CFD - Bluff Body - Diffuser Angle
Recruiting is over. After talking to the people that applied it became apparent it would be easier to handle it all myself. The most promising applicant couldn't guarantee long term support, and since they used different software than mine (mine is all Open Source) it didn't seem right since I want to maintain continuity through the different CFD runs. The downside is I can only handle mesh sizes below 2 million. Such is life.
If I get enough requests I will write a more detailed technical article. Let me know if that's something you'd be interested in.
I have seen plenty of people do CFD on full F1 cars, and other various things that seemed more exciting than what I'm doing. They seemed to fit that niche just fine, so I decided to go the way of George Costanza. Why not start from the very beginning and look at things like Ride Height, Diffuser Angle, Skirts, etc from the most direct and elementary way. Once the trends are found, the F1 Race Car can be Designed with additional complexities and better understandings of the model parameters can be achieved.
So I made a Bluff Body to represent a Race Car. It could be any car, such as a Formula One Car. Since I stand the highest chance of getting real life data from a DSR class car or similar, I chose to make it to size according to those regulations.
There will be many more Articles in this series. This one focuses just on Diffuser Angle.
The point of a Bluff Body (or Blunt Body) is simplicity. We want to extract the data trends accurately in accordance with a single changing variable. In this case it was Diffuser Angle. It wouldn't make much sense to design an entire F1 car at this point, since all of the upstream elements would interfere with our data on a simple clean Diffuser.
I made the model by taking an old DSR (now SR2) model I got from a manufacturer seeking a little help in positioning the rear wing. I took some measurements and rounded them off to make the body. The overall body width is 1.6m, height is .36m, and length is 3.8m.
I set the initial Ride Height at 37.5mm.
The maximum Diffuser Length is 1105mm. Once the Diffuser Angle reaches 10 degrees the Diffuser Length begins to degrease. The reason for this is if the Diffuser Length were left at 1105mm, when the Diffuser Angle reached 20 degrees the Bluff Body would be chopped to a shorter length and have a much more idealized shape, which would throw off the L/D measurements more than I'd like. I transition from Diffuser Height variability to Diffuser Length variability at 10 degrees due to a small edge that would be formed otherwise which would be difficult to mesh given my meshing limitations.
Let me reiterate the fact that I am not trying to model a full F1 car, but rather a simple body that will allow clean airflow and consistent results. The goal is to see how changing a single parameter at a time affects the performance of the vehicle represented as a Blunt Body.
All Bluff Bodies were subjected to a wind speed of 40 m/s (a medium speed in F1 terms). It was a difficult choice between 40 m/s and 30 m/s, since reference papers used both. In the end, since I have so much old data from models being run at 40 m/s, I decided it was best to go that route to have more models to compare to.
Since the computer I am running this on is 5 years old, I decided to aim for mesh sizes of around 750,000 and Residuals of 1E-4 (my references tend to go for 1E-6, but occasionally ran in to convergence issues). They also mentioned that little change was observed for runs with a poor residual, and that flow patterns converged quicker than the residuals suggested. They did however warn that high error was found in Lift and Drag calculations, and that only the trends and qualitative data were relevant.
A mesh size of 750,000 might not be sufficient to accurately mesh a full F1 car with all the bells and whistles, but it is moderately sufficient to model a Bluff Body representing the F1 Race Car.
Convergence failed for Diffuser Angles of 16 degrees and 20 degrees. The data for 16 were clearly wrong, but the data for 20 were included since the Residual kept getting to 1.1E-4 in a constant oscillation. The data at 20 degrees should be taken with a grain of salt.
The images are from ParaView. Rather than posting a BUNCH of images, I decided to make them in to animations using more Open Source software. If you have trouble loading these, let me know.
The above image shows how the Diffuser Length of the SR2 Blunt Body was changed. It can be observed that as the Diffuser Angle of the F1 Car Bluff Body is increased, Flow Separation becomes more and more apparent. The contours represent Velocity.
Here the low pressure region created by the Vortex can be seen. At a Diffuser Angle of zero, the pressure is essentially constant along the underbody of the Race Car. However as the Diffuser Angle is increased a change in Pressure Distribution can be observed.
The Vortex is clearly visible in the Animation above. The Vortex aids in keeping flow attached, and increases the Diffuser Angle achieved before Flow Separation begins. This flow will change drastically as I add in more F1 features such as a Rear Wing, Front Wing, and Wheels etc. But as I have said, we are starting from the beginning in this series of articles since others have already designed a full F1 car and run CFD on them.
It was shown that optimal Lift to Drag Ratio was at a Diffuser Angle of around 10 degrees. In future articles this may change as I vary parameters such as Ride Height. It was also observed that the highest Downforce generated was at a Diffuser Angle of 12 degrees for this given Blunt Body.
Diffuser Area Ratio is a value looked at by my references. I can't get my hands on their data relating to Area Ratio, so I figured I could post mine here since I couldn't find this type of data anywhere else. Keep in mind, the above chart has not been validated, just simply generated. If you would like a more in depth explanation of Area Ratio as it relates to F1 Diffusers, let me know.
Meaningful results were achieved through a substandard (by todays standards) laptop. For the given geometry it was found that the most efficient Downforce in terms of L/D were at a Diffuser Angle of 10 degrees, while the maximum Downforce generated was at a Diffuser Angle of 12 degrees. Convergence and mesh size were an issue, but not insurmountable.
In the future, tests should be completed to see the performance of the Bluff Body when changing Ride Height, Rake Angle, and Geometry.