Virtual is Never as Good as Real

Call me old-fashioned. I have always been a fan of real books. I simply cannot read an on-line article of more than two pages before I start clicking on something else. Yet I can sit down and read through a good paperback from page one through three hundred. I love the feel of a book, the smell of it, the satisfied “thump” when I close a hardcover after finishing it.

What does this have to do with process control? Well, we seem to be in a prolonged world-wide movement away from the “real” world, and into the “virtual” world. It’s not just books. Audio has moved from real vinyl (The grooves actually show the music!) to CD to the virtual iTunes. A walk in the woods has been replaced by Discovery Channel in HD. And control system vendors are actively seeking to create more and more “Virtual Plants”, for training, and even operation.

But we have to be very careful about this. The Virtual Plant will never be as good as the real plant. As my friend Ben Weinstein taught me “All models are wrong. Some are useful.” Models can never tell you as much as the real plant.

As a young engineer, I spent months training with maintenance technicians and operators who knew how to listen to the real plant. Even as we implemented DCS systems, they insisted on routine walk-arounds through the process. Why? Because they could employ all of their senses.

These people didn’t just look at the video screen. They listened to the hum of the equipment. They felt the vibration of the floor. They listened to motors by holding the handle of a screwdriver to their ear, and touching the blade to the motor. They actively smelled all parts of the process. (OK, so that wasn’t always my favorite part!)

And from all of this sensual, tactile experience, they learned about how the process was behaving. They could immediately identify when something was not quite right. They could measure things for which there was no instrument.

I like technology, I really do. And there have been tremendous strides in modeling, simulation, and the creation of virtual plant worlds. But we should always try to become better process control gurus. And a great way to do that is to get out of the control room, and walk around in the real plant. Pay attention. Listen to the frequencies in the hum. Look closely at things. Feel the floor vibrate. You will be surprised at what you can learn in the real world.

Never forget that all virtual worlds are based on models, and “Models are always wrong!”

——————–

Thanks to Nancy Bartles from Control Magazine, for posting a link to this insightful blog post, which was the inspiration for this post.

 

First Order Plus Time Delay

All process control gurus cut their teeth on the first-order-plus-time-delay (FOTD) process.  Let’s take a look at what this means.

A first order process is one that shows an exponential response to a step input.  Let’s take, for example, a pot of warm water on a heater.  If you increase the heat input to a higher level, then the temperature will rise.  The temperature rises quickly at first, then it slows down as it levels off at a new, higher temperature.  The shape of the response curve can be represented by an equation:

DeltaT * [ 1-e^(-t/T) ]

where t is time and T is the effective “lag time” of the process.  The longer this lag, the longer it takes the process to fully respond.  99+% of the process response will be seen after 4 lag times have passed.

First Order Plus Time Delay Response to Step Change

The “time delay” is the part at the beginning, where nothing happens for a little while.

Many, many industrial processes behave similar to this FOTD response.  For example, flow controls, pressure controls, and some simple temperature controls all have FOTD responses.  So we can use this same model to understand the process behavior, tune control loops, and train operators about how the process will behave.

In a future post, we’ll get into the mathematical modeling of this process with 3 parameters: Process Gain, Dead Time, and Time Constant.

(image courtesy of ExperTune, Inc. – http://www.expertune.com)