Aerodynamics Update Summer 2022: Iterative Design and CFD
Hey guys. Heath here. Today I wanted to review our iterative design process, along with our CFD practices. We left off the last blog post after we had set our goals for the season. As a reminder, they are;
Make changes underneath the Downforce to Drag ratio of 1:2.333
Make changes underneath the Downforce coefficient to Mass (kg) ratio of 0.005:1
Keep the center of pressure behind the center of gravity at all times, and 10% behind at 40MPH coasting (Chosen as it was slightly above average FSAE speeds, and a good representation of many aerodynamic limited, meaning an increase in downforce would increase speed, sections of a typical endurance track)
Implement an Active Aerodynamic system which reduced lap time
Manufacture an aerodynamic package that weighs under 25 lbs
The primary tool in this design process was Solidworks CFD, which no one on the team was experienced with. We had to learn this tool, with the help of some former team members. Eventually, we felt familiar enough with the process to trust out results (To a point. Solidworks CFD is generally regarded as unreliable, but its all we had access to at the time. More on this and how to use Solidworks CFD later.)
The first thing we did was set up a procedure using the full car CFD model from 2019, as it was largely similar to the car used in 2022. The initial metrics we decided to take from the sims were the overall downforce and drag from the car, as well as the downforce and drag for the front wing, floor, and rear wing independently. This was done by setting goals of normal force on each of the surfaces in the downforce and drag axis.
Having decided how to run our first sim, we now needed to standardize the results reporting, so we could keep track of changes over extended periods of time. We created a master spreadsheet where we kept track of every sim run. This spreadsheet had tabs for each type of simulation, for example: Full Car, Rear Wing, Front Wing, Floor etc. Within each tab, rows were dedicated to specific simulations, and columns dedicated to different data values. A screenshot of this spreadsheet is seen below.
One important data column which took a while to figure out was the aerodynamic balance of the car. Eventually, we decided on setting a coordinate axis both the front or rear axle, then using a surface parameter to determine the torque around the axle, and finally solving a statics problem to determine the vertical load at each axle (Quick tip: to see the components of the torque around the coordinate system in the surface parameter, right click>edit feature>show).
We also wanted to document any pertinent details and images with each simulation not captured in this spreadsheet. To do this, we created a document template where a few things were noted: The changes made to the model in the simulation, significant changes in the numerical results, relevant imagery (ground pressure, centerline pressure, flowlines etc.).
With a standard for results reporting set, we moved forward with iterative design. This consisted of running sims and reporting their values in the above method constantly. We were mostly limited by simulation power in this facet of the design process, as full car sims could take up to 36 hours to complete. Luckily, we were able to use both D-class computers, which help keep this time to a minimum.
As a general process for deciding what changes to make, we met every week to discuss the previous weeks results, and choose new changes to make. These changes were generally chosen based on discussion around a variety of factors. Some of these factors include ideas leaned at competition, ideas off the internet, results from the previous week sims, manufacturability, and our design goals.
Once we had decided on changes we wanted to make, we would often run a few test sims on the part in free flow to reduce computation time. We did this to help identify the better ideas we had. This had to be done very carefully, paying attention to anywhere we anticipated major differences between a free flow simulation and a full car simulation, and skipping this step in the case there might be major differences.
Once we had decided on the better ideas, we moved them to the full car simulation, changing one variable at a time. We made changes on the car moving front to back. This is the direction of airflow, meaning changes to the front wing could affect every other part, so we tried to finalize the front wing before moving rearwards on the car, and repeated this process. Below, we are going to list changes we made, as well as the final design for the 3 major parts on the car. (NOTE: The design of the floor had been finalized during summer before a change in the suspension geometry caused a redesign due to packaging issues at the rear diffuser in October).
Changes tested were as follows
V1-2022 Front Wing
V2-Added outwashing endplates (figure 1 below)
Added with hopes of directing airflow around the tire, the outwashing endplates accomplish this, reducing drag, and also increase downforce, likely due to an extension of the low pressure zone underneath the wing caused by the "buffer" between the low pressure zone and the outside airflow caused by the footplate.
V3-Remade outwashing endplates
V4-removed outwashing endplates, added dive planes (figure 2 below)
Dive planes hope to increase downforce by extending the wing outside the endplate. However in our case, they also reduce the span of the wing, which is limited by the rules, so they are not effective. (Dive planes are typically used in sports car aerodynamics, where the body shape is limited, and they don't reduce the span of the front wing)
V5-Removed dive planes, Added inboard 3rd element (figure 3 below)
Because the rules allow the aerodynamic devices above 250mm inboard of the front tires, we hope to increase downforce by making the wing more aggressive in this space
V6- Re-Added outwashing endplates (figure 4 below)
V7-adjusted width to fit with 2023 car parameters (figure 5 below)
V8-Reduced AoA and airfoil gaps (figure 6 below)
CFD indicated the airfoils were stalling. This was discovered using the streamlines surface plot feature in solidworks, where swirling was noticed on the second and third elements, thus the angle of attack and slot gaps in the airfoil were reduced
V9-Further Reduced AoA and airfoil gaps
Continued with the reduction due to curiosity. An increase in downforce was not expected or noticed
V10-Implemented adjustable 3rd Element and returned to V8 AoA
All further changes in the front wing were not focused on aerodynamics, and instead related to mounting the wing to the car. After reaching v10, we had run out of ideas and time with which to improve the front wing, so we decided on v10 as the final aerodynamic change to the front wing and turned out focus to the floor. The final iteration of the front wing was v12, which is pictured below. This version of the front wing generated 51.386 lbf of downforce at 40 MPH, compared to last years wing which generated 36.846 lbf for an increase of 14.84 lbf, or ~40%. The 2022 version created 5.656 lbf of drag while our final version had 8.59 lbf of drag at 40 MPH, for an increase of 2.934 lbf, or ~48%. This represents a ratio of 14.84:8.59, or 1:.578, well within our defined maximum of 1:2.333.
Next focus was moved to the Floor. Again I will describe the changes made and show some images before highlighting the final summer design. I'll then describe the changes made to accommodate the change in the chassis geometry. Before showing the final floor which is being manufactured.
Floor (Undertray and Diffuser)
The changes made are as follows:
Floor V1-Started from scratch
Floor V2-Divided rear diffuser and added side diffuser. Added Ramped intake
Rear Diffuser must be divided into two separate tunnels to add space for the jacking bar. Side diffusers are added due to noticing them being common among teams
Floor V3-Added Gurney Flaps
Gurney flaps are the vertical surfaces at the end of the diffusers. They create low pressure right behind and outside the diffuser, which can help keep the air attached to the diffuser longer, and increase downforce with minimal increases in drag (sometimes even a reduction)
Floor V4-Smoothed Side Diffusers
Smooth surfaces tend to separate later then kinks or angles
Floor V5-Added Vortex Generators
These attempted to make a vortex along the side of the wing which sealed the edge from air spilling into the floor
Floor V6-Removed Gurney Flaps and vortex generators
Floor V7-Smoothed Rear Diffuser and removed lateral expansion
Analysis of the rear diffuser showed airflow entering the side of the diffuser and the creating separation traveling inward, which this version attempted to fix
Floor V8-Readded Gurney Flaps
Floor V9-Increased Side Diffuser Volume
Floor V10-Added Strakes
Strakes aim to seal the floor further by creating channels for the air to flow, and making it harder for the air to travel from outside the diffuser to inside
Floor V11-Shortened rear tunnels to investigate tunnel separation
Floor V12-Changed ramp intake shape
Floor v13-Adjusted Side Diffuser intake
Floor v14-Narrowed Ramp Intake
Floor v15-Returned to V6 with strakes
Floor v16-Changed intake geometry
Floor v17-Flattened Intake and raised side diffusers
After investigating the ramp intake, we determined that it was not actually helping downforce, and in fact reducing it and creating drag. This is because the air hitting the front of the intake forced it up and back.
Floor v18-Added Side Diffuser Strakes
Floor v19-Raised Side diffuser again
Floor v20-Added Footplate
Similar to the front wing, the footplate looks to reduce airflow entering the side of the diffuser
Floor v21-Added Footplate Lips
much like the vortex generators in v5, these lips aim to seal the floor by making it harder for the air above the floor to wrap around the side and enter below, reducing the pressure and therefore downforce.
Floor v22-Removed Diffuser Strakes
Floor v23-Removed all strakes, added rear diffuser strakes back (FINAL BEFORE CHASSIS CHANGES)
Floor v24-Widened Rear Diffuse gap and rounded front edges
Here the decision was made for the chassis to extend further back than was originally intended for optimum suspension geometry. Therefore, we had to reduce the size of the rear diffuser to fit with the new chassis.
Floor v25-Narrowed Rear Diffuser Gap to closer match chassis geometry
Floor v26-Made Front Edges Square
Floor v27- Final Version Exported for Manufacturing
At 40 MPH, last years floor (not pictured) generated 12.04 lbf of downforce and 2.84 lbf of drag, while this years floor make 9.626 lbf of downforce and 1.708 lbf of drag. This represents reductions of 2.414lbf and 1.132lbf and percentages of 20% and 40% respectively. These is a decrease of 1:.469 in downforce to drag. Because we are decreasing, we would like this ratio to be above 1:2.333, so this years floor is not preforming as well as last years model. This is due to the change in chassis geometry which limited our ability and time to design.
For the final summer design element, images of the Rear Wing are shown below.
Changes made are as follows
V2-Added Upper 4th Element
This is a feature we had seen on a few FSAE cars, and something we know the team looked into in the past. We wanted to run some sims on this configuration to see if we could get it to work for us. Ultimately, our CFD showed that this element did not add any downforce, and increased the drag. Upon looking at a pressure plot of the car centerline, we saw that the high pressure caused by upper surface of the main element was negating any low pressure created by the new 4th element.
V4-Reverted to V1 (Added Swan Necks in simulation)
V5-No Change from v4
V6-Extended Endplates Lower
The goal of extending the endplates lower was to help contain the low pressure zone underneath the wing. Its important that this low pressure zone be protected, as it is a major component in downforce and drag, so keeping it in shape is beneficial in 2 ways. First it makes the low pressure zone larger and therefore stronger. The extension does not interfere with any of the others components on the car, and increases downforce and decreases drag
V7-Swapped Airfoils for s1223
The Selig 1223 airfoil is a common choice for many FSAE team to use in there wings. We were curious about it, and tried to create a rear wing similar to our current wing in order to compare. FSAE airfoils are often chosen for 2 unique qualities. This first is high stall angle of attack at lower speeds. The second is high coefficient of lift (Seems like this is obvious, but it also usually comes with a drag penalty. Because FSAE operates at low speed and high power to weight ratio, this tradeoff is usually beneficial, and we saw that is our optimum lap simulations and incorporated into our goals for the package). The s1223 airfoil preforms well in both categories.
V8-s1223 wing with leading edge slat
The leading edge slat is commonly used in large airliners at low speed in order to increase lift. It's purpose is to direct more airflow underneath the bottom or the multi element airfoil to increase the downforce, as well as increase the effective chord (length front-back)
V9-NACA 9412 wing with leading edge slat
Attempted the leading edge slat again reverting back to last years airfoil, the NACA 9412.
V10-Removed Leading Edge Slat and Readded the upper 4th element
V11-Removed upper 4th element, Added Anti-drag vortex generator
The Anti-Drag Vortex Generator is the cutout in the upper rear corner of the endplate. Its purpose is to generate a vortex in the opposite direction to the major vortex in hopes to reduce its strength. It accomplishes this by allowing the medium pressure outside the endplates to spill over into the low pressure under the airfoil creating a vortex before the high pressure spills into the medium and low pressure, creating the major vortex in the opposite direction and the initial one. These will cancel out, reducing the strength of the final vortex, and therefore drag
V12-Adjusted Wing Width to fit 2023 rules
The final version of the RW did not have much improvement over the previous year. The only major changes incorporated were a vertical extension of the endplates and a anti-vortex generator (Seen on the last slide of the above slideshow). Last year's wing made 44.47 lbf of downforce at 40 MPH, and 19.406 lbf of drag. The final version for 2023 made 50.41lbf of downforce and 21.292 lbf of drag. These are increases of 5.94 lbf and 1.886 lbf respectively, representing increases of 13% and 10%. The ratio for this increase is 1:.317, again well within our defined limit of 1:2.333.
Full Car Analysis
The complete car made 50.80 lbf of drag and 93.88 lbf of drag at 40 MPH in 2022. The Aerodynamic balance of the package was 57/43 rear, meaning 57% of the force was on the rear tires, and 43% was on the front tires. The 2023 package made 52.15 lbf of drag, and 100.84 lbf of downforce. This is an increase in downforce of 6.98 lbf and an increase in drag of 1.35 lbf. This is a ratio of downforce to drag of 1:0.193, again well inside the limit of 1:2.333 determined using optimum lap.
These changes followed our goal for downforce vs drag numbers, despite several other changes on the car limiting the effectiveness of the aerodynamic package. One of these is mentioned earlier in the extension of the chassis rearward, but other include an increase in ride height and a decrease in track width. Having kept manufacturing in mind during the design of the parts, we had a plan to make them all during the year.
Despite this goal being met, we failed to met the aerodynamic balance goal for the car. The 2023 package had a rear-front balance of 35.35/64.75. This is well off our goal of 70/30 during coasting (10 percent behind the vehicle goal of 60/40 rear balance). We set this goal because of research on the effect of aerodynamic balance. The reason it is preferred to be rearward is a reduction in oversteer due to higher loading on the rear tires, and therefore less is on the front tire. The risk with too much load on the front tire is they "bite" too hard and turn the car to quickly, then the reduced load on the rear tires leads to less grip, and a higher likelihood they begin sliding. This is a reduction in the drivability of the car.
In the end, we felt that the reduction in drivability due to missing the goal of aerodynamic balance was worth the increase in downforce. To help lower the risk of this decision, we built adjustability into the 3rd element of the front wing, which would move the balance rearward at the cost of a reduction in downforce.
With this potential solution, we were comfortable the the aerodynamic package and were comfortable moving forward with initial manufacturing and the final stages of design.