Professional Documents
Culture Documents
Structures
journal homepage: www.elsevier.com/locate/structures
A R T I C LE I N FO A B S T R A C T
Keywords: In this study, analytical investigation and numerical simulations are utilized to examine the responses of re-
Reinforced concrete inforced concrete bridge columns (RCBC) against vehicle collisions. Based on the numerical results, a simplified
Bridge approach is developed for analysis and design of RCBCs to resist vehicle collisions. RCBCs impacted by a medium
Vehicle truck and a heavy truck trailer at different velocities are considered. Based on the numerical results, empirical
Collisions
equations to determine the maximum shear force and bending moment at column critical sections are proposed.
Impact loads
Degree of freedom
A single-degree-of-freedom (SDOF) system is employed to predict the dynamic response of the column. A pro-
cedure to design RCBCs under vehicle collision with either flexural bending or brittle shear failure governed
response of the column is proposed. Two design examples of RCBC under medium truck impact and heavy truck
impact are given in this study to demonstrate the proposed procedure.
1. Introduction in a contact event. CEN [3] and SA/SNZ [4] estimate the maximum
static force from collision events by considering the vehicle velocity,
In recent decades, a number of vehicle collision accidents with vehicle mass, and deformation of both column and vehicle model.
bridge structures have been documented in the open literature and However, many studies have indicated that these design approach
media [1,2]. Collisions from heavy-duty trucks or high-velocity vehicles based on ESF analysis could result in un-conservative designs since the
may cause failures of substructures, cost human life, and paralyze influences of the dynamic responses of structures and high loading rate
transportation systems in urban areas. Therefore, it is crucial to un- of the impact force have been completely neglected [9,12]. Moreover,
derstand and consider the responses of reinforced concrete (RC) col- the ESF method could not predict some of the failure modes of RCBCs as
umns/bridge piers under vehicle collisions in the design stage. Studies observed in real vehicle collisions and mentioned in the previous re-
on the performance and response of structures under impact loads and ports [1,12], e.g. diagonal shear closes to the column top and combined
vehicle collisions have attracted a number of research interests and flexural-shear damage at the column mid-height. In addition, the actual
efforts. Currently, three methods including an equivalent static force dynamic response of RCBCs is also completely different from a pre-
(ESF) [3–8], damage assessment of column structures [9,10], and dy- diction by using the ESF method, especially during the impact force
namic analysis [11–18] have been utilized to examine the response of phase. Therefore, concerns are still persisted about the applicability of
column structures under vehicle collisions. those design methods and recommendations.
Among these three methods, the ESF is commonly used in design To overcome the limitations of the ESF, the damage assessment
specifications and guides since it is straightforward for engineers to use method has been proposed based on failures of reinforced concrete
in design analysis. For instance, AASHTO [8] suggests the ESF of 2668 columns under various loading conditions [9,10]. Sharma et al. [9]
kN applied to the column at 1.5 m above the ground level to design used four different vehicle models ranging from 8 ton to 50 ton together
bridge columns under vehicle impacts. CEN [19] recommends an with various impact velocities between 65 km/h and 161 km/h to im-
equation to predict the maximum contact force from the vehicle colli- pact on the RCBCs. Based on the dynamic shear force from the collisions
sion based on the initial kinetic energy of the truck model and the and the dynamic shear capacity of the column, the impact performance
stiffness of the softer one of the column structure and the vehicle model of the RCBC has been categorised in three groups, i.e. fully operational
⁎
Corresponding author.
⁎⁎
Correspondence to: H. Hao, Center for Infrastructural Monitoring and Protection, School of Civil and Mechanical Engineering, Curtin University, Kent Street,
Bentley, WA 6102, Australia.
E-mail addresses: tin.v.do@postgrad.curtin.edu.au (T.V. Do), thong.pham@curtin.edu.au (T.M. Pham), hong.hao@curtin.edu.au (H. Hao).
https://doi.org/10.1016/j.istruc.2019.08.011
Received 20 March 2019; Received in revised form 1 July 2019; Accepted 12 August 2019
2352-0124/ © 2019 Published by Elsevier Ltd on behalf of Institution of Structural Engineers.
T.V. Do, et al. Structures 22 (2019) 213–229
Lspan Lspan
0.5 m
RC column
H
Ground level
Footing
2000
1500
1200
1000
800
with no concrete damage, an operational structure with concrete da- behaviours and responses of the RCBC under vehicle collision. The
mage, and total collapse of structures. However, the mentioned study study indicated that with different initial conditions, which causes a
mainly considered the failure at the column base whereas the flexural different PIF, the column could exhibit different failure modes from
failure or shear failure at the top or the mid-height of the column as minor damage due to flexural response to diagonal shear failure or
observed in some real accidents was not considered. Zhou and Li [10] punching shear failure. Empirical relations of the PIF and the total
used the damage index, λ, which was defined by dividing the local ESF impact impulse as a function of the initial velocity of the vehicle model,
to static shear capacity of the column, to categorize the damage of the engine mass, and total vehicle mass were proposed based on intensive
column in four groups, i.e. slight damage (0 ≤ λ ≤ 0.2), moderate numerical simulation results. In a subsequent study, the equations to
damage (0.2 ≤ λ ≤ 0.6), severe damage (0.6 ≤ λ ≤ 1) and collapse estimate the entire impact force profile including vehicle bumper's
(λ ≥ 1). It should be noted that the local ESF in the latter study is the impact, engine impact, vehicle trail's impact, and cargo impact together
averaged integration of the impact force time histories in 50 ms during with the column dynamic shear capacity have been proposed by Do
the impact duration. However, in that study, neither the behaviours of et al. [14]. Based on the PIF from a collision event and the dynamic
the RCBC under high impact velocity (higher than 80 km/h) nor the shear capacity of RCBCs, the shape of the shear force and bending
dynamic effects associated with the high-speed and high peak impact moment distributions along the column and the column failure mode
force (PIF) were considered. have been divided into two separated groups, i.e. flexural responses and
In the third approach, detailed finite element (FE) models were used shear responses [14]. Although the detailed FE model simulations were
and the time histories of the impact force and the dynamic response of proven yielding accurate predictions of column responses [12,14], they
the column such as shear force, bending moment, and inertia force have are not straightforward to use in design analysis. Therefore, a
been predicted [11,12,20]. Based on simulations of vehicle model im- straightforward procedure to reliably predict dynamic responses of
pacted on a rigid column, response spectra for the PIFs from the engine RCBC under vehicle impact is still required for design analysis.
and cargo were proposed by Chen et al. [11]. The time histories of the This study aims to propose a design procedure of RCBC to resist
reaction force at the column base were then estimated. Since the vehicle collisions by taking into consideration the vehicle impact con-
column was assumed rigid in the simulation, the contribution of the dition, vehicle-column interaction, and dynamic effects on column re-
vehicle-column interaction and inertia resistance which significantly sponses. By adopting the impact force time histories models from the
affect the shear force and bending moment of the column were not truck impact and classifications of the column failure proposed by Do
considered in the study. The dynamic response of the column such as et al. [14], a procedure to estimate the column internal forces and
shear failure, shear cracks or flexural response was, therefore, not predict the column failure mode is proposed in this study. The proposed
mentioned and discussed. Do et al. [12] developed detailed FE models procedure avoids detailed FE model simulations but yields accurate
and carried out numerical simulations to investigate the dynamic estimations of the maximum shear force and bending moment at
214
T.V. Do, et al. Structures 22 (2019) 213–229
Table 1
Column dimensions and properties.
No Column section Dimensions (mm) Reinforcements Pdynmax (kN)
column critical sections and lateral displacement of the RCBC under 2.2. Model verification
vehicle impacts. Two design examples of RCBCs under vehicle collisions
are presented in this study to demonstrate the proposed procedure for It is worth mentioning that currently no available vehicle impact
its use in design analysis. test on the full-scale RCBC has been reported in the open literature due
to the safety requirements, as well as the high costs associated with
such experimental tests. Thus, to verify the numerical model of the
2. Simulation of vehicle collision of RCBC and its verification vehicle impact on RC structures, many previous studies used the impact
tests on scaled structures [9,10], or the vehicle collision on the steel
2.1. Numerical model of RCBC and vehicle model bollards [28]. In this study, the pendulum impact test on a scaled RC
column [23] and the real vehicle collision on a RCBC [1] (Accident
In this study, to develop the procedure for estimating the column number 18) are simulated to validate the reliability of the above
responses and verifying its accuracy, a numerical simulation of a full- modelling techniques, material models, strain rate effects, and contact
scale bridge structure is built in LS-DYNA [21]. The bridge structure algorithm.
consists of a single RC column, two spans of superstructures, and con- Firstly, based on the experimental test set up [23] and the presented
crete abutments (see Fig. 1a). In numerical analyses, responses of five numerical methodology, a FE model of a scaled RC column impacted by
square columns with the size of 800 mm, 1000 mm, 1200 mm, a solid steel impactor is built. The column which had a cross-section
1500 mm, and 2000 mm are considered under multiple impact condi- area of 100 × 100 mm2 and a height of 800 mm was impacted by the
tions of two vehicle models as shown in Fig. 1b and Table 1. The cross- solid steel impactor with the mass of 300 kg. The impact velocity was
section dimensions of the superstructures are adopted from Megally 0.64 m/s. The comparisons of the simulation results and experimental
et al. [22] but its span length, Lspan, is varied with column cross-section results are presented in Fig. 3. The figure demonstrates that the simu-
dimensions to keep the total dead load from the superstructures to be lation results match well with the test results in terms of the maximum
10% of the column compression capacity of each column model in the impact forces (23.7 kN versus 22.0 kN), impact duration (35 ms versus
analysis. The slenderness of these columns keeps at 8, similar to that 30 ms), maximum displacement (7.6 mm versus 7.5 mm), and residual
considered in the experimental studies by Zhang et al. [23] and Pham displacement (1.8 mm versus 1.5 mm). Moreover, the column damage
et al. [24]. The column is buried under the ground level with a depth of including flexural cracks at the vicinity of the impact point, diagonal
0.5 m (see Fig. 1a). The superstructures are designed to sit on the top of cracks at the column base and damage to the column at the top is also
the cap beam and concrete abutments, modelled with a surface to reasonably predicted by the FE model.
surface contact with a friction coefficient of 0.6 [25]. In this study, the Secondly, a numerical model of the full-scale RCBC column which
concrete is simulated by hexahedral elements with one integration was collided by the heavy truck trailer (30 ton) with the speed of
point while the material named *Mat_072RL3 is employed to model the 80 km/h [1] (Accident number 18) is modelled to further validate the
dynamic behaviours of the concrete with uniaxial compressive strength reliability of the numerical simulation associated with full-scale struc-
of 34 MPa. In addition, the dynamic increment factor (DIF) for concrete tures. The diameter of the impacted column was 800 mm. Since the
strength which was suggested by Hao and Hao [26] is selected in the column was accidentally impacted, no column displacement and impact
simulation to quantify the strength increment of the concrete under force time histories were documented in the collision event. Hence,
dynamic loads. Longitudinal and transverse reinforcements which have only the column damage and failure mode are used to verify the nu-
a nominal yield strength of 500 MPa, are modelled by 3-node beam- merical simulation (see Fig. 4). The figure illustrates that concrete
elements (Hughes-Liu with cross section integration). An elastic-plastic crushing at the local contact area, the diagonal shear failure from the
material model (Mat_024) is adopted to model the behaviour of these contact point to the column base, and flexural-shear cracks in the im-
reinforcements while the DIF for steel reinforcements proposed by pacted side of the column observed in the real accident can be re-
Malvar and Crawford [27] is chosen. The superstructures, concrete produced by the simulation. It also should be highlighted that by using
abutments, and column footing are simulated by hexahedral elements the above material models, contact algorithm, and the FE mesh, the
with the elastic material model (Mat_001) being used. failure of the RCBC under the real vehicle collisions [1], including
Two different vehicle models, i.e. a medium truck model named flexural cracks (Accident number 2), a punching shear failure (Accident
Ford truck single unit (8 ton) and a heavy truck model (30 ton), as number 7), and a diagonal shear in the vicinity of the column top
presented in Fig. 2, are employed in this study to collide on the RCBCs. (Accident number 4) and at the impact area (Accident number 10) were
These vehicle models have been widely used in the open literature all well reproduced by Do et al. [12].
[5,6,9,28] to examine the dynamic responses and failures of structures The comparisons indicate that the above modelling technique can
under vehicle collisions. The initial conditions of these vehicle models be able to simulate the contact force, displacement, concrete damage,
considered in this study are presented in Table 2. The contact keyword and failure modes of RCBCs with reliable predictions of both small-scale
namely Automatic_Surface_to_Surface is used to simulate the vehicle – structures and full-scale columns under wide ranges of impact velocities
column interaction. It should be noted that this contact algorithm al- and masses.
lows simulating the impact force time histories between two impacting
parts in collision events [21].
215
T.V. Do, et al. Structures 22 (2019) 213–229
3. Impact force profile model and classification of column impact duration and amplitude of each impact force stage of these two
response different vehicle models can be estimated with respect to the impact
conditions, i.e., vehicle model, vehicle mass, engine mass, vehicle ve-
As observed from many previous studies [5–7,20,28–31], the impact locity, and column properties, i.e. cross-section dimension and concrete
force time histories of vehicle impact events generally include four strength, as given in Table 3. For an example, the impact force time
continuous stages, i.e. vehicle's bumper impact, engine impact, vehicle histories from the proposed impact force model, AASHTO [8], CEN
rail impact, and vehicle cargo impact. In these four impact stages, the [19], and the numerical simulation are compared in Fig. 5b. In this
engine and the cargo parts of the vehicle model usually generate two condition, the vehicle mass, engine mass, and impact velocity are 11
highest values of the impact force while the other two stages show a ton, 0.64 ton, and 100 km/h, respectively (see Case C20 – Table 2). As
moderate impact force with relatively long duration, as illustrated in can be seen in the figure that the proposed impact profile model can
Fig. 5a. In the present study, the time histories of the impact force on accurately predict the time histories of the impact force as compared to
the RCBC from the medium truck and heavy truck collisions derived by the numerical results while the AASHTO [8] and CEN [19] under-
Do et al. [14] are employed to design the RCBC. It should be noted that estimate the PIF force. It is, therefore, necessary to mention that the use
the time histories of the impact force proposed by these models show a of the equivalent static force from design standards, i.e. AASHTO [8]
good agreement with the results from the numerical simulations. The and CEN [19], may lead to an un-conservative prediction of the impact
Table 2
Initial conditions of the vehicle model and numerical results.
Case Column section Vehicle condition PIF Response Shear force Bending moment
(kN) (kN) (kNm)
Note: V is the vehicle velocities; m is the total mass of the vehicle; and me is the engine mass.
216
T.V. Do, et al. Structures 22 (2019) 213–229
30 12.5
Experiment Experiment
25 Simulation Simulation
10.0
Displacement (mm)
Impact force (kN)
20 7.5
15 5.0
10 2.5
5 0.0
0 -2.5
0 15 30 45 60 75 90 0 100 200 300 400 500 600
Time (ms) Time (ms)
(a) Impact force time histories (b) Lateral displacement at the column mid-height
force from vehicle collisions. failure in the negative side at the column mid-height (the red curve).
Furthermore, according to the impact force profile (i.e. PIF, dura- Therefore to design the RCBC under vehicle impacts, the dynamic shear
tion, and impulse) and the column dynamic shear capacity, Pdynmax, the force and bending moment along the column and at the column ends
response of the RCBC against vehicle collisions can be categorised into need to be properly estimated, which cannot be achieved based on ESF.
flexural governed and shear governed responses [14] (see Fig. 6). The In the following sections, the column displacement and the maximum
PIF is the highest peak impact forces related to either the engine impact value of internal forces, i.e. shear force and bending moment at critical
or cargo impact. Based on the concrete strength and the cross-section sections when its response is governed by the flexural response are
dimension of the column, the column dynamic shear capacity can be estimated. It should be noted that when the failure is governed by the
estimated as [14] shear response mode, these values are not needed since the column is
directly damaged by the shear failure at the base, i.e. diagonal shear
max 6.5 W D ft and punching shear failure, because the PIF is greater than the column
Pdyn = (kN )
1, 000 (1) dynamic shear capacity.
where W and D are the column width and depth, respectively (mm),
and ft is the concrete tensile strength (MPa). 4. Internal forces and column responses of RCBC
The previous study [14] reveals that when the PIF is smaller than
0.5Pdynmax, the column response and failure are governed by the flex- 4.1. Maximum shear force
ural mode in which the maximum bending moment at the column ends
appear on the impacted side (negative side) while the remaining sec- 4.1.1. Shear force at the column base
tions of the column have the maximum bending moment on the rear Without loss of generality, taking Case 20 (C20 in Table 2) as an
side (positive side), see Fig. 6a – the red curve. On the other hand, when example here, from the numerical results, the typical time histories of
the column response and failure are governed by the shear mode the impact force and shear force at the column base are shown in
(PIF ≥ 0.5Pdynmax), see Fig. 6b, the maximum bending moment of the Fig. 7a. It can be seen that the maximum value of the shear force is
RCBC at the intermediate section shifts from the rear side of the column smaller than the PIF from the collision, and shear force oscillates
to the impacted side which may cause the flexural-shear cracks and quickly as compared to the impact force. This is because of the
217
T.V. Do, et al. Structures 22 (2019) 213–229
Fig. 4. Numerical model verification of the full-scale bridge column in a real vehicle collision accident.
influence of the inertia force [12,20,32]. At the PIF, the loading acting From Eqs. (2) and (3), the maximum reaction force at the column
on RCBC includes impact force, inertia force, and reaction force, as il- base corresponding to the PIF can be written as follows:
lustrated in Fig. 7b. Therefore, at the PIF, the equilibrium equation of
H1
the horizontal force applied to the column can be expressed as
Rbase = PIF − ∫ mh a (h) dh = PIF (1 − tan α ) + 2, 000
H1 0 (5)
PIF = Rbase + ∫ mh a (h) dh
0 (2) It was previously observed that the maximum shear force at the base
where Rbase is the maximum shear force at the base of the column; of the column is almost unchanged when the PIF causes the shear
H1 failure at the impact area [12,20] because the shear force has reached
∫ mh a (h) dh is the total inertia force distributed in a portion of the the column dynamic shear capacity. Therefore, when the PIF is larger
0
column, H1 (see Fig. 7b); mh is the mass density per unit length of the than 0.5Pdynmax which results in the diagonal shear failure from the
column; and h is the distance measured from the column base. It should contact point to the column – footing connection, the shear force at the
be noted that this equation is valid because during this stage of vehicle column base can be predicted by the following equation:
impact, i.e., usually engine impact, the top part of the column is not max
Rbase = 0.5 Pdyn (1 − tan α ) + 2, 000 (6)
activated yet to resist the impact force as observed in numerical si-
mulations. The impact force is balanced by the base shear and the in-
ertia resistance from the part of the column that has been activated to
resist the vehicle impact.
4.1.2. Shear force at the column top
From the numerical results, the relation between the PIF and the
As presented in Fig. 9a, after the shear force at the base of the
total inertia force distributed on the column is presented in Fig. 8a. It
column increases to the highest value within about 5–10 ms, the shear
should be mentioned that the total inertia force in the figure is defined
force at the top of the column also rises to its peak due to the stress
by subtracting the maximum shear force at the base from the PIF given
wave propagation from the impact area to the column top. The shear
in Table 2. As can be seen in the figure, under the similar PIF, the
force value at that section then oscillates around the zero level. The
column with larger dimension has a smaller contribution of the inertia
shear force at the top of the column in some cases is also considerably
force because of the larger contribution of the column resistance re-
large and may cause damage as observed in previous studies [12].
presented by Rbase in Eq. (2). From the numerical results (see Fig. 8a),
Fig. 9b shows the relation of the maximum shear force at the base and
the best-fitted relation for estimating the total inertia force along the
the top of the column. The figure illustrates that the maximum shear
column at the PIF is
force at the column top, Rtop, can be estimated by the following best-
H1
fitted equation:
∫ mh a (h) dh = PIF × tan α − 2, 000 ≥ 0
0 (3) 1 1
Rtop = Rbase = [PIF (1 − tan α ) + 2, 000]
2 2 (7)
α
= {28.50 (D < 1, 100 mm)450 − 0.015 D (1, 100 From Eqs. (5) and (7), and the PIF (see Table 3), the predicted shear
force at the two ends of the columns are compared to the numerical
mm ≤ D < 3, 000 mm) (4)
results in Fig. 10. The figure illustrates that the empirical equations can
where α is the slope coefficient which represents the effects of the reliably predict the maximum shear force at the base and the top of the
column stiffness, as presented in Fig. 8b. column as compared to the numerical simulation results.
218
T.V. Do, et al. Structures 22 (2019) 213–229
(b) Comparison of the proposed impact force model with those from design codes (Case C20)
Fig. 5. The impact force profile of a vehicle model collided to RCBCs.
Table 3
Impact force profile of the medium and heavy truck model (16.7 m/s < V < 40 m/s) (Do et al. [14]).
Impact phase Medium truck (m ≤ 12 ton) Heavy truck trailer (m > 12 ton)
4.2. Maximum bending moment highest value [20] and the bending moment diagram is illustrated in
Fig. 11a. After about 1–2 ms, the bending moment at the base of the
Many previous studies observed that at the peak value of impact column then increased to its peak as shown in the figure. After that, the
force, just a portion of structures responds to the impact force [32–34] bending moment distributed in the entire column and varied sig-
and it causes the largest flexural bending moment at the local impact nificantly with time due to the column vibration and the effect of the
area. A similar observation was also seen in the RCBC under vehicle inertia force. The envelope of the column bending moment in an impact
collisions when the impact force caused by the engine impact increased event is illustrated in Fig. 11b. As illustrated in the figure, the negative
to the PIF, the bending moment at the impact area increased to the bending moment at the column base is the highest value in the
219
T.V. Do, et al. Structures 22 (2019) 213–229
Fig. 6. Classifications and the corresponding maximum bending moment and shear force distributions of RCBC under vehicle impact.
impacted side while the positive bending moment at the top portion of and C2000, respectively. From the numerical results, the envelope
the column is almost similar to that at the impact point (see Fig. 11b). It curve of the bending moment is about 0.6375PIF when the flexural
should be noted that when the flexural response governs the column crack does not happen. From Eq. (8), the coefficient, kI, is
0.6375 0.6375
response, the shape of the bending moment envelope is consistent re- Le / 4
= 3 / 4 = 0.85. The maximum positive bending moment at the
gardless of the different loading scenarios as observed in the previous impact point is, therefore, expressed as
study [20]. Therefore, the maximum positive and negative bending
max Le
moments at these two sections, i.e. column base and impact point are MIP = 0.85 PIF (kNm)
4 (9)
used for the design of the column.
Since the bending moment diagram in the column at the instant of To estimate the maximum bending moment at the column base, the
the PIF is a triangle (see Fig. 11a), it is reasonable to assume the impact force and inertia forces are required. As observed in the ex-
boundary conditions of the column as simply supported. Thus, the perimental and numerical studies on concrete structures subjected to
maximum positive bending moment at the impact point, MIPmax, can be impact loading, the inertia forces oscillate with high frequencies
estimated by: [32,34] so that it is difficult to estimate the maximum bending moment
at the column base from an analytical solution. Thus, the maximum
max Le bending moment at the base of the column is predicted based on the
MIP = kI PIF (kNm)
4 (8)
numerical results, as given in Table 2. Fig. 12b shows the bending
moment at the base of the column with respect to the PIF. From the
where kI is the coefficient representing the effects of the inertia force on
numerical results, the maximum bending moment at the column base,
the bending moment; Le is the effective length of the column at the PIF.
Mbasemax, can be predicted by the following equation:
In this study, the vehicle models impact on the RCBC at about 1.5 m
above the footing, thus the effective length of the column at the PIF is max
Mbase = 0.45 PIF + 800 (kNm) (10)
assumed as 3 m. The relation between the maximum bending moment
at the impact point (MIPmax) and the PIF from the numerical simulation
is presented in Fig. 12a. It should be noted that in the numerical si- 4.3. SDOF model
mulation when the bending moment at the impact point reaches the
bending moment capacity, [M], the maximum bending moment then In the design of structures subjected to dynamic loads, the SDOF is
keeps constant, although the PIF continues increasing (see Fig. 12a). In commonly used to predict the dynamic response of structures [35–38].
Fig. 12, [MC800], [MC1000], [MC1200], [MC1500], and [MC2000] are the For instance, the SDOF was employed to predict the impact response of
bending moment capacities of the column C800, C1000, C1200, C1500, RC beams [38]. In this analytical method, the elastic stiffness, plastic
10,000
Shear force at the base Lumped Mass
8,000 Impact force
6,000
Force (kN)
4,000
2,000
PIF
-4,000
0 50 100 150 200 250
Reaction force - Rbase
Time (ms)
(a) (b)
Fig. 7. (a) Impact force and shear force time histories (C20); (b) simplified illustration of the column free-body diagram at the PIF.
220
T.V. Do, et al. Structures 22 (2019) 213–229
Fig. 8. (a) Total inertia force – PIF relation; (b) relation between the column dimension and α.
stiffness, crack section, and residual displacement of the beam can be 24,000
taken into account. The analytical result shows a good agreement with Rbase
the numerical simulation and experimental test in terms of the global o
Rtop 45 benchmark
response of the beam under impact loads. Furthermore, Sha and Hao
18,000
Predicted (kN)
[37] used a SDOF system to predict the response of bridge piers under
barge impacts. The bridge pier was assumed as a nonlinear SDOF
system in which both elastic and plastic response of the bridge piers
were considered. This analytical method can give a reasonable pre- 12,000
diction of the maximum lateral displacement of the pier. To estimate
the displacement response of the RCBC under vehicle collisions, the
SDOF approach is also adopted in this study, which is briefly discussed 6,000 54 data points
below. 2
Based on the proposed impact force profile and column properties,
R = 0.964
an equivalent SDOF model of the column under vehicle collision as il-
0
lustrated in Fig. 13 can be derived. The equation of motion of the SDOF
0 6,000 12,000 18,000 24,000
system is written as follows:
Numerical results (kN)
•• •
Me x (t ) + Ce x (t ) + K e x (t ) = Fe (t ) (11) Fig. 10. Comparison of the maximum shear force at the column ends between
the predicted results and numerical results.
where Me, Ce, and Ke are the equivalent mass, damping coefficient, and
column stiffness of the SDOF system, respectively; Fe(t) is the equivalent H
load on the SDOF system; t is time and x is lateral displacement. Me = ∫0 mh ϕ2 (h) dh + madd (12)
The equivalent mass of the lumped-mass system, as given by Biggs
[39], can be expressed as where mh is the mass density per unit length of the column; ϕ(h) is the
assumed deflection shape function with the displacement at the column
4,000 9,000
Rtop (kN)
2,000
6,000
0
3,000 y=0.5x
-2,000 2
(R =0.94)
-4,000 0
0 60 120 180 240 300 0 6,000 12,000 18,000 24,000
Time (ms) Rbase(kN)
(a) (b)
Fig. 9. (a) Typical time histories of the shear force at the column ends (C20); (b) relation between the maximum shear force at the base and the top.
221
T.V. Do, et al. Structures 22 (2019) 213–229
max
PIF MIP
max
Mbase
At the PIF 1 - 2 ms post PIF Flexural response
(a) (b)
Fig. 11. Simplified column bending moment: (a) at the PIF; (b) envelop bending moment.
top normalized to unit as shown in Fig. 14; madd is the added mass at the density, added weight, moment of inertia, Young's modulus, and
column top; h is the distance measured from the column base; and H is column height [39–41]. To achieve the equivalent mass and con-
the column height. centrated force in the lumped-mass system, the deflection shape func-
Because the displacement at the column top is considered, the tion was usually assumed by a simple linear function as discussed in
equivalent column stiffness is the actual stiffness of the column. By previous studies [37,39]. However, the linear function does not really
assuming a free top end as shown in Fig. 13a, the equivalent stiffness is reflect the actual shape of the structural response and thus does not
provide good predictions [37]. By varying the column height, cross-
3EI
Ke = section, and the added weight at the column top, the dimensionless
(H + Hadd )3 (13)
mode shapes of the column can be derived as presented in Fig. 14. It
where E = 4700 fc′ is the Young's modulus and fc' the concrete com- should be noted that these deflection shape functions are extracted from
pressive strength; I is the moment of inertia of the column; Hadd is the the modal analysis [42] in which the boundary condition of the column
distance from the top of the column to the centroid point of the added is fixed at the column base and free at the column top. From these
weight. curves, the best fitted normalized deflection shape function of the
In addition, the equivalent impact force on the SDOF system is es- column is
timated by:
h 2 h
n ϕ (h) = 0.6 ⎛ ⎞ + 0.4 ⎛ ⎞
⎝H⎠ ⎝H⎠ (15)
Fe (t ) = ∑ Fj (t ) ϕj
j=1 (14)
To solve the equation of motion of the SDOF system, the central
where Fj is the impact force at a location j; ϕj is the value of the de- difference algorithm is adopted. The velocity and the acceleration of the
flection shape function at the location j [39]. SDOF system can be approximated by
The above equivalent mass and concentrated impact force highly • 1
depend on the deflection shape function of the column which is de- x= (x t + Δt − x t − Δt )
2Δt (16)
termined from the column properties, i.e. cross-section area, mass
12,000
14,000 [MC1500] y=0.45x+800
2
(R = 0.95)
8,000
m ax
m ax
[MC1200]
7,000 [MC1200]
4,000 [MC1000]
[MC1000]
[MC800]
[MC800]
0 0
0 8,000 16,000 24,000 32,000 0 8,000 16,000 24,000 32,000
PIF(kN) PIF(kN)
Fig. 12. Maximum bending moment: (a) at the impact point; (b) at the column base.
222
T.V. Do, et al. Structures 22 (2019) 213–229
Mass: madd
Column stiffness: K c Me Ke
Damping coefficient: C F e(t)
Impact force
F(t) Ce
Fig. 13. (a) Simplified model of the column under impact force; (b) equivalent SDOF system.
1.0 by the medium truck model (8 ton) with the velocity of 80 km/h. The
heights of the column and the cap beam are 9600 mm and 1500 mm,
Normalized column length
respectively. The added mass of 800 ton is placed on top of the cap
0.8 beam with Hadd equals to 2100 mm. Since the displacement at the top of
the cap beam is of interest and the impact location is 1.5 m above the
0.6 footing, the value of the deflection shape function at the impact point is
0.065 (Eq. (15)). The equivalent mass, damping ratio, and column
stiffness of the SDOF system are 800 ton, 3.5%, and 8870 kN/m, re-
0.4 C1200-L=9.6m, Madd=800 ton
spectively. The figure illustrates that the analytical model is able to
C1200-L=9.6m, Madd=0 ton estimate the maximum lateral displacement and the dynamic response
C1200-L=4.8m, Madd=800 ton of the column under vehicle collisions with a reliable prediction. It is
0.2 C2000-L=9.6m, Madd=2720 ton worth mentioning that a slight difference in the lateral displacement of
Simplified equation the column, as shown in Fig. 15, is caused by the local deformation of
0.0 the column at the contact area between the vehicle model and the
0.0 0.2 0.4 0.6 0.8 1.0 column which cannot be predicted by the SDOF model. It should be
Normalized shape function value noted that the column response predicted by using a linear assumption
of shape function [37] is also presented in Fig. 15. The result shows that
Fig. 14. The dimensionless mode shape function of the first mode of the using the linear assumption of shape function over-estimates the re-
column.
sponse of the column under vehicle collision.
•• 1
x = (x t + Δt − 2x t + x t − Δt )
Δt 2 (17) 5. Design example
The above equations can be solved with initial conditions of the
• ••
column at the start time t = 0: xt−Δt = 0, x = 0 , and x = 0 . The first 5.1. Design procedure
step of the solution starts to determine xt+Δt with the corresponding
impact force obtained from the proposed impact force profile given in Based on the column properties i.e. cross-section width, W, cross-
Section 3 and Table 3. section depth, D, compressive strength of concrete, fc (MPa), tensile
To verify the analytical method, the comparisons of the analytical strength, ft (MPa), and reinforcement area As (mm2), yield stress fy
result and simulation result are shown in Fig. 15. In this case, the (MPa), and vehicle parameters, i.e. total vehicle mass, m (ton), vehicle
column which has a square section of 1200 × 1200 mm2, is impacted velocity, V (m/s), and engine mass, me (ton), the dynamic shear capa-
city, Pdynmax (kN), shear force Rbase (kN) and Rtop (kN), bending moment
MIP (kNm) and Mbase (kNm), and the maximum column displacement,
12 Proposed analytical model Δmax, can be estimated by the above equations, as summarized in
Numerical model Fig. 16. [V], [M], and [Δ] are the shear capacity, bending moment
8 Linear assumption of shape function capacity, and the maximum allowable lateral displacement of the
Displacement (mm)
223
T.V. Do, et al. Structures 22 (2019) 213–229
START
max
No Pdyn
Shear force (Eqs. 6 and 7) PIF <
Shear failure 2
No R < [V ]
M< M
]
Yes
FINISH
224
T.V. Do, et al. Structures 22 (2019) 213–229
(Original column)
SR 1 SR 2 SR 3
(All columns survive)
Fig. 17. Crack patterns of the RCBCs under medium truck collision.
max
columns survive the impacts from the truck collisions without diagonal Mbase = 0.45 × 7, 886 + 800 = 4, 349 (kNm) (23)
shear cracks at the base. However, as shown in the figure, although the
From the above-calculated shear force and bending moment, de-
columns survive the diagonal shear failure, some intensive flexural
signs of transverse and longitudinal reinforcements of these three col-
cracks are observed (see Fig. 17), indicating the column might experi-
umns are derived and given in Table. 4. The numerical results of these
ence flexural damage. Therefore designs of longitudinal and transverse
three columns with the new reinforcements are re-simulated. The
reinforcements of these three columns are thus required to avoid flex-
ural damage. From the above-proposed equations, the maximum shear
Table 4
force at the base and the top of the column are:
Design of reinforcements.
Rbase = 7, 886 × (1 − tan 28.5o) + 2, 000 = 5, 604 (kN ) (20) Column Transverse reinforcements Longitudinal
reinforcements
1 1 At the base At the top
Rtop = Rbase = × 5, 604 = 2, 802(kN )
2 2 (21)
SR1 Four-leg 16-mm- Two-leg 14-mm- 28d36
Furthermore, from Eqs. (9) and (10), the maximum bending mo- diameter @100 mm diameter @100 mm
SR2 Three-leg 20-mm- Two-leg 14-mm- 32d40
ment at the base and the impact point are:
diameter @100 mm diameter @100 mm
3 SR3 Four-leg 16-mm- Two-leg 14-mm- 28d36
max
MIP = 0.85 × × 7, 886 = 5, 027(kNm) diameter @100 mm diameter @100 mm
4 (22)
225
T.V. Do, et al. Structures 22 (2019) 213–229
plastic strain of these columns after design modification is shown in than a half of the dynamic shear capacity of the column
Fig. 18. The simulation results show that the flexural cracks in the (0.5Pdynmax = 27, 463 (kN)), the column thus survives the direct vehicle
columns are greatly reduced. This example demonstrates that the pro- impact. Therefore, only the flexural capacity is checked.
posed approach can give reliable predictions of the capacity of RC From Eqs. (5) and (7), the maximum shear force at the column base
columns to resist vehicle impact and provide feasible solutions to im- and column top are:
prove the column design.
Rbase = 20, 635 × (1 − tan(45 − 0.015 × 1, 300)) + 2, 000 = 12, 793 (kN )
226
T.V. Do, et al. Structures 22 (2019) 213–229
7,000 kN
1300
1300
10,000
5,000
0 1300 1300
0 80 160 240 320
Time (ms) At the column base The remaining part
Displacement (mm)
10
Axial force (kN)
40,000
0
20,000 -5
0 -10
0 5,000 10,000 15,000 20,000 0 500 1000 1500 2000 2500
Bending moment (kNm) Time (ms)
(c) Column interaction diagram (d) Displacement time histories at the column top
at time when lateral impact force reaches the maximum, i.e. PIF, the sufficient to resist the vehicle impact. To examine the reliability of the
axial compression force in the column is increased due to the stress above design, the numerical model of the designed column is then built
wave propagation from the contact area to the top and the base of the and impacted by the heavy truck trailer. The bending moment and axial
column [12]. Therefore, the maximum dynamic axial force, Adyn, in the force from the numerical simulation are also compared to the analytical
column at the moment of PIF is [12]: solutions (see Fig. 19c). The lateral displacement time histories at the
column top from SDOF model and numerical simulation is also pre-
0.2 × 13002 × 50
Adyn = + 8 × 10−6PIF + 0.32PIF = 26, 910 (kN ) sented in Fig. 19d in which the maximum lateral displacement and the
1, 000
natural period of the column are well predicted by the analytical so-
(28) lution. Meanwhile, the plastic strain of the designed column is also
The interaction diagram of the column, as followed by ACI [25], checked by using numerical simulation where no shear failure or flex-
together with the internal forces caused by the heavy truck trailer ural failure occurs, as shown in Fig. 19e. The results show that the
collision at different instants with time step 0.5 ms during the impact analytical method can provide a useful tool and feasible application to
are presented in Fig. 19c. It shows the column flexural capacity is design the column under vehicle collision with good predictions as
227
T.V. Do, et al. Structures 22 (2019) 213–229
228
T.V. Do, et al. Structures 22 (2019) 213–229
[35] Hao H, Wu C. Effects of simultaneous ground shock and airblast force on structural [39] Biggs JM. Introduction to structural dynamics. New York: McGraw-Hill; 1964.
response. 5th Asia-Pacific conference on shock and impact loads on structures. [40] Dutta R, Ganguli R, Mani V. Swarm intelligence algorithms for integrated optimi-
2003. Hunan, China. zation of piezoelectric actuator and sensor placement and feedback gains. Smart
[36] Ngo T, Mendis P, Gupta A, Ramsay J. Blast loading and blast effects on struc- Mater Struct 2011;20:105018.
tures–an overview. Electron J Struct Eng 2007;7:76–91. [41] Jou JM. Theory and simulation analysis of the mode shape and normal shape ac-
[37] Sha Y, Hao H. A simplified approach for predicting bridge pier responses subjected tuators and sensors. Open Journal of Acoustics 2014;4:184–203.
to barge impact loading. Adv Struct Eng 2014;17:11–23. [42] Bathe K-J. Finite element procedures: Klaus-Jurgen Bathe. 2006.
[38] Pham TM, Hao H. Influence of global stiffness and equivalent model on prediction [43] Pham TM, Hao H. Effect of the plastic hinge and boundary conditions on the impact
of impact response of RC beams. Int J Impact Eng 2018;113:88–97. behavior of reinforced concrete beams. Int J Impact Eng 2017;102:74–85.
229