Snethen Expansive Soil PDF
Snethen Expansive Soil PDF
Snethen Expansive Soil PDF
June 1977
Interim Report
Document is available to the public through
the National Technical Information Service,
Springfield, Virginia 22161
Prepared for
FEDERAL HIGHWAY ADMINISTRATION
Offices of Research & Development
;~ ington, D. C. 20590
~OTICE
The contents of this report reflect the views of the U. S. Army Engineer
Waterways Experiment Station, which is responsible for the facts and the
accuracy of the data presented herein. The contents do not necessarily
reflect the official views or policy of the Department of Transportation.
This report does not constitute a standard, specification, or regulation.
FHWA~RD-77..94
~
FHVTA Contract Manager--Carl Ealy (HRS-21)
16. Abstract
19. Security Claosil. (of !hi$ roport) 20. Sec<~riiy Clossif. (of this page) 21. No. of Pages 22. Price
Unclassified Unclassified 48
Reproduction of completed pt:!ge authori l<ed
PREFACE
ii
TABLE 'OF CONTENTS
PREFACE . . . . . . ii
CONVERSION FACTORS, U. S. CUSTOMARY TO METRIC (SI) UNITS
OF MEASUREMENT v
INTRODUCTION 1
REVIEW OF IDENTIFICATION TECHNIQUES 2
Purpose of Identification Techniques 2
Definition of Potential Swell 3
Summary of Indirect Techniques . . 4
EVALUATION OF IDENTIFICATION TECHNIQUES 8
Selection of Identification Techniques 8
Field Sampling and Laboratory Testing 16
Analysis of Data 18
SUMMARY •• 36
RECOMMENDED USAGE 38
REFERENCES 41
LIST OF FIGURES
LIST OF TABLES
iii
7. Comparison of WES Categories and Classification by Pub-
lished Identification Techniques for 20 Sampling Sites . 27
8. Summary of Independent Variables Used in Statistical
Comparisons . . . . . . . . . . . . . . . . . . . . 29
•
9. Summary of Correlation Coefficients Greater (or Less)
Than +0.7 (-0.7) from Statistical Comparisons . . . . 32
iv
CONVERSION FACTORS, U. ·s. CUSTOMARY TO METRIC (SI)
UNITS OF MEASUREMENT
Multiply By To Obtain
v
AN EVALUATION OF EXPEDIENT METHODOLOGY FOR IDENTIFICATION
OF POTENTIALLY EXPANSIVE SOILS
INTRODUCTION
1
r
REVIEW OF IDENTIFICATION TECHNIQUES
.I
I
Purpose of Identification Techniques
2
Since the purpose of this report is to evaluate identification tech-
niques, two of the three categories, i.e. indirect and combination, will
provide the methodology for evaluation.
3
the best simulation of in situ conditions practical. At a minimum, the
definition should specify the initial conditions of the specimen, such
as water content, dry density, fabric, and structure, as well as the
stress conditions relative to the specimen, such as vertical stress and
lateral confinement conditions. Ideally, the amount and rate of water
applied to the specimen should simulate actual conditions such as ground-
water influences and surface infiltration. The current state .of the art
allows for simulation of most of these conditions, except compromise is
required for the lateral confinement and water application conditions.
With this in mind, the definition of potential swell which satisfies the
largest portion of the field simulation requirements is:
Potent,ial swell is the equilibrium vertical volume change
or deformation from an odometer-type test (i.e. total lat-
eral confinement), expressed as a percent of original
height, of an undisturbed specimen from its natural water
content and density to a state of saturation under an ap-
plied load equivalent to the in situ overburden pressure.
4
Soil composition indicator group
7. The soil composition indicator group is totally concerned
with the determination of the type and, to a lesser extent, amount of
clay mineral present in the soil. The identification of the clay min-
eral is positive indication of potential problem; however, the clay
mineralogy does not present an adequate correlatfon with respect to the
amount of volume change that is likely to occur. A point that is of
more concern from the practical operational standpoint is that most of
the methodology (i.e. equipment and trained personnel) is not readily
available on a cost-effective basis for routine use. In summary, the
determination of the clay mineralogy is a reasonably accurate indicator
that problems could exist; however, very little can be determined about
the relative magnitude of the problem. This lack of correlation with
potential swell and the logistics problems with routine use have ham-
pered and will continue to hamper the use of the methods included in the
soil composition indicator group.
Physicochemical
property indicator group
8. The physicochemical property indicator group provides simpler
testing procedures than the soil composition group; however, the sim-
plicity is offset by poor correlations with measured volume change. The
published relationships show, at best, a general qualitative indication
of potential swell. In other words, the magnitude of potential swell
generally increases with increasing cation exchange capacity and is in-
fluenced by the type and amount of cation present.
Physical property indicator group
9. The physical property indicator group provides several proper-
des which have significant influence on the volume change characteristic,
but individually the properties provide little or no indication of poten-
tial swell from a relative magnitude standpoint. The colloidal content
2
has been used in combination with other index properties to categorize
potential swell. The other properties--specific surface area, soil
fabric, and soil structure--influence the amount and rate of volume
1
change. The influences are explained in more detail elsewhere.
5
Index property indicator group
10. The most widely used indicator group for identification/
classification of expansive soils is the index properties group. The
major factors which result in the popularity of this group are (a) the
practicality from the standpoint that most of the properties involved
are routinely determined by all State Highway Agencies and (b) experi-
ence has shown that potential swell correlates reasonably well with sev-
eral of the simple index properties. In most cases, the index property
group variables, i.e. Atterberg limits or shrinkage properties, are cor-
related with past experiences and used individually to categorize the
potential problem from expansive soils. Examples of these simple cate-
gorizations are given in Table 5, Reference l. On occasions where mea-
sured potential swell values are available, the index properties are
used individually or combined to correlate with potential swell. The
result is several multiproperty categorizations of relative magnitude of
potential volume change. This constitutes the combination category
previously discussed. This property indicator group includes the largest
number of published identification/classification techniques which will
be considered in the evaluation of methodology.
Soil classification indicator grbup
11. The soil classification system indicator group is another
group which deals in simple properties of the soils, but does not pro-
vide more than a general indication that a problem might exist. In the
American Association of State Highway and Transportation Officials
(AASHTO) classification system, the A-6 and A-7 subgrade groups consti-
tute a majority of the potentially expansive soils. In the Unified Soil
Classification System (USCS), the CL and CH categories generally cover
the range of potential expansivity with the possibility of some MH soils
showing expansive characteristics to a lesser degree. The Soil Conserva-
.
tlon S.ervlce
. Cl assl"f"leat"lon Sys t em or SOl"l Taxonomy B-ll as l"t lS
· now
called, is the most detailed classification system in current use. Soil
taxonomy uses many of the basic properties as well as temperature and
moisture regimes and climate to describe soils. In soil taxonomy, the
vertisol order includes all of the expansive soils. Within the vertisol
6
order, the major suborders are Torrerts, Uderts, Usterts, and Xererts.
The system is based on several formative elements which provide indi-
vidual meaning to the total descriptive term. In the vertisol order,
the basic formative element is "ert"; therefore, when these three
letters appear in a soil descriptive name, that material is generally
considered to be potentially expansive.
7
EVALUATION OF IDENTIFICATION TECHNIQUES
Louisiana Depart-
12
ment of Transportation
14. The Louisiana Department of Transportation uses the Atterberg
limits (liquid limit (LL) and plasticity index (PI)) balanced with field
experience to identify potentially expansive soils. The criteria used
for identifying and classifying potential swell are:
8
Potential Swell
LL, % PI, % . Classification
. hway
Kansas H1g comm1ss1on
. . 13
15. The Kansas Highway Commission (KHC) also uses the Atterberg
limits (PI) to indicate potentially expansive soils. In addition, KHC
generally follows up with an odometer-type volume change test (1-psi
surcharge) to better estimate the quantity of anticipated swell. The
KHC criteria for potential swell are:
Potential Swell
PI, % Classification
<15 Low or none
15-35 Moderate
>35 High
14
Raman
16. This method uses the Atterberg limits (LL, PI, and shrinkage
limit (SL)), but in a different configuration. Baman defines the shrink-
age index (SI) as the difference between the LL and the SL (LL- SL).
The criteria recommended by Raman for identification/classification are:
Degree of
PI 2 % SI 2 % Expansion
<12 <15 Low
12-23 15-30 Medium
23-32 30-40 High
>32 >40 Very high
Sowers 15 , 16
17.Sowers' early work1 5 used only the PI as an indicator of
16
potential swell. In his later work he combines the PI with the SL to
provide additional accuracy. The criteria are:
Potential
SL 2 % PI 2 % Volume Change
>12 <15 Probably low
10-12 15-30 Probably moderate
<10 >30 Probably high
9
One interesting note in Sowers' work is his reference to the water-
plasticity ratio or liquidity index (LI). His data indicate that little
swell will occur when the soil moisture reaches a value which results in
a LI of 0.25.
Dakshanamurthy and Raman 1 7
18. This method is based on a modification of Casagrande's
plasticity chart, which includes PI and LL, with the addition of the SI
(LL- SL). Figure 1 is a graphical representation of the recommended
criteria. A simplification of the procedure using the LL is:
Potential Swell
LL, % Classification
0-20 Nonswelling
20-35 Low swelling
35-50 Medium swelling
50-70 High swelling
70-90 Very high swelling
>90 Extra high swelling
60 60
50 50
...z ...z
UJ UJ
u u
ll ll
UJ UJ
.._ 40 400.
)( x·
UJ UJ
0 0
z
~
~ 1 30 30 ~
Q <
... "
z
"'<J it
.. 20 20
ili
10 10
0
LIQUID LIMIT, PERCENT
10
18
Anderson and Thomson
19. The authors of this method use the PI alone, but rather than
qualitatively describe the degree of expansion alone, they add a cate-
gorization of measured potential swell. The recommended criteria are:
Potential Degree of
PI, % Swell, % Expansion
<20 <1. 5 Low
20-31 1. 5-4.0 Medium
31-39 4.0-6.0 High
>39 >6.0 Very high
Ranganatham
19
and Satyanarayana
20. This method uses the Atterberg limits (LL and SL) in combina-
tion to define the SI (LL - SL). These. authors were the first to use
the SI for identification of expansive soils. The criteria are:
Potential Swell
SI, % Classification
<20 Low
20-30 Medium
30-60 High
>60 Very high
20
Saito and Miki
21. This method defines the plasticity ratio (PR) as the PI di-
vided by the PL and uses this ratio to correlate with measured swell.
The corresponding criteria for potential swell are:
Potential
Volume Potential Swell
PR Change 2 % Classification
<0.6 <3 Low
0. 6-1.0 3-10 Medium
1. 0-2.0 10-50 High
>3.0 >50 Very high
2
U. S. Blireau of Reclamation (USBR)
22. This method involves direct correlation of observed volume
change with colloidal content, PI, and SL. The degree of expansion and
limits of correlated properties are shown in the following tabulation:
11
Data from Index Tests Probable
Colloid Content Expansion
%-1 lJm PI, % SL~ % % Expansion
<15 <18 >15 <10 Low
13-23 15-28 10-16 10-20 Medimn
20-31 25-41 7-12 20-30 High
>28 >35 <11 >30 Very high
Experience has shown that this method correlates reasonably well with
expected behavior and provides a good indicator of potential volmne
change. The major criticisms of the method are that the colloidal con-
tent indicates amount but not the type of clay constituents and that the
hydrometer test is not a routine test in many agency laboratories.
21
Altmeyer
23. In a discussion to Holtz's paper presenting the USBR method,
Altmeyer brought out the major criticisms of the method and suggested a
method based on correlations between percent swell and the SL and linear
shrinkage. The results of his recommendations are as follows:
Linear
Shrinkage Probable Degree of
% SL 2 % Swell 2 % Ex:Eansion
<5 >12 <0. 5 Noncritical
5-8 10-12 0. 5-l. 5 Marginal
>8 <10 >1. 5 Critical
One minor criticism of Altmeyer's method is its lack of application to
in situ behavior since the data were collected on remolded samples.
-3 4
Seed 2 Woodward, and Lundgren '
24. The potential swell of an expansive soil is defined from cor-
relations of percent swell from odometer tests using laboratory prepared
and compacted samples with percent clay size (-2 llm) and soil activity.
A statistical relationship is defined for potential swell in terms of
clay content and activity and compared with measured volmne change. The
potential swell may be categorized as follows:
12
22
Chen
25. In an effort to simplify the USBR method (i.e. eliminate the
need for hydrometer analysis) and to provide some relative measure of
soil density, a correlation was made between odometer swell data and
percent passing the No. 200 sieve, LL, and standard penetration resis-
tance. The resulting classification of the degree of expansion is as
follows:
23
Vijayvergi;y:a and Ghazzal;y:
26. The method defines a swell index for an expansive soil as
the ratio of the natural water content w. to the LL and correlates it
l
with odometer swell and swell pressure data. Rather than a specific de-
gree of expansion, limits of probable swell and swelling pressure are
defined as shown in-the following tabulation:
24
Vijayvergi;y:a and Sullivan
27. The method is correlation of odometer swell data with LL
and dry density. Here again, degree of expansion is not defined; in-
stead, a family of curves relates the parameters with quantitative vol-
ume change (Figure 2). The basic material data for the correlation are
good; however, experience with application of the system is somewhat
limited.
13
II.
u
D..
f.-
l:
C)
iii
~
f.-
z tOO
:>
a::>-
0
95
75~------~-------L--------L-------~------J-------~--------L-----~
75
45 50 55 60 65 70
35 40
LIQUID LIMIT, PERCENT
14
25
Sorochan
28. The correlation involves relating the swelling index (void
ratio, e , after free expansion divided by the initial sample void
ratio, e ) to the PI. The resulting degrees of expansion with regard
0
to correlated parameters are as follows:
where
sp = predicted swell percentage
c = clay content, percent
w.
l
= initial moisture content
and
2
p = (3.5817 X 10-2)(PI)l.l2 f_ + 3.7912
p 2
w.
l
where p
p = predicted swelling pressure, psi
15
2
Komornik and David 7
30. This is another statistical comparison of measured data which
provides a relationship for predicting swelling pressure using 11, natu-
ral dry density w. The relationship for predicted swelling
l
pressure is:
2
The dry density and swelling pressure are in kg/m 3 and kg/cm ,
respectively.
26 27
31. The last two procedures ' are actually empirical predic-
tion techniques, but were included in the evaluation of identification/
classification since they represent correlations between Atterberg
limits, physical properties, and measured volume change. Since no cate-
gories of volume change are presented with the techniques, comparisons
will be made between the predicted and measured swell or swell pressure.
16
Table 1
Summ~ of General Site Information and Soil Classifications
Natural Natural
Boring/ Moisture Dry
Site Sample No./ Content Density Soil Classification
No. Site Location Geolosic Formation De,Eth 1 ft % ,ECf uses AASHTO
1 Jackson, Miss. Yazoo U-2/1/1.0-3.2 40.3 80.3 CH A-T-5 (81)
2 Hattiesburg, Miss. Hattiesburg U-2/1/1.0-2.9 25.6 93.8 CH A-T-6 (36)
3 Monroe, La. Alluvial material U-2/2/3.5-4.9 39-5 81.8 CH A-T-5 (68)
4 Lake Charles, La. Prairie terrace material U-2/1/1.0-3.1 20.6 105.6 CH A-T-6 (33)
5 San Antonio, Tex. Taylor U-2/4/3.5-5.1 24.0 95.4 CH A-T-5 (35)
6 Vernon, Tex. Vale U-l/4/4.8-T.2 13.8 119.T CL A-6 (11)
T Durant, Okla. Washita U-2/2/3.5-4.T 15.4 llT.9 CL A-T-6 (30)
8 Hennessey, Okla. Hennessey U-1/2/3.5-5.6 12.5 124.9 CL A-T-6 (25)
......
-.;J 9 Holbrook, Ariz., No.1, I-40 Chinle U-2/1/2.5-4.2 9.4 lll.5 CL A-6 (10)
10 Holbrook, Ariz., No. 2, SH 180 Chinle U-2/1/2.0-4.3 19.3 101.5 CH A-T-6 (24)
11 Price, Utah Mancos U-1/1/1.2-3.2 9.4 10T.8 CL A-T-6 (21)
12 Hays, Kans. Blue Hill U-2/1/1. 4-3.4 26.1 9T.2 CH A-T-6 (58)
13 EJ.Jlsworth, Kans. Graneros U-2/1/2.0-4.3 16.T ll2.8 CL A-T-6 (22)
14 Limon, Colo., No.1, I-TO Pierre U-1/5/4.2-6.3 26.4 98.2 CH A-T-6 (35)
15 Limon, Colo., No.2, I-TO Laramie U-1/1/3.4-5.0 38.2 T8.8 CH A-T-6 (44)
16 Denver, Colo. Denver U-3/4/5.T-T.8 16.0 103.9 CL A-6 (10)
lT Newcastle, Wyo., No. 1, SH 16 Mowry U-2/1/3.0-5.2 .26.4 9[.5 CH A-T-6 (31)
18 Newcastle, Wyo., No. 2, US 85 Pierre U-1/2/3.0-3.9 15.T 9[.5 CH A-T-6 (25)
19 Billings, Mont. Bearpaw U-2/1/3.1-4.6 11.6 110.8 CH A-T-6 (50)
20 Reliance, S. Dak. Pierre U-2/1/1. T-3.9 34.5 83.4 CH A-T-5 (42)
sample description, description of climate, and other pertinent informa-
tion, are given in Reference 28.
33. The laboratory testing program encompassed the determination
of a majority of the physical, mechanical, physicochemical, and mineral-
ogic properties currently used by engineers to characterize soils in gen-
eral and expansive soils in particular. Reference 28 gives a detailed
description of the entire laboratory testing program as well as a summary
of the data in tabulated form (Tables 6-13, Reference 28). Tables 7, 8,
and 11 from Reference 28 are repeated here as Tables 2, 3, and 4, respec-
tively, to provide the data required to use the published techniques.
The data in Table 4 are somewhat different than Table 11 because the
data have been replotted using an expanded scale and a slightly different
and more consistent criterion for interpreting the T-W plots to obtain
the soil moisture suction parameters.
34. During the overburden swell testing program, overburden pres-
sure was supposed to reflect in situ conditions of wet density and depth
to test specimen, thus yielding a measured deformation corresponding to
the potential swell as defined earlier. However, because of an error in
testing assignment, a constant overburden pressure of 0.28 tsf was used
for all specimens. For test specimens whose actual and tested overbur-
den pressures did not agree, a second specimen was tested. The results
of the reruns and those tests not requiring additional testing are given
in Table 5. With the reruns completed, all of the overburden swell test
data conform with the definition of potential swell. Briefly, the over-
burden swell test procedure begins by applying a very small seating load
followed by application of the overburden pressure and inundation of the
specimen. The specimen is allowed to swell to equilibrium (end of swell
condition), then the specimen is consolidated in increments to the void
ratio corresponding to overburden conditions and rebounded in decrements
to the seating.load (end of test condition).
Analysis of Data
18
Table 2
Summarl of S:Eecific Gravit;l: 2 Grain-Size Distribution 2
and Atterberg Limits and Indices
Liq_uid Plastic
Site Specific Grain Size: % Limit Limit Plasticity Liq_uidity
No. Gravit;l: No. 200 ~ % % Index Activitl Index
1 2.69 98 82 104 36 68 0.83 0.06
2 2.68 84 38 61 20 41 1.08 0.14
3 2.72 96 57 96 38 58 1.02 0.03
4 2.72 82 37 56 17 39 1.05 0.09
19
Table 3
. 1
Summary of Measured Shrinkage ProEerties
Bar
Shrinkage Linear
Site Limit Shrinkage Shrinkage* Shrinkage
No. % Ratio Index %
1 9.8 1.83 94.2 23.2
"
2 14.5 1.86 46.5 19.2
3 20.5 1.70 75.5 19.5
4 12.4 1.93 43.6 17.7
5 16.2 1.73 41.8 18.8
6 14.8 1.89 19.2 20.0
7 20.3 1.61 27.7 12.1
8 18.9 1.76 28.1 14.4
9 14.5 1.85 19.5 13.4
10 16.8 1. 77 37 .2. 12.8
11 19.8 1.72 26.2 26.2
12 17.6 1. 75 57.4 20.2
13 23.6 1.58 25.4 13.6
14 20.2 1.68 35.8 14.4
15 18.3 1.72 44.7 15.6
16 15.5 1.83 22.5 11.0
17 19.1 1.75 35.9 12.0
18 19.0 1.72 31.0 12.6
19 12.8 1.92 56.2 19.6
20 27.3 1.49 52.7 16.0
20
Table 4
Summary of Soil Moisture Suction Data
w -r*
Moisture Soil T-W Equation
Site Content Suction Coefficients**
No. % tsf A B
21
Table 5
Summary of Data from Overburden Swell Tests
Note: e =void ratio, w =moisture content (%), S =degree of saturation (%), y =dry density (pcf).
* Corresponds to overburden conditions on e-log p plot.
** 6e = er - e 0 •
t %swell= 6e/(l + e 0 )
data collected during the laboratory testing program were used in con-
junction with the published identification/classification techniques to
compare the indicatec qualitative magnitude of swell with the measured
potential swell as previously defined and measured in the overburden
swell test. Second, the measured potential swell and laboratory data
were analyzed using the statistical analysis program to obtain the best
correlations between the two groups of data.
Establishment of
potential swell categories
36. It was stated earlier that the evaluation of the published
techniques is basically an evaluation of the techniques as they are ap-
plied rather than how they were developed since there was a large varia-
tion in the definition of potential swell used. Evaluation of the ac-
curacy of the published techniques requires a basis or standard for com-
parison, or, more precisely, a categorization of the potential swell as
previously defined and measured in the laboratory testing program.
Table 5 summarizes the measured swell from the odometer swell tests on
undisturbed samples under loads equivalent to the in situ overburden
pressure. The swell varies from 0.01 to 12.7 percent, with the majority
of the measured values between 0.14 and 3.65 percent. During previous
discussions with representatives of the State Highway Agencies it was
indicated that a two- or three-category classification of potential
swell would be preferable to the four- to six-category classifications
published in the literature. With this in mind, the measured potential
swell was divided into three categories. The three categories and their
corresponding potential swell limits are:
23
-
'
24
Table 6 summarizes the classification of potential swell of the 20 sam-
pling sites as indicated by each of the published techniques. The N&C
and K&D techniques are more directly related to prediction of volume
change rather than identification/classification; however, they are in-
cluded here since they are combination techniques based on Atterberg
limits and physical properties. Classifications of potential swell were
not given for either of these techniques, therefore, actual values were
calculated for comparison with measured values. The K&D technique uses
swell pressure as a measure of potential swell rather than percent swell.
For each value calculated using the K&D technique, the number directly
below it in parentheses in Table 6 is the measured swell pressure from
the laboratory testing program.
Discussion of results
38. Careful examination of Table 6 shows that no published
identification/classification technique provides a generally applicable
methodology for identifying and qualitatively classifying potential
swell. However, some of the techniques provide reasonably accurate and
consistent indication of problem conditions. For example, the LDOT pro-
cedure agrees with the U. S. Army Engineer Waterways Experiment Station
(WES) categorization for classification of potential swell of 10 of the
20 sampling sites and is conservative in its classification for nine of
the remaining 10 samples. Using this rationale, the relative accuracy of
the published techniques may be compared as shown in Table 7. Ideally,
the best technique should result in a majority of responses in the
"agreement" and "conservative" columns and as few as possible in the
"nonconservative" column when compared with the WES categorization. The
higher the percentage of "agreements" versus the percentage of "conserva-
tive" responses, the better the technique.
39. Of the 17 techniques evaluated, four balance accuracy and
conservatism best, namely: LDOT, Sowers, Altmeyer, and V&G. All four
of these techniques involve Atterberg limits in some form, either indi-
vidually or in combination. Within this group of techniques, the rela-
tive accuracy and conservatism is best for the LDOT procedure and de-
creases in the order of V&G, Sowers, and Altmeyer. In other words, the
25
Table 6
Swmnary of Indicated Potential SWell Using Published Identification Techniques
Potential
Site Swell Alt- Soro- N&C K&D
~ USBR ~ Chen* chan __%__
Ntnnber % ~ KHC ~ Savers ~ ____M[_ S&M meyer V&G** V&B** ~
1 12.7 Sev High V. High High E. High V. High V. High High V. High Crit V. High V. High Med Lov High 25.94 0.49
(3.82)
2 1.32 High High V. High High High V. High High High High Mar High High Med Lov Non 13.51 0.28
(0.96)
3 0.96 Sev High V. High High E. High V. High V. High High High Mar V. High V. High Med Lov Non 16.61 0.25
(0.43)
4 0.82 High High V. High Mod High V. High High High High Mar High High Med Lov Non 12.67 0.22
(0.35)
5 0.64 High Mod High Mod High High Higt High High Mar High High Med Lov Non 11.37 0.28
(0.90)
6 0.02 Lov Lov Med Lov Lov Lov Lov Med Med Mar Lov Med Med Lov Non 8.98 0.31
(0.66)
7 0.43 Lov Mod Med Mod Med Med Med High High Mar Med High High Med Non 13.99 0. 52
(0.65)
8 0.01 Lov Mod Med Lov Med Med Med High High Mar Med High High High Non 16.85 0.71
(0.06)
9 1.31 Lov Lov Med Lov Lov Low Lov Med Med Mar Lov Med High Lov Non 9.50 0.33
(1.17)
10 3.65 High Mod High Mod High Med High Med Lov Mar Med High High Lov Non 8.30 0.46
(0.93)
1\) 11 0.33 Lov Mod Med Mod Med Med Med High Med Mar Med High V. High Med Non 22.8o 0.89
0\ (0.17)
12 1.02 Sev High V. High High V. High V. High High High High Mar V. High V. High High Lov Non 22.33 0.61
(0.95)
13 0.35 Lov Mod Med Lov Med Med Med Med Med Mar Med High High Lov Non 10.00 0.45
(1.16)
14 0.86 High Mod High Mod High High High High High Mar High High Med Lov Non 11.93 0.24
(0.8o)
15 0.14 High High V. High Mod High V. High High High High Mar High V. High Low Low Non 14.16 0.10
(0.38)
16 3.18 Lov Mod Med Lov Med Lov Med Med Med Mar Med Med Med Lov Non 10.03 0.22
(1.00)
17 0.81 High Mod High Mod High Med High High Med Mar High High Med Lov Non 9.27 0.22
(1.07)
18 2.70 High Mod Med Lov High Med High Med Med Mar Med High High Lov Non 12.15 0.44
(2.37)
19 0.36 High High V. High High High V. High High High High Mar High· V. High High Med Non 20.67 1.05
(0.59)
20 1.87 Sev High V. High High V. High V. High High High High Mar High V. High Med Low Sli 17.84 0.39
(2.54)
Note: Sev = Severe~ Mod = Moderate, V. High = Very High, Med = Medium, E. High = Extra High, Crit = Critical, Mar :;;: Marginal, Non ;: ;- Nonswelling, Sli = Slight.
* No Standard Per..etration data available.
** Categories of potential swell vere not given in reference. Values used were assigned by author.
Table 7
Comparison of WES Categories and Classification by Published
Identification Techniques for 20 Sampling Sites
Number of Sites
for which Classi-
fication in Number of Sites Number of Sites
Published Agreement With on on
Technique WES Categories Conservative Side Nonconservative Side
LDOT 10 9 1
KHC 6 10 4
Raman 4 14 2
Sowers 9 7 4
D&R 5 13 2
A&T 4 12 4
R&S 5 13 2
S&M 3 14 3
USBR 4 13 3
Altmeyer 9 7 4
SWL 3 13 4
Chen 5 14 1
V&G 9 8 3
V&S 3 4 13
Sorochan 1 0 19
N&C 0 20 0
K&D 2 3 15
27
most consistent indicator of potential swell is first, the LL and PI;
second, the LL and natural w.l combined; third, SL and PI; and finally,
the SL and linear shrinkage. As will be discussed later, this is con-
sistent with the analysis of the WES categorization and laboratory data.
28
Table 8
Summary of Independent Variables Used in Statistical Comparisons
Variable Description
29
Table 8 (Concluded)
Variable Description
w./PL Ratio of w. (OST) to plastic limit
1 1
w./LL Ratio of w. (OST) to liquid limit
1 1
A - (B • PL) Sum of indicated variables
A/B Ratio of indicated variables
CEC activity Cation exchange capacity/percent minus No. 200 sieve
30
the Thornthwaite Moisture Index (TMI). 30 • 31 This climatic group had
three subgroups:
Humid Subgroup (TMI > 20)--Sites l-4
Moist Subhumid Subgroup (0 < TMI < 20)--Sites 7,8,12,13
Dry Subhumid Subgroup (-20 < TMI < O)~-S1tes 5,6,9-11,14-20
The second climatic group was based on the 1975 TMI since the samples
were obtained and tested during 1975. This group also had three
subgroups:
Humid Subgroup (TMI > 20)--Sites l-4,7,8
Moist Subhumid Subgroup (0 < TMI < 20)--Sites 5,6,12,13,19
Dry Subhumid Subgroup (-20 < TMI < 0)--Sites 9-11,14-18,20
In both climatic groups, site ll was actually classified in a semiarid
subgroup (TMI < -20); however, since analysis of a subgroup with a
single observation was not practical or significant, site ll was in-
cluded in the Dry Subhumid Subgroup.
42. Table 9 summarizes the correlation coefficients greater (or
less) than +0.7 (-0.7) for the three groups based on physiography and
climate. For the combined group the correlation coefficients shown are
greater (or less) than +0.5 (-0.5) since no r values meeting the pre-
vious criteria were obtained.
43. As expected the individual subgroups resulted in higher cor-
relation coefficients; however, it is difficult to determine whether the
variables actually correlate or the small number of observations is an
overriding factor. For variables which result in significant r values
in four or less of the subgroups, the small number of observations and
the relative magnitudes of the data are likely the reasons for the
higher r values. For variables resulting in significant r values in
five or more of the subgroups, it is likely that the variable actually
correlates with potential swell.
44. Based on the previous discussion, six variables (Table 8)
yield significant correlation coefficients when considering the nine sub-
groups, namely: percent - 2 ~m, LL, PI, SI, BLS, and Tnat• One, per-
cent - 2 ~m, shows a contradiction (i.e. both + and -r values) which
raises some doubt about its actual correlation even though experience
31
i
I
J
I,
Table 9 I,
8\@!N or Correlation Coetticiente Greater (or Leee) Than +0.1(-o.l)
Prte Statistical Pa!pariaonrs
,~it!th!l.!/.1l/.1.!./.~111!1/.~h~!tj;/l.ll.~~f
Combined Group•
0.52 0.50 0.60 0.51 0.51
II
I
20 Sampling Sites
i .
0
~
Colorado Plateau Subgroup
(n • 3)
o.96 ..0.79 -0.92 o.80 0.91 -0.11 0.78 0.75 0.80 0.78 0.99 0.99 0.85 0.83
w
~
Great PlainS Subgroup
0.83
(n • 12)
1'0
H~
...e
~
Humid Subgroup (n • 4) 0.9() 0.70 0. 79 -0.97 o.so 0.95 0-95 0-95 o. 76 0.86 -0.78
aH
~"'
Dry Subbumid Subgroup
H
u
(n • 9)
-0-75
. - -- - - -
• Correlation coe!flc:ients given are greater (or lea a) th&n +0. 5 ( -0.5).
has shown that percent clay has an influence on volume change behavior.
Of the remaining five variables, three resulted in "relatively signifi-
cant" correlation coefficients when the data were analyzed as a combined
group, namely: LL, PI, and SI. ("Relatively significant" does not re-
flect significance from a statistical point of view, rather it indicates
the highest correlation coefficients obtained in the analysis.) Al-
though BLS and Tnat were not "relatively significant" in the combined
group analysis, two factors which are related to soil suction were; that
is, C - and A/B. C - is the suction index, or more precisely, the
Ta Ta
rate of change of soil suction with void ratio based on the vertical com-
pressibility factor, a. A/B is the ratio of the intercept (A) to the
slope (B) of the soil suction versus water content curve.
45. As. indicated earlier, the evaluation of the published tech-
niques and analysis of potential swell and laboratory data agree on the
properties which are the best indicators of potential swell. They in-
clude LL, PI, SI (LL- SL), and BLS. In addition, the analysis of labo-
ratory dat~ shows that soil suction is a very good indicator of potential
swell.
Establishment of criteria
for identifying potential swell
46. Although four of the published identification/classification
techniques were shown to be reasonably accurate in achieving their pur-
pose of identifying problem soils, it was decided to modify the cate-
gories and the relative ranges of the properties included in the cate-
gories to conform better with the definition of potential swell presented
in this report. Two alternatives were available for establishing the
criteria; one was to develop a general criterion based on all the data
collected and analyzed as a combined group and the other was to develop
individual criteria for either the physiographic or climatic groups.
The former alternative (i.e. one general criterion) was selected because
the latter alternative posed several problems as follows:
a. The alternative would result in several different cate-
gories which would make implementation more difficult
and exchange of information virtually impossible.
33
b. The number of observations within the subgroups was
small as compared with the land area represented by the
subgroup, thus leaving doubt about the accuracy and ap-
plicability of the criteria, as well as the accepta-
bility of the statistical analysis.
c. The TMI is a transient value which varies from year to
year and, although it will most likely be useful for
prediction of final moisture content, it does not show
promise for climatic classification for identification
purposes.
d. Physiographic delineation 29 was shown to be a useful
tool for discussion of occurrence and distribution of
expansive soils, but no direct genetic implication
exists.
e. Indications are that identification of expansive soils
is not as critical a problem as testing the soil and
estimating the anticipated volume change.
Potential
Potential Swell
"[ , tsf
LL, % PI 2 % nat Swell 2 % Classification
The SI was included in the trial criteria with the same ranges as the PI
and did not reduce or increase the accuracy of the classification, there-
fore, it was not included in the final criteria. The relative merits
of using the three properties combined (LL, PI, 'T t) are evident in the
na
following tabulation:
34
Not
Pro:12erties Accurate Conservative Conservative
LL 11 5 4
LL + PI 8 8 4
LL + PI + T
nat
12 6 2
LL + PI + SI 8 8 4
LL + PI + SI + T
nat
12 6 2
35
SUMMARY
36
The ranges of potential swell are low as compared with many of the pub-
lished techniques; however, they are consistent with identification/
classification techniques which used comparable magnitudes of applied
load in the potential swell testing programs, i.e., Altmeyer.
21
37
-
RECOMMENDED USAGE
The classification system above may be used without the soil suction
criteria; however, it should be noted that the accuracy and conservatism
of the system are reduced when the LL and PI are used without the soil
suction criteria. For soils which exhibit a low classification of
potential swell, the pavement design may be completed using basic
38
CONDUCT ROUTINE
CLASSIFICATION TESTS
AND MEASURE SOIL
SUCTION. DEFINE
SWELL POTENTIAL.
YES
39
strength parameters with the confidence that expansion problems will be
minimal and normal construction procedures will further minimize the
limited problems. For soils which exhibit a high classification of po-
tential swell, all locations should be tested (undisturbed samples) and
an estimate of the•anticipated volume change made. Specific testing and
prediction techniques are not recommended at this time since they are
the subject of a separate research task and will be discussed in detail
in a subsequent report and the next step of the decision process. For
soils which exhibit a marginal classification of potential swell, addi-
tional testing at that location should be judged on the conditions at
the specific site. The marginal category is indicative of a moderate to
high capacity for volume change, but characteristic properties preclude
the development of swell. For example, a marginal classified soil which
has a relatively high w. (i.e., greater than the PL) and a low natural
l
density will most likely not cause serious problems and therefore not
require testing. However, if conditions are such that the properties
(water content or density or both) are likely to change during and fol-
lowing construction, then additional testing and an estimate of the vol-
ume change corresponding to the new conditions are necessary.
40
REFERENCES
41
13. Snethen, D. R., "Visit to Kansas Highway Commission," Memorandum
for Record,_ 26 September 1974, Soil Mechanics Division, Soils and
Pavements Laboratory, U. S. Army Engineer Waterways Experiment Sta-
tion, CE, Vicksburg, Miss.
14. Raman, V., "Identification of Expansive Soils from the Plasticity
Index and the Shrinkage Index Data," The Indian Engineer, Calcutta,
Vol 11, No. 1, Jan 1967, pp 17-22.
15. Sowers, G. F., "High Volume Change Clays of the Southeastern Coastal
Plains," Vol III-7, Proceedings of Third Pan American Conference on
Soil Mechanics and Foundation Engineering, Caracas, Venezuela, 1967,
pp 99-120.
16. Sowers, G. B. and Sowers, G. F., Introductory Soil Mechanics and
Foundations, 3d ed., Macmillan Company, 1970.
17. Dakshanamurthy, V. and Raman, V., "A Simple Method of Identifying
an Expansive Soil," Soils and Foundations, Japanese Society of Soil
Mechanics and Foundation Engineering, Vol 13, No. 1, Mar 1973,
pp 97-104.
18. Anderson, K. 0. and Thomson, S., "Modification of Expansive Soils
of Western Canada with Lime," Proceedings, Second International
Research and Engineering Conference on Expansive Clay Soils, Texas
A&M University, College Station, Tex., Aug 1969, pp 175-182.
19. Ranganatham, B. V. and Satyanarayana, B., "A Rational Method of
Predicting Swelling Potential for Compacted Expansive Clays,"
Proceedings, Sixth International Conference on Soil Mechanics and
Foundation Engineering, Vol 1, 1965, pp 92-96.
20. Saito, rT. and Miki, G., "Swelling and Residual Strength Character-
istics of Soils Based on a Newly Proposed Plastic Ratio Chart,"
Soils and Foundation, Journal of Japanese Society of Soil Mechanics
and Foundation Engineering, Vol 15, No. 1, Mar 1975, pp 61-68.
21. Altmeyer, W. T., "Discussion of Engineering Properties of Expansive
Clays," Proceedings, American Society of Civil Engineers, Vol 81,
Separate No. 658, Mar 1955, pp 17-19.
22. Chen, F. H., nThe Use of Piers to Prevent the Uplifting of Lightly
Loaded Structures Founded on Expansive Soils," Proceedings, First
International Research and Engineering Conference on Expansive Clay
Soils, Texas A&M University, College Station, Tex. 1965, pp 152-171.
23. Vijayvergiya, V. N. and Ghazzaly, 0. I., "Prediction of Swelling
Potential for Natural Clays," Proceedings, Third International
Research and Engineering Conference on Expansive Clay Soils~ Haifi,
Israel, Aug 1973, pp 227-234.
24. Vijayvergiya, V. N. and Sullivan, R. A., "Simple Technique for Iden-
fying Heave Potential," Proceedings, Workshop on Expansive Clays
and Shales in Highway Design and Construction, D. R. Lamb and S. J.
Hanna, ed., prepared for Federal Highway Administration, Washington,
D. C., Vol 1, May 1973, pp 275-294.
42
25. Sorochan, E. A., "Certain Regularities of the Swelling of Soils,"
Journal, Soil Mechanics and Foundation Engineering, Indian National
Society, Vol 9, No. 3, Jul 1970, pp 293-304.
26. Naya.k, N. V. and C.hristensen, R. W., "Swelling Characteristics of
Compacted Expansive Soils," Clays and Clay Minerals, Vol 19, No. 4,
1974, pp 251-261.
27. Komornik, A. and David, D., "Prediction of Swelling Pressure of
Clay," Journal, Soil Mechanics and Foundations Division, American
Society of Civil Engineers, Vol 95, SMl, Jan 1969, pp 209-225.
28. Snethen, D. R., Johnson, L. D., and Patrick, D. M., "An Investiga-
tion of the Natural Microscale Mechanisms that Cause Volume Change
in Expansive Clays" (in preparation), Federal Highway Administra-
tion, Washington, D. C., Jun 1977.
29. Patrick, D. M. and Snethen, D. R., "An Occurrence and Distribution
Survey of Expansive Materials in the United States by Physiographic
Areas," Report No. FHWA-RD-76-82, Federal Highway Administration,
Washington, D. C., Jan 1976.
30. Thornthwaite, C. W., "An Approach Towards a Rational Classification
of Climate," Geographical Review, Vol 38, pp 55-94, 1948.
31. Russum, K. and Coleman, J. D., "Th.e Effect of Climatic Factors on.
Subgrade Moisture Conditions," Geotechnique, Mar 1961, pp 22-28.
43
1 1 1 1 1 1 1 1 1 1 1 1 ~1 ~ l ilf~l l l l l l l l l l l l l
3 9526 00005132 5
Snethen, Donald R/An evaluation of exped
RESEAACHfTA710.5 .S64 1977