Safety Valve and Actuator Proof Test Coverage

I saw an email string with a question asking if it’s possible to detect all possible failure modes in a safety valve if very intensive and intrusive tests, such as full-stroke testing, leak testing, etc. are performed. Would it be conceivable to use 100% proof test coverage when performing full repair upon detection of a failure?

Emerson’s Mike Boudreaux noted that many people would not claim 100% proof test coverage, because they cannot fully test all failure scenarios in operating conditions. He noted that it’s common to use 90+% proof test coverage for valves. If one needs 100% coverage then they would need to ensure that they are not only testing all of the dangerous failure modes, but also that they are testing them properly without possibility of human errors.

Mike noted that several Fisher valves and actuators have been certified to the IEC 61508 international functional safety standard. Here is an Exida functional safety assessment report for GX Series valves and actuators, Series ED, ET, EZ, and HP valves, Type 657 and 658 actuators, Vee-Ball Series V150, V200, and V300 valves, and Type 1051 and 1052 actuators. He noted that safety manuals on these products provide insight into the certification for these valves and the related proof test procedures.

In an earlier post, Credit for Partial Stroke Tests in Verification Process, I discussed how partial stroke testing could extend the diagnostic coverage for the safety valve and actuator.

Emerson’s Riyaz Ali shared his process safety experience in the email thread and noted that full-stroke test coverage cannot be claimed at 100%, until and unless the valve is re-built with a proper seat. Generally, mechanical items unlike electronics follow a bathtub curve.

Assuming that SIS shutdown on-off valves do not have as much energy throttling compared with basic process control system (BPCS) valves, one would expect the SIS shutdown valve to have better erosion protection. Still, its seat area needs to be re-built and certified to be as good as its original state, in order to claim 100%. Riyaz counsels to be very cautious in using 100% proof test coverage. He also notes that many leading companies with whom he’s worked have never claimed more than 95%, even after the valve seat has been re-built.

For those of you who do the PFDavg calculations for your organization, what proof test coverage percentage do you use?

MP3 | iTunes

Audio clip: Adobe Flash Player (version 9 or above) is required to play this audio clip. Download the latest version here. You also need to have JavaScript enabled in your browser.

10 comments

  1. bloguman says:

    Why not just perform on-line, fault-tolerant, FULL-STROKE valve testing using a technology like that provided by SafetySIL?

    • Thanks for your comment, Bloguman. By fault-tolerant, are you suggesting a bypass valve around the shutdown valve? I received a note from Riyaz that this is generally not accepted practice. Without the bypass, the full stroke will cause a process shutdown. Riyaz believes that industry consensus is to use the PST only and not full stroke testing for on-line in service operation.

  2. Wayne McCandless says:

    When we say “Fault Tolerant” do we mean maintain the ability to shut off flow despite a single failure, or the ability to shut off or allow flow after a failure? The latter case implies use of a series-parallel architecture that is more complex, costly, and harder to test.

  3. Bloguman, Please do send me more information at Jim.Cahill@emerson.com… thanks!

  4. Bloguman, I did also share your comment with Riyaz. He notes safety engineers will ultimately have to evaluate the option of PST diagnostics on a shutdown valve with a Digital Valve Controller versus having two separate valves with their associated maintenance over time, piping and additional space for the two valves, and the additional risk of having bypass or redundant valves.

    • GPBaker says:

      Yes, I agree that the application would have to be evaluated by safety engineers, as would any loop to be certified with a Safety Integrity Level.
      I will forward to you information on the SafetySIL technology, including a comparative study of the system authored by Bill Gobel of Exida. (I’ll be using a company e-mail address; bloguman is my blogging gmail address).
      I look forward to hearing your comments.

  5. JR CONSTANS says:

    Hi,
    Just a comment, imho, mechanical components does not follow the typical bathtub curve. Electronics components are following bathtub curve (observed failure rate is constant over the life time, meaning in between infant mortality and wearing) while mechanical components are always subject to wearing. That is why using the exponential law for mechanical components is rather a simplistic and easy to use approach but not trully modelling reality. For example, a safety valve used very often (high mode of demand, once every day) will probably be failing after 2 years due to wearing and therefore great attention should be given to the maximum operations allowed for SDVs… Weibull law would be closer to reality but calculation would then be more complex.
    Regards,

  6. Jim Cahill says:

    JR, Thanks for sharing your perspectives. I passed them along to Riyaz for his thoughts. I did a lookup of the Weibull distribution, http://en.wikipedia.org/wiki/Weibull_distribution and its application in reliability engineering, http://en.wikipedia.org/wiki/Reliability_engineering and think your point is very valid.

  7. Jim Cahill says:

    JR, I wanted to pass along Riyaz’s thoughts:

    “It is true that electronics components follow truly bath tub curve. Same physics apply to mechanical valve body as well. It is true that mechanical items wear can be dependent on process parameters, fluid, application, environmental conditions, etc., and can be approximated based on available data.

    However, mechanism still remains same as immature failure due to casting issues, flaws in casting pattern, pin hole porosity, weld repair etc. (assuming that valve metallurgy, sizing, selection etc., have been done right), in the beginning and then useful life as dependent on various factors including its usability based on application, process fluid, process delta p and other related physical data with mechanical wear till it reaches to mature life which will be other side of bath tub curve.”

Leave a Reply