DETACHMENT AND DEPOSITION IN A SIMULATED RILL. A Thesis. Submitted to the Faculty. Purdue University. Thomas Arey Cochrane

Transcription

1 DETACHMENT AND DEPOSITION IN A SIMULATED RILL A Thesis Submitted to the Faculty of Purdue University by Thomas Arey Cochrane In Partial Fulfillment of the Requirements for the Degree of Master of Science in Agricultural Engineering December 1995

2 ii ACKNOWLEDGEMENTS I would like to express my gratitude to Dr. Dennis Flanagan, my major professor, for his guidance, help, and research funds throughout the course of my research. I would also like to express my gratitude to the other members of my advisory committee, Dr. Bernard Engel and Dr. William Wood, for information provided in lectures and discussions during the final stages of the research. I would specially like to thank Steve Parker for all the valuable help and friendship during all stages of the research. I would also like to acknowledge the help provided by Scott McAfee during the construction of the equipment and my undergraduate assistants Rich Niemoeller and Jessica Paull for their valuable help during the experimental runs. Many thanks go to my parents, who are always there for me giving me wise advise and encouragement, and my friends and officemates for their constant support. Finally I would like to thank all the staff of the NSERL for providing an excellent research environment.

3 iii TABLE OF CONTENTS Page LIST OF TABLES... vi LIST OF FIGURES... viii ABSTRACT...x ACKNOWLEDGEMENTS...II TABLE OF CONTENTS... III LIST OF TABLES... VI LIST OF FIGURES...VIII ABSTRACT...X CHAPTER 1. INTRODUCTION Erosion Mechanics Rill Erosion Objectives Definition of Terms...3 CHAPTER 2. LITERATURE REVIEW Current Erosion Models and Equations The Effects of Rainfall on Flow Depth Field Studies of Crop-Row Furrows Laboratory Rill Erosion Mechanics Studies Literature Review Summary Objectives of Study as Related to Literature Review...17 CHAPTER 3. EQUIPMENT DESIGN AND MATERIALS The Rill Simulator Flume...19

4 iv Main structure of the flume Infiltration Sediment feeder Sediment Data Acquisition System Data acquisition board Pressure sensors Flow sensors Endplate Design Profile meter (LVDT) Rain Simulators Laser Scanner...31 CHAPTER 4. EXPERIMENTAL STUDIES: SETUP AND PROCEDURE Overview and Purpose of Studies Rill Simulator Experiment Setup and Procedure Pre-run setup Pre-calibrated constants Measurements Laser scan data Procedure Detachment Study Deposition Study...38 CHAPTER 5. RESULTS AND DISCUSSION Detachment Study Water flow Outflow sediment concentration Deposition / detachment on bed Detachment rate and transport capacity Detachment modeling Deposition Study Data Analysis Rainfall effects on sediment bed Laser scan data Effects of rainfall intensity and flow rate on deposition Rainfall intensity effects on sediment discharge The influence of rainfall intensity on flow velocity Infiltration effects on deposition and sediment discharge Deposition modeling...79 CHAPTER 6. SUMMARY AND CONCLUSIONS Summary of setup and experiments General Conclusions...83

5 v Detachment study conclusions Deposition study conclusions...84 CHAPTER 7. RECOMMENDATIONS...85 LIST OF REFERENCES...87 APPENDICES Appendix A. Data acquisition system electronics...91 Appendix B. Rill simulator program for data acquisition system Appendix C. Rainfall simulator calibration Appendix D. Detachment and deposition data Appendix E. Detachment modeling procedure...147

6 vi LIST OF TABLES Page Table 1. Sediment Transport Capacities (g/min) for Sands.(Meyer et al. 1983)...14 Table 2. Crystal Sand Screen Analysis (Ottawa Foundry Sand, U.S. Silica Co.) Table 3. Crystal Sand Physical Properties...23 Table 4. Data Acquisition Board Properties Table 5. Pressure Sensors Table 6. Water Inflow Sensor Table 7. Infiltration Flow Sensor...27 Table 8. Linear Variable Differential Transducer Table 9. Detachment Treatments Table 10. Deposition Treatments...38 Table 11. Water Flow Means and Standard Deviations Table 12. Sediment Concentration in Outflow Table 13. Two Factor ANOVA. Water and Sediment Inflow Effects on Outlet Sediment Concentration...44 Table 14. Water Flow Means and Standard Deviations with 0 g/s Feedrate...63 Table 15. Water Flow Means and Standard Deviations with 5.3 g/s Feedrate...64 Table 16. Water Flow Means and Standard Deviations with g/s Feedrate...65

7 vii LIST OF FIGURES Page Figure 1. Applicability of transport formulas (Alonso et al., 1981)...9 Figure 2. Main frame with stands...20 Figure 3. Galvanized sheet metal forming rill...21 Figure 4. Rill simulator...21 Figure 5. Data acquisition system...25 Figure 6. Endplate design...28 Figure 7. Pulley system...29 Figure 8. Profile meter...30 Figure 9. Laser scanner movable frame...31 Figure 10. Laser scan area...34 Figure 11. Experiment AAA, Rep1 (top) and Rep 2 (bottom) L/min water inflow, 0 g/sec sediment feedrate...48 Figure 12. Experiment AAB, Rep1 (top) and Rep 2 (bottom) L/min water inflow, 5.3 g/sec sediment feedrate...48 Figure 13. Experiment AAE, Rep 1 (top) and Rep 2 (bottom) L/min water inflow, g/sec sediment feedrate...49 Figure 14. Experiment AAC, Rep 1 (top) and Rep 2 (bottom) L/min water inflow, 0 g/sec sediment feedrate...49 Figure 15. Experiment AAD, Rep 1 (top) and Rep 2 (bottom) L/min water inflow, 5.3 g/sec sediment feedrate...50 Figure 16. Experiment AAF, Rep 1 (top) and Rep 2 (bottom) L/min water inflow, g/sec sediment feedrate...50 Figure 17. Influences of flow and sediment feedrate on first scan interval detachment (64 to 91 cm from top of bed)...51 Figure 18. Influences of flow and sediment feedrate on second scan interval detachment (91 to 118 cm from top of bed)...51 Figure 19. Experiment AAA (22.7 L/min, 0 g/s feedrate) detachment modeling...55 Figure 20. Experiment AAB (22.7 L/min, 5.3 g/s feedrate) detachment modeling...56 Figure 21. Experiment AAC (30.3 L/min, 0 g/s feedrate) detachment modeling...57 Figure 22. Experiment AAD (30.3 L/min, 5.3 g/s feedrate) detachment modeling...58 Figure 23. Experiment AAE (22.7 L/min, g/s feedrate) detachment modeling...59 Figure 24. Experiment AAF (30.3 L/min, g/s feedrate) detachment modeling...60

8 Figure 25. Example of deposition study scan images (experiment BAA, rep. 1, rain 80 mm/hr, inflow 22.7, and feedrate g/s)...69 Figure 26. Influences of rainfall intensity on detachment/deposition on bed with g/s feedrate and 30.3 L/min inflow (average of two reps)...70 Figure 27. Influences of rainfall intensity on detachment/deposition on bed with 5.3 g/s feedrate and 22.7 L/min inflow (average of two reps)...71 Figure 28. Influences of rainfall intensity on detachment/deposition on bed with 5.3 g/s feedrate and 30.3 L/min inflow (average of two reps)...72 Figure 29. Influences of flow rate and rainfall on first laser scan interval (64 to 91 cm from top of bed) with 0 g/s feedrate...72 Figure 30. Rainfall intensity effects on sediment discharge for water flow rate of 22.7 L/min and feedrate of g/s...74 Figure 31. Effects of rainfall on flow velocity with g/s feedrate...75 Figure 32. Effects of rainfall on flow velocity with 5.3 g/s feedrate...75 Figure 33. Effects of rainfall on flow velocity with 0 g/s feedrate...76 Figure 34. Effects of infiltration on deposition, 5.3 g/s feedrate...77 Figure 35. Effects of infiltration on deposition, g/s feedrate...77 Figure 36. Effects of infiltration on sediment discharge, 5.3 g/s feedrate...78 Figure 37. Effects of infiltration on sediment discharge, g/s feedrate...78 Figure 38. Flow depth and rainfall vs. β value...81 viii

9 ix ABSTRACT Cochrane, Thomas Arey. M.S.Ag.E., Purdue University, December Detachment and Deposition in a Simulated Rill. Major Professor: Dennis C. Flanagan. An experimental setup was designed to simulate conditions in an agricultural rill. The setup included a 3.6 m long flume with a sediment feeder and rainfall simulators, a laser scanner, and a data acquisition system. The experimental setup was named the rill simulator and was designed to study the effects of different parameters such as slope, rill width, sediment input from the top and sides, infiltration, sediment type, flow depth, and rainfall intensity on detachment and deposition in a rill. The laser scanner was used as a tool to measure the surface of the sediment bed before and after each experimental run and obtain a digital image of the deposition and detachment that had occurred as a result of the experimental treatment. The data acquisition system was used to control the different sensors and equipment which measured and controlled the different variables in the rill such as water inflow, infiltration, and endplate height adjustment. Two preliminary experiments were conducted to test the equipment and to direct further research. The first experiment was named the detachment study and studied the effects of water and sediment inflow on a 25 cm wide fine sand bed at 5% slope with no rainfall and no infiltration. Results indicated that the equipment showed consistency in the experiments and that qualitative observations of the detachment that occurred in the rill were comparable with those predicted by the detachment equation in WEPP (Foster et al., 1995). For low sediment inflow rates, the detachment equation could be tuned to predict reasonable results of detachment in the rill, but it was found that the different parameters such as the shear stresses, erodibility, and transport capacity were difficult to

10 x calculate correctly. The second study was named the deposition study and examined the effects of rainfall, sediment inflow, and flow rate on a 25 cm fine sand bed at 1 % slope. The results also showed that there was consistency in the functionality of the rill simulator. Further results indicated that for the flow conditions studied, increases in rainfall rate did not show a consistent effect on either deposition or sediment transport. It appeared that the interaction of rainfall s effects on turbulence (enhanced transport) and flow velocity (decrease transport) caused the mixed results. Infiltration also may have affected both deposition and transport. Modeling was attempted using the WEPP deposition equation (Foster et al., 1995) which indicated that the turbulence factor for the conditions tested ranged from 0.05 to 0.2 (WEPP currently uses 0.5) and was correlated with flow depth. Fall velocity and transport capacity were important factors in the equation and found to be difficult to estimate. It was recommended that further research be conducted to establish methods to estimate these values correctly. Finally it was recommended that further research be conducted to study the effects of flow depth, rainfall intensity, finer sediment, and slope on both detachment and deposition in rills with sediment also being introduced from the sides to simulate natural conditions.

11 1 CHAPTER 1. INTRODUCTION 1.1.Erosion Mechanics Soil erosion by water is a complex mechanism which depends on various landuse, weather, hydrologic, topographic, and soil properties. The combination of these factors dictate the rates at which sediment detachment, transport, and deposition occur. Although there has been much study in most of the aspects of soil erosion, the overall picture is not complete. One area that needs further study is the effect of rainfall and flow depth on sediment transport and deposition in rills. This focused study of erosion mechanics is important in agriculture, particularly to the problem of sediment transport and deposition between row crops. An important aspect of the study of erosion mechanics is that it will lead to the development and improvement of erosion prediction technologies. Erosion prediction models, such as the Water Erosion Prediction Project (WEPP) developed by the United States Department of Agriculture (Flanagan and Nearing, 1995), rely partly on data obtained through erosion mechanics research. The accuracy of the equations and parameters used in such models also depend on their validation with field and laboratory experiments. It is therefore of utmost importance that these studies be conducted in a way which is representative of the natural conditions and variables being studied. Accurate erosion models would allow the farmer, agronomist, conservationist, or engineer to predict the rate and location of erosion and deposition at a particular site.

12 2 Alternative management practices as well as the impact of various erosion control structures can then be evaluated. 1.2.Rill Erosion On overland flow areas, soil erosion by water can be divided into two distinct components: the rill erosion processes and the interill erosion processes. Interill erosion is the detachment, deposition, and sediment transport that occurs on land due to the effects of rainfall and sheet flow of water, whereas rill erosion can be defined as the detachment, deposition, and sediment transport from water running though concentrated flow paths. Since this thesis will concentrate on the study of rills, it is important to define them. The definition of a rill given by the FAO is that a rill is a channel that is small enough to be removed by normal tillage operations such as agricultural crop furrows (FAO, 1965). Although this definition gives us an idea of the rill s size limitations, it does not describe the actual rill. A better description states that there are two types of rills, incised channels caused by net erosion, and shallow channels lined with their own sediment caused by net deposition. Both types of rills were simulated in the experiments conducted for this study. 1.3.Objectives The primary objective of this study was to construct an experimental device in which rill erosion can be accurately simulated. This involved the design and construction of a flume which has the variables of slope, rill width, infiltration rate, and side slope steepness. The experimental setup would also feature variable water inflow, sediment inflow, and rainfall. A few laboratory studies that involved the construction of a flume to simulate conditions in a rill have been done, but none have incorporated all the variables in this study. Testing of the accuracy and reliability of these parameters was also part of the primary objective.

13 3 The secondary objective was to study some aspects of rill erosion mechanics. Specifically the detachment of sediment in a rill with shallow flow depths and the deposition and transport of sediment in a rill as influenced by rainfall, infiltration, and flow depth. To fulfill these objectives, two groups of experiments were conducted, and these have been labeled the detachment study and the deposition study. Gaining more knowledge on these subjects will help us to better understand the mechanics of sediment transport in shallow rill flow and will enable the further development of erosion models. this thesis: 1.4.Definition of Terms The following is a definition of all the abbreviations used throughout the text of ANSWERS:...Areal Nonpoint Source Watershed Environment Response Simulation. ARC/INFO:...A GIS created by the Environmental Systems Research Institute, Inc. CREAMS:...Chemicals, Runoff, and Erosion from Agricultural Management Systems. EPIC:...Erosion Productivity Impact Calculator. FAO:...Food and Agriculture Organization. FORTRAN:...Computer programming language. GIS:...Geographic Information System. GRASS:...Geographical Resources Analysis Support System, a GIS. LVDT:...Linear Variable Differential Transducer. NSERL:...National Soil Erosion Research Laboratory. QBasic:...Microsoft Quick Basic computer programming language. WEPP:...Water Erosion Prediction Project.

14 4 CHAPTER 2. LITERATURE REVIEW Erosion mechanics is a subject that has intrigued many scientists and has gained perspective in current times due to the environmental conservation and water quality issues that our society faces. Much of the research done in erosion mechanics is geared towards creating models that can predict erosion under different types of situations. These models are then used to remedy or solve environmental concerns. In the following sections a few of these models will be described as well as the different experiments that helped create such models. A greater emphasis will be given to studies involving rill sediment transport and deposition for sand particles. 2.1.Current Erosion Models and Equations During the 1930s, a man by the name of Hugh Hammond Bennett lobbied towards the creation of an organization whose mission would be to protect the agricultural soils of the United States. This lead to the creation of the Soil Conservation Service. Original erosion prediction equations were then developed in response to needs by the Soil Conservation Service and others to protect and manage the land resource. One of the most well known models that was developed to predict erosion in agricultural fields was the USLE (Wischmeier and Smith 1960). The USLE is an empirically based model which was created by collecting vast amounts of field data and uses the collected data to calculate parameters for the different factors involved (A=RKLSCP). The large amount of data collected enabled the approximate prediction of

15 5 erosion in different types of agricultural fields. Although the USLE is useful to give a general idea of the erosion problem, it can not be used to estimate deposition or sediment delivery from fields where deposition has occurred, because these processes were excluded from the experimental database. Meanwhile other studies provided a more theoretical view of the processes involved in soil erosion. Meyer and Wischmeier (1969) suggested that soil erosion by water is a process that can be described by the following separate but interrelated phases: soil detachment by rainfall, transport by rainfall, detachment by runoff, and transport by runoff. Detachment by runoff D F, was assumed to be a function of the tractive force and thus proportional to the average of the initial S 2/3 S Q 2/3 S and final S 2/3 E Q 2/3 E, where S is the bed slope and Q is the water flow rate. The following equation was presented in which S DF is a constant expressing the soil s susceptibility to detachment by runoff as a function of its properties and A I is the area of increment down the slope: 1 23 / 23 / 23 / 23 / DF = SDFAI ( SS QS + SE QE ) 2 [1] Since the detachment is limited by the ability of the flow to carry sediment, a transport capacity term was needed. Transport capacity (T F ) of runoff was assumed proportional to S 5/3 Q 5/3. The constant S TF is a term used to account for the soils transportability as shown in the equation: T = S S Q F TF 53 / 53 / [2] These initial concepts on the processes of soil erosion brought about better ideas towards the development of theoretically based equations that would calculate erosion rates more accurately. As a consequence, Foster et al. (1977) derived a set of equations that would set a standard for our current erosion prediction technology. These erosion equations were based on the concept of dividing the erosion process into rill and interrill

16 6 erosion (Foster and Meyer, 1975). As defined in the introduction, rills are the areas where the flow of runoff concentrates to form small channels. The interrill areas are the areas between such rills. The erosion equations developed by Foster are derived from the continuity equation for sediment transport. It is assumed that a quasi-steady state flow is present: dg dx = D + q [3] r s G = sediment load per unit width (ML -1 T -1 ) x = distance along the channel (L) D r = net detachment rate or net deposition rate by channel flow (ML -2 T -1 ) q s = lateral sediment inflow from adjacent contributing broad shallow flow areas (ML -2 T -1 ) The D r term can either be a net detachment or a net deposition rate. It is believed that both detachment and deposition are always occurring simultaneously, but that net detachment or deposition occur depending on the channel flow conditions. As flow goes down the slope, net detachment can occur. This can be represented by the following equation: D r G = Dc( 1 ) [4] T c T c = transport capacity of the flow (ML -1 T -1 ) D c = detachment capacity of the flow (ML -2 T -1 ) The transport capacity of flow (T c ) is an important term that is used in determining the maximum allowable sediment in the flow for the given hydraulic

17 7 conditions. This term can be computed in many ways and a common way of doing it will be presented later in this section. Detachment capacity is computed by the following equation (Foster et al., 1995): D = K ( τ τ ) [5] c ch c where K ch =channel erodibility parameter (TL -1 ) τ = average shear stress acting on the soil (FL -2 ) τ c = critical shear stress required for detachment to occur (FL -2 ) As seen here, the detachment capacity is dependent on soil erodibility properties and the shear stress acting on the sediment by the flow. This shear stress can be calculated in the following way (Foster et al., 1984): τ = γrs f [6] γ = specific weight of water (FL -3 ) R = hydraulic radius (L) S f = friction slope (L/L) When the transport capacity has been reached for a certain flow, the system is said to be in equilibrium with net detachment being equal to net deposition. As flow conditions change, either detachment or deposition will be favored. The following equation shows how net deposition is calculated (Foster and Meyer, 1972): D = α( T G) [7] r c D r = net deposition rate (ML -2 T -1 )

18 8 α = first-order reaction coefficient (L -1 ) which is computed for a single particle size distribution as (Foster, 1982): β α = V q s [8] β = dimensionless turbulence parameter V s = effective particle fall velocity (LT -1 ) q = Q/w c = flow rate per unit channel width (L 2 T -1 ) w c = channel width (L) The β constant is a gather all parameter that includes the turbulence of the rainfall and other such turbulence and is defined to range from zero for shallow flows and high turbulence to one for deep flows and low turbulence. Its value is usually designated as one for channelized flow, but it is a parameter that could be improved by taking into account different sources of turbulence. One important aspect in the prediction of erosion and deposition is to be able to accurately calculate the sediment transport. To do this, there exist a wide range of sediment transport equations, but only a few can be used for the shallow flows and low flow rates that agricultural erosion prediction requires. An important study that showed the applicability of different transport formulas was conducted by Alonso et al. (1981). Sediment transport predictions of nine different formulas were compared to measurements of flume and field data. The formulas used were the total load formulas of Ackers and White (1973), Engelund and Hansen (1967), Yang (1973), Laursen (1958), and Einstein-Meyer EM-1 and EM-2 (1950); and the bed load formulas of Meyer-Peter and Muller (1948), Bagnold (1956), and Yalin (1963). The study concluded that the formulas that give the best predictions are the Yalin, Yang, and Laursen formulas. The range of applicability of these formulas is shown in Figure 1. As seen in the figure, the

19 9 Figure 1. Applicability of transport formulas (Alonso et al., 1981) best equation for transport in shallow flows seems to be the Yalin formula. This equation has been analyzed by many including Foster and Meyer (1972), and based on its assumptions, derivation, and experiments, it seems to work well for predicting sediment transport in shallow flows. The Yalin equation (Yalin, 1963) is represented by the following set of equations: P = ( W / S ρ dv ) = δ [ 1 ( 1/ σ ) Ln( 1+ σ )] σ = A δ δ = ( Y / Y ) 1 (when Y < Y, δ = 0) A = 2.45S Y 2 Y = V /[( S 1) gd] V * * s g W CR -0.4 g g 05. CR = ( gr s) = ( τ / ρ ) * 12 / 12 / f cs W CR The Yalin Transport Formula [9] P = a nondimensional sediment transport capacity, W s = the transport capacity (mass per unit width per unit time, T c for uniform sediment), S g = particle specific gravity, ρ w = mass density of water, d = diameter, V * = shear velocity, Y CR = ordinate

20 10 from the Shields Diagram, g= acceleration due to gravity, R f =hydraulic radius, s = slope of the energy gradeline, and τ cs = shear stress acting on the soil. Over time, and after detailed studies, suggestions have been made to change a few of the values for the constants. For example, for sand with d = 342µm and S g =2.65, Davis (1978) found that 0.88 was a better value for the constant These sets of equations were incorporated into more process-oriented models to simulate soil erosion, chemical movement, and runoff. One of the models that had a great success was the CREAMS model. It was released in 1980 (Knisel, 1980) to predict chemical runoff and erosion from agricultural fields. Although the model includes many other equations and parameters, the main component of the erosion prediction process is described by the equations presented before. However, USLE parameters were still used extensively in CREAMS. Current models such as WEPP (Flanagan and Nearing, 1995) use process-based erosion prediction technology. Erosion is modeled on a continuous basis with different components estimating the hydrologic, erosion, and related processes. The model utilizes the steady state set of equations developed by Foster (1995). These equations divide erosion into rill and interrill erosion components to predict overall erosion and deposition as seen before. It is worth noting that there have been a variety of approaches to the modeling of soil erosion. One of these approaches has been to use the unit stream power concept instead of the shear stress to calculate sediment detachment (Moore and Burch, 1986). Other studies use a more statistically oriented method to account for variability. Yet another approach uses finite element analysis to study the detachment of soil via the boundary layers of the sediment and the water flow. All these studies contribute in some way to the overall understanding of erosion mechanics. It is also worth mentioning that a variety of other models have been created to simulate erosion. Among the most important empirical models are EPIC and RUSLE. EPIC (Williams et al., 1983) uses a modification of the USLE, and RUSLE (Renard et al., 1991) is a revision of the USLE. Other models include ANSWERS (Beasley et al.,

21 ) and PRORIL (Lewis et al., 1994) which takes a statistical approach to account for the stochastic variability of rill networks. Although these models are important and useful, the approaches and models presented in this section are by far the most studied and used. 2.2.The Effects of Rainfall on Flow Depth Many studies in related areas touch on the subject of rainfall and flow depth effects on sediment transport and deposition, but only a few studies have been made that concentrate wholly on this subject. Experiments conducted in these studies generally demonstrate that high intensities of rainfall on shallow flows increases the sediment transport capacity of the flow. Experiments were conducted by Moss and Green (1983) that indicated that transport rates peak when raindrops impact flows that are between two and three drop diameters deep. It was also found that transport rates tend to vary linearly with flow velocity and that rainfall does not affect sediment transport for slopes greater than 9%, probably due to the high flow velocity at those slope steepnesses (Moss et al. 1979). In a study done by Schmidt (1991) he stressed that the direct impact of rainfall on sediment transport decreases with slope length, because the flow depth increases down the slope. Some studies seem to indicate that there is a definite relationship between rainfall intensity, slope, and sediment transport. These relationships incorporate other effects like flow velocity changing due to rainfall intensity, but they do not take into account the flow depth. As Kinnell (1991) mentioned, the results of many of these experiments may have been misinterpreted because the effects of flow depth on erosion by rain-impacted flow were ignored. In that paper he verifies the work done by Moss and Green (1983) by stating that flow depth has a significant influence on the transport of particles when flow depths are shallower than three drop diameters. He also proposes an analytical theory for the transport of particles by raindrop-induced flow transport (RIFT) to examine the effects of rain, flow, and particle characteristics on the movement of soil material in overland flow. The term raindrop-induced flow transport (RIFT, Kinnell, 1988) was

22 12 introduced to describe the effects of rainfall impacts that induce flow to transport particles which the flow would not be able to by its own means. Although the experiments conducted by Moss and Green (1983 and 1988) and Kinnell (1991) made significant contributions to the understanding of the influences of rainfall on shallow flow, they were directed to simulation conditions of overland flow. Although the results are still valuable for the study of shallow flow in rills, a closer look at the effects of channelized flow on sediment transport is needed. The following section will present different studies actually conducted on rills in the field that show other influences of channelized flow on sediment transport, detachment and deposition. 2.3.Field Studies of Crop-Row Furrows Field experiments are very important in the study of erosion mechanics because they provide a realistic view of what is happening in actual field conditions. A variety of these studies have been conducted on both overland erosion and rill erosion and the most significant ones will be presented here. Meyer and Harmon (1985) conducted several field experiments on a silt loam soil to study the sediment losses from cropland furrows at different gradients. The sediment losses from the bedded rows were evaluated for furrow slopes ranging from 0.5 to 6.5% at three different rainfall intensities (26, 70, and 108 mm/hr) and row lengths. They concluded that soil loss for these furrows is dependent on their gradient and row length. They also stated that for low gradients (0.5 to 5.0%) most of the sediment lost was from the row sideslopes. Increasing the rainfall intensity increased the soil loss. It was observed that an increase of intensity from 26 to 70 mm/hr caused a 5 fold increase in soil loss and an increase of intensity from 70 to 108 mm/hr resulted in a doubling of soil loss. A field study was conducted by Flanagan (1989) in which sediment transport and deposition in a rill were studied and numerical relationships and equations were established based on the results of the experiments. This study provided important information on transport relationships in actual field situations. It was shown that soil

23 13 texture and sediment characteristics had a large effect on sediment transport. The application of rainfall also appeared to significantly increase the sediment transport for shallow flows. Finally it was recommended that future studies be made to improve our understanding of these relationships and to validate the relationships by means of a laboratory study in which parameters could be better controlled. A suggestion was also made to improve the current WEPP deposition equation by defining a functional relationship between rainfall intensity, flow depth, and the rainfall turbulence factor. 2.4.Laboratory Rill Erosion Mechanics Studies Laboratory studies are necessary to study the different parameters that affect soil erosion, due to the fact that they enable the scientist to control influential factors very closely. It is for this reason that a variety of studies have been conducted to simulate rills in laboratory conditions. The most important and significant ones will be described in this section. Meyer et al. (1983) conducted a series of experiments to measure the transport capacity of sands along crop-row furrows. These experiments were conducted on a noninfiltrating parabolic flume 1.8m long with a w=3600d, width to depth ratio. The flume was set to the desired furrow grade and flow rate while sediment was added until the transport capacity was reached. Combinations of four furrow gradients, four flow rates, and four particle diameter groups were tested with (122 mm/hr) and without rainfall. The water inflow was reduced by the same amount as the rainfall addition so that the water outflow would be equal to the runs without rainfall. Partial data results, presented by Meyer et al. (1983), are shown in Table 1. Based on the study of the principal independent variables, channel gradient (S) and water discharge (Q), and the secondary independent variable, particle diameter (D), and the boundary conditions, gravity (g) and fluid kinematic viscosity (υ), the following general nondimensional equation was presented:

24 14 Qs Q = f S Dg 13 /, 23 / υ [10] Q s is defined as the sediment transport capacity for runoff in rills without the influence of rainfall. Table 1. Sediment Transport Capacities (g/min) for Sands.(Meyer et al. 1983) Rainfall rate is 122mm/hr. flow rate Furrow Grade D= µm (g/min) D= µm (g/min) D= µm (g/min) (kg/min) (%) Rain No Rain Rain No Rain Rain No Rain The study concluded that furrow gradient was a very influential factor on sediment transport. As furrow gradient increased, sediment transport capacity increased rapidly. To a lesser extent, flow rate and particle size were also influential to the sediment transport capacity. Transport capacity increases as particle size decreases and as flow rate increases. Meyer et al. (1983) reported flow depths between 5 and 15 mm, and that the rainfall seemed to have a positive impact on transport for smaller particles and a negative impact on transport for larger particles. It was also mentioned that there could be a transition zone in which rainfall goes from enhancing the transport, by the introduction of turbulence, to rainfall inhibiting the flow, resulting in a reduction of sediment transport. A study of the magnitude and range of velocities and shear stresses that occur in rills was studied by Foster et al. (1984). This study was conducted using a full scale

25 15 fiberglass fixed bed, which replicated a rill on an eroded soil. It was concluded that the flow characteristics along the rill are very nonuniform. Flow velocity along the rill varied as a normal distribution. Mean velocity was affected slightly by rainfall. The shear stresses in the rill were estimated by the use of a hot-film sensor and anemometer. It was believed that rill form roughness had an overall greater influence on shear stress than grain roughness although there was a lot of nonuniformity. Results also indicated that for larger slopes this difference decreased significantly. In a study conducted by Cassol (1988) it was shown that particle density was a dominant variable for the segregation of particles during deposition. His study was conducted on a concave shaped 0.5m by 3m cross-sectional bed in which sediment entering, leaving and deposited on the bed were collected and analyzed. Rainfall was also applied, but for sand particles the effects were undetectable for the rainfall intensity of 50 mm/hr. An effort to study the influences of rainfall on sediment transport in rills was made by Khan (1989). He conducted an indoor lab study to simulate the sediment transport in a ridge furrow system in which he used a non-infiltrating parabolic flume. In this study he also came to the conclusion that in general, sediment transport capacity increased with an increase of rainfall on the flume bed, but since he used a noninfiltrating bed, the increase in rainfall intensity resulted in an increased flow depth and shear stress which by theory results in a larger transport capacity due to flow. He also pointed out that sediment concentrated at the center of the flume and moved predominantly as bed load. It was suggested that future studies be made in which sediment is fed from the sides along the flume bed to better simulate the natural ridge furrow system sediment transport process. His experiments showed an exceptional relationship between dimensionless transport capacity and shear stress and that transport capacity is related to stream power.

26 Literature Review Summary The studies and research mentioned above lead to the following summary of statements that describe how certain parameters influence rill transport, detachment, and deposition of sands: a) Slope factor was definitely seen as having a major effect on erosion, as slope increases, sediment transport capacity increases (Meyer et al. 1983). b) Flow has a great influence on erosion, but the velocity of the flow is somewhat more important than the quantity (Foster et al. 1984). c) Flow characteristics such as velocity and shear stress are usually nonuniform throughout natural rills (Foster et al. 1984). d) As slope length increases during rain, flow and depth of flow increase and the effect of rain intensity decreases (Schmidt 1991). e) Mean velocity is slightly affected by rainfall intensity (Foster et al. 1984). f) Form roughness of sands has a greater effect on shear stress than grain roughness for slopes less than 6% (Foster et al. 1984). g) Rainfall influences transport if flow depth is less than 3 raindrop diameter sizes and slope is less than 10% (Moss and Green 1983, and Kinnell 1991). h) For sands, bed load movement predominates (Khan 1989). i) Particle density and shape have a large effect on transportability (Cassol 1988). These statements and others, along with experimental values, are used to build or validate equations that can be used for modeling erosion. As seen from the research and models mentioned before, there has been significant progress in the study of erosion mechanics, but our understanding and our ability to predict erosion rates still needs improvement. A few of the aspects requiring further study are the following: 1. The influence of rainfall intensity and shallow flow depth in rills on the sediment transport capacity of the flow.

27 17 2. The effects of infiltration rates on detachment, transport, and deposition of sediment in the flow. 3. The effects of shallow flow and rainfall on the form roughness of sand in a rill. 4. The influence of sediment and flow from the side slopes of a furrow on sediment transport, detachment and deposition in the rill. Research in these topics will further enhance our knowledge of erosion mechanics and will open up the way to new questions about the transport, detachment, and deposition of sediment. 2.6.Objectives of Study as Related to Literature Review Based on the previously mentioned studies and recommendations from those studies, a set of objectives for this study is described below: 1. Create a flexible laboratory experimental setup in which to study the effects of water flow, rainfall intensity, and infiltration on detachment, transport, and deposition of sands (as well as other types of sediment in future studies). 2. Test the validity of Equation 4, more specifically the (1-G/T c ) term. This term states that erosion should decrease down slope as sediment is incorporated into the flow until transport capacity is reached and there is no more net detachment. 3. Observe and report the effects of rainfall and infiltration on deposition of sand sediment. It is believed that rainfall will increase turbulence which in turn will increase the transport capacity of the flow resulting in lower net deposition. Infiltration could have two effects on the sediment transport: a) the matric suction forces during infiltration could tend to strengthen the bed surface, thus changing the force needed to detach the sediment or b) it could change the flow characteristics by changing the turbulence in the flow. The effects of infiltration in shallow flows could have an important effect on sediment transport.

28 18 4. Test the validity of Equation 7, more specifically the relationships between transport capacity and deposition. New values for the β parameter in Equation 8 would be calculated based on rainfall and flow depth effects. 5. Suggest ways to improve the experimental setup and procedures for further studies on sediment transport in rills.

29 19 CHAPTER 3. EQUIPMENT DESIGN AND MATERIALS 3.1. The Rill Simulator Flume The main purpose of constructing the flume was to simulate as closely as possible the sediment transport, deposition and detachment that occurs in a rill after or during a rainfall event. In order to accomplish this, the following design variables and restraints had to be taken into account a) The rill width had to be adjustable. Experiments could then be conducted with different rill widths. b) The slope of the rill should be variable from 0 to 20 %. c) Infiltration through the bed should be controllable. d) Sediment and water mixture inflow to the top of the rill should be controllable. e) The flume should have the option of introducing sediment from the sides of the rill. f) The end of the rill should have a collection box and an endplate which can be adjustable to different heights. g) The length of the rill should be at least 3 meters. Following these specifications a flume was built which has been named the rill simulator.

30 Main structure of the flume The main frame was built with 3 cm steel angle iron welded together to form a long trapezoidal shaped structure which is held up by two stands, near each end (272 cm apart). The front stand is used as the pivot point of the whole frame and it elevates the frame to a fixed height of 75 cm above the ground. The back stand is a setup of two worm gear jacks that can be raised to change the slope of the rill. These jacks rise a total height of 55 cm giving the rill a slope range from 0 to 20 percent. Figure 2 depicts the main frame with the stands. The actual rill was built with six 3.66 meter long pieces of galvanized sheet metal (two 366 cm by 35 cm pieces and four 366 cm by 20 cm pieces) attached parallel to one another by long hinges. These are represented with dashed lines in Figure 2 and a full view is shown in Figure 3. The two wider pieces (366 cm by 35 cm) are used for the side slopes of the rill. This allows for the introduction of sediment Pivot Point Worm Gear Jacks Figure 2. Main frame with stands from the side of the rill at different angles. The second and fifth pieces are held vertical and parallel to each other with one sheet welded to the frame while the other is held by a retracting mechanism used to change the rill width. This setup allows for the change in rill width as well as the possibility to run experiments with either rectangular channel rills or with trapezoidal shaped rills. A picture of the entire setup is shown in Figure 4.

31 21 Figure 3. Galvanized sheet metal forming rill Figure 4. Rill simulator

32 Infiltration A very important aspect of the rill simulator is its ability to simulate water infiltration through the sediment. This was achieved by drilling 34 evenly spaced one centimeter diameter holes through the two center bottom pieces of galvanized sheet metal. Plastic nozzles (0.64 cm diameter) were placed in these holes and a porous sponge material was placed in them to prevent sand from flowing out. A network of 1.27 cm diameter polyurethane clear tubing was fitted to the nozzles which converged to a single outlet. By changing the height of the outlet, infiltration rates can be controlled; altering the height of the infiltration outlet causes changes in water tension which drains water from the bed at desired rates. The lower the outlet is placed, the higher the water tension and the higher the infiltration rate, and vice versa Sediment feeder An old sediment feeder used in previous experiments (Davis, 1978) was modified and adapted for use as the main source of sediment input to the rill simulator. The main modification was the construction of a water inlet trough and a smooth tray underneath the sediment feeder conveyor belt. With these new additions, water in the water inlet trough overflows over the edge onto the tray creating a uniform sheet flow of water onto which dry sediment is introduced. The dry sediment gets mixed with the flowing water and is directed to a water shoot and then into a box which routes the water and sediment mixture to the top of the rill. The rate of sediment feeding is controlled by changing chain sprockets which adjust the speed of the conveyor belt. Two sprockets were used for the study, one delivering 5.3 g/s and the other one delivering g/s of sediment. 3.2.Sediment The sediment chosen for all the experimental runs was a fine sand processed by the U.S. Silica Company. This sand comes in kg (100 lb) bags and is identified as Crystal Silica Sand. The screen analysis and physical properties are shown in Table 2.

33 23 Additional bulk density measurements were taken using a laser scanner, which will be described later, as the measuring tool. These measurements were done by scanning a 15 cm 2 surface of flat sand and scanning it again after a certain volume of sand was removed. The sand was weighed and the volume was computed by subtracting the initial laser scan from the final laser scan. (A simple program written in Q-Basic was used.) As seen in Table 3, the measurements were consistent with the product information. The sand used was selected for the following reasons: a) The sand is readily available in most hardware shops and the price per bag is low. b) The particle size and other properties of the sand such as infiltration and bulk density were ideal to test out the equipment and approximate a fine sandy soil. Table 2. Crystal Sand Screen Analysis (Ottawa Foundry Sand, U.S. Silica Co.). Sieve Size Particle size (µm) Factory measurements NSERL measurements <1 < < <1 2.6 Average size = 326 µm Table 3. Crystal Sand Physical Properties. Product Information Lab measurements Grain Shape rounded Spec. Gravity 2.65 Base Perm. 290 Bulk Density (Comp.) g/cm 3 saturated sand in bed 1.61 g/cm 3 Bulk Density (Loose) g/cm 3 dry sand in bed 1.58 g/cm 3

34 Data Acquisition System A data acquisition system was designed and built to gather and collect data from the different sensors and controllers on the rill simulator. The system is composed of a data acquisition board installed in a computer that is linked to two pressure sensors, two flow sensors, two motor controllers, and a profile meter. Appendix A contains the detailed design and descriptions of the data acquisition electronic connections Data acquisition board The major component in the data acquisition system is the data acquisition board. The critical specifications for choosing a data acquisition board were its processing speed and the number of digital and analog channels available for data input and output. Keeping this in mind a suitable data acquisition board was found that met and exceeded all the requirements. This board was manufactured by Computer Boards, Inc. and has the properties given in Table 4. Table 4. Data Acquisition Board Properties. Name: CIO-AD16 Analog Input Channels: 8 differential or 16 single ended Resolution: 12 bits, 1 part in 4095 Accuracy: 0.01% of reading, ± 1 bit Acquisition time: 1µSec to 1% of full scale step Dig. to Analog Converters: Two channels. Counters: 3, 16 bit down counters Digital Input/Output: 24 bits, three 8 bit ports Current settings 8 differential input channels Bipolar current of +/- 10 VDC (+/- 2048bits) 4.88bits/milliVolt resolution (4095bits max)

35 25 The board was mounted in a Mhz Tandy (Radio Shack) PC and 37 pin extension cables were extended out to Analog and Digital connector boards. A diagram of the data acquisition system is shown in Figure 5. Figure 5. Data acquisition system

36 26 The data acquisition system is controlled by a program written in QBasic utilizing the Call subroutines provided by the board manufacturer. This program converts the readings from all the sensors to physical values, keeps track of time, controls the movement of the LVDT motor and the end plate motor, and handles all the required analog to digital conversions. A copy of this program is presented in Appendix B Pressure sensors Two water pressure sensors were placed along the flume 120 cm apart and 20 cm below the edge of the sides of the flume. The purpose of these sensors was to test the possibility of measuring water height when no infiltration is present though the sand and to identify infiltration trends when infiltration is occurring. The pressure sensors had to be carefully chosen to meet the requirement of measuring small changes in water height (1 mm changes), and to be compatible with the data acquisition board (have an output voltage range in between +/- 10V). The sensor characteristics are described in Table 5. Table 5. Pressure Sensors. Manufacturer: Kavlico Corporation Type: Model P592-5-D-A-1-A 0-13 cm H 2 O pressure range 5 VDC input voltage 0-4 VDC output voltage 0.5% accuracy Flow sensors Flow sensors were used to measure the water inflow to the top of the flume and to measure water infiltrating out of the bed. For the water inflow, a paddlewheel flow sensor was chosen due to its ability to measure large quantities of water flow. This sensor was installed underneath the water trough of the sediment feeder and measured all the water that entered the flume. The sensor characteristics are described in Table 6.