Sensible Sensor Speed–Part 1

May 13, 2009
Greg McMillan and Stan Weiner Talk Sensor Speeds
This article was printed in CONTROL's May 2009 edition.

Greg McMillan and Stan Weiner bring their wits and more than 66 years of process control experience to bear on your questions, comments, and problems. Write to them at [email protected].

By Greg McMillan and Stan Weiner, PE

Greg: It is best to have lots of field experience to provide the reality check and confidence to go beyond the rote duplication of designs. About 40 years ago, I was sent immediately out into the field after an intensive 12-week instrument engineering course to lead the construction and checkout for the automation systems of five rubber chemical plants. These were difficult batch processes with nasty chemicals. I was lucky to be learning so much so fast, but felt like the three-legged pig named “Lucky.” I graduated to helping rebuild, check out and start up the replacement of a reactor that blew up, and with the help of Stan to become the lead design and construction (D&C) engineer for project where the catalyst would explode if the bag was dropped, and the lead D&C again for a project for a Texas City plant with the world’s largest hydrogen cyanide production rate as a byproduct. 

Stan: We all learned the hard way what worked and didn’t work. When it comes to sensor speed, we got sensitized by compressor surge control, furnace pressure control, exothermic reactor temperature control and neutralizer pH control.

Greg: The fastest automation requirements were compressor surge because going into surge was like falling off a cliff. There can be a precipitous drop in flow in less than 0.03 seconds. Once into surge, there were flow reversals every 2 seconds, too fast for any feedback loop to deal with. You can get an overview of such super-fast applications in the “Analog Control Holdouts” entry on April 2, 2007, and the fastest upsets in the “Stuff That Comes at You Too Fast” entry on Jan. 22, 2007, on my site www.modelingandcontrol.com. To see the effect of transmitter speed on damping settings on surge cycles, check out Figure 7-3 on page 84 of the E-book Centrifugal and Axial Compressor Control posted on April 3, 2009, on the same website.

Stan: In the days of pneumatic controllers, transmitters and tubing, the measurement and control system could be as slow as the valve. For fast pressure control, we used field regulators. For compressor control, we used hydraulic controls and shut down if things were not right. When analog electronic controllers and transmitters came on the scene, we could get rid of the field regulators and hydraulics. The slowest components in the loop were the control valves that had pneumatic actuators and positioners. Unless you had a variable-speed drive, the inherent speed of the sensor and control system were not the limiting factors. When digital controllers came on the scene, some were designed to sample at 0.1 and 0.25 seconds. Users of big modern DCSs typically ran the execution time at once per second, which was fine for the flow and noncritical pressure loops. They didn’t realize they could go with 5 to 10 seconds for level, composition and temperature control. The downside of a faster-than- necessary DCS scan/execution time is controller loading and A/D noise, but with the processor speeds and A/D resolutions used today, these should no longer be issues. For wireless, we can go with a variable update time that is as fast as 1 second (see Greg’s "Is Wireless Process Control Ready for Prime Time?")I was ready to take a nap, confident in the future of process control until I heard a rumor that the proposed design for a next-generation control system doesn’t understand the importance of speed. The question in that case is, “Do you feel lucky?”

Greg: Let’s hope that’s just a bad joke and concentrate on making some sense out of sensor speed requirements. Sensor speed can be set in a lot of places. The sensor design may have some inherent lag time (time constant). Usually this is 0.2 seconds or less for flow, pressure and level sensors—fast enough for all but the most extreme applications. The default filter time setting in the analog input and PID blocks is 0 seconds. Signal filtering exists in the transmitter in terms of a damping adjustment. In some cases, the default filter time is 2 to 3 seconds. This high default setting, combined with reduced visibility and knowledge of a damping setting and the insidious nature of signal smoothing, has caused unexpected deterioration when the field measurements were upgraded.  A user describes the possible dramatic consequences.

User: In the past few years, I have twice seen major equipment accepted into service and then malfunction. One was the lube oil system of an FCCU wet gas compressor where the standby pump didn’t start quickly enough. The other was when two CNG gas export compressors had serious problems with the anti-surge system and load-sharing functions. In both cases, the problem came down to pressure transmitters that had been installed with factory defaults for the transmitter damping of a generous 2 seconds. My control and instrumentation colleagues were quick to point fingers in all other directions, as they had tuned the loops personally (or instructed their reports to do so) and could see no problems.

I have twice found damping left at factory presets to be the major or only cause of problems. None of the data relating to the instruments makes a big deal about this. Is this where the problem lies? I wish I knew where all the similar problems in the world were because I am sure I could be a very wealthy person based just on this knowledge alone. Based on my findings, the savings are probably in excess of $2 million per annum.

Greg: The total of the sensor, transmitter and DCS filter time lags should be less than 1/5 of the process dead time and process time constant in the loop if you want to avoid a degradation of more than 10% in the integrated absolute error from a disturbance. This rule of thumb assumes that about one-half of the time lag effectively becomes loop dead time as a result of time constants in series, and that the controller was properly tuned for disturbance rejection. For compressor and header pressure control, the process dead times and time constants are in the 1-second ball park, which means the transmitter damping setting should be about 0.2 seconds or less.

Stan: It used to be the rule to set the damping setting at a minimum (e.g., zero) and use the signal filter in the DCS to set the smoothing just enough to keep the excursions of the controller output from measurement noise within the resolution setting of the control valve (e.g., 0.2%).

Greg: With wireless measurements we have a new ball game. The transmitter damping setting will be used to keep measurement noise from exceeding the resolution setting and triggering an unnecessary transmission to the DCS. Presently, the fastest update is once per second, which is good enough for most flow and all level, temperature, composition, percent solids and pH loops. For compressor, furnace and header pressure control without variable-speed drives, 0.1 second wired transmitter update time and 0.1 DCS scan\execution time are needed to keep the measurement and control faster than the response of a control valve with a diaphragm or piston actuator.

Stan: Now Randy Reiss offers his “Top 10 Ways to Get Noticed at Work.”

Top 10 Ways to Get Noticed at Work

10. inny dipping at a company picnic. Note: Getting noticed is not always a good thing.
 9. Always be at the coffee machine to help the plant manager with his espresso.
8. Dedication, working extra hours and sacrificing your personal life to better the project…Ha! Who am I kidding? That never works.
7. Seak up at meetings with new ideas. The boss will notice your initiative. Everyone else will hate you.
6. Every time new ideas are mentioned say, “We tried that and it didn’t work!” Then suggest the same idea an hour later, but claim it as yours.
5. Pretend English is your second language, and everyone will think you are real smart when you stutter and spew incomprehensible gibberish.
4. Rewrite history: Half-heartedly support a project until you see that, through other people’s hard work, it’s going to succeed. Then jump in and declare that it works just like you kept telling people.
3. Act dumb. This could be easy.
2. Act smart. Emphasis on “act.” Real smarts make management nervous.
1. Figure out what the plant manager thinks and repeat it to him. It won’t increase production, but you’ll get noticed and, possibly, a raise.