mcweiner
mcweiner
mcweiner
mcweiner
mcweiner

Year End Puzzler Bonanza

Dec. 10, 2007
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. Write to them at [email protected].
Greg: We have been saving up our puzzler answers from way back yonder to provide a year end bonanza that is a hoot and holler (Texas talk).

Stan: People are probably waiting with baited breath (Fish talk).

Greg: Hunter Vegas got to the heart of the matter in the July puzzler “Are the dead times identified from a step change in the controller output and set point identical?”

Hunter: A step change in the output of a controller will provide an indication of the combined dead time of the valve, process and transmitter. This value can then be used to help calculate the proper integral and derivate tuning parameters. If a step change is made in the set point, then the resulting dead time will be a combination of all of the items above, plus whatever dead time the poorly tuned controller might add.  The resulting dead time from that test could be used to generate still worse tuning constants!

Greg: The dead time from backlash (e.g., deadband) and resolution limits (e.g., stick-slip) for a set point change is the deadband or resolution divided by the rate of change of the controller output. The dead time is zero for a step change in the controller output. If the controller has sluggish tuning, the dead time will be way too large for a set point change. If this tuning is used and the set point test done again, the dead increases again. The real additional valve dead time (τdv) from deadband (DB) or resolution limits (DB/2) can be estimated from the controller output (open loop) test dead time (τdo), the Lambda factor (λf) and the change in controller output for a typical disturbance (∆CO), per the following Equation 2-50 in the ISA book, Advanced Control Unleashed. This equation prevents the downward spiral in settings while making the tuning more commensurate with the valve response capability. To positively insure the controller does not out run a big and sloppy valve or damper, you can use a read-back signal of actual position as the external reset signal and enable dynamic reset limiting for the PID so reset action is not faster than the valve. Of course it is better to make the valve more responsive and responsible for its own actions.

Stan: We got a succinct answer from Robert Reiter to our August puzzler “How many pipe diameters downstream of desuperheaters should a temperature sensor be mounted?”

Robert: Temperature measurement should be located at least 20 pipe diameters downstream of any attemperator. However, a Sarco installation drawing says 30 feet. It needs to be far enough away for all the water to boil, but close for faster response.

Greg: A straight run of at least 10 feet between the outlet of the attemperator and the first elbow before the temperature sensor is required to prevent damage from the impact of water droplets.
Meanwhile, Walter Farr had a good answer to our September puzzler question, “When should you not use a valve positioner?”

Walter: Never use a valve positioner as a characterizer for a valve that has no character. Use a valve with character.

Greg: It is better to address the source of the problem rather than trying to treat the symptoms. However, if I were implementing signal characterization to compensate for the flattening of an installed valve characteristic that I was stuck with at high flows, I would do it in the distributed control system rather than in the valve for greater visibility and coordination.

Stan: In the days of pneumatic positioners, the calibration changed and required special skills and too much time to keep in good shape. Plus, analog controllers were faster than these pneumatic positioners on big valves and fast loops, violating the cascade rule requiring the secondary loop (valve positioner) to be five times faster than the primary loop.

Greg: With distributed control systems and digital valve controllers, these issues of tuning and calibration are largely gone. In my book, the diagnostics and read back of actual position justify the general use of smart digital positioners. The biggest mistakes I have made in my career mostly centered around attempts to save money by cheapening the automation system even though these costs were in the noise band of the project and would ultimately come back to haunt me. On the other hand, without valve position read-back, maybe operations will never track down the source of variability to the valve.  Hopefully they haven’t heard that limit cycles from stick-slip can be 10 times larger for valves without positioners, or read the article “Improve Control Loop Performance” in the November issue of Chemical Processing.

Stan: Any answers probably got lost in the spam to the October puzzler, “When does a feed-forward control system do more harm than good?”

Greg: If the feed-forward signal arrives much too soon, it can cause inverse response, confusing the controller, and if it is way late, it can cause a second disturbance, unsettling the controller. Then there’s the question of unmeasured disturbances driving the process back to set point, and feed-forward action oblivious to the trajectory would undo the approach to the set point. For more details check out my September blogs on Feedforward Techniques in the continuous control category at http://ModelingandControl.com.

Stan: We conclude with selections from one of our all-time favorite lists, “Believe it or Don’t,” from our book, Dissipating Heat Through Conviction.

Believe It or Don’t
  • A young engineer told the CEO that there were too many presidents. The CEO thanked him profusely and immediately transferred the presidents to the plants to configure DCSs.
  • A project manager found a major design flaw in the control system and immediately authorized the purchase of the instruments necessary to solve the problem.
  • After major problems meeting product quality specs, a process engineer was heard to say, “The instruments are right, therefore, it must be the process that is screwed up.”
  • A vendor refused a purchase order and recommended the purchase of transmitters from a competitor because the vendor’s meters had an excessively high failure rate.
  • A reagent control valve for low viscous flow was sized according to laminar flow.
  • A DCS’s computational power was fully utilized to do advanced control strategies.
  • A book was published that clearly explained how to install all types of valves and sensors.
  • A plant standardized on the type of instruments supplied with packaged equipment.
  • A control valve spec actually stated that the valve must respond.
  • A chemical plant was designed, checked out and started up without any changes.
  • An engineer was given raises and promotions throughout his career based on the performance of his applications.
  • Mocha banana milkshakes are the missing link in the unified force theory.
  • This column’s authors are sane.

This Month’s Puzzler:

What are the biggest opportunities in process control improvement?

Send an e-mail with your answer to the Puzzler, CONTROL questions, or comments to [email protected].