A thesis presented to. the faculty of. the Russ College of Engineering and Technology of Ohio University. In partial fulfillment

Transcription

1 Shear Strength Correlations for Ohio Highway Embankment Soils A thesis presented to the faculty of the Russ College of Engineering and Technology of Ohio University In partial fulfillment of the requirements for the degree Master of Science Jeffrey M. Holko March 2008

2 2 This thesis titled Shear Strength Correlations for Ohio Highway Embankment Soils by JEFFREY M. HOLKO has been approved for the Department of Civil Engineering and the Russ College of Engineering and Technology by Teruhisa Masada Professor of Civil Engineering Dennis Irwin Dean, Russ College of Engineering and Technology

3 3 ABSTRACT HOLKO, JEFFREY M., M.S., March 2008, Civil Engineering SHEAR STRENGTH CORRELATIONS FOR OHIO HIGHWAY EMBANKMENT SOILS (227 pp.) Director of Thesis: Teruhisa Masada Highway embankment design is typically done with the assistance of empirical correlations created by researchers in the past. Many of these correlations, however, were originally developed for soils not commonly found in Ohio. This lead to embankment designs that are either overly conservative or less conservative. This thesis details a study conducted that looked at the shear strength of Ohio soils. C-U triaxial compression tests were performed along with the standard penetration test, unconfined compression test, sieve analyses, and index property tests to create useful shear strength correlations for the soils found in Ohio. As a result, three levels of guidelines were proposed for all three major AASHTO soil types found in the state. Approved: Teruhisa Masada Professor of Civil Engineering

4 4 ACKNOWLEDGMENTS I would like to thank my advisor, Dr. Teruhisa Masada, for helping me get through the Master s Program. I would also like to thank him for encouraging me to work hard on this thesis and giving me the skills that will help me down the road in my career. I would also like to thank Dr. Shad Sargand, Dr. Deborah McAvoy, and Dr. Greg Springer for being part of my thesis committee, all of whom I previously took classes with. Finally, I would like to thank my parents who supported me in many ways throughout the years I worked in the Master s program.

5 5 TABLE OF CONTENTS Page ABSTRACT... 3 ACKNOWLEDGMENTS... 4 LIST OF TABLES... 8 LIST OF FIGURES CHAPTER 1: INTRODUCTION Background Objectives of Thesis Outline of Thesis CHAPTER 2: LITERATURE REVIEW General Shear Strength of Soil Pore Water Pressure of Soil Consolidation Stability of Highway Embankments Soil Classifications Review of Literature in Ohio Bedrock Glaciers Alluvium and Relief Climate Biota Standard Penetration Test SPT-General SPT-Equipment SPT-Procedure Energy Corrections Normalization of SPT-N Values Static Forces and Stresses in the SPT Existing SPT Correlations Triaxial Compression Test Setup and Equipment Back Pressure Saturation Consolidated-Drained (C-D) Test Consolidated-Undrained (C-U) Test Unconsolidated-Undrained (U-U) Test... 44

6 2.4.6 Unconfined Compression Test Statistical Analysis of Geotechnical Data CHAPTER 3: RESEARCH METHODOLOGY General Site Selection Criteria Subsurface Exploration Protocol SPT Hammer Calibration SPT Hammer Procedure Laboratory Soil Testing Protocol Soil Index Property Testing Unconfined Compression Test Triaxial Compression Test Triaxial Test Equipment C-U Test Procedure Statistical Analysis Protocol CHAPTER 4: RESEARCH DATA AND RESULTS Introduction Subsurface Exploration Work Subsurface Exploration Data for I-275 Site in Hamilton County Subsurface Exploration Data for USR 35 Site in Fayette County Subsurface Exploration Data for SR 2 Site in Lake County Subsurface Exploration Data for USR 33 Site in Athens County Subsurface Exploration Data for I-71 Site in Morrow County Laboratory Index Properties and Sieve Analyses Soil Index Properties for Site No. 1 (Hamilton County) Soil Index Properties for Site No. 2 (Fayette County) Soil Index Properties for Site No. 3 (Lake County) Soil Index Properties for Site No. 4 (Athens County) Soil Index Properties for Site No. 5 (Morrow County) Shear Strength Properties Shear Strength Properties for Site No. 1 (Hamilton County) Shear Strength Properties for Site No. 2 (Fayette County) Shear Strength Properties for Site No. 3 (Lake County) Shear Strength Properties for Site No. 4 (Athens County) Shear Strength Properties from Site No. 5 (Morrow County) CHAPTER 5: STATISTICAL ANALYSIS AND GEOTECHNICAL GUIDELINES Evaluations of Empirical Correlations SPT-N vs. Unconfined Compression Strength by Terzaghi SPT-N vs. Unconfined Compression by Dept. of Navy Effective Friction Angle vs. Plasticity Index by Terzaghi Soil Type vs. Effective Friction Angle by Dept. of Navy

7 5.2 Linear Regression Analysis A-4a Soil A-6a Soil A-7-6 Soil All Three Soil Types Nonlinear Regression A-4a Soil A-6a Soil A-7-6 Soil All Three Soil Types Multi-Variable Linear Regression Analysis A-4a Soil A-6a Soil A-7-6 Soil All Three Soil Types Preliminary Geotechnical Guidelines CHAPTER 6: SUMMARY AND CONCLUSIONS Summary Conclusions Recommendations REFERENCES APPENDIX A: SPT CALIBRATION TEST DATA & C-U TRIAXIAL COMPRESSION TEST INSTRUCTIONS APPENDIX B: SUBSURFACE EXPLORATION DATA APPENDIX C: TRIAXIAL COMPRESSION TEST DATA & PLOTS APPENDIX D: NONLINEAR CORRELATION PLOTS

8 8 LIST OF TABLES Page Table 2.1: AASHTO Classifications for Fine-Grained Materials...23 Table 2.2: (N 60 ) 1 vs Unconfined Compressive Strength...37 Table 2.3: (N 60 ) 1 vs Unconfined Compressive Strength...38 Table 2.4: Effective Friction Angle vs Plasticity Index Terzaghi...39 Table 4.1: Uncorrected SPT-N Values (Hamilton County Site)...71 Table 4.2: Hamilton County Site (N 60 ) 1 Values...73 Table 4.3: Uncorrected SPT-N Values (Fayette County Site)...75 Table 4.4: Fayette County Site (N 60 ) 1 Values...77 Table 4.5: Uncorrected SPT-N Values (Lake County Site)...78 Table 4.6: Lake County Site (N 60 ) 1 Values...79 Table 4.7: Uncorrected SPT-N Values (Athens County Site)...80 Table 4.8: Athens County Site (N 60 ) 1 Values...81 Table 4.9: Athens County Site (N 60 ) 1 Values...82 Table 4.10: AASHTO Classifications for Fine-Grained Materials...84 Table 4.11: Index Properties of Soils (Hamilton County Site)...86 Table 4.12: Sieve Analysis Results (Hamilton County Site)...86 Table 4.13: Index Properties of Soils (Fayette County Site)...87 Table 4.14: Sieve Analysis Results (Hamilton County Site)...87 Table 4.15: Index Properties of Soils (Lake County Site)...87 Table 4.16: Sieve Analysis Results (Lake County Site)...88 Table 4.17: Index Properties of Soils (Athens County Site)...88 Table 4.18: Sieve Analysis Results (Athens County Site)...88 Table 4.19: Index Properties of Soils (Morrow County Site)...89 Table 4.20: Sieve Analysis Results (Morrow County Site)...89 Table 4.21: Unconfined Compression Test Results (Hamilton County Site)...90 Table 4.22: C-U Triaxial Compression Test Results (Hamilton County Site)...90 Table 4.23: Unconfined Compression Test Results (Fayette County Site)...91 Table 4.24: C-U Triaxial Compression Test Results (Fayette County Site)...92 Table 4.25: Unconfined Compression Test Results (Lake County Site)...93 Table 4.26: C-U Triaxial Compression Test Results (Lake County Site)...93 Table 4.27: Unconfined Compression Test Results (Athens County Site)...94 Table 4.28: C-U Compression Test Results (Athens County Site)...95 Table 4.29: Unconfined Compression Test Results (Morrow County Site)...96 Table 4.30: C-U Compression Test Results (Morrow County Site)...96 Table 5.1: Evaluation of Terzaghi s Correlation for A-4a Soil...99 Table 5.2: Evaluation of Terzaghi s Correlation for A-6a Soil Table 5.3: Evaluation of Terzaghi s Correlation for A-7-6 Soil Table 5.4: Comparison of Dept. of Navy & ORITE Data Table 5.5: Correlation Paths for Data Analysis Table 5.6: Path 1 Linear Correlations for A-4a Soil...111

9 Table 5.7: Path 2 Linear Correlations for A-4a Soil Table 5.8: Path 3 Linear Correlations for A-4a Soil Table 5.9: Path 4 Linear Correlation for A-4a Soil Table 5.10: Path 5 Linear Correlations for A-4a Soil Table 5.11: Path 6 Linear Correlations for A-4a Soil Table 5.12: Path 1 Linear Correlations for A-6a Soil Table 5.13: Path 2 Linear Correlations for A-6a Soil Table 5.14: Path 3 Linear Correlations for A-6a Soil Table 5.15: Path 4 Linear Correlation for A-6a Soil Table 5.16: Path 5 Linear Correlations for A-6a Soil Table 5.17: Path 6 Linear Correlations for A-6a Soil Table 5.18: Path 1 Linear Correlations for A-7-6 Soil Table 5.19: Path 2 Linear Correlations for A-7-6 Soil Table 5.20: Path 3 Linear Correlations for A-7-6 Soil Table 5.21: Path 4 Linear Correlation for A-7-6 Soil Table 5.22: Path 5 Linear Correlations for A-7-6 Soil Table 5.23: Path 6 Linear Correlations for A-7-6 Soil Table 5.24: Path 1 Linear Correlations for All Three Soil Types Table 5.25: Path 2 Linear Correlations for All Three Soil Types Table 5.26: Path 3 Linear Correlations for All Three Soil Types Table 5.27: Path 4 Linear Correlation for All Three Soil Types Table 5.28: Path 5 Linear Correlations for All Three Soil Types Table 5.29: Path 6 Linear Correlations for All Three Soil Types Table 5.30: Exponential Model Correlations for A-4a Soil Table 5.31: Logarithmic Model Correlations for A-4a Soil Table 5.32: Power Model Correlations for A-4a Soil Table 5.33: Hyperbolic Model Correlations for A-4a Soil Table 5.34: Reciprocal Model Correlations for A-4a Soil Table 5.35: Second-Degree Polynomial Model Correlations for A-4a Soil Table 5.36: Exponential Model Correlations for A-6a Soil Table 5.37: Logarithmic Model Correlations for A-6a Soil Table 5.38: Power Model Correlations for A-6a Soil Table 5.39: Hyperbolic Model Correlations for A-6a Soil Table 5.40: Reciprocal Model Correlations for A-6a Soil Table 5.41: Second-Degree Polynomial Model Correlations for A-6a Soil Table 5.42: Exponential Model Correlations for A-7-6 Soil Table 5.43: Logarithmic Model Correlations for A-7-6 Soil Table 5.44: Power Model Correlations for A-7-6 Soil Table 5.45: Hyperbolic Model Correlations for A-7-6 Soil Table 5.46: Reciprocal Model Correlations for A-7-6 Soil Table 5.47: Second-Degree Polynomial Model Correlations for A-7-6 Soil Table 5.48: Exponential Model Correlations for All Three Soil Types Table 5.49: Logarithmic Model Correlations for All Three Soil Types Table 5.50: Power Model Correlations for All Three Soil Types

10 Table 5.51: Hyperbolic Model Correlations for All Three Soil Types Table 5.52: Reciprocal Model Correlations for All Three Soil Types Table 5.53: Second-Degree Polynomial Model Correlations for All Three Soil Types Table 5.54: Multi-Variable Linear Regression Correlations for φ and φ (A-4a Soil) Table 5.55: Other Significant Multi-Variable Linear Correlations (A-4a Soil) Table 5.56: Multi-Variable Linear Correlations with 3 and 4 Independents (A-4a Soil) Table 5.57: Multi-Variable Linear Regression Correlations for φ and φ (A-6a Soil) Table 5.58: Other Multi-Variable Correlation (A-6a Soil) Table 5.59: Multi-Variable Linear Regression Correlations for φ and φ (A-7-6 Soil) Table 5.60: Other Multi-Variable Linear Correlations (A-7-6 Soil) Table 5.61: Multi-Variable Linear Correlations with 3 and 4 Independents (A-7-6 Soil) Table 5.62: Multi-Variable Linear Regression Correlations for φ and φ (All Soil Types) Table 5.63: Other Multi-Variable Linear Correlations (All Soil Types) Table 5.64: Average and Standard Deviation of Differences for φ

11 11 LIST OF FIGURES Page Figure 2.1: Failure Envelope for Soil...17 Figure 2.2: Failure Cases for an Embankment...21 Figure 2.3: Ohio s Soil Regions...25 Figure 2.4: Soil Deposits Throughout Ohio...26 Figure 2.5: SPT Drill Rig on the Back of a Truck...28 Figure 2.6: Augering into the Soil...29 Figure 2.7: Split-Spoon Sampler Detached from the Drill Rig...29 Figure 2.8: Forces and Stresses Acting on the Split-Spoon Sampler...34 Figure 2.9: Correlation Between f and Plasticity Index Terzaghi...39 Figure 2.10: Mohr s Circle Created for Three C-U Triaxial Tests...42 Figure 2.11: Example of a p-q Diagram...44 Figure 2.12: Unconfined Compression Test Machine...45 Figure 2.13: Mohr s Circle for an Unconfined Compression Strength Test...46 Figure 3.1: Shelby Tubes Sampling Plan...54 Figure 3.2: Liquid Limit Testing Equipment...57 Figure 3.3: Unconfined Compression Test System...59 Figure 3.4: Triaxial Compression Test System...62 Figure 3.5: Correlations for Project...68 Figure 4.1: Highway Embankment Site No. 1 on I-275 (Hamilton County)...70 Figure 4.2: Original Shelby Tube Sampling Pattern...72 Figure 4.3: Modified Shelby Tube Sampling Pattern...72 Figure 4.4: Highway Embankment Site No. 2 on USR 35 (Fayette County)...74 Figure 4.5: Modified Shelby Tube Sampling Pattern...76 Figure 4.6: Highway Embankment Site No. 4 on USR 33 (Athens County)...79 Figure 4.7: Highway Embankment Site No. 5 on I-71 (Morrow County)...82 Figure 4.8: Modified Plan for Shelby Tube Sampling...84 Figure 5.1: Evaluation of Dept. of Navy Data for All Three Soil Types Figure 5.2: Evaluation of Dept. of Navy Data for A-4a Soil Figure 5.3: Evaluation of Dept. of Navy Data for A-6a Soil Figure 5.4: Evaluation of Dept. of Navy Data for A-7-6 Soil Figure 5.5: Comparison of Terzaghi & ORITE Data (All Soil) Figure 5.6: Comparison of Terzaghi & ORITE Data (A-4a Soil) Figure 5.7: Comparison of Terzaghi & ORITE Data (A-6a Soil) Figure 5.8: Comparison of Terzaghi & ORITE Data (A-7-6 Soil)...108

12 12 CHAPTER 1: INTRODUCTION 1.1 Background Highway embankments constitute some of the most common geotechnical facilities being built by civil engineers. The design and construction of highway embankments is of great importance to transportation costs and safety. When the embankment is not properly designed and/or constructed, problems such as slope instability and excessive settlement can arise. These problems of highway embankments are generally controlled by five key factors: (1) the embankment soil's shear strength, (2) the soil's moist unit weight, (3) the height of the embankment, (4) the angle of the embankment slope, and (5) the pore pressures in the soil. Das (2002) defines the shear strength of soil as the internal resistance per unit area that the soil mass can offer to resist failure and sliding along any plane inside it. There are two important shear strength parameters for soils, the angle of internal friction (φ) and cohesion (c). The φ angle indicates the degree of friction and interlocking among the soil particles, and the cohesion represents the ionic attraction and chemical cementation between the soil particles. Both of these parameters can be determined in a geotechnical laboratory by performing shear strength tests. Also, there are a few test methods that can be performed in the field to estimate shear strength properties of in-situ soils. In Ohio, highway embankments are typically built using silty and clayey soils found at/near the construction site. In some areas of Ohio, the embankments are also

13 13 constructed largely using weathered shale material. It has been known that some cohesive soils found in Ohio have low to medium shear strengths and also that weathered shale material may undergo further weathering over time. These factors require the embankment design engineers in Ohio to carefully study the on-site fill materials and specify their engineering properties, so that slope stability failure and other problems do not occur. However, in reality, detailed investigations of engineering properties of fill material are rarely conducted due to cost and time constraints. Instead, highway embankment engineers in Ohio consult outside sources such as Design Manual 7.2 by U.S. Dept. of Navy (1982), which present correlations between shear strength properties and in-situ or laboratory index test results, to estimate shear strength properties of embankment fill materials. In some embankment projects, unconfined compression strength tests may be performed on relatively undisturbed samples of the fill material to determine strength properties of the soils. These practices can lead to either very conservative or improper designing of the embankments, since the outside sources examined soils from completely different regions of the country or world. There is a need to develop reliable shear strength correlations for embankment fill materials found in Ohio. 1.2 Objectives of Thesis The study carried out in this thesis had five objectives. They are listed below: Conduct a literature review to document information relevant to the design

14 14 and construction of highway embankments in Ohio; Identify several highway embankment sites in Ohio, which can supply representative samples of major soil types existing in Ohio; Perform field soil testing and sampling at the selected highway embankment sites in Ohio; Obtain detailed engineering properties of soil samples recovered from the first five highway embankment sites by performing standard index property and shear strength tests in the laboratory; and Perform a variety of statistical analysis on the field and laboratory test data accumulated for the first five highway embankment site soils to develop correlations between shear strength properties and in-situ soil test data, between shear strength properties and index properties. 1.3 Outline of Thesis Chapter 1 laid out background information for and objectives of the study carried out in the current thesis. The background information described the current state of practice in Ohio and problems associated with it. Chapter 2 presents results of a literature review conducted, which are relevant to both highway embankment design and construction in Ohio. In this chapter, information on the types of soil found in Ohio is given. This information is essential for locating several highway embankment sites in this thesis work. Journal articles involving standard penetration test (SPT) and triaxial compression test are also discussed in Chapter 2. Chapter 3 focuses on the research

15 15 methodology utilized in the current study. The current study consisted of four phases preliminary work (literature review), field soil testing & sampling, laboratory soil testing, and statistical data analysis. This chapter describes the methodology used in each of these phases. The aim of Chapter 4 is to present all the field and laboratory test results obtained for the soils encountered at the first five highway embankment sites. The results are presented for each embankment site and include those from the SPT, the laboratory soil index tests, and the laboratory soil strength tests. The index properties here consist of specific gravity, natural moisture content, Atterberg limits (liquid limits, plastic limits), and grain size distribution. The strength tests refer to the unconfined compression and triaxial compression tests. The last part of Chapter 4 discusses briefly geographical and profile distribution of different soil types and differences in basic properties among the soils encountered in the study. Chapter 5 presents the results of a variety of statistical analysis performed on the assembled data. The chapter first describes a few different statistical approaches (linear regression, nonlinear regression, multi-variable regression) that can be taken to analyze sizable geotechnical data. Then, it presents the outcome of each analysis approach, delineating meaningful correlations produced during the process. This chapter also proposes geotechnical guidelines for highway embankments in Ohio, based on the results of the statistical data analysis. Chapter 6 provides a summary of and conclusions drawn from all phases of the thesis work. Finally, a few appendix sections follow the references list. This was necessary to provide essential supplementary materials.

16 16 CHAPTER 2: LITERATURE REVIEW The current research project is related to soil shear strength, highway embankment stability, Ohio regional geology, SPT, empirical correlations, and statistical analysis of geotechnical data. The aim of this chapter is to present both general information and research findings on these relevant topics, which were assembled through a relatively extensive literature review conducted. 2.1 General Shear Strength of Soil The basic definition of shear strength was given in Chapter 1. Also mentioned were two important parameters, the angle of internal friction (φ) and cohesion (c). Shear strength of soil is a function of the normal stress applied, the angle of internal friction, and the cohesion. The angle of internal friction describes the interparticle friction and the degree of the particles' mechanical interlocking. This characteristic depends on soil particle gradation and shape and the void ratio. Cohesion describes soil particle bonding caused by electrostatic attractions. So, with normal stress, the angle of internal friction, and cohesion, the following equation, known as the Mohr-Coulomb theory, can be used to find the shear strength of soil under a certain condition: τ = c + σ (tan φ) (2.1)

17 17 where σ = normal stress applied. This equation can be plotted on an x-y graph with shear stress on the ordinate and normal stress on the abscissa. This is known as a failure envelope and is shown in Figure 2.1. Figure 2.1: Failure Envelope for Soil In reality, however, the failure envelope is rarely a linear relationship. The degree of electrostatic attraction and cementation of cohesive particles in the soil can cause a slight concave downward curve to form instead Pore Water Pressure of Soil Saturated soils have water filling all of their void spaces. This leads to the concept of effective and normal stress. When a column of saturated soil is subjected to

18 load, the total stress is carried by both the soil particles and the water filling the voids. The equation given below describes this: 18 σ = σ + u (2.2) where σ = effective stress; and u = pore water pressure. The effective stress is the soil particle acting as a skeleton to support the load. Therefore, the effective stress is often directly proportional to the total stress. Also, the shear failure envelope formula, Equation 2.1, can be addressed in terms of effective stresses for saturated soils as: τ' = c + σ (tan φ ) (2.3) where c = the effective cohesion; and φ = the effective angle of internal friction. Many times in the field, however, soil may not be fully saturated. Bishop et al. (1960) gave the following equation to describe the shear strength of unsaturated soils: σ' = σ u a χ (u a u w ) (2.4) where u a = pore air pressure; χ = degree of saturation; and u w = pore water pressure.

19 Going back to Equation 2.3 and adding new variables, the shear strength at failure for unsaturated soil can be written as: 19 τ f = c + [σ u a + χ (u a u w )] (tan φ ) (2.5) For soil that is completely dry (χ = 0), soil that is 50% saturated, and soil that is 100% saturated, the following three equations result, respectively: τ f = c + (σ u a ) (tan φ ) (2.6) τ f = c + (σ 0.5u a 0.5u w ) (tan φ ) (2.7) τ f = c + (σ u w ) (tan φ ) (2.8) Typically, u a is less than 0 and u w is greater than 0. Experiments done by Casagrande & Hirschfeld (1960) revealed that unsaturated soil has greater shear strength than the same soil in a saturated condition. In some cases the unsaturated state may be temporary, and the soil may become eventually saturated due to surface precipitation and subsurface drainage events. Therefore, it is conservative to design highway embankments using the shear strength of saturated soils.

20 Consolidation As mentioned before, saturated soil will have part of its support coming from the soil skeleton and part of it from the pore water pressure. When loads are applied to clay that has low hydraulic conductivity, the pore pressure will increase greatly. Gradually, the pore water pressure and the effective stress will increase, resulting in a volume reduction. This can happen over a period of days, months, or years, depending on the type of soil and the corresponding drainage paths (Das 2002). This leads to a discussion on the overconsolidation ratio (OCR) for soils. The equation for OCR is given below: σ c ' OCR = (2.9) σ' where σ c = the highest past overburden stress for a soil; and σ = the current overburden stress for a soil. Essentially, if the current overburden stress for a soil is the highest stress it has ever been subjected to, then the OCR will be 1. Soils under this condition are referred to as normally consolidated. Soils with an OCR above 1 are overconsolidated. This means they have been subjected to greater stresses than the current overburden one (Das 2002). The consolidation of soils and their past stress histories are important for triaxial compression testing.

21 Stability of Highway Embankments As it was mentioned in Chapter 1, the five factors that influence stability of an embankment are the (1) shear strength of the soil used, (2) the unit weight, (3) the embankment height, (4) the slope steepness, and (5) the pore pressures within the soil. With this in mind, failure generally occurs in two ways, which are the concerns of geotechnical design engineers. The first case (Case 1) is by the physical sliding action of the embankment. This can occur when the base (below the embankment) is very firm and the embankment soil and subsoils do not interlock well together. The second case (Case 2) is by shear failure deep within the base layer. This typically occurs when the subsurface soils are softer. This type of failure happens most frequently in the short-term period after construction when excess pore pressures are still existent. Figure 2.2 diagrams each of these cases. Figure 2.2: Failure Cases for an Embankment

22 22 Another concern when building road embankments is the use of both cohesive soils and rock fragments. This could occur in an unglaciated region and presents longterm stability concerns due to gradual breakdown (i.e., weathering) of the rock fragments Soil Classifications Soils are classified into groups based upon their engineering behavior. Soil engineers currently use two systems, the United Soil Classification System (USCS) and the American Association of State Highway and Transportation Officials (AASHTO) system. The USCS first groups soils based on whether they are gravels and sands or silts and clays. Next, further sieve analysis is done on the gravels and sands to get a more detailed classification until a group name is given for the soil. There are a total of 36 group names for gravels and sands under the USCS. For silts and clays, the first divider is the liquid limit value. Next, the plasticity index and further sieve analysis is done to classify the silts into one of 35 group names. The AASHTO system is different. Soils are divided into 7 groups initially based upon sieve analysis. The groups A-1, A-2, and A-3 contain mostly granular materials. Groups A-4, A-5, A-6, and A-7-6 contain mostly silty and clayey materials. Liquid limit and plasticity index values are then used to further classify the soils. A group index number can also be used with the silty and clayey groups of soils. This number is based upon the percent of soil going through the No. 200 sieve, the liquid limit, and the plasticity index. Table 2.1 outlines these fine grained soil classifications.

23 23 Table 2.1: AASHTO Classifications for Fine-Grained Materials Group classification A-4 A-5 A-6 A-7-6 Percentage passing No. 200 sieve 36 min. 36 min. 36 min. 36 min. Liquid limit 40 max. 41 min. 40 max. 41 min. Plasticity index 10 max. 10 max. 11 min. 11 min. A-4 soils and A-6 soils can be broken down further into the categories of A-4a, A- 4b, A-6a, and A-6b. A-4a soils are A-4 soils that have between 36 and 49 percent of their particles passing through the No. 200 sieve. A-4b soils are A-4 soils that contain a minimum of 50 percent of its particles passing through the No. 200 sieve. A-4a soils contain mostly sands and silts while A-4b soils contain mostly silt. A-6a soils are A-6 soils that have a plasticity index range of A-6b soils are A-6 soils that have a plasticity index greater than Review of Literature in Ohio The soil found throughout Ohio formed over thousands of years. Bedrock, glaciers, streams, relief, climate, and biota were all contributing factors. Because of this, soil types differ throughout the state. In Figure 2.3, Ohio s seven soil regions can be seen. Lake deposit soils tend to be A-4 when looked at using the AASHTO Classification System. These are seen throughout northern and northeast Ohio. A-7-6 soils, which contain silt and clay, are found throughout central and western Ohio in the glacial till. A-6 soils are found in the eastern and southeastern portion of the state, the

24 unglaciated region. They contain silts, clays, and rock fragments. These soil deposits in Ohio are shown in Figure Bedrock Western Ohio bedrock contains mostly limestone and dolomite. Some calcareous shale can be found also. Eastern Ohio is mostly sandstone and silaceous shale Glaciers Glaciers covered all of Ohio except for the eastern and southeastern portions of the state. The unglaciated portion is shown as Soils in Sandstone and Shale, from the Ohio s Soil Regions map. Many of the deposits found in northern and western Ohio contain rock fragments that originated from Canada because of the glaciers. Portions of the state that were subjected to glaciers characterize two types of drift. The first, stratified glacial drift, is seen by layers in the soil. Geological features such as

25 Figure 2.3: Ohio s Soil Regions (Source: Johnson 1975) 25

26 26 Figure 2.4: Soil Deposits Throughout Ohio kames, eskers, and outwash plains, display this layered characteristic. The second drift, known as nonstratified, results from the four documented glacial events which occurred in Ohio. Glaciers picked up bedrock and soils along their path and deposited them when they melted in random patterns. Sand and gravel are found in these areas Alluvium and Relief Another big factor for soil development in Ohio is streams and the overland flow of water. In many lower elevated areas, near rivers, finer soil particles, such as silts and clays, are found. This is the result of years of erosion and deposition during floods. This sequence has also caused lower elevated areas to have very deep bedrock depths, which is critical for construction purposes.

27 Climate Related to erosion and deposition is the climate in Ohio. Rainfall is a big factor affecting soil types in Ohio. There is an average runoff of 12 inches per year around the state (Johnson 1975). This runoff is seen as both groundwater and streams and rivers. The groundwater works to leach many elements from bedrock into streams. Also, freezing and thawing has had an effect on the breaking of bedrock in some places. This can cause bedrock particles to mix with the nearby soils Biota Organisms have a great effect on the soil throughout Ohio. Elements such as calcium, sodium, magnesium, potassium, and carbon have been added to soil over the years from dying organisms. Different parts of the state have varying soil characteristics because of this. 2.3 Standard Penetration Test SPT-General The SPT is the oldest and most commonly used test method for subsurface exploration. The general process consists of augering a hole in the ground and then hammering a hollow tube through the soil at the bottom. The hammering is done using a large truck with a drill rig attached to the back. The resistance given off by the soil during hammering provides engineers valuable information on the characteristics of the soil. This section will describe in detail the SPT.

28 SPT-Equipment As mentioned earlier, the SPT is performed by using a drill rig attached to the back of a large truck. Figure 2.5 shows this. Figure 2.5: SPT Drill Rig on the Back of a Truck An eight inch hole is created in the ground using augers attached to the rig. Then, a split-spoon sampler is attached to the rig after removing the augers. Augers in use and a split-spoon sampler are shown in Figures 2.6 and 2.7, respectively. In some testing procedures, investigators will want to bring up soil specimens wider than those found in the split-spoon sampler. In this case, a Shelby tube will be attached to the drill rig and pushed into the soil. A Shelby tube is a hollow steel tube about 30 inches long and 3

29 inches wide. It brings to the surface undisturbed specimens that can be used for laboratory testing. 29 Figure 2.6: Augering into the Soil Figure 2.7: Split-Spoon Sampler Detached from the Drill Rig SPT-Procedure Once a hole has been augered into the ground and the split-spoon sampler is attached to the rig, a hammer is dropped onto steel rods connected to the sampler. Throughout the years, three types of hammers have been used: the donut hammer, the safety hammer, and the automatic hammer. In the procedure, the 140 pound hammer is dropped 30 inches onto the steel rods. This process is done until the sampler moves 18 inches through the ground. The blows from the hammer it takes to move the sampler through each 6 inch interval are recorded. The blow counts from the bottom two 6 inch intervals are then added together, giving the raw SPT-N value.

30 30 Despite the available hammers, the automatic hammer has become the most commonly used in recent years for reasons of safety and efficiency, as Drumright et al. (1996) points out. Their study concluded that the automatic hammer transferred about 50% more energy to the sampler than the safety hammer. The automatic hammer also reduces the probability of human error involved in the process since the rig does all of the work Energy Corrections As mentioned in the previous section, different hammers transfer different amounts of energy to the split-spoon sampler even if they each drop 140 pounds over 30 inches. Therefore, it is important to correct SPT-N values to a standard measurement. This standard measurement is the 60% free-fall energy value (N 60 ). Essentially, this is 60% of the energy that would theoretically be transferred by the hammer. In most cases, however, the transfer energy is somewhere between 60 and 100%. Therefore, the following series of equations is used to convert raw SPT-N values to N 60 : EMX = F(t) V(t) dt (2.10) where F(t) = force measured at time t; and V(t) = velocity measured at time t. The value of Equation 2.10 is then put into the numerator for Equation 2.11, given below:

31 Energy transfer ratio (ETR) = EMX / (Theoretical SPT Hammer Energy) (2.11) 31 where Theoretical SPT Hammer Energy = 0.35 kip-ft. Finally, the energy transfer ratio can be used to find N 60 in Equation N 60 = ETR x (raw SPT-N value) (2.12) 60 This process will be described more in detail in Chapter 3 and Appendix A Normalization of SPT-N Values In addition to energy transfer corrections, raw SPT-N values are also normalized using a variety of methods. Using the current overburden stress, the N 60 value is normalized to an overburden stress of 13.9 psi. There are five different normalization factors presented in this section. The first is Peck et al. (1974): C N = 0.77 log 20 σ ' 0 (2.13) where σ 0 ' = effective overburden stress (tsf). The second method is given as Terzaghi et al. (1996):

32 32 C N = 100 σ ' 0 (2.14) The third method is given as Bazaraa (1967): C N = σ 0 ' (2.15) where σ 0 ' = effective overburden stress (ksf). Also, this equation is only used when the effective overburden stress is less than or equal to 1.5 ksf. If it is above 1.5 ksf, use the following correction factor: C N = σ 0 ' (2.16) The fourth correction factor is given as Seed et al. (1975): σ ' C N = log 0 (2.17) 2000 Finally, the fifth correction factor is given as Skempton (1986):

33 C N = 2 σ 0 ' 1+ ( ) 2000 (2.18) 33 where σ 0 ' = effective overburden stress (psf) Static Forces and Stresses in the SPT To understand the static forces and stresses involved in the SPT, one must understand how each component works in the process. It can begin by looking at a simple equation, presented by Schmertmann (1979): F + W = F e + ( F o + F i ) (2.19) where F = the force transferred from the hammer to the sampler; W = the weight of the rods and sampler; F e = the reaction force given by the ground onto the bottom surface to the sampler; F o = the frictional reaction force on the outside of the sampler; and F i = the frictional reaction force on the inside of the sampler. A diagram of a split-spoon sampler used in a SPT and the forces acting on it is shown in Figure 2.8.

34 34 Figure 2.8: Forces and Stresses Acting on Split-Spoon Sampler (Source: Schmertmann 1979) Next, to better understand the process, some variables will be added to Equation An assumption is made that the unit friction acting inside and outside of the sampler is the same and will be designated with the variable f. The unit bearing pressure acting on the bottom of the sampler will be designated as q. Also, the standard splitspoon sampler s base area is 10.7 cm 2. Using these three new values, Equation 1 can be changed to the following (Schmertmann 1979): F + W = 10.7 q + ( d i + d o ) π L f (2.20)

35 where d i = inside diameter of the sampler; d o = outside diameter of the sampler; and L = the depth of the sampler into the ground. 35 Next, in Equation 2.20, q, the unit bearing pressure on the bottom of the sampler, will be replaced with the product: C 1 q c. Also, f, the unit frictional force on the sampler will be replaced with the product: C 2 f c. C 1 and C 2 are constants with no units. q c and f c are both in units of force per area. With these assumptions, Schmertmann (1979) gives the following equation: F + W = C 1 q c A e + ( d i + d o ) π L C 2 f c (2.21) Now, with the introduction of another variable, the friction ratio, R f, which is equal to f c /q c, Schmertmann (1979) gives this equation: F + W = [C 1 A e + (d i + d o ) π L C 2 R f ] q c (2.22) The left side of this equation contains the two components that will push the sampler into the ground (hammer energy and weight of equipment). The right side contains the reaction forces. As the sampler is pushed into the ground, L is the only variable on the right side (reaction force side) that changes. Likewise, as the sampler is pushed into the ground, the left side of the equation must change too. Since the weight of the equipment is fixed, then F must increase. Also, as mentioned before, the blowcount over a six inch

36 36 interval is the result of the SPT. As the sampler is pushed further into the ground, more force is used and the blowcount is increased. Therefore, this equation (Equation 2.23), given by Schmertmann (1979) is logical since F avg (the average force used through the six inch interval) and ΔL (the length of sample pushed into the ground) are directly proportional to an increase in blowcount: ΔN ~ F avg ΔL (2.23) Finally, a comparison will be made between the blowcounts experienced in the three intervals: (0 inches 6 inches), (6 inches 12 inches), and (12 inches 18 inches). If it is assumed that the average depth of the sampler while testing the top interval is 3 inches, while testing the middle interval is 9 inches, and while testing the bottom interval is 15 inches, each of these values can be put into Equations Also, replacing F on the left side of Equations with ΔN (since they are directly proportional), the following three relations can be made (Schmertmann 1979). ΔN ΔN [(10.7C = [(10.7C C C 2 2 R R f f ) q ) q c c W' W' (2.24) ΔN ΔN [(10.7C = [(10.7C C C 2 2 R R f f ) q ) q c c W' W' (2.25)

37 37 ΔN ΔN [(10.7C = [(10.7C C C 2 2 R R f f ) q c ) q c W' = 1 W' (2.26) Essentially, under the assumption the soil being testing throughout the entire 18 inch interval has the same frictional and bearing capacity characteristics, the blowcounts will increase with each lower interval. The reason they will increase is because more soil is adhering and rubbing against the inside and outside of the split-spoon sampler, even though that soil may be from a higher up interval. While testing the bottom interval, the soil from the top and middle intervals is affecting the sampler. The sampler is only affected by the soil in the top interval when this section is being tested. This explains, why, in many SPTs, the bottom 6 inch interval is highest even if the soil is very consistent Existing SPT Correlations Currently, there are a few correlations involving SPT-N values and friction angles. The first one given is between corrected SPT-N values and unconfined compressive strength for cohesive soils. This is shown in Table 2.2. Table 2.2: (N 60 ) 1 vs. Unconfined Compressive Strength (N 60 ) 1 Stiffness Strength (psi) < 2 very soft < soft medium soft stiff very stiff > 30 hard > 58 [Reference] Terzaghi et al. (1996)

38 38 Essentially, as the soil gets harder, it takes more blows to push the sampler 12 inches. Likewise, the harder and better interlocking between soil particles there is, a higher unconfined compressive strength will arise. The next set of correlations, given by Dept. of Navy (1982) in Table 2.3, uses the unconfined compressive strength again, but also factors in the plasticity of the soil. Table 2.3: (N 60 ) 1 vs. Unconfined Compressive Strength (N 60 ) 1 q u (psi) of clays (low plasticity) & clayey silts q u (psi) of clays (medium plasticity) q u (psi) of clays (high plasticity) [Reference] Dept. of Navy (1982) As previously seen in the Terzaghi et al. (1996) correlations, an increase in SPT- N value leads to an increase in unconfined compressive strength. Also, the higher the plasticity of a soil, the larger the increase in strength typically is. The last correlation given is between the effective angle of internal friction and the plasticity index. This is shown in Table 2.4 and found in Terzaghi et al. (1996). The general trend is a decreasing effective friction angle with an increasing plasticity index. Figure 2.9 shows the values of Table 2.4 in an x-y plot.

39 39 Table 2.4: Effective Friction Angle vs Plasticity Index Terzaghi Plasticity Index φ' (degrees) Note: The actual φ' value may be off by at least +/- 3 degrees Effective Friction Angle (degrees) Range Plasticity Index (%) Figure 2.9: Correlation Between φ and Plasticity Index - Terzaghi Finally, a correlation between the undrained shear strength of clay and the energy corrected SPT-N value is given in the following equation from Stroud et al. (1975):

40 s u = f 1 p a N 60 (2.27) 40 where f 1 = 0.045; and p a = 14.7 psi. This equation can only be used if the plasticity index is greater than Triaxial Compression Test The triaxial compression test is a useful method for obtaining shear strength parameters from undisturbed soil specimens. Currently, there are three types of tests used. They all use the same equipment but vary in procedure and effectiveness Setup and Equipment The test begins by extracting a soil sample between 5.6 and 7.0 tall from a Shelby tube. The specimen is then wrapped in a rubber membrane and placed on the base of a cylinder (the bottom platen). A small plastic piece (the top platen) is then placed on top of the specimen. There is drainage lines built into both plastic pieces. These drainage lines allow for saturation and consolidation to take place Back Pressure Saturation In a triaxial compression test, saturation of the specimen is done by back pressure of water through the drainage lines. As the specimen is surrounded by a rubber membrane on its sides and plastic pieces at the top and bottom, water is pushed in to fill

41 41 the void spaces. Saturation can be checked by finding the specimen s b-value. This is found by closing the drainage valves and increasing the confining pressure and recording the corresponding increase in pore pressure. This ratio is known as the b-value: b = increase in pore pressure / increase in confining pressure (2.28) full saturation. If this value is over 0.95, then it can be assumed that the specimen has reached Consolidated-Drained (C-D) Test In this test, the specimen is extracted, saturated, and then put through a consolidation process. Consolidation is done by opening drainage lines and removing any back pressure. Then, a confining pressure acts on the specimen, causing all of the pore pressures to be removed. After this, an axial stress slowly compresses the specimen with drainage valves open. Bishop et al. (1960) point out that this prevents any excess pore pressures from developing, which is important, since this test looks at the long term stability of soil when dissipation has already occurred. These tests do take a long time to carry out, however, which is why they are not used very frequently Consolidated-Undrained (C-U) Test The C-U compression test differs from the C-D test in a few ways. First, during consolidation, there is a back pressure being applied to the specimen through the drainage

42 42 lines. This is typically done for a 24 hour period. Also, because there is back pressure applied, the pore pressure in the specimen will not reduce to zero. So, after consolidation is completed, the drainage lines are closed off and an axial stress is applied to the specimen. The axial stress is applied by a strain rate that is determined from consolidation data. This type of test typically lasts for a few hours. Three different C-U tests are done on the same type of soil, each with different confining pressures. This will give three different Mohr's circles on a shear stress-axial stress diagram. Using these Mohr's circles, the φ angle can be found as shown below in Figure This was shown previously in Figure 2.1. Bishop et al. (1960) also point out that for normally consolidated silts and clays, cohesion is approximately zero. This is why it is important the effective consolidation stress be higher than the highest past overburden stress. The effective consolidation stress will be discussed more in Chapter 3. Figure 2.10: Mohr s Circles Created for Three C-U Triaxial Tests

43 43 There is also another method to find the angle of internal friction for a soil without drawing Mohr s circles as in Figure 2.7. It is done by using a p-q or p -q diagram. To construct a p-q diagram, the total major (σ 1fail ) and total minor (σ 3fail ) principal stresses at failure are put into the following equations: p = 0.5 (σ 1fail + σ 3fail ) (2.29) q = 0.5 (σ 1fail - σ 3fail ) (2.30) Then, they are plotted on an x-y graph with p on the abscissa and q on the ordinate. The same procedure can be used for effective stresses. Figure 2.11 shows an example of a p-q diagram. In this diagram, the angle between the best-fit line and the abscissa will be referred to as α. The angle of internal friction can be found by the following equation in units of degrees. φ = sin -1 ( tan α ) (2.31)

44 44 Figure 2.11: Example of a p-q Diagram Unconsolidated-Undrained (U-U) Test This is the third type of triaxial compression test in use. It is typically used on undisturbed samples of clay and silt to measure the existing strength of natural strata (Bishop et al. 1960). After back pressure saturation is complete, the drainage lines are closed off to the specimen and loading begins. Deviator stress is applied until the specimen fails, at which point the test is over. This type of test is done very fast. Also, in a U-U test, the shear strength is independent of the confining pressure. Because of this, the total stress Mohr s circles will produce an angle of internal friction of zero.

45 Unconfined Compression Test The unconfined compression (UC) test is similar to the triaxial compression test except for the lack of a confining pressure. It is performed using a soil specimen of similar size. The specimen is placed between two loading platens and then stress is applied to compress the soil. A typical machine used for this test is shown in Figure Since there is no confining pressure and no membrane around the specimen, only cohesive soils can be used for this. During a test, a stress-strain curve will be created. The highest stress applied on this curve is defined as the unconfined compressive strength (q u ). Plotting this on a Mohr's circle diagram is shown below in Figure The undrained shear strength of the soil is simply the unconfined compression strength divided in half. Figure 2.12: Unconfined Compression Test Machine (Source: Sieken)

46 46 Figure 2.13: Mohr s Circle for an Unconfined Compression Strength Test 2.5 Statistical Analysis of Geotechnical Data Researchers have been compiling and analyzing geotechnical data for many years to provide supporting evidences for new theories, develop new useful empirical correlations, or validate existing theories/relationships. Several different mathematical functions (or models) were applied to best represent the correlations existing among geotechnical data. Linear functions were used to represent the relationships between the plasticity index and the liquid limit in the plasticity chart (Casagrande 1932), between the plasticity index and % clay (Skempton 1953), between the specific discharge and the hydraulic gradient for clean sands in the laminar flow domain (Darcy 1856), and between the shear strength and normal stress in the Mohr-Coulomb failure criterion. Terzaghi et al. (1996) examined the relationship between the effective angle of friction and the plasticity index for a wide range of fine-grained soils and summarized the results by a nonlinear function. Semi-log functions were relied upon to describe the relationships between the moisture content and the blows by the falling cup device (for the determination of liquid limit) and

47 47 between the void ratio and effective stress for clays. Duncan (1980) utilized a hyperbolic function to express the initial tangent modulus of soil in terms of the deviatoric stress and axial strain. Recently, Masada et al. (2006) analyzed a set of laboratory resilient modulus test data obtained for fine-grained soils in Ohio and concluded that a hyperbolic function can describe the correlation between the resilient modulus and deviatoric stress well. Other functions (ex. exponential) were also utilized by geotechnical researchers in the past to describe, for example, the relationship between the specific discharge and the hydraulic gradient for granular soils in the turbulent flow domain and the relationship between effective friction angle and the SPT-N 60 for granular soils (Schmertmann 1975).

48 48 CHAPTER 3: RESEARCH METHODOLOGY 3.1 General The current research work was performed by the ORITE and a private geotechnical consulting firm, BBCM Engineering, Inc. (Dublin, OH). The ORITE was the leading researcher institution, and BBCM served as a subcontractor. This arrangement was necessary, since the ORITE does not possess any capability to perform augering, SPT, and Shelby tube sampling. Also, the joint venture between the academic unit and the industry was encouraged by the sponsor of the project (Ohio Department of Transportation) for maximizing benefits of the research to the engineering community. The project consisted of four phases --- preparations phase, field testing/sampling phase, laboratory soil testing phase, and data analysis phase. This chapter describes general methodology employed in each phase and roles played by each member of the research team (ORITE, BBCM). 3.2 Site Selection Criteria A set of criteria was established in the preparations phase of the current project to select a total of nine (9) sites in Ohio, which can represent a range of highway embankment soil conditions typically detected in Ohio. The criteria were: Criterion #1: Criterion #2: Criterion #3: Embankment fill height over 25 ft Site location on major highway Site estimated to consist of desired soil type(s)

49 49 Criterion #4: Site location highly recommended by ODOT district geotechnical engineers or subcontractor Criterion #5: Site location in unique geographical and/or geological area within the state Criterion #6: Criterion #7: Criterion #8: A lack of gravel size particles and rock fragments No guardrails close to the pavement edge Relatively large and level grassed area The first three criteria were proposed during the initial meeting between the Ohio Department of Transportation and the ORITE. Criterion #5 was added by the ORITE researchers after studying geological maps of Ohio. The last four criteria were devised by the subcontractor (BBCM) to minimize potential problems during the planned field soil testing/sampling work. It was decided during the initial meeting that the embankment age will not be an issue. It was also decided early on that any of the sites selected should not have a history of slope instability or other problems. This was to ensure safe access to the site, reliable SPT results, and high quality soil samples. Any embankment site chosen for the project should have an overall height of at least 25 ft, so that a relatively large volume of SPT results can be collected within the soil fill. SPT should not be performed into the foundation soil layers. The sites should be located mostly on major highways such as

50 50 Interstate highways and U.S. routes, due to their relative importance over lower class roadways. As part of the preparations phase, the ORITE contacted the ODOT district geotechnical engineer in each ODOT district to briefly describe the research project and request for a few recommended highway embankment sites in the region. Also, geotechnical engineers at BBCM, who have supervised subsurface exploration work at numerous locations in Ohio, were consulted to come up with a list of recommended highway embankment sites. Any sites recommended highly by the ODOT geotechnical engineers and/or BBCM geotechnical engineers obviously received a serious consideration in the current project. According to ODOT, the three major soil types (in terms of the AASHTO classification system) found in Ohio are A-4, A-6, and A-7-6. Therefore, the sites selected for the project must consist of these major soil types. The sites should be spread throughout the state, covering the northeastern, northwestern, central, southeastern, and southwestern regions. As it was presented in Chapter 2, geological setting in the state of Ohio is divided into glaciated and unglaciated regions. The ODOT Districts 5, 9, 10, and 11 are mostly in the unglaciated region, while other ODOT Districts are in the glaciated plains. It has been found in the past that silty A-4 soils (lake deposits) are abundant in the area surrounding the shorelines of Lake Erie. Clayey A-7-6 soils have been found in the northwestern portion of the state (ODOT Districts 1 and 2). A-6 soils, which are silty clay with possible rock fragments, can be found in the unglaciated eastern and southeastern parts of the state. Based on these reports, it may be ideal to have two sites in

51 the A-4 soils (lake deposits) zone, at least three sites in the unglaciated region, and three or four sites in the glaciated region Subsurface Exploration Protocol All the subsurface exploration work in this project was conducted by the subcontractor (BBCM), with the ORITE researchers involved as decision makers. During the initial meeting, it was decided that a dedicated truck-mounted drilling rig equipped with a calibrated automatic hammer should be assigned to the project, along with dedicated crew, to minimize undesirable equipment-to-equipment or human-factor variability during the SPT SPT Hammer Calibration The automatic hammer attached to the BBCM drilling rig identified for the project was calibrated by GRL Engineers, Inc. (Cleveland, Ohio), prior to the field work at the first site. The calibration testing was done by hammering the sampler into the ground according to the normal SPT procedure. AWJ rods were used to connect the automatic hammer to the split barrel sampler. Hammering was done at depths of 1, 4.5, 9, 14, and 19 ft with corresponding AWJ rod lengths of 6, 9, 14, 19, and 24 ft, respectively. As mentioned in Chapter 2, the SPT was done by dropping a 140-lb hammer over 30 inches. Assuming no frictional losses, this operation should produce 0.35 kip-ft of free-fall energy.

52 52 GRL Engineers used a PAK model Pile Driving Analyzer to measure the strain and acceleration exerted on the sampler. The analyzer converted the strain and acceleration measurements into force and velocity, so that the results could be easily interpreted. The average energy transferred from the hammer to the sampler was 0.290, 0.277, 0.277, 0.290, and kip-ft, for the depths of 1, 4.5, 9, 14, and 19 ft, respectively. Dividing each of the above energy values by 0.35 kip-ft gives the transfer ratio at each depth. The average energy transfer ratio for the five depths resulted at (81.7%). This means that about 81.7% of the free-fall energy generated by dropping the hammer weight was transferred to the sampler as it was pushed into the ground. The calibration test report by GRL Engineers is included in Appendix A SPT Hammer Procedure The ORITE researchers decided to have at each field site a continuous SPT performed through embankment soil fill to the depth of 25 ft. This was necessary to collect comprehensive subsurface soil profile data, which can be used to establish detailed soil boring logs and aid in selecting the depth ranges for soil sampling. In a typical geotechnical project, SPT is performed at 5 ft intervals. A standard split-spoon sampler, with a retainer, inside liners, and sampling length of 18 inches, was used during the SPT. The hammering was done automatically for the depth ranges of 1.0 to 2.5, 2.5 to 4.0, 4.0 to 5.5, 5.5 to 7.0, 7.0 to 8.5, 8.5 to 10.0, 10.0 to 11.5, 11.5 to 13.0, 13.0 to 14.5, 14.5 to 16.0, 16.0 to 17.5, 17.5 to 19.0, 19.0 to 20.5, 20.5 to 22.0, 22.0 to 23.5, and 23.5 to 25.0 ft.

53 53 During the SPT, the BBCM drill team kept a soil boring log. The blow counts over each 18-inch penetration interval were recorded. Whenever the sampler was brought to the ground surface after each SPT, it was split-open to reveal the types and thicknesses of soil layers present at the tested depth range. While logging the soils, a hand penetrometer tip was pushed against each soil layer to record the estimated bearing capacity value in tons per square foot (tsf). Soil samples were broken up into sections and placed into separate sealed glass jars for transportation and later inspections in the laboratory. Once the continuous SPT was performed, the depth vs. raw SPT blow counts data was quickly analyzed by the ORITE team. Since the main objective of the current project was to correlate SPT N-values to other soil properties, it is desirable to find three depth ranges that differ from each other in terms of SPT-N values. For example, depths at which the SPT-N value was approximately equal to 10, 20, and 30 might be suitable for obtaining Shelby tube samples. Here, it is better to rely on the SPT-N values corrected for the overburden soil pressure effect. Several different correction methods were described for the SPT-N value in Chapter 2. To complete the field work at any site, four soil sampling holes were placed about 3 ft away from the location of the continuous SPT. The short offset distance was necessary to stay close to the soil conditions encountered during the continuous SPTs. This arrangement would assure reliable input data when seeking correlations between the SPT-N values and the other soil properties. Figure 3.1 shows the ideal Shelby tube sampling plan to be used in the field.

54 54 The procedure for pushing three Shelby tube samples in each soil sampling hole was as follows. First, the hole was located according to the plan shown in Figure 3.1 Next, the hole was augered with continuous-flight augers to the shallowest depth at which soil sampling was planned. At that point, the BBCM drill team cleaned out the bottom of the hole, attached a Shelby tube to the tip of the AWJ rods, and pushed the Shelby tube hydraulically 2 ft into the ground. It was preferable that the Shelby tube be pushed 2 ft into the ground. However, this did not always happen since some soils gave a great deal Figure 3.1: Shelby Tubes Sampling Plan of resistance to the Shelby tube penetration. If this was the case, then the drill team pushed the tube as deep as possible. After the first Shelby tube was recovered to the ground surface, removed from the rods, and labeled properly (along with its actual soil sample length), the hole was augered down to the middle sampling depth. Here, the

55 55 second Shelby tube was pushed hydraulically. Next, augering continued down to the final depth, where the third Shelby tube captured a relatively undisturbed soil sample. The Shelby tube sampling procedure described above was repeated precisely in the three remaining holes. When soil sampling efforts were not successful (low sample recovery, crushing of Shelby tube) at one of the four hole locations, an alternative hole was randomly located near the initial continuous SPT hole to progress through the soil sampling program. Since there were three tubes obtained per hole, a total of twelve Shelby tubes were recovered. At the end of the soil sampling work, both ends of each Shelby tube were sealed with wax and tight plastic caps. Nine of the tubes (three tubes at each sampling depth) were transported to the ORITE laboratory at Ohio University. The remaining three tubes were kept by BBCM and taken to their soils laboratory. It was important that each Shelby tube retained by the ORITE team had a soil recovery length of 10 inches or more. This was because at least one good triaxial test specimen had to be trimmed out of the soil inside each tube to perform a C-U triaxial test. A triaxial compression test specimen should have a length of approximately 6 inches. Here, the actual recovery should be much more than 6 inches, since the sample ends were usually uneven and somewhat disturbed from trimming. With this requirement met, three C-U triaxial tests could be performed at each soil sampling depth. Each tube taken by BBCM also had to have a soil recovery length of at least 10 inches, so that they could secure a 6- inch length soil specimen for unconfined compression strength test and use the rest for index property tests.

56 Laboratory Soil Testing Protocol In the current research project, a wide variety of laboratory soil tests was performed by BBCM and the ORITE for soil samples recovered from each highway embankment site. The joint efforts were necessary to complete a large number of tests within a reasonable amount of time. The ORITE research team performed triaxial compression tests, while BBCM focused mainly on index property tests Soil Index Property Testing The soil index property tests, as mentioned in Chapter 2, included the specific gravity test, natural moisture content test, liquid limit test, plastic limit test, mechanical sieve analysis, and hydrometer test. A laboratory technician at BBCM measured the specific gravity of selected soil samples according to the ASTM D-854 method. Split spoon sampler soil samples, broken up and sealed in jars, were used to determine the natural moisture content of the soils found at each field site. Liquid limit and plastic limit tests were both performed according to the ASTM D-4318 protocol. The falling cup method was used to determine the liquid limit. Figure 3.2 shows the liquid limit test equipment. Once the Atterberg limits were found, they provided the plasticity index.

57 57 Figure 3.2: Liquid Limit Testing Equipment (Source: Bowles 1992) Grain size analysis consisted of the mechanical sieve analysis and the hydrometer test. The mechanical sieve analysis was performed according to the ASTM D-422 method. The main outcome of this test was the grain size distribution curve, which provided percent gravel, percent sand, percent fines (silt + clay), and key particle sizes (D 60, D 30, and D 10 ). The hydrometer test was conducted by following the ASTM D-421

58 58 test method. This test provided further breakdowns of the fines into silt and clay size particles. The results from the Atterberg limit and grain size analysis tests were then combined together to arrive at the AASHTO soil classification designation for each soil sample tested. For soils classified as either A-4 or A-6, the additional steps proposed by ODOT were applied to group them into A-4a, A-4b, A-6a, or A-6b. The soil index property test reports issued by BBCM are included in Appendix C Unconfined Compression Test In addition to the index property tests, BBCM performed unconfined compression tests on Shelby tube specimens recovered from each highway embankment site. The unconfined compression test was performed according to the ASTM D-2166 method. Figure 3.3 shows an unconfined compression test machine typically used by soil testing laboratories. Each test was performed in a strain-controlled mode. The loading rate typically ranged between and inches per minute. The test produced load vs. displacement data until a sign of specimen failure was observed. The raw data were then converted into stress vs. strain plots, with unconfined compression strength (undrained shear strength) and strain at failure. The additional data obtained during each unconfined compression test included moist and dry unit weights, moisture content, degree of saturation, and void ratio.

59 59 Figure 3.3: Unconfined Compression Test Machine Triaxial Compression Test Accurate determination of shear strength properties of embankment soils commonly encountered in Ohio constituted one of the most important tasks identified in the current research project. The ORITE research team performed all the consolidatedundrained (C-U) triaxial compression tests in the project, using the Shelby tube soil samples recovered from the first five highway embankment sites. The following sections provide details on the triaxial test equipment and test procedures.

60 Triaxial Test Equipment The triaxial compression test system housed in the ORITE laboratory comprised of many state-of-the-art pieces of equipment to permit a careful and high-precision C-U test to be carried out by trained laboratory personnel. The important system components are listed below: Vacuum Pump Water Tank Load Frame This was used to pull air out of the soil specimen and deair water. This cylinder shaped tank was used to hold deaired water. This device pressed a loading piston downward against the soil test specimen to load it axially. Test Cell This cylinder shaped cell held the soil test specimen and pressurized water around it. The top plate allowed a loading piston to penetrate into the cell. The bottom assembly connected pressure transducers and drainage/saturation lines to the soil specimen or chamber water. Sensors (a). Linear Position Sensor (LPS): This sensor measured the axial displacement of the soil specimen during the test. (b). Load Cell: This sensor measured the reaction force on the soil specimen as it is compressed. (c). Pore Pressure Transducer: This sensor measured the pore pressure within the soil specimen.

61 61 (d). Cell Pressure Transducer: This sensor measured the confining pressure surrounding the soil specimen. Panel This multi-function unit contained a vacuum regulator and pressure regulator. Three large burettes mounted on the panel held pressurized water and were connected to the cell water and soil specimen. It controlled the confining pressure and back pressure during testing. Also, the panel has tubes connecting it to a tap water and air pressure supply. Others (a). Network Module: This device regulates the flow of commands and data between the computer and the sensors on the load frame. (b). PC: A standard IBM-compatible PC ran a special software prepared by the manufacturer of the triaxial test system, so that the sensor readings acquisition and test management will be automatic once the soil specimen is conditioned in the test cell. Figure 3.4 shows a photograph of a compression test setup and the equipment used C-U Test Procedure The C-U triaxial compression test procedure followed the guidelines set fourth by ASTM Standard D The guidelines, however, were fairly general in their descriptions. Major efforts were made to translate some of the specifications outlined in

62 the ASTM method to practical steps applicable to the actual test equipment being used in the laboratory. The following list maps out the steps taken in running the C-U test: 62 Figure 3.4: Triaxial Compression Test System

63 63 Step-1: Fill the water tank with tap water up to about 1 inch below the top. Apply a vacuum pressure of 13 psi to the water tank for 4 hours to remove most of the dissolved air present in the tap water. Step-2: Initiate the specimen extraction process by cutting the Shelby tube into an approximate 6 inch length section, using a circular blade saw. ASTM guidelines require the actual soil specimen length to be between 5.6 and 7.0 inches. They also require the diameter of the test specimen to be close to 2.8 inches. Make sure that the inner diameter of the Shelby tube is indeed approximately 2.8 inches. Mount the Shelby tube section on a hydraulic jacking device. Extract the soil specimen out of the tube (in the direction the soil entered into the tube in the field) by slowly advancing the hydraulic piston. Care is needed to prevent bending or fracturing of the soil specimen during the extraction process. Step-3: If the specimen does not have smooth and flat end surfaces, place it sideway on a special curved block and slice off thin uneven sections. Obtain with a caliper the average diameter and length of the soil specimen. Weigh the specimen on an electronic scale, so that the initial moist unit weight is known. Use a small amount of soil remaining inside the tube or the trimmed portion of the soil specimen to determine the initial (natural) moisture content of the soil.

64 64 Step-4: Place the soil specimen on the bottom platen attached to the base assembly of the test cell. Position the top platen on top of the soil specimen. Envelop the specimen fully with a thin rubber membrane. Seal the ends of the membrane using rubber O-rings. Assemble the test cell by placing the plexiglass cylinder cell wall and the top assembly. Attach flexible tubings coming from the panel to the base assembly ports. Fill the space between the specimen and the cell wall with the de-aired water by applying positive pressure to the water in the water tank. Observe that the water flows out of the tube connected to the top assembly. Step-5: Apply a positive water pressure to the bottom of the soil specimen and a negative air (or vacuum) pressure to the top of the soil specimen. This is done to remove air out of the specimen during the specimen saturation stage. Step-6: Initiate the saturation process by applying back pressure to the top and bottom of the specimen. This is done by setting the confining pressure (pressure applied to the chamber water) to 32.0 psi and the water pressure going into the top and bottom of the specimen to 30.0 psi. Leave the specimen subjected to this state for a period of time until a B-value of 0.95 is reached. This is done by monitoring the pore water pressure reading frequently. B-value is calculated by dividing the change in the pore water pressure reading by the chamber pressure.

65 65 Step-7: Once the specimen is saturated, the consolidation process can be started. Increase the confining pressure so that the difference between the confining pressure and back pressure matches the desired effective consolidation stress. The effective consolidation pressure should be equal to or higher than the estimated overburden pressure that existed in the field. This is to assure that the soil specimen will not exhibit overconsolidated behaviors during the test. The specimen is left in this state for 24 hours. Record the burette readings and the pore pressure reading at specified times. Also, measure the axial compression experienced by the specimen using a caliper. Use these data to verify the completion of the consolidation process and determine the loading rate for the triaxial test based on the t 50 value. The ASTM D-4767 states that the loading rate should be set at 4% divided by the t 50 value magnified by 10, so that pore pressure can achieve equilibrium during each increment of the test. Step-8: After consolidating the soil specimen, close off the drainage paths in and out of the specimen. Bring the loading piston down, so that its tip is in contact with the top platen. Go into the computer software and set the loading rate to the specified value. Begin the loading process. During the test, the computer will record automatically all of the sensor readings frequently and update key plots on the computer screen. The actual test duration will depend on the loading rate and behaviors of the soil specimen. According to ASTM D-4767, the test is to be

66 terminated at 15% axial strain, a 20% decrease in deviatoric stress, or 5% additional strain beyond the deviatoric stress peak. 66 Step-9: Shortly after the triaxial test, drain the water from the test cell. Disassemble the cell and carefully remove the soil specimen. Photograph and sketch the final conditions of the test specimen. If a shear plane is visible, measure its inclination angle. Determine the final moisture content of the soil by placing the entire specimen in the laboratory oven. This completes the basic protocol for running the C-U triaxial compression test. More detailed instructions can be found in Appendix A. 3.5 Statistical Analysis Protocol The main objective of the current research work was to develop for highway embankment soils commonly found in Ohio correlations between shear strength properties and in-situ soil test data and correlations between shear strength properties and index properties. This was done by first analyzing detailed analysis of each triaxial test data, grouping the triaxial and all of the other test data (including the original and corrected SPT-N values) according to the AASHTO soil types, and performing a variety of statistical analyses on the assembled data using computer software. Data produced by each C-U triaxial test were processed to produce p-q and p -q diagrams. A linear curve was fit to the data points on each diagram, providing an

67 67 equation and r 2 value. The constants in the equations (m, α, m, and α ) were converted to actual shear strength parameters (c, φ, c, and φ ). Before getting into the statistical analysis, the data produced in the project were first used to examine the previously published correlation between plasticity index (PI) and effective friction angle (φ ) by Terzaghi et al. (1996) and between unconfined compression strength and SPT-N value by Dept. of Navy (1982). This was important, because many practicing geotechnical engineers in Ohio had relied on these published relationships to estimate shear strength properties of Ohio soils for their highway embankment design work. For each data set grouped for a specific AASHTO soil type, simple or X-Y correlations were sought along several different paths, which are listed below, and shown again in Figure 3.5: Path 1 - Correlations between SPT-N values and index properties Path 2 Correlations between triaxial test results and index properties Path 3 Correlations between triaxial test results and unconfined compression strength Path 4 Correlations between unconfined compression strength and SPT-N values Path 5 Correlations between unconfined compression strength and index properties Path 6 Correlations between triaxial test results and SPT-N values

68 68 Figure 3.5: Correlations for Project With the aid of computer software, many mathematical models (such as linear, 2 nd degree polynomial, logarithmic, power, exponential, hyperbolic, and reciprocal) could be easily applied to the data set to identify the best model and strongest correlations that appear to exist for the shear strength characteristics of major highway embankment soils in Ohio. Once the simple correlations are exhausted, linear multi-variable correlations can be explored within each data set. Incremental forward scheme was adopted to yield the best correlation case. Further details on the analytical phase and the results of the statistical data analysis can be both found in Chapter 5.

69 69 CHAPTER 4: RESEARCH DATA AND RESULTS 4.1 Introduction The data for the current research project were mainly produced during the field subsurface exploration and laboratory soil testing phases. In this chapter, the results from these two major activities will be presented in detail for the first five highway embankment sites in Ohio. The results will be presented in three separate sections. The first section will focus on the subsurface exploration work. The second section will provide the index properties of the soils determined at the BBCM soil laboratory. The third section will present strength test results of the soil samples, which includes unconfined compression test results by BBCM and C-U triaxial test results by the ORITE. Each section will have a subsection laying out the results from an individual site. The order of the sites presented will be: (1) Interstate 275 site in Hamilton County, (2) U.S. Route 35 site in Fayette County, (3) State Route 2 site in Lake County, (4) U.S. Route 33 site in Athens County, and (5) Interstate 71 site in Morrow County. A brief description and pictures taken at each site will accompany the results. 4.2 Subsurface Exploration Work Subsurface Exploration Data for I-275 Site in Hamilton County The first highway embankment site can be found in the southwestern part of Ohio, near the Ohio River. The site selected was located alongside Interstate Highway 275,

70 70 about 10 miles northwest of downtown Cincinnati, in Hamilton County. A photograph showing a general view of the site is given in Figure 4.1. This was one of a few sites highly recommended for the current project by the ODOT geotechnical engineer serving this area. Figure 4.1: Highway Embankment Site No. 1 on I-275 (Hamilton County) Standard penetration tests (SPT) were performed continuously down to a depth of 19 ft, using an automatic SPT hammer attached to the BBCM drilling rig. The planned maximum depth of 25 ft could not be reached due to weathered shale found from the depth of 16.5 ft. This was surprising to the field team, because the plan drawings obtained from the ODOT did not indicate the bedrock to be located at such a shallow

71 71 depth. During the filed work, the split-spoon barrel brought samples of relatively uniform silty clay soil to the ground surface. No water table was encountered during the field work. The original (or uncorrected) SPT-N values are tabulated against depth in Table 4.1. The SPT-N value showed a general trend of increasing steadily with depth. Table 4.1: Uncorrected SPT-N Values (Hamilton County Site) Depth Range (ft) Uncorrected SPT-N Value (blows/ft) Based on the original SPT blow counts, it was decided that Shelby tubes would be pushed at the depth ranges of 2.5 to 4.5, 4.5 to 6.5, and 10.0 to 12.0 ft. As it was mentioned earlier, correlations with N values is a major objective of this project. Therefore, selecting a wide array of values is most desirable. Here, values of 7, 13, and 20 can be used for making correlations since they correspond to the soil that will be brought up by the Shelby tubes. As it was discussed in Chapter 3, the plan shown in Figure 4.2 represented the ideal pattern in which Shelby tube soil samples should be recovered at this site. However, when Hole A was drilled, a large amount of gravel was recovered. This forced

72 a change in the plan. The modified Shelby tube sampling plan, shown in Figure 4.3, was then adapted and executed to produce all twelve tube samples. 72 A 3 C 3 SPT 3 Hole D 3 B Figure 4.2: Original Shelby Tube Sampling Pattern SPT Hole 3 A 3 C 3 B 3 D Figure 4.3: Modified Shelby Tube Sampling Pattern

73 73 After extracting all twelve Shelby tubes, the ORITE personnel inspected each tube and selected nine of them to go to the ORITE laboratory. The soil recovery and notes on each tube kept by ORITE is included in Appendix B as Table B.1. Also included in Appendix B are five more pictures taken at the first site (Figures B.1-5). After the field testing was completed, a series of corrections were done to the original SPT-N values. The first correction made was for the energy transfer to the automatic hammer attached to the SPT truck. This correction was already discussed back in Chapter 2. Also, details on the automatic hammer calibration are given in Appendix A. Next, five more corrections were performed. These are the Peck, Terzaghi, Bazaraa, Seed et al., and Skempton corrections. These correction methods were also given in Chapter 2. Table 4.2 presents the corrected SPT-N values from the I-275 site. According to the table, the correction method by Seed et al. produced values closest to the overall average. Detailed information on the correction factors given above is included in Appendix B as Tables B.2-4. Depth (ft) Original SPT-N Table 4.2: Hamilton County Site (N 60 ) 1 Values Energy Correction Only Peck Terzaghi Bazaraa Seed et al. Skempton [Note] The value Avg. is simply the rounded average of the five previous columns (Peck, Terzaghi, Bazaraa, Seed et. al., and Skempton. The actual values from Peck, Terzaghi, Bazaraa, Seed et. al., and Skempton can be found in Table B.4 in Appendix B as the Summary for each depth range. This is how the Corrected N-values tables will work in the following subsections also. Avg.

74 Subsurface Exploration Data for USR 35 Site in Fayette County The second highway embankment site can be found in the central-southwestern part of Ohio in Fayette County. This site, near Jeffersonville, was located on the old USR 35 embankment about 100 ft away from a bridge abutment. The abutment supported a bridge that went over the new USR 35. Figure 4.4 shows the general view of the site. This site was identified as one of the potential sites, while searching for a site in the central region of Ohio. It was recommended strongly by BBCM based on their prior drilling in this area. Figure 4.4: Highway Embankment Site No. 2 on USR 35 (Fayette County)

75 75 Standard penetration tests (SPT) were conducted to a depth of 25 ft. During the filed work, the split-spoon barrel brought samples of hard silt with clay and sand to the ground surface. No water table was encountered during the field work. The original (or uncorrected) SPT-N values are tabulated against depth in Table 4.3. The SPT-N value fluctuated mostly between 10 and 25 in the top 20-ft depth, increased with depth from the depth of 20 to 23 ft, and declined to 20 at the maximum depth of 25 ft. Table 4.3: Uncorrected SPT-N Values (Fayette County Site) Depth Range (ft) Uncorrected SPT-N Value (blows/ft) Based on the SPT-N values, it was decided to utilize Shelby tubes at depth ranges of 5.5 to 7.5, 8.5 to 10.5, and 14.5 to 16.5 ft. At these depths, the original SPT-N values were 18, 23, and 10. The original plan for the Shelby tube sampling was shown previously in Figure 4.2. While pushing the tubes, Holes A and B produced good recovery at each depth. However, Hole C gave very little recovery at the depth range of

76 to 10.5 ft and no recovery at the 14.5 to 16.5 ft range. This led the field team to modify the plan to the one illustrated in Figure 4.5, by adding the fifth sampling hole (Hole E). This hole was located far from Hole C to avoid more problems with soil in that area. Holes D and E gave moderate recoveries at each depth range. In total, fifteen Shelby tubes were recovered at the second site. Nine of the tubes with good sample recovery were kept by the ORITE. The soil recovery and notes on each tube are included in Appendix B as Table B.5. After field testing was complete, a series of corrections were applied to the original SPT-N values. This was done in a Figure 4.5: Modified Shelby Tube Sampling Pattern similar manner to the ones for the first (Hamilton County) site. Table 4.4 presents the corrected SPT-N values from the Fayette County site. Detailed information on the correction factors is included in Appendix B as Tables B.6-8.

77 77 Table 4.4: Fayette County Site (N 60 ) 1 Values Depth (ft) Original SPT-N Energy Correction Only Peck Terzaghi Bazaraa Seed et. al. Skempton Avg Subsurface Exploration Data for SR 2 Site in Lake County The third highway embankment site can be found in northeast Ohio, along Lake Erie, in Lake County. The site was located on an embankment supporting two bridges carrying State Route 2 over State Route 615. No site photographs are available for this site. This site was placed in this region with an intention of examining A-4 soils that are abundant along the shores of Lake Erie. Standard penetration tests (SPT) were performed continuously down to a depth of 25 ft, as planned. During the filed work, the split-spoon barrel brought samples of hard silt and clay to the ground surface. No water table was encountered during the field work. The uncorrected SPT-N value at each depth range is listed in Table 4.5. The raw SPT-N values fluctuated between 10 and 30 without exhibiting any clear trend with depth.

78 78 Table 4.5: Uncorrected SPT-N Values (Lake County Site) Depth Range (ft) Uncorrected SPT-N Value (blows/ft) Based on the original SPT blow counts, it was decided to obtain Shelby tube samples at depth ranges of 1.0 to 3.0, 4.0 to 6.0, 14.0 to 16.0 ft. At these depths, the uncorrected SPT-N values were 10, 25, and 16, respectively. Shelby tube soil sampling work went according to the plan (illustrated in Figure 4.2) with very few problems and good recovery for each tube. Nine of the twelve total tubes were retained by the ORITE. The recovery and notes on these tubes are included in Appendix B in Table B.9. After the completion of the field work, corrections were applied to the original SPT-N values. The new, corrected SPT N-values for the Lake County site are shown below in Table 4.6. Detailed information on the correction factors is included in Appendix B as Tables B

79 79 Table 4.6: Lake County Site (N 60 ) 1 Values Depth (ft) Original SPT-N Energy Correction Only Peck Terzaghi Bazaraa Seed et. al. Skempton Avg Subsurface Exploration Data for USR 33 Site in Athens County The fourth highway embankment site was located along U.S. Route 33 in Athens County. It was on top of a large embankment, approximately five miles south of Athens on a two-lane portion of the road. Figure 4.6 provides a general view of the site location. This site was identified jointly with the ODOT District 10 Office in an attempt to examine typical embankment materials in the unglaciated region of Ohio. Figure 4.6: Highway Embankment Site No. 4 on USR 33 (Athens County)

80 80 Field work at this site started with a continuous SPT to a depth of 25 ft, as usual. This went forward with no problems. A few different types of soil (or different mixtures of clays and silts) were encountered during the subsurface exploration work. No water table was encountered during the field work. The uncorrected SPT-N values recorded at this site are tabulated against depth in Table 4.7. The raw SPT-N values fluctuated between 15 and 45 without exhibiting any clear trend with depth. Based on the SPT blow counts, it was decided that Shelby tubes be pushed at depth ranges of 4.5 to 6.5, 8.0 to 10.0, and 19.0 to 21.0 ft. This gave the uncorrected SPT-N values of 33, 17, and 21, respectively. At this site, Shelby tube pushing went according to plan (illustrated in Figure 4.2), with no problems. Nine of the Shelby tubes Table 4.7: Uncorrected SPT-N Values (Athens County Site) Depth Range (ft) Uncorrected SPT-N Value (blows/ft)

81 81 were retained by the ORITE, and the remaining three were taken by BBCM. The recovery and notes on the nine tubes are included in Appendix B in Table B.13. Corrections were made to the original SPT-N values similar to the other field sites. They are shown in Table 4.8. Detailed information on the correction factors is included in Appendix B in Tables B Depth (ft) Original SPT-N Table 4.8: Athens County Site (N 60 ) 1 Values Energy Correction Only Peck Terzaghi Bazaraa Seed et al. Skempton Avg Subsurface Exploration Data for I-71 Site in Morrow County The fifth highway embankment site was located in the median of Interstate Highway 71 in Morrow County, about 60 miles north of Columbus. The field operation took place on an embankment about 30 feet high. The embankment supported two bridges for I-71 as it traveled over a small creek and local road at the bottom of a valley. The general view of the site is seen in a photograph inserted here as Figure 4.7. At this location, a continuous SPT was done to a depth of 25 ft. During the filed work, the split-spoon barrel brought samples of hard silt and clay to the ground surface. No water table was encountered during the field work. The uncorrected SPT-N values

82 obtained at this site are given in Table 4.9. Although the blow counts oscillated, they exhibited a general trend of increasing with depth. 82 Figure 4.7: Highway Embankment Site No. 5 on I-71 (Morrow County) Table 4.9: Uncorrected SPT-N Values (Morrow County Site) Depth Range (ft) Uncorrected SPT-N Value (blows/ft)

83 83 After analyzing the above data, the ORITE team decided to push Shelby tubes at depth ranges of 10.0 to 12.0, 13.0 to 15.0, and 17.5 to 19.5 ft. This gave the uncorrected SPT-N values of 9, 17, and 31, respectively. The original soil sampling plan, shown in Figure 4.2, had to be modified. The SPT truck was setup in the median of the freeway, in the center of the drainage path. There had also been substantial rain in the area the past few days. The soil was saturated at the surface, and it was very difficult for the truck to move around. Figure 4.8 shows the modified pattern. A total of twelve tubes were pushed with ORITE taking nine of them. Details on the tubes taken by ORITE are given in Appendix B in Table B.17. Corrections, as done with the previous field sites, were also done with this site. The corrected SPT-N values are shown below in Table Detailed information on the correction factors is in Appendix B in Tables B

84 84 Figure 4.8: Modified Plan for Shelby Tube Sampling Depth (ft) Original SPT-N Table 4.10: Morrow County Site (N 60 ) 1 Values Energy Correction Only Peck Terzaghi Bazaraa Seed et al. Skempton Avg.

85 85 This concludes data collected from the subsurface explorations at the first five sites. The next section will contain information obtained by BBC&M. The data includes soil index properties and sieve analyses. 4.3 Laboratory Index Properties and Sieve Analyses Index properties of soils encountered in the current project were determined using the Shelby tube samples obtained in the field. The index properties included a wide range of properties such as natural moisture content, unit weights (dry, moist), Atterberg limits (plastic limit, liquid limit, plasticity index), specific gravity, and grain size characteristics (percentages of gravel, sand, silt, and clay). These results will be presented for each site in the following subsections Soil Index Properties for Site No. 1 (Hamilton County) Four sets of index property testing were performed by BBCM on the soil samples recovered from the first (Hamilton County) site. Two sets were done on Shelby tube soil samples taken in the depth range of 2.5 to 4.5 ft, one set was done on a Shelby tube sample from the depth range of 4.5 to 6.5 feet, and one more set was done on a Shelby tube sample from the depth range of 10.0 to 12.0 ft. The results of the index tests are summarized below in Tables 4.11 and 4.12.

86 86 Table 4.11: Index Properties of Soils (Hamilton County Site) Depth of Soil (ft) Natural w (%) Moist Unit Wt (pcf) Dry Unit Wt (pcf) Specific Gravity Liquid Limit Plastic Limit Plasticity Index N/A N/A Table 4.12: Sieve Analysis Results (Hamilton County Site) Depth of Soil (ft) % Gravel % Sand % Silt % Clay AASHTO Soil Class A A A A Soil Index Properties for Site No. 2 (Fayette County) Four sets of index property testing were performed by BBCM on the soil samples recovered from the Fayette County site. One set was done on a Shelby tube sample taken from the depth range of 5.5 to 7.5 ft, two sets on two separate Shelby tubes in the depth range of 8.5 to 10.5 ft, and one set was done on a Shelby tube sample taken in the depth range of 14.5 to 16.5 ft. As it was mentioned earlier, a total of five Shelby tubes sampling holes were created at this site. This allowed for an extra tube being available at each soil sampling depth. Hence, two tubes were tested at the mid-depth range. The results of the index tests are summarized in Tables 4.13 and 4.14.

87 87 Table 4.13: Index Properties of Soils (Fayette County Site) Depth of Soil (ft) Natural w (%) Moist Unit Wt (pcf) Dry Unit Wt (pcf) Specific Gravity Liquid Limit Plastic Limit Plasticity Index N/A N/A Table 4.14: Sieve Analysis Results (Fayette County Site) Depth of Soil (ft) % Gravel % Sand % Silt % Clay AASHTO Soil Class A-6a A-4a A-4a A-4a Soil Index Properties for Site No. 3 (Lake County) Five sets of index testing were done by BBCM on the soil samples recovered from the Lake County site. One set was done on a Shelby tube sample obtained in the depth range of 1.0 to 3.0 ft, two on a Shelby tube sample taken in the depth range of 4.0 to 6.0 ft, and two on a Shelby tube sample from the depth range of 14.0 to 16.0 ft. The results of the index tests are summarized in Tables 4.15 and Table 4.15: Index Properties of Soils (Lake County Site) Depth of Soil (ft) Natural w (%) Moist Unit Wt (pcf) Dry Unit Wt (pcf) Specific Gravity Liquid Limit Plastic Limit Plasticity Index N/A N/A N/A

88 88 Table 4.16: Sieve Analysis Results (Lake County Site) Depth of Soil (ft) % Gravel % Sand % Silt % Clay AASHTO Soil Class A-6a A-4a A-4a A-4a A-4a Soil Index Properties for Site No. 4 (Athens County) Five sets of index tests and sieve analyses were done by BBCM on the Athens County site. One set was done on a Shelby tube in the depth range of 4.5 to 6.5 ft, one was done on a Shelby tube in the depth range of 8.0 to 10.0 ft, and three were done on a Shelby tube in the depth range of 19.0 to 21.0 ft. The soil varied greatly throughout the tube at the lowest depth. This is why three tests were done on it. The results of the index tests are summarized in Tables 4.17 and Table 4.17: Index Properties of Soils (Athens County Site) Depth of Soil (ft) Natural w (%) Moist Unit Wt (pcf) Dry Unit Wt (pcf) Specific Gravity Liquid Limit Plastic Limit Plasticity Index N/A N/A N/A Table 4.18: Sieve Analysis Results (Athens County Site) Depth of Soil (ft) % Gravel % Sand % Silt % Clay AASHTO Soil Class A-6a A-6a A-6b A-6b A-7-6

89 Soil Index Properties for Site No. 5 (Morrow County) Four sets of index tests and sieve analyses were done by BBCM on the Morrow County site. Two sets were done on a Shelby tube in the depth range of 10.0 to 12.0 ft, one was done on a Shelby tube in the depth range of 13.0 to 15.0 ft, and one was done on a Shelby tube in the depth range of 17.5 to 19.5 ft. The results of the index tests are shown below in Tables 4.19 and Table 4.19: Index Properties of Soils (Morrow County Site) Depth of Soil (ft) Natural w (%) Moist Unit Wt (pcf) Dry Unit Wt (pcf) Specific Gravity Liquid Limit Plastic Limit Plasticity Index N/A N/A Table 4.20: Sieve Analysis Results (Morrow County Site) Depth of Soil (ft) % Gravel % Sand % Silt % Clay AASHTO Soil Class A-4a A-6a A-6a A-6a 4.4 Shear Strength Properties In this section, the shear strength properties for the soils obtained at each site will be given. This includes data from the unconfined compression and C-U triaxial compression tests.

90 Shear Strength Properties for Site No. 1 (Hamilton County) Four unconfined compression tests were performed by BBCM on the soil samples taken from this site. Two were done on Shelby tubes from the highest depth range, one from the middle depth range, and one on the lowest depth range. Table 4.21 summarizes the test results. Table 4.21: Unconfined Compression Test Results (Hamilton County) Avg. Depth of Specimen (ft) Moisture Content (%) Dry Unit Weight (pcf) Unconfined Comp. Strength (psi) Strain at Failure (%) Also, a total of eight C-U triaxial compression tests were done on soil at this site. Three were done at the highest depth range, three were done at the middle depth range, and two were done at the lowest depth range. Specimen depth, t 50, φ angles, and effective consolidation stress for each specimen are given in Table Six of the specimens Table 4.22: C-U Triaxial Compression Test Results (Hamilton County) Specimen (Depth) t 50 (min) φ (degrees) φ' (degrees) Effective Consolidation Pressure (psi) A-1 (2.5' - 3.0') A-1 (3.1' - 3.6') D-1 (2.5' - 3.0') A-2 (5.1' - 5.6') C-2 (4.9' - 5.4') D-2 (4.6' - 5.1') A-3 (10.3' ') D-3 (10.2' ')

91 91 tested went to 15% axial strain without failure. Two of them were tested to less strain (Specimen A-1 ( depth) to 13.39% and Specimen A-1 ( depth) to 10.2%). Large rocks (larger than 1/6 of the diameter of the specimen) were also found in some of the specimens that could have affected the results. Soil recovery was poor at the lowest depth range for this site. That is why only two tests were done there. In addition, a variety of plots are in Appendix C related to the data just given. Figures D.1 through D.8 give stress-strain curves for each specimen, and figures D.9 through D.14 give p -q and p-q plots for each depth range Shear Strength Properties for Site No. 2 (Fayette County) Four unconfined compression tests were performed on soil from this site by BBCM. One was done on a Shelby tube from the highest depth range, two were done from the middle depth range, and one on the lowest depth range. Table 4.23 summarizes the test data. Ave. Depth of Specimen (ft) Table 4.23: Unconfined Compression Test Results (Fayette County) Moisture Content (%) Dry Unit Weight (pcf) Unconfined Comp. Strength (psi) Strain at Failure (%) Also, a total of nine C-U triaxial compression tests were done on soil at this site. Four were done at the highest depth range, three were done at the middle depth range,

92 92 and two were done at the lowest depth range. Specimen depth, t 50, φ angles, and effective consolidation stress for each specimen are given Table Every C-U triaxial test specimen went all the way to 15% axial strain without showing any failure characteristics. Rocks were also found in some of the specimens after testing. Table 4.24: C-U Triaxial Compression Test Results (Fayette County) Specimen (Depth) t 50 (min) φ (degrees) φ' (degrees) Effective Consolidation Pressure (psi) A-1 (5.7' - 6.2') D-1 (6.6' - 7.1') E-1 (6.3' - 6.7') E-1 (5.5' - 6.0') A-2 (9.2' - 9.7') D-2 (9.2' - 9.7') E-2 (9.2' - 9.7') B-3 (14.7' ') B-3 (15.4' ') Soil recovery was poor at the lowest depth range for this site also. That is why only two tests were done there. In addition, a variety of plots are in Appendix C related to the data just given. Figures C.15 through C.23 give stress-strain curves for each specimen, and Figures C.24 through C.29 give p -q and p-q plots for each depth range Shear Strength Properties for Site No. 3 (Lake County) Five unconfined compression tests were performed on the relatively undisturbed soil samples recovered from this site by BBCM. One was done on a Shelby tube from the highest depth range, two were done from the middle depth range, and two were done on the lowest depth range. Table 4.25 summarizes the test results.

93 93 Ave. Depth of Specimen (ft) Table 4.25: Unconfined Compression Test Results (Lake County) Moisture Content (%) Dry Unit Weight (pcf) Unconfined Comp. Strength (psi) Strain at Failure (%) A total of nine C-U triaxial compression tests were done on soil at this site. Four were done at the highest depth range, three were done at the middle depth range, and two were done at the lowest depth range. Specimen depth, t 50, phi angles, and effective consolidation stress for each specimen are given in Table Every specimen at this site was loaded to a 15% axial strain without exhibiting any failure conditions. Very few rocks were found in the specimens after testing also. Table 4.26: C-U Triaxial Compression Test Results (Lake County) Specimen (Depth) t 50 (min) φ (degrees) φ' (degrees) Effective Consolidation Pressure (psi) A-1 (1.6' - 2.1') A-1 (1.0' - 1.5') D-1 (1.1' - 1.6') A-2 (4.1' - 4.6') D-2 (4.0' - 4.5') D-2 (4.7' - 5.2') C-3 ( ) A-3 (14.6' ') D-3 (14.6' ')

94 94 In addition, a variety of plots are in Appendix C related to the data just given. Figures C.30 through C.38 give stress-strain curves for each specimen, and Figures C.39 through C.44 give p -q and p-q plots for each depth range Shear Strength Properties for Site No. 4 (Athens County) Five unconfined compression tests were performed on soil from this site by BBCM. One was done on a Shelby tube from the highest depth range, one was done from the middle depth range, and three were done at the lowest depth range. Table 4.27 summarizes the test results. Ave. Depth of Specimen (ft) Table 4.27: Unconfined Compression Test Results (Athens County) Moisture Content (%) Dry Unit Weight (pcf) Unconfined Comp. Strength (psi) Strain at Failure (%) A total of nine C-U triaxial compression tests were done on soil at this site. Three were done at each depth range. Specimen depth, t 50, φ angles, and effective consolidation stress for each specimen are given in Table Eight of the nine specimens were tested to 15% axial strain without showing any signs of failure. Specimen B-3 ( depth) failed at 12.72% strain. A few small rocks and shale fragments were found after testing, but they do not seem to be large enough to affect the results. Also, it should be

95 95 mentioned that two tests were done with soil from different tubes. The first specimen listed in Table 4.28 is given as A-1 ( ) & B-1 ( ). Here, because there was not enough soil in each of the tubes to make a specimen of proper height, two smaller sections were placed on top of each other. The same procedure was done with the specimen listed as B-2 ( ) & D-2 ( ). In addition, a variety of plots related to the data just given are in Appendix C. Figures C.45 through C.53 give stress-strain curves for each specimen, and Figures C.54 through C.59 give p -q and p-q plots for each depth range. Table 4.28: C-U Triaxial Compression Test Results (Athens County) Specimen (Depth) t 50 (min) φ (degrees) φ' (degrees) Effective Consolidation Pressure (psi) A-1 ( ) & B-1 ( ) B-1 (5.5' - 6.0') D-1 ( ) B-2 (8.8' - 9.3') D-2 (9.0' - 9.5') B-2 ( ) & D-2 ( ) A-3 (20.0' ') B-3 (20.0' ') D-3 (20.0' ') Shear Strength Properties from Site No. 5 (Morrow County) Four unconfined compression tests were performed on soil from this site by BBCM. Two were done on a Shelby tube from the highest depth range, one was done from the middle depth range, and one was done at the lowest depth range. Table 4.29 summarizes the test results.

96 96 Ave. Depth of Specimen (ft) Table 4.29: Unconfined Compression Test Results (Morrow County) Moisture Content (%) Dry Unit Weight (pcf) Unconfined Comp. Strength (psi) Strain at Failure (%) Also, a total of nine C-U triaxial compression tests were done on soil at this site. Three were done at the top depth range, three were done at the middle depth range, and three were done at the lowest depth range. Specimen depth, t 50, and φ angles for each specimen are given in Table All of the specimens were tested to 15% axial strain without reaching any failure conditions. There were also a few small rocks found in some of the samples, but they likely did not affect the final results. Table 4.30: C-U Triaxial Compression Test Results (Morrow County) Specimen (Depth) t 50 (min) φ (degrees) φ' (degrees) Effective Consolidation Pressure (psi) B-1 (10.5' ') C-1 (10.5' ') D-1 (10.5' ') D-2 (13.3' -13.8') C-2 (13.8' ') C-2 (13.3' ') B-3 (17.9' ') D-3 (18.2' ') C-3 (17.6' ')

97 97 In addition, a variety of plots related to the data just given are in Appendix C. Figures C.60 through C.68 give stress-strain curves for each specimen, and Figures C.69 through C.74 give p -q and p-q plots for each depth range.

98 CHAPTER 5: STATISTICAL ANALYSIS AND GEOTECHNICAL GUIDELINES 98 This chapter analyzes the data collected during experimentation. Empirical correlations presented in Chapter 2 will be evaluated in light of the data collected in the current study. Then, meaningful correlations between the different soil properties are sought using various linear and nonlinear mathematical models and a multi-variable regression analysis method. Based on the outcome of these data analyses, preliminary guidelines are recommended for estimating shear strength properties of embankment soils encountered in Ohio. 5.1 Evaluations of Empirical Correlations SPT-N vs. Unconfined Compression Strength by Terzaghi The first empirical correlation to be evaluated is the one between the fully corrected SPT-N value and unconfined compression strength proposed by Terzaghi et al. (1996). This correlation was previously presented in Table 2.2. Below, in Table 5.1, the unconfined compressive strengths of A-4a soils measured in the current study are entered into the chart prepared by Terzaghi et al. (1996), along with the corresponding (N 60 ) 1 values.

99 99 Table 5.1: Evaluation of Terzaghi s Correlation for A-4a Soil SPT Unconfined Compressive Strength (psi) (N 60 ) 1 Terzaghi Values Within Range Values Outside Range < 2 < , 46.1 > 30 > , , 41.0 The values of 30.2, 46.1, 71.3, and 79.0 psi were obtained on the soil samples recovered at the Lake County site. These are all within the range given by Terzaghi et al. (1996), along with the value of 20.3 psi which came from the Morrow County site. The values of 45.1, 47.2, and 41.0 psi were all found at the Fayette County site. These are all outside of the range. The value of 45.1 psi is above the range, and the other two values are below the range. Next, the unconfined compression strengths of A-6a soil are compared to the Terzaghi et al. (1996) empirical SPT-(N 60 ) 1 vs. unconfined compression strength relation, as shown in Table 5.2. All the measured values listed in Table 5.2 are falling out of the given range. The values of 47.8, 19.1, and 20.8 psi came from the Morrow County site. The values of 25.8 and 38.0 psi came from the Athens County site, the value of 36.6 psi came from the Fayette County site, and the value of 57.3 psi came from the Lake County site. Among the seven data points, only one (the value of 47.8 psi) is above the range specified by Terzaghi et al. (1996). Others are below the reported ranges.

100 100 Table 5.2: Evaluation of Terzaghi s Correlation for A-6a Soil SPT Unconfined Compressive Strength (psi) (N 60 ) 1 Terzaghi Values Within Range Values Outside Range < 2 < , 19.1 > 30 > , 57.3, 38.0, 20.8 Finally, the unconfined compression strengths of A-7-6 soil samples are applied to the empirical correlation of Terzaghi et al. (1996), as seen in Table 5.3. Only two of the measured unconfined compression test values are staying within the range reported for A-7-6 soils. The values of 30.6, 24.8, 18.7, and 46.9 psi came from the Hamilton County site, while the value of 41.8 psi came from the Athens County site. Table 5.3: Evaluation of Terzaghi s Correlation for A-7-6 Soil SPT Unconfined Compressive Strength (psi) (N 60 ) 1 Terzaghi Values Within Range Values Outside Range < 2 < , , 18.7 > 30 > The results presented in Tables 5.1 through 5.3 indicate that the empirical correlation between the SPT-(N 60 ) 1 and unconfined compression strength published by Terzaghi et al. (1996) is, overall, not well suited to the highway embankment soils encountered in Ohio.

101 SPT-N vs. Unconfined Compression by Dept. of Navy The next correlation to be assessed is also concerned with the link between the SPT-(N 60 ) 1 value and the unconfined compression strength. It was presented by Dept. of Navy (1982), as summarized in Table 2.3. The correlation here involves the lower and upper bounds, depending on the value of liquid limit. The lower bound is given by the values in Table 2.3 listed as low plasticity. The upper bound is given by the values in Table 2.3 listed as high plasticity. The actual unconfined compression strengths measured during the current study can be plotted into the correlation chart. Figure 5.1 shows this for all three soil types (A-4a, A-6a, and A-7-6) Unconfined Compressive Strength (psi) Navy (low value) Navy (high value) ORITE SPT-N Value Figure 5.1: Evaluation of Dept. of Navy Data for All Three Soil Types

102 102 A total of twenty-one data points are shown in Figure 5.1. Fifteen of these points fall within the upper and lower bounds given by Dept. of Navy (1982). This means that over seventy-one percent of the measured SPT and unconfined compression data for all three major Ohio soil types (A-4a, A-6a, A-7-6) appear to be mostly compatible with the empirical correlations reported by Dept. of Navy (1982). To evaluate this correlation further, the data compiled for each major soil type are entered into the correlation chart. Figure 5.2 shows a plot of unconfined compressive strength against (N 60 ) 1 values for A-4a soil samples. In Figure 5.2, all eight values fall within the upper and lower bounds set by Dept. of Navy (1982). Thus, the SPT-N vs. unconfined compression strength data of A-4a soil found in Ohio appear to conform to the established empirical correlation. Next, Figure 5.3 shows a similar plot of unconfined compressive strength against (N 60 ) 1 values for A-6a soil samples examined in the current study. In Figure 5.3, only three of the eight data points are positioned between the upper and lower bounds specified by Dept. of Navy (1982). This implies that the empirical correlation established by Dept. of Navy (1982) may not be quite applicable to the A-6a soil found in Ohio.

103 Unconfined Compressive Strength (psi) Navy (low value) Navy (high value) ORITE SPT-N Value Figure 5.2: Evaluation of Dept. of Navy Data for A-4a Soil Unconfined Compressive Strength (psi) Navy (low value) Navy (high value) ORITE SPT-N Value Figure 5.3: Evaluation of Dept. of Navy Data for A-6a Soil

104 104 Finally, in Figure 5.4, the unconfined compression strength vs. SPT-(N 60 ) 1 data compiled for A-7-6 soil is compared with the empirical correlation of Dept. of Navy (1982). Four of the five data points in Figure 5.4 are staying within the upper and lower bounds given. Thus, the SPT-N vs. unconfined compression strength data of A-7-6 soil found in Ohio appear to conform to the empirical correlation established in Dept. of Navy (1982) Unconfined Compressive Strength (psi) Navy (low value) Navy (high value) ORITE SPT-N Value Figure 5.4: Evaluation of Dept. of Navy Data for A-7-6 Soil In summary, although the amount of data is still lacking, the empirical correlation between the SPT-(N 60 ) 1 and unconfined compression strength reported in Dept. of Navy

105 (1982) appears to be reasonably applicable to A-4a and A-7-6 soils but not to A-6a soils found in Ohio Effective Friction Angle vs. Plasticity Index by Terzaghi The third empirical correlation to be tested here is the one between the effective friction angle and the plasticity index. This was established previously by Terzaghi et al. (1996), as shown in Table 2.4 and Figure 2.9. All of the data produced in the current study are added to Figure 2.9 to see how well engineering properties of the Ohio embankment soils obey the established correlation. This is shown in Figure 5.5 for all three major soil types (A-4a, A-6a, and A-7-6) encountered in the study Effective Friction Angle (degrees) ORITE Values Terzaghi's Range Plasticity Index (%) Figure 5.5: Comparison of Terzaghi & ORITE Data (All Three Soil Types)

106 106 Looking at the results presented in Figure 5.5, it is noted that thirty-one (or 70.5%) of the data points produced in this study fall inside the correlation band reported by Terzaghi et al. (1996). Thirteen of them (or 29.5%) are falling outside the band. The correlation band is 6 deep, with the upper and lower curves located at + 3 of the central curve. Most of the data points located outside the band seem to be positioned within + 5 of the central curve. The results shown in Figure 5.5 can also be broken down further into each major soil type to examine which soil types conform to the Terzaghi et al. (1996) φ -PI correlation more closely than others. Figure 5.6 shows such a plot for the A-4a soil samples tested in the current study Effective Friction Angle (degrees) ORITE Values Terzaghi's Range Plasticity Index (%) Figure 5.6: Comparison of Terzaghi & ORITE Data (A-4a Soil)

107 107 In this plot, six (or 54.5%) of the data points are falling inside of the correlation band of Terzaghi et al. (1996). Five out of eleven (or 45.5%) stayed outside the band. A- 4a soils were found only at the Fayette County and Lake County sites. The data points located outside the band are all positioned within + 5 of the central curve. In Figure 5.7, the measured properties of the A-6a soil samples are plotted in terms of the effective friction angle against the plasticity index. Nineteen of the twentytwo data points (or 86.4% of the soils tested) are falling inside the band. Only three points (13.6% of the cases) are landing outside of the band. This means that over eightysix percent of the A-6a soil tested is similar to the Terzaghi et al. (1996) results. The A- 6a soil samples were recovered from the Fayette County, Lake County, Athens County, and Morrow County sites. Most of the outside data points are within + 5 of the central curve. Finally, in Figure 5.8, the measured properties of the A-7-6 soil samples are plotted over the Terzaghi et al. (1996) empirical correlation chart. Here, six of the eleven data points (or 54.5% of the soils tested) are landing inside the reported band, and most of the outside data points are within + 5 of the central curve. In summary, it can be stated that the established empirical φ -PI correlation appears to be reliable for most of the soils encountered in the current study.

108 Effective Friction Angle (degrees) ORITE Values Terzaghi's Range Plasticity Index (%) Figure 5.7: Comparison of Terzaghi & ORITE Data (A-6a Soil) Effective Friction Angle (degrees) Terzaghi's Range ORITE Values Plasticity Index (%) Figure 5.8: Comparison of Terzaghi & ORITE Data (A-7-6 Soil)

109 Soil Type vs. Effective Friction Angle by Dept. of Navy The last empirical correlation that can be evaluated here involves the soil type and effective friction angle, as reported by Dept. of Navy (1982). This relation is shown in Table 5.4, along with the average effective angle of friction determined for each major soil type in the current study. Table 5.4: Comparison of Dept. of Navy & ORITE Data Soil Type φ' (degrees) Dept. of Navy (1982) φ' (degrees) ORITE Values A A A According to this table, the actual measured average φ' value and the Dept. of Navy (1982) φ' value are fairly close to each other for A-4 soil. For A-6 soils, the measured average φ' value is substantially higher than the φ' value listed by Dept. of Navy (1982). For A-7-6 soil, the measured φ' value is just below the upper bound of the range reported by Dept. of Navy (1982). 5.2 Linear Regression Analysis In Section 3.5, it was stated that many mathematical models (such as linear, 2 nd degree polynomial, logarithmic, power, exponential, hyperbolic, and reciprocal) would be applied to the data set to identify the best model and strongest correlations that appear to exist for the shear strength characteristics of major highway embankment soils found in Ohio.

110 110 First, linear regression analysis was performed for the soils tested. As mentioned in Chapter 3, six paths of correlations were formulated. These paths were illustrated in Figure 3.5. They are described again in Table 5.5. Table 5.5: Correlation Paths for Data Analysis Path Dependent Variable vs Independent Variable 1 Corrected SPT-N Values vs Laboratory Soil Index Properties 2 Laboratory Triaxial Test Results vs Laboratory Soil Index Properties 3 Unconfined Compressive Strength vs Laboratory Triaxial Test Results 4 Corrected SPT-N Values vs Unconfined Compressive Strength 5 Unconfined Compressive Strength vs Laboratory Soil Index Properties 6 Corrected SPT-N Values vs Laboratory Triaxial Test Results The following equation was applied in all of the linear regression analyses: y = mx + b (5.1) Throughout this chapter, the correlations will be listed with the strongest one at the top of the table and getting weaker as they go down. Any correlation with a coefficient of determination (R 2 ) value equal to 0.80 or above will be viewed as a strong meaningful correlation A-4a Soil Table 5.6 summarizes the results of the linear regression analysis performed for A-4a soil samples along Correlation Path 1. None of the correlations listed in the table

111 yielded the R 2 value higher than The only correlation which may be considered mildly strong is the one between the SPT-N value and the plasticity index. 111 Table 5.6: Path 1 Linear Correlations for A-4a Soil Dependent Variable y Independent Variable x R 2 Equation SPT-(N 60 ) 1 Plasticity Index y = x SPT-(N 60 ) 1 Hand Penetrometer y = x SPT-(N 60 ) 1 Liquid Limit y = x SPT-(N 60 ) 1 Specific Gravity y = x SPT-(N 60 ) 1 % Gravel y = x SPT-(N 60 ) 1 Initial Dry Unit Weight (UC Test) y = x SPT-(N 60 ) 1 % Clay y = x SPT-(N 60 ) 1 % Silt y = x SPT-(N 60 ) 1 Plastic Limit y = x SPT-(N 60 ) 1 % Sand y = x SPT-(N 60 ) 1 Initial Moisture Content (UC Test) y = x Tables 5.7 through 5.11 rank numerous linear correlations established for A-4a soil along Correlation Path 2, Path 3, Path 4, Path 5, and Path 6, respectively. No strong correlations are seen from Table 5.6 through Table The two highest are between the unconfined compressive strength and both the plasticity index and the liquid limit, identified along Correlation Path 5. These both come in with a R 2 value of just under 0.80.

112 112 Table 5.7: Path 2 Linear Correlations for A-4a Soil Dependent Variable y Independent Variable x R 2 Equation Eff. Friction Angle φ' % Sand y = x Friction Angle φ % Silt y = x Friction Angle φ Plastic Limit y = x Friction Angle φ Liquid Limit y = x Friction Angle φ Initial Moisture Content (UC Test) y = x Friction Angle φ % Gravel y = x Friction Angle φ % Clay y = x Friction Angle φ Initial Dry Unit Weight (UC Test) y = x Eff. Friction Angle φ' Initial Moisture Content (UC Test) y = x Eff. Friction Angle φ' Plasticity Index y = x Eff. Friction Angle φ' Plastic Limit y = x Eff. Friction Angle φ' % Clay y = x Friction Angle φ Plasticity Index y = x Friction Angle φ Hand Penetrometer y = x Eff. Friction Angle φ' % Silt y = x Friction Angle φ % Gravel y = x Friction Angle φ Specific Gravity y = x Friction Angle φ % Sand y = x Eff. Friction Angle φ' Liquid Limit y = x Eff. Friction Angle φ' Initial Dry Unit Weight (UC Test) y = x Eff. Friction Angle φ' Specific Gravity y = x Eff. Friction Angle φ' Hand Penetrometer y = x Table 5.8: Path 3 Linear Correlations for A-4a Soil Dependent Variable y Independent Variable x R 2 Equation Unconfined Compressive Strength Final Moisture Content (UC Test) y = x Unconfined Compressive Strength Initial Dry Unit Weight (UC Test) y = x Unconfined Compressive Strength Friction Angle φ y = x Unconfined Compressive Strength Eff. Friction Angle φ' y = x Unconfined Compressive Strength 50% Consol. Time t y = x

113 113 Table 5.9: Path 4 Linear Correlation for A-4a Soil Dependent Variable Independent Variable R 2 Equation SPT-(N 60 ) 1 Unconfined Compressive Strength y = x Table 5.10: Path 5 Linear Correlations for A-4a Soil Dependent Variable y Independent Variable x R 2 Equation Unconfined Compressive Strength Plasticity Index y = x Unconfined Compressive Strength Liquid Limit y = x Unconfined Compressive Strength % Silt y = x Unconfined Compressive Strength % Gravel y = x Unconfined Compressive Strength % Clay y = x Unconfined Compressive Strength Initial Dry Unit Weight (UC Test) y = x Unconfined Compressive Strength Plastic Limit y = 3.702x Unconfined Compressive Strength Initial Moisture Content (UC Test) y = x Unconfined Compressive Strength Hand Penetrometer y = x Unconfined Compressive Strength Specific Gravity y = x Unconfined Compressive Strength % Sand y = x Table 5.11: Path 6 Linear Correlations for A-4a Soil Dependent Variable y Independent Variable x R 2 Equation SPT-(N 60 ) 1 Eff. Friction Angle φ' y = x SPT-(N 60 ) 1 50% Consol. Time t y = x SPT-(N 60 ) 1 Friction Angle φ y = x A-6a Soil Linear regression was also performed for the A-6a soil data along each correlation path. Tables 5.12 through 5.17 present the outcome. As was the case with A-4a soils, no strong correlations are seen along any correlation path. The highest one is seen in Table 5.16 between the unconfined compressive strength and initial dry unit weight (measured

114 during the triaxial compression test), along Correlation Path 5. This linear correlation has the coefficient of determination R 2 of Table 5.12: Path 1 Linear Correlations for A-6a Soil Dependent Variable y Independent Variable x R 2 Equation SPT-(N 60 ) 1 Plastic Limit y = x SPT-(N 60 ) 1 % Clay y = x SPT-(N 60 ) 1 % Silt y = x SPT-(N 60 ) 1 % Gravel y = x SPT-(N 60 ) 1 Plasticity Index y = x SPT-(N 60 ) 1 Hand Penetrometer y = x SPT-(N 60 ) 1 Initial Moisture Content (UC Test) y = x SPT-(N 60 ) 1 Liquid Limit y = x SPT-(N 60 ) 1 Initial Dry Unit Weight (UC Test) y = x SPT-(N 60 ) 1 % Sand y = x SPT-(N 60 ) 1 Specific Gravity y = x Table 5.13: Path 2 Linear Correlations for A-6a Soil Dependent Variable y Independent Variable x R 2 Equation Friction Angle φ Specific Gravity y = x Friction Angle φ Plasticity Index y = x Friction Angle φ % Gravel y = x Eff. Friction Angle φ' Specific Gravity y = x Friction Angle φ Liquid Limit y = x Friction Angle φ % Clay y = x Eff. Friction Angle φ' % Sand y = x Eff. Friction Angle φ' Plasticity Index y = x Friction Angle φ Plastic Limit y = x Friction Angle φ Initial Moisture Content (UC Test) y = x Eff. Friction Angle φ' Liquid Limit y = x Eff. Friction Angle φ' Hand Penetrometer y = x Eff. Friction Angle φ' % Clay y = x Friction Angle φ % Sand y = x Eff. Friction Angle φ' Initial Dry Unit Weight (UC Test) y = x Eff. Friction Angle φ' Plastic Limit y = x Friction Angle φ Hand Penetrometer y = x

115 115 Table 5.13: Path 2 Linear Correlations for A-6a Soil (cont.) Dependent Variable y Independent Variable x R 2 Equation Friction Angle φ Initial Dry Unit Weight (UC Test) y = x Eff. Friction Angle φ' Initial Moisture Content (UC Test) y = 9.551x Friction Angle φ % Silt y = x Eff. Friction Angle φ' % Silt y = x Friction Angle φ % Gravel y = x Table 5.14: Path 3 Linear Correlations for A-6a Soil Dependent Variable y Independent Variable x R 2 Equation Unconfined Compressive Strength Final Moisture Content (UC Test) y = x Unconfined Compressive Strength 50% Consol. Time t y = x Unconfined Compressive Strength Initial Dry Unit Weight (UC Test) y = x Unconfined Compressive Strength Eff. Friction Angle φ' y = x Unconfined Compressive Strength Friction Angle φ y = x Table 5.15: Path 4 Linear Correlation for A-6a Soil Dependent Variable y Independent Variable x R 2 Equation SPT-(N 60 ) 1 Unconfined Compressive Strength y = x Table 5.16: Path 5 Linear Correlations for A-6a Soil Dependent Variable y Independent Variable x R 2 Equation Unconfined Compressive Strength Initial Dry Unit Weight (UC Test) y = x Unconfined Compressive Strength % Silt y = x Unconfined Compressive Strength Specific Gravity y = x Unconfined Compressive Strength % Gravel y = x Unconfined Compressive Strength Initial Moisture Content (UC Test) y = x Unconfined Compressive Strength % Clay y = 1.451x Unconfined Compressive Strength Hand Penetrometer y = x Unconfined Compressive Strength % Sand y = 3.023x Unconfined Compressive Strength Liquid Limit y = x Unconfined Compressive Strength Plastic Limit y = x Unconfined Compressive Strength Plasticity Index y = x

116 116 Table 5.17: Path 6 Linear Correlations for A-6a Soil Dependent Variable y Independent Variable x R 2 Equation SPT-(N 60 ) 1 Eff. Friction Angle φ' y = x SPT-(N 60 ) 1 50% Consol. Time t y = x SPT-(N 60 ) 1 Friction Angle φ y = x A-7-6 Soil The project data tied to A-7-6 soil, found only at the Hamilton County and Athens County sites, was also analyzed with the linear regression model. Path 1 correlations are shown in Table A strong correlation, with a R 2 value of 0.902, is seen between the (N 60 ) 1 value and specific gravity. Table 5.18: Path 1 Linear Correlations for A-7-6 Soil Dependent Variable y Independent Variable x R 2 Equation SPT-(N 60 ) 1 Specific Gravity y = x SPT-(N 60 ) 1 Plastic Limit y = x SPT-(N 60 ) 1 % Silt y = 1.1x 9.4 SPT-(N 60 ) 1 % Gravel y = x SPT-(N 60 ) 1 Hand Penetrometer y = x SPT-(N 60 ) 1 % Clay y = x SPT-(N 60 ) 1 Initial Moisture Content (UC Test) y = x SPT-(N 60 ) 1 Plasticity Index y = x SPT-(N 60 ) 1 % Sand y = x SPT-(N 60 ) 1 Liquid Limit y = x SPT-(N 60 ) 1 Initial Dry Unit Weight (UC Test) y = x Tables 5.19 through 5.23 summarize the results of the linear regression analysis performed along the remaining five correlation paths for A-7-6 soil samples. No strong

117 correlations were found anywhere. The best case along the second correlation path was presented between the friction angle and % gravel (with R 2 of 0.707). 117 Table 5.19: Path 2 Linear Correlations for A-7-6 Soil Dependent Variable y Independent Variable x R 2 Equation Friction Angle φ % Gravel y = x Friction Angle φ % Sand y = x Friction Angle φ % Clay y = x Friction Angle φ Initial Moisture Content (UC Test) y = x Friction Angle φ Plastic Limit y = x Friction Angle φ Hand Penetrometer y = x Friction Angle φ Plasticity Index y = x Friction Angle φ Initial Dry Unit Weight (UC Test) y = x Friction Angle φ Specific Gravity y = x Friction Angle φ Liquid Limit y = x Eff. Friction Angle φ' Specific Gravity y = x Friction Angle φ % Gravel y = x Eff. Friction Angle φ' Plastic Limit y = x Friction Angle φ % Silt y = x Eff. Friction Angle φ' Hand Penetrometer y = x Eff. Friction Angle φ' % Silt y = x Eff. Friction Angle φ' Plasticity Index y = x Eff. Friction Angle φ' % Sand y = x Eff. Friction Angle φ' Initial Dry Unit Weight (UC Test) y = x Eff. Friction Angle φ' Liquid Limit y = x Eff. Friction Angle φ' Initial Moisture Content (UC Test) y = x Eff. Friction Angle φ' % Clay y = x

118 118 Table 5.20: Path 3 Linear Correlations for A-7-6 Soil Dependent Variable y Independent Variable x R 2 Equation Unconfined Compressive Strength Friction Angle φ y = x Unconfined Compressive Strength 50% Consol. Time t y = x Unconfined Compressive Strength Initial Dry Unit Weight (UC Test) y = x Unconfined Compressive Strength Eff. Friction Angle φ' y = x Unconfined Compressive Strength Final Moisture Content (UC Test) y = x Table 5.21: Path 4 Linear Correlation for A-7-6 Soil Dependent Variable y Independent Variable x R 2 Equation SPT-(N 60 ) 1 Unconfined Compressive Strength y = x Table 5.22: Path 5 Linear Correlations for A-7-6 Soil Dependent Variable y Independent Variable x R 2 Equation Unconfined Compressive Strength Plastic Limit y = x Unconfined Compressive Strength % Gravel y = x Unconfined Compressive Strength Specific Gravity y = x Unconfined Compressive Strength Plasticity Index y = x Unconfined Compressive Strength % Sand y = x Unconfined Compressive Strength Liquid Limit y = x Unconfined Compressive Strength % Silt y = x Unconfined Compressive Strength Initial Moisture Content y = x (UC Test) Unconfined Compressive Strength Hand Penetrometer y = x Unconfined Compressive Strength Initial Dry Unit Weight y = x (UC Test) Unconfined Compressive Strength % Clay y = x Table 5.23: Path 6 Linear Correlations for A-7-6 Soil Dependent Variable y Independent Variable x R 2 Equation SPT-(N 60 ) 1 Eff. Friction Angle φ' y = x SPT-(N 60 ) 1 Friction Angle φ y = 0.42x SPT-(N 60 ) 1 50% Consol. Time t y = x

119 All Three Soil Types Finally, the data compiled for all three soil types tested (44 C-U triaxial compression tests) were analyzed through the linear regression model concept. Path 1 correlations are shown in Table No significant correlations exist in Table Path 2 correlations are, then, shown in Table The correlations, overall, are higher than those determined along Path 1, but none are very strong. Six of them come in with a regression value over Path 3 and 4 correlations are shown in Tables 5.26 and No significant correlations are seen along Paths 3 and 4. Tables 5.28 and 5.29 show Path 5 and Path 6 correlations. No strong correlations are detected in Tables 5.28 and Table 5.24: Path 1 Linear Correlations for All Three Soil Types Dependent Variable y Independent Variable x R 2 Equation SPT-(N 60 ) 1 % Gravel y = x SPT-(N 60 ) 1 Plasticity Index y = x SPT-(N 60 ) 1 % Sand y = x SPT-(N 60 ) 1 Liquid Limit y = x SPT-(N 60 ) 1 Initial Moisture Content y = x (UC Test) SPT-(N 60 ) 1 Plastic Limit y = x SPT-(N 60 ) 1 Initial Dry Unit Weight y = x (UC Test) SPT-(N 60 ) 1 % Silt y = x SPT-(N 60 ) 1 Specific Gravity y = x SPT-(N 60 ) 1 Hand Penetrometer y = x SPT-(N 60 ) 1 % Clay y = x Table 5.25: Path 2 Linear Correlations for All Three Soil Types Dependent Variable y Independent Variable x R 2 Equation Friction Angle φ Liquid Limit y = x Friction Angle φ Plasticity Index y = x Friction Angle φ % Clay y = x

120 120 Table 5.25: Path 2 Linear Correlations for All Three Soil Types (cont.) Dependent Variable y Independent Variable x R 2 Equation Eff. Friction Angle φ' % Clay y = x Friction Angle φ % Sand y = 0.835x Eff. Friction Angle φ' % Sand y = x Eff. Friction Angle φ' Liquid Limit y = x Friction Angle φ Initial Moisture Content (UC Test) y = x Friction Angle φ Plastic Limit y = x Eff. Friction Angle φ' Plasticity Index y = x Eff. Friction Angle φ' Plastic Limit y = x Friction Angle φ Initial Dry Unit Weight y = x (UC Test) Eff. Friction Angle φ' % Silt y = x Eff. Friction Angle φ' Initial Moisture Content (UC Test) y = x Friction Angle φ % Silt y = x Table 5.26: Path 3 Linear Correlations for All Three Soil Types Dependent Variable y Independent Variable x R 2 Equation Unconfined Compressive Strength Final Moisture Content (UC Test) y = x Unconfined Compressive Strength Initial Dry Unit Weight (UC Test) y = x Unconfined Compressive Strength Friction Angle φ y = x Unconfined Compressive Strength Eff. Friction Angle φ' y = x Unconfined Compressive Strength 50% Consol. Time t y = x Table 5.27: Path 4 Linear Correlation for All Three Soil Types Dependent Variable y Independent Variable x R 2 Equation SPT-(N 60 ) 1 Unconfined Compressive Strength y = x

121 121 Table 5.28: Path 5 Linear Correlations for All Three Soil Types Dependent Variable y Independent Variable x R 2 Equation Unconfined Compressive Strength Initial Dry Unit Weight y = x (UC Test) Unconfined Compressive Strength % Sand y = 1.189x Unconfined Compressive Strength Plasticity Index y = x Unconfined Compressive Strength Liquid Limit y = x Unconfined Compressive Strength Hand Penetrometer y = x Unconfined Compressive Strength Initial Moisture Content y = x (UC Test) Unconfined Compressive Strength % Silt y = x Unconfined Compressive Strength Specific Gravity y = x Unconfined Compressive Strength Plastic Limit y = x Unconfined Compressive Strength % Clay y = x Unconfined Compressive Strength % Gravel y = x Table 5.29: Path 6 Linear Correlations for All Three Soil Types Dependent Variable y Independent Variable x R 2 Equation SPT-(N 60 ) 1 Eff. Friction Angle φ' y = x SPT-(N 60 ) 1 Friction Angle φ y = 0.515x SPT-(N 60 ) 1 50% Consol. Time t y = x Nonlinear Regression With the outcome of the linear regression analysis rather disappointing, nonlinear regression analyses were performed extensively on the geotechnical data compiled in the current study to undercover correlations useful to geotechnical engineers in Ohio. These analyses applied six different nonlinear models. The models were the exponential, logarithmic, power, hyperbolic, reciprocal, and second-degree polynomial. These are defined in the equations below:

122 y = b e mx Exponential (5.2) 122 y = b + m(ln(x)) Logarithmic (5.3) y = b x m Power (5.4) b + mx y = Hyperbolic (5.5) x 1 y = b + m Reciprocal (5.6) x y = a 0 + a 1 x + a 2 x 2 2 nd Degree Polynomial (5.7) Relatively strong correlations identified in Section 5.2 are analyzed with each nonlinear model for the different soil types. The results are presented with the strongest (highest R 2 ) correlation at the top A-4a Soil For this soil type, eight mildly strong correlations were identified during the linear regression analysis. Each of these correlations are put through the nonlinear mathematical models to find out whether any of the correlations can become stronger. The results are presented in Table 5.30 through Table 5.35.

123 123 Table 5.30: Exponential Model Correlations for A-4a Soil Dependent Variable y Independent Variable x R 2 Equation Unconfined Compressive Strength Plasticity Index y = e x Friction Angle φ % Silt y = e x Eff. Friction Angle φ % Sand y = e 0.041x SPT-(N 60 ) 1 % Gravel y = e x Unconfined Compressive Strength Liquid Limit y = e x SPT-(N 60 ) 1 Unconfined Compressive Strength y = 5.714e x Unconfined Compressive Strength % Gravel y = e x Unconfined Compressive Strength Initial Dry Unit Weight (UC Test) y = e x Table 5.31: Logarithmic Model Correlations for A-4a Soil Dependent Variable y Independent Variable x R 2 Equation Unconfined Compressive Strength q % Gravel y = Ln(x) Unconfined Compressive Strength q Plasticity Index y = Ln(x) Unconfined Compressive Strength q Liquid Limit y = Ln(x) Eff. Friction Angle φ % Sand y = Ln(x) Friction Angle φ % Silt y = Ln(x) SPT-(N 60 ) 1 Unconfined Compressive Strength y = Ln(x) Unconfined Compressive Initial Dry Unit Weight Strength q (UC Test) y = Ln(x) SPT-(N 60 ) 1 % Gravel y = Ln(x) Table 5.32: Power Model Correlations for A-4a Soil Dependent Variable y Independent Variable x R 2 Equation Unconfined Compressive Strength % Gravel y = x Unconfined Compressive Strength Liquid Limit y = 0.212x Eff. Friction Angle φ % Sand y = x Unconfined Compressive Strength Plasticity Index y = x Friction Angle φ % Silt y = 3 E-05x Unconfined Compressive Strength Initial Dry Unit Weight (UC Test) y = 1 E+22x SPT-(N 60 ) 1 Unconfined Compressive Strength y = 0.158x SPT-(N 60 ) 1 % Gravel y = x

124 124 Table 5.33: Hyperbolic Model Correlations for A-4a Soil Dependent Variable y Independent Variable x R 2 Equation Eff. Friction Angle φ % Gravel y = (34.872x )/x Unconfined Compressive Strength q % Gravel y = (31.008x )/x Friction Angle φ % Gravel y = (31.663x )/x Eff. Friction Angle φ % Sand y = (64.921x 819.5)/x SPT-(N 60 ) 1 Unconfined Compressive Strength y = (91.951x )/x Unconfined Compressive Strength q Plasticity Index y = (159.67x )/x Eff. Friction Angle φ % Clay y = (30.638x )/x Unconfined Compressive Strength q Liquid Limit y = (159.72x )/x SPT-(N 60 ) 1 Plasticity Index y = (140.77x 877)/x Eff. Friction Angle φ Plasticity Index y = (42.01x )/x Unconfined Compressive Strength q % Clay y = (97.67x )/x Unconfined Compressive Final Moisture Content Strength q (C-U Triaxial Test) y = (104.55x )/x Friction Angle φ % Silt y = (125.9x )/x Eff. Friction Angle φ Plastic Limit y = (27.234x )/x Eff. Friction Angle φ Initial Moisture Content (UC Test) y = (24.94x )/x Unconfined Compressive y = ( x + % Silt Strength q 12619)/x Unconfined Compressive Strength q Plastic Limit y = (110.53x )/x Eff. Friction Angle φ Liquid Limit y = (30.96x )/x SPT-(N 60 ) 1 50% Consol. Time t y = (35.05x 11.65)/x SPT-(N 60 ) 1 Plastic Limit y = (120.14x )/x SPT-(N 60 ) 1 % Clay y = (68.29x )/x Friction Angle φ % Clay y = (16.21x )/x Unconfined Compressive Initial Moisture Content y = (89.75x 3.96)/x Strength q Unconfined Compressive Strength q (UC Test) Initial Dry Unit Weight (UC Test) y = ( x )/x

125 125 Table 5.34: Reciprocal Model Correlations for A-4a Soil Dependent Variable y Independent Variable x R 2 Equation Unconfined Compressive Strength q % Gravel y = (1/x) Unconfined Compressive Strength q % Silt y = 12841(1/x) Eff. Friction Angle φ % Sand y = (1/x) Unconfined Compressive Strength q Liquid Limit y = (1/x) SPT-(N 60 ) 1 Plasticity Index y = (1/x) Unconfined Compressive Strength q Plasticity Index y = (1/x) Friction Angle φ % Silt y = (1/x) SPT-(N 60 ) 1 Unconfined Compressive Strength y = (1/x) SPT-(N 60 ) 1 % Gravel y = (1/x) Unconfined Initial Dry Unit Weight Compressive Strength q (UC Test) y = 71057(1/x) Friction Angle φ Initial Moisture Content (UC Test) y = (1/x) Unconfined Compressive Strength q % Clay y = (1/x) Unconfined Compressive Strength q Plastic Limit y = (1/x) Unconfined Final Moisture Content Compressive Strength q (C-U Triaxial Test) y = (1/x) Table 5.35: Second-Degree Polynomial Model Correlations for A-4a Soil Dependent Variable y Independent Variable x R 2 Equation Unconfined Compressive Strength Plasticity Index y = x x Unconfined Compressive Strength Eff. Friction Angle φ y = x x Unconfined Compressive Strength % Gravel y = 0.677x x Unconfined Compressive Strength Liquid Limit y = x x Eff. Friction Angle φ % Sand y = x x Unconfined Compressive Strength % Silt y = 1.878x x Unconfined Initial Dry Unit Weight Compressive Strength (UC Test) y = x x SPT-N Value Plasticity Index y = x x

126 Table 5.35: Second-Degree Polynomial Model Correlations for A-4a Soil (cont.) Dependent Variable y Independent Variable x R 2 Equation Unconfined Unconfined Compressive Strength Compressive Strength y = x x Unconfined Compressive Strength % Clay y = x x SPT-(N 60 ) 1 Liquid Limit y = x x SPT-(N 60 ) 1 Eff. Friction Angle φ y = x x Friction Angle φ % Silt y = x x Friction Angle φ Initial Dry Unit Weight (UC Test) y = x x SPT-(N 60 ) 1 % Gravel y = x x SPT-(N 60 ) 1 % Silt y = x x According to Tables 5.30 through 5.32, the first three nonlinear models (exponential, logarithmic, power) did not improve the correlations previously seen through the linear regression analysis. In contrast, the application of the hyperbolic function introduced significant changes to the correlations. The hyperbolic model produced many fairly strong correlations. They are shown in Table Eleven of the twenty-four correlations listed in the table received the R 2 value of above Next, the outcome of the reciprocal model applications is shown in Table One strong correlation is emerging in Table This is between the unconfined compressive strength and % gravel with a R 2 value of Finally, correlations based on the second-degree polynomial function are listed in Table 5.35 for A-4a soils. Five strong correlations are found in the table with R 2 values of , , , , and These results show that both hyperbolic and the 2 nd degree polynomial functions are more effective in expressing the correlations existing among the geotechnical data

127 obtained for A-4a soil in Ohio. Also, Appendix D contains a plot of each type of nonlinear correlation (Figures D.1 D.6). The highest R 2 value plots are represented A-6a Soil A-6a soils were found at every site except in Hamilton County. Only two mildly significant correlations were found previously for A-6a soil, using the linear regression model. These two cases are reanalyzed here, using the first three nonlinear functions (exponential, logarithmic, power). The results are shown in Tables 5.36 through Table 5.36: Exponential Model Correlations for A-6a Soil Dependent Variable y Independent Variable x R 2 Equation Unconfined Compressive Strength q % Silt y = e x Unconfined Compressive Initial Dry Unit Weight Strength q (UC Test) y = e x Table 5.37: Logarithmic Model Correlations for A-6a Soil Dependent Variable y Independent Variable x R 2 Equation Unconfined Initial Dry Unit Weight y = Ln(x) Compressive Strength (UC Test) Unconfined % Silt y = Ln(x) Compressive Strength Table 5.38: Power Model Correlations for A-6a Soil Dependent Variable y Independent Variable x R 2 Equation Unconfined Initial Dry Unit Weight Compressive Strength (UC Test) y = 5 E-11x Unconfined % Silt Compressive Strength y = 1 E+08x

128 128 Only minor improvements can be seen between the linear and nonlinear regression model results so far for A-6a soils. Next, the hyperbolic model is applied to eighteen correlations concerning A-6a soil, as presented in Table Table 5.39: Hyperbolic Model Correlations for A-6a Soil Dependent Variable y Independent Variable x R 2 Equation Eff. Friction Angle φ % Gravel y = (33.49x )/x SPT-(N 60 ) 1 50% Consol. Time t y = (35.74x 10.59)/x Eff. Friction Angle φ Plasticity Index 0.84 y = (37.97x 54.21)/x Eff. Friction Angle φ Initial Moisture Content (UC Test) y = (35.21x 0.23)/x Unconfined Initial Dry Unit Weight Compressive Strength (UC Test) y = (234.11x 23067)/x Eff. Friction Angle φ % Clay y = (30.54x )/x Friction Angle φ % Gravel y = (17.88x )/x Eff. Friction Angle φ % Silt y = (34.78x 49.64)/x Eff. Friction Angle φ % Sand y = (42.53x )/x Friction Angle φ % Clay y = (31.31x )/x Unconfined Compressive Strength % Gravel y = (48.52x 68.94)/x Unconfined Compressive Strength Specific Gravity y = (772.86x )/x Unconfined Compressive Strength % Silt y = (-95.44x )/x Eff. Friction Angle φ Liquid Limit y = (43.82x )/x SPT-(N 60 ) 1 Plasticity Index y = (174.68x )/x Eff. Friction Angle φ Plastic Limit y = (30.34x )/x SPT-(N 60 ) 1 % Clay y = (98.28x )/x Eff. Friction Angle φ Initial Dry Unit Weight (UC Test) y = (29.62x )/x Once again, the hyperbolic model demonstrated its ability to fit to the geotechnical data better than the linear regression model for a number of cases. Five strong correlations are seen in the above table. Next, in Table 5.40, correlations based on the reciprocal model are presented for A-6a soil.

129 129 Table 5.40: Reciprocal Model Correlations for A-6a Soil Dependent Variable y Independent Variable x R 2 Equation Unconfined Initial Dry Unit Weight Compressive Strength (UC Test) y = (1/x) Unconfined Compressive Strength % Silt y = (1/x) Unconfined Compressive Strength Specific Gravity y = (1/x) SPT-(N 60 ) 1 Plastic Limit y = (1/x) The strongest correlation in the above table is seen between the unconfined compressive strength and the initial dry unit weight recorded during the C-U triaxial test, with a R 2 value of Finally, the second-degree polynomial model is applied to a few A-6a cases, as presented in Table The highest R 2 value here is just below 0.80 and is found again between the unconfined compressive strength and the initial dry unit weight (from the U-C test). Overall, it is noted that the reciprocal and 2 nd degree polynomial functions were not able to express the correlations existing among the A-6a soil data as well as the hyperbolic function. Table 5.41: Second-Degree Polynomial Model Correlations for A-6a Soil Dependent Variable y Independent Variable x R 2 Equation Unconfined Initial Dry Unit Weight Compressive Strength (UC Test) y = x x Unconfined Compressive Strength % Silt y = x x Unconfined Compressive Strength Specific Gravity y = x x

130 A-7-6 Soil The data linked to the A-7-6 soil, which was found at the Hamilton County and Athens County site, was analyzed using the nonlinear models. The four mildly to strongly significant correlations, identified during the initial linear regression analysis, are reanalyzed here using the exponential, logarithmic, and power models. The results are presented in Tables 5.42 through Table 5.42: Exponential Model Correlations for A-7-6 Soil Dependent Variable y Independent Variable x R 2 Equation SPT-(N 60 ) 1 Specific Gravity y = 5 E+09e x Friction Angle φ % Gravel y = e x Unconfined Compressive Strength Plastic Limit y = e x Friction Angle φ % Sand y = e x Table 5.43: Logarithmic Model Correlations for A-7-6 Soil Dependent Variable y Independent Variable x R 2 Equation SPT-(N 60 ) 1 Specific Gravity y = Ln(x) Friction Angle φ % Gravel y = Ln(x) Unconfined Compressive Strength Plastic Limit y = Ln(x) Friction Angle φ % Sand y = Ln(x) Table 5.44: Power Model Correlations for A-7-6 Soil Dependent Variable y Independent Variable x R 2 Equation SPT-(N 60 ) 1 Specific Gravity y = 5 E+09x Friction Angle φ % Gravel y = x Unconfined Compressive Strength Plastic Limit y = 6 E-07x Friction Angle φ % Sand y = x 0.518

131 131 Only mild improvements are brought about by the three nonlinear models. The correlation between the (N 60 ) 1 value and specific gravity remained high. Next, Table 5.45 presents the performance of the hyperbolic model for the A-7-6 soil correlations. Twenty-two correlations are re-examined through this model. Table 5.45: Hyperbolic Model Correlations for A-7-6 Soil Dependent Variable y Independent Variable x R 2 Equation Eff. Friction Angle φ' % Gravel y = (28.07x 3.824)/x Eff. Friction Angle φ' % Sand y = (26.29x )/x Friction Angle φ % Gravel y = (10.46x )/x Friction Angle φ % Sand y = (22.44x )/x Eff. Friction Angle φ' Plasticity Index y = (28.60x 35.41)/x SPT-(N 60 ) 1 Specific Gravity y = ( x )/x SPT-(N 60 ) 1 50% Consol. Time t y = (24.70x )/x SPT-(N 60 ) 1 Unconfined Compressive Strength y = (31.53x )/x Eff. Friction Angle φ' Initial Moisture Content (UC Test) y = (26.55x )/x Unconfined Compressive Strength % Sand y = (55.66x )/x SPT-(N 60 ) 1 % Gravel y = (18.24x )/x SPT-(N 60 ) 1 % Sand y = (23.32x )/x Unconfined Compressive Strength Friction Angle φ y = (70.18x )/x Eff. Friction Angle φ' Liquid Limit y = (28.82x 75.13)/x Unconfined y = (225.60x Plastic Limit Compressive Strength )/x SPT-(N 60 ) 1 Friction Angle φ y = (30.11x 64.77)/x Unconfined Compressive Strength % Gravel y = (23.41x )/x Friction Angle φ Initial Moisture Content (UC Test) y = (23.72x )/x SPT-(N 60 ) 1 % Silt y = (57.15x )/x SPT-(N 60 ) 1 Plastic Limit y = (83.74x )/x Eff. Friction Angle φ' % Silt y = (24.70x )/x Eff. Friction Angle φ' % Clay y = (27.19x )/x

132 132 Ten correlations in the above table possess the R 2 value of above 0.8. This is significant from what was seen previously using the linear regression model and the first three nonlinear models. Table 5.46 gives the outcome of the reciprocal model correlations for A-7-6 soil. Table 5.46: Reciprocal Model Correlations for A-7-6 Soil Dependent Variable y Independent Variable x R 2 Equation SPT-(N 60 ) 1 Specific Gravity y = (1/x) Friction Angle φ % Gravel y = (1/x) Unconfined Compressive Strength Plastic Limit y = (1/x) Friction Angle φ % Sand y = (1/x) Unconfined Compressive Strength Friction Angle φ y = (1/x) Unconfined Compressive Strength 50% Consol. Time t y = (1/x) One strong correlation, which is between the (N 60 ) 1 value and specific gravity is seen in the table. Finally, the second-degree polynomial model is applied to some A-7-6 soil correlations, as presented in Table Table 5.47: Second-Degree Polynomial Model Correlations for A-7-6 Soil Dependent Variable y Independent Variable x R 2 Equation SPT-(N 60 ) 1 Specific Gravity y = -1250x x SPT-(N 60 ) 1 % Gravel y = x x Friction Angle φ % Gravel y = x x Unconfined Plastic Limit y = x x Compressive Strength Friction Angle φ % Sand y = x x Friction Angle φ % Clay y = x x SPT-(N 60 ) 1 % Sand y = x x Unconfined Compressive Strength Friction Angle φ y = x x

133 133 The (N 60 ) 1 value against both specific gravity and % gravel produced good regression values, above 0.90, using the second-degree polynomial models. These were the only strong correlations noticed using the model All Three Soil Types Once again, all soil tested was combined for analysis. Exponential, logarithmic, and power model results are shown in Tables 5.48 through 5.50, respectively. Table 5.48: Exponential Model Correlations for All Three Soil Types Dependent Variable y Independent Variable x R 2 Equation Friction Angle φ Plasticity Index y = 33.09e x Friction Angle φ Liquid Limit y = e x Friction Angle φ % Sand y = e x Friction Angle φ % Clay y = e x Eff. Friction Angle φ' % Clay y = e x Eff. Friction Angle φ' Liquid Limit y = 42.21e x Unconfined Compressive Strength q Friction Angle φ Plastic Limit y = e x Initial Moisture Content (UC Test) y = e x Table 5.49: Logarithmic Model Correlations for All Three Soil Types Dependent Variable y Independent Variable x R 2 Equation Friction Angle φ Liquid Limit y = Ln(x) Friction Angle φ Plasticity Index y = Ln(x) Friction Angle φ % Clay y = Ln(x) Friction Angle φ % Sand y = Ln(x) Friction Angle φ Initial Moisture Content y = Ln(x) (UC Test) Eff. Friction Angle φ' % Clay y = Ln(x) Eff. Friction Angle φ' Liquid Limit y = Ln(x) Unconfined Compressive Strength Plastic Limit y = Ln(x)

134 134 Table 5.50: Power Model Correlations for All Three Soil Types Dependent Variable y Independent Variable x R 2 Equation Friction Angle φ Liquid Limit y = x Friction Angle φ Plasticity Index y = x Friction Angle φ % Sand y = x Friction Angle φ % Clay y = x Eff. Friction Angle φ' % Clay y = x Eff. Friction Angle φ' Liquid Limit y = x Friction Angle φ Unconfined Compressive Strength Initial Moisture Content (UC Test) Plastic Limit y = x y = 8 E-07x Small differences are noticed between the linear model correlation results and the results produced by the three nonlinear models. No strong correlations are seen in these tables. Next, the correlations are reformatted by the hyperbolic model using the combined data for all three soil types. Table 5.51 lists the outcome. Seven strong correlations are detected, with their R 2 values all above In fact, φ' against % gravel, % sand, and plasticity index all gave the R 2 values above Table 5.51: Hyperbolic Model Correlations for All Three Soil Types Dependent Variable y Independent Variable x R 2 Equation Eff. Friction Angle φ' % Gravel y = (33.36x )/x Eff. Friction Angle φ' Plasticity Index y = (24.31x )/x Eff. Friction Angle φ' % Sand y = (39.90x )/x SPT-(N 60 ) 1 50% Consol. Time t y = (24.79x )/x Eff. Friction Angle φ' % Silt y = (47.22x )/x Eff. Friction Angle φ' % Clay y = (21.08x )/x Eff. Friction Angle φ' Liquid Limit y = (20.46x )/x Friction Angle φ % Gravel y = (27.05x 40.27)/x Friction Angle φ % Sand y = (33.88x )/x Eff. Friction Angle φ' Initial Moisture Content (UC Test) y = (21.28x )/x

135 135 Table 5.51: Hyperbolic Model Correlations for All Three Soil Types (cont.) Dependent Variable y Independent Variable x R 2 Equation Unconfined Compressive Strength q % Gravel y = (40.29x 9.63)/x Eff. Friction Angle φ' Initial Dry Unit Weight (UC Test) y = (52.61x )/x SPT-(N 60 ) 1 Unconfined Compressive Strength y = (48.80x )/x Friction Angle φ Initial Dry Unit Weight (UC Test) y = (72.27x )/x Friction Angle φ % Silt y = (41.14x )/x Friction Angle φ Plasticity Index y = (6.53x )/x Unconfined Compressive Initial Dry Unit Weight Strength q (UC Test) y = (174.11x 15743)/x (N 60 ) 1 Plasticity Index y = (19.66x )/x Unconfined Compressive Strength q Friction Angle φ y = (50.79x )/x Unconfined Compressive Strength q % Sand y = (60.02x )/x SPT-(N 60 ) 1 % Gravel y = (20.08x )/x SPT-(N 60 ) 1 Friction Angle φ y = (39.87x )/x SPT-(N 60 ) 1 Eff. Friction Angle φ' y = (73.28x )/x Table 5.52 shows reciprocal correlations for all the soil tested in the project. No strong correlations are seen in Table Finally, the second-degree polynomial model is applied to the correlations for all the soils. Fourteen high R 2 values resulted. They are given in Table No substantial changes in the R 2 values are found using the seconddegree polynomial model as compared to the linear model. Table 5.52: Reciprocal Model Correlations for All Three Soil Types Dependent Variable y Independent Variable x R 2 Equation Friction Angle φ Liquid Limit y = (1/x) Friction Angle φ Plasticity Index y = (1/x) Friction Angle φ % Sand y = (1/x) Friction Angle φ Initial Moisture Content (UC Test) y = 2.394(1/x)

136 136 Table 5.52: Reciprocal Model Correlations for All Three Soil Types (cont.) Dependent Variable y Independent Variable x R 2 Equation Friction Angle φ % Clay y = (1/x) Eff. Friction Angle φ' % Sand y = (1/x) Friction Angle φ Plastic Limit y = (1/x) 4.92 Eff. Friction Angleφ' % Silt y = (1/x) Eff. Friction Angleφ' Liquid Limit y = (1/x) Friction Angle φ Initial Dry Unit Weight (UC Test) y = (1/x) Eff. Friction Angle φ' % Clay y = (1/x) Unconfined Initial Dry Unit Weight Compressive Strength (UC Test) y = (1/x) Table 5.53: Second-Degree Polynomial Model Correlations for All Three Soil Types Dependent Variable y Independent Variable x R 2 Equation Friction Angle φ Liquid Limit y = x x Friction Angle φ Plasticity Index y = 0.021x x Eff. Friction Angleφ' % Clay y = x x Friction Angle φ % Clay y = x x Eff. Friction Angle φ' Plastic Limit y = x x Friction Angle φ % Sand y = x x Eff. Friction Angle φ' % Sand y = x x Friction Angle φ Initial Moisture Content (UC Test) y = x x Eff. Friction Angleφ' Liquid Limit y = x x Friction Angle φ Plastic Limit y = x x Eff. Friction Angle φ' % Silt y = x x Eff. Friction Angleφ' Plasticity Index y = x x Unconfined Initial Dry Unit Compressive Strength Weight (UC Test) y = x x Friction Angle φ % Silt y = x x Multi-Variable Linear Regression Analysis Up to now, linear and nonlinear correlations were explored between a dependent variable and a single independent variable. There were some moderately strong to very strong correlations that emerged from these relatively simple linear and nonlinear

137 137 regression analyses. The next logical step is to look at correlations between a dependent variable and two or more independent variables. This section presents results of the multi-variable linear regression analysis performed for each major soil type and all three soil types combined A-4a Soil One of the main objectives of this project was to correlate the shear strength parameters (φ and φ angles) with SPT-(N 60 ) 1 values and basic soil index properties. In Table 5.54 various linear correlations are addressed for the friction angles involving two independent variables. A total of twelve correlations are examined in the analysis. In the first six, the φ angle is specified as the dependent variable. Sieve analysis values along with the corrected (N 60 ) 1 values are looked at in the first four while plasticity index and the initial dry unit weight are used in the next two with the (N 60 ) 1 value. The next six correlations specify the φ angle as the dependent variable with the same pattern of independent variables. The twelve correlations are ranked in terms of the R 2 value in Table The strongest correlation for the friction angle φ is attributed to % silt and SPT-(N 60 ) 1 value. The strongest correlation for the effective friction angle φ is attributed to % sand and SPT-(N 60 ) 1 value. Table 5.54: Multi-Variable Linear Regression Correlations for φ and φ (A-4a Soil) Dependent Variable Independent Variables R 2 Equation φ % Sand & (N 60 ) φ = (%Sand) (N 60 )

138 Table 5.54: Multi-Variable Linear Regression Correlations for φ and φ (A-4a Soil) (cont.) Dependent Variable Independent Variables R 2 Equation φ % Silt & (N 60 ) φ % Gravel & (N 60 ) φ % Clay & (N 60 ) φ % Clay & (N 60 ) φ Initial Dry Unit Weight (UC Test) & (N 60 ) φ Plasticity Index & (N 60 ) φ % Gravel & (N 60 ) φ Initial Dry Unit Weight (UC Test) & (N 60 ) φ Plasticity Index & (N 60 ) φ % Silt & (N 60 ) φ % Sand & (N 60 ) φ = (%Silt) (N 60 ) φ = (%Gravel) (N 60 ) φ = (%Clay) (N 60 ) φ = (%Clay) (N 60 ) φ = (IDUW-UC) (N 60 ) φ = (PI) (N 60 ) φ = (%Gravel) (N 60 ) φ = (IDUW-UC) (N 60 ) φ = (PI) (N 60 ) φ = (%Silt) (N 60 ) φ = (%Sand) (N 60 ) While analyzing shear strength parameters is the most vital to this project, other significant correlations will be examined here as well. To do this effectively, the outcome of the linear correlations presented in Section 5.2 can be revisited. The A-4a soil correlations that produced relatively high R 2 values are put together and are shown in Table In this table, the independent variables consisted only of sieve analysis and index property testing values. Another meaningful correlation, with the R 2 value of 0.855, is identified for the effective friction angle φ. It has % sand and % silt listed as the independent variables. For the unconfined compression strength, three promising

139 139 correlations have emerged. Among them the strongest one, with the R 2 value of 0.916, expresses the unconfined compression strength in terms of % gravel and the liquid limit LL. Table 5.55: Other Significant Multi-Variable Linear Correlations (A-4a Soil) Dependent Variable Independent Variables R 2 Equation Unconfined Compressive q = (LL) % Gravel & Liquid Limit Strength (q) (%Gravel) Unconfined Compressive % Gravel & Plasticity q = (PI) Strength (q) Index (%Gravel) φ % Sand & % Silt φ = (%Sand) (%Silt) Unconfined Compressive Liquid Limit & Initial Dry q = (LL) (IDUW Strength (q) Unit Weight (UC Test) UC) Unconfined Compressive % Gravel & Initial Dry q = (%Gravel) Strength (q) Unit Weight (UC Test) (IDUW-UC) φ % Sand & Plasticity Index φ = (%Sand) (PI) φ % Sand & % Silt φ = (%Sand) (%Silt) φ % Gravel & % Clay φ = (%Gravel) (%Clay) Finally, it was deemed necessary to look at equations involving more than two independent variables. Table 5.56 does this with the φ and φ angles as the dependent variables. In this table, eight new correlations are examined statistically. Six of the cases addressed here involve the SPT-(N 60 ) 1 value as an independent variable. Two of the cases do not involve the SPT-(N 60 ) 1 value and instead use a combination of other values that can be obtained in the lab. According to Table 5.56, seven strong correlations exist for estimating the shear strength parameters of A-4a Ohio soil. The highest correlation, with a R 2 value of 0.983, does not even rely on a SPT-(N 60 ) 1 value.

140 Table 5.56: Multi-Variable Linear Correlations with 3 and 4 Independents (A-4a Soil) Dependent Variable φ φ φ φ φ Independent Variables R 2 Equation % Sand, % Silt, Plasticity Index, & Initial Dry Unit Weight (UC Test) % Gravel, % Silt, Plasticity Index, & (N 60 ) 1 % Gravel, % Sand, Plasticity Index, & (N 60 ) 1 % Gravel, % Silt, Plasticity Index, & (N 60 ) 1 % Gravel, % Sand, Plasticity Index, & (N 60 ) φ % Sand, % Silt, & (N 60 ) φ % Sand, % Silt, Plasticity Index, & Initial Dry Unit Weight (UC Test) φ % Sand, % Silt, & (N 60 ) φ = (%Sand) (%Silt) (PI) (IDUW-UC) φ = (%Gravel) (%Silt) (PI) (N 60 ) φ = (%Gravel) (%Sand) (PI) (N 60 ) φ = (%Gravel) (%Silt) (PI) (N 60 ) φ = (%Gravel) (%Sand) (PI) (N 60 ) φ = (%Sand) (%Silt) (N 60 ) φ = (%Sand) (%Silt) (PI) (IDUW-UC) φ = (%Sand) (%Silt) (N 60 ) A-6a Soil Similar to the previous section, the study data linked to A-6a soil samples were analyzed by the multi-variable linear regression method, with the φ and φ angles as the dependent variables. The independent variables again, consisted of the (N 60 ) 1 value paired with sieve analysis, plasticity index, and initial dry unit weight values for the soil. The outcome is given in Table None of the cases in the table is strong enough to be meaningful. Earlier in Section 5.2, only two significant correlations were identified for A-6a soil. Therefore, only one other significant correlation is looked at here. This is

141 141 shown in Table The unconfined compression strength of A-6a soil is strongly correlated to % silt and initial dry unit weight. Table 5.57: Multi-Variable Linear Regression Correlations for φ and φ (A-6a Soil) Dependent Variable Independent Variables R 2 Equation φ Plasticity Index & (N 60 ) φ % Gravel & (N 60 ) φ % Clay & (N 60 ) φ Plasticity Index & (N 60 ) φ % Sand & (N 60 ) φ % Clay & (N 60 ) φ % Silt & (N 60 ) φ Initial Dry Unit Weight (UC Test) & (N 60 ) φ % Gravel & (N 60 ) φ % Sand & (N 60 ) φ Initial Dry Unit Weight (UC Test) & (N 60 ) φ % Silt & (N 60 ) φ = (PI) (N 60 ) φ = (%Gravel) (N 60 ) φ = (%Clay) (N 60 ) φ = (PI) (N 60 ) φ = (%Sand) (N 60 ) φ = (%Clay) (N 60 ) φ = (%Silt) (N 60 ) φ = (IDUW-UC) (N 60 ) φ = (%Gravel) (N 60 ) φ = (%Sand) (N 60 ) φ = (IDUW-UC) (N 60 ) φ = (%Silt) (N 60 ) Table 5.58: Other Multi-Variable Linear Correlation (A-6a Soil) Dependent Variable Independent Variables R 2 Equation Unconfined Compressive % Silt & Initial Dry Unit q = (%Silt) Strength (q) Weight (UC Test) (IDUW-UC)

142 142 An attempt was made to find additional multi-variable linear correlations for A-6a soils, using three or four independent variables and φ or φ angle as the dependent variable. However, the attempt was unsuccessful. None of the cases tried produced a R 2 value above This is not too surprising, however, since the regression values in Table 5.57 are very low A-7-6 Soil The same type of multi-variable linear regression analysis was carried out, using the project data tied to A-7-6 soil type. Table 5.59 summarizes the outcome for the cases which involved φ or φ as the dependent variable. The independent variables consist of sieve analysis, index property, and initial dry unit weight values for A-7-6 soil. Table 5.59: Multi-Variable Linear Regression Correlations for φ and φ (A-7-6 Soil) Dependent Var. Independent Variables R 2 Equation φ % Sand & (N 60 ) φ = (%Sand) (N 60 ) φ % Gravel & (N 60 ) φ = (%Gravel) (N 60 ) φ % Clay & (N 60 ) φ = (%Clay) (N 60 ) φ Initial Dry Unit Weight φ = (IDUW-UC) (UC Test) & (N 60 ) 1 (N 60 ) φ Plasticity Index & φ = (PI) (N ) 1 + (N 60 ) φ Sand & (N 60 ) φ = (%Sand) (N 60 ) φ Gravel & (N 60 ) φ = (%Gravel) (N 60 ) φ Clay & (N 60 ) φ = (%Clay) (N 60 ) φ Initial Dry Unit Weight φ = (IDUW-UC) (UC Test) & (N 60 ) 1 (N 60 )

143 Table 5.59: Multi-Variable Linear Regression Correlations for φ and φ (A-7-6 Soil) (cont.) Dependent Var. Independent Variables R 2 Equation φ Plasticity Index & φ = (PI) (N 60 ) 1 (N 60 ) φ Silt & (N 60 ) φ = (%Silt) (N 60 ) φ % Silt & (N 60 ) φ = (%Silt) (N 60 ) Using information from Section 5.1, ten other significant multi-variable linear correlations were examined. This is shown in Table Here, the dependent variables are the friction angle φ, unconfined compressive strength, and corrected SPT-(N 60 ) 1 value. Three statistically strong correlations were found for SPT-(N 60 ) 1, and one strong correlation for the unconfined compression strength. Table 5.60: Other Multi-Variable Linear Correlations (A-7-6 Soil) Dependent Variable Independent Variables R 2 Equation (N 60 ) 1 Plastic Limit & Specific (N ) 1 = (PL) Gravity (Gs) (N 60 ) 1 % Sand & Specific (N ) 1 = (%Sand) Gravity (SG) (N 60 ) 1 % Gravel & Specific (N ) 1 = (%Gravel) Gravity (Gs) Unconfined Compressive q = (%Sand) % Sand & Plastic Limit Strength (q) (PL) φ % Clay & Plastic Limit φ = (%Clay) (PL) φ % Gravel & % Sand φ = (%Gravel) (%Sand) φ Unconfined Compressive φ = (q) Strength (q) & % Gravel (%Gravel) φ % Sand & Plastic Limit φ = (%Sand) (PL) Unconfined Compressive q = (%Gravel) + % Gravel & Plastic Limit Strength (q) (PL) Unconfined Compressive % Sand & Specific q = (%Sand) Strength (q) Gravity (Gs)

144 144 Additional multi-variable linear correlations were investigated for A-7-6 soil in Ohio, with three or four independent variables. The outcome is given in Table The R 2 value of all four correlations is slightly below The friction angle can be predicted fairly well for A-7-6 soils using the equations shown in Table One of the correlations does not even use the (N 60 ) 1 value, which allows it to be of extreme convenience to a design engineer. The same independent variables (% gravel, % sand, % clay, PI, dry unit weight, SPT-N) were tried using φ as the dependent variable. However, these cases all resulted in the R 2 values of about Table 5.61: Multi-Variable Linear Correlations with 3 and 4 Independents (A-7-6 Soil) Dependent Variable φ φ φ φ Independent Variables R 2 Equation % Gravel, % Sand, % Clay, & (N 60 ) 1 % Gravel, % Sand, Plasticity Index, & (N 60 ) 1 % Gravel, % Sand, Plasticity Index, & Initial Dry Unit Weight (UC Test) % Sand, % Gravel, & (N 60 ) φ = (%Gravel) (%Sand) (N 60 ) φ = (%Gravel) (%Sand) (PI) (N 60 ) φ = (%Gravel) (%Sand) (PI) (IDUW-UC) φ = (%Gravel) (%Sand) (N 60 ) All Three Soil Types The last multi-variable linear regression analysis was performed using the entire project data (i.e., looking at all of the soils combined), since they were all from Ohio. The set of correlations using shear strength parameters as the dependent variables is shown in Table Unfortunately, no statistically strong multi-variable linear correlations emerged for either φ or φ.

145 145 Other significant correlations that include shear strength parameters and unconfined compressive strength in as the dependent variable were also investigated. This is shown in Table Once again, the attempts met no major success. Table 5.62: Multi-Variable Linear Regression Correlations for φ and φ (All Soil Types) Dependent Variable Independent Variables R 2 Equation φ % Clay & (N 60 ) φ Plasticity Index & (N 60 ) φ % Clay & (N 60 ) φ % Sand & (N 60 ) φ % Sand & (N 60 ) φ Plasticity Index & (N 60 ) φ % Silt & (N 60 ) φ Initial Dry Unit Weight (UC Test) & (N 60 ) φ % Silt & (N 60 ) φ Initial Dry Unit Weight (UC Test) & (N 60 ) φ % Gravel & (N 60 ) φ % Gravel & (N 60 ) φ = (%Clay) (N 60 ) φ = (PI) (N 60 ) φ = (%Clay) (N 60 ) φ = (%Sand) (N 60 ) φ = (%Sand) (N 60 ) φ = (PI) (N 60 ) φ = (%Silt) (N 60 ) φ = (IDUW-UC) (N 60 ) φ = (%Silt) (N 60 ) φ = (IDUW-UC) (N 60 ) φ = (%Gravel) (N 60 ) φ = (%Gravel) (N 60 ) Table 5.63: Other Multi-Variable Linear Correlations (All Soil Types) Dependent Var. Independent Variables R 2 Equation φ % Sand & Liquid Limit φ = (%Sand) (LL) φ % Clay & Liquid Limit φ = (%Clay) (LL)

146 146 Table 5.63: Other Multi-Variable Linear Correlations (All Soil Types) (cont.) Dependent Variable Independent Variables R 2 Equation φ % Clay & Plasticity φ = (%Clay) Index (PI) φ % Sand & Plasticity φ = (%Sand) Index (PI) φ % Sand & % Clay φ = (%Sand) (Clay) φ % Clay & Liquid Limit φ = (%Clay) (LL) Unconfined Compressive Strength (q) Unconfined Compressive Strength (q) Plasticity Index & Initial Dry Unit Weight (UC Test) % Sand & Initial Dry Unit Weight (UC Test) q = (PI) (IDUW-UC) q = (%Sand) (IDUW-UC) Once again, as it was done previously, further correlations were sought after using three or four independent variables. However, after going through a series of regressions, no combination of variables was able to produce a strong correlation. Most of the R 2 values ranged from 0.55 to Preliminary Geotechnical Guidelines The outcome of the data analysis presented in this chapter can be combined to formulate a set of guidelines that geotechnical engineers can apply to estimate more confidently shear strength properties of highway embankment soils commonly encountered in Ohio. The guidelines may be given at multiple levels to allow varying degrees of sophistication involved in the estimation process. The guidelines presented here are of preliminary nature, since the data from the final four embankment sites will not be available for a few months.

147 147 Level 1: Use the following default φ values for the three major embankment soil types found in Ohio: A-4a. φ = 30.2 (5.1) A-6a. φ = 30.5 (5.2) A φ = 24.5 (5.3) Level 2: Estimate the effective friction angle of any major embankment soil type (A-4a, A-6a, A-7-6) using the empirical φ -PI correlation chart established by Terzaghi et al. (1996). It is noted here that the actual φ value may be within + 3 to + 5 of the average value extracted from the center region of the band. Level 3: For any of the three major soil types, use any of the following mathematic- al equations to estimate the effective friction angle (in degrees). (% ) G 8.28 φ = (5.4) (% G) (% ) S φ = (5.5) (% S) ( ) PI φ = (5.6) ( PI)

148 148 Level 3 (Alternative): For each specific major soil type, use any of the following mathematical equations to estimate the effective friction angle (in degrees). A-4a Soil: (% ) G φ = (5.7) (% G) (% ) S φ = (5.8) (% S) ( ) PI φ = (5.9) ( PI) φ = 0.527(%S) (%S) (5.10) φ = 2.254(%S) (%M) (PI) 1.566(γ d ) (5.11) φ = 1.325(%S) {SPT-(N 60 ) 1 } (5.12) A-6a Soil: (% ) G φ = (5.13) (% G) ( ) PI φ = (5.14) ( PI) A-7-6 Soil: (% ) G 3.82 φ = (5.15) (% G) (% ) S φ = (5.16) (% S)

149 ( ) PI φ = (5.17) ( PI) 149 where %G = % gravel (by mass); %S = % sand (by mass); PI = plasticity index (%); %M = % silt (by mass); γ d = dry unit weight (lb/ft 3 ); SPT-(N 60 ) 1 = SPT-N value fully corrected to energy ratio and overburden stress level (blows/ft). The Level 1 and Level 2 correlations are simply the empirical correlations given by Dept. of Navy (1982) and Terzaghi et al. (1996) with some modifications or additional comments introduced. The Level 3 correlations, however, which are shown in the many tables throughout Chapter 5, were analyzed further. In Table 5.64, the average difference between the predicted φ value (calculated using the Level 3 correlations) and the actual φ' value (given in Chapter 4) is given in the second column. The third column gives the standard deviation of the differences between the predicted and actual values. Table 5.64: Average and Standard Deviation of Differences for φ Equation Average Difference for φ (deg.) Standard Dev. of Differences for φ (deg.)

150 150 Table 5.64: Average and Standard Deviation of Differences for φ (cont.) Equation Average Difference for φ (deg.) Standard Dev. Of Difference for φ (deg.) The results presented in this table clearly indicate the advantage of taking the Level 3 (alternative) over any of the other options. Some of the equations, especially Eqns. 5.8, 5.10, 5.14, 5.16, and 5.17, listed under the Level 3 (alternative) yielded very small average and standard deviation values for the difference between the predicted and measured φ values.

151 151 CHAPTER 6: SUMMARY AND CONCLUSIONS 6.1 Summary For many years, geotechnical engineers in Ohio have had to rely on empirical correlations found in textbooks or journal articles to design highway embankments. However, many of these correlations were found by testing soil from other areas of the U.S. or the world. This has caused engineers to use overly conservative estimates in design. The project at hand looked to create reliable correlations for geotechnical engineers created from testing on Ohio soils. To accomplish this, the following objectives were given: Conduct a literature review related to highway embankments in Ohio; Identify several embankment locations around Ohio for detailed field and laboratory testing; Perform the continuous SPT and push multiple Shelby tube samples at each site; Perform index property tests, sieve analyses, and shear strength tests on the obtained soil; and Analyze field and laboratory test results to develop correlations regarding the shear strength of the soil.

152 152 The first task was completed by looking through journal articles and texts relating to the shear strength of soil, pore water pressure in soil, highway embankment stability, and AASHTO classifications. The different soil regions in Ohio and the types of soil found there were examined. Field testing methods used at embankments and laboratory testing methods used on soil from the embankment were also given. The second task was completed by studying old geotechnical reports from many locations around the state. Figuring out which types of soil were found in certain locations, the project team decided where it would be best suited to perform testing operations. A total of five different locations in the state (Hamilton, Fayette, Lake, Athens, and Morrow Counties) were used for testing. Each location also had to meet the criteria set out in Chapter 3. The third and fourth tasks were completed using a combined effort from BBCM and the ORITE. Proper field and laboratory test methods were used to perform these tests. Finally, the fifth task was completed through a series of statistical analyses. Some of these analyses proved to be useful and some did not. They did provide, however, some guidelines to be used by Ohio engineers, which was a major objective of the project. 6.2 Conclusions After completing the objectives, a series of conclusions were drawn out. First, in Chapter 4, SPT-(N 60 ) 1 values were given. These values were all energy corrected and then normalized by using an average of the five normalization factors. Looking at the

153 153 final (N 60 ) 1 value given at each site, it should be noted that it is extremely close to the value given by Seed et al. (1975). The Skempton (1986) factor was also very close to the final averaged (N 60 ) 1 value, but not as close at Seed et al. (1975). A conclusion can be drawn that the Seed et al. (1975) factor is very effective in normalizing the effects of overburden pressure compared to the other ones. It is also interesting to look what types of soil, in regards to AASHTO classifications, were found at each site. A-7-6 soils were found only at the Hamilton and Athens County sites. These sites are the most southern located of the five. It is possible that Southern Ohio contains a good amount of A-7-6 soil regardless of the whether the area is glaciated (Hamilton County) or not (Athens County). Also, A-4 soils were plentiful at the Lake County site with smaller deposits found also at the Fayette and Morrow County sites. A preliminary conclusion is that A-7-6 soils are found predominantly in Southern Ohio, A-4 soils are found predominantly in Northern Ohio, and A-6 soils are scattered, with most though being found in the Central Region. Next, comparisons made between the BBCM and ORITE unconfined compression test results and SPT-(N 60 ) 1 values and those results given through empirical correlations by Terzaghi et al. (1996) and Dept. of Navy (1982) were mixed. Ohio A-4a soils appear to conform to both the Terzaghi et al. (1996) and the Dept. of Navy (1982) correlations plotting the unconfined compressive strength against SPT-(N 60 ) 1 values. A- 6a soils in Ohio do not appear to conform to either the Terzaghi et al. (1996) or the Dept. of Navy (1982) correlations between unconfined compressive strength and SPT-(N 60 ) 1 values. Finally, using the results of Table 5.3 and Figure 5.4, Ohio A-7-6 soils do not

154 154 conform to the Terzaghi et al. (1996) results but they appear to conform to the Dept. of Navy (1982) results. Another reason that some of the unconfined compression test results are not conforming well to the correlations given in Chapter 2 is that this test is not a very reliable one to begin with. The unconfined compression test set-up does not simulate field conditions. Even with experienced people doing the work and good quality control, the results may be very scattered, as was the case in the current study. Strict quality control was used, also, with the SPT. The same drill rig and field crew worked at each site. This was done to obtain consistency in the results. The same quality control was also used by the ORITE for triaxial compression testing, whose results will be discussed next. Next, effective friction angle and plasticity index data obtained from the ORITE and BBCM experimentation was compared to that from Terzaghi et al. (1996). Overall, the comparisons turned out to be moderately compatible. Six out of eleven of the A-4a soils fell into the Terzaghi et al. (1996) + 3 range and all of them fell within a + 5 range of the central curve. Looking at A-6a soils, nineteen of the twenty-two C-U triaxial tests done gave values within the Terzaghi et al. (1996) + 3 range with the remaining three inside the + 5 degree range. Finally, with the Ohio A-7-6 soils, six of eleven values fell into the Terzaghi et al. (1996) + 3 range. The remaining five were within a + 5 range of the central curve. Overall, the use of the Terzaghi et al. (1996) correlation between effective friction angle and plasticity index is moderately reliable for A-4a and A-7-6 soils and very reliable for A-6a soil in Ohio. These were presented as Level 2

155 155 correlations in Section 5.5. Also presented in Section 5.5, as a Level 1 correlation, are the effective friction angles for A-4, A-6, and A-7-6 soils. These correlations are conservative estimates that represent the lower range of the results given in Chapter 4. However, they should not be used very frequently, as most engineers should turn to the Levels 2 and 3 correlations. In the early sections of Chapter 5, many other correlations were given using ORITE and BBCM data. Linear, exponential, logarithmic, power, and reciprocal models gave only a few strong correlations. The use of hyperbolic, second-degree polynomial, and multi-variable linear regression models, however, gave much stronger correlations while using the same dependent and independent variables, especially the hyperbolic. This was apparent with each type of soil tested. Hyperbolic functions again appeared in Section 5.5 giving Level 3 correlations. The first set of functions given can be used for Ohio soil on any of the three AASHTO types tested in the project. % Gravel, % Sand, or the Plasticity Index can be used to find the effective friction angle of a soil effectively. On the other hand, independent variables such as saturated moisture content, dry unit weight, and specific gravity, showed weaker correlations to the effective friction angle. Analysis was able to go further, too, by looking at each AASHTO classification. These are listed as the Alternative Level 3 correlations. Hyperbolic functions that were found in Section 5.3 are given for specifically A-4a, A-6a, and A-7-6 soils. Correlations regarding the effective friction angle were so good for A-4a soils, a second-degree polynomial model and two multi-variable linear regression models were made available for use also. The

156 156 five strongest equations of the fourteen were Equations 5.8, 5.10, 5.14, 5.16, and These were chosen as the best because of their extremely low average difference value and low standard deviation values. These equations need to be considered by Ohio engineers when performing highway embankment design. 6.3 Recommendations A few recommendations need to be made for future studies. First, many types of statistical analyses were done on the results. Although many good correlations were found, multi-variable nonlinear regression analysis is another model that can be used. This method has the possibility to produce better correlations. Also, the types of soils (AASHTO classification) to be obtained from the final four sites needs to be addressed. With the first five sites, there were 44 C-U triaxial tests done. Half of these contained A-6 soils. Therefore, it would be in the best interest of ORITE and ODOT to obtain mostly A-4 and A-7-6 soils from here on out. The site with the most A-4 soils from the first five was in Lake County. This is located in the Low Lime Glacial Lake Sediment area on Figure 2.3, the Ohio Soils Regions Map. It would, therefore, be desirable to obtain soils from one or two more sites in that area, possibly Lorain, Cuyahoga, or Ashtabula Counties. Also, majority of the A-7-6 soils tested came from the Hamilton County site. This is located in the Glacial Drift of Illinoian Age on the Soils Regions Map. To obtain more A-7-6 soils, it would be desirable to look at future field sites in parts of Knox and Perry Counties or in Highland, Brown, and Clermont Counties.

157 157 REFERENCES ASTM (2004). Standard Test Method for Consolidated Undrained Triaxial Compression Test for Cohesive Soils. Designations D , West Conshohocken, PA, pp Bazaraa, A. R. S. S. (1967). "Use of the Standard Penetration Test for Estimating Settlements of Shallow Foundations on Sand, Ph. D. Dissertation, Civil Engineering Dept., University of Illinois, Urbana-Champaign, IL. Bishop, A. W., Bjerrum, L. (1960). The Relevance of the Triaxial Test to the Solution of Stability Problems. Proceedings of Research Conference on Shear Strength of Cohesive Soils, ASCE, pp Bowles, J. E. (1992). Engineering Properties of Soils and Their Measurement, 4 th Edition, McGraw-Hill, New York, NY, pp Casagrande, A. (1932). The Structure of Clay and its Importance in Foundation Engineering, Proc. Contributions to Soil Mech., , Boston Society of Civil Engineers, Boston, MA, pp Casagrande, A., and Hirschfeld, R. C. (1960). Stress Deformation and Strength Characteristics of a Clay Compacted to a Constant Dry Unit Weight. Proceedings of Research Conference on Shear Strength of Cohesive Soils, ASCE, pp Darcy, H. (1856). Les Fontaines Publiques de la Villa de Dijon, Dalmont, Paris.

158 158 Das, B. M. (2002). Principles of Geotechnical Engineering, 5 th Edition, Brooks/Cole, Pacific Grove, CA, 743 pp. Dept. of Navy (1982). Soil Mechanics Design Manual, NAVFACDM-7.1, Alexandria, VA. Drumright, E. E.. Pfingsten, C. W., Lukas, R. G. (1996). Influence of Hammer Type on SPT Results. Journal of Geotechnical Engineering, American Society of Civil Engineers (ASCE), Vol. 122, No. 7, pp Duncan, J. M., Byrne, P., Wong, K. S., and Mabry, P. (1980). Strength, Stress-Strain and Bulk Modulus Parameters for Finite Element Analysis of Stresses and Movements in Soil Masses. Report No. UCB/GT/80-01, College of Engineering, University of California at Berkeley, California. Johnson, G. O. (1975). Engineering Characteristics of Ohio Soil Series, Report No. OHIO-DOT-75, 3 Volumes, Columbus, OH. Masada, T., Sargand, S. M., and Liao, Y. (2006). Resilient Modulus Prediction Model for Fine-Grained Soils in Ohio: Preliminary Study. Proceedings of the International Conference on Perpetual Pavements (ICPP). Peck, R. B., Hanson, W. E., and Thornburn, T. H. (1974). Foundation Engineering, 2 nd Edition, Wiley & Sons, Inc. New York, NY Schmertmann, J. H. (1975). Measurement of In-Situ Strength. Proceeding of the Conference on In-Situ Measurement of Soil Properties, American Society of Civil Engineers, pp

159 159 Schmertmann, J. H. (1979). Statics of SPT, Journal of the Geotechnical Engineering Division. Vol No. GT5. pp Seed, H. B., Arango, I., and Chan, C. K. (1975). Evaluation of Soil Liquefaction Potential During Earthquakes, Report No. EERC 75-28, Earthquake Engineering Research Center, University of California, Berkeley. Sieken, Inc. < Accessed January 20, Skempton, A.W., (1953). The Colloidal Activity of Clay. Proceedings of the Third International Conference on Soil Mechanics and Foundation Engineering, London, Vol. 1, pp Skempton, A. W. (1986). Standard Penetration Test Procedures and the Effect in Sands of Overburden Pressure, Relative Density, Particle Size, Aging and Overconsolidation, Geotechnique, Vol. 36, No. 3, pp Stroud, M. A. and Butler, F. G. (1975). The Standard Penetration Test and the Engineering Properties of Glacial Materials. Proc. Symp. On Engineering Properties of Glacial Materials, Midlands Geotechnical Society, Birmingham, pp Terzaghi, K, Peck R. B., and Mesri, G. (1996) Soil Mechanics in Engineering Practice, 2nd Edition, Wiley & Sons, Inc. New York, NY pp. 549.

160 160 APPENDIX A: SPT CALIBRATION TEST DATA & C-U TRIAXIAL COMPRESSION TEST INSTRUCTIONS Below is the report from GRL on equipment calibration.

161 The following is a detailed set of instructions on running a CU triaxial compression test. 161

162 162 C-U Triaxial Test Instructions - Shelby Tube sample of clayey-silty material 1) Turn on the system (This means the computer, panel, etc.). Then, turn on the water and air pressure using the valves on the wall. Turn the panel pressure up to 60 psi. Next, drain the crappy water from the wall, through the FILL CELL port, into the water drain bucket, so that it won t be used in the test. Check that all the tubing within the cell is clean so that nothing will clog during the test. Fill the deairing water tank (to 1 from the top) with water from the wall. Turn on the vacuum. Then, switch to the VACCUUM position for the top TANK CONTROLS valve on the panel. This will begin deairing the water. Deair the water for at least 4 hours. 2) Measure off on the Shelby Tube where it is to be cut so that the experimenter obtains the desired sample height. 3) Cut the tube with the electric saw. 4) Using the Dremel rotary tool, shave off any pieces of metal around the perimeter of the Shelby tube that will affect the sample when it is being removed. Jack the sample out of the Shelby Tube with the jacking device. (The jacking process may take a few steps. It may be necessary to place porous stones and circular weights between the sample and the jacking device.) If the sample comes out of the tube in relatively good form, move on to Step 5. If not, go to Step 4a. *If it appears that it will be difficult to remove the specimen from the Shelby tube in good form, it may be necessary to cut the Shelby tube lengthwise, and, then, peel off the pieces of metal. 4a) In this case, the soil has come out of the Shelby Tube in poor testing form. This means that the experimenter will need to remold the specimen. So, for remolding the specimen, first, smash up the soil on the floor using a concrete cylinder. If the soil is fairly wet already, the smashing process will not need to be done. Then, put the soil into a large metal bowl. Add water to the bowl in order to obtain the desired water content (if the desired water content is already present, do not add any water). For a remolded specimen, the desired water content is the optimum moisture content. It s typically difficult to measure the desired/optimum moisture content. It can be approximated, however, by feeling the soil and kneading it together. When mixing soil with water in the large metal bowl, slowly add water. Keep kneading the soil with the water. When the soil sticks together in the experimenter s hand (no crumbling of dry pieces), the soil is at approximately the optimum moisture content. It is important to not go over the optimum moisture content. If the experimenter squeezes some of the soil in their hand and a muddy water/slime mixture comes out of the hand, the soil is over its optimum moisture content. So, if water has been added to the soil and the optimum

163 moisture content has been reached, go on to Step 4b. If no water was added to the soil, and the soil is simply sitting in the bowl, go on to Step 4c b) If water has been added to the soil, place Saran Wrap over the bowl and tape it down on the sides so that it forms an airtight seal for the soil inside of the bowl. Here, the experimenter wants the new water added to the bowl to mix in to the soil more. Let the soil sit in the bowl with the airtight seal for at least 16 hours. After at least 16 hours, the experimenter can remove the Saran Wrap and go on to Step 4c. 4c) Next, it s time to physically form the remolded specimen. First, weigh the mold, porous stones, and plastic limit plate separately. Then, place wet paper towels on the insides of the mold so that they actually stick to the insides of the mold. Clamp the two pieces of the mold together. Then, get out a plastic limit plate. Place a porous stone in the middle of the plastic limit plate. Place a paper towel over the porous stone. Place the mold over the porous stone and paper towel. Then, begin compacting the specimen in the mold. Six layers should be formed with the compaction process. Scarify the top of each layer before adding a new layer. The desired soil density can be reached by measuring the mass of the soil sample plus the mold, porous stones, and plastic limit plate during or after compacting the soil. The volume of the mold and its mass, along with the mass of the plate and porous stones, should be known, so a simple calculation can be done to find the density. While compacting the specimen, make sure that the tamping device used is less than half of the interior diameter of the mold. Finally, after the desired height and density of the specimen is reached, remove the mold from the specimen and move on to Step 5. (Note: After molding the soil, it may appear that the specimen will not be able to stand on its own. If this is the case, place a paper towel over the top of the specimen and place a porous stone over the paper towel. Do not place anything on the porous stone. Let the specimen sit for two hours in the mold. After two hours, take off the stone and paper towel and touch the top of the specimen. If the specimen feels drier and more stable, remove the mold. If the specimen does not feel any drier, place the paper towel and stone back on and let it sit for a few more hours. The experimenter should do this process until they feel certain that the specimen can stand on its own. Once the mold is removed from the specimen, it is difficult to place back on. Likely, the experimenter will need to completely remold the specimen.) 5) Now, measure the height and diameter of the specimen and find its mass. (The mass of the specimen should be the same as was found in the Step 4c density calculation, if Step 4c had been done). When measuring the height of the specimen, do three measurements along the longitudinal axis, each measurement 120 degrees apart, to get an average height of the specimen. Then, do three separate diameter measurements at the quarter points of the specimen height. This will get you the average specimen diameter. Record the readings. 6) The specimen should be approximately 6 inches tall and exactly 2.8 inches (A specimen that comes out of the Shelby tube will be 2.8 inches in diameter and will not

164 164 need its diameter shaven down.) in diameter (The height to diameter ratio needs to be between 2 and 2.5.). Even if the specimen meets the proper height requirements, it is desirable to saw off a small amount of soil on one end of the specimen. This soil will be used for a moisture content test. An initial moisture content test must be performed with soil used in the test. So, soil shavings must be obtained from the jacked out sample prior to mounting by either shaving down the height of the specimen or shaving down its diameter. Shaving down the height of the specimen is done most easily with a miter box and saw. Also, no particle can be within the specimen that is larger than 1/6 th of the specimen diameter. The specimen cannot be used if it does not meet this criteria. The experimenter can find a large particle by simple visual inspection. Also, an individual reading of the specimen height or diameter should not deviate by more than 5% from the average reading. (For example; if the three readings of specimen height are 5.7, 6.2, and 5.7, the average is ; the reading of 6.2 deviates from the average by 5.68%.) If this occurs, saw down the specimen so that there is no longer a large deviation with the average height or diameter. (If, during the sawing of the specimen, small pieces of soil fall out leaving voids, carefully remold the specimen at the voided areas with shavings of the soil.) Save the shavings from the sample in a tin to be used in Step 7 (Before placing the trimmings in the tin, place the tin on a balance and record the mass of the empty tin.). Finally, find the specimen mass again and measure its height and diameter a final time. Record these readings. From these readings, calculate the specimen s volume and unit moist unit weight. 7) Perform a water content determination on material trimmed from the sample. This test needs to be in accordance with ASTM Test Method D The mass of the tin should already be recorded. So, next, find the mass of the wet soil trimmings that are in the tin. Next, place the soil trimmings and the tin into the oven. (Note: The rest of Step 7 will likely be completed some time during the saturation of the soil specimen.) The trimmings should remain in the oven for at least 16 hours. After 16 hours, record the mass of the tin with the soil trimmings. Then, place the tin and soil trimmings back into the oven for at least 4 more hours. After this 4 hour period, record the mass again. If the change in mass over the 4 hour drying period is less than 0.1%, the drying process of the soil trimmings is completed. If the change is greater than 0.1%, continue the process of drying for 4 hours and recording the mass until the change between the two mass readings is less than 0.1%. Finally, the water content of the soil trimmings can be found by the equation: water content ={ [ (mass wet trimmings) (mass of dry trimmings) ] / [mass of dry trimmings] } x 100 Record the water content on a sheet along with the Shelby tube sample it corresponds to. The water content value needs to be recorded to the nearest 0.1%.

165 165 8) Next, find the effective consolidation stress (The effective consolidation stress is the chamber pressure minus the back pressure during consolidation. For example, if we have a back pressure of 30.0 psi and a chamber pressure of 45.0 psi during consolidation, then the effective consolidation stress will be 15.0 psi.) of the soil sample. The first step here is figuring out the past overburden stress of the soil specimen. This is found by multiplying the unit weight of the soil by the depth of the midpoint of the specimen. Our effective consolidation stress needs to be significantly greater than the past overburden stress. 9) Put the two porous stones into the oven for one hour. Then, put the stones into a bowl filled with desiccate pellets for 30 minutes to cool them down. 10) Mount the soil sample: a) Put the dry stone on top of the plastic piece on the bottom of the triaxial cell. Put a piece of dry filter paper on top of the stone. b) Put a membrane onto the metal o-ring slider (Have the membrane cover the inside of the slider and pull the ends of the membrane onto the outside surface of the slider.). After the membrane is secured on the slider, hook a hose up to the hole on the side of the slider. Then hook up the other end of that hose to the VACUUM HOSE port on the panel. c) Turn on the vacuum. Toggle the VACUUM HOSE port. The membrane should stick to the inside of the slider. d) Put the soil specimen on its side on a green glass plastic limit plate. Then, put a piece of stripped filter paper around the specimen (The stripped filter paper helps decrease the length of the drainage path for the specimen). Small pieces of tape can be used to secure it around the sides of the specimen. Make sure that the filter paper is shorter than the length of the specimen. This will prevent excess resistance when shearing the specimen. Carefully cut the ends of the filter paper down if it is too long. e) Put the clay specimen on top of the filter paper and bottom porous stone. Put the slider (with the membrane sticking to the insides of it) over the soil specimen and the plastic piece at the bottom of the cell. Raise the slider a little bit and slide the bottom of the membrane off onto the plastic piece at the bottom of the cell. Then, put a piece of dry filter paper on top of the specimen. Then, place a dry porous stone over the paper. Next, place the top plastic piece on the porous stone. Slide the top of the membrane over the top plastic piece. Pull the slider over the plastic piece and place it off to the side. Finally, turn off the VACUUM HOSE toggle and unhook the vacuum hose from the panel and the slider.

166 166 f) Put two o-rings onto each end of the metal o-ring slider (4 o-rings in total). g) Put the slider over the specimen and membrane and slide two o-rings onto the plastic piece at the bottom of the cell. The two o-rings on the bottom form a water-tight seal over the bottom of the membrane. Then, pull the slider up a little bit and slide two o-rings onto the plastic piece at the top of the specimen. These two o-rings form a water-tight seal at the top of the membrane. h) Put a thin coat of vacuum grease around the outside of the ends of the two 1/8 tubes coming out of the bottom of the cell. Make sure to not clog the ends of the tubes. Then, put the tubes into the holes in the plastic piece. Make sure the tubes are secure in the holes in the top plastic piece. Pressurized water will travel through these pipes into the soil specimen. i) Make sure all 4 valves are closed. Hook up a 1/8 tube from POSITION 1 to valve 4 on the cell. Then, dial up a vacuum pressure of 5 psi with the vacuum regulator. Finally, on POSITION 1, turn the top valve to VACUUM, turn the bottom valve to ON, make sure valves 1, 2, and 3 are closed, and open up valve 4. 11) Put an o-ring and vacuum grease on the bottom of the cell and on the top of the cell. 12) Put the cylinder over the specimen and onto the bottom o-ring. 13) Put on the top and lock it onto the top of the cylinder. Also, at this point, lock the piston into place on top of the plastic piece on top of the sample. Putting the piston into the plastic piece on top of the specimen is an essential way to check the eccentricity. If the alignment of the specimen is off, then first turn off the vacuum, then very gently adjust the cap and specimen by lightly tapping the piston on the specimen cap so that the alignment works. Once the specimen is aligned, you can turn the vacuum back on. Before going onto Step 14, reduce the vacuum by 0.2 psi if the effective consolidation stress is below 5.2 psi. If the effective consolidation stress is above 5.2 psi, don't change anything. 14) Fill the cell. This is done by, first, hooking up the tube with two quick-connects to the FILL CELL port and the bottom port on the cell. Then, hook up a tube (this tube should have a quick connect on one end and be open on the other end) to the top of the cell. Next, open the FILL CELL toggle. Water should start pouring into the cell. When the water is about 1 inch from the top of the cell, physically tilt the cell back so that the port on the top of the cell is higher than the rest of the top. Tilting the cell greatly reduces the amount of water that can get trapped in the cell.

167 167 Finally, water should start spurting out of the open-ended tube. Make sure the open-ended tube can drain into the water drain bucket. Once the open-ended tube has no more air bubbles in it, turn off the FILL CELL toggle. Then, unhook the top tube. *Note: Water comes out of the FILL CELL port at a high velocity. Once the water starts coming out of the open-ended tube, it might start spraying everywhere. Try to use one hand to hold down the end of the tube inside the water drain bucket. Use the other hand to tilt the cell. Once it appears all the air bubbles are out of the open ended tube, rest the cell on its three legs on the table and use that hand to close the FILL CELL toggle. 15) Now, the cell should be filled with water. Fill the burette of POSITION 3 (pipette and annulus) with water to the 21 ml marking. Make sure the cell is resting on the table where the panel is resting (not the load cell). Take the double quick-connect tube (the one used to fill the cell) out of the FILL CELL port and put it into the port of POSITION 3. Make sure the top valve of POSITION 3 is on VENT. Then, turn the bottom valve of POSITION 3 to ON. With the cell on the table, water at the 21 ml marking of POSITION 3, and having POSITION 3 vented to the atmosphere, the top of the cell is at a pressure of approximately 0.0 psi and the bottom of the cell is at a pressure of approximately 0.40 psi (The reason for reducing the vacuum in Step 13 was to prevent overconsolidation of the specimen that would result from the water column in the cell; the middle of the specimen is exposed to approximately 0.2 psi pressure from the unpressurized chamber). Now, increase the vacuum to its maximum pressure. The Welch brand vacuum used has a maximum pressure of about psi (1 in Hg = psi). This means that the experimenter can increase the vacuum to maximum pressure only if the effective consolidation stress is greater than psi (13.25 psi psi = psi; the middle of the specimen will be exposed to an approximate pressure of 0.2 psi from the confining water.). However, if the effective consolidation stress is lower than psi, the vacuum needs to be adjusted accordingly to not overconsolidate the specimen. For example, if the effective consolidation stress is 8.5 psi, then the vacuum pressure should be decreased to 8.3 psi. Now, let the specimen sit with the new vacuum pressure on it and the sides being confined by water for 10 minutes. While the specimen is sitting, open up the Sigma-1 CU program on the computer, and, then, zero the pore pressure sensor and the cell pressure sensor. Hook the pore pressure sensor up to valve 3 making sure there is access to the bleed port screw. Keep valve 3 closed. Next, fill up the pipette of POSITION 2. Apply a pressure of 2.5 psi to the pipette of POSITION 2. Then, turn the vacuum pressure down to 2.0 psi. Essentially, the experimenter will create a pressure difference of 4.5 psi between the top and bottom of the specimen (positive 2.5 psi from the bottom and negative 2.0 psi from the top). Next, turn the bottom valve of POSITION 3 to OFF. Then, take the double quick-connect tube out of the bottom port of the cell and put it in the top port of the cell. Then, put the cell pressure sensor into the bottom port of the cell.

168 168 Now, turn the top valve on POSITION 3 to PRESSURE and slowly dial up a pressure of 2.8 psi (2.8 psi coming in at the top of the cell will act as approximately 3.0 psi in the middle of the specimen). Switch the bottom valve of POSITION 3 to ON. Now, there should be a 2.8 psi pressure coming from POSITION 3 and a 2.0 psi vacuum pressure at the top of the specimen. Then, hook up a 1/8 diameter tube to the port of POSITION 2 (don t hook up the open end of the tube; instead, aim it towards the water drain bucket). Slowly turn the valve on the bottom of POSITION 2 to ON. Water should start coming through the tube and draining into the bucket. Once it appears that there are no air bubbles in the tube, turn the bottom valve of POSITION 2 to OFF. Finally, hookup the 1/8 tube from POSITION 2 to the bridge valve. Then, open up valve 2. Finally, turn the bottom valve on POSITION 2 to ON. The end result should be a 2.8 psi confining pressure, a 2.0 psi vacuum pressure at the top of the specimen, and a 2.5 psi positive pressure acting at the bottom of the specimen. Note: The reason for adding the 2.8 psi confining pressure is so that the specimen does not expand when water is pushed into it through the bottom. A 2.8 psi confining pressure coming out of POSITION 3 will act as an approximate 3.0 confining pressure (hydrostatic pressure) in the middle of the specimen. 16) Loosen up the bleed port screw on the pore pressure sensor. Then, open up valve 3 on the cell. Water should start coming out of the bleed port. Watch the 1/8 tube that goes into valve 3. There will likely be some bubbles going through it as the bleed port screw is open. After about 20 seconds all the bubbles should be out. At this point, snug the bleed port screw. Remember to not overtighten the screw, just make sure it s snug. Also, at this point, if for some reason the water in POSITION 3 starts decreasingly rapidly, then turn off the bottom valves of POSITION 1, then POSITION 2, and then POSITION 3 (in that order). Then, turn the top valve of POSITION 3 to VENT, and fill up the burette with deaired water. Finally, turn the top valve on POSITION 3 back to PRESSURE. Then, turn the bottom valve to ON for POSITION 3, then POSITION 2, and then POSITION 1 (in that order). During the course of the experiment, the three positions should be turned on and off in the manner just described if water ever needs to be drained or added to POSITIONS 1, 2, or 3. Also, it s necessary to keep observing the pipette of POSITION 1. The time will vary for every experiment (It could be 5 minutes, or it could be 3 hours.), but, eventually, water will start rising in the pipette of POSITION 1. Once water appears to rise in the pipette of POSITION 1, turn the bottom valve of POSITION 2 to OFF. Now, there is a 2.8 psi confining pressure and a 2.0 psi vacuum pressure on top of the specimen. Next, turn the vacuum pressure from POSITION 1 down to 0.0 psi. Next, turn the top valve on POSITION 1 to VENT. Then, turn the pressure on POSITION 3 up to 4.8 psi. (Note: While it's difficult, the experimenter should try to simultaneously decrease the vacuum pressure while increasing the chamber pressure. This is done so that the pore pressure at the top of the specimen and the pore pressure at the bottom of the specimen can equalize. The process should be done slowly. It may be advisable to reduce the

169 169 vacuum pressure by 0.5 psi, and, then, increase the chamber pressure by 0.5 psi, and, then, keep that process going.) Finally, observe the pore pressure reading on the computer screen. Once it appears to have stabilized (Stabilization can take 10 minutes, 60 minutes, etc.; it all depends)., close valves 2 and 4. Again watch the pore pressure reading on the screen. If it changes by less than 5% of the value of the chamber pressure, then stabilization has been reached and Step 17 can begin. 17) Now, it s time saturate the soil specimen. Close off the bottom valves of all three positions. If at anytime during the experiment, POSITION 3 needs to be refilled or drained, it s imperative to first close POSITION 1, then POSITION 2, then POSITION 3 (in that order). So, at this point, drain all the water from the annulus of POSITION 3. Leave the pipette of POSITION 3 about half full. Next, fill the pipettes of POSITIONS 1 and 2 about halfway. Now, there should be three pipettes all filled approximately halfway. Next, dial up a chamber pressure of 32.0 psi to POSITION 3. Dial up a back pressure of 30.0 psi to POSITIONS 1 and 2 (the back pressures should be the same). Make sure that valves 2 and 4 are open. Finally, open POSITION 3 (Open up POSITION 3 very slowly. It's likely that the 32.0 psi pressure will send the water shooting down the POSITION 3 pipette very fast. So, watch out for this. If it happens, make sure to close the bottom valve on POSITION 3, turn the top valve of POSITION 3 to VENT, and, then, turn one of the middle valves on POSTION 3 to tank. Fill the pipette of POSITION 3 about halfway. Then, turn the top valve on POSITION 3 to PRESSURE, and turn the bottom valve on POSITION 3 to ON. Keep doing this process until the pipette of POSITION 3 can maintain a 32.0 psi pressure without sending all of the water out.), then POSITION 2, and then POSITION 1. Record the pipette readings for each position. Recording the pipette readings allows the experimenter to monitor the saturation process closely and possibly determine when saturation is complete. 18) Over the first few hours of attempted saturation, it s important to visually monitor the process. Record the readings approximately every three hours. Ideally, water levels in POSITIONS 1 and 2 should be decreasing. The water from these positions should be going into the specimen. While saturation is taking place, one of the positions may become close to drained or filled completely. If this happens, simply turn off POSITION 1, then POSITION 2, and then POSITION 3 (record the pipette readings before turning off the positions). Drain or refill the proper pipette, then turn on POSITION 3, then POSITION 2, and then POSITION 1 (in that order). Record the pipette readings, at this point. Overall, the saturation process can take anywhere from one day to five days. 19) After monitoring the pipette readings for many hours or days, it may appear that saturation is complete. A B-check needs to be done to make sure that saturation is indeed complete. This is performed by, first, closing off POSITIONS 1 and 2 (this will make the sample undrained). Then, watch the pore pressure readings from the Sigma-1 CU program on the computer. Once the pore pressure readings have stabilized, the B-check

170 170 can be done. So, at this point, click on the Tools button at the top left of the Sigma-1 CU program. Select B Check from the menu. A little window will pop up on the screen. Click the Start button on the window. There is a timer on the window that will start counting up. At this point, turn the chamber pressure up by about 10.0 psi on the panel. On the computer screen, a number should appear for the B coefficient. After 2 minutes of taking readings, record the highest B-value that shows up on the screen. If this value is at 0.95 or higher, then saturation has been achieved. If 0.95 is not reached, then the experimenter must continue back pressure saturation until a 0.95 B-value is achieved. Follow one of the two options below. 1 - Saturation has been achieved: Simply click Done on the B-check window. Then, move onto Step Continuation of back pressure saturation: After the B-check is done, close out of the window. Then, reduce the chamber pressure back to 32.0 psi. Next, turn the bottom valve of POSITION 2 to ON, then turn the bottom valve of POSITION 1 to ON. Finally, monitor saturation the same way it was done before. Do not move onto Step 20 until saturation has been achieved. * It's important to note that some soils will simply not reach a b-value of 0.95 after many days of saturation. This will be noticed if for three or four days the b-value does not increase. If this occurs, stop the saturation process and go onto Step 20. Also, make a note of this on the checklist or write a note up on the computer and talk about what happened. 20) Now, it s time to consolidate the specimen. It s good at this point to make a table where readings from the pipettes can be made. Record pipette readings, pore pressure readings, and cell pressure readings, and caliber readings, at times of 0.1 minutes, 0.2 minutes, 0.5 minutes, 1 minute, 2 minutes, 4 minutes, 8 minutes, 15 minutes, 30 minutes, 1 hour, 2 hours, 4 hours, 8 hours, 16 hours, and 24 hours. Make sure that the bottom valves on POSITIONS 1 and 2 are still closed. Turn the bottom valve on POSITION 3 to OFF. Increase the pressure in POSITION 3 so that it is more than the back pressure in POSITIONS 1 and 2 (POSITIONS 1 and 2 need to be at a pressure of 30.0 psi) by the amount of the effective consolidation stress. Next, take an initial reading of the pipettes, cell pressure, pore pressure, and caliber. Now, turn the bottom valve of POSITION 3 to ON, then turn the bottom valve of POSITION 2 to ON, and, finally, turn the bottom valve of POSITION 1 to ON. Then, start making readings at 0.1 minutes. Continue making readings for the specimen at the specific times just listed. When 24 hours of consolidation has amounted, turn off the bottom valves of POSITIONS 1 and 2. Finally, it's important for the pore pressure to stabilize. After closing the valves on POSITIONS 1 and 2, watch the pore pressure reading. It should be fairly stable after the 24 hour consolidation process. After about 5 minutes, compare the pore pressure reading at the 24 hour consolidation reading to the current reading. If the change in pore pressure is less than 5% of the chamber pressure, the sample is assumed to have a stable

171 171 pore pressure and the experimenter can go onto Step 21. If the change is greater than 5%, the drainage valves (valves of POSITIONS 1 & 2) need to be reopened and consolidation needs to continue until pore pressure readings appear stable. Once again, stabilization can be checked by closing the valves to POSITIONS 1 and 2 and measuring the change in pore pressure. If it's less than 5% of chamber pressure, then stabilization has been reached. 21) Now, it's necessary to calculate the strain rate for the loading of the specimen. Assuming that failure will occur after 4% strain, the strain rate can be found using the following equation: ε = 4 / [(10)(t 50 )] The term, t 50, can be found in the following procedure. Finding t 50 : Plot the burette readings on the y-axis and the log of time in minutes on the x-axis. This can be done on Excel. It s best to choose the Excel graph function that places a curved line on the points, connecting them. Print out the plot. Next, draw a straight line through the points that represent the final readings on the plot. These points should already be in an approximate straight line and at a constant slope. Then, draw a straight line through the steepest part of the plot. Make sure that these two straight lines cross each other. The intersection of these two lines represents d 100 and t 100. Next, the experimenter needs to find the deformation representing 0% primary consolidation. First, find two points on the plot that have a time ratio of 1 to 4. These two points need to occur before the steepest part of the curve occurs. It is suggested to use the points at times of 0.5 minutes and 2 minutes. Next, find the deformations at these two points. Find the difference between the two deformations. Then, subtract the difference between the two readings from the reading of the lower time (Essentially, if using times of 0.5 minutes and 2 minutes, find the difference between the deformation at 2 minutes and 0.5 minutes and subtract this difference from the deformation at 0.5 minutes). It is possible that the final result (the deformation representing 0% primary consolidation) will be above (at less deformation) than the deformation at 0.1 minutes. This is fine and expected. Finally, the average d 100 and the deformation representing 0% primary consolidation. This deformation is d 50, the deformation representing 50% primary consolidation. Find the point on the specimen s deformation plot where d 50 corresponds. This point is also where t 50 is. Now that t 50 has been found, it can be plugged into the above equation to find ε. ε is the strain rate used while loading the specimen.

172 172 When calculating the strain rate, t 50 is in minutes. If, for example, t 50 is 5 minutes, then the calculation will look like the following: 4 / (10 5) = 0.08 This gives a strain rate of 0.08% per minute or 4.8% per hour. If additional assistance is needed for finding t 50, go to ASTM D Look at sections 12.3, , , , and This will give some helpful advice. If the experimenter believes that the specimen will fail at a strain lower than 4%, the term 4 in the above equation can be changed to the proper, lower strain value. 22) Place the triaxial cell on the loading device and raise it to about 0.1 inches from the load cell. The cell can be raised using controls from the Sigma-1 CU computer program. Zero the loading cell and the DCDT using the Sigma-1 CU computer program. Then, on the top left of the Sigma-1 CU window, click on File. Go to Specimen Data. Fill in all of the proper entries (Remember: The height of the specimen is the initial height minus the change after consolidation.). Make sure to save the shear test data to the proper folder. Then, go back to File, and click on Test Data. Fill in all of the proper entries here. Finally, check the entries in the top right of the screen in the Sigma-1 CU program to make sure they are correct (height, stain rate, etc.) Then, click on Start Test. A window will pop up that says Unlock the cell piston and click, OK. So, physically unlock cell piston and let it rise (slowly if you can) so it touches the load sensor. Then, click OK on the window that popped up on the computer. Note: The chamber pressure used in consolidation is the same pressure that needs to be used in shear testing. So, for example, if the effective consolidation stress of the sample is 15.0 psi, the chamber pressure during consolidation should be 45.0 psi. After consolidation was completed, POSITIONS 1 and 2 needed to be closed. So, now, with the shear test starting, the chamber pressure should still be at 45.0 psi with POSITIONS 1 and 2 closed off. 23) After OK has been clicked, a window will pop up that says Measuring piston correction. Then one will pop up that says Seating in progress. If something else pops up, go to the notebook Sigma-1 5K: Automated Load Test System 5K, and on page 57 and 58, there are some helpful instructions.

173 173 24) At this point, unless something is wrong as mentioned in the previous step, the test should be running. You can view plots or other information with the Sigma-1 CU program on the computer. While the test is running, it is important to observe the graphs showing the principal stresses on the screen. There are three modes of failure while running the CU compression test. The three modes are shown below. 1) Stop the test once axial strain has reached 15 %. 2) Stop the test after the behavior seen below occurs. 3) Stop the test after the behavior seen below occurs.

174 174 25) Once the test is finished, there should be a bulge in the center of the specimen. Lock the piston. Lower the platen using the controls on the computer program so it s possible to get the cell off of the platen and onto the table. Take the cell pressure sensor out of the bottom port of the chamber. Next, drain the chamber. This is done by placing the open ended tube with one quick-connect into the bottom port of the chamber. There should still be a chamber pressure being applied by POSITION 3 going into the top of the chamber. Water should now be spurting out of the tube from the bottom port of the chamber. Once all of the water is out of the chamber, turn the bottom valve of POSITION 3 to OFF. Take the double quick-connect ended tube out of the top port on the triaxial chamber and the port of POSITION 3. Next, close valve 3 on the triaxial chamber. Then, take the pore pressure sensor off of valve 3. Also, remove the 1/8 tubes from the bridge valve and valve 4. Hold these tubes over the water drain bucket and drain the water from them. Now, drain the water from the pipettes of POSITIONS 1 and 2. There should already be a pressure of 30.0 psi on each of the pipettes. Then, take the chamber apart. Pull the 1/8 tubes out of the top plastic piece. Next, remove the top o-rings that cover the rubber membrane, slide down the top of the membrane, pull off the top plastic piece, remove the porous stone, and take the specimen off of the bottom porous stone. Take all of the filter paper off of the specimen. Place the specimen on the scale and record its mass. Set the specimen on a table and make a sketch of it on paper. Finally, place the specimen in a large tin and place it in the oven. Measure to find the mass of the tin before putting it and the specimen into the oven. Next, take the rubber membrane and the bottom o-rings off the plastic piece on the bottom of the cell. Also, remove the two large o-rings from the top and bottom of the cell. Clean all of the vacuum grease off of the cell (chamber) and o-rings also. Essentially, at this point, clean up everything and put it back to where it came from. This

175 includes the tubing, porous stones, pieces of the chamber, o-rings, etc. Also, at this point, the experimenter can reduce the pressures in each of the positions to 0.0 psi. Make a note of any unique features noticed with the specimen at this point also ) The test is over, and the results can be viewed on Excel. Close out the Sigma 1- CU program on the computer. The results can be viewed by opening up the CU triaxial test template and importing and calculating the proper file. The results will show graphs containing the principal stresses, the shear stresses, pore pressures, etc. 27) After 24 hours, remove the specimen from the oven and find its mass. Then, put it back into the oven for another 4 hours. Then, measure the specimen's mass again. If the change in mass is less than 0.1% over the 4 hour period, then the specimen is dry. If the change is over the 4 hour period is more than 0.1%, then continue the process of drying and measuring mass of the specimen for four more hours until the change in mass is less than 0.1%. Using the mass of the dry specimen and the mass of the wet specimen from Step 25, find the moisture content. Next, a series of calculations will be found as required by ASTM. Using the Excel spreadsheet titled Step 27 calculations, do the following: a) Initial specimen properties - Fill in values for the initial specimen volume (from Step 6), initial water content (from Step 7), the initial moist mass of the specimen (from Step 6), the initial unit weight of the soil (from Step 6), and the specific gravity supplied by BBC&M. Filling in the proper cells will give the other needed values. b) Specimen properties after consolidation Fill in values for the initial height of the specimen (from Step 6), the change in specimen height from consolidation (from Step 20; this can be done by using the change in caliber length), the final water content of the soil, and the wet mass of the specimen after shearing (from Step 27). Filling in the proper cells will give the other needed values. c) Principal stress difference and induced pore-water pressure versus strain curves Fill in values from the Sigma-1 CU template for the axial strain at failure, effective major principal stress at failure, the effective minor principal stress at failure, and the pore pressure at failure. Filling in the proper cells will give the other needed values. Also, after drying it may be necessary to take apart the specimen. Take any notes on the specimen and any unique features in it (shear failure plane, large rocks, etc.). Other necessary calculations that can be done are explained in Section 10 of ASTM D Section 11 of ASTM D gives information on Test Data Sheet(s)/Form(s). This includes Mohr-Coulomb diagrams and p-q diagrams.

176 176 APPENDIX B: SUBSURFACE EXPLORATION DATA Table B.1: Recovery and Notes from Shelby Tubes Taken by ORITE from the Hamilton County Site Tube Depth range (ft) Recovery (in) Notes A Bottom end is somewhat crushed. B D A C D Tube is slightly pushed inward along one side. A B D Figure B.1: SPT Truck (Hamilton County Site)

177 177 Figure B.2: Continuous SPT Being Done (Hamilton County Site) Figure B.3: Logging of Soil in the Split-Spoon Sampler (Hamilton County Site)

178 178 Figure B.4: BBC&M Workers Attaching the Shelby Tube (Hamilton County Site) Figure B.5: BBC&M Worker Filling Shelby Tubes with Wax (Hamilton County Site