Published in the Proceedings of the GEO Construction Quality Assurance/Quality Control Conference ‘Sponsored by ADSC: The International Association of Foundation Drilling, Editors: D.B. Bruce and Cadden, A.W, November 6-9, 2005, Dallas, TX, 596 p. INTEGRITY TESTING OF DRILLED SHAFTS - EXISTING AND NEW TECHNIQUES Naresh C. Samtani, PE, PhD', Frank Jalinoos” and Dennis M. Poland, RG? ABSTRACT The use of drilled shafts as a cost-effective deep foundation system has grown ‘tremendously in the last couple of decades. With the advent of non-redundant large diameter single shaft foundation systems it has become critical to ensure that the constructed product meets the design requirements. Non-destructive tests (NDT) are commonly used to evaluate the structural integrity of the constructed drilled shafts. The two most common NDTs are performed using the Cross-hole Sonic Logging (CSL) and the Gamma-Gamma Density Logging (GDL) methods. In addition to discussing these common NDT methods, this paper presents their newer variations and how they are used to identify anomalies. A roadmap is presented to progress from the identification of anomalies to the verification and imaging of defects and finally to methods for remediation of defective shafis. NON-DESTRUCTIVE TESTS (NDT) Non-destructive test (NDT) methods are used for Quality Assurance (QA) integrity testing of drilled shaft foundations to identify anomalies. An anomaly is a deviation from an assumed uniform geometry of the shaft and/or shaft properties (e.g., homogeneity). Typical anomalies may include necking or bulbing, “soft bottom” conditions, voids or soil intrusions, poor quality concrete, debonding, lack of conerete cover over the reinforcement and honey-combing, NDT testing techniques can be categorized as external or internal, External NDT techniques are used at the surface of the concrete structure when access to the interior of the concrete is not available. Examples of external NDT techniques include Sonic * President, NCS Consultants, LLC, 640 W Paseo Rio Grande, Tucson, AZ 85737; [Corresponding Author: E= ‘ail: Nsutani@imsn,com ; Tel: ($20) 544.2876) 2 President, InfaSeis, Inc, 6504 Moss Circle, Arvada, CO 80007 » Director of Marketing, Anderson Drilling, 10303 Channel Road, Lakeside, CA 92040 329Echo (SE), Impulse Response (IR) or Ultra-seismic (US), Internal NDT techniques are used when testing equipment can access the interior of a concrete structure through either cast-in-place access tubes or cored access paths or through cast-in- place equipment within the concrete (e.g., strain gages). Commonly used internal NDT techniques include standard Cross-hole Sonic Logging (CSL) with zero-ofiset measurements and Gamma-Gamma Density Logging (GDL). _Cross-hole Sonic Logging Tomography (CSLT) using multi-offset CSL method is a logical newer extension of the CSL technique and is starting to gain acceptance. The Perimeter Sonic Logging (PSL) is yet another new variation in which zero-offset or multi-offset, CSL may be performed in PVC tubes attached to the outside of the reinforcing cage. Other more specialized intemal NDT techniques include the Neutron Moisture Logging (NML) and Temperature Logging (TL) (Jalinoos ef al., 2005). Descriptions of the above-mentioned methods can be found in O’Neill and Reese (1999), Samtani (2001) and Jalinoos et al. (2005). Brief descriptions of the commonly used intemal NDT techniques such as CSL and GDL methods are provided here for the sake of completeness. The CSLT and PSL variations of the CSL test are also briefly described. A roadmap for using these methods is proposed based on our observations of the current industry practices. Finally, a brief overview of the possible shaft remediation methods is presented. THE STANDARD CSL TEST In the standard CSL test method, an ultrasonic transmitter (or source) and receiver probes are first lowered to the bottom of a pair of water-filled cast-in- place access tubes as shown in Figure 1. It is common industry practice to locate the access tubes inside the reinforcing cage. The two probes are then pulled up simultaneously such that the probes are level with each other and near horizontal (or zero-offset). The travel time of the ultrasonic wave between the tubes is recorded along with the amplitude of the signal as a function of every inch of depth, This . test procedure is repeated for all | Figure 1: Schematic of CSL Test possible paired combination of access tubes along the outer perimeter as well as across the inner diagonal of the shaft as shown in the inset Plan View in Figure 1. The measured travel time, t, between two tubes with a known center to center distance, d, is expressed in terms of velocity as: V (~d/t). This computed velocity, V, is compared with the theoretical compressional wave velocity, Vc, in a homogeneous, isotropic and elastic medi 330Ve=(aE/p)* with a= [(1-W/{(+H)1-20)}] and p= y/g a where E, 4, p and y are the modulus of elasticity, Poisson’s ratio, mass density, and unit weight of the elastic medium, respectively, and g is the gravitational acceleration (32.2 fl/sec’). With a typical reinforcement density of 1 to 2% (by volume), the primary elastic medium in case of drilled shafts is concrete. The current industry practice uses conerete with a 28-day unconfined compressive strength, f, in the range of 3,000 to 5,000 psi. For common sand-aggregate-cement based concrete mixes, Poisson's ratio ranges from 0.1 to 0.2 and the unit weight of hardened concrete, 7, ranges from 145 to 150 pef with a corresponding mass density, p. of approximately 4.5 to 4.7 slugs/ft (note 1 slug = Ib-sec*/ft). For concretes with f. < 6,000 psi, the modulus of elasticity of concrete, E,, can be expressed as follows: Ec (psi) = 57,000 f° (Note: fe is in units of psi) (2) Based on Eqs (1) and (2) the theoretical ultrasonic wave velocity in competent conerete with fe in the range from 3,000 to 5,000 psi would be approximately 10,000 to 11,500 fi/see, respectively. In reality, the velocity increases with frequency and for a 40 kHz transmitted frequency the velocities are about 15% higher. As a comparison, the sonic velocity in water and air is approximately 5,000 fi/sec and 1,000 fu/sec, respectively. ‘The computed velocity is compared with the theoretical velocity and expressed in terms of velocity reductions, VR = (1-V/Vc)(100)%. A qualitative rating is assigned to the concrete based on VR, as follows: VR (Rating): 0-10% (Good); 10-20% (Questionable) and >20% (Poor) The ratings are partially based on the estimated reduction in strength of concrete in anomalous zones. For example, the correlations in Eqs (1) and (2) indicate that while there is a second power relationship between velocity and unit weight, there is a fourth power relationship between velocity and unconfined compressive strength. Based on this fourth power relationship, it can be deduced that if VR=10% at a given location in a shaft, then the f at that location is approximately 65% of the nominal 28-day fe value of the conerete in that shaft. Similarly, a concrete with VR=20% implies that fz at that location is 40% of the 28-day strength. With the exception of voids and possibly honeycombs, the locations of poor conerete can be confirmed by checking the signal amplitudes. Weaker concrete absorbs the energy of the sonic wave and this phenomenon is reflected in lower signal amplitudes. Thus, if the measurements in the shaft indicate lower velocity and lower signal amplitudes then they typically point to anomalous zones due to soil intrusions or poor quality concrete. An example single plot display format that includes velocity and signal amplitude profiles is shown in Figure 2. In this particular case, it can be seen that a soft bottom condition in the shaft is reflected at the very bottom of the profile by a drastic change in both in the velocity and amplitude profiles. 331Line Legend: 10% Velocity — - - = Drop 120% Velocity _ Drop Velocity Curve Amplitude: If the tubes debond from the concrete, ie., there is a small air gap, then the CSL. test will record a partial or complete loss of signal depending on the extent of the debonding around the perimeter of the tube at that location, Debonding can occur with Schedule 40 PVC access tubes particularly near the ground surface or above the groundwater table where temperature gradients are generally greater. For Schedule 40 PVC tubes, the debonding may occur within a week after placement of concrete as the concrete sets and tends to shrink away from the tubes. Thus, if Schedule 40 PVC tubes are used then it is generally recommended to perform the CSL tests within 2 to 3 days after concrete placement. A thicker wall PVC tube, such as a Schedule 80 tube, may help extend this timeframe because it is able to withstand the higher temperature gradients better than a thinner PVC tube. Longer time frames can be achieved by the use of steel tubes which experience minimal to no debonding. Therefore, many owners tend to specify steel tubes to alleviate the debonding problems. However, in doing so, the owners are giving up an advantage of the PVC tubes in that they can serve as access paths to repair the shafts should an anomaly be identified by the CSL test since the PVC tubes can be cut open at any depth by use of a “water knife” that utilizes a high velocity water jet. Use of a water knife is much 332more difficult, if not impossible, in steel! due to the practical limitation of generating a very high water velocity at depth within the access tubes. THE CSLT TEST While the location of an anomalous zone can be defined using the zero-offset CSL method, the volume of the zone cannot be defined with confidence, The CSLT test can be used to define the volume of the anomalous zones. The principles of the CSL and CSLT tests are the same. ‘The difference is in the location of the transmitter and the receiver. In the CSLT method, data is collected between a pair of tubes by running a zero-ofiset log in combination with several positive offset (receiver is shallower) and negative offset (source is shallower) logs as shown in Figure 3. This procedure is repeated in all possible paired tube combinations to form a three- dimensional (3-D) tomography dataset. A typical format for presentation of 3-D data is shown in Figure 4. In Figure 4a, poor quality concrete zones are located at a depth of approximately 17m. ‘The corresponding estimated location and size (volume) of the soft bottom condition extending into the shaft interior is shown in Figure 4b. _F iy Figure 3: Zero-Offset Versus Multi-Offset Data Collection Procedure. (a) Zero-Offset Test, (b) Ofiset Tests, (c) Integration of All Data The process for generating the data shown in Figure 4 involves complex numerical procedures that are based on several assumptions which affect the predictive capabilities of the model (Jalinoos, ef al., 2005). In particular, “artifacts” can be generated due to edge effects. Artifacts are erroneous velocity values produced by the numerical inversion processes mostly due to the inadequate scanning of the test volume. These artifacts mostly occur near the image boundaries (top and the bottom of the shaft). There are several specialized software packages currently in use that, for the same dataset, may generate slightly different images and therefore 333may give several different interpretations regarding the volume of the anomalous zones. The user should not use CSLT results, such as that shown in Figure 4, as an end product to require repairs but rather use it in conjunction with other engineering analytical tools (e.g. LPILE program), as will be described later. 123 1 7 & : Tube Poston ® ©) re 4: CSLT Data Display Format THE GDL TEST A typical field setup for the GDL test is shown in Figure 5. In this test a weak Cesium-137 (radioactive) source emits gamma rays into the surrounding medium. A small fraction of the gamma ray photons are reflected back to the probe due to Compton scattering. The intensity of the reflected photons is recorded by a Nal scintillation crystal as counts per second (cps). The ‘measured count rate (cps) depends on the electron density of the surrounding medium, which is proportional to the mass per unit volume. The instrument is calibrated by placing the probe in an environment of known density in order to convert the measured count rate (cps) into the units of density or unit weight, e.g. Ib/ft(pef). ‘igure 5: Schematic of GDL Test In the GDL test, the radius of the investigation is largely governed by ¥4 of the source-detector spacing. Good concrete conditions will result in a near continuous alignment of the data. Anomalous zones due to soil intrusions, poor concrete or voids are characterized by low density which leads to a high count rate. 334A typical GDI log is shown in Figure 6. In a GDL log, the measured gamma ray intensity count rate (cps) is presented in terms of unit weight (pef). In Figure 6, the results are plotted in 4 separate sub-plots from the tested access tubes. Each individual sub-plot depicts the GDL results from a 14-inch source-detector separation (corresponding to about 5- to 6-inch radius of investigation) presented in a magnified density scale of 130-180 pef. Also, in each sub-plot, the mean as well as the minus 2 (2) and minus three (-3) standard deviation (SD) from mean curves are displayed as vertical guidelines. Depths, in feet, are measured from the top of the shaft and are shown on the vertical axis. __Tubel Tube2 Depth, ft Tube3__ Tube 4 Density Curve: Line Legend: Mean (M) =—— -2 Std Dev (SD) from Mean -3 Std Dev (SD) from Mean 130 peat TRO pet 180 Figure 6: Data Display Format for GDL test. The results of GDL tests are used to define “questionable” concrete conditions as a zone with reduction in unit weight between -2SD and -3SD and “poor” conerete conditions as a zone with reduction in unit weight of greater than -3SD from the mean (M). These criteria are based on the observation that a cps data set approximates a standard normal distribution probability function in which 99.73% of the data is within M&3SD. Therefore, when data points are identified beyond 3SDs, they are considered to represent an anomaly. While these definitions are generally accepted, it is not widely recognized that the computation of M and SD is inconsistent during presentation of the results by various testers/agencies. Some testers or agencies define the M and SD with respect to a given tube while others may define these quantities based on all tubes within a shaft (i. ignore the variation of steel density and hole geometry) or all tubes from a group of shafts that may form a single overall foundation element for a superstructure. Obviously, the definition of the concrete quality will be different based on the definition of the M and SD. Therefore, the user should be careful with the interpretation of the GDL test data. 335Unlike the CSL test, the GDL is not affected much by debonding of the tubes from the concrete. Therefore, a PVC tube is generally used, although steel can also be used with GDL testing. It must be recognized, however, that the thicker or denser the tube material, the lower the measured counts per second (cps) since the tube itself is absorbing some of the electrons. Therefore, the user of the data should review the calibration data and check whether the tube type used during calibration is consistent with that used in the actual shaft and the density of the shaft reinforcement. THE PSL TEST Figure 7 shows a schematic of a shaft where both CSL and GDL tests are performed. Since the standard CSL test is performed with test tubes located inside the cage, it can only verify the condition of concrete within the cage and, more specifically, between any given pair of tubes (see oblong zones between tubes in Figure 7). The GDL tests only a limited zone | _ Figure 7 within a radius of 5- to 6-inches around the test tube (see filled circles in Figure 7). It can be seen from Figure 7, that even if both tests are performed, there is a large core area and an even more important region immediately adjacent to the reinforcing cage that remains untested. It is in the region near the cage perimeter that problems may occur and go undetected by both the CSL and GDL tests. This problem may be acute when the reinforcing cage is dense and the concrete cannot easily flow through it. In the PSL test, the CSL technique is implemented in water-filled Schedule 40 PVC tubes located on the outside of the reinforcing cage with the specific intent of checking the condition of the concrete immediately near the perimeter of the reinforcing cage. Moving the tubes from inside to outside of the cage may not seem to be much but it is enough to pick up variations in concrete quality near the reinforcing cage since the sonic waves are traveling closer to the cage perimeter. The first author recently implemented the PSL test on a bridge project in Tueson where the condition of the shaft near the cage perimeter was a concern but the project budget did not allow for implementation of a dual CSL and GDL test program. The second author performed the testing program. The shafis were constructed using polymer slurry in a dry hole to maintain stability of raveling sandy formations in the Cafiada Del Oro Wash. The PSL method identified defects in several shafis. The contractor exposed the shafts to a depth of 30-ft to verify the presence of the defects. Figure 8 shows photos of the large defect that is clearly observed. From the photos it is evident that the concrete within the cage is intact and of good quality. If standard CSL tests with tubes on the inside of the cage had been performed (with no GDL test) the anomaly would not have been detected. The PSL test was performed on 20 pier shafis on the project in Tucson. Numerous defects such as those shown in Figure 8 were discovered and several shafts were exposed as shown in Figure 8 to confirm the anomalies. Each suspected anomaly was confirmed. This validated the repeatability of the test process. 336igure 8: Large Defect Outside the Cage Detected by the PSL Test. One shaft was fitted with PVC tubes inside and outside the cage and both PSL and standard CSL tests were performed. This shaft also had anomalies similar to those shown in Figure 8. With respect to the PSL data, as expected, it was found that the signal amplitudes were considerably altered due to the sonic waves alternately passing through concrete and steel. Not only energy drops of up to 70% were measured, but the data was noisier (making it less suitable for CSLT imaging). Therefore, rather than quantify the signal amplitudes, the results were interpreted with respect to maximum signal amplitudes that are assumed to reflect no anomalies, Similarly, anomalous zones were located by identifying distinct decreases in velocity data compared to the local averages within a given tube. LOCATION OF TUBES AND THEIR EFFECT ON NDTs AND ANOMALIES. The conventional wisdom in the drilled shaft industry is that all test tubes should be located inside the reinforcing cage. This is a good practice if one wishes to concentrate only on the condition of the concrete within the cage and in localized zones around the tubes as shown in Figure 7. Tubes on the inside of the cage allow the collection of a dataset that, for the CSL test technique, can be interpreted using the theory of elasticity assuming homogeneous, isotropic and elastic media discussed earlier since the sonic waves will travel only through uniform concrete if there are no defects. If the sonic waves have to pass through the cage then these assumptions may not be fully valid. However, if one recognizes these limitations and is careful in interpretation of the velocities and the signal amplitudes, then it is possible to identify major defects by conducting PSL tests with the tubes outside the cage as discussed earlier. Further research is needed to quantify the velocities and signal amplitudes through altemating steel reinforcement and concrete as would be the case in PSL tests. For example, higher energy and slightly lower frequency sourees may be better for PSL tests. Therefore, at this time, interpretation of the data collected from access tubes located outside the cage is best left to qualified and experienced personnel. To minimize the potential for anomalies, it is recommended that (a) the concrete has a high slump (> 7") so that it is flowable, (b) a minimum slump of 4-inches is maintained within the rising concrete column at all times during the placement of concrete in the shaft, (c) the minimum opening size in the cage is 5-inches, ie, bundle the longitudinal reinforcement if necessary, (d) centralizers are not located 337near the tubes, (¢) a minimum 6-inch concrete cover is provided, and (f) the tremie pipe is not breached. Often the reinforcement cage is dense in the sense that the openings between the longitudinal and transverse reinforcement are such that the concrete may have difficulty in passing through the reinforcement. In such cases, the presence of test tubes may contribute further to the problem if they are attached directly to the cage thereby effectively reducing the opening size. This problem is further compounded if the tubes are located near the centralizers. In such cases, a self- [Figure 9: Schematic and consolidating (highly fluid) concrete mix may be [Photo of Cradles. used. If the major concern is verification of the concrete quality within the cage then the tubes may be tied to a “cradle” as shown in Figure 9 to alleviate the problem. Placing the tubes away from the cage via cradles also permits better flow of concrete through the cage and mitigates the potential development of the anomalies. CURRENT STANDARD OF PRACTICE, The authors performed a limited survey of the NDT practices in several states including AR, AZ, CA, CO, DE, HI, KS, LA, MA, MO, MT, NM, NJ, NV, OR, RI, SD, VT, WI and WA. It was found that the NDT practices varied considerably. Some states have a minimal testing program (e.g., CDOT) while others have a strong preference for CSL tests (e.g., NMDOT) or GDL tests (e.g. CALTRANS) or both (eg. ADOT). While the CSL test method is based on ASTM D6760-02 there is no ASTM “standard” test method for GDL for agencies to follow. A lack of guidelines to evaluate anomalous shafts make the “decision makers” nervous who then require unnecessary and costly shaft repairs. Therefore, there appears to be a need for a roadmap for using the NDT techniques efficiently to define potentially defective concrete drilled shaft foundations. Such a roadmap will also help in making appropriate integrity evaluation within days of field testing rather than within wweeks/months as in the current practice. A PROPOSED ROADMAP FOR USING NDT METHODS A 3-step roadmap, adapted from Jalinoos, ef al, (2005), is presented here to assist owners, agencies and engineers in evaluating shaft anomalies. The first step is analogous in medical practice to the initial examination and diagnosis by a general practitioner. ‘The second step is analogous to the verification of the diagnosis through imaging (x-ray, MRI, CAT scan, ete.) of an unhealthy member. The third step is analogous to conducting external or internal remedies by a specialist. Step I Anomaly Identification (and Verification) — In this step, the owner initially performs dual CSL and GDL testing at any time within 2 to 5 days after concrete placement to sereen for sound and anomalous shafts ~ as a whole - and independently 338verify the existence of the anomalies. Dual testing, i.e., CSL and GDL, is required to assess shaft integrity both inside (“core” of the shaft) and outside the rebar cage (cage “cover”) and to determine rebar’s exposure to soil/moisture, Dual testing is also required to independently verify anomalies and discern “false positives” (or “false negatives”) of a particular test method. Conereting curve from concrete placement data should be used for a preliminary confirmation of the anomalies. In eritical cases, such as non-redundant shafts constructed using wet (slurry) methods, PSL tests may be considered where the budget does not allow for dual CSL/GDL testing, Step IL Strength Imaging of Defects — For shafts suspected to be anomalous in the Step I, CSLT (tomographic imaging) is performed at least 7 days after the concrete placement. A curing time of 7 days will allow the shaft concrete to attain properties that can be correlated with known concrete properties either through laboratory tests or published literature. It is suggested that before a specimen is subjected to strength test, a relatively cheap (@ $20/sample) ultrasonic pulse velocity test measurement be performed to allow for a more reliable calibration of the CSL velocity data. In this, way, the sonic velocities in the tomogram will be similar to the laboratory ‘measurements and a more realistic velocity-strength correlation will be obtained for developing the final strength images. As an alternative, CSLT testing may be done at the same time as CSL/GDL testing and less than 7 days after the concrete placement. Concrete velocity-strength correlations are obtained in the same way as indicated earlier but after consideration for maturity (curing). ‘The tomographic imaging software should be capable of performing a true 3-D numerical analysis, Step IIL. Shaft Integrity Evaluation and Remediation ~ In this step, the sirength image is used to adjust the original design assumptions and perform appropriate re- analysis for axial and lateral load capacity to determine if the shaft is still serviceable, whether it can be repaired by remediation, or if it is unacceptable. In this way, a final integrity (and pay-factor) can be assessed by the engineer more quantitatively than allowed by the present practice. ANOMALY ANALYSIS AND REMEDIATION MEASURES An indication of an anomaly from a NDT program does not automatically mean that a shaft needs to be repaired. In fact, many anomalies may have minimal effect on the structural integrity of the shaft. This is because the reinforcing cage is typically fall length designed to service the maximum moment and/or shear force and this design steel quantity is not varied along the cage. This means that unless the anomaly is located near the point of maximum moment and/or shear, or the cage is subject to corrosion due to its contact with soil, it is likely that the shaft has excess structural capacity that may be able to overcome the effects of the anomaly, Where an anomaly is such that the cage is in contact with the soil, corrosion can lead to a long-term deterioration of the structural capacity. In this case the corrosion loss rates may be evaluated based on the electrochemical properties of the soils near the location of the anomaly. The loss of steel section may then be evaluated over the design life of the 339structure and an effective steel area may be estimated. Depending on the type and size of the anomaly, a detailed re-analysis of the axial and lateral load capacity should be performed to evaluate of the effects of an anomaly in relation to the structural and geotechnical demand rather than arbitrarily requiring shaft repair. If repair is deemed necessary then several remedial measures may be considered. For anomalies extending to the exterior of a shaft and located at shallow depth, the shaft may be exposed and patched. For deeper anomalies located near the tubes, remediation may be performed by cutting open the PVC tubes at appropriate locations bya “water knife.” If steel tubes were used, then additional core holes will have to be drilled for performing remediation. For soft bottom conditions, the bottom cap can be pierced by a sharp object and remediation performed through the bottom of the tube. To facilitate remediation of soft-bottom conditions, consideration may be given to fitting the bottom of steel tubes with plastic caps that can be easily cut if necessary. For anomalies in the shaft core or deep seated anomalies, the repair method will likely be pressure, compaction or jet grouting. Micro-fine cement should be used for pressure grouting, while for compaction grouting the grout slump should not be more than I-inch. Grouting pressures should be selected such that they do not exceed the 2/3 to 1 times the overburden pressure depending on the soil type to prevent detrimental effects such as heave or adverse subsurface lateral movement. In any case, it is advisable to clean all loose or weak materials from the anomalies by controlled air or water flushing before any repair is attempted. CLOSURE This paper provides a discussion of common NDT methods as well as some new techniques. A roadmap to maximize their benefits is suggested. It is hoped that the discussions herein and the roadmap will provide guidance to the drilled shaft industry. ACKNOWLEDGEMENTS The authors wish to acknowledge the review performed by Edward Nowatzki, PE, PhD, of NCS Consultants, LLC. REFERENCES Jalinoos, F., Mekie, N., Grimm R.E. and Hanna K. (2005). Defects in Drilled Shaft Foundations ~ Identification, Imaging and Characterization. Publication No. FHWA-CFL/TD-05-002, Federal Highway Administration, Central Federal Lands Highway Division (CFLHD), Lakewood, CO, 138 p. O'Neill, M. W. and Reese, L.C. (1999). Drilled Shafts: Construction Procedures and Design Methods. Report No. FHWA-IF-99-025, Federal Highway Administration, Washington, D.C., 758 p. Samtani, N. C. (2001). In Situ Tests and Optimization of Drilled Shaft Design. Design Manual and One-day Short Course for Arizona Dept of Trans., 176 p. 340