I hereby license SAMPE to publish this paper and to use it for all SAMPE‟s current and future

publication uses. Copyright 2010 by Sigmatex. Published by Society for the Advancement of

Material and Process Engineering with permission.

Improving Output Rates Of Constant Cross Section Carbon Fiber Woven 3D

Profiles

Christopher McHugh

Sigmatex

Manor Farm Road

Runcorn, Cheshire WA71TE

ABSTRACT

With the requirements to develop rapid processing of carbon fiber composites and emphasis on

fiber deposition rates, 3D weaving presents many opportunities. Optimization of weaving

technology can present many ways of producing constant cross section profiles for stiffening

applications. Current standards are expensive and have constraints in manufacture, limiting

production output. Although improvements can be made, processing limitations will restrict

output rates. Multiple profiles can be manufactured across the width of a wide weaving loom

enabling a substantial increase in production rates for popular T and Pi section structures. This

paper will demonstrate the advantages to be gained from this method of 3D component

manufacture using this advanced weaving process and will examine opportunities for future

developments using textile processing techniques.

1. INTRODUCTION

This work is concerned with the process of manufacturing 3D woven constant cross section

profiles to improve output rates based on current technology. Within the scope of the paper

analysis of the Pi structure will be undertaken and results demonstrated on how 3D Pi section

compares to conventionally sections made from layup techniques. Included in the work will be

explanations of the methods used to design and manufacture the preforms, the infusion method

and test method. Pi relates to the Greek symbol π as outlined in a comprehensive study on joints

by Schmidt et al.1. The work provides excellent detail as to how the preforms can be used as an

enabling technology and spans a decade of work on the subject. The work contained in this paper

utilizes a different method of manufacture which allows various constant cross section profiles

including the Pi and T sections to be manufactured at a higher rate, providing more opportunities

in a broad range of applications where increased throughput is required. The need for rapid

processing techniques is acknowledged throughout the composites industry which is further

substantiated with the UK government investing heavily in new enabling technologies.

1.1 Purpose

The main purpose of this work is to demonstrate how improved output rates can be realized with

the use of optimized 3D weaving techniques. It is also anticipated that performance benefits can

be achieved by incorporating key features into the woven 3 Dimensional profiles. During the

course of the paper, design of components will be demonstrated along with infusion method and

ultimately test results showing a comparison between conventional lay up structures and 3D

woven structures.

2. WEAVING MANUFACTURING METHODS

With the introduction and development of carbon fiber in the 1960‟s, methods of converting the

fiber to fabric quickly developed. Although there are many methods of weft insertion systems

available, few are utilized for standard carbon fiber weaving due to fiber control. As carbon is a

brittle fiber, handling and thread transfer are of primary importance during the weaving process2.

When considering weaving of 3D structures there are even fewer suitable manufacturing

processing routes. There are however options and each insertion method provides benefits and

constraints in the manufacture of 3D constant cross section profiles.

2.1 Current Preferred Method of weft insertion for Constant Cross Section Profiles

The Pi performs which are currently used for potential aerospace projects based on the Lockheed

Patent 3, are manufactured using weaving technology where the fiber carrier is passed through

the warp fibers during each weft insertion. Shuttle weaving although used in fabric manufacture

since 1733, is a method of weaving which enables the weft or fill transfer in this way. There have

been improvements in shuttle technology, however this method of weft insertion has limitations

on the amount of fill fiber the pirn or carrier can hold as it traverses from each side of the woven

material. The speed at which the fiber is laid down is also limited by the picking mechanism of

the loom. Figure 1 illustrates the method of creating a single T section profile on a shuttle loom.

There are narrow shuttle looms available enabling multiple insertions across a given width4,

however the speed and fiber capacity is still limited.

Figure 1 - T Profile Manufacture

2.2 Alternative methods

The manufacture of 3D profile fabrics is not limited to conventional insertion methods. Biteam

boasts the Dual-Directional shedding system for the manufacture of profiles, whilst other

bespoke machines also have the capability. The high speed Dornier rapier weaving machines as

used at Sigmatex offer a wide range of capabilities especially when combined with complex

shedding mechanisms. By utilizing high speed rapier looms weaving at a wider width, increased

rates of production are possible. The method of manufacturing the Constant Cross Section

Profiles in this work has been achieved by designing and manufacturing multiple cross sections

across a wide loom. The width of the loom allows a greater number of samples to be produced

per weft insertion, with up to 18 profiles at 7.6 cm width capable of being accommodated as

opposed to 4 profiles on a 4 head shuttle loom. If we compare the output rates of a state of the art

4 head shuttle, with a rapier loom of 140 cm width, manufacturing a 7.6 cm profile, the output

rates are much higher. For example at similar Fill densities of 42 threads per cm and both looms

weaving at 150 insertions per minute the high speed rapier loom will produce 4.5 times more

yards of profile than the shuttle.

3. DESIGN FOR MANUFACTURE

To manufacture the constant cross section profiles using a full width Jacquard with individual

end control, there are a number of considerations. Primarily the loom setup will be configured to

enable the profiles to be manufactured. The width of the constant cross section profile will

determine the number of profiles able to be accommodated across the width of the loom. For

each section with a bound edge there will be a selvage formation created by using a non carbon

thread. In the image Figure 2, the edges are visible by the series of Kevlar threads, interwoven

with the profile legs. Woven edge seams interlocking the legs of the Pi provide stability for the

cutting process, after which the separate section is ready for further processing as a dry preform

for the infusion process.

Figure 2 - Edge binding of profiles

4. PREFORM DESIGN

In order to manufacture 3D structures which provide the required performance criteria, ultimate

shedding control is required. The Lockheed Pi section has multiple layers of woven carbon

interlocked together. The vertical legs are also engineered to provide the best performance by

placing the fibers in the desired location all of which are controlled by Jacquard technology. The

complexity and development of the patented cross section has led to the licensing of the

technology to various weavers, however the design and method of manufacture is prescribed and

cannot be altered without further investment. For this reason the work in this paper will compare

Pi profiles which have the same fabric aerial weight but will have different methods of

interlacing. All materials are manufactured from the same fiber, woven on the same weaving

machine and infused with the same epoxy resin system to minimize the effects of process and

materials.

4.1 Design method

Three Pi profiles have been manufactured, a hand layup Pi profile, an Orthogonal Pi with

interwoven Z binding and a Bound Pi, where the layers of the legs are interwoven. All the

profiles were manufactured using Carbon fiber type FT300B 6000 50B at a warp density of

35.46 threads per cm. The difference in each perform was due to the fiber paths, with the

baseline layup version having no „Z‟ direction interwoven elements.

4.2 Design Theory

In order to design a Pi joint which would potentially add extra strength, it is important to

understand the failure mode of current joints using standard non 3D fabrics. For example when

considering the load of the woven Pi profile during a pull out test (See Figure 9),Figure 9 -

Tensile Test Pullout setup interlaminar failure ultimately results in the joint failure Figure 15. It

is expected that by introducing through thickness reinforcement at the legs and base the

interlaminar properties will be improved and ultimate pullout strength increased.

4.3 Baseline Pi profile

The baseline hand layup version was manufactured using a 5 harness weave style which was

woven as an unbound multilayer structure. As with all 3D woven structures manufactured at

Sigmatex the offset shedding mechanism was utilized to reduce fiber degradation5. The layup of

the profile is demonstrated in Figure 3. The image shows 4 layers of fabric layered to form the

required shape, with each blue line signifying one layer of material.

Figure 3 - Baseline Fabric Lay up

4.4 Orthogonal Woven Pi Preform

The Orthogonal woven Pi is formed into a preform as demonstrated in Figure 4, with layers of

Orthogonal structure bound at the centre point during the weaving process with „Z‟ woven fibers.

The „Z‟ fibers are expected to provide through thickness reinforcement at the base of the Pi. The

image shows the thread paths of the fill threads colored in blue with the binder yarns in black.

Figure 4 - Orthogonal Structure showing ‘Z’ stitching of Pi

4.5 Bound Pi Preform

The bound Pi preform is designed so that the fibers which form the horizontal legs cross over at

an interchange point to form the vertical leg. This allows fiber continuity between the base and

the upright legs potentially adding extra strength against pull out. Conventional layup structures

would be reliant on the strength of the matrix to prevent delamination and failure of the joint,

however by crossing the fibers at the interchange the fiber strength also resists the pull out under

tensile loading.

Figure 5 - Fiber paths of Bound Pi

5. INFUSION METHOD

5.1 Equipment Summary

Vacuum Pump - Single phase piston pump,

Valve Assembly - Assembly of connected vacuum valves incorporating sensor for vacuum

gauge, condenser and electromagnetic safety valve. Assembly also incorporates branch vacuum

line to resin de-gassing chamber.

Resin Trap

Control Panel – Vacuum pump control, mould heating elements and control, emergency switch

and vacuum gauge display dial.

Flat Mould – Polished surface 1 inch thick aluminum mould with controllable heating element.

5.2 Infusion Method

A Huntsman two part Epoxy resin system was used which allowed for a low temperature cure of

8 hours at 80 °C. Prior to infusion the aluminum base-plate was pre-heated to 50 °C. The 2 parts

were mixed at room temperature then degassed for 5 minutes prior to infusion. The Pi sections

were prepared on the base plate with a 3D woven flat structure inserted between the upright legs

of the Pi to allow the parts to be co-infused. The image Figure 6 shows a schematic of a Pi with a

3D insert panel highlighted in a red color.

Figure 6 - 3D insert and 3D Pi Preform

Flow medium was placed vertically at the insertion side and along the top surface of the joints to

provide a good flow of resin from the inlet. The flow medium was yellow knitted monofilament

material and is shown in Figure 7. The part including the flow medium was bagged using

standard film and then infused. The viscosity of the resin was such that the entire structure was

totally wetted out and the resin witnessed after 8 minutes. To provide an upright version of the

leg, box section pieces were clamped to the infused samples. The sample was left on the heated

tool for 8 hours.

Figure 7 - Sample prepared for infusion

5.3 Sample Preparation

All samples were prepared to provide the same base width. The orthogonal version as seen in

Figure 4 was prepared in such a way that the repeat of the orthogonal Z binder was the same for

each sample. Unlike conventional flat 2D layup materials, the through thickness binding repeat is

essential to ensure consistent test results. The image in Figure 8 illustrates a „Z‟ binder repeat

showing the finer through thickness Z binder. It is essential that when samples are prepared for

the orthogonal version that the same amount of binders are in each sample. If the samples were

prepared with only dimensions taken into consideration then there is a possibility that the amount

of through thickness reinforcement would vary even though the sample size was the same. It is

for this reason that for testing relating to 3D woven structures the weave repeat size and

dimensions need to be considered.

Figure 8 - Design repeat showing 'Z' binder

6. JOINT STRENGTH COMPARISON

6.1 Test Overview

All the samples were tested using a simple pull out test method based loosely around ASTM D

3039/D. For comparative testing this was deemed suitable and would provide comparative pull

out performance relating to load and elongation. The primary objective of the test analysis is to

determine the relationships between the through thickness fiber, failure mechanism and ultimate

load.

6.2 Setup

The samples were mounted into the jaws of the 200kN Instron Tensile Testing machine as

illustrated in the image below. A cross head speed of 4mm/min was used for the test and samples

were loaded and clamped to ensure there was no interference with the radius at the base of the Pi.

Figure 9 - Tensile Test Pullout setup

7. TENSILE PULLOUT TEST RESULTS

7.1 Pull Out Test Results

The test results show the maximum load achieved for the various cross section weaves and the

hand layup sample. The results demonstrate only the maximum load achieved for the different

weave styles.

Figure 10 - Orthogonal Load / Elongation

Table 1 - Orthogonal 3D Pi Preform Tensile Pullout results

The failure mode of the Orthogonal Pi is totally different from all other Pi test results. As can be

seen in the image (Figure 11), and the photograph (Figure 12), the through thickness binding is

pulled out during the tensile tests, with the fibers protruding from the failure location. Given that

the ultimate strength is provided by the strength of the „Z‟ binders there is opportunity to

increase strength in a number of ways. Introducing higher strength „Z‟ fibers, increasing the

density of „Z‟ binders and increasing the footprint size will all contribute to higher maximum

pullout load. A further short analysis has been carried out and is discussed in the conclusions.

Figure 11 - Orthogonal Failure Mode

Figure 12 - Photograph of Orthogonal failure

Figure 13 - 3D Bound Load / Elongation

Table 2 - 3D Bound Pi Preform Tensile Pullout results

Figure 14 - Hand Lay Up Load / Elongation

Table 3 - Hand Lay Up Pi Preform Tensile Pullout results

Figure 15 illustrates the failure mode of the Hand layup version with the inter laminar plies

separating under loading.

Figure 15 - Hand Layup Failure Mode

8. CONCLUSIONS

8.1 Manufacturing

The primary focus of this work was to demonstrate the increased output rates that can be realized

by optimizing machine and design setup. Previous work1 demonstrates how Pi technology can be

manufactured using 3D weaving methods and be successfully used in aerospace applications.

However the advantages of the technology are yet to be fully realized, with much resistance due

to the cost associated with 3D woven structures. This cost is not prohibitive when utilizing

conventional manufacturing methods, as used in this work. By increasing manufacturing output

rates as with commodity products, the cost of manufacture reduces. As composite parts are

manufactured out of autoclave and using resin transfer methods as opposed to the prepreg route,

3D structures become a more attractive solution. The comparative output rates demonstrated in

Figure 16 show what improvements can be realized by using the different technologies which

will have the effect of improving the affordability of manufacturing complex composites.

Figure 16 - Output rates Comparison

8.2 Testing

The tests carried out as part of this work provide an indication of the improvements in ultimate

strength that can be gained with the use of 3D woven structures. The strength demonstrated in

the Orthogonal Pi structure relies only on the through thickness „Z‟ fiber for pullout resistance.

With the relatively low amount of through thickness re-inforcement provided, it is feasible that

by making the footprint of the base of the Pi joint thicker, the increase in through thickness will

provide very high load capability. Further work will be done on the relationships between

maximum pullout load and the amount of 3D woven through thickness „Z‟ fiber re-inforcement.

The graph below (Figure 17), demonstrates the effect of an Orthogonal Pi with 50% less binding

(Specimen 6), and as can be seen the resultant maximum load is approximately 50% lower at

2.5kN. The graph is only an indication and further work will be required to substantiate this

initial finding.

Figure 17 - Comparison of Orthogonal Pi Preforms with 50% less binding

8.3 Further Work

With the growth in composite manufacture, the interrelationships between software become

more important. For example the component geometry will transfer from engineering software to

textile design software then predictive analysis software as provided by ESI6. Whilst a great deal

of work has been publicized around the prediction of 3D material properties7,8

, further work is

required to provide the required detail for complex 3D components. Ultimately the ability to

provide accurate material properties of virtual components created using textile design software,

from fiber / fabric level, to 3D modeling software for finished part would be preferred.

Although this paper has demonstrated variations of 3D woven Pi structures, there are many more

feasible designs yet to be realized. As design and manufacturing methods are further developed

for 3D woven structures and end users are familiar with the fiber paths and resultant stresses in

the preforms, greater advances will result. It is foreseeable that ultimately the designer will be

able to pre-weave the 3D components, inputting the weaving characteristics, then forming the

required preform with all of the inherent deformations without weaving a single thread. Work

has begun on this methodology, however a great deal of development and computing power will

be required if the flow of predictive analysis is to become reality. The image illustrated in Figure

18 demonstrates how the weaving process is beginning to be modeled for simple structures. The

work was part of the Technology Strategy Board i-Composites Project, which also co-funded the

research outlined in this paper.

Figure 18 - ESI Weave Stress Simulation

9. REFERENCES

1 Schmidt.R.P, Lee.S.M, Cooke.L.M. 3D Woven Pi Preforms Joints: An Enabling Technology for Large Composite

Structure. SAMPE CD 2008

2 Lindauer-Dornier Web Page, Accessed 28-03-11 http://www.lindauerdornier.com/home-fr/archives/june-2008-50-

000th-dornier-rapier-weaving-machine

3 Schmidt.R.P et al. US Patent No. US6,874,543 B2. “Woven Preform for Structural Joints” Sept.2002

4 Mageba Textilmaschinen Web page. Accessed 20-11-09 http://www.mageba.com/eng/shuttle.htm

5 McHugh.C. The Use of Recent Developments in Conventional Weaving Shedding Technology to Create 3D One

Piece Woven Carbon Preforms. SAMPE Journal. 45(6) (2009):33-41.

6 ESI Web Page. Accessed 28-03-11http://www.esi-group.com

7 Crookston JJ K.S., Warrior NA, Jones IA and Long AC, 2007, "3D textile composite mechanical properties

prediction using automated FEA of the unit cell", presented at the 16th International Conference on Composite

Materials (ICCM 16), Kyoto, 2007

8 Biragoni P. and Hallett S.R., 2009, "Finite element modelling of 3D woven composites for stiffness prediction",

presented at the 17th International Conference on Composite Materials (ICCM 17), Edinburgh, 2009