Doesn't whirpooling with a pump stir up all the trub or does it stay sitting in the centre of the kettle?
yes, it can do: it's taken me a few goes to get the nozzle right. As Dr. K said, the trick is to introduce the wort tangentally, and at the right velocity. My whirlpool nozzle is a 1/2" tube in a long radius 90degree bend, so it returns the wort parallel to the wall of the vessel. I'll go out later, take a photo and post it.
There's been lots of studies done on whirlpool design, and there are lots of variants- usually with regard to the shape of the bottom of the whirlpool and the area where the cone collects. The general concensus is that the velocity of the tangentally returning wort needs to be in the range of 2-3 m/s (off the top of my head- I'll retrieve a citation for this figure when I post a photo later). Different studies and organisations debate the exact rate, and it is dependant on the size and shape of the whirlpool, as well as how high up the nozzle should be.
I was aiming for 2.5m/s, but for a 1/2" tube this works out to close to 40 L/min, which is a pretty big flow. I can only do 15 L/min, so I need to start it moving with the paddle first. I've found pump only from start to be ineffective in my setup, but once it's moving the rate I supply it back seems to keep it going nicely.
The other thing about pumping is eddy formation, which tends to disturb the cone, rather than help it form. Eddies or vortices are the little mini whirlpools you get that trail the paddle, or your hand as you move it through a body of liquid (or air for that matter- you just can't see it). Internals in a whirlpool are bad, because they shed eddies. The jet from the return nozzle can also induce eddies, which makes the return port/nozzle, velocity, and the operation of the whirlpool important. The whirlpools that I've studied in micros and big breweries use a pump to get the whole volume in the vessel moving. Once it's up to speed, they turn of the pump and let it sit for a "whirlpool rest". Once the pump is off, the flow inside the vessel becomes laminar (no eddies) and it forms the cone then, gently slowing down to a stop. Because DMS can continue to be formed above 60degrees, usually the wort is cooled during the recirculation step.