Estimating Maximum and Minimum Void Ratio From Index Parameters by Pires-Sturm 2022

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 12

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/369668601

Estimating Maximum and Minimum Void Ratio from Index Parameters

Conference Paper · September 2022

CITATIONS READS
0 321

3 authors:

Alex Pires-Sturm Robert Jaeger


California Department of Water Resources California Department of Water Resources
10 PUBLICATIONS   60 CITATIONS    23 PUBLICATIONS   188 CITATIONS   

SEE PROFILE SEE PROFILE

Jason T. Dejong
University of California, Davis
270 PUBLICATIONS   10,510 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Bio-inspired self-burrowing in-situ testing probe View project

Microbial Induced Calcite Precipitation View project

All content following this page was uploaded by Robert Jaeger on 31 March 2023.

The user has requested enhancement of the downloaded file.


Estimating Maximum and Minimum Void Ratio
from Index Parameters
Alex Pires-Sturm, PhD, PE, Design Engineer, CA Division of Safety of Dams;
Robert Jaeger, PhD, PE, Supervising Engineer, CA Division of Safety of Dams; and
Jason DeJong, PhD, Professor, University of California, Davis

Abstract--This paper proposes correlations to estimate the maximum and minimum void ratio (emax, emin) from
commonly measured index parameters. The coefficient of uniformity (Cu) is used as the primary input parameter
while particle roundness (R) can be used to refine estimations, if available. The correlations were developed via
regression of empirical data from 15 publications and are applicable for cohesionless, quartz or silica soils with fines
content from 0 to 10%, mean particle diameter (D50) from 0.1 to 115 millimeters, and Cu from 1.2 to 334. The
correlations can be used to estimate emax and emin in soil deposits when site-specific measurements are not available
or are too costly to obtain. The estimated emax and emin values can be used to calculate site-specific relative density
(DR), which can improve the characterization of cohesionless soils for the design or re-evaluation of existing
structures. This paper concludes with an example from a seismic stability evaluation at a dam in California.

I. INTRODUCTION
The maximum and minimum void ratios (emax, emin) represent the loosest and densest states a given coarse-
grained soil mass can obtain. These parameters are primarily used to calculate relative density (DR, Equation 1),
which is a key parameter for predicting the behavior of cohesionless soils. DR has a wide variety of applications in
dam safety engineering including evaluating liquefaction triggering potential, estimating drained and post-
liquefaction undrained strengths, and calibrating constitutive models used in non-linear deformation analyses [1, 2].

𝑒𝑒𝑚𝑚𝑚𝑚𝑚𝑚 − 𝑒𝑒
𝐷𝐷𝑅𝑅 = (1)
𝑒𝑒𝑚𝑚𝑚𝑚𝑚𝑚 − 𝑒𝑒𝑚𝑚𝑚𝑚𝑛𝑛

While standardized methods exist to measure emax and emin, they are seldom used in practice because of the
associated costs and challenges (e.g., specialized equipment, experienced technicians, large sample size, etc.).
Many engineering applications circumvent measuring site-specific values by adopting typical values. This typically
isn't an issue for clean quartz or silica sands, where the standard of practice is well established and emax and emin
take on a relatively narrow range of values. However, it is an issue for gravelly soils where emax and emin can be
considerably lower than they are for clean sands. It is also more challenging to measure to measure emax and emin
in gravelly soils due to soil segregation, particle size effects, etc. Little practical guidance exists to account for the
changes in emax and emin with gradation, which makes it difficult to accurately predict DR in gravelly soils that are
frequently encountered in the foundations and shells of embankment dams.
The purpose of this paper is to leverage available literature to develop correlations to estimate emax and emin
from commonly measured index parameters.

II. BACKGROUND

A. Laboratory Testing
ASTM D4254-16 [3] governs the measurement of emax. Three methods are described to obtain the loosest
particle packing: using a funnel pouring device, extracting a soil filled tube, and inverting a graduated cylinder. Often
technicians will need to attempt all 3 methods to identify which is optimal for a given soil sample. All 3 methods are
extremely sensitive to sample disturbance; therefore, measurements must be repeated until subsequent trials are
within 2%. Reference [3] is applicable to cohesionless soils with less than 15% fines and a maximum particle size
Copyright © 2022 Association of State Dam Safety Officials, Inc. All Rights Reserved Page 1 of 11
of 3 inches; however, the required mold size and sample mass increase significantly if the maximum particle size
exceeds 0.75 inches.
ASTM D4253-16 [4] governs the measurements of emin. A calibrated vibratory table is used to densify soil
samples in a standard mold under a known confining stress. The test can be conducted on dry or saturated soil
samples with laboratories often trying both methods to identify which achieves the densest particle packing.
Measurements must be repeated multiple times until subsequent trials are within 2%. Reference [4] is applicable to
cohesionless soils with less than 15% fines and a maximum particle size of 3 inches; however, the required mold
size and sample mass increase significantly if the maximum particle size exceeds 0.75 inches.
It is challenging to measure emax and emin, even when following ASTM standards. The parameters take on a
relatively narrow range of values and are sensitive to random and systematic errors. The ASTM standards explicitly
state they do not obtain the absolute loosest and densest particle packings; however, they do provide a standard
basis for comparison when used appropriately. Unfortunately, historic ASTM standards, international standards,
and non-standards methods have been used and reported throughout the literature, which adds to the uncertainty
when comparing multiple datasets.
B. Historic Correlations
In 1948, Reference [5] argued that the most important factor controlling minimum and maximum density (ρmin,
ρmax) in cohesionless soils is the range of particle sizes (CR). CR is defined as the overall slope of the gradation
curve (Equation 2) and is used as the primary variable to estimate ρmin and ρmax via Figure 1. The shape of the
gradation curve and the size/shape of the soil particles were found to have a secondary effect and can be used to
refine the ρmin and ρmax estimates. Figure 1 predicts ρmin and ρmax will approach an asymptotic upper bound as CR
increases. The trends of Figure 1 are not used in current practice because they are imprecise and rely on outdated
input parameters; however, they represent an early recognition of the primary variables controlling ρmin and ρmax,
which are inversely proportional to emax and emin.

𝐷𝐷100
𝐶𝐶𝑅𝑅 = 2 ∙ log(𝐷𝐷0
) (2)

Figure 1: Historic correlations of Reference [5] to predict ρmin and ρmax from index parameters.

In 1973, Reference [6] conducted extensive laboratory testing to study the influence of gradation, particle shape,
and particle size on emax and emin. Figure 2a shows the proposed correlations to predict emax and emin as a function
of coefficient of uniformity (Cu, Equation 3) and particle roundness (R), which was visually estimated using Figures
2b and 2c. Notably, Reference [6] disproved the prevailing notion at the time that emax and emin are dependent on

Copyright © 2022 Association of State Dam Safety Officials, Inc. All Rights Reserved Page 2 of 11
particle size and argued that past studies had not properly isolated the interdependency between particle size and
shape that is typical of natural soils. Figure 2a predicts emax and emin will asymptotically decrease as Cu increases.
Furthermore, angular soils are expected to pack looser than the equivalent round soils since touching asperities will
increase void space. The role of particle shape is shown to be more pronounced for emax than emin. The authors of
this paper hypothesize that the energy input during emin testing overcomes some of the effects of angularity by
allowing particles to re-arrange and the fill void space between touching asperities. Since emax testing minimizes the
energy input to the system, there is nothing to counteract the role of angularity, thus its more pronounced effect.

𝐷𝐷
𝐶𝐶𝑢𝑢 = 𝐷𝐷60 (3)
10

Figure 2: a) Historic correlations of Reference [6] to predict emax and emin from Cu and R, b) representative particle shapes for
visual classification, and c) the proposed particle shape classification system.

The correlations of References [5] and [6] were primarily developed for clean sands; however, they can inform
expectations when developing correlations for gravelly soils. Gradation is the primary variable governing emax and
emin, with both parameters decreasing as the range of particle sizes (Cu or CR) increases. Per Reference [6], particle
shape is the second most important variable governing emax and emin, with both parameters decreasing as R
increases. Lastly, emax and emin approach an asymptotic limit as the gradation becomes broader.

III. CORRELATION DEVELOPMENT

A. Literature Review
A literature review was conducted to identify publications that reported emax, emin, and Cu for cohesionless soils.
A comprehensive database of 159 datapoints was compiled from 15 publications [6] - [20]. Table 1 summarizes the
database, which includes quartz and silica soils ranging in fines content from 0 to 10%, mean particle diameter (D50)
from 0.1 to 115 millimeters, and Cu from 1.2 to 334. Several publications [6, 16, 18] reported data for non-quartz
soils that tended to have higher emax and emin values; however, the information was insufficient to characterize the
role of soil minerology in any detail. As such, the correlations proposed herein were developed using data for quartz
and silica soils only, which are most abundant in the literature and most commonly encountered in practice. Figure
3 shows the variation in emax and emin as a function of Cu for the soils included in the literature database.
One limitation of the literature database is that not all publications used the current ASTM standards [3, 4] to
measure emax and emin. Some publications [12, 13] used international standards, which have been shown to yield
comparable results to ASTM standards in sandy soils. It is assumed that the international standards will also be
comparable to ASTM standards for gravelly soils. Reference [9] used non-standard methods since the soils being
Copyright © 2022 Association of State Dam Safety Officials, Inc. All Rights Reserved Page 3 of 11
tested were primarily rockfill, which exceeded the maximum particle size allowed by ASTM standards. Reference [6]
also used a non-standard method to measure emin, which has been shown by Reference [21] to produce slightly
denser samples than current ASTM standards. Finally, several publications did not report the methods used to
measure emax and emin. The variation in testing method is expected to contribute to the overall scatter in the literature
database; however, the authors searched and did not detect any bias with respect to test method.

Table 1: Summary of parameter ranges covered by each reference in the literature database.

Figure 3: Variation in emax and emin as a function of Cu including measurements from the literature database, proposed
correlations (Equations 5 and 6), and uncertainty bounds (Table 2).
B. Functional Form
Based on Reference [5] and [6], it is hypothesized that emax and emin are primarily dependent on soil gradation,
which can be quantified by Cu (Equation 3). As the range of particle sizes increases with Cu, the particles can pack
more efficiency into the available void space. Previous correlations have modeled this process with a power function
[6]; however, this forces the greatest rate of change in emax and emin to occur at low Cu values. This somewhat
contradicts the standard of practice, which assumes that emax and emin are about constant for clean quartz and silica
sands with low Cu values. For example, the constitutive model PM4Sand assigns typical emax and emin values of 0.8
Copyright © 2022 Association of State Dam Safety Officials, Inc. All Rights Reserved Page 4 of 11
and 0.5 to "produce reasonable agreement with the trends in typical design correlations" [2]. The model can be
calibrated if site-specific measurements are available; however, it is generally acceptable to assume typical values
for clean quartz and silica sands where the standard of practice is well-established and emax and emin take on a
relatively narrow range of values.
This is supported by Figure 4, which shows the range of emax and emin values for clean quartz and silica sands
with a Cu less than 2 included in the literature database. Some of the variation observed in Figure 4 can be attributed
to the secondary effects of R, predicted by Reference [6]. For example, blasting sand is more angular than the other
soils and has the highest emax and emin values of all the clean sands included in Figure 4.

1.1 Blasting
1.1
Sand [11] [13] [14] [15] [16] [19] [20]
Toyoura
1 Ticino Sand
Takahagi 1
Sand #1 Sand Sand

0.9 100B
0.9
Ottawa F-110 Sand

0.8 ASTM
100A 0.8
Ottawa Graded Sand Blasting
F-65 Sand Sand
0.7 Ottawa
20-30 Sand
0.7
#1 Sand Toyoura Takahagi
Ticino Sand
emax

emin
Sand
0.6 0.6 Sand
100A
Ottawa
F-110
0.5 0.5 Ottawa
Sand
100B
Ottawa ASTM
20-30 Sand F-65 Sand Graded Sand
0.4 0.4

0.3 0.3

0.2 0.2

0.1 0.1
1 1.2 1.4 1.6 1.8 2 1 1.2 1.4 1.6 1.8 2
Cu Cu
Figure 4:Variation in emax and emin as a function of Cu for clean quartz and silica sands included in the literature database.

A hyperbolic function was used to fit the correlations proposed herein to allow for a lesser rate of change in emax
and emin at low Cu values. The authors argue that this lesser rate of change is more consistent with the standard of
practice for clean quartz and silica sands than the power function used by Reference [6]. Equation 4 provides the
generic form of the hyperbolic function where Y is the dependent variable (emax or emin), X is the primary predictor
variable (Cu), and α (alpha), ω (omega), λ (lambda), and β (beta) are fitting parameters, where α controls the
maximum Y value as X approaches 1, ω controls the minimum Y value as X approaches infinity, λ controls the X-
position of the transition zone, and β controls the shape of the transition zone.

𝛼𝛼 − 𝜔𝜔
𝑌𝑌 = 𝛼𝛼 − 𝛽𝛽 (4)
1 + �𝜆𝜆�𝑋𝑋�

C. Proposed Correlations
Equations 5 and 6 are proposed to estimate emax and emin as a function of Cu. The solid lines in Figure 3 compare
the proposed correlations to the literature database. Initially, the correlations were fit via least squares regression;
however, the authors recognize that the literature database is not comprehensive and contains significant scatter;
therefore, the following adjustments were made to the best-fit parameters based on engineering judgement.

0.93 − 0.46
𝑒𝑒𝑚𝑚𝑚𝑚𝑚𝑚 = 0.93 − 2.6 (5)
1 + �4�𝐶𝐶 �
𝑢𝑢

Copyright © 2022 Association of State Dam Safety Officials, Inc. All Rights Reserved Page 5 of 11
0.53 − 0.23
𝑒𝑒𝑚𝑚𝑚𝑚𝑚𝑚 = 0.53 − 2.6 (6)
1 + �4�𝐶𝐶 �
𝑢𝑢

First, α was set equal to 0.93 and 0.53 in Equations 5 and 6, respectively. This controls the emax and emin values
when Cu equals 1. As seen in Figure 3, there is significant scatter in the measured emax and emin data at low Cu
values; therefore, the authors did not want to rely solely on the literature database to constrain α. As discussed in
detail below, much of the scatter at low Cu values is likely attributable to particle shape. As such, α was set equal
to the emax and emin values predicted by Reference [6] when Cu equals 1 and R equals 0.35 (Figure 2a). An R value
of 0.35 corresponds to a subrounded/subangular soil (Figure 2b, 2c), which was selected to represent a typical soil
encountered in engineering practice. This refinement provides good agreement with the literature database (Figure
3) and is consistent with the previous work of Reference [6] (Figure 2a).
Next, ω was set equal to 0.46 and 0.23 in Equations 5 and 6, respectively. This controls the emax and emin values
as Cu approaches infinity. As seen in Figure 3, scatter in the measured emax and emin decreases with increasing Cu;
therefore, the authors were comfortable using the literature dataset as the only means to constrain ω. As such, ω
was set equal to the median emax and emin values of the literature database for Cu greater than or equal to 20.
With α and ω fixed, λ and β were fit via least square regression with the constraint that they be equal between
Equations 5 and 6. This was necessary to ensure the difference between emax and emin was sensible and always
greater than zero. Ultimately, λ and β were set equal to 4 and 2.6 in Equations 5 and 6.
D. Secondary Effects
Based on Reference [6], particle shape is expected to have a secondary effect on emax and emin. Particle shape
is best quantified by R, which can be visually estimated (Figure 2b, 2c) or quantified using particle photographs and
image analysis software. Angular particles will not pack efficiently since their asperities will touch and create void
space. Conversely, the lack of asperities in round particles will allow them to pack more efficiently with less void
space.
Reference [6] predicts that R will have a similar effect on emax and emin at all Cu values between 1 and 10
(Figure 2a); however, the authors of this paper expect the role of particle shape will diminish as Cu increases. As
the range of particle sizes (e.g., Cu) increases the smaller particles can fill any void space left between touching
asperities; thereby reducing the effect of particle shape. Furthermore, it becomes more difficult to quantify R as Cu
increases since different sized particles are likely to have different shapes. For these reasons, the correlations
proposed below were developed such that R had no impact on the predicted emax and emin values for Cu greater
than about 10.
Equations 7 and 8 can be used to refine the recommended α values in Table 2 at sites where the particle shape
is expected to be important. Most of the references in the literature database did not report particles shape;
therefore, Equations 7 and 8 were not developed using the empirical data. Instead, they were developed such that
emax and emin at Cu = 1 would match the R-dependent trends proposed by Reference [6] (Figure 2a).

𝛼𝛼|𝑒𝑒𝑚𝑚𝑚𝑚𝑚𝑚 = 0.041(𝑅𝑅)−1.72 + 0.68 (7)

𝛼𝛼|𝑒𝑒𝑚𝑚𝑚𝑚𝑚𝑚 = 0.008(𝑅𝑅)−2.78 + 0.45 (8)

Figure 5 compares the default correlations of Equation 5 and 6 to the correlations which have been adjusted to
account for the secondary effect of particle shape (Equation 7 and 8). Figure 5 also shows measured values from
the literature database that reported particle shape. The proposed trends are in reasonably good agreement with
the limited empirical data, especially when considering the overall scatter observed in Figure 3.
It is noted that the R-dependent trends of Figure 5 (Equations 7 and 8) are not as robust as the default
correlations of Figure 3 (Equations 5 and 6). The authors recommend using the R-dependent correlations only for
estimation purposes at sites where particle shape is expected to be important.

Copyright © 2022 Association of State Dam Safety Officials, Inc. All Rights Reserved Page 6 of 11
Angular (R=0.21) Subangular (R=0.30) Subrounded (R=0.41) Rounded (R=0.59) Proposed Fit
Angular Subangular Subrounded Rounded Well Rounded
(0.17<R<0.25) (0.25<R<0.35) (0.35<R<0.49) (0.49<R<0.70) (0.70<R<1.00)

1.1 1.1

1 1

0.9 0.9

0.8 0.8

0.7 0.7
emax

emin
0.6 0.6

0.5 0.5

0.4 0.4

0.3 0.3

0.2 0.2

0.1 0.1
1 10 100 1 10 100
Cu Cu
Figure 5: Variation in emax and emin as a function of Cu including measurements from the literature database that reported
particle shape, default correlations (Equation 5 and 6), and R-dependent correlations for refinement (Equations 7 and 8).
E. Uncertainty
The dashed lines in Figure 3 show the proposed correlations ±1 standard deviation (σ). Table 2 provides the
adjusted fit coefficients, which can be used with the generic function form (Equation 4) to quantify this uncertainty
and develop emax and emin ranges for sensitivity analyses.
As seen in Figure 3, scatter in the measured emax and emin values decreases with increasing Cu. To capture this,
two σ values were computed: one for Cu ≤ 10, and one for Cu > 10. The calculated σ values were added/subtracted
from α and ω to achieve a smooth transition over the entire Cu range (rather than an abrupt step at Cu = 10). As
discussed previously, α controls emax and emin at low Cu values; therefore, α was adjusted by the σ values for Cu ≤
10 (0.12 for emax and 0.09 for emin). Similarly, ω controls emax and emin at high Cu values; therefore, ω was adjusted
by the σ values for Cu > 10 (0.05 for emax and 0.04 for emin). The authors confirmed that this approach resulted in
68% of the measured values falling between ±1σ, as is expected for the normally distributed residuals.
At first glance, the increased scatter at low Cu values seems to suggest that there is greater uncertainty in emax
and emin for clean sands, which would contradict the standard of practice assumption that emax and emin equal about
0.8 and 0.5 for these soils [2]. Upon further inspection, the authors hypothesize that much of the scatter at low Cu
values may be attributed to secondary effects of particle shape. As disused previously, and as shown in Figure 5,
particle shape has a pronounced impact on emax and emin at low Cu values, but no impact for Cu greater than about
10. This trend mirrors the narrowing of the uncertainty bounds seen in Figure 3. Unfortunately, most of the
references included in the literature dataset did not report particle shape; therefore, it is difficult to separate the
variation due to particle shape from the random and systematic errors inherent to empirical data.

Table 2: Fit coefficients for proposed correlations and uncertainty bounds.

Copyright © 2022 Association of State Dam Safety Officials, Inc. All Rights Reserved Page 7 of 11
IV. EXAMPLE APPLICATION
An 87-foot-tall, zoned embankment dam in California is the subject of a seismic stability evaluation. During
original construction, a 240-foot-wide core trench was excavated to bedrock while the natural alluvium was left in-
place beneath the upstream and downstream shells. Figure 6 shows a representative cross-section while Figure 7
shows representative gradations for the embankment fill (u/s shell, d/s shell, core) and alluvium.

20′
109
Maximum Potential Water Surface Elevation
98
Restricted Reservoir Elevation
90
Elevation (ft)

u/s Shell Core


d/s Shell
30
Alluvium Alluvium
0
Bedrock
Figure 6: Representative cross-section for the example embankment dam.

100
Embankment Fill
90 Alluvium

80

70
Percent Passing (%)

60

50

40

30

20

10

0
1000 100 10 1 0.1 0.01
Particle Size (mm)
Figure 7: Representative gradations for the embankment fill (u/s shell, d/s shell, core) and alluvium.

The embankment fill was well-compacted and is expected to meet performance objectives during the design
seismic event. The alluvium is presumed to be loose and potentially liquefiable; however, the abundance of large
particles makes it challenging to characterize and evaluate using traditional methods developed for clean sands.
During a previous site exploration, both the Standard Penetration Test (SPT) and Cone Penetration Test (CPT) met
refusal; however, the gradation and in-situ void ratio of the alluvium was measured via two, large-diameter ring
density tests. As summarized in Table 3, the alluvium has an in-situ void ratio of 0.326 to 0.338 with corresponding
Cu values of 30.6 and 22.9.

Table 3: Summary of measured and calculated parameters for example application.

Copyright © 2022 Association of State Dam Safety Officials, Inc. All Rights Reserved Page 8 of 11
Equations 5 and 6 were used in conjunction with the measured parameters to calculate the DR of the alluvium.
Table 3 summarizes the calculated parameters: emax ranges from 0.462 to 0.465, emin ranges from 0.222 to 0.223,
and DR ranges from 52 to 57% (specific gravity assumed to be 2.65). This DR estimate supports the assumption
that the alluvium is relatively loose compared to the compacted embankment and is potentially liquefiable. However,
DR can also be used as a key parameter to quantify the seismic performance of the alluvium as part of a preliminary
desktop study. Using standard correlations, the DR estimate of about 55% could be used to evaluate liquefaction
triggering potential, estimate drained and post-liquefaction undrained strengths, or calibrate a constitutive model for
non-linear deformation analyses [1, 2]. While the emax, emin, and DR values calculated via Equations 5 and 6 are
approximate in nature, they are invaluable for planning purposes given the limited information at this site.
The emax, emin, and DR estimates were particularly helpful at this site when planning a future exploration.
Reference [22] proposed Equations 8 and 9, which allowed SPT N1 to be estimated from the calculated parameters
in Table 3. It is noted that Equations 8 and 9 were developed based on non-energy normalized blow counts;
therefore, they should be applied with caution. Reference [22] used higher driving energies in the gravelly soils;
therefore, Equation 8 may underpredict CD at low emax and emin values, which are typical of gravelly soils. Future
work is needed to update Equations 8 and 9 to be consistent with the current standard of practice.

9
𝐶𝐶𝐷𝐷 = (8)
(𝑒𝑒𝑚𝑚𝑚𝑚𝑚𝑚 − 𝑒𝑒𝑚𝑚𝑚𝑚𝑚𝑚 )1.7

𝑁𝑁1 = 𝐶𝐶𝐷𝐷 𝐷𝐷𝑅𝑅2 (9)

As summarized in Table 3, CD is expected to range from 100 to 102, which corresponds to an N1 of 27 to 33. The
ability to estimate N1 is extremely helpful at this site to help determine if an SPT blow count was artificially elevated
due to the abundance of large particles. The estimated N1 is also helpful as a secondary check on the
measurements obtained from large penetrometers, such as the Becker Penetration Test (BPT), which was utilized
at this site given the historic instances of SPT and CPT refusal.
This example also highlights the fact that CD is dependent on gradation. Much like the standard of practice for
emax and emin, CD is assumed to have a typical value of about 46 in clean quartz sands [1]. If the typical CD value of
46 was used with the estimated DR of about 55%, N1 would be estimated to be about 14. This is much less than the
N1 estimate of about 30 predicted via Equations 8 and 9. If the project team did not account for the role of gradation
on CD, they might measure an N1 value much higher than 14, assume it corresponds to a higher DR, and conclude
the alluvium is non-liquefiable. Based on the DR estimate of 55% the alluvium is considered potentially liquefiable;
however, future research is needed to develop analysis procedure to evaluate liquefaction potential of gravelly soils.
Overall, this example demonstrates just one of the many ways gravelly soils are unique from clean sands.
Engineers must use the appropriate correlations when analyzing the behavior of gravelly soils that fall outside the
standard of practice developed for clean quartz sands.

V. CONCLUSIONS
Gravelly soils are unique from cleans sands and require special consideration when designing or re-evaluating
embankments constructed from or founded upon these soils. Gravelly soils exhibit a wide range of particle sizes,
which is the primary factor controlling emax and emin. Ideally, site-specific emax and emin measurements would be
obtained at every site where they are needed; however, the cost and challenges associated with sampling and
testing can be prohibitive, especially for gravelly soils.
Equations 5 and 6 are proposed to estimate emax and emin directly from Cu, which can be easily measured via
sieve analysis. These equations were developed for and are applicable to quartz and silica soils with fines content
from 0 to 10%, D50 from 0.1 to 115 millimeters, and Cu from 1.2 to 334. At low Cu values, the correlations yield emax
and emax estimates that are consist with the standard of practice for clean sands while also capturing the reduction
in emax and emin as Cu increase for gravelly soils. Equations 5 and 6 are representative of subrounded to subangular
soils typically encountered in engineering practice. The uncertainty in Equations 5 and 6 can be accounted for by
adjusting the standard coefficients via Table 2 to develop emax and emin ranges for sensitivity analyses.
Finally, Equations 8 and 9 are proposed to consider the secondary effects of particle shape. These equations
are not as well supported by empirical data and should be used for further estimation purposes only at sites where
particle shape is expected to be important.

Copyright © 2022 Association of State Dam Safety Officials, Inc. All Rights Reserved Page 9 of 11
VI. ACKNOWLEDGMENT
Any opinions, findings, and conclusions or recommendations expressed in this paper are those of the authors
and do not necessarily reflect the views of the CA Department of Water Resources, Division of Safety of Dams.

VII. REFERENCES
[1] Idriss, I.M., and Boulanger, R.W. (2008). “Soil liquefaction during earthquakes.” Earthquake Engineering
Research Institute. Oakland, CA.
[2] Boulanger, R.W., and Ziotopoulou, K. (2017). “PM4Sand (version 3.1) a sand plasticity model for earthquake
engineering applications.” Report No. UCD/CGM-17/01. Center for Geotechincal Modeling, University of
California, Davis.
[3] ASTM International. (2016). “Standard test methods for minimum index density and unit weight of soils and
calculation of relative density.” D4254-16.
[4] ASTM International. (2016). “Standard test methods for maximum index density and unit weight of soils using
a vibratory table.” D4253-16.
[5] Burmister, D. (1948). “The importance and practical use of relative density in soil mechanics.” Proceedings
ASTM, West Conshohocken, PA.
[6] Youd, T. L. (1973). “Factors controlling maximum and minimum densities of sands.” ASTM Special Technical
Publication 523, 98–112.
[7] Koerner, R. M. (1968). “The behavior of cohesionless soils formed from various minerals.” Duke University.
PhD Dissertation.
[8] Skermer, N. A., and Hillis, S. F. (1970). “Gradation and shear characteristics of four cohesionless soils.”
Canadian Geotechnical Journal, 7(1), 62–68.
[9] Marsal, R. J. (1973). “Mechanical properties of rockfill.” Embankment Dam Engineering, John Wiley & Sons.
[10] Fang, H. (1991). “Foundation engineering handbook, second edition”. Van Nostrand Reinhold. New York,
New York. 261–273.
[11] Vaid, Y. P., Fisher, J. M., Kuerbis, R. H., and Negussey, D. (1990) “Particle gradation and liquefaction”
Journal of Geotechincal Engineering, 116(4), 698–703.
[12] Kokusho, T., and Tanaka, Y. (1994) “Dynamic properties of gravel layers investigated by in-situ freezing
sampling.” ASCE Geotechincal Special Publication No. 44: Ground Failures Under Seismic Conditions. 121–
140.
[13] Kokusho, T., Yoshida, Y., and Tanaka, Y. (1995) “Shear wave velocity in gravelly soils with different particle
gradings.” ASCE Geotechincal Special Publication No. 56: Static and Dynamic Properties of Gravelly Soils.
93–106.
[14] Rashidian, M., Ishihara, K., Kokusho, T., Kanatani, M., and Okamota, T. (1995) “undrained shear behavior of
very loose gravelly soils.” ASCE Geotechincal Special Publication No. 56: Static and Dynamic Properties of
Gravelly Soils. 77–91.
[15] Cho, G. C. (2001). “Unsaturated soil stiffness and post-liquefaction shear strength.” Georgia Institute of
Technology. PhD Dissertation.
[16] Santamarina, J. C., and Cho, G. C. (2001). “Determination of critical state parameters in sandy soils - simple
procedure.” Geotechnical Testing Journal, 24(2), 185–192.
[17] Menq, F.-Y. (2003). “Dynamic properties of sandy and gravelly soils.” University of Texas at Austin. PhD
Dissertation.
[18] Harehdasht, S. A., Karray, M., Hussien, M. N., and Chekired, M. (2017). “Influence of particle size and gradation
on the stress-dilatancy behavior of granular materials during drained triaxial compression.” International Journal
of Geomechanics, 17(9).
[19] Sturm, A. P. (2019). “On the liquefaction of potential of gravelly soils: characterization, triggering, and
performance.” University of California, Davis. PhD Dissertation.
[20] B. L. Kutter, Manzari, M. T., Zeghal, M., Arduino, P., Barrero, A., Carey, T. J., Chen, L., Elgamal, A., Ghofrani,
A., Montgomery, J., Ozutsumi, O., Qiu, Z., Taiebat, M., Tobita, T., Travasarou, T., Tsiaousi, D., Ueda, K.,
Ugalde, J., Wada, T., Yang, M., Zheng, B. – L., and Ziotopoulou, K., (2020). “Numerical sensitivity study
compared to trend of experiments for LEAP-UCD-2017”, in Model Tests and Numerical Simulations of
Liquefaction and Lateral Spreading, B. Kutter, Manzari, M., and Zeghal, M. Cham, Switzerland. 219-236.
[21] Mehdiratta, G. R., and Triandafilidis, G. E. (1978). “Minimum and maximum densities of granular materials.”
Geotechnical Testing Journal, 1(1), 34–40.
Copyright © 2022 Association of State Dam Safety Officials, Inc. All Rights Reserved Page 10 of 11
[22] Cubrinovski, M., and Ishihara, K. (1999). “Empirical correlation between SPT N-value and relative density for
sandy soils.” Soils and Foundations, 39(5), 61–71.

VIII. AUTHOR BIOGRAPHIES


Alex Pires-Sturm, PhD, PE
Design Engineer
CA Division of Safety of Dams
2720 Gateway Oaks Blvd, Ste 300
Sacramento, CA, 95833
(916) 565-7888
alexander.pires-sturm@water.ca.gov

Alex Pires-Sturm is a registered professional engineer with 3 years of experience as a design engineer at the
California Division of Safety of Dams. He specializes in geotechnical engineering and has experience with seismic,
hydrologic, hydraulic, and structural evaluations of dams and their appurtenant structures. Alex received his PhD in
Civil Engineering from the University of California, Davis in 2019 with research specializing in gravelly soil
characterization, including the development of the instrumented Becker Penetration Test (iBPT).

Robert A. Jaeger, PhD, PE


Supervising Engineer
CA Division of Safety of Dams
2720 Gateway Oaks Blvd, Ste 300
Sacramento, CA, 95833
(916) 565-7851
robert.jaeger@water.ca.gov

Robert Jaeger is a supervising engineer with the California Department of Water Resources’ (DWR) Division of
Safety of Dams (DSOD). Robert has eleven years of combined engineering experience, including work as a
consulting engineer and presently a dam safety engineer at DSOD. His work experience has focused primarily on
dam and levee projects across the country, including subsurface characterization; seepage, stability, and seismic
stability analyses; design; and risk assessments. Robert obtained his Master of Science and Doctoral degrees in
civil engineering with an emphasis in geotechnical engineering from the University of California at Davis.

Jason DeJong, PhD


Professor
University of California
One Shields Ave.
Davis, CA 95616
(530) 754-8995
jdejong@ucdavis.edu

Jason DeJong is a Professor at UC Davis with 25 years of experience in soil and site characterization, particularly
in challenging soils such as gravels, intermediate soils, tailings, and offshore clays. He directs research in site
characterization, geotechnical earthquake engineering, sustainability, and biogeotechnics, and has served as a
technical advisor and reviewer for several industry projects.

Copyright © 2022 Association of State Dam Safety Officials, Inc. All Rights Reserved Page 11 of 11

View publication stats

You might also like