Posted by: Harold Ennulat | April 6, 2010

PID verses Advanced Controls

For those that are members of the LinkedIn “Automation & Controls Engineering” group there is an interesting discussion about the reasons that advanced controls are not used more often in industry and why the PID controller is so dominantly in use.

The full discussion is at the link below:

My comment which is included in the discussion thread is also shown below:

A couple of reasons for sticking to PI control where you can.

1.  as mentioned it is the best understood control.

2.  it is generally fairly forgiving when the process changes over time or those valves start sticking because generally PI control is not anywhere near critically tuned due to being applied to systems with lags.

3.  PI control implemented in PLC’s and PAC’s etc, can be readily enhanced with additional features to handle a myriad of special cases like startup, system upsets, handling known process changes, and so forth.

Model, Predictive, and other advanced controls are implemented in a black box sort of way.  It seems to me that this doesn’t easily allow any of the parameters of the model to relate to the physical application.  Also, even slight changes to the process can require an entirely new model or predictive or advanced controller.  PID controls can usually just be retuned to handle the process changes.

It seems then, that in order to use advanced process algorithms effectively the expertise of the engineer needs to be captured inside the controller somehow.

The model needs to model not just the entire process as a black box, but I would suggest that it needs to be implemented by modeling the parts of the process and connecting these building blocks into the completed process model.  In this way when things go wrong, the model can be adjusted, say if the valve gets changed out for example, then the valve portion of the model can be retuned according to (hopefully) well understood rules.

A story of my experience working with advanced controls…  I worked for a company that had a guru that enhanced a well known predictive modeling control algorithm.  Tuning the controller in actual systems turned into largely “educated trial and error” or so it appeared to me.  This meant that there was no expertise or rules available to retune the system or make other adjustments.  It was back to trial and error.  When I had to make some modifications to part of a process that required me to port this controller I learned during startup that the controller no longer had any impact on the process, but it was just doing its thing not really benefitting anything.  Of course no one had a clue including me till I tested it during commissioning after I had been so careful to preserve the functionality.

For an article on PID control about when to use the D in a PID controller, see:

Updated: 4/19/2010 7:50pm CST  |  Created:  4/6/2010 7:00pm

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s


%d bloggers like this: