In an earlier study a full-scale furniture calorimeter was used to determine the heat release rates for upholstered chairs containing various construction materials. Samples of these same material combinations have now been tested in a bench-scale apparatus, the cone calorimeter. A correlation was established between bench-scale and full-scale data. Thus, it appears that prediction of flashover potential of a single upholstered item may be possible by using bench-scale results.

National standards for flammability have, in most countries, been established for many years. Thus, they are based on an understanding of fire physics which may be several decades old. In more recent times, however, significant strides have been made in developing test methods which are based on an improved understanding of building fires. The best-possible estimate of the flammability of building products or contents would be, by definition, a full-scale fire experiment. These are recently being standardized, and will serve an important reference function. Because of cost and practical difficulties, however, it is generally desirable to do the majority of product evaluations by benchscale tests. In this paper the rational bases for achieving validated bench-scale tests for flammability are examined, and a number of recent examples are cited where such a process has been followed. It is also shown that these test results have a correct relationship to the fullscale phenomena. BACKGROUND When trying to describe the flammability, or "reaction-to-fire" of combustible products, the most relevant and fundamental understanding must be obtained from full-scale fires, real or experimental. Experimental fires have, of course, been conducted throughout history. It is striking to note, then, that the actual standardization of methods for conducting full-scale room fire tests is so recent a development. The draft method from the International Organization for Standardization eISO) dates only to 1986 [1], developed on the basis of a proposal from the American Society for Testing and Materials of 1982 [2] and a comparable one from NORDTEST of 1986 [3]. National flammability standards-based on various small-scale testing devices-meanwhile, go back several decades. The existing ones [~. g., 4] have generally been developed on an ad hoc basis, to exclude some specific materials known to be poor performers, and not on a detailed understanding of fire behavior. More than ten years ago, Prof. Emmons obtained the results of flammability tests on a number of materials, when tested according to various national flammability standards [5]. Not surprisingly, he found that the relationship between the test results according to the different standards was almost completely random. Ostman and Nussbaum [6] have very FIRE SAFETY SCIENCE-PROCEEDINGS OF THE SECOND INTERNATIONAL SYMPOSIUM. pp. 813-822

The life safety hazard issues associated with flaming fires of mattress and up- holstered furniture are explored. It is shown that full-scale heat release rate (HRR) is the dominant variable which needs to be controlled. This can be determined directly by full-scale measurement. In many cases, full-scale tests are not convenient to conduct. Thus, it is desirable that bench-scale procedures be available which can be used to predict some of the important features of the full-scale test. Such procedures have been developed at NIST for upholstered furniture during several prior studies. In the present work, differences between the behavior of mattresses and of upholstered furniture are explored. Mattresses and upholstered chairs are soft goods which are constructed in a somewhat similar way: both use padding foams or battings, covered by upholstery fabric. There are differences in construction, however. Mattresses are flat, whereas upholstered chairs normally have seats, backs, and side arms. Also, an upholstered chair is constructed normally on a wood frame, whereas a mattress has no structural components, or else has steel innersprings. The quantitative knowledge of mattress behavior is still not as advanced as that for upholstered furniture. Nonetheless, based on a recent set of tests, the behavior of mattress fires can be quantified initially. Especially, data are now available to predict whether or not a particular mattress construction will lead to a propagating fire. Similarly as for upholstered furniture, such a limit value can be used to determine whether certain regulatory pass/fail criteria are met. The relationship obtained is incomplete, however, because the known roles of ignition source power level (that is, kilowatts output) and geometrical configuration are not yet quantified. Also, there is not yet a detailed explanation for differences between the observed relationships for mattresses and for uphol- stered chairs. Thus, future work will need to be done to address and further quantify these effects.

The preliminary work describing the Cone Calorimeter dates to 1982 when NIST (then NBS) presented the results of their research to ASTM Committee E5. In the next few years, the hardware was improved and finalized. ASTM first adopted a draft standard (P 190) in 1985, followed by full adoption as ASTM E 1354 in 1990. By now, more than 100 laboratories have adopted Cone Calorimeters, and several hundred papers have been written by researchers. The present paper reviews some of the highlights of the Cone Calorimeter development and its R&D utilization. It then discusses regulatory and specification usage of the apparatus and finally describes some potential applications yet to be realized.