which were at 78. 1 mg/L and 67.6 mg/L as CaCO3 for Baton Rouge site and Cincinnati site, respectively. As presented in Table 3-1, Cu, Cd, Pb and Zn in urban rainfall runoff from both sites were found at sufficient concentrations which would violate local water qual ity criteria and state standards based on these criteria. Sometimes Cu, Cd and Zn in the dissolved phase were also present above worst-case-based acut e toxic levels. This fi nding indicated that there is a potential
83 for Cu, Cd, Pb and Zn in urban storm water runo ff to be toxic to aquatic life in receiving water systems and suggests that efforts are requi red to minimize the negative effects. Bioaccumulation Mechanism and Metal Speciation For many agents, such as metals, the toxic e ffects that follow a single exposure are quite different from those produced by repeated expo sure. The relationship between the elimination rate and exposure is shown in Figure 3-2. Chemical A has the slowest elimination compared to chemical B and chemical C. Chronic toxic eff ects may occur, therefore, if the chemical accumulates in the biological system (rate of ab sorption exceeds the rate of biotransformation and/or excretion), if it produces irreversible toxic effects, or if there is insufficient time for the system to recover from the toxic damage within the exposure. Bioaccumulation can occur only if the rate of uptake of a chemical by an or ganism exceeds its rate of elimination. Metal elements in urban runoff partition be tween dissolved and pa rticulate-bound fractions and the partitioning is influenced by the paveme nt residence time, rainfall pH, physiochemical properties of solids and solubility of metal el ement (Sansalone and Buchberger 1997). It is commonly accepted that metal partit ioning as well as metal specia tion are crucial to evaluate a metals bioavailability. Figure 3-3 presents the ma jor inorganic speciation of dissolved Cd, Cu, Pb and Zn as a function of pH levels for both DI matrix fresh water (left side) and urban rainfallrunoff (right side). Calculations were performed using Visual MINTEQA2 at a temperature of 25C and based on the cations and anions concentrations for both matrixes. The distributions of free meta l ions and other major complex species for Cu, Zn, Pb and Cd as a function of pH are presented in Figure 33. According to the pH range of urban rainfallrunoff generally occurred from 6 8 (grey area in the plots), free metal ions appear to be the dominant form over others, in particular, for Bato n Rouge site with pH at 6.5 7.0 on average, although poorly buffered source pavement area wa s contributed to runo ff (Sansalone and Kim
84 2007). In addition, a similar pattern for those di stributions was observed for DI matrix fresh water and urban rainfall-runoff. However, the per centage of free metal ions dropped to a certain degree in urban runoff matrix, sugge sting a slight overestimation bu t more conservative toxicity is likely to occur by using DI matrix due to less competitive metal species presented. Furthermore, the MeCO3 (aqueous) species of Cu and Pb we re found to represent 40% and 20% of the total dissolved Cu and Pb respectively under neutral conditions. It is generally accepted that the free metal ion is more bioavailable, although some meta l-ligand complexes may also be sufficiently lipophilic to accumulate as well. Th erefore, as one of the primary targets for immobilization by control strategies for urban runo ff, Cd and Zn would receive more attention in terms of bioavailability. Campbe ll (1995) found that the order of toxic potential is free hydrated metal ion (Me2+) > inorganic complexes (e.g. with Cl-, SO4 2-, CO3 2-, OH-) > organic complex (with humic, hydrophilic acids). For i norganic complex in runoff, MeCO3 and MeOH become more abundant than other inorganic complexes as pH increases which poi nts to a decrease of toxicity with increasing pH. Metal Bioaccumulation After nine days of exposure, all test fi sh had healthy performance and no mortality occurred during the experimental period. Water qua lity analysis results showed the temperature, pH, ammonia and nitrite were all in the safe range. In addition, p values of analysis of variance (ANOVA) indicated that there wa s no significant difference (p > 0.05) among Cu, Zn, Pb and Cd treatment regarding water quality (Table 3-2) The accumulation patterns of minnows exposed to Cu, Zn, Pb and Cd are shown in Figure 3-4, Figu re 3-5, Figure 3-6 and Figure 3-7, respectively. Results indicated that metals in fish body burden were significantly elevated by day 9. However, the increasing trend of Cu and Zn uptake began to rise slowly until a plateau was reached after two days of exposure. This phenomenon indicated the different biological uptake mechanism for
85 Cu and Zn as the essential metals which tend to be more easily incorporated into metabolic processes crucial for the survival and growth of organisms (Linder 1991). Pb exhibited a similar result but further bioaccumulation seems possible indicated by the accumulation data point (solid square) shown in Figure 3-6. Minnows exposed to Cd did not reach a steady-state by the end of the 9 day uptake phase. In addition, our results showed that the fathead minnows were able to eliminate Cd and Zn from their body to a certai n extent during the 9 days of exposure, and further elimination seems possible. In general, the one-compartment model accurate ly described metals kinetics in fathead minnows in this study by comparing the measured data of the accumulation profiles obtained from the 9 day laboratory bioaccumulation bioassa y (solid square) to the modeled line (solid lines). The compartment describing the accumulate d metal fraction stored shows an exponential rise over the exposure period. Kine tic parameters values for Cu, Zn, Pb and Cd, calculated from the uptake and elimination patte rns of minnows exposed to corre sponding metal solution, are given in Table 3-3. The uptake rate (Ku) and elimination rate (Kd) was highest for Cu with values of 0.48 and 0.22 respectively. Generally, the uptak e of Cu, Pb and Cd was depurated from the fish much faster than the Zn. For most metals, biological uptake kinetics wa s shown to be faster than the chemical release kineti cs in sediments. Marinussen et al. (1997) found that Cu excretion is a biphasic process, distinguished by fast elimination kinetics followed by slow release ascribable to the inert stored metals. The relationships between metal exposure time and accumulation by employing kinetic and dynamic modeling were investigated to elucid ate the major mechanisms that account for the observed data and utilized the model to descri be the kinetics of me tals in minnows under different exposure concentrati ons (Figure 3-8). Results show ed that Cu, Zn, Pb and Cd
86 accumulation profiles differed among different leve ls of exposure concentration. For urban rainfall-runoff, metal concentrations in terms of EMC varied at even one order of magnitude depending on the rainfall intensity, traffic frequency, antecedent dry period and the characteristics of the watershed. Therefore, r unoff discharged at a low metal concentration may occur sometimes and seems to pose low or even no impact according to the modeled results. However, the accumulation of metals will neve r end in the receiving water body from a long term point of view, resulting in a significant bioaccumulation of h eavy metals in years or decades through multiple paths, including the food chain. During the initial stage of exposure, the accumulation mainly occurs in the kidney and liver of fish. When the amount of accumulated Cd exceeds the ability of the fish to synthesize the detoxifying metallo thioneins, localization of Cd to other organs of the fish, such as muscle, occurs. In brief, the accumulation factor in the exposed fish was highest for Cu and Pb, intermediate for Cd and low for Zn. The quantified metal Mass Ratios (MR, an inde x of dissolved metal uptake), during the 9 day uptake period in minnows exposed to Cu, Zn, Pb and Cd were illustrated in Figure 3-9. After 9 days of exposure, the MR of Cu, Pb and Cd increased significantly and reached to 0.26, 0.69 and 1.26 respectively. It is interesting to note that Cd accumulated quickly in fathead minnows and a similar quick response to aqueous Cd was reported in other laborat ory studies. However, a similar but less pronounced trend was also seen for Zn from the MR values which were approximately one magnitude less compared to the other three metal species. The depuration of metals from fathead minnows transferred to clean water after the 9 day bioaccumulation test, and is summarized in Table 3-4. The R2 values indicated that the log linear relationship between metal residue concentration and depuration time fit reasonably well. It was noted that only Pb demonstrated no significant depuration, it had a p va lue greater than 0.05.
87 Among the rest of the three metal species, the de puration rate of Zn was highest and Cd was the lowest in this study although all b values tende d to be very small. This phenomenon might be explained by the low metabolism of test organism after 9 days and the relatively strong binding between metal and protein. Metal Joint Toxicity Figures 3-10, Figure 3-11 and Figure 3-12 demonstrate the specific growth of green algae affected by the presence of Cu, Cd and Zn at a seri es of concentrations, re spectively. In general, the normalized growth rates based on the growth rates of the contro l group indicated that the expected inhibitory effect on growth became st rong as the metal concentration increased. The measured and modeled effective concentrations (EC) with 95% CI and median effective concentrations (EC50) determined from the probit model ar e also given in Figures 3-10 through 3-12. From the results, Cu was found to be the most toxic among the three metals tested individually with green algae, and th e toxicity order ba sed on the 48-hour EC50 was: Cu > Cd > Zn. The EMC of Cu, Zn, and Cd for dissolved and pa rticulate fractions were also demonstrated in Figures 3-10, Figure 3-11 and Figure 3-12, resp ectively. The grey hatched area represents the 95% CI of EMCs measured for the storm even t on June 30, 2005. No significant decrease of normalized growth rates was observed for Zn exposure until the concen trations exceeded 40 g/L, after which a sharp decrease occurred in terms of the normalized growth rate. Unlike the situation with Zn, normalized growth rates of gr een algae decreased gradually as metal exposure concentrations increased for Cu and Cd. The EC50 of Cu (90 g/L) was found to fall in the range between the EMC of the dissolved Cu (23 g/L) and particulate Cu (195.3 g/L) which indicated the potential toxicity due to the partitioning from particulate ph ase to dissolved phase based on the EC50 as a critical index. In other words, although the appa rent value of dissolved Cu
88 concentration is lower than EC50, the particulate bound Cu will be released into the water column once the dissolved Cu is assimilated by green al gae gradually. A similar result was found for Zn, which suggests that this metal cont ributed to the toxicity of the r unoff and affected the growth of green algae. The EC50 of Cd (150 g/L) happened to fall beyond the toxic zone and was considered to have less or no toxic effect during this event. It should be noted that the metal concentration also varies among events and si tes in addition to the partitioning process. Generally, as a heterogeneous mixt ure, runoff contained inorganic (NO3 -, SO4 2-, CO3 2etc.), organics (PAHs) and other metal species, which may a lter the toxicity in terms of the growth rate of green algae. Thus, in this case, the removal of particulates potential ly helps reduce toxicity, from the treatment perspective. The binary mixture of Cu and Cd is shown in Figure 3-13. If the mechanism of inhibition/toxicity is assumed to be the same fo r the two toxicants, the toxic effect would exhibit the same functional dependency on the con centration for both. According to the EC50 values of Cu and Cd, which were already known from Figure 3-10 and Figure 3-11, a series of concentrations of Cu-Cd mixture so lutions based on the ratio of Cu-EC50 and Cd-EC50 were employed to get the EC50 of the mixture from the normalized growth rate curve shown in Figure 3-13. The solid circle and curved line represen t the experimental data and probit modeled data, respectively. The growth rate ge nerally decreased with exposure to the Cu-Cd concentration, at a ratio of 1 : 1.7 (90:150) tested. The EC50 of the Cu-Cd mixture was calculated to be 69.4 g/L based on probit analysis and the project ed Cu and Cd fractions of Cu-Cd EC50 were plotted as solid square symbols, shown in Figure 3-13. It is obvious that the symbol was below the additive effect zone which was referred by the linear additiv e line and the area of 95% CI. Therefore, it is
89 possible that a synergistic effect (more than a dditive) exists between Cu and Cd on green algae growth. Synergism could also arise extracellularly when metals less toxic than Cu are able to displace Cu from inorganic or organic ligands oc curring in seawater and thus increase the free ion activity of Cu. It has recently been demonstr ated that Cu(II) ion activities rather than concentrations, determine Cu toxicities. Similar results were found for Cu-Zn mixture of and ZnCd mixture (Figure 3-14 and Figur e 3-15). Therefore, the observed toxic effect for all metal mixtures was substantially higher than that pred icted by assuming a simple additivity effect of the individual metals. Conclusions This study investigated the bioaccumulation and joint toxicity action of copper, lead, zinc and cadmium in urban rainfall-runoff generated from a small instrumental MS4 watershed in Baton Rouge, Louisiana. Two EPA re ference species (fathead minnows Pimephales promelas and green algae Selenastrum capricornutum ) were used in this study because of their economical and ecological importance. The following conclusions can be drawn based on the results in this study. First, the metals in urba n area and highway rainfall-runoff from various municipalities in the U.S. are at sufficient concentrations to cause viol ations of local water quality criteria and/or toxics rule criteria which were de veloped by the U.S. EPA for the State. Site Mean Concentrations (SMC) of Cu, Cd, Pb, Zn excee ded the acute and chronic critical values recommended by the Louisiana Department of Environmental Quality (LDEQ) and the Ohio Environmental Protection Agency (OEPA) for both sites. In addition, nearly all dissolved metal levels were above the acute toxic levels for both sites, which indicates th e potential short term toxic impact that the runoff may have and suggests that the effo rts such as Best Management Practices are required to minimize negative effects. Second, the free ion species for all the metals
90 are dominant species for Cu, Zn, Pb and Cd in urba n rainfall-runoff. pH is an important factor in determining the metal specia tion. A sharp decrease of Cu2+ and Pb2+ concentration with increasing pH above 6 and that of Cd2+ and Zn2+ with increasing pH above 8. As one of the primary targets for immobilizati on by control strategies, cadmium and zinc should receive more attention in terms of their bioavailability. Th ird, in general, the one-compartment model accurately described metals kinetics in fa thead minnows. The compartment describing the accumulated stored metal fraction stored shows an exponential rise over exposure period. The accumulation factor in the exposed fish was highest for Cu and Pb, intermediate for Cd and low for Zn. Although no toxicity endpoints were measured in the current study, the presence of Cd in these fractions would be expected to be associated with some biological im pairment, especially if concentrations were to increase beyond 7 days. F ourth, copper was shown to be the most toxic among the three metals tested individually with green algae and the toxicity order based on the 48-hour EC50 was: Cu > Cd > Zn. Although the apparent value of dissolved Cu concentration is lower than EC50, the particulate bound Cu will be rele ased into the water column once the dissolved copper is assimilated by green algae gra dually. The observed toxic effect for all metal mixtures was substantially higher than that pr edicted by assuming a simp le additivity of toxic effects for individual metals.
91 Table 3-1. Dissolved and particulate fractions of metal species in urban rainfall-runoff and their corresponding toxic rule s given by LDEQ and OEPA. Cu Cd Pb Zn As Cr LDEQ TR [ g/L] [ g/L] [ g/L] [ g/L] [ g/L] [ g/L] Acute 15.8 21.852.292.8339.8448.2(Cr3+)/16(Cr6+) Chronic 9.9 0.91.784.1150.0 145.4(Cr3+)/11(Cr6+) Interstate (I-10) runoff MC [ g/L], Baton Rouge, LA Dissolved 23.0 19.317.746.11.911.1 Particulate-bound 195.3 52.5117.71857.034.461.1 Total SMC 218.3 71.8135.31903.036.372.2 fd 0.1 0.30.10.10.10.1 Cu Cd Pb Zn As Cr OEPA TR [ g/L] [ g/L] [ g/L] [ g/L] [ g/L] [ g/L] Acute 9.3 2.758.884.1340.0413.5(Cr3+)/16(Cr6+) Chronic 6.4 184.108.40.20650.053.8(Cr3+)/11(Cr6+) Interstate (I-75) runoff MC [ g/L], Cincinnati, OH Dissolved 92.8 4.016.0946.0--11.9 Particulate-bound 42.2 3.048.4375.0--9.1 Total SMC 135.0 7.064.41321.0--21.0 fd 0.7 0.60.31.0--0.6 Note: LDEQ-TR represents toxic rules in surfac e water quality standard s given by Louisiana Dept. of Environmental Quality (adjusted by SMC of hardness at 78.1 mg/L as CaCO3). OEPATR represents toxic rules in statewide wate r quality criteria given by Ohio Environmental Protection Agency ((adjusted by SMC of hardness at 67.6 mg/L as CaCO3).
92 Table 3-2. Summary of water quality (Mean, Mi n, Max) for metal bioaccumulation test on fathead minnows. Metal species Temp. (oC) pH NH3--N [mg/L] NO2--N [mg/L] Cu (21.6, 20.8, 22.3)(7.34, 7.25, 7.45)(0.08, 0.00, 0.14) (0.04, 0.0, 0.1) Zn (21.5, 21.0, 22.0)(7.31, 7.18, 7.40)(0.07, 0.00, 0.12) (0.06, 0.0, 0.2) Pb (21.4, 21.0, 22.0)(7.24, 7.08, 7.41)(0.07, 0.00, 0.14) (0.04, 0.0, 0.1) Cd (21.4, 20.2, 22.0)(7.29, 7.12, 7.51)(0.08, 0.00, 0.12) (0.06, 0.0, 0.2) Control (21.1, 20.0, 22.2)(7.30, 7.15, 7.45)(0.07, 0.00, 0.11) (0.05, 0.0, 0.1) p-value 0.67200.27520.9507 0.8394
93 Table 3-3. Summary of kinetics parameters for the one-compartment bioaccumulation model. ku and kd represent the rate of uptake and depuration respectively. R2 indicate the goodness-of -fit for the model. Metal species Cu Zn Pb Cd Conc. [ g/L] 600 1200300 100 ku 0.48 0.060.21 0.14 kd 0.22 0.010.13 0.06 R2 0.97 0.950.98 0.89
94 Table 3-4. Summary of the depuration m odel kinetics parameters after 9 days bioaccumulation test. R2 indicate the goodness-of -fit for the model. Metal species Cu Zn Pb Cd Log(Me) 2.262.132.21 1.98 a 2.211.722.20 1.99 b 6.1 x 10-4 2.7 x 10-31.7 x 10-4 5.5 x 10-5R2 0.970.950.98 0.89
95 Figure 3-1. Isobole of metal A and B projec ted EC50 values for toxicity joint action assessment. The solid line represents the th eoretical values for additive effect. 95% CI 95% CI EC50 (B) 95% CI 95% CI EC50 (A) Plot (projected EC50-A, projected EC50-B) AD: additive AT: antagonistic SY: synergistic SY AD AD AT EC50 (A+B) a/(a+b) EC50 (A+B) b/(a+b)
96 Figure 3-2. Diagrammatic view of the relationshi p between dose and concen tration at the target site under different conditions of expo sure frequency and elimination rate. A B C A B C Single Dose Repeated Doses Concentration Range of Toxic Response (shaded area) Concentration at Target Site Time Time Bioaccumulation Bioaccumulation Low High
97 Cd2+ Species distribution (%) 0 20 40 60 80 100 CdCO3(aq) CdOH+ CdCl+CdSO4(aq) Cu2+ Species distribution (%) 0 20 40 60 80 100 CuCO3(aq) CuOH+ CuSO4(aq) Cu(CO3)2 2Pb2+ Species distribution (%) 0 20 40 60 80 100 Pb(OH)2(aq) PbCO3(aq) PbOH+ PbSO4(aq) Zn2+ pH 5678910 Species distribution (%) 0 20 40 60 80 100 Zn(OH)2 ZnCO3(aq) ZnOH+ ZnSO4(aq) (aq) Cd2+ Species distribution (%) 0 20 40 60 80 100 CdCO3(aq) CdOH+ CdCl+CdSO4(aq) Cu2+ Species distribution (%) 0 20 40 60 80 100 CuCO3(aq) CuOH+ CuSO4(aq) Cu(CO3)2 2Pb2+ Species distribution (%) 0 20 40 60 80 100 Pb(OH)2(aq) PbCO3(aq) PbOH+ PbSO4(aq) Zn2+ pH 5678910 Species distribution (%) 0 20 40 60 80 100 Zn(OH)2 ZnCO3(aq) ZnOH+ ZnSO4(aq) (aq) DI Matrix Stormwater Matrix Figure 3-3. Distribution of inor ganic metal species as a functi on of pH for DI matrix and stormwater matrix.
98 Cud Exposure time (day) 0246810 Uptake [ g/L] 0 60 120 180 240 Exposure concentration [ g/L] 0 200 400 600 800 Accumulation (%) 0 20 40 60 80 Uptake Concentration Accumulation Modeled 95% CI Figure 3-4. Uptake, accumulation and concentra tion change of Cu for juvenile fat head minnows as function of exposure time in a 2-L aerated batch test chamber. The dash lines are 95% CI of modeled curves.
99 Znd Exposure time (day) 02468 Uptake [ g/L] 0 50 100 150 200 250 Exposure concentration [ g/L] 1000 1100 1200 1300 1400 1500 Accumulation (%) 0 5 10 15 20 25 Uptake Concentration Accumulation Modeled 95% CI Figure 3-5. Uptake, accumulation and concentra tion change of Zn for juvenile fat head minnows as function of exposure time in a 2-L aerated batch test chamber. The dash lines are 95% CI of modeled curves.
100 Pbd Exposure time (day) 0246810 Uptake [ g/L] 0 50 100 150 200 Exposure concentration [ g/L] 0 100 200 300 400 Accumulation (%) 0 20 40 60 80 Uptake Concentration Accumulation Modeled 95% CI Figure 3-6. Uptake, accumulation and concentra tion change of Pb for juvenile fat head minnows as function of exposure time in a 2-L aerated batch test chamber. The dash lines are 95% CI of modeled curves.
101 Cdd Exposure time (day) 0246810 Uptake [ g/L] 0 30 60 90 120 150 Exposure concentration [ g/L] 0 30 60 90 120 150 Accumulation (%) 0 20 40 60 80 100 Uptake Concentration Accumulation Modeled 95% CI Figure 3-7. Uptake, accumulation and concentra tion change of Cd for juvenile fat head minnows as function of exposure time in a 2-L aerated batch test chamber. The dash lines are 95% CI of modeled curves.
102 Cu Metal Bioaccumulation mg/g D.W. 0 100 200 300 400 600 g/L 500 g/L 400 g/L 300 g/L 200 g/L Pb 500 g/L 400 g/L 300 g/L 200 g/L 100 g/L Cd Exposure time (hour) 050100150200250300 500 g/L 400 g/L 100 g/L 50 g/L Zn Exposure time (hour) 050100150200250 0 50 100 150 200 250 1200 g/L 1000 g/L 800 g/L 600 g/L 400 g/L Figure 3-8. Metal bioaccumulation for juvenile fat head minnows as function of exposure time based on the one-compartment model.
103 Exposure Time (day) 0246810 Metal mass ratio (Meu : Mew) 0.001 0.01 0.1 1 10 Cu Zn Pb Cd Figure 3-9. Influence of metal Mass Ratio (MR) on exposure time in a 9-day for Cu, Zn, Pb and Cd on fathead minnows. Meu indicates the uptake of metal in mass, Mew indicates the amount of metal mass re maining in the aqueous solution.
104 Figure 3-10. The EC50 (median effective concentration) values of Cu on the growth rate of green algae. The hatched areas indicate the 95% CI of EMC values for dissolved and particulate fractions of Cu for June 30, 2005 event. Concentration [ g/L] 101001000 Probit 0.0 0.2 0.4 0.6 0.8 1.0 Normalized growth rate (%) 0 20 40 60 80 100 modeled Cu EC modeled Cu EC 95% LOCL modeled Cu EC 95% UPCL measured Cu EC CP 195.3 EMCD 23Cu EC50= 90 g/L Partitioning
105 Concentration [ g/L] 101001000 Probit 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Normalized growth rate (%) 0 20 40 60 80 100 120 140 EMCd 19.3 EMCP 52.5 modeled Cd EC modeled Cd EC 95% LOCL modeled Cd EC 95% UPCL measured Cd EC EC50= 150 g/L Cd Figure 3-11. The EC50 (median effective concentration) values of Cd on the growth rate of green alage. The hatched areas indicate the 95% CI of EMC values for dissolved and particulate fractions of Cd for June 30, 2005 event.
106 Concentration [ g/L] 101001000 Probit 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Normalized growth rate (%) 0 20 40 60 80 100 120 140 EMCd 46.1 EMCP 1857.1 modeled Zn EC modeled Zn EC 95% LOCL modeled Zn EC 95% UPCL measured Zn EC EC50= 250 g/L Zn Figure 3-12. The EC50 (median effective concentration) values of Zn on the growth rate of green algae. The hatched areas indicate the 95% CI of EMC values for dissolved and particulate fractions of Zn for June 30, 2005 event.
107 Joint toxicity of (Cu + Cd) Cu concentration [ g/L] 050100150200250300 Cd concentration [ g/L] 0 100 200 300 400 Cu-Cd (1:1.7) effective concentration [ g/L] 050100150200250300 Normalized growth rate (%) 20 40 60 80 100 120 Synergistic Cu, Cd EC50 95% FL Cu, Cd EC50 Cu, Cd EC50 95% FL measured Cu-Cd EC modeled Cu-Cd EC Fraction of joint Cu-Cd EC50 Figure 3-13. The projected Cu and Cd fractions of EC50 values determined from the EC50 value of Cu and Cd mixture based on the EC50 ratio (90:150 = 1:1.7) was plot on Cu axis and Cd axis and indicate the synergistic action.
108 Cu concentration [ g/L] 050100150200250300 Zn concentration [ g/L] 0 100 200 300 400 500 600 Cu-Zn (1:2.8) effective concentration [ g/L] 050100150200250300Normalized growth rate (%) 0 20 40 60 80 100 120 synergistic Cu, Zn EC50 95% FL Cu, Zn EC50 Cu, Zn EC50 95% FL measured Cu-Zn EC Modeled Cu-Zn EC Fraction of joint of Cu-Zn EC50 Figure 3-14. The projected Cu and Zn fractions of EC50 values determined from the EC50 value of Cu and Zn mixture based on the EC50 ratio (90 : 250 = 1 : 2.8) was plot on Cu axis and Zn axis and indi cate the synergistic action.
109 Cd concentration [ g/L] 0100200300400500600 Zn concentration [ g/L] 0 200 400 600 800 1000 Cd-Zn (1:1.7) effective concentration [ g/L] 0100200300400500600Normalized growth rate (%) 0 20 40 60 80 100 120 synergistic Cd, Zn EC50 95% FL Cd, Zn EC50 Cd, Zn EC50 95% FL measured Cd-Zn EC Modeled Cd-Zn EC Fraction of joint of Cd-Zn EC50 Figure 3-15. The projected Cd and Zn fractions of EC50 values determined from the EC50 value of Cd and Zn mixture based on the EC50 ratio (150 : 250 = 1 : 1.7) was plot on Cd axis and Zn axis and indi cate the synergistic action.
110 CHAPTER 4 STORMWATER ADSORPTIVE-FILTRAT ION TESTING OF A RADIAL FLOW CARTRIDGE SYSTEM Introduction Urban storm water runoff from impervious ur ban areas leaches, mobilizes and transports dissolved, colloidal and particulate constituents in a heterogeneous mixture, which includes metals, phosphorus and organic constituents (San salone et al. 1998; Lee and Bang 2000; Muller and Sigg 1990; Igloria et al. 1997; Sansalone and Buchberger 1997). The chemical species partitioning between particulate and dissolved phases depend on the hydrology, hydrodynamics and chemistry of the event and watershed (San salone and Buchberger 1997; Liu et al. 2004; Dean et al. 2005). In addition, it is also impor tant to understand that particulate delivery and granulometry which play an important role in partitioning and separa tion of PM (Stumm and Morgan 1996; Sansalone et al. 1998). From a water quality perspective, PM, having reactive sites and large surface-to-volume ratios, can me diate partitioning and transport of chemical species while serving as reservoirs for many reactive constituents (Sansalone et al. 1998). While the chemical species partition and distribute acro ss the entire PM size gradation, the suspended PM is potentially more mobile a nd acutely bio-available than the coarser settleable and sediment fractions (Cristina and Sansalone 2003). A variety of in situ treatment strategies (u nit operations and processes, UOPs) have been developed to control the quality and quantity of storm water in ur ban areas over th e last several decades, such as found in settling basins a nd filtration (Schueler 1987; Li et al. 1999). Traditionally, settling basins have been the most prevalent unit operation since such basins can provide hydrologic control and PM separation. While such basins function to provide hydrologic and PM control, studies have repo rted that it may be difficult for ba sins to achieve target effluent suspended PM concentrations due to re-suspension of PM, short-circuiting or scour due to lack
111 of maintenance or uncontrolled inflow rates, un less such basins have sufficient surface area and volume (Henderson and Bromage 1988; Johnson and Chen 2006; Dumas and Bergheim 2001; Piedrahita 2003). The increasing cost of urban land, in part, has led to consideration of small unit operations such as hydrodynamic separators. A wide range of hydrodynamic separators (HS) have been developed as preliminary unit operati ons to separate gross solids and debris from stormwater flows and combined sewer overflows (CSO) with regular maintenance (Brombach et al. 1993; Pisano and Brombach 1994; Andoh and Sa ul 2003). However, the HS is designed for separation of gross solids (Wong et al. 1997; Allison et al. 1998; Walker et al. 1999) and not for settleable and sediment PM. In addition, such HS systems do not provide any form of hydrologic restoration. Recently, there has been increasing interest in stormwater filtration th at removes particles, particulate-bound constituents and dissolved chem ical species that primary unit operations such as basins otherwise discharge as primary efflue nt. Media for such systems function to provide suspended PM separation, pH control, adsorp tion, and/or surface complexation to provide effective mass transfer and c ontrol of chemical species (L iu et al. 2005; Sansalone and Buchberger 1997; Revitt et al. 1990). Objectives While the role of flow rate, head loss, medi a size and influent gra nulometry is generally understood for stormwater adsorptive-filtration beha vior, the concepts of surface loading rates, mean residence time and the role of media propert ies and pollutant phase on filter behavior is not as commonly measured or described. Theref ore, this study examined and quantified these factors under conditions of consta nt hydraulic, granulometric and chemical phase loadings. In order to investigate the adsorptiv e-filtration behavior of a single radial flow cartridge system, three objectives were carried out. These objectives were separated into two series of
112 experiments, one set of experiments focused on physical filtration and the other set focused on PM and total phosphorus (TP) removal by combining adsorption, filtration a nd partitioning. The first objective was to examine the PM mass remova l of a single radial flow cartridge under a range of flow rates from 16.7% 133.3% of de sign flow rate (1.14 L/s) for engineered media (AOCM)p and perlite, the control media. The sec ond objective was to examine the change of particle size distributions (PSDs) and head loss for the silt-size PM influent at different flow rates and bed volumes treated. The third objectiv e was to examine the total phosphorus (TP) and total dissolved phosphorus (TDP) ma ss and concentration reduction by a radial flow cartridge for each media type. Methodology Testing Configuration and Description As shown in Figure 4-1 the testing systems major component was a single radial flow cartridge. The experimental system consisted of a slurry tank, single car tridge test tank, and constant level control tank. The slurry tank wa s designed to feed the well-mixed slurry PM solution into the injection line and load the diluted slurry at the ta rget concentration level into the cartridge test tank. The test tank was used to hold the cartridge and served as a container to stabilize the inflow. A constant level control tank downstream was insta lled as a container to control the effluent water level and keep the head and outflow re latively constant. Details of the setup, dimensions of cartridge and characteristics of engineered media used are summarized in Figure 4-1. The water source was potable water from a water main tap connection located 50 meters from the 40,000 L influent water tank, which served as the raw water supply to the radial flow cartridge system. The filter system pump had a capacity of 6.97 L/s and a PVC delivery system was used for the range of flow rates utilized in this study. Th e flow was monitored using a 50.8
113 mm ( ) water supply meter (DLJ multijet) connect ed upstream of the influent and slurry injection line. The flow monitoring system was ca librated over all target flow rates ranging from 0.19 L/s to 1.51 L/s (16.7% to 133.3% of design flow rate). Prior to the initiation of each experiment, tanks, cartridge, test chamber and th e entire experimental system were thoroughly cleaned using potable water. This cleaning invo lved washing and flushi ng with clean potable water throughout the entire system and was intended to ensu re that there was no PM remaining in the system. The 40,000 L potable water tank c ontained potable water conditioned to a pH of 6.5 established with hydrochloric acid, repres entative of conditions for source area pavement runoff (Sansalone and Buchberger 1997; Sansalone et al. 1998; Dean et al. 2005). Before an experiment the entire volume of the cartridge wa s pluviated with a measured dry mass of clean media. Pressure transducer input to the data logger was checked before each experiment and head measurements manually verified. The flow rate was adjusted off-line of the experiment by isolating the test cartridge thr ough valves until a steady-state targ et flow rate was achieved. Once steady experimental conditions were achie ved an experiment was initiated, and data collection and sampling were started. Immediat ely after any experiment that utilized PM the cartridge was back-washed while in the test cham ber and all backwashed PM was recovered. For all experiments a mass balance check wa s conducted on PM and also phosphorus. PM captured by the single radial flow cartridge syst em from each run was completely recovered as a wet slurry from the test chamber and media cartri dge. PM slurries were completely dried at 40 oC and weighed to obtain the mass removal effici ency and mass balance check. Mass balance error (MBE) was required to be within the range of % by mass and determined by the following equation 4-1: 100 Load) Mass (Influent solids)] Trapped Load Mass (Effluent Load) Mass [(Influent (%) MBE (4-1)
114 Radial Flow Filtration Sy stem and Engineered Media The filter system used in these experiments is a passive, radial flow cylindrical cartridge that was deployed with several media. The diamet er and height of the cartridge are 457.2 mm and 558.8 mm, respectively. The media was held be tween an outer screen and inner mesh with plastic covers on the top and bottom. The stab ilized inflow passing th rough the media comes into the 76.2 mm ( ) central PVC drainage pipe which c onnected with the storm drain line directly. Two media types were examined in this study, an aluminum oxide-coated pumice (AOCM)P and a control media, perlite. The (AOCM)P had an alumino-silicate substrate of pumice that contains potassium, sodium, ma gnesium, calcium, and lesser amounts of iron. Pumice is a naturally-occurring po rous volcanic stone with pore volume up to 85%, and specific surface area (SSA) in the range of 5 to 30 m2/g (Kitis et al. 2007). Pumice has been utilized as a filter media for the removal of targeted pollutants such as particulate matter, metals, organics and phosphorus from water (Farizoglu et al., 2003, Kitis et al. 2007). Pu mice has also been used as a biofilm support material in wa ter and wastewater treatment (K ocadagistan et al. 2005). Once coated with aluminum oxide, the (AOCM)P had a mass-based median size (d50) of 3.56 mm (SD = 0.79); a specific gravity of 2.35 g/cm3; an SSA of 0.94 m2/g with a standard deviation of 0.17 m2/g, a porosity of 36.7 % and a dry bulk density of 0.68 g/cm3 at a pluviation height of 10 mm. The single radial flow cartridge co ntained 49830.4 dry grams of (AOCM)P media in a volume of 72975 cm3. One bed volume for (AOCM)P media represented 54.1 L Perlite is a naturally occurring volcanic glass used as a filter aid in many applications (Acemioglu 2005). But only few published studies can be found on the use of perlite as an adsorbent. Perlite of differe nt types (expanded and unexpanded) and of different origin would have different properties because of the di fferences in composition (Mathialagan and
115 Viraraghavan 2002). The perlite we used in this study had a mass-based median size (d50) of 3.81 mm (SD = 1.03), a specific gravity of 2.30 g/cm3, an SSA of 5.6 m2/g with a standard deviation of 1.3 m2/g, a porosity of 83.4% and a dry bulk density of 0.053 g/cm3 at a pluviation height of 10 mm. The single radial flow cartridge cont ained 4,108.5 g dry grams of perlite media in a volume of 72,975 cm3. One bed volume for conventional perlite media represented 68.0 L. Before loading the cartridge with perlite, perlit e had to be prepared by passing the raw perlite through the No. 8 (2.36 mm) sieve si nce perlite is very friable and has a tendency to crush into a very fine powder. The sieved perlite was then washed with potable water with the pH adjusted to around 6.5. After each experiment, both media was recovered for later analysis. Data Acquisition and Management Submersible pressure transducers and a data lo gger were utilized to monitor and record water level and determination of head data. Water levels inside the media cartridge (inner central PVC pipe) and outside the media cartri dge (in the manometer) we re monitored by using 6.9 Kpa pressure transducers as illustrated in Fi gure 4-1. The pressure transducers were hardwired to the datalogger, which wa s configured to enable data m onitoring with 1 second sampling intervals and was powered for constant operatio n by AC power for continuous data logging and DC backup power. Each sensor utilized in the study was assigned appropriate multipliers and offsets in order to obtain data in the require d depth formats (mm or feet). The flowrate was determined volumetrically using a DLJ Multi-Jet water meter (50.8 mm DLJ 200). Calibration Procedures for Flow and Head Data A potable water supply tank with a capaci ty of 40,000 liters served as a reservoir using a directly metered connection to the watermain. Using this water supply, a submersible electric pump with a peak flow rate of 7 L/s at 4 meters of head, a valved re-cir culation system for flow adjustment, and a calibrated and graduated 200 L tank, the flow meter was calibrated
116 volumetrically across the entire flow rate range. The flow meter was calibrated from 0.19 to 1.51 L/s. A calibration curve was developed for the 5.08 cm ( ) water meter (DLJ Multijet). The volumetric calibration procedure produced a coefficient of determination of R2 = 0.998. The equation was Y = 0.9684 X + 0.6094 where Y = volum etric flow rate (L/min); and X = meter reading flow rate (L/min). Both pressure transducers (Druck CS 420-L Model PDCR 1830-8388) with 6.9 kPa were connected to a CR1000 datalogger (Campbell Scientific, Inc.). The pressure data was monitored real-time through a data acquisition field co mputer connected to CR1000 datalogger. The pressure data was recorded as mV readings whic h were converted to water level and head data through a calibration curve for each pressure tran sducer. Pressure transducer calibration was checked across a range of static water depth measurements at 20oC. These physical depth measurements were paired with mV readings ob tained from each transducer. After the mV and static depth readings were plotted, a linear regression pr ocedure provided the necessary multiplier and offset parameters. The multiplier and offset were used as calibration parameters in the data acquisition program, in order to obtai n data in depth units (mm). The procedure was repeated for each pressure transducer utilized in this study. The multiplier was calculated as M = 1/Sensitivity (mV/V/psig) (1 psi = 6.9 kPa). Multiplier is the slope of the regression line plotting voltage measurement (mv/V) as a function pressure (depth); Sensitivity is defined here as the voltage reading (mv/V) pe r unit pressure (depth). The linear calibrations for each pressure sensor were: Y = 0.973 + 0.022 and Y = 0.965 0.014 where Y = measured depth (mm); X = actual depth (mm). Across the entire measured range of depths for each sensor the coefficients of determination exceeded 0.997 for each linear calibration.
117 PM Filtration Experiment The PM suspension used in these experiments was prepared by using silt-size gradation, Sil-co-Sil 106 (U.S. Silica Company, Berkeley Springs, WV). This PM gradation was produced from high purity silica (99.8% SiO2) which is precision ground to microcrystalline sizes. The gradation of Sil-co-S il 106 has a mean particle size of 16.3 m (d50m) and specific gravity of 2.65. Sil-co-Sil 106 is a typical PM gradation spec ified for filtration testin g (Galicki et. al. 2003). Table 1 summarizes the gradation and physical properties of this PM. The parameters considered in this experimental matrix were se ven flow rates (0.19 L/s, 0.38 L/s, 0.57 L/s, 0.76 L/s, 1.14 L/s, 1.32 L/s and 1.51 L/s) and media type. The filtration geometry, surface area and volume, the number of bed volumes (75 B.V.) tr eated, the influent particle size distribution (PSD), pH, and the temperature were held constant. The experimental matrix for the filtration experiments is summarized in Table 4-2. Accordi ng to the total target volumes for each run, the concentrated stock Sil-co-Sil 106 solution was prepared in the 20 L slurry tank with a mixer and recirculated with the slurry to ensure homogeneity of the influent PSD and slurry concentration. The slurry was then introduced into the delivery line using a pe ristaltic pump at a known flow rate adjusted to ensure that the slurry was dilute d by tap water at a chosen influent flow rate and target influent concentration level (as suspended sediment c oncentration, SSC) at a nominal gravimetric concentration of 200 mg/L. The flow rates ranged from 16.7% to 133.3% of a nominal design flow rate of 1.14 L/s (18 gallons per minute). After each filtration run, media was replaced and fresh media utilized for the next experiment. Composite samples were ge nerated from 10 individual samples taken at equal increments in time during each experiment. Each incremental sample was taken manually and sampled through the full cross section of fl ow was a 2 L sample and all samples for an experiment were composited in separate 20 L ca rboys for influent and effluent. PSDs and SSC
118 were measured after collection and mass bala nce analyses were then performed after backwashing and PM recovery. SSC was determined based on the method of D 3977-97 (ASTM 2000), which uses the entire aqueous-particulate sample volume for subsequent operations and analyses (Gray et al. 2000). A mass balance was conducted as an application of conservation of mass to the analysis of the treatment system. Math ematically, the amount of particle load will be same as the particles leave the system and accumulate in the system. A mass balance error (MBE) analysis was conducted after each run to ensure mass conservation and QA/QC within the range of 10%. Particle size distributions (PSDs) were carried out with laser diffraction analysis (LISST portable, SEQUOIA Scientific, Inc. Bellevue, WA). Filter ripening, clogging and head loss will incr ease as a function of bed volumes treated. Monitoring by pressure sensors was used to quantify head loss increases. Head loss developments with time were calcula ted using the following equation 4-2: 0 1 2H H H Ht (4-2) In this equation Ht = head loss (mm), H0 = initial head loss (head lo ss measured for clean bed, mm), H1 = outside cartridge water level (mm) and H2 = inside cartridge water level (mm). Head loss for the clean media cartridge was modeled using the Ergun equation 4-3. g V v a k V v a g L hm m m m 2 3 2 2 3 21 ) 1 ( 17 4 (4-3) In this expression, H/L = head lo ss per unit depth of bed (mm/mm); = fluid viscosity, (Ns/m2); = fluid density (g/cm3); g = gravitational constant (m3.Kg-1.S-2); m = macro porosity; a/v = media surface area per unit of media volume; V = superficial velocity (L/s); k2= dimensionless constant. Ergun reported a k2 value of 0.48 for crushed porous solids. The Ergun equation is a pore-structure based equation that takes into account th e tortuous nature of the flow
119 patterns through the pore spaces (Wu and Yu 2007). The left hand term of this equation quantifies the head loss due to laminar flow through a packed bed and the right hand term quantifies the head loss due to turbulent flow. Phosphorus Adsorption and Adsorption-filtration Experiments For the phosphorus adsorption experiments two sets of influent conditions were examined. In one set of experiments the influent load co nsisted of phosphate only an d in the other set of experiments, phosphate and Sil-co-Sil 106. The fi rst set of experiments examined phosphorus adsorption without phosphorus partitioning to in fluent suspended PM. The second set of experiments was designed to examine phosphorus adsorptive-filtration un der conditions where adsorption, filtration and partitioning occurred si multaneously. The nominal concentration levels of the influent were 1.0 mg/L as PO4-P for phosphorus and 200 mg/L as SSC for Sil-co-Sil 106 in this study. These levels are representative of the source area watershe d where the radial flow cartridge was tested and later deployed in Baton Rouge, Louisi ana (Dean et al. 2005; Sansalone 2005). For each experiment concentrated slurry stock solutions were prepared first and then transferred into the slurry tank. Flow rates, PM and/or phosphate concen trations of the slurry were adjusted to ensure that the nominal influent concentrations were achieved. Four test flow rates (0.19, 0.38, 0.76, 1.14 L/ s) was used in this portion of the study. 10 individual replicated (A and B) infl uent and effluent incremental samples were collected manually at a constant time interval until 100 be d volumes were treated. Phosphorus and SSC (for adsorptive-filtration) were analyzed immedi ately after collection. Phosphorus was analyzed using Standard Method 4500-P-B, the molybdenum blue method. After each experiment media was sampled from the top, middle and bottom la yer of the media cartridge. Media was aciddigested to conduct a mass balance for phosphorus a nd evaluate the variati on in the mass transfer zone from the top to bottom of the cartridge.
120 Evaluation Metrics The efficiency ratio (ER) is defined in te rms of the partial event mean concentration (PEMC) or EMC of a constituent shown in e quation 4-4 and equation 4-5 (Sansalone and Buchberger 1997; Sansal one et al. 1998). EMC inlet average EMC outlet average EMC inlet average ER (4-4) n i i n i i iV V C EMC1 1 (4-5) In this expression V is volume of flow during period I, C is average concentration associated with period I, and n is total number of measurements taken during event. While effluent concentration, mass and PSD are the appropriate metric for filter effluent, mass removal should also be quantified as a pe rcent removal (PR) for any constituent on an event basis using the inflow and outflow loads. The following equation 4-6 was utilized for PM and phosphorus phases. 100 ) ( ) ( ) ( (%)1 1 1 IN i n i IN i EFF j m j EFF j IN i n i IN iC V C V C V PR (4-6) In this equation Vi-IN and Vj-EFF are the volumes of influent flow and effluent flow during sampling period i and j, Ci-IN and Cj-EFF are the mean concentrations associated with period i and j, and n and m are the total numbers of influe nt and effluent measurements taken during the event, respectively.
121 Results and Discussion SSC Removal Efficiencies A series of full-scale single radial flow car tridge tests were conduc ted with engineered media (AOCM)P and control media (perlite) loaded by Sil-co-Sil 106 gradation. Particulate matter (PM) removal efficiency (measured by SSC removal) for the system with (AOCM)P filled are summarized in Table 4-3, along with surf ace loading rate, median residence time and corresponding operating flow rate s. Results indicate that PM ma ss removal performances for the tested Sil-co-Sil 106 gradation are influenced by hydraulic loading. As the operating flow rate increased, the particle removal efficiencies s howed a decreasing trend, from 86.0% to 71.4% with a relatively constant SSC loading at 200 mg/L on average (6.35, SD). In general, the single radial flow cartridge with filled (AOCM)P showed a good capability in terms of mass removal (> 70% mass removal) for such gradation of particles (Sil-coSil 106) with a d50 value of 16.3 m. It should be noted here that performance on SSC removal is based on Sil-co-Sil 106 gradation, 200 mg/L influent loading and sp ecific geometric configuration a nd dimension for single radial flow cartridge. In other words, these results might not be app licable in the situation that real rainfall-runoff is treated, where influent particle gradation and density a nd runoff flow rate are not constant any more but dynami cally changing within a single event and PM in terms of concentration and mass load varied depending on hydrological character istics of individual event. The total solids concentration expressed as SSC for both influent and effluent are summarized in Table 4-3. Generally, with respect to the relatively constant influent concentration, SSC concentrations of effluent are above 32 mg/L, which was contributed by fine part of the gradation including suspended and co lloidal particles. Treatment run results also indicated that effluent particle concentrations from th e radial flow cartridge increased from 32.3
122 to 57.1 mg/L as flow rate increased from 0.19 to 1. 51 L/s, respectively. It is recognized that, with 200 mg/L SSC, effluent concentration was impr oved by approximately 24.8 mg/L when flow rate was reduced from 1.51 to 0.19 L/s. In additi on, the longer duration of the filtration time or larger bed volume for a given flow rate results in lower effluent concen tration as well as higher removal correspondingly before the cartridge fi lter reaches th e breakthrough. This phenomenon was likely associated with the clogging and head loss occurring throughout th e individual run. It is commonly accepted in the stormwater treatmen t industry that higher lo ading concentrations promote higher removal efficiencies. There is al so evidence (Grizzard et al. 1986) that TSS and other constituent removal efficiencies can be sign ificantly affected by the initial concentrations of the constituent. Laboratory and field data us ing stormwater show that less energy was required to remove over 80% of SSC when its initial concentrations are hi gh (e.g., > 400 mg/L) and difficult to remove even 50% of SSC when its in itial concentrations are low (e.g., < 20 mg/L). Based on the performance data from the te sts, SSC removal was also found to be a function of the surface loading ra te, which is often used as a design parameter for a filter treatment and basically indicates how much flow rate the filter can handle per given area. As shown in Table 4-4, recovered PM mass increased from 526.8 to 790.0 g as the surface loading rate decreased from 189.6 to 24 L/(m2.s) for a given influent mass loading (776.3 5.34 g) except the run at the flow of 0.19 L/s. While th is consistent increasing trend of recovered PM mass from the system is in close accordance with the behavior of PM accumulation in the media as a function of filtration rate. Moreover, recall the performance data shown in Table 4-3, mass reduction can reach over 70% and 80% as long as the surface loading rates drop below 189.6 L/(m2-min) and 72.0 L/(m2-min), respectively. Generally, the higher surface loading rate, the shorter residence time was involved in PM removal.
123 In this study, controlled flow and sampling of inflow and outflow allow for the calculation of a mass balance that helped verify monitoring data and/or measured flow rates. Table 4 4 summarizes the mass balance for each run in term s of SSC. Most values are negative since a certain amount of solid are lost during the backwa shing and recovering processes. Estimations of efficiency of a best management practice (B MP) system can be based on these mass balance calculations coupled with sampling data. In general, the error less than 10% indicates a reasonable mass balance check. However, while th ese results were generally well-behaved, the positive values shown in a couple of runs were unexpected. Particle Size Distribution Figure 4-2 illustrates the PSDs) or gradations (and mass) fo r influent and effluent at constant flows across 16.7 to 133.3 % of design flow rate at a nominal influent concentration of 200-mg/L with Sil-co-Sil 106 tested gradation. Influent PSDs were kept relatively constant for each run. It was clearly demonstr ated that PSDs moved toward the right hand side of the plot as influent particles settled dow n to the bottom of the test chamber and passed through the engineered media filter, indicating that coarse particles tended to be captured by gravitational settling first and finer particles were able to trapped through filtrati on under a given hydraulic loading rate. Influent PSD demonstrated very litt le variation as a function of flow rate, with a d50m that varied within the range of approximately 10m to 20m, while d50m of effluent exhibited significant sm aller values (< 5 m). In the case of stormwater treatment, the gradation will be significantly different, and higher remova ls are expected since coarser particles will predominantly contribute to the total mass of solids. From Figure 4-2, the particle si ze gradations of effluent partic les are observed to tend to lie atop each other while influent flow rates were abov e the design flow rate, especially with longer
124 filtration time and larger filtration volume, indicating that be yond design flow rate, the particle separation by filtration occurring in the radial flow cartridge reached close to a maximum capacity in terms of the fraction of particles that can be filtered, which were predominantly fine particles. The particle size based on mass grad ation with respect to effluent decreased as operating flow rate increased up to 1.51 L/s as long as the unit is loaded by the consistent influent particle gradation and density. An interesting phenomenon observed was that a significant amount of colloidal par ticles was found in the influent or water source and apparently it also exhibited a relatively large amount in number of particles for effluent. Table 4-5 summarizes the goodness-of-fit (GOF) of gamma di stributions for gradat ion curves shown in Figure 4-2. In general, the experimental PSD data fit the modeled data well indicated by the Chisquare values. Head Loss as a Function of Flow Rate Head loss results from increased fluid drag, pore constriction, and in creased interstitial velocities caused by particle deposition. Figure 4-3 illustrates that the head loss change as function of bed volume and opera ting flow rate. It was found th at the two parameters linear model could describe the head loss as a func tion of filtration time with good correlation (R2 > 0.94). These linear curves showed the similar trend or slope which indicated the gradual accumulation of particles trapped by the media filter as bed volume increased. Obviously, the use of test media with median size around 3.56 mm in this study will affect the rate of the head loss development. In addition to that, head loss is spread more evenly through the filter in large particle diameter media. Therefore, capturing part icles with larger diameter media result in less head loss than capturing them with smaller media (Clark, et al. 1992). Head loss development was typically linear with filtration time, and, for suspensions of uniform particle size, it usually
125 was higher than the head loss for suspensions of the smaller particle with mixed sizes (Tobiason, et al. 1993). According to the gradation of Sil-co-Sil 106, it wa s classified as silt si ze fine particles with relatively high corresponding surface area. Head loss is determined by the filters surface area, and small particles like Sil-co-Sil 106 may cause more head loss because of their high surface area per unit volume. From Figure 4-4, filter head loss is found to be directly proportional to velocity in media filters which had new and clean beds before each run based on the Ergun equation. But it may not apply to the ripened filte r beds. In addition, filter head loss also was found to be very dependent on porosity and medi a size, and reduction in porosity and media size causes the head loss increase. Phosphorus Concentration Change Using (AOCM)P The results of total phosphorus concentrations and pH change with increased elapsed time for each individual experimental run across the ra nge of operating flow from 0.19 L/s to 1.14 L/s are presented in Figure 4-5. The initial pH sel ected lies in the range 6.5 to 7.0 and influent exhibited relatively constant pH and TP concentra tions throughout the run. In contrast, effluent pH was significantly lower than the influent pH at the beginning of the test and gradually increased up to the initial value, indicating alum inum coated on the surface of test media will neutralize the alkalin ity of the source water and release ca rbon dioxide as well. The same pattern was not seen in the run at 1.14 L/s flow rate because it was conducted after a long time cleaning while adjusting the flow and wate r level as the first phosphorus run. However, the pH values still stay constant throughout the run. Compared to th e source water used in the tests, stormwater contains relatively high alkalinity so that pH will not be reduced below about 6.0 to 6.5. The effect of pH on phosphorus removal by adsorption is critical due to the chemical reactivity of phosphate through associated su rface ligand exchange.
126 It can be observed that total phosphorus concentrations of e ffluent are greater than 0.8 mg/L at 100% of design flow rate and less than 0.6 mg/L as the fl ow rate decreased to 16.7% of design flow rate, indicating the ch emical adsorption is not independe nt from flow rate. In other words, lower flow rates increase the residence tim e, which is the important factor to determine the kinetics of adsorptio n and desorption. An interesting phenomena observed was that the effluent TP concentrations tended to follow the same pattern as infl uent TP did at the higher flow rates (0.76 L/s and 1.14 L/s) while another di fferent pattern with incline phase followed by decline phase for effluent occurred at the lower flow rates (0.19 L/s and 0.38 L/s) regardless the variation of influent TP concen trations over time. This result infers that lower residence time associated with higher flow rate was not able to provide sufficient chemisorption under the specific condition in this study. C onversely, the chemical adsorpti on became more efficient when flow rates were kept lower. Once the capacity of the media surface reach ed a certain level, effluent TP concentration was expected to increase until breakthrough occurs. Under the same flow rates, total dissolved phosphorus was separated from total phosphorus as required when Sil-co-Sil 106 was added into the slurry fed with phosphorus. The results of TP and TDP concentrations over elapsed time are dem onstrated in Figure 4-6. As fractionation was used in this test, total dissolved phosphorus and total phosphorus concentr ations are seen more distinctly and the final effluent TP concentration tends to be lowe r than those in the test without Sil-co-Sil 106 addition. With the flow rates ranging from 0.19 to 1.14 L/s, the effluent TDP concentrations were around 0.2 mg/L or less as influent TDP concentrations were kept around 0.85 mg/L due to the partitioni ng which occurred between the dissolved phase and particulate phase. Compared to the results in a previous test without Sil-co-Sil 106, TP concentrations dropped to 0.4 mg/L or less at flow rate of 0. 19 L/s and 0.6 mg/L at flow rate above 0.38 L/s
127 with the initial influent concentration at 1.00 mg/L. At this point, if we recall the phenomenon found in PSD results, then it is not likely that significant lo wer total phosphorus reduction in contrast with total dissolved phosphorus reductio n after fractionation wa s contributed to the relative amount of colloidal particles in the water source which could serve as substrate for phosphorus to a certain degree. Head Loss for Phosphorus Run Using (AOCM)P The results of head loss over time across flow rates ranging from 0.19 to 1.14 L/s are presented in Figure 4-7. The initial head loss va lues are 3.66, 6.71, 19.81, 40.84 mm at flow of 0.19, 0.38, 0.76, 1.14 L/s respectively and exhibi t relatively constant values throughout the run using phosphate alone. Therefore, it can be infe rred that the chemical coating on the media surface might not result in head loss developmen t. It was found that the quadratic polynomial model could describe head loss as a func tion of flow rate with good correlation (r2 = 0.999). On the contrary, the particle deposition or pr ecipitation onto the surface of media will lead to clogging and development of head loss as well An increase of head loss over time was found across the flow range from 0.19 to 1.14. Anal ysis of Covariance (ANCOVA) was employed and it was noticed that head loss development at hi gher flow rates (0.76 L/s and 1.14 L/s ) were significantly higher (p < 0.05) than those at lower flow rates (0. 19 L/s and 0.38 L/s). This result reinforces the hypothesis that the head loss develo pment is typically linear as a function of time and higher flow results in bigger head loss. Phosphorus Removal Using (AOCM)P Phosphorus reduction performance by adsorption and filtration was studied as a function of surface loading rate. Total phosphorus removal r eached 57 68 % and 80% for total dissolved phosphorus at the flow rate of 0.19 L/s. Appa rently, phosphorus remova l increased as surface loading rate decreased as shown in Table 46. The significantly higher dissolved phosphorus
128 removal compared to the total phosphorus remova l could be explained by either the presence of colloidal particles or the phosphate precipitation after adso rption. Considering event mean concentration (EMC) of P in urban rainfall -runoff is typically below 2 mg/L, AOCM may demonstrate a reasonably good P adsorpti on capacity for low P concentration. Table 4-7 summarizes the vertical distribu tion of phosphorus removal by adsorption on media filter. We compared three layers (top, mi ddle and bottom) of test media loaded in the cartridge on specific mass of phosphorus tra pped on unit of media expressed as mg PO4 3--P / 10 g media. It appears that more phosphorus was tr apped by the bottom layer of media than the top layer of media which indicated that the distri bution of phosphorus adsorp tion is not even across the cartridge vertically. That also suggested an un-uniformly distri buted flow occurred in the test chamber and cartridge as well. Phosphorus and PM Removal Using Perlite Perlite was used as a comparison with (AOCM)P in this study. Table 4 8 summarizes the total phosphorus concentrations of influent and effluent over filt ration time at three flow rates (0.19, 0.38 and 0.76 L/s) with two different in itial phosphorus loadings, 1.0 mg/L and 0.1 mg/L as PO4 3--P. Generally, breakthrough occurred at the begi nning of the test (< 10 B.V.) and effluent total phosphorus concentrations tend ed to increase back to the infl uent initial con centration level of 1.0 mg/L as PO4 3--P. However, it was noticed that ther e was no significant difference in the total phosphorus concentrations of effluent compared with influe nt when initial loading was 0.1 mg/L as PO4 3--P. This is due to the fact that lo wer phosphorus concentration reduces the competitive ability of particulate bounding. A similar trend was found in another set of runnings with both phosphorus and Sil-co-Sil 106 loading at the concentration of 1.0 mg/L as PO4 3P and 200 mg/L as SSC with same flow rates above shown in Table 4-8. It is not surprising to see that the total phosphorus removals at
129 the three operating flow rates are less than 10% in general, which proved the conventional media may not be efficiently used in phosphorus reduc tion, especially for the dissolved portion when compared with engineered media like (AOCM)P used in previous tests. Solids removals for the three operating flow ra tes shown in Table 4-8 were above 70% and exhibit a reasonable increase as the flow decrea ses. The head profile showed less difference between inside and outside cartridge (from 6.2 to 12.5 mm) across the flow range (from 0.19 to 0.76 L/s) throughout the tests. Conclusions This study examined the filtration and adsorp tion performance through a series of fullscale single radial flow car tridge tests conducted with engineered media (AOCM)P and control media (perlite) of similar media size. In general, these performances of the stormwater radial flow cartridge (457.2 mm in diameter 558.8 mm in height) were strongly dependent on surface loading rate, hydraulic residence time, phase of phosphorus and selection of media. From this study we can conclude the following: As the operating flow rate increased, solid mass removal efficiency for the tested Sil-co-Sil 106 gradation showed a decreasing trend under the specified loading conditions. In the case of stormwater treatment, the gradation will be si gnificantly different and higher removals are expected since coarser particles will predominantly contribute to the tota l mass of solids. Filter head loss was found to be directly proportional to velocity in media filters which had new and clean bed before each run. In addition, filter head loss also was found to be very dependent on porosity and media size, and reduc tion in porosity and media size causes the head loss increase. Lower flow rate increases the residence time, which is the important factor to determine the kinetics of adsorption and desorption. Lower residence time associated with higher flow rate
130 was not able to provide sufficient chemisorp tion under the specific c ondition in this study. Conversely, the chemical adsorption became more efficient when flow rates were kept lower. More phosphorus was trapped by the bottom laye r of the media than the top layer of the media, which indicated that the distribution of phosphorus adsorption is not even across the cartridge vertically. That also suggested an un-uniformly distributed flow occurred in the test chamber and cartridge as well. The total phosphorus removals at three operating flow rates are less than 10% in general which proved the conventional media may not be efficient in phosphorus reduction, especially for the dissolv ed portion when compared with engineered media like (AOCM)P used in previous tests.
131 Table 4-1. Summary of particle size distribution characteristics, granulometric parameters and physical properties for Sil-co-Sil 106 util ized in the single ra dial flow cartridge filtration test. d5 ( m) 0.7Central tendency 50 = 5.9 d16 ( m) 1.6Silt (4 < 50 < 8) d25 ( m) 4.3Skewness Sk = -0.2 d50 ( m) 16.3Coarse skewed (-0.10 to -0.30) d75 ( m) 44.0Kurtosis KG = 0.91 d84 ( m) 72.2Mesokurtic (0.90 1.11) d95 ( m) 118.7Uniformity I = 2.49 dg ( m) 10.8Poorly sorted (2 < I < 4) Hardness (MOHS) 6.1 pH 7.2 Physical properties Reactivity Inert
132 Table 4-2. Experimental matrix for single ra dial flow cartridge filtration, adsorption and adsorption-filtration runs by using (AOCM)P under various operating flow capacity. Filtration run (as SSC) Flow rates gpm (L/s) Operating Flow Capacity Sil-Co-Sil 106 PO4 3P Bed volume (B.V.) 3 (0.19) 16.7% 200 mg/L0 mg/L75 6 (0.38) 33.3% 200 mg/L0 mg/L(1 B.V. = 54.1 L) 9 (0.57) 50.0% 200 mg/L0 mg/LExperiment design 12 (0.76) 66.7% 200 mg/L0 mg/L 18 (1.14) 100.0% 200 mg/L0 mg/L 21 (1.32) 116.7% 200 mg/L0 mg/L 24 (1.51) 133.3% 200 mg/L0 mg/L One way ANOVA (SSC)(flow)(B.V.) (1x7x1) Non-repeated measurements Phosphorus run (as PO4 3P, SSC) Flow rates gpm (L/s) Operating Flow Capacity Sil-Co-Sil 106 PO4 3P Bed volume (B.V.) Adsorption run (P only) 100 3 (0.19) 16.7% 0 mg/L1.00 mg/L(1 B.V. = 54.1 L) 6 (0.38) 33.3% 0 mg/L1.00 mg/LExperiment design 12 (0.76) 66.7% 0 mg/L1.00 mg/L 18 (1.14) 100.0% 0 mg/L1.00 mg/L Adsorption-filtration run (P + SSC) 3 (0.19) 16.7% 200 mg/L1.00 mg/L Two way ANOVA (SSC)(flow)(B.V.) (1x7x1) Non-repeated measurements 6 (0.38) 33.3% 200 mg/L1.00 mg/LOthers 12 (0.76) 66.7% 200 mg/L1.00 mg/LpH ~ (6 7) 18 (1.14) 100.0% 200 mg/L1.00 mg/LTemp ~ (14.4 21.1 oC)
133 Table 4-3. Summary of SSC results for single ra dial flow cartridge filtration run under various operating flow rates and 75 Bed Volume with (AOCM)P. The target influent concentration is 200 mg/L Sil-co-Sil 106 as SSC. Flow rates Surface Loading Rate Median Contact Time Filtration Time Liquid Volume Filtered Influent SSC Effluent SSC SSC Removal gpm (L/s) gpm/ft2 (L/m2-min) 50 (min) (min) (L) [mg/L] [mg/L] (%) 3 (0.19) 0.58 (24.0) 4.95400.04368.4205.032.3 86.0 6 (0.38) 1.17 (48.0) 2.79170.03293.3199.835.4 85.0 9 (0.57) 1.75 (72.0) 1.75117.53634.0189.841.2 80.7 12 (0.76) 2.33 (95.4) 1.1985.03855.4196.243.5 78.4 18 (1.14) 3.50 (142.8) 0.8456.03611.3209.949.7 76.7 21 (1.32) 4.08 (165.6) 0.7148.83715.4200.650.0 75.1 24 (1.51) 4.66 (189.6) 0.6442.53611.3199.757.1 71.4
134 Table 4-4. Summary of particles mass results for single radial flow cartridge filtration run at various operating flow rates and 75 Bed Volume with (AOCM)P. The target influent concentration is 200 mg/L Sil-co-Sil 106 as SSC. Flow rate Surface Loading Rate Median Contact Time Influent Particle Mass Effluent Particle Mass Recovered Particle Mass Mass Balance gpm (L/s) gpm/ft2 (L/m2-min) 50 (min) (g) (g) (g) (%) 3 (0.19) 0.58 (24.0) 4.951008.0141.0790.0 -7.64 6 (0.38) 1.17 (48.0) 2.79777.3116.5716.8 7.20 9 (0.57) 1.75 (72.0) 1.75775.6149.6625.2 -0.10 12 (0.76) 2.33 (95.4) 1.19779.0167.9587.0 -3.09 18 (1.14) 3.50 (142.8) 0.84771.0179.4557.9 -4.37 21 (1.32) 4.08 (165.6) 0.71779.0194.0531.6 -6.85 24 (1.51) 4.66 (189.6) 0.64775.6221.8526.8 -3.48
135 Table 4-5. Summary of Goodness-of-Fit ( GOF) of gamma distribution for (AOCM)P filtration runs. Treatment SSE 2 GOF Influent 0.496 59.8964.043.32< 2 0.05, 29 (42.56) 0.19 L/s 0.501 9.40150.642.69< 2 0.05, 29 (42.56) 0.38 L/s 0.555 7.86210.914.54< 2 0.05, 29 (42.56) 0.57 L/s 0.687 6.21236.005.80< 2 0.05, 29 (42.56) 0.76 L/s 0.686 8.15105.402.99< 2 0.05, 29 (42.56) 1.14 L/s 0.448 28.21656.1110.58< 2 0.05, 29 (42.56) 1.32 L/s 0.537 15.57520.488.92< 2 0.05, 29 (42.56) 1.51 L/s 0.495 24.13562.1811.99< 2 0.05, 29 (42.56)
136 Table 4-6. Summary of test parameters in cluding phosphorus removal for phosphorus runs with (AOCM)P at various target flow rates. The initial total phosphorus target concentration is 1.00 mg/L as PO4 3-P. Flow rate (L/s) Parameters 0.190.380.76 1.14 Volume (L) 460949874756 4813 Time (min) 440220110 73 SLR (L/m2.min) 24.0048.0095.40 142.80 Bed Volume (B.V.) 100100100 10050 (min) 4.652.331.16 0.78 TPP only (%) 57.3242.0633.35 15.41 TDPP+SSC (%) 79.7978.1774.14 73.77 TPP+SSC (%) 68.6550.0045.36 37.70
137 Table 4-7. Non uniform adsorption of phosphorus acr oss single radial flow cartridge vertically for phosphorus runs with and without 200 mg/L Sil-co-Sil 106 loaded under various operating flow rates with (AOCM)P. Flow (L/s) 0.190.380.76 1.14 Depth (mm) P only (mg PO4 3P/ 10g media) 76.2 0.5980.4510.353 0.161 254.0 0.6040.4670.368 0.184 457.2 0.7520.5120.422 0.265 Depth (mm) P + Sil-co-Sil 106 (mg PO4 3P/10g media) 76.2 0.5930.4970.413 0.305 254.0 0.6150.5220.436 0.327 457.2 0.7920.6200.531 0.449
138 Table 4-8. Summary of the total phosphorus an d SSC removal by using perlite under different influent loading a nd flow conditions. Flow rate pH B.V. Phosphorus P SSC SSC (L/s) (-log [H+]) (#) [mg/L] (%) [mg/L] (%) 0.19 6.35 500.14.9--0.38 6.42 500.15.8--0.76 6.58 500.1-1.7--0.19 6.48 551.02.8200 75.4 0.38 6.31 501.08.9200 74.8 0.76 6.52 601.04.9200 70.6 Note: -not available (phosphorus only)
139 Figure 4-1. Section view of the single radial flow car tridge and design info rmation of cartridge and media. The volume of the cartridge indicates the volume occupied by media (AOCM)P filled.
140 Particle diameter ( m) 110100 Percent finer by mass (%) 0 20 40 60 80 100 Influent 0.19 L/s 0.38 L/s 0.57 L/s 0.76 L/s 1.14 L/s 1.32 L/s 1.51 L/s Modeled Figure 4-2. Particle gradations of influent with the target conc entration of 200 mg/L Sil-co-Sil 106 and effluent under 7 operating flow ra tes ranged from 0.19 to1.51 L/s for filtration runs over 75 Bed Volume (B.V.) with (AOCM)P.
141 Figure 3 Head loss increase over bed volumes (1 B.V. = 54.1 L) under different operating flow Figure 4-3. Change of the head loss with bed vo lume(1 B.V. = 54.1 L) for filtration runs with (AOCM)P at 0.38, 0.57, 0.76, 1.14, 1.32 and 1.51 L/s as operating flow rate. The linear lines showed the best fit of the trend. Bed Volume 020406080 Head Loss (mm) 0 20 40 60 80 100 0.38 L/s 0.57 L/s 0.76 L/s 1.32 L/s 1.14 L/s 1.51 L/s
142 Filter head loss (Ergun equation) Flow rate (L/s) 0.00.51.01.52.0 Head loss (mm) 0 10 20 30 40 50 d 50 = 3.56 mm Macro Porosity, m 0.30.40.5 Head loss (mm) 0 10 20 30 40 0.19 L/s 0.38 L/s 0.57 L/s 0.76 L/s 1.14 L/s 1.32 L/s 1.51 L/s m = 0.30 m = 0.35 m = 0.40 m = 0.45 m = 0.50 Media diameter (mm) 0123456 Head loss (mm) 0 40 80 120 (0.76 L/s, 1.59 L/m2-s) m = 0.30 m = 0.35 m = 0.40 m = 0.45 m = 0.50 d50 = 3.56 mm Figure 4-4. Filter head loss change over flow rate, macro porosity and media diameter based on the calculation of Ergun equation.
143 Figure 4-5. Total phosphorus (TP) concentration and pH change for influent and effluent over bed volumes (1 B.V. = 54.1 L) for phosphorus adsorption runs with (AOCM)P at selected flow rates (0.19, 0.38, 0.76 and 1.14 L/s). Bed Volume 020406080100 pH 1 2 3 4 5 6 7 TP [mg/L] 0.6 0.7 0.8 0.9 1.0 1.1 1.2 Bed Volume 020406080100 pH 1 2 3 4 5 6 7 TP [mg/L] 0.6 0.7 0.8 0.9 1.0 1.1 1.2 Bed Volume 020406080100 pH 2 4 6 8 TP [mg/L] 0.4 0.6 0.8 1.0 1.2 1.4 Bed Volume 020406080100 pH 2 4 6 8 TP [mg/L] 0.4 0.6 0.8 1.0 1.2 1.4 Influent pH Effluent pH Influent TP Effluent TP 1.14L/s 0.76 L/s 0.38 L/s 0.19 L/s
144 0.19 L/s Phosphorus [mg/L] 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 0.38 L/s pH 0 2 4 6 0.38 L/s 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.19 L/s pH 0 2 4 6 0.76 L/s 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.14 L/s Bed Volume 020406080100 pH 0 2 4 6 1.14 L/s Bed Volume 020406080100 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.76 L/s pH 0 2 4 6 Phosphorus [mg/L] Phosphorus [mg/L] Phosphorus [mg/L]Influent TDP Effluent TDP Influent pH Effluent pH Influent TP Effluent TP Figure 4-6. Total phosphorus, total dissolved ph osphorus concentrations and pH change of influent and effluent ove r Bed Volume with (AOCM)P.
145 Figure 4-7. Change of the head loss over Bed Volume (1 B.V. = 54.1L) for phosphorus adsorption runs with (AOCM)P at 0.19, 0.38, 0.76 and 1.14 L/s flow. Quadratic polynomial curve fit the relationship between head loss and flow rate. Bed Volume 020406080100 Head Loss (mm) 0 20 40 60 80 100 0.19 L/s 0.38 L/s 0.76 L/s 1.14 L/s Flow Rate (L/s) 0.00.30.60.91.21.5 Head Loss (mm) 0 10 20 30 40 50 Phosphorus Run 100 B.V.999 0 85 28 89 0 36 22 2 R V V h h V
146 CHAPTER 5 RESPONSE OF A SOURCE AREA WATE RSHED AND VOLUMETRIC CLARIFYING FILTER TO RAINFALL-RUNOFF AND ANTHROPOGENIC LOADINGS Introduction The proliferation of urban impervious surf aces has significantly altered the natural hydrologic cycle by reducing the hydrologic components of infiltration, evaporation and storage. This hydro-modification results in increased the volume and peak flow of runoff and reduced lag times promoting more effective transport of pollutant loads generated by anthropogenic urban activities and associated infras tructure (Cristina and Sansalon e 2003; Arnold and Gibbons 1996; Mishra et al. 2004; Hamilton a nd Harrison 1991). Since the passage of the 1972 Clean Water Act and the NPDES Storm Water Ph ase I regulations in the 1980s, there has been a large number of suggested structural BMPs developed to mitig ate the effects of stormwater based largely on volumetric control. Furthermore, water quality st andards are challenged by the stochastic nature of interactions between hydrol ogy and the physical and biochemical processes which might affect the water quality in volumetric controls and urban receiving wa ters (Wong et al. 2002). A relatively recent UOP is the combination of detention and spatially-integral filtration component. This study examines such a volumetri c clarifying filter (VCF) system intended to separate PM and phosphorus while providing so me level of hydrologic control based on the volumetric capacity of the system. The res ponse of the VCF system is dependent on the volumetric and treatment processes promoted w ithin the system, which are dependent on the hydraulics of the inflow and outflow, the geom etrics and hydrodynamics of the system and the filters, as well as the filter media properties. Furthermore, head loss in a filter increases proportionally with the square of the flow rate (Farizoglu et al. 2003). Reddi (1997) also reported that transport of fine particles from urban rainfall-runoff into the filtration system
147 eventually results in clogging, which in turn le ads to head loss increase and adversely affect the hydraulics of the system. In addition to hydrological and PM parameters the removal of chemical constituents transported in the runoff also va ries as a function of pH and re dox conditions due to detention or retention. Speciation and partitioning studies of chemicals in urban rainfall-runoff found that the metal partitioning vary with water chemistry para meters such as pH and redox potential that are coupled and change during detention (Glenn et al. 2001; Pettersson 1999; Wilson et al. 2001). For example, under reducing conditions approximate ly 77% of the soluble iron and over 90% of soluble zinc is in a complex form (Patri ck and Verloo 1998). In the study conducted by Charlatchka and Chamber, lower redox potential enhances the mobility of metals by dissolution of Fe/Mn oxides (Charlatchka and Chamber 2000). However, nitrogen and phosphorus has received less attention for the detention mana gement of UOPs under field conditions, and their speciation behavior is hypothesi zed to be distinct under dete ntion/retention conditions. Retention/detention times generally vary seasonally and seasonality is an important characteristic of the hydro-period. A longer retention time was found to be correlated positively with a higher biodegradation rate (Butler and Davies 2000; Scholz 2004). Stored urban stormwater runoff has the potential to change the pH and redox level as a function of residence time in a BMP, rendering many soluble toxins available from th e storage pool so that they can have an immediate effect with discha rge on receiving water systems, both in situ and potentially downstream (Cooke 1991). Objectives This study focuses on the response between ra infall and runoff for an urban paved source area watershed dominated by traffic loadings, th e corresponding response of the VCF system to direct event-based runoff loadings, and the change in nitrogen and phosphorus speciation
148 through the rainfall and runoff dete ntion process. There are four objectives in this study. The first objective is to quantify the rainfall-runo ff relationship for the source area watershed on an event basis for 19 events. The second objectiv e is to examine the transport behavior classification for SSC as a measure of PM on an ev ent basis. The third objective is to examine the hydraulic response of VCF system including the change of flow, head loss and surface loading rate (SLR) as a function of elapsed time for low, modera te and high intensity events. The final objective is to examine redox-pH coupling as a function of detention time and the coupled nitrogen and phosphorus speciation as a functi on of detention time in the VCF system. Methodology Site and VCF System Configuration The instrumented system receive d runoff directly from Intersta te-10 over City Park Lake in urban Baton Rouge, Louisiana. The drainage system was desi gned to intercept the lateral pavement sheet flow from the concrete-paved watershed consisting of two identical eastbound and westbound catchments, each having a contributing area of 544 m2. The average daily traffic (ADT) for eastbound and westbound I-10 lanes wa s 142,000 vehicles. Rainfall was recorded with a tipping bucket rain gage and data-logger in increments of 0.254 mm (0.01 inch). Mean annual precipitation at the site is 1460 mm/year. Detailed descripti ons of the experimental site can be found elsewhere (Dean et al. 2005; Sansalone et al. 2005). During 2006 the Baton Rouge site was loaded by 59 rainfall-runoff events; of these 19 events were treated and monitored from April through August. Since the VCF was an o ff-line unit, events not tr eated continued to be directly discharged to City Park Lake without flowing through the VCF. The main components of the VCF system are detention and media f iltration cartridges, shown in Figure 5-1. The five cartridges with filled (AOCM)P are housed in a 116.8 cm by 212.2 cm detention vault structure. This structure c ontains the inlet and outlet pipes as well as an
149 internal manifold that delivers treated runoff to the effluent drop box of the system. Five pressure transducers were installed in fi ve different locations from upstr eam to downstream locations; the influent Parshall flume, system vault, cartridge center pipe, effluent box and effluent V-notch system as shown in Figure 5-2. Stormwater runoff was collected through pipi ng from the watershed catch basins and expansion joint and entered the VCF through the influent delivery system which includes a 5.08 cm (2-inch) Parshall flume and a 36.6 cm by 79. 2 cm drop box for sampling. Runoff entering the VCF influent box was diverted by a vertical 15.4 cm PVC pipe that directs flow to the bottom of the vault beneath the cartridges. When the wa ter surface elevation in the vault reaches the operating level (slide gate is in the closed positio n), a float valve will then be triggered and the orifice plate opens gradually. Water flows through the media filter cartridges driven by the changing head and drains into the perforated drai n tubes located in the center of the cartridges and flows to the collector manifold through the flex ible pipes. The manifold is plumbed to a float controlled slide gate that sets the overall operational control of the VCF system to achieve a balance between flow and driving head level. The float is designed to fully open the slide gate as the water level reaches the top of the cartridges. After the storm event has ended, the remaining water is slowly released through e ach cartridge and the slide gate until the vault is drained to the outlet pipes invert level. When stormwater runoff flows recede, the float controlled slide gate will close until the next triggering runoff event. Data Acquisition and Management Submersible pressure transducers were desi gned for monitoring a nd recording of water level or pressure data. The datalogger was used as the real-time data monitoring and collection unit. Two 6.9 kPa pressure transducers were in stalled in the manometer of Parshall flume and Vnotch system for influent and effl uent flow measurement, respectivel y. Head for the inside of the
150 cartridge (inner central PVC pi pe) and outside the cartridge (manometer) were monitored by using two 17.2 kPa pressure transducers as show n in Figure 5-2. A 34.5 kPa pressure sensor was installed on the bottom of effluent drop box fo r water level measurement. The pressure transducers were hard-wired to the datalogger. Each sensor utilized in the study was assigned appropriate multipliers and offsets in order to ob tain data in the required depth formats (ft or mm). The datalogger was configured to enable da ta acquisition at ten second intervals. Each reading was converted to flow rates using a ca librated head-discharge relationship for the 50.8 mm Parshall flume and 60o V-notch weir summarized in equati on 5-1 and 5-2, respectively. The VCF stage-storage equation is prov ided in equation 5-3. The sp atial relationship between these components is illustrated in Figure 5-1. 8619 28 304 62 74 2 tan 2 15 8 Z h g C Qu e (5-1) 6548 18 304 07 31 Z Kh Qu (5-2) Z A Ve Ae = 2.48 m2 (Z < 0.1780 m) Z A Ve 061 0 Ae = 2.13 m2 (0.1780 m < Z < 0.5840 m) Z A Ve 540 0 Ae = 1.31 m2 (0.5840 m < Z< 1.1428 m) (5-3) Z A Ve 397 0 Ae = 2.13 m2 (Z > 1.1428 mm) In these expressions Cd = effective coefficient of disc harge, g = acceleration of gravity [L/T2], = notch angle, h = head [L], u = power i ndex, Z = head [L] above VCF invert (datum) and Q = discharge [L3/T], K = dimensional coefficient, V = storage volume (m3), and Ae = effective surface area of the VCF system (m2). An overall volume balance was examined for each event to ensure the volume conservation and QA/QC based on influent, effluent volume, sa mple volume and storage volume
151 retained by the unit. A volume balance error (VBE) criterion was set at 10% according to equation 5-4. 100 (%) INF s S EFF INFV V V V V VBE (5-4) In this expression VINF, VEFF, VS and Vs represent volume of influe nt, effluent storage and samples, respectively. Gamma Distribution of Hydrograph and Hyetograph Cumulative hydrographs and hyetographs were modeled as gamma distributions. The probability density function (pdf) and cumula tive density function (cdf) of the gamma distribution are described in equation 5-5 and equation 5-6, respectively, where and are the scale factor and shape factor. The optimization of and parameters were estimated by minimizing the sum of squared errors (SSE), ma ximizing the coefficient of determination while ensuring a statistical p-value (p > 0.05) between the modeled dist ribution and experimental data. xe x x f1) ( (5-5) /xx F (5-6) Residence Time and Contact time For each event the time datum was set at the star t of rainfall. On th e watershed the initial rainfall would wet the pavement surface, fill pave ment depression, while also being re-entrained and deflected by traffic, resulti ng in a time lag from initial rainfa ll to initial runoff; this time difference designated as initial pa vement residence time (IPRT). The time series of rainfall and runoff for each event was used to compute pave ment residence time and flow indices. Average pavement residence time (APRT) and IPRT were calculated using the methodology illustrated in
152 Figure 5-3. The rainfall-runoff lag time was dete rmined as the centroidal difference between the hyetograph and corresponding runo ff hydrograph (Sansalone and Bu chberger 1997, Sansalone et al. 1998). The same concept can be applied to determine the event residence time (RT) for runoff in the VCF system and the runoff contact tim e (CT) in the filtration media. While RT is defined as the difference between centroid of time of influent hydrograph to centroid of time of effluent hydrograph, CT was defined as the di fference between centroid of time of storage hydrograph to the centroid of time of effluent hydr ograph. The computation of centroids for the rainfall hyetograph and the runoff hydr ograph are illustrated in Figure 5-3. Runoff Coefficients The runoff coefficient in this study is a volum etric based coefficient ranging from 0 to 1 which represents the fraction of runoff that will result from a given volume of rainfall (Cristina and Sansalone 2003). The equation for the rationa l method coefficient can be expressed as following equation 5-7: n w n nA I Q C (5-7) where Qn represents volumetric flow rate of the n th interval observed at the outlet in [L3/T]; In is rainfall intensity of the n th interval in [L/T]; and Aw is contributing area of the watershed, respectively, in [L2]. Rational runoff coefficients Cn were computed for each time tn using the cumulative volume of rainfall and the cumulative volume of flow observed from time 0 to time tn yielding a trajectory of the coe fficient as a function of elapsed time (Bedient and Huber 1992, Cristina and Sansalone 2003). Hydrological VCF System Model as a Function of Time As shown in the following equations, the time ra te of change of storage dS1/dt is equal to the difference between the influent flow (I) and cart ridge inflow (Q), and the time rate of change
153 of storage dS2/dt is equal to the difference betw een the cartridge flow (Q) and effluent flow (E) as shown in equations 5-8 and 5-9. Therefore, both forward influent routing and backward effluent routing methods can be used to determin e the incoming flow to the cartridges. For given I at time t, storage of influent can be dete rmined based on the stage-storage curve which basically described the volume change in the syst em vault as water level changes. Thus the flow Q represents the difference between I and S1. On the other hand, with known effluent flow at time t, the filled bed volume can be determined based on the water level in the cartridge. The cartridge flow Q can be also determined by a dding the E and S2. The instantaneous residence time Tr and contact time Tc are the durations for a water molecule to pass through the VCF system vault and filter cartridges of the hydrologi cal cycle. As shown in equations 5-10 and 511, it can be calculated by dividing volume of wa ter S1 and S2 in storage by the flow rate Q (cartridge inflow) and E (effl uent flow), respectively. t Q t I dt dS 1 (5-8) t E t Q dt dS 2 (5-9) Q S Tr 1 (5-10) E S Tc 2 (5-11) Detention Water Chemistry Water chemistry parameter including pH and redox potential were monitored to demonstrate the dynamics of water chemistry with in a VCF system on site during the dry period (DP) between actual storm events. To examine the changes in water chemistry in the stormwater
154 BMP unit, nineteen actual storm events were captur ed at the site, allowing sludge to accumulate, and the initial time t0 in this study started at th e end of the last event (09-August). After the end of runoff the unit was left with an un-drained vo lume typical of runoff volume retained in most below grade BMPs. After the end of the last runoff event the unit was isolated from the catchment drainage to simulate a dry period for this study. This study was carried out separately from the treatment testing. Samples were taken at 0 h, 0.5 h, 1 h, 6 h at the end of runoff and 12 hour interval after the event for eight days. Wa ter quality including pH and redox potential were recorded at 1 hour interval. All samples were replicated and pres erved according to the standard methods prior to the analysis. Dissolved fractions were separated using the 0.45 m membrane filter immediately after collection due to the partitioning. Temperature, Redox potential and pH were meas ured in duplicate usi ng a calibrated Orion 5-star variable combination electrode. Total nitrog en (TN) after on-site filtration was determined after digestion (HACH 10022), wher eas dissolved inorganic nitrog en (DIN) incl uding nitrate (NO3-) (HACH 8039), nitrite (NO2-) (HACH 10 019) and ammonia (NH3) (HACH 8038) were determined without digestion. Total phosphorus (TP) and total dissolved phosphorus (TDP) were determined using the persulfate digestion met hod (Standard Methods for examination of water and wastewater 4500-P) and spectro photometer analysis (HACH 8048). Results and Discussion Event Hydrology Summary Event-based hydrological indices including prev ious dry hours (PDH), rainfall depth (P), duration of event (D), rainfall intensity (I), runoff flow (Q) and runoff volume (V) were measured and recorded for total 19 storm even ts occurring between April 21, 2006 and August 09, 2006 as shown in Table 5-1 and Table 5-2. The observed storm varied in duration from 20 minutes to 147 minutes and peak flow rates ranged from 0.3 to 25.3 L/s. In addition, events
155 varied in total runoff volume ranging from 495 L on 07 May 2006 to 48,306 L on 29 April 2006. In general, rainfall depths of those monitored st orm events, which were in the range of 1.02 to 71.37 mm, were representative of the water quality volume (WQV) re quirements for treating runoff from the locality. The volum e balance errors (VBE) were ge nerally in the range of 10% except 06 May event which only generated 495 L runoff at an event mean flow rate of 0.1 L/s, leading to challengeable condition to the detect limit of pressure transducer in the system. A bypassing overflow occurred in 29 April event yiel ded 275 L untreated volum e into the effluent box which accounted to 0.6 % of total runoff vol ume for the event. Additionally, it was found that almost all of the cumulative hydrographs and hyetographs could be modeled using gamma distribution. A total of nineteen rainfall runoff events were characterized by fitting gamma model on rainfall hyetographs a nd runoff hydrographs in th is study. The shape fact ors, scale factors and goodness-of-fit of statistics were summarized in Table 5-3. Time of Concentration and Runoff Coefficient The initial pavement residence time (IPRT) or time of concentration referred here is the time required for rainfall to overcome the factors including pavement surface wetting and depression storage filling as well as the airborne resuspension resulted from traffic wheel suction before runoff was generated. Although the IPRT wa s highest for the lowest intensity event of 06 May, in comparing all 19 storm events IPRT re sults generally varied in the range of 3 to 16 minutes (Table 5-1). In addition, IPRT was also controlled by a complex combination of previous dry hours (PDH) and vehi cle during storm (vds) (Sansalone at al. 2005). Furthermore, the average pavement residence time (APRT) values varied from 0.9 to 13.7 minutes by comparing all events monitored. The highest APRT was observed to correspond to the lowest intensity event of 06 May while the lowest APRT value was associated with the highest intensity event of 29 April.
156 In order to quantify the continuous abstraction associated with traffic, incremental volumebased runoff coefficients for 19 events were calc ulated and are presented in Figure 5-4. Eleven storms were grouped in the top plot and compar ed as the high volume events. The box plot indicated that most events generated more than 4,000 L runoff volume. For storms of this nature, there is an asymptotic approach to a maximum runoff coefficient for each storm that typically fell between 0.6 and 0.8. For the low volume events in which total runoff volumes were less than 4,000 L, the abstraction attributab le to traffic was increased resulting in a lower value for the runoff coefficient that ranged from 0.4 to 0. 6. A study by Cristina and Sansalone (2003) on a small urban catchment in Cincinnati, Ohio reveal ed similar runoff coefficient, ranging between 0.2 and 0.4 for low intensity events and 0.6 to 0.8 for high intensity events. Dry period, Hydrograph and Hyetograph The historical rainfall and dry period data was collected from National Oceanic and Atmospheric Administration ( NOAA) and summarized in Fi gure 5-5. According to NOAA 1971-2006, the monthly precipitation of May, June and July in 2006 was observed to be less than the median value which indicated a drought seas on occurred from April to August 2006 when the overall 19 storm events were monitored in Baton Rouge site. On the other hand, the dry period (DP) and annual DP for Ba ton Rouge (BTR) indicated that the mean DP in last 35 years for Baton Rouge was approximately 4.0 days. The mean DP in 2006 was found to be 10% higher than the mean value which may indicate a possibl e drought year. In addition to that, the previous dry days (PDD) for 19 storm events collected in Baton Rouge was 6.8 day on average which also indicated a drought season. Based on the statistics of flow and volume am ong overall 19 storm events, three events were picked up randomly to represent low flow (LF), medium flow (MF) and high flow (HF) events with event mean flow rate less than 0.8 L/s, 0.8-2.0 L/s and greater
157 than 2.0 L/s, respectively (Figure 5-5). In additi on, a drainage management (DM) event was used to examine the hydrological response of the system under drained condition. The hydrology of the site was described for each of four representa tive events by rainfall hyetographs and runoff hydrographs plotted in Figure 5-6. The elapsed time was initialized to zero at the beginning of rainfall The hydrographs were observed to respond quickly to the onsite fluctuations in rainfall intensity. As a result, these hyetographs provide greater precision than constant rainfall intensity by specifying the precipitation variability over time, allowing for simulation of actual rainfall events for flood risk analysis and hydrographic models calibration. Once rainfall starts, it usually takes longer for an urbanized watershed to reach saturation before runoff is generated when rainfall intensity is low. On the impervious pavement, storm runoff may be mainly contributed by the saturated overland flow where sh eet flow may concentrate. In most cases, the runoff hydrograph increases in magnitude shortly after the start of the rainfall event and reaches a peak after the maximum rainfall intensity has occurred. The sharp hydrograph peaks and steep recession segments indi cate that a large part of water, which has fallen in the catchments directly, enters the coll ection system and the treatment unit as runoff. The peak flow rates exceeded the VCF system designed flow capacity during high intensity event, indicating the likelihood that some bypass occurred during those storm periods. From the hydrograph figures shown for MF a nd HF types of events, neither much peak attenuation nor much volumetric attenuation were observed. With this particular device, the storm was actually run through while the volumet ric capacity of the device was filled when no drainage strategies were adopted. The only attenu ation was observed in the storm was near the beginning of the event where the low flow and low volume occurred thus even a small peak was attenuated at the beginning but when it came tim e for a larger peak, the unit was not able to
158 attenuate that peak because no additional volumetri c capacity in the system was left. Therefore it is not surprising to see that ther e is no significant lag between the peak of influent and the peak of effluent especially for those storm events wi th relatively high intensity Furthermore, a slight transition in time was also observed between th e falling limb of influent hydrograph and falling limb of effluent hydrograph for MF event. Unlike the situation above, the system was started in a drained condition for DM event, and a signi ficant attenuation was observed between the influent and effluent flow. Sim ilar response was found in the LF event. But in most cases, the system was not very effective in terms of hydrological control because the management strategy with no volumetric attenuation was chosen for the treatment. Hydrological Response of VCF System Known head is critical whenever filters were designed and used as the BMP unit since because in a treatment system it impacts hydrauli cs. According to Darcys law, the hydraulic gradient for a given flow path determine the dr iven head across the porous media. As with any saturated filter medium with radial inflow and out flow from the bottom of the cartridge, pressure varies across the depth of the cartridge. The exis tence of a pressure gradient as a function of cartridge depth appears to be isobaric distributions with depth. Basically, as the system filled up from its initial level, what was happening is just that the increasing head through the system was measured. Once it reached to the certain level whic h was close to the top of the cartridges, some drainage would occur and head started back down due to the outflow from the system. And then a declining head was observed after the storm passes slowly over time. One of the biggest concerns here is the h ead loss which comes from the water being filtered through the media. Figure 5-6 also showed the difference of head profiles in millimeters between inside and outside of the cartridges wh ich sit 58.42 cm above the floor of the vault. Usually a sudden spike was found in head loss cau sed by some resistance to flow the porous
159 system initially because the media started dry a nd it has already some suspended solids trapped on from previous storm. Therefore the minut e the water level started to move up on the cartridges, the initial wetting has to be got over and that takes certain amount of energy before the radial freatic type of flow was started. In many cases the intensity of flow exceeded the hydraulic conductivity of the media so a momentar y was built up for head while the high flow rate was coming in. Therefore after the head dr opped back down and a peak head loss across the media filters might occur during the peak of the storm. Then throughout the storm the head level dropped once the hydrograph dissipated. The micro porosity in the media during the storm may not be very effective because the convective fl uxes cannot access those fine pores in a short amount of time while the water is in the filter cartridges. Ther efore, only the macro porosity was preferentially available for the storm. It is only when flows are low so a diffusive flux into the micro pores of the media may occur which is im portant because the diffu sive flux over longer period of time at low flows is important for the adsorption of phosphorus or metals if the system is designed accordingly. As depicted in Figure 5-7, the variability of head loss on event ba sis ranked in either chronological order or the orde r of increasing mean flow rate illustrated the difficulty in obtaining consistent trend with stoc hastic nature of the individual event in terms of varied flow, PDH and PM loading as compared to the steady flow and constant PM loading conditions in previous studies conducted on single radial flow cartridge tests. A dditionally, head loss and flow rate for overall 19 events were found to fo llow exponential distri bution and log-normal distribution, respectively. Surface Loading Rate, Residen ce Time and Contact Time Surface loading rate (SLR), contact time (C T) and residence time (RT) change as a function of elapsed time in normalized scale we re summarized in Figure 5-8. SLR represented
160 the amount of flow that can be handled by un it area of cartridges corr esponding to the flow. Although the response varied as a re sult of cartridge submergence, these plots demonstrated that the pattern of surface loading rate was driven by effluent flow as shown in Figure 5-8. In addition, results showed that maximum SLR of HF event was found to reach approximately 500 L/min-m2, while those were less than 100 and 50 L/min-m2 for MF and LF events. Therefore, low intensity and low flow events tended to have a lower SLR compared to high intensity and high flow events, indicating a better performance with respect to filtrati on on average. It should be noted here that SLR would change in proporti onal to flow as the water level reached above the top of the cartridges. Figure 5-8 also demonstrated the temporal relationship between instantaneous residence time (RT) and instantaneous contact time (CT). While contact time represented the time for the runoff filter through cartridges, residence time represented the time for runoff flow through the unit or VCF system. Thus, residence time was exp ected to be higher than contact time as shown in Figure 5-8. Results indicated that both RT and CT change inve rsely proportional to the inflow hydrographs for cartridges which were determined by forward influent routing and backward effluent routing. The difference of cartridge flow determination between two routing methods is not significant according to the re lative percentage difference (RPD) value given in Figure 5-9. Although there existed a reasonabl e agreement between two routi ng methods, the forward inflow routing method was most likely to overestimate or underestimate the flow to the cartridges since the relatively big water surface ar ea compared to the changing de pth of water level in the unit might accounted for the larger variation. On the other hand, the high surface loading rate usually resulted in the low residence time and contact time, indicating the lower removals due to the
161 shorter time for coarser particles to settle and less time for dissolved pol lutant to react with surface of media. Detention Test As illustrated in Figure 5-10, after 15 hour s of acclimation period where redox potential remains relatively constant in the runoff remaining in the unit, a rapid and significant decrease in the redox potential was observed through 48 hours of retention residen ce time. Concomitantly, consistent increase of ammoni a concentration was identified while nitrate concentration exhibited opposite (decreasing) pa ttern as a function of residen ce time. Results also indicated that a retention residence time had a minor in fluence on partitioning of phosphorus that was primarily particulate-bound. Following initial 48 hours of residence time decreasing pattern of redox potential was stabilized at a lower rate and redox reached a plateau of around -140 mV at 120 hours of residence time in the unit. For the ba lance of the 8 day monitoring period the redox remained at approximately -140 mV. These redox results demonstrated that electrochemical condition inside of a BMP detaining runoff and particulate matter may be converted to an anoxic/anaerobic condition and acidic pH within a few days of de tention time (48 hours in the present case study). This result is in agreem ent with the common understanding that rainfallrunoff collection system or treatment operati on/process units where extended periods of residence time (hours to days) often occur undergo a transformation to anoxic/anaerobic conditions. The sludge trapped by BM Ps contain a wide array of or ganic and inorganic pollutants such as metals, sulfur and nutrients (N, P). The presence of these constituents indicate that coupled changes in redox and pH due to pr ogression down the redox ladder to anaerobic conditions causes significant alte ration of mobility and species of N, P and metals. The potential mobilization of these pollutants fr om sludge to the water column in the BMP results in a scour
162 or first-flush of these more mobile soluble po llutant species in addition to the scour illustrated for solids. Speciation of Nitroge n and Phosphorus In order to assess the specia tion of the nitrogen and phosphorus as a function of detention time, site mean concentration (S MC) of nitrate and phosphate were utilized in this study and the species distribution of N and P as a function of pH for rainfall, runoff under aerobic condition (t = 0 hr) and anaerobic condition (t = 48 hr) was plotted in Figure 5-11. Across the typical range of rainfall observed in Baton Rouge site, both nitrate and ammonium were found to be constant in this study. Examination of speciation profiles for nitrogen and phosphorus revealed that pH did not significantly influence th e composition percentages in the rainfall matrix. However, the percentage distribution may va ry among events and sites. In urban rainfall runoff, nitrogen speciation was found to be influenced primarily by pH and redox potential. MINTEQ modeling results de monstrated that the dominant species of nitrogen was nitrate. However, it should be noted that nitr ate decreased as redox potential reduced to anaerobic level after 48 hours retention in the unit. As a consequence, ammonia level increased over retention time and also increase d significantly at pH above 8 which may occur for runoff generated from poorly buffered pavement Similar trend for nitrite was observed as a function of retention time but no significant ch ange was found over the pH typically ranged from 6 to 10 for runoff. Retention time appear to have little influence on phosphorus speciation. However, partitioning of phosphorus from sediment ation bay to the water column in the system vault might be facilitated under anaerobic condition. Conclusions Rainfall depths of 19 monitored storm events which were in the range of 1.02 to 71.37 mm, were representative of the water quality volume requirements for treating runoff from the
163 locality. The hydrograph was observed to respond quickly to the onsite and fluctuations in rainfall intensity. The hyetographs can provide greater precision th an constant rainfall intensity by specifying the precipitation va riability over time and allow for simulation of actual rainfall events for flood risk analysis and hydrographic models calibration. The lag time from a burst of rainfall to the runoff starting was 8 minutes in general. For this site, the amount of rainfall needed to satisfy initial abstra ction prior to the generation of runoff is around 0.43 mm. For high volume storm ev ents (> 4,000 L), there was an asymptotic approach to a maximum runoff coef ficient for each storm that typi cally fell between 0.6 and 0.8. For the low volume events in which total runoff volumes were less than 4,000 L, a lower value for the runoff coefficient was found within the range from 0.4 to 0.6. During many events, a sudden spike was found in head loss caused by some resistance to flow the porous system initially because the media starts as dry and it has already some suspended solids trapped on from previous sto rm. The micro porosity in the media during the storm may not be very effective because the conv ective fluxes cannot access those fine pores in a short amount of time while the water is in the f ilter cartridges. In contrast, only macro porosity was available for storm. In addition, high surf ace loading rate usually resulted in the low residence time and contact time, i ndicating the lower removals due to the shorter time for coarser particles to settle and less time for dissolved pollutant to react with surface of media. The redox change as a function of residen ce time was in agreement with the common understanding that rainfall-runoff collection system or treatment operation/process units where extended periods of residence time (hours to days) often occur undergo a transformation to anoxic/anaerobic conditions. Sludge pumps are generally operated on a 15 to 45 minute frequency in many wastewater clarifiers to ensu re stable clarification and reduce sludge scour.
164 By extension, frequent management is required for BMPs such as the volumetric clarifying filtration system to address misbehavior such as scour and re-partitioning. Ultimately, the hydrologic restoration and pollutant source control is required gi ven that this level of BMP management to ensure intended BMP perf ormance is simply not sustainable.
165Table 5-1. Hydrological indices for 19 st orm events treated by the VCF system fo r a 1088 m2 watershed in Baton Rouge, LA. PDH Drain Dinf Deff Ps IPRT APRT vds Hmax H50 ninf neff Event (hr) (min) (min) (min) (mm) (min) (min) (#) (mm) (mm) (#) (#) 21 April 927 49 47554.17.01.1 4429195.620.81314 29 April 84 147 15617071.44.20.9 8938132.140.61818 06 May 157 69 45361.039.013.7 542843.217.81113 07 May 21 41 33506.416.32.3 4482111.826.91112 27 May 482 22 314220.127.116.11 279556.925.11012 28 May 23 27 28403.13.01.5 240455.920.11012 16 June 451 76 707114.29.01.2 5084152.444.51113 19 June 79 36 32418.104.22.168 366861.512.71113 04 July 352 29 46522.214.171.124 4233176.35.61313 05July 25 48 64835.65.01.4 613446.85.71414 09 July 69 27 28381.84.02.8 274922.81.61011 10 July 16 64 7210030.53.03.0 5512154.854.91010 14 July 89 58 748126.96.36.199 5952104.611.01011 16 July 45 56 699188.8.131.52 652874.727.01011 18 July 26 20 30492.09.05.9 360752.111.31011 04 August 191 76 114135184.108.40.206 8475154.452.11413 05 August 19 50 65829.16.04.7 459695.811.81112 07 August 25 23 415220.127.116.11 3906136.411.21013 09 August 23 68 849918.104.22.168 6104174.148.71312 Mean 163.4 51.9 59.472.712.98.03.9 5001.3105.423.71213 Median 69.0 49.0 47.055.05.66.03.4 4596.0104.620.11112 SD 236.3 29.7 32.935.622.214.171.124 1774.953.416.8---RPD (%) 144.6 57.2 55.448.9146.1101.581.5 35.550.771.0---Note: PDH, Drain, Dinf, Deff, Ps, IPRT, APRT, vds, Hmax, H50, ninf and neff represent previous dry hours, duration of rainfall, duration of runoff, duration of effluent, rainfall depth, initial pavement residence time, averag e pavement residence time, vehicle during storm, maximum head loss, mean head loss, number of infl uent and number effluent samples, respectively.
166Table 5-2. Hydrological indices for 19 storm events treated by the VCF system on a 1088 m2 watershed in Baton Rouge, LA. Imax Imin Iavg Qinf-max Qinf-avg Qeff-max Qeff-avg Vin Vo+s VBE SSCinf SSCeff Event (hr) (mm/hr) (mm/hr) (L/s) (L /s) (L/s) (L/s) (L) (L) (%) (g) (g) 21 April 91.4 1.1 5.013.31.20.60.2 292728662.14161.7119.7 29 April 259.1 2.2 29.025.35.121.54.2 48306435629.810466.21678.7 06 May 1.0 0.5 0.90.30.10.20.1 49542214.7172.213.0 07 May 61.0 1.7 126.96.36.199.41.1 385235777.11630.1132.9 27 May 30.5 1.7 188.8.131.52.41.0 262824795.71219.193.5 28 May 30.5 5.1 184.108.40.206.41.2 209619377.3320.339.7 16 June 30.5 1.7 220.127.116.11.72.1 993892756.75363.6401.3 19 June 15.2 1.3 5.14.31.03.20.6 18161910-5.2541.440.1 04 July 30.5 2.2 18.104.22.168.80.8 27792907-4.6831.8100.8 05July 30.5 1.7 7.07.91.06.00.8 38383934-2.51065.1138.7 09 July 15.2 1.1 4.01.60.30.40.2 6746257.3204.731.8 10 July 76.2 1.3 28.6154.315.24.5 25189245312.69869.2962.8 14 July 30.5 1.3 22.214.171.124.60.6 330432780.81398.4126.8 16 July 15.2 1.7 126.96.36.199.40.2 9458579.3246.724.4 18 July 30.5 3.8 188.8.131.52.70.2 104710331.3276.121.9 04 August 198.1 5.1 41.320.34.521.46.0 36990352844.6150251384.3 05 August 61.0 1.7 11.09.01.37.01.4 64216649-3.51080.1156.4 07 August 76.2 3.8 184.108.40.2060.41.6 602257764.189080.6 09 August 91.4 5.1 21.714.82.015.01.9 12502114388.52550.2383.3 Mean 61.8 2.3 220.127.116.11.51.5 9040.58544.24.03016.4312.1 Median 30.5 1.7 18.104.22.168.41.0 3304.03278.04.61080.1119.7 SD 65.3 1.5 10.96.61.47.01.6 13312.412300.35.44235.4485.4 RPD (%) 105.6 63.7 87.371.181.692.9109.1 147.3144.0134.0140.4155.5 Note: Imax, Imin, Iavg, Qinf-max, Qinf-avg, Qeff-max, Qeff-avg, Vin, Vo+s, VB, SSCinf, and SSCeff represent maximum rainfall intensity, minimum rainfall intensity, average rainfall intensity, maximum influent fl ow rate, average influent flow rate, maximum effluent flow r ate, average effluent flow rate, runoff volume, outflow and storage volume, volume balance, SSC of influent a nd effluent, respective ly.
167Table 5-3. Gamma model parameters of hydrograph and hyetograph, RT and CT for 19 storm events treated by VCF system. Hyetograph Inflow Hydrograph Outflow Hydrograph RT CT Event SSE GOF SSE GOF SSE GOF (min.) (min.) 21 April 2.1 3.6 0.09 p > 0.053.21.50.01p > 0.05 3.86.70.01p > 0.0511.14.7 29 April 0.6 72.7 0.18 p > 0.051.055.00.02p > 0.05 1.255.40.02p > 0.051.50.8 06 May 5.0 8.7 0.02 p > 0.0511.53.80.02p < 0.05 10.24.10.02p > 0.0520.711.4 07 May 44.1 0.5 0.02 p > 0.0546.70.50.02p > 0.05 27.41.10.03p > 0.055.52.1 27 May 17.7 0.6 0.01 p > 0.057.71.20.01p > 0.05 8.31.80.03p > 0.056.22.6 28 May 1.4 6.6 0.02 p > 0.057.30.90.01p > 0.05 5.02.50.01p > 0.056.72.4 16 June 2.5 13.8 0.02 p > 0.057.02.40.02p > 0.05 7.22.80.02p > 0.052.91.0 19 June 4.4 3.3 0.01 p > 0.056.91.70.01p > 0.05 6.92.80.02p > 0.057.54.1 04 July 2.5 4.6 0.03 p > 0.053.24.40.05p > 0.05 12.42.50.01p > 0.0517.38.5 05July 17.7 1.5 0.08 p > 0.0523.61.30.03p > 0.05 19.41.80.03p > 0.056.71.9 09 July 4.9 1.6 0.09 p > 0.0519.71.10.04p > 0.05 4.94.40.01p > 0.0516.59.4 10 July 3.1 8.0 0.04 p > 0.053.67.60.01p > 0.05 4.17.30.01p > 0.051.20.6 14 July 1.8 16.2 0.16 p > 0.053.59.40.07p > 0.05 18.32.40.03p > 0.057.64.3 16 July 4.5 1.3 0.01 p > 0.052.25.90.11p > 0.05 2.219.40.02p > 0.0510.86.4 18 July 3.8 3.7 0.07 p > 0.052.216.40.12p > 0.05 9.33.30.01p > 0.0515.89.2 04 August 6.5 5.4 0.03 p > 0.055.96.50.01p > 0.05 7.15.70.01p > 0.053.71.8 05 August 5.5 4.1 0.06 p > 0.056.83.90.01p > 0.05 8.43.60.01p > 0.055.72.1 07 August 8.8 1.3 0.01 p > 0.059.91.40.01p > 0.05 12.41.40.02p > 0.054.71.7 09 August 4.9 2.5 0.04 p > 0.0510.21.60.04p > 0.05 16.0 1.30.01p > 0.053.61.2 Mean 7.5 8.4 0.1 --22.214.171.124-126.96.36.199--7.63.6 Median 4.5 3.7 0.0 --188.8.131.52-184.108.40.206--6.72.4 SD 10.1 16.1 0.0 --10.712.30.0-6.712.40.0--5.33.2 RPD (%) 134 191 95 --111185102-6818146--7088 Note: SSE and GOF represent scale factor, shap e factor, sum of square of error and goodness-of-fit of gamma model. RT and CT represent the event residence time and event contact time in the VCF system.
168 Figure 5-1. Plan view of experimental site a nd Volumetric Clarifying Filter (VCF) system, CB represents catch basin and 304.8 in the e quation represents unit conversion (mm to ft) factor. Data lo gg e r 5.1 cm Parshall Flume Watershed PCC pavement 1088 m2 (2 x 544 m2) 2% surface slope ADT=142,000 (east and west) Tee 10.2 cm PVC pipe from east CB @ 10 % slope Dro p box 6 % 31 cm sloped open PVC trough under expansion joint 10.2 cm PVC pipe from west CB @ 10 % slope 15 cm PVC pipe @ 6 % slope 212.1 cm 45.7 cm Effluent 1 2 Influent 9 3 4 5 6 7 8 10 11 116.8 cm 1. Influent box 2. Radial flow cartridge 3. Baffle 4. Vault drainage pipe 5. Float valve 6. Effluent drop box 7 Orifice 8 Effluent pipe 9 Influent delivery pipe 10. Effluent V-notch weir 11. Effluent drainage pipe 5.1 cm Parshall flume Discharge (L/s) 0 2 4 6 8 10 6548 18 304 07 31 X Y60 o V-notch weir Stage (mm) 0.00.10.20.30.40.5 Discharge (L/s) 0 2 4 6 8 10 0306090120150 8619 28 304 62 74 X Y VCF Stage-Storage 030060090012 0 Storage (L) 0 500 1000 1500 2000 2500 V overflow = 2180 L Stage (mm)
169 Figure 5-2. Location of the pressure transducers installed in the Volumetric Clarifying Filter (VCF) system. d50, k and m represent the median size, hydraulic conductivity and macro porosity of the media (AOCM)P used in this study. 393.7 mm H5 184 mm H4 177.8m 558.8mm H3 H2 H1 Data logger Parshall flume System vault Cartridge center draina g e pipe Effluent box V-notch weir
170 Figure 5-3. Methodology for determining average pavement reside nce time (APRT), initial pavement residence time (IPRT), event residence time and contact time. 1 2 Rain be g ins i1 i2 Rainfall intensity + 0 0 IPRT I1 I2 + Time r1 r2 c1 c2 + + + Influent QI Storage QS Effluent QE S1 S2 E1 E2 k k kA t A j h j Ij jAPRT n APRT APRTj nj E j Ir nj E j Sc iI S E distance between centroi d s kth interval of hydrograph or h y eto g ra p h k j th h y dro g ra p h, h y eto g ra p h j ela p sed time t area of h y dro g ra p h, h y eto g ra p hA area centroi d residence time contact timer c i, I, S and E re p resent rainfall, influent, stora g e and effluent, res p ectivel y watershed filtration detention rainfall h y eto g ra p hi VCF VCF VCF
171 High volume events 050100150200 Runoff Coefficient C 0.0 0.2 0.4 0.6 0.8 1.0 16 June 29 April 27 May 28 May 04 July 05 July 10 July 04 August 05 August 07 August 09 August Elapsed Time (min) 050100150200 Runoff Coefficient C 0.0 0.2 0.4 0.6 0.8 1.0 Low volume events Runoff Volume (L) 0200004000060000 Runoff Volume (L) 0200040006000 > 4000 L < 4000 LPrevious data Dean et al. 2005 Cristina and Sansalone 2003 Dean et al. 2005 Cristina and Sansalone 2003 Previous data 19 June 09 July 14 July 16 July 18 July 21 April 06 May 07 May Figure 5-4. Incremental volumetric runoff coeffi cients plots showing ta ngential approach to maximum runoff coefficient value as func tion of elapsed time for both high volume and low volume storm events.
172 NOAA 1971-2006 Monthly precipitation (mm) 0 100 200 300 400 500 600 700 April May June July August 2006 data Volume (L) 1e+2 1e+3 1e+4 1e+5 Flow rate (L/s) 0.1 1 10 18 July 14 July 04-August 04 July (LF) (MF) (HF) NOAA 1971 2006 NOAA 1971-2006 NOAA 2006 BTR DP Annual BTR DP BTR DP 4.4 n = 2218 n = 35 n = 59 BTR Storms 2006Dry period (day) 0510152025 n = 19 BTR PDD = 4.0 S.D. = 3.71 = 4.0 S.D. = 0.75 = 4.4 S.D. = 4.00 = 6.8 S.D. = 9.85 Figure 5-5. Historical rainfall and dry period data for Baton Rouge site and representative events among 19 storm events monitored based on flow and volume statistics.
173 I max = 30.5 mm/hr Influent flow Rainfall intensity Effluent flow I ave = 4.9 mm/hr 0.0 0.5 1.0 0.5 0.0 I max = 259.8 mm/hr I ave = 30.2 mm/hrNormalized I, Q0.0 0.5 1.0 0.5 0.0 0.00.20.40.60.81.0 0.00.20.40.60.81.0 Normalized I, Q0.0 0.5 1.0 0.5 0.0 Normalized Q0.0 0.5 1.0 0.5 0.0LF event HF event Inf-Q ave = 5.08 L/s Inf-Q max = 25.3 L/s Eff-Q ave = 4.15 L/s Eff-Q max = 21.5 L/s I max = 30.5 mm/hr Inf-Q ave = 0.79 L/s Inf-Q max = 3.80 L/s I ave = 7.3 mm/hr Eff-Q ave = 0.58 L/s Eff-Q max = 3.62 L/s MF event I max = 30.5 mm/hr Inf-Q ave = 1.15 L/s Inf-Q max = 6.20 L/s I ave = 7.9 mm/hr Eff-Q ave = 0.78 L/s Eff-Q max = 3.79 L/s DM event LF event -low flow event HF event -high flow event MF event -moderate flow event DM event -drainage management event Inf-Q ave = 0.31 L/s Normalized time Normalized I, Q Eff-Q ave = 0.17 L/s Inf-Q max = 3.30 L/s Eff-Q max = 0.71 L/s Flow rate (L/s) 0.0 7.5 15.0 22.5 30.0 Head loss (mm) 0 50 100 150 200 Filter Head Loss Flow rate (L/s) 0.0 7.5 15.0 22.5 30.0 Head loss (mm) 0 50 100 150 200 Flow, Q (L/s) 0.0 7.5 15.0 22.5 30.0 Head loss (mm) 0 50 100 150 200 LF event DM event MF event T D = 69 min H ave = 13.2 mm H max = 52.1 mm T D = 79 min H ave = 10.0 mm H max = 104.6 mm T D = 62 min H ave = 5.6 mm H max = 176.5 mm PDH = 26 min PDH = 89 min PDH = 352 min Normalized time Flow rate (L/s) 0.0 7.5 15.0 22.5 30.0 Head loss (mm) 0 50 100 150 200 T D = 150 min H ave = 52.1 mm H max = 154.4 mm PDH = 191 min HF event (04 August) (04 July) (14 July) (18 July) (04 August) (04 July) (14 July) (18 July) Figure 5-6. Hyetographs, hydrographs and head loss as a function of normalized elapsed time for four events which represent low, medi um and high flow events and event with fully drained condition treated by VCF system.
174 Flow rate (L/min) 0.01 0.1 1 10 100 1000 Head loss (mm) 0.1 1 10 100 1000 06-May 18-July 09-July 16-July 14-July 05-July 19-June 04-July 05-August 21-April 07-August 28-May 27-May 16-June 09-August 07-May 10-July 04-August 29-April Flow rate Head loss Log-normal distribution Flow rate (L/min) 01020304050 Frequency 0 200 400 600 800 n = 7217 Head loss (mm) 010203040 0 300 600 900 1200 n = 7217 Exponential distributionbXae Y 9 93R2 = 0.83 b = 0.85 a = 936.7 2) 32 1 / ln( 5 0 b Xae YR2 = 0.93 b = 0.99 a = 659Frequency Flow rate (L/min) 040080012001600 Head loss (mm) 0 50 100 150 200 250 05e+41e+52e+52e+5 Head loss (mm) 0 50 100 150 200 250 Volume (L) p > 0.05 p > 0.05 Figure 5-7. The relationship between flow rate and head loss for a volumetric clarifying filter system for 19 storm events.
175 Normalized time Flow rate (L/s) 0 6 12 18 24 30 0 100 200 300 400 0 200 400 600 800 1000 Normalized time Flow, Q (L/s) 0 6 12 18 24 30 0 100 200 300 400 Cartridge Submergence (mm) 0 200 400 600 800 1000 Flow, Q (L/s) 0 6 12 18 24 30 Surface Loading Rate (L/min*m 2 ) 0 100 200 300 400 0 200 400 600 800 1000 Influent Flow Surface loading rate Effluent Flow Cartridge depth Cartridge height 558.8 mm Cartridge height 558.8 mm Flow rate (L/s) 0 6 12 18 24 30 Surface Loading Rate (L/min*m 2 ) 0 100 200 300 400 500 Cartridge submergence (mm) 0 200 400 600 800 1000 Cartridge height 558.8 mm Cartridge height 558.8 mm LF event HF event DM event MF event LF event -low flow event HF event -high flow event MF event -medium flow event DM event -drainage management event 0.00.20.40.60.81.0 Normalized time Flow rate (L/s) 0 6 12 18 24 30 RT and CT (min) 0.1 1 10 100 1000 Cartridge inflow (Influent Based) Cartridge inflow (Effluent Based) Flow rate (L/s) 0 6 12 18 24 30 RT and CT (min) 0.1 1 10 100 1000 Flow rate (L/s) 0 6 12 18 24 30 RT and CT (min) 0.1 1 10 100 1000 Flow rate (L/s) 0 6 12 18 24 30 RT and CT (min) 0.1 1 10 100 1000 LF event HF event DM event MF event 0.00.20.40.60.81.0 Contact time Residence time (18 July) (14 July) (04 August) (04 July) (18 July) (14 July) (04 August) (04 July) Figure 5-8. Surface loading rate residence time and contact ti me change as a function of normalized elapsed time of the four storm events which represent low, medium and high flow events and event with fully drained condition trea ted by VCF system.
176 Inflow for cartridge based on forward inflow routing and backward outflow routing Backward outflow r outing flow (L/s) 051015202530 Forward inflow routing flow (L/s) 0 5 10 15 20 25 30 RPD = 7.6% n = 7217 Figure 5-9. Model comparison between inflow r outing and outflow routing for incoming flow rate to volumetric filter cartridges.
177 Redox (mv) -200 -100 0 100 200 300 pH 7.0 7.2 7.4 7.6 7.8 8.0 Redox pH Time 0 represents the cessation of runoff from the 09 August 2006 event NH 3 -N, NO 2 -N [mg/L] 0.0 0.3 0.6 0.9 1.2 NO 3 -N [mg/L] 0 1 2 3 4 Ammonia NH 3 Nitrate NO 3 Nitrite NO 2 Detention residence time (hour) 050100150200 TDP [mg/L] 0.1 0.2 0.3 0.4 TP [mg/L] 0.0 0.1 0.2 0.3 0.4 TP TDP Figure 5-10. Redox potential, pH, nitrogen and phos phorus species change as a function of detention residence time.
178 09 August Rainfall pH 3456 Species distribution (%) 0 20 40 60 80 100 NO3 NH4 + Runoff (t = 0 hr) Species distribution (%) 0 20 40 60 80 100 NO3 NH4 + 09 August Rainfall 3456 NH3(aq) Runoff (t = 48 hr) Species distribution (%) 0 20 40 60 80 100 NO3 -NH4 + NH3(aq) NO2 H2PO4 Runoff (t = 0 hr) CaPO4 HPO4 2CaHPO4(aq) H2PO4 Runoff (t = 48 hr) CaPO4 HPO4 2CaHPO4 (aq) H2PO4 -H3PO4HPO4 2Runoff (t = 96 hr) 678910 Species distribution (%) 0 20 40 60 80 100 NO3 -NH4 + pH NH3(aq) NO2 678910 H2PO4 Runoff (t = 96 hr) CaPO4 HPO4 2CaHPO4(aq) pH Temp. = 28.6 oC Redox = 243.5 mv Temp. = 28.9 oC Redox = 302.7 mv Temp. = 28.5 oC Redox = -27.6 mv Temp. = 29.6 oC Redox = -97.6 mv [N] = 3.6 x 10-5 M [P] = 1.4 x 10-7 M [N] = 6.1 x 10-5 M[P] = 1.2 x 10 -6 M Figure 5-11. Species distribution of nitrogen and phosphorus as a function of pH for rainfall, runoff at time 0 hr (time for end of the 09 August event runoff), 48 hr and 96 hr (48 and 96 hours retention af ter 09 August event).
179 CHAPTER 6 TESTING OF VOLUMETRIC CLARIFYING FILTER SYSTEM FOR URBAN RAINFALLRUNOFF TREATMENT ON PARTICULATE MATTER REMOVAL Introduction Stormwater runoff from highway s and other roadways has been recognized as a specific source of contamination, poten tially impacting receiving-wat er ecosystems (U.S. EPA 1984). The chemical composition has been reported for such water quality constituents as solids, metals, nutrients, and petroleum-related organic compounds, including polycyclic aromatic hydrocarbons(PAHs) (Marsalek 1999; Sansalone et al. 1998). Particulate matter is a potential concern not only because of the environmental and ecological issues related to solid itself, but also considered as a significant reservoir for bot h chemical constituents and toxicity (Morrison et al. 1990; Makepeace 1995; Sansalone 2002). For ex ample, even though the dissolved loads of heavy metals are significant, most of the polluta nt burden appears to be transported by solids (Lee et al. 1997; Marsalek 1999). Th erefore, when these solid cont aminants remain in the water column, they are toxic to fish and other aquatic life. Even when they settle out they still can cause damage to the aquatic system. Anthropogenic activities and infrastructure, pa rticularly transporta tion and maintenance regimes, are significant c ontributory sources of particulate matte r, ranging in size from smaller than 1 m (colloidal) to greater than 10,000 m (gravel) (Sansal one et al.1998). It is generally accepted that a coarser fraction of solids is predom inant on mass basis, so it is often the first type of stormwater pollutant targeted in urban stormw ater management. However, as expected, finer particles can provide higher sp ecific surface area ( SSA), which is a critical index when considering the partitioning of constituents on the surface of particulate matter. Thus the granulometric characteristics of these particles, including particle size dist ributions (PSDs), play
180 an important role in determining their transpor t and fate, and consequent ly in the operation and viability of treatment systems designed to mediate water quality as well as toxicity. Urban runoff problems, because of their peri odic and stochastic nature, need to be addressed through a combination of source control, regu latory and structural best management practices (BMPs). Since the National Pollutant Discharge Elimination System (NPDES) Storm Water Phase II permitting regulations in 2003, there has been a proliferati on of BMPs. Structural BMPs eventually require engineering design BMPs (James 1999; Sansalone 2005). Limited land area and the increasing cost of ur ban land have led to considerat ions of smaller Unit Operation and Processes (UOPs) such as Hydrodynamic Se parator (HS) and filtration systems over the gravitational sedimentation widely used in stormwater controls. However, in many cases, a HS was proved to be effective only in the removal of gross solids and de bris (Wong et al. 1995; Walker et al. 1999; Schwarz and Wells 1999). In addition, according to our previous study on the scouring test on HS, re-suspensi on of trapped sediments resulted in higher pollutant effluent concentration due to the unsteady nature of processes in eventbased treatment tests, including rainfall runoff, mobilization, pa rtitioning, and delivery of polluta nts (Kim et al 2007; Liu et al. 2001). Installation of filter insert s into storm drains at street level is a convenient BMP for controlling urban runoff due to its potential specificity and ease of placement (Hipp et al. 2006), but such systems are generally not effective unles s maintained or replaced on a frequent basis. This is challenging when such systems are genera lly located 300 feet apart along most roadways. In addition, such systems do not provide any hydr ologic restoration. On the other hand, off-line engineered adsorptive filtration systems, which include unit pro cesses such as adsorption and precipitation onto high surface area materials and un it operations such as filtration, can represent viable stormwater treatment for combined unit op erations of filtration and processes of surface
181 complexion, in relation to the removal of part iculate-bound and dissolved pollutants, respectively (Liu et al. 2001). In general, filtration technologies are typica lly accomplished by a replaceable filter media capable of removing finer partic les than screening. Effective removal of TSS and other water quality parameters from stormwater has been achieved with sand filters; however, these systems are expensive to construct and require freque nt maintenance to en sure proper hydraulic performance. To challenge this condition, a Vo lumetric Clarifying Filter (VCF) system filled with engineered coated pumice ( 3.56 0.79 mm) was utilized in this study, intended to separate both coarser and finer particles and associated polluta nts without providing hydrological restoration. While in practice nearly all simila r systems do not provide hydrologic restoration, proper design can allow these systems to provide it. This device is also designed to require less space and less maintenance than traditional filters. As expected, filtration performance was associ ated with the effluent water quality (SSC, TSS, turbidity, PSD etc.), water production ( unit filter run volume), and head-loss development (Clark 2000). In addition, it also depends on ma ny factors such as the surface loading rate, residence time, and the physicochemical characteri stics of the media (type, size, porosity, and surface charge) (Liu et al. 2001; Liu et al. 2004; Teng and Sansalone 2004). Media selection should be based on the environmental factors and co st of the project. Ease of media replacement, foreign debris minimization, and feasibility of installation are several design considerations that enhance the effectiveness of th e system. Optimum size, sorting, mixture and packing can also significantly affect the performance of rainfall runoff filtration (Wesley 2001; Kitis et al. 2006). Pumice is a light, highly porous (pore volume up to 80%) volcanic rock that is formed during explosive eruptions. It resembles a sponge b ecause it consists of a ne twork of irregular or
182 oval shape internal voids/pores or vesicles, so me of which are interconnected and open to the external surface, while others are isolated insi de the particle (Gunduz et al. 1998; Wesley 2001; Kitis et al. 2006). Pumice has been tested and used in various environmental applications mainly as an adsorbent, filtration medi a, biofilm or catalyst support. It exhibited a high potential for use as a filter bed material for tu rbidity removal under rapid filtra tion conditions (Farizoglu et al. 2003). The specific surface areas of various natu ral pumices are generally larger (about 5 m2/ g) than those of conventional sand used in filtration processes (Kitis et al. 2006). Objectives There are five objectives of this research. First objective is to compare the representativeness of sampling behavior between ma nual and auto. The second objective is to present the results of the concentration change and removal of three fractions of particulate matter (PM). The third one is to examine the particle size gradation for runoff before and after the VCF system. The fourth objective is to exam ine the SSC change and the relationship with turbidity. The last objective is to identify the role of gravitational settling and filtration on PM removal. Methodology Volumetric Clarifying Filter Sy stem Structure and Configuration The VCF system was located at an urban NP DES Phase II site in Baton Rouge, Louisiana and collected runoff directly from the elevated pavement area of the Interstate-10 Highway bridge over City Park Lake. Rainfall-runoff was collected directly through the drainage delivery systems piped to the VCF system as shown in Figure 6-1. Once collected by the drainage delivery system, influent flow depth was measured in a 50.8 mm (2-inch) Parshall flume. Flow leaving the Parshall flume wa s transported through a 152.4 mm ( 6-inch) PVC pipe for 2 m to the influent drop box, then into another 152.4 mm ( 6-inch) PVC pipe, which then transports
183 the flow vertically to the botto m of the system vault of the VC F system, where heavy settleable and sediment size particles settle out. A schema tic plan of the volumetric clarifying filter system is also shown in Figure 6-1. Fi ve cartridges with filled (AOCM)p were arranged in parallel and the drainage pipes of cartridges we re connected to the effluent co llection pipe, which opens to an orifice. Each filter cartridge is approximately 558.8 mm (22 inches) in height and 457.2 mm (18 inches) in diameter (corresponding to an approximate cartridge surface area of 0.80 m2), and has a design hydraulic capacity of 1.14 L/s gpm. During each storm, the center annular tube of each filter cartridge allows gravitational drainage from the media and transports the flow to a central drainage manifold as shown in the side view of VCF system in Figure 6-1. This gravitational mechanism is common to all filter cartridge systems and, as with this system the outflow from the cartridge is commonly controlled by valve or orifice controls immediately downstream of the filter cartridges. In this system, as the flow is treated, a valve is trigge red and the downstream control is modified so effluent flows from the cartridge s are not restricted. Flow in excess of the hydraulic capacity bypasses the filter cartridges and flows directly to the outlet drop box. Rainfall and Hydrological Data Collection An outflow measurement system beyond the dr op box was established for effluent flow measurement. In addition, calibrated pressure transducers were located in the VCF system, the Parshall flume, and effluent flow system for head measurements of influent outside and inside of cartridges, water level in efflue nt box and the effluent. The pressu re transducers were hard-wired and the water depth and flow levels were tran smitted to the datalogger (CR 1000) every second. The CR1000 datalogger was configured to enable it to average data over ten seconds to provide a mean value for each ten seconds interval. The te n seconds flow level average values for both influent and effluent were conve rted to flow rates using head-d ischarge relationship for Parshall
184 flume and V-notch weir. Rainfall data were coll ected by using a standard tipping bucket rain gage. Rainfall data were recorded every one minute and was downloaded after every event and every month. Experiment Sampling Design Nineteen (19) discrete rainfall events occurring between 21 April 2006 and 09 August 2006 were monitored, and the entire volum e of runoff for each event was collected. Based on the real-time radar image information provided, sample s were then taken at 2 to 5 minute interval during the rising limb of the hydr ograph and thereafter at inte rvals ranging from 10 to 30 minutes. Sampling continued for the duration of runoff. All samples were pre-labeled in duplicate 500 mL, 1 L, 4 L bottles and 20 L containers for both influent and effluent manual sample collection. Sample analyses included particle size distributions (PSDs) through laser diffraction (down to 0.05 m) and mechanical sieve analysis (up to 9500 m), suspended sediment concentration (SSC) a nd particulate matter (PM) fracti ons (suspended, settleable and sediment). Because of volumetric attenuation wi thin the VCF the duration of effluent sampling was extended beyond the influent sampling. Change s in chemical or particulate loads were determined from influent and effluent concentra tions and flow data to quantify the performance of the system. Two sampling methods, manual sampling and auto sampler, were used simultaneously to compare the representative of the sample in practice. Instead of comparing individual incremental samples, a flow-weighted composite sample was prepared by collecting sample aliquots and combining them based on the pr oportion of flow represented by each. Thus, the composite sample attempts to integrate the effects of many variations in stormwater quality that occur during a storm event. Flow-weighted comp osite samples are more suitable for estimating
185 event mean concentrations (EMC) and polluta nt loads. EMC was determined only from calculations based on discrete sample and increm ental flow analysis. When a bypass or overflow did occur, the volume and water chemistry of the bypass were characterized throughout the bypass duration. PM fractions, SSC and Turbidity According to the size range and more important ly the settling behavior, solids in rainfallrunoff were categorized into f our fractions: dissolved, suspen ded, settleable and sediment, according to their size range and settling behavior. The nominal 0.45-m size is considered the cut-off for dissolved solids and other fractions. Se ttleable PM was determined using Imhoff cone according to the Standard Method 2540 F. In this study, those particulate matters settled out in 1hour Imhoff cone settling test are considered as settleable PM. Settleable fraction is an indication of treatability through gravitational separation. Sediment size particles were retained on a # 200 sieve and recovered, dried and analy zed for solid mass determination according to standard methods. The ASTMs SSC (suspended sediment c oncentration) Method D3977-97B (ASTM 1997) measures the amount of PM contained in the whole sample through filtering the water, and drying and weighing the residue left on the filter. It was reported as the more representative and conservative approach compared to the EPA T SS and Standard TSS met hods, which may not be able to sub-sample coarser particles suspende d in the entire water column by using magnetic stirring (Gray et al. 2000). In addition to SSC, turbidity m easurements were performed with a nephelometric turbbidimeter. The relationship between PM and turbid ity depends on various factors including particle size, shape, composition (organic verses inorganic), and to a degree on water color.
186 Particle Size Distribution Two methods were utilized for PSD analysis in this study. For the suspended fraction, where number concentration dominated mass concen tration and particles c ould remain entrained during analyses, laser diffraction was utilized. Due to the capabilities of the laser diffraction analysis system, particle number concentration wa s also determined for the settleable and very fine sediment fractions. The laser diffracti on system provided total volume concentrations (TVC) in 100 increments with analysis limits that ranged from 0.02 to 2000 m. In contrast, for the sediment fraction where mass concentration dominated number concentration, mechanical sieve analysis was utilized. The mech anical sieve analysis ranged from 9500 m down through the settleable size fraction to 25 m, and utilized 17 discrete size increments. Conversion between the two systems of measurement can be made through knowledge of particle density and particle volume concentration. The entire particulate matter gradation, from accumulated particulate mass trapped by the filter and sedimentation bay, went through a series of granulometric analyses. After air-drying at 40C, particulate matter obtaine d was disaggregated and sieved through a set of graded mechanical sieves ranging fr om 9.5-mm (#3/8) through 25m (#500). Sieve analysis followed ASTM D422-63 (ASTM 1993). Based on laser di ffraction size and number analyses of the suspended and settleable fractions, the settleable fraction was determined to have a lower size limit of approximately 25 m (based on the cumulative d90m of the suspended fraction) and an upper limit of approximately 75 m. This upper limit corresponds to a US Standard #200 sieve size and is the nominal size that differentiates si lt and sand from a fine and coarse designations for soils and sediments (ASTM 1993). The lower limit of 25 also corresponds to the #500 sieve size. Particle size is defined by sieve diameter, which is the width of the minimum square
187 aperture through which the par ticle passes (Allen 1990). Dry so lids separated on each of the stainless steel sieves were weighed and stored separately in round clear sample bottles. Particle Size Indices Several indices can be used to describe the shape and dimension characteristics of particles. Innman (1952) suggested using the geometric mean of the particle gradation for nonsymmetrical or lognormal distributions The geometric mean diameter dg is calculated in equation 6-1. 84 16d d dg (6-1) The standard deviation g is also obtained from d16 and d84 in equation 6-2: 16 84d dg (6-2) The -size is often used in descri ption of particle size. The -size is the negative natural logarithm value of corresponding diameter. Central tendency 50 is given in equation 6-3. ) ( log50 2 50d (6-3) The other factors, incl uding uniformity factor I symmetry factor Sk and normality factor KG, are determined respectively in equation 6-4, equation 6-5 a nd equation 6-6. ) 6 6 4 (5 95 16 84 I (6-4) ) ( 2 2 ) ( 2 25 95 50 95 5 16 84 50 84 16 kS (6-5) ) ( 44 225 75 5 90 GK (6-6) In general, those parameters can be used to examine the granulometry of the PM transported in the urban rainfall-runoff.
188 Evaluation Metrics Process efficiencies or removal rates can be determined from in fluent and effluent contaminant concentration and flow data to quan tify the performance of the system. Event mean concentration (EMC) is a single index used to characterize the stormwater runoff pollution for the entire storm event as shown in equation 6-7: (6-7) In this expression V is volume of flow during period i; C is average concentration associated with period I; and n is total number of measurements taken during event. While effluent concentration, mass and PSD are the appropriate metric for filter effluent, mass removal should also be quantified as a pe rcent removal (PR) for any constituent on an event basis using the inflow and outflow loads. The following modified equation has been often suggested as the basic equation for calculating the percent removal rate of PM. 100 ) ( ) ( ) ( (%)1 1 1 IN i n i IN i EFF j m j EFF j IN i n i IN iC V C V C V PR (6-8) In equation 6-8, Vi-IN and Vj-EFF represent the volume of influe nt flow and effluent flow during sampling period i and j ; Ci-IN and Cj-EFF represent the mean concen tration associated with period i and j; and n and m are the total number of influent and effluent measurements taken during event, respectively. Quality Assurance and Quality Control (QA/QC) In this study a mass balance analysis was also conducted after all events to ensure mass conservation and QA/QC based on influent, effluent mass load and recovered mass in the unit n i i n i i iV V C EMC1 1
189 and trapped by the filters. Mass balance error wa s required to be within the range of % by mass and determined by the following equation 6-9: 100 M )] M M ( M [ (%) MBEinf rec eff inf (6-9) In this equation, Minf and Meff are the accumulative mass load of influent and effluent across the entire event; Mrec is the total mass of PM recovered after the event which including the PM trapped by the sedimentation bay and by the filters through sedimentation and filtration respectively. Results and Discussion Storm Events Description and Mass Fractions Overall, nineteen individual storm events were monitored and examined from April 2006 through August 2006, and entire volume of runoff for each event was treated by the VCF system. Basic event description was summarized in Tabl e 6-1 for each monitored event, which includes rainfall intensity (mm), previous dry hours (PDH, h) event flow (L/s), peak flow (L/s) and total runoff volume (L). During the entire research period, even t rainfalls were monitored from 1.02 to 71.4 mm (0.04 to 2.8 inches) and the total runo ff volume generated from the instrumented MS4 watershed varied from 495 to 48,306 L, which we re representative of the water quality volume requirements for treating runoff from the locality. The PDH varied from 16 to 927 hours, with lower values occurring duri ng back-to-back storm events (06 May 2006 and 07 May 2006, 27 May 2006 and 28 May 2006, 04 July and 05 July, 04 August and 05 August). The event flow rates were observed from 0.1 to 5.1 L/s indicat ing over one order of magnitude difference as the similar pattern occurred for rainfall intensity. The highest peak flow occurred in the highest intensity event (29 April 2006), and 275-L overflow was bypassed in to the effluent box at the peak flow (25.3 L/s) during the event due to the limited designed hydraulic capacity. However,
190 the percentage of bypass was relatively small (0 .57% of 48,306 L total runoff volume) in the high intensity storm. Table 6-1 also summarizes the accumulative ma ss of suspended, settleable and sediment PM and SSC for each individual event and total nineteen storm events. Total mass load of suspended settleable and sedi ment PM captured for all events reached to 8,862.2 g, 26,362 g and 22,088.3 g, respectively. In contrast, the accu mulative mass of filtered suspended, settleable and sediment PM discharged in effluent was 2, 999.1 g, 2,494.9 g and 436.3 g, respectively. Total mass of PM in terms of SSC for all events was 57,311.9 g for influent and 5,930.7 g for effluent, indicating a approximately 88% mass reduction by the VCF system. There is significant variability observed in the ratios of suspended, settleable and sediment fractions to total PM among events. Solid fractions of influent and efflue nt for all 16 events were calculated based on mass and results were shown in Figure 6-2. As seen in Figure 6-2, suspended solid fractions ranged from 5 to 32 % with an average value of 15.4% for influent and significantly increased to a higher percentage that ranged from 14 to 77% with an average value of 50.5% for effluent. Settleable solids generally made up of 42% of solids mass on average for influent and 46% for effluent. Sediment mass fractions dropped from 39% on average fo r influent to 7% on average for effluent. Experimental Sampling Sequential as well as composite sampling approa ches were examined regarding their field application and representativene ss in this study as shown in Figure 6-3. The agreement between the experimental data set and linear fitti ng curves is indicated by high coefficients (R2) of determination for comparison of measured and cal culated EMC values. As a comparative index, EMC is often used to characterize concentrations that represent a flow average concentration for the event (Huber 1993, Sansalone 2005). Theoretical ly, the EMC values of SSC are expected to
191 be identical to the result of the mass quantific ation based on concentratio n and incremental flow, therefore the sub-sample from the discrete sa mples became a contentious issue during composite sample preparation in terms of the representati veness, in particular for influent. Results indicated that the composite samples were found to be underestimated significantly (p < 0.05) in terms of EMC values of SSC for influent, while the difference between composite and theoretical values was not statistically significant (p > 0.05) for effluent. Furthermore, in spite of cost of data acquisition, composite sampling base d on flow may provide adequate information of SSC yield from a small watershed as well as concentration and mass reduction for BMPs, but unpredictable intensity and duration of the storm often made flow in terval set up difficult in this study and reasonably representative of the entire volume captured in many cases. In contrast, the practical aspects of the sequential sampling are that of the detail and significance of the information obtained from the unsteady hydrolog y and variation in po llutant concentration during the event. For instance, if control of hi gh concentrations is targeted, the sampling design for the early portion of the event is critical. However, if mass is the primary focus, the sampling intervals will be based on a priori knowledge of the seasonal w eather patterns and information compiled from radar imagery systems (Sansalone and Cristina 2004; Appel and Hudak 2001). EMC values were not used in evaluation of process, but only tabulated as a nominal index, since an EMC cannot impart any info rmation regarding process behavior but only endpoint information. As shown in Figure 6-2, EMC values were determined from automated sampler composites, and these EMC values deviat ed significantly (ANCOVA, p < 0.01) from the representative discrete sample analyses and supp orting mass balances. It should be noted that the particles transported through the delivery system are not evenly distri buted across the crosssectional area, therefore the commonly used sampling procedures that employ automatic
192 peristaltic pumps to draw samples would bias th e total solids load to lower values because of their inability to sample the entire gradation of particles, especially large grain size particles carried in stormwater which settl ed quickly (Bent et al. 2000). Another problem may be the use of the auto sampler because the first flush or init ial runoff of an event might be missed; a certain allowable minimum volume of flow is required by the auto sampler to initiate effective sampling. The limitation on the number of samp les could be another constraint for a high intensity storm with long duration. As a rational approach to the development of a representative sampling program, manual sampling proved to be the most conservative way compared to auto sampling. Mass Concentrations of PM Fractions A statistical summary of even-based mass c oncentrations of suspended, settleable and sediment PM for 19 qualified storm events were shown as box plots in Figure 6 4, Figure 6 5 and Figure 6 6, respectively. The number of sa mples (10 18) and inherent variability across the event in EMC data led to the selection of box plots as a nonparame tric method of data presentation to display the full range of data without requiring para metric assumptions of normality (Devore 1991). Boxes represent the range from the lower bound of the second quartile to the upper bound of the third (a distance sometimes described as the interquartile range), with the line between marking the median. The cross ma rks represent the event mean values and data within a distance of 1.5X int erquartile range of e ither boundary. In general, event mean concentrations (EMCs) of influent suspended PM transported tend to be higher ( > 60 mg/L) in lower flow events (< 100 L/min) while lower suspended EMCs (< 60 mg/L) were observed in higher flow events (> 100 L/min), as presen ted in Figure 6-4. Both of 19-June 2006 and 05August 2006 events appear to be the exceptions for the lower EMCs of suspended PM, possibly due to the short previous dry period since 16-June and 04-August events were caught ahead of
193 them, respectively. Times to even one order of magnitude lower EMCs of suspended PM in effluent indicate the positive re moval to a certain degree by VCF system. Figure 6 5 and Figure 6 6 show the settleable and sediment PM mass concentrations of both influent and effluent on event basis from low flow to high flow. While effluent mass concentr ations yielded around 1020 mg/L and remain relatively constant with less variance indicated by the length of the boxplots, influent mass concentrations generated a di fferent pattern. It is generally accepted that settleable and sediment PM cont ributed more to the mass, compared to finer suspended PM. However, in spite of the concentrations of inco ming influent, PM varied by orders of magnitude among storm events; the effluent mass concentrati ons are almost ten time s less for settleable and sediment PM than that for suspended PM, which also indicates the different removal mechanisms that occurred and their ro les of contribution to PM reduction. Flow rate is a critical variable to the tr ansport of heterogeneous material including PM. Consequently, the delivery of PM was very depe ndent on flow, as well as on traffic intensity and duration. High flow volumes resulting from urba n impervious nature can promote effective transport of PM during a high intensity, high vol ume runoff event, and PM mass is found to be directly interacted with th e hydrology (Sansalone et al. 1998; Singh 1997). Although response varies as a result of previous dry hours, elapsed time of the even t and the loading of transported PM to the system, for the nineteen storms obser ved, these figures demons trate that no significant correlation had been found between hydraulic load ing and PM mass concentration on an event basis. Removal Efficiencies and Mass Fractions To further investigate relative contributi on of these hydrologic and granulometric characteristics of rainfall-runoff event to sepa ration performance of the VFC system and PM concentration in runoff, additional bar plots we re generated showing results of mass removal
194 efficiencies for individual even t as shown in Figure 6 3 through 6 5. Generally, the mass removal of suspended PM reached over 60% except 05-August and 07-August events with influent EMCs around 20 mg/L, which almost reach to the effluent limit of the system. The removals of settleable and se diment solids ranged from 81% to 98% and from 93.7 to 99.7%, respectively. Relatively lower removals occur for suspended solids in all storm events, and the removals vary significantly from 52.4% to 88.5%. According to the removal mechanism, gravitational settling plays a major role in sediment size (> 75 m) particles, and filtration is the dominant process in suspended solids removal. Mo re than 10% of the settleable solids remain in suspension during the events althou gh they might settle afterwards. The detail will be discussed in recovered solids gradation results. Additionally, removal efficiency is highly relate d to the concentration and load of influent. Standards based on removal efficiency may pena lize cleaner sites and do not take into account effects on receiving water quality. For example, cer tain sites may have low PM concentrations due to increased efforts at pollu tion prevention or other upstream BMPs. Removal of 80% of PM from clean sites would be much more difficult than from a site with higher inlet concentrations. For these situations a target outlet maximu m concentration may be more appropriate. SSC and Particle Size Distribution Figure 6 7 illustrate that event mean SSC values for the nineteen targeted storms analyzed at the Baton Rouge site ranged from l79.2 to 526. 4 mg/L. It is recognized that, while influent SSC typically vary by one order of magnitude, ef fluent concentrations were in a narrow range from 20 to 40 mg/L, except the fi rst event that occurre d in April 21, 2006, which had a long PDH in relation to the generation and deposition of fi ne particles (< 25 m) on the pavement surface prior to the event. The fact of influent SSC and effluent SSC ranges adds further support for the hypothesis that the system can pr ovide reasonable removal for PM in terms of SSC generated
195 from small watersheds. Figure 6 7 also demonstrat ed that there was not a consistent change in SSC on an event basis with increasing event mean flow rate. Lower event flow rate and higher PDH did not always ensure the tr ansport of higher SSC from the urban pavement to the system. It appears that event me an SSC is considered as the joint result from numer ous hydrological and hydraulic parameters, including event mean flow peak flow, PDH, event duration and runoff volume etc. The distributions of SSC for both influent and effluent were investigated for all 19 events and depicted as a probability density function ( pdf) with statistical categories of SSC values shown in Figures 6 8. All pdfs could be de scribed by an optimized log normal distribution function, and Pearsons correlation coefficien ts indicate the Goodnessof-fit (GOF) for both influent and effluent SSC samples. Results indi cate that SSC values generally distributed in a range of one order of magnitude across the median value. The pdf for SSC were determined from 224 and 243 samples for influent and effluent respectively. A summ ary of the mean and standard deviation are shown in the table inserted into Figure 6 8. Standard deviation for influent SSC was found to be significantly bi gger than that for effluent SSC, illustrating the potential reasonable removal of c onsiderable range gradation of PM, in particular the coarser PM. Particle size distributions ( PSDs) based on percentage fine r by mass across the gradation range from 0.1 to 2000 m for nineteen storm events are summarized in bottom plot in Figure 6 7. The d50m values of PSDs are not based on the comp lete mass gradations, since any particles larger than 2000 m were removed before analysis due to the requirement of the equipment (Malvern PSD analyzer). Therefore, the d50m values tend to be underestim ated in this case. While there is some variability in the PSD s trend, results indicated that d50m did not change
196 significantly (p > 0.05) with in creasing flow rate on event ba sis except the 06 May 2006 and 29 April 2006 events. However, the trend showed an increased pattern with these two events included. These results suggest that the captured PSDs are influe nced by geological and physical characteristics of the catchment and granulometri c characteristics of PM source, available on the surface of catchment prior to the start of the event for a given range of hydrology. Additionally, the PSDs tend to skew toward finer or coarser in terms of d50m while the corresponding flow rates are below or above the certain range significantly. Turbidity and SSC Turbidity is primarily influenced by suspended matter such as clay, silt, plankton, or microscopic organisms (APHA 1998). Storm wate r monitoring sometimes involves turbidity monitoring rather than, or in addition to, su spended solids monitoring. The advantage to performing turbidity monitoring is that turbidity may be measured using in situ, continuous methods. The relationship between SSC and turbidity was reported in many studies as well, since SSC included suspended solids that play an importa nt role in turbidity. Gippel (1995) indicates that the relationship between tu rbidity and particle size may follow two empirical equations. If the particle size and composition do not systemati cally vary with respect to the suspended solids concentration, the relationship takes the general linear relationship, otherwise following a log linear relationship. In addition, correlation between them tends to be highly site specific and may vary within the same storm due to compositional variations. As we can see in Figure 6 9, the 3-D plan e indicates that turbidity (NTU), suspended solids (TSS) and settleable and sediment soli ds (SSC-TSS) follow a discernable log linear pattern in which turbidity increases as TSS and SSC increases in different level. The strength of this relationship is measured using the correlation coefficient r. As the value of R approaches 1, there is a high correlation between two parameters The correlation coefficient can be calculated
197 by taking the square root of the coefficient of determination (R2) from the regression analysis. The correlation coefficient r for effluent was de termined as 0.94, indi cating there was a strong relationship compared to that for influent with the R2 value 0.85. From the figure, turbidity was found to follow the SSC pattern very well for effl uent in all events a nd less accuracy of fit between them for influent, which suggests the larger size particles including settleable and sediment solids may contribute less compared to suspended solids. This conclusion also can be derived from the slope values in the equations describing the role played by coarser particles (> 25 m) and finer particles (< 25 m) on turbidity. Recovered Solids Gradation a nd Particle Characteristics Cumulative gamma distribution cu rve is constructed by plotting the percent of PM finer by mass against the sieve diameter or particle di ameter. Figure 10 shows the cumulative particle size distribution for solids recovered from sedime ntation bay and cartridges collected out of 19 rainfall-runoff events at the experimental site. Ch aracteristics of particle size distribution were summarized in Table 6 2. The average median diameter based on mass for the solids recovered from sedimentation bay and top, middle and bottom layers of cartridges are 299.5 m, 63.2 m, 54.7 m, and 34.3 m corresponding to a size of 1.74, 3.98, 4.19 and 4.87, respectively. These values indicate that particles in the sedimentatio n bay are much coarser than particles trapped in cartridges. Among the three layers of cartridges, the solids recove red from top layer seem to be finer than middle and bottom layers, suggesting th e unevenly distribution and filtration from the top to bottom of cartridge. Gravel size and medium sand size particulates in the sedimentation bay can be separated from the runoff st ream through gravitational settling. From the gradation curve for solids recove red from sedimentati on bay (Figure 6 10), settleable and sediment solids were approximately 96% of the total mass, and representing
198 70.4% of the total settleable and sediment solid s from the influent for all storm events. It indicates that less than 30% of settleable solids were trapped through filtration; sediment solids settled very quickly. This phenomenon also can be found from solids gradation recovered from cartridges. From Figure 6 9, suspended solid s recovered from cartridges was less than 40% based on mass, which suggests that some smaller size settleable particles and possibly a small portion of sediment particles may also contribute to the rest of mass percentage. The total mass removal for 19 storm events captured was 89.4% by VCF system, and 77.1% of this removal was contributed by gravitational settling and 22.9% by filtration. Although the contribution to mass removal is less than a quarter, the removal by f iltration still played an important role here because not only is it a very necessary process to reduce the suspended solids concentration and turbidity in effluent close to the target level, but also it helps a great deal with potential toxicity reduction on the high specific su rface area of fine particles and metal and other toxicants associated with them. Conclusion Anthropogenic activities and infrastructure, pa rticularly transporta tion and maintenance regimes, are significant contri butory sources of particulate matter ranging from sub-micron particles to gravel size material. Unit operations such as filtration can represent viable storm water treatment for particles and particulate-bound pollutants. Installation of a volumetric filter system is a feasible way addressing and contro lling urban runoff due to the systems potential specificity and ease of placement. Several conclusions were summarized as follows: Manual sampling proved to be the most conser vative way compared to auto sampler. Autosampler usually cannot sample the entire gradatio n of particles, especi ally large grain size particles, and employing automatic peristaltic pumps to draw samples would underestimate the total solids.Removal efficiency is highly related to the concentration and load of influent.
199 Standards based on removal efficiency may penali ze cleaner sites, and they do not take into account effects on receiving water quality. For these situations a ta rget outlet maximum concentration may be more appropriate. Turbidity (NTU), suspended solids (TSS) and settleable and sediment solids (SSC-TSS) follow a discernable log linear pattern; settle able and sediment solid s may contribute less on turbidity compared to suspended solids. Correlation between them tends to be highly site specific and may vary within the same storm due to comp ositional variations. The mass concentrations of particle size distribution were a pproximately one magnitude less in effluent compared to those in influent. The volumetric filter system was effective in removing an average of 89.4% of the incoming SSC, and 77.1% of this removal was contributed by gravitatio nal settling and 22.9% by filtration. Uneven distribution and filtration oc curred from the top to bottom of the cartridge. Although the suspended solids mass removed by filtr ation is less than a quarter, it still played an important role here because not only is it a very necessary process to reduce the suspended solids concentration and tu rbidity in effluent close to the target level, but also helps a great deal with potential toxicity reduction due to the high specific surface area of fine particles and metal and other toxicant s associated with them.
200Table 6-1. Summary of mass of each partic ulate matter (PM) fraction (suspended, settl eable and sediment) and SSC for 19 storm events. P PDH Qave Qmax V Suspended (g) Settleable (g) Sediment (g) SSC (g) Event (mm) (hr) (L/s) (L/s) L Mi Me Mi Me Mi Me Mi Me 21 April 4.06 927 1.2413.32927663.176.51924.3 39.11574.44.14161.7119.7 29 April 71.37 84 5.0825.3483062659.61039.13993.2 461.43813.4178.210466.21678.706 May 1.02 157 0.070.349539.98.972.0 2.460.31.6172.213.007 May 6.35 21 2.469.13852158.955.41146.3 69.9324.97.61630.1132.927 May 3.56 482 1.816.52628157.556.4687.2 29.3374.47.81219.193.528 May 3.05 23 1.547.4209685.019.9120.2 12.5115.17.3320.339.716 June 14.22 451 2.0614.69938266.255.31851.2 280.03246.265.95363.6401.319 June 3.05 79 1.014.3181650.614.7257.9 18.5232.96.9541.440.104 July 3.81 352 1.156.22779204.078.6356.3 19.3271.52.9831.8100.805 July 5.59 25 0.997.93838225.572.5517.6 60.2322.06.01065.1138.709 July 1.78 69 0.311.667470.123.070.4 8.064.20.8204.731.810 July 30.48 16 4.2615251891268.2465.74375.8 462.94225.234.19869.2962.814 July 5.59 89 0.793.83304359.182.9834.5 39.2204.84.81398.4126.816 July 2.54 45 0.832.894579.817.185.4 6.981.60.4246.724.418 July 2.03 26 0.313.3104749.112.6169.3 8.757.60.5276.121.904 August 52.32 191 4.5020.3369901837.4672.97729.1 668.75458.842.7150251384.305 August 9.14 19 1.269.06421132.362.9465.4 88.3482.45.21080.1156.407 August 8.89 25 1.4510.76022111.246.6512.9 30.4266.03.6890.080.609 August 16.26 23 2.0414.812502444.7138.11193 189.2912.655.92550.2383.3 245.10 1717698862.22999.126362 2494.922088.3436.357311.95930.7 Note: P, PDH, Qave, Qmax, V represent Rainfall depth, previous dry hours, mean flow rate, peak flow rate and runoff volume on event basis. Mi Me represent total mass of each particulate matter fraction for influent and effluent on event basis.
201 Table 6-2. Summary of statistica l characteristics of particle size distribution for the particulate matter (PM) recovered from sedimentation bay and top, middle and bottom layers of cartridges. Source of particulate matter recovered Parameters Sedimentation bay Cartridge bottom layer Cartridge middle layer Cartridge top layer d50 ( m) 299.563.254.7 34.3 50 1.74 (medium sand) 3.98 ( fine sand) 4.19 (silt) 4.87 (silt) dg ( m) 269.7541.9846.28 21.58 g 3.893.362.64 2.88I 1.92 (poorly sorted) 1.75 (poorly sorted) 1.46 (poorly sorted) 1.56 (poorly sorted) Sk -0.04 (symmetrical) -0.24 (coarse skewed) -0.22 (coarse skewed) -0.36 (coarse skewed) KG 0.93 mesokurtic 1.09 mesokurtic 1.46 Leptokurtic 1.08 mesokurtic
202 Figure 6-1. Plan view of site and configuration of Volumetric Clarifying Filter (VCF) system. 10.2 cm PVC pipe from west Data logger 5.1 cm Parshall Watershed PCC pavement 1088 m2 (2 x 544-m2) 2% surface slope ADT=142,000 (E+W) Tee Influent Baffle Media cartridge Influent box Influent Effluent box Drainage pipe Effluent outlet Orifice plate 45.7cm 53.3 cm 187.2 cm 169.2 cm Oil baffle Media cartridge Drainage pipe Effluent drop box Effluent pipe Float valve Orifice Influent delivery pipe Influent box 212.1 cm 45.7 cm 116.8 cm Influent Effluent Effluent collection pipe 212.1 cm Trough @ 6% Trough @ 6%
203 Particulate fraction (%) 0 20 40 60 80 100 Particulate fraction (%) 0 20 40 60 80 100 06-May 08-July 09-July 14-July 16-July 05-July 19-June 04-July 05-August 21-April 07-August 28-May 27-May 16-June 09-August 07-May 10-July 04-August 29-AprilSuspended Settleable Sediment Influent Effluent Figure 6-2. Suspended, settleable and sediment fractions of influent a nd effluent based on mass for 19 storm events.
204 0100200300400500600 SSC as a measured EMC [mg/L] 0 100 200 300 400 500 600 Manual composite Auto composite Theoretical agreement Y= a+ bX (n = 19) abr 2 11.70.840.98 6.30.720.90 INFLUENT SSC as a calculated EMC [mg/L] 020406080100 SSC as a measured EMC [mg/L] 0 20 40 60 80 100 Y= a+ bX (n = 19) abr 2 11.50.850.91 6.90.820.90 EFFLUENT Manual composite Auto composite Theoretical agreement Figure 6-3. Event based comparison between com posite auto sample and manual sample. Thick line with slope = 1 represents the theoretical values.
205 06-May 18-July 09-July 16-July 14-July 05-July 19-June 04-July 05-August 21-April 07-August 28-May 27-May 16-June 09-August 07-May 10-July 04-August 29-April Mass removal (%) 0 20 40 60 80 100 Flow rate (L/min) 0.01 0.1 1 10 100 1000 Flow rate Mass removal (%) 1 10 100 1000 1 10 100 1000Effluent suspended [mg/L] Influent suspended [mg/L] Effluent suspended Influent suspended Figure 6-4. The coupled relationship between flow rate, influent and effluent suspende d solid mass concentration and mass remov al (as an index of VCF system performance) by clarif ication in a small watershed for 19 storm events.
206 06-May 18-July 09-July 16-July 14-July 05-July 19-June 04-July 05-August 21-April 07-August 28-May 27-May 16-June 09-August 07-May 10-July 04-August 29-April Mass removal (%) 0 20 40 60 80 100 Flow rate (L/min) 0.01 0.1 1 10 100 1000 Flow rate Mass removal (%) Effluent settleable Influent settleable 1 10 100 1000 10000 1 10 100 1000 10000Influent settleable [mg/L] Effluent settleable [mg/L] Figure 6-5. The coupled relationship between flow rate, influent and effluent settleab le solid mass concentration and mass remo val (as an index of VCF system performance) by clarif ication in a small watershed for 19 storm events.
207 06-May 18-July 09-July 16-July 14-July 05-July 19-June 04-July 05-August 21-April 07-August 28-May 27-May 16-June 09-August 07-May 10-July 04-August 29-April Mass removal (%) 0 20 40 60 80 100 Flow rate (L/min) 0.01 0.1 1 10 100 1000 Flow rate Mass removal (%) Effluent sediment Influent sediment 0.1 1 10 100 1000 10000 0.1 1 10 100 1000 10000Influent sediment [mg/L] Effluent sediment [mg/L] Figure 6-6. The coupled relations hip between flow rate, influent and effluent sediment solid mass concentration and mass remov al (as an index of VCF system performance) by clarif ication in a small watershed for 19 storm events.
208 Influent PSD ( m) 0.1 1 10 100 1000 10000 Flow rate (L/min) 0.01 0.1 1 10 100 1000 Flow rate Influent PSD Influent SSC [mg/L] Effluent SSC [mg/L]10 100 1000 10000 10 100 1000 10000 Effluent SSC Influent SSC 06-May 18-July 09-July 16-July 14-July 05-July 19-June 04-July 05-August 21-April 07-August 28-May 27-May 16-June 09-August 07-May 10-July 04-August 29-April Figure 6-7. The coupled relationship between flow rate, PSD and influent and effluent SSC by clarifi cation in a small watershed for 19 storm events.
209 SSC Concentration [mg/L] 1101001000 Probability density function 0.000 0.005 0.010 0.015 0.020 Influent Effluent R2p 285.431.6 169.7 18.1 0.930.92 > 0.05> 0.05 E(X) : Mean S(X) : Standard deviation SSC Concentration [mg/L] 02004006008001000 Frenquency of Occurrence % 0 5 10 15 20 R 2 = 0.93 Influent Frequency pdf 020406080100120 Frequency pdf SSC Concentration [mg/L] R 2 = 0.92 Effluent E(X) S(X) [mg/L] [mg/L] [mg/L] [mg/L] n224243 Figure 6-8. Log-normal distributi ons of influent and effluent SSC concentrations for total 19 storm events.
210 1 2 3 4 5 6 7 0 1 2 3 4 5 6 0 2 4 6 8l n ( N T U )l n ( T S S )l n (S S C -T S S ) 1 2 3 4 5 6 7 2 3 4 5 6 0 2 4 6 8l n ( N T U )l n ( T S S )l n ( S S C T S S ) ln(NTU)= 0.8503 + 0.7706ln(TSS)+0.1111ln(SSC-TSS) R2 = 0.85 ln(NTU)= 0.7634 + 0.7616ln(TSS)+0.0459ln(SSC-TSS) R2 = 0.94 Influent Effluent Figure 6-9. Relationship among turbidity (NTU), Suspended solid concentration (TSS) and settleable and sediment solids concentration (SSC-TSS).
211 Particle size ( m) 1 10 100 1000 10000 Percentage finer by mass (%) 0 20 40 60 80 100 sedimentation bay (m = 37300.2 g) bottom layer (m = 4292.5 g) middle layer (m = 4279.5 g) top layer (m = 2618.5 g) Filtration 22.9%Sedimentation 77.1%mass removal Figure 6-10. Particle size distri bution (mass gradation) for solid s recovered from sedimentation bay and volumetric clarifying filters. Three layers (bottom, middle and top) were gently washed and solids were recovered separately. The solid lines are cumulative gamma distribution fitting curves.
212 CHAPTER 7 VOLUMETRIC CLARIFYING FILTRATION FOR PHOSPHORUS IN URBAN RAINFALL-RUNOFF Introduction As the pollutant of primary concern to the ecological health of fresh waters, phosphorus has been long recognized as the limiting nut rient for eutrophication (Correll 1998), which commonly plagues lacustrine systems due to the elevated concentr ations of phosphorus introduced and accumulated in stagnant surface waters (Wendt and Corey 1980; Sharply 1999; Welch 1992). One of the most profound effects of eutrophication is the depletion of dissolved oxygen in the receiving water column with the consequent fish kills due to hypoxia condition (Kofinas and Kioussis 2003; Zhao and Sengupta 1998). In addition, blooms can detract from the aesthetic value of the water re source and create taste and odor problems in drinking water from reservoirs. Having these adverse consequences, eutrophication imposes a high environmental, ecological and health cost on many sect ors of economy (Pretty et al. 2003). A significant load of phosphorus can be tran sported in stormwater runoff including many types of non-point rainfall-runoff discharges from urban areas (EPA 1993). Several origins of phosphorus in stormwater are known today such as the discharges from automobile exhaust and other combustion processes, th e burn-up of organic material releases phosphorus into the atmosphere and the mineral liberated from pa vement surface during weathering (Strecker 1994). Lawns and streets were identified as the larges t sources of total phosphorus being transported from urban areas (Garn 2002; Waschbusch et al 1999). In many cases, phosphorus is introduced into the aquatic environment in a number of differ ent chemical forms, and has been described in general as being present in the aqueous phase an d in the solid phase and may change according to a dynamic equilibrium (Compton et al. 2000). In addition to a simple partitioning concept between dissolved and particulate phases, various species of phosphorus are often associated
213 with and distribute across a wide gradation of particulate matter (PM) in rainfall-runoff ranging in size from smaller than 1 m to greater than 9500 m (Sansalone et al. 1998; Sansalone and Kim 2007). The knowledge of the amount of pollutant s associated with di fferent particle size ranges is important in evaluating the efficiency of treatment facilities or BMPs designed to target the most polluted PM sizes (Vaze and Chiew 2004). Various structural Best Manage ment Practices (BMPs) have been developed in order to mitigate the adverse impact of phosphorus from st ormwater runoff. However, some stormwater BMPs rely on vegetation, soil infiltration, or bo th to remove dissolved phosphorus, while others have less or no mechanism to remove dissolved phosphorus. Generally, it is challengeable to efficiently remove dissolved phosphorus from st ormwater runoff compared to the particulate bound phosphorus removal through conventional sedime ntation and physical filtration processes (Jenkins et al. 1971), but cost-e ffective technologies are needed to remove phosphorus on target of both dissolved and particulate phases from urba n rainfall-runoff to efficiently perceive total phosphorus (TP) and total dissolved phosphorus (TDP) reduction whil e avoiding the high construction and operation costs of a stormwater treatment facility or mechanical treatment process (Erickson et al. 2007). A volumetric clarifying filter (VCF) employed in this study is one tool for addressing and controlling urban r unoff aqueous and particulate chemistry of phosphorus. Unit processes such as adsorption and precipitation onto high surface area materials or unit operations such as filtration can repres ent viable storm water treatment for dissolved pollutants and particulate-bound pollutants, respectiv ely in this combined system (Liu et al. 2001). In urban rainfall-runoff, the transport and ev entual fate of phosphor us is controlled by reactions with solid surfaces. Characterized by la rge surface area to solid volume ratios, natural
214 mineral oxides and oxide-colated media can serv e as reservoirs for dissolved phosphorus with the amphoteric surface. This leads to the idea th at engineered amphoteric oxide coated surfaces on stormwater runoff filter media may have la rge surface areas for adsorption, precipitation, absorption and surface complexation to effec tively remove phosphorus (Ma et al. 2005). According to the definitions given by Berkheiser et al. (1980), a sorption reaction involves the removal of phosphate from solution by its concentr ation in a solid phase. This may be reversible physical sorption or irreversible chemisorption. Precipitation is the re moval of two or more components from a solution by their mutual comb ination a new solid-phase compound (Holtan et al. 1988). In presence of water, oxide surfaces such as Ca and Fe are covered with surface hydroxyl groups, protons and coordinated water mol ecules. Recent studies indicated that other material such as activated alumina, is also able to remove phos phorus efficiently through adsorption and ion exchange (Hano et al. 1997; Seida and Nakano, 2002; Zhao and Sengupta 1998). A number of sorbent media are available fo r the transfer of dissolved or entrained particulate bound phosphorus elements from the liquid phase onto or into the surface of the porous media. Furthermore, cost-effective sorp tive materials are particularly needed in stormwater BMPs because stormwater program s are usually budget-constrained (Zhang, 2006). Filter sand has a long history as an effective me dia to remove particulate-bound contaminants from water and wastewater (Yao et al. 1971; Sansalone 1999), however, the phosphorus control relies heavily on sand filtration tends to have ve ry intensive maintenance requirements. In this study, pumices were used as engineered granul ar support media to inve stigate their adsorptive phosphorus removal from stormw ater runoff under field conditi on. Pumice is a light, highly porous (pore volumes up to 85%) volcanic stone formed during explosive eruptions and consist
215 of a network of irregular or ova l shape internal voids/pores or vesicles, some of which are interconnected and open to the external surface, wh ile others are isolated inside the particle (Gunduz et al. 1998; Wesley 2001 ). The specific surface area of various natural pumices are generally larger (about 5-15 m2g-1) than those of conventional sand used in filtration processes; thus, pumice may provide more surface area for oxides coating, which may enhance the adsorptive removal of phosphorus (Kitis et al. 2007).Pumice also exhibited a high potential for use as a filter bed material for turbidity re moval under rapid filtration condition, which may suggest a high potential for particulate bound phosphorus reduc tion (Farizoglu et al. 2003). Field and laboratory studies ha ve been undertaken by several researchers to determine the phosphorus removal by using engineered media filtra tion systems. Han et al. (2003) showed that the soluble iron species deposited on the junipe r fiber acted as the phos phorus adsorbent with sorption capacity about 2.2 mg/g. Erickson et al (2007) found the steel-enha nced sand filtration retained between 25 and 99% of dissolved phosphorus removal. Th e filtration system should be sized and placed according to local guidelines with proper pretreatment to settle some influent solids before entering the filters. This study propos es Volumetric Clarifying Filter (VCF) system to cost effectively remove TP and TDP in ur ban rainfall-runoff by sedimentation and filtration and through adsorption, respectively. Objectives There are four objectives in this chapter. Th e first objective is to present results on the concentrations and removals of TP and TDP on ev ent basis. The second objective is to examine the cumulative mass loading rate and first flush phenomena of TP and TDP. The third one is to examine the particulate phosphorus fractions and their removals associated with suspended, settleable and sediment PM in VCF system. The last objectiv e is to examine the phosphorus specific capacity in relation to suspended, settleable and sediment fractions.
216 Previous Work Bench Scale Study In this study, the aluminum oxide coating was applied by treating the gr anules of concrete ( = 2 mm) with aluminum nitrate. Column studi es were conducted where columns packed with these granules were used to remove phosphate fr om simulated stormwater at concentrations of 100, 50, 25, and 10 mg/L. A control study without the aluminum nitrate coating was also performed as a comparison. The effectiveness of oxide coated media in phosphorus removal has been proven on a microscope scale in bench scale studies. In addition, the inferences from X-ray absorption spectroscope indicated that the phosphorus K edge spect rum of PM from stormwater showed a higher peak weight for 25 m particle s compared to the other two treatments (75 and 800 m). The spectra also showed a pronounced di fference between the conc rete with aluminum coating and without aluminum coating. Radial Flow Cartridge Test A series of full-scale single radial flow car tridge tests were conduc ted with engineered media (AOCM)P to evaluate the performance of th e filter cartridge on TP and TDP removal at constant flows from 16.7 (0.19 L/s) to 133.3% (1.51 L/s) of design flow rate (1.14 L/s). Phosphorus was introduced at a constant nominal influent total phosphorus (TP) concentration of 1.0 mg/L. Results found that while filter efflue nt TP and TDP ranged from 0.31 to 0.62 mg/L and 0.20 to 0.26 mg/L across th e flow rate range for (AOCM)P respectively, indicating a reasonable concentration reduction in general. Methodology Experimental Site and System Configuration All data presented in this study were collected at an ur ban experimental facility in Baton Rouge, LA, which receives runoff from two 554-m2 (44.6-m long, 12.2-m wide) concrete-paved
217 drainage areas of elevated Interstate 10 (eas tbound and westbound) over City Park Lake. Each section has a transverse slope of 0.018-mm-1 across the length of the pavement and a longitudinal slope of 0.0099 mm-1. Descriptions of the experimental s ite as well as its hydrological and water chemistry characteristics are prov ided in detail elsewhere (Dean et al. 2005; Sansalone et al. 2005; Sansalone and Kim 2007). The VCF system was constructed and connected to the existing rainfall-runoff capture and delivery system as shown in Figure 7 1. The runoff volume was collected from the 1088 m2 catchment and drained through the gr avitational collection system in to the system directly. A 1psi pressure transducer (Druck Inc.) was insta lled in the monometer connected to the standard 50.8 mm (2-inch) parshall flume to record the influent depth at 1-second increments. A section of 152.4-mm (6-inch) PVC pipe tr ansports influent flow from the Parshall flume drop box to influent drop box followed by another 152.4-mm (6 -inch) PVC pipe, which transports the flow vertically to the bottom of a 212.1 cm by 116.8 cm vault, where heavy settleable and sediment size particles settled out. A bank of five radial flow cartridges filled with engineered (AOCM)p (Aluminum oxide coated pumice) was installed in parallel appr oximately 584 mm (23 inches) above the bottom of system vault. The confi guration of the cartridges and granulometric characteristics of the media were summarized in Tabl e 7 1. A float valve is triggered as flow is treated by the filter and the orifice control downstream is opened gradually as a consequence. The outlet flow was me asured through a 60o V-notch weir by another ha rd-wired 1-psi pressure transducer so that effluent depth and flow le vels can be transmitted to the datalogger CR1000 every second, where it was configured to enable average over ten seconds to provide a mean value for each ten seconds interval. Both influe nt and effluent flow was calculated by using a head-discharge algorithm for the Parshall flume and V-notch weir.
218 Experiment Setup and Sampling A storm event for this study was defined as a rain event that is preceded and followed by at least 6.0 hours of dry weather conditions. Once an incoming stor m was identified to meet the qualification of a storm event, the HDPE bottle s (500-ml, 1-L and 4-L), 20-L (5-G) buckets and treatment system were set to begin sampling. Samp le collection for chemical analysis occurred at two sampling locations as shown in Figure 1 a nd sampling across the entire cross sectional area of flow is required to ensure the representativen ess of the PM including the coarser sediment size particles. A minimum of 10 samples in replicat e were taken in order to provide a reasonable estimate of temporal constituent concentrations (Thompson et al. 1997; Dean et al. 2005). Usually samples were taken at 2-minute interv als in the beginning part of the event and thereafter at intervals ranging from 5 to 10 minutes for the duration of the pavement runoff. For anticipated long-term event, 30-minute interv als may be accepted after 10 samples were completed. Sample handling and preservation followed procedures established by Standard Methods 2540-D and 4500-P for particulat e matter and phosphorus analysis. Particulate Fractions In presented study, particulate matter (PM) is categorized into three fractions by size and more importantly by the settling mechanisms: susp ended, settleable and sediment fractions. The sediment fraction includes all par ticles retained on the # 200 sieve with nominal sizes larger than 75 m (ASTM 1993). The suspended and settleable fractions are separated based on the 60 minutes quiescent settling in Imhoff cone (APHA 1998). Settleable particles are indicated as the part of the particles passed through the # 200 me chanic sieve and settled out after 60 minutes which are generally larger than 25 m but smaller than 75m. On the contrary, the particles remained in the Imhoff supernatant are defined as the suspended particles with size less than 25
219 m in most cases. The Suspended Sediment Con centration (SSC) is defined as the combination of the suspended, settleable and sediment solids concentrations. Particulate fractions can be quantified by a method that involves collection of PM on a membrane filter or Petri dish, drying, and weighi ng the loaded filter or dish to get the total weight. The final weight of PM is the difference between the total weight and the tare weight of the empty filter or dish. For SSC measurement, us ually a sample aliquot rather than the whole bottle is transferred to the filter for filtration by using TSS method. Therefore it is sometimes difficult to take a representative sub-sample. To circumvent error generated by the sub-sampling, the entire sample is filtered, on mu ltiple filters if needed, rather than aliquots of it. In this study, the phosphorus bound to those PM fractions (suspende d, settleable, and sediment) were analyzed separately for each incremental influent and effl uent sample in replicate to infer treatability behavior for phosphorus with respect to these pa rticle size classes for a given combination of separation system and loading. Phase Fractionation Dissolved phosphorus (phosphate) is define d as the amount of phosphorus that passes through 0.45 m membrane filter (AHPA/AWWA/WEF 1995), which can be a large or small portion of total phosphorus depending on the partitioning process (Erickson et al. 2007). Total phosphorus (TP) is defined by a lab procedure th at first digests the entire aqueous sample, presumably oxidizing all P to a soluble reactive form that produces a signal in the standard test. Total dissolved phosphorus (TDP) is determined by the same procedure on filtered samples and particulate phosphorus (PP) is the difference between TP and TDP. A pressure filtration method was applied in th is study to fractionate rainfall-runoff samples into dissolved and particulate phase for further phosphorus analysis. The samples were filtered
220 by using a stainless steel pressure filter, contai ning a filter support of stainless steel, and a supported cellulose acetate membrane filter with 0.45 m pore diameter membrane filters. Membrane filters were washed by running 100-ml deionized water through them because they may contribute significant amounts of phosphorus to samples containing low concentrations of phosphate. For each filtered sample (dissolved fraction), 50 mL was saved for TDP measurement. Since dissolved phosphorus forms ar e to be differentiated, samples need to be filtered immediately after collection. Pres erve samples by freezing at or below -10 oC. It is important to note here that only phosphorus-free detergent can be used in all cleaning procedures related to phosphorus analysis. Analysis of Phosphorus In the total phosphorus (TP) determinati on, the sample was thoroughly mixed and a suitable portion (50 mL is recommended) was transf erred into a flask for digestion prior to the phosphorus analysis. Total dissolved phosphorus (TDP) was analyzed after a filtered sample was digested. Phosphorus acid digestion followed the Persulfate Digest ion Method (Standard Methods 1995). The analysis of phosphorus wa s performed using the ascorbic acid molybdenum blue method (EPA 365.2) us ing a HACH DR-4000 Spectrophotometer. Suspended, settleable and sediment fractions of PM illustrated above were carefully saved in the aluminum pans or glass petri dishes For the phosphorus bounded to suspended fraction, the membrane filter was cautiously unwrapped from the aluminum pan and transferred into the individually labeled flask which was treated by 5% HNO3 acid bath overnight and rinsed at least three times using DI water. For settleable and sediment bound phosphorus, the whole amount of particulate matter was placed into the respective dige stion flasks and the dishes should be rinsed by using DI water to ensure the residue was incl uded. If the dry solid mass was greater than 1
221 gram, a representative portion (approximately 0.5 to 1 g) of particulate matter was weighted to 0.1 mg and placed in a 100 ml labeled digestion flask. These prepared samples were digested according to same procedure (TP and TDP analysis) to ensure th e consistency of phosphorus measurement. The phosphate analysis was required to be conducted right away to avoid the repartitioning after certain residence time. A blank (DI water) was included for background leve l verification. The Dilution might be used depending on the turbidity and over-range readin g if necessary. All measurement and analyses were replicated. Event Mean Concentrations and Removal Efficiencies Process efficiencies or removal rates can be determined from in fluent and effluent contaminant concentration and flow data to quan tify the performance of the system. Because the concentration of phosphorus transported in urban runoff can vary by orders of magnitude during individual storm event, EMC is often used to characterize the flow av erage concentration of phosphorus for the entire storm event as shown in equation 7-1: r rt tdt t q dt t q t c C V M EMC0 0 (7-1) In this expression, M is the total mass of phosphorus over entire event duration; V is the total volume of flow over entire event duration; C is the flow-weighted average concentration for entire evtn; c(t) is the timevariable phosphorus fraction; q(t) is the time-variable flow; tr is the duration of the event and t is time. While effluent concentration and mass are the appropriate metric for filter effluent, mass removal should also be quantified as a percen t removal (PR) for any constituent on an event
222 basis using the inflow and outflow loads. Th e following modified equation has been often suggested as the basic equation for calcula ting the percent removal rate of phosphorus. 100 ) ( ) ( ) ( (%)1 1 1 IN i n i IN i EFF j m j EFF j IN i n i IN iC V C V C V PR (7-2) In equation 7-2, Vi-IN and Vj-EFF are the volu me of influent flow and effluent flow during sampling period i and j ; Ci-IN and Cj-EFF are the mean concentration associated with period i and j; and n and m are the total number of influent and effluent measurements taken during event, respectively. Categorical Analysis Translation of percentage and count data directly into an odds ratio can be employed to quantify the likelihood of an outcome and compare two groups of individual s with respect to a particular outcome. The proporti onal odds and odds ratio can be de scribed as following equation 7-3: 2 2 1 1 2 11 / 1 / odds odds (7-3) Equation 7-3 describes the ratio of odds from tw o rows in a two-way contingency table. In this table, the probabili ty of one outcome is 1 in row 1 and 2 in row 2 and the variables of row and column are independent when the conditional di stributions are identical in the two rows (i.e., 1 = 2). Results and Discussion Events Summary Totally nineteen storm events were monitored over a period of 4.5 months from April 3, 2006 to August 23, 2006, which typically represen t 42% of the annual rainfall for the Baton
223 Rouge area based on the data provided by Weather Wunderground (www.wunderground.com). The hydrological characteristics of storm events were summarized in Table 7 2 including rainfall intensity (P, mm), previous dry hours (P DH, hr), mean and peak flow rate (Q, L/s), runoff volume (V, L), duration of event (TD, min), surface loading rate (SLR, L/m2-min), residence time (RT, min) and contact time (CT, min). In general, the events sampled are able characterize the broad range of st orm events typical of East Bat on Rouge and fulfilled the project criteria for storm events. The highest intensity event (29-April, 2006) generated 48,306 L runoff in 165 minutes and 275-L overflow was bypassed dur ing the peak flow at around 25.3 L/s which was approximately 0.57% of the total volume during this event. In contrast, the lowest intensity storm occurred in 06-May which only generate d 475-L runoff in 75 minutes with peak flow around 0.3 L/s. It is generally accepted that adequate residen ce time and contact time must be provided for in the design of the system for both TP and TDP removal proce ss to occur. In terms of the TDP removal, longer contact time betw een media surface and di ssolved phosphorus was expected to provide improved adsorption under th e field condition in this study. SLRs varied from low to relatively high values as function of hydrology due to the unsteady flow loadings to the filters can be another important factor to a ffect the performance of the system. In addition, other parameters also contribu te to the removal including pr evious dry hours, influent concentrations, PM particle si ze distributions (PSDs) etc. TP and TDP Concentration The measured TP and TDP concentrations ar e presented as the box-and-whisker plots in Figure 7 2 and Figure 7 3 based on the event m ean flow rates (EMF) for total nineteen storm events monitored in this study. The box plots su mmarized the statistic measures of median, upper and lower quartiles and minimum and maximu m values of the incremental TP or TDP concentrations across each individual event. As a valuable index of polluta nt reduction, the event
224 mean concentrations (EMC) are pr esented as cross marks in all cases. Results indicated that the EMCs of TP and TDP varied by an order of ma gnitude for both influent and effluent over a range of hydrological events. However, these tre nds of variation did not follow the increasing order of EMF, possibly due to the hydrologic co mplexity and stochastic response for a small MS4 watershed subjected to various traffic loadings and antecedent dry period during and between events. In addition, investigation into the temporal change of influent TP and TDP concentrations with respect to the EMF across each rainfall event revealed that longer tail of the upper bound in box plot tended to occur in the high er flow rate events in terms of EMF in general, indicating the exponentia l decay of TP and TDP influe nced more significantly by high hydrological transport as primary driving force. Several exceptions ex ist including 09-August, 10-July and possibly 07-May events, which mi ght resulted from the previous washoff (07August, 09-July and 06-May) as well as short ant ecedent dry periods prior to the event initiated. For the four sets of back-to-back events ( 06-May and 07-May, 27-May and 28 May, 04-July and 05-July, and 04-August and 05-August events), the relatively higher TP and TDP concentrations occurred in the previous events of each set as expected in most cas es. It should be stated that the highest TP and TDP EMCs occurred in 21-April event is more likely associated with the prolonged dry period (927 hours) as significant amount of phos phorus has been accumulated during long-term drought. As exhibited in Figure 7 4, the probability density functions were developed for TP and TDP collected from all 19 events to determine whether any statistical difference of them between influent and effluent in a reasonable and accura te manner. Generally, sample populations of influent TP or TDP (n = 224) a nd effluent TP or TDP (n = 243) from all events plotted in lognormal distributions showed goodness-of-fit (GOF) indicated by correlation coefficient values
225 (R2) as shown in Figure 7 4. The mean values of influent and effluent TP concentrations were 0.262 and 0.058 mg/L respectively, indicati ng a reasonable TP removal achieved by VCF system. In contrast, the mean values of influent and effluent TDP concentrations were 0.045 and 0.025 mg/L respectively, which yielded a relative ly low removal of TDP as a consequence. It was reported that the event mean concentration for total phosphorus in urban runoff is 0.33 mg/L and dissolved phosphorus is 0.12 mg/L (USEPA 1 983) and Brown et al. (2003) reported that a total phosphorus concentration of 0.3 mg/L is adequate to de scribe both new and old urban development. However, it should be noted he re that the phosphorus concentration and partitioning are highly site specific. TP and TDP Removal Apparently one of the interests many people may have is the beha vior of phosphorus and treatment of phosphorus. Usually EMC values were not used in evaluation of process, but only tabulated as a nominal index, since an EMC ca nnot impart any information regarding process behavior; only endpoint information. Thus, cha nges in phosphorus loads were determined from influent and effluent concentra tions and flow data to quantify the performance of the system. The mass removals of TP and TDP for each event were also summarized in Figure 7 2 and Figure 7 3 respectively to provi de the insight of VCF system performance on several aspects: dissolved phosphorus removal th rough chemical adsorption and particulate phosphorus (PP) removal by physical filtration and gravitational se ttling. Results showed that the TP and TDP removals ranged from 46.8 to 91.1 % and from 46.9 to 90.5 %, respectively. Minton et al. (1996) claims that values for mean removal can be misleading as the range of removal rates for each individual event can be very large. Among 19 storm events monitored in this study, 13 events reached a TP removal above 70% and 12 events reached a TDP removal above 60%. In this respect, the VCF system with filled (AOCM)p illustrated fair removal capability for
226 dissolved phosphorus and particulate phosphorus in spite of various hyd raulic and hydrologic event characteristics including the flow rate, prev ious dry period, the duration and intensity of the storm, and the traffic frequency. Particular attention is genera lly being given to flow rate which has critical influence on phosphorus removal. As shown in Figure 7 2, re latively low percentage mass removals of TP (less than 70%) are observed in 16-June, 09-August, 07-May, 10-July, 04-August and 29-April events which generated consider ably high flow (greater than 100 L/min) compared to other events monitored in this study. Such a phenom enon does not happen to TDP removals presented in Figure 7 3, indicating the di fferent delivery patterns that TP and TDP followed. In general, dissolved phosphorus removal by solid media is more likely controlled by adsorption onto surface sites, which depends on surface area, pore size and polarity, and precipitation of inorganic phosphorus minerals from solution. On the other hand, particulate phosphorus removal is expected to correlate with the PM gradations as well as PM removals. Additionally, removal efficiency is highly related to the concentratio ns and chemical characteristics of the influent. However, no clear trend was found between infl uent concentration and phosphorus removal. TP and TDP Cumulative Mass The temporal transport of TP and TDP were examined for 16 events (rainfall depth greater than 2.54 mm) as shown in Figure 7 5 and Figu re 7 6. It should be noted here that the cumulative mass loading corresponding to the cumulative volume for each event describes the phosphorus mass loading along with elapsed time sin ce rainfall initiated (tim e 0). While there are some exceptions, general trend is found that mass loading of TP and TDP tend to increase rapidly with volume and then gr adually dropped till exhaustion as a function of elapsed time on event basis. This result also suggests TP and TDP transport exhibited first flush phenomena in most cases. The idea of a first flush is based on the premise that much of the material that
227 accumulates on the surfaces of urban environment during dry period is swept up in the first wave of runoff by a following rainfall (Characklis and Wiesner 1997; Sansal one and Cristina 2004). On the other hand, the strong first flush curve in 29 April 2006 event illustrated there is less amount of phosphorus contributed to the cumulativ e mass loading not only for TDP, but also for TP. In addition, the relative difference between cu mulative mass of influent and effluent for both TP and TDP also indicated the removal on a cumulative basis. Investigation into th e change of cumulative phosphorus mass with respect to the cumulative volume shown in those figures also re vealed a quicker exhaustion of TDP than that of TP in 29-April event, further indicating a different delivery pattern between dissolved phosphorus and particulate phosphorus. In addition to that, a strongest firs t flush was exhibited by TP and TDP generated from such a high inte nsity (71.37 mm) and long duration (165 min) event. However, increases of cumulative mass for TP and TDP along with cumulative volume to a lesser extent were observed for the relatively low intensity or short events in which either dissolved or particulate phos phorus was not washed off sufficiently. According to the quantitative criterion used for de fining the occurrence of a first fl ush (Sansalone and Buchberger 1997), such phenomena of TP and TDP were furt her demonstrated in general except the low flow and low volume events for TDP which were considered as flow-limited events compared to mass-limited events with high volumes. Similar pa tterns occurred with effluent TP and TDP, although less amount of phosphorus in terms of mass was discharged after a combination of unit operations and chemical processes (UOPs) includ ing sedimentation, filtration and adsorption. Removal of Phosphorus Associated with PM Ideally, TP was evenly contributed by the par ticulate matter as well as TDP. This would have provided a direct relation between the cumu lative mass of particulate phosphorus (PP) and cumulative PM mass. Mass removals of phosphorus associated with suspended, settleable and
228 sediment PM (PPsus, PPset and PPsed) were ex amined and plotted as a function of PM mass removals in Figure 7 7. There seems to be a trend that mass removal of PP followed a positive linear pattern as a function of PM mass removal in general fo r suspended, settleable and sediment fractions. The linear re lationship of mass removals betw een PP and PM is expected to be perfect (slope = 1) by assuming the uniform di stribution of particle size for each PM fraction. However, in reality, mass percentage of PM tends to present a gamma dist ribution as function of particle size and finer particle s usually provide higher specific surface area (SSA) for phosphorus binding. Results indicated that PPsus was found to have higher potential removal for given PM mass removal compared to PPset and PPsed, although the mean removals of phosphorus bound to settleable and sediment PM (76.5% and 88. 1%) are higher due to th e greater removal of settleable and sediment size particles in mass ( 91.2% and 97.7%) as summarized in Figure 7 7. An interesting phenomenon observed on variation of the linear patterns in Figure 7 7 suggests that the greater heterogeneous si ze range of settleable and sedime nt PM appears to be largely responsible for the pronounced di screpancy of phosphorus removals associated with settleable and especially sediment PM in response to the corresponding PM removal. In brief, particulate matter, which is an economically measurable constitu ent of rainfall runoff, could be utilized as a surrogate of P loading. Such a matter would be very useful in terms of effective prediction. Fractions of Phosphorus Associated with PM To determine the event mass fractions of PP, incremental values for PPsus, PPset and PPsed were examined from sample analysis and summed over the entire event based on incremental runoff volumes. Figure 7 8 provide s a comparison of PP mass fractions between influent and effluent for 19 storm events. The dashed linear line represents that the mass fractions of PPsus, PPset and PPsed in influent are identical to t hose in effluent. Results indicate that PP fractions increase in effluent as they incr ease in influent. Furtherm ore, PPsus in effluent
229 was observed to be more predominant than it wa s in influent and as a consequence, the mass percentages of PPset and PPsed in effluent were smaller than they were in influent. While the mean percentage of PPsus increased from 81.6 % in influent to 88.1% in effluent, the percentage of PPset and PPsed dropped from 12.2 % to 8. 5 % and from 3.1 % to 1.7 % on average for 19 events, respectively. This phenomenon suggests th at to effectively remove PP and TP, the unit operation and processes (UOPs) faci lities must be able to remove the suspended particles which are less than 25 m. Phosphorus Specific Capacity (PSC) Phosphorus specific capacities (PSC) on particulate matter (mg/g) on individual sample basis were categorized into five groups (10-100, 1-10, 0.1-1, 0. 01-0.1 and < 0.01) based on their log linear scale. Categorical data consisted of frequency counts of PSCs occurring in the response categories, which were summarized in the two-way contingency table shown in Table 7 3. Generally, majority of PSC s on overall particulate matter were in the range of 0.01 to 10 mg/g. While suspended PSCs were predominant in the range of 0.01 to 10 mg/g, settleable and sediment PSCs were mainly in the range of 0.01 to 1 mg/g. As a categorical analysis tool, odds ratio (OR) was utilized to compare the relative odds of high PSC in each PM fraction. An odds ratio of 1.0 implies the perfect similarity between the two PM fractions. The more the ratio deviated from 1.0, the greater the dissimilarity. For an example, two groups were separated by a critical level of PSC at 10 mg/g, greater than 10 mg/g (high PSC) and less than 10 mg/g (low PSC) as shown in Table 7 3. Suspended frac tion has an approximately 6% chance of having high PSC and settleable fraction has less than 1% chance of having high PS C, equivalent to the odds of 0.055 (23 to 417) and 0.002 (1 to 439) resp ectively. Thus, the odds ratio of 24.21 implies that the odds of identif ying with suspended fraction over sett leable fraction is approximately 23 times higher in high PSC group than with lo w PSC group. Similar result was found for
230 suspended fraction over sediment fraction, but it appeared that no diffe rence between settleable fraction and sediment fraction wi th an odds ratio of 0.99. The odds ratios for other different groups using 1, 0.1 and 0.01 mg/g as critical leve l of PSC were also summarized in Table 7 3. In general, those results indi cated that significan t amount of phosphorus were likely to bind suspended PM over settleable and sediment PM. In addition, the capability of settleable PM was found to be slightly (1-2 times ) higher than sediment PM regarding the phosphorus binding. These phenomena further supported the results above that the efficient rem oval of TP, especially PP required an efficient removal of su spended PM with size down to 25 m. Conclusion This study examined the in-situ removal of TD P, TP as well as the PP associated with suspended, settleable and sediment PM as f unction of hydraulic and hydr ological characteristics of nineteen storm events monitored on the basis of representative manual sampling and appropriate phosphorus analysis. Such a study would help unde rstanding how particulate bound fraction of phosphorus, which is a predominant ove r dissolved fraction, wa s transported in the urban rainfall-runoff events, evaluating the potentia l fate of TP, TDP and PP as well as shedding sight on effective in-situ cont rol of phosphorus which was co nsidered a major cause of eutrophication. A number of conc lusions have been drawn from th e results of the present study. Transport of particulate bound P during the urban rainfall runoff is a complicated phenomenon. As the primary driving force, hydrol ogical transport is considered as the most contributing factor to the exponential decay of TP and TDP, especially in the high intensity events. Such phenomena may occur in a less exte nt when the event has a short antecedent dry periods followed by a heavily previous washoff. In this case, the subsequent events tend to have relatively low EMCs and mass removal.
231 The separation effectiveness of the VCF sy stem on particulate-bound phosphorus is highly related to the particulate matte r removal for each particle-size fraction. Phosphorus associated with settleable and sediment tends to have higher removal on event basis, but PM Phosphorus associated with suspended PM has higher potenti al removal for given PM mass removal. Such a manner makes it possible to economically pred ict P loading through measuring the more assessable particulate matter and also facili tates better unit operation and process (UOP) selection. Approximately one order of magnitude highe r of PSC for phosphorus associated with suspended solids indicated that significant amou nt of phosphorus were likely to bind suspended PM over settleable and sediment PM. Therefore, for particulate phosphorus removal suspended particles were considered as a major concern regarding the treatment unit performance. This phenomenon suggests that to effectively rem ove PP and TP, the unit operation and processes (UOPs) facilities are required to be able to remove the suspended particles which are less than 25 m. The P removal mechanisms by VCF system are a combination of sedimentation, filtration and chemical adsorption. In general, the VCF system with filled (AOCM)p illustrated fair removal capability over a long period for disso lved phosphorus and part iculate phosphorus in spite of various hydraulic and hydrologic conditions.
232 Table 7-1. Summary of cartridge and media information utilized in Volumetric Clarifying Filter system. Cartridge information Media information Number 5Diameter (mm) 3.56 Height (mm) 558.8 s (g/cm3) 2.35 Diameter (mm) 450.9 t 0.705 Volume (L) 73.0 i 0.371 Mesh size (mm) 2.1m (kg) 49.80 Datum 0 Invert of VCF system Bottom of cartridge 584.2 mm Note: s t i m represent specific gravity, total por osity (including bed porosity), internal porosity and mass of the media.
233 Table 7-2. Summary of hydrological indices for 19 storm events treated by Volumetric Clarifying Filter system on a sma ll watershed in Baton Rouge, LA. I PDH Qave Qmax V TD SLR RT CT Event (mm) (hr) (L/s) (L/s) (L) (min) L/min-m2 min min 21 April 4.06 927 1.2413.32927 6521.9 11.1 4.7 29 April 71.37 84 5.0825.348306 168169.9 1.5 0.8 06 May 1.02 157 0.070.3495 7515.5 20.7 11.4 07 May 6.35 21 2.469.13852 63104.7 5.5 2.1 27 May 3.56 482 1.816.52628 4598.9 6.2 2.6 28 May 3.05 23 1.547.42096 4077.0 6.7 2.4 16 June 14.22 451 2.0614.69938 79107.1 2.9 1.0 19June 3.05 79 1.014.31816 4559.5 7.5 4.1 04 July 3.81 352 1.156.22779 6249.3 17.3 8.5 05 July 5.59 25 0.997.93838 7554.5 6.7 1.9 09 July 1.78 69 0.311.6674 9216.1 16.5 9.4 10 July 30.48 16 4.261525189 105165.5 1.2 0.6 14 July 5.59 89 0.793.83304 7936.6 7.6 4.3 16 July 2.54 45 0.832.8945 10075.4 10.8 6.4 18 July 2.03 26 0.313.31047 6921.8 15.8 9.2 04 August 52.32 191 4.5020.336990 120187.9 3.7 1.8 05 August 9.14 19 1.269.06421 7155.8 5.7 2.1 07 August 8.89 25 1.4510.76022 4775.1 4.7 1.7 09 August 16.26 23 2.0414.812502 8997.2 3.6 1.2 Note: I, PDH, Qave, Qmax, V represent rainfall depth, previous dry hours, mean flow rate, peak flow rate and runoff volume on event basis. TD, SLR, RT, CT represent th e duration of the event, event mean surface loading rate, event mean re sidence time and event mean contact time on event basis.
234 Table 7-3. Phosphorus specific capacity (PSC) frequency and odds ratio for suspended, settleable and sediment fractions. PSC Frequency Conditional probability for PSC (mg/g) Suspended Settleable Sediment Suspended Settleable Sediment 10-100 23 110.06< 0.01 < 0.01 1-10 294 69350.670.17 0.08 0.1-1 120 2551670.270.58 0.38 0.01-0.1 2 114198< 0.010.25 0.45 < 0.01 1 138< 0.01< 0.01 0.09 440 4404391.001.00 1.00 Mass ratio Category frequency Odds ratio (mg/g) Suspended Settleable Sediment Sus/Set Set/Sed Sus/Sed > 10 23 11 < 10 417 439438 24.210.99 24.16 > 1 317 7036 < 1 123 370403 13.622.12 28.85 > 0.1 437 325203 < 0.1 3 115236 51.543.29 169.35 > 0.01 439 439401 < 0.01 1 138 1.0041.60 41.60
235 Figure 7-1. Plan view of experimental site and Volumetric Clarifying Filter (VCF) system. 10.2 cm PVC pipe from west Data logger 5.1 cm Parshall Flume Watershed PCC pavement 1088 m2 (2 x 544-m2) 2% surface slope ADT=142,000 (E+W) Tee 212.1 cm 45.7 cm Effluent 1 2 Influent 9 3 4 5 6 7 8 10 11 116.8 cm 1 Influent box 2 Radial flow cartridge 3 Baffle 4 Vault drainage pipe 5 Float valve 6 Effluent drop box 7 Orifice 8 Effluent pipe 9 Influent delivery pipe 10. Effluent V-notch weir 11. Effluent drainage pipe
236 06-May 18-Jul 09-Jul 16-Jul 14-Jul 05-Jul 19-Jun 04-Jul 05-Aug 21-Apr 07-Aug 28-May 27-May 16-Jun 09-Aug 07-May 10-Jul 04-Aug 29-Apr TP mass removal (%) 0 20 40 60 80 100 Flow rate (L/min) 0.01 0.1 1 10 100 1000 Flow rate TP removal (%) 0.001 0.01 0.1 1 10 0.001 0.01 0.1 1 10Effluent TP [mg/L] Influent TP [mg/L] Effluent TP Influent TP Figure 7-2. The coupled relationship between flow rate, influent and effluent TP conc entration and mass remo val for VCF in a sm all watershed on event basis for 2006 events.
237 06-May 18-Jul 09-Jul 16-Jul 14-Jul 05-Jul 19-Jun 04-Jul 05-Aug 21-Apr 07-Aug 28-May 27-May 16-Jun 09-Aug 07-May 10-Jul 04-Aug 29-Apr TDP removal (%) 0 20 40 60 80 100 Flow rate (L/min) 0.01 0.1 1 10 100 1000 Flow rate TDP removal (%) 0.001 0.01 0.1 1 10 0.001 0.01 0.1 1 10 Effluent TDP Influent TDP Influent TDP [mg/L] IEffluent TDP [ m g /L ] Figure 7-3. The coupled relationship between flow rate, influent and effluent TDP c oncentration and mass removal for VCF in a small watershed on event basis for 2006 events.
238 Probability density function (pdf) Frequency of occurency (%) 0 5 10 15 20 25 30 Influent TP frequency Influent TP pdf Effluent TP frequency Effluent TP pdf R2p 0.122 0.032 0.930.89 > 0.05> 0.05 E(X) S(X) mg/L mg/L mg/L mg/L n224243 E(X) : Mean S(X) : Standard deviation 0.262 0.058 InfluentEffluent PO4-P [mg/L] 0.0010.010.1110100 Probability density function (pdf) Frequency of occurency (%) 0 5 10 15 20 25 Influent TDP frequency Influent TDP pdf Effluent TDP frequency Effluent TDP pdf0.00 0.05 0.10 0.15 0.20R2p 0.038 0.020 0.770.98 > 0.05> 0.05 E(X) S(X) mg/L mg/L mg/L mg/L n224243 E(X) : Mean S(X) : Standard deviation 0.045 0.025 InfluentEffluent0.25 0.00 0.05 0.10 0.15 0.20 0.25 0.30 Figure 7-4. Log-normal distribut ions of influent and effluent TP and TDP concentrations for total 19 storm events.
239 29-April 01000020000300004000050000 TP, TDP mass (mg) 0 500 1000 1500 2000 Influent TP Effluent TP Influent TDP Effluent TDP 21-April 050010001500200025003000 TP, TDP mass (mg) 0 1250 2500 3750 5000 27-May 050010001500200025003000 TP, TDP mass (mg) 0 300 600 900 1200 07-May 01000200030004000 TP, TDP mass (mg) 0 375 750 1125 1500 16-June 0200040006000800010000 TP, TDP mass (mg) 0 700 1400 2100 2800 28-May 0500100015002000 TP, TDP mass (mg) 0 120 240 360 480 04 July volume (L) 050010001500200025003000 TP, TDP mass (mg) 0 300 600 900 1200 19-June volume (L) 0500100015002000 TP, TDP mass (mg) 0 100 200 300 400 Figure 7-5. Mass delivery of TP and TDP in influent and effluent runoff volume on accumulative basis for eight urban rainfall-runoff events (from April 21 to July 04, 2006) treated by VCF system.
240 10-July 0500010000150002000025000 TP, TDP mass (mg) 0 800 1600 2400 3200 Influent TP Effluent TP Influent TDP Effluent TDP 05-July 01000200030004000 TP, TDP mass (mg) 0 350 700 1050 1400 16-July 02004006008001000 TP, TDP mass (mg) 0 70 140 210 280 14-July 0100020003000 TP, TDP mass (mg) 0 250 500 750 1000 05-August 0200040006000 TP, TDP mass (mg) 0 250 500 750 1000 04-August 010000200003000040000 TP, TDP mass (mg) 0 1000 2000 3000 4000 09-August volume (L) 030006000900012000 TP, TDP mass (mg) 0 200 400 600 800 07-August volume (L) 0100020003000400050006000 TP, TDP mass (mg) 0 400 800 1200 1600 Figure 7-6. Mass delivery of TP and TDP in influent and effluent runoff volume on accumulative basis for eight urban rainfall-runoff events (from July 05 to August 09, 2006) treated by VCF system.
241 Particulate Phosphorus removal (%) 50 60 70 80 90 100 Suspended 50 60 70 80 90 100 Settlable Particulate matter removal (%) 5060708090100 50 60 70 80 90 100 Sediment Sus P Sus PMmin max s 65.6 69.2 10.2 8.9 48.5 52.4 91.0 88.5 Set P Set PMmin max s 76.5 91.2 10.3 4.6 54.1 81.0 94.6 96.7 Sed P Sed PMmin max s 88.1 97.7 9.0 1.7 64.5 93.7 96.2 99.7 High potential but high energy Low potential but low energy Figure 7-7. The relationship between PM re moval and particulate phosphorus removal for suspended, settleable and sediment fractions on event basis.
242 Influent particulat e TP fraction (%) 020406080100 Effluent particulat e TP fraction (%) 0 20 40 60 80 100 Influent Effluent 3.1 1.7 1.7 0.9 12.2 4.6 8.5 4.0 81.6 6.3 88.1 4.8 Sediment TP Settleable TP Suspended TP n = 19 (%) (%) Figure 7-8. Particulate phosphorus fractions associated with susp ended, settleable and sediment PM for influent and effluent ba sed on mass for 19 storm events.
243 CHAPTER 8 GLOBAL CONCLUSIONS This dissertation investigated the toxicity of particulate matter (PM) and metals transported in urban rainfall-runoff generated from a small instrumental MS4 watershed located in Baton Rouge, Louisiana. A volumetric clarifying filter system was tested, it consisted of five radial flow cartridges filled with engineered media (AOCM)P and it was installed into a steel vault in parallel. Its removal mechanisms were examin ed and evaluated it for removal efficiency associated with particulate matter (PM) a nd phosphorous. This was done with respect to examining methods of toxicity re duction in rainfall-runoff and al so because of requirements in the National Pollutant Discharge Elimination System Stormwater Phase II. Results indicated that urban rainfall-runoff can exhibit acute toxicity depending on the source, storm characteristics, timing during the storm, and the overall drainage design. Generally, fine particles contribut e most to the lethal and sublethal toxicity due to their larger specific surface area which allows more contaminants, including metals, to bind with them. In addition, fine solids can remain suspended in the water column, which increases organisms exposure time. The larger particles are considered to exhibit a chronic effect after they settled. It is generally accepted that larger particles (>75 m) play a minor role in the determination of acute toxicity. Additionally, sampling efforts focuse d on the first flush of a rainfall-runoff event are likely to produce more toxic responses, but volumes and durations of such flows are relatively small. In this study, early life stage organisms were more sensitive to runoff samples due to both possible multiple routes for toxicant entry, a nd the young organisms incompletely developed elimination system. The epithelial tissue in gills of fathead minnows were found to trap suspended particles less than 25 m, which resu lted in mucus precipitation on the gills surface
244 at sufficiently high concentrati on of TSS (300 mg/L), even though these particle sizes are not generally acutely toxic to juveni le and adult stage fish. As a consequence, the observed increase in the oxygen consumption rate of juvenile fath ead minnows when exposed to storm water runoff tests might be associated with both particulate and dissolved fractions. This is because a dynamic partitioning process occurred continuously between dissolved and particulate phases and it is very difficult to isolate them or evaluate each individually. The total site mean concentrations of copper, lead, cadmium and zinc exceeded acute and chronic critical values recommended by the Loui siana Department of Environmental Quality (LDEQ) and the Ohio Environmental Protection Agency (OEPA) for I-10 site in Baton Rouge and I-75 site in Cincinnati, respectively. In add ition, nearly all dissolved metal levels are above the acutely toxic levels for both sites, which indi cate the potential toxic impact of the runoff, and suggest that efforts are required to minimize its negative effect. Fu rthermore, pH is an important factor in determining metal speciation and in the typical pH range of urban rainfall runoff (6.5 7.5), cadmium and zinc would receive more atte ntion in terms of their bioavailability, although free ion species are found to be dominant. In ge neral, the one-compartment model describing the accumulated metal fraction stored shows an exponen tial rise over nine days of exposure to each metal species at the recommended concentrations based on the water quality characteristics of the site. On the other hand, copper was shown to be the most toxic among the three metals tested individually with green algae, and th e toxicity order based on the 48-hr EC50 was: Cu > Cd > Zn. In addition, the observed toxic effect for tested binary metal mixtures (Cu-Zn, Cu-Cd, Cd-Zn) was substantially higher than that predicted by as suming a simple additivity of toxic effects for individual metals.
245 Before the volumetric clarifying filter system testing was conducted, a series of full-scale single radial flow cartridge tests was carried out to examine the filtration and adsorption performance on both an engineered media (AOCM)P and a control media (perlite) of similar media size. From this study, solid mass removal e fficiency for the tested Sil-co-Sil 106 gradation was found to follow a decreasing trend under the sp ecified loading conditi ons as the operating flow rate increased. Filter head loss was found to be directly proportional to velocity in media filters which had new and clean bed before each r un. In addition, filter head loss was also found to be very dependent on porosity and media size, and reduction in porosity and media size causes the head loss increase. Another important aspect is that lower residence time associated with higher flow rate was not able to provide suffi cient chemisorption under the specific condition in this study. Conversely, the chemical adsorption b ecame more efficient when flow rates were kept lower. It should be noted here that more phosphorus was trapped by th e bottom layer of media than the top layer, which indicated uneven ve rtical distribution of phosphorus adsorption across the cartridge. This also suggested that a nonunifo rmly distributed flow occurred in the test chamber and cartridge as well. In contrast, with perlite the total phosphor us removals at three operating flow rates were less than 10% in genera l. This proved that th e conventional media may not be efficiently used in phosphorus reduction, especially for the dissolved portion when compared to engineered media like (AOCM)P used in previous tests with TP and TDP removal over 51%. Installation of a volumetric fi lter system proved to be a feasible way to address and controll urban runoff due to its potential specific ity and ease of placement. This system was effective at removing an average of 89.4% of the incoming SSC; 77.1% of this removal was contributed to by gravitational settling and 22.9% by filtration. Uneven distribution and filtration
246 occurred from the top to the bot tom of the cartridge. Removal effi ciency is highly related to the concentration and load of the influent. Standa rds are based on removal efficiency, which may penalize cleaner sites, and they do not take into account effect s on the receiving water quality. For these situations, a target outlet maximum concentration may be more appropriate. Although the suspended solids removed by filtration were le ss than 25%, the filtration still played an important role. Reducing the concentration of su spended solidsand the turbidity of effluent to close to target levels is a necessary proce ss. Yet its also necessary to reduce the potential toxicity that results from the high specific surface area of fine particles and the associated metals and toxicants associated with these particles. The comparison between manual sampling a nd auto sampling revealed that manual sampling proved to be the most conservative me thod, while autosampling was unable to sample the entire gradation of particles, especially large grain sized partic les. Further, employing automatic peristaltic pumps to draw samples unde restimated the total so lids. In this respect, particular attention needs to be given to the type of sampling method chosen when particulate matter or pollutant associated with particulate matter is targeted. Comparing the SSC method to the traditional total suspended so lids method, turbidity (NTU), T SS and settleable and sediment solids (SSC-TSS) follow a discernable log linear pattern, and it was found that settleable and sediment solids may contribute less to turbid ity compared to suspended solids. Correlation between them tends to be highly site specific and may vary within the same storm due to compositional variations. This study also examined the in-situ removal of TDP, TP as well as the PP associated with suspended, settleable and sediment PM as a function of hydraulic and hydrological characteristics of nineteen monitored storm ev ents. Hydrological transport, as the primary
247 driving force, is considered to be the most infl uential factor to the e xponential decay of TP and TDP, especially in high intensity events. Such phenomena may occur to a less extent when the event has a short antecedent dry period followed by a heavy washoff prior to the next event. In this case, the subsequent events tend to have relatively low EMCs and mass removal. The separation effectiveness of th e VCF system on particulate-bo und phosphorus was highly related to the particulate matte r removal for each partic le-size fraction. Phosphorus associated with settleable particles and sediment tended to have a higher removal efficiency on an event basis, but PM Phosphorus associated with suspended PM had a higher potential removal for given PM mass removal. Such an association makes it po ssible to economically predict P loading through measuring the more assessable particulate matter, and it also facilitates better unit operation and process (UOP) selection. Approximately one order of magnitude highe r of PSC for phosphorus associated with suspended solids indicated that a significantl y greater amount of phosphorus was likely to bind with suspended PM in comparison to settleable and sediment PM. Therefore, removals of suspended particles are thought to be a major concern regarding the treatment unit performance when removal of particulate phosphorous is a concern. This binding phenomenon suggests that to effectively remove PP and TP, the unit operati on and processes (UOPs) facilities should be required to remove the suspended particles that are less than 25 m. In general, the VCF system with filled (AOCM)p illustrated a fair removal capability over years for dissolved phosphorus and particulate phosphorus, in spite of various hydrau lic and hydrologic conditions. Generally, events and constituent phase transpor t behavior can be classified as either masslimited (first-order transport) or flow-limite d (zero-order) based on volume. But different constituents and constituent pha ses could result in different tr ansport behavior. Results have
248 significant implications for decisi on making with respect to volume tric capture and water quality volume (WQV) determinations. For high volume stor m events (> 4000 L), there is an asymptotic approach to a maximum runoff coef ficient for each storm that typi cally fell between 0.6 and 0.8. For the low volume events in which total runoff volumes were less than 4000 L, a lower value for the runoff coefficient was found within the range from 0.4 to 0.6. During many events, a sudden spike was found in head loss caused by some resistance to flow the porous system initially because the media starts as dry and it already has some suspended solids trapped on from previous stor m. The micro-porosity in the media during the storm may not be very effective because the conv ective fluxes cannot access those fine pores in a short amount of time while the water is in the f ilter cartridges. In contrast, only macro-porosity was available for storm. In addition, a high su rface loading rate usually resulted in a low residence and contact time, indi cating the lower removals due to the shorter time for coarser particles to settle and less time for dissolved po llutants to react with the surface of the media. The redox change as a function of residen ce time was in agreement with the common understanding that rainfall-runoff collection sy stems or treatment operation/process units undergo a transformation to anoxic/anaerobic co nditions when extended periods of residence time (hours to days) often occur. Redox was found to change as a function of residence time, and a significant drop occurred in the first 24 hours. Concomitantly, consistent increase of ammonia concentration was identified while nitrate con centration exhibited an opposite (decreasing) pattern as a function of residenc e time. However, there was no significant change for phosphorus species distribution in this case. In general, this volumetric clarifying filter sy stem proved to be a feasible water chemistry and particulate treatment system for SSC, TP, TD P and associated toxicity reduction but not a
249 hydrologic restoration system. Such systems must provide hydrologic restor ation if our coupled urban anthropogenic cycles of chemistry and hydrology are to be restored. In addition, management is required to address misbehavior such as scour and repartitioning. Ultimately, hydrologic restoration and polluta nt source control is required, given that this level BMP management to ensure performance is simply not sustainable from a long term point of view.
250 LIST OF REFERENCES Acemioglu, B. (2005). Batch kinetic study of sorption of methylene blue by perlite. Chem. Eng. J., 106:73-81. Allan, R. J. (1986). The role of particulate matte r in the transport and buria l of contaminants in aquatic ecosystems. In: Hart, B.T. (Ed), The Role of Particulate Matter in the Transport and Fate of Pollutants. Water Studies Center, Chisholm Institute of Technology, Melbourne, pp. 1-55. Allen, T. (1990). Particle Size Measurement, 4th ed. Chapman & Hall, 125-127. Allison, R. A., Walker, T. A., Chiew, F. H. S., ONeill, I. C., and McMahon, T. A. (1998). From roads to rivers gross pollutant removal from urban waterways. Cooperative Research Centre for Catchment Hydrology Report, 46. American Society Testing and Materials. (1993). Standard Practice for Dr y Preparation of Soil Samples for Particle Size Analysis and Determination of Soil Constants. ASTM D 421-85, West Conshohocken, Pa. Vol. 04.08, 8-9. American Public Health Association. (1995). Standard methods for the Examination of Water and Wastewater, 19th ed, American Public Health Association, Washington. DC. American Society Testing and Materials. (1997). Standard test methods for determining sediment concentration in water samples. ASTM Designation: D-397797, West Conshohocken, Pennsylvania. American Public Health Association. (1998). Standard methods for the examination of water and wastewater, 20th Ed., A.D. Eaton, L.S. Cles ceri, and A.E. Greenberg, eds., Water Environment Federati on, Washington, D.C. Andoh, R. Y. G., and Saul, A. J. (2003). The use of hydrodynamic vortex separators and screening systems to improve water quality. Water Sci. Technol., 47(4), 175. Appel, P. L., and Hudak, P. F. (2001). Automat ed sampling of stormwater runoff in an urban watershed, north-central Texas. J. Envir. Sci. Health., A36(6), 897. Barrett, M. E., Zuber, R. D., Collins, E. R., Mali na, J. F., Jr., and Charbeneau, R. J. (1993). A review and evaluation of literat ure pertaining to the quantity and control of highway runoff and construction. Technical Report CRWR 239. Center for Research in Water Resources, University of Texas at Austin. Bedient, P. B., and Huber, W. C. (1992). Hydrology and Floodplain Analysis. Addison Wesley Publishing, Massachusetts. Bent, G. C., Gray, J. R., Smith, K. P., and Gl ysson, G. D. (2000). Measuring sediment in highway runoff. U.S. Geological Survey Open File, Rep. 00-497, 51.
251 Berkheiser, V. E., Street, J. J., Rao, P. S. C., and Yuan, T. L. (1980). Partitioning of inorganic orthophosphate in soilwater systems. CRC Crit. Rev. Envir. Control. 10, 179-224. Breault, R. and Granato, G. (2000). A synops is of technical issu es of concern for monitoringtrace elements in highway and urban runoff. U.S. Geological Survey Open File Report 00-422, Northborough, Massachusetts. Brezonik, P. L., King, S. O. and Mach, C. E. ( 1991). The influence of water chemistry on trace metal bioavailability and toxic ity to aquatic organisms. In: Metal ecotoxicology. Concepts and applications. Newman, M.C. and McIntosh, A.W., Eds., Lewis Publishers, Michigan. 399. Brombach, H., Xanthopoulos, C., Hahn, H. H., a nd Pisano, W. C. (1993). Experience with vortex separators for combined sewer overflow control. Water Sci. Technol., 27 (5-6), 93104. Brown, T., Schueler, T., Wright, T., Winer, R., and Zielinkski, J. (2003). Maryland Chesapeake and Atlantic Coastal Ba ysCritical area 10% rule guidance manual, Center for Watershed Protection, Ellicott City, Md. Buckler, D. R., and Granato, G. E. (1999). A ssessing biological eff ects from highway-runoff constituents. U.S. Geological Survey Open File Report 99-240, 45. Butler, D., and Davies, J. W. (2000). Urban Drainage. E & FN Spon, London, UK. Campbell, P. G. C. (1995). Interactions between trace metals and aquatic organisms: A critique of the free-ion activity model. In: Metal speciation and bioavaila bility in aquatic systems, John Wiley and Sons Ltd., 45-102. Characklis G. W., Wiesner, M. R. (1997). Particles, metals, and water quality in runoff from large urban watersheds. J. Environ. Eng., 123, 753-759. Charlatchka, R., and Chamber, P. (2000). Influ ence of Reducing Conditions on Solubility of Trace Metals in Contaminated Soils. Water, Air, Soil Pollut., 118 (1-2), 143-167. Clark, S. C., Lawler, D. F., and Crushing, R. S. (1992). Contact Filtration: Particle Size and Ripening. Journal of American Water Works Association. 84(12), 61-71. Clark, S. E. (2000). Urban stormwater filtratio n: Optimization of design parameters and a pilot-scale evaluation. PhD dissertation, Univ. of Alabama at Birmingham, Birmingham, Ala. Clark, S., and Pitt. R. (1999). Stormwater Runoff Treatment: Ev aluation of Filtration Media. EPA-600/R-00/010, U.S. Environmental Prot ection Agency, Water Supply and Water Resources Division, National Risk Manageme nt Research Laboratory, Cincinnati, OH.
252 Compton, J., Mallinson, D., Glenn, C. R., F ilippelli, G., Follmi, K., Shields, G., and Zanin, Y. (2000). Variations in the global phosphorus cycle, SEPM Special Publication, 66, 21. Cooke, S. S. (1991). The effects of urban stormwater on wetland vegetation and soils-A long-term ecosystem monitoring study. In Puget Sound Research Proceedings, Seattle, Washington, 43-51. Correll, D. L. (1998). The role of phosphorus in the eutrophication of receiving waters: a review. J. Environ. Qual., 27, 261-266. Cristina, C. M., and Sansalone, J. J. (2003). Fir st Flush, Power Law and Particle Separation Diagrams for Urban Storm-wate r Suspended Particulates. J. Environ. Eng.. 129(4), 298-307. Cristina, C. M., and Sansalone, J. J. (2003). Kin ematic wave model of urban pavement rainfallrunoff subject to traffic loadings. J. Environ. Eng., 129 (7), 629-636. Davis, J. A., Kent, D. B., Rea, B. A., M aest, A. S., Garabedian, S. P. (1993). In Metals in groundwater; Allen, H. E., Perdue, E. M., and Brown, D. S., Eds., Lewis Publishers, Chelsea, 223-273. Dean, C. M., Sansalone, J. J., Cartledge, F. K., and Pardue, J. H. (2005). Influence of hydrology on storm water metal element speciation at the upper end of the urban watershed. J. Environ. Eng., 131(4), 632. Deneer, J. W., Sinnige, T. L., Seinen, W., and Herm ens J. L. M. (1988). The joint acute toxicity to Daphnia magna of industrial organic chemical s at low concentrations. Aquat. Toxicol., 12: 33. Devore, J. L. (1991). Probability and statistics for engineering and the sciences, 3rd Ed., Brooks/Cole Publishing Co., Pacific Grove, CA. Dumas A., and Bergheim A. (2001). Effluent Tr eatment Facilities and Methods in Fish Farming: a Review. Bullettin of Aquaculture Association Canada, 100, 33-38. Environmental Protection Agency. (1993). Handbook of Runoff Pollution, 6. Office of Research and Development, Washington, D.C. Environmental Protection Agency. (1991). Technical Support Document for Water Qualitybased Toxics Control, U.S. Environmental Protection Ag ency, Office of Water, EPA/505290-001. Erickson, A. J., Gulliver, J. S., and Weiss, P. T. (2007). Enhanced sand filtration for storm water phosphorus removal. J. Environ. Eng., 133 (5), 485-497.
253 Farizoglu, B., Nuhoglu, A., Yildiz, E., and Keskin ler, B. (2003). The performance of pumice as a filter bed material under rapid filtration conditions. Filtration + Separation. Sep. 40, 41 46. Foster, I. D. L., and Charlesworth S. M. (1996). Metal species in the hydrological cycle: Trends and explanation. Hydrolog. Process., 10, 227. Galicki, S., Johnson, A. and Williams, A. (2003) Final report, Sediment removal from stormwater runoff AbTech industr ies Ultra-Urban Filter Series in laboratory flume tests, Millsaps College. Garn, H. S. (2002). Effects of lawn fertilizer on nu trient concentration in runoff from lakeshore lawns, Lauderdale Lakes, Wisconsin. Water-Resources Inves tigations Report 02-4130. U.S. Geological Survey. Garton, E. R. (1977). The effects of highway c onstruction on the hydrogeologic environment at Bowden, West Virginia. Hydrologic Problems in Karst Regions, Dilamarter, R. R., and Csallany, S. C., Eds. Western Kent ucky University, Bowling Green, 439-449. Gippel, C. J. (1995). Potential of turbidity moni toring for measuring the transport of suspended solids in streams, in Hydrological Processes 9, 83-97. Gjessing, E., Lygren, E., and Anderson, S. (1984). Acute toxicity and chemical characteristics of moderately polluted runoff from highways. Sci. of the Total Envir., 15(33), 225-232. Glenn, D W. III., Liu, D. F., and Sansalone, J. J. (2001). Influence of highway runoff chemistry, hydrology and residence time on nonequilibrium part itioning of heavy metals implications for treatment at highway shoulder. Transportation Research Record, 1755, 129-140. Gray, J. R., Glysson, G. D., and Turcios, L. M. (2000). Comparability an d reliability of total suspended solids and suspended-sediment concentration data. U.S. Geological Survey Water-Resources Investigations Report 00-4191, 14. Grizzard, T. J., Randall, C. W., Wea nd, B. L. and Ellis, K. L. (1986). Effectiveness of extended detention ponds, urban runo ff quality impacts and qua lity enhancement technology, Edited by B. Urbonas and L. Roesner, ASCE, New York, NY. Grobler, E., Vuren, J. H., Dupreez J. U. ( 1989). Routine oxygen consumption of Tilapia sparrani (Cichilidae) followi ng acute exposure to atrazine. Comp. Biochem. Physiol. 93C(1), 37-42. Gunduz, L., Sariisik, A., Tozacan, B., Davraz, M., Ugur, I., and Cankiran, O. (1998). Pumice Technology, Vol. 1. ISBAS A.S. and SDU College of Engineering, Isparta, Turkey. Gustafsson, O., and Gschwend, P.M. (1997). Aqu atic colloids: concepts, definitions, and current challenges. Limnology and Oceanography., 42(3), 519-528.
254 Hamilton, R. S., and Harrison, R. M. (1991). Highway pollution, Elsevier, New York. Han, J. S., Hur, N., Choi, B., and Min, S. H. (2003). Removal of phosphorous using chemically modified lignocellulosic materials. In: 6th InterRegional Confer ence on EnvironmentWater, Land and Water Use Planning and Management, Albacete, Spain, 1. Hano, T., Takanashi, H., Hirata, M., Urano, K., and Eto, S. (1997). Removal of phosphorus from wastewater by activated alumina adso rbent. Wat. Sci. Tech., 35(7), 39-46. Heath, A. G. (1987). Water pollution and fish physiology. CRC Press, Inc., Boca Raton, Florida. Henderson, J. P. and Bromage, N. R. (1988). Optimising the removal of suspended solids from aquaculture effluents in settlement lakes. Aquacultural Engineering, 7, 167-188. Herbes, S. E. and Beauchamp, J. (1977). Toxic in teraction of mixtures of two coal conversion effluent components (resorcino l and 6-methylquinoline) to Daphnia magna. Bull. Environ. Contam. Toxicol., 17, 25. Hipp, J. A., Ogunseitan, O., Lejano, R. and Smith C. S. (2006). Optimization of stormwater filtration at the urban/watershed interface. Environ. Sci. Technol. (40), 4794-4801. Holtan, H., Kamp-Nielsen, L., and Stuanes, A. O. (1998). Phosphorus in soil, water and sediment: an overview. Hydrobiologia, 170: 19-34. Horner, R. R., Skupien, J. J., Living ston, E.H., and Shaver, H. E. (1994). Fundamentals of Urban Runoff Management: Technica l and Institutional Issues. Terrene Institute, Washington, D.C. Howell, R. (1985). Effect of zinc on cadmium toxici ty to the amphipod Gammarus pulex. Hydrobiologia, 123, 245. Huber, W. (1993). Contaminant Transport in Surface Water, D. R. Maidment, ed., Handbook of Hydrology, McGraw-Hill Inc., New York, NY., 14.1-14.50. Igloria, R. A., Hathhorn, W. E., and Yonge, D. R. (1997). NOM and trace metal attenuation during storm water infiltration. J. Hydrol. Engrg., 2(3), 120. Innman, D. L. (1952). Measures for descri bing the size distribution of sediments. Journal of Sedimentary Petrology, 22(3), 125-145. James, R. (1999). Solids in Storm Wate r Runoff. Water Resources Management. http://www.stormwater-resources.co m/library/070BMeasuringSolids.doc James, R., Sampath, K., and Selvamani, P. ( 1998). Effect of EDTA on reduction of copper toxicity in Oreochromis mossambicus. Bull. Environ. Contam. Toxicol. 60, 487-493. Jenkins, D., Ferguson, J. F., and Menar, A. B. (1971). Chemical processes for phosphate removal. Water Res., 5, 369.
255 Johnson, W., Chen, S. (2006). Performance Evalua tion of Radial/Vertical Flow Clarification Applied to Recirculatin g Aquaculture Systems. Aquacultural Engineering, 34(1): 47-55. Kargin, F. and Cogun, H. Y. (1999). Metal inte ractions during accumulation and elimination of zinc and cadmium in tissues of the freshwater fish, Tilapia nilotica. Bull. Environ. Contam. Toxicol. 63, 511-519. Kim, J., Pathapati, S., Liu, B. and Sansalone, J. J. (2007). Treatment and Maintenance of Stormwater Hydrodynamic Separators: A case Study. EWRI 2007 proceeding paper. Kime, D. E., Ebrahimi, M., Nysten, K, Roelants, I ., Rurangwa, E., Moore, H. D. M. and Ollevier, F. (1996). Use of computer assisted sperm an alysis (CASA) for mon itoring the effects of pollution on sperm quality of fish, applica tion to the effects of heavymetals. Aquatic Toxicology. 36, 223-237. Kitis, M., Kaplan, S. S., Karakaya, E., Yigit, N. O. and Civelekoglu, G. (2007). Adsorption of natural organic matter from waters by ir on coated pumice. Chemosphere. 66, 130. Kobriger, N. O., and Geinopolos, A. (1984). Sources and migration of highway runoff pollutants-Research Reports, Vollume III. Report, FHWA/RD-84/059. (PB86-227915) FWHA, US Department of Transportation. Kocadagistan, B., Kocadagistan, E., Topcu, N., Demircioglu, N. (2005). Wastewater treatment with combined upflow anaerobic fixed-bed and suspended aerobic reactor equipped with a membrane unit. Process Biochemistry, 40, 177-182. Kofinas, P., Kioussis, D. R., (2003). Reactiv e phosphorous removal from aquaculture and poultry productions systems using polymeric hydrogels. Environ. Sci. Technol. 37 (2), 423 427. Kotze, P., Du preez, H. H. and Van vuren, J. H. J. (1999). Bioaccumulation of copper and zinc in Oreochromis mossambicus and Clarias gariepinus, from the Olifants River, Mpumalanga, South Africa. Water SA. 25 (1), 99-110. Langston, W. J. (1984). Availabili ty of arsenic to estuarine and marine organisms: A field and laboratory evaluation. Mar. Biol., 80, 143-154. Lee, J. H., and Bang, K. W. (2000). Characterization of urban stormwater runoff. Water Res., 34(6), 1773. Lee, P.K., Touray, J. C., Baillif, P. and Ildefons e, J. P. (1997). Heavy metal contamination of settling particles in a retention pond along the A-71 motorway in Sologne, France. Sci. Tot. Env., 201, 1-15.
256 Lee, S. Z., Weng, C. H., Allen, H. E. (1994). In Groundwater contamination and control, Zoller, U., Ed., Marcel Dekker, New York, 241-255. Ledin, M. (2000). Accumulation of metals by mi croorganisms processes and importance for soil systems. Earth-Science Reviews, 51, 1. Lenat, D. R., Penrose D. L., and Eagleson K. W. (1981). Variable effects of sediment addition on stream benthos. Hydrobiologia, 79, 187. Lenat, D. and K. Eagleson. (1981). Ecological Effects of Urban Runoff on North Carolina Streams. North Carolina Divisi on of Environmental Managemen t, Biological Series #104. North Carolina Dept. of Natural Resources and Community De velopment, Raleigh, NC. Li, Y., Buchberger, S. G. and Sansalone, J. J. (1999). Variably Saturate d Flow in Storm-water Partial Exfiltration Trench. J. Environ. Eng.,125(6), 556. Linder ,M.C. (1991). The Biochemistry of Copper. New York: Plenum Press. Liu, D., Teng, Z., Sansalone and J., Cartledge, F. K. (2001). Surface characteristics of sorptivefiltration stormwater media II: higher specific gravity (rs > 1.0). J. Environ. Eng., 127(10), 880-888. Liu, D., Sansalone, J. J., and Cartledge, F. K. ( 2004). Adsorption characteri stics of oxide coated buoyant media (rs < 1.0) for storm water treatment. II : Batch equilibria and kinetics. J. Environ. Eng., 130(4): 383-390. Liu, D., Sansalone, J. J., and Cartledge, F. K. (2005). Adsorptive Kinetics for Urban RainfallRunoff Metals by Composite Oxide-Coated Polymeric Media. J. Environ. Eng., 131(8): 1168-1177. Liu, D., Sansalone, J. J., and Cartledge, F. K. (2005). Comparison of Sorptive Filter Media for Treatment of Metals in Runoff. J. Environ. Eng., 131(8), 1178-1186. Ma, J., Roy, A., Sansalone J. J., Pardue, J. (2005). Partitioning and Particulate-bound Distribution of Phosphorus in Rainfall-Runoff, J. Bennett Johnston Sr., Center for Advanced Microstructures and Devices, Baton R ouge, Louisiana, Annual Users Meeting. Makepeace, D., Smith, D., and Stanley, S. (1995) Urban rainfall-runoff quality: Summary of contaminant data. Critical Reviews in Environmental Science and Technology, 25(2), 93139. Marinussen, M. P. J. C., van der Zee, S.E.A. T.M., de Haan, F.A.M., Bouwman, L.M. and Hefting M. M.. 91997). Heavy metal (copper, lead and zinc) accumulation and excretion by the earthworm, Dendro baena veneta. J. Environ. Qual. 26, 278.
257 Marsalek, J., Rochfort, Q., Brownlee, B., Mayer, T. and Servos, M. ( 1999). An exploratory study of urban runoff toxicity. Wat. Sci. Tech., 39 (12), 33-39. Mason, Y., Ammann, A. A., Ulrich, A., and Si gg., L. (1999). Behavior of Heavy Metals, Nutrients, and Major Components during Roof Runoff Infiltration. Environ. Sci. Technol., 33(10), 1588-1597. Mathialagan, T. and Viraraghavan, T. (2002). A dsorption of cadmium fr om aqueous solutions by perlite. J. Hazard. Mater., B94, 291-303. McKenzie, D. J., and Irwin, G. A. (1983). Wate r-quality assessment of st ormwater runoff from a heavily used urban highway bridge in Miami, Florida. USGS Water Resources Investigations Report 83-4153, Tallahassee, FL. Means, J. C., Wijayaratne, R. (1982). Role of natural colloids in the trans port of hydrophobic pollutants. Science, 215, 968-970. Minton, G., Leif, B., and Suther land, R. (1996) PNW Experience with Vegetated (Bio)swales, Stormwater Treatment Northwest, 2-4. Mishra, S. K., Sansalone, J. J. and Singh, V. P. (2004). Partitioning Analog for Metal Elements in Urban Rainfall-runoff Overland Flow Using the Soil Conservation Service Curve Number Concept. J. Environ. Eng., 130 (2), 145-154. Morrison, G. M., C. Wei, and Engdahl, M. (1993). Variations of environmental parameters and ecological response in an urban river. Water Sci. Tech., 27, 191. Morrison, G., Revitt, D., and Ellis, J. (1990). M etal speciation in sepa rate rainfall-runoff systems. Wat. Sci. Tech., 22(10/11), 53-60. Muller, B., and Sigg, L. (1990). Interaction of trace meals with natural particle surfaces: Comparison between adsorption experiments and field measurement. Aquat. Sci., 52(1): 1015. Munawar, M., Munawar, I. F., and Mayfield, C. I. (1987). Differential sensitivity of natural phytoplankton size assemblages to metal mixture toxicity. Arch. Hydrobiol. Beih., 25, 123 139. Munger, C., Hare, L., Craig, A. and Char est, P. M. (1999). Influence of exposure time on the distribution of cadmium within the cladoceran Ceriodaphnia dubia. Aquatic Toxicology. 44: 195-200. Muschack, W. (1990). Pollution of street run-off by traffic and local conditions. Sci. of the Total Envir., 93, 419-431.
258 Newton, R.B. (1989). The effects of stormwater surface runo ff on freshwater wetlands: A review of the literature and annotated bibliography. Prepared for the Massachusetts Department of Environmental Protection, Office of Research and Standards, by the Environmental Institute, University of Massachusetts at Amherst, MA. Niimi, A. J. and Kisson, G. P. (1994). Evaluat ion of the critical body burden concept based on inorganic and organic mercur y toxicity to rainbow tr out (Oncorhynchus mykiss). Arch. Environ.Contam. Toxicol. 26, 169-178. Patrick, W. H., and Verloo, M. (1998). Distrib ution of Soluble Heavy Metals Between Ionic and Complexed forms in a Saturated Sediment as Affected by pH and Redox Conditions. Water Sci. Technol., 37 (6-7), 165-172. Pelgrom, S. M. G J., Lamers, L. P. M, Garritsen, J. A. M, Pels, B. M.; Lock, R. A. C., Balm, P. H. M. and Wendelaar bonga, S. E. (1994). I nteractions between copper and cadmium during single and combined exposur e in juvenile tilapia Oreochromis mossambicus: Influence of feeding condition on whole body metal accumulati on and the effect of the metals on tissue water and ion content. Aquatic Toxicology. 30, 117-135. Petterson, T. J. R. (1999). Effect s of Variation of Water Qualit y on the Partitioning of Heavy Metals in a Stormwater Pond. Doktorsavhandlingar vid Chalmers Tekniska Hogskola, 1542, 1943-1946. Piedrahita R. H. (2003). Reducing the Potentia l Environmental Impact of Tank Aquaculture Effluents through Intensifi cation and recirculation. Aquaculture, 226, 35-44. Pisano, W. C., and Brombach, H. (1994). Operati onal experience with vorte x solids separators for combined sewer overflow (CSO) control. Water Sci. Technol., 29 (1-2), 383. Playle, R.C., Goss, G.G., and Wood, C.M. (1989). Physiological disturbances in raibow trout (Salmo gairdneri) during acid and aluminum exposures in soft water of two calcium concentrations. Can J. Zool. 67, 314-324. Pretty, J. N., Mason, C. F., Nedwell, D. B ., Hine, R. E., Leaf, S., and Dils, R. (2003). Environmental costs of freshw ater eutrophication in England and Wales. Environ. Sci. Technol., 37, 201-208. Ran, Y., Fu, J. M., Sheng, G. (2000). Fractiona tion and composition of colloidal and suspended particulate materials in rivers. Chemo., 41, 33-43. Rani, E. F., Elumalai, M., and Balasubramanian, M. P. (1997). The toxicity of mixtures of monocrotophos and ammonium chloride to a freshwater fish Oreochromis mossambicus. Biomed. Lett. 55, 193-198.
259 Reddi, L. N. (1997). Particle Transport in So ils: Review of Signi ficant Porcesses in Infrastructure Systems. Journal of Infrastructure Systems, Vol 3 (2), 78-86. Revitt, D. M., Hamilton, R. S., and Warren, R. S. (1990). Transport of heavy metals within a small urban catchment. Sci. Total Environ., 93, 359. Sansalone, J. J. (1999). In-situ performance of a passive treatment system for metal source control, Water Sci. Technol. 39(2), 193. Sansalone, J. J. (2005). Perspective on the synt hesis of unit operations and process (UOP) concepts with hydrological controls for rainfall-runoff. J. Environ. Eng., 131(7), 995-997. Sansalone, J. J. and Buchberger, S. G. (1997). Par titioning and first flush of metals and solids in urban highway runoff. J. Envir. Engrg. Div., ASCE, 123(2), 134-143. Sansalone, J. J. and Cristina, C. M. (2004). Fir st Flush Concepts for Suspended and Dissolved Solids in Small Impervious Watershed. J. Environ. Eng., 130(11), 1301-1314. Sansalone, J. J., Hird, J. P., Cartledge, F. K ., and Tittlebaum, M. E. (2005). Event-based stormwater quality and quantity loadings from elevated urban infrastructure affected by transportation. Water Environ. Res., 77(4), 348. Sansalone, J. J. and Kim, J. (2007). Suspended particles destabilizati on in retained urban stormwater as a function of coagulant dosage and redox conditions. Water res., in process. Sansalone, J., Koran, J., Smithson, J., and Buchbe rger, S. (1998). Physical characteristics of highway solids transported during rainfall. J. of Envir. Engrg., 124(5), 427-440. Sansalone, J. J. and Teng, Z. (2004). In-situ storm water treatment a nd recharge through infiltration: Quality and Quantity Attenuation, J. of Envir. Engrg., 130(9), 990-1007 Sansalone, J. J. (2002). Physi cal and Chemical Nature of Urban Storm Water Runoff Pollutants, in Management of Wet-Weather Flows in the watershed, 43-65. Sansalone, J. J., Koran, J. M., Smithson, J. A., and Buchberger, S. G. (1998). Physical characteristics of urban roadway solids transported during rain. J. Environ. Eng., 125(5), 427. Sansalone, J. J., and Teng, Z. (2005). Transient Ra infall-runoff Loadings to a Partial Exfiltration System: Implications for Urban Water Quantity and Quality. J. Environ. Eng., 131(8), 11551167.
260 Sansalone, J. J., and Tribouillard, T. (1999). V ariation in characteristics of abraded roadway particles as a function of part icle size Implications for water quality and drainage. Transportation Research Record 1690, Transportation Research Board, Washington, D.C., 153. Schueler, T. R. (1987). Contro lling Urban Runoff: A Practic al Manual for Planning and Designing Urban BMPs. Metropolitan Washi ngton Council of Governments, Washington, DC. Scholz, M., (2004). Treatment of gully pot effluent contai ning nickel and copper with constructed wetlands in a cold climate. J. Chem. Technol., Bio., 79(2), 153-62. Schultz, T.W. and T.C. Allison. (1979). Toxicity and toxic interaction of aniline and pyridine. Bull. Environ. Contam. Toxicol., 23, 814. Schwarz, T. and Wells, S. (1999). Storm water pa rticle removal using cr oss-flowfiltration and sedimentation. Advances in Filtration and Separation Technology, American Filtrations and Separations Society, 219-226. Seida, Y., Nakano, Y. (2002). Removal of phosphate by layered double hydroxides containing iron. Water Res., 36, 1306-1312. Seymore, T. (1994). Bioaccumulation of Metals in Barbus marequensis from the Olifants River, Kruger National Park and Lethal Levels of Manganese to Juvenile Oreochromis mossambicus. M.Sc-thesis, Rand Afrikaans University, South Africa. Sharpley, A. N., Daniel, T., Sims, J. T., Lem unyon, J., Stevens, R., and Parry, R. (1999). Agricultural Phosphorus and Eutrophication; U.S. Department of Agriculture, U.S. GPO, Washington, D.C. Singh, V. P. (1997). Kinematic Wave Modeling in Water Resources. In Environmental Hydrology. J. Wiley and Sons, New York. Sonzogni, W. C., Chesters, G., Coote, D. R., Jeffs D. N., Konrad, J. C., Ostry, R. C., and Robinson, J. K. (1980). Pollution from la nd runoff. Envir. Sci. Tech. (14), 148-153. Spehar, R. L. and Fiandt, J. T. (1986). Acute an d chronic effects of wate r quality criteria-based metal mixtures on three aquatic species. Environ. Toxicol. Chem., 5, 917. Standard methods for the examination of wate r and wastewater. (1995). 19th Ed., American Public Health Association (APHA), Ameri can Water Works Association (AWWA), Water Environment Federation (W EF), Washington, D.C. Strecker, E.W. (1994). Constituents and Methods for Assessing BMPs In Proceedings of the Engineering Foundation C onference on Storm Water Monitoring. August 7-12, Crested Butte, CO.
261 Stockdale, E. C. (1991). Freshwater wetlands, urban stor mwater, and nonpoint source pollution control: A literature review and annotated bibliography., Washington State Department of Ecology. Stumm, W. and Morgan, J. (1996). Aquatic chemistry, Wiley, New York. Teng, Z. and Sansalone, J. J. (2004). In-situ storm water treatment a nd recharge through infiltration: Particle transpor t and separation mechanisms, J. Environ. Eng., 130(9), 10081020. Thomson, N. R., McBean, E. A., Snodgrass, W. and Monstrenko, I. B. (1997). Highway stormwater runoff quality: Development of surrogate parameter relationships. Water, Air and Soil Pollution, 94, 307-347. Tobiason, J. E., Johnson, G. S., Westerhoff, P. K., and Vigneswaran, B. (1993). Particle Size and Chemical Effects on Cont act Filtration Performance. J. Environ. Eng., 119(3), 520-539. U.S. EPA. (1984). Nonpoint sources of pollution in the U.S. U.S. Environmental Protection Agency Report to Congress, Washington, D.C. Vaze, J., and Chiew, H. S. F. (2002). Experiment al study of pollutant accumulation on an urban road surface. Urban Water, 4, 379-389. Walker, T. A., Allison, R. A., Wong, T. H. F., and Wootton, R. M. (1999). Removal of suspended solids and associated polluta nts by a CDS gross pollutant trap. Report No. 99/2, Cooperative Research Centre for Catchment Hydrology, 32. Waller, W.T., Acevedo, M.F., Allen, H. J., Kennedy, J. H., Dickson, K. L., Ammann, L. P., and Morgan, E. L. (1995). The use of remotely se nsed bioelectric action potentials to evaluate episodic toxicity events and ambient t oxicity. Water for Texas Conference, Proceedings of the 24th Water for Texas Conference: Research Leads the Way, Texas Water Development Board, Texas Water Resources Institute, Texas Water Conservation Association, Austin, TX., 726 Waschbusch, R. J., Selbig, W. R., and Bannerm an, R. T. (1999). Sources of phosphorus in stormwater and street dirt fr om two urban residential basi ns in Madison, Wisconsin. Water Resources Investigations Report 99-4021, U.S. Geological Survey. Wepener, V., Van vuren, J. H. J and Du preez, H. H. (2001). Uptake a nd distribution of a copper, iron and zinc mixture in gill, liver and plasma of a freshwater teleost, Tilapia sparmanii. Water SA., 27 (1), 99-108. Welch, E. B. (1992). Ecological effects of wastewater, 2nd Ed., Chapman and Hall, New York.
262 Wendt, R. C. and Corey, R. B. (1980). Phos phorus variations in surface runoff from agricultural lands as a function of land use. Journal of Environmental Quality, 9(1), 130136. Werner, E. (1983). Effects of highways on karst springs: An example from Pocahontas County, West Virginia. In Environmental Karst, Dougherty, P.H., ed. Geospeleo Publications, Cincinnati, Ohio, 3-13. Wesley, L. D. (2001). Determination of specific gravity and void ratio of pumice materials. Geotech. Test. J., 24, 418. Wilson, A R., Lion, L. W., Nelson, Y. M., Shul er, M. L., and Ghiorse, W. C. (2001). The Effects of pH and Surface Composition on Pb Ad sorption to Natural Freshwater Biofilms. Environ. Sci. Technol., 35(15), 3182-3189. Witeska, M., Jezierska, B. and Chaber, J. ( 1995). The influence of cadmium on common carp embryos and larvae. Aquaculture, 129: 129-132. Wong, T., Coleman, J., Duncan, H., Fletcher, T., Siriwardena, L., and Wooton, R. (2002). Model for urban stormwater improvements conceptualisatio, Version 1.0, User Manual, CRC for Catchment Hydrology, Urban Stormwater Quality Program. Wong, T. H. F. (1997). Continuous deflective se paration: its mechanism and applications. Proc. 70th Water Environment Federation Conf., Chicago, Illinois, USA, 18-22 October., 1 12. Woodward-Clyde Consultants. (1991). The use of wetlands for cont rolling stormwater pollution. Draft. Prepared for U.S. Environmental Pr otection Agency, Region V, Water Division, Watershed Management Unit, Chicago, IL U.S. Environmental Protection Agency, Washington, D.C. Wu, J., and Yu, B. (2007). A fractal resist ance model for flow through porous media. International Journal of Heat and Mass Transfer, 50, 3925-3932. Yao, K. M., Habibian, M. T. and OMelia, C. R. (1971). Water and wa ste water filtration: Concepts and applications. Envi. Sci. Tech., 5(11), 1105. Zhang, W. (2006). Improvement of Phosphorus and Heavy Metal Retention in Stormwater. Treatment. MS thesis. Stillwater, Oklahoma: Oklahoma State University. Zhao, D., Sengupta, A. K. (1998). Ultimate rem oval of phosphate from wastewater using a new class of polymeric ion exchangers. Water Res., 32(5), 1613.
263 BIOGRAPHICAL SKETCH Bo Liu was born in Jiaozuo, P. R. China and he came to United States of America in Spring 2000 after received his Bachelors degrees in environmental science in P. R. China. He got his Masters degree in environmental t oxicology in December 2003 under the guidance of Dr. Robert Romaire in the School of Renewable Natural resources Louisiana State University. Bo Liu will receive the degree of Doctoral of Philosophy in environmental engineering from University of Florida in December 2007. His doctoral research was focused on toxicity and clarification of anthropogenic c onstituents transported by urban rainfall-runoff. He worked under the guidance of Dr. John J. Sansalone in the Department of Environm ental Engineering and Sciences.