Three New Journal Articles Investigate Methods for Obtaining Properties for Fire Growth Models

Three new peer-reviewed research papers were recently published in ASTM International’s STP1642 on Obtaining Data for Fire Growth Models. The papers were presented at a virtual symposium sponsored by ASTM International Committee E05 on Fire Standards and Subcommittee E05.32 on Research on December 14-15, 2021 by research engineers Mark McKinnon and Matt DiDomizio, from the Fire Safety Research Institute (FSRI), part of UL Research Institutes.

In Impact of Specimen Preparation Method on Thermal Analysis Testing and Derived Parameters, DiDomizio and McKinnon investigated the impact of specimen preparation methods on the results of milligram-scale tests for measurement of properties used in fire models. Tests included thermogravimetric analysis, differential scanning calorimetry, and microscale combustion calorimetry. Samples for these tests are commonly prepared by cryogenic milling of bulk materials. The authors systematically varied the duration of milling applied to polyvinyl chloride and polymethyl methacrylate samples and found that property data derived from the milligram-scale tests could be significantly affected by the milling duration. The impact of this finding was extrapolated to a fire model using these properties as input, and it was demonstrated that burning rate predictions were sufficiently sensitive to these changes to raise concerns on the suitability of some property data in fire modeling applications. This work highlights the need for appropriate standardized testing in the fire science field.

DiDomizio, McKinnon, and McCoy applied a novel approach for quantifying the uncertainty in fire model predictions attributed to uncertainty in model input parameters in Propagation of Uncertainty in a Thermal Analysis Model Using Polynomial Chaos Expansion. While a direct approach for the propagation of uncertainty in fire model predictions has not traditionally been possible due to the number of variables involved in complex fire models, the authors demonstrated the use of a computationally inexpensive algorithm (referred to as polynomial chaos expansion) to propagate the uncertainties of property data inputs in a pyrolysis model. This work demonstrates the importance of error propagation in fire model predictions, and provides a benchmark case for the application of generalized polynomial chaos theory in this field.

Measurement of Flame Heat Flux Using the Cone Calorimeter, a collaborative work between the University of Waterloo and FSRI, explores methods for measuring flame heat flux using a cone calorimeter. The authors carried out a study encompassing solid (polymethyl methacrylate) and liquid (methanol) gram-scale samples exposed to varying levels of irradiance using a cone calorimeter. A methodology was presented for estimation of flame heat flux from simultaneous measurements of the rates of heat release and mass loss of the samples; heat feedback fractions of 0.065 and 0.076 were determined. A second methodology for estimating flame heat feedback fraction from measurements of heat flux gauges embedded within the samples was also explored. It was found that this latter approach could not produce accurate measurements of the flame heat feedback, thereby providing guidance to the fire science community in selecting an appropriate method for measurement of this model input parameter.

ASTM International’s E1591 Guide for Obtaining Data for Fire Growth Models serves as an important reference for fire safety engineers on the measurement and use of data in fire models. FSRI’s work will help in developing a future roadmap for the positioning of this guide, and provide a valuable reference to experimentalists model practitioners in the fire science field.