I hate to speak for Peter but he may have intended to link to this paper:
http://www.apicsllc.com/apics/Ie_cdc01/Ie_cdc01.htm
The equations talked about in Peter's link are pretty standard fair for most decent winders. The stuff talked about in the link I posted is quite a bit more advanced. As a side note, I think the guys at apICS performed the anbalysis and wrote the pseudocode for the Allen-Bradley plc based winder function in the PLC5. I think they also helped with the AB 1336 Impact and Force auto-tune function.
I always react a little bit when people say that projects like this can't be done in a plc. We put men on the moon with less processing power than you have in your calculator. At the speeds you are running you should be able to easily keep up with roll build in a plc. I've done both low speed and high speed winding with nothing more than a plc. However, if you haven't done this before you are probably better off using a more purpose built tool for the job. Motion controllers have built-in tools to handle axis synchonization much better than most plcs. Your implementation and deployment would be much faster with a motion controller.
I don't necessarily think you need to go with servomotors on this job from a performance standpoint. However, you will need to use high quality AC drives in this application. A standard volt/hertz inverter won't have the capability to do what you want. The funny thing is that when you size this the servomotor system may be about the same price as a good AC drive system.
I do agree that you need to wind under some tension. If you don't wind with some minimum amount of web strain you will end up with a roll that is too soft to do anything with. It will act like multiple sheets of material hung over a bar. See if you can fit a load cell roll between the conveyor and the winder. Since you want to wind at a low tension and at a low speed you can probably get away with a small idler roll. Just make sure if you use a load cell roll that the angle of wrap around the load cell roll doesn't change with roll diameter.
You need to make sure you have a good web speed reference for your winder. That will help alot with staying on tension.
I don't think that a torque based system will work well in this case unless you up-size the motor quite a bit. At the speeds you are running the gear ratios will be fairly high if you size the motor to the load. High gear ratios will prevent you from separating the torque losses in the drivetrain from the torque required for tension. Torque control works best for very high tensions or very low gear ratios.
Good luck.
Keith