RESILIENT M O D U L U S FOR F I N E - G R A I N E D

SUBGRADE SOILS

By Dingqing Li, 1 Associate Member, ASCE,

and Ernest T. Selig, 2 Fellow, ASCE

ABSTRACT" A method has been developed for the estimation of resilient modulus
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

of compacted fine-grained subgrade soils. The method takes into account the in-
fluence of soil physical state, stress state, and soil type, The effect of soil physical
state is quantified by combinations of two equations relating resilient modulus to
moisture content. One equation is for paths of constant dry density and the other
is for iaaths of constant compactive effort. The effect of stress state is determined
by equations relating resilient modulus at optimum moisture content to deviator
stress so that the equation parameters represent the effect of soil type and its
structure. Means to estimate the resilient modulus at optimum moisture content
are suggested in the absence of actual test data. Examples of applications of this
method showed that it is simple and versatile and also gives consistency between
predicted resilient modulus and resilient modulus test results.

INTRODUCTION

The soil and aggregate layers forming the foundation of railway tracks
and highway pavements are subjected to r e p e a t e d traffic loading. U n d e r
individual cycles of loading, the layers behave essentially elastically, while
plastic deformation accumulates with r e p e a t e d cycles. This p a p e r considers
one of these properties, resilient modulus, which is recognized as the prop-
erty to characterize the elastic stiffness of soil and aggregate material.
Resilient modulus is usually d e t e r m i n e d by r e p e a t e d load triaxial tests
with constant confining pressure, a3, and with the deviator stress cycled
between the hydrostatic state and some positive deviator stress (~1 - (r3).
For these conditions, the term resilient modulus is defined as follows:

M r = Or_._.4 (1)
Er
where Mr = resilient, modulus; r d = r e p e a t e d deviator stress (r - r
e, = recoverable (i.e., resilient) strain in the direction of axial stress ql
(major principal stress) with confining stress r 3 (minor principal stress)
constant.
Resilient modulus as defined by (1) is equivalent to resilient Young's
modulus, which m a y be designated by the symbol Er.
Subgrade soil is an important c o m p o n e n t of highway-pavement and rail-
way-track support systems. A knowledge of the resilient modulus is nec-
essary for calculating resilient stresses, strains, and deflections in these lay-
ered systems as well as for analyzing the system performance.
Before 1986, studies of resilient modulus of various types of aggregates
~Sr. Res. Engr., Marston Hall/Civ. Engrg., Univ. of Massachusetts, Amherst, MA
01003.
2Prof. of Civ. Engrg., Marston Hall/Civ. Engrg., Univ. of Massachusetts, Amherst,
MA.
Note. Discussion open until November 1, 1994. To extend the closing date one
month, a written request must be filed with the ASCE Manager of Journals. The
manuscript for this paper was submitted for review and possible publication on
February 16, 1993. This paper is part of the Journal of Geotechnical Engineering,
Vol. 120, No. 6, June, 1994. 9 ISSN 0733-9410/94/0006-0939/$2.00 + $.25
per page. Paper No. 5632.

939

J. Geotech. Engrg. 1994.120:939-957.

 and soils were mainly motivated by the fact that the failure of highway
pavement could result not only from the excessive accumulated permanent
deformation of subgrade under repeated traffic loading, but also from the
fatigue cracking of the asphalt concrete surface caused by the repeated
resilient deformation. The characterization of resilient deformation of subgrade
soils under repeated stress application requires an understanding of their
resilient moduli. Since 1986 when AASHTO published a new design guide
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

(AASHTO 1986) for pavement structures that uses the resilient moduli to
characterize material properties, it has become essential for pavement design
to understand and quantify the characteristics of the resilient modulus of
each layer. Computer models for characterizing the resilient response of
railway track have also become of interest for analyzing track performance
and may play an important role in future maintenance decisions. The subgrade
resilient modulus is an essential parameter in these models.
Because of the large variability of subgrade soil properties and the cost
involved in obtaining and testing undisturbed soil samples, very little, if
any, resilient modulus testing is likely on any particular design project. In
the absence of data on the resilient modulus of a particular soil, a method
to predict the resilient modulus based on previous tests is desired. Fur-
thermore, a comprehensive study to develop such a method, which includes
an examination of many different test results, may reveal some of the char-
acteristics of resilient modulus and the factors that influence it.
This paper will present a new approach to estimate the resilient modulus
for fine-grained subgrade soils in an attempt to overcome some drawbacks
of existing prediction models and approaches. The data to develop and
verify this approach are based on many different test results on fine-grained
subgrade soils from the literature. Therefore the approach is generally ap-
plicable to fine-grained subgrade soils. Since both railway subgrade and
highway subgrade are subjected to similar stress and environmental actions,
the establishment and development of the principles and methods through
this research can be beneficial to both fields,

MODELS FOR RESILIENT MODULUS

As found in many different tests, the resilient modulus of fine-grained
soils is not a constant stiffness property, but is dependent upon many dif-
ferent factors. A change of resilient modulus from 14,000 kPa to 140,000
kPa (2,000-20,000 psi) for a fine-grained subgrade soil, which leads to a
dramatic difference of subgrade response under traffic loads, can result for
the same soil from the change of factors such as stress state or moisture
content.
Factors that have a significant influence on the magnitude of resilient
modulus can be grouped into three categories: (1) Loading condition or
stress state, which includes the magnitude of deviator stress and confining
stress, and the number of repetitive loadings and their sequence; (2) soil
type and its structure, which primarily depends on compaction method and
compaction effort for a new subgrade; and (3) soil physical state, which is
defined in this paper by moisture content and dry density and which is
subject to the change of environment. Strain level is also important, but in
this paper it is considered to be represented by stress level.
From the first category, the most important factor affecting resilient mod-
ulus is the repeated deviator stress. Although the resilient modulus increases
with increasing confining stress, it was found (Tanimoto and Nishi 1970;
Barksdale 1975; Fredlund et al. 1975; Townsend and Chisolm 1976) that
940

J. Geotech. Engrg. 1994.120:939-957.

 confining pressure has a much less significant effect than deviator stress for
fine-grained subgrade soils, especially clay soils. There is also some influence
from the number of stress applications, but the resilient modulus will tend
to become constant with increasing number of stress applications if the
deviator stress is below a certain level with regard to failure. Therefore,
constitutive relationships are primarily established between the resilient
modulus and the deviator stress for fine-grained subgrade soils.
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

Many different constitutive models have been proposed to simulate the

characteristic trend of resilient modulus with deviator stress for fine-grained
soils. The main ones are the following:

1. Bilinear model:

Mr = KI + K2crd when crd < crdi (2a)

Mr = /(3 + K4cr, when gd > crdi (2b)
where crag = deviator stress at which the slope of Mr versus cr, changes;/(1,
K2, /(3, and K, = model parameters dependent upon soil type and its
physical state (K2 and/(4 are usually negative).
This model was proposed by Thompson and Robnett (1976). A breakpoint
resilient modulus, Mri at crdi, was often used to characterize the resilient
properties of subgrade soils.
2. Power models:

Mr = kcr~ (3)
where k and n = parameters dependent on soil type and its physical state
(n is usually negative).
Moossazadeh and Witczak (1981) adopted this model and obtained good
agreement with test results on three fine-grained soils from San Diego,
Illinois, and Maryland with the determination of k = 0 to 200 and n =
- 1.0 to 0 for resilient modulus (ksi) (1 ksi = 6,895 kPa) and deviator stress
(psi) (1 psi = 6.895 kPa). Pezo et al. (1991) also used this model and
obtained a range of k = 6,000 to 55,000 and n = - 0 . 3 4 to - 0 . 0 4 for
Austin soil (A-7-6) for resilient modulus and deviator stress in units of psi.
Furthermore, Brown et al. (1975) and Brown (1979) proposed a similar
model, but with the consideration of effective confining stress (~r~) for sat-
urated overconsolidated soils as follows:

\cr;/ (4)
3. Semilog model:

Mr = 10 (k-"~) (5a)
or

log(Mr) = k - nor a (5b)

Fredlund et al. (1977) proposed this model for a moraine glacial till and
obtained the range of parameters k = 3.6 to 4.3 and n = 0.005 to 0.09 for
resilient modulus and deviator stress in units of kPa. Raymond et al. (1979)
941

J. Geotech. Engrg. 1994.120:939-957.

 proposed a similar model with replacement of X = (Td/(Td(failure) for Crd and
also got consistent results with tests on Leda clay.
4. Hyperbolic model:
k + ntr d
Mr - - - (6)
O"d
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

This model was proposed by D r u m m et al. (1990) for fine-grained soils.

For the Tennessee soils tested, the parameters ranged from k = 2-70 and
from n = 2-12 for resilient modulus (ksi) and deviator stress (psi).
5. Octahedral model:

Mr = k erect (7)
m
T oct

where %ct and Toot = octahedral normal and shear stresses. This model was
derived by Shackel (1973) and is more difficult to apply.

As indicated in the literature, each model was able to fit the relationship
between resilient modulus and stress state for the soils tested by the re-
spective researchers.
Comparisons have been made between the models 1 - 4 and test data as
shown in Fig. 1. The data adopted in Fig. 1 are typical of the resilient
modulus values for fine-grained subgrade soils from the test results by Seed
et al. (1962) and Thompson and Robnett (1979), respectively. Four models
are all fit to the actual data points shown in the figure. The results are
summarized in Table 1. All models can be considered to be able to represent
the test results by the fitting of model parameters, although there exists a
ranking of models in terms of their coefficients of correlation, as listed in
Table 1. The best representation in Fig. l(a) and Fig. l(b) is the bilinear
model, followed by the power model, then the semilog model, and last the
hyperbolic model.
The bilinear model gave the best fit but requires five parameters. The
power model .was second best and only requires two parameters. The re-
maining models showed no apparent advantages over the other two. Thus
the power model was chosen for primary consideration.
It is obvious that the correct prediction of resilient modulus using these
models is strongly dependent upon the values of the model parameters. As
will be discussed in the following section, the soil physical state has an
influence on the resilient modulus that is similar to or even greater than
that of stress state. Therefore these values of the parameters can be signif-
icantly dependent upon the soil moisture content and dry density in addition
to the soil type.

INFLUENCE OF SOIL PHYSICAL STATE ON RESILIENT MODULUS

The soil physical state is represented in this paper by two quantities:
moisture content and dry density. These two quantities can always be related
to the soil-compaction curve because a pair of values defines the coordinates
of a point on the compaction curves for a given compaction method. There-
fore the influence of soil physical state on the resilient modulus will be
established by means of the compaction curves.
Many test results indicating the influence of moisture content and dry
density on the resilient modulus of fine-grained subgrade soils are available
942

J. Geotech. Engrg. 1994.120:939-957.

 a d (kPa)
50 100 150 200
14000 I ] ] ]
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

12000 Eq.(2)
80
'~\ --- Eq.(3)
i0000
}',~ ......... Eq.(6)
~;',~ Correlations are 6O
8000 "{~-'\
-- l i s t e d in T a b l e 1
',,\ , v
6000 "% ,9 40

4000
~dm ......... ~
20
2000
(a) D a t a f r o m S e e d e t al ( 1 9 6 2 i
0 I I I I I I
0 5 10 15 20 25 30
~a (psi)
a d (kPa)
0 50 100 150
14000 I I ]

.,~.. Eq.(2)
12000 Eq.(3) 80
..... Eq.(5)
i0000 ......... Eq.(6)
'~, Correlations are 60 "-"
8000 "",,~~ed in T a b l e 1

6000 ~""--"-.
"A~ 40
v

4000
20
2000
(b)Data from Thompson e t al ( 1 9 7 9 )
0 i i i i l 0
0 5 I0 15 20 25
ad ( p s i )
FIG. 1. Comparisons of Different Models for Resilient Modulus Prediction

943

J. Geotech. Engrg. 1994.120:939-957.

 TABLE 1. Comparisons of Different Models for Resilient Modulus Prediction

Coefficients
Models of correlation
(1) (2)
(a) Data
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

Eq. (2) Mr = 109,000 - 1,000t~a, and M, = 18,100 + 15.4~d rz = 1.0

Eq. (3) Mr = 705,000o'd-~ r2 = 0.92
Eq. (5) Mr = 10~5~ ~176176 r2 = 0.89
Eq. (6) Mr = (1,630,000 + 12,900ad)/trd r2 = 0.89
(b) Data
Eq. (2) Mr = 118,000 - 1,450(r~, and Mr = 73,800 - 281crd r2 = 1.0
Eq. (3)Mr = 383,000o'a-~ r2 = 0.97
Eq. (5) Mr = 10(4"94-~176176
) /,2 = 0.95
Eq. (6) Mr = (1,330,000 + 27,4000"d)/O'd r2 = 0.91
Note: Both Mr and crd are in kilopascals (kPa).

in the literature. A systematic a p p r o a c h to consider their influences in pre-

dicting resilient modulus has been d e v e l o p e d in this study.
Soil physical state can be changed by the effect of environment and the
effect of compaction caused by traffic. The moisture content and dry density
varies with time in many different ways. H o w e v e r , any change of the soil
physical state can be r e p r e s e n t e d by a combination of two basic paths as
shown in Fig. 2, which are s u p e r i m p o s e d on curves of constant compactive
effort. Fig. 2(a) shows the paths of moisture content variation with constant
dry density, while Fig. 2(b) shows the paths of moisture content variation
with constant compactive effort for each test. The difference between these
two paths is due to the fact that the variation of dry density may also lead
to a significant change of resilient modulus. T h e consideration of the effect
of moisture content variation on the resilient modulus must be accompanied
by information on the variation of dry density. The influence of moisture
variation with and without the variation of dry density, as will be indicated
later, may be significantly different.
To reduce the effect of other factors on the relationships between resilient
modulus and the change of moisture content in terms of these two paths,
or in other words to m a k e these relationships relatively unique, the change
of resilient modulus for any individual soil from the literature at any physical
state is normalized by a reference resilient modulus obtained at o p t i m u m
moisture content and m a x i m u m dry density. It is felt that the choice of
resilient modulus at o p t i m u m moisture constant and m a x i m u m dry density
as a reference value is logical because of its importance in geotechnical
engineering. F u r t h e r m o r e , all o t h e r p a r a m e t e r s , including stress conditions
and soil type, are kept constant for the resilient moduli along each path
when either dry density or compactive effort is constant. In other words,
the change of resilient modulus is considered primarily caused by soil phys-
ical state along each path.
Following this approach, available r e p e a t e d load triaxial test results from
the literature on various fine-grained subgrade soils were used to develop
the equations that describe the relations between the resilient modulus and
the change of soil physical state.
Fig. 3 shows the relation between resilient modulus and moisture content
944

J. Geotech. Engrg. 1994.120:939-957.

 Reference points ~ 1 0 0 % Saturation

g
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

Moisture Content
(a)

Reference points \100% Saturation

Constant compactive effort \

Moisture Content
(b)
FIG. 2. Paths of Moisture Content Variation: (a) Constant Dry Density; and (b)
Constant Compactive Effort

with values of dry density the same as for the reference resilient modulus
as illustrated previously in Fig. 2(a). The correlation included 27 repeated
load triaxial test results on 11 fine-grained soils from the literature (Seed
et al. 1962; Sauer and Monismith 1969; Culley 1971; Robnett and Thompson
1976; Fredlund et al. 1977; Edil and Motan 1978; Kirwan et al. 1979; Elfino
and Davidson 1989). The best fit polynomial equation for these data is:
Rml = 0.98 - 0.28(w - Wopt) + 0.029(w - Wopt) 2 (8)

where R,,,1 = Mr/Mr(opt)for the case of constant dry density; Mr = resilient

modulus at moisture content w (%) and the same dry density as Mr(opt);
Mr(opt) = resilient modulus at maximum dry density and optimum moisture
content Wopt (%) for any compactive effort. The correlation coefficient, r 2,
is equal to 0.76.
Fig. 4 shows the relation between resilient modulus and moisture content
with the compactive efforts the same as for the reference resilient modulus
945

J. Geotech. Engrg. 1994.120:939-957.

 Rml
5
z~
A
4
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

AZ~
zx" z~

I , J , l , l , l , l m , I

-6 -5 -4 -3 -8 -1 1 2 3 4 5 6
(~-~ropt)(z)
FIG. 3. Relation between M, and w with Constant Dry Density

as illustrated previously in Fig. 2(b). The correlation included 26 repeated

load triaxial test results of 10 fine-grained subgrade soils from the literature
(Seed et al. 1962; Kallas and Riley 1967; Shifley and Monismith 1968;
Tanimoto and Nishi 1970; Edris and Lytton 1976; Fredlund et al. 1977;
Kirwan et al. 1979; Pezo et al. 1991). The best fit polynomial equation for
these data is:
Rm2 = 0.96 - 0.18(w - Wopt) + 0.0067(w - Wopt)2 (9)
where Rm2 = Mr~Mr(opt.) for the case of constant compaction effort; Mr =
resilient modulus at moisture content w (%) and the same compactive effort
as Mr(opt). The correlation coefficient, r 2, is equal to 0.83.
Correlations were also performed between Mr~Mr(opt)and the ratio W/Wopt
for both cases. The trends shown by these correlations are similar to the
trends as indicated by the aforementioned correlations. Because no better
coefficients of correlation than those of (8) and (9) were obtained, those
correlations are not presented in this paper.
As illustrated in Figs. 3 and 4, both cases indicate that an increase of
moisture content will lead to a significant decrease of soil stiffness. If a
value of 69,000 kPa (10,000 psi) for resilient modulus at optimum moisture
content is assumed, an increase of moisture content by 8% can lead to a
decrease of a resilient modulus f r o m a value of 280,000 kPa (40,000 psi) to
a value of 28,000 kPa (4,000 psi).
A comparison between these two cases shows that the case with constant
dry density causes a greater change of resilient modulus with change in
moisture content below optimum than the case with constant compactive
effort. The difference between the two cases becomes smaller when moisture

946

J. Geotech. Engrg. 1994.120:939-957.

 Rm2
5

4
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

z~ 8
A. ' ~ Az~

A z~
z~
I i I , I , I , I , I , T, , ,'~, ~, ,'?,
-6-5-4-3-8-1 0 1 8 3 4 5 6

(W--Wopt)(~*)
FIG. 4. Relation between Mr and w with Constant Compactive Effort

Water content,W, increases

Dry Density
FIG. 5. Influence of Dry Density on Resilient Modulus

content is above the optimum. This trend can be explained by the effect of
dry density on resilient modulus.
Fig. 5 illustrates the trend for the influence of dry density on the resilient
modulus based on tests by Seed et al. (1962). Whether the resilient modulus
increases, decreases, or does both with increase in dry density depends upon
the moisture content. In general, at lower moisture content the resilient
modulus tends to increase with increasing dry density, whereas at higher
moisture content the resilient.modulus tends to decrease with increasing dry
density.
The trends indicated in Fig. 5 can explain the differences between the
947

J. Geotech. Engrg. 1994.120:939-957.

 two cases as shown in Figs. 3 and 4. If the moisture content is low as below
optimum, a decrease of moisture content is accompanied by a decrease of
dry density along the path of constant compactive effort. According to the
trend shown in Fig. 5 for lower moisture content, a decrease of dry density
generally leads to a decrease of resilient modulus. Therefore, a decrease of
moisture content accompanied by a decrease of dry density along the path
of constant compactive effort should result in less increase of resilient mod-
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

ulus due to the opposite effects of dry density and moisture content on
resilient modulus than a decrease of moisture content along the path of
constant dry density. Similarly, if the moisture content is high, that is above
optimum, an increase of moisture content along the path of constant com-
pactive effort is accompanied by a decrease of dry density. According to
the trend shown in Fig. 5 for higher moisture content, a decrease of dry
density generally leads to an increase of resilient modulus. Therefore, an
increase of moisture content above optimum along the path of constant
compactive effort should lead to less decrease of resilient modulus due to
the offsetting influence of dry density on resilient modulus than an increase
Of moisture content along the path of constant dry density.
The two relatively unique correlations (8) and (9) between the resilient
modulus and the soil physical state for the two cases will form the basis to
predict the change of resilient modulus with a change in the soil physical
state.

PREDICTION OF RESILIENT MODULUS

As mentioned before, the resilient modulus is primarily dependent upon
three factors in most situations: (1) Stress state; (2) soil type and its structure;
and (3) soil physical state. As a result, the prediction of resilient modulus
should take into account these three aspects. The following two basic as-
sumptions are proposed to predict the resilient modulus:

1. The relationships between the resilient modulus and soil physical state
from (8) and (9) are of the form
Rrn 1 = fl(W -- Wopt) (10)
for w variation with constant dry density, and
Rm2 = f2(142 - Wopt) (11)
for w variation with constant compactive effort.
2. The bilinear model or the power model is established at optimum
moisture content and maximum dry density for any compactive effort. The
model parameters will only depend upon soil type and its initial structure.
For the same soil type and initial structure, they are constants that are not
dependent upon soil physical state. Because the power model [(3)] involves
fewer parameters than the bilinear model [(2)], the power model is used in
the remaining sections although the bilinear model is also discussed when
the concept of the breakpoint is used.

The following are the steps to predict the resilient modulus for fine-
grained subgrade soils:

1. Obtain resilient modulus at maximum dry density and optimum mois-

948

J. Geotech. Engrg. 1994.120:939-957.

 ture content Mr(opt) either directly from tests, or using (2) or (3) if model
parameters are available from previous tests on similar soil.
2. If the soil for which resilient modulus is needed has another optimum
moisture content corresponding to another compactive effort, or exists in
a physical state as defined by moisture content and dry density other than
that for step 1, then (8) or (9) or their combinations need to be used, as
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

explained in the following:

a. If the physical state of the soil for which resilient modulus is needed
differs from the known state in a manner shown in Fig. 2(a) or Fig.
2(b), then get the resilient modulus from either

Mr = RmlMr(opt) (12)
or

Mr = Rm2Mr(opt) (13)

b. If the physical state of the soil for which resilient modulus is needed
is neither on the same compaction curve nor at the same dry density
as that for which the resilient modulus is known from step 1, then
calculation of resilient modulus requires several steps.
Let us assume that Mr at point Q in Fig. 6 is needed and that Mr
at point O is known, i.e., conversion from one optimum to another
optimum. Path OQ is the sum of paths OA and AQ. First get Mr at
point A by

MrA = f l ( W 1 -- Wopt)Mr(optl ) (14)

Then get Mr at point Q by
MrA
(15)
MrQ ----f2(Wl -- Wopt2)

\
Reference points \ 1 0 0 % Saturation line
Q

Wl W2 W3 Wop~2Woptl
Water Content
FIG. 6. Paths for Determining Resilient Modulus at Any Soil Physical State

949

J. Geotech. Engrg. 1994.120:939-957.

 c. Repeat example b but estimate Mr at point P from Mr at point O.
Path OP is the sum of paths Oil, AQ, and QP. First get Mr at point
Q as in example b. Then get Mr at point P by
Mre = f2(w2 - Wopt2)MrQ (16)
d. Finally consider the most general case in which Mr at point C is known
and Mr at point P is desired. Path CP is the sum of paths CO, 0.4,
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

AQ, and QP. First get Mr at point O from Mr at point C by

Mrc
Mr(optl) -- fz(W3 _ Woptl) (17)

Then find Mr at point P using the steps in example c.

It can be seen that the application of method just discussed can give an
estimation of resilient modulus at any stress state and any soil physical state
by a straightforward procedure. The application of the method and com-
parisons with different test results will be shown later.

RESILIENT MODULUS AT REFERENCE POINT AND

MODEL PARAMETERS

In order to estimate resilient modulus of fine-grained soil at any soil

physical state under any stress condition by using the method discussed in
previous sections, a reference value of resilient modulus is required either
directly from tests or from (2) or (3). When (2) or (3) is used, the model
parameters, which are defined at optimum moisture content and are de-
pendent upon the soil type and its structure but not the soil physical state,
are needed to carry out the calculation.
It is logical to first find the resilient modulus at optimum moisture content
and maximum dry density as a reference value because of the common use
of this reference state in practice. For this study, values of these model
parameters of constitutive equation (3) for various soil types have been
compiled from the literature. The available data are mostly for compacted
soil specimens and thus the values of the model parameters compiled are
primarily for these compacted soils. If the Mr values from a test were not
at optimum moisture, then the Mr at optimum moisture was calculated using
the approach discussed in the preceding sections, as long as the moisture
content and compaction curve for the tested soils were available in the
literature. Values of the model parameters for 48 soil types, which were
back-calculated from available test results in the literature using (3), have
been compiled and are given in a separate research report (Li and Selig
1991). In general, for various fine-grained subgrade soils, the k values range
from 50,000 to 6,000,000 while the n values range from - 0 . 1 to - 0 . 9 , with
the unit of kPa for resilient modulus and deviator stress.
An equation developed by Thompson and LaGrow (1988) significantly
contributes to the approach developed in this study, although it can not
replace the compilation of model constants because it is just applicable to
a specific stress state: the breakpoint of the bilinear model. The following
is the relationship found by Thompson and LaGrow between the breakpoint
resilient modulus at optimum moisture content and 95% of maximum dry
density:
Mr(opt) = 30,800 + 677(% clay) + 821(PI) (18)
950

J. Geotech. Engrg. 1994.120:939-957.

 where Mr~op0 = resilient modulus (kPa) at optimum water content and 95%
maximum dry density for A A S H T O T-99, % clay = % particles finer than
2 Ixm, and PI = plasticity index.
By application of the correlation in (18) together with (8) and (9), the
resilient modulus at any physical state for any soil type can be evaluated
for that specific stress state, i.e., the breakpoint. In other words, the resilient
modulus at this specific stress state for any soil type under any physical state
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

can be determined by the following steps: (1) Perform index tests and
compaction tests to obtain the percentage of clay, the plasticity index, and
the compaction curve; (2) determine the resilient modulus at optimum mois-
ture content using (18); (3) compare the soil moisture content and dry
density at which the resilient modulus is desired with known Woptand %'a(m~);
and (4) use (8) or (9) or their combinations in terms of the paths of variation
of soil physical state to evaluate the desired resilient modulus.

EXAMPLES OF NEW METHOD

Fig. 7 shows the standard compaction curve for a clay soil (called "Vicks-
burg clay" or "Buckshot clay"). Suppose that for the three different physical
states designated by A, B, and C, resilient modulus values are to be derived.
For the method developed in this study to predict resilient modulus for these
three different physical states, the prediction paths are illustrated in this
figure by arrows. The point O is the reference point, corresponding to the
resilient modulus at optimum moisture content and maximum dry density,
which is determined by (3) with the adoption of the model parameters k =
621,000 and n = - 0 . 4 2 for resilient modulus and deviator stress in units
of kPa (Li and Selig 1991). These model parameters at the reference point
were compiled from the tests by Thompson (1990) on Buckshot clay. Thus,
Mr(opt) -- 621,000 O'd -0"42. For the prediction of the resilient modulus at point
A from the reference resilient modulus at O, the path OA is adopted similar

I I 1.68

Vicksburg clay
I00
zero air void line 1.58

90
/ < 1.48 >..

(D

1.38
t....,

80 , I , I , I , 1.28
15 20 25 30 35
Water Content (7o)
FIG. 7. Typical Compaction Curve and Prediction Paths of M,

951

J. Geotech. Engrg. 1994.120:939-957.

 to the path illustrated in Fig. 2(a). Eq. (8) is then used to obtain the Mr at
the state A. For the prediction of Mr at the state B or C, the path OB or
OC consists of two parts, OB' or OC' similar to the path illustrated in Fig.
2(b), and B ' B or C'C similar to the path in Fig. 2(a). Both (9) and (8) are
then used to obtain the Mr at the state either B or C.
Fig. 8(a) shows the comparisons of predicted resilient modulus from the
procedure just discussed with the resilient modulus measured by Thompson
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

(1990) for this Vicksburg clay using repeated load triaxial tests for three
different soil physical states under different stress states. As can be seen,
the agreement between predicted and measured resilient modulus is rea-
sonable.
Fig. 8(b) shows further comparisons of predicted resilient modulus using
the approach developed in this study and tested resilient modulus on the same
Vicksburg clay by Townsend and Chisolm (1976). Repeated load triaxial tests
were used for the soil specimens compacted with two different moisture con-
tents (w = 24% and 27%, respectively), but on the same compaction curve.
The data points shown in the figure are the results from the tests for these
two different soil states. For the prediction of resilient modulus, the reference
values of resilient modulus at optimum moisture content (w = 22%) and
maximum dry density were determined through (3) with model parameters of
k = 5,980,000 and n = - 0 . 7 6 for resilient modulus and deviator stress in
units of kPa (Li and Selig 1991). Again, these model parameters at the ref-
erence point were compiled from the tests by Townsend and Chisholm on
buckshot clay. Thus, Mr(opt) = 5,980,000 O'd-0"76. Then the prediction path as
illustrated in Fig. 2(b) was used with the application of (9) for the prediction
of the resilient modulus at these two states. The predicted results are also
shown in the figure. The predicted and the tested values agree well with each
other. Moreover, a comparison between Figs. 8(a) and 8(b) indicates that
tested results by two different agencies showed similar trends of resilient mod-
ulus change caused by both soil physical state and stress state for this same
type subgrade soil.
Fig. 9 shows the repeated load triaxial test results and compaction test
results on the A A S H T O subgrade soil by Seed et al. (1962). The upper part
of this figure shows the variation of resilient strain with different water
content under different compactive efforts by kneading compaction at a
deviator stress of 69 k P a (10 psi). From the upper figure, the test resilient
modulus can be determined in terms of (1), i.e., Mr = 10/er at any point
on this upper part. The lower part of this figure shows the corresponding
soil physical state, i.e., water content and dry density, for different com-
pactive efforts.
In addition to the test results shown in this figure, the prediction paths
of resilient modulus through the method developed in this study are also
shown in the figure. Point A is taken as a reference point for the prediction
of resilient modulus of other points such as B, C, D, and E. As indicated
by these paths, only (8) for constant dry density path is required to predict
the resilient modulus at points B or C, while both (8) and (9), the second
of which is for constant compactive effort path, are required for points D
or E, i.e., the prediction path for D or E consists of A B and BD, or A C
and CE.
The test results of resilient moduli based on the resilient strains through
(1) at these points A ' , B', C', D ' , and E' are listed in Table 2. The prediction
results of resilient moduli at these points B, C, D, and E based on the
952

J. Geotech. Engrg. 1994.120:939-957.

 ~d (kPa)
0 80 40 80 80 100 120
10000
! predicted by new method 3'
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

- 0 test (w=28~ Td=1 952 Mg/m 3 )

~ e test (w=30% T d = l . 4 7 Mg/m3) 60
8000 O~est (w=32Vo 7d=1.43 Mg/m ) 50

"-- 6000
9 0 40

30 .
~ 4000

2000 ~ 10

(a) Data from Thompson (1990)

0 i i i I i i i i 0
0 8 4 6 8 10 12 14 16 18
Grd (psi)
~d (kPa)
0 50 100 150
40000
' predicted by new method_
9 test ( w = 2 4 % 7 =1 55 M g / m ~) 250
35000 d " 3
9 test (w=27N 7d=1.52 Mg/m )
30000
200
"~
9 25000 9 @

150
80000
%

15000 100

10000 v ~ ~ ~ _ ~ 9 50

5000
(b) Data from Townsend et al (1976]
0 i i i I i 0
0 5 10 15 20 25
~d (psi)
FIG, 8. Comparison of P r e d i c t e d and Measured Resilient Modulus

953

J. Geotech. Engrg. 1994.120:939-957.

 0.7
c 3 = 24 kPa (3.5 psi)
~d = 69 kPa (10 psi)
0.6

9I 05
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

"~'~ 0.4

0.3

~ 0.2

o.1 .Z J: ;B' ',

128 ' ', 2.02

124 ', , ,',
, 1.96
120 / ; ~ . . ~ % ~, ~-

116
/ ~ E ~ ' ~ . ~
1.841.90

"~
112~ / / ~ ~
104108 17.81~17.26.6
"~
~r

100 / Data from Seed, et al (1962) 1.60

9~ I I i J i I t I 1.54
10 11 12 13 14 15 16 17 18 19 20
Water Content (%)
FIG. 9, Relationship between Dry Density, Moisture Content, and Resilient Strain
with Superimposed Prediction Paths

TABLE 2. Comparison of Predicted and Tested Resilient Modulus

Soil physical Tested Mr Predicted Mr
state (psi) (kPa) (psi) (kPa)
(1) (2) (3) (4) (5)
Aa 3,000a 21,000 a 3,100 a 21,000 a
B 7,100 49,000 5,500 38,000
C 12,000 83,000 8,300 57,000
D 4,000 28,000 4,800 33,000
E 6,900 48,000 6,500 45,000
aReference resilient modulus at optimum water content.

954

J. Geotech. Engrg. 1994.120:939-957.

 reference value of resilient modulus through (8) and (9) are also listed in
Table 2. The resilient modulus at the reference point A is determined directly
from the resilient strain at point A' through (1). Again, as can be seen from
Table 2, the predicted resilient moduli for different physical states of this
AASHTO test subgrade soil agree well with the measured resilient moduli.
Overall, as can be found in all these cases, the application of the new method
for the prediction of resilient modulus for different soil physical states under
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

different stress states is relatively simple and gives good predictions.

SUMMARY AND CONCLUSIONS

A method for the prediction of resilient modulus of fine-grained soils has
been developed, which can take into account the influences of s0il physical
state, soil stress state, and soil type.
The relationship between resilient modulus and deviator stress is estab-
lished at a reference point corresponding to optimum moisture content and
maximum dry density for any compactive effort. Thus, the model parameters
can be assumed to be a function only of the soil type and its structure. As
a result of comparisons made among different models presented in the
literature, the power model was selected for the representation of the re-
lationship between resilient modulus and deviator stress, even though the
bilinear model is also recommended when the concept of resilient modulus
at breakpoint is used.
The influence of the variation of soil physical state on resilient modulus
has been quantified by two correlations developed based on many different
test results from the literature on various fine-grained subgrade soils. The
prediction of resilient modulus for a given soil at any soil physical state can
be obtained through the use of these two equations and their combinations
based on the reference resilient modulus for that soil.
The proposed method overcomes a deficiency of previously used models
that combine the effects of soil type and soil physical state in the model
parameters. In this previous approach, there is no way to assess the effect
of changing soil physical state, which is significant. This is an important
limitation because the soil physical state obviously is subject to change with
environment.
The applications of the method are illustrated and prove to be straight-
forward and simple. The comparisons between the predicted and tested
resilient moduli for several tests indicate the versatility of this approach and
the consistency of predicted results with test results. Although the approach
and principles developed in this research may have general application, the
correlations and parameters developed and compiled were based on com-
pacted fine-grained subgrade soils.

ACKNOWLEDGMENTS
The research presented in this paper was sponsored by the Association
of American Railroads (AAR) as part of the A A R geotechnical fellowship
program at the University of Massachusetts, under the cognizance of Dr.
A. J. Reinschmidt, A A R assistant vice president. Dr. M. R. Thompson,
professor of civil engineering at the University of Illinois at Urbana provided
useful information for this study. The authors also thank the many re-
searchers whose test results were used in this research.
955

J. Geotech. Engrg. 1994.120:939-957.

 APPENDIX I. REFERENCES
AASHTO guide for design of pavement structures. (1986). American Association of
State Highway and Transportation Officials, Washington, D.C.
Barksdale, R. D. (1975). "Test procedures for characterizing dynamic stress-strain
properties of pavement materials." Spec. Rep. 162, Transp, Res. Board, Wash-
ington, D.C.
Brown, S. F. (1979). "The characterization of cohesive soils for flexible pavement
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

design." Proc. Design Parameters in Geotech. Engrg., British Geotechnical So-

ciety, London, England, 2, 15-22.
Brown, S. F., Lashine, A. K. F., and Hyde, A. F. L. (1975). "Repeated load triaxial
testing of a silty clay." Geotechnique, 25(1), 95-114.
Culley, R. W. (1971). "Effect of freeze-thaw cycling on stress-strain characteristics
and volume change of a till subjected to repetitive loading." Can. Geotech. J.,
8(3), 359-371.
Drumm, E. C., Boateng-Poku, Y., and Pierce, T. J. (1991). "Estimation of subgrade
resilient modulus from standard tests." J. Geotech. Engrg., ASCE, 116(5), 774-
789.
Edil, T. B., and Moran, S. E. (1979). "Soil-water potential and resilient behavior
of subgrade soils." Transp. Res. Rec., 705, 54-63.
Edris, E. V., Jr., and Lytton, R. L. (1976). "Dynamic properties of subgrade soils,
including environmental effects," TTI-2-18-74-164-3, Texas Transp. Inst., Texas
A&M Univ., College Station, Tex., May.
Elfino, M. K., and Davidson, J. L. (1989). "Modeling field moisture in resilient
moduli testing." Resilient moduli of soils: laboratory conditions, ASCE Geotech.
Special Publication, No. 24, ASCE, New York, N.Y., 31-51.
Fredlund, D. G., Bergan, A. T., and Sauer, E. K. (1975). "Deformation charac-
terisation of subgrade soils for highways and runways in northern environments."
Can. Geotech. J., 12(2), 213-223.
Fredlund, D. G., Bergan, A. T., and Wong, P. K. (1977). "Relation between resilient
modulus and stress conditions for cohesive subgrade soils." Transp. Res. Rec.,
642, 73-81.
Kallas, B. F., and Riley, J. (1967). "Mechanical properties of asphalt pavement
materials." 2nd Int. Conf. on Struct. Design of Asphalt Pavements, University of
Michigan, 931-952.
Kirwan, R. W., Farrell, E. R., and Maher, M. L. J. (1979). "Repeated load param-
eters of a glacial till related to moisture content and density." 7th European Conf.
on Soil Mech. and Found. Engrg., England, 2, 69-74.
Li, D., and Selig, E. T. (1991). "Evaluation of resilient modulus for fine-grained
subgrade soils." Geotech. Report No. AAR91-392R, Dept. of Civ. Engrg., Univ.
of Massachusetts, Amherst, Nov.
Moossazadeh, J., and Witczak, M. W. (1981). "Prediction of subgrade moduli for
soil that exhibits nonlinear behavior." Tramp. Res. Rec. No. 810, 9-17.
Oregon Department of Transportation. (1990). "Proceedings of the workshop on
resilient modulus testing." FHWA-TS-90-031, Oregon State Univ., Corvallis, Oreg.,
Mar.
Pezo, R. F., Kim, D.-S., Stokoe, K. H., and Hudson, W. R. (1991). "Aspects of a
reliable resilient modulus testing system." Transp. Res. Board Preprint, 70th An-
nual Meeting, Washington, D.C., Jan.
Raymond, G. P., Gaskin, P. N., and Addo-abedi, F. Y. (1979). "Repeated com-
pressive loading of Leda clay." Can. Geotech. J., 16(1), 1-10.
Robnett, Q. L., and Thompson, M. R. (1976). "Effect of lime treatment on the
resilient behavior of fine-grained soils." Transp. Res. Rec., 560, 11-20.
Sauer, E. K., and Monismith, C. L. (1968). "Influence of soil suction on behavior
of a glacial till subjected to repeated loading." Hwy. Res. Rec., 215, 8-23.
Seed, H. B., Chan, C. K., and Lee, C. E. (1962). "Resilience characteristics of
subgrade soils and their relation to fatigue failures in asphalt pavement." Proc.
First Int: Conf. on Struct. Design of Asphah Pavements, University of Michigan,
Ann Arbor.
Seed, H. B., Chan, C. K., and Monismith, C. L. (1955). "Effects of repeated loading
956

J. Geotech. Engrg. 1994.120:939-957.

 on the strength and deformation of compacted clay." Highway Res. Rec., 34, 541-
558.
Shackel, B. (1973). "The derivation of complex stress-strain relations." Proc. 8th
Int. Conf. on Soil Mech. and Found. Engrg., Moscow, 353-359.
Shifley, L. H., Jr., and Monismith, C. L. (1968). "Test road to determine the
influence of subgrade characteristics on the transient deflections of asphalt concrete
pavements." Report No. TE 68-5, Office of Res. Services, Univ. of California,
Berkeley, Aug.
Downloaded from ascelibrary.org by New York University on 05/11/15. Copyright ASCE. For personal use only; all rights reserved.

Tanimoto, K., and Nishi, M. (1970). "On resilience characteristics of some soils
under repeated loading." Soils Found., 10(1), 75-92.
Thompson, M. R. (1990). "Results of resilient modulus of Vicksburg clay for FAST
test center." Internal Report for American Association of Railroads, University of
Illinois, Urbana, II1., Dec.
Thompson, M. R., and LaGrow, T. G. (1988). "A proposed conventional flexible
pavement thickness design procedure." FHWA-IL-UI-223, Illinois Cooperative
Highway and Transportation at Urbana-Champaign, Dec.
Thompson, M. R., and Robnett, Q. L. (1976). "Final report, resilient properties of
subgrade soils." FHWA-IL-U1-160, Univ. of Illinois, Urbana, I11, June.
Thompson, M. R., and Robnett, Q. L. (1979). "Resilient properties of subgrade
soils." J. Transp. Engrg., ASCE, 105(1), 71-89.
Townsend, F. C., and Chisolm, E. (1976). "Plastic and resilient properties of heavy
clay under repetitive loadings." Tech. Rep. S-76-16, Soils and Pavement Lab.,
U.S. Army Engrg. Waterways Experiment Station, Vicksburg, Miss., Nov.

APPENDIX II. NOTATION

The following symbols are used in this paper:

k = model p a r a m e t e r used for various models;

Mr = resilient modulus;
Mri = breakpoint resilient modulus;
Mr(opt)= resilient modulus at reference;
n = model p a r a m e t e r used for various models;
PI = plasticity index;
Rml = Mr/Mr(opO for case of constant dry density;
Rm2 = Mr/Mr(op 0 for case of constant compactive effort;
w = water content;
Wopt = o p t i m u m water content;
~/d ---- dry density;
Er = resilient strain in direction of axial stress;
ere = r e p e a t e d deviator stress;
or, = axial stress ( m a j o r principal stress);
cr3 = confining stress (minor principal stress);
cr~ = effective confining stress;
~rdi = deviator stress at which slope of Mr versus ~d changes;
t~oct = octahedral n o r m a l stress; and
%ct = octahedral shear stress.

957

J. Geotech. Engrg. 1994.120:939-957.