A Risk-Based Approach to Assess the Effectiveness of Sprinklers in Buildings with Combustible Cladding

Abstract

This study investigates a risk-based approach to evaluate the effectiveness of sprinklers in residential buildings to offset the risk premium imposed by combustible cladding (expanded polystyrene and aluminium composite panels) installed on such buildings in Victoria, Australia. This approach builds upon the Initial Fire Spread in Cladding Assessment Number (IF-SCAN), a concept pioneered by Cladding Safety Victoria as a triage tool in their rectification program. The analysis uses published data from real fires in buildings with and without sprinkler systems installed. It considers three criteria: death rates, injury rates, and construction cost. The construction cost was determined using an existing costing model currently employed in Victoria. The results of this study suggest a higher risk tolerance can be accepted for combustible cladding on buildings equipped with sprinkler systems over that set out in government policy. More specifically, it was found that a building fully protected by sprinklers can generally counterbalance the fire risk posed by combustible cladding spanning up to seven apartments, while a span of up to ten apartments could be considered for buildings without balconies or a private courtyard.

1. Introduction

Sprinkler systems greatly reduce the risk to life in building fires in general. This is evidenced and recognised around the world. Recent changes to various building codes around the world, including those in Japan, the United Kingdom, and Australia, further supports this. Fire safety engineers often incorporate sprinkler systems into their fire safety strategies when devising solutions for buildings that do not fully adhere to prescriptive building code provisions. Sprinkler systems make buildings inherently safer. When compared to buildings without such systems installed, the risk posed by any fire is mitigated, resulting in less severe fires, smaller extent of property damage, less severe injuries, and fewer fatalities.

On the other hand, combustible cladding installed on residential apartments buildings could have the potential to result in an unwanted and adverse outcome if exposed to a fire. Some of the factors that would impact what such an outcome could result in include, but are not limited to, the amount, extent, location, and type of cladding material, as well as the wall make-up and how the material is fixed in place.

This paper looks at quantifying the benefits provided by sprinkler protection in a building to “offset” the risk posed by the combustible cladding installed to the external walls of that building, particularly:

• The relative cladding risk in a sprinkler-protected building compared to a general fire risk in the residential building cohort that is not sprinkler-protected and does not have combustible cladding; and
• The relative cladding risk in a sprinkler-protected building compared to other buildings in the Elevated risk category that are not sprinkler-protected and do not have combustible cladding.

This work is part of the Australian State of Victoria’s government response to understanding, assessing, and mitigating risk from combustible cladding installed onto residential apartment buildings. Cladding Safety Victoria’s focus was a risk-based approach to advise on and mitigate said potential risk [1]. Buildings with combustible cladding were initially categorised as either Low, Elevated, or Unacceptable buildings. Ministerial Guideline 15, which was issued in Victoria, Australia under the Building Act of 1993 on 21 September 2023 [2], sets out and defines the risk thresholds for Low, Elevated, and Unacceptable risk pertaining to cladding. To ensure a large number of buildings could be assessed in a succinct and repetitive manner, a robust framework had to be developed. This framework is referred to as the Protocols for Mitigating Cladding Risk (PMCR). This framework is not designed to assess new design or buildings. In this framework, the two most popular combustible cladding types in Victoria, namely aluminium composite panels (ACPs) and expanded polystyrene (EPS), are considered. Further development of other aspects of PMCR, including the investigation on non-sprinkler protected buildings and the review on risk assessment methods, can be found on Cladding Safety Victoria’s website, Protocols for Mitigating Cladding Risk—Cladding Risk Prioritisation Model Methodology [3,4]. In line with the overall development of the Protocols, the definition of cladding risk is considered as the “risk of fire spread across the combustible cladding” of different sole occupancy units [3]. The method of risk assessment is presented in Section 2 and the determination of risk is described in Section 2.6 The results of the risk assessment are presented in Section 3 followed by the discussion in Section 4.

2. Methods

The methodology was a review of statistical data for fires, sprinkler efficiency, and fire death and injury rates, in combination with the increased likelihood of exposure of fire to occupants in sole occupancy units (SOUs) due to combustible cladding externally connected in their façades. The cladding risk is quantitatively assessed based on the likelihood of fire initiation, fire spread, and the possible consequence of fire spread (cladding risk = likelihood of fire starting $\times$ likelihood of fire spreading $\times$ consequence). The consequence focuses on the life safety of occupants, namely the death and injury rates. External combustible cladding products that continuously overlay part of the floor plan of one or more SOUs are referred to as a cladding cluster [4].

2.1. Data and Statistics

This review was conducted with different sources, including academic publications, public databases from government agencies, and inter-government communications. Data and statistics from Australia, in particular from the State of Victoria, was sought and reviewed at the first instance. The review was then expanded to cover data and statistics from other countries, including United States of America (US), England, Canada, and New Zealand. These countries were seen, by the authors, to have a building stock and culture similar to that of Australia. Additionally, the construction culture in these countries, especially the US, is seen to share similarities with Australia, including stringent governing regulations, industry structures, and productivity [5]. The datasets and statistics available from Australia and most of these other countries were more limited and/or outdated than US data, making them less robust for the calculations and assessments. The authors therefore decided to rely on data from the United States for all inputs. Using data from a single country was seen as a way to minimise the potential for discrepancies in the results. This was done after confirming that the data and statistics from the other countries, including Australia, appeared to correlate to what the US data was portraying. An example to contrast the availability of data in Australia vs. US was evident from Frank et al.’s work, where Australian data on sprinkler’s effectiveness was dated back to 1986 [6]. It should be noted that when comparing to the other countries, in particular Australia, the data from the United States appears to be of a conservative level.

2.2. Likelihoods of Fire Starting and Spreading

2.2.1. Fire Initiation

Fire initiations were split into two possibilities, either an internal SOU fire, or an external balcony fire.

For the period 2016 to 2020, there were on average 343,100 fires in home structures according to data from the National Fire Protection Association (NFPA) [7], of which 5%, or 17,155, were classified as “Unclassified outside area” fires.

As part of the assessment we have, therefore, conservatively assumed that all fires that were classified as “Unclassified outside area” fires were balcony fires. As the value incorporates balconies as well as other areas outside, this is a conservative assumption that increases the likelihood of a potential fire impacting the combustible cladding.

Data from Canada spanning the period between 2005 and 2015 resulted in Garis and Chris stating that “10% of the multi-residential building fires originated from an outside area (either the exterior balcony (including open porch or deck) or court/patio/terrace area)” [8].

Similarly to internal fires, a fire originating on a balcony is expected to be subdued or even extinguished if a sprinkler system is installed. There appears, however, to be little data for balcony fires and even less data pertaining to the reliability of sprinklers installed on balconies. Sprinklers on balconies can, however, be expected to improve safety and reliably mitigate most of the risk posed by fire on a balcony [9]. Irrespective of this fact, it is conservatively assumed that each of those balcony fires would lead to the ignition of the combustible cladding, regardless of the presence of sprinklers. This approach will be used in the methodology via a conservative assumption that the probability of cladding ignition in the case of an external fire is 1.0. The location of the fire on the balcony is hence ignored and simply assumed to be adjacent to the cladding, resulting in cladding ignition. This effectively ignores the benefit that sprinkler protection on the balconies would have.

To ensure conservatism in the analysis, a higher estimate of 10% for fires originating on balconies was used in the cost calculations. The remaining 90% of fires were assumed to start internally within a single occupancy unit (SOU). For the injury and death calculations, raw data was used directly. However, due to the inability to extrapolate injury and death rates from fires originating inside an SOU to those starting externally (on balconies), recalculating these rates for comparison between buildings with and without balconies was not feasible. Instead, the analysis assumes, based on available data, that 5–10% of fires occur on balconies and that each SOU is equipped with an attached balcony.

It is assumed that the likelihood of a fire starting in any given SOU will be identical for each and every SOU, i.e., that it is not plausible that two SOUs (or balconies) will catch fire at the same time from different and independent ignition sources.

Ignition sources other than a fire originating in an apartment or on a balcony/private open space, such as a vehicle parked in close proximity to the cladding, is not considered in this analysis. This is because the impact of such potential fire sources is assessed separately in another component of the PMCR. Typically, this will be a fire source located at ground level and near combustible cladding. As it is at or near ground level, it would generally also be easily accessible, meaning that removal of said cladding would be relatively easy (if deemed necessary).

2.2.2. Fire Spread

To analyse the benefits of sprinkler protection against the potential impact of combustible cladding installed, as well as for conservatism, in the assessment, a fire is assumed to always have the potential to grow to the point that it reaches flashover and involves the combustible cladding. Every internal fire is, in other words, conservatively assumed to exit the SOU to reach and involve the combustible cladding 100% of the time when the sprinkler system is presumed to not be effective.

Fire spread via the cladding is then conservatively assumed to reach all connected SOUs using an upward spread probability of 100% of the time and downward spread probability of 90% of the time.

Sideways fire spread was not looked at separately. It is, however, acknowledged that horizontal fire spread will be significantly less than vertical fire spread. This can be observed from cladding fires that have occurred. For the purposes of this paper, horizontal fire spread has not been calculated separately and is instead conservatively considered identical to vertical fire spread.

All possible fire spread scenarios are then added up together to find the overall likelihood of fire occurrence within the cluster.

It should be noted here that these are very conservative assumptions being added together to ensure a conservative outcome. This is confirmed through statistics, which show that the majority of fires occurring in buildings with sprinkler protection installed will be controlled or even extinguished following activation of the sprinkler system. Data reported by the NFPA [10] reported that 96% of home fires were confined to the object or room of origin when a sprinkler system was present. This is proof that fire spread is efficiently reduced when sprinklers are installed. This means that it is less likely that the fire would reach or impact on combustible cladding external to the building. The authors acknowledge the uncertainty in data collection, and particularly the exclusion of wind and other environmental factors from this analysis, noting that the statistics relied upon are data from actual events. In response to such uncertainties, the assumptions on fire spread and other conservative assumptions have been consistently used throughout the study.

2.3. Consequences from Fire

2.3.1. Death

Based on data published by NFPA [10], the death rates per 1000 fires has been assumed to be 1 and 8.3 in sprinkler-protected and non-sprinkler-protected SOUs, respectively. The reduction in death rates is, therefore, 88%, which is in line with the 90% ± 4% reduction reported in the UK [11]. The same NFPA document states that “when sprinklers were present, almost all of the home structure fires were confined to the object or room of origin. Most of the civilian deaths and injuries resulting from fires in homes with sprinklers were caused by these fires. In home fires that lacked AES [Automatic Extinguishing System], 72 percent of the fires were confined to the object or room of origin. Less than a fifth of the deaths and less than half of the injuries in home fires with no AES present resulted from such fires”. Further data from [10] is presented in Figure 1.

In the calculations, it has been conservatively assumed that the ratio of death/injury beyond the room of fire origin follows the ratio of death/injury in the room of fire origin.

Table 1. Summary of AES presence and type in reported home structure fires, excluding properties under construction [10].

	Fires		Civilian Deaths		Civilian Injuries		Damage (in Millions)
Sprinkler system present	23,675	7%	22	1%	550	5%	$249	3%
No suppression system present	310,677	92%	2578	99%	10,454	94%	$7295	96%

includes both apartment buildings and one- and two-family residential buildings. In July 2021, the US Federal Emergency Management Agency (FEMA) presented values in a report titled “One- and Two-Family Residential Building Fires (2017–2019)” [12], which is presented in

Table 2. Loss measures for one- and two-family residential building fires (3-year average, 2017–2019) [12].

Measure	One- and Two-Family Residential Building Fires	Confined One- and Two-Family Residential Building Fires	Nonconfined One- and Two-Family Residential Building Fires	Residential Building Fires (Excluding One- and Two-Family)
Average Loss
Fatalities/1000 fires	7.9	0.0	13.2	3.3
Injuries/1000 fires	25.3	5.5	38.4	25.7
Dollar loss/fire	$22,030	$290	$36,390	$13,190

. The report presents property loss values for both one- and two-family buildings, as well as other residential building. It gives insights into deaths and injuries per 1000 fires and loss (cost) per fire.

The number of households in the US, based on the type of household, e.g., 1 housing unit, 2 to 4 housing units, and condominiums, is estimated from US census data [13].

2.3.2. Injury

The extracts from NFPA above appear to signal that the initiating event itself and/or first-aid firefighting by residents themselves is the governing contributor to the injuries and deaths.

The same document also shares that between 2017 and 2021, there was an annual average of 23 injuries per 1000 fires where sprinkler systems are installed vs. 33 injuries per 1000 fires where no sprinkler system is present.

The presence of sprinklers also has an effect on the severity of the injuries reported. NFPA [14] states that “sprinkler presence is associated with a 41% reduction in total cost of injuries per 100 reported home fires”.

To be able to mathematically compare the effects of sprinklers benefit vs. cladding risk, when calculating and comparing the injury component, the calculations discount the overall number of people to account for the reduction of injury severity.

2.3.3. Cost

In 2012, the Fire Protection Research Foundation, in support of the NFPA, presented a study on “sprinkler impact on fire injury” [14]. This research confirms the reduction of the cost associated with a fire when the property is sprinkler-protected.

From the roughly 350,000 fires recorded, the research concluded that if a sprinkler system had been installed in those buildings, there would have been savings of:

• $0.2 billion a year on medical costs associated with civilian fire injuries (a 53% reduction);
• $0.7 billion a year pertaining to civilian fire injury cost (a 41% reduction);
• $10.4 billion a year pertaining to death (6 lives saved per 1000 fires); and
• $4.8 billion a year pertaining to property damage (a 69% reduction).

In 2017, the NFPA undertook a similar study [15]. The Australian Building Codes Board’s (ABCB) subsequently lowered the building height threshold/“trigger point” for when sprinklers are required in residential apartment buildings under the prescriptive deemed to satisfy provisions of the National Construction Code of Australia. The ABCB’s decision for lowering the threshold was at least in part due to the NFPA study.

2.4. Benchmarks

The Victorian government, in Minsters Guideline 15 (MG-15) [2], set out the accepted thresholds for what is referred to as “Low” risk and “Elevated” risk. In addition to these benchmarks, the risk to (the individuals within) an individual SOU as well as the individuals within a cluster were used as ultimate benchmarks.

MG-15 refers to the Cladding Risk Mitigation Framework for the definitions of “Low Cladding Risk” and “Elevated Cladding Risk”, which are defined as follows for a non-sprinkler-protected building:

• Low: “Risk of fire spread across the Combustible External Cladding of ≤1 SOU”
• Elevated: “Risk of fire spread across the Combustible External Cladding of 2 SOU”

2.4.1. At the SOU Level

The calculations that follow later in this document benchmark the individual SOU risk to each of the (individuals within) SOUs in a sprinkler-protected building which has combustible cladding connecting those SOUs, against the risk to (the individual within) an SOU of a non-sprinkler-protected building where there is no cladding connection to adjoining SOUs.

2.4.2. At the Cluster Level

The risk at the cladding cluster level speaks to the overall societal risk within a cluster that is connected by combustible cladding. This was benchmarked against the combined risk of the equivalent number of SOUs in a non-sprinkler-protected building, were there is no combustible cladding.

2.5. Sprinkler Protection and Sprinkler Failure

2.5.1. Intent

The intent of the sprinkler system is to control or extinguish an internal fire before such a fire spreads to the combustible cladding. It is also a means for activating the general fire alarm for the building. In addition to this, where fire spreads from the cladding to another SOU, the sprinkler system is a mitigating measure to control or extinguish such a fire within that SOU.

2.5.2. Efficacy

Where a sprinkler system installed, it is generally considered to control or even extinguish a fire. It is considered a conservative assumption, based on statistical data, that where pyrolysis continues that such a fire would keep burning at a persistent intensity [16]. With sprinklers actively capping or controlling the energy released by a fire, it is reasonable to assume that fire severity is decreased. For the purposes of fire dynamic calculations and modelling for example, reductions in parameters such as flame temperature and emissivity would be considered appropriate and permissible for sprinkler-protected areas [17].

2.5.3. Reliability and Effectiveness

Sprinklers operated in 95% of fire scenarios, based on research conducted in the US by the NFPA in multiple building types, including residential, commercial, and industry properties [18]. The same research found that sprinklers were effective in controlling and suppressing the fire in 92% of cases.

The level of maintenance afforded to a sprinkler system as well as the robustness of the initially installed system strongly affect the effectiveness of the system. The variation in sprinkler effectiveness across various independent studies internationally, as well as the reasons for sprinkler system failure, is presented by Frank et al. [6] and indicates that the largest culprit for sprinkler failure is that the system was shut off (73%), followed by the system being inappropriate (14%) and lack of maintenance (10%). The variation in the definition of ‘sprinkler effectiveness’ across the key studies [15,16,17,18] reviewed is understood to influence the results, in particular for the upper and lower limits of the data contained within them. The study recommended a peak between 90% and 95% for the sprinkler effectiveness if using a probabilistic model, with upper and lower bounds estimated from the study.

Greater sample sizes appear to capture higher nominal effectiveness values with the highest showing 99.5% [19]. Australian data remains higher than most other international datasets for effectiveness.

2.5.4. Design Capacity

The intent of this investigation into the benefits of sprinkler protection was on suppressing apartment fires, as opposed to other fires in commercial and industry settings. This research is predominantly focused on what is accepted as a built environment (buildings) with a lower level of cladding risk, compared to the high-rise towers fully covered in highly combustible cladding from top to bottom that people generally think of when they think of, namely, buildings where the extent of external fire spread is confined in its potential maximum fire size. The data of building fires where the building was constructed with a limited amount of cladding is not readily available for Australia. This could be because such fire events have not occurred, or if they did occur, they were small (as anticipated) and not regarded as “news worthy”.

The value of sprinklers can nevertheless be seen through real life examples of fires with combustible cladding, including two high-rise fires that occurred in Victoria in the last decade. In the Lacrosse fire of November 2014, it was concluded from witness reports and investigations carried out after the fire that the installed sprinkler system was effective in reducing the severity of the fire and reducing the spread of fire internally. It is noteworthy that the installed system was designed for the combined operation of 4 sprinkler heads and 2 fire hydrants. However, during the fire, 26 sprinkler heads operated, across 16 levels in the building, which is well above the intended design capacity [20]. In this study, it is assumed that the sprinkler system operates in all the affected SOUs when it is working (in line with the assumption that the fire has spread inside all connected SOUs).

2.5.5. Sprinkler System Installed on Balconies

Despite the positive impact of sprinklers, as outlined above, externally located sprinkler heads do have other factors, such as wind effects, that could influence the operation and or effectiveness of such sprinkler heads. Because of this, a conservative approach was taken in that sprinkler protection located externally on the balcony is assumed to not have an effect or not be present. This is, of course, in reality, a conservative oversimplification.

2.6. Risk Assessment Method

Using datasets from the US, the death and injury rates per 1000 fires have been estimated for apartment buildings and further down to the SOU level.

Deaths per 1000 fires (subsequently referred to as death rate) have been further discretised into two cohorts, these being (A) one- and two- SOUs per building and (B) apartment buildings. The first (A) corresponds to detached and semi-detached houses and the second (B) to the buildings that are the focus of this analysis.

If $d_1$ and $d_2$ are the death rates and $n_1$ and $n_2$ are the number buildings/houses in cohort (A) and (B), respectively, then the death rate in the residential cohort is:

$$d={\displaystyle \frac{\sum _{i=1}^2{d}_i{n}_i}{\sum _{i=1}^2{n}_i}}$$

(1)

As a result, the death rate in each apartment per building is:

$$d_2={\displaystyle \frac{d\sum _{i=1}^2{n}_i-d_1{n}_1}{n_2}}$$

(2)

Assuming that it is not credible for two independent fires to originate from two independent ignition sources within in the same building at the same time, the death rate in apartment buildings per SOU is calculated as:

$$d_2^{SOU}={\displaystyle \frac{d_2{n}_2}{n_2^{SOU}}}$$

(3)

where $n_2^{SOU}$ is the number of households in cohort B.

Research by the NFPA (2012) on “Sprinkler Impact on Fire Injury” showed that sprinkler-protected buildings could reduce the cost of injury by 41%. This included medical cost, lost work time, and pain and suffering [14]. This adjustment was incorporated in the calculation, acknowledging a possible difference of these costs between the US and Australia.

The benchmarks for elevated risk have been normalised and represent the lower value of the discrete values within the cluster, i.e., the most conservative values. For cladding risk values in or to a cluster, the highest value, being the most conservative, was used. This being the risk to the topmost SOU in a vertical cluster.

If $\alpha$ and $\beta$ are the likelihood of a fire starting either “within a SOU” and “on the balcony/private courtyard area of an SOU”, respectively, then the benchmark for Low of individual SOU risk ($R_2^{SOU}$) and societal SOU risk ($R_2$) are:

$$R_2^{SOU}=(\alpha +\beta )d_2^{SOU}$$

(4)

$$R_2=\sum _{i=1}^N{R}_2^{SOU,i}$$

(5)

where $N$ is the number of SOUs in the cluster.

The above calculation reasonably assumes that there is no fire spread beyond the SOU of fire origin.

The cladding risk posed to a cluster of $N$ SOUs, in a sprinkler-protected building, that are connected by combustible cladding is calculated as:

$$R_{2,SPRK}^{SOU,i}=\left(\alpha +\beta \right)d_{2,SPRK}^O+\sum _{i=1}^{N-i}\left(\alpha +\beta \right){\delta }_u{d}_{2,SPRK}^B+\sum _{i+1}^N\left(\alpha +\beta \right){\delta }_d{d}_{2,SPRK}^B$$

(6)

$$R_{2,SPRK}^{Cluster}=\sum _{i=1}^N{R}_{2,SPRK}^{SOU,i}$$

(7)

where

• $R_{2,SPRK}^{SOU,i}$ is the individual SOU cladding fire risk in sprinkler-protected buildings of SOU $i$ within the cluster;
• $R_{2,SPRK}^{Cluster}$ is the societal SOU cladding fire risk in sprinkler-protected buildings of the cluster;
• $d_{2,SPRK}^O$ is the death rate which occurs in the SOU of fire origin in sprinkler-protected buildings;
• $d_{2,SPRK}^B$ is the death rate which occurs beyond the SOU of fire origin in sprinkler-protected buildings;
• ${\delta }_u$ and ${\delta }_d$ are the probability of a cladding fire to spread upwards and downwards between SOUs in the cluster. It has conservatively been assumed that a cladding fire always spreads upwards (${\delta }_u=1$) and that is highly likely to spread downwards (${\delta }_d=0.9$).

The relative cladding fire risk in a sprinkler-protected building can be compared to the “Low” benchmark above as:

$$I_{2,SPRK}^{individual}={\displaystyle \frac{{\mathrm{m}\mathrm{a}\mathrm{x} (R}_{2,SPRK}^{SOU,i})}{R_2^{SOU}}}withi=\overline{1-N}$$

(8)

$$I_{2,SPRK}^{societal}={\displaystyle \frac{R_{2,SPRK}^{Cluster}}{R_2}}$$

(9)

The individual SOU risk ($R_{2,E}^{SOU}$) and societal SOU risk ($R_{2,E}^{Cluster}$), when comparing to the “Elevated” benchmark noted earlier, can then be calculated as:

$$R_{2,E}^{SOU}=\left(\alpha +\beta \right)d_2^{SOU}+\left(\alpha +\beta \right){\delta }_u{d}_2^{SOU}$$

(10)

$$R_{2,E}^{Cluster}=\sum _{i=1}^N{R}_{2,E}^{SOU,i}$$

(11)

noting again that the individual SOU risk uses the value of the uppermost SOU in a cluster, which results in the most conservative value, and $N$ is two for the benchmark calculation.

The relative cladding fire (RCF) risk in a sprinkler-protected building compared to the “Elevated” benchmark is:

$$I_{2,SPRK,E}^{individual}={\displaystyle \frac{{\mathrm{m}\mathrm{a}\mathrm{x} (R}_{2,SPRK}^{SOU,i})}{R_{2,E}^{SOU}}}$$

(12)

$$I_{2,SPRK,E}^{societal}={\displaystyle \frac{R_{2,SPRK}^{Cluster}}{R_{2,E}^{Cluster}}}$$

(13)

3. Results

The calculated results using Equations (1)–(3) are presented in

Table 3. Deaths and injuries per 1000 fires in apartments in non-sprinklered buildings.

	One-Two-SOUs (per Building)	All Residential	Apartment (per Building)	Apartment (per SOU)
Deaths per 1000 fires	7.90	8.30	11.20	3.36
Injuries per 1000 fires	25.30	33.65	94.50	28.36
Cost per 1000 fires	$22,030	$23,481	$34,055	-
No. of building	$6.68\times {10}^7$	$7.60\times {10}^7$	$9.17\times {10}^6$	-

while the RCF risk of sprinkler-protected buildings when compared to the “Low” and “Elevated” benchmarks have been set out in

Table 4. Relative cladding fire risk in sprinkler-protected buildings compared to “Low” and “Elevated” benchmark.

	Compared to “Low” Benchmark		Compared to “Elevated” Benchmark
	$I_{2,SPRK}^{individual}$	$I_{2,SPRK}^{societal}$	$I_{2,SPRK,E}^{individual}$	$I_{2,SPRK,E}^{societal}$
	IF-SCAN = 3		IF-SCAN = 3
Deaths	32.5%	13.4%	17.1%	13.4%
Injuries	53.9%	49.5%	28.4%	21.8%
Cost	-	55.2%		25.4%
	IF-SCAN = 7		IF-SCAN = 20
Deaths	44.9%	31.4%	44.6%	26.9%
Injuries	77.9%	51.7%	82.0%	44.3%
Cost	-	91.9%	-	97.0%
	IF-SCAN = 10		IF-SCAN = 25
Deaths	54.1%	31.7%	52.7%	27.8%
Injuries	95.8%	52.2%	97.7%	45.9%
Cost	-	>100%	-	>100%
	IF-SCAN = 24		IF-SCAN = 54
Deaths	97.1%	32.0%	99.6%	32.2%
Injuries	>100%	>100%	>100%	49.6%
Cost	-	>100%	-	>100%

.

For injuries, by comparing an IF-SCAN of 3 to the “Low” benchmark, an RCF risk of ~54% is obtained. The highest IF-SCAN value which results in a RCF risk that is less than 100% is 10. When comparing the death rate, and IF-SCAN of 24 provides an equivalent death rate.

For the “Elevated” benchmark, an IF-SCAN of 25 corresponds to an equivalent injury rate, while an IF-SCAN of 54 gives an equivalent death rate.

The loss (cost) per fire was derived from both sprinkler-protected and non-sprinkler-protected buildings. Relying on the above-mentioned datasets and assumptions, it was found that IF-SCAN values of 7 and 20 result in an RCF of ~92% and 97% for the “Low” and “Elevated” benchmarks, respectively.

The above includes for the reliability of the sprinkler system, fires starting both inside sole occupancy units (SOUs) and outside (balcony), damage to the building (external damage due to combustible cladding), and the fire damage when an external fire spreads to the other SOUs in the cluster.

4. Discussion

4.1. Smoke Alarms Detectors and Occupant Warning Systems

In order to provide occupants of a building with early notification, the activation of a sprinkler system should always result in a building-wide fire alarm.

Smoke from any fire in a building could potentially spread to a bedroom in an adjoining sole occupancy unit (SOU). This is true irrespective of any combustible cladding being present externally. Such smoke spread could occur both before and after any activation of the sprinkler system has occurred.

The provision of detecting smoke and providing local alarm inside bedrooms will therefore further assists in mitigating risk posed by smoke from any fire in the building. This is because the bedrooms house potentially sleeping occupants who would be unaware of any smoke. A local smoke alarm can alert such sleeping occupants, potentially prior to a building-wide alarm sounding [21]. It is, therefore, recommended to consider a local alarm inside the bedrooms to add value to early warning for occupants complimentary to the sprinkler system’s involvement in generating the building-wide fire alarm, as discussed in Section 2.5.1.

4.2. Cladding Fire Risk from Other External Ignition Sources

The analysis above considers the cladding fire risk where fire is assumed to start from an SOU (either indoor or outdoor, i.e., balconies and private courtyard). Other external ignition sources such as that of a burning vehicle parked close to combustible cladding was not be specifically addressed in this document. External fires which could feasibly ignite combustible cladding that is either part of a cladding cluster or would result in a single exit/evacuation pathway being compromised, should generally be dealt with on an “elimination” basis. As a result, the residual cladding fire risk remains with the ignition sources considered in the analysis.

4.3. Assessing the Above the Results

Data for deaths per 1000 fires show a clear case for the betterment to the life safety of people within buildings with a sprinkler system is installed. The calculations show that up to 24 sprinkler-protected SOUs connected by combustible cladding equate to the equivalent risk of a single SOU without a sprinkler system and without combustible cladding installed (the “Low” benchmark). When comparing to the “Elevated” benchmark, up to 54 sprinkler-protected SOUs can be connected by combustible cladding.

When comparing injury rates, up to 10 sprinkler-protected SOUs provides a calculated risk that is similar to the “Low” benchmark, and up to 25 sprinkler-protected SOUs for the “Elevated” benchmark.

For property damage, up to seven connected SOUs results in a calculated risk that is just below the “Low” benchmark.

The figure and tables above are calculated values of relative risk based on the likelihood of the fire sources considered and the consequences relating to human safety (deaths and injuries). For certain sprinkler-protected buildings, other consideration will also need to weigh in on the decision for retaining combustible cladding, such as (but not limited to) the buildings’ height and firefighting capabilities. The data uses reasonably comparable database from the US, which may also lead to some variation when compared to Australia; note that the data and statistics reviewed indicate that the IF-SCAN values presented above would increase due to better reported data from Australia.

It should also be noted that all apartments are assumed to have balconies, which increases the resultant risk outcomes. If the building does not have balconies (i.e., $\beta =0$), the calculations would result in a higher number of apartments being connected via combustible cladding being equivalent to the respective benchmarks, e.g., an IF-SCAN cluster of 26 is at 98.6% of the comparable “Low” benchmark for cost. In this case, the comparative risk corresponding for injuries (IF-SCAN of 10) determines the equivalence to the “Low” benchmark.

4.4. Type of Sprinkler System

This work was conducted on the basis of the subject buildings having sprinkler systems installed in accordance with AS 2118.1, AS 2118.4, or AS 2118.6, and that activation of the sprinkler system will initiate a building-wide fire alarm.

Buildings which have FPAA 101D and FPAA 101H automatic fire sprinkler systems installed are also seen to provide a significant level of suppression, including lowering the injury and fatality rates, and could be considered as one of the options for interventions to mitigate cladding fire risks for existing buildings. These systems are found to currently have lower market penetration in the context of the State of Victoria, compared to the systems mentioned in the previous paragraph.

5. Conclusions

In this paper, a risk-based approach was employed to quantitatively assess the effectiveness of sprinkler systems in residential buildings, specifically in mitigating the risk premium associated with combustible cladding. The comparative methodology involved reviewing statistical data on fire incidents, sprinkler efficiency, and fire-related injury and death rates, while considering the increased likelihood of occupant exposure to fire due to sole occupancy units (SOUs) being interconnected by combustible cladding and associated consequence to physical safety.

Fire death and injury rates for apartment buildings, and subsequently SOUs, were extrapolated from data published in the United States. Two benchmarks were established: a “Low” benchmark representing a single SOU without cladding or sprinklers, and an “Elevated” benchmark representing two SOUs with cladding but no sprinklers. A mathematical model was developed to account for fire spread via cladding to interconnected SOUs, as well as the effectiveness of sprinkler systems within those SOUs.

Additionally, a cost analysis was performed, comparing fire-related costs in clusters of SOUs with combustible cladding and sprinkler protection against a single SOU without cladding or a sprinkler system for the “Low” benchmark.

Key findings of this comparative study include:

• Extensive cladding removal is not necessary for sprinkler-protected buildings with an IF-SCAN value of 7 or lower. The risk to individuals in these buildings, both per apartment and across the entire cluster, is lower than that of the reference “Low” benchmark building.
• If the SOUs within a cluster lack balconies, the number of connected SOUs (as represented by IF-SCAN values) increases significantly.
• In cases where the cluster is situated at a significant height or in locations where firefighters may have limited access, additional measures may be required. These could include partial cladding removal, extending sprinkler protection to balconies, or other mitigating actions.

The findings of this study provide scientific inputs into further policy decisions in Cladding Safety Victoria’s Protocols for Mitigating Cladding Risk [4]. Further details of this development are publicly available on the organisation’s website. This study adopts conservative assumptions regarding cladding involvement in SOU fires, specifically applying to highly flammable cladding types such as aluminium composite panels and rendered expanded polystyrene. Some cladding materials, such as aluminium composite panels with flame-retardant cores, may exhibit less severe burning behaviour. However, this study does not provide a detailed assessment of the associated risk reduction for such materials.