«

»

Mar
25

Can’t Get No Satisfaction

Reset action is the principle source of source of overshoot, interaction, and limit cycling because reset is never satisfied. The PID integral mode will continue to ramp for insignificant offsets from setpoint. If the offset is less than the resolution limit of the measurement or valve, the best action is no action. Also, integral action in the traditional PID is acting on old information. For large update times whether due to slow wireless update rates or analyzers with large sample-cycle times, the integral mode in the traditional PID is acting on stale possibly erroneous information.

The integral mode also has no sense of direction. The measurement could be accelerating toward the setpoint. Integral action would promote the acceleration (proportional and derivative action would try to decelerate the approach). Since operators normally see numbers rather trajectories and humans often don’t have the patience and anticipation necessary in the process industry due to deadtime, reset gives the operators what they want. Consequently loops tuned to meet operator expectations looking at digital values have too much reset action (Are we Misleading our Operators? and Tips and Techniques (T-N-T) – Reset has no Sense of Direction ). This tendency to have too small of an integral time (too small of a reset time or too many repeats/min) is particularly true for temperature loops on large volumes. Often the simplest thing to do for these temperature loops is to increase the integral time by an order of magnitude. You often find you can then increase the gain by a factor of ten unless there is measurement noise and you have no noise band setting.

I do not mean to imply we should go with P-only controllers. Reset is needed for load disturbances and low controllers gain settings. I would just like to tell reset “Give it a Rest”, “Get Real”, “Have a Sense of Proportion”, “Patience is the better part of Valor”, and finally just “Chill.” The enhanced PID developed for wireless with a noise band setting to prevent updates from noise only makes an integral calculation based on new information stopping the continual ramping offering the following benefits:

(1) Elimination of limit cycles from valve backlash and stick-slip

(2) Elimination of unnecessary crossing back and forth of the split range point

(3) Elimination of unnecessary switching between override controllers

(4) Less wear of valve packing and seals by the elimination of dither, limit cycles, and unnecessary stroking

(5) Prevention of self-inflicted disturbances by eliminating reaction to noise

(6) Elimination of interaction between loops by eliminating reaction to insignificant changes

(7) Better valve position control for dual valve (big small valve) control by the elimination of limit cycles and interaction between the VPC and process PID

(8) Better valve position control for optimization of feed rate and utility use by the elimination of limit cycles and interaction between the VPC and process PID

Many of the above benefits have been demoed and discussed in Deminars #1, #2, and #12 or will be in Deminar #13.