EVALUATION OF ANIONIC POLYACRYLAMIDE AS AN EROSION CONTROL MEASURE USING INTERMEDIATE-SCALE EXPERIMENTAL PROCEDURES Except where reference is made to the work of others, the work described in this thesis is my own or was done in collaboration with my advisory committee. This thesis does not include propriety or classified information. ___________________________________ Alexander Lee Shoemaker Certificate of Approval: ______________________________ ______________________________ T. Prabhakar Clement Wesley C. Zech, Chair Professor Assistant Professor Civil Engineering Civil Engineering ______________________________ ______________________________ Rod E. Turochy George T. Flowers Associate Professor Dean Civil Engineering Graduate School EVALUATION OF ANIONIC POLYACRYLAMIDE AS AN EROSION CONTROL MEASURE USING INTERMEDIATE-SCALE EXPERIMENTAL PROCEDURES Alexander Lee Shoemaker A Thesis Submitted to the Graduate Faculty of Auburn University in Partial Fulfillment of the Requirements for the Degree of Master of Science Auburn, Alabama August 10, 2009 iii EVALUATION OF ANIONIC POLYACRYLAMIDE AS AN EROSION CONTROL MEASURE USING INTERMEDIATE-SCALE EXPERIMENTAL PROCEDURES Alexander Lee Shoemaker Permission is granted to Auburn University to make copies of this thesis at its discretion, upon request of individuals or institutions and at their expense. The author reserves all publication rights. ______________________________ Signature of Author ______________________________ Date of Graduation iv VITA Alexander L. Shoemaker, son of William L. Shoemaker and Colleen P. Shoemaker, was born on May 24th, 1984, in Auburn, AL. He graduated high school from Hudson High School in Hudson, OH in 2003. After graduating, he attended Auburn University in Auburn, AL and graduated with a Bachelor of Civil Engineering degree in May of 2007. That same month, he entered the Graduate School at Auburn University to pursue a Master of Science in Civil Engineering. v THESIS ABSTRACT EVALUATION OF ANIONIC POLYACRYLAMIDE AS AN EROSION CONTROL MEASURE USING INTERMEDIATE-SCALE EXPERIMENTAL PROCEDURES Alexander Lee Shoemaker Master of Science in Civil Engineering, August 10, 2009 (B.C.E, Auburn University, 2007) 220 Typed Pages Directed by Wesley C. Zech The United States Environmental Protection Agency (EPA) has been working to establish stricter guidelines and limitations pertaining to stormwater and the various pollutants transported by runoff. This transport is primarily prevalent in runoff exiting exposed and disturbed land common to construction sites. Therefore, understanding the effectiveness of many different erosion and sediment control technologies is becoming a high priority to the construction industry. In this research, the effectiveness of a chemical stabilizer known as anionic polyacrylamide (PAM) was examined using intermediate- scale testing procedures that mimic conditions similar to a highway embankment with a compacted 3:1 fill slope. The first phase of this research focused on intermediate-scale testing procedures that were developed from previous research efforts and further vi modified to enable researchers to rapidly generate large quantities of valuable data. The rain regime selected for this research consisted of a 2-year, 24-hour storm event for Montgomery, AL which was divided into four 15 minutes rain events with 15 minute breaks in between that produced 1.10 in. of rainfall per event and a total cumulative amount of 4.4 in. The second phase of research focused on conducting intermediate-scale experiments to examine the effectiveness of PAM with different application rates and application methods. These different application methods included: (1) dry granular PAM applied directly to test plots and (2) dry PAM mixed with water to form a liquid spray application. Application rates were determined through manufacturer recommendations (i.e. 25 to 35 lbs/acre) and this research conducted additional experiments to examine the performance at PAM at rates lower than the recommended rate (i.e. 15 lbs/acre). Liquid PAM applications were not allowed to dry prior to being subjected to rainfall to simulate a ?worst-case scenario? for treatments. The results from this phase of research showed that dry PAM applied at the recommended rate of 35 lbs/acre performed the best out of the various PAM treatments by significantly reducing initial turbidity levels by 97% and eroded soil by 50% when compared to the bare soil control condition. Collected runoff samples indicated that runoff from test plots treated with dry PAM applied at 35 lbs/acre reduced turbidity to the proposed EPA effluent limits of 13 NTU within 20 seconds. vii ACKNOWLEDGEMENTS The author would like to extend a special thanks to Dr. Wesley C. Zech, Dr. T Prabhakar Clement, and Dr. Rod E. Turochy for their time and guidance throughout this research effort. The author would also like to thank the Alabama Department of Transportation (ALDOT) for funding the project, the Auburn University Civil Engineering Department for their financial support, Mr. Steven Iwinski of Applied Polymer Systems, Inc. for supplying the polymers used in the research, and everyone at the National Center for Asphalt Technology (NCAT) Test Facility who graciously provided an innumerable amount of time and assistance throughout the entire research project. The author would also like to thank Mr. David Graham, Mr. Elliot Smith, Mr. Nick Hall, Mr. David Johnson, Mr. Robert Devore, Mr. Benjamin Bethel, Mr. Wesley Wilson, and Mr. Wesley Donald for the help they provided in preparing and conducting experiments. Finally, the author would like to especially thank his parents, Lee and Colleen Shoemaker for everything, because without them, none of this would have been possible. viii Style manual or journal used Auburn University Graduate School Guide to Preparation of Master?s Thesis Computer software used Microsoft Word 2007, Microsoft Excel 2007, Microsoft PowerPoint 2007 and MicroStation v8 ix TABLE OF CONTENTS LIST OF TABLES ??????????????????????????.... xi LIST OF FIGURES ?????????????????????????? xiii CHAPTER ONE INTRODUCTION 1.1 BACKGROUND .......................................................................................... 1 1.2 EROSION AND SEDIMENTATION PROCESSES ................................... 2 1.3 BEST MANAGEMENT PRACTICES (BMPs) ........................................... 4 1.4 RESEARCH OBJECTIVES ......................................................................... 5 1.5 ORGANIZATION OF THESIS ................................................................... 7 CHAPTER TWO LITERATURE REVIEW 2.1 INTRODUCTION ........................................................................................ 8 2.2 PROCEDURES AND METHODS USED FOR EXPERIMENTATION.... 9 2.3 ANALYSES ................................................................................................ 34 2.4 LITERATURE REVIEW SUMMARY ...................................................... 37 CHAPTER THREE INTERMEDIATE-SCALE METHODS AND PROCEDURES 3.1 INTRODUCTION ...................................................................................... 40 3.2 INTERMEDIATE-SCALE TESTING ....................................................... 40 3.3 EXPERIMENTAL DESIGN ...................................................................... 43 3.4 EXPERIMENTAL PROCEDURES ........................................................... 57 3.5 SUMMARY ................................................................................................ 75 CHAPTER FOUR INTERMEDIATE-SCALE EXPERIMENTS RESULTS AND DISCUSSION 4.1 INTRODUCTION ...................................................................................... 77 4.2 EXPERIMENTAL RESULTS.................................................................... 77 4.3 SUMMARY .............................................................................................. 108 CHAPTER FIVE CONCLUSIONS AND RECOMMENDATIONS 5.1 INTRODUCTION .................................................................................... 111 5.2 INTERMEDIATE-SCALE METHODS AND PROCEDURES .............. 112 5.3 DRY AND LIQUID PAM EROSION CONTROL .................................. 114 5.4 RECOMMENDED FUTURE RESEARCH ............................................. 116 REFERENCES???????????????????????????? 118 x APPENDICES .................................................................................................................122 APPENDIX A: Manufacturer?s Specifications for Rainfall Simulator Components .............................................................................123 APPENDIX B: Rainfall Intensity-Duration-Frequency Curves for Alabama ....129 APPENDIX C: Manufacturer?s Specifications for Equipment used during Experimentation .......................................................................132 APPENDIX D: Experimental Results ...............................................................136 APPENDIX E: ANOVA Tables ........................................................................179 APPENDIX F: Further Research on Liquid PAM .............................................183 xi LIST OF TABLES Table 1.1 Examples of Different BMPs for Protecting Against Erosion, Sedimentation, and Stormwater Discharge ................................................................................ 4 Table 2.1 Summary of Experimental Rainfall Regimes from Literature......................... 17 Table 2.2 Summary of Experimental Testing and Procedures ........................................ 28 Table 2.3 Summary of PAM Treatments Applied to Mild Slopes (i.e. < 10%) .............. 32 Table 2.4 Summary of PAM Treatments Applied to Steep Slopes (i.e. > 10%) ............. 33 Table 3.1 Christian Uniformity Coefficients for Rainfall Simulator ............................... 52 Table 3.2 Rain Regime Table for Different Intensities and Storm Durations ................. 56 Table 3.3 Percent Composition and Classification of Experimental Soil ........................ 57 Table 3.4 Calculated Dry Unit Weight (pcf) and Required Number of Drops ................ 61 Table 3.5 PAM Application Rates ................................................................................... 67 Table 3.6 Breakdown of Collected Data Totals ............................................................... 73 Table 4.1 Average Surface and Cumulative Runoff for Each Test [Dry PAM] .............. 80 Table 4.2 Average Surface and Cumulative Runoff for Each Test [Liquid PAM] .......... 80 Table 4.3 Tukey-Kramer Multiple Comparisons on Average Surface Runoff [Test 1] .. 82 Table 4.4 Tukey-Kramer Multiple Comparisons on Average Surface Runoff [Test 2] .. 82 Table 4.5 Tukey-Kramer Multiple Comparisons on Average Surface Runoff [Test 3] .. 83 Table 4.6 Tukey-Kramer Multiple Comparisons on Average Surface Runoff [Test 4] .. 83 Table 4.7 Average Initial Turbidity Results for Surface Runoff [Dry PAM] .................. 86 xii Table 4.8 Average Initial Turbidity Results for Surface Runoff [Liquid PAM] .............. 87 Table 4.9 Tukey-Kramer Multiple Comparisons on Average Initial Turbidity [Test 1] ........................................................................................................... 89 Table 4.10 Tukey-Kramer Multiple Comparisons on Average Initial Turbidity [Test 2] ........................................................................................................... 89 Table 4.11 Tukey-Kramer Multiple Comparisons on Average Initial Turbidity [Test 3] ........................................................................................................... 90 Table 4.12 Tukey-Kramer Multiple Comparisons on Average Initial Turbidity [Test 4] ........................................................................................................... 90 Table 4.13 Summary of Turbidity vs. Time Measurements ............................................ 96 Table 4.14 Average Soil Loss due to Surface Runoff [Dry PAM] ................................ 102 Table 4.15 Average Soil Loss due to Surface Runoff [Liquid PAM] ............................ 102 Table 4.16 Tukey-Kramer Multiple Comparisons on Average Soil Loss [Test 1] ........ 104 Table 4.17 Tukey-Kramer Multiple Comparisons on Average Soil Loss [Test 2] ........ 104 Table 4.18 Tukey-Kramer Multiple Comparisons on Average Soil Loss [Test 3] ........ 105 Table 4.19 Tukey-Kramer Multiple Comparisons on Average Soil Loss [Test 4] ........ 105 xiii LIST OF FIGURES Figure 3.1 Test Facility Exterior. ..................................................................................... 42 Figure 3.2 Test Facility Interior. ...................................................................................... 42 Figure 3.3 Design and Specifications of a Test Box. ....................................................... 45 Figure 3.4 Runoff Collection Device. .............................................................................. 45 Figure 3.5 Constructed Test Boxes used for Experimentation. ....................................... 46 Figure 3.6 Rainfall Simulator Design. ............................................................................. 48 Figure 3.7 Illustration of Constructed Rainfall Simulator. .............................................. 50 Figure 3.8 Rainfall Distribution (in ml) Verifying Uniformity. ...................................... 51 Figure 3.9 Rainfall Intensity (in/hr) as a Function of Rainfall Simulator?s Operating Pressure (psi). ................................................................................................ 54 Figure 3.10 Particle Size Distribution for Experimental Soil. ......................................... 58 Figure 3.11 Proctor Curve for Experimental Soil. ........................................................... 59 Figure 3.12 Determining Required Compaction Rate with Hand-tamp. ......................... 60 Figure 3.13 Compaction of Soil using Hand-tamp. ......................................................... 61 Figure 3.14 Flowchart of Experimental Organization ..................................................... 64 Figure 3.15 Obtaining OMC for Experimentation ........................................................... 65 Figure 3.16 Using Hand Tamp to Compact 1 inch of Test Soil ...................................... 66 Figure 3.17 Application of Dry Granular PAM. .............................................................. 67 Figure 3.18 Preparing Liquid PAM Mixture for Application. ......................................... 68 xiv Figure 3.19 Application of Liquid PAM.......................................................................... 69 Figure 3.20 Plots Placed in ?Left? and ?Right? Position................................................... 69 Figure 3.21 Recording Volume and Mass of Surface Runoff. ........................................ 71 Figure 3.22 ANALITE NEP 160 Turbidity Meter and Probe. ........................................ 71 Figure 3.23 Hayward Filter Bags with Sediment-Laden Water. ..................................... 72 Figure 3.24 Soil Particle Distribution and Sieves. ........................................................... 72 Figure 4.1 Average Surface Runoff vs. Time. ................................................................. 78 Figure 4.2 Average Cumulative Surface Runoff vs. Time. ............................................. 79 Figure 4.3 Average Initial Turbidity of Surface Runoff vs. Time. .................................. 85 Figure 4.4 Average Recorded Turbidity for All Samples vs. Time. ................................ 92 Figure 4.5 Treatment?s Average Recorded Turbidity from 5 and 10 Minute Samples [Dry PAM]. .................................................................................................... 93 Figure 4.6 Treatment?s Average Recorded Turbidity from 5 and 10 Minute Samples [Liquid PAM]. ................................................................................................ 94 Figure 4.7 Average Soil Loss versus Time. ..................................................................... 98 Figure 4.8 Average Cumulative Soil Loss vs. Time. ....................................................... 99 Figure 4.9 Average Percent Sediment in Surface Runoff vs. Time. .............................. 101 Figure 4.10 Average Particle Size Distribution of Stock Pile Soil and Surface Runoff.107 Figure 4.11 Average Initial Turbidity vs. Average Eroded Soil. ................................... 108 1 CHAPTER ONE INTRODUCTION 1 1.1 BACKGROUND The discharge of sediment laden stormwater from construction sites proves to be a major environmental concern. Stormwater discharge is a form of nonpoint source pollution (NPS), which is the result of rainfall or snowmelt traversing though an area and can carry natural and non-natural pollutants, into lakes, rivers, and ground water. In addition, the effect of stormwater discharge can significantly increases the occurrence of erosion and sedimentation (two specific consequences of NPS) when occurring over disturbed land, such as construction sites. Therefore, to provide a level or protection for humans, wildlife, and the environment, federal and state regulations require construction site owners and operators to manage stormwater discharge and prevent NPS. In 1987, Congress passed the Clean Water Act, of which Section 319 established a national program focusing on the control of nonpoint sources of water pollution. Due to Section 319, all States have adopted management programs to help in the control and reduction of NPS (U.S. EPA 2003). Since active construction sites, or any other land-disturbing activities, are identified as one of the leading contributors to NPS, a great deal of effort goes into 2 eliminating the dangers attributed to stormwater and generated runoff. For the U.S. alone, it is estimated that the total amount of sediment washed from construction sites into surface bodies of water is over 80 million tons each year (Novotny, 2003). Much of this sediment laden runoff could have been mitigated through the use of effective erosion and sediment control programs. 1.2 EROSION AND SEDIMENTATION PROCESSES Construction site erosion and sedimentation are two primary contributors to NPS pollution in the construction industry. This section will cover the basics pertaining to the process of erosion and sedimentation. The Alabama Soil and Water Conservation Committee (ASWCC) identifies the main factors contributing to the erosion process to be climate, topography, soils and vegetative cover (ASWCC, 2003). Climate includes rainfall, temperature, and wind, with rainfall being the main contributor to erosion. The amount, duration, and intensity of rainfall are all major factors that can increase the severity of runoff. However, with proper ground cover, such as vegetation, the amount of erosion and sedimentation will be greatly reduced. To assist in understanding and reducing these processes that increase the amount of erosion and sedimentation, the ASWCC published a handbook entitled, ?Alabama Handbook for Erosion Control, Sediment Control, and Stormwater Management on Construction Sites and Urban Areas?, to help contractors with their erosion control plans. The handbook states that the process of erosion is, ?when the land surface is worn away by the action of water, wind, ice or gravity? (ASWCC, 2003). Rainfall water tends to be the common source of erosion. The effect rainfall has on soils includes the force of 3 raindrops impacting the ground and physical movement of surface stormwater runoff, resulting in shear forces on the soil. Both of these actions cause soil particles to become detached and increase potential for NPS pollution. However, since physical properties involving raindrop impacts and surface runoff are different, multiple methods to address, reduce, and eliminate the amount of erosion and sedimentation on construction sites are available (ASWCC, 2003). The other concern of stormwater discharge is sedimentation. Once the velocity of stormwater runoff decreases, detached soil particles begin to settle out of suspension. The larger and heavier particles (i.e. gravel and sand) in runoff will settle out faster. However, smaller particles, (i.e. silt and clay), require longer detention times to promote settlement. To compound the issue further, these smaller particles are also more easily detachable from the surface due to runoff. This process of soil particles settling out of suspension is known as sedimentation. To measure the severity of sedimentation, a unit of measure called turbidity is used. If water is highly turbid, it indicates that more suspended particles are present in the runoff (ASWCC, 2003). It is important to control/reduce sedimentation because these suspended particles can be detrimental to aquatic wildlife. The U.S. Environmental Protection Agency (EPA) has a program available to assist with the management of stormwater discharge from construction sites that result in the process of erosion and sedimentation, known as the National Pollutant Discharge Elimination System (NPDES). This program requires that any construction site disturbing an area of one acre or larger must obtain authorization to discharge stormwater 4 under a NPDES construction stormwater permit (U.S. EPA, 2008a). This program also requires operators of regulated construction sites to implement erosion and sediment control best management practices (BMPs) to prevent NPS pollution. Therefore, the need to develop cost effective measures that help control and prevent erosion and sedimentation is becoming an important issue in the industry. 1.3 BEST MANAGEMENT PRACTICES (BMPs) Best management practices (BMPs) is the process of implementing methods to assist with minimizing erosion and sedimentation on disturbed land. These BMPs are outlined in detail in the ASWCC Handbook and provides contractors with the available methods to assist with controlling stormwater discharge, such as installation and maintenance instructions. Specific examples of these BMPs are listed below in Table 1.1. Table 1.1 Examples of Different BMPs for Protecting Against Erosion, Sedimentation, and Stormwater Discharge Techniques Best Management Practices Surface Stabilization Chemical Stabilization; Erosion Control Blankets; Mulching; Permanent Seeding; Retaining Walls; Sodding; Temporary Seeding Runoff Conveyance Check Dams; Diversions; Drop Structures; Outlet Protection; Subsurface Trains; Swales Sediment Control Brush/Fabric Barriers; Drop Inlet Protection; Filter Strips; Floating Turbidity Barriers; Inlet Protection; Sediment Barriers; Sediment Basins; Sediment Traps Stormwater Management Porous Pavement; Stormwater Detention Basins Stream Protection Buffer Zones; Channel Stabilization; Streambank Protection; Temporary Stream Crossings 5 The examples of BMPs shown in Table 1.1 only provide a small sampling of the available measures. This research focuses on one type of chemical stabilization that use anionic polyacrylamide (PAM). PAM is a negatively charged chemical and when applied to soil surfaces it bonds with soil particles to help maintain soil structure and reduce erosion. In addition to acting as an erosion control measure, PAM also serves as a binding agent to flocculate soil particles that have become detached during the erosion process. This flocculation of fine particles occurs when the negative charge of PAM polymers combine together with suspended soil particles. The resulting increase in combined particle sizes aids flocculation. Therefore, PAM applied as a chemical stabilization technique will assist in reducing erosion and sedimentation caused by stormwater. The U.S. Environmental Protection Agency (U.S. EPA) states in its chemical stabilization section of the NPDES, that PAM effectiveness as a stabilization method can range from 70% to 90% (2006). Anionic PAM can be applied in three different forms: (1) dry, granular form, (2) liquid form (granular PAM/water mixture), and (3) emulsified form. 1.4 RESEARCH OBJECTIVES The research presented herein is based on a continuing effort by the Department of Civil Engineering at Auburn University to study and test erosion and sediment control BMPs that are typically used on highway construction sites. The intent of this research effort is to expand on a previous study that tested anionic PAM in the dry granular form. This research incorporates a new facility, specifically designed to test various erosion and sediment control measures. The facility is located at the National Center for Asphalt 6 Technology (NCAT) Test Facility near Opelika, AL. The focus of this research was to conduct experiments for the purpose of examining the effectiveness of both dry and liquid PAM applications for erosion and sediment control. The research effort was divided into two phases. Phase 1 focuses on the development of the test facility, rainfall simulator, testing apparatus, and new methods to evaluate treatments. Phase 2 of the research focuses on the experimentation of the anionic PAM as either applied in dry or liquid form with different application rates. The specific phases of this research effort are described in detail below. PHASE 1: TESTING FACILITY 1. Design and develop a testing apparatus that will allow for efficient experimental setup and to achieve reproducible results. 2. Design and construct a rainfall simulator that models realistic rainfall events. 3. Develop a unified testing methodology to obtain reproducible results for the comparison of all possible experiments. PHASE 2: EXPERIMENTATION 1. Examine the effectiveness of various application rates of dry and liquid PAM for use as an erosion control measure on compacted, 3:1 slopes. 2. Compare experimental results to provide recommendations for using PAM on highway construction sites. 7 1.5 ORGANIZATION OF THESIS This thesis is divided into five chapters. Following this chapter, Chapter 2: Literature Review, examines the body of knowledge pertaining to research and experiments conducted to evaluate PAM as an erosion and sediment control measure. This chapter discusses the designs, procedures, and experimental results that were presented in previous research efforts. Also discussed are advantages and disadvantages of performing experiments at different scales (i.e. field vs. intermediate). Chapter 3: Intermediate-Scale Methods and Procedures, outlines the designs, methods, procedures used in conducting intermediate-scale experiments, and methods of analysis used in this research. This chapter includes details on the design of the new test facility, test plots, rainfall simulator, and procedures used to prepare and perform experiments. Chapter 4: Experimental Results, presents the results generated from all the experiments performed and includes ANOVA statistical analyses to determine if experimental results were statistically significant. Chapter 5: Conclusions and Recommendations, provides insight on the use and performance of PAM (i.e. dry vs. liquid application) as an erosion and sediment control technology. This chapter also provides recommendations for future research using PAM in conjunction with other erosion and sediment control measure to determine the optimum combination of technologies to prevent NPS pollution. 8 CHAPTER TWO LITERATURE REVIEW 2 2.1 INTRODUCTION The use of anionic polyacrylamide (PAM) as an erosion control measure is a topic that has been gaining attention recently. Several researchers have investigated the use of PAM in a multitude of applications, however, testing methods vary, and there exists a lack of uniform agreement on the effectiveness of PAM. Differences include different methods for applying PAM (i.e. how PAM should be applied and in what quantities), whether PAM is effective when applied alone without any other BMPs, or whether PAM should be used in conjunction with additional erosion control measures. In addition, there are concerns on the effectiveness of PAM on steep slopes, which are primarily used in conjunction with construction projects. Previous research conducted by McDonald (2007) covered the initial investigation performed by Auburn University to test the effectiveness of anionic PAM applied in a granular form as an erosion and sediment control BMP and is further reviewed in this chapter. One of the primary goals of this research was to develop new procedures for testing the effectiveness of PAM and provide a uniform methodology for future research. Therefore, the focus of this literature review will examine: (1) procedures and methods 9 used when testing anionic PAM, (2) experimental results, and (3) types of analysis used in quantifying experimental data. 2.2 PROCEDURES AND METHODS USED FOR EXPERIMENTATION Procedures used for testing anionic PAM vary from study to study without a unified system for evaluation. This section provides an overview of procedures, methods, and testing apparatus? used throughout the literature. This includes examining different methods used to generating runoff via rainfall, experimental scale, experimental setup, data collection procedures, and data analyses. 2.2.1 Generation of Runoff For the erosion process to occur, a source of runoff needs to be generated, which will cause soil particles to become detached. This detachment of soil particles and resulting transport in stormwater is one of the main contributors to NPS pollution and a source of concern in practice. Sources of runoff could include: (1) natural rainfall, (2) simulated rainfall, (3) irrigation, or (4) channelized flow. All four methods of generating runoff can be used to test the effectiveness of erosion control BMPs. A project?s research goals and objectives will be the main driving force in selecting the optimal method for generating runoff for experimental purposes. The research documented in this report focuses on typical highway embankments (i.e. fill slopes) exposed to surface runoff (i.e. sheet flow), therefore the use of simulated and natural rainfall from previous research studies was primarily examined and is discussed in the following sections. 10 2.2.1.1 Simulated Rainfall Designs Rainfall simulators can be used to produce and represent natural rainfall conditions, with the added benefit of having complete control over the system. The desired rainfall intensity, droplet impact velocity, duration, and resulting storm event can all be selected and controlled depending on experimental requirements. This level of control provides a distinct advantage over natural rainfall. However, an improperly designed rainfall simulator can lead to unrealistic rainfall conditions, jeopardizing the authenticity of any experimental results. The following five components when designing a rainfall simulator are critical to overall performance: (1) appropriate density of drop formers under constant head and intensity; (2) application uniformity; (3) droplet size distribution; (4) control of application rate; and (5) obtaining terminal velocity for the droplet distribution (Regmi and Thompson, 2000). Research performed by Hall (1970) investigated three problems in simulating rainfall: (1) controlling application rate in both time and space; (2) reproduction of drop size distribution observed in different intensities; and (3) reproduction of terminal velocities of drops. This study found that a general purpose rainfall simulator could not adequately be developed that addresses all three of these problems equally. The research indicated that by controlling one of the issues; the other issues would be adversely affected. Hall recommends that a rainfall simulator should be able to create artificial conditions that begin the hydrologic process and any laboratory results should be extrapolated to field conditions. 11 2.2.1.2 Simulated Rainfall Applied on Experiments While research and experiments examining erosion and sedimentation issues vary procedurally, many used a simulated rainfall to generate runoff on their experiments (Petersen et al., 2007; McLaughlin and Brown, 2006; Sepaskhah and Bazrafshan- Jahromi, 2006; Benik et al., 2003; Flanagan et al., 2002a; Peterson et al., 2002; Bjorneberg et al., 2000; Roa-Espinosa et al., 1999; Flanagan et al., 1997a; Flanagan et al., 1997b). While it appears using simulated rainfall is generally accepted, the actual designs and methods involved with individual rainfall simulators are varied. Experiments conducted by McLaughlin and Brown (2006) and Petersen et al. (2007) referenced the design of a solenoid-operated, variable intensity rainfall simulator by Miller (1987). Miller?s design included a WSQ nozzle threaded directly into a solenoid valve. The original simulator design included a one or three nozzle setup. Tests conducted by Miller (1987) indicated that a one nozzle configuration performed slightly better than multiple nozzles, in regards to uniform spray coverage, attributed to the lack of overlapping caused by a three nozzle configuration. The final component of the Miller rainfall simulator uses the solenoid valve during operation by changing the duration of an open/closed cycle to reduce the intermittent nature of simulated rainfall. The intensity of rainfall simulated by McLaughlin and Brown was 34 mm/hr, which was obtained by an on/off-cycle at 10-second intervals. An optional pressure regulator was installed to maintain a constant water pressure of 34 kPa. This rainfall event lasted 5 minutes after runoff was initiated. The rainfall intensity reported by Petersen et al. was 75mm/hr and simulations were conducted for 30 minutes after the 12 initial start of runoff. The water pressure of the system was not specified in the literature by Petersen et al (2007). A similar rainfall simulator used in the research conducted by Peterson et al. (2002) was modeled from recommendations provided by Foster et al. (1982). This design also included the ability to program the simulator for different types of rainfall. This configuration allowed rainfall durations and intensities to vary throughout an experiment. During experimentation of dry and liquid PAM applications as an erosion control measure, four test runs were used, modeling different rain events. The first test run (i.e. ?sub-run?) used an intensity of 75 mm/hr for one hour. After a one hour break, another test run (i.e. ?wet run?) was conducted with the same rainfall intensity used previously, over a 30 minute period. Following a 30 minute break with no rainfall, a third ?very wet? run was broken down into three stages: (1) 75 mm/hr for 15 minutes; (2) 28 mm/hr for 15 minutes; and (3) 100 mm/hr for 15 minutes. These different test runs were selected to simulate the effect of the treatments on different soil conditions (i.e., dry soil, moist field, saturated field). Peterson states that the initial dry run represented a 1- hr, 100-yr rain event (for West Lafayette, Indiana) and the wet and very wet test runs both represented events exceeding a 100-yr return period for their respective durations. A rotating-boom simulator was used in research conducted by Benik et al. (2003), which referenced a design recommended by Swanson (1979). This device applied rainfall in a relatively uniform circular pattern of 15.2 m. The simulator used spray nozzles attached to pressure regulators to maintain a constant pressure of 55 kPa. This configuration provided a rainfall intensity of approximately 60 mm/h. Rainfall was 13 applied in two different intervals, classified as a ?dry? and ?wet? run of 90 and 60 minutes, respectively. Another rainfall simulator used for experiments consisted of an oscillating sprinkler as described by Meyer and Harmon (1979) (Bjorneberg et al., 2000). This setup contained a Veejet nozzle mounted approximately 3 m above the test surface. The desired rainfall intensity was 80 mm/hr at a constant pressure of 76 kPa providing droplet sizes of 1.2 mm. The rainfall duration for their experiments lasted 15 minutes. A non-specified rainfall simulator design was used in research conducted by Roa- Espinosa et al. (1999). In this research, it states that the simulator produced a rainfall intensity of 64 mm/hr for an average duration of 40 to 50 minutes (or until the runoff collection tank was filled). No additional information on the rainfall simulator was provided. The rainfall simulator designed by Foster et al. (1982) was used on multiple experiments that examined the effect of different soil treatments on infiltration, surface runoff, and erosion, including the use of PAM (Flanagan et al., 1997a; Flanagan et al., 1997b; Flanagan et al., 2002a). This design was programmable, which allowed an instantaneous change in rainfall intensity by controlling the frequency of nozzle oscillations. The research conducted in 1997 used a rainfall intensity of approximately 64 mm/hr until a generated runoff hydrograph had achieved a steady-state for a time period of at least 5 minutes (Flanagan et al., 1997a and 1997b). For the study done in 2002, a more sophisticated setup was used that consisted of multiple test runs. The initial test run (i.e. ?dry? run) consisted of a target intensity of 64 mm/hr for a one hour duration. 14 Following a one hour break, the second test run (i.e. ?wet? run) had an intensity of 64 mm/hr for one hour. Following a 30 minute break, a third stage ?very-wet run? was conducted using intensities of 75 mm/hr, 28 mm/hr, and 100 mm/hr, for a duration of 15 minutes each (Flanagan et al., 2002a). The initial dry run was representative of a 25-yr storm even for west-central Indiana. The succeeding wet runs represented a return period greater than a 100-yr storm event. 2.2.1.3 Natural Rainfall used in Experimental Tests Instead of using simulated rainfall, an experiment could be designed to use natural rainfall. Natural rainfall has inherent advantages and disadvantages in comparison to simulated rainfall. One advantage of natural rainfall is that experimental results will accurately model real-life or actual field conditions. The potential treatment is exposed to realistic conditions, similar to when the product would be used in practice. One primary disadvantage is that by using natural rainfall, the research is at the mercy of weather, possibly being delayed due to long periods without rainfall. An area could experience drought conditions, severely hampering the experiment. These unpredictable qualities of natural rainfall can lead to undesirable experimental results and conditions. A study conducted by Hayes et al. (2005) used three field test locations exposed to natural rainfall. Each site was set up in succession to handle the effort of post-storm data sampling. This also ensured no overlap between the three test sites existed. Rainfall intensity for all subsequent storm events was not reported, but the accumulated rainfall for each event was recorded. For their first test location, eight different storm events produced amounts ranging from 1 mm to 66 mm of rain, with a failure of the sediment 15 collection device on the last storm event. The second experiment test location received six rain events ranging from 8 mm to 42 mm. The final test location experienced seven rain events ranging from 5 mm to 34 mm of accumulated rainfall (Hayes et al., 2005). The last study examined here was a continuation of the research conducted by Flanagan et al. (2002). As discussed earlier, the first part of the study used simulated rainfall to test PAM under controlled conditions (Flanagan et al., 2002a). This research was expanded to see how PAM preformed under natural rainfall using 26 test plots spread over two locations. The first site encountered 9 different storm events for a total of 185 mm of rainfall. The second site experienced 17 storm events for a cumulative rainfall of 636 mm (Flanagan et al., 2002b). The results of this study showed similar trends in runoff depth when compared to the previous research effort. 2.2.1.4 Rainfall Summary The application of water to a test site to generate runoff is a critical component to the overall design, methodology, and success of an experiment. Different methods exist to achieve this depending on a project?s overall research goals. Therefore, care must be taken in selecting the proper design to generate runoff. When designing a rainfall simulator, Hall (1970) and Regmi and Thompson (2000) identified key components to ensure simulated rainfall accurately models natural rainfall. Many of the designs presented here adequately addressed these issues to justify the quality of their rainfall simulators. By taking steps to properly design a rainfall simulator, the quality of reported results could be considered accurate. None of the studies indicated any major issues or concerns pertaining with simulated rainfall versus 16 natural rainfall. It is important to note, that the design of any rainfall simulator could affect the size of an experiment, due to limited spray area. A simulator design with one nozzle may constrain the size of test plots. Conversely, simulators can be designed to work on large test section, but may require a greater amount of resources and materials. Table 2.1 provides a summary of the five different factors discussed in this section: (1) type of rainfall (i.e. simulated vs. natural), (2) simulator height, (3) rainfall intensity, and (4) storm duration. Selection of different types of rainfall to use depends on the resources available and overall research goals. Typically, the convenience associated with having a rainfall simulator outweighs the use of natural rainfall. The random nature and uncertainty associated with natural rainfall events (e.g. undesirable amount of rainfall being produced or periods of no rainfall) could seriously impact an experiment. This was seen in some of the studies documented in this section where an unanticipated amount of rainfall occurred, causing failures in the experimental design. Conversely, simulated rainfall might not accurately reflect the properties of natural rainfall, so care must be taken to ensure a simulator is correctly designed, operates effectively, and realistically models natural rainfall events. 17 Table 2.1 Summary of Experimental Rainfall Regimes from Literature Study Type Height (m) Rainfall Intensity (mm/hr) Duration Petersen et al. (2007) Simulated UNK 75 30 min after start of runoff McLaughlin and Brown (2006) Simulated 4 34 5 min after start runoff Sepaskhah and Bazrafshan-Jahromi (2006) Simulated 2.65 96 15 min Hayes et al. (2005)1 Natural UNK 1 to 66c UNK 8 to 42c UNK 5 to 34c UNK Benik et al. (2003) Simulated NA 60 90 min 60 60 min Flanagan et al. (2002)a Simulated 2.4 64 1 hr followed by 1 hr break 64 1 hr followed by 30 min break 64 15 min 28 15 min 100 15 min Flanagan et al. (2002)b Natural UNK 185a,c UNK UNK 636b,c UNK Peterson et al. (2002) Simulated 2.4 75 1 hour followed by 1 hour break 75 1 hr followed by 30 min break 75 15 min 28 15 min 100 15 min Bjorneberg et al. (2000) Simulated 3 80 15 min Roa-Espinosa et al. (1999) Simulated UNK 64 40-50 min or until collection tank was filled Flanagan et al. (1997)a Flanagan et al. (1997)b Simulated UNK 64 Until runoff hydrograph on flume chart recorder leveled off for at least 5 min Notes: ?UNK? indicates that data was not specified in the literature. ?1? Three field locations were tested ?a? Total of 9 events (Location 1) ?b? Total of 17 events (Location 2) ?c? Cumulative Rainfall 2.2.2 Experiments and Procedures In this section, the different devices and methods researchers used to test the effectiveness of PAM on soil surfaces are investigated, along with results reported from the experiments conducted. One method for experimental setup is accomplished by 18 using test plots outdoors that represented field conditions on pre-existing slopes (Petersen et al., 2007; Hayes et al., 2005; Leib et al., 2005; Benik et al., 2003; Flanagan et al., 2003a; Flanagan et al., 2003b; Flanagan et al., 2002a; Flanagan et al., 2002b; Lentz et al., 2002; Peterson et al., 2002; Roa-Espinosa et al., 1999). The site location can vary in size and configuration depending on resources available and the goals established by the researchers. These sites have the ability to produce results that would realistically represent processes observed in the field. Conversely, small or intermediate-scaled test plots (usually conducted indoors under laboratory conditions) allow for a more controllable setup (Sepaskhah and Bazrafshan-Jahromi, 2006; McLaughlin and Brown, 2006; Bjorneberg et al., 2000). Either method has advantages and disadvantages associated with each setup. In addition to looking at the equipment and setups, procedures and methods used during experimentation will be examined. These will include plot preparation procedures, data collection methods, and any additional pertinent information. The following sections will investigate these topics and review results of using PAM as an erosion control BMP included within each research effort. 2.2.2.1 Field-Scale Experiments Typically, experimental setups pertaining to outdoor field conditions are large in size when compared to similar experiments conducted in laboratory conditions. Field- scale setups could be supplemented with large rainfall simulators to offer a level of control similar to that used in an indoor, laboratory setting (Peterson et al., 2007; Benik et al., 2003; Flanagan et al., 2002a; Flanagan et al., 2002b; Peterson et al., 2002; Roa- Espinosa et al., 1999; Flanagan et al., 1997a; Flanagan et al., 1997b). If a rainfall 19 simulator was not used during a field-scale experiment, then the research was dependent on natural rainfall, a type of channelized flow, or irrigation to generate the required runoff (Hayes et al., 2005; Peterson et al., 2003; Flanagan et al., 2002b; Bjorneberg et al., 2000). A study conducted by Peterson et al. (2007) used two field test sites for testing new liquid and emulsified PAM formulations for use on agricultural fields. These sites had on average, slopes ranging from 7% to 9%. At both locations, test plots were created by using sheet metal to border the testing area. These enclosed areas had dimensions measuring 2.0 m x 0.75 m. Soil surfaces were raked to a uniform level. Two test plots were created at each site for a total of four sections. At the down-slope end of each test plot, a gutter system was installed to collect runoff and PVC pipes were used to transport and convey the runoff to collection buckets (Peterson et al., 2007). The PAM was applied using a hand sprayer at a rate of 5 kg/ha. Tests were performed at three different times with two replications, at 2-days, 3-weeks, and 10 weeks from the initial PAM application. With the exception of the 2-day tests, the site was exposed to natural precipitation in between experiment times. Data collection consisted of runoff samples collected every 5 minutes. The samples were dried in an oven for 24 hours and weighed afterwards. Following this, additional runoff samples were taken at 5 minute intervals to determine silt, clay, and sand content. Since this experiment was concerned primarily with agricultural applications, additional runoff samples were taken at 5, 15, and 25 minute intervals for phosphorus analysis (Petersen et al., 2007). Finally, total runoff was collected and 20 measured at the end of each test. Results of the experiment indicated that runoff volumes were reduced 100% at 2-days, 59% at 3-weeks, and 55% at 10-weeks. Sediment loss on test plots also experienced similar reductions of: 100%, 80%, and 74% at their respective time periods of 2-days, 3-weeks, and 10-weeks. Research by Hayes et al. (2005) also used multiple site locations for setting up test plots. The purpose of this research was to examine the effectiveness of two dry granular PAM products, Soilfix and Siltstop 705, mixed with water to form a liquid solution applied with and without seed and mulching. Recommended application rates of Siltstop 705 and Soilfix consisted of 10.5 kg/ha and 1.5 kg/ha, respectively. In addition to examining theses rates, researchers performed tests to determine the effectiveness of the PAM products at half the recommended rate. Slope used for experimentation were similar to those found on construction sites (i.e. 20% to 50%). At one test site, plots were constructed with dimensions of 6 m by 1.5 m. These plots were separated by a 15 cm high plastic barrier driven into the soil (Hayes et al. 2005). Additional plastic barriers were placed down-slope to channel the runoff to a drain pipe connected to collection buckets. The other sites varied slightly by plot spacing and the use of different methods for runoff collection. Data collection consisted of analyzing the runoff in the collection buckets at the end of each storm event. Samples were taken to measure the turbidity and suspended solids. Results collected indicated that the addition of PAM to the seed and mulch provided no significant benefits to volume and sediment reduction. However, researchers 21 acknowledge that steep slopes may require higher application rates than used in this research. Two studies performed by Flanagan et al. (2002a; 2002b) were conducted using field conditions consisting of 9 test plots measuring 2.96 m wide by 9.14 m long. Slope steepness for plots ranged from 32% to 45% (Flanagan et al., 2002a; Flanagan et al., 2002b). PAM treatments, of Percol 336 with an application rate of 80 kg/ha, were mixed with water to form a liquid spray. One of the PAM mixtures included an addition of gypsum. Runoff was collected with a metal trough which directed all runoff for analysis. The first study, using simulated rainfall, collected samples at intervals based upon on the rate of runoff and did not exceed 3 minute time intervals. Data of interest included runoff rate, sediment concentration, and sediment yield rate. Following the completion of all rainfall simulations, additional soil samples were taken from the test plots to determine soil surface conditions (Flanagan et al., 2002a). Results for this study found that treatments of PAM reduced runoff and sediment yield 40% and 83%, respectively. PAM treatments with gypsum performed slightly better, with runoff reduced by 52% and sediment yield by 91%. The second study, using the same experimental setup as the first, was examined under natural rainfall conditions. Test plots were constructed at two locations. Runoff was collected in barrels placed at the end of each test plot. The researchers used multiple methods to collect the runoff, depending on the volume of runoff generated. Since samples were not collected at specific time intervals, data consisted of cumulative amounts of runoff and sediment yield (Flanagan et al., 2002b). Performance of different 22 PAM treatments were similar to the experiments that used simulated rainfall. The first test location experienced runoff and sediment reductions for PAM and PAM with gypsum treatments of: 33%, 54%, 33%, and 45%, respectively. The second test location also experienced similar reductions of runoff and sediment volume of: 15%, 40%, 28%, and 53%. Another study conducted by Flanagan et al. (1997a; 1997b) was a two-part project which looked at PAM?s effectiveness on field plots using simulated rainfall. Three test sites were constructed for experimentation. The first part of Flanagan?s study examined the effect different soil amendments had on infiltration and runoff. One of the treatments was a liquid solution of PAM applied at an application rate of 20 kg/ha. The other treatment consisted of fluidized bed combustion bottom ash (FBCBA). The second part of this research focused on how different treatments affected soil erosion specifically. Test plots were divided to include three, small interrill plots measuring 0.8 m wide x 0.6 m long and three, larger interrill plots measuring 0.8 m wide and 10.7 m long (Flanagan et al., 1997a; Flanagan et al., 1997b). Test slopes ranged from 6% to 9%. Runoff sampling used different size collection containers depending on flow rate generated. Data collection for the first part of the study focused on measuring the flow rate by timing the leading edge flow velocity with a fluorescent dye (Flanagan et al., 1997a). In addition, cross sectional measurements were taken using a laser scanner. Total runoff was calculated from the generated hydrograph and infiltration rate was measured by taking the difference of runoff rates from measured rainfall rate. 23 The second part of Flanagan?s study also recorded leading edge flow velocity. Sediment samples were collected every 3 minutes and concentration was determined by gravimetric analysis (Flanagan et al., 1997b). Cross sectional measurements were also taken to evaluate any change in the plot elevation. Total runoff and sediment rates were calculated from the flow and sediment discharge curves (Flanagan et al., 1997b). Researchers found that for both studies, PAM was effective at significantly increasing infiltration, while reducing runoff and sediment transport. Peterson et al. (2002) used 12 testing plots at one site location to examine the effectiveness of dry and liquid PAM applications of 60kg/ha each. Slopes at the site were graded to approximately 17%. Soil depth was raked and leveled by hand to 0.3 m, then roto-tilled afterwards. Plots were also seeded with an unspecified grass mixture (Peterson et al. 2002). Individual test plots were constructed measuring 9.1 m x 3 m and separated with sheet metal. Samples were collected every 3 minutes and discharge rates were measured by collecting runoff at the outlet trough in buckets at an unspecified amount of time (Peterson et al., 2002). The total runoff was calculated from discharge measurements over time and sediment yield was determined from the discharge rate and average sediment concentration. The research concluded the liquid PAM applications were more effective, by reducing runoff and sediment yield with values ranging from 62% to 76% and 93% to 98%, respectively. Dry granular PAM was observed to have almost no benefit during experimentation The final study using field-scale plots examined was performed by Roa-Espinosa et al (1999). The PAM treatments consisted of: (1) liquid PAM applied to dry soil, (2) 24 dry granular PAM applied to dry soil, (3) liquid PAM with seed and mulch, and (4) liquid PAM applied to wet soil. This study is unique because while the plots were installed outdoors in field conditions, the size of the plots was smaller than any previously conducted research discussed. The study used 15 plots measuring 1 m x 1 m with an unspecified depth on a 10% slope. These smaller plots were easier to set-up and allowed researchers three replications, opposed to a single experiment on a larger scale. Runoff was collected in 1 minute intervals by diverting it into specified collection containers to determine sediment yield (Roa-Espinosa et al., 1999). A representative sample taken from the total collected runoff was oven dried and weighed to determine an average sediment load from each test run. Results from the experiments showed that all soil treatments were effective at reducing sediment yield. The test plots treated with liquid PAM on dry soil reduced sediment yield by 87% and 57%. Plots treated with dry granular PAM on dry soil had an average reduction of sediment yield measuring 34%. The liquid PAM with seed and mulching treatment was determined to perform the best, with an average sediment yield reduction of 93%. Runoff and infiltration were unaffected by the addition of PAM for all replications. 2.2.2.2 Intermediate-Scale Tests A testing configuration smaller in scale, when compared to field-scale testing, potentially allows for more control in the overall experiment. These controls can include a more manageable setup, and one that is less resource dependent when compared to a field-scale test. In addition, smaller laboratory tests could allow for one particular 25 experiment to be replicated, at a faster rate, ensuring an accurate testing procedure and results. A study conducted in Iran used seven steel boxes measuring 1.4 m x 1.4 m (Sepaskhah and Bazrafshan-Jahromi, 2006). The metal boxes had a depth of 0.09 m, and can be filled with any type of soil. At the down-slope end of each box, a flume was constructed to funnel runoff to a collection point. The slope of the boxes could be changed to 2.5%, 5.0%, or 7.5%. PAM treatments were applied through the rainfall simulator at applications rates of 1, 2, 4, and 6 kg/ha for the first test and subsequent tests used untreated water. Data collection consisted of measuring the runoff rate per unit area and infiltration rate. Researchers concluded that test plots configured with a steeper slope required higher application rates of PAM to reduce erosion. Experiments indicated that PAM treatments were more effective in reducing erosion, opposed to runoff and infiltration. Research presented by McLaughlin and Brown (2006) examined the effectiveness of ground cover practices with and without the addition of PAM. Treatments included: (1) PAM on bare soil, (2) straw, (3) wood fiber, (4) straw erosion control blankets (ECB), and (5) mechanically bonded fiber matrix (MBFM). Soil treatments were tested in field and laboratory conditions. The study conducted in the field examined ground cover with and without PAM and its effect in establishing long-term vegetation under natural rainfall. Test plots were constructed on a 4% slope in 1 m x 1 m sections. Laboratory experiments, using a rainfall simulator, included four intermediate-scaled boxes constructed out of wood. Each box measured 1 m wide x 2 m long with an allowable soil 26 depth of 0.09 m. When soil was added, it was leveled by hand and no additional compaction was performed. The rainfall simulator could only be used over two plots, so this allowed researchers to setup the proceeding experiment while the initial one was running. The slope of the boxes could be changed to either 10% or 20%. Small drain holes were drilled along the bottom of the down-slope side to reduce the amount of ponding at the lower end (McLaughlin and Brown, 2006). Runoff was collected with plastic gutters attached to the down-slope end of each plot. These were used to collect runoff volume, turbidity, and suspended solids. Samples were then oven dried to determine the quantity of eroded soil. Results from the field-scale setup indicated that the performance of the straw, wood fiber, ECB, and MBFM all significantly reduced runoff volume, turbidity, and total sediment loss. However, the addition of PAM provided little or no added benefits, with no effect on establishing vegetation. The intermediate-scaled laboratory test produced similar results, demonstrating that the addition of PAM had no significant result in the performance of the ground cover practices examined. PAM was effective in reducing turbidity for the first and second rain events simulated, but did not consistently improve runoff quality with additional tests (McLaughlin and Brown, 2006). The last study discussed here that used intermediate-scaled laboratory tests was conducted by Bjorneberg et al (2000). In this research, PAM was tested in conjunction with different methods of applying straw. Six steel test boxes were constructed with dimensions measuring: 1.5 m long, 1.2 m wide and 0.2 m deep. Soil was filtered through a sieve and mixed prior to leveling. This resulted in a slightly compacted surface. At the 27 down-slope side, the depth measured 0.15 m to provide for the installation of a runoff trough to funnel runoff into containers (Bjorneberg et al., 2000). Test plots were hinged to allow for varying slopes from 0 to 15%. Additional drainage tubes were placed to allow any excess water through infiltration to be collected, though later experiments proved this was not necessary. Tests examined the effectiveness of different amounts of straw coverage with and without an application of PAM. PAM was applied at a rate of 2 and 4 kg/ha with an irrigation sprinkler and allowed to dry for 7 to 10 days. After the experiment was conducted, the collected runoff was weighed and examined for any straw. The rest of the runoff was filtered to determine total amount of eroded soil. Researchers observed that PAM treated test plots with straw coverage of 70% reduced runoff by 75% to 80%. PAM treated plots on bare soil only reduced runoff with values ranging from 30% to 50%, which were similar to plots tested with only 30% straw coverage. 2.2.2.3 Procedures and Methods Summary Many different methods exist for testing the effectiveness of erosion control practices. Along with different procedural steps in the experimental procedure, the physical construction and configuration of experiments varied. Scale is one of these physical components, which are necessary to define during experimental design. Table 2.2 summarizes the different parameters and configurations used in the literature discussed in this section. By choosing to conduct experiments out in the field, conditions experienced will accurately model real-life situations and performance of the treatment will be similar to when it is used for real world applications. However, these setups tend 28 to be large in scale and require additional resources and expenses. Time to construct and install these testing plots is greater and makes it difficult to obtain results which could be easily reproduced. This may affect any future comparison of results with additional experimental or actual treatments used in practice. Experiments conducted in a laboratory setting offers researchers an extra level of control not found in field-scale testing. Desired soil can be selected depending on research goals and can be easily monitored throughout an experiment. Plot size can vary and usually smaller than what is seen out in the field-scale tests. In addition, the required resources to construct these intermediate-scale setups can be easily acquired and built. Table 2.2 Summary of Experimental Testing and Procedures Study Type Test Plots Slope Length (m) Width (m) Depth (m) Petersen et al. (2007) Field Test (2)* 4 3% to 9% 2 0.75 UNK Sepaskhah and Bazrafshan-Jahromi (2006) Steel Boxes 7 2.5, 5.0, 7.5% 1.4 1.4 0.1 McLaughlin and Brown (2006) Wood Boxes 2 10% to 20% 2 1 0.1 Hayes et al. (2005) Field Test (3)* 30 20% to 50% 6 1.5 UNK Flanagan et al. (2002)a Field Test 9 32% to 45% 9.14 2.96 UNK Flanagan et al. (2002)b Field Test 9 32% to 45% 9.14 2.96 UNK Peterson et al. (2002) Field Test 12 17% 9.1 3 0.3 Bjorneberg et al. (2000) Steel Boxes 6 0% to 15% 1.5 1.2 0.2 Roa-Espinosa et al. (1999) Field Test 15 10% 1 1 UNK Flanagan et al. (1997)a Field Test 6 6% to 9%% 0.6 and 10.7 0.8 0.3 Flanagan et al. (1997)b Field Test (2)* 6 6% to 9%% 0.6 and 10.7 0.8 0.26 Notes: ?UNK? indicates that the data was not specified in the literature. ?*? indicates number of site locations used 29 Large-scale field test require grading equipment and experienced operators. Intermediate-scale tests allow researchers to setup and conduct experiments more quickly than when compared to field-scale. This allows for a greater frequency of experiments, which more data can be produced, analyzed, and compared. However, depending on the experimental design, a laboratory setting may not accurately reflect actual conditions found in the field. Care and effort must be used when deciding on setup and procedures used for testing. Overall, the issue of scale is relevant to the available resources and research goals. Both methods have distinct advantages and disadvantages associated with each. As long as the methods use sound engineering judgment and provide justification, the experiment should result in viable data. Methods used for an experimental setup are another part of the design process that is heavily dependent on resources and research goals. Generally, experiments pertaining to agricultural purposes used disturbed soil which had little or no compaction. Also, slopes used in such research tended to be mild, not exceeding 10%. These experiments provide valuable insight into similar procedures and methods, which can be adapted and modified to fit needs of research outside the realm of agricultural applications. 2.2.2.4 Summary of Results While many experiments conducted in previous research vary in procedures, all are interested in the performance of PAM as an erosion control measure. PAM has been documented as a means to reduce runoff through promoting infiltration and reducing sediment yield either by improving flocculation or maintaining soil surface structure. 30 However, results generated from research provide a widely varying account of the effectiveness of PAM. Table 2.3 and Table 2.4 summarize results presented from the literature reviewed in this section. The use of PAM as a liquid (i.e. dry PAM/water mixture or emulsified PAM) appears to be the predominant method of application. Most likely this is driven by applications methods contractors use in the field. Typically, the primary goal of any erosion control plan is the long-term establishment of vegetation. Seed mixtures are commonly applied in the field by using a hydro-seeder, so no additional equipment is required to apply a treatment of PAM to a site. However, very little research has been conducted analyzing and comparing the use of dry granular PAM applied directly to the soil surface versus liquid PAM. Only two studies presented examined this comparison (Roa-Espinosa et al., 1999; Peterson et al., 2002). In both cases, PAM was effective in reducing erosion and sedimentation, with liquid PAM performing better than dry PAM. Flanagan et al (2003) concluded that when PAM applied as a liquid spray and allowed to dry will perform better than dry granular PAM at immediately reducing erosion. However, this time to allow for drying may not be feasible, depending on climate conditions. Previous research does not indicate liquid PAM?s performance when subjected to rain soon after application. The severity of slope is another factor that needs to be taken under consideration when examining PAM?s effectiveness. Typically, slopes associated with a construction sites are compacted and steeper than slopes found in agricultural settings. Some researchers doubt whether PAM can function effectively on steep slopes, while it was 31 observed in some studies that conditions with steeper slopes required higher application rates of PAM. Table 2.3 illustrates results generated from experiments configured for mild slopes, (i.e. less than 10%). Generally, the application rates used were low when compared to values pertaining to steep slopes (i.e. greater than 10%), as reported in Table 2.4. This can also be observed when comparing studies prepared by McLaughlin and Brown (2006) and Hayes et al. (2005). Both of these studies observed that PAM had no significant effect on soil plots; however, these experiments were conducted with low application rates on steep slopes configurations. Finally, comparing results from Table 2.3 and Table 2.4, no appreciable difference was observed when comparing an experimental design using simulated versus natural rainfall, or large field-scale versus intermediate-scale setups. The main unifying feature common in all studies presented in this section indicates that experiments were conducted on un-compacted soil, which is atypical to conditions present on construction sites. 32 Table 2.3 Summary of PAM Treatments Applied to Mild Slopes (i.e. < 10%) Study Rainfall Plot Slope Application Rate Application Type Results Flanagan et al. (1997a, b) Simulated 6% to 9% 20 kg/ha Liquid Significantly reduced runoff and sediment transport Bjornberg et al. (2000) Simulated 2.40% 2 and 4 kg/ha Liquid PAM with 70% straw coverage: runoff reduced by 75 to 80% PAM on bare soil: runoff reduced by 30 to 50% Sepaskhah and Bazrafshan- Jahromi (2006) Simulated 2.5%, 5.0%, 7.5% 1, 2, 4, and 6 kg/ha Liquid Steeper slopes required higher PAM applications. PAM was effective in reducing erosion Petersen et al. (2007) Simulated 3% to 9% 5 kg/ha Liquid Runoff volume was reduced 100% after 2 days, 59% after 3 weeks, and 55% at 10 weeks. Sediment loss was reduced 100% after 2 days, 80% after 3 weeks, and 74% at 10 weeks. McLaughlin and Brown (2006)1 Natural 4% 19 kg/ha Liquid 2 PAM did not provide a significant benefit in most cases Lentz et al. (2002) Irrigation 1.50% 1.8 kg/ha Liquid 3 Sediment loss reduced by an average of 82% Peterson et al. (2003) Channelized 1% 80 kg/ha Liquid Sediment reduction ranged from 93 to 98%. PAM was effective in controlling erosion in earth channels, according to cross-section measurements. Notes: ?1? Research examined PAM on field conditions (natural rainfall) and laboratory conditions (simulated rainfall). ?2? PAM treatments: (1) PAM on bare soil, (2) straw with and without PAM, (3) wood fiber with and without PAM, (4) straw erosion control blankets (ECB) with and without PAM, and (5) mechanically bonded fiber matrix (MBFM) with and with PAM. ?3? PAM added to furrow irrigation 33 Table 2.4 Summary of PAM Treatments Applied to Steep Slopes (i.e. > 10%) Study Rainfall Plot Slope Application Rate Application Type Results Roa-Espinosa et al. (1999)1 Simulated 10% 22.5 kg/ha Dry and Liquid All soil treatments significantly reduced sediment yield of test plots. PAM with seed and mulch was most effective. Flanagan et al. (2002a)2 Simulated 32% 80 kg/ha Liquid PAM treatment reduced runoff by 40% and sediment yield by 83% PAM & gypsum treatment reduced runoff by 52% and sediment yield by 91% Peterson et al. (2002)3 Simulated 16.6 % (? 3%) 60 kg/ha Dry and Liquid Liquid PAM was more effective, reducing runoff by 62 to 76% and sediment yield by 93 to 98%. Dry PAM provided almost no benefit. McLaughlin and Brown (2006)4 Simulated 10% and 20% 19 kg/ha Liquid 5 PAM did not provide a significant benefit in most cases Flanagan et al. (2002b)2 Natural 35% and 45% 80 kg/ha Liquid PAM reduced runoff by 33% (15%)6 and sediment yield by 54% (40%)6 PAM & gypsum reduced runoff by 33% (28%)6 and sediment yield by 45% (53%)6 Hayes et al. (2005) Natural 20% and 50% 10.5 kg/ha* and 1.5 kg/ha* Liquid 7 PAM had no significant result on runoff volume, turbidity, and sediment loss Notes: ?1? Treatments: (1) liquid PAM applied to dry soil, (2) dry granular PAM applied to dry soil, (3) liquid PAM with seed& mulch, and (4) liquid PAM applied to wet soil. ?2? PAM treatments: Liquid PAM with and without the addition of gypsum. ?3? PAM treatments: (1) liquid PAM and Nutra-Ash, (2) dry granular PAM and Nutra-Ash, and (3) liquid PAM with SoilerLime. ?4? PAM treatments: (1) PAM on bare soil, (2) straw with and without PAM, (3) wood fiber with and without PAM, (4) straw erosion control blankets (ECB) with and without PAM, and (5) mechanically bonded fiber matrix (MBFM) with and with PAM. ?5? PAM added to furrow irrigation ?6? Values in ( ) indicate results from second field location. ?7? Two types of PAM used: Soilfix and Siltstop 705. PAM treatments: seed/mulch with and without PAM. ?*? PAM was also tested at half this rate 34 2.3 ANALYSES This section examines the different ways researchers examined and analyzed data collected from erosion control experiments. This will focus on how researchers used different statistical tests to evaluate the collected data and how conclusions were developed from reported analyses. 2.3.1 Statistical Testing Statistical testing provides researchers with powerful tools to examine collected data and quantify a particular treatment or experiment to determine its overall effectiveness. Determining if experimental results were statistically significant provides a common and widely accepted method of comparing data. Many different types of statistical test are available and used throughout the research presented. Therefore, care must be taken when choosing which type of analysis should be used to ensure that results are accurately analyzed and reported. This section will examine different methods and techniques used when analyzing data collected from erosion control experiments. In the study by Sepaskhah and Bazrafshan-Jahromi (2006), to determine if the runoff depth and amount of eroded soil was statistically significant, a Duncan multiple range test with a 5% level of probability was used. This test method uses, ?multiple comparisons in which the group means are ranked from smallest to largest, and then the number of steps that two means are apart in this ranking is used to compute a range statistic for each comparison,? (Colman, 2001). In addition to using this statistical test, a multiple regression analysis was conducted on soil erosion and slope to determine if there was a relationship between the two results. Similarly, another regression analysis was 35 conducted on the relationship between soil loss and runoff depth (Sepaskhah and Bazrafshan-Jahromi, 2006). Research conducted by McLaughlin and Brown (2006) reported on whether or not the PAM treatment had a significant effect on the tested parameters. The specific test used to determine significance was not specified. The level of significance was reported to be 5% and sometimes 10% if required. Data tested included runoff amount, eroded soil, turbidity, and grass cover (i.e. as observed in field-scale experiments). Hayes et al., (2005) used an analysis of variance (ANOVA) to determine if the results were significant. ANOVA is used to compare multiple treatments by assessing equality of several recorded means (Ramsey and Schafer, 2002). This type of statistical test provides researchers a powerful tool to compare multiple treatments, and further analyses can be used to examine individual comparison within multiple treatments. In this research conducted by Hayes et al., (2005) treatment pairs were compared using a Tukey-Kramer test to determine a statistically significant difference between runoff volume, turbidity, and sediment loss. Experiments performed by Leib et al., (2005) also used ANOVA to test significance between the different treatments by examining the mean separation by using least square difference (LSD). The collected data was transformed before analysis. Level of significance used for testing was 5%. After the ANOVA analysis, results were transformed back to the original units. In the two part research by Flanagan et al., (2002a; 2002b) statistical analysis was performed on runoff volume and sediment yield using ANOVA procedures. Additional analyses were performed on individual runs, examining cumulative runoff volume, runoff 36 rate, sediment concentration, sediment discharge rate, and total sediment loss. This was performed by examining the mean separation using LSD method. Level of significance selected for testing collected data was 5%. Previous research by Flanagan et al., (1997a, 1997b) also used ANOVA procedures for both parts of the effort. The first part used LSD and a level of significance of 10% on runoff and infiltration rates. The second part of the study used LSD if treatments were found to be significant at the 10% level. 2.3.2 Analyses Summary Statistical analysis can be a valuable tool to researchers when working with data collected from experiments. By using proven methods to analyze data, the accuracy of experimental results can be validated and compared. In addition to using a widely accepted practice of analyzing data, the results from statistical tests can be easily conveyed when presenting conclusions and determining the effectiveness of different treatments. By quantifying the results through these techniques, researchers can determine which treatments perform the best. However, many differences and preferences exist when it comes to statistical testing. Some of the literature reviewed for this research reported statistically significant results, without describing which tests or procedures were used in the analysis (McLauglin and Brown, 2006; Bjorneberg et al., 2000). While results of their prospective experiments were presented, other researchers performing similar experiments do not have the ability to compare results, make it difficult to evaluate the procedures used in their experiments, while also making comparisons to research results. Other literature reviewed made no effort to use or report on statistical methods to analyze 37 data (Petersen et al., 2007; Peterson et al., 2003; Lentz et al., 2002; Peterson et al., 2002; Roa-Espinsoa 1999). 2.4 LITERATURE REVIEW SUMMARY Experimental procedures can differ on a wide array of issues. The two main factors attributed to experimental procedures include available resources and research goals. These effectively control the way an experiment is conducted. Available resources can refer to construction, materials, and necessary labor associated with a project. Smaller research efforts may not have the necessary funds and means to construct a large testing facility. Additionally, goals that the intended research may be interested in examining may require such a setup. So these items can determine how testing will proceed. In this chapter, researchers examined some of the advantages and disadvantages associated with different types of experimental setups used and the performance of PAM as an erosion control measure. This included the differences in how runoff can be generated and applied to an experiment. Generally this is accomplished through exposing an experiment to natural rainfall or by constructing a rainfall simulator. It was observed that both methods are widely used. It was accepted that a rainfall simulator can effectively provide simulated rainfall, given that the design accurately models natural rainfall conditions. The other key part of an experiment design and the procedures depends on how an experiment was prepared. This can include the scale and location of an experiment. Intermediate-scale experiments, which can be conducted in laboratory settings, can allow for more controllable conditions. Large-scale experiments conducted in the field provide 38 conditions that accurately mimic those found in practice. The physical construction of test setups also varied and depended on different experimental parameters. Researchers? opinions on the use and effectiveness of PAM were observed to be widely different. Overall, it is accepted that liquid PAM can be an effective means to control erosion. Many studies indicate that the amount of eroded soil was significantly less when compared to treatments without an application of PAM. The use of liquid PAM appears to be widely accepted as the preferred method of application, while research examining the effect of dry PAM reported that it did not perform as well as when applied in a liquid form. The physical layout of a soil surface also appears to affect how PAM will perform. When steeper slopes were examined, a higher application rate of PAM was required to achieve satisfactory results. The final part of this chapter examined different analyses used in erosion control experiments. By using statistical tests, results generated can be effectively compared using a unified and widely accepted method of analysis. However, which statistical test depends mainly on the experimental design. Overall, many researchers did not include any form of statistical analysis, and some of the literature reviewed did not specify details involved in the reported statistics. It was also observed that ANOVA procedures to compare multiple treatments were found to be most effective in analyzing experimental data. Based on the research examined in the literature review, the difference in performance between the various PAM applications methods (i.e. dry or liquid form) have not been thoroughly documented and additional research should be conducted to determine if one application method is superior to the other with varying conditions. 39 Additionally, existing research efforts investigating the addition of PAM as an erosion and sediment control measure were primarily focused on agricultural purposes. As reported, construction sites are one of the main contributors to NPS pollution, so research focusing on condition representative of that found on construction sites needs to be expanded upon to include testing procedures focusing on steep, compacted slopes. Therefore, the research presented in this report will focus on these areas to contribute additional results on the use and performance of PAM, and its effect in reducing erosion and sedimentation from surface stormwater runoff, similar to that generated on construction sites. This would be accomplished using intermediate-scale tests with a rainfall simulator. This experimental setup was shown to have attractive features that help contribute to quality data generation, which will enable researchers to analyze and determine the effectiveness of PAM. 40 CHAPTER THREE INTERMEDIATE-SCALE METHODS AND PROCEDURES 3 3.1 INTRODUCTION Upon reviewing previous research, it became apparent that many different techniques, setups, designs, and procedures are available and used for erosion and sediment control experimentation. The research presented in this report is a continuation of research previously conducted; therefore similar ideas and practices were used and modified to appropriately reflect new concepts, designs, and conditions. Therefore, research conducted by Halverson (2006) and McDonald (2007) at Auburn University was used as a general guideline for this research effort. However, potential limitations found in the original designs were identified and addressed for future experimentation. These included areas pertaining to: (1) designing and constructing new intermediate-scale test boxes, (2) designing, constructing, and testing a new rainfall simulator, (3) test preparation procedures, (4) and data collection methods. 3.2 INTERMEDIATE-SCALE TESTING At the onset of this research it was decided to conduct experiments using an intermediate-scale testing apparatus. This was done to examine the effects of PAM in a more controlled environment, while using methods that were less-resource dependent. 41 By using intermediate-scale test plots, experiments could be set-up and conducted with relative ease, speed, and accuracy. A goal was established during the experimental design process to develop a means to conduct any erosion and sediment experiment quickly to allow for replications, ultimately resulting in more data for analysis. The first step that was taken in creating a new experimental design involved development of a dedicated space for this research. In this section, the design and specifications of the newly constructed test facility for intermediate-scale testing will be discussed. 3.2.1 Intermediate-Scale Test Facility A dedicated space was constructed for this research?s intermediate-scale experiments. This space was custom built and contained features specifically tailored for this research. Land was made available by the NCAT test facility located near Opelika, Alabama. The proposed building measured 20 x 30 ft (6.1 x 9.1 m) with two drum roll up doors at opposite ends of the structure, and a concrete slab poured for the building?s foundation, as pictured in Figure 3.1(a). Water was supplied by a nearby underground well and fed into the building using two faucets located at the northeast and southeast corners. Electricity was installed and provided indoor lighting and electrical outlets. Jersey barriers were arranged outside the building as a means for storing test soil as pictured in Figure 3.1(b). When soil was stored in these bins, a tarp was used to cover and protect soil stock piles against rainfall and excess moisture. The facility provided ample space for conducting experiments and could easily be expanded upon. The intermediate-scale test building includes room to store equipment, 42 testing apparatus, and material required throughout the study. Additionally, due to the close proximity to the test track facility, WI-FI internet access is available as well. Figure 3.2(a) and 3.2(b) illustrate the interior of the building and the laboratory setup, including equipment. (a) Testing Facility Structure (b) Storage Bins for Soil Figure 3.1 Test Facility Exterior. (a) Facing East Side of Building (b) Facing West Side of Building Figure 3.2 Test Facility Interior. 43 3.3 EXPERIMENTAL DESIGN After the new building was constructed, researchers were able to focus on elements of a new experimental design. The new design for experiments was derived from research originally conducted by Halverson (2006), where his work focused on testing different configurations of silt fences on an intermediate-scale representation of a highway embankment. As stated earlier, this design was further modified by McDonald (2007) to determine dry granular PAM?s effectiveness as an erosion control measure on three intermediate-scale test plots. However, designs used for both research efforts resulted in experimental setups that were time consuming and labor intensive, due to the size of the testing apparatus. In addition, researchers were not satisfied with the rainfall simulator?s overall performance. Therefore, it was decided to address these issues and develop a new experimental design for use at the new testing facility. 3.3.1 Intermediate-Scale Test Plots The first part of a new experimental methodology was to redesign the test plots. It was decided that the added benefits of intermediate-scale testing allowed for more flexibility and control in conducting experiments. Also, with intermediate-scale test plots, effort required to prepare an experiment would be reduced dramatically. To ensure that a new design would meet these requirements, the smallest acceptable plot size was determined to have an approximate width of 24 inch (61 cm) with and a length of 48 inch (122 cm). These dimensions were approximately half that used previously by McDonald (2007). The previous design allowed for a 6 inch (15 cm) depth for soil. The bottom 3 44 inches (7.6 cm) was filled with an Expanded Polystyrene (EPS) material. This allowed researchers to reduce the overall load on the supporting structure, in addition to assisting in facilitating infiltration, due to the very high hydraulic conductivity of the EPS material. On top of the EPS layer, a geotextile was installed, and then an additional 3 inches (7.6 cm) of the desired test soil. This whole process was time consuming to set-up and reset for additional experiments. During experimentation, no appreciable infiltration data was collected. Therefore, it was determined that the addition of an EPS material and geotextiles could be removed in favor of a smaller overall depth. The new selected depth allowed for approximately 3 inches (7.6 cm) of compacted soil. The design and construction of the boxes used were similar to test plots recommended by McLaughlin and Brown (2006), but with smaller dimensions. The specifications and design drawings for the new test plot are illustrated in Figure 3.3. Two test plots were built for simultaneous experimentation. Plots were constructed from pressurized timber, consisting of a 1/2 inch (1.27 cm) plywood base and two-by- fours to form the perimeter. At the down-slope end, a metal strip with 3/8 inch holes drilled was installed to prevent ponding, as recommended by McLaughlin and Brown (2006). A PVC pipe was cut in half and installed below the metal strip to potentially collect any infiltration. However, it was observed during preliminary testing that any resulting infiltration was insignificant and no additional care was taken to collect this data. Also, at the down-slope end, commercial plastic gutters were fabricated and installed to act as a runoff collection device. Figure 3.4 shows the gutters attached to one of the test boxes. Gutters were installed and angled to promote runoff discharge. A 45 gutter rainfall guard was also attached to prevent any rainfall from interfering with surface runoff volumes, as displayed in Figure 3.5(a). (a) Top view (b) Front view (c) Profile view Figure 3.3 Design and Specifications of a Test Box. Figure 3.4 Runoff Collection Device. 46 By using the intermediate-scale test boxes, a greater amount of flexibility was made available, due to the reduction in weight and ease of setup. Two commercial sawhorses were purchased to support the test boxes. These sawhorses are adjustable, allowing a wide range of available slopes for testing. Cinderblocks were used to support the down-slope end and raise the elevation of test boxes to allow for collection buckets to be placed under the gutter discharge point. This testing setup allowed for a test slope of 3:1 to be established. Photos of the completed test plots are displayed in Figure 3.5(a) and 3.5(b). (a) Constructed Test Boxes with Rain Guards (b) Adjustable Slopes with Saw Horses Figure 3.5 Constructed Test Boxes used for Experimentation. This size of these test plots allowed for a more manageable setup, where a single person could lift and adjust an empty box, and only required two people to lift a fully loaded box, ready for experimentation. These boxes helped reduced the time needed for set-up and clean-up, when compared to previous experiments. This allowed more experiments to be conducted, increasing the amount of data available for analysis. 3.3.2 Rainfall Simulator The rainfall simulator used by Halverson (2006) and McDonald (2007) produced a large spray area to ensure coverage over the entire experiment. This was achieved by 3 1 Rain Guards 47 using ? in. (1.9 cm) schedule 40 PVC pipes and consisted of six 1/8HH-3.6SQ FullJetTM spray nozzles and an F-405 Series In-Line Flow meter. However, by using this design, the six nozzles created areas in which rainfall overlapped, resulting in areas receiving concentrated amounts of rainfall. Also, researchers observed that once flow was shut off after the conclusion of a test, water remaining in the system would continue to fall on test plots. To address these issues, a new rainfall simulator was designed for use in the intermediate-scale test building. The issue of overlapping sprays was an area of great concern for producing uniform rainfall distribution. By allowing certain areas to receive greater amounts of rainfall, the authenticity of the collected results could be affected. Therefore, one of the goals for a new rainfall simulator design was to eliminate the possibility of overlapping spray areas. To achieve this goal, one nozzle would be used in the design, eliminating any possibility of overlap. A one nozzle rainfall simulator would allow for a more uniform spray area and no significant areas of concentrated rainfall would fall on the test plots. Another area in the previous rainfall simulator design was related to water flow. Researchers observed that rate of water flow varied and would not remain constant throughout the duration of an experiment. Steps were taken to help control water flow by including a pressure regulator into a new design. By having a pressure regulator, the inflow of water could be kept at a constant pressure, allowing researchers to have greater control over rainfall. This includes the ability to change the rain regime and test different storm events. 48 The third and final change made in the design was to address the issue of water remaining in the system once the simulator was shut off. As stated earlier, water would continue to fall from nozzles onto test plots after the simulator was turned off. This additional water was undesirable and was identified as an area which could be improved upon. To solve this problem, a solenoid valve was introduced into the design. This valve, when shut off, instantaneously closes; ensuring that the flow of water will cease at the conclusion of an experiment. With these key issues from the original design identified, a new rainfall simulator was designed using similar features recommended by McLaughlin and Brown (2006). Drawings for the new design are shown in Figure 3.6. (a) Plan View (b) Profile View Figure 3.6 Rainfall Simulator Design. 49 The construction of the rainfall simulator, as indicated in Figure 3.6, consists of two-by-fours, a ? inch (1.27 cm) diameter steel pipe, support braces, a garden hose, and electrical wiring for the solenoid valve. The rainfall simulator was attached to steel support members in the building?s structure using a ? inch (1.27 cm) bolt. The overall rainfall simulator extended 5 ft (1.5 m) from the building wall. The maximum allowable height (as restricted by the building?s ceiling) was 10 ft (3 meters) from floor level. Multiple holes were drilled into the support member to allow for an adjustable height, with a lowest allowable height of 6 feet (1.8 meters) from floor level. A hose was attached to the pressure regulator down to a faucet in the northeast corner. The pressure regulator installed was a NorgrenTM R43-406-NNLA with ? inch (1.27 cm) port sizes. Attached to the pressure regulator was a gauge to observe the operating water pressure. This allowed for any necessary adjustment to be made during operation to ensure a constant flow was achieved. Four feet (1.2 m) of ? inch (1.27 cm) steel piping connected to the pressure regulator ran along the length of the simulator into the solenoid valve. An ASCOTM 2-way 8210 series solenoid valve was installed to control water shutoff. This valve, when de-energized, remained closed. Care was taken to ensure the electrical wiring for the solenoid was water tight, which lead to a nearby wall switch. Connected directly into the solenoid valve was a FullJetTM ?HH - 30WSQ nozzle, with a wide angle uniform square spray area, and medium to large drop size distribution, similar to natural rainfall. Detailed specifications for the pressure regulator, solenoid valve, nozzle, and raindrop sizes can be seen in Appendix A. Figure 3.7 illustrates the constructed rainfall simulator as installed in the testing facility. 50 Figure 3.7 Illustration of Constructed Rainfall Simulator. With the new rainfall simulator complete, additional tests were conducted to determine the overall performance of this new design. These test examined the nozzle manufacturer?s claim of uniformity and determining a rain regime best suited for this research. To determine and analyze rainfall uniformity, an 8 ft x 8 ft (2.4 m x 2.4 m) grid was placed beneath the simulator. This area was approximately the total spray area generated by the rainfall simulator. At 1 ft (0.3 m) intervals, using intersecting gridlines, marks were placed on the slab and 1 quart (946 ml) containers were placed at these locations to collect rainfall, as pictured in Figure 3.8(a). The uniformity of rainfall distribution was collected during three trial runs at three different operating pressures of 5, 10, and 20 psi for a total duration of 10 minutes. The average recorded volumes (in ml) for the three trials performed at 10 psi can be seen in Figure 3.8(b). 51 ml A B C D E F G H I 1 75 75 92 100 60 2 65 75 100 112 132 147 100 3 83 100 117 140 150 150 150 150 100 4 100 140 150 175 188 170 157 150 122 5 132 153 200 227 207 200 150 150 125 6 147 160 200 225 200 150 150 133 100 7 150 157 150 150 140 132 115 8 123 135 132 123 107 100 75 9 92 75 60 (a) Collection Cups on 8? x 8? Grid (b) Average Volume in Collection Cups at 10 psi Figure 3.8 Rainfall Distribution (in ml) Verifying Uniformity. The chart in Figure 3.8(b) is color-coded to coincide with volume of rainfall, with red indicating higher volumes and blue indicating lower volumes. The squares in gray color were grid marks with no cups placed on them, due to the limited number of cups available for use. The internal square, denoted by the four corners, C3, G3, C7, and G7, was representative of the location of two test plots with an area of 16 ft2 (1.5 m2) as if they were placed under the rainfall simulator during an experiment. To check uniformity, the Christian Uniformity Coefficient, shown as Equation 1, was used (ASAE Standards, 2000). This equation determines a value that quantifies rainfall distribution uniformity as a percentage. The calculated values using Equation 1 for the rainfall simulator in this study are shown in Table 3.1. Generally, a calculated uniformity coefficient ranging from 80% to 100% is considered an acceptable amount for quantifying a rainfall spray area as uniform. 52 1 1 100 1 n i i c n i j v v CU v = = ? ?? ? ? ? ?= ? ? ? ? ? ?? ? ? ? (1) where, CUc = Christian Uniformity Coefficient, n = number of collectors, vi = volume of water in the ith collector, and ?v = mean volume of water in all collectors. Table 3.1 Christian Uniformity Coefficients for Rainfall Simulator Trial Entire 8? x 8? Spray Area Test Plot Area 5 psi 1 63% 87% 2 67% 88% 3 67% 88% Average 66% 88% 10 psi 1 76% 84% 2 76% 88% 3 76% 85% Average 76% 86% 20 psi 1 73% 83% 2 74% 85% 3 73% 84% Average 73% 84% Note: ?Test Plot Area? is the 16 ft2 area located directly under rainfall simulator nozzle. Table 3.1 displays the calculated Christian Uniformity Coefficient for each of the trial runs, including averages. Uniformity of rainfall was examined over the entire spray area, in addition to previously mentioned interior area that was representative of the test plots location. Overall, uniformity for the entire spray area was generally unacceptable at each of the different operating pressures, with values ranging from 63% to 76%. The pattern of spray distribution can be observed in Figure 3.8(b), as recorded volumes on the 53 spray-area-border were less when compared to amounts collected towards the center. However, uniformity of rainfall where testing plots would be placed was much higher, as also seen in Figure 3.8 (b) and Table 3.1. The Christian Uniformity Coefficient for this area was calculated and ranged from 83% to 88%, which was deemed as an acceptable amount of uniform distribution for simulating rainfall during experimentation. Therefore, rain falling specifically on the test plots was determined to be uniform and suitable for experimentation. The next step in analyzing the rainfall simulator was to determine the rate of water outflow. In the manufacturer specifications for the FullJetTM nozzle, capacity of outflow, in gallons per minute, was given as a function of internal water pressure. These rates were verified by researchers and determined to be consistent with what was provided by the nozzle manufacturer. Flow rate was translated as rainfall intensity to determine potential rain regimes. Capacities in gallons per minute were converted into rainfall intensity by using unit conversions and a spray area, with dimensions of 8 ft x 8 ft (2.4 m x 2.4 m), to obtain rainfall intensities in inches per hour. These different calculated rainfall intensities were plotted as a function of different operating pressures and can be seen in Figure 3.9 below. It was observed that the rate of water exiting the nozzle had a linear relationship to water pressure. A regression line was determined and plotted as seen in Figure 3.9. The regression line equation was used to determine any future rainfall intensities. Researchers recommended that pressures during experimentation not exceed 30 psi, due to the limited range of the pressure gauge attached to the pressure regulator, which had a maximum observable pressure of 30 psi. With this graph, resulting rainfall intensities 54 and operating pressures required could be used to select a rain regime for experimentation. y = 0.159x + 2.800 R? = 0.980 0 1 2 3 4 5 6 7 8 5 10 15 20 25 30 35 Ra inf all In ten sit y ( in/ hr ) Pressure (psi) Figure 3.9 Rainfall Intensity (in/hr) as a Function of Rainfall Simulator?s Operating Pressure (psi). 3.3.3 Rain Regime During the literature review, it was observed that rainfall regimes varied. Therefore, for this research, a rainfall regime was selected based off the stormwater inspection guidelines provided by the Alabama Department of Transportation (ALDOT). These guidelines state that an inspection of any erosion BMP will occur within 72 hours of a ?qualifying event?, where ALDOT identifies a ?qualifying event? as 0.75 inches (1.9 cm) of rain accumulation within a 24 hour period (ALDOT, 2004). This 0.75 inches was used as a baseline for the selected rainfall regime simulated during experimentation. Therefore, these inspection requirements provide an amount of rainfall that should be achieved during an experiment to provide data which is representative of a qualifying event experienced in the field. Table 3.2(a) shows the potential rainfall intensities 55 generated from the rainfall simulator with respective water pressure, and calculates total accumulated rainfall based on corresponding rainfall durations. Since 0.75 inches was selected as the minimum amount of rainfall required, researchers selected an operating pressure of 10 psi, which generates an intensity of 4.39 in/hr. A total duration of 15 minutes would produce an amount of rainfall approximately 1.10 inches, which is above the established baseline of 0.75 inch and deemed acceptable for experimentation. Details for this rain regime are illustrated in bold in Table 3.2(a). In addition to using ALDOT?s inspection guidelines, researchers examined rainfall intensity-duration-frequency (IDF) curves from Technical Paper #25, for Montgomery, Alabama (USDC, 1955). IDF curves used for this research are included in Appendix B. These curves relate a rainfall intensity, in inches per hour, with a given rain duration, to determine a return period (i.e. storm event), in years. Table 3.2(b) shows the relationship between these different storm events from the IDF curves and corresponding intensities generated by the rainfall simulator with respective durations. The rain regime for an individual 15 minute test would be representative of a 2-year, 15-min storm event, as illustrated in bold in Table 3.2(b). In ASWCC?s handbook on erosion and sediment control BMPs, many of the different technologies are designed to withstand specific storm events. While the guidebook does not provide a specific design storm for chemical stabilization (i.e. PAM), a 2-year, 24 hour storm is commonly used in many of the BMPs outlined (ASWCC, 2003). This storm event is a relatively common occurrence and is used in the design and selection of many different technologies currently used for erosion and sediment control. Therefore, research conducted with these rainfall parameters 56 would be similar to those an erosion control product or BMP would be exposed to in actual practice. Using the Rainfall Frequency Charts presented in Technical Paper #40, included in Appendix B, it was observed that a 2 year, 24 hour rain event for mid-Alabama would produce a cumulative rainfall amount ranging from 4 to 4.5 inches of rainfall (USDC, 1961). The selected intensity of 4.4 in/hr would need to last for one hour to adequately model this specific storm event. Therefore, the final rain regime used for experimentation would consist of four, 15 minute events, representative of a 2-yr, 15-min event. These four events would be required to achieve the desired 2-yr, 24-hr storm event and will permit researchers to examine the long-term effectiveness of PAM treatments. A period of no rainfall would be observed to allow researches time to collect data in between events. This period of would last for a duration of 15 minutes. Table 3.2 Rain Regime Table for Different Intensities and Storm Durations (a) Total Accumulated Rainfall Based Off Intensity and Duration Pressure (psi) Intensity (in/hr) Total Accumulated Rainfall (in) 5 min 10 min 15 min 20 min 30 min 40 min 60 min 5 3.60 0.30 0.60 0.90 1.20 1.80 2.40 3.60 10 4.39 0.37 0.73 1.10 1.46 2.20 2.93 4.39 15 5.19 0.43 0.86 1.30 1.73 2.59 3.46 5.19 20 5.98 0.50 1.00 1.50 1.99 2.99 3.99 5.98 25 6.78 0.56 1.13 1.69 2.26 3.39 4.52 6.78 30 7.57 0.63 1.26 1.89 2.52 3.79 5.05 7.57 (b) Representative 24-Hour Storm Events Based Off IDF Curves for Montgomery, AL Pressure (psi) Intensity (in/hr) Storm Events 5 min 10 min 15 min 20 min 30 min 40 min 60 min 5 3.60 -- -- 2-yr 2-yr 5-yr 10-yr 25-yr 10 4.39 -- 2-yr 2-yr 5-yr 10-yr 50-yr 100-yr 15 5.19 2-yr 2-yr 5-yr 10-yr 50-yr 100-yr 100-yr 20 5.98 2-yr 5-yr 10-yr 25-yr 100-yr 100-yr 100-yr 25 6.78 5-yr 25-yr 50-yr 100-yr 100-yr 100-yr 100-yr 30 7.57 25-yr 100-yr 100-yr 100-yr 100-yr 100-yr 100-yr 57 3.4 EXPERIMENTAL PROCEDURES Experimental procedures and methods used in previous experiments were examined and modified where improvements could be implemented. Modifications from previous research procedures included: (1) detailed analyses of test soil, (2) preparation methods for conducting compaction on test plots, (3) general procedural changes pertaining to new test boxes and experiment preparation, and (4) data collection methods. 3.4.1 Soil Analysis Soil used for this research was provided by a local grading contractor from a construction site near the NCAT test track in Opelika, Alabama (32o33?5? N, 85o20?28? W). A sample of this material was sent to the Auburn University Soil Testing Laboratory to determine soil composition. Table 3.3 shows the percent composition and United States Department of Agricultural (USDA) textural classification of the soil sample used for this research. Table 3.3 Percent Composition and Classification of Experimental Soil % Sand % Silt % Clay Classification 58.6 12.5 28.9 Sandy Clay Loam The test soil was found to be composed primarily of sand, with smaller amounts of clay and silt. In addition to identifying soil composition, particle distribution was determined using a sieve analysis. Figure 3.10 illustrates distribution of particles and researchers determined that the test soil was well-graded. 58 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 0.0010 0.0100 0.1000 1.0000 Pe rce nt Pa sse d Particle Diameter (in) Figure 3.10 Particle Size Distribution for Experimental Soil. 3.4.2 Compaction Analysis Following tests to classify and identify soil characteristics, additional information pertaining to compaction was analyzed. For this research, test plots were designed to model similar characteristics of a typical highway embankment. The main characteristics of interest were 3:1 compacted fill slopes and how this configuration affects the performance of PAM. Previous research conducted by Halverson (2006) and McDonald (2007) used a metal roller to compact test soils to a compaction rate of 90% of the maximum density. However, to reach a higher level of compaction required by ALDOT for fill slopes, a new method of compaction was considered. ALDOT specifies in its Standard Specification for Highway Construction (2002) that in-place density requirements for highway embankments must be compacted to 95%. The different method examined here used hand-tamps dropped on the test plots to achieve the required compaction rate. 59 To determine the number of drops required to compact the soil using hand-tamps, multiple compaction tests were conducted. A modified Proctor test, as specified in American Society for Testing and Materials (ASTM) D1557, was used to generate a Proctor curve relating moisture content (MC) with dry unit weight, as seen below in Figure 3.11. The chart in Figure 3.11 shows specific moisture contents required to obtain different dry unit weights for the tested material. The optimum moisture content (OMC) was determined by locating the maximum dry unit weight on the Proctor curve, which is 114 pcf (1826.1 kg/m3), and would be achieved with a MC of 15%. The dashed-line at 108 pcf (1730 kg/m3) denotes the minimum acceptable dry unit weight to achieve 95% compaction. Since compaction would be accomplished with hand-tamps, procedures stated in ASTM D1557 specifications were modified for a new mold custom built for the hand-tamps, as pictured in Figure 3.12(a) and 3.12 (b). 100 105 110 115 10 12 14 16 18 20 Dr y U nit W eig ht (p cf) Moisture Content (%) Calculated Curve Points Data Points Maximum Density Point 95% Compaction Figure 3.11 Proctor Curve for Experimental Soil. 60 (a) Hand-tamp with Mold (b) Hand-tamp Compacting Soil in Mold Figure 3.12 Determining Required Compaction Rate with Hand-tamp. The footprint of the hand-tamps used measured 10 in. x 10 in. (25.4 cm x 25.4 cm), so the mold was constructed to contain soil for that area (approximately 12 in. x 12 in.). The sides of the mold were constructed so the top inch of excess soil from compaction could be leveled off. This would provide a 1 inch (2.5 cm) depth of compacted soil with a known volume. This known volume of compacted soil was used to determine a corresponding unit weight pertaining to an amount of compaction generated with the hand-tamps. Soil was compacted with five different setups using a hand-tamp, dropped at approximately 12 inches above the soil surface, at the required OMC of 15%. These different setups corresponded to a hand-tamp dropped 10, 20, 30, 50, and 60 times. After the corresponding number of drops, a dry unit weight was calculated and plotted as a function of hand-tamp drops, as seen in Figure 3.13. 61 y = 36.01x0.243 R? = 0.983 50.0 60.0 70.0 80.0 90.0 100.0 110.0 0 10 20 30 40 50 60 70 Dr y U nit W eig ht (p cf) Number of Drops Figure 3.13 Compaction of Soil using Hand-tamp. A regression line was plotted as illustrated in Figure 3.13. A power function was selected to reflect the nature of compacting soil. Soil will reach a point where it can no longer be compacted, no matter how much energy is applied. Therefore, the regression line would level off at this point. Using this equation, the number of drops required to obtain different unit weights was calculated and shown in Table 3.4. To obtain a 95% compaction rate, at least 90 drops of a hand-tamp would be necessary to reach a unit weight around 108 pcf (1730 kg/m3). Table 3.4 Calculated Dry Unit Weight (pcf) and Required Number of Drops Number of Drops Dry Unit Weight (pcf) 10 63.1 20 74.7 30 82.5 40 88.5 50 93.4 60 97.7 70 101.4 80 104.8 90 107.8 100 110.6 62 3.4.3 Polymer Selection To determine which formulation of PAM was to be used, additional tests were conducted on soil samples prior to experimentation. These tests determine the optimum mixture of PAM and recommended application rate. Therefore, a sample of the sandy clay soil was sent to Applied Polymer Systems (APS) in Woodstock, GA. APS conducted analyses on the soil sample, recommended a PAM formulation and application rate, and supplied the PAM for experiments. APS recommended using the 712 Silt Stop powder, applied at an application rate ranging from 25 to 35 lbs/acre. Application guidelines for PAM were available from APS and were used for this research (APS, 2006). For this study, different application rates of PAM at 35 lbs/acre, 25 lbs/acre, and 15 lbs/acre, in both a dry and liquid form were examined to analyze differences in overall effectiveness with different application rates as an erosion control measure. Both treatment methods (i.e. dry and liquid) would be tested under similar conditions, focusing on the immediate effectiveness of each product after its initial application. Therefore, the liquid PAM applications would not be given time to dry prior to the start of an experiment. 3.4.4 Experiment Organization Figure 3.14 illustrates a flowchart which describes the overall experimental plan for this research. The flowchart shows the terminology and organization used to identify each experimental setup. This research examined six different treatment options and compared results to a control setup, containing bare soil (i.e. conditions). The six 63 treatments options (i.e. experiments) included three different application rates of dry and liquid PAM, (i.e. 15 lbs/acre, 25 lbs/acre, 35 lbs/acre) respectively. Within each treatment option, two experiments were conducted as a means to check reproducibility between setups. Each experiment contained two test plots, identified by its random placement under the rainfall simulator (i.e. ?left? or ?right? position). These locations were randomly assigned using a coin flip, which produced an exact, 50/50 distribution of test boxes to the left and right positions. To obtain the desired amount of rain, as indicated in Section 3.3.3, four ?tests? were simulated within an experiment. Between each test, there was a 15 minute break between storm events. This rain regime also allows researchers to examine the long-term effectiveness of PAM tested under various rainfall events over an extended period of time. 64 Control ? Bare Soil Experiment 1 Experiment 2 Left Plot Test 1 15 min event 15 min break Test 2 15 min event 15 min break Test 3 15 min event 15 min break Test 4 15 min event Right Plot Test 1 15 min event 15 min break Test 2 15 min event 15 min break Test 3 15 min event 15 min break Test 4 15 min event Left Plot ? 4 Tests Right Plot ? 4 Tests Treatment 4 ? Liquid PAM @ 35 lbs/acre Treatment 5 ? Liquid PAM @ 25 lbs/acre Treatment 6 ? Liquid PAM @ 15 lbs/acre Treatment 2 ? Dry Granular PAM @ 25 lbs per acre Treatment 3 ? Dry Granular PAM @ 15 lbs per acre Treatment 1 ? Dry Granular PAM @ 35 lbs per acre CONDITION(s) EXPERIMENT(s) TEST(s) Note: 14 Total Experiments, 28 Replicates, 56 Tests Figure 3.14 Flowchart of Experimental Organization. 65 3.4.5 Test Plot Preparation For each experimental setup, test plot preparation consisted of five parts: (1) determine existing OMC of stockpile, (2) soil compaction, (3) treatment application, (4) plot placement under rainfall simulator, and (5) saturate test plots. To reach a desired level of compaction, test soil must be near the recommended moisture content as determined in Section 3.4.2. Twenty-four hours before an experiment, moisture content for the soil was checked to determine if additional water was required. When the moisture content was lower than the required amount, the necessary ratio of test soil and water were weighed and measured, as seen in Figure 3.15(a). Soil and water were thoroughly mixed in a wheelbarrow, and added into a test box, as seen in Figure 3.15(b). (a) Preparing Soil for an OMC of 15% (b) Soil and Water Thoroughly Mixed Figure 3.15 Obtaining OMC for Experimentation. Approximately 2 inches of soil was placed in a box, because it was observed that when compacted, the depth of un-compacted soil would be approximately halved when compacted. It was also determined that two, 5 gallon buckets, as seen in Figure 3.15(a), would be sufficient to achieve this un-compacted amount for one layer. Therefore, this 66 amount of un-compacted soil would be reduced to a target depth of 1 inch when compacted. To compact one layer of soil, a hand tamp was used to compact 8 subsections, as illustrated in Figure 3.16(a). To achieve a 95% compaction rate, one of these subsection required 90 drops of a hand-tamp. This was performed over all 8 subsections to compact approximately a 1 inch layer of soil, as seen in Figure 3.16 (b). This entire process was repeated one more time to fill a box with approximately 2 inches of compacted material. (a) 8 Sections to Compact One Layer (b) First Layer Compacted Figure 3.16 Using Hand Tamp to Compact 1 inch of Test Soil. After compaction, the test plots could be treated with the different PAM applications required for experimentation. The two methods of PAM application analyzed for this research was (1) dry, granular PAM applied directly to the plots and (2) liquid PAM; prepared using granular PAM mixed with water and applied with a sprayer. Both types of applications were examined using different recommended rates of 35, 25, and 15 lbs/acre. The necessary amount of PAM required for all experiments was converted to experimental scale. Table 3.5(a) and 3.5(b) show the application rates determined for intermediate-scale experiments. Note that in Table 3.5(b), the water corresponds with the 1 2 3 4 5 6 7 8 67 necessary amount to mix with dry granular PAM to create liquid PAM for treatments. The recommended amount of water required for mixing with PAM to achieve the proper water to PAM ratio was provided by APS. Table 3.5 PAM Application Rates (a) Application Rates for Dry PAM Dry Rate1 (lbs/acre) Dry Rate2 (grams/plot) 35 2.80 25 2.00 15 1.20 (b) Application Rates for Liquid PAM Dry Rate1 (lbs/acre) Water1 (gal/acre) Dry Rate2 (grams/plot) Water2 (gal/plot) 35 3000 2.80 0.53 25 2143 2.00 0.38 15 1286 1.20 0.23 Notes: 1. Field Scale | 2. Intermediate Scale Application of the dry granular PAM was accomplished by utilizing a salt shaker, as shown in Figure 3.17(a). PAM was applied uniformly to both test plots, verified through observation. Figure 3.17(b) illustrates the uniform coverage of dry PAM as it was applied to a test plot. (a) 712 Silt Stop and Salt Shaker (b) Dry PAM Applied on Test Plot Figure 3.17 Application of Dry Granular PAM. 68 The process to apply liquid PAM required the use of a Maruyama MS074 backpack sprayer with a built in agitator. Specifications for the backpack sprayer can been seen in Appendix C, included in this report. This type of backpack sprayer was necessary because the dry PAM needed to be introduced slowly and mixed thoroughly with water prior to application. Without an agitator, the mixture would become too viscous and clog a traditional sprayer. Therefore, care was taken to ensure that dry granular PAM was added slowly to the water. The mixture of dry PAM and water was allowed to mix for approximately 30 minutes prior to application. Figure 3.18(a), 3.18(b), and 3.18(c) demonstrate this process for preparing liquid PAM for experimentation. (a) Adding Water to Sprayer (b) Slowly adding Dry PAM (c) Observing Mixing Figure 3.18 Preparing Liquid PAM Mixture for Application. The backpack sprayer was used to uniformly apply the liquid PAM, coating both test plots simultaneously. Care was taken to control the application of PAM, so the sprayer was set to apply a consistent amount of liquid PAM and was applied equally to both plots until all the mixture was completely used. Figure 3.19(a), 3.19(b), and 3.19(c) show the backpack sprayer applying liquid PAM to two test plots prior to experimentation. PAM was applied outdoors to keep the liquid PAM off the concrete slab, due to potential safety hazards with slippery conditions. 69 (a) Backpack Sprayer used for Liquid PAM Application (b) Spraying Liquid PAM on Plot (c) Close-up of Spray Nozzle Figure 3.19 Application of Liquid PAM. Once both test plots were treated with a PAM application, they were moved to a location under the rainfall simulator. A test plot location was randomly assigned using a coin-flip and placed under the simulator to control for any potential bias. Therefore, test plots were classified as either the ?left? position or ?right? position, as shown in Figure 3.20. Figure 3.20 Plots Placed in ?Left? and ?Right? Position. With test plots setup and placed under the rainfall simulator, a brief saturation period was conducted to facilitate the start of runoff prior to experimentation. The ?Left? Test Plot ?Right? Test Plot 70 rainfall simulator was set to the desired properties as discussed in Section 3.3.3 and started. Plots were briefly exposed to rainfall until the runoff was initiated. Plot preparation at this point was completed and experiments could commence. 3.4.6 Data Collection Data collection for this study was similar to previous conducted research by Halverson (2006) and McDonald (2007), with slight modifications to reflect changes made in the experimental design. However, improvements were made in recording turbidity from runoff samples. Additional information was also collected, which was not examined from previous research (i.e. runoff mass and particle sizes). Photographs of each test plot were taken to document a visual condition prior to an experiment and at the end of each of the four tests. In addition, the entire experiment duration was recorded with a video camera, which allowed for time-lapse footage of the erosion process to be documented. The primary concern during data collection pertained to runoff generated from test plots during rainfall events. Runoff samples were collected every minute during experimentation in clear, five quart buckets with volume markings denoted on the side of each. Volume and mass for these buckets were recorded for each ?left? and ?right? test plot, as illustrated in Figure 3.21. Instantaneous turbidity was recorded using an ANALITE NEP160 turbidity meter with a ANALITE NEP 260 probe, as seen in Figure 3.22. This probe was capable of measuring turbidity levels from 0 to 4,000 NTU. Detailed specifications for the meter and probe can be found in Appendix C. Prior to recording initial turbidity, collected samples were stirred to represent turbidity of the 71 surface runoff as it was leaving the test plots. This new turbidity meter allowed researchers a direct means to test water samples for turbidity during an ongoing experiment, reducing overall time involved with experimentation and data collection. Instantaneous turbidity was recorded at minute intervals during each test run. Along with instantaneous turbidity, the meter and probe were used to collect and record turbidity over time, which illustrated the rate of particle settling. At the 5 and 10 minute interval for each of the four tests on each plot, samples were collected in one quart cups and researchers recorded turbidity over time. An example of this process is illustrated in Figure 3.22, where the probe was suspended in the runoff sample, and allowed to collect data over a set period of time. This period of time was set at 10 minutes for the ?control? experiments, and 3 minutes for subsequent PAM treated experiments, due to the difference in length of time required for particles to settle out between the control and treated experiments. Figure 3.21 Recording Volume and Mass of Surface Runoff. Figure 3.22 ANALITE NEP 160 Turbidity Meter and Probe. Next, surface runoff samples were poured into Hayward single-length filter bags with one micron sized pores, as pictured in Figure 3.23. This was done every 3 minutes, for a total of 5 bags per test, totaling 40 bags for an entire experiment. At the conclusion 72 of an experiment, these sediment-laden filter bags were placed in an oven at 160o F (71.1o C) and dried for 24 hours. These dried bags were weighed to determine the amount of eroded soil from each test plot contained within each bag. Dried soil samples from each bag were collected and combined for sieve analysis to determine particle size of the surface runoff, as illustrated in Figure 3.24. Figure 3.23 Hayward Filter Bags with Sediment-Laden Water. Figure 3.24 Soil Particle Distribution and Sieves. By following the above mentioned experimental procedures for data collection, a large sample size of data was produced. Table 3.6 illustrates the totals of data collected during experimentation. For runoff observations, in which mass and volume were recorded, a total of 1,680 observations were recorded (e.g. 7 conditions ? 4 test plots ? 4 tests ? 15 observations per test = 1,680 total recorded measurements). The ?conditions? parameter, as shown, represents the multiple treatments examined (i.e. ?control?, ?dry PAM 35?, ?dry PAM 25?, etc) and ?observations per test? represent measurements recorded for every minute during a 15 minute test. For data pertaining to soil loss, samples were recorded every 3 minutes, resulting in 5 observations per test plot. This produced a total of 560 measurements for analysis. Therefore, using these new Largest Particles Smallest Particles 73 experimental procedures, more experiments were conducted, which produced more data for researchers to examine and analyze. Table 3.6 Breakdown of Collected Data Totals Conditions Test Plots Tests Observations per Test Runoff Observations1 Turbidity Observations1 Soil Loss Observations2 7 4 4 15 1,680 1,680 560 Notes: 1. Observations were recorded every minute 2. Observations were recorded every 3 minutes 3.4.7 Statistical Analyses One-way analysis of variance (ANOVA) was the primary statistical method used to analyze experimental data presented in this research. Typically, ANOVA is used with testing three of more independent groups of data, where a standard t-test is generally used for comparing two independent groups. This is attributed to the fact that if a t-test is performed on more than two independent groups, there is a chance of incorrectly rejecting a null hypothesis or failing to accept the null hypothesis (i.e. Type I and II errors). A 5% level of significance indicates that there is a 95% chance of correctly accepting a null hypothesis. If more than two groups were tested, than the probability of correctly accepting the null hypothesis decreases. This decrease is a function of the number of groups being tested (e.g. if five groups are tested using a standard t-test with a 5% level of significance, the probability of correctly accepting the null hypothesis is approximately [0.95]5 or 0.77). ANOVA procedures compensate for this effect and multiple independent groups can be tested equally without the possibility of compounding these errors. Most statistical software packages contain tools to conduct an ANOVA analysis, due to the complex mathematical computations required to analyze a 74 large number of groups. The one-way ANOVA tool provided by Microsoft?s ExcelTM 2007 was used for this research effort. However, ANOVA analysis alone is not sufficient to determine statistical significance between individual pairs. ANOVA procedures only provide the capability to determine if all tested means are equal. A typical null and alternative hypothesis used during this research for ANOVA analysis is illustrated in Equation 3 and 4, respectively. Ho: ?0 = ?1 = ?2 = ?3 = ?4 = ?6 = ?7 (3) Ha: all means are different (4) where, Ho = null hypothesis Ha = alternative hypothesis ?i = mean values of each data set ?i?, i = independent groups [i.e. (0) control, (1) dry 35, (2) dry 25, etc] The null and alternative hypothesis statements for ANOVA are not sufficient to discern statistical significant difference between individual pairs. Therefore, additional ad hoc tests needed to be performed on different combinational pairs of groups. These test that examine multiple means incorporate measures to assist in reducing the risks associated with Type I error, as outlined above. For this research, the Tukey-Kramer procedure for multiple comparisons was selected for analysis, since comparisons were conducted on all pair-wise differences between means. To determine if statistical significance was observed using Tukey-Kramer procedures, a confidence interval (CI) was calculated using the following equations: 75 ( ) ?? ? ? ??? ? +???= ji pji nnsmultiplierKramerTukeyCI 11 %95 ?? (5) where, CI95% = 95% confidence interval, ?i - ?j = difference of means for ?i? and ?j? groups, sp = pooled standard deviation, and ni,j = sample sizes of ?i? and ?j? groups The pooled standard deviation as shown in Equation 5 can be derived from taking the square-root of the Mean Square Error (MSE), which is calculated during creation of an ANOVA table. A Tukey-Kramer multiplier can be determined using Equation 6: 2 ,, ?anaqmultiplierKramerTukey ?= (6) where, q = upper percentile for a studentized range distribution, a = total number of groups, n = sample size, and ? = level of significance (5%) To determine if the two test groups are statistically significant, the calculated confidence interval are examined and if zero is contained within the upper and lower bounds of the interval, than the two groups are not statistically significant. 3.5 SUMMARY In total, fourteen experiments were conducted, examining the effectiveness of two different treatment options, with varying application rates, to a bare soil condition. A 76 new facility was built to house this research, which provided researchers the necessary equipment and space for experimentation. Experiments were designed using methods and procedures from previous research as a general guideline. Modifications were made where researchers felt that certain areas needed improvements. One of these areas was designing and constructing a new testing apparatus used to contain experimental soil. These newly designed test plots allowed for an experimental setup that required less time to prepare, allowing researchers the ability to conduct more experiments and collect more data than the previous research could feasibly produce. Another area that was redesigned was the rainfall simulator. A new simulator was designed with new components which produced controllable rainfall events with a uniform spray area. Procedures were developed to reflect differences made in the new experimental design. Soil was compacted using hand-tamps, which provide the necessary energy to effectively compact the test soil to a rate of 95%. New equipment was used during data collection to assist with reducing the amount of time needed for each experiment. This included a turbidity meter and probe which could be used to collected turbidity instantaneously during an experiment. Additional data was collected from experiments, which included runoff mass and particle size distribution of surface runoff. These new experimental designs and procedures allowed researchers to develop the means to uniformly setup, conduct, and analyze erosion control BMPs with laboratory conditions using intermediate-scale plot sizes. This new design reduced the overall time and the amount of effort required for conducting an experiment with this setup which produces a high quality data set for conducting an ANOVA analysis to determine if results were statistically significant. 77 CHAPTER FOUR INTERMEDIATE-SCALE EXPERIMENTS RESULTS AND DISCUSSION 4 4.1 INTRODUCTION Intermediate-scale experiments provided means to perform multiple tests with fewer resources, when compared to field-scale experiments, and also generated a large amount of data for analysis. This chapter will present the results generated from experiments and the statistical tests used for analyses pertaining to the performance of PAM as an erosion and sediment control measure. 4.2 EXPERIMENTAL RESULTS Data collection methods and features, as outlined in Section 3.4.6, provided researchers with an abundant amount of raw data. The data collected and recorded includes: (1) surface runoff volume, (2) surface runoff mass, (3) initial turbidity, (4) runoff samples (turbidity versus time) and, (5) amount of soil eroded from test plots. Raw data collected from each experiment is included in Appendix D. Data from replicated test plots were compared to determine if any anomalies were present and an overall average was calculated for further analysis. The raw data was scaled to represent practical field units for the purpose of reporting in subsequent sections. 78 4.2.1 Surface Runoff Surface runoff generated from test plots was collected at 1 minute intervals during an experiment?s 60 minute total duration. Volume and weight for each sample were measured and recorded. Overall averages of runoff volume for the control and six PAM treatments are illustrated in Figure 4.1 Upon visual inspection, Figure 4.1 demonstrates no observable difference between different treatments tested. Specifically, no increase or decrease was discerned in the amount of surface runoff generated. The higher amount of runoff reported in the dry PAM at 35 lbs/acre was attributed to a slightly higher operating pressure of the rainfall simulator, rather than any effect the treatment had on the amount of runoff generated. 0 500 1000 1500 2000 2500 3000 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( ga l/a cre ) Time (min) Control Dry 35 Dry 25 Dry 15 Liquid 35 Liquid 25 Liquid 15 Test 1 Test 2 Test 3 Test 4 * * * Note: ?*? denotes 15 minute break in between tests Figure 4.1 Average Surface Runoff vs. Time. Figure 4.2 illustrates the cumulative amount of average surface runoff for each of the four tests (i.e. Test 1, Test 2, Test 3, and Test 4). As seen earlier with Figure 4.1, the addition of PAM appears to have no substantial effect in the total amount of runoff accumulated, which ranged from 30,000 to 36,000 gal/acre. These figures show that 79 similarities in data confirm the researchers? experimental design goals by rapidly generating reproducible results. 0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( 10 00 ga l/a cre ) Time (min) Control Dry 35 Dry 25 Dry 15 Liquid 35 Liquid 25 Liquid 15 Test 1 Test 2 Test 3 Test 4 * * * Note: ?*? denotes 15 minute break in between tests Figure 4.2 Average Cumulative Surface Runoff vs. Time. Specific values for surface runoff for the dry and liquid PAM treatments are shown below in Table 4.1 and Table 4.2, respectively. Both tables illustrate the calculated average of runoff for each 15 minute test interval and each test?s corresponding cumulative amount. Average runoff and standard deviation were calculated from a total sample size of 60 measurements per test (e.g. 4 test plots ? 15 min per test = 60 total samples). 80 Table 4.1 Average Surface and Cumulative Runoff for Each Test [Dry PAM] Condition Runoff a (gal/acre) Standard Deviation b (gal/acre) Percent Reduction c Cumulative Runoff d (gal/acre) Test 1 Control 2279.0 236.9 - 34184.4 Dry 35 2448.0 328.9 -7.4% 36719.7 Dry 25 2102.0 135.6 7.8% 31530.0 Dry 15 2044.1 182.9 10.3% 30662.2 Test 2 Control 2331.1 236.6 - 34967.1 Dry 35 2432.1 289.8 -4.3% 36481.5 Dry 25 2158.7 251.0 7.4% 32380.7 Dry 15 2088.4 188.6 10.4% 31325.8 Test 3 Control 2326.6 243.8 - 34899.0 Dry 35 2324.3 263.4 0.1% 34865.0 Dry 25 2215.4 212.5 4.8% 33231.5 Dry 15 2087.3 198.0 10.3% 31308.8 Test 4 Control 2335.7 241.2 - 35035.2 Dry 35 2227.9 195.7 4.6% 33418.7 Dry 25 2215.4 197.9 5.1% 33231.5 Dry 15 2141.7 231.2 8.3% 32125.5 Notes: ?a? Average surface runoff vs. time for each test ?b? Standard deviation of surface average runoff vs. time ?c? Denotes values normalized by control condition ?d? Average cumulative surface runoff for each 15 min. test Table 4.2 Average Surface and Cumulative Runoff for Each Test [Liquid PAM] Condition Runoff a (gal/acre) Standard Deviation b (gal/acre) Percent Reduction c Cumulative Runoff d (gal/acre) Test 1 Control 2279.0 236.9 - 34184.4 Liquid 35 1984.0 264.9 12.9% 29760.3 Liquid 25 2127.0 194.4 6.7% 31904.3 Liquid 15 2128.1 194.4 6.6% 31921.3 Test 2 Control 2331.1 236.6 - 34967.1 Liquid 35 2140.6 251.4 8.2% 32108.5 Liquid 25 2197.3 197.8 5.7% 32959.3 Liquid 15 2200.7 191.6 5.6% 33010.3 Test 3 Control 2326.6 243.8 - 34967.1 Liquid 35 2212.0 221.2 4.9% 33180.5 Liquid 25 2237.0 220.4 3.9% 33554.8 Liquid 15 2276.7 217.2 2.1% 34150.4 Test 4 Control 2335.7 241.2 - 34967.1 Liquid 35 2210.9 222.6 5.3% 33163.5 Liquid 25 2264.2 215.5 3.1% 33963.2 Liquid 15 2206.4 205.3 5.5% 33095.4 Notes: ?a? Average surface runoff vs. time for each test ?b? Standard deviation of surface average runoff vs. time ?c? Denotes values normalized by control condition ?d? Average cumulative surface runoff for each 15 min. test 81 Percent reductions, as seen Table 4.1 and Table 4.2, were normalized with the control condition for comparison. Negative values, as observed with the 35 lb/acre dry PAM treatment, indicate an increase in recorded runoff. The largest reduction of runoff occurred during test 1 with the 35 lbs/acre application of liquid PAM at 12.9%. Conversely, dry PAM applied at 35 lbs/acre for test 1 had a recorded increase of 7.4% when compared to the control. 4.2.1.1 Statistical Analysis: Surface Runoff To determine if there was a statistically significant difference between calculated averages for control vs. treatment surface runoff, an ANOVA table was generated for each test (e.g. test 1, test 2). These ANOVA tables are included in Appendix E within this report. An ANOVA analysis showed that test 1, test 2, and test 3 rejected the null hypothesis, indicated that differences between control and treatments existed. Test 4 ANOVA analyses determined that the null hypothesis should be accepted, resulting in no differences occurred between all test groups. Tukey-Kramer confidence intervals were calculated using the Equations 5 and 6, outlined in Section 3.4.7, to determine any specific significant differences between conditions tested, as illustrated in Table 4.3, Table 4.4, Table 4.5, and Table 4.6. 82 Table 4.3 Tukey-Kramer Multiple Comparisons on Average Surface Runoff [Test 1] Comparison ?i - ?j CI [LB] CI [UB] Significantly Different Control vs. Dry 35 169.0 -22.9 361.0 No Control vs. Dry 25 234.8 42.9 426.8 Yes Control vs. Dry 15 177.0 -15.0 368.9 No Control vs. Liquid 35 294.9 103.0 486.9 Yes Control vs. Liquid 25 150.9 -41.1 342.8 No Control vs. Liquid 15 152.0 -39.9 343.9 No Dry 35 vs. Dry 25 403.8 211.9 595.8 Yes Dry 35 vs. Dry 15 346.0 154.0 537.9 Yes Dry 35 vs. Liquid 35 464.0 272.0 655.9 Yes Dry 35 vs. Liquid 25 319.9 128.0 511.8 Yes Dry 35 vs. Liquid 15 321.0 129.1 513.0 Yes Dry 25 vs. Dry 15 57.9 -134.1 249.8 No Dry 25 vs. Liquid 35 60.1 -131.8 252.1 No Dry 25 vs. Liquid 25 83.9 -108.0 275.9 No Dry 25 vs. Liquid 15 82.8 -109.1 274.7 No Dry 15 vs. Liquid 35 118.0 -74.0 309.9 No Dry 15 vs. Liquid 25 26.1 -165.8 218.0 No Dry 15 vs. Liquid 15 25.0 -167.0 216.9 No Liquid 35 vs. Liquid 25 144.1 -47.9 336.0 No Liquid 35 vs. Liquid 15 142.9 -49.0 334.9 No Liquid 25 vs. Liquid 15 1.1 -190.8 193.1 No Notes: [LB] signifies lower bound of confidence interval [UB] signifies upper bound of confidence interval qcrit = 4.26 Table 4.4 Tukey-Kramer Multiple Comparisons on Average Surface Runoff [Test 2] Comparison ?i - ?j CI [LB] CI [UB] Significantly Different Control vs. Dry 35 101.0 -103.5 305.4 No Control vs. Dry 25 242.8 38.3 447.2 Yes Control vs. Dry 15 172.4 -32.0 376.8 No Control vs. Liquid 35 190.6 -13.8 395.0 No Control vs. Liquid 25 130.5 -74.0 334.9 No Control vs. Liquid 15 133.9 -70.6 338.3 No Dry 35 vs. Dry 25 343.7 139.3 548.1 Yes Dry 35 vs. Dry 15 273.4 69.0 477.8 Yes Dry 35 vs. Liquid 35 291.5 87.1 495.9 Yes Dry 35 vs. Liquid 25 231.4 27.0 435.8 Yes Dry 35 vs. Liquid 15 234.8 30.4 439.2 Yes Dry 25 vs. Dry 15 70.3 -134.1 274.7 No Dry 25 vs. Liquid 35 52.2 -152.2 256.6 No Dry 25 vs. Liquid 25 112.3 -92.1 316.7 No Dry 25 vs. Liquid 15 108.9 -95.5 313.3 No Dry 15 vs. Liquid 35 18.1 -186.3 222.6 No Dry 15 vs. Liquid 25 42.0 -162.4 246.4 No Dry 15 vs. Liquid 15 38.6 -165.8 243.0 No Liquid 35 vs. Liquid 25 60.1 -144.3 264.5 No Liquid 35 vs. Liquid 15 56.7 -147.7 261.1 No Liquid 25 vs. Liquid 15 3.4 -201.0 207.8 No Notes: [LB] signifies lower bound of confidence interval [UB] signifies upper bound of confidence interval qcrit = 4.26 83 Table 4.5 Tukey-Kramer Multiple Comparisons on Average Surface Runoff [Test 3] Comparison ?i - ?j CI [LB] CI [UB] Significantly Different Control vs. Dry 35 2.3 -209.7 214.3 No Control vs. Dry 25 239.4 27.3 451.4 Yes Control vs. Dry 15 111.2 -100.8 323.2 No Control vs. Liquid 35 114.6 -97.4 326.6 No Control vs. Liquid 25 49.9 -162.1 261.9 No Control vs. Liquid 15 89.6 -122.4 301.6 No Dry 35 vs. Dry 25 237.1 25.1 449.1 Yes Dry 35 vs. Dry 15 108.9 -103.1 320.9 No Dry 35 vs. Liquid 35 112.3 -99.7 324.3 No Dry 35 vs. Liquid 25 47.6 -164.4 259.7 No Dry 35 vs. Liquid 15 87.3 -124.7 299.4 No Dry 25 vs. Dry 15 128.2 -83.8 340.2 No Dry 25 vs. Liquid 35 124.8 -87.2 336.8 No Dry 25 vs. Liquid 25 189.4 -22.6 401.5 No Dry 25 vs. Liquid 15 149.7 -62.3 361.8 No Dry 15 vs. Liquid 35 3.4 -208.6 215.4 No Dry 15 vs. Liquid 25 61.3 -150.8 273.3 No Dry 15 vs. Liquid 15 21.6 -190.5 233.6 No Liquid 35 vs. Liquid 25 64.7 -147.4 276.7 No Liquid 35 vs. Liquid 15 25.0 -187.1 237.0 No Liquid 25 vs. Liquid 15 39.7 -172.3 251.7 No Notes: [LB] signifies lower bound of confidence interval [UB] signifies upper bound of confidence interval qcrit = 4.26 Table 4.6 Tukey-Kramer Multiple Comparisons on Average Surface Runoff [Test 4] Comparison ?i - ?j CI [LB] CI [UB] Significantly Different Control vs. Dry 35 107.8 -90.8 306.4 No Control vs. Dry 25 194.0 -4.6 392.6 No Control vs. Dry 15 120.2 -78.4 318.9 No Control vs. Liquid 35 124.8 -73.8 323.4 No Control vs. Liquid 25 129.3 -69.3 327.9 No Control vs. Liquid 15 71.5 -127.1 270.1 No Dry 35 vs. Dry 25 86.2 -112.4 284.8 No Dry 35 vs. Dry 15 12.5 -186.1 211.1 No Dry 35 vs. Liquid 35 17.0 -181.6 215.6 No Dry 35 vs. Liquid 25 21.6 -177.1 220.2 No Dry 35 vs. Liquid 15 36.3 -162.3 234.9 No Dry 25 vs. Dry 15 73.7 -124.9 272.3 No Dry 25 vs. Liquid 35 69.2 -129.4 267.8 No Dry 25 vs. Liquid 25 64.7 -134.0 263.3 No Dry 25 vs. Liquid 15 122.5 -76.1 321.1 No Dry 15 vs. Liquid 35 4.5 -194.1 203.2 No Dry 15 vs. Liquid 25 9.1 -189.5 207.7 No Dry 15 vs. Liquid 15 48.8 -149.8 247.4 No Liquid 35 vs. Liquid 25 4.5 -194.1 203.2 No Liquid 35 vs. Liquid 15 53.3 -145.3 251.9 No Liquid 25 vs. Liquid 15 57.9 -140.8 256.5 No Notes: [LB] signifies lower bound of confidence interval [UB] signifies upper bound of confidence interval qcrit = 4.26 84 Many of the observed differences within tests 1, 2, and 3 occurred with comparisons made to dry PAM at 35 lbs/acre. The Tukey-Kramer results for test 4, shown in Table 4.6, found no statistical difference between any combinations of groups. Overall, no statistical significant results were observed pertaining to PAM treatments reducing the amount of runoff generated on the control. While dry PAM at 35 lbs/acre appeared to be performing poorly at first, no other treatments experienced a similar effect. Other PAM applications (with the exception of liquid 35) experience no significant difference when compared to the control condition as well. When different PAM applications were compared to each other, no significant results were observed. Therefore, the addition of PAM, in either a dry or liquid form, at different application rates, had no appreciable effect on surface runoff volume. 4.2.2 Initial Turbidity Turbidity measurements were recorded at 1 minute intervals from thoroughly stirred runoff samples, collected in 5 qt buckets. This provided researchers with a total of 1,680 turbidity measurements. These measurements were identified as an initial turbidity reading for surface runoff, recorded from 1 minute intervals samples from each condition?s data was averaged together, as illustrated in Figure 4.3. Distribution of initial turbidity was observed to be consistent, with dry PAM at 35 lbs/acre performing the best. Throughout the duration of an experiment, dry PAM at 35 lbs/acre was able to reduce turbidity levels and maintain low readings. The dry PAM granular was exposed to water, at which point the PAM molecules were ?activated?. As PAM was slowly introduced into the runoff, the activated PAM molecules bonded with the suspended soil particles present 85 in the stormwater. This process promotes particles to flocculate and settle out as PAM and soil particles become larger. As PAM application rates decreased, initial turbidity measurements increased. This trend continued throughout all PAM treatments observed. Liquid PAM at 25 and 15 lbs/acre experienced a specific increase approximately at 40 minutes into the total experiment duration, with a greater increase occurring for test 4. This increase of initial turbidity was identified as the point at which these PAM treatments were no longer effective at reducing turbidity, as the PAM treatments were being effectively washed away by the runoff. The PAM molecules washed away were still capable of reducing sedimentation, but not as effectively as dry PAM treatments 0 500 1000 1500 2000 2500 3000 3500 4000 4500 0 5 10 15 20 25 30 35 40 45 50 55 60 Tu rb idi ty (N TU ) Time (min) Control Dry 35 Dry 25 Dry 15 Liquid 35 Liquid 25 Liquid 15 Test 1 Test 2 Test 3 Test 4 * * * Note: ?*? denotes 15 minute break in between tests Figure 4.3 Average Initial Turbidity of Surface Runoff vs. Time. 86 Table 4.7 and Table 4.8 show average turbidity measurements, standard deviation of the average turbidity, and a percent reduction, normalized for the control condition. As seen here, dry PAM at 35 lbs/acre performed the best with a reduction of approximately 97% for all tests. The increase of turbidity can also be observed in Table 4.8, as the percent reduction decreases from 49.4 % to 30.9 % and 36.9% to 32.1% for liquid PAM 25 and 15 lbs/acre, respectively. A slight reduction of turbidity occurs with all three applications of the dry PAM treatments over the duration of all four tests. Table 4.7 Average Initial Turbidity Results for Surface Runoff [Dry PAM] Condition Average Turbidity a (NTU) Standard Deviation b (NTU) Percent Reduction c Test 1 Control 3414.7 513.6 - Dry 35 103.4 28.9 97.0% Dry 25 620.0 113.8 81.8% Dry 15 1153.5 171.2 66.2% Test 2 Control 3405.4 395.6 - Dry 35 99.0 14.7 97.1% Dry 25 563.8 110.3 83.4% Dry 15 967.0 198.4 71.6% Test 3 Control 3553.6 304.3 - Dry 35 96.1 15.7 97.3% Dry 25 553.8 85.4 84.4% Dry 15 1018.1 156.0 71.4% Test 4 Control 3636.6 233.5 - Dry 35 95.5 15.2 97.4% Dry 25 496.1 54.3 86.4% Dry 15 1008.7 132.6 72.3% Notes: ?a? Average of initial turbidity vs. time for each test ?b? Standard deviation for average initial turbidity vs. time ?c? Denotes values normalized by control condition 87 Table 4.8 Average Initial Turbidity Results for Surface Runoff [Liquid PAM] Condition Average Turbidity a (NTU) Standard Deviation b (NTU) Percent Reduction c Test 1 Control 3414.7 513.6 - Liquid 35 784.8 167.3 77.0% Liquid 25 1726.4 262.6 49.4% Liquid 15 2153.1 417.3 36.9% Test 2 Control 3405.4 395.6 - Liquid 35 776.7 176.0 77.2% Liquid 25 1907.8 281.0 44.0% Liquid 15 2245.4 279.6 34.1% Test 3 Control 3553.6 304.3 - Liquid 35 775.5 187.4 78.2% Liquid 25 2036.2 279.0 42.7% Liquid 15 2240.9 274.3 36.9% Test 4 Control 3636.6 233.5 - Liquid 35 789.6 145.1 78.3% Liquid 25 2513.3 300.5 30.9% Liquid 15 2470.8 282.9 32.1% Notes: ?a? Average of initial turbidity vs. time for each test ?b? Standard deviation for average initial turbidity vs. time ?c? Denotes values normalized by control condition 4.2.2.1 Statistical Analysis: Initial Turbidity Differences observed between the control and treatments for the surface runoff?s initial turbidity measurements indicated that PAM treatments were having an effect on the amount of sediment contained in the runoff. To verify this observation and determine if this difference was statistically significant, ANOVA tables were created (Appendix E) and Tukey-Kramer tests were conducted to determine statistical significance between 88 individual pairs of groups, as illustrated in Table 4.9, Table 4.10, Table 4.11, and Table 4.12. As expected, these tables demonstrate that the initial turbidity averages had statistically significant differences between all possible pairs tested. Therefore, it can be observed that the addition of PAM had a significant effect on its ability to reduce initial turbidity from surface runoff generated from test plots. Also, statistically significant differences were observed between individual PAM treatments, which indicate that different application rates were statistically significantly different when compared. Examining the mean differences, reported in Table 4.9, Table 4.10, Table 4.11, and Table 4.12, it can be seen which treatment was more effective. Dry PAM applied as 35 lbs/acre performed the best out of the tested treatments in reducing initial turbidity for the bare soil control during all 4 tests. Dry PAM 35 also clearly performed well when compared to other treatments and respective applications rates. Even though it was observed that the effectiveness of liquid PAM 25 and 15 lbs/acre began to diminish during test 4, reductions in initial turbidity were still observed to be statistically significant. 89 Table 4.9 Tukey-Kramer Multiple Comparisons on Average Initial Turbidity [Test 1] Comparison ?i - ?j CI [LB] CI [UB] Significantly Different Control vs. Dry 35 3311.3 3187.5 3435.1 Yes Control vs. Dry 25 2261.2 2137.5 2385.0 Yes Control vs. Dry 15 2794.7 2670.9 2918.4 Yes Control vs. Liquid 35 2629.9 2506.2 2753.7 Yes Control vs. Liquid 25 1261.6 1137.8 1385.3 Yes Control vs. Liquid 15 1688.3 1564.6 1812.1 Yes Dry 35 vs. Dry 25 1050.1 926.3 1173.8 Yes Dry 35 vs. Dry 15 516.6 392.9 640.4 Yes Dry 35 vs. Liquid 35 681.4 557.6 805.1 Yes Dry 35 vs. Liquid 25 2049.7 1926.0 2173.5 Yes Dry 35 vs. Liquid 15 1623.0 1499.2 1746.7 Yes Dry 25 vs. Dry 15 533.4 409.7 657.2 Yes Dry 25 vs. Liquid 35 368.7 244.9 492.5 Yes Dry 25 vs. Liquid 25 999.7 875.9 1123.4 Yes Dry 25 vs. Liquid 15 572.9 449.1 696.7 Yes Dry 15 vs. Liquid 35 164.7 41.0 288.5 Yes Dry 15 vs. Liquid 25 1533.1 1409.3 1656.9 Yes Dry 15 vs. Liquid 15 1106.3 982.6 1230.1 Yes Liquid 35 vs. Liquid 25 1368.4 1244.6 1492.1 Yes Liquid 35 vs. Liquid 15 941.6 817.8 1065.4 Yes Liquid 25 vs. Liquid 15 426.8 303.0 550.5 Yes Notes: [LB] signifies lower bound of confidence interval [UB] signifies upper bound of confidence interval qcrit = 4.26 Table 4.10 Tukey-Kramer Multiple Comparisons on Average Initial Turbidity [Test 2] Comparison ?i - ?j CI [LB] CI [UB] Significantly Different Control vs. Dry 35 3306.4 3203.1 3409.8 Yes Control vs. Dry 25 2438.4 2335.1 2541.8 Yes Control vs. Dry 15 2841.6 2738.2 2944.9 Yes Control vs. Liquid 35 2628.7 2525.3 2732.0 Yes Control vs. Liquid 25 1164.5 1061.1 1267.8 Yes Control vs. Liquid 15 1497.6 1394.2 1600.9 Yes Dry 35 vs. Dry 25 868.0 764.6 971.4 Yes Dry 35 vs. Dry 15 464.9 361.5 568.2 Yes Dry 35 vs. Liquid 35 677.8 574.4 781.1 Yes Dry 35 vs. Liquid 25 2141.9 2038.6 2245.3 Yes Dry 35 vs. Liquid 15 1808.8 1705.5 1912.2 Yes Dry 25 vs. Dry 15 403.2 299.8 506.5 Yes Dry 25 vs. Liquid 35 190.3 86.9 293.6 Yes Dry 25 vs. Liquid 25 1273.9 1170.6 1377.3 Yes Dry 25 vs. Liquid 15 940.8 837.5 1044.2 Yes Dry 15 vs. Liquid 35 212.9 109.5 316.3 Yes Dry 15 vs. Liquid 25 1677.1 1573.7 1780.4 Yes Dry 15 vs. Liquid 15 1344.0 1240.6 1447.3 Yes Liquid 35 vs. Liquid 25 1464.2 1360.8 1567.5 Yes Liquid 35 vs. Liquid 15 1131.1 1027.7 1234.4 Yes Liquid 25 vs. Liquid 15 333.1 229.7 436.5 Yes Notes: [LB] signifies lower bound of confidence interval [UB] signifies upper bound of confidence interval qcrit = 4.26 90 Table 4.11 Tukey-Kramer Multiple Comparisons on Average Initial Turbidity [Test 3] Comparison ?i - ?j CI [LB] CI [UB] Significantly Different Control vs. Dry 35 3457.5 3350.2 3564.8 Yes Control vs. Dry 25 2535.6 2428.2 2642.9 Yes Control vs. Dry 15 2999.9 2892.6 3107.2 Yes Control vs. Liquid 35 2778.2 2670.9 2885.5 Yes Control vs. Liquid 25 1082.8 975.5 1190.1 Yes Control vs. Liquid 15 1517.4 1410.1 1624.7 Yes Dry 35 vs. Dry 25 921.9 814.6 1029.2 Yes Dry 35 vs. Dry 15 457.6 350.3 564.9 Yes Dry 35 vs. Liquid 35 679.3 572.0 786.6 Yes Dry 35 vs. Liquid 25 2374.7 2267.4 2482.0 Yes Dry 35 vs. Liquid 15 1940.1 1832.8 2047.4 Yes Dry 25 vs. Dry 15 464.3 357.0 571.6 Yes Dry 25 vs. Liquid 35 242.6 135.3 349.9 Yes Dry 25 vs. Liquid 25 1452.7 1345.4 1560.0 Yes Dry 25 vs. Liquid 15 1018.2 910.8 1125.5 Yes Dry 15 vs. Liquid 35 221.7 114.4 329.0 Yes Dry 15 vs. Liquid 25 1917.1 1809.8 2024.4 Yes Dry 15 vs. Liquid 15 1482.5 1375.2 1589.8 Yes Liquid 35 vs. Liquid 25 1695.4 1588.0 1802.7 Yes Liquid 35 vs. Liquid 15 1260.8 1153.5 1368.1 Yes Liquid 25 vs. Liquid 15 434.6 327.3 541.9 Yes Notes: [LB] signifies lower bound of confidence interval [UB] signifies upper bound of confidence interval qcrit = 4.26 Table 4.12 Tukey-Kramer Multiple Comparisons on Average Initial Turbidity [Test 4] Comparison ?i - ?j CI [LB] CI [UB] Significantly Different Control vs. Dry 35 3541.1 3397.7 3684.5 Yes Control vs. Dry 25 2627.9 2484.5 2771.3 Yes Control vs. Dry 15 3140.5 2997.1 3283.9 Yes Control vs. Liquid 35 2847.0 2703.6 2990.4 Yes Control vs. Liquid 25 492.0 348.6 635.4 Yes Control vs. Liquid 15 1123.3 979.9 1266.7 Yes Dry 35 vs. Dry 25 913.2 769.8 1056.6 Yes Dry 35 vs. Dry 15 400.6 257.2 544.0 Yes Dry 35 vs. Liquid 35 694.1 550.7 837.5 Yes Dry 35 vs. Liquid 25 3049.1 2905.7 3192.5 Yes Dry 35 vs. Liquid 15 2417.8 2274.4 2561.2 Yes Dry 25 vs. Dry 15 512.6 369.2 656.0 Yes Dry 25 vs. Liquid 35 219.1 75.7 362.5 Yes Dry 25 vs. Liquid 25 2136.0 1992.6 2279.4 Yes Dry 25 vs. Liquid 15 1504.6 1361.2 1648.0 Yes Dry 15 vs. Liquid 35 293.5 150.1 436.9 Yes Dry 15 vs. Liquid 25 2648.6 2505.1 2792.0 Yes Dry 15 vs. Liquid 15 2017.2 1873.8 2160.6 Yes Liquid 35 vs. Liquid 25 2355.1 2211.6 2498.5 Yes Liquid 35 vs. Liquid 15 1723.7 1580.3 1867.1 Yes Liquid 25 vs. Liquid 15 631.4 488.0 774.8 Yes Notes: [LB] signifies lower bound of confidence interval [UB] signifies upper bound of confidence interval qcrit = 4.26 91 4.2.3 Turbidity vs. Time In 2008, the EPA began the process of introducing effluent limits on construction runoff. These limitations propose that stormwater runoff must contain a turbidity level less than or equal to 13 NTUs (U.S. EPA, 2008b). However, while these limitations are currently being subjected to review, a potential exists for a required turbidity level. Therefore, for the purposes of this research, runoff samples were examined to determine whether PAM treatments could achieve EPA?s proposed effluent limitation of 13 NTUs. Samples were collected during each of the four tests at predetermined times of 5 and 10 minutes to observe turbidity over time. This was accomplished for each test plot, providing researchers with 224 runoff samples (e.g. 7 conditions ? 4 test plots ? 4 tests x 2 samples = 224 runoff samples) to measure and record turbidity over time. This collected data was representative of time require for suspended soil particles in the runoff to settle out. This collected data for turbidity versus time can be observed in Figure 4.4. The time selected to observe turbidity over time was originally 10 minutes, with measurements occurring at 1 minute intervals, producing a curve that shows turbidity slowly decreasing over time. However, examining PAM-treated samples; the decrease in recorded turbidity was nearly instantaneous. Therefore, the observational time was reduced to approximately 3 minutes; with readings measured every 10 seconds to better capture the speed at which particles were settling. As illustrated in Figure 4.4, sediment in PAM treatments was capable of settling out much more quickly than the control. By the end of the 10 minute observational period, recorded turbidity for the control had not decreased to 500 NTUs. Only dry PAM applied at 35 and 25 lbs/acre were capable of 92 obtaining EPA?s effluent limit of 13 NTU within the observed time period. Dry PAM 35 obtained 13 NTUs within 20 seconds and dry PAM 25 in 3 minutes and 20 seconds. 0 500 1000 1500 2000 2500 3000 3500 0 1 2 3 4 5 6 7 8 9 10 Tu rb idi ty (N TU ) Time (min) Control Dry 35 Dry 25 Dry 15 Liquid 35 Liquid 25 Liquid 15 Figure 4.4 Average Recorded Turbidity for All Samples vs. Time. Further investigations were conducted on turbidity versus time measurements to determine relative performance between treatments and respective application rates. Data for each treatment and application were graphed for their respective collection times of 5 and 10 minutes. Figure 4.5 and Figure 4.6 illustrate these results for dry and liquid PAM treatments, respectively. It was observed in Figure 4.5 that dry PAM treatments initial turbidity readings perform consistently throughout an experiment?s duration, dince all four test produce graphs with similar shapes. 93 0 500 1000 1500 2000 2500 3000 3500 4000 0 20 40 60 80 100 120 140 160 180 200 Tu rb idi ty (N TU ) Time (sec) Test 1 Test 2 Test 3 Test 4 0 500 1000 1500 2000 2500 3000 3500 4000 0 20 40 60 80 100 120 140 160 180 200 Tu rb idi ty (N TU ) Time (sec) Test 1 Test 2 Test 3 Test 4 (a) 35 lbs/acre at 5 Minutes (b) 35 lbs/acre at 10 Minutes 0 500 1000 1500 2000 2500 3000 3500 4000 0 20 40 60 80 100 120 140 160 180 200 Tu rb idi ty (N TU ) Time (sec) Test 1 Test 2 Test 3 Test 4 0 500 1000 1500 2000 2500 3000 3500 4000 0 20 40 60 80 100 120 140 160 180 200 Tu rb idi ty (N TU ) Time (sec) Test 1 Test 2 Test 3 Test 4 (c) 25 lbs/acre at 5 Minutes (d) 25 lbs/acre at 10 Minutes 0 500 1000 1500 2000 2500 3000 3500 4000 0 20 40 60 80 100 120 140 160 180 200 Tu rb idi ty (N TU ) Time (sec) Test 1 Test 2 Test 3 Test 4 0 500 1000 1500 2000 2500 3000 3500 4000 0 20 40 60 80 100 120 140 160 180 200 Tu rb idi ty (N TU ) Time (sec) Test 1 Test 2 Test 3 Test 4 (e) 15 lbs/acre at 5 Minutes (f) 15 lbs/acre at 10 Minutes Figure 4.5 Treatment?s Average Recorded Turbidity from 5 and 10 Minute Samples [Dry PAM]. 94 0 500 1000 1500 2000 2500 3000 3500 4000 0 20 40 60 80 100 120 140 160 180 200 Tu rb idi ty (N TU ) Time (sec) Test 1 Test 2 Test 3 Test 4 0 500 1000 1500 2000 2500 3000 3500 4000 0 20 40 60 80 100 120 140 160 180 200 Tu rb idi ty (N TU ) Time (sec) Test 1 Test 2 Test 3 Test 4 (a) 35 lbs/acre at 5 Minutes (b) 35 lbs/acre at 10 Minutes 0 500 1000 1500 2000 2500 3000 3500 4000 0 20 40 60 80 100 120 140 160 180 200 Tu rb idi ty (N TU ) Time (sec) Test 1 Test 2 Test 3 Test 4 0 500 1000 1500 2000 2500 3000 3500 4000 0 20 40 60 80 100 120 140 160 180 200 Tu rb idi ty (N TU ) Time (sec) Test 1 Test 2 Test 3 Test 4 (c) 25 lbs/acre at 5 Minutes (d) 25 lbs/acre at 10 Minutes 0 500 1000 1500 2000 2500 3000 3500 4000 0 20 40 60 80 100 120 140 160 180 200 Tu rb idi ty (N TU ) Time (sec) Test 1 Test 2 Test 3 Test 4 0 500 1000 1500 2000 2500 3000 3500 4000 0 20 40 60 80 100 120 140 160 180 200 Tu rb idi ty (N TU ) Time (sec) Test 1 Test 2 Test 3 Test 4 (e) 15 lbs/acre at 5 Minutes (f) 15 lbs/acre at 10 Minutes Figure 4.6 Treatment?s Average Recorded Turbidity from 5 and 10 Minute Samples [Liquid PAM]. 95 Figure 4.6 illustrates that turbidity over time increased for liquid PAM treatments through an experiments individual rainfall events (i.e. test 1 through 4), as a function of application rate. Specifically, both liquid PAM 25 and 15 lbs/acre, as observed in Figure 4.6 (c) through 4.6(f), experienced an increase of turbidity measurements during test 3 and test 4 for both treatments. This is indicative that the performance of liquid PAM at these application rates was not as effective as reducing turbidity levels as seen with dry PAM treatments. This also confirms the previous observation that liquid PAM 25 and 15?s effectiveness had begun to diminish during test 3 and test 4 when examining initial turbidity readings. While recorded turbidity levels were still lower than the control condition, the data shows that higher application rates of PAM were more effective in reducing settling time of sediment particles in surface runoff as a function of time. Table 4.13 illustrates turbidity measurements taken at 200 seconds, averaged from all four tests. These measurements were the final data point collected while observing turbidity versus time. It can be seen that final measurements for dry PAM at 35 lbs/acre reached low turbidity levels around 10 NTUs. Dry PAM 25 and 15 lbs/acre also reached low levels, around 15 and 20 NTUs, respectively. These values show how effective dry PAM was at reducing sedimentation during experimentation. Liquid PAM treatments, while still capable of reducing turbidity levels, were higher than observed with dry PAM treatments, indicating that liquid PAM was not as effective at controlling sedimentation. 96 Table 4.13 Summary of Turbidity vs. Time Measurements Condition Turbidity (NTU) 1 5 min 10 min Dry 35 10.2 9.2 Dry 25 15.2 14.7 Dry 15 21.8 20.1 Liquid 35 40.4 51.5 Liquid 25 78.3 84.2 Liquid 15 77.0 73.7 Note: 1. Turbidity measurement at 200 sec. This observed difference between the performances of dry PAM compared to liquid PAM was attributed to how the dry granules of PAM slowly dissolve during experimentation. These dry granules of PAM were activated once water (i.e. rainfall and surface runoff) was applied to the test plots. As the rainfall passed over the granules, the dry PAM treatment was slowly and consistently introduced into the runoff. The liquid PAM treatments were sprayed on and ?distributed? to the soil and were washed away more quickly. This was observed specifically during test 3 and 4 of the liquid PAM treatment at 25 and 15 lbs/acre. This effect was magnified since this research examined the ?worse-case scenario? and did not allow liquid PAM treatments time to dry after the initial application. Therefore, the dry PAM provided long-term protection and enabled the dry PAM treatments to perform better than the liquid PAM treatments at these conditions. Liquid PAM that is formed through mixing dry granular with water becomes activated at this point. Once applied on the soil, the PAM molecules will bond with the soil surface and provided a protective layer. However, the effectiveness of this protective layer is dependent on two factors: (1) uniform coverage and (2) time allotted for drying. Therefore, since this research focused on a ?worse-case scenario? for liquid PAM, (i.e. no 97 time was allotted for drying), the uniform coverage of the spray area was the main controlling factor in the performance of liquid PAM. Appendix F in this report covers additional experiments that examined the effect of liquid PAM that was allowed to dry for 48 hours in comparison to the other PAM treatments presented in this chapter. 4.2.4 Soil Loss Soil samples, which were representative of the amount of eroded soil from each test plot, were collected from surface runoff every three minutes for a total of 560 observations (e.g. 7 conditions ? 4 test plots ? 4 tests ? 5 observations per test = 560 total recorded measurements)for all experiments conducted. Samples were oven dried and weighed to determine a soil loss for each experimental setup. Figure 4.7(a) illustrates the average values of eroded soil during an experiment?s duration. The control condition and dry PAM applications experienced an initial surge of sediment contained within the runoff. Following this surge, eroded soil levels achieved a steady state and remained relatively constant throughout the four tests. It was observed that dry PAM treatments consistently produced levels of sediment that were less than the bare soil. The application rate of 35 lbs/acre performed better than the 25 and 15 lbs/acre treatments at reducing the amount of eroded soil from the plots. All three liquid PAM treatments produced levels of eroded soil that were less than the bare soil control, but only during test 1. Subsequent tests showed that the liquid PAM treatments had similar soil losses when compared to the control. However, liquid PAM treatments were not capable of producing less eroded soil than dry PAM treatments. Figure 4.7(b) and 4.7(c) display the amount of eroded soil for both the dry and liquid PAM treatments, respectively. 98 0 500 1000 1500 2000 2500 0 5 10 15 20 25 30 35 40 45 50 55 60 M ass (lb /ac re) Time (min) Control Dry 35 Dry 25 Dry 15 Liquid 35 Liquid 25 Liquid 15 Test 1 Test 2 Test 3 Test 4 * * * (a) Dry and Liquid PAM Treatments vs. Control 0 500 1000 1500 2000 2500 0 5 10 15 20 25 30 35 40 45 50 55 60 M as s ( lb/ ac re) Time (min) Control Dry 35 Dry 25 Dry 15 Test 1 Test 2 Test 3 Test 4 * * * 0 500 1000 1500 2000 2500 0 5 10 15 20 25 30 35 40 45 50 55 60 M as s ( lb/ ac re) Time (min) Control Liquid 35 Liquid 25 Liquid 15 Test 1 Test 2 Test 3 Test 4 * * * (b) Dry PAM Treatments (b) Liquid PAM Treatments Note: ?*? denotes 15 minute break in between tests [applies to all figures] Figure 4.7 Average Soil Loss versus Time. Figure 4.8 illustrates the cumulative amount of eroded soil during each 15 minute rain event. Dry PAM applied at 35 lbs/acre produced less total sediment for each test when compared to all other treatments. All three of the different application rates of dry PAM remained constant during an experiment, maintaining a relatively consistent level of protection. The effectiveness of liquid PAM treatments during test 1 can be observed in Figure 4.8. However, subsequent tests show that the liquid PAM treatments produced quantities of eroded soil comparative to that of the control, bringing into question the effectiveness of liquid PAM?s long term potential. 99 0 2000 4000 6000 8000 10000 0 5 10 15 20 25 30 35 40 45 50 55 60 M ass (lb /ac re) Time (min) Control Dry 35 Dry 25 Dry 15 Liquid 35 Liquid 25 Liquid 15 Test 1 Test 2 Test 3 Test 4 * * * (a) Dry and Liquid PAM Treatments vs. Control 0 2000 4000 6000 8000 10000 0 5 10 15 20 25 30 35 40 45 50 55 60 M as s ( lb/ ac re) Time (min) Control Dry 35 Dry 25 Dry 15 Test 1 Test 2 Test 3 Test 4 * * * 0 2000 4000 6000 8000 10000 0 5 10 15 20 25 30 35 40 45 50 55 60 M as s ( lb/ ac re) Time (min) Control Liquid 35 Liquid 25 Liquid 15 Test 1 Test 2 Test 3 Test 4 * * * (b) Dry PAM Treatments (b) Liquid PAM Treatments Note: ?*? denotes 15 minute break in between tests [applies to all figures] Figure 4.8 Average Cumulative Soil Loss vs. Time. Table 4.14 and Table 4.15 contain the specific values of average soil loss and cumulative soil loss for dry and liquid PAM treatments, respectively. The dry PAM treatments, at 35 lbs/acre, produced on average, less eroded soil at around 700 to 900 lbs/acre for each test. Lower application rates for dry PAM had higher amounts of eroded soil measuring around 1,000 lbs/acre. Liquid PAM treatments saw a higher amount of soil loss, with all treatments producing amounts of soil measuring from around 1,200 to 1,500 lbs/acre. Dry PAM 35 lbs/acre performed the best among of all PAM treatments by reducing soil loss by about 50% for all tests. Other application rates of dry PAM were 100 capable of reducing soil loss with measurements ranging from 10 to 30% reductions. Liquid PAM treatments saw a similar reduction of soil loss, but only during test 1 (i.e. first 15 min). Tests 2 through 4 showed that liquid PAM treatments had no effect on the amount of eroded soil produced. As was seen with turbidity results, these differences in the amount of eroded soil was attributed to the effectiveness of the dry granules being slowly and consistently dissolved and introduced into the runoff providing long-term performance when compared to liquid PAM treatments. To further analyze the effect PAM treatments had at reducing erosion, the average recorded amount of eroded soil was compared with its corresponding runoff sample to generate ?percent sediment?, as illustrated in Figure 4.9(a) and (b) for dry and liquid PAM treatments, respectively. This percent sediment shows the amount of sediment contained within a runoff sample during the overall experiment duration when compared to the bare soil condition. The initial surge of sediment is illustrated in these figures, and reaches a steady state around 6 minutes during the first test. As already stated, dry PAM at 35 lbs/acre was observed to perform the best, with collected runoff samples that contained around 1 to 1.5% of sediment. Dry PAM 25 and 15 lbs/acre had sediment levels ranging from 1.75 to 2.25% sediment, which was still observed to be less than recorded with the control. However, liquid PAM treatments appeared to have no effect at reducing the percent of sediment contained within the runoff, with values consistent with the control. 101 0.5% 1.0% 1.5% 2.0% 2.5% 3.0% 3.5% 4.0% 4.5% 0 5 10 15 20 25 30 35 40 45 50 55 60 Pe rce nt Se dim en t Time (min) Control Dry 35 Dry 25 Dry 15 * * * Test 1 Test 2 Test 3 Test 4 (a) Dry PAM Treatment 0.5% 1.0% 1.5% 2.0% 2.5% 3.0% 3.5% 4.0% 4.5% 0 5 10 15 20 25 30 35 40 45 50 55 60 Pe rce nt Se dim en t Time (min) Control Liquid 35 Liquid 25 Liquid 15 * * * Test 1 Test 2 Test 3 Test 4 (b) Liquid PAM Treatments Note: ?*? denotes 15 minute break in between tests [applies to all figures] Figure 4.9 Average Percent Sediment in Surface Runoff vs. Time. 10 2 Table 4.14 Average Soil Loss due to Surface Runoff [Dry PAM] Condition Soil Loss a (lbs/acre) Standard Deviation b (lbs/acre) Percent Reduction c Test 1 Control 1663.8 693.8 - Dry 35 798.9 293.5 52.0% Dry 25 1048.6 300.6 37.0% Dry 15 1052.8 102.0 36.7% Test 2 Control 1210.6 189.7 - Dry 35 674.6 262.9 44.3% Dry 25 982.5 175.0 18.8% Dry 15 1084.0 149.7 10.5% Test 3 Control 1420.7 236.3 - Dry 35 740.6 308.2 47.9% Dry 25 1024.5 206.1 27.9% Dry 15 1014.9 117.1 28.6% Test 4 Control 1506.5 194.0 - Dry 35 787.5 334.6 47.7% Dry 25 1039.6 204.8 31.0% Dry 15 998.7 92.3 33.7% Notes: ?a? Average of eroded soil vs. time for each test ?b? Standard deviation for average soil loss vs. time ?c? Denotes values normalized by control condition Table 4.15 Average Soil Loss due to Surface Runoff [Liquid PAM] Condition Soil Loss a (lbs/acre) Standard Deviation b (lbs/acre) Percent Reduction c Test 1 Control 1663.8 693.8 - Liquid 35 1307.2 377.5 21.4% Liquid 25 1185.4 230.8 28.8% Liquid 15 1276.6 178.0 23.3% Test 2 Control 1210.6 189.7 - Liquid 35 1337.9 322.9 -10.5% Liquid 25 1243.6 172.7 -2.7% Liquid 15 1284.4 145.9 -6.1% Test 3 Control 1420.7 236.3 - Liquid 35 1419.5 256.0 0.1% Liquid 25 1390.7 172.0 2.1% Liquid 15 1463.3 205.8 -3.0% Test 4 Control 1506.5 194.0 - Liquid 35 1421.9 233.5 5.6% Liquid 25 1372.7 147.8 8.9% Liquid 15 1496.9 185.4 0.6% Notes: ?a? Average of eroded soil vs. time for each test ?b? Standard deviation for average soil loss vs. time ?c? Denotes values normalized by control condition 103 4.2.4.1 Statistical Analysis: Soil Loss Continuing the statistical analyses used throughout this research, ANOVA procedures with a Tukey-Kramer multiple comparisons were used for the recorded amounts of soil loss. Statistically significant results are reported in bold in Table 4.16 through Table 4.19. During test 1, only dry PAM at 35 lbs/acre produced statistically significant results when compared in comparison to the control, as illustrated in Table 4.16. In addition to being statistically significant to the control, dry PAM at 35 lbs/acre was determined to be significant when compared to all three applications of liquid PAM. This indicates that dry PAM at 35 lbs/acre outperformed liquid treatments for test 1. As experimental duration increases through tests 2, 3, and 4, dry PAM at 25 and 15 lbs/acre are observed to be statistically significant. This delay in effectiveness observed with dry PAM 25 and 15 could be attributed to a longer required time for treatments to reach necessary levels of PAM concentrations capable of reducing soil loss and sedimentation. Test plots treated with 35 lbs/acre of PAM had higher amounts of PAM applied, resulting in more product available for reducing erosion and sedimentation. Therefore, lower application rates of PAM required more time to become active and provide a level of effectiveness. Differences were also observed between the lower rates of dry PAM when compared to the liquid PAM treatments in later tests, indicating that even the lower rates of dry PAM were capable of outperforming liquid PAM. For all test conducted, no statistical significance between liquid PAM treatments and the bare soil condition was observed with respect to soil loss. 10 4 Table 4.16 Tukey-Kramer Multiple Comparisons on Average Soil Loss [Test 1] Comparison ?i - ?j CI [LB] CI [UB] Significantly Different Control vs. Dry 35 386.5 127.1 386.5 Yes Control vs. Dry 25 132.6 -126.8 132.6 No Control vs. Dry 15 136.8 -122.6 136.8 No Control vs. Liquid 35 121.8 -137.6 121.8 No Control vs. Liquid 25 91.2 -168.2 91.2 No Control vs. Liquid 15 0.0 -259.5 0.0 No Dry 35 vs. Dry 25 253.9 -5.6 253.9 No Dry 35 vs. Dry 15 249.7 -9.8 249.7 No Dry 35 vs. Liquid 35 508.4 248.9 508.4 Yes Dry 35 vs. Liquid 25 477.8 218.3 477.8 Yes Dry 35 vs. Liquid 15 386.5 127.1 386.5 Yes Dry 25 vs. Dry 15 4.2 -255.3 4.2 No Dry 25 vs. Liquid 35 254.5 -5.0 254.5 No Dry 25 vs. Liquid 25 223.9 -35.6 223.9 No Dry 25 vs. Liquid 15 132.6 -126.8 132.6 No Dry 15 vs. Liquid 35 258.7 -0.8 258.7 No Dry 15 vs. Liquid 25 228.1 -31.4 228.1 No Dry 15 vs. Liquid 15 136.8 -122.6 136.8 No Liquid 35 vs. Liquid 25 30.6 -228.8 30.6 No Liquid 35 vs. Liquid 15 121.8 -137.6 121.8 No Liquid 25 vs. Liquid 15 91.2 -168.2 91.2 No Notes: [LB] signifies lower bound of confidence interval [UB] signifies upper bound of confidence interval qcrit = 4.26 Table 4.17 Tukey-Kramer Multiple Comparisons on Average Soil Loss [Test 2] Comparison ?i - ?j CI [LB] CI [UB] Significantly Different Control vs. Dry 35 569.0 352.7 569.0 Yes Control vs. Dry 25 159.7 -56.7 159.7 No Control vs. Dry 15 261.1 44.8 261.1 Yes Control vs. Liquid 35 94.2 -122.1 94.2 No Control vs. Liquid 25 40.8 -175.5 40.8 No Control vs. Liquid 15 0.0 -216.3 0.0 No Dry 35 vs. Dry 25 409.3 193.0 409.3 Yes Dry 35 vs. Dry 15 307.9 91.6 307.9 Yes Dry 35 vs. Liquid 35 663.2 446.9 663.2 Yes Dry 35 vs. Liquid 25 609.8 393.5 609.8 Yes Dry 35 vs. Liquid 15 569.0 352.7 569.0 Yes Dry 25 vs. Dry 15 101.4 -114.9 101.4 No Dry 25 vs. Liquid 35 253.9 37.5 253.9 Yes Dry 25 vs. Liquid 25 200.5 -15.9 200.5 No Dry 25 vs. Liquid 15 159.7 -56.7 159.7 No Dry 15 vs. Liquid 35 355.3 139.0 355.3 Yes Dry 15 vs. Liquid 25 301.9 85.6 301.9 Yes Dry 15 vs. Liquid 15 261.1 44.8 261.1 Yes Liquid 35 vs. Liquid 25 53.4 -162.9 53.4 No Liquid 35 vs. Liquid 15 94.2 -122.1 94.2 No Liquid 25 vs. Liquid 15 40.8 -175.5 40.8 No Notes: [LB] signifies lower bound of confidence interval [UB] signifies upper bound of confidence interval qcrit = 4.26 10 5 Table 4.18 Tukey-Kramer Multiple Comparisons on Average Soil Loss [Test 3] Comparison ?i - ?j CI [LB] CI [UB] Significantly Different Control vs. Dry 35 650.0 421.6 650.0 Yes Control vs. Dry 25 375.7 147.3 375.7 Yes Control vs. Dry 15 366.1 137.7 366.1 Yes Control vs. Liquid 35 28.8 -199.6 28.8 No Control vs. Liquid 25 72.6 -155.8 72.6 No Control vs. Liquid 15 0.0 -228.5 0.0 No Dry 35 vs. Dry 25 274.3 45.8 274.3 Yes Dry 35 vs. Dry 15 283.9 55.4 283.9 Yes Dry 35 vs. Liquid 35 678.8 450.4 678.8 Yes Dry 35 vs. Liquid 25 722.6 494.2 722.6 Yes Dry 35 vs. Liquid 15 650.0 421.6 650.0 Yes Dry 25 vs. Dry 15 9.6 -218.8 9.6 No Dry 25 vs. Liquid 35 404.5 176.1 404.5 Yes Dry 25 vs. Liquid 25 448.4 219.9 448.4 Yes Dry 25 vs. Liquid 15 375.7 147.3 375.7 Yes Dry 15 vs. Liquid 35 394.9 166.5 394.9 Yes Dry 15 vs. Liquid 25 438.7 210.3 438.7 Yes Dry 15 vs. Liquid 15 366.1 137.7 366.1 Yes Liquid 35 vs. Liquid 25 43.8 -184.6 43.8 No Liquid 35 vs. Liquid 15 28.8 -199.6 28.8 No Liquid 25 vs. Liquid 15 72.6 -155.8 72.6 No Notes: [LB] signifies lower bound of confidence interval [UB] signifies upper bound of confidence interval qcrit = 4.26 Table 4.19 Tukey-Kramer Multiple Comparisons on Average Soil Loss [Test 4] Comparison ?i - ?j CI [LB] CI [UB] Significantly Different Control vs. Dry 35 585.2 406.8 585.2 Yes Control vs. Dry 25 373.9 195.6 373.9 Yes Control vs. Dry 15 333.1 154.7 333.1 Yes Control vs. Liquid 35 49.2 -129.2 49.2 No Control vs. Liquid 25 124.2 -54.1 124.2 No Control vs. Liquid 15 0.0 -178.4 0.0 No Dry 35 vs. Dry 25 211.3 32.9 211.3 Yes Dry 35 vs. Dry 15 252.1 73.7 252.1 Yes Dry 35 vs. Liquid 35 634.4 456.0 634.4 Yes Dry 35 vs. Liquid 25 709.4 531.1 709.4 Yes Dry 35 vs. Liquid 15 585.2 406.8 585.2 Yes Dry 25 vs. Dry 15 40.8 -137.6 40.8 No Dry 25 vs. Liquid 35 423.1 244.8 423.1 Yes Dry 25 vs. Liquid 25 498.2 319.8 498.2 Yes Dry 25 vs. Liquid 15 373.9 195.6 373.9 Yes Dry 15 vs. Liquid 35 382.3 204.0 382.3 Yes Dry 15 vs. Liquid 25 457.4 279.0 457.4 Yes Dry 15 vs. Liquid 15 333.1 154.7 333.1 Yes Liquid 35 vs. Liquid 25 75.0 -103.3 75.0 No Liquid 35 vs. Liquid 15 49.2 -129.2 49.2 No Liquid 25 vs. Liquid 15 124.2 -54.1 124.2 No Notes: [LB] signifies lower bound of confidence interval [UB] signifies upper bound of confidence interval qcrit = 4.26 106 4.2.5 Soil Particle Size As stated in Section 3.4.1, a particle size analysis was conducted on the soil prior to experimentation. These recorded data were used during the analysis and identification procedures, in addition to comparison with soil particles transported during runoff. This allowed researchers to observe any differences in particle size between the initial soil and soil transported by runoff. Therefore, samples of soil contained within the runoff were taken and resulting particle sizes were determined. These samples of runoff soil were obtained from the oven dried soil used for determining quantities of eroded soil. Figure 4.10(a) illustrates particle size distribution for the initial soil, prior to experimentation (i.e. stock pile soil), and soil contained in the runoff generated during testing. Figure 4.10(b) and 4.10(c) show particles sizes that were determined for dry and liquid PAM treatments, respectively. As seen in Figure 4.10, soil contained within the runoff was smaller in size when compared to the stock pile soil. This was expected, because smaller particles could be more easily detached and transported by runoff. However, there was no appreciable difference observed between PAM treatments and resulting particle sizes, as all recorded particle distributions produced similar values and curves. 107 0 10 20 30 40 50 60 70 80 90 100 0.0010 0.0050 0.0250 0.1250 0.6250 Pe rce nt Fi ne r Particle Size (in) Stock Pile Control Dry 35 Dry 25 Dry 15 Liquid 35 Liquid 25 Liquid 15 (a) Dry and Liquid PAM Treatments vs. Control and Stock Pile Soil 0 10 20 30 40 50 60 70 80 90 100 0.0010 0.0050 0.0250 0.1250 0.6250 Pe rce nt Fin er Particle Size (in) Stock Pile Control Dry 35 Dry 25 Dry 15 0 10 20 30 40 50 60 70 80 90 100 0.0010 0.0050 0.0250 0.1250 0.6250 Pe rce nt Fin er Particle Size (in) Stock Pile Control Liquid 35 Liquid 25 Liquid 15 (b) Dry PAM Treatments (b) Liquid PAM Treatments Figure 4.10 Average Particle Size Distribution of Stock Pile Soil and Surface Runoff. 4.2.6 Initial Turbidity vs. Soil Loss To determine if any relationship existed between the average recorded initial turbidity and average measured soil loss, values were plotted together, as demonstrated in Figure 4.11. Initially, it appears that turbidity and soil loss may be linearly related, but distinct groupings of data pertaining to individual treatments were observed. The control condition, with high levels and turbidity and large amounts of eroded soil appear together on the right side of the graph. Conversely, dry PAM at 35 lbs/acre, with recorded low turbidity levels and low soil loss appear on the left. This relationship could be used as a method to determine which treatment was most effective at reducing turbidity and soil 108 loss. Therefore, it was observed that dry PAM at 35 lbs/acre was the most effective treatments for reducing erosion and sedimentation. Other application rates and treatments can be seen in their respective group in relationship to turbidity and soil loss. 0 500 1000 1500 2000 2500 0 500 1000 1500 2000 2500 3000 3500 4000 So il L os s (l bs /ac re) Turbidity (NTU) Control Dry 35 Dry 25 Dry 15 Liq 35 Liq 25 Liq 15 Figure 4.11 Average Initial Turbidity vs. Average Eroded Soil. 4.3 SUMMARY Data collected from experimentation provided researchers information that was used to evaluate which PAM treatments and application rates were most effective at reducing erosion and sedimentation for this particular test soil. Data that were collected included: (1) surface runoff, (2) initial turbidity, (3) turbidity versus time, (4) soil loss, and (5) particle size distribution. ANOVA procedures were conducted on the data to assist in determining the effect different treatments had and if any statistical significant results were observed. Runoff samples were collected every minute for all four rain events simulated. These samples were measured to determine volume of surface runoff generated by the rainfall. A total of 1,680 runoff samples were collected and measured during all 14 109 experiments. Following an ANOVA analysis, it was determined that PAM treatment had no significant results on the amount of runoff generated. An initial turbidity reading was measured from each of runoff sample collected every minute. Samples were stirred to reflect a condition of runoff as it immediately left a test plot. This produced a total of 1,680 measurements for initial turbidity. When this data was observed, dry PAM at 35 lbs/acre performed the best in reducing turbidity, with calculated reduction of 97% from the control. Initial turbidity readings were observed to increase as the PAM treatment?s application rates decreased. Liquid PAM applied at 25 and 15 lbs/acres noticed increases in turbidity during test 4, indicating these treatments were losing their effectiveness. In addition to recording an initial turbidity, samples were collected from each test at 5 and 10 minutes to measure turbidity over time. This provided researchers with data that showed how treatments performed in reducing turbidity over time. All PAM treatments were observed to reduce the time required for turbidity to decrease when compared to the control, which after 10 minutes had turbidity levels recorded around 500 NTUs. As seen with the initial turbidity recordings, liquid PAM 25 and 15 lbs/acre effectiveness over time also began to decline during test 3 and test 4. Soil samples were collected every 3 minutes from surface runoff to determine an amount of eroded soil from the test plots. Runoff was filtered and oven dried for 24 hours to show soil loss from each test plot. A total of 560 samples were collected and weighed to determine the eroded soils. Researchers observed that dry PAM at 35 lbs/acre was most effective at reducing soil loss, by 50%. Liquid PAM treatments were only 110 capable of reducing soil loss during the first test, with PAM in subsequent tests performing similarly to the control. The soil samples which were collected to determine eroded soil were also analyzed to determine particle size distribution. This was compared with previous analyses conducting during the soil classification procedures. Researchers observed that surface runoff contained soil particles that all had similar distributions, which were recorded to be smaller than the stock pile soil, which was recorded earlier. Additionally, no observable difference in particle size distributions existed between PAM treatments when compared to each other. Overall, it was observed that dry PAM at the recommended application rate of 35 lbs/acre performed the best for reducing erosion and sedimentation. This was observed when analyzing recorded turbidity measurements and the weight of eroded soil transported from each test plots by the surface runoff. 111 CHAPTER FIVE CONCLUSIONS AND RECOMMENDATIONS 5 5.1 INTRODUCTION Research presented in this report focused on evaluating the effectiveness of anionic polyacrylamide (PAM) as a temporary erosion control measure. Experiments were established to simulate conditions representative of a typical compacted highway embankments with a fill slope of 3:1. The first task of this research focused on examining the previous experimental procedures established previously by Halverson (2006) and McDonald (2007) and making necessary modifications to improve methods for conducting erosion control experiments. These changes reflected researchers? goal to: (1) design and develop a testing apparatus that will allow for an efficient experimental setup and to achieve reproducible results, (2) design and construct a rainfall simulator that can model realistic rainfall events, and (3) develop a unified testing methodology to obtain reproducible results for the comparison of all possible experiments pertaining to erosion and sediment control best management practices (BMPs). The second component of this research uses the unified test procedures developed to evaluate the effectiveness of anionic PAM as an erosion control measure. Specific objectives established were to use the newly developed testing procedures to: (1) test 112 various application rates of dry and liquid PAM as an erosion control measure on 3:1 slopes and (2) compare experimental results to provide product recommendations for use of PAM on highway construction sites. 5.2 INTERMEDIATE-SCALE METHODS AND PROCEDURES A new facility was constructed at the National Center for Asphalt Technology (NCAT) Test Facility for conducting research on erosion and sediment control BMPs. A set of new intermediate-scale test boxes were constructed to streamline the original process that can enable researchers to setup and conduct experiments faster. Overall size of each test plot was reduced and measured approximately 2 ft (0.61 m) wide, 4 ft (1.2 m) long, and a depth of 2 in (5.08 cm) compacted test soil. Runoff was generated using a newly developed rainfall simulator, designed to address deficiencies identified in previous research. Improvements include: (1) eliminating overlapping spray areas, (2) maintaining a constant water pressure, and (3) developing a more efficient method to shutoff simulated rainfall. Overlapping spray areas were eliminated through designing a one-nozzle rainfall simulator with a uniform spray area. Uniformity was verified using the Christian Uniformity Coefficient and determined that the area consisting of two test plots achieved a uniformity ranging between 84% and 88%. Water pressure within the rainfall simulator was controlled using a pressure regulator, which allowed researchers to simulate different rain events, depending on the internal pressure achieved. Finally, a solenoid valve was added to the design, so water flow in the system could be stopped, preventing any additional water falling on test plots at the conclusion of a test. 113 The rainfall simulator was set-up to simulate a storm event common to Alabama and was divided into four separate ?tests?, which produced approximately 1.10 in. of rainfall, individually. This amount of rainfall coincides with the Alabama Department of Transportation?s (ALDOT) inspection guidelines that state that any erosion and sediment control device shall be inspected following an accumulated amount of rainfall measuring 0.75 in. In between each rain event, 15 minute breaks were observed for data collection. The selected rain regime allowed researches to analyze the long-term effectiveness of PAM using a 2-year, 24 hour storm event that produced total rainfall amount of 4.4 in. as experienced in Montgomery, AL. Additional modifications in experiment preparation included a new method to compact test soil to the required rate of 95%, as established by ALDOT standard specifications. It was determined that using hand-tamps on the test soil, with an OMC of 15%, would achieve the required rate of compaction for the test soil. A turbidity meter was used to record turbidity and could be used during experimentation, rather than collecting samples and measuring turbidity at a later date. The data collected for this research included the mass of surface runoff and particle size distribution for each experiment. Procedures developed for intermediate-scale experiments allowed researchers to conduct a greater number of experiments and produce large quantities of data for analysis. One control and six different PAM treatments (i.e. dry and liquid PAM at 35, 25, and 15 lbs/acre) conditions were examined. Each one of these experiments was replicated to provide the means for evaluating the effectiveness of experimental 114 procedures by examining the results for reproducible data and validating these new experimental procedures. 5.3 DRY AND LIQUID PAM EROSION CONTROL Fourteen experiments were conducted to examine two different PAM treatment methods at three application rates (i.e. 35, 25, and 15 lbs/acre). PAM treated test plots were compared with a bare soil control condition. Data that was collected from these experiments included: (1) surface runoff volume and mass, (2) initial turbidity, (3) turbidity versus time, (4) soil loss, and (5) particle size distribution. An initial turbidity reading was measured from each of runoff sample that was collected every minute. Samples were stirred to reflect a condition of runoff as if it had immediately left a test plot. It was observed that dry PAM at 35 lbs/acre performed the best in reducing turbidity, with a calculated reduction of 97% from the control. Initial turbidity readings were observed to increase as the PAM treatment?s application rates decreased. At a point between test 3 and test 4 (i.e. 40 minutes into an experiment), liquid PAM applied at 25 and 15 lbs/acre began to lose its effectiveness in reducing initial turbidity levels. Additional samples were collected from each test at 5 and 10 minutes to measure turbidity over time. All PAM treatments were observed to reduce the time required for turbidity to decrease when compared to the control, which after 10 minutes had turbidity levels recorded around 500 NTUs. As seen with the initial turbidity recordings, liquid PAM 25 and 15 lbs/acre effectiveness over time also began to wear off during test 3 and test 4. Dry PAM at 35 lbs/acre reached EPA?s proposed effluent turbidity requirement of 115 13 NTU within 20 seconds of measurements. Dry PAM at 25 lbs/acre also reached this limit at 3 minutes, 20 seconds. Runoff was filtered and oven dried for 24 hours to quantify soil loss from each test plot, every 3 minutes. Researchers observed that dry PAM at 35 lbs/acre was effective in reducing soil loss by 50%. Dry PAM at 25 and 15 lbs/acre, on average, reduced soil loss by 28.7% and 27.4%, respectively. Liquid PAM treatments were only capable of reducing soil loss during the first test, with subsequent tests performing poorly with results similar to the control. This indicates that liquid PAM was incapable of performing consistently over a long period. The soil samples collected were analyzed to determine their particle size distribution. Researchers observed that all the surface runoff had similar distributions, which contained smaller particle sizes when compared to the stock pile soil distribution. Additionally, no observable difference in particle size distributions occurred between PAM treatments when compared to each other. The results from this research suggest that dry granular PAM could perform as an effective erosion and sediment control technology, when applied at the recommended application rate. PAM formulations are very site specific and laboratory test must be conducted to determine which formula of PAM will perform the best for a given construction site. These additional tests are usually provided at no additional cost and will provide contractors with the correct type of PAM product for use, with the recommended application rates required. However, PAM is rarely used alone and this research has shown that long-term exposure to rainfall will inhibits the product?s effectiveness over time. By using PAM in conjunction with other erosion and sediment 116 control technologies (i.e. erosion control blankets, mulching, etc) would provide the most ideal protection against NPS pollution. Test plots treated with dry PAM in this research demonstrated that the amount of erosion could be significantly reduced from construction sites with 3:1 compacted file slopes. However, liquid PAM applied did not perform as well when compared to dry PAM. Liquid PAM will only performed effective when the spray coverage is 100% uniform. So the effectiveness of liquid PAM applications is dependent on the quality of applications. Evaluating the turbidity results indicated that both PAM treated test plots were effective in reducing sedimentation in the runoff, with dry PAM at the recommended rate performing the best. This was attributed to how dry PAM granules perform over long periods of time and the way PAM is introduced slowly and consistently into the runoff. 5.4 RECOMMENDED FUTURE RESEARCH Results presented in this report show that PAM can perform as an effective means in reducing erosion and sedimentation caused by sediment laden runoff. However, PAM is rarely ever used on its own and is more commonly used in conjunction with additional erosion and sediment control BMPs. Therefore, further research should be conducted to examine how the addition of PAM could potential improve existing technologies, such as erosion control blankets (ECBs). Also, the work conducted during this research effort represents intermediate-scale test plots. It would be beneficial if the performance of PAM was documented under field-scale conditions to validate the intermediate-scale procedures discussed. Also, 117 field-scale experiments could provide additional results to develop recommendations for actual practice. Large-scale test plots could also be used to further the research conducted on ECBs, with and without the addition of PAM. With advantages of conducting experiments at both scales (i.e. intermediate and field), any erosion and sediment control technology could be analyzed thoroughly and performance recommendations could be made for use in practice. 118 REFERENCES 1. Alabama Department of Transportation (ALDOT). Standard Specifications for Highway Construction. Montgomery: Alabama, 2002 2. Alabama Department of Transportation (ALDOT). Stormwater Inspection Guidelines. Montgomery: Alabama, 2004 3. Alabama State. Alabama Soil and Water Conservation Committee (ASWCC). Alabama Handbook for Erosion Control, Sediment Control, and Stormwater Management on Construction Site and Urban Areas: Volume 1. Montgomery: Alabama, 2003. 4. Alabama State. Alabama Soil and Water Conservation Committee (ASWCC). Alabama Handbook for Erosion Control, Sediment Control, and Stormwater Management on Construction Site and Urban Areas: Volume 2. Montgomery: Alabama, 2003. 5. American Society of Agricultural Engineers (ASAE). (2000). Testing procedure for determining uniformity of water distribution of center pivot and lateral move irrigation machines equipped with spray or sprinkler nozzles. ASCE-S436.1, 48h Ed., St. Joseph, Mich. 6. Applied Polymer Systems, Inc., University of Central Florida, Stormwater Management Academy (2006). Polymer Enhanced Best Management Practice (PEBMP) Application Guide. 7. Benik, S. R., B.N. Wilson, et al. (2003) Performance of erosion control products on a highway embankment. Transactions of the ASAE, 46(4): 1113?1119. 8. Bjorneberg, D. L., J. K. Aase, et al. (2000). Controlling sprinkler irrigation runoff, erosion, and phosphorus loss with straw and polyacrylamide. Transactions of the ASAE, 43(6): 1545-1551. 9. Colman, A. M., (2001). Duncan?s multiple range test. A Dictionary of Psychology. Encyclopedia.com. Date Accessed: 8 Feb. 2009 . 10. Flanagan, D. C., K. Chaudhari, et al. (2002a). Polyacrylamide soil amendment effects on runoff and sediment yield on steep slopes: Part I. Simulated rainfall conditions. Transactions of the ASAE, 45(5): 1327-1337. 119 11. Flanagan, D. C., K. Chaudhari, et al. (2002b). Polyacrylamide soil amendment effects on runoff and sediment yield on steep slopes: Part II. Natural rainfall conditions. Transactions of the ASAE, 45(5): 1339-1351. 12. Flanagan, D. C., L. D. Norton, et al. (2003). Using polyacrylamide to control erosion on agricultural and disturbed soils in rainfed areas. Journal of Soil and Water Conservation, 58(5): 301-311. 13. Flanagan, D. C., L. D. Norton, et al. (1997a). Effect of water chemistry and soil amendments on a silt loam soil - Part 1: Infiltration and runoff. Transactions of the ASAE, 40(6): 1549-1554. 14. Flanagan, D. C., L. D. Norton, et al. (1997b). Effect of water chemistry and soil amendments on a silt loam soil - Part 2: Soil erosion. Transactions of the ASAE, 40(6): 1555-1561. 15. Flanagan, D. C., L. D. Norton, et al. (1997b). Effect of water chemistry and soil amendments on a silt loam soil - Part 2: Soil erosion. Transactions of the ASAE, 40(6): 1555-1561. 16. Hall, M. J. (1970). A Critique of Methods of Simulating Rainfall. Water Resources Research, 6(4): 1104-1114. 17. Halverson, J.L. (2006). Use of a Small-Scale Erosion Control Model in Determining the Design and Effectiveness of Silt Fence Tiebacks Along Highway Projects. MS Thesis, Auburn University. 18. Hayes, S. A., R. A. McLaughlin, et al. (2005). Polyacrylamide use for erosion and turbidity control on construction sites. Journal of Soil and Water Conservation, 60(4): 193-199. 19. Lentz, R. D., R. E. Sojka, et al. (2002). Fate and efficacy of polyacrylamide applied in furrow irrigation: Full-advance and continuous treatments. Journal of Environmental Quality, 31(2): 661-670. 20. Leib, B. G., C. A. Redulla, et al. (2005). Erosion Control Practices Integrated with Polyacrylamide to Reduce Sediment Loss in Furrow Irrigation. Applied Engineering in Agriculture, 21(4): 595-603. 21. McDonald, J. S. (2007). Evaluation of Erosion and Sediment Control Best Management Practices: Use of Silt Fence Tieback Systems and Anionic Polyacrylamide on Highway Construction Sites. MS Thesis, Auburn University. 120 22. McLaughlin, R. A. and T. T. Brown (2006). Evaluation of erosion control products with and without added polyacrylamide. Journal of the American Water Resources Association, 42(3): 675-684. 23. Miller, M. P. (1987). A solenoid-operated, variable intensity rainfall simulator. Soil Science Society of America Journal, 51: 832-834. 24. Novotny, Vladimir. Water Quality: Diffuse Pollution and Watershed Management. New York: John Wiley & Sons, Inc., 2003. 25. Petersen, A. L., A. M. Thompson, et al. (2007). A New Polyacrylamide (PAM) Formulation for Reducing Erosion and Phosphorus Loss in Rainfall Agriculture. Transactions of the ASABE, 50(6): 2091-2101. 26. Peterson, J. R., D. C. Flanagan, et al. (2002). PAM application method and electrolyte source effects on plot-scale runoff and erosion. Transactions of the ASAE, 45(6): 1859-1867. 27. Ramsey, F. and D. Schafer (2002). The Statistical Sleuth: A Course in Methods of Data Analysis. New York: Duxbury Press. 28. Regmi, T. P. and A. L. Thompson. (2000). Rainfall Simulator for Laboratory Studies. Applied Engineering in Agriculture, 16(6): 641-647. 29. Roa-Espinosa, A., G.D. Bubuenzer, and E.S. Miyashita, 1999. Sediment and Runoff Control on Construction Sites Using Four Application Methods of Polyacrylamide Mix. American Society of Agricultural Engineers Annual Meeting Paper No. 99- 2013, ASAE, St. Joseph, Michigan. 30. Sepaskhah, A. R. and A. R. Bazrafshan-Jahromi. (2006). Controlling Runoff and Erosion in Sloping Land with Polyacrylamide under a Rainfall Simulator. Biosystems Engineering, 93(4): 496-474. 31. United States Department of Commerce (U.S.D.C.). Technical Paper #25: Rainfall Intensity-Duration-Frequency Curves. Washington D.C., 1955 32. United States Department of Commerce (U.S.D.C.). Technical Paper #40: Rainfall Frequency Atlas of the United States. Washington D.C., 1961 33. United States Environmental Protection Agency (U.S. EPA). Nonpoint Source Program and Grants Guidelines for States and Territories. EPA Federal Register, 68(205): 60653-60674. Washington D.C., 2003. 121 34. United States Environmental Protection Agency (U.S. EPA). National Pollutant Discharge Elimination System (NPDES): Chemical Stabilization. Washington D.C., 2006. Date Accessed: 9 Sept. 2008. . 35. United States Environmental Protection Agency (U.S. EPA). National Pollutant Discharge Elimination System (NPDES): Stormwater Discharges from Construction Activities. Washington D.C., 2008a. Date Accessed: 9 Sept. 2008. . 36. United States Environmental Protection Agency (U.S. EPA). Effluent Limitations Guidelines and Standards for the Construction and Development Point Source Category. EPA Federal Register, 73(230): 72561-72614. Washington: D.C., 2008b 122 APPENDICES Appendix A: Manufacturer?s Specifications for Rainfall Simulator Components Appendix B: Rainfall Intensity-Duration-Frequency Curves for Alabama. Appendix C: Manufacturer?s Specifications for Equipment used during Experimentation Appendix D: Experimental Results Appendix E: ANOVA Tables Appendix F: Further Research on Liquid PAM 123 APPENDIX A MANUFACTURER?S SPECIFICATIONS FOR RAINFALL SIMULATOR COMPONENTS 124 Pressure Regulator 125 Solenoid Valve (a) 126 Solenoid Valve (b) 127 Rainfall Simulator Nozzle 12 8 Raindrop Sizes 129 APPENDIX B RAINFALL INTENSITY-DURATION-FREQUENCY CURVES FOR ALABAMA. 130 IDF Curves for Alabama 131 (a) 2-yr, 24-hr Cumulative rainfall for the United States (b) 2-yr, 24-hr Cumulative rainfall for Alabama 132 APPENDIX C MANUFACTURER?S SPECIFICATIONS FOR EQUIPMENT USED DURING EXPERIMENTATION 133 Backpack Sprayer 134 Turbidity Meter (a) 135 Turbidity Meter (b) 136 APPENDIX D EXPERIMENTAL RESULTS 137 Experiment: Control (1) Date: 9/28/08 Treatment: Bare Soil Exp #: 1 Left Test Plots Right Test Plots (a) Initial (b) 15 min (c) Initial (d) 15 min (e) 30 min (f) 45 min (g) 30 min (h) 45 min (i) 60 min (j) 60 min Photographs Documenting Test Plot Conditions during Experimentation. 138 Experiment: Control (1) Date: 9/28/08 Treatment: Bare Soil Exp #: 1 0 0.5 1 1.5 2 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt ) Time (min) Left Plot Right Plot 0 5 10 15 20 25 30 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt ) Time (min) Left Plot Right Plot (a) Runoff Volume (b) Cumulative Runoff Volume 0 250 500 750 1000 1250 1500 1750 2000 0 5 10 15 20 25 30 35 40 45 50 55 60 M as s ( g) Time (min) Left Plot Right Plot 0 5000 10000 15000 20000 25000 30000 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive M as s ( g) Time (min) Left Plot Right Plot (c) Runoff Mass (d) Cumulative Runoff Mass Surface Runoff Measurements vs. Time. 0 1000 2000 3000 4000 5000 0 5 10 15 20 25 30 35 40 45 50 55 60 Tu rb idi ty (N TU ) Time (min) Left Plot Right Plot Surface Runoff Initial Turbidity vs. Time. 139 Experiment: Control (1) Date: 9/28/08 Treatment: Bare Soil Exp #: 1 0 500 1000 1500 2000 2500 3000 3500 4000 0 1 2 3 4 5 6 7 8 9 10 Tu rb idi ty (N TU ) Time (min) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right 0 500 1000 1500 2000 2500 3000 3500 4000 0 1 2 3 4 5 6 7 8 9 10 Tu rb idi ty (N TU ) Time (min) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right (a) Turbidity vs. Time at 5 Minutes (b) Turbidity vs. Time at 10 Minutes Turbidity vs. Time for Samples Collected at 5 and 10 Minutes. 0 50 100 150 200 250 300 350 400 0 5 10 15 20 25 30 35 40 45 50 55 60 So il L os s ( g) Time (min) Left Plot Right Plot 0 100 200 300 400 500 600 700 800 900 1000 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive So il L oss (g ) Time (min) Left Plot Right Plot (a) Soil Loss vs. Time (b) Cumulative Soil Loss vs. Time Soil Loss and Cumulative Soil Loss vs. Time. 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 0.0010 0.0100 0.1000 1.0000 Pe rce nt vP as sed (% ) Particle Diameter (in) Original Left - TR1 Left - TR2 Left - TR3 Left - TR4 Right - TR1 Right - TR2 Right - TR3 Right - TR4 Surface Runoff Sediment Particle Distributions. 140 Experiment: Control (2) Date: 10/03/08 Treatment: Bare Soil Exp #: 2 Left Test Plots Right Test Plots (a) Initial (b) 15 min (c) Initial (d) 15 min (e) 30 min (f) 45 min (g) 30 min (h) 45 min (i) 60 min (j) 60 min Photographs Documenting Test Plot Conditions during Experimentation Photograph Not Available Photograph Not Available 141 Experiment: Control (2) Date: 10/03/08 Treatment: Bare Soil Exp #: 2 0 0.5 1 1.5 2 2.5 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt ) Time (min) Left Plot Right Plot 0 5 10 15 20 25 30 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt ) Time (min) Left Plot Right Plot (a) Runoff Volume (b) Cumulative Runoff Volume 0 250 500 750 1000 1250 1500 1750 2000 0 5 10 15 20 25 30 35 40 45 50 55 60 M as s ( g) Time (min) Left Plot Right Plot 0 5000 10000 15000 20000 25000 30000 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive M as s ( g) Time (min) Left Plot Right Plot (c) Runoff Mass (d) Cumulative Runoff Mass Surface Runoff Measurements vs. Time. 0 1000 2000 3000 4000 5000 0 5 10 15 20 25 30 35 40 45 50 55 60 Tu rb itit y ( NT U) Time (min) Left Plot Right Plot Surface Runoff Initial Turbidity vs. Time. 142 Experiment: Control (2) Date: 10/03/08 Treatment: Bare Soil Exp #: 2 0 500 1000 1500 2000 2500 3000 3500 4000 0 1 2 3 4 5 6 7 8 9 10 Tu rb idi ty (N TU ) Time (min) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right 0 500 1000 1500 2000 2500 3000 3500 4000 0 1 2 3 4 5 6 7 8 9 10 Tu rb id ity (N TU ) Time (min) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right (a) Turbidity vs. Time at 5 Minutes (b) Turbidity vs. Time at 10 Minutes Turbidity vs. Time for Samples Collected at 5 and 10 Minutes. 0 50 100 150 200 250 0 5 10 15 20 25 30 35 40 45 50 55 60 So il L os s ( g) Time (min) Left Plot Right Plot 0 100 200 300 400 500 600 700 800 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive So il L os s ( g) Time (min) Left Plot Right Plot (a) Soil Loss vs. Time (b) Cumulative Soil Loss vs. Time Soil Loss and Cumulative Soil Loss vs. Time. 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 0.0010 0.0100 0.1000 1.0000 Pe rce nt vP as sed (% ) Particle Diameter (in) Original Left - TR1 Left - TR2 Left - TR3 Left - TR4 Right - TR1 Right - TR2 Right - TR3 Right - TR4 Surface Runoff Sediment Particle Distributions. 143 Experiment: Dry PAM 35 (1) Date: 10/13/08 Treatment: 35 lbs/acre Exp #: 3 Left Test Plots Right Test Plots (a) Initial (b) 15 min (c) Initial (d) 15 min (e) 30 min (f) 45 min (g) 30 min (h) 45 min (i) 60 min (j) 60 min Photographs Documenting Test Plot Conditions during Experimentation. 144 Experiment: Dry PAM 35 (1) Date: 10/13/08 Treatment: 35 lbs/acre Exp #: 3 0 0.5 1 1.5 2 2.5 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt ) Time (min) Left Plot Right Plot 0 5 10 15 20 25 30 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt) Time (min) Left Plot Right Plot (a) Runoff Volume (b) Cumulative Runoff Volume 0 250 500 750 1000 1250 1500 1750 2000 0 5 10 15 20 25 30 35 40 45 50 55 60 M ass (g ) Time (min) Left Plot Right Plot 0 5000 10000 15000 20000 25000 30000 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive M as s ( g) Time (min) Left Plot Right Plot (c) Runoff Mass (d) Cumulative Runoff Mass Surface Runoff Measurements vs. Time. 0 50 100 150 200 250 0 5 10 15 20 25 30 35 40 45 50 55 60 Tu rb itit y ( NT U) Time (min) Left Plot Right Plot Surface Runoff Initial Turbidity vs. Time. 145 Experiment: Dry PAM 35 (1) Date: 10/13/08 Treatment: 35 lbs/acre Exp #: 3 0 50 100 150 200 250 300 0 50 100 150 200 Tu rb idi ty (N TU ) Time (sec) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right 0 50 100 150 200 250 300 0 50 100 150 200 Tu rb idi ty (N TU ) Time (sec) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right (a) Turbidity vs. Time at 5 Minutes (b) Turbidity vs. Time at 10 Minutes Turbidity vs. Time for Samples Collected at 5 and 10 Minutes. 0 50 100 150 200 250 0 5 10 15 20 25 30 35 40 45 50 55 60 So il L oss (g ) Time (min) Left Plot Right Plot 0 100 200 300 400 500 600 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive So il L oss (g ) Time (min) Left Plot Right Plot (a) Soil Loss vs. Time (b) Cumulative Soil Loss vs. Time Soil Loss and Cumulative Soil Loss vs. Time. 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 0.0010 0.0100 0.1000 1.0000 Pe rce nt vP as sed (% ) Particle Diameter (in) Original Left - TR1 Left - TR2 Left - TR3 Left - TR4 Right - TR1 Right - TR2 Right - TR3 Right - TR4 Surface Runoff Sediment Particle Distributions. 146 Experiment: Dry PAM 35 (2) Date: 10/17/08 Treatment: 35 lbs/acre Exp #: 4 Left Test Plots Right Test Plots (a) Initial (b) 15 min (c) Initial (d) 15 min (e) 30 min (f) 45 min (g) 30 min (h) 45 min (i) 60 min (j) 60 min Photographs Documenting Test Plot Conditions during Experimentation. 147 Experiment: Dry PAM 35 (2) Date: 10/17/08 Treatment: 35 lbs/acre Exp #: 4 0 0.5 1 1.5 2 2.5 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt) Time (min) Left Plot Right Plot 0 5 10 15 20 25 30 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt) Time (min) Left Plot Right Plot (a) Runoff Volume (b) Cumulative Runoff Volume 0 250 500 750 1000 1250 1500 1750 2000 0 5 10 15 20 25 30 35 40 45 50 55 60 M ass (g ) Time (min) Left Plot Right Plot 0 5000 10000 15000 20000 25000 30000 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive M ass (g ) Time (min) Left Plot Right Plot (c) Runoff Mass (d) Cumulative Runoff Mass Surface Runoff Measurements vs. Time. 0 20 40 60 80 100 120 140 160 0 5 10 15 20 25 30 35 40 45 50 55 60 Tu rb itit y ( NT U) Time (min) Left Plot Right Plot Surface Runoff Initial Turbidity vs. Time. 148 Experiment: Dry PAM 35 (2) Date: 10/17/08 Treatment: 35 lbs/acre Exp #: 4 0 20 40 60 80 100 120 140 0 50 100 150 200 Tu rb idi ty (N TU ) Time (sec) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right 0 20 40 60 80 100 120 140 0 50 100 150 200 Tu rb idi ty (N TU ) Time (sec) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right (a) Turbidity vs. Time at 5 Minutes (b) Turbidity vs. Time at 10 Minutes Turbidity vs. Time for Samples Collected at 5 and 10 Minutes. 0 50 100 150 200 250 0 5 10 15 20 25 30 35 40 45 50 55 60 So il L oss (g ) Time (min) Left Plot Right Plot 0 100 200 300 400 500 600 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive So il L oss (g ) Time (min) Left Plot Right Plot (a) Soil Loss vs. Time (b) Cumulative Soil Loss vs. Time Soil Loss and Cumulative Soil Loss vs. Time. 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 0.0010 0.0100 0.1000 1.0000 Pe rce nt vP as sed (% ) Particle Diameter (in) Original Left - TR1 Left - TR2 Left - TR3 Left - TR4 Right - TR1 Right - TR2 Right - TR3 Right - TR4 Surface Runoff Sediment Particle Distributions. 149 Experiment: Dry PAM 15 (1) Date: 10/20/08 Treatment: 15 lbs/acre Exp #: 5 Left Test Plots Right Test Plots (a) Initial (b) 15 min (c) Initial (d) 15 min (e) 30 min (f) 45 min (g) 30 min (h) 45 min (i) 60 min (j) 60 min Photographs Documenting Test Plot Conditions during Experimentation. 150 Experiment: Dry PAM 15 (1) Date: 10/20/08 Treatment: 15 lbs/acre Exp #: 5 0 0.5 1 1.5 2 2.5 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt) Time (min) Left Plot Right Plot 0 5 10 15 20 25 30 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt) Time (min) Left Plot Right Plot (a) Runoff Volume (b) Cumulative Runoff Volume 0 250 500 750 1000 1250 1500 1750 2000 0 5 10 15 20 25 30 35 40 45 50 55 60 M ass (g ) Time (min) Left Plot Right Plot 0 5000 10000 15000 20000 25000 30000 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive M ass (g ) Time (min) Left Plot Right Plot (c) Runoff Mass (d) Cumulative Runoff Mass Surface Runoff Measurements vs. Time 0 200 400 600 800 1000 1200 1400 1600 0 5 10 15 20 25 30 35 40 45 50 55 60 Tu rb itit y ( NT U) Time (min) Left Plot Right Plot Surface Runoff Initial Turbidity vs. Time. 151 Experiment: Dry PAM 15 (1) Date: 10/20/08 Treatment: 15 lbs/acre Exp #: 5 0 200 400 600 800 1000 1200 1400 0 50 100 150 200 Tu rb idi ty (N TU ) Time (sec) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right 0 200 400 600 800 1000 1200 1400 0 50 100 150 200 Tu rb idi ty (N TU ) Time (sec) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right (a) Turbidity vs. Time at 5 Minutes (b) Turbidity vs. Time at 10 Minutes Turbidity vs. Time for Samples Collected at 5 and 10 Minutes. 0 50 100 150 200 250 0 5 10 15 20 25 30 35 40 45 50 55 60 So il L oss (g ) Time (min) Left Plot Right Plot 0 100 200 300 400 500 600 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive So il L oss (g ) Time (min) Left Plot Right Plot (a) Soil Loss vs. Time (b) Cumulative Soil Loss vs. Time Soil Loss and Cumulative Soil Loss vs. Time. 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 0.0010 0.0100 0.1000 1.0000 Pe rce nt vP as sed (% ) Particle Diameter (in) Original Left - TR1 Left - TR2 Left - TR3 Left - TR4 Right - TR1 Right - TR2 Right - TR3 Right - TR4 Surface Runoff Sediment Particle Distributions. 152 Experiment: Dry PAM 15 (2) Date: 10/22/08 Treatment: 15 lbs/acre Exp #: 6 Left Test Plots Right Test Plots (a) Initial (b) 15 min (c) Initial (d) 15 min (e) 30 min (f) 45 min (g) 30 min (h) 45 min (i) 60 min (j) 60 min Photographs Documenting Test Plot Conditions during Experimentation. Photograph Not Available Photograph Not Available 153 Experiment: Dry PAM 15 (2) Date: 10/22/08 Treatment: 15 lbs/acre Exp #: 6 0 0.5 1 1.5 2 2.5 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt) Time (min) Left Plot Right Plot 0 5 10 15 20 25 30 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt) Time (min) Left Plot Right Plot (a) Runoff Volume (b) Cumulative Runoff Volume 0 250 500 750 1000 1250 1500 1750 2000 0 5 10 15 20 25 30 35 40 45 50 55 60 M ass (g ) Time (min) Left Plot Right Plot 0 5000 10000 15000 20000 25000 30000 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive M ass (g ) Time (min) Left Plot Right Plot (c) Runoff Mass (d) Cumulative Runoff Mass Surface Runoff Measurements vs. Time. 0 200 400 600 800 1000 1200 1400 0 5 10 15 20 25 30 35 40 45 50 55 60 Tu rb itit y ( NT U) Time (min) Left Plot Right Plot Surface Runoff Initial Turbidity vs. Time. 154 Experiment: Dry PAM 15 (2) Date: 10/22/08 Treatment: 15 lbs/acre Exp #: 6 0 200 400 600 800 1000 1200 1400 0 50 100 150 200 Tu rb idi ty (N TU ) Time (sec) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right 0 200 400 600 800 1000 1200 1400 1600 0 50 100 150 200 Tu rb idi ty (N TU ) Time (sec) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right (a) Turbidity vs. Time at 5 Minutes (b) Turbidity vs. Time at 10 Minutes Turbidity vs. Time for Samples Collected at 5 and 10 Minutes. 0 50 100 150 200 250 0 5 10 15 20 25 30 35 40 45 50 55 60 So il L os s ( g) Time (min) Left Plot Right Plot 0 100 200 300 400 500 600 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive So il L oss (g ) Time (min) Left Plot Right Plot (a) Soil Loss vs. Time (b) Cumulative Soil Loss vs. Time Soil Loss and Cumulative Soil Loss vs. Time. 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 0.0010 0.0100 0.1000 1.0000 Pe rce nt vP as sed (% ) Particle Diameter (in) Original Left - TR1 Left - TR2 Left - TR3 Left - TR4 Right - TR1 Right - TR2 Right - TR3 Right - TR4 Surface Runoff Sediment Particle Distributions. 155 Experiment: Dry PAM 25 (1) Date: 10/27/08 Treatment: 25 lbs/acre Exp #: 7 Left Test Plots Right Test Plots (a) Initial (b) 15 min (c) Initial (d) 15 min (e) 30 min (f) 45 min (g) 30 min (h) 45 min (i) 60 min (j) 60 min Photographs Documenting Test Plot Conditions during Experimentation. 156 Experiment: Dry PAM 25 (1) Date: 10/27/08 Treatment: 25 lbs/acre Exp #: 7 0 0.5 1 1.5 2 2.5 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt) Time (min) Left Plot Right Plot 0 5 10 15 20 25 30 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt) Time (min) Left Plot Right Plot (a) Runoff Volume (b) Cumulative Runoff Volume 0 250 500 750 1000 1250 1500 1750 2000 0 5 10 15 20 25 30 35 40 45 50 55 60 M as s ( g) Time (min) Left Plot Right Plot 0 5000 10000 15000 20000 25000 30000 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive M as s ( g) Time (min) Left Plot Right Plot (c) Runoff Mass (d) Cumulative Runoff Mass Surface Runoff Measurements vs. Time. 0 100 200 300 400 500 600 700 800 0 5 10 15 20 25 30 35 40 45 50 55 60 Tu rb itit y ( NT U) Time (min) Left Plot Right Plot Surface Runoff Initial Turbidity vs. Time. 157 Experiment: Dry PAM 25 (1) Date: 10/27/08 Treatment: 25 lbs/acre Exp #: 7 0 100 200 300 400 500 600 700 0 50 100 150 200 Tu rb idi ty (N TU ) Time (sec) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right 0 100 200 300 400 500 600 700 0 50 100 150 200 Tu rb idi ty (N TU ) Time (sec) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right (a) Turbidity vs. Time at 5 Minutes (b) Turbidity vs. Time at 10 Minutes Turbidity vs. Time for Samples Collected at 5 and 10 Minutes. 0 50 100 150 200 250 0 5 10 15 20 25 30 35 40 45 50 55 60 So il L os s ( g) Time (min) Left Plot Right Plot 0 100 200 300 400 500 600 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive So il L oss (g ) Time (min) Left Plot Right Plot (a) Soil Loss vs. Time (b) Cumulative Soil Loss vs. Time Soil Loss and Cumulative Soil Loss vs. Time. 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 0.0010 0.0100 0.1000 1.0000 Pe rce nt vP as sed (% ) Particle Diameter (in) Original Left - TR1 Left - TR2 Left - TR3 Left - TR4 Right - TR1 Right - TR2 Right - TR3 Right - TR4 Surface Runoff Sediment Particle Distributions. 158 Experiment: Dry PAM 25 (2) Date: 10/29/08 Treatment: 25 lbs/acre Exp #: 8 Left Test Plots Right Test Plots (a) Initial (b) 15 min (c) Initial (d) 15 min (e) 30 min (f) 45 min (g) 30 min (h) 45 min (i) 60 min (j) 60 min Photographs Documenting Test Plot Conditions during Experimentation 159 Experiment: Dry PAM 25 (2) Date: 10/29/08 Treatment: 25 lbs/acre Exp #: 8 0 0.5 1 1.5 2 2.5 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt) Time (min) Left Plot Right Plot 0 5 10 15 20 25 30 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt) Time (min) Left Plot Right Plot (a) Runoff Volume (b) Cumulative Runoff Volume 0 250 500 750 1000 1250 1500 1750 2000 0 5 10 15 20 25 30 35 40 45 50 55 60 M ass (g ) Time (min) Left Plot Right Plot 0 5000 10000 15000 20000 25000 30000 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive M ass (g ) Time (min) Left Plot Right Plot (c) Runoff Mass (d) Cumulative Runoff Mass Surface Runoff Measurements vs. Time. 0 200 400 600 800 1000 1200 0 5 10 15 20 25 30 35 40 45 50 55 60 Tu rb itit y ( NT U) Time (min) Left Plot Right Plot Surface Runoff Initial Turbidity vs. Time. 160 Experiment: Dry PAM 25 (2) Date: 10/29/08 Treatment: 25 lbs/acre Exp #: 8 0 100 200 300 400 500 600 700 800 0 50 100 150 200 Tu rb idi ty (N TU ) Time (sec) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right 0 100 200 300 400 500 600 700 800 900 0 50 100 150 200 Tu rb idi ty (N TU ) Time (sec) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right (a) Turbidity vs. Time at 5 Minutes (b) Turbidity vs. Time at 10 Minutes Turbidity vs. Time for Samples Collected at 5 and 10 Minutes. 0 50 100 150 200 250 0 5 10 15 20 25 30 35 40 45 50 55 60 So il L oss (g ) Time (min) Left Plot Right Plot 0 100 200 300 400 500 600 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive So il L oss (g ) Time (min) Left Plot Right Plot (a) Soil Loss vs. Time (b) Cumulative Soil Loss vs. Time Soil Loss and Cumulative Soil Loss vs. Time. 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 0.0010 0.0100 0.1000 1.0000 Pe rce nt vP as sed (% ) Particle Diameter (in) Original Left - TR1 Left - TR2 Left - TR3 Left - TR4 Right - TR1 Right - TR2 Right - TR3 Right - TR4 Surface Runoff Sediment Particle Distributions. 161 Experiment: Liquid PAM 35 (1) Date: 11/7/08 Treatment: 35 lbs/acre Exp #: 9 Left Test Plots Right Test Plots (a) Initial (b) 15 min (c) Initial (d) 15 min (e) 30 min (f) 45 min (g) 30 min (h) 45 min (i) 60 min (j) 60 min Photographs Documenting Test Plot Conditions during Experimentation. 162 Experiment: Liquid PAM 35 (1) Date: 11/7/08 Treatment: 35 lbs/acre Exp #: 9 0 0.5 1 1.5 2 2.5 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt) Time (min) Left Plot Right Plot 0 5 10 15 20 25 30 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt) Time (min) Left Plot Right Plot (a) Runoff Volume (b) Cumulative Runoff Volume 0 250 500 750 1000 1250 1500 1750 2000 0 5 10 15 20 25 30 35 40 45 50 55 60 M ass (g ) Time (min) Left Plot Right Plot 0 5000 10000 15000 20000 25000 30000 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive M ass (g ) Time (min) Left Plot Right Plot (c) Runoff Mass (d) Cumulative Runoff Mass Surface Runoff Measurements vs. Time. 0 200 400 600 800 1000 1200 0 5 10 15 20 25 30 35 40 45 50 55 60 Tu rb itit y ( NT U) Time (min) Left Plot Right Plot Surface Runoff Initial Turbidity vs. Time. 163 Experiment: Liquid PAM 35 (1) Date: 11/7/08 Treatment: 35 lbs/acre Exp #: 9 0 200 400 600 800 1000 1200 1400 0 50 100 150 200 Tu rb idi ty (N TU ) Time (sec) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right 0 200 400 600 800 1000 1200 0 50 100 150 200 Tu rb idi ty (N TU ) Time (sec) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right (a) Turbidity vs. Time at 5 Minutes (b) Turbidity vs. Time at 10 Minutes Turbidity vs. Time for Samples Collected at 5 and 10 Minutes. 0 50 100 150 200 250 0 5 10 15 20 25 30 35 40 45 50 55 60 So il L oss (g ) Time (min) Left Plot Right Plot 0 100 200 300 400 500 600 700 800 900 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive So il L oss (g ) Time (min) Left Plot Right Plot (a) Soil Loss vs. Time (b) Cumulative Soil Loss vs. Time Soil Loss and Cumulative Soil Loss vs. Time. 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 0.0010 0.0100 0.1000 1.0000 Pe rce nt vP as sed (% ) Particle Diameter (in) Original Left - TR1 Left - TR2 Left - TR3 Left - TR4 Right - TR1 Right - TR2 Right - TR3 Right - TR4 Surface Runoff Sediment Particle Distributions. 164 Experiment: Liquid PAM 35 (2) Date: 11/10/08 Treatment: 35 lbs/acre Exp #: 10 Left Test Plots Right Test Plots (a) Initial (b) 15 min (c) Initial (d) 15 min (e) 30 min (f) 45 min (g) 30 min (h) 45 min (i) 60 min (j) 60 min Photographs Documenting Test Plot Conditions during Experimentation. 165 Experiment: Liquid PAM 35 (2) Date: 11/10/08 Treatment: 35 lbs/acre Exp #: 10 0 0.5 1 1.5 2 2.5 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt) Time (min) Left Plot Right Plot 0 5 10 15 20 25 30 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt) Time (min) Left Plot Right Plot (a) Runoff Volume (b) Cumulative Runoff Volume 0 250 500 750 1000 1250 1500 1750 2000 0 5 10 15 20 25 30 35 40 45 50 55 60 M ass (g ) Time (min) Left Plot Right Plot 0 5000 10000 15000 20000 25000 30000 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive M ass (g ) Time (min) Left Plot Right Plot (c) Runoff Mass (d) Cumulative Runoff Mass Surface Runoff Measurements vs. Time. 0 200 400 600 800 1000 1200 0 5 10 15 20 25 30 35 40 45 50 55 60 Tu rb itit y ( NT U) Time (min) Left Plot Right Plot Surface Runoff Initial Turbidity vs. Time. 166 Experiment: Liquid PAM 35 (2) Date: 11/10/08 Treatment: 35 lbs/acre Exp #: 10 0 200 400 600 800 1000 1200 1400 0 50 100 150 200 Tu rb idi ty (N TU ) Time (sec) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right 0 200 400 600 800 1000 1200 0 50 100 150 200 Tu rb idi ty (N TU ) Time (sec) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right (a) Turbidity vs. Time at 5 Minutes (b) Turbidity vs. Time at 10 Minutes Turbidity vs. Time for Samples Collected at 5 and 10 Minutes. 0 50 100 150 200 250 0 5 10 15 20 25 30 35 40 45 50 55 60 So il L oss (g ) Time (min) Left Plot Right Plot 0 100 200 300 400 500 600 700 800 900 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive So il L oss (g ) Time (min) Left Plot Right Plot (a) Soil Loss vs. Time (b) Cumulative Soil Loss vs. Time Soil Loss and Cumulative Soil Loss vs. Time. 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 0.0010 0.0100 0.1000 1.0000 Pe rce ntv Pa sse d ( % ) Particle Diameter (in) Original Left - TR1 Left - TR2 Left - TR3 Left - TR4 Right - TR1 Right - TR2 Right - TR3 Right - TR4 Surface Runoff Sediment Particle Distributions. 167 Experiment: Liquid PAM 15 (1) Date: 11/17/08 Treatment: 15 lbs/acre Exp #: 11 Left Test Plots Right Test Plots (a) Initial (b) 15 min (c) Initial (d) 15 min (e) 30 min (f) 45 min (g) 30 min (h) 45 min (i) 60 min (j) 60 min Photographs Documenting Test Plot Conditions during Experimentation. 168 Experiment: Liquid PAM 15 (1) Date: 11/17/08 Treatment: 15 lbs/acre Exp #: 11 0 0.5 1 1.5 2 2.5 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt) Time (min) Left Plot Right Plot 0 5 10 15 20 25 30 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt) Time (min) Left Plot Right Plot (a) Runoff Volume (b) Cumulative Runoff Volume 0 250 500 750 1000 1250 1500 1750 2000 0 5 10 15 20 25 30 35 40 45 50 55 60 M ass (g ) Time (min) Left Plot Right Plot 0 5000 10000 15000 20000 25000 30000 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive M ass (g ) Time (min) Left Plot Right Plot (c) Runoff Mass (d) Cumulative Runoff Mass Surface Runoff Measurements vs. Time. 0 500 1000 1500 2000 2500 3000 3500 4000 4500 0 5 10 15 20 25 30 35 40 45 50 55 60 Tu rb itit y ( NT U) Time (min) Left Plot Right Plot Surface Runoff Initial Turbidity vs. Time. 169 Experiment: Liquid PAM 15 (1) Date: 11/17/08 Treatment: 15 lbs/acre Exp #: 11 0 500 1000 1500 2000 2500 3000 3500 4000 4500 0 50 100 150 200 Tu rb idi ty (N TU ) Time (sec) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right 0 500 1000 1500 2000 2500 3000 3500 4000 4500 0 50 100 150 200 Tu rb idi ty (N TU ) Time (sec) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right (a) Turbidity vs. Time at 5 Minutes (b) Turbidity vs. Time at 10 Minutes Turbidity vs. Time for Samples Collected at 5 and 10 Minutes. 0 50 100 150 200 250 0 5 10 15 20 25 30 35 40 45 50 55 60 So il L oss (g ) Time (min) Left Plot Right Plot 0 100 200 300 400 500 600 700 800 900 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive So il L oss (g ) Time (min) Left Plot Right Plot (a) Soil Loss vs. Time (b) Cumulative Soil Loss vs. Time Soil Loss and Cumulative Soil Loss vs. Time. 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 0.0010 0.0100 0.1000 1.0000 Pe rce nt vP as sed (% ) Particle Diameter (in) Original Left - TR1 Left - TR2 Left - TR3 Left - TR4 Right - TR1 Right - TR2 Right - TR3 Right - TR4 Surface Runoff Sediment Particle Distributions. 170 Experiment: Liquid PAM 15 (2) Date: 11/19/08 Treatment: 15 lbs/acre Exp #: 12 Left Test Plots Right Test Plots (a) Initial (b) 15 min (c) Initial (d) 15 min (e) 30 min (f) 45 min (g) 30 min (h) 45 min (i) 60 min (j) 60 min Photographs Documenting Test Plot Conditions during Experimentation. 171 Experiment: Liquid PAM 15 (2) Date: 11/19/08 Treatment: 15 lbs/acre Exp #: 12 0 0.5 1 1.5 2 2.5 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt) Time (min) Left Plot Right Plot 0 5 10 15 20 25 30 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt) Time (min) Left Plot Right Plot (a) Runoff Volume (b) Cumulative Runoff Volume 0 250 500 750 1000 1250 1500 1750 2000 0 5 10 15 20 25 30 35 40 45 50 55 60 M ass (g ) Time (min) Left Plot Right Plot 0 5000 10000 15000 20000 25000 30000 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive M ass (g ) Time (min) Left Plot Right Plot (c) Runoff Mass (d) Cumulative Runoff Mass Surface Runoff Measurements vs. Time. 0 500 1000 1500 2000 2500 3000 3500 4000 0 5 10 15 20 25 30 35 40 45 50 55 60 Tu rb itit y ( NT U) Time (min) Left Plot Right Plot Surface Runoff Initial Turbidity vs. Time. 172 Experiment: Liquid PAM 15 (2) Date: 11/19/08 Treatment: 15 lbs/acre Exp #: 12 0 500 1000 1500 2000 2500 3000 3500 0 50 100 150 200 Tu rb idi ty (N TU ) Time (sec) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right 0 500 1000 1500 2000 2500 3000 3500 0 50 100 150 200 Tu rb idi ty (N TU ) Time (sec) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right (a) Turbidity vs. Time at 5 Minutes (b) Turbidity vs. Time at 10 Minutes Turbidity vs. Time for Samples Collected at 5 and 10 Minutes. 0 50 100 150 200 250 0 5 10 15 20 25 30 35 40 45 50 55 60 So il Lo ss (g ) Time (min) Left Plot Right Plot 0 100 200 300 400 500 600 700 800 900 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive So il L oss (g ) Time (min) Left Plot Right Plot (a) Soil Loss vs. Time (b) Cumulative Soil Loss vs. Time Soil Loss and Cumulative Soil Loss vs. Time. 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 0.0010 0.0100 0.1000 1.0000 Pe rce nt vP as sed (% ) Particle Diameter (in) Original Left - TR1 Left - TR2 Left - TR3 Left - TR4 Right - TR1 Right - TR2 Right - TR3 Right - TR4 Surface Runoff Sediment Particle Distributions. 173 Experiment: Liquid PAM 25 (1) Date: 12/1/08 Treatment: 25 lbs/acre Exp #: 13 Left Test Plots Right Test Plots (a) Initial (b) 15 min (c) Initial (d) 15 min (e) 30 min (f) 45 min (g) 30 min (h) 45 min (i) 60 min (j) 60 min Photographs Documenting Test Plot Conditions during Experimentation. 174 Experiment: Liquid PAM 25 (1) Date: 12/1/08 Treatment: 25 lbs/acre Exp #: 13 0 0.5 1 1.5 2 2.5 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt) Time (min) Left Plot Right Plot 0 5 10 15 20 25 30 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt) Time (min) Left Plot Right Plot (a) Runoff Volume (b) Cumulative Runoff Volume 0 250 500 750 1000 1250 1500 1750 2000 0 5 10 15 20 25 30 35 40 45 50 55 60 M ass (g ) Time (min) Left Plot Right Plot 0 5000 10000 15000 20000 25000 30000 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive M ass (g ) Time (min) Left Plot Right Plot (c) Runoff Mass (d) Cumulative Runoff Mass Surface Runoff Measurements vs. Time. 0 500 1000 1500 2000 2500 3000 3500 0 5 10 15 20 25 30 35 40 45 50 55 60 Tu rb itit y ( NT U) Time (min) Left Plot Right Plot Surface Runoff Initial Turbidity vs. Time. 175 Experiment: Liquid PAM 25 (1) Date: 12/1/08 Treatment: 25 lbs/acre Exp #: 13 0 500 1000 1500 2000 2500 3000 3500 4000 4500 0 50 100 150 200 Tu rb idi ty (N TU ) Time (sec) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right 0 500 1000 1500 2000 2500 3000 3500 4000 4500 0 50 100 150 200 Tu rb idi ty (N TU ) Time (sec) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right (a) Turbidity vs. Time at 5 Minutes (b) Turbidity vs. Time at 10 Minutes Turbidity vs. Time for Samples Collected at 5 and 10 Minutes. 0 50 100 150 200 250 0 5 10 15 20 25 30 35 40 45 50 55 60 So il L os s ( g) Time (min) Left Plot Right Plot 0 100 200 300 400 500 600 700 800 900 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive So il L oss (g ) Time (min) Left Plot Right Plot (a) Soil Loss vs. Time (b) Cumulative Soil Loss vs. Time Soil Loss and Cumulative Soil Loss vs. Time. 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 0.0010 0.0100 0.1000 1.0000 Pe rce nt vP as sed (% ) Particle Diameter (in) Original Left - TR1 Left - TR2 Left - TR3 Left - TR4 Right - TR1 Right - TR2 Right - TR3 Right - TR4 Surface Runoff Sediment Particle Distributions. 176 Experiment: Liquid PAM 25 (2) Date: 12/5/08 Treatment: 25 lbs/acre Exp #: 14 Left Test Plots Right Test Plots (a) Initial (b) 15 min (c) Initial (d) 15 min (e) 30 min (f) 45 min (g) 30 min (h) 45 min (i) 60 min (j) 60 min Photographs Documenting Test Plot Conditions during Experimentation. Photograph Not Available Photograph Not Available 177 Experiment: Liquid PAM 25 (2) Date: 12/5/08 Treatment: 25 lbs/acre Exp #: 14 0 0.5 1 1.5 2 2.5 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt) Time (min) Left Plot Right Plot 0 5 10 15 20 25 30 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e ( qt) Time (min) Left Plot Right Plot (a) Runoff Volume (b) Cumulative Runoff Volume 0 250 500 750 1000 1250 1500 1750 2000 0 5 10 15 20 25 30 35 40 45 50 55 60 M ass (g ) Time (min) Left Plot Right Plot 0 5000 10000 15000 20000 25000 30000 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive M ass (g ) Time (min) Left Plot Right Plot (c) Runoff Mass (d) Cumulative Runoff Mass Surface Runoff Measurements vs. Time. 0 500 1000 1500 2000 2500 3000 3500 0 5 10 15 20 25 30 35 40 45 50 55 60 Tu rb itit y ( NT U) Time (min) Left Plot Right Plot Surface Runoff Initial Turbidity vs. Time. 178 Experiment: Liquid PAM 25 (2) Date: 12/5/08 Treatment: 25 lbs/acre Exp #: 14 0 500 1000 1500 2000 2500 3000 3500 4000 4500 0 50 100 150 200 Tu rb idi ty (N TU ) Time (sec) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right 0 500 1000 1500 2000 2500 3000 3500 4000 0 50 100 150 200 Tu rb idi ty (N TU ) Time (sec) TR1 - Left TR1 - Right TR2 - Left TR2 - Right TR3 - Left TR3 - Right TR4 - Left TR4 - Right (a) Turbidity vs. Time at 5 Minutes (b) Turbidity vs. Time at 10 Minutes Turbidity vs. Time for Samples Collected at 5 and 10 Minutes. 0 50 100 150 200 250 0 5 10 15 20 25 30 35 40 45 50 55 60 So il L os s ( g) Time (min) Left Plot Right Plot 0 100 200 300 400 500 600 700 800 900 0 5 10 15 20 25 30 35 40 45 50 55 60 Cu mu lat ive So il L oss (g ) Time (min) Left Plot Right Plot (a) Soil Loss vs. Time (b) Cumulative Soil Loss vs. Time Soil Loss and Cumulative Soil Loss vs. Time. 0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0 90.0 100.0 0.0010 0.0100 0.1000 1.0000 Pe rce nt vP as sed (% ) Particle Diameter (in) Original Left - TR1 Left - TR2 Left - TR3 Left - TR4 Right - TR1 Right - TR2 Right - TR3 Right - TR4 Surface Runoff Sediment Particle Distributions. 179 APPENDIX E ANOVA TABLES 180 Analysis of Variance (ANOVA) Tables for Surface Runoff Hypothesis: Ho: ?1 = ?2 = ?3= ?4 = ?5 = ?6 = ?7 Ha: All ?i are not equal Source of Variation SS df MS F P-value Fcrit Hypothesis Test 1 Between Groups 2204123 6 367353.9 12.18 3.43E-10 2.19 Ha Within Groups 2956310 98 30166.42 Total 5160433 104 Test 2 Between Groups 1284069 6 214011.6 6.25 1.37E-05 2.19 Ha Within Groups 3353084 98 34215.14 Total 4637153 104 Test 3 Between Groups 610288.3 6 101714.7 2.76 0.02 2.19 Ha Within Groups 3607099 98 36807.13 Total 4217387 104 Test 4 Between Groups 319003 6 53167.17 1.65 0.14 2.19 Ho Within Groups 3165506 98 32301.08 Total 3484509 104 Where, ?i = Mean for ith group [ e.i. (1) Control, (2) Dry 35, ? ,(7) Liquid 15] SS = Sum of Squares, df = Degrees of Freedom, MS = Mean Square, and F = F-value Accept Null Hypothesis: Fcrit > F Reject Null Hypothesis: Fcrit < F 181 Analysis of Variance (ANOVA) Tables for Initial Turbidity Hypothesis: Ho: ?1 = ?2 = ?3= ?4 = ?5 = ?6 = ?7 Ha: All ?i are not equal Source of Variation SS df MS F P-value Fcrit Hypothesis Test 1 Between Groups 111870479.2 6 18645079.87 1486.55 7.4E-94 2.193 Ha Within Groups 1229167.264 98 12542.52311 Total 113099646.5 104 Test 2 Between Groups 119261622.2 6 19876937.03 2271.97 8.4E-103 2.193 Ha Within Groups 857380.6778 98 8748.782426 Total 120119002.8 104 Test 3 Between Groups 136062642.2 6 22677107.03 2405.26 5.2E-104 2.193 Ha Within Groups 923956.6102 98 9428.128675 Total 136986598.8 104 Test 4 Between Groups 177342235.6 6 29557039.27 1755.24 2.3E-97 2.193 Ha Within Groups 1650254.49 98 16839.33153 Total 178992490.1 104 Where, ?i = Mean for ith group [ e.i. (1) Control, (2) Dry 35, ? ,(7) Liquid 15] SS = Sum of Squares, df = Degrees of Freedom, MS = Mean Square, and F = F-value Accept Null Hypothesis: Fcrit > F Reject Null Hypothesis: Fcrit < F 182 Analysis of Variance (ANOVA) Tables for Soil Loss Hypothesis: Ho: ?1 = ?2 = ?3= ?4 = ?5 = ?6 = ?7 Ha: All ?i are not equal Source of Variation SS df MS F P-value Fcrit Hypothesis Test 1 Between Groups 904331.5 6 150721.9 9.028 1.59E-05 2.445 Ha Within Groups 467480 28 16695.7 Total 1371812 34 Test 2 Between Groups 1617987 6 269664.5 23.23 1.15E-09 2.445 Ha Within Groups 325011.1 28 11607.5 Total 1942998 34 Test 3 Between Groups 2329786 6 388297.6 30 5.89E-11 2.445 Ha Within Groups 362425.9 28 12943.8 Total 2692212 34 Test 4 Between Groups 2161449 6 360241.5 45.65 3.49E-13 2.445 Ha Within Groups 220947.8 28 7891.0 Total 2382397 34 Where, ?i = Mean for ith group [ e.i. (1) Control, (2) Dry 35, ? ,(7) Liquid 15] SS = Sum of Squares, df = Degrees of Freedom, MS = Mean Square, and F = F-value Accept Null Hypothesis: Fcrit > F Reject Null Hypothesis: Fcrit < F 183 APPENDIX F FURTHER RESEARCH ON LIQUID PAM 184 F.1 INTRODUCTION Experiments previously conducted (i.e. dry vs. liquid PAM applications) by researchers focused on a worst case scenario for testing PAM treatments. This was identified by performing experiments and exposing test plots to rainfall soon after the initial application of both dry and liquid PAM. However, as seen in the literature reviewed in this report, many researchers provided time for liquid PAM treatments to dry prior to performing any testing. The drying times allowed varied and ranged in duration between 1 to 10 days. In the research that had allow treatments to dry prior to testing and had also examined the differences in performance between both dry and liquid PAM treatments, reported that liquid PAM was more effective (Roa-Espinosa et al., 1999 and Peterson et al., 2002). While allowing liquid PAM applications to dry may provide better protection by allowing PAM molecules to bond with the soil surface, the time to permit PAM to dry and provide this quality of protection may be unfeasible due to weather constraints. As presented in this report, the effectiveness of both dry and liquid PAM applications were examined and compared to determine which treatments were more effective as an erosion and sediment control measure. It was concluded that in these experiments, dry PAM performed significantly better than liquid PAM applications by improving water quality and providing some measure of erosion control. To further examine these results, researchers conducted ancillary experiments with the goal to 185 determining if by allowing liquid PAM treatments time to dry prior to exposure to a storm event, that a better protection against erosion and sedimentation could be provided. Test plots were set-up using the exact same procedures as outlined in Chapter 3 except that liquid PAM treatments were given 48-hours to dry prior to conducting an experiment (herein referred to as the condition of 48-hr liquid PAM). Two experiments were conducted to provide researchers the same amount of data that were generated in previous conditions examined. The following sections cover the results generated from these additional experiments. F.2 EXPERIMENTAL RESULTS Data collection for these additional experiments followed the same procedures as outlined earlier. This included measuring and recording: (1) surface runoff volume, (2) surface runoff mass, (3) initial turbidity, (4) runoff samples (turbidity versus time) and, (5) amount of soil eroded from test plots. Particle size information was not collected since no observable differences occurred between treatments in past experiments. The following sections focus solely on results generated by the recommended application rate of 35 lbs/acre for the three various application methods (i.e. dry granular PAM, liquid PAM, and 48-hr liquid PAM). F.2.1 Surface Runoff Surface runoff volumes recorded for each replication are illustrated in Figure F1. Test plots treated with '48-hr liquid PAM were observed to have slightly lower amounts of surface runoff. As stated from the previous experimental results, any difference in runoff amounts was attributed to fluctuations in the operation of the rainfall simulator 186 (i.e. slight pressure increases or decreases). This was confirmed through the fact that runoff amounts did not vary during individual experiments. Specifically, no observable increase or decrease in surface runoff occurred during an experiment duration and the runoff rates remained constant. Therefore, since the runoff rate remained steady throughout an experiment, it was concluded that PAM treatments had no effect on the amount of runoff exiting a test plot. 0 500 1000 1500 2000 2500 3000 0 5 10 15 20 25 30 35 40 45 50 55 60 Vo lum e (g al/ ac re) Time (min) Control Dry 35 Liquid 35 48-hr Liquid 35 * * * Test 1 Test 2 Test 3 Test 4 Note: ?*? denotes 15 minute break in between tests Figure F1 Average Surface Runoff vs. Time. Table F1 shows the specific values for the average runoff values generated from the additional tests. These values confirm what was illustrated with the runoff volumes shown in Figure F1. The percent reductions in runoff showed that 48-hr liquid PAM experienced an average difference of approximately 11.6% when compared to the control condition. As observed in the previous experiments, this difference in runoff volume was not considered indicative of PAM treatments? performance qualities, but rather the 187 fluctuating conditions of the rainfall simulator, as further confirmed using statistical analyses. Table F1 Average Surface and Cumulative Runoff for Each Test Condition Runoff a (gal/acre) Standard Deviation b (gal/acre) Percent Reduction c Cumulative Runoff d (gal/acre) Test 1 Control 2279.0 236.9 - 34184.4 Dry 35 2448.0 328.9 -7.4% 36719.7 Liquid 35 1984.0 264.9 12.9% 29760.3 48-hr Liquid 35 1978.4 107.7 13.2% 32636.0 Test 2 Control 2331.1 236.6 - 34967.1 Dry 35 2432.1 289.8 -4.3% 36481.5 Liquid 35 2140.6 251.4 8.2% 32108.5 48-hr Liquid 35 2044.1 81.2 10.3% 32227.6 Test 3 Control 2326.6 243.8 - 34899.0 Dry 35 2324.3 263.4 0.1% 34865.0 Liquid 35 2212.0 221.2 4.9% 33180.5 48-hr Liquid 35 1982.9 113.1 13.0% 31802.2 Test 4 Control 2335.7 241.2 - 35035.2 Dry 35 2227.9 195.7 4.6% 33418.7 Liquid 35 2210.9 222.6 5.3% 33163.5 48-hr Liquid 35 2051.0 65.7 10.0% 36481.5 Notes: ?a? Average surface runoff vs. time for each test ?b? Standard deviation of surface average runoff vs. time ?c? Denotes values normalized by control condition ?d? Average cumulative surface runoff for each 15 min. test F.2.1.1 Statistical Analysis: Surface Runoff Following similar procedures to analyze the collected data, an ANOVA analysis with Tukey-Kramer multiple comparisons was used to determine if there was a statistically significant differences observed between treatment pairs. Tables F2 through F5 illustrate the results generated from the Tukey-Kramer analyses and whether or not a specific condition was statistically significant. During the course of the 4 tests, different treatments conditions were observed to have statistically significant differences. 188 However, these differences varied between tests with little consistency. Therefore, since the different treatment pairs that were statistically significant appear to have no appreciable consistency, it was concluded that the previous assumption that PAM treatments had no affect on surface runoff amounts was validated. Table F2 Tukey-Kramer Multiple Comparisons on Average Surface Runoff [Test 1] Comparison ?i - ?j CI [LB] CI [UB] Significantly Different Control vs. Dry 35 169.0 -16.7 169.0 No Control vs. Liquid 35 294.9 109.2 294.9 Yes Control vs. 48-hr Liquid 35 300.6 114.9 300.6 Yes Dry 35 vs. Liquid 35 464.0 278.3 464.0 Yes Dry 35 vs. 48-hr Liquid 35 469.6 283.9 469.6 Yes Liquid 35 vs. 48-hr Liquid 35 5.7 -180.0 5.7 No Notes: [LB] signifies lower bound of confidence interval [UB] signifies upper bound of confidence interval qcrit = 3.75 Table F3 Tukey-Kramer Multiple Comparisons on Average Surface Runoff [Test 2] Comparison ?i - ?j CI [LB] CI [UB] Significantly Different Control vs. Dry 35 101.0 -64.3 101.0 No Control vs. Liquid 35 190.6 25.3 190.6 Yes Control vs. 48-hr Liquid 35 287.0 121.7 287.0 Yes Dry vs. Liquid 35 291.5 126.2 291.5 Yes Dry vs. 48-hr Liquid 35 388.0 222.7 388.0 Yes Liquid 35 vs. 48-hr Liquid 35 96.4 -68.9 96.4 No Notes: [LB] signifies lower bound of confidence interval [UB] signifies upper bound of confidence interval qcrit = 3.75 189 Table F4 Tukey-Kramer Multiple Comparisons on Average Surface Runoff [Test 3] Comparison ?i - ?j CI [LB] CI [UB] Significantly Different Control vs. Dry 35 2.3 -165.0 2.3 No Control vs. Liquid 35 114.6 -52.7 114.6 No Control vs. 48-hr Liquid 35 343.7 176.4 343.7 Yes Dry 35 vs. Liquid 35 112.3 -55.0 112.3 No Dry 35 vs. 48-hr Liquid 35 341.4 174.1 341.4 Yes Liquid 35 vs. 48-hr Liquid 35 229.1 61.8 229.1 Yes Notes: [LB] signifies lower bound of confidence interval [UB] signifies upper bound of confidence interval qcrit = 3.75 Table F5 Tukey-Kramer Multiple Comparisons on Average Surface Runoff [Test 4] Comparison ?i - ?j CI [LB] CI [UB] Significantly Different Control vs. Dry 35 107.8 -52.4 107.8 No Control vs. Liquid 35 124.8 -35.4 124.8 No Control vs. 48-hr Liquid 35 284.7 124.6 284.7 Yes Dry 35 vs. Liquid 35 17.0 -143.2 17.0 No Dry 35 vs. 48-hr Liquid 35 177.0 16.8 177.0 Yes Liquid 35 vs. 48-hr Liquid 35 159.9 -0.2 159.9 No Notes: [LB] signifies lower bound of confidence interval [UB] signifies upper bound of confidence interval qcrit = 3.75 F.2.2 Initial Turbidity Turbidity levels were recorded from surface runoff that was collected every minute. These recorded levels were referred to as an initial turbidity as surface runoff exited each test plot. The averaged values from experimental replications are illustrated below in Figure F2. It was observed that the 48-hr liquid PAM initial performed similarly to dry PAM, but quickly began to reach turbidity levels that were recorded from liquid PAM treatments without a drying period. This trend continued during the first two tests, but subsequent tests (i.e. test 3 and test 4) showed that turbidity was higher than 190 recorded in the liquid PAM that had no drying period. This difference was observed to be approximately 500 NTU or higher. 0 500 1000 1500 2000 2500 3000 3500 4000 4500 0 5 10 15 20 25 30 35 40 45 50 55 60 Tu rb idi ty (N TU ) Time (min) Control Dry 35 Liquid 35 48-hr Liquid 35 * Test 1 Test 2 Test 3 Test 4 * * Note: ?*? denotes 15 minute break in between tests Figure F2 Average Initial Turbidity of Surface Runoff vs. Time. Table F6 shows the specific average values from these experiments. On average, 48-hr liquid PAM reduced turbidity levels over time by approximately 69%. Following test 2, these turbidity levels began to increase, as indicated by the decrease in percent reduction to the control condition over time. This increase in turbidity observed in the 48-hr liquid PAM was attributed to how PAM molecules interact with the soil surface when applied. When dry PAM granules are added to water, they become ?activated?. The application is sprayed on to a test plot, and the drying period permits the PAM to bond with surface soil particles and creates an extremely thin layer of protection. This molecular bonding occurred once the PAM was activated by the initial mixture with water, and as rainfall was initiated, the 48-hr liquid PAM remained active in keeping soil attached on the surface. However, once PAM activates, it cannot ?reactivate? and as the PAM was washed away by the runoff, it was incapable of acting as an effective 191 sedimentation control measure by bonding with the suspended soil particles, as witness with dry PAM applications. Once the protective layer of PAM was washed away, the test plots are effectively ?untreated? and result in higher turbidity levels, as observed. Table F6 Average Initial Turbidity Results for Surface Runoff Condition Average Turbidity a (NTU) Standard Deviation b (NTU) Percent Reduction c Test 1 Control 3414.7 513.6 - Dry 35 103.4 28.9 97.0% Liquid 35 784.8 167.3 77.0% 48-hr Liquid 35 481.3 545.6 85.9% Test 2 Control 3405.4 395.6 - Dry 35 99.0 14.7 97.1% Liquid 35 776.7 176.0 77.2% 48-hr Liquid 35 892.3 613.0 73.9% Test 3 Control 3553.6 304.3 - Dry 35 96.1 15.7 97.3% Liquid 35 775.5 187.4 78.2% 48-hr Liquid 35 1343.5 453.8 60.7% Test 4 Control 3636.6 233.5 - Dry 35 95.5 15.2 97.4% Liquid 35 789.6 145.1 78.3% 48-hr Liquid 35 1498.3 365.5 56.1% Notes: ?a? Average of initial turbidity vs. time for each test ?b? Standard deviation for average initial turbidity vs. time ?c? Denotes values normalized by control condition Conversely, liquid PAM treatments that were not given time to dried had recorded turbidity levels that were lower than 48-hr liquid PAM. Since these treatments were not given the time to dry, the molecular bonding with the soil surface did not adequately form. Therefore, liquid PAM treatments were washed off without the means to provide protection on the soil surface. However, the PAM molecules were still active and were able to bond with suspended soil particles in the surface runoff and act as a more 192 effective sedimentation control measure than liquid PAM applications that had dried for 48 hours. F.2.2.1 Statistical Analysis: Initial Turbidity Tables F7 through F10 illustrate the results from the Tukey-Kramer multiple comparison. As observed in previous experiments, differences between all possible pairs of condition options were statistically significant. This shows that the all possible combinations of PAM treatments had a significant effect on initial turbidity levels. Combinations with higher values between the mean differences indicate which treatments had a greater effect at reducing turbidity. As observed, dry PAM applied 35 lbs/acre performed the best as a sediment control measure over both different liquid PAM application methods. Table F7 Tukey-Kramer Multiple Comparisons on Average Initial Turbidity [Test 1] Comparison ?i - ?j CI [LB] CI [UB] Significantly Different Control vs. Dry 35 3311.3 3165.1 3311.3 Yes Control vs. Liquid 35 2629.9 2483.7 2629.9 Yes Control vs. 48-hr Liquid 35 2933.4 2787.2 2933.4 Yes Dry 35 vs. Liquid 35 681.4 535.2 681.4 Yes Dry 35 vs. 48-hr Liquid 35 377.9 231.7 377.9 Yes Liquid 35 vs. 48-hr Liquid 35 303.5 157.3 303.5 Yes Notes: [LB] signifies lower bound of confidence interval [UB] signifies upper bound of confidence interval qcrit = 3.75 193 Table F8 Tukey-Kramer Multiple Comparisons on Average Initial Turbidity [Test 2] Comparison ?i - ?j CI [LB] CI [UB] Significantly Different Control vs. Dry 35 3306.4 3202.3 3306.4 Yes Control vs. Liquid 35 2628.7 2524.5 2628.7 Yes Control vs. 48-hr Liquid 35 2513.1 2409.0 2513.1 Yes Dry 35 vs. Liquid 35 677.8 573.6 677.8 Yes Dry 35 vs. 48-hr Liquid 35 793.3 689.2 793.3 Yes Liquid 35 vs. 48-hr Liquid 35 115.6 11.5 115.6 Yes Notes: [LB] signifies lower bound of confidence interval [UB] signifies upper bound of confidence interval qcrit = 3.75 Table F9 Tukey-Kramer Multiple Comparisons on Average Initial Turbidity [Test 3] Comparison ?i - ?j CI [LB] CI [UB] Significantly Different Control vs. Dry 35 3457.5 3355.4 3457.5 Yes Control vs. Liquid 35 2778.2 2676.1 2778.2 Yes Control vs. 48-hr Liquid 35 2210.2 2108.1 2210.2 Yes Dry 35 vs. Liquid 35 679.3 577.2 679.3 Yes Dry 35 vs. 48-hr Liquid 35 1247.3 1145.2 1247.3 Yes Liquid 35 vs. 48-hr Liquid 35 568.0 465.9 568.0 Yes Notes: [LB] signifies lower bound of confidence interval [UB] signifies upper bound of confidence interval qcrit = 3.75 Table F10 Tukey-Kramer Multiple Comparisons on Average Initial Turbidity [Test 4] Comparison ?i - ?j CI [LB] CI [UB] Significantly Different Control vs. Dry 35 3541.1 3470.2 3541.1 Yes Control vs. Liquid 35 2847.0 2776.1 2847.0 Yes Control vs. 48-hr Liquid 35 2138.3 2067.4 2138.3 Yes Dry 35 vs. Liquid 35 694.1 623.1 694.1 Yes Dry 35vs. 48-hr Liquid 35 1402.8 1331.8 1402.8 Yes Liquid 35 vs. 48-hr Liquid 35 708.7 637.7 708.7 Yes Notes: [LB] signifies lower bound of confidence interval [UB] signifies upper bound of confidence interval qcrit = 3.75 194 F.2.3 Turbidity vs. Time Additional surface runoff samples were collected at every 5 and 10 minutes periods during each test. These samples were used to determine turbidity over time. The three PAM treatments discussed in this section and the control are illustrated together in Figure F3. As observed with previous experiments, the average turbidity levels over time for all of the PAM treatments performed much better than the control condition, by reducing turbidity much quicker. However, when the different PAM treatments were compared together, as seen in Figure F4(a) through F4(f), dry PAM at 35 lbs/acre was able to reduce turbidity levels over time better than both liquid PAM applications throughout all four tests. 0 500 1000 1500 2000 2500 3000 3500 0 1 2 3 4 5 6 7 8 9 10 Tu rb idi ty (N TU ) Time (min) Control Dry 35 Liquid 35 48-hr Liquid 35 Figure F3 Average Recorded Turbidity for All Samples vs. Time. As observed with the recorded initial turbidity levels, 48-hr liquid PAM was able to reduce turbidity during the initial tests, but subsequent tests show that turbidity over time increased, as the PAM treatment were being washed away. Liquid PAM with no drying period was able to provide better sedimentation protection, since the PAM 195 molecules were still relatively active as they were washed away. These figures confirm that dry PAM performs better as a sedimentation control measure than liquid PAM applications. 0 500 1000 1500 2000 2500 3000 3500 4000 0 20 40 60 80 100 120 140 160 180 200 Tu rb idi ty (N TU ) Time (sec) Test 1 Test 2 Test 3 Test 4 0 500 1000 1500 2000 2500 3000 3500 4000 0 20 40 60 80 100 120 140 160 180 200 Tu rb idi ty (N TU ) Time (sec) Test 1 Test 2 Test 3 Test 4 (a) 35 lbs/acre at 5 Minutes [Dry PAM] (b) 35 lbs/acre at 10 Minutes [Dry PAM] 0 500 1000 1500 2000 2500 3000 3500 4000 0 20 40 60 80 100 120 140 160 180 200 Tu rb idi ty (N TU ) Time (sec) Test 1 Test 2 Test 3 Test 4 0 500 1000 1500 2000 2500 3000 3500 4000 0 20 40 60 80 100 120 140 160 180 200 Tu rb idi ty (N TU ) Time (sec) Test 1 Test 2 Test 3 Test 4 (c) 35 lbs/acre at 5 Minutes [Liquid PAM] (d) 35 lbs/acre at 10 Minutes [Liquid PAM] 0 500 1000 1500 2000 2500 3000 3500 4000 0 20 40 60 80 100 120 140 160 180 200 Tu rb idi ty (N TU ) Time (sec) Test 1 Test 2 Test 3 Test 4 0 500 1000 1500 2000 2500 3000 3500 4000 0 20 40 60 80 100 120 140 160 180 200 Tu rb idi ty (N TU ) Time (sec) Test 1 Test 2 Test 3 Test 4 (e) 48-hr 35 lbs/acre at 5 Minutes [Liquid PAM] (f) 48-hr 35 lbs/acre at 10 Minutes [Liquid PAM] Figure F4 Treatment?s Average Recorded Turbidity from 5 and 10 Minute Samples. 196 F.2.4 Soil Loss The soil transported from each test plot was collected and filtered at three minute intervals. These soil samples were oven dried and a weight was measured to determine the amount of soil that was eroded from test plots during experimentation. Figure F5 illustrates average soil loss from the test plots versus time and Figure F6 shows cumulative soil loss during an experiment?s duration. The additional experiments with 48-hr liquid PAM indicated that the amount of eroded soil from a test plot was much lower than previous experiments of dry PAM and liquid PAM with no drying period. During test 1, 48-hr liquid PAM experienced very little soil loss, but as the duration of an experiment increased, so did the amount of eroded soil. These increased levels of eroded soil were still observed to be less than previous experiments as shown in Figure F5. Figure F6 illustrates that the total accumulated soil from 48-hr liquid PAM was approximately 2,300 lbs/acre per 15 minute test interval, compared to dry PAM at 35 lbs/acre eroded soil of approximately 4,000 lbs/acre. 0 500 1000 1500 2000 2500 0 5 10 15 20 25 30 35 40 45 50 55 60 M as s (l b/a cre ) Time (min) Control Dry 35 Liquid 35 48-hr Liquid 35 Test 1 Test 2 Test 3 Test 4 * * * Note: ?*? denotes 15 minute break in between tests Figure F5 Average Soil Loss versus Time. 197 0 2000 4000 6000 8000 10000 0 5 10 15 20 25 30 35 40 45 50 55 60 M as s (l b/a cre ) Time (min) Control Dry 35 Liquid 35 48-hr Liquid 35 Test 1 Test 2 Test 3 Test 4 * * * Note: ?*? denotes 15 minute break in between tests Figure F6 Average Cumulative Soil Loss versus Time. Table F11 shows the specific values of the average soil loss from the experiments. It was observed that 48-hr liquid PAM reduced soil loss when compared to the control by an average of approximately 76%. This is much higher than the dry PAM at 35 lbs/acre by approximately 28%. This indicates that when liquid PAM was allotted a period to dry, it was capable of performing better than dry PAM as an erosion control measure. This is attributed to how liquid PAM bonds with the soil surface as previously stated. Since the liquid PAM was activated when the dry granules were mixed with water, it was able to bond with the soil surface when it was applied. By providing a period of 48 hours for the application to dry, the liquid PAM treatments in this case had ample time to bond with the soil surface and provide a layer of protection against erosion. This layer of protection kept soil particles from being detached from the soil surface and transported in the stormwater. The slight increase observed in soil loss as an experiment?s duration increased indicated how this protective layer was slowing being washed away by surface 198 runoff. Therefore, liquid PAM (when allowed to dry) performed better as an erosion control measure, when compared to dry PAM treatments, which performed better as a sediment control measure by improving water quality and reducing turbidity levels. Table F11 Average Soil Loss due to Surface Runoff Condition Soil Loss a (lbs/acre) Standard Deviation b (lbs/acre) Percent Reduction c Test 1 Control 1663.8 693.8 - Dry 35 798.9 293.5 52.0% Liquid 35 1307.2 377.5 21.4% 48-hr Liquid 35 225.1 157.0 86.5% Test 2 Control 1210.6 189.7 - Dry 35 674.6 262.9 44.3% Liquid 35 1337.9 322.9 -10.5% 48-hr Liquid 35 403.9 105.5 75.7% Test 3 Control 1420.7 236.3 - Dry 35 740.6 308.2 47.9% Liquid 35 1419.5 256.0 0.1% 48-hr Liquid 35 476.6 120.8 71.4% Test 4 Control 1506.5 194.0 - Dry 35 787.5 334.6 47.7% Liquid 35 1421.9 233.5 5.6% 48-hr Liquid 35 499.4 160.7 70.0% Notes: ?a? Average of eroded soil vs. time for each test ?b? Standard deviation for average soil loss vs. time ?c? Denotes values normalized by control condition F.2.4.1 Statistical Analysis Soil Loss Tables F12 through F15 illustrate the Tukey-Kramer analyses on the soil loss for the multiple test plots. For all four tests, dry PAM at 35 lbs/acre and 48-hr liquid PAM at 35 lbs/acre were observed to be statistically significant when compared to the control and other treatments. Liquid PAM 35 did not provide any significant results when compared to the control, indicating that it had no effect in reducing soil loss. The differences in 199 means for 48-hr liquid PAM indicate that this treatment option performed the best out the different combinations. Table F12 Tukey-Kramer Multiple Comparisons on Average Soil Loss [Test 1] Comparison ?i - ?j CI [LB] CI [UB] Significantly Different Control vs. Dry 35 386.5 216.0 386.5 Yes Control vs. Liquid 35 121.8 -48.7 121.8 No Control vs. 48-hr Liquid 35 960.3 789.8 960.3 Yes Dry 35 vs. Liquid 35 508.4 337.8 508.4 Yes Dry 35 vs. 48-hr Liquid 35 573.8 403.3 573.8 Yes Liquid 35 vs. 48-hr Liquid 35 1082.2 911.6 1082.2 Yes Notes: [LB] signifies lower bound of confidence interval [UB] signifies upper bound of confidence interval qcrit = 4.05 Table F13 Tukey-Kramer Multiple Comparisons on Average Soil Loss [Test 2] Comparison ?i - ?j CI [LB] CI [UB] Significantly Different Control vs. Dry 35 569.0 458.0 569.0 Yes Control vs. Liquid 35 94.2 -16.7 94.2 No Control vs. 48-hr Liquid 35 839.7 728.7 839.7 Yes Dry 35 vs Liquid 35 663.2 552.3 663.2 Yes Dry 35 vs. 48-hr Liquid 35 270.7 159.7 270.7 Yes Liquid 35 vs. 48-hr Liquid 35 933.9 822.9 933.9 Yes Notes: [LB] signifies lower bound of confidence interval [UB] signifies upper bound of confidence interval qcrit = 4.05 Table F14 Tukey-Kramer Multiple Comparisons on Average Soil Loss [Test 3] Comparison ?i - ?j CI [LB] CI [UB] Significantly Different Control vs. Dry 35 650.0 553.9 650.0 Yes Control vs. Liquid 35 28.8 -67.3 28.8 No Control vs. 48-hr Liquid 35 914.1 818.0 914.1 Yes Dry 35 vs. Liquid 35 678.8 582.7 678.8 Yes Dry 35 vs. 48-hr Liquid 35 264.1 167.9 264.1 Yes Liquid 35 vs. 48-hr Liquid 35 942.9 846.8 942.9 Yes Notes: [LB] signifies lower bound of confidence interval [UB] signifies upper bound of confidence interval qcrit = 4.05 200 Table F15 Tukey-Kramer Multiple Comparisons on Average Soil Loss [Test 4] Comparison ?i - ?j CI [LB] CI [UB] Significantly Different Control vs. Dry 35 585.2 502.0 585.2 Yes Control vs. Liquid 35 49.2 -34.0 49.2 No Control vs. 48-hr Liquid 35 873.3 790.1 873.3 Yes Dry 35 vs. Liquid 35 634.4 551.2 634.4 Yes Dry 35 vs. 48-hr Liquid 35 288.1 204.9 288.1 Yes Liquid 35 vs. 48-hr Liquid 35 922.5 839.3 922.5 Yes Notes: [LB] signifies lower bound of confidence interval [UB] signifies upper bound of confidence interval qcrit = 4.05 F.2.5 Initial Turbidity vs. Soil Loss The amount of eroded soil was plotted together with its respective turbidity level and this relationship is illustrated in Figure F7. The control condition with all the tested PAM treatments is displayed below and shows the overall performance. As stated, it was observed that dry PAM applications were more effective as a sediment control measure by examining turbidity versus time measurements, and this is also observed in Figure F7. Dry PAM treatments are grouped together towards lower turbidity levels (200 NTU to 1400 NTU). The groupings of data are also more aligned vertically, rather than horizontally, indicating that a relatively more consistent range of turbidity was achieved, while variations occurred in the amount of eroded soil. Conversely, liquid PAM treatments were more distributed over turbidity measurements, while more consistent levels of soil loss were achieved, especially as observed with 48-hr liquid PAM. This relationship between turbidity and soil loss, as shown, indicates which treatments perform better as either an erosion or sediment control measure. The liquid PAM applications, specifically the 48-hr liquid PAM, were much 201 more efficient at reducing soil loss during the storm duration. Dry PAM, specifically applied at 35 lbs/acre, and was tightly grouped around lower turbidity levels, indicating that this treatment measure was more effective at reducing sedimentation. 0 500 1000 1500 2000 2500 0 1000 2000 3000 4000 Er od ed So il ( lbs /ac re) Turbidity (NTU) Control Dry 35 Dry 25 Dry 15 Liq 35 Liq 25 Liq 15 48-hr Liquid 35 Figure F7 Average Initial Turbidity vs. Average Eroded Soil. F.3 SUMMARY These results validate the researcher?s claim on how PAM treatments performance was dependent on the method of application. Dry granular PAM performed better as a sediment control measure, due to the fact that as rainfall and surface runoff activated the PAM molecules, they were introduced into the stormwater and bonded with suspended soil particles, promoting flocculation and settling. Liquid PAM that was allowed to dry for 48 hours on the soil surface had time to adequately bond with soil particles and seal the surface, providing a thin layer of protection against erosion. This layer was effective at keeping soil particles from being detached and becoming transported in the stormwater. However, since the PAM was already activated during the 202 application process and bonded with the surface soil particles, any PAM that was washed away in the surface runoff would not bond with the suspended particles, resulting in the higher observed turbidity levels. The liquid PAM applications that were not given time to dry, were initially activated when applied, but did not have the time to bond to the surface. This resulted in a higher amount of eroded soil from test plots. Since the PAM was still active as it was being washed away by the surface runoff, it was still capable of bonding with some of the suspended soil particles present in the runoff. This resulted in low turbidity levels but were not as effective as dry PAM treatments. This was also the reason why liquid PAM treatments with no drying period performed poorly when it came to erosion control. F.4 CONCLUSIONS Two additional experiments conducted by researchers examined the effectiveness of liquid PAM spray applications with a 48 hour dry period prior to exposing test plots to rainfall. Experiments were set-up using the procedures and methods created to simulate the effectiveness of different erosion and sediment control technologies on compacted, 3:1 slopes, representative of typical highway embankments. Liquid PAM applications were allowed to dry for 48-hours prior to the initiation of rainfall. The period of drying time permitted the PAM molecules to adequately bond with soil particles on the soil surface. This bonding occurred once the PAM was activated by water during its initial mixing. Once the PAM had been activated and bonded with the soil surface, an extremely thin layer protected the surface of the soil and 203 assisted in preventing erosion. This effect was confirmed during experimentation, with the observed values of soil loss being the lowest recorded amount, with an average reduction of 76% when compared to the control condition. This was approximately a 28% increase in difference between dry PAM at 35 lbs/acre, which was reported as the most effective treatments from previous experiments. However, when looking at the quality of the water in surface runoff, the 48-hr liquid PAM did not perform as well as the dry and liquid PAM treatments (with no drying time). As stated, PAM treatments become active once water is introduced to the PAM molecules. When the liquid PAM treatments were permitted to dry, this activation of molecules had already occurred, and would not occur again. Therefore, there was little bonding occurring with the suspended soil particles in surface runoff, resulting in the higher turbidity levels as observed in the 48-hr liquid PAM treatments. The other liquid PAM treatments were also activated during application, but since the PAM molecules were not given time to bond with the soil surface, they remained active during an experiment?s duration. As treatments were washed off the test plots, PAM molecules were still active and could bond with the suspended soil particles and promote flocculation, resulting in better turbidity levels than the PAM treatments that had been given time to dry. Since this activation of PAM molecules occurs once water is introduced, the dry PAM treatments could not bond with the soil surface during initial application. During the initiation of rainfall the PAM molecules were slowly and consistently introduce into the surface runoff. These newly activated PAM molecules in the runoff could then bond 204 with suspended soil particles, increasing the particle size, and promoting flocculation. This resulted in the very low observed turbidity levels in the surface runoff. These results indicate that liquid PAM applications (that were given 48-hours to dry) were more effective at reducing the amount of soil loss from test plots. Conversely, dry PAM applications were more effective in reducing turbidity, resulting in improved water quality. These differences in results illustrated that liquid PAM performs better as an erosion control measure, while dry PAM application perform better as an sediment control measure. F.4.1 Recommendations for Practice The following recommendations, as outlined below, represent the researcher?s opinions on the proper methods for using PAM as an erosion and sediment control measure based on the research presented in this report. Additional recommendations can be found in Applied Polymer?s, ?Polymer Enhanced Best Management Practice (PEBMP) Application Guide? for various products with and without PAM. RECOMMENDATIONS FOR USING LIQUID PAM (EROSION CONTROL): 1. Obtain soil sample(s) to determine which PAM formulation to use for a specific site location 2. To apply PAM, a type of hydro-seeder with a method of agitation is required. 3. Ensure that PAM application will have time allotted for drying after application. 4. When preparing treatment, ensure that the recommended rate of dry PAM granules are slowly added to the water and remain agitated to prevent clogging. 5. Ensure that PAM has been fully mixed in water prior to application 205 6. Spray application must be uniformly sprayed with 100% coverage of the soil surface; else treatment will not be effective. 7. Additional sediment control technologies must be used in conjunction with liquid PAM treatment to ensure maximum protection. RECOMMENDATIONS FOR USING DRY PAM (SEDIMENT CONTROL): 1. Obtain soil sample(s) to determine which PAM formulation to use for a specific site location. 2. The recommended dry PAM granules must be spread (by hand or mechanically) uniformly on the soil surface. 3. Additional erosion control technologies must be used in conjunction with dry PAM to ensure maximum protection is provided. F.4.2 Recommendations for Future Research As stated in the recommendations for practice, PAM should never be used by itself as an erosion and sediment control measure. This was confirmed through the research presented, by demonstrating that neither application method was fully capable of providing both erosion and sediment control effectively. Therefore, additional research should be conducted to investigate which combination of PAM and additional erosion and sediment control technologies will provide the best solution from preventing excess NPS pollution, through cost-effective measures. Examples of these erosion and sediment control technologies to test with and without PAM in regards to slope stabilization include: (1) Organic Erosion Control Blankets (ECBs), (2) Inorganic ECBs, (3) Soft- Armoring, (4) Mulching, (5) Jute matting, and (6) Geo-synthetic materials. 206 By examining these products, with and without the addition of PAM, best management practices (BMPs) could be recommended and selected for practical use based on the cost and quality of performance. Analyses could focus the long-term effectiveness for each product and how this would relate to maintenance costs and the necessary time for more permanent products to be implemented (i.e. vegetative cover). This research would provide an invaluable tool for contractors, with the potentially stricter EPA effluent guidelines currently under review, which would require more proactive responses to erosion and sedimentation issues, rather than many reactive responses current in use today. Therefore, these erosion and sediment control technologies would be examine, analyzed, and evaluated using stringent scientific and engineering methods to ensure the best solution, for all interested parties, were selected.