Composites: Part A 90 (2016) 489–501

Contents lists available at ScienceDirect

Composites: Part A
journal homepage: www.elsevier.com/locate/compositesa

Effect of fiber orientation on energy absorption characteristics of glass

cloth/epoxy composite tubes under axial quasi-static and impact
crushing condition
Dayong Hu a,b,c, Chong Zhang a, Xiaobing Ma d,⇑, Bin Song a
a
Department of Aircraft Airworthiness Engineering, School of Transportation Science and Engineering, Beihang University, Beijing 100191, China
b
Aircraft/Engine Integrated System Safety Beijing Key Laboratory, Beijing 100191, China
c
Advanced Vehicle Research Center (AVRC), Beihang University, Beijing 100191, China
d
School of Reliability and Systems Engineering, Beihang University, Beijing 100191, China

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

Article history: The collapse characteristics and energy absorption capability of composite tubes made of 759/5224
Received 19 May 2016 woven glass cloth/epoxy with different fiber orientations were studied in the present article under axial
Received in revised form 29 June 2016 quasi-static and impact crushing condition. The effects of fiber orientation and loading condition on the
Accepted 17 August 2016
crushing modes and energy absorption capability were discussed in detail. The fiber orientation could be
Available online 20 August 2016
found to have significant influences on energy absorption performance. Based on results, the energy
absorption capability could be improved by selecting proper fiber orientation. The energy absorption
Keywords:
capability in impact crushing tests could be found to be slightly lower than that in quasi-static crushing
Impact crushing
Crushing mode
tests.
Specific energy absorption Ó 2016 Elsevier Ltd. All rights reserved.
Crashworthiness

1. Introduction considered that this material was very suitable for lightweight
crashworthiness design, taking into account the merits of full recy-
Helicopters are widely used in civil and military fields. As heli- clability and shorter processing times. Esnaola et al. [10] had per-
copters usually fly at a low altitude, they are extremely vulnerable formed quasi-static crushing tests to study semi-hexagonal
to the atrocious weather condition or ground attack, therefore composite structures with different stacking sequences of E-glass
leading to a very high probability of serious and catastrophic crash and basalt fibers. They found that specimens with unidirectional
accidents. Structure crashworthiness, which means that the struc- and bidirectional E-glass fibers could achieve the highest SEA,
ture should protect occupants from injuries in the case of poten- whereas specimens with the hybrid configuration using E-glass
tially survivable crash accidents, has become one of the most and basalt fiber/polyester showed lower SEA. Hu et al. [6] had
important requirements in the helicopter design [1–3]. In order investigated the hybrid effect of two kinds of brittle fibers on
to improve the crashworthiness performance without sacrificing energy absorption characteristics. It was shown that the energy
too much weight, composite materials with excellent SEA are very absorption performance could be optimized by selecting a suitable
attractive for lightweight crashworthy structures, compared to hybrid ratio of two kinds of brittle fibers. Ma et al. [11] studied
conventional metallic materials [4–8]. hybrid carbon-aramid/epoxy composite tubes with different tem-
A considerable amount of literature has been carried out to perature treatments. It was found that the optimum SEA of near
investigate many influence factors on the energy absorption char- 100 J/g could be achieved by optimizing hybrid method, ratio,
acteristics of composite materials, such as fiber and matrix prop- and reasonable geometry shape. Liu et al. [12] performed axial
erty, hybridization, trigger mechanism, fiber orientation, layup, quasi-static tests to investigate the failure mode and energy
geometry dimension, and filler. For example, Boria et al. [9] inves- absorption of square carbon-fiber reinforced plastic (CFRP) tubes.
tigated a new material of full thermoplastic composite, where both They considered that wall thickness was a critical parameter
the reinforcement and matrix were made in thermoplastic. They affecting the failure mode and energy absorption capability, lead-
ing to an increasing trend in Pmax and SEA. Trigger mechanism
⇑ Corresponding author. effects also had been analyzed in order to obtain progressive col-
E-mail address: [email protected] (X. Ma). lapse modes [13,14].

http://dx.doi.org/10.1016/j.compositesa.2016.08.017
1359-835X/Ó 2016 Elsevier Ltd. All rights reserved.
490 D. Hu et al. / Composites: Part A 90 (2016) 489–501

Nomenclature

Pmax peak load, maximum load of the load-displacement M mass of the specimen
curve after crushing initiated m mass of the destroyed part of a specimen, m = MDL/L
L initial length of the specimen. SEA specific energy absorption, SEA = AE/m
DL the destroyed length of specimens q tube density
AE total absorbed energy, measured by integrating the A cross-sectional area of tube
load-displacement curve t tube wall thickness
Pm mean crush load, Pm = AE/DL h fiber orientation with respect to the axis direction of
SSCS specific sustained crushing stress, SSCS = Pm/qA, equal to tube specimen
SEA D tube diameter
CLE crush load efficiency, CLE = Pm/Pmax

In the view of composite structure design, fiber orientation is 0–100 s 1. It was shown that the energy absorption was found to
considered as one of the most important design parameters. So it be independent of strain rate. They attributed the reason to the
is of great interest to obtain the trend between the energy absorp- energy absorption mechanism of fiber-dominated fracture, which
tion characteristics and h. Mahdi et al. [15] studied the effect of h was independent of strain rate within the studied strain rate range.
on the energy absorption of E-glass/epoxy composite tubes. It A more complex variation trend in SEA could be found in the liter-
was shown that tubes with h of 15°/ 75° or 75°/ 15° had the high- ature [17], in which for specimens with h of 15° and 30°, SEA of
est energy absorption capability. Hull [16] investigated the effect of impact tests was lower than that of quasi-static tests, whereas
h on energy absorption behavior of filament-wound glass/polyester for specimens with h higher than 45°, SEA of impact tests was
tubes with h in the range from 35° to 90°. It could be observed that higher than that of quasi-static tests. The reason could be attribu-
SSCS increased with increasing h from 35° to 65° and then ted to the failure mechanisms. For Mode I, the energy dissipation
decreased with further increasing h. The reason could be attributed mechanism was controlled by matrix cracking, which could prop-
to the change in crushing modes: tubes with h in the range from agate more easily due to impact loading, resulting in a decrease of
35° to 55° failed in splaying crushing mode (Mode I), whereas SEA, while for fragmentation crushing mode, it was controlled by
tubes with h higher than 65° failed in transverse shearing or frag- fiber/matrix fracture mechanisms and the strain rate effects might
mentation crushing mode. These results were similar to those by result in material stiffness augmentation, which was a chief contri-
Song et al. for filament-wound glass/epoxy tubes [17]. Ochelski bution to causing the increase of SEA [17]. These conflicting results
and Gotowicki [18] had studied the energy absorption capability mentioned above show that energy absorption characteristics of
of carbon/epoxy and glass/epoxy composites conical tubes. How- axially crushed composite tubes are still insufficient and not thor-
ever, their results revealed that increasing h could lead to an oughly understood. Therefore, more efforts should be paid on an
increase in the energy absorption performance due to the fibers in-depth study to enrich the scientific knowledge on this area,
oriented in the circumference direction and a decrease in the num- which is one of the motivating factors behind the present study.
ber of interlaminar crack. Daniel et al. [19] had investigated the The primary purpose of present work is to experimentally
effect of h on SSCS of E-glass/ polyester laminate with h in the range investigate the effect of h on the energy absorption characteristics
from 15° to 75°. Their results were comparable in magnitude close of composite tubes made of 759/5224 woven glass cloth/epoxy
to literature data of Hull [16], but no significant trend could be under axial quasi-static and impact crushing condition in order
obtained. to provide an effective reference for the design of helicopter crash-
Most of the previous studies on the energy absorption charac- worthy subfloor structure. The schematic diagram of crashworthy
teristics of composites are carried out under the quasi-static crush- subfloor structure can be seen in Fig. 1, where the composite tubes
ing condition, which is simple and easy to control and can obtain are used as primary energy absorption units for this crashworthy
the main failure feature of crushing characteristics. However, the subfloor structure. The price of 759/5224 is significantly lower
practical accidents, such as the car impact and aircraft crash, are than carbon/epoxy materials as well as the manufacture process-
dynamic in nature accompanied with a high speed. Accordingly, ing is relatively easy. It is of great interest to investigate the energy
in order to meet crashworthiness requirements, it is necessary to absorption characteristics of the composite tubes made of
study the crushing failure process and evaluate energy absorption 759/5224 for reducing costs. In this research, the emphasis of the
capability of composite tubes under the dynamic condition. Mama- present work will focus on the effects of h and loading condition
lis et al. [20] reported that the dynamic collapse could overesti- on crushing mode, Pmax, Pm, CLE and SEA and then an optimal selec-
mate the static collapse by about 20% as far as crashworthy tion of h is recommended for the design of EA.
characteristics were concerned probably due to higher values of
dynamic friction coefficients. Ref. [7] also showed an increasing
trend in SEA under the impact loading condition. On the contrary, 2. Experimental setup
a decreasing trend in the energy absorption capability was
reported by Boria et al. [9,21]. They showed that the values 2.1. Materials and specimens
obtained from the dynamic loading condition were less than those
from the corresponding quasi-static cases. Liu et al. [22] had stud- The tube specimens were manufactured using 759/5224 woven
ied the dynamic responses and crushing characteristics of double glass cloth/epoxy prepreg with [±h]28 layup, provided by Beijing
hat shaped tubes made of CFRP. Results showed that the dynamic Institute of Aeronautical Materials, which was widely used in air-
impact tests also resulted in lower SEA as compared to the quasi- craft structures. Epoxy 5224 was a mid-toughen high-
static tests. Similar observations could also be found in the temperature curing epoxy resin matrix. The mechanical properties
literatures [6,23,24]. Besides, Chiu et al. [25] had investigated the of 759/5224 laminate were presented in Table 1 [6]. All tube spec-
behavior of a composite energy absorber (EA) made of carbon/ imens were manufactured with the same length of 125 mm and
epoxy under the static and dynamic loading with a strain rate of inner diameter of 50 mm. The specimens had h of ±15°, ±30°,
 D. Hu et al. / Composites: Part A 90 (2016) 489–501 491

Fig. 3. Curing cycle for 759/5224 composite tubes. (For interpretation of the
references to color in this figure legend, the reader is referred to the web version of
this article.)

ination. Then specimens with the length of 125 mm were cut from
the cured tubes to make the samples ready for the crushing tests.
Fig. 1. Schematic diagram of the crashworthy subfloor in helicopter fuselage. (For The fabricated tube and geometry sketch were shown in Fig. 4
interpretation of the references to color in this figure legend, the reader is referred
and their detailed information was tabulated in Table 2. Under
to the web version of this article.)
axial crushing condition, the composite tubes more likely tended
to fail in a catastrophic failure mode, which was characterized by
Table 1 a very high peak load but very low energy absorption. This mode
Mechanical properties of 759/5224 lamina [6]. was not of interest to the crashworthy structure design [9,26]
Young’s modulus, tension in direction 1/(GPa) E1t 21.0
and must be avoided. To address this issue, trigger mechanisms
Young’s modulus, compression in direction 1/(GPa) E1c 24.0 were usually used to reduce the peak load and ensure a progressive
Shear modulus in plane 1–2/(GPa) G12 3.5 failure [9–11,16,26,27]. Accordingly, in the present work, one end
Poisson’s ratio in plane 1–2 l12 0.16 of each tube was chamfered with an external angle of 45° as a trig-
Tensile strength in direction 1/(MPa) XT 402.0
ger, which was the most widely used method of trigger mecha-
Compressive strength in direction 1/(MPa) XC 453.0
Shear strength in plane 1–2/(MPa) SXY 100.0 nisms [4,10,11,18,27].

2.2. Axial quasi-static and impact crushing test facility

±45°, ±60°, ±75° with t = 3 mm, respectively. Here, the abbreviation Axial quasi-static crushing tests were conducted on a standard
of QA15, QA30, QA45, QA60, and QA75 stood for the specimens servo-hydraulic system MTS 880 with the loading capacity of
under quasi-static loading condition, while DA15, DA30, DA45, 250 kN, as shown in Fig. 5. The specimens were placed centrally
DA60, and DA75 stood for the ones under impact loading between the polished moving platen and stationary platen, while
condition. the chamfered end of every specimen was upward to contact with
The prepreg materials of 759/5224 were wrapped with the tube the moving platen. All specimens were crushed about 100 mm at
axis around a metal mandrel with the outer diameter of 50 mm, as the load rate of 10 mm/min at a temperature of 20 °C. The curve
shown in Fig. 2 and then the assembly was placed into an auto- of the crush load with respect to the moving platen displacement
clave to cure. The curing cycle was shown in Fig. 3. After curing was recorded automatically into a computer.
and demoulding, an ultrasonic flaw detector was used to inspect Axial impact crushing tests were performed by using a drop
any imperfection, such as internal crack, debond, void, and delam- hammer testing system (Fig. 6(a)) to investigate energy absorption

Fig. 2. Manufacturing process of tube specimens. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)
492 D. Hu et al. / Composites: Part A 90 (2016) 489–501

Fig. 4. Real product photo and schematic diagram with the geometry dimension. (For interpretation of the references to color in this figure legend, the reader is referred to
the web version of this article.)

characteristics. The drop hammer with the mass of 81.5 kg was From Fig. 7, it could be observed that the crush load-
lifted by the pulley to a height of 5.31 m and then released through displacement curves of both quasi-statically and dynamically
the trip gear to reach a velocity of 10.2 m/s. The specimens were tested specimens exhibited fairly similar trends characterized by
located at the center of the base of the drop hammer testing sys- a fluctuation pattern with peaks and troughs, except that the shape
tem, directly opposite to the mass center of the hammer. A force of load-displacement curves in the impact tests showed more sev-
sensor and a laser displacement sensor were mounted on the base ere serration and higher Pmax than the corresponding ones in quasi-
to measure the impact force and stroke distance between the ham- static tests, particularly at the early stage of the impact process.
mer and base, respectively. Then the force and displacement This observation was very consistent with the results in our previ-
response signals could be recorded by a digital oscilloscope (Tek- ous work [5,6] and other studies [21,22,28,29]. The higher Pmax
tronix, see Fig. 6(b)) at a sampling frequency of 100,000 HZ and could be due to both inertia effect [29] and strain rate effect
stored by a computer. Thus the crush load versus displacement [7,30], while the impact nature of loading was responsible for the
curve could be obtained from the recorded data and an SAE 1000 higher serration [5,29].
filter was adopted to remove signal noises. In addition, during As seen from Fig. 7, the crush load versus displacement curves
the impact tests, a high-speed camera (MS55K, see Fig. 6(c)) was could be divided into two distinctive stages: pre-crushing stage
used to capture the specimen crush deformation with a rate of and post-crushing stage. At the pre-crushing stage, when the cross-
2000 frame/s. head (or the hammer) contacted the trigger-side specimen end, the
To ensure the experimental repeatability, at least two replicated crush load increased rapidly up to Pmax, which corresponded to the
tests were conducted for each type of specimens under each load- microfracture nucleating with small interlaminar and intralaminar
ing condition. cracks observed at the location of the chamfer trigger due to the
localized stress concentration. Then a progressive failure process
followed to develop after Pmax at the post-crushing stage. In this
3. Experimental results and discussion stage, the crush load dropped sharply as the initiation of fracture
and buckling occurred and subsequently stabilized around Pm with
3.1. Crush load-displacement responses small oscillations. This post-crushing stage with progressive col-
lapse was the major energy dissipation process.
All the specimens exhibited excellent promising crush behav-
iors with good repeatability and were progressively crushed from
the chamfered end. The crush load versus displacement curves 3.2. Failure mechanisms
obtained from axial quasi-static and impact crushing tests were
presented in Fig. 7. The series of these diagrams allowed the first The crush zone morphologies of all the crushed specimens
comparison of the specimen responses between quasi-static and under the quasi-static and impact loading condition were pre-
impact compressive loading and visual evaluation of the dynamic sented in Figs. 8 and 9, respectively. As h increased from 15° to
effect. 75°, the specimens displayed different failure modes, which could
 D. Hu et al. / Composites: Part A 90 (2016) 489–501 493

Table 2
Specimens geometry dimension and test results under quasi-static and impact crushing condition.

Specimen number L (mm) DL (mm) M (g) m (g) Pmax (kN) Pm (kN) AE (J) SEA (J/g) CLE
QA15
No. 1 125.4 100 119.0 94.9 137.9 77.8 7781.9 82.0 0.56
No. 2 125.1 100 118.0 94.3 114.9 68.4 6835.6 72.5 0.59
Mean 126.4 73.1 7308.8 77.2 0.58
QA30
No. 1 125.1 100 118.0 94.3 102.0 68.3 6827.1 72.4 0.67
No. 2 125.1 100 115.0 91.9 107.9 66.2 6623.7 72.1 0.61
Mean 105.0 67.3 6725.4 72.2 0.64
QA45
No. 1 125.1 100 115.0 91.9 78.5 54.0 5400.9 58.8 0.69
No. 2 124.9 100 109.0 87.3 70.7 47.0 4704.8 53.9 0.67
Mean 74.6 50.5 5052.9 56.3 0.68
QA60
No. 1 125.1 100 113.0 90.3 75.6 57.9 5787.6 64.1 0.77
No. 2 125.0 100 111.0 88.8 84.9 65.2 6523.9 73.5 0.77
Mean 80.3 61.6 6155.8 68.8 0.77
QA75
No. 1 125.0 100 118.0 94.4 111.4 60.3 6033.9 63.9 0.54
No. 2 125.0 100 120.0 96.0 87.4 67.0 6697.4 69.8 0.77
Mean 99.4 63.7 6365.7 66.8 0.66
DA15
No. 1 124.9 66.0 118.0 62.4 243.2 58.9 3887.5 62.4 0.24
No. 2 125.0 67.7 119.0 64.5 256.0 62.5 4231.7 65.7 0.24
Mean 249.6 60.7 4059.6 64.0 0.24
DA30
No. 1 125.1 66.0 115.0 60.7 208.0 62.5 4125.3 68.0 0.33
No. 2 125.0 62.3 118.0 58.8 220.8 67.4 4202.0 71.4 0.31
Mean 214.4 65.0 4163.6 69.7 0.32
DA45
No. 1 125.0 84.6 109.0 73.8 124.8 47.6 4031.5 54.6 0.38
No. 2 125.2 92.8 115.0 85.2 153.6 44.0 4087.0 47.9 0.29
Mean 139.2 45.8 4059.3 51.3 0.33
DA60
No. 1 125.1 67.2 111.0 59.6 172.8 61.5 4129.6 69.2 0.38
No. 2 125.1 64.5 113.0 58.2 176.0 65.2 4203.6 72.2 0.37
Mean 174.4 63.4 4166.6 70.7 0.37
DA75
No. 1 125.0 65.8 120.0 63.2 201.6 57.7 3794.8 60.0 0.29
No. 2 125.0 67.2 118.0 63.4 182.4 63.5 4268.5 67.3 0.35
Mean 192.0 60.6 4031.6 63.7 0.32

be classified into two distinct failure modes, named as splaying or fragments formed just at the top side of the central crack due to the
lamina bending crushing mode (Mode I) and brittle fracturing fiber and matrix crushing of the bent lamina bundles against ham-
crushing mode (Mode II) in accordance with the classification by mer platen [28,32]. As compressive loading proceeded further, the
Farley [31], during axial quasi-static and impact crushing process. annular debris wedge was compressed to penetrate the tube wall,
resulting in the subsequent formation of the internal and external
fronds. During this process, it was considered that the frictional
3.2.1. Mode I resistance between the wedge and the internal and external fronds
Mode I was a typically matrix-controlled crushing mode could provide a significant contribution to energy dissipation [28].
[16,31], characterized by the progressive end crushing with inter- For Mode I, it was inferred that the dominated energy dissi-
nal and external fronds forming as a petal-like shape. Fig. 10 pre- pated mechanisms governing this crushing mode could be identi-
sented crushing failure process of DA60 captured by the high- fied as follows (see Figs. 8(c), (d) and 9(c), (d)): (1) Fronds and
speed camera, corresponding to the crush load versus displace- lamina bundles bending inward and outward; (2) Friction between
ment curve. When the hammer impacted the trigger end of the the fronds and loading platen; (3) Friction between the annual
specimen, a local buckling deformation was provoked by the com- debris wedge and the fronds; (4) Circumference fracture of internal
bination of bending and asymmetrical compression induced by the and external fronds; (5) Interlaminar cracks growth for delamina-
chamfer trigger [5]. Subsequently, as crushing process advanced tion; (6) Crack growth in the longitudinal direction and axial splits
continuously, splaying of lamina bundles occurred, causing the of tube wall.
tube wall axial splitting and the internal and external fronds form-
ing a petal-like shape followed by a sharp drop in crush load. A sig-
nificant amount of debris ejected from the tube wall obscuring the 3.2.2. Mode II
view of the high-speed camera. As seen from Fig. 8(d), some fronds Mode II was a combination of the transverse shearing crushing
broke off completely from the crushed specimens. An evident cen- mode and Mode I [33–35], which was considered to be a major
tral crack could also be observed at the crushing zone along the crushing mode for a large percent of crushing tests of brittle
mid-plane of the tube wall. An annual debris wedge of pulverized fiber-reinforced composite tubes by Farley and Jones [34].
494 D. Hu et al. / Composites: Part A 90 (2016) 489–501

of laminate thicknesses in Mode I [34]. Lamina bundles forming in

Mode II fractured with a small radius of bending curvature near the
base of the lamina bundles. The primary energy absorption mech-
anism of Mode II was governed by fiber/matrix fracture [33,34],
while interlaminar and longitudinal crack growth and friction
between fronds and loading platen were secondary contributors.

3.2.3. Effect of fiber orientation and crushing speed on failure

mechanism
The circumference stiffness (stiffness in the circumferential
direction of the tube) could have a significant influence on
intralaminar crack grown [31]. As h increased, the increase of cir-
cumference stiffness could restrain crack opening [16,31] and thus
reduced the crack length which decreased the lamina bundle
length. Hamada and Ramakrishna [37] considered that increasing
h above certain value tended to cause a transition from Mode I to
a shear failure mode. They reported the change in crushing modes
occurred at 25° for AS4 and IM7 carbon/PEEK and 20° for S2 glass/
PEEK, respectively. According to Farley and Jones [38], they also
reported a crushing mode transition from Mode I to transverse
shearing mode occurred as h increased from 45° to 65°. This trend
was also confirmed by Song et al. [17]. They reported that the
crushing mode transformed from Mode I to local buckling and then
to transverse shearing mode for filament-wound glass/epoxy cir-
cular tubes, as the winding angle increased from 15° to 90°.
Fig. 5. MTS 880 test machine. (For interpretation of the references to color in this
In present work, in the case of quasi-static tests in Fig. 8, QA45
figure legend, the reader is referred to the web version of this article.) and QA60 crushed in Mode I and QA75 crushed in Mode II. There
was an evident transition of the crushing mode from Mode I to
Mode II when h increased from 60° to 75°. From Fig. 8(c) and (d),
Ghasemnejad et al. [36] thought Mode II was an ideal crushing it could also be observed that the frond length of QA60 was shorter
energy dissipated mechanism with excellent energy absorption than that of QA45 due to the increase of circumference stiffness
capability for composites. In this mode, such as QA15, the lengths with increasing h. As for QA15 and QA30, they both crushed in
of the central cracks were between 1 and 10 times of laminate Mode II as shown in Fig. 8(a) and (b). As h increased from 30° to
thickness as contrasted with the only 1 times of laminate thickness 45°, the crushing mode changed from Mode II to Mode I. This
in the transverse shearing crushing mode and in excess of 10 times seemed to be not consistent with the trend as described by

Fig. 6. Drop hammer test facility: (a) drop hammer; (b) digital oscilloscope; (c) high-speed camera. (For interpretation of the references to color in this figure legend, the
reader is referred to the web version of this article.)
 D. Hu et al. / Composites: Part A 90 (2016) 489–501 495

Fig. 7. The crush load versus displacement of specimens with different h under quasi-static and impact loading condition. (For interpretation of the references to color in this
figure legend, the reader is referred to the web version of this article.)

Hamada and Ramakrishna [37]. It was inferred that the reason terms of QA15 and QA30, once the higher crush load was applied
could be attributed to the combined interaction of axial and cir- on the fronds, these circumferential cracks propagated easily
cumference stiffness which were both as a function of h: the axial through the fronds. When these cracks propagated through the
stiffness decreased with increasing h, whereas the circumference whole frond, lamina bundles fractured and peeled off from the tube
stiffness increased with increasing h. On the one hand, as seen from wall. Based on the above results, it could be concluded that h had a
the crush load-displacement curves (Fig. 7(a)–(c)), the crush loads noticeable influence on the crushing mode of the composite tubes
of QA15 and QA30 were observed to be higher than that of QA45, made of 759/5224 woven glass cloth-epoxy under the quasi-static
which mainly attributed to the larger axial stiffness of QA15 and loading condition.
QA30. On the other hand, it could be observed from Fig. 8 In the case of impact tests in Fig. 9, the progressive crushing pro-
(a) and (b) that other than the longitudinal cracks and intralaminar cess of the specimens were very similar to those of corresponding
cracks, there was a smaller number of cracks in the circumferential specimens in quasi-static tests, which reflected that high crushing
direction on the formed fronds of the crushed QA15 and QA30. In speed did not cause any transition in the crushing mode for the same
496 D. Hu et al. / Composites: Part A 90 (2016) 489–501

Fig. 8. Crush zone morphologies of specimens under quasi-static crushing condition. (For interpretation of the references to color in this figure legend, the reader is referred
to the web version of this article.)

specimen. However, there were still some differences observed in [6,12,25,40]. Fig. 11(a) showed the effect of h on Pmax under both
crush zone morphologies between the quasi-static and impact tests. quasi-static and impact crushing tests, respectively. Clearly, it
It was clearly seen that specimens in the impact tests possessed could be found that h had a significant influence on Pmax and the
longer fronds in comparison with those in the quasi-static tests. Par- similar trend was in both quasi-static and impact tests: Pmax
ticularly, the fronds of DA15 and DA75 were much longer than those decreased with the increase of h from 15° to 45° and then increased
of QA15 and QA75, respectively. The reason might come from the with further increasing h. Specimens with h of 45° had the lowest
matrix brittleness increase at high-speed impact, which led to the Pmax in quasi-static and impact crushing tests, respectively.
decrease in fracture toughness and thus less resistance to the longi- A comparison of Pmax between quasi-static and impact tests
tudinal crack and interlaminar crack [39]. showed that the dynamic Pmax was about 1.7–2.2 times higher than
the quasi-static one due to the inertia and strain rate effect. The
3.3. Crashworthiness characteristics effect of loading condition on Pmax was consistent with the trend
in our previous work [5,6] and other literature [22,28,29].
The energy absorption characteristic parameters of all the spec-
imens calculated from Fig. 7 were tabulated in Table 2 and plotted 3.3.2. Mean load
in Fig. 11. Pm was a good indicator to compare overall energy absorption
performance [25]. Fig. 11(b) showed the effect of h on Pm under
3.3.1. Peak load both quasi-static and impact crushing tests, respectively. It could
Pmax was an important factor, which was related directly to be observed that in the quasi-static crushing tests, Pm tended to
structure or human tolerance in order to avoid fatal injuries decrease with increasing h from 15° to 45° and this trend reversed
 D. Hu et al. / Composites: Part A 90 (2016) 489–501 497

Fig. 8 (continued)

with further increasing h from 45° to 75°. This trend of Pm was sim- Comparing CLE between the quasi-static and impact tests, it
ilar to that of Pmax. QA15 had the highest Pm, whereas QA45 had the was clearly that CLE in the impact tests was remarkably lower than
lowest Pm. As these specimens crushed in different crushing those in the quasi-static tests. A reduction in CLE between the
modes, the decrease of Pm with increasing h from 15° to 45° might quasi-static and impact tests was between 40% and 60%. As a
be due to the loss of axial stiffness and compressive strength result, it could be concluded that the loading condition had a sig-
[5,26,31], while the increase of Pm with increasing h from 45° to nificant influence on CLE of specimens fabricated of 759/5224,
75° could be attributed to the circumference stiffness [31]. which was very consistent with the trend of carbon/epoxy tubes
In the case of the impact crushing tests, no evident trend could in our previous work [5,6] and other literature [4].
be observed. DA30 and DA60 had the highest and the second high-
est Pm, respectively, whereas DA45 still had the lowest Pm. Pm of
3.3.4. Specific energy absorption
DA15 showed a noticeable decrease by the maximum error of
SEA, as an index of energy absorption efficiency, was usually
24% as compared with that of QA15. For the rest specimens, the
considered to be the most important parameter to evaluate the
discrepancy of Pm between quasi-static and impact crushing tests
crashworthiness in vehicles [4,12]. Fig. 11(d) showed the effect of
was less pronounced in the range of 1–12%. This observation could
h on SEA under both quasi-static and impact crushing condition,
be proved by their crush load vs displacement curves (see Fig. 7),
respectively. Clearly, the results showed excellent SEA performance
where the sustain crush loads of these specimens at the post-
in a range from 47.9 J/g to 82.0 J/g, which was significantly higher
crushing stage were very closer between the quasi-static and
than SEA of AerMet 100 (a high-strength steel with excellent
impact tests than that of the specimen with h of 15°.
energy absorption capability) with the value of 41.2 J/g at the
speed of 15 m/s [41].
3.3.3. Crush load efficiency As seen from Fig. 11(d), there was no obvious trend observed in
CLE was an important characteristic to measure the perfor- both quasi-static and impact crushing test results. The specimens
mance of an EA [27] and evaluate the stability of crushing process with h of 30° and 60° exhibited good SEA under both quasi-static
[4,25]. It was highly desirable to get the value of CLE close to unity and impact loading condition, while the specimens with h of 45°
for a good EA. Fig. 11(c) plotted CLE with respect to h under both had the lowest SEA probably due to less frond fracture. In the case
quasi-static and impact crushing condition, respectively. Clearly, of the quasi-static tests, the variation of SEA with the increase of h
the same trend for both quasi-static and impact crushing tests agreed well with the results for glass/polyester composites with h
could be observed that CLE increased with increasing h from 15° from 15° to 75° by Daniel et al. [19]. For h in the range of 35–75°,
to 60° and then decreased with further increasing h from 60° to the variation of SEA was also consistent with the literature by Hull
75°. Specimens with h of 60° had the highest CLE, whereas speci- [16], where a peak of SEA occurred at 65°. Moreover, the results in
mens with h of 15° had the lowest CLE due to their very high Pmax the present article were comparable in magnitude in most
(see Fig. 11(a)). This implied that specimens with h of 15° more instances to literature data by Song et al. [17] for filament-
likely tended to a catastrophic failure. wound glass/epoxy tubes, but whereas they reported a nonlinear
498 D. Hu et al. / Composites: Part A 90 (2016) 489–501

Fig. 9. Crush zone morphologies of specimens under impact crushing condition. (For interpretation of the references to color in this figure legend, the reader is referred to the
web version of this article.)

increase trend in SEA with the increase of h under both quasi-static 3.3.5. Energy absorber design
and impact crushing condition. Such trend was not found in our Based on the results from Fig. 11, it could be concluded that h
work. Zeng et al. [42] reported that for 3d braided composite tubes was a critical parameter affecting Pmax, Pm, CLE, and SEA during
with h in the range of 20 –60°, SEA decreased firstly and then the quasi-static and impact tests. It could be revealed that in fact,
increased again with increasing braid angle h, which was partly the impact loading condition led to a worse energy absorption per-
similar to the trend in our work. They also found the minimum formance, because a higher Pmax and lower Pm, CLE, and SEA was
value of SEA occurred at h = 45°. caused by the high crushing speed, whereas a good EA was highly
By comparing the experimental results under both quasi-static desirable to have the highest Pm, CLE, and SEA as well as the lowest
and impact crushing condition, it could be found that except the Pmax. Consequently, in consideration of the crash condition of the
specimens h of 15°, SEA of the rest specimens in quasi-static tests composite subfloor structure, it was strongly recommended to
was only slightly higher than those in impact tests. SEA of QA30, use impact test results as a more effective reference for EA design.
QA45, QA60 and QA75 was 1–6%, 7–11%, 2–8% and 3–6% higher Here, some choices could be provided as follows:
than that of DA30, DA45, DA60 and DA75, respectively. The dis-
crepancy of SEA for 759/5224 tubes between the quasi-static and (1) The specimen with h of 60° was more suitable for the design
impact crushing tests was less pronounced than those of carbon/ of EA which needed to dissipate the kinetic energy with a
epoxy tubes in our previous work [5,6]. higher CLE, SEA and Pm as well as a lower Pmax. It should be
 D. Hu et al. / Composites: Part A 90 (2016) 489–501 499

Fig. 9 (continued)

to investigate the effect of h on the energy absorption perfor-

240 1 2 3 4 5 6 7
mance. All specimens exhibited excellently progressive failure
Rebound

220
process at the quasi-static speed of 10 mm/min and high crush-
200 ing speed of 10.2 m/s. The images from the crush zone showed
180 specimens with h from 15° to 75° deform in two distinct crush-
160
1 ing modes.
Experimental results showed that the effect of h was proved to
Crush Load (KN)

140
have a great influence on the energy absorption performance:
120 2

100 3 (1) Under both quasi-static and impact condition, Pmax

Final shape
80 4 5 6 7 decreased with increasing h from 15° to 45° and then
60
increased with increasing h from 45° to 75°.
(2) The trend of Pm was similar with that of Pmax in the quasi-
40
static tests, but no evident trend could be found in the
20 impact tests. DA30, DA60, and DA75 had relatively higher
0 Pm, while DA45 had the lowest Pm.
0 10 20 30 40 50 60 70
(3) CLE showed a first increase trend with increasing h from 15°
Displacement (mm)
to 60° and then decrease trend with increasing h from 60° to
Fig. 10. Load-deformation history of DA60 with corresponding impact crushing 75° under both quasi-static and impact condition.
process. (For interpretation of the references to color in this figure legend, the (4) SEA did not exhibit any significant trend with the increase of
reader is referred to the web version of this article.) h. The specimen with h of 45° had the lowest SEA in both
quasi-static and impact tests, while the specimens with h
noted that this was a compromise optimal solution with of 30° and 60° had relatively high SEA.
consideration of the comprehensive energy absorption
performances. Comparing the experimental results between the quasi-static
(2) The specimen with h of 45° was recommended for the design and impact tests, the high crushing speed caused a higher Pmax
of EA with the lowest Pmax, but not necessarily with the high- and lower Pm, CLE, and SEA, which resulted in a worse energy
est energy absorption capability. absorption performance. Thus it was highly recommended that
the impact test results should be used as a reference for EA design.
4. Conclusion Based on the experimental results, it was found that the energy
absorption performance of composite tubes could be improved by
In this study, axial quasi–static and impact crushing tests of proper selection of h in consideration of crashworthiness
composite tubes fabricated of 759/5224 had been carried out optimization.
500 D. Hu et al. / Composites: Part A 90 (2016) 489–501

Fig. 11. Pmax, Pm, CLE and SEA with respect to h in quasi-static and impact crushing tests. (For interpretation of the references to color in this figure legend, the reader is
referred to the web version of this article.)

Acknowledgements [8] Yan L, Chouw N. Crashworthiness characteristics of flax fibre reinforced epoxy
tubes for energy absorption application. Mater Des 2013;51:629–40.
[9] Boria S, Scattina A, Belingardi G. Experimental evaluation of a fully recyclable
The present work is supported by the National Natural Science thermoplastic composite. Compos Struct 2016;140:21–35.
Foundation of China (Grant Numbers: 11102017 and 61473014) [10] Esnaola A, Ulacia I, Aretxabaleta L, Aurrekoetxea J, Gallego I. Quasi-static crush
energy absorption capability of E-glass/polyester and hybrid E-glass–basalt/
and the Fundamental Research Funds for the Central Universities,
polyester composite structures. Mater Des 2015;76:18–25.
Beihang University. The authors wish to gratefully acknowledge [11] Ma Y, Sugahara T, Yang YQ, Hamada H. A study on the energy absorption
these supports. properties of carbon/aramid fiber filament winding composite tube. Compos
Struct 2015;123:301–11.
[12] Liu Q, Xing HL, Ju Y, Ou ZY, Li Q. Quasi-static axial crushing and transverse
bending of double hat shaped CFRP tubes. Compos Struct 2014;117:1–11.
References [13] Siromani D, Henderson G, Mikita D, Mirarchi K, Park R, Smolko J, Awerbuch J,
Tan TM. An experimental study on the effect of failure trigger mechanisms on
[1] Hu DY, Yang JL, Hu MH. Full-scale vertical drop test and numerical simulation the energy absorption capability of CFRP tubes under axial compression.
of a crashworthy helicopter seat/occupant system. Int J Crashworthiness Compos A Appl Sci Manuf 2014;64:25–35.
2009;14:565–83. [14] Eshkoor RA, Oshkovr SA, Sulong AB, Zulkifli R, Ariffin AK, Azhari CH. Effect of
[2] Hu DY, Meng KP, Yang ZY. Numerical investigation of the energy absorption trigger configuration on the crashworthiness characteristics of natural silk
characteristics of a fan-shaped deployable energy absorber. Int J epoxy composite tubes. Compos B Eng 2013;55:5–10.
Crashworthiness 2014;19:126–38. [15] Mahdi E, Hamouda AMS, Sebaey TA. The effect of fiber orientation on the
[3] Hu DY, Song B, Wang DF, Chen ZY. Experiment and numerical simulation of a energy absorption capability of axially crushed composite tubes. Mater Des
full-scale helicopter composite cockpit structure subject to a bird strike. 2014;56:923–8.
Compos Struct 2016;149:385–97. [16] Hull D. A unified approach to progressive crushing of fibre-reinforced
[4] Xu J, Ma Y, Zhang QJ, Sugahara T, Yang YQ, Hamada H. Crashworthiness of composite tubes. Compos Sci Technol 1991;40:377–421.
carbon fiber hybrid composite tubes molded by filament winding. Compos [17] Song HW, Du XW, Zhao GF. Energy absorption behavior of double-chamfer
Struct 2016;139:130–40. triggered glass/epoxy circular tubes. J Compos Mater 2002;36:2183–98.
[5] Wang YF, Feng JS, Wu JH, Hu DY. Effects of fiber orientation and wall thickness [18] Ochelski S, Gotowicki P. Experimental assessment of energy absorption
on energy absorption characteristics of carbon-reinforced composite tubes capability of carbon-epoxy and glass-epoxy composites. Compos Struct
under axial quasi-static and impact crushing condition. Compos Struct 2009;87:215–24.
2016;13:356–68. [19] Daniel L, Hogg PJ, Curtis PT. The relative effects of through-thickness
[6] Hu DY, Luo M, Yang JL. Experimental study on crushing characteristics of properties and fibre orientation on energy absorption by continuous fibre
brittle fibre/epoxy hybrid composite tubes. Int J Crashworthiness composites. Compos B Eng 1999;30:257–66.
2010;15:401–12. [20] Mamalis AG, Manolakos DE, Demosthenous GA, Ioannidis MB. The static and
[7] Song HW, Wan ZM, Xie ZM, Du XW. Axial impact behavior and energy dynamic axial collapse of fibreglass composite automotive frame rails. Compos
absorption efficiency of composite wrapped metal tubes. Int J Impact Eng Struct 1996;34:77–90.
2000;24:385–401.
 D. Hu et al. / Composites: Part A 90 (2016) 489–501 501

[21] Boria S, Scattina A, Belingardi G. Axial energy absorption of CFRP truncated [31] Farley GL. Relationship between mechanical-property and energy-absorption
cones. Compos Struct 2015;130:18–28. trends for composite tubes. NASA TP-3284, ARL-TR-29; 1992.
[22] Liu Q, Ou ZY, Mo ZW, Li Q, Qu DP. Experimental investigation into dynamic [32] Mamalis AG, Manolakos DE, Ioannidis MB, Papapostolou DP. Crashworthy
axial impact responses of double hat shaped CFRP tubes. Compos B Eng characteristics of axially statically compressed thin-walled square CFRP
2015;79:494–504. composite tubes: experimental. Compos Struct 2004;63:347–60.
[23] Jackson A, Dutton S, Gunnion AJ, Kelly D. Investigation into laminate design of [33] Ghafari-Namini N, Ghasemnejad H. Effect of natural stitched composites on
open carbon–fibre/epoxy sections by quasi–static and dynamic crushing. the crashworthiness of box structures. Mater Des 2012;39:484–94.
Compos Struct 2011;93:2646–54. [34] Farley GL, Jones RM. Crushing characteristics of continuous fiber-reinforced
[24] Kalhor R, Case SW. The effect of FRP thickness on energy absorption of metal- composite tubes. J Compos Mater 1992;26:37–50.
FRP square tubes subjected to axial compressive loading. Compos Struct [35] Hadavinia H, Ghasemnejad H. Effects of mode-I and mode-II interlaminar
2015;130:44–50. fracture toughness on the energy absorption of CFRP twill/weave composite
[25] Chiu LNS, Falzon BG, Ruan D, Xu SQ, Thomson RS, Chen B, Yan WY. Crush box sections. Compos Struct 2009;89:303–14.
responses of composite cylinder under quasi-static and dynamic loading. [36] Ghasemnejad H, Blackman BRK, Hadavinia H, Sudall B. Experimental studies
Compos Struct 2015;131:90–8. on fracture characterisation and energy absorption of GFRP composite box
[26] Jacob GC, Fellers JF, Simunovic S, Starbuck JM. Energy absorption in polymer structures. Compos Struct 2009;88:253–61.
composites for automotive crashworthiness. J Compos Mater 2002;36:813–50. [37] Hamada H, Ramakrishna S. Effect of fiber material on the energy absorption
[27] Kim JS, Yoon HJ, Shin KB. A study on crushing behaviors of composite circular behavior of thermoplastic composite tubes. J Thermoplast Compos Mater
tubes with different reinforcing fibers. Int J Impact Eng 2011;38:198–207. 1996;9:259–79.
[28] Mamalis AG, Manolakos DE, Ioannidis MB, Papapostolou DP. On the response [38] Farley GL, Jones RM. Energy-absorption capability of composite tubes and
of thin-walled CFRP composite tubular components subjected to static and beams. DTIC document; 1989.
dynamic axial compressive loading: experimental. Compos Struct [39] Farley GL. The effects of crushing speed on the energy-absorption capability of
2005;69:407–20. composite tubes. J Compos Mater 1991;25:1314–29.
[29] Mamalis AG, Manolakos DE, Demosthenous GA, Ioannidis MB. The static and [40] Sebaey TA, Mahdi E. Crashworthiness of pre-impacted glass/epoxy composite
dynamic axial crumbling of thin-walled fibreglass composite square tubes. tubes. Int J Impact Eng 2016;92:18–25.
Compos B Eng 1997;28:439–51. [41] Hu DY, Meng KP, Jiang HL, Xu J, Liu RR. Strain rate dependent constitutive
[30] Eshkoor RA, Ude AU, Sulong AB, Zulkifli R, Ariffin AK, Azhari CH. Energy behavior investigation of AerMet 100 steel. Mater Des 2015;87:759–72.
absorption and load carrying capability of woven natural silk epoxy–triggered [42] Zeng T, Fang DN, Lu TJ. Dynamic crashing and impact energy absorption of 3D
composite tubes. Compos B Eng 2015;77:10–8. braided composite tubes. Mater Lett 2005;59:1491–6.