Sie sind auf Seite 1von 70

TALLINN TECHNICAL UNIVERSITY

Faculty of Mechanical Engineering


Department of Machinery

Kristjan Tabri

Local Impact Strength of Sandwich Panels


Master’s Thesis

Supervisor: Petri Varsta, Professor


Instructors: Jaan Metsaveer, Professor Emeritus
Martin Eerme, Doctor of Philosophy

Tallinn 2003
AUTHOR’S DECLARATION

I assure that this master’s thesis is a result of my personal work and that no other than
the indicated aids have been used for its completion. Furthermore I assure that all
quotations and statements that have been inferred literally or in a general manner from
published or unpublished writings are marked as such. Beyond this I assure that the
work has not been used, neither completely nor in parts, for the passing of any
previous examinations.

Tallinn, February 7, 2003

Kristjan Tabri
TALLINN TECHNICAL UNIVERSITY ABSTRACT
Title: Local Impact Strength of Sandwich Panels
Author: Kristjan Tabri
Place: Tallinn
Date: 07.02.2003
Number of pages: 68
Number of figures: 47
Supervisor: Petri Varsta, Professor
Instructors: Jaan Metsaveer, Professor Emeritus; Martin Eerme, PhD.

Keywords: Sandwich panels, impact load, bending energy, membrane energy,


laboratory experiments, FE simulations, Cowper-Symonds model,
The purpose of the study is to understand the behaviour of I-core steel sandwich panel subjected to a
lateral impact load. Furthermore, the aim is to derive an analytical model describing panel’s behaviour and
the consequences of impact. Due to the impact, faceplate of the panel is deformed in high velocity. It
means that dynamic behaviour of materials should be considered. To verify proposed analytical model
data is obtained by laboratory experiments and by finite element calculations.

The behaviour of sandwich panels is studied in a series of laboratory tests, where sandwich panels with
four different configurations are tested. General structure of tested panels remains unchanged during the
tests and the only changing parameter is the thickness of the faceplate. The effect of core material is
investigated by filling some of the panels with urethane foam. In the laboratory tests panels are hit by an
impact head, which has some predetermined mass and velocity. The most important results of the
laboratory experiments are plastic energy absorption of the panel and the extent of deformation.

In addition to the laboratory experiments, impacts are simulated by finite element method using program
LS-Dyna. FE simulations provide a possibility to determine what happens in a sandwich panel during the
impact. The FE simulations are used to obtain information about the velocity of the faceplate and core
displacements. This analysis gives the transversal velocity profile, which can be approximated by linear
line. The decrease of the velocity is shown to be slightly non-linear. Plastic energy absorption and the
extent of the deformation are determined also in FE simulations. Several assumptions made in derivation
of the analytical formulation are verified by the FE calculations.

Derived analytical model assumes that all the energy is absorbed by the faceplate of the panel, as
displacements at steel core are small compared to the displacements of the faceplate and can thus be
neglected. Furthermore, it is assumed that the panel has infinite length and the global bending of the
faceplate does not occur. The maximum extent of the deformation is assumed to be equal to the span of
inner supports. Formulations for energy absorption are derived separately for membrane and bending
energy. Both elastic and plastic deformation energies are considered. The effect of filling material is taken
into account by using Winkler’s foundation.

Comparison with laboratory experiments and FE simulations support the purposed analytical model as
scatter between the results obtained by different methods is small. In the case of plastic deformation
energy the scatter is at worst 10%. Scatter is slightly larger in the case of total deformation energy. In that
case the analytical model overestimates the deformation energy in lower deformation values. The reason
for that is the methodology of calculation of the elastic energy. An improved solution is suggested for
further research. Guidelines how to describe the behaviour of sandwich panel more precisely and thus how
to limit the number of assumptions are also suggested.
TALLINNA TEHNIKAÜLIKOOL RESÜMEE
Pealkiri: “Sandwich” paneelide lokaalne tugevus löökkoormuste korral
Autor: Kristjan Tabri
Koht: Tallinn
Kuupäev: 07.02.2003
Lehekülgede arv: 68
Jooniste arv: 47
Järelvaataja: Professor Petri Varsta
Juhendajad: Emeriitprofessor Jaan Metsaveer, vanemteadur Martin Eerme
Võtmesõnad: “Sandwich” paneelid, löökkoormus, paindeenergia, membraanenergia,
lõplike elementide meetod, Cowper-Symond’i mudel,
Käesoleva töö eesmärgiks on tutvuda “sandwich” paneelide käitumisega löökkoormuste korral. Paneeli
käitumise ja löökkoormuse mõjul tekkinud tagajärgede kirjeldamiseks on tuletatud analüütilised valemid.
Löökkoormuse tulemusena deformeerub paneeli ülemine plaat suurel kiirusel, mis eeldab dünaamiliste
materjaliomaduste kasutamist. Analüütilise mudeli õigsust on kontrollitud laborikatsetest ja lõplike
elementide meetodil tehtud arvutustest saadud informatiooni kasutades.

“Sandwich” paneelide käitumist uuriti laborikatsete abil, kus testiti nelja erineva konfiguratsiooniga
paneeli. Paneelid erinesid plaadistuse paksuse ja täiteaine poolest. Testitud paneelidest kolm ei sisaldanud
täiteainet ja üks oli täidetud uretaanvahuga. Laborikatsetes lasti paneelile kukkuda ümaral kehal, millel oli
kindlaksmääratud mass ja kiirus. Laborikatsetest saadud tähtsamad suurused olid plastne
deformatsioonienergia ja vigastuse ulatus.

Laborikatsetele lisaks simuleeriti kuuli ja paneeli kokkupõrget lõplike elementide meetodil kasutades
programmi LS-Dyna. Lõplike elementide simulatsioonid annavad võimaluse jälgida paneeli käitumist
kokkupõrke ajal. Simulatsioonide abil on võimalik saada informatsiooni paneelis aset leidvate kiiruste ja
siirete kohta. Analüüs osutas, et paneeli ülemise plaadi deformeerumiskiiruse põik- ja pikisuunalist jaotust
saab aproksimeerida lineaarsete sirgete abil. Samuti ilmnes, et kiiruse vähenemise kirjeldamiseks ei piisa
vaid lineaarsest aproksimatsioonist. Sarnaselt laborikatsetele arvutati ka lõplike elementide meetodil
tehtud simulatsioonide abil plastne deformatsioonienergia ja vigastuse ulatus. Mitmete analüütiliste
valemite tuletamisel tehtud oletuste õigsust on kontrollitud simulatsioonidest saadud informatsiooni abil.

Tuletatud analüütiline mudel oletab, et kogu löögist saadud energia neeldub paneeli ülemises
plaadis kuna paneeli jäigastajates aset leidvad siirded on väikesed võrreldes plaadi siiretega.
Samuti on oletatud, et paneel on lõpmatu pikkusega ja löökkoormus ei tekita ülemises plaadis
laiaulatuslikku läbipainet. Vigastuse maksimaalseks ulatuseks põiksuunas on võetud paneeli
sisemiste jäigastajate vahekaugus. Valemid nii elastse kui ka plastse deformatsioonienergia
arvutamiseks on tuletatud eraldi painde- ja membraanenergia jaoks. Uretaanvahu mõju on võetud
arvesse kasutades Winkler’i teooriat.

Laborikatsete, lõplike elementide meetodil tehtud arvutuste ja analüütilise mudeliga saadud


tulemuste kokkulangevust võib lugeda heaks kuna erinevused eri tulemuste vahel on väikesed.
Plastse deformatisoonienergia korral erinevus on halvimal juhul 10%. Erinevused on suuremad
koguenergia korral, mil analüütiline mudel ülehindab neeldunud eneriat väikeste vigastuste puhul.
Erinevuse tekib elastse deformatioonienergia arvutamisel kasutatud metoodika. Edasiseks
uurimiseks on välja pakutud parandatud mudel elastse energia täpsemaks kirjeldamiseks. Samuti
on antud soovitusi tuletatud mudeli parandamiseks ja tehtud oletuste mõju vähendamiseks.
Local Impact Strength of Sandwich Panels Kristjan Tabri

PREFACE

This work is done for the EU-project entitled: Advanced Composite Sandwich Steel Structures.
This project started on 1.04.2000 and its duration is three years. The SANDWICH project will
develop products utilising sophisticated lightweight steel sandwich panels for primary load
carrying structures. The Ship Laboratory of Helsinki University of Technology (HUT)
participates in the project as a partner.

I am grateful to supervisor, Professor Petri Varsta, and to the instructors Professor Emeritus
Jaan Metsaveer and Ph.D. Martin Eerme for valuable and essential guidance and
encouragement they gave me throughout the study.

I would like to express my gratitude to Dr.Tech. Pentti Kujala and Lic.Tech. Hendrik Naar for
giving me vital instructions in many fields. I also wish to thank the personnel both in HUT and
in Tallinn Technical University for pleasant and versatile contribution. Last but not least, I
would like to thank Hannele for the support she gave me throughout the study.

3
Local Impact Strength of Sandwich Panels Kristjan Tabri

CONTENTS

ABSTRACT ..............................................................................................................................................................1

KOKKUVÕTE .........................................................................................................................................................2

PREFACE .................................................................................................................................................................3

CONTENTS ..............................................................................................................................................................4

NOTATIONS ............................................................................................................................................................6

1 INTRODUCTION ...........................................................................................................................................8

1.1 BACKGROUND OF THE STUDY ....................................................................................................................8

1.2 RESEARCH PROBLEMS AND THE PURPOSE OF THE STUDY ...........................................................................8

1.3 LIMITATIONS OF THE STUDY ....................................................................................................................11

2 EXPERIMENTAL SETUP ...........................................................................................................................13

2.1 TESTED STRUCTURES AND MATERIAL PROPERTIES...................................................................................13

2.2 TEST EQUIPMENT, DATA ACQUISITION AND STORAGE ..............................................................................16

2.3 MEASURED / CALCULATED QUANTITIES ..................................................................................................18

2.3.1 Velocity of the impact head before the impact....................................................................................19

2.3.2 Permanent deflection of the faceplate ................................................................................................20

2.3.3 Deformation energy of the panel ........................................................................................................20

2.4 RESULTS OF THE LABORATORY TESTS......................................................................................................21

3 FINITE ELEMENT ANALYSIS .................................................................................................................26

3.1 GEOMETRY OF THE FE MODEL AND THE SIMULATION PROCEDURE ..........................................................26

3.2 MATERIAL PROPERTIES OF THE MODEL ....................................................................................................29

3.3 RESULTS OF THE FE ANALYSIS ................................................................................................................30

4 ANALYTICAL FORMULATIONS.............................................................................................................35

4.1 BACKGROUND AND MAIN ASSUMPTIONS..................................................................................................35

4.2 ANALYTICAL DESCRIPTION OF THE DEFORMATION SHAPE .......................................................................37

4.3 STRAIN RATE ...........................................................................................................................................39

4.4 ENERGY ABSORPTION OF THE PANEL .......................................................................................................42

4.4.1 Elastic energy absorbed by bending...................................................................................................42

4.4.2 Elastic energy absorbed by membrane mechanism ............................................................................46

4.4.3 Plastic energy absorbed by bending...................................................................................................47

4.4.4 Plastic energy absorbed by membrane mechanism ............................................................................49

4.4.5 Energy absorbed by core filling .........................................................................................................49

4.4.6 Approximate solution for membrane energy.......................................................................................50

4
Local Impact Strength of Sandwich Panels Kristjan Tabri

4.5 SOLUTION PROCEDURE ............................................................................................................................54

5 COMPARISON BETWEEN THE LABORATORY TESTS, FE CALCULATIONS AND THE


ANALYTICAL FORMULATIONS .....................................................................................................................58

6 CONCLUSIONS ............................................................................................................................................64

REFERENCES .......................................................................................................................................................67

5
Local Impact Strength of Sandwich Panels Kristjan Tabri

NOTATIONS

a, b, c, d constants
B breath
cV ratio between initial and average velocity
C1 , C2 constants determining the shape of deformation
CS coefficient used to scale yield stress
D constant describing dynamic behaviour of material
E Young’s modulus, energy
F force
G shear modulus of steel material
k foundation constant
K constant describing material properties
L length
m mass
MP plastic moment
pF support reaction
q constant describing dynamic behaviour of material
r radius
R width of deformation
t plate thickness
v velocity
V volume
w deflection

maximum deflection

angle

S angle of deformation

V angle of velocity profile

strain
strain rate
6
Local Impact Strength of Sandwich Panels Kristjan Tabri

distributed load

Poisson's ratio of steel material

F compressive strength of filling material

Y static yield stress


D
Y dynamic yield stress

Subscripts
0 initial
A average
B bending
EF effective
F filling
I impact body
M membrane

Superscripts
* simplified equation
E elastic
P plastic
D dynamic

Abbreviations
FE Finite Element

7
Local Impact Strength of Sandwich Panels Kristjan Tabri

1 INTRODUCTION

1.1 Background of the study

The weight of structures has significant importance in ships and in other forms of
transportation. Decreased structural weight allows vessel to transport larger amount of goods
and passengers with a lower expenses. In luxurious cruise ships more and more attractions
should be added to ship in order to keep customers satisfied. All additional recreational
alternatives increase the lightweight of the and in order to keep the buoyancy in same level, the
weight of structures should be increased. The importance of ship buoyancy can be hardly
overestimated as it has straight impact to the resistance and thus also to the energy consumption
of vessel.

Though the weight is one of the most important parameters in design, there are still a lot of
other requirements and demands for structures, which should be satisfied. Especially in marine
structures attention should be paid to strength, noise, vibrations, safety, manufacturing and
installation of structures. Large amount of requirements have made it almost impossible to
satisfy all the demands just by improving conventional structures.

Increasing demand for the lighter and more efficient structures has challenged the engineers to
invent new solutions to improve the structures and satisfy the demands.

1.2 Research problems and the purpose of the study

The weight of structures can be decreased using lighter materials, new constructions or
combining them. In nowadays industry sandwich structures are used to overcome the increased
demands. General drawing of the sandwich panel is given in Figure 1. Two outer layers, skins

8
Local Impact Strength of Sandwich Panels Kristjan Tabri

or faceplates, are made of material that gives enough strength and stiffness, abrasive and
corrosive resistance, noise isolation and easy production. In order to increase the thickness of
the panel, and thus to increase the stiffness, without using heavy materials in the skins, a light
core material is placed between the plates. Several criteria should be considered when selecting
the core material. Density, mechanical properties, bond properties, fire isolation are just few
examples.

Figure 1. I-core steel sandwich panel.

Combination of high stiffness and low weight was first used in aircrafts during the Second
World War. Combination of balsa in core and veneer in skins was used because of the lack of
high strength materials. Nowadays sandwich panels are used even in space research industry
where beside the other properties also high impact resistance is appreciated.

Improved welding techniques, especially laser welding, have made it possible to connect very
thin sheets to each other and so to manufacture panels where thin faceplates are welded to steel
core structure. In marine industry the combination of new welding possibilities, material and
strength properties of sandwich panels have made them to be good substitution for
conventional structures. Good examples are balconies, decks and bulkheads where sandwich
panels replace conventional stiffened plating. Figure 2 presents some possible uses for
sandwich panels.

Sandwich panels are efficient in means of global response as panel’s thickness and sectional
modulus are bigger compared to conventional stiffened plating. Moment caused by bending is
carried by the faceplates while light and low-strength core sustains shear forces. Core
contributes to the global response also in other ways. It makes it possible to increase the span of
the faceplates without loosing local stiffness. Core also supports the faceplates and distributes
stresses to larger area and so prevents the global bending of the faceplate. In other hand the
contribution allows to reduce the thickness of the faceplates and to decrease weight.

9
Local Impact Strength of Sandwich Panels Kristjan Tabri

conventional ship structure ship structure with sandwich panel

teak coated sandwich panel

Figure 2. Usage of the sandwich panels in contemporary ship structures.

Weight reduction by using thinner skins introduces a new problem. Though the required global
bending resistance can be achieved by using very thin faceplates, still it can be weakened even
when relatively small body strikes the panel and causes permanent damage. Local deflection in
the faceplate of the panel can decrease the bending resistance significantly. Again core is one
possibility to prevent the local deflections, but also the use of some faceplate coatings or new
steel core structures can be effective to prevent the serious consequences caused by any kind of
impacts on sandwich panels. It should be noted that impact not only causes local deflections to
the faceplate, but may also cause widespread global bending of the faceplate. The global
bending of the faceplate already has crucial effect to the bending resistance and the whole
structure can be close to the collapse. Figure 3 shows the typical local damage of sandwich
panel as a result of strike by a spherical object.
10
Local Impact Strength of Sandwich Panels Kristjan Tabri

Figure 3. Local deflection in sandwich panel.

Above described phenomena indicate that together the global behaviour also the local
behaviour of sandwich panels should be considered. The purpose of the study is to investigate
the local behaviour of the sandwich panels subjected to lateral impact load and to derive
analytical formulations describing the behaviour. More precisely the purpose includes the
following matters:

learn about the local impact behaviour of sandwich panels,

study the influence of the faceplate thickness and material properties,

study the effect of core material.

In one hand formulations are to be simple and easy to use, but still they have to take into
account all the major phenomena concerning the impact event. Attention should be paid to the
strain-rate sensitive behaviour of materials; elastic deformation energy of a panel can be quite
high in a dynamic process and cannot be neglected; the shape of deflection caused by impact
load is different from deflection, which is caused by static load etc. To verify the results of the
analytical formulations, series of laboratory experiment and finite element (FE) simulations are
carried out.

1.3 Limitations of the study

The number of different designs of sandwich panels is large and it is obvious that single study
cannot embrace all of them. This study includes only one design, where material properties and
the thickness of the faceplate are changed. The effect of filling material is studied by urethane
foam. Lateral impact load is caused by spherical impact head, which is used to strike the

11
Local Impact Strength of Sandwich Panels Kristjan Tabri

panels. Several impacts are simulated using FE program LS-Dyna, but as numerical simulations
are time consuming, number of FE simulations is smaller compared to the laboratory
experiments.

Analytical formulations are derived assuming infinite panel dimensions and limited extent of
deformation. Tearing and global bending of the faceplate are not considered in analytical
model. Strain-rate behaviour of the materials is considered by using Cowper-Symonds
constitutive equation.

12
Local Impact Strength of Sandwich Panels Kristjan Tabri

2 EXPERIMENTAL SETUP

In order to obtain data to verify the analytical formulations series of laboratory experiments are
carried out. This chapter gives an overview of tested structures and the equipment and methods
used to conduct the laboratory experiments. Also the results of the laboratory experiments are
presented.

2.1 Tested structures and material properties

Altogether 96 impact tests are made for different sandwich panels. General structure of tested
panels remains unchanged throughout the tests and is given in Figure 4. The only changing
parameter is the thickness of the faceplate, which can be 1 to 3 mm with 1 mm spacing.

Figure 4. General drawing of the sandwich panel.

13
Local Impact Strength of Sandwich Panels Kristjan Tabri

Panels have 4 mm thick I-profiles with 120 mm span as steel core. Material properties of the
steel plates are determined by carrying out tensile tests for specimens cut from the faceplates.
Tensile tests are carried out for the following specimens:
(i) three specimens cut from 1 mm plates,
(ii) one specimen cut from 2 mm plate,
(iii) three specimen cut from 3 mm plate.

Information obtained from the tensile tests is gathered into Table 1.

Table 1. Results of the tensile tests.


Name of the Breath of the Thickness Ultimate Ultimate
0.2
panel specimen strength strain
[mm] [mm] [N/mm2] [N/mm2] A5, %

5a-6 24.95 3.03 385 485 33.5

I03 24.95 3.08 388 541 33.5

5a-6 25 3.03 370 482 36

N6 12.5 2.02 428 520 28

5a-13 10.98 1.00 159 288 60

5a-17 10.96 1.00 157 291 64

5a-5 12.4 1.00 179 290 47

Tensile test show that 1 mm thick steel plates are made of material, which yield stress is
significantly lower compared to materials used in 2 and 3 mm plates. For brevity, in following
discussions just low and high yield is used instead of exact values. Obtained yield stress values
are used to predict the strain-rate sensitive behaviour of materials. This behaviour is considered
by using Cowper-Symonds (Jones, 1989) equation, which uses constants D and q to describe
the behaviour. For mild or low yield steel, those constants can easily be found from the
literature. For high strength steels the information about the strain-rate behaviour is scarce and
difficult to get. Some investigations carried out in automotive industry have revealed that high
strain-rate increases the yield stress of high strength steels approximately 20%. According to
that material constants are also calculated for high-yield materials and gathered into Table 2.
More detailed description of Cowper-Symonds model is given in Chapter 4.3.

14
Local Impact Strength of Sandwich Panels Kristjan Tabri

Table 2. Strain rate properties of used steels.

Panel Material Y [MPa] D q

t=1 [mm] Mild steel 179 40.4 5


t=2 [mm] High-yield steel 428 300000 6
t=3 [mm] High-yield steel 379 300000 6

The effect of core material is investigated by filling some of the panels with 2 mm faceplates
with urethane foam. Mechanical properties of the urethane filling are obtained according to the
measured density from literature (Kolsters; Romanoff, 2000) and a graph given in Figure 5,
which presents the relation between the density and the compressive strength of the urethane
foam /see reference 29/.

Figure 5. Properties of the urethane foam.

15
Local Impact Strength of Sandwich Panels Kristjan Tabri

Properties of urethane foam are gathered into Table 3.

Table 3. Properties of the urethane foam.

Density Compressive Shear modulus Young’s


strength F modulus E
G
[kg/m3] [MPa] [MPa] [MPa]

72 0.62 4.8 21

Good overview about the laboratory tests can be given by test matrix, which is presented in
Figure 6.

Core type I-core

Faceplate
thickness [mm] 1 2 3

Yield stress Low High High

Filling
Foam

Void Void
Void

Figure 6. Test matrix.

2.2 Test equipment, data acquisition and storage

The test equipment mainly consists of a test stand and of a data acquisition system. A
schematic picture of the test stand can be seen in Figure 7. The impact system includes a bar,
supported vertically by rollers to allow sliding movement, a replaceable extra mass, a
replaceable nozzle and three sensors. Impact head with conical nozzle is presented in Figure 8.
The impact head, having some predetermined mass, is dropped on the panel and data is
gathered into a computer and saved as a text file. The energy of the impact head is changed
altering its mass and dropping height. Tested panels are hit by a spherical impact nozzle, which
is made of 25-millimeter bearing ball and is shown in Figure 9.

16
Local Impact Strength of Sandwich Panels Kristjan Tabri

Figure 7. Test system.

The parameters of the impact system are the following:


(i) mass from 4.5 to about 37 kg
(ii) dropping height up to 1250 mm
(iii) velocity at the moment of impact up to 5 m/s,
(iv) potential energy from 2 up to 450 J.

Figure 8. Impact head with conical nozzle.

17
Local Impact Strength of Sandwich Panels Kristjan Tabri

Figure 9. Spherical impact nozzle.

During the impact three quantities as a function of time are measured:


(i) force acting between the impact head and the panel,
(ii) acceleration of the impact system,
(iii) displacement of the impact system.

The test system uses one sensor for each quantity. General information about the sensors is
gathered into Table 4. Data acquisition system uses one channel per sensor. During the impact,
the system reads a value from one channel and switches to another channel in 7.5 s intervals.
Due to that information from one channel is registered in 22.5 s intervals. This equals to a
sampling rate of a little over 44 kHz.

Table 4. Sensors used in the impact system.

Quantity Manufacturer Model Range Type


Force HBM U9B 50 kN Strain cage

Acceleration B&K 3073 2000 G Piezo electric

Displacement Midori CPP-45 ----- Potentiometer

2.3 Measured / calculated quantities

In the tests, the following information is registered:


(i) dropped mass,
(ii) dropping height,
(iii) displacement, acceleration and force signals,
(iv) bouncing height of the impact head,

18
Local Impact Strength of Sandwich Panels Kristjan Tabri

(v) permanent deflection at the faceplate,


(vi) impact duration,
(vii) impact coordinates.

According to the registered data the following quantities are calculated:


(i) velocity of the impact head before the impact,
(ii) kinetic energy of the impact head before the impact,
(iii) plastic deformation energy of the panel.

In following sections some of the main calculations are explained.

2.3.1 Velocity of the impact head before the impact

Velocity of the impact head before the impact head is calculated from the dropping height
using the energy principle. The kinetic energy of the head just before the impact is certain
amount smaller than the potential energy of the impact head before the drop, since some of the
potential energy goes to the revolving motion of the rollers. If all four of the rollers would
follow the movements of the impact head, the rollers would eventually give back their kinetic
energy, but because of the clearance between the rollers and the sliding bar it is assumed that
only two of the rollers follow the bar. The velocity before the impact could also be calculated
by the time-derivative of the registered displacement, but mentioned energy principle
calculations give more accurate results because of the scatter in the displacement measurement.
The calculation is verified by taking the time-derivative of the displacement signal from
repeated trials. The resulting average velocity was then compared to the value obtained by the
energy principle.

19
Local Impact Strength of Sandwich Panels Kristjan Tabri

2.3.2 Permanent deflection of the faceplate

The permanent deflection of the faceplate or shortly dent depth is measured manually with a
digital dial indicator. The indicator is set to zero at the assumed drop point before every test. As
only the first hit is under consideration, the impact head is stopped after the first hit to prevent
repetition.

2.3.3 Deformation energy of the panel

Computer programme uses three different methods to calculate the plastic deformation energy
of the panel:
(i) difference in potential energies between the dropping and bouncing heights,
(ii) numerical integration of displacement-force curve. Displacement and force are
calculated from the signal received from the acceleration sensor,
(iii) numerical integration of displacement-force curve. Displacement and force are
calculated from the signals received from the acceleration sensor and the force
transducer.

In the first method, it is simply assumed that the elastic deformation energy of the panel returns
to the kinetic energy of the impact head. Due to that impact head bounces from the panel and
the bouncing height is measured. According to the bouncing height elastic deformation energy
can be calculated.

Other two methods employ the similar principle. Since the velocity of the impact head just
before the impact and the acceleration as a function of time are known, the motion of the
impact head during the impact can be calculated. On the other hand also the force acting
between the impact head and the panel is known, which means that the deformation energy can
be calculated by integrating force-displacement curve. Force-displacement curve is shown in
Figure 10 where the plastic deformation energy is the area under the curve. Since also the
displacement sensor is employed the deformation energy could be calculated using the signal
from the displacement sensor, but because of the resolution and the mechanical construction of
the displacement sensor this is considered to be inaccurate.
20
Local Impact Strength of Sandwich Panels Kristjan Tabri

Force-displacement curve
3
Plastic deformation
2.5 energy

2
Force [kN]

1.5

0.5

0
0.2 0.25 0.3 0.35 0.4 0.45 0.5
Displacement [cm]

Figure 10. Force-displacement curve.

In case of the first method the bouncing height is determined by the signal from the
displacement sensor. As the accuracy of the displacement sensor is low the first calculation
method is considered to give imprecise results. In case of the second method where the
acceleration signal is integrated, the electronic filtering of the signal distorts the signal and
causes some error to the calculated value. Electronic filtering is used to smoothen the signal,
which is affected by the high frequency vibrations induced to the impact system due to the
collision. Considering these facts the third method is considered to be the most accurate. The
plastic deformation energy presented in Chapter 2.4 is calculated by using the signals received
from the acceleration sensor and from the force transducer

2.4 Results of the laboratory tests

In laboratory experiments panels are hit to the centre of two middlemost compartments, as can
be seen in Figure 11 and in Figure 12. During the impact panel is lying in the floor, which can
be assumed as infinitely rigid compared to the panels.

21
Local Impact Strength of Sandwich Panels Kristjan Tabri

Figure 11. Permanent deformations.

Investigation of the tested panels revealed that the width of the deformation in the faceplate is
limited by the span of the inner supports as can be seen from Figure 11 and Figure 12. Left
picture in Figure 12 presents the panel where the faceplate of the section subjected to the
impact load is very close to global bending, but plating of the adjacent sections remains
undamaged and no deformations can be observed. Bending of the faceplate is considered to be
global when length of the deformation is large compared to the width of the deformation. Right
picture of the same figure also reveals that width of the deformation does not exceed the span
of the supports. Deformation has circular shape until the faceplate bends globally and the
circular shape is stretched to oval.

Figure 12. The extent of the deformation.

Laboratory tests also showed that the impact does not cause noticeable permanent deformations
in inner supports even when the global bending of the faceplate occurs as depicted in Figure 13.

22
Local Impact Strength of Sandwich Panels Kristjan Tabri

Figure 13. Global bending of the faceplate.

Results of the laboratory experiments are presented in Figure 14 to Figure 17. In the figures,
single test is marked with rhomb. In addition, to get a better picture of the panel’s behaviour
also trendlines are presented for every panel type. Initial energy of the impact head or in other
words the total deformation energy of the panel as a function of permanent deflection is
presented by red rhombi and by red solid line, while blue colour presents the plastic
deformation energy. Global bending of the faceplate is marked by a red rectangular. Results for
the panels with 1 mm faceplates are presented in Figure 14. Big scatter of the test results may
be due to the dispersion of material properties as can be seen from Table 1.

Figure 14. Initial energy of the impact head and the plastic deformation energy in case of
the panels with 1 mm plating.

23
Local Impact Strength of Sandwich Panels Kristjan Tabri

Figure 15. Initial energy of the impact head and the plastic deformation energy in case of
the panels with 2 mm plating.

Dispersion of the test results is much smaller in case of the panels with 2 mm faceplates and
single tests shows good agreement with the trendline. Both energy levels are significantly
higher compared to the panels with 1 mm faceplates and also the global bending of the
faceplate occurs later.

Figure 16. Initial energy of the impact head and the plastic deformation energy in case the
panels with 2 mm plating and the urethane foam filling.

24
Local Impact Strength of Sandwich Panels Kristjan Tabri

When the sandwich panel with 2 mm faces is filled with urethane foam the global bending of
the faceplate is prevented and the energy level increases a little, see Figure 16. Shape of the
deformation is similar to one is case of the unfilled panels.

Figure 17. Initial energy of the impact head and the plastic deformation energy in case of
the panels with 3 mm plating.

Results of the tests made on panels with 3 mm plates also agree well with the trendline and the
dispersion of the results is small. In case of the panels with 3 mm plating, global bending of the
faceplate was not observed during the laboratory tests.

25
Local Impact Strength of Sandwich Panels Kristjan Tabri

3 Finite element analysis

Though the laboratory tests provide the verification data, they do not give any information
about the inner mechanics of the sandwich panel during the impact. Finite element simulations
allow to follow the impact process and to obtain the information about the behaviour of the
sandwich panel during the impact. Main purpose of the FE simulations is to verify the
assumptions made in the derivation of the analytical formulations.

For the FE analysis four different sandwich panels are modelled. For modelling and three-
dimensional meshing pre-processor LS-Ingrid is used. LS-Ingrid is also used as a translator to
convert a text file into input file for the finite element program LS-Dyna950d. The main
solution method in LS-Dyna bases on explicit time integration. Explicit solution method
exploits the idea that equilibrium equation is always satisfied. At the beginning of the time-step
every node has initial coordinate, velocity and force applied to the system. By the equilibrium
acceleration is found for every node. As the acceleration is known the new velocity and the
displacement of the node can be calculated by using kinematics. New equilibrium force is
calculated by the nodal displacements. Calculated values are used as new initial values for the
next calculation step.

3.1 Geometry of the FE model and the simulation procedure

Though the configuration of the panels is quite simple, it is still not reliable to model the whole
panel as the size of the model also affects the calculation time. Missing part of the panel can be
compensated by boundary conditions.

26
Local Impact Strength of Sandwich Panels Kristjan Tabri

Figure 18. Geometry and dimensions of the modelled panel.

Lets now consider Figure 18 to understand the use of boundary conditions. In order to
determine which parts of the sandwich panel should be modelled and where the boundary
conditions can be used, several simulations are carried out with different models. Calculations
give that using boundary conditions on sides AB and CD, which are transverse to the inner
supports, causes some overestimation of panel’s stiffness. To prevent the use of the boundary
conditions on transversal sides panel is modelled on its full length. Also the inner supports are
not fixed at the ends. Lower plate of the panel is not modelled as it does not contribute to the
energy absorption but only supports the inner members. As the supporting of the inner
members can easily be described by the fixed boundary conditions on lines E F and G H, the
modelling of the lower plate is unnecessary. Simulations also showed that the breath of the
modelled faceplate should be at least two times bigger than the span of the inner supports. Too
narrow faceplate causes some overestimation of the panel’s stiffness. Remaining part of the
panel is compensated by fixing edges A-C and B-D. Furthermore, it is assumed that laser welds
on lines E*F* and G*H* are rigid and do not deform during the impact. It means that the weld
is modelled just by connecting nodes of the faceplate and inner supporting member along the
lines E*F* and G*H*.

When the panel with the urethane filling is under consideration, compressible low-density foam
is modelled inside the panel. To reduce the calculation time only the middle section of the
panel is filled with the foam. As the foam in the other sections prevents the movements of the
inner supports, fixed boundary conditions are used on surfaces E E* F* F and G G* H* H.
Bottom of the foam is fixed to compensate the absence of the lower plate. Information about
the boundary conditions is gathered into Table 5.

27
Local Impact Strength of Sandwich Panels Kristjan Tabri

Table 5. Boundary conditions of the modelled panels.


STRUCTURAL
ELEMENT SIDE/SURFACE EMPTY PANEL FILLED PANEL

Faceplate A-B free free


C-D free free
A-C all dof.* fixed all dof. fixed
C-D all dof. fixed all dof. fixed

Support inner E-E' free free


G-G' free free
F-F' free free
H-H' free free
E-F all dof. fixed all dof. fixed
G-H all dof. fixed all dof. fixed

Urethane filling E-E'-F'-F - all dof. fixed


G-G'-H'-H - all dof. fixed
E-F-H-G - all dof. fixed
*dof.- degree of freedom

Density of the element mesh depends on the location. Near to the impact zone element
dimensions are the smallest- 1x1 mm. The biggest element dimensions are 4x4 mm. Figure 19
gives a better picture about the mesh and the element sizes. Steel plates are modelled by using
two-dimensional four node shell elements with thickness- known as Belytschko-Tsay elements.
This element type is one of the most commonly used elements in numerical analysis of crash
mechanics of thin-walled structures.

Figure 19. Element mesh.

28
Local Impact Strength of Sandwich Panels Kristjan Tabri

Element mesh and size of the urethane filling coincides with the mesh of the faceplate, see
Figure 20. Urethane filling is connected to the metal sheets by connecting the nodes, which
have the same coordinates. Urethane filling is modelled by using eight node hexahedron solid
elements.

Figure 20. Modelled panel with urethane filling.

To simulate an impact event, spherical impact head similar to one depicted in Figure 9 is
modelled. Impact head is modelled as a non-deformable rigid body. Energy of the striking body
is given by its mass and by the velocity at the moment of impact.

3.2 Material properties of the model

Steel plates of the sandwich panel are modelled by using LS-Dyna material model no 24
(Piecewise Linear Isotropic Plasticity). This material model is chosen as it works both in
elastic and in plastic region, capable to use non-linear material properties and can consider
strain-rate sensitive behaviour of the material. In elastic region material behaviour is
determined by Young’s modulus and by Poisson’s constant. Material behaviour in plastic
region is determined from the tensile tests. For a purely plastic response without fracture or
plastic localization, it is straightforward to determine the plastic parameters straight from the
tensile tests. Figure 21 presents the results of the tensile test and approximated true stress-strain
curve for LS-Dyna in case of the 3 mm specimens.

29
Local Impact Strength of Sandwich Panels Kristjan Tabri

Figure 21. Tensile test and approximated true stress-strain curve for LS-Dyna.

Though the tearing of the faceplate did not occur during the laboratory tests, the possibility of
the tearing is still foreseen in FE calculations. The initiation and propagation of fracture in the
structure can be modelled in LS-Dyna by deleting elements from the system once plastic strain
has reached a certain level. To determine that certain level, an equivalent fracture criterion for
the prevailing element is calculated. For the calculation a specimen is modelled and several
tensile tests with different failure criteria are carried out in LS-Dyna. The failure criterion is
evaluated by comparing the real and calculated stress-strain curves. When those two curves
coincide the correct failure criterion is found.

Urethane foam filling is modelled by using material no 14 (Soil and Crushable Foam with
Failure). That material model is selected as it provides a simple model for foams whose
properties are not well characterized. Necessary input variables for the selected material model
were given in Table 3.

3.3 Results of the FE analysis

Finite element simulations provide a possibility to obtain information that is hard to get from
laboratory experiments. Addition to deformation energies, the following characteristics are
determined by FE simulations:
(i) velocity profile,
(ii) velocity of the faceplate as a function of time,

30
Local Impact Strength of Sandwich Panels Kristjan Tabri

(iii) shape of the deformation,


(iv) displacements at the core.

Following discussion bases on simulation where the sandwich panel with 2 mm faceplates is hit
by the sphere with mass of 20 kg and velocity at the moment of contact is 3.13 m/s. Velocity
profile obtained by the finite element simulations is shown in Figure 22 by blue line and the
shape of the deformation by red line. Profiles in Figure 22 are drawn assuming that initial
contact between the impact body and the panel takes place at the origin.

Figure 22. Shape of the deformation and velocity profile.

Velocity profile is evaluated by analysing velocity time histories for every node between the
nodes FP-1 and FP-4, see Figure 24. Profile presents the average velocity values and is made
dimensionless by dividing it with the average velocity of the middle node FP-1. Figure 22
shows that the velocity profile can be approximated by linear line without a significant decrease
in preciseness.

Figure 23. Velocity of a node FP-1 as a function of time.


31
Local Impact Strength of Sandwich Panels Kristjan Tabri

Velocity as a function of time is presented in Figure 23. Red solid line presents the velocity of
the node FP-1 and red dashed line shows the calculated average velocity. It should be noted
that the same average velocity is used to turn velocity profile into dimensionless mode. Figure
23 also reveals that at the beginning of the impact node FP-1 obtains the same velocity with the
impact head. Velocity starts to decrease but the decrease is not exactly linear, but little
smoother at the beginning and slightly sharper at the end of the impact. Linear approximation is
presented by blue dashed line. Simple operation shows, that ratio between the initial and the
calculated average velocity is approximately 1.5. The same value is later used in analytical
calculations to describe the change of velocity.

Figure 24. Nodes at the cross-section of the panel.

It is obvious that most of the impact energy is absorbed by the faceplate, but the significance of
the steel core displacements should still be investigated. For that the transversal displacements
of the steel core are compared with the displacements of the faceplate. Comparison is done by
carrying out the impact simulation for the panel with 3 mm faces. Panel with 3 mm faceplates
is selected for the investigation as thicker faceplate causes greater displacements of the inner
supporters. Described panel is hit by the sphere with velocity of 3.13 m/s and mass of 30 kg.
Results are presented in Figure 25 and the nodes used in comparison were depicted in Figure
24.

32
Local Impact Strength of Sandwich Panels Kristjan Tabri

Figure 25. Displacements of the faceplate and core.

In Figure 25 red lines present the transversal displacements of the faceplate nodes and blue line
presents the displacements at the core multiplied by 100. Figure 25 shows that core
displacements are more than hundred times smaller compared to the displacements of the
faceplate and therefore can be ignored.

Initial and plastic energy as a function of permanent deflection are shown in Figure 26 for
panels with 1 and 3 mm plates. Figure 27 presents the results of FE simulations for the
sandwich panels with 2 mm faces.

(a) (b)

Figure 26. Results of FE simulations. Initial energy of the impact head and plastic
deformation energy as a function of permanent deflection in case of the empty panels
with 1 mm (a) and 3 mm (b) faceplates.

33
Local Impact Strength of Sandwich Panels Kristjan Tabri

(a) (b)

Figure 27. Results of FE simulations. Initial energy of the impact head and plastic
deformation energy as a function of permanent deflection in case of the empty panels
with 2 mm plating (a) and the urethane filled panels with 3 mm faceplates (b).

34
Local Impact Strength of Sandwich Panels Kristjan Tabri

4 Analytical formulations

Aim of the analytical formulations is to provide a possibility to calculate deformations in the


panel when the properties of the striking body are known. Extent of the deformation can be
evaluated by equalizing the kinetic energy of the impact body with the deformation energy of
the panel. As it is easy to calculate the kinetic energy of the striking body the main task is to
describe the energy absorption of the sandwich panel.

4.1 Background and main assumptions

As a result of impact, faceplate of the sandwich panel stretches in all possible in-plane
directions to resist impact loads and can attain large permanent deflections. When plate starts to
deform under lateral load, bending plays a major role for small deformations. With an increase
in transversal deformation, the importance of bending diminishes and the membrane force
quickly develops. At sufficient large deformations, the membrane force dominates the
behaviour. This is known as string response.

Furthermore, impact energy is absorbed not only by the faceplate, but also by the inner
supports, lower plate and by the filling if there is any. To consider all the deformation
mechanisms by analytical single model is complicated and even not necessary. The most of the
impact energy is absorbed by the mechanisms where it is done in most efficient way. To
simplify the model several assumptions should be made and verified.

One of the main assumptions is about the displacements of steel core. When inner supports of
the panel are much stiffer compared to the plates and there is no filling inside, most of the
energy is absorbed by the faceplate. Considering the dimensions of the tested panels and the
test matrix given in Figure 6, it becomes obvious that longitudinal bending stiffness of the I-

35
Local Impact Strength of Sandwich Panels Kristjan Tabri

profile supports is much higher compared to the faceplate. It introduces the first assumption-
inner structure of the tested panels can be considered as rigid and deformation energy is
absorbed only by the faceplate and by the filling. Convenient way to verify that assumption is
to measure core displacements in FE simulations. Measurements showed that displacements at
the core are more than hundreds of times smaller compared to the displacements at the
faceplate.

Second assumption considers the extent of the deformation. As laboratory experiments and FE
calculations have shown the maximum extent of the deformation is equal to the span of the
inner supports. The minimum extent is not limited and should be determined by minimizing the
energy. Furthermore, it is also assumed that the length of the panel is infinite. Assumption
agrees well with the actual use of sandwich panels where one dimension of the panel is often
much larger compared to the others. Importance of the mentioned assumption is that global
bending of the faceplate as can be seen in laboratory tests does not occur and the shape of the
deformation is assumed to be circular. In reality, some global bending of the faceplate occurs
also in the case of infinitely long panels, but the extent of the global bending is small compared
to the panel length. Global bending of the infinitely long panels reveals in deformation shape,
which takes more oval form.

Conclusively the main assumptions are:


(i) majority of the impact energy is absorbed by bending and membrane stresses at the
faceplate as deformations at inner supports and lower plating are small and can be
neglected,
(ii) the maximum width of the deformation is equal to the span of the inner supports,
(iii) length of the panel is infinite, which allows to use circular shape to describe
deformation.

Before proceeding to the derivation of energy absorption formulations, analytical description of


the deformation shape is given in Chapter 4.2. Formulations connected to the calculation of
strain rate are given in Chapter 4.3.

36
Local Impact Strength of Sandwich Panels Kristjan Tabri

4.2 Analytical description of the deformation shape

In order to be able to calculate the energy absorbed by the different deformation mechanisms,
shape of the deformation should be known. Circular shape of the deformation is described by
two coordinates. Coordinate r is pointed to radial direction and w to the direction of deflection.
The laboratory tests and the finite element simulations presented that it is convenient to divide
the deformation of the faceplate into two parts as shown in Figure 28:
(i) Linear line B-C
(ii) Curve A-B, which can be described by polynomial

Figure 28. Deformation shape.

Extent of the linear line is determined by two constants C1 and C2. C1 determines the extent of
the linear line in w direction and C2 is used to determine the extent of deformation in r-
direction. The linear part is

r
w C1 1 C2 R r R. (1)
R

Polynomial part is described by third-order polynomial given and is valid for 0 r C 2 R :

w( r ) a r3 b r2 c r d. (2)

37
Local Impact Strength of Sandwich Panels Kristjan Tabri

Constants a, b, c and d are determined by the following boundary conditions:

wr 0
d ,

w r 0
0 c 0,
(3)
wr C2 R C1 (1 C 2 ) ,

w r C2 R
C1 .
R

After evaluating constants a and b, shape of the deformation can be written as

a r3 b r2 if 0 r C2 R
w( r ) r (4)
C1 1 if C2 R r R,
R

where

C1 C 2 2 C1 2
a
C 23 R 3
(5)
2 C1 C 2 3 C 1 3
b .
C 22 R 2

Inclination of the deformation shape is determined by taking the first derivative of Eq. (4)

3 a r2 2 b r if 0 r C 2 R
S (r ) C1 (6)
if C2 R r R .
R

Later, when deriving equations for the energy absorption, also the change in inclination is
needed. It is evaluated by taking the second derivative of Eq. (4)

d (r) 6 a r 2 b if 0 r C 2 R
(7)
S

dr 0 if C2 R r R.

38
Local Impact Strength of Sandwich Panels Kristjan Tabri

To calculate the energy absorbed by the core filling, compressed volume of the filling material
should be evaluated by using expression

2
V r w dw . (8)
0

As it is laborious to derive the relation where radius r is given as a function of coordinate w by


using polynomial, two linear lines are used instead and the shape of the deformation can be
written as

w
1 R if 0 w 1 C 2 C1 R
C1
r ( w)
C2 R ( w) (9)
if 1 C 2 C1 R w .
1 1 C 2 C1 !

By substituting Eq. (9) into Eq. (8) and carrying out the integration, compressed volume of the
filling material can be calculated by

V R 2 C1 C 22 C1 C 22 .
3 (10)

4.3 Strain rate

As a result of the impact, the panel is deformed in relatively high velocity and possible effect of
the high strain rate to the material behaviour should be considered. The strain rate sensitivity of
the materials is considered by using Cowper-Symonds constitutive equation (Jones, 1989)
given by Eq. (11). Cowper-Symods model simply scales the static yield stress value Y by
considering strain rate and predetermined material constants q and D.

1
' $
% ".
q
Y 1 (11)
D
Y
% D "
& #

39
Local Impact Strength of Sandwich Panels Kristjan Tabri

To obtain the formulation for the strain rate, velocity profile over the cross-section of the panel
should be known. The FE simulations gave that it is sufficient to approximate the transversal
velocity profile by linear line, see Figure 29.

Figure 29. Approximated velocity profile.

Inclination of the velocity profile is

v
V . (12)
R

Consider cross sectional element of the faceplate (Figure 30) to derive formulations for the
strain rate. Non-deformed length of the element is dr. As the result of the impact the plate
deforms and obtains the deflection that can be calculated as S dr. The engineering strain in the
element is calculated from

2
dr 2 S dr dr (13)
r .
dr

By expanding Eq. (13) to series and neglecting high order terms, equation takes a following
form:

2
dr
S
. (14)
2 dr 2

40
Local Impact Strength of Sandwich Panels Kristjan Tabri

Figure 30. Deformed plate element

Strain rate is obtained by taking time derivative of Eq.(14). It should be remembered, that S dr
describes the deflection of the plate and the time derivative of S dr is the deformation velocity
of the plate. Deformation velocity at any point inside the deformed area can be calculated by
using the maximum velocity value at the point of first contact and the inclination of the velocity
profile. The dimensionless strain rate can be written as

2 dr dr
S V
. (15)
2 dr 2
S V

Note that S and V should be used as a dimensionless shape functions and the actual values
for the deflection and the velocity are given by the amplitudes and V0. V0 is the velocity of
the impact body at the beginning of the impact. Velocity time dependence is described by a
single constant cV, which is used to divide the initial velocity to get average velocity

v0
vA . (16)
cV

By substituting Eq. (12) into Eq. (15) the strain rate can be described as

S (r) v0
. (17)
cV R

In Eq. (17) denotes the final permanent deflection of the faceplate.

41
Local Impact Strength of Sandwich Panels Kristjan Tabri

4.4 Energy absorption of the panel

As stated above, it can be assumed that the steel core and the lower plate do not contribute to
the energy absorption and all the energy is turned into the deformation energy by the faceplate
and by the filling material. Formulation are derived both for the absorption of elastic energy as
well as for the plastic energy. Energy absorption in both cases is divided into two parts:
(i) energy absorbed by bending,
(ii) energy absorbed by membrane deformations.

In case of the filled panels also the energy absorbed by the filling is added to the plastic energy.

In analytical calculations it is assumed that the material behaves as elastic, perfectly plastic
material as given in Figure 31. Effect of the high strain rate is considered only in case of the
membrane mechanism.

Figure 31. Elastic, perfectly plastic material.

4.4.1 Elastic energy absorbed by bending

To derive the formulations for elastic energy absorbed by bending, it is assumed that
deformation has circular shape with radius R and deflection w at the middle of the panel
(Figure 32). It is obvious that the bending moment obtains its maximum value at the yield line
where r=R.

42
Local Impact Strength of Sandwich Panels Kristjan Tabri

Figure 32. Deformed panel.

Amount of the elastic energy is evaluated by equalizing the bending moment at the yield line
by plastic moment MP of the panel:

t2
MP Y
. (18)
4

Corresponding force or so-called collapse load FBE and deflection wBE in the middle of the
plate can be evaluated by using plate theory (Ikonen, 1990). Elastic energy can be calculated by
using relation

FBE wBE
E . (19)
2

Deflection of the circular plate subjected to a lateral distributed load can be written as

1 dr ' dr $
w C1 C 2 r 2 r ( r )dr " rdr , (20)
K r %& r #

where K describes the material and is equal to

E t3
K . (21)
12 (1 ( 2 )

43
Local Impact Strength of Sandwich Panels Kristjan Tabri

Distributed load is handled by introducing the concept of the effective radius rEF. Using the
effective radius distributed load can be used as a constant and relation between the force F and
distributed load becomes

F
2
. (22)
rEF

Using Eq. (22) first integral in Eq. (20) can be evaluated as

rEF 2
F rEF F
r ( r )dr rdr 2
rEF 2
. (23)
0
2 2 rEF 2

Remaining three integrals are evaluated as follows:

1 F F
dr ln( r ) , (24)
r2 2

F F 1 2 1 2
ln( r ) rdr r ln( r ) r , (25)
2 2 2 4

F 1 1 2 1 2 1F 2
r ln( r ) r dr r ln( r ) 1 . (26)
2 r 2 4 8

Deflection of the plate takes a form

1F 2
w( r ) C1 C 2 r r ln( r ) 1 . (27)
8

Constants C1 and C2 should be solved by using boundary conditions for clamped circular plate:

(i ) w( r ) r R
0,
d (28)
(ii ) w( r ) r R
0.
dr

44
Local Impact Strength of Sandwich Panels Kristjan Tabri

From the second boundary condition C2 can be solved and can be written as

1 r 2 2 ln( R ) 1
C2 F . (29)
16 K

By substituting Eq. (29) into the first boundary condition, C1 becomes

1 F
C1 R2. (30)
16 K

Final form for the deflection is defined as

1 F
w( r ) )R 2
2 ln( r ) 2 ln( R ) 1! r 2 *. (31)
16 K

Bending moment of the circular plates is given by

d 2w dw
M (r) K . (32)
dr 2 r dr

Derivatives in Eq. (32) are

dw 1F r
ln R ln( r )! , (33)
dr 4 K

d 2w 1F
ln( r ) 1 ln R ! . (34)
dr 2 4 K

Bending moment at the yield line is obtained by substituting Eq. (33) and (34) to Eq. (32)

1F
M (r) r . (35)
R
4

Now the collapse load FBE can be calculated as

45
Local Impact Strength of Sandwich Panels Kristjan Tabri

FBE 4 MP. (36)

Corresponding deflection at the middle of the plate (r=0) when subjected to load FBE is
obtained by replacing Eq. (36) to Eq. (31):

FBE
w BE R2 . (37)
16 K

Elastic energy absorbed by the bending can now be calculated by using Eq. (19) and is obtained
from
2
t4
R2 . (38)
E y
E B
32 K

4.4.2 Elastic energy absorbed by membrane mechanism

To calculate the amount of the elastic energy absorbed by membrane deformation, the same
idea is employed as in case of the elastic bending energy. Stresses in every element inside the
assumed deformed area are equalized by the yield stress of the material. In case of membrane
stress dynamic behaviour of the material plays important role in high strain rates. Due to that
the dynamic yield stress D
Y should be used. When stress in every point of the panel is known
the deformation energy can be obtained from

'D
E MP t Y dA . (39)
A

Assuming that Hooke’s law holds and the relation between the stress and the strain in elastic
region can be expressed as

Y E . (40)

Considering Hooke’s law in Eq. (39) the elastic energy absorbed by the membrane mechanism
is given by
46
Local Impact Strength of Sandwich Panels Kristjan Tabri

R
E 2 t D 2
E M Y r dr . (41)
E 0

4.4.3 Plastic energy absorbed by bending

Concept of plastic hinges is introduced to derive the formulations for plastic energy absorbed
by the bending mechanism. Figure 33 presents a rectangular plate with breath B and thickness t
subjected to a lateral load F.

Figure 33. Plastic hinge.

Due to the load, the panel is deformed and plastic hinge is formed at the point A. Deformation
energy absorbed in forming that plastic hinge can be evaluated from

E BP L MP , (42)

where is the angle and L is the length of plastic hinge. Deformation angle as a function of r is
given by Eq. (6). As work is done only in forming the plastic hinge, plastic energy can be
evaluated by using the change of the angle, given by Eq. (7). Absorbed energy is found by
integrating the change over the radius r

r2
d (r)
E P
B 2 MP S
r dr , (43)
r1
dr

47
Local Impact Strength of Sandwich Panels Kristjan Tabri

d (r)
where absolute value of S
should be used as the energy absorption does not depend on
dr
the direction of deformation angle. Derivation of Eq. (43) is convenient to carry out in two
parts:
(i) energy absorption when 0 < r < C2 R by using Eq. (43),
(ii) Energy absorption when C2 R < r < R by using Eq. (42).

When the first part is considered Eq. (43) takes a form

C2 R '

E P
B 2 MP 6 a r 2 b r dr (44)
r0

with constants a and b as given by Eq. (5). By carrying out the integration, Eq. (44) becomes

' 2 d3$
E BP 2 M P %(1 C1 ) , (45)
& 27 c 2 "#

where

c C1 C 2 2 2 C1
(46)
d 2 C1 C 2 3 3 C1 .

Second part of the energy absorption is obtained by using Eq. (42). The angle of the plastic
C1 C1
hinge is sin , but as C1 is small compared to R angle can be evaluated as
R R
without a deterioration in preciseness. Bending energy absorbed in region C2R < r < R’
becomes

(47)
E BP 2 C1 M P .

48
Local Impact Strength of Sandwich Panels Kristjan Tabri

As all the variables in Eq. (47) are always positive, there is no need to take absolute value. By
adding Eq. (45) and (47), the total plastic energy absorbed by the bending can be calculated
from

2 d3
E BP 2 MP 1 . (48)
27 c 2

4.4.4 Plastic energy absorbed by membrane mechanism

Plastic energy absorbed by the membrane mechanism is calculated similarly as presented in


Chapter 4.4.2, but engineering strain is used instead of the relation obtained by the Hooke’s
law. Consider again an deformed plate element in Figure 30. Due to the impact, element is
stretched and obtains the strain as given by Eq. (13) and (14):

2
dx 2 dx dx 2
S
+ S
.
dx 2

By substituting the strain into Eq. (39) the plastic energy absorbed by the membrane
mechanism is given by

(r) 2 R
E MP t D
Y (r) S
dA t D
Y (r) S ( r ) 2 r dr . (49)
A
2 0

4.4.5 Energy absorbed by core filling

Effect of the filling is considered by Winkler’s foundation, where support reaction caused by
the core filling can be written as

pF k w( r ) , (50)

where k describes the foundation.

49
Local Impact Strength of Sandwich Panels Kristjan Tabri

According to the Winkler’s model the energy absorbed by the core filling can be calculated as a
product of compressed volume V and the compressive strength F of the foundation

EF V F R 2 C1 C1 C 22 C 22 F . (51)
3

4.4.6 Approximate solution for membrane energy

Eq. (41) and (49) present the precise solution for energy absorbed by the membrane
deformations. As in Eq. (41) and (49) yield stress of the material is a complicated function of
coordinate, it is laborious or almost impossible to carry out the integration. Dependence of the
coordinate of yield function is given by multiplying the yield stress by

1
(r) q
(52)
CS ( r ) 1 .
D

By finding a new scaling constant CS* without a coordinate dependence material yield stress
becomes also independent from the coordinate and Eq. (41) and (49) can be integrated.
Consider Eq. (41) for the elastic energy absorbed by the membrane deformations. Assuming
that

D
Y (r) Y CS * (53)

integration can be evaluated as follows:

2
E* 2 t
R
D 2 2 t * 2
R
CS * t
E r dr CS r dr Y
R2.
M
E
Y
E
Y
E (54)
0 0

Integration in Eq. (49) is more complicated and should be carried out in two parts:

50
Local Impact Strength of Sandwich Panels Kristjan Tabri

R R
P* 2 *
E M t D
Y S (r) r dr t Y CS S ( r ) 2 r dr
0 0
(55)
'C2 R R
C1
2
$
t Y CS % ( 3 a r 2
*
2 b x) 2
r dr r dr ",
&% 0 C2 R
R #"

where
C1 C 2 2 C1 2
a 3 3
C 2 R

2 C1 C 2 3 C1 3
b 2
.
C 2 R2

After the integration and some simplifications absorbed plastic membrane energy is obtained
from

t CS * ' 2 11 $
E MP * Y
% C2 6 C2 C12 6 C2 1 C1 3" . (56)
5 & 5 #

Consider Figure 34 and Figure 35 to evaluate the new scaling constant CS*. Figure 34 presents
the effect of the strain rate. In Figure 34a red line presents the absorbed membrane energy
where the effect of the strain rate is considered. Also denotation E(CS) implies the dependence
of strain rate. Blue line is calculated by taking CS=1. In other words it means that the effect of
the strain rate is not considered. Dashed line shows the relation between the two energies.
Figure 34a shows that the ratio of two energies is almost constant and the strain rate effect can
be considered by using a single constant that depends on the initial velocity of impact body and
on the material properties. The same reveals by considering the Figure 34b where first
derivative of membrane energy or briefly energy rate is presented.

51
Local Impact Strength of Sandwich Panels Kristjan Tabri

(a)

(b)

Figure 34. a) Absorbed membrane energy; b) Rate of energy absorption.

Figure 34b reveals that roughly half of the energy is absorbed by that part of the panel where
the deformation is described by polynomial. In that part of the panel rate of the energy is
several times higher compared to the energy rate in linearly described part. Higher rate of
energy absorption is due to the higher scaling factor CS that depends on strain rate, see Figure
35.

52
Local Impact Strength of Sandwich Panels Kristjan Tabri

(a)

(b)

Figure 35. a) Strain rate; b) Scaling factor CS.

The new scaling constant CS* should consider both the polynomial and linear part. Considering
an plate element as depicted in Figure 36. Impact causes permanent deflection to the
faceplate.

Figure 36. Stretched plate element

53
Local Impact Strength of Sandwich Panels Kristjan Tabri

Strain rate in the element becomes

v
A 2
. (57)
L

Assuming that the velocity of the faceplate decelerates according to the constant cV, simplified
equation for the evenly distributed strain rate becomes

* v0
(58)
cV R2

and the energy can be scaled by a constant

1
q
v0 (59)
CS * 1 .
cV R 2 D

Now the approximate Eq. (54) and (56) can be used instead of complicated Eq. (41) and (49).

4.5 Solution procedure

The shape of the deformation and absorbed energy is determined by equalizing the initial
energy of the impact body with the deformation energy of the panel. Flow chart in Figure 38
presents the solution procedure. Flow chart reveals that only the constant C1 is found by the
minimization. To understand the reason for that consider Figure 39, which presents non-
dimensional shape of the total deformation energy as a function of constants C1 and C2. Figure
39a gives that the energy can be minimized respect to C1 as a local minima can be found.
Behaviour of C2 is different and minimization tries to use the lowest possible value. To obtain
the best solution constraints should be used. C1 can obtain any value between 0 and 1, but for
C2 a limit for the lower bound should be evaluated. Lower bound for the C2 is derived from the
geometry given in Figure 37 and can be expressed as

54
Local Impact Strength of Sandwich Panels Kristjan Tabri

1 C1 C1 2 rI
C2 + . (60)
R

At the beginning of the first calculation step initial guess values are given for C1 and RA.
Constant C2 can directly be obtained by using these guess values.

Figure 37. Theoretical shape of the deformation.

Figure 38. Flow chart of the solution process.

55
Local Impact Strength of Sandwich Panels Kristjan Tabri

The deformation energy of the panel as a function of deflection is equalized by the energy of
the impact body

E ME E BE E MP E BP EI , (61)

where EI is initial energy of the impact body defined as

m I v I2
EI (62)
2

and VI is the velocity of the impact body at the moment of the impact and MI is the mass of the
impact body.

(a)

(b)

Figure 39. Non-dimensional shape of the total deformation energy of the panel as a
function of constants C1 and C2.
56
Local Impact Strength of Sandwich Panels Kristjan Tabri

Permanent deformation is calculated numerically by using Eq. (61). Numerically calculated


deflection and Eq. (61) are now used in minimization process to calculate the constant C1.
Calculated and C1 are used as new initial values for the next calculation step. Calculation
loop continues until the values for and C1 do not change anymore and the equilibrium is
found. Calculated values and C1 provide sufficient information to determine the shape of the
deformation and to calculate the energy absorbed as the result of the impact.

57
Local Impact Strength of Sandwich Panels Kristjan Tabri

5 Comparison between the laboratory tests, FE calculations and the


analytical formulations

Comparison between the experiments, FE simulations and analytical formulations is presented


in Figure 40 to Figure 45. Analytical calculations are carried out both by using precise
formulations and also by using simplified expressions derived in Chapter 4.4.6. The results of
the precise formulations are presented by red dots while blue rhomb mark the result of the
simplified equations. Trendlines are drawn in corresponding colour by using second order
polynomial. For readability initial energy of the impact head and plastic deformation energy are
presented in separate figures. Constants and material properties used in analytical calculations
are presented in Table 6.

Table 6. Used constants and material properties for different panels.


Name Unit t=1 [mm] t=2 [mm] t=2 [mm], t=3 [mm]
urethane foam

E [GPa] 210 210 210 210


[-] 0.3 0.3 0.3 0.3

R [mm] 60 60 60 60

t [mm] 1 2 2 3

Y [MPa] 179 428 428 380

D 40.4 300000 300000 300000

q 5 6 6 6

cV* 1.5 1.5 1.5 1.5

[MPa] - - 0.62 -
F

* constant cV is obtained by the finite element calculations

58
Local Impact Strength of Sandwich Panels Kristjan Tabri

The results for the panels with 1 mm plating are shown in Figure 40 and in Figure 41.

Figure 40. Plastic deformation energy in case of the panels with 1 mm plates.

Figure 40 gives that the analytical formulations underestimate the plastic deformation energy
approximately by 10 %. Scatter may be explained by the fact that low yield steels are quite
sensitive to the strain rate effect and the small impreciseness in velocity or in strain rate
calculations may already produce a significant error. Good agreement can be seen between the
simplified and precise analytical equations, which indicates that it is sufficient to save
calculation time and to use simplified expressions.

Consider now Figure 41 for initial energy of the impact head as a function of permanent
deflection. Reveals that LS-Dyna seems to overestimate the energy in case of the high
deflection values. Reason for that may be too stiff boundary conditions that delay the global
bending of the panel, or imprecise material properties, especially yield stress. Analytical
calculations give poor results in low and very good results in high deformation values.
Impreciseness in low deformation values is due to the methodology of elastic energy
calculation. In analytical calculations, the amount of the elastic energy does not depend on the
deformation depth but only on the strain rate effect. If there is no strain-rate effect, amount of
the elastic energy remains constant.

59
Local Impact Strength of Sandwich Panels Kristjan Tabri

Figure 41. Initial energy of the impact head as a function of permanent deflection in case of
the panels with 1 mm plates.

Analytical model always calculates the elastic deformation energy for the predetermined
deformation shape and extent. Model assumes that material inside the predetermined area r=R
is stretched and bent until the end of the elastic region. In some cases, especially in the case of
very small deformations, the latter assumption may not be true as in some areas close to the
boundary, deformations may not reach to the end of the elastic region. According to described
calculation methodology the total deformation energy of the panel can never be zero, but has
some value even when depth of permanent deformation is zero. Such situation may occur when
energy of the impact head is small and the impact causes only elastic deformations. The same
behaviour can also be seen in case of finite element calculations.

Comparison of results obtained by the different methods in case of the panels with 2 mm
plating is presented in Figure 42 and Figure 43. In case of the plastic deformation energy some
scatter between the experimental data and the analytical calculations can be noticed in higher
deformation values. In higher deformation values also the global bending starts to affect the
energy absorption and as the analytical model cannot consider the effect of the global bending,
it may be the source of some impreciseness. Again the simplified analytical formula and
precise formula show very good agreement.

60
Local Impact Strength of Sandwich Panels Kristjan Tabri

Figure 42. Plastic deformation energy in case of the panels with 2 mm plates.

Agreement is not that good when initial energy of the impact head as a function of permanent
deflection is under consideration (Figure 43). Overestimation of total energy in lower
deflection values is obviously caused by the same effect as it is in case of panels with 1 mm
plating. Overestimation in high deformation is probably caused by the fact that analytical
formulations cannot consider the buckling of the faceplate. In case of a buckling, shape of the
deformation is changed to mode where energy is absorbed in more effective way. In brief it
means that oval shape should be used instead of circular shape.

Figure 43. Initial energy of the impact head as a function of permanent deflection in case of
the panels with 2 mm plates.
61
Local Impact Strength of Sandwich Panels Kristjan Tabri

The results for the urethane filled panels with 2 mm plating are given in Figure 44 and in
Figure 45. As the effect of the urethane filling is small compared to the effect of the faceplate
deformation, overall picture remains the same as it was in case of empty panels with 2 mm
faces.

Figure 44. Plastic deformation energy in case of the urethane filled panels with 2 mm
plates.

Figure 45. Initial energy of the impact head as a function of permanent deflection in case of
the urethane filled panels with 2 mm plates.
62
Local Impact Strength of Sandwich Panels Kristjan Tabri

Figure 46 and Figure 47 present the results for the sandwich panels with 3 mm plating. In case
of the plastic energy absorption good agreement can be seen between the results of analytical
calculations and the laboratory experiments.

Figure 46. Plastic deformation energy in case of the panels with 3 mm plates.

Figure 47. Initial energy of the impact head as a function of permanent deflection in case of
the panels with 3 mm plates.
63
Local Impact Strength of Sandwich Panels Kristjan Tabri

6 Conclusions

Comparison of the results obtained by the different methods reveals that the analytical model
describes the plastic energy absorption of the sandwich panel subjected to a lateral impact load
with sufficient preciseness. The biggest scatter between the laboratory test and analytical
calculations is in case of sandwich panels with 1 mm faceplates, when the difference is at worst
10 %. Difference may be due to the strain-rate sensitive behaviour of low steel material and
also due to the yield stress of the used material, which has some deviation from the mean value.
Non-constant material properties may have had impact also to laboratory experiments, where
the deviation of the results is biggest in case of the panels with 1 mm faces. In the case of
materials, which are very sensitive to the strain-rate effect too rough assumptions like imprecise
velocity profile or wrong deceleration constant cV may have serious impact on the results.
Thicker faceplates are made of materials that are not that sensitive to the strain-rate effect and
also the material properties are not changing in that big range. This reflects also in calculation
results, where the scatter is much smaller and the agreement is good in case of panels with 2 or
3 mm faceplates. In the case of the total deformation energy differences are little bigger,
especially in lower deformation values. Reason for that is the methodology of elastic energy
calculation, which always presents the whole elastic deformation energy consumed by the
predetermined polynomial-linear deformation shape with radii r=RA.

The usage of Winkler’s foundation seems to work well as the agreement between the results
obtained by different methods is good. Still it should be remembered that the effect of the
urethane filling is small compared the effect of the faceplate. Also the behaviour of the
urethane foam is relatively simple as the compression in one direction does not cause
significant stresses in other directions. Urethane filling also does not change the stiffness of the
sandwich panel significantly and the same assumptions that are used in case of the empty
panels are considered to hold.

64
Local Impact Strength of Sandwich Panels Kristjan Tabri

The comparison of the calculation results reveals that precise and simplified analytical
formulations give practically same results. In other word, it is possible to use some calculated
average value for the strain-rate and to assume that it does not depend on coordinate. Indeed, to
obtain that calculated value accurately FE calculations should be carried out. Without a
possibility of FE calculations also linear approximations will do, especially in case of materials
that are not very sensitive to the strain-rate effect. Simplified formulas help to save calculation
time as precise integration over the deformation shape takes significantly more time compared
to the simplified model.

Though the results obtained by the analytical calculations are satisfactory, there are still several
subjects that can be recommended for further research. One of the most important is the
possibility to consider global bending of the faceplate. Analytical model should be improved to
be able to use both oval and circular deformation shapes. The actual deformation shape can be
found by looking for the most effective way for energy absorption. Also in case of the panels
where stiffness of the core structure is not very high compared to the stiffness of the faceplate,
possibility of the core deformations should be added to the analytical model. As there are
several different mechanisms how the core structure may deform, the most probable and
effective one should be determined and described to the analytical model. For panels with thin
faces and stiff core structure also the tearing of the faceplate may occur. Problems considering
the fracturing and the tearing are complex as several parameters are affecting the initiation and
the propagation of fracture. One possible solution could be the determination of strains and
stresses in critical areas and once those have obtained some predetermined critical value tearing
occurs.

Furthermore, the formulations used in the current analytical model to calculate the elastic
energy absorption and the energy absorbed by the filling have several disadvantages and base
on rough estimations. In both cases the most effective way to prove the situation is better
description of the actual deformations and the inner mechanics and thus to reduce the effect of
the approximations. In the case of the elastic deformation energy actual extent and shape of the
deformation should be used instead of rough assumptions used presently. Radius of the actual
deformation and the extent of elastic zone may be several times smaller than the radius of
deformation assumed currently. The exact extent of the elastically deformed zone should be
determined by precise calculation of stress values in every point of a panel Similarly, the
formulations used currently to calculate the energy absorbed by the filling base on rough
65
Local Impact Strength of Sandwich Panels Kristjan Tabri

estimation and consider stresses only in one direction while the stresses in other directions are
ignored. That assumptions works relatively well with the urethane foam, but in case of several
other filling materials like balsa wood, stress situation becomes more complicated. More
advanced material models should be implemented to analytical model to consider more
sophisticated stress situations and orthotropic material properties.

Conclusively it can be said that despite the several disadvantages of the analytical model, thesis
fulfilled its goal. Basic understanding of the sandwich panel’s behaviour during the impact is
obtained and the analytical formulations describing it are derived. Also main ideas and usable
techniques to solve problems containing relatively large plastic deformations are investigated
and understood to some extent. One of the most complicated problems seems to be the
description of strain rate sensitive behaviour of materials. Mentioned behaviour can have
significant effect in case of impact loads, but is hard to describe precisely. Also accomplished
finite element simulations have introduced the main problems and critical areas that should be
considered in numerical modelling. Both, FE simulations and analytical calculations indicate
the importance of material properties. In case of impact loads also dynamic material properties
should be known in addition to static properties. That may cause some complicated problems,
as information about the dynamic properties is scarce and hard to obtain from literature. In the
other hand, dynamic tensile tests are expensive and hard to accomplish. Despite the problems
and complications, the study helps to understand complicated problems by using simple
structure and gives a good basis to continue with more complicated problems with more
parameters.

66
Local Impact Strength of Sandwich Panels Kristjan Tabri

References

1. Allen, H.G., Analysis and Design of Structural Sandwich Panels, Pergamon Press, 1969,
283 p.
2. Arve, S., Sandwich Constructions, Det Norske Veritas, Research Division, Technical
Report 1979, 140 p.
3. Chen, W.F and Han, D.J., Plasticity for Structural Engineers, Springler-Verlag, New York,
1987. 606p.
4. Gibson, L.J. and Ashby, M.F., Cellular Solids: Structure and Properties, Second edition,
Cambridge University Press, Cambridge, 1997, 510 p.
5. Goldsmith, W., Impact. The Theory and Physical Behaviour of Colliding Solids, California,
1959, 380 p.
6. Ikonen, K., Plate and Shell Theory, in Finnish, Otatieto Oy, 1990, 333 p.
7. Jones, N., Structural Impact, Cambridge University Press, 1989, 575 p.
8. Kolsters, H. and Zenkert, D., “Numerical and experimental validation of a stiffness model
for laser-welded sandwich panels with vertical webs and low density core”, Royal Institute
of Technology, Stockholm
9. Kreyszig, E., Advanced Engineering Mathematics, 8th edition, John Wiley & Sons, 1999,
1156 p.
10. Lorna, J.G., Ashby, M. F., Cellular Solids. Structure and properties, Cambridge University
Press, 1997, 510 p.
11. Naar, H., All Steel Corrugated Core Sandwich Panels Under Patch Loading, Otaniemi
1997, 54 p.
12. Naar H., Kujala, P., Simonsen, B.C. & Ludolphy, H., Comparison of the crashworthiness of
various bottom and side structures, Marine Structures, 2002, pp. 443-60.
13. Ohtsubo, H. and Wang, G., An Upper-bound Solution to the Problem of Plate Tearing,
Journal of Marine Science and Technology, 1995, pp. 46-51.
14. Paik, J.K. and Pedersen, P.T., Ultimate and Crushing Strength of Plated Structures, Journal
of Ship Research, 1995, pp. 250-261.
15. Pennala, E., Strength of Materials, in Finnish, Otatieto, Helsinki, 2000, 400 p.
16. Prager, W., An Introduction to Plasticity, Addison-Wesley, London, 1959, 148 p.
17. Romanoff, J. The Effect of a Filling Material to the Local Ultimate Strength of an All Steel
Sandwich Panel, Picaset Oy, 2000, 80 p.
67
Local Impact Strength of Sandwich Panels Kristjan Tabri

18. Romanoff, J. and Kujala, P., Formulations for the strength analyses of all steel sandwich
panels, Picaset Oy, 2002, 106 p.
19. Simonsen, B.C., Lauridsen, L.P., Energy Absorption and Ductile Failure in Metal Sheets
under Lateral Indentation by a Sphere, report No. 630, Technical University of Danmark,
Lyngby, 2000, 21 p.
20. Simonsen B.C., Mechanics of Ship Grounding, Technical University of Denmark, Lyngby,
1997.
21. Tärno, Ü., Plate and Shell Theory, in Estonian, TTÜ kirjastus, Tallinn, 2000, 137 p.
22. Wang, G., Some Recent Studies on Plastic Behaviour of Plates Subjected to Large Impact
Loads, Kournal of Ocean Mechanics and Arctic Engineering, 2002.
23. Wang, G., Arita, K. and Liu, D., Behaviour of a Double Hull in Variety of Stranding or
Collision Scenarious, Marine Structures, 2000, pp. 147-187.
24. Wang, G., Ohtsubo, H. and Arita, K., Large Deflection of a Rigid-plastic Circular Plate
Pressed by a Rigid Sphere, Journal of Applied Mechanics, 1998, pp. 533-535.
25. Ylinen, A., Theory of Elasticity and Strength of Materials, in Finnish I, Werner Söderström,
Helsinki, 1965, 476 p.
26. Zenkert, D., Sandwich Construction, Chamelon Press LTD, London, 1995
27. Zhang, S., Plate Tearing and Bottom Damage in Ship Grounding, Marine Structures, 2002,
pp. 101-117.
28. LS-DYNA_950d Theoretical Manual, Livermore Software Technology Corporation, 1998
29. http://www.pipingtech.com

68

Das könnte Ihnen auch gefallen