How to Identify and Troubleshoot Control Valve Problems 'On the Fly'
The best time to determine control valve performance is on-line. Here are some simple techniques to determine control valve performance during normal operation.
Control valves with stiction problems often display a 'sawtooth' pattern when the controller output is trended.
I f instruments are the senses and controllers are the brains, then control valves are the muscles of the process industries. These workhorses are required to move quickly and precisely to desired positions, 24 hours a day, while requiring as little maintenance as possible.
Since control valves are mechanical devices, their performance is less than ideal, and degrades over time. For a plant to perform optimally, control valve performance must be tracked and maintained.
Typically, many valves are inspected or rebuilt during off-line outages. Off-line and bench-testing procedures can be infrequent, inadequate and costly. Testing the valves in-place, during normal operation, can reduce rebuild costs, improve process operation, and focus maintenance efforts.
Types of control valve problems
There are many ways for control valves to degrade process performance. Most common are stiction, hysteresis and backlash, and improper valve sizing.
Stiction -formed by combining the words stick and friction-refers to the extra effort required to get a valve moving from a dead stopped position. Stiction is the result of the sum of all the static friction in the moving parts of the valve. Contributing factors include packing gland torque, viscosity of process fluids, and plug and seat characteristics.
For more articles about control valves, loop tuning, and process analysis,
Stiction is frequently observed in control loops operating in automatic mode as a continuous cycling 'sawtooth' pattern of the controller output (CO) and a square wave pattern of the process variable (PV) that produce process upsets and unnecessary valve wear. (See Stiction diagram.)
Hysteresis and backlash in manual mode diagram
When trends of control valves with hysteresis and backlash problems are developed with the controller in manual mode, the quantity and magnitude of process variable changes isn't consistent with controller output changes.
If CO is moving more than 0.5% each cycle, there is a problem. It should be noted that, to adequately decipher control valve trends and patterns, it may be necessary to adjust PV and CO scaling, or use special products, such as those available from ExperTune (Hubertus, Wis.) or Techmation (Scottsdale, Ariz.).
Fixing stiction problems requires:
Ensuring the valve actuator and positioner are properly sized for the force required to move the valve;
Verifying that the air supply to the valve meets the valve manufacturer's recommendation; and
Checking the torque on the valve packing gland.
If these measures do not work, visually inspect the valve internals for signs of scaling, scarring or excessive wear and replace valve trim as needed. If using a viscous or sticky process fluid contributes to the problem, evaluate the location of the valve, its orientation, and whether a valve is less susceptible to stiction, such as a high-performance butterfly valve, might be appropriate.
Hysteresis refers to overall response and backlash refers to that portion of hysteresis caused by lost motion on valve and positioner mechanical parts. By far, the most common causes of hysteresis and backlash are loose or worn mechanical linkages between the positioner, actuator, and/or valve.
Hysteresis and backlash in automatic mode diagram
Trends of a control loop in automatic mode can provide evidence that hysteresis and backlash problems exist, but not the magnitude of the problem.
Effects of hysteresis and backlash are seen in process cycling around the setpoint, slower controller response, and/or a PV behavior different on 'up' movements of the controller output than on its 'down' moves.
Detecting hysteresis and backlash with the loop in manual (open-loop) mode requires introducing at least two small changes to the controller's output in each direction, and then observing the results. (See Hysteresis and Backlash in Manual diagram.)
Hysteresis of 1% or less is normal in a well performing control valve; hysteresis greater than 3% should be corrected as soon as practical.
Hysteresis and backlash will appear in control loops, with aggressive tuning and operating in automatic (closed-loop) mode, as an overshoot and cycling following setpoint changes and may display an ever-increasing cycling period. (See Hysteresis and Backlash in Automatic diagram.)
It is not possible to determine the amount of hysteresis with the loop in automatic mode, though it is possible to observe its existence.
Adding a positioner to a control valve can remove or minimize the impact of hysteresis. If a positioner exists, conduct a thorough physical inspection of sources of lost motion such as positioner-, actuator- and valve-linkages, and repair as needed.
Valve sizing problems diagram
Depending on the magnitude of valve sizing problems, trend results can vary from instability and imprecise control to a process variable not reaching its full range.
Valve sizing problems exist when the overall process gain is less than 0.3 or greater than 3. Valves are often sized based on a future maximum process design plus a 'safety factor.' This leads to specifying, buying, and maintaining more valve than is needed, and results in imprecise control and instability.
Undersized valves are often the result of inaccurate or 'guesswork' process specifications. This results in control loops that don't work well in automatic, as well as process bottlenecks.
Sizing problems can be detected by conducting a few output changes with the controller in manual mode, or by using setpoint changes in automatic mode. Make at least two process changes in each direction, and the larger the change made, the better. (See Valve Sizing Problem diagram.)
Using data from the last change, calculate the steady-state change of the PV. Steady-state change is simply the final value of the PV, minus the PV value before the bump. Convert this to percent of scale, using the engineering unit limits for the controller. Then divide the percent change in PV by the percent change in CO. If the number is greater than 3 %PV/%CO or less than 0.3 %PV/%CO, then there is a valve sizing problem or a transmitter-scaling problem.
Depending on the type of valve, correcting the problem requires changing the entire valve or changing the valve trim.
Over 30% of control valves have problems, many of which can be detected during normal operation using simple tests. Correcting valve problems can significantly improve bottom-line results.
George Buckbee is a registered professional engineer, trainer, and consultant on process control and loop tuning problems. Mr. Buckbee has a Master's Degree in Chemical Engineering from the University of California at Santa Barbara and 14 years industrial experience in the application of process control.
For more suppliers, go to www.controleng.com/buyersguide. For more info, go online at www.controleng.com/freeinfo ; ExperTune at www.expertune.com , Techmation at www.protuner.com and TopControl at www.topcontrol.com .