The concept looks quite reasonable.
The needle bearing idea will probably work, but not as easily as you think. In a perfect world, that being, all the materials came in at size, you will have some interference. A bit of machining will solve the problem. I would also consider spherical UHMW bearings, instead of the roller, if you think you need a bearing at all. All you really need is a retainer if it is floating.
Don't expect your schedule 80 grey pvc to be super rigid.
PVC is also a very poor bearing material (not even listed as a bearing material). In Cecil's application, the PVC is acting more in the fashion of a retainer to keep the axis of rotation, well, retained, and not as an actual bearing. Put a cylinder in water with external forces without retaining the axis of rotation and it will be uncontrollable. I personally would not worry about PVC as a retainer, but don't use it in a load carrying bearing application.
What media are you using?
Since you posted this. I was playing around with the idea of using 10MM twin wall corrugated polypropylene as media. That would work quite well. A 2ft diameter x 4ft long RBC would be in excess of 1100 sq ft, plus it is buoyant.
Look into the history of RBC's and you will find that they were Engineered for treating municipal waste water, with good success. The problem with the axles snapping and bearing/mechanical failures came about when some fish guy's, without engineering skills tried to duplicate this. They use to use plywood for media. These however, were the pioneers in Aquaculture, and a Big Thanks to them!!!
RBC's for Aquaculture have come a long way since then, and everyone has their flavor.
Nice device for doing the task at hand. And as I said earlier, you can thank the Aquaculture Pioneers for this.
Let us know how things are going.
