High Flash Point Liquids – does the approach for hazardous area classifications need to change?

High flash point liquids are widely used throughout the industry for various purposes such as heating, lubrication, and hydraulics.

Often high flash point liquids are classified as “non-hazardous” as they are handled at temperatures below their flash point (and if there is no way of heating them towards the flashpoint) and at atmospheric pressure. However, in some cases, they are stored or used under pressures which may cause them to generate a flammable mist with behaviours similar to flammable gases, which may be ignited, even at low temperatures.

For hazardous area classifications as part of DSEAR (Dangerous Substances and Explosive Atmosphere Regulations) assessments, the Energy Institute Model Code of Safe Practice Part 15 (EI15) provides an approach to determine the extent of the hazardous area resulting from the pressurised release of a high flash point liquid.

However, recent studies and reports published by the HSE have indicated that these zones may be over-conservative in some instances, resulting in potentially unnecessarily high costs with regards to the avoidance of ignition sources in hazardous areas.

We interviewed Gexcon UK Senior Process Safety Engineer, Hannes Engel who will be presenting a paper entitled ‘Flammability Testing for Heavy Oil Mists’ at this year’s Hazards 31, and Gexcon AS Fire and Explosion Testing Engineer, Adam Armstrong, who worked in collaboration on this project.

Q: Could you tell us briefly about yourself and your experience within the process safety field? 

Hannes:I am a Senior Process Safety Engineer with Gexcon UK and have been with the company for 3 years.  My involvement with process safety started when I was working as an Energy Engineer in a previous role where I would design control systems for refrigeration systems. This enabled me to be part of a HAZOP team, first as a scribe and later as a facilitator.  I enjoyed this work so much that I changed my focus and now I am involved in Process Safety on a daily basis conducting DSEAR Assessments and HAZOPs full time.

Adam: I am an engineer with the fire and explosion testing department of Gexcon AS and have been with the company for 2 years. My involvement with process safety started with Gexcon where I primarily work with instrumentation and data analysis for our large-scale explosion testing facility.  I am also part of the team at our jet fire testing facility working on projects involving dust and gas explosions as well as equipment testing for ATEX certification.

Q: What is this year’s HAZARDS 31 presentation about?

Hannes: The presentation is about a series of tests that were carried out in Norway as part of a project for a large power station client here in the UK who uses heavy fuel oil in boilers on site, they also have hazardous areas associated with their distribution pipework.  The tests investigate the credibility of generating flammable mists when high flashpoint liquids are released through a small orifice under pressure. 

Q: What was the motivation for choosing this topic?

Hannes: High flashpoint liquids are widely used in industry, for example for heating purposes as a fuel as well as for lubrication purposes and in hydraulics units.  When conducting a Hazardous Area Classification as part of a DSEAR assessment, there is always the question of whether the scenario of generating a flammable mist is credible for such applications.  Current guidance on this topic is not always clear on this, which can result in uncertainty when conducting a DSEAR assessment, potentially resulting in assessments that are overly conservative, or even worse not reflect the scenario of a flammable mist generation at all. 

Q: Could you tell us briefly about the experiments? How, where and by whom was it being conducted? 

Adam: The experiments were carried out at Gexcon’s test facility at Steinsland, which is located on the island of Sotra outside of Bergen, Norway.  Myself, along with the assistance of Chris Joakim Vågen were responsible for conducting the experiments which involved testing the flammability of oil mists generated with varying nozzle sizes, reservoir temperatures, and distances to the ignition source.  

There were two types of tests, one where the mist was generated in a relatively open space, with plenty of room for the mist cloud to form and move, and one where the mist was contained in a small cylinder, so that the mist generated would be confined to the area of the ignition source.

Heavy oil was filled into a heated reservoir, where the oil would be heated to either 65 o C or 145 oC. This reservoir would then be pressurized to a given pressure, and an oscillating electrical spark would be turned on at various distances away from the oil mist nozzle, located below the oil reservoir. A pneumatic valve would open between the oil reservoir and the misting nozzle, allowing the pressure to press the heated oil through the nozzle, and out as a mist spray towards the ignition source. If any flames were noted as the mist travelled over the spark, it would be determined that the distance was close enough to be hazardous.

For the open space tests, assuming that there was oil mist ignition, the next test would then be carried out with the same conditions, but with the ignition source moved 25 cm further away. This would continue until no ignition was observed, and then that distance would be repeated 10 times to attempt to see how probable it was that this distance would ignite the mist. In many cases, several of the tests would still ignite, meaning that the actual minimum “safe” distance is slightly further out. For the confined space tests, there would not be any change to the ignition distance.

Q: What caught your interest the most from the experiments? 

Hannes: The most interesting part was to compare the test results with the current hazardous area classification guidance as provided in EI15.  The tests showed that no ignition occurred for release conditions which would usually be classed as “non-hazardous” when following current guidance, which is of course a good result as it demonstrates that for these specific release conditions the guidance is not too optimistic. 

Equally when an ignition occurred it was interesting to compare against current guidelines in relation to distances where ignition took place. In some cases, the guidance matched the test results quite well, but in others, it showed the guidance to be overly conservative.

Adam:What I found most interesting about the testing was attempting to fine-tune the conditions of each test, in order to be absolutely certain that a given distance was run at the worst-case scenario. I quickly found out that there are many factors involved which may not seem obvious at first, but all play a central role.

For example, when the dispersion pressure was increased it seemed logical that the mist would be pushed further away and capable of travelling a greater distance to an ignition source which would result in an ignited oil.  However, this was only partially true.   Eventually, the pressure becomes high enough that it begins to increase the overall volume of the mist cloud, which is then heavily affected by drag.

What I ended up seeing was that the 20 barg dispersion pressures were capable of longer ignition distances than the 50 barg driving pressures depending on the dispersion nozzle used. This relationship between driving pressure and nozzle size had never occurred to me, and I found it quite interesting to explore.

Q: What was the biggest challenge during the experiment? 

Hannes:Some of the test results were surprising and required additional investigation.  So did the maximum ignition distance reduce with increasing pressure in some cases.  After many additional tests, it was concluded that the higher dispersion pressure resulted in a wider spray angle which likely contributed to reduced ignition distances. 

Adam: What made the oil dispersion interesting was how it defied expectations, which leads to significantly more testing than you had initially planned for. To be certain that the anomalous results are due to real phenomena, and not an underlying issue in your systems such as nozzle clogging, incorrect temperatures, or leaks, I would repeat the same experiment 4-5 times trying to take diagnostics on the system before finally accepting that the system was fine, but the results were simply what they were, and it was a combination of the parameters of the test which was giving these rather unique results.

Q:  What are the next steps that should be taken after this study? 

Hannes:  Additional experiments further investigating the credibility of generating a flammable mist for varying high flashpoint fluids and varying release conditions would further inform current guidance and potentially reduce uncertainty. 

Adam:  There are so many different avenues of testing available to determine the underlying principles at work. Everything from mapping nozzle dimension to dispersion pressure ignition lengths, determining mist droplet size at various pressures and temperatures, to even something as basic as determining the relationship between mist throw (that is, the linear distance from the nozzle to where the cloud stops) and cloud width/volume for given pressures and nozzles.

A more detailed presentation of the “Flammability Testing for Heavy Oil Mists” Paper will be presented during the Hazards31 Virtual Process Safety Conference which will be held on 16 – 18 November 2021. Don’t miss Hannes’ presentation and join the event!