Narrow vs wide tyres - Bring data

Back by popular demand, the general all-things Road forum!

Moderator: robbosmans

6fu
Posts: 90
Joined: Thu Nov 23, 2023 10:59 am
In good conditions with perfect road narrower is faster both in terms of rr and aero. That's why they still use narrow tires on track bikes.

In real world conditions with imperfect roads, traffic, weather, bigger tire run at lower pressure makes more sense because increase in comfort and bigger contact patch = more grip.

Lower rolling resistance and smaller contact patch with bigger tires is just a marketing gimmick. That may only happen if pressure is not properly adjusted to account for increased air volume but then running a wider tire would not make any sense because you would end up with more uncomfortable tire with lower grip negating all the benefits of a bigger tire

Jaisen
Posts: 767
Joined: Fri Oct 07, 2022 2:01 am
Nereth wrote:
Wed Jul 24, 2024 5:21 am
Hey all,

A lot of questions about what "Equal vertical stiffness" means, and "what about the different air volume, and the increase in pressure when you hit a bump" (paraphrased heavily)

To address (investigate the possibility of) "equal vertical stiffness", I'll see if I can run an FEA later and get force-displacement curves for a 25mm tyre and a 28 as well. But that's gonna require our license at work to be free because I'm not kicking the other guys off for this...

To address volume change/pressure rise when you hit a bump, is a lot easier at a very rough level. I've modelled up a 25mm tyre and wheel, taken the internal volume (996700mm3), then I've just deleted a slice from the internal volume to generate an appropriate contact patch for 40kg of vertical force on it at 80PSI of pressure (target 711 mm2), and taken the volume again (995987 mm2), to get delta volume (713mm^3) and thus rise in pressure (+0.06PSI).

I've taken some screenshots below to bring you all along for the ride.

Tyre volume change.png

I'd argue 0.06 PSI is quite negligible considering 1) in the real world, the side of the tyre bulges out so you get some volume back and 2) the 28mm tyre also will experience most of that 0.06 PSI. So the real difference in during-bump-pressure due to additional internal volume of a 28mm tyre is probably like, 0.04PSI rise in a 25mm vs 0.03 PSI rise in a 28mm tyre = 0.01PSI difference in behaviour. That last bit is a series of ass-pulls as I haven't modelled a 28mm tyre yet and we don't know the effect of the bulge, but it really doesn't matter, I can be off by an order of magnitude and the conclusion is still the same.

Again, if I get time today/tomorrow, I'll run an FEA and get an actual force-displacement curve between this tyre/wheel setup (25mm blowing up to 26.5 on a 20.5 rim) and a 28mm-tyre-blowing-up-to-30-mm-on-23-mm-internal-width-rim, so we can see how bump reaction might actually look on a "normalised for vertical stiffness" pair of tyre setups.
Thanks for the interesting analysis and the nice screenshots. I look forward to the further analysis. That being said something seems off to me. On the Geek Warning Podcast from Escape Collective on March 22, they were speaking to Zipp engineers about the whole blow out scenario with the De Gendt crash. At the ~44 minute mark, the Zipp guys were saying that in their independent testing they saw increases in tire pressures up to 4 psi as a result of impacts. This was mentioned in the context that the instaneous increase in tire pressure wouldn't be enough to blow off a tire from a hookless rim. I mention it here since that is 2 orders of magnitude higher than your initial estimates, and using Boyle's law that represents almost a 5% change in air volume on a 700x25 tire. I know Boyle's law isn't a great simplification since some of the energy would be dissipated into heat, but I just wanted to keep it simple for the moment to highlight the vast discrepancy compared to your calculations.

Visit starbike.com Online Retailer for HighEnd cycling components
Great Prices ✓    Broad Selection ✓    Worldwide Delivery ✓

www.starbike.com

Nereth
Posts: 302
Joined: Wed Jan 18, 2023 10:18 am
4 PSI gain on a starting pressure of say, 60PSI would require a 6%+ loss of volume.

The extreme case of hitting a FLAT SURFACE so hard you deform all the way to the rim, does in fact lead to a 6% loss in volume, the new volume of 930354mm^3 divided by the initial volume of 996700 is 94.3%.

It checks out, but it seems to me that Zipp's independant testing was to basically hydraulically force the wheel into a flat plane to generate as large a pressure rise as possible to justify their blow-offs. because no human is holding that reaction force back.

Jaisen
Posts: 767
Joined: Fri Oct 07, 2022 2:01 am
Nereth wrote:
Wed Jul 24, 2024 7:54 am
4 PSI gain on a starting pressure of say, 60PSI would require a 6%+ loss of volume.

The extreme case of hitting a FLAT SURFACE so hard you deform all the way to the rim, does in fact lead to a 6% loss in volume, the new volume of 930354mm^3 divided by the initial volume of 996700 is 94.3%.

It checks out, but it seems to me that Zipp's independant testing was to basically hydraulically force the wheel into a flat plane to generate as large a pressure rise as possible to justify their blow-offs. because no human is holding that reaction force back.

Rim strike.png
Again really interesting stuff and I was actually just thinking about how much volume change and tire deformation would be needed to bottom out the tire. So I especially liked this screen shot. In case it is relevant, Zipp said the 4 psi was the max they saw on a 700x28 tire on their NSW 353's with 25mm ID. They mentioned the average increase across all impacts (testing up to failure) was 3.5 psi.

Nereth
Posts: 302
Joined: Wed Jan 18, 2023 10:18 am
So just for anyone interested - FE Model of the 25 v 28 tyres to work out relative stiffness/response to a bump is coming along.

Model solves nicely though I suspect there is some discretization error going on. There's also limitations to the material model I've used to speed things up (you can see it as some rippling of the tyre carcass in the image below). I'm currently running some iterations of the model to confirm validity of boundary conditions, simplifications, make sure this stuff doesn't have a big effect, etc.

What you're looking is 2x half models of the inner boundary surface of the tyre. The carcass itself is not modelled. The wheel is treated as rigid and not shown. The half model concept is a common FEA trick to reduce comptuation time. You should actually be able to do a quarter-model and solve even faster in this analysis, but there is a software limitation that prevents it, and it was faster to just accept fate and solve a half model than sort that out.
A fair number of additional cases and a lot of postprocessing needs to be done before this can usefully add to the conversation unfortunately, so we're a while off.

I've actually seen papers before of similar analyses taking place so it's possible someone can do some google-fu and beat me to it.

Edit a few hours later
: No bueno - as I drop the pressure (this is now about 36 PSI) and raise the deformation, the tyre's skin buckles in a way I'm not convinced is realistic. Will need a more realistic material model to better approximate a tyre's fabric carcass. No longer a problem for today.
Skin buckling.png (81.21 KiB) Viewed 862 times

Requiem84
Posts: 217
Joined: Sat Oct 17, 2020 5:07 pm
Interesting video with some field testing by Peak Torque (quite a renowned tester I'd say). His summary based on his field testing is that a wider tire on a wider rim only results in a very marginal additional drag. The difference between a 28mm and 32mm at 35km/u is approx 0.9 watt total on this wider rim.

If you put the wider tire on a smaller rim, the CdA impact is much higher. You lose about 6watts there. Overall, the narrower rim with narrower tire is the fastest combination in his field test.

Step from a GP5k tire to a Corsa Next however is a muchhhh bigger loss in rolling resistance. That's a wopping 19watts at 35km/h.

apr46
Posts: 388
Joined: Thu Aug 26, 2021 1:46 pm
In Tom Anhalt's spreadsheet there is an equation to callculate the PSI needed for 15% drop for a given load based on Frank Berto's data. My understanding though is that this data is old. I think the Rene Herse pressure calculator is based nearly entirely on that formula.

That said, its awesome to see armchair engineering opinions put to the test, even if those opinions are mine.

AJS914
Posts: 5511
Joined: Tue Jan 28, 2014 6:52 pm
GCN did the ultimate hysterisis test seven years ago. They tested a road bike, a cross bike, and a mountain bike over the a 3 minute section of cobblestons. The wider tire won.

https://youtu.be/QvO74sZxVs4?si=ig2HoXYd0Vmmf3Sd&t=388

cajer
Posts: 769
Joined: Sun Jul 14, 2013 1:26 am
Requiem84 wrote:
Wed Jul 24, 2024 4:00 pm
Interesting video with some field testing by Peak Torque (quite a renowned tester I'd say). His summary based on his field testing is that a wider tire on a wider rim only results in a very marginal additional drag. The difference between a 28mm and 32mm at 35km/u is approx 0.9 watt total on this wider rim.

If you put the wider tire on a smaller rim, the CdA impact is much higher. You lose about 6watts there. Overall, the narrower rim with narrower tire is the fastest combination in his field test.

Step from a GP5k tire to a Corsa Next however is a muchhhh bigger loss in rolling resistance. That's a wopping 19watts at 35km/h.

Outdoor testing is not going to have the resolution to detect a <1 W aero difference at 35 kph espically with other counfounding factors like different tire sizes. For aero, you'll want to test at much higher speeds so that the aero signal is larger.

cajer
Posts: 769
Joined: Sun Jul 14, 2013 1:26 am
Nereth wrote:
Wed Jul 24, 2024 12:40 pm
So just for anyone interested - FE Model of the 25 v 28 tyres to work out relative stiffness/response to a bump is coming along.

Model solves nicely though I suspect there is some discretization error going on. There's also limitations to the material model I've used to speed things up (you can see it as some rippling of the tyre carcass in the image below). I'm currently running some iterations of the model to confirm validity of boundary conditions, simplifications, make sure this stuff doesn't have a big effect, etc.

What you're looking is 2x half models of the inner boundary surface of the tyre. The carcass itself is not modelled. The wheel is treated as rigid and not shown. The half model concept is a common FEA trick to reduce comptuation time. You should actually be able to do a quarter-model and solve even faster in this analysis, but there is a software limitation that prevents it, and it was faster to just accept fate and solve a half model than sort that out.

25v28 FE Model.png

A fair number of additional cases and a lot of postprocessing needs to be done before this can usefully add to the conversation unfortunately, so we're a while off.

I've actually seen papers before of similar analyses taking place so it's possible someone can do some google-fu and beat me to it.

Edit a few hours later
: No bueno - as I drop the pressure (this is now about 36 PSI) and raise the deformation, the tyre's skin buckles in a way I'm not convinced is realistic. Will need a more realistic material model to better approximate a tyre's fabric carcass. No longer a problem for today.

Skin buckling.png
What does it look like at higher pressures and how are you modeling the tire?

If you're adding any shear (not just a vertical displacement) I would think buckling is to be expected espically at low, 36 psi, pressure.

EtoDemerzel
Posts: 425
Joined: Sun Dec 17, 2023 4:13 pm
Nereth wrote:
Wed Jul 24, 2024 5:21 am
I've modelled up a 25mm tyre and wheel, taken the internal volume (996700mm3), then I've just deleted a slice from the internal volume to generate an appropriate contact patch for 40kg of vertical force on it at 80PSI of pressure (target 711 mm2), and taken the volume again (995987 mm2), to get delta volume (713mm^3) and thus rise in pressure (+0.06PSI).
"taking a chunk out of it[the tire]" isn't what happens irl. The shape changes, and not just on the contact patch.

cajer
Posts: 769
Joined: Sun Jul 14, 2013 1:26 am
EtoDemerzel wrote:
Wed Jul 24, 2024 7:56 pm
Nereth wrote:
Wed Jul 24, 2024 5:21 am
I've modelled up a 25mm tyre and wheel, taken the internal volume (996700mm3), then I've just deleted a slice from the internal volume to generate an appropriate contact patch for 40kg of vertical force on it at 80PSI of pressure (target 711 mm2), and taken the volume again (995987 mm2), to get delta volume (713mm^3) and thus rise in pressure (+0.06PSI).
"taking a chunk out of it[the tire]" isn't what happens irl. The shape changes, and not just on the contact patch.
If anything he's being conservative as with increased pressure you'll get expansion of the other parts of the tire leading to a smaller than what he saw pressure increase.

Also he's doing fea to more accurately simulate it.

youngs_modulus
Posts: 694
Joined: Wed Sep 20, 2006 1:03 am
Location: Portland, OR USA
Nereth wrote:
Wed Jul 24, 2024 12:40 pm
So just for anyone interested - FE Model of the 25 v 28 tyres to work out relative stiffness/response to a bump is coming along.
Nice work!
Nereth wrote:
Wed Jul 24, 2024 12:40 pm
Model solves nicely though I suspect there is some discretization error going on. There's also limitations to the material model I've used to speed things up (you can see it as some rippling of the tyre carcass in the image below). I'm currently running some iterations of the model to confirm validity of boundary conditions, simplifications, make sure this stuff doesn't have a big effect, etc.
It looks like you're using ANSYS. Which version? What materials are you using, and what material models? What element number are you using for the tire? (SHELL181? SOLID186? SOLSH190?)

Tires are among the most challenging structures to model, as you may know. They need not only fully anisotropic material properties, but also some really complicated hyperelastic material data.

Bias-ply tires are especially tricky. Per Jobst Brandt, the plies are typically at about 45º to each other prior to inflation. As the tire is inflated, those plies move relative to each other, decreasing the bias angle. This is what causes spoke tension to decrease as clincher tires are inflated—the tires compress the rim circumfirentially.

Are you modeling all that? If so, I'm impressed.
Nereth wrote:
Wed Jul 24, 2024 12:40 pm
What you're looking is 2x half models of the inner boundary surface of the tyre. The carcass itself is not modelled. The wheel is treated as rigid and not shown. The half model concept is a common FEA trick to reduce comptuation time. You should actually be able to do a quarter-model and solve even faster in this analysis, but there is a software limitation that prevents it, and it was faster to just accept fate and solve a half model than sort that out.
The above makes it sound like you're modeling none of that. (Please correct me if I'm wrong). The rest of the boundary conditions seem OK to me. If you are using ANSYS, what software limitation is stopping you from using quarter-symmetry? I'm not aware of one off the top of my head.
Nereth wrote:
Wed Jul 24, 2024 12:40 pm
Edit a few hours later[/b]: No bueno - as I drop the pressure (this is now about 36 PSI) and raise the deformation, the tyre's skin buckles in a way I'm not convinced is realistic. Will need a more realistic material model to better approximate a tyre's fabric carcass. No longer a problem for today.
This again suggests to me that the anisotropic bias-ply casing isn't being modeled at all. Some quick points about that:

- It's totally reasonable to start from isotropic, linear elastic material properties as you build your model. Crawl, walk, run, as they say. But without modeling the bias-ply casing, this model is limited in what it can tell us about casing deformation.

- Ideally, you'd be using frictional contact between the surface and the tire. Shear between these two is a nontrivial factor in rolling resistance. But of course, everything depends on what one is trying to learn from the model at hand—I don't think frictional contact is worth modeling for our purposes.

- Despite the point above, you don't need to model the bias ply structure in order to get the volume/pressure change due to large deformations. I think your Solidworks cut-off model captured that just fine.

FE analysts build entire careers around modeling tires...they are highly nonlinear in every way. It would be unreasonable to expect someone to build a detailed tire model in their spare time. There's a lot to capture, and not all of it is publically available information. So validation is another problem.

youngs_modulus
Posts: 694
Joined: Wed Sep 20, 2006 1:03 am
Location: Portland, OR USA
EtoDemerzel wrote:
Wed Jul 24, 2024 7:56 pm
Nereth wrote:
Wed Jul 24, 2024 5:21 am
I've modelled up a 25mm tyre and wheel, taken the internal volume (996700mm3), then I've just deleted a slice from the internal volume to generate an appropriate contact patch for 40kg of vertical force on it at 80PSI of pressure (target 711 mm2), and taken the volume again (995987 mm2), to get delta volume (713mm^3) and thus rise in pressure (+0.06PSI).
"taking a chunk out of it[the tire]" isn't what happens irl. The shape changes, and not just on the contact patch.
You're not wrong about this in a strictly technical sense, but Nereth's approach is right anyway—the shape changes don't affect the change in pressure in a measurable way. And as another posted out, Nereth's method is conservative—he's subtracted the largest volume possible for contact with a flat surface, throwing out the casing that he's "cut off." In real life, that casing contains some small air volume, resulting in a smaller pressure increase than Nereth reports. I would speculate that the difference is very small—maybe 3.89 PSI instead of 4 PSI for a full, rim-bottoming impact.

Does Silca say how they measured the pressure change? If they were using very high-frequency sampling, they could totally detect pressures higher than Boyle's law would predict. An impact creates pressure waves that echo through the tire, and the leading edge of those waves would produce a higher instantaneous peak pressure than could be measured with lower-frequency gear.

I don't know Josh Poertner personally, but I gather he's the kind of person who might want to know the magnitudes of both short-duration (microsecond) pressure changes and longer (millisecond) pressure changes. Piezoelectric pressure sensors are used to measure peach chamber pressures in firearms, so they're more than adequate to handle transient peaks in bicycle tires. Maybe he used something similar.

Visit starbike.com Online Retailer for HighEnd cycling components
Great Prices ✓    Broad Selection ✓    Worldwide Delivery ✓

www.starbike.com

EtoDemerzel
Posts: 425
Joined: Sun Dec 17, 2023 4:13 pm
This has turned into a 3d modelling tutorial
I honestly don't care how this stuff is modelled or the challenge to create something that is realistic. Though it's his thread so run with it.

Maybe the actual discussion on what is faster has run it's course.