A number of years ago the SAFTENG team was asked to aid in an investigation involving a flash fire that injured a worker. The State Fire Marshall was able to establish the flash fire was ignited OUTSIDE of the atmospheric vessel. We were challenged with finding out how and why. This incident was traced back to an IMPROPER change to the vessel which was NOT identified via the MOC that was completed for the change. Here are the details…
The process was built in the late 1980’s so it is relatively new and was a well designed process. A couple of years prior to this accident the business acquired a new product line. The business was relatively small so only one of the process vessels would be modified in order to support the new business. Because of quality concerns one of the raw materials for this new product needed to have its own delivery system to the process vessel. This meant adding a new line to the vessel.
This new raw material was NOT a flammable (nor any of the listed HHC/EHS in PSM/RMP). In fact, it was fairly low-risk material, so the discussion as to whether an MOC would even be needed began. Once the EHS professional informed them that the modification to the vessel was the driving force behind the MOC and not the addition of a non-has ingredient the first error was made and the dominos aligned!
From that point on, the engineer(s) had on blinders as to the impact this “change could have on the process safety and safety and health of the workers”. This vessel was PSM covered due to the processing of flammable liquids. The PSI stated this vessel was built to conform to several widely recognized RAGAGEPs for flammable liquid vessels, including NFPA 30 and API 2000. Yet this was the ORIGINAL design basis and as we learned this MOC changed a great deal in the way of process safety.
Mistake #1
Since this material was not a flammable liquid, the engineer(s) felt it was OK to “splash fill” this material into the vessel once it was full of a non-conductive flammable liquid. This was being done without the aid of any inerting of the atmosphere or any other attempt to control static. Keep in mind though that there was substantial evidence that the vapor was ignited OUTSIDE of the vessel, so even though this was a poor design and NOT to code, it did NOT play a role in this particular incident.
Mistake #2
Since this material was not a flammable liquid, the engineer(s) lost sight of flammable liquid processing RAGAGEPs and it was this loss of focus that leads to the presence of an ignitable mixture outside of the manway. The original design of this vessel included a vent sizing based on two (2) 1.5” inlets. Both of these inlets originated from a solvent delivery header which meant that even though there were two inlets, ONLY one inlet could be used at a time. On top of this, the normal vent was sized at 2” by either luck or an engineer providing some safety margin in the vent design. We could not find any PSI that explained the normal vent sizing!!
1910.106(b)(2)(iv)(b) covers our vent sizing MINIMUM requirements for flammable liquid tanks. There are some that will lay claim this requirement applies ONLY to “storage tanks”. They may be right and may beat OSHA at the compliance game, but I am more about process safety and good engineering so I say it applies to any vessel handling flammable liquids! The standard states…
Normal vents shall be sized either in accordance with:
1910.106(b)(2)(iv)(b)(1) The American Petroleum Institute Standard 2000 (1968), Venting Atmospheric and Low-Pressure Storage Tanks, which is incorporated by reference as specified in Sec. 1910.6; or
1910.106(b)(2)(iv)(b)(2) other accepted standard; or
1910.106(b)(2)(iv)(b)(3) shall be at least as large as the filling or withdrawal connection, whichever is larger but in no case less than 1 1/4 inch nominal inside diameter.
(b)(3) is the one that I’d like to highlight. It is a GREAT rule of thumb in process design and could have spared the life of many a storage tanks over the years! Basic design states the “breathing vent” will be as big as the largest connection, but never smaller than 1/4” inside diameter.
If we look just a little further down in the standard we come across 1910.106(b)(2)(iv)(d), which states…
If any tank or pressure vessel has more than one fill or withdrawal connection and simultaneous filling or withdrawal can be made, the vent size shall be based on the maximum anticipated simultaneous flow.
BULLSEYE! In the incident, we found that the facility’s MOC did not consider the change as one that could impact code design basis. The new inlet was added and the P&ID was updated as a matter of “PSM paper work” (their statement – not mine). Since this new raw material had its own delivery system (for quality reasons) it was permitted to be in operation at the same time as the flammable liquid was being filled (i.e. “simultaneous filling”). With a 2” vent, but a now having a total of 3.5” total inlets, the 2” vent was nearly 50% too small. The vapor seeked the least resistance path and this was the UNSEALED manway cover. The facility had NOT classified their process areas PROPERLY and thus the 5’ bubble around the manway was not treated as a Class 1 Div 1. The facility also made little to no attempt to control static electricity so we could never identify the actual ignition source; however, our in-place testing proved to us how the flammable vapor found its way outside of the vessel in the work zone of the employee.
Although a change may seem minor since it is not directly involving a hazardous material, we can not loose site of how the change will IMPACT the safety and health of the workers (1910.119(l)(2)(ii) and the impacts it can have on our design/codes. I would be amiss if I did not mention that the employee was wearing Flame Retardant Clothing (FRC) which minimized his/her injuries. Although it is the last line of defense, FRC can and has proven to be invaluable in a flash fire event.