Computers and Geotechnics 91 (2017) 2738

Contents lists available at ScienceDirect

Computers and Geotechnics

journal homepage: www.elsevier.com/locate/compgeo

Research Paper

Frost jacking characteristics of screw piles in seasonally frozen regions

based on thermo-mechanical simulations
Tengfei Wang a, Jiankun Liu a,b,, Bowen Tai a, Chuanzhen Zang a, Zhichun Zhang c
a
School of Civil Engineering, Beijing Jiaotong University, Haidian District, Beijing, China
b
Qinghai Research and Observation Base, Key Laboratory of Highway Construction & Maintenance Technology in Permafrost Region, Qinghai Research Institute of
Transportation, Xining, Qinghai, China
c
Shenshuo Railway Branch of China Shenhua Energy Co., Ltd, Yulin, Shanxi, China

a r t i c l e i n f o a b s t r a c t

Article history: In this paper, a thermo-mechanical model is proposed to simulate the frost jacking behaviour of screw
Received 17 December 2016 piles subjected to frost heave, and the results are further validated by laboratory tests. The calculated
Received in revised form 27 June 2017 results show that large multi-helix piles yield the least frost jacking when the freezing depth reaches half
Accepted 29 June 2017
the embedment depth of pile. Based on the modified cylindrical shear method and individual bearing
method, the optimal geometric parameters of screw piles are determined by a series of numerical calcu-
lations. The numerical approach is expected to serve as a reference for designing effective and economical
Keywords:
pile types in practice.
Screw pile
Frost jacking
2017 Elsevier Ltd. All rights reserved.
Thermo-mechanical model
Optimal design
Seasonally frozen regions

1. Introduction
Seasonally frozen soil is defined as soil or rock having a monthly
mean temperature below 0 C during cold seasons for at least one
year and is characterized by a discontinuous distribution [1]. In
China, the seasonally frozen region is approximately
5.13
106 km2, accounting for 53.5% of the total land area, and it
is mainly distributed in the northern and eastern parts. To date,
many artificial structures have been built in these seasonally fro-
zen regions, e.g., photovoltaic generation projects in Inner Mongo-
lia. Photovoltaic frames with panels are generally installed in light-
rich areas, which extensively overlap with the seasonally frozen
regions. Consequently, widespread frost heave of soil occurring in
winter imposes severe challenges to the normal operation of the
photovoltaic panels. According to some researches on tower foun-
dations along the Qinghai-Tibet Power Transmission Line, uneven
frost jacking of the foundations with the development of the
ground freezing process may lead to total failure of the superstruc-
tures [2,33,34]. As a result, systematic investigations of frost dam-
age to the foundations of photovoltaic structures are required.
To address these problems, screw piles have been widely
applied because they can be quickly installed, are reusable, remov-
Corresponding author at: School of Civil Engineering, Beijing Jiaotong Univer-
sity, Beijing 100044, China.
E-mail address: jkliu@bjtu.edu.cn (J. Liu).
able and cost effective, produce little noise and vibration during
installation, require minimal dewatering and equipment, and offer
high tensile capacity, which is the most distinctive superiority in
this study (mitigating frost jacking) [38]. Initially, the screw piles
were utilized to bear tensile loads (e.g., foundations of transmis-
sion towers and pipelines), and then their applications were fur-
ther extended to structures and infrastructure exposed to
compressive and lateral loads [5]. In contrast to conventional piles,
screw piles are commonly made of steel with helices fixed to the
shaft at a certain spacing, and at most cases, the existence of a
pointy toe guarantees better installation into the ground [9]. Screw
piles with various helix and shaft diameters, helix spacing, embed-
ment depths, etc., are adopted for diversified working conditions.
The mechanical interlock between the helices of the screw piles
and the soil greatly improves the strength of the pile-soil system,
thus screw piles have unique advantages in anti-jacking projects
in seasonally frozen regions.
Mohajerani et al. [10] reviewed the analysis and design meth-
ods of screw piles, which have mainly been centred on vertically
loaded piles in compression and tension in both cohesive and
non-cohesive soils. A series of laboratory tests were conducted to
evaluate the anti-frost jacking abilities of several screw piles with
various types of helices [11]. Abdrabbo and Wakil [12] studied the
responses of horizontally loaded screw piles with different helix
diameters, numbers, and spacing. A numerical method was pre-
sented to simulate screw displacement piles interacting with

http://dx.doi.org/10.1016/j.compgeo.2017.06.018
0266-352X/ 2017 Elsevier Ltd. All rights reserved.
28 T. Wang et al. / Computers and Geotechnics 91 (2017) 2738
non-cohesive subsoil during the transfer of compression load by
using the Plaxis 2D program [13]. The uplift behaviour of helical
anchors in clay was investigated using finite element limit analysis,
and the factors that influence the model uncertainty of the cylin-
drical shear method were also examined [14]. A finite element
method was performed to simulate nonlinear responses of a steel
pile embedded in frozen ground at a field test site in Alaska [15].
Liu et al. developed a numerical model to simulate the stress-
strain relationships in the pier and in the surrounding frozen soil
[36]. However, there is little information in the literature on
numerical simulations regarding screw piles and frost jacking.
The main objective of this paper is to compare the frost jacking
abilities of different screw piles and determine the optimal geo-
metric parameters of screw piles subjected to frost heave. The
paper is organized as follows. First, a numerical model is estab-
lished for the screw piles in seasonally frozen regions on the basis
of heat transfer, mechanical deformation and temperature-related
properties of soil. Then, several types of screw piles adopted in lab-
oratory tests, including one with no helix, are taken as examples to
simulate the freezing process and frost jacking behaviour of piles
using a numerical approach. Finally, the theoretical calculation
methods for the uplift capacities of screw piles in frozen soil are
modified, through which optimal designs are determined to resist
frost jacking. Inherent mechanisms are explained from the per-
spectives of the modified cylindrical shear method and individual
bearing method. This study offers a better understanding of the
frost jacking process of screw piles under one-dimensional freezing
conditions.
2. Mathematical formulation
2.1. Thermo-mechanical model
In freezing soil, heat convection is negligible compared with
heat conduction and the latent heat of water-ice phase changes
[16,17]; hence, the energy conservation equation in soil can be
written as
@T @hi
C r  krT Lqi 1
@t @t
where C denotes the heat capacity per unit volume, T is tempera-
ture, t represents time,  denotes the divergence, k denotes thermal
conductivity, hi represents the volumetric ice content, qi denotes ice
density, and L denotes the latent heat of ice-water phase change.
As no water transport equation is applied in the model, Q is
introduced to replace the last term in Eq. (1), and it is defined as
the heat change per unit volume of the soil due to the water-ice
phase change [1],
Q Lqd w  wu 2 method (Fig. 2b) should be employed for S/D > 3.
where qd is the dry density of the soil, w denotes the gravimetric
moisture content, and wu represents the gravimetric content of
unfrozen water. For common thermal cases, wu of the cohesive soil
can be expressed by [1]
wu KT  w 3
where K denotes the correction coefficient varying with
temperature.
Introducing Eq. (2) into Eq. (1), yields
@T @Q
C r  krT 4
@t @t
The equilibrium equation is given as follows:
r  rij qg i 0 5
where  denotes the divergence, rij represents the total stress ten-
sor, q denotes the density of the material, and gi is the gravity vec-
tor. The stress tensor can be written as [18]
rij C ijkl eij Ikl einel 6
where Cijkl denotes the stiffness tensor, which is associated with the
material parameters such as Youngs modulus E and Poissons ratio
t, eij is the total strain tensor, Ikl is the identity matrix, and einel rep-
resents the strain induced by temperature variation, which is
expressed as follows:
einel aT  T 0 7
where a denotes the thermal expansion coefficient of the material,
and T0 is reference temperature. In this study, nonlinear relation-
ship is established between einel and temperature, which is realized
by parameter a. That is to say, the thermal expansion coefficient a is
treated as a linear expansion coefficient varying with negative tem-
perature as follows:
einel aT  1 8
The volumetric strain ev is defined as
DV
ev 9
V0
where DV and V0 represent volume change and initial volume,
respectively. Then, the relationship between a and ev can be
obtained on the basis of equivalence principle:
p
a 3
ev T 1  1 10
The volumetric strain ev used in this analysis is determined
from a series of laboratory tests by wax coating method at a set
temperature.
The frost heave ratio g is defined as
Dh
g 11
Hf
where Dh is confined frost heave, Hf denotes the freezing depth.
2.2. Design method and modification: tensile capacity of screw piles in
cohesive soil
Fig. 1 shows the basic elements and parameters regarding the
screw piles, where D = diameter of pile helix, d = diameter of pile
shaft, S = helix spacing, H = embedment depth of the pile to the
top helix, t = helix thickness, and uh = declination angle of the helix.
In most cases, two possible failure mechanisms for screw piles are
considered depending on the relative spacing of the helical plates
(S/D). To calculate the tensile bearing strength of the screw piles,
it is suggested that the cylindrical shear method (Fig. 2a) be used
for close helix spacing S/D  3, whereas the individual bearing
The cylindrical shear method was introduced by Mooney et al.
[19] for screw piles in silt and clay. The possible failure mode is
that a cylindrical shear failure surface will develop connecting
the top and bottom helices as shown in Fig. 2a. The tensile bearing
strength Qt will be a combination of shear resistance along the soil
between the helices and the top helix plate bearing and the fric-
tional resistance offered by the soil-shaft interface [10,14], given
by
Q t pDSf n  1Scu Ah Nu cu c0 H pdHeff a0 cu 12
where D is pile helix diameter, Sf is the spacing ratio factor, n
denotes the number of helix plates, S represents the spacing
between two adjacent helix plates, cu denotes the undrained shear
strength, Ah is projected area of the helix, c0 is effective unit weight
of soil, H is the embedment depth of the top helix, d denotes the
diameter of the pile shaft, Heff = HD, which is the effective length
of the pile above the top helix due to the shadowing effect of a loss
 T. Wang et al. / Computers and Geotechnics 91 (2017) 2738 29

Fig. 1. Typical helical anchor arrangement: (a) elements of a screw pile; (b) parameters of interest for screw pile analysis.

Fig. 2. Screw piles axially loaded in tension: (a) cylindrical shear method; (b) individual plate bearing method.

in shaft adhesion above the top helix [20], a0 represents the adhe-
sion factor, and Nu is the uplift bearing capacity factor for cohesive
soils, calculated by [21]
H
Nu 1:2 69
D
The soil weight between the helices and the pile weight are not
explored in this study. In this case (Fig. 3a), when the top helix is
exactly below the freezing depth, the shear strength of soil above
the top helix will greatly increase. Meanwhile, the original shaft
friction above the top helix is replaced by the tangential frost heav-
ing force. This is induced by the frost heave of soil and freezing
action on the pile-soil interface, and the total uplift force can be
acquired as follows [22]
Z Adams and Klym [23] first assumed that each helical plate
Qu pdgf qf dx 0 6 x 6 Hf
where Qu is the total tangential frost heaving force, gf denotes the
influence coefficient of freezing depth, which take a value of 1.0
when the freezing depth is less than 2.0 m [22], qf represents the
shear stress due to frost heave, which can be expressed as a function
13 of temperature, moisture content and soil properties, and dx is the
infinitesimal element within the range of freezing zone.
According to the above analyses, Eq. (12) will then become
Q t;f pDSf n  1Scu Ah Nu cu jf c0 H 15
where Qt, f represents the tensile capacity of the piles exposed to
frost jacking, and jf denotes the freezing factor of the shear
strength. In addition, the shaft friction should not be ignored when
the freezing depth is relatively small.
14 works independently when the helix spacing is sufficiently large.
It is supposed through the individual bearing method (Fig. 2b) that
30 T. Wang et al. / Computers and Geotechnics 91 (2017) 2738

Fig. 3. Screw piles subjected to negative temperature: (a) modified cylindrical shear method; (b) modified individual plate bearing method.

each helical plate behaves independently under tension, thus the
total uplift capacity of screw piles is determined by a combination
of all the individual bearing capacities and shaft friction [57]. The
ultimate tensile capacity of a pile in cohesive soil depends on the
helix bearing area as well as the disturbed soil above it, which
can be calculated as follows [24,25]:
X
Qt Q h Ah c0 H
tent of 22%). Three model piles were embedded into the soil at a
spacing of 30 cm. To complete the model, a cooling tank was placed
on the soil surface. The tank was then connected to a cooling bath
via inlet and outlet pipes through which coolant at a set tempera-
ture was circulated in order to guarantee a stable boundary tem-
16

in which
Q h 9Ah cu 6 Q s 17
where Qh represents the individual helix bearing capacity, and Qs is
the upper limit determined by the helix strength.
As indicated in Fig. 3b, when the top helix is below the freezing
depth, the freezing factor jf is introduced into Eq. (16), which
yields
!
X
n1
Q t;f Q h 9Ah cu jf Ah c0 H 18

3. Finite element analysis

Finite element simulations with the software ABAQUS were

performed in this paper as an analysis and validation tool. A field
site of photovoltaic generation project located in Inner Mongolia,
a seasonally frozen region, is taken as an example to investigate
frost jacking behaviour of screw piles. The screw pile employed
at this site has an embedment length of 1.8 m, and the maximum
freezing depth during winter is approximately equal to half of that.

3.1. Small scale model experiment

To compare the results obtained from the thermo-mechanical

analysis of a pile-soil system exposed to negative temperature with
those of indoor model-scale tests under the same condition, the
geometrical parameters are scaled down to 1/3 compared with
that of the pile in practice (embedment depth of the pilot point
L = 0.6 m). The model-scale test was conducted on an
independently-developed multifunctional test box (Fig. 4a) by
adopting soil samples from a field site in Inner Mongolia, China.
Sensors were implanted at predetermined positions (Fig. 5) during Fig. 4. Multifunctional test box (a) and schematic of the displacement measure-
the placement and compaction of the soil in layers (moisture con- ment system (b).
 T. Wang et al. / Computers and Geotechnics 91 (2017) 2738 31

Fig. 5. Profile of the test box and sensor layouts.

perature. Finally, thermal insulation materials were wrapped out- where Tf and DTf are the freezing temperature and temperature
side the box. When the test began, the cooling tank was arranged range for the phase change (from unfrozen to frozen), respectively;
to freeze the soil at a given negative temperature until the freezing the superscripts + and  represent the unfrozen state and fro-
depth became stable. During the process, pile head displacements zen state, respectively. This implies that the values vary linearly
and soil temperatures were measured by dial gauges and temper- with temperature in the phase change process.
ature sensors (PT100), respectively (Figs. 4b and 5). Four types of In view of the relatively light weight of the superstructures of
screw piles, namely, large (9.4 cm diam.) double helix, large photovoltaic frames, no axial load is applied to the pile head.
multi-helix, small (7.4 cm diam.) double helix and small multi-
helix piles, and one pile with no helix, which served as a control, 3.2.2. Phase change process
were considered in this study (Fig. 6). In the numerical analysis, According to [1], most unfrozen water will freeze during phase
the soil mass is approximated to a cylinder with a depth of change process. Thus, K(T) in Eq. (3) is assumed as the following
Hs = 1 m and a diameter of Ds = 0.6 m (i.e. 15d). Temperature- expression:
displacement swept meshes are employed to describe the helical 8
plates, while fine tetrahedral meshes are introduced to the remain- >
<1 T  Tf
ing parts of the model (e.g., the large multi-helix pile, as shown in KT f T T f  DT f < T < T f 21
Fig. 7). >
:
0:3 T 6 T f  DT f

where f(T) is a function, which is not used in this analysis. The resid-
3.2. Idealization
ual correction coefficient is set to 0.3 according to [35].
Equivalent latent heat Leq is introduced into Eq. (2) to describe Q
3.2.1. Physical parameters and external loading
as follows
The numerical model is idealized considering the following
assumptions. The soil mass is treated as a linear elastic material, Q Lqd w  wu Leq  qd 22
whose Youngs modulus Es [26] and Poissons ratio ms [27] both
vary with the temperature in the frozen state (T < 0 C). Leq L  w  wu Lw  1  KT 23
0:6
Es a1 b1  jTj It is implied that Leq is totally determined by the initial gravimetric
19
ms a2 b2  jTj moisture content w.

where a1, b1, a2, and b2 are constants associated with soil properties.
3.2.3. Frost heave model
The soil volumetric heat capacity Cs and thermal conductivity ks
The volume deformation of the soil due to frost heave is sum-
are expressed in a simplified form as follows:
marized in the calculation of einel (Eqs. (8) and (10)), which has
8
> Cs ; T  Tf been determined by the variation of temperature based on the lab-
>
<  oratory tests of volumetric strain ev, as depicted in Fig. 8.
C s C s CsDC s
T  T f ; T f  DT f < T < T f
>
>
Tf
: 
Cs ; T 6 T f  DT f 3.2.4. Interface model
8 20 Within frozen volume, the residual adfreeze strength of pile-soil
> ks ; T  Tf
>
< interface is composed of cohesive strength and friction force,

ks ks ksDk s
T  T f ; T f  DT f < T < T f which are affected by negative temperature and normal pressure,
>
>
Tf
:  respectively [28]. In this analysis, the static coulomb friction model
ks ; T 6 T f  DT f is introduced to describe the shear behaviour of interface. Com-
32 T. Wang et al. / Computers and Geotechnics 91 (2017) 2738

Fig. 6. Model-scale piles adopted in the laboratory tests: (a) small double helix pile; (b) large double helix pile; (c) large multi-helix pile; (d) small multi-helix pile; (e) pile
with no helix.

0.08
Volumetric strain v (-)

0.06

0.04

0.02

Linear interpolation is adopted

between points
0.00

0 -2 -4 -6 -8 -10 -12

Temperature ( C)

Fig. 8. Measured volumetric strain varying with negative temperature.

Fig. 7. Mesh discretization in the 3-D analysis of a large multi-helix pile-soil
system.
bined with Eq. (14), the shear stress due to frost heave can be
expressed as follows
q f l  rn c
where rn denotes the normal pressure, c is the cohesive strength,
and l represents the equivalent friction coefficient.
As indicated in the experiments, the shearing stress increases
linearly with the absolute value of temperature at a constant nor-
mal pressure [28,29]; to simplify, it is assumed that the impact of
negative temperature is only associated with the variation of l, as
shown in the following equation:
qf lT  rn c0 25
lT a3 b3  jTj 26
where c0 is set as a constant (=50 kPa), a3 and b3 are experimental
parameters.
The adfreeze bond strength sp for the interface model is defined
as the maximum value of shear stress that can be carried by the
24 interface before the surfaces begin to slide, and it is derived from
the laboratory direct shear tests.
3.3. Material properties and parameters of interest
The soil properties considered in the analysis are listed in Tables
1 and 2. Leq used in this analysis is equal to 51.5 [kJ/kg] based on Eq.
(23). Table 3 reports parameters related to the geometrical proper-
ties of the piles.
 T. Wang et al. / Computers and Geotechnics 91 (2017) 2738 33

Table 1 0.35
Soil properties used for the numerical analyses.

Soil parameters Value Soil parameters Value 0.30

Es [MPa] 12b
DTf [C] 0.7 [31]
ms [] 0.35b C+s [kJ/(m3C)] 2.50 [1] 0.25

Freezing Depth (m)

qd [kN/m3] 1.60a C 3
s [kJ/(m C)] 2.07 [1]
w [] 22.0a k+s [W/(mC)] 1.80 [1]
0.20
Tf [C] 0.3a k+s [W/(mC)] 2.20 [1]
L [kJ/kg] 334.5 [1] sp [kPa] 250a
0.15
a
Derived from laboratory tests. No-helix pile
b
Parameter varies throughout the simulations (Eq. (19), Table 2). Large double helix pile
0.10
Large multi-helix pile
Small double helix pile
Table 2 0.05
Small multi-helix pile
Material constants adopted in this study.
Experimental Data
0.00
Constant Es ms l
a1 b1 a2 b2 a3 b3 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
Value 9.39 5.38 0.35 5.15
103 0.30a 0.01a Normalized Elapsed Time (t/tend)
[1,26] [1,26] [1,27] [1,27]
a
The values are determined by laboratory direct shear test. Fig. 9. Computed and measured freezing depth varying with time, tend: total time.

initial soil temperature between simulations and experiments.

Table 3
Based on the results, the controlling factor of freezing depth for
Parameters of interest for each pile (see also Fig. 1b).a all cases is satisfied, where we are able to compare the ultimate
frost jacking between different piles under the same freezing depth
Pile type n D d pitch S/D H/D t uh
(mm) (mm) (mm) (mm) ()
conditions.
To validate the numerical model, a comparison between the
Large double helix 2 94 40 56 2.6 3.8 2 30
computed and measured frost jacking displacements is shown in
Large multi-helix 5 94 40 0.6 3.2 2 30
Small double helix 2 74 40 56 4.0 4.9 2 30 Figs. 10 and 11.
Small multi-helix 5 74 40 0.8 4.1 2 30 It is clear from Fig. 10 that, in general, the frost jacking displace-
No helix 40 ments varying with time computed using the numerical model are
a
The values are the same as the model piles in the experiments. in good agreement with the measured data [32]. The displace-
ments gradually increase with time at an average rate ranging from
The steel pile is assumed as linearly elastic with a Youngs mod- 0.47 to 1.52 mm/day and finally trend towards a constant. How-
ulus, Ep of 30 GPa and Poissons ratio, mp of 0.3 [30]. The thermal ever, a deviation of the computed frost jacking from the test results
conductivity kp and volumetric heat capacity Cpi of the pile material can be observed at the initial stage, which can be explained by the
are set to be 50 [W/(mC)] and 3.6 [kJ/(m3C)], respectively. differences in the initial temperature distributions. The frost jack-
ing of the small double helix pile obtained from the test results is
3.4. Boundary and initial conditions slightly higher than that from the numerical model. The reason for
this inconsistency is that the soil surrounding the helices is mod-
Restrictions are applied to the horizontal displacement on the elled as a linear elastic material, i.e. does not yield.
side (roller boundary) and to both the vertical and horizontal dis- As is depicted in Fig. 11, most of the calculated ultimate frost
placements on the base of the model (pinned boundary). Adiabatic jacking of different piles is consistent with that derived from the
boundaries are employed on both the side and base of the model.
The initial stress state due to gravity in the soil-pile system is
9
considered to be geostatic. The influence of the installation process
C M
of piles is not considered in these elements or adjacent soil. The 8 Large double helix pile
predefined temperature field of the entire model is set to be 1 C. Large multi-helix pile
In the time-dependent analysis, the temperature of the upper 7 Small double helix pile
boundary decreases uniformly from 1 C to 10 C within two Small multi-helix pile
Frost Jacking (mm)

hours, and it remains constant afterwards. 6

5
4. Results and discussion
4
4.1. Thermo-mechanical simulations compared with laboratory tests
3
At the beginning of the simulation process, a temperature
2
decrement of DT = 11 C was applied to the upper surface of soil
for five days. The freezing depth is determined from the position of
1
an isotherm of 1 C, which is also specified as the end of the phase
change process. Fig. 9 reveals the development of freezing depths 0
throughout the simulations. The freezing depths increase nonlin- 0.0 0.2 0.4 0.6 0.8 1.0
early with time at an average rate of 0.06 m/day. It is shown that Normalized Elaspsed Time (t/tend)
the freezing depth of measured data is greater than those of the
numerical analyses in the latter half of the time period, yet all tend Fig. 10. Comparisons between measured (M) and computed (C) frost jacking of
towards 0.31 m [32]. This can be attributed to the inconsistency of screw piles varying with elapsed time.
34 T. Wang et al. / Computers and Geotechnics 91 (2017) 2738

0.026

le
pi
25 lix Numerical Simulation
he

Experimental data 0.024

o

22.11
N
Ultimate Frost jacking (mm)

20

le
0.022

Frost heave (m)

17.7

pi
ix
el
eh
15

bl
0.020

le

ou

le
pi

pi
ld
le
lix Large double helix pile

lix
pi

al
he

Sm
lix

he
Large multi-helix pile

t i-
10
le

he
0.018
ub

ul
8.46 No-helix pile
ti-

lm
do

ul
7.58
em

Small double helix pile

al
e
rg

Sm
rg
La

0.016 Small multi-helix pile

La

5
3.72 3.65
3.02 2.79
2.35 2.33
0.014
0 0.00 0.05 0.10 0.15 0.20 0.25 0.30
Distance from the pile (m)
Fig. 11. Calculated ultimate frost jacking of different piles compared with labora-
tory test results.
Fig. 13. Ultimate frost heave of the soil surface at different distances from the pile.

laboratory tests, except in the case of the pile with no helix, which
has an error of 20%. This notable difference can be attributed to the
interface elements used in the numerical model, which cannot
disclose the practical conditions to the maximum extent when
the relative displacements are too large. When the freezing depth
is equal to 0.31 m, the ultimate frost jacking of screw piles is smal-
ler than that of the no helix pile, which illustrates the positive
effects of helices in mitigating frost jacking. The values of frost
jacking for the screw piles range from 2.4 to 7.6 mm, showing a
decrease of up to 87% compared to that of the no helix case.

Fig. 12. Profile of vertical displacements at the end of calculation: (a) small multi-helix pile; (b) large double pile; (c) large multi-helix pile; (d) small multi-helix pile; (e) no
helix pile.
 T. Wang et al. / Computers and Geotechnics 91 (2017) 2738 35

Fig. 14. Maximum principal stress at the end of the freezing process: (a) small multi-helix pile; (b) large double helix pile; (c) no helix pile; (d) large multi-helix pile; (e) small
double helix pile.
36 T. Wang et al. / Computers and Geotechnics 91 (2017) 2738
The uplift force in the frozen volume is derived from the frost
heave of soil and enhanced friction due to freezing on the pile-
soil interface, while the helices below the freezing depth resist
the jacking process. According to Fig. 11, the frost jacking increas-
ing order is as follows: large multi-helix pile < large double helix
pile < small multi-helix pile < small double helix pile < no helix
pile. Both the number of helices and the helix diameter play a role
in the mitigation of frost jacking, and the latter factor is considered
to be dominant because the frost jacking of large double helix pile
is lower than that of the small multi-helix pile. By comparisons
between Eqs. (15) and (18), the individual bearing capacity derived
from the bottom helix of large double helix pile is larger than cylin-
drical shearing resistance provided by the helices underneath top
helix of small multi-helix pile.
Fig. 12 illustrates the vertical displacements across the soil
domain at the end of the freezing process. The thermally induced
deformation of soil is transmitted to the embedded pile, resulting
in the jacking phenomenon. It is clear from Fig. 13 that the average
frost heave of the soil surface is 2.5 cm. Based on Eq. (11), the ulti-
mate frost heave ratio g is equal to 8.16%, which is close to the
measured data from one-dimensional freezing tests on soil sam-
ples. Furthermore, the vertical displacements from the five cases
almost share the same pattern in the frozen layer, whereas below
the freezing line, differences are found due to the helix layout. The
soil above the helical plate moves upward with the screw pile in
the jacking process, the displacement of which is generally higher
than that characterizing the soil below the helical plate or out of
the range of the anchored zone.
The evolution of the ultimate vertical displacement of the adja-
cent soil at the ground surface varying with horizontal distance is
plotted in Fig. 13. In general, the pile has a significant impact in
restricting frost heave of the adjacent soil. This effect becomes less
pronounced as the distance from the pile increases. Given that the
impact fades away when the vertical displacement trends towards
a constant (i.e. 0.0240.025 m), it demonstrates the rationality of
the model dimensions as the soil at the side boundary can freely
undergo frost heave. For the experiment, the boundary effect of
the test box on the piles is also negligible considering that the
box dimensions are 1000 mm
1000 mm
420 mm. The compar-
ison between the experiment and simulations is credible on the
basis of this fact.
The maximum principal stress of each pile at the end of freezing
process is shown in Fig. 14. The stress concentration is detected for
Temperature (C)
-10 -9 -8 -7 -6 -5 -4 -3 -2
Depth (m)
121.28h
65.72h
30.72h
10.17h
2.78h
Fig. 15. Temperature distributions along the pile at different times.
3.5
Large multi-helix pile
3.0 n=4
Ultimate frost jacking (mm)
2.5
n=5
2.0
n=7
1.5
0.040 0.045 0.050 0.055 0.060 0.065 0.070
Helix spacing (m)
Fig. 16. Relationship between the helix spacing S and the ultimate frost jacking.
the top helix of the screw pile, the value of which is much greater
than the rest. Due to the frost heave of soil and the pile-soil inter-
action, the stress of pile shaft in the frozen region is higher than
that between helices. In general, the helices of double helix piles
bear more loads than those of the multi-helix piles. For large and
small multi-helix piles, the anchoring force is mainly provided by
the top and bottom helices, which implies that a cylindrical shear
surface, which connects the top and bottom helices, is formed and
functions as a resistance against jacking (Fig. 3a).
Fig. 15 illustrates the development of soil temperature along the
pile. The intersections at the vertical axis indicate the freezing
depths of certain moments. The curve shows a strong linear rela-
tionship in the range of 10 to 1 C; however, the temperature
variation is not obvious when the soil is not completely frozen.
This difference is mainly induced by ice-water phase change.
4.2. Optimal designs for the geometric parameters of screw piles
According to the analyses above, the large multi-helix and large
double helix piles perform better than the other piles. A necessary
and economic scheme should be adopted to control ultimate frost
jacking of the foundations. Considering the possible installation
technology in practice and the fact that a larger helical plate diam-
eter (D) results in greater ultimate uplift capacities of screw piles, S
shall be optimized with D = 94 mm. When the freezing depth is
-1 0 1 known from field monitoring, all helices are recommended to be
0.0 deployed below the freezing depth (unfrozen layer), otherwise
the helices in frozen regions will aggravate the frost jacking pro-
cess [11]. The cylindrical shear method performs well for the large
0.1
multi-helix pile (Eq. (15)); however, it is more reasonable to intro-
duce the individual bearing method for the large double helix pile
0.2 (Eq. (18)).
In the first case, n and S exhibit a relationship as follows, thus
only parameter S is considered in this study:
0.3
L  Hf
S 27
n1
0.4
where S is the helix spacing, L denotes the embedment depth of the
pilot point (Fig. 1a), Hf is the freezing depth, and n is the number of
0.5 helical plates. It should be noted that the helix spacing is closely
associated with spacing ratio factor Sf in Eq. (15), which will lead
0.6
to variation of uplift capacities. If the helix spacing S is set to be
0.04 m, 0.05 m, 0.056 m, and 0.07 m, then the ultimate frost jacking
is shown in Fig. 16. The ultimate frost jacking decreases nonlinearly
 T. Wang et al. / Computers and Geotechnics 91 (2017) 2738 37

Fig. 17. Configuration of screw pile with separated helices for optimal design.

Table 4 predicting the frost jacking characteristics of five different piles

Ultimate frost jacking for screw piles with separated helices. (including a pile with no helix) under one-dimensional freezing
n 1 2 (0.24) 2 (0.30) 3 conditions.
To calculate the tension capacities of screw piles in seasonally
Ultimate frost jacking (mm) 7.40 3.02 2.85 2.14
frozen regions, a freezing factor of shear strength jf is proposed,
and the traditional cylindrical shear method and individual bearing
method have been modified. When the freezing depth reaches half
with decreasing helix spacing S. It is effective and economical to
the embedment length of a pile, the large multi-helix pile performs
adopt S = 0.05 m, with ultimate frost jacking equal to 1.97 mm,
the best in terms of anti-jacking ability. Frost jacking of screw piles
because the ultimate frost jacking changed slightly by reducing S
is significantly less than that of the no helix case. Special attention
from 0.05 to 0.04 m.
should be paid to the weak point at top helix, which is subjected to
For the screw pile with separated helices, four types, including
the largest load against jacking. The multi-helix pile with
one single helix, two double helix and one triple helix, were com-
S = 0.05 m and the triple helix pile with S = 0.15 m are identified
pared (Fig. 17). All the piles have one helix fixed at the bottom, and
as optimal layouts for minimizing frost jacking. Geometric param-
the ultimate frost jacking is illustrated in Table 4.
eters of screw piles for the field scale can be obtained through lin-
According to Eq. (18), the tensile capacity of screw piles
ear computations.
increases with an increment of n. Furthermore, the difference
Although the computational method presented in this paper
between the double helix piles with S = 0.24 m and 0.30 m is
has yielded good results, it may only be treated as a simplified pro-
attributed to the freezing factor of shear strength jf, which plays
totype and not the exact solution. Further developments regarding
a role in improving the uplift capacity when the top helix is below
the simulation of the frost jacking mechanism should include
the freezing layer. As a result, the double helix pile with a top helix
moisture migration during the freezing process, and an advanced
closer to the freezing area has generated less ultimate frost jacking.
constitutive model for unfrozen soil considering elastoplastic
It is recommended to adopt a triple helix pile with ultimate frost
behaviour should be established to describe strains on the pile-
jacking of 2.14 mm.
soil interface. Notwithstanding such limitations, this is an appeal-
In view of the analyses above, a multi-helix pile with S = 0.05 m
ing step towards evaluating the in-service performance of screw
and a triple helix pile with S = 0.15 m have been demonstrated to
piles in seasonally frozen regions, and thereafter, design of effec-
be suitable for mitigating frost jacking. This serves as a useful ref-
tive and economical types is recommended.
erence for designing effective and economical pile types in prac-
tice. For the field scale, first, we calculate the freezing depth on
the basis of climate data; then, design a screw pile with the depth Acknowledgements
to top helix (i.e. H in Fig. 1b) slightly larger than freezing depth, and
the helix layout can be easily inferred through linear computations. This research was supported by the National Natural Science
Foundation of China [grant numbers 51378057 and 41371081],
5. Concluding remarks and the scientific and technological development project of Shen-
shuo Railway [grant number 2015-10]. We would like to express
In this paper, the frost jacking performances of screw piles our gratitude to the reviewers for their great efforts in improving
adopted for a photovoltaic generation project in seasonally frozen this article.
regions were simulated using a thermo-mechanical model. Physi-
cal properties of the soil mass varying with temperature were con-
References
sidered using simplified relationships in the calculation process.
Comparisons with experimental results allow one to conclude that [1] Xu Xuezu, Wang Jiacheng, Zhang Lixin. Physics of frozen soil. Beijing: Science
the computational approach described in this study is capable of Press; 2001 [in Chinese].
38 T. Wang et al. / Computers and Geotechnics 91 (2017) 2738
[2] Guo Lei, Xie Yanli, Yu Qihao, You Yanhui, Wang Xinbin, Li Xiaoning.
Displacements of tower foundations in permafrost regions along the
Qinghai-Tibet Power Transmission Line. Cold Reg Sci Technol
2016;121:18795.
[3] Zhang DJY, Chalaturnyk R, Robertson PK, Sego DC, Cyre G. Screw anchor test
program (Part I & II): instrumentation, site characterization and installation.
In: Proceedings of the 51st Canadian geotechnical conference. Edmonton;
1998.
[4] Schmidt R, Nasr M. Screw piles: uses and considerations. Struct Mag 2004:29
31.
[5] Livneh B, Naggar MHM. Axial testing and numerical modelling of square shaft
helical piles under compressive and tensile loading. Can Geotech J 2008;45
(8):114255.
[6] Sakr M. Performance of helical piles in oil sand. Can Geotech J 2009;46
(9):104661.
[7] Sakr M. Installation and performance characteristics of high capacity helical
piles in cohesionless soils. Deep Found J (DFI) 2011;5(1):3957.
[8] Perko HA. Helical piles: a practical guide to design and installation. 1st
ed. USA: John Wiley & Sons Inc.; 2009.
[9] Arup Geotechnics. Design of screw piles: assessment of pile design
methodology. London: Ove Arup & Partners Ltd.; 2005.
[10] Mohajerani Abbas, Bosnjak Dusan, Bromwich Damon. Analysis and design
methods of screw piles: a review. Soils Found 2016;56(1):11528.
[11] Wang TF, Liu JK, Zhao HG, Shang YL, Liu XQ. Experimental study on the anti-
jacking-up performance of a screw pile for photovoltaic stents in a seasonal
frozen region. J Zhejiang Univ Sci A 2016;17(7):51224.
[12] Abdrabbo FM, El Wakil AZ. Laterally loaded helical piles in sand. Alexandria
Eng J 2016;55(4):323945.
[13] Krasinski A. Numerical simulation of screw displacement pile interaction with
non-cohesive soil. Arch Civ Mech Eng 2014;14(1):12233.
[14] Tang Chong, Phoon Kok-Kwang. Model uncertainty of cylindrical shear method
for calculating the uplift capacity of helical anchors in clay. Eng Geol
2016;207:1423.
[15] Gu Quan, Yang Zhaohui, Peng Yi. Parameters affecting laterally loaded piles in
frozen soils by an efficient sensitivity analysis method. Cold Reg Sci Technol
2016;121:4251.
[16] Harlan RL. Analysis of coupled heat-fluid transport in partially frozen soil.
Water Resour Res 1973;9(5):131423.
[17] Jame YW, Norum DI. Heat and mass transfer in freezing unsaturated porous
media. Water Resour Res 1980;16(4):8119.
[18] Alessandro F, Loria Rotta, Laloui Lyesse. The interaction factor method for
energy pile groups. Comput Geotech 2016;80:12137.
[19] Mooney JS, Adamczak S, Clemence SP. Uplift capacity of helix anchors in clay
and silt. Am Soc Civ Eng 1985:4872.
[20] Tappenden K, Sego D, Robertson P. Load transfer behavior of full-scale
instrumented screw anchors. In: Contemporary topics in deep foundations.
USA: ASCE; 2009. p. 4729.
[21] Meyerhof GG. Bearing capacity and settlement of pile foundations. ASCE J
Geotech Eng 1976;102(GT3):197224.
[22] Ministry of Housing and Urban-Rural Development of the Peoples Republic of
China. Technical code for building pile foundation (JGJ 94-2008). China: China
Architecture & Building Press; 2008 [in Chinese].
[23] Adams JI, Klym TW. A study of anchors for transmission tower foundations.
Can Geotech J 1972;9(1):89104.
[24] Chance. Basic guidelines for designing helical piers for underpinning. USA: A.B.
Chance Company; 1993.
[25] Aydin M, Bradka T, Kort D. Osterberg cell load testing on helical piles. Geo-
Frontiers 2011:6674.
[26] Tsytovich HA. Frozen soil mechanics. In: Changqing Zhang, Yuanlin Zhu,
editors. Trans. Beijing: Science Press; 1955/1985 [in Chinese].
[27] Wu Ziwang, Liu Yongzhi. Frozen subsoil and engineering. Beijing: China Ocean
Press; 2005 [in Chinese].
[28] Liu Jiankun, Lv Peng, Cui Yinghui, Liu Jingyu. Experimental study on direct
shear behavior of frozen soilconcrete interface. Cold Reg Sci Technol
2014;104105:16.
[29] Shi Quanbin, Yang Ping, Wang Guoliang. Experimental study on adfreezing
strength of the interface between artificial frozen sand and structure. Chin J
Rock Mech Eng 2016;35(10):214251 [in Chinese].
[30] Tehrani Faraz S, Nguyen Phuong, Brinkgreve Ronald BJ, van Tol A Frits.
Comparison of press-replace method and material point method for analysis of
jacked piles. Comput Geotech 2016;78:3853.
[31] The Editorial Committee of Engineering Geology Manual. Engineering geology
manual. 4th ed. China: China Architecture & Building Press; 2007 [in Chinese].
[32] Wang TF, Liu JK, Tian YD, Lv P. Frost jacking characteristics of screw piles by
model testing. Cold Reg Sci Technol 2017;138:98107.
[33] Wen Z, Yu QH, Zhang MY, Xue K, Chen LZ, Li DS. Stress and deformation
characteristics of transmission tower foundations in permafrost regions along
the Qinghai-Tibet Power Transmission Line. Cold Reg Sci Technol
2016;121:21425.
[34] Li GY, Yu QH, Ma W, Chen ZY, Mu YH, Guo L, et al. Freezethaw properties and
long-term thermal stability of the unprotected tower foundation soils in
permafrost regions along the Qinghai-Tibet Power Transmission Line. Cold Reg
Sci Technol 2016;121:25874.
[35] Mao WN, Liu JK. Different discretization method using in coupled water and
heat transport mode for soil under freezing conditions. Eng Mech 2013;30
(10):12832 [in Chinese].
[36] Liu H, Zubeck HK, Schubert DH. Finite-element analysis of helical piers in
frozen ground. J Cold Reg Eng 2007;21(3):92106.