08 Mar Response by Dr. Alexander Morgan to “Flame retardants in UK furniture increase smoke toxicity more than they reduce fire growth rate” Chemosphere 2018, 196, 429-439.
The text below entitled “Response by Dr. Alexander Morgan to “Flame retardants in UK furniture increase smoke toxicity more than they reduce fire growth rate” Chemosphere 2018, 196, 429-439” is the full editorial transcription from the Fire Safety & Technology Bulletin published in February 2018.
The Fire Safety & Technology Bulletin is available by electronic subscription only and the editors are Marcelo M. Hirschler (GBH International, US) and Stephen J. Grayson (Interscience Communications, UK).
Dear I recently read an article related to flame retardant use with a title that I find incorrect. Given the publicity the article is receiving, I feel a need to comment on it. While there are parts of the paper I find to be done carefully, and I agree with, there are some fundamental flaws in the paper’s experiments that do not support the title. Indeed, I believe some results in the paper show that the flame retardants are working as intended and do actually reduce fire growth rate when tested against the ignition source they were designed to protect against. A better title for the paper would be “Flame Retardants in UK Furniture Can be Overwhelmed by Stronger Fire Sources and May Not Address All Fire Hazards”.
First, let me comment on the parts of the paper with which I agree. I do believe that fire safety regulations should change as we discover new fire hazard scenarios, or discover data that suggests the approach should change. There have been numerous studies showing that heat release is the most important consideration in enabling escape from a burning room, and those studies show that heat release causes a high number of deaths, probably more so than toxic gases. However, for those who cannot easily escape a burning room (very young, disabled, very old), I suspect toxic gases can be equally lethal, if not more so, than the heat release issue. So, in that aspect, I agree with the authors that smoke toxicity is something worth considering in fire safe material development, and is probably something that we as fire safety scientists should take a look at and try to develop fire safe materials that lower heat release and toxic gas emissions. Materials that inherently char and have inherently low heat release are capable of doing both, but they’re currently relegated to high performance (high cost) applications such as aircraft and passenger train cars. I also agree with the authors that halogenated flame retardants, due to how they work in a flame, will create more non-combusted products and more smoke. This is very well known and not a surprise. However, if the sample never ignites, then much of these smoke and toxic gas issues go away. I’ll return to this point later in my discussion, as this is important to understand regarding the full-scale experiments the authors conducted. Further, I agree with the cone calorimeter (bench scale tests) experiments that the authors carried out, in which they made miniature furniture assemblies of padding and surface fabrics to see how the combined system behaved when exposed to a forced combustion fire hazard scenario. Indeed, testing of combined foam and fabric systems is more relevant to real world fire behavior because it is how everything works together in a fire, not the individual components, that matters. The results from the cone calorimeter tests show, not surprisingly, that the highly flame retarded material has the lowest heat release out of all systems tested (a good thing), while having the 2nd highest amount of CO released and 2nd highest amount of HCN released (not a good thing). But for the reasons I described above, this is expected – flame retardants inhibit combustion, so one should expect to see higher amounts of incomplete combustion products. The less flame retardant present, the hotter the potential flame from polyurethane, and therefore the flames get hot enough to start burning up some of the toxicants. By cone calorimeter data alone, we are presented with an interesting finding: heat release is lowered, so flame growth is lowered, but the tradeoff is an increase in toxic gases. Not surprising to those of us in the field, but certainly a sign that we must work toward a better performance: equally low heat release/reduced flame growth rate, and lowered smoke toxicity. I do have some questions on how the authors collected their gases from the cone calorimeter, as I know from personal experience this isn’t easy. I did not see enough details in the experimental section to answer my questions, but I’ll trust the authors cleaned out their ductwork between runs and had the necessary heated transfer lines to collect emissions at the right place in the cone calorimeter exhaust system. One final comment that the authors touch on, but don’t really drive home, is about some of the flame retardants used in this study. While the authors expressed surprise that decabromodiphenyl ether was found in the products, when it was thought to be absent from the market, this should have been a call for more product inspections on imports to the United Kingdom (UK). I feel strongly that as chemicals are found to have negative persistence, bioaccumulation, and toxicity data, they should be removed from use, and we should enforce those regulations that require them to be removed from use. There are better fire safe materials to use which provide meaningful fire safety and minimize environmental impact.
When looking at the large-scale tests, however, I find experiments that are not reproducible. Further, the results appear to have artifacts of the actual experimental conditions and are not relevant to real world conditions. Finally, the large-scale experiments, by the authors’ own admissions, use an ignition source stronger than that called for in the UK legislation. This is an important point from a fire safety perspective and undermines the authors’ arguments that flame retardants increase smoke toxicity more than they reduce fire growth rate. Let me walk through the full-scale experiment and the problems with it that make the title misleading.
First, the authors used a modified shipping container as the “room” in which to conduct their tests. I believe there are numerous labs throughout the European Union that could have provided a standard ISO room for fire testing, to understand how the fire grew and to control the air flow rates, which would in turn affect the fuel/air mixtures and toxic gases/smoke released. A shipping container is not a representative space for mattress use. I suppose people could live in shipping containers, but I would think they are the minority in the UK, and so the shape of the container and how or whether the steel walls affected heating of the room isn’t considered in this analysis either. The authors state that for their testing setup “The outlet was twice the area of the inlet so that only cool air flowed into the container through the inlet, and only hot effluent left through the outlet.” The authors state later in their paper that they did not directly measure gas temperatures going into and out of their modified shipping container, but instead assumed them from another paper they cited. Given the nature of their setup, this should have been validated with actual measurements. But more importantly, the authors indicate in their paper: “Reasonable reproducibility was obtained for each pair of apparently identical mattresses, despite the different weather conditions and wind directions on the day of each test. The UKFR1 and CHFR1 tests were the only two tests performed on the first day, in significantly windier and wetter conditions; visual observation showed the wind moving the crib flame away from the back of the sofa on the first two tests; they showed longer ignition delay times than the subsequent tests where calmer, more stable weather conditions prevailed, until the end of the test programme.” Humidity is well known to affect fire growth and smoke emissions, as are air flow rates and air temperature. This is why controlled rooms are used: to see heat release and emissions properly, as a function of controlled conditions which would be relevant to actual fire hazard scenarios. The conditions in this experiment are uncontrolled, and therefore do not produce reliable conclusions. This is why there are both ASTM and ISO standards regarding room burn tests, to ensure one is testing a representative room under controlled conditions, with artifacts removed.
Second, there is the issue of the ignition sources used in the large-scale test. The authors tried igniting all 4 of their mattress compositions with the “No. 5 crib”, which is the wooden ignition source currently used in UK regulations. This regulation tests individual components, not composite systems of foam and fabric, so it has its limits. Very interestingly, the authors found that none of the samples could be ignited by the No. 5 crib ignition source, and therefore, there would be no toxic smoke, or than the small amount generated by the wood crib. However, due to the above issues with the uncontrolled room, one wonders if any of the samples would have ignited with the No. 5 crib ignition source, thus changing the outcome and conclusions of the paper. Had the flame retarded samples resisted ignition and the non-flame retarded samples ignited, that would tell us that the flame retardants did reduce fire growth rate. If, in a controlled room, none of the samples ignited with the No. 5 crib source, that would tell us that the existing UK test is only appropriate for protecting against fire on exposed polyurethane foam, and is not appropriate for real world furniture composed of foam and fabric assemblies. The authors further muddled the conclusions of this paper by moving to a No.7 wooden crib source, which has a much higher intensity fuel load (125 g of wood to burn, vs. 17 g for the No. 5 wooden crib), to get the samples to ignite. Now the authors have forced the samples to burn, against a more rigorous fire source, in uncontrolled conditions, thereby moving outside the existing fire hazard scenario that the UK regulations address. I would suggest that the failure of the No. 5 crib to ignite these mattresses proves that (a) the large-scale experiments are not controlled or meaningful and conclusions from them cannot be relied upon; and (b) maybe the real issue here is that the existing UK standard should be testing against composites, rather than single components, to provide meaningful fire safety. But we do have to ask ourselves what we are trying to protect against when we design fire safe materials. One cannot provide fire protection against every fire hazard scenario out there. Where would we stop in fire protection for furniture? Crib 7? Flamethrower? Arson (can of gasoline)? Just because I can force a flame retarded material to burn by overwhelming it doesn’t mean that it provides no benefit at all, as the title of the paper suggests. The authors in the introduction touch on the flaws of the existing BS5852 test, and perhaps that should have been the focus of their work: showing that composite materials of foam and fabric are more meaningful for study than testing individual components. If the statistics in the paper pointed to the need to test against large fire sources, then I think testing with a No. 7 crib would have been meaningful – again if it had been done in a controlled room.
To conclude, the title is not supported by the data because the full-scale experiments are not reliable. Further, the title is misleading and is exactly the sort of title that undermines the credibility of scientists everywhere. It’s attention grabbing, but not supported by the data, thus confusing the public and making them distrustful of our work. The conclusions from the cone calorimeter data are sound, as is the concept of testing actual mattresses of known composition, but the work should be withdrawn, re-tested under controlled conditions, and presented again.
Speaking philosophically about the paper and what I believe was its desired goal, I have to ask the question: does it make sense to ask for both improved fire protection AND low emission toxicity? If the material never ignites against an ignition source found to be meaningful from a fire hazard perspective, then emissions are not a problem. But if we have a goal as a society to protect all people against all fire threats, then we need to look at all aspects of the problem, including the fire test and the cost/benefit analysis of making changes. If we achieve better fire safety, but most people cannot afford it, we’ve headed in the wrong direction. I personally would like to see inherently low heat release char forming materials used more often since this would concurrently address fire and emission toxicity issues, but I do not think the market will accept or bear the costs of aviation-grade materials replacing commodity goods in our homes. Maybe in the future the price of these materials will come down, but they’re not there yet. I further have no issue with more natural materials being used. My own analysis suggests that many natural materials can have lower inherent flammability than some synthetics. But synthetics have their place as well, and it’s not my place to tell society what they can and cannot use. Society (government, civilians, and scientists combined) must decide what level of fire risk it can live with, and once that is decided, then I as a fire safety scientist can run the right experiments to find out whether existing solutions work. This paper isn’t the right way to go about it. It addresses about half of the scientific problem, but then falls short due to the failure of the full-scale experiments. Let’s get the data in a controlled manner and see what it says, and then make conclusions about whether current flame retardant approaches are appropriate for future use.
Alexander B. Morgan, Ph.D. – Dayton, OH USA
Dr. Morgan is a fire safety scientist with 22 years of experience. He is a Group Leader at the University of Dayton Research Institute (UDRI) and the Editor-in-Chief of the Journal of Fire Sciences (JFS). This editorial represents his opinion and not those of UDRI or of JFS.
Fire Safety & Technology Bulletin is available by electronic subscription only at the rate of $300 for individual subscriptions or $1200 for library/corporate subscriptions. Enquiries regarding subscriptions and advertising should be addressed to Marcelo M. Hirschler or Stephen J. Grayson, using the above contact numbers. No part of this publication may be reproduced in its entirety without express permission from the editor.
US Editorial Office:
2 Friar’s Lane
Mill Valley CA 94941 USA
Tel: 1 415-388-8278
FAX: 1 415-388-5546
European Editorial Office:
West Yard House
10A Guildford Grove
London SE10 8JT UK
Tel: +44 208 692 5050
FAX: +44 208 692 5155