Login required to started new threads

Login required to post replies

Prev Next
stumpyx13 wrote:
You're right in saying that 15 degrees in a 30 mph hour wind vs. a 25 mph wind is similar but not equal. For a steady state, one could compare the Reynolds number (Re) for each part in question and determine how similar the flow might be. It could be very similar or very different - the issue with aerodynamics in cycling is that the Re and yaw angles tend to be near "break points". For example, everyone adds flow trips (the "rough" curve) to reduce flow separation at lower Re but the effect happens at a very specific point and if the relative air velocity is too low, you may never reach this point to take advantage:

If someone tests in the wind tunnel on "one side" of the curve and rides in the same regime, then their results are likely to be very close, but if not, they can differ significantly.

One other thing that most people forget when thinking about "sailing" in the wind is that gusts that change the yaw angle significantly can cause flow separation that remains even after the gusts go away due to flow separation hysteresis. Basically you can not have separation at say 15 degrees at first, but if a gust comes and changes the angle to 20 degrees and causes separation, the separation bubble will remain even after returning to 15 degrees. This may be why in the real-world people don't experience significant sailing and why real-world results *can* be very different then what is experienced in the tunnel. And for most wings, this effect is around 12-15 degrees (lift on top, drag on bottom):

For the smooth vs rough, I understand what you are trying to say, but the effect is also strongly a function of the nature of the roughness, and placement of the trip wire.

For a point of reference, a 2 inch (25.4cm) diameter sphere in a cross-flow of 40 kph wind, would correspond to a Reynolds number of about 40,000 (hope my math is right). This is the onset of drag crisis (big drop in drag) for a dimpled sphere. I agree 100% with the hysteresis. It is certainly overlooked, but for low Reynolds number cases doesn't it rely on the flow actually being laminar at the separation point? This is something that would depend largely on the nature of the incoming flow and surface roughness too.
MTM wrote:
jeffp wrote:
i have run back to back different tires over the same first 4 miles of a TT. results obtained in tunnel correlated pretty freaking well in the field.

I flatted and had to restart. used different tire second run.(tested both in tunnel)

But I don't believe you test at 15 degrees in the tunnel? ;)

There seems to be two questions in this thread. One is whether we actually experience the yaw angles that marketing departments are happy to tell us about, the other is whether wind tunnel results apply to the "real world". The latter is genuinely believed (and proven) to be true (with small caveats like the ones rruff is mentioning). The first is a little more open for debate, but the general concensus is that yaw angles outside ~10 degrees are usually only seen if you are either riding decently slow (<35 kph) and/or on specific courses/days (high wind, open fields or coast lines, riding perpendicular to the wind direction).

Note for Damon: The last point is why we would still like drag vs. yaw graphs and not just a "dumbed down" weighted average drag so you have no idea about in which conditions specific products excel and vice versa ;)

If we assume an equal probability that wind can come from any direction, then I obtain the following graph which represents the % of wind angles for which yaw is greater than 10 degrees.
Code is here: Feel free to implement and check the math.
clear all;
V = [5 7.5 10 12.5 15 17.5 20 22.5 25 27.5 30 32.5 35 37.5 40 42.5 45 47.5 50 52.5 55 57.5 60];
w = [0 2.5 5 7.5 10 12.5 15 17.5 20 22.5 25 27.5 30 32.5 35 37.5 40];

for j = 1:length(w)
for k = 1:length(V)
count = 0;
for i = 1:360
wan(i) = i*pi/180;
VR(i,j,k) = V(k)*sqrt(1+2*(w(j)/V(k))*cos(wan(i))+(w(j)/V(k))*(w(j)/V(k)));
A(i,j,k) = (w(j)/V(k))*sin(wan(i))/(1+(w(j)/V(k))*cos(wan(i)));
psi(i) = 180*atan(A(i,j,k))/pi;
if psi(i)>10 || psi(i)<-10
count = count +1;
end
end
P(j,k) = 100*count/360;
end
end

surf(V,w,P);
hold on;
xlabel('Ground Speed [kph]');
ylabel('Wind [kph]');
zlabel('% Above 10 Deg Yaw');
Last edited by: AeroTech: Mar 8, 18 14:12
MTM wrote:
Note for Damon: The last point is why we would still like drag vs. yaw graphs and not just a "dumbed down" weighted average drag so you have no idea about in which conditions specific products excel and vice versa ;)

Noted with gusto. That was the same response I had: even a very nice average does not replace the drag-yaw data! (I even considered putting more than one exclamation point on there!!!)

The nice average helps us decide which design is faster in general. We still need the drag-yaw data to optimize equipment selection for specific events.

Damon Rinard
Engineering Manager,
CSG Road Engineering Department
Cannondale & GT Bicycles
(ex-Cervelo, ex-Trek, ex-Velomax, ex-Kestrel)
damon_rinard wrote:
The nice average helps us decide which design is faster in general. We still need the drag-yaw data to optimize equipment selection for specific events.
Add to that - equipment is not the only thing we can select for specific conditions.

http://www.cyclecoach.com
http://www.aerocoach.com.au
How silly of me. I coded to only look for wind angles greater than 10 degrees, but did not include angles less than -10 degrees. So the results simply double (I changed the plot in the above post to reflect this).
Keep in mind, this plot is based on equal probability of wind occurring from any direction (reasonable assumption as any I suppose).
Some reference line plots are shown for different wind speeds that commonly occur. With a ground speed of 30kph, and wind of 5kph, there is no possible way to obtain a yaw angle above 10 degrees. That same ground speed with a 10kph wind, produces a yaw angle greater than 10 degrees at >65% of all possible wind angles. If on average, wind is experienced in all directions with equal probability, it would seem that the % values can be stated at time values. More than 65% of the time when you are traveling at 30kph with a wind of 10kph, your wind angle experienced will be greater than 10 degrees (as an example).

Last edited by: AeroTech: Mar 8, 18 14:19
Interesting discussion. I've been thinking about this for a long time, well before this post from 2012. And I spent about a zillion hours working with this sucker (a calculator where you can enter your speed, the wind speed, and your true angle to the wind and get your apparent wind angle - or yaw as it's called in bike world) when we really started looking into aero wheels. I have a let's say much more than incidental sailing background so I'd used that calculator a ton before when planning sail selection and likelihood matrices for offshore races.

A couple of points to think about:
1. Bontrager did a huge study whose results substantially agree with Flo's results. The link I had for it is dead but if you Google "Trek Aeolus White Paper" the .pdf will come right up
2. The cluster of incidence at low yaw angles is startling, but the distribution once you get past about 7.5* is also notable. You're basically as likely to experience 45* as you are 20*. What's optimized for 20* can't be optimized for 45* and vice versa.
3. The one premier triathlete with whom I've discussed this at length (he's been 2nd in Kona) had no interest in testing beyond 10* when I talked to him about it, which coincidentally was just before he went to test all of his stuff for the upcoming year. We've never tested past 20* and I've never seen anyone else do it. I don't think A2 or any of the other prominent facilities even go out past 20*. So what any piece of equipment will do way out there in the angles is anyone's guess, right?
AeroTech wrote:
How silly of me. I coded to only look for wind angles greater than 10 degrees, but did not include angles less than -10 degrees. So the results simply double (I changed the plot in the above post to reflect this).
Keep in mind, this plot is based on equal probability of wind occurring from any direction (reasonable assumption as any I suppose).
Some reference line plots are shown for different wind speeds that commonly occur. With a ground speed of 30kph, and wind of 5kph, there is no possible way to obtain a yaw angle above 10 degrees. That same ground speed with a 10kph wind, produces a yaw angle greater than 10 degrees at >65% of all possible wind angles. If on average, wind is experienced in all directions with equal probability, it would seem that the % values can be stated at time values. More than 65% of the time when you are traveling at 30kph with a wind of 10kph, your wind angle experienced will be greater than 10 degrees (as an example).
Does this account for the vertical wind gradient?

At rider COM height wind velocity is ~ half of the wind station velocity (taken at 10m).

Wind velocity at wheel axle height is ~ 70% of that at rider COM height (about 36% of wind velocity at 10m).

http://www.cyclecoach.com
http://www.aerocoach.com.au
AlexS wrote:
AeroTech wrote:
How silly of me. I coded to only look for wind angles greater than 10 degrees, but did not include angles less than -10 degrees. So the results simply double (I changed the plot in the above post to reflect this).
Keep in mind, this plot is based on equal probability of wind occurring from any direction (reasonable assumption as any I suppose).
Some reference line plots are shown for different wind speeds that commonly occur. With a ground speed of 30kph, and wind of 5kph, there is no possible way to obtain a yaw angle above 10 degrees. That same ground speed with a 10kph wind, produces a yaw angle greater than 10 degrees at >65% of all possible wind angles. If on average, wind is experienced in all directions with equal probability, it would seem that the % values can be stated at time values. More than 65% of the time when you are traveling at 30kph with a wind of 10kph, your wind angle experienced will be greater than 10 degrees (as an example).

Does this account for the vertical wind gradient?

At rider COM height wind velocity is ~ half of the wind station velocity (taken at 10m).

Wind velocity at wheel axle height is ~ 70% of that at rider COM height (about 36% of wind velocity at 10m).

This would be under an assumption of zero wind gradient. The way I am thinking about it, is with the use of an on-board aero sensor measuring the wind. Indeed, the station wind will over-predict the true wind experienced by the rider, but that is not what I am focused on here. If the rider experiences (say using an on-board aero sensor) a 5kph or 10kph wind, what does that mean for the possible yaw angles assuming equal probability of wind coming from any direction.
As for wind velocity at COM vs wheel axle, I never really characterized that so I am not sure about the 70%. I personally focus on the velocity at the frontal area centroid, and whether that velocity is a good estimate for the majority of the frontal area. I have found over 90% of the frontal area exists above the wheel axle, so the wind measurement from a sensor at or close to the frontal area centroid is a very good estimate of that experienced by the rider.
AlexS wrote:
Does this account for the vertical wind gradient?

At rider COM height wind velocity is ~ half of the wind station velocity (taken at 10m).

Wind velocity at wheel axle height is ~ 70% of that at rider COM height (about 36% of wind velocity at 10m).

On a side note, is there a particular wind gradient model that's considered best for the close-to-ground levels that riders experience, as well as the types of "tarmac?"

My wiki-fu says that the "log wind profile" model is best for the "lowest 10-20m of the planetary boundary layer." And I assume a "roughness length" of around ~0.001m for typical asphalt roads?

Or is some other model used for vertical wind gradient in cycling?
Bio_McGeek wrote:
Take a look at the figure from Martin et al 97 posted by Tom Anhalt earlier in this thread. During those trials the conditions were about as challenging as they could possibly be; the wind was high, gusty, and constantly changing directions.

On the other hand it appeared to be just one set of conditions in one location. And one surface material. And pretty far away from any structures or vegetation. That's not precisely "real world" conditions, compared to, say, the Flo experiments which used actual race courses.

Maybe that doesn't matter to the end result, though.
Last edited by: trail: Mar 8, 18 15:49
1. One set of conditions, but by chance rather extreme. If you wanted to expand that envelope, you would have to be able to control wind speed and direction... which requires a wind tunnel!

2. The exposed nature of the taxiway means that the subjects were fully exposed to any shifts in wind speed and/or direction. IOW, this means it was a more, not less, extreme test.

3. The road surface has nothing to do with things.
Andrew Coggan wrote:
3. The road surface has nothing to do with things.

I should clarify. I was looking at the "surface roughness" parameter used in vertical wind profile gradient models. If seems that if you're riding in any area with many small obstacles on the ground (maybe riding next to a curb) the elevation at which the wind can assumed to be effectively zero raises from 0.0m to something higher than that. Now for most purposes that would probably be in the noise, particularly since most rider-bike surface area would be much higher. But it might fall into that 1-3% error range in some cases (e.g. riding next to a curb).
rruff wrote:
lanierb wrote:
I'm finding the opposite: more wind generally means higher CdA, even crosswinds. My least aero race last year was in a very gusty crosswind. Now, all that doesn't mean it's not happening sometimes. I think it's just not happening enough to make a big difference to anything.

That is the case for me also. Wind is not my friend. But for some a crosswind seems to have a slight benefit.

Some people (and their setups) "sail" better than others ;-)

For example, look at the shapes of the curves for Cervelo bikes with and without "Foam Dave". Mr. Z apparently "sails" really well.

http://bikeblather.blogspot.com/
A row of hedges, houses, trees, cars parked along the side of the street - anything that blocks the path of the wind will divert it to its path of least resistance, which is often up. Beyond that, whatever obstacle to the wind doesn't even have to be between you and the wind. Go stand on the windward side of a building on a windy day. It'll be windy, but nowhere near as windy as it is if you walk 100m away from that building. Not only are the reporting stations 10m up, they have a good clear unobstructed path for the wind in 360*.

And gradient is a big variable. Cold surfaces create massive boundary layers. Sailors talk all the time about "wind weight" and how "mixed" the wind is. In the spring, with a sea breeze, you can look out on the water and it's glassy. A sailboat with a 15m tall mast might be bobbing like a cork in no breeze at all, while a boat with a 25m mast will be moving right along (and it's no fun when you're the boat with the 15m mast and you're racing against the boat with the 25m mast). That's because the cold water keeps the wind from mixing down to the surface.

Wind direction is also not even close to stable. Sure, a weather station might say it's been blowing north all day, but that could be anywhere between 345* and 015* - a HUGE range. Here in New England and on the east coast in general, any wind from the west is going to be pretty squirrely, northwest directions being the worst. Sea breezes are more stable directionally, but they still oscillate throughout the day and they veer (meaning they move to the right) as they get established. And they are building until they peak and then they start dying, which of course changes their contribution to your yaw angle vector.
Good stuff, thanks.
Bio_McGeek wrote:
Take a look at the figure from Martin et al 97 posted by Tom Anhalt earlier in this thread. During those trials the conditions were about as challenging as they could possibly be; the wind was high, gusty, and constantly changing directions. We used just the average wind speed and direction for each trial. The model still accounted for 97% of the variability with a standard error of less than 3 watts. I reckon if we had continuous values for wind speed and direction we would have cleaned up a lot of that final 3%.

Thanks Jim!

But I don't know if that paper answers the question in the title of this thread. Your measured drag coefficient in the tunnel was:

0deg, .269
5deg, .265
10deg, .265
15deg, .255

With about +-.008 uncertainty. There is a general drop in CdA with yaw (nominally ~5%), but not nearly as dramatic is other sources. For instance in the Cervelo test with the DZ mannequin, there was a 15-25% drop from 0 to 15 deg yaw. I think the OP is trying to ascertain if these large reductions in CdA with yaw can be replicated on the road.

jeffp wrote:
i prefer 180deg :)

Now this is something nobody will disagree with

http://www.pb3coaching.com
The values we reported are for CdA along the axis of the bike. To get CdA, one must account for the velocity of the air along that axis, which is less than the velocity of air in the tunnel. The drag numbers in that figure you showed may or may not be corrected for air velocity along the axis of the bike. If not, they reflect both yaw and reduced wind speed. Consequently, the reductions would be inflated.
Perhaps whoever took those data could comment.
Cheers,
Jim

rruff wrote:
Bio_McGeek wrote:
Take a look at the figure from Martin et al 97 posted by Tom Anhalt earlier in this thread. During those trials the conditions were about as challenging as they could possibly be; the wind was high, gusty, and constantly changing directions. We used just the average wind speed and direction for each trial. The model still accounted for 97% of the variability with a standard error of less than 3 watts. I reckon if we had continuous values for wind speed and direction we would have cleaned up a lot of that final 3%.

Thanks Jim!

But I don't know if that paper answers the question in the title of this thread. Your measured drag coefficient in the tunnel was:

0deg, .269
5deg, .265
10deg, .265
15deg, .255

With about +-.008 uncertainty. There is a general drop in CdA with yaw (nominally ~5%), but not nearly as dramatic is other sources. For instance in the Cervelo test with the DZ mannequin, there was a 15-25% drop from 0 to 15 deg yaw. I think the OP is trying to ascertain if these large reductions in CdA with yaw can be replicated on the road.

rruff wrote:
Bio_McGeek wrote:
Take a look at the figure from Martin et al 97 posted by Tom Anhalt earlier in this thread. During those trials the conditions were about as challenging as they could possibly be; the wind was high, gusty, and constantly changing directions. We used just the average wind speed and direction for each trial. The model still accounted for 97% of the variability with a standard error of less than 3 watts. I reckon if we had continuous values for wind speed and direction we would have cleaned up a lot of that final 3%.

Thanks Jim!

But I don't know if that paper answers the question in the title of this thread. Your measured drag coefficient in the tunnel was:

0deg, .269
5deg, .265
10deg, .265
15deg, .255

With about +-.008 uncertainty. There is a general drop in CdA with yaw (nominally ~5%), but not nearly as dramatic is other sources. For instance in the Cervelo test with the DZ mannequin, there was a 15-25% drop from 0 to 15 deg yaw. I think the OP is trying to ascertain if these large reductions in CdA with yaw can be replicated on the road.

Like I mentioned above, DZ apparently "sails" really well...whether or not the reductions in drag at yaw can be realized depends on how well the rider also "sails".

http://bikeblather.blogspot.com/
So, they aren't in terms of "CxA"?

For modeling purposes, it's much easier to use CxA (since it can be multiplied directly by the apparent wind speed ^2) to get the retarding force in the direction of travel.

Bio_McGeek wrote:
The values we reported are for CdA along the axis of the bike. To get CdA, one must account for the velocity of the air along that axis, which is less than the velocity of air in the tunnel. The drag numbers in that figure you showed may or may not be corrected for air velocity along the axis of the bike. If not, they reflect both yaw and reduced wind speed. Consequently, the reductions would be inflated.
Perhaps whoever took those data could comment.
Cheers,
Jim

rruff wrote:
Bio_McGeek wrote:
Take a look at the figure from Martin et al 97 posted by Tom Anhalt earlier in this thread. During those trials the conditions were about as challenging as they could possibly be; the wind was high, gusty, and constantly changing directions. We used just the average wind speed and direction for each trial. The model still accounted for 97% of the variability with a standard error of less than 3 watts. I reckon if we had continuous values for wind speed and direction we would have cleaned up a lot of that final 3%.

Thanks Jim!

But I don't know if that paper answers the question in the title of this thread. Your measured drag coefficient in the tunnel was:

0deg, .269
5deg, .265
10deg, .265
15deg, .255

With about +-.008 uncertainty. There is a general drop in CdA with yaw (nominally ~5%), but not nearly as dramatic is other sources. For instance in the Cervelo test with the DZ mannequin, there was a 15-25% drop from 0 to 15 deg yaw. I think the OP is trying to ascertain if these large reductions in CdA with yaw can be replicated on the road.

http://bikeblather.blogspot.com/
Tom A. wrote:
So, they aren't in terms of "CxA"?
Not sure what you mean by CxA Tom. CdA is Coefficient of Drag x Area. So, if that's what you mean, then yes, that's what we report. And, yes, that is what gives force when multiplied by 1/2 x air density x air speed^2.
What I was saying is that those drag force values in the figure might or might not be corrected for the wind velocity relative to the axis of the bike. With greater yaw angle, the component of the wind tunnel air velocity along the bikes front to back axis is reduced. If not corrected they will represent a combination of actual sail effect and reduced velocity. The effect of reduced velocity will be further exaggerated because force is related to v^2. I'm sure you know all this so we may just have a terminology issue.
Cheers,
Jim
Tom A. wrote:
So, they aren't in terms of "CxA"?

Although pendants sometimes insist on CxA, most people studying the aerodynamics of ground-based vehicles still refer to it as just CdA.
Bio_McGeek wrote:
Tom A. wrote:
So, they aren't in terms of "CxA"?
Not sure what you mean by CxA Tom. CdA is Coefficient of Drag x Area. So, if that's what you mean, then yes, that's what we report. And, yes, that is what gives force when multiplied by 1/2 x air density x air speed^2.
What I was saying is that those drag force values in the figure might or might not be corrected for the wind velocity relative to the axis of the bike. With greater yaw angle, the component of the wind tunnel air velocity along the bikes front to back axis is reduced. If not corrected they will represent a combination of actual sail effect and reduced velocity. The effect of reduced velocity will be further exaggerated because force is related to v^2. I'm sure you know all this so we may just have a terminology issue.
Cheers,
Jim

Yeah...I think we're talking the same thing.

Also, seeing as how that data is from the San Diego LSWT, I'm fairly certain the "beta yaw correction" HAS been applied. Their standard reporting form us set up to automatically apply it, IIRC.

http://bikeblather.blogspot.com/
Well, maybe that's the reason for this whole thread. That seems like a LOT of sail effect. I'd trust it if I could have the raw data and run the calculations myself.

Tom A. wrote:
Yeah...I think we're talking the same thing.

Also, seeing as how that data is from the San Diego LSWT, I'm fairly certain the "beta yaw correction" HAS been applied. Their standard reporting form us set up to automatically apply it, IIRC.
IIRC, from 0 to 15 deg my CdA fell by ~15% when measured at TAMU on my Hooker.

I have also always seemed to do better relative to my (typically stockier*) competition when it is breezy instead of dead calm.

*You were the first to point out to me this apparent relationship.

Prev Next