Professional Documents
Culture Documents
Brandenberg, S. J. (2009) - Different Approaches For Estimating Ground Strains From Pile Driving Vibrations at A Buried Archeological Site
Brandenberg, S. J. (2009) - Different Approaches For Estimating Ground Strains From Pile Driving Vibrations at A Buried Archeological Site
Abstract: Ground strains were estimated from vibrations measured during pile driving operations at a buried, prehistoric archeological
Downloaded from ascelibrary.org by Tokyo Univ Seisan Gijutsu on 05/14/15. Copyright ASCE. For personal use only; all rights reserved.
site to monitor potential construction impacts. Subsurface characteristics of the site were investigated using multiple cone penetration test
共CPT兲 soundings and the shear wave velocity profile was measured using the seismic CPT method. Embedded geophones and surface
accelerometers were then used to measure ground vibrations during pile driving. Displacement gradients were estimated from the
vibrations using the following three methods: 共1兲 the difference between adjacent displacements divided by sensor spacing; 共2兲 peak
particle velocity divided by depth-dependent wave velocity 共i.e., at the depth where the sensor was placed兲; and 共3兲 peak particle velocity
divided by frequency-dependent wave velocity from a measured dispersion curve. Methods 共1兲 and 共3兲 agreed well, while method 共2兲
caused errors that depended on depth of embedment of the sensors and distance from pile driving. Errors in 共2兲 were attributed to a
mismatch between the depth-dependent wave velocity and the wave velocity on the frequency band that carried the largest velocity pulse
through the dispersive soil profile. Ground strains were related to displacement gradients based on theoretical solutions of harmonic body
waves and Rayleigh waves in dispersive elastic media. The peak estimated ground strains were smaller than the threshold volumetric shear
strain, but a few centimeters of settlement were nevertheless observed at the site. The spatial extent of the settlement is characterized using
attenuation rules fit to the vibration data, and by calibration with a settlement gauge. Ground cracking and vertical offsets that could
potentially mask the archaeological history of the site were neither observed nor predicted from the observed vibration amplitudes.
Estimated impact on archeological interpretation of artifacts in their stratigraphic context was likely insignificant except in the immediate
region where the piles were driven. This insight will assist in future planning at sites with similar subsurface stratigraphy.
DOI: 10.1061/共ASCE兲GT.1943-5606.0000031
CE Database subject headings: Vibration; Pile driving; Archaeology; Wave measurement; Foundation construction.
Archaeological Background
Geotechnical Background
Caltrans determined that construction operations required to
widen I-5 near Lathrop, Calif., would impact a National-Register- Ground vibrations are commonly monitored due to potential im-
eligible prehistoric occupation site 共CA-SJO-003 or the Mossdale pacts on sensitive structures and equipment, and annoyance to
site兲. The Mossdale site occupies approximately 13 ha within the people nearby 共Ju et al. 2007; Svinkin 2002; Hwang et al. 2001兲.
flood plain of the San Joaquin River. Radiocarbon and obsidian Relations between vibration amplitudes and impacts are important
hydration dating on samples taken from the site indicate occupa- for establishing tolerable vibration limits for a given vibration
tion at the site ranged from 155 to 1,500 years ago at a time when source. For example, the Caltrans vibration manual 共Jones and
the North Valley Yokuts are believed to have inhabited the area. Stokes 2004兲 distinguishes between transient vibration sources
The site contains varied artifact assemblages that are characteris- and continuous sources 共such as pile driving兲, and specifies a
tic of prehistoric habitation: a significant amount of faunal re- threshold particle velocity of 2 mm/ s from continuous sources for
mains such as bone and shell, burned bone, charcoal, impressed fragile historical structures, and specifies that particle velocities
clay, clay pipes, bone awls, shell beads, pendants, ground stone smaller than about 0.3 mm/ s from continuous sources are barely
pestles and mortar fragments, obsidian, and chert lithics. Human perceptible. Vibrations that exceed threshold values can be miti-
remains have also been disinterred from the site in the past. gated using wave barriers 共Comina and Foti 2007; Hung et al.
Construction operations at the Mossdale site were governed by 2004兲.
environmental laws in place to protect cultural resources, includ- Whereas the influence of vibrations on nearby equipment and
ing National Environmental Policy Act, Section 106 of the Na- structures has been extensively studied, vibration-induced soil
tional Historic Preservation Act, and the California settlement has received less attention. Vibrations induce shear
Environmental Quality Act. A “reasonable and good faith effort” strains in soil, which can cause cohesionless soil particles to slip
must be made to identify cultural resources that may be affected past one another as the soil becomes denser thereby inducing
by an undertaking, evaluate the significance of each resource, and settlement. Dowding 共1996兲 summarized several case histories
determine how the undertaking will impact the resource. Signifi- where pile driving caused ground settlements in sandy soil depos-
cant impacts must be mitigated. Impacts can either be direct 共e.g., its. Pile driving vibrations with peak particle velocities of
excavation of artifacts at the site兲 or indirect 共e.g., settlement 2.5– 5.0 mm/ s caused the soil near the pile driving operations to
induced by nearby construction vibrations兲. Direct impacts are settle as much as 75 mm. Drabkin et al. 共1996兲 summarized case
easy to identify, but indirect impacts are often subjective. histories where vibration amplitudes of 1.0– 18.0 mm/ s were re-
Two basic principles help archaeologists interpret the artifacts sponsible for settlements of 8 – 250 mm. Stewart et al. 共2001兲
and features they discover. The Law of Association states that documented seismic performance of hillside fills during the 1994
artifacts found together result from a common set of human ac- Northridge earthquake, and compiled 250 cases where structures
tions, and should be interpreted together to tell a story. The Law were damaged by large settlements and ground cracks with ver-
of Superposition states that young strata are deposited over old tical offsets approaching 100 mm. They predicted particle veloci-
strata, which is the guiding principle of stratigraphy. Stratigraphy ties of hundreds of mm/s were responsible for the observed
is the single best method that archaeologists have for determining settlements. The case studies from construction vibrations and
relative ages of archaeological materials, and artifacts outside of from earthquakes help to bracket the nature of the settlement that
their stratigraphic context have little value. Impacts that alter soil can be anticipated for different strain levels. The cases docu-
stratigraphy, whether direct or indirect, are therefore significant mented by Dowding 共1996兲 and Drabkin et al. 共1996兲 were asso-
and must be mitigated. Past measures to mitigate indirect impacts ciated with smoothly varying settlements and no extensive
have, for example, involved placing fill on top of cultural re- cracking, while the cases documented by Stewart et al. 共2001兲
sources during construction operations to prevent direct contact were associated with ground cracking and large differential settle-
with construction equipment. However, it is not clear the extent to ments. Differential settlements like those observed by Stewart et
which this mitigation approach is effective, and the Federal High- al. 共2001兲 could potentially distort a site’s stratigraphy and mask
way Administration recently stated that an undertaking’s effects its archaeological history.
冉 冊
which is widely adopted as a general rule of thumb for sand.
However, settlements have been observed at sites where ground 2qk2 −sz
uz = A e − qe−qz ei共t−ky兲 共5兲
strains are lower than 0.01%, indicating that perhaps a large num- s2 + k2
ber of loading cycles can cause settlements even when ground
strains are smaller than ␥tv, or that ␥tv may be lower at frequen-
冉 2iqsk −sz
冊
Downloaded from ascelibrary.org by Tokyo Univ Seisan Gijutsu on 05/14/15. Copyright ASCE. For personal use only; all rights reserved.
cies that are higher than tested in their laboratory program. Mas- uy = A e − ike−qz ei共t−ky兲 共6兲
s2 + k2
sarch 共2000兲 presents a guideline wherein strains below 0.001%
would be associated with very low risk of settlement, and strains The strain components yy, zz, ␥yz can be computed by solving
larger than 0.01% would be associated with a high risk of settle- the appropriate partial derivatives
ment. These guidelines reasonably capture the case history obser-
vations. Drabkin et al. 共1996兲 present an equation for predicting
settlement based on soil type, vibration amplitude, static shear
zz =
uz
z
冉2qsk2
冊
= A − 2 2 e−sz + q2e−qz ei共t−ky兲
s +k
共7兲
冉 冊
stress, confining stress, and number of loading cycles based on an
extensive laboratory test program. The equation reasonably pre- u y 2qsk2
yy = = A 2 2 e−sz − k2e−qz ei共t−ky兲 共8兲
dicted settlements from case histories. Shear strain is easy to mea- y s +k
sure in laboratory experiments, but difficult to measure in the
field, whereas particle velocity is easy to measure in the field and
is therefore often used as a proxy for shear strain. ␥yz = 冉 uz u y
y
+
z
冊 冉
2qsk2
= 2A 2 2 e−sz −
s +k
k2 + q2 −qz i共t−k y兲
2
e e R 冊
共9兲
Relations between Shear Strain and Particle The peak shear strain, ␥max, taken as the radius of the Mohr circle
Velocity of strain, can then be computed using Eq. 共10兲
It is convenient now to devise relations between particle velocity ␥max = 冑共zz − yy兲2 + ␥2yz 共10兲
and shear strain before presenting results from the experimental
study in this paper. A common rule of thumb is that shear strain is The vertical velocity can be computed by differentiating Eq. 共5兲
equal to particle velocity divided by wave speed. This relation with respect to t
was derived for harmonic shear waves propagating through a ho-
mogeneous elastic medium, as shown in Eqs. 共1兲–共4兲, where u
= displacement; A = displacement amplitude; = angular
u̇z =
w
t
冉
2qk2
冊
= iA 2 2 e−sz − qe−qz ei共t−ky兲
s +k
共11兲
frequency= 2f; t = time; x = position; k = wave number 共k = / vs兲; The relation between ␥max and u̇z depends on Poisson ratio, ,
vs = shear wave velocity; and u̇ = particle velocity because the relation between Rayleigh wave velocity and s- and
p-wave velocity depends on . However, the displacement gradi-
u共x,t兲 = A cos共t − kx兲 共1兲
ent, uz / dy, is equal to u̇z / vR, which can be verified by taking the
partial derivative of Eq. 共5兲 with respect to y.
u
u̇ = = A cos共t − kx兲 共2兲 Fig. 2 presents the ratio of peak engineering shear strain 共␥max兲
t to the ratio of peak vertical velocity at the ground surface divided
by Rayleigh wave velocity 关u̇z共z = 0兲 / vR兴 as a function of z / ,
u where wavelength = 2vR / . The ratio is near 1.0 at the ground
␥= = − kA cos共t − kx兲 共3兲
x surface and decreases quickly to a depth of about 0.06 – 0.07,
and subsequently increases again with depth before attenuating.
u̇ u̇ u̇ Niemunis 共1995兲 formulated the equations for relating engineer-
␥= ·k= · = 共4兲 ing shear strain to vertical velocity at the ground surface for Ray-
vs vs
leigh waves. However, Fig. 2 is different than the relation he
While Eq. 共4兲 provides a simple and useful relation between par- presented due to an error in his implementation 共Niemunis, per-
ticle velocity and shear strain, it is limited to harmonic shear sonal communication兲, and engineering shear strain near the
waves in homogeneous media. Errors arise when waveforms are ground surface is actually larger than implied in his paper. As a
more complex than harmonic body waves and/or the medium is general rule of thumb, peak vertical velocity divided by Rayleigh
heterogeneous. In the field, vibrations are often measured on the wave velocity will provide accurate shear strain estimates at the
ground surface where the ground motion is characterized by a ground surface, and will be within about 0.3 to 1.1 times the true
combination of body waves and surface waves, and at heteroge- strain value within the upper 0.6 for planar harmonic Rayleigh
neous sites where a constant wave speed may not adequately waves in elastic homogeneous soil.
characterize the dynamic properties of the soil. Layered soil profiles exhibit frequency-dependent wave speed
Vibration sources 共e.g., pile driving兲 are often characterized by 共and are therefore said to be dispersive兲 because waves of differ-
body waves within the soil near the source, and primarily by ent wavelength mobilize different combinations of soil layers. To
Site Investigation
Fig. 4. Site plan showing measurement locations G1, G2, and G3 and positions of CPT soundings in relation to pile driving operations
Fig. 5. CPT soundings recorded at positions 7, 9, 10, 10.5, and 16 共see Fig. 4兲
tial position rather than characterizing horizontal trends across the higher frequency vibrations whose wavelength was on the same
site 共as is a more typical objective for CPT site investigations兲. order of size as the cavity 共Razin 1995兲.
The soil profile consists of approximately 1 – 3 m of gravelly sand The shear wave velocity profile at the site was measured using
fill overlying alluvial silty sand to sandy silt with lenses of clay. the geophone embedded in the CPT probe 关i.e., using the, seismic
Many of the soundings resulted in refusal in the fill layer, and cone penetration test 共SCPT兲 method兴 at 0.25 m intervals for
significant amounts of coarse gravel were encountered during depths from 0.75 to 6.5 m. Wave travel time measured at the first
hand augering operations. The groundwater table was not encoun- peak of the waveforms was plotted versus depth and shear wave
tered in the deepest CPT sounding so pore-water pressure read- velocity was computed as the slope of travel distance versus
ings are not presented. The fill layer is highly heterogeneous with travel time 共Fig. 6兲. The profile was divided into three distinct
CPT tip resistance varying sharply over short distances horizon- regions based on the slope of travel time versus depth, and the
tally and vertically. The CPT records in Fig. 5 were separated by average shear wave velocities for the regions, from shallowest to
as little as 0.5 m, yet exhibit significantly different tip resistance deepest, were 135, 175, and 250 m / s. The shear wave velocity
and sleeve friction due to the large heterogeneity at the site. The profile increases with depth, which is a typical effect caused by
underlying alluvium was more uniform than the fill, but still ex-
hibited significant scatter in penetration resistance. The heteroge-
neity of these soil layers complicates the ability to compare
before and after CPT soundings to estimate settlement induced by
pile driving operations because it is impossible to discern whether
differences in records taken in close proximity to each other were
caused by soil settlement or by natural variability in the soil. A
region of very low tip resistance was encountered between depths
of 2 – 5 m for soundings 7 and 16, indicating the presence of a
very soft layer or a cavity. However, the cavity was not encoun-
tered in any other soundings, indicating that its lateral extent must
be on the order of 1 m or less. The presence of a very soft zone or
cavity was subsequently verified during hand augering operations
共the hand auger could be pushed through the soil by hand without
rotating the cutting teeth兲 at position G2 in Fig. 2, but not at
positions G1 or G3. The cavity may have influenced vibrations at
the site by affecting modes of wave propagation, particularly for Fig. 6. Shear wave velocity measurements using SCPT equipment
3.7, and 4.6 m as the augered holes were advanced at those loca- vertical direction 共see Fig. 7兲. The recorded data show that blows
tions. When pile driving operations moved to pier 2 the geophone from the pile driving hammer occurred approximately at 1.5 s
was removed from hole G1 and placed into hole G3, and vibra- intervals, and generated pulses of velocity that lasted approxi-
tions were measured between G2 and G3 at a depth of 4.6 m. mately 0.1 s. The recorded velocity pulses are similar for each
Measurements at other depths were not made between positions pile blow within a given record, though the amplitudes of the
G2 and G3 due to time constraints. For all cases, surface accel- velocities were observed to vary with depth and with distance
erometers were shallowly buried near the location of the holes in from pile driving. Table 3 summarizes the average peak velocity
which the geophones were installed, and surface vibration data for the different depths and distances tested. The peak velocities
were recorded simultaneously with the geophone data. exceed the values of 2.5– 5.0 mm/ s that have been associated
Figs. 9 and 10 show velocity records collected at a depth of with settlement in past case histories, hence some settlement
4.6 m during pile driving at distances of 40 m 共Fig. 9兲 and 10 m could be anticipated to accompany the pile driving vibrations.
共Fig. 10兲. A pair of figures is shown for each record. The first of Frequency content of the vibrations is an important consider-
the pair shows a sequence of six pile driving vibrations illustrat- ation for wave propagation because a dispersion curve can only
ing the repeatability of the vibration measurements, and the sec- be computed accurately on frequency bands with significant data
ond shows a zoomed in view of only one hit to show the signal content, and because wavelength is a function of frequency and
quality in more detail. Records from both geophones are shown in wave velocity. Fig. 11 shows Fourier spectra for the three com-
the zoomed in view to show any difference in the recorded vibra- ponents of ground velocity recorded at the ground surface and at
tions for adjacent sensor locations. These differences are impor- a depth of 4.6 m for a single pile driving strike at a distance of
tant because they form the basis for computing a dispersion curve 40 m from the sensors, and a single pile driving strike at a dis-
and for computing shear strains, as discussed in the next section. tance of 10 m from the sensors. The spectra indicate that the
The coordinate system is defined based on a line running through frequency content is distributed over a band lying primarily be-
the two sensors as follows: y- is parallel to that line, x- is perpen- tween 5 and 70 Hz. The amplitudes of the spectra are larger at a
dicular to the line in the horizontal direction, and z- is in the distance of 10 m than at 40 m because waves attenuate with dis-
Fig. 9. Velocity records at depth of 4.6 m and distance of ⬃40 m from pile driving
Fig. 10. Velocity records at depth of 4.6 m and distance of ⬃10 m from pile driving
tance. The sharp cutoff at 80 Hz is due to the low-pass antialias- 2005; Brandenberg et al. 2008兲. The phase, wave velocity, and
ing filters in the Q330 data loggers. Data on frequency bands wavelength are plotted versus frequency in Fig. 12 for sensors
higher than 70 Hz were verified to be lowamplitude based on embedded 4.6 m in the ground with pile driving at a distance of
measurements with a different data logger with higher sampling 10 m. The soil at this site was dispersive since the wave velocity
rate and no antialiasing filter. varies with frequency. The very large and sometimes negative
wave velocities below about 3 Hz are the result of low signal-to-
noise ratio at these low frequencies, and such errors were also
Calculation of Dispersion Curve observed above about 70 Hz 共not shown in Fig. 12兲. These errors
did not adversely affect the calculation of displacement gradients
A dispersion curve relates wave velocity to frequency, and can be
computed by measuring phase lag between two adjacent vibration
records. The relation between wave velocity and frequency is a
function of the soil profile, and dispersion curves form the basis
for spectral analysis of surface waves 共Stokoe et al. 1994兲. The
dispersion curve was computed based on the phase of the cross
correlation of adjacent vertical acceleration recordings. Calcula-
tions were performed in the frequency domain using the Fourier
transform of the signals as outlined in Fig. 12. The signal-to-noise
ratio was improved by averaging the Fourier spectra from a se-
quence of 300 pile driving hits prior to computing the dispersion
curve for each distance-to-pile-driving/depth combination. Signal
stacking improves signal-to-noise ratio because the noise is pre-
sumably random, while the desired portion of the signal is well
correlated among multiple measurements 共Santamarina and Fratta
Table 3. Average Peak Strains and Particle Velocities from Recorded Data
Average peak displacement gradient Average peak velocity
of settlement by Massarch 共2000兲. Hwang, J.-H., Liang, N., and Chen, C.-H. 共2001兲. “Ground response dur-
Mitigation of direct impacts in the immediate vicinity of the ing pile driving.” J. Geotech. Geoenviron. Eng., 127共11兲, 939–949.
pile groups, required by Section 106 of the National Historic Jones and Stokes 共2004兲. Transportation- and construction-induced vi-
Preservation Act, was provided by data recovery of representative bration guidance manual 共J&S 02-039兲, California Department of
deposits conducted prior to pile driving. In addition, data suggest Transportation, Noise, Vibration, and Hazardous Waste Management
that differential settlement in the form of ground cracks and ver- Office, Sacramento, Calif.
tical offsets at this site was unlikely. Rather, the settlement pattern Ju, S.-H., Lin, H.-T., and Chen, T.-K. 共2007兲. “Studying characteristics of
varied smoothly with distance from pile driving. Therefore, the train-induced ground vibrations adjacent to an elevated railway by
archeological interpretation of artifacts in their stratigraphic and field experiments.” J. Geotech. Geoenviron. Eng., 133共10兲, 1302–
spatial context was probably affected only due to the direct im- 1307.
Massarch, K. R. 共2000兲. “Settlements and damage caused by
pacts in the immediate region where the pile foundations were
construction-induced vibrations.” Proc., Int. Workshop Wave 2000,
constructed.
Bochum, Germany, N. Chouw and D. Schmid, eds., Taylor and Fran-
The data presented in this paper demonstrate a valuable
cis, London, 299–315.
method for quantifying and documenting indirect impacts to in-
Niemunis, A. 共1995兲. “On the estimation of the amplitude of shear strain
tact archaeological deposits. Previous methods for mitigating in- from measurements in situ.” Soil Dyn. Earthquake Eng., 14, 1–3.
direct impacts, such as placing fill over the site, provided Razin, A. V. 共1995兲. “Elastic wave propagation in a randomly stratified
improvements that were difficult to quantify and record. Data solid medium.” Waves Random Media, 5, 137–143.
from this study have already been used in projects with similar Santamarina, J. C., and Fratta, D. 共2005兲. Discrete signals and inverse
subsurface stratigraphy to identify and plan for indirect effects on problems: an introduction for engineers and scientists, Wiley, Chich-
other cultural resources. ester, U.K.
Santamarina, J. C., Klein, K. A., and Fam, M. A. 共2001兲. Soils and
waves: Particulate materials behavior, characterization and process
Acknowledgments monitoring, Wiley, Chicester, U.K.
Stewart, J. P., Bray, J. D., McMahon, D. J., Smith, P. M., and Kropp, A.
Resources used in this study were from the National Science L. 共2001兲. “Seismic performance of hillside fills.” J. Geotech. Geoen-
Foundation-funded George E. Brown Jr. Network for Earthquake viron. Eng., 127共11兲, 905–919.
Engineering Simulation 共NEES兲. Assistance by the nees@UCLA Stokoe, K. H., II, Wright, S. G., Bay, J. A., and Roesset, J. M. 共1994兲.
staff, particularly Alberto Salamanca, Steve Keowen, and Steve “Characterization of geotechnical sites by SASW method.” Proc.,
Kang, is gratefully acknowledged. Funding for this study was 13th ISSMFE Technical Committee 10 for ICSMFE, Geophysical
provided by Caltrans through a subcontract to Far Western An- Characteristics of Sites, Balkema, Rotterdam & Brookfield, The Neth-
thropological Research Group. erlands, 785–816.
Svinkin, M. R. 共2002兲. “Predicting soil and structure vibrations from
impact machines.” J. Geotech. Geoenviron. Eng., 128共7兲, 602–612.
References Trifunac, M. D., Todorovska, M. I., and Ivanovic, S. S. 共1996兲. “Peak
velocities and peak surface strains during Northridge, California,
Brandenberg, S. J., Kutter, B. L., and Wilson, D. W. 共2008兲. “Fast stack- earthquake of 17 January 1994.” Soil Dyn. Earthquake Eng., 15, 301–
ing and signal processing for shear wave velocity measurement in a 310.