Surface design

I guess you could say this is the most important part. Not least because it finally stops people asking "is that a go-kart?"

Design of the aerodynamic surfaces is probably the least important part of this first year in our five year plan. As long as we got something generating enough forces that it had non-negligible effect on the suspension and general vehicle behaviour, and something that allowed us to easily test and improve CFD correlation, we would satisfy most of our goals for this year and we would be happy enough. With this in mind, I didn't want to waste any time with the likes of automated optimisation, or 3D aerofoils, as these are all "last 10%" sort of things. Pushing to get the last bit of performance out of a floor where force magnitudes aren't that high isn't worth it for us, especially when we don't have wings on the car; if we simply wanted more downforce we'd get much more value for our time by slapping on some unoptimised wings instead. The priority was rather to get something designed, built, and on the race track as soon as possible. 

In terms of flow field targets, I had nothing specific in mind as we have no useful flow features from upstream to utilise, or aero surfaces downstream to consider for the floor design. Generally, I was expecting something like this:

Instead of setting overall flow field goals, I made a list of aspects for the team to focus on or otherwise consider:

These are in addition to the requests I had at the start of the year for the base of the chassis to be narrowed, and chassis node angles to be minimised. This was to give maximum space for side-mounted aerodynamic devices, and promote flow attachment and thin boundary layers over the whole chassis respectively. 

I had always planned to focus on a surface-based floor (i.e. not aerofoil-based, with no thickness other than that of the composite material layup), both for ease of manufacture and keeping weight minimal to align with our design philosophy for this year (in simple terms, where two or more design options offer the same estimated vehicle performance within 1%, choose the one with lightest weight). I still wanted to explore an aerofoil-based option in case it turned out to be vastly superior, but I had doubts. As well as being lighter and easier to make, by creating a custom floor surface profile the ground proximity can be extended for as long as desired to make best use of the Venturi/ground effect.. 

Like all good aero design processes, we started at the front to decide where a good initial guess for the floor inlet should go. The fore/aft plane was positioned as far forward as possible to maximise available plan-view area for the floor. The height, width, and position were chosen to utilise the maximum amount of clean mass flow based on analysis of total pressure under the four main test cases. The wheel wake from the four cases was superimposed (Figure 1), and the first inlet shape was constructed to entirely fill the clean window between the wheel, lower wishbone, and rocker wakes. Another interesting feature in Figure 1 is the huge difference in lower wake and tyre squirt shapes and sizes between the four scenarios, resulting from different dynamic cambers, steering angle, tyre deformations, and cross-flow, highlighting the extreme sensitivities I've mentioned in earlier posts. You may also notice the mid wake no longer extends inboard anywhere near as far as it did on EV23; this is a combined effect of running slightly more toe on EV24, and moving the brake calliper from the rear to the front of the shell, both contributing to increased flow through the wheel shell from inboard to outboard. Relocating the brake calliper had two other benefits, in easier manufacture of the upright and in creating more space to mount a new and improved (albeit slightly larger) laser wheel speed sensor. 

Figure 1: Superimposed total pressure "wake" and vehicle positions for the four main CFD test cases, revealing a common envelope of "clean" air to utilise for the floor entry

Next up, the outlet shape was specified. Here, the main limitations were keeping the width inboard of the rear tyre to avoid a blockage effect, while maximising the height and cross-sectional area to get the largest area ratio (and therefore peak suction) without inducing separation. The rest of the floor was set up to have transition down to minimum cross-section just forward of the midpoint, maintaining both smooth surface transition and smooth area transition. Width was expanded at the centre of the floor (and height reduced accordingly) to maximise both the ground effect and the surface area for low pressure to act on. A skirt was extended out to the track width regularity to utilise the high pressure build-up in front of the rear tyre, and keep front tyre wake above the floor. This first design is shown in Figure 2. It's pretty tame, as I would rather start with well-behaved flow and slowly push it harder than start with something messy and have to work backwards. 

Figure 2: First floor design surface and skeleton before any simulation testing, based on existing geometry constraints, bare-car flow field analysis, and a bit of educated guesswork. 

Having the front edge of the flat skirt section set back slightly was purely rules-based to keep outside the front wheel exclusion box. Before adding any furniture or flap elements to the floor, I was targeting at least a CLA of -0.7 for all four test cases as a KPI to indicate the design and simulations are progressing as expected, and that the floor is well worth its own weight, cost, and drag in case we run out of design and/or manufacture time for the planned additions. I wasn't expecting much oomph from this first "guess" design, and indeed it didn't deliver a blinding performance estimate, at a CLA of just over -0.3 (to be fair, it did have to also overcome the 5 points of lift that the bare car was making). 

I won't go through the whole design process of getting from here to the final product, but I'll mention a few key things.

At the start, changes were very coarse and almost trial-and-error based as we looked for any promising trends to follow. Once finding a few design features that seemed to promise or provide performance improvements, we started to post-process results more and analyse exactly what the flow was doing, what was changing when we started following positive trends, what an "ideal" flow would look like for the current trend, and how we could adjust the surfaces to alter the flow field to achieve this. Generally, the straight airflow scenario was used to identify trends as it was the least computationally expensive to run. The more critical driving scenarios were then simulated to fine-tune different areas of the floor's design. With limited access to CFD and rather slow computational resources, this sort of critical and analytical thinking was key to still being able to use simulation effectively and efficiently despite the limitations. Once we were on a good track, I imposed a rule that no simulations were to be run without brief documentation of what the aim was, how any changes to surface design in the simulation were expected to achieve this, and if it was not achieved then what flow features or forces were to be looked at to suggest why it hadn't been an improvement or otherwise behaved as expected. This reduced time wasted on unhelpful simulations and increased the efficiency of post-processing analyses. 

Eventually we arrived at a design with an expected CLA of -0.74 in straight airflow, -0.71 in the median corner scenario, -0.72 in the skidpan scenario, and 0.68 under braking. Losing less than 5% total downforce in the cornering cases was a promising result. We accepted this and begun adding and optimising additional features such as strakes and flap elements. The scope for surface design wasn't too limited, but I also didn't want to go overboard given we have no validation data for complex flows or vortex behaviour/interaction yet. For this same reason though, I wanted some interesting flow features so we could actually analyse them experimentally. The final design (after 5 weeks of CFD and concept sketches) is shown in Figure 3. 

Maximum theoretical performance in an ideal, straight airflow, ride-height scenario:

CLA -1.08 (-0.82 with DRS)
CDA 0.61 (0.53 with DRS)
%Rear 84 (87 with DRS)

Using the uncertainty from the EV23 CFD setup, this puts predicted lift somewhere in the range of -1.05 to -1.11, including a larger factor of safety of 50% as a crude way of accounting for the somewhat questionable verification processes used to-date (see Certainly not for a new and much-improved verification process and discussion).

Maximum downforce efficiency is achieved in the skidpan scenario thanks to some cornering-specific design optimisation, in which CLA reaches a predicted -1.14, with a minor drag increase and 1% forward aero balance shift. 

I'll cover design and intent of the DRS system in the next post.

You'll notice these figures don't really match our downforce goals (although a downforce/drag ratio of 1.78 with just side-floors isn't too shabby), and this is due to a number of limitations we encountered that weren't foreseen at the start of the year due to having no previous data to base these goals on. I'll get to these limitations shortly, but first I'll briefly run through the surface design and the functions of each design feature.

Figure 3: Final floor design (left side) from the first CFD design-optimisation stage (spoiler: this wasn't actually the final design - see Figures 6 and 7 in Structural considerations ) 

Figure 4: "Final" floor design in context on the CFD CAD model

Underfloor entry and leading edge

Justification for the position of the floor inlet was given earlier, but also important is its shape. The aggressive drop in height behind the leading edge gets the cross-section area down as quickly as possible to both maximise ground effect over the available length, and keep the aero balance as far forward as possible. Meanwhile, although the inboard edge has to sit at a 45 degree angle to avoid ingesting the front suspension rocker wake under the floor, the bottom of the leading edge reaches in sharply below the wake to capture that little extra bit of air volume. To support this curvature, the underfloor suspension bodywork has an out-washing design to force air from under the chassis onto the top surface of the floor just behind the leading edge to reduce separation.

The outboard (vertical) section of the leading edge is aligned with the local flow direction; flow here is generally being pushed outboard by a combination of the high pressure at the floor inlet and low pressure behind the front wheel. CFD suggests the flow vectors here are not particularly sensitive to different driving scenarios. The rounded base of the outboard vertical leading edge is simply to reduce damage in case of contact with the ground. 

Brow wing

The purpose of the brow wing is to hold the boundary layer to the top surface behind the leading edge of the floor, which would otherwise separate due to the high angle of attack. The local lift created by the brow wing is a small compromise that yields huge high pressure gains on the pressure side of the floor by maximising available energy downstream, and keeping flow attached to best utilise the surface curvature. Separation on the pressure side was less of an issue when testing aerofoil-based floors, but the reduce curvature of the aerofoil shape meant overall performance was still less, while weighing a lot more. Additional performance benefits are also spawned from the high pressure zone the wing creates above the floor entry, which drives more mass flow down under the floor. If we had a front wing, we could instead do away with this odd-looking features by instead using a strong vortex off the inboard edge of the front wing to downwash the floor entry.

No endplates were used as we found more benefit from creating strong counter-rotating vortices that continued forcing mass flow onto the floor surface downstream. These vortices are visible in the below image, where blue indicates clockwise and red indicates counter-clockwise vortices in the -Z direction. 

All of this performance on the pressure-side of the floor is made extra critical for us given the majority of our downforce will come from just the floor structure. The suction side is naturally very sensitive to changes in ride-height, roll, and pitch of the vehicle, so we want to maximise the downforce from the much less sensitive pressure side to dampen these effects. 

Squirt suppression element

This is a small downwashing element attached to the lower front wishbone, designed to suppress front tyre squirt by pushing it down and behind the contact patch. Tyre squirt isn't much of a problem when driving straight, as it's held against the tyre by the high negative toe; in this case, the squirt suppression element makes no difference to performance estimates. During high-load cornering, the steering angle of the outside front wheel combined with the cross-flow from the vehicle slip angle has the opposite effect, and pulls the tyre squirt and lower wake into a large cloud that sits substantially inboard of the wheel (see Figure 1). Without the squirt suppression element, this gets sucked very quickly into the main floor section and creates blockage and separation issues. By keeping the squirt pushed against the contact patch, the cleaner flow under the floor gave a 9 point downforce advantage with only a 1 point drag increase. 

Skirt

The main purpose of the skirt is to sit low to the ground and keep the majority of front wheel wake on top, preventing it from leaking into the suction side of the main floor section. There are still some performance gains from treating the skirt as a separate floor and diffuser, but it can't be pushed anywhere near as hard as the inboard section as there is less energy/total pressure to exploit when positioned in a wake region, and with a significant adverse pressure gradient in front of the rear wheel. A subtle diffuser in front of the rear wheel expands the flow slightly, which combined with pressure propagation from the main floor section creates a fair bit of local downforce on the skirt underside. On the pressure side, the large frontal area of the rear wheel meant that drag was inevitable anyway, so a simple vertical "wall" adds negligible additional drag while allowing the high pressure to be projected onto the skirt, creating significant local force. There may be performance benefits from a more optimise flap element design here, but this fell outside our manufacture scope for this year to ensure timelines were kept to. The wall is also a nice structural element to support the outboard free edge of the skirt. The curved shape is partly aesthetic, but this area generally had lower pressure magnitudes and was too far from the floor to effectively contribute to local downforce. 

At the leading edge of the skirt, the inboard section was raised to accommodate the vortex pair from the squirt suppression element. Previously, the leading edge directly split the vortices in half, leading to them blowing up and reducing downstream performance on both the pressure and suction sides. By lifting the edge to "capture" the vortices underneath, they remain intact for much longer and their local suction is utilised for downforce. The slight high pressure created by the downwashing inlet combined with the outwashing main floor edge at this location both contribute to pushing front wheel wake out and off the skirt surface, meaning the majority of surface flow on the pressure side remains attached. The downward mass flow created by the vortex pair ensures flow remains attached through the adverse pressure gradient created behind the leading edge.

Underfloor area

The general shape of the floor was designed to maximise both ground effect and plan-view area (i.e. the area that converts pressure to downforce). Given there are very few geometry constraints for a side-mounted floor, the point of minimum cross-section area was positioned as far forward as possible to try and overcome the drag moment and get the front-rear balance as close to the theoretically ideal 75% as possible. The width of the floor is expanded along nearly the entire length, to maximise the suction area around the midpoint as well as to increase the expansion ratio, as relying too much on purely vertical expansion produced local separation due to the resulting high curvature. At the rear of the floor towards the diffuser, vertical expansion is prioritised to maximise high pressure on top of the floor. Here, the width contracts to maintain a smooth increase in cross-section area, effectively supporting the higher vertical expansion. 

To seal the main tunnel and subsequently increase velocity and enhance ground effect, a strake (that I'm calling the "skate") extends below the floor surface, separating it from the skirt area. This also has the benefit (if managed correctly) of creating a strong vortex that can be used for local suction. Just aft of where the vortex naturally starts to form, the skate is trimmed slightly to create a raised edge. This  sheds the vortex and ensures there is always some cleaner flow available to "feed" the second vortex that starts to form on the new shedding edge, even when the front half of the skate nears or contacts the ground. This avoids the vortex being starved of clean mass flow and blowing up in the diffuser under high compression (track bumps, grade changes, or high roll during cornering). It also has the benefit of avoiding a "jet" effect where air is sucked under the skate at high velocity, causing the vortex to be pushed into the middle of the diffuser and create a blockage. 

Sitting in the diffuser itself is another strake, specifically targeting tight corner scenarios. In straight airflow, the vortex sits nicely against the skate and tunnel wall, and so this additional strake just adds drag and some minor separation where it meets the diffuser roof. In tight corners though, the flow curvature and slip angle cross-flow were found to push the vortex off the wall of the tunnel on the outside of the corner. This created a blockage effect as the vortex expanded, and the rotational mass flow started causing separation on the diffuser roof in the outboard half of the tunnel. The strake keeps the vortex pinned in place, which in combination with a curved edge to the diffuser, keeps flow attached even under high roll and yaw angles. The diffuser strake, and two skate vortices can be seen in blue in the image under the 'Skirt" dropdown. 

Diffuser flap

The diffuser flap's main job is to further expand the flow from under the floor, increasing peak velocity and suction. The angle and location is limited by the cockpit height and rear suspension clevis location, so the main parameter for optimisation was the chord length of the element, using a (roughly) fixed trailing edge position. Small adjustments to optimise the angle and slot-gap parameters were also carried out. By changing the chord length with a fixed trailing edge location, the position of the slot-gap and trailing edge of the main floor diffuser is shifted forwards or backwards. Downforce is maximised by having the main floor section extend rearwards and upwards as far as possible without inducing flow separation, as this reduces the effective porosity of the surface in the area where the downward component of pressure is greatest. I wanted to add DRS (or at least the ability to adjust its position manually) to this flap element though, both because it's cool (engineers can have some fun, right?), but also critically because our overall vehicle design is quite drag-sensitive with our new lower-torque motors (see the limitations discussed below). Reducing drag by 10% above a certain speed in our lap simulator yields a predicted additional 15% competition points boost over the points added by the floor itself. This complicates the matter, as now by moving the floor trailing edge forwards and decreasing downforce, the flap becomes larger and thus the DRS effect is greater. It goes without saying that the best compromise between downforce and DRS for predicted overall competition performance was what we ended up selecting.

With potential inclusion of a rear wing in the future, this flap would likely be removed and the floor overall made much more modest (at least for the inboard section) to avoid kicking up losses towards the rear wing.

A little note on CFD too before I move on:

All floor aero surfaces utilised a wall-resolved approach targeting y+ = 1 (the actual values when plotted ended up looking approximately log-normally distributed with the bulk of cells being between 0.5 and 2: see Figure 5). The choice of a resolved approach came from a combination of two factors:

Prism layers were initially set to extend to the expected local boundary layer thickness and later adjusted in areas where boundary layer thickening or separation were found to occur. CAD thicknesses of all structural surfaces were 3mm based on the structural layup (more in this in a future post) and furniture surfaces were kept to 0.5mm. Trailing edges were given non-zero thickness to reflect manufacturing limitations, and this also gave an easy way to apply a smaller mesh size in these areas to more precisely model shed flow.

Figure 5: y+ distribution for all wall-resolved surfaces 

Right, time to get onto the limitations we found.

My downforce goal ended up being a little high, but only just. We would have easily achieved it with a lower ride height giving increased ground effect (aero mapping showed a 6 point downforce increase just by lowering the ride height 5mm all around), but we have to maintain quite a high ride height (currently about 50mm) with our car to accommodate the bumps and high compression zones on our hilly test track without the compromise of overly stiff suspension. We also want to be able to clear the very steep top of our workshop driveway, and straddle the concrete curbs at our test track to avoid catastrophic damage to the chassis or accumulator/motor mounts on the underside if a driver were to lose control on what is a very narrow (~5m) track. 

A similar issue was the higher than expected drag moment trying to rotate the car back onto the rear wheels. While an aero balance of 75% seems perfectly achievable by focussing downforce near the centre of the car, quite a bit of force is required to overcome the moment applied by drag. With a very small amount of downforce, this results an aero balance of more than 100% rearward, with the front axle experiencing reduced normal force. With the modest downforce a floor provides, this means the peak downforce needs to be focussed much further forwards than one might initially expect. I had planned for this by pushing the throat of the floor forwards and by positioning the whole structure as far forward as possible, but it simply wasn't enough to overcome the drag and bring the aero balance to the desired 75%. The biggest improvement here which will be brought up for next year's chassis design is to bring the front of the chassis down lower, lowering the stagnation point on the nose and bringing more of the bodywork into increased ground effect. Quick and dirty tests suggest this alone could bring the aero balance forward from 84 to 79 with negligible change to drag and a 2 point downforce increase. The current nose is only the height it is because of the ergonomics design being carried forward from the previous vehicle iteration, where that year's design philosophy saw ergonomics getting priority over smaller chassis optimisations for mass or VCG (and when aerodynamic performance was not considered). 

Overall I'm pretty happy with the design. It's got decent performance, and it's not too complex while being unique enough (especially with the brow wing and passive DRS) to hopefully get people talking about it and provide some interesting flows with which to test and improve CFD correlation. 

Figure 6: Some total pressure sweeps of the floor design. The rear view of the skidpan case (bottom) does a good job of showing the intended effects of the brow wing and lifted skirt inlet.