The crisscrossing of the split range point is a common problem due to the severe discontinuity and stick-slip at the closed position where seating friction and changes in the process gain is the greatest. The crisscrossing from steam to coolant for temperature control, acid to base for pH control, and vent to supply for pressure control causes a significant loss in process efficiency besides introducing process variability. An integral deadband in the PID controller and gap in the split range point may help but is difficult to adjust, slows down the transition on a set point change, and messes up tuning tests and methods.
The enhanced PID developed for wireless applications have the potential to eliminate oscillations across the split range point and provide a fast response to set point changes. A wireless transmitter may actually improve the setpoint response since the controller gain can be set equal to the inverse of the process gain per white paper but is not essential to get the benefits from the enhanced PID.
The enhanced PID has been demonstrated to eliminate the limit cycle from stick-slip or deadband in Emerson Exchange short course EmersonExchange2010Session204ProcessControlLabs.pdf and in the white paper DeltaV-v11-PID-Enhancements-for-Wireless.pdf. Check out the August 5 entry on this website for a summary. No adjustment or tuning is required. A similar benefit is realized at the split range point. To suppress oscillations across the split range point, the PIDPlus option in DeltaV V.11 is just selected in FRSIPID_OPTS.
The split range point can be set to help compensate for the change in process gain when moving from heating to cooling, acid to base, and vent to supply. A more general solution that also compensates for the change in process time constant and dead time is to use adaptive control and gain scheduling (DeltaV Insight).
For jacket temperature control the transition from chilled water addition to steam injection to go from cold to hot water can shock the temperature probe and cause a kick back to chilled water. The signal might also get noisy from bubbles hitting the sensor. To prevent this overreaction and noise, the jacket temperature sensor should be on the outlet of the jacket. The jacket volume then moderates the transition. Shinskey has shown that the dynamics in terms of what the primary reactor temperature loop sees is about the same whether the secondary temperature loop measurement is on the inlet or outlet of the jacket. A jacket inlet temperature loop may provide a faster reaction to coolant temperature upsets by elimination of the transportation delay to the jacket outlet in the secondary loop but this is not as important as dealing with steam shock.