Technical Note

Tension Cracks in a Compacted Clay Embankment

R. L. Handy, M.ASCE 1; and Michael Lustig, M.ASCE 2

Abstract: In a spirit of adventure, a boring was made and soil moisture contents measured along a tension crack near the crest of an earth
Downloaded from ascelibrary.org by THE UNIVERSITY OF NEWCASTLE on 03/03/17. Copyright ASCE. For personal use only; all rights reserved.

embankment. Moisture contents increased linearly with depth and then sharply decreased at what is interpreted to be the bottom of the crack.
Desiccation does not appear to have been a factor because of construction delays due to rains as the embankment still was under construction.
It is hypothesized that the crack was created by quasi-elastic rebound of the recently compacted soil directed normal to the intersecting ground
surfaces. A second increase in moisture content that appears to be linear with depth was discovered as the boring had been extended all of the
way through to the bottom of the embankment. This change in moisture contents suggests a possible basal crack that can be attributed to
differential settlement, and is assumed to connect to the upper crack as it was encountered in the same boring. Hydraulic connection may have
been impaired by pinching action related to differential settlement but still could allow some water to go through. A comparison of the
two mechanisms suggests that the upper crack probably formed first. A basal crack that remains open should be relevant for earth dams.
DOI: 10.1061/(ASCE)GT.1943-5606.0001667. © 2017 American Society of Civil Engineers.
Author keywords: Cracking; Tension; Embankments; Differential settlement; Water pressure; Landslides; Earth dams.

Introduction and Review of Literature predictor of crack depths. Chowdhury and Zhang (1991) modified
the Terzaghi method by including a pore-pressure parameter and
Tension cracks commonly occur near crests of slopes and embank- using iterative searches for the lowest factor of safety, but did not
ments, where the cracks are ideally situated to intercept surface include field measurements.
runoff water and funnel it down into the soil (Handy 1995). Zhou
et al. (2009) documented the formation of tension cracks in a soil
slope and noted that the cracks continued to grow with increasing Investigation
precipitation. A parametric study by Sasekaran (2011) indicated
that tension cracks affect seepage patterns and lower the factor Construction of a 10-m (33-ft) high lagoon embankment at Baxter,
of safety against slope failure. Duncan et al. (2014) indicated that Iowa was delayed by rains and then halted and repairs initiated after
the influence of tension cracks on slope stability depends on depths tension cracking and widening of the cracks warned of an imminent
and locations of the cracks. shear failure. The embankment consisted of a compacted mixture
Irregularities in crack surfaces can make depths difficult to of Wisconsin-age loess and paleosol developed in pre-Illinioan gla-
measure. Terzaghi (1943) suggests that tension cracks probably ex- cial till, CL and CH, respectively, in the Unified Soil Classification
tend about halfway through the depth of an embankment. Vertical System. The dominant clay mineral in these soils is Ca-smectite
tension cracks often run parallel to vertical excavations, and on the (formerly montmorillonite). Four sets of Atterberg limits from
basis of this failure geometry, Terzaghi (1943) related crack devel- two forensic test borings including Test Boring TB-1 gave a liquid
opment to a Rankine active failure state in the underlying soil. He limit ðLLÞ ¼ 40–48% and a plastic limit ðPIÞ ¼ 23–31, all in the
defined a critical height, Hc , of a vertical slope in cohesive soil as central range for a CH soil, indicating loess as the main constituent
at the locations and depths of the cracks. The LL > 40% also is
4c indicative of highly expansive clay soil under conditions of low
H c ¼ − tan ð45 þ ϕ=2Þ ð1Þ
γ physical restraint (Handy 2002). The foundation soil for the em-
bankment is recently deposited, soft alluvial CH clay. The un-
where γ = unit weight of soil penetrated by the crack; and c and drained shear strength was determined to be 38 kPa (5 psi) prior
ϕ = shear-strength parameters of the underlying soil. H c = total to embankment construction.
height, and the inferred crack depth was estimated to be about All moisture contents and nuclear densities measured at the time
one-half of that amount. of compaction are shown in Fig. 1. Data obtained from samples
Eq. (1) frequently is applied to inclined as well as vertical slopes from Test Boring TB-1, which extended down a crack after it
(Duncan et al. 2014), but several authors have noted that with had opened, are shown by squares in the same figure, and indicate
effective stress soil-strength parameters, it is not an accurate a general increase in moisture content in soil along the crack, with
two of the six data points at or nearly at saturation.
1
Distinguished Professor Emeritus, Dept. of Civil, Construction and When the cracks were first measured and mapped, widths varied
Environmental Engineering, Iowa State Univ., Ames, IA 50010 (corre- from 19 to 70 mm, averaging about 40 mm (0.16 in.) at the ground
sponding author). E-mail: [email protected] surface. Shearing was not indicated along the crack as the two sides
2
Lecturer, Dept. of Civil, Construction, and Environmental Engineer-
were even. The embankment construction had reached the ap-
ing, Iowa State Univ., Ames, IA 50010. E-mail: [email protected]
Note. This manuscript was submitted on February 29, 2016; approved
proximate elevation of the test boring when the inclinometer was
on October 25, 2016; published online on March 3, 2017. Discussion per- installed. Measurements indicated that sliding already had initiated
iod open until August 3, 2017; separate discussions must be submitted for along the base of the embankment where it was in contact with the
individual papers. This technical note is part of the Journal of Geotech- underlying alluvial soil. Over the next 20 days, the crack and nearby
nical and Geoenvironmental Engineering, © ASCE, ISSN 1090-0241. extensions of the crack became wider. No direct measurements of

© ASCE 06017006-1 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., -1--1

 Soil-moisture contents measured in borings that were not close
to tension cracks did not show a systematic change with depth and
were representative of those measured at the time of compaction.
The apparent relationship between moisture content and hydro-
static pressure from water filling a crack may be a function of
the rate of horizontal penetration of water out of the crack and
into the soil, aided by hydration of the expansive clay mineral fol-
lowing the release of lateral pressure.

Testing Eq. (1)

Downloaded from ascelibrary.org by THE UNIVERSITY OF NEWCASTLE on 03/03/17. Copyright ASCE. For personal use only; all rights reserved.

Four triaxial test data sets were obtained from soil samples from
different depths in Boring TB1. Two data sets from above the in-
ferred crack depth gave a soil unit weight of 19.9 kN=m3 and a
moisture content of 23%, giving a wet unit weight of 24.5 kN=m3 .
The average strength parameters were c 0 ¼ 6.3 kPa and ϕ 0 ¼
Fig. 1. Compaction control and moisture-density data from soil sam- 22.6°. Substituting these values into Eq. (1) and dividing by 2 gives
ples obtained by drilling down a vertical tension crack located along the Hc ¼ 1.5 m, which significantly underestimates the depth of 5.9 m
upper slope boundary of a recently compacted clay embankment from Fig. 1. Back-calculation of a computer-generated slope failure
analysis gave a depth of 3.0 m, still significantly less than the in-
dicated depth. The opinion that a tension crack can extend down-
crack depths were attempted as it was not anticipated that there ward about half of the thickness of an embankment probably comes
might be a relationship to the soil-moisture content. Full depths also closest, as the thickness of the embankment at boring TB1 was
8.5 m. That would give a crack depth of 4.25 m, still an underpre-
can be difficult to determine if the width narrows to zero with in-
diction by about 30%. Such estimates become moot if connecting
creasing depth.
tension cracks extend all of the way down through an embankment.

Near-Surface Crack Cracking Attributed to Rebound

The crack moisture-content data of Fig. 1 are graphed versus depth Common locations of tension cracks along the crests of embank-
in Fig. 2 and reveal a surprising relationship. Despite the fact that ments suggest a possible relationship to quasi-elastic rebound fol-
moisture contents were only reported to the closest 1%, there is a lowing compaction and directed normal to intersecting ground
linear increase down a depth of about 5.5 m (19 ft), which suggests surfaces. Rebound is routinely measured during the unloading
a linear relationship between soil-moisture content and hydro- phase of an odometer test, and time-related rebound was reported
static pressure from water filling the crack. This interpretation is by Seed et al. (1962) after unloading of compacted sandy clay.
supported by extrapolation of the moisture-content data to the Carder et al. (1977), cited by Duncan et al. (1991), found a gradual
ground surface, Point A in the graph, which indicates zero gain relaxation of lateral soil pressures after compaction, and an ap-
at zero depth. That being the case, the depth at which the soil- proach to at-rest values after 4 months. This was attributed to
moisture content sharply decreased is interpreted as representing dissipation of positive pore-water pressures that should follow a
the bottom of the crack. small and time-dependent vertical expansion of the soil skeleton.
A significant decrease in seismic velocity has been observed to
occur 2 days after compaction (J. M. Hoover, unpublished research,
1967). X-ray diffraction studies have shown that unrestrained
calcium-saturated smectite (montmorillonite) clay will be fully ex-
panded at 100% relative humidity (Senich et al. 1967).
Micrographs of a flocculated structure show a contorted boxlike
arrangement of clay particles linked by edge-to-face bonding that
may contribute to rebound. Flocculation has been shown to en-
hance quasi-elastic rebound of kaolinite clay by a factor of about
3 (Mitchell and Soga 2005). A flocculated structure is characteristic
of calcium-saturated expansive clay that dominates Iowa loessial,
glacial, and alluvial soils
Soil in the embankment contains expansive clay minerals and
was compacted in accordance with standard engineering specifica-
tions to a minimum density of 95% of the Standard Proctor Density.
A self-propelled sheepsfoot roller was used with nominal 0.25 m
(9 in.) lifts. The dominant clay mineral in Iowa loessial soils is
Caþþ smectite (montmorillonite). The hydraulic conductivity of
compacted samples was determined to be quite low, the average
from seven tests being 8.7 × 109 cm=s. The liquid limits from four
Fig. 2. Soil moisture contents measured along the tension crack and
samples ranging in depth from 2.4 to 7 m varied from 40 to 48%,
evidence that a deeper connecting crack may extend all of the way
averaging 43.8%, and plasticity indices varied from 23 to 31, aver-
down through the embankment
aging 26.5, representative of plastic loess. Particle-size analyses

© ASCE 06017006-2 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., -1--1

 relevance for earth dams, as it should intersect any transverse basal
cracks and deconcentrate the hydraulic pressure. A basal crack also
should provide a useful avenue for grouting.
A similar confluence of surficial and basal tension cracks is
believed to contribute to a columnar structure in lava flows, as
shrinkage from cooling proceeds from the top and from the bottom.

Conclusions

The following conclusions can be drawn from the tests conducted

as part of this study:
Downloaded from ascelibrary.org by THE UNIVERSITY OF NEWCASTLE on 03/03/17. Copyright ASCE. For personal use only; all rights reserved.

• Longitudinal tension cracks at or near the crest of a slope or

earth embankment do not appear to relate to the development
of a Rankine failure state in the underlying soil, as conjec-
tured by Terzaghi;
• Moisture contents measured in soil along a tension crack in
expansive clay soil were found to increase linearly with depth
down to a point where the relationship abruptly ends, which
appears to define the depth of the crack;
Fig. 3. Diagram suggesting how the upper and lower cracks may be • Typical locations of surface tension cracks suggest a relation-
sequential and involve two different causes: (a) hypothesized surface- ship to quasi-elastic rebound of soil following compaction or
tension crack related to elastic rebound of recently compacted clay soil; unloading by erosion. Rebound directions normal to ground sur-
(b) hypothesized basal-tension crack from differential settlement on a faces should diverge along the crest of a slope or embankment;
compressible soil layer • A second increase in moisture content below the inferred bottom
of the upper tension crack suggests the existence of a connecting
basal tension crack;
indicated about 41% 0.005 mm clay, the remainder being silt. Data • A basal crack can be attributed to differential settlement of the
from four triaxial tests gave ϕ 0 ¼ 27.7–33.4°, averaging 30.2°, and embankment on compressible soil, in which case the upper
c 0 ¼ 8–11 kPa. Strength tests were carried to failure and rebound crack probably formed first. If upper and lower cracks join at
was not measured. the ends, they in effect can cut the embankment in two;
Open arrows in Fig. 3(a) indicate hypothetical rebound direc- • Hydraulic connection between the conjoined cracks is impaired,
tions that diverge along the upper slope boundary where the slope probably as a result of downward rotational settlement of the
angle changes. parts of the embankment; and
• A basal crack could be particularly significant in an earth dam,
and provide an avenue for grouting of basal transverse cracks
Basal Tension Crack? related to cross-valley differential settlement.

More problematic is a second increase in soil-moisture content,

D-E in Fig. 1, which also extrapolates to the average moisture Acknowledgments
content measured at the time of compaction, Point C. Extrapolation The field investigation was discontinued after a slope failure and re-
upward indicates that the increase may begin at the same level as construction of the embankment. Data are courtesy of Geotechnical
the bottom of the upper crack, Point B in the figure. If the same Services, Inc., Des Moines, Iowa. The soil borings were conducted
causal factor is involved in the apparent increase in soil-moisture by the co-author, Zachary Thomas, and Dean Heldt. This study uses
contents CDEF in Fig. 1, it would indicate a connection since both an observational method that has its roots in geology and was
cracks were encountered in the same boring. On the other hand championed by Terzaghi. As Mark Twain (1874) observed, “There
the sudden shift in moisture contents indicates an impaired hy- is something fascinating about science. One gets such wholesale
draulic connection, as if soil between the cracks was being squeezed returns of conjecture out of such a trifling investment of fact.”
together but still allowed some water to come through. Fig. 3(b)
shows two conjoined tension cracks inferred from the moisture-
content data. References
Carder, D. R., Pocock, R. G., and Murray, R. T. (1977). “Experimental
Mechanisms retaining wall facility—Lateral stress measurements with sand backfill.”
Rep. No. LT 726, Transportation and Road Research Laboratory,
As the cracks appear to involve two different causal mechanisms, Crowthorne, U.K.
they probably develop independently, with the upper crack forming Chowdhury, R. N., and Zhang, S. (1991). “Tension cracks and slope fail-
first, so as a consequence of stress concentration it became a target ure.” Slope stability engineering: Developments and applications,
for the lower crack. Differential settlement can be expected because Thomas Telford, London, 27–32.
Duncan, J. M., Williams, G. W., Sehn, A. L., and Seed, H. B. (1991).
of uneven loading from the embankment and compressibility of the
“Estimation earth pressures due to compaction.” J. Geotechnical
underlying alluvial soil. Basal cracking has been reported by Lee Eng., 10.1061/(ASCE)GT1943.5606, 1833–1847.
et al. (1988) on the basis of finite-element modeling. A consequent Duncan, J. M., Wright, S. G., and Brandon, T. L. (2014). Soil strength and
inward rotation of sides of the embankment could explain a pinch- slope stability, 2nd Ed., Wiley, New York.
ing action that would inhibit hydraulic connection between the Handy, R. L. (1995). “The day the house fell: Homeowner soil problems from
two cracks. The presence of a longitudinal basal crack may have landslides to expansive clays and wet basements.” ASCE, Reston, VA.

© ASCE 06017006-3 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., -1--1

 Handy, R. L. (2002). “Geology, soil science, and the other expansive clays.” Seed, H. B., Mitchell, J. K., and Chan, C. K. (1962). “Swell and swell
Geotech. News, 20(1), 40–45. pressure characteristics of compacted clays.” Highway Res. Board Bull.,
Lee, F. H., Lu, K. W., and Lee, S. L. (1988). “Tension crack development in 313, 12–39.
soils.” J. Geotech. Eng., 10.1061/(ASCE)0733-9410(1988)114:8(915), Senich, D., Demirel, T., and Handy, R. L. (1967). “X-ray diffraction
915–929. and adsorption isotherm studies of the calcium montmorillonite-H2 O
Mitchell, J. K., and Soga, K. (2005). Fundamentals of soil behavior, system.” Transp. Res. Rec., 209, 23–54.
3rd Ed., Wiley, New York. Terzaghi, K. (1943). Theoretical soil mechanics, Wiley, New York.
Sasekaran, M. (2011). “Impact of permeability and surface Twain, M. (1874). Life on the Mississippi, Harper Brothers, New York.
cracks on soil slopes.” M.S. dissertation, School of Mechanical, Zhou, Y. D., Cheuk, C. Y., and Tham, L. G. (2009). “Deformation and
Aerospace, and Civil Engineering, Univ. of Manchester, Manchester, crack development of a nailed loose fill slope subjected to water
U.K. infiltration.” Landslides, 6(4), 299–308.
Downloaded from ascelibrary.org by THE UNIVERSITY OF NEWCASTLE on 03/03/17. Copyright ASCE. For personal use only; all rights reserved.

© ASCE 06017006-4 J. Geotech. Geoenviron. Eng.

J. Geotech. Geoenviron. Eng., -1--1