Fire Safety and Technology Editorial

 

Unrealistic expectations for fire testing

Fire tests can be used for a whole variety of purposes, but, unfortunately, they can also be misused. I have discussed in the past cases where companies purposeful misuse a fire test to get "passing" results. What I want to talk about today is the use of fire tests with unrealistic expectations, and the inevitable ensuing disappointments. I will address in this editorial predictions from the cone calorimeter to large scale fire tests. In future editorials I plan to discuss other types of potentially unrealistic expectations people may have when considering conducting fire tests.

One of the most widely used and most useful small scale fire tests is the cone calorimeter, which has been standardized as ASTM E 1354 and ISO 5660, and into a number of application standards. The cone calorimeter assesses ignitability, heat release, smoke release, mass loss and combustion product release by burning a small (100 mm x 100 mm, by up to 50 mm thick) test specimen, which can be a material or a representation of a "mostly planar" product. The results, in engineering units, are of properties that are of key importance in fire safety. People run this test because it is relatively easy to conduct, needs small amounts of material and is consequently fairly inexpensive.

In some cases the cone calorimeter is actually the regulatory tool. In such cases, conducting cone tests is the perfect solution and the cone test report will contain the answer needed. That is not, however, the only case where cone tests are conducted.

Studies and publications have shown that heat release from the cone calorimeter is an excellent predictor of heat release in full scale tests, for several types of materials/products. The key studies have involved wall/ceiling linings (compared to a room-corner test, such as NFPA 286, NFPA 265 or ISO 9705, and even compared to an intermediate scale test such as the SBI), upholstered furniture and mattresses (compared to a full scale test of one product, such as CA TB 133, CA TB 129, their ASTM equivalents [ASTM E 1537, ASTM E 1590], 16 CFR 1633, the US Navy's mattress purchase specification NAVSEA 05L PD 4-02 or the CBUF tests) and electric or fiber optic cables (compared to the various vertical cable tray tests, such as UL 1581, UL 1685, CSA FT4, IEC 60332-3 or FIPEC). It has also been shown that cone calorimeter smoke results can be used to predict (with a lower degree of confidence) smoke release in the same types of tests. This means that cone testing can ascertain with a high degree of confidence whether these types of materials or products are fairly safe or not. This is a key fact and will be particularly important when assessing whether to bring a new material to market or whether to sue a material for a new (or even an existing) application. It will also help when considering product liability issues.

However, even with these types of materials or products, the predictability of the cone calorimeter is usually not a simple comparison of two numbers. Typically models have been developed that use more than one data point, or even data from tests under different conditions, to predict performance in the larger scale test. Therefore, it is unrealistic to expect that a number in test results from one (or one set) of cone tests will be able to tell us whether a material will perform well or not.

Another issue to consider is the fact that predictions are best when the fire performance of materials is most distinct. For example it is very easy to predict that a material with a cone peak heat release rate of 1,500 kW/m2 will perform better in a large scale test than a material with a cone peak heat release rate of 50 kW/m2. Furthermore, a material with a very low cone peak heat release rate is likely to do well on most criteria. However, this is a comparison that manufacturers are usually looking for: people rarely care about extremes in fire performance. More typically, a manufacturer wants to tweak a composition to get a compound with a fire performance that's good enough while minimizing cost. Now we are talking about comparing materials perhaps with cone peak heat release rates of 300 kW/m2 and of 400 kW/m2; things are much tougher because we are likely to be close to a threshold and the degree of predictability is much poorer.

Things can get even more complicated. There are many instances where the tests mentioned above are not the tests that people want to predict, because what they want is to pass a mandatory test requirement. In the US it is very common for people to want to predict Steiner tunnel test results. The tests are ASTM E 84 (for flat sheets or materials that can be made into flat, or "mostly planar" test specimens), NFPA 262 (for plenum cables), UL 1887 (for sprinkler pipes), UL 1820 (for pneumatic tubing) and UL 2024 (for raceways). Again, the cone still does an excellent job of predicting that a material or product with very low heat release is likely to end up with low flame spread (or flame spread index) while one with very high heat release is likely to perform poorly.

When considering smoke production in the cone and the Steiner tunnel, everything is even more complicated, primarily because: (a) the repeatability/reproducibility of the Steiner tunnel is mediocre at best and the precision is particularly poor with regard to smoke, (b) the cone is a small scale test and smoke is a property which is based on a combination of intrinsic smoke per unit area or mass (which small scale tests assess) and the amount of material burnt (which small scale tests don't assess) and (c) the predictability of the cone with regard to smoke is less satisfactory than with regard to heat release or similar properties. Therefore, it is likely that when a comparison between borderline candidate materials is made based on cone results, the information may be insufficient to make fail-safe decisions.

The cone calorimeter is an excellent tool for understanding fire performance of materials or products, as long as no unrealistic expectations are placed on the results. Comments on my opinion are very welcome.

Marcelo M. Hirschler