Ungrateful Dead Band

March 13, 2003
Our new column has a little fun with process engineering. A sticky situation made worse, 10 reasons not to retire, and a new puzzler: Dumb DP Transmitter

Greg McMillan ([email protected]) and Stan Weiner, PE, bring their wit and more than 66 years of process control experience to bear on your questions, comments, and problems.

Greg: We had a good analysis of the reluctant valve from Jack Funderburk, Harley Jeffrey, and Michael McFadden. In this January puzzler, the control valve with linear trim didnt open until controller output was greater than 40% after equal percentage characterization was added.

Harley: If there was no dead band, it wouldn't really matter where this linearizing process occurred. But with dead band in the valve assembly, it makes a lot of difference. Assume that a load disturbance has occurred, which has resulted in a process variable error that normally requires a 1% change in controller output. Due to the characterization effect, the characterization may rescale this 1% signal down to as low as one-eighth percent or less. This means that the valve assembly must now attempt to respond to a very small signal. Assume that the positioner actuator has a dead band of 2% or more, which is not unusual for butterfly valve assemblies. This means that the process variable would have to get 16 times larger before the valve would respond and make a flow correction. If the linearization was in the valve trim, however, the controller output would only have to double in order to get through the dead band. As you can see, dead band detrimentally affects process variability, but putting the linearization ahead of the dead band can make the situation even worse.

Jack: When the controller's output is 40%, the equal-percentage-characterized positioner's output is a little over 10%. The problem is apparently exacerbated by excessive valve packing friction. Since the valves are new, I assume that the stems aren't galled or rusty. That leaves improperly adjusted or specified packing as the likely source of the stiction. There are a number of companies that insist on graphite packing for all control valves (regardless of service), so this problem isn't uncommon.

Michael: An equal percentage trim on a valve would provide about 10-12% maximum flow at 40% open, compared to the flow at 100% open at the same differential pressure. Whereas a linear trim would provide 40% maximum flow at 40% open. By choosing an equal percentage positioner, this engineer has, in effect, moved the characteristic of the equal percentage curve from the valve to the positioner. Now, the valve opening will be about 10-12% with a controller output of 40%. Couple this with any stiction or looseness in the valve and you have the valve just cracking open when the controller output reaches 40%.

Stan: If the signal characterization makes the change in signal smaller, the valve can get hung up from backlash and stiction. If the signal characterization is set up just for the flat portion of an installed characteristic so that the change in signal is larger, the valve can zip through its dead band and resolution limit. Rotary valves often have backlash from gaps in linkages. Also, the friction of high temperature or tightened packing and the seating friction of many tight shutoff valves cause excessive stick-slip, particularly near the closed position.

Greg: Valves are not perfect and neither are engineers, so we will take this opportunity to make the following apologies.

Stan: I apologize for giving problems to summer interns that havent been solved in decades.

Greg: I apologize for wearing plaid bell-bottom pants in the '60s.

Stan: I apologize to anyone who saw me dance in the '60s.

Greg: I apologize to all the CEOs who were the butt of my jokes in my top 10 lists.

Stan: I apologize for being the longest successive winner of the agitator of the month award.

Greg: I apologize for island hopping while my former associates are glued to their computer screens. Just in case they are envious, here are the top 10 reasons why they shouldnt retire:

Top 10 Reasons Not to Retire

10. Vintage laptops for those quiet moments waiting for a model to run.

9. Vintage operating systems to save time learning new features.

8. Exotic food and cheap prices--or is it cheap food and exotic prices?--by the new contractor for lunch services.

7. Learning to "just say no" when faced with unreasonable schedules and budgets.

6. Unlimited opportunities to earn frequent flyer awards and experience firsthand the new airport security systems. Just remember not to eat a powdered doughnut while in line or have instruments or dirty underwear in your carry-on.

5. Chance to "guess what floor" your office will be on next year. Is there a basement or utility closet big enough for engineering?

4. Great stock options and valuable company-matched funds for 401(k) plans.

3. Biannual package lottery. You might be a big winner and not a big loser like me.

2. Creative deals by your sole source of software. Does a sole supplier have no soul?

1. Cozy cubicles.

Stan: The work place has changed. Nowadays there are two types of engineers, male and female. These engineers want to be referred to by the proper pronoun. Males prefer "he" and females prefer "she" but I noticed both will respond to "hey you."

Greg: Thank you for that astute observation. This column thrives on such hard-hitting, up-to-date perspectives. How about the puzzler to further expand our consciousness?

Stan: Long ago (at least 10 years), a dumb DP transmitter managed to provide tight vent condenser pressure control but a trend of the controller output showed a large 24-hour cycle. Why was the improvement like night and day when it was replaced with a smart transmitter?

Greg: Now for this months disclaimer: "In the 1960s, the authors thought the lyrics of the Moody Blues had profound messages, but they never had much of an impact on our performance reviews."

This Months Puzzler:

Dumb DP Transmitter

A conventional DP transmitter managed to provide tight vent condenser pressure control, but a trend of the controller output showed a large 24-hour cycle. Why was the improvement like night and day when it was replaced with a smart transmitter?