112 TRANSPORTATION RESEARCH RECORD 1546 Condition Assessment of Installed Timber Piles by Dispersive Wave Propagation SHUNYI CHEN AND Y. RICHARD KIM Timber piles are widely used for supporting bridges, piers, wharves, and other marine structures. As they age, it becomes critical that their in situ condition be assessed so their remaining service life can be evaluated. Current inspection methods involving visual examinations and sounding tests are unable to quantitatively disclose a pile’s degree of deterioration, depth of penetration, or remaining load-bearing capacity. Years of exposure to wood-decomposing fungi and weathering may have substantially decreased a pile’s effective cross-sectional area, so that the pile can no longer function as originally intended. A study was conducted in which nondestructive dispersive wave propagation tests were applied to both laboratory pile models and field timber piles. The laboratory models consisted of acrylic cylinders having different wall thicknesses to simulate various levels of internal deterioration and timber posts with drilled holes to simulate damage created by marine borers. Seven installed and two uninstalled field timber piles were tested. Laboratory experiments indicated that the dispersive wave propagation test is an excellent means for evaluating the degree of hollowness and borer damage. Field experiments further verified the applicability of the dispersive wave propagation in finding the condition of the installed timber piles. The approaches found to be promising for the condition assessment were (a) phase velocity comparison between the first pass and the return pass and (b) wave speed versus test location. The general trend of the dispersion field in condition assessment is the higher the wave speed, the better the condition. Timber pilings are one of the most commonly used structures in bridge construction because of the relatively low cost of the raw wood and installation. These economic advantages are partially offset by the timber’s proneness to decay and deterioration from insects, marine organisms, fungi, and weathering. As timber piles age, it becomes critical that their in situ condition be assessed so their remaining service life can be evaluated. While there are a number of schemes to protect timber piles from biological damage, a widespread practice is to simply replace all the piles at regular intervals. A preferable criterion for replacement should be based on a reliable evaluation of the structural integrity of a pile. Unfortunately, the current practice of visual inspection and sounding does not provide a quantitative measure of the pile’s condition. Inaccuracy in these approaches has resulted in actual failure of the piling in service (safety consideration) as well as undocumented economic loss (economic consideration), which can be a significant amount considering the tens of thousands of pilings in service on the coast of the United States. S. Chen, 2775 Lexington Ave., No. 123, Roseville, Minn. 55113. Y. R. Kim, Department of Civil Engineering, North Carolina State University, Raleigh, N.C. 27695-7908. Presented here are results from two research projects conducted at North Carolina State University to determine whether dispersive wave propagation methods can be used for assessing the condition of installed timber piles. The techniques involve dispersive transverse and surface wave propagation tests to determine whether the condition can be related to wave velocities, wave intensities, or signal’s changing spectral content. In working toward a solution to this problem, several methods have been proposed and carried out by different organizations (1–3). These methods use a longitudinal wave. However, the application of the longitudinal wave method requires either open access to the top of a pile (1) or a modified approach to create a longitudinal wave (2). As a consequence, the method is either costly or not feasible in most cases because of the presence of superstructure. The dispersive wave propagation technique was used in this study. The surface and transverse wave groups traveling along the length of a pile resulting from a transverse blow to the side of a pile are analyzed. Tests were conducted using both laboratory models and field timbers. Dispersive data were analyzed by both the Fourier transform method and the short kernel method (4–10). It is hoped that the findings of this work will lead to an inexpensive method for evaluating a timber pile’s condition. Such evaluations could then lead to early identification of questionable piles, less frequent replacements, and the avoidance of possible dangerous structural failures. USING DISPERSIVE WAVES TO EVALUATE PILE CONDITION The wave speeds of a dispersive signal depend on a pile’s material properties and geometry. A change in its properties, as may be caused by deterioration or changing cross-sectional area, leads to different wave speeds for the different frequencies in a signal. Whereas stress waves propagating through undamaged timber will have characteristic wave intensities and wave velocities, damaged timber cannot support stress wave propagation in the same manner. For this reason, wave speeds of a dispersive signal can be used to determine a pile’s condition. Fortunately, the dispersive surface waves and transverse waves created by a strike (thump) on the side of a pile will contain most of the energy put into the signal and will be the easiest to record and study. The equipment used in the dispersive wave propagation test included a digital oscilloscope, accelerometers, power supplies, tools for creating the signals, and a laptop computer for storing the data. Field test setup is shown in Figure 1. Chen and Kim 113 each hole was 6 mm with 13 mm equal spacing in the vertical and horizontal directions. Field Timber Piles Nine timber piles, located in Kure Beach Fishing Pier, North Carolina, were tested for the condition study. Table 1 shows the field records of the tested piles. Among these piles, seven are installed piles with varying conditions, and the remaining two are uninstalled piles. Of the uninstalled piles, one is an uninstalled good pile and the other is a damaged pile that had been installed in 1984 and extracted in November 1993. This extracted pile had been damaged severely near the ground line by shipworms. Field investigation indicates that the extracted pile does not have enough chromated copper arsenate (CCA) treatment. The required CCA treatment, according to the American Wood Preserver’s Association standard, is 40.0 kg/m 3 (2.5 lb/ft3) for timber piles that are used along the coasts of the United States. FIGURE 1 Field test setup. MATHEMATICAL BASIS FOR EVALUATING PILE CONDITION The available digital signal processing tools used in this research for wave propagation analysis are the Fourier transform (FT) method (11–13) and the short kernel method (8–10). SPECIMENS Laboratory Specimen—Acrylic Cylinders Internal deterioration in a timber pile cannot be easily detected under visual inspections or sounding tests. Three hollow acrylic cylinders with different wall thicknesses (3.2, 4.8, and 6.4 mm) were used to simulate different levels of internal deterioration (hollowness). In addition, a solid cylinder was used to simulate a sound pile. All cylinders have the same length, of 1.8 m, and the same outer diameter, of 38 mm. Their percentage of hollowness (H) was defined by Equation 1 for use when graphing the results. 2 i 2 0 A d H (%) − (100) i = (100) A0 d (1) where Ai 5 cross-sectional area of the hollow portion of each cylinder, di 5 diameter of the hollow portion of each cylinder, A0 5 entire cross-sectional area of each cylinder, and d0 5 diameter of each cylinder. FT Method Information extracted from the Fourier transform are wave speeds and dispersion field. A dispersion field presents the relationship between wave speeds and wavelengths. Wave speed at a particular frequency is computed by ∆φ C = (GL) ⋅ T 2π T = 1 f Df 5 (f2 2 f1) 1 (n ? 2p) A Douglas fir timber post, 100 mm by 100 mm by 5.5 m, was used to simulate damaged timber. Transverse holes were drilled in the middle of the post to simulate the damage created by marine borers. The damage was made by drilling a total of 185 holes in the middle of the wood in an array of 5 rows by 37 columns. The diameter of (3) and the corresponding wave length is computed by λ = C⋅T = C⋅ 1 f where Laboratory Specimen—Timber Post (2) C 5 wave speed, GL 5 physical distance of two accelerometers (gauges), T 5 period, f 5 frequency, Df 5 phase angle difference in two signals, f1 5 phase angle obtained from the Gauge 1 signal, f2 5 phase angle obtained from the Gauge 2 signal, (4) 114 TRANSPORTATION RESEARCH RECORD 1546 TABLE 1 Field Records of Tested Piles n 5 number of 2p jumps of the phase angle traveling from Gauge 1 to Gauge 2 at a particular frequency, and l 5 wavelength. Short Kernel Method The short kernel method (SKM), a digital signal–processing algorithm developed by Douglas (5–7) at North Carolina State University, can identify the entrance of individual phases. Information from the SKM can be used to compute wave speeds and to construct the dispersion field. The SKM is based on the cross-correlation procedure described by Bendat and Piersol (14). Mathematically, a single value of the SKM at a particular frequency can be stated as follows: n −1 Sk ( j∆t ) = ∑ x[( j + i )∆t ] fk [i∆t ] j = 0, 1, 2, . . . , N − 1 (5) i=0 where Sk fk x Dt N n 5 SKM output at an assigned kth frequency, 5 SKM kernel, 5 real time record, 5 time step, 5 number of data points in x, 5 number of data points in fk. method using different frequencies. Figure 3 shows the SKM dispersion field indicating that each cylinder possessed a different dispersion curve. Figure 4 presents the relationship between percentage of hollowness and SKM wave speed at a selected frequency shown in Figure 3. On the basis of the pattern in the data, a heuristic equation could be established in the following form: H 5 A ? C p2 1 B ? Cp 1 D where H 5 percentage of hollowness; Cp 5 SKM wave speed; and A, B, and D 5 real scalars determined by a curve-fitting procedure. The percentage of hollowness can also be defined as the ratio of the volume of the hollow portion to the total volume of each cylinder. This definition is useful in field tests because of the existing inconsistent hollowness along a pile’s longitudinal axis. However, the models and the equations developed herein are valid for both the area- and the volume-oriented environments. A detailed description of how to perform SKM transformation, calculate wave speed, and calculate overall length by SKM can be found elsewhere (6–10). LABORATORY TEST RESULTS Hollowness Assessment—The Acrylic Cylinders Each acrylic cylinder was tested in the test configuration shown in Figure 2, and the wave speeds were calculated by the SKM (6) FIGURE 2 Test setup—the acrylic cylinders. Chen and Kim 115 phase angle difference (Df) between the first and the second accelerometer locations caused by the Fourier transform’s inability to determine a unique phase angle above 2p radians (9,10). It may be attempting to display a different limb of the dispersion field. FIELD TEST RESULTS FIGURE 3 SKM dispersion field—the acrylic cylinders. Damage Detection—5.5-m Timber Post Figure 5 shows the bending wave test setup, and Figures 6 and 7 show the SKM and FT dispersion fields, respectively. In Figure 5, Tests 1 and 2 provide data that were collected from the undamaged and the damaged wood, respectively. In both Figures 6 and 7, the solid line is the dispersion curve obtained from the undamaged section, and the dashed line is the dispersion curve from the damaged section. Both analysis methods indicated the same pattern: a given frequency’s wave speed is higher when traveling through the undamaged wood compared with the damaged wood. It is emphasized that in the FT dispersion field, there is a region of dramatic variation in the vicinity of 4 kHz (see Figure 7). This phenomenon is due to the uncertain Fourier FIGURE 4 Two types of analyses were carried out for the installed timber pile condition assessment. The fist type is called overall structural integrity analysis. The idea is that if the created wave can travel down to a pile’s tip (toe) and back to the pile’s head again, then the time record will have the first pass and the return pass of the same wave. The variation of the wave speeds between the first and the return pass may be able to disclose the structural integrity of the pile; more specifically, the overall structural integrity in the region between the accelerometer and the pile’s tip. In field tests, two accelerometers were spaced at a known distance with the thumps made above the accelerometer locations. Heavier hammers were used to create low-frequency signals. These tests provided data for the computations of overall condition of the pile. The second type of analysis is called local condition analysis. Overall structural integrity analysis can indicate the average condition of the pile that has been tested. However, the failure of piles is quite frequently governed by local conditions that cannot be measured accurately using the overall condition analysis. Local condition analysis is needed to detect severe local damage. In field tests, two accelerometers were spaced at a known distance and systematically moved down along the pile length. Light hammers were used to create high-frequency signals. This series of tests was used to determine whether internal flaws or deteriora- Hollowness versus SKM wave speed—the acrylic cylinders. 116 TRANSPORTATION RESEARCH RECORD 1546 (Figure 8). The trends found from these two dispersion fields are reversed. The reason for the increase in return wave speed for the good pile is not clearly understood. One possible reason is the effect of the much smaller amplitude in the return wave on the strain level dependent modulus of materials. That is, the smaller strain levels in the return wave may have resulted in a higher modulus and therefore a higher wave speed. Overall Structural Integrity Analysis—Installed Piles FIGURE 5 Test setup—the timber post. tion existed by comparing the wave speeds at different testing locations. Figures 10 and 11 show the SKM dispersion fields that were computed from the two time records of installed piles. Pile 7 in Figure 10 was in very poor condition at the time of testing and extracted several months after testing. Inspection after the extraction revealed severe hourglass with a 152-mm-diameter internal hole. Figure 10 displays 8 to 16 percent reduction in the return wave speed compared with the first pass wave speed. On the contrary, Pile 9, which has been identified as a sound pile, has a wave speed increase of 0.1 to 5 percent (Figure 11). The trends in dispersion fields found from these two piles are reversed, which coincides with the results obtained from the uninstalled timber piles. Overall Structural Integrity Analysis—Uninstalled Piles Local Condition Analysis—Installed Piles A comparison was made between a new uninstalled good pile (Pile 3) and an extracted damaged pile (Pile 4). Test results indicate that a severely damaged pile will have a significant wave speed reduction in the return pass. Figures 8 and 9 show the SKM dispersion fields that were computed from both the good and the damaged piles’ time records, respectively. The damaged pile has a wave speed reduction in the range of 28 to 35 percent in the selected frequencies (Figure 9). On the contrary, the good pile has a wave speed increase in the range of 0.1 to 21 percent in the selected frequencies The local condition assessment was performed on a good pile (Pile 1) and a damaged pile (Pile 2) by moving the test area systematically from the head to the tip. The results are presented by the SKM dispersion fields in Figures 12 and 13. In Pile 1, the wave speed comparison between the highest and the lowest test locations indicates a wave speed reduction of 9 to 14 percent (Figure 12). Pile 2, with an hourglass damage in the lowest test location, indicates a wave speed reduction of 33 to 36 percent between the highest and the lowest locations (Figure 13). Interpre- FIGURE 6 SKM dispersion field—the timber post. Chen and Kim 117 FIGURE 7 FT dispersion field—the timber post. tation of the results from the installed piles is more complicated because the lower portion of the pile is saturated with water, which also contributed to the wave speed reduction. Assuming that most of the wave speed reduction in Pile 1 is from the effect of water, the additional wave speed reduction in Pile 2 can be attributed to the effect of damage. The reduction of wave speeds in two different test areas is mainly due to the change in diameter, water content, and damage. To accurately represent the condition of the tested pile due to damage, the effects of changing diameter and water content need to be corrected. A correction procedure has been developed by Chen (10) for adjusting wave speeds in sections with FIGURE 8 pile. varying diameters to the wave speed of a reference diameter. Chen also indicates, in the laboratory, that significant wave speed loss occurs from the presence of water. The procedure for correcting the water effect, however, is more involved and is currently under investigation by the authors. CONCLUSIONS Dispersive wave propagation tests were found to be a promising means of evaluating the degree of hollowness by the way in which SKM dispersion field, first pass versus return pass, Pile 3—good 118 TRANSPORTATION RESEARCH RECORD 1546 FIGURE 9 SKM dispersion field, first pass versus return pass, Pile 4—bad pile. their individual dispersion curves shift. A heuristic equation relating the degree of hollowness and wave speed was derived. Simulated borer damage to a 5.5-m timber post was successfully detected by the SKM dispersion field and the FT dispersion field, with the SKM analysis demonstrating more consistent results. Two different experimental approaches were found to be promising for field condition assessment: (a) overall condition analysis by comparing wave speeds between the first pass and the return pass and (b) local condition analysis by comparing wave speeds at different locations. Test results, from both uninstalled and installed FIGURE 10 poor piles, indicated that the return wave speed decreased significantly compared with the first wave speed. The second approach revealed that the wave speed determined from the areas with more damage was lower than that of sound areas. The condition of the embedded portion, however, can be detected only by an overall condition analysis approach. The determination of the degree of damage of the installed pile requires a correlation between the laboratory-developed heuristic equations and the field-collected data, which is currently under investigation by the authors. SKM dispersion field, first pass versus return pass, Pile 7—bad pile. Chen and Kim 119 FIGURE 11 pile. SKM dispersion field, first pass versus return pass, Pile 9—good ACKNOWLEDGMENTS This paper presents results of two research projects conducted in the Civil Engineering Department of North Carolina State University. Principal funding was provided by the Sea Grant program of the National Oceanic and Atmospheric Administration. The authors FIGURE 12 good pile. acknowledge the important contribution of Robert A. Douglas of North Carolina State University in providing many helpful suggestions and discussions. The authors also acknowledge the valuable assistance of J. Darrin Holt, FDH Inc.; Spencer M. Rogers, North Carolina Sea Grant Marine Advisory Services; and Mike Robertson, Kure Beach Fishing Pier. SKM dispersion field, wave speeds at different locations, Pile 1— 120 TRANSPORTATION RESEARCH RECORD 1546 FIGURE 13 bad pile. SKM dispersion field, wave speeds at different locations, Pile 2— REFERENCES 1. Goble, G. Pile Driving Analyzer Manual. Pile Dynamics, Inc., 1987. 2. Antony, R. W., G. E. Phillips, and J. Bodig. Nondestructive Strength Assessment of In-Situ Timber Piles. SBIR Phase I Final Report. Engineering Data Management, Inc., Ft. Collins, Colo., Sept. 1989. 3. Aggour, M. S., A. Hachichi, and M. A. Mayer. Nondestructive Evaluation of Timber Bridge Piles. In Evaluation and Upgrading of Wood Structures: Case Studies, American Society of Civil Engineering, 1986, pp. 82–95. 4. Douglas, R. A., and G. L. Eller. Nondestructive Pavement Testing by Wave Propagation: Advanced Methods of Analysis and Parameter Management. In Transportation Research Record 1070, TRB, National Research Council, Washington, D.C., 1986, pp. 53–62. 5. Douglas, R. A., J. L. Eddy, and H. E. Wahls. On Transforms and the Dispersion Computations Used for Evaluating Layer Properties. Nondestructive Testing of Pavements and Back Calculation of Moduli. ASTM STP 1062 (A. J. Bush III and G. Y. Baladi, eds.), ASTM, Philadelphia, Pa., 1989, pp. 612–627. 6. Douglas, R. A., and J. D. Holt. Determining Length of Installed Timber Pilings by Dispersive Wave Propagation Methods. Final Report, Research Project 23241-92-2. North Carolina Department of Transportation, FHWA, U.S. Department of Transportation, June 1993. 7. Holt, J. D., S. Chen, and R. A. Douglas. Determining Lengths of Installed Timber Piles by Dispersive Wave Propagation. In Trans- 8. 9. 10. 11. 12. 13. 14. portation Research Record 1447, TRB, National Research Council, Washington, D.C., 1994, pp. 110–115. Holt, J. D., and R. A. Douglas. A Field Test Procedure for Finding the Overall Lengths of Installed Timber Piles by Dispersive Wave Propagation Methods. Institute for Transportation Research and Education, University of North Carolina, March 1994. Holt, J. D. Comparing the Fourier Phase and Short Kernel Methods for Finding the Overall Lengths of Installed Timber Piles. Ph.D. dissertation. North Carolina State University, 1994. Chen, S. Condition Assessment of Installed Timber Piles and Comparison of Three Transform Techniques Using Stress Wave Method. Ph.D. dissertation. North Carolina State University, 1995. Bracewell, R. N. The Fourier Transform and Its Applications. McGrawHill, New York, 1986. Papoulis, A. The Fourier Integral and Its Applications. McGraw-Hill, New York, 1962. Ramirez, R. W. The FFT Fundamental and Concepts. Prentice-Hall, Inc., 1985. Bendat, J. S., and A. G. Piersol. Engineering Applications of Correlation and Spectral Analysis. John Wiley and Sons, Inc., New York, 1980. Publication of this paper sponsored by Committee on Foundations of Bridges and Other Structures.