Thanks for this post and for the analysis... this is why we don't yet have 'an app' that does all this for people.. We discuss this exact topic in the marginalgainspodcast episode on asymmetry and have more to come in the next one, but you are spot on that these blanket statements, 'wider is faster', 'narrower is faster' etc, are such gross oversimplifications that they just aren't that useful for any of our purposes anymore, and this data shows how convoluted this can get when you try to look deeper!
This is why doing the testing and keeping a log can create a substantial advantage in the real world... consider that 80% of the people don't know the general rule of thumb and then maybe 5% of people are really willing to do the reading of the current data, and that less than 0.5% of people are willing to actually do the testing.. you can create some pretty asymmetric advantages for yourself if you just dig a little deeper and do some testing. This is why I still keep a full spring schedule of traveling all over Europe working with teams on this stuff over and over again... each year we know more, but if has yet to come anywhere close to 'automatic'..
My couple of thoughts to add here... using tire drop percentage was ground breaking and relevant when Berto did it originally and it has served us well for a long time, but for this purpose it is neither accurate nor really useful when comparing tires at a range of widths. There are nonlinearities at play which make this method unreliable for the purposes of Crr and it tends to under inflate larger tires from my experience.
Second, the bicycle rolling resistance site does awesome work, but doesn't have nearly enough roughness to approximate even good pavement, nor does it have the capabilities to measure impedance, much less the ability to look at impedance at a range of roughnesses. I really like what he did adding Crr at same initial drop of 4.5mm but even this isn't quite accurate from my experience and data as wider tires recruit casing adjacent to the contact patch much more readily than narrower ones, so this model also probably under inflates the larger tires by a few % though is better than the 15% drop method.
Third, much of what is generally missed in the 'wider is lower Crr' is that historically for most tires that had a range of widths for 'same model' of tire, the tread was made separate from the casing and then glued together. Much of the benefit of the wider casings was that for manufacturing purposes the constructor only has a few widths of tread, so the percentage of tread, which is a thicker, higher rolling resistance zone of the tire, tends to get smaller at higher widths as the tread remains the same.. so for Vittoria tires, the 21, 23 and 25 tires of old all used the same tread width.. so the 25 had that much more casing, similar for the Dugast and FMB so beloved by the pros. This is true with any number of brands, but will not exactly be true with the new Vittoria TLR or GP5000 as each tire is produced from it's own casing and rubber layup in it's own specific mold.. now from the data I'd imagine that the 23 and 25mm probably share the tread strip or puncture strip pieces during the layup and similarly the 28 and 32 likely share one or both of these parts as you can see the grouping in the data that they are very similar to each other but different from the other group... and of course this is before we even begin considering the differences between individual layups and the molds used to make each tire. I can tell you from experience that one of the GP4000 molds was considerably more aero than the others and another one of them was consistently lower Crr than the others and I'm sure that the same effects are at play here as well as relatively small manufacturing differences such as how much excess rubber squeezes out of the tool and pressure differentials within a given tool can have measurable effects in the finished product despite the tools all being 'the same' from a production point of view.
Finally, there are also very real aero penalties to larger tires, so the key is to be using them at the right pressures on the right surfaces, this is why we are still running 20mm tires on indoor velodromes, 21 and 23mm tires in TT's on good surfaces and 24-26 for most road racing surfaces with 27-30mm for rougher/harsher conditions and cobbles. Much of the real benefit of the wider tires on rough surfaces is that they widen the target pressure zone, sort of flattening out the hockey stick that we talk about in the podcast.. you just have a little larger margin for error, while also having a bit more safety from an impact damage perspective.. and of course that is also remembering that most of the tires we are using at the ProTour level are using the same tread strip between 27 and 30mm... so there is a bit more of a natural Crr benefit than you see with a molded tire like the GP5000
http://www.SILCA.cc Check out my podcast, inside stories from more than 20 years of product and tech innovation from inside the Pro Peloton and Pro Triathlon worlds!
http://www.marginalgainspodcast.cc