Normally, valves are pressure tested by the factory to check for leaks before they’re shipped. The type and method of testing conforms to one or more of many different standards. The purpose of these tests and the expected results are often misunderstood and misapplied, resulting in unnecessary delays and unanticipated costs.
This article identifies the most common mistakes. Its primary purpose is to inform and guide end users and engineers that write valve specifications so those parties can avoid some of the mistakes.
Numerous valve standards define leak testing requirements. Often, the difference between them is unclear, which leads to confusion over which one to use and why. The place to begin is to understand the two types of leaks undergoing tests:
- External is leakage from pressure inside the valve out to the atmosphere. This is verified by the shell test, and no leakage is allowed at pressure equal to or greater than the valve rating.
- Internal is leakage around the valve seat or through the closure member when the valve is closed. Internal leakage is normally contained within the piping system. Such leakage can be external when a valve is used to dead-end a pipe. Internal leakage is checked with a seat leakage or a closure test. The allowable leakage rate depends on the valve test standard.
External and internal leak test methods are defined in the numerous standards, and there are too many to cover in a single article. Below, however, is a brief explanation of some of the more common standards:
- American Society of Mechanical Engineers (ASME) B16.34, Valves – Flanged, Threaded and Welding End: This is the most widely used valve design standard. It contains charts that identify the working pressures used in conjunction with other test standards, such as API 598. ASME B16.34 contains test methods, but does not define allowable seat leakage.
- American Petroleum Institute (API) 598, Valve Inspection and Test: This is the most widely used test specification in the world. It covers many types of valves, but was developed primarily for consideration of isolation valves. The standard includes leakage rates and testing criteria for both metal-seated and resilient-seated valves.
- Manufacturers Standardization Society (MSS) SP61, Hydrostatic Testing of Steel Valves: This is similar to API 598, but has subtle differences in test holding times and leakage rates. It was one of the first valve test standards, but is no longer widely used. However, it also was the basis for the more recent standards.
- International Standards Organization (ISO) 5208, Industrial Valves, Pressure Testing of Valves: This is ISO’s primary valve test standard, and covers many types of valves. It has 10 levels of allowable internal leakage, and the acceptance criteria is to be negotiated between the purchaser and manufacturer.
- American National Standards Institute (ASNI) FCI 70-2, Control Valve Seat Leakage: This contains detailed test procedures and leakage rate classes for control valves. The leakage classes are also occasionally referenced by other documents and used as acceptance criteria. It has six different leakage levels from Class I up to Class VI. An important note here is that ANSI FCI 70-2 is for internal testing only. It does not define external testing, and it refers users to API-598 for testing isolation valves.
- International Society of Automation (ISA) S75, Hydrostatic Testing of Control Valves: This provides a procedure for external testing of control valves. Closure testing and acceptance criteria are out of the scope of this document and usually are covered by referencing ANSI FCI 70-2.
Usually, the first stop in the production process for an assembled valve is the external test, known as a hydrostatic shell test. This test validates the soundness and strength of the valve body. It also pre-stresses the valve body, which aids in avoiding fatigue with cyclic pressure. For safety, this test is normally done with water at one-and-a-half times the valve body pressure rating. It also is typically done with the valve fully assembled and in the partially open position.
For shell testing, API-598 refers to ASME B16.34 for body pressure rating, which varies by body material. For example, a Class 150 valve in carbon steel (e.g., WCB) is tested at 31 bars. The same valve in stainless steel (e.g., CF8M) is tested at 29 bars. EN (European Committee for Standardization) and DIN (German Institute for Standardization) standards consider the body pressure rating only. For all body materials, a PN10 valve is tested at 15 bars and a PN40 is tested at 60 bars. It is not uncommon for customers conducting business internationally and used to working with EN and DIN standards to claim the wrong test pressure was applied when a valve is shell tested per API or ASME.
In addition, even a test as simple as the shell test has issues including:
- Sensitivity: In some cases, valve body castings have porosity too small for water to leak through, which is commonly called micro-porosity. This is most likely to occur in Monel and bronze castings. To detect this type of casting defect, the valve must be tested with air first, and the valve immersed in water or the exterior covered with soap solution. Such porosity can be blocked, in which case it will not be detected if the valve is pressurized with water first since the water is sufficient to block the leak path. Valve standards do not mention this additional test, but it is wise to use such tests with higher-risk cast materials going into hazardous and lethal gas service.
- Clamping forces: Split body valves are typically compressed in a hydraulic press to seal off the ends for pressure testing. If not carefully controlled, the clamping force can easily exceed the body joint gasket load generated by the bolted body joint. When this happens, the body joint gasket can be over-compressed, leading to loose body joint bolts and possibly external leakage.
- Special clean valves: Valves for use in services such as oxygen, chlorine and hydrogen peroxide require an extra level of cleaning and inspection to ensure their safe use. Water inside the fully assembled valve can introduce undesirable and unsafe contaminates, and any residual water remaining after the test can be hazardous when installed into service. For example, residual water remaining in valves for cryogenic service can freeze and prevent valve operation. In such cases, it is best to test the valve body prior to valve assembly.
- Corrosion: Residual water inside carbon steel valves can lead to crevice corrosion, a particularly aggressive type of corrosion that can damage critical sealing surfaces and result in both internal and external leakage. Attempts to remove the water through baking or heating the valve at elevated temperatures are generally ineffective.
- Chloride content: Concerns regarding stress-corrosion cracking of stainless steel has resulted in restrictions in the chloride content of test water, defined in several standards. Typically, the limit is around 50 parts per million (ppm), and some standards go as low as 30 ppm or lower. Low chloride levels are particularly difficult to consistently achieve if the factory is using a public or city water supply.
- Safety: When considering the problems of testing with water in special clean valves as well as the risk of corrosion, it is tempting to believe testing with air is a better choice. Sometimes this is true. However, since air is compressible, the volume of air contained that is compressed within the test valve can be very hazardous if something ruptures and the energy is suddenly released. According to the American Society for Nondestructive Testing’s Nondestructive Testing Handbook, exposure to a relatively small shock wave of air pressure can lead to the following injuries:
- Service repairs: A question exists of whether or not, after a valve has been repaired, the shell test must be repeated. If the original pressure-containing shell has not been altered during maintenance, it is not necessary to repeat the shell test since the soundness and strength of the valve body remains unchanged. These were already verified in the original factory test. However, if severe corrosion has occurred or if weld repairs have been performed, then a shell test is needed.
While the arguments regarding external leakage or shell testing appear to be significant, they are less complex than discussions over internal leakage. With internal leakage, it is imperative to understand that terms such as zero leakage and bubble tight are often misinterpreted—creating common mistake Number 1. The expressions “bubble tight” and “zero leak” should be further vetted. Table 2 defines measurable leak rates and explains them in everyday terms.
Most valve internal leak testing is done with air for a relatively short duration, from a few seconds to a few minutes. This is sufficient time to allow quick visual observation of a significant leak without being overly burdensome and impairing the factory’s capacity to build and ship valves.
While it is possible to observe bubbles rising in water with a leak rate of 1 x 10–4 standard cubic centimeters (cc) per second, factory tests are short in duration so the sensitivity of typical factory leak tests is around 1 x 10–3 standard cc per second. This is roughly equivalent to four cc (four game dice or marbles) of leakage per hour. Generally, this amount of leakage is insignificant because it would take years for the gas to simply fill the volume of pipe downstream from the valve.
Extending the duration from a few minutes to an hour brings the sensitivity of the test into the 1 x 10–4 standard cc per second range. This is equivalent to one game dice or marble volume of leakage in three hours, which is not a significant decrease in leakage considering the additional time it would take to test the valve.
Frequently, an inexperienced technician will apply a soap solution to the closure element and wait an hour or more before observing some foam developing. This test method is achieving 1 x 10-5 level of sensitivity, equivalent to one game dice or marble of leakage in one day. Considering some of the hazardous gasses the valve may be required to contain, it is reasonable to expect body joints and other static seals in the valve to be leak free when tested using this method. However, it is not reasonable to expect closure elements, which are dynamic seals, to achieve this level of tightness.
There are many different valve types and types of closure design, with equally varying materials of construction. Thus, many test standards exist to address the unique behaviors of each.
Failure to read and understand the reasoning behind the test methods leads to the second most common mistake, which is believing the factory test is done to prove the valve will work at its rated pressure. API and ISO standards do a thorough job of identifying the risk of this way of thinking. Each contains an important note warning that high-pressure testing of resilient seat valves can degrade performance in low-pressure service. The real purpose of the factory internal leak test is to verify the quality of the product; it is not to validate the strength of the design. The internal leak test should be appropriate for the type of valve, and it should be the most discerning and difficult test that valve must pass. Resilient-seated valves can have pressure-energized seals, meaning higher pressure increasing the sealing forces, potentially masking any dimensional or surface finish flaws. For such valve types, a low-pressure test is more difficult to pass and is a better indication of quality build.
For isolation valves, ASME, API and ISO standards are suitable when the valves are used in on/off service. They can also be the appropriate choice for control valves that are expected to provide isolation in service. For example, butterfly valves are frequently used for both control and isolation. If the standard factory test is based on API or ISO, there is no need to conduct an additional test per ANSI FCI 70-2 because the quality of the valve assembly has already been verified.
IN THE FIELD
The third common mistake is believing the factory production test is an indication of how the valve will perform in service. Often, end users will specify that the valve be tested under their process design pressures and temperatures. While this is not a problem in a laboratory equipped to conduct such tests, it is not practical to do this in a typical factory environment. Such tests are a validation of the valve design, not of the build quality.
Life would be simple if all valves were limited to no leakage for the duration of a test. But this is not technically possible. For valve pressure testing, it is wise to use the standard practice employed by the manufacturer. The manufacturer does not want to ship defective product and is naturally motivated to use the most discerning test methods with the least risk of damage to the valve.