Manufacturing of Composite Parts Reinforced Through-Thickness by Tufting

G. Dell’Anno*, J.W.G. Treiber†, I.K. Partridge‡

*The National Composites Centre, Bristol and Bath Science Park, Bristol, BS16 7FS, United Kingdom(Corresponding author: giuseppe.dellanno@nccuk.com, +44 0117 9560762)

† Manufacturing and Materials Department, Cranfield University, Cranfield, Bedfordshire, MK43 0AL, United Kingdom

‡ Advanced Composites Centre for Innovation and Science, Department of Aerospace Engineering, University of Bristol, Bristol, BS8 1TR, United Kingdom

AbstractThe paper aims at providing practical guidelines for the manufacture of composite parts reinforced by tufting. The need for through-thickness reinforcement of high performance carbon fibre composite structures is reviewed and various options are presented. The tufting process is described in detail and relevant aspects of the technology are analysed such as: equipment configuration and setup, latest advances in tooling, thread selection, preform supporting systems and choice of ancillary materials. Effects of the process parameters on the preform fibre architecture and on the meso-structure of the reinforced component are discussed. Special emphasis is given to the different options available in terms of tuft insertion and loops management.

Potential fields of application of the technology are investigated as well as the limitations of its applicability in relation to preform nature and geometry. Critical issues which may arise during the manufacturing process concerning thread insertion, loops formation, alteration to the fabric fibres layout or local volume fraction are identified.

Keywords: Tufting; Composite; Stitching; Through-Thickness Reinforcement; TTR; Preform; 3D reinforcement

1. IntroductionThe growing use of composite materials for structural and semi-structural components in the aerospace, defence, transportation, civil and energy sectors has dictated the need for the development of automated manufacturing systems capable of producing, at the required rates, large, often complex, and high-performance parts. The unique manufacturing processes involved in composite parts production require the development of specialised technologies and the design of dedicated machinery. Generally speaking, the production line must account for the transformation of long filament fabrics (typically available as broad goods) into 3D shaped, resin impregnated, solid parts: this involves cutting plies out of the fabric roll, stacking them up over suitably shaped moulds, forming each ply individually to the required geometry, and finally curing the polymeric resin to embed the fibrous reinforcement is a consolidated matrix. The dexterity needed for such operations makes it challenging to devise automated processes which would replicate the functions of skilled operators; nevertheless the adoption of robotic manipulators and much specialised end-effectors is making this possible. Engineering companies now focus on the development and commissioning of full lines for the preforming, curing, machining and assembly stages of the composite part production.

In conventional composite laminates, the 2D fibre arrangement across the individual plies and the inherent brittleness of highly cross-linked matrix resin, makes the part subject to cracking in the interlaminar region between adjacent fibrous plies. This is the most likely consequence not only of direct out-of-plane loads, but also of high

1

energy impacts. Good impact resistance is achieved through the minimisation of the damage extent within the composite structure, which in turn makes the structure more damage tolerant, by reducing the likelihood of progressive growth of the crack (or cracks) to a size which will cause structural failure. Several techniques are currently available to enhance the delamination resistance of polymer matrix composites. The manufacturer’s selection of a particular technique over the others depends also, among other factors, on the primary composite manufacturing method involved and must be taken into account in the design of the automated system for the composite component production.

One well established method for improving damage and delamination resistance of composites is through resin toughening [1], which promotes phenomena like crack blunting, crazing, particle cavitation, shear banding and void coalescence as energy absorbing mechanisms [2,3]. Systems based on blends of polyethersulphone (PES) and polyetherimide (PEI) were identified in the 1980’s [4,5]. Other effective resin systems were developed in those years, such as polyetheretherketone (PEEK) and polyphenylenesulphide (PPS), which would also satisfy the requirement of the primary structure for high stiffness and strength. However, these systems were both high cost and difficult to process, having high melting temperature and needing high mould tooling pressures [6]. An alternative approach was developed by Toray in the early 1990’s for the Boeing 777 aircraft in which thermoplastic particles were applied to the surface of epoxy resin/carbon fibre prepreg [7]: the prepreg type Toray 3900-2 proved very effective for damage resistance, although at a higher cost than standard prepregs.

Alternative, non-resin type toughening approaches use embedded reinforcing elements, sometime referred to as micro-fasteners, through either part of the laminate or the complete thickness of the assembly to reduce the risk of plies delaminating or disbonding. Once a three-dimensional fibre architecture is obtained, delamination or disbonding requires the pull-out or breakage of such micro-fasteners [8-13].

One technique which provides the part to reinforce with an extra load carrying medium through its thickness is Z-pinning [8], which was developed in the U.S. by Foster Miller, then Aztex, now Albany International, and involves embedding an array of thin1, rigid pins (or Z-Fiber®) through the laminate or assembly before its final cure in autoclave. Once embedded in the cured part, the pins must be broken, pulled out or at least heavily deformed to allow crack growth (Figure 1). This has been proven to be a very efficient technique, and cost effective compared to using conventional mechanical fasteners [14]. However, it is a complex method, involving mainly manual operations and specifically developed for prepreg structures, since the inserted pins are held in place by the uncured matrix.

Figure 1: On the left: SEM of chamfered portion of carbon Z-pin after frictional pull-out from the embedding carbon/epoxy laminate. On the right: cross-section of the Z-pin partially failed in shear.

Three-dimensional weaving technologies certainly address the problem of delamination at its root albeit being cost-effective mainly at higher production volume, when manufacturing large scale, continuously 3D reinforced

1 Available diameters are 0.28mm or 0.51mm.2

laminates. Their inability to provide preform taper without unacceptable fibre wastage or to provide easily tows at 45 degrees severely limits their field of application.

Stitching is well suited to preform-type processing: it is both simple and relatively low cost, while lending itself to localised Z-reinforcement of sections in the composite with high out-of-plane stress state [15]. Early stitching research focused on conventional sewing, using two interlocking threads [16-19]. This was proven to be very effective for damage resistance by the NASA stitched wing programme in the late 1980’s, but required extremely high sewing machine investment. The high cost of the unit was due not only to the scale and complexity of the equipment but also to the need of accessing both sides of the preform being reinforced to interlock the threads [20]. More recently, variations on this theme have been exploited such as stitching technologies which only require access from a single side of the structure, usually referred to as one-sided stitching technologies (Figure 2).

Figure 2: Examples of one-sided stitching technologies: (a) and (b) use two straight needle and two threads or a single thread, respectively. The system in (c), also called ‘blind-stich’, uses one curved needle and a single self-interlocked thread. Adapted from [21-24].

Tufting represents the simplest version of the one-sided stitching approaches and it is specifically designed for the dry preform/liquid resin moulding process route. It represents a further stage, prior to the resin infusion process, in the manufacturing procedure; nevertheless, it may be considered a relatively economical method of obtaining a three-dimensional fibre architecture [25].

Originally an ancient method for carpet and warm garments manufacturing, tufting has now become a commercially available technology for through-thickness reinforcement (TTR) of thermosetting polymer matrix composites. It involves the insertion of additional tows, via a single needle, through the layers of a laid-up dry preform. The tows can be fully inserted (Figure 3a) or applied to a partial depth through the preform thickness (Figure 3b), orthogonal to the preform surface or angled (Figure 3c). When the needle penetrates the whole preform thickness, a loop of yarn is formed on the underside of the structure. The loops are not tied or inter-locked and the tufts remain in position because of the natural friction between the fabric and the thread. This technology requires access from a single side of the preform, which makes it ideally suitable for local, tailor-made reinforcement of complex, three dimensional shapes. In this paper a feasible manufacturing procedure will be outlined, in the form of practical guidelines, for the production of tufted composite materials. The mechanical performance and the failure mechanisms of the tufted composite are covered elsewhere [26-30].

Figure 3: Schematic of the thread arrangement in a tufted preform: (a) details the sequential steps of a full insertion, (b) illustrates the option of partial penetration while (c) shows angled insertion.

3

2. Scope of Through-Thickness Reinforcement methodsIn principle, all TTR technologies represent valid methods to reinforce structural joints, as well as a potential alternative to mechanical fastening and bonding. On this basis, some authors have proposed a classification of stitching based on the function of the stitch: ‘fixing and positioning’ seam, ‘assembly’ seam, and ‘structural’ seam [31]. The first involves joining two or more reinforcing fabric layers in a two-dimensional preform, the second is used to hold together various sub-components in a single three-dimensional preform, and the third consists of the use of the stitch as a reinforcing element to change the mechanical properties of the composite. The use of TTR elements, including tufts, within complex 3D structures appears particularly relevant as they might accomplish more than one of these tasks at the same time.

Stiffening elements with I, C, T, Π or Ω geometries are commonly used in composite components manufacturing and their failure mode often involves delamination between the surfaces in contact, i.e. flanges and skin [32]. There are examples in the literature of T stiffeners whose flange-to-skin joint is reinforced by the use of stitches [33-36], Z-pins [37, 38] or tufts [39], although the use of 3D reinforced composite components in the aerospace industry is currently limited, mainly because of the lack of mature and reliable modelling tools [40].

Normally, joining between aircraft skins and spars, stiffeners or ribs is achieved by the use of either a riveting or bolting process. When metallic type fastening is applied to carbon fibre composite structure, manufacturing cost is invariably very high. The use of TTR elements for replacing mechanical fasteners offers the possibility of manufacturing the structures by component integration, which involves joining the sub-components together in the dry state, placing them in the same mould and then co-curing the assembly as a single preform, avoiding or significantly reducing the need for mechanical fastening. Co-curing represents the lowest cost process [41] when compared to either moulding the detail parts individually and bonding (post-bonding) or incrementally curing detail parts onto pre-cured larger parts (co-bonding). Co-curing requires the fewest mould tools, has the minimum process stages, with correspondingly less labour time and quality assurance, and requires the least surface preparation for bonding. Given its flexibility and the need to access the preform only from one side, tufting seems ideally suited to exploit this trend further.

3. Tufting unitIn line with the rising expectation of cost effective manufacturing in the composites industry, the tufting process has been automated. Commercial tufting cells consist of a tufting end-effector interfaced to articulated robot arms, typically with six rotary joints. The degrees of freedom of the set-up can be increased further by adding tracks for the robot to slide on, or mandrels to manipulate the tool. Early models of tufting heads, such as the KSL KL150 in Figure 4a, are mechanically driven by a single motor, linked by belts to a crank mechanism which controls simultaneously the needle and the presser foot strokes, as well as the oscillating movement of the main shaft, designed to keep the needle and presser foot in a quasi-static position while the former is inserted in the preform [27]. The relative motion of all these components is, consequently, interlinked by the internal mechanisms of the tool. This type of units are capable of tufting at rates up to 500 tufts per minute; rate which has been tested successfully on 5mm thick unbindered or lightly bindered preforms and that is considered acceptable not only in the research environment but also by industrial standards.

4

Figure 4: (a) Kawasaki FS 20N robot unit with KSL KL150 tufting head while tufting a large preform with double curvature surface. (b) Kuka KR 240-2 robotic arm interfaced to KSL RS522 tufting head while tufting a simple single curvature preform.

Current models of tufting heads, like the KSL RS522 in Figure 4b, have abandoned the single motor belt transmission in favour of a fully geared mechanism, where the movements of the needle, foot and shafts are controlled independently with three electronic motors providing a much more accurate control over the tuft placement as well as delivering more power during the insertions. Such modern units have been shown to be capable of tufting 10mm thick, bindered and highly consolidated preforms, previously impossible to penetrate automatically. The electronic control over the individual parts of the end-effector (needle, foot and shafts) allows continuous feedback to the controlling PC as to the needle position at all times. This has increased considerably the level of repeatability of the process compared to earlier systems, where the needle spatial arrangement had to be estimated with some approximation by the program while the routine was executed, leading to a slight discrepancy between the predicted and the actual position of the tufts, especially when tufting large areas.

The robot arm provides the unit with a high level of flexibility, allowing adapting the reinforcement to the specific requirements of the different geometries and a variety of assembly setups. The ease of adjusting the tuft placement layout and pattern depends on the robot control software and hardware. In the setup in Figure 4a, the trajectory tracking has been achieved using the dedicated Kawasaki software and the robot controller language. This is a basic, high-level, text based machine programming language, not in line with the requirements of user friendliness expected from a modern programming package and intended to be used in heavy industry applications where the simple, repetitive actions performed by the robot are unlikely to be changed often during its lifetime. The unit setup does not allow an easy connection to other, more flexible, programming tools such as LabView®. The solution shown in Figure 4b offers more up-to-date capabilities in terms of system control, making these systems relatively manageable also by non-experienced users. Software capable of importing the shape of the surface to be tufted directly from CAD files is now available, which allows fuller exploitation of the versatility and adaptability of the system to different needs.

Apart from the geometric arrangement and pattern of the tufts or blocks of tufts, which is an evident process parameter, the accuracy to which these are placed across the preform also determines the efficiency of the reinforcement. Tufting tool suppliers offer the option to provide their units with high accuracy robot models such as the Kuka KR100-HA which, nominally, can control placement of the tufts within +/- 0.05mm.

In order to maintain such a level of accuracy on setups working with several interchangeable end-effectors, it is critical that the robot arm and joints are very stiff, that the robot maximum payload is sufficient to withstand the forces acting on the tufting tool during the repeated needle insertions, and that the hardware connection between the tool and the robot arm is designed with very strict tolerances. In those setups where it is impossible to realign the connecting flanges exactly in the same position when the end-effector is taken off and reinstalled, a calibration procedure is needed to teach the robot the offset between the centre of the flange at the end of the arm and the tip of the tool. While this may be acceptable in the lab environment, it would be inadequate to any industrial

5

arrangement. One effective, although costly solution involves equipping the robot with fully automated tool loading systems which use specifically designed racks and pneumatically driven connections and interfaces to guarantee accurate and repeatable tool/robot attachment. This option is currently available on the newest tufting units.

4. Tufting threadThe development and the adoption of specialised continuous yarn tufting threads are essential aspects of this technique. The thread must be not only suitable for tufting but also compatible with the liquid resin moulding type processes for composites manufacture and with the subsequent mechanical and durability performance demands on the final composite. Although sometime non-specialised continuous fibre tows have been used for tufting [42], designing a thread grade specifically conceived to be used in conjunction with such technology is likely to improve both the manufacturing feasibility and the mechanical performance of the tufted composite. An ideal thread product would have a relatively high twist level, given the need of high flexibility in bending to withstand the sharp kink from the needle during insertion. This is particularly important when the filaments adopted are fragile in bending, as in case of carbon fibre. Based on current experience, threads made with two or three highly twisted yarns (up to 300 turns per meter) in which each single yarn is twisted in the opposite direction over 250 times per meter, are flexible enough to be successfully tufted or stitched [30].

Table 1 lists the grades of thread tested to date which proved suitable for tufting, with the aramid being the most robust. The brittle nature of the carbon filaments makes the carbon thread susceptible to local splitting under high curvature in the needle eye especially when used with highly packed preforms. One grade of stretch-broken carbon fibre was successfully tufted but, being thinner than the equivalent long filament grade, exhibited lower mechanical properties, without offering particular advantages to the manufacturing process. Threads made of bundles of short fibres like the carbon/Zylon® thread developed by Schappe Techniques2 can in principle represent an alternative solution, although the available grades currently available have too large a diameter and are incompatible with the tufting needles available.

Table 1: Details of a selection of thread grades suitable for tufting

Thread type Carbon Stretch-broken Carbon Aramid Glass

Thread specification Tenax® Carbon Nm25/2 Carbon Kevlar® Tkt 30 EC9 68x3 S260Fibre HTA40 HTA40 Kevlar® 29 EC9 68 Z28

Manufacturer Schappe Techniques® Schappe Techniques® Somac Threads Saint Gobain Vetrotex

Linear weight[g/km] 140 80 92 204

Filament count 2 x 1000 2 x 590 4 x 134 3 x 411Dry cross-section area[mm2] 0.077 0.045 0.065 0.078

Carbon thread damage was observed on the tuft loops (Figure 5a) as well as at the insertion point (Figure 5b) into relatively loose non-crimped fibre (NCF) carbon fibre preforms. While the former is non-critical, the latter has potentially detrimental effects on the mechanical bridging performance of the tuft and must be avoided. In the selection of the thread material a compromise must be found between suitability for the manufacturing process and the mechanical performance in the cured composite.

2 Twisted carbon fibre thread wrapped in Zylon® (polybenzoxazole) filaments. Zylon® content is 11% and the total thread weight is 4400m/kg.

6

Figure 5: Evidence of damage on carbon fibre thread loops (a) and at the insertion point (b)

The thread can be simply pulled out of the spool held close to the robot. If this configuration is chosen, a cone shaped bobbin rather than a cylindrical one facilitates the operation. Current systems offer the option of a spool directly installed on the tufting head, as close as possible to the needle and with some form of controlled feeding rate which simplifies installation procedures and minimises the risk of thread entanglement and breakage.

5. Tufting needleTufting needles have been specially designed with an inclined hole at the tip of the needle (‘eye’) to enable the insertion of thread loops into the dry preform without interlocking. The choice is limited to the two types shown in Figure 6: the silver coloured alternative (on the left hand side in the picture) is claimed to reduce the damage to the thread during insertion. It has approximately square section of 2.3mm side and rounded edges, and it is slightly larger than the model on the right, which has a maximum width of 1.47mm and a depth of 1.85mm.

Figure 6: Front view (top), side view (bottom) and longitudinal cross section (in the bottom right corners) of the two types of needle tested.

These dimensions are rather large in comparison with the usual unit cell of a dry preform but they are required for robustness in repeated application. The profile section of both needles is ‘C’ shaped and provides a channel, terminating in the needle eye, on the side of the shaft facing the tufting direction. While this penetrates the stack of fabric, the thread is pulled from the feeding spool and runs through the channel with minimum friction, facilitating tuft formation. The size of the channel and of the needle eye obviously dictates a limit on the thread dimension: the needle in Figure 6a can accommodate larger threads as its eye is 1mm in diameter versus the 0.68mm of the other model. The use of threads with a nominal diameter exceeding 500µm is not recommended as this has shown to increase both the level of damage to the thread and the number of breakages during the process.

Some thread grades offer the option of a lubricating sizing applied on the yarns to reduce friction with the needle and preform; while reducing the amount of potential damage to the thread during insertion, such an option should

7

be avoided as it might interfere with the mechanical performance of the tuft by affecting the tuft/matrix interaction in the cured composite.

There is certainly significant scope for improving the design of the needle, for example by modifying its internal channel geometry or the tip shape: new models might improve the needle wear-out, reduce damage to the thread and minimise disruption to the preform fibres. Current experience suggests that, when tufting with short pitch (i.e. less than 3mm), a shorter and thinner needle might help limiting the fabric disruption and improve tuft quality. However, to date, no research work has been specifically dedicated to this aspect of the technology.

6. Thread insertionThe penetration depth of the needle can be adjusted to suit different needs, up to a maximum of about 38mm. This limit is posed by the needle length as well as by its stroke, which may be either independently controllable or, in case of fully mechanical tufting heads, constrained by the geometry of the tool internal mechanism.

If loosely woven dry preforms are used, then the size of the needle seems to pose little problem in terms of fibre breakage: only few filaments are damaged as the fibres are able to move out of the way of the needle (Figure 7). However, significant fibre damage can be expected to result in the tufting of highly bindered preforms. The effect of such damage may potentially result in a reduction of the strength of the final composite, although review papers on this topic highlight variability in the conclusions of different studies, with some reporting reduction, others no variation and some improvements in strength within a range of 15-20% [43].

Figure 7: Damaged filaments after needle insertion in a dry carbon fibre preform (a) and evidence of filament breakage in the infused composite (b)

Experience to date indicates that knitted fabrics are unsuitable for use with this technology, as the interlocked tows arrangement makes it difficult for the mesh to widen and accommodate the needle [30]. Woven fabrics are relatively easy to tuft, whilst NCF fabrics appear ideally suited for tufting, although some problems have been encountered when tufting thick layups of NCF fabric incorporating glass non-structural stitching yarns, as these prevent the in-plane movement of the filaments increasing significantly the resistance to needle penetration. Depending on the nature of the preform and on the amount of binder, the needle might need to be changed frequently. Needle wear becomes eventually apparent because of the distinctive noise the blunt or even chipped tip generates during the insertions, however, a needle change would be advisable before this stage is reached to minimise potential damage to the preform. Needle re-sharpening should be discouraged as modification of the peculiar tip geometry might affect the efficiency of the insertions. Given its dependence on the particular substrate being tufted, the optimal interval between needle changes is to be defined on a case-by-case basis, upon execution of experimental trials and systematic microscopy of the needle tip. Broadly speaking, the tip wears out after a few working hours if the preform is particularly thick or the fabric particularly tight. A quantitative analysis of the stress and strain to which the needle is subjected while tufting is available in [44].

8

The additional possibility of inserting the thread at an angle between 45° and 135° to obtain oblique tufts was implemented and tested (Figure 8). The standard presser foot used to hold the fabric down and steady while it is tufted is relatively wide and designed for perpendicular insertions only. The option of removing it altogether might appear as an easy, although not ideal, solution when tufting at non-right angles. However, experience has demonstrated that removal of the foot element may jeopardise the quality of the tufts in that its role is not only to hold the preform down while the needle is withdrawn, but also to lock the last inserted tuft against being pulled back during the next insertion. For non-perpendicular insertions, a custom made foot or at least modification of the standard foot shape to fit the particular angle of insertion should be considered; this would allow retaining the functionality of the foot while maintaining the freedom of selecting the penetration angle.

Figure 8: Angled insertion of tufts at 45° for tufting of 3D woven Pi-stiffener to 5-harness woven skin. The picture on the right shows detail of angled support.

The relevance of the pressure exerted by the foot has been investigated on spring driven presser foot of the KSL KL159 and it has been observed that equipping it with a stiffer spring, able to exert a pressure of up to 20N when fully compressed, ensures the formation of more even loops on the backside of the preform, which is an essential prerequisite for loops length control. In this model, the level of pressure exerted by the foot on the fabric can be adjusted by moving the head closer to or further from the preform surface, therefore compressing more or less the spring. The selection of a high pressure level has the added advantage of pre-compressing the preform to a thickness closer to that of the cured part. The preform is then held in this compacted state by the tufts themselves, essentially reducing its bulk factor (defined as the ratio of the thickness of the preform to the thickness of the fully cured composite). A smaller bulk factor lowers, in turn, the risk of the tufts being bent or kinked when the preform is forced in the mould cavity.

While the use of high foot pressure levels has advantages, on very loose fabrics, the dragging action of the foot element can shift the top plies tows significantly from their original position (Figure 9). In extreme cases such an effect can change the local fibre volume fraction of the cured composite. However, an adequate fabric clamping system will avoid fibre distortion and misplacement altogether.

9

Figure 9: Fibre bundles dragged out of the carbon fibre woven preform during tufting with glass fibre thread. The arrows points at the misplaced carbon tows. This effect is due to a combination of loose preform insufficiently clamped to the supporting tool and high pressure from the presser foot.

7. Tufting processIn its standard configuration (i.e. when the needle fully penetrates the preform), the tufting process requires the use of sacrificial material to support the preform. The role of this supporting pad is to accommodate and hold in place the portion of thread forming the loops on the underside of the fabric stack (Figure 3a). A nylon film placed between the backing layer and the preform helps pulling out the loops when the thread insertion is completed. The pad is to be disposed of after a single or very few uses because of the numerous holes punched by the needle during the operation. One of the main aspects in the manufacture procedure development is the choice of the sacrificial material for the preform backing layer. Selection of unsuitable backing materials for a given thread/preform combination can lead to uneven loops formation or, in the worst case, to unsuccessful thread insertion: insufficient grip from the backing layer can cause the thread to be pulled out of its site after insertion, resulting in loop-free areas. On the other hand, a suitable materials combination (to be determined experimentally on a case-by-case basis) may make it possible to obtain loops which are only a few millimetres long and barely appear on the preform underside.

Silicone based materials are in general stiff enough to offer an adequate support and yet sufficiently resilient to exert a good grip on the thread while the needle is withdrawn. The very high elasticity of such materials offers the added advantage of making a single pad reusable several times as the punctures close back once the needle is withdrawn. Among the vast range of options, two different materials were selected. Silastic® 3481 with Silastic® 81T curing agent is a room temperature curing silicone rubber from Dow Corning®. Being available in liquid, uncured form, it can be moulded to the desired shape and represents a good solution when tufting three dimensional, complex shaped preforms. SIL16 is a silicone foam from Samco®; it is softer and cheaper than Silastic® and more suitable for thinner and more delicate threads. It is supplied in sheets of various thicknesses and it is an appropriate solution when tufting flat panels. Despite being effective, the silicone based backing materials have the tendency to release debris in the form of loose particles which can be drawn into the preform by the needle and become embedded in the part. Microscopy on infused and cured composite revealed the presence of silicone particles with a diameter up to 0.5mm within the fibre-free zone around the tuft.

Solid closed cell foams of different nature usually do not suffer from this drawback; different grades have been tested from extruded polystyrene to polyurethane, polyvinylchloride (PVC) or polymethacrylimide (PMI). Airex® R63 types with densities up to 90kgm-3 have been tufted successfully: this is a PVC, damage tolerant foam, which fits very well the tufting setup, and allowed obtaining very good control over the loops length and consistency. This grade of supporting pad provides good stiffness as well as high elongation at break, which allows easy loops removal after tufting, even when the backing foam can only be pulled parallel to the laminate surface, for example when tufting

10

sleeves or pipes. This represents a particularly difficult situation in which very brittle backing materials (like the polystyrene foams) or pads that tear easily (like the silicones) would be unsuitable for the job.

The tuft loop length (measured between the top surface and the reversal point of the thread) appears to be linearly dependent from the needle penetration depth (Figure 10), whether the loop is formed within the preform thickness or in the supporting foam. It is apparent that the needle has to advance between 12 and 14mm further into the preform/support assembly than the intended final loop length. The main sources of such offset are the distance between the needle tip and the needle eye (ineffective in terms of thread deposition) and the partial extraction of the already inserted tuft loops when inserting subsequent tufts.

For a fixed needle depth, glass tufts were typically 1 to 1.5mm longer than carbon thread ones, probably because of the higher flexural stiffness of the two-stranded carbon thread (thread grades are detailed in Table 1).

Figure 10: Tuft length vs. needle penetration depth in 14.5mm thick, biaxial NCF carbon fibre preform supported by Airex® R63.

The use of resin compatible foams becomes particularly relevant if designing closed, foam cored structures where the tuft loops are intentionally formed in the foam to anchor the composite external layer to the core. A relevant application of this concept is in sandwich materials where the tufts can not only improve the skin/core adhesion but also create a truss-like structure between the sandwich skins with advantages for its mechanical performance [45, 46]. A practical application of such concept is in the patented products of NidaCore® (France), now 3M Composite Materials, and Acrosoma® (Belgium): in this case, the loops become integral constituents of the composite part and ‘functional’ to the structure performance. Apart from such niche applications, such a ‘functional’ loop approach is not general practice yet, however, experience indicates that it may represent a viable method to exploit fully the potential of tufting. For such particular applications, a foam grade able to withstand the typical curing temperatures of an epoxy resin should be selected. Rohacell® WF is a PMI rigid foam which has been designed for aerospace applications and can tolerate temperatures of up to 180°C. Grades with densities up to 110kgm -3 have been used successfully as tufting backing material.

The use of honeycomb sheets as a substrate for supporting the laid-up plies during tufting was also tested both on flat and single-curvature preforms. Paper honeycomb is not stiff enough and it collapses locally, whereas the option of aluminium honeycomb proved unfeasible because the thread is sheared very easily when the needle penetrates the dry laminate in proximity of the vertical walls of the honeycomb cell. For obvious geometrical reasons, this effect is more evident on curved preforms.

An alternative option to the use of disposable backing substrates is the adoption of solid supporting rigs, designed to maintain the shape of the preform while locally providing empty volumes on the underside of the preform where the loops can form freely. These tufting rigs can be manufactured by machining slits (or grooves of adequate depth) in the areas where the tufts insertion is planned, on replicas of the metallic moulds used for the next resin injection stage. An example of a solid supporting rig with slits is shown in Figure 4b. Given that the preform remains locally

11

unsupported, this option is only viable when the areas to be tufted are limited. Experience has shown that, especially when relatively thin preforms are being processed with this setup, it is necessary to equip the rig with a rigid and strong clamping system to hold the preform in place and avoid it being pushed into the grooves by the presser foot and the needle. The use of backing materials cannot always be avoided; supporting rigs only work if there is enough interaction between the thread and the dry fabric to lock the former in place. A particularly thin thread together with a very loose preform might not offer enough grip to promote loops formation.

Another option which does not require the use of substrates is the partial reinforcement of the composite. This can be obtained by stopping the needle penetration before the preform underside is reached. The concept is illustrated in Figure 3b and shown in an experimental trial in Figure 11. In this case no external loop is formed, therefore the correct tuft placement can only be confirmed either by monitoring accurately the amount of thread inserted or, after cure, using non-destructive scanning tests. Very careful materials selection is required in order to ensure that a sufficient friction from the fabric releases the thread within the preform thickness. Broadly speaking, this option is only realistically applicable to laminates of substantial thickness (>15mm) and tight enough (because of fibre packing or bindering) to hold the thread in place upon needle withdrawal. No defined set of processing parameters is available today to define the envelope of applicability of such an option; its definition can only be addressed by systematic experimental trials with the specific material combination.

Partial reinforcement opens to the possibility of tufting the preform while it sits in the metallic resin injection mould [47]. Although this represents certainly a desirable manufacturing solution, it should be noted that a few plies on the underside of the component will have to be left necessarily unreinforced. In fact, the tip of the needle is not effective for releasing the thread as the needle eye is placed approximately 5mm back from the shaft tip. This implies that the tuft length is always shorter than the penetration depth and that, in order to insert a tuft through the full preform thickness, a shallow channel has to be machined on the mould surface, to be filled eventually with resin during the following injection stage.

Figure 11: Cross section of tuft loops fully embedded within the thickness of a carbon fibre laminate

Apart from the configurations in which no free loops are formed, an obvious question arises as to whether they should be removed from the panel prior to resin infusion and, if so, how. If the looped side of the part is facing the tool surface during resin impregnation and cure, then the surface finish is not expected to be affected by them. Nevertheless, the manual removal of the tufts both by ordinary scissors and by a commercially available electric trimmer has been attempted and it was proven to be doable (loops were effectively sheared and no tufts were dislodged), but practically unfeasible, being too time demanding. To date no realistic solution has been found to this problem although the market offers a vast array of industrial shearing machines commonly used for carpet production which potentially might represent a commercially viable method of loop removal. Further investigation on this front is required in the future.

12

8. Resin infusion of tufted preformsThe extra yarns will have to be accommodated within the two dimensional fibre architecture of the unreinforced preform and the total fibre content (fabric + thread) is therefore locally increased. Hence, the presence of the tufts and, in general, of a three-dimensional fibre architecture is expected to have an effect on the resin flow and impregnation [48]. When resin transfer moulding (RTM) technologies are involved, given that the cavity size of the mould is fixed, the continuous fibre have to rearrange themselves both within the panel thickness (with the plies more compacted to accommodate surface seams and loops) and across the panel main plane (with the tows being spread apart by the tufts).

Apart from the obvious observation regarding the resin rich layer formed on the looped side of the component, which is believed to have an effect on the bending stiffness of the tufted composite, it was estimated that the extra fibre compaction in RTM cavities can increase locally the fibre volume fraction by up to 18% in the immediate vicinity of the tuft. This has consequences on the resin impregnation of the part; in fact evidence of localised air trapping was found in preform infused by RTM with in-plane resin flow. This can have extreme consequences when relatively large portions of a part are tufted with a short pitch pattern: the high fibre compaction can hinder the resin flow, leading to the formation of large dry spots and poorly impregnated areas (Figure 12a). This effect is quantified in Figure 13, where the applied pressure versus the achieved dry preform thickness is plotted for tufted and untufted (control) samples, made of 32 layers of 0°/90° NCF with an areal weight of 440g/m2. The plot shows that the presence of both 0.5% carbon and glass tufts equally raises the required consolidation pressure of the Z-reinforced preform by up to 40% to achieve the same thickness as the untufted preform for a typical 2D equivalent fabric fibre volume fraction of 55%. This is in agreement with compression results on conventional stitched dry fabric preforms in [49, 50].

Figure 12: Dry spot within a densely tufted area of a cured RTMed panel (a) and cross section of a vacuum infused panel showing increase in thickness in the tufted region (b)

Figure 13: Compression force vs. dry preform thickness for tufted and untufted (control) NCF preform

When the accommodation of the extra thread in the fixed sized cavity mould is likely to pose a problem to the impregnation process, once the minimum effective tuft density is identified, reducing the loops length to the

13

minimum may represent a further way to avoid excessive fibre compaction. Alternatively, it may be necessary to modify the cavity geometry (and, in turn, the cured component size and tolerances) to allow extra room for the tuft loops.

Some evidence of trapped air next to the tufts was found also in preforms impregnated under vacuum infusion process conditions, when the local perturbations to fabric permeability become important, nevertheless macroscopic dry spots were never observed as the bagging film can easily adjust to variations in the preform geometry. However, in this case an increase in thickness must be expected in the tufted regions of the cured part (Figure 12b).

It appears that, when a technology involves the insertion of an extra load bearing, fibrous medium through the thickness of the material, a correct evaluation of the local variations in fibre volume fraction becomes critical. The total fibre content can be estimated on the basis of simple geometric considerations, either assuming a nominal loops length or monitoring the amount of thread effectively inserted. However, the independent evaluation of the reinforcing fibre content and the in-plane fibre content is crucial for the correct interpretation of the mechanical test data. In the case of the complex fibre structure of a tufted laminate there is the further complication of if and how to consider the portion of the thread forming the loops. In terms of the effects on the mechanical properties, it appears more appropriate to consider only the functional portion of the thread, neglecting the surface stitches and the loops.

9. ConclusionsThe technique of robotic through-the-thickness reinforcement of dry carbon fabric stacks by glass, carbon or aramid fibre threads has proved to be relatively easy to introduce in the laboratory environment. Some guidelines for implementing and exploiting the advantages offered by the technology have been outlined and the main practical challenges identified. Apart from the issues of interfacing and programming of the robot arm and the commercial tufting head, the major practical challenges are in ensuring sufficient anchorage of the tuft loops either in the backing material or within the thickness of the tufted preform.

Possible solutions to critical and potentially problematic aspects such as controlling and managing the loops and selecting and configuring raw and ancillary materials have been suggested, which will facilitate the transfer of the technology into industrial and commercial settings.

AcknowledgementsThis study was sponsored by the ADVITAC project, within the European Commission 7 th Framework Programme, and by Cranfield IMRC, Projects 37 and 102.

14

References1. A.J. Kinloch, R.J. Young, Fracture behaviour of polymers, Applied Science Publishers, London, 1983.

2. L.A. Utracki, Polymer blends handbook, Kluwer Academic Publishers, Dordrecht, 2002.

3. R.A. Pearson, A.F. Yee, The effect of cross-link density on the toughening mechanisms of elastomer-modified epoxies, in: Tough composite materials, Noyes Publications, Park Ridge, 1985, pp. 157-177.

4. M. Akay, J.G. Cracknell, Epoxy resin-polyethersulphone blends, J. Appl. Polym. Sci. 52 (1994) 663-688.

5. C.B. Bucknall, A.H. Gilbert, Toughening tetrafunctional epoxy resins using polyetherimide, Polymer 30 (1989) 213-217.

6. L. McKague, Thermoplastic resins, in: D.B. Miracle, S.L. Donaldson, S.D. Henry (Eds.), ASM Handbook, Volume 21: Composites, Revised edition, ASM International, Ohio, 2001, pp. 135-139.

7. R. Griffiths, Boeing sets pace for composite usage in large civil aircraft, High Performance Composites [URL: www.compositesworld.com/articles/boeing-sets-pace-for-composite-usage-in-large-civil-aircraft] (2005).

8. I.K. Partridge, D.D.R. Cartie, Delamination resistant laminates by Z-fiber® pinning: Part I manufacture and fracture performance, Compos. Part A-Appl. S. 36 (2005) 55-64.

9. K.A. Dransfield, C. Baillie, Y.W. Mai, Improving the delamination resistance of CFRP by stitching: A review, Compos. Sci. Technol. 50 (1994) 305-317.

10. A.P. Mouritz, Review of Z-pinned composite laminates, Compos. Part A-Appl. S. 38 (2007) 2383-2397.

11. A.P. Mouritz, M.K. Bannister, P.J. Falzon, K.H. Leong, Review of applications for advanced three-dimensional fibre textile composites, Compos. Part A-Appl. S. 30 (1999) 1445-1461.

12. J. Wittig, Recent development in the robotic stitching technology for textile structural composites, in: A. Falcone, K.M. Nelson, R. Alvers (Eds.), 33rd International SAMPE Technical Conference - Advancing Affordable Materials Technology, Society for the Advancement of Material and Process Engineering, Seattle, WA, 5 - 8 November 2001, p. 540.

13. L.C. Dickinson, G.L. Farley, M.K. Hinders, Translaminar reinforced composites: A review, J. Compos. Tech. Res. 21 (1999) 3-15.

14. L.K. Anon, Z-pins strengthen the super hornet, save weight and cost, The Integrator, 3 (2001) 1-2.

15. M.K. Bannister, Development and application of advanced textile composites, Proc. Inst. Mech. Eng. L J. Mater. Des. Appl. 218 (2004) 253-260.

16. A.P. Mouritz, K. Leong, I. Herszberg, A review of the effect of stitching on the in-plane mechanical properties of fibre-reinforced polymer composites, Compos. Part A-Appl. S. 28A (1997) 979-991.

17. U. Beier, J.K.W. Sandler, V. Altstädt, H. Spanner, C. Weimer, Mechanical performance of carbon fibre-reinforced composites based on stitched and bindered preforms, Compos. Part A-Appl. S. 40 (2009) 1756-1763.

18. M. Dow, B. Dexter, Development of stitched, braided and woven composite structures in the ACT program and at Langley Research Center (1985 to 1997) - Summary and bibliography, Report No. TP-97-206234, NASA, Hampton, VA, 1997.

15

19. A. Baker, S. Dutton, D. Kelly, Composite materials for aircraft structures, 2nd ed., AIAA Education Series, Reston, VA, 2004.

20. A.V. Hawley, Preliminary design of an advanced technology composite wing for a transport aircraft, SAWE 53 rd Annual Conference, Society of Allied Weight Engineers, Inc., Long Beach, CA, 23-25 May 1994.

21. T. Grundmann, T. Gries, M.T. Kordl, B. Corves, Robot-supported joining of reinforcement textiles with one-sided sewing heads, Technische Textilien 49 (2006) 262.

22. K. Klopp, K.U. Moll, B. Wulfhorst, Stitching technology for technical textiles and composites, Technische Textilien 43 (2000) 278.

23. T. Gries, K. Klopp, Füge- und oberflächentechnologien für textilien: Verfahren und anwendungen, Springer Verlag, Berlin, 2007.

24. T. Thurm, Applications of one-sided stitching techniques for resin infusion preforms and structures, SAMPE Journal 41 (2005) 64-67.

25. A.R. Mills, J. Jones, Investigation, manufacture, and testing of damage-resistant airframe structures using low cost carbon fibre composite materials and manufacturing technology, Proc. Inst. Mech. Eng. G J. Aerosp. Eng. 224 (2010) 489-497.

26. A.P. Mouritz, B.N. Cox, A mechanistic interpretation of the comparative in-plane mechanical properties of 3D woven, stitched and pinned composites, Compos. Part A-Appl. S. 41 (2010) 709-728.

27. G. Dell'Anno, D.D. Cartie, I.K. Partridge, A. Rezai, Exploring mechanical property balance in tufted carbon fabric/epoxy composites, Compos. Part A-Appl. S. 38 (2007) 2366-2373.

28. J.W.G. Treiber, Performance of tufted carbon fibre/epoxy composites, PhD dissertation, Cranfield University, Cranfield (UK) (2011).

29. M. Colin de Verdiere, A.K. Pickett, A.A. Skordos, V. Witzel, Evaluation of the mechanical and damage behaviour of tufted non crimped fabric composites using full field measurements, Compos. Sci. Technol. 69 (2009) 131-138.

30. G. Dell'Anno, Effect of tufting on the mechanical behaviour of carbon fabric/epoxy composites, PhD dissertation, Cranfield University, Cranfield (UK) (2007).

31. A. Ogale, P. Mitschang, Tailoring of textile preforms for fibre-reinforced polymer composites, J. Ind. Text. 34 (2004) 77-96.

32. E.E. Theotokoglou, T. Moan, Experimental and numerical study of composite T-joints, J. Compos. Mater. 30 (1996) 190-209.

33. S.S. Suh, N. Han, J.M. Yang, H.T. Hahn, The effect of stitching on damage tolerance of stiffened composite panel, in: Proceedings of the American Society for Composites, 17th Technical Conference, CRC Press, West Lafayette, IN, 2002.

34. P.B. Stickler, M. Ramulu, P.S. Johnson, Experimental and numerical analysis of transverse stitched T-joints in bending, Compos. Struct. 50 (2000) 17-27.

35. P.B. Stickler, M. Ramulu, Investigation of mechanical behavior of transverse stitched T-joints with PR520 resin in flexure and tension, Compos. Struct. 52 (2001) 307-314.

16

36. P.B. Stickler, M. Ramulu, Parametric analyses of stitched composite T-joints by the finite element method, Mater. Design 23 (2002) 751-758.

37. G. Freitas, C. Magee, P. Dardzinski, T. Fusco, Fiber insertion process for improved damage tolerance in aircraft laminates, J. Adv. Mater. 25 (1994) 36-43.

38. K.L. Rugg, B.N. Cox, R. Massabó, Mixed mode delamination of polymer composites laminates reinforced through the thickness by z-fibers, Compos. Part A-Appl. S. 33 (2002) 177-190.

39. D.D.R. Cartie, G. Dell'Anno, E. Poulin, I.K. Partridge, 3D reinforcement of stiffener-to-skin T-joints by Z-pinning and tufting, Eng. Fract. Mech. 73 (2006) 2532-2540.

40. E. Greenhalgh, M. Hiley, The assessment of novel materials and processes for the impact tolerant design of stiffened composite aerospace structures, Compos. Part A-Appl. S. 34 (2003) 151-161.

41. J. Russell, Market trends: The composites affordability initiative: Part I, High Performance Composites [URL: www.compositesworld.com/columns/market-trends-the-composites-affordability-initiative-part-i] (2007).

42. V. Koissin, J. Kustermans, S.V. Lomov, I. Verpoest, B. Van Den Broucke, V. Witzel, Structurally stitched NCF preforms: Quasi-static response, Compos. Sci. Technol. 69 (2009) 2701-2710.

43. A.P. Mouritz, B.N. Cox, A mechanistic approach to the properties of stitched laminates, Compos. Part A-Appl. S. 31 (2000) 1-27.

44. E. Chehura, G. Dell'Anno, T. Huet, S. Staines, S. James, I.K. Partridge, R. Tatam, On-line monitoring of multi-component strain development in a tufting needle using optical fibre Bragg grating sensors, Submitted to Smart Mater. Struct.

45. A. Henao, M. Carrera, A. Miravete, L. Castejón, Mechanical performance of through-thickness tufted sandwich structures, Compos. Struct. 92 (2010) 2052-2059.

46. P. Potluri, E. Kusak, T.Y. Reddy, Novel stitch-bonded sandwich composite structures, Compos. Struct. 59 (2003) 251-259.

47. J. Wittig, In-mold-reinforcement of preforms by 3-dimensional tufting, in: Proceedings of the 47 th International SAMPE symposium and exhibition, Society for the Advancement of Material and Process Engineering, Long Beach, CA, 12 - 16 May 2002.

48. N. Patel, L. James Lee, Modeling of void formation and removal in liquid composite molding, Part II: Model development and implementation, Polym. Composite. 17 (1996) 104-114.

49. M. Schuster, A. Ogale, L. Peetz, J. Schuster, P. Mitschang, Analysis of sewed preforms by visual on-line monitoring of stitch-hole variations under compaction, Compos. Sci. Technol. 68 (2008) 312-320.

50. A. Ogale, P. Mitschang, Compaction behavior of assembled fiber reinforced preforms, J. Ind. Text. 37 (2007) 15-29.

17