Velocite-RT50-2nd-gen-front_r3qVelocite is constantly pushing the envelope of design, releasing the 2nd generation of the RT50 wheel set. Aimed at being aero, but not just in a straight line, the design story on these is long and technical, with Velocite breaking out computer hardware and software not typically used in the cycling industry.

They say the main objective in this redesign was to fine tune the handling of deep profile rims in crosswinds. The methods employed were cool to see, click through to get the scoop…

Velocite RT50 CFD simulation 12_degress_y_slice

Computational Fluid Dynamics is a type of simulation software to simulate airflow over an object in the computer. Velocite says it is like a wind tunnel, except that using CFD allows the experimentation to happen before the product is created and designed. They say this creates better products because it designs the product for the flow of the air, where as a wind tunnel can only validate or verify something that is already designed. This software was created with the help of the University of Geneva in Switzerland. It is so complex, that less than 1 second of simulation results in an enormous volume of data, and they can see all airflow around the wheel when compiled and played back in slow motion.


A seamless brake track is another new feature that is integrated into the rim profile, and makes for a simpler and stronger layup of the rim. This is made with a special resin with a higher transformation point of 377º Fahrenheit, minimizing the chance of a brake track failure due to overheating.


The rim profile is a popular 25mm wide, but not for the same reasons as other makers. Velocite still strongly advises the use of a 23mm tire on the wider rim, because it completes the airfoil shape they used in simulations to maximize the aerodynamics. Using a wider tire would negate that effect and add weight, neither of which are their intent for this wheel set. The rim design is also for road tubeless tires, and a liquid sealant should be used.


The wheel set is completed with the Gram Triplet hubs that use a24-count, 2:1 rear spoke ratio, meaning twice as many spokes on the drive side of the rim. The front has 20 evenly distributed straight pull spokes.

Velocite-RT50-front_edge_onIn the video below, this is the data taken from the CFD analysis, stretched out to 7 seconds to highlight what can be seen. The full development story can be found here, and if you understand the technical bits, it’s very fascinating. Velocite’s RT50 is available now, direct from the company for $1,889 per pair. That includes spare spokes and nipples, brake pads and warranty card, and they come pre-taped and have tubeless valve stems installed.

Weight is 1,660g, which isn’t the lightest 50mm deep wheel we’ve seen, but you save 100g or so by not needing a tube. Plus, the decals are intentionally easy to remove, which’ll save a few grams and clean up the looks a bit.


  1. jinknobat, to be fair they have done a bit more than Hed did 7 years ago. It’s more in line with what Enve and Zipp did 4 years ago.

  2. Yeah, was just going to say that. Despite the fancy computer modeling, still came out pretty similar to every other wheel out there. Not to be a hater, just don’t see what’s so extraordinary about the end product.

  3. CFD software is used by most of the top end aero companies on the market. Not sure why these guys think they’re special in that regard.

  4. Tyler, can you speak to the 23mm tire at all? What I find interesting is that while the 23mm tire might complete the aero profile best, is it actually going to be faster overall when you consider that most roads are that smooth?

  5. That’s what everybody has been doing for at least 5 years on wheels, frames, helmets, etc. And many did that at least for some projects since the 90s.

    But whatever… for the vast majority of us most of these gains are more than cancelled by what we ate 2 days or 2 hours ago.

  6. @AlanM, pro teams have gone to 25mm, especially on the rear wheel, because it reduces rolling resistance and improves grip and comfort. Thin=fast is an oldskool road cycling myth. That’s only true on the smoothest of roads, which barely exist in the real world.

  7. @JeroenK, I’m very aware of the switch to 25mm from my work with some teams. It would be interesting to see data that is more relevant, not just from a wind tunnel, which only takes into account one of the factors a rider faces. A 23mm tire might be more aero in this setup, but is a 25mm tire still faster overall? Just the things I think about.

    @Dude, aero improvements might beat rolling resistance improvements, but there is always so much more at play than static data from a computer. How much of an aero gain are we talking between 23 and 25mm? What about rider fatigue from a harsher ride? Again, I know there isn’t going to be one answer, but food for thought when companies use computer data that only focusses on a single element.

  8. these guys actually show what they’re doing with significant details.
    just saying “we use CFD” and not showing anything doesn’t tell much. everyone can use CFD. Not all software is equal, and not all uses are the same either.

    For the state of the art CFD you can’t just run it on your workstation because its very resource consuming and you dont wanna wait 2 weeks for the results. What i find interesting though is that they seem to be using real hardware instead of say, AWS, for scaling.

  9. It often times is not easy enough to say a 23mm or 25mm tire is faster. In the little data that’s out there, width seems to mostly affect stall. Also, rolling resistance increases in proportion to speed, whereas aero drag increases moreso. So whether a lower RR, or more aero tire is faster often depends on rider speed and the wind conditions.

    All that said, the Conti gp4000s has proven again and again to roll fast, and be very aero in the 23mm version soon. And really it runs close to 24mm, particularly on wide rims.

  10. Anyone can do a CFD analysis. The question is whether you did it right or not. A lot of people will do a CFD run on a part, but how did you know you did it right? The only way to tell if your model is right or not is through wind tunnel verification.

  11. Horses for courses. Even if 23 might be more aero, it’s going to work differently under a 170lb rider vs a 140lb rider and even if it’s more aero, that’s not necessarily faster overall if you’re less confident in fast corners or on rough or loose stuff and lose the wheel in front of you. Even so, it’s not like it’s *not* going to be aero with a 25mm tire, and it’s not like 23mm is “un-raceable”under most circumstances.

    Either way 25mm+ works really nicely for me personally and it would be cool to see more stuff optimized for that size. Wondering if that would make for a tight fit in the chainstay/bb area on a lot of current frames though.

  12. I don’t believe them at all. I think they are lying and are simply buying generic Taiwanese molded carbon rims and adding their own branding.

  13. Haha there is a 25 and also 23 mm wide tubeless carbon farsports rim now. Not saying I think they’d do that.

    Tubeless carbon is interesting I agree with Tyler on that point. More players in the game is fine with me I never understand why people get all hyped up saying who’s idea it was first, waste of time.

    Regarding aero vs crr, this is the hot debate in tires – wider versions of same tires (new zipp, spec tires) have better crr, while narrower (24 vs 27mm) are slightly more aero. You’d have to measure both sizes for aero and crr, and then convert that to time over a distance to find the ideal. I would say that the 24-25mm width of the fast crr tires are the fastest. Currently my favorite combo is the new zipp slspeed tubulars 24mm – that’s what I have on my 404 tubs, and on clinchers I have 23mm Conti 4000s II in front and Zipp Speed Clincher 25mm in back.

    Look at this guy’s data for some interesting data –

    It’s just roller crr testing, but it’s fun to look at.

  14. Bang on, @scotty. The software is only as good as the guys interpreting the data. I do get a heck of a kick out of this article, though. To the earlier points, who ISN’T using CFD these days? Maybe next we’ll hear about a breakthrough company making bikes from (gasp!) CARBON FIBER!

  15. Thank you Tyler for the post and thank you commenters for your interest.

    The word CFD is now almost required when describing any bicycle product with pretensions for aerodynamics. Unfortunately (or fortunately if you look at it from our perspective), not all CFD is created the same. Legacy, multipurpose CFD software is indeed difficult to set up as the operator needs to predefine regions of interest and ensure that the solid 3D model is adequately represented by a high quality surface mesh. Thus, when using legacy CFD software getting a result is sometimes not difficult, but it is very hard to ascertain whether the result is physically valid or not.

    Our CFD software has a single purpose: simulate external airflow over 3D objects at “low” airspeeds (actually Reynolds number < 1 million). It uses a Lattice-Botlzmann method with immersed boundary "meshing" for solid objects and Large Eddy Simulation turbulence modeling. This methodology is just about perfect for the situations encountered by bicycles moving through air. It is not used widely (or by anyone in the bike industry) as it is computationally very expensive, and it is not widely supported, if at all by legacy CFD software on which engineers trained.

    The advantage of this method is that there is no simulation setup as such. We do not have to predict what the airflow will do before we simulate it. Our software just simulates everything. The entire simulation domain is resolved in the same resolution. The other advantage is that the method scales very well across additional computation nodes, so we have our own multi core in-house computation cluster, whose capacity and power increases as we add additional compute nodes.

    There is also no meshing requirement, however we have to make sure that the input file (STL in our case) is just about perfect in quality. In traditional terminology, our surface "mesh" for the RT50 rim is about 6 million cells.

    The last big advantage of our CFD method is that we simulate moving objects directly. The model meshes are moved, not the frame of reference (ie. air), thus we do not need to be concerned about correctly allowing for air inlet directions and downstream processing. We just declare at what rate which component is rotating, or periodically moving.

    Thus what we do is very new, it is very powerful and we will have more new products to show you that are a result of our CFD and R&D process. In fact we also offer our CFD services to other companies. Having our own equipment and expertise allows us to do any number of iterations and "what if" experiments with the sole focus on discovery, not cost or time.

    Wind tunnel testing of the RT50 will happen soon, however I first need to verify that our wind tunnel service provider is instrumented for measuring torque. Aerodynamic drag among the latest series of well developed aero wheels is pretty close now.

    …and some of the commenters (Matt and greg) are eerily prescient 🙂 We have a new product under way that is made of carbon, but not as it's even been applied by the bike industry, and we do in fact have a FEA project too, again not the usual stuff and it will hopefully tell us just how much strain energy is stored (and power dissipated, if any) due to flexing of individual bike components.

    Victor, CEO Velocite Tech.

  16. I am starting to think I am an idiot for not going tubeless. I have had 3 flats in the last month. Can someone please tell me why I should not go tubeless?

  17. I used to have fun riding bikes. However, now with the switch from 23 to 25 I fear that it will become much more difficult with the added areo drag and rolling resistance. It is with great sadness that I may need to abandon my beloved sport as it has now spiraled out of control to place that I’m not physically capable of.

  18. Folks, I wonder how many of you own a math or science degree. It doesn’t take much IQ to ride a bike, but it does take some above average knowledge to build them and to explore the possibilities in the technologies. The fact that Victor takes the time to create and verify everything with the latest, and some “before their time”, measurements and tests, before he offers them to the public is rare. There is no hype, no marketing. No almighty bottom line, either. Just science and improved products through diligent use of available technologies and real life testing. Sure other companies test and use CFD, wind tunnels, etc. to what end? For many it’s a marketing tool to the ignorant. Don’t be ignorant. Take the time to do some homework and maybe add some useful knowledge so you can debunk the BS coming from your coffee house riding group. Don’t just be mediocre and regurgitate loads of crap you’ve heard, or read. You’re only as fast as your body and fitness level allows. The improvements in bicycle technology allows you to go a bit faster with the same wattage output. I rather see real science applied to real world situations and then simulated to produce the best parts available at any price. Wouldn’t you?

  19. Velocite is the real deal. It isn’t some generic part rebadging company. I ride one of their frames and it is outstanding!

    As Tyler pointed out, full carbon tubeless is very rare. I think only one other company produces such a rim.

    My only knock against tubeless is the rolling resistance of the best tubeless tyres still needs some improvement.

  20. tl;dr summary of my excessively technical post below: this is not an especially creative or interesting use of CFD, and the analysis includes some questionable decisions. Finally, Velocite may believe this is “extraordinary” only because they’re unfamiliar with the state of simulation in general.

    As others have pointed out, there’s nothing “extraordinary” about this application of CFD. The biggest difference is the use of a lattice-Boltzmann solver, though that is a technical detail; the analysis and its results are not different from anything solved conventionally (via the Navier-Stokes equations). Velocite is using the Palabos solver, which is open-source and available to anyone who cares to download it.

    Velocite makes a big deal about how they don’t increase their mesh density in areas of interest. It’s strange to claim this as an advantage, especially because Palabos doesn’t allow mesh refinement in 3D models. Velocite uses a constant mesh size not because it’s a good idea but because they have to.

    Additionally, their mesh size of 2.56 mm is far too coarse to capture the boundary layer, especially at the tire/rim junction.

    Finally, the use of FEA to quantify strain energy and power dissipation is also not especially new. In fact, I did just such a simulation for the Fairwheel Bikes crank review a year and a half ago:

    I’m not saying that the CFD modeling is wrong, but without knowing more about it I can’t say that it’s right. The mesh size issue gives me pause.

    But neither the use of a lattice-Boltzmann solver for CFD nor the attempt to quantify strain energy in FEA is especially novel.

  21. @Jason

    tl/dr version:

    Sorry for the lack of technical information in my reply, and on our blog post. You and anyone else familiar with or interested in CFD, FEM/FEA and computer aided engineering and research in general are welcome to get in touch with us directly and I’ll be happy to discuss everything in as much detail as you would like.

    We are hardly unfamiliar with the “state of CFD simulation in general”. We have used various solutions all the way from custom DNS (direct numerical simulation), “standard” Navier-Stokes (Fluent, Comsol, OpenFoam and derivatives), had and have lengthy discussions with CD-Adapco, trialed XFlow, worked with Janusz and Sailfish, and many others including more custom GPU based solvers thanks to our relationship with the NCHC (national supercomputing center) and The NCKU (very good university in Taiwan).

    Palabos is a library, not really a ready to use application. What we have is a custom made low speed virtual wind tunnel application built using the Palabos library in an arrangement more aptly described as commercial open source. In my mind this is the best of both worlds – we get commercial grade software and support, but the solver code and the mathematical models are open to all researchers in the field to see and contribute to. Lattice-Boltzmann field is very active and new research is continually refining and adding to the body of knowledge. Both Palabos and Sailfish incorporate new breakthroughs as they become available. Legacy, closed source code by default does not benefit from this scientific (or crowd-sourced) process.

    Regarding the use of constant resolution domain, yes it is both a limitation, but also a benefit as for wake resolution the only mathematically consistent way to do it is to resolve the entire domain to the same resolution. Modified LBM solvers like XFlow and a few others have models for adaptive domain resolution, and one of the custom codes that we used on a GPU did that as well, but it is not optimal and in my mind does not capture the full benefit of LBM when it comes to its ability to accurately resolve wake. A simpler and better way is to increase computational resources and grow the cluster or buy some supercomputing or compute cloud time.

    Regarding the boundary layer, completely true, but the boundary layer thickness depends on the Reynolds number so we make sure that we stay within the range where we can resolve it. We are not simulating fast moving vehicles or aircraft. Bicycles are perfectly fine.

    Regarding the work that Fairwheelbikes did, we are actually using it as one of the very few reference points for our work. There really isn’t much research in this field. The difference with what we are doing is that we are simulating a bike, not just a crankset so we will hopefully be able to contribute towards answering the question “does frame/wheel stiffness matter, if it does, how much?” We are using Comsol for this.

    Victor, CEO (and R&D Director) Velocite Tech.

  22. @Jason again

    After I replied I noticed that you were actually the person who did that Ansys FEA work on that Fairwheel Bikes article. I was meaning to contact you, and (still will 🙂 ) once we have completed our simulations to show you what we did. Besides using your work as a reference, it was also an inspiration for me to finally start our own project.

    I have been meaning to get this done for a while…there should be some remnants of my discussions in that one Slowtwitch forum post about out work with the NCHC from a couple of years ago or so.

    We have the empirical data on our (and others’) frame stiffness, and the prevailing manufacturing and end user position that stiffness is good for efficiency and “power transfer”, but I am not familiar with any hard data. Even in your work there was some dissent in the comments section so I want to be able to finally address that and focus the arguments into refining the physical model and the maths, rather than staying with the same old argument about whether the energy returns to the bicycle, or whether it is wasted.


  23. Hi Victor,

    I was perhaps too blunt when I wrote that. Also, I left out an important part. What I meant to write was that the article made you sound like you were “unaware of the state of CFD and simulation in general *in the bike industry.*”

    The headline of the BikeRumor article was pretty breathless, and I still maintain that using different (but very interesting) tools does not by itself imply an extraordinary use of computer design. So my point was simply that anyone who would think that the application of CFD to bicycle aerodynamics was extraordinary simply hasn’t followed the industry. But again, this was probably too blunt; I apologize.

    Furthermore, I imagine that you would agree that the application of CFD to bicycle wheels is not in itself extraordinary. That was the point I meant to make, and I could have done it better.

    I have to say that your use of LES made for some nifty looking plots. My previous wheel simulations (in CFX) have all used K-Epsilon turbulence models because LES was too computationally expensive for the hardware I was running on. I’d be interested to see whether LES models give more realistic results when there’s interaction between the leading-edge rim wake and the trailing edge of the rim.

    I would be glad to discuss your FEA work on strain energy whenever you’re ready. I’d be interested to see how you guys handled certain problematic aspects of this sort of modeling…I’d love to know if you guys figured out a better way to handle them.

    I’m not inclined to post my contact info here, but if you get in touch with Fairwheel, they should be able to put us in touch with one another. Or you could just get an account at the Fairwheel forums and PM me there.

  24. @Jason

    no problem and I understand. Contrary to one of the earlier comments, we did not write this article, but I am glad that the technology that we used inspired the authors and hopefully the readers.

    The unique aspects of what we do are:

    – apply full LBM and LES to bicycle models through a custom, single purpose application. I am not aware of anyone using LBM and LES for bicycle work. There is some limited resolution XFlow work which uses modified LES, and some work was done with DES model with another commercial solver.
    – we have the compute cluster and resources in-house
    – our capabilities are fully scalable and growing.

    You can find some of our simulation videos showing the wake interactions on our YouTube channel here:

    I will get in touch with you regarding the FEA work through Fairwheel, thanks!


  25. @ everyone

    after researching the CFD servicing industry in particular with Ansys it was very clear to us that inexperienced one dimensional non-specific computer aided programmers were simply supplying hyped-marketing via a white elephant generic data interpretation smoke screen !

    hence we were ready to begin working with the highly accredited Ansys & the experienced skilled team that implements their software until we contacted Victor Major the real deal in house bike detective whom we have been effectively collaborating with for the last month !

    their is absolutely no question about his authenticity, nor his honestly & pure passion for transparent scientific methodology while being based in Taiwan the heart land of the bicycle industry, where his integrity, wealth of knowledge & demand increase on a daily basis !!!

  26. I believe that these tubeless rims are exactly the same as the ones used for the Julius CC50 wheel set ( However, those wheels only cost €800.

    This means that the CFD analysis is performed afterwards by Velocite and the rims are not designed by them.

COMMENT HERE: (For best results, log in through Wordpress or your social media account. Anonymous/fake email comments may be unapproved or deleted. ALL first-time commenter's posts are held for moderation. Check our Comment Policy for full details.)

This site uses Akismet to reduce spam. Learn how your comment data is processed.