Thanks for coming through mi amigo ;) Good video, what happened at 10:49 tho with the voice, it's completely messed up and changing pitches afterwards...
You're welcome, and thanks for the idea! Oh, when I was recording, I had some other software running on the computer (a simulation), and the recording software got slow. So, for some reason my voice got recorded really low. Sorry.
Even garbage CFD can still tell you there's a drag decrease, which opens another can of worms. Can you spot a general trend with CFD even if it's inaccurate? Yes, and on the other side, you may see effects that don't show up in reality, or your CFD's inherent limitations hide actual artifacts that show up in real experiments. Another issue is the over or underestimation of effects the CFD tells you. Now if your goal is to just reduce drag, it doesn't matter if you're under or over estimating as long as you get the desired result, it's a winner, at least short term. Long term having inaccuracies can lead you to incorrectly assuming the direction you're going is worthwhile, which means then you waste more time back tracking.
That's a really good point. Nothing is 100% accurate and nothing is 100% wrong. Accuracy is a scale. So, it comes down to a couple of factors. The first is how much random uncertainty there is and how much systematic error there is. The systematic error could be corrected for, but then the question becomes, why is there this error to begin with. What in the flow has caused it. As for the random uncertainty, that cannot be corrected for because we don't know what it is. So, if the values you are looking at fall inside the random uncertainty, then you cannot tell anything because the values could really be anything within this range. For this simulation, the random uncertainty, as I calculated from their data, is between 12% and 24%. The differences in the drag coefficient throughout the paper were below 5%. This is much lower than the random uncertainty, therefore, the data from the CFD can't be used because these differences are being engulfed by the random uncertainty. Now, the second factor is what flow features you're looking at and how well they are replicated in the CFD. If they are well replicated, then you can use the CFD to get a general understanding of how the flow features are affected. In this simulation, the flow around the front wheel was not well captured because the wake was about twice as large. Considering that this study is how poking the wheel out into the flow affects the aerodynamics, this is a major error because the flow the wheel sees is now different. So the results will be different. As such, this CFD can't be used to determine these effects either. The only way to make use of the CFD is to show that some car, somewhere creates the front wheel flow they found. But, that was not shown in the paper. What we saw was a car that had a significantly different front wheel flow.
@@PremierAerodynamics That's exactly my point, say your CFD says something happens, and experimentally you get the same results when you measure(ie pressure), but for different reasons. You get led towards the wrong conclusion, so if you try to further exploit that wrong conclusion you see you make no progress or go backwards. Take for instance LBM, which is good at rarefied flows, low density, and compressibility effects, but is less good at other things that a standard CFD package does better. Do you live with that inaccuracy when on the other hand it delivers on other parts of your project with good accuracy, parts which are ultimately more important, and the relatively quick processing speed allows you to rapidly iterate? Or do you go for the more computationally expensive method, and sacrifice iteration speed, for slightly more accuracy and resolution globally, and comparable where it's really important?
Absolutely, you know what's interesting about tire wake. You only really have one option, because no matter what you do, you end up with the same result. If you outwash flow around the tire, the tire sheds a vortex outboard of the wheel. This is the only way that tire wake can work on a car, to reduce drag and increase downforce. If you've noticed, transport trucks have started putting bargeboards on their trailers, all in an effect to outwash air around the tires. Outwashing air around the tires is the broad geometrical intention, but aerodynamically, so much more happens. It's not just the physical deflection of air around the tires, it's the reduction of the tire's turbulence under the car, the diverging geometry is also a diffuser, which increases the mass flow under the trailer, which then can be used to blow the rear wake more effectively. The interesting bit is there are several effects on airflow from even simple devices.
@@rolandotillit2867 Oh right! I see. Definitely. In my experience, and opinion, when you know that you have an inaccuracy in the simulation, or experiment (or any research you're doing), you either have to always keep it in mind or try to get rid of it. Otherwise, you end up wasting resources later on trying to find something that isn't there. There is one example I saw which was there was this piece of research done at this automotive manufacturer that I was working at at the time. It concluded something about a certain aerodynamic device. Another researcher had this spreadsheet of various configurations that indicated a contradictory conclusion. So, we ended up doing a $500,000 project over a year looking more into these devices, only to conclude the original finding of that piece of research earlier. It turns out that the researcher with the spreadsheet wasn't taking many variables into account and was changing other things at the same time too. So, their data was inaccurate and hence making the new project a waste of time and money.
@@rolandotillit2867 Definitely, getting the tire wake, and any wake, away from the underbody and diffuser is much better for aerodynamics. And shielding the wheels from the outside flow also helps reduce drag.
Wow! That makes a lot of sense. And also, I would imagine that the bulge would change shape as well, with one side being bigger than the other, and then the jetting vortex would also get affected.
Thanks for coming through mi amigo ;) Good video, what happened at 10:49 tho with the voice, it's completely messed up and changing pitches afterwards...
You're welcome, and thanks for the idea!
Oh, when I was recording, I had some other software running on the computer (a simulation), and the recording software got slow. So, for some reason my voice got recorded really low. Sorry.
@@PremierAerodynamics horror aerodynamics lectures
New to CFD here. So, what was the Outcome? Yes to wheels being moved outward of the wheel wells by 10mm? or no?
It isn't as bad as you might think.
Even garbage CFD can still tell you there's a drag decrease, which opens another can of worms. Can you spot a general trend with CFD even if it's inaccurate? Yes, and on the other side, you may see effects that don't show up in reality, or your CFD's inherent limitations hide actual artifacts that show up in real experiments. Another issue is the over or underestimation of effects the CFD tells you. Now if your goal is to just reduce drag, it doesn't matter if you're under or over estimating as long as you get the desired result, it's a winner, at least short term. Long term having inaccuracies can lead you to incorrectly assuming the direction you're going is worthwhile, which means then you waste more time back tracking.
That's a really good point. Nothing is 100% accurate and nothing is 100% wrong. Accuracy is a scale.
So, it comes down to a couple of factors.
The first is how much random uncertainty there is and how much systematic error there is. The systematic error could be corrected for, but then the question becomes, why is there this error to begin with. What in the flow has caused it.
As for the random uncertainty, that cannot be corrected for because we don't know what it is.
So, if the values you are looking at fall inside the random uncertainty, then you cannot tell anything because the values could really be anything within this range.
For this simulation, the random uncertainty, as I calculated from their data, is between 12% and 24%.
The differences in the drag coefficient throughout the paper were below 5%. This is much lower than the random uncertainty, therefore, the data from the CFD can't be used because these differences are being engulfed by the random uncertainty.
Now, the second factor is what flow features you're looking at and how well they are replicated in the CFD.
If they are well replicated, then you can use the CFD to get a general understanding of how the flow features are affected.
In this simulation, the flow around the front wheel was not well captured because the wake was about twice as large.
Considering that this study is how poking the wheel out into the flow affects the aerodynamics, this is a major error because the flow the wheel sees is now different. So the results will be different.
As such, this CFD can't be used to determine these effects either.
The only way to make use of the CFD is to show that some car, somewhere creates the front wheel flow they found. But, that was not shown in the paper. What we saw was a car that had a significantly different front wheel flow.
@@PremierAerodynamics That's exactly my point, say your CFD says something happens, and experimentally you get the same results when you measure(ie pressure), but for different reasons. You get led towards the wrong conclusion, so if you try to further exploit that wrong conclusion you see you make no progress or go backwards.
Take for instance LBM, which is good at rarefied flows, low density, and compressibility effects, but is less good at other things that a standard CFD package does better. Do you live with that inaccuracy when on the other hand it delivers on other parts of your project with good accuracy, parts which are ultimately more important, and the relatively quick processing speed allows you to rapidly iterate? Or do you go for the more computationally expensive method, and sacrifice iteration speed, for slightly more accuracy and resolution globally, and comparable where it's really important?
Absolutely, you know what's interesting about tire wake. You only really have one option, because no matter what you do, you end up with the same result. If you outwash flow around the tire, the tire sheds a vortex outboard of the wheel. This is the only way that tire wake can work on a car, to reduce drag and increase downforce. If you've noticed, transport trucks have started putting bargeboards on their trailers, all in an effect to outwash air around the tires.
Outwashing air around the tires is the broad geometrical intention, but aerodynamically, so much more happens. It's not just the physical deflection of air around the tires, it's the reduction of the tire's turbulence under the car, the diverging geometry is also a diffuser, which increases the mass flow under the trailer, which then can be used to blow the rear wake more effectively.
The interesting bit is there are several effects on airflow from even simple devices.
@@rolandotillit2867 Oh right! I see. Definitely.
In my experience, and opinion, when you know that you have an inaccuracy in the simulation, or experiment (or any research you're doing), you either have to always keep it in mind or try to get rid of it.
Otherwise, you end up wasting resources later on trying to find something that isn't there.
There is one example I saw which was there was this piece of research done at this automotive manufacturer that I was working at at the time.
It concluded something about a certain aerodynamic device.
Another researcher had this spreadsheet of various configurations that indicated a contradictory conclusion.
So, we ended up doing a $500,000 project over a year looking more into these devices, only to conclude the original finding of that piece of research earlier.
It turns out that the researcher with the spreadsheet wasn't taking many variables into account and was changing other things at the same time too. So, their data was inaccurate and hence making the new project a waste of time and money.
@@rolandotillit2867 Definitely, getting the tire wake, and any wake, away from the underbody and diffuser is much better for aerodynamics.
And shielding the wheels from the outside flow also helps reduce drag.
Camber has a decent effect on drag, because it changes the aerodynamics of the contact patch, instead of being rectangular it's more trapezoidal.
Wow! That makes a lot of sense. And also, I would imagine that the bulge would change shape as well, with one side being bigger than the other, and then the jetting vortex would also get affected.
@@PremierAerodynamics are you able to look into that? I have -4 degrees of camber on the front of my car lol
It's a shame about the bunk CFD
Yeah, it happens...
Its a shame that such an interesting paper is so full of mistakes and someone is going to take them for granted instead of fact checking like you did.
It happens. At the very least, there are some general ideas in it that can be used, and at least for inspiration. But, oh well.