Jump to content
LegacyGT.com

4th gen aero


boxkita

Recommended Posts

What's been done. what works. what doesn't. What's not known. What's unproven.

 

Interesting youtubes:

engineering explained:

kyle.drives: https://www.youtube.com/user/Kyleengineers

graysgarage series: https://www.youtube.com/playlist?list=RDCMUCHqgnBEbgkljvHQUg9MQfKA&feature=share&playnext=1

 

Interesting books:

Joseph Katz - Racecar Aerodynamics: https://www.amazon.com/Race-Car-Aerodynamics-Engineering-Performance/dp/0837601428

Simon McBeath - Competition Car Aerodynamics: https://www.amazon.com/Race-Car-Aerodynamics-Engineering-Performance/dp/0837601428

Julian Edgar - Modifying the aerodynamics of your road car: https://www.amazon.com/Modifying-Aerodynamics-Your-Road-SpeedPro/dp/1787112837

 

 

 

Interesting posts/threads:

from wdydtyfgt thread: https://legacygt.com/forums/showpost.php?p=5965986&postcount=2845

Wool tufted wagon: https://legacygt.com/forums/showpost.php?p=5640078&postcount=306

wagon roof at various speeds (expand thread as several more posts occur on this test session): https://legacygt.com/forums/showpost.php?p=5640123&postcount=771

 

People with useful skills:

awfulwaffle - does cfd work: https://legacygt.com/forums/showpost.php?p=5944879&postcount=2059

 

3d models:

$0 lgt - https://sketchfab.com/3d-models/subaru-legacy-25-gt-af6ca49f8e244b78a1c26781d4e82a5d

$96 lgt jdm facelift - https://hum3d.com/3d-models/subaru-legacy-station-wagon-2008/

$100 outback - https://www.3dcadbrowser.com/3d-model/subaru-outback-wagon-2005

catalog: https://3dmdb.com/en/3d-models/subaru%20legacy/

 

software:

solidworks - https://www.solidworks.com/how-to-buy

proE/creo (student) - https://www.ptc.com/en/academic-program/academic-products/free-software/creo-college-download

onshape - https://cad.onshape.com/help/Content/exporting-files.htm

 

file formats needed

step

iges

 

3d scanning:

photogrammetry- https://all3dp.com/1/best-photogrammetry-software/

 

Gold plated bling:

Edited by boxkita
added software notes
Link to comment
Share on other sites

Component to model : modeler : file format : file location.

 

2005 usdm lgt 2d body @ centerline : awfulwaffle

2005 usdm lgt wagon. : awfulwaffle

2005 usdm lgt hood with scoop opening

2005 usdm na hood with no scoop opening

2005 usdm lgt scoop only

2005 usdm lgt front bumper

2005 usdm lgt grill

2005 usdm left headlight

2005 usdm right headlight

2005 udsm lgt undertray

2095 usdm lgt radiator support structure

2005 usdm lgt front structure

2005 usdm lgt wagon tailgate

2005 usdm lgt sedan trunk lid

2005 usdm lgt sedan spoiler

2005 usdm lgt wagon spoiler

2005 usdm lgt rear bumper

2005 usdm lgt underbody

 

jdm:

pre-facelift front components

pre-facelift rear components

post-facelift front components

post-facelift rear components

s204 body parts

 

aftermarket:

ebay lip

jdm sport front bumper skin

madrig's wagon wing

madrig's front fenders

madrig's rear wagon fenders

Edited by boxkita
add components
Link to comment
Share on other sites

I'll have to take a look. I'm no good with freeform CAD modeling, and getting all the surfaces right in an engineering CAD package will be tedious. Kind of why I was hoping Nonamedude would find the time to tweak whatever models - I can get into it if I have to though.

 

An alternative - what are the odds someone on the forum owns or has access to a handheld 3D scanner?

Link to comment
Share on other sites

i posted a link to a model I found. What formats can you use?

 

I talked with a local company a few years ago about doing a full scan of my wagon. They quoted about 2700$. Ultimately passed as they were flakey to deal with. I didn't have suitable space for the scan and they wouldn't commit their space.

Link to comment
Share on other sites

NX or ProeE/Creo .prt files would be preferable since that's what I use natively at the day job and we own the interface licenses to get them straight into the CFD mesher.

 

Otherwise, I have access to Solidworks at home so .sldprt/.sldasm would be fine, would just have to be converted to IGES or STEP before meshing. At the end of the day IGES or STEP would be fine to start with, would just be clunkier to modify than a nice parametric model in one of the above formats (if we can even find one).

 

On the 3D scanning side - I've been thinking about getting one of these on and off for a while now:

 

https://www.matterhackers.com/store/l/3d-systems-sense-scanner/sk/MZ9TC0LX?rcode=GAT9HR&gclid=Cj0KCQjw-_j1BRDkARIsAJcfmTFYzdPWg6qHYDH1RnAgFwUyFhx-C03kNI5NSrbwUt36MjWFbtOt2BIaAmBPEALw_wcB

Link to comment
Share on other sites

NX or ProeE/Creo .prt files would be preferable since that's what I use natively at the day job and we own the interface licenses to get them straight into the CFD mesher.

 

Otherwise, I have access to Solidworks at home so .sldprt/.sldasm would be fine, would just have to be converted to IGES or STEP before meshing. At the end of the day IGES or STEP would be fine to start with, would just be clunkier to modify than a nice parametric model in one of the above formats (if we can even find one).

 

On the 3D scanning side - I've been thinking about getting one of these on and off for a while now:

 

https://www.matterhackers.com/store/l/3d-systems-sense-scanner/sk/MZ9TC0LX?rcode=GAT9HR&gclid=Cj0KCQjw-_j1BRDkARIsAJcfmTFYzdPWg6qHYDH1RnAgFwUyFhx-C03kNI5NSrbwUt36MjWFbtOt2BIaAmBPEALw_wcB

 

Randy of Hkcspeed used to have a cf shop and was looking at scanning parts. They bought one of these and found the resolution too low to be useful. Quote from listing "We found that this scanner works particularly well for capturing scans of people’s faces". ymmv.

 

From my research, you needed 1mm accuracy in order for the cfd to be meaningful. However not an expert.

Link to comment
Share on other sites

To me it sounds like we could build up a useful dataset for folks to use when setting up their cars in the future. I'm just getting into running on the track, but one of the co-owners of my shop space has been at it for a while. Talking to him on the subject a while back, it really seems to me like it'd be useful to understand (at least in a broad sense) the impact certain choices in setup will have on the flowfield around the car, and implications on drag, downforce, stability, etc. It should put us all one step above guess-and-check when it comes to getting things set up, or at the very least provide a starting point for fine tuning in the real world.

 

Also, it's fun to nerd out.

Link to comment
Share on other sites

Randy of Hkcspeed used to have a cf shop and was looking at scanning parts. They bought one of these and found the resolution too low to be useful. Quote from listing "We found that this scanner works particularly well for capturing scans of people’s faces". ymmv.

 

From my research, you needed 1mm accuracy in order for the cfd to be meaningful. However not an expert.

 

Dang, that's a shame. Good to know though.

 

At the end of the day, the accuracy you'd need really depends on the conditions you're running and what exactly you're after. IMO, if we were able to get a smooth model that's pretty close to the net shape of the actual car within a few mm, we'd be OK. Unless you're willing to shoulder the computational expense, you're going to be making a bunch of simplifications and approximations (both physically and numerically), and CFD becomes less of a tool that gives you a dead-nuts answer and more of one that gets you pretty close and allows you to have confidence in the trends you uncover as you make incremental changes in what you're analyzing.

 

FWIW, I don't think the boundary layer thickness on one of our cars will stay thinner than 1mm for very long anyways, at the speeds in question. I'll do a quick calc over lunch : D

Link to comment
Share on other sites

finding a model in 3dmax or something similar seems pretty easy. Converting to the formats you listed requires Solid Works, although I thought there is a solidworks interface to 3dmax?

 

I looked at photogrammetry instead of the expensive 3d scanners. Never got far because clunky computer. No excuses anymore. Don't have solid works tho. Hopefully student license can be obtained during quarantine.

Link to comment
Share on other sites

FWIW, I don't think the boundary layer thickness on one of our cars will stay thinner than 1mm for very long anyways, at the speeds in question. I'll do a quick calc over lunch : D

 

for the roof, its pretty good. Testing an inch above the roof at 100mph showed laminar flow. Even at low speeds it's really good

Link to comment
Share on other sites

That's pretty surprising (not that I don't believe you, data is king). A Reynolds number calculation along a 3m plate at standard conditions and 100 mph shows that the the laminar/turbulent flow threshold should be crossed within a few centimeters. Corresponding turbulent boundary layer thickness calc shows a thickness ranging from a few mm to a few cm by the end of the plate.

 

The acceleration of the air over the roof of the car will suppress the BL growth versus the flat plate, but I didn't think it'd be that much. Guess I need to recalibrate my intuition :lol:

 

Either way, the entire BL doesn't need to be resolved for CFD to give us reasonable results. I really think what matters is that the contour is close and relatively smooth - ie, no sudden jumps or slopes that are far in excess of the actual body of the car.

 

Wonder how the folks who put together the CAD models you linked took all their measurements...

Capture.PNG.095d588411016dba5e9e12545a2ca82e.PNG

Link to comment
Share on other sites

Here's a thought - could start with a simple 2D run along the centerplane of the car. Could derive the contour from a photo - we'd just need a pretty squared-up picture of one of our cars in profile, which would be pretty simple to spline out in CAD.

 

I could do the wagon if someone wanted to do a sedan.

Link to comment
Share on other sites

this : https://legacygt.com/forums/showpost.php?p=4737198&postcount=91

in the body.pdf has an exact side view

 

this : https://sketchfab.com/3d-models/subaru-legacy-25-gt-af6ca49f8e244b78a1c26781d4e82a5d is free and has a pretty good number of triangles. Not sure the format but should be convertible thru blender to something importable.

 

Not trying to dodge the work, just haven't any idea how any of this works.

Link to comment
Share on other sites

That's pretty surprising (not that I don't believe you, data is king). A Reynolds number calculation along a 3m plate at standard conditions and 100 mph shows that the the laminar/turbulent flow threshold should be crossed within a few centimeters. Corresponding turbulent boundary layer thickness calc shows a thickness ranging from a few mm to a few cm by the end of the plate.

 

The acceleration of the air over the roof of the car will suppress the BL growth versus the flat plate, but I didn't think it'd be that much. Guess I need to recalibrate my intuition :lol:

 

Either way, the entire BL doesn't need to be resolved for CFD to give us reasonable results. I really think what matters is that the contour is close and relatively smooth - ie, no sudden jumps or slopes that are far in excess of the actual body of the car.

 

Wonder how the folks who put together the CAD models you linked took all their measurements...

 

this post shows roofline flow at speeds every 15 mph or so to 100

https://legacygt.com/forums/showpost.php?p=5640123&postcount=771

Link to comment
Share on other sites

Rad, that's great to have.

 

Starting with a scoopless wagon over my lunch break here. With utc_pyro's data above, I can add in a 2D representation of a scoop and pipe off some air there to explore the effect on the flow in 2D. Hopefully the 2D runs give us some ideas that we can explore further in 3D.

Link to comment
Share on other sites

Alright LGT nerds, quick first run I managed to get set up and fired off over lunch and in the background over the last hour or so. Some static pressures, velocity, and turbulent kinetic energy of a wagon scooting at 100 mph. Oh, and a plot showing convergence trend.

 

To Boxkita's point - I set the mesh height to 1 mm along the surface of the wagon, and analysis shows that this edit: [DID NOT] result in the boundary layer being resolved down to the viscous sublayer (the thin laminar sheet at the very bottom of the boundary layer). If you take a look at the turbulent kinetic energy distribution, the flow along the roof has some turbulent dissipation but juuuust barely. I had to zoom and change the colormap to see it. I think this is in agreement with the results in the linked thread above.

 

I'll noodle on things that would be useful to explore in 2D later when I get home. Let me know if you guys can think of anything! We can post-process pretty much anything that can be expressed in terms of temperatures, pressures, velocities, etc. Downforce and drag come to mind.

wagon_trace.PNG.6528735e7b460798dcb70ab50ec368c9.PNG

conv.thumb.PNG.f9601cbf614d26498f626b5cbbe507ed.PNG

100_mph_Pstatic.PNG.48d096b130efa80ed2fe33c665769266.PNG

100_mph_vel.thumb.PNG.22ef479d84378a5522abcb22d0badac2.PNG

100_mph_vel_zoom.PNG.5338263df304a7a85646f988b3ca5379.PNG

tke.png.0e1c51c3241b3fbe0a91db5226f20dc3.png

tke_zoom.png.45f54dee93f078a03fcf3a9335081203.png

Edited by awfulwaffle
Updated images and text
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...

Important Information

Terms of Use