A balance between keeping weight and cost low while not having the floor fold in half at high speed
For aerodynamic components, weight and cost are easy parameters to quantify even before starting to manufacture any composite parts. We know how much material we are likely to use and approximately how much it weighs once combined with resin and various stiffening structures, mounting brackets, and fasteners. Our lap simulator will then give us an estimate of how this weight affects both average and instantaneous performance. Cost event points are awarded linearly, so we also can tell exactly how many points we will gain or lose compared to a baseline design without needing to know how well the other teams are manging their costs. Composite part strength is still a bit of an unknown on the team as we don't often make composite parts with a critical strength requirement, and when we do the philosophy has previously been to use an extremely thick core material, pile on the plies, and add aggressive external stiffening later if it still deflected too much. While developing composite simulation methods and data wasn't in the scope for this year, we decided that we would at least adopt a more structured approach (good pun?) to designing the layup. In other words, a non-zero amount of thought was actually put in before we broke out the gloves and hardener.
Maximum downforce expected on each of the two separately-mounted floors at any point was in the ballpark of 22kg, with drag at 15kg. The permanent chassis floor provided the rest of the downforce between the two side-mounted floors. The mounting for the side floors was designed both to adequately support these forces, and accommodate the desired easy-install/remove method.
Traumatised from once having to deal with an older team car with an undertray that took 2 people, 10 minutes, and a jacking bar to remove from the vehicle, I was determined to create a mounting system that required a single person, no jacks/stands, and could be done in a matter of seconds. This wasn't just for my sanity, but a number of other reasons too:
The Australasia competition has a tight schedule, so time between events has to be maximised. Access to the accumulator mounts was blocked when the floor was installed, so being able to remove it quickly directly impacts the available charging time between events.
Our trailer ramp and driveway curb outside the workshop are two obstacles that the car needs to be able to roll over with sufficient ground clearance, lest we would otherwise have to get 4 people to lift the whole car every time. The side floors extend slightly below the lowest point on the chassis, so having them be easily removable will encourage people to do so, limiting the chance they are left on and damaged due to rolling the car over large bumps or sharp angles.
The corridor into our workshop is only 150mm wider than the car. With the seat normally filled with all sort of parts, tools, helmets and bags, the only way to steer the car is by standing next to it, between the front and rear wheels. The implication and intent here is pretty much the same as the previous point.
The solution I came up with was a forwards/backwards slide-mount design using pins and slots, with longitudinal motion secured with a single bolted mount on each side floor (see the last sketch in Figure 1). This design would support very little bending moment however, so to support the outer extremity of the floor, 1mm steel cable with quick-disconnect hooks was used at the front and rear. Manufacture of the floors, their mounting, and the external suspension cover took two of us about three and a half weeks full-time in October, although much of this was spent experimenting, researching, and fixing mistakes as we improved our understanding and knowledge of best practises and helpful tricks.
Figure 1: Some original concept sketches for the chassis slide-mount design that I made once I knew roughly what I wanted to achieve
The stiffness of the floor was designed around the strength and location of the supporting mount brackets and cables. Some flexure is acceptable as the floor was intentionally designed to be insensitive to small changes, and a buffer was left between the "rigid" floor position and the ground at maximum steady-state vehicle roll. Too much deflection though and there could be excessive ground contact in high speed corners and a significantly increased expansion ratio resulting from the lesser-supported centre of the floor sagging down under load. This could lead to unexpected separation, or higher than expected forces which could snowball to the point of breaking a mount or permanently bending or otherwise damaging the floor.
From the start, the plan was to use carbon fibre twill as the sole composite fabric. Carbon was preferred over glass due to its higher stiffness per weight, and its more favourable layup behaviour compared to the softer and easily deformable glass fiber twill or even standard weave. Lap simulation suggested there was negligible difference in points between carbon and a structurally equivalent amount of glass, when considering both weight and cost. The only real downside of carbon over glass was the real-world cost the team had to pay, but this was only a minor consideration thanks to having a local supplier and ongoing team sponsor. Twill was preferred as the plan was to have at least three plies at any given point on the floor, meaning there was scope for multiple weave angles to be used, supporting stiffness in all key axes. This then left us with the benefit of the easier workability of twill weave, particularly in areas of complex (multi-axis) curvature and sharp radii down to 5mm.
The aerodynamic surface design utilised a "2D" flat sheet layup with no internal volume/mould or multi-surface construction. This was done to both minimise weight and simplify manufacturing, given the team had little experience in making high quality composite parts. This design practically mandated the use of a core included in the layup to add thickness to the floor to support bending. There were two mainstream options for this: cork or foam. Foam had the benefits of being cheap, and easily conformable to more complex surface curvatures. The downside was it was quite compressible, so didn't support bending as effectively as cork did in our tests, with results showing increasingly poor relative stiffness as bending load increased, leaving the risk of a sudden otherwise-undetectable failure on-track if loads were excessive. In the end I decided to use both materials. The foam would be used on the main inboard tunnel/diffuser section, and cork would be used on the flatter outboard skirt section. The inboard section was directly supported by the three chassis mounts and had naturally greater stiffness due to its multi-axes curvature, while the flat and less supported skirt needed a bit extra help. Layup diagrams were produced to communicate the final selected structure to anyone involved in the manufacture; details of a few key sections are shown in Figure 2.
Figure 2: (Top) Ply details from the composite manufacture drawings of the side-floors, and (Bottom) a visualisation of how the different twill ply angles stack up to provide the required stiffness in critical directions
In general, 3 layers of fabric were used. One on the underside of the core, and two on top to increase resistance to buckling given the floor had greater support at the ends than towards the centre. A strip of thin copper mesh was used surrounding the foam core in the area that lay within 100mm from the high voltage accumulator connections to meet grounding rules.
In Figure 2, detail A shows the interface between the main tunnel's foam core and the skirt's cork core. The thicknesses of each were 1mm different (2mm cork was sufficient and 3mm cork was difficult to source locally in a suitable quality) so the interface utilised a strip of carbon fabric to act as both a sealer for that area of the cork, and to reduce the aggressiveness of the step up to a more gentle ramp by supporting the main top-surface plies. Detail B shows the trailing edge of the main diffuser, with the foam core tapering off and two additional plies introduced to both seal the taper and add thickness. The four plies that form the trailing edge gave a thickness of 1.1mm when vacuum formed. Detail C shows the attachment of the smaller continuous strake (the "skate") that runs the length of the underside of the floor. This piece used an additional flexible foam mould after the main flat layup had cured and been removed from its mould.
Peel ply was used on the top "pressure" surface of the floor to provide a uniform, easily post-workable surface. Perforated plastic film was later used for the same purpose on the strake (which was not vacuum sealed for practical reasons), for which a smoother finish was desired due to the difficulty of conducting post-work on the thin part. The simple 90-degree flange and minimal other bends or curvature on the strake meant the plastic film was a valid option, unlike the main floor surface.
Figure 3: MDF moulds used for the layup, sans a final sealing/smoothing coat and the separate vertical mould blocks used to form the endplate and rear skirt wall
Additional separately attached parts consisted of the squirt suppression wishbone element, the brow wing, the diffuser flap, and the large diffuser strake. The diffuser strake was simple as it was intentionally designed to be perfectly flat. This meant it could be pre-cured in a flat compression mould giving a perfect finish. The cured part was then trimmed to shape using a printed template and mounted to the diffuser roof using a single strip of carbon tape and resin on each side. This simple attachment method was sufficiently strong given the combined air pressure on either side of the strake was expected to be close to equal under all driving scenarios, including high yaw angles. CFD suggested a maximum of 200 grams of lateral force on the strake, which it held just fine in an experimental test setup. The brow wing and flap were manufactured the same way: a 2D polystyrene foam core cut with a hot wire and templates/guides, sealed with spray adhesive to prevent excessive resin absorption, then covered with a single layer of fibreglass fabric (to avoid the grounding rules issues that come with using carbon) and perforated plastic film in a vacuum bag. The squirt element was similar but did away with the fibreglass layer and just had resin applied to seal the part, given the low local forces experienced. The trailing edge was then trimmed to approximately 1mm thickness, adjusting slightly if necessary to ensure symmetry between each part for both side floors. By trimming the trailing edge after manufacture instead of adjusting the hotwire template to cut a 1mm trailing edge, any accidental light damage before laying up and curing the glass became inconsequential.
The flap was bolted securely to the floor on both sides as it formed a necessary addition to the overall stiffness of the floor, particularly in the diagonal axes. After deciding against incorporating the passive DRS design, its mounting was semi-permanent, but adjustable to a single pre-set angle for the acceleration event (it was kept in maximum downforce configuration for endurance after consulting the lap simulator). The brow wing and squirt suppression element on the other hand were the complete opposite. Both were at risk of being hit at speed by cones, and without adding weight by significantly bolstering the strength of these parts and their mounts (and the floor too in the case of the brow wing), this would pretty much guarantee some sort of damage. Given this was the case, I decided to go down the route of making the mounting for these parts as weak as I could get away with. The idea was that in the case of a cone collision, either the part or its mount would break free entirely from the car. For the brow wing especially, this was important to avoid irreparable damage to the floor which has several hundred dollars of material per side and nearly 40 man-hours of manufacture and post-work. A simple, small, 2D wing profile though? I made half a dozen spares of each in a few hours with some foam from the recycling bin and $15 of fibreglass.
Finally, the mounts were aligned in-situ and inserted between the plies in a section of removed foam core material before being glued and riveted in place.
Figure 4: Underside of one of the side-floors showing the main diffuser strake, front-rear "skate" behind, and a rake mounted in the diffuser in the background for upcoming tests. The sharp leading edge on the strake showed no separation issues in any CFD test case, but flow-vis tests will reveal whether this is reflected in reality.
Figure 5: Trailing edge and underside of diffuser flap, which was painted black to hide the fibreglass
Figure 6: Finished floor assemblies weighing just 1.0kg each post-curing in the sun at at a worryingly windy test track
Overall, the manufacture and structural aspect of the floor were impressively successful given this was our first attempt at trying something of this scale and quality. As with all things though, we identified a few areas for future improvement.
Cable mounting locations were not ideal as they didn’t really consider the effects of deflection in the floor which was a mistake. The deflection allowed the cables to move perpendicular to their tension direction, observed under high speed testing with the front of the floor lifting and shifting rearwards, allowing the centre section to sag more than intended. Cable mounting is still the preferable option over heavier and larger rigid options (e.g. carbon rods), but it will benefit from some more careful placement of the pickup points both on the chassis and the floor, with respect to the axis of lowest stiffness.
The chassis mount brackets and studs worked well in concept for their ease of use but the aluminium brackets had significant deflection under load due to the loading not being perfectly aligned vertically. Resulting drawbacks were increased deflection in the floor and increased friction when attempting to slide the floor into or out of its position. This will be solved for next year by simply increasing the tab thickness. Potential exists to have the chassis-stud on the inboard side of the chassis and have the floor mount bracket protrude up through the chassis floor will be explored, as this will reduce the disruption to airflow on the pressure side of the floor.
The diffuser flap mount will also be in the scope for improvement next year, with the ideal implementation seeing the mounting interface on the side of the floor rather than the top, to reduce disruption to the flow over the suction side of the flap.
Lastly, quality of the clear coat was severely lacking from a combination of limited prior practise and running out of time to build up the required thickness. While it's not expected to have a significant detrimental effect on performance (given the worst offending location is the pressure side of the skirt), it's simply not pleasing to look at. This will just come down to more time and experience, and possibly a more suitable resin consistency and curing environment, instead of just re-using the same laminating resin.
Figure 7: Ten months of work; still plenty to do yet...