Professional Documents
Culture Documents
Abstract: Premature failures in chemically stabilized expansive soils cost millions of dollars in maintenance and repair. One reason for these
failures is the inability of existing stabilization design guidelines to take into account the complex interactions between the clay minerals and the
stabilizers. It is vital to understand these complex interactions, as they are responsible for the improved strength and reduction of swelling/shrinking
in these soils, which in turn affects the overall health of the infrastructure. This research study examined the longevity of chemically stabilized
expansive soils subjected to wetting/drying conditions, with a major focus on clay mineralogy. Eight different natural soils with varying clay
mineralogy were subjected to wetting/drying durability studies after they were stabilized with chemical additives including quicklime and cement.
Performance indicators such as volumetric strain and unconfined compressive strength trends were monitored at regular intervals during the
wetting/drying process. It was observed that clayey soils dominant in the mineral montmorillonite were susceptible to premature failures. It
was also noted that soils dominant in other clay minerals exhibited early failures at lower additive contents. For the first time, an attempt
was made to address the field implications of the laboratory studies by developing a correlation that predicts service life in the field based
on clay mineralogy and stabilizer dosage. DOI: 10.1061/(ASCE)GT.1943-5606.0001796. © 2017 American Society of Civil Engineers.
Author keywords: Soil stabilization; Durability; Seasonal changes; Clay mineralogy; Expansive soils; Volumetric strain.
Note: CEC = cation exchange capacity; PI = plasticity index; SSA = room, the plastic wrap was removed, and durability tests were
specific surface area; TP = total potassium; %I = illite percent in performed.
clay fraction; %K = kaolinite percentage in clay fraction; %M =
montmorillonite percentage in clay fraction.
Durability Testing
The procedure outlined by the ASTM D559 method was closely
stabilized with lime and the TEX-120E (TxDOT 2017a) method followed; it simulates both wet and dry cycle conditions close to
for the soils stabilized with cement. For the soils stabilized with the field conditions in a reasonably short period. According to the
lime, an initial estimate of the amount of lime was made via a pH ASTM D559 method, the soil specimens should be prepared and
test or the Eades and Grim test (Eades and Grim 1966). Once an cured and then submerged in water for 5 h for the wetting cycle, and
initial estimate was made, the optimum moisture contents (OMC) then oven-dried at 71°C (160°F) for 42 h for the drying cycle. A
and maximum dry unit weights (MDUW) were obtained for vary- schematic of the test setup used in this research is presented in
ing soil-lime combinations with an additive content higher than Fig. 1. After removal from the oven, the specimen was subjected
that determined by the pH test. The OMC and MDUW for each to volume-change and moisture-content measurements. The pro-
soil-additive combination were determined by conducting the cedure was repeated for a total of 21 wet-dry cycles or until the
standard Proctor’s compaction test, as per ASTM D698 (ASTM sample failed. The soil specimens were allowed to swell and shrink
2012). unconfined compressive strength (UCS) tests, as per ASTM in both lateral and vertical directions during the wetting and the
D2166 (ASTM 2016), were performed on samples prepared at the drying cycles, as prior studies by Punthutaecha et al. (2006) noted
OMC and MDUW to determine the appropriateness of the stabi- that the volumetric swell/shrink strains measured in the laboratory
lizer and its content on the basis of the strength requirements per with no lateral or vertical restraints were in closer agreement with
TEX-121E, which recommends a UCS value of more than 345 kPa the field-measured values than those obtained by restraining the
(50 psi). A similar procedure was followed for the soils stabilized lateral movement. The vertical movement was measured with the
with cement, except that no pH testing was performed and the pro- help of a dial gauge, whereas the radial movements were measured
cedure outlined in TEX-120E was followed. Table 3 presents the using a pi tape.
type and the amount of stabilizer selected for each of the eight soils A sample is deemed to have failed when (1) the UCS of the sam-
used in this research, along with their dominant clay minerals, ple tested is close to zero; (2) the sample crumbles under its own
MDUW, and OMC. weight, and hence no further testing can be performed; and (3) the
sample becomes so sensitive that it can be crumbled/crushed during
sample-handling activities, including moving the sample from the
Sample Preparation oven to the water bath and/or making volumetric measurements.
The specimens of each soil/additive combination to be tested for the In this study, a mix design or soil/additive combination was
durability studies were prepared at the optimum moisture content considered to be effective if the soil samples prepared using a par-
and maximum dry unit weight condition, obtained during the ticular mix design lasted all 21 cycles and retained at least 50% of
mix design process. The samples were prepared using the static their initial strength at the beginning of wetting/drying. Unconfined
compressive strengths were performed on control and stabilized
soil samples as per ASTM D2166. Because the UCS tests were
destructive tests, a total of five soil samples were prepared so that
Table 3. Type and Amount of Stabilizer Selected for Each Soil along with each sample could be tested at 0, 3, 7, 14, and 21 W/D cycles for
Their Dominant Clay Mineral each soil-additive combination. Duplicate specimens were used
Amount of for obtaining each data point; thus, a total of 10 samples were cast
additive, for each of the soil/additive combinations. Each data point in the
Serial Dominating Type of (% by MDUW OMC plots is an average of two test results. The UCS tests were per-
number Soil clay mineral additive weight) (kN=m3 ) (%) formed after the corresponding wetting cycle, and because the sam-
1 Austin Montmorillonite Lime 6 17.1 20.4 ples were saturated after this cycle, the control sample at 0 cycles
2 Bryan Kaolinite Lime 8 15.3 22.7 was also tested after saturating the sample.
3 El Paso Illite Lime 8 18.2 15.1
4 Fort Montmorillonite Lime 6 14.6 23.1 Additional Durability Studies
Worth The soils that did not perform well after stabilization using the cur-
5 Keller Kaolinite Lime 6 19.1 13.4 rent design procedures were further analyzed with different additive
6 Paris Montmorillonite Lime 8 13.7 24.3 types in varying dosages to determine which stabilizer dosage
7 Pharr-A Montmorillonite Lime 4 13.4 31.0 combination would be effective for these soils. On the basis of
8 Pharr-B Kaolinite Lime 3 15.1 24.7
these observations, an alternative procedure was developed for
Soil Sample
Porous Stone
Downloaded from ascelibrary.org by MVGR College of Engineering on 02/24/19. Copyright ASCE. For personal use only; all rights reserved.
WETTING DRYING
Pi-tape
determining the stabilizer type and dosage according to the miner- Summary of W/D Test Results
alogy and other index properties of a particular soil.
For example, the Austin soil, which has a PI of 34, was not sta- Volumetric strain changes were measured at the beginning and
bilized effectively when 6% lime was used as a stabilizer, due to the the end of each wetting and drying cycle. Percentage changes in
high montmorillonite content (40%) in its clay fraction. This soil volume after the corresponding cycle of W/D were calculated for
was retested with an increased dosage of lime (8%) and two dos- each sample on the basis of its initial dimensions before the W/D
ages of cement (3 and 6%). After different cycles of durability, the cycles started. Strength tests were performed after 3, 7, 14, and 21
soil specimens were tested for the retained strength. Similar tests cycles of W/D. The percentage of retained strength after a corre-
were conducted on the Pharr-A, Pharr-B, and Fort Worth soil spec- sponding W/D cycle was calculated by dividing the current UCS
imens; Table 4 summarizes the various mix designs performed on strength by the 0 cycle UCS strength before the first W/D cycle and
these soils. Following the procedures described previously, these after seven days of curing.
soil specimens were subjected to alternate wetting and drying For conciseness, the volumetric strain results are presented
studies. in three groups, based on their mineral dominance in their clay
6% lime 31.5 12.2 are presented in Figs. 2–4, respectively. These three figures show
8% lime 32.5 11.9 that the treated soils had lower volumetric strains than the untreated
3% cement 32.0 12.1
soils, which was expected. Soils that survived all 21 cycles with min-
6% cement 32.5 11.9
imal volumetric changes (<10%) were considered to have performed
Fort Worth Control 24.0 14.4 well under the wetting/drying durability conditions.
6% lime 25.5 14.1 Figs. 2(a and b) show that the Austin and Fort Worth soils
3% cement 24.5 14.1 survived for a total of 12 and 11 cycles of durability, respectively,
6% cement 25.0 14.0 when treated with 6% lime. Similarly, the Paris soil samples treated
Note: MDUW = maximum dry unit weight; OMC = optimum moisture with 8% lime survived for only seven durability cycles, and this
content. can be observed in Fig. 2(c). Fig. 2(d) shows that the Pharr-A soil
Fig. 2. Volumetric strain change with W/D cycles for Group I soils: (a) Austin soil; (b) Fort Worth soil; (c) Paris soil; (d) Pharr-A soil
Fig. 3. Volumetric strain change with W/D cycles for Group II soils: (a) Bryan soil; (b) Keller soil; (c) Pharr-B soil
samples survived for four W/D cycles when they were treated with
4% lime. It is apparent that none of the Group I soils mentioned
survived the entire W/D testing. Premature failure in one soil
(Pharr-A) was attributed to the low initial dosage used. However,
all Group I soils experienced early failures as these soils are dom-
inant in the montmorillonite mineral.
The Keller and Bryan soils, which fell under Group II, survived
all 21 W/D cycles when they were treated with 6 and 8% lime,
respectively. However, the Pharr-B soil samples, which belonged
to the same group, survived only eight W/D cycles when they were
treated with 3% lime; this is clearly shown in Fig. 3(c). Finally, the
El Paso soil, which belonged to Group III, survived all 21 W/D
cycles when it was treated with 8% lime. Fig. 2 also shows that the
treated soils did not survive all 21 cycles, whereas Figs. 3 and 4
shows that all the soils survived the 21 W/D cycles, except for the
Pharr-B soil shown in Fig. 3(c).
Fig. 5. Unconfined compression strength variation with W/D cycles for Group I soils: (a) Austin soil; (b) Fort Worth soil; (c) Paris soil;
(d) Pharr-A soil
dosage did not retain any strength after 12 and 11 W/D cycles, re- Effect of Soil Type
spectively. The Paris soil treated with 8% lime lost all of its strength
Table 5 shows that the Austin, Fort Worth, Paris, and Pharr clays,
after eight W/D cycles, as shown in see Fig. 5(c), whereas the
which have high amounts of the montmorillonite mineral in their
Pharr-A soil specimens treated with 4% lime lost all their strength
clay fractions, did not survive all 21 W/D cycles. The clays from El
after four durability cycles, as shown in Fig. 5(d). Figs. 6(a and b)
Paso, Bryan, and Keller, which have other clay minerals (illite or
show that both the Bryan and Keller soil specimens treated with 8
kaolinite) as the dominant mineral in their clay fraction, survived
and 6% lime, respectively, lasted all 21 W/D cycles and retained over
all 21 cycles. This indicates that soils containing montmorillonite
85% of their initial strength. The Pharr-B soil specimen treated with
as a dominant mineral are more susceptible to premature strength
3% lime failed to retain any strength after eight durability cycles, as
failures after chemical stabilization when they are exposed to
shown in Fig. 6(c). Finally, Fig. 7 shows that the El Paso soil speci-
volume changes caused by moisture movements. It should be noted
men treated with 8% lime retained 85% of its initial strength after
that although the Pharr-B soil is dominant in the mineral kaolinite,
21 W/D cycles. Table 5 summarizes the retained strength measure-
ments, along with the maximum volumetric strain change at the end it lasted for only eight cycles, primarily because of the low addi-
of the number of W/D cycles survived, for all the eight soils. tive dosage (3% lime). To further represent the effect of the min-
eral montmorillonite on the durability of stabilization, the retained
strengths and volumetric strain changes after seven W/D cycles for
Analysis and Discussion all the soils treated with 8% lime were plotted, as shown in Fig. 8.
This figure shows that as the percentage of montmorillonite in-
As explained previously, on the basis of the results, the four soils creased in the clay fraction, the retained strength after seven W/D
that did not perform well under the W/D durability were further cycles decreased whereas the volumetric strain change increased.
tested; those results are summarized in Table 6. All the above re- This is an important finding, as it shows the influence of the clay
sults were analyzed for the effects of clay mineralogy, additive type, mineralogy on the durability of the chemical stabilizer in providing
and additive dosage in the following sections. sustained strength over a long period of time. A similar figure was
Fig. 6. Unconfined compression strength variation with W/D cycles for Group II soils: (a) Bryan soil; (b) Keller soil; (c) Pharr-B soil
8% lime 21a 6 67
3% cement 14 10 0
6% cement 21a 4 80
Pharr-A Montmorillonite 4% lime 4 30 0
6% lime 9 9 0
8% lime 14 7 0
3% cement 9 9 0
6% cement 21a 5 83
Fort Montmorillonite 6% lime 10 15 0
Worth 8% lime 14 8 57
3% cement 14 11 42
6% cement 21a 5 57
a
Sample still intact. Fig. 9. Effect of plasticity index on durability of lime stabilization
ture variations and volumetric movements due to these moisture design. It is highly important to include the clay mineralogy infor-
fluctuations. Two soils were selected for this analysis: Pharr-A, mation in the stabilizer design process to achieve a durable and sus-
dominant in the mineral montmorillonite; and Pharr-B, dominant tainable subgrade stabilization.
in the mineral kaolinite. The Pharr-A was treated with 4, 6, and
8% of lime, whereas the Pharr-B was treated with 3, 6, and 8%
of lime. It is shown by the table that as the lime dosage increased, Field Service Life Implications
the stabilization effectiveness of the Pharr-B soil improved consid- On the basis of the results obtained in this study, an effort was
erably, making the samples last longer with the increase in dosage. made to understand the implications of these laboratory studies
In the case of the Pharr-A soil, the stabilization effect lasted longer in field service life assessments. For that purpose, the researchers
when the dosage was increased from 4 to 8%; however, even at 8% attempted to predict the service life of stabilization, defined as field
lime, the samples lasted only 14 W/D cycles. Hence, this exercise service years (FSY), using the ratio between the percentage of mont-
shows that for soils dominant in the kaolinite mineral (in this case, morillonite in the clay fraction of the soil and the amount of lime
the Pharr-B soil), the lower dosage levels can lead to premature used to stabilize the soil, defined as the montmorillonite-lime ratio
failures, while a higher dosage can assure long-term performance. (MLR). This analysis is approximate and is dependent on the clay
The same is not true for the montmorillonite-dominant soils (in mineralogy and the additive dosages. The test database used is not
this case, Pharr-A), where a higher dosage did not result in better comprehensive, and more W/D studies with detailed mineralogy in-
performance. formation will enhance the correlation developed in this research.
One important observation regarding the montmorillonite-rich In the laboratory, the samples were dried for 42 h at 60°C, and
soils is that if these soils are restrained in volume changes due to the samples were submerged under water for 5 h, during which the
higher confinements (with the assumption that treatments are car- samples reached a saturation of 85–95%. No studies have been con-
ried out at medium-range depths of 1–2 m), the durability perfor- ducted so far to assess how many of these wetting and drying sce-
mance will be improved (Pedarla 2009). If the treatments are narios occur in a year; this study used the assumption that in a given
carried out at shallow depths within the top 0.6 m of the subgrade, year there were two such wetting/drying scenarios, although there
it is recommended to replace the montmorillonite-rich soils with a might be several rainfall events followed by dry periods in a given
quality fill material. Other methods of amending the natural soils year. The basis for this conservative assumption is that the labora-
with sand or other inert materials may also provide better W/D per- tory conditions in the study were extreme, whereas in the field,
formance, but those studies are not considered in this research and the soil rarely becomes completely dry unless there are prolonged
therefore no recommendations on those amendments can be made. drought conditions. This kind of approximation is area-specific and
varies among regions. However, the analysis attempted here pro-
Limitations of Existing PI-Based Design Methodologies vides a simple method that correlates the laboratory W/D cycles
with the actual field performance data. Hence, in this research,
So far in this research study, the soils stabilized using the current
on the basis of these approximations, the study’s 21 W/D cycles
stabilization design procedures have not survived the seasonal mois-
correspond to approximately 10 years of the field conditions.
ture fluctuations studied by conducting the W/D durability tests.
Fig. 11 presents a plot of the ratio of the percentage of mont-
The current stabilizer selection criterion, followed by various state
morillonite and the percentage of lime used for stabilization, de-
and federal agencies including TxDOT, FHWA, and the United
fined as the montmorillonite-lime ratio (MLR), versus the field
States Army Corps of Engineers, are based on the plasticity index
survival years (FSY). The FSY can be equated to the number of
(PI) and the gradation characteristics of the soil. These organizations
W/D cycles survived, divided by 2 (assuming two cycles per year).
have reported problems with subgrade failures that occur when a
With the assumption that the montmorillonite percentages in the
soil range from 10 to 80 in the clay fraction and the lime percent-
Table 7. Comparison of Results between the Additive Dosages ages range from 1 to 10, the MLR values can range from 1 to 80.
An MLR value of 80 indicates a worst-case scenario in which the
Amount Percentage
of Number retained percentage of montmorillonite in the soil is 80% and is treated
Soil Mineral Additive additive of cycles strength after with only 1% lime, whereas an MLR value of 1 indicates a best-
type dominance type (%) survived 7 W/D cycles case scenario in which the percentage of montmorillonite in the soil
is only 10% and the soil was treated with 10% of lime. Other com-
Pharr-A Montmorillonite Lime 4 4 0
6 9 29 binations are possible, but for the discussion, not all scenarios are
8 14 24 considered. The typical MLR values for Texas soils range from 2 to
20, based on the soils tested in this research and on standard prac-
Pharr-B Kaolinite Lime 3 8 32 tices. From Fig. 11, it can be observed that for the lower MLR val-
6 14 58 ues, the FSY was higher (at least 10 years) whereas for the higher
8 21 63
MLR values, the FSY values decreased. On the basis of the soils