Critical Check Valve Reliability
Following the release of the most recent Fireside Chat on pressure relief critical controls, we received a question specific to critical check valves, or non-return valves (NRVs), and our thoughts on taking credit for them in risk analyses. We thought it worthwhile to explore check valves a little deeper and provide our perspective on why we are not typically able to take credit for check valves in risk analysis studies, unless additional conditions are met.
In general, NRVs are not credited as an independent layer of protection due to their reliability (or rather, a significant amount of data, anecdotal or otherwise, that says as a class of equipment, check valves don’t work completely a fair number of times when they are called upon). This approach is analogous to the general stance for NRVs in evaluation of overpressure protection – without further evidence to the contrary, positive action of NRVs is not considered as a means of preventing a cause of overpressure due to reverse flow. The ‘further evidence’ piece is where critical check valves come into play. A critical check valve (or a series of them) is one in which there is additional evidence that indicates their reliability is sufficiently commensurate with the risks involved that they can be credited as a layer of protection.
In some cases, this ‘credit’ involves simply a reduction in the anticipated rate of reverse flow, and in other cases, the credit is for prevention of an overpressure scenario altogether; both of these are discussed in API Standard 521 §22.214.171.124. The means of providing that additional evidence is also discussed in API Standard 521 – primarily direct maintenance and inspection. The purpose of the maintenance and inspection is to ensure the reliability of the NRV if called upon to act. If the reliability can be established in accordance with the operator’s acceptable risk criteria, credit can be taken per the above referenced section in API Standard 521. It is important to note here that the API guidance states that, with respect to overpressure protection, complete failure of a single check valve is expected regardless of maintenance and inspection. So, for credit to be taken there must be at least two check valves in series.
Historically, industry had put more faith in NRVs than actual experience has shown to be warranted. This is not to say that all check valves don’t work, just that special effort needs to be taken to demonstrate their reliability for cases where failure on demand may have significant adverse consequences. It is left to the operator to determine appropriate procedures to determine reliability, and API Standard 521 provides guidance for what may need to be considered. Once the reliability has been established, there should not be any impediment to using that reliability in various risk analyses, whether they be LOPA or determination of credible overpressure scenarios. It is important to note that the reliability is only credible if the programs used to establish the reliability are maintained, which ties back to the original post on pressure relief critical controls.