Ambient Air Temperatures
While ambient temperatures were measured from floor to ceilings (Table 1), most firefighter activities on a fire ground take place between the 0.3 to 0.9 m heights as firefighters crawl, crouch, kneel and then move to 1.5 m as firefighters eventually stand during overhaul. However, some fire departments’ tactics dictate walking in the structure during the firefight to allow more efficient movements, so conditions at the higher levels may be experience by some firefighters throughout the firefight. At the 0.9 m level, average temperatures in the left and right chambers were 37.0 ± 5.5 °C and 37.1 ± 5.1 °C respectively with peak values of 54.9 ± 6.8 °C and 60.9 ± 7.6 °C. These values are well aligned with data collected from the simulated Fireground study, where search firefighters in Interior Attack trials average helmet temperatures were 39.7 ± 4.6°C with peak values of 63.2 ± 13.0°C . At the 1.5 m level average temperatures were 62.4 ± 8.5°C and 63.7 ± 8.7°C, which compares well with the simulated Fireground study’s Attack firefighters where average helmet temperatures of 57.6 ± 7.0°C were recorded when conducting Interior attack .
Figure 2 depicts the variability in the temperatures over the 40 mannequin test trials. This variability in test conditions can be both a benefit and a limitation of the technique. Due to the ability to test multiple mannequins (with 24 independent sets of PPE) in a given trial, the tool provides the ability to compare many different iterations at one time with excellent repeatability within that group of samples. However, there is some day to day variability in FES ambient conditions due to wind and environmental conditions as a result of the prop’s location outdoors. A portion of this variability may be reduced by utilizing the prop inside a large environmental enclosure or test laboratory such as that present at UL, NIST or other large fire laboratories around the world. However, it is important to note that any live-fire scenario can have important variability in conditions even when using identical fuel packages [26, 27]. Thus, conducting serial trials using the FES prop can expose PPE samples a range of conditions that may be present during live fire events.
PAHs and VOCs in the Air
Of the 40 mannequin trials, measurements of airborne PAHs (n = 12) and VOCs (n = 8) were collected during 4 trials. PAH and VOC data from a subset of firefighters completing the FAS protocol were collected during 12 separate trials with 6 participants in each. As expected, both total PAHs and benzene concentrations in the environment were higher on the standing mannequins. Time-averaged median total PAH concentration in the environment (Table 2) was 45,800 µg/m3 for the human subject trials where firefighters were crawling and staying lower in the structure, while median values were 196,000 µg/m3 for the standing mannequins. A recent study by Keir et al. (2020) collected personal air samples from firefighters during emergency fire suppression and found much lower mean total PAH concentrations (2725 µg/m3) . However, the average sample collection time from Keir et al.  was 67 min (max = 420 min), nearly 6 times longer than the current study. Overall, the sampling time for this study ranged from 4 to 11 min. In general, sample times were lower for the mannequin portion as pump faults were more prevalent, limiting how long the pumps ran. Data from Keir et al. were likely influenced by the sampling pumps running on the way to the firefight and possibly after suppression was completed, including time outside of the burning structure that are common in response scenarios . The FES prop timing was designed to focus on simulating the short exposure time typical with fire suppression timeframes. From the simulated Fireground study , median personal concentrations of total PAHs for the attack firefighters were 23,800 µg/m3 (7460–78,200 µg/m3) and 17,800 µg/m3 (9770–43,800 µg/m3) for the search firefighters. Thus, firefighters conducting the FAS protocol in the FES prop resulted in PAH concentrations that are within the upper ranges measured in the simulated Fireground study. At the standing level, PAH concentrations are nearly an order of magnitude higher than the measurements from the simulated Fireground study. Consequently, contamination on the upper portions of the jacket for stationary mannequins will be much higher than what might be experienced for firefighters operating in a typical manner. However, this may also provide a high-concentration challenge to identify possible leakage paths in the PPE, which is similar to some traditional laboratory simulant tests.
VOC concentrations in the FES exposure chamber were dominated by benzene, which is similar to the results in the simulated Fireground study . The median air concentration of benzene in the environment (Table 2) was 72,200 ppb for the trials where firefighters were crawling and staying lower in the structure during these simulated firefighting activities, while the median concentration was 234,000 ppb for the stationary, standing mannequins. These median values are within the ranges measured from attack firefighters (12,400–322,000 ppb) and search firefighters (12,000–306,200 ppb) in the simulated Fireground study , but higher than benzene concentrations reported by Austin et al. (11,000 ppb)  and Jankovic et al. (22,000 ppb) .
Some PAH measurements from the trials with human subjects were stratified by specific compound (Table 3). Four PAH compounds—the lower molecular weight PAHs, like naphthalene, fluorene, phenanthrene, and anthracene—were heavily partitioned into the vapor phase (> 75%). Naphthalene was the predominant PAH compound in air (48% of the total PAHs). From the simulated Fireground study , naphthalene was also the predominant PAH species accounting for 50% of the total. The relative ratio of the other PAH compounds and partition into vapor and solid phase in the exposure chamber appears to match that of a typical structure fire quite closely. Keir et al. (2020) also found that naphthalene was the most abundant PAH in the air in their study, with this compound accounting for 72% of all PAHs . Furthermore, probable and possible carcinogens (IARC Class 2A and 2B) accounted for 77% of the total PAHs compared to 62% here, while benzo(a)pyrene (IARC Class 1) was typically around 1% in the Keir et al. study . Some of these discrepancies may be attributed to the use of different analysis methods, including the type of sampling media (PUF and quartz tube vs. OVS-XAD-7 tubes). Overall, Keir’s analysis did not result in solid-phase PAHs at the same level reported in the current study. For example, in the current study, fluoranthene and dibenzo(a,h)anthracene accounted for over 14% of the total PAHs and were mostly captured in solid phase on the filter (Table 3). These two analytes only made up about 2.5% of the total PAHs reported in Keir et al. Overall, the trends in PAH distribution from the FES prop were similar to fireground conditions.
The variability in thermal and area air samples (Total PAHs and Benzene) were analyzed for the mannequin scenarios where measurements were conducted at a similar height throughout each scenario (1.5 m from the floor). Independent correlation analyses that included temperature, Total PAHs and Benzene suggested a weak relationship between each measurement. Benzene had a negative weak correlation with both Temperature (r = −0.303) and Total PAH (r = −0.230), while Total PAH and Temperature comparison showed a weak, positive correlation (r = 0.315). The peak temperatures measured in the exposure chambers for these scenarios were quite tightly grouped (118.2 ± 8.1°C) resulting in a coefficient of variation (CV) of 6.8%. However, the Benzene (CV = 22.5%) and Total PAH (CV = 45.6%) were significantly more variable. While this variability is quite high, it is similar to Total PAH measurements made during the Controlled Residential Fire Study where CV for Attack and Search groups were 65% and 45% respectively. It is important to note that different pump fault times for PAH and Benzene measurements, which also occurred during the data reported in , likely contribute to this high variability.