Prediction of the Vibration Levels

Generated by Pyrotechnic Shocks

Using an Approach by Equivalent
Mechanical Shock
This paper is concerned with the numerical simulation of mechanical structures subjected
to pyroshocks. In practice, the methodology is applied on the pyroshock test facility,
which is used by Thales to qualify the electronic equipment intended to be embarked
onboard of spatial vehicles. This test facility involves one plate or two plates linked by
screw bolts. The tested device is mounted on one side while the explosive charge is
David Wattiaux applied on the other side. The main issue of this work is to be able to tune, by simulation,
e-mail: [email protected]
the parameters of the facility (number of plates, material of plate, number of bolts,
amount of explosive, etc.) so as to get the required level of solicitation during the test.
Olivier Verlinden The paper begins by an introduction presenting the state of the art in terms of pyroshock
e-mail: [email protected]
modeling, followed by a description of the shock response spectrum (SRS) commonly
used to represent the test specifications of an embarked equipment. It turns out that there
Calogero Conti is a lack of computational techniques able to predict the dynamic behavior of complex
e-mail: [email protected]
structures subjected to high frequency shock waves such as explosive loads. Three sec-
tions are then devoted to the simulation of the pyrotechnic test, which involves on one
Faculté Polytechnique de Mons,
hand a model of the structure and on the other hand an appropriate representation of the
31 Boulevard Dolez,
impulsive load. The finite element method (FEM) is used to model the dynamic behavior
7000 Mons, Belgium
of the structure. The FEM models of several instances of the facility have been updated
and validated up to 1000 Hz by comparison with the results of experimental modal
analyses. For the excitation source, we have considered an approach by equivalent me-
Christophe De Fruytier chanical shock (EMS), which consists in replacing the actual excitation by a localized
Thales Alenia Space ETCA,
101 Rue Chapelle Beaussart,
force applied on the FEM model at the center of the explosive device. The main origi-
6032 Mont-sur-Marchienne, Belgium
nality of the approach is to identify the amplitude and duration of the EMS by minimizing
e-mail: [email protected]
the gap between the experimental and numerical results in terms of the SRS related to
several points of the facility. The identification has been performed on a simple plate
structure for different amounts of explosive. The methodology is then validated in three
ways. Firstly, it is shown that there is a good agreement between experimental and
numerical SRS for all the points considered to identify the EMS. Secondly, it appears that
the energy injected by the EMS is well correlated with the amount of explosive. Lastly, the
EMS identified on one structure for a given amount of explosive leads to coherent re-
sponses when applied on other structures. A parametric study is finally performed, which
shows the influence of the thickness of the plate, the material properties, the localization
of the EMS, and the addition of a local mass. The different obtained results show that our
pyroshock model allows to efficiently estimate the acceleration levels undergone by the
electronic equipment during a pyroshock and, in this way, to predict some eventual
electrical failures, such as the chatter of electromagnetic relays.
关DOI: 10.1115/1.2827985兴

Keywords: pyroshock, equivalent mechanical shock, shock response spectrum

1 Introduction account because the manufacturers of electronic equipment

thought that the duration of the explosion was too short to damage
Nowadays, the space industry uses more and more pyrotechnic
devices, such as pyrotechnic valves 共Fig. 1兲 or mild detonating the onboard electronic devices. However, Moening 关1兴 has shown
fuses 共Fig. 2兲, to carry out various operations such as separation of that many observed breakdowns on the American launchers were
structural elements 共booster separation, etc.兲, unlocking mecha- caused by the pyroshocks.
nisms 共unfolding solar panels, etc.兲, or activation of onboard op- At present time, the pyroshock resistance of the electronic
erating subsystems 共Table 1兲. The shock wave generated by the equipment is mainly checked experimentally because of the diffi-
blast of these pyrotechnic devices produces severe vibrations in- culties to approach the problem with computational techniques,
side the space shuttles. For several years, the effects of these especially concerning the modeling of pyrotechnic excitation. In
pyrotechnic shocks 共so-called pyroshock兲 have not been taken into practice, simplified resonant fixtures, such as beam or plate as-
semblies, are used to reproduce a vibratory environment equiva-
lent to the actual one.
Contributed by the Technical Committee on Vibration and Sound of ASME for A new test campaign always begins by the choice of an ad-
publication in the JOURNAL OF VIBRATION AND ACOUSTICS. Manuscript received May 22,
2007; final manuscript received November 27, 2007; published online July 14, 2008. equate test facility. A trial-and-error process is applied on the test
Review conducted by Sotirios Natsiavas. assembly, loaded with a dummy of equipment, so as to satisfy the

Journal of Vibration and Acoustics Copyright © 2008 by ASME AUGUST 2008, Vol. 130 / 041012-1

Downloaded 23 Mar 2011 to 210.117.158.180. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 ior of complex structures. FEM and BEM are deterministic meth-
ods for which all parameters of the structure must be known ac-
curately. FEM and BEM have good performances in the low
frequency range. However, these deterministic methods have
some deficiencies at higher frequencies. As the wavelength de-
creases with the frequency, the number of elements has to be
increased in the same way. This makes these methods, at high
frequencies, costly in memory resources, modeling work, and
postprocessing time.
At present time, the statistical energy analysis 共SEA兲 is the
most widely used theoretical framework for the analysis of the
Fig. 1 Pyrotechnic valve „left view: before activation-right dynamic response of complex systems in high frequency range
view: after activation…
关5,6兴. In the SEA method, the structure is modeled as an assem-
blage of discrete subsystems that receive, dissipate, and transfer
test specifications of the launcher. When the desired vibratory en- vibrational energy. The SEA approach is based on two basic hy-
vironment is achieved, the nominal test is performed on the real potheses: the internal dissipation in a subsystem is proportional to
equipment. Obviously, such a procedure is rather inefficient and the subsystem energy and the energy flow between subsystems is
expensive. Consequently, it is useful to develop a mathematical proportional to the difference in modal energy. The main advan-
model of the test facility to predict the vibration levels generated tage of SEA is the small size of the model, which is not related to
by pyroshocks. the excited wavelengths, but only to the number of subsystems;
consequently, the solution of a SEA model can be done at high
2 Overview of the State of the Art frequencies with a low computational cost. However, the major
drawback of SEA is the difficulty in establishing an appropriate
The numerical prediction of the vibration levels generated by model and particularly the choice of subsystems and the evalua-
pyroshocks requires an accurate dynamic model of the test facility tion of the input parameters 共coupling loss factors, internal loss
as well as a mathematical description of the excitation sources.
factors, modal densities, power inputs兲, which are not directly
2.1 Modeling of the Structure. For simple configurations of related to the physical properties that are commonly used in prac-
the test facility, as for example an assembly of plates or beams, tice. Another major drawback of SEA is the loss of information on
several authors, such as Hampton et al. 关3兴 and Sad 关4兴, suggest the spatial distribution of the vibrational energy inside each sub-
discretized quasianalytical models, derived from a continuous system. Finally, the SEA method does not provide the time history
analytical model 共Euler–Bernoulli theory兲. Unfortunately, these of the acceleration.
models are not easy to implement, particularly for more complex Several alternative methods to SEA have been developed in
structures. order to try to overcome its limitations. Bodin 关7兴 and Brevart 关8兴
The finite element method 共FEM兲 and the boundary element present a prediction method of the response of an electronic
method 共BEM兲 are the most conventional predictive tools and are equipment assembly submitted to high frequency shocks. His
widely used in low frequency range to predict the dynamic behav- method combines the use of deterministic calculations and SEA.
The FEM provides the low frequency content of the acceleration
and SEA, coupled to a local random phase reconstruction concept,
provides the high frequency content. This approach is similar to
the one followed by Dalton 关9,10兴 but based on a distinct phase
synthesis. Dalton’s method uses a virtual mode synthesis 共VMS兲
method, which assumes that the modes are distributed over fre-
quency according to the modal density estimation and that these
modes collectively produce the frequency response envelope in
each frequency band. The virtual mode residues are obtained by
comparing the Frequency Response Function 共FRF兲 magnitude of
the virtual system with the one obtained by SEA.
Because of the difficulty to verify the underlying hypotheses
and to give a physical interpretation of the parameters of SEA and
VMS models, we have decided, in this work, to only consider the
deterministic methods to describe the dynamic behavior of the
pyroshock test facility.
2.2 Modeling of the Excitation Sources. The dynamic simu-
lation of mechanical structures requires an accurate knowledge of
the excitation forces. In the case of pyroshocks, these forces are
unknown because they cannot be directly measured. Therefore, it
Fig. 2 Mild detonating fuses „MDFs…
is essential to have access to the applied forces. The most wide-
Table 1 Pyrotechnic applications in astronautics †2‡ spread identification procedures are the inverse methods, such as
the regularization methods which can be applied in the time and
Number of installed frequency domains 关11兴. These methods allow in theory to iden-
Program pyrotechnic devices tify the unknown excitation forces from transient response mea-
surements as far as a dynamic model of the structure is available.
Mercury 46 The drawback of inverse methods is their great sensitivity to mea-
Gemini 139
Saturn ⬇150
surement noise, which can produce an important drift of the iden-
Apollo 共CSM/SLA/LM兲 314 tified forces. Different drift reduction techniques, which consist in
Apollo 共CSM/SLA兲 for Skylab 249 imposing constraints on the force profile, can be found in Ref.
关11兴. Nevertheless, the application of these identification tech-

041012-2 / Vol. 130, AUGUST 2008 Transactions of the ASME

Downloaded 23 Mar 2011 to 210.117.158.180. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 Fig. 3 Single degree of freedom system Fig. 4 Examples of SRS specifications

niques to pyroshocks does not lead to accurate results; the differ-

ence between experiment and simulation being greater than the
tolerances is generally admitted by the equipment manufacturers 3 Test Specifications
关12兴. 3.1 Shock Response Spectrum. The pyroshock specifica-
The investigations of Brossard et al. 关13兴 and Dharaneepathy et tions are generally expressed in terms of the shock response spec-
al. 关14兴 on the effects of air blast on shell structures, such as trum 共SRS兲. The principle of the SRS consists in replacing the real
concrete plates or cooling towers, show that the pressure wave structure by an array of independent single degree of freedom
load can be described by a sine-exponential model depending on systems and to calculate the maximum response of each resonator
two parameters: the angle of the incident shock ␣ and the reduced when its foundation is animated by a motion corresponding to the
parameter ␭, which represents the ratio between the radial dis- shock time history 关18,19兴. The motion of each single DOF sys-
tance R from the center of the explosion and the cube root of the tem 共Fig. 3兲 is characterized by the following classical differential
chemical energy E. Their model has, unlike the inverse methods, equation:
the advantage to take into account the propagation of the pressure
wave along the structure. However, the model is exclusively us- ␦¨ + 2␰␻0␦˙ + ␻20␦ = − ÿ 共1兲
able for reduced parameters ␭ between 0.35 and 12, which lead,
for typical pyrotechnic explosives 共penthrite兲, to distances from where ␦共t兲 = x共t兲 − y共t兲 represents the relative displacement, ␻0
0.07 m to 2 m between the explosive and the structure. Com- = 冑k / m the natural pulsation, and ␰ = c / 2m␻0 the damping ratio,
monly, this condition is not verified during pyrotechnic campaigns which is generally fixed to 5% for pyroshock specifications 关2,20兴.
given that the explosive device is fixed straight on the test facility. The SRS is the plot of the absolute maximum of the relative
In this paper, we suggest an identification procedure of the py- displacement ␦共t兲, denoted by Sd, in relation to the natural fre-
rotechnic excitation using an approach by equivalent mechanical quency f 0 = ␻0 / 2␲.
shock EMS. The EMS method replaces the actual excitation by a The pseudoacceleration spectrum Spa is generally used 关21兴 to
localized force applied on the FE model at the center of the ex- define pyroshock specifications. The pseudoacceleration spectrum
plosive device. The amplitude and duration of the force are tuned Spa is obtained by multiplying the displacement response spectrum
so as to generate equivalent acceleration fields. It is not unreason- Sd by ␻20:
able to envisage such a method given that several investigations
关2,15–17兴 have shown that the use of metal-metal impact devices,
Spa = ␻20Sd 共2兲
such as dropping mass, air gun or pneumatic actuator, allows to
reproduce experimentally the acceleration levels of the far-field The pseudoacceleration Spa has the units of acceleration but it
pyroshocks 共distance between the pyrotechnic device and the does not represent the absolute acceleration of the mass m, except
equipment higher than 15 cm兲. for a zero damping coefficient.

Fig. 5 Examples of typical pyroshock test facilities „Thales Alenia

Space ETCA…

Journal of Vibration and Acoustics AUGUST 2008, Vol. 130 / 041012-3

Downloaded 23 Mar 2011 to 210.117.158.180. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 Fig. 6 View of the pyrotechnic device
Fig. 8 Experimental setup and location of the explosive device

3.2 Examples of Test Specifications. Equipment specifica-

tions are commonly expressed in terms of acceleration SRS speci-
fied from a low frequency limit of a few hundreds of hertz to a
high frequency limit of 10 kHz 共Fig. 4兲; sometimes, this superior 4.2 Shock Devices. The excitation of the test facilities can be
limit can reach 25 kHz. The acceleration levels are generally de- carried out by a mechanical impact or an explosive charge. In the
fined in the three orthogonal directions and can vary widely from case of mechanical shock, the impact device is a pneumatic ac-
one specification to another one. Figure 4 shows some examples tuator, which is screwed on the test fixture 共Fig. 5共b兲兲.
of SRS specifications for the direction perpendicular to the equip- In the case of pyrotechnic shock, the excitation is generated by
ment mating plane. Vibration levels are often more difficult to an explosive device composed of a nonelectrical 共NONEL兲 deto-
achieve for in-plane directions. Nevertheless, a small difference is nator and a detonating cord 共12 g / m penthrite兲 assembled on a
generally admitted between observed and required SRS. A typical thin rectangular aluminum plate with the help of adhesive tip 共Fig.
example of tolerance is given in Ref. 关21兴: 6兲. The detonating cord length can vary from 0 to 1 m depending
• ⫾6 dB for natural frequencies 艋3000 Hz on the desired excitation level. A zero detonating cord length cor-
• +9 dB/ −6 dB for natural frequencies ⬎3000 Hz responds to the single detonator. Figure 7 illustrates the arrange-
ment of the explosive device for different lengths of the cord.
Thales Alenia Space ETCA sometimes considers a tolerance NONEL detonators are used for safety issues but also because
band of ⫾3 dB about a straight-line approximation of the SRS they generate less electromagnetic pulse than electrical detonators.
specifications. This is especially important when operating electronic units are
tested.
4 Thales Alenia Space ETCA Pyroshock Test Facility 4.3 Reference Experimental Data. The test fixture is a
4.1 Test Facility Description. The pyroshock test facilities square steel plate 共1 ⫻ 1 ⫻ 0.015 m3兲, which is vertically sus-
used by Thales Alenia Space ETCA 共Charleroi-Belgium兲 are reso- pended to a tubular structure with steel cables 共Fig. 8兲. The explo-
nant fixtures, which can be excited either by an explosive charge sive charge is fixed on the center of the plate by means of adhe-
or by a mechanical impact 共dropping mass, sledgehammer, air sive tip. We have considered lengths of the detonating cord
gun, or pneumatic actuator兲 关2,20兴. The resonant fixtures are made varying from 0 cm to 50 cm. The accelerations are measured with
of steel or aluminum plates supported by a tubular structure with piezoelectric sensors, which are directly screwed on the steel
steel cables. Different configurations can be considered: simple plate. Their location is given in Fig. 9. The signals of the accel-
plate or double plate linked by screw bolts either in vertical or in erometers are finally filtered by a low pass electronic filter with a
horizontal orientation. The tested item is screwed on the test fa- cutoff frequency of 10 kHz.
cility and is submitted to the shock wave generated by the exci- Figure 10 gives a summary of the main experimental observa-
tation and to the response of the fixture. Figures 5共a兲 and 5共b兲 tions.
show two classical configurations of the test facility used by Tha-
• The energy of a pyroshock is mainly injected perpendicu-
les Alenia Space ETCA.
larly to the plate 共Fig. 10共a兲兲.
Among the parameters influencing the SRS, the most important
• The maximum amplitude of the SRS increases with the
ones are the following:
length of the detonating cord 关Fig. 10共b兲兴.
• the configuration of the test facility 共number of plates and • Acceleration levels weakly vary with the location of the
their orientation in space兲 sensor 共Fig. 10共c兲兲.
• material, area, and thickness of the plates • The repeatability between two pyroshocks, which are real-
• the intensity of the shock ized in the same conditions, is satisfactory; the mean differ-
• the location of the shock device ence between the two SRS is below 2 dB 共Fig. 10共d兲兲.

Fig. 7 Arrangement of explosive device for different lengths of the detonating cord

041012-4 / Vol. 130, AUGUST 2008 Transactions of the ASME

Downloaded 23 Mar 2011 to 210.117.158.180. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 Table 2 Physical characteristics of the plate

Steel plate

E 2.02E11 N / m2
␳ 7800 kg/ m3
M 117 kg
␯ 0.3
h 0.015 m
f 10 kHz
␭ 0.121 m

this same range.

It is recommended to define an element size that is lower than
the smallest bending wavelength. In an infinite uniform plate, the
wavelength ␭ of the bending waves is given by 关22兴
Fig. 9 Location of the explosive device and piezoelectric sen-
sors „circles…
␭= 冑 冉 冊
2␲ D
f Ms
1/4
共3兲

5 Finite Element Model

Eh3
5.1 Description. The model of the test facility has been built D= 共4兲
12共1 − ␯2兲
under a classical FE software 共ANSYS 8.1兲 and has been validated
and updated from an experimental modal analysis. 3D structural where M s is the surface mass density, E Young’s modulus, ␯ Pois-
solid elements 共SOLID45兲 have been used for the plate. The ele- son’s coefficient, h the plate thickness, and f the frequency of the
ment size is a significant parameter in the simulation and is cho- bending wave. Table 2 gives the geometric and physical charac-
sen according to the analyzed frequency range. As the SRS speci- teristics of the studied plate. The bending wavelength ␭ corre-
fications are generally defined in the range 关0 – 10 kHz兴, the FEM sponding to a frequency of 10 kHz is equal to 0.121 m.
should be able to describe the dynamic behavior of the plate in In the plane of the plate, a 56⫻ 56 element grid has been de-

Fig. 10 Main experimental observations

Journal of Vibration and Acoustics AUGUST 2008, Vol. 130 / 041012-5

Downloaded 23 Mar 2011 to 210.117.158.180. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 Table 3 Correspondence between experimental and numerical
modal characteristics

f E 共Hz兲 f S 共Hz兲 ⌬k 共%兲 MAC

47 49 3.4 0.98
92 89 3.7 0.85
124 127 2.5 0.69
231 224 2.4 0.88
282 282 0.6 0.98
457 445 2.6 0.83
488 477 2.2 0.77
493 498 1.2 0.84
564 560 0.73 0.98
622 635 2.1 0.80
738 730 1.0 0.64
790 771 2.5 0.85
Fig. 11 Definition of the EMS
796 803 0.9 0.73
897 919 2.4 0.68
898 899 0.1 0.66
ing the pyroshock as, for example, the interaction between the
shock wave, generated by the explosion, and the geometry of the
room, which controls the numerous reflection waves.
The EMS corresponds to the mechanical force that has to be
fined in order to respect a number of six elements per wavelength
applied to the FE model to obtain equivalent acceleration levels.
up to a bending wave of 10 kHz. We have defined three elements
In our case, the force is applied on the node corresponding to the
in the direction of the thickness of the plate.
center of the explosive charge. Although the pyroshock is a three-
5.2 Model Validation. A model validation has been realized dimensional excitation source, we have considered an EMS acting
by comparing the modal properties of the test facility deduced only in the direction perpendicular to the plate because the energy
from the model and those experimentally identified from mea- is mainly injected in this direction 共Fig. 10共a兲兲. Different impact
sured frequency response functions. The frequency response func- profiles can be used to describe the shock 共rectangular, half sine,
tions Hij共␻兲 have been measured in the direction perpendicular to versed sine, etc.兲. For a given impact duration ␶, the shape of the
the plate, with the help of an impact hammer, and in the frequency excitation does not influence the SRS calculations as far as the
range 关0 – 1000 Hz兴 with a frequency resolution of 0.625 Hz. The integral 兰␶0Fdt, which represents the energy injected in the system,
modal characteristics have been identified with EasyMod, which is constant 共Fig. 12兲. In our work, we will consider only triangular
is a MATLAB toolbox of modal analysis developed by the Depart- symmetrical profiles equivalent to those observed during a ham-
ment of Theoretical Mechanics of the Facult Polytechnique de mer impact. Consequently, the EMS is completely defined by two
Mons 关23兴. The Least Square Complex Exponential 共LSCE兲 关24兴 parameters 共Fig. 11兲:
method has been used to identify the resonant frequencies f k, • the intensity Fmax of the impact
damping factors ␰k, and modal vectors 兵␺k其. • the duration ␶ of the impact
The finite element model was updated from the experimental
results by minimizing the relative difference between natural fre- The parameters Fmax and ␶ of the EMS are deduced by an
quencies calculated as follows 关24兴: optimization process that minimizes the difference between ex-
兩f Ek − f Sk 兩 perimental and simulated SRS:
⌬k = 共5兲 10 kHz NSRS
f Sk
where the superscripts E and S are used for experimental and
⑀ = min 兺 兺 兩SRS
Fmax,␶ f=1000 Hz j=1
measured
j − SRSsimulated
j 共Fmax, ␶兲兩2 共7兲
model data, respectively.
The experimental and numerical mode shapes are matched from where SRSmeasured
j and SRSsimulated
j represent the shock response
the modal assurance criterion 共MAC兲 defined by 关24兴 spectrum at node j of the measured and simulated accelerations,
respectively. NSRS is the number of measurement points on the
共兵␺Ek 其T兵␺Sk 其兲2 plate 共Fig. 9兲 and f the frequency.
MACk = 共6兲
共兵␺Ek 其T兵␺Ek 其兲共兵␺Sk 其T兵␺Sk 其兲 Such a procedure is laborious because of the calculation time:
for each couple 共Fmax , ␶兲, it is necessary to evaluate the accelera-
where 兵␺E其 and 兵␺S其 denote the experimental and simulated
tion fields with the FE model. To simplify the optimization pro-
modal vectors, respectively.
cedure, we have considered durations varying from
Table 3 summarizes the first 15 experimental and analytical
modes, correlated with a MAC value greater than 0.6 and a rela- 20 ␮s to 200 ␮s by discrete steps of 20 ␮s. For each duration ␶, a
tive frequency gap lower than 10%. These results allow to validate reference SRS corresponding to a shock with a unitary intensity
the FE model up to 1000 Hz. Mode shapes at higher frequencies has been calculated and the optimal intensity Fmax minimizing the
are much more difficult to identify due to the high modal density. difference between the experimental and simulated SRS has been
Nevertheless, we have assumed that it can be extrapolated at determined by optimization:
higher frequencies 共until 10 kHz兲. 10 kHz NSRS

⑀共␶兲 = min 兺 兺 兩SRS

Fmax f=1000 Hz j=1
measured
j − j 兩
FmaxSRSref 2
共8兲
6 Equivalent Mechanical Shock
The parameters of the EMS correspond to the couple 共Fmax , ␶兲
6.1 Definition. The computations of the pyroshock response for which the function error ⑀ is minimal.
require a dynamic model of the test facility and a mathematical Let us mention that in low frequencies, the pyroshock measure-
description of the excitation sources. The latter is the main diffi- ments are generally perturbed by zero-shift problems: A decreas-
culty in the modeling because it cannot be properly measured. ing exponential, originating from electric effects, adds to the real
Moreover, a lot of complex physical phenomena can appear dur- acceleration. An error of some percent in the time domain gives

041012-6 / Vol. 130, AUGUST 2008 Transactions of the ASME

Downloaded 23 Mar 2011 to 210.117.158.180. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 Fig. 12 Influence of the shape of the excitation on the SRS

more than one order of magnitude of error in the SRS at low 7 Model Validation
frequencies. In the literature, several authors, such as Smallwood
and Cap 关25兴, suggest different curve fitting techniques to correct 7.1 Agreement Between Experimental and Numerical
the original data. Neverthless, it is important to underline that Shock Response Spectrum. In order to verify and quantify the
these methods must be employed carefully because they can lead correspondence between experimental and simulated SRS, we
to unphysical corrections if they are applied without sufficient have used some statistical indicators:
engineering judgment. That is why, although we have applied a • ⌬i共f兲, which represents the difference at frequency f be-
zero-shift corrections, the frequency range begins at 1 kHz in Eqs.
tween experimental and simulated SRS in terms of fre-
共7兲 and 共8兲.
quency for node number i:
⌬i共f兲 = 兩SRSisimulated共f兲 − SRSimeasured共f兲兩 共9兲
• ␮共⌬i兲 and ␴共⌬i兲, which correspond to the mean and the
standard deviation, respectively, of the indicator ⌬i共f兲 along
the frequency range:

兺 f 兩⌬i共f兲兩
␮共⌬i兲 = 共10兲
N

␴共⌬i兲 = 冑兺 1
N f
共⌬i共f兲 − ␮共⌬i兲兲2 共11兲

where N represents the number of samples of the frequency

vector.
• ␮G and ␴G relate to the mean and the standard deviation,
respectively, of the frequency difference between experi-
mental and simulated SRS considered on the whole set of
measured nodes:

␮共⌬i兲
Fig. 13 Evolution of the product Fmax␶ in relation to the length
␮G = 兺N i SRS
共12兲

of the explosive cord

Table 4 Characteristics of the reference EMS

Length Fmax 共N兲 ␶ 共␮s兲 Fmax ␶ 共N s兲 ␮G 共dB兲 ␴G 共dB兲

0 83,518 60 5.01 0.82 0.64

4 cm 129,830 60 7.79 0.76 0.57
10 cm 203,980 60 12.24 0.88 0.68
20 cm 199,260 80 15.94 0.84 0.67
30 cm 191,210 100 19.12 1.32 1.83
50 cm 240,870 100 24.09 1.27 1.16

Journal of Vibration and Acoustics AUGUST 2008, Vol. 130 / 041012-7

Downloaded 23 Mar 2011 to 210.117.158.180. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 Table 5 Accuracy between experimental and simulated SRS ing ratio of 0.1%, which corresponds to the mean value measured
experimentally in the frequency range 关0 – 1000 Hz兴. The rms
0 cm 30 cm
value of the acceleration in each 1 / 3 octave band is relatively well
Length cord ␮共⌬i兲 ␴共⌬i兲 ␮共⌬i兲 ␴共⌬i兲 reproduced.

Node 1 1.30 0.89 3.13 4.02 7.2 Application to Other Configurations. We have studied
Node 14 0.70 0.52 0.86 0.61 in detail three other configurations of the test facility.
Node 11 0.67 0.44 0.83 0.54 Configuration 1. Steel plate 共1 ⫻ 1 ⫻ 0.015 m3兲 on which an
Node 6 0.76 0.48 1.15 0.94 aluminum block of 15 kg is screwed, simulating an electronic
Node 15 0.67 0.53 0.88 0.65
Node 12 0.70 0.66 0.83 0.63
device 共Fig. 16共a兲兲.
Node 5 0.91 0.65 1.58 0.98 Configuration 2. Double plate facility composed by a square
steel plate 共1 ⫻ 1 ⫻ 0.015 m3兲 and by a rectangular aluminum
plate 共0.8⫻ 0.6⫻ 0.006 m3兲, the two plates are linked by screw
bolts 共Fig. 16共b兲兲.

␴G = 冑 1
NSRSN 兺 兺 共⌬ 共f兲 − ␮
i f
i G兲
2
共13兲
Configuration 3. Same configuration as the previous item but
where a dummy of electronic equipment 共⬇4 kg兲 has been added
at the center of the aluminum plate 共Fig. 16共c兲兲.
For each of the three configurations, we have developed a FE
For the whole set of experimental data that we have previously model, which has been validated and updated from a modal analy-
described, we have identified the EMS for each length of the sis in the frequency range 关0 – 1000 Hz兴. The plates have been
detonating cord. modeled by 3D structural solid elements 共SOLID45兲 and the
Table 4 summarizes the characteristics of the EMS for the con- screw bolts by beam elements 共BEAM4兲 with equivalent geomet-
sidered excitation levels and gives the values of statistical indica- ric properties.
tors ␮G and ␴G. Whatever the length of the explosive cord, the The same pyroshock excitation 共length of detonating cord of
EMS reproduces in a satisfactory way, in terms of SRS, the dy- 0 cm兲 has been applied on all configurations and the EMS has
namic behavior of the plate; the mean frequency difference ␮G is been identified. Table 6 gives the characteristics of the identified
widely below the tolerances that are generally admitted by the
EMS from the different studied configurations and the statistical
equipment manufacturers. The mean frequency difference ␮共⌬i兲
indicators ␮G and ␴G. On average, the duration of the impact is of
and the standard deviation ␴共⌬i兲 calculated at seven measured
about 80 ␮s, which is coherent with the typical values found in
nodes are given in Table 5 for detonating cord lengths of 0 cm and Refs. 关4,26兴. Although the intensity of the impact varies from one
30 cm. Figures 14共a兲 and 14共b兲 show some comparisons between
configuration to another one, the product Fmax␶, which represents
experimental and simulated SRS. The evolution of the product
an image of the energy injected in the system, is close to 5 N s
Fmax␶ in relation to the length of the explosive cord is given in
whatever the EMS. In order to compare the different EMS, we
Fig. 13.
Although the SRS is the most frequently used tool to quantify a have applied the reference EMS 共␶ = 60 ␮s and Fmax = 83,518 N兲
vibratory environment, the comparison between experimental and to the three configurations. Table 7 gives the global deviation
simulated SRS is not a sufficient criterion to validate the EMS between simulated and measured SRS for each configuration.
model because different acceleration profiles can lead to the same The agreement between experience and simulation is less sig-
SRS 关15兴. Consequently, it is essential to make sure that the model nificant for the double plate Configurations 共Configurations 2 and
allows to reproduce also the experimental acceleration fields. 3兲; the mean difference ␮G is higher than 3 dB. Nevertheless, the
Figures 15共a兲 and 15共b兲 represent at Nodes 1 and 6 the time vibration levels of the steel plate, on which the excitation source is
history and the 1 / 3 octave band spectrum of the experimental and applied, are very well reproduced 共Fig. 17共a兲兲. The model has
simulated acceleration fields, respectively. The modal superposi- some difficulties to reproduce the acceleration levels of the alu-
tion method has been used to predict the transient response of the minum plate, particularly at high frequencies 共Fig. 17共b兲兲. This
structure. We have introduced in our FE model a constant damp- observation can be explained by the filter effect of beam elements

Fig. 14 Comparison between experimental and simulated SRS

041012-8 / Vol. 130, AUGUST 2008 Transactions of the ASME

Downloaded 23 Mar 2011 to 210.117.158.180. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 Fig. 15 Comparison between experimental and simulated acceleration fields

Fig. 16 The other analyzed configurations

Table 6 Characteristics of the EMS for more complex configurations

Fmax 共N兲 ␶ 共␮s兲 Fmax␶ 共N s兲 ␮G 共dB兲 ␴G 共dB兲

Configuration 1 58,580 80 4.69 2.43 1.62

Configuration 2 89,092 60 5.34 3.36 2.35
Configuration 3 65,500 80 5.24 3.35 2.88

Journal of Vibration and Acoustics AUGUST 2008, Vol. 130 / 041012-9

Downloaded 23 Mar 2011 to 210.117.158.180. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 Table 7 Application of the reference EMS to other used to describe the dynamic behavior of the screw bolts. In the
configurations framework of this work, we have not verified this assumption.
␮G 共dB兲 ␴G 共dB兲

Configuration 1 2.74 2.19 8 Parametric Analysis

Configuration 2 3.47 2.40 The model can be exploited to estimate the influence of some
Configuration 3 4.09 3.07 parameters such as the localization of the explosive device, the
geometric and physical characteristics of the plate, or the addition
of localized masses.

Fig. 17 Comparison between experimental and simulated SRS-double plate in vertical configuration

Fig. 18 Influence of some operating parameters

041012-10 / Vol. 130, AUGUST 2008 Transactions of the ASME

Downloaded 23 Mar 2011 to 210.117.158.180. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm
 The acceleration levels vary significantly with the thickness 关5兴 Keane, A. J., and Price, W. G., 2005, Statistical Energy Analysis: An Overview,
With Applications in Structural Dynamics, Cambridge University Press, En-
共Fig. 18共a兲兲 and the material properties of the plate 共Fig. 18共b兲兲; gland.
an increase of about 10 dB is obtained when the thickness is re- 关6兴 De Langhe, K., 1996, “High Frequency Vibrations: Contributions to Experi-
duced by half or when the steel plate is replaced by an aluminum mental and Computational Sea Parameter Identification Techniques,” thesis,
plate. It also turns out that the vibrations are slightly influenced by Katholieke Universiteit Leuven 共KUL兲, Leuven, Belgium.
关7兴 Bodin, E., 2001, “Comportement Dynamique d’un Quipement Lectronique
the location of the EMS 共Fig. 18共c兲兲. Soumis des Chocs Mcaniques ou Pyrotechniques,” thesis, Université de Tech-
As illustrated in Fig. 18共d兲, the addition of a judiciously local- nologie de Compigne, Compigne, France.
ized mass 共10 kg兲 allows to lessen the influence of some reso- 关8兴 Bodin, E., and Brévart, B., 2004, “Pyrotechnic Shock Response Predictions
nance peaks. In this example, the SRS are calculated from the Combining Statistical Energy Analysis and Local Random Phase Reconstruc-
tion,” J. Acoust. Soc. Am., 112共1兲, pp. 156–163.
time history of the acceleration simulated at Node 6. 关9兴 Dalton, E. C., “Ballistic Shock Response by an Extension of Statistical Energy
Analysis,” Proceedings, 63rd Shock and Vibration Symposium, 1992.
9 Conclusions 关10兴 Dalton, E. C., 1999, “Overview of the High Frequency Shock Problem in
Aerospace,” Huntsville, AL, coursebook.
This paper has presented some pyroshock test facilities that are 关11兴 Hadjit, R., 2001, “Methodes Inverses Adaptes l’Identification de Forces
commonly used by Thales Alenia Space ETCA to check the py- d’Excitation de Structures Mecaniques,” thesis, Faculté Polytechnique de
roshock resistance of electronic units dedicated to space applica- Mons 共FPMs兲, Mons, Belgium. http://mecara.fpms.ac.be.
tions. Different configurations of the test facility have been stud- 关12兴 Algrain, H., Hadjit, R., and Wattiaux, D., 2003, “Etude sur les Chocs Pyro-
techniques dans le Domaine du Spatial,” Facult Polytechnique de Mons
ied and modeled by the FE method. The FE model has been 共FPMs兲, Technical Report, Projet Région Wallonne-Convention R & W PYR
updated and validated at low frequencies from an experimental 99/013.
modal analysis. 关13兴 Brossard, J., Desrosier, C., Purnomo, H., and Renard, J., 1995, “Pressure
An approach by EMS has been used to identify the excitation Loads on a Plane Surface Submitted to an Explosion,” Proceedings of the 19th
International Symposium on Shock Waves, Marseille, France.
sources generated by pyrotechnic explosions. This approach con- 关14兴 Dharaneepathy, M. V., Keshava, R. M. N., and Santhakumar, A. R., 1995,
sists in replacing the actual excitation by a localized force applied “Critical Distance for Blast-Resistant Design,” Comput. Struct., 54共4兲, pp.
on the FE model at the center of the explosive device. 587–595.
From a simplified configuration of the pyroshock test facility 关15兴 Eriksson, J., 1999, “Measuring and Analysis of Pyrotechnic Shock,” thesis,
共simple plate suspended vertically兲, the EMS has been identified Chalmers University of Technology 共CTH兲, Gothenburg.
关16兴 Bai, M., and Thatcher, W., 1979, “High Pyrotechnic Shock Simulation Using
for several excitation levels. This identification procedure pro- Metal to Metal,” The Shock and Vibration Bulletin, 49共1兲, pp. 96–100.
vides accurate results; the average differences between experi- 关17兴 Sutra, M., Combes, B., Berlioz, A., and Mesnier, D., 2005, “Dveloppement
mental and simulated SRS are below the tolerances that are gen- d’une Dmarche de Simulation des Quipements Spatiaux Soumis des Chocs
erally admitted by the equipment manufacturers. Pyrotechniques,” Proceedings of the 17me Congrs Franais de Mcanique,
Troyes, France, Sep.
Our pyroshock model allows to predict the influence of several 关18兴 Irvine, T., 2002, “An Introduction to the Shock Response Spectrum,” www.vi-
operating parameters of the test facility and in this way it can brationdata.com.
orientate the experimental procedure of the pyroshock testing. 关19兴 Lalanne, C., 1999, Chocs Mcanique, Vibrations et Chocs Mcaniques Vol. 2,
We have applied our methodology to more complex test facili- Hermes, Paris. http://www.hermes-science.com.
关20兴 Filippi, E., Cambier, F., and Conti, C., 1998, “Development of the Alcatel Etca
ties. This analysis has shown that, for a given amount of explosive Pyroshock Test Facility,” Proceedings of European Conference on Spacecraft
charge, similar EMSs have been identified when different configu- Structures, Materials and Mechanical Testing, Stadthalle Braunschweig, Ger-
rations are used. Consequently, the approach by EMS is promising many, Nov.
to estimate the acceleration levels undergone by the electronic 关21兴 Mulville, D. R., 1999, “Pyroshock Test Criteria,” NASA Technical Standard
NASA-STD-7003, National Aeronautics and Space Administration 共NASA兲,
equipment during a pyroshock and, in this way, to predict some
May, http://standards.nasa.gov.
eventual electrical malfunctions, such as the chatter of electro- 关22兴 Gorman, D. J., 1982, Free Vibration Analysis of Rectangular Plates, Elsevier,
magnetic relays 关27兴. Paris.
关23兴 Kouroussis, G., and Wattiaux, D., 2006, Toolbox Matlab Pour l’Analyse
Modale Exprimentale, Facult Polytechnique de Mons 共FPMs兲, Mons, Belgium,
References http://mecara.fpms.ac.be.
关1兴 Moening, C. J., 2001, “View of the World of Pyrotechnic Shock,” Shock and 关24兴 Maia, N., Theoretical and Experimental Modal Analysis, Research Studies
Vibration bulletin, 56共3兲, pp. 3–28. Press, England.
关2兴 Filippi, E., Attouoman, H., and Conti, C., 1999, “Pyroshock Simulation Using 关25兴 Smallwood, D. O., and Cap, S. C., 1999, “Salvaging Pyrotechnic Data With
the Alcatel Etca Test Facility,” Proceedings of the first European Conference Minor Overloads and Offsets,” Journal of the Institute of Environmental Sci-
on Launcher Technology, Toulouse, France, Dec., CNES. ences and Technology, 42共3兲, pp. 27–35.
关3兴 Hampton, E., Nygren, P., and Li, H., 2006, “Analytical Shock Response of a 关26兴 Derumaux, M., 2005, “Sur la Modlisation et la Simulation de Liaisons
Transversely Point-Loaded Linear Rectangular Plate,” Proceedings of the Soumises à des Chocs Pyrotechniques,” thesis, Ecole Normale Supérieure de
Eighth Biennial ASME Conference on Engineering Systems Design and Analy- Cachan, Cachan, France.
sis (ESDA2006), Torino, Italia, Jul. 关27兴 Wattiaux, D., Conti, C., and Verlinden, O., 2006, “Prediction of the Dynamic
关4兴 Sad, D., 1998, “Etude Thorique et Numrique des Vibrations de Structures Behaviour of Electromagnetic Relays Submitted to Mechanical Shock,” Pro-
Soumises des Chocs Pyrotechniques,” thesis, Ecole Normale Suprieure de ceedings of the Eighth Biennial ASME Conference on Engineering Systems
Cachan, Cachan, France. Design and Analysis (ESDA2006), Torino, Italia, Jul.

Journal of Vibration and Acoustics AUGUST 2008, Vol. 130 / 041012-11

Downloaded 23 Mar 2011 to 210.117.158.180. Redistribution subject to ASME license or copyright; see http://www.asme.org/terms/Terms_Use.cfm