Ah. Okay. I think I'm understanding better now. Sorry to have hit you with what you already know. Your motivation is a noble one, but you just can't get around the physics/thermodynamics that won't let your idea generate energy from a source not being depleted.
Let's say you aimed the magnetic pole of one magnet directly outward from each fan blade, meaning parallel to the axis of rotation around the axle that the blades are attached to. When you start, let's say you use your muscle power to start with the magnets almost directly across from each other so that their poles are strongly pushing. When you let go of the fan blades the magnets would push away from each other and cause the fan blades to rotate. This is because you've started the system with some potential energy that you gave it by positioning it yourself, by the way.
As the magnets rotated away from each other and the field decreased, let's say there was enough speed for it to go around enough to then reach near another magnet. The next magnetic field would repel the one coming around, right? So the device would slow down, then likely oscillate back toward the starting position.
Let's say you tried to "aim" the magnets to get them to turn the blades. So their fields are no longer parallel with the axis of the axle but are instead sorta skew to that, sort of as it might be if you had glued a fridge magnet to each of the fan blades.
In that case I think it's easiest to approach the solution by trying to imagine the full magnet field for each magnet, emanating from both ends of the magnet. You might at first think "well, at first the poles are repelling each other like they were before and there's also some of the other side being attracted toward the next pole, so it's even better this way." I'd understand that motivation too. Think of what happens as the fan blade rotates toward the next magnet though. The back of one magnet and the back other one approaching will repel each other. Even if you get get the fan blade to go far enough then the fronts would repel each other but the direction that their curvy fields hit each other wouldn't end up with a force vector that points in a positive direction.
It's pretty hard for me to do verbally, but don't take my word for it! I encourage you to try building one of these if you're really interested in it and see for yourself what happens.