The Latest

We also hand build custom alloys. Rims by Pacenti, Stans, HED and Kinlin. Hubs by Miche, WI, Chris King, DT, Tune and PowerTap.

Site Search

Disc Brakes in The Mythical 40k TT

Now that has posted the complete data for the wind tunnel test of disc brakes versus rim brakes, it's time to take a look at the time gaps you could expect to see across the mythical 40k TT.

As you see in the Velonews graph, the speed difference between the disc bike and the rim bike is greatest when the wind is blowing from the drive side of the bike. The drive side of the bike is normally the faster side of a bike with rim brakes. The dynamic that causes this is also nearly certainly responsible for this difference - to put it in the simplest terms we can, you're adding sail area to the disc side. This helps to overcome some of the drag added by the disc setup, and makes the disc side of the disc bike relatively faster than the drive side of the disc bike. 

However, this difference is most profound at greater wind angles - wind angles that don't commonly come into play. There have been a bunch of studies that show that even in fairly windy locations, at the speeds we typically ride, we are most often riding with an apparent wind angle of 10* or less. This study that Trek did while developing their SpeedConcept bike is a great read on this (section 3.2 has what you're looking for). Our convention has been to use the weighting that Tour Magazine uses for their tests, which places a similar emphasis on wind angles of 10* or less, but logically brings wider angles into play at slower bike speeds. The slower you go, the more likely you are to experience wider wind angles, although wind angles of greater than 12* or so are still quite rare. 

Between -10 degrees (the drive side is the negative side) and +10 degrees, the extra watts of drag that the disc bike incurs are fairly symmetrical. As you get out to the wider angles, the drive side becomes relatively slower, while the non-drive side becomes very equal to the rim-braked counterpart. 

Thanks to this distribution, the delta between the disc and rim bikes in the mythical 40k TT are similar no matter which side the wind is from. Imagining a flat, out-and-back 40k TT course where the wind is blowing from the left on the way out and the right on the way in, the time costs for the disc brake bike are:

@30mph - 3.6 seconds out, 4.7 seconds in - 8.3 seconds total

@25mph - 4.5 seconds out, 4.5 seconds in - 9 seconds total

@20mph - 3.5 seconds out, 4.5 seconds in - 8 seconds total

My New Years resolution (I haven't made one in about a dozen years) is to use subjective language as infrequently as possible, so we will simply let those numbers speak to themselves. We're sure that the forums will host fierce battles for all perspectives. Our objective (see what I did there?) in doing this exercise was merely to quantify the difference between the two setups in order to give people the information that allows them to make the decision that best suits their purpose.

I've had lots of articles published in mags, but never a pic. It's cool


Wind Tunnel Test of Discs v Rim Brakes, pt 1

On June 24th, we had a good idea. We had a trip to test a bunch of stuff at A2 planned out, the results of much of which we've published previously. Since we'd seen a rise in road disc wheel activity, we thought the time was right to include Rails with discs. It's such a small step from ignorant guessing to knowing, all it takes is a test.

Well said, sir!

The good idea was when we thought "you know, no one's really published anything remotely definitive about the aerodynamics of disc versus rim brake wheels - maybe some bigger media outlet wants to work with us on the story?" Before lunch, we were hooked up with Velo, with the proviso that the story would be about the whole package - wheels in the bike. 


Since we didn't have "except for brake format" race-oriented rim and disc brake frames of our own at the time (we do now), Caley Fretz at Velo offered to arrange getting two frames to A2, and off we go. 

For those keeping track at home, June 24th is a challenging date on the cycling calendar. Life is in FULL swing. Caley was off to France to cover a bike race, we're going like mad trying to keep on top of orders, but we were behind ourselves getting to the tunnel and didn't want to delay it anymore.  We scheduled the tunnel for July 28th (which is already a lot less hectic for us than June), which would give everything plenty of time to happen even in light of the TdF and everything else. 

The only thing more expensive than paying for testing in a wind tunnel is paying for not testing in a wind tunnel. Despite confirmation that frames had shipped, frames hadn't shown up and there was no good info on where they were or when they might arrive at A2. Since I was driving down and had made a bunch of arrangements to do other stuff in concert with the trip, plus our desire not to delay the trip anymore, we kept the schedule even as the frame component started to look a little questionable. We rearranged schedules by a day to give the frames an extra day to arrive, and kept on. Fortunately we were able to be super productive during the day and a half that we were there when the frames were supposed to hit, because the frames never showed. We never doubted the Velo component, but it won't be a surprise when I say it felt like not everyone involved was playing it straight up. What are you going to do? We did our testing, made our contingencies, and when the frames never actually showed up, we went home without that piece accomplished. 

Thanks to Caley's persistence, the frames eventually showed up at A2. Plan A was just to have A2 run the tests on our tab, with me "present" by remote connection- basically Facetime. Plan A never works. When A2 unpacked the bikes, there was a lot of work to do in order to net out differences in the frames. The disc bike was Di2, the rim brake was mechanical. Seats were different. Bars were wrapped differently. Too much noise. Another high and hard fastball, the degree to which this was within the pitcher's control is up for debate. So I saddled up a jumbo jet and flew back down to A2 to equalize the bikes as much as possible. When the bikes went into the tunnel, they were as equal as they possibly could have been - the only differences were the differences elemental to disc versus rim brake bikes. A2 sent the data files directly to Velo, and I shipped a bunch of photos off.

As alike as they can be made

With this accomplished, the only difficult part was keeping mum about what we had done and learned. You spend that kind of dosh to make that kind of a leap in your understanding of things, and your instinct is to start shouting about it post-haste pronto. Nope. Gag order until the December issue dropped, which was scheduled to happen in early to mid November. Tough, but worth the eventual exposure. Patience is not my strong suit, this was agonizing. 

Then the December issue came out, and it was the awards issue, no sign of our test. After my coronary event subsided, I learned that there had been a shuffle in the editorial calendar and it would be in the January issue. Not ideal, but okay, just a couple of weeks tacked on. And then, Monday, this beauty landed in my inbox. 

Of course nothing is ever even that straightforward, right? Of course not. Flipping through Twitter last night, I see a "how much do discs really slow you down?" tweet.



Hmm, we didn't take a video so what's with the YouTube link? Oh. Specialized decided to publish their own test, from their tunnel. Coincidentally, one day after Velo drops the issue with our story. That sure is one heck of a coincidence, huh? They've got the resources do it, and we're certainly in favor of more info being out there for consumers. The problem is that their test sucks - they left sloppy differences in place between the two bikes, and they only tested with wind from one side. If you guessed that the differences from one side to the other are absolutely nothing alike - congratulations, you win! 

Those of you who get Velo will have seen or will soon see the data for yourselves, and we'll be able to talk a heck of a lot more about it soon. At this point, we're glad that objectivity is starting to displace conjecture, and happy to be at the forefront of the discussion.


Wheel Design Charts and Graphs

If anyone's left standing, I've got a couple of quick graphs that help explain two of the more important considerations that go into our wheel specs.

The first is simply the % of spokes you are gaining or losing when you add or subtract spokes. This helps illustrate the point of diminishing returns in spoke counts. The series of points starts at 16 spokes and ends at 40. % change in # of spokes going from 16 through 40 spokesAs you can see, when you go from 20 to 16 spokes, you take away an impressive percent of the spokes in the wheel. Going from 36 to 40? Not so much. The points start to become more clustered at 24 and really get tight at 28. Early on, each spoke you add is a big increment of the number of spokes you had. Later on, each additional spoke becomes less significant.

The next chart is of a concept I call "unsupported span." This is simply the distance between spokes.  Again, as you add spokes early on, you chop that unsupported span down quite quickly. Later on, the gains are smaller. This will have an impact on both radial and lateral stiffness of the wheel. Rim stiffness comes into play a lot here, in a way which I think of as "bridging."  Think of an 8 foot long 2x4 resting on a set of saw horses that are 5 feet apart - there's going to be almost no "sag" deflection in that 2x4. Now think of an 8' long sheet of plywood resting on the same saw horses. Huge sag. Put another saw horse halfway between the two original ones, though, and the sag goes away.  

16 thru 40 spokes once again, based on a rim ERD of 588

The law of diminishing returns is evident here again, coincidentally with the inflection coming in that 24 to 28 spoke zone. Must be something to that sucker...

Last is a chart of weight gain, as a percentage of the total wheel weight, from adding the last 4 spokes (so 16 spoke weight gain % is from a 12 spoke wheel). This is based on a 450g rim, a 240g hub, and 5g spoke + nipples.  

Again we see a line that is not straight, but there is no pronounced inflection. 

The old convention of 36 spoke wheels (the time when 32 spoke wheels were those new-fangled get offa my lawn weight weenie wonders was not that long ago at all) was based on very shallow and very soft rims. Those days are long gone, and today's deeper rims made from harder metals change the dynamic of how wheels work. However, increasing structural imbalances in the move to ever more dished rear wheels work against that. You want to build a strong, durable, stable, stiff wheel, but you balance those features against aerodynamics and weight.  

If the tryptophan doesn't get you tomorrow, this might do the trick. Have a nice Thanksgiving.  


The comment that became a post

Yesterday's blog inspired a comment question so involved that the answer requires its own post. We've never done this before, but it gives us a chance to articulate how we tie all of the testing that we do into better recommendations and, hopefully, better wheels for you. I've (severely) edited/paraphrased the questions for brevity, but you can read the whole comment at yesterday's post. Incidentally, we probably have to spend some time figuring out how to catalog all of this info.

Q: How would 2:1 lacing affect the parameters that you describe here?

A. We’ve actually previously laced a wheel 16:8 on a 32 hole rim and tested it for tension balance and lateral stiffness. The tension balance was exceptional, the lateral stiffness was exactly the same as if it had been a 24 spoke wheel laced 2x/2x (it was compared directly to one of those). For radial stiffness, we haven't measured it. Presumably it's got a bit of an advantage because there are more drive side spokes, and so they're in closer proximity to one another.  The quantity and bracing angle of non-drive spokes seems to affect wheel lateral stiffness as much/more than anything else. I think that wheel is still built, if it is I will give it a check when I have a chance. Not sure that my measuring system is sensitive enough to parse out that difference. It's also built on a Kinlin rim, so that would be noise in the answer. Here is where we talked about that setup 


Q: Does spoke count affect lateral stiffness?

A: We've done much more measuring of lateral stiffness than radial. Here is the most recent post about how spoke count affects lateral stiffness. Lateral stiffness is more affected than radial stiffness by hub selection, and is also affected by rim stiffness.  In the post prior to this one we quantified the lateral stiffness of all the rims we currently use - those aren't guesses they are quite precise measurements.  The short answer is that every rim/hub combo has a point of diminishing returns for lateral and radial stiffness.  For lateral stiffness, that depends on the “recruitment point” as I think of it – given the rim’s stiffness, how far can the support imparted by one spoke span. This is easy to see in building wheels – on carbon rims, tightening one spoke will loosen the adjacent spokes. That’s recruitment in action, as that spoke’s tension is spanning past its neighbors. 


Q: How about quantifying the tradeoff between aerodynamic penalties and structural gains from more spokes?

A: This would be an incredibly involved piece of measurement, and I don’t think you could ever make it totally accurate.  I’m going to do what I rarely do here, and break my arm patting us on the back here. We’ve done tunnel tests of a front wheel at 20 spokes versus an otherwise identical wheel with 24 spokes (a small but measurable difference. We’ve published tunnel tests of bikes with 34/34, 34/52, and 52/52 wheel combos,which is instructive on the relative importance of front wheel aerodynamics versus rear (front wheel is way more important). We’ve done lateral testing up the wazoo, and now radial testing. So we’ve tested every discrete component to that answer. I don’t see where anyone else has published even one of those component tests, and quite frankly I doubt they do them.  The rub is that you have to credibly quantify performance impact of radial and lateral stiffness in order to gauge that versus aerodynamic gains and losses. Given the tests we’ve done, I have no problem claiming that we are better prepared to say that adding four spokes to a rear wheel does precisely jack squat to your system’s aerodynamic performance, so that there is no value in trading for fewer spokes until you reach the point of diminishing stiffness returns. In effect, every wheel recommendation we make to a customer or inquiry already encompasses this info. That’s why we do as much testing as we do.  Admittedly though, we are making that tradeoff in light of durability vs performance. I don’t know how to do it as performance vs performance, at least in a way that will reach consensus. 

On alloy wheels, the 20/24 convention is fashion, pure and simple. There is no objective support for this convention at all, but it does engender a lot of positive eyeball aerodynamics estimates, and I suppose helps bikes look cool on web sites. 20 spoke fronts are often desirable and defensible because they are adequately stiff and strong as is (front wheel structure is better), so we often recommend them. For carbon rims, 20/24 is very often at the sweet spot anyway.  For very very deep rims we’d go fewer spokes, but we have no plans for a rim that deep.


Rim Strength And Lacing

More charts and graphs today! I finally figured out how to illustrate a point that's been banging around in my head for a long time, which is how rim strength affects lacing. We've talked about lateral stiffness in rims a bunch of times (enough to earn ourselves the clearly affectionate sobriquet of "fun sponges"), but radial strength is as big a deal. 

I took 2 rims, 2 built wheels, and a large clamp, and simply applied pressure at 12 and 6 o'clock to show what happens to rims and wheels when they are loaded this way.  The two rims were a Rail 34 and a Pacenti SL23. The two wheels were a Pacenti/Miche 28h rear, and a Pacenti/WI 24h rear, both laced 2x/2x with CX Rays. I chose these rims because they are both stiff relative to their category, plus I already had the Pacenti wheels built up for something else.

Wheel testRim test

To do this test to lab standards would require time and fixtures/equipment not currently at our disposal, but as an illustration of the concept this worked beautifully.

Part 1 - Compress the rims: With a rim loaded into the clamp as shown above, I applied clamp pressure by the trigger until I couldn't squeeze anymore. With the Rail rim, that point was reached once the clamp's pads compressed. The Pacenti rim was much easier to squeeze. I didn't want to ruin the rim but compressing it 1/4" was easy. Once unloaded, it snapped back into roundness. These results were completely as expected. A heavier and/or deeper alloy rim would have resisted compression better, but is that worth it?  Read on..

Part 2 - Compress the built wheels: The exact same protocol was used for wheels as rims - squeeze the trigger until I couldn't, this time measuring the change in spoke tension that resulted. This admittedly inexact loading was used simply for lack of the equipment to precisely repeat the loads, but since a large part of my life is spent squeezing and plucking spokes I'm exceptionally well calibrated for this kind of thing, and, well, this loading was pretty close.  

As expected, the 24h wheel's spokes both gained more tension in the 9 o'clock - 3 o'clock axis and lost more in the 12 to 6 axis, along which the load was imparted.  This is simply the axiom of "many hands make light work" in action - with the adjacent spokes closer to the spokes most directly affected to the load, they were able to help the most affected spokes carry the burden.  If I was a graphics ninja I would graph this but instead you can interpolate the graphs below.

28h DS spoke tension loaded v unloaded24 DS spoke tension loaded v unloaded

As you can see, the spokes in the 24h wheel saw a bigger change than the spokes in the 28h wheel. The NDS spokes acted exactly the same way, to no surprise. In both cases, the NDS spokes at 12 and 6 o'clock went completely slack, but the 3 and 9 o'clock spokes gained less tension in the 28h wheel than they did in the 24h wheel.  

The more spoke tensions are cycled in this manner, the more quickly the wheel will wear out. This will manifest as either non-drive spokes breaking (usually at the hub), or drive side spoke holes in the rim cracking from stress (or, less likely, spokes breaking at the hub). This whole dynamic is why we are generally in favor of more spokes than others when it comes to alloy rear wheels. A heavier rim, as mentioned, would counteract this behavior - hence the number of 20h rear wheels (OEM and otherwise) that don't explode under riders. In our opinion, adding weight at the rim is a bad trade there. Adding 4 spokes makes an easily seen difference at a cost of 20 grams - to us, that's the best trade.

This isn't a perfectly precise mapping of everything that goes on in the wheel (wheels get loaded between the wheel perimeter and the hub, for instance), but it's a great illustration.