Basic PID; Can someone give me the lowdown? NO
What you seek is an “easy out”. Many people generalize that a single PID loop should fix all their current controls problems.
There is talk of “auto-tuning” PID controllers, and purchased “one-size-fits-all” auto-tuning software. There is talk of analogies of the “cruise-control” in your automobile as a prime example of a simple PID.
I call “Bull-****!” One size fits all. (poorly). Each system(s) have fixed mechanical dynamics, and more importantly, each process has different speed and accuracy requirements.
There are dozens of variables that can deem a loop simple, or not. If a loops Pv is highly responsive, then a simple PI loop with little effort can net great results.
BUT, when there are large lag times between CV output and PV response, then, each loop needs to be tuned to customer requirement.
Is the loop bi-polar where it increases with one set of gains, and decreases with an alternate set of gains?
Does the process require FAST response, and tolerant to overshoot, or does the process require NO overshoot, but time is important? You cannot have your cake and eat it too. There are some physical mechanics that you cannot solve with one simple PID loop. Sometimes, the customer asks to get to a physical destination, in a time frame that is NOT achievable, PERIOD.
Recently, we launched some equipment that had some thermal requirements to satisfy the exacting process to achieve expert control to make REPEATABLE process at temperature. We collected data during launch to understand the mechanics of the system and deemed a single PID with large setpoint changes from Ambient to Idle, to Process, was NON-LINEAR.
The take-away observation was, that it was not prudent to try one PID, with one set of terms, to satisfy the entire spectrum of the physical make-up of the system.
Based on our observations, we noted a stable ramp of PV regardless of how hard the CV was kicked.
Our solution was not to do a single large step change to the Setpoint and let a single PID attempt the task, but to merely ramp the setpoint based on the typical dynamic physics of the system. When we did this it was observed that the PV climbed the ramp with a predictable lag. During the ramp, we used a soft integral gain. Once the ramped setpoint flat-lined at the desired setting, we dynamically increased the integral gain, and the PV walked right up to target, without overshoot.
Executive Summary: We abandoned the pipe dream of Old-School, Single PID to solve our customer requirement. We instead, used empirical data from our real system, and created a programmatical, dynamic, change on the fly PID, solution to achieve highly repeatable performance results.
