The turbine rotor typically reacts to temperature changes faster than the shell, so it will grow or shrink faster (in length):
the shell (turbine case) is pinned (anchored) at a given spot so it will grow and shrink relative to this pinned point (varies from turbine style to turbine style and from OEM to OEM, but usually around the low pressure shell or turning gear area)... the turbine rotor itself is generally fixed by the thrust bearing -- this is all designed so that under normal conditions, as these two entities grow and shrink, the turbine clearances (spaces between rotating and stationary parts) will remain within acceptable ranges (some clearances are bigger than others due to the relative growth differences in the shell vs turbine due to these pinned points)
the difference between these (growth and shrinkage) is refered to as differential expansion: typically, rotor long occurs when the rotor heats up (too fast) faster than the shell and the clearances on the generator side of the rotating parts gets smaller (assuming the generator is toward the low pressure end of the turbine) -- this can occur if the start-up is too fast and the rotor comes way up on temperature while the shell is still soaking it in... or sometimes the slide plates haven't been greased and won't allow the shell to expand as designed (like under the front standard)
rotor short is just the opposite: the turbine shrinks too fast and the clearances on the governor end of the rotating parts gets smaller -- typically when the steam temperature is drastically reduced due to water, attemperator problems, etc... (or due to hot starts when boiler temp isn't up sufficiently and the turbine & case are still hot, but the rotor will cool off too fast)
ideally, the start-ups (and to a lesser degree) the shutdowns have established guidelines for matching steam temperatures and turbine temperatures along with soak periods etc. to keep the differential expansion within acceptable guidelines... in this case you will still see the rotor expand (and contract) relative to the shell, but the clearances are never even close enough to create rubs...
