I think shielding may not be such a hot idea. If you start capacitively coupling the signal to a ground shield, you may well attenuate the signal, the very opposite effect you're trying to achieve. The twisted pair really simply minimizes the introduction of external EMI to the signal since any magnetic field cuts both conductors and is cancelled out, the whole purpose of twisted pair.
I think process control with all the EMI in that environment simply adjusts the drive of the signals to compensate for any losses from the shielding. Also, the frequencies involved in the control signals may not be as sensitive to coupling to ground, lower frequencies will have less attenuation for a given capacitance to ground.
