UFDC Home  myUFDC Home  Help 



Full Text  
PAGE 1 UPDATING FLORI DA DEPARTMENT OF TRANSPORTATION'S (FDOT) PILE/SHAFT DESIGN PROCEDURES BASED ON CPT & DTP DATA By ZHIHONG HU A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2007 1 PAGE 2 2007 Zhihong Hu 2 PAGE 3 To my wife for her strong support, endless encouragement, and help to make this happen 3 PAGE 4 ACKNOWLEDGMENTS I first thank to my wife, Meiyu, for her endless support, advice, and encouragement of my research and my dissertation. Without her support, this dissertation could not be finished. I would also like to thank to my parents, who also encouraged me to finish my study and work. I want to say thank you for Dr. Bloomquist for his innovate ideas, helpful advice, and guidance to fulfill this dissertation. I also want to thank to Dr. McVay for his insight, knowledge, patient discussions, suggestions, and good research model. There are also many other professors who devoted time and energy to my research, such as Dr. Townsend, Dr. Hiltunen, Dr. Consolazio and Dr. Flood. I would like to thank them, as well. Finally, I would like to thank to my classmates, Jeongsoo Ko, Scott Wasman, Patrick Dunn, Luis Campos, Adrian Viala, Heath Forbes, and Mark Styler. Thank you for your help with my research. 4 PAGE 5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ...............................................................................................................4 LIST OF TABLES ...........................................................................................................................7 LIST OF FIGURES .........................................................................................................................8 ABSTRACT ...................................................................................................................................15 CHAPTER 1 INTRODUCTION..................................................................................................................17 2 LITERATURE REVIEW.......................................................................................................20 Cemented Sand.......................................................................................................................20 Cone Penetration Test (CPT)..................................................................................................23 CPT Based Pile Capacity Prediction Methods.......................................................................25 Schmertmann Method.....................................................................................................26 De Ruiter and Beringen Method......................................................................................27 Penpile Method................................................................................................................28 Prince and Wardle Method..............................................................................................28 Tumay and Fakhroo Method...........................................................................................29 Aoki and De Alencar Method..........................................................................................29 Philipponnat Method.......................................................................................................30 LCPC (Bustamante and Gianeselli) Method...................................................................30 Almeida et al. Method.....................................................................................................30 MTD (Jardine and Chow) Method..................................................................................31 Eslami and Fellenius Method..........................................................................................35 Powell et al. Method........................................................................................................36 UWA05 Method.............................................................................................................37 Zhou et al. Method..........................................................................................................38 3 MATERIALS AND METHODS...........................................................................................41 Dual Tip Penetrometer (DTP)................................................................................................41 Locate Cemented Sites...........................................................................................................42 Axial Ultimate Pile Capacity Prediction Methods..................................................................42 Load and Resistance Factor Design (LRFD)..........................................................................43 Modified First Order Second Moment (FOSM) Approach.............................................43 Limit state equation..................................................................................................44 Reliability index.......................................................................................................45 Resistance factor ..................................................................................................45 Differentiate Ultimate Skin Friction and Tip Resistance from Load Test Data.....................49 The Proposed UF Method.......................................................................................................51 5 PAGE 6 4 RESULTS...............................................................................................................................59 Identification of Cemented Sand............................................................................................59 DTP & CPT Test Data at FDOT Bridge Sites........................................................................59 Archer Landfill Site.........................................................................................................60 I95 at Edgewood Avenue Site........................................................................................61 Apalachicola River Bridge Site.......................................................................................61 West Bay Bridge Site......................................................................................................62 Port Orange Relief Bridge...............................................................................................62 University of Central Florida...........................................................................................63 I295 at Blanding Blvd Site.............................................................................................63 I295 at Normandy Blvd. Site.........................................................................................63 White City Bridge Site (Pier 5).......................................................................................64 White City Bridge Site (Pier 8).......................................................................................64 Identifier for Cemented Sand Summarized from DTP Data..................................................64 Assessing LRFD Resistance Factors, Based on Reliability (Risk)....................................66 Criteria Used to Quantify Pile Capacity from Load Test Data.......................................66 LRFD Resistance Factor, for Florida Soil..................................................................67 LRFD Resistance Factor, for Louisiana Soil..............................................................68 Evaluate the Prediction Methods Using the Bootstrap Method..............................................69 5 CONCLUSION.....................................................................................................................110 Conclusion............................................................................................................................110 Future Work..........................................................................................................................111 APPENDIX MATHCAD PROGRAM....................................................................................113 LIST OF REFERENCES.............................................................................................................164 BIOGRAPHICAL SKETCH.......................................................................................................167 6 PAGE 7 LIST OF TABLES Table page 31 Ultimate unit tip resistance factor k b .................................................................................53 32 Ultimate unit skin friction empirical factor, F s ..................................................................53 41 Predicted ultimate skin friction for 14 CPT methods (Florida soil)..................................71 42 Predicted ultimate tip resistance for 14 CPT methods (Florida soil).................................72 43 Predicted Davisson capacity for 14 CPT methods (Florida soil).......................................73 44 LRFD resistance factors,, for CPT methods (ultimate skin friction, Florida soil)..........74 45 LRFD resistance factors,, for CPT methods (ultimate tip resistance, Florida soil)........74 46 LRFD resistance factors,, for CPT methods (Davisson capacity, Florida soil)..............75 47 Predicted ultimate skin friction for 14 CPT methods (Louisiana soil)..............................76 48 Predicted ultimate tip resistance for 14 CPT methods (Louisiana soil).............................77 49 Predicted ultimate pile capacity for 14 CPT methods (Louisiana soil).............................78 410 LRFD resistance factors,, for CPT methods (ultimate skin friction, Louisiana soil).....79 411 LRFD resistance factors,, for CPT methods (ultimate tip resistance, Louisiana soil)....79 412 LRFD resistance factors,, for CPT methods (ultimate pile capacity, Louisiana soil)....80 51 Bridge sites where load test data are available................................................................112 7 PAGE 8 LIST OF FIGURES Figure page 21 Regular cone penetrometer................................................................................................40 31 Dual tip penetrometer........................................................................................................54 32 The locations of 21 sites with load test data and CPT data...............................................55 33 MathCAD program for Philipponnat method....................................................................56 34 Static load test, Apalachicola Bay Bridge (pier 3).............................................................57 35 Separate the ultimate skin friction and tip resistance.........................................................58 41 CPT and DTP test data from Archer Landfill site.............................................................81 42 Friction ratio of CPT and DTP rest from Archer Landfill site..........................................82 43 CPT and DTP test data from I at Edgewood Avenue site............................................83 44 Friction ratio of CPT and DTP test from I at Edgewood Avenue site.........................84 45 Tip2/Tip1 ratio and Qc/N ratio of CPT and DTP test from I at Edgewood Avenue site......................................................................................................................................85 46 CPT and DTP test data from Apalachicola River Bridge site...........................................86 47 Tip2/Tip1 ratio and Qc/N ratio of CPT and DTP test from Apalachicola River Bridge site......................................................................................................................................87 48 CPT and DTP test data from West Bay Bridge site...........................................................88 49 Tip2/Tip1 ratio and Qc/N ratio of CPT and DTP test from West Bay Bridge site............89 410 CPT and DTP test data from Port Orange Relief Bridge site............................................90 411 Tip2/Tip1 ratio and Qc/N ratio of CPT and DTP test from Port Orange Relief Bridge site......................................................................................................................................91 412 CPT and DTP test data at the University of Central Florida site.......................................92 413 Tip2/Tip1 ratio and Qc/N ratio of CPT and DTP test at the University of Central Florida site.........................................................................................................................93 414 CPT and DTP test data at I295 at Blanding Blvd site......................................................94 415 Tip2/Tip1 ratio and Qc/N ratio of CPT and DTP test at I295 at Blanding Blvd site.......95 8 PAGE 9 416 CPT and DTP test data at I295 at Normandy Blvd. site...................................................96 417 Tip2/Tip1 ratio and Qc/N ratio of CPT and DTP test at I295 at Normandy Blvd. site....97 418 CPT and DTP test data in White City site (pier 5)............................................................98 419 Tip2/Tip1 ratio and Qc/N ratio of CPT and DTP test in White City site (pier 5).............99 420 CPT and DTP test data in White City site (pier 8)..........................................................100 421 Tip2/Tip1 ratio and Qc/N ratio of CPT and DTP test in White City site (pier 8)...........101 422 Typical load test curve in Louisiana soil.........................................................................102 423 Comparisons (ratio: measured/predicted) for 14 methods using Florida soil..................103 424 Comparisons (ratio: measured/predicted) for 14 methods using Louisiana soil..............104 425 Frequency of of the proposed method for Florida soil.................................................105 426 Frequency of the sample means using Bootstrap method for Florida soil (100,000 resampling runs, the proposed method)..............................................................................105 427 Frequency of the sample standard deviations using Bootstrap method for Florida soil (100,000 resampling runs, the proposed method)..........................................................106 428 Frequency of of the proposed method for Louisiana soil.............................................106 429 Frequency of the sample means using Bootstrap method for Louisiana soil (100,000 resampling runs, the proposed method)..........................................................................107 430 Frequency of the sample standard deviations using Bootstrap method for Louisiana soil (100,000 resampling runs, the proposed method)....................................................107 431 Frequency of of the Schmertmann method for Florida soil..........................................108 432 Frequency of the sample means using Bootstrap method for Florida soil (100,000 resampling runs, Schmertmann method)............................................................................108 433 Frequency of the sample standard deviations using Bootstrap method for Florida soil (100,000 resampling runs, Schmertmann method).........................................................109 9 PAGE 10 LIST OF ABBREVIATIONS Empirical factor to calculate tip resistance in Zhou et al. method C Penetrometer to pile friction ratio in clay in Schmertmann method S Penetrometer to pile friction ratio in sand in Schmertmann method s Empirical factor to calculate skin friction in Philipponnat method A S Surface area for the calculation of skin friction ASTM American society for testing and materials Adhesion factor in De Ruiter and Beringen method and Zhou et al. method T Target reliability index C S Pile skin friction factor in Eslami and Fellenius method CPT Cone penetration test COV (Q) Load coefficients of variation COV (R) Resistance coefficient of variation COV QD Dead load coefficient of variation COV QL Live load coefficient of variation f Pilesoil interface friction angle at the maxmum shear stress cv Constant volume interface friction angle between sand and pile D Diameter or side length of pile D CPT Diameter of cone penetrometer D int The internal diameter for pipe pile DTP Dual tip penetrometer r Interface dilation rd The net dilatant component LRFD resistance factor F b Empirical factor to calculate tip resistance in Aoki and De Alencar method 10 PAGE 11 F s Empirical factor to calculate skin friction in Aoki and De Alencar method F S Empirical factor to calculate skin friction in Philipponnat method FDOT Florida department of transportation FORM First Order Reliability method FOSM First Order Second moment f L Loading coefficient f S Pile unit skin friction f sa Average CPT sleeve friction Friction angle G The sand shear modulus, failure equation in terms of random variables G max Maximum shear modulus D Dead load factor L Live load factor h The hight above the pile tip I295 Interstate 295 I95 Interstate 95 K c Earth pressure coeffient after equilization k Pile tip resistance factor in MTD method k 1 Pile skin friction factor in Almeida et al. method and Powell et al. method k 2 Pile tip resistance factor in Almeida et al. method and Powell et al. method k b Pile tip resistance factor in Prince, Wardle method Philipponnat method, LCPC method, and the Proposed method k S Pile skin friction factor in Prince and Wardle method L Pile embedment length LCPC Laboratoire central des ponts et chausses 11 PAGE 12 LRFD Load and resistance factor design i Load factors QD Dead load bias factor QL Live load bias factor R The mean bias Ri The ratio of measured to predicted pile capacities MTD Marine technology directorate MN/m 2 Mega Newton per square meters m Pile skin friction factor in Tumay and Fakhroo method N The number of cases, or SPT blow count N C Bearing capacity factor N k Cone factor in De Ruiter and Beringen method N kt Cone factor in Powell et al. method N s Cone factor to estimate sensitivity NCHRP National cooperative highway research program OCR Over consolidation ratio PI Plasticity index Pa The atmosphere pressure PPC Precastprestressedconcrete PLAID Pile load settlement analysis from insitu data psi Pound per square inches Q Random variable for load Qc/N The ratio of CPT tip resistance to the SPT blow count Q I Force effects Q D Dead load 12 PAGE 13 Q Davisson Davisson capacity Q L Live load ratio Q D /Q L Dead to live load ratio Q S Pile total tip resistance Q S ult Ultimate pile skin friction Q T ult Ultimate pile tip resistance q c CPT tip resistance q ca Average CPT tip resistance q e The average of the effective cone resistance within the calculation layer q eg The geometric average of the effective cone resistance q eq (tip) Average of tip resistance within 1.5 D above and 1.5 D below the pile tip after eliminating abnormal data (out of the range of 30% of average value) q t Pile unit tip resistance R The diameter of the pile in MTD method, random variable for resistance R cla The piles centerlineaverage roughness R design Design capacity R mi Measured capacity from load test data R ni Predicted capacity form CPT data R n Nominal resistance R.D. Relative density SMO State material office SPT Standard penetration test S t Clay sensitivity S u Undrained shear strength R The standard deviation of Ri 13 PAGE 14 ' rc Radial effective stress on side after equalization rf Radial effective stress at maximum shear stress v0 The total overburden stress v0 The effective overburden stress T1 DTP first tip resistance T2 DTP second tip resistance tsf Tons per square feet UF The University of Florida UWA The university of Western Australia VAR Variance YSR Yield stress ratio y The distance between the surface and the skin friction calculating point Load modifier for importance, redundancy and ductility 14 PAGE 15 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy UPDATING FLORI DA DEPARTMENT OF TR ANSPORTATION'S (FDOT'S) PILE/SHAFT DESIGN PROCEDURES BASED ON CPT & DTP DATA By Zhihong Hu December 2007 Chair: David Bloomquist Cochair: Michael C. McVay Major: Civil Engineering The Florida Department of Transportation (FDOT) began developing a geotechmaterialconstruction database that includes information on piles and drilled shafts, specifically, insitu data (SPT, CPT, etc.) and load test data in the early 1990s. More recently, FDOT sponsored novel research to develop a new insitu device, the dual tip penetrometer (DTP), to identify cemented soils and thereby contribute to the database data. My research evaluated current pile design methodologies (Schmertmann, LCPC, etc.) using CPT, DTP, and modified current methods and proposed a new method to improve future driven pile design. My research also involves identifying cemented soils using the DTP, since cementation is a critical issue in pile design procedures. My research explores 14 pilecapacitydesign methods based on cone penetration test (CPT) and assesses load and resistance factor design (LRFD) resistance factors for each using 21 cases from Florida and 28 from Louisiana. The resulting resistance factors were not satisfactory for any of the methods. A new design method was proposed, taking into account cementation and other issues. The LRFD resistance factor was also assessed for this new method. DTP tests were performed at cementedsoil sites to verify the cementedsoil identification (T2/T1, [Tip 15 PAGE 16 1/Tip 2], and Friction Ratio) from DTP. The ratio T2/T1 was finally determined to be an excellent identifier in locating cemented sand. My new method provides better LRFD resistance factors for both Florida soil and Louisiana soils. It could be a promising method to improve pile design in the future. From the DTP tests in cemented soils, it was concluded that the DTP could be an efficient tool to identify cemented sand and thereby better predict pile capacity. 16 PAGE 17 CHAPTER 1 INTRODUCTION From the 1960s to the 1980s, FDOT sponsored research at the University of Florida to evaluate the methods used for calculating static pile capacity based on the CPT (cone penetration test). After years of research, Dr. John H. Schmertmann (1978) proposed the method which was later named after him. This method was put into the FDOT pile design software and termed, PLAID. This method has been used successfully in the districts of Florida on 16 to 18 precastprestressedconcrete (PPC) piles. However, during the last two decades, the size of the pile has increased to 24 to 30, primarily due to higher strength concretes and steel as well as larger pile driving equipment. The Schmertmann method became conservative in evaluating pile capacity based on the comparison between the predictions and static load test results. Compare to drilled shaft, driven piles have their own advantages. There are significant reductions of lateral capacity for drilled shaft due to torsional loading (Hu et al. 2006). Since driven piles are usually in group and there will be no reduction of lateral capacity due to torsional loading. That is the reason why more and more CPT based pile capacity prediction methods have been proposed. During the 1980s, many methods were developed around the world. These included the Aoki and de Alencar method (1975), Penpile method (Clisby, M.B., et al. 1978), de Ruiter and Beringen method (1979), Philipponnat method (1980), Bustamante and Gianeeselli method (LCPC) (1982), Price and Wardle method (1982), and Tumay and Fakhroo method (1982). Most of these methods were generated by matching the CPT and load test database in the local area. None of the methods have been evaluated in Florida soils. The Louisiana Transportation Research Center evaluated these methodologies in predicting the axial ultimate capacity of square PPC piles driven into Louisiana soils. Based on the result, the de Ruiter and Beringen and Bustamante and Gianeselli (LCPC) methods showed the best 17 PAGE 18 performance in predicting the pile capacity. However, that may or may not be the case in Florida. From 1990 until the present, many new methods have been proposed. They include the Almeida et al. method (1996), Jardine and Chow method (1996), Eslami and Fellenius method (1997), Powell et al. method (2001), and UWA05 method (Lehane, B.M., et al. 2005). Most of these methods take into account the pore pressure from CPTU to improve their accuracy. The Zhou et al. method (1982) was proposed in 1982 using load test data and CPT performed in the eastern China. However, it had not been evaluated in other areas outside of this region. Again, none of these methods have been evaluated in Florida. Details of these methods are given in the following chapters. There are a total of 21 cases (load test data with CPT close to them) in Florida and 28 in Louisiana. All previous methods were evaluated by using these cases. The LRFD resistance factor for each method was calculated and compared. One of the most accurate and simplest methods, the Philipponnat method, was chosen and modified to form the proposed UF method. Both Florida and Louisiana soil data were used to validation. One of the more challenging soil types in Florida is cemented sand. Cementation in sands improves strength, but the strength increase depends on the degree of cementation. The degree of the bond strength in the cemented sands should be considered when designing foundations on or in cemented sands. Since this material cannot be identified by a CPT test, none of the methods mentioned above had taken into account the cementation issue. This means they may overestimate the pile capacity, which may produce a serious design flaw. Recently, the FDOT funded the University of Florida to develop a new cone penetrometer, the dual tip penetrometer (DTP), to be able to identify cemented sands. UFs new proposed design method takes into 18 PAGE 19 account the cementation issue and results in the highest R (the ratio of resistance factor to the mean bias 0.62 for Florida soil and 0.67 for Louisiana soil) among all the methods as well has the lowest coefficient of variation (0.27 for Florida soil and 0.23 for Louisiana soil). 19 PAGE 20 CHAPTER 2 LITERATURE REVIEW Cemented Sand The term cemented sand is a general term used for a wide variety of soils. R.W. King (Lunne et al. 1997) proposed a classification system for a variety of cemented carbonate soils. One of the main problems of this system is that degree of cementation is only a function of penetrometer resistance q c and does not take into account the relative density of sand. For example, noncemented dense sand may have cone resistance q c higher than 10 MN/m 2 with no cementation. It will be very useful for design engineers, if better identifiers or parameters for cemented sand can be found. Cemented sands exist in many areas of the United States, including California, Texas, Florida, and along the banks of the lower Mississippi River. They also exist in Norway, Australia, Canada, and Italy (Puppala et al. 1995). Calcareous cemented sands are a feature of warm water seas mainly due to the sedimentation of the skeletal remains of marine organisms (Lunne et al. 1997). Cemented sand as the name implies, is a cohesionless material in which a calciumcarbonate chemical bond develops to some extent. This chemical bond is the result of the deposition of calcium at the particletoparticle contacts and the chemical reaction between calcium and sand over time. The strength of these chemical bonds depends on the degree of cementation as well as the distribution. This kind of cementation leads to a significant increase in modulus (Briaud). Ahmadi et al. used computer modeling for CPT penetrating process and found that the modulus of sand is the key factor in CPT tip resistance (Ahmadi et al. 2005). This is why cementation tends to increase tip resistance. This phenomenon was also found by Puppala et al (Puppala et al. 1995). 20 PAGE 21 From a mechanical point of view, cemented sand belongs to an intermediate class of geomaterials placed between classical soil mechanics and rock mechanics. Often, no physical or mathematical models are able to integrate this kind of material in a consistent and unified framework (Gens and Nova 1993). During loading, cemented sand shows a very stiff behavior before yielding, which is governed by cementation. After stress reaches yielding stress, it suddenly changes into a ductile material. Leroueil and Vaughan (1990) discovered that the structure of chemical bonds and its effects on soil behavior is a very important factor in determining the soil stressstrain behavior as well as other factors, such as the relative density, over consolidation ratio etc. However, the structure of chemical bonds is an unpredictable and very difficult to identified, let alone quantify. An understanding of the effect of a low degree of cementation on the sands strength is increasingly important in geotechnical engineering design and analysis. In the current design procedure, the effect of cementation is often neglected because cementation often improves the strength. However, preliminary studies indicate that light cementation increases the tip and friction resistances while decreasing the friction ratio of CPT (Rad and Tumay 1986). This could be explained by the bonds increasing the resistance during the penetration. However, the bonds tend to break during pile driving, and the CPT could not sense this reduction of strength. The degree of cementation in sands can be an issue for geotechnical engineers. For well cemented sands, the strength can be so high that engineers will neglect the cementation issue. However, in these cases, even though the CPT could not totally break the chemical bond within the sand particles, large diameter driven piles may indeed do so, resulting in a much lower pile capacity than that predicted by the CPT results. For lightly cemented sand, a breakdown of cohesion bonds can occur from a disturbance such as an earthquake, cone penetrating, or pile 21 PAGE 22 driving. One such example is when the Loma Prieta San Francisco earthquake caused slope failures along the cemented northern Daly City bluffs (Puppala et al. 1995). Other similar slope failures have occurred due to earthquakes and heavy rains (Rad and Tumay, 1986). If the sands tested with a CPT test are not known to be cemented, the high bearing readings may be misinterpreted as being due to high relative densities. This can lead to an underestimation of the liquefaction potential of the soil and an overestimation of the ultimate pile capacity. None of the CPT prediction methods evaluated to date take into account the reduction of strength in cemented sand. The concern was proved legitimate by the CPT and DTP testing performed at Port Orange Relief Bridge in Port Orange Florida. A comparison between the predictions using the prediction methods and the load test result shows that most of the methods overpredict the pile capacity (some by over 100%). Researchers have performed laboratory tests on cemented sands obtained in the field. Both Clough et al. (1981) and Puppala et al. (1998) have tested naturally cemented sands in triaxial tests and unconfined compressive strength tests. The cemented sands were obtained by trimming samples using an SPT split spoon sample. The retrieval of undisturbed, lightly cemented sands was quite difficult since the bonds tended to break under light finger pressure. Due to the difficulty in sampling, insitu testing has become a more popular method of testing naturally cemented sands. The CPT test is a popular device for testing the cemented sands. The CPT test has been used to test both naturally occurring cemented sands in the field (Puppala et al., 1998) and artificially cemented sands in calibration chambers (Rad & Tumay, 1986; and Puppala et al. 1995). In both the 1985 and 1996 calibration chamber studies, Monterrey No. 0/30 sand was cemented with 1% and 2% Portland cement. An attempt was made to relate the tip bearing and friction sleeve values to the sand properties, including cement 22 PAGE 23 content, relative density, confining stress and friction angle. Puppala (1995) did this by using the bearing capacity equations of Durgunoglu & Mitchell (1975) and Janbu & Senneset (1974). To include the effect of cementation or cohesion on tip bearing, the other parameters that affect the tip bearing, mainly relative density and confining stress, needed to be known and included in the equations proposed by Puppala. Even though the calibration chamber study is timeefficient and makes it easy to control the cementation ratio, there are several drawbacks; first, the cementation structure in the nature is almost impossible to simulate in the chamber and, as is discussed above, this is a very important issue to determine a soils strength. Secondly, the stress state in the field is not the same as those in the calibration chamber, especially for deeply occurring cemented soil, due to size limitation of the chamber. Therefore, the literature indicates that the best approach in dealing with this issue is to test materials insitu (e.g., CPT test) and somehow identify when cementation is present. Since the main problem of cemented sands is providing cohesion on tip bearing resistance, if its effect could be removed from the tip bearing resistance it may be possible to obtain more accurate bearing capacity predictions. One way to accomplish this would be to design an insitu device that could measure the bearing strength of the cemented sands both before and after the cohesive bonds have been broken up. This is the rationale that led to the development of the dual tip penetrometer developed at the University of Florida. Since it is simply an enhanced CPT, a brief history of this versatile instrument is provided below. Cone Penetration Test The cone penetration test is considered one of the most costeffective and reliable method for soil classification. The CPT (Figure 21) test pushes a cone into the soil at a constant rate by means of cylindrical rods that are connected in series with the cone located at the base of the string of rods. During the test, the sleeve friction and tip resistance are measured and recorded. 23 PAGE 24 These two parameters are used to classify soil and to estimate strength and deformation characteristics of soils. In 1917, the Swedish railways introduced the CPT. Ten years later, Danish railways started to use CPT. The first apparatus was simply a cone and a string of outer rods. In 1936, the Dutch Mantle cone was introduced. This cone has an area of 10 cm 2 and an apex angle of 60, which is similar to the currently ones in use. But the cone was pushed by hand and there was a limitation on the capacity and penetration depth. In addition, it could not penetrate very dense sand or cemented soils. In the 1940s and early 1950s, hydraulic jacks were introduced that allowed for much more reactive force being applied, thereby increasing penetration depths. This advancement dramatically increased CPT usage. In 1948, the first electric cone penetrometer was developed. Strain gages were used to measure the soil resistance, which increased its accuracy dramatically, since the bridge circuit made it more sensitive to small changes in soil resistance. The most important feature of electric CPT is that it can provide a continuous reading of a soils resistance during the test (typically logged every 5 cm). This provides a wealth of subsurface information for geotechnical engineers. One of the most important improvements of the CPT was made in 1953. Begemann proposed the use of a separate sleeve located just behind the tip that allows the penetrometer to measure both tip resistance q c and sleeve friction resistance f s The friction sleeve has an area of 150 cm 2 and was used in conjunction with the traditional Begemann mechanical cone in the late 1950s. In 1968, an electric cone penetrometer with the friction sleeve was developed in Australia. 24 PAGE 25 The first ASTM standard (ASTM D344175T) for the cone penetrometer was published in 1975. In 1979 and 1986, ASTM D344179 and ASTM D344186 were published to revise the previous standard. In 1988, an international reference test procedure was developed by the International Society of Soil Mechanics and Foundation Engineering. Currently, there are two diameters for the cone: 1.41 in (10 cm 2 cross section) and 1.71 in (15 cm 2 cross section) with both having a 60 angle. The first one is the most commonly used. CPT Based Pile Capacity Prediction Methods Using CPT data for design is considered one of the most promising methods to predict the pile capacity for the following reasons: 1. The shape of a cone penetrometer is very similar to a cylindrical driven pile except at the bottom. However, during the ultimate failure of a pile, the soil under the pile tip is densified and forms a coneshaped failure envelop similar to the cone penetrometers 60 tip. 2. The soil state during penetration is comparable to that during pile driving. 3. The testing process is quasistatic, which is more representative of a static load test compared to other insitu tests. 4. Because the cone penetrometer actually penetrates the soil, causing an ultimate failure (punching failure) condition, it should be possible to predict the ultimate failure of the pile including ultimate skin friction and ultimate tip resistance. These two predictions can also be useful during pile driving in order to prevent damage during the driving process. 5. The speed of conducting a test allows for more CPT soundings at a particular site and coupled with load test data make it possible to generate improved pile capacity prediction methods. There are also issues involved in the prediction of pile capacity using CPT which have to be solved by empirical correlations: 1. The scale effect caused by the difference between the diameter of the penetrometer and that of the piles. This will influence the soil densification. The larger the diameter, the more densification the soil can achieve. It will also influence the size of the stress 25 PAGE 26 ball which will influence the zone of resistance soil near the pile tip. If the soil is not uniformly distributed and the soil layer is not horizontal, one CPT may not be able to representative to the pile particularly large diameter piles. 2. The CPT can not be used to identify cemented soils. It tends to misidentify the material as simply a denser soil state, due to the high q c values. 3. When a CPT is performed in saturated clayey soils, especially with low permeability, high excess pore pressure will be generated during penetration and will cause higher q c value. However, when a pile is driven into the same soil, much higher excess pore pressure will be generated and will dissipate slowly depending on the permeability. In order to propose a better method to predict pile capacities, many existing methods have been investigated. An extensive literature research was conducted, specifically looking for axial pile prediction methods based on CPT cone soundings. The following methods were identified as those used by a number of DOTs, consultants or contractors: the Schmertmann method, the de Ruiter and Beringen method, the Penpile method, the Price and Wardle method, the Tumay and Fakhroo method, the Aoki and De Alencar method, the Philipponnat method, and the LCPC (Bustamante and Gianeselli) method. Most of the above methods were developed in 1980s. From 1990 until now, many new methods have been proposed. They are the Almeida et al. method, the MTD (Jardine and Chow) method, the Eslami and Fellenius method, the Powell et al. method, and the UWA05 method. The Zhou et al. method was proposed in 1982 using load test data and CPT performed in eastern China. A discussion of each of the methods is presented in the following section of this chapter. Schmertmann Method This method was first proposed by Schmertmann in 1978. It uses both tip resistance and sleeve friction to predict the pile capacity. The piles unit tip capacity is calculated by the minimum path rule. Schmertmann set an upper limit of 150 tsf for the unit tip capacity. The piles unit skin friction: 26 PAGE 27 In clay: f1.2 scfsatsf where: c is a function of f sa In sand: where: S is a function of pile depth to width ratio. De Ruiter and Beringen Method This method is proposed by de Ruiter and Beringen from their study of the soil near the North Sea. It uses both tip resistance and sleeve friction to predict the pile capacity. The piles unit tip capacity: In clay: where: N C = 9, constant, bearing capacity factor; q c (tip) is the average cone tip resistance around the pile tip similar to Schmertmann method (minimum path rule); N k = 15~20, constant, cone factor, (20 was used in the current study since it yielded better results). In sand: The calculation of tip capacity is simular to Schmertmann method. The piles unit skin friction: In clay: where: constant, adhesion factor, 1 for N.C., 0.5 for O.C., (1 was used in the current study). q c (side) is the average cone tip resistance within the calculated layer along the pile. Sutip()qctip()Nk q tNcSutip() qtqc1qc2 2 150ts f Suside()qcside()Nk fsSuside() Qy ss08Dy8D fsaAs8DLysaAs f fs1.2tsf 27 PAGE 28 In sand: where: f sa is the average sleeve friction within the calculated layer along the pile. fsminfsaqcside()300 compression()qcside()400 tension()1.2tsf Penpile Method This method was invented by Clisby et al. for the Mississippi Department of Transportation. It uses both cone tip resistance and sleeve friction to predict the piles axial capacity. The piles unit tip capacity: In clay: qt0.25qc a In sand: qt0.125 qc a where: q ca : the average of three cone tip resistances close to the pile tip. The piles unit skin friction: where: f sa : the average sleeve friction within the calculated layer along the pile. fs fsa1.50.1fsa f s f sa are expressed in psi (lb/in 2 ). Prince and Wardle Method This method uses both the CPT tip resistance, q c and sleeve friction, f s to predict the axial pile capacity. The piles unit tip capacity: qtkbqcatip() 150ts f fsksfsa 1.2ts f The piles unit skin friction: where: k b and k S are factors that depend on pile type; k b = 0.35 for driven pile, 0.3 for jacked pile; k S = 0.53 for driven pile, 0.62 for jacked pile, and 0.49 for drilled shaft; 28 PAGE 29 q ca (tip) is the average CPT tip resistance within 4D below and 8D above the pile tip (there is no reference about the influence zone, therefore for better results, 4D below and 8D above were chosen). Tumay and Fakhroo Method This method was proposed by Tumay and Fakhroo for estimating pile capacity in clayey soil. In order to see how this method performed for Florida soil, it was also evaluated. It uses both tip resistance and sleeve friction to predict the pile capacity. The piles unit tip capacity: The calculation is similar to Schmertmann method except leting y equal to 4. qtqc1qc2 2 150ts f The piles unit skin friction: f smfsa 0.72ts f m0.59.5e9fs a where: f sa is the average sleeve friction within the calculated layer along the pile with the unit tsf (ton/ft 2 ). Aoki and De Alencar Method This method only uses the CPT tip resistance to predict the pile capacity. The piles unit tip capacity: qqca ttip()Fb 150ts f The piles unit skin friction: where: F b F s are empirical factors that depend on pile type, S is a function of soil type. q ca (tip) is the average CPT tip resistance within 4D below and 8D above the pile tip. fsqca side()sFs 1.2ts f 29 PAGE 30 Philipponnat Method This is another method which uses tip resistance, q c to predict the axial pile capacity. The piles unit tip capacity: qtkbca qtip( ) The piles unit skin friction: side()sFs 1.27ts f fsqca where: q ca (tip) is the average tip resistance of 3D below and 3D above the pile tip; k b and F S are functions of soil type; s is determined by pile type, = 1.25 for precast prestressed concrete piles. LCPC (Bustamante and Gianeselli) Method This method only uses cone tip resistance for predicting axial pile capacity. It was proposed by Bustamante and Gianeselli for the French Highway Department after the study of 197 piles in Europe. It is also called the French method. The piles unit tip capacity: qtkbeq qtip() where: q eq (tip) is the average of tip resistance within 1.5 D above and 1.5 D below the pile tip after eliminating abnormal data (out of the range of 30% of average value); k b is a function of soil and pile type. The piles unit skin friction is a function of pile type, soil type and cone tip resistance. Almeida et al. Method This method was proposed by Almeida et. al. based on the analysis of 43 load tests on driven and jacked piles in clay in Norway and Britain. Most of the load tests were performed in 30 PAGE 31 tension and only 4 tests in compression. The parameters used in this prediction method are penetrometer tip resistance and overburden stress. The piles unit skin friction: k111.814.0logqc00' f sqc0k1 where: q c is CPT tip resistance, with pore pressure correction for piezocones, v0 is the total overburden stress, v0 is the effective overburden stress. In order to calculate the effective overburden stress, hydrostatic pressure has been used. A reduction in k 1 needs to be applied if L/D >60. The reduction factor is recommended by both Semple and Rigden (1948) and is included in the procedure suggested by Randolph and Murphy (1985). qtqc0k2 The piles unit tip capacity: where: k 2 is a function of both pile type and material (k 2 = 2.7 for driven pile = 1.5 for jacked pile in soft clay, and =3.4 for jacked pile in stiff clay). In order to prevent a nonrealistic result in sand, a limitation of highest unit skin friction is set at 1.2 tsf. MTD (Jardine and Chow) Method The method proposed by Jardine and Chow is from intensive field tests using 4 inch (102 mm) diameter, closedended instrumented piles at two sand sites in France. In addition, data acquired from field tests on highquality instrumented displacement piles in a large range of clay 31 PAGE 32 soils performed by MIT, Oxford University, NGI and Imperial College over 15 years was utilized. The piles unit tip capacity: In clay: q tkqc a where: k = 0.8 for drained loading, =1.3 for undrained loading. q ca is the average cone tip resistance within 1.5D above and 1.5D below the pile tip. qt10.5logDDCPT qc a In sand: where: D is the diameter of the pile; D CPT is the diameter of cone penetrometer which is 1.4 inch (36 mm). q t has the lower bound value of 0.13* q ca when D is greater than 6.56 ft (2 m). The piles unit skin friction: In clay: fsfL Kc0'tanf Kc2.20.016YSR0.870logStYSR0.42hR 0.20 where: f L : loading coefficient, = 0.8; K c : earth pressure coeffient after equilization; YSR: yield stress ratio (yield stress determined in an oedometer test divided by the vertical effective stress). In case the YSR is not available, Lehane et.al (2000) provides the following relationship between YSR and cone tip resistance; YSR0.04427qc0' 1.66 7 32 PAGE 33 S t : clay sensitivity, and in case the S t is not available, Robertson and Campanella (1983) proposed the relationship between S t and friction ratio; St10Rf%() h: the hight above the pile tip. In order to prevent too large a K C value, h/R 8; R: the diameter of the pile; v0 : effective overburden pressure; f : pilesoil interface friction angle at the maxmum shear stress. Because the large variations are possible, it is recommended by Lehane et.al (2000) to use a ring shear test to obtain the direct measurement; in case direct measurement is not available, Jardine and Chow proposed a relationship between and clay plasticity index for steel pile. f will be between peak ( peak ) and ultimate ( ultimate ) depending on relative displacement between pile and soil. If the PI of the deposit is not available for determining f for clay, Schmertmann (1978a) suggests assuming an average normally consolidated ratio of 0.33 for most postpleistocene clay which is corresponding to the PI equal to 0.59. From the relationship between PI and tan( f ), 0.2 was determined to be tan( f ). In sand: For compression pile fsrf'tanf rfrcrd For tension pile ''' rf0.8rcrd ''' 33 PAGE 34 rc'0.029qc0'Pa 0.13hR 0.3 8 rd'4GRclaR Gqc0.02030.00125qcPa0' 1.216106qc2Pa0' 1 where: rf : radial effective stress at maximum shear stress; rc : radial effective stress on side after equalization; rd : the net dilatant component; Pa: the atmosphere pressure; G: the sand shear modulus; R cla : the piles centerlineaverage roughness. It is qual to 10 5 for steel pile, 10 4 for very rough casing of concrete pile and 3*10 5 for prestressed concrete pile; f : pilesoil interface friction angle at the maxmum shear stress. It is recommended by Jardine and Chow (1996) to use an interfacedirect or a ringshear test with the same roughness and hardness as the pile material and same effective normal stress as the field; in case direct measurement is not possible, Jardine and Chow recommended to use the relationship between cv (critical state interface friction angle) and sand mean particle size (d 50 ) for steel pile and assume f is equal to cv From correspondence with the authors, it was found that they are currently conducting sets of 34 PAGE 35 interface shear tests on sands sheared against concrete but have not finished yet. For now, it is recommended to assume f between concrete pile and sand is not so different from f between steel pile and sand. Since D50 of Florida soil is somewhat between 0.1 mm and 0.3 mm, it was decided to use 30 as f Eslami and Fellenius Method This method was proposed by Eslami and Fellenius from the study of 102 cases around the world. This is the method that uses cone tip resistance (q c ) and pore pressure (u) to predict the axial pile capacity. CPT sleeve friction is only used to identify the soil type. q t = q eg The piles unit tip capacity: where: q eg is the geometric average of the effective cone resistance. The effective cone resistance is calculated by subtracting the hydrostatic pressure from the cone resistance if pore pressure data is not available. The influence zone proposed by the Eslami and Fellenius is as follows: 2D above and 4D below the pile tip when the pile is installed through a dense soil into a weak soil. 8D above and 4D below the pile tip when the pile is installed through a weak soil into a dense soil. The piles unit skin friction: f s = cs q e where: C S is functions of soil type. q e is average of the effective cone resistance within the calculation layer. 35 PAGE 36 The effective cone resistance is calculated by subtracting the hydrostatic pressure from the cone resistance if pore pressure data is not available. Powell et al. Method This method was proposed by Powell et al. from the study of 63 steel driven or jacked piles. The soil condition ranged from soft normalconsolidated clay to stiff overconsolidated clay and two sand sites. The parameters used by this method are cone tip resistance (q c ) and pore pressure (u), undrained shear strength(s u ), and a soil profile to predict the axial pile capacity. The piles unit skin friction: k110.513.3logqc00' fsqc0k1 where: q c is CPT tip resistance, with pore pressure correction for piezocones, v0 is the total overburden stress, v0 is the effective overburden stress. In order to calculate the effective overburden stress, hydrostatic pressure has been used. A reduction in k 1 needs to be applied if L/D >60. The reduction factor is recommended by both Semple and Rigden (1948) and is included in the procedure suggested by Randolph and Murphy (1985). The piles unit tip capacity: k2Nkt9 q tqc0k2 where: N kt is cone factor, range from 10 to 20 based on local experience. 15 was used in current study. 36 PAGE 37 UWA05 Method This method was proposed by Lehane et al. in 2005. This method is especially used to predict pile ultimate capacity in sand. qtqca0.150.451Dint2D2 The piles unit tip capacity: where: q ca is calculated by minimum path rule. D int is the internal diameter for pipe pile, D is the outer diameter of the pile. The piles unit skin friction: fs0.03qc1Dint2D2 0.3maxhD 2 0.54GrD tancv Gqc185qcPa 'v0Pa 0.7 where: q c : average CPT tip resistance within the calculated soil layer; Pa: the atmosphere pressure; G: the sand shear modulus; h: the hight above the pile tip. In order to prevent too large a K C value, h/R 8; v0 : effective overburden pressure; r: interface dilation, 0.02 mm was used by current study; cv : constant volume interface friction angle between sand and pile. 37 PAGE 38 Since the large variations are possible, it is recommended to use ring shear test to obtain the direct measurement; In the absance of lab tests, The trends between cv and D50 recommended by ICP05 with the upper limit 0.55 for the tan( cv ) are considered reasonable. Since D50 of Florida soil is somewhat between 0.1 mm and 0.3 mm, it was decided to use 29 as cv which will give the tan( cv ) 0.55. Zhou et al. Method This method was proposed by Zhou et al. after the study of 96 precast driven concrete piles in several eastern Chinese provinces. It provides a satisfactory predictions (80% of the predicted errors are within 20% of the true load test results). The soil condition ranged from sand to clayey soil. The parameters used by this method are cone tip resistance (q c ) and sleeve friction (f s ) to predict the axial pile capacity. One of the interesting points about this method is that it predicts limit load capacity in stead of ultimate load capacity. The limit load is defined as the load near the starting point of the straight line portion on the load test curve, the point where the shaft resistance of pile would be fully mobilized, while the end resistance only partially mobilized. If the point is not obvious form the data, they recommend using the load at a relative settlement of 0.4 0.5. The piles unit tip capacity: where: q ca is average CPT tip resistance within 4D above and 4D below the pile qtqc a tip; is the function of soil type and q ca use the following equations to calculate value; 0.71qca0.2 5 1.07qca0.3 5 Soil Type I: Soil Type II: 38 PAGE 39 Soil type is defined as: Soil Type I: q ca > 2 MPa and f sa / q ca < 0.014 Soil Type II: other than Soil Type I. The piles unit skin friction: f sfsa where: f sa is average CPT sleeve friction along the calculated soil layer; is the function of soil type and f sa use the following equations to calculate value. 0.23fsa0.4 5 Soil Type I: 0.22fsa0.5 5 Soil Type II: 39 PAGE 40 Figure 21. Regular cone penetrometer 40 PAGE 41 CHAPTER 3 MATERIALS AND METHODS Dual Tip Penetrometer (DTP) The DTP is the latest version of a series of devices developed at the University of Florida intended for identifying cemented sands. Daniel Hart developed the first version of the device in 1996 while he was a graduate student at the University of Florida. The device at that time had a lip welded onto the top of the friction sleeve. However, this meant that the bearing reading measured by the lip was added to the frictional component measured by the friction sleeve strain gauge. The lip also made it difficult to remove the cone from the ground. In 1998, Randell Hand eliminated the welded lip and welded a bearing annulus onto the friction reducer coupler. The annulus was therefore located about 20 inches above the top of the cones friction sleeve. Strain gages were used to measure resistance in the annulus. The voltage output was translated into a second q c (tip resistance) reading. Steve Kiser and Hogentogler & Co. Inc. improved on this design and came up with the dual tip penetrometer in 1999. In 2004, Hogentogler & Co. Inc. converted the DTP from analog into digital cone, which is the latest version of DTP equipped in the cone truck at the State Material Office. The DTP is similar to conventional cone penetrometers except for a second tip (actually an annulus) just above the friction sleeve as shown in Figure 31. The second tip has the same angle (60) and bearing area (10 cm 2 ) as the regular cone. The first tip was originally designed to break down the cohesive bonds of cemented sand while the second tip was meant to measure the residual or brokenup bearing resistance. Based on the relationship between Tip 1 and Tip 2 along with the soil profile, through a large number of experiments, cemented sand identifiers would be identified. 41 PAGE 42 Locate Cemented Sites Cemented sands exist in many areas of the United States, including California, Texas, Florida, and along the banks of the lower Mississippi River. Lightly cemented sands are usually misidentified in the CPT test, which can cause design problems. SPT borings log are one resource used to identify cemented sand. However, if the cementation is very weak and can be easily broken by finger pressure, it might not be noticed by the field technicians. In order to incorporate cemented sands into the new proposed design method, strongly cemented and lightly cemented sand sites were identified using two databases, FDOT and UF. There are hundreds of insitu and load test data in these two databases. Both SPT data and boring logs were searched to identify cemented sand sites. CPT data were also searched and combining with SPT N (Qc/N) to locate cemented sand sites. All previous projects reports were reviewed to find soil and load test data. In case of sites where load test and SPT data were available, but no CPT data, the sites were flagged for future CPT and DTP testing. There were a total of 21 cases where load test, SPT and CPT (DTP in some cases) are all available. Figure 32 shows the relative locations of these 21. These cases were used to calibrate the ultimate pile capacity prediction methods and their corresponding LRFD resistance factors Axial Ultimate Pile Capacity Prediction Methods A total of 14 prediction methods have been analyzed in my research. They are: the Schmertmann method, de Ruiter and Beringen method, Penpile method, Price and Wardle method, Tumay and Fakhroo method, Aoki and De Alencar method, Philipponnat method, LCPC (Bustamante and Gianeselli) method, Almeida et al. method, MTD (Jardine and Chow) method, Eslami and Fellenius method, Powell et al. method, UWA05 method, and Zhou et al. method. Because most of the methods involve complicated calculation and digital CPT data 42 PAGE 43 make these calculations time intensive, a MathCAD program was used to calculate each of predictions. For each method there is one MathCAD program that was presented in the appendix. Figure 33 shows the program for the Philipponnat method. The colored fields are inputs. The CPT data input is an Excel file formatted with four columns (depth, tip resistance, sleeve friction, and pore pressure). Other inputs are diameter or edge length of pile, embedment length, layer depths. After all required fields have been inputted, the program will calculate ultimate tip resistance, ultimate skin friction and Davisson capacity (1/3 of ultimate tip resistance + ultimate skin friction). Load and Resistance Factor Design (LRFD) Over the past two decades, Load and Resistance Factor Design (LRFD) have been incorporated in structural and geotechnical designs. One of the benefits of LRFD is its consistent reliability in design practice. Many state DOTs, including FDOT, are now implementing AASHTO (American Association of State Highway and Transportation Officials) LRFD Specifications. Hence, the object of my research is to update FDOTs pile/shaft design procedure based on CPT and DPT data and access the LRFD resistance factor for each pile capacity prediction method. Even though LRFD requires both load and resistance factors, the resistance factor is considered as a variable for each static pile capacity prediction method whereas load factors are typically constants, based on local experience. Modified First Order Second Moment (FOSM) Approach The LRFD approach used in my research is the modified FOSM (First Order Second Moment Approach). The modification was developed at the University of Florida (Styler, 2006) due to the difference between FORM (First Order Reliability method) and FOSM resistance factor in NCHRP Report 507. The modified portion in FOSM is the term COV (Q). The 43 PAGE 44 previous FORM assumes COV (Q) = COV (Q D ) + COV (Q L ). It was found that this equation was incorrect and a modified COV (Q) was formulated as: COVQ()QD2QL2 QD2COVQD2QL2COVQL2QD2QL2 QD22QDQL QDQLQL2 where: Q D /Q L = Dead to live load ratio, varies from 1.0 to 3.0 (spans, L = 57170 ft. Since it is not very sensitive, a value of 2.0 used herein), QD QL = Dead load and live load bias factors, QD = 1.08, QL = 1.15 (recommended by AASHTO 1996/2000), COV QD COV QL = Dead load and live load coefficients of variation, COV QD = 0.128, COV QL = 0.18 (recommended by AASHTO). Based on this modified FOSM, it was concluded that the difference between FORM and FOSM was little if any. Since FORM involves complicated calculation process, the modified FOSM is the approach used in my research. The following section provides a detailed deviation of LRFD resistance factor using modified FOSM approach. Limit state equation In this approach, load and resistance are assumed to be lognormal distribution. Therefore, the limit state equation is as follows: 44 PAGE 45 GlnR()lnQ() EG()ElnR()()ElnQ()() ElnR()()lnER()()12 ln1COVR()()2 ElnQ()()lnEQ()()12 ln1COVQ()()2 EG()lnER()1COVQ()()2 EQ()1COVR()()2 It is assumed that R and Q are statistically independent, therefore G : GVARlnR()()VARlnQ()() Gln1COVR()()2 1COVQ()()2 where: COV(R), COV(Q) = Resistance, load coefficient of variation. Reliability index EG()G lnER()1COVQ()()2 EQ()1COVR()()2 ln1COVR()()2 1COVQ()()2 ER()EQ()expln1COVR()()2 1COVQ()()2 1COVQ()()21COVR()()2 Resistance factor The LRFD equation: R n i Q i 45 PAGE 46 where: = Resistance factor R n = Nominal Resistance = Load modifier for importance, redundancy and ductility i = Load factors Q i = Force effects Load modifier is set equal to 1 which leaves all uncertainty on the resistance factor For driven pile design, two load effects, dead load Q D and live load D L are considered. Therefore, D and L are considered as load factors for dead load and live load, respectively. The LRFD equation becomes: R n D Q D + L Q L DQDLQLRn ER()RR n Since: where: R = Mean bias (mean of resistance bias factor); R1NiRiN RiRm i Rni where: R mi = Measured capacity from load test data R ni = Predicted capacity form CPT data N = The number of cases DQDLQLER()R 46 PAGE 47 By inserting E(R) into the above equation, the follow equation can be derived: RDQDLQL1COVQ()()21COVR()()2 EQ()expln1COVR()()2 1COVQ()()2 Since the dead load and live load are considered as statistically independent, E (Q) can be expressed as follows: QDQDLQ L EQ()QDQDQLQ L where: QD QL = Dead load and live load bias factors; QD =1.08, QL = 1.15 (recommended by AASHTO 1996/2000) By inserting E(Q) into the equation for resistance factor, and divide both numerator and denominator by Q L the follow equation results: RDQDQL L 1COVQ()()21COVR()()2 QDQDQL QL expln1COVR()()2 1COVQ()()2 This equation was traditionally used to calibrate the resistance factor using FOSM in AASHTOs specifications by assuming COV (Q) 2 = COV (QD) 2 + COV (QL) 2 However, as mentioned previously, this assumption is incorrect. The correct derivation is as follows: VARQ()QD2VARD QL2VARL QDDQD QLLQL 47 PAGE 48 COVQD()2VARD QD2QD2QD2 COVQL()2VARL QL2QL2QL2 VARLQL2COVQL()2 VARDQD2COVQD()2 COVQ()2VARQ()EQ()2 VARQ()QD2QD2COVQD()2QL2QL2COVQL() 2 COVQ()2QD2QL2 QD2COVQD()2QL2COVQL()2QD2QL2 QD22QDQL QDQLQL2 By inserting COV(Q) into the resistance equation, the following equation can be obtained: RDQDQL L 1QD2QL2 QD2COVQD()2QL2COVQL()2QD2QL2 QD22QDQL QDQLQL2 1COVR()()2 QDQDQL QL expln1COVR()()2 1QD2QL2 QD2COVQD()2QL2COVQL()2QD2QL2 QD22QDQL QDQLQL2 48 PAGE 49 where: D = Dead load factor (1.25, recommended by AASHTO (1996/20001)), L = Live load factor (1.75, recommended by AASHTO 1996/2000), Q D /Q L = Dead to live load ratio, varies from 1.0 to 3.0 (spans, L = 57170 ft, is not very sensitive, a value of 2.0 used herein) R = Mean of resistance bias factor, COV (R) = Resistance coefficient of variation QD QL = Dead load and live load bias factors, QD = 1.08, QL = 1.15 (recommended by AASHTO 1996/2000), COV QD COV QL = Dead load and live load coefficients of variation, COV QD = 0.128, COV QL = 0.18 (recommended by AASHTO) T = Target reliability index, AASHTO and FHWA recommend values from 2.0 to 3. Of major importance in estimating LRFD resistance factor, are the resistances bias factor ( R ), and the resistances coefficient of variation (COV(R)). Both are computed from: 1) the nominal predicted resistance, R ni (predicted pile capacity), and 2) the measured pile capacity, R mi i.e. load test results. Based on the ratio of measured to predicted pile capacities, Ri for each of the sites, the mean, R standard deviation, R and coefficient of variation, COV(R), were determined for all 14 CPT methods. Using these inputs, the LRFD resistance factors, were determined with a reliability index, of 2.5 for each method. Because driven piles are usually in groups and redundancy will require lower reliability, 2.5 was used in my research. Differentiate Ultimate Skin Friction and Tip Resistance from Load Test Data All of the load tests were conventional topdown static load tests performed for which loadsettlement data exist (see Figure 34.). Using FDOTs load testing protocol (i.e. 49 PAGE 50 Specification Section 455), the Davisson Capacity was assessed for each load versus settlement curve. The Davisson Capacity, referred to as the measured capacity, R m generally occurs for settlements that are tolerable (i.e. less than 1), under service loading conditions. These values were subsequently used with the predicted capacities, R n (where R n = Q S ult + 1/3 Q T ult ) to assess the LRFD resistance factor, for each of the fourteen CPT methods. However, the load versus settlement curve does not provide the ultimate skin friction and tip resistance. In order to evaluate the accuracy of each prediction method for ultimate skin friction and tip resistance separately, each needs to be estimated using this load test curve. Figure 35 shows the proposed process that was developed to separate the two attributes. For the majority of the tests, the piles were not loaded sufficiently to induce a plunging failure. While the ultimate skin friction is likely fully mobilized at small displacements (0.1), the tip resistance is not. Thus, it is not possible to determine the ultimate tip resistance (i.e., the tests ended after one inch displacement was reached, whereas two inches of displacement or approximately D/10 [D is in inches] is typically needed to induce a plunging failure) unless the load test curve is extrapolated to the two inch value. The idea of extending the load test curve follows from deBeers method for determining pile capacity. The proposed differentiate method is as follows (see Figure 35 for an example of the process): a. The load test is plotted in loglog space. b. Two straight trend lines are drawn among the data points, with the second line extended or extrapolated to two inches. c. Two distinct loads are indentified: the first occurs at the intersection of the two sloped lines and the other at the presumed displacement of two inches. d. Since the first load typically occurred at a displacement of approximately 0.1 to 0.2 inches, (i.e., 5% to 10% of the extrapolated value), it was assumed that 5% of 50 PAGE 51 the ultimate tip resistance was mobilized. Therefore, the first load is assumed to be the sum of the ultimate skin friction and 5% of the tip resistance, while the second load is the ultimate skin plus tip resistance. Therefore, the separate contributions of ultimate tip and skin friction can be calculated. The Proposed UF Method The proposed method uses the following equation to estimate the ultimate pile unit tip resistance, q t from the CPT tip resistance, q c : q t = k b q ca (tip) 150 tsf where: k b is a factor that depends on the soil type as shown in Table 31. The soil type was determined using the soil classification chart for the standard electronic friction cone (Robertson et al, 1986) which includes tip resistance and sleeve friction. Soil cementation was determined by SPT samples, DTP tip2/tip1 ratio or SPT qc/N ratio (>10). q ca (tip): the average CPT tip resistance, which is calculated as follows: q ca (tip) = (q ca above + q ca below ) / 2 q ca above : average q c measured from the tip to 8 D above the tip; q ca below : average q c measured from the tip to 3 D below the tip for sand or 1D below the tip for clay; Impose the condition: q ca above q ca below which means if q ca above q ca below let q ca (tip) be equal to q ca below The proposed method uses the following equation to estimate the ultimate skin friction resistance of the pile, f s from the CPT tip resistance, q c : f s = q ca (side) *1.25 / F s 1.27 tsf 51 PAGE 52 where: F s : friction factor that depends on the soil type as shown in Table 32. The following criterion was used to determine the sand state based on relative density: loose sand (R.D. < 40 %), medium dense sand (40 % < R.D. < 70 %), and dense sand (R.D. >70 %). q ca (side) : the average q c within the calculating soil layers along the pile. 52 PAGE 53 Table 31. Ultimate unit tip resistance factor k b Well Cemented Sand Lightly Cemented Sand Gravel Sand Slit Clay 0.1 0.15 0.35 0.4 0.45 1 Table 32. Ultimate unit skin friction empirical factor, F s Well Cemented Sand Lightly Cemented Sand Gravel and Dense Sand Medium Dense Sand Loose Sand Silt, Sandy Clay, Clayey Sand Clay 300 250 200 150 100 60 50 53 PAGE 54 54 Figure 31. Dual tip penetrometer PAGE 55 55 Figure 32. The locations of 21 sites with load test data and CPT data PAGE 56 Figure 33. MathCAD program for Philipponnat method 56 PAGE 57 Apalachicola River Bridge Pier 3010020030040050060000.20.40.60.811.21.4Displacement (inches)Load (tons) 479 tons Figure 34. Static load test, Apalachicola Bay Bridge (pier 3) 57 PAGE 58 Jacksonville Industrial Zone 11010010000.010.1110Displacement (inches)Load (tons) 352.5 tons 249 tons Example: S + 5 % T = 249 tons S + T = 352.5 tons S = 243.5 tons T = 109 tons S : Ultimate Skin Friction T : Ultimate Tip Resistance Figure 35. Separate the ultimate skin friction and tip resistance 58 PAGE 59 CHAPTER 4 RESULTS This chapter includes two parts. The first provides a synopsis of the DPT and CPT results from various sites and the discussion of the cemented sand identifier. The second part is the evaluation of current axial ultimate pile capacity prediction methodologies using LRFD statistical method and the proposed method, including its verifications. Identification of Cemented Sand Cemented sands exist in many areas of the United States, including California, Texas, Florida, and along the banks of the lower Mississippi River. They also exist in Norway, Australia, Canada, and Italy (Puppala et al. 1995). If the sands found with a CPT test are not known to be cemented, the high bearing readings may be misinterpreted as being due to high relative densities. This can lead to an underestimation of the liquefaction potential of the soil and overestimate the ultimate pile capacity. It is a very challenging soil type for geotechnical engineers. It is therefore important to identify such soil before the design of foundation. In previous practice, engineers used SPT borings and CPT (Qc/N) to identify cemented sand. However, there are two drawbacks: firstly, lightly cemented sand may not be able to be identified by field technicians since the strength between sand particles is very weak, and secondly, the SPT provides a discontinuous data so that engineer judgment is required to estimate the N value between data points for calculating Qc / N ratio. It would be very beneficial if the DTP could be used to identify cemented sand. DTP & CPT Test Data at FDOT Bridge Sites Many FDOTs bridge sites have been revisited and had DTP and CPT tests performed. DTP and CPT tests were usually performed in pairs for comparison. All field tests were conducted by personnel from the State Material Office (SMO). The sites were chosen so that 59 PAGE 60 load tests and SPT data were available, and several sites also had cemented sand presented. The sites include: Archer Landfill (Archer, FL), I95 at Edgewood Avenue (Jacksonville, FL), Apalachicola River Bridge, Pier 3 (Apalachicola, FL), West Bay Bridge, Pier 20 (Bay County, FL), Port Orange Relief Bridge (Port Orange, FL), University of Central Florida (Orlando, FL), I295 at Blanding Blvd (Jacksonville, FL), I295 at Normandy Blvd. (Jacksonville, FL), and White City Bridge, Pier 5 and Pier 8 (White City, FL). Archer Landfill Site This site is one of the first sites used for DTP testing. The Archer landfill is an excellent site for initial scrutiny. Located in Archer, Florida, this site contains a very clean, fine sand deposit peripheral to the outside of the clay liner boundary. This area has been used for many years as a borrow area, where clean sands were mined for daily cover of the landfill debris. No SPT data is available for this site. Figure 41 shows the CPT tip resistance, DTP Tip 1 and Tip 2 resistances. The Tip 1 and CPT tip resistance are very close to each other. Therefore, the second tip (the one above the friction sleeve) does not influence the reading of the first tip (the one at the same location as the CPT). This phenomenon was found in all paired CPT and DTP tests, which means the distance between these two tips is sufficient enough to eliminate the interaction between each other. Another finding is that the relative magnitudes between DTP Tip 2 reading (T2) and Tip 1 reading (T1) for the clear fine sand equal 0.5. Figure 42 shows the comparison between friction ratios of CPT and DTP. From the plot, it can be found that the friction ratio of the DTP is much higher than that of CPT for sand (~ 3 times). The main reason for this increase is the normal stress adjacent to the friction sleeve due to the second tip. Therefore, the shear stress near the friction sleeve is also increased. The discrepancy in friction ratios between the DTP and CPT does not allow for soil classification unless a new soil classification chart for the DTP is created. However, the main purpose of DTP 60 PAGE 61 is to identify cemented sand using cemented sand identifiers (T2/T1 ratio). This ratio is presented below in the following analysis. I95 at Edgewood Avenue Site This site is located in Jacksonville and is a FDOT bridge site. Figure 43 shows the CPT tip resistance, DTP Tip 1 and Tip 2 resistances. The Tip 1 and CPT tip resistance are very close to each other. It can be found that the T2/T1 ratio varies with depth. Since the depth is function of soil type, T2/T1 is a function of soil type, depth, or both. Through later test results, it was found that the T2/T1 ratio is solely a function of soil type. Figure 44 shows the comparison between the friction ratios of the CPT and DTP. From the plot, it can be observed that the DTPs is again higher than the CPTs for sandy soil (about 3 times), but it is close to the CPTs for clayey soil. Since the friction ratio of DTP is different from that of CPT and hence could not be used to classify the soil, the friction ratio of following cases will not be presented. Figure 45 shows the comparison between the T2/T1 ratio and Qc/N ratio with depth. Generally speaking, when the Qc/N ratio is less than 3, the soil is considered as a clayey, silty soil. With Qc/N ratios between 3 to 10, the soil is considered a regular sand, and above 10, cemented sand may be encountered. For the I95 at Edgewood site, the soil profile is as follows: loose to medium dense sand with a thin layer of silt (0 feet 27 feet), clay (27 feet 36 feet), and dense, silty fine sand (36feet 59feet). This was confirmed by the Qc/N ratio. The T2/T1 ratio is 0.5 for loose to medium dense sand, greater than or equal to 1 for clayey silty soil, and greater than 1 for very dense, silty sand. Apalachicola River Bridge Site This site is also a FDOT bridge site located in Apalachicola, FL. Figure 46 shows the CPT tip resistance, DTP Tip 1 and Tip 2 resistances. The Tip 1 and CPT tip resistance are very 61 PAGE 62 close to each other. The soil profile in this site is as follows: two lightly cemented sand layers (0 feet 25 feet), sandy silt (25 feet 45 feet), silty clay (45 feet 60 feet), and lightly cemented sand layer (under 60 feet). Figure 47 gives the comparison between the T2/T1 ratio and Qc/N ratio with depth at the Apalachicola River Bridge site. The T2/T1 ratio is less than 0.5 for lightly cemented sand, between 0.5 to 1 for sandy silt and silty clay soil. West Bay Bridge Site This site is a FDOT bridge site located in Bay County, FL. Figure 48 shows two CPT tip resistances, DTP Tip 1 and Tip 2 resistances. The reason for presenting two CPTs is to show that there is significant spatial variability at this site. These two CPTs are within 100 feet and are totally different. The DPT Tip 1 and CPT1 Tip resistances are very close under 68 feet but DTP T1 is much higher than CPT1 tip resistance above 68 feet. However, DTP T1 is very similar to the CPT2 tip resistance at the depth above 64 feet but significant smaller at the depth below 64 feet. In this site, there are two lightly cemented sand layers and one well cemented sand layer. Figure 49 provides a comparison between T2/T1 ratio and Qc/N ratio along the depth at this site. The T2/T1 ratio is less than 0.5 for lightly cemented sand, and close to 1 for well cemented sand. Port Orange Relief Bridge This site is a FDOT bridge site located in Port Orange, FL. Figure 410 shows CPT tip resistance, DTP Tip 1 and Tip 2 resistances. The DPT Tip 1 and CPT1 Tip resistances are very close to one another. At this site, there are two well cemented sand layers with one clayey silt and one lightly cemented sand layer in the middle. Figure 411 shows a comparison between T2/T1 and Qc/N with depth. The T2/T1 ratio is less than 0.5 for lightly cemented sand and clayey silt and close to 1 for well cemented sand. The interesting finding in this figure is that the Qc/N ratio is higher for lightly cemented sand than that for well cemented sand. The lightly cemented sand is loose to medium dense sand with 62 PAGE 63 very low N blow count; therefore, it does not require a high tip resistance (q c ) to acquire very large Qc/N ratio. Therefore, this ratio cannot indicate just how strong the cementation is. University of Central Florida This site is located at the University of Central Florida, Orlando, FL. Figure 412 shows CPT tip resistance, DTP Tip 1 and Tip 2 resistances. The DPT Tip 1 and CPT1 Tip resistances are similar. At this site, there are two lightly cemented sand layers (T2/T1 <0.5) and four well cemented sand layers (T2/T1 1) with one clayey silt layer with shell (38 feet 53 feet) and two clay layers (4 feet 6 feet, 30 feet 34 feet). Figure 413 gives a comparison between T2/T1 and Qc/N with depth. The T2/T1 ratio is less than 0.5 for lightly cemented sand, close to 1 for well cemented sand, and less than 0.5 for clay and clayey silt. The interesting finding at this site is that the T2 values in clay and clayey silt are very low and even negative. That means that the clay is highly sensitive and loses its shear strength dramatically after the disturbance of the first tip. I295 at Blanding Blvd Site This is a site in Jacksonville, FL. Figure 414 shows CPT tip resistance, DTP Tip 1 and Tip 2 resistances. At this site, there are two sand layers (0 feet 32 feet, 53 feet 58 feet) with four lightly cemented sand layers within (T2/T1 <0.5) and one clayey silt layer (32 feet 53 feet) with one lightly cemented sand within. Figure 415 shows the comparison between T2/T1 and Qc/N with depth. The T2/T1 ratio is less than 0.5 for lightly cemented sand, close to or greater than 1 for clayey silt. I295 at Normandy Blvd. Site This is another site in Jacksonville, FL. Figure 416 shows CPT tip resistance and DTP Tip 1 and Tip 2 resistances. At this site, there are three lightly cemented sand layers (T2/T1 63 PAGE 64 <0.5). Figure 417 shows a comparison between T2/T1 and Qc/N with depth. The T2/T1 ratio is less than 0.5 for lightly cemented sand. White City Bridge Site (Pier 5) This is a site in White City, FL. Figure 418 shows CPT tip resistance and DTP Tip 1 and Tip 2 resistances. At this site, there are three lightly cemented sand layers, two silty sand layers and three clay layers. Figure 419 indicates the comparison between T2/T1 and Qc/N with depth. The T2/T1 ratio is less than 0.5 for lightly cemented sand, is between 0.5 1 for silty sand and close to or greater than 1 for clay. White City Bridge Site (Pier 8) This is the second pier at White City Bridge site in White City, FL. Figure 420 shows CPT tip and DTP Tip 1 and Tip 2 resistances. There are three lightly cemented sand layers, one silty sand layer, one clay layer, and one sand layer. Figure 421 gives the comparison between T2/T1 and Qc/N with depth. The T2/T1 ratio is less than 0.5 for lightly cemented sand, is between 0.5 and 1 for silty sand, greater than 1 for clay and 0.5 for sand. Identifier for Cemented Sand Summarized from DTP Data Based on the data collected to date, the T2 /T1 ratio appears to be a reasonable predictor or identifier for cemented sand. The following ranges for each soil type are based on the limited data collected in Florida. It may be different elsewhere and more data needs to be collected to verify these values. T2/T1: 0.5 NonCemented Sand (Loose and Medium Dense Sand) = 0.3~0.5 Lightly Cemented Sand 1 Strongly Cemented Sand = 0.5~1 Silty Sand, Sandy Silt and Silty Clay 64 PAGE 65 < 0.5 Highly Sensitive Clay > 1 Dense Silty Sand 1 Clayey Silt, Silty Clay and Clay The reasons for the identifier having different values for different soil types may seem ambiguous. However, the following is one possible explanation that will have to be verified by additional testing. For noncemented sand (loose and medium dense sand), T2 is less than T1 due to the compression of the sand after the first tip penetrates the soil. The value 0.5 for the identifier is function of the location of the second tip. Therefore, it is the unique character for the DTP used in my research. However, for dense sand, due to the dilation of sand particles after the disturbance of the first tip, T2 is greater than T1. For highly sensitive clay, the disturbance of the first tip dramatically reduces the shear strength of clay that results in the T2/T1 of less than 0.5. For lightly cemented sand, the cementation bonds are totally broken after penetration of the first tip so that T2 is the same as that for noncemented sand. However, T1 for lightly cemented sand is higher than that for noncemented sand, which causes T2/T1 ratio to be less than 0.5. For well cemented sand, the cementation bonds are so strong that they cannot all be broken or are only partially destroyed. Therefore, T2 is close to T1 for wellcemented sand. For low sensitive clay, clayey silt and silty clay, the penetration of the first tip may reduce the shear strength of the soil and the reduction depends on the strength of clay. However, the excess pore pressure generated by the intrusion of the first tip also increases T2. The resultant effects cause T2/T1 ratio to be less than, close to or greater than 1. For a mixed soil such as silty sand and sandy silt, T2/T1 is between 0.5 and 1 due to the combining effects. 65 PAGE 66 Assessing LRFD Resistance Factors, Based on Reliability (Risk) Two set of data were used for the LRFD assessment. They include 21 Florida cases and 28 in Louisiana. The data for Florida and Louisiana are analyzed separately to see how each method works for their respective soil type (Florida: predominantly sand, Louisiana: predominantly clay). Criteria Used to Quantify Pile Capacity from Load Test Data The criteria used to quantify pile capacity from load test data are the same as provided in FDOT specification for 21 cases in Florida. It is defined as the load that causes a pile tip deflection equal to the calculated elastic compression plus 0.15 inches plus 1/120 of the pile width in inches for piles 24 inches or less in width. For piles greater than 24 inches, it is equal to the calculated elastic compression plus 1/30 of the pile width (FDOT Specification 2007, 4552.2.1). The criteria are similar to Davisson capacity except for piles greater than 24 inches. All methods predict the ultimate pile tip and skin capacity, therefore the nominal capacities are calculated by the sum of ultimate skin capacity and 1/3 of ultimate tip capacity (R n = Q S ult + 1/3 Q T ult ). The ratio of measure capacity (FDOT failure load) to the nominal capacity is used to assess LRFD resistance factor. Since clay is the predominant soil type in Louisiana, a typical load test curve is shown in Figure 422. The ultimate pile capacity is less than the peak load. Therefore, it is not appropriate to compare the predicted nominal capacities (R n = Q S ult + 1/3 Q T ult ) with Davisson capacity (Q Davisson ). Thus it was decided to compare the predicted ultimate pile capacity (R ult = Q S ult + Q T ult ) with the ultimate capacity from the load test (load at 2 inches of displacement). According to FDOT specification, the allowable load of any pile tested must be either 50 % of the maximum applied load or 50 % of the failure load, whichever is smaller. 66 PAGE 67 LRFD Resistance Factor, for Florida Soil Using the 14 CPT methods with the cone data from the 21 load test sites in Florida, see Tables 41, 42, and 43 the ultimate skin friction, ultimate tip resistance, and Davisson capacity were determined for each test pile. Also shown in these tables are the measured ultimate skin friction, ultimate tip resistance, Davisson Capacity (from load test curve and the proposed process to separate skin and tip). The highlighted cases represent cemented sand sites. Shown in Figure 423 is a plot of the ratio (measured Davisson capacity/predicted Davisson capacity) for each method for all 21 piles. In this plot, the cases are sorted by diameter. No correction between the ratio and diameter. It was attemped to sort the case by length and there still was no correction between the ratio and length. The methods above the value of one are conservative. Next, the LRFD resistance factors, were assessed for each method. Based on the ratio of measured to predicted pile capacities, Ri for each of the sites, the mean, R standard deviation, R and coefficient of variation, COV R were determined for all 14 CPT methods. Using the computed mean, R and coefficient of variation, COV R the LRFD resistance factors, were determined with a reliability index, of 2.5 (recommended for redundant foundation elements) for each method. Tables 44, 45, and 46 are the LRFD resistance factors for ultimate skin friction, ultimate tip resistance, and Davisson capacity. The last method is the one proposed by UF and was created by modifying the most promising method, the Philipponnat method (1980). Evident from Tables 44, 45, and 46, many of the methods result in higher resistance factors, However, if the method has a substantial amount of scatter (a high COV R ), Table 44 (e.g. Schmertmann, Prince, etc.), then the LRFD resistance factor, will be adjusted downward. 67 PAGE 68 Of strong interest is a ranking of the 14 CPT methods investigated. The latter may be determined as follows: R design = R n (i.e. predicted capacity from individual method) However R = R m (i.e. Davisson) / R n (predicted capacity) Substituting the above equation into the R design equation for R n R design = (/ R ) R m The term (/ R ) in the above equation identifies the percent of measured Davisson capacity that is available for design. Obviously, the higher the (/ R ) term, the better the method. Shown in Tables 44, 45, and 46 are (/ R ) terms for each method. Clearly, Philipponnat & LCPC are the better methods. Interestingly, both the Philipponnat & LCPC methods use just the CPT tip resistance, q c, to predict axial pile capacity. Finally, the proposed UF method gives the value of 0.617 for / R which means 61.7% of load test measured capacity can be used for design. LRFD Resistance Factor, for Louisiana Soil Tables 47, 48, and 49 show the ultimate skin friction, ultimate tip resistance, and ultimate pile capacity using the 14 CPT methods with cone data from 28 load test sites in Louisiana. Shown in Figure 424 is a plot of the ratio (measured ultimate capacity/predicted ultimate capacity) for each method. Again, methods above one are conservative. Tables 410, 411, and 412 show the LRFD resistance factors for ultimate skin friction, ultimate tip resistance, and ultimate pile capacity. Shown in Tables 410, 411, and 412 are the (/ R ) terms for each method. From each table, it is found that the ranking of prediction methods are different from the Florida data. The 68 PAGE 69 Jardine & Chow (MTD) & Philipponnat methods are the better methods. However, both the Jardine & Chow (MTD) and Philipponnat methods use only the CPT tip resistance, qc, to predict axial pile capacity. The proposed UF method provides a value of 0.673 for / R which is second best method, and only slightly less than the MTD method. In summation, it appears that the proposed UF method works very well for both sands (Florida soil) and clays (Louisiana soil). Evaluate the Prediction Methods Using the Bootstrap Method After calculating the i (Measured/Predicted) for each method and performing an LRFD resistance factor analysis, one question remains: how representative is the data to the population? Is there sufficient data to predict the pile capacity with a certain level of confidence? Fortunately, a very powerful tool is available, termed the Bootstrap method. This method estimates the sampling distribution of an estimator by resampling with replacements from the original sample. For instance, there are total of 21 cases involving Florida soil. We have 21 s which are original samples from the whole population of s. We want to know how good the samples mean and standard deviations are. There are three steps involved in the method. The first is to resample with replacement 21 s from our original samples, which means picking one by one and after each pick it is placed back into the pool for the next pick. Therefore, in this new set of samples, some 'smay appear more than once, but some not at all. The second step is to calculate the mean and standard deviation of the new sample set. The last step is to repeat the first and second steps numerous times (at least 1,000). Finally, one obtains the distribution of the mean and standard deviation of the new sample sets. From these one can determine how representative the samples are to the population. 69 PAGE 70 70 Figures 425, 426, and 427 show the results of the Bootstrap analysis for Florida soils. It can be seen that the standard deviation of the resampled mean and resampled standard deviation are quite small (0.061, 0.042 respec tively). Figures 428, 429, and 430 show the analysis for the proposed method for Louisiana soil. It also shows very small sta ndard deviation (0.042, 0.027) for the resampled mean and standard devi ation. Therefore, the proposed method does provide a higher level of quality predictions. Figures 431, 432, and 433 show the fre quency diagram of the original sample, resampled mean and resampled standard deviat ion for Schmertmann method for Florida soils. As it shows, the standard devi ation of resampled mean and resampled standard deviation are somewhat larger (0.148, 0.243), which means th e sample may not be representative of the population. In other words, in order to perfor m more accurate analysis to Schmertmann method, more data is required. PAGE 71 Table 41. Predicted ultimate skin friction for 14 CPT methods (Florida soil) Load Test No.Site NamePier (Bent/Slab) No.Diameter (in)Length (ft)SchmertmannDeRuiter & Beringe n Bustamante & Gianeselli (LCPC)Aoki & de Alenca r PenpilePhilipponnatPrince & WardleTurmay & FakhrooAlmeida et al.Eslami & FelleniusJardine & Chow (MTD)Powell et al.UWA05Zhou et al.Proposed methodSkin friction ( Load Test ) 1FT Myers1467110172183926615079106288140110296591361501332Apalachicola river bridgeBent 161860.2139151222112851771101342641431702341171931541233Port orange relief bridgeBent 191833.51291101661206519893992321282122331861141261034West palm I95Bent 41852.61111202681516627375953681512583682221241921625West palm I95Bent 9 1845.8118109246117741911001103231461843241271371951876Apalachicola bay b ridgeBent 221868939223512765194721153721271793731201261941657Trout River, I95 @ Edgewood DriveBent 11848.5848421683551576398293106135294100116157147.58Jacksonville industrial1204615018334722089343119115368228393368305163280243.59Jacksonville industrial2203613613621416485257132125286165284286214153247119.410Apalachicola river bridgePier 3A2488.419112653421711437613310465627339166723923430831111Apalachicola river bridgePier 3B2488.432029846925117739323328364134643665224137930731112Choctawhatchee b ay bridgeSlab 32481.716926526715411125711621457626422960411127123123713Choctawhatchee b ay bridgeSlab 26246914323433514190254941884592031904679322125537414Overstreet bridgePier162465.620824832819210229013421637025130338115023124820515Bayou Chico BridgePier 152426.7831032301284820155642461292342462169320125016Black water bay b ridgeBent 202484.9237187437216114348149204530250400534266224348360.517Escambia river b ridgeBent 5248731636572538916460520424575240362175535132360538218Escambia river b ridgeBent 772461.41612174732308938810618547425435648321518538869519White City BridgePier 52437.280.5170167775112470130174125183.517911212011410220White City BridgePier 82428.57116117067411374888183102134190102.5931209621West bay bridgePier 2030103.62241462681911593021633619733492941008157365302390 71 PAGE 72 Table 42. Predicted ultimate tip resistance for 14 CPT methods (Florida soil) Load Test No.Site NamePier (Bent/Slab) No.Diameter (in)Length (ft)SchmertmannDeRuiter & BeringenBustamante & Gianeselli (LCPC)Aoki & de AlencarPenpilePhilipponnatPrince & WardleTurmay & FakhrooAlmeida et al.Eslami & FelleniusJardine & Chow (MTD)Powell et al.UWA05Zhou et al.Proposed metho d Tip resistance (Load Test)1FT Myers146763431.53261381.43.59672Apalachicola river b ridgeBent 161860.22572571261704714210425711020915017915412246773Port orange relief b ridgeBent 191833.52652651862835720217026518748722130415915151454West palm I95Bent 41852.63383383163381043272243382303043763383092352202665West palm I95Bent 9 1845.81431438113724102962518216410713386112912616Apalachicola bay b ridgeBent 221868338338211258971901583381592272322582061571541607Trout River, I95 @ Edgewood DriveBent 11848.52132131191814012089214941791411531281081041108Jacksonville industrial12046276276263383962772483342497242834031662042351099Jacksonville industrial22036417417216297852151824171854852353002681857615910Apalachicola river b ridgePier 3A2488.43583582073275723318736219449021930821521620227311Apalachicola river b ridgePier 3B2488.42542542272846922617430116435224026615320821827312Choctawhatchee b ay bridgeSlab 32481.714617164496365543837878.566141213Choctawhatchee b ay bridgeSlab 262469254254154213611491373371307014321015515814624214Overstreet bridgePier162465.623102832121881764771331757321614192434515Bayou Chico BridgePier 152426.760060044260014849537960040384844860056236438337616Black water bay b ridgeBent 202484.920220219430959239189300189543216306121190190169.517Escambia river b ridgeBent 52487455455202334722582406002385992243852732372658518Escambia river b ridgeBent 772461.477778812814136136135139196862244614013310519White City BridgePier 52437.2600600507459.517239928160029415447047644230232132320White City BridgePier 82428.5600600440434159357266600279244426452.540927930428121West bay bridgePier 2030103.6255255107163361211162751082899617615314812653 72 PAGE 73 73 Table 43. Predicted Davisson capacity for 14 CPT methods (Florida soil) Load Test No.Site NamePier (Bent/Slab) No.Diameter (in)Length (ft)SchmertmannDeRuiter & BeringenBustamante & Gianeselli (LCPC)Aoki & de Alenca r PenpilePhilipponnatPrince & WardleTurmay & FakhrooAlmeida et al.Eslami & FelleniusJardine & Chow (MTD)Powell et al.UWA05Zhou et al.Proposed MethodDavisson Capacity (Load Test)1FT Myers14671121741849466150.680.2108289141113296601441521402Apalachicola river b rid g eBent 161860.22242372641691012241452193012122203331673151701653Port orange relief b rid g eBent 191833.5217198228213842641501872952902863342392651431034West palm I95Bent 41852.62232323742631003821502084442533834813253592662505West palm I95Bent 9 1845.8166157273169822251321933512012203681552492252666Apalachicola bay b rid g eBent 221868205205305213972571252274252022574591892842452137Trout River, I95 @ Edgewood DriveBent 11848.51551552551316819792169324166182345142.52231921948Jacksonville industrial120462422754343551213582012264514694875023603673582839Jacksonville industrial2203627527528626311332919326534732736238630433827218510Apalachicola river b rid g ePier 3A2488.4311245.5603.5319133443194.8224.8720436464769310.542637547911Apalachicola river b rid g ePier 3B2488.440538354534620046229138369646351674029259538047912Choctawhatchee b a y brid g eSlab 32481.717426827318911327213823659427724263311433723624913Choctawhatchee b a y brid g eSlab 26246922931938719711128114030050322623853714537930348014Overstreet bridgePier162465.621625233728810635319337541531032845315542426225015Bayou Chico Brid g ePier 152426.72833033783289736618126438041138344640345732939316Black water bay b rid g eBent 202484.930425550131913441121230459343147263630641441143817Escambia river b rid g eBent 52487468517792520188691284445831603695884442.555969342518Escambia river b rid g eBent 772461.41872435023049443215123052032038555823132643273519White City BridgePier 52437.2280.5370336230108256.5164330272176.5340338259421.522133220White City BridgePier 82428.52713613172129425613628827618327634123937222125021West bay bridgePier 2030103.6309.523130425517138220245210094463261066208.5513344425 PAGE 74 Table 44. LRFD resistance factors,, for CPT methods (ultimate skin friction, Florida soil) Analysis methodRCOV RRSchmertmann (1978)1.6190.5070.5480.339de Ruiter and Beringen (1979)1.4600.5090.4920.337Bustamante and Gianeselli(LCPC) (1982)0.7770.3580.3820.491Aoki & de Alencar (1975)1.4920.3780.6970.467Penpile Method (Clisby et al. 1978)2.7370.5310.8730.319Philipponnat (1980)0.8990.3610.4390.488Prince and Wardle (1982)2.2850.5700.6630.290Tumay and Fakhroo (1982)1.6530.5370.5200.314Almeida et al. (1996)0.5880.4260.2430.414Eslami & Fellenius (1997)1.1900.4050.5200.437Jardine & Chow (MTD) (1996)0.9260.4530.3580.387Powell et al. (2001)0.5820.4230.2430.418UWA05 (2005)1.5020.5850.4200.280Zhou et al. (1982)1.3260.5370.4170.315Proposed Method (2006)0.9750.2920.5650.580 Table 45. LRFD resistance factors,, for CPT methods (ultimate tip resistance, Florida soil) Analysis methodRCOV RRSchmertmann (1978)0.7550.6630.1750.233de Ruiter and Beringen (1979)0.9861.0230.1060.107Bustamante and Gianeselli(LCPC) (1982)1.0010.6550.2380.237Aoki & de Alencar (1975)0.7950.6690.1820.229Penpile Method (Clisby et al. 1978)3.1720.7470.6080.192Philipponnat (1980)0.8790.7140.1820.207Prince and Wardle (1982)1.0900.7450.2100.193Tumay and Fakhroo (1982)0.5300.5750.1520.287Almeida et al. (1996)1.2761.1550.1060.083Eslami & Fellenius (1997)0.8231.0540.0830.101Jardine & Chow (MTD) (1996)0.8290.6110.2180.263Powell et al. (2001)0.8261.2620.0570.069UWA05 (2005)1.2020.7080.2520.210Zhou et al. (1982)0.8850.5670.2590.293Proposed Method (2006)1.1410.5040.3900.341 74 PAGE 75 75 Analysis methodRCOV RRSchmertmann (1978)1.3270.5220.4330.327de Ruiter and Beringen (1979)1.2240.4740.4490.367Bustamante and Gianeselli(LCPC) (1982)0.8520.3090.4730.555Aoki & de Alencar (1975)1.2890.3850.5910.459Penpile Method (Clisby et al. 1978)2.8690.4781.0450.364Philipponnat (1980)0.9690.3420.4950.511Prince and Wardle (1982)1.9350.4560.7440.385Tumay and Fakhroo (1982)1.2290.4710.4550.370Almeida et al. Table 46. LRFD resistance factors,, for CPT methods (Davisson capacity, Florida soil) ( 1996 ) 0.6920.3960.3090.447Eslami & Fellenius (1997)1.0940.4460.4300.394Jardine & Chow (MTD) (1996)0.9820.4150.4190.4260.3750.2950.471 Powell et al. (2001)0.628UWA05 ( 2005 ) 1.492Zhou et al. (1982)0.873Proposed Method (2006)1.079 0.5160.4940.3310.4360.3530.4040.2670.6650.617 PAGE 76 Table 47. Predicted ultimate skin friction for 14 CPT methods (Louisiana soil) Load Test No.Site NamePier (Bent/Slab) No.Diameter (in)Length (ft)SchmertmannDeRuiter & BeringenBustamante & Gianeselli (LCPC)Aoki & de AlencarPenpilePhilipponnatPrince & WardleTurmay & FakhrooAlmeida et al.Eslami & FelleniusJardine & Chow (MTD)Powell et al.UWA05Zhou et al.Proposed Metho d Skin Friction ( Load Test ) 1Houma_ICW_BridgeTP 11480181187208878313211115322119211523055208132106.42Houma_ICW_BridgeTP 2147099201164475989661171291108614029.515889433Houma_ICW_BridgeTP 31480140159140686810884.514717417610718242178108116.54Houma_ICW_BridgeTP 4148112315613857619371151136.51408314536175931055Bayou_Boeuf_Main_SpanTP 21470892271514550106531201491418516136142106976Bayou_Boeuf_Main_SpanTP 5148059150105423272301619614764.510427109721157Bayou_Boeuf_West_ApproachTP 11489.51311771195676.590821311201318513131206901058Bayou_Boeuf_West_ApproachTP 31463.591138115465380.560107119130721264314580.5125.59Bayou_Boeuf_East_ApproachF 1146888.5138.587.547.5516952107921425910025146699010Bayou_Boeuf_East_ApproachF 2147191131864050635011687122609524149639011Bayou_Boeuf_East_ApproachF 31477.599176.511350558756129117149.57612830164878212Bayou_Boeuf_East_ApproachF 41479107.524818158591146012515714894171361711149513Bayou_Boeuf_East_ApproachF 514799618511340538153135.51301687514231161817014Houma_ICW_BridgeTP 51671.5111127884462.5646412685106659323180648515Bayou_Boeuf_West_ApproachTP 4167012616614066791029812914015290149372031028016Bayou_Ramos_BridgeTP 1167811816313659651016814816515888175431861019817Bayou_Ramos_BridgeTP 7167712916912661711037814013614989145.53419898107.618Houma18105187316202.5102102173106215248.52571462685229417316019Tickfaw_River_BridgeTP 13059.3283606412145170286238251387307.52804186739628640720Bayou_Boeuf_West_ApproachTP 230110429492374.51702422782813444373873014738362527828521Bayou_Ramos_BridgeTP 2308843740752319821339828334361744235463515649739836522Bayou_Ramos_BridgeTP 33010445355146625325649034140677850038080514160349035323Bayou_Ramos_BridgeTP 43099.346454245121221349728541161444439364118852349745424Bayou_Ramos_BridgeTP 53011343455737520821734526740442145636545217058434530725Bayou_Boeuf_East_ApproachF 630110332676487154186213.5197378.544051931147973.5532313.528026Gibson_Raceland_ HighwayTP 130116319553377169169315175435428455300463.59350431536427Gibson_Raceland_ HighwayTP 430124412648447211203400245.550552152635855913557340038728Luling_Bridge30112491516408209228420.5311500698582362724160582420.5424 76 PAGE 77 Table 48. Predicted ultimate tip resistance for 14 CPT methods (Louisiana soil) Load Test No.Site NamePier (Bent/Slab) No.Diameter (in)Length (ft)SchmertmannDeRuiter & BeringenBustamante & Gianeselli (LCPC)Aoki & de Alenca r PenpilePhilipponnatPrince & WardleTurmay & FakhrooAlmeida et al.Eslami & FelleniusJardine & Chow (MTD)Powell et al.UWA05Zhou et al.Proposed MethodTip Resistance ( Load Test ) 1Houma_ICW_BridgeTP 1148014698.5475143122058161532.62Houma_ICW_BridgeTP 21470525.5643.545231233109123Houma_ICW_BridgeTP 3148012588375123112157151484Houma_ICW_BridgeTP 414811157736411.529163714125Bayou_Boeuf_Main_SpanTP 21470241129.5261519162616316125143121196Bayou_Boeuf_Main_SpanTP 514802.51853432.5111821.5987Bayou_Boeuf_West_ApproachTP 11489.59.546.511218716413.514761918.58Bayou_Boeuf_West_ApproachTP 31463.593.593.562.564.5225539.5984064836456524539.59Bayou_Boeuf_East_ApproachF 1146815710948615413.521.569171610Bayou_Boeuf_East_ApproachF 214711258837512310175714.51311Bayou_Boeuf_East_ApproachF 31477.51258746.55123101847141312Bayou_Boeuf_East_ApproachF 4147919.59131269.5719.55.5172991219913Bayou_Boeuf_East_ApproachF 51479126983.675123111957151414Houma_ICW_BridgeTP 51671.511587364112917.537151215Bayou_Boeuf_West_ApproachTP 4167017811124971751524810201816Bayou_Ramos_BridgeTP 11678484823476.428297025.56630412940314317Bayou_Ramos_BridgeTP 71677157151251071551235.58.6921108.418Houma181052411171771410256213791528285619Tickfaw_River_BridgeTP 13059.31225590793868.548.5125431161956973123617720Bayou_Boeuf_West_ApproachTP 23011011311312311948887311767150.5101109681468821Bayou_Ramos_BridgeTP 23088212212107252151551542121553809525012716817716222Bayou_Ramos_BridgeTP 3301041831831511735613110618710224611316611019912514723Bayou_Ramos_BridgeTP 43099.3787815131529175193264.5176.54391412864726817514624Bayou_Ramos_BridgeTP 53011345545538151116345231372929630335948027337833626825Bayou_Boeuf_East_ApproachF 6301109744676027469710024901453958964626Gibson_Raceland_ HighwayTP 13011625225217032258220197521164162162.5265151246.519327527Gibson_Raceland_ HighwayTP 430124310310302414.590310254668217211.526035218635925225428Luling_Bridge30112378.5378.539664154390392.5937.533038334753422739637276 77 PAGE 78 78 Table 49. Predicted ultimate pile capacity for 14 CPT methods (Louisiana soil) Load Test No.Site NamePier (Bent/Slab) No.Diameter (in)Length (ft)SchmertmannDeRuiter & BeringenBustamante & Gianeselli (LCPC)Aoki & de Alenca r PenpilePhilipponnatPrince & WardleTurmay & FakhrooAlmeida et al.Eslami & FelleniusJardine & Chow (MTD)Powell et al.UWA05Zhou et al.Proposed MethodUltimate Capacity ( Load Test ) 1Houma_ICW_BridgeTP 114801951942179587135117167224204134.8235632241471392Houma_ICW_BridgeTP 2147010420317053629270122131113981433216898553Houma_ICW_BridgeTP 31480151.51641487671.5115.589159177187128.5187491931221254Houma_ICW_BridgeTP 414811351611456464997516213914998148431891051055Bayou_Boeuf_Main_SpanTP 2147011323718171.565.512569146164171146186501741271166Bayou_Boeuf_Main_SpanTP 5148061151113.547367633164971488310728.5118801157Bayou_Boeuf_West_ApproachTP 11489.5140182125677910888.5146.512414499138372241091058Bayou_Boeuf_West_ApproachTP 31463.518423117811075136100205159194156191991971261659Bayou_Boeuf_East_ApproachF 114681031459757557758122961558010733.6163859010Bayou_Boeuf_East_ApproachF 214711031369448537055129901337710031163769011Bayou_Boeuf_East_ApproachF 31477.511118212157599460.514112016094132371781008212Bayou_Boeuf_East_ApproachF 41479127257194706512368144163165123180481911239513Bayou_Boeuf_East_ApproachF 514791081901224856.58858148133.517993146.539176957014Houma_ICW_BridgeTP 51671.512213295.5506669.568.513887115829629195768515Bayou_Boeuf_West_ApproachTP 416701431741517883.5111105146.5145166.5113157472231208016Bayou_Ramos_BridgeTP 1167816721115910772130972181902241182167222713214017Bayou_Ramos_BridgeTP 71677144175.5142737611385155141161124.51544321910811618Houma18105211327220119109186116.52402542781832776632220121619Tickfaw_River_BridgeTP 13059.3405660.5502224207.535528737643042447548714051934748420Bayou_Boeuf_West_ApproachTP 230110542606497.5289290366354461.550553840258215177136728521Bayou_Ramos_BridgeTP 2308864961863644922855343755677282244988528366557552722Bayou_Ramos_BridgeTP 33010463673561742631255444759388074649397125180261550023Bayou_Ramos_BridgeTP 43099.3542620602527241672478676790883533.592723579167257824Bayou_Ramos_BridgeTP 5301138891012756719381796579.51133717.5759724.593244396268157725Bayou_Boeuf_East_ApproachF 630110429.572055421421336023447846460945651813262836028026Gibson_Raceland_ HighwayTP 130116571805547.5491227535372955.559261746372924575050863927Gibson_Raceland_ HighwayTP 4301247229587496262937104991173738.573861891132193265264128Luling_Bridge30112870895804850282810704143810279657091258387978793500 PAGE 79 Table 410. LRFD resistance factors,, for CPT methods (ultimate skin friction, Louisiana soil) Analysis methodRCOV RRSchmertmann (1978)0.9100.3220.4890.537de Ruiter and Beringen (1979)0.6080.2690.3730.613Bustamante and Gianeselli(LCPC) (1982)0.8000.2550.5070.634Aoki & de Alencar (1975)1.8460.2391.2180.660Penpile Method (Clisby et al. 1978)1.6900.3150.9240.547Philipponnat (1980)1.0530.2480.6790.645Prince and Wardle (1982)1.5040.3780.7030.467Tumay and Fakhroo (1982)0.8030.2830.4760.593Almeida et al. (1996)0.7340.2820.4360.594Eslami & Fellenius (1997)0.7130.2590.4480.628Jardine & Chow (MTD) (1996)1.1450.2400.7520.657Powell et al. (2001)0.6840.2750.4130.604UWA05 (2005)2.9830.3061.6670.559Zhou et al. (1982)0.6140.2900.3570.582Proposed Method (2006)1.0400.2480.6700.644 Table 411. LRFD resistance factors,, for CPT methods (ultimate tip resistance, Louisiana soil) Analysis methodRCOV RRSchmertmann (1978)1.0730.6780.2410.224de Ruiter and Beringen (1979)1.8691.0000.2090.112Bustamante and Gianeselli(LCPC) (1982)1.3420.7280.2690.200Aoki & de Alencar (1975)1.1330.9100.1520.135Penpile Method (Clisby et al. 1978)4.0240.7270.8070.201Philipponnat (1980)1.5190.8560.2290.151Prince and Wardle (1982)1.8640.9370.2370.127Tumay and Fakhroo (1982)0.8760.8440.1360.155Almeida et al. (1996)2.7211.1630.2240.082Eslami & Fellenius (1997)1.1680.9130.1560.134Jardine & Chow (MTD) (1996)0.9410.5900.2610.277Powell et al. (2001)1.7101.1760.1370.080UWA05 (2005)1.7940.6780.4030.225Zhou et al. (1982)0.8890.5780.2530.285Proposed Method (2006)1.1070.4470.4350.393 79 PAGE 80 Table 412. LRFD resistance factors,, for CPT methods (ultimate pile capacity, Louisiana soil) Analysis methodRCOV RRSchmertmann (1978)0.8380.3200.4520.539de Ruiter and Beringen (1979)0.6130.2570.3870.631Bustamante and Gianeselli(LCPC) (1982)0.7760.2500.4980.641Aoki & de Alencar (1975)1.4170.2800.8450.596Penpile Method (Clisby et al. 1978)1.7400.3120.9580.551Philipponnat (1980)0.9850.2310.6630.673Prince and Wardle (1982)1.3760.3720.6520.474Tumay and Fakhroo (1982)0.6980.2630.4340.622Almeida et al. (1996)0.7660.2660.4730.618Eslami & Fellenius (1997)0.6760.2460.4380.648Jardine & Chow (MTD) (1996)0.9710.2220.6680.687Powell et al. (2001)0.6780.2600.4240.626UWA05 (2005)2.3170.2761.3960.603Zhou et al. (1982)0.5970.2860.3510.587Proposed Method (2006)0.9640.2300.6490.673 80 PAGE 81 Archer Landfill0510152025303540450100200300400Qc(tsf)Depth(ft) DTP T1 DTP T2 CPT Figure 41. CPT and DTP test data from Archer Landfill site 81 PAGE 82 Archer Landfill051015202530354045012345F. R. (%)Depth (ft) DTP CPT Figure 42. Friction ratio of CPT and DTP rest from Archer Landfill site 82 PAGE 83 Slightly Silty Fine Sand to Fine Sand Dense Silty Fine Sand Clay Slightly Silty Fine Sand to Fine Sand Slightly Silty Fine Sand to Fine Sand Dense Silty Fine Sand Dense Silty Fine Sand Clay I95 @ Edgewood01020304050600100200300400Qc (tsf)Depth (ft ) DTP T1 DTP T2 CPT Figure 43. CPT and DTP test data from I at Edgewood Avenue site 83 PAGE 84 I95 @ Edgewood0102030405060012345F.R.(%)Depth (ft ) DTP CPT Figure 44. Friction ratio of CPT and DTP test from I at Edgewood Avenue site 84 PAGE 85 I95 @ Edgewood010203040506000.511.52Tip2/Tip1Depth (ft ) DTP I95 @ Edgewood0102030405060051015Qc/NDepth (ft ) DTP CPT Figure 45. Tip2/Tip1 ratio and Qc/N ratio of CPT and DTP test from I at Edgewood Avenue site 85 PAGE 86 Apalachicola River Bridge Pier 301020304050607080901000100200300Tip Resistance(tsf)Depth (ft ) Tip 1 Tip 2 CPT Lightly Cemented Sand Lightly Cemented Sand layer Sandy Silt SiltyClay Lightly Cemented Sand Lightly Cemented Sand Lightly Cemented Sand layer Lightly Cemented Sand layer Sandy Silt SiltyClay Lightly Cemented Sand Figure 46. CPT and DTP test data from Apalachicola River Bridge site 86 PAGE 87 010203040506070809010000.511.5Tip 2/Tip 1Depth (ft ) Apalachicola River Bridge Pier 3010203040506070809010005101520Qc/NDepth (ft ) DTP CPT Lightly Cemented Sand LightlyCemented Sand Sandy Silt SiltyClay LightlyCemented Sand Lightly Cemented Sand LightlyCemented Sand LightlyCemented Sand Sandy Silt SiltyClay LightlyCemented Sand Figure 47. Tip2/Tip1 ratio and Qc/N ratio of CPT and DTP test from Apalachicola River Bridge site 87 PAGE 88 West Bay Bridge, DTP vs CPTs0102030405060708090100110120130140150160050100150200250Tip Resistance (tsf)Depth (ft) DTP T1 DTP T2 CPT1 CPT2 Lightly Cemented Sand layer Well Cemented Sand West Bay Bridge, DTP vs CPTs0102030405060708090100110120130140150160050100150200250Tip Resistance (tsf)Depth (ft) DTP T1 DTP T2 CPT1 CPT2 Lightly Cemented Sand layer Well Cemented Sand Figure 48. CPT and DTP test data from West Bay Bridge site 88 PAGE 89 010203040506070809010011012013014015000.511.5Tip2/Tip1Depth (ft ) Lightly Cemented Sand layerCemented Sand layer DTP T1010203040506070809010011012013014015005101520T1/NDepth (ft ) Figure 49. Tip2/Tip1 ratio and Qc/N ratio of CPT and DTP test from West Bay Bridge site 89 PAGE 90 0510152025303540450100200300400500Tip Bearing (TSF)Depth (ft ) DTP Tip1 DTP Tip2 CPT Well Cemented SandClayey SiltWellCemented Sand Lightly Cemented Sand Well Cemented SandClayey SiltWellCemented Sand Lightly Cemented Sand Figure 410. CPT and DTP test data from Port Orange Relief Bridge site 90 PAGE 91 05101520253035404505101520Qc/NDe p th ( ft ) DTP CPT Tip2/Tip105101520253035404500.511.5De p th ( ft ) Well Cemented SandClayey SiltWellCemented Sand Lightly Cemented Sand Well Cemented SandClayey SiltWellCemented Sand Lightly Cemented Sand Figure 411. Tip2/Tip1 ratio and Qc/N ratio of CPT and DTP test from Port Orange Relief Bridge site 91 PAGE 92 Well Cemented San d Clayey Silt with Shell Well Cemented San d Clay Sand University of Central Florida0102030405060700100200Qc(tsf)Depth(ft) DTP T1 DTP T2 CPT Well Cemented Sand Lightly Cemented San d Sand Well Cemented Sand Clay Lightly Cemented San d Figure 412. CPT and DTP test data at the University of Central Florida site 92 PAGE 93 051015202530354045505560657000.511.5Tip2/Tip1Depth (ft) 01020304050607005101520Qc/NDepth(ft) CPT, Qc/N Well Cemented Sand Lightly Cemented Sand Figure 413. Tip2/Tip1 ratio and Qc/N ratio of CPT and DTP test at the University of Central Florida site 93 PAGE 94 Lightly Cemented Sand I295 @ Blanding Blvd0102030405060700100200300400Qc(tsf)Depth(f t DTP T1 DTP T2 CPT Lightly Cemented Sand I295 @ Blanding Blvd0102030405060700100200300400Qc(tsf)Depth(f t DTP T1 DTP T2 CPT I295 @ Blanding Blvd0102030405060700100200300400Qc(tsf)Depth(f t DTP T1 DTP T2 CPT Figure 414. CPT and DTP test data at I295 at Blanding Blvd site 94 PAGE 95 T2/ T105101520253035404550556000.511.52Depth(f t 05101520253035404550556005101520Qc/NDepth(f t DTP CPT Lightly Cemented Sand T2/ T105101520253035404550556000.511.52Depth(f t 05101520253035404550556005101520Qc/NDepth(f t DTP CPT Lightly Cemented Sand Figure 415. Tip2/Tip1 ratio and Qc/N ratio of CPT and DTP test at I295 at Blanding Blvd site 95 PAGE 96 Tip Resistance0510152025303540450100200300400Qc (tsf)Depth (ft) ` DTP T1 DTP T2 CPT Lightly Cemented Sand Tip Resistance0510152025303540450100200300400Qc (tsf)Depth (ft) ` DTP T1 DTP T2 CPT Lightly Cemented Sand Lightly Cemented Sand Figure 416. CPT and DTP test data at I295 at Normandy Blvd. site 96 PAGE 97 05101520253035404505101520Qc/NDepth (ft) DTP CPT Tip2/Tip105101520253035404500.511.5Depth (ft ) Lightly Cemented Sand 05101520253035404505101520Qc/NDepth (ft) DTP CPT Tip2/Tip105101520253035404500.511.5Depth (ft ) Lightly Cemented Sand Figure 417. Tip2/Tip1 ratio and Qc/N ratio of CPT and DTP test at I295 at Normandy Blvd. site 97 PAGE 98 Lightly Cemented San d Sandy Clay Silty Sand Lightly Cemented Sand Clay Silty Sand Clay White City BridgePier 50510152025303540050100150200Qc(tsf)Depth (ft ) DTP T1 DTP T2 CPT Lightly Cemented San d Figure 418. CPT and DTP test data in White City site (pier 5) 98 PAGE 99 051015202530354000.511.5Tip2/Tip1Depth (ft ) White City BridgePier 5051015202530354005101520Qc/NDepth(ft) DTP CPT Lightly Cemented San d Figure 419. Tip2/Tip1 ratio and Qc/N ratio of CPT and DTP test in White City site (pier 5) 99 PAGE 100 White City BridgePier 80510152025303540050100150200250300350400Qc(tsf)Depth (ft ) DTP T1 DTP T2 CPT Lightly Cemented San d Silty Sand Lightly Cemented San d Clay Lightly Cemented San d Sand Figure 420. CPT and DTP test data in White City site (pier 8) 100 PAGE 101 Figure 421. Tip2/Tip1 ratio and Qc/N ratio of CPT and DTP test in White City site (pier 8) 0510152025303540 0 0.511.5Tip2/Tip1Depth (ft ) White City BridgePier 80510152025303540505101520Qc/NDepth(ft) DTP CPT Lightly Cemented San d 101 PAGE 102 Bayou_Boeuf_East_Approach_F40.0020.0040.0060.0080.00100.00120.000.00.20.40.60.81.01.21.41.61.82.0Displacement (in)Load (tons ) Q Davisson Figure 422. Typical load test curve in Louisiana soil 102 PAGE 103 00.511.522.533.54123456789101112131415161718192021Case No.i Schmertmann Method deRuiter & Beringen Method LCPC Method Aoki & Alencar Method Penpile Method Philipponnat Method Prince & Wardle Method Turmay & Fakhroo Method Almeida et al. Method Eslami & Fellenius Method Jardine & Chow Method Powell et al. Method UWA05 Method Zhou et al. Method Proposed Method 00.20.40.60.811.21.41.61.82123456789101112131415161718192021Case No.i Schmertmann Method deRuiter & Beringen Method LCPC Method Aoki & Alencar Method Penpile Method Philipponnat Method Prince & Wardle Method Turmay & Fakhroo Method Almeida et al. Method Eslami & Fellenius Method Jardine & Chow Method Powell et al. Method UWA05 Method Zhou et al. Method Proposed Method Figure 423. Comparisons (ratio: measured/predicted) for 14 methods using Florida soil. A) Yscale from 0 to 4. B) Yscale from 0 to 2. 103 PAGE 104 00.511.522.533.5412345678910111213141516171819202122232425262728Case No.i Schmertmann Method deRuiter & Beringen Method LCPC Method Aoki & Alencar Method Penpile Method Philipponnat Method Prince & Wardle Method Turmay & Fakhroo Method Almeida et al. Method Eslami & Fellenius Method Jardine & Chow Method Powell et al. Method UWA05 Method Zhou et al. Method Proposed Method 00.20.40.60.811.21.41.61.8212345678910111213141516171819202122232425262728Case No.i Schmertmann Method deRuiter & Beringen Method LCPC Method Aoki & Alencar Method Penpile Method Philipponnat Method Prince & Wardle Method Turmay & Fakhroo Method Almeida et al. Method Eslami & Fellenius Method Jardine & Chow Method Powell et al. Method UWA05 Method Zhou et al. Method Proposed Method Figure 424. Comparisons (ratio: measured/predicted) for 14 methods using Louisiana soil. A) Yscale from 0 to 4. B) Yscale from 0 to 2. 104 PAGE 105 0.60.811.21.41.600.511.52 Mean = 1.079Stdev = 0.288 Figure 425. Frequency of of the proposed method for Florida soil 0.830.870.910.9611.041.081.121.161.21.251.291.331.370246 Mean = 1.079Stdev = 0.061 Figure 426. Frequency of the sample means using Bootstrap method for Florida soil (100,000 resampling runs, the proposed method) 105 PAGE 106 0.110.130.160.180.210.230.260.280.310.330.360.380.4102468 Mean = 0.278Stdev = 0.042 Figure 427. Frequency of the sample standard deviations using Bootstrap method for Florida soil (100,000 resampling runs, the proposed method) 0.40.480.570.650.740.820.910.991.081.161.251.331.4200.511.522.5 Mean = 0.963Stdev = 0.224 Figure 428. Frequency of of the proposed method for Louisiana soil 106 PAGE 107 0.80.820.850.870.90.920.940.970.991.011.041.061.091.110246810 Mean = 0.963Stdev = 0.042 Figure 429. Frequency of the sample means using Bootstrap method for Louisiana soil (100,000 resampling runs, the proposed method) 0.10.120.140.150.170.190.210.220.240.260.280.290.310510 Mean = 0.219Stdev = 0.027 Figure 430. Frequency of the sample standard deviations using Bootstrap method for Louisiana soil (100,000 resampling runs, the proposed method) 107 PAGE 108 00.310.620.921.231.541.852.152.462.773.083.383.69400.511.5 Mean = 1.327Stdev = 0.693 Figure 431. Frequency of of the Schmertmann method for Florida soil 0.911.11.21.31.41.51.61.71.81.922.112.2100.511.522.5 Mean = 1.328Stdev = 0.148 Figure 432. Frequency of the sample means using Bootstrap method for Florida soil (100,000 resampling runs, Schmertmann method) 108 PAGE 109 0.130.230.330.430.530.630.730.830.931.031.131.231.331.430123 Mean = 0.631Stdev = 0.243 Figure 433. Frequency of the sample standard deviations using Bootstrap method for Florida soil (100,000 resampling runs, Schmertmann method) 109 PAGE 110 CHAPTER 5 CONCLUSION Conclusion A total of 21 cases (load test data with CPT close to it) in Florida and 28 cases in Louisiana have been used to assess LRFD resistance factor, for 14 pile axial capacity prediction methods based on CPT results. One of the best methods, the Philipponnat method, was chosen and modified to form the proposed UF method. Ten field sites were revisited with paired CPT and DTP tests performed. This was an attempt to develop identifiers for cemented sand, which is the one of the most challenging soil types for geotechnical engineers. Based on the analysis of the test data and resistance factors of each method, the following conclusions can be drawn: Current pile axial capacity prediction methods using CPT results overestimate pile capacities in cemented soil. This issue needs to be addressed with additional testing. The proposed procedure for separating skin friction and tip resistance is a useful tool to better evaluate the pile capacity prediction methods. Of the fourteen pile axial capacity prediction methods analyzed, the Philipponnat method, LCPC method and MTD method were shown to have the highest value, indicating more accurate results. The proposed UF method has the highest for Florida soil and second highest for Louisiana soil (0.62 for Florida soil and 0.67 for Louisiana soil) among all the methods. In addition, it has the lowest coefficient of variation for Florida soil and second lowest for Louisiana soil (0.27 for Florida soil and 0.23 for Louisiana soil). Therefore, the proposed UF method works very well for both sands (Florida soil) and clays (Louisiana soil). (Hu et al. 2007a) The dual tip penetrometer can be a potential tool to identify cemented soil by using the identifier, T2/T1. The ranges of T2/T1 for different soil type are shown below, based on limited test data (10 cases) and more data need to be collected to confirm these figures. (Hu et al. 2007b) 0.5 NonCemented Sand (Loose and Medium Dense Sand) = 0.3~0.5 Lightly Cemented Sand 1 Strongly Cemented Sand, 110 PAGE 111 = 0.5~1 Silty Sand, Sandy Silt and Silty Clay < 0.5 Highly Sensitive Clay > 1 Dense Silty Sand 1 Clayey Silt, Silty Clay and Clay Using the bootstrap method on field data shows that the standard deviation of the resampled mean and resampled standard deviation are quite small for the proposed method. This means that the proposed method provides higher quality predictions. Future Work There still needs work to confirm the above conclusions. For example: Since the sleeve friction reading of DTP is not identical to CPT and it hinders the ability of DTP to identify soil type, a redesign of DTP is recommended to eliminate the influence of the second tip to the sleeve friction. More bridge sites should be revisited and tested with both the CPT and DTP. Table 51 shows the bridge sites where load test data are available. Validate the cemented sand identification values from the DTP (T1/T2), and compare them with SPT & CPT data. Evaluate the proposed pile capacity method based on a larger CPT, DTP, and load test database. This will provide confidence in recommending a realistic LRFD resistance factor, and in turn further the stage of geotechnical engineering practice. 111 PAGE 112 Table 51. Bridge sites where load test data are available Cemented Sand SiteLoad Test No.Site NameProject NumberPier (Bent/Slab) No.StationPile Embeded Length (ft)Desired CPT Depth (ft)1T1 N 1533290 E63146146532T6 N 1533447 E63154167723T7 N 153325 E63156846.2534T14 N 1533572 E63161736435TP2 N /A33386TP10 N /A63687Siesta Key Sarasota N /A N /A N /A16.320.383A N /A36.54493B N /A4653103C N /A606511B20 N 2909 W6492647012B21 N 4080 W6905465313B4 ( Pier4 ) 264+27.2352.65914B9 ( Pier9 ) 271+25.6745.85215C2 ( Pier2 ) 1981+87.414349NonCemented Sand SiteLoad Test No.Site NameProject NumberPier (Bent/Slab) N o.StationPile Embeded Length (ft)Desired CPT Depth (ft)1TP1240+1041.2502TP38378+2923.6323Pile #2 N /A64684Pile#3 N /A49.2585Pile #4 N /A23.8306Beaches of LongboatN/AN/AN/A47.3547I275, 34th Street Pinellas N /A N /A N /A103.61148I295/103rd StreetN/APier 1R216+14.2585.7949I295/CSXN/ABent 2RN/A61.37010I295/I10N/APier 1494+0053.46411I295/Melvin RDN/APier 2E244+0589.310012I295/MEM. PKN/ABent 2W429+3580.89113I295/Ortega RiverN/APier 3R64+80809014Julington Cree k N /ABent 3245+00616715 N W Conn. OC Retrofit N /APile #2 N /A52.361Blount Island Marine TerminalN/ACape Canaveral891199.1St. John's River(ASCE)N/AMarco Island7661096801Ballona CreekN/AWest Palm Beach I9593220347349th Street BridgeN/A 112 PAGE 113 APPENDIX MATHCAD PROGRAM Pile Capacity Prediction Methods Schmertmann Method (1978) 113 PAGE 114 114 PAGE 115 115 PAGE 116 De Ruiter and Beringen Method (1979) 116 PAGE 117 117 PAGE 118 118 PAGE 119 119 PAGE 120 Penpile Method (1980) 120 PAGE 121 121 PAGE 122 122 PAGE 123 Prince and Wardle Method (1982) 123 PAGE 124 124 PAGE 125 Tumay and Fakhroo Method (1982) 125 PAGE 126 126 PAGE 127 127 PAGE 128 Aoki and De Alencar Method (1975) 128 PAGE 129 129 PAGE 130 Philipponnat Method (1980) 130 PAGE 131 131 PAGE 132 LCPC (Bustamante and Gianeselli) Method (1982) 132 PAGE 133 133 PAGE 134 134 PAGE 135 135 PAGE 136 136 PAGE 137 Almeida et al. Method (1996) 137 PAGE 138 138 PAGE 139 139 PAGE 140 MTD (Jardine and Chow) Method (1996) 140 PAGE 141 141 PAGE 142 142 PAGE 143 143 PAGE 144 Eslami and Fellenius Method (1997) 144 PAGE 145 145 PAGE 146 Powell et al. Method (2001) 146 PAGE 147 147 PAGE 148 UWA05 Method (2005) 148 PAGE 149 149 PAGE 150 150 PAGE 151 151 PAGE 152 Zhou et al. Method (1982) 152 PAGE 153 153 PAGE 154 Calculation of LRFD Resistance Factor 154 PAGE 155 155 PAGE 156 Bootstrap Analysis Schmertmann Method (Florida Soil) 156 PAGE 157 157 PAGE 158 158 PAGE 159 159 PAGE 160 The Proposed Method (Florida Soil) 160 PAGE 161 161 PAGE 162 162 PAGE 163 163 PAGE 164 LIST OF REFERENCES AASHTO. (1996/2000). LRFD Bridge Design Specifications. 2nd LRFD Edition2000 Interim Revisions, AASHTO, Washington, DC. Ahmadi, M.M., Byrne, P.M., and Campanella, R.G. (2005). Cone tip resistance in sand: modeling, verification, and applications. Can. Geotech. J., 42, 977. Almeida, M.S.S., Danziger, F.A.B., and Lunne, T. (1996). Use of the piezocone test to predict the axial capacity of driven and jacked piles in clay. Can. Geotech. J., 33 (1), 2341. Aoki, N., and De Alencar, D. (1975). An approximate method to estimate the bearing capacity of piles. Proc., 5th PanAmerican Conf. of Soil Mechanics and Foundation Engineering, Buenos Aires, Vol. 1, 367. Baldi, G., Bellotti, R., Ghionna, V., Jamiolkowski, M., and Pasqualini, E. (1986). Interpretation of CPTs and CPTUs in drained penetration of sands. 4th International Geotechnical Seminar, Field Instrumentation and In Situ Measurements, Part 2, Singapore, pp. 143156. Briaud, J.L. Introduction to soil moduli. Dept. of Civil Engineering, Texas A&M Univ., College Station, TX, USA. < http://ceprofs.civil.tamu.edu/briaud/Intro%20to%20Moduli/ Intro%20to%20Moduli.pdf>, Sep. 21, 2007. Bustamante, M., and Gianeselli, L. (1982). Pile bearing capacity predictions by means of static penetrometer CPT. Proc., 2nd European Symp. on Penetration Testing, ESOPTII, Amsterdam, The Netherlands, Vol. 2, 493. Campanella, R. G., and Robertson, P. K. (1988). Current status of the piezocone test. Proc., 1st Int. Symp. on Penetration Testing, ISOPT1, Orlando, FL, Vol. 1, 93. Clisby, M. B., Scholtes, R. M., Corey, M. W., Cole, H. A., Teng, P., and Webb, J. D. (1978). An evaluation of pile bearing capacities. Final Report, Mississippi State Highway Department, Volume 1. Clough, G.W., Sitar, N., Bachus, R.C. and Rad, N.S. (1981). Cemented sands under static loading. J. of the Geotech. Engrg. Division, ASCE, 107 (GT6), 799817. Davisson, M. T. (1972). High capacity piles. Proc., Lecture Series on Innovation in Foundation Construction, American Society of Civil Engineers, ASCE, NY, 81. De Ruiter, J., and Beringen, F. L. (1979). Pile foundations for large North Sea structures. Mar. Geotech., 3(3), 267. Durgunoglu, H.T., Mitchell, J.K. (1975). Static penetration resistance of soils 1 analysis. Proc. of the Conf. on Insitu Measurement of Soil Properties, 151 171. Eslami, A., and Fellenius, B. H. (1997). Pile capacity by direct CPT and CPTU methods applied to 102 case histories. Can. Geotech. J., 34, 886. 164 PAGE 165 Florida Department of Transportation. (2007). Standard specifications for road and bridge construction 2007. Sec. 4552.2.1. Tallahassee, FL. PAGE 166 Price, G., and Wardle, I. F. (1982). A comparison between cone penetration test results and the performance of small diameter instrumented piles in stiff clay. Proc. of the 2nd European Symp. on Penetration Testing, Amsterdam, The Netherlands, Vol. 2, 775. Puppala A.J., Acar, Y.B. and Tumay, M.T. (1995). Cone penetration in very weakly cemented sand. J. Geotech. Engrg., Vol. 121, No. 8, 589600. Puppala, A.J., Arslan, S., Tumay, M.T. and Yalcin, B.A. (1998). Cone penetration testing in cemented soils: Comparisons between field and laboratory chamber test results. Proc. of the 1st International Conf. on Site Characterization, Atlanta, GA. Rad, N.S., and Tumay, M.T. (1986). Effect of cementation on the cone penetration resistance of sand. Use of In Situ Tests in Geotech. Engrg, ASCE, 926948. Robertson, P.K., Campanella, R.G., Gillespie, D. and Grieg, J. (1986). Use of piezometer cone data. Proc., insitu '86, ASCE Specialty conf., Blacksburg, VA, United States, 1263. Schmertmann, J.H. (1978a). Guidelines for cone penetration test, performance and design. Rep. No. FHWATS78209, U.S. Department of Transportation, Washington, D.C., 145. Styler, M.A. (2006). Development and implementation of the DIGGS format to perform LRFD resistance factor calibration of driven concrete piles in Florida. Masters thesis, Univ. of Florida, Gainesville, FL. Titi, H. H., and AbuFarsakh, M. Y. (1999). Evaluation of bearing capacity of piles from cone penetration test data. Rep. No. FHWA/LA.99/334, Louisiana Transportation Research Center, Baton Rouge, LA. Tumay, M. T., and Fakhroo, M. (1982). Friction pile capacity prediction in cohesive soils using electric quasistatic penetration tests. Interim Research Rep. No. 1, Louisiana Department of Transportation and Development, Research and Development Section, Baton Rouge, LA. Zhou, J., Xie, Y., Zuo, Z.S., Luo, M.Y. and Tang, X.J. (1982). Prediction of limit load of driven pile by CPT. Proc. of the 2nd European Symp. on Penetration Testing, Amsterdam, The Netherlands, Vol. 2, 957961. 166 PAGE 167 BIOGRAPHICAL SKETCH Zhihong Hu was born in Kaifeng, Henan Province, China. He spent his childhood in that small, beautiful city and finished his primary school and middle school there. He moved to Zhengzhou with his parents and studied his high school. He was accepted by Civil and Architectural Engineering, Shanghai Jiaotong University in 1996 and spent 4 years in this university. He got his two bachelors degrees (civil engineering and applied electrical engineering) in September 2000. He found a job in a construction company and gained his first working experience there. After some time in the work, he realized that his knowledge was far from enough to deal with the real work problem. So he decided to go abroad to get advanced education. He was accepted by the Department of Civil and Coastal Engineering in the University of Florida and went to the US in January 2002. He had studied as well as being doing centrifuge research supervised by Dr. Michael McVay for two years. The study and research strengthened his background in geotechnical engineering and made him decide to devote the rest of his life in this field. He got his masters degree in December 2003 and continued his Ph.D. in this school in the following 4 years. During these 4 years, he not only finished another FDOT project and his Ph.D. degree, but also met his wife, Meiyu, and got married in May 2006. 167 