Fire: characteristics, origins, and environmental effects

¶ … Hadjisophocleous, G. & Kia, Q. (2009). "Comparison of FDS Prediction of Smoke Movement in a 10-Storey Building with Experimental Data." Fire technology 45(2)

This study used real data from an experimental fire in a specially designed ten-storey tower to determine the accuracy of the Fire Dynamics Simulator, a computer modeling system that had been developed by the National Institute of Standards and Technology in Canada. The primary consideration of the study, and the facet of the fire that was most closely observed, was the movement of smoke through the building; temperature as well as certain gas levels (primarily oxygen gas and carbon dioxide) were measured on each floor to determine the spread of the smoke. The experiment showed that the Fire Dynamics Simulator was reasonably accurate in its predictions, and could be made more accurate through the inclusion of some simple adjustments for unaccounted-for features.

The research problem that Hadjisophocleous and Kia were addressing in this study was determining the reliability and validity of a computer simulation of a fire when dealing with the issue of a real (though controlled) fire event. That is, the researchers were not primarily concerned with discovering the movement patterns and rapidity of spread of some during a fire such as the one created, but rather with how well the NIST-developed Fire Dynamics Simulator predicted smoke movement, as well as other elements of the fire. Addressing this research question involved a semi-indirect approach, where the results of a real fire had to be compared to the predictions of the computer model in order to secure reliable and meaningful data confirming -- or finding fault in -- the model's predictions.

In order to conduct the research, a ten-storey tower was designed and erected. This tower had a compartment on each floor to serve as a "room" and a corridor, to approximate the spaces that exist in a typical building. A complete staircase running all ten floors was also included, as was an elevator shaft. Through the design of the building itself, the researchers were attempting to come as close as possible to real-world conditions, for which the Fire Dynamics Simulator was designed. Sensors for heat and gas levels were placed on every floor of the tower, relaying their information to external computers. A fire was then started in the second-floor compartment of the building and allowed to burn while the instruments on each floor collected data regarding the movement of heat and gases -- specifically tracing the movement of smoke -- as the fire burned out.

There were definitely some flaws in the research design, some of which could have been foreseen and prevented, and others of which would have presented practically insurmountable difficulties to the researchers. In the former category, more accurate data could have been achieved through the use of more widespread data collection devices, and their employment in a variety of areas (within the corridors as well as the compartments; on ceilings as well as floors and possibly walls, in the staircase and elevator shaft, etc.). This would lead to a more comprehensive understanding of how smoke, heat, and gases moved throughout the building, and would have allowed for more specific and finely-tuned adjustments to be suggested for the Fire Dynamics Simulator than the current study with its more limited data collection schema provided to the researchers.

The design and construction of the tower itself is somewhat problematic as well, though the issues here could not readily be addressed while keeping this experiment possible from both financial and other practical concerns. The materials used in the construction of the tower and the layout of the completed tower both caused significant differences from a real world fire, and though the Fire Dynamics Simulator still managed to make relatively accurate predictions of the spread of temperatures, gases, and smoke when compared to the experimental data, these did not bear much direct relation to real-world conditions. The study, therefore, did not test the reliability or validity of the Fire Dynamics Simulator's modeling capabilities in as significant a way as might have been liked. As the conducting of such a controlled experiment on a real-world scale would be highly impractical if not impossible, however, these flaws must be considered unavoidable.

The analysis of the data showed that the Fire Dynamics Simulator very closely predicted the patterns of movement of both smoke and gases and the rising temperatures in various areas, though it consistently predicted greater rises in both temperatures and gas levels, as well as earlier occurrence. That is, the rises, peaks, and falls of the experimental data match those of the Fire Dynamic Simulator's modeled predictions almost exactly in most cases, but the rises occur later in time and the peaks are generally lower than those predicted. This means that, although the model definitely has strong predictive powers that closely mirror the timeline and severity of a fire and its spread of temperature, smoke, and gases, it tends to predict more heat and greater levels of toxicity than actually exist.