Numerical simulation and experimental study of Parallel Seismic test for piles

(1)

Numerical simulation and experimental study

of Parallel Seismic test for piles

Shu-Tao Liao

a,*

, Jian-Hua Tong

b

, Cheng-Hao Chen

c

, Tsung-Tsong Wu

d

a

Department of Civil Engineering, Chung Hua University, Hsinchu 30067, Taiwan, ROC b

Department of Computer Science and Information Engineering, Hungkung University, Taichung 400, Taiwan, ROC c_{Der-Bao Engineering Consultant, Inc., Tainan 700, Taiwan, ROC}

d_{Institute of Applied Mechanics, National Taiwan University, Taipei 106, Taiwan, ROC} Received 28 December 2004; received in revised form 19 March 2005

Available online 13 May 2005

Abstract

The theoretical capabilities of the Parallel Seismic (PS) test for determining the length of piles and characterizing possible defects are investigated in this paper. In deriving the theory, a correction factor is proposed in this paper to enhance the accuracy of the prediction. An axisymmetric finite element (FE) model was developed to carry out a series of parametric studies which included the effects of the pile length, the pile-to-borehole distance and the stiffness of sur-rounding soil. A miniature experiment using an aluminum bar embedded in epoxy prism was also designed and tested in the lab to verify the theory and to demonstrate the use of the correction factor. The results from the experiment and the numerical FE model were also compared to expose the potential of this nondestructive method to in situ application. The model was further modified to simulate the PS test on defective piles with axisymmetric necks and bulges. It can be concluded that the geometric configuration of a pile and the possible significant defect can be characterized with this nondestructive test.

Keywords: Nondestructive test; Pile; Parallel Seismic method; Integrity test

1. Introduction

The importance of assessing the construction quality of a pile foundation has long been recognized. To achieve this goal, various kinds of destructive and nondestructive tests were developed. In this aspect,

*

Corresponding author. Tel.: +886 35186757; fax: +886 35372188. E-mail address:shutao@chu.edu.tw(S.-T. Liao).

(2)

nondestructive testing (NDT) was considered to possess high potential for application because it is featured as cost-effective, damage-free, time-saving, and thus can cover the whole population to avoid the disadvan-tages of sampling (Baker et al., 1991). The primary goal of carrying out a nondestructive test on a pile is the acquisition of the geometric configuration of the pile. This information, describing the detailed geometry of the pile and the embedded defects if they exist, can first be used to evaluate the integrity of the pile and then be served as the input of a subsequent nonlinear finite element analysis to estimate the bearing capacity (or the ultimate strength) of the pile. This philosophy exposes the vast space that NDT can apply in the future. However, all these potential applications rely on the success of NDT in obtaining the detailed geometry of the intact or defective piles (Liao and Roesset, 1997a). Many researchers have been working on the NDT of piles. Most of the tests can be classified into two categories, i.e., (1) reflection and (2) direct transmission methods. Among them some commonly used examples are Impulse Response (IR) (Davis, 2003; Finno and Gassman, 1998) or Transient Dynamic Response (TDR) (D5882 of ASTM, 2004; Davis and Dunn, 1974), Sonic Echo (SE) (Rausche et al., 1991), Impedance Log (IL) (Paquet, 1991; Rix et al., 1993) and Impact Echo (IE) (Lin et al., 1997; Sadri and Afshin, 2003) tests as categorized in reflection methods and Cross-hole Sonic Logging (CSL) and Parallel Seismic (PS) (Olson et al., 1995) tests as categorized in direct transmission methods. In reflection methods, the force to introduce a disturbance to a pile and the receiver to collect the response of the pile are basically applied at the pile head, whereas in direct trans-mission methods the receiver is placed at intervals in a tube or borehole which must be cored or drilled be-fore testing. Therebe-fore reflection methods are in general faster and more cost-effective. However, direct transmission methods can obtain the information in a more direct way and they can be applied to piles with much larger length.

Among the reﬂection methods,Lin et al. (1997)used the IE tests to successfully detect the lengths of piles and the depths of the embedded necks. Using the ﬁnite element method,Liao and Roesset (1997b) devel-oped one, two, and three-dimensional models to simulate the IR test for intact piles. A series of parametric studies were carried out to demonstrate the capabilities of this method in determining the lengths and cross sectional areas of intact piles under various kinds of soil conditions. The numerical models of IR tests were later extended to study piles with axisymmetric irregularities such as necks and bulges. The results indicated that by analyzing the mechanical admittance (or mobility) curves, the depth and length of the necks (or bulges) can be determined in good precision. However, the methodology to determine the widths of the defects seems to remain pending. The results of this paper primarily indicated that this goal can be achieved by the PS test, which has not been thoroughly studied due to limited application for its relatively high cost of operation. However, this method plays an important role in the second line for settling down the dispute on suspicious yet critical piles. This paper will also propose a correction factor to be incorporated in the traditional procedure of this method to enhance the accuracy of the estimation on the length of piles.

The determination of the substantial length of a pile exhibits a high value in practical application, par-ticularly when evaluating the integrity of existing foundations such as piles with cap. This point can be illus-trated with an engineering problem arose after the disastrous earthquake, the Ji-Ji Earthquake, struck central Taiwan on September 21, 1999. A lot of structures were damaged or collapsed in the attack. The post-earthquake evaluation on the integrity of the structures played an important role in the success of the later retrofitting work. In this aspect, the integrity evaluation of the existing foundations is among the most difficult. Shown in Fig. 1is the in-built bridge system of a superhighway near central Taiwan. To the time when the earthquake struck, the supporting system of the bridge had just been finished as shown in the picture. There was no problem for the engineers to make the decision to replace the super-structure of the system as the huge damages being discovered in the inspection. However, the difficult sit-uation was to determine how seriously the foundations were damaged and whether the supporting piles should be replaced. To settle down this problem, a painful decision was made to carry out the coring test on some piles. The sampling rate is 28 out of 350 piles in the region. Most of the cored cylinders are con-crete until a depth of 13 m and soil beyond the depth. The designed length of the piles is 25 m and this result

(3)

indicated that the piles were twisted seriously to break at about the mid-height, as shown inFig. 2. As a matter of fact, this conclusion can be reached in a much safer and economic way if the substantial length of the piles can be obtained with nondestructive testing methods. The goal of this kind of applications may be summarized to the determination of the substantial length of single piles, or much more diﬃcult cases, Fig. 1. In-built bridge system of a superhighway in central Taiwan. (Photo courtesy of Taiwan Area National Expressway Engineering Bureau.)

(4)

piles with cap. To reach this goal, all the tools such as the Parallel Seismic method should attract deeper study than before.

2. Regular procedure of Parallel Seismic test

In Parallel Seismic tests, the top or side of the pile is struck by an instrumented hammer and the prop-agated stress wave is sensed by a hydrophone placed in a cased, water-filled borehole drilled next to the pile (Olson et al., 1995). As schematically shown inFig. 3, the receivers are placed at different depths along the borehole to record the time histories of the motion caused by the induced stress waves. As an example,Fig. 4shows the velocity time history sensed by a receiver. The time of the first arrival of the P wave, defined as the earliest time when the response curve starts to fluctuate, is then identified from the waveform. For each receiver this arrival time, as marked with ‘‘T0’’ inFig. 4, is plotted versus the depth of the corresponding

receiver on an x–y plot, as shown inFig. 5(Liao et al., 1998). Usually the data points can be grouped into two families and approximately ﬁtted by two straight lines. The y-coordinate, i.e, the depth, of their inter-section point indicates the length of the pile. The theoretical background of this regular procedure will be explored in the next section to verify the validity of this method. In doing so, a correction factor to enhance the accuracy of this method can be concluded.

3. Theoretical derivation and correction factor

Consider a pile partially or fully embedded in soil, as shown inFig. 6. The pile system consists of a col-umn, which is assumed to be made of concrete, and the soil. The pile has a circular cross-sectional area with a radius r. The length of the pile above the ground is denoted by LAand the total length is L. The distance

between the lateral surface of the pile and the receiver borehole is denoted by D. An external impact is

(5)

applied at the center of the top surface of the pile. The contact area is relatively small compared to the cross sectional area of the pile so that the impact may be treated as a point source. The receivers in the borehole can be categorized into two groups based on whether the receiver is located above or below the level of the pile bottom. Let H1denote the distance between the level of the pile head and a receiver B1which belongs to

the ﬁrst group. Using the ray path theory a stress wave is assumed to start from the center of the loaded area O, passing through a point A1, which is located at the lateral surface of the pile, and arriving at the

receiver B1. Let S1denote the travel length of this path O! A1! B1. Let V1and V2denote the

propaga-tion velocities of the waves in pile and soil, respectively. The refracpropaga-tion at the interface of the pile and the surrounding soil should obey SnellÕs law, which states that sin h1/V1= sin h2/V2where h1and h2denote the

angles between the normal vector of the interface and the ray paths in the two media. Then,

S1¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðH1 X Þ 2 þ r2 q þpffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiD2þ X2_{¼ ðH} 1 X Þ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1þ r H1 X 2 s þpffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiD2þ X2 _ð1Þ 0.000 0.005 0.010 0.015 0.020 0.025 0.030 Time (sec) -1.5E-6 -1.0E-6 -5.0E-7 0.0E+0 5.0E-7 1.0E-6 1.5E-6 Ve loc ity (m /se c) T0

Fig. 4. Identiﬁcation of ﬁrst arrival time of stress wave from velocity time history.

0.000 0.005 0.010 0.015 0.020 0.025 Time (sec) 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20.0 Dept h (m) L1 L2 Z

(6)

where X ¼ D tan h2 ð2Þ h2¼ sin1 sin h1 V2 V1 ð3Þ In the case where (H1 X) r, Eq.(1) becomes

S1 ðH1 X Þ þ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi D2þ X2 p ¼ ðH1 X Þ þ F ð4Þ where F ¼pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiD2_{þ X}2 _ð5Þ

Thus the total travel time T1for the P wave to propagate through the path S1and then arrive at B1would

be

(7)

T1¼ ðH1 X Þ V1 þ F V2 ð6Þ Referring to the arrival time versus depth plot as shown inFig. 5, the equation of the straight line L1

which is selected to represent the data points of the ﬁrst group can be given as follows: L1: T ¼ ðH X Þ V1 þ F V2 ð7Þ In the same manner, let H3denote the distance between the level of the pile head and a receiver B3which is

placed at a large depth and belongs to the second group. The travel distance S3of the path O! An! B3

would be S3¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi L2þ r2 p þ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi D2þ ðH3 LÞ 2 q ¼ L ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1þ r L 2 r þ ðH3 LÞ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1þ D H3 L 2 s ð8Þ In the general cases where (H3 L) D and L r, S3can be approximated by

S3 L þ ðH3 LÞ ð9Þ

Therefore the travel time T3associated with S3would be

T3¼

L V1

þðH3 LÞ V2

The equation of the straight line L2for the second group inFig. 5can thus be given by

L2: T ¼ L V1 þðH LÞ V2 ð10Þ Let HIdenote the depth of the receiver associated with the intersection point of the two lines L1and L2,

solving Eqs.(7) and (10) may reach HI ¼ L þ V1 V1 V2 F V2 V1 V2 X ð11Þ That is, L¼ HI R ð12Þ where R¼ V1 V1 V2 F V2 V1 V2 X ð13Þ

is deﬁned as the correction factor in this paper. In comparison with the conventional strategy which approximates the length of the pile with HI, it will be shown later that the accuracy in predicting the true

length L of the pile can be enhanced when the correction factor R is taken into account.

4. Finite element model

An axisymmetric finite element model was used to model the pile and the surrounding soil as shown in Fig. 7. The complete domain of the problem can be obtained by rotating this figure about the centerline of the pile. The four-noded, quadrilateral, and isoparametric (Q4) element was selected to use in this model. After discretization, the equations of motion in matrix form for finite analysis can be obtained. The central difference scheme was then adopted to carry out the integration for the equations of motion to obtain the

(8)

nodal displacements at each time step (Liao and Roesset, 1997b). It should be noted that the central dif-ference method is conditionally stable so that for stability, the time step Dt should be smaller than a critical value, Dtcr, which is controlled by the maximum natural frequency (or minimum natural period) of the

ﬁnite elements for the system. A program was developed to implement the direct iteration scheme to deter-mine the value of Dtcr.

The loading pressure P(t) used in this paper was deﬁned as a function of time by

PðtÞ ¼ P0sin xt for 0 6 t 6 Td ð14Þ

and P(t) = 0 for t > Tdwhere x¼ p=Td. This represents a sinusoidal impact of half a cycle duration with a

period of 2Td. In this paper, Td= 1.5 ms = 1.5· 103s and P0= 15,570 N/m2. The radius of the loaded

area is set at rL= 0.0254 m.

5. Experimental veriﬁcation

To verify the capability of the PS method in determining the length of the pile and to evaluate the potential of the correction factor in enhancing the accuracy of estimation, a miniature experiment was

(9)

carried out in the laboratory. The experimental results will be also compared to those obtained with the ﬁnite element simulation so that the numerical model can be veriﬁed for later use in parametric study. 5.1. Experimental results

Consider an aluminum cylinder embedded in an epoxy bulk, as shown inFig. 8. The height of the epoxy prism is 38 cm. The dimension of the square cross section of this prism is 8 cm by 8 cm. The radius and the total length of the aluminum bar are 1.3 cm and 31 cm, respectively. The length of the bar above the top level of the epoxy prism is 4 cm. The axis of the bar coincides with the center line of the prism. The YoungÕs modulus EAl, PoissonÕs ratio mAl, and mass density qAlof the aluminum bar are provided inTable 1. The

propagation velocities of the P (primary) and S (shear) waves, denoted by VAl-Pand VAl-S, are also listed in

the table. In the table, the theoretical velocities of P and S waves propagating in a material were calculated by VP¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð1 mÞE ð1 þ mÞð1 2mÞq s ð15Þ VS¼ ffiffiffiffi G q s ¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi E 2ð1 þ mÞq s ð16Þ

(10)

where G is the shear modulus of the material. For comparison purpose, the YoungÕs modulus EEp, PoissonÕs ratio mEp, and mass density qEpof the epoxy bulk, as well as the P and S wave velocities, VEp-Pand VEp-S, are also provided in the table. It can be seen that the ratio of the P wave velocities in aluminum and epoxy is about 2.6, a number much lower than that of those in concrete (Wu and Fang, 1997) and normal soil, which is about 9. This condition places the outcome on a conservative side since now it is more diﬃcult to deter-mine the length of the aluminum bar in the laboratory condition than the length of a concrete pile in prac-tical site with the PS method.

An impact was introduced by triggering the falling of a small steel ball at the center of the top surface of the aluminum bar. In this study, a laboratory-built vertically polarized conical transducer (Tong et al., 2002) was utilized to measure the vertical component of the surface displacement signals. The received volt-age signal from the conical transducer was amplified by a preamplifier and recorded by a 100 MHz digital oscilloscope (LeCroy 9314L) (Tong et al., 2001). It is noted that the preamplifier must be embedded in the transducer assembly to minimize the electric noise. The first receiver was placed 10 cm down from the top level of the aluminum bar at the side surface of the epoxy prism, as shown inFig. 8. Then the receiver was moved down at the side surface of the epoxy prism at an interval of 5 cm until 40 cm deep down from the top level of the aluminum bar. The displacement amplitude recorded at the first receiver, which is placed at 10 cm down from the top level of the aluminum bar, is presented inFig. 9as a function of time. The next step in the PS method is to identify the arrival time of the first P wave from this response curve. However, this work can be done more easily and accurately from velocity time history because it is sharper. The direct way to get the velocity is to obtain the time derivative of the displacement. However, the high-frequency fluctuations resulted from in situ noises and the digitalization of the data may cause a problem in obtaining Table 1

Material properties of aluminum bar and epoxy prism and the corresponding stress wave velocities

Aluminum bar Epoxy prism

YoungÕs modulus (N/m2₎ _7.24 · 1010 _3.94 · 109 PoissonÕs ratio 0.35 0.38 Mass density (kg/m3₎ ₂₈₀₀ ₁₁₆₀ P wave velocity (m/s) 6390 2500 S wave velocity (m/s) 3100 1110

0.00E+0 3.00E-5 6.00E-5 9.00E-5 1.20E-4 Time (sec)

Relative amplitude of displacement

(11)

this derivative. This phenomenon can be seen as the circled part in the curve ofFig. 9is blown up and pre-sented by a dashed line inFig. 10. Using the Moving Average (MA) method to manipulate the original displacement, a much smooth curve can be obtained as shown in the figure with a solid line. Based on this smooth displacement, the corresponding velocity time history was obtained and presented by the dashed curve inFig. 11. Similarly, the velocity recorded at the side surface of the epoxy prism at the distances of 20 cm down from the top level of the aluminum bar is plotted as the dashed curves inFig. 12. The next step in the PS method is to identify the arrival times of the first P waves in each figure and then plot the depth-to-arrival-time curve for determining the length of the aluminum bar. This result is presented in Fig. 13. The arrival time of the first P wave is identified by comparing the signal and an undisturbed datum line (Tong et al., 2002). InFig. 13, data points A and B are then selected to form the first straight line and points C and D, the second. The intersection point of these two lines indicates that the length of the bar is 33.6 cm, as predicted with the regular procedure of the PS method. The error is 8.4% since the true length of

5.00E-5 5.20E-5 5.40E-5 5.60E-5 5.80E-5 6.00E-5 Time (sec) ¥-·Æ³B²z«e ¥-·Æ³B²z«á before after Rela ti ve am p li tude of di sp lac eme nt

Fig. 10. Displacement amplitude before and after the processing of moving average scheme.

0.00E+0 3.00E-5 6.00E-5 9.00E-5 1.20E-4 Time (sec) ¼Æ-È¼ÒÀÀ ¹êÅé¸ÕÅç Numerical Experimental Relative amplitude of velocity

(12)

the bar is 31 cm. However as the correction factor, which is calculated to be R = 3.9 cm, is taken into ac-count, the predicted length is modiﬁed to be 29.7 cm, which is only 4.2% in error.

5.2. Comparison of numerical and experimental results

To cultivate a numerical model for the use of parametric study for this research, the finite element model developed and introduced in this paper was used to simulate the PS test carried out in the lab as mentioned above. The system for the numerical model to simulate is shown inFig. 14. It is noted that the epoxy prism shown inFig. 8as for the experiment is now replaced by an epoxy cylinder with a diameter of 8 cm because the finite element model to be used is axisymmetric. This modification, which resulted from the limitation of the developed finite element model, surely will affect the comparableness of the results with those obtained from the experiment. However, the effect is expected to be small because of this small discrepancy on the

0.00E+0 3.00E-5 6.00E-5 9.00E-5 1.20E-4

Time (sec) ¼Æ-È¼ÒÀÀ

¹êÅé¸ÕÅç Numerical Experimental

Relative amplitude of velocity

Fig. 12. Comparison of numerical and experimental results in velocity obtained at 20 cm from bar head.

D C B A

0.00E+0 3.00E-5 6.00E-5 9.00E-5 1.20E-4

Time (sec) 10.0 15.0 20.0 25.0 30.0 35.0 40.0 De p th (cm )

(13)

boundary condition. Furthermore, what matter in a PS test are the arrival times of the ﬁrst P waves at the receivers. The simulation results of the ﬁnite element model are then plotted together with those from the experiment. The velocity time histories at the receivers located at distances of 10 and 20 cm down from

Fig. 14. Geometric conﬁguration of numerical model for simulating the experiment.

0.00E+0 3.00E-5 6.00E-5 9.00E-5 1.20E-4

Time (sec) 10.0 15.0 20.0 25.0 30.0 35.0 40.0 Depth (cm)

(14)

the top level of the aluminum bar are plotted with the solid curves inFigs. 11 and 12, respectively. It is very easy to note that the results from the numerical model match well with those from the experiment at the initial stage and then the accordance deteriorates with time, as expected. The arrival times of the first P waves in each figure are then identified to plot the depth-versus-arrival-time curve, as shown inFig. 15. Two straight lines can easily be defined to fit these data points and their intersection point indicates that the length of the bar is 34.2 cm, using the regular procedure of the PS method. The error is 10.3% since the true length of the bar is 31 cm. However as the correction factor is taken into account, the predicted length is modified to be 30.3 cm, which is only 2.3% in error. ComparingFigs. 15 and 13reveals that form-ing the two straight lines is even more straightforward for the numerical than experimental results. And also the prediction error is even smaller.

6. Parametric studies on intact piles

Once the validity of the developed finite element model had been verified with the experimental results, a series of parametric studies were then carried out to study the capability of the PS tests and the proposed correction factor. The parameters to be varied include pile length, pile to borehole distance, and soil stiff-ness. Referring toFig. 6, consider an intact pile with a total length of L = 12 m. The pile has a circular cross-sectional area with a radius r = 0.5 m. The lengths of the pile above and below ground are 1 and 11 m, respectively. The material properties of the soil and the concrete pile are given inTable 2, in which Ec, mc, qcand Es, ms, qsare the YoungÕs modulus, PoissonÕs ratio, and mass density of the concrete pile and

surrounding soil. The theoretical wave velocities of P and S waves in the concrete and soil are also calcu-lated with Eqs.(15) and (16)and provided in the table. The pile-to-borehole distance D is set at 1 m. The receiver is lowered into the borehole and takes the measurements in velocity at an interval of 1 meter until a depth of 19 m from the ground (or 20 m from the top level of the pile).

6.1. Eﬀect of correction factor

Consider the example mentioned above. The measurements of the particle velocity in the vertical direc-tion were made at an interval of 1 m down along the receiver borehole. Using the finite element model, the velocity response at the depths of 12 m from the pile head is presented for example in Fig. 16. The first arrivals of the P waves are selected directly from the finite element results and marked with ‘‘PP’’ in the figures. The approximate times of arrival of the waves which travel with the P wave in the pile and then the S wave in the soil are marked with ‘‘PS’’ in the figures for reference. The depths versus the arrival times are plotted inFig. 17. The two data points (marked with ‘‘A’’ and ‘‘B’’ in the figure) associated with 11 and 12 m depth are chosen to define the first straight line and points 19 and 20 (marked with ‘‘C’’ and ‘‘D’’) are chosen to define the second. After calculation, the depth of the intersection of these two straight lines is HI= 12.95 m, based on the conventional procedure of the Parallel Seismic test. The error is 7.9%. Using

Table 2

Concrete and soil properties and corresponding wave velocities

Concrete Soil Unit

Ec 3.31· 10 10 Es 1.8· 10 8 N/m2 qc 2300 qs 1925 kg/m3 mc 0.2 ms 0.4 VP 4000 VP 450 m/s VS 2450 VS 183 m/s

(15)

Eqs.(3), (2), (5) and (13), the value of the correction factor for this case can be calculated to be 1.12 m. Thus the length of the pile is recovered to be

L¼ H R ¼ 12.95 1.12 ¼ 11.83 The error is now reduced from 7.9% to 1.4%. 6.2. Eﬀect of pile length

To examine the capability of the correction factor, the embedded length of the pile is ﬁrst decreased from 11 m to 9 and 10 m and then increased to 12 and 13 m so that the total length of the pile is 10 to 14 m. The results are presented in the depth versus arrival time plot as shown inFig. 18. The results before and after taking the correction factor into account are presented inTable 3. The error in predicting the total length of the pile with the conventional procedure decreases from 10.3% to 6.5% as the length of the pile increases. After taking the correction factor into account, the errors for all cases are within 1.5%.

0.000 0.005 0.010 0.015 0.020 0.025 0.030 Time (sec) -1.2E-6 -8.0E-7 -4.0E-7 0.0E+0 4.0E-7 8.0E-7 1.2E-6 Veloc ity (m/sec) PP _PS

Fig. 16. Velocity response of intact pile to the PS test obtained at 10 m from pile head.

0.000 0.005 0.010 0.015 0.020 0.025 Time (sec) 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20.0 Dep th (m ) L1 L2 A B C D

(16)

6.3. Eﬀect of pile-to-borehole distance

The distance between the receiver borehole and the lateral surface of the pile is now increased from D = 1 m to 1.5, 2.0, 2.5, and 3 m. The results are presented in Fig. 19 and summarized in Table 4. It can be seen that the error in predicting the total length of the pile increases as the distance between the bore-hole and the pile increases. This error increases from about 8% for D = 1 m to 20% for D = 3 m if conven-tional procedure is used. The errors are about 1.4–8.2% if the correction factors are taken into account. The trend that the error increases as D increases can be expected in deriving the correction factor where (H L) was assumed to be far greater than D.

6.4. Eﬀect of soil stiﬀness

To study the eﬀect of soil stiﬀness, the YoungÕs modulus of the soil Esis increased so that the P wave

velocity in the soil is increased from VP= 450 to 1350, 2250, 3150, and 4050 m/s. In other words, the ratio

of the P wave velocities in the pile and soil is decreased from 8.9 to 3.0, 1.8, 1.3, and 1.0. The results are presented inFig. 20and summarized inTable 5. The ﬁgure indicates that determining the intersection point becomes harder as the stiﬀness of the soil increases toward that of the pile. In the cases where the ratio of the P wave velocities in the pile and soil is less than 2, it is unpractical to predict the length of the pile with this method. As shown inTable 5, the error in predicting the length of the pile ranges is about 8–16% and 1–8% before and after using the correction factor.

0.000 0.005 0.010 0.015 0.020 0.025 Time (sec) 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20.0 Depth (m) L = 10 m L = 11 m L = 12 m L = 13 m L = 14 m

Fig. 18. Depth-versus-time plot of intact piles with lengths of 10, 11, 12, 13, and 14 m.

Table 3

Estimation and error of length prediction for piles with diﬀerent lengths Pile length, L (m) 10 11 12 13 14 Initial estimate, Z (m) 11.03 12.04 12.95 13.98 14.91 Error % 10.3 9.5 7.9 7.5 6.5 Correction factor, R (m) 1.12 1.12 1.12 1.12 1.12 Modiﬁed length, L (m) 9.91 10.92 11.83 12.86 13.79 Error % 0.9 0.7 1.4 1.1 1.5

(17)

0.000 0.005 0.010 0.015 0.020 0.025 Time (sec) 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20.0 Depth (m) D = 1.0 m D = 1.5 m D = 2.0 m D = 2.5 m D = 3.0 m

Fig. 19. Depth-versus-time plot of intact piles with pile-to-borehole distances of 1, 1.5, 2, 2.5, and 3 m.

Table 4

Estimation and error of length prediction for piles with diﬀerent pile-to-borehole distances Pile-to-borehole distance, D 1.0 1.5 2.0 2.5 3.0 Initial estimate, Z (m) 12.95 13.38 13.78 14.06 14.38 Error % 7.9 11.5 14.8 17.2 19.8 Correction factor, R (m) 1.12 1.68 2.24 2.79 3.36 Modiﬁed length, L (m) 11.83 11.70 11.54 11.27 11.02 Error % 1.4 2.5 3.8 6.1 8.2 0.000 0.005 0.010 0.015 0.020 0.025 Time (sec) 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 18.0 20.0 Depth (m) = 450 m/s = 1350 m/s = 2250 m/s = 3150 m/s = 4050 m/s

(18)

7. Numerical simulation on defective piles

To study the capability of the PS method in identifying and quantifying the defects embedded in piles, the ﬁnite element model was used to simulate the response of defective piles. Two kinds of defects were con-sidered. They are axisymmetric neck and bulge, which result in a reduction and expansion in the cross sec-tional area of a pile, respectively.

7.1. Pile with axisymmetric neck

Consider the intact pile discussed in previous section. Now a neck is introduced into the pile at a distance between 5 m and 8 m from the pile head. The original radius of the pile is 0.5 m. To study the eﬀect of a neck in the response of a pile to the PS test, consider that the radius of the pile in the range of the neck is reduced to Fr= 0.4, 0.3, 0.2, and 0.1 m, respectively. The depth versus arrival time curve of these PS tests is

presented inFig. 21. It can be seen in all defective cases that a neck results in a delay in the arrival time and thus the data points associated with the defective area deviate to the right from the original straight line. Furthermore, the deviation becomes larger as the neck-in becomes worse. The arrival times will go back to the original straight lines as the depth of the receivers goes deeper than the neck. In other words, the depth, the extent as well as the thickness of the neck can be characterized from the result of a PS test, at least in theoretical sense.

Table 5

Estimation and error of length prediction for pile systems with diﬀerent wave velocities in soil P wave velocity, VP 450 1350 2250 3150 4050 Initial estimate, Z (m) 12.95 13.35 13.56 13.93 12.91 Error % 7.94 11.22 12.96 16.10 7.58 Correction factor, R (m) 1.12 1.42 1.89 2.87 NA Modiﬁed length, L (m) 11.83 11.93 11.67 11.06 NA Error % 1.39 0.60 2.72 7.80 NA 0.000 0.002 0.004 0.006 0.008 0.010 Time (sec) 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 Depth (m) = 0.1 m = 0.2 m = 0.3 m = 0.4 m = 0.5 m

(19)

7.2. Pile with axisymmetric bulge

Next consider that the neck discussed above is replaced by a bulge. For comparison purpose, the radius of the pile in the range of the bulge is increased from the normal one, Fr= 0.5 m, to 0.6, 0.7, 0.8,

and 0.9 m, respectively. The depth versus arrival time curve is presented in Fig. 22. To the contrary, a bulge results in an early arrival of the stress wave and thus the data points associated with the bulge area deviate to the left from the original straight line. Furthermore, the deviation increases with the radius of the bulge. Therefore, it is feasible to identify the depth, the extent and the thickness of the bulge through the PS test.

8. Conclusions

The potential of the Parallel Seismic test in determining the length of the pile and locating a possible defect as a neck or bulge is evaluated in this paper. A correction factor is proposed to be incorporated into the regular procedure of the method in recovering the length of the pile. The results indicated that the accuracy in predicting the length of the pile can be effectively enhanced when the correction factor is taken into account. A miniature experiment was designed and carried out to verify the feasibility of this method and to demonstrate the effectiveness of this correction factor. The results from the experiment indicated that the error in predicting the bar length with the PS method can be reduced from 8.4% as with the traditional procedure to 4.2% as the correction factor being taken into account. An axisymmet-ric finite element model developed for simulating the PS test was used to cross check with the experimen-tal results. With this numerical model, a series of parametric studies were carried out to study the effects of these parameters in the accuracy of prediction. From the study, it is concluded that the error of the estimate increases with the pile-to- borehole distance but decreases with the pile length. In all cases, the correction factor plays a good role in enhancing the accuracy of the estimation. Another observation is that the predicting work becomes harder as the stiffness of the surrounding soil increases toward that of the pile. The numerical model was finally used to simulate the response of a defective pile with a neck or bulge. The results indicated that the capability of the PS method in locating and characterizing the neck or bulge is promising. 0.000 0.002 0.004 0.006 0.008 0.010 Time (sec) 0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0 16.0 Depth (m) = 0.5 m = 0.6 m = 0.7 m = 0.8 m = 0.9 m

(20)

Acknowledgments

The work presented in this paper was ﬁnancially supported by the National Science Council of Taiwan, ROC, under the grants NSC89-2625-Z-216-002 and NSC90-2211-E-216-006. These helps are gratefully acknowledged.

References

ASTM Book of Standards, 2004. D5882, vol. 04.09.

Baker Jr., C.N., Drumright, E.E., Mensah, F.D., Parikh, G., Ealy, C.D., 1991. Use of nondestructive testing to evaluate defects in drilled shafts: results of FHWA research. Integrity Testing of Foundations, Transportation Research Record 1331, 28–35. Davis, A.G., 2003. The nondestructive impulse response test in North America: 1985–2001. NDT & E International 36, 185–193. Davis, A.G., Dunn, C.S., 1974. From theory to ﬁled experience with the nondestructive vibration testing of piles. Proceedings of the

Institution of Civil Engineers, Part 2, 571–593.

Finno, R.J., Gassman, S.L., 1998. Impulse response evaluation of drilled shafts. Journal of Geotechnical and Environmental Engineering, ASCE 124 (10), 965–975.

Liao, S.-T., Roesset, J.M., 1997a. Identiﬁcation of defects in piles through dynamic testing. International Journal for Numerical and Analytical Methods in Geomechanics 21 (4), 277–291.

Liao, S.-T., Roesset, J.M., 1997b. Dynamic response of intact piles to impulse loads. International Journal for Numerical and Analytical Methods in Geomechanics 21 (4), 255–275.

Liao, S.-T., Roesset, J.M., Chen, C.-H., 1998. Numerical simulation of nondestructive Parallel Seismic test for piles. In: The Sixth East Asia-Paciﬁc Conference on Structural Engineering and Construction, Taipei, Taiwan, 1997–2003.

Lin, Y., Sansalone, M., Carino, N.J., 1997. Impact-echo response of concrete shafts. Geotechnical Testing Journal 14 (2), 121–137. Olson, L.D., Jalinoos, F., Aouad, M.F., 1995. Determination of Unknown Subsurface Bridge Foundations, Report submitted to

NCHRP, Transportation Research Board, National Research Council, pp. 129–148.

Paquet, J., 1991. A new method for testing integrity of piles by dynamic impulse: the impedance log. International Colloquium on Deep Foundations, Paris, March, pp. 1–10.

Rausche, F., Shen, R.K., Likins Jr., G.E., 1991. Comparison of pulse echo and transient response pile integrity test methods. Integrity Testing of Foundations, Transportation Research Record 1331, 21–27.

Rix, J.R., Jacobs, L.J., Reichert, C.D., 1993. Evaluation of Nondestructive test methods for length, diameter, and stiﬀness measurements on drilled shafts. Field performance of structure and nondestructive evaluation of subsurface infrastructure. Transportation Research Record 1415, 69–77.

Sadri, Afshin, 2003. Application of impact-echo technique in diagnoses and repair of stone masonry structures. NDT & E International 36, 195–202.

Tong, J.-H., Wu, T.-T., Lee, C.-K., 2001. A portable transient elastic wave system for in situ nondestructive evaluation of concrete. The e-Journal of Nondestructive Testing & Ultrasonics 6 (6).

Tong, J.-H., Wu, T.-T., Lee, C.-K., 2002. Fabrication of a piezoelectric impact hammer and its application to the in situ nondestructive evaluation of concrete. Japanese Journal of Applied Physics 41 (Part 1, No. 11A), 6595–6600.

Wu, T.-T., Fang, J.-S., 1997. A new method for measuring in situ concrete elastic constants using horizontally polarized conical transducers. Journal of Acoustical Society of America 101 (1), 330–336.