Tuesday, April 19, 2016

More details on the UH-Manoa incident: explosion probably due to sparking pressure gauge

Credit: KHON (Picture of "before" by Jian Yu, "after" by the Honolulu Fire Department)
(Via Jyllian Kemsley's tweet about Honolulu TV station KHON's report). A partial PDF copy of the report from the Honolulu Fire Department about the explosion in a University of Hawaii - Manoa biofuels laboratory is pretty clear as to the cause of the explosion: there was a pressure gauge that was not explosion proof that sparked in a flammable gas environment. From the report: 
Accidental causes were considered and only one probable cause remained. 
The accidental cause of this explosion was caused by the detonation of compressed gasses to include: Hydrogen, Carbon Dioxide, and Oxygen within the air tank. A digital pressure gauge used to check the pressure within the tank was not rated or designed (not Intrinsically Safe) to be in a flammable gaseous atmosphere. When the OFF button was pressed, an electrical arc/spark created within the gauge detonated the flammable gas within the tank causing the explosion. 
Level of Certainty: Probable, with a reasonable degree of fire investigative certainty.
Please forgive the gruesome details that I am going to share next, but I think they're essential to understanding the incident:
On this particular day, she moved and filled the tank with a set amount of gasses using the Ashcroft 300 psi digital gauge that is battery operated. This gauge is a push button type for ON and OFF. When she disconnected the hoses used to fill the tank, she checked the pressure in the tank to verify the amount to be 117 psi. She then pressed the OFF button and the tank exploded. Prior to the explosion, she did not hear any sounds of escaping or leaking gasses from any of the fittings or pressure relief valve. 
She did not lose consciousness or hit her head; she was aware that she lost her arm in the explosion. She couldn’t open the door to the lab, the door was stuck closed. A person by the name of Savannah was there to help get the door open and help her out of the lab. 
She added that earlier in the week, she was conducting another experiment using a smaller one gallon size air tank assembly nearly as identical as the one that failed using similar components to include the Ashcroft 100 digital pressure gauge and the premixed gasses. The tank pressure was set to 27 psi. After reading the gauge, she pressed the OFF button and a small internal explosion occurred. She related to me that there was evidence of a soot and smoke stains. 
Static shock also appeared to have been a problem as Ms. Ekins-Coward would get shocked on occasion when touching the tank. 
She brought this information to the attention of Mr. Yu who she said told her don’t worry about it.
What immediately bothers me about this incident is that I have been simplistically trained to think about the "fire triangle": that oxygen, fuel and a heat source are needed for a fire. In this case, because the Yu Laboratory and Dr. Ekins-Coward were running mixtures of hydrogen, oxygen, you have both fuel and oxygen, so most of the preventative measures should have been aimed at reducing the risk of a spark. That there were near-misses is supremely tragic - this didn't have to happen.

(Readers, what do you think here? Is there another root cause I'm missing? Should attention have been paid elsewhere?)

To make this even more interesting, it appears that we have a disagreement as to who designed the system. I think this sort of thing is bound to happen during police investigations; two people will disagree about something key to the event. From the report's interview with Dr. Yu (emphasis mine):
The following was related to me by the named individual above: 
The victim Thea Ekins-Coward was working as a Post-Doctoral Fellow trained in Marine Science and a Chemical Engineer. 
She was hired by Dr. Yu October 2015 to conduct research in bioplastics and biofuels. 
In response to my questions he related that: 
The tank that failed was Ms. Ekins-Coward's design. She bought the equipment (tank, digital gauge, pressure relieve valve, and fittings) between November 2015 and January 2016. The tank was to have been rated at 10 bar or 150 psi. When the tank was assembled with its parts, a pressure test was done using the buildings air which produced 91.2 psi. Several leaks were detected. So the tank assembly was taken to the Universities maintenance for help in stopping the leaks. 
This particular tank was used to contain a mixture of gases to include: 70% Hydrogen, 10% Carbon dioxide, and 20% Oxygen in that order and to a normal operating pressure of 50 psi. 
The tank was not grounded and was normally moved from a stored location to areas where it could be filled. The tank would be moved approximately 3 feet to fill it with Hydrogen and Carbon Dioxide, then moved approximately 13' so it can be filled with Oxygen.
And here's the relevant paragraph from the interview with Dr. Ekins-Coward:
The air tank and assembly that failed was purchased between November 2015 and January 2016. The parts were assembled and pressure tested with the building supplied air. This output of air was 91.2 psi which was sufficient enough to cause leaks at the connections so the tank assembly was taken to the Universities maintenance shop where they helped stop the leaks. The design of this tank assembly was that of Dr. Jian Yu. The tank was rated at 145 psi.
So, it's time for some questions that will ultimately need to be answered:
  • Who designed the tank and selected the pressure gauge? 
  • Is it accurate that both the PI and postdoc were trained as chemical engineers, as brief Google searching seems to indicate? 
  • What is the role of the institution and its EH&S office in this? Did the EH&S office know about the Yu Laboratory's experimental systems? 
  • This was clearly an unsafe experimental setup - will the comments about this experiment being safely performed since 2008 be walked back? 
  • What disciplinary actions are going to be carried out by the department or institution? 
  • Will there be civil penalties from the state occupational health and safety regulatory agency or from the local fire authorities?  
  • Will there be criminal penalties? 
The current lesson to be learned: when dealing with flammable gas mixtures, inherently safe equipment will go far to prevent fires, explosions and serious injury.

UPDATE 160419 4:49 PM: Jyllian has a story up about the Honolulu Fire Department report at C&EN's website; some background on the experiment follows:
The gas mixture was “food” for bacteria being used to produce biofuels and bioplastics. Ekins-Coward was working for the Hawaii Natural Energy Institute under researcher Jian Yu. A 2013 paper by Yu indicates a set-up in which gases are plumbed through a mixing device called a gas proportioner directly into the bioreactor (Int. J. Hydrogen Energy 2013, DOI: 10.1016/j.ijhydene.2013.04.153). The gas gauge identified in the paper is an “intrinsically safe” model designed to prevent ignition. 
But after Ekins-Coward started in the lab last fall, she purchased a 49-L steel gas tank, a different gauge not rated as intrinsically safe, a pressure-relief valve, and fittings, and she put them together, Yu and Ekins-Coward told fire department investigators, according to the report. Ekins-Coward would add the gases to the portable tank, which would then be connected to the bioreactor. She was using a mixture of 70% hydrogen, 25% oxygen, and 5% carbon dioxide for her experiments, the report says. 
In the week before the incident, a similar set-up with a 3.8-L tank resulted in a “small internal explosion” when Ekins-Coward pressed the off button on the gauge, the fire department report says. 
Also, she links to a longer version of the report.  


  1. The key point seems to be who designed and made the equipment. I wonder if they'll subpoena lab notebooks - I imagine knowing when the apparatus was first used might clarify who came up with the design and implementation (if it was there before Ekins-Coward got there, she didn't design it, while if it was first used when she was there, then it could either have been her advisor or her).

    Considering how advisor-advisee conflicts normally work out, I am unoptimistic.

  2. Regardless of who designed the experimental setup up, should the following statement have any validity, I think Dr. Yu is in trouble:

    " Static shock also appeared to have been a problem as Ms. Ekins-Coward would get shocked on occasion when touching the tank.

    She brought this information to the attention of Mr. Yu who she said told her don’t worry about it."

  3. If they were really both chemical engineers, then they should have known better. 2 seconds of google searching shows this is in the flammable range for hydrogen, and any components of any sort of hydrogen gas system need to be Class 1 Div 1.

    1. I'm not a chemical engineer but I've worked with many and I would venture a guess that they would say that the 70% H2/20% O2 gas mixture is inherently dangerous and that making such a mixture is simply unacceptable practice. In fact, I would bet money that any process engineer worth his or her salt would tell you that you just plain don't do that. Any ChemEs here?

    2. Eric Reed: Then would you agree that the likely more fundamental problem here was the purposeful preparation of an explosive gas mixture that never should have been conducted in the first place?

    3. Exactly. To quote the late Trevor Kletz, "what you don't have can't leak". Or explode in this case. It would have been far safer (and simpler experimentally) to buy three mass flow meters and deliver the gases directly to the bioreactor in the correct proportions.

  4. Any competent electrician would have thrown a red flag on this one. Not using intrinsically-safe components around hydrogen is just plain stupid.

  5. Not using intrinsically-safe components around *high-pressure hydrogen and oxygen together in a known-to-be leaking bomb* is just plain uninformed/naive/terrible.

    1. Looks to me like it was not due to a leak - the explosion was located in the interior of the tank. The tank contained a flammable mixture and was ignited from the interior portions of the electronic gauge. The write up suggests that the leaks had been located and fixed.

  6. What research was being conducted that necessitated a pressurized mixture of hydrogen and oxygen?

  7. If EH&S at Hawaii is anything like EH&S at other universities I have been in, then I doubt they knew about this, or had anything useful to say. In my (limited) experience they tend to be a bit toothless and more for the university to say "See, we trained people!" than to actually promote a culture of safety.

  8. Here's the part that has given me pause: "She related to me that this is the first time the use of premixed gas in a portable tank to conduct experiments was being done. In the past, each individual gas would have a separate hose line and be premixed at the reactor eliminating the use of the portable tank."

    So while growing the bacteria with this gas mixture has been done in that lab for quite some time, it previously hadn't been done this exact way. I'd like to see a full walk through of how they previously did this set up, and what changed doing this premix. And why they changed to this premix. Pretty much, I want to know the nitty gritty so when I talk to the labs I cover that work with gases like these, we have a better idea of how to learn from this accident.


  9. I think that trying to assign blame or criminal liability on one person or entity isn't very productive here - OSHA will ascribe some responsibility and collect some penalty, but will not get to the root of the problem. While Yu is ultimately responsible for the work done in his lab, in reality, the experimental design was likely a collaborative one - one of them may have had the vision and the other performed the implementation - and my guess is that together they discussed the set-up and execution. The failure here, once again, is that the academic community has not learned the importance of robust risk assessment and incident investigation processes. No matter how brilliant the researcher or the PI, they usually can't think of everything (even if the errors seem obvious in retrospect). If multiple group members ruminate on an experimental plan while sitting in a room together, there is a much better chance of identifying the hazards, quantifying the consequences, and implementing risk reduction steps where needed. If the pressure gauge was still missed in that process, there should have been an investigation after explosion #1 - note that an 'internal explosion' is not a near miss, it is an incident. Not sure how to make these processes mandatory, but they are long overdue.

    1. Hi, Beth:

      Thanks for your informed comment. I should note here when I ask "Will there be criminal penalties?", the emphasis is on the "will", i.e. does anyone predict that (say) the Honolulu district attorney will see fit to bring charges, and not on "should there be criminal penalties?" (which is a discussion that will not result in a testable hypothesis.)

      - Chemjobber

    2. Sorry. I guess I wasn't assuming that blame was going to be sufficient but that because of the injuries and costs, someone would be assessing them (and that the point of disagreement would be important in doing so).

    3. So, just a few thoughts to contribute here:
      1. Did anyone field verify the design or current set up?
      Given my experience with academia, I doubt it, but if someone has actually been checking the system after construction it may have caught this problem.
      2. I certainly agree a group should have evaluated the design and its implementation, but its also important to have the right constituents in that group - an experienced engineer/chemist, electrician, or even a decent EHS employee could be valuable members of such a group. That is to say, a group of students could easily miss something.

  10. If you are working with something inherently dangerous in lab, then you should have an SOP in place. We are required by our institution's EHS to write explicit SOPs and properly train students performing dangerous tasks/handling reactive chemicals (and subsequently document said training). While I have never worked in industry, I imagine they are even more thorough with their safety training.

    I'm more inclined to assign the majority of the blame to the advisor, given that this was not the first time this inherently dangerous activity was being performed. At this point, it's all hearsay since nothing was likely ever documented...

  11. I wonder if the intrinsically safe pressure gauge cited in the 2013 paper was selected on purpose. If it was, then it's not clear to me why Drs. Yu and Ekins-Coward didn't realize the current gauge was inappropriate, especially after the small explosion in the 3.8-L tank.

  12. Is explosion proof electrical equipment not standard in labs? All the stuff in my lab was explosion proof, no sparks anywhere. And earthing metal tanks and containers, well that should also be standard.

    1. If your area classification does not require it, many people will not spend the extra money

  13. But these explosions in microbiology labs growing microbes under hydrogen have been happening before - I remember a report few years back, nobody got injured in that explosion but the lab was demolished. One would think the risk would have been recognized by now...

    1. It would be interesting to know if the people involved here even had a clue that the gas mixture they prepared was almost certainly in the explosive range (see for example http://onlinelibrary.wiley.com/doi/10.1002/ceat.200403174/abstract, especially Table 2). Maybe biochemists/engineers just don't think about these things. I worked for a very large US chemical (e.g., petrochem) company that bought a biotech company. The biotech company was piloting an aerobic (O2 required) fermentation process to make a volatile hydrocarbon. The petrochem engineers from the chemical company realized that the mixture of O2 required to run the fermentation and the evolved hydrocarbon constituted a flammable/explosive mixture. The engineering requirements to render the process safe killed the economics and thus the project. Unfortunately it wasn't until after tens of millions of dollars had been spent on the R&D and pilot facility. In this present case the researchers were working with an explosive mixture and, as such, their process never would have been practiced - at least not the way they were doing it.

    2. If you're thinking of the University of Missouri explosion, that was about anaerobic glove box procedure. The bacteria needed an anaerobic environment but weren't grown under hydrogen.

    3. A couple of people were injured in that one: http://blogs.sciencemag.org/pipeline/archives/2010/06/30/how_not_to_do_it_hydrogen_gas_mixtures

      Still, it's a pretty good warning that if you're using H2 at concentrations within its explosion limits, you ought to be really careful.

  14. Speaking of "intelligence-essential facts" in the previous blog post - Prof. Yu should have his PhD revoked for being unaware that a mixture of H2 and O2 is essentially a bomb!

  15. This comment has been removed by the author.

    1. Professor Bradley's comment was accidentally deleted by him. Here is his comment:

      This has been good coverage of a horrifying accident. For those of us running labs which make use of hydrogen gas mixtures, this gives important food for thought and should certainly prompt a review of safety procedures. A few more points:

      (1) My immediate impulse upon hearing about a tank full of an H2/O2 mix was-- that's the Hindenburg, ready to blow. You don't need a PhD for that level of general safety knowledge.

      (2) Not every PI running a lab is a callous monster only interested in results at any cost. Some of us actually care quite a bit about the well-being and safety of our students and colleagues, and strive to maintain a safe and healthy working environment. However you have to pay attention; the worst part about this story is the fact that there had already been a precursor explosion, which was more-or-less ignored.

      (2) I agree that university safety people sometimes pass the buck; they offer very general training, but they often don't have the detailed knowledge about the specific hazards in a given lab (to be fair, how could they ? The range of lab work being done and possible hazards at a major research university is too great). So the most important thing is to have a proper functioning safety culture within a lab- this starts with the PI

      Mike Bradley

      Michael P. Bradley, PhD (MIT), P.Eng.
      Associate Professor, Physics & Engineering Physics
      University of Saskatchewan
      Saskatoon, SK Canada S7N 5E2
      Tel. 306-966-6399

    2. Mike Bradley's comments are excellent. A few things to think about in retrospect. One, did the PI or his post-doc consider the ramifications of using such large volumes of an explosive gas mixture? If so, why did they proceed and what precautions should have been taken. Two, wasn't there a better set up possible that did the gas mixtures at the point of the experiment itself, rather than having a very large reservoir of a gas mixture well inside hydrogen's explosive range? Three, did they do an energy calculation? I did a quick, back of the envelope calculation using the gas mixture, pressure, and hydrogen content and the results, in stored energy, are quite startling. They basically had a bomb in their lab ready to go off. That is anything but "routine".
      Those of us who have students or junior scientists in our care need to pay attention.

  16. Those who do not learn from history are bound to repeat it! Mixing rocket fuel? They should have contacted NASA on how they do it on ton scale! Very tragic.

  17. With so many fascinating avenues to explore without making bombs, why even do such a project? Why fund someone doing such work? Why give such a position to someone doing this kind of work?


looks like Blogger doesn't work with anonymous comments from Chrome browsers at the moment - works in Microsoft Edge, or from Chrome with a Blogger account - sorry! CJ 3/21/20