Real World Engineering

This is a blog from the trenches—written by engineers at Maverick Technologies who are implementing and upgrading control systems every day across every industry. This isn’t what they teach you in engineering school. These are lessons learned from years on the job, encountering the obstacles and issues that are part of the real world of control and process engineering.

Real World Engineering

Science, it just works

Some operational myths seem to live on even in the face of evidence to the contrary. How many of them are changing the way you work?

November 18, 2013


The actual quote is a bit harsher but the sentiment is the same: Science works and mythology doesn’t. Yet in my experience there is still a lot of mythology and misunderstanding at work in the application of process control that could be dispelled with proper scientific and engineering analysis, though sometimes the people involved just don’t understand. A case in point: I once had a boiler operator ask me when I was going to fix the electronic indicator on his control panel to match the reading on the drum level gauge glass. I told him that the gage glass was installed too low and that the electronic indicator was right. He replied that a gage glass couldn’t be wrong and proceeded to repeat his request every day. Finally, one morning I walked in and he thanked me for fixing the indicator. I didn’t have the heart to tell him that the pipe fitters had finally gotten around to reconnecting the gage glass correctly.

Other common pronouncements you might hear are:

• “That might work other places but it won’t work here”
• “We tried that but it doesn’t work on our process”
• “Our operators could never learn to use that”
• “We’ve tried tuning that loop but nothing works,” and
• “Remember, I can make paper in manual.”

These kinds of comments always have me asking, “Why?”

Responses often start with, “Well we tried that but…” The response implies that a proper system analysis was done prior to trying whatever it was, but often the reality does not back this up. Of course the one thing you really can’t do anything about is an overly optimistic sales person who doesn’t understand the limitations of his or her equipment, like trying to use a magnetic flow meter to measure deionized water. Oh, and you’ll never set a production record trying to run your process in manual.

When I’m at a customer’s site I’ll often take a quick survey to see how many loops that are in manual mode do not have stable measurements. If the measurement isn’t stable when the loop is in manual, you will not be able to tune it into stability. An unstable measurement can have many causes from bad sensor location, poorly operating equipment, a manifestation of odd piping geometry, thermodynamic phenomenon like shrink/swell, or liquid flow phenomenon like the Coanda effect.

Some people will try and make the instability go away by using a filter, assuming that it’s just normal process noise. But without the use of something more sophisticated than a first order filter, like a Kalman filter, the results will be less than acceptable. These effects may be exacerbated by over or under sized equipment, or may be the result of design decisions made to work around limitations in the process piping. Regardless of the cause, the loop cannot be tuned successfully. For loops like these, step one is to find the source of the instability, then determine why it’s unstable and fix it. Very few processes are inherently unstable and even those that are can usually be made to be stable or at least controllable. Of course you could take the approach one operator did of putting a toothpick in a chart recorder to keep the pen from moving around during his shift (except we don’t use chart recorders any more).

Another sign of process issues are loops that have unresponsive tuning. The myth is that if it’s tuned more aggressively, it will overshoot too much.  There can be several reasons to de-tune a loop, but the most typical are a badly chosen or oversized control element, or a non-linear response in the process. The non-linear process response can usually be addressed by using adaptive gain, as is often applied to pH control, but that can be a challenge to tune properly.

The oversized or badly chosen control element can actually be a compound issue. If you’ve specified an equal percentage trim but you’ve put in too large a fudge factor, you’ll end up with a linear action from the valve. Worse, if you’ve specified a linear trim but oversized it, you’ll end up with a quick opening action. Early in my career I discovered that the mechanical department would put a “safety factor” on their systems to ensure that the plants would meet their design basis, plus we knew that our customers typically tried to run their plants at 110% of design. Until I found out what production level they really wanted, I would routinely add 10% to the maximum flow rate when I was sizing a valve compounding the over sizing of the process equipment. At startup we’d often discover that valves were running less than half-open at rated operating conditions. A few that had also been selected with quick opening trip were running very near their seats causing other issues. The right fix is to replace the trim with reduced trim if it is available or replace the valve if reduced trim isn’t available. In the mean time, the loop should stay de-tuned.

So what myths have you debunked at your plant? How do you help your operators make sense of what they see on the screen when it runs counter to what they think it should be doing?

This post was written by Bruce Brandt. Bruce is a technology leader at MAVERICK Technologies, a leading automation solutions provider offering industrial automation, strategic manufacturing, and enterprise integration services for the process industries. MAVERICK delivers expertise and consulting in a wide variety of areas including industrial automation controls, distributed control systems, manufacturing execution systems, operational strategy, business process optimization and more. 



No entries

Nothing found in the guestbook.