1

WATER MIST (FINE SPRAY) FIRE PROTECTION INLIGHT HAZARD OCCUPANCIES

Robert G. Bill, Jr., Richard Ferron and Antonio BragaFactory Mutual Research Corporation

1151 Boston-Providence TurnpikeNorwood, MA 02062

ABSTRACT

The capability of impingement type water mist (fine spray) nozzles, with operating pres-sures lower than 12 bar, was investigated in full-scale fire tests using light hazard scenar-ios. In these tests, four prototype nozzles were evaluated; however, only one of the nozzles(Nozzle A) provided satisfactory protection. The results of the tests, with Nozzle A, arereported as an indication of the capabilities of impingement type water mist nozzles inlight hazard occupancies. Nozzle A was typically operated at a flow rate of 15 Lpm anda pressure of 10 bar. The fire tests were conducted in a room 5.5 m x 5.5 m with openingsto two adjacent rooms (and in one test, an opening to a corridor). The fuel packagesused in the fire tests were: the residential fuel package of Underwriters Laboratories (UL),a living room fuel package, a bedroom fuel package, and a shielded heptane pool fire.The UL and the living room fuel packages were used in a room corner and along a wallbetween two nozzles. Gas temperature measurements indicated that Nozzle A providedacceptable protection in the fire tests of this study with a total water flow rate of 45 Lpm.Based on the results of this study and a review of previous work with water mist systems,recommendations for approval testing and installation standards have been made forlight hazard scenarios.

INTRODUCTION

Water Mist (fine spray*) systems for fire protec-tion have been extensively investigated in thelast several years, due in part to the phasing outof halon and the need to find alternatives. For

example, halon alternatives are being sought forshipboard engine rooms, telecommunication cen-tral offices, combustion turbine enclosures, and

, other enclosures with flammable liquid hazards.Proposed water mist systems have been dis-cussed at specialized technical conferences2,3 andat conferences where details of other halon alter-natives or other suppression systems have beenpresented (see, e.g., Refs. [4] and [5]). A compre-hensive review of water mist technology wasrecently prepared by Mawhinney and Richardson.~ 6

*Fine spray would seem to be a more appropriate term thanmist since the term mist is typically used for water dropletsthat are much smaller than those generated by current com-mercial systems for fire protection. For example, Reference 1indicates that drop sizes of mists range from 0.01 to10 microns, while droplets for sprays range from 100 micronsto about 3000 microns, a range that encompasses currentwater &dquo;mist&dquo; technology.

The potential of water mist for fire protection hasbeen recognized for many years. For example, theFactory Mutual Engineering Division investi-gated the use of water mist for the protection offlammable liquid spray fires in the 1940s.’ It wasrecognized that small water droplets could beentrained into the fire, producing cooling andoxygen dilution in the combustion zone. While

potential advantages over the use of traditionalsprinklers were revealed, the higher requiredoperating pressures were considered to be a prac-tical barrier at that time.

In studies such as that of Rasbash, Rogowski,and Stack8, the role of drop size in flame coolingand oxygen dilution has been quantitativelyinvestigated; however, understanding of theextinguishment mechanisms of water sprayremains insufficient to design water mist sys-tems. A review of water mist technology byMawhinney, Dlugogorski, and Kim9 states thatthe major extinguishment mechanisms are heatextraction, oxygen displacement, and blocking ofradiant heat (reduced thermal feedback to burn-ing and unburned surfaces), but no design

-2-

method is given. Observation of water mist pro-tection of combustion turbine enclosures by theFactory Mutual Research Corporation (FMRC)&dquo;suggests that flame destabilization through theturbulent or mean flows generated by water mistsystems may also be an important mechanismfor some applications. Another important consid-eration in the performance is the oxygen con-sumption by the fire. Bill et all’ observed in simu-lated engine room test fires with flammable liq-uids that extinguishment did not occur unlessthe compartment oxygen concentration wasreduced to about 17%. The complex nature of thedelivery of water mist to the combustion zone isprimarily responsible for the difficulty in devel-oping design criteria for water mist compared toinerting agents.

Because of the recognition that smaller dropletsthan those generated by traditional sprinklerswould more effectively extract heat from the com-bustion zone (see, e.g., Ref. [8]), it has been antic-ipated that water mist systems, in addition tobeing a halon alternative, might provide protec-tion superior to sprinklers in some occupanciesby requiring smaller water supplies. Such con-siderations have led to the development and test-ing of water mist systems for cabins and publicaccommodation areas on passenger ships, whereweight is an important consideration in thedesign of new ships or in retrofitting of fire pro-tection onto existing ships. Examples of such sys-tems are discussed by Turner (in Ref. [2]) andArvidsonlz. The fire testing associated with cab-ins on passenger ships represents the most rele-vant information concerning the use of watermist in light hazard occupanciesl3. Furnishingsin modern passenger ship cabins are similar tothose of residential occupancies, such as homesor hotels. Fire experience on ships indicates thefire hazards are also similar in that fires can befast growing and involve fire propagation alongwalls.

Fire test procedures for evaluating water mistsystems have been developed by the Interna-tional Maritime Organization 14 (IMO). These firetests have been designed to establish whetherwater-based systems will provide protectionequivalent to sprinkler systems regulated underSafety of Life at Sea (SOLAS) regulations.15 Forcabins between 12 m2 and 50 m2, the fuel pack-ages from the FMRC16 residential fire test, or

equally acceptable, the Underwriters Labora-tories (UL) residential fire test,17 are prescribed.In the current study, the appropriateness ofthese tests for generic light hazard occupanciesincluding residential occupancies was investi-gated.

In 1994, the U.S. Fire Administration sponsoreda program’8 to investigate the use of water mistsystems for residential protection. In an evalua-tion of five different water mist systems coveringdifferent technologies, it was concluded that &dquo;thefeasibility of water mist as a residential fire sup-pression system has been demonstrated under alimited number of residential fire scenarios.&dquo;

The objective of the current study was to continueto evaluate water mist protection for light hazardand residential occupancies and to develop a firetest protocol appropriate to test the performanceof these systems in light hazard occupancies andto develop the related appropriate installationstandards.

WATER MIST NOZZLECHARACTERISTICSWater mist can be generated by a variety of tech-nologies. These include impingement (sprinkler-like) nozzles at low pressures (less than 12 bar)and high pressure single fluid nozzles, and airatomization nozzles. A review of these technolog-ies is given in Refs. [2] and [9]. This study onlyinvestigates the performance of prototype lowpressure nozzles because low pressure singlefluid nozzles can, in principle, use much of theequipment associated with sprinklers. Four pro-totypes were investigated in this study; however,only one nozzle (Nozzle A) provided acceptableperformance. Details and test results for theother nozzles are available in the FMRC watermist study’9 for the USFA. The failure of theother nozzles is indicative of the need to evaluatewater mist systems through full-scale fire test-ing. Characteristics of Nozzle A are given inTable 1.

Impingement nozzles are similar to sprinklersin that droplets are generated by the impinge-ment of a water stream onto a deflector. Theselow pressure water mist nozzles may have a

sprinkler-like appearance, but they have special

3

Table 1. Characteristics of Nozzle A

*

Based on distance of corner nozzles from walls.

small orifices, i.e., K-factors* of the order of14 Lpm/(bar)1/2 or less and special deflectors togenerate a water mist. While drop sizes are gen-erally larger than for other water mist generat-ing technologies, sufficient numbers of smalldroplets may be generated to allow flame coolingand oxygen displacement to occur. The existenceof larger drops may be beneficial in that penetra-tion of fire plumes and suppression through sur-face cooling, as with sprinklers, may also occur.Thus the impingement nozzle may be thought ofas a hybrid water mist/sprinkler nozzle.

Nozzle A was a prototype which was commer-cially produced. The nozzles discharged verti-cally downward. In order to attempt to under-stand some aspects of the nozzle performance,characteristics of the water mist and the thermal

response of the nozzle actuating elements havebeen measured. Although NFPA 7502° statesthat &dquo;currently no general design method is rec-ognized for water mist protection,&dquo; NFPA 750recommends that cumulative volumetric distri-butions of water droplets, horizontal water dis-charge distribution, profiles of the spray enve-lope, and spray thrust force be measured as ameans of developing scientifically based systemdesigns in the future. NFPA 750 defines watermist as a spray in which 99% of the volumetric

flow, at a point 1 m below the nozzle in the coars-est part of the spray, is associated with dropletsless than 1 mm in diameter. FMRC, however, inits approval process, has no restriction on drop

*The nozzle K-factor is a discharge coefficient defined as theratio of the flow rate to the square root of the operating pressure.

size. Rather, water mist systems are identifiedthrough adequate suppression and/or extin-guishment in fire tests. In particular, acceptableperformance in shielded fires as described laterin this paper is taken as evidence that the systemresponds as a water mist system.

Measurements analogous to those recommendedby NFPA have been made, along with others foruse in the current program. Table 1 lists charac-teristics of Nozzle A: the K-factor, typical flowrates, operating pressure, nozzle spacing, andthermal response parameters (RTI, C, and nomi-nal operating temperature). These thermalresponse parameters were measured followingthe ISO procedures for automatic sprinklers.21

Cumulative drop size distributions for Nozzle A,supplied by the manufacturer 3 were given attwo pressures, 9 bar and 11 bar, at a point 1 mbelow the nozzle. Note that these two pressuresbracket the operating pressures in this study.The volume median drop sizes were measuredto be 180 microns at 9 bar and 225 microns at11 bar. At the operating pressure used in thisstudy, the volume median drop size is thereforeapproximately 200 microns. The drop sizes cor-responding to a cumulative volume of 99% were360 and 400 microns, respectively, for the 9 and11 bar operating pressures. The increase of dropsize with operating pressure appears to be ananomaly given the expected decrease in drop sizewith pressure. This result is possibly due to mea-surement at a single point and the change inspray pattern with increasing pressure.

4

Cumulative drop size distributions were alsoreported by the manufacturer at radii of 0.36 mand 0.78 m in the same plane. As would beexpected due to the increased effect of drag onthe radial throw of smaller droplets, the mediandrop size increases with radius. The median dropsize increased to approximately 280 and500 microns, respectively.

The horizontal water densities of Nozzle A were

measured in one quadrant of the coverage areaof each nozzle at a distance of 2 m below thenozzle. This elevation was selected because theseat of the upholstered chair used in the livingroom fire test is at this elevation. The tests were

conducted in a room approximately 7.0 m x7.0 m, with densities collected in 0.093 m2 pans.The flow rate for each nozzle was 15.1 Lpm, asshown in Table 1. The results are given in Fig. 1.(Note that the orientation of the frame arms isshown schematically.)

The thrust forces of the nozzles used in the studywere measured using a load cell as in Ref. [22].The force of the spray on a plate 0.33 m in diame-ter was measured approximately 25 mm from thetip of the nozzle. The weight of the water wasestimated and subtracted from the thrust force.

Figure 1. Horizontal water flux distribution for Nozzle A,2 m below the deflector (flux in mm/min). (: indicatesorientation of frame arms).

The thrust force was measured to be about 5 Nat 10 bar.

FIRE TEST FACILITIES

Test CompartmentFire testing was conducted within the &dquo;HUD&dquo;

building located at FMRC headquarters in Nor-wood, MA. The facility had been previously usedfor fire testing, leading to the development of aprototype Limited-Water-Supply sprinkler23.The test area used in that study was re-compart-mented for use in the current program. A sche-matic of the test facility is shown in Fig. 2 alongwith the nozzle configuration. The frame armsof nozzles were parallel to the north wall in alltests, except those in which the fuel package wasplaced between two nozzles, as shown in Fig. 2.In those tests, the frame arms were parallel tothe east wall. It was observed that the framearms did not significantly obstruct the waterspray pattern.

Fire testing was conducted within the largestroom in the facility. The room was 5.5 m x 5.5 min area, with a ceiling height of 2.44 m. In addi-tion to the main fire test room, the facility wasconstructed so that fire products could ventilateinto connecting rooms (shown in Fig. 2 as anoptional opening). A 12 m long, 1.5 m wide corri-dor was also constructed to simulate hallwayscommon in offices, condominiums and apartmentbuildings with multiple residences. The purposeof the corridor, as with the rooms connecting tothe main fire room, was to evaluate the effectsof natural draft on the performance of the watermist system (number of actuating nozzles andrequired water supply).

The test room was fabricated using 5 cm x 10 cmstuds and 5 cm x 20 cm ceiling joists. The wallsand ceiling were covered with 1.3 cm thick plas-ter board. Combustible paneling and ceiling tileswere also used as part of the fuel packages, andwill be discussed in the following section.

Fire Test PackagesFour different fuel packages were used in thecurrent study: the UL 1626 17 corner residentialpackage, an FMRC corner residential packagels,a bedroom fuel scenario, and a kitchen fuel sce-nario.

5

Figure 2. Schematic of test facility showing nozzle locations for Nozzles A (dimensions in meters).

-6-

The UL 1626 corner package consists of a woodcrib and polyether foam attached to wood sup-ports. Wood paneling with an ASTM E84 flamespread number of 200 is used along the walls,and combustible paneling with a flame spreadnumber of 25 is used along the ceiling. Completedetails of the fuel package are given in Ref. [16].A schematic of the package as installed in thenortheast corner of the main fire test room is

shown in Fig. 3.

Originally, the FMRC residential fuel package 16was based on a living room scenario using actualfurnishings. Subsequent to this study, the FMRC

residential fuel package has been revised to con-sist of simulated furnishings, such as used inUL 1626, rather than actual furnishings. Theoriginal FMRC residential fuel package has beenused throughout this study. (It will be referredto herein as the FMRC residential fuel pack-age.)16 It consisted of a customized vinyl coveredchair with polyurethane foam padding, two lay-ers of curtains, a wooden end table, and a simu-lated end of a sofa. As in the UL package, combus-tible paneling and ceiling tiles are used. A sche-matic of the FMRC residential fuel package 16 isshown in Fig. 4. An analysis of sixty full-scaleresidential fire tests in the Los Angeles Residen-

Figure 3. Schematic of UL 1626 corner residential fuel package (dimensions in meters).

7.

Figure 4. Schematic of FMRC corner residential fuel package (dimensions in meters).

tial Test Program24 has shown that the FMRCresidential fire test represents a worst case firescenario for residential sprinkler protection. Inaddition to these fire tests, FMRC has also inves-

tigated fire scenarios in hotel rooms25 andoffices.26 A comparison of sprinklered fire testresults indicates that the corner living room sce-nario of the Los Angeles Residential Test Pro-gram is an appropriate worst case scenario forlight hazard occupancies. FMRC has also usedthis scenario in the development of FMRCApprovals criteria for use with extended cover-age sprinklers in light hazard occupancies. 27

Details covering the burning characteristics ofthe upholstered chair are given in Refs. [23] and[28]. A listing of contents of the FMRC residentialfuel package&dquo; is given in Table 2.

The bedroom fuel package scenario consisted ofa standard commercially available bed placed inthe northeast corner of the main fire test room.The contents of the bed are listed in Table 3.

A fourth fuel package consisted of a shielded hep-tane pool fire on a counter above the floor. This

8

Table 2. Description of FMRC Residential Fuel Package

Table 3. Standard Bed Contents

hazard was much more severe than would be

expected for light hazard occupancies and pro-vided an indication of the capabilities of thewater mist system.

Water Supply SystemWater was supplied in all fire tests to the watermist nozzles by an electric motor (20 HP) drivencentrifugal pump rated for 492 Lpm at a pres-

9-

sure of 13 bar. Water was supplied to the watermist nozzles through a 2.54 cm (1 in.) CPVC feedline connecting to 2.54 cm (1 in.) branchlinesabove the ceiling. The water mist nozzles, all

discharging with their deflectors about 50 mmbelow the ceiling, were connected to the watersupply system using 1.27 cm (1/2 in.) CPVC cou-plings connecting to 1.27 cm (1/2 in.) CPVC nip-ples ; CPVC tees and elbows were used and alljoints were bonded with CPVC cement. The loca-tions of water mist nozzles are shown in Fig. 2.Flow rates were controlled by monitoring watersupply pressure. The appropriate system pres-sure was determined by flowing open nozzles andmeasuring the volume of flow over a 1 min timeinterval. This established the operating pressurefor the desired flow of a single nozzle. When mul-tiple nozzles operated, the pressure was manu-ally adjusted to maintain the supply pressure.The centrifugal pump was operated prior to noz-zle actuation by the use of a bypass line andby flow through an orifice (external to the testfacility) at the end of the main feed line. Flow andpressure determinations were made with theseestablished flows.

InstrumentationInstrumentation to measure gas temperature,ceiling temperature, and gas concentrations ofoxygen, carbon monoxide, and carbon dioxidewere installed in the test facility. Thermocouplesfabricated from Inconel sheathed 30 gaugechromel-alumel wire were used to measure gastemperatures. Thermocouples were installed inthe center of the main fire test room 76 mm belowthe ceiling and 1.5 m above the floor. In each firescenario, a thermocouple was installed over theignition point, 76 mm below the ceiling. A ther-mocouple was also embedded 1 mm in the ceilingover ignition.

Gas sampling occurred at an elevation of 1.5 min the room center. The gas flow rate was 8 Lpmwith a lag time to the gas analyzers of about 30 s.The gas samples were drawn through a tube witha glass wool filter, and a desiccant. Concentra-tions of carbon monoxide and carbon dioxidewere obtained using infrared analyzers. Oxygenconcentrations were obtained using a paramag-netic analyzer.

The times of nozzle actuation were determined

through recording of a voltage drop due to the

breaking of an electrical circuit caused by theactuation of the nozzle. Data was obtained

through computerized data acquisition at a rateof 1 scan per second.

DESCRIPTION OF’ FIRE TESTS

The four fuel packages described previously wereused in six fire scenarios. These scenarios werethe FMRC corner residential fire test,16 theFMRC residential fire test with fuel packagebetween two sprinklers, the UL 1626 corner resi-dential fire test,17 a fire scenario using a modifiedUL fuel package between two sprinklers, a bed-room scenario, and a high challenge scenario con-sisting of a shielded heptane pool fire.

The procedure for supplying water to the watermist systems was similar in all tests. Water flowto the nozzles was maintained for at least 10 minunless damage to the test facility was antici-pated. If it appeared, through monitoring of thegas temperature over ignition, that the fire wasextinguished, the water flow was discontinued10 min after the first nozzle actuation with datafrom the room being acquired for up to a totaltest time of 20 min. Otherwise, the water supplypressure was maintained after initial nozzleactuation up to a total test time of 20 min.

The FMRC residential fuel package 16 is shownin Fig. 4. Ignition was achieved through the useof two 0.30 m long cotton wicks which hadabsorbed 40 mL of methanol. The wicks were

placed adjacent to the chair, as shown in Fig. 4.Ignition was initiated with a match. In this testscenario, the optional opening to the corridorshown in Fig. 2 was closed.

In a second scenario, the FMRC test packagewas installed along the east wall between twosprinklers, as shown in Fig. 2. In this scenario,however, the simulated sofa end was not used.The chair and cotton wicks used for ignition werelocated so as to be equidistant from the closestnozzles. The optional door opening, shown inFig. 2, was used in this scenario.

The UL 162617 corner residential package, shownin Fig. 3, was installed in the northeast cornerof the main burn room. As described in Ref. [17],0.2 L (8 oz) of n-heptane were placed in a pan

10

directly below the wood crib, and 113 g (4 oz) ofexcelsior were located on the floor adjacent toeach section of simulated furniture (foamattached to wood supports). The n-heptane wasignited and, 40 s later, the excelsior was alsoignited.

As with the FMRC residential fuel package&dquo;, asecond scenario between two nozzles was investi-

gated using the UL residential fuel package. Inthis scenario, the crib was located as in the sec-ond FMRC scenario, i.e., between the two nozzlesas indicated in Fig. 2. The wood crib was located25 mm from the wall, as in the corner test withthe crib center equidistant from the two closestnozzles (Fig. 5). The wooden supports, with poly-ether foam attached, were located as shown inFig. 5. The distance between the supports alongthe wall was 1.4 m. Excelsior was placed on thefloor as in the corner test. Ignition of the fuelpackage was as in the standard UL residentialcorner test.

The bedroom scenario was conducted with thebed (contents described in Table 3) located in thenortheast corner of the main fire test room. Thehead of the bed was 76 mm from the north walland the right side of the bed was 0.36 m fromthe east wall (see Fig. 2). A trash can (180 mm indiameter) was filled with 10 sheets of newspaperand placed adjacent to the head of the bed on theright side. Ignition was by a match placed inthe bottom of the trash can through a 25 mm x25 mm opening on the side.

The final scenario was a high challenge testincorporating a shielded heptane pool fire placedon a counter with simulated cabinets above. Thescenario is not representative of residential orother light hazard scenarios, however, it pro-vided a means of investigating the limitations ofwater mist systems. The counter/cabinet assem-bly was placed in the northeast corner of themain fire test room with the counter/cabinets

running 2.1 m longitudinally along the northwall. The counter top was 0.91 m high and 0.61 mdeep. The cabinets were 0.5 m above the countertop and were 0.91 m high and 0.4 m deep.

Using the counter/cabinet assembly, a high chal-lenge shielded flammable liquid fire was devel-oped. A 0.3 m diameter pan, filled with 600 mLof heptane, was raised 0.1 m above the counter

top. The pan, placed 0.2 m from the east wall,was situated so that the bottom of the cabinets,0.4 m above the pan, covered one half of the pan.A cabinet door, 0.6 m by 0.6 m, was simulatedabove the pan with a vertical 0.63 cm ( 1/4 in.)plywood panel which was attached to the cabi-nets’ surface on one edge and was propped openat a 30° angle with respect to the cabinet. Thesimulated door opening shielded a portion of thefire from droplets discharged by the water mistnozzles. (Note, however, no opening was cut intothe cabinet surface).

The fire tests, utilizing Nozzle A, are listed inTable 4. The table indicates the fire scenario,number of nozzles actuated, and fire test results.Unless otherwise noted, nozzle flow rates, spac-ing and other characteristics were as in Tables 1and 2. Details of all the fire tests are discussedin the following section.

FIRE TEST RESULTS

As indicated in Table 4, eleven fire tests wereconducted in the program using Nozzle A (and,in Test 10, a prototype limited-water-supplysprinkler) and the fire scenarios were describedin the previous section. The fire tests weredesigned to last 20 min; however, some testswere aborted earlier using hose streams to avoidexcessive damage to the test facility (Tests 8band 15). When significant suppression occurred,the water supply was disconnected 10 min afterfirst nozzle actuation (Tests 2, 3, 6, 7, 10 and 12).The 10 min period corresponds to the minimumsprinkler water supply duration under NFPA 13D29for one- and two-family dwellings. (The watersupply requirement under NFPA 133° for lighthazard occupancies is 30 min). In Test 7, fireregrowth occurred and the water supply wasreconnected, resulting in fire suppression. Thewater supply was also reconnected in Test 3, butroom tenability was not maintained. In theremaining tests (Tests 4, 9, and 13), the watersupply pressure was maintained after nozzleactuation throughout the entire test period.

Test results were judged following the room tena-bility criteria of the Los Angeles Residential TestProgram:24 gas temperatures over ignition,76 mm below the ceiling, are not to exceed 316°C;the ceiling surface temperature over ignition is

11

Figure 5. Schematic of modified UL 1626 residential package with ignition between two nozzles (dimensions in meters).

not to exceed 260°C; and the gas temperature inthe room center at an elevation of 1.5 m is notto exceed 93°C. Gas temperatures at a height of1.5 m in the room center did not exceed 51°C in

any test. Although water spray did not impingedirectly on the thermocouple at that location, ageneral build-up of fine droplets in the room mayhave wetted the thermocouple; hence, theseresults are not considered to be reliable and arenot used further in the analysis of the testresults. The maximum allowed carbon monoxideconcentration is 3000 ppm. In Table 4, whenthese criteria are satisfied, the test result is lis-ted as &dquo;suppression&dquo; or &dquo;extinguishment.&dquo; Extin-guishment implies no flames were observed at

the conclusion of the test. The nozzle actuationtimes in each test are given in Table 5. Maximumtemperature and gas concentrations (minimumfor oxygen) are given in Table 6.

In the first test of Nozzle A, Test 2, the FMRCcorner residential package 16 was used and rapidsuppression occurred after the actuation of thenozzle closest to the fire. The water supply wasdisconnected 10 min after nozzle actuation. Nofire regrowth occurred and the fire was seen tobe extinguished at the end of the 20 min test.Temperature results are shown in Fig. 6.Nozzle A extinguished the FMRC residential fuelpackage at a flow rate of 15.1 Lpm. For compari-

12-

Table 4. Fire Test Summary

1 Initial period plus additional time if fire regrowth occurred and water turned on again.2Flow rate varied between 11 and 15 Lpm per nozzle (below minimum specified by manufacturer).3 Bulb actuating element with 68°C temperature rating.4Test aborted because of severe fire development.

son, fire regrowth has been observed to occurin this fire scenario with residential sprinklersoperating at 68 Lpm.24 The severity of the fuelpackage is indicated by the Required DeliveredDensity (RDD) of the FMRC chair which hasbeen measured to be between 9 and 15 mm/min.28

Tests 4 and 13 were conducted at the manufac-turer’s specified flow rates using Nozzle A withthe UL 1626 corner package, and provided a com-parison between the FMRC residential fuel pack-age and that used by UL. In Test 4, the firegrowth was arrested when the first nozzle actu-ated at 1 min after ignition; however, the fire sizewas maintained. A second nozzle then operated1 min 34 s after ignition, and fire suppressionoccurred followed by regrowth at about 4 min(see Fig. 7). Because the fire size was significant(as judged from the gas temperature) 10 min intothe test, the water supply pressure was main-tained throughout the remaining test period. The

fire was suppressed and no other nozzles actu-ated.

Because of the symmetry of the nozzles withrespect to the fire location in Test 4 (see Fig. 2),it was expected that a third nozzle might operate.To investigate this possibility, the conditions ofTest 4 were repeated in Test 13. In this test, thethree nozzles closest to the fire actuated. After

actuation, the fire was initially suppressed; how-ever, the fire then regrew, followed again by sup-pression.

The maximum gas temperature over ignition inTest 13 was 399°C, 84°C above the tenabilitylimit. However, the maximum ceiling surfacetemperature and carbon monoxide concentra-tions were acceptable (216°C and 2086 ppm).Given the brief nature of the temperature peak(see Fig. 8) and the acceptable value of the othertenability factors, the fire is considered to be sup-

13-

Table 5. Actuation Times (min:s)

* See Fig. 2 for location of nozzles.** Nozzle No. 9 operated (see Fig. 2).DNO = Did not operate.

Table 6. Gas Concentration and Temperature Results

* Instrument malfunction

pressed for the purposes of this study. The differ-ence in the results between Test 4 and 13 maypossibly be due to slight differences in the nozzlesdue to their fabrication as prototypes rather thanproduction nozzles.

In order to provide some insight into the perfor-mance of the water mist nozzles, Test 10 wasconducted with the UL corner package using pro-totype Limited-Water-Supply sprinklers. Theprototype was developed by FMRC in a program23

14

Figure 6. Temperature results for FMRC residential corner test (Test 2); -, ceiling gas temperature; -.-, ceilingsurface temperature.

sponsored by the USFA to develop a sprinklersystem for manufactured homes requiring a totalwater supply of 380 L. The sprinklers wereinstalled at the same locations as Nozzle A,shown in Fig. 2. The design flow rate of 38 Lpmwas provided upon sprinkler actuation. The firewas suppressed with only a single sprinkler actu-ation, and suppression was maintained when thewater supply was discontinued after 10 min.This contrasts sharply with the test reported inRef. [23], in which three sprinkler actuationsoccurred with the FMRC package and controlwas lost after the water supply was discontinued.

The above results are indicative of the different

suppression mechanisms employed by sprinklersand water mist systems. In the UL scenario, theburning crib, being close to the wall, is a reliableignition source for the combustible paneling. Incontrast, in the FMRC scenario, the chair is theprimary hazard with the paneling becominginvolved only after considerable fire development

in the chair. Sprinklers with relatively largedroplets, and a spray pattern designed to havehigh wall wetting, may easily prevent propaga-tion up the wall paneling if sprinkler actuationis early enough. Water mist nozzles, however,with smaller droplets and lower flows provideless wall wetting, and are therefore more vulner-able to vertical fire propagation on combustiblesurfaces. The superior performance of the watermist system in the FMRC scenario suggests thatthe cooling of flames above the chair, absent withlarger sprinkler droplets, is significant to sup-pression in this scenario.

Compared to the FMRC residential fuel pack-agel6, the UL package is more easily specified andconstructed, and represents a more challenginglight hazard fire test for water mist systems. (TheFMRC residential package appears to be moresevere for automatic sprinklers). It was thereforerecommended that a UL 1626 type corner test

15-

Figure 7. Temperature results for UL residential corner test (Test 4); -, ceiling gas temperature; -.-, ceilingsurface temperature.

be adopted as part of a water mist system FMRCApproval test.

Two other tests, Tests 3 and 9, were conductedusing the UL 1626 corner scenario with Nozzle A.In these tests, the flow rate was not steady dueto varying supply pressures. The flow variedbetween 11 and 15 Lpm. In both tests, 3 nozzlesactuated and the fire was not controlled. For com-

parison, results of Test 9 are shown in Fig. 9.This indicates that a minimum flow rate higherthan 11 Lpm is needed for acceptable perfor-mance by Nozzle A. In Test 15, Nozzle A wasinstalled with a 68°C temperature rated bulbcompared to 79°C used in the other tests withthis nozzle. Although the nozzle flows were thesame as in successful Tests 4 and 13, the testhad to be aborted to avoid damage to the facility.It is unclear why this failure occurred, particu-larly considering the successful use of 68°C tem-perature rated water mist nozzles in similar test-ing3l. Test 15 appears to be an anomaly which

unfortunately could not be further investigatedin this program.

A maximum of three nozzles actuated in any ofthe corner tests using Nozzle A. In order to inves-tigate if additional nozzles could actuate in otherscenarios, the FMRC package and a modified ULpackage were used in two tests with ignitionbetween two nozzles. The details of the fire sce-nario are described in the previous section. InTest 7, the FMRC package was used and NozzleType A was installed as in Fig. 2. The nozzle inthe northeast corner actuated 1 min 54 s after

ignition (Fig. 10). The fire was rapidly sup-pressed ; however, fire regrowth occurred and asecond nozzle actuated 4 min 40 s after ignition.The fire was again suppressed and the water flowwas discontinued 10 min after the first nozzleactuation. Fire regrowth occurred 13 min afterignition and the fire was suppressed when thewater flow was reestablished.

16-

Figure 8. Temperature results for UL residential corner test (Test 13); -, ceiling gas temperature; -.-, ceilingsurface temperature.

Because this scenario was more challenging forNozzle A than the FMRC corner scenario, a mod-ified UL package was installed in Test 12 asshown in Fig. 5. In this test, three nozzles actu-ated. Nozzles 3, 6 and 9, as shown in Fig. 5, actu-ated at 55 s, 46 s, and 55 s, respectively. The firewas suppressed and the water supply discon-nected 10 min after the first actuation (Fig. 11).A comparison of the results shown in Figs. 7 and11 indicates that the UL corner test was a moresevere challenge than the modified UL testbetween two nozzles. The three-nozzle actuationin Test 12, however, underlines the conclusionfrom Test 13 that, for Nozzle A, a water supplyof at least 15 Lpm for each of three nozzles isneeded for adequate performance.

Two other fire scenarios were conducted withNozzle A. In Test 6, the bedroom scenario wasused. One nozzle actuated and extinguished thefire. Water flow was discontinued in this test,10 min after nozzle actuation.

In Test 8b, a shielded heptane pool fire was used,as described in the previous section. It was envi-sioned that the small droplets would beentrained into the pool fire and result in extin-guishment. About 6 min into the test, the woodpaneling of the cabinet, shielded by an open cabi-net door, began to burn vigorously. Nozzle actua-tions occurred at 7 min 1 s and 7 min 47 s; how-

ever, the fire was not suppressed. Despite gastemperatures exceeding 900°C under the ceilingover ignition, no additional nozzles operated. Theceiling surface temperature, however, was ashigh as 673°C and the carbon monoxide concen-tration maximum was 3923 ppm. The test wasaborted 18 min after ignition using a hosestream.

The small number of actuated nozzles attest tothe very effective cooling of the small dropletsgenerated by the water mist nozzles. After thetest, the nozzles were actuated to see if dropletswere impinging on the unactuated nozzles.

17

Figure 9. Temperature results for UL residential test (Test 9) with low water flow rate; -, ceiling gas temperature;-.-, ceiling surface.

Although no droplet impingement was visible, ageneral build-up of water mist in the room wasobserved, which may account for the small num-bers of nozzles actuated. For residential protec-tion with a limited water supply or light hazardoccupancies where it is critical to limit waterdamage, this may be advantageous in that thefire is localized while cooling throughout theroom reduces nozzle actuation.

DISCUSSION OF RESULTS

The fire test results presented above indicatethat water mist systems have the potential toprovide protection for light hazard occupanciescomparable to sprinklers as used in commercialor residential applications. The maximum num-ber of nozzles operating in any test was threewith a total application of 45 Lpm (Tests 3 and12). The tests in which three nozzles operatedinvolved simulated furnishings; however, whenrealistic furnishings were used (Tests 2 and 7)

the performance was improved. The excellentperformance of the water mist systems with realfurnishings provides a high level of confidencethat water mist systems have the capability ofprotecting light hazard occupancies at reducedwater supply levels compared to traditionalsprinkler systems. Finally, in Test 8b, in whicha shielded heptane spray fire was used, only twonozzles operated due to the high cooling at theceiling even though the hazard was well beyondthat of a light hazard occupancy. This test resultsuggests that a further benefit of water mist sys-tems compared to sprinklers is reduced area ofwater damage.

In contrast to the water mist performance sum-marized above, current FMRC sprinkler stan-dards for unconfined light hazard occupanciesrequire a total water flow of 568 Lpm (4.1 mm/min over 139 m2 area). FMRC Approvedextended coverage (EC) sidewall sprinklers maybe used in compartments up to 150 m2 in area, at

18-

Figure 10. Temperature results from testing using FMRC fuel package placed between two nozzles (Test 7);-, ceiling gas temperature; -.-, ceiling surface temperature.

a density of about 4.1 mm/min and a maximumcoverage area of 37 m2. (Sidewall sprinkler watersupplies are governed by end pressures). 15 For acompartment the same size as that tested in thisstudy (30 m2), the required total flow rate wouldbe 124 Lpm.

In order to establish nozzle flow and spacingrequirements along with total flow requirementsfor water mist systems in light hazard occupanc-ies, FMRC has established three full scale firetests: 1) a compartment test similar to thosereported here in which the UL residential pack-age is installed in the room corner, 2) a fire testwith bunk beds which demonstrates the capabil-ity of the water mist,to suppress fires wheredirect spray impingement is not possible, and3) an &dquo;unconfined&dquo; space test using simulatedcouches. These tests have been adopted from theIMO standard for water based systems.&dquo; Twotypes of FMRC Approval are available for watermist systems protecting light hazard occupanc-

ies : systems restricted for use in compartmentsto a maximum size of 37 m2 and systems whichcan be installed without any restriction in regardto area. In the former case, only the first twotypes of fire tests must be conducted. All threetypes of fire tests must be conducted for systemswith no restriction on compartment area. Keydetails of the three tests are given below. Furtherdetails are available in the FMRC Water Mist

Light Hazard Standard.32

The bunk bed test is adopted from a series ofsuch tests intended to simulate small cabin firesin the IMO standard. A fire is ignited in a lowerbunk bed mattress in a room 12 m2. The mattressis made of polyether foam with a cotton fabriccover. In the FMRC test, a single nozzle is placedin the room center. Damage to the cushion of thelower bunk bed is to be limited to 40% by volume.In addition, the maximum ceiling surface andgas temperature (76 mm below the ceiling) overignition are to be less than 260°C and 315°C,

19-

Figure 11. Temperature results from testing using modified UL residential fuel package between two nozzles(Test 12); -, ceiling gas temperature; -.-, ceiling surface temperature.

respectively. Testing3 has shown that sprinklerswill not typically provide this level of protection;hence, this test is a means of differentiatingwater mist systems and sprinklers in light haz-ard applications.

A compartment test, similar to the corner testswith the UL fuel package described in this study,is to be conducted in a square room no greaterthan 37 m2and a ceiling height of 2.5 m. The fuelpackage, however, should have more precise andcontrollable flammability characteristics, mea-surable using small-scale calorimetry. The com-partment size, selected by the manufacturer, willbe the largest size compartment that can be pro-tected by the mist system unless the unconfinedspace fire test is conducted. Nozzles are to beinstalled at the maximum spacing and suppliedwith the minimum water flow specified by thewater mist system manufacturer. A nozzle is alsoto be installed in a doorway diagonally acrossfrom the simulated furniture fuel package. For

systems only FMRC Approved for protection ofcompartments 37 m2 or less (i.e., without the&dquo;unconfined&dquo; space test), the doorway nozzle isnot to actuate. Water supply requirements arebased upon all the nozzles in the compartmentactuating. Other criteria for approval are thatthe gas temperature at 76 mm below the ceilingand the ceiling surface temperature over ignitionshould not exceed 315°C and 265°C, respectively.Based on these tests, water mist systems are tobe FMRC Approved for use in compartmentswith ceiling heights no greater than 3 m, a mod-est extrapolation of the test condition.

Water mist systems to be FMRC Approved foruse in spaces without area restriction are to passthe IMO Public Space Testl4. In this test, simu-lated couches with polyether cushions and cottonfabric covers are placed below nozzles in threeceiling configurations: couches centered underone nozzle, between two nozzles, and betweenfour nozzles. In tests with a ceiling height of

20

2.5 m at the Swedish National Testing andResearch Institute (SP), it was observed that thePublic Space Test resulted in the actuation of sixstandard sprinklers [RTI = 300 (m-s)1/2] installedat a 4 m spacing using a density of 4.4 mm/minwhen the couches were centered between two

sprinklers.33 This indicates that the Public SpaceTest is a severe test for light hazard occupancies.

The ceiling height in the FMRC Approval testfor &dquo;unconfined&dquo; spaces is 5 m. No more than fivenozzles are allowed to actuate with at least onenozzle observed to remain unactuated beyondeach actuated nozzle. In addition, over ignitionsthe maximum ceiling surface temperature andgas temperature at 76 mm below the ceiling areto be no more than 260°C and 315°C, respec-tively. Damage to the couches is restricted to 50%by volume.

WATER SUPPLY AND SYSTEMDEMAND

In the successful tests conducted during thisstudy, water supply duration varied from 10 minto 19 min. However, because in some fire testsextinguishment did not occur, a minimum watersupply of 60 min is required for the operation ofthe water mist system protecting light hazardoccupancy occupancies, the same as required byFMRC for sprinklers. The water supply is to becapable of supplying the hydraulically mostremote nine nozzles for mist systems approvedfor use in &dquo;unconfined&dquo; spaces. This nine nozzle

design approach corresponds to a design involv-ing a ring of nozzles operating beyond that of afire centered under one nozzle. Such a designmay prove conservative, as indicated by the testresults of this study, but a margin of safety isappropriate until further experience is gainedwith this new technology. In the case of mistsystems that are only approved for use in com-partments, the water supply must be sufficientto supply the maximum number of nozzles in anycompartment.

z

CONCLUSIONS

This study shows low pressure water mist tech-nology to be a viable alternative to automaticsprinklers in the protection of areas of light haz-

ard occupancies, such as in offices, hotel rooms,residences, and other similar areas. When com-pared to automatic sprinkler protection in com-partmented areas, tests conducted by FMRCshowed that low pressure water mist systemshave potential for offering protection using lesswater, hence, with reduced water damage atequivalent fire damage areas. Water mist nozzledesign characteristics and operating parameterswere found to have significant impact on the abil-ity of this new technology to suppress or extin-guish the test fires; consequently, a full-scale firetest protocol is required for evaluation of differ-ent nozzles and system designs for light hazardoccupancy applications.

Based on the results of the FMRC fire tests in a

compartmented area and on a review of the IMOfire test protocol for water mist in unconfinedareas, a comprehensive fire test protocol hasbeen recommended in this paper for water mist

systems in compartmented and &dquo;unconfined&dquo;areas of light hazard occupancies.

ACKNOWLEDGMENTS

This study was sponsored by the U.S. FireAdministration (USFA) under contract EMW 94-C-4447 and Factory Mutual Research Corpora-tion (FMRC) as part of its Water Mist TechnologyProgram. The support of Mr. Larry Maruskin,Program Officer (USFA) and Dr. H-C. Kung,Director of Protection Research at FMRC, isgratefully acknowledged. The authors furtherwish to acknowledge the technical support ofMr. Edward Hill, Jr. and Mr. William Brownof FMRC.

REFERENCES

1. McCormick, P.Y., Lucas, R.L. and Wells,P.F., "Gas Solid Systems," in Chemical Engi-neering Handbook (J.H. Perry ed.), McGrawHill, 1963, p. 20-66.

2. Proceedings of the Water Mist Fire Suppres-sion Workshop, March 1-2, 1993, NISTIR5207 (K.A. Notarianni and N.H. Jason, Eds.),Building and Fire Research Laboratory,NIST, Gaithersburg, MD, March 1-2, 1993.

3. Proceedings of the International Conferenceon Water Mist Fire Suppression Systems,

21

Swedish National Testing Institute, Boras,Sweden, November 4-5, 1993.

4. Halon Options Technical Working Confer-ence, The Univ. of New Mexico, New MexicoEngineering Research Institute, Albuquer-que, NM, 1995.

5. Proceedings of International Fire Researchand Engineering, sponsored by Building andFire Research Laboratory, NIST, Gaithers-burg, MD, and Society Fire Protection Engi-neers, Boston, MA, at Orlando, FL,September 10-15, 1995.

6. Mawhinney, J.R. and Richardson, J.K., "AReview of Water Mist Fire SuppressionResearch and Development, 1996," FireTechnology, Vol. 33, No. 1, 1997, pp. 54-97.

7. Thompson, N.J., Fire Behavior and Sprin-klers, 3rd Edition, National Fire ProtectionAssociation, Quincy, MA, 1974.

8. Rasbash, D.J., Rogowski, Z.W. and Stark,G.W.V., "Mechanisms of Extinction of LiquidFuel Fires with Water Sprays," Combustionand Flame, Vol. 4, 1960, pp. 223-234.

9. Mawhinney, J.R., Dlugogorski, B.Z. andKim, A.M., "A Closer Look at the Fire Extin-guishing Properties of Water Mist," FireSafety Science—Proceedings of the FourthInternational Symposium, 1994, pp. 47-60.

10. Ural, E.A. and Bill, R.G., Jr., "Fire Suppres-sion Performance Testing of Water Mist Sys-tems for Combustion Turbine Enclosure,"Halon Options Technical Working Confer-ence, NMERI, Albuquerque, NM, May 9-11, 1995.

11. Bill, R.G., Jr., Hansen, R.L. and Richardson,K., "Fine Spray (Water Mist) Protection ofShipboard Engine Rooms, Fire Safety Jour-nal, Vol. 29, 1997, pp. 317-336.

12. Arvidson, M., "Luxury Cabin, Open AreaPublic Space and Corner Public Space FireTests for Grinnell Fire Protection UsingAqua Mist AMS Nozzles," Test ReportR30455a, Swedish National Testing Labora-tory, Boras, Sweden, August 1994.

13. Bill, R.G., Jr., "Fire Performance for FineSpray (MIST) Systems in Passenger ShipPublic, Accommodation and Service Areas,"Technical Report J.I. 0X0N7.RA, July 1993.

14. "Guidelines for Approval of Sprinkler Sys-tems Equivalent to that Referred to inSOLAS II -2/12," MSC 64/22/Add. 1, Interna-tional Maritime Organization, London,UK, 1994.

15. SOLAS, The International Convention forSafety of Life at Sea, International MaritimeOrganization, London, UK, First Edition,1992.

16. "Approval Standard for Residential and Lim-ited Water Supply Automatic Sprinklers,"Class 2030, Factory Mutual Research Corpo-ration, Norwood, MA, January 1983.

17. "Standard for Residential Sprinklers for FireProtection Service," UL 1626, UnderwritersLaboratories, Northbrook, IL, December1990.

18. "Final Report, Feasibility Study of WaterMist Application for Residential Fires," Con-tract No. EMW-93-4247, U.S. Fire Adminis-tration, Emmitsburg, MD, February 1995.

19. Bill, R.G., Jr., Stavrianidis, P., Hill, E.E., Jr.and Brown, W.R., "Water Mist Fire Protec-tion in Residential Occupancies," TechnicalReport J.I. 0Y1N9.RA/ 0Z0J1.RA, FactoryMutual Research Corporation, Norwood,MA, November 1995.

20. "Water Mist Fire Suppression Systems,"NFPA 750, 1996 Annual Meeting Report onProposals, National Fire Protection Associa-tion, Quincy, MA, 1995.

21. "Fire Protection—Automatic Sprinkler Sys-tems, Part 1, Requirements and Methods ofTests for Sprinklers," ISO 6182-1, Interna-tional Standard, First Edition, Geneva,Switzerland, 1992.

22. Bill, R.G., Jr., Kung, H-C., Brown, W.R., Hill,E.E., Jr., Spaulding, R.D., Stavrianidis, P.,Vincent, B. and Yu, H-Z., "Evaluation of theSuppression Capability of Residential and

-22-

Quick Response Sprinklers," TechnicalReport J.I. 0Q5N0-8.RA, Factory MutualResearch Corporation, Norwood, MA,December 1990.

23. Bill, R.G., Jr., Kung, H-C., Hill, E.E., Jr.,Chan, T-S. and Brown, W.R., "The Develop-ment of a Limited-Water-Supply Sprinklerfor Mobile Homes," Technical Report J.I.0T1N7.RA, Factory Mutual Research Corpo-ration, Norwood, MA, April 1992; see alsoBill, R.G., Jr. and Kung,H-C., "LimitedWater Supply Sprinklers for Manufactured(Mobile) Homes," Fire Technology, Vol. 29,No. 3, 1993, pp. 203-225.

24. Kung, H-C., Spaulding, R.D. and Hill, E.E.,Jr., "Field Evaluation of Residential Proto-type Sprinkler, Los Angeles Fire Test Pro-gram," Technical Report J.I. 0E0R3.RA(1),Factory Mutual Research Corporation, Nor-wood, MA, February 1982.

25. Bill, R.G., Jr. and Kung, H-C., "Evaluationof an Extended Coverage Sidewall Sprinklerand Smoke Detectors in a Hotel Occupancy,"Journal of Fire Protection Engineering,Vol. 1, No. 3, 1989, p. 77.

26. Vincent, B.G., Kung, H-C., Hill, E.E., Jr. andSymonds, A.P., "Sprinklered Office FireTests, " Technical Report J.I. 0M3N5.RA(1),Factory Mutual Research Corporation, Nor-wood, MA, March 1987.

27. Bill, R.G., Jr., Allard, R., Braga, A. and Fer-ron, R., "Standardized Extended CoverageSprinklers for Light Hazard Occupancies,"Journal of Fire Protection Engineering,Vol. 7, No. 4, 1995, pp. 133-151.

28. Bill, R.G., Jr., Kung, H-C., Brown, W.R., Hill,E.E., Jr. and Spaulding, R.D., "Predictingthe Suppression Capability of QuickResponse Sprinklers in a Light Hazard Sce-nario, Part I: Fire Growth and RequiredDelivered Density (RDD) Measurements,"Journal of Fire Protection Engineering,Vol. 3, No. 3, 1991, pp. 79-91.

29. "Installation of Sprinkler Systems in One-and Two- Family Dwellings and Manufac-tured Homes," NFPA 13D, National FireProtection Association, Quincy, MA, 1994.

30. "Installation of Sprinkler Systems,"NFPA 13, National Fire Protection Associa-tion, Quincy, MA, 1996.

31. Arvidson, M.A., "Cabin, Corridor, LuxuryCabin and Public Space Fire Tests for Grin-nell Fire Protection Using Aqua Mist AM6,AM11 and AM22 Nozzles," Test ReportR30608, Swedish National Testing andResearch Institute, Boras, Sweden, Octo-ber 1995.

32. "FMRC Draft Performance Requirements forWater Mist Systems for the Protection ofLight Hazard Occupancies," Factory MutualResearch Corporation, Norwood, MA,August 1997.

33. Arvidson, M., "Luxury Cabin, Open AreaPublic Space, and Corner Public Space FireTests for Grinnell Fire Protection usingAquaMist AM5 Nozzles," Test ReportR30455a, Swedish National Testing andResearch Institute, Boras, Sweden, June1994.