Impact Resistance of Marine Sandwich Structures
Tiago da Silva Rodrigues Castilho
Dissertação para obtenção do Grau de Mestre em
Engenharia e Arquitectura Naval
Júri
Presidente: Doutor Carlos António Pancada Guedes Soares
Orientador: Doutor Leigh Stuart Sutherland
Vogal: Doutor Yordan Ivanov Garbatov
Junho de 2014
iii
Acknowledgements
A great part of my gratitude goes to Dr. Leigh Sutherland, for the guidance, and shared knowledge and
interest during the development of the work. Would also like to thank for giving me the freedom to make
mistakes and learn with them.
To Dr. Carlos Guedes Soares, by the guidance through the whole course.
To Estaleiros Navais de Peniche, for the great support, providing all the materials and human resources
with immense know-how, both extremely important during the development and manufacturing of the
sandwich panels.
To Amorim Cork Composites, for gently providing the Corecork panels and sharing the interest and
knowledge to solve unexpected problems related with the use of cork on the panels.
To DeCivil laboratory staff, for the support solving all kind of problems during the preparation and test
stages, especially, cutting the specimens.
To my friends and colleagues, for the criticism, but also for the support and friendship.
Last but not least, a special thanks to my family, especially my parents, José and Belmira, for all the
support and sacrifices.
v
Resumo
Este trabalho é motivado pela crescente utilização de estruturas em materiais compósitos sandwich.
É apresentada uma revisão de literatura, com foco no impacto em estruturas marinhas. A segunda parte
do trabalho consiste na produção e ensaio à flexão, indentação e impacto de uma série de compósitos
marinhos com estrutura em sandwich.
Quatro tipos de núcleo são utilizados (PVC, Balsa, Corecork NL10 e NL20) para produzir painéis em
sandwich, com faces de polyester reforçadas com fibra de vidro.
Os testes de impacto são realizados até que se obtenha a rotura da segunda camada do provete, por
perfuração ou separação da camada do núcleo. Os provetes de PVC e NL20 mostram resultados
repetíveis, enquanto os provetes de NL10 falham de maneiras diferentes, sem que haja uma relação
com a espessura da primeira camada ou com a velocidade do impacto. Os provetes de NL10
apresentam uma má cura, que leva a uma baixa rigidez mas a uma grande capacidade de plastificação
e absorção de energia.
Apesar de as forças máximas relacionadas com a rotura das primeira e segunda camada de fibra de
vidro serem 1.5 vezes superiores nos testes de impacto, o comportamento global dos provetes de PVC
e NL20 é bem aproximado pelos testes de indentação. Por outro lado, o comportamento dos provetes
de NL10 varia significativamente, aumentando 3 vezes a energia absorvida.
Este trabalho indica que as estruturas compósitas sandwich com cortiça têm potencial em aplicações
com solicitações de impacto, com a desvantagem de apresentarem menor rigidez e maior peso.
Palavras chave: Resistência ao impacto; Compósitos marinhos; Compósitos sandwich; Cortiça; Testes
de indentação; Testes de impacto
vii
Abstract
This work is motivated by the ever increasing range of applications of sandwich composite materials in
the marine industry. A brief literature review is presented, focused in areas with special interest for
marine impact. The second part of the work consists in the manufacture and flexure, quasi-static and
impact tests of a series of marine sandwich composites.
Four different core materials (PVC, Balsa Corecork NL10 and NL20) are used to produce a sandwich
laminate, with E-glass/polyester skins.
Drop-weight tests are performed until the failure of the second skin, either by penetration or separation
from the core. PVC and NL20 specimens show predictability and repeatability of results, while NL10
present different failure modes, without any relation with skin thickness or incident velocity. NL10
specimens present bad cure of the skin, which leads to low stiffness but high plasticization and energy
absorption capabilities.
Apart from the peak forces related with the failure of both skins, that are around 1.5 times higher in the
impact tests, the overall behaviour of the PVC, Balsa and NL20 specimens is well predicted by quasi-
static tests. On the other hand, NL10 specimens’ behaviour change dramatically from static do impact
test, increasing 3 times the absorbed energy.
This work indicates that cork sandwich composites have potential in applications with impact
requirements, with the downside of lower stiffness and higher weight.
Keywords: Impact resistance; Marine composite; Sandwich; Cork; Quasi-static test; Drop-weight test
ix
Contents
Acknowledgements ............................................................................................................................... iii
Resumo .................................................................................................................................................. v
Abstract ................................................................................................................................................ vii
Contents ................................................................................................................................................ ix
List of Tables ......................................................................................................................................... xi
List of Figures ...................................................................................................................................... xiii
List of abbreviations ............................................................................................................................ xix
1. Introduction ..................................................................................................................................... 1
1.1. Motivation ................................................................................................................................... 1
1.2. Aim and structure of the dissertation .......................................................................................... 1
2. State of the art ................................................................................................................................. 3
2.1. General impact ........................................................................................................................... 3
2.2. Marine impact ............................................................................................................................. 8
2.2.1. Impact modeling ..................................................................................................................... 8
2.2.2. Material selection and structural solutions ............................................................................ 12
2.2.3. Impact behaviour comparisons of Classification Society Rules ............................................ 14
2.2.4. Residual strength .................................................................................................................. 14
2.2.5. Water absorption .................................................................................................................. 16
2.3. Conclusions .............................................................................................................................. 17
3. Experimental procedures .............................................................................................................. 19
3.1. Typical marine sandwich........................................................................................................... 19
3.2. Relevant impact scenario ......................................................................................................... 20
3.3. Selection of the experimental tests ........................................................................................... 20
3.3.1. Bending test .......................................................................................................................... 20
3.3.2. Static indentation test ........................................................................................................... 20
3.3.3. Impact test ............................................................................................................................ 21
3.4. Specimen preparation............................................................................................................... 21
3.4.1. Material selection and sandwich design ............................................................................... 21
3.4.2. Panel manufacturing ............................................................................................................. 22
3.4.3. Specimen cutting and preparation ........................................................................................ 25
3.5. Specimen tests ......................................................................................................................... 29
3.5.1. Bending tests ........................................................................................................................ 29
3.5.2. Quasi-static indentation tests ................................................................................................ 34
3.5.3. Impact tests .......................................................................................................................... 37
x
4. Result analysis .............................................................................................................................. 43
4.1. Bending tests ............................................................................................................................ 43
4.2. Quasi-static indentation ............................................................................................................ 48
4.3. Impact tests .............................................................................................................................. 54
4.3.1. Evolution of failure with increasing energy ............................................................................ 54
4.3.2. Quasi-static testing vs Impact testing ................................................................................... 59
5. Conclusions ................................................................................................................................... 65
5.1. Future work ............................................................................................................................... 66
6. References .................................................................................................................................... 67
Appendices ........................................................................................................................................... 70
A. List of marine impact references by place and research area .................................................. 70
B. Specimen dimensions............................................................................................................... 72
C. Impact test data: Load vs displacement and Absorbed energy vs Displacement ..................... 76
D. Post-test specimen photographs .............................................................................................. 85
D.1 Quasi-static tests .............................................................................................................. 85
D.2 Impact tests ...................................................................................................................... 87
xi
List of Tables
Table 1 - Relevant mechanical properties of the core materials utilized ............................................... 22
Table 2 - Materials utilized in the manufacture of the sandwich panels ................................................ 23
Table 3 - Experimental vs manufacturer shear strength ....................................................................... 47
Table 4 - Absorbed energy at different instants during indentation test ................................................ 53
Table 5 - Marine impact references, by place and research area ......................................................... 70
Table 6 - Relevant dimensions and estimated shear strength for specimens used in bending tests .... 72
Table 7 - Representative dimensions for specimens used in quasi-static tests .................................... 73
Table 8 - Representative dimensions for specimens used in impact tests ............................................ 74
Table 9 - Dimensions, weight and density of specimens from each manufactured panel ..................... 75
Table 10 - Summary of peak forces and corresponding absorbed energy - PVC, NL20 and Balsa
specimens ............................................................................................................................................ 83
Table 11 - Summary of peak forces and corresponding absorbed energy - NL10 specimens .............. 84
xiii
List of Figures
Figure 1 - Impact regimes (from left to right): Ballistic impact, Intermediate velocity impact and Low
velocity impact, [2] .................................................................................................................................. 4
Figure 2 - Vacuum bagging technique [53] ........................................................................................... 23
Figure 3 - Resin flow to the top of the panel, during vacuum bagging. Corecork NL10, NL20 and
Divinycell H100 (from left to right) ........................................................................................................ 24
Figure 4 - Detail of a beam cut with a toothed disk ............................................................................... 25
Figure 5 - Stone cutting machine .......................................................................................................... 26
Figure 6 - Thickness of the manufactured panels ................................................................................. 26
Figure 7 - Skin thickness of the manufactured panels .......................................................................... 27
Figure 8 - Average density of the manufactured panels ....................................................................... 28
Figure 9 - Hydraulic test machine (Leigh Sutherland) .......................................................................... 29
Figure 10 - Beam supports attached to the machine base (Leigh Sutherland) ..................................... 30
Figure 11 – Load vs displacement for different spans and loading pad configurations (unfiltered results)
............................................................................................................................................................. 30
Figure 12 - FNL11-1 test, showing excessive bending ......................................................................... 31
Figure 13 – Load vs displacement for first round of tests ..................................................................... 31
Figure 14 - Load vs displacement for all PVC beams ........................................................................... 32
Figure 15 - Load vs displacement for all NL10 beams ......................................................................... 32
Figure 16 - Load vs displacement for all NL20 beams ......................................................................... 33
Figure 17 - Load vs displacement for all Balsa beams ......................................................................... 33
Figure 18 - Simple support frame ......................................................................................................... 34
Figure 19 - Force vs Displacement for quasi-static tests on PVC specimens ....................................... 34
Figure 20 - Force vs Displacement for quasi-static tests on NL1 specimens ....................................... 35
Figure 21 - Force vs Displacement for quasi-static tests on NL2 specimens ....................................... 35
Figure 22 - Force vs Displacement for quasi-static tests on BAL specimens ....................................... 35
Figure 23 - Contact force for quasi-static tests on PVC specimens ...................................................... 36
Figure 24 - Contact force for quasi-static tests on NL10 specimens .................................................... 36
Figure 25 - Contact force for quasi-static tests on NL20 specimens .................................................... 36
Figure 26 - Contact force for quasi-static tests on Balsa specimens .................................................... 37
Figure 27 - Rosand IFW5 falling-weight impact machine (Leigh Sutherland) ....................................... 37
Figure 28 - Load vs Displacement for impact tests on PVC specimens – 50J-200J energy range ....... 39
Figure 29 - Load vs Displacement for impact tests on NL10 specimens – 50J-200J energy range ..... 39
Figure 30 - Load vs Displacement for impact tests on NL20 specimens – 50J-200J energy range ..... 40
Figure 31 - Load vs Displacement for impact tests on Balsa specimens – 50J-200J energy range ..... 40
Figure 32 - Load vs Displacement for impact tests on all specimens - 300J range .............................. 41
xiv
Figure 33 - Load vs Displacement for impact tests on NL10 specimens - 450J range ......................... 41
Figure 34 - Overview of the typical behaviour for each core material ................................................... 43
Figure 35 - Test of the specimen FPVC1-2: Rupture of first skin (left) and specimen shape after removal
of the load (right) .................................................................................................................................. 44
Figure 36 - Test of the specimen FNL11-2: Maximum deflection (left) and specimen shape after removal
of the load (right) .................................................................................................................................. 44
Figure 37 - Test of the specimen FNL21-2: Failure of core due to shear (left) and specimen shape after
removal of the load (right)..................................................................................................................... 44
Figure 38 - Test of the specimen FBAL1-2: Failure of core due to shear and material discontinuity (left)
and specimen shape after removal of the load (right) .......................................................................... 45
Figure 39 - Shear yield strength for PVC .............................................................................................. 46
Figure 40 - Shear yield strength for NL10............................................................................................. 46
Figure 41 - Shear yield strength for NL20............................................................................................. 47
Figure 42 - Absorbed energy vs Displacement for quasi-static tests on PVC specimens ..................... 48
Figure 43 - Quasi-static indentation frame sequence (SPVC1-3): a)-Elastic domain; b)-Perforation of
first skin; c)-Perforation of second skin; d)-End of test ......................................................................... 48
Figure 44 - Absorbed energy vs Displacement for quasi-static tests on NL10 specimens.................... 49
Figure 45 - Quasi-static indentation frame sequence (SNL11-2): a)-Elastic domain; b)-Elastic domain
with significant first skin deformation; c)-Perforation of first skin; d)-Perforation of the core, with first skin
approaching the original shape; e)-Perforation of second skin; f)-End of test ...................................... 50
Figure 46 - Absorbed energy vs Displacement for quasi-static tests on NL20 specimens.................... 51
Figure 47 - Quasi-static indentation frame sequence (SNL21-2): a)-Elastic domain; b)-Perforation of first
skin; c)-Perforation of second skin; d)-End of test ................................................................................ 51
Figure 48 - Absorbed energy vs Displacement for quasi-static tests on Balsa specimens ................... 52
Figure 49 - Quasi-static indentation frame sequence (SBAL1-3): a)-Elastic domain; b)-Perforation of first
skin; c)- Perforation of second skin; d)-End of test (second skin almost completely separated) .......... 52
Figure 50 - Load vs Displacement for impact tests on PVC specimens - 50J-300J energy range ....... 54
Figure 51 - Load vs Displacement for impact tests on NL10 specimens - 50J-450J energy range ...... 55
Figure 52 - Initial stiffness vs 1st skin thickness for DNL10 specimens ................................................ 55
Figure 53 - Initial stiffness vs Initial velocity for DNL10 specimens....................................................... 55
Figure 54 - Load vs Displacement for impact tests on NL20 specimens - 50J-300J energy range ...... 56
Figure 55 - Load vs Displacement for impact tests on Balsa specimens - 50J-300J energy range ...... 57
Figure 56 - Condition of the core in the specimens DBAL1-6, DBAL1-10 and DBAL1-13 .................... 58
Figure 57 - Absorbed energy to reach failure of specimen skins - Quasi-static tests ........................... 59
Figure 58 - Absorbed energy to reach failure of specimen skins - Impact tests.................................... 60
Figure 59 - Load vs Displacement for PVC specimens – Quasi-static and impact ............................... 60
Figure 60 - Load vs Displacement for NL10 specimens – Quasi-static and impact .............................. 61
Figure 61 - Load vs Displacement for NL20 specimens – Quasi-static and impact .............................. 61
Figure 62 - Load vs Displacement for Balsa specimens – Quasi-static and impact ............................. 62
xv
Figure 63 - Absorbed energy at failure of first skin vs Initial velocity .................................................... 63
Figure 64 - Maximum force at failure of first skin vs Initial velocity ....................................................... 63
Figure 65 - Displacement at failure of first skin vs Initial velocity .......................................................... 63
Figure 66 - Load vs Displacement and Absorbed energy vs Displacement for PVC - 50J ................... 76
Figure 67 - Load vs Displacement and Absorbed energy vs Displacement for NL10 - 50J .................. 76
Figure 68 - Load vs Displacement and Absorbed energy vs Displacement for NL20 - 50J .................. 76
Figure 69 - Load vs Displacement and Absorbed energy vs Displacement for Balsa - 50J .................. 77
Figure 70 - Load vs Displacement and Absorbed energy vs Displacement for PVC - 100J ................. 77
Figure 71 - Load vs Displacement and Absorbed energy vs Displacement for NL10 - 100J ................ 77
Figure 72 - Load vs Displacement and Absorbed energy vs Displacement for NL20 - 100J ............... 78
Figure 73 - Load vs Displacement and Absorbed energy vs Displacement for Balsa - 100J ................ 78
Figure 74 - Load vs Displacement and Absorbed energy vs Displacement for PVC - 150J ................. 78
Figure 75 - Load vs Displacement and Absorbed energy vs Displacement for NL10 - 150J ................ 79
Figure 76 - Load vs Displacement and Absorbed energy vs Displacement for NL20 - 150J ................ 79
Figure 77 - Load vs Displacement and Absorbed energy vs Displacement for Balsa - 150J ................ 79
Figure 78 - Load vs Displacement and Absorbed energy vs Displacement for PVC - 200J ................. 80
Figure 79 - Load vs Displacement and Absorbed energy vs Displacement for NL10 - 200J ................ 80
Figure 80 - Load vs Displacement and Absorbed energy vs Displacement for NL20 - 200J ................ 80
Figure 81 - Load vs Displacement and Absorbed energy vs Displacement for Balsa - 200J ................ 81
Figure 82 - Load vs Displacement and Absorbed energy vs Displacement for PVC - 300J ................. 81
Figure 83 - Load vs Displacement and Absorbed energy vs Displacement for NL10 - 300J ................ 81
Figure 84 - Load vs Displacement and Absorbed energy vs Displacement for NL20 - 300J ................ 82
Figure 85 - Load vs Displacement and Absorbed energy vs Displacement for Balsa - 300J ................ 82
Figure 86 - Load vs Displacement and Absorbed energy vs Displacement for NL10 - 450J ................ 82
Figure 87 - SPVC1-2 after test ............................................................................................................. 85
Figure 88 - SPVC1-3 after test ............................................................................................................. 85
Figure 89 - SNL11-1 after test .............................................................................................................. 85
Figure 90 - SNL11-2 after test .............................................................................................................. 85
Figure 91 - SNL21-1 after test .............................................................................................................. 86
Figure 92 - SNL21-2 after test .............................................................................................................. 86
Figure 93 - SBAL1-1 after test .............................................................................................................. 86
Figure 94 - SBAL1-2 after test .............................................................................................................. 87
Figure 95 - DPVC1-2 after test ............................................................................................................. 87
Figure 96 - DPVC2-5 after test ............................................................................................................. 87
Figure 97 - DPVC1-1 after test ............................................................................................................. 88
Figure 98 - DPVC2-6 after test ............................................................................................................. 88
xvi
Figure 99 - DPVC1-3 after test ............................................................................................................. 88
Figure 100 - DPVC2-7 after test ........................................................................................................... 88
Figure 101 - DPVC2-4 after test ........................................................................................................... 89
Figure 102 - DPVC2-8 after test ........................................................................................................... 89
Figure 103 - DPVC2-9 after test ........................................................................................................... 89
Figure 104 - DPVC2-10 after test ......................................................................................................... 89
Figure 105 - DNL11-3 ........................................................................................................................... 90
Figure 106 - DNL12-6 ........................................................................................................................... 90
Figure 107 - DNL11-1 ........................................................................................................................... 90
Figure 108 - DNL11-2 ........................................................................................................................... 90
Figure 109 - DNL12-7 ........................................................................................................................... 91
Figure 110 - DNL12-4 ........................................................................................................................... 91
Figure 111 - DNL12-8 ........................................................................................................................... 91
Figure 112 - DNL12-10 ......................................................................................................................... 91
Figure 113 - DNL12-5 ........................................................................................................................... 92
Figure 114 - DNL12-9 ........................................................................................................................... 92
Figure 115 - DNL12-11 ......................................................................................................................... 92
Figure 116 - DNL12-12 ......................................................................................................................... 92
Figure 117 - DNL12-13 ......................................................................................................................... 93
Figure 118 - DNL12-14 ......................................................................................................................... 93
Figure 119 - DNL12-15 ......................................................................................................................... 93
Figure 120 - DNL21-2 ........................................................................................................................... 93
Figure 121 - DNL21-5 ........................................................................................................................... 94
Figure 122 - DNL21-1 ........................................................................................................................... 94
Figure 123 - DNL22-6 ........................................................................................................................... 94
Figure 124 - DNL21-3 ........................................................................................................................... 94
Figure 125 - DNL22-7 ........................................................................................................................... 95
Figure 126 - DNL21-4 ........................................................................................................................... 95
Figure 127 - DNL22-8 ........................................................................................................................... 95
Figure 128 - DNL22-9 ........................................................................................................................... 95
Figure 129 - DNL22-10 ......................................................................................................................... 96
Figure 130 - DBAL1-4........................................................................................................................... 96
Figure 131 - DBAL1-7........................................................................................................................... 96
Figure 132 - DBAL1-1........................................................................................................................... 96
Figure 133 - DBAL1-8........................................................................................................................... 97
Figure 134 - DBAL1-11 ......................................................................................................................... 97
xvii
Figure 135 - DBAL1-12......................................................................................................................... 97
Figure 136 - DBAL1-5........................................................................................................................... 97
Figure 137 - DBAL1-9........................................................................................................................... 98
Figure 138 - DBAL1-6........................................................................................................................... 98
Figure 139 - DBAL1-10......................................................................................................................... 98
Figure 140 - DBAL1-13......................................................................................................................... 98
Figure 141 - DBAL1-14......................................................................................................................... 99
Figure 142 - DBAL1-15......................................................................................................................... 99
xix
List of abbreviations
ABS – Acrylonitrile Butadiene Styrene
CAI – Compression After Impact
CP – Cross ply
CSM – Chopped strand mat
DNV – Det Norske Veritas
DSP – Digital speckle photography
FEM – Finite Element Method
FRP – Fibre Reinforced Plastic (or Polymer)
GRP – Glass Reinforced Plastic (or polymer)
IMO – International Maritime Organization
MOSAIC - Materials On-board: Steel Advancements and Integrated Composites
NDE – Non Destructive evaluation
NDI – Non Destructive Inspection
PC - Polycarbonate
PE - Polyethylene
PU - Polyurethane
WR - Woven roving
1
1. Introduction
1.1. Motivation
Composite materials have been widely used in marine industry, in a wide, and ever increasing range of
applications. Fiberglass reinforced plastic (FRP) was the first modern composite material used in the
construction of small military craft, after World War II. The advantages were enormous: easy production,
low maintenance, good strength-to-weight ratios, non-magnetic properties,...
During the 1960s, the use of FRP moved towards recreational craft building, and was the main cause
of the increase in boat ownership. In the last years, and depending on the type of craft, new materials
and methods have been used. The concern with the weight of the hull of racing craft, lead to the
construction of sandwich hulls, instead of single skin hulls.
Sandwich construction allows a panel to have less weight than a single skin panel with the same bending
stiffness. This is achieved through the concept of similar to an I beam. The thin skins of the laminate
reproduce the role of the flanges of an I beam, carrying most of the tensile and compressive loads due
to bending. The web is reproduced by the core material, whose main role is to keep the skins together
and carry shear loads from one skin to the other, where maximum shear is located.
Since the application of composite materials is very often in light-weight and high speed craft, which are
often subjected to high impact loads such as slamming or collision with floating objects, impact
resistance properties of these laminates become an even bigger concern.
Although there is a large amount of literature on impact on composite materials, typically concerning
aerospace materials and scenarios, only a small part of those is applicable to the marine field. There is
a need to discuss this literature together with the work developed specially for the marine industry.
1.2. Aim and structure of the dissertation
The aim of this dissertation is to extend the work already developed in CENTEC, from single skin to
sandwich laminates. Is also desired to develop a comprehensive literature review of impact on marine
composites and a material and impact testing program that will allow for the comparison of marine
laminates in terms of impact resistance, for a given relevant scenario to marine conditions. To achieve
the aims, the dissertation will be developed according to the following structure:
Brief review of the state of the art in general impact on composites, followed by a review of the
state of the art in impact on marine composites;
Identification of a typical sandwich laminate to be studied, as well as a relevant impact scenario;
Selection of material properties and impact testing methods;
Manufacturing and testing of the sandwich laminate;
Result analysis and conclusions.
3
2. State of the art
During the lifetime of a general structure, it will certainly be subjected to impact loads, even if the purpose
of the structure is not to handle impact loads. Manufacturing and maintenance impacts, like a tool
dropped by a worker or an impact during a transportation operation are very common.
Laminated composite structures are more susceptible to impact damage than similar metallic structures.
Composite materials show different micro and macroscopic properties than isotropic materials, as steel
and aluminum. Among other concerns, composite materials have the ability of developing internal
damage, almost impossible of being detected, and that can lead to the failure of the structure when
subjected to a load far below the ones it was designed to support, or reduce its service life by fatigue
failure.
In this chapter, it will be presented a brief review on general impact of composite materials, addressing
the main methodologies and concerns in this area of studies.
However, most of the last developments have more impact in the “state of the art” composites and/or
industries, namely, aerospace industry and composites used for that purpose. Most of the advanced
construction techniques and materials used in these industries are also used in marine industry, but with
small expression (racing boats and components). The majority of the marine applications are based in
low tech fiberglass composites, with some advances in the last years, where mainly luxury craft builders
are evolving into the use of sandwich composites, production methods with better final properties
(vacuum bagging and infusion) and other materials than fiberglass and PVC foam.
Attending to this and other particularities, it will be presented a detailed review on the work developed
in the last years, concerning marine impact.
2.1. General impact
First of all, it is important to define the difference between low-velocity impact and high-velocity, or
ballistic impact. There are some definitions of ballistic impact. One of them differentiates the two
classifications by the penetration of the plate. However, this definition is a bit confusing, because two
impacts with the same velocity would be classified differently, depending on the resistance to penetration
by the plate or object under test. So, a different definition will be used. The low-velocity impacts are
those in which stress wave propagation through the thickness of the specimen has no significant
importance [1]. In other words, in high-velocity impacts, damage is introduced in the plate before its
motion is established, while in low-velocity, damage occurs after the establishment of plate’s motion.
The intermediate velocity impact is characterized by the existence of flexural waves, and the low velocity
impact can be treated as a quasi-static indentation [2]. In Figure 1, three impact regimes are shown.
As an example, for epoxy resin, the transition velocity is somewhere between 10 and 20 [m/s].
4
Figure 1 - Impact regimes (from left to right): Ballistic impact, Intermediate velocity impact and Low velocity impact, [2]
Most of the reviews in general and marine impact on composites will focus on low-velocity impact. Most
of the drop-weight tests and impacts during structure life don’t generate that level of impact velocities.
In terms of general impact of composite materials, some effort has been done in the last years in order
to group and systematize the various theories and methodologies [1], [2]. The first book is more
concerned in analysis and modelling methodologies, while the second as a more practical approach,
addressing major engineering applications.
When there is a collision between two bodies, a certain pressure distribution is developed in the surface
of both bodies, and can be represented by a contact law. The contact laws depend on geometry and
material properties, and can be calculated statically with reasonable accuracy, when dealing with low-
velocity impacts.
The indentation, α, is defined as the difference in the displacement of the projectile and that of the back
face of the laminate. The force between the two bodies, during the indentation process, can be described
by the Hertzian contact law:
𝑃 = 𝑘𝛼
32
( 1 )
where k is a coefficient that depends on geometry (radius) )and material properties of both bodies
(Young Modulus, E and Poison’s ratio, ν). The law was derived for homogeneous and isotropic materials,
but can also be applied to composite materials, regarding fiber and matrix damage in the contact zone.
In this case, k is better determined by experimental procedures.
After indentation, or loading phase, occurs the unloading phase. Due to permanent indentation, α0, the
unloading phase is different from the loading phase, and can be approximated by the following equation:
𝑃 = 𝑃𝑚 [
𝛼 − 𝛼0𝛼𝑚 − 𝛼0
]2.5
( 2 )
where Pm is the maximum force during the loading phase and αm is the maximum indentation.
Before study the analytical impact models, it is important to refer the main beam and plate theories that
will support them. The main beam theories are the Bernoulli-Euler Beam Theory, Timoshenko beam
Theory and Higher Order Beam Theory. The main plate theories are the Classical Plate Theory, First
Order Shear Deformation Theory and Higher Order Plate Theories.
5
To model the dynamics of an impact event, different models can be used, depending on the desired
degree of accuracy or simplicity of the problem.
Spring-mass models can be used to model the impact problem. One or two degrees of freedom can be
used, as well as a variety of stiffness constants (linear stiffness of structure, non-linear membrane
stiffness and non-linear contact law). Solving the problem for the initial conditions, it is possible to
compute indentation, force and deflection along time. Considerations about the mass ratios and impact
speed can also be derived [3].
Other approach is to study the impact event through energy considerations. The initial kinetic energy of
the projectile will equal the final energy stored in the structure, in the different modes (bending, shear,
membrane and local damage on indentation region).
1
2𝑀𝑉2 = 𝐸𝑏 + 𝐸𝑠 + 𝐸𝑚 + 𝐸𝑐 = ∫ 𝑃 𝑑𝛼
𝛼𝑚𝑎𝑥
0
= 2
5𝑘𝛼𝑚𝑎𝑥
52
( 3 )
Ultimately, based on the analytical beam and plate theories, complete models can be derived. This way,
dynamic behaviors will be described accurately, according to the accuracy of the base analytical theory.
Projectile dynamics and local indentation are also taken in consideration. This methods result in the
solution of a matrix system, depending on the discretization of the structure. However, the solution can
be computationally expensive, depending on geometry, discretization and boundary conditions.
The impact models described so far, are important for the initial assessment about maximum forces,
indentation and impact duration, for instance. However, due to the complexity of an impact event and
the great amount of factors that influence the impact resistance of a certain structure, it is important to
perform experimental testing, allowing for a better knowledge of the damage initiation and propagation.
About the experimental testing, one should distinguish impact testing from material properties testing.
Although both are very common in composite industry, the first is focused on the study of the impact
resistance of a certain structure to certain impact criteria, while the second is focused on the mechanical
properties of the various materials that constitute the composite.
Generally, impact testing consists in striking a structure with an indenter or projectile according to a
certain mechanism that allows for the calculation of the impact energy and speed. The type of test must
simulate the typical impact scenario the structure will suffer. Gas guns, pendulum and drop-weight are
the most common impact testing types. Gas guns shoots a projectile, and therefore, is suitable to
simulate medium or high velocity impacts, usually, with small projectiles. In case of high mass low
velocity impacts, drop-weight testing is preferable. Whenever it is possible, tests are instrumented, to
register as much information as possible. Simple LED devices are used to accurately measure the
velocity before impact, and load cells can be attached to the falling weights or pendulums to register
force history. Other impact tests can be performed, to simulate less common impact scenarios, as
explosions and slamming.
6
American Society for Testing and Materials has developed several standard tests for damage resistance
of composites structures, both for static and impact loads, as well as material properties.
The fact that composite structures are heterogeneous and anisotropic, leads to a large variety of failure
modes. The four main failure modes are:
Matrix damage - Matrix damage usually takes the form of matrix cracking and fibre/matrix
debonding. In the upper layer, matrix cracks are induced by stress concentration at the contact
edges of the impactor. In intermediate layers, shear cracks are formed by high shear stresses
through the material, while in the back layer tensile stresses generated by plate bending are the
main cause of matrix damage. Matrix crack tend to arrange in a complicated and unpredictable
way. However, they do not contribute significantly to the reduction of residual properties of the
laminate. Usually, thin laminates tend to start matrix damage in the back face, due to excessive
bending, while thick laminates start matrix damage in the impact face, due to high impactor
contact forces.
Debonding/delamination – The debonding or delamination of two adjacent laminas present a
great concern in composite impact, since it represents a significant reduction in the strength of
the laminate. Delaminations occurs in the resin rich area between two adjacent plies and are
initiated by matrix cracks and induced by mismatching bending stiffness of those plies, due to
different reinforcement orientations and its orthotropic properties. After a small energy threshold,
related with matrix crack initiation and development, delamination area varies linearly with
impactor kinetic energy.
Fibre breakage – Fibre failure is the last failure mode before penetration. It occurs on the impact
face, due to high stresses and shear forces related with indentation and in the back face, due
to high bending stresses.
Penetration – Is the last failure mode. When fibre failure has reached all the laminate thickness,
it allows the impactor to penetrate the laminate. Shear-out and impactor friction are factors that
can absorb a large percentage of the initial energy.
These failure modes are relative to single skin laminates. When dealing with sandwich structures, core
failure modes have a great importance on the overall laminate damage. The failure modes in a
composite plate can be classified as follows [4]:
Face crushing
Face shear failure
In-plane failure of faces
Flexural failure of faces
Core buckling/crushing
Core shear failure
Core/face debonding
Nevertheless, before damage can be detected by visual inspection, internal cracks, delaminations and
even fibre failure can be developed. An apparently intact structure can have internal damage that will
7
promote a premature failure. To avoid those situations, some damage assessment techniques shall be
used. Nondestructive evaluation techniques (NDE) are necessary to analyze the existence, location and
extent of internal damage in in-service composite structures.
When dealing with translucent structures (glass or aramid single skin reinforced structures), damage
can be located and measured by using a strong backlight. However, this technique is difficult to use in
complex structures or in structures that have coatings, as is the case of hulls and other marine parts.
Other simple NDE technique is the acoustic impact, also known as “coin tap”. The evaluation requires
the inspector or recording device to distinguish changes in the frequency of the sound the material
emanates when tapped with a hard object. A variety of other NDE techniques, typically with use of
sophisticated technology, such as ultrasonic testing, X-ray, thermography,… Each method has its pros
and cons, but all of them present several challenges when applied to evaluate sandwich composites.
When the location and extent of a certain damage is known, it is then important to relate the damage
with the decrease of the capacity to withstand the loads it was design to carry. That assessment is
usually based in results from residual strength tests. They are similar to the normal strength tests, that
test the mechanical properties of a certain structure (plate, beam,…), but where the structure being
tested has previously suffered a certain level of damage.
8
2.2. Marine impact
Although there is a vast number of studies on impact on composite materials, only a small part of those
is applicable to the marine field. Marine composites are generally less sophisticated than composites
used in automotive and aerospace industries. Materials, manufacturing techniques and available design
and inspection tools can be very different. Even the design parameters and safety factors associated
with the demands for low cost and low maintenance solutions require different approaches in the
research of impact resistance.
This lead to the development of impact studies with special focus on marine composite structures,
approaching different concerns and methodologies.
A table summarizing the representative investigations into impact on marine structures that will be
reviewed in this chapter is presented in Appendix A. Most part of the research was done in institutions
dedicated to marine subjects, as Centre for Marine Technology and Engineering (CENTEC), Det Norske
Veritas (DNV), Institutt Français de Recherche pour l'exploitation de la Mer (IFREMER) and VTT
Technical Research Centre of Finland (VTT), among others. On the other side, some of the research
presented was not performed with the objective of study marine composites, as is the case of
mechanical and aerospace institutes and centers. However, due to the similarity of structures with a
wide range of applications, as is the case of sandwich structures, some results from aerospace
structures can be applied to marine structures. The expected failure modes for such kind of structures
can be the same, even if the modelling of the loads and other parameters are completely different.
The studies were divided into 5 main arbitrary sections (impact modeling, material selection, society
rules, residual strength and water absorption), although there is not a defined boarder between these
areas.
2.2.1. Impact modeling
The first main concern when studying the impact resistance of marine composite structures must be to
correctly model, numerically or experimentally the loads that the structure is expected to resist. In case
of a composite vessel, those loads can be due to accidental impacts during construction phase and
docking operations, or impacts that occur during navigation, as impact with floating objects and
slamming.
On this subject, Choqueuse et al. [5] has developed experimental research regarding different impact
scenarios in marine structures, with the main objective of getting comparative results with the developed
predictive methods. Glass reinforced plastic with PVC foam sandwich was submitted to a low energy
drop-weight test. The results were reasonably predicted by FEM analysis. The second series of tests,
consisted of carbon fibre reinforced plastic single skin and sandwich (with PVC foam and aluminum
honeycomb) were tested against slamming, using a variety of solid impactors and water bladders. In the
last series of tests, the application of composite materials in offshore applications as a substitute to steel
9
floors was proved, as the composite floors performed fairly against the accidental drop of a steel
container …
A very typical collision situation in marine structures, with special importance in high speed craft, is the
collision with drift wood or other small/medium size floating objects. A simplified formulation for impact
loads prediction was developed by Toyama [6]. The geometry of the wood log was simplified, and the
problem was solved, considering energy of impact and elastic deformations. In case of high energy, the
wood will break, and, modal analysis is required.
Ping et al. [7] developed research regarding high speed vessels. Besides collision with floating objects,
other local loadings are considered, as dropped objects on deck and low speed collisions during berthing
in a quay. The effect of these local loadings can be assessed by performing drop-weight tests, with
appropriate force, energy and impactor. In case of some specific vessels, as is the case of
minesweepers and other naval vessels, underwater or air explosions must be taken into consideration.
It was showed that much of the intensive work developed for steel can be derived for composite
structures.
DNV presented methodologies regarding the dynamic response of sandwich panels [8]. The main goal
of this methodology is to calculate natural frequencies and maximum displacement. The governing
equation assumes undamped motion and linear elastic sandwich plate with thin isotropic skins. In case
the panel is submerged, water inertia shall be included. However, damping can be neglected in the
majority of the cases, as the maximum displacements of the panel occur in the first half cycle of
movement. Additionally are presented considerations about modeling of air blasts, underwater
explosions and slamming loads.
In the field of experimental testing, VTT has presented several publications in the 1990’s, regarding
different themes. Two publications were presented, regarding experimental test methods [4] and
empirical methods for strength prediction [9]. Standardized (ISO 6603) and non-standardized (pyramidal
shape impactor and quasi-static cylindrical) test results were compared. Non standardized tests are
used to better represent the impact scenarios the structures are subjected to. However, pyramidal shape
impactor produces a variety of failure modes, and penetration of the skins is not easily identified.
Empirical estimation methods were developed, relating absorbed energy at inner face penetration with
sandwich total thickness, panel weight and reinforcement weight. A relation considering reinforcement
weight and core thickness and material was also achieved. Lastly, a semi-empirical method was
developed, considering the 3 main stages observed in more than 70 samples: 1)-penetration of the outer
face through transverse shear failure; 2)-bending of outer face and core crushing and 3)-bending of
outer face, core crushing and penetration of inner face through transverse shear failure. Each stage
requires some mechanical properties of the different constituents as input. The results are reasonably
accurate. However, they are only based in pyramidal indenter test data.
Johnson et al. [10] performed finite element analysis with progressive damage models in large simply
supported glass woven vinylester panels, and the results compared with full scale tests. Damage
initiation is assessed using strain based functions, while damage degradation is assessed by strain
10
functions for in-plane properties and by stress functions for transverse properties. Additionally, ply
delaminations were modeled using cohesive elements. The full 3D damage model proved to be suitable
to analyze thick plates, where most of the models suggested by the literature do not include all failure
modes.
FEM can also be used to analyze sandwich panels and beams. A study to simulate static indentation
and unloading was performed by Zenkert et al. [11]. Firstly, an analytical model assuming elastic-plastic
compressive behaviour of the foam core was developed. After removal of the load, elastic unloading
response of the foam core is assumed. After, a FE model was developed, with the use of “crushable
foam” material model and higher element density near the facesheet. Input values for analytical model
and FEM were obtained from a first set of uniaxial compressive tests. Indentation tests were also
performed on actual beams, and compared with analytical and FEM results. Both models could predict
the response of sandwich beams under static indentation and the residual indentation in the face sheet
after unloading. The predicted residual strains were compared with the experimental ones, through DSP
technique.
Although numerical models can give reasonable accurate results for both impact tests, they often require
material properties not easily available in the literature and difficult to obtain from experimental tests, for
instance, ASTM standards.
The use of cheaper quasi-static tests to assess the impact resistance of marine GRP was studied by
Sutherland and Guedes Soares [12]. WR, CSM and CP single skin laminates with several thicknesses
were impacted and quasi-statically tested. For all material systems considered, quasi-static tests gave
a good approximation of the initiation of delamination and dynamic force-deflection behaviour at low to
medium incident energies and displacements. However, if the study concerns about fibre failure or
absorbed energy, the results will be conservative, especially in the thinnest specimens.
Quasi-static indentation tests on E-glass/polyester marine laminates was also performed in CENTEC
[13]. A Hertzian contact law fitted well the initial response. Irregularities on the surface were responsible
for a contact force lower that predicted, especially in WR specimens with small diameter indenters.
These irregularities are mainly resin-rich surfaces that reduce the contact stiffness. At higher loads the
response became linear as damage became significant. It was presented the hypothesis that this
transition is due to delamination. During the study was also verified the correlation of the transition load,
from Hertzian to linear behaviour, with the indenter radius.
Other important factor in experimental testing, namely in impact and quasi-static testing are the scale
effects. Scale modeling is attractive because of the lower costs and the ease, comparing with full scale
testing.
Scaling issues concerning impact on single skin GRP laminates were also studied in CENTEC [14]. In
a simplified case, when target and impactor head are geometrically scaled by a scaling factor s, the
response should also be scaled geometrically by s. The duration of the impact is scaled by s, the force
by s2 and absorbed energy by s3. However, many aspects were not taken into consideration: non-
homogeneous and non-isotropic nature of composites, effect of damage, contact stiffness and probably
11
strain rate effects. If these effects are seen in the model but not in the prototype, the scaling will be
distorted. After a set of impact tests, was proven that larger specimens present relatively lower load and
displacement at fibre failure, less membrane effects and more irreversibly absorbed energy (even before
fibre failure) than smaller specimens.
Davies [15] studied the effect of scaling in impact testing of GRP-PVC sandwich laminates, with two
different ratios between size and thickness, to replicate shear and bending predominant behaviours The
geometrical scaling factor was visible on the experimental results, however, better for the panels where
shear is predominant. These results showed a distorted model when bending and membrane effects
are predominant.
Zenkert et al. [16] developed a methodology to assess the damage tolerance of carbon fibre composite
sandwich panels with localized damage. A series of experimental tests (impact, indentation and residual
compressive strength) were the basis of comparison to the prediction models of residual strength of
panels based on the damage. These prediction models were then integrated in a damage assessment
model. The model estimates the load carrying capacity of a certain panel after a damage, and if it is
enough to withstand the local loads. In case it is not enough, the effect on the global structure is
assessed. If it is possible to reduce the operational loads, operational restrictions are imposed to the
ship, and the damages repaired later. Otherwise, an emergency repair has to be done. Besides strength,
the methodology can be applied to other operational requirements, as watertightness, fire division
effectiveness, among others.
Hayman [17] developed a similar work concerning the assessment of damage in marine FRP sandwich
structures. A response monitoring systems based in sensors can be used to indicate when the loads on
the structure have exceeded a level that is likely to have caused damage, as indication of actual damage
is far more difficult. Based on the expected damage of the structure, some operational restrictions can
be imposed to the ship.
Collombet et al. [18] used an experimental method called the response surface method, to study the
response of laminated and sandwich composite structures under impact loading (low velocity-high
mass). The method allows to calculate coupling coefficients between mass and velocity, for each
response (contact force, displacement of impactor, contact time and structure damage). Apart from
contact time, in the other responses it seems that mass and velocity are coupled. In terms of sandwich
structures, although the mechanical behaviour and failure modes are different, local core crushing is not
very affected by the core properties (ductile or rigid foam). However, rigid foam induces more
delamination in lower skin than ductile foam. For same energy, impactors with more mass generate
more delamination on lower skin of rigid foam sandwich.The results underline that the kinetic energy of
the striker should not always considered as the representative parameter of the tests.
Guillaumat [19] developed a similar work for single skin, based on the same principles as previous
studies [18]. It is also suggested that for single skin, the energy of the impact may not be sufficient to
describe the impact loading.
12
2.2.2. Material selection and structural solutions
An extensive experimental work regarding single skin composites was performed over the last years in
CENTEC [14], [20]. It was proven that the high performance of a certain solution under impact depends
on many parameters, thus the difficulty to standardize impact test and design. Besides material
properties and manufacturing techniques, structural configuration has a main importance on the damage
failure modes.
Extensive work considering material selection has been developed by DNV [21], [22]. A series of tests
was performed on hand lay-up panels with a variety of materials (glass fibre, aramid, carbon and
aluminum). Aramid and carbon plates withstand less energy impact than glass fibre plates. Aramid and
carbons’ performances increase when considering panel’s weight. However, aluminum stills present the
highest impact strength, especially when considering plate thickness.
In a more generalized way, several materials utilized in marine industry were tested according to the
ISO 6603 rule by Hildebrand [23]. Plywood, FRP-Plywood, FRP, ABS, PE, PC and aluminium were
tested and compared in terms of impact strength and specific impact strength. Aluminium absorbs more
energy, but taking into consideration material density, PC presents almost the double the specific
absorbed energy of the aluminium, due to higher plastic deformation. However, PC is only used in the
manufacturing of windows and hatches.
Wiese et al. [24] performed oblique impact tests on a series of sandwich panels, with different
thicknesses and cores (PVC foam and balsa). According to this study, core material has little influence
on the impact strength, in terms of energy required to penetrate de outer skin. Balsa panels present
delaminations on the back skin earlier (in terms of energy) than PVC panels. One reason for this is the
different core failure mode for PVC foam (conical) and balsa (cylindrical).
Atas and Sevim [25] performed series of single and multi-impact tests on E-glass epoxy sandwich
composites, with balsa and PVC foam cores. The results were similar to the ones obtained by Wiese et
al. [24]. Balsa core tends to promote more back skin delaminations than PVC, while PVC promotes
more delaminations on the impact skin. Balsa panels also show a core/back skin debonding, for higher
energies. Compression tests of the core showed core densification at around 60-70% strain. A study
relating the number of impacts until perforation for several energies was also performed. The number of
repetitions until perforation increases with decreasing impact energy. However, the total absorbed
energy by the sandwich tends to increase with the increasing number of repetitions.
Gustin et al. [26] developed an experimental comparison on the impact strength of sandwich composites
with honeycomb core, epoxy resin and skins made from carbon, aramid and hybrid reinforcements. The
replacement of carbon by aramid in the outer skin improved the maximum absorbed energy and
minimized the loss of residual compressive strength on panels with complete perforation.
The application of cork in high performance sailing craft sandwich structures was studied by Dumont
and Blake [27]. Corecell M80 and Corecork NL20 were used to manufacture fibreglass epoxy sandwich
panels. The panels were compared in a 3-point bending test and in a drop slamming test. In the bending
13
tests, apart from the different failure modes, the panels with cork core are almost equivalent to the PVC
ones. The dynamic tests showed a lower rigidity of the cork panels, hence the higher defections of the
panels. Due to the higher strains the cork panels were subjected, and also due to the higher weight of
the panels (+53%), it was concluded that these panels are not suitable for high performance
applications.
Echtermeyer et al. [28] developed a study concerning composite structures for high speed craft.
Slamming loads are the more common impact loads on high speed craft. However, it was concluded
that collision with floating objects is more critical than slamming loads. Was developed a method for
designing lighter hybrid glass-aramid laminates, maintaining the impact strength of pure glass fibre
panels. Assuming pure bending of the panel and maximum strain accepted by glass and aramid, glass
fibre in the middle of the panel was replaces by aramid.
Hildebrand [29] developed and tested methods for improving the impact strength of sandwich panels for
ship applications The basic laminate under study was a glass-epoxy sandwich with aluminium
honeycomb core. Some parameter were changed during the experiment: skin core bonding, skin
reinforcements, skin matrix, increase of 1 [mm] in skin thickness and also the addition of 1 or 2 [mm] of
PU coating. After impact test with a pyramidal impactor, elastomeric resins (elastomeric epoxy and
rubber modified vinylester) tend to increase the impact strength at penetration of outer skin. Also, PU
elastomeric coatings proven to be more effective that additional thickness, when considering specific
impact strength.
Hildebrand [30] presented an intensive literature review on the effects of strain-rate on the impact
strength of FRP sandwich skin and core materials. Typically, the strength of the materials increase with
increasing strain rate. The fact that most of material properties for the design of composites are based
on quasi-static tests can induce the choice a material that is not the best for the impact scenario. With
the increase of strain rates from quasi-static to impact scenarios (typically, 4 decades increase for
collision with floating object and slamming), the increase in strength can be up to 40 [%] for glass
polyester skins and up to 80 [%] for some low density PVC core foams.
Baral et al. [31] studied the use of through-thickness reinforced foams to improve the performance of
structures subjected to water impact. Carbon fibre sandwich panels were built with low density PVC
foam, honeycomb and PVC foam with through-thickness pultruded carbon fibre pins. The large clamped
panels were impact tested with a large elastomeric ball. The test results show that pinned foams perform
better then honeycomb under slamming or distributed impact scenarios. However, the pin diameter and
spacing can be further improved.
Mitra [32] proposed a method to improve the shear strength of marine grade sandwich laminates. A
shear key was embedded in the sandwich structure, through groves filled with glass fibre and
subsequent vacuum bagging. The specimens were tested according to ASTM C273 standard and the
results compared with numerical simulations. The method proven to be a cost effective way to increase
in-plane shear stiffness and strength. It is also mentioned the possibility of using this method to improve
the face-core bonding resistance.
14
Imielinska et al. [33] studied the effect of face-core bonding in the impact resistance of glass-PVC
sandwich panels.
Perrot et al. [34] studied the sensitivity of glass reinforced plastic to the brittleness of the low styrene
resins. Low styrene and low emission styrene polyester resins were compared with traditional polyester
and vinylester resins. A wide variety of tests was performed both in the resins and composites (fracture
toughness, debonding, impact,…) Panels with low styrene and low emission styrene reveal larger
projected damage area than traditional resins and present easier damage propagation than traditional
polyester and vinylester. However, damage initiation doesn’t seem to be affected. Through the variety
of performed tests, it was proved that is the brittleness that causes the low damage resistance.
Hayman et al. [35] presented a series of results from a European project, concerned with the
development of a large-scale use of fibre composites in naval ships. The wide and comprehensive
variety of results approaching material performance, static and extreme dynamic loads, inspection and
damage repair helped to overcome major limitations and is considered to be a large step towards the
increase of composites in naval applications.
2.2.3. Impact behaviour comparisons of Classification Society Rules
Aamlid [36] performed a series of impact tests regarding aluminium, single skin and sandwich FRP
laminates. The results were compared with DNV’s High Speed Light Craft rules. In collision with small
objects, the impact load is carried out by the plate. However, the minimum thickness structural criteria
(by the rules) is dependent on the stiffener spacing and craft length. This will induce a great difference
in the impact resistance of two FRP craft with different sizes (20 [m] and 100 [m], for instance).
Concerning sandwich constructions, the minimum thickness applies to both skins. When the skin of a
sandwich is very thin, it withstands lower impact energies than a single skin laminate with the same
thickness. However, the impact resistance of the outer skin in a sandwich plate increases rapidly with
the increase in thickness.
Nowadays, HSLC rules’ thickness criteria are based only on ship length, and do not represent an actual
thickness. Instead, the rules specify the minimum weight of reinforcement, depending of the hull
structural member.
Pedersen and Zhang [37] developed a mathematical formulation to obtain the critical impact energy for
a given shell structure, when colliding with a solid floating object, with mass much smaller that the craft.
The formulation assumes that the critical energy induces the maximum admissible strain of the panel.
The results were compared with the previous experimental work by DNV [21], and showed reasonable
agreement.
2.2.4. Residual strength
Davies et al. [38] developed a series of compressive tests on single skin FRF panels, previously
subjected to drop-weight impact tests. Two concepts of damage tolerance are approached. The first
concept, are the thresholds (normally, in terms of impact energy) that state the beginning of structural
15
damage. Ultrasonic C-scan and cross sectioning of impacted plates showed that this structural damage
starts with the outer plies shear out, with consequent drop in impact force. The second concept of
damage tolerance is related with the load carrying capacity of the plate after impact, typically,
compressive strength. Especially in thin single skin laminates, the compressive strength is very much
affected by the increase of damage and impact energy.
Auerkari [39] presented a study where the diminution of the lateral buckling load is related with impact
energy (W) and number of repetitions (N):
𝑃 = 𝑃0
(
1−𝑊
√𝑁𝐾)
( 4 )
The decrease of buckling load can be also related with the ratio of damage diameter (du) and effective
width under loading (d):
𝑃 = 𝑃0 (1 − (1.5
𝑑𝑢𝑑)3
) ( 5 )
Both equations fit well experimental results of carbon-PVC sandwich plates previously subjected to
impact loads. These and other prediction methods are especially useful for evaluating the residual
strength of sandwich structures, where NDT methods hardly present accurate results.
Shipsha and Zenkert [40] performed CAI tests on foam-cored sandwich panels subjected to low-velocity
semi-spherical impact. The main impact damage was core crushing with facesheet permanent
indentation. The CAI results present a linear load-compressive displacement load, with a small
decreasing in stiffness near the break point, and the maximum compressive load revealed a linear
relation with the dent diameter. During the compression tests, it was stated that the facesheet dent
increased. The final compressive failure was the complete separation of the facesheets from the core.
It was suggested that the failure was induced by an outward displacement in the neighborhood of the
facesheet dent, which, under compression, lead to an out-of-plane tensile failure between facesheet
and core.
Most of the studies regarding residual strength of plates consist in the measure of a concentrated load,
transversal or axial. However, for marine purposes, it is important to know the residual strength in
presence of a distributed load. Auerkari and Pankakoski [41] measured the residual strength of glass-
epoxy balsa sandwich panels with the use of a rubber bladder filled with water. The panels were
previously impacted with a pyramidal impactor. The main failure mode induced by the bladder pressure
was the extension of the delaminations until the panel´s edges. Comparison with intact panels showed
that the loss of strength was less than 20 [%]. However, when the impact damage reached the panel
edges, the loss was more severe. This indicates that the results depend on the panel size, and possibly,
on the type of impactor.
16
Typically, impact tests consist in 3D approaches to the impact problem (sphere colliding with a plate).
Although 3D approaches represent better the real cases of collisions, the problem can be studied using
a 2D model (cylinder colliding with a beam) [42], [43].In the first part of the study, PVC cored sandwich
beams with E-Glass skins were impacted with a steel cylinder with energies up to 40 [J]. In the higher
energy tests, some delaminations occurred in the face sheet, associated with some loss of bending
stiffness (< 5 [%]). The core presented a cavity (core crushing) right under the impact zone, followed by
a zone of compacted core. The beams were also statically indented and the compressive strain was
measured using DSP technology. These two types of tests allowed to conclude that serious damage
can occur in the core, even in cases where there is no visible damage in the facesheet. This core
damage affected the shear strength and critical buckling stress of the beams. In the second part of the
study, FEM modeling of the impacted beams was performed considering the geometry of the core
crushing. The FEM predictions of residual shear strength were good. However, buckling stresses were
overestimated with respect to the experimental results.
Shipsha and Zenkert [44] studied the effect of cyclic constant loads on composite beams with sub-
interface impact damage. The beams and impact testing followed previous work [42], [43]. The cyclic
load was chosen from static residual strength, and two load ratios were tested (tension-tension and
tension-compression). After the fatigue of the “bridged zone” (zone were crushed core connects to
facesheet), the behaviour of the impact damaged specimens was similar to specimens built with face-
core localized delamination. From this point, once the fatigue crack initiates, it rapidly propagates
through the thickness of the core, leading to failure. During the fatigue tests, it was registered a small
decrease in stiffness. Finally, using results from intact and damaged beams, at different cyclic loads, S-
N curves were built. For beams with impact damage, the estimated fatigue threshold load was
approximately 35% of the ultimate static strength.
2.2.5. Water absorption
Gu and Hongxia [45] studied the effect of water absorption on the delamination behaviour of glass-
polyester composites. Two layer composites were tested for peeling strength. It was noticed that the
peeling strength increased with immersion days. Water may act as a plasticizer, making the matrix more
even, and thus, increasing the peel strength. However, as time passes, the water starts to degrade the
matrix, leading to a decrease in peeling strength.
Boukhoulda et al. [46] studied the effect of water absorption in the impact resistance of glass-polyester
composites. The plates were aged with temperature and moist environment, and impact tested
afterwards. It was noted that maximum impact force decreases with increasing water absorption, and
that maximum deflection increases with increasing water absorption. In terms of delaminations, it was
seen that the delamination area decreases with increasing absorbed water, due to plasticization.
However, after 0.48 [%] of absorbed water, the delamination are tends to increase, because of
irreversible damage between fibre and matrix, induced by the water absorption.
Imielinska and Guillaumat [47] developed a similar study concerning glass-aramid epoxy laminates.
Besides the effect of water absorption in plasticization and delamination reduction, it was studied the
17
effect on the residual compressive strength. The compressive tests revealed a significant reduction on
compressive strength, namely, 28 [%] for intact plates and 42 [%] for impacted specimens.
Weitsman et al. [48] studied the effects of water on polymeric foams and respective sandwich
composites. Typically, sea water causes the PVC foam cells to swell. By means of drying the foam
samples previously exposes to water, it is stated that foams can absorb large quantity of water,
depending on the foam density. To assess the effect of water in the face-core bond, peel tests were
performed in glass-PVC polyester sandwich composite, with significant drop in peel strength.
2.3. Conclusions
During the research prior to the revision of the state of the art, it was noticeable the large amount of
research on the theme of impact in composite materials. That amount and variety of research proves
that this is an area with several challenges and with a lot of specificities and variables, making each
study different from the previous. Those specificities make that most of this research is not directly
applicable to marine scenarios, either because the utilization of completely different materials and
manufacturing methods or because the different requirements and actual impact scenarios.
Still, was presented a significant amount of knowledge not specifically developed for marine
applications, but whose results can be considered.
The studies with a marine scope cover almost all significant areas, but still, there is a lot to do and to
understand, in case of this work, concerning impact in sandwich composite structures
19
3. Experimental procedures
This chapter has the main goal of explain the process that lead to the final data and results. All design
criteria and assumptions will be presented. However, this was not a straight forward process. As in most
of the design processes, it required an iterative process between sandwich layout, material mechanical
properties and experimental tests parameters, which are related to the relevant impact scenario.
3.1. Typical marine sandwich
From the previous literature review, one can state that there is an extensive variety of composite
structures layouts. Even considering only the group of marine sandwich composites, there is variety of
resins, reinforcement and core materials and structural forms. Besides that, the selection of the
fabrication method is also a parameter that affects the final product.
So, what can be considered a “typical marine sandwich composite”? Apart from high performance
applications, such as racing and sailing craft, marine applications tend to require low-cost and low-
maintenance solutions, as shown by the extensive use of GRP in all kind of recreational crafts and small
work and fishing boats. So, a typical marine sandwich composite would have glass-polyester skins, and
would use a relatively cheap core material, as PVC foam.
Currently, there is a European project being developed by 11 partners, to which Instituto Superior
Técnico (IST) and Estaleiros Navais de Peniche (ENP) belong. The main goal is to improve the response
of merchant and passenger ships to wave loads and corrosion, through the right selection of materials
and structural solutions. The solutions being studies are the use of HSLA (High Strength Low Alloyed
Steels) in critical areas where cracking is supposed to become an issue, and the use of composite
materials in some parts of the ship, to reduce weight (superstructures, transverse bulkheads and partial
decks).
The sandwich layout under study in MOSAIC [49] is a sandwich with 30 mm core (PVC foam and end-
grain balsa) with approximately 2.5 [mm] E-glass reinforced skins (variations of 2 to 4 plies, hand lay-
up and vacuum bagging, epoxy and vinylester resins).
It was decided to use a similar layout for the sandwiches under study in this work by several reasons:
it is a layout that is being studied, with several applications in commercial solutions.
the panels will be manufactured by ENP, that already manufactured some MOSAIC panels. This
is reflected in very helpful know-how during manufacturing and specimen preparation.
the MOSAIC panels were designed/dimensioned so they can be tested accordingly to the
standards.
20
3.2. Relevant impact scenario
The state of the art presented a large variety of impact scenarios that marine structures can face during
its lifespan. However, with such a variety of impact scenarios, it may be difficult to select one. When
studying a local impact, such a tool drop or a pier collision, the ideal would be to perform tests in panels
with typical size and considering the reinforcement spacing. When considering slamming in a small high
speed boat, it would be good to have a scale model of the boat and have it tested in a tank test. But in
a majority of the cases, there are no resources that allow to perform these kind of tests.
For the development of this work, there is a hydraulic test machine and a drop-weight machine, and
therefore, the selection of an impact scenario must take that into consideration. This equipment requires
the use of small specimens, which will certainly not match the actual behaviour of the panels in a real
life situation. The case of a small rectangular plate centrally impacted with a small diameter
hemispherical impactor was also selected to reproduce a severe case of impact.
In the end, the tests will be a basis of comparison for the different core materials, showing how the
different materials behave in a sandwich structure. The results will allow the designer to select the best
material for each application.
3.3. Selection of the experimental tests
3.3.1. Bending test
ASTM C 393 [50] was developed to experimentally determine the core shear properties of flat sandwich
constructions. The test consists in a 3 or 4 point bending, with recording of displacement and force.
More than checking the bending stiffness of the beam, the test’s main goal is to provide information
about how the core behaves when applied in a short beam structure.
The specimen must be considered a short beam specimen, and the test conditions must induce the
failure mode in core shear or core-to-face bond. Hence, the standard recommends the verification of
two criteria related with the core shear and compressive properties. Besides the guidelines to design
the specimens, the standard provides general information for the test procedure. For instance, data
recording and test speed.
3.3.2. Static indentation test
Quasi-static indentation tests consist in the application of a concentrated load, through the compression
of an indenter against a specimen. ASTM D 6264 [51] was developed to provide guidance when testing
monolithic composite laminates. However, it may be useful to test other types of composite materials,
as the ones in studied in this work.
21
This test method was selected to develop a basis of comparison for the impact tests that will be
performed later. Failure modes and energy absorption are expected to vary significantly from one
situation to another.
3.3.3. Impact test
Impact tests consist in measuring the damage resistance of a structure to a sudden load. Typically, that
load is applied in the specimen through a falling weight or a pendulum.
ASTM D 7136 [52] presents a general procedure and guidance to perform tests to measure the impact
resistance of general fibre reinforced composites in a drop weight event. It also provides guidance about
data recording and post processing.
3.4. Specimen preparation
3.4.1. Material selection and sandwich design
From the standard test ASTM C 393, core compressive and shear properties must comply with the
following criteria:
𝐹𝑠 ≤
2𝑘𝜎𝑡
(𝑆 − 𝐿)
(6)
𝐹𝑐 ≥
2(𝑐 + 𝑡)𝜎𝑡
(𝑆 − 𝐿)𝑙𝑝𝑎𝑑
(7)
where:
𝑆 = 𝑠𝑢𝑝𝑝𝑜𝑟𝑡 𝑠𝑝𝑎𝑛 [𝑚𝑚]
𝐹𝑠 = 𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 𝑐𝑜𝑟𝑒 𝑠ℎ𝑒𝑎𝑟 𝑠𝑡𝑟𝑒𝑛𝑔𝑡ℎ [𝑀𝑃𝑎]
𝐹𝑐 = 𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 𝑐𝑜𝑟𝑒 𝑐𝑜𝑚𝑝𝑟𝑒𝑠𝑠𝑖𝑣𝑒 𝑠𝑡𝑟𝑒𝑛𝑔𝑡ℎ [𝑀𝑃𝑎]
𝜎 = 𝑒𝑠𝑡𝑖𝑚𝑎𝑡𝑒𝑑 𝑓𝑎𝑐𝑖𝑛𝑔 𝑢𝑙𝑡𝑖𝑚𝑎𝑡𝑒 𝑠𝑡𝑟𝑒𝑛𝑔𝑡ℎ [𝑀𝑃𝑎]
𝑡 = 𝑓𝑎𝑐𝑖𝑛𝑔 𝑡ℎ𝑖𝑐𝑘𝑛𝑒𝑠𝑠 [𝑚𝑚]
𝑐 = 𝑐𝑜𝑟𝑒 𝑡ℎ𝑖𝑐𝑘𝑛𝑒𝑠𝑠 [𝑚𝑚]
𝐿 = 𝑙𝑜𝑎𝑑𝑖𝑛𝑔 𝑠𝑝𝑎𝑛 𝑙𝑒𝑛𝑔𝑡ℎ [𝑚𝑚] (0 𝑓𝑜𝑟 3 𝑝𝑜𝑖𝑛𝑡 𝑙𝑜𝑎𝑑𝑖𝑛𝑔)
𝑘 = 𝑓𝑎𝑐𝑖𝑛𝑔 𝑠𝑡𝑟𝑒𝑛𝑔𝑡ℎ [𝑀𝑃𝑎] (0.75 𝑟𝑒𝑐𝑜𝑚𝑒𝑛𝑑𝑒𝑑)
𝑙𝑝𝑎𝑑 = 𝑑𝑖𝑚𝑒𝑛𝑠𝑖𝑜𝑛 𝑜𝑓 𝑙𝑜𝑎𝑑𝑖𝑛𝑔 𝑝𝑎𝑑, 𝑖𝑛 𝑠𝑝𝑒𝑐𝑖𝑚𝑒𝑛 𝑙𝑒𝑛𝑔𝑡ℎ𝑤𝑖𝑠𝑒 𝑑𝑖𝑟𝑒𝑐𝑡𝑖𝑜𝑛 [𝑚𝑚]
22
Equation 6 represents the maximum shear that the beam can carry without exceeding the ultimate
strength of the facings. If the skins are very thin or the core has a high shear strength, the skins will fail
prematurely. Hence, the expected failure of the core will not be visible. On the other hand, the core must
have enough compressive strength to resist the pressure applied by the loading pad. This can be
outlined by increasing the radius of the loading pad.
MOSAIC specimens were tested with a span of 280 [mm]. However, in this work, the materials used
present a wide range of mechanical properties, with a great difference between the properties of
Corecork NL10/NL20 and Baltek 100, especially considering the compressive strength (Table 1). It was
impossible to set a length for the bending test that could comply with the criteria for all core materials.
However, the length of the specimens must be the same, enable valid comparisons. It was decided to
produce the beams long enough, and then, experiment what would be the bending span, so that the
shear failure is visible in all specimens.
About the width of the specimen, the standard specifies that the width shall not be less than twice the
total thickness, nor greater than one half of the span length. A width of approximately 70 [mm] was used.
Table 1 - Relevant mechanical properties of the core materials utilized
Corecork NL10
Corecork NL20
Divinycell H100
Baltek 100
Density [Kg/m³] 120 200 100 153
Shear modulus [Mpa] 5.9 5.9 35 309
Shear strength [Mpa] 0.9 0.9 1.6 3
Compressive modulus [Mpa] 5.1 6 135 4005
Compressive strength [Mpa] 0.3 0.5 2 12.9
3.4.2. Panel manufacturing
After the gathering of all the materials, the panels were manufactured in the facilities of ENP. Amorim
Cork Composites handed over plates of Corecork NL10 and NL20. Besides the facilities and technical
support during manufacturing, ENP also provided the remaining core materials, glass reinforcement,
resin and catalyst, consumables for vacuum bagging and safety equipment.
Table 2 presents the constituent materials of the several manufactured panels. Due to availability of
materials and other factors, it was decided to produce 2 panels with each one of the core material.
Regarding balsa, only one panel was produced. However, the area of the panel was larger than the
others, allowing to produce a reasonable number of specimens.
Besides the constituent materials, some consumables were utilized. The panels were manufactured
through vacuum bagging technique (Figure 2), requiring peel ply, breather fabric, vacuum bag film,
sealant tape and plastic tube.
23
Table 2 - Materials utilized in the manufacture of the sandwich panels
Material Quantity
Corecork NL10 2 * 1000 [mm] * 500 [mm] * 30 [mm]
Corecork NL20 2 * 1000 [mm] * 500 [mm] * 30 [mm]
Divinycell H100 2 * 1000 [mm] * 500 [mm] * 30 [mm]
Baltek 100 1 * 1219.2 [mm] * 609.6 [mm] * 31.75 [mm]
E-Glass LT800 E 10ª 0/90 Biaxial 35 [m] * 1.4 [m]
Resin Crystic 489 PA BT LV 22.40 [Kg]
Catalyst Butanox M-50 424 [mL]
Figure 2 - Vacuum bagging technique [53]
The panels were built in two different stages, corresponding to both skins. The sequence for each skin
is described under:
The reinforcements, peel ply, breather and vacuum bag were cut previously, with larger
dimensions than the core plates (+20 [mm] for reinforcements, +100 [mm] for peel ply and
breather). The vacuum bag was cut much larger than the core plates (+500 [mm]), to allow the
bag to crimp without stretch and eventually tear. The sealant tape was applied to the bag,
without removing the paper
To the core plates, resin dissolved in acetone was applied with a paint roller, to reduce the
porosity and increase adhesion. Three holes were drilled in the centerline of the plates, to allow
the resin to flow easily from bottom to top.
The resin was mixed with the catalyst, with and approximate ratio of 20 [mL/Kg].
On the lamination platform, previously cleaned, the 4 layers of reinforcement, corresponding to
one skin, were laminated with a roller.
After lamination of the skin, the core plate was placed on top, followed by the peel ply and
breather. The bag was placed on top and glued to the table, making use of wrinkles on the
corners and around the vacuum tube.
24
The vacuum was applied, with a relative pressure of -0.8 [bar], and was kept during nearly 6
hours. The vacuum compressed the core plate against the skin, and removed the excess of
resin to the breather.
After the setting of the resin, the vacuum was removed, as well as the bag and peel ply. The
edges of the panel were cut and grinded.
The process was repeated for the second skin of each panel. Up to 4 sides were laminated
simultaneously, due to the use of a vacuum pump with 4 intakes.
During and after the completion of the manufacture of the panels, some facts particularities were
observed:
The cork panels absorb resin, especially the low density panel (NL10). In Figure 3, it is possible
to state the differences. In Corecork NL10 panel, almost no resin is visible on the breather fabric,
while on the Corecork NL20 panel, there is a significant amount of resin on the fabric, near the
edges, but also scattered in the middle of the panel. In the Divinycell panel, which does not
absorb resin at all, the resin flows to the top through the edges and holes in the middle of the
panel. Although the resin amount in each skin prior to application of vacuum may vary, one can
state that the less dense cork is more porous that the high dense one, and that reflects in the
amount of resin absorbed.
The skin texture presents a rather irregular surface. It is believed to be related with some small
breaches in the sealant tape, which let some air in. Also, the fact that the reinforcement is
bidirectional, with large diameter bundles may explain this phenomena. Typically, one or more
low to medium density CSM layers are applied against the mold, to create a smoother surface.
The panels were named with the type of core material (NL10, NL20, PVC and Balsa) followed by a
number (1 or 2).
Figure 3 - Resin flow to the top of the panel, during vacuum bagging. Corecork NL10, NL20 and Divinycell H100 (from left to right)
25
3.4.3. Specimen cutting and preparation
After the manufacture of the panels, they needed to be cut into beams and small plates, to perform the
tests. This revealed one of the major challenges during the progress of the work.
Typically, single skin fiberglass can be easily cut using a universal saw, with a diamond surrounded disk.
However, with the increase in thickness or in a sandwich panel, this method is not very effective. The
disk cannot extract all the dust, starts to overheat and can easily jam. The attempts with this technique
proved to be impossible to cut more than a few centimeters through the cork panels. In the PVC and
balsa panels, they could be cut with much effort, but the cut was of very poor quality (not straight nor
completely perpendicular to the face).
To overcome this problem, a wood saw was used, with a toothed disk. This way, the previous problems
disappear. This type of disk had no problem with the very fine and sticky dust from cork cores, and the
alignment of the cut was perfect. However, the teeth of the disk tend to rip the reinforcement fibres from
the lower skin. This is more or less severe depending on the properties of the skin. MOSAIC specimens
were cut with this saw without any problem, but the attempt to cut specimens from the panels of this
work revealed that the fibers were ripped, as seen in Figure 4. This shows that the resin has much
influence in the skin and skin’s surface properties.
One could use this method to cut the rectangular specimens for static indentation and impact test, as
the ripped fibers do not affect the behaviour of the specimen. However, for the beams, it was thought
these imperfections could induce a premature failure of the beams. Because of this, the search for a
suitable method to cut both specimens continued.
Figure 4 - Detail of a beam cut with a toothed disk
Finally, the cut of the specimens was tried in a water refrigerated diamond disk saw (Figure 5), typically
used to cut stone and concrete. Due to the water flow on the disk, the dust from the skin and core can
be easily removed from the center of the panel, avoiding the jamming of the machine. The only
disadvantage of this method is the use of water, which can be absorbed by the core materials, specially,
26
in the balsa panels. In these panels, the excess of water seemed to affect the skin-core bond. This was
prevented by reducing the water flow to the disk when cutting balsa panels, avoiding to soak it.
Figure 5 - Stone cutting machine
After the cut was performed, the specimens were stacked and left to dry for some days. The dimensions
of each specimen were then measured with a digital caliper. For the beams, 3 readings were taken for
width, total thickness, 1st skin and 2nd skin. For the rectangular plates, only the thicknesses were
measured. Additionally, some rectangular specimens were weighted and measured in length and width,
to estimate the average density of the panels.
Appendix B presents the properties of the specimens that were tested during this work.
The dimensions of the specimens may be used to analyze and compare results. However, in this
chapter, the dimensions, namely the thicknesses, are used to characterize the panels. Figure 6 and
Figure 7 show the thickness and the thickness of each skin of each panel, respectively.
Figure 6 - Thickness of the manufactured panels
27
Figure 7 - Skin thickness of the manufactured panels
First of all, it is easily seen the relation between the thickness of the panels and the thicknesses of the
skins. Usually, there is not much variation on the thickness of the core plates. However, there is some
variation on the thickness of the skins. The first and second skin were defined according to the common
practice in composite manufacturing of hulls. The outer skin (here called first skin, or skin on the impact
side) is typically thicker than the inner skin (second skin).
The variability in thickness of the skins is typically dependent on the amount of resin and pressure
applied with the roller on the laminate. The vacuum bagging technique is applied to bond the core to the
skin. The vacuum bag applies pressure against the core, “squeezing” the resin from the skin. This way,
the variability of the thickness should be less than in an equivalent hand lay-up skin. When
manufacturing panels with the use of vacuum bag, the viscosity of the resin and porosity of the core can
also contribute to the final thickness of the skin.
In addition, there seems to be a relation between the thickness of the skins and the porosity or capacity
to absorb resin in the core. The higher the porosity is, the thinner the skin is. NL10 and balsa panels are
the ones with more porosity and present the smallest thicknesses in the skin. The variation on the
thickness from one skin to the other, in the same panel, can also be explained by the capacity to absorb
resin. One could say that when producing the second skin, the core has less porosity, thus leaving more
resin on the skin.
In Figure 3 is possible to see the resin flow to the top of the panels. In PVC panels, the resin flows
through the edges and holes in the core, since it is non-permeable. In the NL20 panel, one can see that
the core is almost saturated with resin, since a considerable amount of the top surface presents resin
flowing from the bottom surface. On the other hand, NL10 core has higher porosity, and the resin
absorbed from the first skin is not enough to saturate the core. This leads to the fact that both skins
present almost the same thickness.
Unfortunately, this hypothesis cannot be confirmed. During the manufacturing and cutting processes,
the track of the first and second skin to be laminated was lost
28
Figure 8 - Average density of the manufactured panels
Figure 8 presents the average density for each manufactured panel. Although the density of the panels
depends in a variety of factors, such as core density and thickness, skin thickness and absorbed resin
in the core, two main remarks can be taken from the results.
The first is that there is a significant difference in the density of cork panels and PVC and balsa.
The second one is that even though the density of the NL10 cork is almost half of the NL20, the densities
of the panels produced from both cores are almost identical. And the same happens with PVC and balsa
panels, although in this case, balsa is only 1.5 times denser than the PVC foam. But the reasons for
these two cases are different. In the cork panels, the NL10 panels absorb more resin than the NL20
ones, therefore increasing the density of the final panel. Comparing the PVC and balsa panels, the
densities are similar because the PVC panels have thicker skins than Balsa ones, which compensates
the fact that Balsa core is denser that PVC.
It is also relevant to mention that NL10 panels present a sticky surface and a strong almond smell,
indicating that something interfered with the cure. Probably, the use of a specific chemical in the
manufacturing of the Corecork NL10 core pieces interfered with the cure, since all the remaining
conditions were similar to all panels manufactured (resin batch, temperature, surface preparation,…).
29
3.5. Specimen tests
For an easy documentation of the specimens and respective results, the reference of each specimen
consists in a sequence of the following codes:
F, S or D – Identifies the type of test: Flexural, Static or Dynamic (impact);
PVC, NL1, NL2 or BAL – Identifies the core material: Divinycell H100, Corecork NL10,
Corecork NL20 or Baltek 100;
1 or 2 – Identifies the number of the panel the specimens were cut from;
XX – The number of that specimen/test, for each type of test (independent of the number of the
panel).
For example, the specimen FPVC1-1 is a specimen for a flexural test, cut from the first panel of Divinycell
H100, and will be the first flexural test of a PVC specimen.
3.5.1. Bending tests
Bending tests were the first to be performed. Figure 9 presents the hydraulic test machine where the
bending and quasi-static indentation tests were done. Figure 10 presents the support that was attached
to the test rig, which allows to adjust the span of the beam under testing.
As previously mentioned, the wide range of mechanical properties of the core materials used in this
work does not comply with the standard ASTM C 393 recommendations for maximum shear strength
and minimum compressive strength. Hence, the selection of the span and loading pad were also based
in some preliminary tests.
Figure 9 - Hydraulic test machine (Leigh Sutherland)
The first set of preliminary tests consisted in variations of different spans and loading pad geometries,
all applied to PVC specimens, and with constant speeds of 0.2 [mm/s].
30
Figure 10 - Beam supports attached to the machine base (Leigh Sutherland)
Figure 11 presents results for the four cases, following the time sequence at which they were performed.
Although the use of a cylindrical roller removes sharp edges where the failure could start prematurely, it
leads to the fact that the load is virtually applied in a line, making the specimen fail prematurely. The use
of plates instead of the roller seems to distribute the load evenly. In the test with the 40 [mm] load plate,
the specimen withstands more deformation, while shear yield is also better observed. However, the use
of a large plate under the cylindrical roller induces high instability of the plate.
Figure 11 – Load vs displacement for different spans and loading pad configurations (unfiltered results)
The test with 200 [mm] span presents a curve similar to the test with 280 [mm] and 20 [mm] plate.
However, in the 280 [mm], the yield is easily visible in the load vs displacement plot. The span of 280
[mm] was then chosen for the remaining tests.
ASTM C393 recommends to set the testing speed to produce the failure within 3 to 6 minutes. For the
span test conditions selected (span equal to 280 [mm] and load plate with 20 [mm]), this correspond to
a speed between 0.07 and 0.03 [mm/s]. The first test, FPVC1-1, was performed with a speed of 0.05
[mm/s].
31
After, FNL11-1 was performed, with the same speed. The test became too long, because the specimen
bends too much without breaking (Figure 12). This proved to be very time consuming, and revealed
some issues with the data acquisition system, due to the sample size.
Figure 12 - FNL11-1 test, showing excessive bending
For the next tests, FNL12-1 and FBAL1-1, it was decided to use a speed of 0.2 [mm/s], which revealed
to be a reasonably good speed for NL20 beams, but fast enough for balsa ones.
Figure 13 presents the results for the first round of tests, at different speeds. These and all the results
presented in this work were “smoothed”, using a moving average with 7 points.
Figure 13 – Load vs displacement for first round of tests
To avoid the existing limitations on the data acquisition system, it was decided to use speeds faster than
the ASTM C 393 recommendation. Three different speeds were selected for the following 4 rounds tests,
depending on core material. FBAL and FPVC tests performed at 0.1 [mm/s], FNL2 at 0.2 [mm/s] and
FNL1 at 0.4 [mm/s].
The results for all bending tests are presented in Figure 14 to Figure 17.
32
Figure 14 - Load vs displacement for all PVC beams
Figure 15 - Load vs displacement for all NL10 beams
33
Figure 16 - Load vs displacement for all NL20 beams
Figure 17 - Load vs displacement for all Balsa beams
34
3.5.2. Quasi-static indentation tests
After the bending tests, quasi-static tests were performed, using a hemispherical indenter (ø=16 [mm]),
attached to the same load cell, using the same test machine as in the bending tests. Test speed was set
to 0.2 [mm/s]. The specimens, with a nominal size of 150 [mm] * 100 [mm] were simply supported on a
“picture frame”, with the inner dimensions of 100 [mm] * 75 [mm] (see Figure 18).
Figure 18 - Simple support frame
First, 2 specimens of each material were tested until complete perforation. The results are presented
from Figure 19 to Figure 22.
Then, 3 specimens of each material were tested until the first failure was noted. The results are
presented from Figure 23 to Figure 26.
Pictures of the specimens, taken after the tests are available in Appendix D.1.
Figure 19 - Force vs Displacement for quasi-static tests on PVC specimens
35
Figure 20 - Force vs Displacement for quasi-static tests on NL1 specimens
Figure 21 - Force vs Displacement for quasi-static tests on NL2 specimens
Figure 22 - Force vs Displacement for quasi-static tests on BAL specimens
36
Figure 23 - Contact force for quasi-static tests on PVC specimens
Figure 24 - Contact force for quasi-static tests on NL10 specimens
Figure 25 - Contact force for quasi-static tests on NL20 specimens
37
Figure 26 - Contact force for quasi-static tests on Balsa specimens
3.5.3. Impact tests
Finally, the impact tests were performed, using a Rosand IFW5 falling-weight machine. Figure 27 shows
the impact mechanism, which consists in an impact head (the machine is equipped with two impact
heads with variable weight) that slides between two vertical rails. The head is hoisted by a steel cable,
connected via an electric release system.
Figure 27 - Rosand IFW5 falling-weight impact machine (Leigh Sutherland)
On the bottom extremity, the indenter is connected to the head through a load cell. The machine is also
equipped with a speed transducer, to accurately measure the speed at the instant prior to impact. Finally,
38
the machine has two pneumatic arms that catch the impact head right before the rebound, to avoid
multiple impacts.
The machine is connected to a computer, were the position of the weight and the data acquisition
process is controlled. The height from where the impact head is released can be directly inserted by the
user, or estimated based on the desired velocity/energy. However, due to friction, the actual speed is
lower than the desired one, hence the need to accurately measure the actual speed before impact.
The data recording system captures 1000 data points, during a previously chosen sample period. This
requires the user to have an estimate of the impact duration. The data points consist in a “force” value,
which is then converted in Newton, through a calibration factor. Speed, displacement and absorbed
energy can then be calculated using the following equations:
𝑣𝑖 = 𝑣𝑜 + 𝑑𝑡(9.81 − (𝐹𝑖+𝐹𝑜
2𝑚) (8)
𝑑𝑖 = 𝑑𝑜 + 𝑣𝑜𝑑𝑡 (9)
𝐴𝐸𝑖 = 𝐴𝐸𝑜 + (𝑑𝑖 − 𝑑𝑜)(𝐹𝑖+𝐹𝑜
2) (10)
where:
𝑣𝑜 = 𝑠𝑝𝑒𝑒𝑑 𝑖𝑛 𝑡ℎ𝑒 𝑖𝑛𝑠𝑡𝑎𝑛𝑡 𝑡 [𝑚
𝑠]
𝑑𝑡 = 𝑠𝑎𝑚𝑝𝑙𝑒 𝑝𝑒𝑟𝑖𝑜𝑑 [𝑠]
𝑣𝑖 = 𝑠𝑝𝑒𝑒𝑑 𝑖𝑛 𝑡ℎ𝑒 𝑖𝑛𝑠𝑡𝑎𝑛𝑡 𝑡 + 𝑑𝑡 [𝑚/𝑠]
𝐹𝑜 = 𝑓𝑜𝑟𝑐𝑒 𝑖𝑛 𝑡ℎ𝑒 𝑖𝑛𝑠𝑡𝑎𝑛𝑡 𝑡 [𝑁]
𝐹𝑖 = 𝑓𝑜𝑟𝑐𝑒 𝑖𝑛 𝑡ℎ𝑒 𝑖𝑛𝑠𝑡𝑎𝑛𝑡 𝑡 + 𝑑𝑡 [𝑁]
𝑚 = 𝑚𝑎𝑠𝑠 𝑜𝑓 𝑡ℎ𝑒 𝑖𝑚𝑝𝑎𝑐𝑡 ℎ𝑒𝑎𝑑 [𝐾𝑔]
𝑑𝑜 = 𝑑𝑖𝑠𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡 𝑖𝑛 𝑡ℎ𝑒 𝑖𝑛𝑠𝑡𝑎𝑛𝑡 𝑡 [𝑚]
𝑑𝑖 = 𝑑𝑖𝑠𝑝𝑙𝑎𝑐𝑒𝑚𝑒𝑛𝑡 𝑖𝑛 𝑡ℎ𝑒 𝑖𝑛𝑠𝑡𝑎𝑛𝑡 𝑡 + 𝑑𝑡 [𝑚]
𝐴𝐸𝑜 = 𝑎𝑏𝑠𝑜𝑟𝑏𝑒𝑑 𝑒𝑛𝑒𝑟𝑔𝑦 𝑖𝑛 𝑡ℎ𝑒 𝑖𝑛𝑠𝑡𝑎𝑛𝑡 𝑡 [𝐽]
𝐴𝐸𝑖 = 𝑎𝑏𝑠𝑜𝑟𝑏𝑒𝑑 𝑒𝑛𝑒𝑟𝑔𝑦 𝑖𝑛 𝑡ℎ𝑒 𝑖𝑛𝑠𝑡𝑎𝑛𝑡 𝑡 + 𝑑𝑡 [𝐽]
For the impact tests, the specimens were the same size as the ones tested under static indentation (150
[mm] x 100 [mm]), and using the same simply supported configuration.
From the absorbed energies in the quasi-static indentation tests, it was decided that the first set of tests
would be performed with an energy range between 50 [J] and 200 [J]. For this, a small impact head with
a mass of 10.853 [Kg] was used The results from the tests with nominal energies of 50, 100, 150 and
39
200 [J] are presented in Figure 28, Figure 29, Figure 30 and Figure 31 respectively. Pictures of the
specimens, taken after the tests are available in Appendix D.2.
Figure 28 - Load vs Displacement for impact tests on PVC specimens – 50J-200J energy range
Figure 29 - Load vs Displacement for impact tests on NL10 specimens – 50J-200J energy range
40
Figure 30 - Load vs Displacement for impact tests on NL20 specimens – 50J-200J energy range
Figure 31 - Load vs Displacement for impact tests on Balsa specimens – 50J-200J energy range
After performing 2 or more tests for each energy step, it was noted that none of the specimens had
suffer penetration of the second skin. With the remaining specimens, additional tests were performed
until the failure of the second skin was reached (Figure 32 and Figure 33). For these tests, a heavier
impact head was used, with 24.269 [Kg].
41
Figure 32 - Load vs Displacement for impact tests on all specimens - 300J range
Figure 33 - Load vs Displacement for impact tests on NL10 specimens - 450J range
43
4. Result analysis
4.1. Bending tests
As previously stated, one of the main reasons to perform bending tests on this work, it to assess how
the different cores behave in a structure where shear domains over bending.
Figure 34 shows the typical behaviour of the beams made of each one of the core materials. From
Figure 35 to Figure 38 are presented frames from the test recording, showing the instant right after
failure of the core, and the beams’ shape after the load is removed. In case of the specimen FNL11-2,
as for all the NL10 specimens, there was no core failure or skin sudden failure and the tests were
stopped at around 35-40 [mm] deflection. There was no point on keep testing the specimens beyond
this deflection, since there was severe core compression, and the skins were far from parallel.
Figure 34 - Overview of the typical behaviour for each core material
From Figure 34 it is noticeable the difference in the behaviour of beams made out of different materials.
PVC specimens present an almost linear relation between load and displacement, until it reaches a
small drop in load, believed to be the shear yield. The force keeps raising at a much lower rate, until the
first skin fails, through the combine action of compressive stress in the skin and the out of plane force
applied by the load pad.
NL20 specimens present a load-displacement curve similar to PVC specimens, with some differences.
The stiffness decreases almost from the beginning and there is no visible shear yield. The specimen
fails when a crack appears in the core.
The behaviour of NL10 specimens is rather unexpected, especially because the material properties
(Table 1) for Corecork NL10 and NL20 are very similar. These specimens present a very low stiffness
and can suffer high deflection without any significant damage. This poor performance may be related
with the incomplete cure of the skins, which leads to very flexible skins.
44
Balsa specimens withstand higher forces than the other specimens, but they fail prematurely (around 3-
4 [mm] deflection). The failure occurs at around half the expected core shear strength, and it consists in
a core crack, aligned with the core discontinuities, followed by large skin-core debonding.
Figure 35 - Test of the specimen FPVC1-2: Rupture of first skin (left) and specimen shape after removal of the load (right)
Figure 36 - Test of the specimen FNL11-2: Maximum deflection (left) and specimen shape after removal of the load (right)
Figure 37 - Test of the specimen FNL21-2: Failure of core due to shear (left) and specimen shape after removal of the load (right)
45
Figure 38 - Test of the specimen FBAL1-2: Failure of core due to shear and material discontinuity (left) and specimen shape after removal of the load (right)
To help analyzing the shear in the specimens, 2% offset curves were calculated, according to ASTM C
393. The curves are presented from Figure 39 to Figure 41. Balsa specimens failed during the elastic
regime and therefore, it is not necessary to do this procedure.
Since the plots don´t present the actual shear strain, a simplification was used, assuming pure shear.
For the test span of 280 [mm], 2% shear strain corresponds to a deflection of 2.8 [mm].
𝑆ℎ𝑒𝑎𝑟 𝑠𝑡𝑟𝑎𝑖𝑛 = tan(𝛼) =
𝐷𝑒𝑓𝑜𝑟𝑚𝑎𝑡𝑖𝑜𝑛
𝑂𝑟𝑖𝑔𝑖𝑛𝑎𝑙 𝑙𝑒𝑛𝑔𝑡ℎ
( 8 )
Yield and offset shear strength presented in Table 3. In PVC specimens, shear failure occurs before the
2% offset shear. Therefore, this is the value used for comparison.
Comparing the results with the manufacturer information, one can conclude that Balsa and NL10 have
a poor performance. NL10 shear strength is about 1/3 the strength specified by the manufacturer. Balsa
shear strength about half the expected strength.
In case of NL10, it is not known a certain reason that can explain such a difference. Again, the
manufacturer properties for NL10 and NL20 are very similar. Maybe the compressive strength of the
core, together with the low stiffness of the skins allow for the compression of the core under the loading
pad. Once the first and second skins are no longer parallel, their effectiveness in carrying axial loads is
significantly reduced and the load bearing capacity of the beam is lower than it would be expected.
The fact that the manufactures properties for shear are obtained using a different standard test, may
also influence the results. ASTM C273 tests sandwich or core specimens in a configuration similar to
pure shear.
That is the expected reason why Balsa specimens fail at half the expected shear strength. Balsa core
panels are built with small rectangular pieces of balsa, glued together. When testing a beam, there will
inevitably exist some discontinuities in the core, along the beam’s length. These discontinuities are a
week point, and will induce shear failure, as seen in Figure 38.
46
That will not happen when performing a test according to ASTM C273. This shows how important it is
to analyze the way materials will be solicited in a structure. The conditions of the laboratory tests may
not occur in “real life” applications, and may induce the designer to select the wrong material.
Figure 39 - Shear yield strength for PVC
Figure 40 - Shear yield strength for NL10
47
Figure 41 - Shear yield strength for NL20
Table 3 - Experimental vs manufacturer shear strength
Load-displacement slope - 0-2.5
[mm]
Maximum force - Offset
Maximum force - Yield
Shear strength -
Offset
Shear strength -
Yield
Shear strength-
Manufacturer
FPVC1-1 1278.60
7398.09 6987.84 1.57 1.48
1.60
FPVC1-2 1327.30
7118.77 6900.56 1.51 1.47
FPVC1-3 1308.30
7258.43 7118.77 1.52 1.50
FPVC1-4 1258.60
7136.23 6839.46 1.52 1.45
FPVC1-5 1250.10
7075.13 6717.26 1.52 1.44
FNL11-1 186.46
1235.74 - 0.26 -
0.90
FNL11-2 194.69
1149.21 - 0.25 -
FNL11-3 236.20
1471.41 - 0.30 -
FNL11-4 195.63
1069.53 - 0.23 -
FNL11-5 205.14
1218.28 - 0.26 -
FNL21-1 1035.20
5475.73 - 1.15 -
0.90
FNL21-2 978.71
5006.47 - 1.05 -
FNL21-3 1031.00
5215.29 - 1.08 -
FNL21-4 979.62
4822.53 - 1.00 -
FNL21-5 1013.10
5102.48 - 1.07 -
FBAL1-1 2488.40
- 6717.26 - 1.32
3.00
FBAL1-2 2960.90
- 8314.58 - 1.62
FBAL1-3 2583.20
- 7485.37 - 1.49
FBAL1-4 2811.60
- 7930.53 - 1.52
FBAL1-5 2436.50
- 7555.20 - 1.54
48
4.2. Quasi-static indentation
To analyze the quasi-static test results, the absorbed energy during the specimen indentation was
calculated. A sequence of frames from the video recording of the tests is also presented, showing the
main instants during one of the tests
Figure 42 - Absorbed energy vs Displacement for quasi-static tests on PVC specimens
Figure 43 - Quasi-static indentation frame sequence (SPVC1-3): a)-Elastic domain; b)-Perforation of first skin; c)-Perforation of second skin; d)-End of test
a)
b) d)
c)
49
Table 4 presents the absorbed energies at the failure of first and second skin, as well as the absorbed
energy at 57 [mm] displacement.
Both PVC specimens present a similar behaviour. There are two force peaks, one on each skin. Between
these two peaks, the force due to friction and core crushing is relatively low.
NL10 presents high deflection of the first skin, prior to skin failure. Again, the incomplete cure of the
skins is the likely explanation for the lack of stiffness of the skins, and consequently, lack of stiffness of
the specimens. The skin then goes back to a shape close to the initial one. In one of the tests (SNL11-
1), the first drop in the load corresponds to the appearance of a crack on the skin, similarly to what
happens in some of the impact test with this core material.
Both NL20 and Balsa present high friction force between the two skins, and the second skin tend to
separate from the core. NL20 specimens present higher force in the second skin penetration, thus
making this specimens the ones which absorb more energy.
One phenomena that was common to all specimens, although with different severity, was the fact that
the second skin’s edges deflect upwards, as consequence of the specimen being longer and wider than
the frame opening. This upward deflection is more intense in the cork specimens and less intense in the
PVC specimens. Its amplitude depends on the core compressive modulus and on the deflection of the
second skin in the middle of the specimen, which depends on the adhesion of the second skin to the
core and also on the deflection of the core.
PVC and NL10 specimens present a good adhesion between skins and core. NL20 specimens present
a lower adhesion compared to NL10. However this fact may be biased by the fact that NL10 panels did
not reach the same level of cure as NL20. The fact that NL10 skins are less rigid and brittle than NL20
and that NL10 panels seem to have absorbed more resin may result in a lower adhesion of skin and
core in the NL20 specimens.
Figure 44 - Absorbed energy vs Displacement for quasi-static tests on NL10 specimens
50
Figure 45 - Quasi-static indentation frame sequence (SNL11-2): a)-Elastic domain; b)-Elastic domain with significant first skin deformation; c)-Perforation of first skin; d)-Perforation of the core, with first skin
approaching the original shape; e)-Perforation of second skin; f)-End of test
Regarding Balsa specimens, they present the lower adhesion between skin and core of all specimens.
The second skin is easily ripped off from the core material
a)
b)
c) f)
e)
d)
51
Figure 46 - Absorbed energy vs Displacement for quasi-static tests on NL20 specimens
Figure 47 - Quasi-static indentation frame sequence (SNL21-2): a)-Elastic domain; b)-Perforation of first skin; c)-Perforation of second skin; d)-End of test
a) c)
d) b)
52
Figure 48 - Absorbed energy vs Displacement for quasi-static tests on Balsa specimens
Figure 49 - Quasi-static indentation frame sequence (SBAL1-3): a)-Elastic domain; b)-Perforation of first skin; c)- Perforation of second skin; d)-End of test (second skin almost completely separated)
a) c)
d) b)
53
Table 4 - Absorbed energy at different instants during indentation test
1st peak 2nd peak 57 [mm]
Disp. Force Energy Disp. Force Energy Force Energy
[mm] [N] [J] [mm] [N] [J] [N] [J]
SPVC1-2 7.92 8630.60 36.91 43.40 7958.41 148.97 1612.30 195.20
SPVC1-3 7.10 7575.12 29.04 41.41 8361.73 157.74 2084.61 212.91
SNL11-1 11.71
(24.69)
4694.73
(6274.32)
29.67
(92.09)
41.99 5164.92 140.54 957.52 171.66
SNL11-2 16.92 5226.00 50.06 41.93 5069.50 103.97 1210.31 138.76
SNL21-1 8.16 8100.66 38.38 44.12 9137.85 217.93 6073.50 302.53
SNL21-2 7.98 7206.86 34.98 44.98 9249.48 207.58 2398.95 261.18
SBAL1-1 4.50 8604.41 19.58 43.22 5706.14 145.87 1429.36 187.16
SBAL1-2 4.41 8772.08 18.90 46.21 6614.79 179.75 1515.45 217.90
54
4.3. Impact tests
4.3.1. Evolution of failure with increasing energy
This chapter is intended to present a qualitative analysis of each one of the 4 types of sandwich and
how they behave with increasing impact energy.
After performing the impact tests, the raw data was processed in an Excel file, with the equations in
Chapter 3.5.3. In Appendix C are presented the plots of absorbed energy vs displacement. Table 10 and
Table 11 summarize the main data of the tests, including incident energy and velocity.
PVC specimens, as expected and seen in the previous tests, present a quite reproducible and
predictable behaviour. Figure 50 presents a sequence of plots of load vs displacement for each one of
the nominal incident energies, to demonstrate the behaviour of this material configuration under impact
scenarios. One can state that the stiffness of the specimen and the maximum force during the impact of
first skin do not vary with rising incident energy, and therefore, do not vary with incident speed as well.
Rupture of the first skin occurs between 8 and 9 [mm] displacement. As in the previous tests, this value
corresponds to the sum of the deflection of the specimen plus the permanent indentation of the
specimen’s skin.
Figure 50 - Load vs Displacement for impact tests on PVC specimens - 50J-300J energy range
After the rupture the first skin, the force drops and is maintained relatively low until the indenter
approached the second skin. The force rises again, up to a maximum force slightly lower that the force
of rupture of the first skin, although with lower stiffness, due to the fact that in the second skin, there is
a densification of the core material prior to the rupture of the second skin. As the energy increases,
second skin delaminations start to appear. With an energy of 300 [J], the second skin is pierced and
some transverse fibre bundles are ripped off during this.
NL10 specimens presented a far less predictable behaviour than PVC specimens, being the sandwich
configuration more tested in this work. Figure 51 presents one or two load plots for each energy,
depending on the failure mode observed in each energy level.
55
Figure 51 - Load vs Displacement for impact tests on NL10 specimens - 50J-450J energy range
Figure 52 - Initial stiffness vs 1st skin thickness for DNL10 specimens
Figure 53 - Initial stiffness vs Initial velocity for DNL10 specimens
56
Unlike the specimens made from the other materials, NL10 specimens present a variation on the initial
stiffness, prior to the failure of the skin, as well as a wide range of maximum load at which the first failure
of the skin occurs (approximately between 7000 and 10000 [N]). Figure 52 and Figure 53 present the
initial stiffness (up to the first skin failure) against 1st skin thickness and initial velocity of the indenter.
With the amount of data acquired and the small variation of both skin thickness and initial velocity, it is
impossible to establish any relation between these two properties and the initial stiffness of the
specimens.
A small number of NL10 specimens behave similarly to PVC and NL20 specimens. The first skin is
perforated without the appearing of skin cracks, followed by a drop in the load, as the indenter pierced
through the core material.
However, the majority of the specimens, the first failure of the first skin is characterized by a transverse
crack (see Figure 105 to Figure 119), with a consequent drop in the load. After, the first failure, the
specimen keeps an “elastic” behaviour. For example, this composite can withstand a 200 [J] impact
without completely penetrating the first skin (see test DNL12-9). When comparing with the other
composites, at this energy level they are already near the failure of the second skin, noticeable by the
increase in the load when the indenter approached the second skin. At 300 [J], the first skin in one of
the specimens is pierced, followed by a sudden drop in the load (DNL12-13). Other specimen does not
reach the complete perforation of the first skin, but developed severe shear cracks near the first skin
(see Figure 116).
Only with an incident energy of 450 [J] the specimen is completely perforated. All specimens showed a
good adhesion between core and skins, with the appearing of delaminations in the middle of the skin
instead of skin/core delamination.
Figure 54 - Load vs Displacement for impact tests on NL20 specimens - 50J-300J energy range
NL20 specimens’ initial behaviour is very similar to PVC specimens’. The indenter’s load increase
linearly until the rupture of the first skin, that occurs at around 10 [mm] displacement and between 12000
57
and 13000 [N]. The failure of the first skin consists in perforation of the skin, with some delaminations
around the hole (see Figure 122 to Figure 124).
The load then drops down to around 3000 [N], being the NL20 specimens the ones with higher energy
absorption rate in the penetration of the core material stage.
As the indenter approached the second skin, compression the cork and deformation of second skin
make the load increases. During the development of this work, it was stated that cork has the ability to
almost completely return to its original shape after compression, similarly to a piece of rubber or a spring.
That does not happen with PVC and Balsa. Although these two materials are stiffer that cork, their cells
will eventually collapse before cork. Balsa collapses at 0,3% strain and PVC at 1.5%, while NL10 fails
at around 5.8% and NL20 at 8.3%, according to Table 1.
The fact that NL20 withstands larger strains before failing may explain the second skin failure mode
present in the 300 [J] tests (specimens DNL22-9 and DNL22-10). In these specimens, the second skin
completely separated from the specimen. Aligned with the indenter, there is a piece of core material
attached to the skin (see Figure 128 and Figure 129). That piece of material transferred the load from
the indenter to the skin, without collapsing. Bellow that piece of core material, the skin was delaminated
and developed cracks.
As the main failure of the second skin corresponds to the delamination of the skin from the core material,
and the specimens have relatively small dimensions, it is not possible to state about the extent of the
delamination in a real life application. However, one can state that if the second skin was not perforated
in a specimen this small it will most certainly not be perforated in a larger panel, where bending
behaviour dominates.
Figure 55 - Load vs Displacement for impact tests on Balsa specimens - 50J-300J energy range
Similarly to the beam specimens, Balsa specimens present higher stiffness than the other composites
for impact tests. The maximum force observed in the failure of the first skin is similar to the one verified
in PVC and NL20 specimens, and slightly below the maximum force measured in the final failure of
NL10 specimens’ first skin.
58
After the penetration of the first skin, the load reduces linearly, identically as in the PVC specimens.
Then, a plateau between 2000 and 3000 [N] is maintained along almost 25 [mm] of penetration,
corresponding to almost the complete thickness of the core. At around 40 [mm] displacement, the load
starts to increase, for incident energies above 200 [J].
Between the 200 [J] and 300 [J] there is a well-defined increase in the stiffness, when reaching the
second skin. Figure 141 and Figure 142 (Appendix D.2) show the massive pyramidal cracks that were
formed in the core material, in the 300 [J] tests. Figure 56 presents the core for the specimens DBAL1-
6, DBAL1-10 and DBAL1-13, after the second skin was ripped off (after the test). One can see that these
specimens also developed similar core cracks, but with less intensity. Still, the difference in the response
of the 200 [J] and 300 [J] Balsa specimens second skin occurs at around 40 [mm], suggesting that this
difference is not related with the core itself, but with the way the second skin fails. In the tests with 200
[J], the skin did not completely separated from the core, as it happened in the 300 [J] tests.
Figure 56 - Condition of the core in the specimens DBAL1-6, DBAL1-10 and DBAL1-13
59
4.3.2. Quasi-static testing vs Impact testing
Once both static and dynamic tests were performed, it would be beneficial to find relations, even that
only in a qualitative basis, between quasi-static and dynamic test results. Quasi-static testing requires
less sophisticated apparatus and is easier to do.
Among the most important information that can be obtained from the quasi-static and impact tests is the
energy that the specimen can withstand up to the occurrence of a certain failure. The natural choice is
to measure the energy absorbed up to the failure of each skin, as they represent important stages in the
integrity of the specimens and panels.
These events consist in a variety of failure modes, as delaminations, matrix cracks and fibre rupture,
and therefore, they occur during a certain amount of time. The absorbed energy at the displacement
where the load reaches a maximum was taken as the absorbed energy up to the failure of first or second
skin, respectively. In the quasi-static test results, those energies are quite simple to obtain.
On the impact tests, it was considered that the first skin damage starts at 50 [J] of incident energy,
although the plots seem to show an elastic behaviour for these level of incident energy. For the second
skin, the energies for the energy levels after and before the failure were considered. They correspond
to 200 [J] and 300 [J] for PVC, NL10 and Balsa, and to 300 [J] and 450 [J] for NL10.
Figure 57 and Figure 58 present the average absorbed energies for quasi-static and impact tests,
respectively. Is also shown the deviation of the experimental data.
Figure 57 - Absorbed energy to reach failure of specimen skins - Quasi-static tests
It is noticeable the increase in absorbed energy from the quasi-static tests for the impact tests, especially
for NL10 specimens. As seen before, NL10 behaviour seems to be governed by dynamic phenomena.
The fact that the absorbed energy until failure of the second skin increased more than 3 times from
quasi-static to dynamic tests, and also the variation of the results around the average value strengthens
that hypothesis.
60
Figure 58 - Absorbed energy to reach failure of specimen skins - Impact tests
To help understanding these results, quasi-static data was plotted together with impact data, from Figure
59 to Figure 62. After a brief analysis, one can see that quasi-static tests present a maximum load for
the first skin of roughly 2/3 of the maximum impact load.
NL20 seems to present the best results, when comparing quasi-static to impact data (see Figure 61).
Apart from the fact that the maximum load for the first skin is around 2/3 of the one for impact tests,
loading and unloading stiffness are very similar in both cases. The main difference is the fact that in the
impact tests, the failure of the skin occurs at a well-defined point, while in the static tests it takes some
millimeters to be completed.
Figure 59 - Load vs Displacement for PVC specimens – Quasi-static and impact
As stated in the previous section, NL20 specimens present high friction on the core material. That friction
matches quite well the friction from impact tests.
61
Figure 60 - Load vs Displacement for NL10 specimens – Quasi-static and impact
Opposing to NL20 results, NL10 present a large deviation between quasi-static and impact results (see
Figure 60). Quasi-static maximum loads are around half the impact maximum loads and the
displacements at which the failures of first and second skin occur are also lower, probably due to the
huge deformations the NL10 specimens suffer during the impact tests. There is also a significant
difference between the initial stiffness of the quasi-static and impact results. Additional tests with incident
velocities between 0 and 3 [m/s] would help to find a trend line in the highly dynamic behaviour of NL10
specimens. The higher deformations and lower stiffness of the NL10 are related with the bad cure of the
skins, among other phenomena. This allows the specimen to absorb more energy, similarly to what
happens with composites where plasticizers were added to the resin, or where elastomeric coatings are
used. The downside of lower stiffness of the skins is the lower stiffness of the specimen or panel, which
may be a limitation for the application of these panels in structures with high static solicitations.
Figure 61 - Load vs Displacement for NL20 specimens – Quasi-static and impact
62
Figure 59 and Figure 62 present the comparison of results for PVC and Balsa. Much of the comparison
made for NL20 results is applicable for these materials too. The main difference is that the friction force
when penetrating the core are not so well defined in PVC and Balsa
Figure 62 - Load vs Displacement for Balsa specimens – Quasi-static and impact
After the qualitative comparison between the main characteristics of the static indentation and impact
tests, the main data relative to the first skin failure was related against the initial velocity.
It was previously mentioned in this section that there was a significant increase in the absorbed energy
from quasi-static to impact tests, without however doing any reference to the initial velocity. Absorbed
energy up to first skin failure increases with incident velocity, for all materials, although more significantly
in Balsa specimens.
The increase in the absorbed energy is mainly related with the increase in the maximum force, rather
than with the increase in displacement at which first skin failure occurs. The increase in the displacement
is not much significant. Actually, the quasi-static tests show that for NL10, there is a significant decrease
in the displacement at which the first skin fails.
The increase of the maximum force appears to be linear between 0 and 4 [m/s], stabilizing after that
speed, indicating that the influence of the speed is much higher between quasi-static tests and impact
tests than between the impact tests themselves. These relations should be further investigated, through
the test of more specimens, and including incident energies between 0 and 3 [m/s] as well.
63
Figure 63 - Absorbed energy at failure of first skin vs Initial velocity
Figure 64 - Maximum force at failure of first skin vs Initial velocity
Figure 65 - Displacement at failure of first skin vs Initial velocity
65
5. Conclusions
The aims of the thesis were reached: present a comprehensive literature review on impact on marine
composites and develop and complete a test program to compare marine laminates in terms of impact
resistance and behaviour.
A brief review of the state of the art was presented, concerning general knowledge about impact,
followed by a revision of a significant number of topics regarding specificities of marine impact,
composed by research focused in marine applications and by work developed for other applications.
It was decided to develop the test program with sandwich laminates similar to the ones used in MOSAIC.
Among the several advantages, is the fact that it is a project with commercial applications. In addition
to the use of PVC and Balsa, cork was also considered as a suitable core material. It is a material with
an ever increasing range of applications, namely shock absorption.
Four different sandwich laminates were then produced, consisting in 30 [mm] of core materials (PVC,
Balsa, Corecork NL10 and Corecork NL20) with 3 [mm] E-glass/polyester skins. To obtain better final
properties of the laminates, vacuum bagging was choose as manufacturing method. During the cutting
of the specimens, some difficulties arisen, due to the complete different characteristics of the core and
skin materials.
The initial set of bending tests show that the different laminates present a wide range of properties and
failure modes. Balsa beams present the higher stiffness but fail at a very low deflection, while NL10
beams withstand large deflections without almost any permanent deformation. The results have also
proven that the mechanical properties provided by the manufacturers may be obtained with tests that
do not replicate the actual conditions the material will be subjected to. This fact may lead the designer
to the choice of the wrong material for a given application.
From the static indentation tests, the main considerations are that all four types of specimens present
the high forces when the skins are penetrated. The friction force in the core is low relatively to the skin
penetration force. NL20 specimens present the highest core friction against the indenter. NL10
specimens’ skin withstand larger deformations until the first failure.
For the impact test, the criteria to consider a specimen’s failure was the penetration or complete
separation of the second skin, whichever happens in first place. However, the separation of the second
skin from the core may not be critical in a large panel, if the extent of that separation is limited to a
certain area. The first conclusion from these tests is that NL10 specimens fail between 300 and 450 [J],
while the other specimens fail before, between 200 and 300 [J].
PVC and NL20 specimens present a good repeatability of the load-displacement behaviour and of the
failure modes, while NL10 specimens present two major failure modes and a far less predictable
behaviour. Balsa specimens show a reasonable repeatability of results up to the start of failure of the
second skin. After that, the fact that the skin-core bond is not very strong may explain the variation of
the final behaviour.
66
In the final stage of the work, impact and quasi-static results were compared. PVC, NL20 and Balsa
present similar results in both tests, but the first and second skin peak forces are around 1.5 times higher
in the impact tests. The absorbed energy at the failure of first and second skin is then slightly higher in
the impact tests. On the other hand, NL10 specimens present a large difference between the two tests.
While in the quasi-static tests, NL10 are the specimens which absorb less energy up to the failure of
second skin, in the impact tests they absorb substantially more than the other specimens, and 3 times
more than the NL10 in quasi-static tests. This significant difference in NL10 results from static to impact
test shows the trade-offs of the high plasticization in NL10 specimens: the lack of stiffness due to the
bad cure of the resin, which is undesirable, allow the specimen to deform more without failing, which
leads to the ability to absorb more energy.
Finally, one can conclude that cork laminates have high potential for applications with impact
requirements.
5.1. Future work
Based on the conclusions and challenges faced during the development of this work, further work can
be done to improve the comparison between the different laminates:
Investigate the interaction between the cork panels (especially Corecork NL10) and the
polyester resin. Check the compatibility of cork with other resins vinylester and epoxy resins.
Investigate the behaviour of the specimens in impact tests with incident velocities between 0
and 3 [m/s].
Repeat a similar experiment with panels with and intermediate cork core density.
Study the effect of speed and energy separately
Measure the actual deflection of the specimens with help of ultrasonic/laser transducers,
especially in the static indentation and impact tests.
Perform tests in larger panels, to see the extent of the second skin delaminations in a “real”
case, and measure the residual strength.
Compare the panels in terms of thermal, acoustic and fire-resistance properties, in order to
validate its use as internal bulkheads and other applications.
67
6. References
[1] S. Abrate, Impact on composite structures, First. Cambridge University Press, 1998.
[2] S. Abrate, Impact Engineering of Composite Structures. Springer Wien New York, 2011.
[3] S. Abrate, “Modeling of impacts on composite structures,” 2001.
[4] M. Hildebrand, “A comparison of FRP-sandwich penetrating impact test methods.” VTT, 1996.
[5] D. Choqueuse, R. Baizeau, and P. Davies, “Experimental studies of Impact on Marine Composites.” IFREMER, 1999.
[6] Y. Toyama, “Drift-wood collision load on bow structure of high-speed vessels,” Mar. Struct., vol. 22, no. 1, pp. 24–41, Jan. 2009.
[7] Y. Ping, L. Weihua, and B. Hayman, “Recent Research of High Speed Vessels in Structural Response to Accidental loads,” J. Wuhan Univ. Technol., vol. 26, 2002.
[8] B. Hayman and D. Mcgeorge, “Dynamic response of sandwich structures.” DNV, 2005.
[9] M. Hildebrand, “Penetrating impact strength of FRP-sandwich panels - Empirical and semi-empirical prediction methods.” VTT, 1996.
[10] H. E. Johnson, L. a. Louca, S. Mouring, and a. S. Fallah, “Modelling impact damage in marine composite panels,” Int. J. Impact Eng., vol. 36, no. 1, pp. 25–39, Jan. 2009.
[11] D. Zenkert, A. Shipsha, and K. Persson, “Static indentation and unloading response of sandwich beams,” Compos. Part B Eng., vol. 35, no. 6–8, pp. 511–522, Sep. 2004.
[12] L. S. Sutherland and C. Guedes Soares, “The use of quasi-static testing to obtain the low-velocity impact damage resistance of marine GRP laminates,” Compos. Part B Eng., vol. 43, no. 3, pp. 1459–1467, Apr. 2012.
[13] L. S. Sutherland and C. Guedes Soares, “Contact indentation of marine composites,” Compos. Struct., vol. 70, no. 3, pp. 287–294, Sep. 2005.
[14] L. S. Sutherland and C. Guedes Soares, “Impact on marine composite laminated materials.” CENTEC, 2011.
[15] P. Davies, “Scale and size effects in the mechanical characterization of composite and sandwich materials,” ICCM12 Pap. 1033, 1999.
[16] D. Zenkert, a Shipsha, P. Bull, and B. Hayman, “Damage tolerance assessment of composite sandwich panels with localised damage,” Compos. Sci. Technol., vol. 65, no. 15–16, pp. 2597–2611, Dec. 2005.
[17] B. Hayman, “Damage assessment and damage tolerance of FRP sandwich structures,” Sandw. Struct. 7 Adv. with Sandw. Struct. Mater., pp. 27–43, 2005.
[18] F. Collombet, L. Guillaumat, J. L. Lataillade, P. Davies, and A. Torres-Marques, “Study of impacted composite structures by means of the response surface methodology,” ICCM12 Pap. 227, 1999.
68
[19] L. Guillaumat, “Reliability of composite structures - impact loading,” Comput. Struct., vol. 76, 2000.
[20] L. S. Sutherland and C. Guedes Soares, “Impact behaviour of typical marine composite laminates,” Compos. Part B Eng., vol. 37, no. 2–3, pp. 89–100, Apr. 2006.
[21] O. Aamlid, “Oblique impact testing of aluminum and composite panels.” DNV, 1995.
[22] O. Aamlid and G. A. Antonsen, “Oblique impact testing of single skin, aramid fibre reinforced plastic panels.” DNV, 1997.
[23] M. Hildebrand, “Local impact strength of various boat-building materials.” VTT, 1997.
[24] M. Wiese, A. Echtermeyer, and B. Hayman, “Evaluation of oblique impact damage on sandwich panels with PVC and balsa core materials,” Fourth Int. Conf. Sandw. Constr., vol. 2, pp. 807–818, 1998.
[25] C. Atas and C. Sevim, “On the impact response of sandwich composites with cores of balsa wood and PVC foam,” Compos. Struct., vol. 93, no. 1, pp. 40–48, Dec. 2010.
[26] J. Gustin, A. Joneson, M. Mahinfalah, and J. Stone, “Low velocity impact of combination Kevlar/carbon fiber sandwich composites,” Compos. Struct., vol. 69, no. 4, pp. 396–406, Aug. 2005.
[27] A. M. Dumont and S. J. I. R. Blake, “Application of novel cork sandwich core for high performance sailing craft.” VTT, 2012.
[28] A. Echtermeyer, R. F. Pinzelli, K. B. Raybould, and N. Skomedal, “Advanced composite hull structures for high speed craft.” DNV, 1994.
[29] M. Hildebrand, “Improving the impact strength of FRP-sandwich panels for ship applications, Part 2.” VTT, 1997.
[30] M. Hildebrand, “The effect of the strain rate on the strength of FRF-sandwich face and core materials.” VTT, 1997.
[31] N. Baral, D. D. R. Cartié, I. K. Partridge, C. Baley, and P. Davies, “Improved impact performance of marine sandwich panels using through-thickness reinforcement: Experimental results,” Compos. Part B Eng., vol. 41, no. 2, pp. 117–123, Mar. 2010.
[32] N. Mitra, “A methodology for improving shear performance of marine grade sandwich composites: Sandwich composite panel with shear key,” Compos. Struct., vol. 92, no. 5, pp. 1065–1072, Apr. 2010.
[33] K. Imielińska, L. Guillaumat, R. Wojtyra, and M. Castaings, “Effects of manufacturing and face/core bonding on impact damage in glass/polyester–PVC foam core sandwich panels,” Compos. Part B Eng., vol. 39, no. 6, pp. 1034–1041, Sep. 2008.
[34] Y. Perrot, C. Baley, Y. Grohens, and P. Davies, “Damage Resistance of Composites Based on Glass Fibre Reinforced Low Styrene Emission Resins for Marine Applications,” Appl. Compos. Mater., vol. 14, no. 1, pp. 67–87, Jan. 2007.
[35] B. Hayman, A. Echtermeyer, and D. Mcgeorge, “Use of fibre composites in naval ships.” DNV, 2001.
[36] O. Aamlid, “Impact properties of different shell structures in relation to rule requirements,” Compos. Sandw. Struct., pp. 87–101, 1997.
69
[37] P. T. Pedersen and S. Zhang, “Minimum plate thickness in high-speed craft,” Pract. Des. ships Mob. units, pp. 959–965, 1998.
[38] G. A. O. Davies, D. Hitchings, and G. Zhou, “Impact damage and residual strengths of woven fabric glass/polyester laminates,” Compos. Part A Appl. Sci. Manuf., pp. 1147–1156, 1996.
[39] P. Auerkari, “Effect of impact face damage on strength of sandwich composites.” VTT, 1993.
[40] A. Shipsha and D. Zenkert, “Compression-after-Impact Strength of Sandwich Panels with Core Crushing Damage,” Appl. Compos. Mater., vol. 12, no. 3–4, pp. 149–164, May 2005.
[41] P. Auerkari and P. H. Pankakoski, “Strength of sandwich panels with impact defects.” VTT, 1995.
[42] A. Shipsha, S. Hallström, and D. Zenkert, “Failure Mechanisms and Modelling of Impact Damage in Sandwich Beams - A 2D Approach: Part I - Experimental Investigation,” J. Sandw. Struct. Mater., vol. 5, pp. 7–31, 2003.
[43] A. Shipsha, S. Hallström, and D. Zenkert, “Failure Mechanisms and Modelling of Impact Damage in Sandwich Beams - A 2D Approach: Part II - Analysis and Modelling,” J. Sandw. Struct. Mater., vol. 5, pp. 33–51, 2003.
[44] A. Shipsha and D. Zenkert, “Fatigue Behavior of Foam Core Sandwich Beams with Sub-Interface Impact Damage,” J. Sandw. Struct. Mater., vol. 5, pp. 147–160, 2003.
[45] H. Gu and S. Hongxia, “Delamination behaviour of glass/polyester composites after water absorption,” Mater. Des., vol. 29, no. 1, pp. 262–264, Jan. 2008.
[46] F. B. Boukhoulda, L. Guillaumat, J. L. Lataillade, E. Adda-Bedia, and a. Lousdad, “Aging-impact coupling based analysis upon glass/polyester composite material in hygrothermal environment,” Mater. Des., vol. 32, no. 7, pp. 4080–4087, Aug. 2011.
[47] K. Imielińska and L. Guillaumat, “The effect of water immersion ageing on low-velocity impact behaviour of woven aramid–glass fibre/epoxy composites,” Compos. Sci. Technol., vol. 64, no. 13–14, pp. 2271–2278, Oct. 2004.
[48] Y. J. Weitsman, X. Li, and A. Ionita, “Sea water effects on polymeric foams and their sandwich layups,” pp. 193–197, 2005.
[49] MOSAIC, “Materials On-board: Steel And Integrated Composites,” 2012. [Online]. Available: http://www.mosaicships.com/.
[50] “Standard Test Method for Core Shear Properties of Sandwich Constructions by Beam Flexure.” ASTM, 2011.
[51] “Standard test method for measuring the damage resistance of a fiber-reinforced polymer-matrix composite to a concentrated quasi-static indentation force,” vol. 15. ASTM, 1998.
[52] “Standard test method for measuring the damage resistance of a fiber-reinforced polymer matrix composite to a drop-weight impact event.” ASTM, 2005.
[53] Gurit, “Guide to Composites.” SP Gurit.
70
Appendices
A. List of marine impact references by place and research
area
Table 5 - Marine impact references, by place and research area
Area Research Centre Year Title
Impact Modeling
CENTEC, Portugal
2005 Contact indentation of marine composites
Skin 2006 Impact behaviour of typical marine composite laminates
Skin 2011 Impact on marine composite laminated materials
Impact Modeling 2012
The use of quasi-static testing to obtain the low velocity impact damage resistance of marine GRP laminates
Water absorption
Djillali Liabès University of Sidi Bel-Abbès, Algeria and Esplanade des Arts et Métiers, France 2011
Aging-impact coupling based analysis upon glass/polyester composite material in hygrothermal environment
Skin and core
DNV, Norway
1994 Advanced composite hull structures for high speed craft
Skin 1995 Oblique impact testing of aluminium & composite panels
Skin 1997 Oblique impact testing of single skin, aramid fibre reinforced plastic panels
Society Rules 1997
Impact properties of different shell structures in relation to rule requirements
Core 1998 Evaluation of oblique impact damage on sandwich panels with PVC and balsa core materials
Skin and core 2001 Use of Fibre Composites in Naval Ships
Impact Modeling 2002
Recent Research of High Speed Vessels in Structural Response to Accidental loads
Impact Modeling 2005 Dynamic Response of Sandwich Structures
Society Rules 2005
Damage assessment and damage tolerance of FRP sandwich structures
Core
Dokuz Eylul University and AERO Wind Industry Inc , Turkey 2010
On the impact response of sandwich composites with cores of balsa wood and PVC foam
Impact Modeling
Esplanade des Arts et Métiers, France 2000 Reliability of composite structures - impact loading
Water absorption Gdansk University of
Technology, Poland, ENSAM, France
2004 The effect of water immersion ageing on low-velocity impact behaviour of woven aramidglass fibre/epoxy composites
Skin and core 2008
Effects of manufacturing and face/core bonding on impact damage in glass/polyester-PVC foam core sandwich panels
Residual strength
IFREMER, France
1996 Impact damage and residual strengths of fabric glass / polyester laminates
Impact Modeling 1999 Experimental studies of Impact on Marine Composites
Impact Modeling 1999
Scale and size effects in the mechanical characterization of composite and sandwich materials
Impact Modeling 1999
Study of impacted composite structures by means of the response surface methodology
Skin and core 2007
Damage Resistance of Composites Based on Glass Fibre Reinforced Low Styrene Emission Resins for Marine Applications
Core 2010 Improved impact performance of marine sandwich panels using through-thickness reinforcement: Experimental results
Impact Modeling
Imperial College London, UK and US Naval Academy, USA 2009 Modelling impact damage in marine composite panels
71
Core
Indian Institute of Technology Kharagpur, India 2010
A methodology for improving shear performance of marine grade sandwich composites: Sandwich composite panel with shear key
Core
IUT Brest, GeM Institut de Recherche en Génie Civil et Mécanique and IFREMER, France 2005 Loading rate effects on foam cores for marine sandwich structures
Water absorption
Louisiana State University and University of Delaware, USA 2004
Determination of Moisture Effects on Impact Properties of Composite Materials
Skin North Dakota State University, USA 2005
Low velocity impact of combination Kevlar/carbon fiber sandwich composites
Residual strength
Royal Institute of Technology, Sweden
2003 Failure Mechanisms and Modelling of Impact Damage in Sandwich Beams A 2D Approach: Part I Experimental Investigation
Residual strength 2003
Failure Mechanisms and Modelling of Impact Damage in Sandwich Beams - A 2D Approach: Part II - Analysis and Modelling
Residual strength 2003
Fatigue Behavior of Foam Core Sandwich Beams with Sub-Interface Impact Damage
Impact Modeling 2004 Static indentation and unloading response of sandwich beams
Impact Modeling 2005
Damage tolerance assessment of composite sandwich panels with localised damage
Residual strength 2005
Compression-after-Impact Strength of Sandwich Panels with Core Crushing Damage
Impact Modeling
School of Marine Science and Technology, Japan 2009 Drift-wood collision load on bow structure of high-speed vessels
Society Rules
Technical University of Denmark, Denmark 1998 Minimum plate thickness in high-speed craft
Water absorption
Tianjin Polytechnic University, China 2008
Delamination behaviour of glass/polyester composites after water absorption
Water absorption
University B.D.Tand Kuvempu University, India 2010
Damage characterisation of glass/textile fabric polymer hybrid composites in sea water environment
Water absorption University of London, UK 2008
The influence of long term water immersion ageing on impact damage behaviour and residual compression strength of glass fibre reinforced polymer (GFRP)
Core University of Southampton, UK 2012
Application of novel cork sandwich core for high performance sailing craft
Water absorption
University of Tenessee, USA 2005 Sea water effects on polymeric foams and their sandwich layups
Residual strength
VTT, Finland
1993 Effect of impact face damage on strength of sandwich composites
Residual strength 1995 Strength of sandwich panels with impact defects
Impact Modeling 1996
Penetrating Impact Strength of FRP-Sandwich Panels - Empirical and Semi-Empirical Prediction Methods
Impact Modeling 1996 A comparison of FRP-sandwich penetrating impact test methods
Skin and core 1997 Local impact strength of various boat-building materials
Skin and core 1997
Improving the Impact Strength of FRP-Sandwich Panels for Ship Applications, Part 2
Skin and core 1997
The effect of the strain rate on the strength of FRP-sandwich face and core materials
72
B. Specimen dimensions
Table 6 - Relevant dimensions and estimated shear strength for specimens used in bending tests
Specimen Breadth Total
thickness
First skin
thickness
Second skin
thickness
Estimated
shear
strength
[mm] [mm] [mm] [mm] [N]
FNL11-1 72.80 35.61 2.65 2.57 4324.52
FNL11-2 70.65 35.65 2.43 2.66 4210.17
FNL11-3 73.62 35.29 2.53 2.47 4344.73
FNL11-4 71.01 35.55 2.66 2.43 4218.62
FNL11-5 71.79 35.64 2.63 2.40 4280.05
FNL21-1 71.37 36.64 3.68 2.84 4288.40
FNL21-2 71.40 36.56 3.63 2.77 4286.78
FNL21-3 72.48 36.61 3.63 2.68 4364.47
FNL21-4 72.17 36.57 3.49 2.79 4342.99
FNL21-5 71.88 36.54 3.77 2.67 4311.51
FNL21-6 70.41 36.54 3.67 2.78 4222.92
FNL21-7 72.13 36.58 3.56 2.65 4346.41
FPVC1-1 71.78 36.06 3.23 3.12 7553.90
FPVC1-2 71.88 35.93 3.29 3.08 7531.52
FPVC1-3 72.62 35.91 3.21 3.04 7617.59
FPVC1-4 71.52 36.06 3.16 3.20 7523.93
FPVC1-5 70.89 36.00 3.24 3.07 7450.42
FPVC1-6 71.28 36.01 3.20 3.12 7494.06
FPVC1-7 73.05 35.99 3.13 3.20 7672.74
FBAL1-1 74.49 36.71 2.56 2.73 15225.01
FBAL1-2 74.70 37.05 2.69 2.70 15401.59
FBAL1-3 72.95 37.14 2.60 2.62 15113.01
FBAL1-4 75.49 37.11 2.63 2.47 15655.81
FBAL1-5 71.33 36.96 2.55 2.50 14736.78
FBAL1-6 71.40 37.06 2.58 2.72 14739.10
73
Table 7 - Representative dimensions for specimens used in quasi-static tests
Specimen Total
thickness
First skin
thickness
Second skin
thickness
[mm] [mm] [mm]
SNL11-1 35.38 2.59 2.50
SNL11-2 35.62 2.67 2.59
SNL11-3 35.56 2.66 2.35
SNL11-4 35.61 2.70 2.45
SNL11-5 35.57 2.75 2.63
SNL21-1 36.68 3.87 2.69
SNL21-2 36.57 3.72 2.70
SNL21-3 36.64 3.45 2.71
SNL21-4 36.07 3.12 2.60
SNL21-5 36.59 3.69 2.77
SPVC1-1 35.65 2.86 2.92
SPVC1-2 36.00 3.23 2.98
SPVC1-3 35.61 2.91 2.80
SPVC1-4 35.75 3.01 2.78
SPVC1-5 35.90 3.15 2.72
SPVC1-6 35.61 2.90 3.15
SBAL1-1 36.90 2.49 2.46
SBAL1-2 37.19 2.61 2.45
SBAL1-3 36.94 2.48 2.38
SBAL1-4 37.10 2.52 2.46
SBAL1-5 36.92 2.72 2.56
74
Table 8 - Representative dimensions for specimens used in impact tests
Specimen Total
thickness
First skin
thickness
Second
skin
thickness
Specimen Total
thickness
First skin
thickness
Second
skin
thickness
[mm] [mm] [mm] [mm] [mm] [mm]
DNL11-1 35.38 2.66 2.55 SPVC1-1 36.02 3.24 3.06
DNL11-2 35.55 2.62 2.55 SPVC1-2 35.98 3.26 3.00
DNL11-3 35.51 2.63 2.51 SPVC1-3 35.65 3.12 2.89
DNL12-4 35.95 2.54 2.58 SPVC2-4 36.12 3.25 3.14
DNL12-5 36.09 2.57 2.53 SPVC2-5 36.24 3.41 3.08
DNL12-6 35.80 2.52 2.54 SPVC2-6 36.10 3.34 3.12
DNL12-7 35.66 2.66 2.62 SPVC2-7 36.17 3.40 3.15
DNL12-8 35.78 2.58 2.59 SPVC2-8 36.00 3.40 3.00
DNL12-9 35.83 2.60 2.55 SPVC2-9 35.95 3.12 3.03
DNL12-10 36.34 2.53 2.45 SPVC2-10 36.26 3.35 3.01
DNL12-11 36.52 2.68 2.67 SBAL1-1 37.16 2.71 2.74
DNL12-12 36.23 2.52 2.52 SBAL1-2 36.91 2.66 2.54
DNL12-13 36.03 2.61 2.55 SBAL1-3 37.08 2.65 2.57
DNL12-14 36.25 2.53 2.68 SBAL1-4 36.88 2.69 2.61
DNL12-15 36.00 2.69 2.49 SBAL1-5 37.03 2.64 2.74
DNL21-1 36.45 3.48 2.72 SBAL1-6 36.59 2.61 2.59
DNL21-2 36.33 3.28 2.76 SBAL1-7 36.93 2.70 2.60
DNL21-3 36.83 3.47 2.77 SBAL1-8 36.76 2.79 2.64
DNL21-4 36.58 3.38 2.69 SBAL1-9 36.87 2.76 2.61
DNL21-5 36.01 3.18 2.81 SBAL1-10 36.91 2.74 2.62
DNL22-6 35.92 3.20 2.77 SBAL1-11 37.04 2.65 2.59
DNL22-7 35.90 3.17 2.72 SBAL1-12 37.09 2.55 2.64
DNL22-8 36.11 3.12 2.75 SBAL1-13 36.94 2.57 2.53
DNL22-9 35.84 3.23 2.56 SBAL1-14 37.03 2.63 2.67
DNL22-10 36.10 3.21 2.65 SBAL1-15 37.03 2.57 2.53
75
Table 9 - Dimensions, weight and density of specimens from each manufactured panel
Panel Specimen properties
Length Breadth Total
thickness
Weight Density
[mm] [mm] [mm] [g] [Kg/m3]
NL11 151.24 100.44 35.57 250.00 462.74
152.10 99.89 35.61 260.00 480.60
151.60 101.32 35.56 251.00 459.49
NL12 149.07 101.57 35.94 256.00 470.44
149.42 103.55 35.93 261.00 469.49
149.59 99.87 35.98 261.00 485.51
NL21 151.90 98.26 36.59 270.00 494.34
150.71 99.39 36.07 258.00 477.47
150.57 99.64 36.64 270.00 491.19
NL22 150.41 98.71 36.36 252.00 466.87
150.05 98.01 36.11 251.00 472.62
148.11 98.08 36.10 247.00 471.09
PVC1 148.89 101.24 35.61 198.00 368.87
152.00 99.60 35.90 202.00 371.70
148.35 99.97 35.75 198.00 373.50
PVC2 148.43 98.23 36.09 197.00 374.43
151.68 99.63 36.09 207.00 379.57
148.73 101.68 36.22 203.00 370.63
BAL1 146.90 96.64 36.92 197.00 375.83
147.03 99.44 37.10 205.00 377.93
147.42 102.50 36.94 202.00 361.86
76
C. Impact test data: Load vs displacement and Absorbed
energy vs Displacement
Figure 66 - Load vs Displacement and Absorbed energy vs Displacement for PVC - 50J
Figure 67 - Load vs Displacement and Absorbed energy vs Displacement for NL10 - 50J
Figure 68 - Load vs Displacement and Absorbed energy vs Displacement for NL20 - 50J
77
Figure 69 - Load vs Displacement and Absorbed energy vs Displacement for Balsa - 50J
Figure 70 - Load vs Displacement and Absorbed energy vs Displacement for PVC - 100J
Figure 71 - Load vs Displacement and Absorbed energy vs Displacement for NL10 - 100J
78
Figure 72 - Load vs Displacement and Absorbed energy vs Displacement for NL20 - 100J
Figure 73 - Load vs Displacement and Absorbed energy vs Displacement for Balsa - 100J
Figure 74 - Load vs Displacement and Absorbed energy vs Displacement for PVC - 150J
79
Figure 75 - Load vs Displacement and Absorbed energy vs Displacement for NL10 - 150J
Figure 76 - Load vs Displacement and Absorbed energy vs Displacement for NL20 - 150J
Figure 77 - Load vs Displacement and Absorbed energy vs Displacement for Balsa - 150J
80
Figure 78 - Load vs Displacement and Absorbed energy vs Displacement for PVC - 200J
Figure 79 - Load vs Displacement and Absorbed energy vs Displacement for NL10 - 200J
Figure 80 - Load vs Displacement and Absorbed energy vs Displacement for NL20 - 200J
81
Figure 81 - Load vs Displacement and Absorbed energy vs Displacement for Balsa - 200J
Figure 82 - Load vs Displacement and Absorbed energy vs Displacement for PVC - 300J
Figure 83 - Load vs Displacement and Absorbed energy vs Displacement for NL10 - 300J
82
Figure 84 - Load vs Displacement and Absorbed energy vs Displacement for NL20 - 300J
Figure 85 - Load vs Displacement and Absorbed energy vs Displacement for Balsa - 300J
Figure 86 - Load vs Displacement and Absorbed energy vs Displacement for NL10 - 450J
83
Table 10 - Summary of peak forces and corresponding absorbed energy - PVC, NL20 and Balsa specimens
Indicent energy
Initial velocity
1st peak 2st peak
Disp. Force Energy Disp. Force Energy
[J] [m/s] [mm] [N] [J] [mm] [N] [J]
50J
DPVC1-2 49.20 3.01 8.21 11361 48.39 - - -
DPVC2-5 55.48 3.20 8.65 11573 51.74 - - -
DNL21-2 49.65 3.02 8.88 10673 49.02 - - -
DNL21-5 55.48 3.20 8.94 11199 52.19 - - -
DBAL1-4 51.48 3.08 3.93 10532 20.77 - - -
DBAL1-7 55.48 3.20 3.61 9913 18.36 - - -
100J
DPVC1-1 100.65 4.31 8.39 11056 50.78 - - -
DPVC2-6 110.37 4.51 8.35 12623 53.80 - - -
DNL21-1 96.88 4.23 9.06 12691 57.52 - - -
DNL22-6 110.37 4.51 9.76 12557 61.58 - - -
DBAL1-1 99.37 4.28 8.21 11493 58.30 - - -
DBAL1-8 108.90 4.48 5.39 10247 29.00 - - -
DBAL1-11 106.04 4.42 7.20 13074 54.00 - - -
DBAL1-12 107.46 4.45 7.14 13105 49.93 - - -
150J
DPVC1-3 159.68 5.42 8.67 13142 58.23 - - -
DPVC2-7 164.99 5.51 8.84 13315 58.86 - - -
DNL21-3 159.68 5.42 9.19 12575 63.01 - - -
DNL22-7 164.99 5.51 10.29 11491 64.18 - - -
DBAL1-5 159.68 5.42 6.34 10600 41.71 - - -
DBAL1-9 164.99 5.51 6.44 10630 42.75 - - -
200J
DPVC2-4 210.92 6.23 8.81 13383 57.81 45.83 5863 204.83
DPVC2-8 219.02 6.35 7.86 12252 45.57 45.96 6327 212.18
DNL21-4 214.91 6.29 9.09 12372 60.33 40.78 5838 213.67
DNL22-8 219.02 6.35 8.99 12019 55.74 40.96 5892 213.30
DBAL1-6 210.92 6.23 7.72 15206 68.71 - - -
DBAL1-10 223.25 6.41 7.38 12236 51.28 51.28 5184 209.37
DBAL1-13 219.02 6.35 7.99 13490 58.02 50.91 5008 190.88
300J
DPVC2-9 310.12 5.06 8.35 13579 55.01 46.64 12456 238.39
DPVC2-10 319.71 5.13 8.24 12021 54.90 47.33 11208 239.30
DNL22-9 314.86 5.09 9.60 13162 64.50 46.02 11056 254.55
DNL22-10 300.95 4.98 8.70 12860 55.63 44.62 11771 250.08
DBAL1-14 310.12 5.06 7.32 13272 58.03 48.97 8947 225.68
DBAL1-15 314.86 5.09 7.44 12103 49.77 50.32 8148 210.35
84
Table 11 - Summary of peak forces and corresponding absorbed energy - NL10 specimens
Indicent energy
Initial velocity
1st peak
2st peak
3rd peak
Disp. Force Energy Disp. Force Energy Disp. Force Energy
[J] [m/s] [mm] [N] [J] [mm] [N] [J] [mm] [N] [J]
50J
DNL11-3 50.10 3.04 11.79 7366 45.43 - - - - - -
DNL12-6 54.96 3.18 8.87 7425 32.84 - - - - - -
100J
DNL11-1 98.11 4.25 12.92 7398 49.79 - - - - - -
DNL11-2 99.37 4.28 12.06 9513 57.86 - - - - - -
DNL12-7 110.37 4.51 10.63 8694 46.49 - - - - - -
150J
DNL12-4 159.68 5.42 11.33 8270 44.93 - - - - - -
DNL12-8 167.75 5.56 10.26 9913 52.2 39.95 4015 163.01 - - -
DNL12-10 159.68 5.42 10.48 7555 37.95 - - - - - -
200J
DNL12-5 214.91 6.29 14.06 8978 66.26 - - - - - -
DNL12-9 223.25 6.41 12.01 9935 59.4 - - - - - -
300J
DNL12-11 314.86 5.09 12 6646 38.77 39.24 16356 277.46 - - -
DNL12-12 319.71 5.13 11.81 7803 45.25 - - - - - -
DNL12-13 314.86 5.09 11.69 8359 48.2 32.06 12468 228.42 49.91 8366 316.4
450J
DNL12-14 462.97 6.18 12.17 7580 45.55 37.6 15707 276.84 51.74 9227 376.1
DNL12-15 471.65 6.23 10.75 8013 45.03 36.47 14579 294 56.55 9641 439.59
85
D. Post-test specimen photographs
D.1 Quasi-static tests
Figure 87 - SPVC1-2 after test
Figure 88 - SPVC1-3 after test
Figure 89 - SNL11-1 after test
Figure 90 - SNL11-2 after test
86
Figure 91 - SNL21-1 after test
Figure 92 - SNL21-2 after test
Figure 93 - SBAL1-1 after test
87
Figure 94 - SBAL1-2 after test
D.2 Impact tests
Figure 95 - DPVC1-2 after test
Figure 96 - DPVC2-5 after test
88
Figure 97 - DPVC1-1 after test
Figure 98 - DPVC2-6 after test
Figure 99 - DPVC1-3 after test
Figure 100 - DPVC2-7 after test
89
Figure 101 - DPVC2-4 after test
Figure 102 - DPVC2-8 after test
Figure 103 - DPVC2-9 after test
Figure 104 - DPVC2-10 after test
90
Figure 105 - DNL11-3
Figure 106 - DNL12-6
Figure 107 - DNL11-1
Figure 108 - DNL11-2
91
Figure 109 - DNL12-7
Figure 110 - DNL12-4
Figure 111 - DNL12-8
Figure 112 - DNL12-10
92
Figure 113 - DNL12-5
Figure 114 - DNL12-9
Figure 115 - DNL12-11
Figure 116 - DNL12-12
93
Figure 117 - DNL12-13
Figure 118 - DNL12-14
Figure 119 - DNL12-15
Figure 120 - DNL21-2
94
Figure 121 - DNL21-5
Figure 122 - DNL21-1
Figure 123 - DNL22-6
Figure 124 - DNL21-3
95
Figure 125 - DNL22-7
Figure 126 - DNL21-4
Figure 127 - DNL22-8
Figure 128 - DNL22-9
96
Figure 129 - DNL22-10
Figure 130 - DBAL1-4
Figure 131 - DBAL1-7
Figure 132 - DBAL1-1
97
Figure 133 - DBAL1-8
Figure 134 - DBAL1-11
Figure 135 - DBAL1-12
Figure 136 - DBAL1-5
98
Figure 137 - DBAL1-9
Figure 138 - DBAL1-6
Figure 139 - DBAL1-10
Figure 140 - DBAL1-13
99
Figure 141 - DBAL1-14
Figure 142 - DBAL1-15