When operators first discovered an internal leak in Valve V6272A, they weren’t particularly concerned. The valve was located beneath a large liquid oxygen (LOX) tank at an air separation unit (ASU), and from the outside, everything looked fine. Industrial oxygen components need maintenance or replacement from time to time, so operators and maintenance staff drained the tank and removed insulation to get a closer look at the piping system.
What they found that day was a most unwelcome surprise. Markings visible on the outside of the pipes (and later discovered inside) suggested a significant oxygen fire had occurred inside the system itself. Thankfully, the internal fire did not breach the piping of the tank. If it had, thousands of gallons of LOX would have drained out into an area where personnel may have been present. A rapid discharge of so much LOX could have led to life-threatening fires, explosions of organic materials, and even severe cryogenic injuries.
How and when had this happened? There was no way to know just how long the valve had been in operation in its current damaged state. Despite the mystery surrounding the event, answers had to be found. If such an incident ever occurred again, those involved might not be so lucky.
In order to find answers, our customer engaged the specialized forensic team at WHA International. For more than 30 years, WHA has performed failure analyses after oxygen fire incidents. Our engineers apply a uniquely data-driven approach, combining decades of forensic experience with the latest tools and technologies.
This analysis was challenging because the fire didn’t burn out of the valve, and the exact date of fire was unknown. Ultimately, WHA’s meticulous evaluation of the fire evidence would help reveal answers, resulting in critical knowledge for the client to develop safety improvements.
WHA utilizes a failure analysis approach specifically for oxygen fires, which is consistent with industry standard ASTM G145. The oxygen fire failure analysis is based on the fire triangle concept, which requires three legs for a fire to occur: oxidizer, fuel, and ignition.
The incident occurred inside one of the two main isolation (drain) valves below the LOX tank pictured above.
This fire incident was particularly challenging to investigate because damage was not discovered until well after the fire occurred. In 2014, without any knowledge of the fire event, operators determined that V6272A was leaking internally, so the LOX tank was emptied and insulation around the piping was removed for visual inspection and disassembly.
Upon discovering external surface patterns consistent with severe heat, personnel removed and disassembled the valve and piping for further inspection. The internal inspection revealed a significant oxygen fire had occurred, which burned large quantities of the piping and valve. Somehow, the fire extinguished and did not lead to burn through, which was an extremely fortunate outcome given the potentially catastrophic consequences.
Once the internal fire damage was discovered, additional details from the system’s history operation and inspection began to come together. Namely, large pieces of metallic flake debris were discovered in the strainer on LOX pump A in 2013 and 2014. These may have originated from the fire and migrated downstream, after fire extinguishment, as resolidified metallic combustion products.
Additionally, suspicious pressure spikes were recorded on February 20, 2010 and on August 12, 2011, but it is also possible that the fire occurred when the tank was first filled in 2009. Incredibly, this represents a 5-year window of time during which the fire could have occurred.
Upon discovery of the fire-damaged valve, WHA was engaged to provide forensic engineering services to help determine what went wrong.
The piping system around the discharge valve was documented during the initial inspection. External surface patterns on the upstream piping were among the first indicators that an internal fire event had occurred. Further internal inspection provided more clues to how the fire propagated.
To learn more about the original condition of the system and the after-effects of the fire, engineers examined the internal surfaces of the bodies of the sister valve located on the adjacent piping train (V6272B) and the manual isolation valve downstream of subject valve (V6403A). Both were inspected with UV light and no residues were visually detected. No fire damage and minimal particulate were observed.
A spare (exemplar) valve (identical to the incident valve) was also obtained from a storage facility on-site for subsequent inspection and reference.
During the inspection and disassembly both onsite and in laboratories, samples were collected for material characterization of organic and metallic materials (including residues and particulate). The material characterization was performed using laboratory analysis techniques, such as Fourier Transform Infrared (FTIR) spectroscopy and Energy-Dispersive X-ray Spectroscopy (EDS).
The following items were documented, and notable findings are discussed below.
The threads of the V6272A seat clamp screws were coated with particulate and grease-like residues. These threads were extracted with solvent and the FTIR analysis of the filtered solvent extraction identified a perfluorinated lubricant and a hydrocarbon-based contaminant with a carbonyl stretch. This FTIR analysis suggested this contaminant was different to the hydrocarbon oil found on the stem/wedge connection and likely represents a different hydrocarbon source.
Reconstruction of the evidence allowed development of potential scenarios under which ignition occurred and the fire progressed inside the system. The evidence of the V6272A and attached piping suggest that the valve was in the closed position when the fire occurred.
Following evidence documentation and reconstruction, the forensic team was able to consider three critical elements associated with an oxygen fire event: oxidizer conditions, fuel consumption, and ignition mechanisms.
Presence of an Oxidizer: The exact timing of the incident remains unknown, but the valve was assumed to be filled with LOX on the upstream side and possibly on the downstream side as well (LOX-full or partially full).
Pressure: Because of the high likelihood that the valve was closed at the time of the event, we can assume that no LOX/GOX (gaseous oxygen) flow occurred past the wedge seat of the valve at the time of ignition. Therefore, the maximum potential pressure at the time of ignition was 2 bar from the head pressure of the tank.
Temperature/Phase: If the ignition occurred during initial filling, then the temperature of the valve would be between ambient temperature and LOX cryogenic temperatures and the phase of the oxygen would be expected to be mixed phase. If the ignition occurred at a later date, the temperature and phase would be consistent with LOX on the upstream side, and likely GOX on the downstream side when the valve was closed because the valve was orientated vertically.
In an oxygen fire event, more flammable materials can spread fire to other less flammable materials in a “kindling chain.”
Based on the fire patterns, material specifications, and material characterization work, the following materials were considered present and exhibited burning in the fire event. They are listed in order from most flammable to least flammable (but not necessarily representing a kindling chain order) based on available oxygen compatibility data.
Based on the understanding of oxidizer conditions and fuel consumption, the following potential ignition mechanisms were identified:
The three mechanical impact ignition mechanisms presented above could translate mechanical energy at multiple locations inside the valve. Hydrocarbon-based materials can become extremely shock sensitive in LOX and can require minimal mechanical impact energy for ignition. In addition, the potential for accumulation of hydrocarbons at the closed seat (i.e., sump) via LOX boiling over many days was theoretically demonstrated.
Based on the evidence, analysis, and testing experience, the failure analysis team determined the most probable fire origin was at the upstream sealing surface and seat of the valve. The most probable ignition mechanism was repetitive mechanical impact (vibrations) at this location, causing ignition of contaminants.
The following observations provide the main support for these conclusions:
WHA’s failure analysis method is uniquely rooted in a data-driven approach — not simply anecdotal opinions. Our team of highly specialized engineers, chemists, and consultants leverage decades of forensic experience combined with the latest tools and technologies.
What’s more, our process is unique because we go beyond analysis results to help clients understand and interpret the data. Based on the failure analysis of this fire incident, the client and WHA generated the following takeaways:
Because of the detailed analysis and resulting recommendations, the client was able to return to operating their systems confidently. Operators were equipped with the awareness and practices they needed to prevent future incidents and help ensure safety.
Together, WHA and the customer worked to publish a detailed paper through ASTM international to help bring awareness of this unique case study to the industry at large. The above practices are now recommended for all similar industrial oxygen systems and components.
If you need answers following an oxygen fire incident or would like to analyze the fire risk of your systems, components, and materials, please reach out to our experienced engineering team.
For more details on this incident see “ASTM STP 1626 Failure Analysis of a LOX Valve: Internal Fire Only” by Gwenael J. Chiffoleau, Derek Miller, Brad Forsyth, Barry Newton. Produced by WHA International in partnership with Air Products & Chemicals, Inc.
Contact us today to learn more about oxygen precision cleaning, inspection, and verification services available from WHA.Contact Us