Examination of the Use of Fire Dynamics Analysis Techniques With Furniture Fueled Fires

Overview
Findings

Overview

This study on fire dynamics consists of three phases:

HRR experiments
Compartment experiments
Assessment of three types of models used for fire dynamics analysis

This research focuses on evaluating the ability of three types of models commonly used in fire investigations to predict characteristics of the fire environment generated from gas burners and a single upholstered furniture items. Results from this research will provide guidance and insight into the uncertainty bounds that should be considered when using HRR and fire environment data as input for a fire dynamics analysis that is part of a fire investigation. In addition to providing HRR and fire environment data sets, the modeling results show how uncertainties in the data propagate through the models.

Context

Fire investigations provide a means to identify the cause of a fire as well as collect data that may provide insight about the development and spread of a fire. By determining the cause of a fire and identifying products and phenomena that contributed to fire spread, investigators may be able to prove guilt or innocence in criminal proceedings, assign blame in civil proceedings, or contribute to the knowledge base that informs the fire protection community for future designs. Data such as the thermal conditions in area of fire origin, the time to flashover of a compartment, and the influence of ventilation on the fire are critical to understanding the cause of the fire.

Computational models are increasingly relied upon in fire investigations, as part of the scientific method, to analyze data or to test hypotheses. The models can be used to qualitatively gain insight on fire phenomena or fire-induced fluid flows. The models can also be used for quantitative analysis if an appropriate range of uncertainty is included. Model results, much like measurements, have varying degrees of uncertainty that can be affected as much by limitations of the model as well as unknowns in the model input parameters.

Current computational models range in complexity from algebraic-based specialized fire dynamics routines derived from fundamental physical concepts and empirical data to generalized, physics-based computational fluid dynamics codes that require a wide range of property values as inputs. The latter may also require specialized knowledge and significant computational resources.

The understanding of the limitations, accuracy, and inherent uncertainties in each model is primarily based on fire measurements generated with well-characterized and, in many cases, steady-state heat sources, such as natural gas-fueled burners or liquid hydrocarbon pool fires. However encountered in fire investigations are often fueled by natural and synthetic solid materials. These fuels are three-dimensional (as opposed to a two-dimensional burner surface), and the foam plastics used in furnishings tend to drip and flow during burning. Fires with these fuels are characterized by non-steady burning where rapid transitions in energy and fuel output are possible.

The Fire and Arson Investigation Technology Working Group Operational Requirements, published in December 2016, addressed several issues regarding input data and fire model validation: (1) repeatability and reproducibility of test measurements of large-scale structure fires, (2) materials property data inputs for accurate computer models, and (3) evaluation of incident heat flux profiles to walls and neighboring items in support of fire model validation.

Objectives

Improve the understanding of the repeatability of key test measures associated with fire dynamics analysis such as heat release rate (HRR), heat flux, gas velocity, and temperature.
Add fire test data on several samples of commercially available furnishings for use with fire models.
Add heat flux profiles to walls of a compartment close to and remote from the fire origin.
Assess the accuracy of a range of predictive fire algorithms and models when provided with HRR data from residential furniture and compared to replicate fire test results.
Provide guidance toward standard best practices for the use of fire dynamics analysis techniques with furniture fueled, compartment fires based on research results.

Research Partners

Examination of Fire Dynamics Analysis Techniques: Heat Release Rate Experiments

The results from the interlaboratory HRR experiments demonstrated that the repeatability and reproducibility of HRR measurements have improved considerably in the past twenty years. Even with uncertainties introduced by a different flaming ignition source, the expanded uncertainty intervals (95 % confidence level) for peak HRRs across the three calorimeters for all of the chair tests were 19% to 22%. The range of expanded uncertainties for the total heat release was 6% to 16%. This is a major improvement from a similar study conducted 20 years ago that showed the mean standard deviation of the peak HRR was 40 %.

This provides a guideline for the uncertainty bounds to use when the HRR for upholstered furniture is needed for use in a fire dynamics analysis. This study also provided HRR data which will be included in the NIST Fire Calorimetry Database available at https://www.nist.gov/el/fcd.

Examination of Fire Dynamics Analysis Techniques: Repeatability of Compartment Fire Experiments

A total of 117 fire experiments were conducted inside a compartment with a single door that was either opened or closed for the entirety of each experiment. Natural gas burners set to different HRRs and two upholstered furniture items were used as fuel sources and were placed at four different locations within the compartment over the course of the experimental series. Multiple iterations of the different experimental configurations were conducted to evaluate the repeatability of the thermal measurements.

Comparisons of measurements between replicate experiments revealed that fire conditions generated by gas burner fuels were generally more repeatable than those generated by furniture fuels. For both the open- and closed-door configurations, temperature and flame height were typically the most repeatable measurements, while the wall heat flux measure was the least repeatable. For the gas burner experiments, the mean values of experimental data collected during the period of steady thermal conditions (final three minutes of the experiment) were used for the repeatability assessment of the experiments with an open door. Table 1 displays the ± 2σ values computed over the normalized steady mean values from the various measurement groups for open door replicate experiments with each gas burner fuel source. The total number of normalized steady mean values, N, over which each ± 2σ value was computed is also included in the table.

Examination of Fire Dynamics Analysis Techniques: Assessment of Predictive Fire Algorithms and Models

The accuracy of the predictions from all the models was found to be better in the gas burner experiments than in the furniture experiments. All the models were sensitive to the definition of the heat release rate and the geometry of the burning item. Figure 2. Provides examples of results, in this case, the layer temperature predictions for the gas burner and the furniture fire experiments.

View the data and video from all three experiment series, visit our Fire Investigation Data Portal.

For questions about this project, please contact:

Published: September 9, 2021