Undertray Design using CFX

[Jaaps]

Hey everyone,

We are going with aero for the first time this year and are doing a massive undertray, but the problem I'm facing now is setting it up in ANSYS CFX.

I've tried to find examples and research papers on how people set up their undertray in CFX, but the one I found just talk about main 'concepts' in undertray design, I think the one I'm referring to is from Michigan written as a thesis, it was very helpful but not what I was after. I also found a thesis from Sydney where they used a bluff body to design a Formula 3 diffuser.

I've ran the undertray model by itself, and found a downforce number of about 100N. then running the same undertray with a bluff body over the top, I got 180N. This is starting to screw with my head as it seems every time I do one model and then run a bluff body over the top it creates a massive difference between each not always the same way.

I understand that the amount of cells/nodes used in a mesh is very important, and have been running about 5-7 million cells, on both so I don't think that it can be my meshing.

I tried contacting people in my university, mainly PHD students and maybe one professor but the problem is that no one really understand how to accurately model CFD.

This is why I'm confused as one doesnt use a bluff body, and another does. Any 'guidance' would be awesome.

Thanks
Jakob

[Jaaps]

Hey everyone,

We are going with aero for the first time this year and are doing a massive undertray, but the problem I'm facing now is setting it up in ANSYS CFX.

I've tried to find examples and research papers on how people set up their undertray in CFX, but the one I found just talk about main 'concepts' in undertray design, I think the one I'm referring to is from Michigan written as a thesis, it was very helpful but not what I was after. I also found a thesis from Sydney where they used a bluff body to design a Formula 3 diffuser.

I've ran the undertray model by itself, and found a downforce number of about 100N. then running the same undertray with a bluff body over the top, I got 180N. This is starting to screw with my head as it seems every time I do one model and then run a bluff body over the top it creates a massive difference between each not always the same way.

I understand that the amount of cells/nodes used in a mesh is very important, and have been running about 5-7 million cells, on both so I don't think that it can be my meshing.

I tried contacting people in my university, mainly PHD students and maybe one professor but the problem is that no one really understand how to accurately model CFD.

This is why I'm confused as one doesnt use a bluff body, and another does. Any 'guidance' would be awesome.

Thanks
Jakob

[tgman]

I can't remember the title off the top of my head but on the sae website there is a paper on physically testing of a bluff body diffuser with and without a moving ground. They took plenty of pressure tappings at a number of ride heights and with varying geometry. Your best bet would be to compare their results with yours / simulate their model and see what you get.

If you have the resources avaliable build a diffuser and test it on the car. Take some data then run the cfd again. It may take 2 or 3 diffusers before it shows promise but in my opinion iterative design is the beat option (its what we will be doing)

EDIT: "The aerodynamic performance of automotive underbody diffusers" SAE technical paper series 980030.

[Jaaps]

Yer was thinking that, but then we decided to go with a new engine aswell. so after that we cant afford to do trial and error without it coming out of my pocket haha but i will definitely be looking for paper on the bluff body diffuser

[Lorenzo Pessa]

Hi Jakob,

I can give you some brief CFD best practise but you need to look deeply in how CFD works.

First of all, for your first guess undertray try to work with half model (use a symmetry plane).

I don't know which kind of bluff body you use. I hope it looks like a Formula SAE car.

Run your simulation with a FSAE car over your undertray. I can be very simple shape for your first analisys (no A-arms, no frame, closed cockpit, ecc...). Use spinning wheels and moving floor. Now you can start to evaluate different undertray shapes.

Remember that with CFD you can't use the absolute values you obtain as good. You can use the difference from a reference model.
It can be useful if you start with a model of a car without diffuser. Then add different diffuser and find what happen.

Talking about the number of cells you need to do a sensibility study. Make different meshes (from draft to fine) on the same model. Try at least 4 different meshes (let's say 2-4-6-7 millions). Then look if your results change (for example drag and downforce). When your results seem to reach an asymptote you can use that kind of mesh.

I haven't use CFX but look if you find some tutorials on web (for StarCCM+ there are a lot).

[Jaaps]

Thanks Lorenzo for your post,

yer I'll have to make a simple shaped FSAE car. I was just using a rounded nose from the front to a flat surface on top, it looked like a bullet lol. I might have to make a quick easy body that I can use for this as you suggested.

I was definitely thinking about doing the mesh idea as you mentioned so I'll definitely be doing that now.

I've found a few things about CFX but for some reason there isn't a lot of tutorials about this feature of ANSYS. If anyone comes across any can you please post the long on here.

Regards,
Jakob

[Jaaps]

<BLOCKQUOTE class="ip-ubbcode-quote"><div class="ip-ubbcode-quote-title">quote:</div><div class="ip-ubbcode-quote-content">Use spinning wheels and moving floor. </div></BLOCKQUOTE>

Another thing, I dont know how to set up rotating wheels, if you could give me a rough explanation I'm sure I'll be able to figure it out. I had a look on google over the last couple of weeks for it, but there is no documents/tutorials on how to achieve rotating wheels.

I am however able to do a moving ground plane, so thats all good.

Cheers.
Jakob

[Scott Wordley]

Hey man, we have developed some tutes that we can send you. Talk to me via email, cheers

Scott

[Jaaps]

Hey Scott,

Can you just let me know whether you received my email?

Regards
Jakob

[Ben A]

Hello,

Hello I have a question about the boundary layer thickness.
What are the assumptions you expected? Any good assumption to calculate it for the flow around and under the car?

What is your construction of the boundary layer near the wall? how many layer do you use? How thick are your individual layers?

[Adam P]

I assume that you are talking about inflation layers in the meshing stage. You need to look at something that any serious CFD software (ANSYS etc.) calls the Y Plus or Y+ value. Go to CFDonline.org. Hands down the best CFD info source out there.