This is probably a better conversation to have on a thread on ES, but here goes for the theory of what options we may have.
The issue with wet conditions is that as the tire rolls through the water, the water is pulled up off the road and remains in contact with the tire. By the time the tire has rotated around to where the friction roller comes in contact with the tire, the water has created a peak by the centrifugal forces. This water is then rolled over by the friction roller. We then expect the water to be squeezed out off this contact region. If enough water is squeezed out then there is enough contact area between motor and tire to sustain the required shear stress to transfer the drive torque.
With a smooth roller, and slick road tires the water ends up just forming a boundary layer and we lose traction very quickly. (In fact this is how bearings pull oil in to the contact region between ball/roller and the inner/outer races to provide lubrication and reduce wear.) With a rough roller, and slick road tire, the water has some valleys and cracks to hide in, while the peaks are able to break through the water to touch the tire and transfer the drive torque.
So option 1 would be to change the tire to something with some tread, to give valleys for the water. But if the roller is still smooth you end up relying on having a quite defined sharp edge to the grip tread. (Just ask the formula 1 guys how important a sharp edge on their tread is for wet weather grip). This sharp edge allows the water to be pinched out when the edge first comes in contact with the roller (or road), and can ensure you get contact direct from tire to roller/road. But during normal dry use this edge is quickly lost, and a friction drive is likely to wear it out too. But it would definitely be better than slicks.
Option 2 would be to use the rough roller. But this has the downside of having higher tire wear during normal operation. Lets explore why that is. Well it turns out that for the flexible tire to conform to the rigid roller, the tire does actually shear across the roller surface slightly, as the convex tire surface get pushed into concave, and the tire bulges sideways in the process. Now with a smooth roller the coefficient of friction between roller and tire is too low to exceed the shear strength of the tire rubber, so it just slides across the can. But with a rough roller the shear force can exceed the shear strength of the rubber, leading to wear. So what can we do to avoid this. (a) change tire material,
- perhaps a compound that can sustain greater shear before failing
(b) change tire profile geometry,
- by change to a tire with a flatter profile, the tire deforms less when conforming to the roller, thus reduce shear stresses, and wear
(c) change to a tire with tread that has tread blocks small enough to "sway" instead of locally shearing the rubber. This would also give space for water to escape. But small tread blocks will not be supported by the rest of the tire jacket like slick, so they will need to sustain more shear forces through the whole block.
Option 3 you suggested was a more compliant drive roller. This compliance could share some of the conformance, and shear loads. Therefore reducing the forces that can wear the tire. But for wet conditions we still need to stop a water boundary layer from forming. So we need to give the water somewhere to go, get rid of it before it reaches the drive, and have surfaces that come in contact that can help stop a boundary layer from forming, like sharp edges. The rubber on rubber contact may have a better coefficient of friction in dry conditions, but unless we stops the water separating the drive and driven surfaces, it won't work in the wet.
Personally I would put my research efforts towards a tire with tread first, and then second a rough roller of some sort that helps pinch through water to ensure contact. Pick some tire treads that give water somewhere to run, a flatter profile, perhaps small blocks not too high, and tightly spaced not an off road knobbly. Not to sure about the small blocks theory, as the blocks will tend to sway, leading to uneven wear...
The other thing we can do is reduce the drive torque when in the wet. Not ideal I know, as this me less assist. But I am in the process of adding a max torque limit to the software of the brain box which could easily do this.
Actually I might go do a bit of a search now and see what I can find.
Please note this is mostly theory at the moment, I have been considering this as a fair weather drive personally, and haven't done a lot of testing in the wet. As most of my electronics during development has not been water proof, so I have generally not taken the bike out in the rain.
Okay here are a few tires I might consider (based on a 10 minute browse of Chain Reaction Cycles)
Road Tires
http://www.chainreactioncycles.com/Models.aspx?ModelID=28140
http://www.chainreactioncycles.com/Models.aspx?ModelID=52622
http://www.chainreactioncycles.com/Models.aspx?ModelID=35763
MTB Tires
http://www.chainreactioncycles.com/Models.aspx?ModelID=28141
http://www.chainreactioncycles.com/Models.aspx?ModelID=8547
http://www.chainreactioncycles.com/Models.aspx?ModelID=53651
http://www.chainreactioncycles.com/Models.aspx?ModelID=24632
http://www.chainreactioncycles.com/Models.aspx?ModelID=3244
http://www.chainreactioncycles.com/Models.aspx?ModelID=51012
http://www.chainreactioncycles.com/Models.aspx?ModelID=53953
http://www.chainreactioncycles.com/Models.aspx?ModelID=11093
If/when I test this I will probably try a combinations of:
- few different tires,
- motor with and without grip tape,
- in the wet,
- possibly with different maximum torque setup, to ensure I have the same torque when I go through puddles independent of speed.
Then see what I learn.
A more novel approach would be to install an air blade (think dyson hand dryer) that blasts a thin high velocity jet of air at the tire jsut before the contact with the motor, to displace the water.
Well that is enough theory for now.
I might post this to my thread, as I may spark some other ideas.
Cheers,
Adrian