Received: 6 February 2020 Accepted: 20 March 2020

DOI: 10.1002/prs.12153

ORIGINAL ARTICLE

Crediting check valves as IPLs? Testing protocol to better

understand check valve reliability

Jody E. Olsen

JE Olsen Consulting LLC, Anchorage, Alaska

Abstract
Correspondence Conventional process safety wisdom assumes that check valves are not reliable
Jody E. Olsen, JE Olsen Consulting LLC,
Anchorage, AK. safeguards. Experience indicates that check valves are prone to failure and that they
Email: jodyo@jeolsenconsulting.com may fail undetected. Therefore, the conservative assumption is that check valves
may be listed in process hazard analyses as safeguards, but they are rarely considered
to meet the standards required of an independent protection layer (IPL). Independent
protection layers must be effective, independent, and auditable. Although indepen-
dence is readily achievable by check valves, confirming and routinely auditing effective-
ness is rarely pursued. And maintenance practices for check valves are often insufficient.
Little data is available from operating companies regarding failure and leakage rates for
different check valve types in various service applications or at various stages of service
life. This paper examines a testing protocol that was put in place in 2014 for the purpose
of testing check valves in order to apply layer of protection analysis (LOPA) credit to
these valves for reverse flow scenarios. In order to understand check valve performance
expectations, leakage allowances for new check valves are reviewed. Industry guidance
and standards regarding consideration of check valves as safeguards or IPLs are also dis-
cussed. The analysis of new valve standards and the assessment of process safety
requirements are the basis for establishing the pass/fail thresholds for the tests. The goal
of sharing this information is that the discussion will stimulate others to consider the
opportunity and the need to set-up similar testing and to begin gathering and sharing a
larger body of data on check valve performance in various applications. Accumulation of
check valve performance data and sharing of that data should lead to better understand-
ing of check valve performance by type, size, age, and service. Better performance may
be achieved where maintenance is improved and where learnings are applied to selec-
tion and design. In instances where requirements are met and credit is due, check valves
may be credited in PHA and LOPA.1

KEYWORDS

check valve, safeguards, independent protection layer (IPL), process hazard analysis (PHA),
layer of protection analysis (LOPA)

1
This article was prepared for presentation at American Institute of Chemical Engineers 2019 Spring Meeting, 15th Global Congress on Process Safety, New Orleans, Louisiana, March 31 – April 3, 2019.

This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium,
provided the original work is properly cited.
© 2020 The Authors. Process Safety Progress published by Wiley Periodicals, LLC on behalf of American Institute of Chemical Engineers.

Proc Safety Prog. 2020;39:e12153. wileyonlinelibrary.com/journal/prs 1 of 19

https://doi.org/10.1002/prs.12153
2 of 19 OLSEN

1 | I N T RO DU CT I O N This paper advocates changing the paradigm. Critical check valves

need to not only be inspected but also tested for leakage rates in
Check valves are a curious equipment component in process plants order to assure their reliability. Where defects are found, check valves
and operations. Check valves are by their very nature a safeguard. should be repaired and returned to acceptable condition. That means
They prevent flow in the wrong direction. However, standard practice that the installation configuration must allow for future inspection,
during process hazard analysis is to assume that the check valve will maintenance, and testing. Maintenance work must be planned, sched-
fail and to give no credit to the safeguard. This long-standing practice uled, and completed. History on the performance of check valves in
seems prudent given experience with serious incidents involving various applications, for various types and sizes, and at various stages
check valve failure. of their service life will allow for better design and selection of check
However, does the assumption of check valve failure become a self- valves for the particular application. Assuring proper function of check
fulfilling prophecy? Does the assumption of failure become an enabler of valves may allow for more robust process safety design by sharing the
failure? Does failure to recognize actual failure modes and frequencies burden of protection across more disparate safeguarding devices and
prevent improvement in check valve performance and reliability? independent protective layers. Understanding the actual performance
In most applications, check valves are not causes of hazard sce- of check valves allows for attributing appropriate credit, where it is due,
narios, they are preventative safeguards. and also allows resources to focus attention on hazards that have a
larger actual gap between target risk levels and current assessed risk.
As per Baldas/Carithers: Check valves are almost exclu- While testing check valves is necessary to allow for potential
sively used as safeguards, rather than as active design crediting of check valves as independent protection layers (IPLs), inspec-
features…. It is not typical that the failure of a check tion programs must also be in place to identify insipient failures. Inspec-
valve is the cause of a [Layer of Protection Analysis,] tions reveal conditions that are precursors to failures such as mechanical
LOPA[,] scenario; the cause of the reverse flow comes damage, fouling, debris, and looseness that are not yet manifested as
from some other failure [or initiating event (IE)]. A check leakage. The particular application, hazards, and consequences must be
1
valve is typically a failure to protect, not a cause. thoroughly understood in order to consider PHA and/or LOPA credit for
individual or redundant check valve installations.
CCPS Guidelines for Initiating Events and Independent Protection This paper discusses an example testing protocol that was written
Layers in Layer of Protection Analysis (CCPS IE/IPL), 2015,2 generally and implemented for several new check valves that were installed in an
supports that characterization. In Section 4.3.3 and Appendix D, a dis- oil and gas processing plant in 2014. In preparation for that testing, stan-
tinction is made between check valves operating in low demand mode dards that define maximum allowable leakage rates for new valves were
versus high demand mode.2 Check valves operating in high demand reviewed. Guidance from industry publications on considering check val-
mode, such as internal to equipment, tend to be design features and ves as safeguards or IPLs was also reviewed. The analysis of valve stan-
may be initiating events. Whereas check valves operating in low dards and the assessment of process safety requirements were the basis
demand mode are generally operating as a safeguard. for establishing the pass/fail thresholds. Key take-aways and experiences
from the preparation and testing are summarized.
Note: API 521 lists check valves as “causes” for over-
pressure scenarios; however, the discussion regarding
reverse flow scenarios is much the same.3 2 | HI STO RY

As such, we list check valves as safeguards in process hazard Early publications of process safety hazard analysis methodology includ-
reviews, but check valves rarely meet the standards required for con- ing CCPS LOPA (2001)5 and CCPS Guidelines for Hazard Evaluation Pro-
sidering the check valve or check valves in series as an independent cedures, Table 7.4, (2008, 1992)6 cautioned practitioners that check
protection layer (IPL). Check valves are recognized for having the valves should not be credited as safeguards or IPLs. With the publication
potential to fail, to fail undetected (ie, latent failure), and to fail sud- of later references, such as CCPS IE/IPL in 2015, guidance has changed
denly or catastrophically. Check valves are also recognized to operate to allow for the usage of check valves as IPLs.2 From CCPS IE/IPL:
in severe service because moving parts are present in the flow stream.
However, lack of testing and inspection data remain a primary obsta- Section 1.3 … At the time of CCPS LOPA (2001), check
cle to being able to better assess and quantify the actual performance, valves were not generally considered to be valid IPLs
failure modes, and failure frequencies for check valves. due to a lack of data supporting their reliability. Since
Although inspection, maintenance, and testing (IMT) are consid- that time, understanding of check valve reliability has
ered essential to assure the performance and reliability of any piece of improved, assisted by more data that substantiates
mechanical equipment, check valves generally do not receive the their reliability. (Refer to Appendix D for Example Reli-
4
same attention as other equipment and components. The lack of rou- ability Data Conversion for Check Valves for more
tine inspection and/or maintenance may in part explain why check information.) Based on this data, check valves have
valves are considered unreliable. been included as IPLs in this book.2
OLSEN 3 of 19

Section 1.7… Appendix D … Check valve design has 1 Single check valve—credited as stand-alone IPL
improved over the years, and data exist that indicate 2 Dual check valves in series—credited as single IPL
that check valves can be effective IPLs when properly 3 Combination IPL comprised of upstream PSV or PSV setting and
specified and maintained.2 downstream check valve(s)
Section 5.2.2.3 … check valves can be considered to be
IPLs when properly specified, designed, installed, and Inspection and testing protocol including threshold leakage rates
2
maintained. must be defined depending on which of the above configurations is
Note: The data referenced in Appendix D2 is based on applicable. When establishing threshold leakage acceptance rates, it is
1987 data. However, the Probability of Failure on important to consider the amount of leakage that is deemed tolerable
Demand (PFD) analysis and generic IPL credit guidance within the particular process system in question. Process equipment
came later. rating and capacity, reactivity hazards, relief device capacity (where
applicable), upstream rotating equipment impacts, and other factors
Little has changed since the initial publication of PHA and LOPA for a given scenario must be considered. As with all IPLs, each check
guidance with respect to opinions, assumptions, and maintenance valve IPL must prevent the consequence of concern.
practices regarding check valve reliability. As such, practices and It is important to note that option 3 references a single IPL that is
assumptions in the process safety arena have remained essentially the comprised of a specific PSV or PSV setting and a specific check valve
same as well. or check valves in series. Inspection, maintenance, and testing of all
In spite of regulation, such as OSHA PSM 1910.119(j) Mechanical components are required to credit the single IPL. These components
7
integrity, (1992), and industry guidance, CCPS Guidelines for Risk may not be used separately as independently credited IPLs. Likewise,
Based Process Safety, 4.3.3 Element—Asset Integrity and Reliability in option 2, two check valves may be credited as a single IPL and both
(CCPS RBPS), (2007),8 that stipulate inspection, maintenance, and test- check valves require independent inspection, maintenance, and
ing requirements to ensure equipment integrity and fitness-for-service, testing.
many check valves remain untagged equipment items indicating that
the check valve will not be included in integrity management programs
and no history will be tracked as to the condition of the valve. All check 3.1 | Crediting check valves, single or dual check
valves may not require testing, but inspection, failure, and repair history valves, as an IPL
should be tracked, at a minimum, for all safeguards. Guidance in API
5709 May 2017 addendum also requires that critical check valves Several reference documents address potentially crediting check val-
“shall” be included in inspection or testing programs.4 ves as safeguards and/or IPLs. CCPS IE/IPL2 Section 5.2.2.3 and
While some operators have migrated to include critical check val- Table 5.3.2 address crediting a single maintained check valve as an
ves in basic inspection and maintenance programs, very few operating IPL. CCPS IE/IPL distinguishes check valves operating in low demand
companies test their check valves. Therefore, very few operating com- mode as IPLs whereas check valves operating in high demand mode,
panies credit their check valves as IPLs. While many standards exist such as integral to equipment, are considered initiating events. How-
for quantifying leakage rates for new check valves provided by manu- ever, the calculated probability of failure is the same.
facturers, little to no quantitative data is available on check valve leak-
age with respect to type, size, age, or service after check valves are Note: In CCPS IE/IPL,2 the probability of failure evalua-
placed in service. Lack of performance data rather than specific data tion for a single check valve in either mode, high demand
indicating unacceptable leakage rates is a primary reason for lack of as initiating event or low demand as IPL, is based on the
movement towards potentially crediting check valves as IPLs. This sit- same data set. The probability of failure on demand (PFD)
uation applies not only to older check valves that may not be well calculated for a single check valve IPL is the same as the
configured for testing, but also to newer check valve installations. initiating event frequency (IEF) for a high demand check
valve, 0.1. CCPS IE/IPL also calculates IEF for dual check
valves in the high demand mode of 0.01. The data set
3 | P H A OR L OP A CR E D I T— W H E N T O used for the probability of failure calculation is based on
C RE D I T A N D H O W M U CH Reliability Data Book for Components in Swedish Nuclear
Power Plants, RKS/SKI 85-25, p 79 (Bento et al. 1987).
When considering giving a check valve or check valves qualitative Calculations are shown in Appendix D of CCPS IE/IPL.
(order of magnitude) credit as a PHA safeguard or semi-quantitative
LOPA credit as an independent protection layer for a hazard scenario, CCPS IE/IPL2 highlights the fact that some check valve leakage
there are several options for crediting that depend on the equipment may be expected. Therefore, the user must define tolerable leakage
and process configuration. rates and assure through inspection and leakage testing that the check
Check valves that meet IPL criteria may be considered for valve can be expected to perform as anticipated and protect against
crediting as IPLs when in the following configurations: the scenario. As per CCPS IE/IPL,2 for consideration as an IPL, the
4 of 19 OLSEN

check valve should be operating in clean, nonfouling, noncorrosive have moved toward applying credit to check valves as IPLs. In “Check
service. Valves as Safeguards: Friend or Foe in Process Hazard Analysis,”
An IPL with calculated PFD for dual check valves is not explicitly Modi,10 Table 9.1 provides a sampling of credit given to check valves
2
discussed in CCPS IE/IPL Section 5.2.2.3. CCPS IE/IPL cautions the by operating companies of various size and sector. Based on this over-
users to consider design, operability, and function of two check valves view, very few companies attribute any credit to a single check valve
in series to assure that both respond appropriately and independently. as a safeguard in PHA analyses. A number of companies attribute a
2
CCPS IE/IPL also reminds the user to consider common cause fail- single order of magnitude credit to dual check valve safeguards in
ures. Manufacturers indicate that failures associated with multiple PHA assessments. But LOPA credit is rarely attributed even to dual
check valves in series may be more likely in the forward flow direction check valves.
due to failure to meet the required cracking pressure. This type of fail- As with any IPL, check valves considered for crediting as an
ure does not impact the reverse flow protective function. IPL must meet all required criteria, including assuring
Although, CCPS IE/IPL2 does not show a separate dual check (a) effectiveness, (b) independence, and (c) auditability. The deci-
valve IPL category, it is assumed that the PFD for a dual check valve sion to credit either a single check valve or dual check valves in
IPL will be similar (0.01) to the initiating event frequency (IEF) for dual series as a safeguard or as an IPL is based on individual company
check valves in high demand mode. Crediting dual check valves with protocol. This protocol should be based on historical experience,
this PFD assumes that the check valves are individually inspected and understanding of check valve failure modes, reliability and perfor-
tested and are unlikely to experience common cause failures. mance, understanding of hazard scenarios involved, and company
Crediting of check valves that meet IPL criteria is also discussed risk tolerance. The lack of credit taken for check valves represents
1
by Baldas and Carithers. They calculated check valve IPL credit for a conservative approach. However, this conservative approach is
multiple valves in series based on a single check valve PFD = 0.1, or also attributable to lack of progress in understanding check valve
Risk Reduction Factor (RRF) = 10. Using this assumption and an performance, failure modes, and failure frequencies in various pro-
assumption for common cause failure rates of 1 in 10, Baldas and Car- cess services.
1
ithers calculated a PFD for dual check valves in series of 0.0181 Refer to Table 1 for a summary of check valve probability of fail-
(RRF = 55) which rounds to approximately two orders of magnitude. ure guidance based on several references discussed above. This table
They further showed via the same calculation that more check valves summarizes potential PHA or LOPA credit that may be given when
in series would not yield a substantially lower combined probability of check valves are maintained, inspected, and tested and considered
failure on demand; therefore, no more than two orders of magnitude with regard to special considerations stipulated in the guidance docu-
credit should be attributed to check valves in series, even if more than ments. Table 1 also shows discrepancies between guidance
two check valves, that are maintained, are present in series.1 This con- documents.
clusion is also supported in API 521 (2014)3 Section 4.4.9.3.3. “a)
Because of potential common mode failures the user is cautioned
against taking a larger credit for more than two check valves in series
3.2 | Crediting check valve(s) and upstream PSV(s)
that are inspected and maintained.”
as an IPL
It is notable that CCPS IE/IPL2 and API 521(2014)3 differ on guid-
ance regarding single check valve performance and probability of fail-
CCPS IE/IPL2 not only references the possibility of crediting check
ure on demand (PFD). Although API 521 (2014)3 only addresses check
valves as IPLs, it also addresses the possibility of crediting a PSV in
valve(s) working together with PSV(s) to comprise the IPL and does
conjunction with a check valve in Sections 3.2.1 and 5.2.2.2.2 as per
not explicitly address mathematical PFD for check valves, inherent
configuration 3 above. API 5213 also references configuration 3. How-
unreliability is implied in this standard. API 521 (2014)3 assumes that
ever, CCPS IE/IPL2 and API 5213 differ regarding guidance as to
a single check valve, regardless of service, history, or testing, is not
whether the combination PSV/check valve IPL may be credited with
sufficient to be credited for even partially restricting flow to the PSV.
only a single check valve in place that limits reverse flow to the PSV.
In other words, this guidance assumes a single check valve should
API 521 (2014)3 does not allow for crediting a single check valve with
receive no credit for being in place, therefore, PFD = 1 (RRF = 1). API
reducing flow to the PSV. API 521 (2014)3 directs the user to assume
521 (2014)3 dictates guidance that all single check valves should be
“complete” check valve failure when a single check valve is in place,
assumed to completely fail and not reduce reverse flow to the PSV.
even if maintained and inspected. CCPS IE/IPL2 references a single
This conclusion is fundamentally different from the position
check valve in scenarios involving an IPL comprised of a PSV and a
supported by data and PFD calculations in CCPS IE/IPL2 Appendix
check valve.
D. Additional data should be gathered that is specific to check valve
type, size, age, and service in order to better define the PFD for a
Note: Regardless of these discrepancies in guidance, it is
given check valve.
important to recognize that the guidance provided in API
Although CCPS IE/IPL2 provides a basis for crediting either single
5213,11 only applies to configuration 3 above. Configura-
or, by extension of IE discussion, dual check valves when IPL require-
tions 1 and 2 are not related to relief scenarios.
ments and other special considerations are satisfied, few companies
 OLSEN

TABLE 1 Potential PHA and LOPA credit

PFD = probability of failure on demand

IEF = initiating event frequency
RRF = risk reduction factor
OM = order of magnitude Independent protection layer Initiating event *Scenario likelihood reduction

Device Safeguard PFD RRF IEF RRF* Reference

Single check valve (low demand mode)** 1 OM 0.1 10 CCPS Guidelines for initiating events and independent
protective layers in layer of protection analysis, Chap 5,
5.2.2.3, Table 5.32 based on Appendix D2
Dual check valves (low demand mode) Not Addressed Not Addressed Not Addressed CCPS IE/IPL2, Not addressed
Single check valve (low demand mode) 1 OM 0.1 10 Baldas/Carithers1, assumed value
Dual check valves (low demand mode) 2 OM 0.01 100 Baldas/Carithers1, calculated, assumption: no common cause
failures
Dual check valves (low demand mode) ~ 2 OM 0.0181 55 Baldas/Carithers1, calculated, assumption: common cause
failures
Three check valves (low demand mode) 2 OM 0.0107 93 Baldas/Carithers1, calculated, assumption: common cause
failures
Single check valve (low demand mode)** 0 OM 1 1 API 521, 2014, 6th Ed,3 Implied based on assumption that
single check valve will always fail
Single check valve (high demand mode) NA 0.1 10 CCPS IE/IPL2, Chap 4, 4.3.3, Table 4.11 based on Appendix D2
Double check valve (high demand mode) NA 0.01 100 CCPS IE/IPL,2 Chap 4, 4.3.3, Table 4.12 based on Appendix D2

Note: All values assume that appropriate inspection and testing is conducted and that special considerations have be reviewed for the scenario and for the application of the generic probability of failure value
given. **See text for discussion of discrepancies.
5 of 19
 6 of 19

TABLE 2 Industry valve standards—maximum allowable seat leakage rate references

Valve maximum allowable seat leakage Check valve maximum allowable seat
rate by size (in diameter)—Table or section leakage rate by size (in diameter)—Table
Standard Valve types included within scope where defined or section where defined
API 598 (2016) Valve Gate; Globe; Plug; Ball; Butterfly; Check API 598 (2016): Table 5—Maximum API 598 (2016): Table 5—Maximum
Inspection and Testing12 allowable leakage rates for closure tests, allowable leakage rates for closure
page 912 tests, page 912
ISO 5208: 2015 Industrial Gate; Globe; Plug; Ball (Floating; Trunion Mounted); ISO 5208: 2015: Table 4—Maximum ISO 5208: 2015: Table 4—Maximum
valves—Pressure testing of Diaphragm; Butterfly (Concentric; Eccentric); Check allowable closure test leakage rate, page allowable closure test leakage rate,
metallic valves13 1113 defined as Rate A, AA, B, C, CC, D, page 1113 Corresponding API 598
E, EE, F, G from most stringent to least limits for check valves are referenced:
stringent requirements. Corresponding • Rate G, using liquid test, and
API 598 limits are referenced: • Rate EE, using gas test. Also see
• Rate A for DN ≤ 50 and soft seated Supporting Information S1
• Rate CC, using liquid test, and
• Rate AA, using gas test. Also see
Supporting Information S1
MSS SP-61-2013 Pressure Gate; Globe; Plug; Ball; Butterfly; Check MSS SP-61-2013: Table 4—Units of MSS SP-61-2013: As per Section 5.7.2,
Testing of Valves14 Leakage per NPS/DN, page 4.14 Also see allows four (4) times leakage rate
Supporting Information S2 shown in MSS SP-61-2013 Table 4—
Units of Leakage per NPS/DN, page 3,
for check valves.14
API 6D (2014) Specification Gate; Plug; Ball; Check API 6D (2014): Regarding seat testing for API 6D (2014): Regarding seat testing for
for Pipeline and Piping valves (other than check valves), check valves, maximum allowable
Valves15 maximum allowable leakage rate is given leakage rate is given as:
as: • ISO 5208 Rate G for liquid test per
• ISO 5208 Rate D for liquid test per Section 9.4.3*
Section 9.4.3 • ISO 5208 Rate EE for gas test per
• ISO 5208 Rate D for gas test per Section Section H.4.3.2*15 *These limits are
H.4.3.215 consistent with API 598.
ANSI/FCI 70-2-2013 Control Control ANSI/FCI 70-2-2013: Table 1, page 2, and Not applicable
Valves Seat Leakage19 Table 2, page 319
OLSEN
 OLSEN

TABLE 3 Valve seat leakage allowance by standard,12-15 liquid test

Note: Also see Supporting Information S3 for expanded tables showing all ISO 5208 leakage rate categories.
7 of 19
 8 of 19

TABLE 4 Valve seat leakage allowance by standard,12-15 gas test

Note: Also see Supporting Information S4 for expanded tables showing all ISO 5208 leakage rate categories.
OLSEN
OLSEN 9 of 19

FIGURE 1 Overview comparing check and block valve seat leakage allowance, liquid test12–16

Configurations 1 and 2 consider the check valve or check internal manufacturer standard that meets or exceeds these standards:
valves as an IPL on its own merits. API 598,12 ISO 5208,13 MSS SP-61,14 or API 6D.15,16 EN-12266-117 is
an equivalent European standard which superseded BS 6755-1.18 The
2
As per CCPS IE/IPL, when a PSV or PSV setting is used in combi- standard chosen depends primarily on the valve materials of construc-
nation with a check valve or check valves as an IPL, the probability of tion, location jurisdiction, industrial sector, or customer requirements.
failure on demand (PFD) attributed to the combination PSV(s) and The standards referenced stipulate not only maximum allowable leakage
check valve(s) IPL should be equal to the PFD of the component hav- rate criteria, but also testing requirements and procedures, including test
ing the greatest probability of failure. CCPS IE/IPL3 references the sin- pressure and duration by valve type.
gle check valve PFD of 0.1, provided in Table 5.32, as the overall PFD Metal-to-metal seated valves of all types, including check valves, are
for the single check valve/PSV combination IPL as an example. expected to have some leakage. Resilient seated valves, also called soft
While check valve PFD is discussed directly or indirectly in CCPS seat valves, have more stringent leakage standards, usually indicated as
IE/IPL2 and API 521,3 frequency of inspection and/or testing of check zero leakage or no visual leakage. Most of these industry standards allow
valves is not defined. Frequency of inspection and testing is left to the for higher seat leakage rates through metal seated check valves as com-
user to define. As with other IPLs, inspection and testing frequency pared to other types of metal seated block valves. However, some manu-
must support and validate the reliability of the component. Shut-down facturers design check valves to leakage standards comparable to other
requirements for testing of check valves must also be considered and valve types. Other valve manufacturers may conform to the check valve
planned in conjunction with defining inspection and test frequency. standards but exceed the minimum requirements.
Table 2 summarizes information from several valve standards.
Table 2 shows which valve types are addressed by each standard and
4 | N E W CH E C K V A L V E S E A T L E A K A G E references tables within those documents that stipulate maximum
R E Q U I R E M E N T S — INDUSTRY STANDA RDS allowable seat leakage rates.
Recent updates to major standards have resulted in some conver-
In order to understand the potential performance capability for check gence of leakage limits. Table 3 and Table 4 show maximum allowable
valves in operational service, we must first understand the performance valve seat leakage rates for various valve sizes as defined by API
expectations for new check valves. Standards have been established that 598,12 MSS SP-61,14 API 6D,15 and ISO 520813 using consistent units
define valve performance requirements including seat leakage rate allow- of measure. Table 3 and Table 4 reference tests conducted using liq-
ances. In the United States, block valves and check valves are generally uid and gas, respectively. Manufacturer performance standards for
tested by manufacturers to one of the following standards or to an two example manufacturers are also shown. Values shown in Table 3
10 of 19 OLSEN

FIGURE 2 Overview comparing check and block valve seat leakage allowance, gas test12–15

and Table 4 are approximate as calculated by author. Refer to As noted, all mechanical valves are prone to leak, regardless of
Supporting Informations S3 and S4, for expanded tables that show valve type or style. Zero leakage rates are only achieved where
seat leakage rates for all ISO 520813 performance levels. Figures 1 resilient or soft seats are included in the valve design. Valves
and 2 provide simplified overview graphics of allowable seat leakage designed for zero leakage rates using soft seats include various
rates using liquid and gas test media, respectively. categories of valves, including check valves. However, temperature
Leakage rates for various types of block valves, including gate, and materials compatibility issues as well as wear and maintenance
ball, plug, globe, and butterfly, as well as check valves, are included requirements must be considered when soft seating materials are
in the standards referenced above. A separate standard, ANSI/FCI specified.
70-2 Control Valve Seat Leakage,19 specifically addresses control
valves. This standard is a useful point of comparison as control val-
ves operated as basic process control systems (BPCS) may also be 5 | IN-SERVICE CHECK VALVE SEAT
considered for credit as IPLs, as per CCPS IE/IPL2 5.2.2.1, Safety LEAKAGE ASSUMPTIONS
Control Loop (Normal Operating Control) Table 5.12, if they meet
IPL criteria. API 521 (2014)3 is considered the definitive reference regarding sizing
and capacity of relief equipment. API 521 (2014)3 provides guidance
2
Note: CCPS IE/IPL Section 5.2.2.3, Check Valves as on assumptions for check valve leakage rates that dictate PSV relief
IPLs, references API 598 (2009b), MSS SP-61(2009), capacity requirements for reverse flow scenarios. Many users refer-
and ANSI/FCI 70-2 as applicable standards that ence this guidance with respect to check valves given the lack of in-
address check valve leakage rates. Specific leakage rate service leakage rate data that is available in industry from other
classes from ANSI/FCI 70-2, Class I through Class VI sources. Section 4.4.9.3.1 of API 521 (2014)3 discusses several sce-
are replicated in Check Valves as IPLs section of CCPS narios for reverse flow conditions, including complete check valve fail-
IE/IPL 5.2.2.3. This reference is misplaced with respect ure, severe leakage, and normal leakage.
to check valves, as ANSI/FCI 70-2 is specific to control
valve leakage. Manufacturers of check valves surveyed
confirmed that they refer to API 598, ISO 5208, API 5.1 | Complete check valve failure
6D, or appropriate MSS standard for requirements on
allowable leakage rates and testing protocol. Some man- As per API 521 (2014),3 complete check valve failure is considered to
ufacturers institute more stringent corporate leakage be “potential for gross reverse flow because the check valve does not
standards. function at all (e.g. it is stuck wide open or the internals are gone).” …
OLSEN 11 of 19

“Complete check valve failure is assumed for all check valves in series These leakage rate assumptions for severe check valve failure
that are not inspected and maintained and for a single check valve provided in API 521 (2014)3 are considered conservative and often
regardless if it is inspected and maintained.”3 form the basis of PSV design calculations where leakage rates are not
known or quantitatively tested.
Note: Discrepancies between API 521 (2014)3 and CCPS CCPS IE/IPL2 Section 5.2.2.2.2 Case 1, PSV with check valve IPL,
IE/IPL2 guidance on check valve PFD were previously dis- references similar language that is taken from API 521 (2008).11
cussed in this article, Sections 3.1 and 3.2. (Note: A single check valve is referenced in Case 1 as described in
CCPS IE/IPL.2)
The assumption of complete failure means that the normal opera-
tion of the check valve has no limiting effect on the reverse flow rate. From CCPS IE/IPL2 … one may estimate the reverse
API 521 (2014) provides three options for estimating reverse flow flow through series check valves as the flow
through a completely failed check valve.3 through a single orifice with a diameter equal to
one-tenth of the largest check valve's nominal diam-
1 Assume no flow resistance in reverse direction.3 eter. A lower value may be used if a condition-
2 Assume the same flow resistance in reverse through the check monitoring system for the check valves … is
valve as in the forward flow direction.3 installed to monitor the condition to ensure that the
3 Assume the failed check valve has an orifice equal to the check leakage rate is below the capacity of the low-
valve flow area without internals.3 pressure side relief device.2

CCPS IE/IPL 5.2.2.2.2 Case 22 speaks to this “complete failure” This reference to API 521 (2008)11 guidance alludes to “severe”
scenario. However, since the check valve is uninvolved in a protective leakage in the beginning of the statement and “normal” leakage as the
capacity in this scenario, the IPL is comprised of only the PSV, not a “lower value.” In practice, most users assume the severe leakage rate
combination of the check valve and PSV. as described above in CCPS IE/IPL2 and API 521 (2008),11 or as
described in API 521 (2014),3 assumption 1 or 2, because leakage
Note: If assumption 2 or 3 above is used for sizing the rates are rarely, if ever, tested directly.
PSV, in the event of removal or modification of the check
valve, the PSV capacity would have to be revalidated Note: In CCPS IE/IPL2 Section 5.2.2.2.2 Case 1, the fol-
through the MoC process. lowing statement is made: “If the ANSI/API 521 (2008)
[or 2014] guidance is followed, the PRV sizing basis would
be reduced to handle just the flow from a slight reverse
5.2 | Severe check valve leakage leak through the check valve.2”

As per API 521 (2014),3 severe check valve failure is considered to be The word slight in this sentence is misplaced as this assumption
“potential for significant reverse flow because of check valve seat for leakage represents the “severe” failure case having “significant”
damage or obstruction.” API 521 (2014)3 advises that it is the respon- leakage rates. Leakage rates discussed later in this paper in conjunc-
sibility of the user to determine the appropriate method for estimating tion with testing protocol also support the position that this reverse
reverse flow, but offers the following guidance. “Where no specific flowrate assumption represents “severe” rather than “slight” or “nor-
experience or company guidelines exist,” API 521 (2014)3 directs the mal” leakage of the check valve.
user to assume complete failure of the smallest check valve and to In order to utilize this guidance and credit the entire IPL,
assume severe failure of all other check valves in series. both the relief valve(s) and the check valves, must be maintained
Two alternatives are provided in API 521 (2014)3 for assuming via company preventative maintenance (PM), inspection, and
reverse flow rate for check valves having severe leakage. The follow- repair programs. It is important to note that this guidance is pro-
ing assumed reverse flow rate may be used for sizing relief devices vided and is often used in lieu of known quantitatively tested
upstream of inspected and maintained check valves.3 check valve leakage rates, which explains the “conservative”
assumptions that are used in API 521 (2014),3 (2008)11 for
1 Assume that the check valve orifice diameter is 10% of the check assumed leakage rates. While “severe” leakage assumptions may
valve nominal diameter (1% of nominal flow area) and calculate be considered conservative based on the magnitude of reverse
reverse flow.3 (See API 521 (2014)3 for properties and parameters flow allowed by the calculation, data is not available that sup-
assumptions. (Note: API 521 (2008)11 references the same flowrate ports the notion that the failure mode described by the API
calculation assumption without parameter details provided.) 521 (2014)3 “severe” failure is more likely to occur than “com-
2 Treat check valve as an orifice sized to pass 10% of normal forward plete” failure as described by full loss of internal parts. From this
flow. Back calculate orifice size. Then calculate reverse flow.3 (See standpoint, using the API 521 (2014)3 “severe” failure assump-
3
API 521 (2014) for properties and parameter assumptions.) tion to calculate reverse flow rate may not be prudent without
12 of 19 OLSEN

seat leakage testing data and/or failure modes and failure fre- Key considerations:
quency data to support that approach.
• Isolation requirements both downstream and upstream of the
check valve
5.3 | Normal check valve leakage • Tap locations, for connecting pressure indication devices/
transmitters
3
As per API 521 (2014), normal check valve failure is considered to be • Tap locations, for connecting pressure source, return line, and
“potential for minor reverse flow due to normal check valve wear.” bleed line
API 521 (2014)3 advises that normal check valve leakage may be
assumed when two of more check valves are in series that are Each check valve assembly involved in the testing program, whether
3
inspected and maintained and demonstrate reliability. API 521 (2014) a single check valve or dual check valves in series, must be configured
does not allow for assumption of normal leakage rates through a sin- with sufficient upstream and downstream taps or connections to allow
gle valve, even if inspected and maintained. (Note: Guidance changed for testing each check valve individually. Isolation of the check valve for
in API 521 (2014),3 as API 521 (2008)11 did acknowledge the ability of testing must also be considered. Many legacy installations lack adequate
a single inspected and maintained check valve to limit reverse flow.) spacing for required connections or isolation locations.
API 521 (2014)3 does not provide the user with assumptions for nor- Taps are required in order to monitor pressure both upstream and
mal leakage rates. For normal leakage, the user is advised that if a pressure downstream of all individual check valves that will be tested. For testing
relief device is involved in the IPL, the relief device loads and sizing includ- purposes, these locations should be configured with pressure transmitters.
ing the check valve reverse flow leakage rates must be determined by the Where an external pressure source is required to deliver the test pres-
user. API 521 (2008)11 alluded to methods for establishing actual leakage sure, taps are also required to provide that pressure source to the down-
rates such as through condition-monitoring using pressure indicators but stream side of the check valve. Taps must be provided on the normally
did not reference specific assumptions or “normal” leakage values. upstream side of the valve, as well, in order to allow for return flow during
Having reliable, repeatable check valve leakage data along with the test if a flow through test is performed. Connections must be provided
historical failure modes and failure frequency data associated with the to bleed the pressure from both sides of the check valve following the test.
check valve type, size, age, and service is essential for users planning Additionally, isolation requirements must be considered. In order to
to assume normal leakage rates. Establishing normal leakage rate data assure that the leakage measured is going in the direction assumed, such
is essential for application of either combination PSV/check valve IPLs as in reverse across the check valve, the system on both sides of the
or standalone check valve IPLs. Normal leakage should be evaluated check valve must be tight and leak tested in advance of performing the
and considered relative to new check valve leakage standards. leak test on the check valve. Obtaining and proving a tight system may
be achieved by using double block and bleed valving on process piping
upstream and downstream of the check valve(s) or by providing isolation
5.4 | Alternative characterization of seat leakage points for installing blinds both upstream and downstream. As always,
the costs and challenges of getting an effective double block and bleed
An alternative characterization to the three check valve leakage con- leak test must be weighed against safety risk and operational require-
3
ditions described in API 521 would be complete failure; degraded ments for opening and blinding. Any laterals that tie into the check valve
condition; and like-new condition. Like-new condition implies that piping involved in the testing have the same isolation requirements.
maximum allowable seat leakage rates given by industry standards, Therefore, these laterals should be minimized or eliminated where possi-
such as API 598,12 ISO 5208,13 or MSS SP-61,14 are met. Degraded ble unless the lateral provides the source pressure.
condition encompasses leakage rates ranging from like-new to com- The entire testing configuration and procedure steps must receive a
plete failure. Leakage rates characterized by API 5213 as severe would full process hazard review before any external connections are made and
fall into the upper range of degraded performance. Leakage rates before any procedural steps are implemented. Likewise, if test pressure
characterized by API 5213 as normal would fall within the lower range is provided via a process or utility pressure source through an alternate
of degraded performance or meet like-new requirements, depending (abnormal) valve alignment, the valve alignment and procedural steps
on interpretation of normal and expectations of check valves. must be evaluated via process hazard review prior to implementation.

6 | TE S TI N G D E SI GN CO NS I D E RA T I ON 7 | T H I S T ES T— P U R P O S E A N D
(GENERAL) DE S C R I P T I O N OF CO N F I G U R A T I O N

In order to implement a quantitative leak testing program for check 7.1 | Purpose
valves, the process and mechanical configuration should be consid-
ered in the design phase of new projects and modifications to existing The check valve configuration and the testing procedure described
installations. here were put in place in 2014. The reason for testing the check
OLSEN 13 of 19

F I G U R E 3 Test 1 configuration.
Test single axial check valve using
nitrogen Check Valve PT

PT

Relief Skid N2 N2
Source FI

Return
Post-Test Bleed
Rotameter Open Vent

valves was to apply layer of protection analysis (LOPA) credit to these pressure. The nitrogen skid and connected equipment was protected
valves for reverse flow scenarios. An added benefit of testing the by a PSV skid designed to protect the equipment involved in this pro-
check valves was to better understand their performance and reliabil- cedure. The check valve piping assembly was fitted with ports to
ity and to quantitatively assess their performance over time through allow for pressuring the downstream side of the valve and relieving
leakage testing. the upstream side of the valve through the return line. Pressure trans-
mitters were available on both the downstream and upstream side of
the check valve(s). The pressure transmitters were reconfigured dur-
7.2 | Description of configuration ing the testing period to allow for gathering of data at very short time
intervals. The leakage could be measured one of two ways: (a) Using a
Five check valves, two different types, were installed and tested in rotometer in the outlet flow path from the upstream side of the check
two different process locations. One location involved testing a single valve to an atmospheric vent location or (b) Closing the outlet path
axial flow check valve in a higher pressure service downstream of a and monitoring the pressure transmitter on the upstream side of the
gas compressor (See Figure 3). This check valve was in series with an check valve for pressure build-up within the closed upstream chamber
existing wafer check valve on the discharge of the compressor setting. using gas laws to calculate the leakage rate associated with the pres-
Procedures were envisioned to test the existing check valve using sure build-up. Plan a would be used initially. And Plan b would be used
process pressure. Testing of both the new and existing check valves in the event that the rotometer method was unsuccessful or not
would allow for crediting as dual check valves rather than a single definitive. The check valve configuration was in an enclosed module,
check valve. Testing procedures for the existing check valve are not but the nitrogen bottle source and the return line discharge from the
included in the scope of this discussion. The other location involved rotometer were both outdoors. Refer to Figure 3 for a simplified
two sets of dual check valves, an axial flow check valve in series with depiction of the testing configuration for Test 1: single axial check
a wafer check, on two parallel process trains in lower pressure gas ser- valve, higher pressure gas service downstream of compressor.
vice (See Figure 4). The testing protocol for dual valves vs the single valve was simi-
The check valves being tested were purchased and installed in lar. The primary difference was that having two valves added com-
the same year. The testing procedure was implemented at the time of plexity and steps to the configuration and procedure. Additionally, in
installation and commissioning to serve as baseline performance test- this particular configuration, isolation of the location was attempted
ing. Follow-up testing was planned at intervals consistent with the via double block and bleed rather than using blinds. Because of the
company LOPA protocol based on the credit that was intended for number of testing points involved and a desire to minimize cost of
these check valves. Testing would occur during planned and sched- additional pressure transmitters (PTs), a tubing manifold and valving
uled shut-down opportunities, either full or partial plant outages. was designed, built, and installed that would allow for usage of certain
The testing procedure involved connecting an external pressure pressure transmitters in more than one testing configuration. How-
source to the system, nitrogen bottles, and supplying pressure at pre- ever, achieving a successful leak test of this manifold proved difficult.
scribed levels to the downstream side of the check valves. The nitro- Successful leak tests were also required of double block and bleed
gen bottles were fitted with a regulator to supply the appropriate valving to the process. Refer to Figure 4 for a simplified depiction of
14 of 19 OLSEN

PT2
PT2 PT2
PT2
PT PT3
PT3 PT1 PT
A B
CHK-01 CHK-02

PT

Relief Skid
V-1

N2 N2
Chk-01 FI

Chk-02
Source Open Vent
Return
Rotameter
Post-Test Bleed

FIGURE 4 Test 2 configuration. Test dual check valves, axial and wafer, using nitrogen one at a time

the testing configuration for Test 2: dual check valves (axial and wafer 521 (2014)3 “severe” failure leakage rates are approached. This situa-
checks), lower pressure gas service. tion is implied given that the failure mode is described as “severe”
with leakage defined as “significant” in API 521 (2014).3 The reference
to “slight” leakage in CCPS IE/IPL3 Section 5.2.2.2.2 Case 1 is in error.
8 | D E F I N I N G M A X I M U M A L L O W A B LE Leakage rates within the vicinity of these “severe” failure values may
SEAT LEAKAGE THRESHOLDS be unlikely in many check valve configurations as the leakage may be
either far below these values (i.e., “normal” leakage potentially repre-
So what level of leakage is the threshold for “acceptable?” In order to sented by values much closer to new valve leakage) or above these
answer this question, the hazard scenario including consequence values (i.e., complete failure). In order to experience “severe” leakage,
severity must be understood. The engineer should recognize the con- failure modes would likely involve significant debris or a foreign object
cept of process safety time with respect to progression towards the preventing closure or may be caused by freeze up such as from Joule-
hazardous event. In selecting the threshold for allowable leakage also Thompson flashing. These failures modes could result in fixing the
called maximum allowable leakage rate (MALR), the engineer or other check valve element at a location off but somewhat near the seat.
personnel who set up the testing procedure should look at several dif- The specific technical origin of these API 521 (2014)3 leakage
ferent points of reference for expected and/or acceptable leakage. rate assumptions for “severe” failure is not clear. Current API
In the sample test procedure described here, Test 1, the following 521 committee members are only aware that the values were
thresholds were identified in the procedure as points of reference. adopted from a member company participating on the committee. It
Rates shown in Table 5 are based on standards for new check valve is possible that due to lack of leakage rate data, very high leakage
allowable seat leakage, rotometer size selected, physical constraints of rate assumptions were chosen using an abundance of caution. No
the equipment included in the test configuration, API 521 (2014)3 data is known that supports the assumption that check valves com-
guidance for “severe” check valve failure, and normal as well as monly fail in a manner that results in these high leakage rates. Also,
expected maximum forward flowrates. A pass/fail limit was selected no data is provided that supports the assumption that “severe” fail-
based on review of these rates. ure is more likely than “complete” failure.
Review of this range of rates highlights a key point. The leakage Since little to no data is available for typical “normal” check valve
rates associated with API 521 (2014)3 “severe leakage” guidance - also leakage rates, establishing an appropriate initial limit is challenging.
referenced in CCPS IE/IPL2 Section 5.2.2.2.2 Case 1—are four to five However, as was made clear by defining the various thresholds, a limit
orders of magnitude larger than the acceptable leakage rates for new well below the “severe” leakage rate referenced in API 521 (2014)3
check valves prescribed by numerous standards for seat leakage. The and at or near new valve allowable leakage rates is an appropriate
check valve has failed from a reliability standpoint long before the API limit to assure that the check valve is performing its function as
OLSEN 15 of 19

TABLE 5 Test 1 flowrates and leakage rates range

intended. Impacts to upstream equipment including pressure build-up 2 Energy isolation and safe-out: Performed as per standard energy
in lower rated low volume equipment and reverse flow impacts to control procedures.
rotating equipment should be considered when selecting the pass/fail 3 Equipment installed for the test:
threshold or MALR. a Nitrogen bottle skid
The MALR for Test 2 was defined based on similar considerations b Regulators, from nitrogen bottle skid
as with Test 1 configuration. c Pressure gauges at nitrogen bottle skid (scaled for each testing
procedure)
d Hose, 5000 psig, inspected and tested. Installed with appropri-
9 | T E S T M E T H O D A N D P R O C ED U R E ate safety devices and signage.
i Connecting supply side:
Preparation for and conducting the testing procedure involved several 1 N2 skid and PSV skid
steps. Below is a description of the primary steps involved in the test 2 PSV skid and check valve downstream tap
procedure for testing dual check valves. ii Connecting return (vent) side
1 Check valve upstream tap to rotometer (or alternatively,
1 Process hazard analysis (PHA for testing configuration) use SS tubing)
a Safeguards and IPLs were considered that remained in place for 2 Rotometer to atmospheric vent point
the entire system as well as safeguards and IPLs specific to the
testing equipment. e PSV skid, sized and with setpoint specific to each test configuration
i A PSV skid was designed and installed for use as an IPL f PTs/PITs on existing process piping (PITs preferred for onsite
for the equipment installed for the testing. pressure display.)
b Alarm setpoints were adjusted on pressure transmitters (PTs) i Reset appropriate alarm setpoints for the testing procedure
including on the upstream chamber to indicate build-up of pres- ii Broadcast of data reset from once every 5 minutes to
sure through leakage. once per second for this testing procedure
c Board operators were engaged in monitoring potential pressure g Meter, rotometer sized at or above anticipated leakage rate
build-up. and MALR
16 of 19 OLSEN

4 Configuration approval and sign-off before each use. Assure engi- applying a pressure source for testing certain check valves particularly
neer sign-off of current job package. on parallel equipment. In this case, quantification of the leakage rate
5 Verify initial positions: valve positions, blind positions, as required could be accomplished by one of several methods. Pressure build-up
by procedure analysis using SCADA systems may be most convenient if upstream
6 Pre-start-up safety review (PSSR): Verify as-installed equipment isolation is available. Acoustic monitoring equipment should also be
and initial valve/blind positions. Visual inspection and checklist considered as a simple alternative which allows for assessing seat
review. leakage rates and check valve performance. Given the nature of check
7 Perform leak testing on installed testing equipment (peripheral valve operation, some type of shut-down, whether full plant turn-
equipment). around, unit outage, train outage, or parallel equipment shut-down
8 Perform leak testing of downstream double block and bleed valves. will be required to assess check valve performance.
If leakage found, safe-out equipment and roll downstream blind. Although not quantitative, diagnostic, or predictive, when check val-
9 Perform leak testing of upstream block valve (by connecting pres- ves are engaged, there may be opportunities to confirm gross closure of
sure source between upstream check valve (CHK-01) and the check valve by operators or mechanics. Operations or maintenance
upstream block valve. If leakage found, safe-out equipment and personnel may have some indication of closure of check valves based
roll upstream blind between vessel and first check valve. audible sound and on response of equipment that cannot tolerate back
10 Prepare for leak test of downstream check valve (CHK-02). Oper- spinning. Equipment that cannot tolerate reverse flow, including certain
ator resets Alerts and Alarms on PTs. Flowmeter is connected to rotating equipment, may be immediately or catastrophically damaged by
check valve upstream connection and vents to atmosphere. Pres- incidences of check valve failure and loss of containment may occur.
sure source is connected to check valve downstream connection. Additionally, check valve failure on a spared centrifugal pump may be
11 Assure PTs at zero pressure reading. Note time for start of test. detected following start-up of a parallel pump when both discharge pres-
12 Perform leak testing as per procedure on check valve (CHK-02). Set sure and flowrate measured on a downstream common line are lower
pressure regulator at prescribed test pressure and pressure up to test than expected.
pressure on the downstream side of downstream check valve.
13 Note and record pressure reading. (PT data available for further
analysis at a later date.) Also note and record pressure readings 11 | K E Y TA K E - A W A Y S A N D F I N D I N G S
downstream and/or upstream of block valves to assure no leakage. FR O M C H E C K V A L V E L E A K A G E TE S T
14 Record flowmeter flow rate data.
15 Review acceptance criteria. If below acceptable leakage rate Because the testing involved only a single instance of testing five sep-
(including below measurable rate), conclude test. If below measur- arate new check valves, the data from the testing is not presumed to
able rate, may consider either using a rotometer with a smaller represent a wider population of in-service check valves. Because of
scale or the pressure build-up method to quantify rate. this fact and because much of the specific data from the testing is not
16 Wait and take another data set after specified time interval. available for sharing, the results of the check valve testing will be dis-
17 Following conclusion of test, de-pressure the downstream and cussed in general terms. Descriptions of results and findings from the
upstream side of the check valve through bleed line and testing are the recollections of the author.
rotometer vent line. (Note: The rotometer should register instan-
taneous flow even if no flow was detected during the check valve 1 The testing procedure worked relatively well in both instances. The
leak test.) testing of the single valve (higher pressure) was more easily accom-
18 Disconnect testing equipment. plished due to the simplicity of the testing procedure. Only a single
19 (If dual check valves being tested, repeat procedure for upstream valve was being tested and isolation was accomplished with blinds
check valve (CHK-01).) rather than double block and bleed valving.
20 Return process equipment to original setpoints and data broad- 2 In the case of the single axial check valve test (higher pressure
cast frequencies. application) isolated by spectacle blinds:
21 Reverse energy isolation procedures. a The rotometer was used for measuring the leakage rate through
22 Perform PSSR and restart process as per normal start-up procedures. the check valve.
b Leakage rates were found to be below the scale of the
rotometer, therefore less than 5.9 SCFH.
1 0 | A L T E R N A T I V E TE S T I N G A N D i This rate of leakage was well within the acceptable range
A SS ES SM E N T OP T I ON S for new check valve allowable leakage criteria, so no fur-
ther investigation or calculation of actual leakage rates
The procedure described represents a direct testing method using an was pursued.
external pressure source. In addition to this testing method, other ii During relief of the trapped pressure from the down-
alternatives should also be considered as appropriate. In some stream system through the rotometer following the test,
instances, use of existing process pressure may be a viable manner for the functionality of the rotometer was confirmed.
OLSEN 17 of 19

3 In the case of the dual check valves, one axial check valve and one b API 521 (2014)3 does not allow for crediting a single check valve
wafer check valve on each train (lower pressure application) iso- with any reduction in flow rate to the upstream PSV, even if the
lated by block valves: check valve is inspected and maintained. This assumption equa-
a Initially achieving a good leak test was challenging primarily due tes to a PFD for a single check valve of 1, or RRF = 1. No data is
to the tubing manifold planned for use. Usage of the tubing man- given to support this assessment of PFD = 1. (Note: API
ifold was abandoned in preference to direct connections (and re- 521 (2008)11 did acknowledge the potential for a single check
positioning of PTs). valve to limit reverse flow when the check valve was appropri-
b Using valving to isolate the test section from the process also ately inspected and maintained.)
required additional steps in the leak test procedure. Leak testing c CCPS IE/IPL2 references API 521 (2008)11 language in 5.2.2.2.2
of block valves was successfully accomplished. No blinds were Case 1 which refers to a “series check valves,” but also refer-
rolled. ences a single check valve in the Case 1 description.
c Check valve leakage rates were determined using PTs and pres-
sure build-up calculations rather than with the rotometer. The discrepancies in CCPS IE/IPL2 and API 521 (2014)3 regarding
d Leakage rates were calculated for both the axial check valve(s) check valve failure likelihood should be reconciled. Check valve proba-
and the wafer check valve(s). Both valves passed the test. Leak- bility of failure on demand (PFD) by valve type, size, and service
age rates were below the threshold stipulated in the procedure. should be based on inspection, testing, maintenance, operational, and
e Wafer check leakage was higher than axial check valve leakage. failure history data.
Both were within established limits. Check valve leakage rates should be determined based on testing
4 As anticipated, selecting rotometers that meet the actual range of and should be associated with valve type, failure mode, size, and ser-
leakage was challenging. However, the rotometers selected were vice. Check valve leakage rates for “severe” and “normal” failure
within the “ballpark” of leakage ranges. With more time and test should be better defined and consistently applied.
data, rotometers could be effectively “dialed in” that would be in
the range of the actual leakage. 2 Leakage rate assumption(s) cited by API 521 (2008)11 and repli-
5 Pressure build-up testing proved to be a viable and simple alterna- cated in CCPS IE/IPL2 Section 5.2.2.2.2 Case 1 represent “severe”,
tive method to determine leakage rates. significant, or high rates of leakage. The check valve has failed from
6 System and block valve leak tests must be conducted in advance of a reliability standpoint at a much lower rate of leakage (four to five
check valve leak testing to assure that isolation from other poten- orders of magnitude less leakage). The reference to “slight” leakage
tial leakage points have been successfully achieved. in CCPS IE/IPL2 5.2.2.2.2 Case 1 should be corrected.
7 The testing requires either a full or partial outage to accomplish.
8 As with any new procedure, appropriate process hazard review of Little data is currently available regarding typical or “normal” leak-
the connected equipment and procedure is required as well as pre- age rates for check valves that have been in operation. Likewise, little
start-up safety review. A personnel job safety analysis (JSA) must data is available regarding check valve failure modes and frequencies.
be conducted as well. This void was a driver for designing and performing these field tests
and for writing this paper.

12 | KEY TAKE-AWAYS FROM REVIEW 3 Numerical estimates for “normal” leakage rates are not offered by
OF REFERENCE MATERIAL API 5213,11 or by CCPS IE/IPL.2 API 5213,11 indicates it is the
responsibility of the user to determine normal leakage rates.
As noted within this paper, several discrepancies were found in refer- 4 “Severe” leakage rates as defined in API 521 (2014)3 exceed new
ence material. The following inconsistencies should be resolved within check valve performance requirements by four to five orders of
reference documentation in order to prevent confusion for users and magnitude. Integrity management programs should strive to
practitioners. achieve performance comparable to new valve standards.
5 In spite of the high rates associated with the “severe” leakage cal-
1 Crediting a Single Check Valve as IPL and using an appropri- culation in API 521 (2014),3 the assumption that check valves fail in
ate PFD: this manner may not be accurate or conservative. Complete failure
a CCPS IE/IPL2 lists a single check valve as a candidate IPL in of the check valve may be a more likely failure mode.
Chapter 5 and provides a generic PFD value of 0.1 (RRF = 10) 6 Testing of check valves that have been in service is a prerequisite
based on data and calculations provided in Appendix to not only understanding and improving check valve performance,
D. Special considerations are listed that must be satisfied. Ini- but also to crediting check valves as IPLs or safeguards.
tiating event frequencies (IEFs) are also given for check valves 7 Existing industry equipment reliability databases such as CCPS Pro-
in high demand modes of 0.1 and 0.01 for single and dual cess Equipment Reliability Database (PERD) have very little data
check valves, respectively, in Chapter 4 based on the same available on check valve performance including failure modes and
data in Appendix D. frequencies. PERD offers guidance for data collection in a manner
18 of 19 OLSEN

that supports reliability analysis. Usage of PERD and other industry Stewart, exida; and Jennifer Mize, Eastman for their review, com-
resources is recommended. ments, and contributions on this paper and presentation. The author
also wishes to thank American National Standards Institute (ANSI) for
providing permission on behalf of International Organization for Stan-
13 | CO NC LUSIO NS dardization (ISO) to reproduce Tbl 4 from page 11 of ISO 5208:2015
and Manufacturers Standardization Society of the Valve and Fittings
Check valves perform an important and often critical function in Industry (MSS) for providing permission to reproduce Tbl 4 from page
process equipment. Therefore, check valve performance and reli- 4 of MSS SP-61-2013.
ability should be a foremost concern for operating companies.
Check valve performance and reliability should be managed OR CID
through maintenance programs that include inspection, testing,
and repair of check valves. Check valves have historically not Jody E. Olsen https://orcid.org/0000-0003-2610-2534

received proper attention in inspection and maintenance programs.

RE FE RE NCE S
And testing programs to quantify leakage have rarely been instituted.
1. A. Baldas and F. Carithers. Check Valves and Risk Assessment: Should
This situation persists in spite of regulatory and industry stipulated
You Take Credit? Available at https://bluefieldsafety.com/2017/09/
requirements for maintaining asset integrity. Regulation, such as
check-valves-and-risk-assessment-should-you-take-credit/, Accessed
OSHA PSM,7 explicitly references piping and piping components, such on December 23, 2018.
as valves, as process equipment covered by the regulation. 2. CCPS. Guidelines for Initiating Events and Independent Protection Layers
More and better data on check valve failure modes, failure likeli- in Layer of Protection Analysis. New York, NY: Center for Chemical
Process Safety, American Institute of Chemical Engineers; 2015.
hood, and performance (i.e., leakage) can lead to better design by man-
3. American Petroleum Institute (API). API Standard 521: Pressure-
ufacturers, better selection for the given application by operators,20 relieving and Depressuring Systems. 6th ed. Washington, D.C: API Pub-
better piping configuration design, and better performance of the check lishing; 2014.
valves. Currently, discrepancies exist in some core industry reference 4. C. Patschke, “Solving the Mysteries of Critical Check Valves,” Available
at: https://inspectioneering.com/journal/2018-06-28/7745/solving-the-
material regarding characterization of check valve probability of failure
mysteries-of-critical-check-valves, Accessed January 10, 2019.
on demand (PFD). Those discrepancies should be reconciled. (See Sec- 5. CCPS. Layer of Protection Analysis: Simplified Process Risk Assessment.
tion 12 for descriptions of these discrepancies.) Center for Chemical Process Safety, American Institute of Chemical
Little data is available regarding check valve performance and fail- Engineers: New York, NY; 2001.
6. CCPS. Guidelines for Hazard Evaluation Procedures. 3rd ed. New York,
ure modes by type and in given service applications. And little pro-
NY: Center for Chemical Process Safety, American Institute of Chemi-
gress has been made on gathering more modern data on check valve cal Engineers, John Wiley and Sons Inc; 2008.
performance. The author encourages operating companies to consider 7. U.S Department of Labor Occupational Safety and Health Administra-
testing and inspection of check valves in critical service as part of their tion (OSHA). Process safety management of highly hazardous
chemicals, 29 CFR. 1910.119, Jun 1, 1992, as amended.
mechanical integrity program, particularly when designing and
8. CCPS. Guidelines for Risk Based Process Safety. Hoboken, New Jersey:
installing new installations where appropriate isolation and testing Center for Chemical Process Safety, American Institute of Chemical
ports may be included in the piping arrangement. The author further Engineers; 2007.
encourages the sharing of check valve inspection and testing data 9. American Petroleum Institute (API). API Standard 570: Piping Inspection
Code: In-service Inspection, Rating, Repair, and Alteration of Piping Systems.
more broadly, such as through the CCPS Process Equipment Reliabil-
4th ed. Washington, D.C: API Publishing; 2016 (Addendum 1, 2017).
ity Database (PERD), in order to achieve industry-wide progress 10. K. Modi. Check Valves as Safeguards: Friend or Foe in Process Hazard
towards improved check valve reliability. Better practices, better data, Analysis. Available at http://jclrisk.com/wp-content/uploads/sites/3/
and better understanding of performance can lead to changing expec- 2017/05/Check-Valves-as-Safeguards-JCL-Risk-Author-Kay-Modi-
tations regarding check valve performance. April-25-2017-at-SynGas-Conf.pdf, Accessed on December 23, 2018.
11. American Petroleum Institute (API). API Standard 521: Pressure-
When check valves are thoroughly and routinely inspected,
relieving and Depressuring Systems. 5th ed. API Publishing:
tested, and maintained, PHA and/or LOPA credit for the check Washington, D.C; 2008.
valve(s) as a safeguard or IPL, respectively, may be achieved. As 12. American Petroleum Institute (API). ANSI/API Standard 598: Valve Inspec-
always, operators must consider applying this credit with a thorough tion and Testing. 10th ed. Washington, D.C: API Publishing; 2016.
13. International Standard Organization (ISO). ISO 5208: Industrial
understanding of the check valve failure modes and their ability to
valves—Pressure testing of metallic valves. 4th ed. Switzerland: ISO
prevent the hazard consequence from being realized. Publishing; 2015.
14. Manufacturers Standardization Society of the Valve and Fittings
ACKNOWLEDGMENTS Industry, Inc. (MSS). Standard MSS SP-61-2013: Pressure Testing of
The author wishes to thank Robert Coover, Praxair, Inc.; Fred Valves. Vienna, Virginia: MSS Publishing; 2013.
15. American Petroleum Institute (API). ANSI/API Standard 6D: Specifica-
Henselwood, NOVA Chemicals Corporation; Hal Thomas, exida; Jan
tion for Pipeline and Piping Valves. 24rd ed. Washington, D.C: API Pub-
Hayes, RMIT University; Barry Bennett, Stepan Company; Angela lishing; 2014.
Summers, SIS-Tech Solutions; Forrest Carithers, Wyatt Field Service 16. American Petroleum Institute (API). ANSI/API Standard 6D: Specification
Company; Bart Lewis, retired Maintenance Planner/Mechanic; Loren for Pipeline Valves. 23rd ed. Washington, D.C: API Publishing; 2008.
OLSEN 19 of 19

17. European Standard. EN-12266-1:2012: Industrial Valves – Testing of SUPPORTING INF ORMATION
Metallic Valves – Part 1: Pressure Tests, Test Procedures and Acceptance Additional supporting information may be found online in the
Criteria – Mandatory Requirements. Brussels: European Committee for
Supporting Information section at the end of this article.
Standardization; 2012.
18. British Standards Institute (BSI). BS 6755-1: 1986. Testing of valves—
Part 1: Specification for production pressure testing requirements.
1986. How to cite this article: Olsen JE. Crediting check valves as
19. American National Standard Institute (ANSI). ANSI/FCI 70-2-2013 Con- IPLs? Testing protocol to better understand check valve
trol Valve Seat Leakage. Cleveland, OH: Fluids Controls Institute; 2013. reliability. Proc Safety Prog. 2020;39:e12153. https://doi.org/
20. M. Johnson, “The Misunderstood Check Valve,” Available at https://
10.1002/prs.12153
www.dft-valves.com/pdf/literature/general-information/
CheckValveBacktoBasicsArticle.pdf, Accessed on December 23, 2018.