Improving sprinkler system reliability
By Samuel S. Dannaway, PE,
President, S.S. Dannaway Associates, Inc., Honolulu
In my January article, I noted that sprinkler system reliability is an important issue for performance-based fire safety engineering. Of course, it is important in any design. However, in prescriptive design, sprinkler reliability is implied or assumed to be reliable, while, in a probabilistic fire risk analysis, the engineer must be able to quantify the reliability.
There are two components to total fire protection or sprinkler system reliability, operational reliability and performance reliability1. Operational reliability is the probability that the sprinkler system will operate as intended when it is needed. Performance reliability is the probability that the sprinkler system, once operated, effectively does the job by controlling or suppressing the fire. Total sprinkler system reliability is equal to operational reliability and performance reliability (i.e. total system reliability = operational 5 performance).
There is a very comprehensive U.S. sprinkler reliability analysis based on National Fire Incident Reporting System (NFIRS) data in the 2006 NFPA report by Dr. John Hall2. Improvements regarding data collection for sprinkler systems in the 1999 edition of NFIRS has made analysis of both operational and performance reliability possible. From the report it is concluded that operational reliability is 93% and performance reliability is 96%. The overall sprinkler system reliability is thus 89% (0.93 5 0.96). In a review of several studies of sprinkler reliability Koffel3 cites figures ranging from 81.3% to 99.5%.
Some in the fire protection industry decry the “mere 89%” overall reliability as a reason to limit sprinkler trade-offs in codes in favor of more passive protection, i.e. more compartmentation, more concrete, gypsum board, fire doors, etc). In my view, an 89% overall effectiveness is pretty darn good. The passive fire protection industry would do well to provide reliability data for their products (If it exists, I could not find it).
Though 89% reliability is good, there is substantial room for improvement. In looking at operational reliability, Dr. Hall reports the following reasons that sprinkler systems failed to operate:
Cause Percentage to Total Failures
System Shut Off 65%
Manual System Intervention
Defeated System 16%
Lack of Maintenance 11%
Wrong Type of System 5%
System Component Damaged 3%
The obvious thing designers can do to minimize the chance of a closed valve is to provide proper supervision for all valves controlling water supplies to sprinkler systems. The best method is to electrically supervise the condition of the valve with a tamper switch and to have the signal report to a constantly attended location.
One way to improve operational reliability by reducing failures due to lack of maintenance is to keep the design simple. For example, do not use preaction sprinklers where wet pipe sprinklers would be acceptable. Wet pipe sprinklers do not rely on a properly functioning detection subsystem for operation. There are also many failures related to dry pipe systems, particularly in storage occupancies. Again, where conditions permit, wet pipe systems should be used.
The Hall report cites the following reasons for ineffective sprinkler performance:
Cause Percentage to Total Failures
Agent Did Not Reach Fire 50%
Not Enough Agent Released 32%
Inappropriate System for Type of Fire 11%
System Component Damaged 4%
Manual Intervention Defeated System 3%
From this information, it is apparent that obstructions to sprinkler discharge are a major problem and that performance will be improved by ensuring that systems are designed and installed to avoid obstructions. In addition to avoiding obstructions, ensuring that 100% sprinkler coverage is provided for the entire building will serve to improve system performance.
The issue with “not enough agent released” points to two factors under the control of the designer. Assuring an adequate water supply to sprinkler systems as part of the design and providing sprinkler design criteria suitable to the hazard will improve performance reliability.
I suppose that I can be accused of stating the obvious in regard to improving sprinkler system reliability through design. But if it is so obvious, why are we having any failures at all?
On a side note, though sprinkler reliability is not specifically addressed in the analysis of residential sprinkler performance in Prince George’s County, Maryland4, the report contains several interesting facts. PG County has required sprinkler systems in all new townhouses and single family residences since 1992. From 1992 to 2007, there were 101 fire deaths in non-sprinklered townhouses and single family dwellings, while there was not a single fire death in sprinklered townhouses and single family homes. Granted that only approximately 10% of these residences in PG County are sprinklered, it is still an impressive testament to the reliability of sprinkler systems.
1. Richard W. Bukowski, P.E., Estimates of the Operational Reliability of Fire Protection Systems, NIST, 2002.
2. John R. Hall Jr., Ph.D., An Analysis of Automatic Sprinkler System Reliability Using Current Data, National Fire Protection Association, February 2, 2006.
3. William E. Koffel, P.E., Reliability of Automatic Sprinkler System, Revised September 2005.
4. Steve Weatherby, Benefits of Residential Fire Sprinklers: Prince George’s County 15-Year History with its Single-Family Residential Dwelling Fire Sprinkler Ordinance, Home Fire Sprinkler Coalition, August 2009.
Samuel S. Dannaway, PE, is a registered fire protection engineer and mechanical engineer and past president and a Fellow of the Society of Fire Protection Engineers. He is president of S. S. Dannaway Associates, Inc. He can be reached via email at SDannaway@ssdafire.com.