Second investigation report released regarding U Hawaii explosion

This steel tank ruptured during the explosion, which severed a researcher’s arm. Credit: Honolulu Fire Department

This steel tank ruptured during the explosion, which severed a researcher’s arm. Credit: Honolulu Fire Department

Earlier today, the University of Hawaii released a second investigation report into the lab explosion that caused a postdoctoral researcher to lose one of her arms. This report was by the University of California Center for Laboratory Safety; the first was by the Honolulu Fire Department. Still to come is the one by the Hawaii Occupational Safety & Health Division.

At the time of the explosion, postdoctoral researcher Thea Ekins-Coward had just finished combining hydrogen, carbon dioxide, and oxygen gases from high-pressure cylinders into a lower pressure container. The mixture was to be used as a feedstock to grow bacteria to produce bioplastics and biofuels.

I’ve only made it through a quick read of the technical part of the report so far, but here are some quotes:

This report was written to serve as a direct call to action for researchers, administrators and EHSO staff not only at the UH, but at all institutions of higher education that conduct research. The recommendations and lessons learned contained herein should be understood and addressed at all universities in order to help prevent laboratory accidents. (page 5)

From the beginning of February until March 16, 2016 the gas storage tank was filled eleven times with varying H2:O2:CO2 mixtures, all in the explosive range, with pressures between 37 and 117 psig (1 atm = 14.7 psig). The experiments were reviewed by the PI and the postdoctoral researcher weekly to discuss improvements of the bacterial culture conditions. They assumed the process to be safe since they stayed well below the maximum pressure for which the gas storage tank was rated (140 psig). The lab received a laboratory safety inspection in January 2016, however, the use of the gas storage tank was not questioned because the inspection used a typical checklist focusing on storage of chemicals and chemical waste, gas cylinder storage, laboratory fume hood certification, and documentation of training. (page 6)

In fact, before accepting the postdoctoral researcher into his lab the PI sent out a written
interview that contained the following question: “What was your duty and responsibility for the Environmental Health and Safety in the laboratories?” … Including safety questions in an interview enables a PI to examine general safety perceptions and attitude of a candidate, which is not commonly done. The Investigative Team is not aware of guidelines for incorporating safety questions into such an interview process, hence the safety concern reflects the PI’s genuine interest in laboratory safety. (page 9)

[The postdoc’s] interest in safety as it directly related to the experiments she conducted were expressed in meeting notes from 10/21/2015. These also reflect her safety training in the United Kingdom where COSHH forms (hazard assessments) are an established component of planning experiments:

  • “Gas supply – Please could we look into this together quickly. Would I combine H2 and CO2 in one cylinder and then add O2 separately?
    • Several papers have discussed the need for the O2 levels to below 4.0-6.9%-lower explosion limit. Is this something we need to consider? If we work with higher limits on a small scale are they safe to scale up?
    • PPE – would I need a flame proof lab coat? If so airgas sell them for $106.
    • Would all gases need to be supplied at the same pressure – due to different Henry’s constants, should just O2 and H2 be pressured?
  • Any SOP’s that have been written for the lab?
    • Should I write my risk assessments / COSHH forms or is there general lab ones that I just
      need to read over?
  • I have now completed lab safety, Haz, and Biosafety training. Can I now start in the lab?
    • Arrange to shadow tomorrow
    • Training on prep, storage and inoculation of microorganisms”

(page 10)

Initially, gases were supplied from three gas cylinders as described above and continuously sparged through the bioreactor. The postdoctoral researcher involved in the accident was tasked with developing a closed gas system to eliminate waste gas and optimize the process for industrial applications and commercialization. Gas mixture was added to the bioreactor and gas supply was closed off once the desired pressure was reached. As the bacteria grew and consumed gases, gas mixture was re-supplied through an automated feedback mechanism. (page 12)

When working on the drain plug [to fix a leak in the tank] the technician noticed rust falling out of the tank. Rust was also observed by the Investigative Team when remnants of the tank were examined (Figure 5). The rust was likely a result of water used for the hydrostatic testing of the tank. During his interview, the technician specifically commented that he assumed the tank would only be used for air and had he been told that it would be used for hydrogen gas, he would not have worked on the tank. He was aware of special requirements for hydrogen gas tanks. (page 15)

The events preceding the UH hydrogen/oxygen explosion bring to light several underlying problems which apply to research laboratories in academic institutions in general. The first is a drive inspired by real and perceived pressures to produce results in order to publish papers and obtain and maintain funding. Researchers’ careers are measured by and are dependent upon publication output and amount of funding they bring to the university. Secondly, innovation is at the core of scientific discovery as researchers constantly adapt or change experimental procedures allowing them to overcome limitations or challenges as part of their research process. Third, work with highly hazardous substances or processes is not necessarily perceived by many researcher to be high risk. These practices can lead researchers to place a higher value on experimental outcomes than on research safety. Furthermore, in contrast to highly hazardous biological materials, physical hazards lack regulatory oversight. The tragic accidents at Texas Tech University, UCLA, and now at the UH are stark reminders that the scientific discovery process, especially when highly hazardous chemicals or processes are employed, gravely underestimates the risks involved. (page 21)

One day before the accident, the postdoctoral research reported a “cracking sound” within the 1 gallon pressure vessel to her PI. Based on the pressure changes, visual observations, and smell, a combustion had occurred. Yet the researchers apparently dismissed the event as an anomaly even though it involved the same gas mixture and style of pressure gauge as the 13 gallon (50 liter) gas tank. The three researchers in the UH lab knew that hydrogen/oxygen gas mixtures were hazardous, so this near miss event should have triggered a shutdown of operations and initiated a thorough investigation of all procedures. But the significance of this near miss event went unrecognized and nothing happened. (page 23)

The force of the explosion caused severe injuries to the researcher including the loss of her lower right arm. The explosive force was so severe that the HFD HazMat 1 entry team was not able to find any part to reattach. Blood and flesh were found scattered in a wide area throughout the ceiling, floor and benches. (page 27)

The second postdoctoral researcher’s oxygen cylinder was blown against the wall (Figure 21). Fortunately, the regulator faced away from the wall when the cylinder hit as there was a significant chance for the brass cylinder valve to have been sheared off. Nevertheless, the force of the impact compressed the cylinder’s regulator spring causing an overpressure and spinning of the gauge. The polyethylene tubing attached to this oxygen cylinder was severed at the compression connection and the cylinder emptied into the lab. It was also fortunate that, an oxygen enriched fire did not occur due to this release. (page 29)

The originally intended gas mixture used for the gas storage tank was 70% H2:20% O2:10% CO2; however, gas mixtures were constantly adjusted to fit experimental needs. At the time of the explosion, the gas mixture in the gas storage tank was 55% H2:38% O2:7% CO2 (Table 1). Both gas mixtures are clearly in the explosive range and near stoichiometric concentrations promoting a reaction between the two gases and increasing the risk of explosion (Figure 25). … Although slightly in the fuel lean area, the estimated energy for the hydrogen/oxygen gas mixture contained in the tank at the time of detonation was approximately the equivalent of 70.5 grams of TNT. … Due to the low ignition energy (<0.02 millijoules) these hydrogen/oxygen mixtures can be easily ignited in a wide variety of ways. NFPA 53 table F.2.1 reported an even lower ignition energy of 0.0012 millijoules when H2 is mixed with O2. These levels are well below perceptible levels of 1 mJ that a human can feel. (page 34)

The Investigative Team concluded that the source of energy that initiated the detonation was a electrostatic charge accumulation that was discharged between the postdoctoral researcher to the gas storage tank or from the gas storage tank to the postdoctoral researcher causing a corona or brush discharge within the pressure gauge stem. We believe the same event occurred in the near miss event with the incubator. Neither the postdoctoral researcher, the gas storage tank or the incubator were properly grounded, therefore allowing charge accumulation. The discharge occurred when the postdoctoral researcher touched the metal housing of the pressure gauge as she attempted to turn the gauge off. (page 47)

More next week. Typos, formatting, and/or copy & paste errors are the result of composing this post after 6 pm PDT on the Friday before a long weekend.

Author: Jyllian Kemsley

Share This Post On


  1. Where’s the whole report from CLS?

  2. While it was indeed a terrible thing to happen to anyone, and we should certainly be sympathetic to her plight, the key paragraph (11th, I think) in this article begins with the most important sentence in the entire report: “One day before the accident, the postdoctoral researcher reported a “cracking sound” within the 1 gallon pressure vessel to her PI. Based on the pressure changes, visual observations, and smell, a combustion had occurred.” The alarm bells should have been ringing loudly at that point, and I am astonished that the “anomaly” was dismissed as insignificant. This accident was entirely preventable and serves as unfortunate example and warning to anyone working with combustible gasses. I hope the post-doc researcher can overcome this and continue her science career.

  3. I’m amazed anyone would mix H2 and O2 into a pressurized vessel. This SOP completely discounts the potential for human error.

    Every researcher I know understands that when working with flammable gases you do so in a way that minimizes the volumes.

    Why not use 3 tanks (H2, CO2 and O2) with mass flow controller, mani folding the gases together just before introduction into the culture?

  4. Many years ago I did my PhD study on pure hydrogen at very high pressure up to 800 bar. We only used small volumes of hydrogen in thick walled steel vessels. We compressed the hydrogen with a manual Mercury hydraulic compressor. At one point a colleague was compressing hydrogen and the manual compressor vessel with 4 cm thick walls split apart with one chunk flying right through a cement block wall. Very fortunately nobody was injured in the incident. An investigation showed that hydrogen has a tendency to penetrate into steel and weaken it. This is a cautionary comment to regularly inspect and safely pressure test vessels that contain pressurized hydrogen.

  5. This could have been avoided if they did a hazard review led by a departmental safety officer, which is pretty standard in industry. A person outside the immediate research group is less likely to bend to the will of the PI.

    Another tragic part of this is that she probably has minimal health care coverage, if any. Most likely she doesn’t have AD&D coverage to help with her expenses. In many regards AD&D is more important to young scientists/engineers than plain life insurance.

  6. becareful

  7. Thanks for sharing Jyllian. In my opinion, reading your excerpts and experience in both industry and academia, the impression that I get from the sentence “PI sent out a written
    interview that contained the following question: “What was your duty and responsibility for the Environmental Health and Safety in the laboratories?” … Including safety questions in an interview enables a PI to examine general safety perceptions and attitude of a candidate” is definitely not one of complete carelessness towards safety, but rather an inability to follow up on how safety measures should be implemented. What do you do when you have “near miss reports”? What sort of safety controls should be in place (hint: PPE should be the LAST resort not the first)? How do you follow up with each student/researcher? PIs need to be given the skills and training on how to do this within their group.

    Something as simple as filling hazard ID forms with group members is a worthwhile exercise for everyone. You’d be surprised which tasks actually come with the biggest risks.

    We cannot undo the tragic accident seen at UH, nor the accidents that happened before, but we can work to make sure they don’t happen in the future. Chemists in the academy: you have plenty of colleagues, former students, connections that went to work in industry – I bet some of them would be willing to help you make industry-like safety plans. I know I am one such chemist.

  8. Even at atmospheric pressure, the presence of oxygen in percentages higher than 20 lead to ready combustion of almost anything given an ignition source, especially at levels of >35%.

  9. I find it remarkable that anyone would mix hydrogen and oxygen in a sealed container. This incident points to a profound ignorance of basic practical chemistry. These incidents are becoming more and more common as career technicians are displaced by post-docs and research assistants whose broader knowledge-base is very limited. This would not have happened 20 years ago when centrally funded technicians were the norm.

  10. Be careful & please check gas storage tank that it is not rusted. Like this accident can happen anywhere, so have to take precaution before happening any other explosion.