Tensile testing of 3D reinforced composites

(1)

Kungliga Tekniska H¨ ogskolan (KTH) Lightweight Structures

Tensile testing of 3D reinforced composites

Maite Gonz´ alez Eizaguirre

Supervisors:

Stefan Hallstr¨ om Fredrik Stig (PhD student)

Master of Science Thesis Stockholm, Sweden 2011

(2)

(3)

Preface

The work presented in this Final Degree Project was carried out at the Depart- mente of Lightweight Structures, Kungligla Tekniska H¨ogskolan (KTH). There- fore I would like to thank both universities, the University of Sevilla and KTH for having the learning agreement that allowed me spend these months working in Sweden as an exchange student.

I would also like to express my gratitude to my supervisors Dr. Stefan Hall- str¨om and Fredrik Stig for the guidance along the work carried out; and particu- larly to Fredrik, that helped me getting started in the lab, introduced me to the other workers at the department as ”the girl who would break things” and the

”Thursday’s fika”.

Furthermore, thanks to my ”office mates” for the discussions in the office and the time spent out the office getting to know Stockholm and Sweden.

Finally, I would also like to thank my parents for their support and because the education of my sister and I came always first for them.

Maite Gonz´alez Eizaguirre Stockholm in May 2011

iii

(4)

(5)

Abstract

3D-reinforced composite materials can be manufactured by 3D-weaving 3D- fabrics. A novel technique for 3D weaving has been used in order to do so.

With the purpose of analyzing the influence on the properties from the parameters involved in the weaving process, 20 specimens were manufactured at the Department of Lightweight Structures, Kungliga Tekniska H¨ogskolan (KTH), in colaboration with Biteam.

The resulting specimens had a limited length, square cross section and high expected ultimate loads; making gripping for tensile tests difficult. Therefore, a gripping solution was first developed. In spite of great efforts, the gripping solution did not succeed. Milled 3D-reinforced specimens were then tested without the use of gripping tabs.

The test results showed that, on the one hand, the ultimate loads were lower than expected. The reason seemed to be dry fibres resulting from the manufacturing process. On the other hand, the results from the tests that went well are very consistent. They show that the straighter the fibre yarns are in the main leading direction the better tensile properties of the material.

v

(6)

(7)

List of Symbols and Abbreviations

Abbreviation Description

D Dimension

E Modulus of elasticity

F Load

G Shear modulus

L Length

PC Polycarbonate

T Transverse

t Thickness

VARI Vacuum Assisted Resin Injection W or w Width

ν Poisson’s ratio ˆ

ε Fracture strain ˆ

σ Fracture stress τ Shear stress

Subscripts:

Abbreviation Description

1,2,3 Direction corresponding to ply material axes

c Compression property f Fibre property

m Matrix property s Shear property t Tensile property ult Ultimate

iii

(10)

iv LIST OF SYMBOLS AND ABBREVIATIONS

(11)

Chapter 1

Introduction

1.1 Background

Fibre reinforced composite materials are being increasingly used in structures in the aeronautics industry. For example, these materials are replacing metals as they can decrease the structural weight.

The reason for this is that composite materials are a macroscopic combination of two or more distinct materials, in this case fibres and matrix. The fibres im- part the mechanical properties in order to enhance the matrix properties. Thus, a structural component made of a fibre reinforced composite material can be designed by selecting the amount of fibres in every in-plane direction; and the type of fibres and matrix for each application. (See Figures 1.1aand 1.1b). Polyester, E-glass, S-glass, carbon and boron are some examples of fibre types, whereas the matrix could be Polyester, vinyl ester, epoxy or polypropylene.

However, a typical fibre reinforced composite material consists of multiple layers of fibre and resin plies bonded together to construct a laminate. In these plies, the fibres are mainly orientated in one direction or in perpendicular directions; and the desired properties of the laminate are gained by orientating each ply in the right direction.

The composite materials explained formerly, have high in-plane stiffness and strength. However, the out-of-plane properties are not better than that of the matrix material, which sometimes limits their use. 3D-woven fibre reinforcements contain through-thickness fibres which could provide delamination-free composites.

What are 3D reinforced composites?

3D reinforcement is characterised by fibres also in the out-of-plane direction, improving both strength and stiffness in that direction. If a laminate was defined

1

(12)

2 CHAPTER 1. INTRODUCTION

(a) Single ply (b) Laminae with its plies orientated in different directions

Figure 1.1: Description of composites.

by three orthogonal axes, these directions would be denoted by length (L), width (W) and transverse (T). Therefore, improving the out-of-plane properties would mean having fibres in the transverse direction.

There are different ways to refer to the 3D-fabrics in the composite mater- als’ industry. Since a way to carry out the 3D-weaving process was developed, the author in [1] proposes a new classification of fabrics. In the following, the terminology suggested by this author is going to be used.

There are three different manufacturing processes that can result in a 3D- fabric: 3D-weaving, 2D-weaving and non-interlacing.

In 3D-weaving, a bi-directional shedding operation takes place. A grid-like set of yarns in the longitudinal direction (warp yarns) is first sheded vertically and another set of yarns (horizontal weft) is picked in its width direction. The warp yarns are then subjected to horizontal shedding and a third set of yarns (vertical weft) is picked in the transverse direction. The result is a fully interlaced 3D-fabric. Figures 1.2ato1.2g describe the operations involved in this process.

In 2D-weaving, only a mono-directional shedding operation takes place. A set of yarns in the longitudinal direction is displaced in a crossed manner in the transverse direction (vertical shedding). Yarns in the width direction are then picked. The result of repeating this operation when a single layer of yarns in the longitudinal direction is used is a 2D-woven 2D-fabric. A 2D-woven 3D- fabric could be obtained if the warp yarns are arranged in several layers in the transverse direction.

In a noobed fabric (non-interlacing orientating orthogonally and binding), a grid of uniaxial yarns in the longitudinal direction are tied with fibre yarns in the width and transverse direction. The transverse yarns are interconnected outside the longitudinal yarns to form the bindings in the fabric’s surface. This type of fabrics contains straight yarns in three perpendicular directions but since there is no shedding it is not considered weaving.

(13)

1.2. OBJECTIVE 3

(a) Grid-like warp yarns

(b) Vertical shedding

(c) Horizontal picking

(d) Interlacing

(e) Horizontal shedding

(f) Vertical picking

(g) Interlacing

Figure 1.2: Description of the 3D weaving process. (Inspired by [2])

The fabric structures are illustrated in Figures 1.3c to 1.3a, courtesy of Fredrik Stig.

(a) 3D-woven 3D-fabric (b) 2D-woven 3D-fabric (c) Noobed fabric

Figure 1.3: Three different 3D textiles manufactured using the different processes 3D-weaving, 2D-weaving and noobing.

1.2 Objective

3D-weaving is a novel technology and it is still under development. It is therefore crucial to obtain reliable strength data of the material, and also to determine which factors influence the strength. To achieve the aim, a parameter study is performed to clarify the effects of:

(14)

• Cross section size: in 3D-weaving, a grid-liked set of warp yarns is used with variable number of rows and columns. (See Figure 1.4a)

• Surface structure: Both sets of wefts turn at the surface making the internal structure of the surface layer different from the structure inside the weave.

The objective is to investigate the influence of this layer. (See Figure1.4b)

• Crimp: The crimp measures the warp yarn’s undulation. In 3D-weaving, this happens in two directions since there is horizontal and vertical shedding. While the manufacturing process is taking place, the fabric is advanced forward by the take-up. A small take-up will result in higher crimp compared to a fabric advanced more frequently. (See Figure 1.4c)

• Stuffer yarns: Straight longitudinal yarns can be inserted in pockets generated in the 3D-fabric. Such yarns are expected to increase the longitudinal stiffness and possibly also the strength. (See Figure 1.4d)

(a) mxn grid-like yarns

(b) Outer-cell contri- bution

(c) Undulation of warp and weft

(d) Possible location for stuffer yarns (Re- produced from [2])

Figure 1.4: Involved factors in a 3D-woven 3D-fabric.

1.3 Scope of this project

Description of the test specimens

Composite specimens reinforced with preforms woven with varying weaving parameters were previously manufactured at the division of Lightweight structures, with the weaving technology of Biteam. All 20 specimens have square cross sections which is not ideal for gripping when subjected to tensile tests.

(15)

1.3. SCOPE OF THIS PROJECT 5

The baseline weave reinforcement was set to be an 8x8 warp yarn 3D weave with a medium size warp crimp and no stuffer yarns. Specimens with 4x4, 6x6 warp yarns were also manufactured together with 8x8 specimens with more and less crimp and with stuffer yarns. There were also 8x8 specimens manufactured using the noobing technique. All but the ones containing the stuffer yarns had the same fibre volume fraction (0.4) so that the test results could be compared in order to clarify the influence of the parameters on the performance of these materials. The specimens’ characteristics are enclosed in Table 1.1.

Clasification Length Size (mm) Expected ultimate

Size Crimp Type (mm) (Squared section) load (kN)^∗

8x8 high 200 10,5 115

medium 200 10 137,5

— Noobed 200 9,4 220

medium Stuffer 200 10,5 140

low 200 9,4 164

6x6 medium 200 7,2 71

— Noobed 200 7,2 73

4x4 medium 200 4,9 33

∗Explained in Subsection2.2.1

Table 1.1: Characteristics of the specimens

In order to be able to analyze the surface effects, some of the specimens were milled down to generate a narrower test section. The new geometry of these is outlined in Figure 1.5. The layer of warp yarns closest to the surface, which contained the turning weft yarns, where milled off, approximately 1,3 mm on each face. 50 mm by each end were saved to preserve the gripping area.

Figure 1.5: Geometry of the milled specimens

Description of the work

The final goal of the project was to perform tensile tests up to ultimate load.

However, in order to study the influence of the factors involved in the manufacturing process, a square cross section was considered the most appropriate.

The square cross section geometry however offered a relatively small surface to

(16)

transfer the load from the loading grips into the specimens, which was a problem.

In addition, the specimens had a limited length, and were very stiff and strong.

Therefore, this project was focused on the development of a suitable gripping solution, performing the test and analyzing the results.

(17)

Chapter 2

Test method development

2.1 Introduction

With the purpose of getting into the context of why it is difficult to test the existing specimens, an illustrative example is firstly presented. Specimens made of fibre reinforced composite materials that are subjected to tensile tests are usually manufactured with a rectangular cross section area. The tensile tests are performed following ASTM D 3039/ D 3038M: Standard Test Method for Tensile Properties of Polymer Matrix Composite Materials [3].

In [3] the use of tabs is strongly recommended not to damage the test specimens due to the gripping. In this way, the tensile stresses in the specimen are transfered as shear stresses by the adhesive that is located between the tabs and the specimen. If tabs are to be used, the geometry can be calculated according to Equations2.1 to2.3from [3].

F_ult(N ) = 2· τ_ult^adhesive· w_tab· L_tab (2.1)

w_tab= w_specimen (2.2)

L_min = L_gripping+ 2· width + L_gauge (2.3)

The 3D reinforced specimens do not have a rectangular cross section but a square one. Therefore, it is explained in the following what happens with the geometry requirements for the use of tabs when the thickness of these specimens is increased.

Hypothetical specimens made of a unidirectional lay-up are analyzed. These are designed to be stronger than 3D-woven specimens with 8x8 warp yarns so the estimated ultimate load will be 120 KN. In Figure2.1the effect in the width, tab length and total specimen length when the thickness is increased is presented.

7

(18)

8 CHAPTER 2. TEST METHOD DEVELOPMENT

For this example it has been considered that the ultimate load and the cross section area will remain constant.

On the one hand, when the thickness of the specimen is increased (x axes), the width will decrease, see blue line (solid line). This is understandable as the cross section area and ultimate load of a will remain constant. On the other hand, the tab length and the total specimen length will increase; see green and red lines (dash-doted and doted lines, respectively). The reason for these is that as the width is decreas- ing the tab length will have to increase so as to avoid the adhesive failure.

Figure 2.1: Influence of the thickness in the width, tab length and total specimen’s length.

What is interesting in this example is that if the thickness is only 1 mm, the specimen should be at least 234 mm, close to the 250 mm recommended in [3].

However, this length is already longer than the one the 3D woven specimens have.

Therefore, in order to be able to perform the final tensile tests with the 3D woven specimens, a successful way to grip them had to be developed. Gripping possibilities were firstly tested in 2D fibre reinforced composite specimens.

Gripping options that were generated by a first brainstorming are enclosed in Figures 2.2a to 2.2h. The purpose of these configurations was to introduce mechanical fastening (Figures 2.2a to 2.2e), load equally the four faces of the specimen (Figure 2.2f), decrease the expected ultimate load (Figure 2.2g) or modify the gripping geometry (Figure 2.2h).

2.2 Method

As already explained, the problem when testing the 3D specimens was the high ultimate loads combined with a square cross section and limited specimen length. Therefore, different gripping configurations were first studied analyti- cally. Some of the solutions were then tested using 2D fibre reinforced specimens manufactured for this purpose. This enabled to obtain the properties of the 2D specimens and, by comparing them to the theoretical ones, study the feasibility or how and improve the solutions.

2.2.1 Calculation of properties for the 3D-woven specimens

The properties of the 3D reinforced composites are not well known and laminate theory cannot be applied. However, in order to estimate the properties,

(19)

2.2. METHOD 9

(a) (b) (c) (d) (e) (f) (g) (h)

Figure 2.2: Gripping possibilities firstly considered. (a)Tabs manufactured from files (b) Gripping with pin bolts and Al tabs (c) ”Screw-assisted” griping (d) Gradual bolts(e)Gradual teeth(f)”V” gripping(g)”Bone” shaped samples(h) Casted edges

Finite Element Micromechanics studies for Stiffness and Strength of Wavy Fiber Composites can be consulted.

Assuming that the warp yarns were completely straight, the ultimate strength was first calculated applying micromechanics as outlined in Equations (2.1) and (2.2) reproduced from [4].

The properties of fibres (Catbon T-700) were aproximated by data from [4]

(Catbon T-300). The resin used was vinyl ester (DION 9500) and the data was obtained from the product’s data sheet. This data and the result from the micromechanical prediction for strength can be seen in Table2.1.

ˆ

σ_1t= ˆεE_fν_f + ˆεE_m(1 − ν_f) (2.4) ˆ

ε = min(ˆε_f, ˆε_m) (2.5)

Fibre characteristics Resin characteristics Material: Carbon T700 (12K) Material: Vinyl Ester

Properties: Properties:

Ef = 230GP a Em= 3100M P a ˆ

ε_f = 1.7% εˆ_m= 9%

νf = 0.45

Estimated strength σˆ1t= 2.2093· 10³M P a

Table 2.1: 3D-woven specimens’ characteristics and micromechanical prediction of strength.

(20)

As a result of the crimp, the strength will be lower. This effect is gained by applying knockdown factors calculated from [5], see Table2.2.

Crimp Knockdown factor

High 1,9

Medium 1,6

Small 1,34

Table 2.2: Knockdown factors according to the crimp.

On the other hand, for specimens containing stuffer yarns in the pockets, the effect of these is added as if they were straight. Regarding noobed specimens, it is considered that warp yarns remain straigth (knockdown factor = 1).

Finally, the ultimate loads presented in Table1.1 are calculated considering the specimens’ cross sections. (10, 7,2 and 4,9 mm width for 8x8, 6x6 and 4x4 warp yarn specimens, respectively.)

2.2.2 Developing 2D fibre reinforced specimens

In order to be able to simulate the behaviour of the 3D reinforced specimens and develop the different gripping possibilities, 2D reinforced laminas were manufactured using VARI (vacuum assisted resin injection). The same type of fibres and matrix as in the 3D specimens were used (Catbon T-700 and vinyl ester).

The properties of a lamina were calculated applying micromechanics following [4]. The results can be seen in Table2.3.

In order to predict the results of different lay-up sequences, for the failure load the Max Stress and Max Strain criteria were used. In addition, the Total degradation Tsai Model from [6] was implemented in which the stiffness of the failed laminas was reduced.

The results produced by this method were checked by comparing the findings of a real case. In [7], some samples of fibre reinforced composites built with the same materials were tested. The analytical results were compared with the ones obtained by the tests. It was predicted that the samples would fail at a higher ultimate load (23% higher) and strain than measured. The reasons for this could be that the matrix was supposed to have a linear behaviour and no imperfections or voids from the manufacturing process were considered. As a result of this, since 2D specimens were meant to behave similarly to the 3D ones- The ultimate load of the last ones was increased proportionally a 30%.

The last step was to come across with a suitable lay-up sequence for the 2D specimens that would give them similar properties and geometry as the 3D ones.

It was decided to reproduce the 8x8 warp yarn specimens since these would reach the highest ultimate loads (approximately 120 kN) and to follow two different

(21)

2.2. METHOD 11

Fibre characteristics Resin characteristics Lamina characteristics Volume faction: 0.4 t = 0.2625 mm Ef 11 = 220.63 GP a Em = 3.100 GP a E11= 111.87 GP a E_{f 22} = 13.790 GP a Gm = 1.1481 GP a E22= 6.1141 GP a G_{f 12}= 8.9632 GP a ν_m = 0.3500 G₁₂= 2.0356 GP a Gf 23= 4.8263 GP a σˆmt= 70 M P a ν12 = 0.2750 ν_{f 12} = 0.2000 σˆmc= 241.32 M P a^∗ σˆ1t = 2349.2 M P a ν_{f 23} = 0.2500 σˆ_ms= 89.632 M P a^∗ σˆ_1c = 1034.2 M P a ˆ

σ_{f t} = 2413.2 M P a εˆ_mt = 0.0900 σˆ_2t = 58.762 M P a ˆ

σ_{f c} = 2068.4 M P a εˆmc = 0.0778 σˆ2c = 202.57 M P a ˆ

ε_f = 0.0210 εˆ_ms = 0.0289 τˆ₁₂ = 73.446 M P a ˆ

ε_1t = 0.0210 ˆ

ε1c = 0.0092 ˆ

ε_2t = 0.0096 ˆ

ε_2c = 0.0331 ˆ

γ12 = 0.0361

∗Data used is Intermediate High strength resin data from [4]

Table 2.3: Micromechanics’ data and results for laminas for the 2D fibre reinforced composites.

sequences depending on the lamina orientations: Lay-up 1 had 0^◦and 90^◦ laminas ([0₆/90₂₇/0₆]); whereas Lay-up 2 had 0^◦ and ⁺₋45^◦ laminas ([0₇/⁺₋45₂₅/0₇]). The non-zero plies were primarily added to achieve the desired specimen’s height.

2.2.3 Gripping configurations and preformed tests

The data from Table 1.1 (Sample characteristics) correspond to the input parameters to design the gripping. Summarizing, the specimens have a length of 200mm and a square section which size depends on the type of specimen. Apart from these, there is an estimated ultimate load.

In the following, some gripping solutions are discussed making reference to ASTM D 3039/ D 3038M: Standard Test Method for Tensile Properties of Poly- mer Matrix Composite Materials [3].

Viable solutions were tested. These tests were performed up to a load in which there was a failure in the specimen, the bonding or the grips. Therefore, data such as maximum strain, load, load/area and elastic modulus could be obtained from them. These results were used to evaluate how good that solution was. The final goal was to test a gripping solution in which the specimen would break as intended so that the test would provide data to estimate the strength.

(22)

The tests were performed using an Instron 4505 test machine with a 100 KN load-cell. The strains were measured using strain gauges.

Plain specimens

Firstly, it has to be mentioned that the test methods recommended in the test standard [3] are for thin flat strips of material having a rectangular cross section which is already incompatible with the type of the specimens being treated here.

In these methods, the use of tabs is strongly recommended when testing unidirectional material or when there is risk for gripping damage. The material under study is not unidirectional; nor balanced and symmetric; nor random discontinu- ous. Therefore, tensile geometry recommendations mentioned are not applicable.

In order to check the influence of not having tabs, two specimens, one of each lay-up, were tested. These tests finished when the specimens failed since there was no possible debonding as no tabs were used. The results of the test show that the Young’s modulus is similar to the theoretical one whereas the ultimate load is at least 15% lower. The reason for this could be that where the grips end there is a stress concentration that could cause an early failure. Therefore, this is not likely to be a feasible gripping configuration.

Adhesive bonding tabs

Adhesive bonding 2 faces

If tabs are to be used, the geometry can be calculated according to Equations 2.1to2.3and input parameters. The results for the particular problem show that the minimum specimen length is greater than the one the specimens actually have, and this solution is therefore not applicable.

However, wedge action grips are used to perform the tests. For this reason, the adhesive is not only under shear but also under compressive stresses and the tests could be performed up to higher loads.

These results were tested. The tests results showed that the performance of the adhesive was improved, as it was expected. It was found that the shear strength of the adhesive (Araldite 2015) improved due to the compressive loads from 22 MPa to 29 MPa.

Adhesive bonding 4 faces

In the former configurations only 2 of the 4 faces where bonded. If all four faces of the specimen are used to bond the tabs, then, in theory, the gripping length could be halved. This calculation can be done considering a 25 or 50 mm

(23)

2.2. METHOD 13

gauge length. The results show that both could work for 4x4 warp yarn specimens since the minimum length required would be 174 and 199 mm, respectively.

How this gripping could be designed is illustrated in Figure 2.3. The aim would be to use a square section hollowed piece of an isotropic material such as aluminum and perform the test using wedge action grips. An isotropic, material is desired as it is stated in reference [8] that using these materials the stress concentration factor at the end of the tabs is reduced. The thickness of the material used should assure that it is able to stand the stress requirements with the overall thickness limit of 12 mm (maximum allowed by the wedge action grips available in the lab). The selected material should also be able to be gripped with the wedge action steel grips. If this happened, the 4x4 warp yarn specimens could be tested and a mechanical fixture for the 6x6 and 8x8 ones could be applied.

Figure 2.3: Geometry description bonding 4 faces of the specimen

On the other hand, there could still be other possibilities to bond four faces.

One of these is described in Figure 2.4. It consists of several tab layers stacked on to the specimen thickness in order not to exceed the adhesive’s ultimate load.

Nevertheless, this possibility only solves the problem for the 4x4 specimens since there is not enough length to bond the 4 faces of the 6x6 and 8x8 warp yarns specimens.

Figure 2.4: Possible configuration bonding four faces

Attending to the results presented formerly, if four faces of the specimen are bonded the 4x4 warp yarn specimens could be tested. Therefore, with the original

(24)

idea of stacking tabs, other configurations were also tested in order to make the manufacturing process easier. See Figures 2.5ato2.5c.

(a) (b) (c)

Figure 2.5: Bonding 4 faces of the specimen. (a) Stacking GFRP (glass fibre reinforced plastic) tabs (b)Foam (c) Wood

It was found out that the tests performed with GFRP failed due to delamination at loads between 24 and 28 kN. On the other hand, the foam and wood failed before the specimen, at 26 and 16 kN respectively.

Since delamination is not possible in the 3D reinforced specimens, this is an option that could have worked using glass laminate stackings. However, this solution requires a meticulous manufacturing process, so it was decided to explore other possibilities. (First the stackings are bonded. Afterwards, they are cut into a similar width to the one of the specimen. Once this is done, they are bonded to the specimen and leveled. Finally, the tabs are bonded.)

Mechanical bonding load transfer

A possibility for mechanical fixture is illustrated in Figure 2.6a, the pins would help transfer some of the shear loads. The Tresca and Von Mises criteria have been implemented in order to calculate the number of pins that would make this alternative possible. The materials considered are aluminum, mild steel and piano wires; having in this order an increasing ultimate strength.

The available specimen’s length is not enough in order to place the Al pins required. Similarly, if steel pins wanted to be used, as it can be seen in Figures 2.7aand 2.7b, the combination of pins of different sizes could make the specimen fail in an undesired way, due to the gripping.

Attending to these results it turns out that it only seems possible to use piano wires. The number of piano wire pins required and the distance between pins a (see Figure 2.6b) are presented in Table 2.4. The results came from the Tresca criteria, being the most restrictive one.

Although mechanical fastening using aluminum pins is not a possible solution, some tests were performed. The objective was to see in which proportion the performance of the tests was improved compared with tests in which just adhesive was used for bonding.

Two aluminum pin configurations were tested. A specimen had 7 pins on each edge (4+3) whereas a second one had 3 (2+1). The maximum loads were 32,8

(25)

2.2. METHOD 15

(a) (b)

Figure 2.6: Geometry of the mechanical bonding. (a) Failure model geometry(b)Distance between pins

T resca : F

2· n· A ≥ σY

2 (2.6) V onM ises :

r3 4· F

n· A ≥ σ_Y (2.7)

and 32,2 kN respectively. This represents that when using few pins, increasing the number of these does not increase the maximum load in the same proportion;

and 120 kN (the worse case for 8x8 specimens) is not close to be reached by this solution.

(a) 8x8 specimen (b) 6x6 specimen

Figure 2.7: Steel pin configurations in the available gripping length.

”V” grips

In fibre reinforced composite specimens that are thin strips of material, only the widest faces are used to bond the tabs. When the specimens have a square

(26)

Diameter Number of pins

a^∗ (mm)

a^† (mm)

1 70 <0 0,11

2 18 1,53 2,18

3 8 4,56 5,94

4 5 7,50 9,58

5 3 12,50 15,63

6 2 17,67 21,83

7 2 17,00 21,17

8 2 16,33 20,50

9 1 28,00 34,25

(a) 8x8 specimens

Diameter Number of pins

a^∗ (mm)

a^† (mm)

1 40 0,68 0,98

2 10 4,35 5,48

3 5 8,80 10,88

4 3 13,95 17,08

5 2 19,27 23,43

6 2 18,60 22,77

7 1 30,40 36,65

8 1 29,90 36,15

9 1 29,40 35,65

(b) 6x6 specimens

∗L_gauge=50mm ^†L_gauge=25mm

Table 2.4: Number of piano wire ”pins” (Ultimate strength 2200 MPa).

cross section, all four faces can be used to distribute the load transfer. This is the purpose for turning the specimen 45^◦ around the longitudinal axis as shown in Figure 2.2f.

Polycarbonate ”V” grips with attached glass tabs

Side slide grips in combination with polycarbonate (PC) ”V” grips were firstly tested. The configuration of the PC ”V” grips can be observed in Figures 2.8b and 2.8c. The glass fibre reinforced tabs are screwed to the PC blocks and bonded to the 2D specimen using Araldite 2015. Figure 2.8ashows the side slide grips used to perform the tensile test.

The results were not satisfactory because the side slide grips were unable to hold the PC blocks without avoiding sliding. In fact, the test was aborted when the load had only reached 5kN (less than 0.1% strain). A picture of the damage in the PC blocks due to the sliding shown in Figure 2.8d .

Modified polycarbonate ”V” grips with attached glass tabs

In order to improve the performance of the PC grips and keep working with the same specimen, the shape of the PC block was modified to replace the side slide grips for wedge action grips. The set-up is presented in Figure 2.9a.

In this case, the tensile test reached a load of 15 kN and 0.5% of strain. The test was stopped at that load because one of the PC grips broke. An upper view

(27)

2.2. METHOD 17

(a) (b)

(c)

(d)

Figure 2.8: Polycarbonate ”V” grips with attached glass tabs. (a)Specimen in the side slide grips ready for the tensile test (b)Specimen with the PC grips(c) Upper view of the ”V” grips (d)Damage in the PC blocks

(a)

(b)

(c)

(d)

Figure 2.9: Modified polycarbonate ”V” grips with attached glass tabs. (a) Milled specimen in the wedge action grips ready for the tensile test(b)Resulting crack in the grip(c)Damage produced during the milling(d)Inexact milled angle in the broken grip before the test

of the longitudinal crack in the grip is enclosed in Figure 2.9b.

Figures 2.9c and 2.9d show two possible causes of the failure. The first one regards the milling and its result is a smaller contact surface. In the second one

(28)

the difference between the milled angle and the one of the grip can be observed.

Both could have led to an unequal distribution of the load that could have caused the crack growth.

Epoxy casted edges

Epoxy casted edges (Vertically casting mould)

Since the possibility of combining the mechanical interlocking with the PC grips was too complicated in order to have an acceptable manufacturing accuracy, the next possibility to be considered was casting the edges of the specimens with the right shape so that they could fit in the wedge clamps. With the purpose of manufacturing a mould, the geometry of the clamps was first studied. The idea was to have as much specimen embedded as it was possible so as to have the largest bonding surface. Therefore, the free length of the specimen was calculated to be able to place a 25 mm strain gauge when performing the tests. Sketches of this and the resulting mould with the ”zero specimen” are enclosed in Figures 2.10a and 2.10b. More information about how to manufacture a silicone mould can be read in Appendix A.

(a) (b)

Figure 2.10: Casting the specimens edges. (a) Sketch of the casted specimen mounted in the test-rig(b)Manufactured silicone mould with the ”zero specimen”

The manufacturing of the casted edges resulted cumbersome. Firstly, it was vertically casted one edge, while the remaining specimen protruded through the base. The specimen was adapted to the cavity’s size by adding plastic layers to its faces. Leaks at this stage were unavoidable and as a result of these, the remaining specimen’s surface was compromised. Once this stage was completed, the second edge was casted following the same procedure. This time apart from the leaks

(29)

2.2. METHOD 19

(the casted edge had to be protected in order not to modify its surface), since the specimen was turned upside down, the first casted edge’s weight made the specimen slide slightly adding more imperfections to the manufacturing process.

Only one specimen casted vertically could be tested. The test was performed up to a load of 10 KN. By the end of the test, two cracks along the secondly casted edge were visible. This early failure could be the result of the misalignment of the specimen in one of the directions. Figures 2.11b and 2.11c show the crack in the secondly casted edge.

(a)

(b)

(c)

Figure 2.11: Vertically casted epoxy edges. (a)Fist specimen to be tested with two casted edges (b)and (c) Resulting cracks in the secondly casted edge

Epoxy casted edges (Horizontally casting mould)

After experimenting with the two-parts silicone mould, an horizontally casting mould was designed. This improved the manufacturing process by reducing time, since both of the specimen’s edges could be casted at the same time and in case of leaks, these happened more slowly as the casted height was smaller and therefore also the hydrostatic pressure.

In the new mould specimens of different sizes would have to be tested (see

(30)

Figure 2.12). In order to be able to do this, the ”narrow pass” would be adjustable. Since it was seen that the inner angles were very sensible to the way the vertical mould was held, it was a good idea to decrease the amount of silicon close to the adjustable part. What the geometry would look like for the different specimen sizes is illustrated in Figure 2.13.

1. Casting cavity 2. Open part

3. Narrow pass adjustable to specimen sizes

4. Foam ensuring the specien is straight

5. Glass laminate that closes the casting cavity

Figure 2.12: Horizontally casting mould design.

(a) (b)

Figure 2.13: Adaptation of the narrow pass to the different specimens. (a)Small specimens(b) Large specimens

Parameters such as the angle or the casted height were still determined by the wedge action grips. 4x4 warp yarns specimens were used to dimension the mould. This time, a free length of 35 mm was considered in order to take strain measurements. Since the wedge angle is the same for all casted specimens, the bigger the specimen’s size, the larger would the free length be. (See Figures 2.14a and 2.14b). This is a positive point of this design as it is recommended to have a free length that is twice the width plus the strain length. Therefore, the bigger the width, the larger the free length should be. This can also be a disadvantage as it also means that the larger the free length is, the smaller the embedded part will be; and large samples have more warp yarns and higher expected ultimate load.

(31)

2.2. METHOD 21

(a) (b)

Figure 2.14: Definition of the geometry according to the test machine. (a) Configuration for 4x4 specimens(b) Configuration for 8x8 specimens

By changing the manufacturing process, the failure load in the tensile tests was increased from 10 to 28 kN, for sample 1033. The results of the first test suggested that having the sample turned 45^◦, could have a negative effect when there were compressive loads. Therefore, a non turned specimen was tested.

After the manufacturing process it was seen that air voids were captured under the specimen. This test failed at a lower ultimate load, 20 kN, and due to debonding. As the results of both tests showed that the casted edges failed due to debonding, new solutions in order to improve the specimen’s surface were studied.

In order to improve the bonding, three possibilities modifiying the specimens’

surfaces were tested. These can be seen in Figures 2.15ato2.15c.

(a) (b) (c)

Figure 2.15: Surface bonding improvement. (a) Horizontal groves (b) Diagonal groves (c)Pins

These test showed that the bonding was not improved by making groves in the specimen’s surface. Furthermore, specimen 1063 failed inside the epoxy edge and specimen 2035 failed due to debonding in the block too. In both cases, the groves drawn in the surface weakened the specimen. On the other hand, the test performed with the specimen with pins failed when the last pins yielded. Which showed that with a larger number of pins, a higher load could have been reached.

(32)

Specimen Max. Load [KN]

1033 28

1061^∗ 20,4

∗Not turned specimen

Table 2.5: Test results for the epoxy casted edges (Horizontally casting mould).

Specimen Max. Load [KN]

1063^∗ 18,23

2035^† 15,28

1052^‡ 48,24

∗Horizontal groves

†Diagonal groves

‡Mechanical bounding with pins

Table 2.6: Test results for the epoxy casted edges with bonding improvements (Horizontally casting mould).

Further tests tried to clarify the best way to reach the highest mechanical load transfer. Other pin configurations, fibre rods and randomly reinforced epoxy were tested. However these attempts lead to a specimen failure; sometimes due to delamination or because the specimen failed in a section in which there was a pin (or fibre rod).

2.3 Discussion

Figure 2.16gathers together the maximum loads reached with the different gripping solutions studied. The variability of the loads is due to the lack of similarity of all the possible solutions.

It can be seen that the highest loads were reached when the specimens were tested without tabs (”plain specimens”). However, there was premature failure so this was not an acceptable solution.

This figure also shows the potential of casting the specimen’s edges. The debonding failure occurred at a very similar load to the one in adhesive bonding (28KN). However, the possibility of including mechanical load transfer almost doubles the failure load.

Some of these tests also showed specimen failure that would not happen with the 3D specimens. Delamination occurred in the case of bonding 4 faces or casting a specimen with diagonal groves trying to avoid debonding. On the other hand, there was also found out specimen’s failure due to the stacking sequence. Since the inner layers in Lay-up 1 where orientated at 90^◦. These failures are illustrated in Figure 2.17.

2.4 Conclusions from trials

While other tests failed without reaching the desired test maximum loads, casting the edges in epoxy seems to be the best solution. Due to the versatility

(33)

2.4. CONCLUSIONS FROM TRIALS 23

Figure 2.16: Maximum loads reached with the different gripping methods

(a) (b) (c)

Figure 2.17: Examples of specimen’s failure during the gripping study. (a) De- lamination when bonding 4 faces (b)Delamination when casting a specien with diagonal groves(c) Specimen failure

of this method, the load can be transfered mechanically without compromising the manufacturing accuracy, as could happen with the PC grips.

On the one hand, using the 2D specimens, which could fail due to delamination, it could not be proved that following this method the tests will be able to be performed successfully.

On the other hand, the findings from the tests with casted edges justified testing the 3D specimens.

(34)

(35)

Chapter 3

Experimental study

The gripping solution that was deemed most successful in the previous chapter was adopted for the 3D-woven specimens. In the following sections the test specimens preparations, their configuration, and the results are described.

3.1 Test specimen preparation

Moulded specimen preparation

In order to perform the tensile tests and as a result of the gripping study, it was decided to cast the specimens edges in epoxy and perform the tests with wedge action grips. The test configuration is shown in Figure 3.1.

The gripping study also showed that the load transfer to the specimen was not possible to be done only by adhesion (epoxy-embedded specimen). Therefore, it was aided mechanically by the use of pins and ”fibre rods” or just pins. These went through the embedded part of the specimen in the transverse and width direction. Examples of these are shown in Figures 3.2a and 3.2b.

The following operations were completed so as to prepare a test specimen.

1. Holes with a separation of 7mm in the embedded part of the specimen were drilled. As it can be seen in Figures 3.2a and 3.2b, these were drilled in alternating perpendicular directions.

2. The pins were then fixed using Araldite 2015. In the case of using fibre rods, these were wet in Epoxy.

3. Before placing the test specimen in the casting mould, casting-cavity closing surfaces were fixed to the specimen. These were cut with the right size (according to each specimen) and with a 90^◦ angle so as to ensure the specimen’s orientation. Figure 3.3 shows the result of these operations.

4. The epoxy (Renlam LY 113, 30% Ren HY 97-1) was then casted.

5. After removing it from the mould, the specimen was post-cured overnight at 80^◦C.

25

(36)

26 CHAPTER 3. EXPERIMENTAL STUDY

Figure 3.1: Set up for the tensile tests

(a) (b)

Figure 3.2: Examples of mechanical aiding. (a)Pins and fibres(b)Pins.

Figure 3.3: Test specimen in the silicone casting mould. (How to manufacture a silicone mould is described in Appendix A)

Plain specimen’s preparation

Since all of the specimens were not tested at the same time, the test results helped to improve future tests. The first tests showed that the strength was lower than expected and since there were bonding problems, it was decided to test the six other milled specimens without the use of tabs, plain, in the wedge-action grips.

(37)

3.2. TESTS SPECIMENS 27

3.2 Tests specimens

In the following, the tested specimens are described.

Moulded specimens

Tensile tests were performed using an Instron 4505 test machine with a 100 kN load-cell. The strains were measured using a 25 mm strain gauge.

These tests were performed up to a load in which there was either a specimen or a gripping failure.

The first specimens to be tested were the ones that had been previously milled.

There were two specimens of each type and two casting moulds; so it was decided to cast one of each type at a time so that in case there was any problem when testing, there could still be another specimen of the same type that could be tested.

Table3.1provides a description of the tested specimens.

(a) Specimen 1

(b) Specimen 2

(c) Specimen 3

(d) Specimen 4

(e) Specimen 5

(f) Specimen 6

Figure 3.4: Mechanical loading configuration of the 3D reinforced casted specimens.

(38)

Specimen 1 Technical name: D 6x6 -1-R-TU6 (11)

Description: 6x6 warp yarns milled specimen

Geometry: L = 192 mm

Milled middle section: 4,76 x 4,79 mm Unmilled edges section: 7,2 x 7,2 mm

Expected ult. load: 27,5 kN

Mechanical loading configuration: Pins and fibres. (See Figure 3.4a) Specimen 2 Technical name: D 8x8 -S-2-C (10)

It also contained stuffer yarns

Expected ult. load: 71 kN

Mechanical loading configuration: Pins and fibres. (See Figure 3.4b) Specimen 3 Technical name: D 6x6 -2-R-TU6 (12)

Expected ult. load: 27,5 kN

Mechanical loading configuration: Pins. (See Figure 3.4c) Specimen 4 Technical name: D 8x8 -8-C-TU10 (5)

Mechanical loading configuration: Pins. (See Figure 3.4d) Specimen 5 Technical name: D 8x8 -4-C-TU6 (3)

Milled middle section: 7,39 x 7,56 mm Unmilled edges section: 10 x 10 mm

Mechanical loading configuration: Pins and rough surface^∗. (See Figure 3.4e) Specimen 6 Technical name: D 8x8 -N1-C-TU6 (7)

Expected ult. load: 98 kKN

Mechanical loading configuration: Pins and rough surface^∗. (See Figure 3.4f)

∗These specimens had surface debonding improvements made by file dentation Table 3.1: Specimen characteristics.

(39)

3.2. TESTS SPECIMENS 29

Plain specimens

Specimen 7 Technical name: D 8x8 -9-C-TU10 (6)

Description: 8x8 warp yarns milled specimen Geometry: Milled middle section: 6,59 x 6,72 mm

Unmilled edges section: 9,4 x 9,4 mm Expected ult. load: 82 kN

Unmilled edges section: 10 x 10 mm Expected ult. load: 89 kN

Specimen 11 Technical name: D 8x8 -N2-C-TU6 (8)

Specimen 12 Technical name: D 8x8 -S2-C (9)

Description: 8x8 warp yarns milled specimen It also contained stuffer yarns

Geometry: Milled middle section: 8,06 x 8,16 mm Unmilled edges section: 10,5 x 10,5 mm Expected ult. load: 72 kN

Table 3.2: Non-molded specimen characteristics.

(40)

3.3 Results

The results from the tensile tests are presented in the following.

Molded specimens

Expected ultimate load (KN)

Max test load (KN)

Failure cause

Specimen 1 27,5 14 Specimen failure at a section that had been milled. Inside one of the embedded edges.

Specimen 2 71 29 Failure at load introduction.

The pins yielded.

Specimen 3 27,5 13,4 Specimen failure at a section that had been milled. Inside one of the embedded edges.

The pins yielded and the specimen at one of the external holes.

Specimen 6 98 20,2 Failure at load introduction.

The small pins yielded and tore the specimen.

Table 3.3: Test results for the 3D reinforced specimens with moulded edges

(a) Specimen 1 (b) Specimen 2 (c) Specimen 6

Figure 3.5: Undesired specimens failure.

(41)

3.3. RESULTS 31

Figure 3.6: Specimens that failed due to pin yielding.

Plain specimens

Expected ultimate load (KN)

Max test load (KN)

Failure cause

Specimen 7 82 40,3 Specimen failure within the

gauge length.

Specimen 10 69 38,2 Specimen failure within the gauge length.

Specimen 11 98 65 Specimen failure within the

gauge length.

Specimen 12 72 66,3 Specimen failure within the gauge length.

Table 3.4: Test results for the 3D reinforced specimens tested without being moulded

(42)

(d) Specimen 10 (e) Specimen 11 (f) Specimen 12

Figure 3.7: Specimens without end tabs showing intended failure.

(43)

Chapter 4

Discussion

Six 3D reinforced specimens were tested with the gripping solution and none of the tests were able to provide stiffness nor strength data.

The gripping solution used to perform the tests was selected because once the bonding was mechanically improved by the use of pins, the maximum load that could be reached during the tests seemed to be better than the ones reached with the other possibilities.

However, instead of finding specimen’s failure by the end of the tests there was a gripping failure. The performed tests could be grouped in three different failure types:

• Cause A: Early specimen failure. This is the case of specimens 1 and 3.

Both failed in an embedded section close to where the milling started. The reason for the failure could be stress concentrations due to the specimen’s milled geometry and the load case. (See Figure 4.1awhere the load case is drawn and the possible grip re-design solution tested in specimens 5 and 6).

The failure loads could be considered representative. However, as a result of the tests, no damage in the surface was observed as could happen with specimens 7 to 12.

• Cause B: Pin yielding. A distributed pin arrangement was applied in the different tested specimens, see Figure 4.1b. However, since the superficial load transfer was stopped early due to debonding, the pins had to carry all the load. This is the reason why the pins could have failed earlier than expected.

• Cause C: Specimen failure due to the mechanical load introduction. Pins tore apart specimens 4 and 6 leaving a visual damage. In both cases, there was also pin yielding (failure Cause B).

33

(44)

34 CHAPTER 4. DISCUSSION

(a) Load case and possible solution

(b) Applied pin density distribution configurations

(c) Tore specimens

Figure 4.1: Gripping failing causes explanations

On the other hand, six more 3D reinforced specimens were tested without the use of tabs in the wedge action grips. In this tests, it was the specimen that failed within the gauge length providing thus reliable data. No conclusions can be withdrawn regarding the number of warp yarn or the outer cell structure since all tested specimens contained 8x8 warp yarns and were milled. However, other results can be analyzed:

1. The ultimate loads seemed to be smaller than the expected ones. This could be due to manufacturing problems since the specimens seemed to have partly dry fibres.

2. As expected, the more crimp the less stiffness. The undulation of the yarns going in the main direction tend to soften the material and allow for greater displacements. For test specimens 8 and 9, the tests were stopped due to the method’s software limits since the maximum displacement was reached.

This is the reason why there are no data regarding the maximum ultimate load for these two specimens.

3. Also related to the crimp, the smaller it is the higher the ultimate loads, as show the results for specimens 7 and 10.

4. Test specimens 11 and 12 reached the highest ultimate loads. The fibres in specimen 11 were noobed, which means that the yarns were straight and this benefits the performance. Specimen 12 contained stuffer yarns. This is translated to more fibre yarns in the main direction that explain the results.

(45)

Chapter 5

Conclusions

The purpose of this project was to perform tensile tests and obtain data of the stiffness and strength of 3D reinforced composites.

3D-woven fibre reinforced composite materials constitute a new class of material, since 3D weaving is a novel technique. Therefore, there are efforts on characterizing them.

The test specimens manufactured at KTH had geometry constrains defined by factors involved in the weaving process (number of yarns, take-up and possibility to contain stuffer yarns). This resulted in square section high performance specimens.

It was required to develop a successful gripping solution that would enable testing up to the ultimate load and ensure that the specimen failure would not be related to the gripping.

A gripping study was carried out. The different possibilities were tested on 2D reinforced specimens manufactured for that purpose. The results showed that the most promising solution was casting the specimens edges in epoxy and performing the tests with wedge action grips. As a particular difference with rectangular cross-section specimens, these ones were tested loading equally all four faces by holding the specimen turned 45^◦ around the longitudinal axis. In order to reach higher loads, it was proved that the load transfer had to be aided mechanically.

The use of pins and fibre bundles were tested in order to study the element that would achieve the best load transfer to the specimen. However, at this point 2D reinforced specimens were failing at lower loads that the expected ones. It was due to delamination, a cause that would no happen with the real test specimens, the 3D reinforced ones.

For this reason, it was decided to start performing tensile tests with the 3D reinforced specimens. The gripping problem would be solved by casting the specimens’ edges in epoxy and aiding the load transfer by the use of pins. The

35

(46)

36 CHAPTER 5. CONCLUSIONS

pin configuration would be designed according to each specimens geometry and with a pin density distribution, higher close to the specimen’s edges.

More and thicker pins than in the gripping study were used. However, lower maximum loads than during the study were reached in the performed tests. It was observed that the casted epoxy cracked earlier in the 3D reinforced specimens.

As a result of this all the load would be carried by the pins, instead of being transmitted by a combination of adhesion and mechanically. This could be the reason why the pins yielded at lower test loads than in earlier trials.

It was not possible to test successfully any of the 3D reinforced specimens with this gripping solution. However, there were still six more milled specimens that were tested without the use of tabs, plain, in the wedge action grips.

These last tests provided relevant information. On the one hand, the specimens seemed to have dry fibres, that can explain why the ultimate loads were smaller than the expected ones. On the other hand, the results also corrobo- rate that the more straight the fibres in the longitudinal direction are after the weaving process (less crimp/ bigger TU), the better the specimens will perform.

This is also confirmed by the noobed fabrics in which the yarns are completely straight. Finally, the possibility to place stuffer yarns in the 3D-woven fabric also improves the final properties of the material.

Regarding the unmilled 3D-reinforced specimens that were not tested, it seems unlikely to be able to test successfully the 8x8 wrp yarn ones. However, these could be milled down in order to provide more results since there were only six specimens properly tested. Milling the surface is also suggested for the 6x6 warp yarn specimens as the ones tested of this kind failed due to the gripping. Finally, there are also two 4x4 warp yarn specimens left that could be tested plain, straight in the wedge action grips.

(47)

Bibliography

[1] N. Khokar. 3D-Weaving: Theory and Practice. Journal of the Tensile Institute, 92(1):193-207. 2001, 2001. [cited at p. 2]

[2] Fredrik. Stig. An Introduction to the mechanics of 3D-Woven fibre rinforced composites. Licenciate thesis. KTH University. 2009. [cited at p. 3, 4, 53]

[3] ASTM D 3039/ D 3039M. Standard Test Method for Tensile Properties of Polymer Matrix Composite Materials. [cited at p. 7, 8, 11, 12]

[4] C.C.C. Chamis. Simplified Composite Micromechanics Equations For Strength, Frac- ture Toughness and Enviromental Effects. SAMPLE. Quaterly, July 1984.[cited at p. 9, 10, 11]

[5] M. R. Garnich and G. Karami. Finite Element Micromechanics for Stiffness and Strength of Wavy Fiber Composites. Journal of Composite Materials February 2004 vol. 38 no. 4 273-292. [cited at p. 10, 45]

[6] J.C. Marn F. Paris, J. Caas and A. Barroso. Introduccin al anlisis y diseo con ma- teriales compuestos. Seccin de Publicaciones. Escuela Tcnica Superior de Ingenieros.

Universidad de Sevilla. ISBN: 978-84-88783-92-9. [cited at p. 10]

[7] F.Stig and S.Hallstrm. Assessment of mechanical properties of a new 3D woven fibre composite material. Compos Sci Technol (2008), doi:

10.1016/j.compscitech.2008.04.047. [cited at p. 10]

[8] J. Degrieck De Baere, W. Van Paepegem. On the Design of End Tabs for Quasi- Static and Fatigue Testing of fibre-Reinforced Composites. Wiley InterScience DOI 10.1002/pc.20564, 2009. [cited at p. 13]

37

(48)

(49)

Appendices

39

(50)

(51)

Appendix A

How to make a two-parts silicone mould

In order to be able to make a two part mould, the things enclosed in the following list are needed:

1. Box with the dimensions of the future mould 2. Plasticine

3. ”Zero specimen”

4. Coat with form separating (Silicone spray) 5. Silicone (In this case Silicotin HB )

The next steps were followed:

1. Manufacture the ”zero specimen”: According to the geometrical requirements, the sample was manufactured out of foam. It was covered with plaster in order to have a smooth surface (without pores) and a future gradual end between the casted epoxy and the test specimen (see Figure A.1a). Absorbent materials such as plaster must be sealed before making the mould. A plastic coat was applied over the ”zero specimen”. The plastic coat was also manufactured by mixing Polyester, 1% Perioxide 1 and less than 2% Liquid wax.

2. Manufacture the box: The box defines the size of the final mould. When manufacturing this, it must be taken into account that for any open surfaces in the mould, the ”zero specimen” will have to be in contact with the box’s walls. In this case, the mould will have two openings, one to pour the casting material and another to exit the test specimen. Thus, the box will have the same length as the specimen as can be seen in Figure A.1c.

41

(52)

42 APPENDIX A. HOW TO MAKE A TWO-PARTS SILICONE MOULD

3. 1^st part: So as to manufacture the first mould part, the specimen that will be copied has to be halve embedded in plasticine. Some guidance holes can be made in the plasticine; this will produce some silicone ”pins” that will close the two parts mould afterwords. Before pouring the silicone it must be ensured that there will not be any leakage. In order to do this, risky spots can be closed with some plasticine. The box, plasticine and

”zero specimen” must be coated too, the silicone spray can be used for this purpose. It is recommended to let the agent dry and polish with a cloth before pouring the silicone. How this step looks can be seen in A.1c.

4. 2^nd part: After removing the plasticine and emptying the box, the second part of the mold will be poured over the ”zero specimen” embedded in the 1^st silicone mould. Before doing so, the box and the content has to be coated once again.

5. Finishing the mould: In order to diminish shrinkage, the silicone manu- facturer may recommend to follow a heat treatment.

(53)

43

(a) (b) (c)

(d) (e)

(f)

Figure A.1: Making a silicone mould (a)”Zero specimen”(b) Mould box(c) 1^st part ready to be poured(d)1^st part dry and unmoulded (e)2^ndpart ready to be poured(f) Finished silicone mould

(54)

(55)

Appendix B

Test protocol and results

The tensile tests were performed at the Department of Lightweight Structure at KTH. The test machine used was an Instron 4505 with a 100 kN load cell.

The test protocol is enclosed in Table B.1^∗.

The next set of pictures, FiguresB.1atoB.6b, show the test specimens before and after performing the tensile tests in the case of moulded specimens. Figures B.7a toB.7f show the results of the specimens that were tested plain.

Test data results are also ploted below. In the case of specimens tested with molded edges, only the load vs machine displacement are enclosed since the strain data was not reliable (Figure B.8a to B.8f). On the other hand, for the other specimens strain data with a correction is also represented. The correction changes the data from a 25mm to a 50mm strain gauge since with this change the results, in terms of elastic modulus, fits with the ones of previous tests. See FiguresB.9a toB.14b.

(a) Before the test

(b) After the test

Figure B.1: Test specimen 1

(a) Before the test

(b) After the test

∗The maximum loads presented in this table have been estimated following [5]. In this procedure the different crimp sizes and that the specimens were milled have been considered.

45

(56)

46 APPENDIX B. TEST PROTOCOL AND RESULTS

(a) Before thetest (b) After the test

(a) Before the test (b) After the test

(57)

47

(d) Specimen 10 (e) Specimen 11 (f) Specimen 12

Figure B.7: Test specimens tested plain after the tests.

Tensile testing of 3D reinforced composites

Kungliga Tekniska H¨ ogskolan (KTH) Lightweight Structures