54 NFPA JOURNAL MARCH/APRIL 2014 Photograph: Shutterstock

A new study puts a dollar amount on the damage
caused by sprinklered fires versus non-sprinklered
fires—and in industrial occupancies, those dollar
differences can get very large
Do Sprinklers Really Make a
Diference? Yes, They Do!
Fred Sanford and Ralph Tiede, Liberty
Mutual Insurance
Monday, June 9, 9:30-10:30 a.m.
For updated information on education
sessions, visit nfpa.org/conference.
t’s been long recognized by property insurers and
firefighting professionals that sprinklered properties
generally suffer less damage in fires than non-sprin-
klered properties. Nowhere is that more true than in
industrial occupancies; fires in these settings tend to be
disproportionately more damaging than fires in other types
of occupancies. It can be argued that without the automatic
sprinkler, industry in the United States would not have
developed as rapidly and to the extent that it has. From the
early models of sprinklers in the mid-19th century, which
kept the wooden floors of New England mills wet by spray-
ing water in every direction, to the standard spray sprin-
klers of the 1950s, which cooled light steel construction by
setting up an umbrella pattern of finer droplets, sprinklers
have fulfilled their promise. Nowadays, in response to
ever-higher storages and modern exposures, early suppres-
sion fast-response (ESFR) sprinklers and other specialized
types of suppression are able to drive larger droplets down
through the fire plume to control and suppress fires in their
incipient stages.
A 14-year study of hundreds of fires by Liberty Mutual
Insurance’s Property Risk Engineering Group underscores
the importance of fire sprinklers in minimizing property
damage, especially in industrial occupancies. The study
damage involving sprinklered and non-
sprinklered manufacturing occupancies
was significant.
These categories bear a somewhat
closer look, however. For rubber/plastics,
the estimated fire damage in non-sprin-
klered occupancies averaged $788,000,
compared to $766,000 in sprinklered
occupancies. We do not believe this is
an aberration, but rather a function of
a fire exposure unique to the industry.
Plastic molding machines combine the
extensive use of high-pressure hydraulic
oils with ready ignition sources. When
high-pressure combustible hydraulic
oil sprays from a leak in the system,
it frequently finds an ignition source
and results in a fireball type of fire that
may be one of the most severe expo-
sures encountered by sprinklers in the
manufacturing classes—it’s no wonder
that large numbers of sprinklers operate
almost instantly in these events. Even
under the best of circumstances, damage
at the ceiling and to adjacent equipment
can be severe. Ten to 20 sprinklers can be
expected to operate in a typical hydraulic
oil fire, whereas fewer than five is typical
for industrial fires overall. The average
estimated fire damage of a sprinklered
hydraulic oil fire in the plastics industry
has been about $1.2 million. Exclu-
sive of hydraulic oil fires, the average
sprinklered fire damage has been about
This illustrates a characteristic of
industrial fires overall: Whereas dam-
age resulting from non-sprinklered
fires tends to be merely a function of
the value of the property, as well as the
combustibility of the construction and
occupancy, the damage from sprin-
klered fires is driven by the intensity
of the hazard in the area of origin.
The most damaging sprinklered fire,
at slightly greater than $5 million, in-
volved combustible oil where a carbon
dioxide extinguishing system operated
first but did not fully extinguish the fire.
In fact, it was a single large hydrau-
lic oil fire in a sprinklered facility that

looks at 946 serious fires that occurred
from 2000 to 2013 across 11 different
occupancy types, with threshold dol-
lar damage estimates of $100,000 and
greater. The total estimated damage
of those fires was $1.2 billion. Of the
fires studied, 322 were industrial fires:
269 in manufacturing occupancies and
53 in warehouse occupancies. While
those fires made up just over a third of
all the fires included in the study, they
were responsible for 49 percent of the
total estimated damage.
The findings supported the notion
that the “heavier” the occupancy, such
as industrial, the proportionately more
damaging fires tend to be. In this study,
for example, while residential fires ac-
counted for 8 percent of the number of
serious fires, they accounted for just 4
percent of the total estimated damage of
those fires. Even within the manufactur-
ing occupancies themselves, such as a
foundry compared to a watch manufac-
turer, the heavier end tends to produce
the more damaging and costly fires.
Overall, in a comparison of estimated
fire damage in non-sprinklered versus
sprinklered industrial occupancies, fires
in non-sprinklered facilities averaged
an estimated $1.9 million in damage
compared to $638,000 in sprinklered
facilities, or a ratio of about 3:1.
While we are unable to share the ac-
tual report with the public due to issues
of confidentiality, we felt an overview
here would lend valuable support in
the cause of automatic sprinklers in
industrial occupancies. The story told
by the study isn’t that non-sprinklered
fires get big—we knew that already. The
critical point is that when sprinklers
are installed in the area of origin and
operate as designed, fires almost always
remain small. Now, we have the dollars-
and-cents comparisons to prove it.
Running the numbers
We segmented our data in four ways:
fires where sprinklers were installed in
the area of origin and operated as de-
signed; fires where sprinklers were not
installed, but clearly could have been
and would have operated; fires where
sprinklers were not, and probably
would not have been, a factor; and fires
where sprinklers were impaired or oth-
erwise adversely affected by physical
conditions, as in instances where fire
gained headway in a non-sprinklered
concealed space or where water sup-
plies were clearly deficient.
We also compiled comparative data
for 11 different categories of manufac-
turing operations: heavy metal classes,
food, rubber/plastics, paper/printing,
textile/apparel, stone/
glass, wood/furniture,
buildings only, light
metal classes, and ag-
ricultural. Again, fires
at the heavier end of
the metal classes were disproportionate-
ly more damaging on average; the heavy
metal classes totaled 43 percent of the
269 manufacturing fires, but accounted
for 60 percent of the $502 million of
estimated damages caused by those fires.
Again, although not universal, the
tendency was for a greater differential in
estimated damage between non-sprin-
klered and sprinklered fires among the
heavier occupancies. In the heavy metal
classes, the average non-sprinklered
fire damage was nearly $3.1 million,
compared to $819,000 for sprinklered
occupancies, or a ratio approaching 4:1.
Similarly, for textile/apparel, the average
non-sprinklered fire damage was slight-
ly more than $1.9 million, compared to
$461,000 for sprinklered occupancies.
In all categories except for two—rub-
ber/plastics and wood/furniture—the
difference between the average fire
THE STORY TOLD by the study is that when
sprinklers are installed in the area of origin
and operate as designed, fires almost always
remain small.
produced by NFPA’s Fire Analysis and Data Division, includes
information on manufacturing and warehouse facilities, the
types of occupancies that are termed “industrial” in the Lib-
erty Mutual study. The data, taken from a survey by NFPA and
the National Fire Incident Reporting System and excerpted
here, supports the findings of the Liberty
Mutual study: that sprinklers in these oc-
cupancies are overwhelmingly successful
at reducing the loss of life and property in
the event of fire.
In the NFPA report, for manufactur-
ing facilities (excluding buildings under
construction) in 2007–2011, 48 percent of
reported structure fires indicated some type of sprinkler was
present: 85 percent were wet pipe, 12 percent were dry pipe,
and 3 percent were other.
Wet pipe sprinklers operated in 91 percent of the fires and
operated efectively in 86 percent of the fires. When failure
occurred, leading reasons were system shutof (62 percent)
and manual intervention defeated system (20 percent). When
operating equipment was inefective, leading reasons were
water did not reach fire (36 percent) and not enough water
released (31 percent).
Only one or two sprinklers
operated in 67 percent of reported
fires where wet pipe sprinklers
In manufacturing facilities,
deaths per thousand reported
fires were 88 percent lower when
wet pipe sprinklers were present,
when compared to fires with au-
tomatic extinguishing equipment
In manufacturing facilities, di-
rect property damage per reported
fire was 38 percent lower when
wet pipe sprinklers were present,
when compared to fires with no
automatic extinguishing equip-
ment present.
For warehouses (excluding cold
storage) in 2007–2011,
32 percent of structure fires reported that some type of sprin-
kler was present: 79 percent wet pipe, 20 percent dry pipe,
and 1 percent other.
Wet pipe sprinklers operated in 86 percent of fires and
operated efectively in 84 percent of fires. Only one or two
sprinklers operated in 73 percent of reported fires where wet
pipe sprinklers operated.
In warehouses excluding cold storage,
deaths per thousand reported fires were
61 percent lower when wet pipe sprinklers
were present, when compared to fires with
no automatic extinguishing equipment
In both manufacturing and warehouse
facilities, estimates of reliability and efectiveness are based
only on fires and installations where the fire should have acti-
vated and been controlled by an operational sprinkler system,
therefore excluding buildings under construction, fires with
sprinklers not in the fire area reported as the reason for failure
or inefectiveness, fires reported as being too small to activate
equipment, and fires reported as being confined to cooking ves-
sel, chimney or flue, fuel burner or boiler, commercial compac-
tor, incinerator, or trash.
Manufacturing and warehouse sprinkler data from NFPA’s
2013 “U.S. Experience with Sprinklers” report
To see the complete “U.S.
Experience with Sprinklers”
report, visit nfpa.org/research.
Photograph: AP/Wide World
A firefighter pulls a hose past a pile of burning molded fiber apple and egg crates at a manufactur-
ing facility in Washington state. NFPA data found that, in manufacturing facilities where sprinklers
were present, 85 percent were wet pipe, which operated efectively in 86 percent of the fires.
THE LIBERTY MUTUAL STUDY documented fires in indus-
trial occupancies where sprinklers were present and should
have operated, or perhaps even did operate, but were shut
of either before the fire event or prematurely during the fire.
These actions, known as impairments, are almost always the
result of human error.
In one instance, an impairment occurred at a warehouse
where, ironically, sprinkler contractors were making improve-
ments to the system. At closing time, they lacked a necessary
part to recharge the system. Instead of getting the part and
completing the recharging process, they left the system of
without notifying anyone and went home. A fire occurred that
night. What may have been damage around $250,000 to
$500,000 amounted to many multiples of that.
Another example involved a fire in a typical New England
textile mill. The fire seemed to exceed what would normally be
expected in a sprinklered building. After the fire, a fire protection
engineer visited the location and found several sprinkler control
valves in the closed position, including the one controlling
sprinklers in the fire area. It turns out they were shut of during
the fire to conserve water escaping through some damaged pipe.
Although the consequences of this action were not fully investi-
gated, it has to be at least assumed that sprinklers in the fire area
were shut of during the fire as well. Again, what may have been
fire damage of around $300,000 was many multiples of that.
Finally, one of the most unusual and well-documented
cases of sprinkler impairment occurred in Roseville, Califor-
nia, in 2010. Although it was in a retail location, this scenario
could just as easily have happened in a large, complex indus-
trial facility. The details were taken from the ofcial City of
Roseville loss report.
As the 1.4 million-square-foot Westfield Galleria at Roseville
Mall, an important local employer, geared up for holiday shop-
ping, a person believed to be carrying a gun barricaded himself
in the storage room of a video game store on the second floor
and started a fire. A standof ensued, the mall was evacuated,
and a suspect was apprehended. Emergency responders deter-
mined that sprinklers in that area of the mall were operating.
But the fire continued to grow in intensity and spread to adja-
cent stores until a partial roof collapse led to the withdrawal
of all emergency personnel. The fire entered a phase of drastic
escalation and fighting it became an external efort.
In the aftermath, the central question was how the fire could
have grown so large when sprinklers should have contained it.
An investigation revealed that, unbeknownst to authorities at
the time, the sprinklers had actually been shut of shortly after
responders arrived at the scene. They did not discover this for
well over an hour, when fire department personnel overheard a
Westfield maintenance employee talking about having been told
by police to shut of one of the sprinkler valves. Upon hearing
this, the fire department ordered him to reopen the valve, which
was done at about 71 minutes after it had been shut—plenty of
time for the fire to intensify and spread. In the final investiga-
tion it appeared that a delivery service driver relayed a “shut the
valve” message from police to the maintenance employee.
The fire was extinguished by the next day, but the Bureau of
Alcohol, Tobacco and Firearms (ATF) did not allow the mall to
reopen until days later. According to ATF, the fire destroyed 20
stores and resulted in an estimated $55 million in damage. The
four anchor stores had been able to shut themselves of from
the mall, which somewhat limited their damages. The man who
started the fire pled guilty to two counts of arson and was sen-
tenced to 15 years in federal prison.

A former mill building used for storage burned out of control in
Massachusetts in 2010. A local fire ofcial was alleged to have ordered
the building’s sprinkler system turned of weeks before the fire.
The Impairment Problem
What can happen when sprinklers meet human error
as impairments, sprinklers were present
and should have operated, or perhaps
even did operate, but were shut off either
before the fire event or prematurely dur-
ing the fire. In almost every instance, the
result of shutting off sprinklers before
the fire is fully controlled is roughly the
same as if sprinklers had not been pres-
ent. It can also be highly embarrassing
for those responsible, which is why it’s
often difficult to establish exactly how
the impairment occurred. (For examples,
see “The Impairment Problem,” page 58.)
These are the exceptions, however.
The study illustrates clearly that when
sprinklers are installed in the area
of origin, and when they operate as
designed—which occurs in the vast
majority of instances—fires almost
always remain small, as reflected by
the cost of the damage associated with
those fires. This is a key reason why
the value of automatic sprinklers for all
occupancies, from residential to heavy
industrial, continues to be more widely
recognized, not just in the United
States but around the world.
It is our hope that that momentum
will continue, aided by the work of
NFPA, major insurers, firefighting
organizations, and other stakeholders
to constantly investigate improvements
in hardware and to upgrade sprinkler
installation codes and standards. That
collective effort will ensure that
automatic sprinklers continue to fulfill
their promise of reduced loss of life and
property damage well into the 21st
century and beyond.
FRED SANFORD has been a P.E. for more than
55 years, and is director of special projects
with Liberty Mutual Insurance’s Property Risk
Engineering Group.
with great effectiveness in spite of
design deficiencies.
However, they do not compensate
well for failures in the post-design
human activities intended to keep sprin-
klers in service. Those activities include
regular inspections
to make sure valves
are open, using an
“impairment” system
that assures sprinkler
systems are restored
promptly after
repairs, and, perhaps
most important, posting a person dur-
ing emergency situations to make sure
sprinkler valves stay open until ordered
shut by an authorized person.
An example of a design/review fail-
ure is where sprinklers are omitted in
concealed spaces, in enclosed process
equipment such as ovens and printers,
from beneath mezzanines, or from
other enclosures. In these cases, fire is
able to gain headway before encounter-
ing sprinklers. On average, fires in this
category tend to be as damaging as if
there were no sprinklers. However, the
balance between fire and water damage
leans toward less fire and more water
damage when sprinklers are delayed.
Another example of a design/review
failure is a fire in an occupancy that
simply overwhelms the ability of sprin-
klers to control it. Usually these types
of fires involve a deficiency in water
supplies and are fortunately few and
far between. The study includes fires in
three warehouses—which stored rub-
ber tires, non-woven textiles, and baled
cotton—where deficient water supplies
were believed to be the critical factor in
the failure to control the fire. All three
fires resulted in the total destruction of
the buildings and contents.
Saddest of all are those post-design/
review cases where the capital expense
of sprinklers has been incurred but the
benefit of sprinkler protection is not re-
alized at the time it is most needed, dur-
ing the fire. In these situations, known
skewed the comparison in a relatively
small sample in the wood/furniture
category. This was the only category
where the estimated average damage
caused by a sprinklered fire was actually
higher than that of a non-sprinklered
fire—$582,000 to $438,000, respective-
ly. Without the single large hydraulic
oil fire, however, the estimated average
damage caused by sprinklered fires
would have been $391,000—a relatively
small differential due to the fortunate
lack of serious non-sprinklered fires.
While the study included warehousing
as an industrial occupancy, there are so
many variables of commodity, storage
height, arrangement, sprinkler design,
and other factors that an enormous
amount of loss data would be needed to
drill into it very deeply. From a loss point
of view, there was not a large number of
fires to generate the data. The study did
find that fires in facilities without sprin-
klers typically did three times more dam-
age than those in sprinklered properties,
which was in line with the 3:1 ratio
generally found across all occupancies.
The human element
The study also revealed cases where
sprinklers were present but were
impeded or impaired. Most of these
cases—where sprinklers were pres-
ent but failed to control the fire within
expected limits—could be traced to
failures in what we termed the “human
element,” which are those activi-
ties, conducted by people, that either
support or defeat the effectiveness of
sprinkler protection.
The human element begins with
proper design and review of sprinkler
system placement and hardware, as
well as competent evaluation of the
water supply to meet the intensity of
the fire anticipated by the occupancy.
For best results, system design should
anticipate the most challenging condi-
tions, not necessarily the least costly
solution. That said, sprinklers are
amazingly resilient and often perform
Photograph: Dave Roback, Springfield-Republican
are omitted in concealed spaces and other
enclosures. Fires in this category tend to be
as damaging as if there were no sprinklers.