Why You Should Not Use On-Off Valves as Control Valves

The following tip is from the ISA book by Greg McMillan and Hunter Vegas titled 101 Tips for a Successful Automation Career, inspired by the ISA Mentor Program. This is Tip #83.

101 Tips for a Successful Automation CareerPrior to the 1970s, you were not in much danger of getting a control valve that would not respond correctly. You simply went to your control valve suppliers and worked with them to get the best valve. In the 1970s, piping valve manufacturers decided they could sell their rotary valves as control valves by slapping on a piston actuator and a spool positioner. Because the piping valves were cheaper, already in the piping spec, and had less leakage, many users, particularly process engineers, were fooled into buying these valves. The problem escalated to the point of absurdity. Most of the oscillations in loops in the 1980s and 1990s were due to the use of piping valves designed for on-off service as control valves.

How did this error perpetuate itself? The valve specifications of the time had an entry for leakage class but no entry for the smallest change in signal that the valve must respond to. The valves were called “High Performance Valves” because they were exceptional at meeting piping specification and leakage class requirements. Who would not want to buy a “High Performance Valve” at a great price? Even the parent companies of control valve suppliers saw the opportunity and proceeded to buy the piping valve companies, so if you go to your valve supplier for a control valve now, you may be offered a suite of former piping valves.

There was no readback of actual position in pneumatic positioners, and the feedback of actuator shaft position did not show if the internal flow element (e.g., ball, butterfly, or plug) actually moved. Typical signal changes used to test a control valve were 20% or larger. For these large changes in signal, nearly all valves will respond. The ISA standard on valve response testing, ANSI/ISA-75.25.01-2000, provides a wakeup call (Tip #81), but today the message is still not clear to the average user. The absence of entries for deadband or stick-slip in a control valve specification essentially means there is no requirement that a control valve actually moves in response to a signal.

In the 1990s, I had several memorable experiences where having smart positioners fooled plant personnel into thinking there would be no control problems with “High Performance Valves.” The digital positioners for ball valves in phosphorous streams, for example, all indicated that the valves were pretty good in responding to signal changes as small as 0.5%, but the process control was horrible for these loops. Testing the valves in the shop, I found that the internal ball did not move until the change was 10% or larger. The actuator shaft responded to small changes but motion was lost in the pin connections of the actuator shaft to the ball stem and in the pin connections of the ball stem to the ball. The valves were great for reliable on-off stroking and isolation, but not for control. A similar problem occurred with large high-performance butterfly valves, again selected by virtue of a plant piping spec. We had wonderful trend charts from the digital positioners saying the valves were quite responsive. Only after I put a travel gauge on a butterfly disc did we realize that the disc was not moving.

Concept: Despite digital positioner tuning and diagnostics, on-off valves are not throttling valves because of high friction in sealing surfaces for tight shutoff, pinned connections resulting in backlash, and shaft windup. The type of piston actuators and positioners often used add to the problem.

Details: Use on-off valves to meet leakage specifications for isolation. Use control valves for throttling, from companies whose heritage is process control and not piping. Specify a % deadband and % stick-slip (resolution or threshold sensitivity) for a specified throttle range. Put the tight shutoff leakage class on the on-off valve specification and not on the control valve specification. Use v-notch ball valves and toothed butterfly valves for better installed flow characteristics. Check the catalog table of flow coefficients at low and high positions to make sure a valve has a suitable inherent flow characteristic. Plot the installed characteristic for a given ratio of valve drop to system drop at operating conditions. Try to get a change in valve gain (slope of the installed characteristic) that is less than 4:1 over the operating range. Consult the checklist in Appendix C to make sure the whole valve package will function well as a control valve and will not add variability to the loop. Test the control valve in the shop, making sure the internal flow element is actually moving per the positioner readback for small signal changes. Repeat the tests when the valve is in service with a sensitive low noise flow measurement.

Watch-outs: If you need to explain valve backlash, deadband, stick-slip, or resolution to the valve supplier, the valve may have originally been a piping valve. Piping and on-off valves are often sold as control valves by control valve suppliers. A supplier may not even offer a real control valve and may not know the difference. If you go out for bids on control valves, the lowest bid will probably be valves originally designed for on-off service. Valves supplied with packaged equipment will usually be the least expensive valves and will have exceptionally poor response. “High Performance Valves” are not high performance in terms of throttling. Operators and maintenance technicians are justifiably concerned about leakage, and you need to make sure you specify on-off valves that will address these concerns. The control valves with the best throttling capability have less friction and consequently more leakage. Not all v-notch ball valves have good inherent flow characteristics. For example, one v-notch ball from what I think was a piping valve manufacturer had a flow coefficient of 0% for 0-15% signal. When asked for response test results, most control valve manufacturers will do the tests at 50% output where the friction is low and the flow characteristic is the best. Valves can get stuck due to build-up of solids, corrosion, and expansion of final flow elements at high temperatures. Periodic pulsing of the valve can keep a valve free to move.

Exceptions: For pulse width control to prevent plugging problems, an on-off valve might be better than a control valve because the on-off valve is designed for fast, large strokes.

Insight: On-off valves should not be used as control valves (for throttling) and control valves should not be used as on-off valves (for isolation).

Rule of Thumb: For throttling, use control valves supplied by a company whose heritage is process control and not piping.

Print Friendly

, , , ,

Pin It on Pinterest

Share This

Share This

Share this ISA post with your friends!