STP 1568-2013

Selected Technical Papers STP1568
Mechanical Properties of Frozen Soils
Editors:
Hannele Zubeck
Zhaohui Yang
ASTM International
100 Barr Harbor Drive
PO Box C700
West Conshohocken, PA 19438-2959
Printed in the U.S.A.
ASTM Stock #: STP1568

Library of Congress Cataloging-in-Publication Data
Mechanical properties of frozen soils / editors, Hannele Zubeck, Zhaohui Yang.
pages cm. -- (STP ; 1568)
“Standards worldwide.”
“ASTM stock #: STP1568.”
“Contains peer-reviewed papers that were presented at a symposium held January 31, 2013 in
Jacksonville, FL; the symposium was sponsored by ASTM International Committee D18 on
Soil and Rock and Subcommittee D18.19 on Frozen Soils and Rock”--Title page verso.
Includes bibliographical references and index.
ISBN 978-0-8031-7556-3 (alkaline paper)
1. Frozen ground--Congresses. 2. Soil mechanics--Congresses. 3. Soils--Testing--Standards--
Congresses. I. Zubeck, Hannele K. II. Yang, Zhaohui (Joey) III. ASTM International
Committee D18 on Soil and Rock. IV. ASTM International Committee D18 on Soil and Rock.
Subcommittee D18.19 on Frozen Soils and Rock.
TA713.M365 2013
624.1’5136--dc23 2013037685
Copyright © 2013 ASTM INTERNATIONAL, West Conshohocken, PA. All rights reserved.
This material may not be reproduced or copied, in whole or in part, in any printed, mechanical,
electronic, film, or other distribution and storage media, without the written consent of the
publisher.
Photocopy Rights
Authorization to photocopy items for internal, personal, or educational classroom use, or the
internal, personal, or educational classroom use of specific clients, is granted by ASTM
International provided that the appropriate fee is paid to ASTM International, 100 Barr Harbor
Drive, P.O. Box C700, West Conshohocken, PA 19428-2959, Tel: 610-832-9634; online:
http://www.astm.org/copyright.
The Society is not responsible, as a body, for the statements and opinions expressed in this
publication. ASTM International does not endorse any products represented in this publication.
Peer Review Policy
Each paper published in this volume was evaluated by two peer reviewers and at least one editor.
The authors addressed all of the reviewers’ comments to the satisfaction of both the technical
editor(s) and the ASTM International Committee on Publications.
The quality of the papers in this publication reflects not only the obvious efforts of the authors
and the technical editor(s), but also the work of the peer reviewers. In keeping with long-standing
publication practices, ASTM International maintains the anonymity of the peer reviewers. The
ASTM International Committee on Publications acknowledges with appreciation their dedication
and contribution of time and effort on behalf of ASTM International.
Citation of Papers
When citing papers from this publication, the appropriate citation includes the paper authors, “paper
title”, STP title and volume, STP number, Paper doi, ASTM International, West Conshohocken,
PA, Paper, year listed in the footnote of the paper. A citation is provided as a footnote on page one
of each paper.
Printed in Bay Shore, NY

September, 2013
Foreword
THIS COMPILATION OF Selected Technical Papers, STP1568, on Mechanical
Properties of Frozen Soils, contains peer-reviewed papers that were presented
at a symposium held January 31, 2013 in Jacksonville, FL. The symposium was
sponsored by ASTM International Committee D18 on Soil and Rock and Sub-
committee D18.19 on Frozen Soils and Rock.
The Symposium Co-Chairpersons and STP Co-Editors are Hannele Zubeck
and Zhaohui Yang, University of Alaska Anchorage, Anchorage, AK, USA.
Contents
Overview ............................................................ vii
Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ix
FREEZE-THAW EFFECTS ON MECHANICAL PROPERTIES

Freeze–Thaw Effects on Stiffness of Unbound Recycled Road Base
O. Bozyurt, A. K. Keene, J. M. Tinjum, T. B. Edil, and D. Fratta ................... 3
Characteristics of Chicago Blue Clay Subjected to a Freeze–Thaw Cycle
C. W. Swan, A. Grant, and A. Kody ....................................... 22
Investigation of Increased Hydraulic Conductivity of Silty Till Subjected to Freeze–
Thaw Cycles
G. P. Makusa, H. Mattsson, and S. Knutsson ................................ 33
TESTING OF MECHANICAL PROPERTIES

Sampling, Machining, and Testing of Naturally Frozen Soils
B. Still, Z. (Joey) Yang, and X. Ge ........................................ 49
Practice of Testing Frozen Soils
F. E. Oestgaard and H. K. Zubeck ........................................ 62
Triaxial Testing of Frozen Soils—State of the Art
T. S. Kornfield and H. K. Zubeck ......................................... 76
Experimental Study on Dynamic Properties of Clay Modified by Aught-Set
Solidifying Agent Subjected to Freeze-Thaw Cycles
J. Liu, J. Zhang, M. Chen, and T. Wang .................................... 86
MECHANICAL PROPERTIES
Mechanisms Controlling the Initial Stiffness of Frozen Sand
G. Da Re, J. T. Germaine, and C. C. Ladd .................................. 97
Mechanical Properties of Naturally Frozen Ice-Rich Silty Soils
X. Ge, Z. (Joey) Yang, and B. Still ......................................... 124
Properties of Elastic Waves in Sand–Silt Mixtures Due to Freezing
J.-H. Park, J.-S. Lee, S.-S. Hong, and Y.-S. Kim .............................. 140
Strength and Creep Properties of a Frozen Coastal Sand in Saltwater
S. J. Vitton and M. R. Muszynski ......................................... 153
CLASSIFICATION AND EFFECTS OF MECHANICAL

PROPERTIES ON PERFORMANCE
Frozen-Soil Classification with Index Testing
B. Still, S. Proskin, H. Zubeck, and Z. Yang ................................. 169
Investigation of the Dynamic Behaviors of Frozen Saline Silt with the Use
of a Spherical Stamp
A. Sinitsyn ......................................................... 180
Overview
The ASTM International Committee D18.19 has jurisdiction over seven frozen
soil standards, and is considering whether more standards are needed; especially
in the area of frozen soil mechanical properties. The symposium offered a forum
for the exchange of ideas on current research (nationally and internationally) as
it relates to testing of mechanical properties of frozen ground; it also provided a
rationale for the various details within new standards for testing of frozen soils.
Thirteen selected technical papers (STP) were submitted for publication and
presented at the symposium. These papers portray ideas of authors from Canada,
China, Norway, South Korea, Sweden and U.S.A. The papers are divided into
four topics: Freeze-thaw Effects on Mechanical Properties, Testing of Mechanical
Properties, Mechanical Properties, and Classification and Effects of Mechanical
Properties on Performance. In addition, the symposium provided a round table
meeting where future needs in the area of frozen soil research were proposed and
discussed.
The success of the symposium was due to a large group of ASTM personnel
including Mary Mikolajewski, ASTM International Manager, Symposia Opera-
tions; Hannah Sparks, ASTM International Administrative Assistant, Sympo-
sia Operations; Heather Blasco, Editorial Office, and numerous volunteers
who devoted many hours in the planning, implementation and execution of the
symposium. The volunteers behind the symposium were D18.96 Symposia Chair
Joe Richey, and a large number of paper reviewers from the following volunteer
groups: D18.19 members, American Society of Civil Engineer’s Technical Council
on Cold Regions Engineering, TRB AFP50 Seasonal Climatic Effects on Trans-
portation Infrastructure, International Society on Soil Mechanics and Geotechni-
cal Engineering - Technical Committee on Frost, TC 8, and International Perma-
frost Association’s Permafrost Engineering Working Group. Sincere appreciation
is extended to the authors, speakers and moderators for their contributions.
Hannele Zubeck and
Zhaohui Joey Yang
vii
Discussion
Participants of the ASTM International Symposium on Mechanical Properties of
Frozen Soil discussed two issues during the symposium: the research needs in
frozen soil related engineering and frozen soil standard development needs.
One of the pressing problems presented by Chinese researchers is related to
the high-speed railway (HSR). In Northeastern China, fine grained soils exist
extensively and frost heaving poses a serious threat to the safe operation of
newly constructed HSR. In fact, HSR speed has to be reduced from 300 km/h
to 200 km/h due to uneven roadbed caused by frost heaving. In Northwestern
China, the Qinghai-Tibet railroad was completed in 2005 and has been operat-
ing since July 2006. The railroad crosses 630 km of warm permafrost. There are
already clear signs (ponds and subsidence of roadbed, etc.) showing degradation
of the warm permafrost along the railway. Participants agree that more stud-
ies are needed regarding the effectiveness of geophysical methods in detecting
permafrost distribution and properties. In addition, participants also felt that
engineering practice guidelines are lacking in dealing with warm permafrost
and that more research is needed regarding the engineering properties of warm
permafrost.
The following standard development needs were listed and the participants
recommend that D18.19 considers developing the following standards: Thaw
consolidation test, shear stress test, triaxial compression test, unfrozen water
content test, segregation potential test, handling-transportation-storage of
frozen samples for laboratory testing, and sample preparation for laboratory
testing.
ix
FREEZE-THAW EFFECTS
ON MECHANICAL PROPERTIES
STP 1568, 2013
Available online at www.astm.org
DOI:10.1520/STP156820130014
Ozlem Bozyurt,1 Andrew K. Keene,2 James M. Tinjum,3

Tuncer B. Edil,4 and Dante Fratta5
Freeze–Thaw Effects on Stiffness of Unbound

Recycled Road Base
REFERENCE: Bozyurt, Ozlem, Keene, Andrew K., Tinjum, James M., Edil,
Tuncer B., and Fratta, Dante, “Freeze–Thaw Effects on Stiffness of Unbound
Recycled Road Base,” Mechanical Properties of Frozen Soils, STP 1568,
Hannele Zubeck and Zhaohui Yang, Eds., pp. 3–21, doi:10.1520/
STP156820130014, ASTM International, West Conshohocken, PA 2013.6
ABSTRACT: A major concern for using recycled pavement material as an
unbound base or subbase layer is the effect of changing seasons on the
properties of the recycled material. Three sources of recycled concrete ag-
gregate (RCA) and recycled asphalt pavement (RAP), and one conventional
base aggregate, were used to investigate the effect of freeze–thaw cycles on
the stiffness of unbound road base/subbase layers. Effects of freeze–thaw
cycling on the mechanical behavior of three gradations (coarse, medium,
fine) of recycled materials were systematically evaluated to determine how
Manuscript received January 17, 2013; accepted for publication July 1, 2013; published online
August 22, 2013.
1
Formerly at Univ. of Wisconsin-Madison, Madison, WI 53706, United States of America,
e-mail: ozlembozyurt@gmail.com
2
Graduate Research Assistant, Civil, Architectural, and Environmental Engineering, Univ. of
Texas at Austin, Austin, TX 78712, United States of America; and formerly at Univ. of
Wisconsin-Madison, Madison, WI 53706, United States of America, e-mail:
akkeene@utexas.edu
3
Assistant Professor, Dept. of Engineering Professional Development, Univ. of Wisconsin-
Madison, Madison, WI 53706, United States of America, e-mail: tinjum@epd.engr.wisc.edu
4
Professor Emeritus, Research Director, Recycled Materials Resource Center and Civil and
Environmental Engineering, Univ. of Wisconsin-Madison, Madison, WI 53706, United States of
America, e-mail: tbedil@wisc.edu
5
Associate Professor, Geological Engineering, Univ. of Wisconsin-Madison, Madison, WI
53706, United States of America, e-mail: fratta@wisc.edu
6
ASTM Symposium on Mechanical Properties of Frozen Soils on January 31, 2013 in
Jacksonville, FL.
Copyright V
C 2013 by ASTM International, 100 Barr Harbor Drive, PO Box C700, West Conshohocken, PA
19428-2959.
3
4 STP 1568 ON MECHANICAL PROPERTIES OF FROZEN SOILS
climatic factors and aging affect the resilient modulus. Sealed specimens
were exposed to 5, 10, and 20 sets of freeze–thaw cycles. Resilient modulus
tests were conducted according to NCHRP 1-28A after the final freeze–thaw
cycle. Freeze–thaw cycling caused a decrease in the stiffness (i.e., the sum-
mary resilient modulus) of RAP samples and class 5 aggregate because of
the effect of the water retained in the pores. An increase in the stiffness of
RCA was observed over 20 freeze–thaw cycles and is attributed to self-
cementitious behavior of crushed concrete. Seismic modulus testing was
used to investigate the continuous rate of change (daily) of the stiffness for
RCA and class 5 aggregate. The seismic modulus test confirmed the trends
observed in resilient modulus testing and served as a non-destructive
method for tracking changes in stiffness over time and freeze–thaw cycling.
KEYWORDS: recycled concrete aggregate, recycled asphalt pavement,
crushed rock aggregate, unbound base/subbase, resilient modulus, seismic
modulus, freeze–thaw
Introduction
The U.S. Geological Survey estimated that 1.19 109 tons of crushed stone
(natural aggregate) is consumed annually in the United States, 82 % as con-
struction material in road base or road-surfacing material [1]. However, rapidly
decreasing sources of natural aggregate, along with limits placed on aggregate
production by environmental regulation and land use policies, have caused the
price of these materials to increase dramatically [2]. The use of recycled mate-
rial as road base/subbase layers in new or rehabilitated roadway construction is
becoming more common in the United States [3]. Significant societal benefits
of using recycled material in pavement construction include reduction of
greenhouse gas emissions and energy consumption, as well as conservation of
natural aggregate and landfill capacity [4,5].
Common recycled materials used as unbound base course in pavement con-
struction include recycled concrete aggregate (RCA) and recycled asphalt pave-
ment (RAP). RCA is produced from demolition of concrete structures, such as
buildings, roads, and runways, and RAP is produced by removal and reprocess-
ing of existing asphalt concrete pavement [6–8]. In addition to the environmental
benefits, use of RAP and RCA in road base applications is financially competi-
tive with natural aggregate [7,8]. Despite increased acceptance of recycled base
material in construction, research concerning the mechanical properties and du-
rability of such materials is limited [2,7]. Schaertl [9] indicated that RCA and
RAP used alone or in blends with natural aggregate can have significantly differ-
ent resilient modulus (Mr), durability and rutting performance, and sensitivity to
stress state as compared to natural aggregates. The objective of this study was to
evaluate how freeze–thaw (F-T) cycles affect the stiffness of unbound (i.e., with-
out treatment or stabilization) base or subbase made of RCA or RAP by using
resilient modulus and seismic modulus testing methods. Resilient modulus tests
are the parameters primarily used for highway design and for MEPDG. Effects
of F-T cycles are shown in Mr tests and can be considered in highway design
BOZYURT ET AL., doi:10.1520/STP156820130014 5
and maintenance. The use of seismic testing in pavement health monitoring is

becoming more popular because it is a non-destructive and relatively cheap
application. Use of it in this study was to validate Mr testing and its potential use
to detect changing mechanical properties.
Background
Unbound RAP and RCA

Pavement systems are designed to withstand, for a given lifespan, stresses imposed
by traffic and damaging effects of environmental factors [10]. In 2009, the
National Asphalt Pavement Association (NAPA) reported that asphalt pavement is
the most recycled material in the United States [11]. The U.S. highway construc-
tion industry annually produces more than 100 million tons of RAP, which are
beneficially reused or recycled into new pavement [11]. According to the Federal
Highway Administration (FHWA), RAP is a valuable and high-quality material
that may demonstrate good performance as a granular road base and a replacement
for more expensive natural aggregates [12]. Kim et al. [13] investigated the stiff-
ness of base course containing different ratios of RAP and natural aggregate and
found that 50 % aggregate-50 % RAP specimens possess stiffness equivalent to
100 % aggregate specimens at lower confining pressures (20 kPa); but at higher
confinement (120 kPa), the RAP specimens were stiffer.
Recycled concrete aggregate used in pavement construction predominantly
replaces granular aggregates from natural sources and cement-treated subbase
layers [4]. Molenaar and van Niekerk [14] and Park [15] investigated the engi-
neering properties of RCA and suggested that good-quality road base/subbase
layers can be built from these materials. The Federal Highway Administration
reported that, when compared to natural aggregates, RCA has lower density,
higher water absorption, and higher soundness mass loss (i.e., lower durabil-
ity). In most cases, properties of RCA are within the specifications for base
course or concrete aggregate [8].
Bennert et al. [3] compared the mechanical properties of two types of
recycled material (RAP and RCA) to dense-graded aggregate base course used
in roadway base applications. The RAP and RCA were mixed at varying per-
centages with dense-graded aggregate base course. Bennert et al. [3] found that
the pure RAP and RCA samples had higher stiffness than the dense-graded ag-
gregate base course; in addition, the stiffness of the base course increased as
RAP and RCA content increased.
Freeze–Thaw Effect on RAP and RCA

Seasonal variation in moisture and temperature occurs in most areas of the
United States. The stiffness of road base/subbase layers tends to change
throughout the pavement’s life because of these seasonal variations. F-T cy-
cling of pavement profiles may significantly influence pavement performance.
When air temperature at the ground surface is lower than the temperature of
the ground, heat is transferred from the ground causing a drop in ground tem-
perature. If the surface temperature is below 0 C, a freezing front advances
into the ground. In cold regions, the resilient properties demonstrate extensive
variation because of frost conditions [16].
The Mr of an aggregate base/subbase generally increases during freezing
and decreases during thawing [17]. Jong et al. [18] concluded that frozen pave-
ment can sustain loads greater than the design loads without being damaged;
however, after thawing, the weakened pavement cannot support the load for
which it was originally designed and damage can occur. Therefore, in regions
where variations in temperature and moisture are appreciable, these factors
should be considered in pavement design [19].
Resilient Modulus
The design of roadway pavement relies on proper characterization of the repet-
itive load-deformation response of the pavement layers [20]. Base and sub-
grade layers deform when subjected to repeated loads from vehicular traffic.
Resilient modulus defines the non-linear elastic response of pavement geoma-
terials, such as unbound aggregate base/subbase, under traffic loading. The
resilient behavior of unbound aggregate layers is affected by the stress state
experienced because of wheel loading and confinement and the physical prop-
erties of the aggregate [21].
Mr is a modulus obtained from dynamic loading, defined as the ratio of
cyclic deviator stress to resilient (recoverable) strain given by
rd
Mr ¼ (1)
er
where er is the recoverable or elastic strain and rd is the applied cyclic deviator
stress. Mr is a key input in NCHRP 1-37 (mechanistic-based pavement design),
which is being evaluated for adoption by state highway agencies [21]. The per-
formance of pavement is dependent on the stiffness of the pavement structure
under specified traffic loads and environmental conditions.
Seismic Modulus
Seismic testing methods are commonly used for field and laboratory investiga-
tions for non-destructive determination of the modulus. Traditional laboratory
seismic modulus testing methods (ASTM C215 [22] and ASTM C1198 [23])
determine the dynamic Young’s modulus, shear modulus, and Poisson’s ratio of
specimens with high stiffness, such as Portland cement concrete and ceramics.
Modified testing protocols can be used for testing of unbound base/subbase mate-
rials through adaptation of equipment and data-acquisition software [24].
There are several procedures for determining elastic moduli of soil specimens
in the laboratory using seismic testing methods. General methods involve excita-
tion of the specimen in a manner that generates p and s waves for calculating con-
strained modulus and shear modulus, respectively. Obtaining these parameters
allows for the calculation of other important material properties, such as the
dynamic Young’s modulus and Poisson’s ratio. Methods appropriate for this type
of laboratory investigation in unbound materials where examined by Pucci [25]
and Toros and Hiltunen [26]; however, pressurized confinement or material sup-
port is required because of the loose nature of coarse-grained materials considered
in this study. The use of seismic modulus (SM) can be advantageous for tracking
the timeline of curing for unbound recycled concrete aggregates, similar to usage
in tracking curing of concrete specimens in ASTM C215 [22].
Materials
The recycled materials used in this study were obtained from various states in
the United States. Three RAPs and three RCAs were collected and named
according to state of origin. The materials represent coarser, medium, and finer
gradations based on their grain size (D50, Cc, and Cu). The reference base
course used as the control in this study was a gravel meeting class 5 aggregate
specifications for base course per the Minnesota Department of Transportation.
A summary of the index properties and soil classifications is shown in
Table 1. The materials used in this study are classified as non-plastic per
ASTM D2487 [27], the Unified Soil Classification System (USCS). The
recycled materials (three RCAs and three RAPs) classified as A-1-a and class 5
aggregate classified as A-1-b according to the AASHTO soil classification sys-
tem (ASTM D3282 [28]). Specific gravity (Gs) and absorption tests were
TABLE 1—Index properties of recycled materials and class 5 aggregate.
D50 AB AC wopt cdmax Gravel Fines

Material States (mm) Cu Cc Gs (%) (%) (%) (kN/m3) (%) (%) USCS
Class 5 aggregate MN 1.0 21 1.4 2.6 — — 8.9 20.1 22.9 9.5 GW-GM
RCA CA 4.8 22 1.4 2.3 5.0 — 10.4 19.9 50.6 2.3 GW

MI 9.7 35 3.9 2.3 5.4 — 8.7 20.8 68.5 3.2 GP
TX 13.3 38 6.0 2.3 5.5 — 9.2 19.7 76.3 2.1 GW
RAP CA 3.0 13 1.2 2.4 2.0 5.7 6.1 20.7 36.8 1.8 SW
MN 1.6 7 0.7 2.4 1.8 7.1 6.7 20.8 26.3 2.5 SP
TX 5.4 11 1.1 2.3 1.3 4.7 8.1 20.3 54.2 1.0 GW
Note: AC, Asphalt content; AB, absorption; MN, Minnesota; CA, California; MI, Michigan; TX, Texas.
FIG. 1—Particle size distribution for RCA, RAP, and class 5 aggregate and
lower and upper limits of RAP/RCA from the literature.
conducted according to AASHTO T85. Asphalt content was determined by
ASTM 6307 [29]. The modified Proctor compaction test (ASTM D1557 [30])
was performed to determine the optimum moisture content (wopt) and maxi-
mum dry unit weight (cdmax). The particle size distributions (PSD) for the
investigated materials were determined according to ASTM D422 [31] and
shown in Fig. 1, along with upper and lower bounds reported in the literature
[3,6,7,32,33].
Methods
Freeze-Thaw Cycling
A method that follows ASTM D6035 [34] for specimen conditioning was used
to assess frost susceptibility, as detailed in Camargo et al. [33]. Test specimens
were compacted in plastic molds at wopt and 95 % cd. Specimens, instrumented
with a thermocouple, were monitored at 19 C for 24 h to confirm that complete
freezing occurred within that time. Details of specimen preparation and compac-
tion procedure can be found in Bozyurt [35]. Specimens were retained in their
plastic mold with ends exposed but carefully wrapped with plastic sheeting to
prevent exposure to moisture during F-T cycling, thus maintaining bulk water
content. Each F-T cycle consisted of exposing each specimen to 19 C for 24 h,
followed by room temperature for another 24 h. The effect of F-T cycling on the
engineering properties of recycled materials was determined by testing the Mr af-

ter selected number of F-T cycles and the seismic modulus after each F-T cycle
in the fully thawed state. For the Mr, after the select number of F-T cycles, speci-
mens were extruded frozen (to maintain specimen integrity), then allowed to
thaw inside the resilient modulus cell. Because of the destructive nature of the
rResilient modulus test, the F-T process had to be repeated on replicate speci-
mens. 5, 10, and 20 cycles were selected for this study.
It is important to note that ASTM D6035 [34] does not enable cryosuction
to absorb additional water into the material while the material is freezing. Con-
sequently, some minor aspects of field behavior may not have been adequately
captured in this study, and laboratory results may represent a slightly higher
stiffness than may exist in the field.
Resilient Modulus
Resilient modulus tests were performed on compacted specimens according to
NCHRP 1-28 A Procedure Ia, which applies to base and subbase. The materials
used in this study classify as type I in NCHRP 1-28 A, which requires a 152-
mm-diameter and 305 -m-high specimen for Mr testing [36]. Specimens were
prepared at wopt and 95 % cdmax.
The displacement corresponding to the applied load in the Mr tests was
measured using internal and external linear variable displacement transducers
(LVDT). The resilient modulus was obtained by averaging the Mr from the last
five cycles of each load sequence. The Mr data were fit to the power function
model proposed by Moosazedh and Witczak [37]:
k2
h
Mr ¼ k1 (2)
Pr
where h is bulk stress, Pr is reference pressure, and k1 and k2 are empirical fit-
ting parameters. For base course, the summary resilient modulus (SRM) corre-
sponds to the Mr at a bulk stress of 208 kPa, as suggested by Section 10.3.3.9
of NCHRP 1-28 a. SRM is used to determine the layer coefficient, which is a
required input in the AASHTO pavement design equation [20]. The power
function (Eq 2) is a simple model widely used for granular material. The esti-
mated SRM per the power function model was compared to the measured mod-
ulus. Statistical analysis indicated that results from the power function model
are significant at a 95 % confidence level, and the model represents the data
reasonably well for RCA (R2 ¼ 0.85) and for RAP (R2 ¼ 0.90) [35].
Seismic Modulus
Because of the lengthy duration of resilient modulus testing (approximately
6 h) in which the specimen is destroyed in the performance of the test, a faster
and non-destructive seismic testing method was performed to calculate the

stiffness of the recycled materials without confining pressure [26]. Seismic
modulus testing was used to evaluate the temporal change in stiffness because
of the effects of freeze/thaw cycling.
In seismic testing, the material stiffness is captured using the elastic wave
propagation method. For the seismic modulus, the testing was done through the
exposed ends, on the same specimens that were to undergo resilient modulus test-
ing after the selected F-T cycles. While each specimen was retained in their plas-
tic mold, seismic waves were generated with an impact hammer (100 gm), which
sends a compression wave (p wave) through the specimen, which is received by
an accelerometer. The p-wave velocity (Vp) is calculated according to
L
Vp ¼ (3)
t2 t1
where L is the length of the specimen, t1 is the departure time of the p wave fol-
lowing excitation, and t2 is the arrival time of the p wave. An oscilloscope was
used to record the wave propagation and travel time within the specimen. From
the p-wave velocity obtained from the SM test, the low-strain constrained mod-
ulus (D) is calculated as
D ¼ Vp2 q (4)
where q is the mass density.
Results
Recycled Asphalt Pavement (RAP)
Resilient Modulus—The effect of F-T cycling on SRM for the RAPs com-
pared with class 5 aggregate is shown in Fig. 2(a). Full test data are given in
Bozyurt [35]. F-T cycling has a relatively small effect (7 %) decrease over
five F-T cycles) on the SRM of class 5 aggregate in comparison to the RAPs.
Camargo et al. [33] also observed a 7 % decrease in SRM for a natural aggre-
gate after five F-T cycles. However, the rate of decrease for class 5 aggregate
over 10 and 20 F-T cycles was 14 % and 21 %, respectively. The SRM of the
RAPs showed the most reduction after the first five F-T cycles, with relatively
small change thereafter. The differences in the effects of F-T cycles on a ma-
terial can be attributed to the differences in material gradation, mechanical
properties, and mineralogy and origin of aggregate. For instance, RAP (TX)
(coarser) exhibited the smallest rate of decrease (28 %) in SRM after 20 F-T
cycles compared with RAP (CA) (medium) (32 %) and RAP (MN) (finer)
FIG. 2—(a) Internal SRM of RAP class 5 aggregate during 0, 5, 10, and 20
freeze–thaw cycles. (b) Normalized SRM of RAP class 5 aggregate during 0, 5,
10, and 20 freeze–thaw cycles.
(53 %). The rate of decrease of SRM for RAP ranged from 20 % to 66 %,
which is similar to the range reported by Rosa [16] for various coarse- and
fine-grained soils. Despite the decrease in SRM of the RAPs over 20 F-T
cycles, the SRM of the RAPs remained greater than that of class 5 aggregate
(see Fig. 2).
The reduction in stiffness of RAP over time may be related to the volume
change of water retained in the pores, the hydrophobicity of asphalt, and weak-
ness in the asphalt binders. A study conducted by Rosa [16] on the F-T effect
on engineering properties of recycled pavement materials (RPM) (a mixture of
recycled asphalt layer and base course material) found that the initial Mr
decreased after five F-T cycles. Rosa [16] confirmed that when pore water
freezes within the unbound base/subbase layer, the volume of voids increases,
and the resulting volume change causes degradation and decreased stiffness.
This was attributed by Arm [38] to poor F-T resistance. Therefore, the pave-
ment moduli change during F-T cycles might occur as a result of changes in
the phase of the pore water over time [18].
In this study, relatively low volume changes were observed because speci-
mens underwent F-T cycles in a closed system (i.e., no external source of
water). Degradation of the material was attributed to volumetric changes
because of freezing and thawing of the water within the pore space of the mate-
rial. However, no net change in weight occurred in the material over the
selected F-T cycles.
An increase in the stiffness of RAP after F-T cycling has been reported in
other studies. For example, Attia and Abdelrahman [39] reported that Mr of
RAP increased after two F-T cycles for specimens that were kept in latex mem-
branes to keep bulk moisture content constant during the test. However, during
their conditioning process, a significant amount of water loss occurred, which
would be a significant factor for the Mr increase. Because there is an asphalt
coating around the fine particles in RAP, water-retention capabilities are less
than those in the natural aggregates that are also included in RPM [40].
Seismic Modulus—Seismic modulus testing was conducted on RAP speci-

mens. There was no identifiable trend during F-T cycling observed in the data
gathered in this study. Identifying factors that contributed to the inability of the
seismic method to identify stiffness in RAP was beyond the scope of this
study.
Recycled Concrete Aggregate (RCA)
Resilient Modulus—The effect of F-T cycling on the SRM of representative

RCA samples compared with class 5 aggregate is presented in Fig. 3(a). For
the RCAs, SRM decreased after five F-T cycles, followed by a consistent
increase. The rate of decrease during the first five F-T cycles varied. As the
FIG. 3—(a) Internal SRM of RCA and class 5 aggregate during 0, 5, 10, and
20 freeze–thaw cycles. (b) Normalized SRM of RCA and class 5 aggregate dur-
ing 0, 5, and 20 freeze–thaw cycles.
source and origin of the RCAs differ (i.e., the gradation, compaction character-
istics, and mechanical properties differ), these variations affect the rate of
change in SRM. The SRM for RCA (TX) decreased 10 % over five F-T cycles
followed by an increase over 20 F-T cycles to 30 % above the initial SRM (i.e.,
0 F-T cycles). RCA (MI) decreased 18 % over five F-T cycles followed by an
increase over 20 F-T cycles of 38 % above the initial SRM over 20 F-T cycles.
The same trend was observed for RCA (CA), with an 11 % decrease over five
F-T cycles followed by an increase over 10 F-T cycles to 5 % above the initial
SRM.
Molenaar and van Niekerk [14] reported that gradation has the least influ-
ence on Mr for unbound base course made with recycled concrete. Therefore,
the difference in SRM between RCA from different states may not be related
to the initial gradation of materials. As seen in Fig. 3(b), normalized SRM val-
ues reveal a trend in percentage change of specimen stiffness over F-T cy-
cling. The self-cementing properties of RCA and fine content generation over
time could explain why an increase in stiffness after five F-T cycles occurred.
These trends are consistent with other research in which the strength of sub-
base prepared with RCA has been found to increase with time [38]. RCA par-
ticles typically have a coarser and more angular shape than natural aggregates
as a result of material crushing and processing operations [6], leaving a signif-
icant amount of mortar adhered to the surface of the particles [6,41,42]. Proc-
essed RCA has hardened cement paste that holds smaller aggregate particles
together [4]. The amount of cement paste attached to aggregate in RCA
depends on the process used to produce RCA and the properties of the original
concrete [43].
Poon et al. [44] stated that unhydrated cement content retained within the
adhered mortar was the cause of self-cementing in RCA used for unbound
base. Arm [38] conducted a field investigation over 2 years on the stiffness of
unbound base layers made of crushed concrete from demolished structures. An
increase in Mr with time was observed and attributed to the self-cementing
properties of RCA. Arm [38] conducted repeated load triaxial tests on crushed
virgin aggregate and concrete specimens after certain storing periods (1, 3, 7,
28, and 90 days). An increase in modulus was observed for crushed concrete
specimens, but not for natural base layers, over time. Arm [38] postulated that
the self-cementing properties of crushed concrete were the reason behind the
increase of stiffness, with time, in unbound base layers made with crushed
demolished concrete.
There was an increase observed in the fines percentage of RCA specimens
after 20 F-T cycles. Recent studies show that an increase in fines has an impor-
tant effect on the stiffness of road base/subbase aggregates. Mishra et al. [45]
investigated the effect of fines on compaction of dolomite samples and found
that MDU increased as the percentage of fines in the sample increase. Because
the addition of fines gradually filled the voids, the aggregate matrix became
denser. They also found that as the fines content increased beyond a certain
point, voids in the uncrushed gravel matrix (rounded aggregate particles having
lower total void space than crushed samples) were filled, and the contact points
among the coarse particles were no longer present within the particle matrix.
This resulted in a reduction in the dry density without a corresponding signifi-
cant decrease in aggregate material matrix strength. This phenomenon was also
observed by Ebrahimi et al. [46] during the investigation of ballast void filling
with fouling materials (i.e., fines and water content). For example, fines content
>12 % may significantly decrease the Mr of unbound granular materials [47].
The fines percentage in the soil matrix likely improves the Mr of unbound
aggregates to a point, after which the matrix starts to be dominated by the fines
and the Mr starts to decrease. Another reason for the observed increase in SRM
for RCA specimens may be because of the increased amount of fines in the ma-
trix of the specimens during F-T cycles.
Seismic Modulus—Seismic modulus testing was used to verify the trend in

changing stiffness of RCA as observed in Mr tests because of F-T cycles. The
SM method was conducted for each specimen that underwent F-T cycling and
Mr testing. The data from SM testing, shown in Fig. 4(a), revealed a trend
among the RCA and class 5 materials similar to that of the resilient modulus
tests. The SM for RCA (TX) decreased 12 % over two F-T cycles, followed
by an increase over 10 F-T cycles to 23 % above the initial SM; RCA
(MI) decreased 47 % over five F-T cycles followed by an increase over 10 F-T
cycles to 70 % above the initial SM; and RCA (CA) had an 11 % decrease
over five cycles followed by an increase over 10 F-T cycles to 60 % above the
initial SM.
Because of the nature of resilient modulus and seismic modulus testing, the
expected results may not directly correlate, but the trends should be consistent
[48]. In this study, the common trend between materials undergoing F-T cy-
cling were detected and confirmed by both methods.
Seismic modulus testing was also used to determine the resulting modu-
lus of the specimens during the frozen and thawed phases, and the results are
shown in Fig. 4(b). The frozen state modulus (Df) is much greater the thawed
state modulus (Dt). Within five F-T cycles, the Df showed a decreasing trend
and the Dt showed an increasing trend, indicating that the modulus at both
states were converging. For RCA (TX), an increase of 99.7 % from the initial
Dt was observed when the specimen was frozen, followed by a decrease of
99.8 % when the material thawed. The % increase from the initial Dt to initial
Df fell to 94 % after the fifth F-T cycle. A similar trend occurred for RCA
(CA) with an initial increase from Dt to Df, of 99.3 %, which fell by the fifth
F-T cycle to 96.4 %. The seismic modulus tests were only conducted for five
F-T cycles, a continuation of these tests could reveal more information
regarding behavior of the Df and Dt over additional F-T cycles.
FIG. 4—(a) Seismic modulus of RCA class 5 aggregate over 0, 5, 10, and 20
frozen and thawed states. (b) Seismic modulus of RCA (CA) and RCA (TX) for
frozen and thawed state.
Conclusions
Freeze/thaw cycling was found to influence the stiffness properties of unbound
recycled pavement and recycled concrete aggregates used for base course. Resil-
ient modulus can be used to investigate the effect of freeze–thaw (F-T) cycles on
unbound road base/subbase layers consisting of natural aggregate, RAP, and
RCA. The seismic modulus (SM) method is non-destructive and thus can be con-
ducted many times on the same specimen exposed to multiple freeze–thaw cycles.
For tracking the effect of freeze–thaw cycles on a recurring basis and determining
long-term effects for RCA and natural aggregate (class 5 aggregate), the SM
method is useful; however, the SM testing method did not work well for RAP.
Further study and corroboration of seismic and resilient modulus testing are under-
way at University of Wisconsin-Madison, Department of Geological Engineering.
The stiffness of RAP decreased over the first five F-T cycles, with smaller
decrease recorded thereafter. This decrease in stiffness of RAP subjected to F-T
cycles may be attributed to particle degradation and progressive asphalt-binder
weakening. For RCA, the exposure to F-T cycles led first to a decrease in stiff-
ness, followed by an increase, which may be attributed to progressive generation
of fines and hydration of cement paste. The seismic modulus method confirmed
the trends of changing stiffness of RCA during F-T cycling. Among the recycled
materials evaluated in this study, quantitative differences in F-T response was
observed, which was reflective of material grading and source. Exposure of the
natural aggregate control (class 5 aggregate) to F-T cycles resulted in relatively
small decreases in stiffness; however, the stiffness of the recycled materials was
always greater than the natural aggregate, even after F-T induced decreases.
Acknowledgments
These results are based on work supported by the TPF-5 [129] Recycled
Unbound Materials Pool Fund administered by The Minnesota Department of
Transportation and the Recycled Materials Resource Center (RMRC), which is
supported by the Federal Highway Administration. The opinions, findings, con-
clusions, or recommendations expressed herein are those of the author(s) and do
not necessarily represent the views of the sponsors. Seismic modulus testing data
were collected by Mr. B. Warren, an undergraduate researcher, Department of
Civil and Environmental Engineering, University of Wisconsin-Madison.
References
[1] USGS, “Crushed Stone: Statistical Compendium,” U.S. Geological Sur-

vey, Reston, VA, 2010.
[2] ACPA, “Recycling Concrete Pavements,” American Concrete Pavement
Association, Washington, DC, 2009.
[3] Bennert, T., Papp, W. J., Maher, A., and Gucunski, N., “Utilization of
Construction and Demolition Debris under Traffic-Type Loading in Base
and Subbase Applications,” Transportation Research Record, No. 1714,
2000, pp. 33–39.
[4] Saeed, A., Hammons, M. I., Feldman, D. R., and Poole T., “Evaluation,
Design and Construction Techniques for Airfield Concrete Pavement
Used as Recycled Material for Base,” Publication IPRF-07-g-002-03-5,
IPRF, Rosemont, IL, 2006.
[5] Lee, J. C., Edil, T. B., Benson, C. H., and Tinjum, J. M., 2010, “Use of
BE2ST in-Highways for Green Highway Construction Rating in
Wisconsin,” Proceedings of the 1st T&DI Green Streets & Highway Con-
ference, Denver, CO, 2010.
[6] Kuo, S. S., Mahgoub, H. S., and Nazef, A., “Investigation of Recycled
Concrete Made With Limestone Aggregate for a Base Course in Flexible
Pavement,” Transportation Research Record, No. 1787, 2002, pp.
99–108.
[7] Guthri, S. W., Cooley, D., and Eggett, D. L., 2007, “Effects of Reclaimed
Asphalt Pavement on Mechanical Properties of Base Materials,” Trans-
portation Research Record, No. 2005, 2007, pp. 44–52.
[8] FHWA, “User Guideline for Byproducts and Secondary Use Materials in
Pavement Construction,” Publication FHWA-RD-97-148, FHWA, U.S.
Department of Transportation, Washington, DC, 2008.
[9] Schaertl, G. J., 2010, “Scaling and Equivalency of Bench-Scale Tests to
Field Scale Condition,” M.S. thesis, Department of Civil and Environ-
mental Engineering, University of Wisconsin-Madison, Madison, WI.
[10] ARA, “Guide for Mechanistic-Empirical Design on New and Rehabili-
tated Pavement Structures,” Publication NCHRP Project 1-37A, NCHRP,
Washington, DC, 2004.
[11] NAPA, “How to Increase RAP Usage and Ensure Pavement Perform-
ance,” National Asphalt Pavement Association, Lanham, MD, 2009.
[12] FHWA, “Reclaimed Asphalt Pavement in Asphalt Mixtures: State of the
Practice,” Publication FHWA-HRT-11-021, FHWA, U.S. Department of
Transportation, Washington, DC, 2011.
[13] Kim, W., Labuz, J. F., and Dai, S., “Resilient Modulus of Base Course
Containing Recycled Asphalt Pavement,” Transportation Research Re-
cord, No. 2005, 2007, pp. 27–35.
[14] Molenaar, A. A. A. and van Niekerk, A. A., “Effects of Gradation, Com-
position, and Degree of Compaction on the Mechanical Characteristics of
Recycled Unbound Materials,” Transportation Research Record, No.
1787, 2007, pp. 73–82.
[15] Park, T., “Application of Construction and Building Debris as Base and
Subbase Materials in Rigid Pavement,” Transportation Research Record,
No. 129, 2003, pp. 558–563.
[16] Rosa, M., 2006, “Effect of Freeze and Thaw Cycling on Soils Stabilized
Using Fly Ash,” M.S. thesis, Department of Civil and Environmental En-
gineering, University of Wisconsin-Madison, Madison, WI.
[17] Kootstra, B. R., Ebrahimi, A., Edil, T. B., and Benson, C. H., “Plastic De-
formation of Recycled Base Materials,” GeoFlorida 2010: Advances in
Analysis, Modeling & Design, 2010, pp. 2682–2691.
[18] Jong, D. T., Bosscher, P. J., and Benson, C. H., 1998, “Field Assessment
of Changes in Pavement Moduli Caused by Freezing and Thawing,”
Transportation Research Record, No. 1615, pp. 41–48.
[19] Zaman, M. M., Laguros, J. H., and Zhu, J. G., “Durability Effects on
Resilient Moduli of Stabilized Aggregate Base,” Transportation Research
Record, No. 687, 1999, pp. 29–39.
[20] Tian, P., Zaman, M. M., and Laguros, J. G., “Gradation and Moisture
Effects on Resilient Moduli of Aggregate Bases,” Transportation
Research Record, No. 1619, 1998, pp. 75–84.
[21] Pan, T., Tutumluer, E., and Anochie-Boateng, J., “Aggregate Morphology
Affecting Resilient Behavior of Unbound Granular Materials,” Transpor-
tation Research Record, No. 1952, 2006, pp. 12–20.
[22] ASTM C215: Standard Test Method for Fundamental Transverse, Longi-
tudinal, and Torsional Frequencies of Concrete Specimens, Annual Book
of ASTM Standards, ASTM International, West Conshohocken, PA, 2008.
[23] ASTM C1198: Standard Test Method for Dynamic Young’s Modulus,
Shear Modulus, and Poisson’s Ratio for Advanced Ceramics by Sonic
Resonance, Annual Book of ASTM Standards, ASTM International, West
Conshohocken, PA, 2009.
[24] Williams, R. R. and Nazarian, S., “Correlation of Resilient and Seismic Mod-
ulus Test Results,” J. Mater. Civil Eng., Vol 19, No. 12, 2007, pp. 1026–1032.
[25] Pucci, M. J., 2010, “Development of a Multi-Measurement Confined
Free–Free Resonant Column Device and Initial Studies,” M.S. thesis,
Department of Civil, Architectural, and Environmental Engineering, Uni-
versity of Texas at Austin, Austin, TX.
[26] Toros, U. and Hiltunen, D. R., “Effects of Moisture and Time on Stiffness
of Unbound Aggregate Base Course Materials,” Transportation Research
Record, No. 2059, 2008, pp. 41–51.
[27] ASTM D2487: Standard Practice for Classification of Soils for Engineer-
ing Purposes (Unified Soil Classification System), Annual Book of ASTM
Standards, ASTM International, West Conshohocken, PA, 2011.
[28] ASTM D3282: Standard Practice for Classification of Soils and Soil-
Aggregate Mixtures for Highway Construction Purposes, Annual Book of
ASTM Standards, ASTM International, West Conshohocken, PA, 2009.
[29] ASTM D6307: Standard Test Method for Asphalt Content of Hot-Mix
Asphalt by Ignition Method, Annual Book of ASTM Standards, ASTM
International, West Conshohocken, PA, 2010.
[30] ASTM D1557: Standard Test Methods for Laboratory Compaction

Characteristics of Soil Using Modified Effort (56 000 ft-lbf/ft3 (2700
kN-m/m3)), Annual Book of ASTM Standards, ASTM International, West
[31] ASTM D422: Standard Test Method for Particle-Size Analysis of Soils,
Annual Book of ASTM Standards, ASTM International, West Consho-
hocken, PA, 2007.
[32] Bejarano, M. O., Harvey, J. T., and Lane, L., “In Situ Recycling of
Asphalt Concrete as Base Material in California,” Transportation
Research Record, 2003, p. 22.
[33] Camargo, F. F., Edil, T. B., and Benson, C. H., “Strength and Stiffness of
Recycled Base Materials Blended With Fly Ash,” Transportation
Research Record, 2009.
[34] ASTM D6035: Standard Test Method for Determining the Effect of
Freeze-Thaw on Hydraulic Conductivity of Compacted or Intact Soil
Specimens Using a Flexible Wall Permeameter, Annual Book of ASTM
[35] Bozyurt, O., 2011, “Behavior of Recycled Asphalt Pavement and
Recycled Concrete Aggregate as Unbound Road Base,” M.S. thesis,
Department of Civil and Environmental Engineering, University of Wis-
consin, Madison, WI.
[36] NCHRP, “Laboratory Determination of Resilient Modulus for Flexible
Pavement Design,” NCHRP Research Results Digest, NCHRP, Washing-
ton, DC, 2004.
[37] Moosazedh, J. and Witczak, M. W., “Prediction of Subgrade Moduli for
Soil that Exhibits Nonlinear Behavior,” Transportation Research Record,
No. 810, 1981, pp. 9–17.
[38] Arm, M., “Self-Cementing Properties of Crushed Demolished Concrete in
Unbound Layers: Results From Triaxial Tests and Field Tests,” Waste
Manage., Vol. 21, No. 3, 2001, pp. 235–239.
[39] Attia, M. and Abdelrahman M., “Modeling the Effect of Moisture on
Resilient Modulus of Untreated Reclaimed Asphalt Pavement,” Transpor-
tation Research Record, No. 2167, 2010, pp. 30–40.
[40] Nokkaew, K., Tinjum, J. M., and Benson, C. H., 2011, “Hydraulic Proper-
ties of Recycled Asphalt Pavement and Recycled Concrete Aggregate,”
GeoCongress. Oakland, CA, March 25–29, 2012, pp. 1476–1485.
[41] Juan, M. S. and Gutierrez, P. A., “Study on the Influence of Attached
Mortar Content on the Properties of Recycled Concrete Aggregate,” Con-
struct. Build. Mater., Vol. 23, 2009, pp. 872–877.
[42] Gokce, A., Nagataki, S., Saeki, T., and Hisada, M., 2011, “Identification
of Frost-Susceptible Recycled Concrete Aggregates for Durability of
Concrete,” Construct. Build. Mater., Vol. 25, 2011, pp. 2426–2431.
[43] Chini, A. R., Kuo, S. S., Armaghani, J. M., and Duxbury, J. P., “Test of
Recycled Concrete Aggregatenin Accelerated Test Track,” Transportation
Research Record, No. 6, 2001, pp. 486–492.
[44] Poon, C. S., Qiao, X. C., and Chan, D., “The Cause and Influence of Self-
Cementing Properties of Fine Recycled Concrete Aggregates on the Prop-
erties of Unbound Sub-Base,” Waste Manage., Vol. 26, No. 10, 2006, pp.
1166–1172.
[45] Mishra, D., Tutumluer, E., and Butt, A. A., “Quantifying Effects of Parti-
cle Shape and Type and Amount of Fines on Unbound Aggregate Per-
formance through Controlled Gradation,” Transportation Research
Record, No. 2167, 2010, pp. 61–71.
[46] Ebrahimi, A., Tinjum, J. M., and Edil, T. B., “Protocol for Testing Fouled
Railway Ballast in Large-Scale Cyclic Triaxial Equipment,” Geotech.
Test. J., ASTM, Vol. 35. No. 5, 2012, pp. 796–804.
[47] Barksdale, R. D. and Itani, S. Y., “Influence of Aggregate Shape on Base
Behavior,” Transportation Research Record, No. 1227, 1989, pp.
173–182.
[48] Schuettpelz, C. C., Fratta, D., and Edil, T. B., “Mechanistic Corrections
for Determining the Resilient Modulus of Base Course Materials Based
on Elastic Wave Measurements,” J. Geotech. Geoenviron. Eng., Vol. 136,
No. 8, 2010, pp. 1086–1094.
STP 1568, 2013
DOI:10.1520/STP156820130015
Christopher W. Swan,1 Alex Grant,2 and Alyssa Kody2
Characteristics of Chicago Blue Clay

Subjected to a Freeze–Thaw Cycle
REFERENCE: Swan, Christopher W., Grant, Alex, and Kody, Alyssa,

“Characteristics of Chicago Blue Clay Subjected to a Freeze–Thaw Cycle,”
Mechanical Properties of Frozen Soils, STP 1568, Hannele Zubeck and
Zhaohui Yang, Eds., pp. 22–32, doi:10.1520/STP156820130015, ASTM
International, West Conshohocken, PA 2013.3
ABSTRACT: This paper presents the preliminary results of a laboratory
study on the impact of one freeze–thaw cycle on the engineering properties
of natural samples of Chicago Blue Clay (CBC). Freezing was performed in a
closed system with freezing occurring simultaneously in all directions. Two
types of consolidation tests were performed: incremental loading and long-
term compression under a constant load. These tests were performed on
specimens subjected to a cycle of freeze–thaw with results compared to tests
on CBC in its natural state. For constant load consolidation tests, freezing
and subsequent thaw of specimens was performed under the applied stress.
Additionally, the Atterberg limits of CBC, in its natural state and after one
freeze–thaw cycle, were also determined. Based on the test results in this
study, it is clear that one cycle of freezing and thawing impacts both the con-
solidation behavior and limits of the samples. Specifically, for both the incre-
mental and constant loading tests there is a significant increase (between
25 % and 40 %) in settlement during consolidation versus results of tests on
CBC in its natural state. There is also a notable increase in the virgin com-
pression ratio (Cc) and coefficient of consolidation (cv) upon thawing. Results
of Atterberg limits indicate a decrease in the CBC’s liquid limit that explains,
in part, the increase in cv. Overall, the results indicate that the freezing and
subsequent thawing of CBC has a severe impact on its consolidation
Manuscript received January 18, 2013; accepted for publication June 7, 2013; published online
August 23, 2013.
1
Dept. of Civil and Environmental Engineering, Tufts Univ., Medford, MA 02155, United States
of America, e-mail: chris.swan@tufts.edu
2
Dept. of Civil and Environmental Engineering, Tufts Univ., Medford, MA 02155, United States
of America.
3
Jacksonville, FL.
Copyright V
19428-2959.
22
SWAN ET AL., doi:10.1520/STP156820130015 23
behavior. The writers recommend that more testing be performed to verify

and expand on the results determined from this preliminary evaluation.
KEYWORDS: freeze–thaw cycle, cohesive soils, consolidation, Chicago
Blue Clay
Motivation
The purpose of this project was to determine the impact that one cycle of
freeze–thaw can have on the consolidation of cohesive soils, specifically, Chi-
cago Blue Clay. This evaluation focused on simulating artificial freezing proc-
esses performed in the field. In geotechnical engineering, artificial ground
freezing is a technique that can contribute greatly to the feasibility of construct-
ing tunnels, mines, subways, foundations, and shafts, especially for complex
construction situations. For example, ground freezing is frequently used as an
earth support system during construction in soft soils because these soils, upon
freezing, are more stable, gaining the consistency of soft rock. Typically, in
projects where ground freezing is used for earth support, circulating pipes are
installed in the construction area. A brine solution (liquid nitrogen can also be
used) is pumped through the pipes at a temperature of 10 C to 20 C, lead-
ing to the freezing of the ground over a relatively short time period. But,
whereas ground freezing successfully provides support during construction, it
is important for designers to understand the behavior of the once-frozen soil
upon, and after, thawing. The relevance and importance of thaw-related settle-
ment is exemplified by the extensive settlements noted for work associated
with the Central Artery/Tunnel project in Boston, MA, commonly referred to
as the “Big Dig” [1]. This case provides a motivation for the investigation into
the impacts of a freeze–thaw cycle on the consolidation behavior of clay.
Big Dig Settlement Case

In the late 1990s, designers believed construction of the U.S. Interstate 90 con-
nector tunnel in Boston would be problematic because of (1) the existing soft
soils and high water table in the area, and (2) unacceptable disruption of the rail
service into South Station, one of Boston’s major transit hubs. Therefore, design-
ers developed a ground-freezing protocol to aid in the excavation support for the
tunneling beneath existing surficial railroad tracks. Both shallow (up to 15 ms
below ground surface) and deep (up to 43 ms) freezing pipes were installed to
successfully facilitate construction. Approximately 700 vertical pipes were in-
stalled and chilled brine was pumped through these pipes for 3 to 4 months prior
to excavation. Approximately 60 000 cubic meters of soil were frozen [2]. Tun-
nel construction in the area was completed and active freezing ended in 2002.
However, over the next 10 years, previously frozen soil thawed and has settled
more than anticipated, reaching surface settlements as much as much as 2.5 ms
at measured rates up to 2.5 cm per month. Though not directed measured, it is
believed that a cavity may have developed beneath the completed tunnel section.
Reports indicate the cavity could be as large as approximately 60 ms long by
18 ms wide. The Massachusetts Department of Transportation has preliminarily
planned to fill the cavity beneath the tunnel with concrete [1].
Previous Research
Mechanics of Freeze–Thaw on Porous Media

During freezing of a moist cohesive soil, negative pore water pressures are gen-
erated, which allows for ice segregation and changes in soil structure [3,4].
Research on thin cross sections of frozen material was performed to examine
ice lens formation [3]. In the direction of freezing, solid vertical ice lenses
formed, whereas ice formed polygon structures perpendicular to the freezing
front. In a study measuring the negative pore water pressure during freezing
[5], suction pressures “as large as 532 kPa” from an initial consolidation pres-
sure of 128 kPa were generated by one directional freezing of plastic clay soil
samples. These negative pore pressures allow for creep-like behavior in the
soil causing, in the case of freeze–thaw, rapid particle rearrangement and set-
tlement under a constant overburden of pressure.
Ice lens formation in a porous medium has been explored by a number of
researchers recently [6–9] and in clayey silts [10]. In total, these research
efforts indicate that when a sub-freezing temperature isotherm is applied to a
saturated soil, a frozen fringe will begin to form by water (or some other pore
liquid) freezing at the applied isotherm interface. This frozen fringe advances
because of the thermal imbalance that has been formed, and water is drawn to-
ward the fringe. A lens will continue to grow as long as the permeability in the
soil is such that water is able to flow across the fringe, and as soon as perme-
ability drops below this threshold, a new lens will begin to form ahead of the
existing ice lens in the soil profile. Efforts have been made to isolate the signifi-
cance of the rate of freezing on heave or post-thaw settlement. Generally, slow
freezing will lead to slower rates of ice lens segregation and ultimately more
heave/settlement [10,11]. Researchers have also postulated that the suction that
develops in the soil is also an important component of ice lens development,
and that this suction in the soil is inversely proportional to the hydraulic con-
ductivity of that soil [12]. Thus, significant heaving and freeze-thaw settlement
are to be expected in silty and clayey soils, whereas more porous sands and
gravels will be less affected.
Methods of Freezing
In situ seasonal ground freezing would be most accurately modeled as a three-
dimensional open system (access to free water during freezing) with a
SWAN ET AL., doi:10.1520/STP156820130015 25
fluctuating applied temperature at the open (top) surface. Given the difficulties
of creating such a system in the laboratory, much of the current literature on
freeze–thaw has been done on a one-dimensional open system with a constant
temperature applied to the top [5,11]. Artificial ground freezing for construc-
tion is performed by inserting a grid of freeze pipes vertically into the ground
and circulating a brine or anti-freeze mixture up and down the pipes. Freezing
in this manner causes zones of frozen soil to radiate outward from the pipes
until the frozen areas meet in the spaces between the pipes, forming a solid
block of soil and ice. Given the anisotropic nature of soils, the direction of
freezing has an influence on the magnitude of particle rearrangement, and thus
settlement, because of freeze–thaw.
For artificial ground freezing, two major laboratory models can be devel-
oped. The first is an open system frozen via a vertical linear isotherm in the
center of the sample to simulate near freeze-pipe conditions. The second is a
closed (no access to free water during freezing) system frozen from all exterior
directions to model the far freeze pipe conditions in the field once free water
has been cut off from accessing the remaining unfrozen soils.
Thawing and Post-Thaw Behavior

The behavior of cohesive compressible soils in consolidation testing after
freeze–thaw cycles has been less rigorously studied than the frost-heaving pro-
cess. The goal of the present study is to build on the work of researchers in
attempting to form an understanding of the settlement behavior of clays after
artificial freezing.
Previous research, evaluating the impact of one cycle of freeze–thaw on
Boston Blue Clay (BBC), found that BBC subjected to various cycles of
freeze–thaw exhibited significant disturbance, in the form of macroscopic
cracking and fissures compared to BBC in its natural state [13]. Their testing
also indicated a change in engineering properties related to consolidation and
shear strength behavior. Similar observations have been noted by other
researchers [14]. It has also been found that for freezing and thawing under an
open-system (i.e., water is available to the specimen during the cycle) the max-
imum volume changes were as high as 30 % under constant loading and that
there was a linear relationship between net volume change and liquidity index
prior to freezing [15]. It has also been found that all-round freezing resulted in
larger volume changes than one-dimensional freezing and slower rates of freez-
ing (0.8 mm/h) yielded higher strains than faster freezing rates (30 mm/h) [15].
Recent research has found a positive relationship between magnitude of
settlement of previously frozen soil and the duration of heating (thawing) it
experiences [16]. Specimens were frozen–thawed under constant loadings of
90 to 290 kPa. The results of this research indicate that “significant amount(s)
of structural damages” can occur during the thawing of clay-dominated soils.
Testing Procedures
The current study focuses on the effects of freeze–thaw on the consolidation of

Chicago Blue Clay (CBC). The CBC was obtained from a New England con-
sulting firm who provided “remnants” of 7.6-cm diameter, undisturbed Shelby-
tube samples used for an undisclosed testing program. Other than being
undisturbed samples, no additional information was made available to the
authors regarding CBC sampling characteristics, e.g., sampling location, depths
of sampling, procedures used in obtaining samples, etc. Whereas the lack of
detailed information is problematic, the value of using natural CBC specimens
was deemed advantageous to the research effort. Specimen of CBC were pre-
pared for consolidation testing with specimen trimmings used for index testing.
Two types of consolidation tests were performed on specimens of CBC:
(1) incremental loading from low (31.7 kPa) to moderate (2020 kPa) stresses,
closely following ASTM D2435-04 [17], and (2) long-term (48 h) constant
stress loadings on specimens. For each type of consolidation test, two speci-
mens were tested; one subjected to a freeze–thaw cycle and the other in its nat-
ural state. All consolidation tests were conducted on cylindrical specimens
trimmed into 6.4-cm in diameter by 1.9-cm high stainless steel consolidation
rings. For incremental loading tests on natural CBC, trimmed specimens were
placed into a computer-controlled consolidation testing apparatus (Load Trac
II unit from Geocomp, Inc.) and loaded to appropriate increments (each held
for 6 h). One unload–reload cycle was performed. The same loading sequence
was used for a specimen trimmed from the same tube but subjected to one
freeze–thaw cycle prior to testing.
Freezing was performed under a “closed-system,” i.e., freezing in all direc-
tions without access to water. This was done to replicate freezing conditions of
the soil between freeze pipes used in ground-freezing projects. Because these
soils would be surrounded by already frozen ground, their access to water flow
would be effectively cut off. The freeze–thaw cycle was accomplished by plac-
ing a trimmed specimen (i.e., CBC in a consolidation ring) into a freezer set to
14 C for approximately 24 h. The frozen specimen was then removed from
the freezer and allowed to thaw at room temperature for 24 h. The thawed spec-
imen was then subjected to consolidation testing.
For the constant stress consolidation testing, a trimmed specimen of natural
CBC was first incrementally consolidated (3-h increments) to an effective
stress of 505 kPa (no unload–reload cycle performed). The freeze–thaw speci-
men was removed from the consolidation device after the 505-kPa stress was
maintained for 3 h. The specimen was then placed into the freezer for 24 hs
under the same stress (maintained via dead weights). After 24 hs, the specimen
was removed from the freezer, immediately placed back into the consolidation
apparatus and stressed to 505 kPa, and then allowed to thaw with the stress
maintained for 24 h. Though temperatures of the specimen were not measured,
SWAN ET AL., doi:10.1520/STP156820130015 27
visual observation indicated that thaw occurred within the first 10 to 20 min af-
ter removal of the specimen from the freezer. The natural state (never frozen)
specimen was subjected to the final 505 kPa effective stress for 48 h.
In addition to the consolidation testing program, Atterberg limits (specifi-
cally the liquid limit and plastic limit) were obtained for the CBC. Test proce-
dures followed those outlined in ASTM D2216-05 [18] and ASTM D4318-10
[19]. The Atterberg limits were also determined after CBC subjected to one
freeze–thaw cycle.
Test Results
Consolidation Results
Figure 1 shows a plot of the void ratio versus effective stress for incrementally
loaded consolidation tests. The plotted results indicate a significant change in
CBC’s consolidation behavior has occurred after a cycle of freeze–thaw. Based
on the results as shown in Fig. 1, there is an average increase in deformation of
approximately 30 % of the freeze–thaw specimen from that of the natural CBC
for the range of stresses used in the tests. The smallest difference in deforma-
tion occurs at low stresses where the over-consolidated specimens exhibit very
similar consolidation behavior. However, at higher stresses, the differences
become more obvious with the maximum difference between the responses of
the specimens occurring at a stress of 1010 kPa where a % increase of 44 %
was obtained.
FIG. 1—Void ratio versus effective stress for Chicago Blue Clay natural state
and one freeze–thaw cycle.
TABLE 1—Summary of consolidation parameters from incremental consolidation tests on Chicago

Blue Clay: natural state and one freeze–thaw cycle.
One Freeze–Thaw Change From

Consolidation Parameter Natural State Cycle Natural State, %
Recompression index, Cr 0.083 0.088 þ5.5
Compression index, Cc 0.284 0.400 þ41.0
Max past pressure, r0 p (kPa) 610 270 56.3
Coefficient of consolidation, cv (cm2/min) 0.26 0.17 þ34.6
Consolidation parameters developed from the incremental tests are pre-

sented in Table 1. It is readily observed that the recompression index (Cr),
compression index (Cc), and the coefficients of consolidation (cv) in the nor-
mally consolidated range of stresses all increase for the freeze–thaw specimen.
In addition, the maximum past pressure (rp0 ), as determined by the Casagrande
method, is substantially reduced from that found for CBC in its natural state.
These results indicate that the freeze–thaw cycle causes significant changes in
the consolidation behavior to where deformations would be greater (higher Cc),
occur at a faster rate (higher cv), and become more significant at lower stresses
(lower rp0 ).
The significant change in the shape of the void-ratio–log effective stress
response also indicates that significant soil disturbance or remolding has
occurred to the CBC subjected to freeze–thaw. Figure 2 shows images the
CBC specimens after incremental testing was completed. These images, taken
along the sides of the specimen, illustrate that even after being consolidated to
stresses as high as 2020 kPa, a change in the soil’s macro-structure has
occurred because of the freeze–thaw cycle. Whereas the CBC consolidated in
FIG. 2—Photographs of specimens of Chicago Blue Clay subjected to incre-

mental consolidation tests—natural state and after one freeze–thaw cycle.
SWAN ET AL., doi:10.1520/STP156820130015 29
TABLE 2—Results of Atterberg limit tests on Chicago Blue Clay.
Liquid Limit, Plastic Limit, Plasticity Index,

LL (%) PL (%) PI ¼ LL-PL (%)
After a After a After a

Natural Freeze-Thaw Natural Freeze–Thaw Natural Freeze–Thaw
Soil State Cycle State Cycle State Cycle
Chicago blue clay 42.6 29.6 19.2 18.3 23.4 11.3
its natural state [Fig. 2(a)] exhibits a noticeably smoother face, indicative of a
relatively homogeneous soil; the frozen–thawed CBC, shown in Fig. 2(b),
exhibits vertical and horizontal cracks and fissures.
The constant-load consolidation tests also provide evidence that the consol-
idation behavior of CBC is impacted even under significant overburden
stresses. For tests where the maintained effective overburden pressure was
505 kPa, the measured deformation was 0.76 mm for the natural CBC but
0.96 mm for the CBC subjected to one freeze–thaw cycle. This is an increase
of approximately 25 %.
Atterberg Limit Results

Results of Atterberg limit tests of CBC in its natural state and after one
freeze–thaw cycle are presented in Tables 2.
Confirmatory of the consolidation results, the Atterburg limit tests show a sig-
nificant difference in the CBC’s liquid limit (and plasticity index) after the soil
undergoes one cycle of freeze–thaw. It is well established that the liquid limit is
correlated to a cohesive soil’s coefficient of consolidation, cv [20] where, the
higher a soil’s liquid limit, the lower its coefficient of consolidation. The decrease
in liquid limit of CBC subjected to a cycle of freeze–thaw could help explain, in
part, the increase in cv values of the CBC subjected to a freeze–thaw cycle.
Summary, Conclusions, and Recommendations for Further Study

Based on the results of the testing program conducted on Chicago Blue Clay, it
is readily apparent that the freezing and thawing of CBC leads to significant
changes in the consolidation behavior and index properties. Consolidation tests
show that one cycle of freeze–thaw Chicago Blue Clay led to a 25 % to 40 %
increase in deformation during consolidation when compared to the deforma-
tion of CBC in its natural state. The liquid limit of the CBC was found to
change significantly after a freeze–thaw cycle.
However, whereas the significance of the change in CBC behavior is read-
ily apparent in this study, there are shortcomings to the current work; for exam-
ple, the lack of temperature measurements during the freezing and thawing
process. There are a number of other alterations which should be considered to

verify or further substantiate the results presented. Possible modifications that
should be considered in future investigations include:
1. The freezing and thawing process used in this study was a closed system
with freezing occurring from all directions. If the freezing/thawing pro-
cess were altered, it could yield very different results. Therefore, several
freezing methods, such as 1D top-down freezing, pure radial freezing
with the top and bottom of the specimen insulated, as well as multidi-
mensional freezing, as used in this study, should be investigated.
2. A thorough visual investigation into the effects freezing and thawing
have on a clay’s structure and layering, including microscopic visual
records of clays during all stages of freezing and thawing, is an appro-
priate and vital test for further investigation. Visual investigations
coupled with variations in freezing rate and method can shed light on
which freezing regime is most destructive to clay structure.
3. In addition to freezing method, variable freezing rates should be consid-
ered, including considering a rate equivalent to an observed field rate of
2 mm/h [15], to rapid, 30 mm/h, more common in artificial freezing
applications.
4. Thermal monitoring during thawing and varying thawing rates during
consolidation could give insight into settlement values as a function
thawing rates.
5. Consolidation testing of one cycle, frozen–thawed clays other than CBC
could provide a basis for comparison of ground freezing feasibility in
different regions of the country underlain by various types of clays.
6. Testing that can freeze samples while simultaneously maintaining con-
solidation stresses could provide information about the relationship
between the swell caused by freezing soils and their subsequent collapse
during thawing under loading.
7. Changes in salinity of clay during freezing/thawing was not examined
in this study. Further testing should be run on clays from the same for-
mation that have different initial salinity values to test the hypothesis
that the salinity of clays is linked to its consolidation potential after
freezing and thawing.
References
[1] U.S. Federal Highway Administration, “Structural Assessment of 1-90

Connector Tunnels at South Station Boston, MA,” FHWA Assessment
Report, FHWA, Washington, D.C., 2011, 12 pp.
[2] Donohoe, J. F., Maishman, D., and Schmall, P., “The Freezing of Soil
Masses as an Aid to Engineering Construction,” ASCE GSP, Vol. 81,
1998, pp. 149–160.
SWAN ET AL., doi:10.1520/STP156820130015 31
[3] Chamberlain, E. J. and Gow, A. J., “Effect of Freezing and Thawing on

the Permeability and Structure of Soils,” Eng. Geol., Vol. 13(1), 1979, pp.
73–92.
[4] Williams, P. J., “Pore Pressures at a Penetrating Frost Line and Their Pre-
diction,” Geotechnique, Vol. 16(3), 1966, pp. 187–208.
[5] Chamberlain, E. J., “Overconsolidation Effects of Ground Freezing,”
Eng. Geol., Vol. 18(1), 1981, pp. 97–110.
[6] Rempel, A. W., “Formation of Ice Lenses and Frost Heave,” J. Geophys.
Res.: Earth Surface, Vol. 2003–2012, 2007, pp. 112(F2).
[7] Rempel, A. W., “Microscopic and Environmental Controls on the Spacing
and Thickness of Segregated Ice Lenses,” Quat. Res., Vol. 75(2), 2011,
pp. 316–324.
[8] Style, R. W., Cocks, A. C., Peppin, S. S., and Wettlaufer, J. S., “Ice-Lens
Formation and Connement-Induced Supercooling in Soils and Other
Colloidal Materials,” Report 11/38, Oxford Centre for Collaborative
Applied Mathematics (OCCAM) Mathematical Institute, Oxford, 2011,
33 pp.
[9] Peppin, S. S. L., Majumdar, A., and Wettlaufer, J. S., “Morphological
Instability of a Non-Equilibrium Ice–Colloid Interface,” Proc. Royal Soc.
A: Math. Phys., Vol. 466(2113), 2010, pp. 177–194.
[10] Konrad, J. M., “Influence of Freezing Mode on Frost Heave Characteristics,”
Cold Regions Sci. Technol., Vol. 15(2), 1988, pp. 161–175.
[11] Eigenbrod, K. D., “Effects of Cyclic Freezing and Thawing on Volume
Changes and Permeabilities of Soft Fine-Gained Soils,” Can. Geotech. J.,
Vol. 33(4), 1996, pp. 529–537.
[12] Konrad, J. M. and Morgenstern, N. R., “The Segregation Potential of a
Freezing Soil,” Can. Geotech. J., Vol. 18(4), 1981, pp. 482–491.
[13] Swan, C. and Greene, C., “Freeze-Thaw Effects on Boston Blue Clay,”
ASCE GSP, Vol. 81, 1998, pp. 161–176.
[14] Qi, J., Vermeer, P. A., and Cheng, G., “A Review of the Influence of
Freeze-Thaw Cycles on Soil Geotechnical Properties,” Permafrost Peri-
glacial Proc., Vol. 17, 2006, pp. 245–252.
[15] Eigenbrod, K. D., “Effects of Cyclic Freezing and Thawing on Volume
Changes and Permeabilities of Soft Fine-Grained Soils,” Can. Geotech.
J., Vol. 33, 1996, pp. 529–537.
[16] Dashjamts, D. and Altantsetseg, J., “Research on the Consolidation of
Frozen Soils Upon Thawing,” 6th International Forum on Strategic Tech-
nology (IFOST 2011), Vol. 2, Harbin, Heilongjiang, Aug 22–24, 2011,
pp. 1295–1300.
[17] ASTM D2435-04: Standard Test Methods for One-Dimensional Con-
solidation Properties of Soils Using Incremental Loading, Annual
Book of ASTM Standards, ASTM International, West Conshohocken,
PA, 2004.
[18] ASTM D2216-05: Standard Test Methods for Laboratory Determination

of Water (Moisture) Content of Soil and Rock by Mass, Annual Book of
[19] ASTM D4318-10: Standard Test Methods for Liquid Limit, Plastic Limit,
and Plasticity Index of Soils, Annual Book of ASTM Standards, ASTM
[20] Naval Facilities Engineering Command, Soil Mechanics Design Manual
7.01, 1986, 388 pp.
STP 1568, 2013
DOI:10.1520/STP156820120139
Gregory P. Makusa,1 Hans Mattsson,2 and Sven Knutsson2
Investigation of Increased Hydraulic

Conductivity of Silty Till Subjected
to Freeze–Thaw Cycles
REFERENCE: Makusa, Gregory P., Mattsson, Hans, and Knutsson, Sven,

“Investigation of Increased Hydraulic Conductivity of Silty Till Subjected
to Freeze–Thaw Cycles,” Mechanical Properties of Frozen Soils, STP 1568,
ABSTRACT: The hydraulic conductivity of silty till increases when the till is
subjected to freeze–thaw cycles. A dramatic increase normally occurs after
the first freeze–thaw cycle, and the magnitude generally depends on the ini-
tial or molding water content. Freezing of silty till causes aggregations of
clods and the formation of macrostructure. The initial or molding water con-
tent determines the number of freeze–thaw cycles required to complete the
agglomeration of clods and the formation of stable macrostructures, which in
turn controls the hydraulic conductivity of compacted specimens frozen and
thawed in the laboratory. The findings of this study show that for specimens
compacted wet of the optimum water content, a significant increase in the hy-
draulic conductivity was measured after the first freeze–thaw cycle. When
specimens were compacted at the optimum water content, a number of
freeze–thaw cycles were required in order to obtain the corresponding signifi-
cant increase in the hydraulic conductivity.
KEYWORDS: hydraulic conductivity, freeze–thaw, silty till, microstructure,
macrostructure, dry cover, clods
Manuscript received November 2, 2012; accepted for publication May 29, 2013; published
online August 22, 2013.
1
Dept. of Civil, Environmental and Natural Resources Engineering, Luleå Univ. of Technology,
SE 97187, Luleå, Sweden (Corresponding author), e-mail: gregory.makusa@ltu.se
2
Dept. of Civil, Environmental and Natural Resources Engineering, Luleå Univ. of Technology,
SE 97187, Luleå, Sweden.
3
Jacksonville, FL.
Copyright V
19428-2959.
33
Introduction
Till has been used in Sweden as a hydraulic barrier in landfills for many years.
Investigations of the hydraulic conductivity of silty till have shown that it is
sufficiently low to prevent the infiltration of water into the waste material. It
has also been reported that the hydraulic conductivity of silty till, like that of
cohesive soils, increases when the soil is subjected to a number of freeze–thaw
cycles [1–3]. Various researchers have postulated a number of reasons for this.
Among other causes, the formation of cracks due to ice lenses, changes in void
ratios, the initial or molding water content, and overburden stress have
been mentioned frequently. Studies have shown that the increased hydraulic
conductivity of compacted cohesive soils is caused by the formation of cracks
in frozen soils. However, Chamberlain and Gow [4] studied the hydraulic con-
ductivity of four fine-grained soils subjected to freeze–thaw cycling, and their
study reported increased hydraulic conductivity in clayey silt specimens that
did not exhibit the formation of cracks. According to Viklander [1], the initial
void ratio has a significant influence on the hydraulic conductivity change of
fine-grained till subjected to freeze–thaw. His results showed that initially
denser till soils exhibited increased hydraulic conductivity after the last cycle
of freeze–thaw, whereas initially loose soils showed decreased hydraulic con-
ductivity after the last cycle of freeze–thaw. These observations were similar to
the findings of Chamberlain and Gow [4] on the hydraulic conductivity of com-
pacted clay, but they are in contrast to those of Kim and Daniel [5], who
observed the largest increase in hydraulic conductivity in specimens of com-
pacted silty sand with a decrease in void ratio. These and other contentious
results from earlier research [1–9] suggest that more mechanisms contributing
to the increased hydraulic conductivity of frozen and thawed soils are involved.
Most of the test results from the investigations referred to above show a high
increase in hydraulic conductivity during permeation following the first
freeze–thaw cycle. No significant (or only small) variations in increased
hydraulic conductivity occur because of subsequent freeze–thaw cycles.
The present study sought to address the aggregation of clods due to freez-
ing and the ultimate formation of macrostructures, which in turn control the
hydraulic conductivity of thawed silty till. The study also investigated the
effect of laboratory sample preparation, surcharge load, and compaction on
the subsequent hydraulic conductivity of silty till subjected to freeze–thaw
cycles.
Materials and Methods

The soil used in this study was a Swedish till frequently used as barrier material
in landfills and earth dams [1–3]. The soil is to be utilized as dry cover for
wastes originating from mining activities at Boliden Mineral AB in the
MAKUSA ET AL., doi:10.1520/STP156820120139 35
Gällivare area in northern Sweden. The soil was characterized with regard to
grain size distribution by means of wet sieving. Sedimentation analysis via the
pipet method showed that the clay content was less than 5 %. A modified Proc-
tor test was carried out to establish the dry density–water content relationship.
The soil samples for the Proctor test were prepared in accordance with Swedish
laboratory test routines, in which representative soil samples are oven dried,
rewetted with the desired water content, and compacted. Figure 1 presents the
particle size gradation of a representative soil sample. The results of the char-
acterization were similar to previous results from other researchers [1–3], and
the studied soil was classified as silty till based on the Swedish soil classifica-
tion system or SM (silty sand) in accordance with the Unified Soil Classifica-
tion System.
Dry Density–Water Content Relationship

The investigated soil had a received water content between 10.43 % and
11.20 % with an average value of 10.80 % by mass, determined after oven dry-
ing portions of representative silty till for 24 h. Using the dried soil samples, a
series of specimens was compacted in a modified Proctor apparatus at molding
water contents of 2 %, 4 %, 6 %, 7 %, 8 %, 10 %, and 12 % by mass. The dry
density–water content relationship curve was plotted. A maximum dry density
(MDD) of 2.14 g/cm3 was obtained at the optimum water content (OWC) of
6.70 %, as shown in Fig. 2. It follows from the Proctor curve (Fig. 2) that the
FIG. 1—Particle size distribution for studied silty till.

FIG. 2—Dry density–water content relationship.
average water content of 10.80 % corresponds to a dry density of 2.03 g/cm3,

which was equivalent to a 95 % degree of compaction.
Sample Preparation for Hydraulic Conductivity Test

A representative soil was processed to form two portions for the hydraulic con-
ductivity test. Based on the methods of sample preparation, the laboratory
experiment setup was divided into two categories, which involved the use of
specimens from either soil sample A or soil sample B.
Sample A was prepared by oven drying a representative sample for 24 h.
The oven-dried sample was then sieved to remove all particles larger than
20 mm prior to being mixed with the desired molding water contents. Two
specimens, namely, specimens A1 and A2, were prepared from this soil
sample. To mimic the received soil water condition, the soil sample for speci-
men A1 was molded with a water content of 10.80 % by mass and compacted
into the permeameter. The soil sample for specimen A2 was molded at OWC
and compacted into the permeameter, resulting in MDD in the Proctor curve.
To minimize the effect of processing the soil sample and preserve the natu-
ral soil suction capacity, another portion of representative silty till was sieved
to remove soil particles larger than 20 mm without oven drying for sample B.
Three specimens, namely, specimen B1, specimen B2, and specimen B3, were
prepared at the received water condition for the hydraulic conductivity test.
Hydraulic Conductivity Test

In order to minimize double handling processes, five Proctor molds were man-
ufactured. Each specimen was compacted in a rigid-wall Proctor mold
101.6 mm in diameter and 112.0 mm in height. Taking into account the initial
and molding water contents, all specimens were compacted to a 95 % degree

of compaction except specimen A2, which was compacted to a 100 % degree
of compaction. A filter was placed at the top and bottom of each specimen. The
specimen molds were covered with top and bottom rigid plastic caps. All speci-
mens were then set for the initial hydraulic conductivity prior to freeze–thaw
cycling using a constant pressure head at a hydraulic gradient of 18 in accord-
ance with ASTM D5856 [10]. The outflow was collected using upward flow,
and the specimens were protected from upward movement by clamps. In order
to simulate field conditions, in which the underlying material would not pro-
duce capillary action, no backpressure was applied to saturate the specimens.
The outflow could be obtained within an hour; nevertheless, the test readings
were taken every three hours during daytime, and a minimum of eight readings
was considered sufficient. The average of the last four steady readings (within
625 %) was reported as the hydraulic conductivity of a particular specimen.
After the initial hydraulic conductivities of all specimens had been established,
the top caps were removed and the molds were moved into a freezing cabinet.
One-dimensional freezing in a closed system was desired. Therefore, the
perimeters of the permeameter were covered with 35 mm of insulation foam.
The bottom caps were placed in a Styrofoam insulator placed at the base of the
freezing cabinet (Fig. 3). No backpressure was applied. Thermocouple instru-
mentation data showed that a duration time of 24 h was sufficient to freeze the
specimens to 15 C and then thaw them to room temperature (Fig. 4). During
freezing and thawing, specimen B2 and specimen B3 were loaded with 10 kPa
and 25 kPa, respectively. Specimens B1, A1, and A2 were all frozen and
FIG. 3—Schematic sketch of experimental device for one-dimensional freezing

and thawing tests.
FIG. 4—Progress of temperature within the silty till specimen from room tem-
perature in a 15 C freezer.
thawed without an overburden stress. The hydraulic conductivity results

reported here are the average values of at least the last four steady readings.
Results
A summary of the test results is presented in Table 1. The results show that
specimens from soil sample B compacted without oven drying the soil had low
initial hydraulic conductivity relative to specimens from soil sample A com-
pacted after the represented soil sample had been oven dried and rewetted to
the desired water content. Thus, the average initial hydraulic conductivity
value of 6.6 10 9 m/s of specimens from soil sample B was considered as the
baseline hydraulic conductivity Kb of the studied silty till. It followed that all
measured hydraulic conductivity values K were normalized over this baseline
TABLE 1—Results of hydraulic conductivity tests.
Soil Sample A Soil Sample B
Freeze–Thaw Specimen A1 Specimen A2 Specimen B1 Specimen B2 Specimen B3

Cycles (0 kPa), K, m/s (0 kPa), K, m/s (0 kPa), K, m/s (10 kPa), K, m/s (25 kPa), K, m/s
8 8 9 9 9
0 2.0 10 1.1 10 7.0 10 6.5 10 6.3 10
7 8 7 8 8
1 1.6 10 1.8 10 1.4 10 5.7 10 5.3 10
7 8 8 8 8
2 1.3 10 1.6 10 7.5 10 5.8 10 4.6 10
8 8 7 8 8
3 8.2 10 1.8 10 1.5 10 5.6 10 4.4 10
8 8 8 8 8
4 9.0 10 2.0 10 8.7 10 5.9 10 4.4 10
8 8 8 8 8
5 7.0 10 2.2 10 6.6 10 5.9 10 4.3 10
8 8 8
6 6.8 10 — — 6.2 10 4.6 10
TABLE 2—Normalized hydraulic conductivity test results (K/Kb) for all specimens.
Soil Sample A Soil Sample B
Freeze–Thaw Specimen A1 Specimen A2 Specimen B1 Specimen B2 Specimen B3

Cycles (0 kPa) (0 kPa) (0 kPa) (10 kPa) (25 kPa)
0 3.0 1.7 1.1 1.0 1.0
1 24.2 2.7 21.2 8.6 8.0
2 19.7 2.4 11.4 8.8 7.0
3 12.4 2.7 22.7 8.5 6.7
4 13.6 3.0 13.2 8.9 6.7
5 10.6 3.3 10.0 8.9 6.5
6 10.3 — — 9.4 7.0
hydraulic conductivity. Table 2 presents the normalized hydraulic conductivity

values. As can be seen, the hydraulic conductivity increased dramatically after
the first freeze–thaw cycle. The increase was in the range of 2.7 to 24.0 times
the baseline value. Specimen B1 and specimen A2 underwent five cycles only,
which was considered sufficient. Svensson [11] carried out a parallel test for
his Master’s thesis using similar testing conditions with only oven-dried soil
samples and obtained comparable results, except for the initial hydraulic
conductivity.
Effect of Sample Preparation

Figure 5 presents the initial hydraulic conductivities of all tested specimens.
Specimens from soil sample B showed low and consistent readings for three
days of permeation, with an average initial hydraulic conductivity value of
6.6 10 9 m/s. Specimens from soil sample A showed higher initial hydraulic
FIG. 5—Initial hydraulic conductivity of specimens from soil samples A and B.

FIG. 6—Normalized hydraulic conductivity of specimens versus number of

freeze–thaw cycles.
conductivity. Specimen A1 compacted wet of the OWC had an average initial

hydraulic conductivity of 2.0 10 8 m/s, which was about three times the
baseline hydraulic conductivity. Specimen A2 had an average initial hydraulic
conductivity of 1.1 10 8 m/s. This value is about 1.7 times higher than the
baseline hydraulic conductivity.
Effect of Freeze–Thaw and Surcharge Load

After being subjected to the first freeze–thaw cycles (Fig. 6), all specimens
exhibited a significant increase in hydraulic conductivity. The hydraulic con-
ductivity of specimen B2 loaded with 10 kPa initially increased by about nine
times the baseline hydraulic conductivity and remained almost the same up to
the sixth freeze–thaw cycle. Similarly, the hydraulic conductivity of specimen
B3 loaded with 25 kPa initially increased by about eight times the baseline
hydraulic conductivity. This initially increased hydraulic conductivity of speci-
men B3 decreased to about seven times the baseline hydraulic conductivity
after the second freeze–thaw cycle, but it remained almost the same for the
subsequent freeze–thaw cycles.
Effect of Freeze–Thaw and Compaction

After the first freeze–thaw cycle, the hydraulic conductivity of specimens of
soil sample A compacted at desired water contents increased (Fig. 6). The
initial increases in hydraulic conductivity for specimen A1 and specimen A2
were 24 and 3 times the baseline hydraulic conductivity, respectively. The
hydraulic conductivity of specimen A1 decreased to about 11 times the
baseline hydraulic conductivity at the end of fifth freeze–thaw cycle. This

increase and then decrease in hydraulic conductivity for specimen A1 was
almost similar to that of specimen B1, which was compacted at the received
water content and to the same degree of compaction as specimen A1. In con-
trast, the hydraulic conductivity of specimen A2 increased by about three
times the baseline hydraulic conductivity at the end of the fifth freeze–thaw
cycle.
Discussion
According to Vapalli et al. [12], fine-grained soils have two levels of struc-
ture, microstructure and macrostructure, both of which are present in natural
and compacted clay soils. Regardless of the mineralogy, texture, and method
of preparation, the resulting macrostructure of specimens prepared at differ-
ent initial water contents will be different. The structure and aggregation of
fine-grained soil is highly influenced by the initial or molding water content.
Clay soils with a high initial water content are more homogeneous, with the
large pore spaces not interconnected or in a closed state. These specimens
offer more resistance to water flow. Consequently, the microstructure of
specimens compacted wet of the OWC controls the water flow. In contrast,
specimens with a low initial water content (i.e., dry of the OWC) contain rel-
atively large pore spaces located between clods of soil. Thus, the macrostruc-
ture controls the initial discharge. Benson and Daniel [13] studied the
influence of clods on hydraulic conductivity of unfrozen highly plastic clay.
Their results show that the outcome of clod and inter-clod pores occurring
during soil processing and compaction controlled the hydraulic conductivity
of the compacted soil. The clod size had a large influence upon the hydraulic
conductivity of the compacted soils. Samples of soil with initially small clods
had lower hydraulic conductivity than samples compacted from material with
initially large clods.
The term “microstructure” herein refers to soil specimens that are domi-
nated by micro pores, whereas “macrostructure” refers to soil specimens in
which the macro pore is dominant. “Clod” refers to the agglomeration of fine
particles into small soil lumps within the specimen.
The differences in the initial hydraulic conductivity observed between
specimens of the two soil samples A and B were due to different aggregations
and soil-water affinities, which were caused by variations in initial water con-
tent and sample preparation. This resulted in the formation of clods with differ-
ent physical characteristics. Specimens compacted from soil sample B had a
higher initial water content. This allowed the saturation of natural clods during
compaction, which flowed to fill the pore spaces. In additional to the flow of
saturated clods, free water became available to fill the remaining pore spaces,
FIG. 7—Enlarged schematic sketch of micro- or macrostructure levels of natu-

ral silty till compacted at different water contents.
thus reducing the size of open inter-grained pores and inter-clod pores. Exam-
ples of the physical characteristics of specimen B are shown in Fig. 7 and
Fig. 8(b); the resulting microstructure controls the hydraulic conductivity. Sat-
urated specimens tend to have low soil suction. Specimen A1, which was
molded and compacted at the received water content, indicated no degree of
saturation during compaction, which means that the molding water content was
not sufficient to mold and saturate the whole specimen. It can be postulated
FIG. 8—Scanning electron photomicrograph with corresponding photographs

of soil compacted with water contents of (a) 12 % and (b) 20 % showing the
physical characteristics of compacted clods (from Ref 13, with permission from
ASCE).
that the molding water was utilized in building up clods. According to Sharma
[14], clods are formed via two mechanisms: coating deflocculated clay paste,
and soil compaction. Consequently, during molding and compaction, stiff clods
were formed with bridges between them, leaving interconnected large pores as
shown in Fig. 7 (specimen A1) and Fig. 8(a), which provided a means for
hydraulic paths. For that reason, macro pores became dominant and the macro-
structure controlled the flow at high soil suction, which resulted in high initial
hydraulic conductivity relative to those of specimen B. According to clod
theory [13], soft, wet clods of soil are easier to remold than hard, dry clods. For
specimens compacted at the optimum molding water content, the fine particles
became sufficiently smooth (lubricated and friable), allowing them to slide past
each other easily without sticking to each other. This allowed the coarse par-
ticles to penetrate, and the result was a uniform distribution of fine particles
between coarse particles, as shown in Fig. 7 (specimen A2). Thus, the sizes of
pore spaces were minimized and the hydraulic conductivity was controlled by
the microstructure. According to Konrad [15], for well-graded Saint-Martyrs-
Canadians till, the lowest fines (clay and silt) content of 12 % is sufficient to
fill the voids between the sand and gravel particles.
Although silty till is classified as a non-cohesive soil [1,3], the water con-
tents within the specimens gave it the ability to expand upon freezing. Accord-
ing to De Bruyn et al. [16], in order for the soil to expand when it freezes, it
must be able to bind some amount of water that is not drained away by the
force of gravity. Under the closed system, the only water that could freeze was
the capillary and bound water and some remaining water in the hydraulic paths.
During freezing, soil suction increased to the point of (or beyond) the air entry
suction value (in the soil–water characteristic curve). At this point or beyond, it
becomes difficult to remove water from soil particles because of high soil-
water affinity [12]. Thus, the negative pore pressures generated between the
bound water and soil particles created inter-particle bonds resulting in the for-
mation of stable clods. The formed clods bridged one another, creating inter-
clod pores (large pores), which became the ultimate hydraulic path at the time
of permeation. Throughout the repeated process of desaturation (freezing), the
aggregations of fines into strings and bulky clods continued, and during satura-
tion (thawing and permeation) the inter-clod pores controlled the flow of water.
According to Konrad [17], the relocation of particles occurs during freezing,
and after thawing the clodded fine particles do not move back to their initial
positions; this leads to permanent changes in the pore-size distribution
characteristic.
Depending on the initial or molding water content and degree of saturation,
the agglomeration of clods and ultimate formation of a stable macrostructure
can occur either in the first freeze–thaw cycle or after subsequent freeze–thaw
cycles. It follows that fine particles in specimens from soil sample B and speci-
men A1, which were compacted wet of optimum, aggregated into stable clods,
turning the whole specimen into macrostructure during the first freezing. Dur-
ing the first freezing, all possible clods were formed in specimens B1, B2, and
B3, and after thawing and permeation the inter-clod pores became full of
free water and the whole specimen became saturated again. Thus, during
subsequent permeation, no extra-unsaturated pores were available to increase
the hydraulic conductivity. The macrostructure controlled the flow at low soil
suction, resulting in decreased or maintained increased hydraulic conductivity.
Specimen A1, in contrast, had to undergo about five cycles before the inter-
clod pores became saturated, after which the ultimate macrostructure con-
trolled the hydraulic conductivity at a magnitude of about 10 times the baseline
hydraulic conductivity. The continued increase in the hydraulic conductivity
exhibited by specimen A2 suggested that the aggregation of fine particles into
clods and the formation of macrostructures continued after five freeze–thaw
cycles. This is because once the hydraulic paths have been defined during the
initial permeation, it might take a long time to saturate soil particles located far
away from these predefined hydraulic paths.
The surcharge loads reduced or maintained the size of inter-clod pores.
Accordingly, specimens B2 and B3 had low increases in hydraulic conductivity
relative to specimen B1. According to Benson and Daniel [13], in order to
achieve low hydraulic conductivity in soils that form clods, large inter-clod
pores must be eliminated during compaction. However, freezing action contrib-
utes to the formation of new clods, as discussed earlier, and sufficient overbur-
den stress might be required in order to prevent the formation of these clods or
completely break the clods and close the inter-clod pores.
Conclusion
The increase in the hydraulic conductivity of silty till due to freeze–thaw
actions was investigated. We considered the effects of laboratory sample prep-
aration, surcharge load, degree of compaction, and initial or molding water
content on the hydraulic conductivity of silty till samples that were frozen and
thawed in the laboratory. The following conclusions are drawn:
• Freezing action on silty till results in agglomerations of clods and the
formation of macrostructures, which in turn increase the hydraulic con-

ductivity of thawed silty till.
• For silty till compacted wet of the OWC, the hydraulic conductivity
after the first freeze–thaw cycle was increased by 8 to 24 times the

hydraulic conductivity of unprocessed silty till. All the significant
increases in hydraulic conductivity were observed after the first
freeze–thaw cycle.
• For silty till compacted at OWC and subjected to laboratory freeze–thaw
action, several freeze–thaw cycles might be required in order to obtain

the ultimate increase in the hydraulic conductivity.
Acknowledgments
The writers acknowledge Mr. Seth Mueller (Boliden Mineral AB) for provid-
ing materials for investigation of the effect of freeze–thaw on hydraulic con-
ductivity of silty till. Special thanks are due to Mr. Thomas Forsberg, Ulf
Stenamn (Soil Laboratory technician), and Jimmy Svensson (Master’s student)
for their daily support and skilled laboratory work.
References
[1] Viklander, P., “Permeability and Volume Changes in Till Due to Cyclic
Freeze-Thaw,” Can. Geotech. J., Vol. 35, 1998, pp. 471–477.
[2] Viklander, P., 1997, “Compaction and Thaw Deformation of Frozen Soil:
Permeability and Structural Effects Due to Freezing and Thawing,” Ph.D.
thesis, Luleå University of Technology, Luleå, Sweden.
[3] Carlsson, E., 2002, “Sulphide-rich Tailings Remediated by Soil Cover:
Evaluation of Cover Efficiency and Tailings Geochemistry, Kristineberg,
Northern Sweden,” Ph.D. thesis, Luleå University of Technology, Luleå,
Sweden.
[4] Chamberlain, E. and Gow, A., “Effect of Freezing and Thawing on the
Permeability and Structure of Soils,” J. Eng. Geol., Vol. 13, Nos. 1–4,
1979, pp. 73–92.
[5] Kim, W. and Daniel, D., “Effects of Freezing on Hydraulic Conductivity
of Compacted Clay,” J. Geotech. Engrg., Vol. 118, No. 7, 1992, pp.
1083–1090.
[6] Benson, C. and Othman, M., “Hydraulic Conductivity of Compacted Clay
Frozen and Thawed in-situ,” J. Geotech. Engrg., Vol. 119, No. 2, 1993,
pp. 276–294.
[7] Chamberlain, E. J., Erickson, E. A., and Benson, H. C., “Frost
Resistance of Cover and Liner Materials for Landfills and Hazardous
Waste Sites,” Special Report 97-29, Corps of Engineers, Department of
the Army, Cold Regions Research and Engineering Laboratory, Hanover,
NH, 1997.
[8] Kraus, J., Benson, C., Erickson, A., and Chamberlain, E., “Freeze-Thaw
and Hydraulic Conductivity of Bentonitic Barriers,” J. Geotech. Geoen-
viron. Eng., Vol. 123, No. 3, 1997, pp. 229–238.
[9] Benson, H. C., Abichou, H. T., Olson. A. M., and Bosscher, P. J., “Winter
Effect on Hydraulic Conductivity of Compacted Clay,” J. Geotech.
Engrg., Vol. 121, 1995, pp. 69–79.
[10] ASTM D5856-95: Standard Test Method for Measurement of Hydraulic
Conductivity of Porous Material Using a Rigid-Wall, Compaction-Mold
Permeater, Annual Book of ASTM Standards, ASTM International, West
[11] Svensson, J., 2012, “Hydraulisk konduktivitet i en morän: inverkan

av frys- och tiningscykler vid olika överlaster och packningsgrader
[Hydraulic Conductivity of Till: Influence of Freeze/Thaw Cycles at
Different Overburden and Degree of Compaction],” M.S. thesis, Luleå
University of Technology, Luleå, Sweden (in Swedish).
[12] Vapalli, S. K., Fredlund, D. G., and Pufhal, D. E., “The Influence of Soil
Structure and Stress History on the Soil-Water Characteristics of a Com-
pacted Till,” Geotechnique, Vol. 49, 1999, pp. 143–159.
[13] Benson, C. H. and Daniel, D. E., “Influence of Clods on Hydraulic Con-
ductivity of Compacted Clay,” J. Geotech. Engrg., Vol. 116, 1990, pp.
1231–1248.
[14] Sharma, K. P., “Clod Formation—A Serious Problem for Wheat Cultiva-
tion in Rice Wheat Cropping Sequence,” Journal of Agricultural Physics,
Vol. 1, 2001, pp. 76–79.
[15] Konrad, J. M., “Frost Susceptibility Related to Soil Index Properties,”
Can. Geotech. J., Vol. 36, No. 3, 1999, pp. 403–417.
[16] De Bruyn, C. M. A., Collins, L. E., and Williams, A. A., “The Specific
Surface, Water Affinity, and Potential Expansiveness of Clays,” J. Clay
Miner., Vol. 3, No. 17, 1957, pp. 120–128.
[17] Konrad, J. M., “Physical Processes During Freeze-Thaw Cycles in Clayey
Silts,” Cold Reg. Sci. Technol., Vol. 16, 1989, pp. 291–303.
TESTING OF MECHANICAL
PROPERTIES
STP 1568, 2013
DOI:10.1520/STP156820130003
Benjamin Still,1 Zhaohui (Joey) Yang,1 and Xiaoxuan Ge1
Sampling, Machining, and Testing of Naturally

Frozen Soils
REFERENCE: Still, Benjamin, Yang, Zhaohui (Joey), and Ge, Xiaoxuan,

“Sampling, Machining, and Testing of Naturally Frozen Soils,” Mechanical
Properties of Frozen Soils, STP 1568, Hannele Zubeck and Zhaohui Yang,
Eds., pp. 49–61, doi:10.1520/STP156820130003, ASTM International, West
Conshohocken, PA 2013.2
ABSTRACT: For designing infrastructure foundations in cold regions, it is
essential to evaluate the mechanical properties and characteristics of natu-
rally frozen soils obtained from the field. However, sampling frozen soils is
a difficult task especially when looking for minimally disturbed naturally fro-
zen soil samples. The less disturbed the samples, the more reliably one
can evaluate the mechanical properties of the frozen soils at its in situ con-
dition. This paper describes the processes used to extract and machine
naturally frozen soil samples for mechanical property testing. The extrac-
tion process involves a chainsaw with a carbide chain to cut a square into
the ground of the desired size and the isolated block is snapped from the
ground by using a wedge. The machining process involves cutting the block
sample into octagons. During this process, particular attention is paid to
the orientation of soil fabric. Once an octagon is made, a lathe is used with
a special holding bit to avoid disturbing the octagon ends. Carbide cutting
tools are needed as standard cutters dull very quickly and heat begins
affecting the soil. The rate of lathing is carefully controlled to minimize the
heat disturbance and roughness of the cylinder surface. A four-jaw chuck is
used to face both sides of the specimen with the lathe to ensure that the
two faces are parallel. This machining process is believed to minimize the
thermal and mechanical disturbance to the specimen. The testing equip-
ment and sensors used to measure stress–strain behavior are described.
Manuscript received January 3, 2013; accepted for publication April 16, 2013; published online
August 19, 2013.
1
School of Engineering, Univ. of Alaska Anchorage, Anchorage, Alaska 99501, United States of
America.
2
Jacksonville, FL.
Copyright V
19428-2959.
49
Testing data obtained from four samples with different orientation angle are
presented.
KEYWORDS: naturally frozen soil, sampling, specimen machining, specimen
orientation
Introduction
The quality of the naturally frozen soil sample depends on the type of frozen
soil sampled, the in situ thermal condition at the time of sampling, the sam-
pling method, the transportation and storage procedures, and the specimen
machining procedure prior to testing. Methods for sampling, transportation,
storage, and machining of frozen soils were proposed by Baker [1]. Baker [1]
investigated the end effects during uniaxial compression tests using different
platens. Ebel [2] discussed the aspect ratio of frozen soil cylinders and end
effects during uniaxial compression testing. De Re et al. [3] proposed triaxial
testing methods and equipment for constant strain rate control for testing fro-
zen soils.
Although many research programs have been conducted over the years, not
many have tackled naturally frozen soils with high strain rates during compres-
sion tests, which present many difficulties because of the specimen’s non-
uniformity. This paper refines methods and tools to minimize the evaporation,
sublimation, and thermal disturbance of naturally frozen samples. Some test
data is provided to validate the effort to minimize the thermal disturbance of
the samples.
Site Preparation and Instrumentation

Two sites are chosen to take frozen soil samples: CRREL Permafrost Tunnel at
Fox, AK and the North Fork of Campbell Creek Bridge in Anchorage, AK,
with the former representing permafrost and the latter seasonally frozen soils.
At the Campbell Creek Bridge site, a steel pipe was drilled into the ground to
help monitor the temperature and depth of the seasonally frozen soils. This will
help to determine the sampling depth and natural temperature of the frozen soil
during sampling. Based on the observation of ground temperature from this
study and a previous study by Li et al. [4] at this site, the seasonally frozen soil
can reach a depth of up to 5 feet and its temperature ranges from 1 C to
10 C.
Sampling
Frozen soil is very stiff and a carbide-chain chainsaw was used to cut out
blocks of frozen soil approximately 1 ft3 as seen in Fig. 1. Where electricity is
readily available or where exhaust is a concern (permafrost tunnel), an electric
chainsaw is used, but at more remote locations including bridge sites a gas
STILL ET AL., doi:10.1520/STP156820130003 51
FIG. 1—Sampling at Campbell Creek Bridge Site (left) and CRREL Permafrost
Tunnel (right).
powered chainsaw is needed. The chainsaws worked well in frozen silt. How-
ever, when small pieces of gravel were encountered in gravelly soils, the blade
was dulled very quickly resulting in increased cutting time and melting the fro-
zen soil. During the extraction of soil blocks, a square was cut in the ground
with the chainsaw and a wedge was pounded into one side of the chainsaw cut,
which snapped the bottom of the block. Typically, the bottom broke irregularly
and could be trimmed in a cold room at a later time. Pulling the block out of
the soil can be difficult, and we typically tried to angle the blade of the chain-
saw slightly to make the block smaller at the base. Nail pullers worked well for
pulling the block out of the soil. The orientation of the frozen soil block is
important, and we noted which side was the surface and whether the surface
had any inclination from the horizontal. Sampling using this method does not
work for obtaining samples at depths much deeper than 2 m. Drilling cores
will be necessary to obtain samples at greater depths. ASTM D7015-07 [5] was
followed for block sampling.
Transportation
After the frozen soil blocks were extracted from the ground, they were wrapped
in cellophane and put in a polyurethane bag with some snow to help minimize
moisture loss. Air was extruded from the bag before it was sealed to help pre-
vent sublimation. The samples were placed in a portable freezer run by a car
battery to minimize thermal disturbance while being transported to the cold
room for storage.
Because of the distance of the CRREL Permafrost Tunnel from the Univer-
sity of Alaska Anchorage Cold Room, we took more samples than the portable
freezer was able to carry. The rest of the samples were placed in a rocketbox
on top of a vehicle for transportation to Anchorage. The temperatures were

warm for the time of year but well below 0 C.
Storage
Frozen block samples were placed in a cold room in their polyurethane bag.
The bags were inspected for holes and places of wear during transportation. If
any bag had been compromised, the sample was resealed in another polyur-
ethane bag. It is important to keep the temperature of the block samples as
close to the in situ temperature as possible during storage. Temperatures in a
typical cold room can fluctuate substantially and insulating the samples can
help reduce this fluctuation.
Machining
ASTM D7300-06 [6] describes a standard test method for lab determination of
strength properties of frozen soils at a constant strain rate. It recommends using
the unconfined compression test by using a cylindrical frozen soil specimen of
certain dimension. Experience indicates that consistent creep and strength
results can be obtained when the height-to-diameter ratio is at least three to
two [2]. So we decided to use specimens 101.6 mm (4 in.) height by 50.8 mm
(2 in.) diameter in this study.
Frozen soil blocks can be large and irregular. A band saw in a cold room is
a great way to cut through these blocks. Good metal cutting blades can be used
to cut through frozen silt, sand, and even the occasional piece of gravel
although this dulls the cutting surface substantially. The outer inch or two of
the block is generally considered to be heat affected and trimmed off the block.
We trimmed the blocks into vertically and horizontally oriented rectangular
prisms of 76.2 mm 76.2 mm 114.3 mm (3 in. 3 in. 4.5 in.). To save time
on lathing, we also trimmed the long corners making octagon-shaped prisms
with a wooden jig as seen in Fig. 2.
After the specimens were made into octagons they were lathed into cylin-
ders using a special holding bit. Using a lathe in a cold room for machining fro-
zen soils presents some special problems. Most lathes are designed to work in
room temperature near 20 C, but the cold room is well below freezing, so the
gears and moving parts of the lathe need to be cleaned and cold weather grease
needs to be applied instead of the factory grease. The shavings of the frozen
soil specimen also work their way into the gears and moving parts creating
unnecessary wear. Using a shop vacuum during the machining of the frozen
soil specimens helps to keep the soil particles out of the gears.
A special holding bit to turn the frozen soil specimen is needed as seen in
Fig. 3. This circular wooden holding bit uses low grit sandpaper to hold the
specimens in place. Ideally, a carbide tip cutter should be used for the lathe.
Standard cutters dull very quickly and heat begins affecting the soil quickly.
FIG. 2—A frozen soil sample being cut into an octagon with the wooden jig by
the band saw.
During the initial lathing, as much as 2.54 mm (100/1000 in.) was trimmed at a
time until the specimen was within 6.35 mm (0.25 in.) of the required diameter.
During final trimming, no more than 0.381 mm (15/1000 in.) was trimmed at a
time typically being reduced to 0.127 mm (5/1000 in.) during the final pass. Af-
ter the specimen was turned to the 50.8 mm (2 in.) diameter, we used the four-
jawed chuck, a standard holding bit, to then face both ends using a framing
square to make sure both sides are parallel and perpendicular to the cylinder
surface. Facing the frozen soil specimen is very important and a sharp cutter is
essential to perform this task. While facing we took no more than 0.381 mm
(15/1000 in.), and the final pass no more than 0.127 mm (5/1000 in.) is taken
off the face. The face is inspected for any defects or inconsistencies before the
specimen is wrapped in cellophane, labeled, and put into a polyurethane bag to
FIG. 3—A frozen soil specimen in the special holding bit on the lathe.
prevent sublimation for storage. Testing is typically done within 1 week of

specimen completion. Samples extracted from cores can use this same smooth-
ing procedure with the lathe to remove the disturbed portion.
Conditioning and Classification of Soils

Frozen soil mechanical properties are very sensitive to temperature, in addition
to other factors. We tested the specimens at a range of temperatures for this
study. A small adjustable freezer was used to condition the specimens to a spe-
cific temperature. Thermocouples were placed on the specimens and then the
specimens were insulated fully and placed in the freezer for a minimum of
24 h. The rest of the freezer was filled with ice in bags to help minimize any
temperature variation. As shown in Fig. 4, the variation of temperature during
the final stage of conditioning is less than 0.1 C.
Before compression testing, several more steps are necessary. Photos are
taken of the different aspects of the frozen soil cylinder including both ends.
Notes are taken about the visual classification of the specimen including loca-
tion, size, and distribution of ice and larger grains of soil. The specimen is
given a frozen soil classification according to ASTM D4083-07 [7]. All sam-
ples contained visible ice lenses and were classified as V with the appropriate
ice descriptor with very few exceptions. Both the length and diameter of the
specimen are measured. The sample is transported to the UTM-100 Machine in
a cooler conditioned to 10 C.
Several sieve analysis and hydrometer tests were conducted for both the
permafrost and seasonally frozen soils. The grain size distribution can be
seen in Fig. 5 for the permafrost and Fig. 6 for the seasonally frozen soil.
Both soils contain large amounts of fines and are silts. The permafrost soil is
FIG. 4—Conditioning curves of two frozen soil specimens.

FIG. 5—Grain size distribution of permafrost.
classified as silt, whereas the seasonally frozen soil is classified as sandy or-
ganic silt with several specimens being classified as peat from their highly
organic nature.
Specimen Orientation
During seismic loading, a pile with its supported superstructure swings back
and forth; hence, the main loading direction is horizontal. Traditionally, soil
specimens, particular bored soil samples, are obtained from a borehole perpen-
dicular to the ground surface. To investigate whether the orientation of the soil
specimen in reference to the ground surface has any impact on the soil mechan-
ical properties, both vertical and horizontal specimens are prepared and tested.
As illustrated in Fig. 7, a vertical specimen (identified by V) indicates that its
FIG. 6—Grain size distribution of seasonally frozen soil.

FIG. 7—Frozen soil specimen orientation.
axis is perpendicular to the ground surface and a horizontal specimen (identi-

fied by H) indicates that its axis is parallel to the ground surface. Assuming
uniform soil properties in the same block, at least one vertical and one horizon-
tal specimen were machined from the center portion of the same block for com-
parison. All specimens are labeled by their sampling site and orientation. For
example, C2H1 indicates #1 horizontal specimen machined from block #2
taken from the Campbell Creek Bridge site (seasonally frozen soil); similarly,
P2V1 indicates #1 vertical specimen machined from block #2 taken from the
Permafrost Tunnel (permafrost).
Testing Apparatus and Instrumentation

The unconfined compression test was completed by the Universal Testing
Machine (UTM-100). The UTM-100 has a temperature chamber that can main-
tain the temperature as cold as 17 C. A fan was installed in the chamber to
improve air circulation, help maintain a more uniform temperature environ-
ment, and minimize thermal disturbance of the frozen soil specimen.
It is recommended that lubricated platens be used whenever possible in the
uniaxial compression and creep testing of frozen soils [2]. The lubricated
platen consists of a circular sheet of 0.8-mm-thick latex membrane, attached to
the loading face of a steel platen with a 0.5-mm-thick layer of high-vacuum sil-
icon grease. The steel platens are polished stainless steel disks about 10 mm
larger than the specimen diameter [6].
Displacement control was used during the compression test at a strain rate
of 0.1 %/s. As the latex sheets and grease layers compress under load, the axial
strain of the specimen should be measured using an extensometer installed
near the center of the specimen. The extensometer setup is shown in Fig. 8. A
load cell was used to measure axial load and, hence, stress on the specimen.
FIG. 8—A frozen soil specimen during loading with an extensometer attached
on the left.
The extensometer and load cell are calibrated before use in these experiments.
The overall testing setup of the UTM-100 machine can be seen in Fig. 9.
Observations and Specimen Modes of Failure

Three failure modes were observed during testing including bending, bulging,
and shearing as can be seen in Figs. 10–12. No collapsing is observed for all
specimens. In most cases, small cracks were observed around pieces of rock or
FIG. 9—UTM-100 machine with computer interface (right).

FIG. 10—A vertically oriented naturally frozen soil cylinder with ice lenses
before testing (left) and after testing to 15 % axial strain (right).
FIG. 11—Specimen P4V2 after failure because of shearing.

FIG. 12—Specimen C2H1 after failing because of bending.
ice lenses on the surface of specimens. Small ice lenses were found throughout
the specimens in both the permafrost and seasonally frozen soil and were dis-
tributed non-uniformly. For example, Fig. 10 shows a vertically oriented speci-
men before and after testing. Note the ice lenses presented in this specimen; it
is believed that the distribution of ice lenses, among other factors, will affect
the specimen’s failure mode. Water contents of the seasonally frozen soil were
mostly above 100 %, but the visually observed amount of ice lenses did not
correspond well with the water content.
Sample Testing Results

Figure 13 shows typical stress strain curves from this study. Four sets of
stress–strain curves from four permafrost specimens are shown. Note the stress
is much higher for a horizontally oriented specimen than for a vertically oriented
specimen, even at higher test temperatures. Similar observations can be made
FIG. 13—Typical stress strain curves.

for the other pair of specimens. This indicates that specimen orientation does
matter. A companion paper by Ge et al. [8] discusses the testing results in detail.
Conclusion
This paper describes the processes used to extract and machine naturally frozen
soil specimens for a research program aimed at investigating the mechanical
properties of naturally frozen soils for the seismic design of bridge deep foun-
dations. It describes the preparation of naturally frozen specimens including
block sampling from the field, transportation, and storage of block samples to
the lab, and a detailed machining process of frozen soil specimens for uncon-
fined compression tests. Attention is paid to specimen orientation. A more
detailed rate of lathing is described allowing for quicker machining of frozen
soil samples without heat affecting the sample. This paper also describes the
instrumentation and compression testing methods used in this study with a few
observation of specimen failure mode. Sample testing results are presented,
and indicates that the specimen orientation does affect the stress–strain curve.
Acknowledgments
This project is jointly funded by Alaska University Transportation Center
(AUTC) and the State of Alaska Department of Transportation and Public
Facilities (AK DOT&PF) under AUTC Project No. 510021. This financial sup-
port is greatly appreciated. Thanks to Tim Kirk, John Brueck, and Donald
Richardson for their assistance in sampling, specimen machining, and testing.
Mr. Elmer Marx from AK DOT&PF is the Project Advisory Comm. chair, and
we appreciate his support for this project.
References
[1] Baker, T. H. W., “Transportation, Preparation, and Storage of Frozen Soil

Samples for Laboratory Testing,” ASTM STP 599, ASTM International,
West Conshohocken, PA, 1976, pp. 88–112.
[2] Ebel, W., “Influence of Specimen End Conditions and Slenderness Ratio
on the Mechanical Properties of Frozen Soils,” Proceedings of the Fourth
International Symposium of Ground Freezing, Sapporo, Japan, Aug 5–7,
1985, pp. 231–236.
[3] De Re, G., Germaine, J. T., and Ladd, C. C., “Triaxial Testing of Frozen
Sand: Equipment and Example Results,” J. Cold Regions Eng., Vol.
17(3), 2003, pp. 90–118.
[4] Li, Q., Yang, Z., Marx, E. E., and Lu, J., “Seasonally Frozen Soil Effects
on the Dynamic Behavior of Highway Bridges,” Sci. Cold Arid Regions,
Vol. 4(1), 2011, pp. 13–20.
[5] ASTM, D7015-07: Standard Practices for Obtaining Intact Block (Cubical
and Cylindrical) Samples of Soils, Annual Book of ASTM Standards,
ASTM International, West Conshohocken, PA, 2007.
[6] ASTM D7300-06: Standard Test Method for Laboratory Determination of
Strength Properties of Frozen Soil at a Constant Rate of Strain, Annual
Book of ASTM Standards, ASTM International, West Conshohocken, PA,
2006.
[7] ASTM D4083-07: Standard Practices for Description of Frozen Soils
(Visual-Manual Procedure), Annual Book of ASTM Standards, ASTM
[8] Ge, X., Yamg, Z., and Still, B., “Mechanical Properties of Naturally Frozen
Silty Soils,” ASTM STP 1568, ASTM International, West Conshohocken,
PA, 2013.
STP 1568, 2013
DOI:10.1520/STP156820130012
Finn E. Oestgaard1 and Hannele K. Zubeck1
Practice of Testing Frozen Soils
REFERENCE: Oestgaard, Finn E. and Zubeck, Hannele K., “Practice of

Testing Frozen Soils,” Mechanical Properties of Frozen Soils, STP 1568,
ABSTRACT: When designing building foundations, infrastructure, or opera-
tions for natural resource retrieval in cold regions, mechanical properties of
frozen soils need to be known in order to avoid sudden bearing capacity fail-
ures, excessive creep settlements, and slope failures. Several methods exist
to measure the mechanical properties of frozen soils, some of them specified
by the ASTM International; however, it is not clear if these test methods are
in current use. Do they need to be modified or, in the light of several new
developments in the Arctic areas, are new test methods needed? To investi-
gate these issues as well as issues with the rest of the D18.19 standards, a
survey was created and sent to laboratories around the world testing frozen
soils. The purpose of this paper is to present the survey results. Responses
were received from Alaska, Canada, Denmark, Norway, Sweden and the
continental United States. According to the survey results, none of the prac-
tices and standard test methods currently under the jurisdiction of the
D18.19 are considered outdated; they are all relevant to current industry
practice, are utilized as references for their corresponding test methods, and
are used by several laboratories. Laboratory tests not currently standardized
by ASTM International are performed on a regular basis in the frozen soil
industry. The standardization of the following frozen soil tests is recom-
mended: Thaw Consolidation Test, Shear Stress Test and Tri-axial Compres-
sion Test. The standardization of the field tests examined in this study is not
recommended until further investigation regarding these test methods are
performed. The current D18.19 standard test methods and practices do not
require major modifications. The only recommended modification for the cur-
rent standards relates to ASTM D4083-89: Standard Practice for Description
Manuscript received January 16, 2013; accepted for publication February 27, 2013; published
1
Civil Engineering Department, University of Alaska Anchorage, Anchorage, AK 99508, United
States of America.
2
Jacksonville, FL.
Copyright V
19428-2959.
62
OESTGAARD AND ZUBECK, doi:10.1520/STP156820130012 63
of Frozen Soils, Annual Book of ASTM Standards, ASTM International, West

Conshohocken, PA, 2007. These results can be used to guide the future de-
velopment of D18.19 standards.
KEYWORDS: frozen, soils, testing, practice, mechanical properties
Introduction
When designing building foundations, infrastructure, or operations for natu-
ral resource retrieval in perennial and seasonal frost areas, knowledge of fro-
zen soil properties is relevant to avoid frost action related problems, sudden
bearing capacity failures, excessive creep settlements, and/or slope failures.
Several methods exist to classify and measure the properties of frozen soils,
some of them specified by the ASTM International. However, it is not clear
if these test methods are in current use, need modification, and in the light
of several new developments in the Arctic areas, if new test methods are
needed. The authors together with the ASTM Subcommittee D18.19 on Fro-
zen Soils and Rock created a survey to investigate these issues. The work
was performed by conducting a literature review and by creation of a sur-
vey. The purpose of this paper is to present the findings of the literature
review and the results of the survey.
Current Frozen Soil Standard Test Methods and Practices

Several test methods are used by industry to characterize frozen soils. The fol-
lowing standards are currently under the jurisdiction of Subcommittee D18.19:
• ASTM D4083-89 (2007) Standard Practice for Description of Frozen
Soils (Visual-Manual Procedure) [1],

• ASTM D5520-11 Standard Test Method for Laboratory Determination of
Creep Properties of Frozen Soil Samples by Uniaxial Compression [2],

• ASTM D5780-10 Standard Test Method for Individual Piles in Perma-
frost Under Static Axial Compressive Load [3],

• ASTM D5918-06 Standard Test Methods for Frost Heave and Thaw
Weakening Susceptibility of Soils [4],

• ASTM D6035-08 Standard Test Method for Determining the Effect of

Specimens Using a Flexible Wall Permeameter [5],
• ASTM D7099-04 (2010) Standard Terminology Relating to Frozen Soil
and Rock [6], and

• ASTM D7300-11 Standard Test Method for Laboratory Determination
of Strength Properties of Frozen Soil at a Constant Rate of Strain [7].

Descriptions for the aforementioned standards can be found in Refs 1–7.
Other test methods not standardized by the ASTM International (or called by dif-
ferent names) are listed here and briefly described at the end of the paper:
• shear stress test,

• tri-axial compression test,
• uniaxial tensile stress test,
• cyclic compression test,
• constant creep test,
• compressibility of thawing soils,
• relaxation test,
• thaw-consolidation test,
• pile pullout test,
• pressuremeter creep test, and
• pressuremeter relaxation test.
Survey Design
Survey Questions
The questions of this survey were designed to investigate the current involve-
ment of ASTM International standards on frozen soil laboratory and field tests.
The questions were intended to find answers to the primary objectives:
• Are the current standards being used?
• Do they require modifications?
• What other test methods and practices are currently being used in indus-
try that are not standardized?

The survey was divided into multiple blocks with a display-flow logic. A
total of 112 questions were included. However, approximately half or fewer
were presented to the user during the survey because of the display logic
attached to each question. Question formats consisted of multiple choice, select
all that apply, and short answer.
Survey Results
Survey Audience
The survey was sent out to laboratories, engineering firms, and universities,
and was available at the University of Alaska Anchorage School of Engineer-
ing web page. Twenty-two responders from professors, Ph.D. students, senior
engineers, military engineers, and laboratory managers were received. These
respondents are from around the globe including the United States, Canada,
Denmark, Norway, Sweden, and Finland. Figure 1 displays the different areas
of industry the tests are conducted for and the number of laboratories that use
frozen soil testing for each area. Academic research and construction site anal-
ysis were the most popular reasons for conducting frozen soil testing. The
“other” purposes in Fig. 1 included consulting, geotechnical engineering/foun-
dation design, design, and preconstruction design.
FIG. 1—Frozen soil testing per purpose (number of responses).

Current ASTM Standards
All of the standards under Subcommittee D18.19 were found to be utilized in
the laboratories surveyed. This block of the survey received 15 responses. Fig-
ure 2 displays each test method or practice described in ASTM International
D18.19 standards pertaining to frozen soil and rock, and the number of
respondents that perform these tests and practices. The practices described in
D4083 and D7099 are utilized the most with ten responses.
Figure 3 shows the results when asked if the corresponding ASTM Interna-
tional standard was utilized as a guide when performing the specific test. All of
the standards are utilized when performing their corresponding tests. The most
common reply when asked why the standard was not utilized as a guide was
because of the testing being performed before the standard was published. The
one other response that was submitted was because of ASTM International
Standards not being in public domain.
Figure 4 displays the results when asked if a standard is a valuable test or prac-
tice for industry and the respondents’ laboratory. The D4083 and the D5520 were
reported the most valuable standards. Comments included a deficiency in regard to
D4083: the deficiency relates to establishing a standardized method and lack of
terms to describe the “volume” of visible ice in a frozen sample. Several respond-
ents expressed interest in assisting with the modification of current standards.
Other Laboratory Tests

This block of the survey received 15 responses. The utilization of laboratory
test methods performed on frozen or thawing soil or rock not standardized by
ASTM International is shown in Fig. 5.
FIG. 2—Number of laboratories using each ASTM International Standard

D18.19.
FIG. 3—Number of laboratories utilizing ASTM International standards as a

guide when performing specific test.
FIG. 4—Importance of ASTM International standards (number of responses).
Out of the test methods not standardized by the ASTM International, the
thaw consolidation test is the most popular, whereas all the others listed are
used as well, except for the relaxation test. Other tests that were not listed in
the survey, but submitted by the respondents are:
FIG. 5—Number of laboratories performing specific tests on frozen or thawing

soils or rock.
• wax densities of frozen soils,

• tube suction test,
• free resonant column test,
• segregation potential of soils,
• frozen state oedometer test,
• heat capacity at a specific temperature,
• unfrozen water content at a specific temperature,
• sound velocity at a specific temperature,
• electric resistivity at a specific temperature, and
• indentation of spherical stamp by Vialov and Tsytovitch test.
Figure 6 displays the responses received when asked if each test method is
a valuable test for industry and the respondents’ laboratory. Tri-axial compres-
sion test and thaw consolidation test are reported valuable by the largest
FIG. 6—Importance of test methods for industry (number of responses).

number of responders; however, all listed test methods are important for at
least two surveyed laboratories.
Figure 7 displays the results when asked if standards were developed or if
pre-published standards were utilized for each test method. Many laboratories
have developed their own test methods, whereas pre-published standards are
also in use. The majority of survey responses declined to share the developed
standards with ASTM International for publication; however, one respondent
for each of the following tests were willing to share their developed standards:
tri-axial compression test, uniaxial tension test, and indentation of spherical
stamp by Vialov and Tsytovitch. The majority of respondents did not submit
information when asked for references for the pre-published standards that
were utilized as a guide. The one reference submitted was for the cyclic load-
ing test: AASHTO T307-99 [8].
Field Tests
This block of the survey received responses from six respondents. The utiliza-
tion of field tests performed on frozen or thawing soil or rock not standardized
by ASTM International is shown in Fig. 8.
The pile pullout tests were used by three of the respondents, whereas the
pressuremeter relaxation test, pressuremeter creep test, as well as the falling
weight deflectometer test (other) were used by one responder (the second re-
sponder for “other” did not list any methods). Figure 9 shows the value of each
test for the industry and the respondents’ organization. One laboratory devel-
oped a method for the pile pullout test (proprietary) and the other laboratories
used pre-published methods for the tests in Fig. 9 (no references were listed).
FIG. 7—Use of prepublished standards (number of responses: yes ¼ prepub-

lished standard was used, no ¼ standards were developed).
FIG. 8—Utilization of field test methods (number of responses).
FIG. 9—Importance of test method to industry (number of responses).
Conclusions and Recommendations

The tests standardized by ASTM International regarding frozen soil and rock
focus on tests and practices that are all currently utilized by laboratories in
industry. None of the tests are outdated; they are all relevant to current industry
practice and are utilized as references for their corresponding test methods and
practices. There is willingness to further improve the test methods.
Laboratory and field tests not currently standardized by ASTM Interna-
tional are performed on a regular basis in the frozen soil industry. The stand-
ardization of these tests by ASTM International would benefit the industry in
the form of more normalized frozen soil properties.
The ASTM International Subcommittee D18.19 standards do not require

major modifications. The only modification suggested relates to the D4083,
and work is already being performed within the D18.19.
The standardization of the following frozen soil tests is recommended:
thaw consolidation test, shear stress test, and tri-axial compression test. The
uniaxial compression test and cyclic compression test are also recommended to
have standards developed; however, these tests are not utilized as much as the
other tests.
Test Method Descriptions

Shear stress test is a laboratory test that determines the shear strength of frozen
soils. The use of the test method is reported by, e.g., Bennett and Nickling [9]
and Yasufuku et al. [10].
This test method requires the utilization of a shear box capable of keeping
the specimen at a predetermined subfreezing temperature or the shear box can
be operated in a cold room. A test specimen is prepared by constructing the
specimen with the desired material makeup, freezing the soil, and machining
the sample to the required dimensions for the shear box. The sample is placed
in the shear box and exposed to a shear stress at a constant strain rate or con-
stant stress.
Tri-axial compression test examines the behavior of frozen soils when
exposed to a compressive force and confined from expansion in the lateral
direction. Results from this test provide information regarding creep and
strength properties of the soil. Several authors have used this method [11–14].
A cylindrical soil specimen is placed into a cell that contains cooled fluid that
will provide the confining pressure, and then loaded into the test apparatus.
The test apparatus is capable of loading the specimen axially, creating a confin-
ing pressure around the specimen through a cell fluid, measuring and recording
the pressure the test specimen is under, and measuring and recording the vol-
ume displacement indirectly by measuring the change in volume of the cell
fluid. The tests are most commonly carried out at constant strain rates or as a
constant stress creep test, however, step loading is another alternative. Stand-
ardization of this test method has been discussed by Baker et al. [11] and sug-
gested by Jesseberger [15] and Bragg and Andersland [16].
Uniaxial tensile stress test is a laboratory test that determines the uniaxial
tensile strength of frozen soil specimens [17–19]. Test specimens are prepared
from core samples or undisturbed blocks. The samples have a specific shape
resembling a dumbbell. Haynes et al. [20] designed conditions for test sample
geometry to minimize stress concentrations. The accuracy of the dimensions
for the test specimen is very important in obtaining precise results. This leads
to the test specimens requiring machining by a lathe to ensure useable data.
The loading apparatus is capable of loading the specimen with a tensile force
at a predetermined rate, measuring the forces exerted on the specimen, record-
ing time, and recording the displacement of the specimen ends. Special atten-
tion is required to ensure that the specimen is aligned accurately and that all
tools that come in contact with the soil specimen are cooled to prevent any sort
of localized thawing of the specimen.
Cyclic compression test is a laboratory test that determines the strength
characteristics of frozen soils when exposed to cyclic compression forces. This
test method utilizes the same apparatus as the triaxial compression test; how-
ever the apparatus is required to load the soil specimen cyclically for a prede-
termined amount of time and at a specific loading rate and amplitude. The test
may be performed at different confining pressures or at a single effective con-
fining pressure. Cyclic triaxial strength test results are used for evaluating the
ability to resist stresses induced in soil mass caused by earthquakes or other
cyclic loading. ASTM International D5311 Standard Test Method for Load
Controlled Cyclic Triaxial Strength of Soil [21] describes the method for unfro-
zen soils.
Constant-stress creep test: Frozen soils under relatively high loads will
deform slowly with time. Constant-stress (creep) (CSC) tests are conducted to
evaluate the creep rate and strength parameters [22]. The soil sample is loaded
to predetermined stress and held at this level for the length of the test. The pre-
determined stress level is carefully selected to obtain representative results in a
reasonable amount of time. Creep curves for the soil sample can be created
with the data obtained by this test by plotting strain versus time. The tempera-
ture at which the test was conducted will have an influence on the shape of the
creep curve so it is important to control the temperature during the test. The
results from this test assist in determining the allowable service loads for the
soil supporting a structure over the expected service life for the structure.
Compressibility of thawing soils: Thawing soils are subject to self-weight
loading and/or applied loads. The Terzaghi consolidation theory is used to
model the compressible behavior of the soil skeleton. The Terzaghi principle
states that all quantifiable changes in stress to a soil are a direct result of a
change in effective stress. The effective stress summed with the pore pressure
is equal to the total stress. Compressibility of thawing soil utilizes an oedome-
ter to measure the compressibility of the sample. This test is similarly executed
to unfrozen soil samples with an extra variable, the rate of thawing. This must
be carefully controlled because of the drastic effects it can have on the soil.
This test is a key in estimating the settlement of structures built on locations
containing sensitive permafrost.
Relaxation test is a laboratory test that measures the stress response of soil
when held at a constant strain. Relaxation testing is an important alternative for
investigating the creep properties of a material. Stress relaxation is fundamen-
tal process by which an effective measure of the state of stress within a solid
can be seen to decrease in time because of the conversion of elastic into inelas-
tic strain [22]. A relaxation test can generally be viewed as the inverse of a
creep test: Instead of maintaining a constant load and recording strain variation
with time, an initial strain level is maintained and the decaying stress is moni-
tored over time. The importance of relaxation testing is the potential it offers
for verifying experimentally and independently a material flow law, based orig-
inally on creep testing [22]. A soil specimen initially unstressed is subjected to
a rapid load. The load is maintained once the targeted strain is achieved and
the stress response of the specimen is measured as a function of time. The
stress–time data from the experiment results in a relaxation curve for each tar-
geted strain, which can be used in the estimation of creep parameters for cer-
tain creep equations.
Thaw-consolidation test is a laboratory test that determines the volumetric
reduction of a thawed soil sample under a steady static pressure [23]. Results
from this test are used to determine the thaw consolidation ratio of the soil
specimen. During freezing, frost heave occurs in saturated fine-grained soils
because of icing of water, and, during the following thawing, thaw-
consolidation takes place [24]. Consolidation of saturated soil is a process of
volume reduction caused by the expulsion of water from the void space [25].
As frozen soil is exposed to temperatures above 0 C, thawing will initiate.
Thawing at a slow rate allows generated water to flow from the soil at about
the same rate as melting occurs. The apparatus utilized for this test is an oed-
ometer that measures the compressibility of a soil. It is a laboratory device ca-
pable of loading a soil specimen to a design pressure and measuring the
volume reduction of the soil because of the pressure applied and displacement
of water. The basic procedure for this test is to place a specimen into the oed-
ometer and apply the design pressure onto the specimen for the desired length
of time. The volume reduction is recorded for each step load and specimen.
The specimens can be thawing or a thawed sample and pore pressure is typi-
cally monitored during the duration of the procedure. Multiple procedure varia-
bles exist when conducting this test and are left up to whoever is performing
the test resulting in discrepancies in similarity results when comparing results
from different sources. Watson et al. [26] conducted multiple experiments on
permafrost core samples, including thaw consolidation, and noted the lack of a
standard procedure for this test.
Pile pullout test is a field test that examines the interaction at the pile–soil
interface when the pile undergoes a tensile load. To obtain accurate results, the
test pile needs to be as similar as possible to the actual pile of interest, which
includes installation method, ground temperature, pile size, pile shape, and
ground makeup. Once the pile is installed regardless of method, driven, drilled,
or slurry with refreezing, a delay of one to two weeks is required to allow for
the stabilization of the frozen soil and its surroundings. Once the pile is under a
load, the loading is usually increased in steps until failure. After a certain
amount of creep displacement, the piles fail by slip along the soil–pile inter-
face. The slip failure is considered to coincide with the onset of tertiary creep
[27]. Displacement with time is the primary measurement taken during pile
pullout tests.
Pressuremeter creep test is a field test that utilizes a special device
(Menard Pressuremeter) that is designed to be inserted into a hole bored into
the frozen soil and exerts a force on the borehole walls. The probe portion that
is inserted into the hole is inflatable. A pressure control device inflates the
probe to a desired pressure which applies a desired force on the borehole walls.
The volume increase of the borehole is measured. Two pressuremeter tests are
required to obtain a clear picture of creep characteristics. The first test is a
single-stage creep test. The pressure of the probe is rapidly brought to the
desired pressure and left at that pressure for as long as possible. The total de-
formation of the borehole is recorded. The second test is a multi-stage creep
test. In this test, the pressure is rapidly brought up to an initial level and
increased to the limit of the probe’s capacity. This is done for several
equal stress increments, where they are each kept constant for a minimum of
15 min [22].
Experience by Ladanyi [28] shows that there are limitations to these creep
tests. Because of the total volume limitation of the pressuremeter, cells furnish
creep data only for relative short creep times and for a medium range of
stresses. These limitations led to the use of pressuremeter relaxation tests.
Pressuremeter relaxation test is a field test using the Menard Pressuremeter
for determining creep properties of frozen soils. The probe is inserted into the
borehole and pressurized until a targeted strain is attained and the decaying
stress is monitored over time. Ladanyi and Melouki [29] describe the pressure-
meter relaxation test as the inverse of a creep test. Instead of maintaining a con-
stant load recording strain variation with time, initial strain level is maintained
and the decaying stress is monitored over time. The pressuremeter relaxation
test has advantages over the pressuremeter creep test resulting in it being uti-
lized more when exploring borehole creep characteristics. The most significant
advantage of the relaxation test is the fact that, in a relaxation test, strain is con-
trolled and stress variation with time is recorded so the total length of testing
time is not limited by the volume capacity of the test cell as in the creep test.
The volume limitation of the probe during creep tests allows the relaxation
tests to test for longer periods of creep and a larger range of stresses.
References
[1] ASTM D4083-89: Standard Practice for Description of Frozen Soils, An-
nual Book of ASTM Standards, ASTM International, West Conshohocken,
PA, 2007.

Creep Properties of Frozen Soil Samples by Uniaxial Compression,
Annual Book of ASTM Standards, ASTM International, West
[3] ASTM D5780-10: Standard Test Method for Individual Piles in Perma-
frost Under Static Axial Compressive Load, Annual Book of Standards,
[4] ASTM D5918-06: Standard Test Methods for Frost Heave and Thaw
Weakening Susceptibility of Soils, Annual Book of ASTM Standards,
[5] ASTM D6035-08: Standard Test Method for Determining the Effect of
Specimens Using a Flexible Wall Permeameter, Annual Book of ASTM
[6] ASTM D7099-04: Standard Terminology Relating to Frozen Soil and
Rock, Annual Book of ASTM Standards, ASTM International, West Con-
shohocken, PA, 2010.
2011.
[8] AASHTO, 2007, “Standard Method of Test for Determining the Resilient
Modulus of Soils and Aggregate Materials, Single User Digital
Publication,” reviewed 4/23/2013, https://bookstore.transportation.org/
item_details.aspx?id¼841.
[9] Bennett, L. and Nickling, W. G., “The Shear Strength Characteristics of
Frozen Coarse Granular Debris,” J. Glaciol., Vol. 30, No. 106, 1984.
[10] Yasufuku, N., Springman, S. M., Arenson, L. U., and Ramholt, T.,
“Stress-Dilatancy Behaviour of Frozen Sand in Direct Shear,” in Perma-
frost, Swets & Zeitlinger, Amsterdam, 2003, p. 1253.
[11] Baker, T. H. W., Jones, S. J., and Parameswaran, V. R., “Confined and
Unconfined Compression Tests on Frozen Sands,” 4th Canadian Perma-
frost Conference, March 2–6th, Calgary, Alberta, 1981, pp. 387–393.
[12] Arenson, L., Johansen, M., and Springman, S., “Effects of Volumetric Ice
Content and Strain Rate on Shear Strenght under Triaxial Conditions for
Frozen Soils Samples,” Permafrost Periglac., Vol. 15, 2004, pp. 261–271.
[13] Parameswaran, V. R. and Jones, S. J., “Triaxial Testing of Frozen Sand,”
J. Glaciol., Vol. 27, No. 95, 1981, pp. 147–156.
[14] Fish, A. M., “Creep and Strength of Frozen Soil under Triaxial
Compression,” CRREL Special Report 94-32, Hanover, NH, 1994.
[15] Jessberger, H. L., “State-of-the-art Report, Ground Freezing Mechanical
Properties, and Design,” in Proc. the 2nd International Symposium on
Ground Freezing, Trondheim, Norway, 1980, pp. 1–33.
[16] Bragg, R. A. and Andersland, O. B., “Strain Rate, Temperature, and Sam-
ple Size Effects on Compression and Tensile Properties of Frozen Sand,”
Proc. the 2nd International Symposium on Ground Freezing, Trondheim,
Norway, 1980, pp.34–47.
[17] Zhu, Y. and Carbee, D. L., “Tensile Strength of Frozen Silt,” CRREL
Report 87-15: Cold Regions Research and Engineering Laboratory, Hano-
ver, NH, 1987.
[18] Erckhardt, H., “Creep Tests with Frozen Soils under Uniaxial Tension
and Uniaxial Compression,” in Roger J. E. Brown Memorial Volume,
Proc. of the 4th Canadian Permafrost Conference, National Research
Council of Canada, Calgary, Canada, 1981, pp. 394–405.
[19] Sopko, J. A., Jr., “New Design Method for Frozen Earth Structures with
Reinforcement,” Ph.D. Dissertation, Michigan State University, East
Lansing, MI, 1990.
[20] Haynes, F. D., Karalius, J. A., Kalafut, J., “Strain Rate Effect on the
Strength of Frozen Silt,” Volume 350 of Research Report, Cold Regions
Research and Engineering Laboratory, Hanover, NH, 1975.
[21] ASTM D5311: Standard Test Method for Load Controlled Cyclic Triaxial
Strength of Soil, Annual Book of ASTM Standards, ASTM International,
West Conshohocken, PA, 2004.
[22] Andersland, O. and Ladanyi, B., Frozen Ground Engineering, 2nd ed.,
Wiley and Sons, Hoboken, NJ, 2004.
[23] Morgenstern, N. R. and Nixon, J. F., “One-dimensional Consolidation of
Thawing Soils,” Dept. of Civil Engineering, Univ. of Alberta, Edmonton,
Alberta, 1971.
[24] Zou, Y. and Boley, C., “Compressiblity of Fine-Grained Soils Subjected
to Closed-System Freezing and Thaw Consolidation,” Min. Sci. Tech.,
Vol. 19, 2009, pp. 631–635.
[25] Crawford, C. B., “Interpretation of the Consolidation Test,” Proceedings
of the American Society of Civil Engineers Journal of the Soil Mechanics
and Foundations Division, Vol. 90, No. SM 5, 1964, pp. 87–110.
[26] Watson, G. H., Slusrchuk, W. A., and Rowley, R. K., “Determination of
Some Frozen and Thawed Properties of Permafrost Soils,” Can. Geotech.
J., Vol. 10, No. 4, 1973, pp. 592–606.
[27] Johnston, G. H. and Ladanyi, B., “Field Tests of Grouted Rod Anchors in
Permafrost,” Can. Geotech. J., Vol. 9, No. 2, 1972, pp. 176–194.
[28] Ladanyi, B., “Borehole Creep and Relaxation Tests in Ice-Rich
Permafrost,” Proc. the 4th Canadian Permafrost Conference, 1982, pp.
406–415.
[29] Ladanyi, B. and Melouki, M., “Determination of Creep Properties of Fro-
zen Soils by Means of the Borehole Stress Relaxation Test,” Can. Geo-
tech. J., Vol. 30, 1992, pp. 170–186.
STP 1568, 2013
DOI:10.1520/STP156820130010
Tamra S. Kornfield1 and Hannele K. Zubeck1
Triaxial Testing of Frozen Soils—State

of the Art
REFERENCE: Kornfield, Tamra S. and Zubeck, Hannele K., “Triaxial Testing

of Frozen Soils—State of the Art,” Mechanical Properties of Frozen Soils, STP
1568, Hannele Zubeck and Zhaohui Yang, Eds., pp. 76–85, doi:10.1520/
ABSTRACT: Mechanical properties of frozen soils are invaluable input pa-
rameters when designing building foundations or infrastructure in perennial
frost areas. Mechanical properties are also important for natural resource de-
velopment in the north, such as for mining and petroleum-industry-related
projects. One of these properties is the shear strength of frozen soil under
varying temperatures and loading times. If shear strength is estimated
instead of measured, risks for failure or overdesign exist. Therefore, it is im-
portant to accurately measure the strength of frozen soils. Two main methods
exist for measuring the shear strength of soils: the direct shear test (ASTM
D3080) and the triaxial compression test (ASTM D4767, ASTM D7181, and
ASTM D2850). Although these tests are routinely used for unfrozen soils, not
much published information exists regarding their use for frozen soils. Yet the
industry needs this property for planning their operations in cold regions.
Therefore, ASTM International Subcommittee D18.19 on Frozen Soils and
Rock has started a process of developing new standards for mechanical
properties of frozen soils. Of special interest is the dynamic triaxial testing of
frozen soils. The purpose of the study reported here was to collect informa-
tion and practices for the current usage of triaxial testing for frozen soil, under
either static or dynamic loading conditions. According to the results of the lit-
erature review, researchers use various modified testing systems and sample
configurations, and unfortunately they do not always describe them fully. So,
standardization of the testing method would be beneficial for creating compa-
rable results between laboratories. The measurement of small strains
and deformations in dynamic tests was reported to be challenging. A
Manuscript received January 14, 2013; accepted for publication May 21, 2013; published online
August 21, 2013.
1
Civil Engineering Dept., Univ. of Alaska Anchorage, Anchorage, AK 99611, United States of
America.
2
Jacksonville, FL.
Copyright V
19428-2959.
76
KORNFIELD AND ZUBECK, doi: 10.1520/STP156820130010 77
Split-Hopkinson pressure bar is currently being developed as a means to

overcome this problem. Another challenge under investigation is the accu-
racy of the strain rate control. The synthesized information can be used as a
starting point in the development of a standard test method for the dynamic
triaxial testing of frozen soils.
KEYWORDS: triaxial testing, frozen soils, mechanical properties, state of
the art
Introduction
The stress-strain-strength behavior of frozen soil is an important design consid-
eration for structures, infrastructure, and natural resource development and ex-
ploration in perennial frost regions. The technique of artificial ground freezing
is also routinely used in warmer regions to control groundwater and support
shaft and underground construction [1]. However, the mechanical behavior of
frozen soil is difficult to predict because of the presence of ice and unfrozen
water within the frozen soil. Ice makes the soil susceptible to creep settlement,
and the magnitude of the creep is strongly affected by temperature. Further-
more, estimations of frozen soil strength do not provide enough accuracy to
ensure structural soundness and environmental stability of frozen soil during
cyclic loading, such as occurs under the vibration of machinery, drilling equip-
ment, or seismic activity. Inaccurate estimations are not cost effective and can
result in overdesign or a risk of failure.
The purpose of the study reported here was to collect information and prac-
tices for the current usage of triaxial testing for frozen soil, under either static or
cyclic loading conditions. This test is used to examine the stress-strain behavior
of frozen soils in order to further understand the behavior of frozen soil. ASTM
International test methods exist for the triaxial testing of unfrozen soils, but
because of the complicated nature of frozen soils, these practices are not
adequate to test or predict the behavior of frozen soils [2]. The following reviews
of literature pertaining to triaxial testing include a description of the equipment
requirements, specimen considerations, testing procedures, data analysis, and
typical results. This information could be used in the development of an ASTM
International standard test method for the triaxial testing of frozen soils.
Aspects of Triaxial Testing

Interest in the mechanical properties of frozen soils began in the mid-20th cen-
tury with an interest in creating stable artificially frozen coal mining shafts [3].
The significance of frozen soil testing stems from the necessity of designing
foundations embedded in or bearing upon frozen ground. Tests are used to pre-
dict time-dependent settlements under a load from piles and shallow founda-
tions, as well as to predict the overall short- and long-term load capacity.
There has not been much change in triaxial test procedures, as shown by a
series of tests on the mechanical behavior of frozen soils conducted in 1970.
Neuber and Wolters [3] show that the significance of the temperature and rate
of loading is great and therefore must be measured accurately. Da Re et al. [4]
report that efforts toward improved control over the accuracy of load applica-
tion, temperature control, and strain rate control have been made since 1995 to
improve small strain behavior measurement (see “Equipment Requirements”
and “Testing Procedures” sections).
Overall, it was observed that there are many variables to account for in fro-
zen soil testing. The equipment used, soil type (e.g., sand or clay), specimen
size, and temperature are just some of the testing considerations. Soil character-
istics such as water and ice content, grain-size distribution, and ice lenses and
layering must also be considered [5]. Externally imposed testing conditions are
also a consideration, including strain rate, temperature, stress and strain history,
and confining pressure. No two articles reported on the same variables. How-
ever, the literature found reports similar testing procedures and agrees on typical
test result trends.
As described in ASTM D7300-06 [6], tests should report on the follow-
ing: a description of soil based on the Unified Soil Classification system for
frozen soils and the Standard Practice for the Description of Frozen Soils
(Visual-Manual Procedure); grain size gradation curve; Atterberg limits
(where applicable); physical properties including total water content, dry unit
weight of soil, specific gravity of soil grains, water/ice saturation (%), and sa-
linity; sampling conditions and specimen preparation, including the sampling
method, ground temperature, temperature changes during transportation and
storage, specimen machining method, and specimen dimensions; and testing
conditions including test temperature, end condition of sample, loading condi-
tions, loading equipment data, description of all tests, and graphs of test
results.
In a typical triaxial test (see Fig. 1), a frozen soil specimen is placed in a
pressure chamber and isotropically loaded by a hydrostatic pressure. Then an
axial load is applied in order to produce shear stresses on the sample under ra-
dial pressure. The axial load is applied either at a controlled loading rate or
cyclically at specified amplitudes and frequencies [7].
Triaxial Testing Equipment

These are some of the types of testing equipment used by researchers: a cryo-
genic triaxial apparatus improved from an MTS material testing machine [2,9],
a Model 1116 Instron universal testing machine of 25 000-kg capacity [10], a
TATW-500 testing machine [11], a regular MTS triaxial testing machine
[11–13], and a screw-driven universal testing machine of 250-kN capacity
[14]. One of the important features of the test is the temperature control of the
apparatus during the test, which may take up to several days. Da Re et al. [4]
used a cold room to control the temperature of their soil specimens, whereas
FIG. 1—Triaxial testing device (a) with sample before test, (b) with sample
after test, and (c) closed during test [8].
other researchers modified their testing systems to fit in a low-temperature
chamber as shown in Fig. 2. Zhang et al. [9] used a cryogenic triaxial apparatus
improved from an MTS-810 material test machine to include a sensing device
for volume change (Fig. 3).
FIG. 2—Universal testing machine with a temperature chamber (photo: Zhao-

hui Yang).
FIG. 3—Cryogenic triaxial apparatus improved from MTS-810 material test

machine. 1: displacement sensing device connected to a data acquisition sys-
tem; 2: dial gauge to monitor the displacement of piston (5); 3: sensing device
for volume change; 4: piston rod; 5: piston; 6: high-pressure oil tube; 7: load-
ing piston; 8: regulated pressure oil supply; 9: two-way valve [9].
Kabir and Chen [7] describe new developments in the measurement of

small strain and deformation in dynamic tests. The authors claim that conven-
tional methods of measurement can adequately record axial load and specimen
deformation for static tests (a line pressure gauge on the hydraulic line to re-
cord the system’s hydrostatic pressure and a linear variable differential trans-
former [LVDT] to measure the deformation of the specimen) but are not
adequate to measure the rapidly changing load and specimen deformation of
dynamic tests. Kabir and Chen [7] describe the use of a Kolsky bar (Fig. 4)
with a 200-ls duration to apply the dynamic axial compression on the speci-
men. Strain gauges are used on the Kolsky bar, and a manganin pressure gauge
is installed in the pressure chamber around the specimen. A capacitive trans-
ducer measures the instantaneous redial dimension change.
Specimen Preparation
The soil types used in the referenced studies are as follows: frozen saturated
Ottawa sand [10,15], Qinghai-Tibet plateau silty clay [5], Manchester fine sand
[4], remolded loess from Lanzhou, China [13], cored alpine permafrost [16],
and artificially prepared frozen soil samples [8,11,13,15,16]. Xu et al. [2] froze
FIG. 4—A schematic of the dynamic triaxial experiment system integrated with
a Kolsky bar [7].
soil samples artificially by quickly freezing them after saturation at 30 C to

avoid frost heave. The specimens were frozen for 48 h and then covered with a
rubber membrane and end platens and kept at the target testing temperature for
12 h so that they could reach a uniform temperature.
ASTM D7300-06 [6] recommends an ideal height-to-diameter ratio of fro-
zen soil specimens of 3:2. Other height:diameter ratios have been also used;
for example, the specimens used by Qi et al. [17], Parameswaran and Jones
[10], and Wang et al. [5] used a ratio of 2:1 (see Table 1).
TABLE 1—Summary of testing details.
Frozen Soil Height/Diameter Axial Confining Temperature,

1
Authors Type Ratio Strain Rate, s Pressure, MPa C
Li et al. [19] Sand 2.4 0 to 1.378 1 to 10
Zhang et al. [9] Sand/silty clay 2.0 1.67 10 4 0 to 18.0 4 and 6
Kabir and Chen [20] Sand 0.5 1000 and 500
Ma and Chang [21] Sand 2.0 1 to 5 2 to 10
Arenson et al. [8] Sand 2.0 0.05 to 0.4 2
5
Parameswaran Sand 2.1 7.7 10 0.1 to 76 10
and Jones [10]
Zhao et al. [11] Clay 2.0 8 to 12 17 to 25
Xu et al. [2] Sand 2.0 0.3 to 18 30 to 2
Testing Procedures
Because the temperature has such a significant effect on the test results, Da
Re et al. [4] recommend using a cold room to keep the temperature constant
during storage and testing. This method limits air temperature fluctuations
around the triaxial cell to 60.17 C from the desired temperature. There is still
a slight temperature gradient in the oil within the triaxial cell, and Da Re
et al. [4] suggest eliminating it with the use of a high-pressure pump to circu-
late the confining fluid. ASTM D5311-11 [18] notes that the interaction of the
specimen, membrane, and confining fluid influences the cyclic behavior of the
soil. As stated above, Kabir and Chen [7] claim that the change in axial
dimension should be measured by an LVDT at low rates and a Kolsky bar at
high rates of loading. This should be accounted for in the testing procedures.
Table 1 summarizes the main aspects of testing procedures used by several
authors.
Presenting Triaxial Testing Data

The data obtained from triaxial testing are described with stress-strain curves
in most research. Arenson and Springman [22] found that creep and constant
strain rate tests can be used to form mathematical formulas to describe the ther-
momechanical behavior of ice-rich frozen soils. As with other studies, a
Mohr–Coulomb failure criterion is used to determine the shear strength of a
material. Di Prisco and Zambelli [23] reiterate this by maintaining that sophis-
ticated modeling approaches are capable of reproducing the quasi-static cyclic
mechanical behavior of frozen soils. With improvements in modeling technol-
ogy, better predictions of the behavior of frozen soils are likely.
Typical Test Results for Certain Frozen Soils

Though tests had various study objectives, soil samples, and test practices, sev-
eral characteristics of frozen soil triaxial tests were agreed upon. The results
presented by Yang et al. [24] and Xu et al. [2] show that the shear strength of
frozen sand increases with an increase in confining pressure (and a decrease in
temperature [not shown]) until it reaches a maximum, and then it decreases
with further increases in confining pressure because of pressure melting and
crushing phenomena, unlike unfrozen soil. According to the results of Xu et al.
[2], the strength of frozen soil changes with increasing confining pressure in
three distinct phases. The first phase shows that pressure melting and crushing
may not occur. The increased confining pressure increases the strength of the
frozen sandy soil by increasing normal pressure on the shear surface and
increases the interlocking frictional force between solid particles by making
them tighter. In the second phase, the strength of the frozen sandy soil
decreases with confining pressure because of crushing and pressure melting in

the specimen. In the third phase the strength of frozen sandy soil is hardly
influenced by confining pressures in excess of 12 MPa. However, some studies
have pointed out that it is important to distinguish between the mechanical
response recorded during small strain cyclic load tests and large strain loading
cycles, as results show that the stress-strain response of soil specimens is insen-
sitive to high load rates [20,23].
Conclusions and Recommendations

Information and practices for the current usage of triaxial testing for frozen soil
were collected. This information can be used in the development of ASTM
International standards for the triaxial testing of frozen soils. The following
findings were obtained from the study:
(1) Researchers used various modified testing systems that were not
always described fully. A standardized testing system is recommended
for producing uniform results.
(2) Researchers use various soil types and mixes for testing specimens, but
not enough published data exist to give typical stress-strain curves for
various soils.
(3) Improvements in strain rate control are being developed to apply and
record accurate levels of axial strain, and advances are being made in
small strain technology. For example, the use of a Kolsky bar is cur-
rently being developed as a means to accurately measure small strain
and deformation in dynamic tests.
References
ASCE, Reston, VA, 2004.
[2] Xu, X., Lai, Y., Dong, Y., and Qi, J., “Laboratory Investigation on
Strength and Deformation Characteristics of Ice-Saturated Frozen Sandy
Soil,” Cold Regions Sci. Technol., Vol. 69, 2011, pp. 98–104.
[3] Neuber, H. and Wolters, R., Mechanical Behaviour of Frozen Soils under
Triaxial Compression, Canada Institute for Scientific and Technical Infor-
mation, Ottawa, ON, Canada, 1970.
[4] Da Re, G., Germaine, J., and Ladd, C., “Triaxial Testing of Frozen Sand:
Equipment and Example Results,” J. Cold Reg. Eng., Vol. 17, 2003, pp.
90–118.
[5] Wang, S., Qi, J., and Yao, X., “Stress Relaxation Characteristics of Warm
Frozen Clay Under Triaxial Conditions,” Cold Regions Sci. Technol.,
Vol. 69, 2011, pp. 112–117.

Strength Properties of Frozen Soil at a Constant Rate of Strain, Annual Book
of ASTM Standards, ASTM International, West Conshohocken, PA, 2004.
[7] Kabir, M. D. and Chen, W., “Measurement of Specimen Dimensions and
Dynamic Pressure in Dynamic Triaxial Experiments,” Rev. Sci. Instrum.,
Vol. 80, 2009, 125111.
[8] Arenson, L., Johansen, M., and Springman, S., “Effects of Volumetric Ice
Content and Strain Rate on Shear Strength Under Triaxial Conditions for
Frozen Soil Specimens,” Permafrost Periglacial Process., Vol. 15, 2004,
pp. 261–271.
[9] Zhang, S., Lai, Y., Sun, Z., and Gao, Z., “Volumetric Strain and Strength
Behavior of Frozen Soils Under Confinement,” Cold Regions Science
Technol., Vol. 47, 2007, pp. 263–270.
[10] Parameswaran, V. and Jones, S., “Triaxial Testing of Frozen Sand,” J.
Glaciol., Vol. 27, No. 95, 1981, pp. 147–156.
[11] Zhao, X., Zhou, G., Chen, G., Shang, X., and Zhao, G., “Triaxial Com-
pression Deformation for Artificial Frozen Clay With Thermal Gradient,”
Cold Regions Sci. Technol., Vol. 67, 2011, pp. 171–177.
[12] Zhu, Z., Ling, X., Wang, Z., Lu, Q., Chen, S., Zou, Z., and Guo, Z.,
“Experimental Investigation of the Dynamic Behavior of Frozen Clay
From the Beiluhe Subgrade Along the QTR,” Cold Regions Sci. Technol.,
Vol. 69, 2011, pp. 91–97.
[13] Wang, D., Ma, W., Wen, Z., and Chang, X., “Study on Strength of Artifi-
cially Frozen Soils in Deep Alluvium,” Tunneling and Underground
Space Technology, Vol. 23, 2008, pp. 381–388.
[14] Baker, T. H. W., Jones, S. J., and Parameswaran, V. R., “Confined and
Unconfined Compression Tests on Frozen Sands,” Proceedings of the 4th
Canadian Permafrost Conference, Calgary, Alberta, Canada, Mar 2–6,
1982, pp. 387–393.
[15] Andersland, O., Baladi, G., and Li, J., “Cyclic Triaxial Tests on Frozen
Sand,” Eng. Geol. (Amsterdam), Vol. 13, 1979, pp. 233–246.
[16] Arenson, L. and Springman, S., “Triaxial Constant Stress and Constant
Strain Rate Tests on Ice-Rich Permafrost Samples,” Can. Geotech. J.,
Vol. 42, 2005, pp. 412–430.
[17] Qi, J., Ma, W., Sun, C., and Wang, L., “Ground Motion Analysis in Seasonally
Frozen Regions,” Cold Regions Sci. Technol., Vol. 44, 2005, pp. 111–120.
[18] ASTM D5311-11: Standard Test Method for Load Controlled Cyclic Tri-
axial Strength of Soil, Annual Book of ASTM Standards, ASTM Interna-
tional, West Conshohocken, PA, 2004.
[19] Li, J., Baladi, G., and Andersland, O., “Cyclic Triaxial Tests on Frozen
Sand,” Eng. Geol. (Amsterdam), Vol. 13, 1979, pp. 233–246.
[20] Kabir, M. D. and Chen, W., “Dynamic Triaxial Test on Sand,” Proceedings
of the Dynamic Behaviour of Materials, Vol. 1, Springer, New York, 2011.
[21] Ma, W. and Chang, X., “Analyses of Strength and Deformation of an

Artificially Frozen Soil Wall in Underground Engineering,” Cold Regions
Sci. Technol., Vol. 34, 2002, pp. 11–17.
[22] Arenson, L. and Springman, S., “Mathematical Descriptions for the
Behaviour of Ice-Rich Frozen Soils at Temperatures Close to 0 C,” Can.
Geotech. J., Vol. 42, 2005, pp. 431–442.
[23] Di Prisco, C. and Zambelli, C., “Cyclic and Dynamic Mechanical
Behaviour of Granular Soils: Experimental Evidence and Constitutive
Modeling,” Geodynamics and Cycling Modeling, Vol. 7, 2003, pp.
881–910.
[24] Yang, Y., Lai, Y., and Li, J., “Laboratory Investigation on the Strength
Characteristic of Frozen Sand Considering Effect of Confining Pressure,”
STP 1568, 2013
DOI:10.1520/STP156820120141
Jiankun Liu,1 Jiannan Zhang,2 Mengqiao Chen,2 and Tianliang Wang3
Experimental Study on Dynamic Properties

of Clay Modified by Aught-Set Solidifying
Agent Subjected to Freeze-Thaw Cycles
REFERENCE: Liu, Jiankun, Zhang, Jiannan, Chen, Mengqiao, and Wang, Tian-
liang, “Experimental Study on Dynamic Properties of Clay Modified by Aught-Set
Solidifying Agent Subjected to Freeze-Thaw Cycles,” Mechanical Properties of
Frozen Soils, STP 1568, Hannele Zubeck and Zhaohui Yang, Eds., pp. 86–94,
doi:10.1520/STP156820120141, ASTM International, West Conshohocken, PA
2013.4
ABSTRACT: Clay modified by a solidifying agent has been used in the recent
construction of high-speed rails in northern China because of a lack of good
fill material. In this study, the dynamic properties of a clay modified by “aught-
set” solidifying agent (a Chinese product) subjected to freeze-thaw cycles
were analyzed. Dynamic triaxial tests were carried out on clay modified by
aught-set solidifying agent that had been subjected to freeze-thaw cycles.
The variation of the critical dynamic stress (CDS) with the cycles and freezing
temperature during freeze-thaw was obtained. The results show that the
CDS of modified clay decreased with the number of freeze-thaw cycles and
decreasing freezing temperature.
KEYWORDS: freeze, thaw, solidifying agent, modified soil, mechanical
properties
Introduction
In recent years, high-speed rail (HSR) has been developing rapidly as an effi-
cient and environmentally friendly mode of transport. Areas subjected to
Manuscript received November 4, 2012; accepted for publication April 24, 2013; published
1
School of Civil Engineering, Beijing Jiaotong Univ., Beijing 100044, China (Corresponding
author), e-mail: jkliu@bjtu.edu.cn
2
School of Civil Engineering, Beijing Jiaotong Univ., Beijing 100044, China.
3
School of Civil Engineering, Shijiazhuang Railway Univ., Shijiazhuang 050043, China.
4
Jacksonville, FL.
Copyright V
19428-2959.
86
LIU ET AL., doi:10.1520/STP156820120141 87
seasonal freezing in China account for more than half of the total land, and
HSR inevitably will be influenced by the freeze-thaw effects of soil. In North-
east China, soils are exposed to many freeze-thaw cycles each year. Such
freeze-thaw cycles change the soil’s moisture content and reduce soil density
and bearing capacity. An effective method for improving soil stability is to
modify the soil by adding a solidifying agent.
Lime, cement, and fly-ash solidifying agents are usually used to modify
soil. There has been much study of the mechanical properties of lime-, cement-
, and fly-ash-modified soil subjected to freeze-thaw cycles. Sauer and Weimer
concluded that the resilient characteristics of glacial till can be improved sig-
nificantly by the addition of lime [1]. Janoo et al. conducted a field evaluation
of subgrade stabilized with Portland cement; the results revealed losses of up
to 50 % in compressive strength caused by freeze-thaw cycles [2]. Through
shear strength tests, Ma and Xu [3] and Xu and Yang [4] found that freeze-
thaw cycles strongly affect the strength and deformation levels of lime-
modified silt. Wei et al. investigated the dynamic properties of fly-ash-modified
soil for subgrade filling and obtained the variation between the dynamic
strength and the number of dynamic loading and freeze-thaw cycles [5]. Yar-
basi et al. showed that lime-, fly-ash-, and cement-modified soils have high
freeze-thaw durability relative to unmodified samples [6]. Liu et al. conducted
dynamic triaxial tests to study the threshold deviator stress and resilient modu-
lus of cement- and lime-modified soils with different blend ratios subjected to
freeze-thaw cycles, and they determined an optimal blend ratio [7].
However, few studies have focused on the mechanical properties of solidify-
ing-agent-modified soil, especially the influence of freeze-thaw cycles. A review
by Paige-Green concludes that there is a rapidly developing market for proprie-
tary soil additives that make use of various chemical reactions during the soil sta-
bilization process [8]. Bobrowski used laboratory and field experiments to
determine the possible benefits obtained by injecting Condor SS, a liquid chemi-
cal stabilizer, into a clay subgrade [9]. Bell chose Poly-Fluoro-Alkoxy (PFA), a
kind of solidifying agent, as an extender with cement and with lime to evaluate
how effective such mixtures were when used to stabilize clay soils [10].
Saboundjian studied the application of an organic soil stabilizer (EMC2) in sub-
grade reinforcement [11]. Huang et al. studied the strength characteristics of sol-
idifying-agent-modified soil subjected to freeze-thaw cycles and analyzed the
influence of the additive ratio, curing age, and number of freeze-thaw cycles
[12]. Dai et al. used Base-Seal stabilizer (BS-100 Model) to stabilize typical
clays found in Changchun, China, and found that Base-Seal-stabilized soil has
high early-stage and long-term strength and excellent frost stability [13].
It can be seen that previous studies of modified soils subjected to freezing
and thawing mainly focused on static properties; few results can be found
related to the dynamic performance and frost stability of solidifying-agent-
modified soil. In this study, a series of triaxial tests were conducted on clay
FIG. 1—Grain size distribution of the studied clay.
modified by aught-set subjected to freeze-thaw cycles, and the dynamic proper-

ties of the clay were analyzed.
Sample Preparation and Testing Procedures
Soil Properties and Sample Preparation

Clay samples were obtained from an HSR construction site located in Anshan,
China. This clay was modified and used in the lower part of the subgrade of the
Harbin-Dalian high speed passenger railway. Figure 1 shows the grain size dis-
tribution curve. Other physical and mechanical indices of the soil are shown in
Table 1.
In the tests, the admixture of aught-set solidifying agent was chosen to
modify the clay soil. The solidifying agent was added to the soils at 3 % of the
dry soil weight, and then the mixtures were compacted at the optimal water
content to obtain stabilized samples, which were stored in a tank for 2 to 4 h.
The compaction process for the modified clay samples was carried out in five
steps at a 95 % degree of compaction. The samples were cylindrical with
dimensions of 39.1 mm in diameter by 80.0 mm in height. All the stabilized
samples were cured for seven days in a moisture tank. At the same time, the
unmodified samples were prepared for comparison with the modified samples.
TABLE 1—Physical properties of studied clay.
Specific Plastic Liquid Plasticity Optimal Water Maximum Dry

Gravity Limit, % Limit, % Index Content, % Density, g/cm3
2.68 21.5 48.6 27.1 18.1 1.80
LIU ET AL., doi:10.1520/STP156820120141 89
Testing Procedure
After seven days of curing, freeze-thaw tests were conducted. In order to simu-
late the freeze-thaw environment found at the HSR construction sites, five dif-
ferent freeze-thaw temperature amplitudes were used: 2 C to 15 C, 5 C to
15 C, 10 C to 15 C, 15 C to 15 C, and 20 C to 15 C. During freezing,
the samples were placed in a freezer at a certain freezing temperature for 12 h.
During thawing, the samples were placed in a freezer at 15 C for 12 h. This
procedure was repeated until the samples had undergone 10 freeze-thaw cycles.
The freeze-thaw cycles for the testing samples took place in a thermo-tank. Af-
ter every 1, 3, 6, 8, or 10 cycles, one sample was taken for testing.
The static triaxial test was conducted using a GDS triaxial test system for
unsaturated soil. A confining pressure of 20 kPa was applied. The loading rate
was 0.5 mm/min, and the maximum axial strain was 15 %. Test data were
recorded every 10 s.
The dynamic triaxial test was conducted using an MTS858.2/TESTSTAR2
testing system. The vibration frequency was 4 Hz for the load type of a sine
wave, and the confining pressure was 20 kPa. According to the Soil Test Tech-
nical Manual (Chinese Ministry of Railway, 2003), the failure criterion is 5 %
for samples with plastic failure, or the turning point when it appears earlier.
Discussion of Test Results
The Effect of Freeze-thaw Cycles on the Static Properties of Modified Soil

Figure 2 shows the stress-strain curves of unmodified and modified soil under a
confining pressure of 20 kPa before freeze-thaw cycling. The static strength of
unmodified soil in this case was 262 kPa, and the peak strength of modified soil
reached 876 kPa, which is three times that of the original, but its residual
strength at 3 % strain was 10 % to 15 % lower than the unmodified one.
The triaxial test results for the static strength of modified soil under a con-
fining pressure of 20 kPa after different numbers of freeze-thaw cycles and at
different freezing temperature amplitudes are shown in Fig. 3. It can be seen
that the static strength of modified soil decreased with increasing freeze-thaw
cycles and decreased with decreasing freezing temperature in freeze-thaw
cycles. In all cases, the first three freeze-thaw cycles had a great influence on
the strength of the modified clay. The peak strength dropped from 876 kPa to
420 kPa after three freeze-thaw cycles when using 20 C as the freezing tem-
perature, a 52 % reduction. When 5 C was used as the freezing temperature,
the strength dropped from 876 kPa to 700 kPa, a 20 % reduction. In all cases,
after six cycles the strength remained nearly unchanged; this is the residual sta-
ble strength after freeze-thaw cycles. The residual stable strength is related to
the freezing temperature. The lower the freezing temperature, the greater the
FIG. 2—The stress-strain curves of the original and modified clay samples
before freeze-thaw cycles.
strength reduction, and the lower the residual stable strength. The lower freez-
ing temperatures also led to a faster reduction of the strength of the tested
modified clay. It can be seen in Fig. 3 that it took three freeze-thaw cycles to
reach the stable residual strength when using 20 C as the freezing tempera-
ture. At 2 C and 5 C, it took four cycles to reach the stable residual
strength.
FIG. 3—The static strength of modified clay under a confining pressure of

20 kPa when subjected to different numbers of freeze-thaw cycles and tempera-
ture amplitudes.
LIU ET AL., doi:10.1520/STP156820120141 91
The Effect of Freeze-Thaw Cycles on the Critical Deviator Stress of Modified

Clay
For soil, it is possible to determine a level of stress above which repeated appli-
cations of a load will cause large permanent deformations, and below which
permanent deformations will be small and negligible for the operation of the
HSR. The dynamic stress that corresponds to a certain critical strain is defined
as the critical dynamic stress (CDS). Figure 4 shows the cyclic triaxial test
results for the original clay [Fig. 4(a)] and for the modified clay [Fig. 4(b)].
The CDS of the original clay was 215 kPa. Brittle failure occurred in the
FIG. 4—Cycle number versus accumulated plastic strain of clay soil without
freeze-thaw cycles: (a) original clay; (b) aught-set modified clay.
FIG. 5—Critical dynamic stress versus the number of freeze-thaw cycles.
modified soil, and its CDS was 675 kPa, which is three times that of the
unmodified soil.
For all studied cases, the CDS was determined experimentally through
cyclic triaxial testing for a certain number of freeze-thaw cycles. The obtained
results are plotted in Fig. 5. The CDS of modified soil was 325 kPa after 10
freeze-thaw cycles when the freezing temperature range was 20 C to 15 C.
This represents a 52 % reduction relative to the sample without a freeze-thaw
cycle. However, it still represents a 51.2 % increase relative to the original
unmodified clay. It can be seen from Fig. 5 that the CDS of modified soil
decreased with increasing freeze-thaw cycles and decreased with decreasing
freezing temperatures in freeze-thaw cycles. When the negative temperature
was 2 C or 5 C, the CDS became stable after six to eight freeze-thaw
cycles. When the freezing temperature was 10 C or 15 C, after three
freeze-thaw cycles the CDS became stable. When the freezing temperature was
20 C, the CDS almost stabilized after one to two cycles.
Conclusion
In this study, the effects of aught-set soil stabilizing agent on the dynamic
properties of clay subjected to freeze-thaw cycles were investigated, and the
following conclusions are drawn.
The static strength of modified soil is three times that of the original
unmodified soil without freeze-thaw cycling. The experiments showed that the
static strength of the modified soil decreased with increasing freeze-thaw
LIU ET AL., doi:10.1520/STP156820120141 93
cycles and with decreasing freezing temperature used in the freeze-thaw

cycles.
The critical dynamic stress (CDS) of the modified soil was three times that
of the unmodified clay without freeze-thaw cycling. The CDS decreased with
increasing freeze-thaw cycles and with decreasing freezing temperature used in
the freeze-thaw cycles. After 10 freeze-thaw cycles using 20 C as the freez-
ing temperature, the CDS of the modified clay was still 51.2 % larger than that
of unmodified clay, meaning the improvement of the soil provided by the solid-
ifying agent was significant.
Acknowledgments
The writers acknowledge the support provided by the National 973 Project of
China (Grant No. 2012CB026104) and the National Science Foundation of
China (NSFC) under Grant No. 41171064.
References
[1] Sauer, E. K. and Weimer, N. F., “Deformation of Lime-Modified Clay af-

ter Freeze-Thaw,” Transp. Engrg. J., Vol. 104(2), 1978, pp. 201–212.
[2] Janoo, V. C., Firicano, A. J., Barna, L. A., and Orchino, S. A., “Field
Testing of Stabilized Soil,” J. Cold Reg. Eng., Vol. 13(1), 1999, pp.
37–53.
[3] Ma, W. and Xu, X. Z., “Influence of Frost and Thaw Cycles on Shear
Strength of Lime Silt,” Chinese J. Geotech. Eng., Vol. 2(21), 1999, pp.
158–160.
[4] Xu, S. and Yang, Y. H., “Experiment Study on Strength Property of Lime
Improved Loess,” Journal of Lan-Zhou Jiao-Tong University (Natural
Sciences Edition), Vol. 6(25), 2006, pp. 97–100.
[5] Wei, H. B., Liu, H. B., Gao, Y. P., Fang Y., and Li, C. Y., “Experimental
Research on Dynamic Properties of Fly Ash Soil Subjected to Freeze-
Thaw Cycles,” Rock Soil Mech., Vol. 28(5), 2007, pp. 1005–1008.
[6] Yarbasi, N., Kalkan, E., and Akbulut, S., “Modification of the Geotechni-
cal Properties, as Influenced by Freeze-Thaw, of Granular Soils With
Waste Additives,” Cold Regions Sci. Technol., Vol. 48(1), 2007, pp.
44–54.
[7] Liu, J. K., Wang, T. L., and Tian, Y. H., “Experimental Study of the
Dynamic Properties of Cement- and Lime-Modified Clay Soils Subjected
to Freeze-Thaw Cycles,” Cold Regions Sci. Technol., Vol. 61(1), 2010,
pp. 29–33.
[8] Paige-Green, P., “Recent Developments in Soil Stabilization,” Proceed-
ings of the 19th ARRB Conference, Sydney, Australia, Dec 7–11, 1998,
ARRB Transport Research Ltd, Vermont, Australia.
[9] Bobrowski, I., “Injection of a Liquid Soil Stabilizer into Subgrade Soil,”
Research Report DHT-30, Texas Department of Transportation, Austin,
TX, 1992.
[10] Bell, F. G., “An Assessment of Cement-PFA and Lime-PFA Used to Sta-
bilize Clay-Size Materials,” Bull. Eng. Geol. Environ., Vol. 44(1), 1994,
pp. 25–32.
[11] Saboundjian, S., “Subbase Treatment Using EMC2 Soil Stabilizer,”
Report No. FHWA-AK-RD-02-08, Department of Transportation and Pub-
lic Facilities, Fairbanks, AK, 2002.
[12] Huang, Z. J., Liang, B., and Sun, C. X., “Study of Characteristic of Clay-
crete Clay Soil Stabilizer Improved Soil Under Thawing Condition,”
Journal of Lanzhou Railway University (Natural Science Edition), Vol.
22(6), 2003, pp. 88–90.
[13] Dai, W. T., Chen, Y., and Chen, X., “Test Study on Road Performance of
Soils Stabilized by BS-100 Model Stabilizer in Seasonally Frozen
Region,” Rock Soil Mech., Vol. 29(8), 2008, pp. 2257–2261.
MECHANICAL PROPERTIES
STP 1568, 2013
DOI:10.1520/STP156820130013
Gregory Da Re,1 John T. Germaine,2 and Charles C. Ladd3
Mechanisms Controlling the Initial

Stiffness of Frozen Sand
REFERENCE: Da Re, Gregory, Germaine, John T., and Ladd, Charles C.,
“Mechanisms Controlling the Initial Stiffness of Frozen Sand,” Mechanical
Eds., pp. 97–123, doi:10.1520/STP156820130013, ASTM International,
West Conshohocken, PA 2013.4
ABSTRACT: This paper presents an extensive laboratory testing program
undertaken to study the influence of interface bonding and particle size,
roughness, and shape on the Young’s modulus of ice-saturated frozen sand
in triaxial compression. The program also included the influence of particle
stiffness by testing plastic beads having a much lower modulus than the
quartz-based sand particles. The study used a high-pressure, low-tempera-
ture, automated triaxial compression testing system having an on-specimen
device capable of consistently measuring displacements of less than 0.1 lm,
corresponding to strains of less than 0.0002 %. Very precise temperature
and strain-rate control systems contributed to the reliability of the small strain
measurements at confining pressures up to 12.5 MPa. Experimental findings
show that the Young’s modulus of ice-saturated frozen materials varies signif-
icantly with particle modulus and increases slightly with particle volume frac-
tion but does not change with confining pressure, strain rate, or temperature.
The modulus, however, depends on the ability of the system to transfer inter-
facial shear stresses between the particle and ice matrix in the form of both
adhesion and mechanical interference. In natural frozen sands, the shear
August 20, 2013.
1
Division Chief, Strategy and Innovation, Inter-American Investment Corporation, Washington,
DC 20577, United States of America.
2
Senior Research Associate, Dept. of Civil and Environmental Engineering, MIT, Cambridge,
MA 02139, United States of America.
3
Edmund K. Turner Professor Emeritus, Dept. of Civil and Environmental Engineering, MIT,
Cambridge, MA 02139, United States of America.
4
Jacksonville, FL.
Copyright V
19428-2959.
97
stress is transferred primarily through surface roughness and particle angu-

larity, and consequently the adhesional bond between the ice matrix and the
sand particles is of secondary importance. In contrast, adhesional bonding
dominates in frozen systems composed of relatively smooth spherical par-
ticles. Reinforcement theories for two phase particulate composite materials
can be used to model the Young’s modulus of ice-saturated frozen natural
sand.
KEYWORDS: adhesion, frozen sand, particulate composite, triaxial com-
pression, Young’s modulus
Introduction
Resource utilization and development in cold regions and the increased use
of ground freezing as a means of temporary support during the construction
of tunnels and other underground structures have prompted the need for a
more comprehensive understanding of the mechanical properties of frozen
soils. Although constitutive relations that model stress-strain-time-temperature
behavior have yet to be developed, considerable experimental data on the
strength and deformation behavior of frozen soils have been reported in
the literature [1–8]. Few studies, however, have specifically addressed
the small strain behavior (ea 1 %) that is important for predicting ground
movements during geotechnical construction. Furthermore, most of the
frozen soil testing was performed with equipment that had excessive
compliance and which lacked the internal strain and force measurements
needed to accurately measure the initial stiffness (Young’s modulus) of
frozen soils.
Research at the Massachusetts Institute of Technology (MIT) has charac-
terized the behavior of a typical frozen sand in triaxial compression over a
large range of strains as a function of relative density, confinement, strain
rate, and temperature [9–11]. In addition, advances in technology related to
the measurement of small strains in the triaxial apparatus allowed the initial
stiffness and yielding behavior of frozen sands to be measured with even
greater detail, accuracy, and reliability [12]. These data showed that the
Young’s modulus of frozen Manchester fine sand was essentially independent
of confinement, strain rate, and temperature but increased slightly with sand
density. The research was then extended to explore the effects of other factors
such as interfacial (ice-particle) bonding and particle characteristics, which
have been postulated to affect the initial stiffness and strength of frozen sand.
Investigation of these fundamental parameters aids our understanding of the
mechanisms controlling the stress-strain behavior of frozen sand, which hope-
fully will assist in the development of improved techniques to model its
behavior.
We investigated the physical mechanisms controlling the initial stiffness of
frozen sand by conducting a series of triaxial compression tests on saturated
DA RE ET AL., doi: 10.1520/STP156820130013 99
frozen specimens that varied the interface adhesion and the size, modulus,
roughness, and shape of the particles, as well as the relative density, confine-
ment, strain rate, and temperature. The experimental data are compared with
existing reinforcement theories developed to predict the composite modulus of
two phase systems. An empirical model to predict the Young’s modulus of
frozen sand is then presented.
Experimental Program
Granular Materials
In order to better understand the physical mechanisms responsible for the
small-strain behavior of frozen sand, six different granular materials were used
for frozen test specimens. Each material was chosen based on its ability to help
quantify the importance of a certain variable in determining the initial stiffness
of frozen materials. Table 1 lists the physical characteristics of the six granular
materials used in the testing program, and Fig. 1 shows their grain size
distributions.
Manchester fine sand (MFS) served as the reference material for the testing
program. It is a quartz-based poorly graded fine sand exhibiting angular to sub-
angular grain structure [Fig. 2(a)]. Other properties are described in Refs 10
and 13.
Commercial industrial quartz was selected as a surrogate material for MFS
for tests investigating the role of grain size. This material has larger grains than
MFS, but it has a similar mineralogy, angularity, and coefficient of uniformity
(Cu ¼ D60/D10). Two grades were chosen for testing, 2010 and 2075. The
four-digit notation XXYY indicates that YY % of the material is retained on a
number XX U.S. standard sieve. Therefore, for the 2010 grade, 10 % of the
TABLE 1—Granular materials used for frozen test specimens.
Surface Treatment
Material Gs Ep, GPa d50, mm Cu ¼ D60/D10 Hydrophobic Roughened

a
Manchester fine sand 2.69 75 0.145 2.08 Yes No
2010 industrial quartz 2.65 75a 0.54 2.07 No No
2075 industrial quartz 2.65 75a 1.00 N/A No No
Small glass beads 2.51b 74c 0.54 1.32 Yes Yes
Large glass beads 2.54b 74c 3.00 1.0 Yes Yes
PMMA beads 1.19b 4.5d 0.62 1.63 No No
a
Estimated from Refs 18, 19, and 20.
b
Quoted from the manufacturer.
c
From Ref 21.
d
From Ref 14 and adjusted to 10 C.
FIG. 1—Grain size distributions for Manchester fine sand, industrial quartz,
small glass beads, and PMMA beads.
material is larger than the number 20 sieve (0.850 mm). Figure 2(b) shows the
size, shape, and surface texture of the 2010 industrial quartz particles.
Polymethylmethacrylate (PMMA) beads were chosen to investigate the
effect of particle modulus on the initial stiffness of frozen materials. PMMA is
an amorphous, transparent, colorless, and hydrophobic thermoplastic that
is hard and stiff at room temperature. Its Young’s modulus in compression is
3.3 GPa [14] at 25 C and increases to 4.5 GPa at 10 C. Inspection under a
scanning electron microscope (SEM) [Fig. 2(c)] shows very smooth particles
with a geometry varying from spherical to egg-shaped.
Two different types of glass beads were used to investigate the roles of par-
ticle size, shape, and roughness. The “small” glass beads had a gradation simi-
lar to that of the 2010 industrial quartz and PMMA materials. According to the
manufacturer’s specifications (Ferro Corporation, Cataphote Division, Jackson,
MS), the beads are predominately spherical in shape with not more than
15 % irregularly shaped particles and not more than 3 % angular particles.
Figure 2(d) shows their size and smooth surface texture. The “large” glass
beads have a uniform diameter of 3 mm and are composed of high-quality soda
lime glass.
Surface Treatments
In order to reduce ice adhesion, a hydrophobic treatment was developed and
applied to both the MFS and the glass beads. Hydrophobic surfaces were read-
ily created through silation, a process that involves the replacement of surface
DA RE ET AL., doi: 10.1520/STP156820130013 101
FIG. 2—SEM images of (a) Manchester fine sand, (b) 2010 industrial quartz,
(c) PMMA beads, (d) small glass beads, (e) chemically roughened small glass
beads, and (f) mechanically roughened large glass beads.
hydroxyl molecules with large non-polar functional groups at the molecular

level through treatment with a silating agent such as Silquest A-137 (octyl-trie-
thoxysilane). For the hydrophobic treatment procedure, we combined 1 kg of
material in a 5 % solution of octyl-triethoxysilane and ethanol and allowed it to
react in a closed container for 24 h. After treatment and solution removal, the
material was oven-dried for two to three days. SEM inspection showed no
change in the surface roughness of the treated materials.
FIG. 3—MIT automated high-pressure, low-temperature triaxial system. A, tri-

axial cell; B, load frame; C, actuators (PVCs); D, motor control box; E, perso-
nal computer; F, cell fluid pump; G, power supply; H, Voltmeter.
Surface roughening techniques were applied to the two sizes of smooth

glass beads. For the small glass beads, roughening was effectively accom-
plished by acid-etching for 30 min with a 48 % hydrofluoric acid solution
diluted 50/50 with distilled de-ionized water. Figure 2(e) shows the resulting
roughened glass beads. Some of these roughened beads were then subsequently
treated to make them hydrophobic using the procedure already described. The
surfaces of the large glass beads were also roughened, but by lightly grinding
the beads between two abrasive corundum stones. The resulting level of sur-
face roughness is shown in Fig. 2(f). A sample of the large glass beads also
received a hydrophobic treatment.
Testing Equipment
Frozen specimens were tested in an MIT-designed and -built high-pressure,
low-temperature triaxial testing apparatus featuring personal-computer-based
closed-loop electrohydraulic feedback control. The apparatus consists of three
main components: a triaxial cell, a load application and control system, and a
small-strain measurement system. Figure 3 shows a schematic of the entire
system.
The triaxial cell, shown in Fig. 4, consists of a steel chamber (20-MPa
capacity) that mates to the triaxial base and contains an enlarged base pedestal
that is designed to accommodate at least 15 % radial deformation of the
DA RE ET AL., doi: 10.1520/STP156820130013 103
FIG. 4—High-pressure, low-temperature triaxial chamber. A, triaxial cham-

ber; B, load cell; C, alignment device; D, triaxial base; E, pressure sensor; F,
test specimen; G, thermistors; H, external LVDT; I, internal LVDTs.
specimen during shear. Because this system is dedicated to testing frozen

specimens, it utilizes an oversized floating top cap together with a hardened
steel ball to ensure concentric loading. Specimen stability during shear was
also increased through the use of small 1.5-mm-diameter pins in the center of
both the base pedestal and top cap. Axial load is applied to the specimen via a
hardened steel piston attached to an internal 45-kN (10 000-lb) shear-beam
load cell. Piston movement is continuously monitored by an externally
mounted linear variable differential transformer (LVDT). Two thermistors
located on the inside wall of the triaxial chamber measure the temperature of
the cell fluid (Dow-Corning 200 silicone oil) at the top and bottom of the
specimen. A fluid circulation system utilizing a small external magnetically

driven high-pressure pump ensures that the chamber fluid is kept at a constant
temperature and that the temperature gradient over the specimen is less than
0.1 C.
A triaxial state of stress is applied to the specimen through two MIT-
designed actuators (pressure-volume controllers) that are capable of controlling
either fluid pressure or volume. One actuator applies the confining pressure to
the test specimen and also monitors cell-fluid volume changes. The other actua-
tor applies the vertical axial force to the specimen through an 89-kN (10-ton)
hydraulic ram in conjunction with a high-capacity benchtop loading frame.
The complete testing system, aside from the computer and motor control unit,
operates inside an environmental enclosure (to dampen temperature fluctua-
tions) within MIT’s Low Temperature Testing Facility, which can achieve test-
ing temperatures as low as 40 C.
The small-strain measurement system uses two miniature LVDTs set dia-
metrically opposite each other on two yokes that clamp to the specimen. Their
averaged output provides displacement measurements up to 2 % axial strain;
thereafter the displacement is computed from the external LVDT. The LVDTs
can operate at temperatures ranging from 55 C to 150 C in fluids pressurized
to 21 MPa (3000 psi). A highly stable, low-noise, MIT-designed signal
conditioning system completes this design. The resulting system easily
resolves displacements of less than 0.1 lm, which corresponds to strains of less
than 0.0002 % for standard-sized (3.56 cm by 7.62 cm) triaxial specimens.
Reference 12 provides performance details of the equipment.
Testing Procedures
Triaxial tests were performed on frozen specimens within two weeks of their
preparation. Reference 15 gives a detailed account of the preparation proce-
dure. All specimens were formed using a multiple sieve pluviation technique
that produces extremely homogeneous specimens having a wide range of rela-
tive densities. Lubricated end conditions, consisting of thin ice caps, were used
in all cases to reduce lateral friction. Latex membranes sealed the specimen
during testing.
Each specimen equilibrated at the desired testing temperature for 24 h prior
to shear. This was performed under test cell pressure in order to establish a
cell-fluid leakage rate for the test, which was subsequently used to correct volu-
metric strain measurements during shear. The leakage rate varied from test to
test depending on the test temperature and level of confinement. Tests on MFS
were conducted at nominal temperatures that varied from 5 C to 25 C in
5 C increments. However, most tests for this study were run at 5 C and
10 C. Confining pressures ranged from 0.1 MPa to 10 MPa.
DA RE ET AL., doi: 10.1520/STP156820130013 105
The axial load was applied at a constant rate of displacement using the
on-specimen axial deformation as the feedback source for a digital propor-
tional-integral-derivative control algorithm. This program used two strain rates,
a “slow” rate (3 106 s1) and a “medium” rate (3.5 105 s1). Faster rates
were not used because of limitations in the data-acquisition system. As previ-
ously described, the computation of axial strain was made using the average of
the strains recorded by the two internal LVDTs up to 2 % strain and then
switched to the external LVDT at larger strains. Most tests were terminated at
about 15 % axial strain.
A rating system was developed, similar to that described in Ref 10, to
assess the quality of the calculated Young’s modulus based on the agreement
between the two internal LVDTs. Each shear was assigned one of four ratings:
poor, fair, good, and excellent. This rating was based on the difference in
slopes between the two transducers during the very early linear portion of the
stress-strain curve, typically less than 0.0025 % axial strain. Only tests with rat-
ings of fair, good, and excellent were used for analyses. Although tests with
poor ratings did not yield reliable values of modulus, they still provided mean-
ingful data at larger strains.
Figure 5 shows representative stress-strain data for a specimen of frozen
MFS that was tested as part of this research. Figure 5(a) shows the complete
stress-strain curve. Figure 5(d) shows that the measuring system was capable
of resolving the strains necessary to capture the initial linear behavior, which
for this test extended to approximately 0.002 % strain. Figures 5(a), 5(b),
and 5(d) represent the average strain of the two LVDTs, and Figs. 5(c) shows
the results for each LVDT transducer. This comparison shows that the two
measurements of the initial stiffness differed by less than 5 % - that is, the
agreement between the two LVDTs was excellent. As the strain level
increased, the two transducers tended to deviate somewhat, but their average
was generally accurate up until the point where the transition was made to the
external LVDT.
Experimental Results
Effects of Density, Confinement, Strain Rate, and Temperature

The variation of Young’s modulus with density, confinement, strain rate, and
temperature has been well established for frozen MFS by two previous experi-
mental programs [9,11]. As described in Ref 10, the modulus of frozen MFS
was found to be relatively insensitive to the effects of these variables over the
ranges considered, except for a small increase with increasing density. This
relationship is shown in Fig. 6, which presents the variation of Young’s modu-
lus with the volume fraction (Vp ¼ particle volume/total volume) for frozen
FIG. 5—Typical triaxial shear data on frozen Manchester fine sand at

various strain resolutions: (a) 0% to 20%; (b) 0% to 2%; (c) 0.00% to 0.02%;
(d) 0.00% to 0.005% (FRS182: Dr ¼ 49.8%; rc ¼ 0.5 MPa; T ¼ 5 C;
e_ ¼ 3.0 106 s1) [12].
specimens of MFS for temperatures ranging from 5 C to 25 C for all con-
fining pressures and strain rates.
As illustrated in Fig. 7, the addition of data for the other quartz-based sys-
tems with a similar particle modulus tested during this program had little effect
on the regression statistics, and this confirms the above conclusion. However,
from a practical viewpoint, the modulus of frozen particulate materials that
have their particles in contact with one another may be considered constant
(that is, increasing little with particle density).
The observed temperature and strain rate independence of the modulus of
frozen sand is interesting, because it is generally accepted that the modulus of
DA RE ET AL., doi: 10.1520/STP156820130013 107
FIG. 6—Variation of Young’s modulus of frozen Manchester fine sand with

volume fraction at varying temperatures for all confining pressures (rc ¼ 0.1 to
10 MPa) and strain rates (_e ¼ 3.0 106 to 5.0 104 s1) (updated from
Ref 10).
polycrystalline ice, a principal component of frozen sand, increases with
decreasing temperature and with faster loading rates [16]. However, for the
ranges investigated for frozen MFS, any effects are probably small and thus
masked by the scatter in the experimental data. Although only a small number
FIG. 7—Variation of Young’s modulus of frozen quartz-based systems with vol-

ume fraction at varying temperatures (T ¼ 5 C to 25 C), confining pres-
sures (rc ¼ 0.1 to 10 MPa), and strain rates (_e ¼ 3.0 106 to 5.0 104 s1).
TABLE 2—Summary of the Young’s modulus obtained for each material type at 5 C and 10 C at
varying volume fractions, confining pressures (rc ¼ 0.1 to 10 MPa), and strain rates (_e ¼ 3.0 106 to
3.5 105 s1).
Mean 6 S.D., GPa

Material Type Ep, GPa T ¼ 5 C T ¼ 10 C
Manchester fine sand 75 26.8 6 7.3 (n ¼ 10) 26.3 6 1.8 (n ¼ 6)
Industrial quartz 75 22.9 6 1.5 (n ¼ 3) 26.8 6 5.0 (n ¼ 19)
PMMA beads 4.5a 4.70 6 0.3 (n ¼ 7) 5.10 6 0.3 (n ¼ 11)
Small glass beads 74 N/A 26.8 6 0.9 (n ¼ 5)
Large glass beads 74 N/A 25.2 (n ¼ 1)
Polycrystalline ice N/A 4.90 6 0.4 (n ¼ 4) 5.30 6 0.3 (n ¼ 4)
a
Computed at 10 C using a temperature correction given in Ref 21.
of tests were performed on specimens of fine-grained polycrystalline ice (see

Table 2), no rate dependence was observed, and only a small temperature effect
was noticed. Furthermore, the measured modulus at 10 C (5.3 6 0.3 GPa,
number of tests n ¼ 4) was substantially lower than the theoretical value
of approximately 9 GPa [16]. This indicates that the testing procedure, which
subjects the test specimens to the applied hydrostatic stress for 12 h before
shearing in order to achieve temperature equilibrium, is probably measuring
the relaxed modulus, which can be as low as 5 GPa [17]. These lower values
are usually attributed to time-dependent delayed elastic strains that manifest
themselves from the moment the load is first applied. Therefore, it is not sur-
prising that specimens of frozen sand tested in a similar manner also showed
no rate or temperature dependence for the modulus.
Effect of Particle Modulus

The effect of particle modulus is shown in Fig. 8, which presents data on each
of the particulate systems tested as a function of volume fraction at 5 C and
10 C for all confining pressures and strain rates investigated. Table 2 summa-
rizes the average values of modulus for each of the materials and temperatures
shown in Fig. 8. MFS and industrial quartz, both predominantly quartz-based
materials, were assumed to have a particle modulus of 75 GPa based on their
similarity to sands used in other programs [18–20]. Both types of glass beads,
being composed of soda-lime glass, were assumed to have a particle modulus
of 74 GPa [21].
Figure 8 and Table 2 clearly show that all the frozen quartz-based and glass
systems had the same composite modulus of approximately 26 GPa. Further-
more, the PMMA test results show that particle modulus had a profound effect
on the composite modulus, which dropped to approximately 5 GPa, or
DA RE ET AL., doi: 10.1520/STP156820130013 109
FIG. 8—Effect of particle modulus on Young’s modulus of frozen systems at

5 C (solid symbols) and 10 C (open symbols) at varying confining pres-
sures (rc ¼ 0.1 to 10 MPa) and strain rates (_e ¼ 3.0 106 to 3.5 105 s1).
essentially that of ice and only slightly larger than that of the particle modulus
(4.5 GPa).
Effect of Particle Grain Size

Evaluating particle size effects necessitates the use of materials that have simi-
lar particle moduli, as the particle modulus can have a significant effect on the
overall composite modulus. This investigation used two sizes of industrial
quartz (i.e., 2010 and 2075) and the two sizes of glass beads for comparison
with the much finer grained MFS. The resulting data, summarized in Table 3,
do not show a consistent influence of particle grain size on Young’s modulus.
TABLE 3—Summary of the variation of Young’s modulus with particle size at 5 C and 10 C at
varying volume fractions, confining pressures (rc ¼ 0.1 to 10 MPa), and strain rates (_e ¼ 3.0 106 to
3.5 105 s1).
Mean 6 S.D., GPa
Material Type Grain Size d50, mm T ¼ 5 C T ¼ 10 C

Manchester fine sand 0.145 26.8 6 7.3 (n ¼ 10) 26.3 6 1.8 (n ¼ 6)
2010 industrial quartz 0.54 22.9 6 1.5 (n ¼ 3) 26.8 6 5.0 (n ¼ 19)
2075 industrial quartz 1.00 20.5 6 3.5 (n ¼ 4) N/A
Small glass beads 0.54 N/A 26.8 6 0.9 (n ¼ 5)
Large glass beads 3.00 N/A 25.2 (n ¼ 1)
This observation agrees with the results of Ref 20, which state that particle size
has little effect on the modulus of glass-filled resins.
If a direct correlation is assumed between the particle size and the resulting
ice grain size, then the insensitivity of the composite modulus to particle size
would indicate that the modulus of the ice matrix is also independent of grain
size. This contradicts the findings of Ref 22, which reports a 40 % decrease in
the true elastic modulus of polycrystalline ice with decreasing ice grain size
(from 5 mm to 1.7 mm). It is believed, however, that this apparent contradiction
might be invalid for three reasons. Firstly, this method of testing incurs sub-
stantial delayed elastic strains that mask the true elastic response of ice. Sec-
ondly, the presence of solid particles greatly increases the composite modulus,
which might mask changes in the modulus of ice with decreasing grain size.
Thirdly, the grain size of ice in all of the specimens, except perhaps within the
large glass beads, is at least an order of magnitude less than the grain sizes
tested by Cole [22].
Effect of Interface Adhesion

The importance of interface adhesion was investigated by applying a hydro-
phobic treatment to the MFS and glass beads such that they would hopefully
become icephobic without any alteration of their physical characteristics.
Inspection of treated particles with an SEM revealed no noticeable changes in
surface roughness that might affect the role of mechanical interference between
the ice matrix and the particles. MFS and the two sizes of glass beads were
chosen because they represent extremes in particle size, roughness, and shape.
Figure 9 plots the values of Young’s modulus with the volume fraction for
frozen MFS and both glass bead materials in their treated and untreated
states at 10 C. A difference can be seen when comparing the data for MFS
FIG. 9—Effect of interface adhesion on Young’s modulus at 10 C at varying

volume fractions and strain rates for (a) Manchester fine sand and (b) small
and large glass beads.
DA RE ET AL., doi: 10.1520/STP156820130013 111
TABLE 4—Summary of the effect of interface adhesion on the Young’s modulus of Manchester fine
sand and glass bead systems at 10 C at varying volume fractions, confining pressures (rc ¼ 0.1 to
10 MPa), and strain rates (_e ¼ 3.0 106 to 3.5 105 s1).
Mean 6 S.D., GPa
Material Type Regular Hydrophobic

Manchester fine sand 26.3 6 1.8 (n ¼ 6) 21.7 6 4.0 (n ¼ 10)
Small glass beads 26.8 6 0.9 (n ¼ 5) 5.70 6 0.2 (n ¼ 4)
Large glass beads 25.2 (n ¼ 1) 12.0 6 0.8 (n ¼ 2)
[Fig. 9(a)]. The statistical summary in Table 4 shows that the hydrophobic
treatment reduced the modulus of MFS by 17.5 % and significantly increased
the scatter in the data. These data indicate that either the hydrophobic treatment
did not entirely remove the particle-matrix adhesional strength for MFS, or the
adhesional strength is far less important to the modulus of frozen MFS.
In contrast to the data on MFS, the hydrophobic treatment reduced the
modulus of the frozen small glass beads by almost 80 % and by over 50 % for
the frozen large glass beads [Fig. 9(b) and Table 4). The single test performed
on untreated large glass beads was considered fairly reliable because its results
were consistent with the other quartz-based moduli. A similar modulus reduc-
tion was expected for both glass systems, and the resulting modulus for the
hydrophobic large glass beads (12 GPa) is thought to be due to a less efficient
surface treatment. The data on small glass beads, however, essentially confirm
that the hydrophobic treatment can make quartz surfaces icephobic, as the
modulus of the treated small beads dropped to that measured for polycrystal-
line ice (Table 2).
Effect of Particle Roughness and Shape

The results in Fig. 9 initiated an investigation into the importance of particle
roughness and particle shape to determine whether the composite modulus of
hydrophobic glass beads (i.e., systems lacking ice-particle interface adhesional
strength) would be increased by surface roughening. The first tests were per-
formed on acid-etched small glass beads [Fig. 1(e)]. The results of tests con-
ducted at 10 C using these roughened and subsequently hydrophobically
treated glass beads are shown in Fig. 10 and summarized in Table 5. Particle
roughening led to a 25 % higher modulus.
Large glass beads (3 mm) were individually mechanically roughened
between two corundum grinding stones in order to roughen their surface.
Figure 2(f) indicates that this technique effectively produced a consistent,
roughened surface. This technique was not successful with the small glass
beads because they had a tendency to become pulverized during the grinding
process. After applying the hydrophobic treatment, the modulus of the
FIG. 10—Effect of particle roughness on Young’s modulus of small and large

glass bead systems at 10 C and moderate strain rate (_e ¼ 3.5 105 s1) at
varying volume fractions and confining pressures (rc ¼ 0.1 to 10 MPa). Note:
larger symbols denote large glass beads.
roughened large glass beads increased markedly from 12.0 to 21.0 GPa. This
very large increase shows that the degree of surface roughness is very impor-
tant in systems with little adhesional bonding between the ice and the particles.
It also suggests that particle roughness, relative to particle shape and angular-
ity, is more important in controlling the composite modulus, given that the
round, yet roughened, icephobic large beads had a modulus similar to that of
hydrophobic MFS and nearly equal to that of untreated quartz based systems.
Discussion
The experimental results indicate that of the eight variables investigated
(volume fraction, confinement, strain rate, temperature, particle size, particle
modulus, particle roughness/shape, and interface adhesion), only volume
TABLE 5—Summary of the effect of particle roughness on the Young’s modulus of glass bead systems
at 10 C and moderate strain rate (_e ¼ 3.5 105 s1) at varying volume fractions and confining
pressures (rc ¼ 0.1 to 10 MPa).
Mean 6 S.D., GPa
Material Type Regular Hydrophobic Hydrophobic Roughened

Small glass beads 26.8 6 0.9 (n ¼ 5) 5.70 6 0.2 (n ¼ 4) 7.10 6 1.0 (n ¼ 9)
Large glass beads 25.2 (n ¼ 1) 12.0 6 0.8 (n ¼ 2) 21.0 6 0.0 (n ¼ 2)
DA RE ET AL., doi: 10.1520/STP156820130013 113
fraction (density), particle modulus, interface adhesion, and particle roughness/

shape are important to the Young’s modulus of frozen particulate systems. The
other variables (particle size, level of confinement, strain rate, temperature)
have been shown to be of minor importance over the ranges investigated,
which supports the conclusions of Ref 10. However, strain rate effects might
become important for dynamic applications such as those involving earthquake
loadings or geophysical investigations. Similarly, temperature considerations
might become significant in problems involving creep and the behavior of
frozen soils at higher temperatures.
One key research objective was to investigate the effects of interface adhe-
sion and particle roughness and shape on the modulus of frozen systems. Most
of the models that exist to describe the behavior of particulate composites
assume that perfect bonding exists between the two phases and that this adhe-
sional strength is sufficient to enable composite action. Particle roughness and
shape, in contrast, have received little attention in the composite material liter-
ature. Although these factors can affect interparticle friction and the develop-
ment of strength in unfrozen granular materials, the resistance of the granular
skeleton has little to no effect on the modulus of frozen material. For example,
drained triaxial compression tests on unfrozen MFS consolidated to high
stresses (up to 10 MPa) show an initial stiffness that is an order of magnitude
less than the value of E ¼ 26 GPa that this and previous experimental programs
reported for frozen MFS.
FIG. 11—Interaction diagram for the influence of mechanical interference and

interface adhesion on the composite Young’s modulus of frozen quartz-based
systems at 10 C. Note: open and solid symbols denote regular and hydropho-
bic systems, respectively.
Figure 11 presents a hypothesis to explain how particle roughness/shape

and interface adhesion combine to affect the composite modulus. Particle
roughness/shape values were objectively assigned based on SEM image analy-
sis and are denoted here by the symbol / on a scale ranging from 0 to 1.
Increasing / reflects greater mechanical interference at the ice-particle inter-
face that resists shearing (relative movement) between the ice matrix and the
particles. In turn, increasing particle roughness, and perhaps angularity, is
thought to enhance this mechanical interference. At low particle roughness
(i.e., with round smooth spheres), the initial stiffness is very much dependent
on the degree of adhesional bonding at the interface, whereas at higher values
of particle roughness the importance of adhesion decreases because of the
increase in mechanical interference generated between the surface of rough
particles and the ice matrix. This hypothesis uses the concept of contours of the
degree of adhesional bonding. Although only the bounds (full and zero adhe-
sion) can be estimated from the experimental data, it is reasonable to assume
that different degrees of adhesion can exist depending on the effectiveness of
the treatment used to modify the bond strength. For example, the higher values
for modulus measured in the tests on the large hydrophobic glass beads indi-
cate that the hydrophobic treatment did not completely eliminate the ice-
particle adhesional bond.
Development of the Initial Stiffness in Frozen Sand

It is well known that the mechanical properties (stiffness, strength, etc.) of two
phase composite materials result from some combination of the mechanical
properties of the individual constituents. For the initial stiffness of frozen mate-
rials, this means that the resulting composite modulus must be that of ice modi-
fied by the presence of rigid inclusions. The two extreme cases are all-matrix
material (polycrystalline ice) and all-quartz (for frozen quartz-based sands).
Therefore, the lowest modulus of frozen sand must be that of ice, and the high-
est possible value must be that of the particle. The difficulty lies in predicting
the variation of the composite modulus between these endpoints as a function
of the amount of inclusion material.
Having established that the composite modulus is dependent on the particle
and matrix moduli and the volume fraction of particles, it is necessary to
investigate the additional importance of interfacial adhesion and mechanical
interference that provides the coupling between phases that is required for
stress transfer between the particles and the ice matrix. It is well established
that improving adhesion at the interface increases the fracture strength [23,24]
and yield strength [25,26] of various composite materials. However, it is
not entirely clear how interfacial adhesion affects the Young’s modulus of
particulate-filled composites, especially when particles of different roughness
and shape are considered.
DA RE ET AL., doi: 10.1520/STP156820130013 115
In untreated frozen quartz-based systems such as those investigated here,

one expects that the ice-particle bonding is predominately due to a combination
of the adhesive strength of ice, which is known to be very strong [27], and the
mechanical interference between the ice matrix and the particle surfaces.
From the experimental results, it is clear that in cases when the mechanical
interference is probably small (smooth spherical particles), the strength of the
adhesional bond between particles and ice matrix is very important. This was
shown for the small glass bead system, in which the modulus decreased to that
of ice after the particles had been treated to make them hydrophobic. In con-
trast, in cases when the mechanical interference is thought to be much higher
(/ ¼ 0.9), as in the system composed of MFS, the transfer of internal shear
stresses at the interfaces is far less dependent on the adhesional bond between
the particles and ice matrix (the decrease in composite modulus was less than
20 % after hydrophobic treatment). A similar result was reported by Ahmed
and Jones [20], who found that in the absence of chemical adhesion between
particles and the matrix, semi-angular-shaped sand-filled resin composites
have higher moduli (by some 50 %) than their spherical glass bead counter-
parts. However, the authors surmise that mechanical interference caused by
particle roughness, or angularity, and not particle shape per se might be respon-
sible for the higher modulus. This is based on the tests conducted on roughened
glass beads that were still very much spherical in shape after having undergone
chemical or mechanical surface roughening. The fact that increasing surface
roughness for the same particle shape in the absence of, or with a greatly
reduced, adhesional bond leads to an increased modulus indicates that the
Young’s modulus is dependent on ice-particle mechanical interference that is
caused by particle roughness.
Put another way, the particle-ice matrix adhesional bond is less important
in frozen natural sands than in frozen smooth spherical particles because natu-
ral sands have rough, often angular surfaces. It is this high degree of particle
roughness that effectively provides mechanical interference and thus tends to
make up for a lack of adhesional bonding between the ice and the particle sur-
face. This means that even if the surface of a rough particle is made hydropho-
bic (for systems involving ice), the modulus can be little affected because of
the mechanical coupling that already exists.
Composite Material Modeling

The similarities between the factors controlling the modulus of frozen sand and
those controlling the modulus of particulate-filled composites suggest that fro-
zen sand can be modeled as a two phase composite material. A number of theo-
ries have been developed to describe the effect of particles in a matrix, but
most assume perfect bonding between the two phases and conditions of equal
stress or equal strain in the individual constituents. As a result of these assump-

tions, discrepancies arise between theoretical predictions and experimental
results. For the comparisons presented here, only those models that assume per-
fect bonding are reviewed. No attempt is made to fit the experimental data for
the systems having undergone some type of surface treatment to reduce adhe-
sion (i.e., hydrophobic MFS and glass beads).
The simplest possible approach for predicting the modulus of a two phase
system is to assume that the materials are coupled either in parallel or in series.
For the parallel arrangement, in which conditions of uniform strain are
assumed, an upper bound solution is given by
Ec ¼ Ep c þ Em ð1 cÞ (1)
For the series arrangement, in which conditions of uniform stress in the two
phases are assumed and a lower bound results,
1 c ð1 cÞ
¼ þ (2)
Ec Ep Em
where:
c ¼ volume fraction (Vp) of inclusions, and
Ec, Ep, and Em ¼ modulus of the composite, particles, and matrix (ice),
respectively.
In an attempt to account for the actual stress-strain distribution in a two
phase system, Ref 28 proposes the following relation, which is basically a
weighted average of Eqs 1 and 2:

1 2Z c ð1 cÞ 2Z 1
¼ 1 þ þ (3)
Ec p Ep Em p Ep c þ Em ð1 cÞ
Reference 28 recommends Z ¼ 0.785 based on tests on concrete. By adjusting

the weighting factor, one can achieve an optimal fit. However, without a
rational basis for adjusting this parameter, it is not possible to use this model in
a predictive sense.
The theoretical curves predicted by these three models are compared in
Fig. 12 with the data for the quartz-based materials (i.e., MFS, industrial
quartz, and glass beads) at 10 C. The predictions used a particle modulus
(Ep) of 75 GPa and the experimentally measured polycrystalline ice matrix
modulus (Em) of 5.3 GPa. The upper and lower bounds from Eqs 1 and 2 are
widely spaced and of little predictive value, whereas the estimation by Eq 3
gives an intermediate solution that, although somewhat high, provides an
acceptable fit to the data.
DA RE ET AL., doi: 10.1520/STP156820130013 117
FIG. 12—Comparison of series, parallel, and Hirsch [28] models to experi-

mental data for quartz-based systems at 10 C at varying confining pressures
and strain rates. *, Manchester fine sand; ^, 2010 industrial quartz; 4, small
glass beads; 5, large glass beads.
A number of other solutions for predicting the composite elastic modulus,

which are neither upper nor lower bounds, have also been derived [18,29–31].
These models assume that the macroscopic stress and strain can be reproduced
in some average sense on a typical unit volume of a specified geometry (usu-
ally cubic) that contains a single inclusion of the stiffer phase. Because most of
these models assume that the inclusion is cubic, the different formulations are
a result of the assumptions made in specifying the state of stress or strain at the
boundary of the inclusion.
Following the work of Refs 29 and 32, Hashin and Shtrikman [33] derived
improved upper and lower bounds for the modulus of a two phase composite
by invoking variational principles in the linear theory of elasticity without
making any assumptions about phase geometry. The Hashin–Shtrikman
formulation requires the calculation of the upper and lower limits of the bulk
modulus K and the shear modulus G given by the following equations:
c
KL ¼ Km þ (4)
1 3ð1 cÞ
þ
Kp Km 3Km þ 4Gm
ð1 cÞ
KU ¼ Kp þ (5)
1 3c
þ
Km Kp 3Kp þ 4Gp
c
GL ¼ Gm þ (6)
1 6ðKm þ 2Gm Þð1 cÞ
þ
Gp Gm 5Gm ð3Km þ 4Gm Þ
ð1 cÞ
GU ¼ Gp þ (7)
1 6ðKp þ 2Gp Þc
þ
Gm Gp 5Gp ð3Kp þ 4Gp Þ
9KG
E¼ (8)
3K þ G
As Fig. 13 shows, these bounds provide a much closer approximation than the
bounds given by Eqs 1 and 2 for the same values of Ep and Em presented previ-
ously and by using Poisson’s ratios for quartz and ice of 0.25 and 0.31, respec-
tively, in the computation of the bulk and shear moduli. The spacing of the
Hashin–Shtrikman bounds, however, is still dependent on the modulus ratio of
particle to matrix, which means that they are still rather widely spaced for the
quartz-based systems. Nevertheless, Fig. 13 shows that this formulation bounds
FIG. 13—Comparison of Hashin-Shtrikman [33], series, parallel, and BNC

models [34] to experimental data at 10 C at varying confining pressures and
strain rates for quartz and PMMA-based systems. *, Manchester fine sand; ^,
2010 industrial quartz; 4, small glass beads; 5, large glass beads; h, PMMA
beads.
DA RE ET AL., doi: 10.1520/STP156820130013 119
the scatter in the quartz system data quite nicely. The fit to the data for the
frozen PMMA is very good, as should be expected given the small difference
in modulus between the PMMA and the ice matrix (Ep ¼ 4.5 GPa and
Em ¼ 5.3 GPa at 10 C) and the fact that the spacing of the bounds is con-
trolled by the modulus ratio of the two phases.
In contrast to Hashin and Shtrikman’s theoretical treatment of the compos-
ite modulus, the Bache and Nepper-Christensen (BNC) model [34] described
by Eq 9 is completely empirical. It gives the approximate mean of the Hashin
and Shtrikman bounds and, although developed for concrete, provides a very
satisfactory fit to the data in Fig. 13.
c
Ep
Ec ¼ Em (9)
Em
The simplicity of the BNC model makes it the most attractive model for pre-
dicting the composite modulus of saturated frozen sands.
Summary and Conclusions

An automated high-pressure, low-temperature triaxial compression apparatus
with accurate on-specimen strain measurements was used to investigate the
mechanisms controlling the initial stiffness (Young’s modulus E) of ice-
saturated frozen granular materials. Prior research [10] has shown that the
modulus of frozen Manchester fine sand (MFS) is independent of confinement
(rc ¼ 0.1 to 10 MPa), strain rate (_e ¼ 3 106 to 5.0 104 s1), and tempera-
ture (T ¼ 10 C to 25 C) but increases slightly with increasing density or
particle volume fraction (Dr ¼ 20 % to 100 % or Vp ¼ 0.55 to 0.65). This
research investigated the influence of particle size, particle roughness/shape,
and degree of ice-particle adhesion of quartz-based materials on the resulting
composite modulus. It also tested spherical PMMA beads to determine the
effect of particle modulus.
Evaluation of the collective data and application of simple composite
material models indicate that the composite modulus primarily depends on the
modulus of the particles (Ep) and of the ice matrix (Em) and the particle volume
fraction (Vp or c). However, the composite modulus also depends on the effec-
tiveness of the transfer of shear stresses at the interface between the particles
and the ice matrix. This shear stress transfer at the particle-ice interface is
thought to have two components: one in the form of adhesional bonding of the
ice to the particles, and the other in the form of a mechanical interference that
resists shearing (relative movement) between the ice matrix and the particles.
In turn, increasing roughness of the particle surface, and perhaps angularity, is
believed to enhance the mechanical interference.
In cases when the mechanical interference is small, the strength of the

adhesional bond is very important for the transfer of shear stress. This is shown
by the data on the small glass beads (i.e., smooth spherical particles). The mod-
ulus decreased from 26.8 GPa to only 5.7 GPa, near to that measured for pure
ice, after the beads had been treated to make them hydrophobic. In contrast, in
cases when the mechanical interference is high, the strength of the adhesional
bond is far less important. This is shown by the data on MFS (i.e., rough angu-
lar particles): hydrophobic treatment caused the modulus to decrease by less
than 20 %, that is, from 26.3 GPa to 21.7 GPa. Another indication of the
importance of surface roughness is shown by the increase in modulus, from
12.0 GPa to 21 GPa, caused by roughening the surface of the hydrophobic large
glass beads. Surface roughening of the small glass beads was far less effective,
however, with the increase being only from 5.7 GPa to 7.1 GPa.
The similarity between the factors controlling the modulus of frozen sand
and those controlling the modulus of particulate-filled composites in general
suggest that frozen sand can be modeled as a two phase composite system.
After evaluating a number of particulate composite material models, all of
which assume perfect “bonding” between the two phases, the empirical BNC
model was found to be the most suitable for predicting the composite modulus
(Ec) of frozen materials.
c
Ep
Ec ¼ Em (10)
Em
where:
Ep and Em ¼ moduli of the particles and the ice matrix, respectively, and
c ¼ Vp ¼ volume fraction of the particles.
The five quartz-based frozen materials that were tested had large variations
in particle size, shape, uniformity, and surface roughness (Table 1, Figs. 1
and 2), yet they all had essentially the same Young’s modulus of 26 6 5 GPa
at Vp ¼ 0.60 6 0.05 (Fig. 7). For these materials, Eq 10 predicts a composite
modulus of 23 to 30 for a particle modulus of 75 GPa and a matrix modulus of
5.3 GPa (the measured value for polycrystalline ice at 10 C).
The empirical Eq 10, which was developed for concrete, obviously does
not account for variations in the ice-particle adhesional bond or the degree of
mechanical interference at the ice-particle interface. These considerations,
however, should not be important for most saturated frozen natural sands, in
which the typical particle angularity and roughness should make up for any
lack of adhesional bond. Regarding the appropriate value for the modulus of
ice, a range of Em ¼ 5 to 9 GPa is suggested, with the higher value being more
appropriate for dynamic problems such as for earthquake loadings. Similarly,
temperature considerations also might prove to be significant in problems
involving the behavior of frozen soils at high temperatures (e.g., permafrost).
DA RE ET AL., doi: 10.1520/STP156820130013 121
Acknowledgments
The funding for this research was provided by the Army Research Office under
Grant Nos. DAAH04-96-1-0042, DAAL03-92-G-0226, and DAAL03-89-
K0023. The writers are extremely grateful for this support and the outstanding
technical guidance of Dr. David M. Cole of CRREL. The writers also acknowl-
edge the extensive contributions of Drs. G. R. Andersen and C. W. Swan in
developing the specimen preparation and freezing protocols and obtaining
most of the data presented on frozen MFS. They also thank Professor M. C.
Santagata and S. Rudolph for their assistance in the design, fabrication, and
testing of the small-strain measurement system.
References
[1] Chamberlain, E., Groves, C., and Perham, R., “The Mechanical Behavior
of Frozen Earth Materials under High Pressure Triaxial Test Conditions,”
Geotechnique, Vol. 22, No. 3, 1972, pp. 469–483.
[2] Alkire, B. D. and Andersland, O. B., “The Effect of Confining Pressure
on the Mechanical Properties of Sand-ice Materials,” J. Glaciol., Vol. 12,
No. 66, 1973, pp. 469–481.
[3] Sayles, F. H., “Triaxial and Creep Tests on Frozen Ottawa Sand,” Pro-
ceedings of the Second International Permafrost Conference, Yakutsk,
Russia, July 13–28, 1973, International Permafrost Assoc., Potsdam,
Germany, pp. 384–391.
[4] Parameswaran, V. R. and Jones, S. J., “Triaxial Testing on Frozen Sand,”
J. Glaciol., Vol. 27, No. 95, 1981, pp. 147–155.
[5] Orth, W., “Deformation Behavior of Frozen Sand and Its Physical Inter-
pretation,” Proceedings of the Fourth International Symposium on
Ground Freezing, Sapporo, Japan, Aug 5–7, 1985, Hokkaido Univ. Press,
pp. 245–253.
[6] Shibata, T., Adachi, T., Yashima, A., Takahashi, T., and Yoshioka, I.,
“Time-dependence and Volumetric Change Characteristics of Frozen
Sand under Triaxial Stress Conditions,” Proceedings of the Fourth Inter-
national Symposium on Ground Freezing, Sapporo, Japan, Aug 5–7,
1985, Hokkaido Univ. Press, pp. 173–179.
[7] Youssef, H., “Volume Change Behavior of Frozen Sands,” J. Cold Reg.
Eng., Vol. 2, No. 2, 1988, pp. 49–64.
[8] Zhu, Y. and Carbee, D. L., “Triaxial Compressive Strength of Frozen Soils
under Constant Strain Rates,” Proceedings of the Fifth International Confer-
ence on Permafrost, Trondheim, Norway, Aug 2–5, 1988, International
Permafrost Assoc., Potsdam, Germany, pp. 1200–1205.
[9] Andersen, G. R., 1991, “Physical Mechanisms Controlling the
Strength and Deformation Behavior of Frozen Sand,” Sc.D. thesis,
Department of Civil and Environmental Engineering, MIT, Cambridge,

MA.
[10] Andersen, G. R., Swan, C. W., Ladd, C. C., and Germaine, J. T., “Small-
Strain Behavior of Frozen Sand in Triaxial Compression,” Can. Geotech.
J., Vol. 32, No. 3, 1995, pp. 428–451.
[11] Swan, C., 1994, “Physical Mechanisms Controlling the Deformation and
Strength Behavior of Unfrozen and Frozen Manchester Fine Sand,” Sc.D.
thesis, Department of Civil and Environmental Engineering, MIT, Cam-
bridge, MA.
[12] Da Re, G., Santagata, M. C., and Germaine, J. T., “LVDT-based System
for the Measurement of the Pre-failure Behavior of Geomaterials,” Geo-
tech. Test. J., Vol. 24, No. 3, 2001, pp. 288–298.
[13] Da Re, G., Germaine, J. T., and Ladd, C. C., “Triaxial Testing of Frozen
Sand: Equipment and Example Results,” J. Cold Reg. Eng., Vol. 17, No.
3, 2003, pp. 90–118.
[14] Ashby, M. F., Materials Selection in Mechanical Design, Pergamon
Press, Oxford, UK, 1992.
[15] Da Re, G., 2000, “Physical Mechanisms Controlling the Pre-Failure
Stress-Strain Behavior of Frozen Sand,” Ph.D. thesis, Department of Civil
and Environmental Engineering, MIT, Cambridge, MA.
[16] Sinha, N. K., “Elasticity of Natural Types of Polycrystalline Ice,” Cold
Regions Sci. Technol., Vol. 17, 1989, pp. 127–135.
[17] Cole, D. M., Personal Communication, USACE Cold Regions Research
and Engineering Laboratory, Hanover, NH, 2000.
[18] Counto, U. J., “The Effect of the Elastic Modulus of the Aggregate on the
Elastic Modulus, Creep, and Creep Recovery of Concrete,” Magazine of
Concrete Research, Vol. 16, No. 48, 1964, pp. 129–138.
[19] Ishai, O. and Cohen, L. J., “Elastic Properties of Filled and
Porous Epoxy Composites,” Int. J. Mech. Sci., Vol. 9, 1967, pp.
539–546.
[20] Ahmed, S. and Jones, F. R., “Effect of Particulate Agglomeration and the
Residual Stress State on the Modulus of Filled Resin. Part II: Moduli of
Untreated Sand and Glass Bead Filled Composites,” Composites, Vol. 21,
No. 1, 1990, pp. 81–84.
[21] Gibson, L. J. and Ashby, M. F., Cellular Solids: Structure and Properties,
Pergamon Press, New York, 1988.
[22] Cole, D. M., “Reversed Direct-stress Testing of Ice: Initial Experimental
Results and Analysis,” Cold Reg. Sci. Technol., Vol. 18, 1990, pp. 303–321.
[23] Sahu, S. and Broutman, L. J., “Mechanical Properties of Particulate
Composites,” Polym. Eng. Sci., Vol. 12, No. 2, 1972, pp. 91–100.
[24] Leidner, J. and Woodhams, R. T., “The Strength of Polymeric Compo-
sites Containing Spherical Fillers,” J. Appl. Polym. Sci., Vol. 18, 1974,
pp. 1639–1654.
DA RE ET AL., doi: 10.1520/STP156820130013 123
[25] Spanoudakis, J. and Young, R. J., “Crack Propagation in a Glass Particle-

filled Epoxy Resin, Part 2: Effect of Particle-matrix Adhesion,” J. Mater.
Sci., Vol. 19, 1984, pp. 487–496.
[26] Moloney, A. C., Kausch, H. H., and Stieger, H. R., “The Fracture of
Particulate-filled Epoxide Resins,” J. Mater. Sci., Vol. 18, 1983, pp.
208–216.
[27] Ryzhkin, I. A. and Petrenko, V. F., “Physical Mechanisms Responsible
for Ice Adhesion,” J. Phys. Chem. B, Vol. 101, 1997, pp. 6267–6270.
[28] Hirsch, T. J., “Modulus of Elasticity of Concrete Affected by Elastic
Moduli of Cement Paste Matrix and Aggregate,” J. Am. Concr. Inst., Vol.
59, 1962, pp. 427–450.
[29] Paul, B., “Prediction of Elastic Constants of Multiphase Materials,”
Trans. Metall. Soc. AIME, Vol. 218, No. 1, 1960, pp. 36–41.
[30] Ishai, O., “The Effect of the Elastic Modulus of the Aggregate on the
Elastic Modulus, Creep and Creep Recovery of Concrete: Discussion,”
Magazine of Concrete Research, Vol. 17, 1965, pp. 148–149.
[31] Ravichandran, K. S., “Elastic Properties of Two Phase Composites,” J.
Am. Ceram. Soc., Vol. 5, 1994, pp. 1178–1184.
[32] Hashin, Z., “The Elastic Moduli of Heterogeneous Materials,” Trans.
ASME, J. Appl. Mech., Vol. 29, 1962, pp. 143–150.
[33] Hashin, Z. and Shtrikman, S., “A Variational Approach to the Theory of
the Elastic Behaviour of Mulitphase Materials,” J. Mech. Phys. Solids,
Vol. 11, 1963, pp. 127–140.
[34] Lydon, F. D. and Balendran, R. V., “Some Observations on Elastic Prop-
erties of Plain Concrete,” Cem. Concr. Res., Vol. 16, 1986, pp. 314–324.
STP 1568, 2013
DOI:10.1520/STP156820130002
Xiaoxuan Ge,1 Zhaohui (Joey) Yang,1 and Benjamin Still1
Mechanical Properties of Naturally Frozen

Ice-Rich Silty Soils
REFERENCE: Ge, Xiaoxuan, Yang, Zhaohui (Joey), and Still, Benjamin,

“Mechanical Properties of Naturally Frozen Ice-Rich Silty Soils,” Mechanical
Eds., pp. 124–139, doi:10.1520/STP156820130002, ASTM International,
West Conshohocken, PA 2013.2
ABSTRACT: Frozen soils, especially seasonally frozen soil, have great
impact on the seismic performance of bridge pile foundations. To account for
its impact on pile foundations during seismic events, it is necessary to evalu-
ate the mechanical properties of naturally frozen soil samples. This paper
focuses on the mechanical properties of naturally frozen silty soil in Alaska.
High-quality specimens of both permafrost and seasonally frozen soil were
prepared by block sampling and machining. Both horizontal and vertical
specimens were prepared to investigate the effect of specimen orientation.
Unconfined compression tests were performed at temperatures ranging from
0.7 C to 11.6 C at a constant deformation rate corresponding to a strain
rate of about 0.1 %/s, which is consistent to the strain level expected in the
frozen soil during a design earthquake in interior Alaska. Testing results
including soil characteristics and mechanical properties such as stress–strain
curves, compressive strength, yield strength, and modulus of elasticity are
presented. The impact of temperature, dry density, water content, and speci-
men orientation on the mechanical properties is discussed.
KEYWORDS: naturally frozen soil, specimen orientation, compressive
strength, Young’s modulus
August 16, 2013.
1
America.
2
Jacksonville, FL.
Copyright V
19428-2959.
124
GE ET AL., doi:10.1520/STP156820130002 125
Introduction
It is well known that frozen soil properties largely depend on temperature, ice/
water content, strain rate, dry density, and soil type. Early studies [1,2] mainly
focused on the creep behavior of frozen soils (sands, silts, and clays) [3], stud-
ied the stress–strain behavior and strength of frozen fine soils (clay and clayey
silt) at varying strain rates [4,5], studied the stress–strain behavior of frozen
Ottawa sand [6], conducted a comprehensive study on the thaw settlement and
strength of permafrost by using large permafrost core samples taken from the
field [7], found that low confining pressure (0 to 0.35 MPa) has little effect on
the compressive strength or axial strain at failure [8], conducted a comprehen-
sive study on the dynamic elastic properties of naturally frozen silts from the
CRREL Permafrost Tunnel, including both horizontally and vertically oriented
specimens by using cored specimens. They found little difference in the
dynamic properties between horizontally and vertically oriented specimens,
and the confining pressure up to 500 MPa had little impact on the dynamic
Young’s modulus [9], carried out a uniaxial compression test program on
remolded frozen Fairbanks silts under various deformation rates and studied
the mechanical properties including uniaxial compressive strength [10], and
discussed small-strain behavior of frozen sand in triaxial compression tests.
Most existing studies except [6] and [8] were based on remolded, artifi-
cially frozen soil samples, which do not necessarily represent the field condi-
tions. There is a lack of study on soil stress–strain behavior at large strain rate
by using naturally frozen samples; nor is there any information on the depend-
ency of frozen soil strength properties on specimen orientation. To evaluate
quantitatively the mechanical properties of undisturbed, naturally frozen silt
for seismic design of pile foundations embedded in frozen soil specimens
including seasonally frozen soil and permafrost of different orientation were
tested at temperatures varying from 0.7 C to 11.6 C at a constant deforma-
tion rate corresponding to a strain rate of about 0.1 %/s, which is consistent
with the strain level expected in the frozen soil during a design earthquake in
interior Alaska. Test results including unconfined compressive strength, yield
strength, Young’s modulus, and shear wave velocity are presented and impor-
tant variables effecting these properties are discussed.
Experiment
Test Specimens and Equipment

The frozen soil specimens tested in this investigation are naturally frozen silts
sampled from the Campbell Creek Bridge site in Anchorage, AK (seasonally
frozen soil) and the CRREL Permafrost Tunnel in Fox, AK (permafrost). Dur-
ing seismic loading, the pile with its supported superstructure swings back and
FIG. 1—Illustration of specimen orientation.
forth; hence, the main loading direction is horizontal. As discussed in [11],

both vertical and horizontal specimens are prepared and tested to investigate
soil orientation impact. As illustrated in Fig. 1, a vertical specimen (identified
by V) indicates that its axis is perpendicular to the ground surface and a hori-
zontal specimen (identified by H) indicates that its axis is paralleled to the
ground surface. Assuming uniform soil properties in the same block, at least
one vertical and one horizontal specimen were machined from the center
portion of the same block for comparison. All specimens are labeled by their
sampling site and orientation. For example, C2H1 indicates #1 horizontal spec-
imen machined from block #2 taken from Campbell Creek Bridge site (season-
ally frozen soil); similarly, P2V1 indicates #1 vertical specimen machined
from block #2 taken from the Permafrost Tunnel (permafrost). More details
about specimen preparation and conditioning procedures can be found in Still
et al. [11].
A total of 45 seasonally frozen soil specimens (22 horizontal and 23 verti-
cal) and 23 permafrost specimens (nine horizontal and 14 vertical) were pre-
pared. Representative properties of tested soils are shown in Table 1. Figures 2
and 3 present the respective grain-size distribution of the permafrost and sea-
sonally frozen soil. Both soils contained large amounts of fines. The permafrost
is classified as silt, whereas the seasonally frozen soil is classified as sandy or-
ganic silt with several specimens being classified as peat because of their
highly organic nature. Dry density of specimen ranges from 320 to 780 kg/m3
TABLE 1—Representative properties of tested soils.
Liquid Limit Plastic Limit Plasticity Index Specific Gravity

Seasonally frozen soil (C) 47 44 3 2.44
Permafrost (P) 39 37 2 2.55
GE ET AL., doi:10.1520/STP156820130002 127
FIG. 2—Grain-size distribution of permafrost.
for seasonally frozen soil and from 534 to 941 kg/m3 for permafrost. Water
content varies from 86 % to 225 % for seasonally frozen soil and from 62 % to
134 % for permafrost. All samples with very few exception contained visible
ice lenses and were classified as V according to ASTM D4083-89 [12]. All
specimens, when thawed, can be classified as organic silty soil with a few
specimens classified as peat.
Testing was conducted according to ASTM D7300-06 [13]. The uncon-
fined compressive strength tests were performed by a UTM-100 machine with
a temperature control chamber, which can maintain the temperature as cold as
17 C. Both stress and strain data were gathered from an independent data-
acquisition system; a load cell was used to record the axial load; and an on-
specimen extensometer was used to record the axial strain.
FIG. 3—Grain-size distribution of seasonally frozen soil.

Test Results
Three failure mechanisms were prevalent during testing, including bending,
bulging (bulging at the top, at the bottom or uniform bulging), and shearing.
There is no collapsing observed for all specimens. In most cases, small cracks
were observed around pieces of rock or ice lenses on the surface of specimen.
A majority of the specimens were loaded to an axial strain of at least 15 %.
However, for safety concerns, some tests were terminated because of excessive
bending of specimens, which was probably caused by non-uniform density and
the asymmetrical mechanical property of the specimen. The test results includ-
ing soil densities, water content (x), test temperature (T), ultimate (rm)
strength, yield strength (ry), strain values corresponding to ultimate and yield
strength, Young’s modulus (E), shear wave velocity (Vs) as interpreted from
shear modulus by assuming a Poisson’s ratio of 0.3, degree of ice saturation as
calculated from specific gravity, among others are summarized in Table 2.
Note that the ultimate strength rm was defined as the peak strength if a peak
occurs in the stress–strain curve, or the stress at 2 % axial strain. The ultimate
strength of these specimens varies from 1.653 to 7.079 MPa and the yield
strength varies from 0.742 to 5.079 MPa. Young’s modulus ranges from 2.071
to 10.569 GPa and the shear wave velocity (Vs) ranges from 809 to 1788 m/s.
Result Analyses
Typical Stress–Strain Curves

Figures 4 and 5 show respective typical stress–strain curves for both seasonally
frozen soil and permafrost. As shown in Fig. 4, some specimens demonstrated
a peak in the stress–strain curve (for example, C2H1), and most others showed
a strain-hardening behavior.
Ultimate Compressive Strength

Figures 6 and 7 show respective ultimate compressive strength-versus-tempera-
ture relationships for seasonally frozen soil and permafrost. As shown in Fig.
6, the ultimate compressive strength of seasonally frozen soil increases with
decreasing temperature. The same trend with more scattering can be observed
for permafrost in Fig. 7. The relationship between the ultimate strength and
temperature for seasonally frozen soil can be described as
rm ¼ 0:35T þ 1:80 (1)
where T is the temperature of the frozen soil. What is interesting to note is that,
for permafrost, the horizontal specimens exhibit substantially higher compres-
sive strength than the vertical specimens at the same temperature, whereas
TABLE 2—Physical and mechanical properties of frozen soil specimens.
Frozen Strain at
Water Bulk Dry Organic Test Ultimate Yield Yield Young’s Degree of
Specimen Content Density Density Content Temperature Strength Strain at e50 Strength Strength Modulus Ice Saturation
ID x (%) (kg/m3) qd (kg/m3) (%) T ( C) rm (MPa) rm (%) (%) ry (MPa) (%) E (MPa) Vs (m/s) (%)
C1 H1a 120 1254 570 18.5 7.0 4.889 1.954 0.087 3.186 0.145 4706 1201 97.33
C1 V1 103 1302 642 35.5 2.5 2.153 1.974 0.081 1.738 0.204 3366 997 97.86
C1 V2 86 1452 780 10.7 6.3 3.990 N/Ac N/Ac 3.414 N/Ac N/Ac N/Ac 100.00
C2 H1a 127 1254 551 21.6 6.0 3.935 0.803 0.083 3.214 0.202 3784 1077 98.57
C2 H2a 121 1269 574 20.5 5.8 3.325 0.650 0.062 2.351 0.117 5035 1235 99.04
C2 V1 120 1276 580 21.5 11.6 4.960 1.950 0.061 2.975 0.088 6806 1432 99.57
C2 V2 94 1343 691 30.5 9.0 4.759 1.990 0.074 2.963 0.113 6275 1341 98.82
C2 V3 115 1273 592 20.4 11.4 4.448 1.970 0.068 2.701 0.100 5501 1289 98.03
C4 V1 200 1109 370 49.2 3.4 2.938 2.014 0.124 1.775 0.178 3275 1066 95.12
C4 V2 162 1167 442 33.3 8.8 4.938 1.998 0.095 2.713 0.115 4598 1231 95.36
C4 V3 154 1177 463 38.6 9.5 4.471 1.986 0.080 2.433 0.095 5230 1037 95.97
C5 H1 128 1230 539 22.8 5.6 4.465 N/Ac N/Ac 4.115 N/Ac N/Ac N/Ac 96.57
C6 H1a 155 1181 464 31.6 6.5 4.794 1.975 0.068 2.119 0.052 7895 1603 96.85
C6 H2a 128 1237 543 13.7 9.5 5.801 1.453 0.112 4.156 0.229 4140 1171 97.49
C6 H3a 205 1111 364 20.4 2.6 3.008 1.828 0.093 1.970 0.162 3159 1046 95.64
C6 H4 225 1038 320 62.6 3.2 2.189 1.921 0.098 1.541 0.191 2071 876 90.37
C6 V1 183 1120 396 35.5 7.7 4.039 1.987 0.090 3.239 0.239 4167 1196 94.34
C6 V2 192 1124 385 44.7 7.2 4.254 1.851 0.111 3.230 0.246 3528 1099 95.71
C6 V3 116 1270 589 23.8 2.5 2.478 2.001 0.116 1.585 0.192 2709 906 98.22
C6 V4 136 1217 515 30.5 2.7 2.279 2.010 0.110 1.508 0.196 2484 886 96.81
GE ET AL., doi:10.1520/STP156820130002
C7 H1a 170 1176 436 19.0 5.8 5.031 0.455 0.058 4.098 0.142 6724 1483 98.41
C7 V1 124 1226 547 18.3 5.8 3.426 1.996 0.038 2.269 0.066 5882 1358 95.34
129
TABLE 2—Continued
Frozen Strain at
c c c c
C8 H2 169 1161 431 38.6 5.9 1.859 N/A N/A 1.690 N/A N/A N/Ac 96.47
C8 H3a 125 1241 543 22.3 3.0 2.538 0.904 0.074 1.941 0.167 3553 1049 95.21
C8 H4 122 1258 566 21.7 11.4 N/Ac N/Ac N/Ac N/Ac N/Ac 6453 1405 98.05
C9 H1a 145 1210 493 18.3 5.8 3.675 0.385 0.036 2.633 0.070 6146 1398 97.69
C9 H2 145 1201 491 18.9 5.8 N/Ac N/Ac N/Ac N/Ac N/Ac 4300 1173 97.20
C9 H4 103 1305 644 62.6 2.7 2.475 2.000 0.069 1.390 0.085 3464 1010 98.27
C9 V1 131 1204 522 20.1 5.8 3.417 1.800 0.049 2.294 0.089 5620 1340 94.87
C9 V2 93 1352 700 16.0 11.4 5.407 2.010 0.055 2.929 0.066 8214 1529 99.55
C9 V3 121 1271 574 17.7 5.8 3.583 1.960 0.070 2.578 0.139 4301 1141 99.04
C9 V4 109 1294 620 16.4 5.8 2.826 1.980 0.064 2.151 0.142 3536 1025 98.80
C9 V5 112 1247 589 17.2 3.3 2.419 1.360 0.063 1.679 0.116 3514 1041 94.83
C9 V6 136 1216 516 20.3 5.3 4.784 1.550 0.093 2.374 0.091 4000 1125 97.05
C10 H1 93 1357 702 N/Ab 4.6 4.087 1.260 0.091 3.107 0.204 5672 1268 99.95
C10 H2 103 1302 640 N/Ab 4.6 3.366 1.630 0.073 1.983 0.103 4075 1097 97.45
C10 H3 96 1334 683 N/Ab 8.7 4.877 1.080 0.087 3.615 0.188 5397 1247 99.30
C10 H4 118 1272 584 N/Ab 8.3 4.537 1.090 0.045 2.534 0.055 10569 1788 98.80
C10 H6 124 1257 562 N/Ab 8.8 5.327 1.430 0.058 2.858 0.068 8944 1654 98.74
C10 V1 118 1271 583 N/Ab 4.6 3.790 1.945 0.062 2.276 0.091 5152 1249 98.57
C11 V1 141 1225 509 N/Ab 8.1 5.380 1.860 0.103 3.959 0.219 4811 1229 98.90
C11 V2 119 1276 583 N/Ab 10.5 7.079 0.972 0.041 4.760 0.077 8438 1595 99.41
C11 V4 130 1271 553 N/Ab 4.5 3.878 2.009 0.122 2.262 0.164 3784 1070 100.00
P4 H1 134 1251 534 N/Ab 1.1 3.044 1.956 0.061 2.439 0.149 4748 1208 98.70
P4 H4a 130 1255 545 N/Ab 1.0 2.414 1.258 0.066 2.048 0.161 2234 827 98.26
P4 V1 100 1341 671 N/Ab 1.3 2.890 1.956 0.067 2.322 0.168 4196 1097 99.30
TABLE 2—Continued
Frozen Strain at
b
P4 V2 62 1319 816 N/A 0.7 2.358 2.012 0.067 1.696 0.140 3291 980 81.13
P4 V3 72 1442 840 N/Ab 0.9 1.653 2.016 0.082 1.198 0.171 2454 809 98.35
P4 V4 76 1420 809 N/Ab 1.0 1.751 2.008 0.123 0.742 0.087 2480 820 98.21
P6 H1 114 1299 606 N/Ab 4.5 4.303 2.000 0.062 4.215 0.199 5013 1218 98.82
P6 H2 72 1460 847 N/Ab 4.5 5.360 2.000 0.116 4.346 0.288 4386 1075 99.58
P6 H3 85 1408 762 N/Ab 7.5 4.650 0.270 0.066 4.345 0.200 5802 1259 100.00
P6 V1 N/Ab N/Ab 743 N/Ab 10.0 2.608 N/Ac N/Ac 2.094c N/Ac N/Ac N/Ac N/Ab
P6 V2 84 1454 792 N/Ab 10.0 2.362 N/Ac N/Ac N/Ac N/Ac N/Ac N/Ac 100.00
P7 V1 N/Ab N/Ab 705 N/Ab 10.0 N/Ac N/Ac N/Ac 1.907c N/Ac N/Ac N/Ac N/Ab
P8 H1 84 1415 770 N/Ab 7.5 6.015 1.120 0.112 5.079 0.292 5166 1185 100.00
P8 H2 93 1405 728 N/Ab 7.5 6.370 1.430 0.075 4.684 0.155 8891 1560 100.00
P8 V1 88 1380 733 N/Ab 6.7 2.509 2.000 0.077 1.880 0.191 4315 1097 98.72
P8 V2 74 1457 839 N/Ab 6.7 3.340 0.781 0.074 2.263 0.131 4650 1108 100.00
P9 H1 106 1319 641 N/Ab 7.5 5.160 0.751 0.069 4.153 0.169 6477 1374 98.98
P9 H2 80 1426 791 N/Ab 7.5 6.010 1.107 0.116 5.036 0.296 4195 1064 100.00
P9 V1 N/Ab N/Ab 683 N/Ab 7.5 3.620 2.000 0.058 3.120 0.157 4093 N/Ac N/Ab
P9 V4 108 1320 634 N/Ab 7.5 3.680 2.000 0.060 2.829 0.133 4631 1162 99.38
P11 V1 75 1460 835 N/Ab 4.5 3.320 2.000 0.110 2.422 0.166 3362 941 100.00
P11 V2 62 1525 941 N/Ab 4.5 3.620 2.000 0.093 2.250 0.145 3574 950 100.00
P11 V3 73 1450 837 N/Ab 7.5 3.190 2.000 0.049 1.599 0.049 5439 1201 99.19
a
Indicates the specimens whose stress–strain curves have peak.
GE ET AL., doi:10.1520/STP156820130002
b
Indicates the corresponding standard test was not conducted for the specimen.
c
Indicates test result could not be obtained because of data-acquisition system error.
131
FIG. 4—Stress–strain curves for seasonally frozen soil.
FIG. 5—Stress–strain curves for permafrost.
FIG. 6—Ultimate compressive strength versus temperature for seasonally fro-

zen soil.
GE ET AL., doi:10.1520/STP156820130002 133
FIG. 7—Ultimate compressive strength versus temperature for permafrost.
there is no such trend for seasonally frozen soil. The respective relationships
for horizontal and vertical permafrost specimens can be described as
ðVerticalÞ rm ¼ 0:44T þ 2:5 (2)
ðHorizontalÞ rm ¼ 0:19T þ 2:1 (3)
The increase in strength for the horizontal permafrost specimens is likely

because of ice wedge formation commonly observed in permafrost. During its
formation process, ice wedges expand and consolidate the surrounding soil lat-
erally, which can possibly lead to much higher compressive strength in hori-
zontally oriented specimens than vertically oriented specimens.
Figure 8 shows the variation of ultimate strength rm with dry density qd for
different temperatures. It can be seen from this figure that the ultimate strength
decreases with increasing dry density for specimens tested at temperatures
ranging from 5.6 C to 6.1 C, and this trend is also visible for specimens
FIG. 8—Ultimate compressive strength versus dry density.

tested at temperatures ranging from 2.5 C to 3.4 C but not nearly as

obvious. The relationship between the ultimate strength and dry density for dif-
ferent temperatures can be described as follows:
At 5:6 to 6:1 C; rm ¼ 0:0097qd þ 9:10 (4)
At 2:5 to 3:4 C; rm ¼ 0:0011qd þ 3:00 (5)
Refs 9 and 14 also looked into the relationship between the ultimate strength
and dry density of artificially frozen specimens and came to similar conclu-
sions. For example, Zhu and Carbee [9] tested remolded frozen silts under a
strain rate of 1.1 103/s at 2 C (almost the same as the strain rate used in
this study), and they found a very moderate trend in the ultimate strength-
versus-dry density curve.
Figure 9 shows the variation of ultimate strength with water content for
specimens tested at two temperature ranges. It can be observed that the ulti-
mate strength increases with increasing water content and the relationships can
be described by the following equations at different temperature ranges:
At 5:6 to 6:1 C; rm ¼ 0:0176x þ 1:66 (6)
At 2:5 to 3:4 C; rm ¼ 0:0003x þ 2:40 (7)
These observations may seem counter-intuitive when compared with

unfrozen soils. In unfrozen soil, the inter-particle friction contributes to the ma-
jority of the strength, and higher density typically means larger friction angle
and, hence, larger strength. In frozen soil, however, it is the pore ice that bonds
the soil particles together and contributes to the majority of the strength. One
would expect the ultimate strength of frozen soils to increase with increasing
FIG. 9—Ultimate compressive strength versus water content.

GE ET AL., doi:10.1520/STP156820130002 135
FIG. 10—Yield strength versus ultimate compressive strength for seasonally

frozen soil.
water or ice content, or increase with decreasing dry density, because the dry
density decreases with increasing water or ice content. Further, as the tempera-
ture increases, there will be more unfrozen water in silty soils, therefore weak-
ening this trend.
Yield Strength
Figures 10 and 11 show the relationships between the yield strength and the
ultimate compressive strength at various temperatures for seasonally frozen
soil and permafrost, respectively. The respective relationships for seasonally
frozen soil and permafrost can be described by the following equations:
ðSeasonalÞ ry ¼ 0:61rm þ 0:27 (8)
ðPermafrostÞ ry ¼ 0:90rm 0:46 (9)
FIG. 11—Yield strength versus ultimate compressive strength for permafrost.

FIG. 12—Young’s modulus versus temperature.
Young’s Modulus
Figure 12 shows the Young’s modulus as a function of temperature. It is
observed from Fig. 12 that the Young’s modulus decreases with increasing
temperature. The horizontal specimens tend to have higher Young’s modulus,
especially for permafrost, although this trend is not nearly as clear as observed
in ultimate strength. A trend line for both seasonally frozen soil and permafrost
are plotted in Fig. 12 and can be described as follows:
E ¼ 411:81T þ 2417 (10)
Shear Wave Velocity

The shear wave velocity is related to shear modulus (G) and bulky density (q).
The shear modulus can be evaluated based on Young’s modulus (E) and Pois-
son’s ratio () by using Eq 11. Poisson’s ratio was not available and assumed to
be 0.3 in this study. The shear wave velocity (Vs) was calculated by using Eq 12.
E
G¼ (11)
2ð1 þ Þ
sffiffiffiffi
G
VS ¼ (12)
q
Figure 13 shows the shear wave velocity versus temperature for both sea-
sonally frozen soil and permafrost. In general, the shear wave velocity
decreases with increasing temperature and the relationship can be described by
the following equation:
VS ¼ 50T þ 890 (13)

GE ET AL., doi:10.1520/STP156820130002 137
FIG. 13—Shear wave velocity versus temperature.
Conclusions
A comprehensive testing program of naturally frozen soils including seasonally
frozen soil and permafrost was carried out to study frozen soil mechanical
properties. All specimens are of organic silty nature and can be classified as
ice-rich; their water content ranges from 62 % to 225 % and their dry density
varies from 320 to 940 kg/m3. Unconfined compression test was carried out
with these specimens under a constant deformation rate corresponding to a
strain rate of about 0.1 %/s at temperatures varying from 0.7 C to 11.6 C.
The ultimate compressive strength varies from 1.65 to 7.08 MPa and the yield
strength varies from 0.74 to 5.08 MPa. The Young’s modulus ranges from 2.1
to 10.6 GPa and the shear wave velocity ranges from 800 to 1800 m/s. The fol-
lowing conclusions can be made based on analyses of the test results:
(1) The ultimate compressive strength of both seasonally frozen soil and
permafrost decreases with increasing temperature; it decreases with
increasing dry density, or increases with increasing water or ice con-
tent. The trend for the latter is clearer at lower temperatures.
(2) There is a correlation between the yield strength and the ultimate
strength for both seasonally frozen soil and permafrost.
(3) The Young’s modulus decreases with increasing temperature. The hor-
izontal specimens tend to have higher Young’s modulus, especially for
permafrost. Similarly, the shear wave velocity of frozen soils decreases
with increasing temperature.
(4) For permafrost, the ultimate compressive strength of horizontal speci-
men is substantially higher than that of vertical specimen at the same
testing temperature. This strength anisotropy is likely due to ice wedge
formation commonly observed in permafrost.
(5) The factors that affect the mechanical properties in descending impor-
tance are: temperature, water content or dry density, and specimen
orientation.
These mechanical properties can be directly used to evaluate frozen soil

lateral resistance in the analyses of pile foundations during seismic loading.
Acknowledgments
The research is supported by the Alaska University Transportation Center and
the State of Alaska Department of Transportation and Public Facilities under
AUTC Project #510021. The writers would like to express their thanks to Drs.
Anthony Paris and Scott Hamel and Mr. Daniel King for their assistance on the
testing equipment and data-acquisition system. Mr. Donald Richardson took
part in these experiments. The writers are thankful to former graduate students
Mr. Xiaoyu Zhang and Mr. Qiang Li from UAA for their assistance with field
sampling. Mr. Elmer Marx from AK DOT&PF is the Project Advisory Comm.
chair, and we appreciate his support for this project.
References
[1] Sayles, F. H., “Creep of Frozen Sands,” Technical Report 190, Cold
Regions Research and Engineering Laboratory, Hanover, NH, 1968.
[2] Sayles, F. H. and Haines, D., “Creep of Frozen Silts and Clay,” Technical
Report 252, Cold Regions Research and Engineering Laboratory, Hano-
ver, NH, 1974.
[3] Akili, W., “Stress–Strain Behavior of Frozen Fine-grained Soils,” High-
way Res. Rec., Vol. 360(1), 1971, p. 8.
[4] Sayles, F. H., “Triaxial and Creep Tests on Frozen Ottawa Sand,”
Proceedings of the 2nd International Conference on Permafrost, North
American Contribution, Yakutsk, U.S.S.R., July 13–28, 1973, National
Academy of Science, 1973, pp. 384–391.
[5] Jones, S. J. and Parameswaran, V. R., “Deformation Behavior of Frozen
Sand-Ice Materials Under Triaxial Compression,” Proceedings of the 4th
International Permafrost Conference, Fairbanks, AK, July 17–22, 1983,
pp. 560–565.
[6] Watson, G. H., Slusarchuk, W. A., and Rowley, R. K., “Determination of
Some Frozen and Thawed Properties of Permafrost Soils,” Can. Geotech.
J., Vol. 10(4), 1973, pp. 592–606.
[7] Baker, T. H. W., Jones, S. J., and Parameswaran, V. R., “Confined and Uncon-
fined Compression Tests on Frozen Sands,” Proceedings of the 4th Canadian
Permafrost Conference, Roger J.E. Brown Memorial, Calgary, Alberta,
March, 1981, pp. 387–393.
[8] Vinson, T., Wilson, C., and Bolander, P., “Dynamic Properties of Natu-
rally Frozen Silt,” Proceedings of the 4th Permafrost International Con-
ference, Fairbanks, AK, July 17–22, 1983, pp. 1315–1320.
GE ET AL., doi:10.1520/STP156820130002 139
[9] Zhu, Y. and Carbee, D. L., “Uniaxial Compressive Strength of Frozen Silt
under Constant Deformation Rates,” Cold Regions Sci. Technol., Vol. 9,
1984, pp. 3–15.
[10] Andersen, G. R., Swan, C. W., Ladd, C. C., and Germaine, J. T., “Small-
Strain Behavior of Frozen Sand in Triaxial Compression,” Can. Geotech.
J., Vol. 32, 1995, pp. 428–451.
[11] Still, B., Yang, Z., and Ge, X., “Sampling, Machining and Testing of
Naturally Frozen Soils,” ASTM STP 1568, ASTM International, West
[12] ASTM D4083-89: Standard Practice for Description of Frozen Soils
2006.
[14] Sayles, F. H. and Carbee, D. L., “Strength of Frozen Silt as a Function of
Ice Content and Dry Unit Weight,” Eng. Geol., Vol. 18, 1981, pp. 55–66.
STP 1568, 2013
DOI:10.1520/STP156820120209
Jung-Hee Park,1 Jong-Sub Lee,2 Seung-Seo Hong,3

and YoungSeok Kim4
Properties of Elastic Waves in Sand–Silt

Mixtures Due to Freezing
REFERENCE: Park, Jung-Hee, Lee, Jong-Sub, Hong, Seung-Seo, and Kim,

YoungSeok, “Properties of Elastic Waves in Sand–Silt Mixtures Due to Free-
zing,” Mechanical Properties of Frozen Soils, STP 1568, Hannele Zubeck
and Zhaohui Yang, Eds., pp. 140–152, doi:10.1520/STP156820120209,
ASTM International, West Conshohocken, PA 2013.5
ABSTRACT: The freezing of soil converts a particulate medium into a contin-
uous solid bonded by ice bridging. The objective of this study is to investigate
the property changes in elastic waves, including shear and compressional
waves, in a sand–silt mixture during soil freezing, using bender elements
and piezo disk elements. Experiments are performed in a small freezing
cell, which is designed to freeze soil specimens from top to bottom. Bender
elements and piezo disk elements are used for the generation and detection
of the shear and compressional waves. Two pairs of bender elements
and two pairs of piezo disk elements are installed in the walls of the freezing
cell at two different depths. A sand–silt mixture, with a silt fraction of 10 %
and a degree of saturation of 40 %, is frozen in a freezer. As the temperature
of a specimen drops from 20 C to 10 C, the velocities, resonant
Manuscript received December 18, 2012; accepted for publication April 5, 2013; published
1
Graduate Student, School of Civil, Environmental and Architectural Engineering, Korea Univ.,
1, 5-ga, Anam-dong, Sungbuk-gu, Seoul 136-713, Korea.
2
Professor, School of Civil, Environmental and Architectural Engineering, Korea Univ., 255-53,
Anam-dong, Sungbuk-gu, Seoul, 136-713, Korea (Corresponding author), e-mail:
jongsub@korea.ac.kr
3
Senior Researcher, Geotechnical Engineering Research Division, Korea Institute of
Construction Technology, Goyang 411-712, Korea.
4
Research Fellow, Geotechnical Engineering Research Division, Korea Institute of Construction
Technology, Goyang 411-712, Korea.
5
Jacksonville, FL.
Copyright V
19428-2959.
140
PARK ET AL., doi:10.1520/STP156820120209 141
frequencies, and amplitudes of shear and compressional waves are continu-

ously monitored. Experimental results show that the properties of the elastic
waves change markedly as the specimen is freezing; the velocities of the
shear and compressional waves dramatically increase. Poisson’s ratio con-
tinuously decreases with the decrease in temperature, and becomes con-
stant after soil freezing. In addition, huge increases are observed in the
resonant frequencies of both elastic waves. This study demonstrates that
the dramatic increases in both velocity and resonant frequency result from
the ice bonding of the particulate mixture as a result of freezing of the pore
moisture.
KEYWORDS: compressional wave, elastic wave amplitude, elastic wave ve-
locity, resonant frequency, shear wave, soil freezing
Introduction
Pure water is a solid at temperature below 0 C at the standard atmospheric
pressure. When moist or saturated soil freezes, it increases in volume, which
causes significant damage to geotechnical structures, including pavement,
railroads, and foundations [1,2]. Several experimental studies have been per-
formed with the objective of characterizing frozen soils and preventing dam-
age to geotechnical structures. One of these studies used elastic waves,
including shear and compressional waves, in liquid kerosene [3–6]. Thus, fro-
zen soils should be submerged in kerosene. The behavior of shear waves in
liquid can be evaluated through mode conversion. This traditional method of
analyzing elastic waves, however, may be used to characterize the properties
of completely frozen soils [7,8]. Thus, a new elastic wave method has been
required for the characterization of soil properties during the process of soil
freezing.
In this study, shear and compressional waves were continuously moni-
tored as the soil temperature dropped from 20 C to 10 C. The experiments
were carried out in a nylon cell equipped with shear and compressional
wave transducers. Bender elements and piezo disk elements were used for
the shear and compressional wave transducers, respectively. Bender ele-
ments are well-known shear wave transducers, because of their tendency to
attach well to soils [9–11]. The specimen, which was prepared by mixing
sand and silt, was frozen in a laboratory freezer. The sands material was of
uniform size distribution to minimize particle-size effects on shear and com-
pressional waves.
In this study, three properties of shear and compressional waves, i.e., wave
velocities, amplitudes, and frequencies, were continuously measured during
the temperature decrease from 20 C to 10 C. In particular, this study concen-
trated on the changes in elastic wave properties near the freezing point of the
soils. The paper describes the experimental setup, measurements, analyses and
conclusions of the study.
Experimental Setup
The specimens were prepared by mixing sand and silt particles. Jumunjin sand
with a particle size of 0.45 mm was used to minimize potential effects because
of variations in particle sizes. The silt consisted of crushed limestone with a
particle size of less than 75 lm. Crushed limestone has been commonly used
for investigating frost susceptibility in fine soils [12]. The sand and silt were
mixed to achieve a silt fraction (% of silt ¼ Wsilt/Wsand 100 %) of 10 % by
weight. The maximum and minimum void ratios of the sand-silt mixture were
0.74 and 0.47, respectively, as determined by test methods ASTM D4253 [13]
and ASTM D4254 [14]. The water content of the sand–silt mixture was 10 %
of dry weight, which ultimately corresponded to a degree of saturation of 40 %.
The mixture specimen was placed into the freezing cell in four layers, and was
compacted by applying the same tamping energy to each layer. The specimen
was compacted to achieve a uniform relative density of 70 %.
Freezing Cell
The nylon cell used for freezing the specimens is shown in Fig. 1. The freezing
cell consisted of four pieces of nylon plates, as shown in Fig. 1(a), to minimize
vibrations propagating through the nylon plates. Bender elements were used as
shear wave transducers, and piezo disk elements were used as compressional
wave transducers. Pairs of bender elements and piezo disk elements for gener-
ating and detecting the shear and compressional waves were attached to the
side walls, as shown in Fig. 1.
Test Procedure
After the sand–silt mixtures were prepared in the cell, the cell was placed in
the laboratory freezer. The temperature inside the freezer was maintained at a
temperature of 30 C. The temperature of the specimen was continuously
measured using k-type thermocouples, as the temperature of the specimen
dropped from 20 C to 10 C. Two thermocouples were installed, one each in
the upper and lower parts of the specimen, as shown in Fig. 2, for monitoring
the specimen temperature along its depth. The side walls of the cell were cov-
ered with insulation material (Styrofoam), so that freezing of the specimens
would progress downward from the top to the bottom.
Measurement Systems
A schematic of the shear and compressional wave measurement systems is
shown in Fig. 3. Input signals, which were generated by a function generator,
were transformed to elastic waves by the source bender elements and piezo
PARK ET AL., doi:10.1520/STP156820120209 143
FIG. 1—Freezing cell: (a) top view; (b) side view along a-a0 section; and (c)
side view along b-b0 section. BE and PDE denote bender elements and piezo
disk elements, respectively.
FIG. 2—Arrangements of the thermocouples along the depth. The shaded part
indicates insulation using Styrofoam.
disk elements. The generated elastic waves were propagated through the
sand–silt mixture and were measured by the receiver bender elements and
piezo disk elements. The captured signals were filtered and amplified using a
filter amplifier, displayed on an oscilloscope and recorded by the computer.
The high-pass cut-off frequency was 500 Hz, and the low-pass cut-off
frequency was at least 10 times the resonant frequency of the elastic waves
[15]. The 1024 signals were stacked to minimize uncorrelated high-frequency
noise.
Results and Analyses
Temperature Change
The shear and compressional waves were continuously measured in the upper
and lower parts of the freezing cell during soil freezing as the specimen tem-
perature dropped from 20 C to 10 C. The temperature of the specimen versus
time is presented in Fig. 4, which shows that the temperature of the upper part
of the specimen decreased slightly faster than that of the lower part of the spec-
imen, because of slightly greater convective heat loss from the top than from
PARK ET AL., doi:10.1520/STP156820120209 145
FIG. 3—Elastic wave measurement system.
the bottom. A constant specimen temperature region was observed over time
during the experiment; the liquid water coexisted with the (solid) ice at a tem-
perature of 0 C for approximately 4 hours because thermal energy was released
by the phase change from the liquid state to the solid state.
FIG. 4—Temperatures change along the depth with time.

Elastic Waves
The elastic waves measured in the upper part of the specimen are plotted in
Fig. 5. Figure 5(a) shows that the first arrivals of the shear waves, which takes
into account the near-field effect [9,10], gradually decrease, as the temperature
decreased from 20 C to 0 C. The first arrivals of the compressional waves,
however, changed little with this same range of temperatures decrease, as
shown in Fig. 5(b). After the soil specimens froze, the first arrivals of shear and
compressional waves dramatically decreased, as shown in Figs. 5(a) and 5(b),
respectively. Furthermore, the wave shapes, including amplitude and fre-
quency, changed greatly. The amplitudes of the shear waves decreased after
the soil was frozen. The amplitude change observed in the shear waves resulted
from the phase change of water within the sand–silt mixture specimen. As the
pore moisture underwent a phase change from water to ice and the ice bonded
FIG. 5—Elastic wave signatures during soil freezing in the upper part: (a)
shear waves; and (b) compressional waves. Note the input signal is a single si-
nusoid with an adjusted resonant frequency according to soil condition. The
small circles on the figure denote the first arrivals of elastic waves.
PARK ET AL., doi:10.1520/STP156820120209 147
the soil particles together, the mixtures of sand, silt, and pore moisture became
a continuous solid material.
Elastic Wave Velocities

The elastic wave velocity (V), which is the ratio of the tip-to-tip distance between
the transducers (Ltip-tip) to the first arrival time (t), is calculated as follows:
Ltiptip
V¼ (1)
t
The calculated shear and compressional wave velocities versus the tempera-
ture change are plotted in Fig. 6. As the temperature decreased from 20 C to
FIG. 6—Elastic wave velocity versus temperature: (a) shear waves; and (b)
compressional waves.
0 C, the shear wave velocity gradually increased, whereas the compressional

wave velocity was almost constant in the upper and lower parts of the speci-
men. The shear and compressional wave velocities dramatically increased
as the temperature passed down through 0 C. When the temperature was
lower than 0 C, the shear and compressional wave velocities were
almost constant. The shear wave velocities were 128 m/s and 2120 m/s in
the upper part of the specimen before and after freezing, respectively. In addi-
tion, the compressional wave velocities in the upper part of the specimen
increased from 250 m/s to 3270 m/s after the specimen froze, as shown in
Fig. 6.
The elastic waves are affected by the effective stress as follows [16,17]:
0
r0 rw þ r0m b
Vs ¼ a 0 ¼a (2)
Pa 2Pa
where, r00 ¼ (rw0 þ rm0 )/2 is the average effective stress on the polarization
plane, rw0 is the effective stress in the direction of wave propagation, rm0 is the
effective stress in the direction of particle motion, Pa is the atmospheric pres-
sure expressed in the same units as r00 , and a and b are empirically set param-
eters. The large change in the shear wave velocity during freezing (as the
temperature dropped through 0 C) in the upper part of the specimen, from
128 m/s to 2120 m/s, demonstrates that the dominant factor affecting the shear
wave velocities changed from one of interparticle effective stress to one of
ice bonding. Note that the compressional wave velocity was also affected by
the effective stress, because of the specimen being an unsaturated soil–-
particle material. After freezing, the specimen became a continuous solid, and
the compressional wave velocity dramatically increased. The shear and com-
pressional wave velocities measured in the lower part of the specimen were
greater than those measured in the upper part of the specimen because the
effective stress was higher in the lower part. Similar results were found in
studies involving soil cementation [18–21]. In this study, the ratio of the com-
pressional to the shear wave velocity decreased from 1.94 to 1.54 as the soil
froze.
Poisson’s Ratio
The Poisson’s ratio can be obtained from the compressional and shear wave
velocities (Vp and Vs, respectively) as follows:
2Vs2 Vp2
¼ (3)
2ðVs2 Vp2 Þ
PARK ET AL., doi:10.1520/STP156820120209 149
FIG. 7—Poisson’s ratio versus temperature.
The Poisson’s ratios calculated from measurements in the upper and lower
parts of the specimen are plotted in Fig. 7. Figure 7 shows that as the tempera-
ture decreased from 20 C to 0 C, the Poisson’s ratio gradually decreased from
0.3 to 0.14 because the shear wave velocity decreased while the compressional
wave velocity remained nearly constant. Note that the Poisson’s ratios in the
lower part are smaller than those in the upper part (during the same temperature
drop from 20 C to 0 C) because the effect of the confining stress on the shear
wave velocity is greater than on the compressional wave velocity. At tempera-
tures below 0 C, the Poisson’s ratios in the lower part are still slightly smaller
than those in upper part.
Resonant Frequency
Resonant frequencies of shear and compressional waves versus temperatures
are presented in Fig. 8. While the resonant frequencies of shear waves slightly
increased as the temperature dropped from 20 C to 10 C, the resonant fre-
quencies of compressional waves stayed nearly constant at temperatures above
freezing. The phase change of pore water from liquid to solid had a significant
influence on the resonant elastic frequencies; the resonant frequencies of both
shear and compressional waves dramatically increased as the temperature
dropped through 0 C, and their resonant frequency remained constant as the
temperature dropped still lower. The change in resonant frequency because of
the temperature change was similar to the change in the elastic wave velocity.
The resonant frequencies measured in the lower part of the specimen were
larger than those recorded in the upper part. The ratios of the resonant frequen-
cies of the compressional waves to shear waves before and after soil freezing
are 1.1 and 1.7, respectively.
FIG. 8—Resonant frequency of elastic waves versus temperature: (a) shear

waves; and (b) compressional waves.
Conclusions
The objective of this study was to investigate the change of elastic wave prop-
erties in partially saturated particulate soil material during soil freezing. The
shear and compressional waves were continuously measured at two locations
in a sand–silt mixture as the temperature dropped from 20 C to 10 C. The
weight fraction of silt in the sand–silt mixture was 10 %, and the degree of sat-
uration was 40 %. The freezing cell, which consisted of four separate walls,
was assembled and insulated to induce freezing of the soil specimen progres-
sively downward from the top to the bottom. Two pairs of bender elements and
piezo disk elements were installed in the upper and lower parts of the freezing
PARK ET AL., doi:10.1520/STP156820120209 151
cell. As the temperature of specimen dropped from 20 C to 10 C, the elastic
waves were continuously monitored
Experimental results show that the shear and compressional wave velocities
dramatically increased as the temperature dropped through 0 C, whereas the shear
and compressional wave velocities remained nearly constant at temperatures
below freezing. Although the amplitudes of compressional waves were constant
near 0 C, those of shear waves decreased. Poisson’s ratio, which was determined
from the elastic wave velocities, decreased as the temperature decreased and
became constant after the soil mixture was completely frozen. The changes in res-
onant frequencies of the shear and compressional waves because of the tempera-
ture change were similar to the changes in elastic wave velocities. The changes in
the elastic waves because of freezing demonstrate that the dominant factor affect-
ing the elastic waves changes from one of effective stress to one of ice bonding.
This study may provide important new information regarding changes in the
behavior of shear and compressional waves in soil during soil freezing.
Acknowledgments
This work is supported by a National Research Foundation of Korea (NRF)
grant funded by the Korean government (MEST) (No. 2012-0005729) and the
“Development of the platform technology for quick construction of the sub-
structure at the polar region,” which is a key development project of the Korea
Institute of Construction Technology for 2011.
References
[1] Andersland, O. B. and Ladanyi, B., Frozen Ground Engineering, 2nd ed.,
ASCE Press/John Wiley & Sons, New York, 2004, p. 363.
[2] Guy, O., Jean-Marie, K., and Marius, R., “Deterioration Model for Pave-
ments in Frost Conditions,” Transportation Research Record 1655, Trans-
portation Research Board, National Research Council, Washington, D.C.,
1999, pp. 110–117.
[3] Fukuda, M. and Sheng, Y., “Elasticity Measurement of Frozen Silt by
Immersion Ultrasonic Sing-Around Method,” IEEE Ultrasonics Symp.,
Vol. 2, 1998, pp. 1207–1210.
[4] Christ, M. and Park, J. B. “Ultrasonic Technique as Tool for Determining
Physical and Mechanical Properties of Frozen Soils,” Cold Regions Sci.
Technol., Vol. 58, 2009, pp. 136–142.
[5] Sheng, Y., Fukudu, M., and Kim, H., “Effect of Unfrozen Water Content
on the Ultrasonic Velocities in Tire-Mixed Frozen Soils,” Chinese J. Geo-
tech. Eng., Vol. 22, No. 6, 2000, pp. 716–720.
[6] Christ, M. and Park, J. B., “Determination of Elastic Constants of Frozen
Rubber-Sand Mixes by Ultrasonic Testing,” J. Cold Regions Eng., Vol.
25, No. 4, 2011, pp. 196–207.
[7] Nakano, Y., Martin, R. J., and Smith, M., “Ultrasonic Velocity of the Di-
lation and Shear Wave in Frozen Soil,” Water Resour. Res., Vol. 8, No. 4,
pp. 1024–1030.
[8] Wang, D. Y., Zhu, Y. L., Ma, W., and Niu, Y. H., “Application of Ultra-
sonic Technology for Physical-Mechanical Properties of Frozen Soil,”
[9] Lee, J. S. and Santamarina, J. C., “Bender Elements: Performance and
Signal Interpretaion,” J. Geotech. Geoenviron. Eng., Vol. 131, No. 9,
2005, pp. 1063–1070.
[10] Lee, J. S. and Santamarina, J. C., “Discussion ‘Measuring Shear Wave
Velocity Using Bender Elements’ by Leong, E. C., Yeo, S. H., and
Rahardjo, H.,” Geotech. Testing J., Vol. 29, No. 5, 2006, pp. 439–441.
[11] Lee, J. S., Lee, C., Yoon, H. K., and Lee, W.–B., “Penetration Type Field
Velocity Probe for Soft Soils,” J. Geotech. Geoenviron. Eng., Vol. 136,
No. 1, 2010, pp. 199–206.
[12] Tester, R. E. and Gaskin, P. N., “Effect of Fines Content on Frost Heave,”
Can. Geotech. J., Vol. 33, 1996, pp. 678–680.
[13] ASTM D4253-00: Standard Test Methods for Maximum Index Density
and Unit Weight of Soils Using a Vibratory Table, Annual Book of ASTM
[14] ASTM D4254-00: Standard Test Methods for Minimum Index Density
and Unit Weight of Soils Calculation of Relative Density, Annual Book of
[15] Lee, J. S., 2003, “High Resolution Geophysical Techniques for Small-
Scale Soil Model Testing,” Ph.D. thesis, Georgia Institute of Technology,
Atlanta, GA, 280 pp.
[16] Roesler, S. K., “Anisotropic Shear Modulus Due to Stress Anisotropy,” J.
Geotech. Eng. Div., Vol. 105, No. 7, 1979, pp. 871–880.
[17] Yu, P. and Richart, F. E., Jr., “Stress Ratio Effects on Shear Modulus of
Dry Sands,” J. Geotech. Eng., Vol. 110, No. 3, 1984, pp. 331–345.
[18] Truong, Q. H., Eom, Y. H., Byun, Y. H., and Lee, J. S., “Characteristics
of Elastic Waves According to Cementation of Dissolved Salt,” Vadose
Zone J., Vol. 9, No. 3, 2010, pp. 662–669.
[19] Lee, C., Truong, Q. H., and Lee, J. S., “Cementation and Bond Degrada-
tion of Rubber–Sand Mixtures,” Can. Geotech. J., Vol. 47, No. 7, 2010,
pp. 763–774.
[20] Yoon, H. K., Lee, C., Kim, H. K., and Lee J. S., “Evaluation of Preconso-
lidation Stress by Shear Wave Velocity,” Smart Struct. Syst., Vol. 7, No.
4, 2011, pp. 275–287.
[21] Truong, Q. H., Lee, C., Kim, Y. U., and Lee, J. S., “Small Strain Stiffness
of Salt-Cemented Granular Media Under Low Confinement,” Geotechni-
que, Vol. 62, No. 10, 2012, pp. 949–953.
STP 1568, 2013
DOI:10.1520/STP156820130034
Stanley J. Vitton1 and Mark R. Muszynski2
Strength and Creep Properties of a Frozen

Coastal Sand in Saltwater
REFERENCE: Vitton, Stanley J. and Muszynski, Mark R., “Strength and

Creep Properties of a Frozen Coastal Sand in Saltwater,” Mechanical Proper-
ties of Frozen Soils, STP 1568, Hannele Zubeck and Zhaohui Yang, Eds., pp.
153–166, doi:10.1520/STP156820130034, ASTM International, West Con-
shohocken, PA 2013.3
ABSTRACT: In 1998, the Port of Los Angeles contracted to have a force
main under their primary shipping channel relocated from a depth of 13 m to
26 m. Access shafts were excavated using frozen ground technology to a
depth of 26 m, with the frozen ground used for support of the excavation.
When the shaft was within 1 m of the required depth, water started to enter
into the excavation. Six hours later, the shaft collapsed. An investigation was
conducted into the strength of the frozen soil. Unconfined compression
strength tests were conducted at strain rates of 0.1 % and 1 %. The samples
were saturated with water from the shaft, which had a salt concentration
measured as equal to that of seawater. Short-term compressive strength
was measured between 1300 and 1900 kPa. Constant stress creep tests
were also conducted, indicating a strength of about 760 kPa. All testing was
conducted at a temperature of 10 C.
KEYWORDS: frozen soil, creep, seawater
Introduction
In an effort to improve shipping at the Port of Los Angeles in Los Angeles,
CA, the Fries Ave. project was undertaken by the Port Authority to lower a
Manuscript received February 14, 2013; accepted for publication July 23, 2013; published online
August 23, 2013.
1
Associate Professor, Dept. of Civil & Environmental Engineering, Michigan Technological
Univ., 1400 Townsend Dr., Houghton, MI 49931-1295, United States of America.
2
Assistant Professor, Dept. of Civil Engineering, Gonzaga Univ., 502 East Boone Ave.,
Spokane, WA 99258-0026, United States of America.
3
Jacksonville, FL.
Copyright V
19428-2959.
153
force main from a depth of 13 m to 26 m. This modification would allow for

deeper dredging in the east shipping channel. The project was started in the fall
of 1998 with the selection of ground freezing as the main form of ground sup-
port. The main jacking shaft was located on Terminal Island, at the beginning
of the East Basin Channel, just off the Main Channel and the Turning Basin. A
freezing subcontractor installed 24 steel pipes, each 76 mm in diameter, at an
angular spacing of 15 to a depth of 70 m. A second row of freezing pipes was
installed to counter the effects of tidal action near the ground surface. The pro-
ject involved microtunneling a 1130-mm-diameter steel casing pipe a horizon-
tal distance of 255 m to a receiving shaft located on Mormon Island. A 760-
mm-diameter ductile iron force main was then installed inside the casing pipe.
The jacking shaft was excavated to a depth of 26 m, maintaining a maximum
diameter of 6 m. When the excavation was within 1 m of the anticipated shaft’s
depth, water began leaking into the shaft from a depth of 18 to 24 m below
grade. For three hours water seeped into the shaft, and then the seeping ceased.
Three hours later, however, the shaft suddenly failed, with an inflow of sea-
water and soil creating a sinkhole at the surface adjacent to the shaft but out-
side the frozen ground section. The water inrush ceased when the water level
inside the shaft equaled the channel water elevation.
An investigation was undertaken to determine the cause of the collapse and
included consideration of the strength of the frozen ground at the time of the
collapse. The shaft soils consisted of dense, fine sand with shell fragments and
mica. Soil borings were conducted and soil and water samples from the shaft
were sent to Michigan Technological University (MTU) for strength, creep,
and salinity testing. The project required a relatively short turnaround time.
This paper provides the details of the testing procedures and results. Additional
details of the frozen ground shaft collapse can be found in a Civil Engineering
Magazine article titled “Deep Freeze” [1].
Site Geology
The Los Angeles Port is located along the coastal part of the Los Angeles Ba-
sin, which is filled mainly with sediments transported from the surrounding
mountains and hills, as well as hydraulic fill. Sections of both Terminal and
Mormon Island, across the East Channel Basin from Terminal Island, had been
extensively hydraulically filled during the development of the coastline. The
islands, however, are composed of late Quaternary deposits commonly present
in the Los Angeles Basin that have been grouped together and mapped as the
Lakewood Formation. The Lakewood Formation deposits consist primarily of
silty sands and sands with some gravel and limited thin clay layers. The granu-
lar soils within the Lakewood Formation commonly have a greenish gray or
olive color.
VITTON AND MUSZYNSKI, doi:10.1520/STP156820130034 155
Soil borings at the location of the Terminal Island shaft revealed that the
shaft’s soils consisted of dense fine to coarse sand with up to 10 % silt content,
with occasional thin stiff to very stiff lean clay layers at depths between 10.6 m
and 12.8 m. The collapsed section of the shaft, which was at a depth of 18 to
24 m, consisted of dense fine sand with silt, with abundant shells and some
mica. The boring log’s standard penetration test raw blow count (N) values for
this section ranged from 50 to 80 with an average value of 60. The in situ satu-
rated unit weights ranged from 19.1 to 21.1 kN/m3, with moisture contents
varying from 20 % to 27 %.
Experimental Program
Soil Analysis and Classification

After the shaft’s collapse, a number of soil borings were made to better delin-
eate the extent of the collapse, as well as to provide in situ soil samples for lab-
oratory testing. Twenty-one soil samples from three boreholes, along with a
3.8 -l bottle of water obtained from the Terminal Shaft, were sent to MTU for
testing. The samples were obtained using Pitcher and Shelby tube samplers
from depths between 16.8 and 25.6 m. Eight soil specimens were tested in the
zone near the main collapse, which was between 19.7 and 23.1 m. The speci-
men identification (ID) numbers, depths, modes of freezing, and test loading
rates are shown in Table 1. The ID prefix “MB3” refers to the soil boring num-
ber from which the samples were obtained. Note that an additional, ninth speci-
men (MB3 18) was damaged and therefore used only for water salinity testing
(see Table 2).
Soil identification and classification included grain size analysis, solid par-
ticle density, salinity, frozen unit weight, and moisture content. Grain-size dis-
tribution curves for seven of the nine samples are shown in Fig. 1. The fines
contents for six of the seven samples averaged 3.6 % with a low of 2.8 % and a
TABLE 1—Identification of soil samples tested, depth, and testing modes.
ID Number Borehole Depth, m Freezing Mode Loading Rate

5 MB3 19.7 Uniaxial 70% creep
6 MB3 19.9 Uniaxial 1%/min
8 MB3 20.4 Uniaxial 40% creep
9 MB3 20.5 Multiaxial 1%/min
13 MB3 21.6 Uniaxial 0.1%/min
17 MB3 23.1 Uniaxial 0.1%/min
TABLE 2—Salinity, particle density, unit weight, and sample moisture.
Specimen Salinity, Freezing Particle Frozen Unit Moisture

ID ppt Mode Density, g/cm3 Weight, kN/m3 Content
MB3 5 31 Uniaxial 2.72 19.82 18.2
MB3 6 32 Uniaxial 2.72 19.95 17.8
MB3 8 31 Uniaxial 2.72 19.12 18.9
MB3 9 Not tested Multiaxial 2.73 19.89 23.0
MB3 13 32 Uniaxial 2.67 18.96 20.1
MB3 14 32 Uniaxial 2.66 19.75 17.8
MB3 16 No water Uniaxial 2.67 Not tested 20.4
MB3 17 No water Uniaxial 2.69 19.59 17.8
MB3 18 32 Uniaxial Not tested Damaged Damaged
high of 4.1 %. One sample (MB3 8) had a fines content of 10.3 %. Digital pho-
tographs of samples MB3 8 and MB3 16 are shown in Fig. 2.
The salinity measurements were made using an ATAGO 2412-W04 salin-
ity refractometer. There were, however, some anomalies noted: specimens 16
and 17 did not contain water when the wax seals were removed from the metal
tubes, and specimen 9 was tested before the salinity refractometer reading had
been obtained. In general, seawater contains about 3 % to 3.3 % salt, or 30 to
33 parts per thousand (ppt). All of the soil water tested ranged between 31 and
32 ppt, indicating that it was seawater. The soil particle density was measured
using a Micrometrics 1330 helium pycnometer. The procedures used with this
instrument are outlined in Ref 2. Frozen unit weight measurements were made
using a caliber to measure the volume of the frozen soil and a mass balance.
The moisture measurements were made following ASTM D2216–10 [3]. The
results of these analyses are provided in Table 2.
FIG. 1—Grain-size analysis.

FIG. 2—Digital photographs of (a) MB3 8 and (b) MB3 16.
Soil Preparation and Freezing

The soil samples were received and stored at approximately 20 C in the soil
mechanics laboratory at MTU. All of the samples were received in metal
extraction tubes in either 76.2-mm-diameter (Pitcher) tubes or 63.5-mm
(Shelby) thin-walled sampling tubes with the ends sealed with wax. Because
the soil consisted of sands, it was decided to freeze the soil in the extraction
tubes to avoid further disturbance and then extract the frozen soil from the
tubes after freezing. Uniaxial and multiaxial soil freezing was conducted in a
walk-in cold room (Model SPIF313, International Cold Storage, Inc.). An esti-
mate of the ground temperature at the time of failure was determined to have
been no less than 10 C. Therefore, the control temperature of the room was
set at 10 C, and subsequently all samples were frozen to 10 C and tested at
10 C.
To obtain a quick estimate of the soil’s frozen strength, specimen MB3 9
was frozen multiaxially by simply being placed in the cold room at 10 C.
The specimen was placed in a vertical position in a pan with porous stone
placed between the tube and the pan, which allowed the entire tube to be
exposed to the cold temperature. During the freezing process, water was
observed being expelled from the tube. The specimen was frozen for approxi-
mately four days. Extraction of the frozen specimen was accomplished by mak-
ing a thin saw cut along the length of the tube with a table saw, which was also
located in the cold room. The saw was adjusted so that it would cut only the
metal tube. A single cut was made along the length of the tube. Once the tube
was cut, it was gently widened and the specimen was removed. Immediately
after removal, the specimen was wrapped in freezer-grade plastic wrap and
placed in a freezer-grade bag. To prevent any sublimation from occurring prior
to or during testing, the specimen was placed in an environmental testing
chamber (Russell Environmental Chamber, Model RB-3-1) that was inserted
into the loading system to a conduct uniaxial compression strength test. The
environmental chamber was set at 10 C and had a temperature deviation of

60.1 C. The specimen was kept at 10 C for one week prior to uniaxial com-
pression strength testing to ensure that the specimen was fully frozen.
The remaining seven samples were frozen uniaxially using the same proce-
dure described in Ref 4 at the Cold Regions Research Center in Hanover, NH.
In this procedure, to freeze the soil, cold temperature is applied to the top of
the sample while a warmer temperature (above 0 C) and constant-head water
supply are maintained at the base of the sample, allowing the specimen to
freeze from the top down. Because of the cohesionless nature of the soil, it was
decided once again to freeze the soil in the metal extraction tubes to avoid dis-
turbing the soil, although it was recognized that the metal tube would act as a
heat conductor as well as induce some compression in the soil during freezing.
Steps were taken, however, to minimize these effects; this is discussed further
elsewhere.
Another important requirement was a constant-head water supply to main-
tain a saturated condition in the soil as it froze and water was drawn upward to
the freezing front as the freezing front advanced downward. As in the Cham-
berlain method, a Marriott water supply was used to provide a constant-head
water supply to the soil during freezing. As shown in Fig. 3, the test setup was
placed into the cold room, which was maintained at 10 C.
The test setup consisted of the following four parts: (1) a constant-head
water supply to provide water during the freezing process in order to maintain
the soil in a saturated condition, (2) a warm temperature bath to keep the base
of the soil samples at a temperature above freezing, (3) an aluminum base plate
to hold the metal extraction tubes and provide access for the water from the
constant-head water supply, and (4) a large block of Styrofoam that was
FIG. 3—Freezing chamber inside cold room with Marriott water supply.
hollowed out to hold the metal tubes steady and brace the insulation surround-
ing the soil specimens while exposing the top of the soil tubes to the cold-room
freezing temperature of 10 C via metal “antennas” extended through the
insulation. A Plexiglas box was used as the base of the test apparatus, where at
the start of the test the water was maintained at þ4 C. The warm water was fed
to the Plexiglas box by an Endocal refrigeration bath, shown in Fig. 4(a), which
was placed outside the cold room with plastic tubes from the bath to the Plexi-
glas box to allow a water–antifreeze solution to circulate into the Plexiglas
box. The aluminum plate was manufactured to hold the eight metal tubes for
freezing as shown in Fig. 4(b). The plate was constructed to fit on the top of
the Plexiglas box but also be in contact with the circulating fluid from the
warm temperature bath. The plate was sealed to prevent the circulating warm-
ing fluid from accessing the metal tubes and the soil samples. The extraction
tube plastic end caps were used to attach the metal tubes to the aluminum plate.
Holes were drilled in the centers of the end caps to allow water to access the
soil from the Marriott constant-head water supply. Plastic connections with
seals were placed in the holes to connect the tubes from the Marriott water sup-
ply, and also to prevent water leakage from the soil tubes. The eight water
FIG. 4—Pictures of experimental setup: (a) constant-head water supply next to

the walk-in cold room; (b) aluminum base plate with end cap holders; (c) metal
tubes with soil and freezing antennas; and (d) insulation system to allow speci-
mens to freeze uniaxially with thermal couple sensors.
supply tubes leaving the environmental chamber were bundled together along
with a water pipe heating cord, which was used to prevent the water supply
tubes from freezing. The eight tubes were then connected to a distribution sys-
tem that was attached to the Marriott water supply as shown in Fig. 3. The Mar-
riott tube was constructed as a closed tube filled with water with a tube inserted
through the top of the Marriott tube and into the water. Before the soil tubes
were placed on the aluminum plate, as much air as possible was eliminated
from the supply tubes by allowing water to move through the system. The
water was extracted from the end caps as shown in Fig. 4(b). Water supplied
from the project site was used in the Marriott tube to supply water to the soil.
To prepare the soil specimens for freezing, the wax from the top of the
tubes was removed, while the bottom cap remained in place. In some tubes the
soil was a number of centimeters below the top of the tube, so a pipe cutter was
used to trim the tube approximately 12 mm above the soil. The tube-cutting
procedure was conducted while the soil tube was kept in a vertical position. It
was assumed that at the bottom of the tube, which was the cutting end, the soil
would be in contact with the end cap. To place the tube on the aluminum plate,
the bottom cap was carefully removed. It was not certain that this procedure
would work, as it was unclear what the condition of the soil in the tube would
be. However, in all cases, the soil remained firmly in the tube and was not dis-
turbed by the tube’s placement on the aluminum plate. Prior to placement of
the tubes on the aluminum plate, a porous stone and filter paper were placed
between the end cap on the plate and the bottom of the soil. Once the soil tube
was in place in the end cap, a silicone sealant was applied around the connec-
tion to prevent leakage. Once the sealant had cured, water from the project site
was added to the soil from the top of the tube to saturate the soil prior to freez-
ing. At this point, the elevation of the Marriott tube was also adjusted to pro-
vide a constant water elevation within the soil tubes. At full saturation,
aluminum antennas were placed on top of the soil as shown in Fig. 4(c). The
aluminum antennas consisted of 60-mm-diameter solid aluminum stock for the
63.5-mm tubes and 75-mm-diameter solid aluminum stock for the 76.2-mm
tubes. A strip of insulation was added to the side of the antenna in contact with
the soil in an attempt to minimize the heat flow between the aluminum anten-
nas and the metal tube. This insulation can also be seen in Fig. 4(c). Once the
antennas were in place, a Styrofoam chamber was placed over the base of the
freezing chamber. Once the Styrofoam container was in place, perlite insula-
tion was placed around the tubes to insulate the metal tubes and allow the soil
to freeze from the top down as shown in Fig. 4(d). Ten thermocouples were
used to monitor the temperature at various locations during the freezing pro-
cess. The thermocouple locations are provided in Table 3.
We began the freezing process by placing the test setup with the soil speci-
mens into the walk-in cold room, which was set at 10 C. At the start of the
freezing process, the temperature at the base of the soil samples was measured
TABLE 3—Thermocouple locations.
Thermocouple Number Location

1 Cold room temperature 12 in. above aluminum antennas
2 In water bath fluid below aluminum plate near specimen 6
3 Inserted in the soil at the bottom of soil specimen 17
4 In water bath fluid below aluminum plate between specimens 17 and 18
5 Inserted in soil at the top of soil specimen 17
6 Placed on aluminum plate between specimens 6, 5, and 16
7 Placed in the Endocal refrigeration bath fluid (outside cold room)
8 On top of aluminum antenna on soil specimen 17
9 Within insulation between soil tubes
10 On top of aluminum antenna on soil specimen 13
at þ4 C, the temperature of the water bath below the soil specimens. During
freezing, the temperature of the soil gradually declined and reached 0 C in
about 26 h. After about 37 h, the bottom soil temperature reached 2 C. At
39 h the temperature of the Endocal refrigeration bath was lowered to 6 C to
complete the freezing of the soil. The soil specimens remained in the freezing
chamber for an additional 20 h; the top of the soil was at approximately
10 C, and the bottom of the soil was now at 8 C. After 60 h of freezing, the
perlite insulation and the Styrofoam box were removed, exposing the full
length of the soil tubes to the 10 C temperature of the cold room for an addi-
tional 24 h.
While still in the cold room, the metal tubes were removed from the freez-
ing setup. To extract the frozen soil from the metal tubes, the tubes were
quickly immersed in a bucket of warm water. The ends of the tubes were sealed
first so no water could access the frozen soil (note that this extraction method
is different from the preliminary method used with specimen MB3 9, as
described previously). The first test for the warm water extraction method was
conducted on specimen MB3 16 having a 76.2-mm-diameter tube. After the
tube had been in the water for about 20 to 30 s, the frozen soil was extracted by
being pushed out of the metal tubes. This method was successfully used in all
subsequent soil specimen extractions. In visual examinations, the extracted soil
specimens appeared to be completely frozen. As soon as the specimens were
extracted from the metal tubes, they were placed in an aluminum block for
trimming to a 2:1 length-to-diameter ratio (L/D) and paralleling of the ends.
Immediately after the specimens were trimmed, they were wrapped in freezer-
grade plastic wrap and placed in freezer-grade ziplock plastic bags and trans-
ferred to the Russell environmental testing chamber, where they were stored at
10 C until testing. During the trimming process, however, specimens MB3
18 and MB3 6 were damaged. Specimen MB3 18 was unusable, but specimen
MB3 16 was recut to a shorter length and was no longer at a 2:1 length-to-
height ratio for testing in uniaxial compression. Of the usable specimens, six
were tested in uniaxial compression at two different loading rates, and two
were tested in creep at two different loads.
Compression and Constant Stress (Creep) Testing

The uniaxial compression strength and creep testing was conducted using a
55-kip MTS 810 Material Test System (MTS) with a TestStar II digital con-
troller. A Russell environmental chamber, which was designed to be inserted
into the MTS, was used to maintain a 10 C temperature for testing. The
testing system is shown in Fig. 5. The uniaxial compression tests were con-
ducted in displacement control, whereas the creep tests were conducted in
load control.
Six of the eight frozen specimens were tested in unconfined compression
using displacement control; four tests were conducted at a strain rate of
1 %/min (MB3 6, 9, 14, and 16), and two tests were at a strain rate of
0.1 %/min (MB3 13 and 17). The strain rate for each test was based on the
length of each specimen, and therefore the rate was changed slightly for each
test because of small variations in the length of each specimen.
The two creep tests were conducted based on 40 % (MB3 8) and 70 %
(MB3 5) of the maximum uniaxial compression strength of the 1 % strain rate
unconfined compression tests. For the creep test, specimens were loaded at a
1 % strain rate until the stress reached 40 % or 70 % of the maximum compres-
sive strength (and then the stress was held constant until failure). The
tests were discontinued after significant displacement had occurred. All of
the tests were conducted, and data were acquired, using the MTS software
Testware SX.
FIG. 5—MTS climate control system for compression testing.

FIG. 6—Stress-strain results for uniaxial compression tests at (a) 1 % strain

rate and (b) 0.1 % strain rate.
Results
The stress-strain results for the 1 % and 0.1 % strain tests are shown in
Figs. 6(a) and 6(b), respectively, as well as in Table 4. The results of the com-
pression tests show that the maximum stress occurred at a strain between 6 %
and 8 % for the 1 % strain rate specimens [Fig. 6(a)], whereas the maximum
stress for the 0.1 % specimens [Fig. 6(b)] occurred at a strain around 4 % to
5 %. Specimen MB3 9 displayed the lowest maximum strength of around
800 kPa, whereas the remaining specimens tested at a 1 % strain rate had
strengths of 1800 kPa and up to 2920 kPa. The specimens loaded at a rate of
0.1 % had maximum strengths of about 1385 kPa.
TABLE 4—Summary of strength and creep testing results.
Specimen Freezing Loading Rate, Compressive Time to

ID Mode Strain/min Strength, kPa (psi) L/D Failure, min
9 Multiaxial 1% 890 (130) 2.01 —
6 Uniaxial 1% 1830 (265) 1.99 —
14 Uniaxial 1% 1900 (275) 2.06 —
16 Uniaxial 1% 2920 (425a) 1.05 —
13 Uniaxial 0.1% 1450 (210) 1.96 —
17 Uniaxial 0.1% 1310 (190) 1.94 —
5 Uniaxial Creep 70% 1290 (185b) 2.00 55
8 Uniaxial Creep 40% 750 (108c) 2.00 800
18 Uniaxial Damaged Not tested — Not tested
a
Specimen was damaged in the freezing process and had to be cut to a shorter length, which resulted in an
L/D of 1.05 instead of the required 2:1.
b
Constant creep load, which was 70% of the uniaxial compressive strength at a 1%/min loading rate.
c
Constant creep load, which was 40% of the uniaxial compressive strength at a 1%/min loading rate.
FIG. 7—Axial compression creep test (MB3 8) at constant stress of 40 % of

maximum compressive strength.
FIG. 8—Axial compression creep test (MB3 5) at constant stress of 70 % of

maximum compressive strength.
TABLE 5—Comparison with compressive strength results from other investigations on frozen saline
soils.
Loading Rate, Temperature, Uniaxial Compressive

Reference Soil %/min C Strength, kPa (psi)
This study Dense fine sand 1.0 10 1865 (270)
[5] Alluvial sand 1.0 10 2016 (292)
[5] Toyoura sand 1.0 10 2460 (356)
[5] Coarse sand 1.0 10 2352 (340)
[6] Ottawa sand 0.6 9.6 1488 (215)
[7] Uniform sand 0.51 11 2208 (320)
[7] Mortar sand 0.31 6.2 912 (132)
This study Dense fine sand 0.10 10 1385 (200)
[8] Uniform fine sand 0.013 10 672 (97)
[8] Very silty sand 0.013 10 768 (111)
[8] Very silty sand 0.013 10 576 (83)
The creep test for 40 % of the maximum compressive strength is shown in

Fig. 7 (MB3 8), and the creep test at 70 % of maximum compressive strength
is shown in Fig. 8 (MB3 5). Specimen MB3 8, tested at 40 % of the maximum
compressive stress (745 kPa), reached tertiary creep, with characteristic large
strains, beginning at about 1200 min, whereas specimen MB3 5, tested at 70 %
of the maximum strength (1300 kPa), reached the onset of tertiary creep at
approximately 70 min.
Discussion and Conclusions

The primary goal of the testing was to assess the strength of the frozen soil and
the creep properties from the collapsed shaft. Although only eight specimens
were tested, the results indicated that the frozen soil had strength consistent
with that of soil frozen with seawater. The water taken from the shaft was
measured to have the same composition as seawater, with a salinity of 3.2 %.
The unconfined compression tests ranged from a low of 896 kPa (130 psi) for
specimen MB3 9 (multiaxially frozen without a water supply) to a high
of 2920 kPa (423 psi) for specimen MB3 16, which was damaged and cut to an
L/D of approximately 1, rather than the target L/D of about 2 (Table 4). The
average for the remaining two tests employing 1 % strain per minute (MB3 6
and 14) was 1865 kPa (270 psi), and for the two 0.1 % strain tests the average
was 1365 kPa (200 psi).
Table 5 provides a comparison of the average compressive strength results
obtained in this testing and results obtained by others when testing at about a
10 C temperature using loading rates of 1 % to as low as 0.013 %. While
there are a number of factors that affect the strength of frozen soil, such as
loading rate and temperature, the results reported in the current study are com-
parable to those obtained by others, as shown in Table 5. Note that the average
compressive strength of 1385 kPa observed in this study for the 0.1 %/min
loading rate [Fig. 6(b)] is notably greater than the results from Ref 8 (see Table
5), which, however, are from constant strain tests and not constant stress tests.
Furthermore, the loading rate of 0.1 %/min for specimens MB3 13 and MB3
17 is approximately eight times greater than the loading rate utilized in Ref 8.
The creep test results from this current testing program show that the creep
compression strength, which is defined as the onset of tertiary creep, was about
760 kPa (110 psi) for both the 40 % and 70 % load cases. However, these creep
tests clearly show the time-dependent character of frozen soils, indicating how
the stress level dramatically affects the time required in order for tertiary creep
and failure to develop in the specimen. With a stress of 70 % of the maximum
compressive strength, tertiary creep/failure took place nearly 20 times more
quickly under these conditions.
Finally, it appears that the specimen preparation procedure used in this
research, in which the soils were frozen while in their extraction tubes, is a
viable method for applying freezing conditions and forcing a “top-down” freez-
ing front. We note that further research is needed in order for the role of con-
finement during the freezing process to be better understood, as well as the
possible changes in soil fabric associated with obtaining Pitcher and Shelby
tube specimens on site and then subsequently refreezing them for later testing.
After the specimens had been frozen, the extraction procedure of subjecting the
frozen specimens to a warm water bath proved to be an effective recovery
method.
References
[1] Berti, D. J., Lindquist, E. S., and Roesner, L., “Deep Freeze,” Civ. Eng.
Mag., February, 2002, pp. 68–74.
[2] Vitton, S. J., Lehman, M. A., and Van Dam, T. J., “Automated Soil Parti-
cle Specific Gravity Analysis Using Bulk Flow and Helium Pycnometry,”
Nondestructive and Automated Testing for Soil and Rock Properties,
ASTM STP 1350, W. A. Marr and C. E. Fairhurst, Eds., ASTM Interna-
tional, West Conshohocken, PA, 1998, pp. 3–13.
[4] Chamberlin, E. J., “A Freeze-Thaw Test to Determine the Frost Suscepti-
bility of Soils,” CRREL Special Report No. 87-1 and U.S. Dept. of Trans-
portation, Federal Aviation Administration, Report No. DOT/FAA/PM-
85/20, U.S. Cold Regions Research and Engineering Laboratory, Hanover
NH, 1987.
[5] Ogata, N., Yasuda, M., and Kataoka, T., “Salt Concentration Effects on
Strength of Frozen Soils,” Proceedings of the Third International Sympo-
sium on Ground Freezing, Hanover, NH, June 22–24, 1982, U.S. Army
Corps of Engineers Cold Regions Research and Engineering Laboratory,
pp. 3–10.
[6] Pharr, G. M. and Merwin, J. E., “Effects of Brine Content on the Strength
of Frozen Ottawa Sand,” Cold Reg. Sci. Technol., Vol. 11, 1985, pp.
205–212.
[7] Andersland, O. B. and Ladnyi, B., An Introduction to Frozen Ground En-
gineering, Chapman & Hall, New York, 1994.
[8] Hivon, E. G. and Sego, D. C., “Strength of Frozen Saline Soils,” Can.
Geotech J., Vol. 32, 1995, pp. 336–354.
CLASSIFICATION AND EFFECTS
OF MECHANICAL PROPERTIES
ON PERFORMANCE
STP 1568, 2013
DOI:10.1520/STP156820130017
Benjamin Still,1 Sam Proskin,2 Hannele Zubeck,1 and Zhaohui Yang1
Frozen-Soil Classification With Index Testing
REFERENCE: Still, Benjamin, Proskin, Sam, Zubeck, Hannele, and Yang,

Zhaohui, “Frozen-Soil Classification With Index Testing,” Mechanical Proper-
ties of Frozen Soils, STP 1568, Hannele Zubeck and Zhaohui Yang, Eds., pp.
ABSTRACT: Classification of frozen soils was first developed by the U.S.
Army Corps of Engineers’ Cold Regions Research and Engineering Labora-
tory (CRREL) together with the Division of Building Research, National
Research Council, Canada. This visual method of classification was adopted
by ASTM in 1983 as designation D4083, currently ASTM D4083-07:
Standard Practice for Description of Frozen Soils (Visual-Manual Procedure),
Annual Book of ASTM Standards, ASTM International, West Conshohocken,
PA. The current visual classification standard does not use any engineering
index testing to classify the soils. This leaves the engineer with a qualitative
assessment of soil–ice mass type, strength, and stability in which a more
conservative and expensive design may be considered and even worse, an
under design may occur. The engineer is in need of some index properties
that can be readily measured to provide a more objective method to classify
the soil–ice mass. This paper investigates the relationship between water
content and density to frozen-soil classification. It was found that the water/
ice content and dry density within the same frozen-soil class varies signifi-
cantly, which may lead to different mechanical behavior or thaw settlement.
Therefore, reporting the water (ice) content and frozen-soil density together
with the frozen-soil classification helps the engineer to better evaluate thaw
settlement and assess need for further testing.
KEYWORDS: frozen soil, classification, visual method, density, water
content
Manuscript received January 18, 2013; accepted for publication June 13, 2013; published online
August 23, 2013.
1
America.
2
NOR-EX Ice Engineering Inc., Calgary, Alberta, T3A 1P7, Canada.
3
Jacksonville, FL.
Copyright V
19428-2959.
169
Introduction
Classification of frozen soils was first developed by the U.S. Army Corps of
Engineers’ Cold Regions Research and Engineering Laboratory (CRREL) with
the Division of Building Research, National Research Council, Canada. This
visual method of classification was adopted by ASTM International in 1983 as
designation D4083, currently ASTM D4083-07 [1]. The current visual classifi-
cation standard does not use any engineering index testing to classify the soil-
ice mass. This leaves the engineer with a qualitative assessment of soil–ice
mass type, strength, and stability in which a more conservative and expensive
design may be considered and even worse, an under design may occur. The
engineer is in need of some index properties which can be readily measured to
provide a more objective method to classify the soil–ice mass. The objective of
the work performed at the University of Alaska at Anchorage is to develop an
index test to improve the current visual classification method. This paper reports
the first part of the work; the investigation of the relationship between the water
(ice) content and dry density of the frozen soils obtained from two sites in Alaska
as a means to quantifying the current visual classification system.
Test Methods
Naturally frozen soil specimens were sampled at two locations, the CRREL
Permafrost Tunnel in Fox, AK and the Campbell Creek Bridge at Elmore Rd.
in Anchorage, AK. Permafrost specimens were obtained from the CRREL Per-
mafrost Tunnel and seasonally frozen soils were obtained from the Campbell
Creek Bridge site. A detailed description of the sampling and machining tech-
niques used can be found in the companion paper in this STP [2], which fol-
lows work done by Baker [3]. Cylindrical samples with a 51-mm diameter by
102-mm length were machined out of block samples collected at the field loca-
tions as seen in Figs. 1 and 2. After machining, these specimens were visibly
classified in accordance with ASTM D4083-07 [1]. The length of the specimen
was measured three times at different locations and averaged and the diameter
was measured three times at three different locations and averaged. The speci-
men measurements were typically within 0.02 mm, which means a nice smooth
surface had been achieved by machining. The above measurements were used
to determine the volume of the cylinders. Uniaxial compression tests were con-
ducted on these specimens for another study and after testing the specimens’
mass was measured and the specimens were put in the oven for at least 24 h to
find the water content following ASTM D2216-10 [4]. The dry density of the
soil was calculated knowing the frozen bulk density of the specimens and the
mass after drying.
Several sieve analyses and hydrometer tests were conducted for both the
permafrost and seasonally frozen soils. Figures 3 and 4 present the respective
FIG. 1—Typical ice-rich seasonally frozen soil specimen.
grain size distributions for the permafrost and seasonally frozen soil. Both soils
contain large amounts of fines. The ASTM D2487 [5] soil classification of the
permafrost sample was silt (ML), whereas the seasonally frozen soil was classi-
fied as sandy organic silt (OL), with several specimens being classified as peat
(PT) because of their highly organic nature.
Data
The data for the testing procedures described above are seen in Table 1. Forty-
five seasonally frozen soils and 23 permafrost specimens were tested in this
FIG. 2—Typical ice-rich permafrost soil specimen.

FIG. 3—Grain size distribution of permafrost specimen.
study. The water content of the specimens ranged between 62 %–225 %. The
frozen-soil classifications of each cylinder were almost all in the V (visible ice)
classification with only a few exceptions classified as Nbe (non-visible,
bonded, excess ice) (Fig. 5). The dry density of the samples ranged between
320–941 kg/m3.
FIG. 4—Grain size distribution for seasonally frozen soils.

TABLE 1—Testing data.
Water Frozen
Specimen Content Bulk Density Dry Density Frozen-Soil Classification
ID (%) (kg/m3) (kg/m3) Per ASTM D4083 [1] USCS
P11 V2 62 1525 941 Vx ML
P4 V2 62 1319 816 Nbe ML
P6 H2 72 1460 847 Nbe ML
P4 V3 72 1442 840 Vx ML
P11 V3 73 1450 837 Vr ML
P8 V2 74 1457 839 Vr ML
P11 V1 75 1460 835 Vr ML
P4 V4 76 1420 809 Vr ML
P9 H2 80 1426 791 Vr ML
P6 V2 84 1454 792 Vr ML
P8 H1 84 1415 770 Vr,s ML
P6 H3 85 1408 762 Vr,s ML
C1 V2 86 1452 780 Vr,s OL
P8 V1 88 1380 733 Vs ML
C9 V2 93 1352 700 Vr OL
C10 H1 93 1357 702 Vr,s OL
P8 H2 93 1405 728 Vr ML
C2 V2 94 1343 691 Vx,s OL
C10 H3 96 1334 683 Vr OL
P4 V1 100 1341 671 Vr,s ML
C10 H2 103 1302 640 Vr OL
C1 V1 103 1302 642 Vr OL
C9 H4 103 1305 644 Nbe,Vx PT
P9 H1 106 1319 641 Vr,s ML
P9 V4 108 1320 634 Vr,s ML
C9 V4 109 1294 620 Vs,r OL
C9 V5 112 1247 589 Vs,r OL
P6 H1 114 1299 606 Vr ML
C2 V3 115 1273 592 Vx,s OL
C6 V3 116 1270 589 Vr,s OL
C10 H4 118 1272 584 Vr,s OL
C10 V1 118 1271 583 Vs,r OL
C1 H1 120 1254 570 Vs OL
C2 V1 120 1276 580 Vs OL
C2 H2 121 1269 574 Vs OL
C9 V3 121 1271 574 Vs OL
C8 H4 122 1258 566 Vs OL
C7 V1 124 1226 547 Vs OL
C10 H6 124 1257 562 Vr OL
C5 H2 125 1251 555 Vr,s OL
C8 H3 129 1241 543 Vs OL
C2 H1 127 1254 551 Vs OL
TABLE 1—Continued
Water Frozen
Specimen Content Bulk Density Dry Density Frozen-Soil Classification
ID (%) (kg/m3) (kg/m3) Per ASTM D4083 [1] USCS
C6 H2 128 1237 543 Vr,s OL
C5 H1 128 1230 539 Vr,s OL
C11 V4 130 1271 553 Vs,r OL
P4 H3 130 1255 545 Vr ML
C9 V1 131 1204 522 Vs,r OL
P4 H1 134 1251 534 Vr,s ML
C6 V4 136 1217 515 Vr,s OL
C9 V6 136 1216 516 Vs,r OL
C11 V1 141 1225 509 Vs,r OL
C9 H1 145 1210 493 Vs OL
C9 H2 145 1201 491 Vs OL
C6 H1 155 1181 464 Vr,s OL
C4 V2 164 1167 442 Nbe OL
C8 H1 162 1179 449 Vr,s OL
C4 V3 154 1177 463 Nbe,Vx OL
C8 H2 169 1161 431 Vr,s OL
C7 H1 170 1176 436 Vs OL
C6 V1 183 1120 396 Vr,s OL
C6 V2 192 1124 385 Vr,s PT
C4 V1 200 1109 370 Vr PT
C6 H3 205 1111 364 Vr,s OL
C6 H4 225 1038 320 Vr,s PT
FIG. 5—The left sample is classified as Nbe while the right is Vs.
FIG. 6—Water contents (%) for soil specimens classified as Vr (ML, OL,
PT ¼ USCS class).
Results
Several authors have shown that frozen soil’s mechanical properties, as well as
possible thaw settlement are dependent on ice content and density (summarized
by Andersland and Ladanyi [6]). For example, compressive strength of frozen
soil first increases and then decreases with increasing ice content. Based on ba-
sic weight–volume relationships, water content (or ice content) is related to the
bulk density and dry density as given in Eq 1
qd ¼ q=ð1 þ xÞ (1)
where qd is the dry density of the frozen-soil specimen, q is the frozen bulk den-
sity of the soil specimen, and x is the water content of the frozen-soil specimen.
Because the water content is related to the dry and bulk density of frozen soil,
one could investigate one or the other depending on the case of interest. The
point that is made here, is that the water (ice) content or the dry density vary sig-
nificantly within a frozen-soil class as can be seen from Figs. 6 and 7. This
means that frozen soils classified with the current visual method will have differ-
ent mechanical properties and will behave differently under thawing. The speci-
mens classified as ML according to the USCS classification method and Vr with
the frozen-soil classification method have water content ranging from 73 % to
130 % and dry densities from 839 to 545 kg/m3 or frozen bulk densities from
1457 to 1255 kg/m3. If one estimates thaw settlement with, e.g., Eq 2 [6] for a
silty soil with 1.2 < qf/qw < 2.0, the DH/Hf varies from 0.32 to 0.52, which indi-
cates over 60 % increase in the thaw strain and could correspond significant dif-
ference in the settlement depending on the thickness of the thawing soil layer,
0:5
DH qf
¼ 0:80 0:868 1:15 (2)
Hf qw
FIG. 7—Dry densities (kg/m3) for soil specimens classified as Vr (ML, OL,
PT ¼ USCS class).
where DH/Hf is the vertical strain of soil element thawed, Hf is the thickness of
the thawing soil layer, and qf and qw are the frozen bulk density of the frozen
soil and the density of water, respectively.
Similar observations could be drawn for other frozen-soil classes. Figures
8 and 9 show water (ice) contents and dry densities for frozen-soil class Vr,s.
FIG. 8—Water contents (%) for soil specimens classified as Vr,s (ML, OL,
PT ¼ USCS class).
FIG. 9—Dry densities (kg/m3) for soil specimens classified as Vr,s (ML, OL,
PT ¼ USCS class).
Figures 10 and 11 show water (ice) contents and dry densities for frozen-soil
class Nbe.
All frozen-soil classes contained the same USCS soils, ML, OL, and PT.
This is, of course, natural as only two sampling locations were used, but signif-
icant in a way that the same soils in some cases had visible ice and in other
cases non-visible ice. Therefore, the USCS classification is not used alone to
estimate the existence of the segregated ice or the ice content in frozen soils. If
water (ice) contents are compared within the V and N class, they varied
between 62 %–225 % in V and 62 %–164 % in N class. Thus, both classes had
FIG. 10—Water contents (%) for soil specimens classified as Nbe (ML, OL,
PT ¼ USCS class).
FIG. 11—Dry densities (kg/m3) for soil specimens classified as Nbe (ML, OL,
PT ¼ USCS class).
specimens within the same ranges, even if the V class soils had higher ice con-
tents. Figure 12 shows water contents for all ML soil samples, and it can be
seen that there is a large variation within the frozen-soil classes. Similar obser-
vation was drawn for OL soil samples.
FIG. 12—Water contents (%) for all ML soil samples.

Conclusions
Frozen-soil water content, dry density, and frozen bulk density were considered
as index properties to quantify the current visual frozen-soil classification sys-
tem. The following conclusions were found:
• A large variation of water contents and densities exist in a particular
frozen-soil class.
• Water contents and densities can be similar in different frozen-soil
classes.
• Based on the water contents (or ice contents in frozen state), soils’ me-
chanical properties and possible thaw settlement can vary significantly

within a frozen-soil class.
• The current visual frozen-soil classification is not adequate alone to clas-
sify frozen soils.

• Including the water content, dry and bulk density with the visual soil
classification helps engineer in thaw settlement estimations and assess-

ing a need for mechanical testing of the frozen soil.
The writers aim to collect more data from various locations to create an
actual new index test or graphical curves to categorize frozen soils having simi-
lar properties.
References
[1] ASTM D4083-07: Standard Practice for Description of Frozen Soils

[2] Still, B., Yang, Z., and Ge, X., “Sampling, Machining and Testing of Nat-
urally Frozen Soils,” ASTM STP 1568, ASTM International, West Con-
shohocken, PA, 2013.
[3] Baker, T. H. W., “Transportation, Preparation, and Storage of Frozen Soil
Samples for Laboratory Testing,” ASTM STP 599, ASTM International,
West Conshohocken, PA, 1976, pp. 88–112.
[5] ASTM D2487: Standard Practice for Classification of Soils for Engineer-
ing Purposes (Unified Soil Classification System), Annual Book of ASTM
Standards, ASTM International, West Conshohocken, PA.
ASCE, Reston, VA, 2004.
STP 1568, 2013
DOI:10.1520/STP156820130016
Anatoly Sinitsyn1
Investigation of the Dynamic Behaviors of

Frozen Saline Silt with the Use of a Spherical
Stamp
REFERENCE: Sinitsyn, Anatoly, “Investigation of the Dynamic Behaviors of

Frozen Saline Silt with the Use of a Spherical Stamp,” Mechanical Properties
of Frozen Soils, STP 1568, Hannele Zubeck and Zhaohui Yang, Eds., pp.
ABSTRACT: The objective of this work was to study the dynamic interactions
between a spherical indenter and frozen soil. Indentation via a spherical
stamp was used for various dynamic applications with equal kinetic energy.
This research is an attempt to add to the theoretical background for pile driv-
ing in permafrost. The focus is on the supposition that pile driving performed
by a heavier hammer falling with a low velocity is more efficient than that
done by a light hammer falling from a greater height at an equal kinetic
energy. Monitoring of the force acting on the stamp during dynamic loading
was performed for the first time with the use of a thin, flexible sensor placed
between the stamp and the frozen soil sample. A plastic ball with a diameter
of 25 mm was used as the spherical stamp, and the dynamic load on the
stamp was provided by a falling dead weight. Artificial samples of frozen sa-
line silt with a height of 35 mm and a diameter of 70 mm were used in the
tests. The results show that the deformation behavior of frozen saline silt was
significantly dependent on the impulse at a constant kinetic energy, and
greater deformation was observed with increasing temperature in the range
of the freezing point. An analytical solution that was previously used for ice
was found to be suitable for the description of the investigated dynamic
processes.
KEYWORDS: dynamic behaviors, frozen saline silt, spherical stamp,
dynamic indentation
August 19, 2013.
1
Dept. of Arctic Technology, The Univ. Centre in Svalbard, Longyearbyen, 9171 Norway.
2
Jacksonville, FL.
Copyright V
19428-2959.
180
SINITSYN, doi:10.1520/STP156820130016 181
Introduction
The results of a study on the mechanical properties of frozen saline silt are pre-
sented in this article. The present research is necessary for providing a theoreti-
cal basis for pile driving in the conditions of frozen soils.
Problems associated with defect-free pile driving in frozen ground condi-
tions have been reported since the 1950s. The practice of pile driving in frozen
soils in the United States and Canada [1–15] illustrates the utility of such
operations.
To date, reports by U.S., Canadian, and Russian authors have not
compared the geocryological conditions and feasible geometric dimensions
of piles, their strength parameters, and the potential characteristics of
driving equipment. In addition, few publications have presented theoretical
studies (with the exception of Refs 16 and 17) and experimental data
addressing the parameters of pile-sinking processes in permafrost conditions.
Nevertheless, it should be noted that this is not presently an active research
area.
The existing methodologies for determining pile drivability are not
adequate for estimating the soil resistance during pile driving in plastically fro-
zen and cooled soils. Studies of the mechanical properties of plastically frozen
and cooled soils under dynamic loading are required for the design of such
methodologies.
Field tests [17] have shown that the “efficiency of a blow (penetration
depth from one blow) increases with an increase in the mass of a hammer at
the same impact energy.” Based on this relationship, one can investigate the
response of frozen ground based on the parameters of a shocking action, or
the impact impulse (by varying the mass of an indenting object and the initial
indentation velocity of an object), at a constant kinetic energy.
Experiment
The experiments were performed at The University Centre in Svalbard (UNIS).
The tests were conducted using frozen saline silt lacking an intact struc-
ture. The soil was tested at the optimal water content. Test specimens were pre-
pared from artificial cylindrical samples with a height of 130 to 140 mm and a
diameter of 68 mm. Soil paste was placed in molds and compacted in layers.
Each sample consisted of five layers, and each layer was compacted by the
same amount of blows provided by a metal disk falling from a given height.
The metal disk had a hole in it and fell along a leading rod; friction between
contacting surfaces was absent. A lining lubricated with silicone grease was
placed inside the molds to facilitate the extraction of the sample from the form
after freezing. The mold was covered in double-layer thermal insulation con-
sisting of polyethylene foam with a total thickness of 20 mm. The form ends
were closed using caps with openings for the drainage of pore fluids; the caps
were fixed by screw-clamps. The mold containing the soil was placed in a
chamber with a temperature of 20 C to 24 C. After freezing (20 C), sam-
ples were removed from the molds and placed in a chamber set at the evalua-
tion temperature and maintained until the temperature was stabilized. The
samples were cut into three sections, from which samples were prepared for
the tests. The height of each sample was 35 mm.
Tests were conducted at temperatures close to the freezing point. The tem-
perature regime in the cold chamber was stable at temperatures of 4 C. The
specimens were saline to provide test temperatures close to the freezing point.
Salt (NaCl) was added to distilled water used for the preparation of the soil
paste. The amount of salt added corresponded to a freezing point of 4 C for a
given water content, grain size distribution, salinity, and type of salt. The freez-
ing points of the specimens were also obtained experimentally using a cooling
diagram.
The physical parameters of the specimens were as follows: soil–silt, den-
sity q ¼ 1.88 g/cm3, density of soil particles qs ¼ 2.80 g/cm3, total water content
Wtot ¼ 0.26, water content at the plastic limit Wp ¼ 0.22, water content at the
liquid limit WL ¼ 0.28, salinity of the pore solution Cps ¼ 4.81 %, and freezing
point hbf ¼ 3.9 C.
Experimental Method
The objective of the experiments was to determine the relationship between the
soil deformation and the impulse parameters.
The spherical stamp indentation method is used as a basis for research
[18]. The method proposed by Tsytovich [19] is based on a Brinell test.
Impulse Indentation Using a Spherical Stamp—The indentation of a spheri-

cal stamp via an impulse action was considered. A part of a sphere was used as
the stamp. The impulse was generated by dropping a dead weight. Three com-
binations of the varied dead weights and initial velocities (height of drop) were
considered, and all combinations provided the same kinetic energy E of a fall-
ing dead weight.
mV 2
E¼ ¼ const (1)
2
where:
m ¼ mass of a moving mass, kg, and
V ¼ velocity of the moving mass, m/s.
SINITSYN, doi:10.1520/STP156820130016 183
The applied ranges of the weights and initial speeds (drop heights) were
experimentally selected based on the following condition (Eq 2):
St 0:05D (2)
The value of St ¼ 0.05 D is within the range of 0.01D to 0.1D, as defined for
long-term tests [20]. It was shown that elastic (but not only plastic) deforma-
tion plays a significant role in the settling of a stamp at very low loads and val-
ues of Slong term 0:01D; at higher loads and at Slong term > 0:1D, the equation
for the average normal stresses becomes very approximate.
A portion of a hard plastic sphere, which had a diameter of 25 mm and a
mass of 0.5 g, was used as the indenter. Three sets of metal screw-nuts (the
nuts were securely joined using tape) were used as the dropping weights.
A kinetic energy of 0.150 J was selected. The kinetic energy was recalcu-
lated to account for the mass of the indenter. The test parameters are presented
in Table 1. The chosen combinations of dead weight and height of drop pro-
vided a wide range of dead weights (2 orders) and initial velocities (1 order)
within the range of allowable deformations (Eq 2). Combinations were desig-
nated regarding the initial velocity of the falling dead weight as “high veloc-
ity,” “medium velocity,” and “low velocity.”
The impulse duration and acting force were registered using thin, flexible
FlexiForce sensors placed under the stamp. This type of sensor functions based
on the resistivity principle. The sensors were calibrated prior to the tests using
dead weights. The calibration allowed for the relative comparison of dynamic
impact readings; an assumption that the measured parameters were absolute
values was not valid because of the static regime of the calibration. The flexible
sensors were connected to a laptop computer. Data were processed using the
software ELF Multi 3.20 [21]. The acting force was displayed as the mass act-
ing on the sensor, and the sampling frequency was 200 Hz. The test configura-
tion is presented in Fig. 1.
Test Results
The temperature of each specimen was measured at its surface and at a depth
of 15 mm in order to calculate the mean temperature hmean. A Testo 177-T3
thermo-logger was used for the temperature measurements. Experiments were
TABLE 1—Dead weight mass and drop height combinations used in the tests.
Initial Velocity, Height of Drop, Recalculated

Combination Dead Weight, g m/s cm E, J
High velocity 7.5 6.3 204 0.140
Medium velocity 62.9 2.2 23.4 0.148
Low velocity 485.6 0.8 3.1 0.149
FIG. 1—Configuration of the impulse indentation instrument using a spherical

stamp: 1, soil specimen; 2, thin sensor; 3, a portion of a plastic sphere with
D ¼ 25 mm (indenter); 4, leading rod for the dropping weight; 5, dropping
dead weight (with a hole in the middle for the leading rod).
conducted at temperature series designated as “warm temperature” (hmean/

hbf ¼ 1.5–1.73), and “cold temperature” (hmean/hbf ¼ 2.28–2.46).
The diameter of the footprint Dfp was measured using a trammel device in
two directions. The settling of the indenter St was calculated based on the mean
footprint diameter.
A typical result of an impulse measurement is presented in Fig. 2: the act-
ing mass m (in grams) is shown on the y-axis, and the duration of the impulse
timp (in seconds) is presented on the x-axis.
The impulse used for the deformation (in newton-seconds) is calculated as
follows:
ð timp
J¼ Fdt (3)
0
SINITSYN, doi:10.1520/STP156820130016 185
FIG. 2—Typical result of an impulse measurement.
where F is the force acting on the sensor during stamp indentation N.

The force is calculated as
F ¼ mG (4)
where G is the gravitational constant.

A test matrix and results are presented in Table 2. The brackets indicate
that the presented value is an average for a given combination.
At a constant soil temperature, the settling of the stamp St and duration of
the impulse timp increase with an increasing drop weight and decreasing veloc-
ity of the dead weight acting on the indenter at a constant kinetic energy (Figs.
3 and 4). The settling of the stamp and duration of the impulse decrease with
decreasing temperature (increasing of relative temperature hmean/hbf) (Figs. 3
and 4). The settling increases with an increase in the impulse J and increased
temperature (decrease of relative temperature hmean/hbf) (Fig. 5). The influence
of temperature on the soil deformation behavior and the duration of the impulse
can be explained by the general condition that frozen soil becomes harder,
stiffer, and more brittle as its temperature decreases.
Discussion
In this study, it is proposed that the analytical Eq 5 [22] can be used to describe
the experiment.
t7=9 M4=9
fmax ¼ 0:71 1=9 (5)
ð2RÞ5=9 3lkp3
where:
fmax ¼ maximal settling of the stamp, m,
R ¼ radius of the surface curvature of the indenting body, m,
M ¼ mass of the impacting body, kg,
t ¼ initial velocity of the blow, m/s, and
3lkp ¼ multiplier characterizing the physical-mechanical properties of ice,
Pa2 s/m.
TABLE 2—Test matrix for the spherical stamp impulse indentation trials.
Test hmean, Dfp, hSti, hJi, J, timp, htimpi,

Number Combination C hhmean/hbfi mm mm Ns Ns s s
Warm temperature
13 High velocity 6.7 1.73 5.95 0.44 0.604 0.0092 0.030 0.032
16 6.7 6.65 0.88 0.035
17 6.7 6.65 1.62 0.034
18 7.0 7 0.64 0.030
1 Medium velocity 5.9 1.50 7.45 0.49 1.89 0.0167 0.036 0.036
2 5.9 6.4 1.62 0.036
3 5.9 6.8 1.60 0.035
11 Low velocity 6.4 1.56 8.5 0.74 4.82 0.0491 0.05 0.051
12 5.9 8.4 5.19 0.05
Cold temperature
1 High velocity 9.7 2.46 6.25 0.30 0.42 0.0065 0.024 0.022
2 9.6 5.2 0.45 0.019
3 9.55 5.05 0.95 0.025
4 9.5 5.45 0.52 0.019
5 9.55 5.7 0.26 0.014
6 9.6 5.05 1.36 0.030
7 Medium velocity 9.6 2.44 6.1 0.36 0.81 0.0070 0.024 0.024
8 9.45 5.5 0.65 0.022
9 9.5 6.15 0.95 0.025
10 9.55 5.7 0.99 0.035
11 9.55 6.45 0.21 0.014
12 9.5 5.75 0.70 0.024
13 Low velocity 9.4 2.28 6.15 0.39 4.26 0.0457 0.049 0.041
14 9.4 5.75 4.80 0.038
15 9.05 6.15 4.53 0.041
16 8.45 6.7 4.65 0.041
17 8.15 6.45 4.89 0.040
18 * 5.95 4.85 0.038
*
not recorded
Equation 5 was developed for obtaining the maximal settling of ice

impacted by a spherical hard body. The objective was to describe the dynamic
character of the contact pressures, which depends on the collision velocity,
mass of the colliding bodies, and configuration of the contact zone.
An expression that relates the contact pressure and the thickness of the ice
layer was obtained in Ref 22. The axisymmetric problem regarding the interac-
tion of an absolutely rigid sphere in a flat ice surface was considered. The depth
of sphere penetration into ice is small relative to the radius of the sphere. An
elastic contact of a rigid body with ice in the initial moment of impact was
SINITSYN, doi:10.1520/STP156820130016 187
FIG. 3—Relationship between the stamp settling and the relative temperature.
FIG. 4—Relationship between duration of impulse and relative temperature.
FIG. 5—Relationship between the stamp settling and the impulse at the studied
temperatures.
assumed. Immediately, therefore, further indentation occurs with the develop-

ment of an intermediate layer. It was considered that the thickness of the inter-
mediate layer is small relative to any another linear dimensions.
Equation 5 was obtained by substituting the expressions for the contact
pressure and thickness of the ice layer in the equation for the motion of a half-
spherical body in an impact process.
Let us define the coefficient a ¼ ð3lkp3 Þ1=9 Pa s (Pa/m3)1/9 for the experi-
ment of dropping weights on a spherical indenter placed on frozen soil; the
rewriting of Eq 5 (with substitution of fmax by St) gives
1=9 t7=9 M4=9

a ¼ 3lkp3 ¼ 0:71 (6)
ð2RÞ5=9 St
where l is the coefficient of the internal friction in a layer, measured in pascal-

seconds. The coefficient kp is defined as follows:
p p0 ¼ kp h (7)
where:
p ¼ contact pressure, Pa s,
p0 ¼ static pressure, Pa s, and
h ¼ thickness of a layer, m.
Average values of the stamp settling values St, coefficient a ¼ ð3lkp3 Þ1=9 ,
and relative temperature hmean/hbf for three combinations of weight and falling
weight velocity are presented in Table 3.
The dependence of the a coefficient on the temperature for each given
combination is presented in Fig. 6. This relationship demonstrates that values
of a were highly similar for all combinations at a given temperature. The coef-
ficient a increases with decreasing temperature (Fig. 6). This correlation can be
explained by the increase in the soil viscosity l (included in a) with decreasing
TABLE 3—Results of the calculation of coefficient a.
Combination Series m, kg t, m/s hhmean/hbfi hSti, mm hai

1 1 0.0075 6.3 1.73 0.44 6046.43
2 2.46 0.30 8871.08
2 1 0.069 2.2 1.56 0.49 6495.69
2 2.44 0.36 8383.36
3 1 0.4856 0.8 1.5 0.74 4569.28
2 2.28 0.39 8696.73
SINITSYN, doi:10.1520/STP156820130016 189
FIG. 6—Dependence of a coefficient on the temperature for the given weight

and initial velocity combinations.
temperature. The experiments presented herein can be accurately described

using analytical Eq 5.
Conclusions
A new methodology for the dynamic testing of frozen soils is presented. An an-
alytical equation that describes the experiment was proposed and verified based
on experimental data. It was demonstrated that the deformation behavior of fro-
zen saline silt is significantly dependent on the impulse at a constant kinetic
energy and increases with increasing temperatures in the range close to the
freezing point. An equation (Eq 5) developed for ice appeared to be generaliz-
able to frozen soil.
Acknowledgments
The writer acknowledges the support of the Research Council of Norway
through the Center for Sustainable Arctic Marine and Coastal Technology
(SAMCoT) at NTNU, UNIS, and SINTEF, Norway. The writer does not have
any affiliation with the manufacturers of the equipment described in this paper;
all trademarks are mentioned only to assist persons interested in the use of such
equipment.
References
[1] Kitze, F. F., “Installation of Piles in Permafrost,” Miscellaneous Paper

18, Arctic Construction and Frost Effects Laboratory, U.S. Army Engi-
neer Division, New England Division, Hanover, NH, 1957.
[2] Johnston, G. H., “Pile Construction in Permafrost,” International Perma-
frost Conference, National Academy of Sciences, Washington, D.C.,
1963, pp. 477–481.
[3] Crory, F. E., Installation of Driven Test Piles in Permafrost at Bethel Air
Force Station, Alaska, Cold Regions Research and Engineering Labora-
tory, Hanover, NH, 1973.
[4] Rooney, J. W., Nottingham, D., and Davison, B. E., “Driven H-Pile
Foundations in Frozen Sands and Gravels,” Symposium of the Second
International Symposium on Cold Regions Engineering, Fairbanks, AK,
Aug 12–14, 1976.
[5] Bendz, J., “Permafrost Problems Beaten in Pile Driving Research,” West-
ern Construction, Vol. 52(11), 1977, pp. 16, 18, 34, 35.
[6] Davison, B. E., Roney, J. W., and Bruggers, D. E., “Design Variables
Influencing Piles Driven in Permafrost,” Proceedings of the Conference
on Applied Techniques for Cold Environments, Anchorage, Alaska, May
17–19, 1978, ASCE, New York, pp. 307–318.
[7] DiPasquale, L., Gerlek, S., and Phukan, A., “Design and Construction of
Pile Foundations in the Yukon-Kuskokwim Delta, Alaska,” Proceedings
of the 4th International Conference on Premafrost, National Academies
Press, Washington, D.C., 1983, pp. 232–237.
[8] Tomas, H. and Mobley, K., “Special Pile Foundation for a Coastal Perma-
frost Site,” Cold Regions Engineering: Proceedings of the Fourth Interna-
tional Conference, Anchorage, AK, Feb 24–26, 1986, Committees of the
National Academy of Sciences and the State of Alaska, pp. 1–9.
[9] Merrill, K. S. and Riker, R. E., “Comparison of Static and Dynamic Test
Results for Driven Steel Pipe Piles in Highly Saline Permafrost,” Pro-
ceedings of the Eighth International Conference on Cold Regions Engi-
neering, Fairbanks, AK, Aug 12–16, 1996.
[10] Phukan, A., “Driven Piles in Warm Permafrost,” Proceedings of the 7th
International Permafrost Conference, National Academies Press, Wash-
ington, D.C., 1998, pp. 891–895.
[11] Vyalov, S. S., Reologicheskie svoistva i nesuschya sposobnost merzlyh
gruntov [Rheological Behavior and Bearing Capacity of Frozen Soils],
Akademizdat, Moscow, Russia, 1959 (in Russian).
[12] Eroshenko, V. N., “Trial Pile Deployment on the Area with Deep Stratifi-
cation of Permafrost in Vorkuta,” Osnovania, Fundamenty i Mehanika
Gruntov (OFMG), Vol. 3, 1965 (in Russian).
[13] Vyalov, S. S. and Tagulyan, U. O., “Prohodka skvazhin i pogruzhenie
svai v vechnomerzlie grunty [Arrangement of Bore Holes and Pile
Deployment in Permafrost],” Osnovania, Fundamenty i Mehanika Grun-
tov (OFMG), Vol. 3, 1968 (in Russian).
[14] Kosterin, E. V., Osnovaniya i fundamenty: Uchebnik dlya vuzov [Base-
ments and Foundations: Study Book for Universities], 3rd ed., Vissh.
Shk., Moscow, 1990 (in Russian).
[15] Targulyan, U. O., Fedorovich, D. I., Gohman, M. F., and Nekludov, V. S.,
“Sovershenstvovanie sposobov ustroystva svainyh fundamentov na
SINITSYN, doi:10.1520/STP156820130016 191
vechnomerzlih gruntah [Improvement of Pile Foundation Construction in

Permanently Frozen Soils],” OFMG, No. 4, 1994 (in Russian).
[16] Weaver, J. S., Pile Foundations in Permafrost, University of Alberta, Ed-
monton, AB, Canada, 1979.
[17] Novozhilov, G. F., Bezdefektnoe pogruzhenie svai v talyh i vechnomerzlyh
gruntah [Defect-free Pile Driving in Frozen Soils], Stroiizdat, Leningrad,
1987 (in Russian).
[18] GOST 24586-81, “Grunty. Metody laboratornogo opredeleniya harakter-
istik prochnosti i deformiruemosti gruntov [Soils. Methodology for Labo-
ratory Testing on Strength and Strain Properties],” Gosstroy, USSR, 1990
(in Russian).
[19] Tsytovich, N. A., “Determining of Cohesion Forces of Frozen Soils on
the Base of Spherical Stamp,” Funds of Institute of Permafrost, Academy
of Sciences of USSR, Moscow, USSR, 1947 (in Russian).
[20] Tsytovich, N. A., Mehanika merzlyh gruntov. Uchebn. posobie, Vyssh.
shkola, Moscow, 1973 (in Russian).
[21] Tekscan, Inc., “Pressure Mapping, Force Measurement, and Tactile
Sensors,” http://www.tekscan.com/index.html (Last accessed 7 Aug
2013).
[22] Heisin, D. E., Lihomarov, V. A., and Kurdumov, V. A., “Obtaining of
Specific Energy of Destruction and Contact Pressures at Blow of Solid
Body on Ice,” Arctic and Antarctic Problems, 1973, pp. 210–218.

STP 1568-2013

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

STP 1568-2013

Uploaded by

Copyright:

Available Formats

Selected Technical Papers STP1568

Mechanical Properties of Frozen Soils

Printed in the U.S.A.

ASTM Stock #: STP1568

Printed in Bay Shore, NY

FREEZE-THAW EFFECTS ON MECHANICAL PROPERTIES

TESTING OF MECHANICAL PROPERTIES

CLASSIFICATION AND EFFECTS OF MECHANICAL

Ozlem Bozyurt,1 Andrew K. Keene,2 James M. Tinjum,3

Freeze–Thaw Effects on Stiffness of Unbound

and maintenance. The use of seismic testing in pavement health monitoring is

Unbound RAP and RCA

Freeze–Thaw Effect on RAP and RCA

TABLE 1—Index properties of recycled materials and class 5 aggregate.

D50 AB AC wopt cdmax Gravel Fines

RCA CA 4.8 22 1.4 2.3 5.0 — 10.4 19.9 50.6 2.3 GW

engineering properties of recycled materials was determined by testing the Mr af-

and non-destructive seismic testing method was performed to calculate the

where q is the mass density.

Recycled Asphalt Pavement (RAP)

Seismic Modulus—Seismic modulus testing was conducted on RAP speci-

Recycled Concrete Aggregate (RCA)

Resilient Modulus—The effect of F-T cycling on the SRM of representative

Seismic Modulus—Seismic modulus testing was used to verify the trend in

[1] USGS, “Crushed Stone: Statistical Compendium,” U.S. Geological Sur-

[30] ASTM D1557: Standard Test Methods for Laboratory Compaction

Christopher W. Swan,1 Alex Grant,2 and Alyssa Kody2

Characteristics of Chicago Blue Clay

REFERENCE: Swan, Christopher W., Grant, Alex, and Kody, Alyssa,

behavior. The writers recommend that more testing be performed to verify

Big Dig Settlement Case

Mechanics of Freeze–Thaw on Porous Media

Thawing and Post-Thaw Behavior

The current study focuses on the effects of freeze–thaw on the consolidation of

TABLE 1—Summary of consolidation parameters from incremental consolidation tests on Chicago

One Freeze–Thaw Change From

Consolidation parameters developed from the incremental tests are pre-

FIG. 2—Photographs of specimens of Chicago Blue Clay subjected to incre-

TABLE 2—Results of Atterberg limit tests on Chicago Blue Clay.

Liquid Limit, Plastic Limit, Plasticity Index,

After a After a After a

Atterberg Limit Results

Summary, Conclusions, and Recommendations for Further Study

process. There are a number of other alterations which should be considered to

[1] U.S. Federal Highway Administration, “Structural Assessment of 1-90

[3] Chamberlain, E. J. and Gow, A. J., “Effect of Freezing and Thawing on

[18] ASTM D2216-05: Standard Test Methods for Laboratory Determination

Gregory P. Makusa,1 Hans Mattsson,2 and Sven Knutsson2

Investigation of Increased Hydraulic

REFERENCE: Makusa, Gregory P., Mattsson, Hans, and Knutsson, Sven,

Materials and Methods

Dry Density–Water Content Relationship

FIG. 1—Particle size distribution for studied silty till.

FIG. 2—Dry density–water content relationship.

average water content of 10.80 % corresponds to a dry density of 2.03 g/cm3,

Sample Preparation for Hydraulic Conductivity Test

Hydraulic Conductivity Test

and molding water contents, all specimens were compacted to a 95 % degree

FIG. 3—Schematic sketch of experimental device for one-dimensional freezing

thawed without an overburden stress. The hydraulic conductivity results

TABLE 1—Results of hydraulic conductivity tests.

Soil Sample A Soil Sample B

Freeze–Thaw Specimen A1 Specimen A2 Specimen B1 Specimen B2 Specimen B3

Soil Sample A Soil Sample B