In an industrial setting such as a chemical plant, oil refinery or a deep-sea oil rig, there are safety instrumented systems in place that are designed to promptly bring about a safe shutdown of a process if certain hazardous conditions are detected. Such safe shutdowns frequently involve isolating a process fluid flow, often accomplished using an ESD (emergency shutdown) valve. If the ESD fails dangerously, when a hazardous condition necessitating a shutdown occurs and the ESD does not perform its function when required, the worst-case consequences can be catastrophic.
ESD valves often include a solenoid valve in the ESD design with the solenoid valve energized in an open position while the process operates normally and with the solenoid valve moving to the closed position to initiate a safe shutdown of process fluid flow. Solenoid valves can fail (be unable to close on command) due to several different conditions. However, the failure mode contributing most significantly to the dangerous failure rate of the solenoid valve is that of sticking or adhesion.
Due to the fact that the solenoid valve, as part of an ESD, is often “at rest,” i.e., stationary, during normal plant operation, the fact that it has become “stuck” in one position cannot be observed. In other words, the dangerous failure is undetectable in normal operation.
What is Stiction?
Many studies and organizations have defined stiction, i.e., static friction, in different ways. Yet, all agree that stiction is the act of being “stuck” by static friction, which prevents one surface from moving against another. Further, if the external force becomes greater than the static friction, the stiction between two surfaces will be overcome and the object will begin to move again.
Stiction in a Solenoid Valve
In normal operation, the O-rings in the solenoid valve maintain a seal even while the plunger or spool is in motion. Because the O-rings are in direct contact with the valve chamber walls, for the plunger/spool to begin to move, it must first overcome the stiction between the O-rings and the chamber walls.
Once in motion, the plunger/spool must overcome the sliding or dynamic friction between the O-rings and walls but, since the dynamic friction is usually significantly less than the nominal level of stiction, it is generally not of concern. By design, the magnetic and spring forces in the solenoid valve are sufficient to overcome the nominal level of static friction and, depending on the design specifics of the solenoid valve, generally have excess force sufficient to overcome about 2 to 2.5 times nominal stiction. Once the stiction is overcome, the plunger/spool will continue to move.
However, a common and troublesome problem can occur when the valve is at rest for a length of time because the stiction between the O-ring and the valve chamber walls increases over time from its nominal level until it reaches some maximum value. It is possible that the increases in stiction may reach a level where the forces generated by the solenoid coil and/or spring are no longer sufficient to overcome the increased stiction. Consequently, when the valve is called upon to close, it is unable to do so. However, if the excessive stiction is overcome, stiction reverts to its original nominal level though it does begin to increase again once the valve stops moving.
Evidence of Stiction
Expert Knowledge: Mechanical engineers and technicians who routinely work with solenoid valves are familiar with the experience of trying to “stroke”, i.e., move through its range of motion, a valve that has been stationary for an extended time. They report that it is not uncommon for extra force to be required to move the valve after it has been stationary for a month or more.
In the one manufacturer’s O-ring handbook, it is noted that the “coefficient of starting friction” (stiction), increases when the O-ring has been stationary for between 1 week and 1 month after which time the stiction plateaus. One specific graph of “break out friction,” (stiction) vs. “delay between cycles” shows the stiction force plateauing at approximately 300 hours. The author of the handbook states: “The theory has been proposed and generally accepted that the increase in friction on standing is caused by the rubber O-ring flowing into the micro-fine grooves or surface irregularities of the mating part.”
ISO 13849: In a declaration regarding the reliability indicators and information for use of a specific solenoid-actuated pneumatic valve with respect to the safety standard EN ISO 13849-1, the text reads “valve must be operated at least once per week or once per shift to insure the intended function.” The data supporting this declaration was obtained by cycle testing during which the solenoid valve was never at rest for any significant time. This statement recognizes that the results of cycle testing are valid only if periodic valve movement is maintained. In the absence of such periodic movement, the failure rates derived from the results of cycle testing cannot be considered valid when applied to components in applications where these components spend significant times at rest.
Solenoid Valve Stroke Testing
In a stroke test, the solenoid coil is de-energized for a short period of time to see if the spring can fully decompress moving the spool in one direction through its full range of motion. Then re-energized, the spool must again move. At the completion of the valve stroke test, either the solenoid valve has been verified to be operable or a dangerous failure due to stiction has been detected.
While valve stroke testing will have beneficial results on Average Probability of Failure on Demand (PFDavg),) even if performed infrequently, ideally we would like not simply to detect failures due to stiction but also to prevent them. So, is stroke testing once per week sufficiently frequent to ensure that stiction buildup will be disrupted before it exceeds the valve’s available excess force? Unlike random failures truly represented by a constant failure rate, failures due to stiction are unlikely to occur very early in the time interval between tests when stiction levels have not increased very much and more likely to occur once stiction levels have increased beyond the level that the excess force of the valve can overcome. Thus, while we know that, as a rule of thumb, the maximum level of stiction is reached at about 275-300 hours, we also need to consider how fast the stiction builds towards its maximum value in order to determine the most appropriate interval for valve stroke testing.
Alternatives to Valve Stroke Testing?
In some end-user circles, there persists the belief that valve stroke testing risks a plant shutdown due to a false trip and that a better safety solution is to provide a much higher safety margin (greater excess force) to overcome increases in stiction. However, those who are well versed in solenoid design realize that a higher safety margin to overcome stiction actually results in a greater false trip rate. This occurs because a higher safety margin to overcome increases in stiction requires a bigger spring, which, in turn, requires more energy to power the coil. The increased energy requirements result in a higher likelihood that the coil burns out, which results in a higher false trip rate. Therefore, if we seek to optimize the false trip rate while improving safety, then valve stroke testing provides a good solution.
Recommendations for Stroke Testing
Based on currently available information consistent over multiple sources, we recommend:
- Begin solenoid valve stroke testing on a weekly basis recording if valve was stuck.
- After a few months of testing, if the number of times the valve is found stuck seems excessive then halve the test interval.
- Repeat process until an optimum test interval is identified such that stroke testing generally verifies valve movement.
Even infrequent solenoid valve stroke testing can have significant beneficial impacts with respect to improvements in reliability and safety performance. However, if a solenoid valve stroke test at a frequency of once per week is feasible within a given application, implementing it could help dangerous failures from occurring in an ESD.