One purpose of setpoint tracking is for a bumpless transfer from auto to manual control.
Suppose you are in auto control mode for level control and current setpoint is at 20%, but your manual SP is at 90%. You don't want your level to shoot up to 90% in a flash do you? This allows you better control.
If you're talking about the old "set-point tracking" controller feature, the advantage is that the controller setpoint tracks the output while in manual so there is no bump when you switch the controller back to auto. Unfortunately the operator then needs to adjust the setpoint back to its original value before the controller was switched to manual. This was sometimes called "bumpless transfer".
If you have a complex control system with internal setpoints that the operator can not get to or doesn't understand, you may have to design your internal control loop to reestablish the setpoint by either:
1 "Not tracking" and tuning the controller to bring the process back to a fixed set point without an unacceptable "bump". or
2 research the feedback options of the controller you are using and see what features are available to index the setpoint to some new value based on the final control element signal and other process conditions. This might include a set point ramping feature, or a sample and hold feature that is watching some other control loop, or a target feature that won't allow a transfere until the output and setpoint are back to their original relative values.
Mark Illian @ synergysvcs.com
Set point tracking allows you to transfer your control from open to closed loop without process upset as the set point will be the same value as the process variable just prior to transfer. Disadvantage is that you have to re-set the set point to your desired value. Another approach, set point ramping, allows the controller to "step" back to original setting after the transfer. Some proprietary algorithms also allow for "power back" function, which ramps set point and automatically adjusts the set point step size using a predictive algorithm, based on process variable rise rate.
Setpoint tracking is a tad different than explained below.
In Auto mode, the controller moves the control valve (or final control element) to keep the process value (PV) at setpoint (SP).
In Manual mode, the human operator moves the control valve directly. The controller may be doing the control calcs in the background, but the controller outputs are ignored and the operator's manipulations take precedence.
The challenge comes when changing the mode from Auto to Manual and back to Auto. When going from Auto to Manual, the control valve goes under the control of the operator at the last position directed by the controller's calcs. The operator may then move the valve to any position at his/her discretion. Since the PV will be changing as the valve is moved or the process changes, the old SP from Auto mode will no longer match the current PV.
Thus when the controller is switched back to Auto mode the current PV and old SP may be far apart causing the controller to make quick, large adjustments to the valve. This can upset certain processes.
Setpoint tracking is an optional controller feature that causes the Auto mode SP to be constantly updated to match the PV while in manual mode. Thus when the controller is switched from manual to auto, the SP equals the current PV and a "bumpless transfer" is made. Little or no movement of the control valve results. The operator would then be expected to slowly shift the SP to whatever is appropriate to return the process to typical conditions.
Gerhardt & Associates
New Orleans - Boston
The other reason for setpoint tracking that hasn't been mentioned is to eliminate integral windup. Some controllers (e.g. AB SLC) keep calculating the integral term of the PID equation even when the loop is in manual. I find PV tracking useful in cascade loops where you start the process up using the slave controller and then switch the master loop in to control in steady state, if you don't have the master loop in PV tracking mode you will find that the slave SP jumps as soon as the master loop is activated even if you have set it up for bumpless transfer.