You are on page 1of 14

Numerical-experimental study of steel

plates subjected to blast loading


Daniel Ambrosini1,2 & Abel Jacinto3

1 CONICET
2 Engineering

Faculty, Nat. Univ. of Cuyo, Argentina


Univ. of Cataluna, Spain

3 Politechnical

Abstract
The main objective of this paper is the comparison between testing and
numerical responses of metallic plates subjected to air blast loads, in order to determine the accuracy of modern calculation methods and computational codes. The secondary objective is to provide data that could be
used for checking the accuracy of a variety of calculation methods. A set
of four tests at full scale is presented on two non-stiffened metallic steel
plates with different boundary conditions (one set up as a cantilever and
another clamped along the four edges), subjected to the action of pressure
waves originating from the detonation of high explosive charges. The time
history of the acceleration at different points on both plates and the pressure waves at selected points, are recorded. In addition, a linear dynamic
analysis of the plate models was carried out (using the codes ABAQUS and
COSMOS). Suggestions about computational modeling of structures under
impulsive loads are discussed arising from the comparison of numerical and
experimental results.
Keywords: air blast waves, unstiffened plates, dynamic response, experimental
analysis
1 Introduction
In recent years explosive loads have received considerable attention because
of different events, accidental or intentional, relating to important structures
all over the world. As a consequence, there has been an increased activity

WIT Transactions on Engineering Sciences, Vol 49


Impact Loading of Lightweight Structures, M. Alves & N. Jones (Editors)
c 2005 WIT Press, www.witpress.com, ISSN 1743-3533

50

Daniel Ambrosini & Abel Jacinto

in explosive loading research in the last decade. Initially, this work was
mostly empirical, but, in recent years, important new methods have begun
to be developed. The dynamic loads originating from explosions result in
strain rates in the material of about 101 to 103 s1 . These extreme loads
produce a special behaviour in the material that is characterized, among
other effects, by an increase in strength, in comparison with normal, static
properties. Galiev [1] and Krauthammer et al. [2], among others, describe
the behaviour of metals under impulsive load.
According to Smith et al. [3], even nominally identical, well-controlled experiments involving explosives can produce results with a significant spread,
making data analysis more uncertain than is desirable. For this reason, good
quality experimental data can be useful for theoretical and numerical investigations. Moreover, Pan and Louca [4] said that while there has been interest
in blast resistance of plates and panels over the past few years, fundamentally for use in topside modules in the offshore industry, there is very little
data available on their response characteristics. However, recently, Louca
et al. [5] and Boh et al. [6] present studies about pro?led panels used as
blastwalls in typical offshore topsides modules to provide a safety barrier
for working personnel and equipment.
The objectives of this paper are, firstly, to determine the accuracy of modern calculation methods and computational codes and, secondly, to provide
good quality data that can be used for checking the accuracy and reliability
of new calculation methods and procedures. To achieve these objectives,
a set of four tests at full scale was performed on two unstiffened metallic
steel plates with different boundary conditions (one set up as a cantilever in
the ground and another clamped around the four edges), subjected to the
action of pressure waves produced by the detonation of explosive charges
equivalent to 1 to 10 kg of TNT. The acceleration time histories at different points on both plates and the pressure waves at selected points, are
recorded. In addition, numerical analysis using the finite element programs
ABAQUS/Standard [7] and COSMOS/M [8] were carried out. In previous
papers [9, 10], some results related with different tests and numerical verification were presented. In this paper, results of others tests are presented
jointly with the comparison of results obtained with different computational
codes.
Related to dynamic response of plates subjected to blast loading, Rudrapatna et al [11] present numerical results for clamped, square stiffened steel
plates subjected to blast loading, Nurick et al [12] present experimental results for the prediction of tearing of clamped circular plates subjected to
uniformly loaded air blast, Louca et al [13] describes numerical results of
nonlinear analysis on both stiffened and unstiffened plates and Shen and
Jones [14] analyze the nonlinear dynamic response and failure of clamped
circular plates. Recently, Yuen and Nurick [15] present valuables experimen-

WIT Transactions on Engineering Sciences, Vol 49


Impact Loading of Lightweight Structures, M. Alves & N. Jones (Editors)
c 2005 WIT Press, www.witpress.com, ISSN 1743-3533

Numerical-experimental study of plates subjected to blast loading

51

tal and numerical studies on the response of quadrangular stiffened plates


subjected to uniform blast load and Langdon et al. [16] analyze the case of
localized blast loading. Moreover, Akus and Yildirim [17] study the effect
of thickness on deformation of plates subjected to blast loading.
2 Experimental tests description
2.1 Experimental set-up
The structures analysed in this paper are two unstiffened steel plates subject
to blast loads caused by the detonation of explosives at different stand-off
distances. The explosive used was Gelamon VF80 with a theoretical TNT
equivalence of 0.8. Four tests with different amounts of explosive were carried out (Figure 1). In order to measure the overpressure generated by the
blast waves, four pressure sensors (Honeywell 180PC) were used, located in
the front and back positions of the plates. In addition, three accelerometers
were used to measure the dynamic response of the plates (Figure 2) and a dynamic strain amplifier amplified the signal generated by the accelerometers.
A 100 kHz data acquisition board was mounted on a notebook computer
in order to record and process the signals. Seven channels were used in all
tests and the signals were sampled at 4000 sps for each channel (28,000 sps
total rate acquisition).
The experimental pressure time history was used as applied load in the
computational analysis of the plate. A typical pressure-time history is given
in Figure 3, from Test 4.
Test 2
(10 kg TNT)

30.0 m

5.65 m

5.0 m

Test 4
(10 kg TNT)

Plate B
Plate A

4.05 m

Test 3
(1 kg TNT)

Test 1
(0.8 kg TNT)

0.75 m

30.0 m
10.0 m

0.75 m

50.0 m

References
Plate A: Metallic plate clamped at the base
Plate B: Metallic plate campled at four edges

Figure 1: Experimental set-up

WIT Transactions on Engineering Sciences, Vol 49


Impact Loading of Lightweight Structures, M. Alves & N. Jones (Editors)
c 2005 WIT Press, www.witpress.com, ISSN 1743-3533

Equipment

52

Daniel Ambrosini & Abel Jacinto

2.2 Plates tested


The steel plate A is 1.0 m wide and 1.5 m high and 2.1 mm thick. The
bottom of the plate was set up as a cantilever on a concrete base (Figure
2). The accelerometer locations in the tests are shown in Figure 2. For the
case of Test 1, accelerometers were positioned at points 1 and 2, and for the
other tests they were positioned at the three points indicated in Figure 2.
The steel plate B is 0.95 m wide, 0.95 m high and 0.9 mm thick. The
four edges of the plate were clamped to a rigid steel frame of I section of
100 x 50 x 4.5 mm. This frame was clamped to a concrete base (Fig. 2).
The accelerometer was located in the centre of the plate (Fig. 2). The experimental dynamic response measured for Test 1 is presented in the next
sections.
3 Numerical analysis
The numerical analysis was carried out using the finite element program
ABAQUS/Standard 5.8 [7] and COSMOS/M 1.7 [8]. The plates were modeled using shell elements. In both cases, the boundary conditions were considered as perfectly clamped. The material properties adopted were: Youngs
E = 180 GPa (experimental value), Poisson coefficient = 0.3, density
= 7850 kg/m3 . The dynamic analysis was performed using the modal superposition method and integration direct method. The integration step by
step was carried out using the Newmark- method ( = 1/4 and = 1/2)
and the time step size was 0.25 ms. The time-history of the acceleration in
the nodes corresponding to the accelerometer position in the test was determined. Only the results corresponding to the modal superposition method
are presented in this paper.
3.1 Metallic plate clamped in the bottom (Plate A)
Numerical models of 150 (10x15 mesh, Model 1) and 600 elements (20x30
mesh, Model 2) were used. The damping coefficient adopted was 0.6 %. This
value, obtained experimentally, is similar to other values found in literature
for steel working under low stresses and was discussed in [10].
The pressure time history shown in Fig. 4b was used as the applied load
for the numerical analysis. This curve was obtained by the vector addition
of the recorded pressures in Sensor 2 (front) and Sensor 4 (back). These
pressure waves have different arrival times (Figure 4a) and are acting in
opposite directions. Moreover, in order to introduce a soft variation in
the input pressure, the oscillations observed in the resultant record of the
shock wave were smoothed, considering only the upper points of the curve
(Figure 4b). In addition, for comparison purposes, the pressure recorded on

WIT Transactions on Engineering Sciences, Vol 49


Impact Loading of Lightweight Structures, M. Alves & N. Jones (Editors)
c 2005 WIT Press, www.witpress.com, ISSN 1743-3533

Numerical-experimental study of plates subjected to blast loading

53

1.0 m

0.5 m

0.5 m

0.5 m

Accelerometers

0.4 m

1.2 m

(a) Plate A

Steel plate

0.95 m

0.95 m

0.2 m

Accelerometer
(b) Plate B

Figure 2: Plates tested and accelerometers location

the front face of the plate was also used as the applied load for the numerical
analysis.
In a previous paper by the authors [9], it was demonstrated that, in this
case, it is equivalent to use the superposition modal method or the direct integration method because, for the pressures involved, the structures behave
in a linearly elastic manner and the results obtained with both methods are
similar.
WIT Transactions on Engineering Sciences, Vol 49
Impact Loading of Lightweight Structures, M. Alves & N. Jones (Editors)
c 2005 WIT Press, www.witpress.com, ISSN 1743-3533

54

Daniel Ambrosini & Abel Jacinto

12.0
10.0

Pressure (kPa)

8.0
6.0
4.0
2.0
0.0
-2.0
-4.0
6.05

6.07

6.09

6.11

6.13

6.15

Time (s)

Figure 3: Record of the time history of the overpressure. Test 4. Sensor 3.

3.2 Metallic Plate clamped around the four edges (Plate B)


Numerical models of 100 (10x10 mesh, Model 1) and 400 elements (20x20
mesh, Model 2) were used. The damping coefficient, experimentally determined, was 2.2 %. Again, this value was discussed in [10]. As in the case of
plate A, the response of the plate was determined using as loading the blast
wave shown in Fig. 4a and the pressure recorded on the front face of the
plate for comparison purposes. The acceleration in the centre of the plate
was determined using the modal superposition method and the response is
shown in the next section.
4 Results and discussion
Firstly, recorded air blast reflected overpressures were compared with predicted values [18] based on empirical findings. Table 1 show measured and
estimated values including the positive phase duration. The agreement between the values, except for test 2, is very good and consistent with hemispherical air blast loading.
4.1 Natural frequencies
In order to validate the numerical models, firstly, measured and numerical
natural frequencies were compared and the results are presented in Table 2
for the first four modes.
WIT Transactions on Engineering Sciences, Vol 49
Impact Loading of Lightweight Structures, M. Alves & N. Jones (Editors)
c 2005 WIT Press, www.witpress.com, ISSN 1743-3533

Numerical-experimental study of plates subjected to blast loading

55

3.0
2.5

Pressure (kPa)

2.0
1.5
1.0
0.5
0.0
-0.5
-1.0
-1.5
5.95

5.96

5.97

5.98

5.99

6.00

5.99

6.00

Time (s)
Sensor 2

Sensor 4

(a) Measured pressures


3.0
2.5

Pressure (kPa)

2.0
1.5
1.0
0.5
0.0
-0.5
-1.0
-1.5
5.95

5.96

5.97

5.98

Time (s)
Resultant

Smoothed

(b) Resultant pressures

Figure 4: Applied load used in the analysis

In the case of the plate A, the second mode (torsional) is missing in the experimental determination and there are some differences between measured
and numerical values, particularly in higher modes. This behaviour can be
explained by the second order effects due to the dimensions of the plate
WIT Transactions on Engineering Sciences, Vol 49
Impact Loading of Lightweight Structures, M. Alves & N. Jones (Editors)
c 2005 WIT Press, www.witpress.com, ISSN 1743-3533

56

Daniel Ambrosini & Abel Jacinto

Table 1: Blast pressure and duration estimated and recorded. Sensor 3


Measured
Test

Estimated

Overpressure

Duration

Overpressure

Duration

Pr (kPa)

td (ms)

Pr (kPa)

td (ms)

4.6

5.5

4.6

5.7

10.0

12.2

7.9

10.8

3.6

6.5

3.5

6.1

14.8

10.0

15.0

10.1

(very low thickness/width ratio) and the support conditions that affect the
stability of the plate and make it sensitive to imperfections. Nevertheless,
the differences between experimental and numerical modes of vibration do
not affect the dynamic response significantly, as is shown in the next point,
probably because the pressure is symmetrical and so the influence of antisymmetrical modes in the response is of minor importance. There are minor
differences between the two numerical modes in this case.
In Plate B, there is a better agreement between the numerical and experimental values and consequently the computational model (mesh and
material properties) can be used for the next step of the analysis, that is
the forced vibration. It is interesting to note that the refineshment of the
numerical model lead to a more flexible model in case of ABAQUS and
more stiff in case of COSMOS. However, in all cases the model obtained
with ABAQUS is dynamically more stiff that the model of COSMOS. This
difference can be explained by the different Shell elements used by both
codes.
4.2 Dynamic response
To meet the objectives of the paper, a comparison between the computational and the experimental results of the acceleration time history will be
carried out at this point. At first, as an example, a complete acceleration
record is showed in Figure 5.
In Figure 6, a comparison between the results obtained experimentally
and with the two codes used, is showed. It can be seen that both codes
overestimate the peak acceleration record but shows a good qualitatively
adjust in the forced vibration zone. In the free vibration zone, more differences between the programs is founded.
In Figures 7 and 8, the experimental values of the time history of the acceleration of plate A are compared with those obtained using the ABAQUS
WIT Transactions on Engineering Sciences, Vol 49
Impact Loading of Lightweight Structures, M. Alves & N. Jones (Editors)
c 2005 WIT Press, www.witpress.com, ISSN 1743-3533

Numerical-experimental study of plates subjected to blast loading

57

Table 2: Natural frequencies of the Plates (Hz)


Plate A
Mode

Meas.

ABAQUS

COSMOS

Model 1

Model 2

Model 1

Model 2

0.72

0.74

0.74

0.74

0.74

2.51

2.51

2.49

2.50

4.99

4.67

4.63

4.59

4.61

7.62

8.52

8.48

8.38

8.44

Plate B
1

7.98

8.45

8.32

8.21

8.26

16.22

17.98

17.14

16.69

16.83

23.70

26.29

25.23

24.34

24.76

27.94

35.03

31.32

29.76

30.15

15.0

10.0

5.0

Acc. (g)
0.0

-5.0

-10.0

-15.0
5.95

6.15

6.35

6.55

6.75

6.95

Time (sec)

Figure 5: Acceleration record. Plate A. Test 1. Location 1.


program. These results were obtained using the modal superposition method
involving 60 modes because it was founded in previous papers [9, 10] that
this amount of modes it is necessary in order to reproduce the measured
response. It is important to note that the response is shown for a long time
(50 ms), that is, approximately 4 and 2.5 times the duration of the applied
WIT Transactions on Engineering Sciences, Vol 49
Impact Loading of Lightweight Structures, M. Alves & N. Jones (Editors)
c 2005 WIT Press, www.witpress.com, ISSN 1743-3533

58

Daniel Ambrosini & Abel Jacinto

20.0
15.0

Acceleration (g)

10.0
5.0
0.0
-5.0
-10.0
-15.0
0.00

0.01

0.02

0.03

0.04

0.05

Time (sec.)
Measured values

COSMOS

ABAQUS

Figure 6: Acceleration time histories. Plate A. Test 1. Location 2.


20.0

Forced vibration

Free vibration

15.0

Acceleration (g)

10.0
5.0
0.0
-5.0
-10.0
-15.0
-20.0
0.00

0.01

0.02

0.03

0.04

0.05

Time (s)
Measured

Numerical

Figure 7: Acceleration time histories. Plate A. Test 3. Location 2.

load for the tests 3 and 4 respectively.


In addition, in Table 3 the acceleration peak values of both responses are
WIT Transactions on Engineering Sciences, Vol 49
Impact Loading of Lightweight Structures, M. Alves & N. Jones (Editors)
c 2005 WIT Press, www.witpress.com, ISSN 1743-3533

Numerical-experimental study of plates subjected to blast loading

59

120.0

Forced vibration

Free vibration

80.0

Acceleration (g)

40.0

0.0

-40.0

-80.0

-120.0
0.00

0.01

0.02

0.03

0.04

0.05

Time (s)
Measured

Numerical

Figure 8: Acceleration time histories. Plate A. Test 4. Location 3.


80.0
60.0

Acceleration (g)

40.0
20.0
0.0
-20.0
-40.0
-60.0
-80.0
0.00

0.01

0.02

0.03

0.04

0.05

Time (s)
Measured

Numerical

Figure 9: Acceleration time histories. Plate B. Test 1.

presented. For both locations 2 and 3 an excellent agreement was achieved


for the peak values of acceleration in the forced vibration zone. For both
WIT Transactions on Engineering Sciences, Vol 49
Impact Loading of Lightweight Structures, M. Alves & N. Jones (Editors)
c 2005 WIT Press, www.witpress.com, ISSN 1743-3533

60

Daniel Ambrosini & Abel Jacinto

tests in the location 3 there are a very good adjustment for the two first
pulses of the response and for the first pulse in the location 2. In the free
vibration zone, for both positions and both tests, there was not complete
coincidence for high frequencies and only a qualitative agreement is founded.
However, these differences are not important for the design stage.

Table 3: Peak acceleration (g). Plate A.


Position

Test 3

Test 4

Measured

Numerical

Diff. (%)

Measured

Numerical

13.44

13.60

1.2

99.40

99.80

Diff. (%)
0.4

14.38

15.18

5.6

100.01

111.52

11.5

In Figure 9 the experimental values of the acceleration time history of


plate B are compared with those obtained using the ABAQUS code when
20 modal forms are considered.
Finally, in Table 4 the acceleration peak values of responses measured and
obtained with ABAQUS and COSMOS codes are presented. Again, there is
a good prediction for the peak value of acceleration as well as the first pulse
of the response and a qualitative agreement in the free vibration zone.

Table 4: Peak acceleration (g). Plate B.


Position
Center

Test 1
Measured

ABAQUS

COSMOS

61.08

57.21

55.25

5 Conclusions
A set of experimental results for unstiffened steel plates, with different
boundary conditions, subjected to air blast loading is presented. The experimental work has provided data on the dynamic response of these structures.
For comparison purposes, a numerical analysis was carried out.
When the superposition modal method it is used, it is extremely important the number of vibration modes considered in the analysis because, in
general, this type of loading, unlike earthquake or wind loading, excites the
high frequencies. Then, for numerical analysis prediction, it is necessary to
WIT Transactions on Engineering Sciences, Vol 49
Impact Loading of Lightweight Structures, M. Alves & N. Jones (Editors)
c 2005 WIT Press, www.witpress.com, ISSN 1743-3533

Numerical-experimental study of plates subjected to blast loading

61

carry out numerical tests to determine the appropriate number of modes


that will be used in the numerical analysis. Moreover, the element size of
computational models should agree with the number of modes that will be
included in the response. As more refined meshes capture the high frequencies with minor error and these frequencies have a significant participation
in the obtained response, it is therefore very important to determine its
values accurately.
According to table 1, the air blast wave was consistent with hemispherical
air blast loading and the pressure applied over the plates could be considered
as uniform, but the results obtained improved significantly when the load
applied was considered as the temporal superposition of the pressure over
the front and back faces of the plates and this effect could not be ignored
in this type of thin structures. However, the definition of the load applied
should be improved, considering a different spatial distribution over both
faces.
Modern computational codes can predict the dynamic response of unstiffened steel plates accurately, specially the forced vibration zone and the first
pulses of the response (see figs. 6-9 and tables 3-4). However, there are differences between different codes because its shell elements definition. These
differences could result in different predictions, particularly in free vibration
zone and in nonlinear studies.
References
[1] Galiev, U., Experimental observations and discussion of counterintuitive behavior of plates and shallow shells subjected to blast loading. International Journal of Impact Engineering, 18(7-8), pp. 783802, 1996.
[2] Krauthammer, T. & Ku, C., A hybrid computational approach for the
analysis of blast resistant connections. Computer & Structures, 61(5),
pp. 831843, 1996.
[3] Smith, P., Rose, T. & Saotonglang, E., Clearing of blast waves from
building facades. Proceedings of Institution of Civil Engineers: Structures & Buildings, volume 134, pp. 193199, 1999.
[4] Pan., Y. & Louca, L., Experimental and numerical studies on the response of stiffened plates subjected to gas explosions. Journal of Constructional Steel Research, 52, pp. 171193, 1999.
[5] Louca, L., Boh, J. & Choo, Y., Design and analysis of stainless steel
profiled blast barriers. Journal of Constructional Steel Research, 60,
pp. 16991723, 2004.
[6] Boh, J., Louca, L. & Choo, Y., Numerical assessment of explosion resistant profiled barriers. Marine Structures, 17, pp. 139160, 2004.
[7] ABAQUS/Standard 5.7-3; Users Manual, 1997.

WIT Transactions on Engineering Sciences, Vol 49


Impact Loading of Lightweight Structures, M. Alves & N. Jones (Editors)
c 2005 WIT Press, www.witpress.com, ISSN 1743-3533

62

Daniel Ambrosini & Abel Jacinto

[8] COSMOS/M; Version 1.71. User Guide, 1994.


[9] Jacinto, A., Ambrosini, D. & Danesi, R., Experimental and computational analysis of plates under air blast loading. International Journal
of Impact Engineering, 25, pp. 927947, 2001.
[10] Jacinto, A., Ambrosini, D. & Danesi, R., Dynamic response of plates
subjected to blast loading. Proceedings of the Institution of Civil Engineers: Structures and Buildings; SB152, volume 3, pp. 269276, 2002.
[11] Rudrapatna, N., Vaziri, R. & Olson, M., Deformation and failure of
blast-loaded stiffened plates. International Journal of Impact Engineering, 24, pp. 457474, 2000.
[12] Nurick, G., Gelman, M. & Marshall, M., Tearing of blast loaded plates
with clamped boundary conditions. International Journal of Impact
Engineering, 18, pp. 803827, 1996.
[13] Louca, L. & Pan, Y., Response of stiffened and unstiffened plates subjected to blast loading. Engineering Structures, 20(12), pp. 10791086,
1998.
[14] Shen, W. & Jones, N., Dynamic response and failure of fully clamped
circular plates under impulsive loading. International Journal of Impact
Engineering, 13, pp. 259278, 1993.
[15] Yuen, S. & Nurick, G., Experimental and numerical studies on the
response of quadrangular stiffened plates. part i: subjected to uniform
blast load. International Journal of Impact Engineering, 31(1), pp. 55
83, 2005.
[16] Langdon, G., Yuen, S. & Nurick, G., Experimental and numerical studies on the response of quadrangular stiffened plates. part ii: localised
blast loading. International Journal of Impact Engineering, 31(1), pp.
85111, 2005.
[17] Akus, Y. & Yildirim, R., Effect of thickness on deformation of plates
subjected to transient high pressures. Proceedings of ESDA 2004: 7th
Biennial Conference on Engineering Systems Design and Analysis,
Manchester, United Kingdom, pp. 16, 2004.
[18] Smith, P. & Hetherington, J., Blast and Ballistic Loading of Structures.
Butterworth-Heinemann Ltd: Great Britain, 1994.

WIT Transactions on Engineering Sciences, Vol 49


Impact Loading of Lightweight Structures, M. Alves & N. Jones (Editors)
c 2005 WIT Press, www.witpress.com, ISSN 1743-3533

You might also like