Pneumatic Testing of Pipelines as an Alternative to Hydrostatic Testing

The site www.eng-tips.com is a technical forum for practicing engineers to discuss relevant topics with other practicing engineers. 

Discussions of static testing come up on eng-tips.com every few months.  Generally they will follow the format of:

This question is generally followed immediately by something like:

Which is pretty funny and quite memorable.  The message?  That pneumatic tests are irresponsible and anyone who proposes one is a cowboy.  While it is right and proper that we have a strong bias in favor of hydrostatic testing over testing with compressed gases, testing with compressed gas testing is far from irresponsible and can be the lower risk alternative in certain specific cases.

The risk being talked about here is that pressurized gas contains significantly more potential energy than pressurized incompressible liquid.  Rapidly converting this potential energy to kinetic energy can be a violent and destructive event.

Strength Testing for Pipelines
When new piping is to be placed in service, various codes and company standards require that it be subjected to a leak test and/or a strength test.  Leak tests are generally done at fairly low pressures and are only intended to prove that the pipe will in fact contain the fluids.  Risks are generally reasonably low and leak tests are done without much consideration of catastrophic failure.

The strength test is done with elevated pressure at some multiple greater than 1.0 of the system maximum allowable working pressure (MAWP) and held for some length of time.  The pressure multiple and time duration vary considerably from one regulatory jurisdiction to another, from one code document to another, and from one company to another.  Those details, while liberally sprinkled in posts on this topic are outside the scope of this discussion.

The primary kinds of tests are "Hydrostatic" or "pneumatic static" (sometimes called "pneumostatic" but that is just too pretentious).  The "static" simply means that during a successful test the fluids under pressure have no net movement relative to a pipe end or the pipe centerline.

A hydrostatic test is done using a largely incompressible fluid like water (hence the prefix "hydro"), oil, glycol, or some mixture (e.g., glycol is often added to hydrostatic-test water to prevent freezing).  In these tests, the line is filled with liquid, entrained gases are allowed to disperse to vents, and the pressure is raised within the system to the required test pressure and held there for the duration of the test.

A pneumatic static test is done using a gas like compressed air, nitrogen, CO2, or methane (tests with CO2 are very rare and very difficult because at elevated pressures the gas can change into a "dense phase" which behaves very differently from either a gas or a liquid).  The issues associated with pneumatic static testing are mostly concerned with stored energy.

Energy Involved in testing
The bulk modulus (i.e., the amount of pressure required to reduce the fluid volume by 1%) of liquids is very large, so even in the most aggressive tests the liquid will have very little compressive energy (e.g., the bulk modulus of water is on the order of 319,000 psi [2,200 MPa], so a 900 psig [6.2 MPa] test would reduce the volume by about 0.3%).  In a test failure, the energy release from this decompression would tend to slightly expand any tearing in the failed material, but is unlikely to create any projectiles.

Tests with gas are the exact opposite.  The density is very low so the gravitational forces are much less significant.  For example air at 900 psig would exert 0.034 psi/ft [0.758 kPa/m] which can be safely ignored. 

While the density of gas is low, compressibility is high enough to cause concern.  Compressing air from atmospheric pressure up to 900 psig at sea level at constant temperature would result in gas fitting into a volume that is 1/63 the initial volume.  Think of this a compressing a spring to 1/63 its length, and you begin to see the magnitude of the stored energy.

The concern in performing pneumatic tests is "explosive decompression".  NASA published a document a few years ago which has come to be known as the "NASA Glenn Research Centre Methodology".  This document was really the first time that anyone had made an effort to quantify the risk of pressurized gas.  It was on NASA's web site for several years but recent attempts to locate it have proven to be unsuccessful.  Several regulations and many company policies were written based on the NASA document.  Basically this 2 page document said:

• A pipeline failure could properly be called an "adiabatic" process (i.e., it occurs at constant entropy and is reversible)
• An adiabatic decompression results in a significant energy release.
• All of the material in the system will participate in the explosive decompression

Calculating the adiabatic energy in a pneumatic test
The adiabatic energy can be calculated by (this is the NASA version, the derivation of this equation requires a "k" in the numerator of the "k-1" term, but let's stick with the NASA version):

Where:

• Wgas --> Work done by the gas (N-m or ft-lbf).  To convert to "tons of TNT", divide the ft-lbf number by 3.086x109 or the N-m number by 4.184x109 (this number is the most common referenced conversion, but some sources use 4.8x109 N-m/ton of TNT)
• Vsystem --> Volume of the system (m3 of ft3)
• Ptest --> Pressure during the test (Pa or lbf/ft2) in absolute units
• Patm --> Local atmospheric pressure (Pa or lbf/ft2) in absolute units
• k --> Adiabatic constant made up of the ratio of specific heat at constant pressure over the specific heat at constant volume (no units, air has a value of 1.4)

This calculation can end up with a very large number.  For example, if you were testing 100 miles [161 km] of 36-inch [914.4 mm] Schedule 40 pipeline to 900 psig [6.2 MPA] at sea level (14.7 psia [101.35 kPa]) with compressed air, the volume of the system would be 3.428x106 ft3 [9.706x104 m3].  This results in total energy storage of 253.8 tons of TNT which is on the scale of a tactical nuclear weapon.  Scary stuff.  I'm not sure that "the next zip code" is far enough.

The problem with the NASA Glenn Research Methodology is that an explosive decompression event is very short duration.  Experiments done at the University of Nebraska-Lincoln for the Department of Energy in 2012 show that the gas temperature in an explosive decompression drops very rapidly to a minimum, and then increases to approximately initial temperature over the next few seconds.  This minimum can be taken to be the end of explosive decompression and the start of depressurization.  The referenced paper does not identify the duration of this nearly vertical temperature transient.  Other, less formal sources indicate it occurs at 10-50 mS after an opening large enough to result in choked flow is created.

Natural events within a gas volume are limited to the speed of sound (Mach 1.0).  This limitation is due to the creation of standing "shock waves" in the flow that inhibit communication from downstream to upstream.  Prior to Mach 1.0 the existence of lower pressure downstream is communicated upstream through a failure to support the higher upstream pressure.  At Mach 1.0 the shock wave is adequate to support the upstream pressure and only allow flow at the speed of sound.

So if we say that the vertical transient is 50 mS and allow half of the available time for the notice of the event to communicate within the system and half of the time for the energy that now "knows" that there has been a failure to participate in the explosion then with the speed of sound being:

Where:

• vsonic --> Speed of sound (m/s or ft/s)
• Rgas --> Specific gas constant (Universal gas constant/Molar mass)
• T --> Gas Temperature (R or K)

For air at 60°F [15.6C], the speed of sound is 1118 ft/s [341 m/s].  That says that over the 25 mS available, the shock wave would travel 28 ft [8.5 m].  Let's assume that the failure happened infinitely far (i.e., more than 28 ft [8.5 m]) from the end of the pipe so the amount of pipe involved is 56 ft [17 m] since stored energy from both sides of the failure participates.  That is a volume of 364 ft3 [10.29 m3] so using the adiabatic energy equation above, the energy is equivalent to 54 lbm of TNT—not a trivial event, but far from a tactical nuclear weapon.  To put it in perspective, 54 lbm of TNT in a properly constructed and properly deployed "cratering charge" would result in a crater 6 ft [1.8m] deep and 25 ft [7.62 m] in diameter which is a volume of earth of about 36.4 yd3 [27.8 m3].

In Thread378-293859, member SNORGY who is a frequent contributor to these discussions shared an Excel spreadsheet that uses the NASA calculations to set a "restricted distance" (i.e., the closest safe point of approach while under test) of 5621 ft [1.7 km] for this test.  Changing the pipe length to the 56 ft calculated above changes the restricted distance to 271 ft—still outrageous, but not over one mile.  This calculator demonstrates the utter fallacy of this approach—if the 100 mile line were operating at 300 psig (half of MAWP) the closest you could ever approach the line in operation would be 3670 ft (1.12 km).

The valve leaked through and pressure built up enough in the vessel to cause it to fail in a very dramatic manner.  This failure is put forth to demonstrate exactly how dangerous and irresponsible pneumatic testing is.  Another view is that you never test against a closed valve without monitoring the downstream conditions.  The failure was one of engineering procedures and/or procedure execution and should not be used to indict pneumatic testing.

Risks and mitigation strategies for hydrostatic tests
Hydrostatic tests are regularly done safely and without environmental consequences.  Successful tests have considered:

• Strength of materials.  Specified minimum yield strength (SMYS) is a measure of the stresses that the material can withstand without beginning to deform.  Different codes and company policies specify different maximum stress as a function of SMYS.  Raw-gas gathering systems are often limited to 20% of SMYS.  Cross country processed-gas transportation often allows stresses much closer to 100% of SMYS.  Lines with a high potential to impact the public are limited to lower fractions of SMYS than lines in open country.  Prior to any decisions on testing, these stresses must be quantified and factored into the decision.
• Environmental/safety considerations. 
• Hydrostatic test water (even without chemical additives) must be treated as industrial waste and cannot be dumped in the roadside ditch.  Successful tests address this issue by defining a disposal point and verifying that that location will accept the water. 
• A failed test will empty all or part of the liquid involved in the test near the failure.  A successful test will anticipate this through temporary berms to protect sensitive locations (i.e., rivers, dry washes, parking lots, office buildings, etc.). 
• Dewatering hydrostatic tests has been the cause of innumerable spills and injuries.  Sending large volumes of liquid through a flexible conduit like a fire hose has the ability to develop very large exit forces on the outlet nozzle which can cause the end of the hose to swing wildly with the risks of personal and property damage.  Successful tests specify means to capture the hose ends.
• Regulatory considerations.  There are jurisdictions where a test plan must be approved by a regulator prior to performance.  Other jurisdictions require notification but not approval.  If roads are going to be closed during the test, then permits are usually required.  Successful tests request the required approvals/permits well in advance of the test.
• Source of liquid.  Every liquid source contains microbes and contaminants, many of which pose long-term integrity management risks to pipelines.  Successful tests have considered that it is very common for a test to leave some amount of liquid behind after the test and specifies required treatment chemicals.
• Weight of the liquid.  When testing lines with above ground sections, it is important to confirm that the pipe supports are adequate to carry the pipe full of liquid (collapsing pipe racks are a common source test failure).
• Terrain.  The test needs to ensure that the test pressure meets a minimum magnitude at the high points without being "excessive" at the low points.  Engineering judgment is required define "good enough" (e.g., is it acceptable to go to 160% of MAWP at the low point to be able to reach 110% of MAWP at the high point?  Or is it better to stay at 150% of MAWP at the low point and accept 90% of MAWP at the high point? Or can you segment the line to stay within ±10% of 150% of MAWP?).
• Line terminations.  If the system under test has already been connected to upstream/downstream piping/vessels then you have to consider how you are going to prevent your test from including this outside piping.  If there is no way to avoid testing against a shut valve, then you need pressure monitoring and over-pressure protection on the connected systems.
• Determining injection/drain, test, and vent points.  These points all have to be accessible and located somewhere that is useful.  For example, if the designated vent point is at a system low point, then it will be difficult to remove any gas that may accumulate at high points.
• System fill.  Any liquid introduced will have the potential to bring entrained gas along with it.  This gas is very compressible and can make a nominally incompressible test very difficult.  A successful test will anticipate this gas and specify soak times after fill and vent frequency during the fill step.
• System pressurization.  Rate of pressurization and minimum temperatures (both ambient and fluid temperatures) must be considered to prevent brittle failure in piping that would otherwise have passed the test.
• Test execution.  All but the shortest tests will experience some amount of temperature change.  Water will change pressure about 100 psig/°F
  [1241 kPa/C].  Fairly small temperature changes cause significant pressure changes.  A successful test will include acceptance criteria.  For example, on hydrostatic tests that I design I specify that fluid can be bled off during the test but cannot be added, and that the test is successful if the final pressure is greater than MAWP.  Others specify a maximum volume that can be added to maintain test pressure.  It all comes down to engineering judgment.
• System drain.  Once the test fluid has been in new piping, it must be treated as industrial waste because of the near certainty that it will pick up oil, grease, and mill scale.  You can't just dump it on the ground.  Also, there have been several incidents of unsecured hoses flopping about and injuring people.  These risks need to be anticipated and minimized.
• System drying.  Many systems will not naturally drain due to undulations in the piping topology.  Generally this residual liquid is removed by running pigs with air.  Successful tests specify how dry the line needs to be prior to turning the line over to operations (e.g., "run foam pigs until one arrives dry", or "purge -40°F nitrogen through the line until water content on a Draeger Tube is less than 7 lbm/MMSCF").
• Clean up.  Tests always require some amount of system modification (e.g., installation of blind flanges and fill equipment) that must be undone prior to the test being called "complete".  Successful tests have detailed lists of the things that need to be done and if there are any temporal dependencies, the order in which they must be done.

Risks and mitigation strategies for pneumatic static tests of pipelines
Many of the issues mentioned above under hydrostatic tests are identical to the pneumatic static tests.  Some are a bit different:

• Strength of materials calculations are the same for pneumatic static tests as for hydrostatic tests above.
• Environmental/safety considerations
• With the high-energy concentration in the gasa failure has the risk of launching debris at high velocity.  For buried lines the primary debris is dirt and rocks, but rocks have been used as projectiles since time immemorial.  For above-ground structures the debris will be pipe or fitting material.  Some of the most damaging failures involved launching a weld-neck flange and blind hundreds of feet.  Successful tests consider "exclusion zones" around buried pipe and a combination of barricades and exclusion zones around above-ground structures.  Consideration is also given to doing tests during times of minimum occupancy of roadways and structures.
• Regulatory considerations are similar to hydrostatic tests, with the exception that there are jurisdictions that have a strong bias against pneumatic static tests.  In those cases it is mandatory that you have done adequate preparation work to demonstrate why you are suggesting a pneumatic static test instead of a hydrostatic test.  "Convenience" or "cost" will rarely carry much weight in this discussion.  You must demonstrate that the potential outcome of a hydrostatic test is measurably worse than the potential outcome of a pneumatic static test (e.g., "impossible to adequately dry", "segmentation points inaccessible").
• Source of gas.  With gases we are not concerned about multiphase issues (i.e., gas in the liquid) or about corrosion.  We are very much concerned with the suitability of the gas for the test.  If the test media is compressed air, then you have to have an air compressor that can move huge volumes at moderate pressure for most of the fill period and then smaller volumes at high pressure for the remainder.  For a nitrogen test you have to pick a source (i.e., bottles or bulk liquid nitrogen) and make sure that you understand the issues of your choice (e.g., changing nitrogen bottles is risky, bottles are able to be emptied less as system pressures increase, bulk nitrogen is in liquid form and must be heated prior to injection).
• Weight of the fluid is not an issue with gas.
• Terrain is not an issue with gas
• Line terminations.  All of the issues are identical to hydrostatic.
• Determining injection/drain, test, and vent points.  You do not need to degas a gas fill, but you do still need fill/drain and test points.
• System fill.  Ambient and gas temperatures are far more critical in pneumatic static tests than in hydrostatic tests.  Both minimum ambient temperature and minimum injection temperature must be specified and monitored.  Also, since the stored energy in a pneumatic static test is so much greater than the stored energy in a hydrostatic test, specifying soak times at specified pressures to allow stresses to equilibrate is required.  On a test I recently designed, we filled the system at 5 psig/min to 50 psig followed by a 30 minute soak period.  After the soak, the pressure was increased at10 psig/min with 30 minute soak periods at 150 psig, and 450 psig.  These pressures, fill rates, and soak periods were determined by calculating stress accumulation.
• System pressurization.  At the end of the fill period, the system is pressurized.
• Test execution.  Pneumatic static tests are much less subject to changing pressure due to temperature variation.  It is rare for the test pressure to increase or decrease significantly due to temperature equilibration.  Just like the hydrostatic test, a successful test will include acceptance criteria. 
• System drain.  At the end of the test the gas will normally be vented to atmosphere.  For air and nitrogen the big concern with the blowdown is Joule-Thomson cooling of the piping into the brittle failure region.  On the test mentioned above we specified a maximum depressurization rate of 25 psig/min (and specified that the rate be determined every 60 seconds).  One significant exception is tests with saleable products.  If I test a CO2 line with CO2 then I can leave the system pressurized for service after the test.  Same way with testing a natural gas line with natural gas.
• System drying is not an issue with pneumatic static tests.
• Clean up issues are the same as hydrostatic tests above.

Discussions on professional forums about pipeline testing

• There was not a single post referencing personal knowledge of a pipeline failure under a pneumatic test.  There was one very interesting post about a valve that failed under a manufacturer's pneumatic test and one about pipe spools that failed in a yard test.  No pipeline test failures were reported in the first person (there was one post where a responder indicated that "he knew a guy that …", but the anecdote just supported the official investigation).
• In all the threads that I reviewed there were only a dozen distinct references to failures under pneumatic tests.  None of the links older than 2007 were still valid, but the ones after 2007 all referred to one of 4 pneumatic test failures.  A couple of posts referenced fatalities related to hydrostatic tests.  Several posts referenced failures and explosions in pressurized systems that were years past their static test (sometimes decades past).
• Every single pneumatic failure with injuries/fatalities could be traced to a failure of engineering (e.g., a 2600 psig pressure source was connected to a 900 psig test without a pressure relief valve between the very high pressure source and the valve being tested) or a failure to properly execute the procedure (e.g., not monitoring the injection temperature from a liquid nitrogen tank, or starting the test with the piping below the specified minimum ambient temperature).  Every injury associated with pneumatic static testing can be traced directly back to these two causes.  If proper procedures are written and followed then a pipe failure in a pneumatic test is just a pipe failure, not an ambulance ride.

My conclusions from reading this concentrated body of work are that:  (1) many people feel that hydrostatic testing is inherently safe and doesn't require any significant analysis; and (2) many people feel that pneumatic static testing is inherently unsafe and cannot be done without creating unacceptable hazards.  The first conclusion is frightening because the human and environmental risks associated with a hydrostatic test are considerable.  They can be managed, but a cavalier attitude towards that much mass and energy is quite dangerous.  The second conclusion precludes competent consideration of a valid technique to mitigate the risks associated with hydrostatic tests.

It is reasonable to say that if disposal, drying, and mass risks of liquid tests can be adequately managed then a hydrostatic test is preferable.  On the other hand it is not unreasonable to say that there are times that the best way to mitigate the risks of a hydrostatic test is to do a pneumatic static test.

About the Author

David Simpson, PE is an Oil & Gas engineering consultant at Muleshoe Engineering.  David is an MVP in the professional forums at www.eng-tips.com and a member of the Engineering Writers Guild. 

Follow David (zdas04) at http://eng-tips.com/userinfo.cfm?member=zdas04