Reliability of Automatic Sprinkler Systems

William E. Koffel, P.E.

Revised September 2005

            Whether one is preparing a performance design or working with a prescriptive code, the reliability of fire protection systems and features must be considered.  Budnick explains that reliability includes both operational reliability and performance reliability.  The operational reliability is a measure of the probability that a system or component will operate as intended when needed.  The performance reliability is a measure of the adequacy of the system once it has operated.  While critical for all fire protection features and systems, this paper will focus on the reliability of automatic sprinkler systems, in particular the operational reliability.

            When the original paper on this subject was prepared by this same author, critics immediately claimed that the data was manipulated and the operational reliability of sprinkler systems was being represented as being too low.  However, many of the critics failed to consider the aspects of uncertainty addressed in the paper.  Since that time, NFPA has released two additional reports, the latter of which specifically confirms that the operational reliability of sprinkler systems, as reported in the original paper, accurately represented the data upon which the paper was based.  The recent NFPA reports utilize more current data which cannot be combined with the original data due to differences in the reporting system.  The more recent NFPA reports are included in this revised paper.

Past Studies

Table 1 provides a list of previous studies in which the reliability of automatic sprinkler systems has been documented.  The scope, breadth, and reporting periods of the various studies vary significantly.  One must also carefully review the scope of each study. 

Table 1

Operational Reliability

            Table 1 includes both domestic and international estimates regarding the reliability of sprinklers.  Many of the studies include limited data bases and are based upon experience over 15 years ago.  A review of more recent fire experience in the United States indicates that the reliability of automatic sprinkler systems, while still good, may not be as high as reported by several of the studies in Table 1.  In an NFPA report , Rohr provides considerable data regarding the fire experience in the United States in buildings protected with automatic sprinklers.

            The NFPA data over a ten year reporting period regarding the operational reliability of automatic sprinkler systems can be summarized as indicated in Table 2.

Table 2

NFPA provided an update on the original report using both the original data reported in Table 2 and data for a period of one year (1999).  Due to differences in the reporting system, the two data sets should not be combined.  Table 3 summarizes the data as reported by NFPA using 1999 data.

Table 3

Although the 1999 data would indicate that the operational reliability of automatic sprinkler systems has decreased slightly from the previous ten year data base, the decrease may not be statistically significant since the data base is substantially smaller.

As with any data collection system, there are some limitations regarding the accuracy of the data.  While identified as a limitation in some of the studies reported in Table 1, it should be noted that the Estimated Number of Fires with Sprinklers Present in Tables 2 and 3 do not include fires which were too small to operate a sprinkler.  For example, if the incident report indicated that the fire was too small to operate a sprinkler, that data point is not included in Tables 2 and 3.

The data in Tables 2 and 3 do not include fires that are not reported to fire departments.  The data does not discern whether the systems have been properly designed, installed, and maintained which would obviously increase the operational reliability of automatic sprinkler systems.  Also not included is the type of sprinkler system provided and as such, it is not clear that sprinklers were present in the area of origin for all the reported fires.  For example, it is possible that sprinklers were present in the building and the incident report may indicate the presence of sprinklers.  However, the area of origin may not be in an area where sprinklers were present and there is no way to discern this from the data.  While not the same data base, a separate NFPA report7 indicated that partial sprinklers where the fire originated in an area that was not sprinklered constitute 7.8% of the sprinkler system failures.

            In the August 2005 report , NFPA utilizes information available in the new data system to better document the fires that occur within an area where sprinklers are not present.  The adjusted data in the August 2005 report deletes all data in which sprinklers were reported as not being present in the area of fire origin from the data base if sprinklers did not operate and if sprinklers operated but were not effective.  The information contained in the report does not allow one to determine if this may result in overestimating sprinkler system reliability.  For example, if a fire occurs in an area in which sprinklers are not present and the reference standard does not require sprinklers to be present, the incident may be eliminated from the analysis based upon the entry that sprinklers were not in the area of fire origin.  This is different than the issue where the only selected areas of a building are protected and the fire occurs in a space that was not intended to be protected by automatic sprinklers.

            Unfortunately the August 2005 NFPA report does not provide the same level of data as provided in previous reports.  Instead, the report merely provides percentage values for the time period 1999-2002.  Therefore, Table 4 does not contain the number of incidents as provided in the previous tables.  The first column of percentages in Table 4, labeled “Nonadjusted,” is provided for comparison with Tables 2 and 3.  The second column of percentages in Table 4, labeled “Adjusted,” provides the data as “corrected” by NFPA.  Where data is not provided in Table 4, the information is not provided in the August 2005 report but was provided in one of the previous reports.

Table 4

            Again, the operational reliability of automatic sprinkler systems as reported by the non-adjusted data is lower than what was reported in the original paper by this author.

Performance Reliability

            Performance reliability is not easily determined using NFPA fire data.  Some of the studies cited in Table 1 use the number of sprinklers operating as a means of evaluating performance reliability.  In a performance-based design, the ultimate evaluation may be whether the outcome is consistent with the expected performance as documented during the design process.

            It is understood that most automatic sprinkler systems are designed to control a fire but not necessarily to completely extinguish the fire.  The NFPA fire data supports the concept that sprinkler systems can control fires but do not necessarily result in complete extinguishment.  Table 5 indicates the percentage of fires where sprinklers are present and that are reported as being extinguished by an automatic suppression system.  Note that the data includes the fires reported to be extinguished by all types of automatic suppression systems and not only those extinguished by automatic sprinkler systems. However, since automatic extinguishing systems other than sprinkler systems constitute only a tiny fraction of protected areas, it is reasonable to assume that the overall automatic extinguishing system data can be interpreted as a relatively accurate indication of sprinkler system data.

The data in Table 5 has not been updated to include the periods from 1999 through 2002.  Instead, the August 2005 report indicates that when sprinkler systems operate they are effective in 96% of the incidents.  Assuming the validity of the data entry used to generate this value, the August 2005 report would be a better means to measure performance reliability than the data in Table 5.

Table 5

            While property loss and life loss are greatly reduced in buildings protected with an automatic sprinkler system, the sprinkler system alone is not providing the entire increased protection.

Summary

While NFPA fire data clearly demonstrates that property loss and life loss are reduced in buildings protected throughout with an automatic sprinkler system, the same data has indicated in the past that sprinklers fail to operate 1 in every 6 fires that are large enough to activate a sprinkler.  More recent data studies indicate that the operational reliability of automatic sprinkler systems may be decreasing.  However, improvements in the data collection system enable a better evaluation of the data and based upon the August 2005 NFPA report, the operational reliability of sprinkler systems may be as high as 93%.
It has been stated that unreported fires may increase the reliability of automatic sprinkler systems.  However, no data has been presented to support that claim.  It is common in the U.S. that current building and fire codes require the water flow alarm from an automatic sprinkler system to automatically transmit an alarm to an alarm receiving facility.  This should have the effect of increasing the percentage of fires reported to fire departments in buildings protected with an automatic sprinkler system.

The original paper indicated that the uncertainty in the data could result in an operational reliability of sprinkler systems in the area of 90%.  In subsequent presentations regarding the paper, this is the value that the author has used.  This is the same value that is proposed to be used for sprinkler system reliability for life safety purposes in a British Standard.   The same British standard proposes a value of 80% for automatic sprinkler system reliability when considering property protection.

The NFPA data indicates that the commonly stated reliability of automatic sprinkler systems in the range of 96% (fails once in every 25 fires) is overstating the reliability of sprinkler systems unless there are assurances that the preventive maintenance on the system is substantially better than that on the average system in a building in which a fire has occurred.  When combining the operational effectiveness and performance effectiveness data as published in the August 2005 NFPA report, the overall reliability of automatic sprinkler systems is 91%.  This value is extremely close to the 90% value previously proposed by this author and the value proposed by the British Standard.

The paper was commissioned by the Alliance for Fire and Smoke Containment and Control, Inc.

William E. Koffel, P.E., is President of Koffel Associates, Inc., a fire protection engineering and code consulting firm with offices in Maryland and Connecticut.  Mr. Koffel has a B.S. in Fire Protection Engineering and he has over 26 years of experience as a practicing fire protection engineer.  Mr. Koffel participates actively in the model code development processes of the International Code Council and the National Fire Protection Association and has served on numerous committees within each process.
He has previous experience with the Maryland State Fire Marshal’s Office and has been a volunteer firefighter.