By Jekel, T. B., Claas, M., and Reindl, D. T., Industrial Refrigeration Consortium, University of Wisconsin-Madison

Editor’s Note

The SRVcalc project is a twophase research initiative to develop a method and tool for determining whether relief valve replacement intervals can be extended or shortened depending on the conditions in which they are installed.

The first phase of the project included the development of a physical testing apparatus and test methodology to generate data on real-life relief valves that can feed into an appropriate statistical analysis.

The second phase of the project developed the appropriate statistical modeling using a technique known as “Weibull Analysis,” to predict the life performance of tested relief valves. Part of the second phase of the project was to develop and make available a software program, called SRVcalc, which can be used as a tool to complete Weibull Analysis for a set of data.

As discussed in the cover story of this issue of the Condenser, “Using Weibull Analysis to Predict Valve Replacement Intervals,” IIAR is preparing to release software to make the analysis tool available for wider use.

Reprinted in the following pages of this issue of the Condenser, is an account of the findings of the University of Wisconsin Madison’s Industrial Refrigeration Consortium concerning the first phase of the SRVcalc project, which were delivered at IIAR’s 2011 conference in San Diego.

The results of the project included a relief valve test rig design and corresponding test procedures suitable for data collection by post-mortem testing of relief valves.

In 2007, IIAR revised its recommended practice for replacing pressure relief valves on industrial ammonia refrigeration systems (Section 6.6.3 of Bulletin 110). In addition to the prescriptive five-year relief valve replacement interval, the revisions to Bulletin 110 added an alternative replacement method based on an evaluation of in-service relief valve life using appropriate testing and data analysis methods.

This paper describes the results of a research project that aimed to validate guidelines for the post-mortem testing of relief valves. The purpose of the data collected by post-mortem testing is intended to support the alternative path to determine the service life of relief valves following their removal from the system prior to their disposal (i.e. post-mortem). The testing procedure and data gathering methods described in this paper are intended for relief valves that have not discharged during their in-service life. A test rig suitable for post-mortem testing of relief valves was designed, constructed, and proof-tested. The function of the experimental rig was established by testing a range of alternative relief valves that included high and low set pressures; high and low capacities; as well as both new and used relief valves. The draft test procedure was modified using the information gathered during the rig proof-test. The results of this project include a relief valve test rig design and corresponding test procedures suitable for data collection by postmortem testing of relief valves.

INTRODUCTION

Pressure relief valves are engineered safety devices designed to automatically function and relieve vapor caused by overpressure as a means of preventing the catastrophic failure of vessels and other equipment. Paragraph UG-125 of the ASME Boiler and Pressure Vessel Code (Section VIII Div. 1) prescribes basic requirements for pressure relief protection applied to vessels and other equipment built in accordance with the Code (ASME 2010).

A number of factors are essential for successful safety relief systems. First, appropriate relief valve selections are required for given equipment and relief scenarios. ASHRAE (2010) and IIAR (2008) provide prescriptive pressure relief valve selection requirements for vessels and positive displacement compressors. For other equipment, design engineers need to prepare appropriate analyses to properly size relief devices for envisioned scenarios per UG-133 (ASME 2010).

The most common basis used for sizing of pressure relief valves for refrigerant-containing vessels assumes a radiant heat load from an external fire condition (Fenton and Richards, 2002). Reindl and Jekel (2009) present principles of relief device capacity determination for other types of equipment. Applied relief devices not only require sufficient capacity (vapor mass flow rate) but also they must have a set pressure no higher than the maximum allowable working pressure (MAWP) for the vessel or equipment they are protecting per UG-134 (ASME 2010). Suitably engineered relief valve inlet and outlet piping is another necessary condition for proper relief device operability. Reindl and Jekel (2006) provide guidance on the principles and practices for properly engineering safety relief systems. Finally, the relief devices themselves must remain operational while installed throughout their life.

A number of factors can, potentially, influence the life of a pressure relief valve but corrosion on the outlet portion of the relief device tends to dominate (Gross, et al. 2006). Section 9.4.9 of ASHRAE Standard 15 (ASHRAE 2010) requires that suitable materials be used for the construction of relief valves to resist corrosion or other chemical action that may be caused by refrigerant exposure.

9.4.9 The seats and discs of pressure-relief devices shall be constructed of suitable material to resist refrigerant corrosion or other chemical action caused by the refrigerant. Seats or discs of cast iron shall not be used. Seats and discs shall be limited in distortion, by pressure or other cause, to a set pressure change of not more than 5% in a span of five years.

In addition, suitable materials for seats and disks must be used such that their set pressure will not change by more than 5% in a span of five years. Traditionally, relief valves were recommended for replacement at least every five years (IIAR Bulletin 110, 1993). In part, Section 6.6.3 of Bulletin 110 stated:

At least every five years pressure relief valves (or cartridges) shall be removed and replaced with new or with overhauled and recalibrated valves (or cartridges).

In 1992, OSHA issued its Process Safety Management (PSM) standard (1910.119) and paragraph (j) of that standard obligated end-users of covered processes, such as ammonia refrigeration systems with more than 10,000 lb of refrigerant inventory, to perform inspections and tests to ensure the ongoing mechanical integrity of their systems and equipment. Specific requirements for inspections and tests are not prescriptively identified in the PSM standard; rather, the details of required tests are based on industry-recommended practices with modifications based on site-specific experience.

In 1997, OSHA issued a series of citations to the Armour Swift-Eckrich plant in Kansas City, MO as a result of deficiencies found in the facilities’ PSM program during an inspection. As part of a settlement following the OSHA inspection, Swift-Eckrich agreed to abate a number of items discovered during the inspection (OSHA 1997). A portion of the abatement action focused on various aspects of the pressure relief protection for its vessels and other equipment such as compressors. In these parts of the settlement, Swift-Eckrich agreed to address the ongoing function of pressure relief valves as a means of assuring their function as part of the facilities’ mechanical integrity program. The following are specific references to the PSM standard and associated activities that Swift-Eckrich agreed to:

Item 28: 1910.119(j)(4)(i)

…

c. The company agrees to perform inspection and testing on pressure relief valves.

Item 29: 1910.119(j)(4)(ii)

…

d. The company agrees that the Mechanical Integrity Program will include procedures for the inspection and testing of pressure relief valves and vent systems that follow good engineering practices. Good engineering practices for the testing and inspection of pressure relief valves and vent systems include:

…

3. A procedure that verifies replacement every five years is an adequate interval. To assure pressure relief valves are still operational when replaced, they must be visually inspected for corrosion or fouling when removed from service, pop tested in the “as received” condition by an authorized testing laboratory, and the testing and inspection or replacement interval reduced, if necessary.

Under item 28 within the agreement, Swift-Eckrich agreed to perform inspections and tests on their pressure relief valves. Under item 29, Swift-Eckrich agreed to develop procedures for the inspections and testing of pressure relief valves that follow good engineering practices. The good engineering practices identified included conducing visual inspection of pressure relief valves for corrosion or fouling as well as pop testing when removed from service. At the time of the settlement agreement, no Recognized as Generally Accepted Good Engineering Practice (RAGAGEP) was available to guide the performance-based service life for pressure relief valves.

In 2001, IIAR revised section 6.6.3 in Bulletin 110 to state:

When a component reliability program is in place to verify relief valve functionality and longevity by history, testing, disassembly and inspection, and periodic statistical review of these activities, relief valves may be replaced at any interval justified by the findings of such a program. In the absence of such a program, each relief valve shall be replaced at the frequency recommended by the relief valve manufacturer. In the absence of both a component reliability program and manufacturers’ recommendations, relief valves shall be replaced every five years if not indicated earlier at annual inspection.

The revision to IIAR Bulletin 110 gave end-users another option to the most common industry practice – the prescriptive replacement of pressure relief valves on a five-year interval. Although the revised IIAR Bulletin 110 guidance established a performance-based interval for replacement, less clear were the specific means for methods of test as part of data collection on relief valves removed from service for replacement as well as the statistical analysis methods to guide altering the interval for replacement.

In 2006, IIAR formed a task force to look consider further modifications to section 6.6.3 of Bulletin 110. The task force considered relief valve inspection and test criteria and benchmarked with related industries. The task force also took action to further revise section 6.6.3 of Bulletin 110. In 2007, IIAR approved revisions to section 6.6.3 of Bulletin 110 prepared by the task force. An excerpt of Section 6.6.3 as-revised states:

Pressure relief devices shall be replaced or recertified in accordance with one of these three options:

1. Every five (5) years from the date of installation. IIAR originally recommended (in 1978) that pressure relief valves be replaced every five years from the date of installation. This recommendation represents good engineering practice considering the design and performance of pressure relief devices; or
2. An alternative to the prescriptive replacement interval, i.e., five years, can be developed based on documented in-service relief valve life for specific applications using industry accepted good practices of relief valve evaluation; or
3. The manufacturer’s recommendations on replacement frequency of pressure relief devices shall be followed.

   Exception: Relief devices discharging into another part of the closedloop refrigeration system are not subject to the relief valve replacement practices.

The task force continued to develop procedures for post-mortem testing of pressure relief valves. This effort was eventually transferred to the IIAR Research Committee which facilitated in the drafting of testing procedures. At the end of 2008, the IIAR Research Committee completed its development of procedures for bench testing relief valves. The test procedure document included a description of the test equipment and procedures that should be followed for the post-mortem testing of pressure relief valves; however, the test rig documented and the associated testing procedures had not actually been validated by practice. In 2009, the IIAR Ammonia Refrigeration Foundation (ARF) funded this project to validate and revise, as needed, the test rig and corresponding procedures for post-mortem testing of pressure relief valves. This paper summarizes the results and findings of this project.

RELIEF VALVE TEST PROCEDURE

The original objective of the IIAR relief valve bench test procedure was to “to quantitatively determine the opening pressure and qualitatively verify operation (i.e. lift or flow) of a reclosing vapor relief valve after removal from service.” The procedure was intended for bench testing reclosing safety relief valves designed to discharge to atmosphere and the procedure allowed the use of compressed air (or nitrogen) as the test medium. The procedure was not intended to test:

• Non-reclosing relief devices(i.e. rupture discs),
• Internal relief valves( i.e valves with  superimposed back-pressures greater than 1 atmosphere),
• Hydrostatic relief valves (i.e. liquid relief), or
• A test medium other tan compressed air or nitrogen (e.g. steam, refrigerant, or incompressible liquids).

Figure 1 shows a schematic of the pressure relief valve test rig as originally proposed. Because a sufficient source of high pressure air needed to test relief valves with high set pressures (e.g. 300 psig) is not readily available in most facilities, the test rig relied on the use of high pressure gas cylinders to supply air or nitrogen for testing. From high pressure compressed gas cylinders, the pressure is regulated down as it flows to an accumulator. The purpose of the accumulator is to buffer the pressure and flow of gas to the inlet of the relief valve being tested. Service valves are located throughout the test rig to allow isolation of the compressed gas cylinders and relief valves for change out. Pressure gauges are used to provide the operator of the test rig accurate information in order to document the relief valve start-to-leak and pop pressure.

Note that it is important to consider and check the capacity of the safety relief valve protecting the accumulator vessel relative to the capacity of the pressure regulator feeding gas from the cylinder into the vessel to prevent the possible over pressurization in the event of the regulator failing open.

The present research project intended to answer a number of questions related to the originally developed test rig and procedure including:

• Is a vessel or accumulator on the inlet of the test relief valve required? If so, is the recommended 12” x 48” adequate?
• What rate of pressure rise result in accurate opening pressure determination?
• Is the bubble method for start to leak reliable and repeatable?
• Is a pop test visible? If so, what safety precautions should be written into the procedure and integrated into the test stand?
• Should be a procedure to be alerted based on the rated capacity or set pressure of the valve tested?
• Can the equipment for the test be readily acquired at a reasonable cost?

The remainder of this paper describes the techniques used to validate the test method, observations and findings associated with testing a range of relief valves using the test rig, and conclusions.

TEST RIG DESIGN

The test rig design was modified from the one proposed in the original procedure drafted by the IIAR Research Committee in order to reduce the amount of piping, streamline the fabrication, and facilitate the changing of relief valves during sequences involving multiple valve tests. In addition to the cost benefits of streamlining the piping, there is an added benefit associated with the reduction of inlet pressure drop during the pop test of relief valves. The number of connections on the inlet piping for accommodating the vent and pressure gauges was also reduced from the original test rig concept design. The simplifications made to the inlet piping however requires that the pressure gauge be changed when testing valves with different pressure set points, but a higher value was placed on the simplification to the piping. Both ranges of pressure gauges could be incorporated easily if desired.

The accumulator vessel is rated in accordance with the ASME Boiler & Pressure Vessel code (ASME 2007) with a Maximum Allowable Working Pressure (MAWP) of 500 psig and a standard Minimum Design Metal Temperature (MDMT) of -20°F. All accumulator vessel connection sizes were oversized to allow flexibility. A ¾ inch connection on the bottom of the vessel serves as a vent/drain; a 1-½ inch connection on the top is for the test relief valve connection; two 1 inch connections on the top of the of the vessel are needed for air inlet from the pressure source supply and for a relief valve to protect the accumulator; and ¾ inch connection allows for the installation of a dedicated pressure gauge (0-600 psig range) for the accumulator vessel itself.

Different inlet connection sizes of the relief valves are accommodated by using a Class 600 flange (ASME/ ANSI B16.5), fitting, and nipple for each size (½”, ¾”, 1, 1-¼”) shown in Figure 2. This arrangement allows the installation of varying sizes of relief valves without the complexity of an elaborate manifold. With a proper experimental plan, this arrangement can decrease the time associated with changing relief valves.

Ball valves are used throughout the test rig for isolation for two reasons. First, the quarter-turn ball valves provide quick actuation. Second, the valves are full area which minimizes gas pressure drop. All valves were pressure rated in excess of the 500 psig MAWP on the vessel. For the rig, 3-piece, carbon steel ball valves with threaded connections and a maximum pressure rating of 1,480 psi WOG were used in all cases except for the bleed valve on the test section which was a two-piece, brass ball valve with threaded connections rated at 600 psi WOG.

The pressure gauge was located on the experimental test section. Its installation at that location does not affect the reading since the pressure of interest is the valve starts-to-leak and/ or pops pressure, both of which are low flow conditions. As such, the pressure read off the gauge in this location is virtually identical to the relief device inlet pressure at the point of relief valve operation (i.e. either simmering or popping). The pressure gauge dedicated to the vessel is intended for monitoring the pressure in the accumulator vessel during periods where the test section itself may be isolated for relief valve change out.

Compressed air was chosen over nitrogen despite its lower price to avoid having to vent the outlet from the relief valve to an outdoor location for personnel safety. If nitrogen is used, or exhaust is required for other reasons, the vent piping should be adequately sized to insure that its presence does not affect the data. If nitrogen or another inert gas is used as the test medium indoors, an area monitor with low oxygen alarm set at 19.5% should be installed. The final proposed design is shown below in Figure 3 and a close-up of the test section is shown in Figure 4. A 3-D schematic of the test rig is shown in Figure 5 and a photo of the completed test rig is shown in Figure 6. The estimated cost of the rig is shown in Table 1. Note that fabrication with all screwed fittings and without the completely flanged test section and flanged inlet connection would lower the cost.

METHOD VALIDATION

With a rig designed and constructed, the project moved on to the test method validation phase. To validate the proposed test method, pressure relief valves with a range of capacities and set pressures were tested. Table 2 shows the combination of valve capacities and set pressures considered during the process of validating the test procedure. A check mark indicates that at least one pressure relief valve with that combination of capacity and set pressure was tested.

The relief valves tested included a mix of new and used (post-mortem) valves. We chose used relief valves from a variety of locations throughout the plants that provided them for the project. Used valves tested came from locations that included machinery room, condenser (outdoors), compressor, oil pot, and low temperature recirculator. The range of locations allowed us to test the “condition report” requirements of the procedure in order to see if the form included within the test procedure adequately captured the necessary information. Ultimately, we tested nine (9) used relief valves and five (5) new relief valves. Most of the used relief valves were in the middle range shown in Table 2 while the newly manufactured relief valves were chosen to reach the bounds for capacity and set pressure in Table 1.

Since the purpose of the tests was to validate the method, no archival data on the tested relief valves was generated. The outcome of this effort was focused on determining whether the test rig was adequate in size and function, and that the test procedure is capable of providing data that can support downstream statistical analysis of relief valve mechanical integrity (service life).

The pressure relief valves were put through the originally proposed procedure to see if the set pressure could reliably be determined with the start-to-leak test. Once the start-to-leak was done, the outlet tubing assembly (relief valve outlet connection size to ¼” connection reducer, ¼” hose barb and ¼” plastic tubing) was removed and the relief valve was popped with no outlet piping.

RELIEF VALVE TEST OBSERVATIONS

General observations

The experimental rig, as-constructed, performed well over the entire range of relief valve capacities and set pressures tested. Even with the compressed air supply regulator fully open, the relief valve inlet pressure rise was slow enough to provide the test rig operator an ample opportunity for reading the pressure gauge accurately. An obvious possible upgrade to the test rig would be to integrate a pressure transducer and computer data acquisition program, but that addition is not necessary to obtain good accuracy.

Start-to-leak tests

Start-to-leak tests initially looked like a viable and preferred option for determining the opening pressure of relief valves being tested. However, as we tested more relief valves, we found that not all pressure relief valves exhibit a simmer characteristic. In cases where a relief valve exhibited flow during a simmer condition, the pressure drop experienced through the outlet fitting on the pressure relief device for start to leak testing (¼ inch hose barb and tubing) was large enough to keep the relief valve from opening. During testing of valves with no simmer characteristic and the start-to-leak equipment installed, relief valve chatter occurred. Chatter is the rapid opening and reclosing of the relief valve which is undesirable for obtaining test data. We found that after a relief valve experienced chattering, its opening pressure during subsequent pop testing was susceptible to drift. In thinking about the start-to-leak test after this observation, the following question was posed “what constitutes verification of the relief valve’s function?” It was agreed that the opening of the relief valve (i.e. lift) was what verified function and that the start-to-leak test provided no additional useful information. Therefore, recommended eliminating the start-to-leak pressure test from the original procedure in favor of pop-testing only.

Pop tests

In response to the previous determination that opening the relief valve verifies its function, we performed pop tests of the relief valves tested. To perform pop tests, requirements for securing the test rig were emphasized and the use of proper personal protective equipment (PPE) for both eye and ear protection during testing. While no direct measurement of pressure relief valve lift is done, actuation of the relief valve is easily determined by the flow through the relief valve. The rig performed very well during these tests. The accumulator vessel provided a buffer for both better control of the pressure rise while approaching the popping pressure as well as adequate air volume to fully open the relief valves over the entire range of capacity.

An unintended side benefit of using the pop test is that the method allowed estimates of the relief valve’s closing pressure, or blowdown pressure to be easily measured. When the relief valve recloses, the pressure on the gauge shows the closing pressure directly. This pressure is an estimate of the closing pressure unless the compressed air cylinder is isolated from the accumulator vessel, because there is still a small inlet flow into the accumulator vessel occurs otherwise. However, the makeup air flow is small compared to even the smallest relief valve tested. Some words of caution are required here. First, the blowdown pressure does not have an impact on the success or failure of the relief valve actuation (it just is additional information). Second, the observed blowdown pressure with no outlet piping will be much greater (i.e. lower closing pressure) than for a relief valve installed with vent piping. This is occurs because the pressure that builds up in the vent piping during flow provides an additional closing force on the relief valve resulting in closing at a higher inlet pressure. This operating characteristic was validated during the course of several experimental tests.

Multiple tests on single relief valve

Multiple tests on some relief valves were also performed. After the performing multiple tests on a given valve, the question was posed “is any additional information gathered?” The answer was always no. It was never the intent of data gathered by this testing procedure to provide justification for elimination of the recommendation to replace a relief valve after it lifts. Truly no experiment can adequately justify that change. The purpose of the pop test is to determine whether or not the relief valve would have actuated during an overpressure event prior to the relief valve’s removal from the system. Multiple pop openings on valves removed from service do not provide additional insight on this determination. In fact, one might pose the scenario, where the pop pressure observed during a subsequent test varies significantly from the first pop. Changing pop pressure during multiple pop tests, it does not indicate relief valve failure. It is our view that multiple pop tests provides no additional value for post-mortem functional testing and, as such, was eliminated from the original test procedure.

CONCLUSIONS

The purpose of the project was to test and propose changes to the draft Bench Procedure for Post-Mortem Testing of Safety Relief Valves developed by the IIAR Research Committee. The main deliverable was to provide recommendations for changes to the procedure, test rig, and data collection sheets. This paper provides an overview of the project and the findings that were the background for the changes proposed.

The test rig was re-designed and performed flawlessly throughout the range of pressure relief valves considered. The procedure was altered to streamline the test by elimination of the start-to-leak test and multiple pop tests.

As a final note, the purpose of this bench test is to determine at what pressure the removed relief valve would have actuated. The next step is to use this data to determine whether or not the relief valve passed or failed. It is not the purpose of this test to recertify the tested relief valve for continued service. In other words, if a relief valve is tested using the equipment and methods described herein, they should not be placed back into service unless a qualified facility re-certified the relief valve.

Ultimately, the data generated from this test procedure can be used with a statistical analysis to justify changing the service life of the pressure relief valves in similar service to insure safe operation (i.e. the relief valve will actuate, that is, open, as expected during its service life).

ACKNOWLEDGEMENTS

This project was funded with funding from the Ammonia Refrigeration Foundation (ARF) with generous donations of equipment and expertise by the following companies:

• Isotherm, Inc. for donating the vessel
• Rohde Brothers, Inc. for donating piping, fittings, and welding service
• Hansen Technologies for donating pressure relief valves
• Refrigerating Specialities, a division of Parker for donating pressure relief valves
• Kraft Foods, Inc. for donating relief  valves for post-mortem testing
• Schoep’s Ice Cream for donating relief valves for post-mortem testing

Also thanks to the IIAR Research Committee for their oversight and guidance throughout this project.

• Table 1: Estimates test rig cost.
• Figure 1: Pressure relief valve bench test rig (as-originally proposed).
• Figure 2: Photos of the flanged relief valve inlet connections.
• Figure 3: Schematic of re-design of test rig section portion of the rig (dimensions shown are in inches). 
• Figure 4: Schematic of the test section portion of the rig (dimensions shown are in inches). 
•  Figure 5: Parts list and 3-d schematic of test rig.
• Figure 6: Photo of the completed test rig.

REFERENCES

ASHRAE, “Standard 15 – Safety Standard for Refrigerating Systems”, American Society of Heating, Refrigerating, and Air Conditioning Engineers, Atlanta, GA, (2010).

ASME, “Boiler and Pressure Vessel Code – Section VIII Division 1”, American Society of Mechanical Engineers, New York, NY, (2007) and (2010). ASME/ANSI B16.5, “Pipe Flanges and Flanged Fittings”, American Society of Mechanical Engineers, New York, NY, (1996).

Fenton, D. and Richards, W., Standard 15- 2001 User’s Manual, American Society of Mechanical Engineers, New York, NY, (2002).

Gross, R. E. and Harris, S. P., “Extending Pressure Relief Valve Inspection Intervals by Using Statistical Analysis of Proof Test Data”, Proceedings of ASME PVP 2006 / ICPVT-11 Conference, Pressure Vessel Technologies for the Global Community, Vancouver, BC, (2006).

IIAR, “Standard 2 – Equipment, Design, and Installation of Closed-Circuit Ammonia Mechanical Refrigerating Systems”, International Institute of Ammonia Refrigeration, Arlington, VA, (2008).

OSHA, “Informal Stipulation and Settlement Agreement for OSHA Swift-Eckrich, INC. d/b/a Armour Swift-Eckrich”, Occupational Safety and Health Administration, http://tiny.cc/p78uf, (1997).

Reindl, D. T. and Jekel, T. B., “Pressure Relief Device Capacity Determination”, ASHRAE Transactions, Vol. 115, No. 1, (2009).

Reindl, D. T. and Jekel, T. B., Engineering Safety Relief Systems, Industrial Refrigeration Consortium, University of Wisconsin-Madison, ( 2006).