Int. J. Impact Engng Vol. 21, No. 10, pp.

827—836, 1998
 1998 Elsevier Science Ltd. All rights reserved
Printed in Great Britain
PII: S0734–743X(98)00034–7 0734—743X/98 $ — see front matter

CRUSHING BEHAVIOUR OF ALUMINIUM

HONEYCOMBS UNDER IMPACT LOADING
HAN ZHAO* and GËRARD GARY
Laboratoire de Mécanique des Solides, Ecole Polytechnique, 91128 Palaiseau, France

Received 12 November 1997; in revised form 27 April 1998

Summary—Understanding the crushing behaviour of honeycombs under dynamic loading is useful

for crash simulations of vehicles and for design of impacting energy absorbers. Available experi-
mental techniques, however, are not always able to provide satisfactory precision for tests on
honeycombs under impact loading. This paper presents a new application of the Split Hopkinson
Pressure Bar (SHPB) for testing honeycombs. Viscoelastic bars are used to improve the accuracy of
measurements, and a generalised two-strain measurement method is applied to obtain a sufficient
measurable maximum crush (up to 80%). Original experimental data (especially in the in-plane
crushing directions) under impact loading are then reported. Differences between quasi-static and
dynamic results are discussed.  1998 Elsevier Science Ltd. All rights reserved

Keywords: Honeycombs; Kolsky’s bar; rate sensitivity; dynamic testing

1. INTRODUCTION
Honeycomb structures are used in various engineering situations which aim, for example, to
improve the weight/strength ratio for applications in the railway, automotive and aircraft
industries [1, 2]. They also are used to absorb energy in accidental impacts. For example,
honeycombs are placed at the front of TGV (high-speed train) locomotives to protect the
survival cell of conductors in accidental shock. Most studies are closely related to those two
major applications.
With regard to weight/strength improvement, interests focus on the mechanical behav-
iour for small deformations (elastic behaviour and failure strength). Analytical and experi-
mental work has been performed under various loading situations because of the
anisotropic nature of honeycombs. The theoretical work is based mainly on a micromecha-
nical analysis to derive global cellular structure response from the study of a single cell [3].
Elastic and fracture models for out-of-plane (axial) crushing [4] (x3-direction), and in-plane
(lateral) crushing [5] (x1 and x2-directions), as well as for transverse shearing [4, 6], have
been developed (see Fig. 1). Related topics such as fracture detection using elastic waves [7],
negative Poisson’s ratio honeycombs [8], and foam-filled honeycombs [9], have been also
reported in the open literature.
For energy absorption, the behaviour of large deformations is desired. Under quasi-static
assumptions, Wierzbicki [10] has developed an out-of-plane large deformation crushing
model that gives an analytical prediction of the crush pressure; Klintworth and Stronge
[11, 12] have formulated a large-deformation behaviour of the in-plane crushing that takes
account of localised deformation band effects.
However, because the energy adsorber works under dynamic loading, it is worthwhile to
study the crushing behaviour under high loading rates. Few theoretical studies have been
performed in this area because of analytical difficulties. Goldsmith and co-workers have
reported some experimental work on out-of-plane crushing [13] and on the ballistic
perforation [14] of honeycombs. They have fired a rigid projectile to a target made of
honeycombs and have shown that mean crushing pressures sometimes increase up to 50%
with respect to the static results, but they have warned that the accuracy of the technique is
not always good. Wu and Jiang [15] have also studied out-of-plane crushing with a similar

* Corresponding author.

827
828 H. Zhao and G. Gary

Fig. 1. Honeycomb configurations.

Fig. 2. Split Hopkinson Pressure Bar arrangement.

experimental technique. Their results confirm significant enhancement of the x3 crushing

strength. Such large differences call for more studies in this domain.
Industry favours numerical crash simulations, which allow for substantial savings in time
and costs. Overall equivalent models for honeycombs have been developed [16] to enable
rapid numerical simulations; nevertheless, the parameters of such models are based on static
experimental data. Accurate experimental data for honeycomb materials under impacting
loading rates are needed to improve the accuracy of these simulations.
This paper presents results of experimental studies on the behaviour of honeycomb
materials under impact loading using the Split Hopkinson Pressure Bar (SHPB, or Kolsky’s
bar) technique [17]. The main advantage of the SHPB is its accuracy in the study of the
constitutive laws of materials at high strain rates. This universal experimental technique
provides nearly constant crushing velocities during the test, which is different to that of
previously mentioned works, in which a small projectile decelerated during the test.
However, a common steel SHPB cannot be used for honeycomb testing because the
honeycomb resistance is quite low (typically, 5 MPa in the x3-direction, 0.2 MPa in the x1-
and the x2-directions) and a large crush (up to 80%) during the test is desired. The
development which enables the application of SHPB to test honeycombs are presented.
Honeycombs are then tested at impacting velocities from 2 to 28 m/s, and original results
are obtained not only in the out-of plane but also in the in-plane directions. Differences with
the quasi-static results are observed and discussed.

2. HONEYCOMB TESTING WITH SHPB

2.1. Split Hopkinson pressure bar test
A typical SHPB set-up is composed of long input and output bars with a short specimen
placed between them (Fig. 2). The impact of the projectile at the free end of the input bar
creates a compressive longitudinal incident wave, e (t). Once it arrives at the bar-specimen

interface, a reflected wave, e (t), is developed in the input bar; a transmitted wave, e (t), is
 
induced in the output bar. From these basic experimental data (incident, reflected and
transmitted waves), forces and velocities at both faces of the specimen can be deduced.
F (t)"S E (e (t)#e (t)), » (t)"C (e (t)!e (t)),


 
 (1)
F (t)"S E e (t), » (t)"C e (t),


 

where E , S , C denote Young’s modulus, cross-sectional area, and wave speed of the bar,




respectively.
 Crushing behaviour of aluminium honeycombs 829

From measurements of the forces and velocities, the mean pressure, p(t), and the
overall crush, *(t) (with the same definition as in the quasi-static case) are calculated as
follows:
p(t)"F (t)/S ,
  (2)
*(t)"  (» (q)!» (q)) dq,
  
where S is the cross-sectional area of the specimen.


2.2. Specific problems in honeycomb testing

The measuring accuracy of SHPB depends on the amplitude of waves in the bars
induced by the resistance of the specimen. Because the honeycomb is very weak (especially
in the x1 direction), the induced strain in an ordinary steel bar is smaller than 10\. In order
to get an accurate measurement, the use of low impedance bars, which are generally
viscoelastic, is proposed [18]. The Nylon bars (Young’s modulus 3.5;10, density
1200 kg/m) are used; they provide an improvement of about 200 times that of a classical
steel bar.
However, the use of viscoelastic bars in a SHPB set-up introduces complications related
to an important wave dispersion effect and requires more difficult calculation of stress and
particle velocity in the bar from the measured strain and a specific projectile length
limitation [18]. The correction of dispersion effects based on the analysis of the wave
propagation in an infinite viscoelastic cylindrical bar [19] is used. Using careful data
processing, the Nylon bar provides satisfactory measuring precision and has been used
successfully in testing fibre-reinforced polymeric composite plates [20].
Another difficulty of the classical SHPB test is achieving a large crush in order to study
the densification part of the response. The SHPB measurement is founded on the propor-
tionality between the strain, stress and particle velocity associated with a single wave
propagation. When the reflection from the ends of bars comes back and superimposes any
incident pulse, this proportionality is lost. The SHPB measuring duration is then limited,
because it uses a short observing window (when the reflection has not return). The
observation duration, *¹, depends on the length of the bars (*¹4¸/C, C being the wave
speed and ¸ the length of the bar). Consequently, the measurable crushing displacement is
limited for a given impact velocity (* 4» *¹). A typical duration for a 3-m long Nylon
 
bar is about 1 ms, so that the maximum observed crushing displacement will not exceed
5 mm for an impact at 5 m/s.
To overcome this measuring limitation, the superimposition of two waves propagating in
opposite directions due to multiple reflections in bars [21] is studied. A two-strain
measurement method has been reported to increase observation duration in elastic bars
[22]. This method can be extended to separate the two elementary waves in the case of
viscoelastic bars provided that correction of wave dispersion is carefully taken into account
[23]. In practice, the strain history in each bar is recorded by two strain gages at distinct
points A and B (Fig. 3). Since the recorded strain is the sum of the contributions of the two
waves propagating in opposite directions, we have
e (t)"e (t)#e (t),
   (3)
e (t)"e (t)#e (t),
 
where e (t), e (t) are waves propagating in the ‘‘ascending’’ direction (artificially de-
 
fined), and e (t), e (t) are those in the ‘‘descending’’ direction at A and B.
 
In order to exhibit ‘‘ascending’’ and ‘‘descending’’ waves from measurements at A and B,
an iterative process is built. For this purpose, strain measurements e (t) and e (t) are divided

into small pieces e (t) and e (t) of a constant time length, *t, as shown in Fig. 3. *t is twice

the time needed by the waves to travel between the two gauges. This operation can be
applied to both virtual elementary ‘‘ascending’’ and ‘‘descending’’ waves at points A and
B to obtain e (t), e (t), e (t) and e (t). As the ‘‘ascending’’ wave at A for the first
   
830 H. Zhao and G. Gary

Fig. 3. Scheme for wave separation.

interval, e (t), is directly measured, the corresponding part of the ‘‘descending’’ wave at

point B, e (t), is calculated. Indeed, on applying Eq. (3), e (t) is the difference between
 
the strain measurement and the ‘‘ascending’’ wave at A shifted to point B.

e (t)"e(t)!e (t). (4)

 
The same process is performed to calculate the ‘‘ascending’’ wave at point A for the next
interval from the knowledge of the ‘‘descending’’ wave given by Eq. (4). In this way, an
iterative formula is constructed, and both ‘‘ascending’’ and ‘‘descending’’ waves can be
calculated for all time intervals. The precision of such an iterative process depends strongly
on that shifting of waves. An accurate propagation theory that takes account of so-called
wave dispersion effects must be then used; this has already been introduced in the SHPB
analysis to improve its accuracy [24].
The use of such an unlimited measuring technique is indispensable for performing tests at
low impact velocities (about 2 m/s). Such tests are usually performed with a fast hydraulic
machine, which requires a suitable measuring technique (often very expensive) to obtain
reliable signals [25].

3. RATE SENSITIVITY OF MECHANICAL BEHAVIOURS OF

HONEYCOMB STRUCTURES
Two Nylon (Pa 6.6) bars (3 m length, 40 mm diameter) were used in our tests on
honeycomb specimens. The two-strain measurement method was applied for tests at 2 m/s.
Simple formula [Eqn. (2)] are used to obtain the pressure-crush curves for all tests because
equilibrium check is satisfactory [26]. Figure 4 shows the input and output forces in
a typical test at 2 m/s.
Cubic specimens (about 36;36;36 mm) are used. Two types of aluminium honeycombs
(Table 1) are tested in the three directions. Because similar results have been obtained, only
the results for honeycomb No 1 are presented. For a given loading rate, the experiment is
repeated at least once. To make figures more legible, only one curve is presented for one
loading rate.
The post-test observations of crushed specimens do not reveal visible differences between
static and dynamic loading. The out-of-plane crushing mode (x3 direction) is a regular
multiple localised folding (Fig. 5a and b-left). For in-plane crushing, a regular folding
is observed in the x1-direction (Fig. 5b-center), whereas irregular patterns are found
in the x2-direction (Fig. 5b-right), similar to those observed by Klintworth and Stronge
[10, 11].
 Crushing behaviour of aluminium honeycombs 831

Fig. 4. Input and output forces for a test on honeycombs.

Table. 1. Characteristics of tested honeycombs

No. Minor cell diameter, S Thickness, h Density

1. 4.7 mm 0.08 mm 130 kg/m

2. 6.2 mm 0.08 mm 100 kg/m

Fig. 5. Post-test observations.

3.1. Out-of-plane behaviour

A summary of experimental data in the x3-direction is shown in Fig. 6. The difference in
the final crush is due to the limitation of the observation duration of standard SHPB
systems (tests at 10 and 28 m/s); hence, the final crush depends on impact velocities.
The significant differences between static and dynamic loading (40%) have been found for
the mean crushing pressure in the x3-direction. However, mean crushing pressures are
nearly the same (about 5.4 MPa) for impact velocities from 2 to 28 m/s. This result differs
from those of Wu and Jiang [15] but agrees with those of Sackman and Goldsmith [14].
The densification point seems independent of the loading rate (at 65% crush).
832 H. Zhao and G. Gary

Fig. 6. Out-of-plane crushing behaviour.

Fig. 7. Reproducibility of testing results (2 m/s in the x3-direction).

A pressure decrease just before the densification points is observed in both the dynamic
and static cases. The three tests at 2 m/s (Fig. 7) show that this decrease is reproducible. This
is somewhat surprising, but its origin is not discussed here. There are no initial pressure
peaks (usually observed for honeycombs), because our supplied specimens have one side
weakly pre-deformed for industrial reasons. This eliminates the initial peak in all our tests in
x3-direction. However, the initial peak does exist, as shown in Fig. 8, if the pre-deformed
short end of the specimen is cut, which corresponds to known theory. Because the main
interest here is the mean crushing strength, this point is not investigated.

3.2. In-plane behaviour

The mean pressure in the x1-direction is nearly the same (0.09 MPa) for dynamic (impact
at 2, 10, 28 m/s) and quasi-static loads (Fig. 9). The densification points at different loading
rates exhibit a spread but are located at about 70% of crush.
The pressure-crush curve in the x2-direction is less regular under dynamic loading (Fig.
10), although the mean pressure is close to that under static loading. This is probably
a consequence of the irregular deformation mode (Fig. 5b). The densification point (about
80% crush) is, again, independent of the loading rate.
 Crushing behaviour of aluminium honeycombs 833

Fig. 8. Discussion on the initial peak.

Fig. 9. In-plane crushing behaviour (x1-direction).

Fig. 10. In-plane crushing behaviour (x2-direction).

834 H. Zhao and G. Gary

4. DISCUSSIONS
A summary of experimental results for the three directions is shown in Fig. 11. (Values in
the out-of-plane x3-direction have a different scale.) Significant differences between static
and dynamic results are only found in the out-of-plane crushing. According to Wierzbicki’s
formula [10], the mean crushing pressure depends only on the flow stress and the so-called
relative cell thickness [Eqn. (5)]. Increases in the mean crushing pressure could, then, be
attributed to a different flow stress under dynamic loading:


h 
p "16.56p (5)
 S

where p , p , h, S are mean crushing pressure, flow stress of foils, cell wall thickness and

minor cell diameter, respectively.
The rate sensitivity of aluminium foils from which the tested honeycombs were made was
then investigated. Static and dynamic tests (with steel SHPB) were performed with
a crushed honeycomb (in the x1-direction and before the full contact of foils). Experimental
curves are given in Fig. 12. The elastic parts are not accurate because of the difficulty in
evaluating initial length at the early test stage. The plastic parts, however, especially the flow
stress level, are reliable. A 10% difference of flow stress between static loading (10\ /s) and

Fig. 11. Summary of test results.

Fig. 12. Rate sensitivity of aluminium foils.

 Crushing behaviour of aluminium honeycombs 835

Table 2. Crushing pressure increase under impact loading

h/S Density Static Dynamic Difference Difference/Density

(mm/mm) (kg/m) (MPa) (MPa) (%) (% /kg/m)

0.08/4.7 130 3.9 5.4 #38 #0.29

0.08/6.2 100 2.8 3.6 #28 #0.28

dynamic loading (600/s) is found. This result indicates that the increase of the flow stress of
aluminium foils is not the major cause of the observed enhancement of the crushing
strength.
Other observations also support such a conclusion. There is no significant enhancement
of the crushing strength in the two in-plane (lateral) directions. This indicates that the large
enhancement in the x3-direction is related to structural effects. In addition, for the two
different tested honeycombs (Table 2), the increase of the crushing strength in the x3-
direction appears to be proportional to the mass density, implying a correlation with
structural inertia.
The factor primarily responsible for this enhancement of crushing strength remains an
open question. Possible factors have been mentioned in theoretical works in the open
literature, for example, a dynamic buckling model of an elastoplastic column [27] has
shown that the buckling mechanism is modified by a stabilisation effect due to lateral
inertia, even when the initial imperfection is significant. A shock model based on structural
inertia effects has successfully explained a similar phenomenon observed in experimental
results of woods [28, 29], and there is also a model based on air trapped in the cells [30].

5. CONCLUSION
The measuring technique presented in this paper provides an original testing method for
honeycombs under impact loading. The use of viscoelastic split bars and the two-strain
measurement method allows accurate measurements of the crushing behaviour and enhan-
ces the understanding of such behaviour in crashworthiness studies involving honeycombs.
Experimental results show that only the out-of-plane crushing behaviour is affected by
the loading rate (up to a 40% increase). The measured difference of the flow stress increase
of the aluminium foil between static and dynamic loadings shows that this flow stress
increase is not primarily responsible for the large enhancement of crushing strength.
A complete explanation of this large enhancement requires further study.

REFERENCES
1. A. Jullien, J. Launay and C. Guillet, Structural sandwiches for railway structures, mechanical test. Me& canique
idustrielle et Mate& riaux 48, 204—207 (1995).
2. I. Nakada and E. Haug, Numerical simulation of crash behaviour of composite structures for automotive
applications. Mate& riaux and ¹echniques, 82, 33—38 (1994).
3. L. J. Gibson and M. F. Ashby, Cellular Solids. Pergamon Press, Oxford, (1988).
4. J. Zhang and M. F. Ashby, The out-of-plane properties of honeycombs. Int. J. Mech. Sci. 34, 475—489 (1992).
5. J. Zhang and M. F. Ashby, Buckling of Honeycombs under in-plane biaxial stresses. Int. J. Mech. Sci. 34,
491—509 (1992).
6. G. Y. Shi and P. Tong, Equivalent transverse shear stiffness of honeycomb cores. Int. J. Solids Struct. 32,
1383—1393 (1995).
7. S. Thwaites and N.H. Clark, Non-destructive testing of honeycomb sandwich structure using elastic waves.
J. Sound »ibration, 187, 253—269 (1995).
8. D. Prall and R. S. Lakes, Properties of chiral honeycomb with a Poisson’s ratio of -1, Int. J. Mech. Sci. 39,
305—314 (1997).
9. C. L. Wu, C. A. Weeks and C. T. Sun, Improving honeycomb-core sandwich structures for impact resistance.
J. Adv. Mater. 26, 41—47 (1995).
10. T. Wierzbicki, Crushing analysis of metal honeycombs. Int. J. Impact Engng 1, 157—174 (1983).
836 H. Zhao and G. Gary

11. J. W. Klintworth and W. J. Stronge, Elasto-plastic yield limits and deformation laws for transversely crushed
honeycombs. Int. J. Mech. Sci. 30, 273—292 (1988).
12. J. W. Klintworth and W. J. Stronge, Plane punch indentation of a ductile honeycomb. Int. J. Mech. Sci. 31,
359—378 (1989).
13. W. Goldsmith and J. L. Sackman, An experimental study of energy absorption in impact on sandwich plates.
Int. J. Impact Engng 12, 241—262 (1992).
14. W. Goldsmith and D. L. Louie, Axial perforation of aluminium honeycombs by projectiles. Int. J. Solids Struct.
32, 1017—1046 (1995).
15. E. Wu and W. S. Jiang, Axial crush of metallic honeycombs. Int. J. Impact Engng 19, 439—456 (1997).
16. H. L. Schreyer, Q. H. Zuo and A. K. Maji, Anisotropic plasticity model for foams and honeycombs. ASCE J.
Engng Mech. 120, 1913—1930 (1994).
17. H. Kolsky, An investigation of the mechanical properties of materials at very high rates of loading. Proc. Phys.
Soc. ¸ondon Ser. B 62, 676—700 (1949).
18. H. Zhao, G. Gary, J. R. Klépaczko, On the use of a viscoelastic split Hopkinson pressure bar. Int. J. Impact
Engng, 19, 319—330 (1997).
19. H. Zhao and G. Gary, A three dimensional analytical solution of longitudinal wave propagation in an infinite
linear viscoelastic cylindrical bar. Application to experimental techniques. J. Mech. Phys. Solids 43, 1335—1348
(1995).
20. H. Zhao and G. Gary, An experimental investigation of compressive failure strength of fibre polymer matrix
composite plates under impact loading. Composite Sci. ¹echnol. 57, 287—292 (1997).
21. J. D. Campbell and J. Duby, The yield behaviour of mild steel in dynamic compression. Proc. Roy. Soc. ¸ondon
A 236, 24—40 (1956).
22. B. Lundberg and A. Henchoz, Analysis of elastic waves from two-point strain measurement. Exp. Mech. 17,
213—218 (1977).
23. H. Zhao and G. Gary, A new method for the separation of waves. Application to the SHPB technique for an
unlimited measuring duration. J. Mech. Phys. Solids 45, 1185—1202 (1997).
24. R. M. Davies, A critical study of Hopkinson pressure bar. Philos. ¹rans. Roy. Soc. A240, 375—457 (1948).
25. A. J. Holzer, A technique for obtaining compressive strength at high strain rates using short load cells. Int. J.
Mech. Sci. 20, 553—560 (1978).
26. H. Zhao and G. Gary, On the use of SHPB techniques to determine the dynamic behavior of materials in the
range of small strains. Int. J. Solids Struct. 33, 3363—3375 (1996).
27. G. Gary, Dynamic buckling of an elastoplastic column. Int. J. Impact Engng 1, 357—375 (1983).
28. S. R. Reid, Response of wood and cellular structures to impact loading. Proc. 1st Int. Symp. Impact Engng
(Edited by I. Maekawa), Sendai, Japan, Vol. 1, pp. 691—698 (1992).
29. S. R. Reid and C. Peng, Dynamic uniaxial crushing of wood. Int. J. Impact Engng 19, 531—570 (1997).
30. P. Seggewiss, Numerical study of aluminium honeycomb structures. Master’s Thesis, Bundeswehr University,
Munich (1996) (in German).