A Simplified Method for Evaluation of Insulation Fire Hazard Potential

Fires caused by runaway oxidation of combustible fluids on high surface area substrates such as insulation, absorbents and carbon beds are relatively common throughout the chemical industry. These types of process safety events have caused serious injuries and property damage. The basis for adequate and cost-effective fire protection strategies and solutions depends on accurate determination of the potential risk. For example, choosing how to insulate a process line or vessel, either with an inexpensive, porous insulation (such as mineral wool) or with an expensive, non-combustible, non-porous insulation (such as cellular glass) will depend on the risk of a leak and fire with the process fluid at that process temperature. Twenty-five years ago, Britton's "Z" parameter method was proposed and validated for predicting spontaneous fires in insulation from any given fluid's autoignition temperature (AIT) and flash point (FP). At Dow, this approach is still utilized along with an internally-developed, alternative, heat-balance method to screen for potential insulation fires. This method, which assigns a potential risk based on heat gains from oxidation vs. heat losses from several cooling mechanisms, is based on knowledge of any given chemical's molecular structure and normal boiling point, or vapor pressure. To validate this approach, a variety of studies were performed using model systems of insulation coated with various liquids at several weight percent levels. The coated samples were placed in a convection oven and heated while thermocouples placed on the edge and at the center of the insulation samples were monitored. In addition to the convection oven studies, thermokinetic data from open-cup adiabatic calorimetry (Accelerating Rate Calorimeter (ARC)) and isothermal calorimetry (SETARAM C80) were collected and modeled to determine if these data could accurately predict the detected onset temperatures. It was found that using the fits of either open-cup ARC or C80 microcalorimeter differential scanning calorimetry data and the Frank-Kamenetskii critical parameters, the detected onset temperature of oxidation measured in a convection oven for insulation samples coated with different organic liquids could be predicted within 1.4-3.9% (depending on the system). The breadth of this approach is illustrated for a number of fluids which have caused insulation fire incidents in the past.