WO2022058702A1

WO2022058702A1 - Method of manufacturing fibre reinforced composite structures

Info

Publication number: WO2022058702A1
Application number: PCT/GB2020/052273
Authority: WO
Inventors: Lorenzo Mencattelli; Silvestre Taveira PINHO
Original assignee: Imperial College Innovations Ltd
Current assignee: Ip2ipo Innovations Ltd
Priority date: 2020-09-18
Filing date: 2020-09-18
Publication date: 2022-03-24
Anticipated expiration: 2023-03-18

Abstract

The present disclosure relates to a method of manufacturing a fibre reinforced composite structure, the method comprising: laying up a plurality of plies of fibre reinforced composite material, wherein each of the plurality of plies comprises a plurality of deviations in the thickness direction of the ply; and curing the plurality of plies.

Description

METHOD OF MANUFACTURING FIBRE REINFORCED COMPOSITE STRUCTURES

FIELD

The present disclosure relates to a method of manufacturing fibre reinforced composite structures.

BACKGROUND

Fibre reinforced composite materials have poor damage resistance, resulting from their inherent brittleness and tendency to fail in a potentially catastrophic and harmful manner. Specifically, fibre reinforced composite materials have poor impact resistance.

Accordingly, there exists a need for improving the damage resistance of fibre reinforced composite materials.

SUMMARY

This summary introduces concepts that are described in more detail in the detailed description. It should not be used to identify essential features of the claimed subject matter, nor to limit the scope of the claimed subject matter.

According to one aspect of the present disclosure, there is provided a method of manufacturing a fibre reinforced composite structure, the method comprising: laying up a plurality of plies of fibre reinforced composite material, wherein each of the plurality of plies comprises a plurality of deviations in the thickness direction of the ply; and curing the plurality of plies.

The plurality of deviations in the thickness direction of the plies provides improved damage resistance. Specifically, the plurality of deviations in the thickness direction results in interfaces between adjacent plies that provide resistance to the development of delamination damage and the formation of mechanical interlocks between adjacent plies, the diffusion of damage and the containment of damage within the region of the plurality of deviations, resulting in enhanced structural integrity.

Laying up the plurality of plies may comprise laying up the plurality of plies such that the alignment of fibres in one of the plurality of plies is angled with respect to the alignment of fibres in an adjacent one of the plurality of plies.

The method may further comprise forming the plurality of deviations in the thickness direction of each of the plies. The plurality of deviations in the thickness direction may be periodic. The periodic deviation may be provided in two orthogonal directions in each ply. The periodic deviation may be sinusoidal.

Forming the plurality of deviations in the thickness direction of each of the plies may comprise placing the plurality of plies on a mould comprising a plurality of protrusions. The plurality of protrusions may be confined to a specific region of the mould. The plurality of protrusions may be periodic (e.g. sinusoidal).

An amplitude and a distribution (e.g. period) of the plurality of protrusions is selected to maximise the out-of-plane deviation in the thickness direction, wherein the selection of the amplitude and the distribution is based on a fibre type and a manufacturability constraint of the mould. Larger values of the ratio of amplitude to period correspond to structures with higher damage resistance.

Each of the plies may comprise a matrix material, and the method may further comprise pre-heating the mould to a temperature to cause a decrease in the viscosity of the matrix material prior to laying up the plurality of plies on the mould. For example, the mould may be pre-heated to a temperature that allows the viscosity of the matrix material to decrease to between about 10⁷ to about 10⁸ cP. Pre-heating the mould at this temperature minimises fibre misalignment and maximises the mouldability of the plies.

The plurality of deviations may extend over an area of each of the plurality of plies. Extension of the deviations over an area increases the mechanical interlocking between fibres in adjacent plies.

The alignment of fibres in each of the plurality of plies may be angled with respect to the alignment of fibres in an adjacent one of the plurality of plies. The angle between adjacent plies provides a dissipative failure mechanism. Each fibre of each ply may be rotated about the longitudinal axis of the structure at an angle relative to the fibres in the adjacent ply, each extending longitudinally over the composite structure.

The angle between the alignment of the fibres in adjacent ones of the plurality of plies may be between 0° and 360°, more particularly between 2° and 20°.

Curing the plurality of plies may comprise curing the plurality of plies while the ply at a first face of the plurality of plies is uncovered. Leaving one face of the structure uncovered during the curing process allows for a straightening of the fibres at the uncovered face, thereby providing a gradual variation in the amplitude of the periodic deviations from one face of the structure to the other face. The gradual variation (e.g. reduction) in amplitude leaves fibres on the tensile side of the laminate straight. This increases their capability of carrying load under impact without incurring an early failure.

Curing the plurality of plies may comprise curing the plurality of plies while the ply at a first face of the plurality of plies is covered by a similar mould having a plurality of protrusions. This allows a structure to be created with constant amplitude of the deviations in the thickness direction.

The method may further comprise forming an impact layer over the ply at one face of the plurality of plies. The impact layer may comprise one or more of: a hard polymer, a soft polymer, and a ceramic material. The impact layer may be toughened with particles having a characteristic length at the nano- to the micro-scale. The toughening particles may be a different material to the material of the impact surface. Where a first face of the plurality of plies is uncovered during curing, the impact layer may be formed over the ply at a second face of the plurality of plies, wherein the second face is opposite the first face. The impact layer provides a tough surface which dissipates energy on the contact surface during loading.

Each of the plies may comprise a matrix material, and curing the plurality of plies may comprise curing the plies at a temperature that allows the matrix material to reach a viscosity of between 10⁷ and 10⁸ cP. The plies may be cured at this temperature for between thirty minutes and one hour (more particularly, for one hour) to maximise the moulding of the uncured plies before curing. Laying up the plurality of plies may comprise: splitting the plurality of plies into a first plurality of groups of plies; and for each of group of plies in the first plurality of groups of plies: placing plies in the group of plies on a mould; and moulding the plies in the group of plies. For each group of plies in the first plurality of groups of plies, moulding the plies in the group of plies comprises placing the group of plies under vacuum for two to ten minutes. The moulding time of two to ten minutes minimises fibre misalignment and maximises the mouldability of the plies. For each group of plies in the first plurality of groups of plies, moulding the plies in the group of plies may comprise compressing the group of plies using an indenter. This improves the quality of the moulding of the plies. Each group of plies in the first plurality of groups of plies may comprise a number of plies that allows for maximum mouldability and minimum fibre misalignment (e.g. six or seven plies).

For each of the plurality of plies, the periodic deviation is located in a first region of the ply, the ply further comprising a second region with no deviation in the thickness direction of the ply. This allows the periodic deviations to be included in specific regions of the structure, meaning that improved damage resistance can be provided in specific locations that may be subject to through-the-thickness loads. The periodic deviations also provide a smooth transition between the laminate region with no deviations in the ply thickness direction and the laminate region that includes the deviations in the ply thickness direction. This prevents damage from being localised at the periphery of the region with the periodic deviations.

According to another aspect of the present disclosure, there is provided a fibre reinforced composite structure obtained by the method described in the above paragraphs. The fibre reinforced composite structure may be a carbon fibre reinforced plastic. Alternatively, the fibre reinforced composite may comprise glass fibre, aramid fibres, nylon fibres, acrylic fibres, natural fibres, basalt fibres, polyethylene fibres, polyparaphenylene fibres, benzobisoxazole fibres, polypropylene fibres, polybenzamidazole fibres, or any combination of the foregoing materials. The fibre reinforced composite structure provides improved damage resistance and structural integrity through the features described above.

BRIEF DESCRIPTION OF FIGURES Specific embodiments are described below by way of example only and with reference to the accompanying drawings, in which:

FIG. 1 is a schematic diagram of a Herringbone microstructure.

FIG. 2 is a schematic diagram of two design requirements for a Herringbone-Bouligand region.

FIG. 3 is a graphical representation of the design requirements shown in FIG. 2.

FIG. 4 is a flow diagram of a method of manufacturing a fibre reinforced composite structure.

FIG. 5 is a schematic diagram of a manufactured fibre reinforced composite structure.

FIG. 6 is a schematic diagram of a Bouligand microstructure.

FIG. 7 shows a mould used for manufacturing a Herringbone-Bouligand fibre reinforced composite structure.

FIG. 8A is a flow diagram of a method of manufacturing a fibre reinforced composite structure according to one example.

FIG. 8B shows pseudo-code used for carrying out the method of FIG. 8A

FIG. 9 shows a pre-moulding set-up used in the method of FIG. 8A.

FIG. 10 shows a curing cycle used in the method of FIG. 8A.

FIG. 11 shows a cured fibre reinforced composite laminate produced using the method of FIG. 8A.

FIG. 12 shows a set-up for creating an impact surface according to the method of FIG. 8A.

FIG. 13 shows a cured fibre reinforced composite laminate structure having an impact surface, produced using the method of FIG. 8A. FIG. 14 shows a Herringbone-Bouligand region within a fibre reinforced composite sample, along with a cross-section through the Herringbone-Bouligand region.

FIG. 15 shows an additional cross-section through the Herringbone-Bouligand region shown in FIG. 14.

FIG. 16 shows a further cross-section through a portion of the Herringbone-Bouligand region shown in FIG. 14.

FIG. 17 is a graph of load against displacement for a Herringbone-Bouligand sample and a Bouligand sample.

FIG. 18 is a graph of peak load against total dissipated energy for a Herringbone- Bouligand sample and a Bouligand sample.

FIG. 19 is a graph of dent diameter for different tests carried out on a Herringbone- Bouligand sample and on a Bouligand sample.

FIG. 20 is a graph of projected delamination area for different tests carried out on a Herringbone-Bouligand sample and on a Bouligand sample.

FIG. 21 shows photographs of the indentation face, back face and delamination damage for different tests carried out on a Bouligand sample.

FIG. 22 shows photographs of the indentation face, back face and delamination damage for different tests carried out on a Herringbone-Bouligand sample.

FIG. 23 shows cross-sections through a Herringbone-Bouligand sample and a Bouligand sample tested up to an applied load of 4 kN.

FIG. 24 shows photographs of the back face of Bouligand and Herringbone-Bouligand samples at the end of a full penetration test.

FIG. 25 is a graph of total dissipated energy for a quasi-isotropic (QI) carbon fibre reinforced plastic (CFRP) laminate, a Bouligand CFRP laminate and a Herringbone- Bouligand CFRP laminate. FIG. 26 is a graph of peak load for a QI CFRP laminate, a Bouligand CFRP laminate and a Herringbone-Bouligand CFRP laminate.

DETAILED DESCRIPTION

As used herein, the term “fibre reinforced composite” is used to refer to a series of fibrous plies stacked up to form a laminate, where each ply may comprise fibres of various lengths and elastic moduli embedded in a matrix with lower elastic modulus compared to the fibres. The term “fibre reinforced composite” may also be used to refer to a stack up of a series of plies in the form of mats with chopped fibres with lengths ranging from 1 mm to 50 mm randomly distributed or with a preferred orientation embedded in a matrix with lower elastic modulus compared to the fibres.

FIG. 1 is a schematic diagram of a Herringbone microstructure. As used herein, the term “Herringbone” is used to refer to a microstructure in which the fibres in each ply include a deviation in the out-of-plane (i.e. thickness) direction, which may be a periodic deviation.

The Herringbone microstructure shown in FIG. 1 includes a plurality of plies, where fibres in a ply in the structure are aligned at an angle A0 in the x-y plane to fibres in an adjacent ply. However, the fibres in each ply of the structure in FIG. 1 also include a periodic out-of-plane (z-thickness direction) fibre component. This results in a plurality of wavy plies (as shown in FIG. 1). The wavy architecture defines bi-sinusoidal interfaces (i.e. sinusoidal interfaces in the x- and y-directions) between neighbouring plies that can be expressed as:

(Equation 1) where A is the amplitude of the Herringbone (periodic) pattern, and A is the wavelength (or period) of the Herringbone pattern.

The Herringbone microstructure shown in FIG. 1 provides a mechanism for stress redistribution, leading to a more uniform through-the-thickness stress state and a higher compressive stiffness. Implementations of the present disclosure provide a method of manufacturing a fibre reinforced composite structure that incorporates improved impact resistance provided by a Herringbone region.

Implementations of the present disclosure are specifically described with reference to carbon-fibre reinforced plastics (CFRPs). However, it will be appreciated that the implementations disclosed herein are also applicable to other fibre reinforced composite structures, such as glass fibre reinforced composites, and materials with different fabric types, layup sequences, fibre types and resin types.

In an example in which the plurality of protrusions is periodic (e.g. as in FIG. 1), two main parameters are used in the manufacture of the Herringbone region. These are: the amplitude A of the Herringbone pattern (as used in Equation 1), and the wavelength A of the Herringbone pattern (as used in Equation 1).

The choice of A and A is driven by two requirements shown in FIGS. 2 and 3. Firstly, the maximum bending strain of the fibre material (e.g. carbon fibre) during the moulding process should not exceed the respective failure strain of the fibre material. This is given by Equation 2:

(Equation 2) where £_f is the failure strain of the fibre material and p is the radius of the fibre material. As one specific example, the values for carbon fibre are £_f = 1.9% and p = 3.4 pm.

The amplitude A (i.e. peak-to-peak amplitude) and wavelength A (or period) of the fibres are schematically shown in FIG. 2. The upper bound on A against A given by Equation 2 is shown as a solid line in FIG. 3.

Secondly, the mould used to shape the plies must be manufacturable. That is, there must be an available tool size to mill the mould, in the event that the mould is manufactured using computer numerical control (CNC). Alternatively, if the mould is manufactured using an etching technique, the tool size will refer to the size of the diode used in the etching procedure. This requirement is given by Equation 3:

(Equation 3) where r is the radius of a cylindrical tool used to mill the Herringbone mould.

The requirement for mould manufacturability is shown schematically in FIG. 2, with the dashed line in FIG. 3 showing the bound on A against A given by Equation 3. As shown in FIG. 3, the mould manufacturability restricts the value of A to a greater extent than the material properties.

In one example (described further below), values of A = 2.5 mm and A = 5 mm (i.e. a ratio of A/A = 0.5) were used. These values were used in the example below to maximise the ratio of amplitude to wavelength that could be achieved given the constraints on mould manufacturability and fibre type.

FIG. 4 is a flow diagram of a method of manufacturing a fibre-reinforced composite structure. At 500, a plurality of plies are oriented to the angles used in the desired stacking sequence of the plies and cut to the desired shape and orientation, which will depend on the chosen stacking sequence.

At 502, the plurality of plies is divided into a first number of groups of plies. Each group of plies includes a similar number of plies.

At 504, a first one of the first number of groups of plies is laid up. In one example, an alignment mould is used to guarantee precision during the layup.

At 506, the first one of the first number of groups of plies is placed on a mould. The mould comprises a region in which a plurality of periodic protrusions is provided. For example, the mould may comprise a sinusoidal pattern of protrusions in an x-z-plane and a sinusoidal pattern of protrusions in a y-z-plane (together, a bi-sinusoidal pattern). The mould may be preheated prior to placement of the first group of plies on the mould, in order to soften the plies. For example, the mould may be heated to a temperature that allows the viscosity of the matrix material to decrease to between about 10⁷ to 10⁸ cP. The group of plies may be placed between two Teflon films and subsequently placed on the mould. At 508, the first one of the first number of groups of plies is moulded. In one example, the group of plies is covered using a breathing-cloth with a central hole. The central hole is positioned to leave the ply area above the periodic protrusion region of the mould uncovered. Moulding the group of plies may comprise placing the covered group of plies under vacuum for a period of time (e.g. two to ten minutes). During application of the vacuum, an indenter may be mechanically pressed down on the uncovered region of the group of plies to compress the first group of plies. This improves the quality of the mould in the uncovered region. Multiple passes with the indenter may be implemented. For example, the moulding process may comprise two passes with the indenter.

At 510, the first one of the first number of groups of plies is rapidly cooled, in order to stabilise the region moulded by the periodic protrusion region of the mould. This region of the mould gives rise to a plurality of periodic deviations in the thickness direction of each of the plies in the group of plies.

At 512, the manufacturing process in 504 to 510 is repeated for the other ply groups in the first number of groups of plies, thereby producing a first number of pre-moulded groups of plies.

At 514, a plurality of the first number of pre-moulded groups of plies are laid up to form a second number of groups of plies.

At 516, the manufacturing process in 504 to 512 is carried out for each of the groups of plies in the second number of groups of plies, thereby producing a second number of pre-moulded groups of plies.

At 518, a plurality of the second number of pre-moulded groups of plies are laid up to form a third number of groups of plies.

At 520, the manufacturing process in 504 to 512 is carried out for each of the groups of plies in the third number of groups of plies, thereby producing a third number of premoulded groups of plies.

At 522, the third number of groups of plies are laid up in sequence to form a stack of pre-moulded groups of plies. At 524, the manufacturing process in 504 to 512 is carried out for the stack of premoulded groups of plies, thereby producing a pre-moulded laminate.

At 526, the stack of pre-moulded groups of plies is cured. In one example the stack of pre-moulded groups of plies is cured in the mould. Curing the stack of pre-moulded ply groups may be implemented using an autoclave. The curing process may comprise leaving the upper face of the laminate (i.e. the face opposite the laminate face in contact with the mould) free. Leaving the upper face of the laminate free encourages a gradual transition in amplitude A of the periodic deviations, from a maximum value in the ply that interfaces with the mould, to a minimum value (e.g. zero) at the upper face. In one example, the curing cycle comprises: (i) a first curing stage at a temperature of 50 °C and a pressure of 5 bar, for one hour; (ii) a second curing stage at a temperature of 80 °C and a pressure of 5 bar, for 30 minutes; and (iii) a third curing stage at a temperature of 125 °C and a pressure of 5 bar, for 1.5 hours.

At 528, the cured laminate is cut to a final shape. For example, the cured laminate may be cut using a waterjet.

At 530, an impact surface is created on the laminate. For example, the impact surface may be created on the surface of the laminate with maximum deviation in the thickness direction of the plies (i.e. the laminate face that was cured in contact with the mould). In alternative examples, the impact surface may be created on the opposite face (i.e. the laminate face that was left free during curing). The impact surface may be created by spreading a layer of toughened-epoxy adhesive on the surface of the laminate, and subsequently curing the adhesive under vacuum at room temperature.

FIG. 5 is a schematic diagram of a manufactured fibre reinforced composite structure in the form of a laminate. The laminate shown in FIG. 5 includes a region in which the plies have a periodic deviation in the thickness direction (z-direction in FIG. 5) of the plies (i.e. a “Herringbone region”) and a region in which there is no deviation in the thickness direction of the plies. FIG. 5 also shows the decreasing amplitude of the periodic deviations from a maximum value at the laminate face in contact with the impact surface to a minimum value at the opposite face of the laminate. As shown in FIG. 5, the impact surface has been created using a toughened epoxy layer spread over the cured laminate. As discussed in detail in the example below, the microstructure shown in FIG. 5 provides improved damage tolerance compared to a microstructure in which there is no deviation in the thickness description (e.g. as shown in FIG. 6, described further below).

Specifically, the microstructure shown in FIG. 5 provides: (i) highly reduced in-plane spreading of damage; (ii) successful containment of damage (including fibre failure, delaminations and matrix cracks) within the Herringbone region; (iii) an increase in peak load, penetration load and total dissipated energy, with a more significant through-the-thickness diffusion of damage; (iv) higher resistance to delaminations, resulting in delayed onset of delamination damage; and (v) a higher degree of through- the-thickness sub-critical damage diffusion, achieved through: (a) the dissipation, by the tough impact surface, of energy on the contact area between the indenter (used in the damage tolerance tests) and the laminate; and (b) the tough bi-sinusoidal interfaces between consecutive plies (i.e. the Herringbone pattern), which resulted in higher resistance to the development of delamination damage and in the formation of mechanical interlocks between neighbouring plies. These important damage-tolerant mechanisms promoted the formation of disconnected delamination areas, deflected delaminations and the accumulation of sub-critical matrix cracks.

In summary, the microstructure shown in FIG. 5 is capable of diffusing damage, resisting delamination and containing failure within the region of periodic deviations in the ply thickness direction (i.e. the Herringbone region), while increasing the load bearing capacity and the energy dissipation capability of the structure. Therefore, the laminate structure shown in FIG. 5 provides a tailorable damage-tolerant solution for applications requiring resistance to through-the-thickness loads. Specifically, regions of periodic deviations in the ply thickness direction can be incorporated into a laminate structure at locations that may be subject to through-the-thickness loads. Given that the region of periodic deviations in the ply thickness direction is capable of containing damage, damage can be prevented from propagating into neighbouring regions of the structure, thereby allowing the neighbouring regions to remain undamaged. Therefore, a barrier to damage propagation is provided, which improves the safety of structures in which the composite laminate is incorporated. In addition, the containment of damage means that laminate structures that include region of periodic deviations in the ply thickness direction can be more efficiently and less expensively repaired in the event of a damage event. In addition, the manufacturing procedure described with reference to FIG. 5 can be used to locally tailor the microstructure of prepreg-based composite materials. This means that improved damage resistance can be provided in specific locations, as mentioned in the preceding paragraph. This technique can be integrated in an automated manufacturing process such as automatic tape placement or additive manufacture.

Variations or modifications to the systems and methods described herein are set out in the following paragraph. It will be appreciated that the manufacturing process set out above is described by way of example only and that various modifications may be made to the process for manufacturing a laminate structure such as that shown in FIG. 5. For example, a different number of pre-moulding stages (at 502 to 524) may be used, or a single pre-moulding stage may be implemented (in which all plies are premoulded together at 502 to 510). In addition, the periodic deviations in the thickness direction may be implemented using other manufacturing techniques besides moulding (such as additive manufacture and automatic tape placement (ATP)), as will be appreciated by the skilled person.

In another example, the moulding process used to imprint in the fibre reinforced composite the plurality of deviations in the laminate thickness direction may be carried out via two rotating rolls in which a plurality of protrusions used to mould the plies are engraved. The rolls may be provided at a temperature that allows the plurality of plies to be softened and allow for increased mouldability. The plurality of plies may be drawn in between the two rotating rolls. One of the rolls may impose a pressure on the plurality of plies (similar to the process at 508) as they are drawn in between the rolls to imprint the plurality of deviations in the laminate thickness direction.

In yet another example, the two pressing rolls may be substituted by a press where one of the faces may be flat and the opposite face may have a plurality of protrusions engraved on its surface. In another example, both faces of the press may include a plurality of protrusions. The press may be provided at a temperature to soften the plurality of plies and allow for increased mouldability. The plurality of plies may be placed between the faces of the press. The press may be closed, and a pressure applied to imprint the plurality of deviations in the laminate thickness direction.

In yet another example, the plurality of plies may comprise a preform made of a plurality of dry-fibre (no resin) fabrics stacked together and stabilised through use of a binding agent or a thermoplastic veil. The plurality of plies may be subsequently moulded using processes 502-524 as well as using the examples presented above (rolls, press). The moulded plurality of dry plies may be infused with resin using a Resin Transfer Molding (RTM) technique with a thick silicone counter mould in which the plurality of protrusions are engraved to imprint in the final fibre reinforced composite. In yet another example, the mould could be made of a foam material such as those used in sandwich structures. The foam core of the sandwich structure may be machined to have the plurality of protrusions engraved, A plurality of plies may be stacked up on the machined foam core which would also act as the mould to imprint the plurality of deviations in the laminate thickness. The moulded plurality of plies may be cured together with the machined foam mould which, in this example, would become an integral part of the final sandwich structure. In yet another example, similar to the example given for the sandwich structure, the mould may be a thick coating of tough material with a plurality of protrusions engraved on one face and a smooth and flat opposite face. The pluralities of plies may be moulded (processes 502-524) on the thick coating mould and cured along with it. In this example the mould would become an integral part of the structure by serving as a thick and tough impact surface.

As a further alternative, the plurality of deviations in the thickness direction may not be sinusoidal. Further, the plurality of deviations in the thickness direction may not be periodic. The advantages described above can be achieved using deviations in the thickness direction of the ply to form mechanical interlocks between adjacent plies. Alternative forms of deviations in the ply thickness direction will be apparent to the skilled person, and may be manufactured using the example methods described above.

Example

Helicoidal lamination sequences such as Bouligand-inspired architectures have been investigated in recent years, leading to successful attempts to enhance the damage tolerance to through-the-thickness loads of fibre reinforced composite materials, including glass, aramid, polyolefin and carbon fibre reinforced composites. As used herein, the term “Bouligand” is used to refer to a microstructure in which the fibres in a ply in a structure are aligned at an angle (in the plane of the ply) to the fibres in an adjacent ply in the structure with the difference between adjacent ply orientation, called herein pitch angle, being constant. This therefore provides a helicoidal layup. An example of a Bouligand microstructure is shown in FIG. 6. As shown in FIG. 6, the Bouligand microstructure includes a plurality of plies. Each ply includes a plurality of aligned co-planar fibres. The fibres in a first ply in the structure are aligned at an angle to the fibres in an adjacent ply in the structure. For example, the fibres in the bottom ply in the stack shown in FIG. 6 are aligned at an angle to the fibres in the next-from- bottom ply in the stack. The angle between fibres in adjacent plies in the structure is indicated as A0. Bouligand structures provide a highly-dissipative failure mechanism. These have been successfully exploited in CFRPs.

With the fibre reinforced composite used in the following example, a pitch angle A0 of 2.5° has been shown to achieve the greatest enhancement in damage tolerance to through-the-thickness loads with respect to existing quasi-isotropic layups. The choice of pitch angle may range between 2° to 20° for other fibre reinforced composites.

In the following example, a tool size of r_min = 0.5 mm was used. A ratio of A to A of A/A = 0.5 was also adopted. This resulted in values of A = 2.5 mm and A = 5 mm, which are compatible with the overall thickness of the Herringbone-Bouligand CFRP laminate (i.e. do not exceed the total thickness of the laminate).

In the following example, the mould was designed such that the Herringbone-Bouligand microstructure was included only in the central part of the sample. Specifically, the Herringbone-Bouligand region in the central part of the sample in the example below extends for an area of 50 mm by 50 mm. The microstructure surrounding the Herringbone-Bouligand region was a Bouligand layup with pitch angle A0 = 2.5°( best performing pitch angles may range between 2° to 20° for other fibre reinforced composites). By including a Herringbone-Bouligand region only in the central part of the sample, a point-by-point tailorable Herringbone-Bouligand microstructure was provided.

FIG. 7 shows an aluminium Herringbone mould with a high-quality finish, which was used to produce the Herringbone-Bouligand region within the sample.

In this example, the Bouligand and Herringbone-Bouligand samples were manufactured using Skyflex LISN20A, a unidirectional prepreg tape with areal weight of 20 gsm. The prepreg constituents are TR30S 3K carbon fibres manufactured by Mitsubishi and K50 epoxy matrix manufactured by SK Chemicals. In addition, Scotch- Weld (RTM) EC-9323, a high-impact-resistant epoxy-based adhesive, was used to create the impact surface of the Herringbone-Bouligand region.

FIG. 8A is a flow diagram of a method used for manufacture of a fibre-reinforced composite structure according to the present example. Specifically, the method shown in FIG. 8A is used to manufacture a Herringbone-Bouligand region within a fibre- reinforced composite structure such as CFRP. The pseudo-code used in the method of FIG. 8A is shown schematically in FIG. 8B.

The methods shown in FIGS. 8A and 8B use a stacking sequence which includes an alignment angle A0 between fibres in adjacent plies. Therefore, a first ply in the stacking sequence is oriented such that the fibres in the first ply are aligned at 0 = 0°. The first ply is then cut to the desired shape of the composite structure (for example, a square panel). The second ply in the stacking sequence is oriented such that the fibres in the second ply are aligned at an angle A0 to the fibres in the first ply. For example, if A0 = 2.5°, then the fibres in the second ply are aligned at 2.5°. The second ply is then cut to the desired shape (which may be the same shape as the first ply, such as a square panel). The third ply in the stacking sequence is oriented with fibre orientation 2A0 (e.g. 5°) to the first ply. The third ply is then cut to the desired shape. The process is repeated until all plies in the stacking sequence have been cut to the desired shape.

The number of plies in the stacking sequence may be chosen such that the fibres in the uppermost ply in the sequence are aligned in the same direction as the fibres in the lowermost ply in the sequence. This can be achieved by using a number of plies such that the angle between the fibres in the uppermost and lowermost plies is 180°. For example, where A0 = 2.5°, 73 plies may be used. Alternatively, the stacking sequence may be continued in order to provide a number of Bouligand units (where the lowermost and uppermost plies in each unit are aligned in the same direction). For example, where A0 = 2.5°, 145 plies may be used in order to provide two Bouligand units, where the fibres in the 1^st, 73^rd and 145^th are aligned in the same direction.

In this example, the method uses 145 plies laid up with a pitch angle A0 = 2.5° between adjacent plies. Specifically, the 145 plies were laid up in a stacking sequence (0°, 2.5°, ... , 177.5°), 180°, (177.5°, 175°, ... , 0°). At 802, each ply in the stacking sequence was oriented to the angle used in the stacking sequence and then cut to the desired shape using an automatic cutting table.

At 804, the 145 plies were laid up into 24 groups of six or seven adjacent plies per group, by breaking down the stacking sequence into 24 parts. An alignment mould was used to guarantee excellent precision during the layup.

Sublaminates ^s? were then produced using a pre-moulding procedure. Specifically, 24 sublaminates ^s? were produced, using the 24 groups of plies. For each sublaminate ^s; the following procedure was used:

At 806, the Herringbone mould shown in FIG. 7 was pre-heated to 50 °C. The co sublaminate was then packed between two thin Teflon films and placed on the mould. The mould was pre-heated to soften the stacks of uncured prepreg.

At 808, the sublaminate

was covered using a breathing-cloth with a central hole, to leave the Herringbone-Bouligand region uncovered, as shown in FIG. 9.

„o

At 810, the covered sublaminate ^&i was placed under vacuum for 10 minutes. During application of the vacuum at 810, an indenter was mechanically pressed down on the co sublaminate ^sr at 812, to improve the quality of the mould in the Herringbone- Bouligand region. Two passes with the indenter were used at 812. An example indenter is shown in FIG. 9.

„o

At 814, the sublaminate was rapidly cooled immediately after moulding, in order to stabilise the moulded Herringbone pattern.

The values of moulding temperature (50 °C), moulding time (10 minutes) and the „0 number of plies used in each sublaminate (six or seven) were chosen to minimise fibre misalignment (due to fibre spreading) and maximising the mouldability of the sublaminates.

„o

At 816, the 24 pre-moulded sublaminates were laid up in adjacent pairs to form 12 sublaminates ^s« ,

and the pre-moulding procedure from 806 to

814 was repeated to produce 12 pre-moulded sublaminates

At 820, the 12 pre-moulded sublaminates were laid up in adjacent pairs to form six sublaminates

, and the pre-moulding procedure from 806 to 814 was repeated to produce six pre-moulded sublaminates

.

At 822, the six sublaminates

were laid up in sequence to form a laminate s, where s — is^sj.sg.s^sg.s^j _anc| t g pre-moulding procedure from 806 to 814 was repeated to produce a pre-moulded laminate.

At 824, the pre-moulded laminate was cured in-mould in an autoclave. While the mould was on the bottom side of the laminate during curing, the upper side of the laminate was left free. This allowed for straightening of the outermost CFRP layers, thereby leading to a gradual transition in amplitude A of the Herringbone pattern from a maximum value (at the interface with the mould) to zero (at the back face of the laminate). The curing cycle at 824 comprised an initial curing step at a temperature of 50 °C and a pressure of 5 bar for one hour, followed by the manufacturer recommended curing cycle (0.5 hours at a temperature of 80 °C and a pressure of 5 bar followed by 1 .5 hours at a temperature of 125 °C and a pressure of 5 bar), as shown in FIG. 10. The initial step of the curing cycle was added to maximise the moulding of the uncured laminate before curing.

At 826, the cured CFRP Herringbone-Bouligand laminate (as shown in FIG. 11) was cut to its final square shape with width 110 mm, using a water-jet. At 828, an impact surface was created on the laminate by spreading a layer of toughened-epoxy adhesive on the Herringbone-Bouligand region. At 830, the impact surface adhesive was cured under vacuum, at room temperature, using the set-up shown in FIG. 12. FIG. 13 shows the cured CFRP laminate with an impact surface (epoxy layer).

FIGS. 14 to 16 show a photograph and several optical micrographs of a manufactured Herringbone-Bouligand sample produced using the method shown in FIG. 8A.

Specifically, the micrograph of section A-A (corresponding to a (y, z)-plane not cutting through the peaks of the Herringbone motif) shown in FIG. 14 shows that a layered CFRP Bouligand microstructure (A0 = 2.5°) with bi-sinusoidal (Equation (1)) ply- interfaces (Herringbone motif) was successfully created. The micrograph shown in FIG. 14 shows that the CFRP Herringbone-Bouligand microstructure is characterised by a progressive variation of the amplitude to wavelength ratio (A/A) decreasing from the top surface of the laminate to the bottom surface. The micrograph of section A-A shown in FIG. 14 also shows that the toughened-epoxy used to create the impact surface of the Herringbone-Bouligand microstructure (at 828 in FIG. 8A) successfully filled the gaps between the peaks of the underlying CFRP Herringbone-Bouligand structure, creating a flat layer with a very good geometrical stability.

In addition, section C-C shown in FIG. 16 (corresponding to a (x, z)-plane not cutting through the peaks of the Herringbone motif) shows the good quality of the transition zone between the Bouligand region and the Herringbone-Bouligand region.

While the overall moulding procedure was successful, three types of manufacturing features were identified in the microstructure: (i) a small amount of fibre misalignment confined to a small region close to the top surface of the laminate, where A/A is maximum; (ii) the limited presence of voids (see section C-C in FIG. 16) on the bottom half of the laminate; and (iii) the presence of not fully-moulded peaks resulting in a periodic flattening of the bi-sinusoidal outermost layers, as shown in section B-B in FIG. 15, which corresponds to an x-plane located at the peaks of the Herringbone motif where the achieved maximum amplitude in the manufactured Herringbone pattern is therefore visible. The latter might have originated during the curing of the laminate due to the presence of trapped air between the top surface of the laminate and the mould (as shown schematically in FIG. 15). FIG. 15 also shows a top view of the cured Herringbone-Bouligand CFRP laminate highlighting the flattened peaks. Section B-B in FIG. 15 shows that while the crests of the Herringbone motif were not fully-moulded, the troughs showed excellent moulding quality with a minimum radius of curvature r_min ~ 0.5 mm, closely matching the nominal r_min of the Herringbone mould shown in FIG. 7. Additionally, although good control over the wavelength of the Herringbone motif was achieved, the presence of not-fully-moulded crests resulted in an outermost profile of the laminate with an amplitude of 1.29 mm, smaller than the nominal value (2.5 mm). As discussed further below, the aforementioned features did not significantly interact with the failure mechanisms and consequently they did not affect the mechanical performances of the Herringbone-Bouligand laminates under quasi-static indentation (QSI) tests.

Overall, the method shown in FIG. 8A led to a good quality of the final laminate and a good geometrical control over the parameters defining the Herringbone-Bouligand microstructure. As discussed further below, the quality of the Herringbone motif achieved in this example allowed a consistent enhancement in damage tolerance for the Herringbone-Bouligand microstructures with respect to tailored Bouligand microstructures to be demonstrated.

To evaluate the damage resistance properties achieved using a Herringbone- Bouligand region, several Herringbone-Bouligand samples were manufactured using the method of FIG. 8A. For comparison with the Herringbone-Bouligand samples, several Bouligand samples (shown schematically in FIG. 6) were also manufactured.

The stacking sequence of the plies used in the Bouligand samples were the same as in the Herringbone-Bouligand samples (i.e. 0°, 2.5°, ... , 177.5°), 180°, (177.5°, 175°, ... , 0°). As with the Herringbone-Bouligand samples, the plies in the Bouligand samples were oriented to the angle used in the stacking sequence and then cut. An alignment mould was used to perform the lay-up. The Bouligand samples were cured using the same autoclave cycle as the Herringbone-Bouligand samples and the samples of both configurations were scanned for defects before testing using ultrasonic C-scans. No high-toughness high-impact-resistant epoxy-based adhesive was spread on the impact surface of the Bouligand samples.

Therefore, the Herringbone-Bouligand and Bouligand samples comprised substantially lay-up sequences and were manufactured using identical autoclave cycles.

Consequently, the Bouligand and Herringbone-Bouligand samples were to a good extent similar, except for the microstructure. Looking at the thickness specifically, the thickness for the Bouligand laminates was te “ 3.48 mm, while, for the Herringbone- Bouligand laminates, the thickness at the trough was tHB^trough “ 2.68 mm (i.e. less than ts), while the thickness at the crest was tHB^crest “ 3.97 mm (i.e. greater than tc). The average thickness of both laminates was therefore similar, and the most significant difference was the microstructure itself.

Full -penetration QSI tests were performed using a custom-made test rig designed for the samples developed in this example. Each square sample (110 mm wide and 3.5 mm thick) was clamped between two aluminium plates (each 25 mm thick) with a central square opening 80 mm wide. The geometry of the opening was defined based on the geometry of the tailored Herringbone-Bouligand region of the Herringbone- Bouligand samples. The resulting span-to-thickness ratio of the QSI samples was 22.8, within the range of values suggested in the ASTM D6264 standard. A 50 kN Instron machine was used to load the samples at 1 mm/min displacement rate, using a hemispherical indenter made of steel 20 mm in diameter. The data were acquired with a frequency of 10 Hz. Additionally, an Imetrum camera was used to record the displacement at the indenter tip and correct for the machine compliance. For both Herringbone-Bouligand and Bouligand microstructures, three samples were tested until full-penetration (test stopped below a measured load of 1 kN) and one sample until an applied load of 4 kN to perform optical microscopy. All samples tested until full-penetration were tested using step-loading cycles. The test was interrupted at an applied load of 2 kN, 4 kN and immediately after catastrophic failure (load-drop). For each laminate, between each loading step, the surface damage was analysed via visual inspection of the impact and back surfaces, and delamination damage was analysed via ultrasonic C-scans performed in an immersion tank.

For the Herringbone-Bouligand and Bouligand CFRP samples tested under QSI, Table 1 shows the average values and standard deviations of peak load, displacement at peak load, penetration load and total dissipated energy. FIG. 17 shows the load vs displacement curves.

Table 1: Mean and standard deviation of peak load, displacement at peak load, penetration load and total dissipated energy for the Herringbone-Bouligand and Bouligand samples tested under QSI.

FIG. 18 shows the peak load and total dissipated energy. FIGS. 19 and 20 respectively show the evolution of the dent diameter (T) and the total projected delamination area at different stages of the test (applied load of 2 kN, 4 kN, critical failure (load-drop) and full penetration (end of test)).

FIGS. 17 to 20 show that, compared to the classical Bouligand microstructure, the Herringbone-Bouligand architecture led to: (i) a stiffer mechanical response during the initial stages of the loading; (ii) a 10% increase in maximum load bearing capability; (iii) a 11% decrease in displacement at critical failure; (iv) a 13% increase in penetration load; (v) a 13% increase in total dissipated energy with the associated scatter partially overlapping with the one measured for the Bouligand samples; (vi) a larger dent diameter (on average 62% larger before critical failure and 22% larger after critical failure); (vii) a greatly reduced (71%) total projected delamination area.

For two representative Bouligand and Herringbone-Bouligand samples, FIGS. 21 and 22 respectively show photographs of the impact face, the back face and the delamination damage detected via ultrasonic C-scans. Schematics of the type of microstructure characterising a certain region in the laminate are shown in these figures. The coloured areas in the C-scans refer to delamination at a certain ply- interface (see colour bar on top of each figure). The photographs and the C-scans images were taken by interrupting the test at an applied load of 2 kN, 4 kN and right after catastrophic failure (load-drop).

FIG. 23 shows photographs of two representative Herringbone-Bouligand and Bouligand samples tested up to a load of 4 kN and then cut for optical microscopy (sections A-A and B-B for the Herringbone-Bouligand and section C-C for the Bouligand sample). The microstructural features described above with reference to FIGS. 14 to 16 did not significantly interact with the damage evolution in the Herringbone-Bouligand laminates.

FIG. 24 shows photographs of the back face of two representative Herringbone- Bouligand and Bouligand samples after full-penetration, i.e. end of QSI test. The location of the clamp line and the boundary of the Herringbone-Bouligand region are also shown in FIG. 24.

FIG. 21 shows that at an applied load of 2 kN, delamination damage was already present in the Bouligand microstructure. On the contrary, no damage could be detected with the ultrasonic probe in the Herringbone-Bouligand microstructure (FIG. 22). Therefore, the Herringbone pattern successfully delayed the onset of delamination damage.

As the load increases, the smaller extent of delamination damage in the Herringbone- Bouligand laminates contributed to a stiffer mechanical response (FIG. 17). On the contrary, the earlier formation of helicoidal delamination damage (FIG. 21) in the Bouligand samples led to a gradual softening (knee region), corresponding to a non- linear load response in FIG. 17. It has previously been shown that the through-the- thickness distribution of delamination damage in Bouligand architectures can be inferred from the through-the-thickness distribution of intralaminar shear stresses Ti 3 and T23, which drive the formation of shearing matrix cracks and delamination in the bulk of the laminate. Specifically, in Bouligand microstructures with small pitch angles (such as A0 = 2.5° in this example), the maximum value of intralaminar shear stress T13 (driving the formation of delaminations) was previously found to be uniform across a large number of plies. This is expected to lead to the failure of several interfaces, hence promoting smooth through-the-thickness helicoidal delamination damage, such as the one observed in the Bouligand samples tested in this example (applied load stopped at 4 kN in FIG. 21). The best performing pitch angles may range between 2° to 20° for fibre reinforced composites different form the ones used in this example.

At an applied load of 4 kN, section C-C in FIG. 23 shows that the helicoidal distribution of delaminations in Bouligand laminates contain various continuous (in-plane (x,y)) delamination areas at the interface between two generic plies. On the contrary, for the Herringbone-Bouligand samples, disconnected delamination areas appeared at the same interface between two generic plies (FIG. 23). As explained above, the Herringbone pattern disclosed herein defines bi-sinusoidal interfaces between consecutive plies, with ratios A/A ranging nominally from 0.5 (at the top surface of the laminate) to 0 (on the bottom surface). Due to the nature of the loading in QSI tests, these interfaces are subjected to dominant Mode II. Consequently, any bi-sinusoidal interface with A/A > 0 forms a mechanical interlock. This results in an important damage-tolerant mechanism which: (i) together with the lower amount of delamination damage contributes to a stiffer mechanical response; (ii) results in a higher resistance to form and grow delaminations (“adhesive” type of failure) leading to the formation of discontinuous delamination areas; and (iii) promotes the activation of sub-critical “cohesive” mechanisms of failure (more prominent in the compression side of the laminate where the values of A/A are larger) such as delamination deflection and accumulation of matrix cracks (as observed in sections A-A and B-B of FIG. 23).

FIG. 19 shows that the dent diameter (p before and after critical failure (load-drop) was larger in the Herringbone-Bouligand than in the Bouligand samples. As explained above with reference to FIG. 8A, the impact surface in the Herringbone-Bouligand samples was created by spreading a layer of a high-toughness high-impact-resistant epoxy-based adhesive characterised by a higher toughness compared to the CFRP composite impact surface of the Bouligand samples. The presence of the impact surface on the Herringbone-Bouligand samples resulted in higher energy dissipation during the contact between the indenter and the Herringbone-Bouligand laminates, hence leading to a larger dent diameter (p. Additionally, the epoxy-based adhesive used to create the Herringbone-Bouligand impact surface is characterised by higher plastic deformation capability and lower Young's modulus compared to the brittle and stiff CFRP impact surface of the Bouligand samples. The epoxy-based adhesive therefore promotes contact stress redistribution which may have prevented the localisation of stress concentrations at the peaks of the underlying Herringbone- Bouligand CFRP structure. This is shown by the absence of premature damage due to stress concentrations observed in the Herringbone-Bouligand laminates (as shown in sections A-A and B-B in FIG. 23).

Overall, the effect of a tough impact surface, together with the disconnected delamination areas (sections A-A and B-B in FIG. 23) and other sub-critical dissipative mechanisms such as: (i) delamination deflection, (ii) accumulation of subcritical matrix cracks in the compression side, and (iii) the formation of several fully-developed sub- critical Bouligand matrix cracks in the tension side, have contributed to increasing the sub-critical damage diffusion capability of Herringbone-Bouligand microstructures with respect to classical Bouligand layups, thereby enhancing the damage tolerance to through-the-thickness loads.

As mentioned above, the Herringbone-Bouligand laminates showed a high through-the- thickness sub-critical damage diffusion characterised by the accumulation of disconnected delaminations and matrix cracks (as shown in FIG. 23). The lack of continuous delamination areas together with the interlocking mechanisms offered by the bi-sinusoidal interfaces of the Herringbone pattern led to a stiffer response (FIG. 17) and to a 10% increase in maximum load bearing capability with respect to Bouligand laminates (FIG. 18). However, the larger in-plane (x-y) spreading of delamination damage of the Bouligand samples was more effective in relieving the back-face tensile stresses, hence leading to a delay in the occurrence of fibre failure (displacement at load-drop 11% smaller in the Herringbone-Bouligand laminates).

Interestingly, for the Bouligand microstructure, FIG. 21 shows that immediately after critical failure, the delamination damage reached the clamping line, occupying most of the available testing area. On the contrary, the ability of the Herringbone-Bouligand laminates in resisting delamination formation and growth resulted in a delamination damage typically contained inside the Herringbone-Bouligand region. Overall, Herringbone-Bouligand laminates achieved a large reduction in total projected delamination area — 71% smaller than in the Bouligand laminates (FIG. 20). Additionally, FIGS. 21 and 22 show that while fibre failure extended up to the clamp line in the Bouligand microstructure, this was entirely contained inside the Herringbone- Bouligand region in the Herringbone-Bouligand samples, further highlighting the high damage tolerance of the Herringbone-Bouligand microstructure.

Table 1 and FIG. 17 show that, after critical failure has occurred, the lower extent of inplane spreading of damage allowed Herringbone-Bouligand laminates to delay penetration at higher loads (11% increase in penetration load). As shown in FIG. 19, for the Herringbone-Bouligand laminates, the dent at this stage of the test was still larger than the one in the Bouligand laminates. This indicates that the presence of a tough impact surface, along with the ability of the Herringbone-Bouligand microstructure in accumulating damage in the compression side, avoided the localisation of failure due to contact throughout the entire duration of the test.

FIG. 18 shows that the Herringbone-Bouligand microstructure dissipated on average 13% more energy during the entire fracture process than the Bouligand microstructure. The average increase in total dissipated energy observed for the Herringbone- Bouligand microstructure was achieved with a simultaneous large decrease in in-plane extent of damage (71% smaller projected delamination area than the one measured for the Bouligand samples). Therefore, although in the Herringbone-Bouligand samples the in-plane extent of damage was greatly reduced, the highly-dissipative sub-critical failure mechanisms activating in Herringbone-Bouligand laminates (discussed above) successfully increased on average the energy dissipation capability compared to classical Bouligand structures.

Additionally, the analysis of the back face of a representative Herringbone-Bouligand laminate at the end of the penetration phase shown in FIG. 24 reveals that the overall damage, including fibre failure, delaminations and matrix cracks, was successfully contained inside the Herringbone-Bouligand region. On the contrary, for the Bouligand laminates, damage extended up to the clamp line during penetration (FIG. 24), resulting in reduced structural integrity.

To summarise, the Herringbone-Bouligand samples were capable of greatly reducing the in-plane spreading of damage and containing damage within the Herringbone- Bouligand region while increasing the load bearing capability and the energy dissipation capability of the structure. We can therefore conclude that the Herringbone- Bouligand design offers a more damage-tolerant solution than tailored Bouligand architectures.

FIGS. 25 and 26 show a comparison of the average total dissipated energy (FIG. 25) and peak load (FIG. 26) for the Herringbone-Bouligand, Bouligand and “conventional” QI (quasi-isotropic) microstructures. FIGS. 25 and 26 show that the Bouligand microstructure outperforms the “conventional” QI lay-up both in terms of total dissipated energy and load-bearing capability (peak load). Additionally, FIGS. 25 and 26 show that the Herringbone-Bouligand microstructure further improved the performances of the Bouligand microstructure both in terms of total dissipated energy and load-bearing capability.

The described methods may be implemented using computer executable instructions. A computer program product or computer readable medium may comprise or store the computer executable instructions. The computer program product or computer readable medium may comprise a hard disk drive, a flash memory, a read-only memory (ROM), a CD, a DVD, a cache, a random-access memory (RAM) and/or any other storage media in which information is stored for any duration (e.g., for extended time periods, permanently, brief instances, for temporarily buffering, and/or for caching of the information). A computer program may comprise the computer executable instructions. The computer readable medium may be a tangible or non-transitory computer readable medium. The term “computer readable” encompasses “machine readable”.

The singular terms “a” and “an” should not be taken to mean “one and only one”. Rather, they should be taken to mean “at least one” or “one or more” unless stated otherwise. The word “comprising” and its derivatives including “comprises” and “comprise” include each of the stated features, but does not exclude the inclusion of one or more further features.

The above implementations have been described by way of example only, and the described implementations are to be considered in all respects only as illustrative and not restrictive. It will be appreciated that variations of the described implementations may be made without departing from the scope of the invention. It will also be apparent that there are many variations that have not been described, but that fall within the scope of the appended claims.

Claims

CLAIMS:

1. A method of manufacturing a fibre reinforced composite structure, the method comprising: laying up a plurality of plies of fibre reinforced composite material, wherein each of the plurality of plies comprises a plurality of deviations in the thickness direction of the ply; and curing the plurality of plies.

2. A method according to claim 1, wherein laying up the plurality of plies comprises laying up the plurality of plies such that the alignment of fibres in one of the plurality of plies is angled with respect to the alignment of fibres in an adjacent one of the plurality of plies.

3. A method according to claim 1 or claim 2, further comprising: forming the plurality of deviations in the thickness direction of each of the plies.

4. A method according to claim 3, wherein forming the plurality of deviations in the thickness direction of each of the plies comprises placing the plurality of plies on a mould comprising a plurality of protrusions.

5. A method according to claim 4, wherein an amplitude and a distribution of the plurality of protrusions is selected to maximise the out-of-plane deviation in the thickness direction, wherein the selection of the amplitude and the distribution is based on a fibre type and a manufacturability constraint of the mould.

6. A method according to claim 4 or claim 5, wherein each of the plies comprises a matrix material, and wherein the method further comprises pre-heating the mould to a temperature to cause a decrease in the viscosity of the matrix material prior to laying up the plurality of plies on the mould.

7. A method according to any of claims 1 to 6, wherein the plurality of deviations extends over an area of each of the plurality of plies.

27

8. A method according to any of claims 1 to 7, wherein the alignment of fibres in each of the plurality of plies is angled with respect to the alignment of fibres in an adjacent one of the plurality of plies.

9. A method according to claim 8, wherein the angle between the alignment of the fibres in adjacent ones of the plurality of plies is between about 2° and about 20°.

10. A method according to any of claims 1 to 9, wherein curing the plurality of plies comprises: curing the plurality of plies while the ply at a first face of the plurality of plies is uncovered.

11. A method according to claim 10, further comprising forming an impact layer over the ply at a second face of the plurality of plies, wherein the second face is opposite the first face.

12. A method according to any of claims 1 to 11, further comprising forming an impact layer over the ply at one face of the plurality of plies.

13. A method according to any of claims 1 to 12, wherein each of the plies comprises a matrix material, and wherein curing the plurality of plies comprises curing the plies at a temperature that allows the matrix material to reach a viscosity of between 10⁷ and 10⁸ cP.

14. A method according to claim 13, wherein curing the plurality of plies comprises curing the plies at the temperature that allows the matrix material to reach a viscosity of between 10⁷ and 10⁸ cP for one hour.

15. A method according to any of claims 1 to 14, wherein laying up the plurality of plies comprises: splitting the plurality of plies into a first plurality of groups of plies; and for each of group of plies in the first plurality of groups of plies: placing plies in the group of plies on a mould; and moulding the plies in the group of plies.

16. A method according to claim 15, wherein for each group of plies in the first plurality of groups of plies, moulding the plies in the group of plies comprises placing the group of plies under vacuum for two to ten minutes.

17. A method according to claim 15 or claim 16, wherein for each group of plies in the first plurality of groups of plies, moulding the plies in the group of plies comprises compressing the group of plies using an indenter.

18. A method according to any of claims 1 to 17, wherein the plurality of deviations is periodic, and optionally wherein the periodic plurality of deviations is sinusoidal.

19. A method according to any of claims 1 to 18, wherein for each of the plurality of plies, the periodic deviation is located in a first region of the ply, the ply further comprising a second region with no deviation in the thickness direction of the ply.

20. A fibre reinforced composite structure obtained by the method of any of claims 1 to 18.