Characterization of Gravelly Alluvium
Characterization of Gravelly Alluvium
Characterization of Gravelly Alluvium
ABSTRACT
Characterizing project sites where gravelly alluvium is present poses particular challenges.
Conventional field and laboratory techniques are significantly limited when large gravel particles
either prevent their operation or adversely influence their measurements. The deposition of
gravelly alluvium often involves complex, energy and sediment load dependent processes that can
result in highly spatially variable deposits comprised of gravel to clay sized particles. This
pervasive variability warrants a comprehensive site characterization approach that begins with
developing a thorough understanding of the geological formation processes and the expected
controlling deformation/failure mechanisms of the geotechnical system. In this context, the site
investigation becomes a confirmatory process, allowing the subsequent idealization of site
conditions for analysis and design to have a firm basis in geology. Recent work at several
gravelly alluvium sites has allowed the development of new approaches and techniques for
improved characterization. In particular, the benefit of continuous sonic sampling is demonstrated
and a systematic method for determining gravel influence on SPT N values is outlined. Finally,
the instrumented Becker Penetration Test (iBPT), a new closed-ended, large diameter
penetrometer with direct energy measurement at the drill string tip is presented.
Introduction
Characterizing project sites where gravelly alluvium is present poses particular challenges.
Despite these challenges, documented liquefaction of gravelly alluvium in historical earthquake
events (e.g. Harder, 1994; and Cao et al., 2013) has demonstrated the importance of reliable
liquefaction assessment for these materials. Additional liquefaction case history sites, including
many in Christchurch, NZ (Cubrinovski et al. 2011a,b), have involved liquefaction of gravel-
free, sand-like layers deposited directly above or below such alluvial gravel layers. As a result,
proper characterization of gravelly alluvium is a persistent need when assessing the liquefaction
potential of a site. The need for characterization extends beyond liquefaction assessment, and
includes static design of deep foundations, tunnels, bridges, levees, dams, and many other
structures founded upon alluvial deposits.
Deposition of gravelly alluvium is a complex process that can result in highly interlayered
deposits comprised of gravel to clay sized particles from varying geologic source materials. In
broad valleys these alluvial gravels are commonly deposited within a braided river architecture,
1
Professor, Civil and Env. Engineering, University of California Davis, Davis, USA, jdejong@ucdavis.edu
2
Doctoral Researcher, Civil and Env. Engineering, University of California Davis, Davis, USA,
apsturm@ucdavis.edu
3
Assistant Professor, Civil Engineering, University of Toronto, Toronto, Canada, mason.ghafghazi@utoronto.ca
an example of which is presented in Figure 1a (Christchurch, NZ). Aerial views capture the
current river structure while the coloring and vegetation of the bounding lands provide insight
regarding historical depositional mechanisms and the lateral extent of river migration. During
high river flow a high-energy depositional environment is present, allowing larger gravel
particles to be transported downstream. Subsequent deposition of gravel particles is an energy
and sediment load dependent process, with deposition occurring as the energy and/or sediment
load decreases. As a result, the deposition of gravel layers is typically concurrent with high water
levels associated with flash flooding and spring snow melt.
Over time, the highly variable depositional process results in subsurface conditions similar to
that sketched in Figure 1b. The braided river architecture consists of a series migrating gravel
layers (commonly referred to as bars when still exposed). Near the end of a depositional cycle,
low-energy flow conditions allow finer sand, silt, and clay size particles to infill the exposed
void space in the gravel layers. This process often results in high fines contents near the top of
gravel layers. Exceptionally high flow events that cause overtopping of the existing channel
results in the formation of overbank deposits. The rapid decrease in energy in this depositional
environment leads to larger particles being deposited close to the existing channel while finer
grained sediments may fall out of suspension in pools of trapped water further away.
The deposition of gravelly alluvium in narrower valleys and canyons, where dams are typically
constructed, generally consist of similar depositional architecture, albeit with additional
complications. The extent of lateral river migration is typically reduced due to steep valley walls,
which provide geographic bounds, and a larger river gradient (slope), which are both inversely
related to the extent of river meander. The water flow level may also be more extreme, possibly
ranging from valley-wide flooding at its highest, to no flow after extended periods of drought. In
addition, colluvium (debris from slope instability of the valley walls) can be deposited at the
valley edges; the colluvium may further constrain river migration and be mixed with the
sediment being transported from upstream.
(b)
(a)
Figure 1 – (a) Arial image of braided gravel river deposit west of Christchurch, NZ and (b)
depositional architecture of braided river deposit (from Nicols 2009).
Spatial Variability
The actual complexity of a natural braided alluvial deposit is more variable and chaotic than
conceptually indicated above. Figure 2a and b present a photograph and a geologic map,
respectively, from a 9 m deep excavation within a braided river deposit, cut perpendicular to the
historic river alignment. Large gravel/cobble sized particles exist in lenses that extend up to 1 m
in thickness and 5 m laterally (their continuity parallel to the river alignment is unknown, but
likely greater). These lenses are interspersed within sandy layers/pockets, which contain up to
30% fine to medium gravel, as well as more continuous layers of silty sand. The extents of most
layers are variable and fingering, with abrupt transitions between some layers and gradual
transitions between others. Collectively, these layers reflect the changing deposition, migration,
and energy of the river, with similar lenses reflecting several realizations of a depositional
process across time and space.
The composition of gravelly alluvium is highly variable within a given project site and between
different project sites. Expectedly, its composition also differs significantly from more uniform
natural materials and artificial fills. Figure 3a presents the percentages of gravel, sand, and fines
present in SPT samples obtained at several different project sites using ternary plots. Site 2, for
example, is a relatively clean sand site with largely less than 20% gravel or fines (as discussed in
more detail later, the SPT test could be expected to perform well at this site). Site 1, in contrast,
includes a substantial number of the samples that contain significant gravel (up to 50%),
indicating that conventional methods for field characterization will not be applicable.
Collectively, the alluvium differs significantly from the natural soft clay deposit and hydraulic
fill materials that have a noticeably small fraction of gravel and significant fines content (Figure
3b). The compacted embankment material, which exists on the downstream side of a dam,
contains a higher percentage of gravel but also a modest percentage of (plastic) fines for
effective compaction and low permeability.
The natural spatial variability of gravelly alluvial deposits makes it difficult to compare and
evaluate the applicability of different field techniques for characterization of a given deposit.
Consider two borings/soundings performed at 4 m spacing in the alluvial deposit mapped in
(a) (b)
Figure 2. (a) photograph and (b) geologic map of 9 m deep excavation within a braided river
deposit cut perpendicular to the historic river alignment (Permission from LADWP. Photograph
and geologic map provided by Amec Foster Wheeler).
Figure 2. At nearly any depth it is unlikely that the exact same material will be encountered. As
a result, it can be very difficult to assess the suitability of one technique relative to another as the
two techniques may encounter different soils at the same exact depth. However, by recognizing
that the same geologic depositional processes formed the subsurface of interest, it may be
possible to identify larger depth intervals where both the materials encountered and penetration
resistances measured, are consistent; therefore, the median penetration resistances between the
two tests may be compared over these larger, consistent intervals. This approach has been
outlined by Ghafghazi et al. (2016) and used successfully in selecting representative soil property
values for liquefaction potential analysis.
The extent of spatial variability that exists within alluvial deposits is often greater than that
recognized by engineers. This is exemplified using two cone penetrometer soundings performed
in an alluvial sand deposit (gravel influence not of concern) at 4 m spacing. As evident in Figure
4a, the two cone tip resistance profiles are relatively similar to each other, generally following
the same trends and containing characteristics typically observed in sand deposits. Presentation
of the same data plotting the pair-wise data points at each measurement depth (Figure 4b) and
then averaged over 0.3m depth intervals (Figure 4c) reveals a significant amount of variability,
with logarithmic coefficient of variation (COV) values of 0.40 and 0.33, respectively. As
expected, averaging of data over 0.3 m (1 ft) depth intervals reduces the COV values.
Nonetheless, it is evident that at a lateral spacing of 4 m the penetration resistance can be
expected to be up to ±40% the value measured at the reference location. This observation
extends to, and has been confirmed in, gravelly alluvium as well; variations in penetration
resistance of up to ±40% may be expected across lateral distances of 4 m.
0 100 0 100
20 80 20 80
40 60 40 60
60 40 60 40
80 20 80 20
100 0 100 0
0 20 40 60 80 100 0 20 40 60 80 100
(a) Gravel (%) (b) Gravel (%)
Figure 3. Ternary plots indicating the percent gravel, sand, and fines in (a) gravelly alluvium at
four different project sites as well as for (b) uniform clay, hydraulic fill, and compacted dam
embankment soils.
CPT qt (MPa) 50
0 10 20 30 40 50
0 40
CPT qt (MPa)
2 30
20
4
10
6
(b) 0
8
Depth (ft)
50
10
40
CPT qt (MPa)
12
30
14
20
16
10
18 (c) 0
0 10 20 30 40 50
20 CPT qt (MPa)
(a)
Figure 4. (a) Example of two cone penetrometer sounding profiles performed in a sandy deposit
at a lateral spacing of 4 m. Reprocessing of the data by plotting depth specific data pairs from
the two soundings for (b) each measurement depth (5 cm increments) and (c) averaged over 0.3m
intervals to illustrate spatial variability and the effect of averaging.
The spatial variability of gravelly alluvium, primarily due to its highly variable energy and
sediment load dependent depositional process, presents challenges not often encountered when
characterizing soils deposited by more stable and widespread processes (e.g. marine clay
deposits). Implementing conventional site characterization practices typically employed in
uniform materials, for example performing a few cone profiles at large spacing across the site,
will be less effective in gravelly alluvium. In fact, this approach may provide unique soil profiles
at every location tested with little or no clear correlation between these different soundings.
Understanding site conditions, in this case, can only be achieved by first developing a site-
specific geologic understanding and determining how the geotechnical system being
analyzed/designed interacts with the geologic deposit.
A systematic methodology for site characterization of gravelly alluvium, or any project with
complex subsurface conditions, is outlined in Figure 5. It contains the primary five steps of
Inductive Reasoning, Scenario Assessment, Site Investigation, Site Idealization, and Analysis &
Design. The steps of Inductive Reasoning and Scenario Assessment are focused on developing
and preliminarily evaluating hypotheses of expected subsurface conditions and controlling
performance mechanisms. The Site Investigation step utilizes techniques to develop a database
while the Site Idealization step is focused on capturing the controlling factors in a simplified
representation that can later be used in detailed analysis. The final Analysis & Design step
consists of a detailed study, eventually leading to project completion.
While the methodology outlined here builds on prior recommendations (e.g. Clayton 1995,
USACE 2001, Mayne et al. 2004, Clayton and Smith 2013), emphasis is placed on the extensive
and detailed analysis of the site project prior to mobilization of equipment for in situ testing, and
drilling and sampling for complimentary laboratory testing. In particular, Site Investigation is
only a single step in the much larger site characterization process, a step that often begins
prematurely, prior to developing a clearly articulated hypothesis of expected site conditions. The
Site Investigation step also often incurs the largest expense of the geotechnical project budget;
delaying the Site Investigation step until after the first two steps are completed may
simultaneously serve to improve characterization and reduce costs by refining the testing efforts
and reallocating a portion of the funds towards more detailed analysis and interpretation.
Site Investigation
Perform site
Identify performance investigation using
mechanisms in-situ & lab tools
Inductive Reasoning
Understand geology
Site Idealization
Assess spatial variability
Develop working & stratigraphic continuity
hypotheses of all
possible mechanisms, Sub-divide into critical
layers, spatial variability zones & assign
& properties representative values
Analysis & Design
Scenario Assessment
Completed
The site characterization process begins by outlining the required performance of the
geotechnical system being analyzed, the geologic setting that the project is being constructed
upon/through, and how the length scales of these compare. The controlling performance
mechanisms are often capacity or deformation based, and their requirements often differ for
loading conditions (static, dynamic) and duration (short term, long term). Once identified, the
potential deformation/failure zones for each mechanism can be estimated, revealing the
mechanism scales and subsurface characteristics that are relevant for the investigation.
The geologic formation processes (as well as subsequent anthropogenic activities) that have led
to the current site conditions dictate system performance. Therefore, the geology must be
understood in detail by leveraging reference sources (e.g. geology reports, aerial/satellite photos,
adjacent geotechnical project documents, etc.) and working closely with local geologists. This
enables estimation of the number, thickness, variability, and soil properties of each of the
geologic strata likely to exist at the project site. It is important to recognize that geotechnical
engineers are typically over-optimistic about the uniformity of soil properties (Duncan 2000).
Therefore, when estimating variability, it is useful to consult values obtained in prior studies (e.g.
Phoon et al. 1995, Phoon and Kulhawy 1999a, 1999b, Baecher and Christian 2003) and to
employ systematic, quantitative methods for estimating variability to the extent possible (e.g.
Duncan 2000).
This collective effort enables development of a set of working hypotheses of all possible
mechanisms that will govern performance, including identification of the particular strata that are
engaged in and influence these mechanisms. Assessing the contribution of particular strata to
different mechanisms requires hypotheses on the expected thicknesses, lateral continuity,
material variability, and soil properties.
Scenario Assessment
The suite of hypotheses, when defined in sufficient detail, can then be analyzed to assess which
scenarios will ultimately control performance. Sensitivity studies, where controlling
conditions/parameters are systematically varied across ranges of uncertainty (based on geology
and/or literature values), can help identify which aspects are controlling performance and how
reducing uncertainty in these areas can improve the estimation of performance. In many respects
this approach has similarities to the Potential Failure Modes Analysis (PFMA) often conducted
for dams (USDOI 2012, USACE 2011). For smaller projects, this stage may consist of a
relatively straight forward spreadsheet analysis, while for larger projects this stage may entail a
suite of finite element simulations.
The outcome of the Scenario Assessment step includes a geologically-based hypothesis of the
subsurface conditions that exist at the site (including major units, vertical and lateral variability,
soil property ranges), a detailed understanding of the primary mechanisms controlling
performance, and knowledge of which conditions/parameters control the governing mechanisms.
When understanding is sufficiently developed at this step the subsequent steps (Site
Investigation, Site Idealization) become a process of confirmation instead of discovery. It is not
that revisions to the developed hypotheses will not be necessary; on the contrary, they may be
expected as more site-specific information is gathered and further analysis is performed. This
revision process, however, corrects and refines the understanding gained during hypotheses
development instead of being the first detailed examination of the site.
Site Investigation
Site investigation, encompassing in-situ testing, drilling and sampling (with complimentary
laboratory testing), is the middle step in the overall site characterization process. It focuses on
obtaining an experimental, site-specific database of conditions/parameters expected to control
system performance and design. The methods and tools employed in this step are extensively
described in the geotechnical literature, with books, manuals, and publications detailing many
different techniques available (e.g. Clayton 1995, USACE 2001, Mayne et al. 2004, Clayton and
Smith 2013).
The primary challenge in Site Investigation is selecting the most appropriate techniques and
determining where they should be performed at the site. The extent of expected spatial variability
will guide the balance between the quantity versus quality of in-situ and laboratory tests
performed; as the site hypotheses are confirmed/refined, the number and type of tests can be
adjusted to optimize collection of critical data. In addition, in-situ and laboratory tests should
follow stress paths similar to those induced by the expected deformation/failure mechanisms
(e.g. Kulhawy and Mayne 1990) to the extent possible, and site specific factors for in-situ test
correlations (e.g. CPT based estimate of Su) should be calibrated to appropriate shearing modes.
The Site Investigation step concludes by compiling the data collected, including interpretation of
measured data to estimate engineering properties, and separation of data by geologic strata
and/or similar soil properties. Normalization of properties with respect to overburden stress, for
example by using normalized strength ratios (Su /σv’) and normalized penetration resistance
values ((N1)60 and qc1N), are particularly useful when compiling data across borings/soundings for
a particular stratum.
Site Idealization
The Analysis & Design of a project requires, as input, idealization of project site conditions into
simplified representations. Site Idealization requires reduction of the site investigation database
into representative parameters that appropriately account for the range in properties measured
and the connectivity of the layers (or weaker zones within a layer). Binning of data from multiple
borings/soundings into geologic strata and sub-intervals where properties (rather than materials)
differ and then plotting cumulative distributions of overburden normalized parameters is
effective in selecting representative percentiles and bounding (e.g. ±1σ) values for subsequent
analysis.
It is recommended that the Site Idealization step, to the extent possible, be separated from the
Site Characterization step in order to explicitly and systematically document each decision as an
idealized representation (representative values, layer thicknesses, ground water table elevations)
is formed. Each step of idealization may build conservatism in the subsequent analysis.
Conservatism is built in, for example, when the 30th percentile value of (N1)60 is selected to be
the representative value for a uniform, liquefiable layer underneath a dam; however, this may be
representative if the layer’s (N1)60 values are highly variable (see Boulanger and Montgomery
2015). The series of decisions made during site idealization can then result in compounding
conservative or unconservative decisions that may not otherwise be clearly conveyed and
documented.
The idealized site conditions, with upper/lower bound parameters for sensitivity studies, form the
primary inputs for analysis and design. The methods employed at this stage are often more
sophisticated and advanced, but do build on, those used during the earlier Scenario Assessment
step. During detailed design and analysis, it may be necessary to revisit the Site Idealization
phase in order to redefine the stratigraphic units and their representative properties or to evaluate
the effects of uncertainty by analyzing multiple realizations of the idealized site conditions. In
some cases, it may be necessary to expand the Site Investigation step, either by performing
additional laboratory tests on samples already obtained or re-mobilizing equipment to site.
Eventually the analysis and design is completed, and hence the overall process of site
characterization is complete.
Summary
Recent work on a series of projects has enabled the development of alternate strategies for
evaluating the potential influence of gravel on different conventional methods and advances in
technology have enabled the development of a new continuous penetration technique in gravelly
alluvium.
The sonic drilling technique is useful when characterizing gravelly alluvium as it provides rapid,
disturbed, continuous core samples, essentially bringing an uninterrupted stratigraphic sample to
the surface for the geologists and geotechnical engineers to inspect (Figure 6). This enables
approximate delineation of the stratigraphic layers, identification of layers where gravel is
present, information to guide selection of in situ and laboratory testing techniques, and disturbed
bulk samples without significant scalping of coarse fraction materials for laboratory testing.
In North America, the “core and case” sonic drilling technique is typically employed to provide a
continuous sample; its use in the characterization of gravels has increased in recent years though
it was developed more than 40 years ago in Canada. The hydraulic oscillator induces vibrational
resonance (Figure 6a), enabling the core barrel (typical 125 mm inside diameter) to easily
penetrate into soil and weak rock (vibration frequency adjusted between 50 and 140 Hz to
optimize penetration as soil density/hardness varies). An over casing is then advanced around the
core barrel to the driven depth to prevent borehole collapse, followed by retraction of the core
barrel for sample recovery; the process is repeated in 3 m intervals to the depth of interest
(Figure 6b). A continuous sample core to 50 m depth (with less than 1% deviation from vertical)
can typically be achieved in one day.
The vibratory sonic drilling technique induces vibrations that can disturb surrounding soil and
heat the soil sample obtained. As a result, common practice is to perform all subsequent
soundings/borings 3 m (10 ft) from the sonic location; if sonic drilling is performed after other
methods then it can be positioned much closer (e.g. 1 m). Heating of the soil sample due to
vibrations can vaporize water in the soil, resulting is sample water contents that are less than in
situ conditions. Vibrations, which are effective in displacing soil, can also pulverize weak rock
and damage/destroy soil structure, cementation, and laminations. Inferred sample depths within
each 3 m core sample can drift by up to 0.3 m due to vibrations and sample handling.
Core Samples
Graphic Log
1. 2. 3. 4. 5.
Depth (ft)
Run / Sample
Identification
Recovery (ft) /
Run Interval
Run Length
S-38 51%
(b)
S-39
Core barrel Over casing Sample 35
penetration advancement recovery
S-40
S-41
40
67%
S-42
S-43
(a) (c) (d)
Figure 6. Images of sonic drill equipment (a) head, (b) drilling schematic (c) rods and clamping
system (image from sonicsampdrill.com), and (d) continuous soil sample core.
Core barrel samples are sleeved in plastic (Figure 6d) and laid out, much like a rock core, to map
the formation. As evident, a continuous profile enables identification of primary geologic strata
as well as subtle variations in the material characteristics within each layer.
The sonic sample can provide valuable information for confirming/refining the geologic
hypotheses, determining which subsequent drilling, sampling, and in situ testing techniques are
appropriate, and developing a rational basis for defining stratigraphic layers in the site
idealization stage. As a result, it can be particularly effective as the first tool mobilized to site.
Penetrometer Measurements
Penetrometer resistance measurements often form the basis for characterizing gravels as
undisturbed sampling and laboratory testing are prohibitive in practice. Relying on penetrometer
measurements for property characterization requires confidence that the measurements obtained
be representative of the soil behavior when mobilized by the controlling deformation/failure
mechanisms. As a result, it is imperative that the gravel particles do not artificially increase the
penetrometer resistance, giving a false measure of material competence.
The conditions at which gravel particles influence penetrometers depend on the gravel properties
(size and percent gravel), penetrometer geometry (outer diameter, inner diameter and/or annulus
area), and penetration mechanisms (open-ended sampler penetration versus closed-ended
penetration). Despite the complexities, several ‘rule of thumb’ recommendations exist. For
example, Mejia (as reported in Idriss and Boulanger 2008) suggested that gravel content above
20% will likely result in elevated SPT blow counts (no specification on gravel size was given).
Bolton et al. (1999) suggested that the penetration resistance changes when the cone
penetrometer diameter to particles size ratio is less than 20:1. Others, such as Daniel et al. 2004
have also provided recommendations. While the exact parameter ratio (e.g. particle size to inner
sampler diameter) values are not agreed upon, there is consensus that gravel particles can
influence the standard penetration test (SPT). This is reflected in the wide range of alternate
larger samplers/penetrometers that have been developed and tested in gravels.
The range of large penetrometer tests (LPT) developed is summarized in Figure 7 (discussion of
the CPT is not included in the figure due to its steady state penetration mechanism). As evident,
the sampler/penetrometer diameter ranges from the SPT (5.0 cm) up to the Becker penetration
test (16.8 cm) (Figure 7a). The reference energy for a given method correspondingly increases
with sampler diameter. The LPTs, like the SPT, use an open-ended sampler and suffer from
unrepeatable results due to an inconsistent penetration mechanism in the presence of large
particles. Large particles may arch across the inner diameter of these probes and change the
penetration mechanism from partial to full displacement, they may slide inside the probe without
causing additional resistance, or they may be encountered by the probe and increase the blow
counts while they are pushed aside or crushed. Closed ended probes such as the CPT, DPT or
BPT are less susceptible to this aspect as the penetration mechanism is always full displacement.
The additional amount of gravel content (in terms of both percent and size) that can be
successfully sampled by these larger penetrometers without influencing blow counts is still
limited. Many of the large penetrometers have diameters approximately 150% of the SPT
diameter, which is still comparable to larger gravel particles (relative scales indicated in Figure
7b). Only the BPT has a diameter that significantly exceeds the gravel particle size range and
provides the driving energy necessary for advancement in both loose and dense deposits.
4
Fine Coarse
Fine Gravel Coarse Gravel Gravel Gravel
Reference Hammer Energy* (kJ)
BPT
3
2
ILPT
SPT BPT
1
JLPT
DPT
NALPT
Cal. 10 cm
SPT Burmister LPT
Mod
0
0 4 8 12 16 20
Sampler Outer Diameter (cm)
5 cm 16.8 cm
* Reference Hammer Energy is 30% of the maximum
hammer energy for BPT and 60% for all other penetrometers.
Figure 7. (a) Images of the SPT sampler and Becker drill tip compared to fine and coarse gravel
and (b) comparison of large scale penetrometers (sampler shape and particle size also indicated
for reference; expanded from Daniel et al. 2004).
The following sections build on this insight. Presented first is a systematic, practical
methodology for assessing the extent to which, if any, gravel present in the soil may adversely
influence SPT penetration resistance. Second, a recently developed instrumented Becker
Penetrometer method for estimating equivalent SPT N60 values in gravelly soils is presented.
A practical, repeatable framework for assessing gravel influence on SPT measurements was
developed based on the analysis of nearly 600 individual SPT measurements (and complimentary
information) in gravelly alluvium from four project sites. Two screening information categories
were used to infer where there was sufficient gravel to adversely influence the SPT
measurement: SPT blows-per-inch (per 25.4 mm of penetration) and physical evidence
(encompassing sample gradation, photos, and logs from the SPT test and/or adjacent Sonic
soundings).
The per-inch (per 25.4 mm) blow counts are used to detect gravel influence, which often
manifests as an increase in per-inch blows over a few inches of penetration. Often these “spikes”
can be corrected for, as depicted in Figure 8a, when either a thin gravel layer is penetrated or a
gravel particle temporarily increases penetration resistance until it is pushed aside. Multiple
changes in the blows-per-inch trend, consistently high penetration resistance, or incomplete
penetration often prohibit the measurements from being salvaged. Detailed guidelines and
examples can be found in Ghafghazi et al. (2016).
The second screening category, physical evidence, is necessary to identify the abundance and
size of the gravel particles in the soil being penetrated and to assess whether it is sufficient to
adversely influence the SPT measurement. The mere presence of gravel sized particles (as
indicated by the physical evidence) is not sufficient to infer gravel influence on SPTs; it is
plausible that small amounts of coarse gravel can influence SPTs while larger amounts of finer
15 No - I
Position (in)
6 Yes No
No - Yes Yes II
Yes No
Yes Yes III
No -
Yes Yes Yes Yes IV
12 30 Yes No Yes Yes V
* Physical Evidence refers to soil gradations, sample photos and field logs from
SPT split spoon samples and/or Sonic cores in the vicinity of the SPT sample
** An influential gravel is one of sufficient size and abundance to have plausibly
18 45 affected SPT penetration measurement.
(a) (b)
Figure 8. (a) Example of an Index III SPT sample showing blows per inch (per 25.4 mm) and the
applied correction, and (b) a table summarizing the five indices describing SPT sample quality.
gravel may not influence the SPT blow counts. Therefore, the framework suggests using the
physical evidence from SPT and sonic testing to screen for the presence of ‘influential gravels’ –
gravel particles of sufficient size and abundance to plausibly affect SPT penetration.
Collectively, the two screening categories are used to assign each SPT measurement an Index of
I to V based on the framework outlined in Figure 8b. Once an Index has been assigned, the
measurements are categorized as either high quality (HQ) or low quality (LQ), with HQ
measurements representing those free of adverse gravel influence. Any sample with an Index of
I-III is considered HQ while Index V samples are rated LQ. Index IV samples having less than
20% gravel content are rated HQ while those containing more than 20% gravel are rated as LQ.
Application of this method is intended to improve the ability to detect whether gravel has
influenced the SPT measurement, provide a reliable process for correcting the influence when
possible, and to increase certainty as to whether or not SPT measurements can be used in
characterizing a given gravelly alluvial deposit.
The Becker Penetration Test (BPT), developed in Canada in the late 1950s, uses a 180 ICE
double-acting diesel hammer to drive a 168 mm diameter, closed-ended steel pile. The BPT, out
of the extensive set of penetrometers presented in Figure 7, is least likely to be adversely
influenced by gravel particles and as such is the most promising tool for characterizing gravelly
alluvium.
Unlike common penetrometers which are performed in over-bored open holes (often using
drilling mud), the BPT’s drive shoe (a closed-ended, 8 toothed, crowd-out bit) is the same
diameter as the drill string, therefore, shaft friction accumulates during driving, contributing to
the measured blow counts (NB). Utilizing prior methods of interpreting BPT measurements (e.g.
Harder and Seed, 1986; Sy and Campanella, 1994) leads to unreliable results that primarily stem
from the inadequate methods employed to account for the effects of shaft friction on BPT
penetration resistance.
Recent technological advances have enabled the development of the instrumented BPT (iBPT,
DeJong et al. 2014, 2016, Ghafghazi et al. 2016), which directly measures the energy delivered
to the drill string head and penetrometer tip (Figure 9a). Downhole sensors (force and
acceleration) and a vibration-isolated data acquisition system continuously record measurements
of hammer blows and digitally transmit them above ground to a data control unit (Figure 9c).
iBPT operation generates a continuous profile of both the per-foot blow count (NB) and average,
per-foot residual energy arriving at the penetrometer tip (Eres,tip) (Figure 10a,b). The delivered
energy ratio (ERT/H), the residual energy delivered to the drill string tip relative to the energy
delivered by the hammer, provides insight into how much energy is absorbed by shaft friction
with penetration depth (Figure 10c); the ERT/H often decreases to less than 25% within 10 m of
driving.
Bounce Chamber
Ram
Exhaust
Fuel
Measured response at
Impact Helmet module sensor locations (b)
Head Section
Ground
Surface
F
V·Z
Tip Section
Drive Shoe
Closed-ended 2L/c t
(a) Crowd-out (d)
Figure 9. (a) Schematic of the iBPT system including the diesel hammer and sample
measurements at the head and tip, (b) drill rig used to conduct Becker testing, (c) above-ground
data control system for the iBPT, and (d) iBPT tip section and closed-ended drive shoe.
An energy normalized tip penetration resistance can be directly calculated using the energy
arriving at the tip and per-foot blow counts:
𝐸𝐸𝑟𝑟𝑟𝑟𝑟𝑟,𝑡𝑡𝑡𝑡𝑡𝑡
𝑁𝑁𝐵𝐵30 = 𝑁𝑁𝐵𝐵
30
where NB is the measured blow count, Eres,tip is the per-foot average residual energy delivered to
the tip as a percentage of the nominal hammer energy (11 kJ), and 30 is the standardized
efficiency of the Becker hammer as a percentage. Since NB30 values from iBPT are normalized
based on the energy arriving at the tip, the influence of shaft friction has been accounted for.
The empirical correlation used to convert iBPT NB30 values to equivalent N60 values is presented
in Figure 11. Each data point on the correlation figure compares the median iBPT NB30 and SPT
N60 values over geologically consistent depth increments, from adjacent, closely spaced borings.
The geologically consistent depth increments are not of consistent thickness for every data pair,
therefore, the size of each point depicts the amount of data supporting it, with larger points
NB Eres,tip (%) ERT/H (%) iBPT N60
0 100 200 300 0 10 20 30 0 25 50 75 100 0 25 50 75 100
30
35
Depth (m)
40
45
50
Figure 10. Example iBPT data output and conversion to equivalent iBPT N60 profile. Note that
iBPT penetration started at 31 m depth after predrilling the boring through a compacted
embankment.
200
160
SPT N60
120
80
40
0
0 20 40 60 80 100
iBPT NB30
Figure 11. Correlation between iBPT NB30 and SPT N60 values based on 110 data points from
four different project sites.
N60 N60 N60
0 50 100 150 200 0 50 100 150 200 0 50 100 150 200
0 35 0
10 40 5
Depth (m)
Depth (m)
Depth (m)
15
20 45 10
25
30 50 15
Figure 12. Example iBPT profiles from three different project sites. Agreement between iBPT
N60 prediction and high quality SPT N60 data and estimated CPT N60 values (using Robertson
2012 conversion). Low quality SPT measurements are due to high gravel content, generally
occur where variation increases, confirming gravel influence.
carrying more weight. Overall, a linear correlation factor of 1.8 exists between iBPT NB30 and
SPT N60 values regardless of material type or saturation conditions, albeit with ±40% scatter. As
shown previously (Figure 4), this scatter is characteristic of the alluvial deposits in which the
correlation was developed.
The iBPT provides repeatable, continuous N60 profiles not adversely influenced by the presence
of gravel. Figure 12 compares iBPT equivalent N60 values to SPT measured and CPT estimated
N60 values (Robertson, 2012) across three profiles from three different alluvium sites. The iBPT
consistently trends with either SPT or CPT, with departures attributed to spatial variability. It is
noted that there are instances where LQ, gravel influenced, SPT measurements (open ‘+’
symbols) are much higher than the equivalent iBPT measurements.
The iBPT system provides a new, reliable method for the characterization of gravelly alluvium.
The iBPT provides energy normalized penetration resistance using energy measured at the drill
tip during driving. A robust correlation, used to estimate equivalent SPT N60 values, is based on
more than 1,000 m of driving, 100,000 recorded hammer blows, and 600 corresponding SPTs.
Itis applicable for gravelly deposits whose composition ranges from gravels to sandy clays.
Conclusions
The characterization of project sites where gravelly alluvium is present poses particular
challenges, but is simultaneously of critical importance for liquefaction potential evaluation as
well as for conventional geotechnical analysis and design. The following observations have been
made:
• The depositional process of gravelly alluvium is a complex, energy and sediment load
dependent process that can result in highly interlayered deposits comprised of gravel to clay
sized particles transported downstream and downslope from different geologic source
materials. As a result, a high level of spatial variability, both horizontally and vertically,
should be expected.
• The extent of spatial variability in alluvial deposits is larger than what many engineers would
estimate, based on examining CPT sounding trends versus depth, for example. Quantitative
analysis indicates that coefficient of variation values on the order of 0.3 to 0.4 are common.
It is critical to recognize and rigorously estimate the extent of variability (particularly when
the site-specific data population is limited) as property values towards the lower part of the
range are often selected as representative values for analysis/design since they often control
performance.
• A comprehensive approach to the characterization of alluvial deposits is recommended to
properly understand the geologic formation processes, interpret data from the site
investigation, and synthesize the collective information to develop an idealized site
realization that will be used for analysis and design. This is particularly important in highly
variable gravelly alluvial deposits as the range of in situ and laboratory tools that are feasible
and appropriate can differ significantly from those conventionally used for sands and clays.
• Extensive effort can be placed in the early Inductive Reasoning and Scenario Assessment
steps of the site characterization process in order to develop a detailed hypothesis of expected
site conditions and the likely deformation/failure modes that will control geotechnical
performance and design. When performed well, the subsequent steps of Site Investigation
and Site Idealization become confirmatory and refinement processes instead of discovery
processes.
• Sonic drilling provide continuous (disturbed) samples in gravelly alluvium that enable
delineation of geologic sequences and stratigraphic layering. The larger sample size captures
gravel size particles and provides less scalping relative to the standard penetration test. The
continuous samples are valuable for confirming initial hypotheses and justifying stratigraphic
delineations during site idealization as well as for selecting other investigation techniques.
• The standard penetration test (SPT) N value, an industry standardized drive sampling index
commonly used to assess liquefaction triggering potential, can be adversely affected by
gravel. A method for systematically evaluating the quality of SPT data, based on blows-per-
inch data as well as sample images, gradations, and soil descriptions, enables evaluation of
gravel influence, if any, and identification of high quality data that can be used with
confidence in analysis and design.
• A novel, robust instrumented Becker Penetration Test (iBPT) that directly measures the
energy delivered to the drill string drive shoe provides a continuous profile of energy
normalized soil resistance with depth that can be used to estimate equivalent SPT N values
for soil characterization and liquefaction potential evaluation. The large diameter equipment
can penetrate gravelly alluvium without concern for particle size influence and can also be
deployed through existing dams and embankments.
Acknowledgments
The authors appreciate the funding and support provided by the Division of Safety of Dams of
the California Department of Water Resources (David Gutierrez, Richard Armstrong) and the
Los Angeles Department of Water and Power (Adam Perez, Craig Davis). The support of Kevin
Kuei, Chase Temple, Dan Wilson, Bill Sluis, and Daret Kehlet of the Department of Civil
Engineering at the University of California Davis is also greatly appreciated. In addition, the
support and collaboration with AMEC, AECOM, and GeoPentech consultants as well as with
Great West Drilling is appreciated.
References
Baecher GB and Christian JT. Reliability and Statistics in Geotechnical Engineering. New York. John Wiley &
Sons, 2013.
Bolton MD, Gui MW, Garnier J, Corte JF, Bagge G, Laue J, Renzi, R. Centrifuge cone penetration tests in sand.
Geotechnique. 1990; 49(4): 543-552.
Boulanger RW, Montgomery J. Nonlinear Deformation Analyses of an Embankment Dam on a Spatially Variable
Liquefiable Deposit. 6thICEGE. Christchurch NZ, 2015.
Cao Z, Youd LT, Yuan X. Chinese Dynamic Penetration Test for Liquefaction Evaluation in Gravelly Soils. ASCE
JGGE, 2013; 139(8):1312-1333.
Clayton C, Matthews M, Simons N. Site Investigation (2nd ed.). Oxford: Blackwell Science, 1995.
Clayton CRI, Smith DM. Effective Site Investigation (2nd ed.). Site Investigation Steering Group, Institution of Civil
Engineers, London: ICE Publishing, 2013.
Cubrinovski M, Bray JD, Taylor M, Giorgini S, Bradley B, Wotherspoon L, Zupan J. Soil liquefaction effects in the
central business district during the February 2011 Christchurch earthquake, Seismological Research Letters, 2011a;
82(6): 893-904.
Cubrinovski M, Bradley B, Wotherspoon L, Green R, Bray J, Wood C, Pender M, Allen J, Bradshaw A, Rix
G, Taylor M, Robinson K, Henderson D, Giorgini S, Ma K, Winkley A, Zupan J, O'Rourke T, DePascale G, Wells
D. Geotechnical aspects of the 22 February 2011 Christchurch earthquake, Bulletin of the New Zealand Society for
Earthquake Engineering, 2011b; 44(4): 205-226.
Daniel CR, Howie JA, Campanella RG, Sy A. Characterization of SPT grain size effects in gravels. ISC’2. Porto,
Portugal, 2004; 8p.
DeJong JT, Ghafghazi M, Sturm AP, Armstrong R, Perez A, Davis C. A New Instrumented Becker Penetration Test
(iBPT) for Improved Characterization of Gravelly Deposits Within and Underlying Dams. ASDSO. The Journal of
Dam Safety, 2014; 12(2): 9-20.
DeJong JT, Ghafghazi M, Sturm AP, Wilson DW, den Dulk J, Armstrong RJ, Perez A, Davis CA. instrumented
Becker Penetration Test: Equipment, Operation, and Performance. ASCE JGGE, 2016 – under review.
Duncan JM. Factors of Safety and Reliability in Geotechnical Engineering, ASCE JGGE, 2000; 126(4): 307-316.
Ghafghazi M, DeJong JT, Sturm AP, Temple CE. Instrumented Becker Penetration Test: The Application of SPT
and iBPT for Liquefaction Assessment in Gravelly Soils. ASCE JGGE, 2016 – under review.
Ghafghazi M, DeJong JT, Wilson, D. Review of Becker Penetration Test Interpretation Methods for Liquefaction
Assessment in Gravelly Soils. ASCE JGGE, 2016 – under review.
Harder LF Jr. Becker test results from gravel liquefaction sites. ASCE GSP, 1994; 44: 201-220.
Harder LF Jr, Seed HB. Determination of penetration resistance for coarse-grained soils using the Becker Hammer
Drill. College of Engineering, University of California, Berkeley. Report No. UCB/EERC-86/06, 1986.
Idriss IM, Boulanger RW. Soil Liquefaction during Earthquakes. Oakland, California. EERI 2008.
Mayne PW, Christopher BR, DeJong, JT. Subsurface Investigations. National Highway Institute. U.S. Department
of Transportation. Federal Highway Administration, 2002.
Nichols, Gary. Sedimentology and Stratigraphy,2nd Edition. Wiley-Blackwell, 2009.
Phoon KK, Kulhawy FH, Grigoriu MD. Reliability based design of foundations for transmission line structures.
Electric Power Research Institute, Palo Alto, Report TR-105000, 1995.
Phoon KK, Kulhawy FH. Characterization of geotechnical variability. CGJ, 1999; 36: 612-624.
Phoon KK, Kulhawy FH. Evaluation of geotechnical property variability. CGJ. 1999; 36: 625–639.
Robertson PK., Interpretation of in-situ tests – some insights. ISC’4. Recife, Brazil. 2012; 22p.
Sy A, Campanella RG. Becker and standard penetration tests (BPT-SPT) correlations with consideration of casing
friction. CGJ, 1994; 31(3): 343-356.
U.S. Army Corps of Engineers. EM 1110-1-1804 Geotechnical Investigations. Washington, DC, 2001.
U.S. Army Corps of Engineers. EM 1110-2-1156 Safety of Dams – Policies and Procedures. Washington, DC,
2011.
U.S. Department of the Interior - Bureau of Reclamation. Best Practices in Dam and Levee Safety Risk Analysis.
Washington, DC, 2012.