Hydro and Air Tests

Hydrostatic Pressure Test

Pressure vessels are tested for leaks with a hydrostatic pressure test.  The test involves filling the vessel with a liquid, usually water, and pressurizing it to the specified test pressure.  In my career, the rule for determining the hydrostatic test pressure has been to simply multiply the Maximum Allowable Working Pressure (MAWP) by 1.5.  That is, suppose a certain vessel has an MAWP of 600 psig; it would be hydrostatically tested at 900 psig.  Of course, there are other details such as the duration of the test, variations based on operating temperature, etc.  But basically, the hydrostatic test pressure has been 1.5 x MAWP for many years – certainly since the 1950s.  However, it has not always been that way and may be about to change.

In Section 10.3.3, API 682 4th Edition requires that pressure casing components be tested hydrostatically with liquid at a minimum of 1.5 times the Maximum Allowable Working Pressure (MAWP) of the pump casing but not less than 20 psig.  Adjustments must be made based on the allowable working stress of the material based on operating temperature.  The hydrostatic test duration is 30 minutes without leaking.

Apparently, in the future, the hydrostatic test pressure ratio will be changed from 1.5 to 1.3 according to the ASME Pressure Vessel Code Section VIII.  This will be applied to both pumps and piping.  The same multiplier will probably be used for API 682 reservoirs such as are used with Piping Plan 52 and 53.  The pertinent 4th Edition clauses for reservoirs now read: The reservoir is part of the pump piping system. Unless otherwise specified or required by local code, the reservoir shall be designed, fabricated, and inspected in accordance with ISO 15649 or ASME B31.3 using piping components. If the reservoir is built entirely of piping components, ISO 15649 or ASME B31.3 can be applied and provides adequate design for the reservoir just as it does for the pump suction and discharge piping. It is the user’s responsibility to ensure that local codes do not require that reservoirs be built in accordance with a pressure vessel code such as EN 13445 or ASME VIII, Division 1.

With this wording, the default is a pipe based reservoir but an ASME certified reservoir is an option.

ASME Certified Pressure Vessels

ASME certified pressure vessel fabricators undergo a rigorous program to ensure compliance with the rules and regulations of the ASME Boiler and Pressure Vessel code.  For unfired pressure vessels, these requirements are given in ASME VIII, Division 1 (EN 13445).  There are two variations of certification:  “U” stamp and “UM” stamp.

“U”stamped pressure vessels are required to have a 3rd party ASME inspector review and approve the calculations as well as inspect certain stages during manufacture of the reservoir.  This inspector also witnesses the ASME hydrostatic test which, apparently, is now 1.3 x MAWP.  This means that U stamped vessels cannot be manufactured in advance; that is, the U stamped vessel is customized for a particular service, has a serial number and therefore cannot be a stocked item. The fabricator undergoes an ASME inspection every three years.

In contrast to the U stamped vessels, there is a “UM” stamped vessel which is also subject to the same ASME specifications.  However, the “UM” stamped vessel is limited to a maximum of 1.5 cubic feet for a 600 psig rating and 600 psig is the maximum permitted pressure for the UM stamp.  ASME 3rd party inspection is not required; therefore “UM” vessels can be manufactured in advance, they do not have a serial number and they can be stocked as an inventory item.  The UM fabricator is inspected annually by ASME.

Pipe Based Reservoirs

Pipe based reservoirs for API 682 sealing systems must be built entirely of piping components and ASME B31.3, “Process Piping”, (ISO 15649) is the governing standard.

Mechanical Seals

It should be noted that the mechanical seal is not considered to be part of the pump pressure vessel and therefore does not fall under pressure vessel rules.  Seal manufacturers have several pressure ratings for their products.  API 682 recognizes a static pressure rating, a dynamic pressure rating and a hydrostatic pressure test rating.

The  API 682 Integrity Test

The American Petroleum Institute standard for mechanical seals, API 682, includes a test of the final cartridge assembly using pressurized air. The purpose of this test is to prove that the cartridge has been assembled properly; however, the test, variously called the “air test” or “integrity test” is widely misunderstood. 


In accordance with API 682, “Pumps—Shaft Sealing Systems for Centrifugal and Rotary Pumps”, seal cartridge assemblies are to be tested using pressurized air prior to shipment to insure that the cartridge has been properly assembled. This Integrity Test has been in place since the first edition of API 682 was published in 1992.

The API Integrity Test has proved to be one of the major benefits of API 682 by virtue of contributing measurably to a decrease in startup problems. In fact, it has been so beneficial that many OEMs and End Users now do the API 682 Integrity Test on all cartridge seals regardless of whether that cartridge is “per API” or not.

The test and procedure that API 682, 1st Edition adopted was a Chevron test and was recommended to Chevron by BWIP (now Flowserve). The procedure had been in use for about three years by Chevron when it was adopted by API 682.  In any case, the procedure was in the first draft of API 682 (June, 1992) in pretty much the same form, pressure, time, volume, etc. as currently exists.

About the same time (early 1990s), Roger Jones had a pump and seal air test rig built at the Shell Deer Park Refinery in Texas.  Roger tested five pumps with seals at 50 psig air pressure. After bubble testing (no bubbles allowed) the various pump and adapter joints, the air supply was blocked in and the time recorded for the pressure to fall by 1 psi.  Roger then took the volume of the seal test rig and proportioned the results to get an acceptable leakage rate for a 1 cubic foot volume during a five minute test period. His work essentially confirmed the usefulness of the proposed API 682 method.

The maximum volume of one cubic foot had been recommended by Gordon Buck in response to Terry Roehm’s (Ashland Oil at the time) comment about pressure drop being related to volume. Absolutely no thought was given to correlation of volume, pressure drop, etc to leakage rate.  Gordon just said something like “How about one cubic foot?” while envisioning the physical size of a test pot that might be needed to hold a big cartridge.  The intent was more along the lines of limiting test volume that it was to specify a volume. The one cubic foot suggestion was immediately adopted; however, it has since been widely criticized.

In 1990, John Crane had developed an air test procedure for cartridge seals being supplied to major pump OEMs. That test proved tremendously valuable as a QC tool.  John Crane provided the details of that test procedure to the 1st Edition Task Force and recommended it as the Integrity Test.  The Crane test was very simple:  hold 50 psig for 30 seconds without loss of 2 psi; no additional system volume was to be used.  That is, the test volume consisted of the volume of the cartridge seal plus a bit of tubing.  Over a two year period, not even one cartridge that had passed the JC test was returned as a leaker traceable to assembly.  The vast majority of seals that did fail the initial JC test were reported to have had either cut O-rings or scratched faces.  The JC test was not adopted by API – mostly because the task force members did not like the 30 second duration although some did not think that 50 psig air pressure was safe and some did not have 50 psi air systems.

After the API 682 Integrity Test was finalized, John Crane double tested seals assembled in their repair facilities for a month using the then existing JC test and the new API test. Every cartridge that passed one test also passed the other and vice versa.

With the acquisitions of Flexibox and Sealol, John Crane suddenly had at least four basic models of air test rigs with several variations of each basic model. Many of the test rigs looked like a pot.  All required many adapters.  When testing a single seal, the test volume is essentially that of the pot.  Pot volumes range from about 1/4 cubic foot to 1/2 cubic foot.  It would be difficult to get test rig volumes much less than about 1/8 cubic foot without making special pots, housings, fillers, etc for various seal sizes and arrangements.  When testing dual seals, some rigs added volume to the buffer/barrier chamber but most did not.  Controls and measurements systems range from fully manual to fully automatic using programmable logic and digital output.

Leakage Rate

The API 682 Integrity Test allows a maximum pressure drop of 0.14 bar (2 psi) over 5 minutes from a maximum 28 liter (1cu ft) reservoir that is pressurized to 1.7 barg (25 psig). Based on the ideal gas laws, this is a leakage rate of 56.9 g/hr of air.  Realistically, this is quite a high leakage rate and, as such, has generated criticism of the Integrity Test.

As strange as it may now seem, during the development of the Integrity Test, there was no attempt to correlate leakage of air to the API 682 Qualification Test or field performance. In fact, the Integrity Test was defined in 1st Edition of API 682 whereas the leakage criterion was not established until 2nd Edition.  Again, the Integrity Test was intended only to test for assembly practices and was based on a successful established program.

For seals within the scope of API 682 to have an air leakage rate of 56.9 g/hr across the seal faces, the face separation would be in the range of 4 to 6 microns (160 to 240 micro-inches). This is quite a large face separation as compared to typical face separations more on the order of a half to one micron (20 to 40 microinches).  Alternatively, a small hole on the order of 0.25mm (0.01 inch) diameter could also produce an air leakage rate of 56.9 g/hr.  (The exact calculated values depend on assumptions and coefficients used.)

Whether the air leakage occurs through the seal faces or through a small hole, the corresponding calculated liquid leakage rates would also be very high.

Although most seals will seal both air and liquids effectively at low pressures, all do not. For example seals designed for high pressures with low balance ratios might seal liquids satisfactorily but leak air.  Seals with special face finishes (such as matte finish) will allow more air leakage than will seals with polished faces.  Non-contacting seals definitely leak more than contacting seals.

It is important to understand that the API 682 Integrity Test is to be done with the seal faces “dry”. At the same time, it is well known that a bit of oil on the seal faces dramatically reduces the air leakage rate during the integrity test – sometimes to essentially zero or “bubble free” leakage.

In practice, the integrity of the seal being tested is virtually always immediately apparent. That is, the seal obviously passes or fails.

Air Test Rigs

The empty volume of a large (Size 10) pump seal chamber is between 2 and 3 liters depending on the depth.  The pump shaft displaces a liter or so.  Therefore, a requirement for less than 2 or 3 liters of air test volume is likely to be somewhat difficult to obtain unless the test chamber is closely matched to the seal size. This means that many test chambers would be required.

On the other hand, the estimated test volumes of dual seal chambers are on the order of 15 to 30 inch^3 (250 – 500 cc).  Therefore, test times for dual seal chambers could potentially be reduced.

Additional considerations include:

  • Was the lapping done correctly?
  • Are the faces polished? (Hard/hard combinations may not be polished.)
  • Have the primary or mating ring been distorted? This particularly applies to shrink fitted assemblies and clamped in mating rings.
  • Is the secondary seal or any other part of the seal head hanging up?
  • Is the carbon “porous”?
  • Have sealing surfaces been badly scratched?
  • Have non-elastomeric secondary seals been fitted correctly?

 Problems in the Field

In spite of the benefits of the Integrity Test, there are times when a seal cartridge passes the Integrity Test but leaks more than expected in the field. How can this happen?

First of all, the Integrity Test does not test the bolts and gaskets used to attach the cartridge to the pump. In fact, the Integrity Test does not even include the gland plate bolts.  The Integrity Test also does not test the gasket surfaces of the pump.

To prevent distortion, API 682 gland plates are required to have metal-to-metal contact with the face of the seal chamber. This contact is required to be both inside and outside the bolt circle.  However, many older pumps are not designed to provide such support and permit gland plate distortion to occur as the bolt are tightened.

Even though API 682 has an allowable leakage rate, sometimes end users expect “bubble free” air tests and zero leakage performance.


The Integrity Test has proved to be one of the major benefits of API 682 by virtue of contributing measurably to a decrease in startup problems. This test has been in place since the 1st Edition of API 682 was published in 1992. Current requirements are described in Section 10.3.4 of API 682 4th Edition.

The Integrity Test is not designed to insure performance; it is the Qualification Test that is designed to demonstrate performance. The Integrity Test is designed to find assembly errors and/or manufacturing defects.  It is practical, easy to implement and has been used successfully for many years.

The Integrity Test of 4th Edition does not differentiate by face technology, size, proportions, arrangement, materials, lapping technology, etc., etc.  The same test is used for all seal designs.  If the existing simple Integrity Test is changed then it is likely that seal design parameters will have to be considered as part of the test procedures and pass/fail criteria.