Vous êtes sur la page 1sur 5

9th National Congress on Theoretical and Applied Mechanics, Brussels, 9-10-11 May 2012

Study of the resistance of bollards against impact loads


B. De Clerck, E. Segers, D. Kakogiannis, J.Vantomme, S. Suleau
Koninklijke Militaire School, Department of Construction and Materials
Renaissancelaan 30 - 1000 Brussel
email: declercb@student.rma.ac.be
AbstractThe objective of this study is to identify a sim-
plied approach that can be used in the initial phases of the
design process of bollards. Simplications are proposed to
evaluate a design without having to execute complicated nu-
merical simulations or full-scale tests. For classical bollards,
an analytical model is proposed, based on a simplication
of the impact load. This analytical model is subsequently
compared with a numerical simulation to assess its useful-
ness. Furthermore, following the increasing popularity of
shallow mount bollard systems (SMB), a parametric study is
performed on such a system in order to determine whether
and when it can be compared to a classical bollard. The use
of such models could limit the initial development time and
thus the cost of a bollard.
Keywords bollards,vehicle impact, analytical modeling,
NCTAM 2012
I. INTRODUCTION
The bollards that are treated in this paper do not concern
the decorative type, but they serve as anti-ram barriers and
are meant to be able to prevent a vehicle, whether its errant
or deliberate, from entering a certain perimeter. Currently
there are no theoretical models or guidelines for the devel-
opment of bollards. This absence can be explained by the
fact that bollards always have to pass a full-scale impact
test before being able to be sold to serve as anti-ram bar-
rier, an approach that is unrealistic for buildings. The tools
that are mostly being used are complex numerical simu-
lations on the one hand and full-scale impact tests on the
other. A brief overview of existing classications methods
will be presented followed by the simplication of the im-
pact load, an analytical model for classic bollards, a com-
parison with numerical results and the parametric study of
a Shallow Mount Bollard (SMB) system.
II. CLASSIFICATION METHODS
Manufacturers use labels in order to be able to identify
the quality of a bollard. Three big different labels can be
distinguished, according to the European or the American
approach.
A. CEN (European Committee for Normalisation) Work-
shop Agreement 16221
This agreement denes the different performance cat-
egories, testing methods and guidelines concerning the
placement of vehicle barriers on a European level. This
document is largely based on the British standards PAS 68
and PAS 69. This is one of the reasons that manufacturers
still have a tendency to advertise their product as full PAS
68 tested. The European approach leaves more details to
the classication due to a larger amount of vehicle types
and more exibility concerning the angle of approach, the
vehicle penetration and the distribution of the debris.
B. Department of State (DoS) K-rating
This classication method is most common in the
United States, but is very strict in comparison with the Eu-
ropean approach. Only perpendicular barrier impacts with
a 6810 kg vehicle are considered at three different speeds
(80, 65 and 48 km/h; K12, K8 and K4 respectively) and
the bed of the truck must not penetrate the barrier by more
than 0,91 m.
C. ASTM guidelines
In response of the very strict DoS rating, the Ameri-
can Society for Testing and Materials (ASTM) proposed a
classication comparable to the DoS guidelines, but whilst
allowing different vehicle types and speeds. Different pen-
etration depths are allowed as well, since it is not always
required to keep this below 1 m.
III. THE LOAD
The nature of an impact problem is dynamic, however
it is common to use an equivalent static load to determine
whether or not structure will fail. The European standard
EN 1991-1-7 gives some indicative equivalent static de-
sign forces due to vehicular impact on members support-
ing structures over or adjacent to roadways. The area of
application for the load is described as well, in function of
the vehicle type. In annex C of this document further indi-
cations are given on how to include explicitly the dynamic
effects, using a simplied formula:
F = v
r

kM (1)
In the above formula, v
r
is the resultant impact velocity,
M the vehicle mass and k the equivalent stiffness of the
vehicle. The impacted structure is supposed to be stiff and
rigid and the vehicle is supposed to deform linearly. The
problem when using this expression is the need to know
1
9th National Congress on Theoretical and Applied Mechanics, Brussels, 9-10-11 May 2012
the equivalent vehicle stiffness. Determining the stiffness
of a vehicle is a process that requires a lot of experimental
data and there is no standardized denition of equivalent
vehicle stiffness. Additionally, the stiffness varies greatly
with the type and the age of vehicle.
In order to tackle the problems related with the deter-
mination of the equivalent static load, an analysis of the
real dynamic load is performed. Two different studies are
consulted, one concerning vehicle impact on bridge piers
[1] and another concerning vehicle impact on bollards [5].
Both studies show the resultant contact force in function
of the time. A typical result is shown in gure 1. This
function is comprised of a relatively low force level that is
sustained over the duration of the impact event combined
with several large spikes. The sharp spikes occur when
stiff or heavy components of the vehicle, such as the chas-
sis or engine block, reach the pier and interact with it. As
the approach speed increases the rst signicant spike oc-
curs earlier in time compared to slower approach speeds.
Fig. 1. Resultant contact force in function of the time for
different impact velocities of a small truck on a bridge
pier [1]
Finally, the following approach was retained to simplify
to load on the bollard. A symmetrical, triangular impulse
load is applied on a specic area of the bollard. The size
of this area is chosen according to the guidelines for vehi-
cle collisions with structures as described in EN 1991-1-7
[2] and depends on the type of vehicle. For cars, the ap-
plication area is 0.25 m by the width of the bollard at 0.50
m above the level of the carriageway. For lorries, the ap-
plication area can be at any height between 0.5 m to 1.5
m above the level of the carriageway and is 0.5 m by the
width of the bollard. Due to the limited height of a bol-
lard, the application area for lorries is always taken at 0.50
m above the level of the carriageway.
The symmetrical impulse load is obtained by neglecting
the smaller spikes in gure 1 in favour of the impact of the
engine block, since this constitutes the most import part of
the contact force. The orders of magnitude for the parame-
ters of the impulse, i.e. the peak value and the duration, are
chosen to correspond with the experimentally validated re-
sults from the previously mentioned publications and thus
correspond with the impact of a small truck, typically used
to obtain the DoS K-rating.
This simplication allows using an equivalent single de-
gree of freedom system and simulations without the use of
a vehicle model and thus drastically lowers the required
computational power.
IV. EQUIVALENT SINGLE DEGREE OF FREEDOM
(SDOF) SYSTEM
A. Simplied bollard model
In this case the bollard is considered as a clamped beam
with a distributed load applied onto it, as shown in gure
4(a). The elastic deformations are determined using clas-
sical Euler beam theory. When the applied load exceeds
the elasticity limit of the beam, plastic deformation will
occur. As a hypothesis, it is posed that a plastic hinge will
form at the base, since this is where the largest moment
occurs. The moment required for plastic deformation can
be determined with the following expression:
M
p
= f
y
Z (2)
In the above equation f
y
denotes the yield strength of the
material used and Z is a value which depends on the shape
of the section. For example for a hollow, circular bollard
the following expression is valid for Z:
Z =
1
6
(D
3
d
3
) (3)
D denotes the outer and d the inner diameter of the hollow,
circular bollard. Additional expressions for other sections
can be found in specialized literature [6]. The resistance
function of the bollard is dened by the force one needs to
apply to obtain a certain deformation. Figure 2(b) shows
this function schematically.
(a) Simplied bol-
lard
Deformati on
F
o
r
c
e
0
ye
Rm
(b) Resistance function
Fig. 2. Simplications regarding classical bollards and the
associated resistance function
2
9th National Congress on Theoretical and Applied Mechanics, Brussels, 9-10-11 May 2012
B. Equivalent SDOF system
Biggs [3] described a method that allows determining
the equivalence between a structure and system with a cer-
tain number of degrees of freedom. Since the movement of
the bollard is unidirectional, a SDOF system is sufcient
to determine the equivalence. This equivalence is kine-
matic in nature i.e. the displacement, velocity and accel-
eration correspond with those of the bollard. In this study,
no damping is taken into account; the equivalence is made
with a mass-spring system. The expressions for the stiff-
ness, the mass and the load of the equivalent SDOF system
have to be determined because the model that is proposed
is not a standard case that is described by Biggs. They are
all function of the mass, the stiffness and the geometry of
the bollard as well as the applied load.
The equivalent SDOF model gives the ratio of the max-
imal dynamic deection and the deection that would be
caused if the same force is applied statically. This ratio
is called the dynamic load factor (DLF). The magnitude
of the DLF is a rst indication whether or not a structure
will resist. Typical values that are found in the literature
cite that in order to avoid excessive deformation the DLF
should be kept below 10 [8]. The strength of this model
lies in the small number of parameters that one must know
(the bollards geometry and material on the one hand and
the parameters of the impulse on the other) and in the very
limited computation time.
C. Inuence of high deformation speed
Metals under high deformation rates typically show a
higher resistance. For the equivalent SDOF model, this
can be taken into account by increasing the yield strength
of the material. This is obtained by multiplying the yield
strength with a so-called dynamic increase factor (DIF).
The value of the DIF depends on the nature of the load.
These values can be found in specialized literature [7].
V. NUMERICAL SIMULATION
A. General
A numerical simulation is performed using the explicit
dynamics module from Ansys [12]. Two different bol-
lard cross section are considered, a hollow circular cross-
section and a hollow square cross-section, with a central
reinforcement. The round bollard has an outer diameter
of 350 mm and an inner diameter of 270 mm. The square
bollard has an outer dimension of 350 mm and an inner
diameter of 270 mm. Different bollard lengths are consid-
ered, i.e. a 750 mm and a 1000 mm long bollard. These
lengths are retained in order to study impact of respectively
cars and lorries, whilst respecting the applications areas as
described in section III. Figure 3 shows the 750 mm model
of both bollard models. All the nodes at the bottom layer
of the bollards are xed in all degrees of freedom in or-
der to create the same effect as a bollard with xed steel
over concrete foundation in real life. This method of x-
ing has already successfully been applied [5] to validate a
numerical model.
Fig. 3. The bollard models that were used
B. Material models
Two different materials are used for the bollards. Firstly
AISI 5000 series steel is used. An elastic perfectly plastic
material model was presumed. The properties that are used
are shown in table I.
TABLE I
MATERIAL PROPERTIES OF AISI 5000 STEEL
Density 7850 kg/m
3
Modulus of elasticity 205 GPa
Poissons ratio 0,29
Bulk modulus 162.7 GPa
Shear modulus 79.46 GPa
Tensile strength, yield 647 MPa
Secondly, in order to take the dynamic material be-
haviour into account, 4340 steel is used. This material
is used in combination with the Johnson Cook strength
model, which takes into account the deformation speed.
This material is selected because the parameters for the
Johnson-Cook model are available [9] and because its
properties in general are suited for bollards.
C. Results
A number of simulations are performed for impulses
with peak values ranging from 3000 to 15000 kN. Each
time the corresponding calculation with the equivalent
SDOF model is performed as well. Proper deformation
is obtained for both bollard models, when compared to
the results from other work. The comparison between
the equivalent SDOF system and the results of the simula-
tion showthat the deformations predicted by the equivalent
model are overestimated. This is due to the hypotheses of
the formation of a plastic hinge at the base, while in reality
a zone of plastic deformation occurs. This phenomenon is
3
9th National Congress on Theoretical and Applied Mechanics, Brussels, 9-10-11 May 2012
visible in gure 4, where a deformed bollard after impact
is shown next to the distribution of the plastic strain.
(a) (b) (c)
Fig. 4. Deformed bollard and plastic strain distribution
(a) Bollard post impact (b) Equivalent plastic strain in
bollard (front view) (c) Side view
D. Limitations
When the simplied impulse load is applied to a bollard
with a hollow square cross-section the approach should not
be used. The impacted face will show unrealistic local de-
formations and fold inward. This is because the impacting
body is replaced by a distributed load. The problem oc-
curred for ratios of wall thickness of the bollard to bollard
width smaller than 0.25. This problem can be avoided by
lling the bollard with concrete, in practice this is often
done as well to avoid buckling of the bollard. However, an
appropriate concrete model should then be used. Another
option is to add a central reinforcement in the square bol-
lard as is shown for the right model in gure 3. For the
circular bollard, this problem did not occur.
VI. SHALLOW MOUNT BOLLARDS
A. General
Most SMB systems share the same basic layout, i.e. a
metal frame xed in the ground with concrete. The bol-
lard is placed in this frame. Following this lay-out, a sim-
ple model, based on an existing product is proposed. The
model that is studied is shown in gure 5. Only one bollard
is placed in the frame at the central cross-beam, because
the presence of the other two only increases the computa-
tion time and they arent impacted. A parametric study is
performed to gain some insights in the working of shallow
mount bollard systems.
B. Materials
In general, the installation of a SMB system does not
require the use of high strength concrete. For the analysis
C35/45 concrete is used. To implement this in the sim-
ulation, the CONC-35MPA material model is used. This
concrete model was designed specically for the dynamic
loading of concrete and incorporates the RHT concrete
strength model [10]. The metal of the frame and the bol-
lard are assumed to be made out of the same steel with an
elastic-perfectly plastic behaviour since this gave good re-
sults for the simple bollard. To be consistent with existing
data, the yield strength is set to 550 MPa.
Fig. 5. The studied SMB system
To further reduce the computation time, the soil be-
neath and the concrete around the systemare not modelled.
However to make sure that the results correspond with re-
sults from more elaborate simulations, different boundary
conditions were evaluated. Fixing the back of the three
cross-beams and the two extremities of the long beam in
all degrees of freedom gives the best results.
The central bollard is submitted to an impulse load as
described in section III. The deformations and plastic
strains that are obtained with this simplication show good
agreement with results obtained in another study [11]. The
absence of the surrounding concrete however allows for
slightly bigger deformations than would occur in reality.
C. Results
A comparison is made between a model with and with-
out concrete in the frame and in the bollard. This shows
that lling the bollard with concrete prevents buckling and
reduces maximal plastic strain in the bollard by up to 30%.
A parametric study is performed where two parameters
are varied, namely the wall thickness of the bollard and the
wall thickness of the frame in the ground. These calcula-
tions require limited computational power, but still allow
to clearly seeing the evolution of the behaviour of the sys-
tem. The relation between the deformations, the maximal
stresses etc. are found to evolve in a non-linear way. The
results of the parametric study are represented on several
curves to make a rst evaluation of a design. As an exam-
ple of such a curve, the inuence of the wall thickness of
the bollard on the maximal plastic strain in the bollard is
shown in gure 6. This curve is based on the results of the
simulations.
At rst sight, the wall thickness of the frame of the
frame in the ground does not have a large inuence on the
4
9th National Congress on Theoretical and Applied Mechanics, Brussels, 9-10-11 May 2012
results, however it needs to be adequately high in order to
limit the shear stresses, which could result in failure. The
wall thickness of the bollard needs to be sufciently high
as well to limit the maximum plastic strain and avoid fail-
ure.
Bollard wall thickness [mm]
M
a
x
i
m
a
l
p
l
a
s
t
i
c
s
t
r
a
i
n
i
n
b
o
l
l
a
r
d


25 50 75 100 125 150
10e!5
10e!4
10e!3
10e!2
10e!1
10e0
10 mm
20 mm
30 mm
Frame wall thickness
Fig. 6. Inuence of bollard wall thickness on maximum
plastic strain in the bollard for different thicknesses of
the frame walls
VII. CONCLUSIONS AND FUTURE WORK
A. Conclusions
In order to reduce computation time in the initial de-
sign phase of bollards, a simplied load function is pro-
posed. This load function also permits the development of
an equivalent SDOF system. The different results that are
obtained serve as a conrmation that this approach could
be used to evaluate a model in its initial phase.
The analogy between a structure and a system with one
or several degrees of freedom as proposed by Biggs is de-
veloped for a classical bollard. The bollard is considered
as a clamped beam. This allowed developing an equiv-
alent SDOF system that gives a rst estimate whether or
not a bollard will resist a certain impulse. The compari-
son between the estimates of the analytical model and the
numerical simulation shows that the former overestimates
the deformations that will occur. This is not necessarily a
bad thing, as this is along the safe side.
The analysis of a SMB system provides some insights
into the workings of such a system and serves as an extra
motivation to use the simplied load. The parametric study
allows creating some interesting curves that can be used to
optimize and assess a certain design.
B. Future work
The results that are obtained are promising, however to
fully be able to rely on this design method, more elaborate
comparisons between the analytical models predictions,
numerical simulations and full-scale impact test results are
necessary. Furthermore, due to the limited available data,
only impulses corresponding with a small truck impact are
used. For a complete evaluation of a bollard to withstand
a certain impacts, it is necessary to determine indicative
values be used for different vehicle types.
REFERENCES
[1] Sherif El-Tawil, P.E. et al., Vehicle Collision with Bridge Piers,
Journal of Bridge Enginering, may/june, 2005.
[2] CEN, EN 1991-1-7: Actions on structures - Part 1-7: General ac-
tions - Accidental actions, 2006
[3] John M. Biggs, Introduction to Structural Dynamics, McGraw-Hill
Book Company, 1964.
[4] J. Vantomme, Benaderende berekening van constructies belast
door schokgolven te wijten aan explosies, Koninklijke Militaire
School, 2010.
[5] Bangalore Krishna-Prasad Protective Bollard Design for High
Speed Impact Energy Absorption, Wichita State University, 2006.
[6] U.S. Army Corps of Engineers, HNDM-1110-1-2: Suppressive
Shields - Structural Design and Analysis Handbook, 1977.
[7] Departements of the Army, the Navy and the Airforce, The Effects
of Accidental Explosions (TM 5-1300), 1990.
[8] U.S. Army Corps of Engineers, Naval Facilities Engineering Com-
mand and Air Force Civil Engineer Support Agency, UFC-3-340-
02: Structures to Resist the Effects of Accidental Explosions, 2008.
[9] Gordon R. Johnson and William H. Cook, A constitutive model
and data for metals subjected to large strains, high strain rates
and high temperatures, 1983.
[10] W. Riedel, Beton unter dynamischen Lasten: Meso- und
makromechanische Modelle und ihre Parameter, Fraunhofer IRB
Verslag, 2004.
[11] John E. Crawford, et al. Development of shallow footing anti-
ram bollard system through modeling and testing, 1
st
International
Conference on Analysis and Design of Structures against Explo-
sive and Impact Loads, 2006.
[12] ANSYS, Inc. ANSYS R Academic Research, Release 13, Help
System, Explicit Dynamics Guide
5

Vous aimerez peut-être aussi