International Journal Impact

8/8/2019 International Journal Impact

http://slidepdf.com/reader/full/international-journal-impact 1/37

www.elsevier.com/locate/ijimpeng

Author’s Accepted Manuscript

Repeated impact response of hand lay-up and vacuum

infusion thick glass reinforced laminates

Giovanni Belingardi, Maria Pia Cavatorta, Davide

Salvatore Paolino

PII: S0734-743X(07)00029-2

DOI: doi:10.1016/j.ijimpeng.2007.02.005

Reference: IE 1467

To appear in: International Journal of Impact

Received date: 29 September 2006

Revised date: 7 December 2006

Accepted date: 28 February 2007

Cite this article as: Giovanni Belingardi, Maria Pia Cavatorta and Davide Salvatore Paolino,

Repeated impact response of hand lay-up and vacuum infusion thick glass reinforced

laminates, International Journal of Impact (2007), doi:10.1016/j.ijimpeng.2007.02.005

This is a PDF file of an unedited manuscript that has been accepted for publication. As

a service to our customers we are providing this early version of the manuscript. The

manuscript will undergo copyediting, typesetting, and review of the resulting galley proof

before it is published in its final citable form. Please note that during the production process

errors may be discovered which could affect the content, and all legal disclaimers that apply

to the journal pertain.

http://www.elsevier.com/locate/ijimpeng

http://dx.doi.org/10.1016/j.ijimpeng.2007.02.005

http://dx.doi.org/10.1016/j.ijimpeng.2007.02.005

http://www.elsevier.com/locate/ijimpeng



A c c e

p t e d m a n

u s c r i p

t

1

REPEATED IMPACT RESPONSE

OF HAND LAY-UP AND VACUUM INFUSION

THICK GLASS REINFORCED LAMINATES

Giovanni Belingardi, Maria Pia Cavatorta∗, Davide Salvatore Paolino

Mechanical Engineering Department – Politecnico di Torino

Corso Duca degli Abruzzi, 24 – 10129 Torino (Italy)

ABSTRACT

Vacuum infusion (VI) is being considered as a viable alternative to more traditional

hand lay-up (HL). Main reason in favor of the more costly technique is the cleaner and

friendlier work environment. Moreover, VI potentially offers another important benefit

over HL in that prepreg levels of resin may be achieved, resulting in stronger and lighter

laminates. The present paper compares the two manufacturing techniques on the basis

of the response to repeated impact loading. The laminate is a thick non-symmetric glass

fiber reinforced plastics intended for nautical application. Four impact velocities (1.5m/s, 2.2 m/s, 3.1 m/s and 3.8 m/s) were considered, and a minimum of four specimens

for any given velocity were subjected to forty repeated impacts or up to perforation. The

impact response was evaluated in terms of damage progression by visual observation of

the impacted specimens, evolution of the peak force and of the bending stiffness with

the number of impacts and by calculating the Damage Index (DI), a damage variable

recently proposed by the authors to monitor the penetration process in thick laminates.

Results point out that for impact velocities for which no perforation occurs within test

duration, the experimental data essentially overlap. On the contrary, for perforation

tests, HL specimens survived more impacts before perforating absorbing more total

energy than VI specimens. Plots of the DI variable against the number of impacts were

observed to exhibit an initial linear portion, owing to a stable process of damage

accumulation within the laminate, and to undergo an unstable growth a few impacts

before perforation. When comparing the VI and HL specimens it was observed that,

given an impact energy, the level of damage at first impact as well as the rate of stable

damage accumulation is alike for the two sets of specimens. On the contrary, it is the

number of impacts of the stable damage accumulation region which is lower for VI

specimens.

KEYWORDS: low velocity impact, damage accumulation, glass fiber reinforced

composite, nautical applications.

INTRODUCTION

In the nautical sector, where large products are involved and where there is no massive

series production, Hand Lay-up (HL) is by large the most widely used manufacturing

∗ Corresponding author. Fax number: +39.011.5646999. e-mail address: [email protected]



A c c e

p t e d m a n

u s c r i p

t

2

technique. However, in the HL technique, a mayor health risk for operators comes from

the resin fumes that exhale from the open mould on which the operator needs to hover

over while saturating the laminate by hand. Producers of fiber reinforced composite

laminates are thus looking for cleaner and friendlier work environment manufacturing

processes. In this respect, Vacuum Infusion (VI) is now being considered as a viable

alternative to the more traditional HL technique.

While in a typical HL, reinforcements are laid into the mould and manually wet out

using brushes or rollers and then vacuum is used to remove the resin in excess, VI takes

a different approach, in that a vacuum is drawn while the materials are still dry. Once

vacuum is achieved, resin is literally sucked into the laminate via carefully placed

tubing. Ideally, any excess resin that is introduced will eventually be sucked out into the

vacuum line.

Therefore, besides environmental issues, VI potentially offers another important benefit

over HL, in that it should allow for a very predictable resin usage approaching prepreg

levels of resin content. Because of the improvement in the fiber-to-resin ratio, laminates

manufactured by VI should be stronger and lighter as compared to laminates

manufactured by HL.

The paper presents data of a comparative experimental study conducted on a glass fiber

reinforced plastics for nautical application, manufactured by HL and VI. Comparison is

made on the laminate response to low velocity repeated impacts, a loading condition of

particular relevance for naval or nautical applications. Recently few papers have

addressed the problem of repeated impact loading [1-5.]

Low velocity impacts on laminates are known to significantly reduce the laminate

strength and stiffness, mainly as a consequence of multiple stacked delaminations that



A c c e

p t e d m a n

u s c r i p

t

3

are produced by the impact loading at a number of interfaces through the thickness of

the composite laminate. In [1,2], typical SN curves for glass fiber reinforced plastics are

drawn. In [1], an endurance limit above 104

impacts was observed. As it is the case for

classical fatigue, the endurance limit corresponds to a condition of no degradation of the

composite mechanical characteristics. In [2], SEM observations revealed that, even

without any visible external damage, microcracks in the resin can produce internal

delaminations, thus reducing the composite laminate strength. For composites, the

damage induced by impact loading is more subtle than in metals, as it is often not

detectable, beginning on the non-impacted surface or in the form of internal

delamination. Matrix cracking is chronologically the first damaging mode.

Delaminations are initiated by critical matrix cracks that are in fact emerging at the

interfaces between layers. Curves of damage evolution against the number of impacts

reveal three regions at various energy levels. At low impact energies, the damage

process is governed by initiation and multiplication of delamination, with the non-

impacted face of the specimen being the first region to delaminate. As the impact

energy increases, saturation of delamination is achieved. Beyond saturation of the

delamination process, there is an acceleration in the damage accumulation until final

failure, which occurs by ply cracking with fiber breakage. In [3], tensile and

compressive static tests were performed on beam-like specimens cut from carbon/epoxy

plates subjected to repeated impact tests. Test results point out that, for tests in which

perforation is not achieved within test duration, a degradation in residual properties

occurs only for specimens that include the damage impact zone. Impacts at higher

energy levels induce more damage (i.e. lower residual compressive and tensile strength)

than a number of lighter impacts. On the contrary, for tests in which perforation is

achieved, the degradation of residual properties is the same regardless of the energy per



A c c e

p t e d m a n

u s c r i p

t

4

single impact. Moreover, the damage sustained in the area surrounding the contact point

is higher for a laminate perforated in several impacts with less energy per impact.

To the authors’ best knowledge all available experimental papers on repeated impact

tests are for thin laminates which rarely find application in the marine industry. As

correctly pointed out in [6], the advancement from thin to thick laminates is not trivial,

as thick composite laminates behave quite differently from their thin counterparts. One

important aspect differentiating thick from thin laminates is the extent of the penetration

process which obliges to a clear distinction between laminate penetration and

perforation. For thick laminates, the energy required by the dart to go through the

laminate can not be neglected.

In the present paper a thick glass reinforced plastics laminate is tested to repeated

impacts, with an emphasis on the comparison between the more traditional HL

manufacturing technique and a viable alternative such as VI. Four falling heights are

considered, corresponding to conditions of no perforation within test duration and to

conditions of laminate perforation. The impact response is evaluated in terms of damage

progression by visual observation of the impacted specimens, evolution of the peak

force (maximum of the load-displacement curve) and of stiffness loss as a function of

impact number, and by calculating the Damage Index (DI), a damage variable recently

proposed by the authors [7] to monitor the penetration process in thick laminates.

THE DAMAGE INDEX

In [8,9] Belingardi & Vadori introduced the Damage Degree (DD) to account for

damage accumulation in thin laminates.

Defined as the ratio between the absorbed energy Ea and the impact energy Ei (Figure

1):



A c c e

p t e d m a n

u s c r i p

t

5

≡=

n penetratioafter E

E

n penetratiotoup E

E

DD

sa

a

i

a

,1

,

(1)

the DD was shown to increase rather linearly with the impact energy to reach the value

of one at laminate penetration. A saturation energy level Esa was defined as the impact

energy at which the DD regression curve reaches the value of one. This energy

threshold is of practical and theoretical interest since it defines the maximum energy

level the laminate can dissipate with no penetration and by means of internal damage

mechanisms only [9].

As pointed out in [6], while a single energy threshold is generally sufficient to define

the impact characteristics of thin laminates, for thick laminates two different threshold

values are to be defined: the penetration threshold Pn and the perforation threshold Pr .

The penetration threshold is identified at the first time the absorbed energy reaches the

level of impact energy and it is therefore conceptually the same as the saturation energy.

For impact energies above the penetration threshold, the impactor moves deeper into the

laminate. Once the impact energy is high enough, perforation eventually takes place.

The impact energy is higher than the absorbed energy and the energy in excess is

retained in the impactor for post-perforation motions. Between the penetration and

perforation thresholds, there exists a range, named by Liu “the range of the penetration

process”, in which the impact energy and the absorbed energy are equal to each other

but which represent different stages of the penetration process with the impactor

moving deeper and deeper into the specimen as the impact energy increases.

However, by its definition the DD is unable to differentiate between conditions of

laminate penetration and perforation since over the entire penetration process the

absorbed energy would be equal to the impact energy and the DD equal to one. To



A c c e

p t e d m a n

u s c r i p

t

6

account for how far the impactor has moved down into the laminate, the maximum

deflection sMAX registered during the impact test was introduced into definition of the

DI:

QS

MAX

s s DD DI = (2)

Normalization by the maximum stroke of the quasi-static penetration test sQS allows for

a non-dimensional quantity. The sQS value as well as the sMAX value in perforation tests

correspond to the stroke value at which the force becomes constant (Figure 2) and equal

to the friction force of the dart sliding through the penetrated specimen. For all the

laminates tested in [7] the sMAX measured in perforation tests was practically equal to

sQS; so that the DI reaches the value of one at laminate perforation.

EXPERIMENTAL

The laminate under study is a glass fiber reinforced plastics. The stacking sequence is

reported in Table 1. Two resin systems are used: vinylester and polyester. Main reason

for the asymmetry of the laminate, in both the resin system and the stacking sequence, is

cost reduction. The interface between the two resins is located at about two thirds of the

laminate thickness from the vinylester face. The fiber reinforcement varies from 35% in

weight in the mat, 45% in the bidirectional lamina and 50% in the unidirectional lamina.

With the purpose of investigating the possibility of replacing the current manufacturing

technique, i.e. HL, with a more environment friendly technique such as VI, static

indentation and impact tests were performed on two sets of specimens manufactured by

the two technologies. Specimens were cut from plates. The thickness of the plates was

measured at several locations showing reduced scatter. However, for both technologies

the achieved thickness values were quite distant from the design protocol. The average



A c c e

p t e d m a n

u s c r i p

t

7

thickness for HL was 10.13 mm (standard deviation: 0.25 mm), while for VI was 9.72

mm (standard deviation 0.05 mm).

For both sets, specimens were impacted on the polyester face of the laminate. Previous

tests performed on HL specimens [10] have indeed shown that, while higher peak forces

are achieved by impacting the vinylester face, the number of impacts to perforation is

greater when impacting the polyester face. The experimental result was explained by

considering that, in the case it is the polyester face to be impacted, the more ductile

resin –vinylester– is on the rear target face where delamination is known to initiate [2],

thus limiting the progression of damage. Moreover, it was observed that the condition

of laminate perforation is associated with the splitting of laminae, which always occurs

at the interface between the two resin systems (Figure 3). Therefore perforation is

delayed in the case the impacted face is polyester, being the interface between the two

resin systems more distant from the rear target face. For repeated impact tests, four

falling heights were considered (125 mm, 250 mm, 500 mm, 750 mm) corresponding to

four impact velocities (1.5 m/s, 2.2 m/s, 3.1 m/s and 3.8 m/s) and a minimum of four

specimens for any given velocity were subject to forty repeated impacts or up to

perforation. Single impact tests were also performed at 4.4 and 6.3 m/s (corresponding

to the maximum height of the drop-dart apparatus) to investigate possible strain-rate

effects.

Impact tests were performed according to ASTM 3029 standard [11] using an

instrumented free-fall drop dart testing machine. The impactor has a total mass of 20 kg,

its head is hemispherical with a radius of 10 mm. Stainless steel was chosen for its high

hardness and resistance to corrosion. The maximum falling height of the testing

machine is 2 m, which corresponds to a maximum impact energy of 392 J. The drop-

weight apparatus was equipped with a motorized lifting track. The collected data were



A c c e

p t e d m a n

u s c r i p

t

8

stored after each impact and the impactor was returned to its original starting height.

Using this technique, the chosen impact velocity was consistently obtained in

successive impacts. Because, the target holder was rigidly attached to the frame of the

testing device, the tup struck the specimen each time at the same location. By means of

a piezoelectric load cell, force-time curves were acquired. The acceleration history was

calculated dividing the force term by the impactor mass. The displacement was obtained

by double integration of the acceleration and thus force-displacement curves were

plotted. By integration of the force-displacement curves, deformation energy-

displacement curves were then obtained. Initial conditions were given with the time axis

having its origin at the time of impact. At time t=0, the dart coordinate is zero and its

initial velocity can be obtained by the well known relationship:

h g v ∆= 20 (3)

where ∆h is defined as the height loss of the center of mass of the dart with respect to

the reference surface [8]. The drop dart machine used in the study is equipped with an

optoelectronic device for measurement of the impact velocity. Agreement between

measured and theoretical values was very good.

Square specimen panels, with 100 mm edge, were clamped with a 76.2 mm inner

diameter, and fixed to a rigid base to prevent slippage of the specimen (Figure 4). The

clamping system makes use of pre-loaded springs to provide an adequate and repeatable

uniform pressure all over the clamping area.

Prior to impact tests, a series of static indentation tests were performed to get

information on the material stiffness and strength characteristics, which serves as a

starting point to decide on the falling heights of impact tests. For quasi-static

penetration tests, specimens were tested using a servo-hydraulic machine with

maximum loading capacity of 100 kN. The hydraulic actuator was electronically



A c c e

p t e d m a n

u s c r i p

t

9

controlled in order to perform constant velocity tests. Signals of the force applied by the

actuator and of the actuator stroke were acquired in time with an appropriate sampling

rate.

To assess the progression of damage, for one series of repeated impact tests at any given

impact velocity, pictures of the specimen impacted and rear face as well as of the

specimen thickness were taken after each impact. A software program was used to

determine the area of delamination, readily seen under light due to a change in the

opacity of the resin. In addition to visual observation of impacted specimens, for all

tests variation of bending stiffness against impact number was evaluated from the force-

displacement curves.

TESTS AND RESULTS

Figure 5a depicts representative force-displacement curves obtained in quasi-static

perforation tests for the two sets of laminates. As it can be observed from the graph, the

first damage force for VI specimens is around 8 kN. The small plateau in the force

associated to the first damage was consistently observed in all tests performed on VI

specimens. For HL, the first damage force is around 10kN and leads to a variation in the

curve slope rather than to a plateau. While the first damage force appears to be slightly

higher for HL than for VI specimens, values for the maximum force are alike. As for the

laminate stiffness, values were calculated from the initial portion of the force-

displacement curves before the first damage takes place and are reported in Figure 5b.

The laminate stiffness was consistently higher for HL specimens. Over a minimum of

three tests for each manufacturing technology, calculated average values for the

laminate stiffness were 6493 N/mm for HL and 5736 N/mm for VI specimens.



A c c e

p t e d m a n

u s c r i p

t

10

In analyzing the static test results, it should be reminded that the VI specimens are about

four tenths of a millimeter thinner than HL specimens. In this respect, it is interesting to

note that the different values obtained for the laminate stiffness are well explained by

the difference in thickness. Indeed, the ratio between the average stiffness values

(6493/5736= 1.13) is equal to the ratio of the thickness elevated to the third power

(10.13/9.72=1.04; (1.04)3= 1.13). Due to the thickness of the laminate under study, it is

the flexural behavior that affects the first damage event rather than the membrane

behavior as observed in very thin laminates [9]. Looking for a relationship between the

first damage force and the thickness is more troublesome. Indeed, for the case under

study, the difference in thickness is so limited that it would be speculative to comment

on a relationship between load and thickness elevated to the power of 3/2, as observed

by many authors with reference to the critical load in impact tests [12]. What can be

said is that for the first damage force (10/8=1.25) the manufacturing technology seems

to play a role in determining the value of the force and the type of damage involved.

Figures 6 show the force-displacement curves for two series of ten repeated impact tests

at 3.1 m/s run on HL and VI specimens. From the two graphs, the Damage Threshold

Load (DTL) [12] can be evaluated and compared for the two sets of specimens. A

correlation between the DTL and the laminate thickness is noticeable (12.8/11.6= 1.10).

The two series of curves depicted in Figure 6 clearly show that the laminate stiffness

diminishes impact after impact and that the highest reduction is achieved in the first few

impacts. The statement is supported by data shown in Figure 7, where the stiffness is

plotted against the impact number for the considered four impact velocities and for HL

and VI specimens. Reasonably, the total loss in laminate stiffness is greater for higher

impact velocities. Data observation points out that for VI specimens (solid symbols) the

stiffness loss is in essence concentrated in the first impact while the loss appears less



A c c e

p t e d m a n

u s c r i p

t

11

abrupt for HL specimens (hollow symbols). However, when looking at the value of the

stiffness at perforation or the asymptotic value of no perforation tests, they are alike for

VI and HL specimens meaning that the total stiffness loss is greater for HL specimens.

Data points on the vertical axis show that the laminate stiffness is not constant at

different impact velocities. To investigate the laminate strain-rate sensitivity, additional

single impact tests were performed at 4.4 and 6.3 m/s (corresponding to the maximum

height of the drop-dart apparatus). Figures 8a and 8b plot average values for the

laminate stiffness and the initial damage load in the form of DTL, respectively, against

the impact velocity. Data obtained in quasi-static penetration tests are reported on the

vertical axis. The maximum available impact energy was not sufficient to perforate the

laminate in single impact tests, therefore no comparison could be carried out in terms of

maximum load. Although the investigated range of impact velocities is rather limited,

data plotted in Figure 8 demonstrate a strain-rate dependency of the stiffness and initial

damage force. Values of R 2 show a good fitness of a linear function on the experimental

data. The observed strain-rate dependency is likely to be owed to the large amount of

resin in the laminate.

In Figures 9 and 10 plots of the peak force and of the damage variable DI against the

impact number are presented. To avoid confusion among the experimental data,

separate graphs have been edited grouping data obtained in no perforation tests (Figures

9) and in perforation tests (Figures 10). For Figure 10 reported data are not average

values but refer to one series of repeated impact tests. Indeed, in perforation tests, the

number of impacts to perforation among different specimens may change by one or two

unities, thus causing scatter in the peak force and DI values in the one-two impacts

before perforation. On the contrary, for no perforation tests single test and average

values agree very well.



A c c e

p t e d m a n

u s c r i p

t

12

Experimental results presented in Figures 9 show that, for no perforation tests, data for

the two sets of specimens are rather similar. A distinction should be made between the

two impact velocities. At the lowest velocity, for which both the peak force and the DI

reach a stable value indicating the reaching of a steady-state condition, data for the HL

and VI basically overlap. For the graphs at 2.2 m/s, in which the peak force continues to

slowly decrease impact after impact while the DI increases, suggesting a slow but

steady accumulation of damage, VI specimens appear to damage more than HL.

The result is confirmed in perforation tests (Figures 10). Main result that can be evinced

from Figure 10b is that the number of impacts to perforation is smaller for VI

specimens. The depicted difference of four impacts between VI and HL is the average

observed gap. When looking at the result, it is important to remind that the VI

specimens are about 4 tenths of a millimeter thinner than the HL (9.72 mm against

10.13 mm). However, taking into account the difference in thickness to give reason for

a difference in the number of impacts to failure is not straightforward, as it was for the

value of the stiffness. A possibility could be comparing the two sets of specimens on the

basis of the total impact energy density at perforation, that is on the basis of the product

between the number of impacts to perforation and the impact energy density per single

event (impact energy divided by the specimen volume). As it can be observed from

Figure 11, the total impact energy at perforation is higher for HL specimens, which

seem to exhibit a greater damage tolerance than VI. From Figure 11 it can also be noted

that, regardless of manufacturing technology, perforation is not associated with a

specific level of total energy. Higher impact energies per single event are more

damaging than a number of lighter impacts.

As for the graphs of the peak force (Figures 9a and 10a), it is interesting to note that for

all considered impact velocities, the maximum of the peak force is not reached in the



A c c e

p t e d m a n

u s c r i p

t

13

first impact. The literature suggests two explanations for the phenomenon. In a series of

repeated impact tests run on carbon/epoxy composite laminate, Wyrick and Adams [3]

commented the initial increase in the peak force as the result of the compaction process

of the thin layer of unreinforced resin at the impacted surface. At low impact energy

levels, damage to the fibers near the surface is minimal and the compaction process

provides a harder surface for the next impact. In a series of single impact tests run at

different impact velocities on glass/epoxy laminates, Liu [6] observed that, even if

delamination takes place very early in the impact event, indentation and local matrix

cracking are the dominant damage modes up to the maximum peak force. The

maximum in the peak force therefore signals a turning point in the dominant damage

mode. In particular, up to maximum peak force the dominant damage modes are

indentation and local matrix cracking around the impacted region; while after being

loaded by the maximum peak force, the damage accumulation process is dominated by

delamination. For impact velocities above the maximum peak force, delamination

becomes the dominant damage mode while matrix cracking, and hence lamina splitting,

continue to grow.

When considering the curves depicted in Figures 9a and 10a, the compaction process

suggested in [3] may well explain the reaching of a steady-state value in the peak force

for the 1.5 m/s tests as well as the initial rise in the peak force observed for the 2.2 m/s

tests (Figure 9a). On the contrary, both visual observation of the impacted specimens

and the laminate stiffness loss (Figure 7) do not seem to support Liu’s findings. Indeed,

for the 3.1 and 3.8 m/s impact velocities, the stiffness loss is concentrated in the first

few impacts; moreover, visual observation of the impacted specimens did not reveal

anything noteworthy in the impacts before and after the maximum of the peak force. In

the authors’ opinion, it should be considered that the value of the peak force is affected



A c c e

p t e d m a n

u s c r i p

t

14

by two opposite mechanisms: on one side the compaction of the resin which provides a

harder surface for the next impact event and on the other side the progression of damage

which reduces the laminate mechanical properties. Initially, the beneficial local surface

effect of resin compaction can determine an increase in the peak force, even if the

global bending stiffness of the laminate has decreased.

Figures 9a and 10a show that at increasing impact velocities, the impact number at

which the maximum of the peak force occurs generally decreases. Also, the decrease in

the peak force after the maximum value becomes sharper. Interestingly, while for

impact velocities causing no perforation the maximum value of the peak force increases

for increasing impact velocity (Figure 9a), in tests where perforation takes place the

maximum in the peak force is alike regardless of the test impact velocity (Figure 10a).

A few comments are worthwhile making on the DI plots of Figures 9b and 10b. In no

perforation tests (Figure 9b), the DI value is always very small. While for the 1.5 m/s

tests it remains constant at the value of 0.1 impact after impact, for the 2.2 m/s tests the

DI increases quite linearly with the impact number, owing to a slow but steady

accumulation of damage. DI values for VI specimens are slightly above those for HL

specimens. In perforation tests (Figure 10b), the DI shows an initial linear region

followed by an unstable growth a few impacts before perforation. During the

penetration process the DI growth is highly non linear. A value of one is reached at

complete laminate perforation. By looking at the DI values plotted in Figure 10b, it is

worth commenting that, given the impact velocity, the initial linear region of the curves

is alike for HL and VI specimens, both in terms of point values and of curve’s slope. In

other words, both the level of damage at first impact and the rate of stable damage

accumulation is the same regardless of the manufacturing technology. On the contrary,

it is the number of impacts sustained by the specimen before the onset of unstable



A c c e

p t e d m a n

u s c r i p

t

15

damage growth which is smaller for VI specimens. As it can be evinced from Figure

10b, the number of impacts in the unstable region does not change between VI and HL

specimens. In [13], it was shown that a strong interfacial bond between the fiber and

resin matrix delays the initiation of severe damage modes, thus improving the overall

resistance to repeated impacts. Similarly, the reduced damage tolerance of VI specimens

could be ascribed to the larger amount of resin of HL specimens and to the fact that

manual wetting out of the prepregs assures good quality and homogeneous fiber

impregnation. One last comment is to be made regarding the slope of the DI vs. impact

number curves at different impact velocities. By looking at Figures 9b-10b, it is

apparent that the slope increases for increasing impact velocity, i.e. increasing impact

energy. Figure 12 plots values of the slope of the DI vs. impact number curves as a

function of the impact energy. A quadratic regression curve well fits the experimental

data. As already pointed out, no significant difference is found between HL and VI

specimens.

Figures 13 and 14 reproduce photographs of the laminate impacted and rear faces after

the first impact as well as after the impact prior to perforation, for HL and VI specimens

respectively. The impact velocity is 3.1 m/s. As it is noticeable from the pictures, the

damage appearance for the two laminates is rather different, in particular for what

concerns the extent of damage visible from the rear face after the first impact.

The extent of the delamination area in the laminate was assessed by the change in the

opacity of the resin. Clearly such evaluation was limited to the laminate faces and

neglects internal delaminations. Observation of the specimen thickness after each

impact event allowed to observe that prior to perforation no significant laminae splitting

occurs, while at perforation splitting of laminae takes place at the interface between the

two resin systems (Figure 3).



A c c e

p t e d m a n

u s c r i p

t

16

By observing the two laminate faces, it can be noticed that in the case of VI specimens

(Figure 13), delamination extends to the whole unsupported region of the specimen

from the first impact. On the contrary, for HL specimens (Figure 14) the amount of

delamination in the rear face after the first impact is rather limited. Delamination

continues to grow as the number of impacts increases but does not appear to saturate the

whole specimen unsupported region even at perforation. The dissimilarity in the extent

of the delamination area visible from the rear face may explain the previously observed

difference in stiffness loss at first impact, i.e. the fact that for VI specimens the stiffness

essentially drops off in the first impact while it decreases less abruptly in HL specimens

(Figure 7). If the damage associated with delamination appears earlier in time for VI

specimens, the energy dissipation mechanism of fiber pull-out is on the contrary more

readily seen in HL specimens. By looking at the rear face of the two specimens it can be

noted that, while in VI specimens fiber damage is first limited to the contact area with

the impactor tup, for HL specimens fiber pull-out extends along the fiber directions

from the first impact.

CONCLUSIONS

The impact characteristics of a non-symmetric thick glass fiber reinforced plastics

laminate were investigated. To look into the possibility of manufacturing the laminate

through VI, two series of repeated impact tests up to forty impacts or to perforation

were run at different impact velocities, comparing VI with the more traditional HL

technology. The impact response was evaluated in terms of damage progression by

visual observation of the impacted specimens, evolution of the peak force and of the DI

with the number of impacts as well as through assessment of laminate stiffness

reduction.



A c c e

p t e d m a n

u s c r i p

t

17

The following statements summarize the experimental results:

• no significant differences exist in the force, energy curves and damage

parameter for tests in which no perforation occurs;

• when the impact energy is such to determine laminate perforation, HL

specimens survived more impacts before perforating absorbing more total

energy;

• visual observation of the impacted specimens shows that in both series of tests

and for all the considered impact velocities, the growth of the delamination area

is concentrated in the first impacts. In particular, for VI specimens, the

delamination area saturates from the first impact, following closely the behavior

of the bending stiffness against the number of impacts;

• for both sets of specimens, perforation of the laminate is associated to laminae

splitting at the interface between the two resin systems;

• the damage variable DI is observed to initially grow linearly impact after impact

owing to a stable accumulation of damage to then undergo an abrupt growth a

few impacts before perforation. During the penetration process the DI growth is

highly non linear. A value of one is reached at complete laminate perforation;

• when calculating the rate of stable damage accumulation (slope of the DI vs.

impact number curves) in the initial linear portion, a quadratic relationship is

found between the rate of damage accumulation and the impact energy;

• given the impact energy, no significant difference in the rate of stable damage

accumulation is observed between HL and VI specimens. However, the number

of impacts sustained by the specimen before the onset of unstable growth is

smaller for VI specimens.



A c c e

p t e d m a n

u s c r i p

t

18

ACKNOWLEDGMENTS

The authors wish to acknowledge AZIMUT Yachts for supplying the specimens.



A c c e

p t e d m a n

u s c r i p

t

19

REFERENCES

[1] Roy, R., Sarkar, B.H. and Bose, N.R., 2001, “Impact fatigue of glass fibre-

vinylester resin composites”, Composites: Part A, 32, pp. 871-876.

[2] Azouaoui, K., Recha, S., Azari, Z., Benmedakhene, S., Laksimi, A. and

Pluvinage, G., 2001, “Modelling of damage and failure of glass/epoxy composite plates

subject to impact fatigue”. Int J Fatigue, 23, pp. 877-885.

[3] Wyrick, D.A. and Adams, D.F., 1998, “Residual strength of a carbon/epoxy

composite material subjected to repeated impact”. J. Composite Materials, 22, pp. 749-

765.

[4] Baucom, J.N. and Zikry, M.A., 2005, “Low-velocity impact damage progression

in woven e-glass composite systems”. Composites: Part A, 36, pp. 658-664.

[5] Kawaguchi, T., Nishimura, H., Ito, K., Sorimachi, H., Kuriyama, T. and

Narisawa, I., 2004, “Impact fatigue properties of glass fiber-reinforced thermoplastics”.

Composite Science and Technology, 64, pp. 1057-1067.

[6] Liu, D., 2004, “Characterization of impact properties and damage process of

glass/epoxy composite laminates”. J. Composite Materials, 38, pp. 1425-1442.

[7] Belingardi, G., Cavatorta, M.P. and Paolino, D.S., “A new damage index to

monitor the range of the penetration process in thick laminates. Submitted to

Composites Science and Technology.

[8] Belingardi, G., Grasso, F. and Vadori, R., 1998, "Energy absorption and damage

degree in impact testing of composite materials", Proceedings XI ICEM (Int. Conf.

Experimental Mechanics), Oxford (UK), pp. 279-285.

[9] Belingardi, G. and Vadori, R., 2003, “Influence of the laminate thickness in low

velocity impact behaviour of composite material plate”. Composite Structures, 61, pp.

27-38.



A c c e

p t e d m a n

u s c r i p

t

20

[10] Belingardi, G., Cavatorta, M.P. and Paolino, D.S., 2006, “Repeated impact

behaviour and damage progression of glass reinforced plastics”. 16th European

Conference of Fracture (ECF16), Alexandroupolis, Greece, July 3-7, 2006.

[11] ASTM D3029 – “Standard Test Method for Impact Resistance of Rigid Plastic

Sheeting or Parts by means of a Tup (Falling Weight)”. American Society for Testing

Materials (1982).

[12] Schoeppner, G.A. and Abrate, S., 2000, “Delamination threshold loads for low

velocity impact on composite laminate”. Composites: Part A, 31(9), pp. 903-915.

[13] Choi, H.Y., Wu, H.Y.T. and Chang, F.K., 1991, “A new approach towards

understanding damage mechanisms and mechanics of laminated composites due to low-

velocity impact: Part II – analysis”. J. Composites Materials, 25(8), pp. 1012-1038.



A c c e

p t e d m a n

u s c r i p

t

21

Table 1. Stacking sequence of the laminate. Thickness (t) of each ply is the one given in

the design protocol.

ply type t (mm) N. of plies

mat 0.74 2

UD 1.25 2

BD 0/90 1.40 2vinylester

mat 0.74 1

BD 0/90 1.40 2

UD 1.25 1polyester

mat 0.74 1

total t (mm) 12.31



A c c e

p t e d m a n

u s c r i p

t

22

FIGURE CAPTIONS

Figure 1. An example of force vs. time and energy vs. time plots. Definition of force

and energy terms.

Figure 2: An example of quasi-static penetration tests and impact tests to perforation.Definition of the stroke variables sMAX and sQS.

Figure 3. Splitting of laminae at the interface between the two resin systems. Impacted

face: vinylester.

Figure 4. Testing fixture for impact testing

Figure 5. Representative force-displacement curves for quasi-static perforation tests.

Complete curve (a); initial portion for calculation of the laminate stiffness (b)

Figure 6. Representative force-displacement curves for repeated impact tests. Impact

velocity: 3.1 m/s. HL (a); VI (b)

Figure 7. Laminate stiffness against number of impacts.

Figure 8: Average laminate stiffness (a) and delamination Threshold Load –DTL- (b) as

a function of impact velocity.

Figure 9. Comparison of the peak force (a) and DI (b) vs. impact number. No

perforation tests.

Figure 10. Comparison of the peak force (a) and DI (b) vs. impact number. Perforation

tests.

Figure 11. Values of total impact energy density at perforation for different impact

velocities.

Figure 12. Slope of the DI vs. impact number curves plotted against the impact energy.

Figure 13. Pictures of a HL specimen impacted at 3.1m/s.

Figure 14. Pictures of a VI specimen impacted at 3.1m/s.



A c c e

p t e d m a n

u s c r i p

t

23

Figure 1. An example of force vs. time and energy vs. time plots.Definition of force and energy terms.

0

5000

10000

15000

20000

25000

30000

0.0 1.0 2.0 3.0 4.0 5.0 6.0 7.0 8.0 9.0 10.0

Time [ms]

F o r c e

[ N ]

0

20

40

60

80

100

E n e r g y

[ J ]

Force

Energy

Impact

Energy

Peak

Force

Absorbed

Energy

Rebound

Energy



A c c e

p t e d m a n

u s c r i p

t

24

Figure 2. An example of quasi-static penetration tests and impact tests to perforation.

Definition of the stroke variables sMAX and sQS.

0

5

10

15

20

25

30

0 8 16 24 32

Displacement [mm]

F o r c e

[ k N ]

0.2mm/s

3.8m/s

0

1

2

3

4

5

6

25 26 27 28 29 30 31 32 33 34

F~constant

sMAX=sqs



A c c e

p t e d m a n

u s c r i p

t

25

Figure 3. Splitting of laminae at the interface between the two resin systems.

Impacted face: vinylester.



A c c e

p t e d m a n

u s c r i p

t

26

φ 20

76.2

Stroke

A

B

CLAMPING AREA

Z

Y

X

Figure 4. Testing fixture for impact testing



A c c e

p t e d m a n

u s c r i p

t

27

(a)

(b)

Figure 5. Representative force-displacement curves for quasi-static perforation tests.

Complete curve (a); initial portion for calculation of the laminate stiffness (b)



A c c e

p t e d m a n

u s c r i p

t

28

(a)

(b)

Figure 6. Representative force-displacement curves for repeated impact tests.

Impact velocity: 3.1 m/s. HL (a); VI (b)

0

5000

10000

15000

20000

25000

30000

0 0.002 0.004 0.006 0.008 0.01 0.012

Displacement [m]

F o r c e

[ N ]

DTL= 11.6 kN

0

5000

10000

15000

20000

25000

30000

0 0.002 0.004 0.006 0.008 0.01 0.012

Displacement [m]

F o r c e

[ N ]

DTL= 12.8 kN



A c c e

p t e d m a n

u s c r i p

t

29

Figure 7. Laminate stiffness against number of impacts.



A c c e

p t e d m a n

u s c r i p

t

30

(a)

(b)

Figure 8: Average laminate stiffness (a) and Delamination Threshold Load –DTL- (b)

as a function of impact velocity.



A c c e

p t e d m a n

u s c r i p

t

31

(a)

(b)

Figure 9. Comparison of the peak force (a) and DI (b) vs. impact number.

No perforation tests.



A c c e

p t e d m a n

u s c r i p

t

32

(a)

(b)

Figure 10. Comparison of the peak force (a) and DI (b) vs. impact number.

Perforation tests.



A c c e

p t e d m a n

u s c r i p

t

33

Figure 11. Values of total impact energy density at perforation for different impact

velocities.



A c c e

p t e d m a n

u s c r i p

t

34

Figure 12. Slope of the DI vs. impact number curves plotted against the impact energy.



A c c e

p t e d m a n

u s c r i p

t

35

1st

impactimpacted face

1st

impactrear face

impact before perforationimpacted face impact before perforationrear face

Figure 13. Pictures of a HL specimen impacted at 3.1m/s.



A c c e

p t e d m a n

u s c r i p

t

1st

impactimpacted face

1st

impactrear face

impact before perforationimpacted face impact before perforationrear face

Figure 14. Pictures of a VI specimen impacted at 3.1m/s.

Documents

International Journal Impact