Upsetting experiences

Control Talk columnists Greg McMillan and Stan Weiner explain why the differential pressure measurement was straight and proffer a new Puzzler: Why did the drum-level control cause more trips?

By Greg McMillan and Stan Weiner, PE

Stan: The puzzler for August that asked: "Why a slurry tank level differential pressure measurement was a straight line,” did not exactly clog up our e-mail inbox, but we did get a couple of solid answers from Hunter Vegas and Meredith Winn.

Hunter: There are a lot of possibilities, but the first that comes to mind is a plugged sensing line on the high side.

Meredith: The nozzle is plugged up with solids in front of differential pressure cell or the tank is running over.

Greg: These guys must have read our mind, which is not advisable, or should at least come with a parental warning of some sort.

Stan: The only straight lines I’ve seen were in simulation results, which are notorious for being devoid of dynamics and noise. Kind of like life in a Sun City retirement village.

Greg: If you didn’t have upsets, it would be sooo boring and there would be little justification for regulatory control.

Stan: If an output drew a straight line it meant the controller was in manual.

Greg: The most common load upset is a feed upset, Even though engineers may claim the feed rate is constant, operators have to deal with equipment constraints and inadequate surge tank volumes. Each operator tends to have a “sweet spot” for the process that shows up as a feed change at a shift change. Most of our plants were running well beyond the original nameplate rating, and capital improvements to cooling towers and tanks were not included in the de-bottlenecking (Editor’note: This is technical term understood by only the most highly educated control engineers) projects because of a lack of understanding of dynamics. Sitting with the operators for a few shifts can reveal these very real limitations.

Stan: Processes with a change in phase are generally the most difficult to run. Bubbles can cause foam, shrink and swell, and erratic measurements. Solids can result in the coating and plugging equipment, sensors, and valves.

Gerald’s Top 10 Signs You’re a High-Tech Redneck

10. E-Mail address ends in “”
9. Your Laptop has a sticker saying “protected by Smith & Wesson.”
8. Value of your truck doubles by installing a cell phone.
7. Your baseball cap reads “DELL” instead of “CAT.”
6. Your computer is worth more than all your cars combined.
5. You use a CD-ROM as a coaster for your beer can.
4. Your screensaver is a bitmap image of your favorite truck, tractor or farm animal.
3. Your wife said “Either that computer goes or I go!”
2. You still do not miss her.
1. You have several computers sitting on cinder blocks in your yard.
Greg: Often loops end up in manual in these difficult to run processes, which in turn increases severity of disturbances in several ways. First, these disabled control loops do nothing to transfer variability from the process variables to the manipulated flows. Second, corrective actions made by operators are larger, more abrupt, sporadic and usually without anticipating the slowness of the processes’ response from lags and delays. Control loops can make gradual and continuous corrections that attenuate and smooth out disturbances.

Stan: Sometimes, process engineers unfamiliar with the advantages of setting up a basic loop, will specify “process actions” that step the position of a control valve, opens or closes a block valve, or turns on and off a motor when a process variable exceeds a limit. Step changes are nice for control studies but are very disruptive to processes. Some classic examples of this misguided attempt to save a few bucks is the use of high and low-level switches to turn pumps on and off or high and low temperature switches to turn cooling tower fans on and off. Properly tuned loops that throttle control valves or manipulate motor speeds are a powerful but simple tool to eliminate the source and propagation of upsets.

Greg: The control system needs to maintain the same ratio of manipulated flow to feed flow as is shown in the process flow diagram or simulation model, whether the controller output is regulating an air, additive, coolant, distillate, reagent, reactant, reflux, or steam flow. Reproducible measurements of these flows are immensely valuable for flow feed forward or ratio control. Even if the feedforward gain (ratio) or timing is off by 10%, there is still a 10:1 reduction in error by pre-emptive action. In order to deal with drift, dynamics, noise, nonlinearities, and offsets, the feedforward signal is almost always corrected by a feed back controller output.

Stan: Most control text books and studies show an upset entering into the output from the process whereas feed and other load upsets in plants are inputs to the process. If they arrive basically at the same point in the process equipment and the flow controllers are tuned properly, timing may not be an issue and dynamic compensation of the feedforward signal an unnecessary complication. This is commonly the case for properly designed blenders, boilers, desuperheaters, furnaces, heat exchangers, neutralizers, and reactors. For columns, the timing depends upon the location of the trays for temperature control and feed relative to the top or bottom of the column.

Greg: The worst scenario is when the feedforward signal arrives too soon because it causes inverse response. Fortunately, this can be corrected by delaying the feedforward signal. If a signal arrives too late, a lead applied to the feedforward signal can help compensate for a lag or to kick a valve with stick-slip. If a lead time is used, a lag or filter time of at least 1/8 the lead time helps avoid amplification of noise. More often, a feedforward signal arrives too late because of a transportation delay. This occurs in pH control because process and mechanical engineers do not realize the horrendous transportation delay in piping and dip tubes possible from extremely small reagent flows. Nothing can be done to compensate for this dead time in the corrective action.

Stan: Exceptions to feedforward rules exist, which leads us to this month puzzler. Why did the addition of three-element drum level control where a preheated feed water flow was ratioed to the steam flow cause more high and low level trips of a boiler pushed beyond its nameplate rating?

Greg: Now for some levity via a list by Gerald Mitchell from my ISA book titled Dispersing Heat Through Conviction.

Stan: Now for this month’s disclaimer. The authors have been disturbed since the 1960s.

 This Month’s Puzzler:
Why Did the Drum-Level Control Cause More Trips?

Why did the addition of three-element drum level control where a preheated feed water flow was ratioed to the steam flow, cause more high and low level trips of a boiler pushed beyond its nameplate rating? Send an e-mail with your answer, other questions or comments to

Greg McMillan and Stan Weiner, PE, bring their wits and more than 66 years of process control experience to bear on your questions, comments, and problems.
Show Comments
Hide Comments

Join the discussion

We welcome your thoughtful comments.
All comments will display your user name.

Want to participate in the discussion?

Register for free

Log in for complete access.


No one has commented on this page yet.

RSS feed for comments on this page | RSS feed for all comments