792

Measuring shear and compression wave velocities

of soil using bender–extender elements
E.C. Leong, J. Cahyadi, and H. Rahardjo

Abstract: Piezoceramic elements have been used for laboratory measurement of wave velocity in soil and rock specimens.
Shear-wave piezoceramic elements (bender elements) are commonly used to measure shear wave velocity for the determi-
nation of small-strain shear modulus. Compression-wave piezoceramic elements (extender elements), on the other hand,
are less commonly used as compression wave velocity is less frequently measured. In this paper, the performance of a pair
of bender–extender elements for the determination of both shear and compression wave velocities is examined with respect
to the resolution of the recorder, bender–extender element size. and excitation voltage frequency. The evaluation showed
that the performance of the bender–extender elements test can be improved by considering the following conditions:
(i) the digital oscilloscope used to record the bender–extender element signals should have a high analog to digital (A/D)
conversion resolution; (ii) the size of the bender–extender elements plays an important role in the strength and quality of
the receiver signal, especially for compression waves; and (iii) using a wave path length to wavelength ratio of 3.33 ena-
bles a more reliable determination of shear wave velocity.
Key words: piezoceramic element, bender–extender, laboratory measurement, shear wave, compression wave.
Résumé : Les éléments piézocéramiques sont utilisés pour les mesures en laboratoire de la vitesse des ondes dans des
échantillons de sol et de roche. Les éléments piézocéramiques pour les ondes de cisaillement (« bender elements ») sont
généralement utilisés pour mesurer la vitesse des ondes de cisaillement afin de déterminer le module de cisaillement à
faible déformation. Les éléments piézocéramiques pour les ondes de compression (« extender elements ») sont moins utili-
sés puisque les mesures de vitesse des ondes de compression sont moins fréquentes. Dans cet article, la performance de la
combinaison des éléments piézocéramiques pour les ondes de cisaillement et de compression est évaluée lors de la déter-
mination des vitesses des ondes de cisaillement et de compression. Cette évaluation est basée sur la résolution de l’enregis-
trement, la taille des éléments piézocéramiques et la fréquence du voltage d’excitation. Les résultats ont démontré que la
performance des éléments piézocéramiques peut être améliorée dans les conditions suivantes : (i) l’oscilloscope digital uti-
lisé pour enregistrer les signaux des éléments piézocéramiques devraient avoir une résolution de conversion analogue–
digital (A/D) élevée; (ii) la taille des éléments piézocéramiques joue un rôle important dans la force et la qualité des si-
gnaux détectés par le récepteur, en particulier pour les ondes de compression; et (iii) en utilisant un ratio de la longueur
de la trajectoire de l’onde versus la longueur d’onde de 3,33, on obtient une mesure plus fiable de la vitesse de l’onde de
cisaillement.
Mots-clés : éléments piézocéramiques, « bender–extender », mesures de laboratoire, onde de cisaillement, onde de com-
pression.
[Traduit par la Rédaction]

Introduction ceiver. Lawrence (1963, 1965) was probably the first to use
Bender elements consist of two sheets of piezoceramic piezoceramic elements bonded to shear plates to measure
plates rigidly bonded to a center shim of brass or stainless shear wave (S-wave) velocity in clay and sand specimens.
steel plate. Due to its piezoelectric properties, bender ele- Shirley (1978) introduced the use of piezoceramic bender el-
ments are capable of converting mechanical excitation into ements to determine S-wave velocity in laboratory tests. The
electrical output and vice versa. When excited by an input incorporation of bender elements for geotechnical laboratory
voltage, the bender element changes its shape and generates testing was described by Dyvik and Madshus (1985).
a mechanical excitation, hence acting as a signal transmitter. Over the years, bender elements have become increas-
When subjected to mechanical excitation, the bender ele- ingly popular for determining S-wave velocity in laboratory
ment emits an electrical output, thus acting as a signal re- soil testing and are now commercially available. However,
the usage of compression wave (P-wave) piezoceramic ele-
Received 14 August 2007. Accepted 12 February 2009.
ments is not as widespread. Two of the reasons for the low
Published on the NRC Research Press Web site at cgj.nrc.ca on usage of P-wave piezoceramic elements in laboratory soil
8 July 2009. tests are (i) the shear modulus is more applicable to describe
deformation behavior of soil as soil is loaded mostly in
E.C. Leong,1 J. Cahyadi, and H. Rahardjo. School of Civil & shear mode and (ii) this is an ineffective method for measur-
Environmental Engineering, Nanyang Technological University, ing P-wave velocity in saturated soils as the P-wave travels
Blk N1, 50 Nanyang Avenue, 639798 Singapore.
through the water phase faster than through the soil skele-
1Corresponding author (e-mail: [email protected]). ton. However, with growing interest in unsaturated soil me-

Can. Geotech. J. 46: 792–812 (2009) doi:10.1139/T09-026 Published by NRC Research Press
Leong et al. 793

chanics, P-wave velocity measurement becomes more im- Fig. 1. Types of piezoceramic elements: (a) x-poled; (b) y-poled.
portant as it can provide an assessment of the degree of sat-
uration as well as the porosity of the soil specimen (Brignoli
et al. 1996).
Usually, it is more common to use a pair of S-wave pie-
zoceramic elements and a pair of P-wave piezoceramic ele-
ments mounted in the same apparatus to measure both S-
wave and P-wave velocities, respectively (e.g., Schultheiss
1981; Bates 1989; de Alba and Baldwin 1991; Brignoli et uration in terms of deflection per unit of applied voltage is
al. 1996; Nakagawa et al. 1996; Fioravante and Capoferri greater by a factor of two as compared with the series con-
2001). Recently, Lings and Greening (2001) introduced a figuration. This is accomplished by a reduction in impe-
single hybrid element called ‘‘bender–extender element.’’ A dance of 4:1 (Germano 2002). In other words, for the same
simple modification is made to the bender element to pro- motion, the parallel configuration requires only half of the
duce an extender element capable of transmitting and re- voltage needed for the series operation. Therefore, the paral-
ceiving P-waves. The key to its modification is the lel configuration is better suited as a transmitter as it gives a
changing of the bender element wiring configuration from a stronger mechanical motion for a given excitation voltage.
two-wire configuration to a three-wire configuration and However, the series configuration is better suited as a re-
then matching it with the corresponding type of polarization. ceiver as it gives a higher output voltage for a given me-
Lings and Greening (2001) further suggested that using a chanical excitation.
single pair of bender–extender elements enables both S- In the bender–extender elements test, a pair of piezocer-
wave and P-wave velocity measurements to be made over amic elements is used. One piezoceramic element is used as
exactly the same soil path, which facilitates the determina- the S-wave transmitter as well as the P-wave receiver, and
tion of the arrival of the S-wave, especially when it is ob- the other piezoceramic element is used as the P-wave trans-
scured by the near-field effect. mitter as well as the S-wave receiver. A schematic diagram
Lings and Greening (2001) have also performed some of the bender–extender elements test set-up inside the triax-
preliminary testing on dry sand specimens to validate the ial cell is shown in Fig. 4. Depending on the type of wave
use of bender–extender elements. The results showed that a that is transmitted and received through the soil specimen,
single pair of bender–extender elements can successfully de- it is possible to obtain the S-wave velocity, Vs, and P-wave
tect both shear and compression waves under various con- velocity, Vp, from the distance between the tip of the piezo-
fining pressures. The intent of this paper is to evaluate the ceramic elements, Ltt, and the travel time, Dt using the fol-
performance of bender–extender elements in the determina- lowing equation for velocity V:
tion of both shear and compression wave velocities in terms
½1 V ¼ Ltt =Dt
of resolution of the recorder, size of the bender–extender el-
ements, and excitation voltage frequency. Readings from a local displacement transducer attached to
the specimen were used to correct Ltt during isotropic com-
Principles of bender–extender element test pression.
Two types of piezoceramic elements are available, x- From the S-wave velocity, the small-strain shear modulus,
Gmax, can be determined from elastic wave theory
poled and y-poled (Fig. 1), which differ with respect to their
direction of polarization. The x-poled piezoceramic element ½2 Gmax ¼ rVs2
refers to the condition where the directions of the polariza-
tion for each of the two layers are pointing in opposite di- where r is the soil specimen’s density.
rections, generally towards each other. The y-poled From the P-wave velocity, the small-strain constrained
piezoceramic element refers to the condition where the di- modulus, Mmax, can be calculated by
rections of the polarization for each of the two layers are
pointing in the same direction. ½3 Mmax ¼ rVp2
These piezoceramic elements can be wired using two
types of configuration, i.e., series and parallel. In the series
configuration, the excitation voltage is applied across the Test set-up
two piezoceramic layers at once, as seen in Figs. 2a and 3b. The bender–extender element test set-up used in this
In the parallel configuration, the excitation voltage is applied study was installed into a triaxial cell. Elements were in-
to the piezoceramic layers and the centre shim as seen in stalled inside the top cap and bottom pedestal. A schematic
Figs. 2b and 3a. To transmit and (or) receive a S-wave and diagram of the bender–extender element test set-up used in
hence, act as a bender, the x-poled and y-poled piezoceramic this study is shown in Fig. 4. To investigate the effect of
elements have to be connected in series and parallel configu- the bender–extender element size, two sets of bender–
ration, respectively, i.e., Fig. 2. To transmit and (or) receive extender elements were used in this study. The first pair
a P-wave and hence, act as an extender element, the x-poled was x-poled PZT5A T220-A4-103X (abbreviated hereafter
and y-poled piezoceramic elements have to be connected in as 103X) and y-poled PZT5A T220-A4-103Y (abbreviated
parallel and series configuration, respectively, i.e., Fig. 3. hereafter as 103Y) piezoceramic elements. The second pair
As the excitation voltage is applied to each of the piezo- was x-poled PZT5A T220-A4-303X (abbreviated hereafter
ceramic layers, the motion sensitivity of the parallel config- as 303X) and y-poled PZT5A T220-A4-103Y piezoceramic

Published by NRC Research Press

794 Can. Geotech. J. Vol. 46, 2009

Fig. 2. Bender element wiring configuration using (a) x-poled series and (b) y-poled parallel configurations (adapted from www.piezo.com,
Piezo Systems, Inc., Woburn, Mass.).

Fig. 3. Extender element wiring configuration using (a) x-poled parallel and (b) y-poled series configurations (adapted from www.piezo.
com, Piezo Systems, Inc., Woburn, Mass.).

Fig. 4. Bender–extender element test set-up.

elements. The piezoceramic elements were from Piezo Sys- A three-wire configuration was adopted to fabricate the
tems Inc., (Woburn, Mass.). The dimensions for the 103X bender–extender elements. The x-poled piezoceramic ele-
and 103Y series were 15.9 mm  3.2 mm  0.51 mm. The ment was installed inside the bottom pedestal, whereas the
dimensions for the 303X series were 15.9 mm  9.5 mm  y-poled piezoceramic element was installed inside the top
0.51 mm. The dimensions of the piezoceramic elements cap. To generate the S-wave, the y-poled piezoceramic ele-
were modified from their original dimensions to accommo- ment was connected in parallel configuration as a transmitter
date their installation into the top cap and bottom pedestal and the x-poled piezoceramic element was connected in ser-
of the triaxial cell. ies configuration as a receiver. To generate the P-wave, the

Published by NRC Research Press

Leong et al. 795

Fig. 5. Insulation of bender–extender element and its (a) series and (b) parallel wiring configuration.

Fig. 6. Typical output signals with and without grounding.

x-poled piezoceramic element was connected in parallel Table 1. Basic properties of soils tested.
configuration as a transmitter and the y-poled piezoceramic
element was connected in series configuration as a receiver. Property Value
Residual soil
Bender–extender elements are very sensitive to undesired Specific gravity, Gs 2.62
electromagnetic signals or noises that may affect their per- Liquid limit, LL (%) 62
formance. One common type of interference is cross-talk. Plastic limit, PL (%) 29
As the type of connection for bender–extender elements Plasticity index, PI (%) 33
will always be in parallel and series modes, cross-talk is ex- Sand grain-size distribution (%) 70
pected (Lee and Santamarina 2005). Therefore, the bender– Silt and clay grain-size distribution (%) 30
extender elements were fabricated with shielding and USCS SC
grounding to minimize the cross-talk effect. The electrical Sand
shielding was provided by insulating the bender–extender el- Specific gravity, Gs 2.64
Minimum dry density (Mg/m3) 1.38
ement with Teflon tape followed by aluminum tape (for
Maximum dry density (Mg/m3) 1.79
grounding attachment). The assembly was then waterproofed Maximum void ratio, emax 0.91
by coating it with an even layer of epoxy resin. The details Minimum void ratio, emin 0.47
can be seen in Fig. 5. Figure 6 shows a comparison of typi- USCS SP
cal output signals with and without grounding for a pair of
Note: USCS, unified soil classification system (ASTM
303X and 103Y bender–extender elements. It can be seen 2006); SC, clayey sand; SP, poorly graded sand.
that grounding greatly minimizes the cross-talk effect.

Published by NRC Research Press

796 Can. Geotech. J. Vol. 46, 2009

Fig. 7. Grain-size distribution of soils tested.

Fig. 8. Standard Proctor compaction curve of Bukit Timah granite transmitted by the transmitter through the soil specimen was
residual soil. detected by the receiver. Both the transmitter and receiver
signals were recorded by the digital oscilloscope.

Soil samples
To investigate the performance of the bender–extender el-
ements, two types of soil were used: sand (dry) and Bukit
Timah granite residual soil (unsaturated) from a site in the
Mandai area, central Singapore. The index properties of the
soils are summarized in Table 1. The grain-size distributions
of the soils are shown in Fig. 7. The sand specimen was pre-
pared in six layers using dry tamping to achieve a total den-
sity of 1.66 Mg/m3, corresponding to a relative density of
about 74%, inside the triaxial cell. The sand specimen had
a diameter of 50 mm and a height of 120 mm. The Bukit
Timah Granite residual soil was compacted using the stand-
ard Proctor method in ASTM standard D698-91 (ASTM
1997b). The standard Proctor compaction curve of the resid-
ual soil specimen is shown in Fig. 8. The compacted resid-
The bender–extender elements were positioned inside a ual soil specimen had a dry density of 1.72 Mg/m3, a water
cylindrical tube with a protrusion length of 3 mm and fixed content of 18.1%, and a degree of saturation of 90%. The
in position using epoxy resin. The cylindrical tubes were compacted condition of the residual soil specimen is also in-
then slotted into the top cap and bottom pedestal of the tri- dicated in Fig. 8. The compacted residual soil specimen was
axial cell (Leong et al. 2005). A tight fit was achieved be- trimmed using a soil lathe to a diameter of 50 mm and a
tween the tube and the top cap as well as the bottom height of 100 mm. After trimming, the residual soil speci-
pedestal by means of O-rings. The gap between the tubes men was transferred to the triaxial apparatus for the
and the porous stone was further sealed with silicon rubber. bender–extender element test.
The test set-up also consisted of a function generator
(Hewlett–Packard (HP) model 33120A), a power amplifier Measurement method and test conditions
(Piezo EPA-104), and a digital oscilloscope (HP model The transmitter element was excited using a single-
54610B, Yokogawa model DL750P). The function generator sinusoidal voltage pulse. Leong et al. (2005) proposed to
produces the excitation voltage (20 V peak-to-peak i.e., adopt a signal-to-noise ratio of at least 4 dB for the receiver
20 Vpp), which is further amplified by the power amplifier signals to improve the interpretation of the bender element
and applied to the x-poled or y-poled piezoceramic ele- test. The signal-to-noise ratio of the receiver signal can be in-
ments. The power amplifier, Piezo EPA-104, is capable of creased by increasing the magnitude of the applied voltage to
amplifying the function generator voltage from 20 Vpp up to the transmitter. In this study, the input voltage applied to the
200 Vpp. However, the excitation voltage applied was lim- bender–extender elements was the rated voltage, i.e., 90 Vpp,
ited by the bender–extender elements rated voltage i.e., which is the highest voltage that can be applied to the ele-
90 Vpp for both the x-poled and y-poled piezoceramic ele- ments without possible deterioration due to depolarization.
ments in parallel configuration. This was done to avoid de- It is well known that the S-wave arrival is influenced by
polarization of the bender–extender elements. The wave the near-field effect. The wave path length to wavelength ra-

Published by NRC Research Press

Leong et al. 797

Fig. 9. Typical shear wave (S-wave) signals recorded on both Hewlett–Packard (HP) 54610B and Yokogawa DL750P digital oscilloscopes
and their corresponding frequency spectrum.

tio (Ltt/l) is an important parameter for the determination of al. (2001), and Leong et al. (2005) computed the Ltt/l ratio
S-wave travel time (Sanchez-Salinero et al. 1986; Mancuso using the input excitation frequency, fin, for l, i.e.,
and Vinale 1988; Brignoli et al. 1996; Arulnathan et al.
1998; Pennington et al. 2001; Arroyo et al. 2003; Leong et ½4 Ltt =l ¼ ðLtt =VÞfin ¼ fin Dt
al. 2005). Brignoli et al. (1996) computed the Ltt/l ratio using In this study, the Ltt/l ratio was determined using fin and
the predominant frequency of the first cycle of the received fout, i.e., finDt and foutDt, respectively. The receiver signal is
waveform, fout, for l. Arulnathan et al. (1998), Pennington et strongly affected by the response of the bender element–soil

Published by NRC Research Press

798 Can. Geotech. J. Vol. 46, 2009

Fig. 10. Typical compression wave (P-wave) signals recorded on both HP 54610B and Yokogawa DL750P digital oscilloscopes and their
corresponding frequency spectrum.

system, which is in turn a function of the input frequency frequencies between 10 and 30 kHz for P-waves as these
and the resonant frequency of this system (Jovičič et al. frequencies showed more significant effects on the receiver
1996; Kawaguchi et al. 2001; Lee and Santamarina 2005). signals.
To investigate the effect of input frequencies, bender– The receiver bender element signal is also dependent on the
extender element tests were performed with input frequen- input waveform. Jovičič et al. (1996) and Blewett et al. (2000)
cies between 1 and 30 kHz. Results are presented for input observed that the receiver signals for square waves do not re-
frequencies between 1 and 20 kHz for S-waves and input semble the original transmitter signal. Leong et al. (2005)

Published by NRC Research Press

Leong et al. 799

Fig. 11. Typical influence of near-field effect on the shear wave for compacted residual soil at an effective confining pressure of 50 kPa.

Published by NRC Research Press

800 Can. Geotech. J. Vol. 46, 2009

Fig. 12. Typical influence of near-field effect on the shear wave for compacted residual soil at an effective confining pressure of 100 kPa.

Published by NRC Research Press

Leong et al. 801

Fig. 13. Typical influence of near-field effect on the shear wave for compacted residual soil at an effective confining pressure of 200 kPa.

Published by NRC Research Press

802 Can. Geotech. J. Vol. 46, 2009

Fig. 14. Typical influence of near-field effect on the shear wave for compacted residual soil at an effective confining pressure of 400 kPa.

Published by NRC Research Press

Leong et al. 803

Fig. 15. Effect of the Ltt/l ratio on shear wave velocity for the compacted residual soil tested under effective confining pressures of 50, 100,
200, and 400 kPa.

found that the square wave causes greater ambiguity in arrival the Yokogawa DL750P oscilloscope without further signal
time than the sinusoidal wave. Therefore, in this study, a si- processing.
nusoidal waveform was selected as the input waveform. Figure 10 shows typical P-wave receiver signals from a
pair of 303X and 103Y bender–extender elements recorded
Results and discussions on both HP 54610B and Yokogawa DL750P oscilloscopes
and their corresponding frequency spectrum for the same
Resolution of the signal recorder dry sand specimen subjected to an isotropic effective confin-
In Leong et al. (2005), the digital oscilloscope used to re- ing pressure of 100 kPa. The excitation pulse for the P-wave
cord both the transmitter and receiver signals was HP model had a frequency of 20 kHz. Again, it can be seen that the
54610B, which has an 8 bit analog to digital (A/D) conver- signal obtained from the Yokogawa DL750P oscilloscope is
sion resolution. The receiver signals obtained from the oscil- much clearer compared with that obtained from the HP
loscope were then transferred to a computer for further 54610B oscilloscope. Furthermore, the frequency spectrum
signal processing. All of the receiver signals were treated of the signal recorded in the HP 54610B does not show a
using a Hanning window to eliminate Gibbs’ phenomenon clear dominant frequency. The dominant frequency of the
(Hamming 1977) and were filtered through two passes of a receiver signal recorded by the Yokogawa DL750P oscillo-
fourth-order, low-pass Butterworth filter with a cut-off fre- scope, on the other hand, was about 20 kHz, close to the ex-
quency above the dominant frequency of the receiver signal citation frequency.
to remove the phase shift due to filtering (Li 1997). Besides A/D conversion resolution, the Yokogawa
Although these procedures improve the receiver signal qual- DL750P oscilloscope allows the user to select a bandwidth
ity, the resolution of the receiver signal is limited by the 8 limit of 5, 50 or 500 kHz, whereas the bandwidth for the
bit A/D conversion resolution of the digital oscilloscope. In HP 54610B oscilloscope is fixed at 30 MHz. The maximum
this study, the Yokogawa model DL750P digital oscillo- voltage sensitivity for HP 54610B oscilloscope was also
scope, which has a 12 bit A/D conversion resolution, was limited to 2 mV/div (1 div = 0.002 V) compared with
employed. 0.05 mV/div for the Yokogawa DL750P oscilloscope.
Figure 9 shows typical S-wave receiver signals from a Hence, it would be difficult for the HP 54610B oscilloscope
pair of 303X and 103Y bender–extender elements recorded to detect the arrival of a receiver signal if its peak-to-peak
on both HP 54610B and Yokogawa DL750P oscilloscopes amplitude was less than 2 mV, as shown in Fig. 10. There-
and their corresponding frequency spectrum for dry sand fore, a digital oscilloscope with a high A/D conversion reso-
specimen. The dry sand specimen was subjected to isotropic lution, multiple bandwidth, and high voltage sensitivity is
effective confining pressure of 100 kPa. The excitation essential for bender–extender element tests. The subsequent
pulse for the S-wave had a frequency of 10 kHz. It can be tests described in this paper employed only the Yokogawa
seen that although the results from both oscilloscopes show DL750P oscilloscope in performing the bender–extender el-
that the receiver signals have the same dominant frequency, ement tests.
the signal-to-noise ratio and the signal amplitude obtained
from the Yokogawa DL750P oscilloscope is much higher Bender–extender element size
than that obtained from the HP 54610B oscilloscope. In ad- Initially in this study, a pair of piezoceramic elements —
dition, the travel time of the S-wave is more readily visible 103X and 103Y — was used. Their dimensions were the
and can be determined from the receiver signal recorded by same i.e., 15.9 mm  3.2 mm  0.51 mm. For the S-wave,

Published by NRC Research Press

804 Can. Geotech. J. Vol. 46, 2009

Fig. 16. Typical compression wave signals under different input frequencies for the compacted residual soil at an effective confining pres-
sure of 50 kPa.

the 103Y and the 103X piezoceramic elements were con- 103X and the 103Y piezoceramic elements were connected
nected in parallel configuration as a transmitter and in series in parallel configuration as a transmitter and in series con-
configuration as a receiver, respectively. For the P-wave, the figuration as a receiver, respectively. Therefore, a single

Published by NRC Research Press

Leong et al. 805

Fig. 17. Typical compression wave signals under different input frequencies for the compacted residual soil at an effective confining pres-
sure of 100 kPa.

pair of piezoceramic elements can be used as a bender and wave velocity, but not P-wave velocity as no signal was de-
extender element by changing the wiring configuration. tected by the receiver 103Y piezoceramic element. Increas-
This pair of piezoceramic elements was able to measure S- ing the input excitation voltage to the maximum value of

Published by NRC Research Press

806 Can. Geotech. J. Vol. 46, 2009

Fig. 18. Typical compression wave signals under different input frequencies for the compacted residual soil at an effective confining pres-
sure of 200 kPa.

90 Vpp and increasing the cantilever length of the transmitter extender during voltage excitation was about one order-of-
103X piezoceramic element also failed to solve the problem. magnitude smaller than the lateral movement of the bender
Lings and Greening (2001) found that the lengthening of the during voltage excitation in their bender–extender element.

Published by NRC Research Press

Leong et al. 807

Fig. 19. Typical compression wave signals under different input frequencies for the compacted residual soil at an effective confining pres-
sure of 400 kPa.

It was thus concluded that the transmitter 103X piezocer- this, the transmitter 103X piezoceramic element was re-
amic element was not creating sufficient P-wave energy to placed with a piezoceramic element, 303X, having a larger
reach the receiver 103Y piezoceramic element. To confirm width (15.9 mm  9.5 mm  0.51 mm). After the replace-

Published by NRC Research Press

808 Can. Geotech. J. Vol. 46, 2009

Fig. 20. Comparison of compression and shear wave signals for undisturbed residual soil specimen S4 at an effective confining pressure of
50 kPa. (a) Compression wave at fin = 30 kHz; (b) shear wave at fin = 1 kHz; (c) shear wave at fin = 16 kHz.

ment, the arrival of the P-wave was successfully detected. same, eq. [5] shows that a larger W would generate a smal-
Hence, the performance of 103Y with the 303X piezocer- ler output voltage per unit force applied to the tip of the
amic element to S-wave was re-assessed. piezoceramic element. Generally, it was observed that the
For the 303X piezoceramic element in series configura- 303X piezoceramic element gives a smaller output voltage
tion, the output generated voltage, Vo, due to a mechanical when measuring the S-wave receiver signals. However, the
excitation is given by (Germano 2002) higher resolution of the Yokogawa DL750P oscilloscope
compensates for the loss in amplitude of the S-wave recei-
   ver signal. Hence, tests described in this paper employed
3 Flb t2 the 103Y piezoceramic element as the S-wave transmitter
½5 Vo ¼ g31 1  s2 K
2 WT T and P-wave receiver and the 303X piezoceramic element as
the P-wave transmitter and S-wave receiver.
where g31 is the piezoelectric voltage constant, F is the ap-
plied force, lb is the cantilever (protrusion) length of the Effect of excitation frequency
bender element, W is the width of the bender element, T is The travel time of the S-wave can be taken as the first ar-
the thickness of the bender element, ts is the thickness of the rival of the receiver signal (Leong et al. 2005). Reversal of
centre shim ( < < T ), and K is an empirical weighting factor the received signal’s polarity when the polarity of the trans-
(‡1). If the net force on the piezoceramic element is the mitting element is inverted is often taken as the arrival of

Published by NRC Research Press

Leong et al. 809

Fig. 21. Comparison of compression and shear wave signals for undisturbed residual soil specimen S7 at an effective confining pressure of
60 kPa. (a) Compression wave at fin = 30 kHz; (b) shear wave at fin = 1 kHz; (c) shear wave at fin = 16 kHz.

the S-wave (Abbiss 1981). However, numerical studies by (l), Ltt/l. The near-field effect amplitude decays rapidly
Mancuso and Vinale (1988) have shown that the first deflec- with increasing number of wavelengths between the trans-
tion of the S-wave signal may not correspond to the arrival mitter and the receiver element.
of the S-wave but to the arrival of the so-called ‘‘near-field’’ Sanchez-Salinero et al. (1986) showed that the near-field
component, which travels with the velocity of the P-wave. effects are not significant when Ltt/l is greater than 2.
In addition, inverting the polarity of the transmitter element Brignoli et al. (1996) found that the near-field effect de-
does not help to eliminate the near-field effect as it produces creases as Ltt/l increases. Arulnathan et al. (1998) found
the reversal of all the waveform components including the that the near-field effect was absent when Ltt/l is greater
near-field components. than 1. They also found that the receiver signals deteriorated
Sanchez-Salinero et al. (1986) showed that the S-wave as Ltt/l increased and attributed the deterioration to the ef-
signal is always accompanied by the propagation of another fect of the ratio of wavelength to cantilever length of the
signal of opposite polarity that travels with the velocity of bender element (l/lb). Arroyo et al. (2003) showed that the
the P-wave. In other words, the bender element that produ- near-field effect is less than 5% if Ltt/l is greater than 1.6.
ces mainly shear displacement will also generate a compo- However, Arroyo et al. (2003) also pointed out that this cri-
nent that travels at the P-wave velocity. This phenomenon terion is only useful in determining the S-wave velocity in
is known as the near-field effect, which is quantified in the frequency domain and not in the time domain. Leong et
terms of the ratio of wave path length (Ltt) to wavelength al. (2005) suggested a higher limit of 3.33 for the Ltt/l ratio

Published by NRC Research Press

810 Can. Geotech. J. Vol. 46, 2009

based on ASTM standard D2845-95 (ASTM 1997a) and ex- Table 2. Basic properties of undisturbed residual soils tested.
perimental observation. Property Value
Figures 11 to 14 show typical S-wave signals produced Residual soil S4
from the bender–extender elements performed on compacted Specific gravity, Gs 2.66
residual soil specimens. The excitation frequencies were var- Liquid limit, LL (%) 45
ied from 1 to 16 kHz. The specimen was subjected to iso- Plastic limit, PL (%) 20
tropic effective confining pressures of 50, 100, 200, and Plasticity index, PI (%) 25
400 kPa. A table of values is given to the left of each re- Sand grain-size distribution (%) 17
corded waveform in Figs. 11 to 14. The values are: (i) input Silt and clay grain-size distribution (%) 83
excitation frequency, fin; (ii) predominant frequency of the USCS CL
first cycle of the received waveform, fout; (iii) the arrival Initial bulk density (Mg/m3) 2.013
time, Dt; (iv) Ltt/l in terms of finDt; (v) Ltt/l in terms of Initial void ratio 0.60
foutDt; (vi) Ltt; and (vii) Vs. Due to the damping properties Natural water content (%) 21.1
of soil and the interaction of the bender–extender elements Residual soil S7
and the soil specimen, the predominant frequency of the re- Specific gravity, Gs 2.65
ceived waveform, fout, is usually lower than the input excita- Liquid limit, LL (%) 38
tion frequency, fin. This is especially the case when the Plastic limit, PL (%) 23
excitation frequency is equal to or greater than 8 kHz. It Plasticity index, PI (%) 15
can be seen that the near-field effect tends to fade away as Sand grain-size distribution (%) 46
the excitation frequency increases or as the Ltt/l ratio in- Silt and clay grain-size distribution (%) 54
creases. It was also observed that the strongest signal was USCS CL
received at an excitation frequency of 4 kHz, except for the Initial bulk density (Mg/m3) 2.112
case where confining pressure was 400 kPa. The strongest Initial void ratio 0.52
receiver signal for this case was at excitation frequency of Natural water content (%) 21.23
8 kHz. Lee and Santamarina (2005) suggested that the exci-
tation frequency of the strongest receiver signal gives the Note: USCS, unified soil classification system (ASTM 2006); CL,
clay of low plasticity.
approximate resonant frequency of the system. Identification
of the arrival time Dt is most difficult for the S-wave. From
the test results in Figs. 11 to 14, the arrival time Dt is given
as the travel time to the first bump maximum as described the strength of the P-wave signals improves with increasing
by Lee and Santamarina (2005) or the first deflection if first confining pressures, an observation also made by Lings and
bump is not visible (Leong et al. 2005). Greening (2001). The strongest receiver signal was obtained
Lee and Santamarina (2005) showed that the transverse from an excitation frequency of about 15 kHz, except for the
directivity of bender elements can affect S-wave measure- case of an effective confining pressure of 400 kPa where the
ments. When excited, bender elements produce compression excitation frequency was about 20 kHz.
and rarefaction wave side lobes normal to their plane and a Comparing the S-wave and P-wave signals at the same ef-
S-wave frontal lobe. The P-waves reflected from the boun- fective confining pressure, e.g., Figs 13. and 18, show that
dary can interfere with the S-wave arrival. Arroyo et al. the initial deflection in the S-wave signal corresponds ap-
(2006) found that specimen size has an effect on the re- proximately to the arrival of the P-wave. The phenomenon
ceiver signal in bender element tests due to reflections from is clearer for undisturbed residual soil specimens as shown
the boundary of the specimen. The effect is more pro- in Figs. 20 and 21. The properties of the undisturbed resid-
nounced in small-diameter soil specimens. Some of these ef- ual soil specimens are summarized in Table 2.
fects may be present in the output signals of Figs. 11 to 14, Therefore, from the above results and discussions, it can
but they are not investigated in this paper. be concluded that a single pair of 303X and 103Y piezocer-
Figure 15 shows the effect of Ltt/l on the S-wave veloc- amic elements is able to perform reasonably well when de-
ities for the compacted residual soil specimen tested under termining both shear and compression wave velocities of
various confining pressures. It can be seen that there is a soil specimens. Having a single pair of piezoceramic ele-
greater discrepancy in the S-wave velocity for a Ltt/l ratio ments also enables the measurement of both shear and com-
below 2. This observation is in agreement with Sanchez- pression wave velocities over the same soil path (Brignoli et
Salinero et al. (1986), Arulnathan et al. (1998), Pennington al. 1996; Lings and Greening 2001).
et al. (2001), Arroyo et al. (2003), and Leong et al. (2005).
Based on the above discussions, the proposal by Leong et al. Conclusions
(2005) that Ltt/l = 3.33 can provide a more reliable estimate
of S-wave velocity is found to be reasonable.
Figures 16 to 19 show the P-wave signals obtained from  Bender–extender elements were fabricated to measure
the same compacted residual soil specimen and tested under shear and compression wave velocities in a triaxial appa-
the same conditions. The input frequencies were varied from ratus. The results show that the bender–extender elements
10 to 30 kHz. It can be seen from Figs. 16 to 19 that the were able to measure both shear and compression wave
arrival times are not influenced by the input frequencies. velocities reliably if certain test conditions were met.
Hence, the P-wave travel time is relatively easier to deter-  It is important to employ a high-resolution digital oscillo-
mine as compared with the S-wave travel time. In addition, scope in performing bender–extender element tests. The

Published by NRC Research Press

Leong et al. 811

use of a high-resolution oscilloscope (‡12 bits) helps to electric transducers. Geotechnical Testing Journal, 19(4): 384–
improve the detection of the arrival of the P-wave as the 397. doi:10.1520/GTJ10716J.
receiver wave has a low amplitude. de Alba, P., and Baldwin, K.C. 1991. Use of bender elements in
 The size of the piezoceramic element plays an important soil dynamics experiments. In Recent advances in instrumenta-
role in the quality of the receiver signal, especially for P- tion, data acquisition and testing in soil dynamics. Edited by
S.K. Bhatia and G.W. Blaney. ASCE Geotechnical Special Pub-
waves. In this study, the performance of PZT5A T220-A4-
lication No 29. American Society of Civil Engineers (ASCE),
103Y and PZTA T220-A4-303X piezoceramic elements as
Reston, Va. pp. 86–101
a pair of bender-extender elements was found to be satis- Dyvik, R., and Madshus, C. 1985. Lab measurement of Gmax using
factory. For the S-wave, the 103Y element was used as bender elements. Norwegian Geotechnical Institute, Oslo, Nor-
the transmitter and the 303X element was used as the re- way. pp. 186–196. Publication No. 161.
ceiver. For the Pwave, the 303X element was used as the Fioravante, V., and Capoferri, R. 2001. On the use of multi-direc-
transmitter and the 103Y element was used as the receiver. tional piezoelectric transducers in triaxial testing. Geotechnical
 The S-wave signal was affected by near-field effects and Testing Journal, 24(3): 243–255. doi:10.1520/GTJ11344J.
transverse directivity. Germano, C.P. 2002. Flexure mode piezoelectric transducers. Mor-
 The near-field effect in S-wave signals diminishes as the gan Electro Ceramics, Southampton, UK. Technical Publication
input excitation frequency increases. Using Ltt/l of at TP-218.
least 3.33 was found to be reasonable to eliminate the Hamming, R.W. 1977. Digital filters. Prentice Hall, Englewood
near-field effect for anS-wave. Unlike the S-wave, the P- Cliffs, N.J.
Jovičič, V., Coop, M.R., and Simič, M. 1996. Objective criteria for
wave arrival time is not influenced by input frequency.
determining Gmax from bender element tests. Géotechnique,
46(2): 357–362.
Acknowledgements Kawaguchi, T., Mitachi, T., and Shibuya, S. 2001. Evaluation of
The work described in this paper is part of a research shear wave travel time in laboratory bender element test. In Pro-
project funded by the Defense Science and Technology ceedings of the Fifteenth International Conference on Soil Me-
Agency, Grant No. PTRC-CEE/DSTA/2006.01, and Na- chanics and Geotechnical Engineering, Istanbul, Turkey, 27–
nyang Technological University, Grant No. RG 81/05. The 31 August 2001. A.A. Balkema, Rotterdam, the Netherlands.
assistance of Ms. Than Than Nyunt in performing some of Vol. 1. pp. 155–158.
the tests is gratefully acknowledged. Lawrence, F.V. 1963. Propagation of ultrasonic waves through
sand. Massachusetts Institute of Technology, Cambridge, Mass.
Research Report R63–08.
References Lawrence, F.V. 1965. Ultrasonic shear wave velocity in sand and
Abbiss, C.P. 1981. Shear wave measurements of the elasticity of clay. Massachusetts Institute of Technology, Cambridge, Mass.
the ground. Géotechnique, 31(1): 94–104. Research Report R65–05.
Arroyo, M., Muir Wood, D., and Greening, P.D. 2003. Source Lee, J.S., and Santamarina, J.C. 2005. Bender elements: perfor-
near-field effects and pulse tests in soil samples. Géotechnique, mance and signal interpretation. Journal of Geotechnical and
53(3): 337–345. doi:10.1680/geot.53.3.337.37277. Geoenvironmental Engineering, 131(9): 1063–1070. doi:10.
Arroyo, M., Muir Wood, D., Greening, P.D., Medina, L., and Rio, 1061/(ASCE)1090-0241(2005)131:9(1063).
J. 2006. Effects of sample size on bender-based axial G0 mea- Leong, E.C., Yeo, S.H., and Rahardjo, H. 2005. Measuring shear
surements. Géotechnique, 56(1): 39–52. doi:10.1680/geot.2006. wave velocity using bender elements. Geotechnical Testing
56.1.39. Journal, 28(5): 488–498.
Arulnathan, R., Boulanger, R.W., and Riemer, M.F. 1998. Analysis Li, X.S. 1997. Discussion: Pulse transmission system for measuring
of bender element test. Geotechnical Testing Journal, 21(2): wave propagation in soils. Journal of Geotechnical and Geoen-
120–131. doi:10.1520/GTJ10750J. vironmental Engineering, 123(9): 883–884. doi:10.1061/(ASCE)
ASTM. 1997a. Testing materials, laboratory determination of pulse 1090-0241(1997)123:9(883).
velocities and ultrasonic elastic constants of rock. ASTM stan- Lings, M.L., and Greening, P.D. 2001. A novel bender/extender
dard D2845-95 Annual Book of ASTM Standards. ASTM Inter- element for soil testing. Géotechnique, 51(8): 713–717. doi:10.
national, West Conshohocken, Pa. 1680/geot.51.8.713.40469.
ASTM. 1997b. Testing materials, laboratory compaction character- Mancuso, C., and Vinale, F. 1988. Propagazione delle onde sis-
istics of soil using standard effort. ASTM standard D698- miche: teoria e misura in sito. Atti del Convegno del Gruppo
91Annual Book of ASTM Standards. ASTM International, West Nazionale di Coordinamento per gli Studi di Ingegneria Geotec-
Conshohocken, Pa. nica, Monselice, Italy. pp. 115–138. [In Italian.]
ASTM. 2006. Standard practice for classification of soils for engi- Nakagawa, K., Soga, K., and Mitchell, J.K. 1996. Pulse transmis-
neering purposes (Unified Soil Classification System). ASTM sion system for measuring wave propagation in soils. Journal of
standard D2487. American Society for Testing and Materials, the Geotechnical Engineering Division, 122(4): 302–308. doi:10.
West Conshohocken, Pa. doi:10.1520/D2487-06E01. 1061/(ASCE)0733-9410(1996)122:4(302).
Bates, C.R. 1989. Dynamic soil property measurement during triax- Pennington, D.S., Nash, D.F.T., and Lings, M.L. 2001. Horizon-
ial testing. Géotechnique, 39(4): 721–726. tally mounted bender elements for measuring anisotropic shear
Blewett, J., Blewett, I.J., and Woodward, P.K. 2000. Phase and am- moduli in triaxial clay specimens. Geotechnical Testing Journal,
plitude responses associated with the measurements of shear 24(2): 133–144. doi:10.1520/GTJ11333J.
wave velocity in sand by bender elements. Canadian Geotechni- Piezo Systems, Inc. 2003. Application data [online]. Available from
cal Journal, 37(6): 1348–1357. doi:10.1139/cgj-37-6-1348. www.piezo.com [cited 10 June 2007].
Brignoli, E.G.M., Gotti, M., and Stokoe, K.H., II. 1996. Measure- Sanchez-Salinero, I., Roesset, J.M., and Stokoe, K.H., II. 1986.
ment of shear waves in laboratory specimens by means of piezo- Analytical studies of body wave propagation and attenuation.

Published by NRC Research Press

812 Can. Geotech. J. Vol. 46, 2009

Civil Engineering Department, University of Texas of Austin, Shirley, D.J. 1978. An improved shear wave transducer. The Jour-
Austin, Tex. Report GR 86–15. nal of the Acoustical Society of America, 63(5): 1643–1645.
Schultheiss, P.J. 1981. Simultaneous measurements of P & S wave doi:10.1121/1.381866.
velocities during conventional laboratory soil testing proce-
dures. Marine Geotechnology, 4(4): 343–367. doi:10.1080/
10641198109379831.

Published by NRC Research Press