We conclude with a ten step implementation procedure for helping you get the most out of your cascade control system.
(1) Pick a fast secondary measurement with enough rangeability to correct for nonlinearities and disturbances. Flow is the most popular secondary measurement because it is relatively fast and can compensate for nonlinear valve characteristics and pressure upsets. However differential head meters may lack sufficient rangeability for some applications. A common triple cascade loop is vessel temperature to jacket temperature to makeup coolant flow, which makes the primary loop linear and corrects for coolant makeup temperature and pressure upsets and non-ideal control valve behavior. If you have a positioner on the coolant valve (highly recommended), you have a quadruple cascade. If you have a digital valve controller (DVC) as your positioner with an inner loop of actuator pressure, you have graduated to a quintuple cascade control system.
(2) If you have a positive feedback network for the integral mode in your secondary PID and have fast reliable feedback of the variable that the secondary loop is manipulating, enable external feedback (“Dynamic Reset Limit”) in the secondary PID and provide the manipulated variable for external reset feedback. Fieldbus read back is fast enough for any valve with a pneumatic actuator whereas HART read back is fast enough for very large pneumatic actuators. Some variable speed drives (VSD) have tachometer feedback creating an inner speed loop. The use of speed for external reset feedback is particularly useful for dealing with overly conservative maximum ramp rate settings in the VSD.
(3) Remove set point filtering on the secondary loop.
(4) Tune the secondary (inner) loop first for a fast response to set point changes. Consider set point feedforward in the secondary loop for a low secondary PID controller gain (< 0.5) to make the secondary response to setpoint changes faster.
(5) If you a positive feedback network for the integral mode, enable external feedback (“Dynamic Reset Limit”) in the primary PID and provide the secondary loop PV for external reset feedback.
(6) Put the secondary loop in the remote setpoint cascade (RCAS) mode.
(7) Make sure the output limits of the primary PID match up with the setpoint limits of the secondary PID.
(8) Add a PV noise filter to the primary loop just large enough keep from unnecessarily moving the secondary loop set point.
(9) Tune the primary (outer) loop for a smooth response. The primary closed loop time constant must be at least five times larger than the secondary closed loop time constant to eliminate any interaction between the primary and secondary loops. If the secondary loop time constant cannot be made faster, you must slow down the primary loop time constant by decreasing the primary controller gain.
(10) Add feedforward signals as necessary to the primary controller output to improve its response to measured disturbances. Add a PV noise filter to the feedforward signal just large enough to prevent unnecessary movement of the secondary loop set point. Add dynamic compensation (delay and/or lead/lag) to the feedforward signal so that the correction by the secondary loop doesn’t arrive too soon or too late relative to the disturbance at the same point in the process.