Development of an automated fibre placement-based hybrid composite wheel for a solar-powered car

Substantial range, handling and acceleration improvements in high-performance vehicles can be achieved by weight reduction. An important area for weight reduction on a car is the wheels. A novel prototype carbon fibre/epoxy wheel has been developed using a combination of automated fibre placement (AFP) and hand layup for the Sunswift 7 solar car. A three-piece wheel design that utilises each process where best suited has been analysed and optimised using the ANSYS ACP PrepPost suite, manufactured, and mechanically tested. The wheel disc was produced using AFP and featured selective reinforcement in the form of spokes. The AFP fibre paths for the disc have been optimised using CGTech’s VERICUT VCP and VCS to minimise gaps and overlaps, resulting in a 98.9% reduction in overlaps when compared with the unoptimised layup. The rim and tyre mounting region of the wheel have been manufactured using hand layup and adhesively bonded to the disc. This hybrid manufacturing approach has demonstrated an advancement in the feasibility of combining traditional and automated composite manufacturing. The final wheel weighed 3352 g, and the wheel deflection under a compressive load has been experimentally verified within 3% of the theoretical value.


Introduction
Fossil fuel remains the predominant energy source for the transportation sector around the world [1]. Ever stricter global emission laws as well as increasing consumer awareness to sustainability and the environmental impact of vehicle emissions [2] have driven the advancement of alternative energy sources for vehicles, with electric vehicles (EVs) being at the forefront of this movement. Solar electric cars represent an advancement over regular EVs, utilising photovoltaic cells to generate electricity from solar irradiance, and have been identified as important for the creation of new photovoltaic infrastructure [3]. Due to the relative inefficiency of current solar panel technology as a means of energy conversion [4], it is critical to find efficiency gains in other areas of the solar car. A primary method for this is weight reduction, as a lower-mass vehicle requires less energy to move. Fibre-reinforced plastics are already wellknown for their applications in the automotive industry, primarily due to their superior strength-weight ratio when compared with traditional materials such as steel and aluminium [5][6][7][8]. There has been a major shift towards using fibre-reinforced composites in solar car components in recent years all over the world [9][10][11]. As wheels have an additional inertia component (rotational) when compared with other parts of the car, they are a key area for optimisation. The weight reduction of wheels not only offers decreased energy consumption, but also improved suspension response and vehicle handling characteristics.
Czypionka and Kienhöfer [12] developed and validated an FE model of a composite car wheel using ANSYS ACP. A prototype wheel was manufactured, and a dynamic cornering fatigue test was conducted. This test applied the static equivalent of a cornering-induced bending load to the wheel, with strain gauges used to measure the resulting deflection. The authors found a good correlation between FEA results and the physical testing, validating their model. Following the final layup optimization, the authors achieved a theoretical weight reduction of 18% over an identical aluminium wheel. Wang et al. [13] conducted a detailed design optimization and analysis of a hybrid wheel using a CFRP rim and aluminium disc. A multi-objective optimization algorithm was employed to optimise layup thickness, sequence and angle. A prototype wheel was constructed and physically validated using a radial fatigue test and a 13° impact test. Importantly, the optimization scheme reduced the Tsai-Wu failure indices by between 14.4 and 25.8% compared with the initial layup, clearly highlighting the benefits of the optimization process. Research into composite wheels in the aerospace industry has also been conducted. Wacker et al. [14] developed a wheel concept for the nose wheel on an Airbus A320 aircraft. Whilst other studies have focused on developing a wheel with minimal components, in this study, the authors use four composite pieces manufactured using hand layup of prepreg fabrics to create a wheel that can be disassembled for ease of maintenance. To accomplish this, the authors designed and physically validated a bolted joint that was used to connect the parts of the wheel. Overall, a theoretical weight reduction of 27% was achieved; however, no physical prototype was constructed. Rondina et al. [15] investigated the automated manufacturing of composite wheels using high-pressure resin transfer moulding (HP RTM). The authors found a reduction in manufacturing time compared with an autoclave method and validated the mechanical performance of the HP RTM material system.
Composite wheels are also becoming commercially available on passenger vehicles. The first company to market with such as wheel was Koenigsegg [16]. Their patented 'Aircore' technology uses hollow spokes to achieve a 5-kg weight reduction per wheel compared with the standard aluminium wheel. Additionally, Australian-based Carbon Revolution [17] designed and produced a carbon fibre wheel using resin transfer moulding that is used by global vehicle manufacturers including Ford, Ferrari, and Renault. Likely due to the proprietary technology used, there are limited technical details available about either the Koenigsegg or Carbon Revolution wheels. Porsche [18] also developed a carbon fibre wheel for the 911 Turbo model. This wheel utilised a combination of hand layup and radial braiding to create a preform that was later impregnated with resin and cured. The finished wheel achieved a weight reduction of 2.15 kg per wheel over the standard forged aluminium wheel and 20% higher rigidity at the time of cracking. However, at a price of $17,600 USD per set [19], these wheels are significantly more expensive than steel or aluminium wheels.
Manufacturing of current commercially available carbon fibre wheels is labour intensive, contributing to their high price. For example, Koenigsegg relies on hand layup of prepreg fabrics to construct their wheels [20]. Efforts have been made to automate the production of composite wheels. In particular, Carbon Revolution uses automated manufacturing processes to create dry fibre preforms that are subsequently placed in a mould and impregnated with resin in an injection process [21]. An alternative automation strategy is the use of Automated Fibre Placement (AFP). This technique uses the selective placement of narrow carbon or glass fibre tapes to construct a composite laminate. Popularised in the aerospace industry, AFP improves productivity whilst providing high accuracy and quality as well as reduced wastage [22]. Hence, AFP has the potential to produce wheels at a lower cost and faster rate than currently used processes.
This study demonstrates the design and analysis of a carbon fibre-reinforced composite wheel for the Sunswift 7 solar racing car made using a hybrid approach by combining AFP and hand layup techniques. This novel prototype design aimed to evaluate the feasibility of incorporating AFP into the wheel production process and simultaneously demonstrate the thick laminate and selective reinforcement manufacturing capability of automated fibre placement. A detailed analysis is carried out using ANSYS ACP Prep-Post. The wheel geometry is optimised to reduce deflection from vehicular loads. VERICUT Composite Programming/ Simulation (VCP/VCS) software was used to check the manufacturability of the optimised design with AFP. The manufactured wheel was tested for deflection and strain and compared with the predicted results. The manufactured wheel weight was 3352 g and provides a valuable and novel insight into thick laminate manufacturing and the integration of AFP into car wheel design.

Conceptual design
The composite wheel was designed to replace the existing metal wheel. For compatibility with the solar car, and to mount the designated Bridgestone Ecopia solar car tyre, the wheel had to conform to a set of prescribed dimensions. The wheel dimensions are provided in Table 1.
The wheel design consisted of three components that were permanently joined during manufacturing. The components were named the inner rim, outer rim, and disc. These components are depicted in Fig. 1a. The disc refers to the part of the wheel that is typically an arrangement of spokes on a traditional car wheel. Both the inner and outer rims refer to the tyre mounting surface, with the inner rim sitting on the inboard (towards the centre of the car) side of the disc. The disc was curved inboard to achieve the required 35 mm mounting offset as well as provide increased rigidity (Fig. 1b). This three-component approach was taken due to the challenges associated with the inability of the AFP tow placement roller to access the highly curved bead seat (R = 3 mm) where the tyre sits. For this reason, the disc was made using AFP, whilst the two rim halves were made using a hand layup of prepreg fabrics. The shape of the two rim halves was driven by the desire to maximise the contact area between them and the disc. To achieve this, each rim half transitioned into a flange for connection to the disc as shown in Fig. 1b. The rims were attached to the disc using a structural adhesive. The bead seat geometry was defined by the ISO 4249-3 Motorcycle Tyre and Rim standard [23], herein referred to as the standard. Notably, each rim angles down towards the connection flanges. This creates a drop in the centre of the wheel that is necessary for tyre installation. Five spokes on the disc were also incorporated to provide additional stiffness to the wheel; these spokes extended out from the centre of the rim until the point where the outer rim overlaps the disc. Five spokes were chosen as it allowed the spoke pattern to line up with each of the 5 mounting holes.
The wheel mounting was designed to fit the solar car (5 bolts on an 86-mm PCD, as described in Table 1). The front and rear faces of the wheel were covered using faceplates made from a 1.6-mm thick aluminium sheet as shown in Fig. 2. This distributed the load from the wheel nuts into the wheel and acted to prevent surface damage from installation and removal of the wheel. Similarly, an aluminium sleeve was inserted into the central wheel bore to protect the carbon fibre. The wheel was secured to the car using 5 × M10 studs

Numerical analysis
Finite element analysis was undertaken in two parts to reduce the computational complexity arising from the large assembly. At first, a global analysis of the wheel was carried out focusing on failure, deflection, and the connection between the disc and rims, followed by the analysis of the wheel mounting, focused on the wheel/hub interaction.

Global analysis
A CAD model was created in SolidWorks for the geometries presented and subsequently imported into ANSYS SpaceClaim where it was converted into shell geometry. A 119,023-element mesh was generated using SHELL181 elements, with refinement around the important bead seat area. Mesh convergence was checked by varying the mesh size between 82,022 and 319,370 elements. Over this range, maximum fibre stress and deformation under the bump load case varied by 2.6% and 1.2% respectively.

Loading conditions and constraints
The wheel was analysed under three different load cases representing forces applied to the car due to braking, cornering, and bumping (vertical force due to hitting a road bump). These loads were pre-existing forces defined for the solar car. A fixed support was applied to the central portion of the wheel. All loads were applied via a remote force at the rolling radius of the tyre of 271.5 mm. Braking and bump loads were applied to a 50° arc of the bead seat area on both halves of the rim. The cornering load was applied to a 180° arc of the bead seat to one-half of the rim. Additionally, 0.55 MPa tyre pressure was applied to all interior surfaces on the rim in all cases. The loading conditions are depicted in Fig. 3, note that each of the three forces was applied one at a time. A 50° arc was selected based on previous experimental research conducted by Stearns et al. [24]. The authors determined that the pressure on the rim bead caused by the loading acts on an 80° arc on a 16″ rim (the same size being designed). This result accounts for the fact that the tyre helps distribute the pressure over the rim. Thus, a 50° arc without factoring in the tyre provides a reasonable estimate.

Carbon fibre laminate
The ANSYS ACP PrepPost module was used to define the layup on the wheel. Three epoxy prepreg materials were defined. For the disc, Park Aerospace HTS45E23/E-752-LT material was used. For the rims, Delta-Preg T700/DT806R UD fabric and T300/DT806R 2 × 2 twill fabric was used. Material properties are given in Table 2. The layup was defined in three zones: rims (inner and outer), disc and spokes (see Fig. 4a). Figure 4b depicts the fibre directions for each of these zones. Given the large number of design variables, the layup optimisation was simplified using geometric constraints. As shown in Fig. 4, the outer rim was sized to fit the tyre valve with clearance for attaching a pump head. The rim thickness was set close to 4 mm to allow for installation of snap-in-valves as per the standard. This then   constrained the disc thickness to approximately 10 mm to achieve the required mounting offset. The spoke thickness was varied to achieve the required strength and stiffness. A parametric study on spoke thickness using the vertical load case was performed, as this case induces the highest stress. The number of spoke plies was varied from 0 to 40 (thickness ranging from 0 to 5.2 mm), and maximum deflection, maximum strain, and inverse reserve factor (IRV) were recorded. For this study, Tsai-Wu failure criteria were used for IRV calculation. The IRV represents an inverse factor of safety, a value of 0.5 translates to a factor of safety of 2, and values over 1 indicate failure.
The results of the comparison are shown in Table 3, note that values in the 'difference' rows are percentage changes from the reference values. As the spoke thickness increased, total deflection, strain, and IRV decreased. Mass increased at a linear rate with spoke thickness. A 30-ply (3.9 mm) thick spoke was selected as it provided a similar decrease in deflection to the 40-ply spoke and closely matched the thickness of the rim, adding to the visual appeal of the wheel. This is shown in Zone 3 of Fig. 4a.
The layup for each zone is given in Table 4, where orientation is denoted as θ . In the rim, 0° fibres were used extensively to resist bending whilst 90° fibres were also incorporated to increase hoop strength and aid with resisting the air pressure load. The disc used a symmetric layup of [0/45/90/ − 45] whilst the spoke plies were oriented radially as shown in Fig. 4b.

Cohesive zone modelling
The adhesive bond between wheel components was simulated using a cohesive zone model (CZM). Contact debonding was set up between the flange on each rim and the accompanying contact area on the disc, corresponding to the red regions in Fig. 1b. The CZM material was defined using a fracture-energies-based method. The values used are provided in Table 5.

Summary of the global analysis results
Maximum deformation occurred at the edge of the inner rim in all cases, closest to where the load was applied. Figure 5 shows the lateral and radial deflection of the wheel for the bump and cornering load cases in the area where the load is applied. The maximum lateral separation of the bead seats occurred on the vertical load case. A maximum lateral deflection of 0.60 mm on the outer rim and a maximum of 0.20 mm in the same direction on the inner rim were observed. The ISO standard [23] used for the design of the wheel specifies a maximum deviation in the bead seat separation of + 1 mm to maintain secure seating of the tyre. Thus, the net deformation of 0.6-0.2 = 0.4 mm satisfies this requirement. Additionally, a maximum radial deformation of 0.85 mm  was found. The ISO standard does not specify a diametral tolerance on the wheel, so this result was deemed satisfactory as it is highly localised to the load application area. The rim-disc bond was analysed using the CZM contact. The results indicated no debonding under any load case. Figure 6 shows the normal pressure and shear stress in both the inner rim/disc and outer rim/disc joint in the bump load case, where maximum stress occurred. A maximum absolute normal pressure of 17.49 MPa (pulling the surfaces apart) and maximum tangential stress of 1.10 MPa fall below the allowable stress limits of 60 MPa and 35 MPa respectively. Additionally, ANSYS reported the CZM contact status as 'sticking' at all locations in all load cases. This indicates that the bond is intact.  Figure 7 shows the IRVs for the bump load case, with the highest IRVs focused on the rim where the wheel is loaded. As failure occurs at 1, this represents a minimum FOS of 3.03. In all load cases, the maximum IRV occurred on the edges of the rim where the tyre mounts.

Wheel mounting analysis
The wheel mounting area was extracted and analysed separately to reduce the computational complexity. The wheel mounting area uses aluminium fittings bonded to the carbon fibre wheel. The simulation of this interaction is complex, so a simplified model was used. The surface pressure on the carbon fibre was evaluated from the bump and cornering loading conditions and compared with the out-of-plane compressive strength for the material. The braking load was not analysed as it is small in comparison to the other loads.

Loading conditions and constraints
The mounting area on the wheel was extracted along with the rear face plate, compression tubes and centre sleeve, as shown in Fig. 8. The front face plate was not included, and loads were applied directly to the carbon fibre; this removes the pressure distribution effect of the faceplate, adding conservatism to the model. A 53,380-element mesh using SOLID186 elements was generated, and the convergence feature in ANSYS Mechanical was used with an allowable change of 2% to check for mesh convergence. For the cornering load, a remote force was applied to a 20-mm diameter circular area around each mounting hole, corresponding to the diameter of an M10 washer. For the bump load, a remote force was applied to the inside faces of the compression tubes. The rear faceplate was set as a fixed support.

Mounting analysis results
The pressure on the wheel was evaluated for the bump and cornering load cases. As shown in Fig. 9, a maximum of 30.84 MPa and 49.25 MPa was found for the bump and cornering load cases respectively. Given the 90° compressive strength of 341 MPa for the material (as provided by the manufacturer), the maximum stress failure criterion coefficient can be found as follows [26]:  Thus, failure is not predicted. It is acknowledged that the maximum stress failure criterion is simplistic and does not account for interaction terms; however, the large FOS of approximately 7 was considered satisfactory for this efficient analysis.

Automated fibre placement-based manufacturing of the disc
Automated fibre placement (AFP) is an advanced manufacturing method for making composite components in which several manufacturing stages are incorporated in the placement head, as shown in Fig. 10. The AFP robot places thin 49.25 341 = 0.144 < 1 tows onto a substrate using a feed system and compaction roller. The incoming material is heated to improve adhesion to existing layers, and a knife cuts the tows, allowing for tailored material deposition. In this study, the automated dynamics-built AFP machine at UNSW Sydney was utilised for making the disc. This AFP machine is a sevenaxis robot platform (a six-axis Kawasaki robotic arm and a coordinated spindle), a thermoset placement head which comprises a compaction roller, prepreg tape dispensing system and heating system (HGT), and a computer controller [22]. The placement head can feed up to four 6.35 mm wide tows simultaneously.

Fibre path generation
VERICUT Composite Programming (VCP) software, from CGTech, was used to generate the fibre paths for manufacturing the AFP disc. Various analyses including ply angle deviation, steering radius violation and gap/overlap  Fig. 11. For the spokes, a guide curve was projected onto the tooling surface, and parallel steered paths were generated based on the guide curve.

Path geometry/layup analysis
Overlap and gap analysis was performed in VCP to ensure there are no excessive tow overlaps or gaps that could affect the overall net shape and integrity of the structure. Initially, rosette role path geometry was used to generate tow path trajectories. However, using this approach, overlaps and gaps of 4281.44 mm 2 and 12.04 mm 2 respectively per ply were simulated for the 0° ply orientation. The analysis of the other path geometry algorithms showed that these values can be optimised using the parallel path geometry technique (e.g., total overlap/ply = 45.06 mm 2 and total gap/ply = 32.23 mm 2 for the 0° ply orientation). Thus, the parallel path approach was used to minimise gaps and overlaps. Ply angle deviation was also studied using VCP. Theoretically, the parallel path approach would not produce significant ply angle deviation on a two-dimensional plane surface. However, as the tool surface is three-dimensional convex, tow trajectory deviation of up to ~ 8° was simulated. Figure 11 illustrates VCP analysis on ply angle deviation where the ply angle variation exceeds 3°. The angle deviation on the spokes is not shown as it was zero. The AFP machine kinematics were finally visualised and analysed in VERICUT Composite Simulation (VCS) (Fig. 12). Before the actual AFP operation, erratic AFP head movements, as well as collision clearances, were checked, and trajectories were replanned and remedied back in VCP. Finally, the placement and process information were exported as numerical control programs for the AFP machine.

Fibre placement using AFP and curing
The disc manufacturing is shown in Fig. 13. A matched male and female mould were manufactured from aluminium. The male mould was cleaned with isopropyl alcohol and prepared with TR-930 release agent before AFP layup. The mould was first secured onto the rotatable paddle, as shown in the simulation in Fig. 12 using double-sided tape. The tooling surface was preheated using a heat gun. The AFP processing parameters of deposition rate, consolidation force and hot gas torch temperature were set to 76 mm/s, 180 N and 200 °C (~ 70 °C at the nip-point) respectively. Fig. 11 Ply angle deviation analysis using VERICUT VCP Fig. 12 Lay-up simulation using VERICUT VCS Three manufacturing issues were encountered during the AFP layup. Firstly, the mould was not exactly aligned on the paddle, causing the incorrect placement of the tows. This was corrected after the first layer; however, it could have been avoided by including some form of alignment feature into the mould design that indexed the circular part to the rectangular paddle. Secondly, the first tows to be placed suffered from heavy distortion (Fig. 13a). This may have been caused by the compaction pressure in subsequent plies moving them on the heavily curved edge of the mould. Thirdly, the inside ends of the spokes had a large degree of variation in their length. Figure 13a shows this area on the disc before curing. Gaps, overlaps and inconsistencies are clearly visible in this region. These errors did not affect the final wheel as this region was removed to create the central bore.
A female mould half was placed over the disc for curing to ensure good control over the final surfaces. Due to uncertainty in how well the spokes would be placed by AFP, the spoke geometry was not machined into the female mould. Instead, inserts made from 3-mm plywood were wrapped in adhesive-backed PTFE fabric and used to control the spoke shape. As shown in Fig. 13b, the inserts were retained using high-temperature tape. The inside area of the spokes where defects occurred was filled with epoxy adhesive and a hand-shaped insert to prevent resin flow into this area. The mould assembly was cured at the material manufacturer's recommended cycle, with an additional 60-min hold at 82.2 °C ± 2.8 °C during the initial temperature ramp to provide more time for the thick laminate to heat homogenously. This was recommended by the material manufacturer. The cured part possessed localised resin extrusions where the plywood mould plants did not correctly align with the geometry and resulted in lower compaction; this produced resinrich build-up areas as shown in Fig. 13c and fibre waviness along the edge of the disc in these regions which had to be removed with a grinder. The centre bore, mounting holes and bevelled outer edge were produced using 5-axis waterjet cutting, as shown in Fig. 13d. The part was also sanded to remove surface imperfections and sharp edges. The final part weighed 2081 g.

Hand lamination of the rim
The manufacturing of the rim was completed using a traditional hand layup process. Two male aluminium moulds (inner and outer rim) were designed and machined. Vertical surfaces on the mould were given a 1 o draft angle to aid with release. Wooden female moulds were machined from plywood sheets glued together using PVA wood glue to form a block. The wooden moulds were machined in 2 halves and joined together, with a laser-cut template used to ensure concentricity. The join line was sealed with epoxy adhesive, and the moulds were coated with 2 layers of polyurethane (Cabot's CPF Floor) to aid release agent application and minimise resin impregnation into the wood. The moulds were prepared for layup in the same way as the disc mould.
Each layer of the rim was segmented into 8 pieces of 45 o . A 2D flat pattern of the mould surface was extracted from the CAD model and used to cut sections from the prepreg fabric using a CNC fabric cutter as shown in Fig. 13 Details of disc manufacturing Fig. 14a. As the thickness of the part increased, the size of the fabric sections was increased appropriately. The rims were laid up as shown in Fig. 14b and vacuum bagged with the metal mould on top, as the weight of the aluminium helped with mould alignment and compaction. The rims were autoclave cured at 90 °C for 3 h, excluding temperature ramps. The parts were hand finished to remove flash lines. The final weights were measured at 617 g and 540 g for the inner and outer rims respectively.

Assembly
Aluminium sleeves were press fit into the 5 mounting holes, and the centre bore and face plates were adhesively bonded to the front and back mounting surfaces. The three wheel components (disc and rims) were assembled in two stages using a jig, as shown in Fig. 15. The jig consisted of a lower piece, upper piece and centre cylinder that held the wheel components concentric whilst fixing the mould surfaces of the inner and outer rims in place. Firstly, the inner rim and disc were bonded, followed by the outer rim and disc. The bond surfaces were sanded with 80-grit sandpaper and cleaned with acetone before 3 M DP460NS Off-White epoxy adhesive was applied. Each joint was left to cure for 24 h at ambient temperature, and the final prototype is shown in Fig. 16. The assembly jig, once tightened, squeezed glue out on the front and rear faces and dried in place indicating good adhesive dispersion in the contact area. This is shown in Fig. 16c where the white adhesive can be clearly seen on around the bond area. With all components attached, including the valve, the final wheel was weighed at 3352 g.

Mechanical testing
To validate the numerical model, a quasistatic compression test was performed using an Instron 8804 testing system. As shown in Fig. 17, the wheel was placed outside face up on a steel plate on the lower jaw of the Instron. Two strain gauges were installed on different spokes, with one close to the centre of the wheel and the other close to the outer edge. The compressive force on the wheel was varied between 200 and 5000 N each 3 times, and the force, displacement and strain results were recorded.
The test results are shown in Table 6. The strain results along the fibre direction were extracted from ANSYS by probing the top position of the outermost spoke ply at the measured strain gauge locations. The results are compared with the measured strain data in Table 5. The outer strain gauge correlated more closely to the prediction than the inner gauge. The difference can be attributed to several factors such as the strain gauge alignment (both angle and centring on the spokes), physical measurement of the gauge location and the fact that the actual gauge takes an areal average value, whereas the values retrieved at ANSYS are from a single point at the centre of the gauge.

Microscopy
Material offcuts from the external edge of the AFP disc were used for the examination of the layup quality. Figure 18a shows a macro photograph of an offcut section   passing through one of the resin-rich build-up areas depicted in Fig. 13d. The fibre waviness is clearly visible in this region.
The ply thickness of the thick disc laminate was also examined. Four samples from the disc edge were extracted from material offcuts and embedded inside epoxy pucks. The pucks were ground and polished to a mirror finish using a Struers TegraPol-15 and photographed at 10 × magnification using a Keyence VK-X200 laser microscope. An example micrograph is shown in Fig. 18b. The average ply thickness across 10 locations was measured to be 0.145 mm with a coefficient of variance of 7.94%. This is a difference of 10.9% to the theoretical value of 0.13 mm used.

Conclusion
A novel hybrid automated fibre placement and hand layup wheel for a solar racing car has been designed, analysed and constructed. The final product successfully demonstrated the incorporation of AFP into the wheel manufacturing process. Only the disc was manufactured using AFP as the robot used was deemed incapable of placing fibre on the complex rim profile in any orientation. The three-piece design assembled well; however, the assembly jig prevented access to remove glue squeeze-out, resulting in a poor aesthetic. Poor mould compaction from the use of plywood plants also created resinrich pockets that had to be removed with secondary machining operations and introduced fibre waviness in small regions of the disc edge. The mechanical performance of the wheel has been successfully validated using experimental methods, and the ply thickness of the thick laminate has been calculated.
Future work using the selective reinforcement capability of AFP should be pursued. It is hypothesised that a highly optimised wheel design could be achieved by using a hand layup technique for the bulk shape of the wheel and AFP for precise and selective reinforcement of key areas such as the spokes. This could still result in increased manufacturing efficiency as shapes that are complex and time-consuming for hand layup, such as the spokes could be replaced with simple flat surfaces that are subsequently reinforced using AFP. Other manufacturing strategies utilising AFP for the disc and a different technique more suited for round parts for the rims such as filament winding or radial braiding should also be investigated.