The area damage in PD 7974-7  is expressed according to various classes of spread of fire where the correspondent m2 of damage and related percentage of fires is provided. A distinction is given between sprinklered and un-sprinklered buildings and two different property types are analysed: in Table A.4 textile industry subdivided in Production area, Storage area and Other areas while in Table A.5 pubs, clubs, restaurants—all areas. The data of which these two classes are composed are referred to the UK Fire Statistics of 1984–1986. Tables A.4 and A.5 from PD 7974-7 are recreated using fire incident data from NFIRS including: property use, fire spread, number of floors below and above ground floor, average size of floor, percentage of area damaged, and presence of automatic extinguish systems.
All classes of property use have been analysed and in the USA fire statistics database, fire spread is classified according to the five different classes stated in Table 1 above. This subdivision for fire spread is similar to PD 7974-7 with some differences. PD 7974-7 fire spread confined to room of origin further specifies if the spread is confined to contents only or if the structure is involved; this is not present in the NFIRS. In the USA fire statistics when fire spreads beyond the room of origin, other specific classes than the simple one referred to the spread beyond the room of origin, are introduced (confined to floor, confined to building and beyond building of origin).
In the USA fire statistics, the damage represents the number of stories damaged by flame spread. Flame damage is the area actually burned or charred and does not include areas receiving only heat, smoke, or water damage. Damage is expressed as banded percentage ranges of area instead of unique m2 of area damage. Since these percentage bands present a range and not a unique value or percentage, the median of each band has been presented with error bars for the upper and lower bands (i.e. for Minor damage, median is 12% damaged area, with lower and upper bounds of 0% and 24%, respectively).
The total weighted area damaged is determined from the number of stories damaged, average floor area, and damage class involved per building. Damage and fire spread are also analysed in this paper related to the presence or absence of automatic extinguishing systems. Finally, the fire frequency for each fire spread classification has been evaluated with the distinction in fires and unclassified events.
Area Damage and Percentage of Fires in the Property Uses of the USA Fire Statistics
In general, the influence of sprinklers not only reduces the fire spread with fires 10% to 30% more likely to be confined to object or room of origin, but sprinklers also limits the damage from a fire, with higher percentages seen in the Minor and Significantly damage bands, rather than the Heavy and Extreme. It should be noted however that there are several fires within the NFIRS databases that have an assigned level of fire spread, however, they have no assigned damage class. These have been represented as unclassified (U) in Tables 5, 6, 7 and Tables 12, 13, 14, 15, 16 in Appendix (legend present in Table 4). The aforementioned tables also show the number of fires for each fire spread level for each damage class, percentages of fire spread per damage class (%F–S, percentages sum horizontally for the specific spread class) and percentages of damage per spread class (%F–D, percentages sum vertically for the specific damage class). The direction to follow to evaluate the two percentages are specified by the arrows in the aforementioned tables.
The property types for which sprinklers are effective and their action limits fire and spread, are Educational; Health care, Detention and Correction; Industrial, Utility, Defence, Agricultural, Mining; and Storage. Table 5 and Fig. 10 show, for Educational buildings, that the presence of sprinkler systems have significant impact on the spread of fire and the amount of damage these fires cause, with approximately 85% of these fires being confined to the room of origin, with a predominantly class of minor damage. When no sprinklers are present the fire has a more than 30% chance of spreading beyond the room of origin. No fires were reported, in the NFIRS data, to fire spread beyond the building of origin in sprinklered buildings, whereas this occurred in 3% of the fires for un-sprinklered educational buildings. The average area damaged for sprinklered buildings across all the four fire spread classifications is of almost 87 m2, compared to approximately 390 m2 across five class of spread when no sprinklers are present. Un-sprinklered educational buildings are also more likely to experience extreme damage compared to sprinklered buildings with 2660 m2 of damage confined to building.
Similar considerations as found for Educational buildings can be applied to Health Care, Detention and Correction. Sprinklers, again are shown to reduce the area damaged with average damage being 70.70 m2 and 94.51 m2 for sprinklered and un-sprinklered buildings, respectively (Figure 16 and Table 12 in Appendix). [NB: one fire incident has not been considered where the fire spread was confined to object first ignited however, it had a median area damaged of 89,186 m2].
Industrial, Utility, Defence, Agricultural, Mining as well as Storage buildings have limited data for sprinkler systems so a rigorous analysis is not possible but similar comments can be deduced for both property types where for un-sprinklered buildings, in more than 50% of cases the fire affects the whole buildings (Fig. 17, 18 and Tables 13 and 14 in Appendix).
Sprinklers limit the damage but not the spread in Assembly; Residential; and Mercantile, Business showing similar trends. Starting with Assembly, a single fire (no sprinklers, confined to floor, significant damage) has been removed from the analysis as the area damaged (in the range of 145,161 to 284,516 m2; 25% to 49% respectively) is significantly higher than others in the same class, and thus skews the data to unrepresentative values.
Table 6 and Fig. 11 show that, when sprinklers are present in Assembly in the USA, damage is predominantly Minor Damage, whereas in un-sprinklered buildings damage is predominantly extreme damage. This consideration supports the idea that automatic extinguish systems can reduce fire spread and can be demonstrated by the values of %F-S. When the fire spreads beyond the room of origin, un-sprinklered structures experience much higher damage and the frequency of fire spread beyond room of origin in un-sprinklered buildings is more than 50% of fires, compared to more than 25% in the sprinklered case. On average, a sprinklered and un-sprinklered Assembly building will experience 25.96 m2 and 59.50 m2 of damage, respectively. Fires are frequently confined within the room of origin for around 70% in sprinklered and 50% in un-sprinklered buildings. The data also indicates that if the fire does go beyond the room of origin, then it is likely to spread to the whole building, rather than being confined to the floor of origin, with sprinklered and no automatic extinguishing systems, having 24.61% and 43.85% of their fire spread, respectively, spreading past the floor of origin (Table 6 and Fig. 11).
Similar comments can describe Residential and Mercantile, Business with average damage being 4.92 m2 and 48.58 m2 for sprinklered compared to 35.69 m2 and 97.34 m2 for un-sprinklered, respectively (Figs. 19, 20 and Tables 15, 16 in Appendix).
In Manufacturing, Processing, sprinklers seem to limit the spread but not the fire damage. Indeed, 69% of fires are confined in the room of origin when sprinklers are present and 37% for un-sprinklered buildings (Table 7 and Fig. 12). For both sprinklered and un-sprinklered buildings the damage class is usually minor, however peaks are seen for both with significantly damage of 2856.61 m2 of area damage confined to building for sprinklers, and 2007.78 m2 confined to building for extreme damage for un-sprinklered buildings. Table 7 and Fig. 12 also show that for both sprinklered and un-sprinklered buildings the highest average area damage is reached for spread confined to building with 843.41 m2 and 839.56 m2, respectively.
It is evident from the analysis of these data that sprinklers can significantly reduce both the spread of fire and the associated damage of the fire. However, the NFIRS database presented the damage by flame only, and there is no indication within the data about the damage that may or may not be caused by the sprinkler systems, or by smoke.
Area Damage and Percentage of Fire Spread in PD 7974-7 vs USA Fire Statistics
Table A.4 in PD 7974-7 summarizes the area damage in m2 and percentage of fires for each category of fire spread in textile industry. Since the exact comparison to the textile industry was not possible with the USA fire statistics, comparisons are made instead to Manufacturing, Processing. Table A.4 is divided into three different fire locations, which have been reproduced in the USA fire statistics as follow in the NFIRS: Production area—Processing/manufacturing area, workroom, assembly area; Storage area—Storage Areas; Other areas—all the other fire origin locations excluding those previously considered.
The method explained in Sect. 6, has been adopted with the only difference that a further classification has been applied regarding fire locations. The influence of sprinkler systems on the response is also considered. The average area damage within Table 8 (last row) is obtained by multiplying the area damage for the related frequency for each class of fire spread, summing all values and dividing by the total frequency.
In recreating Table A.4 for Production area using NFIRS data, for sprinklered buildings, 7 fire incidents have been recorded for spread confined to floor, 14 for spread confined to building and 2 for the one beyond building. For un-sprinklered buildings, only 4 incidents related to fire confined to floor and 2 for spreads beyond the building of origin. One sprinklered building damage value (spread confined to building) has been ignored value in the dataset as the median damage is 85,935 m2, which is not comparable with other values.
Frequency of fire, in Table A.4 of PD 7974-7, not confined to the room of origin is equal to 4% for sprinklered and 12% for un-sprinklered, indicating good compartmentation. However, in the NFIRS, the spread beyond the room of origin is much greater at 24.07% for sprinklered and 51.73% for un-sprinklered (Table 8). When considering damage, PD 7974-7 has identical areas of damage for sprinklered and un-sprinklered buildings, whereas in the NFIRS analysis the damage for sprinklered buildings is always greater than for un-sprinklered. This is believed to be due to the small number of fire recorded in the case of sprinklered buildings.
For Storage areas, in the PD 7974-7 table, there are two mistakes: firstly the area damage confined to room, for contents only and for structure involved in sprinklered buildings, both have identical values of 19 m2 and 24% of frequency. If these values were real for both classes, then the cumulative frequency would exceed 100%. In this paper, we have combined the two classes and assigned them the 19 m2 and 24% damage and frequency, respectively. The second mistake is that the average area damage for buildings without automatic extinguish systems is wrongly evaluated since it should be equal to 533 m2 while in the PD 7974-7 table is 539 m2.
In the USA fire statistics analysis, one fire has been neglected as its median area damaged is too high compared to the others 16,350 m2. In the NFIRS, there is no data for fire spread beyond the building of origin for sprinklered buildings, and limited fires for confined to floor, only 5, and confined to building there are only 7. Moreover, fire departments have attended an overall of 50 fires associated with buildings without sprinklers. The results should therefore be taken with caution. In Table A.4 of PD 7974-7, fire frequency decreases as the fire spread class, and damage, increase. In the USA fire statistics, the fire frequency for spread within room of origin is 68.97% for sprinklered and 19.35% for un-sprinklered buildings (Table 8). There is also a significant 58.06% of fires confined to building of origin in un-sprinklered buildings. In general, as seen in Sect. 6.1, the presence of sprinklers decreases the damage and spread of fire as can be seen in Table 8 where for sprinklered see 31% of fires spread beyond room of origin compared to 80% for un-sprinklered buildings. This results in the average damage in un-sprinklered buildings being more that 1.49% of sprinklered buildings.
Last part of Table A.4 regards Other areas. In the PD 7974-7 table, fire frequency and area damage have the same trend as for Storage areas: fire frequency decreases with the increase of the spread of fire class while the area damage increases with the increase of the extension of fire (Table 8). In the NFIRS, for sprinklered building, only 6 fire incidents are recorded for fire beyond the building of origin. In the USA fire analysis, the spread within the room of origin is equals to 66.67% and 40.19% respectively for sprinklered and un-sprinklered buildings. There is also a 43.93% for fires confined to building of origin for un-sprinklered buildings. The average damage caused by the fire in the USA analysis is slightly lower in sprinklered than in un-sprinklered fires. The reason(s) behind this needs further investigation, but it is possible to say that sprinklers in other areas do limit the spread of fire as seen by the spread beyond room values of 33.33% and 59.82% for sprinklered and un-sprinklered structures, respectively.
Pubs, Clubs and Restaurants
Table A.5 is focused on pubs, clubs and restaurants and in order to recreate a direct comparison with the USA fire statistics, within the property type of assembly, the following subclasses have been considered and the results for each of them summed together: Athletic or health club (includes YMCA or YWCA, lodge, swimming, and baths); Clubhouse associated with country club that includes golf, tennis, hunting, fishing, and riding activities; Yacht club (includes boating and yacht club facilities; excludes marinas, boat mooring facilities; boat repair/refuelling facilities; or boat sales, services, and repairs); Casino, gambling clubs (includes bingo halls—use only where primary use is for gambling); Clubs, others; Restaurant or cafeteria, places specializing in on-premises consumption of food (includes carryout and drive-through restaurants); Bar, nightclub, saloon, tavern, pub; Eating, drinking places, others.
As for Table A.4, Table A.5 presents fire frequency decreasing with increasing fire spread classes and increasing area damaged. In USA fire statistics, in presence of sprinklers, only 7 fires occurred for the class of fire beyond the building of origin. Fire frequency, from the USA statistics, of spread confined to the room of origin is 70.23% and 50.38% for sprinklered and un-sprinklered buildings, respectively (Table 9). Average area damage for both countries present lower values when buildings are equipped with sprinklers, and is significantly less when considering damage beyond room of origin.
The data recreated for Tables A.4 and A.5 show that the PD 7974-7 statistics may be no longer representative of the damage and frequency of fires as observed in fire statistics from the USA.