Department of Transportation Office of Materials and Research GDOT Research Project No. 0404 FINAL REPORT Estimating Critical Shear Stress of Bed Sediment for Improved Prediction of Bridge Contraction Scour in Georgia Submitted by Terry W. Sturm, Seung Ho Hong, and Paul Hobson School of Civil and Environmental Engineering Georgia Institute of Technology Atlanta, GA 30332 December 2008 1.Report No.: FHWA-GA-08-0617 2. Government Accession No.: 3. Recipient's Catalog No.: 4. Title and Subtitle: Estimating Critical Shear Stress of Bed Sediment for Improved Prediction of Bridge Contraction Scour in Georgia 5. Report Date: December 2008 6. Performing Organization Code: 7. Author(s): Terry W. Sturm, Ph.D. 9. Performing Organization Name and Address: Georgia Institute of Technology School of Civil & Environmental Engineering Atlanta, GA 30332-0355 12. Sponsoring Agency Name and Address: Georgia Department of Transportation Office of Materials & Research 15 Kennedy Drive Forest Park, GA 30297-2534 8. Performing Organ. Report No.: 10. Work Unit No.: 11. Contract or Grant No.: SPR00-0007-00-073 13. Type of Report and Period Covered: Final; July 2006 December 2008 14. Sponsoring Agency Code: 15. Supplementary Notes: Prepared in cooperation with the U.S. Department of Transportation, Federal Highway Administration. 16. Abstract: The influence of sediment properties and the flow field on bridge contraction scour are investigated in this study. Sediment properties of samples obtained from Shelby tube cores at 5 bridge sites in Georgia are tested using an erosion flume and a rheometer. Relationships for the critical shear stress are developed from this data to be used in contraction scour formulas. The effect of the flow field on contraction scour is investigated by conducting a physical model study of the SR 42 bridge over the Towaliga River in central Georgia. Physical model results for contraction scour depths are shown to match field measurements rather well. Based on this favorable comparison, conclusions are drawn concerning the interaction of local scour and contraction scour with suggestions for further research on scour near an abutment under submerged bridge flow conditions. 17. Key Words: bridge, sediment, scour, shear stress, contraction 18. Distribution Statement: 19. Security Classification (of this report): Unclassified 20. Security Classification (of this page): Unclassified 21. Number of Pages: 61 22. Price: i ACKNOWLEDGMENTS The authors gratefully acknowledge the expert assistance and encouragement provided by David Jared, Special Research Engineer; Susan Beck, Bridge Design Group Leader (Hydraulics); and Tom Scruggs, Assistant State Materials and Research Engineer (Geotechnical Bureau), Georgia DOT. The project was made possible through the support of GDOT officials Georgene Geary, State Materials and Research Engineer; Rick Deaver, Chief of Research and Development; and Paul Liles, State Bridge and Structural Design Engineer. The able assistance of Andy Udell, Laboratory Manager in the School of CEE at Georgia Tech, in support of the laboratory experimental setup and instrumentation was invaluable in completion of the project. River cross-section surveys were expertly performed by Georgia Tech students Ben Allread and Jeremiah Kimbell. ii EXECUTIVE SUMMARY Previous GDOT research by the principal investigator has resulted in an approximate method for estimating critical shear stress for initiation of motion of river bed sediment to be used in bridge scour prediction methods applied to Georgia bridges. In this study, Shelby tube samples were taken at five additional bridge sites and tested for critical shear stress in an erosion flume. The results validated and expanded the applicability of the previous relationship given in terms of a Shields parameter for critical shear stress as a function of a dimensionless median particle diameter with percent fines by weight as a third parameter. In addition, the fine fraction of the Shelby tube samples was separated out for determination of the yield stress in a rheometer. The yield stress followed a relationship similar to that for the critical shear stress obtained from the erosion flume, and reflected the influence of interparticle forces on the erodibility of the fine material. A relationship between critical shear stress and yield stress was obtained indirectly; however, additional testing is needed to verify it. The new method for estimation of critical shear stress can be applied in contraction scour formulas, but the other part of the problem is predicting the flow field and its interaction with the bed sediment as a result of the bridge flow contraction. In order to attack this aspect of the problem, a physical model study of the Towaliga River bridge at SR 42 near Forsyth, GA was conducted. This bridge experienced significant contraction scour during Tropical Storm Alberto in 1994 as did some 125 other bridges that were damaged by the flood. The bridge and river bathymetry were modeled at a 1:60 scale with a sediment size chosen according to a methodology developed in previous research by the principal investigator. The results showed excellent agreement with field observations of scour that were made during and immediately after the flood for a discharge reached just prior to overtopping of the bridge. Separation of the measured scour into local pier scour and contraction scour components followed by comparison of the results with predictions from accepted scour formulas exposed the weakness of those formulas. The arbitrary assumption of adding scour components as though they were independent processes resulted in an overprediction of scour and an oversimplification of the complex spatial distribution of the scour components, particularly near the toe of the abutment where abutment scour, pier scour, and contraction scour occurred simultaneously. Some suggestions are made for improving the existing idealized contraction scour formula by conducting systematic experiments on contractions of various degrees, including both vertical contraction due to pressure flow as well as lateral contraction due to embankment blockage of floodplain flow. iii TABLE OF CONTENTS ABSTRACT ACKNOWLEDGMENTS EXECUTIVE SUMMARY LIST OF FIGURES LIST OF TABLES CHAPTER 1. INTRODUCTION BACKGROUND PREVIOUS RESEARCH RESEARCH OBJECTIVES RESEARCH APPROACH ORGANIZATION OF THE REPORT CHAPTER 2. INVESTIGATION OF CRITICAL SHEAR STRESS AND YIELD STRESS INTRODUCTION SHELBY TUBE SAMPLE LOCATIONS EXPERIMENTAL METHODS Erosion flume Rheometer RESULTS ANALYSIS ERODIBILITY CLASSIFICATION OF GEORGIA SOILS RECOMMENDED FORMULA FOR CRITICAL SHEAR STRESS CHAPTER 3. PHYSICAL MODEL STUDY OF TOWALIGA BRIDGE INTRODUCTION SITE SELECTION EXPERIMENTAL TECHNIQUES EXPERIMENTAL PROGRAM EXPERIMENTAL RESULTS DISCUSSION OF RESULTS CHAPTER 4. SUMMARY AND CONCLUSIONS SUMMARY CONCLUSIONS RECOMMENDATIONS FOR FUTURE RESEARCH REFERENCES APPENDIX A. SEDIMENT CORE TEST RESULTS iv Page No. i ii iii v vii 1 1 2 3 3 5 6 6 6 7 7 10 12 13 19 20 22 22 22 25 27 29 46 50 50 51 53 54 56 LIST OF FIGURES Figure Page No. 2.1. Previous and current sample locations and Georgia physiography (Alhadeff et al., 2000) 8 2.2. Erosion flume (Hobson 2008) 9 2.3. Stress controlled rheometer apparatus 11 2.4. Determining the lower yield stress from a stress-strain flow curve 11 2.5. Determining the upper yield stress from a strain rate-stress flow curve 12 2.6. Shields diagram with natural sediments from previous study and the present study 14 2.7. Measured vs. predicted Shields parameter (Eq. 2.1) for previous flume data and current flume data 15 2.8. Measured vs. predicted Shields parameter by Eq. 2.2 for previous flume data and current flume data with regression applied to combined data set (Error Limits =0.5 - 2.0 times predicted value). 16 2.9. Measured vs. predicted critical shear stress (Eq. 2.2) for combined data set 16 2.10. Critical Shields parameter as a function of Fines content applied to both data sets 17 2.11. Critical Shields parameter for upper and lower yield stress as a function of d* 19 2.12. Erodibility classification of Georgia soil samples 20 3.1. Towaliga R. Bridge at SR 42 crossing 24 3.2. Towaliga River rating curve at SR 42 and extension 25 3.3. Location of stations for scour time monitoring and approach velocity measurement 29 3.4. Approach velocity profiles 30 3.5. Time history of scour development for Run 4 31 3.6. Time history of scour development for Run 5 32 3.7. Time history of scour development for Run 6 32 3.8. Run 4 Cross section comparison 35 3.9. Run 5 Cross section comparison 36 v Figure 3.10. Run 6 Cross section comparison 3.11. Bridge cross-section comparison after scour 3.12. Bed contours before and after scour for Run 4 3.13. Photographs of bed after scour for Run 4 3.14. Bed contours before and after scour for Run 5 3.15. Photographs of bed after scour for Run 5 3.16. Bed contours before and after scour for Run 6 3.17. Photographs of bed after scour for Run 6 3.18. Velocity vectors at 40 percent of the depth for Runs 5 and 6 Page No. 36 37 38 39 41 42 43 44 45 vi LIST OF TABLES Table Page No. 2-1. Candidate Shelby tube sampling sites 7 2-2. Critical shear stress and erodibility classification of various sand sizes (after Julien 1995) 20 3-1. Possible sites for physical model 23 3-2. Final bridge model selection 24 3-3. Experimental conditions 28 3-4. Experiment results 34 3-5. Comparison of predicted and measured scour depths in left floodplain for Run 6 48 A.1. Sediment Core Results for Meriwether County Samples 57 A.2. Sediment Core Results for Monroe County Samples 58 A.3. Sediment Core Results for Houston County Samples 59 A.4. Sediment Core Results for Stewart County Samples 60 A.5. Sediment Core Results for Candler County Samples 61 vii CHAPTER 1 INTRODUCTION BACKGROUND Bridge scour is a significant transportation problem because of the structural damage, economic disruption and possible loss of life that can occur when it results in bridge foundation failure (Stamey 1996, Parola et al. 1998, Richardson and Davis 2001, Morris and Pagan-Ortiz 1999, Wardhana and Hadipriono 2003). The damage caused by tropical storm Alberto in Georgia in 1994 is a case in point. Tropical storm Alberto dumped as much as 28 in. of rainfall in parts of central and southwest Georgia from July 3-7, 1994, and caused numerous bridge failures and highway closings as a result of the 100-yr flood stage being exceeded at many locations along the Flint and Ocmulgee Rivers. Approximately 18 ft of foundation scour occurred at the U.S. Highway 82 crossing of the Flint River near Albany, Georgia (Stamey 1996). Total damage to the Georgia Department of Transportation highway system was approximately 130 million dollars (Richardson and Davis 2001). Prevention of bridge scour damages and possible loss of life hinges on having the capability of predicting expected bridge scour. Unfortunately, such predictions remain a challenging problem because of the complex interaction of the river flow with the obstruction presented by the bridge foundation and with the erodible bed of the river. As the approach flow encounters a bridge foundation it separates and rolls, in reponse to the foundation-induced adverse pressure gradient, to form multiple large-scale vortices whose axes may be parallel (horseshoe vortices) or perpendicular (tornado-type vortices) to the bed. These unsteady, largescale flow structures in conjunction with a broad range of turbulent scales control the sediment transport in the vicinity of the foundation and are responsible for the growth of scour holes (Raudkivi 1986, Dargahi 1989, Melville 1997, Melville and Coleman 2000). The complexity of 1 such local scour-inducing flow phenomena has hampered satisfactory analysis and prediction of scour in the past. In addition, contraction scour may occur simultaneously with local scour within the bridge opening (Mueller and Wagner 2005). Current FHWA procedures require separate estimates of local scour and contraction scour as though they were independent processes, but they may interact so that this addition of the two types of scour may be overly conservative. The other side of the scour problem is quantifying the resistance to scour offered by the erodible bed of the river. A common measure of the erosion resistance of the soil is the critical shear stress that is just sufficient to cause particle movement. The critical shear stress can also be used to calculate a critical flow velocity using Keulegan's equation for fully-rough turbulent flow (Sturm 2001). While the Shields diagram is well known as a method of predicting the critical shear stress of noncohesive sediments (sand and gravel), the presence of fine-grained sediments (silt and clay) introduces interparticle physicochemical forces that are difficult to quantify. These forces depend on the type of clay, the porewater chemistry, and the bulk density of the sediment, for example, all of which can be quite variable (Ravisangar et al. 2005). PREVIOUS RESEARCH Previous GDOT research by the principal investigator has resulted in an approximate method for estimating critical shear stress for initiation of motion of river bed sediment to be used in bridge pier scour prediction methods applied to Georgia bridges (Sturm et al. 2004). A related research need is to improve bridge contraction scour estimation, which depends not only on characterizing critical shear stress properties of the sediment but also on predicting the flow field and its interaction with the bed sediment as a result of the flow contraction caused by the bridge opening (Sturm et al. 2006, Hong et al. 2006). Contraction scour was the dominant failure mechanism for some 125 bridges that had to be replaced in Georgia after Tropical Storm Alberto in 1994. Critical shear stress of sediment was investigated in the previous study using Shelby tube cores that were taken by GDOT at 10 bridge sites spread out among the physiographic regions of 2 Georgia with the purpose of developing regional sediment properties. These samples were tested in a specially-designed flume at Georgia Tech to measure the critical shear stress of the sediment, and the results were correlated with a wide range of sediment properties. It was found that a firstorder estimate of critical shear stress could be made as a function of the dimensionless median particle size and the percent by weight of fine sediment in the sample as a surrogate for the magnitude of the effect of interparticle electrochemical forces. Somewhat unexpected was the degree of variability in the values of critical shear stress even within the same sediment core that made establishing a regional basis for critical shear stress estimation rather difficult except in qualitative terms. RESEARCH OBJECTIVE It is the purpose of the research described in this report to further refine the procedure for estimating critical shear stress and predicting scour depths at bridge foundations, especially for smaller bridges that were found to be vulnerable to failure by combined contraction and local scour during Tropical Storm Alberto in Georgia in 1994. Such bridges are too small to instrument for continuous scour monitoring and too expensive to model on a case-by-case basis in the laboratory using the techniques developed in the previous GDOT research (Sturm et al. 2006) because of their relatively large number. The present research seeks to both improve the predictive relationship for critical shear stress in terms of easily measured sediment properties and apply it in a laboratory model of contraction scour so that a procedure for estimating contraction scour in the presence of local scour can be developed. RESEARCH APPROACH The research approach consisted of the collection of 10 additional sediment cores by GDOT at five bridges that were constructed over smaller streams than in the previous study, at least some of which had to be repaired or replaced because of tropical storm Alberto. The sediment samples 3 were tested in the erosion flume at Georgia Tech with additional testing of the erosion properties of the fine sediment fraction. The fine sediment fraction was separated from a sampled portion of the sediment cores tested in the flume, and then tested in a rheometer to determine their yield stress. The rotational rheometer applies a continuously increasing stress to the sediment in an annular space and measures the strain rate from which the yield stress can be determined as the value at which the sample begins to "flow". The principal investigator has shown in previous research that there is a close correlation between yield stress and critical shear stress of kaolinite clay, for example (Ravisangar et al. 2005). This additional information on erodibility of the fine sediment fraction of the sediment cores tested in the flume has provided further index data to help explain the role of the fine-grained fraction of the sediment in contributing to the critical shear stress estimates. In effect, the variability in critical shear stress due to clay mineralogy and resultant particle structure that may be more regional in nature can be explained by the rheological behavior of the fine-grained fraction of the sediment. In addition, the expanded database of additional sediment core samples tested in the erosion flume provided an opportunity to validate and extend the previous relationship for estimating critical shear stress. In the second part of the research, a laboratory model of a typical small bridge that failed during tropical storm Alberto was constructed and tested as a forensic validation case. Combining the modeling techniques developed in previous research led by the principal investigator (Lee et al. 2004, Hong et al. 2006) and the improved critical shear stress relationship from the present research, the contraction scour failure conditions were reproduced successfully in the laboratory model. The results are used to validate and further improve scour predictions for the special case of smaller bridges with foundations in sediments having a significant percentage of fine-grained particles. Ultimately, the result will be a bridge scour estimate that is more realistic and a bridge foundation design that is more economical for both new bridges and bridges that may have to be replaced in the future in Georgia. 4 ORGANIZATION OF THE REPORT In Chapter 2, the collection of Shelby tube cores at 5 bridge sites in Georgia is described. The analysis of the soil properties, including rheologic properties, leads to a proposed relationship for predicting critical shear stress, which is a parameter required to estimate clear-water bridge contraction scour. Chapter 3 is devoted to reporting the results of a physical model study of floodplain contraction scour at a bridge over the Towaliga R. in central Georgia. A scour estimation procedure using the relationship for critical shear stress and the results of the physical model study is suggested. Conclusions are given in Chapter 4 along with suggestions for unknotting the intricate intertwining of contraction scour, abutment scour, and pier scour near a bridge abutment in the presence of submerged bridge flow conditions, even when a riprap apron is used as a scour countermeasure. 5 CHAPTER 2 INVESTIGATION OF CRITICAL SHEAR STRESS AND YIELD STRESS INTRODUCTION Shelby tube core samples were collected by GDOT from five bridge sites in middle and south Georgia. These samples were tested in the Georgia Tech erosion flume to determine critical shear stress. In addition, several other soil properties were determined for each soil sample including liquid limit, plastic limit, particle size distribution, organic content, specific gravity, and bulk density. For those samples consisting of a significant fraction of fine-grained particles, the fine fraction was separated and tested for yield stress in a rheometer. SHELBY TUBE SAMPLE LOCATIONS Candidate Shelby tube sample sites were selected based on: (1) old boring logs showing blow counts that allowed a sample to be collected; (2) geographic distribution to increase coverage of some physiographic regions not well represented in previous research; (3) ease of getting drilling rig in place; and (4) inclusion of the chosen bridge modeling site. The candidate sampling sites are listed in Table 2-1, and the final five sites that were selected are marked with an asterisk. The Shelby tube sample locations are shown in Figure 2.1. The black circles represent the sites selected in this study (identified by numbers in Table 2-1) while the white circles indicate the samples previously tested by Navarro (2004) as detailed in Sturm et al. (2004). Two of the sample locations in this study were in the Southern Piedmont, one in the Sea Island physiographic province, and two in the Sea Island/East Gulf Coastal Plain. Two to three tubes were taken at each sampling location. 6 Table 2-1. Candidate Shelby tube sampling sites. Site No. in Stream Road Fig. 2.1 Crossing Tobesofkee Cr. SR 74 Towaliga R. SR 87 South Fork, Broad River SR 8 Yellow Jacket Cr. SR 54 15 Canoochee R.* SR 23 14 Hodchodkee Cr* US 27 12 Towaliga R.* SR 42 11 Red Oak Cr.* SR 362 13 Bay Cr.* West Perry Pkwy. *selected site County Bibb Monroe Madison Troup Candler Stewart Monroe Meriwether Houston EXPERIMENTAL METHODS Erosion Flume The flume used for erosion testing of the Shelby tube samples is shown in Figure 2.2. The Shelby tube sample is placed below the bed of the flume, and the soil is extruded into the flume by a piston while water is flowing. The flume has been calibrated so that a known shear stress can be applied to the sample by adjusting the flow rate, flume slope, and flow depth. For a given shear stress, the sample is gradually extruded while it erodes so that the erosion surface remains level with the flume bed. A cable-pull potentiometer is used to measure the piston position as a function of time from which an erosion rate can be determined after the dry density of the sample is measured. Several pairs of shear stress and erosion rate data are collected in this way by increasing the shear stress. From the plot of erosion rate vs. shear stress, the data are fit by a straight line that is extended to zero erosion rate. The shear stress intercept is then the critical shear stress for the sample. Additional details of the experimental procedure are given by Navarro (2004) and Hobson (2008). 7 11 12 13 14 15 Navarro (2004) Hobson (2008) Sample 1 2 3 4 5 6 7 8 9 10 County Murray Towns Habersham Haralson Wilkinson Bibb Effingham Decatur Berrien McIntosh Sample 11 12 13 14 15 County Meriwether Monroe Houston Stewart Candler Figure 2.1. Previous and current sample locations and Georgia physiography (Alhadeff et al., 2000). 8 Figure 2.2. Erosion Testing Flume (Hobson 2008). All samples were characterized by conventional geotechnical tests with material recovered directly from the Shelby tube. Additional material was sampled when there was a 9 significant change in the erodibility or appearance of the material during an experimental run. Soil is a natural material; thus, heterogeneity was expected and found in nearly all samples. The geotechnical tests included the following: dry and bulk densities, water content, specific gravity of soil grains, organic matter content, Atterberg Limits, and grain size analysis by sieve and hydrometer. The activity index of each sample, defined as the plasticity index divided by the percent clay, was calculated from the measurements to suggest the type of clay that might be present. Rheometer The yield stress experiments were performed with a Haake RheoStress RS65 stress controlled rheometer. A rheometer is a device designed to measure the viscosity and/or yield stress of a fluid, or sediment slurry in this case, as a result of appplying an increasing shear stress with time. The instrument consists of a cup and a concentric cylinder that is submerged and then rotated in the fluid. The test is used to as a means of directly estimating the strength of the interparticle forces of the fine fraction of a sediment. The experimental setup is shown in Figure 2.3. The fine sediment samples were mixed with varying amounts of water and allowed to settle for 24 hours in the rheometer cup before testing (Hoepner 2000). Settling characteristics of each prepared sample were observed. Based on the clarity of the sediment-water interface, the slurry could be classified as flocculated or dispersed following Ravisangar et al. (2005). If the settling interface was clear, it was classified as a flocculated suspension, whereas a muddy interface indicated a disperse suspension. Yield stresses of the reconstituted and settled fine sediment samples were determined as a function of prepared water content. Then, characteristic values of the yield stress were estimated for each Shelby tube sample depth from a best fit of the data at a common water content, w= 150%, to provide a relative index for comparison. In several cases, insufficient fine sediment was present to conduct rheometer tests. 10 Figure 2.3. Stress controlled rheometer apparatus. In a previous study of estuary mud, Hoepner (2000) programmed the rheometer to logarithmically increase the shear stress from 0.04 Pa to 12 Pa over a period of 300 seconds. The same logarithmic rate was used in this study. The lower yield stress (y1) is determined from the stress-strain flow curve by fitting power curves to the initial, low-strain section and also to the brief transition to the high strain portion of the flow curve. The intersection of these two curves is defined as the lower yield stress, as shown in Figure 2.4. Figure 2.4. Determining the lower yield stress from a stress-strain flow curve. 11 The upper yield stress, y2 , as defined by Otsubo and Muraoka (1988), is similarly determined by fitting straight lines to the initial steep and final flatter parts of the stress vs. strain-rate flow curves, rather than the stress-strain curves. This procedure is illustrated in Figure 2.5. Figure 2.5. Determining the upper yield stress from a strain rate-stress flow curve RESULTS Complete results for geotechnical properties and the measured critical stresses and yield stresses for all five bridge sites are given in Appendix A. Samples taken at Red Oak Creek from 7 to 12 ft in depth were sandy silty clay with critical shear stresses in excess of 21 Pa (1 Pa = 0.0209 lb/ft2), which is the limit that can be measured in the flume. However, the upper yield stresses were measured in the rheometer to be in the range of 10 to 13 Pa. The samples from 15 to 17 ft deep were silty and poorly graded sand with critical shear stresses from the erosion flume of 1.6 and 2.6 Pa. There was insufficient fine material in these samples for rheometer tests. The Towaliga River sample from 5 to 7 ft in depth consisted of a silty sand with a critical shear stress of 3.77 Pa. In contrast, the material from 20 to 22 ft in depth exhibited alternating 12 thin layers of slowly erodible clay to highly erodible sand with critical shear stresses of 18.4 Pa and 8.8 Pa, respectively. The upper yield stress for the clay layer was 13.2 Pa. The activity and Atterberg limits of the clay portion of the deeper sample indicate the presence of kaolinite and possibly illite. The shallow sample from Bay Creek at a depth of 10 to 12 ft was a clayey sand and the deeper sample from 25 to 27 ft was a silty clay. Samples from both depths had critical shear stresses in excess of 21 Pa while the yield stresses were 15.9 Pa and 17.7 Pa for the shallow and deep samples, respectively. A poorly graded sand was found at a depth of 10 to 11 ft at the Hodchodkee Creek site, while clayey sand was present from 11 to 17 ft depths. Unfortunately, the Shelby tubes were corroded and prevented sample extrusion except for the deepest sample which had a critical shear stress of 1.6 Pa. Insufficient fines were present for rheometer testing. The Canoochee River samples were hard, stiff marine clays that were impossible to erode in the flume. Upper yield stresses of 15.9 Pa and 17.7 Pa were measured with the rheometer. The activity values and Atterberg limits indicated that these samples were a montmorillonite clay that settled in a dispersed manner. All site samples with the exception of the Canoochee River samples exhibited activity and Atterberg limit values that strongly suggested kaolinite clays, perhaps with some illite, and they settled as flocculated slurries. ANALYSIS The work of Navarro (2004) in previous research (Sturm et al. 2004) developed a Shields type of diagram as a means of predicting critical shear stress of sediment samples taken from existing or potential bridge sites. The proposed relationship depends only on measuring a complete size distribution of the sediment including a sieve test and a hydrometer test. The Shields parameter was correlated with a dimensionless sediment diameter and the percent fines, defined as the 13 percent by weight of silt plus clay in the size distribution. The relationship suggested by Navarro (2004) is given by *c = 0.584 10 2.68Fines d -0.377 * (2.1) in which the definitions of the Shields parameter, *c, and the dimensionless particle diameter, d*, are given in Figure 2.6 where s = specific weight of the sediment; w = specific weight of water; d50 = median particle sediment size from particle size analysis; Gs = specific gravity of sediment; g = gravitational acceleration; = kinematic viscosity of water; and Fines = decimal fraction by weight of silt plus clay fraction of particle size distribution. The new data points collected from the flume studies in this project can be observed to be consistent with Navarro's relationship in Figure 2.6. Figure 2.7 further indicates that the data from this study validate the previous data for the critical value of the Shields parameter. Figure 2.6. Shields diagram with natural sediments from previous study and the present study 14 R2 = 0.89 Figure 2.7. Measured vs. predicted Shields parameter (Eq. 2.1) for previous flume data and current flume data. If the two data sets are combined, the best-fit regression relationship is nearly the same as before and is given by *c = 0.644 10 2.48Fines d -0.409 * (2.2) The coefficient of determination increases slightly from 0.89 to 0.90 as shown in Figure 2.8. Error intervals are shown in Figure 2.8 for the limits of 0.5 to 2.0 times the predicted value of Shields parameter, and 86% of the data fall between these limits. The measured vs. predicted data are re-plotted in Figure 2.9 in terms of critical shear stress in Pa (1.0 Pa = 1.0 N/m2 = 0.0209 lb/ft2). Predictions for critical shear stresses in excess of 15 Pa (0.31 lb/ft2) vary by a relatively large amount, but if the prediction is limited to values less than 15 Pa, the standard error is 2.0 Pa as shown in Figure 2.9. The inherent experimental error in measuring the critical shear stress using the erosion flume is of the order of 0.5-1.0 Pa (Hobson 2008). 15 100 *c=0.644102.48Fines d*-0.409 R2 = 0.90 10 Predicted Shields Parameter 1 0.1 0.01 0.01 Navarro 2004 Hobson 2008 Series4 0.1 1 10 100 Measured Shields Parameter Figure 2.8. Measured vs. predicted Shields parameter by Eq. 2.2 for previous flume data and current flume data with regression applied to combined data set (Error Limits =0.5 - 2.0 times predicted value). 30 25 Predicted Critical Shear Stress, Pa 20 15 10 5 s.e. = 2.0 Pa 0 Navarro 2004 Hobson 2008 Series4 0 5 10 15 20 25 30 Measured Critical Shear Stress, Pa Figure 2.9. Measured vs. predicted critical shear stress (Eq. 2.2) for combined data set. 16 Although the dimensionless diameter, d*, does explain some of the variance in the data, a second regression model was applied with the dimensionless Shields parameter as a function of fraction of Fines alone. The best-fit equation is given by *c = 0.211103.35Fines (2.3) and it is shown in Figure 2.10. (It must be emphasized again that Fines is the decimal fraction of silt+clay in the sample by weight). The coefficient of determination of Equation 2.3 given in Figure 2.10 is very similar to that in Figure 2.9 so that this approach seems to be statistically acceptable, but it no longer includes the viscous effect represented by d*. As shown in Figure 2.6, the best-fit curves for *c as a function of d* and Fines are nearly parallel to the laminar flow/smooth turbulent flow portion of the Shields curve (d* < 2). From Equation 2.2, the best-fit slope is -0.41, which is very close to the accepted value of -0.39 for the laminar/smooth turbulent portion of the Shields curve (Sturm 2001). R2 = 0.88 Figure 2.10. Critical Shields parameter as a function of Fines content applied to both data sets. 17 Upper and lower yield stresses were nondimensionalized in the same manner as for the Shields parameter using the median particle size of the entire size distribution. When regressed against the dimensionless particle size, d*, also defined in terms of the median particle size for the full size distribution, the results as shown in Figure 2.11 are *y1 = 1.45 106 d -3.04 * R 2 = 0.85 (2.4) *y2 = 5.30 106 d -2.78 * R 2 = 0.97 (2.5) in which *y1 = dimensionless lower yield stress, and *y2 = dimensionless upper yield stress. The upper yield stress appears to be more representative of the critical shear stress determined in the erosion flume. Unfortunately, insufficient data were available to relate the yield stress directly to the critical shear stress, but the fact that both can be related to d* shows that they are related to each other. For example, since both the dimensionless upper yield stress and the Shields critical shear stress parameter are functions of d*, they can be related to each other using Equations 2.2 and 2.5 to produce the following relationship: ^*c = 0.0661102.48Fines ( *y2 )0.15 (2.6) in which the upper yield stress reflects the influence of the specific cohesive characteristics of the type of clay present in the fines on the critical value of the Shields parameter. Yield stress of fine grained sediments can be measured from rheometer tests for cases of highly nonerodible samples for which critical shear stress cannot be measured in the flume. What is needed to obtain a direct experimental confirmation of Equation 2.6 is a set of samples for which both yield stress from the rheometer and critical shear stress from the flume can be measured and compared. 18 Figure 2.11. Critical Shields parameter for upper and lower yield stress as a function of d*. ERODIBILITY CLASSIFICATION FOR GEORGIA SOILS The classification of erosion resistance by Hanson and Simon (2001) can be used to show the distribution of critical shear stress values in Georgia. For this purpose, data points collected by both Navarro (2004) and Hobson (2008) are included. Samples which had critical shear stress values exceeding 21 Pa as the limit of the flume testing apparatus are placed into the category of "resistant" (10-99.9 Pa) in Figure 2.12. Additional data points obtained by Navarro (2004) for which erosion rates could not be measured but which had critical shear stresses estimated at the point of incipient motion are also included. A total of 44 samples are represented in Figure 2.12 at the sampling locations identified previously in Figure 2.1. More than half of all samples fall into the "resistant" erosion category with very few samples in the "erodible" class. For comparison, Table 2-2 gives the critical shear stress and corresponding erosion category for various sizes of sand such as medium sand, which is classified as "very erodible". All samples tested were significantly more resistant to erosion than medium sand. 19 Percent of Total Samples 100 80 60 54.6 38.6 40 20 6.8 0 0 0.01-0.374 0.375-1.99 2.0-9.99 10-99.9 very erodible erodible moderately resistant resistant Critical Shear Stress, Pa 0 >99.9 very resistant Figure 2.12. Erodibility classification of Georgia soil samples. Table 2-2. Critical shear stress and erodibility classification of various sand sizes (after Julien 1995). Size class d50, mm d* of sand *c c , Pa c , lb/ft2 Class Very coarse >1 25 0.029 0.47 0.010 Erodible Coarse >0.5 12.5 0.033 0.27 0.0056 Very erodible Medium >0.25 6.3 0.048 0.19 0.0040 Very erodible Fine >0.125 3.2 0.072 0.14 0.0029 Very erodible Very fine >0.0625 1.6 0.109 0.11 0.0023 Very erodible RECOMMENDED FORMULA FOR CRITICAL SHEAR STRESS Based on the foregoing analysis, it is recommended that Equation 2.2 be used to estimate critical shear stress of soil samples collected in Shelby tubes at the actual bridge site rather than using the classification of "medium sand" for all bridge sites. The formula is repeated below for reference: *c = 0.644 102.48Fines d*-0.409 (2.2) in which the variables are defined by *c = ( s c - w )d50 d* = (Gs - 1) gd 3 50 2 1/ 3 20 Fines = decimal fraction of silt+clay by weight in soil sample c = critical shear stress in Pa (N/m2) or lb/ft2 (1 lb/ft2 = 47.9 Pa) s = specific weight of sediment = 26,000 N/m3 or 165 lb/ft3 = specific weight of water = 9810 N/m3 or 62.4 lb/ft3 d50 = median sediment size in meters or ft Gs = specific gravity of sediment = 2.65 as an average value g = acceleration due to gravity = 9.81 m/s2 or 32.2 ft/s2 = kinematic viscosity of water = 1.0010-6 m2/s at 20 C or 1.08 10-5 ft2/s at 68 F The approximate limits on the use of Equation 2.2 are 0.03 mm < d50 < 1.0 mm 0.05 < Fines < 0.7 Georgia soils only If the predicted critical shear stress from Equation 2.2 is less than 15 Pa, then the uncertainty is approximately 2 Pa which is similar to the experimental uncertainty in a relationship obtained for the Powder River Basin in Wyoming by Thoman and Niezgoda (2008). Predictions of critical shear stress greater than 15-20 Pa by Equation 2.2 must be viewed as unreliable because they are outside the range of the samples tested. For soils with high percentages of clay, it is recommended that additional experimental work be conducted using a rheometer to further develop Equation 2.6 in which the yield stress is a measure of the cohesiveness of the type of clay present. 21 CHAPTER 3 PHYSICAL MODEL STUDY OF TOWALIGA RIVER BRIDGE INTRODUCTION In this chapter, the results of the physical model study are given and discussed. The site selection was carried out systematically according to a set of criteria developed in concert with the GDOT Bridge Design Office. The final site selected was the SR 42 crossing of the Towaliga River near Forsyth, Georgia. A model of the bridge and river bathymetry was built in the Georgia Tech Hydraulics Laboratory at a scale of 1:60. Velocities and scour depths were measured for a total of six experimental runs, some of which were needed to establish a technique for building a model erodible embankment that mimicked the observed scour behavior of the existing bridge during Tropical Storm Alberto. During this historic event, severe overtopping of the bridge and significant scour around the pier bents in the left floodplain occurred, but the embankment and bridge remained sufficiently intact until repairs could be made to the scoured area. In what follows, the site selection process is described first. Then the experimental techniques are established, and the experimental data for observed local and contraction scour are presented and discussed. SITE SELECTION A total of nine bridge sites were inspected in the field to determine their suitability for the purposes of the physical model study. A relatively small bridge was sought in comparison to the Ocmulgee River bridge in Macon, which was modeled in a previous study, because a large number of smaller bridges failed during Alberto. Partial failure of the bridge due to contraction scour and subsequent repair was considered desirable, because the existing bridge could be 22 modeled under conditions of severe scour in an attempt to reproduce observed scour measurements. Finally, streamflow data and field measurements of observed scour were needed for comparisons with the laboratory data, but this criterion was the most challenging since small bridges tend not to be instrumented by the USGS. The bridges that were inspected in the field as possible candidates for a model study are listed in Table 3-1. Table 3-1. Possible sites for physical model Site Route Stream County Gage A1 SR 96 Mossy Cr Peach A2 SR 247 Echeconee Houston Cr A3 SR 74 Tobesofkee Bibb Cr A4 SR 87 Towaliga R Monroe B1 SR 90 Turkey Cr Dooly B2 SR 11 Mossy Cr Houston B3 SR 42 Towaliga R Monroe B4 SR 16 Towaliga R Butts B5 SR 74/85 Line Cr Coweta No No Yes: 1967-78, 1990, 1994 No Yes: 1951-2006, real time Yes: 1979 - 2006 Yes: 1945 - 1966, 1991, 1994 Yes: 1961 - 1994 Yes: 1965 - 2006, real time D.A. (mi2) Alberto Q Stage (cfs) (ft) Return Interval (yr) New Bridge ? Yes Yes 156 26 30,000 >100 No Yes 45 14.3 5,820 >100 Yes 93 19.9 24,000 >100 Yes 315 ?? 60,000 >100 No 105 26.5 23,000 >100 No, widened 101 20.1 28,400 >100 No Of the sites listed in Table 3-1, the selection was narrowed down to two bridges as shown in Table 3-2. The SR 42 bridge over the Towaliga River was chosen as the bridge to be modeled in the laboratory primarily because the existence of contraction scour was more evident. Photographs of the site are shown in Figure 3.1. The bridge crosses between two bluffs with a solid rock outcropping forming the steep right bank just downstream of the bridge. The drainage area of the river at the SR 42 bridge is 315 mi2. Discharge was estimated as 60,000 cfs by the USGS for Tropical Storm Alberto (Stamey 1996). The previous bridge was instrumented and the USGS rating curve is shown in Figure 3.2. The peak discharge for this period of record was 15,000 cfs which has a water surface elevation below the low chord of the existing bridge that was built in the 1970s. 23 Table 3-2. Final bridge model selection Selection Criteria SR 42 over Towaliga R. Damaged during Alberto due to Yes. Bridge maintenance cross- contraction scour? sections clearly show contraction scour in left floodplain (probably includes pressure scour effects) Existence of stream gauging Yes, but only for bridge that station to determine hydraulic conditions and historical cross- was replaced by the current bridge which was built in 1971. section changes? Old bridge data from 19291950. No current streamflow records. Alberto discharge estimated with no stage measurement. No velocity measurements. Repair rather than replacement of Yes. Some good bridge bridge and plans of bridge? maintenance cross sections but not extremely detailed. Original cross section from bridge plans also available. Accessible for surveying cross sections? Yes. Wooded but not impossible. Very steep right Relatively small bridge compared to 5th St. bridge over Ocmulgee R. (D.A. = 2,240 mi2) bank downstream. Yes. D.A. = 315 mi2. Located above the fall line in Monroe County. SR 74 over Tobesofkee Cr. Probably. Main channel bed currently appears to be lower than original bridge plans and USGS cross sections from the 1970s. Yes, some USGS cross sections from the 1970s available as well as velocity measurements. Stream gauge no longer in operation but measurement of both peak stage and discharge available for Alberto. Yes. Cross section available from original bridge plans. Current streambed is mostly rock. Yes. Some wading required and heavily wooded floodplains. Yes. D.A. = 156 mi2. Located above the fall line in Bibb County. (a) Left bank of main channel (pier bent 5) (b) Left floodplain under bridge (bents 5, 6, 7) (c) Looking upstream (pier bents 3 and 4) (d) Left floodplain on upstream side Figure 3.1. Towaliga R. Bridge at SR 42 crossing. 24 Gaging Station 02211500 Towaliga River at SR 42 Datum is 410 ft above NGVD29 100 Gage-Height, ft 10 1 1000 10000 Discharge, cfs 100000 Figure 3.2. Towaliga River rating curve at SR 42 and extension (USGS) EXPERIMENTAL TECHNIQUES All experiments were conducted in the hydraulics laboratory of the School of Civil and Environmental Engineering at the Georgia Institute of Technology, Atlanta, GA. The physical model of the Towaliga R. bridge was built inside a 4.2 m wide by 24.2 m long horizontal flume at a geometric scale of 1:60. The model had a mobile bed that reproduced the river bathymetry in addition to the complete bridge and pier geometry. The approach channel upstream of the bridge was 7.3 m long followed by a working mobile bed section with a length of approximately 6.1 m in which the bridge model was placed. The templates for the river model cross sections in the approach channel were cut from plywood sheets placed vertically at regular intervals with elevations scaled from detailed field measurements of river bathymetry. The spaces between the 25 templates were filled with bed sediment and carefully leveled to the elevations established by the templates. The approach channel bed was then fixed with polyurethane. In the mobile bed section, thin aluminum templates were used to reproduce the bed bathymetry and then removed for the scour tests. The model of the embankment was also constructed as an erodible fill with rock riprap protection in order to reproduce the influence of erosion of the endroll on the abutment scour and pier scour in the region of the toe of the embankment. (Ettema et al. 2006). The water supply to the flume was provided from a large constant-head tank through a 30.5 cm diameter pipe that can deliver up to 0.3 m3/s to the head box of the flume. A flow diffuser, overflow weir, and baffles in the flume head box produced stilling of the inflow and a uniform flume inlet velocity distribution. A flap tailgate controlled the tailwater elevation. Water recirculated through the laboratory sump from which two pumps continuously provided overflow to the constant-head tank. In the supply pipe, discharge was measured by a magnetic flow meter with an uncertainty of 0.001 m3/s. An instrument carriage was mounted on horizontal steel rails and was moved along the flume on wheels driven by a cable system and electric motor. Approach velocities were measured with a SonTek 16 MHz acoustic Doppler velocimeter (ADV) that was attached to the instrument carriage on a mobile point gauge assembly that could be accurately positioned in all three spatial dimensions. A 3D down-looking probe was used to measure velocity profiles across the deeper portions of the cross section while a 2D side-looking probe was selected to measure velocity profiles in the shallow floodplain areas. The water depth and bed elevations before and after scouring were measured by the point gauge and the ADV. The ADV can generally measure the distance from the center of the sampling volume to a solid boundary with 1 mm uncertainty. The sampling frequency of the ADV was chosen to be 25 Hz with a sampling duration of 2 minutes at each measuring location. Each of the pier bents consists of two in-line rectangular columns having a width of 3 ft, and they were modeled inclusive of pile caps and piles at a geometric scale of 1:60. Abutment or 26 end bents were also modeled and buried in the erodible embankment which was protected by rock riprap. The bridge deck was modeled but made removable for scour measurements. The mobile bed sediment was a uniform sand with d50 = 0.53 mm which was selected based on the scour modeling methodology developed in previous research (Lee et al. 2004, Hong et al. 2006, Sturm et al. 2004, 2006). In this approach, the ratio of approach velocity to critical velocity for initiation of sediment motion, V1/Vc, and the Froude number are the same in model and prototype for contraction scour. Pier scour adds the further requirement that V1/Vc < 1.0 in the model by the amount necessary to compensate reduction in dimensionless scour depth (ds/b) at very large values of the ratio of pier width to sediment diameter, b/d50, in the field in comparison to the laboratory. For the measured prototype critical shear stress of 3.8 Pa and d50 = 0.11 mm, the model sediment size of d50 = 0.53 mm and geometric scale ratio of 1:60 satisfied all these potentially conflicting requirements within about 15% to provide a reasonable expectation of being able to reproduce prototype scour depths in the laboratory. There may be some question as to whether equilibrium scour conditions were reached in the field, but backwater from the Ocmulgee River, to which the Towaliga River is a tributary, is estimated to have persisted at the modeled bridge site for as much as 15 days based on the hydrographs given by Stamey (1996). EXPERIMENTAL PROGRAM The full scope of the physical modeling program is summarized in Table 3-3. Preliminary HECRAS modeling indicated that the maximum velocity under the bridge in the left floodplain would occur at 45,000 cfs with subsequent increases in discharge contributing primarily to overtopping relief. The first three experimental runs were an attempt to model the embankment as erodible with the rock riprap placed according to GDOT standards. In Runs 2 and 3, a vertical sheet pile was buried inside the embankment. In all three runs significant erosion and failure of the embankment occurred contrary to the observed prototype behavior. The sheet pile only served as a fixed barrier that exacerbated abutment scour. Runs 4, 5, and 6 were conducted with an erodible 27 embankment that was more carefully compacted and completely covered with riprap before putting the removable model roadway in place. This approach was successful in maintaining the general integrity of the embankment. Experimental Runs 4 and 5 were conducted at 45,000 cfs. The prototype tailwater was reproduced in Run 5 and overtopping of the embankment occurred as expected. Run 4 was intended as a comparison of the scour at the same flow rate but with lower tailwater which prevented any significant overtopping; the flow type was submerged orifice flow referred to as pressure flow. Run 6 was a pressure flow case with the same tailwater as Run 4 but with a discharge of 39,000 cfs that corresponded to this prototype tailwater. Table 3-3. Experimental conditions ( Q =discharge, V1m =approach flow velocity in main channel, Vcm =critical velocity in main channel, V1 f =approach flow velocity in floodplain, Vcf =critical velocity in floodplain, W .S.u = water surface elevation upstream and W .S.d = water surface elevation downstream of bridge) Run Q (cfs) V1 m (ft/s) Vcm (ft/s) V1 m Vcm V1 f (ft/s) Vcf (ft/s) V1 f Vcf W .S.u W .S.d (ft) (ft) Condition Overtopping 1 45,000 3.27 8.156 0.401 2.72 9.024 0.301 443.88 443.21 w/erodible embankment 2 45,000 3.27 8.156 0.401 2.72 9.024 0.301 443.88 443.21 Overtopping w/sheet pile 3 45,000 5.44 8.074 0.674 3.96 8.873 0.446 440.88 440.58 Pressure w/sheet pile 4 45,000 5.44 8.074 0.674 3.96 8.873 0.446 440.88 440.58 Pressure w/erodible embankment Overtopping 5 45,000 3.27 8.156 0.401 2.72 9.024 0.301 443.88 443.21 w/erodible embankment 6 37,000 3.22 8.012 0.401 2.45 7.426 0.330 440.88 440.58 Pressure w/erodible embankment 28 The approach velocity distributions were measured upstream of the bridge at the location shown in Figure 3.3. Scour depths were monitored with respect to time at Points A, B, and C which can be seen also in Figure 3.3. Cross sections after scour were measured immediately upstream of the bridge, just inside the bridge (by removing the bridge deck), and immediately downstream of the bridge at the longitudinal stations indicated in the figure. Lateral Station (ft) Approach Velocity Measurement Location 0 200 400 Point A 600 Point B Point C 800 2100 2300 2500 2700 2900 3100 3300 Longitudinal Station (ft) Cross-section Measurement Location x=2734.8ft, 2743.2ft and 2782.8ft Prototype Bed Elevation (ft) 450 446 442 438 434 430 426 422 418 414 410 406 402 398 394 Figure 3.3. Location of stations for scour time monitoring and approach velocity measurement. EXPERIMENTAL RESULTS The approach velocity distributions measured at a cross section located approximately 500 ft upstream of the bridge are shown in Figure 3.4 for Runs 4, 5, and 6. Velocities are clearly higher for Run 4, which had the same discharge as Run 5, the overtopping run, but with the tailwater deliberately lowered to ensure that all the flow went under the bridge. The resulting mean velocities in the left floodplain and main channel for Run 4 are 3.96 ft/s and 5.44 ft/s, 29 respectively, which are approximately 50% greater than in Runs 5 and 6 as shown previously in Table 3-3. Accordingly, the largest scour is expected for Run 4. z,Bed Elevation (ft) Velocity (ft/s) 460 7.0 Run 4 Run 5 Run 6 450 5.0 440 3.0 430 1.0 420 -1.0 410 -3.0 400 0 -5.0 100 200 300 400 500 600 700 800 Lateral Station (ft) Figure 3.4. Approach velocity profiles The time history of scour development was measured at three points immediately upstream of the bridge as shown previously in Figure 3.3. Point A is located at the toe of the abutment, while Point B is on the left bank of the main channel. The last point (Point C) is near the centerline of the main channel. Figures 3.5 to 3.7 show the scour time history results for Runs 4, 5, and 6. In all three experimental runs, the maximum scour occurred at the left bank of the main channel. The scour at Points A and B occurred rapidly during the first 10 to 20 hrs after which the rate of scour began to decrease. In all three experimental runs there was almost no scour at the centerline of the main channel (Point C). At the toe of the abutment (Point A), the rate of scour development in Runs 4 and 5 was very gradual after the initial rapid development, but in Run 5 there was an abrupt increase again at about 50 hours followed by another leveling off with respect to time. This event in Run 5 is associated with gradual sliding of a small portion 30 of the abutment end fill and riprap at the toe into the scour hole which slowed the rate of time development initially. The scour infill area was then scoured out again starting at a time of 50 hours, perhaps because the deposited riprap was transported out of the scour hole. In Run 6, there was less sliding of the abutment fill and riprap into the scour hole near the toe of the abutment, and the scour developed continuously. In general, the scour near the main channel bank in Runs 4 and 5 continued increasing for a longer duration than scour at the toe of the abutment, perhaps because of the larger scour depth near the bank. In Run 6 at a lower flow rate, and without the complication of partial failure of the abutment, the scour at the toe of the abutment and at the bank of the main channel occurred at very similar but gradual rates and did not reach equilibrium until about 100 hours of continuous flow in the laboratory. Previous research on contraction scour in the Ocmulgee River at Macon (Sturm et al. 2006) also showed a much longer duration for equilibrium development of contraction scour than for local scour. Scour depth (ft) 8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0 0 -1.0 20 40 60 80 100 120 Time (hr) time development at location A location B location C Figure 3.5. Time history of scour development for Run 4. 31 Scour depth (ft) Scour depth (ft) 8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0 0 -1.0 8.0 7.0 6.0 5.0 4.0 3.0 2.0 1.0 0.0 0 -1.0 20 40 60 80 100 120 Time (hr) time development at location A location B location C Figure 3.6. Time history of scour development for Run 5. 20 40 60 80 100 120 Time (hr) time development at location A location B location C Figure 3.7. Time history of scour development for Run 6. 32 Results for scour depth are given in Table 3-4. The scour was separated into contraction and pier scour as described previously by Sturm et al. (2006). First, the pier scour holes are removed by using the concurrent ambient scour surface next to the local scour holes. Then the average depth of the cross section is adjusted relative to an undisturbed longitudinal bed profile through the bridge before scour. The difference between the average bed elevation after scour and the adjusted bed elevation before scour is the contraction scour. Note that contraction and pier scour were calculated separately for the main channel and floodplain in Table 3-4. Only the maximum pier scour is reported in the table. The maximum pier scour in the main channel occurred at Pier Bent #4, while the maximum pier scour in the floodplain was found at Pier Bent #5. These two pier bents are near the left bank of the main channel which failed. The contraction scour depths in the left floodplain for Runs 4 and 5 are 9.8 ft and 7.6 ft, respectively, and are greater than the corresponding maximum local pier scour depths of 4.8 ft and 5.9 ft at Pier Bent #5 as shown in Table 3-4. Contraction scour dominates local pier scour in the floodplain because of the floodplain flow that is forced through the bridge opening in addition to the vertical contraction due to bridge submergence. The contraction scour in both the floodplain and main channel decreases from Run 4 to Run 5 because of the relief offered by bridge overtopping in Run 5 at the same discharge. The contraction scour in the floodplain continues to decrease in Run 6 to a value of 4.9 ft because of the lower discharge at the same tailwater as Run 4. The maximum floodplain and main channel pier scour depths in Runs 5 and 6 are comparable due to similar approach flow velocity distributions that have approximately the same magnitude near the left main channel bank in the vicinity of Pier Bents #4 and#5 (see Figure 3.4). Because of the smaller contraction scour in the main channel compared to the floodplain, local pier scour dominates contraction scour in the main channel for all three experimental runs. 33 Table 3-4. Experiment results ( z m-ref =average reference bed elevation in main channel, z f -ref = average reference bed elevation in floodplain, z m =average bed elevation in main channel after scour, z f = average bed elevation in floodplain after scour, d m-c =average contraction scour depth in main channel ( z m-ref - z m ), d f -c =average contraction scour depth in floodplain ( z f -ref - z f ), d m- p =maximum pier scour depth in main channel, d f - p =maximum pier scour depth in the flood plain) Run z m-ref z f -ref z m z f (ft) (ft) (ft) (ft) 4 410.82 422.06 407.90 412.24 5 410.82 422.06 409.17 414.43 6 410.82 422.06 409.65 417.16 (1)Pier bent #4; (2)Pier bent #5 d m-c (ft) 2.92 1.65 1.17 d f -c (ft) 9.81 7.63 4.90 d m- (1) p (ft) 11.95 7.94 7.13 d (2) f -p (ft)) 4.82 5.93 7.67 Condition Pressure w/erodible embankment Overtopping w/erodible embankment Pressure w/erodible embankment A better idea of the scour distribution at the bridge cross sections can be seen in Figures 3.8, 3.9, and 3.10. The maximum total scour depth occurred in Run 4 as expected because of the higher velocities for this case. The best comparison between measured bed elevations in the laboratory and observed bed elevations in the field, some of which were measured during Alberto while others were reported after the flood, is shown in Figure 3.10 for the case of pressure flow just before overtopping (Run 6, Q = 37,000 cfs). Bed elevations at Pier Bents #4 and #6 in Run 6 were reproduced in the laboratory within a few tenths of a foot. The same cross sections just inside the upstream side of the bridge for Runs 4, 5, and 6 are shown in Figure 3.11 for comparison. The maximum scour at Pier Bent #4 for Run 4 is evident as is the exposure of the pier caps at Pier Bents #5, #6, and #7 for all three experimental runs. Of particular interest is the scour at the left abutment which is greatest for Run 4 and least for Run 6. 34 z,Bed Elevation (ft) 450 #8 #7 #6 #5 #4 #3 #2 #1 430 410 390 300 400 500 600 700 800 y,Lateral Station (ft) Initial C.S. Measurement during Alberto Field (Jan-93) Bridge inside C.S. after run 4 Field (Oct-94) Post-scour report Bridge cross-section comparison just inside upstream side of bridge (Run 4:Q=45,000cfs, W.S. elev. upstream of bridge=440.88ft, Pressure flow with erodible embankment) Figure 3.8. Run 4 Cross section comparison. 35 450 #8 #7 #6 #5 #4 #3 #2 #1 430 z,Bed Elevation (ft) 410 390 300 400 500 600 700 800 y,Lateral Station (ft) Initial C.S. Measurement during Alberto Field (Jan-93) Bridge inside C.S. after run 5 Field (Oct-94) Post-scour report Bridge cross-section comparison just inside upstream side of bridge (Run 5:Q=45,000cfs, W.S. elev. upstream of bridge=443.88ft, Overtopping with erodible embankment) Figure 3.9. Run 5-Cross section comparison 450 #8 #7 #6 #5 #4 #3 #2 #1 430 z,Bed Elevation (ft) 410 390 300 400 500 600 700 800 y,Lateral Station (ft) Initial C.S. Measurement during Alberto Field (Jan-93) Bridge inside C.S. after run 6 Field (Oct-94) Post-scour report Bridge cross-section comparison just inside upstream side of bridge (Run 6:Q=37,000cfs, W.S. elev. upstream of bridge=440.88ft, Pressure with erodible embankment) Figure 3.10. Run 6 Cross section comparison 36 450 #8 #7 #6 #5 #4 #3 #2 #1 430 Bed Elevation (ft) 410 390 300 400 500 600 700 800 Lateral Station (ft) Initial C.S. Run 6 C.S. Run 4 C.S. Run 5 C.S. (Run 4: Q=45,000, W.S. elev.=440.88ft (Pressure) / Run 5:Q=45,000, W.S. elev.=443.88ft (Overtopping) / Run 6: Q=37,000cfs, W.S. elev=440.88ft (Pressure)) Figure 3.11. Bridge cross-section comparison after scour Bed contours before and after scour are shown for Run 4 in Figure 3.12, and the corresponding photographs at the end of scour can be seen in Figure 3.13. The partial failure of the abutment can be observed from the contours and from exposure of the top of the end bent in Figure 3.13. The cupping out of the left floodplain bed under the bridge by contraction scour is quite striking in Figure 3.13. In addition, the failure of the left bank of the main channel is evident. The contraction scour extends downstream of the bridge between Pier Bents #5 and #6 as shown by the contours in Figure 3.12. 37 Lateral Station (ft) (a) Before Scour 0 200 400 Point A 600 Point B Point C 800 2200 2400 2600 2800 3000 Longitudinal Station (ft) (b) After scour 0 3200 Prototype Bed Elevation (ft) 450 446 442 438 434 430 426 422 418 414 410 406 402 3400 398 394 200 400 600 800 2200 2400 2600 2800 3000 Longitudinal Station (ft) 3200 3400 Prototype bed elevation contours before and after scour ( Run 4: Q=45,000cfs, W.S. elev. upstream of bridge=440.88ft, Pressure with erodible embankment) Figure 3.12. Bed contours before and after scour for Run 4. Lateral Station (ft) 38 (a) (b) Figure 3.13. Photographs of bed after scour for Run 4. 39 Similar features of the bed can be seen in Figures 3.14 and 3.15 for Run 5, which was the roadway overtopping case. Some riprap can be observed to have slid into the scoured area in the vicinity of Pier Bent #7 near the toe of the abutment. The embankment remained fairly intact in spite of overtopping. The scoured area extends downstream of Pier Bent #5 as shown by the contours in Figure 3.14. In Run 6, for which bed contours and photographs of the scoured bed are shown in Figures 3.16 and 3.17, it appears that abutment scour, contraction scour, and pier scour all occurred in the vicinity of the toe of the abutment. This experimental run was the one that best matched the prototype scour depths observed for Alberto, and the flow conditions were pressure flow without overtopping at a discharge of 37,000 cfs. The end bent and endroll remained relatively intact for this case. The abutment scour appears at the upstream corner of the endroll as a separate scour hole similar to abutment scour holes observed in previous research by the principal investigator (Sturm 2006). In Figure 3.17, local scour can be seen around the individual columns of Pier Bent #7 at the toe of the abutment while contraction scour is manifested between Pier Bents #7 and #6. Measured velocity vectors are shown in Figure 3.18 for Runs 5 and 6 prior to the occurrence of scour. In particular, results of Run 6 in Figure 3.18b illustrate the classic flow pattern around an abutment as the floodplain flow turns and accelerates around the toe of the abutment to reach much higher velocities under the bridge between pier bents. Here there is the local confluence of abutment scour, pier scour, and contraction scour that so confounds scour prediction efforts in this zone. 40 Lateral Station (ft) (a) Before Scour 0 200 400 Point A 600 Point B Point C 800 2200 2400 2600 2800 3000 Longitudinal Station (ft) (b) After scour 0 3200 Prototype Bed Elevation (ft) 450 446 442 438 434 430 426 422 418 414 410 406 402 3400 398 394 200 400 600 800 2200 2400 2600 2800 3000 Longitudinal Station (ft) 3200 3400 Prototype bed elevation contours before and after scour ( Run 5: Q=45,000cfs, W.S. elev. upstream of bridge=443.88ft, Overtopping with erodible embankment) Figure 3.14. Bed contours before and after scour for Run 5. Lateral Station (ft) 41 (a) (b) Figure 3.15. Photographs of bed after scour for Run 5. 42 Lateral Station (ft) (a) Before Scour 0 200 400 Point A 600 Point B Point C 800 2200 2400 2600 2800 3000 Longitudinal Station (ft) (b) After scour 0 3200 Prototype Bed Elevation (ft) 450 446 442 438 434 430 426 422 418 414 410 406 402 3400 398 394 200 400 600 800 2200 2400 2600 2800 3000 Longitudinal Station (ft) 3200 3400 Prototype bed elevation contours before and after scour ( Run 6: Q=37,000cfs, W.S. elev. upstream of bridge=440.88ft, Pressure with erodible embankment) Figure 3.16. Bed contours before and after scour for Run 6. Lateral Station (ft) 43 (a) (b) Figure 3.17. Photographs of bed after scour for Run 6. 44 Lateral Station (ft) Run 5 (Q=45,000cfs, W.S. El at Bridge Upstream=443.88 ft) 0 Reference Velocity Vector (ft/s, Prototype) 7.21 Prototype Bed Elevation (ft) 450 446 442 200 438 434 430 400 426 422 418 600 414 410 406 402 800 398 2200 2400 2600 2800 3000 3200 3400 394 Longitudinal Station (ft) (a) Run 5 Run 6 (Q=37,000cfs, W.S. El at Bridge Upstream=440.88 ft) 0 Reference Velocity Vector (ft/s, Prototype) 7.21 Prototype Bed Elevation (ft) 450 446 442 200 438 434 430 400 426 422 418 600 414 410 406 402 800 398 2200 2400 2600 2800 3000 3200 3400 394 Longitudinal Station (ft) (b) Run 6 Figure 3.18. Velocity vectors at 40 percent of the depth for Runs 5 and 6. Lateral Station (ft) 45 DISCUSSION OF RESULTS Given the complex three-dimensionality of multiple scour processes near the toe of the abutment where bridge failure due to scour seemed most likely to occur during Alberto, it is worthwhile to question the reasonableness of adding contraction scour to local scour as independent processes. Because the model study reported herein was able to reproduce observed scour in the field for experimental Run 6, and to separate pier and contraction scour, the predicted local pier scour and contraction scour using currently accepted formulas can be compared with the model study results. The CSU pier scour formula (HEC-18) was applied in the left floodplain using the measured depth and approach velocity in the model scaled up to the prototype for Run 6. The result is a predicted pier scour depth in the floodplain of 8.5 ft at Pier Bent#7 as shown in Method 1 of Table 3-5. Using the HEC-18 pressure scour formula for vertical contraction scour, the result is negative, so a value of zero is entered for pressure scour in Method 1 in Table 3-5. Lateral contraction scour for the left floodplain is calculated from the HEC-18 formula for a floodplain width of Wf = 155 ft and floodplain discharge of Qf = 14,000 cfs inside the bridge as determined by the laboratory measurements of the velocity distribution there. The effective particle size is taken as dm = 1.25d50 with an assumed value of d50 = 0.375 mm for a medium sand as in current GDOT practice (personal communication, Susan Beck; GDOT Drainage Design Manual 2005). The result for flow depth in the floodplain at the bridge section after scour is y s = 0.0077(Q f /W f d 2/ m 3 )2 3/ 7 = 0.0077(14,000 /155) (1.25 0.375/ 305)2 / 2 3 3/ 7 = 37 ft (3.1) The depth of flow at the bridge before scour is 14 ft as determined from the laboratory model measurements. The resulting estimated contraction scour assuming a medium sand as the soil type is given by ds = ys - y0 = (37 - 14) = 23 ft (3.2) 46 The total scour in Table 3-5 becomes 31.5 ft taken as the sum of the different types of scour and assuming that there is no contribution of abutment scour. In a second approach to scour estimation summarized as Method 2 in Table 3-5, the actual measured critical shear stress of the soil at the bridge site (c = 3.8 Pa or 0.079 lb/ft2) from the flume measurements described in Chapter 2 is substituted into the contraction scour formula in the form given by: (HEC-18) ys = n2 (Q f /W f K n2 c )2 3/ 7 = 0.032 (14,000 /155) 1.492 (0.079) 2 3/ 7 = 5.0 ft (3.3) in which ys = flow depth at the bridge after scour; n = Manning's resistance coefficient in the floodplain at the bridge section; Qf = floodplain discharge inside the bridge opening; Wf = floodplain width inside the bridge opening; c = critical shear stress of the bridge foundation soil; and Kn=1.49 for English units or 1.0 for SI units. (Equation 3.3 is the same as Equation 3.1 for estimating contraction scour except that it is given in terms of the original variables before substitution of relationships for n and c in terms of d50. Equation 3.3 should be used for estimating contraction scour in fine-grained soils for which the critical shear stress has been determined as described in Chapter 2.) The result for ys from Equation 3.3 using Method 2 is approximately 5 ft which is less than the flow depth before scour (14 ft); therefore, the estimated contraction scour is zero instead of 23 ft as in Method 1 (see Table 3-5). The difference is in guessing the soil to be medium sand vs. using the actual measured critical shear stress. However, there is further uncertainty in estimating the floodplain value of Manning's n inside the bridge where it is affected by vegetation and the bridge piers. Because of the failure of the HEC-18 formula to predict a nonzero value of pressure scour in Method 1, the pressure scour formula suggested by Lyn (2008) was used in Method 2. The Lyn pressure scour formula is given by 47 d s y1 = 0.21 Va Vc 2.95 for Va Vc 1.43 or d s = 0.6 for Va > 1.43 y1 Vc (3.4) in which Va = flow velocity inside the bridge; Vc = critical velocity inside the bridge section; y1 = approach flow depth; and ds = pressure scour depth. For a measured flow intensity value of Va/Vc = 0.95 and y1 = 21 ft just upstream of the bridge in Run 6, the pressure scour depth from Equation 3.4 is 3.8 ft as shown in Table 3-5. The total scour depth for Method 2 becomes 12.3 ft. On the other hand, the measured pier scour at Pier Bent #7 was 3.2 ft and the contraction scour due to vertical and lateral contractions was 4.9 ft as shown in Method 3 in Table 3-5. It is important to recognize that the total scour of 8.1 ft for the model study was not determined by adding pier and contraction scour, but rather by separating the measured total scour into pier and contraction scour components. Abutment scour is not calculated in Table 3-5 using HEC-18 since the endroll was protected by rock riprap including an apron. Nevertheless, some riprap slid into the scoured area near the toe of the abutment and some local abutment scour occurred, albeit at a different location than the pier and contraction scour. If calculated abutment scour had been included in the HEC-18 estimate of total scour, it would have obviously been even larger than shown in Table 3-5. Table 3-5. Comparison of predicted and measured scour depths in left floodplain for Run 6. Method Pier #7 Pressure Contraction Total Scour (ft) Scour (ft) Scour (ft) Scour (ft) 1. HEC-18 w/medium sand 8.5 0 23.0 31.5 2. HEC-18 w/measured c plus 8.5 3.8 Lyn pressure scour formula 0 12.3 3. Model study measurements 3.2 (included in 4.9 8.1 (in agreement with prototype) contraction scour) The discrepancy between predicted and observed values of total scour in Method 1 and Method 3 in Table 3-5 is partly due to using incorrect soil properties in this case by assuming the soil to be a medium sand, and partly due to a problem with the HEC-18 pressure scour formula. 48 However, after correcting these difficulties in Method 2, there are more subtle reasons for the remaining discrepancy between Method 2 and the observations. The local pier scour occurred in the presence, and simultaneously with, a lowering of the surrounding bed under the bridge due to vertical contraction scour. While it is true that the contraction component of scour took longer to reach equilibrium than the pier scour component, one possibility to consider in lieu of the HEC18 procedure is to apply the contraction scour first and then to estimate the pier scour using adjusted values of the approach velocity and depth. Another suggestion based on observations in this study is to apply a reduction to the estimated contraction scour for an idealized long contraction that takes into account the fact that the bridge is an abrupt, short contraction. Both of these suggestions go to the heart of the problem, which is that the local and contraction scour processes are occurring simultaneously and may not have maximum values that coincide spatially or temporally. Further experiments are needed to resolve this issue, especially in the vicinity of the abutment under pressure flow conditions. 49 CHAPTER 4 SUMMARY AND CONCLUSIONS SUMMARY Contraction scour is the result of flow acceleration under the bridge due to floodplain flow being blocked and forced through the bridge opening, natural geometric contraction of the main channel width, contraction of the main channel and floodplain widths due to a large number of pier bents, and vertical contraction due to submergence of the bridge leading to orifice flow with or without roadway overtopping. The formula used to predict contraction scour is an oversimplified, onedimensional treatment of an idealized contraction (Laursen 1960), but the model study results in Chapter 3 have emphasized the complex three-dimensionality of the contraction scour process and its interaction with local scour. Even if we accept the one-dimensional formula, estimating the required formula parameters in the field contributes to its misapplication and outright failure to correctly predict prototype contraction scour much of the time (Mueller and Wagner 2005). Two of the most important aspects of contraction scour prediction are (1) the estimation of a critical shear stress or critical velocity of the sediment in the scour zone; and (2) the current practice of adding contraction scour depths to local scour depths as though they were independent processes. Chapter 2 has dealt with the first problem by developing a regression relationship for estimating critical shear stress from erosion testing of undisturbed Shelby tube samples taken from all across the State of Georgia. The second problem was analyzed in Chapter 3 by using modeling techniques developed in previous research (Sturm et al. 2004, 2006) to demonstrate that a laboratory model can reproduce measured contraction and pier scour depths in the field from extreme hydrologic events, and in so doing exhibit more of the complex reality of the contraction scour process. Comparisons of measured scour components in the model study with predicted values using accepted HEC-18 procedures exposed the weakness of these formulas which is that 50 they do not account for interaction of the local and contraction scour nor the lack of coincidence of the spatial distribution of different types of scour. Furthermore, it is apparent that calculating contraction scour without knowledge of the actual critical shear stress of the soil can cause significant discrepancies between measured and predicted total scour depths. For example, an assumed medium sand having a critical shear stress of 0.23 Pa (0.0048 lb/ft2) in comparison with a moderately erosion resistant soil having a critical shear stress of 5 Pa (0.104 lb/ft2) produces a flow depth at the bridge after scour that is 3.7 times larger than for the actual soil. The corresponding change in contraction scour depth would depend on the value of the undisturbed floodplain flow depth. CONCLUSIONS 1. Erosion testing of Shelby tube samples from five bridge sites has validated and improved the relationship developed in previous GDOT research (Sturm et al. 2004) for critical shear stress which is an important parameter used in estimating contraction scour. The recommended relationship for the critical value of Shields' shear stress parameter depends on a dimensionless diameter and the percent of fine sediments by weight (silt + clay) in the soil sample. 2. Rheometer testing of the fine fraction of the Shelby tube samples has resulted in a proposed relationship for the upper yield stress that reveals a connection with the critical shear stress obtained from erosion testing in the flume. The critical value of the Shields parameter is shown to depend on percent fines in the soil and on the upper yield stress from the rheometer, which indicates the influence of the mineralogy of the fine fraction of the soil. Further experiments are needed to solidify this relationship which is especially important because of the common occurrence of mixed-layer soils with a significant fine size fraction in Georgia. 3. A physical model study of the SR 42 crossing of the Towaliga River near Forsyth, Georgia has shown that introduction of an erodible embankment protected by riprap, and a choice of 51 model sediment size guided by principles developed in previous GDOT research (Sturm et al. 2004, 2006), can successfully reproduce observed scour depths in the field for extreme hydrologic events such as Tropical Storm Alberto. 4. A significant component of the total scour in the left floodplain of the Towaliga River bridge was attributable to contraction scour which appeared to be due primarily to the vertical contraction of the flow associated with pressure scour. Although Alberto led to overtopping of the bridge, the maximum depth of scour occurred for submerged orifice flow just prior to overtopping. This is an important design flow that should be considered in estimating bridge scour rather than relying on just the 100-year peak discharge. 5. Using medium sand as a representative soil can lead to serious overestimation of contraction scour since 94% of the Shelby tube samples tested in this research can be classified as either "moderately resistant" or "resistant" to erosion while medium sand, which is used as a representative sediment size in scour calculations by GDOT, is classified as "very erodible". 6. The HEC-18 pressure scour formula may need to be improved as suggested by Lyn(2008) although additional data from longer duration experiments are needed. 7. Contraction scour developed over a longer period of time than did pier scour, but their simultaneous occurrence showed that using scour prediction formulas for separate components of scour in an additive fashion seriously overestimates the total scour. 8. The most vulnerable scour areas in the model study were the left floodplain and the left bank of the main channel where an unacceptable depth of scour exposed the pilings. Contraction scour was dominant on the floodplain, while pier scour was most significant on the channel bank which failed. Modeling of the embankment as erodible showed that partial failure of the riprap protection in the vicinity of the toe of the endroll moderated somewhat the pier scour there but exposed an isolated area of abutment scour that was not protected. 9. Contraction scour formulas might be improved by taking into account the modification of flow conditions as contraction scour occurs, and the introduction of a reduction factor that 52 takes into account the fact that a bridge is a short contraction rather than an idealized long contraction. RECOMMENDATIONS FOR FUTURE RESEARCH Further testing of artificially mixed soil samples to determine the upper yield stress from rheometer tests of the fine size fraction which can be related directly to critical shear stress obtained from flume tests is warranted. This approach would allow a more systematic study of the influence of the fine fraction on erodibility and would expand the applicability of the relationship for critical shear stress developed in this research. Building upon the successful efforts of the model study to reproduce contraction scour observed in the field under extreme hydrologic conditions, it is recommended that the physical model be used to systematically study simultaneous local scour and contraction scour for varying embankment lengths in order to develop a reduction factor for the idealized long contraction scour formula, especially under conditions of submerged orifice flow as well as free surface flow. Pressure scour with an erodible embankment protected by riprap should be studied further to determine the interaction of pier scour, abutment scour, and contraction scour near the abutment because abutment riprap protection may be insufficient to prevent scour failure of the pier closest to the abutment. 53 REFERENCES Alhadeff, S. J., Musser, J. W., Sandercock, A. C., and Dyer, T. R. (2000). Digital Environmental Atlas of Georgia, USGS, Atlanta, GA, 2 Disks. Dargahi, B. (1989). "Turbulent flow field around a circular cylinder." Exp. in Fluids, 8(1-2), 1-12. Ettema, R., Yorozuya, A., Muste, M., and Nakato, T. (2006). "Design estimation of abutment scour depths," Proceedings Third International Conference on Scour and Erosion, CURNET, Gouda, The Netherlands. GDOT Manual on Drainage Design for Highways (2005). Ayres Associates, Atlanta,Ga. Hoepner, M. A. (2000). "Stability of cohesive sediments from flume and rheometer measurements,"M.S. Thesis, Georgia Institute of Technology. Hobson, P. (2008). "Rheologic and flume erosion characteristics of Georgia sediments from bridge foundations," M.S. Thesis, Georgia Institute of Technology. Hong, S., Gotvald, A., Sturm, T. W., Landers, M. (2006). "Laboratory and Field Measurements of Bridge Contraction Scour," Proceedings Third International Conference on Scour and Erosion, CURNET, Gouda, The Netherlands. Laursen, E. M. (1960). "Scour at bridge crossings," J. Hydraul. Div., ASCE, 86(HY2), 39-54. Lee, SeungOh, Sturm, T. W., Gotvald, A., and Landers, M. (2004). "Comparison of Laboratory and Field Measurements of Bridge Pier Scour," Proceedings. 2nd Int. Conf. on Scour and Erosion, ed. by Y. M. Chiew, S. Y. Lim, and N. S. Cheng, Singapore, 231-239. Lyn, D. A. (2008). "Pressure-flow scour: A reexamination of the HEC-18 equation." J. Hydr. Eng., 134(7), 1015-1020. Melville, B. W. (1997). "Pier and abutment scour: integrated approach." J. Hydr. Eng., 123(2), 125-136. Melville, B. W., and Coleman, S. E. (2000). Bridge scour, Water Resources Publications, Highlands Ranch, Colo. Morris, J.L., and Pagan-Ortiz, J.E. (1999). "Bridge scour evaluation program in the United States", Stream Stability and Scour at Highway Bridges, E.V. Richardson and P.F. Lagasse, eds., ASCE, 61-70. Mueller, D. S., and Wagner, C. R. (2005). Field observations and evaluations of streambed scour at bridges, U.S. Department of Transportation, Federal Highway Administration, TurnerFairbank Highway Research Center, McLean, Va. Navarro, H. R. (2004). "Flume measurements of erosion characteristics of soils at bridge foundations in Georgia," M.S. Thesis, Georgia Institute of Technology. 54 Otsubo, K. and Muraoka, K. (1988). "Critical shear stress of cohesive bottom sediments," J. Hydr. Engrg., 114(10):12411256. Parola, A. C., Hagerty, D. J., and Kamojjala, S. (1998) Highway infrastructure damage caused by the 1993 upper Mississippi River basin flooding, NCHRP report 417, Transportation Research Board, Washington, DC. Raudkivi, A. J. (1986). "Functional trends of scour at bridge piers." J. Hydr. Eng., 112(1), 1-13. Ravisangar, V., Sturm, T. W., and Amirtharajah, A. (2005). "Influence of Sediment Structure on Erosional Strength and Density of Kaolinite Sediment Beds," J. Hydr. Engrg., ASCE, 131(5), 356-365. Richardson, E. V., and Davis, S. R. (2001). "Evaluating scour at bridges." HEC-18, U.S. Federal Highway Administration, Washington, DC. Stamey, Timothy C. (1996). "Summary of Data-collection activities and effects of flooding from Tropical Storm Alberto in parts of Georgia, Alabama, and Florida, July 1994," U.S. Geological Survey Open-File Report 96-228, 23p. Sturm, T. W. (2001). Open-channel hydraulics, McGraw-Hill, Boston. Sturm, T. W., Sotiropoulos, F., Landers, M., Gotvald, T., Lee, S. O., Ge, L., Navarro, R., and Escauriaza, C. (2004). Laboratory and 3D numerical modeling with field monitoring of regional bridge scour in Georgia, Georgia DOT Final Report, Atlanta, GA. Sturm, T. W., Sotiropoulos, F., Landers, M., Gotvald, T., Lee, S. O., Hong, S. H., and Escauriaza, C. (2006). Laboratory and 3D numerical modeling with field monitoring of regional bridge scour in Georgia, Phase 2, Georgia DOT Final Report, Atlanta, GA Sturm, Terry W. (2006). "Scour around bankline and setback abutments in compound channels," J. Hydr. Engrg., ASCE, 132(1), 2006, 21-32. Thoman, R. W., and Niezgoda, S. L. (2008). "Determining erodibility, critical shear stress and allowable discharge estimates for cohesive soils: Case study in the Powder River Basin of Wyoming," J. Hydr. Engrg., ASCE, 134(12), 1677-1687. Wardhana, K., and Hadipriono, F. C. (2003). "Analysis of recent bridge failures in the United States." Journal of Performance of Constructed Facilities, 17(3), 144-150. 55 APPENDIX A SEDIMENT CORE TEST RESULTS 56 Table A.1. Sediment Core Results for Meriwether County Samples Depth (BGS) 7 ft to 9 ft 10 ft to 12 ft 15.25 ft to 15.75 ft Critical Shear Stress Lower Yield Stress Upper Yield Stress Sediment Group Name Color Void Ratio Bulk Density >21 Pa 0.532 Pa 12.8 Pa CL Lean clay Gray with brown 1.48 1951 kg/m3 >21 Pa 0.396 Pa 10.1 Pa CL Lean clay Gray with brown 0.787 1910 kg/m3 1.59 Pa NP NP SP-SM Poorly graded sand with silt Gray 0.926 2233 kg/m3 Dry Density 1435 kg/m3 1246 kg/m3 1765 kg/m3 Specific Gravity Organic Matter Water Content Liquid Limit Plastic Limit Plasticity Index Activity d50 2.84 3.15% 25.7% 31.0 21.6 9.4 0.34 51 m 2.74 4.10% 29.1% 31.7 20.7 11.0 0.36 25 m 2.68 1.56% 26.5% NP NP NP NP 0.20 mm Sand Silt Clay 45% 36% 27% 33% 28% 31% 93% ~3.5% ~3.5% 16 ft to 16.75 ft 2.63 Pa NP NP SP Poorly graded sand Orange 0.928 1989 kg/m3 1478 kg/m3 2.78 1.77% 26.5% NP NP NP NP 0.90 mm 95% ~2.5% ~2.5% 57 Table A.2. Sediment Core Results for Monroe County Samples Depth (BGS) 5 ft to 7 ft 20 ft to 22 ft (fine/coarse) Critical Shear Stress Lower Yield Stress Upper Yield Stress Sediment Group Name Color Void Ratio Bulk Density 3.77 Pa 1.71 Pa 18.1 Pa SM Silty sand Orange with brown 0.681 2050 kg/m3 18.38 Pa/8.76 Pa 1.08 Pa 13.2 Pa ML Inorganic silt Orange with brown 1.89 2081 kg/m3 Dry Density 1717 kg/m3 1345 kg/m3 Specific Gravity Organic Matter Water Content Liquid Limit Plastic Limit Plasticity Index Activity d50 2.96 1.77% 16.3% 28.5 22.8 5.7 0.57 0.11 mm 2.79 4.80% 36.0% 45.8 34.3 11.5 0.44 27 m Sand 62% 25% Silt 17% 49% Clay 20% 26% 58 Table A.3. Sediment Core Results for Houston County Samples Depth (BGS) 10 ft to 12 ft 25 ft to 27 ft Critical Shear Stress >21 Pa >21 Pa Lower Yield Stress 2.67 Pa 0.309 Pa Upper Yield Stress 15.9 Pa 17.7 Pa Sediment SC CL Group Name Clayey sand Silty Clay Color Red to dark red Light gray with dark streaks Void Ratio Bulk Density 0.640 2363 kg/m3 0.416 1995 kg/m3 Dry Density 2025 kg/m3 1428 kg/m3 Specific Gravity Organic Matter Water Content Liquid Limit Plastic Limit Plasticity Index Activity d50 2.76 4.75% 18.6% 33.4 14.8 18.6 0.50 89 m 2.79 5.30% 24.7% 35.4 22.1 13.2 0.33 7 m Sand 53% 27% Silt 10% 33% Clay 37% 40% 59 Table A.4. Sediment Core Results for Stewart County Samples Depth (BGS) 10 ft to 11 ft 11 ft to 12 ft 15 ft to 17 ft Critical Shear Stress Lower Yield Stress Upper Yield Stress Sediment Group Name Color Void Ratio Bulk Density -NP NP SP Poorly graded sand Dark brown 0.924 2125 kg/m3 ---SC Clayey sand Dark brown 0.963 1927 kg/m3 1.60 Pa NP NP SC Clayey sand Light brown to white 0.943 2436 kg/m3 Dry Density 1580 kg/m3 1258 kg/m3 2080 kg/m3 Specific Gravity Organic Matter Water Content Liquid Limit Plastic Limit Plasticity Index Activity d50 2.68 0.97% 28.8% NP NP NP NP 0.35 mm 2.68 9.81% 34.7% 26.6 21.4 5.2 0.15 0.10 mm 2.75 1.22% 14.6% NP NP NP NP 0.22 mm Sand Silt Clay 92% 55% 85% 2% 10% 3% 6% 35% 12% 60 Table A.5. Sediment Core Results for Candler County Samples Depth (BGS) 10 ft to 10.5 10.5 ft to 12 ft 17 ft to ft 17.5 Critical Shear >21 Pa >21 Pa >21 Pa Stress Lower Yield 6.02 Pa 2.68 Pa -- Stress Upper Yield 15.9 Pa 17.7 Pa -- Stress Sediment MH OH OH Group Name Color Void Ratio Bulk Density Inorganic silt Bluish gray 1.07 1794 kg/m3 Organic clay Blue to greenish gray 1.85 1410 kg/m3 Organic clay Bluish gray 1.32 1848 kg/m3 Dry Density 805 kg/m3 635 kg/m3 729 kg/m3 Specific Gravity Organic Matter Water Content Liquid Limit Plastic Limit Plasticity Index Activity d50 2.54 4.09% 71.5% 71.7 61.6 10.1 0.20 2 m 2.45 5.44% 122.0% 131.5 94.6 36.9 0.38 0.2 m 2.67 3.31% 60.5% 57.8 37.1 20.8 0.49 4 m Sand 25% Silt 25% Clay 50% 1% 27% 1% 31% 98% 42% 17.5 ft to 19 ft >21 Pa 0.494 Pa 4.83 Pa MH Inorganic silt White to yellowish gray 1.23 2030 kg/m3 1144 kg/m3 2.60 3.45% 53.2% 55.9 43.3 12.6 0.18 0.5 m 7% 21% 72% 61