Carbon Monoxide Generation in Partially Suppressed Fires
By: Nils Johansson, Lund University, Sweden
Within the quarter of a century, numerous pieces of literature have been published noting the significant increase in carbon monoxide concentrations, the most fatal toxic by-product, when fires are partially suppressed through a water spray. In a master thesis presented at Lund University Haydn Lewis has studied the effect on carbon monoxide production in partly suppressed fires in small-scale experiments.
Records of fire fatalities within the built environment show that the “dominant toxicant in fire deaths” is CO accounting for approximately two thirds of fire deaths. Within the field of fire safety engineering, the ventilation conditions have traditionally been used to determine the yields of toxic species with little, or no, consideration of the influence of suppression on fire chemistry and the resulting environmental conditions. Fire suppression systems, including both traditional sprinklers and mist systems, are increasingly prevalent as more complex buildings and infrastructure are constructed.
The suppression of fire by water sprays is a complex physio-chemical process, involving several competing mechanisms, all of which have not yet been fully understood. For the purposes of the work presented here, extinguishment is defined as the ceasing of all combustion, suppression as the act of controlling combustion to a reduced the HRR and partial suppression as an inefficient application of suppression in which the fire continues to steadily burn (a shielded or high challenge event).
The sizes, and relative discharge speeds, of the droplets within water sprays are key variables involved in weighting the potential influence of each mechanism to interact with the combustion process. A water spray consists of many millions of droplets having a distribution of different diameters (so-called polydisperse sprays). The diameters of droplets produced by mist suppression nozzles are an order of magnitude smaller than those produced via a sprinkler nozzle, and this enables water mist systems to interact directly with the flame reaction zone. The small droplets enable the water spray to remove large amounts of energy directly from the flame. This reduction in temperature results in a reduction in the effectiveness of the combustion. Further, the inert water vapor generated through evaporation of droplets within the reaction zone, displaces part of the oxygen required to oxidize the fuel.
The aim of the conducted thesis was to contribute to the knowledge of fires subject to suppression by water. Specifically, the intent is to analyse the interaction of fine water droplets on the gas phase chemistry of fire, the interruption of the combustion chemical process and the resulting generation of CO.
Previous work
A literature review was undertaken to study the previous studies in the area and to get inspiration for the development of an experimental study. Many previous studies present an influence of characteristic water droplet size on CO concentration, with smaller droplets also being shown to result in higher increases in CO concentrations. In most of the reviewed studies a single characteristic diameter of droplet size, with no information regarding the distribution of droplet sizes present within each spray, was used. This limits the appropriateness of drawing comparisons between the data sets. Consequently, it was concluded that more work was needed to investigate how the shape of droplet distributions of these polydisperse sprays influence the results obtained.
Experimental setup
The experimental set-up was designed to focus on the effect of water droplet interaction on the gas phase chemistry of the fire. The setup was positioned under an oxygen consumption calorimetry extraction hood to capture and analyze the products of combustion. In addition, a water collection tray was utilized to quantify and compare the volume of water that did not vaporize within the flame (see Figure 1).
Figure 1: Experimental setup.
A key point of difference between this study and analogous studies studied was the perpendicular orientation of the nozzle compared to the flame (see Figure 1). This was done to focus the interaction of the droplets on the gas phase chemistry of the fire and prevent flare-up because of the vaporisation of water when interacting with the surface of the fuel.
Twenty-four unique tests were undertaken within the study, which varied fuel type, nozzle type and water pressure. Heptane and propane were utilized as fuels within this study based on their well-known degree of incomplete combustion within well-ventilated free-burn conditions, and to ensure that any changes to combustion during suppression were promoted. The heat release rates from the fires ranged from 40 to 70 kW.
To decouple the effect of droplet size distribution and water flow rate effects on the CO production, two different nozzles with different spray properties were utilized in the study. The mist suppression sprays had droplets of volume median diameters of between 163 μm and 287 μm and water flow rates of 1.5 L/min and 3.5 L/min. For further information about the setup and the experimental tests conducted the reader is referred to the thesis report [1].
Results
The change in CO concentration due to partial suppression being initiated following a period of free burning, was measured. The results were average over the three repetitions undertaken for each set of variables. The results show a CO concentration increase of up to 250% when sprays containing the distribution of smallest droplet diameters and highest flow rates are applied to the largest fires considered within this study. Whereas the smallest recorded change in CO concentration was 14% for sprays containing the distribution of largest droplet diameters and lowest flow rates applied again to the largest fires considered within the study.
Figure 2 provides an example of the variation in measured CO levels recorded for each of the three test repetitions for both a heptane and propane fire. The level of fluctuation in CO concentration, around the mean, is comparatively small during the baseline period compared to when the mist suppression system is active, when additional turbulence is introduced to the combustion process.
Figure 2: Example variation in CO across the three test repetitions. Left (T1): 80 kW Heptane; Right (T13): 80 kW Propane.
Regarding reduction of heat release rate the results show that the measured heat release rate of the heptane fires experiences a slight decrease upon activation of the mist suppression, and then remain relatively stable. On the other hand, the heat release rates of the propane fires were much less influenced by the water mist, as the cooling effect of the droplets do not impact on the rate of release of combustible gases.
Implications for fire safety engineering
Whilst these experiments consist of several simplifications, it is considered that this study establishes a proof of concept that typical product yields that are frequently adopted fire safety engineering designs, may not be appropriate when fires are subject to mist suppression that does not provide a significant reduction to heat release rate.
The results have shown that upon activation of the mist suppression, the volume fraction of CO within the fire environment significantly increases, by up to 250%. Fire safety engineering designs developed based on predicted toxicity dose taken up by egressing occupants, in which mist suppression systems are installed, are potentially significantly underestimating the levels of toxic exposure. However, it should be acknowledged that properly designed mist suppression systems are likely to significantly reduce the fire heat release rate, and in, turn limit the volume fraction of CO within the fire environment.
The scope of this work is not broad enough to draw conclusions regarding an appropriate adjustment factor to apply to typical species product yields when partial suppression scenarios could be expected. However, the results indicate a need for closer examination of how water sprays influence toxic species production. Typical fire scenarios in which partial suppression could be anticipated include:
- Shielded vehicular fire within a tunnel featuring a fixed firefighting system
- Fires within a rack storage structure with ceiling mounted sprinkler heads.
The master thesis work demonstrates that the commonly applied free burning CO yields have the potential to be significantly lower than the yields present in scenarios in which fires are partially suppressed. Therefore, until more expansive studies are undertaken, and knowledge of this phenomenon is developed, it is recommended that fire safety engineers consider whether it is appropriate to adopt a more conservative species yield in relevant scenarios.
References
- Lewis, H., Factors influencing the generation of carbon monoxide in fires partially suppressed through water mist application, IMFSE master thesis, Lund University, Sweden, 2020, http://lup.lub.lu.se/student-papers/record/9027810.