Modifying your car? Be very wary of CFD

[@RwP223]

Automotive Engineer here, this is a meme where the CFD Engineers need empirical testing to help "teach" their model to be more realistic. I'm doing the empirical testing :)

[@robertpalin5878]

Sadly, having done both, I find neither to be truly reliable with respect to on-road, real environment performance, and that the best pragmatic approach is to use each to their strengths, and to 'fill in' the weaknesses of the other.

[@bobguarnieri280]

Very good discussion, thank you. I have never run a CFD analysis, but have done many structural analyses over my career, and I have to say that it took years of training, learning the nuances of the software, being able to do hand calculations to verify I was in the ballpark (and to understand what was going on), and comparing the analyses to physical results to get to a position of doing valuable work. An amateur with no training/experience, and no engineering background would have produced laughable, and dangerous results. My favorite aero book is Racing Car Aero by Katz, but I do have yours also and find that it complements the Katz book very well.

[@CurboroughSprinter]

Back when I was a kid my parents used to say “don’t believe everything you read in the papers”. Sometimes in the modern world it seems we’ve forgotten to critique things and just accept what people say on social media. Luckily there are people like you to set things straight. To be honest all his makes me think I won’t bother with CFD and just jump to ancient tech of wool tufts and magnehelic gauges. 👍

[@JulianEdgar]

Yes, if you can accurately measure the real thing it's always going to be better than a model of a model.

[@paolonieri473]

@@JulianEdgar kzbin.info/www/bejne/l5LWl4StgJqJhK8

[@paolonieri473]

​@@JulianEdgarA very interesting video on CFD

[@JanTonovski]

Thanks for this video Julian, I was about to invest a whole bunch of time in a CFD programme. I think I am going to buy some wool and duct tape instead 😂

[@Testtestnotest]

Currently my 4th year of automotive engineering engineering degree and have to say every small detail you give in these videos gives me that much more help on my finishing it. Thank you very much

[@JulianEdgar]

Great to hear!

[@robhills2613]

I purchased a cheap 3D model of a Mazda MX-5 NC (Mk3) and used AirShaper to run some CFDs. While the 3D model was clearly deficient in many areas, I did nonetheless think that the CFD artefacts were interesting, and the CFD-derived Cd was in fact very similar to the value obtained by a German car magazine in a full scale windtunnel. Next step is to buy the car and try using some of the methods that Julian has provided in his many books!

[@JulianEdgar]

See my book for how different the CFD-derived drag and lift values were on the Mercedes compared to the CFD...

[@AirShaper]

​@@JulianEdgar @robhills2613 There is no simple "yes or no" answer to the question if CFD is accurate / useful or not. Julian rightfully points out that it highly depends on the 3D model, which is true. And that was the major reason those drag values were off on the Mercedes. We've done simulations on a highly accurate scan of the Rivian R1T and got very good results (which were endorsed by Andrew McAllister of Rivian himself). Just to show that when the model is correct (and the selected accuracy at AirShaper is high enough) we can get very close: kzbin.info/www/bejne/gqm7dIKKaq53aJI Does that mean CFD is useless for anything else? No, not necessarily. Through automation we can take away some of the errors that are made by many users. Also, if you're looking for overall flow patterns to get insights into for example the angle of attack towards a wing, then 3D models off the internet can still make sense. But I fully agree - you need to look at CFD results with a healthy dose of skepticism.

[@JulianEdgar]

@AirShaper In fact the angle of attack of a wing is a good example of how CFD should *not* be used to look at aero elements in isolation; the best angle may well depend on the car to which it is fitted!

[@AirShaper]

@@JulianEdgar Yes, the best scenario is to fit the wing on the car and run a full car analysis - completely agree, and we also saw this working on the Pikes Peak car - wing worked well in isolation and was bad when placed on the car. But if you're for example shopping for after market components and you're on a tight budget, it's still valuable to know what the "base" airflow pattern around your car is like, to know where & how to put a wing (and yes, that wing will influence the base airflow pattern in turn, but you'll still learn a lot).

[@JulianEdgar]

Yes, and finding the "the 'base' airflow pattern around your car" can be achieved in 30 min and at no cost... no CFD needed!

[@hoapham7389]

your book is not available in Google Play bookstore. I did buy the previous edition, looking forward for current one to also to come into Play Store

[@JulianEdgar]

It won't ever be available on Google Play Store.

[@ivanmelnikovabramov2177]

For real, your channel its a gold mine for me, i will try to buy some of your books!

[@JulianEdgar]

Awesome, thank you!

[@HadiKamalzadeh-bo7fq]

I realized many new things about aerodynamics and CFD. thanks a lot.

[@JulianEdgar]

Not a problem!

[@kcwolffan6900]

When driving on highway,I'd like to watch other cars on raining days.So I can "see" the airflow by observing the flying water mist.😆

[@JulianEdgar]

It's a good technique, if you know what to look for.

[@kcwolffan6900]

😄😄@@JulianEdgar

[@jasonrivers7518]

I once removed the factory rear spoiler from my car and drove it, with the bootlid unlatched, on an empty road at 100, and it was still unlatched when I stopped. I then repeated the process with the spoiler refitted, and the boot latched at not quite 80- I heard it. I thus concluded that my rear spoiler wasn't entirely ornamental... Thinking that this is an effective, if wholly unscientific method of testing.

[@JulianEdgar]

Very good!

[@kewinczarny5985]

As the person who works a lot in CFD, i need to add something really important about your first point - CAD model doesnt need to be precise to every mm, because before you open the CFD program you still need to simplify the model to make the simulation in reasonable time (like 8 hours). That does not mean that precise models are not used - they are but in fields that require you to calculate downforce right to 0,001 N. And also good mesh is really important. With bad quality mesh you will never have right results.

[@JulianEdgar]

Yeah, but we're talking CAD models with no underside detailing (just flat floors), steps in body shape that aren't there, wheels that aren't detailed (and sometimes don't turn) etc.

[@LoganLeGrand]

One thing I’ve already disliked was lack of radiator airflow modeling, most just block it off lol a solid or free flow through it like it’s open. Also for track cars they always model them with windows up but in the US tracks require windows down. The window opening is a massive influence on down stream airflow and it’s never modeled!

[@JulianEdgar]

Yes, both very valid points.

[@indopleaser]

Looks like they scanned your cad model from a dicast car instead of the full scale car. Not sure

[@JulianEdgar]

I guess that is possible but it was the best free model I could find... just like any other amateur wanting to do CFD.

[@jamessever8936]

What about measuring scaled models? You can 3d print different iterations that way. The only issue I can see is that the Reynolds number will be different due to the size difference. But I guess you could change the wind speed for that

[@JulianEdgar]

Reynolds numbers a very large problem. 1/10th scale model - 1000 km/h wind speed needed. No good. How big does your model need to be? Also, surface roughness a real issue.

[@jamessever8936]

@JulianEdgar I see. My main concern was just seeing if an addition made a positive or negative impact to the Cd, but not trying to see the scale of the impact. I assume that cfd MIGHT be good for that since I am not trying to get perfect results. But even then a scale model might be better. But I think the largest I could print on my machine is only .42 meters...

[@JulianEdgar]

Making a positive or negative impact on Cd is highly based on Reynolds numbers and surface finish. If you can print at 0.42 metres (1/10th scale?) then you will not be able to test it very well. Also, wouldn't a printed model need a massive amount of work to give the correct surface finish? Maybe you could try testing it in a water tunnel - higher density fluid will allow slower speeds. Personally, I think making a full size mockup will, overall, give better results for less work.

[@piedpiper1172]

@@JulianEdgar3D printed surfaces have gotten massively better, but, I think you’d still have to at the absolute minimum sand it smooth and put automotive clear coat over it, which to get that well applied-You’d basically need a real car paint booth. Unless painting cars happens to be your hobby, I think you’re right about full scale mock-ups.

[@OgamiItto70]

Good points. Ancient computer technology saying: "Garbage in, garbage out." Computer modeling is only as effectively predictive as the data and programming that drives it. Also, be careful not to misinterpret the result(s). Use critical thinking. It's always good to be skeptical in matters of engineering (and many other facets of life).

[@twinforce_fusion6560]

That is a big question I'm always asking, does a automobile company ad's a spoiler for look or efficiency (hopefully both)? I'm personally debating on removing my factory spoiler to see if I could gain any MPH/KM's down the track (1/4 and half mile). Next Spring planning to do the full mile, maybe that would be the perfect place to test it! Already got taller tires (%5), drop the car 0.9", now the big decision will be with OEM spoiler or not?

[@JulianEdgar]

Covered in detail in my book - www.amazon.com/Vehicle-Aerodynamics-Modification-Development-alternative/dp/B0C87VYVL8 or Amazon in your country.

[@hectorae86]

Question.... If the wind is coming in from the side at say 60 miles per hour and you're driving forwards at 60 miles an hour..... Isn't the air then flowing over your car diagonally?

[@JulianEdgar]

Yes, and you don't even need a 60 mph wind to have yaw airflow - and all its effects.

[@hectorae86]

@@JulianEdgar do you have videos about that effect on KZbin too?

[@JulianEdgar]

I thought I did but I can't quickly find it. All covered in my book - www.amazon.com/Vehicle-Aerodynamics-Modification-Development-alternative/dp/B0C87VYVL8 or Amazon in your country

[@hectorae86]

@@JulianEdgar knee deep in psychology books right now, but I added them to my wishlist .

[@piedpiper1172]

Have you ever experimented with an automatically variable front air dam/grill? My modified little Gen 1 Veloster hatchback runs cool until it doesn’t, and I think it would be quite the fun pursuit to devise a way to have a speed and temperature controlled air dam, but I’m wondering if you’ve ever looked into it, or if you can think of a method that would be remotely cost viable. Once my intercooler and/or oil coolers Heatsoak, it needs all the deep breaths it can get. Just the nature of squeezing fun and horse power out of a 1.6L 4 banger. Don’t worry, I know your book will teach me how to test any implementation of such a harebrained idea, and I’ve already ordered it. More just a question of feasibility.

[@JulianEdgar]

Pretty easy - just use temp switch and solenoid.

[@piedpiper1172]

@@JulianEdgar lol. I meant the moving bits, sensors are the easy part. But fair enough, I wasn’t clear in how I asked. To me, the hard part is devising something that can open/close all or part of the grill, fits inside the engine bay, and can survive the heat environment. Once all that is in, sensors to tell it when to move seems the easy bit

[@JulianEdgar]

Get the grille shutter from another car. I have one from a Ford Focus that looks pretty adaptable.

[@piedpiper1172]

@@JulianEdgar Ty!

[@matteoricci9129]

Another KZbinr Bjorn Nyland tested a car with a roof box, without, and with a trailer itch box ( just a box hanging from the hook not a trailer with wheel). He was measuring the efficiency reported on the same road, quite accurately, not as much as you, but still good . The end result was that the difference between none and trailer itch box whas minimal Maybe you should collaborate To improve

[@nigelsmith7366]

There is a reason why automotive companies spend millions of dollars on RnD for most vehicles since mid 80's

[@JulianEdgar]

Yes, but that misses the point. The point is that aero measurement on the road is vastly better in accuracy, cost and ease than amateur level CFD.

[@jasonrivers7518]

The reason car companies spend all that money is because the government makes them, mate- they would much prefer not to!

[@THEANIMALGUY]

So a question is eyeballing(like proper eyeballing with attention to detail)better than a for the lack of a better word crappy cfd?

[@JulianEdgar]

No. Eyeballing can be completely wrong. Testing is best.

[@THEANIMALGUY]

@@JulianEdgar so even bad cfds are better than eyeballing with detail?

[@JulianEdgar]

Whaaat? Both approaches are crap - and why would you use either, when you can easily, accurately and cheaply measure the real vehicle?

[@THEANIMALGUY]

@@JulianEdgar doesn't that count aS eyeballing since you still have to look Yeah my apologies I seem to have abnormal explanationz for words 😅

[@JulianEdgar]

Measurement is nothing like just standing there looking at a car ("eyeballing").

[@nigelsmith7366]

I would be very interested in your thoughts on Aero packages on hill climb and super lap cars

[@JulianEdgar]

My book covers in detail how to develop (and measure) downforce - www.amazon.com/Vehicle-Aerodynamics-Modification-Development-alternative/dp/B0C87VYVL8 or Amazon in your country. It's all pretty easy.

[@potatosfortots9909]

Whenever people bring free cfd up to me when i talk about pressure testing I always have to explain to them the simple fact that theyre relying on a free program to emulate the laws of physics, and be able to interpit it. Its just not reasonable to be able to emulate the laws of physics that we still arent 100% sure of on a dell laptop with a free program and a video game model of your car that has a flat underside

[@eddieelectric4938]

Hey i am very curious about ur thoughts on spoilers on fwd, especially for sedan shape (notch back), do u believe a low angle spolier is good for them to give a bit off stablity on the rear during corners to them but not to much because it is fwd ?

[@JulianEdgar]

Makes no difference whether the car is front, rear or all wheel drive. Spoilers can be effective if well designed. All covered in my book!

[@eddieelectric4938]

@@JulianEdgar may I know which book it is that would cover it the most, might try to get it than to get a better understanding, thank youuu so much for your content by the way

[@JulianEdgar]

www.amazon.com/Vehicle-Aerodynamics-Modification-Development-alternative/dp/B0C87VYVL8 or Amazon in your country. Shows how to develop and measure rear spoilers, including on a sedan shape.

[@indopleaser]

Do a video of flow viz, put it on a car would make a great video

[@JulianEdgar]

Never had any success with it - and tufts are less messy.

[@neilbreen6998]

Beautiful garbage in beautiful garbage out

[@tomast9034]

only fool who totally believes in any cad/cae / climate/ finacial market simulation is the head of sales in any company. of course if you have decades expertise doing similar stuff and able to test it in real life it different. you have to know what you do and what the program can do.