I was encouraged to omit positioners in one of my first projects in the 1970s as lead instrument engineer. A former power system maintenance technician taking a new career path in instrument maintenance decided that positioners would add unnecessary complexity and aggravation. Fortunately, the request that valves not have positioners occurred after the project was completed. In the days of pneumatic positioners, adjusting the links and pneumatic relays required special skills. Positioners got out of calibration within six months and the lack of position readback made maintenance more difficult and the value of the positioner more questionable.
The lead instrument engineer of the system integrator for my next project proposed various ways to reduce project costs. While he had good intentions and was very experienced, all of his suggestions that were implemented created problems that had to be corrected during commissioning and after start-up. One suggestion was that we should put positioners only on selected control valves based on a frequency response study from a noted technologist that showed that boosters should be used instead of positioners on fast loops. However, I found that many rotary valves did not open until the current to pneumatic (I/P) transducer output was 30 percent. During startup, we put positioners on most of the valves.
When I was promoted into Engineering Technology, my areas of expertise came to include exceptionally fast loops such as compressor surge control and phosphorous furnace pressure control. These loops were so fast that they required analog or exceptionally fast digital PID execution times. On particularly large compressors, I recommended that the 24-inch butterfly surge valves have their positioners replaced with boosters (as indicated by the frequency response study mentioned above). The surge valves inexplicably slammed shut. When I went to the field the next day, the technician showed me how he could manually move a big butterfly that had a booster by just grabbing the stem, whereas the butterfly with a positioner could not be moved by hand. I realized that putting a booster on an actuator without a positioner could create positive feedback, where the booster would assist to continue movement.
The booster had assisted some turbulence-induced movement to the closed position. I subsequently dug up a 1958 article (Reference 34) which showed that you should not eliminate the positioner and should put the booster on the output of the positioner with a bypass valve slightly open. A few years later, when I had specified a fast stroking time for furnace pressure control, the supplier put boosters without positioners on the big butterfly valves. In a test at the factory, I showed how I could manually move the big valves and showed them the article. The boosters were then put on the positioner output and the booster bypass valves were tuned for a fast, stable response.
Concept: Today’s smart digital positioners hold their calibration and offer diagnostics and flexible tuning. The speed of response of a valve can be increased by putting a booster on the positioner output and tuning the bypass valve and positioner response settings. However, the tuning rules are not extensively documented. Field support and training schools by the supplier currently provide the required expertise to get the most out of a positioner. External-reset feedback and fast readback can prevent a process PID controller from changing faster than the valve can respond due to size or tuning.
Details: The threshold sensitivity is about 0.1% for digital positioners, 0.2% for two stage high gain relay pneumatic positioners, and 2% for spool type pneumatic positioners. Use digital positioners. Tune a positioner on the valve before installation. After installation at operating temperature and pressure, tune the positioner again. Avoid integral action. If integral action must be used, set the integral deadband in the positioner to stop limit cycles. If the valve must be speeded up, install a booster with a bypass on each positioner output that goes to the actuator.
Double-acting piston actuators require two outputs and hence two boosters. Use a booster with a deadband for piston actuators. For all types of actuators, open the booster bypass valve just enough to prevent rapid cycling of the valve (e.g., 1 cycle per second). Specify a pre-stroke deadtime and slewing rate for large valves. For fast loops, test the positioner tuning to ensure that the valve response is faster than the process controller response. If the control valve response cannot be made fast enough, add an exceptionally fast position readback with external-reset feedback per Tip #85 or detune the process controller to prevent a burst of oscillations from the process controller output changing faster than the control valve can respond in the event of large setpoint changes or big disturbances.
Watch-outs: Poor actuator or valve design or sizing can limit what a positioner can do (Tip #81). Positioner readback is on actuator shaft position and not internal flow element (e.g., ball, butterfly, or rotary plug) position. For on-off valves, the difference in position between the internal flow element and the actuator shaft can be as large as 8%, leading to a deadband or stick-slip of 8% from backlash and shaft windup (Tip #83). For on-off valves showing deadband and stick-slip, diagnostics and trend charts within the positioner may indicate that the deadband and stick-slip are an order of magnitude less than actual. Even the smartest and most advanced positioner will give erroneous performance results under such conditions. Thus, putting a great positioner on a control valve may not solve all response problems.
Exceptions: If a PID execution time is smaller than 0.1 seconds or an analog controller is required for good control, a variable speed drive with no deadband or rate limiting in the drive setup is needed. Loops that are too fast for digital PID and hence for control valves (analog control holdouts) are noted in the Control Talk blog What is the Best PID Execution Time?
Insight: Readback control of valve position compensates for a whole host of unknowns in connection with valve response but a positioner, like a controller, must be tuned.
Rule of Thumb: Use tuned positioners on all control valves and add boosters with tuned bypass valves on positioner outputs as necessary to speed up valve response.