I just stumbled onto this, but I'll give it a shot...
UpFromOne wrote:1. Since the hub flanges must be offset from the centerline in order to accept the cassette, the DS and NDS hub flanges experience different stresses. One goal is to equalize these flange stresses as much as possible.
Nope... nobody cares about that. No reason to. Just make the flanges strong enough for each load.
2. Another goal is to equalize the spoke tension on each side, to better keep the wheel true.
No... we are stuck with DS offsets of ~19mm max with S hubs and ~17mm with C. We need a much larger NDS bracing angle to get decent lateral stiffness and stability.
3. Another goal is to equalize the stresses at the rim holes. (The pulling force from each side of the hub)
That would be a good idea since rims are generally uniform all around.
1. A smaller (lower) hub flange experiences less stress than a high flange.
Not sure what you mean by that.
2. A longer spoke requires more tension than a shorter spoke to do the job (resist external forces on the wheel).
No. Bracing angle and tension are what matter... not length.
The assumption in a crossed DS + radial NDS is that the torque force starting from the cassette will reach the DS flange more so than the NDS flange. I don't agree. Modern hubs have enough material between the flanges, and are strong enough to almost completely resist torque equally at each flange.
Do the calculation and tell me what you find. You are assuming that the hub body is very stiff relative to the spoke pattern... and it generally isn't. The DS will carry the majority of the torque load even if the flange sizes and spoke pattern are the same on each side. The NDS will carry some of the torque if it is cross-laced, but there are tradeoffs to this.
It appears that the radial NDS pattern is trying (among other things) to reduce the NDS spoke length. After all, a radial spoke is shorter than a 2x spoke. Combined with a 2x DS, this seems to be an attempt to equalize the spoke lengths used on each side, with the implied objective of equalizing spoke tension because of more equal spoke lengths.
Nope, none of that matters... and it doesn't equalize spoke tension at all. The tension ratio depends only on the ratio of bracing angles.
The problem is that this only addresses spoke tension, without dealing with hub flange tension, or tension at the rim. A radial spoke, being shorter, will require less tension. Therefore, it aggravates the inherent difference in tension between the DS & NDS due to the uncurable offset. This is further aggravated by the flange design, which is typically lower on the NDS. Again, I have to assume that this parameter is an attempt to cure the spoke angles. The narrower DS angle will be relieved with a higher flange, thus angling the DS spoke out just a hair more.
The misconceptions and irrelevancies multiply...
So these modern hubs designed only for NDS radial spokes (especially the straight-pull spokes with an even lower effective flange diameter) are missing the overall effect. If I had to have radial on one side, I'd actually rather have it on the NDS (like ZIPP) than the DS.
There are a few good ways to design a rear hub, and 2x NDS with radial DS can work well... but I don't think Zipp's design is very good, because the NDS flange has too small a diameter.
A wheel is a pretensioned structure, and so loses a great deal of strength if that pretension is lost. One of your biggest concerns is to avoid spokes going slack. There is no concern about DS spokes going slack if they are used to transfer torque, because they are at such a high tension. The flange tends to already be large to clear the cassette mechanism, so that helps also. But the NDS spokes must be ~half the tension to get decent lateral stability... so you need to be careful about using them to transfer torque... especially if they do so exclusively. In some circumstances torque loads are far from trivial. A typical rider going steadily up a steep climb would cause ~+- 30kg tension change with 12 crossed spokes taking the load. A strong heavy rider on a steep climb would be significantly worse. Since NDS tension tends to be in the 45-60kg range, you are flirting with spokes going slack when you add in radial and lateral loads as well.
If you *wanted* to design a hub for 2x NDS and radial DS, you'd make the NDS flange large to reduce the tension excursions due to torque, and make the NDS offset fairly low. Of course it is best on any hub to use every bit of DS offset you can. The advantage of 2x NDS and radial DS is that the peak spoke forces at the rim would be lower. You could theoretically make the rim lighter... but you'd also gain weight at the hub by making the NDS flange large.
Getting back to your original question of 2x 2x vs 2x DS radial NDS, the differences are pretty subtle. By cross lacing the NDS you reduce the torque excursions on the DS a little bit, so that is a fine idea if you do a lot of climbing. But it isn't going to really help unless your rim is weak and prone to spoke hole cracks. Radial spoking looks slicker and is a little more aero and lighter. That's about it.