June 28, 2002
A CENTURY OF CYLINDERS – TEST PROGRAM
Introduction
This year marks our 100th anniversary of manufacturing steel seamless high-pressure cylinders for the compressed gas industry. It is a fitting occasion to reflect on the history of that endeavor and examine issues that affect it’s future. As our industry advances into the 21st century, it is incumbent on manufacturers, regulators, gas producers, and end-users of high-pressure gas cylinders to contemplate the implications of indefinite use of gas cylinders. We believe that the current industry practice of using cylinders indefinitely needs to change.
From the beginning, safety has been the guiding principle in the design, material selection, manufacturing and inspection of high-pressure cylinders. As technical advances were made, the safety of high-pressure cylinders improved.
Essentially, there have been four generations of cylinder technology over the past 100 years.
The first generation ran from 1902 to approximately 1930. Cylinders were made from carbon steel, the principle material for use at that time.
By 1930 the second generation of cylinders, manufactured from intermediate manganese steel, was in use.
The development of the third generation of cylinders was driven by U.S. military demands for cylinders that would not fragment when ruptured like earlier cylinders. The result was the development of quenched and tempered chrome molybdenum alloy steel cylinders designated as 3AA.
The fourth generation of cylinders are high strength cylinders manufactured from enhanced alloy steels. These cylinders are currently manufactured under DOT exemptions; such as E9421 and E9909.
Concurrent with cylinder material and technology developments, innovations in the machinery and process employed by the steelmaking industry have further contributed to the enviable safety record of today’s cylinders. The progression from Bessemer converters to open-hearth furnaces to present day electric furnaces and BOF’s (basic oxygen furnaces) all factored into current, enhanced cylinder composition and design.
Each advance from carbon steel to intermediate manganese steel to the modern chrome-moly alloy steel has resulted in the enhanced toughness of seamless steel cylinder products. In addition to allowing thinner walls, the increased toughness equates to cylinders that exhibit improved flaw tolerance and fracture performance. For example,
an antiquated carbon steel cylinder may fragment on failure as opposed to the predictable, leak before break attribute of a modern chrome-moly cylinder. Decades of advancements in cylinder design, manufacturing technology and inspection expertise have also contributed to a modern cylinder population with improved fracture performance.
We have conducted a series of tests to exhibit the fracture performance of generation I, II, and III cylinders. The outstanding fracture performance of generation IV cylinders has been previously documented in Taylor-Wharton’s exemption application to DOT and, more recently, in the work of ISO/TC58/SC3/WG14.
Performance Tests
Hydro-pneumatic Burst Test
The Hydro-pneumatic Burst Test was selected to exhibit the fracture performance of cylinders under adverse service conditions that would result in failure of the cylinder while pressurized with a gas. Under all such conditions, the cylinder is subjected to general or localized stresses that go well beyond the design conditions and exceed the ultimate tensile strength of the steel. Such conditions could include but are not limited to physical damage of the cylinder, wall thinning due to corrosion, cyclic fatigue, over-pressurization, impact loading due to dropping, etc., and exposure to a fire.
A cylinder to be tested is charged with air or nitrogen to the marked service pressure. Note: Old cylinders without a marked service pressure, i.e. ICC 3, were assumed to have a service pressure of 1800 psi. Water is then incrementally pumped into the cylinder to increase the pressure until failure occurs.
The failure of a cylinder under gas pressure is a violent event and adequate precautions need to be taken to protect anyone in the area. The cylinder or cylinder fragments may travel a substantial distance (even as much as 300 yards). When failure occurs, the failure pressure is recorded and the cylinder or its remains are retrieved and examined.
Acceptable results are:
- The failure initiates in the sidewall
- The cylinder remains in one piece
- Minimum burst pressure attained (usually WP x 2.25)
As an alternate test, a flaw can be machined into the cylinder wall to add the aspect of flaw tolerance to the test.
Hydrostatic Burst Test
The Hydrostatic Burst Test is used to verify critical design parameters such as the safety factor, the ratio of burst pressure to service pressure. It can also give an indication of ductility and fracture toughness.
Comparing the Hydro-pneumatic Burst Test to the Hydrostatic Burst Test, a cylinder would be expected to have the same burst pressure in either test. The burst appearance, however, would be expected to be very different. In the Hydrostatic Burst Test, the hydraulic pressure that is the driving force for the failure dissipates the instant that failure initiates, removing the driving force that causes further tearing of the sidewall as in the Hydro-pneumatic Burst Test.
A cylinder that fragments in the Hydro-pneumatic Burst Test would most likely remain in one piece in the Hydro-pneumatic Burst Test while exhibiting a brittle burst profile with branching or fishtailing at the ends of the rupture.
A cylinder to be tested is filled with water and then pressurized by pumping water under increasingly higher pressure into the cylinder until failure occurs.
Acceptable results are:
- The failure initiates in the sidewall
- The fracture face shall be a 450 (shear) surface as opposed to a 900 surface indicating brittleness
- Minimum burst pressure attained (usually WP x 2.25)
As with the Hydro-pneumatic Burst Test, an alternate test can be performed by machining a flaw into the cylinder wall to add the aspect of flaw tolerance to the test.
Gunfire Test
This test, designed for cylinders intended for military service, evaluates the nonshatterability of the cylinder and indicates ductility at a high strain rate.
A cylinder to be tested is charged with an inert gas to the marked service pressure. The cylinder is shot at a right angle on the longitudinal centerline ± 1 inch near the vertical center of the cylinder. The bullet is a 0.50 caliber armor piercing projectile that is either fired straight on or tumbled by passing through a 450 tumble plate. The projectile strikes the cylinder at 2800 feet per second ± 100 feet per second.
Acceptable results are:
- The cylinder breaks into no more than 2 pieces, not counting pieces less than 2 inches in diameter coming from the area of the point of entry and exit.
Material Tests
Transverse Charpy Impact Test
This test gives a relative indication of resistance to crack propagation and also of material toughness. It provides a numerical value of the energy required to break a standard test specimen with a pre-machined notch and thus serves as a comparison of the notch toughness of various materials and aids in predicting failure modes in service.
Steel toughness generally is reduced as the testing temperature is lowered. The temperature at which a reduction of 50% of the values for room temperature occurs is commonly referred to as the transition temperature. Testing a material over a range of temperature can be used to determine minimum operating temperature for that material.
Charpy impact tests traditionally have been performed in the longitudinal direction which usually result in higher values than testing in the transverse direction. However, in a cylinder, the highest stresses under pressure are in the circumferential or "hoop" direction. Transverse Charpy impact testing aligns the test direction with the highest stresses and more accurately represents a cylinder in service.
To perform the test, specimens (usually in sets of three) are brought to the testing temperature in an agitated liquid bath. A specimen is then placed in the anvil which provides support for the specimen to receive the blow of the moving mass. A pendulum of known mass is then released which swings through the anvil and breaks the specimen. An indicator attached to the pendulum displays the energy absorbed by the broken specimen. The fracture faces of the specimens are examined and evaluated for % fibrous fracture.
Acceptable results are:
- 13 ft-lb for a 5 mm x 10 mm specimen
- 100% fibrous fracture (0% brittle)
- Transition temperature < -500 C
Cylinder Test Results
Hydro-pneumatic Burst Test Results – without flaw
Cylinder Serial Number 33562
Specification ICC3A2000
Manufactured June 1915
Generation I
Burst Pressure 6000 psi
An attempt was made to confine the cylinder in a steel box buried in the ground. The explosion ripped the box from the ground, tore the door off, and scattered fragments over several hundred feet. Thirteen pieces were recovered comprising most or all of the cylinder. This is the most dramatic illustration of the potential fragmentation hazard.
Hydro-pneumatic Burst Test Results – without flaw (2)
Cylinder Serial Number K225940
Specification ICC3A2015
Manufactured February 1917
Generation I
Burst Pressure 5900 psi
Here is another unacceptable result. Three pieces were recovered comprising about 67% of the cylinder. Secondary cracking nearly produced more, smaller pieces.
Hydro-pneumatic Burst Test Results – without flaw (3)
Cylinder Serial Number TW02-519883
Specification DOT3AA2265
Manufactured May 2002
Generation III
Burst Pressure 6750 psi
The cylinder remained in one piece and showed evidence of considerable bulging prior to failure. Crack propagation arrested in a relatively short distance. This is an example of an ideal result under hydro-pneumatic burst conditions.
Hydro-pneumatic Burst Test Results – with 3" long x 0.140" deep flaw
Cylinder Serial Number 255822
Specification ICC3A2000
Manufactured May 1918
Generation I
Burst Pressure 4200 psi
The cylinder fragmented into three pieces with numerous secondary cracks. The flaw reduced the burst pressure about 30% but the brittle burst still tore the cylinder into three pieces.
Hydro-pneumatic Burst Test Results – with 3" long x 0.140" deep flaw (2)
Cylinder Serial Number 17471
Specification ICC3A2015
Manufactured November 1939
Generation II
Burst Pressure 3800 psi
The cylinder showed an extreme amount of tearing, but remained in one piece. The material is ductile, but the flaw tolerance is low. The flaw reduced the burst pressure about 37%.
Hydro-pneumatic Burst Test Results – with 3" long x 0.140" deep flaw (3)
Cylinder Serial Number TW02-519882
Specification DOT3AA2265
Manufactured May 2002
Generation III
Burst Pressure 4100 psi
The cylinder remained in one piece and showed evidence of considerable bulging prior to failure. Another example of an ideal result.
Hydrostatic Burst Test Results – without flaw
Cylinder Serial Number H52147
Specification ICC2015
Manufactured February 1925
Generation I
Burst Pressure 5100 psi
A very brittle result for a hydrostatic burst. This cylinder would be expected to fragment under pneumatic or hydro-pneumatic burst conditions. The result is unacceptable.
Hydrostatic Burst Test Results – without flaw (2)
Cylinder Serial Number 312156
Specification ICC3A2015
Manufactured August 1926
Generation I
Burst Pressure 5900 psi
This cylinder shows some ductility with bulging at the initiation point, but the fracture surface is mainly brittle. There would be possible fragmentation under pneumatic or hydro-pneumatic burst conditions. Unacceptable result.
Hydrostatic Burst Test Results – without flaw (3)
Cylinder Serial Number 412647
Specification ICC3A1800
Manufactured December 1939
Generation II
Burst Pressure 4400 psi
A ductile burst approaching the performance of Generation III cylinders. Acceptable result.
Hydrostatic Burst Test Results – without flaw (4)
Cylinder Serial Number TW02-519886
Specification DOT3AA2265
Manufactured May 2002
Generation III
Burst Pressure 6870 psi
A ductile burst with considerable bulging and very little crack extension beyond the initiation point. An ideal result.
Hydrostatic Burst Test Results – with 3" long x 0.140" deep flaw
Cylinder Serial Number 42-3233
Specification ICC3A1800
Manufactured September 1920
Generation I
Burst Pressure 3300 psi
A very brittle result for a hydrostatic burst. This cylinder would be expected to fragment under pneumatic or hydro-pneumatic burst conditions. Unacceptable result.
Hydrostatic Burst Test Results – with 3" long x 0.140" deep flaw (2)
Cylinder Serial Number 23-4145
Specification ICC3A2015
Manufactured August 1923
Generation I
Burst Pressure 2800 psi
A mostly brittle result for a hydrostatic burst. This cylinder could possibly fragment under pneumatic or hydro-pneumatic burst conditions. Unacceptable result.
Hydrostatic Burst Test Results – with 3" long x 0.140" deep flaw (3)
Cylinder Serial Number 12322
Specification ICC3A1800
Manufactured March 1928
Generation I
Burst Pressure 2400 psi
A brittle result for a hydrostatic burst. This cylinder would be expected to fragment under pneumatic or hydro-pneumatic burst conditions. Unacceptable result.
Hydrostatic Burst Test Results – with 3" long x 0.140" deep flaw (4)
Cylinder Serial Number L1486
Specification ICC3A2015
Manufactured August 1943
Generation II
Burst Pressure 2400 psi
A mostly brittle result for a hydrostatic burst. This cylinder could possibly fragment under pneumatic or hydro-pneumatic burst conditions. Unacceptable result.
Hydrostatic Burst Test Results – with 3" long x 0.140" deep flaw (5)
Cylinder Serial Number TW02-519885
Specification DOT3AA2265
Manufactured May 2002
Generation III
Burst Pressure 4000 psi
The 3A Carbon (Generation I) cylinders had extremely low impact properties at all temperatures. This indicates that these cylinders are subject to failure from relatively low impact loading. The results are consistent with the fragmentation that occurred in the burst tests. The fracture faces of all 3A Carbon specimens tested were 100% brittle.
The 3AA Cr Mo (Generation III) cylinders had excellent impact properties at all temperatures. This indicates that these cylinders are fracture tough and is consistent with the performance of these cylinders in the burst tests. The fracture faces of all 3AA Cr Mo specimens tested were 100% fibrous.
The testing demonstrates the benefits of 100 years of technology evolution.
Generation I cylinders exhibited poor fracture performance, fragmenting under hydro-pneumatic failure conditions and revealing brittle fracture profiles under hydraulic failure conditions.
Generation II cylinders exhibited better fracture performance under all burst conditions tested, but were far short of the performance of the Generation III cylinders.
Generation III cylinders exhibited excellent fracture performance under all conditions tested.
Generation IV cylinders were not tested as pert of this investigation. The outstanding fracture performance of Generation IV cylinders has been documented in Taylor-Wharton’s exemption application to DOT and , more recently, in the work of ISO/TC58/SC3/WG14. The fracture performance of Generation IV cylinders is comparable to that of Generation III cylinders even though the strength levels are significantly higher.
The gunfire test did not discriminate between the different generations of cylinders. There was no significant difference between any of the results. All of the results were acceptable indicating that the critical flaw length for all of the cylinders tested is greater than the length of the projectile used.
Taylor Wharton recommends that in the interest of public safety, only seamless steel high-pressure cylinders employing state-of-the-art technology (third and fourth generations) be used.
As a result, Taylor-Wharton believes that users of the Generation I cylinders should be warned of the potential fragmentation hazard associated with those cylinders and that the potential consequences of a failure are greater with Generation I cylinders.
Transverse Charpy Impact Test Results