In reply to Taiden:
If it is a steam system, which I believe it is, the t-stat when satisfied will close a solenoid shutoff valve to the rad or supply piping. Steam when deadheaded in a pipe, valve etc. will start to condense as it cools. A single drop of condensate when caught up in the inrush of steam (100 mph kinda inrush) will cause the water hammer as the drop hits any obstruction as pipe fittings, coil etc.... yea, a single drop. If you leave the t-stat turned up so the valve doesn't close I bet the hammer stops.
Any large campus will have a central steam boiler where high pressure steam is piped to individual buildings and pressure reduced to operating pressure at point of use, a quarter mile or more is not uncommon. Not much you can do to eliminate the condensate (and hammer) at the rad inlet unless there is a bypass around the rad to the trap, which you ain't got.