TRANSVERSE IMPACT OF CIRCULAR MARINE COMPOSITE

PLATES

Sutherland, L.S.1and Guedes Soares, C.2

1
Bolseiro de Pós-Doutoramento, 2 Prof. Catedrático,
Unidade de Engenharia e Tecnologia Naval
Instituto Superior Técnico
Av. Rovisco Pais, 1049-001 Lisboa, Portugal
+351 218 417 468 / [email protected]

ABSTRACT
Falling-weight impact tests on circular low fibre-volume E-glass / polyester composite plates
have been performed. Three behaviour ‘regimes’ have been defined; ‘Undelaminated’,
‘Delaminated’ and ‘Fibre damage’. A fracture mechanics model describes extremely well the
sudden onset of delamination, and gives good scaling between specimen sizes. Bending and
membrane effects are significant for thin laminates. For thick laminates, especially after the
sudden onset of delamination, the response is shear-dominated. Delaminated behaviour was
well described by an energy balance model. Final failure appears to be due to back-face
strains.
floating debris and grounding, all of which
1 INTRODUCTION
are low-velocity impacts.
The use of composite materials in a marine
The response of a composite material to
environment is advantageous due to their
impact is highly complex involving not
ease of forming double-curvatures,
only dynamic effects but also numerous and
resistance to corrosion and rot, and high
interacting damage modes including
specific material properties. However, a
internal delamination, surface micro
known disadvantage of these materials is
buckling, fibre fracture and matrix
that they are susceptible to impact damage,
degradation (Richardson and Wiseheart,
especially that due to out of plane impact.
(1996)). The impact behaviour and the
This disadvantage is exacerbated by the fact
damage modes and paths are themselves
that the behaviour under impact loading is
dependant in a complex manner on the
not well understood, and this lack of
almost infinite material permutations
confidence often leads to increased safety
including fibre and resin types, quantities,
factors and loss of potential weight savings.
architectures, and interface and on the
Mouritz et al (2001) quote safety factors of
production method used (Cartié and Irving
up to 10 applied when marine composite
(2002), Caprino and Lopresto (2001), Hirai
structures will be subjected to impact loads.
et al. (1998)).
In a marine environment, common impact
events are collisions with other craft, docks, It is also difficult to use the term ‘impact
response’ as it is possible to define the
1
tensile, compressive or bending behaviour of a clamped rectangular plate with a small,
of a composite material since the impact steel hemispherical impacter was
event is defined by many variables such as considered as a relatively severe case. Here
impacter and target geometries, impact the aim is to further this work by studying
speed and energy (Christoforou (2001), the impact of theoretically simpler circular
Sutherland and Guedes Soares (2003)). plates and hence to further characterise the
various impact damage mechanisms
Hence, there has been a large amount of
occurring.
work in the area. Abrate (1998) provides
both a comprehensive review and a 2 EXPERIMENTATION
classification of the different areas of the Flat panels of 1m by 1m were laminated by
problem. The analysis of the behaviour of hand using orthophthalic polyester resin
an impacted laminate is usually split into and 500gm-2 E-glass balanced woven
two parts; the localised contact problem and roving. A fibre mass-fraction of 0.5
the overall target deflection. Hertzian (equivalent to a fibre volume fraction of
contact law is usually used to model approximately 0.35) was stipulated as
indentation at the surface (e.g. Sun et al. representative of that commonly achieved
(1993), Yang and Sun (1982)). Complete under production conditions in the marine
models may be used to exactly describe the industry. 1, 2 and 3% by mass of
deformation of the target using beam or accelerator, catalyst and paraffin
plate theories (e.g. Chen and Sun (1985)) respectively were used, in an ambient
for simple cases for small deflections. temperature of between 18 and 21 degrees
However, these models rapidly become too Celsius, to cure the resin. In order to ensure
computationally expensive when a full cure, all specimens were stored at
considering more complex architectures, room temperature for four months before
large deflections or cases with significant testing.
shear deformations. They are also not Specimens were cut from the panels using a
effective for the consideration of damage. A diamond-surrounded circular saw. Panels of
more approximate and realistic approach in 3, 5, 10, and 15 plies were cut into ‘Small’
these latter cases is to use theories to 100mm square specimens. ‘Large’ 200mm
describe the overall response of the square specimens were cut from panels of
composite such as the energy balance and 5, 10, 15 and 20 plies. Four thickness
spring-mass methods as described in Abrate measurements were made on each specimen
(2001), where a concise overview of all of
before testing.
the main modelling techniques used may
also be found. A fully instrumented Rosand IFW5 falling
weight machine was used for the impact
The majority of the work in the area testing (Figure 1). A hemispherical ended
concerns the high-cost, high fibre-fraction, cylindrical impacter is dropped from a
usually pre-impregnated carbon-fibre epoxy known, variable height between guide rails
based autoclave or vacuum-bag produced onto a clamped horizontally supported plate
composites associated with the aerospace target. A much larger, variable mass is
materials. However, the composite attached to the impacter and a load cell
materials used in the marine industry are between the two gives the variation of
much lower fibre-faction, hand laid-up impact force with time. The data was
glass-reinforced polyester resins and there
filtered with a 2kHz low pass filter. An
is very little data available concerning these optical gate gives the incident velocity, and
lower cost composites. hence the impacter displacement and
Previous work (Sutherland and Guedes velocity and the energy it imparts are
Soares (1999a and b)) considered the low- calculated from the force-time data by
energy impact of such marine composites. successive numerical integrations. Since the
Specifically, the out of plane, central impact impacter is assumed to remain in contact
2
with the specimen throughout the impact 2.853kg and 10.853kg for the small and
event, the impacter displacement is used to large specimens respectively, but the mass
give the displacement and velocity of the used for the some of the higher incident
top face of the specimen, under the energy impacts of the small specimens was
impacter. By assuming that frictional increased to give perforation at the
effects are negligible, the energy imparted maximum velocity attainable. After testing
by the indenter is that absorbed by the the damage modes were observed and
specimen. Thus, this energy value at the noted.
end of the test is that irreversibly absorbed
3 RESULTS
by the specimen.
Damage occurred at all but the very lowest
incident energies, including matrix
Weight cracking, matrix degradation, permanent
indentation, internal delamination, partial
surface micro-buckling delamination of the
Load Cell upper ‘front-face’ laminate, front-face fibre
damage, fibre damage on the lower ‘back-
Impacter face’, and perforation. These modes form a
complex overall damage pattern, but the
progression of damage with increasing
incident energy is similar for all specimens:
1. ‘Un-delaminated’: At extremely low
incident energies damage is slight and
mainly restricted to matrix cracking.
2. ‘Delaminated’: At a low critical
incident energy delaminations
suddenly appear, which then spread
with increasing impact severity.
Specimen 3. ‘Fibre Damage’: At higher energies
fibre failure occurs, leading to
perforation.
Importantly, since the damage seen is
directly responsible for the behaviour seen,
Figure 1: Falling Weight Impact Machine these definitions also allow simple
The specimens were fully clamped between characterisation of the impact response.
two thick annular circular steel plates. The However, some differences in damage were
seen between ‘thin’ and ‘thick’ specimens
clamping force was applied using a
(in this case diameter to thickness ratios of
pneumatic actuator and a long lever arm to
greater than and less than 15 respectively)
achieve high forces. The ‘small’ and ‘large’
as illustrated in Figures 2 and 3 (where the
specimens were tested using clamps with
front-face is shown above the bottom-face).
inside diameters of 50mm and 100mm
The main differences were that the thinner
respectively. A 10mm and a 20mm
specimens incurred much less internal
diameter impacter were used to strike the
delamination, but were more prone to back
small and large specimens respectively.
face fibre damage.
Tests were performed for a range of
The impact response also differs between
increasing incident energies either up to
thin and thick specimens as shown in the
perforation where possible, or to the
force-displacement and force-time plots
maximum attainable by the machine.
Nominally the impact masses used were

3
Figures 4 and 5. Each graph shows a family
of curves for increasing incident energy.
0.74ms-1 / 0.78J 2.33ms-1 / 7.74J 3.24ms-1 / 14.97J

Figure 5: Impact Response Thick Laminate

Figure 2: Impacted Thin Laminates (3-ply 50mm

4 ANALYSIS
Diameter) 4.1 Indentation
The geometry of the impact event is defined
1.43ms-1 / 11.10J 2.99ms-1 / 48.51J 5.71ms-1 / 176.9J in Figure 6. The measured displacement of
the impacter () is made up of the sum of
the plate deflection (w) and the indentation
().
i.e.   w  (1)

 w

Figure 3: Impacted Thick Laminates (20-ply

100mm Diameter)
Thinner laminates show the increase in Figure 6: Impact Geometry
stiffness with displacement due to
membrane effects until fibre damage gives The Hertzian contact law relates the contact
a sharp drop in force. Thicker laminates force P to the indentation. For a rigid
show a bi-linear response due to the onset spherical indenter and a transversely
of internal delamination. Despite the fact isotropic half-space this may be expressed
that internal delamination also occurred in as (Tan and Sun (1985)):
the thinner laminates, no effect of this is P  k 3 / 2 (2)
seen in Figure 4.
Where k is a contact coefficient depending
on target and sphere material properties,
and the sphere radius. Complementary
work (Sutherland et al (2004)) has shown
this to be the case at lower contact forces,
but then as contact damage occurs the
relationship becomes linear. The
experimental data from the indentation
work was used here to obtain the plate
deflections from the displacement data.
4.2 Internal Delamination
Internal delamination is the first damage
Figure 4: Impact Response Thin Laminate leading to a significant reduction in material

4
properties. It is important to predict the data for internal delamination due to
onset of this since a pigmented gel coat or impact.
paint covering will hide such damage in 3
y = 1.65x - 1.59

Ln (Delamination Force (kN))

practice. Such damage gives a local 2.5 R2 = 1.00
stiffness reduction, which may cause a 2 y = 1.61x - 1.71
future failure when an abnormally large R2 = 0.99
1.5
loading is encountered and/or the damage
1 50mm Diam Impact
grows with cyclic loading. The fact that 100mm Diam Impact
delamination occurs here at such low 0.5 50mm Diam Static
100mm Diam Static
incident energies indicates that many 0
vessels will have considerable undetected 1 1.5 2
Ln (Thickness (mm))
2.5 3

delamination due to even minor, everyday

impacts. Figure 7: Delamination Threshold

Davies et al (1993) use a simple mode II 4.3 Plate Deflections

fracture analysis to describe the critical load Here both large deflections of thin plates
for the unstable onset of a single circular where membrane effects become important,
delamination in an isotropic material: and also deflections of thick plates where
1 the effects of shear must be considered.
2 2  EGIIc  2 3 2 Shivakumar et al, 1985 provide the
Pc    h
3  1  2  (3) relationship between impact force P and
Where E is Young’s modulus, GIIc is the plate deflection w in terms of bending,
mode II strain energy release rate,  is shear and membrane stiffness (Kb, Ks and
Poisson’s ratio, and h is laminate thickness. Km respectively) for a centrally loaded
Davies and Zhang (1995) used this circular plate:
approach fairly successfully for carbon P  K bs w  K m w 3 (4)
epoxy, but satisfactory results were not
obtained for high fibre-volume fraction
glass polyester laminates (Zhou and Davies
Kb K s (5)
(1995)). Cartié and Irving (2002) obtained K bs 
such good correlation using this method for Kb  K s
CFRP that they advocated the use of impact Expressions for the stiffness are also given
testing to determine GIIc. Christoforou for the four permutations of clamped or
(2001) also used equation simply supported specimens with
( 3 ) to define the onset of delamination. immovable or movable edges. These may
be expressed in the form:
A logarithmic plot of Pc against h for the
tests carried out here is made in Figure 7. K m  Ao h ; K b  Bo h 3 ; (6)
The data fits the theory extremely well, as
K s  Co h
shown by the high ‘goodness of fit’ R2
value, although the slope is slightly higher Where h is laminate thickness, Ao, Bo and
than the theoretical value of 1.5. Co are constant for a given material and
Importantly, the theory scales extremely plate diameter, and in the case of Co,
well between the small and large assuming the contact area ac is constant.
specimens. The static values shown are
from quasi-static replications of the impact Here the aim is to present the data in such a
set-up. Again very good correlation with form that it is possible to separate and
the theory is shown, with delamination characterize the various impact mechanisms
occurring at a slightly lower force than for occurring. Based on the results of previous
the impact tests. This indicates that the use work (Sutherland (1999b)) the response
of far simpler quasi-static testing may will be assumed to be shear controlled and
provide a practical, if slightly conservative, then deviations from this simplified model

5
will be used to distinguish the various types 2.5 50mm Diameter

of behaviour.

Max Force / Thickness (kN/mm)

2

Un-Delaminated
Firstly, assuming that membrane effects are 1.5 Fibre
Delaminated Damage
negligible and combining equations ( 4 ), (
1
5 ) and ( 6 ) gives: 3 Ply
5 Ply

 
0.5
P h2 (7) 10 ply

 C o  2 w

15 Ply


0
h  h C o B o  0 1 2 3 4 5 6 7
Deflection w at Max F (mm)

Hence, if membrane effects are not

significant, plots of P/h against w for Figure 8: Small Specimens Maximum Force
4
100mm Diameter
different specimen thickness will give

Max Force / Thickness (kN/mm)

linear relationships whose slopes increase 3
Fibre Damage
with thickness (for a given material and

Un-Delaminated
Delaminated
plate diameter). For very thick laminates 2

shear deflections dominate and equation ( 7 5 Ply

10 Ply
) may be further simplified to give: 1
15 Ply
20 Ply
P (8)
 Co w
0
0 2 4 6 8 10 12
h Deflection w at Max F (mm)

Hence, if shear deflections dominate, plots Figure 9: Large Specimens Maximum Force
of P/h against w for different specimen
4.4 Energy Balance
thickness will give linear relationships
whose slopes are independent of thickness. In order to characterise the relationship
In Figures 8 and 9 the maximum force between the severity of the impact event
normalized by thickness is plotted against and the force-deflection behaviour an
the corresponding plate deflection for each energy balance approach is taken. At the
test. In all graphs that follow points maximum deflection the specimen absorbs
containing a cross indicate no delamination all of the incident energy.
and those containing a dot indicate fibre Calculating the energy absorbed by
failure. integration of the force-deflection response,
The thinnest specimens show stiffening noting that for all tests where fibre damage
with increased deflection due to membrane is not significant the maximum force and
effects. A bi-linear response is seen for the maximum deflection occur almost
thicker specimens, indicating that simultaneously, and normalising both sides
membrane effects are not significant in by thickness gives:
these cases. A stiffer undelaminated wP Max
(9)
IKE P
response is followed by a less stiff
delaminated response. For the h
 0
h
dw

undelaminated response the stiffness

increases with thickness, but is approaching Again, shear deflections are assumed to
a constant value for the thickest specimens, dominate and then deviations from this
indicating that shear deflections dominate behaviour used to identify the different
for thicker specimens even before mechanisms. The force-deflection
delamination. After delamination the behaviour for thicker specimens was bi-
thicker specimens are approximately linear (Figure 5), and equation ( 8 ) may be
equally stiff, indicating that shear modified accordingly:
deflections dominate the delaminated
behaviour.

6
 10 100mm Diameter
P
 C 0 w for w  wC (a)

(Force/Thickness)2 (kN2/mm2)
8
h ( 10 )
P 6
 C1 w  (C 0  C1 ) wC (b)
h 4
5 Ply
for w  wC 10 Ply
2 15 Ply
where C0 and C1 are the ‘thickness- 20 Ply
normalised stiffness’ of the un-delaminated 0
0 5 10 15 20
and delaminated behaviours respectively Incident Energy / Thickness (J / mm)

and wc is the deflection corresponding to Pc.

Figure 11: Large Specimens Energy Balance Plot
Substituting for P/h from equations
( 10 ) into equation ( 9 ), integrating and 4.5 Fibre Damage
then using equations For the highest IKE test in Figure 4 the
displacement increases, as fibres are
( 10 ) to express the result in terms of
broken, after the maximum load has been
maximum force gives the bi-linear
reached. That is, the maximum load and
relationship:
displacements no longer occur
2
 PMax   IKE  simultaneously. Hence, the difference
   2Co   (a)
 h   h  between the maximum displacement and
for P  PC the displacement at maximum force is used
2 here as an indicator of the onset of fibre
 PMax   IKE 
   2C1  ( 11 ) damage.
 h   h 
Perforation could be initiated by one of two
 C  C1  PC 
2
  o  
 (b) plausible damage paths; bending giving
 Co  h  back-face fibre tensile failure, and high
for P  PC contact forces giving front-face fibre
In Figures 10 and 11 (PMax/h)2 is plotted ‘shear-out’ failure. If the former is the
against IKE/h. In both graphs the un- controlling mechanism then fibre failure
delaminated behaviour forms only a small should occur at a critical plate deflection for
initial part of the response. The upward- a given plate diameter. Conversely, if the
curving trend of the thinnest specimens due latter initiates perforation then fibre failure
to membrane effects is again evident. should occur at a critical contact force,
However, the main point is that the data independent of plate diameter. Figures 12
collapses very nicely onto a common linear and 13 indicate that the start of fibre failure
trend for the delaminated case. is bending controlled since this starts at an
approximately constant deflection.
4 50mm Diameter However, Figure 14 shows that fibre failure
(Force/Thickness)2 (kN2/mm2)

of the small 10-ply and 15-ply specimens

3
occurs at approximately the same maximum
force. This provides some, but
2
inconclusive, evidence that contact force
1
3 Ply
5 Ply
initiates perforation for the thickest
10 Ply laminates.
15 Ply
0
0 2 4 6 8 10
Incident Energy / Thickness (J / mm)

Figure 10: Small Specimens Energy Balance Plot

7
 18 50mm Diameter show the durations of the impacts of the
16
small and large specimens respectively. All
Max Disp - Disp at Max F (mm)

14
12 but the thinnest specimens show an increase
10 in impact duration as delamination occurs.
8
This is followed by fairly constant impact
6
4
3 Ply
5 Ply durations for the delamination-controlled
10 Ply
2 15 Ply behaviour, until the impact event is further
0 prolonged by fibre damage. In the case of
0 2 4 6 8 10
Deflection w at Max F (mm)
the thickest specimens the undelaminated
behaviour is reflected by an approximately
Figure 12: Fibre Damage Parameter vs. Deflection, constant duration at low IKE.
Small Specimens
However, the thinnest laminates show a
marked decrease in impact duration until
10
5 Ply 100mm Diameter
fibre damage occurs, even though
Max Disp - Disp at Max F (mm)

8
10 Ply delamination was present. This is because
15 Ply
20 Ply of the non-linear membrane effect not
6 allowed for in equation ( 12 ). Tests on thin
4 specimens at higher impact energies lead to
higher deflections and hence membrane
2
effects which increase stiffness and thus
0 reduce impact durations. This reduction in
0 2 4 6 8 10 12 14 duration could also be in part due to strain-
Deflection w at Max F (mm)
rate dependence of material properties
(Zhou and Davies 1995, Hancox and Mayer
Figure 13: Fibre Damage Parameter vs. Deflection, 1993) since as incident energy is increased
Large Specimens the maximum deflection increases whilst
7 50mm Diameter 3 Ply
the duration reduces giving an increase in
5 Ply
10 Ply
strain rate. However, since a similar effect
Max Disp - Disp at Max F (mm)

6
15 Ply
5
was not seen for thicker specimens any
4
such effects are not thought to be
3
significant.
2

1
10
3 Ply
0 5 Ply
0 2 4 6 8 10 12 14 16 10 Ply
8
Max F (kN) 15 Ply
Impact Duration (ms)

6
Figure 14: Fibre Damage Parameter vs. Maximum
Force, Small Specimens
4
4.6 Impact Duration
2
A spring-mass model approach assuming
geometrical non-linearities and indentation 0
are negligible gives the result for contact 0 20 40 60
Incident Energy (J)
80 100

time TC as (Abrate 1998):

Figure 15: Impact Durations, 50mm Diameter
TC   ( M / K bs )1 2
( 12 ) Specimens

Hence, the impact duration should be

independent of incident energy, increase
with impacter mass M but decrease with
increasing stiffness Kbs. Damage will
extend the impact event. Figures 15 and 16
8
 20
5 Ply
 Fibre failure leading to penetration is
10 Ply
15 Ply
thought to be back-face tension
15 20 Ply controlled, although fibre shear-out
Impact Duration (ms)

may become important for very low

10 diameter to thickness ratios.
6 ACNOWLEDGEMENTS
5
This work has been financed by Fundação
para a Ciência e a Tecnologia, Lisbon,
0 Portugal, through its pluri-annual funding
0 50 100 150 200 250
Incident Energy (J) awarded to the Unit of Marine Technology
Figure 16: Impact Durations, 100mm Diameter
and Engineering. The first author has been
Specimens financed by a scholarship from the
Fundação para a Ciência e a Tecnologia,
5 CONCLUSIONS under contract number SFRH / BPD / 1568
Drop-weight impact tests have been carried / 2000.
out for low fibre-volume glass-polyester 7 REFERENCES
laminates for two geometrically scaled
circular specimen sizes and for a range of Abrate S. Impact on composite structures.
diameter to thickness ratios. The impact Cambridge University Press, 1998.
behaviour has been characterised as a Abrate S. Modelling of impacts on
progression of three damage stages with composite structures. Composite Structures;
increasing incident energy, namely ‘Un- 51: 129-138, 2001.
delaminated’, ‘Delaminated’ and ‘Fibre
damage’. The following points for the Caprino G. and Lopresto V. On the
impact behaviour of marine composites penetration energy for fibre-reinforced
have been concluded: plastics under low-velocity impact
conditions. Composite Science and
 Impact damage occurs in two main Technology; 61: 65-73, 2001.
stages: hidden internal delamination
damage at low incident energy, and Cartié D.D.R. and Irving P.E. Effect of
then at high incident energies resin and fibre properties on impact and
perforation failure. Hence, high strain compression after impact performance of
to failure reinforcement fibres will CFRP. Composites Part A; 33: 483-493,
improve the perforation resistance, 2002.
but good interlaminar shear Chen J.K., Sun C.T. Dynamic large
properties are also required to resist deflection response of composite laminate
and reduce delamination. subject to impact. Composite Structures;
 Bending and membrane effects are 4(1): 59-73, 1985.
significant for thin laminates. For Christoforou A.P. Impact dynamics and
thick laminates, especially after the damage in composite structures. Composite
sudden onset of delamination, the Structures; 52: 181-188, 2001.
response is shear-dominated.
Davies GAO. and Zhang X. Impact damage
 A fracture mechanics model prediction in carbon composite structures.
describes extremely well the sudden Int J Impact Eng; 16(1):149-170, 1995.
onset of internal delamination, and
Davies GAO., Hitchings D. and Zhou G.
gives good scaling between specimen
Impact damage and residual strengths of
sizes.
woven fabric glass/polyester laminates.
 An energy balance approach gives Composites Part A;27A:1147-1156, 1993.
good correlation between impact
force and incident energy.

9
Hancox, NL. and Mayer RM. Design Data Tan TM. and Sun CT. Use of statical
for Reinforced Plastics. Chapman and Hall, indentation laws in the impact analysis of
1993. laminated composite plates. Trans. ASME;
52:6-12, 1985.
Hirai Y., Hamada H. and Kim J.K. Impact
response of woven glass-fabric composites Yang S.H., Sun C.T. Indentation law for
– I. Effect of fibre surface treatment. composite laminates. ASTM STP 787: 425-
Composite Science and Technology; 58: 449, 1982.
91-104, 1998. Zhou G, Davies GAO. Impact response of
Mouritz AP, Gellert E, Burchill P and thick glass fibre reinforced polyester
Challis K. Review of advanced composite laminates. Int J Impact Eng; 16(3):357-374,
structures for naval ships and submarines. 1995.
Composite Structures; 53:21-41, 2001.
Richardson M.O.W. and Wisheart M.J.
Review of low-velocity impact properties
of composite materials. Composites Part A;
27A: 1123-1131, 1996.
Shivakumar KN., Elber W. and Illg W.
Prediction of impact force and duration due
to low-velocity impact on circular
composite laminates. J. Appl.
Mech.;52:674-680, 1985.
Sun C.T., Dicken A., Wu H.F.
Characterization of impact damage in
ARALL laminates. Composite Science and
Technology; 49: 139-144, 1993.
Sutherland L.S. and Guedes Soares C.
Impact tests on woven roving e-
glass/polyester laminates. Composites
Science and Technology;59:1553-1567,
1999a.
Sutherland L.S. and Guedes Soares C.
Effects of laminate thickness and
reinforcement type on the impact behaviour
of e-glass/polyester laminates. Composites
Science and Technology; 59:2243-2260,
1999b.
Sutherland L.S., Machado Santos F. and
Guedes Soares G. “Indentation of marine
composites.” 5º Encontro Nacional de
Análise Experimental de Tensões e
Mecânica Experimental, Universidade de
Coimbra 21-23 January 2004
Sutherland LS. and Guedes Soares C. The
effects of test parameters on the impact
response of glass reinforced plastic using an
experimental design approach. Composites
Science and Technology; 63:1-18, 2003.

10