Occupant Tenability in Single Family Homes: Part I—Impact of Structure Type, Fire Location and Interior Doors Prior to Fire Department Arrival

Nine of the experiments took place in a one-story structure, designed to be representative of a home constructed in the mid-twentieth century with walls and doorways separating all of the rooms and 2.4 m ceilings throughout. The one-story structure had a floor area of 112 m²; with three bedrooms, one bathroom and eight total rooms (Fig. 1). The house was wood framed and lined with two layers of gypsum board (Base layer 16 mm, Surface layer 13 mm) to protect the structure and allow for multiple experiments. All of the windows were filled with removable inserts so that window failure did not occur in any scenario. The leakage area determined from a blower door test was found to be approximately 0.1 m².

The two-story structure had an area of 297 m²; with four bedrooms, 2.5 bathrooms and twelve total rooms (Figs. 2, 3). The structure incorporated features common in twenty-first century construction such as an open floor plan, two-story great room, and open foyer. The house was a wood framed structure and lined with two layers of gypsum board (Base layer 16 mm, Surface layer 13 mm). All of the windows in this structure were filled with removable inserts so that window failure did not occur in any scenario prior to fire department intervention (see Part B). The leakage area determined from a blower door test was found to be approximately 0.2 m².

Figures 1, 2 and 3 include 3-dimensional renderings of the floorplan in each house with ignition and furniture locations (Table 1). The living room in the one-story house as well as the family room and living room in the two-story house were furnished similarly; with television stand, television, end table, lamp with shade, coffee table, chair, ottoman, two sofas, two pictures, and two curtains. The floor was covered with polyurethane foam padding and polyester carpet. The fuel loading was approximately 29 kg/m². To describe the potential energy of the fuel package, a test with the living room furnishings was performed in a compartment with a large opening (6.5 m² of ventilation area) under a cone calorimeter resulting in a maximum heat release rate of 8.8 MW and a total of 4060 MJ of heat released [3]. The scenarios reported in Part I of this series are conducted with all windows and door closed, resulting in underventilated conditions and lower heat release rates. This manuscript focuses on occupant exposures from the fire prior to ventilation and flashover did not occur in this timeframe (though flashover did occur after ventilation). Part II will analyze conditions after fire service ventilation [20].

Bedroom 1 in both houses was furnished with a queen bed comprised of a mattress, box spring, wood frame, two pillows and comforter. The room also contained a dresser, television stand and television. The floor was covered with polyurethane foam padding and polyester carpet. The remainder of the bedrooms (2 to 4) in both houses were furnished with the same bed, armoire, television and flooring compliment as well as a smaller dresser, headboard, and a framed mirror. A heat release rate experiment was conducted with the bedroom furnishings in a compartment with a large opening (6.5 m² of ventilation area) under the cone calorimeter. The maximum heat release rate was 9.4 MW and the total heat released was 3580 MJ [3].The dining room of both houses was furnished with a solid wood table and four upholstered chairs. The kitchens were furnished with the same type of table and chairs as the dining room, as well as a dishwasher, stove, refrigerator and wood upper and base cabinets with cement board counters. The floors of the dining rooms and kitchen were also cement board to simulate a tile floor.

The same make and model of all of these fuels (with the exception of Experiment 17) were purchased from the same supplier and stored in an environmentally controlled warehouse before they were used in an identical layout in the structure for each experiment. Experiment 17 was conducted using a different fuel package than the remaining experiments. While the room layout was the same, the natural fiber furnishings are intended to provide a relative risk from a common structure fire of 50+ years ago.

While significant amounts of instrumentation were included in each of these structures (Figs. 1, 2, 3), this manuscript will focus on gas temperature and concentration data collected in the fire rooms and bedrooms at a height of 0.9 m from the floor. Gas temperature was measured with bare-bead, type K thermocouples, with a 0.5 mm nominal diameter in locations shown in Figs. 1, 2 and 3. The uncertainty in type K thermocouple measurements is less than 1% to 2% of the measured value for temperatures up to 1250 K [21].

Gas concentrations of oxygen, carbon monoxide, and carbon dioxide were measured using Ultramat 23 NDIR from Siemens at 0.9 m from the floor adjacent to the front door and in bedrooms 1, 2 and 3 for both houses. The uncertainty of the measured concentration is 1% of the maximum concentration measurement. The maximum concentration measurements were 1% by volume for CO and 10% by volume for CO₂. The gases were extracted from the corners of rooms to minimize transport length from sample location to the sensor and reduce the risk of damage during firefighting operations. All data was collected at a frequency of 1 Hz.

For this study, tenability was calculated based on the measurements of air temperature and CO/CO₂. Other factors could contribute to increased FED values and lower times to untenability, including the effect of radiant heat (particularly in the fire room) and the presence of HCN and other gases in the structure. However, due to experimental limitations, these factors are not considered here. The FED values from HCN should scale with CO/CO₂. So although are values may be conservative, there is consistency in the comparisons.

All of the experiments started with the exterior doors and windows closed, the roof vents closed, and all of the interior doors open except for Bedroom 3 in the one-story and Bedroom 2 in the two-story structure. The fire was ignited on a sofa in the living room (one-story) or family room (two-story), in a trash can next to the bed, or in a coffee maker on the kitchen countertop (Figs. 1, 2, 3). The ignition of each experiment was performed with a set of matches that were spark ignited on a fuel source (couch in the living/family room, trash can next to the bed in the bedrooms, and towels under a cabinet in the kitchen) in the room of interest [3].

A flaming fire was allowed to grow until ventilation operations were simulated. Fire service ventilation for each scenario was determined based on three factors; time to achieve ventilation limited conditions in the house, potential response and intervention times of the fire service, and window failure times from previous window failure experiments [5]. Times to arrival on-scene vary greatly based on fire department capabilities and response distance. NFA 1710 suggests that departments should provide for the first arriving engine company to be on-scene within approximately five minutes (80 s for turnout, 240 s for travel time) after alarm handling (which includes 15 s for alarm handling [95% of the time], and 64 s for processing [90% of the time]) [22]. According to NFPA 1720, the goal for fire emergency response for volunteer departments is to arrive at the scene at a maximum of 9 min in an urban area (~384 people/km²), 10 min in a suburban area (192 people/km² to 384 people/km²), 14 min in a rural area (~192 people/km²) and directly related to driving distance for remote areas greater than 8 miles from the closest fire station [22]. Of course, these times do not include the time to detection and notification, which can also vary greatly (60 s to 310 s [23]). To account for this variation, while still achieving objectives for other components of this study, firefighter intervention was largely determined on achieving ventilation limited conditions for each scenario, within the realistic timeframes provided by NFPA standards. In all cases, temperatures were relatively stable (see Fig. 4) such that accumulation of additional exposure and increased FED is estimated to be constant. Therefore, to allow estimation of changes in FED if ventilation were delayed beyond the intervention time used here, the instantaneous rate of FED per second will be reported at the time when initial ventilation was provided.

For living room fires in the one-story structure, ventilation began at 8 min after ignition for most experiments (to simulate quick fire department arrival and due to the fire stabilizing under ventilation-limited conditions). The two exceptions were Experiment 15 at 6 min (ventilated to study the impact of flow path and fire spread) and Experiment 17 at 24 min (to allow for the legacy fire to become ventilation-limited), while the two-story house was ventilated 10 min after ignition for the family room fires. The additional time in the two-story structure enabled ventilation limited conditions, as more time is needed for oxygen to be consumed in a larger volume. In all bedroom scenarios, ventilation occurred at 6 min after ignition due to the smaller fire room compartment and simulated window failure, while kitchen fires were ventilated 10 min after ignition (to allow for ventilation-limited conditions). As an example, Fig. 4 shows a typical evolution of temperature with time in the one-story structure from Experiment 3, collected at a height of 0.9 m. This experiment began with ignition on the couch in the living room, which grew until the fire became ventilation-limited. The temperatures then began to decrease until 8 min into the experiment when an initial ventilation opening was created, in this case, by opening the front door. The data analyzed here focuses on tenability prior to Fire Service intervention, and will largely consider temperature and gas concentrations up to the times listed above. However, for reference and to contextualize this data, some of the following tables do include exposures after fire department intervention (Tables 2, 3).

Occupant tenability, which is the survivability of occupants in the fire environment, is a primary concern for any firefighting operation. Two standard measures of occupant tenability were used during these experiments—temperature and gas concentration-based upon the fractional effective dose methodology (FED) from ISO 13571 [15]. This methodology provides a method to calculate the time to incapacitation based on an accumulated exposure to either toxic gases: or local ambient air temperaturewhere \( \upsilon_{{CO_{2} }} \) is a frequency factor to account for the increased rate of breathing due to carbon dioxide, \( \varphi_{{CO_{2} }} \) and φ _CO are the mole fractions (%) of carbon dioxide and carbon monoxide, T is the temperature near the occupant (°C), and \( \Delta t \) is the time increment of the measurements made in the experiments in minutes (1/60 in these experiments). According to ISO 13571, the uncertainty in Eq. 1 is ±20% and the uncertainty in Eq. 2 is ±35%. Equation 3 only applies for temperatures greater than 120°C, which is taken as the lower limit to this method. Gas concentration measurements did become saturated for some of the experiments (1% by volume for CO and 10% by volume for CO₂), so the FED values that we report are conservative estimates. Tables 2, 3, 4 and 5 will indicate which specific samples were saturated.

FED relates to the probability of the conditions being non-tenable for a certain percent of the population through a lognormal distribution. For reference, FED = 0.3 is the criterion used to determine the time of incapacitation for susceptible individuals (young children, elderly, and/or unhealthy occupants) and corresponds to untenability for 11% of the population, and FED = 1.0 is the value at which 50% of the population would experience untenable conditions.

FED’s were calculated at an elevation of 0.9 m above the floor for both houses, representative of exposures that would be experienced by a person crawling on the floor. The time to exceed the thresholds for all of the experiments in each house for both heat (only convection considered as no radiant heat flux measurements were made) and carbon monoxide/carbon dioxide are calculated for both houses in living rooms and bedrooms, both with doors open and closed. It should be noted that the values assume the occupant was in that location for the duration of the experiment up to ventilation. These estimates may be considered lower bound scenarios as additional thermal risks may be present from exposure to large radiant heat exposures or from the additive effects of exposure to a variety of different hazardous gases.

While these measurements estimate exposures for the most likely case of an occupant crawling in smoky conditions, it is acknowledged that both heat exposure and toxic gas exposure will be larger with increasing elevation in the structure. If an occupant is standing or attempting to walk out of the structure, higher FED values and lower times to untenability will likely result.

Four experiments in the single story structure (Experiments 1, 3, 5, and 7) are all identical in terms of structure and fuel package layout and materials, ignition location (Living Room), and ventilation conditions up to the 8 min where firefighters intervened. Environmental conditions were held to a high level of repeatability in terms of temperature, moisture and air velocity (basically still). However, outside of the closed bedrooms, FED values ranged dramatically. For the fire room (living room), FED_temp ranged from 4.0 to 6.8, while FED_CO in Bedroom 2 ranged from 1.2 to 4.5. For the two-story structure, a similar series of experiments were run with the Family Room as the ignition location. Nearly identical fuel packages were ignited in experiments 2,4,6,8, and 12. For the fire room, FED_temp ranged from 2.4 to 9.8, while FED_CO in Bedroom 3 ranged from 0.2 to 0.5.

Time to untenability for these ‘room and contents’ fires with limited ventilation is much improved on the second floor of the two story structure compared to the one-story structure, largely due to the increased volume of the structure. If the same volume of CO is generated by these fires, the two-story structure will have a smaller CO concentration because of the increased dilution with air in the enclosed space. Additionally, carbon monoxide generation typically increases as the fire becomes oxygen-limited. Since the two-story structure has more oxygen available inside the enclosed space at the start of the fire, CO generation is likely to increase more slowly than in the one-story structure. Importantly, while the second floor bedrooms are more tenable than the first floor spaces, egress through the interior of the structure would likely require exposure to the highly untenable conditions on the first floor. Firefighters should consider this fact in the risk–benefit analysis when employing vent-enter-isolate-search techniques to rapidly access victims on the second floor from the exterior as opposed to attempting rescue through the high temperature environment on the interior of the first floor.

For the single story structure, the FED_CO values in the target bedrooms with open doors were consistently higher in Bedroom 2 compared to Bedroom 1. The trend for FED_temp was not as clear, though the values were higher in Bedroom 2 compared to Bedroom 1 for two experiments and similar in the remaining three living room fire experiments. This affect could possibly be attributed to the smaller volume of Bedroom 2, the orientation of the door at the end of the hallway, or the distance from the heat source which results in lower temperatures further from the heat source and thus less stratification of the gas layer. For the two-story structure, FED_CO was similar in Bedroom 1 and 3. However, FED_temp was significantly higher in Bedroom 3, exceeding 0.3 for all 5 scenarios while never exceeding 0.2 in Bedroom 1 for any scenarios. Interestingly, the largest FED in the second floor bedrooms was due to thermal effects for Bedroom 3, but due to CO in Bedroom 1. Again, this may be due to the distance from the heat source resulting in lower temperatures but also less stratification of the gas layer. Additional research would be required to fully understand and decouple these potentially interacting effects.

The bedroom fire scenarios each transition to ventilation limited conditions more rapidly than the Living Room fires, and result in higher fire compartment temperatures. FED at the time of fire department intervention for the bedroom scenarios was well in excess of 30, suggesting that more than 99.9% of the population would be incapacitated. These fires are ignited in smaller compartments providing significant re-radiation and more rapid growth. For the single story structure, FED_CO values produced by the bedroom scenarios were similar to or larger than those produced by the Living Room fire despite fire department ‘intervention’ 2 min earlier than the Living Room scenarios. At the same time, the temperature increase in the non-fire rooms is relatively small, with the maximum FED_temp = 0.21 in the adjacent bedroom.

For the two story scenarios, where the fire was ignited in the second floor bedrooms, the FED_temp values are almost three times higher than similar fires in the single story structure, even with identical furnishings and similar room size. However, this is partially attributed to the longer times to firefighter ‘intervention’ in those experiments. Additionally, for open bedrooms on the second floor, the FED_CO values were 20× to 30× higher in the bedroom fires than those measured during the family room fires. As with the living room/family room scenarios, the largest risk for remote victims is again gas exposures, but the risk is relatively more elevated in the bedroom fire scenarios because the fires become locally ventilation-limited due to their confined nature. It is also likely that since these scenarios resulted in higher ambient fire room temperatures more rapidly, they were able to sustain the combustion process even at lower oxygen concentrations, producing relatively larger amounts of CO. On the first floor of the structure, there was little measureable impact on tenability, most likely due to the buoyant nature of the combustion products. Furthermore, without a ventilation location for combustion products to escape, air from the first floor is not as easily entrained into the oxygen-limited fire on the second floor.

The kitchen fuel package (Experiments 13 and 16) resulted in low FED compared to the living room and bedroom fuel packages in both structures. For the kitchen fire scenarios—and typical of common structures in the US—the majority of the fuel is wood cabinets and countertop appliances of hard plastic. Fewer soft and/or foamed polymers are typically found in the kitchen. At the time of fire department intervention (even delayed to 10 min), FED_temp < 0.01 in all target rooms. The worst case bedroom FED_CO = 0.5, which is equivalent to the lowest value measured for the living room fire. The fuel sources in the kitchen fires consisted mostly of wood cabinets and countertops that burn slower and take longer for the structure to reach ventilation-limited conditions. CO production increases significantly when the fire reaches ventilation-limited conditions [24], and thus there is more CO produced by the bedroom and living room fires. Kitchen fires are the most common source of residential fires (43%), but fortunately appear to be the most survivable based on results from this study.

Finally, experiment 17 was added to the test series to provide a comparison with furnishings constructed from mostly natural materials (sometimes referred to as legacy furniture [5]) as opposed to the largely polymer based furnishings that are currently common in US households. The average times to achieve untenable conditions for Experiment 17 were well beyond the timeframe of initial fire service intervention (24:30 for FEC criteria of 0.3). This is an increase of approximately 20 min compared to the experiments with living room furnishings common in the twenty-first century in the one-story structure. Compared to the NFPA 1710 and 1720 based response timeframes, it is apparent that firefighters of the past responding to fires with these fuels were likely to find survivable victims more readily than fires involving fuel loads typical of today’s structures. At the same time, fire-related occupant fatalities have continued to decline over the past several decades in apparent contrast to the tenability data presented here. Thanks to progress in public education, fire safety initiatives, widespread use of smoke detectors, and increasing installation of active fire sprinkler system, fire protection engineers have been successful at not only keeping pace with this increasing tenability risk, but actually affecting an improvement in life safety.

While improved detection, suppression and public education have helped to drive down fire related injuries and fatalities, a complimentary initiative can be supported by the notable tenability levels in Bedroom 3 of the one-story structure and Bedroom 2 of the two-story structure. Both of these rooms had the interior doors closed for the duration of the experiments, physically separating these spaces from the fire room. Importantly, the times to untenability found in Tables 2 and 4 suggest that occupants in compartments with closed doors never receive an FED > 0.3, even though an immediately adjacent bedroom may reach FED = 0.3 in approximately 5 min or less. In every case, thermal FED in the bedroom behind closed doors remained less than 0.01. The maximum FED based on CO exposure in these rooms was measured for living room fires at 0.11, which would be considered untenable for 1.4% of the population. For this same scenario, the adjacent bedroom with open door resulted in a measured FED_CO = 4.51, which would be untenable for 93% of the population.

In order to quantitatively characterize the improvement in tenability behind closed doors, FED ratios at the time of firefighter intervention were calculated and they are reported in Tables 9 and 10. The FED ratio is calculated for the nearest bedroom of the same dimensions compared to the closed door bedroom (BR2/BR3 for one-story and BR3/BR2 for two-story). Two scenarios for the two-story structure utilized bedroom 3 as the fire room, so in this case, the bedroom 1 is the open bedroom control. This bedroom is farther away from the fire room than bedroom 2 and larger, so should provide a conservative FED estimate. In some scenarios, the FED behind closed doors is very small, so a lower limit of FED = 0.01 is utilized for these calculations to bound the calculation. In all cases, the maximum FED—based on either temperature or gas—is utilized for each room.

Due to the relatively small FED in the closed bedroom, the FED ratio varies widely even for the same ignition location. However, for the single story structure, the FED ratio ranged from 7.3 for the kitchen scenario (which resulted in FED <0.5 throughout the structure) to over 400 for a Living Room scenario. The median value (for all 17 experiments) was 46—a potential trapped victim behind a closed door would be exposed to a 46× lower FED than those in a bedroom with an open door.

In both cases, the lowest FED ratio behind closed doors was for the Kitchen scenarios, which were significantly longer and had relatively low temperatures compared to the other tests. For the two story structure, the largest FED ratio was found for the Bedroom fires, which occurred on the same level as the other bedrooms. For the one story structure, there was little difference in the FED ratio from the Bedroom to Living Room fires.

Once again, this data suggests the importance of teaching the public the value of a comprehensive fire safety plan in residential structures. As mentioned earlier, the rapid accumulation of an incapacitating FED in a timeframe that is well within the 2 min to 16 min RSET analysis of Proulx et al. [22] highlights the need for rapid fire detection and notification throughout a structure as well as active suppression systems that can control the fire. At the same time, certain individuals will not feasibly be able to respond rapidly enough to self-evacuate, in which case the critical message of sheltering behind a closed door should be shared. The tables included in this manuscript show the unequivocal improvement in tenability behind closed doors, particularly for those who may be susceptible to smoke exposure and also have a long RSET (young, elderly, mobility impaired). Furthermore, for those individuals whose means of egress may be cut off by the progression of a fire, the value of sheltering behind closed doors should be reinforced based on this data.