The invention relates to a composite material structures where the composite is composed of structural fibres embedded within a matrix polymer resin system. The invention particularly but not exclusively relates to methods and apparatus for manufacture of the structures. Use of the invention enables the construction of web type structures employing unidirectional fibres in which the webs pass through each other by the creation of a node. The invention also facilitates integration of the web structures into various different products and to products obtained by use of the methods.

Historically web type structures have been used to provide structural stiffness, for example in structures made of steel, cast iron, many alloys and plastics. Typical components incorporating web type structures include gear boxes, manhole covers, cabinet doors and open mesh grating. Processes such as moulding, machining from solids and casting can produce web structures effectively but in the case of steel and other materials supplied in lengths of constant cross section a method of connecting the webs is required. This incurs additional complexity, effort and cost. Composite materials consisting of structural fibres such as glass or carbon fibres can be made efficiently by a pultrusion process. The sections made are typically of constant cross sectional shape. However it is difficult to efficiently make them into a finished product which effectively uses their structural properties. Ladders are a typical product made from pultruded glass fibre composite sections. However, they require additional effort in the form of drilling, bonding and mechanical fixing methods to complete a finished ladder.

The format of fibres typically found in composite products may fall into three main categories:

1. random fibres either in chopped or chopped strand mat format;

2. woven roving or stitch bonded fabrics;

3. unidirectional fibres where all the fibres are in the same direction. The format and quantity of fibres significantly influences the tensile strength of the product. In practice the fibre volume fraction is limited as each fibre should be coated with the matrix resin and the fibres have to be packed together. The packing density is influenced by the fibre format. Random fibres have the least packing density.

The process of manufacturing high strength composites typically found in the aeronautical industry involves taking woven laminates from which profiles are cut to predefined shapes. The laminates are then directly applied to a mould surface or assembled into a preform and then placed into a mould. The laminates may be pre- impregnated with a matrix resin and are then heat cured or contain only fibre and are infused with a matrix resin. The laminates typically contain fibres in two or three directions and are laid down to provide predetermined fibre orientations. Laminates of this type are referred to as bi-directional laminates and can have high tensile strength values, typically 300Mpa for glass fibre and 440Mpa for a carbon fibre laminate.

EP 1917401 discloses structures which achieve tensile strength values in the range 300 to 380MPa.

Our application WO-2013/067560 and WO-2013/067564 disclose a method of formation of nodes in composite material structures and apparatus for performing the method. The disclosures of these specifications are incorporated into the present specification by reference for all purposes.

Composite structures made from uni-directional fibres have significantly higher tensile strength values than bi-directional fibres, typically 800Mpa for glass fibre and lOOOMpa for carbon fibre. Products made from uni-directional fibres have been limited to items such as golf club shafts, fishing rods and spokes supporting cryogenic units. To date the technology to build more complex structures where elements of the structure are required to pass through each other has not been available. To build the type of structures that this technology can facilitate requires creation of fibre preforms with a complex fibre architecture. It is an object of the present invention to provide a method and apparatus for automatically assembling uni-directional fibres into a preform of fibres than can be handled, placed into a mould or become part of a more complex preform.

According to a first aspect of this invention, a method of forming a fibre reinforced polymer structure comprises the steps of:

providing a first bundle of fibres arranged in an array and having a first fibre volume fraction (FVF);

forming a node region in the bundle wherein a transverse dimension of the first bundle is increased and a second perpendicular dimension of the bundle is decreased so that the first FVF is maintained at a substantially constant value;

providing a second bundle of fibres extending in angular relation to the first bundle, arranged in an array and having a second FVF;

forming a node region in the second bundle wherein a transverse dimension of the second bundle is increased and a perpendicular dimension of the second bundle is decreased so that the second FVF is maintained at an approximately constant value; wherein the node region of the second bundle overlies the node region of the first bundle to form an assembly;

wherein the second bundle extends from a single side of the first bundle;

infusing the assembly with a polymer resin; and

curing the resin to form the structure.

According to a second aspect of the present invention, fibre reinforced polymer structure manufacturing apparatus comprises:

a first feeder for providing a first bundle of fibres in a first direction;

a second feeder for providing a second bundle of fibres in a second direction; wherein the first and second directions are disposed in angular relation;

and arranged so that the first and second bundles are overlaid in a node region; wherein the second bundle extends from a single side of the first bundle;

the apparatus further comprising first and second dies, each having a variable lateral and perpendicular dimension, arranged so that the first and second bundles pass through a respective die at or adjacent the node region to control the lateral and perpendicular dimensions of the bundle;

wherein the node regions are superimposed to form an uncured assembly; and heating means arranged to provide heat and pressure to bond the assembly to form a fibre reinforced structure.

Preferably, the step of overlying first and second bundles is repeated to form a structure having a multi-layered node.

The bundles preferably comprise thin layers of fibres, for example, having a thickness of about 1mm.

A joint in accordance with this invention comprises a T-shaped or L-shaped joint. The joint may not be perpendicular but will be referred to in this specification as a T-shaped or L-shaped joint for brevity.

In a further embodiment, one or more bundles may be joined end to end, for example, to form an elongate or annular structure.

The T-shaped joint may be located at an edge of a lattice structure or a joint between a beam and cross-member.

In preferred embodiments, the second bundle extends perpendicularly from a first side of a first bundle to form a T-shaped joint.

An L-shaped joint in accordance with this invention may be a corner, for example in a rectangular or otherwise shaped frame or support. A frame or support may define an aperture with rectangular or otherwise L-shaped joints which may be used to receive and engage a panel, cover or other component.

Manufacture of L-shaped or T-shaped structures of this invention may be significantly cheaper than comparable metal structures formed as pressings or moldings. The second side of the first bundle may have a straight edge so that the fibres of the second bundle do not extend beyond the second side of the first bundle.

Ends of the fibres of the second bundle may be cut and arranged to prevent them extending beyond the second side of the first bundle. Alternatively, the fibres of the second bundle may be laid with the free ends arranged perpendicular to the second side in a method as disclosed in PCT/GB2013/067560.

In an alternative preferred embodiment, the fibres of the second bundle may be folded at the second side of the first bundle. The fibres may be folded around fibres of the first bundle. In this way, the second bundle may extend across the structure twice or repeatedly, avoiding a need for cutting of the fibres of the bundle.

In an L-shaped or corner joint, fibres of both bundles may be successively folded over fibres of adjacent bundles. Such an arrangement has the effect of providing a coherent and hardwearing structure to the exterior of the corner.

Fibres of each bundle may be folded alternately over fibres of adjacent bundles. Alternatively, fibres of a first bundle may be folded over two or more second bundles. Such an arrangement may have the advantage of forming a smooth, coherent surface on the edge of the composite structure due to regions of the fibres extending across the outer surface. The outer surface may have the appearance of a woven material, for example a twill pattern.

Folding of the fibres is particularly advantageous in perpendicular arrangements. However, lattices in which the cross-members extend at 45° or 60° and 30° to the edge or second bundle may be used to form an array having rhomboid or diamond-shaped apertures with a peripheral frame or border. Lattice structures comprising an outer frame and cross members defining rhomboid apertures may be provided.

The second or straight edge may be located facing outwardly of the joint, for example to form a structure having linear straight sides. Alternatively, the second edge may be located inwardly to form a frame having a linear sided socket in order to receive a square or rectangular component, such as a closure or access cover.

Further embodiments in which ends of the bundles are joined together to form a circular, ovoid or otherwise shaped loop may be provided.

The transverse dimension of each bundle is preferably increased from the transverse dimension or width of an internodal body of the bundle to a transverse nodal dimension or width, with an intermediate portion of increasing transverse dimension or width;

wherein the perpendicular dimension or height decreases from the perpendicular dimension or height of the internodal bundle to a perpendicular nodal dimension or height with an intermediate portion of decreasing transitional perpendicular dimension or height.

Advantageous methods include the step of providing an array of thermally fusible fibres within the node region. Preferably, the fusible fibres are provided within each bundle or interleaved between adjacent bundles.

The fusible fibres may be used to secure the bundles together by microbonding to form a preform.

The array of fusible fibres may be laid upon the first bundle as a web or as a warp or preferably as a weft of parallel or unidirectional fibres.

Alternatively, the fusible fibres may be formed as a skeleton having a configuration of the completed structure. In a further alternative embodiment, a fusible sheet is applied between the first and second bundles.

The fusible fibres may be deployed on a first bundle before laying a second bundle onto the fusible fibres. This results in the fusible fibres being sandwiched between adjacent bundles.

Alternatively, the fusible fibres may be interleaved between rovings of each bundle, preferably extending transversely of the bundles. For example, the fusible fibres may be layed as weft fibres. Use of a woven arrangement has the advantage of binding the fibres together throughout the fibre structure.

Alternatively, individual fusible fibres or bundles of fusible fibres may be laid parallel to a bundle or may be formed into a skeleton or thin sheet for application between adjacent bundles.

The method may further comprise the step of applying heat and pressure to the assembly to form a microbonded assembly of fibres. This assembly may be suitable for handling and placing into a mold.

The assembly may be passed through a heated nip, for example between heated rollers or between a heated roller and support surface. Alternatively, a heated press may be used.

Preferably cooling means are provided for rapidly solidifying the fusible fibres. Air jets or other gaseous cooling means may be employed. Alternatively, the structure may be passed between cooled rollers. In an embodiment in which the assembly comprises an array of nodes, for example a grating, cooling gas may be passed through apertures between the nodes to promote rapid solidification of the composite structure.

The structure may comprise a preform for an engineering component. The apparatus of this invention allows manufacture of preforms with a high degree of structural and dimensional accuracy and consistency. Complex structures may be formed.

The die of the apparatus may advantageously comprise a channel together with means for adjusting a lateral and perpendicular dimension of the channel in the node region. For example the width and depth of the channel may be adjusted.

In a particularly preferred embodiment, the die comprises a rotatable cam, preferably a cylindrical rotatable cam having a circumferential channel with a variable lateral and perpendicular dimension. Conveniently the channel has variable width and depth.

A stepper motor or other reversibly controllable drive means may be arranged to rotate the cam in both a forward or reverse direction with respect to the direction of fibre feed so that a channel with variable dimensions is presented to the bundle as it reaches the node region. The rate and angular extent of the rotation in either direction may be controlled.

In the preferred method, each die may be rotated either with or against the direction of travel of the bundle to alter the dimensions of the channel through which the bundle passes, the die being returned to a starting point after each node has been formed and laid upon the preceding bundle. The vertical dimension or depth of the channel may be selected so that a bundle is urged into engagement with the underlying bundle.

Alternatively, a die having an adjustable orifice may be employed. The cross- sectional area or shape of the orifice may be adjusted. Preferably the cross-sectional area is maintained while the dimensions of the orifice are adjusted to control the dimensions of the bundle.

Apparatus for manufacture of grating or lattice structures may have two or more arrays of feeders disposed in angular relation and arranged to lay bundles successively to build up a multi-layer node structure.

Structural fibres used in the production of high strength composite products are commonly supplied in creels which are wound from a continuous length of direct roving. A typical creel may have a weight of 20Kg. The direct roving consists of a number of fibres lightly bonded together by a sizing agent and are normally measured in a unit such as Tex. For example, a direct roving defined as 1200 Tex would have a weight of 1200 grams per kilometre (g/km).

The invention may further comprise a fusible fibre feeder arranged to lay an array of fusible fibres upon each bundle of first and second fibres in the node region. In a preferred embodiment, the bundle comprises a plurality of parallel rovings or tapes of fibres laid with an array of fusible fibres alternating above and below adjacent rovings. For example, a bundle having a width of 5mm may comprise 12 rovings of 600 Tex. The fusible fibres may be interleaved or woven between adjacent tapes. The fusible fibres may be provided as a skeleton laid between adjacent tapes. The skeleton preferably has a width equal to or less than the width of the bundle.

The manufacturing process commences with laying dry fibres in alternate layers to create the webs. As the fibres enter the node they are spread into their appropriate shape necessary to pass through each node. The shape is maintained by either microbonding or stitching the fibres together and in this way a preform of fibres is constructed. The preform is then placed in either a mould or may be assembled into a more complex preform which is then assembled into a mould. The mould is closed and sealed and then infused with catalysed resin. On completion of the curing stage the mould is opened and the component is removed. At this point the component will only require removal of flash and if required the addition of ancillary components.

The invention provides a construction method that allows webs of composite material to pass though each other by the creation of a node, within which bundles of structural fibres that form the webs and typically comprise unidirectional fibres, are permitted to change their cross sectional shape so that in both webs their cross sectional area and therefore their fibre volume fraction remains constant as each web passes through the node. In this manner the fibre volume fraction can be maintained at its desired value.

According to a third aspect of the present invention, a fibre reinforced polymer structure manufactured by a method in accordance with the first aspect of this invention or using apparatus in accordance with the second aspect of this invention comprises a plurality of first and second bundles of substantially unidirectional fibres; a first bundle overlying a second bundle in angular relation to form a node; wherein the second bundle extends from a single side of the first bundle;

wherein the fibre volume fraction (FVF) of each bundle remains approximately constant as it passes through the node. This invention may provide a fibre reinforced polymer structure comprising a generally T or L-shaped junction where two webs pass through each other that permits the tensile strength properties of each web to be retained.

The fibre volume fraction (FVF) may be defined as the volume of fibre as a proportion of the total composite volume or as the ratio of the cross-sectional areas. FVF may be determined by chemical matrix digestion, photomicroscopy or by calculation.

A structure in accordance with the present invention may have a FVF of about 40% to about 60%, preferably about 50% to about 60%.

In some embodiments of the invention, the FVF values of the first and second bundles are the same or approximately the same. In this case the FVF of the node may be approximately the same as the FVF of each bundle. This avoids unnecessary interlaminar shear problems.

Embodiments in which the FVF of each of the bundles is the same are advantageous because homogenous infusion of the resin is facilitated. Furthermore, compressive forces within the mold should be evenly balanced to maintain the configuration and relative alignment of the fibres during compression. This may serve to create a more isotropic structure than may be achieved using typical composite materials.

Alternatively, the FVF of the first bundle may be different to the FVF of the second bundle, with the total FVF of the node being a mean or intermediate value of the components.

For the purposes of simplicity, the composite structure will be described with horizontally extending bundles arranged in a vertical stack. However, overlying bundles may extend vertically or may extend in a curved direction, for example if the structure is assembled on a roller or other cylindrical surface. The bundles may also be assembled on a conical or spherical surface. The FVF may change vertically across the node although the FVF of each bundle remains approximately constant. Preferably, the total FVF remains approximately constant across all bundles within a node.

By use of the terms substantially constant or approximately constant it is intended that the FVF may have a tolerance or margin of variation of ± 10%, preferably ± 5%, more preferably ± 2%.

The angle between the first and second bundles is preferably 90°. Alternatively, an angle of 30°, 40°, 60° or other suitable angle may be employed.

The angle between adjacent bundles is preferably constant in the vertical direction from the top to the bottom of the node. However, second bundles may be overlaid at two or more different angles, for examples 30° and 60°, relative to the direction of the first bundles.

In order to maintain a constant FVF, the height of each bundle may decrease at the node with the dimension in the transverse direction or width increasing in inverse proportion. In a first embodiment in which two similar bundles are overlain at 90°, the height of each bundle may be reduced by 50% at the node with the width increasing by 100% in order to maintain a constant FVF. Preferably, a constant cross sectional area is maintained in each bundle.

The fibres of each bundle may be splayed or displaced laterally as the fibres approach the node. Each bundle of fibres may have a median, longitudinal axis, the displacement of each fibre being proportional to the distance from the axis. Fibres, particularly glass fibres, may be displaced at an angle up to 30°, typically 20°, without kinking or deformation.

The incremental angular displacement between adjacent fibres is small so that interlaminar shear forces are minimised. Preferably, the angular displacement is less than 5°, typically 1% or less. Small incremental angular displacements reduce interlaminar shear forces so that a value approaching zero is obtained. In order to achieve a desired fibre structure at a junction where webs pass through it is necessary for fibres to make angular changes relative to each other. These changes of angle can be made without loss of strength properties, providing that the change between individual fibres is small. For example, if a layer of fibres is 5mm wide and the fibre architecture requires that the fibre of the outer layer on one side is to deviate by 20 degrees from the fibre at the opposite side, then the angular change can be evaluated. Assuming the fibre diameter is 17 microns, then across the width of the layer of fibres there will be 294 fibres. The angular change of each fibre relative to its adjacent fibre is equal to 20/294, producing an angular change of 0.068 degree between each fibre. This can be considered as an infinitesimally small change of angle.

In particularly preferred embodiments, the following requirements may be met;

1. the fibres within each bundle are laid parallel to one another;

2. the fibres are crimp free;

3. the fibre volume fraction of each bundle is approximately constant; and

4. a change of direction of a fibre relative to an adjacent fibre of the bundle is less than 1°.

In a particularly preferred embodiment, a node is characterised as a region in which all of the fibres of a first bundle pass over all of the fibres of a second bundle, wherein all of the fibres of each bundle are parallel, crimp free and have a constant fibre volume fraction (FVF) with no change in direction of each fibre relative to adjacent fibres in the bundle.

It is particularly advantageous if the change in direction of a fibre relative to an adjacent fibre of the bundle is less than 0.5°, especially less than 0.01°.

The distance from the axis of the fibres of a first bundle may increase in a direction towards the node until a region of overlap (defined as a node) with fibres of a second bundle is reached. The region of overlap or node may be rectangular or square in plan view for perpendicular bundles with the fibres of each bundle extending in parallel spaced relation perpendicular to the fibres of an adjacent bundle. Where the bundles are the same size, the region of overlap may be square. The presence of transverse fibres for non-perpendicular arrangements or perpendicular fibres for rectilinear arrangements strengthens the node by providing a restoring force reducing the tendency of the fibres in a bundle to move closer when under tension. Interlaminar strains are relieved and the FVF within the node remains constant.

Various types of fibre may be employed, including aramid, carbon fibre, S- glass and E-glass. The matrix polymer resin may be a thermosetting polymer, or a thermoplastic polymer or blend having sufficient melt flow to permit the polymer to be shaped at an elevated temperature below the temperature at which it degrades. Thermosetting polymers may be selected from polyester, acrylic, polyurethane and hybrid resin combinations. Polyethers, polyetherketones and polyetheretherketones, for example as sold under the registered trade mark VICTREX may be used. An additional advantage of the invention is that webs can be formed from more than one fibre type allowing specific properties and economic criteria to be achieved.

Fibres may be provided as tapes comprising a monolayer of fibres or a thickness of a small number of fibres, for example 1 to 5 or 1 to 10 fibres. These tapes may be laid alternately to provide a closely integrated nodal structure, for example in which each row of fibres is adjacent rows of fibres extending in a different direction.

Co-extruded fibres such as DYNEEMA (registered trade mark of DuPont) may be employed. Such fibres may be thermally set in a moulding tool.

In a particularly preferred embodiment, the node further comprises an array of thermally fusible fibres located between first and second bundles. The fusible fibres may be arranged to bond the bundles, for example, by microbonding.

The fusible fibres may be laid as an array between the first and second bundles, for example as a skeleton, web, warp or weft during assembly of the node. Alternatively, a bundle may incorporate a proportion of fusible fibres, for example located on an outer surface of the bundle and arranged to contact an adjacent bundle during assembly of the structure.

The fusible fibres may be composed entirely of a fusible material having a melting temperature in the range 60°C to 160°C. Various polymers or blends such as polyester, copolyester or polyamide may be used. Alternatively, combination yarns may be used, comprising a high tenacity carrier thread coated with a fusible coating such as low melting point polyamide. Suitable fibres are manufactured by EMS- Chemie AG under the registered trade mark GRILON.

Functional devices may be incorporated into one or more of the bundles. For example, sensors or identity devices may be included to allow remote checking of the security or location of the structure, for example of a cover for an access chamber.

The dimensional configuration of the node may be precisely determined by the need to maintain the fibre volume fraction of the fibres in the webs as they pass over each other in the node and accommodate the changing width of the webs as they approach the node. If the webs are of different thickness the node shape may be changed accordingly. In the event that webs approach each other at an angle of less than 90° the node shape is modified so that the fibre volume fraction can be maintained at a constant value.

In preferred embodiments of the invention the fibre volume fraction of the web is maintained as it passes through the node. However individual design requirements may dictate that some webs do not have the same structural requirement and may employ a different fibre volume fraction compared to the web they pass through. In such a case, the node design may be adapted.

In a further embodiment of the invention it may be advantageous to incorporate a larger web in one direction and a smaller web in another direction and only where the smaller web passes through the larger web will a node be required. The use of a node through which the fibres of both webs pass serves to maintain the structural properties of the web as there is no change in fibre volume fraction. There may be a deviation from linearity of the fibres as they approach the junction within the node where the fibres from both webs pass over each other. Tensile testing of webs has shown no reduction in strength of the webs and that the strength of the nodes benefits from the interaction of fibres from each web.

According to a fourth aspect of the present invention, an engineering component comprises a fibre reinforced polymer structure in accordance with the third aspect of this invention.

According to a fifth aspect of the present invention, a fibre preform for an engineering component comprises a fibre reinforced structure in accordance with the third aspect of this invention, wherein the bundles are bonded together. The bundles may be bonded, for example, by microbonding.

The engineering component may comprise a structural member or load bearing panel, joist, beam or support. A load bearing work surface or cover may comprise the structure together with a suitable surface layer.

Engineering components in accordance with this invention find applications in manufacture of automobile chassis members and structural components, aerospace components and civil engineering components. Marine or other water resistant components may be provided.

The engineering component may comprise one or more first bundles and an array of second bundles, the second bundles being arranged in parallel spaced relation, for example, to form an edge of a lattice or grating. Such gratings may be used for access chambers.

Alternatively, the engineering component may comprise a frame defining a socket for a cover, closure or other component. The invention is further described, by means of example but not in any limitative sense, with reference to the accompanying drawings, of which:-

Figure 1 is a perspective view of a single structural element;

Figure 2 is a perspective view of a T-shaped joint;

Figure 3 is a perspective view of an L-shaped joint;

Figure 4 is a plan view of bundle of fibres as they spread prior to a node;

Figure 5 is a plan view of the first bundle;

Figure 6 is an elevation of the first bundle shown in Figure 5;

Figure 7 is a plan view of the second bundle;

Figure 8 is a side view of the second member;

Figure 9 is a plan view of a T-shaped joint;

Figure 10 is an isometric view of a T-shaped joint;

Figure 11 is a cross-section of a joint with the second bundle folded around the first bundle.

Figure 12 is a plan view of the joint shown in Figure 1 1 ; Figure 13 is a plan view of an oblique T-shaped joint; Figure 14 is a partial section on F-F of Figure 13; Figure 15 is a plan view of an L-shaped joint; Figure 16 is cross-sectional view of the joint shown in Figure 14:

Figure 17 is an isometric view of a structure in accordance with the invention; Figure 18 is an isometric view of a frame in accordance with the invention;

Figure 19 is apian view of apparatus in accordance with this invention; Figure 20 is an end elevation of a feeder of the apparatus; Figure 21 is a side elevation of the feeder shown in Figure 20;

Figure 22 is a schematic view of the feeder apparatus; Figure 23 shows a skeleton of fusible fibres;

Figure 24 is a cross-sectional view of the preform;

Figure 25 is a side elevation of the apparatus; Figure 26 shows a hot and cold shoe assembly;

Figure 27 is a side view of the assembly; and

Figure 28 is an end view of the cold shoe. Unidirectional composite structures have excellent tensile properties along the length of the structure and an example of a typical structure is the flat bar shown in Figure 1 , which may be described as a web.

On its own the web shown in Figure 1 finds very limited application in general load bearing structures. Unlike a steel structure it cannot be welded and can only be integrated into a structure by the use of mechanical methods such as bonding or fastening together by some means. To make effective use of the high tensile strength values offered by unidirectional fibres, it is necessary to be able to manufacture structures with a connection system that facilitates the basic web element shown in Figure 1 to be produced with T-shaped or L-shaped connection types as shown in Figure 2 and Figure 3.

The invention provides a structural fibre architecture at the junction where two webs are integrated into each to form a connection and for that junction to retain the tensile strength properties of each individual web.

A unidirectional composite implies that the fibres must be parallel to each other and crimp free. High fibre volume fractions are also desirable as they achieve the highest tensile strength values and it is important to maintain a constant fibre volume fraction within a structure to reduce stress concentrations.

Where it is necessary for fibres to make angular changes relative to each other, then these changes of angle can be made without loss of strength properties providing that the change between individual fibres can be defined as infinitesimally small. For example, if a layer of fibres is 5mm wide and the fibre architecture requires that the fibre of the outer layer on one side is to deviate by 20 degrees from the fibre at the opposite side, then we can evaluate the angular change as follows. See Figure 4. Assuming the fibre diameter is 17micron then across the width of the layer of fibres there will be 294 fibres. The angular change of each fibre relative to its adjacent fibre is equal to 20/294 producing an angular change of 0.068 degrees between each fibre and this can be considered as an infinitesimally small change of angle. (This would represent a circle divided into 5292 segments).

Figure 2 shows a T joint. Consider a layer of fibres that will form part of a web having a width of "a", see Figure 5. The fibre architecture of the layer is then modified commencing at line "A" and as the fibres progress to line "B" the fibres are progressively and evenly spread out so that at line "B" their width is "2a". Between line "B" and line "C" the fibres remain parallel and the width "2a" is constant. Between line "C" and "D" the fibres are progressively drawn together so that at line "D" the width of the fibres returns to "a". The fibres then continue at constant width until they approach the junction with another web. The angle of the extreme edge fibre relative to the web is defined as 0.

It has been found experimentally that glass fibres in their dry fibre state can easily be made to take up an angularangular change of up to approximately 20° and as previously shown, the angular change from one fibre to the other remains very small.

Figure 6 shows a side view of the layer of fibres. At line "A" the thickness of the layer is "b" and would typically be in the region of 1mm. At line "b2 the %FVF will be the same as at line "A" as it is essential that the % FVF remains constant within the web. Therefore the thickness at line "B" and between lines "B" and "C" will be "b/2".

A second layer of fibres may be placed at a right angle to the first layer. The thickness of this layer is also "b" the same as the first layer and preferably will have the same % FVF, although this is not essential.

A plan view of the second layer is shown in Figure 7 and a side view in Figure 8. The layer has a width of "c" which is the width of the web. The fibre architecture of the layer is then modified commencing at line "G" and as the fibres progress to line "F" their width is "2c". Between line "F" and point "E" the fibres remain parallel and the width "2c" is constant. The angle of the extreme edge fibre relative to the web is defined as 0 and its value is constrained in the same way as it is for the first web.

Figure 9 shows the second layer placed on top of the first layer and the web width "a" and "c" have the same value. In this instance the distance between lines "B" and "C" is 2c and the distance between lines "F" to "E" is 2a.

The area bound by lines BC, EF, CB and BE is referred to as the node. Within the node the fibres within each web are parallel to each other and of constant FVF. The fibres of the first layer will be subsequently bonded by the matrix resin to the second layer and each layer in turn bonds to and stabilises the fibres in the two adjacent layers it is in contact with. Figure 10 is an isometric view of webs in Figure 9.

Webs having different thickness can be joined by adopting the same geometry as previously described, with the result that the node then becomes a rectangle rather than a square which occurs when the two webs are of equal thickness.

The webs are formed by laying down alternative layers of fibres until the dimensions of the webs reach their desired dimensions. The fibre structure described can be made by the apparatus disclosed in International patent application number PCT/EP2013/067560.

In Figure 10 it can be seen that the fibres of the second layer, Figure 7, stop at the outer edge of web, Figure 5. An alternative approach is to fold back the fibres of the second web over the first web as shown in Figure 1 1. This approach requires a more complex fibre assembly machine, but based on the same concept, as that described in International patent application number PCT/EP2013/067560. In order to accommodate the fold back of material at the outer edge and retain a constant fibre volume fraction, it is necessary to modify the dimensions of the node, see Figure 12. The dimension "E" to "F" which was previously "2a" now becomes "2a + b/2".

Figure 2 shows a T-shaped structure in which a web at right angles to the web it is connecting to. Webs entering the web to which they are joined can enter at angles less than 90°, as shown in Figure 13. To accommodate this variation, the node is modified by increasing the dimension from "B" to "C" so that the fibres in both webs remain parallel and of constant fibre volume fraction within the node.

Figure 14 shows a partial section on line "F" of three layers from each web assembled in their respective positions. As the fibres from the first web change their width from line "A" to line "B", at which point they have a thickness of b/2, they create an opening of thickness b/2 which extends between line "B" and "C" and which can accommodate a layer of fibres from the alternate web.

As the fibres change dimension from lines "A" to "B", a gap "Y" opens up which has a maximum depth of b/2. This appears to have no effect on the strength of the webs because by keeping the individual web layers thin, typically in the range of lmm, the gap is 0.5mm at its maximum point. The changes to the fibre direction are therefore kept very small and can be described as infinitesimally small. The %FVF in the fibre layers remains constant and at the node the layer of fibres in each web subsequently become bonded together by the matrix resin.

An L-shaped joint is shown in Figure 3. Referring to Figure 15, two layers of fibres are shown. The first layer commencing at line "A" the fibres progressively spread until at line "B" their width has increased from width a to 2a. From line "B" to "C" the fibres remain parallel.

The second layer of fibres commencing at line "F" progressively spread until at line "E" their width has increased from c to 2c. From line "E" to line "D" the fibres remain parallel. The fold back method shown in Figure 11 and 12 can also be applied to the "L" connection joint.

The area bounded by lines BC, DE, CB and ED is referred to as the node and complied with the conditions previously defined.

Figure 16 is a side view of the joint type shown in Figure 15 and shows how as the fibres are spread from line "F" to line "E" and their width increases from c to 2c their thickness decreases from b to b/2 and in doing so the fibre volume fraction remains constant.

Figure 17 shows a typical structure incorporating both "T" and "L" type connection joints. The structure comprises a rectangular frame having parallel long sides (50) and short sides (51), the cavity being divided by a cross-member (52). The frame has rectangular outer facing sides, L-shaped nodes (53) at each corner and T- shaped nodes (54) connecting the cross-member (52) to the cross sides (50).

Figure 18 shows a further embodiment wherein rectangular sides (55,56) are joined by L-shaped nodes (57) to form a smooth sided rectangular opening within which a closure, for example, the frame shown in Figure 17, may be received. In combination the structures shown in Figures 17 and 18 may form an access chamber and housing.

Figures 9 to 16 show embodiments of the composite material structure of the invention. In each embodiment of the invention the composite structures comprise a plurality of layers of fibres. In the first embodiment shown in Figures 9 and 10 the composite structure comprises two layers of fibres (5,6); wherein the first layer (5) of fibres extends in a 0° direction and a second layer (6) of fibres extend in a direction perpendicular to the first layer (5) of fibres. For simplicity of illustration, the layers of fibres are shown in rectilinear format. The layers of fibres can be of any shape, including cylindrical. The composite structure (5,6) has four layers of fibres (7,8,9,10) wherein the first layer (7) and third layer (9) of fibres extends in a 0° direction and a second layer (8) and fourth layer (10) of fibres extend in a direction perpendicular to the first and third layers.

In a perpendicular embodiment, the layers of fibres of the composite structure extending in the 0° direction overlap the layers of fibres extending 90° at a node point (1 1). In each of the embodiments the shape of the layers of fibres (7,8,9,10) is altered at the node point (1 1) to enable the layers to overlap each other without varying the fibre volume fraction. In Figure 10, the shape of layers of fibres is described in relation to the second and fourth layers of fibres (8,10). This description is applicable to all layers of fibres. The second and fourth layers of fibre (8,10) each have a specific height and width in the composite structure. As the layer of fibres (8,10) extends into a node point (1 1), the layer of fibres (8,10) gradually flattens out so that the height of the layer of fibres is decreased and the width) is increased relative to each other so that at the node point (1 1), although the shape of the layer of fibres is different, the volume of space occupied by the layer of fibres remains the same. In this particular embodiment the effective height of the layer of fibres (8) has halved, whilst the effective width has doubled. In this way the cross sectional area occupied by the fibres remains constant, as does the fibre volume fraction of the composite structure. Where the layers of fibres overlap at the node point (1 1) they are equivalent to the sum of the fibre volume fraction of each of the layers of fibres (7,8,9,10). The layers of fibres (7,8,9,10) progressively revert to their original shape as they extend beyond the node (11). As the layers of fibre (7,8,9,10) progressively change shape as they approach the node (1 1) small resin rich areas (14) develop within the structure. In order to keep the size of these resin rich areas to a minimum, the thickness of each fibre layer (7,8,9,10) is kept small, typically 1.0mm, and hence at the node (1 1) the height of the resin rich area (14) may be 0.5mm. It has been found that the fibre volume fraction of each layer remains constant as they approach the node and even though their shape changes.

Layers of both carbon fibres and glass fibres may be used. This construction results in a useful combination as the carbon fibres may be up to twenty five times more expensive than the glass fibres. Placing the carbon fibres at the extreme edges of the web optimises the increased performance provided by carbon fibre. Such fibre blending can be accommodated very effectively using the invention.

Figure 9 shows a plan view of a perpendicular node. A bundle comprising a single layer of parallel fibres of rovings (34) approaching a node (35). As the node (35) is approached, the transverse dimension of the bundle increases from a transverse dimension of the internodal bundle (34) through an intermediate portion (36) of increasing transverse dimension, to a transverse nodal dimension as shown in the nodal region (37). The fibres extend in parallel across the nodal region (37) to a second intermediate region (38) and return to an internodal transverse dimension at region (39). In addition to fibres of the bundle (34), a proportion of fusible fibres (40) is distributed on the surface and throughout the bundle (34).

A second bundle (41) overlies the first bundle (34). The second bundle (41) has an intermediate region (42) and a nodal region (35). In the nodal region the second bundle of fibres extend in parallel, perpendicular to the first bundle of fibres, to form a square overlapping node. The fusible fibres (40) are fusibly engaged and secured to the fibres of both of the first and second bundle in the nodal region, for example, at (45,46). The layers of fibres may be further secured when the preform is moulded into the finished engineering component or other product. The fibres of one bundle therefore provide perpendicular support to the fibres of adjacent bundles, relieving forces which may urge the fibres to move towards each other when the structure is under tension.

In the illustrated embodiment, the first and second bundles have the same dimensions so that the node (35) is square in plan view. The angle (47) between the parallel internodal fibres (35) and the fibres in the intermediate region (38) may be 20°-30°. This angle is preferably maintained at a minimum value to reduce interlaminar strains. However, nodal bonding (45) of adjacent fibres allows use of relatively large angles (47).

Figures 11 and 12 show an arrangement in which fibres of the second bundle are folded around the second side of the first bundle. In Figure 1 1, fibres of adjacent first bundles (7,9) are laid with fibres of the first bundle (20) laid between the first bundles (7,9) and are folded at the outer edge around the outermost fibres of first bundle (9) and then passed outwardly to form the second member (22) extending perpendicularly away from the first side (23) of the first bundle. In this way a smooth and coherent edge is formed by adjacent loops (21).

Figure 12 shows the arrangement in plan view. A gap (24) along the second side of first member (7) is provided to accommodate the width of the folded fibres.

Figures 13 and 14 show the configuration of a node in which the bundles are not perpendicular but are arranged at, for example, an angle of 110°. In this case, the intermediate regions are shorter in length on one side (19) and are longer on the opposite side (20). The node (21) has a parallelogram or rhomboid configuration.

Figure 15 is a plan view of an L-shaped corner section wherein outer edges of the corner are linear.

Figure 16 is a cross-sectional view of the L section joint as shown in Figure 15. Figure 17 shows a rectangular member having four corners and two mid joints as previously described. Figure 18 shows an alternative arrangement wherein inner surfaces of the corners are linear with the nodes extending outwardly.

The apparatus of this invention may permit a number of bundles or layers of fibres to be assembled with the transverse and perpendicular dimensions changed at each node in order to achieve a desired configuration. As each layer of fibres is laid down, it may be microbonded to retain the shape relative to the previously laid down layers in order to form a complete fibre preform.

Figure 19 is a schematic view of apparatus in accordance with this invention. A platen (60) is mounted on a support to provide linear motion in the X, Y and Z directions and rotation in the Z axis, if necessary. A tray (61) is located on the platen (60). A preform (62) is assembled on the tray (61). The preform (62) may be removed from the tray (61 ) and stored prior to completion of the moulding process, as required.

Two arrays of fibre bundle laying heads (63,64) are located along edges of the platen in the X and Y directions. The fibre bundles may be delivered from the heads in the form of tapes. The bundle laying process is carried out by positioning and raising the platen (60) under the row of X direction heads (64). The platen (60) is moved an appropriate distance in the Y direction to lay a series of parallel bundles (66) on the tray (61) to produce the Y axis webs of the preform. The bundles are then cut by cutting apparatus (not shown). The platen (60) is then lowered in the Z direction and a tray is positioned under the Y heads. The bundle laying process is then carried out in a similar manner to produce the X axis webs of the preform (65). The process may be repeated as many times as necessary to build up the layers of preform to a suitable thickness and configuration.

In an alternative embodiment, a single array of tape heads is employed with the platen being rotated through 90° between alternate bundle laying stages.

In Figures 20 and 21, a bundle or tape laying head is illustrated. A guide wheel (67) is mounted on a horizontal axis (68) and may be rotated in either direction ("A"). The circumferential face of the guidewheel (67) is provided with a main channel (69). The channel (69) has variable width and depth around the circumference of the wheel. The inner axially extending surface of the channel has a series of minor grooves or ridges (70). The axial spacing of the grooves or ridges varies in proportion to the width of the channel (69). The grooves or ridges serve to assist in guiding the structural fibres as the transverse dimension changes at the intermediate portions of the node following rotation of the wheel.

In Figure 22 the feeding of fibres to the guidewheel (67) is shown. Structural fibres (71 ,72) are brought from the creels (73,74) of the feeder and are directed into channel (69) of guidewheel (67) by a guideplate (75). A skeleton of fusible fibres (76) is supplied from a feeder creel (77).

The preform assembly (78) passes downstream from the guidewheel (67) (to the left as shown in Figure 20).

The structure of the skeleton (17) of fusible fibres is shown in Figure 23. Parallel longitudinally extending portions (79) are joined by spaced apart transverse members (80) to form an endless ladder configuration which is laid between alternate bundles of fibres (71,72) as shown in Figure 24 The structural fibres (71,72) and fusible fibre skeleton (79,80) are arranged so that the transverse fusible fibres (80) travel across the bundle as it is laid down so the skeleton is located on both sides of the bundle as shown in Figure 24.

When the skeleton has been heat treated the structural fibres of bundle (78) are held in the correct position. The consecutive layers of bundles of the assembly may be bonded together to form a semi-rigid preform.

The rotation of the guidewheel (67) is synchronised to the linear movement of platen (60) and tray (61). This permits the fibre architecture to be configured in a precisely controlled predetermined manner to form a microbonded preform.

Figure 25 shows the bundle laying head in further detail and a method of laying the tapes of fibres (71,72) to form a bundle (78) having one or more nodes. The structural fibres (71,72) are restrained in the channel (69) of the guidewheel (67) at the point where they reach the surface of tray (61 or the upper surface of a previously laid bundle (not illustrated). The bundle (78) then passes under a heated shoe (81). The shoe has a surface facing downwardly towards the tray (61) and arranged to engage the upper surface of bundle (78). The shoe (81) applies light pressure to hold the bundle (78) in position and also provides heat to activate the fusible fibres of skeleton (79,80). When the fibres and skeleton are located in the channel (69) in the guidewheel (67), the transverse and perpendicular dimensions of the bundle are fixed and correspond to the dimensions of the channel at the particular radial orientation of the guidewheel (67).

As the structural fibres of bundle (78) continue to be laid down as a result of the linear movement of the platen (in the direction of arrow C in Figure 23) the heated bundle, including the fusible fibres, passes under a cold shoe (82). The cold shoe (82) cools the bundle and solidifies the fusible fibres of skeleton (79,80). The hot shoe (81) is separated from cold shoe (82) by a layer of insulating material (83).

In Figures 26, 27 and 28 the hot shoe (81) and cold shoe (82) are shown in greater detail. Cold shoe (82) has an inlet (84) for compressed air which is fed through a transverse bore (85) to an array of outlet holes (86) with openings in the lower surface. The openings contact the bundle of fibres (78). Compressed air penetrates the bundle (78), rapidly removing the heat provided by the hot shoe (81), to cause solidification of the fusible fibres and thereby forming a rigid skeleton holding the fibres of the bundle in the dimensions determined by the channel (69) of the guidewheel (67).

Hot shoe (81) includes heater elements (87) to maintain a controlled elevated temperature. Alternatively or in addition, an array of holes may be provided to permit circulation of heated air through the bundle (78) in order to provide a more rapid and controllably responsive force of heat.

When an appropriate length of bundle (78) has been laid, it is cut to length. The process is then repeated for the X and Y axes as necessary to build up a desired bundle of layers microbonded together and laid out in each direction. The X, Y and Z motion of the platen (60) is provided by stepping motors (not shown). By this means, the discreet location of the platen (60) in all three axes is known and may be controlled by a microprocessor control unit during the manufacturing process. The angular orientation of the guidewheel (12) is also controlled by a stepping motor. Linking of the four motions by the control unit allows the cross-sectional dimensions of each bundle to be controlled at any particular location within the preform (79,80).

Specific programmes for the microprocessor and the number and location of the guidewheels may be changed to facilitate production of a wide variety of composite structures.

A typical gully grating made in accordance with the invention may provide a clear opening of 450 x 450 mm and be suitable to support a load of 400 KN as defined in European Standard El 24: 1994.