1999

Keski-Rahkonen, Olavi; Mangs, Johan; Turtola, Antti

This paper reviews electrical ignition phenomena from a wide perspective through statistical, modelling and experimental tools. A rather comprehensive concept of electrical ignition phenomena has been described. Several databases indicate that defective cables leading to short circuit and ground shorts, as well as loose connections leading to overheating, are the most common reasons for electrical ignitions. For modelling an overheated cable a mathematical model has been proposed, which compares favourably with a limited set of experimental data. Experiments on PVC cables showed self-heating a possible but improbable cause of initial ignition. The literature review of physical models of electrical arcs established conditions where ignition of cables might be possible. A limited set of tests under poorly controlled conditions succeeded, not producing long lasting arcs amenable to sustained ignition. The reason for experimental failure is believed to be too violent release of energy, which blew off the flames. Existing semiquantitative models of flame spread are shown to be able to describe salient features of cable ignitions despite clear deviations of the initial assumptions of the model. Laboratory tests of electronic components heavily or destructively overloaded did not generally lead to ignition of adjacent material because of sudden release and subsequent destruction of the component. Only power transistors heavily mounted on printed cards seemed able to start ignition of the card. The phenomenon can be modelled as a piloted ignition similar to flame spread on cables.