International Journal of Impact Engineering 36 (2009) 965–974

Contents lists available at ScienceDirect

International Journal of Impact Engineering

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

Blast response simulation of an elastic structure: Evaluation of the

fluid–structure interaction effect
Kolluru V. Subramaniam a, *, Weimin Nian a, Yiannis Andreopoulos b
a
Civil Engineering Department, T-110 Steinman Hall, City College of the City University of New York, Convent Avenue at 140th Street, New York, NY 10031, USA
b
Mechanical Engineering Department, City College of the City University of New York, New York, NY 10031, USA

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

Article history: Blast pressure wave interaction with an elastic structure is investigated using a numerical analysis
Received 13 March 2008 approach, which considers fluid–structure interaction (FSI) within an Arbitrary Lagrange Euler (ALE)
Received in revised form framework. Approximate numerical procedures for solving the Riemann problem associated with the
6 January 2009
shock are implemented within the Godunov finite volume scheme for the fluid domain. The structural
Accepted 7 January 2009
Available online 19 January 2009
displacement predicted by ignoring FSI is larger than the corresponding displacement considering FSI.
The influence of the structural and blast pressure wave parameters on the importance of FSI is studied
using an analysis of variables. Two non-dimensional parameters corresponding to the ratios of blast
Keywords:
Blast duration to the time period of the structure and the velocity of the structure to the particle velocity of the
ALE incident blast pressure wave are identified. It is shown that for a given blast pressure wave, the error in
Transient the maximum displacement predicted by ignoring FSI effect during structural motion is directly
Dynamic proportional to the ratio of the structure velocity to the particle velocity of the incident blast pressure
Fluid–structure interaction wave. There is a continuous exchange of energy between the structure and air during the structural
motion, which is significant when the structural velocity is significant compared to the particle velocity
of incident blast pressure wave. FSI effect become insignificant when the ratio of velocities starts
approaching zero.
Ó 2009 Elsevier Ltd. All rights reserved.

1. Introduction material models at high strain rates. Several examples of this

In typical air blast analysis, the blast over-pressure, which is
determined as the reflected pressure of a fixed rigid surface is
imposed on the structure as an external pressure loading with
a known evolution in time [1]. The actual pressure variation is
further simplified using triangular or exponential functions [1–5].
During the analysis, it is usually assumed that there is no coupling
between the fluid and the solid and the response of the solid can be
obtained independently of the fluid by directly prescribing a known
pressure on to the structural element. The results presented in the
literature broadly differ in the level of sophistication adopted for
representing the structural element. A popular approach comprises
of transient dynamic analysis of the structural element to
a prescribed pressure time history using a numerical scheme such
as the finite element method. The response of the structure is
accurately represented using the appropriate elements and
* Corresponding author. Tel.: þ1 212 560 6569; fax: þ1 212 650 6965.
E-mail address: ksubram@ce.ccny.cuny.edu (K.V. Subramaniam).
approach for studying the blast response of structural elements
such as slabs, walls and beams can be found in the literature [3–5].
The second approach recommended by the U.S. Department of
Army in TM5-1300 [1], which is more practice-oriented, is based on
substituting the structural element with an equivalent single
degree-of-freedom (DOF) spring-mass system. The procedures for
obtaining the equivalent stiffness and mass for the single DOF
spring are reviewed in detail by Morison [6]. This rather simplistic
approach is often preferred because it provides a satisfactory basis
for studying the dynamic response of a structural element sub-
jected to transient loading of short duration [7].
In recent years there has been considerable interest in evalu-
ating the role of fluid–structure interaction (FSI) in the blast
response of a structure. Fluid–structure interaction occurs when
the blast pressure wave causes deformation of the structure and,
thus alters the flow of the fluid. In an early attempt, Taylor (1963)
studied the momentum transferred to a free-standing plate from
a pressure wave with an exponential profile [8]. It was shown that
changes in the reflected over-pressure are associated with move-
ment of the boundary which results in a decrease in the

0734-743X/$ – see front matter Ó 2009 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ijimpeng.2009.01.001
966 K.V. Subramaniam et al. / International Journal of Impact Engineering 36 (2009) 965–974
transmitted impulse from the pressure wave to the free-standing
plate. The use of compliant structures in mitigating the influence of
underwater shock loading has been explored for developing
sandwich-plate structure for improved blast performance [9,10].
The role of FSI, evaluated using Taylor’s approach, was shown to be
significant in decreasing the transmitted impulse in such struc-
tures. Taylor’s approach, however, involves simplistic assumptions,
such as, no change in the fluid density and linear superposition of
waves, which are acceptable for studying weak shock waves. Such
an approach is not suitable for studying the interaction of a strong
shock wave in air blasts which involves non-linear, finite amplitude
disturbances propagating in a compressible medium [11]. Recently,
from a numerical evaluation of the response of a free-standing plate
subjected to air blast it has been shown that the FSI effect in
reducing blast impulse is significant only at large displacements,
which limits its practical use [12]. These results however cannot be
directly applied to the case of air blast response of structures, where
the resistance to motion is derived from both inertial and stiffness
effects. A careful evaluation of the role of the structural parameters
on determining the influence of FSI in the dynamic response of
a structure is required.
A blast pressure wave is associated with a leading shock front,
which brings an abrupt finite pressure change, followed by
decaying pressure. A simplistic understanding of the role of FSI
when a blast pressure wave in a compressible fluid medium is
incident upon the surface of a moving boundary can be obtained
from the classical work of Courant and Friedrichs on shock waves
[13]. In a moving shock wave, the leading shock front is followed by
constant pressure. If the shock wave is incident on a fixed rigid
boundary, it is assumed a pressure Pro is developed on the surface of
the boundary immediately following the reflection of the leading
shock front. If the boundary starts to move with a velocity, Vb, the
motion of the boundary would alter the pressure at its surface.
The relation between the resultant pressure amplitude at the face
0 and the velocity of the boundary, V is
of the moving boundary, Pro b
related by the simple wave relation, given as
0   2g
Pro g  1 Vb g1
¼ 1þ
Pro 2 ar
where ar is the local sound velocity after the reflected shock, which,
can be shown to monotonically increase with the intensity of the
shock wave and g is the ratio of heat capacities of air [13, p. 95].
When the motion of the boundary is away from the direction of
reflected shock front, for Vb less than ar, there will be a decrease in
the resultant pressure at the face of the moving boundary, when
compared with Pro .
From the simplistic analysis it appears that the pressure on the
surface of a moving boundary is dependent on its velocity. If the
blast pressure wave is incident on an elastic structure, the velocity
of the structure would depend upon the applied pressure loading
history and structural parameters such as stiffness and mass. This
therefore presents a coupled fluid–structure problem, where the
movement of the structure influences the applied pressure, which
in turn effects its movement. It is clear that when a blast pressure
wave is incident on an elastic structure, the role of FSI depends
upon both the blast wave and structural parameters. Conventional
analysis ignores this interaction and the reflected pressure
obtained from a fixed rigid wall, which assumes that the boundary
does not move during the time associated with the passage of the
blast wave, is directly prescribed on the structure.
There are some important questions which remain unanswered,
such as: (1) when is the FSI effect in air blast incidence on a struc-
ture important? (2) what is the error in ignoring the air blast–
structure interaction in terms of the predicted stresses and
amplitude of displacements in the structural element? and (3)
what are the structural and blast parameters which help to decide
when the role of FSI is important? Answers to these questions are
investigated using a single DOF spring-mass system within
a framework which considers the change in the fluid domain due to
the deformation of the solid and the variation in the blast over-
pressure due to movement of the solid–fluid boundary. A numerical
analysis framework for predicting the response of a 1-DOF
structure, which considers solid–fluid interaction within an Arbi-
trary Lagrange Euler (ALE) framework, was developed. An accurate
Riemann Solver with appropriate spatial and temporal discretiza-
tion was implemented for capturing the singularity associated with
the shock. Numerical simulations were performed to study the
interaction between a blast pressure wave and an elastic structure.
2. Objectives
The objectives of this paper are:
1. To develop a numerical framework for studying the response of
an elastic structure subjected to an incident blast pressure
wave considering FSI.
2. To determine the relationship between the structure and blast
parameters for which the influence of FSI is important.
3. To develop a procedure to estimate the difference between the
structural response parameters when FSI is not considered in
the analysis.
3. Problem statement
The FSI was studied using a 1-DOF structure placed in the
path of an incoming blast pressure wave generated using a shock-
tube arrangement as shown in Fig. 1. A spring-mass system was
used to represent the elastic structure. The mass is idealized as
a piston in the path of an incoming shock front (Fig. 1a). The
piston is assumed to fit perfectly inside the tube, which is fric-
tionless. In this one-dimensional idealization, the shock front is
planar and there is no flux of air perpendicular to the direction of
(1)
blast wave propagation. The analysis is performed considering
a
reflecting boundary
Non-transmissive
K
Shock
X
us(t)
Frictionless tube piston
b
Pressure
P2 , ρ 2 p1 = 1atm
X
c Expansion
Contact
waves
discontinuity
Time
Leading shock
X
Fig. 1. The configuration used for evaluating 1-D response of the structure subjected to
blast pressure wave: (a) initial conditions for creating the blast pressure wave; (b) X–t
diagram showing the formation of the blast pressure wave; and (c) the spring-mass
system in a shock-tube arrangement.
 K.V. Subramaniam et al. / International Journal of Impact Engineering 36 (2009) 965–974
a unit cross-sectional area of the mass and assuming that the
motion of the mass only produces an extension or a compression
of the spring. The blast pressure wave is created using an initial
condition where air at higher pressure and density is separated
from air at ambient conditions (Fig. 1b). A reflective boundary is
placed behind the high pressure air. This corresponds to the
classical shock-tube arrangement used for generating a blast
pressure wave (shown schematically in Fig. 1c) [14,15]. Starting
from the initial pressure discontinuity, a compression pressure
pulse travels along the positive X coordinate, while expansion
waves travel in the negative X direction. Upon reflection at the
left boundary, the expansion waves start traveling in the positive
X direction. The expansion waves catch up with the traveling
compression pulse to produce a blast pressure wave. In this setup
the different parameters of the blast pressure wave such as the
shock pressure amplitude and duration can be varied by changing
the initial conditions, the location of the initial discontinuity, the
location of the reflective back wall and the initial location of the
mass.
The spring-mass system within a shock-tube offers a powerful
platform to evaluate the influence of different variables such as the
structural mass, the structural stiffness, the shock wave speed, blast
pressure duration and shock amplitude on the different aspects of
the response of the structure, such as, the energy transferred from
the shock front, the displacement, velocity and acceleration of the
structure at different time scales.
4. Numerical framework
In this section, the numerical framework for studying the shock
wave interaction with the spring-mass system is described in
detail. In particular, the following topics are discussed: (a) the blast
pressure wave propagation in the fluid; (b) the transient dynamic
analysis of the spring-mass system; and (c) the fluid–solid coupling
algorithm. The shock wave propagation in the air is modeled using
the numerical discretization of the conservation equations stated in
the arbitrarily moving Eulerian framework. The transient dynamic
response of the spring-mass system was obtained from the dis-
cretization of the momentum equation within the Lagrangian
framework. The fluid–solid coupling was achieved by applying the
congruence conditions at the air–mass interface using a staggered
algorithm. While the fundamental aspects of these topics are well
established, a brief description of each is provided for
completeness.
4.1. Numerical framework for fluid domain
In a general moving coordinate system the conservation laws
are described by the Arbitrary Lagrange Euler (ALE) formulation
[16,17]. The one-dimensional integral form of the conservation
equations in the ALE formulation is given as
Z Z *
v * vF
UdUðtÞ þ dUðtÞ ¼ 0
vt vx
UðtÞ UðtÞ
*
where the integral domain UðtÞ is a function of time, U is the
*
conservative state vector, and F is the flux vector, which in ALE
formulation represents the net mass momentum and energy flux
passing through the moving boundary vUðtÞ.
8 9 8 9
< r = < rV =
* *
U ¼ rn and F ¼ rnV þ p
:
rE ; :
rEV þ np ;
The contravariant velocity, V is given as
967
V ¼ n  Vc (4)
where Vc is the prescribed coordinate velocity.
In the numerical implementation, the entire domain occupied
by the fluid was subdivided into a computational grid of N cells,
which were centered around grid/node points. For each cell, the
conservative state vector was defined at the grid point located in
the center of the cell and the flux vector was computed at the
interface between two adjoining cells. In the numerical scheme,
the node/grid points were identified with integers (1  i  N). The
discretized form of the conservation equations obtained by inte-
grating over the finite volumes was used [18,19]
n o
* * * *
Dxnþ1
i U nþ1
i  Dxni U ni ¼ Dt F niþ1=2  F ni1=2 (5)
* *
where Dx is the grid size (the size of the cell) and F ni1=2 and F niþ1=2
are the interfacial numerical fluxes at i/(i  1) and i/(i þ 1) inter-
faces, respectively in the time step n.
The Godunov finite volume methodology for shock propagation,
which has been developed for the Euler-based hyperbolic system,
was directly applied to the ALE-based formulation. An approximate
Riemann solver of Roe was used. The interfacial numerical flux was
calculated using the weighted averaged-flux (WAF) scheme [20,21].
The total variation diminishing (TVD) constraint was applied to
minimize the spurious oscillations produced in the vicinity of high
gradients. In the actual implementation, the SUPERBEE limiter was
used in the WAF scheme.
4.2. Treatment of computational domain
The computational domain for the finite volume calculations is
shown schematically in Fig. 2. The left boundary of the air tube was
treated as a non-transmissive, fixed, reflecting boundary. The right
boundary of the fluid domain always coincided with the position of
the piston, which is a non-transmissive, moving, reflective
boundary. At any given time the entire fluid domain was discretized
into N elements of equal size. The size of elements however varied
during the analysis depending upon the motion of the right
boundary.
4.2.1. Treatment of boundary nodes
To advance the solution of the 1st and Nth nodes/grid points,
numerical fluxes at the boundary were evaluated using a 2nd order
accurate scheme by adding two fictitious grid points on the other
side of each boundary, whose states were reflective symmetric to
the inside grid point/node [20, p. 491]. It should be noted that
while the right and left boundaries represent conditions of zero
mass flux, the momentum and energy fluxes need not be zero
across these boundaries. For the left boundary at x ¼ 0, the condi-
tions in the fictitious cells denoted by i ¼ 1 and i ¼ 0 are reflective
symmetric of the conditions in cells i ¼ 2 and i ¼ 1, respectively. The
states of the fictitious cells denoted by i ¼ N þ 1 and i ¼ N þ 2 at
the right boundary are given as
(2)
(
rnNþ1 ¼ rnN ; nnNþ1 ¼ nnN þ 2u_ ns ; pnNþ1 ¼ pnN
(6)
rnNþ2 ¼ rnN1 ; nnNþ2 ¼ nnN1 þ 2u_ ns ; pnNþ2 ¼ pnN1
where u_ ns is the velocity of the mass or piston ðnnNþ1 ; rnNþ1 ; rnNþ1 Þ
and ðnnNþ2 ; rnNþ2 ; rnNþ2 Þ are the velocity, density and pressure at
nodes N þ 1 and N þ 2.
To evaluate the total force that is applied on the mass at each
time step, the pressure at the right boundary (p*) was calculated
(3)
using the exact solution for the Riemann problem constructed
across the moving boundary. As shown in Fig. 3, there are two
possible cases which need careful consideration.
968 K.V. Subramaniam et al. / International Journal of Impact Engineering 36 (2009) 965–974

pressure
Initial condition at t =0 P2, 2 p1 = 1atm
X

n-1/2 n+1/2 us(t)

Left boundary
1 2 3
X n N-1 N
(non-transmissve, fixed,
reflecting boundary)
Right boundary
Computational cell (non-transmissive moving boundary)

Fig. 2. The computational domain used for finite volume calculations.

Case (a) – when the velocity of air at the Nth node is less than the where ms, us(t) and u€ s ðtÞ are the mass, the displacement and the
velocity of the structure ðnN < u_ s Þ, two rarefactions, one going to acceleration, respectively, and fs(t) is the external force applied on
right the other one going to left with respect to the boundary, are the mass by the air, defined as
formed. The change of state across each series of rarefactions obeys 
* * * *
the simple wave relationship, i.e. the states U L / U *L and U R / U *R , fs ðtÞ ¼ p* ðtÞ  Po A (10)
obey the simple wave relationship. The two states UR* and UL* are
separated by a single contact discontinuity and it is known that only where Po is the ambient pressure, p* is the pressure applied by the
density changes across the contact discontinuity, while there is no air in the shock tube and A is the cross-sectional area of the piston.
change in the velocity and pressure. This is also referred to as the The equation of motion of the mass (piston) was numerically
congruence condition, given as p*L ¼ p*R ¼ p* and n*L ¼ n*R ¼ n* . integrated in time using the Newmark-beta method (with b ¼ ¼
The pressure and velocity p* and v* adjacent to the boundary are and g ¼ ½) where the acceleration is assumed to be constant in
obtained by utilizing the relations across the rarefactions and a time interval and equal to the mean of the beginning and the end
congruence conditions [20, p. 223]. of the interval.
Case (b) – when the velocity of air at the Nth node is greater than  
the velocity of the structure ðnN > u_ s Þ, two shocks, one traveling to 4 4 4
2
ms þ ks unþ1
s ¼ fsnþ1 þ 2 ms uns þ ms u_ ns þ ms u
€ ns (11)
the left and the other to the right of the boundary, are formed. In Dt Dt Dt
this case, the pressure and velocity adjacent to the moving
where fsnþ1 is presumed to be known in the solid solver.
boundary are obtained from the Hugoniot relationship and the
congruence conditions [20, p. 224].
The boundary pressure for both cases is obtained as 4.4. Fluid–structure coupling
8   g2g
>
> 1 C 1
In this paper, the staggered method is used to account for the
>
< p* ¼ pN 1 þ ðg  1Þ N for nN < u_ s
2 aN (7) fluid–structure coupling in time domain [18,19,22,23]. In the
> h i1  staggered method, the fluid and structure solvers are run inde-
>
> CN 2 þ 4A ðB þ p Þ 2
: p* ¼ pN þ 2A C N þ C N N N N for nN > u_ s
N pendently; the equation of motion for the solid and the conser-
where vation laws for the fluid are alternately integrated in time (shown
2 ðg  1Þ schematically in Fig. 4). The interaction between the air and the
AN ¼ ; BN ¼ p ; CN ¼ nN  u_ s (8)
ðg þ 1ÞrN ðg þ 1Þ N solid mass (piston) is accounted for enforcing the congruence
conditions of pressure and velocity at the air–piston interface; the
and the solution of boundary velocity v* is automatically equal to u_ s . interface velocity and pressure are equal for both the fluid and
the structure. Starting at t ¼ tn, when the state of the fluid and the
4.3. Transient dynamic analysis of the elastic structure
structure are known, the fluid–structure system is integrated in
time to obtain the solution at time tnþ1 as:
The motion of the mass is described by the momentum equation
for the single DOF system, given as,
(i) The displacement of the structure at the end of the current
€ s þ ks us ¼ fs ðtÞ
ms u (9) time step (t ¼ tnþ1) is predicted.

a Contact
b Contact
discontinuity
discontinuity
shock t rarefactions t rarefactions
→* →*
shock
UL UR

→
UL
→
UR

x x
Two shocks case Two rarefactions case
Fig. 3. Two possible cases for Boundary Riemann Problem (the dotted line represents the motion of the boundary): (a) the two rarefactions case; and (b) the two-shock case.
 K.V. Subramaniam et al. / International Journal of Impact Engineering 36 (2009) 965–974 969

equal length. The initial pressure discontinuity was created by

Integrate prescribing an initial condition with high pressure in the first 30
Fluid cells. The initial pressure in the remaining 70 cells was equal to the
Int
e
Str grat ambient pressure. In the analysis the CFL number was chosen to be
uct e
ct t
ru dic

ure
e
0.6. Shock waves of different amplitudes were created using
ur
St re
P

different initial pressure conditions. A comparison of the pressure

amplitude of the reflected shock obtained numerically with the
corresponding value obtained from Eq. (13) is shown in Fig. 5. It can
n = n+1 be seen that the pressure amplitude predicted numerically agrees
well with the exact solution thereby verifying the accuracy of the
tn tn+1 numerical scheme for the Riemann Solver at the boundary.
Fig. 4. Schematic representation of the staggered algorithm for fluid–structure
coupling.
6. Numerical results for blast–structure interaction

(ii) The fluid is integrated to tnþ1 using the predicted displacement
of the structure. The interfacial pressure is obtained from the
fluid solver.
(iii) The structure is updated to the next time level (tnþ1) using the
interfacial pressure of step ii.
The crucial part in this algorithm is the prediction of the
structural state in step (i). Within a time–step, if the acceleration is
assumed constant, the structural velocity is a linear function in time
and the displacement can be obtained exactly using the trapezoidal
integration of the velocity. The interface boundary velocity in the
fluid solver is then obtained as
1 n
n*n ¼ u_ s þ u_ nþ1
s (12)
2
5. Verification of analysis procedure
Results of the analysis procedure were verified considering
interaction of a shock wave with a fixed rigid wall. In a shock wave,
the leading shock front is followed by constant pressure. Incident
shock when reflected from a fixed rigid wall creates a reflected
shock traveling in the opposite direction. The amplitude of the
reflected shock immediately following reflection from a fixed rigid
surface (Pro) for a given incident shock amplitude (Pio) can be
calculated as (when g ¼ 1.4 in the case of air and for the range of
blast intensities of interest) [13, p. 153]
8Pio  Po
Pro ¼ pio (13)
Pio þ 6Po
Numerical analyses were performed considering shock waves
with different pressure amplitudes incident on a fixed rigid
boundary. In the shock-tube arrangement the elastic structure was
replaced with a fixed reflective boundary. Shock waves of varying
pressure amplitude were created by varying the pressure in the
initial high pressure region. The total length of the shock tube was
taken equal to 1 m. The entire domain was divided into 100 cells of
The origin of the fixed global coordinate system was coincident
with the left reflecting boundary. The entire fluid domain was
discretized into 100 cells (elements). The number of cells was kept
constant. The size of the cells was varied to account fro the moving
boundary such that the entire fluid domain was divided into 100
cells of equal size at any given instant. The structural mass coin-
cided with the right boundary of the 100th cell. The incident blast
pressure wave was created considering the location of the initial
discontinuity (in pressure and density) at the interface between the
5th and 6th cells. Initial conditions in the first five cells were:
vL ¼ 0 m/s, pL ¼ 15 atm (1.519 MPa) and rL ¼ 18.375 kg/m3. The
initial conditions in the remaining cells were: vR ¼ 0 m/s,
pR ¼ 1.0 atm and rR ¼ 1.225 kg/m3. Extra source term was added to
the conservation equation of energy in the first time step to
simulate the sudden release of energy associated with a detonation.
The source term was added to the cells in the high pressure region
and was a constant equal to 5  107 J. The initial conditions resulted
in an incident blast pressure wave at the initial location of the mass
with the following characteristics: 14.9 atm (pressure amplitude of
the shock front); 931 m/s (particle velocity upstream of the shock);
and 64.5 ms (duration of the blast). Since the exact duration of the
blast pressure wave, based on a precise determination of the value
of the over-pressure decreasing to the ambient value is difficult, the
duration of blast pressure was taken equal to the time required for
the pressure to decrease to 10% of the shock pressure amplitude.
The same procedure has been used in all subsequent calculations
for determining the blast duration.
Results of analysis considering two different values of spring
stiffness are shown in Fig. 6. For the first case (case A), the
equivalent structural stiffness and mass used in the analysis were
obtained considering a steel plate which is 4.5 m in length,
2.25 m wide and 2.5 cm thickness. Fixed-edge and pinned
boundary conditions were considered along the 2.25 m and 4.5 m
edges, respectively. The equivalent structural mass and stiffness
of the fundamental mode of vibration of plate per unit cross-
sectional area were used in the analysis and are equal to
148.2 kg/m2 and 1.397e6 N/m/m2, respectively. The time period of
vibration (TN) for the equivalent spring-mass system is equal to

6
exact analytical
5 numerical
Pro/Pio

4
3
2
1
0 5 10 15
Pio/Po

Fig. 5. Comparison of numerical and analytical solutions for the reflected shock pressure amplitude.
970 K.V. Subramaniam et al. / International Journal of Impact Engineering 36 (2009) 965–974

a 8.0E+06
with FSI w/o FSI
b 8.0E+06
with FSI w/o FSI
7.0E+06 7.0E+06

Pressure (Pa)
Pressure (Pa)
6.0E+06 6.0E+06
5.0E+06 5.0E+06
4.0E+06 4.0E+06
3.0E+06 3.0E+06
2.0E+06 2.0E+06
1.0E+06 1.0E+06
0.0E+00 0.0E+00
0 0.05 0.1 0.15 0 0.05 0.1 0.15
Time (sec) Time (sec)

c 6.0E+02
with FSI w/o FSI
d 2.5E+01
with FSI w/o FSI
2.0E+01
4.0E+02
1.5E+01

Velocity (m/s)
Velocity (m/s)

2.0E+02 1.0E+01
5.0E+00
0.0E+00
0.0E+00
0 0.05 0.1 0.15
-2.0E+02 -5.0E+00 0 0.05 0.1 0.15
-1.0E+01
-4.0E+02
-1.5E+01
-6.0E+02 -2.0E+01
Time (sec) Time (sec)

e 8.0E+00
with FSI w/o FSI
f 1.0E-01
with FSI w/o FSI
6.0E+00 8.0E-02

Displacement (m)
Displacement (m)

6.0E-02
4.0E+00
4.0E-02
2.0E+00
2.0E-02
0.0E+00
0.0E+00
-2.0E+00 0 0.05 0.1 0.15
-2.0E-02
0 0.05 0.1 0.15
-4.0E+00 -4.0E-02
-6.0E+00 -6.0E-02
Time (sec) Time (sec)

Fig. 6. Comparison of the pressure, velocity and displacement histories from analyses by considering and by ignoring FSI effect during the motion of the solid. (a), (c) and (e) for
Case A; and (b), (d) and (f) for Case B.

0.064715 s. For the second case (case B), the equivalent structural
stiffness and mass used in the analysis were obtained considering
a steel plate which is 8 m in length, 4 m wide and 25 cm thick-
ness. Fixed-edge boundary conditions were considered along the
4 m sides, while edges with the longer dimension of 4.5 m were
treated as pinned supports. The equivalent structural mass and
stiffness of the fundamental mode of vibration of plate per unit
cross-sectional area used in the analysis are equal to 1482 kg/m2
and 1.397e8 N/m/m2, respectively. The TN for this case is equal to
0.020465 s.
The pressure and the velocity time histories of the structural
mass are shown in Fig. 6 for the two cases. The blast over-pressure
obtained after reflection from a fixed rigid wall is also shown along
with the actual pressure time history for comparison. It can be seen
that the pressure immediately after reflection reaches the same
value equal to 72.78 atm in all cases. This suggests the amplitude of
the shock front immediately after reflection from a solid substrate,
initially at rest, is independent of the stiffness of the structure. The
amplification in the blast pressure amplitude upon reflection from
an elastic structure, initially at rest, is equal to Pro/Pio, which is
obtained after reflection from a fixed rigid wall.
The pressure, which follows the shock front after reflection from
an elastic structure, shows a cyclic variation about the reflected
pressure history obtained from a fixed rigid wall. It can be seen that
for the case of the structure with the larger stiffness, there is no
significant difference in the two pressure histories (with and
without considering FSI). For the structure with the lower stiffness,
however, the actual pressure experienced by the structure is
significantly different from that obtained of an immovable
boundary. From the pressure profile it appears that the duration of
the blast over-pressure produced at the face of the elastic structure
is relatively unaffected by its motion.
The velocity and displacement histories of the spring-mass
systems are shown in Fig. 6. The motion of the structure
ignoring FSI (w/o FSI in the graphs) was obtained using the
reflected pressure history from a fixed rigid wall. It can be seen
that for the case of the structure with the larger stiffness, the
role of FSI in the structural motion is not significant. For the
structure with the lower stiffness and mass, however, the actual
velocity and displacement of the structure are significantly
different from the corresponding values obtained when FSI is
ignored. The effect of FSI during the structural motion is
significant in this case; the maximum velocity and displacement
of the structure considering FSI are significantly lower than the
corresponding value when FSI is ignored. Further, the damped
response predicted considering FSI suggests significant energy
dissipation in the motion of the structure as a result of energy
exchange between the solid and the fluid medium during the
motion of the structure. For low stiffness system the resulting
damping is more significant. Therefore, an analysis procedure
which ignores FSI would not consider this exchange of energy
and can produce significant error in predicting the motion of the
structure for low stiffness systems. Looking at the results it is
also apparent that the velocity of structural motion immediately
following the leading shock directly influences the pressure
history experienced by the structure; the structure with smaller
stiffness and mass has a larger velocity and hence experiences
a significantly variable pressure history. The variations in the
velocity profile within the duration of the blast correspond well
with the variation in the pressure about the pressure reflected
off of a rigid boundary. The maximum velocity during the early
motion of the structure occurs approximately at TN/4. The
maximum displacement, during the structural motion corre-
sponds closely in time with TN/2.
 K.V. Subramaniam et al. / International Journal of Impact Engineering 36 (2009) 965–974
7. Discussion
The reflection of a blast pressure wave from a moving boundary
is shown schematically in Fig. 7. The incident shock of amplitude Pio
is reflected as a shock of amplitude Pro traveling in the opposite
direction. The pressure experienced by the moving boundary at any
given time, pr(t) is equivalent to the reflected pressure wave. The
incident blast pressure history, pi(t) cannot be accessed directly at
the surface of the moving boundary since the measured pressure
corresponds to pr(t). For a given blast, the incident shock pressure
amplitude and the pressure history can be estimated knowing the
strength of the blast at the source and the distance from the blast
source [TM 51300]. pr(t) can be obtained from the incident pressure
history, pi(t) as a function of the boundary velocity.
From the results of the numerical analysis a relevant question
which needs to be addressed is: for a given incident blast pressure
wave (identified by the peak pressure and duration), what is the
range of structural parameters for which the FSI is important? The
key structural parameters of concern for an elastic structure are
the structural stiffness, ks, and the structural mass, ms. Pressure at
the face of a moving boundary is known to be directly related to its
velocity. In the case of an elastic structure, the applied pressure
history (pr(t))and the velocity of the structure within the duration
of the blast ðu_ s ðtÞÞ are thus intimately related. For a given incident
blast pressure wave, the pressure history experienced by the elastic
structure (pr(t)) would therefore depend upon the structure
velocity in addition to the blast wave properties and can be
expressed symbolically as

pr ðtÞ ¼ pr Pio ; tB ; u_ s ðtÞ; f1 ; f2 ; .; fN
where Pio and tB are the pressure amplitude and the duration of the
incident blast pressure wave, respectively, f1, f2,., fN are variables
which define the pressure variation behind the shock front. It
should be noted that the functional relationships between these
variables are intractable. The exact variation in the pressure history
behind the shock front involves the solution of non-linear differ-
ential equations, which does not allow the development of
analytical expressions.
Incident blast pressure wave at
location of O
Pio
Pi(t)
tB
Pro
Pr(t)
O blast, the importance of FSI in determining the structural response
tB
Reflected blast pressure wave at O
Fig. 7. Schematic representation of the incident and reflected blast pressure waves
from a moving boundary.
971
There is a unique relationship between the Pio and the particle
velocity of air immediately upstream of the shock front of blast
wave speed (vio), given as [20, p. 190]
sffiffiffi
Pio
2 po  1
nio ¼ ao qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi (15)
g ðg  1Þ þ ðg þ 1ÞPio
po
pffiffiffiffiffiffiffiffiffiffiffi
where Po is the ambient pressure and ao ¼ gRTo is the sound
speed under ambient conditions. pr(t) can now be written as
pr ðtÞ ¼ pr nio ; tB ; u_ s ðtÞ; f1 ; f2 ; .fN (16)
The velocity of the structure subjected to an applied pressure
history, pr(t) can be written symbolically as
u_ s ðtÞ ¼ u_ s ðks ; ms ; pr ðtÞÞ (17)
Considering Eq. (16), an implicit expression for the velocity of
the structure is obtained
u_ s ðtÞ ¼ u_ s ks ; ms ; nio ; tB ; f1 ; f2 ; .; fN ; u_ s ðtÞ (18)
An incident blast pressure wave defined by Pio and tB, and with
a defined incident pressure history given by fi, would, upon
reflection from a fixed rigid surface produce a well-defined,
reflected pressure wave. If we now consider the motion of the
structure obtained by directly applying the reflected pressure
history from a fixed rigid surface to the structure, there is a unique
relationship between the time period of vibration (TN) and the
maximum velocity in the first half cycle during the vibration ðu_ max
s Þ
(14) and (ks, ms). The velocity of the structure following the incidence of
a blast pressure wave can thus be expressed equivalently in terms
of TN and u_ max as
s

u_ s ðtÞ ¼ u_ s TN ; u_ max
s ; nio ; tB ; f1 ; f2 ; .fN ; u_ s ðtÞ (19)
While functional form of the expression is intractable and may
not readily allow for developing an expression for u_ s ðtÞ in terms of
non-dimensional ratios, two key ratios can nevertheless be iden-
tified, which allow for relating structural parameters with the blast
wave characteristics. Eq. (19) can thus be rewritten as
!
TN u_ max
u_ s ¼ u_ s ; s ; t ; n ; f ; f ; .; fN ; t (20)
tB nio B io 1 2
It should be noted that in Eq. (20), the variables following the
ratios define the pressure variation in the blast wave.
For a given incident blast pressure wave the FSI depends upon
the velocity of the structure, which in turn depends upon two ratios
relating the structural and blast parameters. The velocity ratio
ðu_ max
s =nio Þ provides a measure of the structure velocity relative to
the particle velocity associated with blast. The second important
parameter, the time ratio (TN/tB), gives an indication of the duration
of the applied pressure and hence is related to the impulse of blast
Us pressure wave. Thus it can be postulated that for a given incident
would depend upon the time and velocity ratios given as TN/tB and
ðu_ max
s =nio Þ, respectively. This is now verified in the subsequent
analysis where the influence of FSI is evaluated in terms of one
structural response parameter obtained with and without FSI for
different values of the two ratios. A numerical evaluation is per-
formed to see if the ratios identified in Eq. (20), allow for an eval-
uation of FSI effects. The influence of other variables related to the
blast pressure wave (highlighted variables) is qualitatively evalu-
ated considering different input blasts (with varying pressure
amplitudes and duration).
972 K.V. Subramaniam et al. / International Journal of Impact Engineering 36 (2009) 965–974
Considering the vibration response of a 1-DOF structure, which
is initially at rest, the maximum displacement and velocity within
the first half cycle occur approximately at TN/4 and TN/2, respec-
tively. As shown in Fig. 8, there are three different cases which need
to be evaluated based on the relative magnitudes of TN/4 and tB.
Case 1: when TN/4 is smaller than tB/2, both maximum velocity
and displacement of the structure occur within tB.
Case 2: when TN/4 is greater than tB/2 but smaller than tB, the
maximum velocity of the structure occurs before tB, but the
maximum displacement in the vibratory motion occurs after tB.
In this case, the maximum displacement within tB is smaller
than the maximum displacement obtained from the subsequent
motion of the structure.
Case 3: when TN/4 is larger than tB, maximum displacement of
the structure occurs after tB. In this case, the maximum velocity
of the structure u_ max
s occurs close to tB. The time for the occur-
rence of u_ max
s will approach tb when the ratio tb/(TN/4)
approaches zero.
The velocity of the structure after the blast pressure has
decreased to the ambient level and has no influence on the pressure
history. Only the velocity of the structure during tB is of significance
in evaluating FSI.
Numerical analyses were performed to evaluate if the velocity
and time ratios allow for assessing the influence of FSI in deter-
mining structural response parameters. The displacement was used
as the structural response parameter for evaluation. The structural
displacement obtained from an analysis considering FSI when
compared with displacement obtained without considering FSI was
used as an indicator of the error in predicted structural response on
ignoring FSI. Analyses were performed for varying values of the
velocity ratio for fixed values of TN/tB. The maximum displacement
of the structure within the duration of the blast pressure wave was
evaluated as function of the velocity ratio. In Fig. 9, the displace-
ment ratio is defined as the ratio of the maximum displacement of
the structure during its vibratory motion considering FSI to the
corresponding displacement obtained from an analysis ignoring FSI
u_ max
is plotted as a function of the velocity ratio ðp sffiffiffiffiffiffiffi =np Þ. Results
ioffiffiffiffiffiffi fromffi
pffiffiffiffiffiffiffiffiffiffiffi
three different values of TN/tB equal to 0:1; 10 and 1000,
which correspond to cases 1, 2 and 3, respectively, are shown in
Fig. 9. For the cases 2 and 3, since the maximum displacement
during the vibratory motion is not attained before tB, the
displacement of the structure at tB was used for computing the
displacement ratio. The u_ max
s used in generating the graph corre-
sponds to the maximum velocity obtained from an analysis without
Pressure
Pro
tB
Velocity
max Case 1 Case 2
Case 3
Us
Time
Fig. 8. Schematic representation of three cases based on relative magnitudes of TN and tB.
FSI (where the reflected pressure history from a fixed rigid wall is
directly applied to the structure).
For each value of TN/tB, line plots (lines in the graph) are
shown for three different incident blast pressure amplitudes. The
incident blast pressure wave characteristics, Pio and vio, for each
wave used for generating the line plots were: (3.28 atm, 323.2 m/s),
(14.9 atm, 931 m/s) and (67.1 atm, 2137 m/s). The tB of all three
waves was equal to 64.5 ms. The reflected pressure amplitudes,
Pro, were equal to 8.57 atm, 72.78 atm, and 438 atm for the three
blast pressure waves. The graph also shows the displacement
ratio calculated as a function of velocity ratio considering blast
pressure waves with the same amplitudes but a smaller tB equal
to 6.45 ms. The results for these cases are shown using hollow
symbols on the graph.
Each curve in the graph was obtained for a given incident blast
pressure wave and a fixed value of TN/tB. To generate one curve, the
values of ks and ms were varied while keeping TN constant. For
a given combination of ks and ms, an analysis was also performed
using the reflected pressure history from the fixed rigid surface to
determine velocity and displacement variation as a function of
time. Finally, the displacement and velocity ratios were obtained by
comparing the corresponding values of displacement and velocity
from an analysis considering FSI.
For all three blast pressure waves considered in this study the
displacement ratio decreases continuously as a function of the
velocity ratio. This implies that the margin by which the displace-
ment obtained by ignoring the FSI effect would over predict the
actual displacement increases as the maximum structural velocity
increases relative to the particle velocity of the blast pressure wave.
The results of the analysis suggest that the displacement of the
structure obtained by ignoring FSI, deviates significantly from the
actual displacement for larger values of velocity ratio. This indicates
that as the velocity of the structure increases relative to the particle
velocity upstream of the incident blast pressure wave, the error in
predicting structural response also increases when FSI is not
considered. Conversely, the influence of FSI in decreasing the
displacement of the structure increases with an increase in
the velocity ratio. Considering the responses obtained from
different values of tB, for a given value of TN/tB, the relationship
between the velocity and displacement ratios is independent of tB.
This suggests that the velocity ratio is a good indicator of the error
in ignoring the influence of FSI for a given value of TN/tB.
It can be seen that for a given TN/tB ratio, the displacement ratio
is relatively independent of the incident blast pressure amplitude.
The three curves for a given value of TN/tB are comparable. The
relationship between the velocity and displacement ratios,
however, seems to depend upon TN/tB. This can pffiffiffiffiffiffibe explained
pffiffiffiffiffiffiffiffiffiffiffiffi by
considering the fact that for TN/tB equal to 10 and 1000, the
maximum displacement used for generating a curve is not the true
maximum displacement of the structure but the displacement at tB.
Since the value of structural displacement at tB would vary
depending upon the magnitudes of TN and tB, the relation between
the velocity and displacement ratios would change for different
values of TN and tB.
Results indicate that for any value of TN/tB, the influence of FSI
is not significant when the velocity ratio is small. For a given blast
pressure wave, the structural velocity is influenced by both
structural mass and stiffness. Combinations of stiffness and mass
which result in small velocity related to particle velocity of air
(relatively stiff structures) would result in an insignificant influ-
ence of FSI. When the structural parameters result in a large
velocity related to the incident particle velocity (compliant
structures), the resulting motion would indicate a significant level
of damping as a result of higher level of energy exchange between
the structure and air.
 K.V. Subramaniam et al. / International Journal of Impact Engineering 36 (2009) 965–974
1
0.95
0.9
0.85
Displacement ratio
(UsFSI/Usno FSI)
0.8
0.75
0.7
0.65
0.6
0.55
0.5
0 0.2
Fig. 9. Relationship between velocity and displacement ratios for different values of TN/tB (lines in the graph for tB ¼ 64.5 ms and symbols for tB ¼ 6.45 ms).
As the TN/tB increases, the duration of the blast as a proportion
of the period of vibration decreases. For an elastic structure with
finite ks and ms, as the ratio of TN/tB increases, the motion of the
structure within the duration of the blast is also a smaller
proportion of the maximum displacement during the motion of
the structure. One would expect the influence of FSI in predicting
the structural motion to decrease. However, there are two cases
which need to be clearly delineated for large TN/tB. These can be
illustrated considering the limit case as TN/tB approaches infinity.
This limit is achieved for the two following conditions: (1) ms
approaching infinity for finite ks; and (2) ks approaching zero for
finite ms. In case 1, the velocity ratio would approach zero and
hence the influence of FSI would be insignificant. Physically case 2
corresponds to the case of a free-standing structure in the path of
the blast wave. The velocity of the free-standing structure
produced by the blast pressure wave depends upon the magni-
tude of mass and the pressure amplitude of the blast pressure
wave. When the mass of the structure is such that it is accelerated
to speeds which are not insignificant when compared to the
particle velocity of the blast pressure wave, then there would be
a significant influence on the resulting pressure. Therefore, for
large velocities the role of FSI would be significant even as TN/tB
approaches infinity. The role of FSI would therefore depend on
the velocity ratio for all values of TN/tB.
The results presented in the graph also provide a means for
evaluating the need for considering FSI in determining the struc-
tural motion. The results indicate that for a given TN/tB, the velocity
ratio give a good indication of the decrease in the maximum
displacement during the motion of the structure upon considering
FSI. For a given incident blast pressure wave, the need for FSI can
now be evaluated using the results presented in Fig. 9. A conven-
tional analysis can initially be performed using the pressure history
reflected from a fixed rigid wall. The relative error in ignoring FSI
can be evaluated using the ratio of u_ max s obtained from the
conventional analysis to vio of the incident blast pressure wave.
Typically, in conventional analysis, only the reflected pressure
history obtained from a fixed rigid surface is available. One can,
however, estimate the incident blast pressure amplitude using the
relation which is obtained from analysis of shock waves [13]
8 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi9
 2
1 <Pro Pro =
Pio Pro
¼ þ1þ þ 1 þ192
po 16: Po Po Po ;
973
TN/tB = 31.6, Pio = 67.1atm
TN/tB = 31.6, Pio = 14.9atm
TN/tB = 3.16, Pio = 3.28atm
TN/tB = 3.16, Pio = 67.1atm
TN/tB = 3.16, Pio = 14.9atm
TN/tB =3.16, Pio = 3.28atm
TN/tB = 0.316, Pio = 67.1atm
TN/tB = 0.316, Pio = 14.9atm
TN/tB = 0.316, Pio = 3.28atm
0.4 0.6 0.8 1 1.2 1.4
Velocity ratio
The corresponding value of vio for determining the velocity ratio
can be calculated using Eq. (15).
8. Conclusions
A numerical scheme for coupled analysis is presented.
A procedure for evaluating the influence of FSI in structural
response parameters during the dynamic motion of the structure is
presented. Based on the results presented in this paper the
following conclusions can be drawn:
1. The pressure amplitude at the face of an elastic structure,
which is initially at rest, immediately following the reflection of
the blast pressure wave is identical to the pressure produced
from a reflection off of a fixed rigid wall. Subsequent motion of
the structure produces changes in the fluid domain, which
influences the pressure experienced by the structure.
2. The displacement of the structure predicted considering FSI
during its motion is smaller than the displacement predicted
using an approach where the reflected pressure history from
a fixed rigid wall is directly imposed on the structure.
3. Considering FSI leads to a damped periodic motion of the
structure. There is a continuous exchange of energy between
the elastic solid and the air during the motion of the structure.
4. The difference in the displacement predicted using an
approach which does not consider the influence of FSI when
compared with the corresponding displacement predicted
considering FSI increases with an increase in the ratio of the
structural velocity to the particle velocity upstream of the
incident blast.
5. The structural motion obtained from an analysis where the
pressure history from a fixed rigid surface is directly prescribed
on the structure would be identical to the motion obtained
from an analysis which considers FSI when the velocity ratio
approaches zero.
References
[1] Structures to resist the effects of accidental explosions. Technical manual
TM5-1300. Washington, DC: US Department of the Army; 1990.
[2] Hause T, Librescu L. Dynamic response of anisotropic flat panels to explosive
pressure pulses. Int J Impact Eng 2005;31:607–28.
(21)
[3] Krauthammer T, Altenberg A. Negative phase blast effects on glass panels. Int J
Impact Eng 2000;24:1–17.
974 K.V. Subramaniam et al. / International Journal of Impact Engineering 36 (2009) 965–974
[4] Lee HK, Lee W, Ki K. Dynamic analysis of nonlinear composite structures under
pressure wave loading. J Comp Matls 2006;40(15):1361–83.
[5] Pandey AK, Kumar R, Paul DK, Trikha DN. Non-linear response of reinforced
concrete containment structure under blast loading. Nucl Eng Des
2006;236:993–1002.
[6] Morison CM. Dynamic response of walls and slabs by single degree-of-
freedom analysis – a critical review and revision. Int J Impact Eng
2006;32:1214–47.
[7] Gantes CJ, Pnevmatikos NG. Elastic–plastic response spectra for exponential
blast loading. Int J Impact Eng 2004;30:323–43.
[8] Taylor G. The pressure and impulse of submarine explosion waves on
plates. In: Batchelor G, editor. The scientific papers of Sir Geoffrey
Ingram Taylor. Aerodynamics and the mechanics of projectiles and
explosions, vol. III. Cambridge, UK: Cambridge University Press; 1963. p.
287–303.
[9] Deshpande V, Fleck N. One-dimensional response of sandwich plates to
underwater shock loading. J Mech Phys Solids 2005;53(11):2347–83.
[10] Hutchinson JW, Xue Z. Metal sandwich plates optimized for pressure impulses.
Int J Mech Sci 2005;47:545–69.
[11] Tan PJ, Reid SR, Harrigan JJ. Discussion: ‘‘the resistance of clamped sandwich
beams to shock loading’’ (Fleck NA, Deshpande VS. ASME J Appl Mech
2004;71:386–401). J Appl Mech 2005;72:978–9.
[12] Kambouchev N, Noels L, Radovitzky R. Numerical simulation of the fluid–
structure interaction between air blast waves and free-standing plates.
Comput Struct 2007;85:923–31.
[13] Courant R, Friedrichs KO. Supersonic flow and shock waves. Springer-Verlag;
1976. pp. 95, 153.
[14] Gong M. Mutual interactions between shock waves and structures. PhD Thesis,
City University of New York, New York, NY; 2006. pp. 187-89.
[15] Saad M. Compressible fluid flow. 2nd ed. Saddle Brook, NJ: Prentice Hall; 1998.
pp. 187–189.
[16] Hint CW, Amsden AA, Cook JL. An arbitrary Lagrangian–Eulerian coupling
method for all flow speeds. J Comput Phys 1974;14:227–53.
[17] Donea J, Giuliani S, Halleux JP. An arbitrary Lagrangian–Eulerian finite element
method for transient dynamic fluid–structure interactions. Comput Meth Appl
Mech Eng 1982;33:689–723.
[18] Blom FJ, Leyland P. Analysis of fluid–structure interaction on moving airfoils by
means of an improved ALE-method. AIAA 1997:97–1770.
[19] Blom FJ. A monolithical fluid–structure interaction algorithm applied to the
piston problem. Comput Meth Appl Mech Eng 1998;167:369–91.
[20] Toro EF. Riemann solvers and numerical methods for fluid dynamics: a prac-
tical introduction. Springer Verlag; 1999. pp. 190, 223–24, 491.
[21] Roe PL, Pike J. Efficient construction and utilisation of approximate Riemann
solutions. In: Glowinski R, Lions J-L, editors. Computing methods in applied
science and engineering VI, INRIA. Amsterdam: North-Holland Publishing Co.;
1984. p. 499-518.
[22] Farhat C. High performance simulation of coupled nonlinear transient aero-
elastic problems (Cosmase Course, EPF-Lausanne, 1995).
[23] Farhat C, Lin TY. A structure attached corotational fluid grid for transient
aeroelastic computations. AIAA J 1993;31(3):597–9.