REAL TIME CONSTRUCTION MONITORING FOR DRILLED SHAFTS SOCKETED INTO FLORIDA LIMESTONE By MICHAEL BLAKE RODGERS A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUI REMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2016
2016 Michael Blake Rodgers
To t he greatest engineer I know, Papa, I dedicate this work to you. Thank you for being such an in sp iration and positive role model To my mom, N ana, and fathers, thank you for providing me with a life filled with love and supp ort. To my brothers, sister, and cousins, thank you for all the laughter and for keeping my spirits high. To the love of my life, thank you for being so patient through this long journey and for being my rock, we finally did it! To my best friends, thank you for always being there for me and keeping me in line, you getting emotional right now. T o my lord and savior thank you for giving me the strength and guidance to achieve my dreams and for the gift of life, a gift I cherish every day. To my Aunt, thank you for having faith in me and helping me begin th is journey.
4 ACKNOWLEDGMENTS central Geotechnical Engineers is greatly appreciated: David Horhota, Jose Hernando, Rodrigo Herrera, Juan Castellanos, and Larry Jones. A special thanks is extended to all the FDOT participants that assisted in the development and production of Gatorock: Dan Pitocchi, John Shoucair, Richard Delorenzo, Patrick Carlton, Patrick Gallagher, Thomas Frank, and Dale Deford. To the FDOT S tate M ate rials O ffice field technicians: Bruce Swidarski, Todd Britton, Kyle Sheppard, and Travis Stevens; thank you for your tireless efforts and for being my eyes and ears throughout the years. I can never thank you enough for all you have done for me and helpi ng make this research a reality. You are all truly amazing! I would also like to thank the University of Florida research team: Jon Sinnreich, Phillip Rodgers, Jerry Paris, Jordan Nelson, Shelby Brothers, Katie Maslak, Mike Stephenson, Tim Copeland, Matt Andrews, Aaron Hendricks, Michael Ferguson, Richard Booze, Ryan Mackey, and Vik Adams. Without your tremendous efforts, this research would not have been possible Caitlin Tibbetts and Stephen Crawford, words cannot describe how thankful I am f or you bo th I literally could n ot have done it without you. To my committee members: Michael McVay, Christopher Ferraro, Ana Mohseni, Esther Obonyo, and Timothy Townsend; thank you for your guidance and support. It has been a true blessing being mentored by ea ch and every one of you. Thank you for helping me grow as an engineer and person, I am forever grateful.
5 TABLE OF CONTENTS page ACKNOWLEDGMENTS ................................ ................................ ................................ .. 4 LIST OF TABLES ................................ ................................ ................................ ............ 7 LIST OF FIGU RES ................................ ................................ ................................ .......... 8 ABSTRACT ................................ ................................ ................................ ................... 14 CHAPTER 1 INTRODUCTION ................................ ................................ ................................ .... 16 2 LABORATORY MEASURE MENTS OF DRILLING PARAMETERS ON SYNTHETIC LIMESTONE ................................ ................................ ...................... 21 Development of Synthetic Limestone, Gatorock ................................ ..................... 21 Development of a Laborato ry Drilling Environment ................................ ................. 25 Coupler System to Monitor Laboratory Drilling ................................ ................. 26 Laboratory Coupler Calibration ................................ ................................ ......... 29 Laboratory Drilling Procedure ................................ ................................ ........... 34 Lab Data Analyses ................................ ................................ ................................ .. 42 q u vs. q t ................................ ................................ ................................ ............. 42 Effects of Bit Diameter ................................ ................................ ...................... 43 Increasing Force and Torque Relationship ................................ ....................... 57 Torque a nd Force vs. Penetration Rate per Rotational Speed Ratio (u/N) ....... 59 Torque and Crowd vs. Compressive and Tensile Strengths ............................. 62 Deve loped D s vs. q u and D s vs. q t ............ 66 Developed e vs. q u and e vs. q t ...................... 68 Comparisons Based on Rotational Speed ................................ ........................ 71 Comparisons Based on Penetration Rates ................................ ....................... 74 3 DRILLED SHAFT FIELD MONITORING TO MEASUR E COMPRESSIVE STRENGTH ................................ ................................ ................................ ............ 81 Development of Equipment for Shaft Installation Monitoring in Real Time ............. 81 Surveying Florida Contracto rs and District Geotechnical Engineers ................ 81 Monitoring Equipment ................................ ................................ ...................... 8 3 Field Monitoring Equipment Setup and Installation ................................ ................. 86 Analysis of Rock Strength from Real Time Field Monitoring ................................ 103 4 DRILLED SHAFT FIELD MONITORING TO MEASURE SHAFT CAPACITY ...... 127 Development of the Florida Geomaterials Equation ................................ ............. 136 Distinguishing Rig Malfunction from Encountered Voids ................................ ....... 148
6 Comparative Skin Friction Analysis ................................ ................................ ...... 153 5 CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE WORK ................. 158 Conclusions ................................ ................................ ................................ .......... 158 Recommendations ................................ ................................ ................................ 159 Summary of Research ................................ ................................ .......................... 159 LIST OF REFERENCES ................................ ................................ ............................. 165 BIOGRAPHICAL SKETCH ................................ ................................ .......................... 167
7 LIST OF TABLES Table page 2 1 Compressi ve strength results for Gatorock cores and cast cylinders. .................... 25 2 2 Instron calibration results. ................................ ................................ ...................... 31 2 3 Equation comparison and results. ................................ ................................ .......... 33 2 4 Torque calibration results showing offsetting axial force values, Ch 2 and 4. ........ 34 2 5 Gatorock projected streng ths (20 RPM). ................................ ................................ 40 2 6 Gatorock actual strengths (20 RPM). ................................ ................................ ..... 40 2 7 Gatorock projected strengths (40 RPM). ................................ ................................ 40 2 8 Gatorock actual strengths (40 RPM). ................................ ................................ ..... 40 3 1 Summary of the drilled shaft survey results. ................................ .......................... 82 3 2 Field monitoring system comparison. ................................ ................................ ..... 84 4 1 Drilled Shaft design skin friction equations. 13 21 ................................ .................... 130 4 2 Prelim inary skin friction comparative analysis at Little River. ............................... 135 4 3 Unit side shear bias analysis summary of statistics. ................................ ............ 157
8 LIST OF FIGURES Figure page 1 1 Spatial profile of boring rock strength at 17 th Street Bridge Fort Lauderdale. ......... 17 2 1 Gatorock mix design c/a vs. q u plot. ................................ ................................ ....... 24 2 2 Instrumented drill rod (Photo courtesy of author). ................................ .................. 28 2 3 Laboratory coupler system (A) and laboratory drilling set up (B). .......................... 28 2 4 Coupler system drilling into a large Gatorock block (Photo courtesy of author). .... 29 2 5 Instron calibration setup (Photo courtesy of author). ................................ .............. 30 2 6 Axial calibration, applied loads vs. measured loads, zero intercept equation. ........ 32 2 7 Axial calibration, applied loads vs. measured loads, intercept equation. ................ 32 2 8 Torque calibration with a moment arm 16 inches in length (Photo courtesy of author). ................................ ................................ ................................ ............... 33 2 9 Water circulation system to represent wet hole shaft installations (Photo courtesy of author). ................................ ................................ ............................. 36 2 10 Sinusoidal wave pattern of real time readi ngs (Photo courtesy of author). .......... 37 2 11 Torque and crowd vs. depth for a single laboratory drilling. ................................ 38 2 12 q u vs. q t plot wi th linear and 2 nd order polynomial curve fitting. ............................ 43 2 13 Torque vs. unconfined compressive strength (N = 20 rpm, u = 0.16 in/min). ....... 44 2 14 Crowd vs. unconfined compressive strength (N = 20 rpm, u = 0.16 in/min). ........ 44 2 15 Torque vs. splitting tensile strength (N = 20 rpm, u = 0.16 in/min). ...................... 45 2 16 Crowd vs. splitting tensile strength (N = 20 rpm, u = 0.16 in/min). ....................... 45 2 17 Torque vs. unconfined compressive strength (N = 20 rpm, u = 0.28 in/m in). ....... 46 2 18 Crowd vs. unconfined compressive strength (N = 20 rpm, u = 0.28 in/min). ........ 46 2 19 Torque vs. splitting tensile str ength (N = 20 rpm, u = 0.28 in/min). ...................... 47 2 20 Crowd vs. splitting tensile strength (N = 20 rpm, u = 0.28 in/min). ....................... 47 2 21 Torqu e vs. unconfined compressive strength (N = 20 rpm, u = 0.40 in/min). ....... 48
9 2 22 Crowd vs. unconfined compressive strength (N = 20 rpm, u = 0.40 in/min). ........ 48 2 23 Torque vs. splitting tensile strength (N = 20 rpm, u = 0.40 in/min). ...................... 49 2 24 Crowd vs. splitting tensile strength (N = 20 rpm, u = 0.40 in/min). ....................... 49 2 25 Torque vs. unconfined compressive strength (N = 40 rpm, u = 0.32 in/min). ....... 50 2 26 Crowd vs. unconfined compressive strength (N = 40 rpm, u = 0.32 in/min). ........ 50 2 27 Torque vs. splitting tensile strength (N = 40 rpm, u = 0.32 in/min). ...................... 51 2 28 Crowd vs. splittin g tensile strength (N = 40 rpm, u = 0.32 in/min). ....................... 51 2 29 Torque vs. unconfined compressive strength (N = 40 rpm, u = 0.56 in/min). ....... 52 2 30 Crowd vs. unconfined compressive strength (N = 40 rpm, u = 0.56 in/min). ........ 52 2 31 Torque vs. splitting tensile strength (N = 40 rpm, u = 0.56 in/min). ...................... 53 2 32 Crowd vs. splitting tensile strength (N = 40 rpm, u = 0.56 in/min). ....................... 53 2 33 Torque vs. unconfined compressive strength (N = 40 rpm, u = 0.80 in/m in). ....... 54 2 34 Crowd vs. unconfined compressive strength (N = 40 rpm, u = 0.80 in/min). ........ 54 2 35 Torque vs. splitting tensile str ength (N = 40 rpm, u = 0.80 in/min). ...................... 55 2 36 Crowd vs. splitting tensile strength (N = 40 rpm, u = 0.80 in/min). ....................... 55 2 37 Torqu e plotted vs. unconfined compressive strength displaying the direct torque rotational speed ............ 57 2 curve fitting. ................................ .............. 58 2 ................................ ................. 58 2 ................................ ................................ ...................... 60 2 ................................ ................................ ....................... 60 2 ................................ ................................ ......................... 61 2 43 ................................ ................................ .......................... 61 2 44 Torque vs. q u with linear curve fitting. ................................ ................................ .. 62 2 45 Crowd vs. q u with linear curve fitting. ................................ ................................ ... 63 2 46 Torque vs. q t with linear curve fitting. ................................ ................................ ... 63
10 2 47 Crowd vs. q t with linear curve fitting. ................................ ................................ .... 64 2 48 D s vs q u plot. ................................ ................................ ................................ ......... 66 2 49 D s vs q t plot. ................................ ................................ ................................ ......... 67 2 s S e and D s (Sori granite, 50 RPM). 2 ........ 67 2 51 Specific energy vs. unconfined compressive strength. ................................ ......... 70 2 52 Specific energy vs te nsile strength. ................................ ................................ ...... 71 2 53 D s vs. q u (grouped by rotational speeds). ................................ ............................. 72 2 54 e vs. q u (grouped by rotational speeds). ................................ ............................... 72 2 55 D s vs. q t (grouped by rotational speeds). ................................ .............................. 73 2 56 e vs q t (grouped by rotational speeds). ................................ ................................ 73 2 57 D s vs. q u (grouped by penetration rates for 20 RPM drillings). ............................. 74 2 58 e vs. q u (grouped by penetration rates for 20 RPM drillings). ............................... 75 2 59 D s vs q u (grouped by penetration rates for 40 RPM drillings). .............................. 75 2 60 e vs q u (grouped by penetration rates for 40 RPM drillings). ................................ 76 2 61 D s vs q t (grouped by penetration rates for 20 RPM drillings). ............................... 76 2 62 e vs q t (grouped by penetration rates for 20 RPM drillings) ................................ 77 2 63 D s vs q t (grouped by penetration rates for 40 RPM drillings). ............................... 77 2 64 e vs q t (grouped by penetration rates for 40 R PM drillings). ................................ 78 2 65 D s vs. q u independent of drilling parameter groupings. ................................ ........ 79 2 66 e vs. q u independent of drilling paramet er groupings. ................................ .......... 79 2 67 D s vs. q t independent of drilling parameter groupings. ................................ ......... 80 2 68 e vs. q t independent of drilling paramete r groupings. ................................ ........... 80 3 1 Jean Lutz monitoring system (Photo courtesy of author). ................................ ...... 85 3 2 Tapping into the IMT depth sensor (Photo co urtesy of author). ............................. 87 3 3 Jean Lutz penetration rate sensor (Photo courtesy of author). .............................. 88
11 3 4 Rotational speed sensor (Photo courtesy of author). ................................ ............. 89 3 5 Tapping into the torque and crowd hydraulic lines (Photos courtesy of the author and Jean Lutz). ................................ ................................ ........................ 90 3 6 Junction box located in electrical compartment (Photo courtesy of author). .......... 90 3 7 DIALOG and B tronic both monitoring a shaft installation in real time (Photo courtesy of author). ................................ ................................ ............................. 91 3 8 External viewing of a monitored shaft installation via Bluetooth (Photo courtesy of author). ................................ ................................ ................................ ........... 92 3 9 Bauer BG 30 crowd specs from serial plate, max crowd and operating pressure (Photo courtesy of author). ................................ ................................ ................. 94 3 10 Bauer BG 30 specs from serial plate, showing model type and operating pressure (Photo courtesy of autho r). ................................ ................................ .. 94 3 11 Bauer BG 30 spec sheet, showing torque operating pressure and maximum crowd. ................................ ................................ ................................ ................. 95 3 12 Bauer BG 30 specs, showing mo del type and maximum torque. ......................... 95 3 13 Elevation vs. rotational speed. ................................ ................................ ............. 97 3 14 Rotational speed frequency distribution. ................................ .............................. 97 3 15 Elevation vs. penetration rate. ................................ ................................ .............. 98 3 16 Penetration rate frequency distribution. ................................ ................................ 98 3 17 Elevation vs. crowd. ................................ ................................ ............................. 99 3 18 Crowd raw data frequency distribution. ................................ ................................ 99 3 19 Crowd converted freque ncy distribution. ................................ ............................ 100 3 20 Elevation vs. torque. ................................ ................................ ........................... 100 3 21 Torque raw data frequency distribution. ................................ ............................. 101 3 22 Torque converted frequency distribution. ................................ ........................... 101 3 23 Little River q u frequency distribution. ................................ ................................ .. 105 3 24 Little River q u cumulative frequency distribution. ................................ ................ 10 5 3 25 Visualization of core data variability. ................................ ................................ .. 107
12 3 26 Kanapaha site investigation locations. ................................ ............................... 108 3 27 Seismic line EW 3 results for P wave, S wave, and Poisson ratio. .................... 109 3 28 Li mestone recovered at a depth of 30 feet with grey clay at the top of the spoon (Photo courtesy of author). ................................ ................................ .... 110 3 29 Highly weathered limestone at a depth of 30 feet (Photo courtesy of author) ... 111 3 30 Kanapaha q u frequency distribution. ................................ ................................ .. 112 3 31 Kanapaha q u cumulative frequency distribution. ................................ ................ 112 3 32 Dry unit weight vs. average q t values from 23 Florida sites. ............................... 115 3 33 Dry unit weight vs. average q u values from 23 Florida sites. .............................. 115 3 34 q u vs. q t from 23 Florida sites (data points from dry unit weight groupings). ...... 116 3 35 q u core data with additional q u values generated using Rodgers method. .......... 117 3 36 Kanapaha q u frequency distribution using Rodgers method. ............................. 118 3 37 Kanapaha q u cumulat ive frequency distribution using Rodgers method. ........... 118 3 38 Little River q u frequency distribution using Rodgers method. ............................. 119 3 39 Little River q u cumulative frequency distribution using Rodgers method. ........... 120 3 40 Overland q u frequency distribution. ................................ ................................ .... 122 3 41 Overland q u cumulative frequency distribution ................................ ................... 123 3 42 East west cross section A 12 ................................ ..... 124 3 43 Overl and elevation vs. q u plot, q u reported in psi. ................................ ............... 124 4 1 Drilled shaft load transfer diagram. ................................ ................................ ...... 127 4 2 Summary of stress states a t rock shaft interface. 13 ................................ .............. 131 4 3 Strength envelope for Florida limestone 13 ................................ ............................ 132 4 4 Little River q t /q u analysis using all bori ng locations with outliers removed. .......... 134 4 5 Core data from Little River indicating dissimilar materials. ................................ ... 136 4 6 q t /q u vs. q u for concrete. 23 ................................ ................................ ..................... 137 4 4 .................. 138
13 4 8 q t /q u vs. q u for clay an ................................ 140 4 9 q t /q u vs. q u developed using only Little River q u data. ................................ ........... 142 4 10 Comparison of q t /q u vs. q u curves. ................................ ................................ ..... 142 4 11 Comparison of q t /q u vs. q u curves with the new Florida geomaterials curve. ...... 144 4 12 Florida geom aterials q t equation. ................................ ................................ ....... 145 4 13 Split tension strength vs. moisture content with predicted q t values for 23 Florida sites. ................................ ................................ ................................ ..... 146 4 14 Dry unit weight vs. split tension strength with predicted and measured q t values from 23 Florida sites ................................ ................................ .............. 146 4 15 Test shaft depth vs. specific energy. ................................ ................................ .. 148 4 16 East shaft depth vs. specific energy. ................................ ................................ .. 149 4 17 Monitored readings indicating rig malfunction. ................................ ................... 151 4 18 Monitored readings indicating a void was encountered. ................................ ..... 152 4 19 Skin friction comparative analysis using methods 1 through 4. .......................... 154 4 20 Skin friction comparative analysis using methods 5 through 8. .......................... 154 4 21 Skin friction comparative analysis summary using McVay et al. 13 ...................... 156 4 22 Unit side shear comparison from all monitoring sites. ................................ ........ 156 4 23 Unit side shear bias analysis (Load Test / Monitoring). ................................ ...... 157
14 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 REAL TIME CONSTRUCTION MONITORING FOR DRILLED SHAFTS SOCKETED INTO FLORIDA LIMESTONE By Michael Blake Rodgers August 2016 Chair: Michael McVay Major: Civil Engineering This research focused on the compressive strength, q u through drilling parameters: crowd, torque, p enetration rate, rota tional speed, and bit diameter, in both the labo ratory and field for assessing drilled shaft capacity Synthetic homogeneous limestone blocks of various strengths were cast in the laboratory an d allowed to cure for 14 days; t he blocks were then drilled with rock augers of two different diameters at multiple penetration rates and drilling rotational speeds. Both crowd and torque were measured in real time. Estimation of rock strength, q u from drilling relationships developed by Karasaw a et al. 1,2 and Teale 3 were strength, q u using multiple drill bit diameters as expected in field drilled shaft construction. In the field, shaft installations were monitor ed at three sites from which rock strength, q u was estimated along with shaft side shear. For side recommended equation for limestone was used with split tension strength initially estimated from the q t /q u ratio established by field c ores; 4 q t /q u vs. q u relationship was used The predicted unit skin friction in multiple rock formations
15 was compared to measurements using conventional methods; provided by instrumented segment s from Osterberg, Statnamic, and top down s tati c load testing. The mean bias, measured/predicted, unit side shear in limestone was 1.00, and the coefficient of variation, CV, was 0.07. The results suggest that est imating rock strength and side shear using drilling parameters is viable, espec ially considering the available rig moni toring equipment
16 CHAPTER 1 I NTRODUCTION Foundation engineering is a discipline shifting from experience based design to a more analytical approach using science and math based principles. Generally, a foundation engineer has to consider multiple layers of soil and rock as well as uncertainty, due to site variability with limited in situ or laboratory data available for design. In the past 10 years, codes, such as Load and Resistance Factor Design, LRFD, have been develo ped based on reliability, which enable engineers to consider different resistance factors based on design and construction practices. For instance, dynamic monitoring or static load testing allows the engineer to improve the resistance factors for both dr iven piles and drilled shafts. For driven piles, the highest resistance factors that can be achieved are associated with monitoring, estimating the static capacity of every pile that is driven on a site. Unfortunately, in the case of drilled shafts the o nly construction process to improve resistance factors and reliability are with static or dynamic load testing of multiple shafts; as monitoring drilled shaft installations is currently limited to visual inspection of drilled debris. As a result, the stat ic capacity of every drilled shaft installed on the site cannot be estimated and spatial variability concerns are left unresolved. This is alarming since the use of drilled shafts in Florida has increased significantly over the past 20 years due to urban s ettings and associated noise and vibration issues encountered w ith driven pile installations. Of great int erest is the assessment of soil and rock strength properties during the drilled shaft construction process, i.e. drilling, to account for subsurface variability. For instance, shown in Figure 1 1 is the mean and standard deviation of rock strength within individual borings at a drilled shaft site in Fort Lauderdale, Florida. Based on the
17 obtained core data, a shaft located at coordinates, x = 9, y = 24, would have nearly half the axial capacity of the shaf t located at, x = 18, y = 11 approximately 16 feet away. Figure 1 1 Spatial profile of boring rock strength at 17 th Street Bridge Fort Lauderdale T ypically, core strengths used for design are rarely recovered within the footprint of each shaft unless the elements are non redundant. As a re sult, the variability of rock strength within Figure 1 1 may be characterized with a reduced resistance factor, or through an increased number of bori ngs closer to each foundation, the latter being less likely. For the case of non redundant shafts, coring within the footprint of each shaft is required, yet only visual inspection of the cores typically occurs. However, if the strength of the soil/rock could be assessed in real time by monitoring the drilling greatly reduced. This would provide quality assurance to the foundation engineer and the drilling contractor. In addition, const ruction costs could be decreased by r educing the time of completion, e.g. elimina ting non redundant shaft coring, and through the use of higher 0 5 10 15 20 25 0 2 4 6 8 10 12 14 16 18 20 y (ft) x (ft) ( ) ( ) ( ) ( ) ( ) ( )
18 LRFD resistance factors leading to a reduction in shaft length, by accounting for the uncertainty and variabil ity of in situ soil/r ock conditions for every shaft. The assessment of rock strength during drilling is an active research area in the on controlling the drilling pro cess to limit damage and improve drilling efficiency, which fields have been developed to determine rock strength from drilling parameters. These include equation develo pment by Wolcott and Bordelon 5 Hoberock and Bratcher 6 Teale 3 and Karasawa et al. 1,2 as well as several others. However, many of these methods require specific roller bit coefficients that are not applicable to drilled shaft installations where a rock a uger is most often used; or they include drilling parameters that may be problematic to continuously monitor in real time such as drilling mud viscosities and densities. Of all the developed drilling equations investigated, only Karasawa et al. 1,2 and Teal e 3 employ parameters that are applicable to monitoring drilled shaft installations with relative ease. Both methods showed that in situ rock strength or compressive strength, q u is correlated to the following five drilling parameters: 1. Crowd or downward a xial force, F 2. To rque applied to the drilling tool T 3. Vertical penetration rate, u 4. Rotational speed of the drilling tool, N 5. Drilling tool diameter d Using either of the proposed methods does not require the use of bit specific coefficients or additiona l drilling parameters to assess the compressive strength of rock in real time. Moreover, an investigation of monitoring capabilities of current drilled shaft
19 rigs revealed that many newer rig types are equipped with sensors to monitor these five drilling parameters and monitoring equipment is readily available for rig types without built in sensors. Therefore it is proposed that continuous monitoring of the drilling process to assess rock strength is a viable option for reducing spatial variability conce rns and provides a means to quantify the quality and length of rock sockets. This will ensure the as built foundation meets or exceeds the engineering design, providing quality assurance to the drilling contractor and foundation engineer. As well, the pr oposed methods will take the first steps towards eliminating spatial variability concerns; which will ultimately lead to the use of increased resistance factors for d rilled shaft design, creating a more efficient and cost effective construction process. In order to measure rock strength in real time to assess shaft capacity, the following three tasks needed to be completed: 1. Develop relationships for rock strength with the monitored drilling parameters in a laboratory drilling environment. 2. Monitor the drill ing process in the field using the same drilling parameters recorded from the drill rig. 3. Validate the monitored soil/rock strength estimates using both field/laboratory core testing and load testing as the basis for comparison During the drilling process, monitoring will provide real time compressive strength estimates, which can be used to ensure the material being drilled meets the expectations of the shaft design It is important to identify monitoring resistance in terms of compressive strength as thi s is the conventional strength property used for estimating drilled shaft capacities. This approach also provides the option to use numerous drilled shaft design methods to estimate shaft capacity during the installation
20 process. This is important because engineers may use different methods to estimate shaft capacity and nearly all of the methods rely on compressive strength data.
21 CHAPTER 2 LABORATORY MEASUREMENTS OF DRILLING PARAMETERS ON SYNTHETIC LIMESTONE Development of Synthetic Limestone, Gatorock To develop a relationship between monitored drilling parameters and rock strength, a series of laboratory drilling tests had to be conduct ed on large homogenous compressive strength s typical of F lorida bedrock. Unfortunately, homogeneity is not a characteristic of Florida limestone. Of concern was the high degree of variab ility of natural Florida limestone over small distances, e.g., Figure 1 1 Therefore, it was necessary to develop a homogeno us drilling medium cast into large scale blocks at various design strengths for testing In past research at the University of Florida, UF synthetic limestone, known as of laboratory experim ents including: B ullock 7 McVay et al. 8 and Sheppard et al. 9 Gatorock provides a simplistic method to model natural Florida limestone with the ability to control the desired compressive strength. Also, by using a synthetic limestone it enables the cr eation of a homoge nous formation with a wide range of des ign strengths typical of Florida li mestone For this research the Gatorock needed to be created in a simplistic manner and easily repeatable for various compressive strengths. It also needed to be developed at 14 d ay strength to provide a quick turnaround for laboratory drillings as well as maintain moisture within the synthetic rock, which is typical of in situ limestone. Gatorock mixtures generally c onsist of limesto ne screenings portland cement and water. Lime stone screenings, are a fine aggregate with 100% passing a #4 sieve, and are ideal for use in Gatorock production as the small er well graded aggregate sizes,
22 <4.75 mm, provide a homogenous mixture. This eliminates the risk that a single larger sized piece of aggregate will greatly influence the overall material properties of the synthetic rock mass. Florida p ortland ce ment combined with water is used as the cementitious material within the synthetic rock matrix, drastically speeding up the natural limesto ne bonding process which typically occurs over thousands to millions of years. Florida Portland cement is an ideal substitute to natural bonding agents as the main ingredient of Portland cement, calcium oxide, CaO is derived from limestone. In the natur al limest one formation process, calcite precipitate, CaCO 3 accumulates to form a carbonate matrix, which holds together larger carbonic sedimentary rocks to form a limestone rock mass. As the calcite precipitate accumulates, inclusions of sand, clay and organic matter are deposited within the matrix. T his produces a carbonate matrix which often includes impurities such as iron, silica, magnesium and in some c ases aluminum These impurities are the additional chemica l ingredients found in Florida p ortl and ceme nt, reinforcing the concept of p ortland cement being an ideal replacem ent to natural bonding agents. The development of Gatorock for this research began using guidelines of a Controlled Low Strength Material, CLSM, reported by ACI Committee 229. C LSM guidelines were used because compressive strengths of a CLSM are relatively low, 1 ,200 psi or less, and Gatorock design strengths for this research were to be 140, 280, 556 and 1,667 psi, which is representative of soft weathered, medium and strong F lorida limestone. Conventional C LSM mixtures consist of water, p ortland cement, fine or coarse aggregate or both and fly ash or similar products. However, the use of standardized materials i s not necessary. ACI 229 10 states, the selection of materials
23 s hould be based on availability, cost, specific application and the necessary characteristics of the mixtur e including flowability strength ex cavatability, and density These characteristics can be defined as slump, compressive strength, drillability, a nd unit weight, respectively. Since limestone screenings are generally considered a waste material, ACI 229 provided a standard method of using non standard materials to develop a homogenous drilling medium that is representative of Florida limestone. The total unit weight of a CLSM ty pically ranges from 115 145 pounds per cubic foot, pcf, which is within the typica l range of Florida limestone recovered in core samples throughout the state. The Gatorock developed using CLSM guidelines produced dry unit we ights ranging from approximately 95 pcf to 125 pcf, which is within range of typical Florida limestone i.e., 95 165 pc f Additionally, the properties of SMs, comp r essive strength of 50 100 psi, exhibit characteristic properties of soils and equate to an allowable bearing capacity of a wel l compacted soil 10 This is beneficial as field drilling will pass through several layers of varying mat erial, sand, clay, limes tone and intermediate geo material, IGM, and laboratory drillings will only focus on drilling into materi al representative of limestone. However, it is possible that lower strength Gatorock could be representative of some higher compacted in situ soils en countered in field drillings such as over consolidated clays and low strength IGM As is typical with CLSM mixture propor tioning, the development of Gatorock for this research was created through a trial and error process. Since proportions and guideline s for creating 14 day Gatorock did not exist, trial mixtures were created and evaluated on th e basis of strength, slump, and unit weight. It was found that by holding
24 the water to aggregate ratio, w/a, at 17%, the desired slump of six to eight inches coul d be achieved ; and that using variable cement to aggregate ratios, c/a, provided an excellent trend with unconfined com pressive strength, Figure 2 1 Therefore a mix design spr eadsheet was created that ensured the w/a = 17%, accounting for the aggregate saturated surface dry specific gravity and water content within the aggregate, and proportioned the mix based on the c/a ratio derived using the regression equation in Figu re 2 1 for each target strength. Figure 2 1 Gatorock mix design c/a vs. q u plo t Once all of the Gatorock mix designs were complete, a method for casting large scale Gatorock blocks, used for laboratory drilling was developed However, the target strength of the large scale blocks had to be verified to ensure the reported comp ress ive strengths matched those of test cylind ers cast from the same mix ; as the cast cylinders were to be used as the strength reference for each block For verification five blocks were cored using a four inch core barrel, each providing a core sample appr oximately y = 0.3751x 0.5366 R = 0.9913 0 5 10 15 20 25 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 2,000 2,200 Cement to Aggregate Ratio, c/a (%) Unconfined Compressive Strength, qu (psi)
25 eight inches long by four inches in diameter, matching the cast cylinder dimensions, and providing a 2:1 ratio whi ch is compliant with ASTM 11 standards for compression testing of field cores. The cores and cast cylinders from each Gatorock block were then subjected t o unconfined compression loading and compared based on their compressive strengths Table 2 1 Table 2 1 Compressive strength results for Gatorock cores and cast cylinders Cylinder 1 Cylinder 2 Cylinder 3 Cylinder 4 Cylinder 5 Cyli nder 6 348.0 334.2 341.0 341.1 381.4 10.57% 462.4 461.8 502.2 475.5 528.0 9.94% 520.6 490.9 445.9 485.8 489.7 0.80% 278.4 293.3 266.7 279.4 315.3 11.39 % 360.2 355.3 360.4 358.7 347.9 3.10% From the core results Table 2 it is evident that the majority of cores produced compressive strengths slightly higher than the cast cylinders. This was considered acceptable as the differences in strength were relatively low generally 10% or less, and inherently creat ed a conservative approach for estimating rock strength. This comp leted the Gatorock development and provided an accurate homogenous synthetic represent ation of Florida limestone used in all laboratory drilling and subsequent analyses Development of a La boratory Drilling Environment In addition to creating a homogenous drilling medium, instrumentation and monitoring equipment also had to be developed t o conduct the measured small scale drillings. This consisted of modifying an existing drill press to me et fiel d drilling standards, fabricating a coupler system to monitor torque and crowd in real time locating small scale drill bits that best represent the cu tting action of a rock auger
26 calibrating all of the equipment and developing a reliable drilling method to best represent field drilling Coupler System to Monitor Laboratory Drilling The majority of the work developing the laboratory drilling environment was designing and fabricating a coupler system to accurately monitor the applied forces, torque and crowd, in real time. The rotational speed and penetration rate were operator controlled parameters maintained by a variable frequency drive, VFD, installed on the drill press for the research The VFD provided rotational speeds and variable penetrati on rates representative of field conditions reported in a research conducted survey of drilled shaft rigs in Florida Since, torque and crowd were a byproduct of the rotational speed and penetration rate settings, both had to be continuously monitored and recorded in real time. To achieve this, two pairs of T element strain gauges and two pairs of torque rosettes were placed on a hollowed aluminum rod, connecting the drill bit to the drill press. The pairs of strain gauges and torque rosettes were setup i n a full bridge system to compensate for temperature effects with gau ges of each type oriented 180 degrees apart to compensate for bending. This provided alternating ga u ge types every 90 degrees as seen in Figure 2 1 All data was wirelessly sent through a transmitter mounted on a PVC sleeve with rubber padding to reduce vibrations The wall thickness of the a luminum rod Figure 2 1 was selected to withstand the applied drilling forces while providing a large enough strain range to reduce noise within the system, ensuring accurate readings. The length of the drill rod was designed to provide an undisturbed portion of the shaft two and a half times the rod diameter from e ach end to the gauges. This e liminate d edge effects, Saint provided uniform readings Finally, the full length of the coupler system was designed
27 to fit the available clearance and ensure the drill bit could reach a drilling depth of 20 inches into the blocks. Figure 2 2 shows the developed laboratory coupler system drill press, and block placement prior to drilling eter and two inch outer diameter constructed from an aluminum rod, fourteen inches in length. The base of the main shaft has male threading which attaches the connection collar. The connection collar connects the drill bit using a pin lock system similar to a drilled shaft rig Kelly bar connection. Between the collar and the drill bit is a steel washer that helps reduce wobbling of the bit at the pin lock connection. At the top of the main shaft a S ch. 80 PVC sleeve was mounted that allows the data tran smitter to be attached. Rubber compression padding was used between the main shaft and the PVC sleeve to reduce vibrations felt by the data transmitter which may disturb readings. The top four inches of t he main shaft have female inner threading that all ows the drill bit chuck to be threaded to the main shaft. The drill bit chuck connects the coupler to the drill press. The remaining ten inches are where monitoring takes place. This provides an undisturbed portion of the drill rod five inches in both d irections meeting the two and a half diameter spacing required to eliminate edge effects. A separate Sch. 80 PVC shield protects the gauge s from being da maged while drilling. Figure 2 3 shows the coupler in acti on, drilling 20 inches into a large Gatorock block. As depicted a dry hole drilling method is being used. This method was investigated but found to be unacceptable, as dry drilling typically produced a 70% increase in torque compared to using a wet hole method which is most often used in Fl orida according to survey results.
28 Figure 2 1 Instrumented drill r od (Photo courtesy of author) A B Figure 2 2 Laboratory coupler system (A) and laboratory drilling set up (B) A) Chuck connecting the coupler to the drill press. B) Wireless data transmitter mounted on PVC sleeve with rubber padding to reduce vibrations. C) PVC shield protecting strain gauges and torque rosettes. D) Alumin um drill rod where torque and crowd are monitored. E) Connection collar between the drill rod and the drill bit. F) Pin lock similar to drilled shaft Kelly bar drill bit connections. G) Spacer to reduce wobbling at the drill bit connection. H) Small s cale rock auger bit with rotating conical carbide teeth (Photos courtesy of author). Torque Rosettes T Element Stain gauges A C D E F G H B
29 Figure 2 3 Coupler system drilling into a large Gatorock block (Photo courtesy of author). Once the coupler system was constructed, equations for measuring torque and crowd in real time were developed for the data transmitter software. This required deriving a transfer function for the aluminum shaft that di rectly converted the strain gauge readings, wirelessly transmitted in bits, to measures of torque and crow d. After the transfer function equations were developed for both torque and crowd a thorough calibration process took place ensuring the readings obtained were highly accurate. Laboratory Coupler Calibration The calibration process i ncluded both axial and torsional loading to ensure measures of torque and crowd were highly accurate. To complete the axial loading calibration the Instron device, locate d on the UF campus, was used to provide consistent sustained loading, Figure 2 4
30 Figure 2 4 Instron calibration setup (Photo courtesy of author). Before loading w as applied, the drill rod level. However, much like the laborat ory drilling process, eccentric loading was expected. As a result, one side of the rod could experience tension while the other side exhibited compression. During laboratory drilling, this behavior was generally a result of the layout of the drill bit, i.e. line of carbide t eeth at different orientations when drilling began. However, by orienting the strain gauges 180 degrees apart, the effects of eccentric loading were eliminated through averaging. For the calibration using the Instron, a program was created to initiate constant loading for two minutes at each load step. The following loads were used: 50, 100, 250, 500, 1,000, and 1,500 lbs. Before and after each load step, a two minute resting period was initiated where no load was applied M easur ements were continuously recorded during the resting period to ensur e readings returned to the initial baseline
31 Averages of the before and after readings were then subtracted from the readings taken during the two minute loading period which provided the actual load measured. In the two minut e loading period, 960 readings, recorded at 8 Hz for 120 seconds, were captured with 800 of the readings used to create the average for each load step Table 2 1 shows the results from the Instron axial calibration. Table 2 1 Instron calibration results Loading Phase Ch 2 Ch 2 Balance Ch 4 Ch 4 Balance Final Average (lbf) Baseline 1 24.62 0.83 250 lbs 175.92 200.78 328.32 325.53 263.15 Baseline 2 25 .10 4.76 500 lbs 408.84 434.73 613.43 603.09 518.91 Baseline 3 26.69 15.93 1,000 lbs 790.00 816.82 1253.09 1237.90 1027.36 Baseline 4 26.95 14.45 1,500 lbs 1075.70 1105.89 2005.15 1987.24 1546 Baseline 5 33.43 21.37 100 l bs 21.16 54.56 171.72 150.57 102.56 Baseline 6 33.36 20.93 50 lbs 4.78 29.30 88.52 68.31 48.80 Baseline 7 34.80 19.49 As seen in the results of Table 2 1 the measured loads from the coupler did not perfectly a lign with the induced loads from the Instron. A load cell integrated into the Instron setup ensured loading was highly accurate with an error of 3 lb. Therefore, a slope adjustment equation was needed to correct the error of the coupler system. From t he collected data two equations were developed, one fitting all the data with a non zero intercept and the other with a zero intercept. Figures 2 6 and 2 7 display the plotted data as measured load v s. applied load for the two scenarios and provides the c alibration equations.
32 Figure 2 5 Axial calibration, a pplied loads vs. measured loads, zero intercept equation Figure 2 6 A xial calibration, a pplied loads vs. m easured lo ads, intercept equation As seen in the two plots, both show an R 2 = 1 confirming a linear fit using either equati on. B oth equations were then compared using the tabular data to determine which provided the best fit, Table 2 3. y = 1.0308x R = 1 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 0 200 400 600 800 1,000 1,200 1,400 1,600 Measured Axial Force, F m (lbf) Applied Axial Force, F a (lbf) y = 1.0302x + 0.6895 R = 1 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 0 200 400 600 800 1,000 1,200 1,400 1,600 Measured Axial Force, F m (lbf) Applied Axial Force, F a (lbf)
33 Table 2 2 Equation comp arison and results (Original Data) (y = 1.0302x + 0.6895) (y = 1.0308x) Applied Force (lbf) Measured Force (lbf) % Difference Adjusted Force (lbf) % Difference Adjusted Force (lbf) % Difference 0 0 0 0 0 0 0 50 48.8 2.40% 46.7 6.60% 47.34 5.31% 100 102.56 2.56% 98.89 1.11% 99.5 0.50% 250 263.15 5.26% 254.77 1.91% 255.29 2.12% 500 518.91 3.78% 503.03 0.61% 503.41 0.68% 1,000 1,027.36 2.74% 996.57 0.34% 996.66 0.33% 1,500 1,546.56 3.10% 1,500.56 0.04% 1,500.35 0.02% Sum 15.05% Sum 10.60% Sum 8.97% Average 2.51% Average 1.77% Average 1.49% From the tabular comparison, the equation with a zero intercept provided the least amount of error and was therefore the equation chosen to make the final adjust ments. Also investigated was the effect a torque load had on the axial force measurements. When applying an axial force the torque rosettes are designed to experience no loading and this was noticed during axial calibration. However, when a torque load was ap plied to the coupler system the axial strain gauge s react ed and this needed to be investigated to ensure the results were being r eported accurately. Figure 2 7 Torque calibration with a moment arm 16 inches in length (Photo courtesy of author).
34 Table 2 3 Torque calibration results showing offsett ing axial force values, Ch 2 and 4 M (in lbs) W (lbs) Ch 1 Ch 2 Ch 3 Ch 4 %Diff 1 3 %Diff 2 4 140.8 8.8 134.34 55.20 143.87 54.59 1.79% 1.10% 281 .6 17.6 283.02 101.77 283.22 99.32 0.07% 2.41% 422.4 26.4 423.09 145.16 422.82 139.65 0.06% 3.80% 563.2 35.2 561.30 186.70 560.20 183.56 0.20% 1.68% Comparing channels 2 and 4, axial force gauges, showed that the percent difference between t hem was very small, i.e. negl igible. Channel 2 is negative for compression, and channel 4 is positive for tension. Since the values were nearly equal for every torque loading, but opposite in sign, this suggested that even though the torque loading does provide axial loading, the two torque induced axial loads neg ated each other out on summing. From both analyses, the average error recorded for applied forces vs. measured forces, were 1.49% for crowd, Table 2 2 and 0.60% for t orque Table 2 3 Channels 1 and 3 This confirmed the derived transfer functions for both torque and crowd were now highly accurate, and the system was functioning properly Laboratory Drilling Procedure Once the calibration was complete, a standard laboratory drilling procedure was developed from a thorough investigation of how to drill in the laboratory. The investigation included comparing wet and dry drilling, investigating the number of drillings that could be obtained per block without disturbance, and the lengths of drill runs to prevent bit bite which causes spike s in torque and crowd readings. From the investigation, it was determined that two holes could be drilled into each block without disturbance, drill runs co uld only be four inches in depth before bit bite set in causing large spikes in readings and a wet hole drilling method was required as dry hole drilling produced a 70% increase in torque on average. In the field, wet hole
35 construction is almost always u sed in Florida and laboratory drillings needed to be represent ative of field conditions. Dry hole drilling was investigated to see if the laboratory drilling process could be simplified. However, the drastic increases in torque from dry drilling compared to wet drilling proved that using a wet hole drilling method in the lab was unavoidable. Therefore, a water circulation system was developed to best repre sent field drilling conditions. Figure 2 8 shows the water circulation sy stem developed to represent the wet hole constructio n method most often used in Florida drilled shaft installations. In the field as the drill bit is advanced, mat erial is collected on the drill bit and brought back to surface where the bit is removed from the hole an d the debris is spun off cleaning the bit for the next advancement. This process is constantly repeated throughout drilling as this is the only way to remove drilled debris from the hole. Unfortunately, this was not practical in the lab as t he process took nearly t en minutes every time the bit was removed for cleaning and added a significant amount of time to the already lengthy laboratory drilling process. Laboratory drilling times ranged from one to four hours depending on the penetration rate used. Therefore a water circulation system was designed that constantly removed water with drill ed debris in suspension from the hole while inject ing clean water. T he rotation of the bit kept the drilled debris in suspension for easy removal. This in combination with shorter drill runs greatly reduced /eliminated bit bite and provided consistent readings.
36 Figure 2 8 Water circulation system to represent wet hole shaft installation s (Photo courtesy of author). T he developed method for laboratory drilling is as follows: 1. Drill to a depth of 8 inches using the dry drilling method 2. Remove the bit from the hole 3. Remove debris from the bit and the hole (removed from hole via shop vac) 4. Reattach the cleaned bit 5. Lower the drill h ead and drill 4 inches with constant water circulation 6. Repeat steps 2 5 until desired drilling depth is reached (20 inches) During the drilling process 12 readings per revolution were recorded and used to provide an average torque and crowd value f or each depth increment which was determined by the penetra tion rate setting. Averages were taken for each revolution to compensate for bending effects. As stated earlier, gauges of each type were only placed in two locations 180 degrees apart. This l ed to bending that was not compensated for in every possible direction. Therefore, readings create d a sinusoidal wave pattern as t he bit rotated seen in Figure 2 9 Dirty effluent water Clean influent water
37 Figure 2 9 S inusoidal wave p attern of real time readings (Photo courtesy of author). By taking the average of a full rotation, peaks and valleys of the sine waves offset one another and provide d an average that accounted for a variable degree of bending during a full revo lution. The pink and red lines in Figure 2 9 are from the strain gauge s t hat record crowd or downward axial force It was noticed durin g drilling that the bit typically put one side of the rod in constant tension and the other in constan t c ompression, and that torque affected both axial force gauge s. However, it was confirmed in the c alibration process that torque e ffects are equally applied to both gauge s and offset one another. Additionally, axial force calibration confirmed that when av eraging the tension side and the compression side the opposite forces offset one another a nd the average value obtained is highly accurate. Again, b ecause the compressive forces are negative and tensile forces are positive the ir average produces the ac tual compressive force applied to the drill bit. For example, the red crowd line i s showing approximately 800 lb and the pink crowd line is sho wing approximately 600 lb When average d, the applied force is 100 lb indicating the drill ro d is being compr essed by 100 lb of force. The torque rosettes were designed to compensate for applied axial forces and only report the actual torque being applied. Therefore, both torque lines, blue and green, nearly plot on top of one another and are only separated by bending effects which are compensated for when an average is taken using both lines. Crowd Torque
38 After each drilling was complete, all of the averages obta ined at each depth increment were then combined to produce a final averag e that represented both torque and crow d for the entire length of the drilling. This resulted in hundr eds to thousands of data points obtained from averaging each rotation, making up the final average which was used in equation development. Figure 2 10 displays the average readings for torque and crowd taken p er revolution and plotted vs. depth for a single drilling Figure 2 10 Torque and crowd vs d epth for a single laboratory drilling From this drilling, the average t orque and crowd displaye d in the summary of statistics box are used with the other drilling parameters, also displayed, to report a single dat a point vs. compressive strength, q u For instance, the compressive strength, 1 ,683.7 psi, was obtained from unco nfined compression testing on three to six cylinders cast from the same mix as the large G atorock block that was drilled, three cylinders were also tested in split tension This was the method used for all drillings to develop a -20 -18 -16 -14 -12 -10 -8 1,000 1,500 2,000 2,500 3,000 3,500 Depth (in) Applied Torque and Crowd Torque (in-lbs) Crowd (lbf) Final Results 1683.7 psi d (in) N (rpm) u (in/min ) 6 20 0.4 Statistics T (in lbs) F (lbf) Average 2277.9 1505.4 Maximum 2659.5 1849.4 Minimum 1890.0 1177.1 Std. Dev. 144.2 151.7 CV 0.063 0.101 Count 840 840
39 new drilling equ ation usin g a rock auger bit type in a drilling medium representative of Florida limestone. Once the laboratory drilling environment and standard drilling procedure were developed, a drilling plan was implemented t o account for variations in the drilling process. Th is included using two rota tional speeds, 20 rpm and 40 rpm t hree penetration rate settings, 0 .008 in/rev, 0.014 in/rev and 0.020 in/rev, two bit diameters, and four design strengths of rock, 10, 20, 4 0 and 120 tons per square foot, tsf Usi ng various dril ling parameters provided a means to analyze how each parameter affected the drilling as a whole. The drilling plan consisted of 48 drillings that were needed to complete the research. The following table s illustrate all d rillings that were planned to me et the requirements of the research ; followed by the actual strengths that were record ed for each designated drilling. The Gatorock blocks used to perform the drillings had a small degree of variability which is typical of concrete design. As well, on some occasions it was not feasible to perform the laboratory drillings on the 14 th day. Therefore, some blocks were tested at later dates, thus increasing the strength of the material. However, the referenced rock strength for each drilled bl ock was from performing unconfined compression testing on at least three cast cylinders the day of drilling. Each of the tested cylinders were cast from the same mix as the large Gatorock blocks. On most occasions, split tension testing was also performe d. However, sometimes the split tension cylinders had to be used for compression testing.
40 Table 2 4 Gatorock p rojected strengths (20 RPM ) 20 RPM u (in/rev) 4.5" bit 6" bit Strength (psi) Strength (psi) 0.008 140 280 556 1,667 140 280 556 1,667 0.014 140 280 556 1,667 140 280 556 1,667 0.02 140 280 556 1,667 140 280 556 1,667 Table 2 5 Gatorock a ctual s trengths (20 RPM) 20 RPM u (in/rev) 4.5" bit 6" bit Strength (psi) Strength (ps i) 0.008 135.4 281.5 498.7 1 548.9 135.4 273.9 498.7 1 548.9 0.014 135.4 296.4 650.6 1 601.2 135.4 282.1 622.6 1 601.2 0.02 135.4 279.4 601.9 1 683.7 135.4 279.4 601.9 1 683.7 Table 2 6 Gatorock p rojected s trengths (40 R PM) 40 RPM u (in/rev) 4.5" bit 6" bit Strength (psi) Strength (psi) 0.008 140 280 556 1,667 140 280 556 1 667 0.014 140 280 556 1,667 140 280 556 1 667 0.02 140 280 556 1,667 140 280 556 1 667 Table 2 7 Gatorock a ctu al s trengths (40 RPM) 40 RPM u (in/rev) 4.5" bit 6" bit Strength (psi) Strength (psi) 0.008 135 .0 273.9 486.6 1 637.7 135 .0 295.1 545.2 1 637.7 0.014 135 .0 284.7 545.2 1 514.2 135 .0 284.7 486.6 1 514.2 0.02 135 .0 317.9 626.5 1 601 .0 135 .0 281.5 622. 6 1 601 .0 In addition to the 48 planned dri llings to meet the research requirements 33 extra drillings at various strengths and drilling parameters were also performed. This provided a total of 81 recorded drilling data points used to develop the field prediction equation. The 33 extra drilling data points were obtained from good data that was recorded while developing the drilling methods and from using an incremental drilling method for the low end strength blocks, i.e., 140 psi.
41 It is was discovered that as G atorock block production was coming to an end, the research would be two blocks short of meeting the requirements of the drilling plan It was then decided that for the lower end projected strength of 140 psi these blocks would be drilled using an incremental method. F or a specified design strength, i.e. 135.4 psi and use a different penetration rate for each 4 in ch increment of drilling. It was previously determined that each drill run should be approximately four inches for the most undisturbed result. Ty pically, the penetration rate was held consta nt throughout drilling and the three inc rements of the drilled block were averaged. However, due to the shortage of blocks, not all the proposed drillings were possible using th is method. Therefore the final four blocks were drilled using t he following method: the first four inches used the slowes t penetration rate, the second four inch drill run used the middle penetration rate, a nd the final four inches were drilled using the fastest penetration rate. second available block was then drilled using the same procedure but with the 20 RPM s etting. By using this method, all needed data points for the designated stre ngths were collected using only two blocks instead of six b locks. With the remaining two blocks the same procedure was performed, reversing the order of penetration rates, i.e. penetration rates in the following order: Fastest, middle, and slowest Not only did this method provide all the needed drillings but it also provided a means to compare drilling parameters at the same exact str ength. As seen in Table s 2 6 through 2 8 the strength s are not identical. The increase or decrease in various block strengt h affected the recorded parameters, torque and crowd used in the comparison. Using multiple
42 drilling parame ters for the same block provided direct comparison of the recorded drilling parameters, torque and crowd at the same specified strength. This allowed comparisons to be made similar to Karasa The incremental method is recommended if future drillings are to take place. A far larger number of design strengths co uld be drilled using the same amount of material used in this research; and t ypically average s from each four inch drill run were very similar. As previously stated there were a total of 81 recorded drillings completed From the 81 drillings, there were 8 1 compressive strength, q u values and 64 tensile strength, q t values available for analyses. Of the 81 q u t bit and 35 we t values was a result of not enough cylinders available for split tension testing due to changes in drilling dates in the early stages of laboratory dri llings, i.e. split tension cylinders had to be used for compression testing the day of drilling as the planned drilling date was pushed back from 14 day drilling to a later date Lab Data Analyses Once all of the laboratory drilling s were complete an analysis of each drilling parameter took place. The ana lysis included investigating the effects of bit diameter on the applied forces torque and crowd, the torque and crowd relationship, the effects of rotational speed and penetration rate on torque and crowd, and the relationship of torque and crowd with c omp ressive and tensile strength q u vs. q t The first step to the analyses was to compare recorded q u values vs recorded q t values to look for trending of the material tested. The following plot provides this
43 comparison using a linear fit equation and 2 nd or der polynomial fit equa tion to describe the trending, b oth h ad their intercepts set to zero, Figure 2 11 Figure 2 11 q u vs. q t plot with linear and 2 nd order polynomial curve fitting As seen from the plot, the trending is quite linear. However the second order polynomial does a better job of describing the variability as it has the higher R 2 value, 0.9774 > 0.9737 From th e results it was confirmed that G atorock mixing and material formation was qui te consistent throughout all the laboratory tests Effects of Bit Diameter Investigated next was the influence bit diameter had on the applied forces, torque and crowd. The goal was to determine if either of the applied forces were more affected by change s in bit diameter than the other, indicating a less reliable drilling parameter. For each bit diameter, the influence of penetration rate and rotational speed were removed by only grouping drilling data points subjected to the same penetration rate and ro tational speed. The data point groupings were then plotted as either torque or crowd vs. unconfined compressive strength for each bit diameter. If changes in bit y = 2E 05x 2 + 0.2287x R = 0.9774 y = 0.1992x R = 0.9737 0 50 100 150 200 250 300 350 400 0 500 1,000 1,500 2,000 Splitting Tensile Strength, qt (psi) Unconfined Compressive Strength, qu (psi)
44 diameter have little to no influence on the applied force, then the two bit diameter groupin gs should plot on top of one another and be accurately defined by a single regression line, indicated by a higher r squared valu e. Figure 2 12 Torque vs unconfined compressive strength (N = 20 rpm, u = 0.16 in/min) Figure 2 13 Crowd vs unconfined compressive strength (N = 20 rpm, u = 0.16 in/min) y = 0.8691x 89.33 R = 0.989 y = 1.0107x 32.739 R = 0.9954 y = 0.9364x 54.778 R = 0.97 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 0 500 1,000 1,500 2,000 Torque, T (in lbs) Unconfined Compressive Strength, qu (psi) 4.5" 6" 4.5" 6" y = 0.29x + 23.15 R = 0.8543 y = 0.7569x 75.915 R = 0.9939 y = 0.522x 23.843 R = 0.771 0 200 400 600 800 1,000 1,200 0 500 1,000 1,500 2,000 Crowd, F (lbf) Unconfined Compressive Strength, qu (psi) 4.5" 6" 4.5" 6"
45 Figure 2 14 Torque vs splitting t ensile strength (N = 20 rpm, u = 0.16 in/min) Figure 2 15 C rowd vs splitting tensile strength (N = 20 rpm, u = 0.16 in/min) y = 4.2645x 110.33 R = 0.9358 y = 4.9617x 36.89 R = 0.9677 y = 4.5815x 65.658 R = 0.9266 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 0 50 100 150 200 250 300 350 Torque, T (in lbs) Splitting Tensile Strength, qt (psi) 4.5" 6" 4.5" 6" y = 1.3681x + 22.243 R = 0.7473 y = 3.7081x 78.283 R = 0.9623 y = 2.5295x 27.324 R = 0.7224 0 200 400 600 800 1,000 1,200 0 50 100 150 200 250 300 350 Crowd, F (lbf) Splitting Tensile Strength, qt (psi) 4.5" 6" 4.5" 6"
46 Figure 2 16 Torque vs unconfined compressive strength (N = 20 rpm, u = 0.28 in/min) Figure 2 17 Crowd vs unconfined compressive strength (N = 20 rpm, u = 0.28 in/min) y = 0.8698x 62.907 R = 0.9898 y = 1.005x + 43.555 R = 0.9991 y = 0.9362x 9.2916 R = 0.9602 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 0 500 1,000 1,500 2,000 Torque, T (in lbs) Unconfined Compressive Strength, qu (psi) 4.5" 6" 4.5" 6" y = 0.2667x + 6.8459 R = 0.8359 y = 0.3895x + 56.739 R = 0.924 y = 0.3273x + 31.958 R = 0.7891 0 100 200 300 400 500 600 700 800 0 500 1,000 1,500 2,000 Crowd, F (lbf) Unconfined Compressive Strength, qu (psi) 4.5" 6" 4.5" 6"
47 Figure 2 18 Torque vs splitting tensile strength (N = 20 rpm, u = 0.28 in/min) Figure 2 19 Crowd vs splitting tensile strength (N = 20 rpm, u = 0.28 in/min ) y = 4.4125x + 9.4658 R = 1 y = 5.1595x + 49.958 R = 0.9961 y = 4.7778x + 44.576 R = 0.9808 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 0 50 100 150 200 250 300 350 Torque, T (in lbs) Splitting Tensile Strength, qt (psi) 4.5" 6" 4.5" 6" y = 1.3589x + 60.812 R = 0.9952 y = 2.0226x + 56.735 R = 0.9427 y = 1.6889x + 66.385 R = 0.8984 0 100 200 300 400 500 600 700 800 0 50 100 150 200 250 300 350 Crowd, F (lbf) Splitting Tensile Strength, qt (psi) 4.5" 6" 4.5" 6"
48 Figure 2 20 Torque vs unconfined compressive strength (N = 20 rpm, u = 0.40 in/min) Figure 2 21 Crowd vs unconfined compressive strength (N = 20 rpm, u = 0.40 in/min) y = 1.3033x 150.33 R = 0.9791 y = 1.353x 43.155 R = 0.9914 y = 1.3285x 96.879 R = 0.9768 0 500 1,000 1,500 2,000 2,500 0 500 1,000 1,500 2,000 Torque, T (in lbs) Unconfined Compressive Strength, qu (psi) 4.5" 6" 4.5" 6" y = 0.5345x 58.414 R = 0.8838 y = 0.924x 88.675 R = 0.9842 y = 0.7293x 73.202 R = 0.8557 0 200 400 600 800 1,000 1,200 1,400 1,600 0 500 1,000 1,500 2,000 Crowd, F (lbf) Unconfined Compressive Strength, qu (psi) 4.5" 6" 4.5" 6"
49 Figure 2 22 Torque vs splitting tensile strength (N = 20 rpm, u = 0.40 in/min) Figure 2 23 Crowd vs splitting tensile strength (N = 20 rpm, u = 0.40 in/min) y = 7.515x 192.27 R = 0.9393 y = 7.9018x 133.91 R = 0.9479 y = 7.7006x 158.11 R = 0.9391 0 500 1,000 1,500 2,000 2,500 0 50 100 150 200 250 300 350 Torque, T (in lbs) Splitting Tensile Strength, qt (psi) 4.5" 6" 4.5" 6" y = 3.037x 68.224 R = 0.8267 y = 5.4016x 151.18 R = 0.9428 y = 4.209x 103.1 R = 0.8168 0 200 400 600 800 1,000 1,200 1,400 1,600 0 50 100 150 200 250 300 350 Crowd, F (lbf) Splitting Tensile Strength, qt (psi) 4.5" 6" 4.5" 6"
50 Figure 2 24 Torque vs unconfine d compressive strength (N = 40 rpm, u = 0.32 in/min) Figure 2 25 Crowd vs unconfined compressive strength (N = 40 rpm, u = 0.32 in/min) y = 0.8124x 106.01 R = 0.9897 y = 0.9403x 79.078 R = 0.9972 y = 0.8813x 100.18 R = 0.9793 0 200 400 600 800 1,000 1,200 1,400 1,600 0 500 1,000 1,500 2,000 Torque, T (in lbs) Unconfined Compressive Strength, qu (psi) 4.5" 6" 4.5" 6" y = 0.2854x 10.189 R = 0.9681 y = 0.6883x 79.508 R = 0.9934 y = 0.4913x 50.356 R = 0.8006 0 200 400 600 800 1,000 1,200 0 500 1,000 1,500 2,000 Crowd, F (lbf) Unconfined Compressive Strength, qu (psi) 4.5" 6" 4.5" 6"
51 Figure 2 26 Torque vs splitting tensile strength (N = 40 r pm, u = 0.32 in/min) Figure 2 27 Crowd vs splitting tensile strength (N = 40 rpm, u = 0.32 in/min) y = 3.9882x 92.034 R = 0.9694 y = 4.5918x 41.208 R = 0.993 y = 4.3236x 78.933 R = 0.9654 0 200 400 600 800 1,000 1,200 1,400 1,600 0 50 100 150 200 250 300 350 Torque, T (in lbs) Splitting Tensile Strength, qt (psi) 4.5" 6" 4.5" 6" y = 1.4128x 9.4338 R = 0.9499 y = 3.3361x 43.323 R = 0.9892 y = 2.4098x 38.332 R = 0.7832 0 200 400 600 800 1,000 1,200 0 50 100 150 200 250 300 350 Crowd, F (lbf) Splitting Tensile Strength, qt (psi) 4.5" 6" 4.5" 6"
52 Figure 2 28 Torque vs. unconfined compressive strength (N = 40 rpm, u = 0.56 in/min) Figure 2 29 Crowd vs unconfined compressive strength (N = 40 rpm, u = 0.56 in/min) y = 1.0274x 148.27 R = 0.9695 y = 1.2966x 107.76 R = 0.9934 y = 1.1592x 142.79 R = 0.946 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 2,000 0 500 1,000 1,500 2,000 Torque, T (in lbs) Unconfined Compressive Strength, qu (psi) 4.5" 6" 4.5" 6" y = 0.3735x 33.938 R = 0.8493 y = 0.8965x 103.7 R = 0.9934 y = 0.6275x 81.715 R = 0.76 0 200 400 600 800 1,000 1,200 1,400 0 500 1,000 1,500 2,000 Crowd, F (lbf) Unconfined Compressive Strength, qu (psi) 4.5" 6" 4.5" 6"
53 Figure 2 30 Torque vs splitting tensile strength (N = 40 rpm, u = 0.56 in/min) Figure 2 31 C rowd vs splitting tensile strength (N = 40 rpm, u = 0.56 in/min) y = 4.7693x 107.15 R = 0.9768 y = 5.9675x 104.26 R = 0.9806 y = 5.3532x 99.982 R = 0.955 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 2,000 0 50 100 150 200 250 300 350 Torque, T (in lbs) Splitting Tensile Strength, qt (psi) 4.5" 6" 4.5" 6" y = 1.7992x 21.105 R = 0.8988 y = 4.106x 99.196 R = 0.9711 y = 2.9345x 54.118 R = 0.784 0 200 400 600 800 1,000 1,200 1,400 0 50 100 150 200 250 300 350 Crowd, F (lbf) Splitting Tensile Strength, qt (psi) 4.5" 6" 4.5" 6"
54 Figure 2 32 Torque vs unconfined compressive strength (N = 40 rpm, u = 0.80 in/min) Figure 2 33 Crowd vs unconfined compressiv e strength (N = 40 rpm, u = 0.80 in/min) y = 1.0243x 105.76 R = 0.9683 y = 1.4535x 139.47 R = 0.9829 y = 1.2343x 122.95 R = 0.9243 0 500 1,000 1,500 2,000 2,500 0 500 1,000 1,500 2,000 Torque, T (in lbs) Unconfined Compressive Strength, qu (psi) 4.5" 6" 4.5" 6" y = 0.4148x 31.072 R = 0.796 y = 0.9098x 141.39 R = 0.9451 y = 0.6554x 84.208 R = 0.7678 0 200 400 600 800 1,000 1,200 1,400 1,600 0 500 1,000 1,500 2,000 Crowd, F (lbf) Unconfined Compressive Strength, qu (psi) 4.5" 6" 4.5" 6"
55 Figure 2 34 Torque vs splitting tensile strength (N = 40 rpm, u = 0.80 in/min) Figure 2 35 Crowd vs splitting tensile strength (N = 40 rpm, u = 0.80 in/min ) y = 4.9566x 78.134 R = 0.9905 y = 6.6887x 140.07 R = 0.9573 y = 5.8369x 96.775 R = 0.9365 0 500 1,000 1,500 2,000 2,500 0 100 200 300 400 Torque, T (in lbs) Splitting Tensile Strength, qt (psi) 4.5" 6" 4.5" 6" y = 2.1743x 54.472 R = 0.9018 y = 4.1311x 124.89 R = 0.9162 y = 3.1688x 75.785 R = 0.797 0 200 400 600 800 1,000 1,200 1,400 1,600 0 100 200 300 400 Crowd, F (lbf) Splitting Tensile Strength, qt (psi) 4.5" 6" 4.5" 6"
56 Evident from the plots, crowd is more affected by changes in bit diameter than torque. This can be interpreted from the lower r squared values obtained for crowd when defining both bit diameter groupings with a single regression line. The crowd plot s also display more of a progressive deviation between the two bit diameter groupings as strength increases; suggesting that crowd is highly dependent on bit diameter and becomes less predictable as compression and split tension strength increases. Conver sely, the two torque groupings defined by a single regression line produced excellent r square values with limited deviation as compression and split tension strength increased. This indicated a strong correlatio n between torque and compression and split tension strength with less dependency on bit diameter. All the plots suggest that torque is a more reliable drilling parameter for estimating rock strength based on changes in bit diameter; as the results were consistent for each of the groupings based on penetration rate and rotational speed. While investigating the trends of torque and crowd based on bit diameter, it was i dentified that torque, rotational speed, and penetration rate have a direct relationship independent of rock strength For instance, plotted in Figure 2 36 are two groupings based on different penetration rates and rotational speeds using the s ame diameter drill bit i.e., 6 inch drill bit I t can be derived that the volume of the excavation from the 40 rpm gr ouping would be double that of the 20 rpm grouping if both excavations were drilled for the same amount of time based on the penetration rates. Therefore, by doubling the rotational speed, twice the amount of material can be excavated using the same amoun t of torque indicating a direct relationship between torque, rotational speed and penetration rate, independent of compressive strength
57 Figure 2 36 Torque plotted vs. unconfined compressive strength displaying the direct tor que rotational speed penetration rate relationship Increasing Force and Torque Relationship Investigated next, was how crowd increased with torque independent of the strength of the material. This was done by plo tting crowd measurem ents vs their respective torque measurement Two plots wer bit and one This was done to see if crowd increased with torque at a different rate when u sing different bit sizes. The two plots are provided in Figu res 2 3 8 and 2 3 9 where it can be seen that crowd does increase at a higher rate as torque increases using a larger bit diameter. Also of interest is the linear trending of crowd increase with respect to to rque. As bit diameter increased, the trend beca me more linear. This was confirmed by the larger R 2 R 2 = 0.9536 > R 2 = 0.9196 y = 0.0003x 2 + 0.9413x R = 0.985 y = 0.0002x 2 + 1.0507x R = 0.9966 0 500 1,000 1,500 2,000 2,500 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 Torque, T (in lbs) Unconfined Compressive Strength, qu (psi) N = 40 rpm, u = 0.80 in/min N = 20 rpm, u = 0.40 in/min N = 40 rpm, u = 0.80 in/min N = 20 rpm, u = 0.40 in/min
58 Figure 2 37 Crowd vs. torque curve fitting Figure 2 38 Crowd vs. torque curve fitting y = 0.401x R = 0.9196 0 100 200 300 400 500 600 700 800 900 1,000 0 500 1,000 1,500 2,000 2,500 Crowd, F (lbf) Torque, T (in lbs) y = 0.6146x R = 0.9536 0 200 400 600 800 1,000 1,200 1,400 1,600 0 500 1,000 1,500 2,000 2,500 Crowd, F (lbf) Torque, T (in lbs)
59 Torque and Force vs. Penetration Rate per Rotational Speed Ratio (u/N) After comparing crowd vs. torque, penetration rate per rotational speed ratios u/N, plotted vs. torque and crowd were investigated. This was simila r to comparisons made by Karasawa in 2002. However, this research did not use the no rmalization that Karasawa used, i.e., 2 comparison 1,2 This was done to eliminate a bias toward a specific equation m oving forward in the development of a final field p rediction equation. The research needed to consider all opt ions; and therefore torque and crowd were compared to the u/N ratios d irectly without normalization. When comparisons began it was noticed that s trength increases/decreases between groupin gs used provided a poor result, i.e. using 270 psi, 290 psi and 310 psi with the same u/N ratio provided a poor result because of the changes in compres sive strength Therefore, the low end 140 psi drillings wer e used for these comparisons. As discussed, t he 140 psi r ange blocks were drilled using three different penetration rates per block side. This provided a means to investigate the trends of u/N vs. torque and crowd without changes in compressive strength affecting the results. These comparis ons can be seen in Figures 2 40 through 2 43 In Figures 2 40 through 2 4 3 the two groupings with the lower rotational speeds required more torque to complete the excavations regardless of the compressive strength of the material, confirming the strong torque rotational speed penetration rate relationship. The same was not found w hen investigating the u/N vs. crowd plots. In fact, the crowd rotational speed relationship appeared to be random or non existent. Althou gh, both crowd and torque did display trends of increasing applied force as the penetration rate was increased.
60 Figure 2 39 u/N vs. Torque Figure 2 40 u/N vs. Crowd y = 0.0006x 0.0147 R = 0.9943 y = 0.0002x 0.0068 R = 0.9301 y = 0.0003x 0.0124 R = 0.8987 y = 0.0004x 0.0159 R = 0.7573 0 0.005 0.01 0.015 0.02 0.025 0 20 40 60 80 100 120 140 u/N (in) Torque, T (in lbs) 135.0 psi (40 RPM) 135.4 psi (20 RPM) 145.9 psi (20 RPM) 155.4 psi (40 RPM) 135.0 psi (40 RPM) 135.4 psi (20 RPM) 145.9 psi (20 RPM) 155.4 psi (40 RPM) y = 0.0003x + 0.0012 R = 0.444 y = 0.0003x 0.0191 R = 0.9752 y = 0.0001x + 0.0261 R = 0.6754 y = 0.0003x 0.0074 R = 0.262 0 0.005 0.01 0.015 0.02 0.025 0 50 100 150 200 u/N (in) Crowd, F (lbf) 135.0 psi (40 RPM) 135.4 psi (20 RPM) 145.9 psi (20 RPM) 155.4 psi (40 RPM) 135.0 psi (40 RPM) 135.4 psi (20 RPM) 145.9 psi (20 RPM) 155.4 psi (40 RPM)
61 Figure 2 41 u/N vs. Torque Figure 2 42 u/N vs. Crowd y = 0.0002x 0.0102 R = 0.9542 y = 0.0001x 0.0114 R = 0.9976 y = 0.0002x 0.0168 R = 0.8357 y = 0.0002x 0.0062 R = 0.9959 0 0.005 0.01 0.015 0.02 0.025 0 50 100 150 200 250 u/N (in) Torque, T (in lbs) 135.0 psi (40 RPM) 135.4 psi (20 RPM) 145.9 psi (20 RPM) 155.4 psi (40 RPM) 135.0 psi (40 RPM) 135.4 psi (20 RPM) 145.9 psi (20 RPM) 155.4 psi (40 RPM) y = 0.0005x 0.008 R = 0.2124 y = 0.0006x 0.0097 R = 0.9466 y = 0.0003x 0.0143 R = 0.8768 y = 0.0002x 0.0061 R = 0.9711 0 0.005 0.01 0.015 0.02 0.025 0 20 40 60 80 100 120 140 u/N (in) Crowd, F (lbf) 135.0 psi (40 RPM) 135.4 psi (20 RPM) 145.9 psi (20 RPM) 155.4 psi (40 RPM) 135.0 psi (40 RPM) 135.4 psi (20 RPM) 145.9 psi (20 RPM) 155.4 psi (40 RPM)
62 From the provided plots a few trends were noticed: Torque and crowd both show increasing values as the u/N ratio increases Lower rot ational speeds provide larger torques. This was noticed for both bit diameters where the 40 RPM drillings produced the lowest torques. The highest strength block, 155.5 psi, drilled at 40 RPMs produced a lower torque average than the 135.4 psi block dril Trends of torque increasing as rock strength increases are seen in both plots as was expected Crowd increases show greater variability with respect to increases in rock streng th Torque and Crowd vs. Compressive and Tensile Strengths Next, torque and crowd measurements vs the respective compressive and tensile strengths of the drilled blocks were investigated If trending of the previous analyses were correct, then crowd shou ld show more variability than torque when compared to compressive and tensile strengths. Figure 2 43 Torque vs. q u with linear curve fitting y = 0.9944x R = 0.907 0 500 1,000 1,500 2,000 2,500 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 Torque, T (in lbs) Unconfined Compressive Strength, qu (psi)
63 Figure 2 44 Crowd vs. q u with linear curve fitting Figure 2 45 Torque vs q t with linear curve fitting y = 0.5117x R = 0.7362 0 200 400 600 800 1,000 1,200 1,400 1,600 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 Crowd, F (lbf) Unconfined Compressive Strength, qu (psi) y = 5.0389x R = 0.8864 0 500 1,000 1,500 2,000 2,500 0 50 100 150 200 250 300 350 400 Torque, T (in lbs) Splitting Tensile Strength, qt (psi)
64 Figure 2 46 Crowd vs. q t with linear curve fitting As expected, t he four plots show that torque has a stronger relationship with compressive and s plitting tensile strength than crowd. This is confirme d by the higher R squar ed values attained for the torque plots compared to c rowd plots The torque vs. compressive strength plot shows that torque is more predictable, displays less scatter, and produ ces an approximate 1:1 linear trend with compressive strength when drilling data points with variations in bit diameter, rotational speed and penetration rate are plotted together. This proves that by using torque, rotational speed, penetration rate and bit diameter in combination, compressive strength can be accurately measured when drilling with a rock auger bit. From the analysis, the following conclusions were drawn for torque and crowd: y = 2.6311x R = 0.7308 0 200 400 600 800 1,000 1,200 1,400 1,600 0 50 100 150 200 250 300 350 400 Crowd, F (lbf) Splitting Tensile Strength, qt (psi)
65 Torque: Shows less dependency on bit diameter; Shows a st rong trend with rotational speed; Shows a strong trend with penetration rate; Shows a strong trend with compressive strength independent of all other drilling parameters Crowd: Shows more dependency on bit diameter; Shows a poor trend with rotation al speed; Shows a good trend with penetration rate; Shows a trend with compressive strength independent of all other drilling parameters From these conclusions it was determined that torque is a far more reliable drilling parameter than crowd when it comes to predict ing rock strength. As identified earlier, any drilling estimation of rock strength needs to be less dependent on bit diameter as field drilling requires significant upscaling, by a factor of 10, compared to laboratory drilling based on sur vey results of current drilled shaft practices in Florida Since c rowd was found to be more dependent on bit diameter and provi ded less correlation with the other drilling parameters more focus should be placed on torque, rotational spee d and penetratio n rate to predict rock strength. As well, bit diameter should only be used to define the area of the excavation to eliminate the effects of Equation 2 3 meets this criteria where on average the thrust component, F/A, only ac counted for 0.13% of the recorded specific energy values, e, recorded during laboratory drilli ng. Th is was not the cas equation, Equation 2 1 where a large degree of emphasis is placed on bit diameter and crowd is grouped in with al l th e other drilling parameters; which does not restrict the applied force from greatly influencing t he final result. This indicated
66 equation should be more reliable, accounting for variations in bit diameter, rotational speed and penetration ra te. The following two sections provide the developed equations for both Karasawa et al. 1,2 and Teale 3 Developed D s vs. q u and D s vs. q t Curves Once all lab oratory drillings were completed and the data reduced, the finalized D s v s. q u and D s vs. q t curves were created. The drillability strength D s was derived ( 2 1 ) Both curves are presented her e with bit sizes analyzed independently and dependently, i.e. an equation and curve fit developed f or each bit size and as a whole : Figure 2 47 D s vs q u plot y = 0.0154x 2 + 107.16x R = 0.8274 y = 0.0005x 2 + 41.299x R = 0.8852 y = 0.0049x 2 + 79.348x R = 0.5854 0 50,000 100,000 150,000 200,000 250,000 300,000 350,000 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 Drillability Strength, Ds (psi) Unconfined Compressive Strength, qu (psi) 4.5" bit 6" bit 4.5" bit 6" bit
67 Figure 2 48 D s vs q t plot From the developed curves it can be highly dependent on the diameter of the bit. Curve fitting shows good trending for each bit diameter independently from one another but poor trending using a single regression equation to define all of the drillings. This can also be seen in his report from 2002: Bit Diameter I s (Mpa) S e (Mpa) D s (Mpa) 101.6 mm (4") 218.0 154.0 108.0 127.0 mm (5") 229.0 137.0 82.0 142.9 mm (5 5/8") 231.0 127.0 69.6 Figure 2 49 Kara s S e and D s (Sori granite, 50 RPM) 2 y = 0.2053x 2 + 596.38x R = 0.8366 y = 0.1502x 2 + 166x R = 0.8669 y = 0.3779x 2 + 316.8x R = 0.576 0 50,000 100,000 150,000 200,000 250,000 300,000 350,000 0 100 200 300 400 Drillability Strength, Ds (psi) Splitting Tensile Strength, qt (psi) 4.5" bit 6" bit 4.5" bit 6" bit
68 From Ka different diameters, were never meant to identify the full effect bit diameter has on rock drillability systematically. His objective was to show one example of how bit diameter affects I s S e and D s 2 This was diameters that were representative of field drilling conditions where no up scaling would be requi would not be an ideal opt ion moving forward and Developed e vs. q u and e vs. q t Curve s s Approach The Concept of Spec ific Energy in Rock Drilling 3 he states that in rotary non percussive drilling, work is done by the thrust, F, and the torque, T. If the rotation al speed is N, the area of the hole or excavation is A and the penetration rate is u (all using the standard u nits as this research ), the total work done in one minute is: (2 2 ) The volume of rock excavated in one minute is Au where A the is cross sectional area and u is the change in penetration depth. Putting e as the specific energy, dividing work by volume, gives (2 3 ) Using subscripts t and r e (2 4 ) and, (2 5 )
69 In T eale that the thrust component, (F/A) is e quivalent to the exerted by the t h rust over a cross sectional area of the hole. Specific energy is, in fact, dimens ionally identica l with pressure or stress. Physically this arises from the fact that if a force F acting on and normal to a surface of area A moves through a distance ds the increment of work done, dW is equal to Fds The volume change effected by the movement, dV is Ads If e is the specific energy at any point, then e = d W/dV = F/A = P the pressure at that point. For a given size of excavation, A is constant so that e t is directly proportional to F It is always small in comparison with e r i.e., sometimes negligible. the thrust component is always small in comparison with the torque component makes sense when applying it to field drilling. From conversations with multiple rig operators, all of them stated that one sh ould always let the torque to do the work. This was also noticed an d recorded during field monitoring as well. In the field, penetration rates are rarely consistent. However, rotational speeds are almost al ways held at a consistent rate, not an exact ra te. This is because the rig provide the cutting action and prevents overcrowding of the bit and possible stall or ndam entally makes sense placing more emphasis on the parameters that are providing the majority of the work. Also of interest, it has been witnessed in the field that sometimes rig operators will fracture a rock layer without using rotation or torque. They s imply pound the rock layer using crowd until it is fractured. T his is not a recommended method would compensate for a zero rotary component by placing all the wor k done on the
70 thrust component, i.e., an index value would still be obt ained without a torque or rotation component; this would not be pos This is because Teale only uses bit dimeter to define the cross sectional area of the excavation whereas Karasawa places additi onal emphasis on bit diameter. By only using the bit diameter to define the area of the excavation, upscaling is sues should be inherently resolved when transitioning to field drilling with larger rock augers. The following are the developed specific energy vs compressive strength plot, e vs. q u and the specific energy vs. split ting tensile strength plot, e vs. q t : Figure 2 50 Specific energy vs unconfined compressive strength y = 0.0101x 2 + 12.546x R = 0.9001 y = 0.0032x 2 + 14.533x R = 0.8938 y = 0.0066x 2 + 13.681x R = 0.8529 0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 Specific Energy, e (psi) Unconfined Compressive Strength, qu (psi) 4.5" bit 6" bit 4.5" bit 6" bit
71 Figure 2 5 2 Specific energy vs tensile strength A seen in both figures, plotting specific energy vs. compressi ve and split ting tensile strength provides a analyzing both bit sizes separately and as a whole. Comparing R 2 values for both compressive and t the better fit, thus ex plaining more of the variation, R 2 = 0.8529 > 0.5854 for compressive strength and R 2 = 0.8432 > 0.5 76 for split ting tensile strength. Comparisons Based on Rotational Speed After comparing Teale comparisons were then made on th e b asis of rotational speed. The following provides the comp arisons based on rotational speed groupings, i.e. 20 RPM drillings are grouped together and 40 RPM drillings are grouped together, regardless of bi t diameter or penetration rate. y = 0.2527x 2 + 66.239x R = 0.8895 y = 0.1314x 2 + 58.525x R = 0.891 y = 0.202x 2 + 59.152x R = 0.8432 0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 0 100 200 300 400 Specific Energy, e (psi) Splitting Tensile Strength, qt (psi) 4.5" bit 6" bit 4.5" bit 6" bit
72 Figure 2 5 3 D s vs q u (grouped by rotational speeds) Figure 2 5 4 e vs q u (grouped by rotational speeds) y = 0.0123x 2 + 109.55x R = 0.5887 y = 0.0135x 2 + 63.08x R = 0.5957 y = 0.0049x 2 + 79.348x R = 0.5876 0 50,000 100,000 150,000 200,000 250,000 300,000 350,000 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 Drillability Strength, Ds (psi) Unconfined Compressive Strength, qu (psi) 20 RPM 40 RPM 20 RPM 40 RPM y = 0.0038x 2 + 17.932x R = 0.8301 y = 0.0083x 2 + 11.228x R = 0.8811 y = 0.0066x 2 + 13.681x R = 0.8539 0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 Specific Energy, e (psi) Unconfined Compressive Strength, qu (psi) 20 RPM 40 RPM 20 RPM 40 RPM
73 Figure 2 5 5 D s vs q t (grouped by rotational speeds) Figure 2 5 6 e vs q t (grouped by rotational speeds) y = 0.68x 2 + 285.15x R = 0.6252 y = 0.2854x 2 + 303.01x R = 0.5525 y = 0.3779x 2 + 316.8x R = 0.5788 0 50,000 100,000 150,000 200,000 250,000 300,000 350,000 0 100 200 300 400 Drillability Strength, Ds (psi) Splitting Tensile Strength, qt (psi) 20 RPM 40 RPM 20 RPM 40 RPM y = 0.2512x 2 + 56.651x R = 0.8593 y = 0.1923x 2 + 53.203x R = 0.8501 y = 0.202x 2 + 59.152x R = 0.8446 0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 0 100 200 300 400 Specific Energy, e (psi) Splitting Tensile Strength, qt (psi) 20 RPM 40 RPM 20 RPM 40 RPM
74 In comparing the two equations based on rot ational speed groupings, it is evident by the obtained R 2 values that far superio r when describing the groupings independently and as a whole. This same trend was also noticed when the drilled data was grouped by penetration rates, both independently and as a whole. Comparisons Based on Penetration Rates The next comparison provides groupings based on penetration r ates for each rotational speed This was done to reduce the plots from being far to cluttered with six different penetration rate regression lines The com parisons can be seen here; with the curve equations and R 2 values for all penetration rates combine d displayed atop the legend : Figure 2 5 7 D s vs q u (grouped by penetration rates for 20 RPM drillings) y = 0.0186x 2 + 89.853x R = 0.658 y = 0.0218x 2 + 124.97x R = 0.6735 y = 0.0164x 2 + 94.941x R = 0.564 y = 0.0123x 2 + 109.55x R = 0.5887 0 50,000 100,000 150,000 200,000 250,000 300,000 350,000 0 500 1,000 1,500 2,000 Drillability Strength, Ds (psi) Unconfined Compressive Strength, qu (psi) 0.16 in/min 0.28 in/min 0.40 in/min 0.16 in/min 0.28 in/min 0.40 in/min
75 Figure 2 5 8 e vs q u (grouped by penetration rates for 20 RPM drillings) Figure 2 5 9 D s vs q u (grouped by penetration rates for 40 RPM drillings) y = 0.0081x 2 + 21.738x R = 0.9472 y = 0.002x 2 + 17x R = 0.9461 y = 0.0052x 2 + 11.102x R = 0.9182 y = 0.0038x 2 + 17.932x R = 0.8301 0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 0 500 1,000 1,500 2,000 Specific Energy, e (psi) Unconfined Compressive Strength, qu (psi) 0.16 in/min 0.28 in/min 0.40 in/min 0.16 in/min 0.28 in/min 0.40 in/min y = 0.0363x 2 + 52.02x R = 0.6688 y = 0.0015x 2 + 79.589x R = 0.6557 y = 0.0068x 2 + 68.607x R = 0.609 y = 0.0135x 2 + 63.08x R = 0.5957 0 50,000 100,000 150,000 200,000 250,000 300,000 350,000 0 500 1,000 1,500 2,000 Drillability Strength, Ds (psi) Unconfined Compressive Strength, qu (psi) 0.32 in/min 0.56 in/min 0.80 in/min 0.32 in/min 0.56 in/min 0.80 in/min
76 Figure 2 60 e vs q u (grouped by penetration rates for 40 RPM drillings) Figure 2 61 D s vs q t (grouped by penetration rates for 20 RPM drillings) y = 0.0089x 2 + 16.654x R = 0.9446 y = 0.0075x 2 + 12.34x R = 0.9573 y = 0.0042x 2 + 10.977x R = 0.9634 y = 0.0083x 2 + 11.228x R = 0.8811 0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 0 500 1,000 1,500 2,000 Specific Energy, e (psi) Unconfined Compressive Strength, qu (psi) 0.32 in/min 0.56 in/min 0.80 in/min 0.32 in/min 0.56 in/min 0.80 in/min y = 0.9534x 2 + 307.35x R = 0.6551 y = 1.1602x 2 + 105.5x R = 0.8199 y = 0.036x 2 + 381.05x R = 0.5228 y = 0.68x 2 + 285.15x R = 0.6252 0 50,000 100,000 150,000 200,000 250,000 300,000 350,000 0 50 100 150 200 250 300 350 Drillability Strength, Ds (psi) Splitting Tensile Strength, qt (psi) 0.16 in/min 0.28 in/min 0.40 in/min 0.16 in/min 0.28 in/min 0.40 in/min
77 Figure 2 6 2 e vs q t (grouped by penetration rates for 20 RPM drillings) Figure 2 6 3 D s vs q t (grouped by penetrati on rates for 40 RPM drillings) y = 0.3408x 2 + 68.402x R = 0.9385 y = 0.0681x 2 + 84.229x R = 0.9414 y = 0.2865x 2 + 33.388x R = 0.9119 y = 0.2512x 2 + 56.651x R = 0.8593 0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 0 50 100 150 200 250 300 350 Specific Energy, e (psi) Splitting Tensile Strength, qt (psi) 0.16 in/min 0.28 in/min 0.40 in/min 0.16 in/min 0.28 in/min 0.40 in/min y = 0.8931x 2 + 270.97x R = 0.6587 y = 0.1901x 2 + 314.89x R = 0.6433 y = 0.18x 2 + 326.71x R = 0.5443 y = 0.2854x 2 + 303.01x R = 0.5525 0 50,000 100,000 150,000 200,000 250,000 300,000 350,000 0 100 200 300 400 Drillability Strength, Ds (psi) Splitting Tensile Strength, qt (psi) 0.32 in/min 0.56 in/min 0.80 in/min 0.32 in/min 0.56 in/min 0.80 in/min
78 Figure 2 6 4 e vs q t (grouped by penetration rates for 40 RPM drillings) From the provided compariso to when grouping results by penetration rate s. Fr om all comparisons between the two developed equa consistently outperformed based on bit diameter, rotational speed and pe netration rate. W hen considering variation s in all of the drilling parameters Figures 2 65 and 2 66 provided better correlation, R 2 = 0.85 > 0.59 T his is attributed to the specific energy equation placing a negligible amount of emphasis on crowd, grouping drilling parameters t hat show strong correlation with one another together and only using bit diameter to define the cross sectional area of the excavation when assessing rock strength, q u Figure 2 66 was chosen fo r use during field monitoring. (2 6 ) y = 0.2396x 2 + 79.565x R = 0.9396 y = 0.1821x 2 + 51.847x R = 0.9571 y = 0.1193x 2 + 44.527x R = 0.9637 y = 0.1923x 2 + 53.203x R = 0.8501 0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 0 100 200 300 400 Specific Energy, e (psi) Splitting Tensile Strength, qt (psi) 0.32 in/min 0.56 in/min 0.80 in/min 0.32 in/min 0.56 in/min 0.80 in/min
79 Figure 2 6 5 D s vs. q u independent of drilling parameter groupings Figure 2 6 6 e vs. q u independent of drilling parameter g roupings y = 0.0049x 2 + 79.348x R = 0.5854 0 50,000 100,000 150,000 200,000 250,000 300,000 350,000 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 Drillability Strength, Ds (psi) Unconfined Compressive Strength, qu (psi) y = 0.0066x 2 + 13.681x R = 0.8529 0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 Specific Energy, e (psi) Unconfined Compressive Strength, qu (psi)
80 Figure 2 6 7 D s vs. q t independent of drilling parameter groupings Figure 2 6 8 e vs. q t independent of drilling parameter groupings y = 0.3779x 2 + 316.8x R = 0.576 0 50,000 100,000 150,000 200,000 250,000 300,000 350,000 0 50 100 150 200 250 300 350 400 Drillability Strength, Ds (psi) Splitting Tensile Strength, qt (psi) y = 0.202x 2 + 59.152x R = 0.8432 0 10,000 20,000 30,000 40,000 50,000 60,000 70,000 0 50 100 150 200 250 300 350 400 Specific Energy, e (psi) Splitting Tensile Strength, qt (psi)
81 CHAPTER 3 DRILLED SHAFT FIELD MONITORING TO MEASURE COMPRESSIVE STRENGTH Development of Equipment for S haft Installation Monitoring in Real Time In addition to a thorough laboratory investigation, field monitoring equipment was needed to monitor shaft installations in the field. However, information regarding the types of drill rigs and tooling used for sh aft installations in Florida was scarce. It was important to understand what was being used in the field before developing equipment to mount on the drill rigs for monitoring the shaft installations. Therefore, an investigation took place to provide more insight. The following section briefly covers the investigation. Surveying Fl orida Contractors and District Geotechnical Engineers In order to gain a better understanding of what types of drill rigs and tooling were being used in the field to install dri lled shafts; a survey was created and presented to leading contractors and district geotechnical engineers from Florida, Alabama, and Georgia that practice in the state of Florida. The intent of the survey was to develop a better understanding of drilled shaft equipment, as well as typical operating parameters of the rigs used in Florida. This included general questions such as: What types of drill rigs are used and who are the manufacturers? How are forces torque and crowd applied to the drill bit? What are the typical rotational speeds and penetration rates used? Are any of the drilling parameters monitored in any way? If so, how? What types of tooling are used and what are the typical bit diameters? What shaft installation method is used, i.e., dry, wet cased? From the results of the survey, field monitoring equipment was developed to create a monitoring system capable of being used on a variety of rig types. The following is a summary of the survey results:
82 Table 3 1 Summary of the drilled shaft su rvey results Rig Type Truck Carrier Crane Crawler % of rigs used 27% 9% 18% 46% Kelly System Telescoping Telescoping Telescoping Telescoping Crowd System Hydraulic Hydraulic Hydraulic Hydraulic Torque System Hydraulic Hydraulic Hydraulic Hydrau lic Crowd Monitoring 67% In cab monitor 100% In cab monitor 50% In cab monitor 60% 75% In cab monitor 25% Digital monitor Torque Monitoring 67% In cab monitor 100% In cab monitor 100% 50% In cab monitor 50% Multiplier 80% 50% In cab mo nitor 25% Multiplier 25% Digital monitor RPM Monitoring 100% In cab monitor 100% In cab monitor 100% 50% In cab monitor 50% Laser sensor 100% 60% In cab monitor 20% Laser sensor 20% Digital monitor Depth Monitoring 100% 100% Weighted tape 50% In ca b monitor 100% Weighted tape 100% 100% Weighted tape 50% In cab monitor 100% 100% Weighted tape 80% In cab monitor 20% Digital monitor In the survey summary table, there are several percentages given for the monitored drilling parameters. The fi rst percentage reflects the portion of returned surveys that answered questions regarding the monitoring of that drilling parameter. The remaining percentages reflect the breakdown of the types of monitoring used. For example, under crawler mounted rigs for crowd monitoring, only 60% of the returned surveys provided and answer. From the 60% returned, 75% of the responses said an in cab monitor is used and 25% said a digital mon itor is used to monitor crowd. From the compiled data, a trend toward s the use of hydraulic powered rigs was observed. The survey results clearly indicate that both applied forces, torque and crowd, are provided via hydraulically driven systems. In most cases, the results indicated that monitoring capabilities are available but no t for every rig type. Therefore,
83 a monitoring system would be needed to record the drilling parameters torque, crowd, rotational speed, and penetration rate in real time. Monitoring Equipment With the understanding that a field monitoring system would be required to monitor the drilling parameters in real time, focus turned to investigating how each drilling parameter could be monitored on the drill rigs. Several different options were explored for each drilling parameter. This section covers the option s chosen to monitor each drilling parameter on the drill rig and the equipment required to provid e the monitoring in real time. In order to measure T, F, N, and u on the drill rig for field monitoring, hydraulic pressure lines controlling crowd and torque need to be tapped with individual pressure transducers. The recorded pressures then need to be converted to physical measures, i.e., lbf for crowd and in lbs for torque, for compatibility with the laboratory drilling equation Measuring rotational speed of the drilling tool, N, requires a proximity sensor to be attached near the rotating collar of the rotary table with no conversion necessary for compatibility. Vertical movement of the drilling tool, u, needs to be monitored from line movement of the cab ling attached to the Kelly bar and main winch using a rotary encoder. Penetration rate is determined as a function of cable movement per unit time with no conversion necessary for compatibility. All drilling parameters and estimated strengths then need t o be recorded and displayed as a function of depth as the drilling tool is advanced. After a thorough investigation of available options, two monitoring systems were compared. A UF developed system built from the ground up and a commercially available sys tem produced by Jean Lutz. The following comparisons were made:
84 Table 3 2 Field monitoring system comparison Jean Lutz System UF Developed System From the comparative analysis, it was decided to acquire the Jean Lutz monitoring system, Figure 3 1 This system was chosen because it was fully
85 operational in its current state and provided hi gh quality sensors to monitor the needed drilling parameters. Additionally, the DIALOG data acquisition model, DAQ, was designed to be highly compatible with many preexisting rig manufacturer installed sensors. Therefore, in some cases the rig manufactur er installed sensors could be used instead of mounting the acquired Jean Lutz sensors on the drill rig, leading to a simplified installation process. For drill rigs with manufacturer installed sensors that do not produce compatible signals with the DIALOG signal conditioners, available from Jean Lutz, can be used to provide compatibility. This makes the system far more universal by providing several different monitoring options: 1. Use Jean Lutz sensors to monitor the drilling parameters 2. Tap into preexisti ng sensors and copy the signals to provide monitoring 3. Use signal conditioners to provide compatibility with the DIALOG for monitoring when preexisting sensors are not compatible 4. Use various comb inations of options 1 through 3 Figure 3 1 Jean Lutz monitoring system (Photo courtesy of author). DIALOG (DAQ) C16400 Pressure Transducer (Torque) C16400 Pressure Transducer (Crowd) V R28 Proximity sensor (Rotary speed) F82 Rotary Encoder (Penetration Rate) Junction Boxes
86 With the Jean Lutz monitoring equipment acquired, focus turned to field monitoring. The following section covers the installation and monitoring equipment setup on the drill rigs, where various combi nations of the monitoring options were used. Field Monitoring Equipment Setup and Installation Throughout the course of the research, drill rigs from three different manufacturers were monitored. Each drill rig was used at a different location. This inclu ded the following three drill rig models and the respective location they were used: 1. IMT AF250, used at the Little River bridge site in Quincy, FL 2. Bauer BG30 Premium Line, used at the Overland bridge site in Jacksonville, FL 3. s Kanapaha site in Gainesville, FL For the IMT drill rig, only a crowd sensor needed to be installed. The remaining sensors were preexisting and installed by the rig manufacturer. To complete the monitoring setup, the torque, rotational speed, and penet ration rate sensors were simply tapped into and connected to the DIALOG v ia the Jean Lutz junction box. The Bauer rig was brand new and equipped with the B tronic monitoring system. The B tronic system provides fully functional sensors with the capability of monitoring and recording all the needed drilling parameters. Integrating the DIALOG required wired connections to be made for the rotational speed and penetration rate sensors at the preexisting terminal connections, located in the electrical unit on the drill rig. For the copy modules were also wired into the preexisting terminal connections located in the electrical unit, and used to bridge the original connections This routed the received signals from each of the sensors to both the B tronic and the DIALOG. While monitoring, both systems were active and recorded the monitored drilling parameters.
87 The SoilMec rig was also equipped with its own fully functional mo nitoring system, the Drilling Mate System, DMS. However, the signals produced from the DMS sensors were not compatible with the DIALOG. Therefore, monitoring was provided through the use of only the Jean Lutz equipment. This required sensors to be insta lled for rotational speed, penetration rate, torque and crowd. Each sensor was routed to the junction box and connected to the DIALOG to provide full monitoring capabilities. Tapping into preexisting sensors typically requires splicing the Jean Lutz cabl ing with a multi pin connection that matches the existing connection. The IMT rig was equipped with extra multi pin connections, greatly simplifying the installation. The following shows the depth sensor tie in connection on the IMT rig. Figure 3 2 Tapping into the IMT depth sensor (Photo courtesy of author). Cabling from the tie in is routed along the same path as the existing sensor cabling and connected to the junction box located in the electrical compartment. Tapping into the IMT rotational s peed and torque sensors was completed using the same method. Tapping into the de pth sensor using a multi pin connection. The Jean Lutz cable is green.
88 The existing depth sensor in Figure 3 2 is a preinstalled rotary encoder integrated into the main cable winch. Similarly, the Jean Lutz sensor is also a rotary encoder, mounted on the outer rim of the main cable winch, Figure 3 3 After installation, the depth sensor is calibrated by comparing the tracked movement with that of the in cab monitor readout as well as physical measurements of vertical drill bit movement. Figure 3 3 Jean Lutz penetration rate sensor (Photo courtesy of author). Monitoring rotational speed is performed using a proximity sensor mounted on a stationary location on the base of the rotary table. Steel bolts are welded to the rotating collar of the rotary head where rotation occurs without wobbling. The proximity sensor Depth Sensor
89 detects each bolt as the collar rotates and the rotational speed can be determined directly. The steel bolts are evenly spaced around the rotary head. The sensor is then calibrated by comparing mea sured rotary speeds to the in cab monitor readout, and through visual inspection by counting the approximate number of rotations over a minute. Figure 3 4 Rotational speed sensor (Photo courtesy of author). The torque and crowd sensors are tied into the hydraulic lines where the existing sensors are located or in locations along the hydraulic lines where differential pressures are not experienced. Cabling from both sensors are routed to the junction box. Proximity Sensor Steel Bolts
90 Figure 3 5 Tapping into the torque and crowd hydraulic lines (Photos courtesy of the author and Jean Lutz). Figure 3 6 Junction box located in electrical compartment (Photo courtesy of author). Cabling from the junction box is routed to the main cab and connected to the DIALOG T ypically there are openings on the floor board or the back wall of the main cab that allow the cabling to be easily routed to the DIALOG. Presented in Figure 3 7 Junction Box Cable running to the cab
91 are both the DIALOG and B tronic systems monitoring a drilled shaft installation in real tim e. Figure 3 7 DIALOG and B tronic both monitoring a shaft installation in real time (Photo courtesy of author). The monitoring equipment is designed and installed in a manner that does not interrupt or interfere with the drilling process. The DIA LOG also provides real time external visualization of the monitored shaft installation away from the drill rig. The monitored data is wirelessly transmitted to an external computer, via Bluetooth technology. This provides a real time data stream of the D IALOG graphical monitor DIALOG B tronic
92 display on an external computer as presented in the following image, and allows monitoring to be conducted from a safe distance. Figure 3 8. External viewing of a monitored shaft installation via Bluetooth (Photo courtesy of aut hor).
93 Once the monitoring equipment is installed and calibrated, the next step is der iving conversion coefficients, K for both the hydraulic torque and crowd measurements The conversion coefficients transform the recorded hydraulic pressures to physical measures of the drilling parameters, i.e., in lbs for torque and lbf for crowd, that are compatible with the developed laboratory drilling equations. This is achieved ollowing provides the development of the Bauer rig conversion coefficients for both torque and crowd. The Bauer BG 30 drill rig was confirmed to use a multi drive system with two available gears. This is important because K coefficients for torque need to be derived for each gear that is used during drilling. However, second gear on the Bauer rig provides higher rotational speeds with less available torque and is only used for spinning off material from the auger bit. In order to monitor the torque, only first gear needed to be considered as this was the gear used for drilling. This was confirmed by the contractors lead drill rig operator. Once the gear setup was confirmed, the next step was to determine the maximum torque, crowd, and hydraulic pressure s available within were able to be determined and confirmed during rig inspection. The maximum crowd was determined to be 330 kN using an operating pressure of 320 bar or 32 MPa. The maximum torque was determined to be 300 kN m using an operating pressure of 350 bar or 35 MPa. The following shows how the values were determined:
94 Figure 3 9 Bauer BG 30 crowd specs from serial plate, max crowd and operating pressure (Pho to courtesy of author). Figure 3 10 Bauer BG 30 specs from serial plate, showing model type and operating pressure (Photo courtesy of author).
95 Figure 3 11 Bauer BG 30 spec sheet, showing torque operating pressure and maximum crowd Figu re 3 12 Bauer BG 30 specs, showing model type and maximum torque Once the needed parameters were determined, the conversions were made using the following equation: Torque or Crowd = K (Operating Pressure Threshold Pressure) ( 3 1 ) The threshold pressure for the Bauer BG30 rig was 5 Bar, determined by checking the pressures recorded in hydraulic lines on the DIALOG and B tronic while the bit was at rest, i.e., no rotation or penetration. This provided a single equation with a single unknown that could be solved straightforward. The following are the known parameters: Max Torque for 1 st gear
96 Maximum Torque = 300 kN m Maximum Crowd = 330 kN Operating Pressure for Torque = 350 Bar Operating Pressure for Crowd = 320 Bar Threshold Pressure = 5 Bar For Crowd: 330 kN = K (320 bar 5 bar), solving for K provides, K = 1.0476 For Torque: 300 kN m = K* (350 bar 5 bar), solving for K provides, K = 0.8696 With the installation, calibration, and conversion coefficients derived, the rig is now ready to begin monitorin g. Duri ng the shaft installations, the DIALOG continuously records the monitored d rilling parameters at a 1,000 Hz sampling rate. Each recorded value is then averaged for every 2 centimeters of penetration. This provides an average value every 2 centimeters tha t is comprised of hundreds to thousands of recorded values for each drilling parameter. The following provides depth vs. drilling parameter plots displaying the large quantity of values that are recorded for each drilling as the drill bit is advanced. Fr equency distributions are also provided to show the spread of data and the variability of each drilling parameter. As previously discussed, crowd and toque must be converted to physical measures using the developed conversion coefficients. Therefore, fre quency distributions are provided for the raw data recorded for both torque and crowd, as hydraulic pressu res, as well as distributions showing the conversions to physical measures.
97 Figure 3 13 Elevation vs. rotational speed Figure 3 14 Rotation al speed frequency distribution 10 15 20 25 30 35 40 45 50 55 60 0 5 10 15 20 25 30 Elevation (ft) Rotational Speed, N (RPM) 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 Frequency Rotational Speed, N (RPM) Stats N (RPM) Average 14.8 Median 10.7 Maximum 26.0 Minimum 0.0 Std. Dev. 5.7 CV 0.384 Count 977
98 Figure 3 15 Elevation vs. penetration rate Figure 3 16 Penetration rate frequency distribution 10 15 20 25 30 35 40 45 50 55 60 0 5 10 15 20 25 30 35 40 Elevation (ft) Penetration Rate, u (in/min) 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 60 100 140 180 220 260 Frequency Penetration Rate, u (in/min) Stats u (in/min) Average 6.0 Median 4.5 Maximum 277.9 Minimum 0.0 Std. Dev. 11.8 CV 1.98 Count 977
99 Figure 3 17 Elevation vs. crowd Figure 3 18 Crowd raw data frequency distribution 10 15 20 25 30 35 40 45 50 55 60 0 5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 Elevation (ft) Crowd, F (lbf) 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 35 40 45 50 55 60 65 70 75 80 85 90 95 100 105 110 115 120 125 130 135 140 145 150 155 160 Frequency Crowd, F (Bar) Stats Crowd (Bar) Average 96.2 Median 95.6 Maximum 155.3 Minimum 32.8 Std. Dev. 18.9 CV 0 .196 Count 977
100 Figure 3 19 Crow d converted frequency distribution Figure 3 20 Elevation vs. torque 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18 5,100 6,442 7,784 9,126 10,468 11,811 13,153 14,495 15,837 17,179 18,521 19,863 21,205 22,547 23,890 25,232 26,574 27,916 29,258 30,600 31,942 33,284 34,626 35,969 37,311 38,653 Frequency Crowd, F (lbf) 10 15 20 25 30 35 40 45 50 55 60 0 500,000 1,000,000 1,500,000 2,000,000 Elevation (ft) Torque, T (in lbs) Stats F (lbf) Average 21,537 Median 21,370 Maximum 37,387 Minimum 4,514 Std. Dev. 5,066 CV 0.235 Count 977
101 Figure 3 21 Torque raw data frequency distribution Figure 3 22 Torque converted frequency distribution 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0 15 30 45 60 75 90 105 120 135 150 165 180 195 210 225 240 255 270 Frequency Torque, T (Bar) All T Values 1st Gear 2nd Gear 0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.10 0 62,000 124,000 186,000 248,000 310,000 372,000 434,000 496,000 558,000 620,000 682,000 744,000 806,000 868,000 930,000 992,000 1,054,000 1,116,000 1,178,000 1,240,000 1,302,000 1,364,000 1,426,000 1,488,000 1,550,000 1,612,000 1,674,000 Frequency Torque, T (in lbs) All T Values 1st Gear 2nd Gear Raw Data Torque, T (Bar) Stats All 1st Gear 2nd Gear Average 146.4 118.6 194.3 Median 142.1 106.3 192.2 Maximum 265. 5 265.5 254.4 Minimum 10.7 10.7 112.7 Std. Dev. 52.7 40.9 32.2 CV 0.360 0.345 0.166 Count 977 618 359 Converted Torque, T (in lbs) Stats All 1st Gear 2nd Gear Average 721,664 743,186 684,613 Median 672,084 666,418 677,194 Maximum 1,663,946 1,663 ,946 896,036 Minimum 66,956 66,956 397,032 Std. Dev. 216,959 256,342 113,582 CV 0.301 0.345 0.166 Count 977 618 359
102 In Figure 3 22 there are torque distributions provided for eac h gear that was used with the multi geared IMT drill rig. When the raw data is presented it appears that the 2 nd gear torque values are higher than the 1 st gear values. This is counterintuitive to what was found in the lab; as 2 nd gear produces higher ro tational speeds and the lab results indicated that higher rotational speeds produce lower torque values. The misleading result stems from each gear requiring a different K coefficient to make the conversions from hydraulic pressures to physical measures. Once the conversions are applied using a unique K coefficient for each gear, the 2 nd gear torque values are reduced and the results are in agreement with laboratory findings. This can be seen in the summary of statistics box that indicates 1 st gear, whic h produces lower rotational speeds, provided higher torque values than 2 nd gear on average. With each drilling parameter properly converted, monitoring is ready to take place. Although, in order to produce average compressive strength values in real time, for every 2 centimeters of penetration, the laboratory drilling equation needs to be rearranged to solve for q u directly using the recorded drilling parameters. Therefore, using equation from the e vs. q u plot in Figure 2 66 y = 0.0066x 2 + 13.681 where, y = e (psi) x = qu (psi) setting the equation equal to zero, 0.0066x 2 13.681x y = 0 using the Quadratic solution to rearrange the equation,
103 (3 2 ) and substituting terms in for a, b, and c, (3 3 ) q u can now be solved directly usin g the drilling parameters with E quation 3 3. Analysis of Rock Strength from Real Time Field Monitoring Throughout the course of the research, three field monitoring opportunities were presented which provided the first field monitoring trials using the labor atory derived drilling equation The locations were a t the Little River bridge site in Quincy, FL, the FL. The sites were chosen because each location had planned drilled shaft installations with subsequent load testing. In addition, all of the test shafts were instrumented with strain gauges along their length to assess skin friction by layer. This provided a means to directly compar e compressive strength and shaft capacity estimates, obtained from monitoring, with reco vered core samples and the actual measured capacity in mobilized portions of each shaft using conventional methods Additionally, each location used a different type of load testing, O cell testing at Little River, top down static load testing at Kanapaha and Statnamic testing at Overland. This provided direct comparative data from three of the most conventional load testing methods used throughout the state. This also provided field monitoring with three variations in the following categories: location, shaft diameter, drill rigs used to install the shafts, drilling crews, drill bits and drill bit tooth configurations. These variable drilling parameters provided great insight as to how the laboratory drilling equation held up when drilling conditions an d rig
104 configurations changed. This section and the remainder of the chapter will cover the comparative analysis of compressive strength values, q u obtained from monitoring and recovered core samples tested in the laboratory at the State Materials Office, SMO. Comparative analysis of skin friction and shaft capacity estimates will be covered in the following chapter. Using the laboratory developed drilling equation, Equation 3 3, measurements of rock strength, q u were obtained at the three monitored loca tions during the installation of each test shaft. Core samples obtained within 100 feet of each test shaft, and tested at the SMO, were used for comparison. This provided comparative core data from four borings obtained at Little River, nine borings at K anapaha, and five borings at Overland. At each location, different degrees of subsurface and site variability were experienced, as well as rock core recoveries, REC%. This provided an excellent opportunity to show the true benefit and insight monitoring drilled shafts installations can provide. At Little River, core data obtained from the entire site indicated there was a high degree of variability, as the coefficient of variability, CV, was equal 1.81. At the site, the subsurface strata was interlaced w ith over consolidated clays, intermediate geo material, IGM, and limestone with a wide range of compressive s trengths, spanning from 4 to 4,3 00 psi. However, the mean core sample recovery was very good, REC% = 85%, providing 37 core samples for comparison within the investigated depth range. Figures 3 23 and 3 24 provide the q u frequency and cumulative frequency distributions for both the recovered core samples and the va lues obtained from monitoring.
105 Figure 3 23 Little River q u frequency distributio n Figure 3 24 Little River q u cumulative frequency distribution 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 40 80 120 160 200 240 280 320 360 400 440 480 520 560 600 Frequency Unconfined Compressive Strength, qu (tsf) Core Data Monitoring 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 40 80 120 160 200 240 280 320 360 400 440 480 520 560 600 Frequency Unconfined Compressive Strength, qu (tsf) Core Data Monitoring Little River qu (tsf) Stats Core Data Monitoring Average 51.4 56.8 Std. Dev. 68.5 70.7 CV 1.33 1.25 Median 7.9 32.5 Max 253.1 542.9 Min 0.3 0.6 Count 37 415
106 In the summary of statistics box, the average q u value obtained from monitoring is in good agreement with the average obtained from the laboratory tested core samples. Additionally, the frequency and cumulative frequency distributions align very well with one another. This indicates that the recovered core samples, used for standard drilled shaft design, are reflective of the strata encountered during the shaft installation, i.e., drill ing. Interestingly, both data sets indicated nearly the same variability, determined by the reported CV values, within the investigated depth range; even though the monitoring provided 415 q u values compared to the 37 recovered cores. This leads to the i dea that both data sets provided the true distribution of rock strength at the site, and the small discrepancy between the reported strength averages and CV values are a result of sampling location. As stated, borings within 100 feet of each test shaft we re used for comparison. At Little River, none of the borings were obtained within the footprint of the monitored test shaft and two of the borings were in average rock s trength is an inherent result of site variability. F or example, if the four Little River core borings were recovered in different quadrants within the footprint of shaft, Figure 3 25 and 37 core samples were recovered in each quadrant ; it is unlikely tha t each boring would produce the same distribution of rock strength. However, the averages obtained should b e fairly similar to one another and with the average obtained from monitoring the shaft installation Since two of the borings were completed appro ximately 80 feet away from the monitored shaft; the difference in average compressive strength is likely a reflection of rock strength variability associated with changes in sampled depth and the horizontal distance between sampled locations.
107 Therefore, t he sampled core data should produce an increased CV value compared to monitoring as the site variability was CV = 1.81 which was a result of changes in strength due to sampled depth and the horizontal distance between sampled locations. Figure 3 25 Vis ualization of core data variability At Kanapaha, there were a total of nine borings performed in close proximity to each of the three shafts installed for the top do w n static load test. Three of the borings were performed within the footprint of each sha ft. The remaining six borings were performed within five feet from the center of each shaft, providing three borings at each shaft location in close proximity, i.e., similar to Figure 3 25 However, only 19 core samples were obtained in the inve stigated depth range due to poor r ecoveries, average REC% = 30%. From previous site investigation completed at Kanapaha, a host of CPT, SPT, auger boring, and core boring data was available to begin the search for an ideal load test location. Previous load test da ta obtained in 1993 was also available. Throughout
108 the course of the new site investigation, 12 seismic test lines, over 20 CPTs, 15 core borings, and 5 SPTs were completed in the search for a viable location, Figure 3 26 Figure 3 2 6 Kanapaha site in vestigation locations The incredible amount of site investigation performed was due to the highly variable and highly weathered nature of the limestone. At the Kanapaha site, the Ocala Limestone formation is encountered, which is one of the oldest format ions found in Florida formed approximately 35 million years ago Additionally, the site is located within the Ocala uplift which provides a karst landscape that is very cavernous due to a high degree of weathering. In most locations where seismic testin g indicated rock was present, SPT and core runs indicated the same. However, the extracted limestone was mostly in a granular form due to the high degree of weathering which compromises the integrity of the rock matrix holding the rock mass together Th is is because the rock matrix is generally the most permeable portion of the rock mass. As a result, acidic rain water and groundwater both erode the carbonate rock matrix which serve s as the -250 -200 -150 -100 -50 0 -100 0 100 200 300 400 500 600 North to South (ft) East to West (ft) CPT SPT Core Borings Auger Borings Seismic Lines Resistivity Fence Lines N
109 binding material, holding the rock mass together For example, the seismic results for EW line 3, Figure 3 27 indicated limestone should be encountered between depths 30 to 35 feet. Figure 3 2 7 Seismic line EW 3 results for P wave, S wave, and Poisson ratio From the SPT run, limestone was encountered a t the predicted depth, however the limestone was highly weathered with the recovered samples obtained in a granular Seismic results displaying higher waves speeds which is indicative of rock beginning around 30 to 35 feet
110 form rather than an intact rock mass as previously discussed. This can be seen in Figures 3 28 and 3 29 Figure 3 28 Limestone recov ered at a depth of 30 feet with grey clay at the top of the spoon (Photo courtesy of author). Highly Weathered Limestone
111 Figure 3 29 Highly weathered limestone at a depth of 30 feet (Photo courtesy of author). In addition to poor intact specimen recoveries encountered throughou t the site, the planned load test shaft diameters were 36 inches or 3 feet. This meant that in order to adhere to ASTM guidelines for a top down static load test, the clear distance between shafts needed to be 15 feet to provide the required 5 D spacing. In total this called for a 39 foot span of competent limestone at nearly the same depth with a similar layer thickness in all three shaft locations. It took a year of thorough site investigation to find an acceptable location which signifies the difficu lties of shaft design in locations with poor recoveries such as Kanapaha The following provides the summary of statistics, frequency distributions, and cumulative frequency distributions for the Kanapaha site.
112 Figure 3 3 0 Kanapaha q u frequency distr ibution Figure 3 3 1 Kanapaha q u cumulative frequency distribution 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 5 10 15 20 25 30 35 40 45 50 55 60 65 Frequency Unconfined Compressive Strength, qu (tsf) Core Data Monitoring 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 10 20 30 40 50 60 70 Frequency Unconfined Compressive Strength, qu (tsf) Core Data Monitoring Kanapaha qu (tsf) Stats Core Data Monitoring Average 20.5 14.1 Std. Dev. 9.9 12.7 CV 0.481 0.897 Median 20.4 8.6 Max 39.1 62.5 Min 5.7 0.3 Count 19 430
113 In Figure 3 30 the summary of statistics shows that the average compressive s trength obtained from core sampling is nearly 40 percent higher than the average obtained from monitorin g. The frequency distributions indicate that core sampling did not recover any of the high end or low end q u values present at the site. This leads to the assumption that due to poor recoveries, the reported core data d istribution is only a portion of th e true distribution within the 39 foot span Typically, this results in an overestim ation of compressive strength because only the most competent limestone is recovered which is generally hig her strength material Therefore, when recoveries are low, est imating rock strength using only the recovered compressive strength samples is inadequate to define the true site variab ility and average rock strength The lack of compressive strength data recovered is also a result of the intact core sample size requir ed to adhere to ASTM guidelines. The guidelines state the sample must provide a 2:1 length/diameter ratio in order to be subjected to unconfined compression loading It can be difficult to obtain core samples that meet the 2:1 ratio at sites like Kanapah a, where a high degree of weathering compromises the likelihood of recovering larger intact samples. Reported recoveries can also be misleading. For example, if the reported recovery of a core run was 20%, it does not mean the remaining 80% of the core r un did not encounter rock i.e., a voided section The reported recovery only takes into account the competent material that was extracted and disregards the lower strength material that may have been compromised during coring and was not extracted. Cons equently, 80% percent of the geomaterial present is not accounted for and results in recovered data that can be quite misleading. Therefore new techniques need to be developed to account for the
114 during this research Rodgers m ethod, uses split tension and REC% data to generate more compressive strength values. Using over 1,200 q t core samples and 700 q u core samples collected from 23 different sites all around Florida; it was found that split tension and unconfined c ompressive strength show an excellent trend when plotted vs. dry unit weight. The following provides the list of sites where the core samples were recovered. 17th Street Causeway Acosta Bridge BR720153 SR 9 (I 95) Overland CR 326 @ Waccasa River HEFT / S R 874 PD&E I 295 Buckman Bridge I 295 Dames Point Bridge I 95 @ I 295 Cloverleaf I 95 Fuller Warren Bridge Jewfish Creek MIC People Mover Project NW 12th Ave (SR 933) Miami River Bridge NW 36th Street Bridge Pump Station at Bal Harbour (96th St & Indian C reek) Radio Tower Everglades Academy (Florida City) SR 10 @ CSX RR (Beaver St. Viaduct), Duval Co. SR 20 @ Lochloosa Creek, Alachua Co. SR 25 @ Santa Fe River SR30/US98 @ Aucilla River (District 3) SR 9 (I 95) Overland Bridge US 90 Victory Bridge (District 3) Verona Ave Bridge Over Grand Canal Wall At Service Road South of Snake Creek The measured q t and q u values from all of the sites combined were grouped by dry unit weight with in a 5 pcf range, e.g., 105 t o 110 pcf, 110 to 115 pcf, etc ., and plotted as dry unit weight vs. split tension and unconfined compression strength separately. Next, the mean and standard deviation were calculated for each strength value within each unit weight range. The standard deviation was used to establish an
115 rength range, one standard deviation above and below the mean, for predicted q t and q u values within each respective unit weight range. Figure 3 3 2 Dry unit weight vs. average q t values from 23 Florida sites Figure 3 3 3 Dry unit weight vs. avera ge q u values from 23 Florida sites y = 40.953x 0.2058 R = 0.9799 85 95 105 115 125 135 145 155 165 175 0 200 400 600 800 1,000 1,200 1,400 Dry Unit Weight (pcf) Splitting Tensile Strength, qt (psi) 95 100 pcf 100 105 pcf 105 110 pcf 110 115 pcf 115 120 pcf 120 125 pcf 125 130 pcf 130 135 pcf 135 140 pcf 140 145 pcf 145 150 pcf 150 155 pcf 155 160 pcf 160 165 pcf Average y = 40.75x 0.1581 R = 0.9862 85 95 105 115 125 135 145 155 165 175 0 2,000 4,000 6,000 8,000 10,000 12,000 Dry Unit Weight (pcf) Unconfined Compressive Strength, qu (psi) 95 100 pcf 100 105 pcf 105 110 pcf 110 115 pcf 115 120 pcf 120 125 pcf 125 130 pcf 130 135 pcf 135 140 pcf 140 145 pcf 145 150 pcf 150 155 pcf 155 160 pcf 160 165 pcf Average
116 From Figures 3 32 and 3 33 the following plot was developed where each data point is the average q t /q u ratio from each dry unit weight range. Figure 3 3 4 q u vs. q t from 23 Florida sites (data points from dry unit weight groupings). Using the re gression equation in Figure 3 34 unconfined compressive strength can now be estimated using split tension data and checked using Figure 3 33 to ensure the estimated strength falls within the acceptable range This provides a significant increase in compressive strength values available for design ; as split tension samples can be tested at a 1:1 length/diameter ratio and therefore provide an increas ed number of testable samples. This is quite evident from the number of q t sa mples available from the 23 Florida sites compared to the number of q u samples. There was nearly twice as many q t samples tested as there were q u samples, i.e., 1,200 q t samples compared to 700 q u samples. However, q t samples are also included in the rec overy reported for each core run and the lower strength material present at the site but not recovered is still not accounted for. In order to account for the lower strength material, the recovery y = 1.5847x 1.2254 R = 0.9947 0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 0 100 200 300 400 500 600 700 800 900 1,000 Unconfined Compressive Strength, qu (psi) Splitting Tensile Strength, qt (psi)
117 of each c ore run is used to generate low end and interme diate q u values. Each q u value, measured and estimated using q t is multiplied by the respective recovery of the core run. This leads to twic e the amount of data generated and produces low end values that are not r eco vered. For example, Figure 3 35 disp lays the core data recovered from 45 to 50 feet in the west reaction shaft at Kanapaha. Indicated in red are the values generated using the new q t q u rela tionship, i.e., Figure 3 34 and the values generated using the recoveries of each core run. Boring q t (tsf) qu (tsf) REC% qu*REC (tsf) 21 25.4 50 12.7 3.7 14.3 7.2 3.0 10.9 5.4 39.1 19.6 4.5 18.0 9.0 5.4 22.4 11.2 25 8.7 40.8 48 19.6 34.4 16.5 10.1 48.6 23.3 4.6 18.7 9.0 8.1 37.2 17.9 1.6 5.1 2.4 28 7.6 34.5 29 10.0 3.2 12.1 3.5 3.5 13.5 3.9 Statistics qt (tsf) qu (tsf) REC% qu*rec (tsf) Average 5.3 25.0 42.3 11.4 Std. Dev. 2.7 13.2 11.6 6.6 CV 0.499 0.528 0.274 0.579 Count 12 15 3 15 Maximum 10.1 48.6 50 23.3 Minimum 1.6 5.1 29 2.4 Figure 3 3 5 q u co re data with additional q u values generated using Rodgers m ethod. Using the new approach, 115 q u values wer e generated from the original 19 q u values. From the new data set, a frequency and cumulative frequency distribution were
118 developed and compared to the original core data set and the monitoring results, Figures 3 36 and 3 37 labeled MBR core data. Figure 3 3 6 Kanapaha q u frequency distribution using Rodgers method. Figure 3 3 7 Kanapaha q u cumulative frequency distribution using Rodgers metho d. 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 5 10 15 20 25 30 35 40 45 50 55 60 65 Frequency Unconfined Compressive Strength, qu (tsf) Core Data Monitoring MBR Core Data 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 10 20 30 40 50 60 70 Frequency Unconfined Compressive Strength, qu (tsf) Core Data Monitoring MBR Core Data Kanapaha qu (tsf) Stats Core Data Monitoring MBR Core Average 20.5 14.1 14.2 Std. Dev. 9.9 12.7 10.6 CV 0.481 0.897 0.746 Median 20.4 8.6 12.1 Max 39.1 62.5 61.8 Min 5.7 0.3 1.2 Count 19 430 134
119 The generated values were checked to ensure they fe ll within the range of values obtained from monitoring and the acceptable range of Figure 3 33 As well, the values were compared to q u values extracted from the 23 Florida sites where the Ocala L imest one formation was encountered. All generated values were found to be acceptable and within the range of strengths obtained from monitoring and present within the Ocala L imestone formation. From Figures 3 36 and 3 37 the new method appears to work very well. T he frequency distribution s, average compression strength and the CV value are more reflective of the monitoring which provides the true site conditions confirmed by load testing However, the method needed to be checked for sites with good recove ries. Therefore, the same procedure was conducted using the Little River data If the method is correct, the average compressive strength should be approximately the same as the original core data set and the CV value should more ref lective of the site C V = 1.81. Figure 3 38 Little River q u frequency distribution using Rodgers method. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 40 80 120 160 200 240 280 320 360 400 440 480 520 560 600 Frequency Unconfined Compressive Strength, qu (tsf) Core Data Monitoring MBR Core Data Little River qu (tsf) Stats Core Data Monitoring MBR Core Average 51.4 56.8 51.0 Std. Dev. 68.5 70.7 79.4 CV 1.33 1.25 1.56 Median 7.9 32.5 4.3 Max 253.1 542.9 398.6 Min 0.3 0.6 0.0 Count 37 415 224
120 Figure 3 39 Little River q u cumulative frequency distribution using Rodgers method. From Figures 3 38 and 3 39 the new approach provided the desired results. The method filled in the gaps of the core data distribution, generated high end values that were not recovered during core sampling, produced a nearly identical average compressive strength value and increased the CV value to be more reflective of the site va riability. Again, the core data CV should be higher than the monitoring CV because the monitoring variability is only based on changes in strength associated with changes in depth; whereas the core data variability accounts for changes in strength due to changes in depth and the horizontal distance between sampled locations. Since the site variability is a reflection of 8 different s ampled locations the CV would be expected to be higher than the variability of a single boring location. Therefore the cor e data distribution used in comparison should be more reflective of the site variability than the monitoring variability, which was obtained at a single location. 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 40 80 120 160 200 240 280 320 360 400 440 480 520 560 600 Frequency Unconfined Compressive Strength, qu (tsf) Core Data Monitoring MBR Core Data
121 F ollowing standard practice, the original Kanapaha core data set would be used for design whi ch would lead to an overestimation of shaft capacity. Although, due to the poor recoveries, the overestimation of shaft capacity would likely be compensat ed for by factor. This inherently provides a large factor of safety for the shaft designs. Unfortunately providing a large factor of safety comes with a large increased cost per shaft installed at the site. Mor factor and increased cost per shaft do not resolve the issues associa ted with the site variability. This reinforces the importance of monitoring drilled s haft installations. Although the designed shaft s may be acceptable in locations where load testing occurs, there is no assurance the shaft designs will meet the needs of the rest of the site. This concept was quite relevant at the Overland bridge site in Jacksonville, Florida. Due to the high degree of variability and poor recoveri over loaded to four times the design capacity and only a few sections of the three load test ed shafts were mobilized. This confirms the shafts are more than capable of supporting the expected loading in the three load tested locations and provides reliability for the design. However, the reliability of the design was achieved at a significant cost per shaft and does nothing to ensure the same s uccess will be achieved throughout the remainder of the site. At the Overland bridge site, monitoring indicated the highest degree of variability was encountered from all three sites, CV = 2.54. As well, the site produced the worst average recovery, REC% = 17%. In fact, the recoveries were so poor within the investigated depth range, which provides direct comparison with load test results, there
122 were only five core samples recovered; even though four of the five borings were performed within 10 feet of th e test shaft. Consequently, the depth range for the compressive strength comparative analysis was extended an additional seven feet to provide at least 10 core samples. In total, 12 core samples were available from the core data set, whereas monitoring p rovided a total 165 values for comparison. Although, 85% of the monitoring values were lower than the lowest core sample recovered Figure 3 40 Indicating the core sampling only recovered the most competent limestone at the site, which tends to be the h ighest strength material present. Since the recoveries were so poor and the recovered material was not representative of the true site conditions, performing Rodgers method provided limited improvement in developing a more representative q u distribution. This implies that current site investigation practices are inadequate for sites such as Overland Figure 3 4 0 Overland q u frequency distribution 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 Frequency Unconfined Compressive Strength, qu (tsf) Core Data Monitoring Overland q u (tsf) Stats Core Data Monitoring Average 127.3 13.9 Std. Dev. 85.7 35.2 CV 0.674 2.54 Median 90.4 4.8 Max 268.8 303.9 Min 30.8 0.4 Count 12 165
123 Figure 3 4 1 Overland q u cumulative frequency distribution When comparing both of the frequency distr ibutions, it is easy to see the drastic difference between the two data sets. From the comparative analyses performed using the Little River and Kanapaha data sets, it can be stated that monitoring provides the true distribution of rock strength encounter ed at each site. Therefore, the core data set at Overland is not even remotely close to the true rock strength distribution and does a poor job of defining the site variability. As well, t he monitoring compressive strength estimates were confirmed to be correct based on skin friction results provided by both load testing and monitori ng. Both methods were in near perfect agreement and reported the unit side shear, or skin friction, to be approximately 1.0 tsf within the investigated depth range A unit s ide shear value this low is indicative of very weak rock or IGM, which was likely encountered. Present in Jacksonville, and along the east coast of Florida, is a young limestone formation of undifferentiated sediments deposited in the Holocene era. Undif ferentiated sediments are poorly cemented limestone 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0 50 100 150 200 250 300 350 Frequency Unconfined Compressive Strength, qu (tsf) Core Data Monitoring
124 formations that have not had time to fully develop, e.g., it is comparable to early strength concrete that has not had time to properly cure. In Jacksonville, the young formation rests atop the Hawthorne Group, Figure 3 42 Therefore, the lower strength mat erial was likely the undifferentiated sediments formation and the higher strength material was likely from the Hawthorne group. The q u profile obtained from monitoring the shaft installation supports this idea Figure 3 43 Figure 3 4 2 East west cross section A 12 Figure 3 4 3 Overland elevation vs. q u plot, q u reported in psi -28 -27 -26 -25 -24 -23 -22 -21 -20 -19 -18 0 200 400 600 800 1,000 1,200 1,400 1,600 Elevation (ft) Unconfined Compressive Strength, qu (psi) Segment 2 Segment 3 qu (psi) Statistics Segment 2 Segment 3 Average 58.1 185.6 Std. Dev. 38.8 263.8 CV 0.669 1.421 Count 68 68 Undifferentiated Sediments Hawthorne Group
125 In Figure 3 43 a clear distinction of strength is noticed between the two monitored se gments of the shaft. Segment 2 was fully mobilized during the load test, and was the only portion of the three Overland test shafts that provided a skin friction comparison with the monitoring results. The lack of comparative load test data was a result of the test shafts being overdesigned which stems from a l ack of data available to properly design the shafts As stated, there were only four q u core samples obtained in the Segment 2 elevation range within 100 feet of the shaft. Two of the four core sa mples were obtained from the four closest borings that were not performed as a part of the shaft design process for the site These borings were performed specifically for this research to provide comparable rock core data which reinforces the notion tha t conventional site investigation and design methods are inadequate for sites with a high degree of variability and where highly weathered or underdeveloped rock formations exist, i.e., Kanapaha and Overland. Therefore, the developed methods for measuring q u during drilling should also be developed as a site investigation tool. This would provide a much better understanding of the true site conditions and alleviate the need to design drilled shafts with a significantly large factor of safety, which greatl y drives up the cost per shaft As discussed throughout this section drilled shaft monitoring eliminates spatial variability concerns in the footprint of each shaft and provides the tr ue rock strength distribution This inherently provides shaft capacity estimates that nearly perfectly align with results obtained from load testing. This is of great importance because load testing is the conventional method for ensuring proper shaft design and is generally the only means of improving LRFD resistance facto rs. Methods for predicting shaft capacity
126 during monitored drillings are covered in the f ollowing chapter, as well as a comparative analysis between the monitoring and load test capacity estimates.
127 CHAPTER 4 DRILLED SHAFT FIELD MONITORING TO MEASURE SHAF T CAPACITY In drilled shaft design, the capacity of each shaft is developed from a combination of end bearing, q b and side shear, f s resist ance mechanism as shown in Figure 4 1 Figure 4 1 Drilled shaft load transfer diagram Load transfer through skin friction is the result of sliding friction along the side of the shaft, and a combination of cohesion and adhesion at the rock shaft inter face. Load transfer through end bearing is the result of compressive loading between the bottom of
128 the drilled shaft and the soil/rock. In Florida, it is common practice to design drilled shafts solely using skin friction. This is due to the complicatio ns associated with load transfer through end bearing. End bearing relies on a thick layer of uniform limestone to be present beneath the base of the shaft, with the same strength, to ensure proper load transfer. Throughout this report, there has been much discussion on the high degree of site variability that is encountered when drilling into Florida bedrock. As stated, it took nearly a year of thorough site investigation to find a 39 foot span of competent limestone, 10 feet thick, at Kanapaha. Dependin g on the shaft diameter and length, as well as the strength of the rock, a layer thickness of 10 feet may not be sufficient to ensure proper load transfer th r ough end bearing. This leads to an unreliable design. As well, proper load transfer through end bearing heavily relies on construction practices. To ensure proper load transfer, the drilling contractor must take special measures to ensure the base of the excavation is free of loose debris, as well as provide a uniform flush surface at the rock shaft tip interface. This can be quite difficult as some drilled shafts in Florida reach depths over 100 feet, according to the conducted drilled shaft survey. This too compromises the reliability of the shaft design. Finally, even if a competent limestone l ayer could be ensured beneath the base of the shaft, and the base of the shaft was properly cleaned and leveled off; the shaft displacement associated with end bearing mobilization is unacceptable for foundation design. It typically requires several inche s of displacement to fully mobilize the end bearing of a drilled shaft. Since one inch of displacement is generally the acceptable limit for
129 foundation settlement, relying on the available strength of end bearing can lead to failure of the structure su ppo rted by the drilled shaft. Conversely, the displacement associated with skin friction mobilization is typically between 0.2 to 0.4 inches. Therefore, if the engineer can ensure this range of foundation settlement; it inherently creates a factor of safety ranging from 2.5 to 5, based on the general foundation settlement limit of one inch. As a result, Florida drilled shaft design typically only considers the available load transferred to the so il/rock through skin friction. When designing drilled shafts, t here are large number of methods for estimating skin friction; and typically each equation is only used for a specified soil type, e.g., sand, clay, or rock. In most cases, skin friction equations developed for rock and intermediate geomaterial, IGM, use compressive strength, q u to estimate the side shear capacity. The equations are generally formed using empirical methods and presented in the following two ways: Using a linear function to develop the equation, (4 1 ) Or using a power function to develop the equation, (4 2 ) where, a and b are empirical constants developed using load test data fro m instrumented drilled shafts. In the pursuit of an a ccurate method to determine skin friction in real time during field monitoring; a large number of leading equations used in Florida drilled shaft design were considered for the analysis, presented in Table 4 1.
130 Table 4 1 Drilled Shaft design skin frictio n equations 13 21 Method Author Design Methodology 1 McVay et al. 13 2a 14 (tsf) 2b Horvath a nd Kenney 15 (tsf) 3 Williams et al. 16 (tsf) 4 Reynolds and Kaderabek 17 (tsf) 5 Gupton and Logan 18 (tsf) 6 Carter and Kulhawy 19 (tsf) 7 Ramos et al. 20 (< 36 ksf) (> 36 ksf) 8 Rowe and Armitage 21 (tsf) clean sockets (tsf) rough sockets A seen in Table 4 1 all methods accept for McVay et a l. 13 use only q u to estimate skin fri ction. McVay et al. 13 incorporates split tension strength to estimate skin friction which ta kes the material formation, q t /q u into consideration Using a parametric finite element method, McVay investigated the maximum side shear at the rock shaft interface, where he indicated the cohesion of rock is a closely approximated underestimate of the s ide shear, Figure 4 2 However, solving for the model presented in Figure 4 2 to determine the cohesion requires knowledge of the frictio n angle, displayed in Equation 4 3, which is not readily available. In order to determine the cohesion of the rock mo re than one laboratory test would need to be performed. For example, multiple triaxial compression tests at different confining pressures would be one option, but it is a very time consuming process. Alternatively, McVay proposed a more simplistic approa ch using results from
131 unconfined compression and split tension testing implemented on field cores, which are readily available from typical site investigation. He found that the failure of rock can be described through a Mohr Coulomb strength envelope; le ading to the development of an alternat ive model based on split tension and compressive strength test data presented in Figure 4 3 Figure 4 2 Summary of stress states at rock shaft interface 13 Where, = Normal Stress (tsf) = Shear Stress (tsf) c = Cohesion (tsf) f su = Skin friction or side shear (tsf) = Friction angle (degrees) and, (4 3 )
132 Figure 4 3 Strength envelope for Florida limestone 13 Based on the new model, the following equation development was completed: (4 4 ) Simplifying, (4 5 ) solving for (4 6 ) substituting ( 4 5 ) into ( 4 6 ) (4 7 ) using the Pythagorean theorem,
133 (4 8 ) substituting Equation s 4 7 and 4 8 into Equatio n 4 3 and simplifying, the final equation is obtained: (4 9 ) When compared to conventional t est methods, McVay et al. 13 found excellent agreement between results from Equation 4 9, using existing q u and q t data, and results obtained from 53 pullout tests and 7 load tests at 14 different sites in Florida. McVay et al. 13 provides a convenient and conservative method to determine the rock socketed skin friction using q u and q t data that is typically gathered d uring a general site investigation; and allows q t adjustments to be made based on the material formation determined by the q t /q u ratio. A dditionally, McVay et al. 13 was developed specifically for Florida limestone socketed shafts; whereas most developed e quations for estimating skin friction were developed for multiple rock types, including limestone, or rock types other than limestone, e.g., shales o r sandstones. McVay et al. 13 is also the recommended design method to determine skin friction for drilled shafts socketed into ation Handbook 22 Since McVay et al. 13 is the SFH recommended design method, measures were taken to ensure the equation could be used for drilled shaft field monitoring. This was the sole reason the e vs. q t equation was developed in the laboratory. However, the material formation of Gatorock was found to be a hybr id of concrete and limestone, as ACI 229 10 suggested providing a higher q t /q u ratio than is typical of Florida limestone. Th is was confirmed in preliminary skin friction analysis at Little River, where McVay et
134 al. 13 used with the laboratory developed e vs. q t equation, consistently produced overestimates. Consequently, alternative methods for determining the q t /q u ratio were investigated. The first method grouped pairs of q u and q t values within one vertical foot of each other, above and below, for each individual boring completed at the site. This was done in an attempt to provide a range of q t values for each recorded q u v alue. Once all the pairs were created, the q t /q u ratio was found for each pair. Any q t /q u value that fell outside of one standard deviation was removed. Remaining pairs from every boring location were then combined, removing the outliers, and used to pl ot q t vs. q u to determine the q t /q u ratio for the entire si te, Figure 4 4 Figure 4 4 Little River q t /q u analysis using all boring locatio n s with outliers removed y = 0.1375x R = 0.914 0 100 200 300 400 500 600 700 800 0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 Splitting Tensile Strength, qt (psi) Unconfined Compressive Strength, qu (psi) All Borings Stats qt (psi) qu (psi) qt/qu Mean 85.7 636.1 0.1368 Std. Dev. 146.1 1008.2 0.0334 CV 1.705 1.585 0.244 Count 87 87 87
1 35 Using the developed q t /q u ratio, skin friction was estimated in four segments along t he test shaft where load test results indicated the side shear was fully mobilized in layers of limestone The following provides the results: Table 4 2 Preliminary skin friction comparative analysis at Little River Little River Skin Friction, f s (ksf) Section Load Test Monitoring % Error SG7 to SG6 21.1 20.8 1.37% SG6 to O cell 20.6 22.9 11.31% O cell to SG5 21.4 20.4 4.81% SG5 to SG4 13.6 13.9 1.91% Average 19.2 19.5 1.76% As seen, using McVay et al. 13 with the developed q t /q u ratio provided an excellent result. In all four mobilized sections of the drilled shaft, the results obtained from monitoring the shaft installation were in near perfect agreement with the Osterberg load test results. This was quite impressive considering the variabil ity encountered at the Little R iver bridge site However, the validity and practicality of the method was in question. Using this a pproach, compressive strength values may be paired with split tension values from two dissimilar materials when developing the q t /q u ratio ; and the q t /q u ratio developed using unconfined compression does not reflect the confined condition experienced at the rock shaft interface for loaded drilled shafts. The former would provide inaccurate q t /q u ratios and shaft capacity esti mates in other locations with less available core data, e.g., Kanapaha and Overland. For example, in Figure 4 5 the highlighted portion indicates a q t q u pair that would be used to determine the average q t /q u ratio for a layer or site to provide q t for us e with McVay et al. 13 However, looking at the dry unit weights and moisture contents of the two test results, it is clear that these
136 are dissimilar materials and should not be combined and used to determine the q t /q u ratio. Boring/ Shaft Core Sample No. T est Type Moisture (%) Dry Density (pcf) Max Load (lbs) q t (psi) q u (psi) 2/1 2 1 T 59.8 63.3 39.6 6.5 2 U 58.2 64.7 128.6 38.1 3 T 60.1 63.2 66.0 8.1 4 U 69.0 59.7 61.1 16.9 5 T 3.2 156.0 9,520.2 975.2 6 T 3.8 135.0 3,180 .6 327.3 7 U 5.1 134.4 5,319.1 1,116.9 8 T 7.6 131.5 2,077.8 218.1 9 T 4.8 149.9 1,031.9 107.3 10 T 61.3 63.2 39.3 4.0 11 T 28.8 91.6 54.5 5.2 Figure 4 5 Core data from Little River indicating dissimilar materials This led t o a more theoretical approach to develop q t estimations criterion 4 which explains the difference s in uniaxial compressive shear strength in a confined and unconfined setting Development of the Florida Geomaterials Equation Johnston 4 investigated the strength of intact geomaterials, where he found that a number of strength criteria can describe the strength of geomaterials and that each criterion is typically limited to certain materials types with a limited range of stress conditions i.e., Mohr Coulomb criterion and Griffith criteria. Johnston proposed a new empirical strength criterion that was applicable to a wide variety of intact geomaterials, from lightly overconsolidated clays to very hard rock for both compressive and tensil e
137 stress regions. His new criterion demonstrated that the strength of these largely different geo materials followed a distinct progressive pattern. 4 4 a q t /q u vs. q u plot was developed by Anoglu et al. 23 for the concrete industry, indicating q t /q u ratios decrease as compressive strength increases and that the trend is nonlinear Figure 4 6 The Gatorock trending was in general agreement. Anoglu tested various concrete samples that were developed using differ ent water to cement ratios, binders, additives, cure times, and curing conditions. This is similar to the various limestone formations found throughout Florida. Each formation comprises different binding materials found within the rock matrix, such as cl ay found in north Florida that is not found in south Florida, various formation ages ranging from less than 1 million years to over 35 million years, various curing conditions such as changing sea level or the amount of overburden present above the formati on, as well as different skeletal remains left behind that act as the aggregate an d provide the main source of binder from calcite precipitate. Figure 4 6 q t /q u vs. q u for concrete 23
138 ort, he indicated the same relationship for tens ile strength and compressive strength can be determined for all geomaterials. 4 However, the q t /q u ratio developed using split tension and unconfined compression data would provide an inaccurate ratio, as platen friction effects induce the development of t en sile stresses within cor e sample s tested in unconfined compression This ultimately leads to axial splitting and a failure mode controlled by the smaller tensile strength of the material. Therefore, the failure involving axial splitting will give a re sult that can significantly underestimate the likely compressive shear strength and lead to higher q t /q u ratio s This was confirmed when the re lat ionship developed in Figure 3 34 was used to estimate q t for use with McVay et al. 13 which produced skin fr iction overestimates. Johnston noted in his report that his most important condition to satisfying his criteria was that compression test results had to be in reasonable agreement with all other test methods, particularly tri axial tests on similar specim ens displayed in Figure 4 7 where c Figure 4 4
139 As seen in Figure 4 7 unconfined compres sion testing i.e., quoted generally provide s a reduced compressive strength estimate compared to the regression curve that fits the trends of direct tension, indirect tension and triaxial test results at various degrees of confinement. As previously stated, this will result in an increased q t /q u ratio and provide overestimates for skin friction predictions when using Mcvay et al. 13 Therefore, u devel oped for Florida geomaterials. The equation s relationship for q t /q u ratio s 4 (4 10 ) where, q t = Uniaxial tensile strength (direct tension) q u = Uniaxial compression s trength (best fit q u Figure 4 7 ) B is a material parameter developed by Johnston that defines th e nonlinearity of the Mohr Coulomb failure envelope and is a measure of confinement effectiveness. B is independent of material type. M is also a Johnston material parameter and defines the changes in failure stresses associated wi th different geomaterial types, i.e., t he relationship between u that would be obtained from multiple triaxial tests. Johnston developed a single equation for B, ( 4 11 ) where q u is measured in kilopascals (kPa), and developed multiple equations for M based on material groupings such as; carbonate materials with w ell developed crystal cleavage, e.g., limestone and dolomite (4 12)
140 and lithified argillaceous materials with strong crystals and poo rly developed crystal cleavage, e .g. clay, claystone, and mudstone ( 4 13 ) With the understanding that Florida field drilling would likely pass through varying layers of over consolidated clays, IGMs, and limestone, both M equations were conside red for the development of the Florida geomaterials, q t /q u vs. q u relationship. Figure 4 8 were developed using q u values ranging from 1 to 10,000 psi, with data points plott ed for q u in increments of 10 psi, e.g., 10 psi, 20 psi, 30 psi, etc. The respective q t values were generated using the relationship q t = q u x (B/M). Note, the carbonate materials are labeled limestone and the lithified argillaceous materials are labeled clay. Figure 4 8. q t /q u vs. q u Evident from the developed regression equations, the shape of the two curves are different and the new curve developed for Florida geomaterials needed to account y = 0.659x 0.204 R = 1 y = 0.593x 0.229 R = 1 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000 10,000 qt/qu Unconfined Compressive Strength Limestone Clay
141 for this By developing the new equation to account for both material types; a reasonable approximation of splitting tensile strength will be generated in real time based on the measured compressive strength value acquired during the monitored drilling. Therefor e, regardless of the Florida geomaterial type encountered, a good approximation of skin friction will be report ed in real time as well. This ensures the as built foundation meets or exceeds the engineering design during construction. It was proposed tha t by using the q u data set from Little River, the shape of the two curves could be combined. As discussed, there was a high degree of variability within the q u core samples that were collected at Little River, CV = 1.81 The high degree of variability wa s a result of the multiple materials types present at the site. This included over consolidated clays, IGMs and limestone with a large compressive strength range, q u = 4.3 to 4,300 psi Therefore, by using the reported q u values from Little River, more emphasis would be placed in different portions of the q u range. This would be more representative of actual Florida site conditions instead of using equally spaced increments of q u to establis h the relationship. To clarify it is highly unlikely that the re would be an equal number of core samples obtained in the 9,000 to 10,000 psi q u range and the 1 to 1,000 psi q u range and the developed curve needed to account for this. As well, by placing more emphasis on low end q u values, from clays and IGMs, the a pproach should reshape the curve to better represent both material groupings. Since the focus of the study is on limestone, the M equation for limestone was used as the basis for development and the following presents the developed curve.
142 Figure 4 9. q t /q u vs. q u developed using only Little River q u data The new Florida geomaterials curve was then compared to the previously developed clay and limestone curves. Figure 4 10 Comparison of q t /q u vs. q u curves y = 0.545x 0.175 R = 0.995 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 4,500 5,000 qt/qu Unconfined Compressive Strength, qu (psi) y = 0.659x 0.204 R = 1 y = 0.593x 0.229 R = 1 y = 0.545x 0.175 R = 1 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000 10,000 qt/qu Unconfined Compressive Strength Limestone Clay FL Geomaterials
143 As seen in Figure 4 10 the shape of the limestone curve changed. However, the curve provided q t /q u ratios much higher than the clay and IGM curve, and were higher than the original limestone curve for q u > 1,000 psi. This was thought to be a result of J equations. Therefore, a correction factor would need to be applied to account for the differences in material formation of Florida limestone compared to limestone found elsewhere throughout the nation. In Florida, for the design of highway pavement materials; the Limerock Bearing Ratio, LBR, is used instead of the California Bearing Ratio, CBR, which is used nationwide. The only difference between the two tests is the compressive strength used to define the standard strength of limestone. For the CBR, q u = 1,000 psi and for the LBR, q u = 800 psi. This same reduction was then applied to establish the correction factor, ( 4 14 ) applying the correction factor to q u (4 15 ) and rearranging the equation to solve for q t (4 16 ) t herefore, the new equati on would be, ( 4 17 ) The new Florida geomaterials equation, with t he correction factor applied, Equation 4 17, was then compared to the original limestone and clay equations aga in
144 Fig ure 4 11. Comparison of q t /q u vs. q u curves with the new Florida geomaterials curve Evident in Fig ure 4 11 the correction factor appeared to work well. The q t /q u ratios for the lower end q u values were quite representative of the clay and IGM curve, and as q u increases the q t /q u ratios become more representative of the original limestone curve. Therefore, by only using the q u values obtained from Little River and applying the correction factor, a curve that represents both mat erial groupings was developed. Interestingly, the q t /q u range is approximately 0.29 to 0.09 for q u ranging from 10 to 10,000 psi, and is close to wh at the range was thought to be 0.3 to 0.1, based on historical FDOT core data. To simplify deriving q t us ing q u with the newly developed equation, q t was plotted vs. q u Figure 4 12 and another equation was derived to calculate q t directly. y = 0.659x 0.204 R = 1 y = 0.593x 0.229 R = 1 y = 0.436x 0.175 R = 1 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4 0.45 0.5 0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000 10,000 qt/qu Unconfined Compressive Strength Limestone Clay FL Geomaterials
145 Figure 4 1 2. Florida geomaterials q t equation From Figure 4 12 estimates of q t can now be derived directly using E quation 4 18 (psi) (4 18 ) Using the new Florida geomaterials equation, q t values were estimated using measured q u values and compared to measured q t values from the core data collected around the state. The basis of the comparison was dry unit weight and water content plotted vs. q t If the predicted q t values were accurate, they should plot as an underestimate of the measured q t values obtained from actual split tension testing. Figures 4 13 and 4 14 prov ide the c omparisons using 1,200 measured q t values and 700 predicted q t values derived using the Florida geomaterials equation with the available 700 compressive strength values from the 23 Florida si tes used to develop Figures 3 32 and 3 33 y = 0.436x 0.825 R = 1 0 100 200 300 400 500 600 700 800 900 1,000 0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000 10,000 Splitting Tensile Strength, qt (psi) Unconfined Compressive Strength, qu (psi)
146 Figure 4 1 3 S plit tension strength vs. moisture content with predicted q t values for 23 Florida sites Figure 4 1 4 Dry unit weight vs. split tension strength with predicted and measured q t values from 23 Florida sites Figures 4 13 and 4 14 show the predicted q t values fit the measured q t data very well for all of Flor ida based on moisture content and dry unit weight I n Figure 4 14 the 0 200 400 600 800 1,000 1,200 1,400 1,600 1,800 2,000 0 10 20 30 40 50 Splitting Tensile Strength, qt (psi) Moisture Content, w (%) Measured qt Predicted qt y = 40.953x 0.2058 R = 0.9799 y = 47.962x 0.1915 R = 0.9863 85 95 105 115 125 135 145 155 165 175 0 200 400 600 800 1,000 1,200 1,400 Dry Unit Weight (pcf) Splitting Tensile Strength, qt (psi) 95 100 pcf 100 105 pcf 105 110 pcf 110 115 pcf 115 120 pcf 120 125 pcf 125 130 pcf 130 135 pcf 135 140 pcf 140 145 pcf 145 150 pcf 150 155 pcf 155 160 pcf 160 165 pcf Measured Avg. Predicted qt
147 predicted q t values nearly all fall wi thin one standard deviation below the mean measured q t values. Therefore, the Florida geomaterials equation, Equation 4 18 should be fairly accurate when used with the developed drilling equation for measuring compressive strength. As the drilling equation was developed using drilling para meters recorded in a confined setting and referenced with the compressive strength of Gatorock, which provided an increase in tensile strength Since the Gatorock provides additional tensile strength compared to limestone with the same compressive strengt h, this inherently provides a reported compressive strength value more similar to u value for actual limestone Therefore, when skin friction is estimated using McVay et al. 13 during drilling, the predicted q t values in Figure 4 14 wi ll be increased and more reflective of the measured q t values; providing an accurate estimation of skin friction. Although it must be noted that the q t crit erion is for uniaxial tension, i.e., direct tension, not split tension. 4 Anogl u used the following to transform uniaxial tension to split tension; (4 19 ) where, = 0.9 23 However, the value of used by Anoglu was derived for concrete, and the value of for Fl orida limestone and IGM is unknown. Therefore, was assumed to be 1, as the current Florida geomaterials equation provided a good cons ervative estimate for tensile strength With the development of the Florida geomaterials e quation, real time skin friction estimates can now be achieved usi ng McVay et al. 13 ; providing a means to use all of the leading drilled shaft equations to estimate shaft capacity with the developed drilled shaft monitoring techniques. However, before the comparative analysis is provided, the topic
148 of rig malfunction and encountering voids while monitoring shaft installations needs to be addressed. Distinguishing Rig Malfunction from Encountered Voids While monitoring, it can be hard to distinguish betwe en voided areas encountered while drilling or whether the depth sensor picked up a rig malfunction produced from an im properly spooled cable winch. mast is misaligned during drilling. Fortunately, this can be checked b y looking at the recorded tabular data. Simply pause the drilling for a few minutes and upload the data onto a USB. The specific energy data can then be investigated on an external computer to make the distinction. The following provides two depth vs. s pecific energy plots from the monitored drillings at Kanapaha. Figure 4 15 Test shaft depth vs. specific energy -50 -49 -48 -47 -46 -45 -44 -43 -42 -41 -40 -39 0 2,000 4,000 6,000 8,000 10,000 12,000 14,000 Depth (ft) Specific Energy, e (psi) Zero specific energy
149 Figure 4 16 East shaft depth vs. specific energy From the two plots provided, it is difficult to determine if the low specific ene rgy recorded was from a small voided section of the excavation or if rig malfunction occurred. Both plots indicate that approximately zero specific energy was recorded in several sections of each shaft. However, only the test shaft actually experienced r ig malfunction. This occurred after the cable on the winch was spooled incorrectly during aligned. The drill rig inclinometer was not functioning and the mast inclin ation had to be checked periodically to ensure proper alignment. As a result, when the bit was lowered back into the hole after clean out, the cable would quickly release and the drill bit would drop uncontrollably. This caused the DIALOG to record verti cal movement that was not yet achieved. Because there was virtually zero crowd or torque applied when the bit dropped, the DIALOG recorded the data in these sections as zero specific energy. Using the DIALOG, if a certain depth has previously been reache d, e.g., spooling too -50 -48 -46 -44 -42 -40 -38 -36 -34 -32 -30 0 2,000 4,000 6,000 8,000 10,000 12,000 14,000 Depth (ft) Specific Energy, e (psi)) Small void
150 far, the DIALOG will not record over the data when advancing the bit to that depth again. Additionally, the quick release causes back pressure in the hydraulic lines which theoretically increases the threshold pres sure and changes th e develo ped K coefficients. Visualization is provided using the hydraulic conversion equation After watching replays of the drillings on the DIALOG, it was noticed the hydraulic pressures eventu ally stabilized to normal conditions. This typically occurred when the bit was removed from the hole and back spinning was completed. However, recorded rock strengths were lower while the back pressure was locked in as the K coefficients are constants Therefore, not only did the rig malfunction produce recorded readings at q u = 0 psi from the quick cable release; until the back pressure was returned to normal, the additional recorded rock strengths were lowered and produced an underestimate of the mate rial. This was a result of using an older drill rig to complete the drilling. This phenomenon appeared to have occu rred in the east shaft as well. However, the east shaft actually experienced voids and the following will explain the difference in readin gs and how to interpret the results to make the determination. In the test shaft around 46 feet, rig malfunc tion occurred as indicated in Figure 4 17 The malfunction was witnessed in the field as the falling bit produced an extr emely loud sound when the cable finally halted the falling bit causing the cable winch to jerk and the depth sensor to track penetration not yet achieved
151 Depth Pen. Rate Rotation Torque Crowd Sp. Energy qu (ft) u (in/min) N (rpm) T (in lbs) F (lbf) e (psi) (psi) 45.9 16.3 6. 4 519716 2525 1268.59 88.9 45.96 5.4 6.5 530608 2926 3946.19 256.7 46.03 27.5 6.6 532757 693 791.67 56.3 46.1 52.0 2.0 73385 408 17.52 1.3 46.16 109.1 0.0 5385 258 0.25 0.0 46.23 114.8 0.0 5507 256 0.25 0.0 46.29 116.5 0.0 5642 257 0.25 0.0 4 6.36 115.0 0.0 5710 259 0.25 0.0 46.42 149.0 0.0 5365 261 0.26 0.0 46.49 15.6 7.1 410576 1873 1153.92 81.2 46.56 8.6 7.1 407152 1809 2084.81 142.6 46.62 7.9 7.3 370936 1787 2101.55 143.7 Figure 4 17 Monitored readings indicating rig malfunction As seen in Figure 4 17 between 46.16 to 46.42 feet, there was zero recorded bit rotation which only occurs when the bit is being lowered into the hole. This also indicates there was zero torque applied to advance the bit. The recorded values for torqu e and crowd are a result of residual pressure s in the hydraulic lines, slightly above the estimated threshold pressure s Therefore, the only specific energy recorded was compensa tes for penetration without bit rotation. This results in virtually zero specific energy required to advance the bit which should not occur. The reading at 46.10 feet was a result of the DIALOG averaging the drilling parameters for every 2 cm of penetrat ion. Consequently, a portion of the average was recorded before the rig malfunction occurred and a portion of the average was recorded during the rig malfunction. This also leads to an underestimate for the reading which needs to be eliminated for the ov erall average of the drilled section. In the east shaft, at a depth of 32.19 feet, a void was encountered.
152 Depth Pen. Rate Rotation Torque Crowd Sp. Energy qu (ft) u (in/min) N (rpm) T (in lbs) F (lbf) e (psi) (psi) 31.99 3.3 5.4 168874 1281 1728.40 11 9.5 32.05 1.2 5.4 159893 1364 4547.36 291.4 32.12 2.0 5.4 159781 1543 2736.05 183.7 32.19 66.4 1.4 34563 271 4.81 0.4 32.25 2.2 4.5 122209 963 1533.51 106.6 32.32 2.4 5.8 170090 2351 2592.85 174.8 32.38 4.7 5.8 195211 1315 1496.66 104.2 Figur e 4 18 Monitored readings indicating a void was encountered Figure 4 18 shows that although the specific energy reading was extremely low, every drilling parameter produced a non zero reading. As a result, large differences in the drilling parameters can be used to distinguish between void detection and rig malfunction. The first indication is the differences in penetration rate. The rig malfunction section consistently produced a rate of penetration nearly double that of the voided section. The rig malfunction penetration rates were also increasing as penetration occurred, indicating acceleration from freefall. The second indication is the voided section recorded rotation, confirming drilling was taking place and torque was being applied. The reco rded torque value from the voided section is also on a higher order of magnitude than the rig malfunction section. Additionally, the torque values in the rig malfunction section are two orders of magnitude lower than the torque values recorded before and after the malfunction occurred. Finally, the specific energy for the voided section produced a result on a higher order of magnitude than the rig malfunction section, where the specific energy value, 0.25 psi, was repeatedly recorded. It is highly unlike ly the same recorded value would occur for specific energy at this degree of precision. This was verified through visual inspection by looking at the variability of specific energy readings before and after the rig malfunction and the voided
153 section. In the test shaft at depths 46.10 and 46.42 feet, a 1.10 psi difference in q u produced nearly a 15 psi difference in specific energy. This led to the conclusion that rig malfunction occurred in the test shaft and voids were encountered in the east shaft. It is recommended that zero specific energy data points be removed when analyzing the data, as they can significantly reduce the estimated shaft capacity. However, if the zero specific energy points were overlooked and included in the shaft capacity estimat ion; the result would only provide a conservative estimate and would be acceptable. Comparative Skin Friction Analysis The developed drilled shaft monitoring method provides a means to measure both unconfined compressive strength and skin friction in real time during the shaft installation process, i.e., during drilling. As well, the developed method allows the foundation engineer to choose from any of the lead ing skin friction equations Table 4 1 that are generally used in drilled shaft design in the st ate of Florida. How ever, a recommended method needed to be determined to complete the research. The following prov ides the list of equations and their respective method number used in the comparative analysis. 1. McVay et al. 13 2. 14 in combin atio n with Horvath and Kenney 15 3. Williams et al. 16 4. Reynolds and Kaderabek 17 5. Gupton and Logan 18 6. Carter and Kulhawy 19 7. Ramos et al. 20 8. Rowe and Armitage clean socket equation 21 Method 2 combines the equations devel 14 and Horvath and Ke nney 15 This was done because each equation is only recommended for u
154 and Horvath and Kenney was used when q u > 20 tsf. Th e clean socket equation from Rowe and Armitage 21 was used because the rough socket equation produced much large r overestimates. The following provides the comparative analysis using each of th e methods at all three monitored shaft locations; where each of th e monitored sections were mobilized and provided direct comparison. Location Section Load Test Method 1 Method 2 Method 3 Method 4 Little River SG8 to SG7 9.9 11.2 9.0 14.7 40.8 SG7 to SG 6 21.1 19.7 8.2 14.0 33.6 SG6 to O cell 20.6 22.1 9.7 15.6 37.2 O cell to SG5 21.4 19.5 8.6 14.3 33.0 SG5 to SG4 13.6 14.0 7.3 13.0 22.4 Kanapaha TS SG1 to SG2 8.0 8.6 5.5 11.0 13.1 TS SG2 to SG3 8.2 8.2 5.3 10.7 12.3 TS SG4 to Base 4.9 4.9 3.3 8.7 7.0 ES SG1 to SG2 2.4 2.4 1.6 5.5 3.3 Overland Segment 2 2.1 1.9 1.3 5.8 2.5 Average Percent Error N/A 0.6% 40.0% 41.6% 78.9% Figure 4 19 Skin friction comparative analysis using m ethods 1 through 4. Location Section Load Test Method 5 Metho d 6 Method 7 Method 8 Little River SG8 to SG7 9.9 27.2 8.8 18.6 20.2 SG7 to SG6 21.1 22.4 8.1 16.3 18.6 SG6 to O cell 20.6 24.8 9.2 15.3 21.2 O cell to SG5 21.4 22.0 8.3 15.3 19.0 SG5 to SG4 13.6 15.0 7.2 12.7 16.5 Kanapaha TS SG1 to SG2 8.0 8.7 5.7 8.6 13.0 TS SG2 to SG3 8.2 8.2 5.5 8.1 12.6 TS SG4 to Base 4.9 4.7 4.1 8.9 9.5 ES SG1 to SG2 2.4 2.2 2.4 4.7 5.6 Overland Segment 2 2.1 1.7 2.4 4.2 5.6 Average Percent Error N/A 19.3% 29.6% 29.6% 62.1% Figure 4 20. Skin friction comparativ e analysis using m ethods 5 through 8. As seen, McVay et al. 13 provided the best result. The method was in excellent agreement with the load test results at all three monitored locati ons. Therefore, McVay
155 et al. 13 was chosen as the recommended method to u se with the developed drilled shaft monitoring method. For convenience, the Florida geomaterials equation was incorporated into the equation developed by McVay et al. 13 so skin friction could be estimated directly from q u The following pr ovides the equa tion development: Substituting the Florida geomaterials equation, (4 18) into the skin friction equation developed by McVay et al., ( 4 19 ) f s can be solved directly using only q u (4 20 ) where, (4 21 ) With the final drilling equation developed to measure skin friction during shaft installations, additional analysis was performed to provide a better understanding of the monitoring accuracy and the variability of the results. The following presents the monitoring results vs. the load test results for skin friction at each monitored location using the new equation. Again, the presented results are in portions of the shaf ts, at each location, wh ere the skin friction was mobilized; thereby, providing direct comparison of monitoring to conventional methods for estimating shaft capacity. As well, a different load test method was used at each of the sites; providing compariso n with the three most conventional load test methods.
156 Location / (Reference) Section Test Type Thickness (ft) Measured (ksf) Predicted (ksf) % Difference Little River (LR 1) SG8 to SG7 Osterberg 10.0 9.90 11.15 12.63% Little River (LR 2) SG7 to SG 6 Osterberg 5.0 21.10 19.67 6.78% Little River (LR 3) SG6 to O cell Osterberg 5.5 20.60 22.09 7.23% Little River (LR 4) O cell to SG5 Osterberg 3.5 21.40 19.46 9.07% Little River (LR 5) SG5 to SG4 Osterberg 5.0 13.60 13.95 2.57% Kanapaha (K 1) SG1 to SG2 Static 3.0 8.02 8.62 7.48% Kanapaha (K 2) SG2 to SG3 Static 3.0 8.22 8.18 0.49% Kanapaha (K 3) SG4 to Base Static 2.0 4.86 4.88 0.41% Kanapaha (K 4) East Shaft Static 5.0 2.36 2.36 0.00% Overland (O 1) Segment 2 Statnamic 5.0 2.06 1.90 7.77% Av erage All All 4.7 11.21 11.23 0.62% Figure 4 2 1 Skin friction comparative analysis summary using McVay et al 13 Figure 4 2 2 Unit side shear comparison from all monitoring sites As seen in Figures 4 21 and 4 22 the developed dri lled shaft monitor ing method was not only successful, but highly accurate. This was confirmed by conducting a bias analysis as measured/predicted, i.e., load test/monitoring. 0 5 10 15 20 25 LR-1 LR-2 LR-3 LR-4 LR-5 K-1 K-2 K-3 K-4 O-1 Unit Side Shear (ksf) Monitoring Location Monitoring Load Test
157 Table 4 3 Unit side shear bias analysis summary of statistics. Statistics Bias Average 1.00 Me dian 1.00 Std. Dev. 0.07 CV 0.0678 Count 10 Figure 4 2 3 Unit side shear bias analysis (Load Test / Monitoring) The results of the analysis show the mean and median bias is 1.00, and the CV is le ss than 0.1, which indicates that monitoring has lim ited variability. From the comparative analysis summary for the entire project, it is evident that drilled shaft construction monitoring is a viable solution to removing spa tial uncertainties and providing accurate measurements of compressive strength and skin friction in real time 0.0 0.2 0.4 0.6 0.8 1.0 1.2 LR-1 LR-2 LR-3 LR-4 LR-5 K-1 K-2 K-3 K-4 O-1 Bias (Measured/Predicted) Monitoring Location
158 CHAPTER 5 CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE WORK Conclusions The following conclusions were drawn from this study: The developed monitor ing method is a viable option for estimating rock strength and drilled shaft capac ity in real time. The results obtained from monitoring the drilled shaft installations were in near perfect agreement with conventional methods. This included compressive strength comparisons with core data obtained from traditional rock coring and skin f riction estimations compared to three of the most widely used load testing methods, i.e., Osterberg, Statnamic, and Top down static load testing. It was found that equipment needed to monitor drilled shaft installations is often standard on new drill rigs and is commercially available for rig types without monitoring equipment. This provides an easy transition to incorporate the developed method into standard drilled shaft practice. The developed method provides a means to quantify the qua lity and length of rock sockets in real time during the drilling process. This ensures the as built foundation meets or exceeds the engineer ing design during construction. The develop method provides quality assurance and quality control to the drilling contractor and fo undation engineer. This research took the first steps towards eliminating spatial variability concerns for structures supported by drilled shafts. Monitored drilling practices will ultimately lead to increased resistance factors used in design. This will provide more efficient and cost effective construction practices by reducing the time of compl etion and cost per shaft based on reduced uncertainty. N ew relationships were developed between material and strength properties of Florida geomaterials. It was found that there is an interdependence between the compressive and tensile strength of geomaterials; which is in agreement with Mcvay et al. 13 4 and Anoglu et al. 23 Correlations were developed between compressive and tensile strength with dry unit weight and moisture content. This gives rise to the concept of index testing, which provides a better understanding of geomaterial strength properties when core data is limited for sites with poor recoveries.
159 Recommendations The following re commendations are based on the findings from this study: Conduct more drilled shaft monitoring with load tests to further validate the developed method. The concept of monitored drilling should be used in more geotechnical engineering applications. The met hod should be developed for ACIP piles where visual inspection of the dr illed cuttings does not occur. The developed method should be used as a site investigation tool on SPT rigs to provide additional data used for geotechnical engineering design purposes This will provide continuous measurements of rock strength and a means to quantify the quality of the coring procedure. Incorporating the developed method into the SPT rig will provide continuous measurements, similar to CPT, with the ability to penetr ate through layers of rock which terminates a CPT. The method should be used while advancing the hole with a roller bit and during coring. This will improve the theoretical eq uations developed in this research and provide more accurate estimations of q t and f s The concept of index testing should be explored. This will provide an excellent reference of strength for sites with poor recoveries. Summary of Research This research focused on developing and evaluating the relationship between rock drilling parameters: crowd, torque, penetration rate, rotational speed and drill bit diameter with rock strength, e.g. unconfined compression, q u This required both a laboratory and fi eld investigation. In the laboratory, synthetic limestone, Gatorock, was developed at four di fferent design strengths, 10, 20, 40, limestone formations. The Gatorock was cast into large blocks, were subsequently drilled using various combinations of the five drilling parameters. The results were wirelessly transmitted in real time to an external computer and
160 recorded for analysis. During the analysis multiple developed drilling equations were specific energy equation for rotary non percussive drilling in rock. The developed relationship provided the strongest correlation between the five drilling parameters and compressive strength. From this, a new equation was developed that could be used to measure rock strength in real time during field drilling. In total, 81 drilling data points were used to develop the equation. In the field, Jean Lutz monitoring equipment was acquired and used to measure the same five drilling parameters on the drill rig. The drilling parameters were recorded and transmitted wirelessly to an external computer in real time away from the drilling. The Jean Lu tz equipment included pressure transducers used to tap into the hydraulic lines providing torque and crowd to the drill bit, a proximity sensor to monitor rotational speed at the rotary table, a rotary encoder mounted on the rim of the main cable winch to monitor penetration rate, a junction box to receive the signals from each sensor, and a data acquisition module to record, display, and transmit the data wirelessly to an external comput er in real time via Bluetooth. During the course of the research proje ct, field monitoring took place at three separate location s where load testing occurred. The locations were at the Little River bridge site in Quincy, F lorida, the Overland bridge site in Jack sonville, FL, and the ida Monito ring in these locations provided direct comparison of laboratory tested core samples with monitoring estimated rock strengths. It was found that monitoring provides highly accurate measures of in situ rock strength, eliminate s spatial variabil ity concerns and produces accurate rock
161 strength distributions with a degree of precision that has never been achieved before, and could not be achi eved using any current method. Field monitoring also provided direct comparison of the estimated shaft cap acity obtained during monitored drilli ng to the actual capacity of installed shafts measured using conventional load testing methods A dditionally, each location use d a different type of load test : O c ell testing at Little River, Statn amic testing at Over land, and a traditional top down static load te st at Kanapaha. This provide d direct comparative data from three of the most conventional load testing methods used throughout the state. The f ield investigation also provided three variations in the followin g categories: location, drill rigs used to install the shafts, shaft diameters, drilling crews, drill bits, drill bit tooth configurations and limestone formations encountered These variable drilling conditions provide d great insight on how well the labo ratory drilling equations perform when drilling conditions, rig configurations and Florida limestone formations change In all locations, the m onitoring results were in near perfect agreement with the load test results. As well the monitoring provided l ittle variability in the results, as the bias analysis conducted produced a mean and m edian bias of 1.00, with a CV less than 0.07. In addition to developing the drilling equation for real time measurement of rock strength, methods to estimate splitting te nsile strength and skin friction in real time were also developed. Initially it was thought that these strength measurements would be estimated using site specific limestone q t /q u ratios post monitoring. However, using 4 and core d at a from around the state a new equation was developed that accurately estimates splitting tensile strength in real time based on compressive strength, the Florida geomaterials equation. Using this equation allows the
162 recommended SFH skin friction equation to be used in real time as well. New relationships between compressive and splitting tensile strength with void ratio, moisture content and dry unit weight were also developed. This gives rise to concept that index testing of these material properties could provide better understanding of limestone rock strengths in locations where rock cores are difficult to obtain for core testing. As is evident in the conclusions, the drilled shaft monitoring methods are more than viable. With this work, the researc h took the first steps towards eliminating spatial variability concerns for structures supported by drilled shafts. In addition, this provides a means to quantify the quality and length of rock sockets during drilling so the as built foundation meets/exce eds the engineering design, providing quality assurance to the drilling contractor and foundation engineer. This would be extremely useful for projects that do not implement load testing to confirm the design of their production shafts. For all of these reasons it is recommended that more drilled shafts with planned load tests be monitored during construction in order to further verify the strength, as well as estimated side shear. Future efforts should focus on locations where different Florida limeston e formations are encountered. Ideal locations would be in the Tampa area where dolomite is encountered, in Miami where oolite is encountered, along the east coast where the young Anastasia formation is found, in the panhandle where higher strength limesto ne is present and in larger metropolitan areas such as Jacksonville where construction is constantly growing and where both limestone and marl formations exists.
163 Since the developed monitoring techniques were capable of determining not only compressive s trength but splitting tensile strength and skin friction in real time, the same drilling techniques could be applied in more geotechnical engineering applications such as auger cast piles, ACIP, or used as a site investigation tool, SPT. In the case of au ger cast piles, interest for the use of these foundations under bridge structures is high. Generally, ACIP piles are installed quickly and they too develop large unit skin frictions in limestone. However, since the continuous flight auger does not go in and out the hole, the excavated material type is unknown; therefore, the developed monitoring techniques could be very useful. In the case of site investigation, drill monitoring from an SPT rig could provide the benefit of continuous data collection simi lar to CPT with the ability to penetrate through layers of rock that terminates a CPT test. Continuous data collection would provide great assistance for sites with poor recoveries and highly variable soil and rock. Additionally, at any time during monit oring from the SPT rig, SPT testing could be implemented or core samples could be extracted for lab testing and compared to drilling strength measurements. Finally, covered in this report was a newly developed correlation between rock strength properties, q u and q t with materials properties of rock such as void ratio, porosity, moisture content and dry unit weight. It is believed that developing correlation between strength and material properties of rock will lead to a better understanding of the materi als being influenced by engineering design. It also gives rise to the concept of index testing where strength estimates could be made based on the material properties of the rock. Collection and testing of limestone cores from around the state could prov ide the basis of the investigation. From this, a stronger correlation between
164 Florida geomaterials and Gatorock could be developed experimentally and compared to th e current theoretical relationship developed for q u and q t This would further improve the real time estimates of splitting tensile strength and skin friction during drill shaft monitoring as well as provide new insight for use in conventional design.
165 LI ST OF REFERENCES 1. Karasawa H, Ohno T Kosu gi M, Rowley JJC. Methods to Estimate the Rock Strength and Tooth Wear While Drilling With Roller Bits, Part 1: Milled Tooth Bits. ASME Journal of Energy Resources Technology 2002;124: 125 132. 2. Karasawa H, Ohn o T Kosugi M, Rowley JJC. Methods to Estimate the Rock Strength and Tooth Wear While Drilling With R oller Bits, Part 2: Insert Bits. ASME Journal of Energy Resources Technology 2002; 124 : 133 140. 3. Teale R. The Concept of Spe cific Energy in Rock Drillin g. Int J Rock Mech Mining Sci 1965; 2 : 57 73. 4. Johnston I. Strength of Intact Ge omechanical Materials. ASCE Journal o f Geotechnical Engineering 1985;111: 730 749. 5. Wolcott D S, Bordelon DR. Lithology Determination u sin g Downhole Bit Mechanics Data. SPE 26492, P resented at the 68th ATCE of th e SPE Houston; TX: 1993 769 778 6. Hoberock LL Bratcher, GJ 1996. A New Approach for Determining In Sit u Rock Strength While Drilling. ASME Journal of Energy Resources and Technology 1996;1 18 : 249 255. 7. Bull ock PJ Insitu Rock Modulus Apparatus Florida Department of Transportation. Re search Report BC354 13 Gainesville, FL: University of Florida ; 2004. 8. McVay, MC Niraula, L Development of P Y Curves for Large Diameter Piles/Drilled Shafts in Limestone f or FBPIER Florida Department of Transportation, Research Report BC354 59 G ainesville, FL: University of Florida ; 2004. 9. Sheppard MD Bloomquis t D, Slagle PM, Renna R Rate of Erosion Properties of Rock and Clay Florida Department of Transpor tation Re search Report BD545 3 Gainesville, FL: University of Florida ; 2006. 10. American Concrete Institute Co ntrolled Low Stre ngth Materials ACI Committee 229, American Concrete I nstitute ACI 229R 99 Farmington Hills; MI : 1999. 11. American Society for Tes ting and Materials Standard Test Method for Unconfined Compressive Strength of Intact Rock Core Specimens ASTM International ASTM D2938 95 West Conshohocken; PA: 2002. 12. Scott TM, Campbell KM, Rupert FR, Arthur JD, Green RC, Means GH M issimer T M Lloyd JM, Yon JW Duncan JG Geolo gic Map of the State of Florida 2001. http://publicfiles.dep.state.fl.us/FGS/FGS_Publications/MS/ms146_geology_of_fl.pdf
166 13. McVay M Townsen d F, Williams R Design of Socke ted Drilled Shafts in Limestone. ASCE Jour nal of Geotechnical Engineering 1992; 118 :1 0 : 1626 1637. 14. R eese LC, O'Neill MW Drilled Shafts: Construction P rocedures and Design Methods, Design Manual. US Department of Transportation, Federal H ighway Administration McLean; VA: 1987. 15. Horvath RG, Kenney TC Shaft Resistance of Rock Socketed Drilled Piers. Symposium on Deep Foundations, ASCE N ational Convention Atlanta; GA: 1979. 182 214. 16. Williams AF John ston IW, Donald IB The Design of S ocketed Piles in Weak Rock. Proc Int Conf on Struct Foundations in Roc k ( Netherlands ): 1980. 327 347. 17. Reynold s RT, Kaderabek TJ Miami Limestone Foundation D esign and C onstruction. ASCE New York; NY : 1980. 18. Gupton C, Logan T. Design Guidelines for Drilled S hafts in Weak Rocks of South Florida. Proceedings of the South Florida Annual ASCE Meeting ASCE: 1984. 19. Carter JP, Kulhawy FH Analysis and Design of F oundations Socketed into Rock. Research Repo rt 1493 4, Geotechnical Engineering Group Ithaca; NY : Cornell University; 1987. 20. Ramos HR, Antorena JA, McDaniel GT Correlations b etween the Standard Penetration Testing (SPT) and the Measured Shear Strength of Florida Natural Rock. Proceedings from FHWA International Conference on Design and C onstruction of Deep Foundations Orlando; FL : 1994. 699 711. 21. Ro we RK, Armitage HH. 1987. A Design Method for Drilled P iers in Soft R ock. Can Geotech J 1987; 24 (1) : 126 142. 22. Florida Department of Transpo rtation ( FDOT ) Soils and Foundation Handbook Florida Department of Transportation, State Materials Office Gainesville; FL : 2015. 23. Anoglu N Girgin Z Anoglu E. 2006. Evaluation of Ratio between Splitting Tensile Strength and Compressive Strength fo r Concrete up to 120 MPa and its Ap plication in Strength Criterion. ACI Materials Journal 2006;103 M03
167 BIOGRAPHICAL SKETCH As a chil d when asked by his grandmother what he wanted to do when he grew Therefore, it seems fitting that Michael pursued a career in engineering. The focus of civil engineering came from growing up building things around the house with his grandfather who was a general contractor for the majority of his life. Michael also be the one who tells in Mi chael and sparked his pursuit of a career in civil engineering, a lifelong dream of his grandfather. To Michael, family and friends are everything and to be able to fulfill his and support, Michael was destined for success. Through his father he found a love for sports and fitness, and to this day Michael actively strives to maintain peak physical fitness and firmly believes a sound mind starts with a sound body. Through his mother Michael has found strength, wisdom, and the mentality to never give up. He largely attributes his success to the lessons learned from his mother Growing up with two younger brothers and a sister Michael has always tried to lead by example. Although, the task has not required much effort as his siblings have all turned out to be truly amazing individuals and leaders in the ir own right. The suc cess of his family is somet hing Michael is very proud of. Michael is a second generation Florida Gator, and is very proud to be a part of the Gator Nation. He has earned a b achelor s degree, m aster s degree, and d octorate
168 at the University of Florida from the College of Civil Engineering. During his undergraduate studies Michael was actively engaged in research. This led to him Scholar Award for his graduating class. Upon ente ring graduate school Michael received a CCE Graduate Scholarship. Through his continued research, hard work, and academic success; Michael eventually earned an Assistantship and later a Graduate Student Fellowship. He has mentored younger undergraduate a nd graduate students during his time at UF and is very p roud of their success as well. Michael firmly believes that his research will spark a revolution in geotechnical engineering and lead to improvements in site investigation and drilled shaft design an d construction. As well, he believes his research will help bring the field of geotechnical engineering into a new age that grows with technological advancements and improve s standard practice. Michael is committed to ensuring this becomes a reality.