It is now well established that hydrogen chloride is unusual among common fire gases in that it decays from the atmosphere. No model of hydrogen chloride transport and decay exists as yet, which has been formulated in such a way that it is generic enough to be used for scenarios different from the one in which the experiments were carried out, and can be incorporated into more comprehensive fire hazard models. The present paper introduces such a model. It deals with the influence of various surfaces [poly(methyl methacrylate) (PMMA), painted gypsum board, ceiling tile, cement block and Marinite], of surface to volume ratio and of humidity, on atmospheric hydrogen chloride concentration, both inside and outside the room of origin of the fire. The present model incorporates generation of hydrogen chloride from poly(vinyl chloride), mass transfer to various wall locations, partition between the atmosphere and the surface, and a combination of diffusion and reaction inside the surface. The parameters in the model were fitted by using a non-linear (Marquardt) optimization procedure. The model was corroborated using various experiments which involved the combustion of poly(vinyl chloride) in large- and small-scale scenarios. It was found that, for a non-sorptive surface such as PMMA, the rate of mass transfer to the surface is much larger than the rates of the various reactions at the surface, in all cases. Such a surface, thus, allows much higher peak hydrogen chloride concentrations and much lower rates of decay than any of the sorptive surfaces investigated. For sorptive surfaces and static systems the rate limiting process is the mass transfer to the surface. The activity of the various surfaces investigated was found to follow the order: ceiling tile > cement > Marinite ≥ painted gypsum board ≫ PMMA The significance of this model is that it can predict hydrogen chloride decay in a real fire scenario. It is relevant to point out that normal construction surfaces are sorptive, and that hydrogen chloride decay will generally be quite fast in a fire, whereas it will be much slower in a small-scale toxicity test exposure chamber.

A series of full-scale cable tray fire tests have been done and the results have been expressed in terms of heat release (rate and amount), smoke release (rate and amount), mass loss and gas emissions, as well as the standard properties of flame spread and extent of charring. These tests were carried out in two different full-scale facilities. The same cables have also been tested in the cone rate of heat release calorimeter, (ASTM Method for Heat and Visible Smoke Release Rates for Materials and Products Using an Oxygen Consumption Calorimeter, E 1354), in a horizontal orientation, and the same prop-erties (except for gas emissions) are measured. Moreover, the combustible materials which make up many of the cables have also been tested in both the cone calorimeter and the Ohio State University (OSU) rate of heat release calorimeter (ASTM Method for Heat and Visible Smoke Release Rates for Materials and Products, E 906). The rate of heat release results of the cone calorimeter tests on cables were well correlated, linearly, with the results of the full-scale tests. This was particularly true when the cone was used at an incident flux of 20 kW/m2. A model has therefore been devised to predict full-scale cable tray results. The amount of smoke obscuration resulting from all full-scale cable tests was heavily dependent on the extent of burning of the cables. Those cables that did not burn extensively and released very little smoke. Similarly, those cables that did not burn extensively released low amounts of combustion gases, notably CO and HCl. As far as smoke release is concerned, total smoke released in the full-scale fires correlated very well with smoke factors measured in the small-scale cone calorimeter tests. The total smoke released in the small-scale tests, following complete sample combustion, was a much less reliable measure of full-scale smoke release than the smoke factor. The two small-scale rate of heat release instruments correlated well with each other, on all properties, except for time to sustained burning. Some fire properties of the cable jacket material alone, in the cone calorimeter at 20 kW/ m2, can be used to give a priori indication of likely cable full-scale fire performance in a certain scenario. The properties most appropriate for this purpose are the peak rate of heat release and the smoke factor. The OSU calorimeter was a somewhat less reliable small-scale predictor than the cone calorimeter, based on jacket material results only. Fire tests with the cone calorimeter can thus be used for preliminary fire hazard assessmentof electrical cables when installed in vertical cable trays.

In recent years electrical wire or cable insulation has been, once more, identified by NFPA statistics as a major material first ignited in residential fires (representing 7.6% of fires and 3.9% of fire fatalities in 1991–95) and the cause of 13% of catastrophic fires (1993–96). This highlights the need for renewed emphasis on fire testing of wires, cables and electrical materials. Cable fire tests can be subdivided into 5 categories: (a) Old fashioned small scale tests, which generally address only ignitability or flame spread, but the results of which are rarely meaningful in terms of real fire performance; (b) Vertical cable tray tests, of which there are a large variety, ranging in heat input from 20 kW up to 154 kW (in the case of riser cables), which address flame spread, and sometimes also smoke and heat release; (c) The Steiner tunnel NFPA 262 (UL 910) test, which determines wind aided horizontal flame spread, and smoke release, under a very high heat input (ca. 90 kW), with a relatively small mass load of cables; and (d) Small scale cable tests, often originally designed for materials, directed at measuring fundamental fire properties, such as heat release or critical fluxes for ignition or flame spread and thermal heating properties. (e) Tests for other cable fire properties, mainly smoke (obscuration, toxicity, corrosivity) and circuit integrity.

Toxicity of smoke is only one of many factors determining the hazard or the risk resulting if a product were involved in a fire in a specific scenario. Other factors include: amount of smoke (i.e., concentration of combustion products in the atmosphere), rate and quantity of heat release, mass loss rate, and flame spread rate, as well as such “environmental” factors as ignition source characteristics, fire detection and suppression devices, building occupancy, and code enforcement. A factor almost specific to the smoke generated from burning poly (vinyl chloride) (PVC) is the decay of hydrogen chloride (HCl) by reaction with building surfaces. The values of smoke toxic potency measured will also be affected by a number of parameters, including combustion mode, exposure mode, toxicological end point, and statistical analysis of results. A crucial factor, often overlooked, is the choice of an animal model appropriate as a surrogate for man, and its validation. Test animals are frequently chosen on the basis of convenience, cost, or other characteristics (e.g., sensitivity) rather than because of their similarity to man. This is particularly important in combustion toxicology, where one test species may not be a good model for all the major combustion products generated. Thus, comparisons of materials producing different major combustion products must be approached with caution to ensure that any apparent differences encountered in tests are not simply an artifact of test species. Over recent years, increasing evidence has surfaced that some rodent species are poor models for the toxic response of man to irritant gases or to smoke-containing irritants. Studies on HCl (as a pure gas) and on the smoke generated from the burning of PVC have indicated that mice are much more sensitive than rats. More importantly, they are much more sensitive than primates. It has also been established that rats are a good model for primates in terms of the lethal effects of irritant products. Although primates have survived 15-min exposures to 10 000 ppm of HCl, 2500 ppm is lethal to mice. Moreover, under the same exposure conditions, mice will die at PVC smoke levels four to seven times lower and HCl levels seven to ten times lower than those at which rats will. In contrast, lethal doses of asphyxiants such as carbon monoxide (CO) are similar in rats and mice. These results indicate that the response of the mouse significantly overestimates the toxic potency of HCl and of PVC smoke to man.

The corrosivity of combustion products has arisen as an issue for both product manufacturers and standards bodies. While many industries can have concerns in this area, electronic communications and control equipment are especially vulnerable to the problem. The best way to manage the smoke corrosivity issue is to avoid a fire. In the event a fire does occur, a number of actions must be taken. An important consideration is that the materials and products used should have been chosen with due consideration of their corrosion-causing potential. To do this requires a suitable corrosivity test. It has traditionally been assumed that smoke corrosivity is directly correlated to the emission of acid gases. The results of recent experiments have shown that materials which do not release acid gases can, nevertheless, cause corrosion of metal surfaces, as determined by metal loss. In addition, for electrical equipment there are two other types of related nonthermal damage from combustion products which must be considered, viz., ohmic bridging and degradation of contacts. Laboratory tests proposed to date to measure the corrosive effects of combustion products all have significant deficiencies: some methods are not performance-based at all and are merely tests for pH. In others, unrealistic specimen heating or unrealistic exposure targets are used. To facilitate the development of a better test, a series of criteria have been developed. A specific test method was evolved from these criteria. The method is performance-based and incorporates realistic fire heating conditions.