J. ofFireProt. Engr., 4(4),1992, pp 117-131

ANALYSIS OF THE HAPPYLAND SOCIAL CLUB FIREWITH HAZARD I

by

Richard W.Bukowski, P.E. and Robert C. SpetzlerBuilding and Fire Research Laboratory, NIST

Gaithersburg, MD 20899

SUMMARY

The paper presents the reconstruction of the Happyland Social Club fire using the HAZARD I firehazard assessment method along with an examination of four potential mitigation strategies:automatic sprinklers, a door at the base of the stairway to the second floor, a second means ofegress from the second floor, or a noncombustible interior finish. The paper concludes that thetraditional second means of egress might have not have eliminated the observed fatalities, andthat the noncombustible interior finish or sprinkler options would, with the former being the morecost effective approach.

BACKGROUND

In the early morning hours of 25 March1990, a tragic fire took the lives of 87 per­sons at a neighborhood club in the Bronx,New York. A ~ew days later, the New YorkCity Fire Department requested the assis­tance of the Center for Fire Research (CFR)in understanding the factors which contrib­uted to this high death toll and to developa strategy that might reduce the risk of asimilar occurrence in the many similar clubsoperating in the city. It was not the purposeof the CFR work to examine cause and ori­gin, nor criminal or negligent actions whichmayor may not have taken place before orafter the incident.

In responding to this request, CFR staffvisited the fire scene on 29 March to obtaininformation needed to perform an analysiswith the HAZARD I Fire Hazard Assess­mentMethodl• Physical measurements takenon-site, along with floorplan drawings andnewspaper accounts, provided the only dataon which this analysis was based. No mate­rial samples were taken, and no testing wasperformed.

This paper is a contribution of the NationalInstitute of Standards and Technology andis not subject to copyright.

THE BUILDING

The floorplan drawings obtained from theNew York City Fire Department are shownon the next page. Wall finish throughout theinterior of the building was 3/16-inch thickwood paneling on furring strips over plas­ter. The ceilings were low density fiber­board tiles in the first floor entry and bar,and gypsum board elsewhere. The fiberboardtiles were installed on furring strips underthe floor joists. Note the partial sprinklersystem on the second floor.

THE FIRE

From information reported in the media andobservations of the CFR staff dqring an on­site investigation, the ignition scenario forthis fire was as follows. One dollar's worth(about three quarters of a gallon) of gaso­line was poured on the floor of the entrywayand ignited2• The door from the entryway to

Author's note: This analysis was performed inthe periodjust after the fire in support of the NewYork City Fire Department's understanding ofthe event and how to deal with other similarestablishments operating in the city. This reportwas provided to the fire department at the time,but publication was withheld at the request ofthe District Attorney until criminal proceedingshad been completed.

- 117-

Bukowski, R. W. & Spetzler, R.C" Analysis of Happy/and Social Club Fire

58'

BAR

-5.5'

BarRoom

6'

13'

08& Y toilet valVft

First 3

Headnotsteps 19" fused

remainingDoorI(sheet rocklsteps 39-

opens Inwood sluds)wtde Sprlnkler breached by

head (fused)

!(~~;;.--4---6'~

~ISupply

line13'10" ¢Head1 fused

Danca54'Aoor

r I

Sprtnklerpiper-- ?Headfused

IS'

Open out steel door& roll-down gate

Exhaust'anGrated window, 1.5' x 2' wide

Floorplan drawings of the Happyland Social Club

the street was believed to be open. The doorfrom the entryway to the bar was believedto be closed. The fire quickly spread to thecombustible interior finish within the entryarea and, at some point someone opened thedoor from the bar to the entryway to go out3•This action opened a path for fire to spreadfrom the entryway into the first floor bararea.

At some point before or after this interiordoor was opened, an employee and a patronescaped3 through the service entrance whichwas reportedly c1osed4• It is unclear whetherthese doors were left open or closed andreopened after the fire department arrived.The fire department entered through thefront doors and quickly brought the fireunder control4• Of the seven sprinkler headson the second floor, four opened, two didnot, and one was missing at the time of thesite visit, but water stain patterns on theceiling indicated that it had been in placeand had activated during the fireS. Based ondamage observations made at the scene, thehollow core door at the base of the front

stairway was believed to be closed throughmost of the fire.

THE VICTIMS

Newspaper accounts indicate that 68 of the87 victims were recovered from the secondfloor where they succumbed to toxic smoke6.The remainder of the bodies were recoveredfrom the first floor (11from the rear restroom)6,each having some burns in additionto smokeinhalation. There were at least three survi­vors and possibly a few more3•

THE ApPROACH

Given the limited scope of the CFR involve­ment, the HAZARD Jl software was used todevelop an approximate reconstruction ofthe fire events. The details of the buildingarrangement and construction, and the ig­nition sequence described above, were en­tered into the HAZARD I routines. A seriesof computer runs was made examining thepossible ventilation conditions associatedwith which doors to the exterior might have

- 118-

been open, and when they might have beenopened. This is critical in understandingthe fire development since this fire quicklybecame ventilation controlled. A ventila­tion controlled fire is one in which the burn­ing rate becomes controlled by the avail­ability of air necessary for combustion ratherthan by the rate at which fuel is vaporized.Both the computer reconstruction ofthe fire

events and the limited extent of burning inthe building5 support this position. Thus,the burning rate was influenced more byventilation than by fuel characteristics (e.g.,type, quantity, and arrangement). The com­bination of door openings predicting resultswhich best match the observed damage con­ditions would be assumed to represent thelikely conditions during the fire.

The next step was the examination of pot en­tial mitigation strategies that could haveinfluenced the outcome ofthis fire. Initially,four possible approaches were identified:• Provide a complete automatic sprinkler

system.• Install a door at the base of the rear

stairs to prevent combustion productsfrom traveling to the second floor.

• Install a second means of egress fromthe second floor.

• Upgrade all interior finish materials tononcombustible materials that would notcontribute to the fire.

For each strategy, the ability to mitigatethe life loss was evaluated and the retrofitcost estimated. Cost is an important deter­minate of whether any strategy is realistic,since high costs may result in a reluctancebybuilding owners to complywith codemandates.

THE ANALYSIS

The room dimensions and estimated firecharacteristics for the interior finish mate­rials were entered into HAZARD I. TheMAKE FIRE routine in FPETOOL' was em­ployed to make estimates of the burningrate of the gasoline and the combustibleceiling and walls of the entryway, based onthe estimated surface areas involved. Thegeneral procedures ofthe chapter on calculational

J. ofFireProt. Engr., 4(4),1992, PP 117-131

procedures in the HAZARD I manual wereused. The resulting potential energy releaserate profile for the gasoline alone and incombination with the finish materials inthe entryway are shown as "GASOLINE FIRE"and "TOTAL POTENTIAL FIRE" in Figure2a. The difference between the potentialrate of heat release and the rates of heat

release predicted by HAZARD I for the cases

analyzed (except the gasoline fire) resultsfrom the impact of the limited air supply.The rate of heat release estimated by HAZ­ARD I for the case where the only combus­tible material was the gasoline is identicalto the potential rate of heat release of thegasoline. This was because the available airsupply was sufficient to freely burn all ofthe gasoline.

The execution of HAZARD I requires spe­cific input data. In addition to the potentialrate of heat release, the data listed in Table1 were used. The area of the gasoline fire isajudgment estimate as no specific data existon exactly how the reported gasoline pour2occurred. The value used represents a cir­cular pool about 3 feet in diameter. The areaof interior finish is approximately that burnedin the fire as estimated by CFR staff5. Theother values are generic values typical forgasoline and plywood or fiberboard interiorfinish materials. The 1% LOI is typical forpost-flashover fires.

The predicted conditions within the build­ing for the incident (labeled b.ase fire) andfor each mitigation strategy are presentedin the Figures 2 through 5. Data for a singlevariable (e.g., temperature, oxygen concen­tration) are presented in each figure con­sisting of a graph for each room with a curvefor each case examined. All graphs for agiven variable use the same scale for easycomparisons.

With these as input data, the FAST modelwas utilized to examine the room tempera­tures and gas concentrations predicted inthe various spaces within the club as a func­tion of whether and when the doors to theservice entrance and those to the entrywaywere open. The results of these calculations

- 119-

Bukowski, R. W. & Spelzler, R.C., Analysis of Happyland Social Club Fire

are presented in Table 1.

Service EntranceThe analysis assumes that the exterior ser­vice entry door and the door between theservice entry and the bar room were ini­tially closed. At some point early in the fire,the two uninjured survivors exited throughthese doors, probably leaving both open3•Since there were no automatic closers oneither door, it is assumed that they did notdelay to close the doors behind them in therush to escape. There was fire damage to theservice corridor and the exterior ofthe buildingabove that door5, so these two doors werecertainly open for some time during the fire.Since the volume of the service corridor issmall compared to the remaining volume ofthe building, the question of interest is whetherboth doors remained open after they wereused for escape or whether either was closed.

The calculations show that the fire develop­ment was much less sensitive to the posi­tion of these doors than to the doors to theentryway. The HAZARDI computations indicatethat the service entrance doors have a no­ticeable effect on increasing burning in thebar room but only a minor effect on burningin the entryway. Thus it was assumed thatthe service entrance doors were open throughoutthe incident.

EntrywayThe assumed sequence of events has theexterior door open throughout and the inte­rior door initially closed. At some time afterthe gasoline was ignited, a person openedthe interior door, saw the fire, and ran outthrough the flames (sustaining serious burns,but surviving) leaving the door open3,4. Thefire damage to the entryway was severe,and there was also fire damage to the build­ing exterior over this door.

Simulations with the exterior entryway doorclosed quickly showed the effects of firestarvation due to lack of oxygen, even if theservice entrance doors were both open. Sincethis is inconsistent with the physical evi­dence, the analyses were based on the casewhere the exterior entryway door was openthroughout the incident5•

The injured survivor confirmed that the innerdoor was initially closed3•4• But it is of in­terest to determine at what point in the firedevelopment he opened that door. It is as­sumed that if he ran through the entrywaywhile the gasoline was burning, he wouldhave picked up burning gasoline on his shoes,and would have received burns to his legs.He did not4. Thus, it was assumed that hemoved through the entryway after the gaso­line burned out-by our estimates, about 90sec.

Table 1. .Assumed Fire Characteristics at Peak Burning

.

Ga~nllnA

Flnlsh*

Area

0.75 m264.5 m2

Mass loss/unit area

0.045 kg/m2 -s..

0.03 kglm2 -s...

Heat of Combustion

44 MJ/kg+12 MJ/kg •••

CO/C02 Ratio

0.005 (.5)++0.005 (.5)++

C/C02 Ratio

1.0++1.0++

limiting Oxygen Index

1%++1%++

Values which were modified after flashover are shown in ( ).• Composite of 3/16-inch thick plywood wall panels and low density fiberboard ceiling material.•• See Reference 8.••• See Reference 9+See Reference 9. ++ See Reference 1.

- 120-

The FAST model was used to examine theimpact of the inner door being opened at 60see, 120 see, and 180 sec. The primary effectof this door being opened at different timesis to shift the point at which the fire's ef­fects began to spread into the building. Asdiscussed above, the door was probably notopened before 90 sec. If the door was openedas late as 180 see, the finish in the entrycould have begun to burn out. Thus, to bemore conservative, we assumed for calcula­tion purposes that the inner door was openedat about 120 see after the gasoline was ig­nited.

BASE FIRE SCENARIO

(HAZARD I CASE-BASE FIRE)

The simulations presented thus far wereused to identify the likely positions of theexterior doors, and therefore the ventila­tion conditions present in the building dur­ing the incident. On this basis the followingconditions were assumed for the balance ofthe study.• The exterior door to the entryway was

open throughout.• The door from the entryway to the bar

was initially closed, but was opened atabout 120 sec (after ignition of the gaso­line).

• Both doors to the service entrance wereopen throughout the incident.

Under these base conditions, occupants ofthe second floor room were predicted to havebeen physically incapacitated by toxicity (acombination of high carbon monoxide andlow oxygen in the presence of CO2 at about5 min. Note that this result is consistentwith the number of victims who died with­out apparent attempts to escape. At thattime, predicted temperatures on the secondfloor were still low. Further, as has beendemonstrated experimentallyll the opera­tion of a sprinkler at the top of a stairswould cool the gases even further, but wouldhave no effect on the CO, CO2, or oxygenconcentrations. Prediction of such effects ofsprinklers is beyond the current capabili­ties of the models used in this analysis.

J. ofFireProt. Engr., 4(4),1992, pp 117-131

It should be noted that the peak upper layertemperatures shown in Figure la are toohigh. Such temperatures reported from ex­periments rarely exceed lOOO·C. Zone mod­els are known to overpredict upper layertemperatures and underpredict lower layertemperatures, because not all mechanismsof energy exchange between layers are in­cluded. FAST is no exception, with suchoverprediction having been estimated at 30­50% under most conditions and higher un­der severe conditions. Since these overpredictionsoccur after the times of death, they havelittle effect on hazard-to-life prediction.

MITIGATION STRATEGIES

Once the base scenario was derived, thepoten tial benefits of the four, previouslyidentified mitigation strategies could beexamined. These are presented in the fol­lowing sections.

Automatic SprinklersWhile the building had a partial sprinklersystem on the second floor, it played only aminor role in this fire since it was so farremoved from the actual fire. If the buildinghad been protected by a complete (operational)sprinkler system, the fire likely would havebeen extinguished in the entryway. While itis not currently possible to demonstrate thiswith the models, there are test data 12 for avery similar condition, a gasoline spill in theentryway to an apartment, ionwhich rapidextinguishment was achieved. There are routinesin FPETOOU that estimate the activationtime of sprinkler heads. The FPETOOL rou­tine FIRE SIMULATOR indicates that a tra­ditional sprinkler head in the entryway wouldhave activated in about 7 see following igni­tion of the gasoline.

Door at Base of Rear Stairway(HAZARD I Case-Rear Stairway Door)The principal avenue of smoke/gas travelimpacting the second floor occupants wasthe rear stairway. If this stair had beencutoff by the addition of a door with anautomatic closer at the bottom, this pathwould have been reduced to whatever leak­age might have existed around that door.

- 121 -

Bukowski, R. W. & Spelzler, R.C., Analysis of Happyland Social Club Fire

:l t~~ .• I:~ 1_ BASE FIRE

•••••••••• GASaJNe FIRE

ISlO ISlO ou" _STAll\WAYOOOR- - - - FIRE ESCAI'E_. _. _ ENCl.OSED EXIT STAIRWEU.

1400 1400~ G~ 1~ W 1~

~ ~m~ ~~~ /IX) __ •• ~":: FIRE ~ /IX)

SlO _ ::-.::: ~E=AYDOOR SlO•••• \ ..•.•.• ~ - ENClOSeD exrr STAIRWElL

400i f \ I 400

~]'/ ~ .., d ~........... -

, --.------------------------o ..... 01 - - ---- -----o 100 ~ 300 400 500 SlO 0 100 aX, :m ..00 500 SlO

~~ ~~a ErtyWa/ b.SeM:l9Enra1ce

~ ~

1/1X) __ RAE 1/1X)1 _ BASERAE

- ClASOUNEFlAE -- 0AS0lINE RAE

lSlO u_u IlEARcrAIIWAYDOOR ISlO ouu IlEARSTAlRWAYDOOR- -- - RAE ESCAl'E - - - - RAE ESCAPE-. - ••• ENClOSeD EXIT STAIRWElL -. -. - ENCl.OSED EXIT STAlRWaL

1400 1400

~ ~~ ~ ~ 1~

~I1~ i 1~/IX) , m /IX)J ....

~\ J'---- ~~ :~

SlO t:::. _:s:..••......_ SlO,; ~

«Xl1 cL ] °1 :j-~==- .

~ _-,~ ~

I .. ': . n--:-- 0100 ~ 300 «Xl 500 SlO 0 100 ~ 300 «Xl 500 SlO

l1ME(5) TIME(5)C. 1st FloorBar d.Bad<StaiIwaIl

~] ~1/1X) I 1/1X)BASEFlAEClASOUNE FlAE

ISlOi I:~-~~=,,~AYDOOR I ISlOENClOSeD EXIT STAIAWEU.

1«Xl 1«Xl

~ ~w~ ~~~ ~

J~ I~m /IX) m /IX)SlO SlO

«Xl «Xl

~l ,-=-- ~ ~r .v· ---------- 000 100 ~ :m ..00 !in SlO 0 100 ~ 300 «Xl 500 SlO

l1ME~ l1ME~e.2rd Floor f. ExitStairwell

Figure 1. Predicted upper layer temperatures.

- 122-

J. ofFireProt. Engr., 4(4),1992, PP 117-131

14OOl1

I\ I1400l

I\I \

1:ml1

I \_ BASEFIU:

l:mll,- BASe FlAf

__ ._. GASOlINE FIRE I \_uu RENlSTAlAWAYOOOA __ H. GASOlINE FlRe

I~ ___ FlRe ESCAPE .u •• RENlSTAlRWAYOOOA, _ ••• _ ENClOSEDEXITSTAlRWaL - - - - FIRE ESCAPE

~ lOOXl

I - _ - POlENTW. FIRE

~

-. -. - ENClOSED EXIT STAlRWaL, lOOXlI ,~

I \

~1m)

I \ 1m)w

I\.--- Poknial Fre

~~

I \I \8XlO I \ 8XlO

•wa: I• a:

~ 400l

I~ iA ~:z:

I G/:z:

400l

:ml 1 '-l.I:ml

--\.---GaooreFieo I'

,:...

..I 001002)):m0400500llll 01002)):m0400500llll

aErty~

llME(s)llME(s)

b. 1st Floor Bar

1400l

1400l

1:ml

IlASEFlRe

'-1§1FIRE

--- ClASOUNEFIAE

• GASOlINE FIRE•••••• AEARSfAIFtWAYDOOR

.u •• AEARSTAlAWAYOOOA- - - - FlRe ESCAPE

- - - - ARE ESCAPE- • ~ • - ENClOSED EXIT STAIRWEll.

-. -. - ENClOSED EXIT STAIRWEU.

~lOOXl ~lOOXli 1m)

i1m)I 8XlO

~8XlO

a:~ 4000

~400l

X X

:ml

:ml

.', :.1'"'01

.... \ ..I 00

1002)):m0400500llll 01002)):m0400500llll

C. Back Staiwefl

llME(S)

d2ndFloor

llME(s)

1400l

Figure 2. Predicted energy release rates.

1:ml

2lXXl

oo 100

a.Outside

:m 0400

llME(s)

500

- 123-

Bukowski. R. W. & Spetzler, R.C., Analysis of Happyland Social Club Fire

6 6

5BASe FIREClA90llNe RAEREAR STAllWAY DOORFIRE ESCAI'EENClOSED EXIT STAIRWEll.

5BASe RAEGASDlJNE FIREREAR STAllWAY DOORFmEESCAI'EENClOSED EXIT STNI\WEU.

'~ •.......•.."'~ ....''''- ..-'- .

3

4

i21; ••••••••----.--.-------- •• ----.-.-I!/-..I

3

4

b. SeMoe Enlranoe

oo 100

aErtryWay

3D

TIME(s)

<lOO 500 8Xlo

o 100 3Xl 3D

11ME(s)

<lOO 500 8Xl

--1 GASDUNE FIRE••••• _sr_YDODR- - - - FIAE EllCAl'E:_. _ •• ENClOSED EXIT srAllWEll.

_.:..:.":.-:.":.-:-::.-:.-:.~:-::::::::::=:

2

6

_RAEGASDUNE _llEARsr_yDODRFIAE EllCAl'EENClOSED EXIT srAllWEll.

3

4

5

., .."' __ ••.__ •.••••~t': ••=-.....,... --- ..-.- ..-

6

e. Secxlrd Fbof

Figure 3. Predicted layer height.

8Xl5003D <lOO

nME(s)f. Exit StaiI'lYell

oo 100 3D

00

1003Xl:m<lOO5008Xl

TIME(s)d. Back Stairwell

65

S e

4

:z: ~

3

t2:;)

8Xl

8Xl

500

500

0:l

_RAEGASDUNE FIREllEARsr_yDODRFIRE EllCAl'EENClOSED EXIT srAlRWEu.

3D

TlME(s)

3D <lOO

TIME(s)

100

3

4

Co 1st Floor Bar

5

6

oo 100 3Xl

- 124-

J. of Fire Prot. Engr., 4 (4), 1992, pp 117-131

25 25

:m11ME(s)

c. 1st Floor Bar

5

oo 1m

10

15

,;::::=---­l.V - --d •••••••••

/.,,/1.1

I.'ItI', I

:mTIME(s)

1m 3Xlo

o

5

15

10

:mTlME(s)

d Back SlaiIWell

5

oo 1m

10

15

:m11ME(s)

,~,,-==-',t~:",~, - ., , ,*,l ",/ Il.' "I ", ,

...••_-'~.##"

-------------­...­..'•.-

100 3Xl

.','.'.'.'-

oo

5

15

10

25

25 25

15

10

58ASEARe0ASClUNE AReREAR SfNlfWAY 0CX)ftFlREESCAPEEHClDSED EXIT STAIRWElL

15

10

5

oo 100

e.2rdFbor

:m <4lXI

11ME(s)

oo 1m 3Xl

f. Exit S1aiIwelI

:m <4lXI

TlME(s)

Figure 4. Predicted oxygen concentrations.

- 125-

Bukowski, R. W. & Spetzler, R.C., Analysis of Happyland Social Club Fire

500

'." ,\,," '.'llo ••••••••••

IlASEFIlE

GASWEFIlEAEARSTAllWAYDOORFIlE ESCAPEENCLOSEDEXITSTAlRWEU.

3J:l

TIME(s)

100 llXl

b. Service EnIJance

oo

•.-.•......

5003J:l

T1ME(s)

'­--I/,,\, :. \

\

'-----------.-------

aEnlyWay

oo 100

lASE_0AS0l.INEFIREAEAIIsr_AYDOOR_ ESCAI'EEHCt.OSEOElaT srAmWEU

I::":'=':

3J:l

TlME(s)d Back StailweII

oo 100 llXl5003J:l

T1ME(s)

,,,,,--, -.,'\ '~\

# \. ,'--'.-'.' \ \. ,. \\ \.,

',9\ ' ...•..''' ."\\v,'~----,1= ~_

,••••• flEMtIfAI/IINAYDOOR_ EllCAP£ENClOSEOElCITsr~

Co 1st FloorSa-

oo 100

lASE FIRE0AS0l.INE RAEI'IENIsr_AY DOOR_ESCAPEENClOSEOElaT sr~

ii

5003J:l

TIME(s)f. Exit Stairwell

oo 1005003J:l «n

T1ME(s)

- 126 ~

e.2rdFloor

oo 100

Figure 5. Predicted carbon monoxide concentrations.

However, this solution would have requiredthat the occupants "ride out the fire," andthat the fire department could respond andextinguish it before any structural failure.

To examine this strategy, a (closed) solidwood door at the first floor entry to the stairwas assumed. An effective crack width of 1.3

cm (0.5 in) was assumed over the height of the

door to represent the gap to the frame. Thesimulation run for these conditions showed a

reduction in the predicted temperatures withinthe rear stairway and the second floor (Figure1, d and e), as expected. The temperature ofthe first floor bar (Figure lc) was also lower,as less flow up the stairs reduced the venti­lation rate through the front door reducingthe oxygen (Figure 4, a and c) and the burningrate (Figure 2, a and c) in both the entry andthe bar. The CO levels upstairs (Figure 5e)were significantly lower, and the levels on thefirst floor (Figure 5c) were reduced due to thelower burning rate.

Second Exit (HAZARD I Cases-Fire Escape and Enclosed Exit Stairwell)The third mitigation strategy involved theaddition of an independent, second exit fromthe upper floor. This would likely be locatedin the front (street) side as either an exte­rior, open fire escape with a drop down lad­der, or as an interior, enclosed exit stairwaydescending in the service entry area. Thedifference between these two is more thanjust the additional cost of the latter. Open­ing a second floor door to a fire escape wouldhave changed the ventilation pattern in thebuilding, possibly resulting in a significantlyincreased flow up the rear stairs to thesecond floor, making occupant survivabilityworse. Thus, both arrangements were ex­amined with the model. The results of thepredictions showed similar results with re­spect to each of the examined parameters.In both cases, opening of the egress door onthe second floor results in increased flow upthe back stairs with an attendant increasein the temperature (Figure Ie) and CO lev­els (Figure 5e), and decrease in the oxygenon the second floor (Figure 4e). However,these effects were somewhat greater in thecase of the fire escape than for the enclosed

J. ofFireProt. Engr., 4(4),1992. pp 117-131

exit stair since its direct opening to theoutdoors allows more flow.

The FPETOOL routine EGRESS estimatesthat it would take between 2 and 5 minutesto evacuate the 87 second floor occupantsusing a single exit, depending on the widthofthe exit stairs. The 2 min evacuation time

is for a 44 in wide stair; the 5 min time is for22 in wide stairs. These are the minimum

widths permitted by the Life Safety Code13

for stairways and fire escapes, respectively.The amount of delay between the time offireinitiation and when most of the occupantswould start to evacuate then becomes a cru­cial unknown. To reach safety the occupantswould have to be able to egress before beingincapacitated at about 290 seconds (by tox­icity). That is within about 3 min after theinner door of the entryway is opened. A narrowfire escape would have likely resulted in aconstriction at the top tread that would haveprevented some of the occupants from es­caping even if evacuation started as soon asthe fire burst into the bar room. Wider stairswould improve the situation, but the degreeof success would depend on the speed ofstartupof evacuation. In the case of a 44 in widestair, the HAZARD I predicted environmentand the EGRESS prediction of evacuationtimes indicate that it would have been nec­essary for the occupants to start evacuationwithin about 3 minutes of ignition (1 minuteafter the inner door was opened).

If the additional means of egress were com­bined with the door for the back stairs,sufficient egress time would be available forall occupants, but notification to begin evacuationbecomes an even more important issue inthe absence of an automatic fire alarm sys­tem. However, it could be argued that para­graph 93.1, exception 1 of the Life SafetyCode would not require the protected verti­cal opening with sufficient egress capacityfrom the second floor, so this combinationwas not examined.

Noncombustible Interior Finish(HAZARD I Case-Gasoline Fire)The last strategy involves the replacementof the combustible ceiling and walls with

- 127 -

Bukowski, R. W. & Spetzler, R.C., Analysis of Happyland Social Club Fire

noncombustible interior finish materials.This would have the effect of limiting thefire to the original gasoline spill. Applyingthe same assumption that the inner doorfrom the entryway was closed until after thegasoline burns out, then the conditions inthe rest of the building remain at ambientthroughout the incident. Thus, we exam­ined this case as if this door were open from

the point of ignition, resulting in the predic­tion of elevated temperatures in the build­ing during the first 120 sec not present inthe oUier scenarios due to the closed doorbetween the entryway and bar.

When the noncombustible finish scenariowas run in HAZARD I, the results indicatedthat temperatures (Figure 1), CO (Figure5), and oxygen concentrations (Figure 4)remained tolerable throughout the fire. Onlya person in the entryway would have expe­rienced lethal conditions, from burns. Withboth doors to the entryway open, the gaso­line fire had sufficient air to burn out with­out producing flames out either door, soignition of furniture in the bar would beunlikely. But if the fire had involved thecontents of the bar (e.g., if the fire wasstarted in the bar), some life loss would belikely, even with noncombustible finish.

Cost EstimatesRough estimates of the retrofit costs of eachof these mitigation strategies were madeusing typical construction cost handbooksto assist in evaluating the degree to whicheach would be realistic to enforce on suchclubs. The estimates developed 14 were:• Automatic sprinklers: $8000• Door for rear stairway: $300• Second exit: $2000 (fire escape), $6000

(enclosed stair)• Non-combustible interior finish $3000

SUMMARY OF OCCUPANT IMPACTS

Once all of the simulation runs were com­pleted, the TENAB model from HAZARD Iwas used to evaluate the time to incapaci­tation and lethality for each room in thehuildingfor each condition examined. Thesepredicted results are presented in Table 2.

A complete explanation of the calculationson which these predictions were made iscontained in the HAZARD I documentation1•

These represent estimates of time to physi­cal incapacitation (or death) for a normaladult within, and not moving from, the in­dicated room from the time of ignition of thefire in the entryway. The Purser models15

are based on incapacitation experiments with

monkeys and do not attempt to predict le­thality. The NIST model is based on lethal­ity experiments with rats, and estimateincapacitation. Persons who are physicallyincapacitated will remain stationary andwill die unless rescued before lethal condi­tions are reached.

CONCLUSIONS

While we believe that the descriptions ofthe course of the fire provided by HAZARDI are representative of the actual conditionswhich occurred, it is important that they beverified to the maximum extent possibleagainst all data obtained in the investiga­tion, including witness statements. In thecourse of this analysis, the authors had noaccess to official reports or statements ofwitnesses, nor were any samples taken orphysical tests run. The primary informationsource was articles in the press.

Based on the analysis reported herein, thefollowing conclusions have beel} reached.

1. HAZARD I predicted conditions quitesimilar to those reported for the actual in­cident, including times to and cause of deathfor the building occupants consistent withobservations.

2. Calculations indicate that an automaticsprinkler head in the entryway would havepromptly actuated and most likely preventedthe spread of lethal conditions from theentryway into the rest of the club.

3. A second means of egress might havereduced the toll, but probably would nothave eliminated all of the fatalities. Thedegree of success would be a function of the

- 128-

Table 2.Summary of Occupant Tenability

J. of Fire Pro!. Engr., 4 (4), 1992, pp 117-131

Time (see) to Base CaseNon-Door on RearFireExit

Incapacitation

Firecombust!InteriorEscapeStairway(death)

IbleFinishStairway

EntrywayFLUX

3030303030

TEMP2

3040303030FED2

40NR404040FED1

60 (100)3060 (100)60(100)60 (100)

Service Entr.FLUX

160NR160160160TEMP2

210NR240NRNRFED2

290NR240350380FED1

NRNRNRNRNR

15t Floor BarFLUX

140NR140140140TEMP2

200NR170200180FED2

200NR170210210FED1

280 (420)NR200 (230)270 (290)290 (320)

Back StairflUX

200NRNR200200TEMP2

NRNRNR270270FED2

270NRNR260270FED1

350 (480)NRNRNR300 (NR)

2nd FloorFED2

320NRNR290300FED1

390 (520)NRNR340 (460)360 (460).Exit StairFED2

NANANANA340FED1

NANANANA420 (590)

FED1 is Fractional Effective Dose based on NIST toxicity model.

FED2 is Fractional Effective Dose based on Purser toxicity model.TEMP2 is limit for convected temperature based on Purser model for incapacitation.FLUX is the threshold of 2nd degree burns to exposed skin causing incapacitation.NR = limiting condition not reached.NA = not applicable since the exit stairway was not present in other than exit stairway cases.FLUX and TEMP2 were not applied to the second floor nor the exit stairwell because of theoperating sprinkler at the top of the back stairs, which was predicted to activate at 240 sec.

- 129-

Bukowski, R. W. & Spelz/er, R.C., Analysis of Happy/and Social Club Fire

speed ofrecognition ofthe danger and promptnessof the start of evacuation; and is highlycoupled to the width of the stairs.

4. Protecting the back stairs would haveprovided additional safe time for occupantsof the second floor, but the structural integ­rity of the building would then become acrucial factor.

5. Noncombustible interior finish appearsto be the least costly strategy for limitingthe life loss in this incident. By limiting thefuel available to the gasoline spill, the im­pact of the fire on the building occupantswould have been minimized. However evenwith noncombustible finish, if the fire hadinvolved the contents of the first floor bar,some life loss would have been likely.

REFERENCES

1. Bukowski, R.W., Peacock, R.D., Jones,W.W., and Forney, C.L., HAZARD I FireHazardAssessment Method, NIST Handbook146, National Institute of Standards andTechnology, Gaithersburg, MD, 1989.

2. "Portrait of Suspect in Social Club BlazeEmerges," New York Times, 26 March1990.

3. "No Escape in Bronx," New York Times,26 March 1990.

4. Personal Communications, Hodgens, J.J., Assistant Bureau Chief, N.Y. CityFire Department.

5. Observed by CFR staff during onsite visit,30 March 1990.

6. "Dream Turns into Nightmare," USA To­day, 26 March 1990.

7. Nelson, H.E., FPETOOL, Fire Protec­tion Tools for Hazard Estimation, NISTIR4380, National Institute of Standardsand Technology, Gaithersburg, MD, 1990.

8. SFPE Handbook of Fire Protection Engi­neering, Figure 21.2(a), 1989, page 2-3.

9. Unpublished NIST data from Cone Calo­rimeter.

10. SFPE Handbook of Fire Protection Engi­neering, Table 1-10.3, 1989, page 1-149.

11. Cooper, L.Y. and O'Neill, J.G., Fire Testsof Stairwell Sprinkler Systems, NBSIR812202, National Institute of Standardsand Technology, Gaithersburg. MD, 1981

12. DiCicco, P., "Fullscale Fire Tests on Row­frame Residential Buildings," Symposiumon Fullscale Fire Tests in Research andDevelopment, Armstrong World Indus­tries, Lancaster, PA., 1974.

13. Life Safety Code, NFPA Standard 101­1988, National Fire Protection Associa­tion, Quincy, MA, 1988.

14. Dodge Manual for Building Construc­tion Pricing and Scheduling, Dodge BuildingCost Services, McGraw-Hill Publishers,New York, NY, 1979.

15. Purser, D. A., "Toxicity Assessment ofCombustion Products," SFPE Handbookof Fire Protection Engineering, NationalFire Protection Association, Quincy, MA

-130 -