Click for next page ( 74


The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 73
74 hydraulically through lines extending from the top unit (sometimes called a rodless drill), which is fixed to the top of the casing. Alternatively, a drill rod may be used to transmit torque from the top unit to the bit. The bit has a central hol- low orifice connected to a flexible line extending back up to the top unit. During drilling, a vacuum pump or air lift is used to draw the drilling fluid with the cuttings upward to a clean- ing plant, from where it is circulated back into the hole. The unit shown in Figure 56 was used to drill 3.35-m-diameter rock sockets in Franciscan Formation sandstone and ser- pentinite. Some manufacturers are now producing reverse circulation units that can be installed on a conventional rotary hydraulic drilling rig to provide similar capability, at a smaller diameter. It is likely that these units will become more common in North America for rock-socket drilling FIGURE 55 Downhole hammer tool for drilling in hard rock. (D. Poland, Anderson Drilling, personal communication, Aug. 2, 2005). Reverse circulation drilling can also be car- A technique used for drilling large rock sockets at the ried out with any type of rotary drill rig equipped with a hol- RichmondSan Rafael Bridge (Byles 2004) is reverse circu- low Kelly bar (drill stem) that allows circulation of the lation drilling with a "pile top" rig. The unit consists of two drilling fluid from the cutting surface up through the bar. main components. A top unit (Figure 56) is fixed to the top of a steel casing. The "bottom hole assembly" (Figure 57) is a drill bit lowered to the bottom of the hole through a casing, FIELD LOAD TESTING submerged in water or other drilling fluid. The bit is operated The most direct method to determine the performance of full- scale rock-socketed drilled shafts is through field load test- ing. Clearly there have been advances in engineers' ability to predict rock-socket behavior. However, there will always be sources of uncertainty in the applicability of analysis meth- ods, in the rock mass properties used in the analysis, and with respect to the unknown effects of construction. Load testing provides direct measurement of load displacement response for the particular conditions of the test foundation, and can also provide data against which analytical models can be evaluated and calibrated. Objectives Field load testing may be conducted with different objectives and this should determine the scope of testing, type of tests, and instrumentation. A partial listing of valid reasons for transportation agencies to undertake load testing of rock- socketed shafts includes: Confirm design assumptions, Evaluate rock resistance properties, Evaluate construction methods, Reduce foundation costs, and Research aimed at evaluating or improving design methods. More than one of these objectives can sometimes be achieved. For example, load tests conducted primarily for FIGURE 56 Toredo T40-4 pile top unit being placed over confirmation of design assumptions (proof test) for a partic- casing for reverse circulation drilling, RichmondSan Rafael ular project can be useful to researchers by contributing ad- Bridge (California) (Byles 2004). ditional data for evaluating empirical correlations proposed

OCR for page 73
75 FIGURE 57 Shrouded bottom hole assembly lifted for placement through the top unit (Byles 2004). for design. Load tests carried to ultimate capacity of the shaft as the basis for design of productions shafts. Items such as are especially valuable not only to the agency conducting the construction method (casing, slurry, dry), type of drilling test or for the specific bridge project, but to the entire deep fluid, cleanout techniques, and others may have influenced foundation engineering community. the behavior of the test shaft. If possible, the construction methods anticipated for production shafts should be used to The costs of conducting field load tests should be offset construct test shafts. by its benefits. The most obvious costs include the dollar amount of contracts for conducting testing. Other costs that are not always as obvious include construction delays, delays Axial Load Testing in design schedule, and DOT person hours involved in the testing. Direct cost benefits may be possible if the testing Conventional Axial Load Testing leads to more economical designs. This requires testing prior to or during the design phase. Numerous case histories in the Until the early 1990s the most common procedure for con- literature show that load testing almost always leads to ducting a static axial compression load test on a deep founda- savings. Lower factors of safety and higher resistance factors tion followed the ASTM Standard Method D1143, referred to are allowed by AASHTO for deep foundation design when herein as a conventional axial load test. Several load applica- a load test has been conducted. tion methods are possible, but the most common involves using either (1) a hydraulic jack acting against a reaction Other benefits may not be so obvious or may occur over beam that is anchored against uplift by piles or (2) a loading time. Construction of the test shaft provides the DOT and platform over the pile top on which dead load is placed. Six all subsequent bidders with valuable information on con- states indicated that they have conducted conventional axial structability that can result in more competitive bids. Refine- load tests on rock-socketed shafts. Conduct and interpretation ment in design methods resulting from information gained by of axial compression and uplift load tests specifically for drilled load testing has economic benefits on future projects. shafts is discussed in detail by Hirany and Kulhawy (1988). Load test results provide the most benefit when they are Axial load tests may be conducted for the purpose of con- accompanied by high-quality subsurface characterization. firming the design load for a specific project, in which case Knowledge of site stratigraphy, soil and rock mass properties, it is typical to load the shaft to twice the anticipated design site variability, and groundwater conditions are essential for load to prove the shaft can support the load with an accept- correct interpretation of load test results. The ability to apply able settlement (a proof load test). This type of test is nor- load test results to other locations is enhanced when subsur- mally conducted under the construction contract and does not face conditions can be compared on the basis of reliable data. yield a measured ultimate capacity, unless the shaft fails, in which case the design must be adjusted. Proof tested shafts Construction factors and their potential effects on shaft normally are not instrumented except to measure load and behavior should be considered when using load test results displacement at the head of the shaft. When the objective of

OCR for page 73
76 testing is to gain information on behavior of the shaft in terms The following case illustrates effective use of conven- of load transfer, the shaft should be instrumented to deter- tional axial load test on rock sockets. Zhan and Yin (2000) mine the distribution of axial load as a function of depth and describe axial load tests on two shafts for the purpose of con- as a function of axial deformation. firming design allowable side and base resistance values in moderately weathered volcanic rock for a Hong Kong tran- Common types of instrumentation for measuring axial sit project. The proposed design end bearing stress (7.5 MPa) load and deformation at specific points along the length of exceeded the value allowed by the Hong Kong Building the shaft include sister bars and telltales. A sister bar is a sec- Code (5 MPa). One of the objectives of load testing was, tion of reinforcing steel, typically 1.2 m in length, with a therefore, to demonstrate that a higher base resistance could strain gage attached in the center. Either vibrating wire or be used. The project involved 1,000 drilled shafts; therefore, electrical resistance-type gages can be used. The sister bar is proving the higher proposed values offered considerable tied to steel of the reinforcing cage and its lead wires are potential cost savings. routed to the surface, where they are monitored by a com- puter-controlled data acquisition unit. The gage signals are Figure 58 shows the load test arrangement, consisting of converted to strain, which is assumed to be equal to the a loading platform for placement of dead load. Figure 59 strain in the concrete and can be used to estimate load using shows details of one of the instrumented shafts. Strain gages the appropriate elastic modulus and section properties of the were provided at 17 different levels, including 4 levels of shaft. A telltale is a metal rod installed within a hollow tube gages in the rock socket. Two telltales were installed, one at embedded in the shaft. The bottom end of the rod is fixed at the base of the socket and one at the top of the socket. Shafts a predetermined depth in the shaft and is the only point on the were excavated through overburden soils using temporary rod in physical contact with the shaft. By measuring vertical casing to the top of rock. When weathered rock was encoun- deformation of the upper end of the telltale during loading, tered, a 1.35-m-diameter reverse circulation drill (RCD) was deformation of the shaft is determined for the depth at which used to advance to the bearing rock, followed by a 1.05-m- the telltale is fixed. By measuring the relative displacement diameter RCD to form the rock socket. For the shaft shown between two successive rods and distance between their in Figure 59, the socket was 2 m in length. A permanent, bottom ends, the average strain in the shaft between the two bitumen-coated casing (to reduce side resistance in the over- telltales can be determined. Further information on these and burden materials) was placed to the top of the socket. The other types of instrumentation is given by Hirany and bottom was cleaned by airlift and concrete placed by tremie Kulhawy (1988) and O'Neill and Reese (1999). (wet pour). FIGURE 58 Axial load testing setup (Zhan and Yin 2000).

OCR for page 73
77 FIGURE 60 Unit side and base resistance versus axial load (Zhan and Yin 2000). Testing Systems" was undertaken to evaluate these and other methods for deep foundations and to recommend interim procedures for their use and interpretation. A draft final report by Paikowsky et al. (2004b) describes these methods in detail. The role of each of these tests for rock-socketed shafts is described here. FIGURE 59 Details of instrumented rock-socketed shaft (Zhan and Yin 2000). Osterberg Load Cell Figure 60 shows the results in terms of mobilized unit side The O-cell is a hydraulically operated jacking device that can and base resistances versus load applied at the head of the be embedded in a drilled shaft by attachment to the reinforc- shaft. Unit side resistance reached a value of 2.63 MPa, well ing cage (Figure 61). After concrete placement and curing, a exceeding the proposed design allowable value of 0.75 MPa. load test is conducted by expanding the cell against the por- Zhan and Yin noted that this value agrees well with Eq. 30 in tions of the shaft above and below it (Osterberg 1995). The chapter three. Load transfer to the base was mobilized im- load is applied through hydraulic piston-type jacks acting mediately upon loading, indicating excellent base conditions, against the top and bottom cylindrical plates of the cell. The and reached a value exceeding 10 MPa. In the other shaft (not shown) a unit base resistance of 20.8 MPa was reached with no sign of approaching failure. The case presented by Zhan and Yin demonstrates how a set of well-instrumented conventional axial load tests can be used to (1) achieve cost savings on a project with a large number of shafts, (2) confirm design allowable values of socket resistance, (3) demonstrate suitability of the construc- tion method, and (4) provide data against which design meth- ods can be evaluated. Conventional axial load testing has largely been replaced by methods that are easier to set up and conduct, require less equipment and space, are safer, less time consuming, and usually less expensive, especially in rock. These methods include the O-cell, Statnamic (STN), and dynamic impact FIGURE 61 O-cell at bottom of reinforcing cage ready for load tests. NCHRP Project 21-08, entitled "Innovative Load placement in a drilled shaft. (O'Neill and Reese 1999).

OCR for page 73
78 maximum test load is limited to the ultimate capacity of hydraulic lines either the section of shaft below the cell, the section above placement channel the cell, or the capacity of the cell. Pressure transducers are used to monitor hydraulic jack dial gages pressures and converted to load. Linear vibrating wire dis- placement transducers (LVWDTs) between the two plates measure total expansion of the cell and telltales are installed to measure vertical movements at the top and bottom of the test sections. The downward movement of the bottom plate Test shaft is obtained by subtracting the upward movement of the top test section from the total extension of the O-cell as deter- SOIL mined by the LVWDTs. Telltale deformations are monitored with digital gages mounted on a reference beam. All of the instrumentation is electronic and readings are collected by a data acquisition unit. The O-cell testing method provides some important ad- vantages. There is no structural loading system at the ground surface. Load can be applied at or very close to the base of a socket for measurement of base resistance. In conventional top load testing, most or all of the side resistance must be mo- ROCK bilized before there is significant load transfer to the base. Some of the cited disadvantages are that the O-cell is sacri- O-cell ficial and requires prior installation, so it is not useful for test- ing existing foundations. Using a single O-cell, it is possible to mobilize the ultimate capacity of one portion of the shaft only, so that other sections of the shaft are not loaded to their ultimate capacity. FIGURE 62 Shaft and O-cell test setup (adapted from Gunnink and Kiehne 2002). According to DiMillio (1998), the majority of load tests on drilled shafts are now being done with the O-cell. This is supported by results of this study, in which 17 of 32 states re- tain or increase load without continuous upward deflection sponding to the survey reported using the O-cell for axial of the top of the shaft, whereas the average base displacement load testing of rock-socketed shafts. Of these, 13 stated that did not change. From these tests, it is not possible to deter- ultimate side resistance was determined and 7 reported that mine ultimate base resistance values. The base load dis- the ultimate base resistance was determined. Five states in- placement curves show an interesting difference. For Shaft dicated the test was used for proof load testing, in which de- No. 1, the downward base movement is small (around 1 mm) sign values of shaft resistance were verified. These responses up to the maximum test load, suggesting a very stiff base and show that the O-cell has become a widely used method for good contact between the concrete and underlying rock. axial load testing of rock sockets. However, the curve for Shaft No. 3 shows downward move- ment approaching 10 mm upon application of the load, fol- A set of O-cell tests reported by Gunnink and Kiehne lowed by a flattening of the curve. This behavior suggests the (2002) serves to illustrate the type of information that is ob- presence of a compressible layer between the concrete and tained from a typical test in which a single O-cell is installed underlying rock, possibly the result of inadequate cleanout of at the base of a rock socket. Figure 62 shows the test setup the hole before pouring concrete. Both shafts were poured for three test shafts socketed into Burlington limestone. As under dry conditions and both were cleaned using the same shown, the shafts extended through soil before being sock- method, reported as "rapidly spinning the auger bit after the eted into limestone. All shafts were 0.46 m in diameter and addition of water and then lifting out the rock cuttings." socket lengths ranged from 3.45 m to 3.85 m. Depth of soil was approximately 4 m. Figure 63 shows test results for two Gunnink and Kiehne (2002) reported that it is common of the shafts (Shaft Nos. 1 and 3), respectively. Each graph practice to design drilled shafts founded in sound Burlington shows two curves, one of the O-cell load versus average mea- limestone for base resistance only, using a presumptive al- sured uplift of the upper portion of the shaft, and the other of lowable unit base resistance of 1.9 MPa. Side resistance is of- the O-cell load versus downward displacement of the base of ten neglected for design. Even the lowest observed base re- the cell. Both figures are typical of failure of the shaft in up- sistance measured by the O-cell tests yielded an allowable lift. At the maximum test load, it was not possible to main- unit base resistance of 5 MPa, assuming a factor of safety

OCR for page 73
79 30 30 25 25 upward displacement Average displacement (mm) of shaft above O-cell upward displ of shaft above O-cell 20 Average Displacement (mm) 20 15 15 10 10 5 5 0 0 -5 -5 downward displacement of O-cell base plate -10 -10 downward displ of O-cell base plate -15 -15 -20 -20 0 500 1000 1500 2000 2500 3000 3500 4000 0 500 1000 1500 2000 2500 3000 3500 4000 Load (kN) Load (kN) FIGURE 63 Results of single O-cell load tests: (left) Shaft No. 1; (right) Shaft No. 3 (Gunnink and Kiehne 2002). of 3. The tests shown in Figure 63 yield ultimate unit side testing results make it possible to achieve more economical resistances of 2.34 and 2.28 MPa, respectively. These tests designs. The O-cell tests also identify construction deficien- illustrate a typical outcome when field load testing is con- cies, such as inadequate base cleanout (Figure 63 left). ducted; that is, measured unit side and base resistances exceed presumptive values, sometimes significantly. Load The tests reported by Gunnink and Kiehne also illustrate a limitation of testing with a single O-cell at the bottom of the socket. The values of ultimate unit side resistance re- ported by the authors are based on the assumption that all of the load was resisted by the rock socket, neglecting any contribution of the overlying soil. It is not known how sig- nificant the error is for this case, but testing with multiple O-cells makes it possible to isolate the section of shaft in rock for evaluation of average side resistance (however, multiple O-cells increase the cost of load testing). For example, if a second O-cell is located at the top of the rock socket, a test conducted with that cell can be used to determine the com- bined side resistance of all layers above the rock. An innova- tive approach based on this concept is illustrated in the testing sequence shown in Figure 64. The figure and description are from O'Neill et al. (1997) based on tests conducted by LOADTEST, Inc., for the Alabama DOT. Arrangement of the O-cells and the 4-step testing sequence depicted in the figure made it possible to measure ultimate base resistance, side re- sistance of the socket (in both directions), and side resistance of the cased portion of the shaft above the socket. It is noted that this arrangement made it possible to measure a total foundation resistance of 80 MN, compared with approxi- mately 11 MN for the largest standard surface jacks. Instal- lation of multiple O-cells makes it necessary to provide a tremie bypass line to facilitate placement of concrete below and around the upper cells. Interpretation of O-cell tests in rock sockets is typically based on the assumption that total applied load at the ultimate condition is distributed uniformly over the shaft/rock side interface, and used to calculate an average unit side resis- tance by Qoc FIGURE 64 Test setup and loading sequence with two O-cells fs = (148) (O'Neill et al. 1997). BD

OCR for page 73
80 where added to obtain the equivalent top load for a downward dis- fs = average unit side resistance (stress), placement of 10 mm and plotted on a load-displacement QOC = O-cell test load, curve as shown in Figure 65b. This procedure is used to B = shaft diameter, and obtain points on the load-displacement curve up to a displace- D = socket length. ment corresponding to the least of the two values (side or base displacement) at the maximum test load. In Figure 65a, The degree to which this average unit side resistance is valid this corresponds to side displacement. Total resistance cor- for design of rock sockets loaded at the head depends on the responding to further displacements is approximated as fol- degree to which side load transfer under O-cell test condi- lows. For the section of shaft loaded to higher displacement, tions is similar to conditions under head loading. Detailed the actual measured load can be determined for each value of knowledge of site stratigraphy is needed to interpret side load displacement up to the maximum test load (in Figure 65a this transfer. is the base resistance curve). The resistance provided by the other section must be estimated by extrapolating its curve O-cell test results typically are used to construct an equiv- beyond the maximum test load. In Figure 65a, the side resis- alent top-loaded settlement curve, as illustrated in Figure 65. tance curve is extrapolated. The resulting equivalent top- At equivalent values of displacement both components of loaded settlement curve shown in Figure 65b is therefore load are added. For example, in Figure 65a, the displacement based on direct measurements up to a certain point, and par- for both points labeled "4" is 10 mm. The measured upward tially on extrapolated estimates beyond that point. and downward loads determined for this displacement are According to Paikowsky et al. (2004b), most state DOT 80 geotechnical engineers using O-cell testing tend to accept 12 the measurements as indicative of drilled shaft performance 60 11 under conventional top-down loading. O-cell test results are 40 10 applied in design by construction of an equivalent top-load 9 side load-deformation curve is measured 8 settlement curve, as illustrated earlier, or by using the 6 7 Movement (mm) 20 3 4 5 measured unit side and base resistances as design nominal 2 1 side resistance curve values. However, some researchers (O'Neill et al. 1997; 0 is extrapolated 1 2 3 Paikowsky et al. 2004b) have pointed out differences be- -20 4 5 6 tween O-cell test conditions and top loading conditions that 7 8 may require interpretation. The most significant difference is -40 9 measured base load- 10 that compressional loading at the head of a shaft causes com- deformation curve -60 11 pression in the concrete, outward radial strain (Poisson's 12 effect), and a load transfer distribution in which axial load in -80 the shaft decreases with depth. Loading from an embedded 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 O-cell also produces compression in the concrete, but a load O-Cell Load (MN) transfer distribution in which axial load in the shaft decreases (a) upward from a maximum at the O-cell to zero at the head of the shaft. It is possible that different load transfer distribu- 30 tions could result in different distributions of side resistance with depth and, depending on subsurface conditions, differ- 25 ent total side resistance of a rock socket. Equivalent Top Load (MN) 12 10 11 9 20 8 6 7 In shallow rock sockets under bottom-up (O-cell) loading 5 base resistance measured, conditions, a potential failure mode is by formation of a con- 15 side resistance extrapolated ical wedge-type failure surface ("cone breakout"). This type 4 3 of failure mode would not yield results equivalent to a shaft 10 2 loaded in compression from the top. A construction detail noted by Crapps and Schmertmann (2002) that could poten- 5 1 tially influence load test results is the change in shaft diame- ter that might exist at the top of a rock socket. A common 0 practice is to use temporary casing to the top of rock, fol- 0 10 20 30 40 50 60 70 80 lowed by a change in the tooling and a decrease in the diam- Downward Displacement (mm) eter of the rock socket relative to the diameter of the shaft (b) above the socket. Top-down compression loading produces FIGURE 65 Construction of equivalent top-loaded settlement perimeter bearing stress at the diameter change as illustrated curve from O-cell test results (a) O-cell measured load- in Figure 66, whereas loading from an O-cell at the bottom displacement; (b) equivalent top-load settlement results. of the socket would lift the shaft from the bearing surface.

OCR for page 73
81 compression SOIL change in diameter at soil/rock interface ROCK FIGURE 67 Comparison of load-displacement curves; O-cell versus FEM (Paikowsky et al. 2004b). FIGURE 66 Perimeter bearing stress at diameter change under top loading. equivalent top-load settlement curve derived from an O-cell Paikowsky et al. (2004b) reviewed the available data that load test may underpredict side resistance for higher dis- might allow direct comparisons between O-cell and conven- placements; that is, the O-cell derived curve is conservative. tional top-down loading tests on drilled shafts. Three sets of Further FEM analyses reported by Paikowsky et al. (2004b) load tests reported in the literature and involving rock sockets suggest that the differences between loading from the bot- were reviewed. However, in two of the cases the test sequence tom (O-cell) and loading in compression from the top are the involved conventional top-down compression loading (Phase result of differing normal stress conditions at the interface, 1) followed by O-cell testing from the bottom up (Phase 2). and that these differences become more significant with in- Mobilization of side resistance in Phase 1 is believed to have creasing rock mass modulus and increasing interface friction caused a loss of bond, thereby influencing results of the O-cell angle. tests and precluding any direct comparison. The third case in- volved STN and O-cell tests of shafts in Florida limestone. These numerical analyses suggest that differences in the Paikowsky et al. stated that several factors, including highly response of rock sockets to O-cell test loading and top-down variable site conditions and factors related to the tests, pre- compression loading may warrant consideration in some vented a direct comparison of results. cases. Ideally, side-by-side comparisons on identical test shafts constructed in the same manner and in rock with sim- FEM reported by Paikowsky et al. (2004b) suggests that ilar characteristics and properties are needed to assess differ- differences in rock-socket response between O-cell testing ences in response. However, it is expected that the potential and top-load testing may be affected by (1) modulus of the differences, if any, will eventually be identified and incorpo- rock mass, EM, and (2) interface friction angle, i. Paikowsky rated into interpretation methods for O-cell testing. In the first calibrated the FEM model to provide good agreement meantime, the O-cell test is providing state transportation with the results of O-cell tests on full-scale rock-socketed agencies with a practical and cost-effective tool for evaluat- shafts, including a test shaft socketed into shale in Wilsonville, ing the performance of rock sockets and it is expected that Alabama, and a test shaft in claystone in Denver, Colorado, the O-cell test will continue to be used extensively. described by Abu-Hejleh et al. (2003). In the FEM, load was applied similarly to the field O-cell test; that is, loading from Instrumentation such as sister bars with strain gages the bottom upward. The model was then used to predict be- makes it possible to better determine the load distribution and havior of the test shafts under a compression load applied at load transfer behavior during an O-cell load test. This infor- the top and compared with the equivalent top-load settlement mation can then be used to make more refined predictions of curve determined from O-cell test results. Figure 67 shows a load transfer behavior under head load conditions. comparison of the top-load versus displacement curves for the Alabama test, one as calculated from the O-cell test and In summary, some of the advantages of the O-cell for ax- the other as predicted by FEM analysis. The curves show ial load testing of rock-socketed shafts include: good agreement at small displacement (<0.1 in. or 2.5 mm); however, the curve derived from FEM analysis is much Ability to apply larger loads than any of the available stiffer at higher displacement. This exercise suggests that the methods (important for rock sockets) and

OCR for page 73
82 With multiple cells or proper instrumentation, it can an overall program leading to increased use and improved de- isolate socket base and side resistances from resistance sign methods for rock-socketed foundations. The Colorado of other geomaterial layers. DOT has also used O-cell testing to improve its design procedures for rock-socketed shafts, as documented by Abu- Limitations of the O-cell test for use by state DOTs include: Hejleh et al. (2003). Shaft to be tested must be predetermined, because it is not possible to test an existing shaft; Statnamic For each installed device, test is limited to failure of one part of the shaft only; The STN load test was developed in the late 1980s by There are possible concerns using test shaft as a pro- Berminghammer Foundation Equipment of Hamilton, On- duction shaft; tario. Its use in the U.S. transportation industry has been Interpretation methods that account for differences in supported by FHWA through sponsorship of load testing loading mode are not yet fully developed; and programs, as well as tests conducted with an STN device There are currently no ASTM or AASHTO standards owned by FHWA for research purposes. specifically for O-cell load tests. In this test, load is applied to the top of a deep foundation Interviews with state DOT engineers for this study show by igniting a high-energy, fast-burning solid fuel within a that the O-cell test has been an integral tool in advancing the pressure chamber. As the fuel pressure increases, a set of re- understanding and use of rock-socketed drilled shafts. The action masses is accelerated upward, generating a downward Kansas DOT (KDOT) experience is representative of several force on the foundation element equal to the product of the other states. The following is based on an interview with reaction mass and the acceleration. Loading occurs over a Robert Henthorne, KDOT Chief Geologist. The geology period of approximately 100 to 200 ms, followed by venting of the western half of Kansas, located in the High Plains of the pressure to control the unloading cycle. Load applied physiographic province, is dominated by thick sequences of to the foundation is monitored by a load cell and displace- sedimentary rocks, mostly sandstone, shale, and limestone. ment is monitored with a photovoltaic laser sensor. The con- Until approximately 1995, virtually all highway bridges were cept is illustrated schematically in Figure 68. STN equipment founded on shallow foundations or H-piles driven to refusal is available for test loads as high as 30 MN. on rock. Drilled shafts were not considered a viable alterna- tive because of uncertainties associated with both design and Processing of the load and displacement time histories is construction. With encouragement from FHWA, KDOT required to convert the STN measurements into an equivalent, engineers and geologists initiated a long-term program of static load-displacement curve. The analysis accounts for training, education, and field load testing to better match dynamic effects that may include damping and inertial foundation technologies with subsurface conditions. Work- effects. The unloading-point method as reported by Horvath shops on drilled shaft design, construction, inspection, and et al. (1993) provides a relatively straightforward method for nondestructive testing (NDT), sponsored by FHWA and the determining static resistance using measurements made at the International Association of Foundation Drilling (ADSC), top of the shaft during a STN test. Test interpretation is also were conducted at the invitation of KDOT. KDOT began us- discussed by Brown (1994) and El Naggar and Baldinelli ing drilled shafts as bridge foundations where appropriate. (2000). Several bridge sites in western Kansas were designed with rock-socketed shafts. To address the lack of experience with these conditions, O-cell testing was incorporated into the larger bridge projects. In almost every case, the O-cell test results showed side and base resistances considerably higher than the values used for preliminary sizing of the shafts, and valuable experience was gained with construction methods, effective cleanout strategies, NDT methods, etc. KDOT now has O-cell test results on rock-socketed shafts from nine projects and has developed in-house correlations between rock mass properties and design parameters for commonly encountered geological formations. Drilled shafts now comprise approximately 70% to 80% of new bridge foundations, and shaft designs are more economical because there is a high level of confidence in capacity predictions, based directly on the load tests. The approach taken by KDOT illustrates how field load testing, in this case with the O-cell, can be incorporated into FIGURE 68 Schematic of STN load test (O'Neill et al. 1997).

OCR for page 73
83 Mullins et al. (2002) recently introduced the segmental Derived Static = REF*UP-derived capacity unloading point method, which uses top and toe measure- REF = 0.96 for rock ments as well as strain measurements from along the length 0.91 for sand of the foundation. The segmental unloading point method 0.69 for silt enables determination of load transfer along various seg- 0.65 for clay. ments of the foundation, an advantage for rock-socketed shafts to separate resistance developed in rock from that developed in portions of the shaft embedded in soil. The analy- Dynamic Impact Testing sis is automated using software provided by the testing firm and equivalent static load-displacement graphs are produced A dynamic compression load test can be carried out by drop- immediately for evaluation. All data are stored for future ping a heavy weight onto the head of the shaft from various analysis and reference. heights. The shaft is instrumented with strain gages and accelerometers to measure the force and impact velocity of During the 1990s, FHWA performed or funded STN and the stress wave generated by the dynamic impact. The mea- correlation studies with conventional static load tests to surements are correlated to driving resistance to predict load develop standardized testing procedures and data interpreta- capacity. A review of various available drop weight systems tion methods (Bermingham et al. 1994). Numerous other and evaluation of the method is given by Paikowsky et al. studies have further expanded the database of case histories (2004c). A typical drop weight system consists of four and performance studies. The result is that STN testing is components: (1) a frame or guide for the drop weight, (2) the now a well-developed technology that is highly suitable for drop weight (ram), (3) a trip mechanism to release the ram, and use by state DOTs for axial load testing of rock-socketed (4) a striker plate or cushion, as shown in Figure 69. Various shafts. STN advantages identified by Brown (2000) include: configurations of modular weights can be used to provide ram weights as high as 265 kN (Hussein et al. 2004) and drop Large load capacity, applied at top of shaft; heights are adjustable up to 5 m (Paikowsky et al. 2004c). A Can test existing or production shaft; rule of thumb given by Hussein et al. is that a ram weight of Economies of scale for multiple tests; 1% to 2% of the expected shaft capacity be available on site. Amenable to verification testing on production shafts; and Drop weight load testing interpretation relies on analysis Reaction system not needed. methods similar to those used in standard dynamic pile test- ing. Strain gage and accelerometer measurements at the top Disadvantages include: of the pile are used to evaluate characteristics of stress wave propagation. If sufficient shaft resistance is mobilized, it is Capacity high, but still limited (30 MN); possible in theory to relate the stress wave characteristics to Rapid loading method, as rate effects can be significant shaft capacity using available PDA (Pile Driving Analyzer) in some soils (less in rock); technologies. Drop weight testing of drilled shafts has not Mobilization costs for reaction weights; and been used extensively on bridge foundations in the United Not currently addressed by ASTM or AASHTO stan- States, in part because other available methods (e.g., O-cell dards. and STN) provide a more direct measurement of static resis- tance. According to DiMillio (1998), test results on FHWA Mullins, as reported in Paikowsky et al. (2004b), analyzed projects have not demonstrated sufficiently good agreement a database of 34 sites at which both STN and static load tests between drop weight and other tests. The drop weight tests were conducted on deep foundations. The data included load reportedly overpredicted measured capacities. tests on four drilled shafts in rock at two sites, one site each in Florida and Taiwan. The objective of the study was to Drop weight testing for rock sockets is suitable for post- develop recommendations for LRFD resistance factors when construction tests at bridge sites where questions arise during axial compression capacity is based on STN testing. The au- construction regarding the performance of as-built founda- thors recommend a resistance factor of 0.74 for all deep foun- tions. This application is illustrated by the case of the Lee dation types in rock (not specific to drilled shafts) when Roy Selmon Crosstown Expressway, in Tampa, Florida. tested by STN. In addition, a rate effect factor (REF) is rec- The columns supporting an elevated section of roadway are ommended to account for rate effects when using STN founded on drilled shafts socketed into limestone. During results by the unloading point method. The REF varies with construction of the superstructure, one of the columns sud- soil or rock type and recommendations are given here. If the denly underwent more than 3 m (11 ft) of settlement as a segmental unloading point method is used (requiring strain result of the failure of the drilled shafts. Subsequent investi- gages), separate REF factors can be applied to each seg- gations determined that the failed shafts were not founded ment to account for different soil or rock types. This analy- in sound limestone as believed, raising questions about the sis addresses the disadvantage cited previously regarding rate capacity of all 218 drilled shafts supporting the elevated effects. roadway. As part of an investigation to determine how many

OCR for page 73
84 FIGURE 69 Schematic of drop weight system (Paikowsky et al. 2004c). shafts might need remediation, dynamic load tests were con- applicable to shafts that satisfy the criteria for rigid behavior, ducted on 12 of the shafts supporting existing columns using given as the pile driving hammer shown in Figure 70. Testing proved the design capacity of 11 of the 12 shafts tested. This case 2 Ec also illustrates the need for thorough subsurface investiga- Er 1 (149) tion when socketing into limestone. In this case, rock eleva- 2D tions were found to be highly variable. Seismic methods B used in combination with borings in the post-failure inves- tigation provided a more detailed geologic model of site in which Ec = modulus of the reinforced-concrete shaft, Er = conditions. rock mass modulus, D = socket length, and B = socket diameter. Interpretation Framework for Static Axial The analysis is applied to two cases: (1) shear socket un- Load Tests der compression or uplift and (2) complete socket under compression. The shape of a load-displacement curve from a Carter and Kulhawy (1988) and Kulhawy and Carter (1992b) load test is modeled in terms of constant slopes (S), which are proposed a method for interpretation of static axial load tests related mathematically to the model parameters described in on rock-socketed shafts. The method involves analyzing a chapter three. Consider the load-displacement curve for a static axial load-displacement curve from a load test accord- shear socket loaded in compression, as shown in Figure 71. ing to the analytical closed-form solutions presented in chap- Three parameters are required to idealize the geometry of ter three (Eqs. 6995). The parameters back-calculated from the curve. S1 is the slope of the initial portion, S2 is the the load test could then be used to evaluate effects of various approximated slope of the full-slip portion of the curve, and design parameters on the load-displacement behavior of trial Qi is the intercept on the vertical axis (wc = 0) of the line with designs that differ from that of the test shaft. The method is slope S2. For a rigid shaft, the measured curve parameters

OCR for page 73
85 theoretically are related to the elastic model parameters by the relationships given here (a): (a) Shear Socket -- Compression or Uplift (1 + r ) Er = S1 D 1 S2 tan tan = 2 S1 - S2 c = ( 2 tan tan + 1) Qi BD with = ln 5 (1 - r ) D (150) B in which r = Poisson's ratio of the rock mass, = interface friction angle, and = interface angle of dilation, and c = in- terface cohesion. For a complete socket under compression in which the base load-displacement is determined (Figure 72), the load- displacement curve is approximated by S1, S2, Qi, and S3, the slope of the base load-displacement curve. The curve parame- ters are related to the elastic model parameters as given in (b), (b) Complete Socket -- Compression (1 + r ) Er = ( S1 - S3 ) D FIGURE 70 Dynamic load testing of shaft-supported column in Eb = ( 1 + b 2 ) S3 Tampa, Florida. B 1 S - S3 tan tan = 2 2 S1 - S2 c = ( 2 tan tan + 1) Qi BD in which Eb = modulus of the rock mass beneath the shaft base. Carter and Kulhawy (1988) applied the technique described to 25 axial load tests reported in the literature by back- calculating values of the model parameters Er, Eb, c, and (tan tan) from load-displacement curves using the equations given previously. A limitation of the model described earlier is that the assumption of rigidity may be less acceptable for shafts in harder rocks where the modulus values for the rock mass and the shaft material are closer. The reader is advised to review the original publications for further assumptions and derivations of the equations. Lateral Load Testing FIGURE 71 Interpretation of a side-shear-only test (Carter and A significant number of states indicated in the questionnaire Kulhawy 1988). that lateral loading governs the design of rock-socketed

OCR for page 73
86 depths corresponding to approximately the first 10 diameters of the shaft. Other important points to consider when con- ducting conventional lateral load tests, as pointed out by O'Neill and Reese (1999) are summarized as follows. The test site conditions and test shaft should be selected and built to match as closely as possible the actual conditions to which they will be applied. Items such as overburden stresses acting in the resisting soil and rock layers, ground- water and surface water conditions, shaft dimensions and re- inforcing, and construction methods all can have a significant influence on the lateral load response of a drilled shaft. To the extent possible, these conditions should be matched by those of the load test. Analysis of the load test results will be interpreted using the analytical methods presented in chapter four. The most widely used method is the p-y curve method, in which p-y curves are fit to obtain agreement with the load test mea- surements. As a minimum, it is therefore necessary to have reliable measurements of ground line shear load, ground line FIGURE 72 Interpretation of a complete socket test (Carter and Kulhawy 1988). deflection, and rotation (requires two deflection points sepa- rated by a known vertical distance). To define p-y curves ac- curately over the length of the shaft requires measurements of the deflected shape of the shaft, which can be done using shafts for a significant percentage of projects (Question 25). slope inclinometer measurements. A more accurate method However, as noted in chapter four, very few lateral load tests to determine p-y curves (or to evaluate any analytical have been conducted on rock-socketed shafts. Methods for method) is to establish bending moment as a function of conducting lateral load tests on deep foundations include depth, which can be done by installing a steel tube with conventional methods, Osterberg load cell, and STN. closely spaced strain gages along the length of the shaft. This approach is most appropriate for tests conducted for applied Conventional Lateral Load Test research; for example, to develop new methods for estab- lishing p-y curves in rock. The conventional method for conducting a lateral load test is given in ASTM D3966 and involves pushing or pulling the Boundary conditions must be considered carefully when head of the test shaft against one or more reaction piles or back-fitting analytical models and then applying the model shafts. A variety of arrangements for the test shaft and reac- for design. In a lateral load test, the boundary conditions at tion shaft are possible and these are given in detail in Reese the head of the shaft will normally be free of any rotational (1984) and Hirany and Kulhawy (1988). One approach is to restraint and have zero applied moment and zero axial load. use two shafts and apply the load such that both shafts are Service boundary head conditions are likely to include some tested simultaneously, providing a comparison between two head restraint and possibly axial load and moment. Also, the shafts. A load cell is used to measure the applied lateral load nonlinear momentEI relationships must be accounted for and dial gages or displacement transducers attached to a ref- both in the load test and in the analysis. erence beam can be used to monitor lateral deformation. Thorough treatment of instrumentation for lateral load tests Four states (California, Massachusetts, New Jersey, and can be found in Reese (1984) and Hirany and Kulhawy North Carolina) reported the use of conventional lateral load (1988). tests on rock-socketed shafts. Although lateral load testing is not as common as axial testing, conventional testing has been Drilled shafts are often used where the designer wishes to the method of choice for lateral. Other methods have, so far, take advantage of their large lateral load capacity, especially been used on a limited basis. These include lateral O-cell that of large-diameter shafts. Analysis often shows that the tests (at least two states, South Carolina and Minnesota) and geomaterials in the upper part of the ground profile have the lateral STN (Alabama, Florida, Kentucky, North Carolina, most significant influence on lateral deformations and lateral South Carolina, and Utah). Several states that did not respond load transfer. A critical part of lateral load testing is to have to the survey are known to have conducted lateral STN (Ohio detailed knowledge of the site stratigraphy, particularly at the and Virginia).

OCR for page 73
87 Lateral Osterberg Load Test allowing the reaction masses to slide on rails, as shown in Figure 74. The lateral STN test can simulate lateral impact The O-cell can be embedded in a drilled shaft and oriented loading such as might occur against a bridge pier from a such that the load is applied in the horizontal direction. The vessel. method is described by O'Neill et al. (1997) for a case in which the Minnesota DOT required representative p-y curves The lateral STN test can also be used to derive the static for a stratum of friable sandstone situated beneath a thick lateral response, but requires appropriate instrumentation layer of normally consolidated clay. Shafts socketed into the and correct analysis of the test results. In tests described by sandstone were to support a bridge undergoing ice loading. Brown (2000), the following instrumentation was used: The test was conducted at a nearby location in which a 26.7 MN O-cell was positioned vertically within a 1.22-m-diameter Load cell, socket, as depicted in Figure 73, and used to thrust the two Displacement transducers, halves of the socket against the rock. Lateral force and de- Accelerometers on top of cap or shaft, flection measurements were used to derive p-y curves. The Downhole motion sensors, authors point out that care must be taken in interpreting Resistance-type strain gages, and the results, because the stressstrain conditions created by Megadac Data Acquisition System. the test are not the same as in a laterally loaded socketed shaft that is loaded at its head and not split. Figure 75a shows the measured dynamic response of the shaft in terms of force, acceleration, and lateral displacements Lateral O-cell testing of rock sockets offers some of the versus time. The curves showing measured lateral displace- same advantages as for axial O-cell load testing, namely the ment from three measurements are identical and cannot be elimination of a structural loading system at the ground level. distinguished in the figure. Dynamic response is separated Also, the test provides the ability to apply lateral loading at pre- into static, inertial, and damping components. A p-y analysis determined depths, such as within the rock socket. Further re- (using LPILE or FBPIER) is fit to obtain a reasonable match search is needed to establish guidelines for proper procedures between the measured load-displacement response for each and to define correct analyses that account for the differences component of force (static, inertial, and damping). Load ver- in boundary conditions, load transfer, and soil and rock resis- sus displacement curves derived are shown in Figure 75b tance, compared with a shaft loaded at its head. It is also worth based on analysis of the dynamic response in Figure 75a. noting that the lateral split socket test may provide a means to measure the in situ rock mass modulus of deformation (EM). The lateral STN test is reported as to be safe, controlled, and economical. Its principal advantage lies in the ability to measure directly the dynamic lateral response and to provide Lateral Statnamic a derived static response. This test is a valuable tool for the design of bridge foundations to withstand dynamic lateral The STN load test has also been adapted for lateral loading. loading from earthquakes, wind, and vessel impacts. The test The device is mounted on steel skids supported on the ground may also be used in place of a conventional static test. Lateral loads up to 18 MN may be possible. FIGURE 73 Top view of O-cell arrangement for lateral split socket test (O'Neill 1997). FIGURE 74 Lateral STN load test (Courtesy: L. Fontaine).