Cold Regions Science and Technology 71 (2012) 3443

Contents lists available at SciVerse ScienceDirect

Cold Regions Science and Technology

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

Design ice load for piles subjected to ice impact

Jiwu Dong , Zhijun Li, Peng Lu, Qing Jia, Guoyu Wang, Guangwei Li
State Key Laboratory of Coastal and Offshore Engineering, Dalian University of Technology, Dalian 116024, China

a r t i c l e

i n f o

Article history:
Received 25 November 2010
Accepted 1 November 2011
Keywords:
Sea ice
Design load
Cylinder piles
Envelope method
Model test

a b s t r a c t
Design loads of offshore structures for ice-covered seas are typically based on ice-crushing approach. It always, therefore, makes the design conservative, especially for the loads on vertical offshore structures
exerted by moderate ice mass. Taking the case of sheet-pile-type pier, we conducted a series of tests of ice
loads on cylindrical piles subjected to the impacts of drifting ice. The objective is to develop the relationship
between ice kinetic energy and the impact force, and nally to establish the design ice load for the piles. An
envelope of logarithm curve was applied to determine the design load on the piles. The envelopes of ice force
with known compressive strengths with respect to ice kinetic energy were used to determine the expected
peak loads. The expected peak loads with different ice uniaxial compression strength were used to form another envelope to establish the design load on an individual pile for a special compressive strength. With the
same approach, the total ice force on one pier unit was obtained from an envelope curve. The design load on
an individual pile can be applied to real ice as its scaled kinetic energy and compressive strength range from
0 to 0.8 J and 30 to 90 kPa. Considering the large force on the corner pile, special tests of force against the pile
were conducted. The results showed that the design load on the corner pile is 0.78 times that of one dock
unit, and 1.69 times that of the other six piles. Therefore, the corner pile should be reinforced in engineering
applications.
2011 Elsevier B.V. All rights reserved.

1. Introduction
In winter production activities in cold regions are usually infested
with sea ice. Consequently, ice load is considered a signicant factor
in designing marine structures in ice-infested waters (Li et al., 2008;
Yang, 2000). The jacket platform on a piled foundation is the most
common structure for oil exploring platforms (Ou et al., 2007). It
has been demonstrated that when large ice masses with high kinetic
energy collide with vertical-sided structures (i.e. lighthouses, bridge
piers and pile-supported offshore platforms), ice oes fail in crushing
at contact surfaces. Intensive researches have been focused on designing and managing these structures against ice damage in recent
years (i.e. Eik, 2011). In ice-covered waters, the design load of vertical
offshore structures is typically based on a static ice-crushing force
(API RP 2N, 1995; MTPRC, 1998), which is only appropriate when
considering the effects of static ice crushing on structures by large
ice oes. However, in situ observations of ice crushing are difcult.
The eld measurements by Cornett and Timco (1998) demonstrated
that ice crushing loading events on the Molikpaq oil platform at the
Amauligak I-65 site in the Canadian Beaufort Sea occupied 1.1% of
the interactions. Impacts occur more easily when ice oes are small
Corresponding author. Tel.: + 86 411 84708271; fax: + 86 411 84708526.
E-mail addresses: [email protected] (J. Dong), [email protected] (Z. Li),
[email protected] (P. Lu), [email protected] (Q. Jia), [email protected]
(G. Wang), [email protected] (G. Li).
0165-232X/$ see front matter 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.coldregions.2011.11.002

or of low kinetic energy. Consequently, the static crushing approach

is possibly overly conservative in ice load design.
Determinations of the impact load from the interaction of ice oes
on offshore structures can be very complex (Ibrahim et al., 2007). The
load depends on geographical location, ice type, interaction scenario
and structural conguration, etc. Several methods have been proposed for determining impact load on piles by previous studies. Statistical techniques (Bekker et al., 2007; Timco and Frederking, 2004) are
thought to be effective in predicting or determining ice loads for offshore structures. Based on the analysis of more than 170 loading
events, Timco and Johnston (2004) suggested that predicted total
load on caisson structures in the Canadian Beaufort Sea corresponded
to a variety of ice failure modes. Enns and Smith (1984) gave an optimal design strength for a platform subjected to collision of ice oes or
icebergs, using an approach of probability distribution and energy
conservation. However, the type of ice-structure impact was not
taken into account. Morland (1996) studied the direct impact of
rigid structures and viscoelastic ice under wind and ocean drag at a
low ow velocity. However, central collision from direct impact seldom occurs under normal circumstances, and eccentric impact is far
more likely to occur. It has been demonstrated that an eccentric impact leads to a notable decrease in the maximum load when compared
with a head-on collision. Furthermore, the shape of the ice oe has
an inuence on the load due to the difference in collision angle
(Yamaguchi et al., 1997). Moreover, Matskevitch (1997) pointed out
that horizontal splits in the ice edge were a principal factor in ice

J. Dong et al. / Cold Regions Science and Technology 71 (2012) 3443

failure mechanism. In fact, the force of oating plates against structures is determined by the ice mass and its drift velocity (Sodhi and
Haehnel, 2003). That is to say, the kinetic energy is one of the critical
factors affecting the impact force.
Laboratory experimentation is one of the most common methods
to determine ice loads on structures. Frederking and Timco (2000)
carried out a series of tests in an ice tank to measure the impact
force on an isolated oe against an instrumented structure. In the
tests, the ice oe was accelerated to a desired speed by a towing carriage, and made to drift under its own momentum into the test structure. The aim of this paper is to evaluate the design ice load on cylinder
piles exerted by moderate ice by using of logarithm curve envelope
method, and then establish the relationship between ice impact forces
on an individual pile and pile groups and the kinetic energy of the ice
oe through the envelope method. As the carrier of drift ice, ow plays
an important role in ice-structure interaction. Therefore, using a pier
structure of sheet-pile type construction as an example and taking
advantage of the return ow system, we conducted tests of moderate
sized ice oe impacts against vertical column piles according to the
method of Frederking and Timco (2000).

2. Experimental setup and DUT-1 model ice

distance of the two closest piles was 5.63 times that of the pile diameter. In general, there is no interaction of the ice on the two closest
piles as the distance between the two centers is greater than ve to
six times that of the pile diameter (Kato and Sodhi, 1984; Timco,
1985). Therefore, we assessed each pile individually. Bidirectional
force transducers were installed on pile #1, and unidirectional ones
on the other six piles, oriented toward the ow. Pile #1 was on the
upstream corner of the dock unit and therefore, always the rst pile
to be impacted.
Non-refrigerated DUT-1 model ice was employed in the experiments. Previous studies have showed (Li et al., 2002, 2003a) that the
density of DUT-1 model ice is between 876 and 926 kg/m 3, consistent
with Bohai Sea ice, and is able to simulate the ice-structure interaction
with a scaling factor of 1:10 to 1:50. The ice strength can be restricted
within a range by regulating the cement content in the model ice (Li et
al., 2003b). The typical exural strength and compressive strength of
the model ice range from 20 to 75 kPa and 20 to 150 kPa, respectively.
The ratio of compressive strength to exural strength is in the range of
1:0.8 to 1:2. It's worth noting that this ratio is always higher than that
of natural sea ice, as many other model ice materials (Zufelt and
Ettema, 1996), because of the differences in material component.
However, many studies using model ice material has proved that
such differences can be ignored in some conditions (i.e. Leivisk
et al., 2001; Zufelt and Ettema, 1996), for example, in this study
where no exural failure is encountered in the tests.

3. Test procedures and measurements

Drifting oes typically attach to or leave vertical marine structures
with a rotation after an impact, and exural failure of sea ice does not
generally take place in the process. Consequently, the compressive
strength of the model ice was considered to be the main mechanical
parameter in the tests. The model ice used for the tests was 3.0 m in
length, 1.9 m in width and averaged of 1.1 cm in thickness. The compressive strength of the model ice in the tests was in the range of 31
to 86 kPa. The drift velocity and equivalent diameter of the ice oe
were 10 to 25 cm/s and 5 to 259 cm, respectively. The kinetic energy
of the ice oe varied from 0.1 to 1.9 J. A static ice oe was accelerated
to impact the piles by the ow with a certain velocity. The ow velocity varied regularly from the low speed to high speed until this oe

487.5

62.5

75

2x112.5=225

75

50

Experiments were carried out in a water tank, with a geometry scale

of 1:40. The tank was 50 m in length, 3 m in width and 1 m in depth. The
total ux of the bidirectional-return ow system was 0.65 m3/s, and
variations of ow velocity were regulated through a ow control system. Two video cameras were employed to record each impact.
A dock is always constructed of several similar units. To provide a
realistic simulation of the impact on any part of the structure, two
units of the dock model were prepared according to the pile arrangement of an existing dock in the north Bohai Sea (Fig. 1). The diameter
of the piles was 20 mm. One model dock unit was rigged with load
sensors to measure the impact force, and the other model dock unit,
without load sensors, was placed in the upstream direction. The upstream model dock unit was taken away while measuring the load
on the corner pile of the force-measured element. By this way, the
corner pile could resist the maximum loading. The two dock units,
with the seven piles keeping a certain distance from the tank bottom,
were xed to the tank bottom with lead ballast. The center-to-center

35

50

6x112.5=675

50

775
Fig. 1. Piles arrangement of one dock element model (in mm). The 15 angle between the oblique line and the vertical axis represents the angle between the inclined piles and
vertical planes.

36

J. Dong et al. / Cold Regions Science and Technology 71 (2012) 3443

Tank

15o
#7
#6
#5

Pier Unit
Model

#4
#3
#2
#1
CCD Cameras

Pier Unit
Model Without
Load Cells
Pier Unit Model

Ice Movement
Direction

Ice Movement
Direction
Tank

Plan View

Side View

Fig. 2. Simplied cross-section of the test set-up. The seven piles of the pier unit model are marked on the plan view, not including the two CCD cameras and other piles. The two
CCD cameras are xed on a shelf beside the tank. And the pier unit model without load cells is also omitted in the side view.

a
Impact ice force on an individual pile

was broken into small pieces. Then the process was repeated by using
another oe. In this way, the oe could be reused for up to 10 to 30
times, hence the different impact loads under the same compressive
strength could be measured. However, the size of the ice sheet
decreased with the frequency of use as a result of the repeated
interactions.
Both the sampling rate for the load cells and CCD cameras were
0.01 s. The angle of the ow direction was 15 from the center line
formed by the seven vertical piles. Due to the limitation of the tank
width (3 m), there was not enough space for the oating ice (1.9 m
in width). The length of the two pier units was about 1.6 m. Therefore,
the two units were placed on the tank bottom at an oblique angle from
ow direction. The angle could not be too wide, so an angle of 15 was
selected for the tests. The direction of the ice oes was monitored by
one vertically down-looking CCD camera, and the collision of the
piles was monitored by another camera facing obliquely-downward
(Fig. 2).

-1
0

Time (s)
Crunching ice force on an individual pile

20
15

10
5
0
-5
0

Time (s)
Fig. 3. Photo of oating ice impacting against the piles. The black triangle is applied to
calibrate the pictures from the vertical-installed CCD camera. Lead blocks on the pier
unit model are used as ballast.

Fig. 4. The comparison of impact and crushing process on an individual pile for the
same reused oe under different kinetic energy. Fig. 4a indicates a typical impact process. Fig. 4b shows a crushing ice force on an individual pile.

J. Dong et al. / Cold Regions Science and Technology 71 (2012) 3443

37

The kinetic energy and compressive strength of the drifting ice

were the key parameters determining the impact force. To get the kinetic energy of an ice oe, its density, size and velocity should be determined. The density of the model ice was obtained by the massvolume method. The ice drift velocity was analyzed from continuous
images according to Lu et al. (2008). A black triangle of known size
was attached as a reference on the ice surface to calculate the real
size of the ice oe from the images (Fig. 3). The mass of the triangle
itself was negligible. The instantaneous speed of ice oe at the impact
against piles was acquired by analyzing the sequence of several
images showing the ice moving to the pier unit. The ice oes were observed to rebound, rotate, deviate or scrape the piles after impact, and
even broke into several pieces under high ow velocity. Because of
the highly random in interaction of the ice oe and the piles, it is difcult to measure the velocity of the ice oe after an impact, so the
kinetic energy of the oe after impact is not considered here.
In order to simplify the method and the process of the design load,
a certain ice condition was taken as an example in this study. Some
ice parameters were measured. These were: a uniaxial compression
strength of 56 kPa, a thickness of 1.1 cm, a diameter of 2.5 m and a
maximum velocity of 12.65 cm/s (the kinetic energy was 0.35 J). A
curve envelope of the impact force was drawn in respect of the limit
value of the kinetic energy at the impact of the ice and the force on
the piles in each test group. The value of the curve corresponding to
0.35 J kinetic energy was dened as the expected impact load on a single pile for the test group.
A design load is always taken to be the limit value in a recurrence
interval in engineering practice. Therefore, a worst case situation has
been considered. The peak values of the forces on different piles at
the same time were used to plot envelopes to ascertain the expected
peak loads on an individual for a 0.35 J kinetic energy, and with the
same methods, the sum of the force against the several piles of the
seven piles at the same time was regarded as the expected peak
total load on one dock unit. The expected peak loads corresponding
to 0.35 J kinetic energy with different compressive strengths were applied to draw another envelope curve with respect to ice compressive
strength. Thus, the design ice load on an individual pile of 56 kPa compressive strength was nally determined. The total design ice load on
one dock unit could be obtained with the same method, as well as that
on pile #1.
4. Results

4.1. Comparison of crushing and impact ice force against an individual

pile
An ice impact on a single pile is different from that of a quasi-static
crushing process (Fig. 4). The same reused oe of 31 kPa in compressive strength was used in the two processes. There are two impact
pulses shown in Fig. 4a. It shows that the load on the piles increases
rapidly from about zero up to a peak value in several tenths of one second as the oe touches the piles. From there on, the force decreases at
a very rapid rate. Occasionally, the ice oe impacts again on the piles
due to ow drag. There are always one or two interactions before
the oe stays on the ice edge while the ow velocity is low. The
width of the pile penetration the oe is typically lower than the pile
diameter. For the crushing process (Fig. 4b), the ice force moves up
and down during which the mean magnitude remains almost constant. This process always occurs with a high indentation speed, and
the width of the contact surface equaled to the pile diameter. It can
be seen by comparing the two curves that the magnitude of averaging

Fig. 5. The impact of the ice oe on the seven piles. (a) Only one pile touches the oe.
(b) The oating ice impacts on the several piles simultaneously. (c) There are two impacts on a pile in one test run.

38

J. Dong et al. / Cold Regions Science and Technology 71 (2012) 3443

Fig. 6. The expected peak ice forces on an individual pile from different compressive-strength oes with 0.35 J in kinetic energy.

J. Dong et al. / Cold Regions Science and Technology 71 (2012) 3443

crushing ice force is higher than the impact loads. In a word, impact
and crushing failure of oes on piles represented incomplete-contact
and sufcient-contact crushing failure, respectively. The major differences between the two processes are the action duration, the width of
contact surface, and the load magnitude.

4.2. Design ice load on an individual pile

The drag force on the piles under several different ow speeds was
gauged before the tests. The results showed that the ow drag was low
and could be ignored. It can be seen that the impact load on the piles is
different for each interaction (Fig. 5). In some interactions, piles withstood the force concurrently, while sometimes several piles experienced loads asynchronously. One of the piles even received actions
twice during a test, because of a rebound, due to the rotation of the
oating ice. However, the force for the second impact was lower
than the rst one, due to the much slower speed.
An impact force with the same ice strength was selected to plot a
curve envelope using the peak value data (Fig. 6). The values on the
curves relating to 0.35 J kinetic energy on the horizontal axis were
the expected peak loads on an individual pile. There is a cluster of discrete data points in Fig. 6, because the data are the peak loads on the
seven piles in any given impact, and there are several different impacts under varying ow velocities as shown in Fig. 5. Furthermore,
the restriction of the sidewall and a varying shape oe are other reasons for the scattered data.
The expected peak loads from the force-energy envelope on an individual pile in Fig. 6, with respect to a kinetic energy value of 0.35 J,
have been used to plot a new envelope curve (Fig. 7). It shows that
the design load value for an individual pile of 56 kPa compressive
strength is 24 N, which is higher than that calculated from Afanasev's
formula for static crushing failure scenario. It seems that the design
load is higher. However, the most serious condition should be considered for the design of structures within an expected service period in
engineering practices. A proper safety factor has to be applied before
applications. Therefore, envelope method was used for the tests. It
has a marked effect on the loads. From the measured impact forces
to the design loads, the ice loads have been amplied for twice. The
maximum values of the impact loads on the piles in each impact process were used to form the envelopes. That is, the data points in Fig. 6
were the peak values in each impact process. The peak impact loads
from different compressive strengths are used to form the envelope
curves. Then the expected peak loads on the above envelopes were
used to form a new curve to determine the design loads.

4.3. Relationship of design ice load on an individual pile, kinetic energy of

drift ice and its compressive strength
Both the kinetic energy and the compressive strength affect the failure of the oes on the ice-pile interface and thus the force on the piles.
Fig. 8 shows the minimum critical kinetic energy of sufcient-contact
crushing failure for a certain strength oe. That is, sufcient-contact
crushing failure occurs as the kinetic energy greater than or equal to
the critical value. The kinetic energy for a sufcient-contact crushing
failure tends to increase with the rising of compressive strength. The
critical values of the energy for different compressive strengths, however, cannot be accurately dened, due to the large scatter data.
The relationship of the design ice load on an individual pile, ice kinetic energy and ice compression strength were plotted, based on all
the data points to give the envelope curves in Fig. 6. It shows that ice
force on an individual pile increases with increasing ice compressive
strength (Fig. 9). However, the kinetic energy of the plates greatly inuences the ice load on a single pile. This is because the crushing force
is the upper limit of the impact force, and therefore will not increase
with rising kinetic energy after reaching a critical value. The collision
force acting on an individual pile under a compressive strength of between 30 and 90 kPa and a kinetic energy ranging from 0 to 1 J can be
determined from the curved surface.
The impact ice force on an individual pile Fm is:
F m 121:09  3:09 lgEm  13:17 lgEm 2  4:88 lgEm 3  0:51 lgEm 4
2

68:28 lg cm  7:91 lg cm

1
where Em is the oe kinetic energy, and cm is oe compression
strength. Note Eq. (1) is only for the inner piles rather than the corner
pile.
This equation has the potential to apply for real ice impact loadings. For a geometry scale factor of , the kinetic energy E and the
compressive strength c of natural ice is 4 and , respectively.
4

b c =

a E=

where a and b are parameters of natural ice.

If the values of a and b are within the limits of 0 to 0.8 J and 30 to
90 kPa, the force of natural ice on a single pile F can be calculated as
follows:
3

F Fm:

Fig. 7. The design ice load on an individual pile with 56 kPa in ice compressive strength.

39

Fig. 8. The kinetic energy for sufcient-contact crushing failure increases with the rising of ice compressive strength.

40

J. Dong et al. / Cold Regions Science and Technology 71 (2012) 3443

Fig. 9. The relationship of design load on an individual pile, kinetic energy and compression strength of the ice oe.

For example, considering an ice oe of 0.92 Mg/m 3 in density

(Timco and Weeks, 2010), 1000 m 2 in area, 0.4 m in ice thickness,
2.0 MPa in compressive strength, 0.2 ms 1 in velocity, and a cylinder
pile of 0.9 m in diameter, the design load on an interior pile is 1.1 MN.
A paper by Timco (1986) mentions another paper (Ralston, 1979),
describing an appropriate strain rate for an individual pile that can be
estimated by:
_ v=2D

where v is the ice velocity, and D is the structure width.

In terms of the strain rate, the relationship between strain rate and
ice compressive strength has to be discussed for natural ice and DUT-1
model ice. It has been demonstrated that the strength for natural ice is
generally related to strain rate when it is less than 10 2 s 1 (Sodhi
and Haehnel, 2003) (Fig. 10). There is a brittle-ductile transition for
the crushing strength with strain rate ranging from 10 4 to
10 2 s 1, and the peak strength is approximately twice of that in brittle stage. The strain rate from 10 2 to 10 1 s 1 or even higher has
little effect on the ice strength. For DUT-1 model ice, it seems that
there is no brittle-ductile transition for its compressive strength as
the strain rate is within the range 1.94 10 3 to 7.30 10 3 s 1 (Li
et al., 2003b) (Fig. 11). Since only a very limited amount of data are
available, it is difcult to provide an accurate quantitative correlation
between compressive strength of the model ice and its strain rate.
The strain rate for the model ice in the tests ranged from 10 0.5 to
10 0.9 s 1 (within the ductile stage for real ice). The extrapolation for
real ice can be directly done by Eq. (4). When the estimated strain
rate for real ice is higher than 10 2 s 1 the larger of the model ice
strength and that of scaled natural ice is employed in Eq. (1).

Fig. 11. The compressive strength of DUT-1 model ice with respect to strain rate.
From Li et al., 2003b.

4.4. Total design ice load on one dock unit

The sum of the forces on the seven outer piles was taken to be the
total ice load of one pier unit, due to the much lower forces exerted
on the other piles. There were only unilateral transducers on the six
measured piles, the resultant forces against them were modied
according to that on pile #1. The forces on pile #1 mainly ranged
from 40 to 60 (accounting for 70% of all the data) counterclockwise
to the ow with an average of 50.8. The resultant forces on the six
piles were considered to be the same as that on pile #1. Therefore,
the correlation of the measured magnitudes of the load cells on the
six piles FIm and the total loads FI on them can be expressed as:
F I F Im =cos

where I is an integer from 1 to 6, representing one of the six piles, and

the angle of the resultant force on pile #1 counterclockwise to the
ow = 50.8.
Similar to the expected peak force on an individual pile, the summation of the amended values was taken to be the expected peak
total loads on one dock unit. Variations of the expected peak forces
with ice kinetic energy under different compressive strengths are
shown in Fig. 12. The envelope curve of the total design loads on
the structure is plotted by the curves in Fig. 13.
Scatter of the data points in Fig. 12 is largely due to the different
number of piles subjected to the force synchronously. The total design
load is 52 N, only 2.16 times of that on an individual pile for a compressive strength of 56 kPa. This is because the extreme load case
for the piles is the sum of the design ice load acting on the seven
piles at the same time, with the maximum load usually found on
pile #1. Furthermore, as the other six piles are behind pile #1 the
force exerted on them is obviously lower. The asynchronous action
on the piles is another reason for the low value of the total design
load.
The design ice loads on an individual pile and the global loads on
one pier unit with respect to a kinetic energy value of 0.35 J have
been determined from the curves in Figs. 6 and 12 and are listed in
Table 1.
4.5. Design load on the corner pile

Fig. 10. Ice compressive strength with respect to strain rate.

After Sodhi and Haehnel, 2003.

Generally, pile #1 in the upstream pier unit was the rst pile to receive the ice impact. Consequently, the dock unit without sensors was
removed from the tank so that the ice force on the corner pile could be
measured. The results show that the design load on pile #1 is 26 N at
an ice compression strength cm = 36 kPa. The design load can be deduced to be 40 N for cm = 56 kPa under a linear relationship. This
force is 1.69 times of that on an individual pile, and 0.78 times of
total design load. Consequently, the corner pile should be reinforced

J. Dong et al. / Cold Regions Science and Technology 71 (2012) 3443

41

Fig. 12. The expected peak ice forces on one pier unit with 0.35 J in kinetic energy for different compressive strength.

in engineering practice. Eccentric collisions were the most common

occurrences in ice-pile impacts. The ice normally became trapped
with a rotation during an ice oe impact, which caused crushing
loads on the piles. There was a visible variation of force direction on
pile #1with time. The direction of maximum force was not parallel

to the ow direction, but was at a direction of 25 counterclockwise

to it (Fig. 14). Timco (1985) has reported a similar nding. The angle
changed to 50.8 when the model dock unit without load cells was
present (Fig. 15). The variation of angle is due to the lateral restriction
of the tank side walls.

42

J. Dong et al. / Cold Regions Science and Technology 71 (2012) 3443

Fig. 13. Global design load on one pier unit with 0.35 J in kinetic energy of oating ice.

Table 1
Design impact load on an individual pile with different compression strength.
Model ice
thickness
(mm)

Compressive strength
of model ice (kPa)

Design load on an
individual pile (N)

Total design load

on one pier unit (N)

12
11
10
10
10
11
11
12
11

31
31
33
39
56
62
62
65
86

13
14
12
13
21
13
26
20
19

30
30
27
13
26
14
54
42
28

5. Discussion and conclusions

Ice crushing failure is always used in design ice load. This is appropriate only when a structure exposes to large ice oes. However, it

has shown to be a conservative design method for oating ice of moderate size. This is evident when comparing crushing and impact
against a single pile. The differences come from the action mode
and the magnitude of the maximum force. Therefore, an envelope
method was applied in this paper to determine the design ice load
on an individual pile and that on a dock unit under moderate ice
oe impacts.
The kinetic energy of the ice oe before it impacts on the piles is
considered to be one of the key factors for determining the load on
the structure, and it can easily be acquired by analyzing the images
from the downward CCD camera, providing the ice is of a known density and thickness.
The ice compressive strength was between 31 and 86 kPa with an
average thickness of 11 mm. The design ice load was gained under the
following conditions: the ice compressive strength was 56 kPa, the ice
model was 2.5 m in diameter, and the maximum velocity of the ice
was 12.65 cm/s (the kinetic energy was 0.35 J). By plotting the envelope
curve of the peak value of expected peak force with the kinetic energy of
oating ice, the value of design load corresponding to 0.35 J was
obtained from the curve. The design load on an individual pile for the
ice conditions was 24 N. As the kinetic energy and compressive strength
are between 0 and 0.8 J and 30 and 90 kPa, respectively, the design load
for an interior pile can be calculated by Eq. (1). It can be scaled to real ice
when the scaled kinetic energy and compressive strength are within the
ranges stated above. However, due to the variation of compressive
strength of natural ice with strain rate, the strain rate should be previously estimated with Eq. (5). The corresponding compressive strength
for real ice can be compared to that of the model ice. If the compressive
strength of model ice is lower than that of natural ice, it should be
replaced by the latter before applying Eq. (1). When there is no compressive strength data, due to the ductile-brittle transition in compressive strength of real ice, the design load should be increased before
extrapolating to full scale.
The total design load of one pier element was only 2.16 times that
on an individual pile. The design load on the corner pile was 1.69
times that of the other six piles, and 0.78 times that of the total structure. Therefore, it is necessary to increase the load capacity of the corner pile. Due to the lateral restrictions of the tank wall, loading

Fig. 14. Time series of ice force on pile #1 and its direction under the oe of 36 kPa in compressive strength. The meaningless curve in Fig. 14b corresponding to zero loads is removed in Fig. 14a.

J. Dong et al. / Cold Regions Science and Technology 71 (2012) 3443

Fig. 15. Sketch of the direction of the resultant force on the corner pile. Note piles #2 to
#6 and the pier unit without sensors are not shown in the gure. F1 and F2 represent
the resultant force on pile #1 for the absence and presence of the non-measured pier
unit, respectively.

direction on the corner pile was not parallel to the direction of ow,
but was at an angle of 50.8 counterclockwise of ow direction. The
angle changed to 25 when the model without cells was removed.
Due to different failure modes and the restriction of the tank wall,
there is a large scatter for the measured data. The envelope for twice
amplications was used to determine the design loads. All the above
possibly lead to the higher design loads. The design load on an individual pile for envelope method is about 1.7 times of the linear least
square method. The loads on the piles in the tests were taken as statistic force. It was found that there were observable vibrations during
the ice-pile impacts. The dynamic response of the pier units has been
ignored in the tests. It is another factor for the loads, which should be
considered in the future studies.
Acknowledgments
The authors would like to thank the funding supported by the
Foundation for Innovative Research Groups of the National Natural
Science Foundation of China (no. 50921001) and the National Natural
Science Foundation of China (no. 50439010 and no. 51079021).
References
API RP 2N, 1995. Recommended Practice for Planning, Designing, and Constructing
Structures and Pipelines for Arctic Conditions, 2nd ed. American Petroleum Institute, Washington, D.C., USA.
Bekker, A.T., Sabodash, O.A., Shubin, O.A., 2007. Probabilistic aspects of modelling of
extreme ice loads on engineering structures for Sakhalin offshore projects. In:

43

Yue (Ed.), Recent Development of Offshore Engineering in Cold Regions. POAC07, Dalian, China, pp. 7589.
Cornett, A.M., Timco, G.W., 1998. Ice loads on an elastic model of the Molikpaq. Applied
Ocean Research 20 (12), 105118.
Eik, K., 2011. Sea-ice management and its impact on the design of offshore structures.
Cold Regions Science and Technology 65 (2), 172183.
Enns, E.G., Smith, B.R., 1984. Optimal platform strength in the presence of moving ice.
Ocean Engineering 11 (3), 239244.
Frederking, R., Timco, G., 2000. Sea ice oe impacts large scale basin experiments.
Proceedings of the 10th (2000) International Offshore and Polar Engineering Conference, ISOPE'00, Seattle, USA, pp. 640645.
Ibrahim, R.A., Chalhoub, N.G., Falzarano, J., 2007. Interaction of ships and ocean structures with ice loads and stochastic ocean waves. Applied Mechanics Reviews 60,
246289.
Kato, K., Sodhi, D.S., 1984. Ice action on two cylindrical structures. Journal of Energy Resources Technology 106 (1), 107112.
Leivisk, T., Tuhkuri, J., Riska, K., 2001. Model tests on resistance in ice-free ice channels. Proceedings of 18th International Conference on Port and Ocean Engineering
under Arctic Conditions, POAC'01, Ottawa, Canada, 2, pp. 881893.
Li, G., Liu, X., Liu, Y., Yue, Q., 2008. Optimum design of ice-resistant offshore jacket platforms in the Bohai Gulf in consideration of fatigue life of tubular joints. Ocean Engineering 35 (56), 484493.
Li, Z., Li, G., Shen, Z., Wang, Y., Qu, Y., 2003a. Physical properties and elastic modulus of
DUT-1 model ice. Journal of Hydrodynamics 15 (5), 98102.
Li, Z., Wang, Y., Li, G., 2002. On the exural strength of DUT-1 synthetic model ice. Cold
Regions Science and Technology 35 (2), 6772.
Li, Z., Wang, Y., Wang, X., Li, G., 2003b. Effect of cement content and curing period on
properties of DUT-1 synthetic model ice. Journal of Offshore Mechanics and Arctic
Engineering 125 (4), 288292.
Lu, P., Li, Z.J., Zhang, Z.H., Dong, X.L., 2008. Aerial observations of oe size distribution in
the marginal ice zone of summer Prydz Bay. Journal of Geophysical Research 113,
C02011. doi:10.1029/2006JC003965.
Matskevitch, D.G., 1997. Eccentric impact of an ice feature: linearized model. Cold Regions Science and Technology 25 (3), 159171.
Morland, L.W., 1996. Dynamic impact between a viscoelastic ice oe and a rigid structure. Cold Regions Science and Technology 24 (1), 728.
MTPRC, 1998. Ministry of Transport of the People's Republic of China, Load Code for
Harbour Engineering JTJ 215-98. China Communications Press, Beijing, China. (in
Chinese).
Ou, J., Long, X., Li, Q., Xiao, Y., 2007. Vibration control of steel jacket offshore platform
structures with damping isolation systems. Engineering Structures 29 (7),
15251538.
Ralston, T.D., 1979. Sea ice loads. Technical Seminar on Alaskan Beaufort Sea Gravek Island Design, Houston, USA.
Sodhi, D.S., Haehnel, R.B., 2003. Crushing ice force on structures. Journal of Cold Regions Engineering 17 (4), 153170.
Timco, G.W., 1985. The force of a moving ice cover on a fair of vertical piles. Proceedings Canadian Coastal Conference, St. John's, Newfoundland, Canada, pp. 349362.
Timco, G.W., 1986. Ice forces on multi-legged structures. Proceedings of 8th IAHR Symposium on Ice, Vol. 2. Iowa City, USA, pp. 321337.
Timco, G.W., Frederking, R., 2004. Probabilistic analysis of seasonal ice loads on the
Molikpaq. Proceedings of the 17th International Symposium on Ice, IAHR'04,
Saint Petersburg, Russia, Vol. 2, pp. 6876.
Timco, G.W., Johnston, M., 2004. Ice loads on the caisson structures in the Canadian
Beaufort Sea. Cold Regions Science and Technology 38 (23), 185209.
Timco, G.W., Weeks, W.F., 2010. A review of the engineering properties of sea ice. Cold
Regions Science and Technology 60 (2), 107129.
Yamaguchi, H., Toyoda, M., Nakayama, H., Rheem, C., Matsuzawa, T., Kato, H., Kato, K.,
Adachi, M., 1997. Inuence of oe shape on behavior of ice oes around a structure.
Proceedings of the 16th International Conference on Offshore Mechanics and Arctic
Engineering/14th International Conference on Port and Ocean Engineering under
Arctic Conditions (Joint OMAE/POAC Conference), Yokohama, Japan, pp. 461468.
Yang, G., 2000. Bohai Sea ice conditions. Journal of Cold Regions Engineering 14 (2),
5467.
Zufelt, J.E., Ettema, R., 1996. Model ice properties. USA CRREL Report 96-1. Hanover,
NH, USA.