Sponsored by -----

This Advanced Research Workshop was organized by:

GIDAI - Fire Safety- Researc/1 and Tcclmofogy UNIVERSIDAD DE CANTABRIA Dpto. de Transportes y Tecnologia de Proyectos y Procesos Avda. Los Castros, sin 39005 Santander. Spain Tf. + 34 942 201826. Fax. +34 942 202276; gidai@unican.es; hllp:llgrupos.unican.es/GIDAI

With the collaboration of: Society ol Fire Protection Engineers National Fire Protection Association

SFPE NFPA

Scientific Committee· Editorial Board:

Jooge A. Capote ......................... ..

Daniel Alvear . ......... ......... ......... ..

Richard Carve! ......... ............. .

Michacl Detichatsios .. ............. ..

Carlos Fem~ndez-Pello .............. ..

Charles M. Fleischmann ............. ..

George Hadjisophacleous ........... .

Morgan J. Hurley ....................... ..

Time Korhonen ........................... .

James A. Milke .... ................. ..

Frederick W. Mowrcr ............. ..

Pau!o Piloto ......................... . .. .

David Purser .............................. ..

James G. Ouintiere .................... ..

Guillarmo Rein ...... .... ............. .

Jose L. Torero ........ ........... .. ..

Arnaud Trouve ..... .. ..... ......... .. .

Sponsored by:

University of Cantabria (Spain)

University or Can!abria (Spain)

University ol Edinburyh (UI<)

University or Ulster (UK)

University ol California, Berkelsy (USA)

University of Canterbury (N. Zealand)

University of Carlelon (Canada)

SFPE(USA)

VTT (Finland)

University ol lvlaryland (USA)

University of Maryland (USA)

Institute Pol. de Bragan~a (Portugal)

Hartlord Environment Research (UK)

University ol Maryland (USA)

University of Edinburgh (UI<)

University ol Edinburgh (UK)

University of Maryland (USA)

International Association lor Fire Safety Science tAFSS

1,11NISTERIO

DE CIENCIA c INNOVACION

Convocaloria de Ayudas para la reahzacion de Proyectos de lnvestigacion, Programas de Actividad lnvestigadora y Acciones Complementarias

Ref.: BIA2009-05701·E

Contents

Index .... ..... ................. ..... ... .......................... ........................ .................. .......... ..... ............. .

Pn:facc ..... .. ................ .... ....................... ...... ... .... .. ....... .... .. ...... ............... .. ............. .. ..... ..... .. ii i

Invited Lecture: Prescribing the Input for the Aset versus Rset Analysis: is this the

way forward for Performance Based Design'! ........................ ........................................... .

Application of Human Behuviour and Toxic Hazard Analysis to the Va lidation of

CFD Modelling for the Mont Blanc Tunnel Fire Incident............ ..... ...... .. ... ....... ... .. ... .... .. 23

Fire Behaviour and Fa<;:ude Flames for Corridor and Tunne l Like Enclosures Fires.... ..... 59

Analysis of Eleven Evacuation Ev~nts in Finland................. ................................ .. ............ 75

Experiments to lnvcstig;1tc Radiant Heat Flux on Adjacent Buildings............ ...... .. ........... R7

Them1al Analysis in Fire-Resistance Furnace.......................... .. .......... .. .................... ....... 103

Low and Medium Power Full-Scale Atrium Fire Tests and Numerical Validation or

FDS ................... ....... ...... .. ...... .................................................. .... .. .... ....... .......................... 113

Landing Distance of Droplets from Water I'VIist Suppress ion Systems in Tunnels with

Longitudinal Vent ilation ........ .. .... ............................. ........................................... :................ 139

!Vlodelo Estocastico de Evacuacit'ln pur:J Trenes de P:1sajeros ....... .. ... .. ...... ................... ... . 155

Influence of the Bench Scale Test in the Calcu lation of the Heat Release Rate fur

Ai rcraft Materials: Cone Ca lorimeter vs. OSU Appnmtus . ............. ................................... . 185

Efkctivcncss Assessment of Road Tuuncl Fi re-Fighting Stmtcgies by Ventilation and

\Vater Mist Systems .. .. ...... ... ............................ .................... ........................ ......... .. ..... ......... . 197

Can Active Fire Protection Systems in Tunnels Prevent Minor Fire Incidents from

becoming Disasters'? 22 1

ADVANCED RESEARCII WORI\SIIOI'

FIR£ f'ROT£CTIUX :IND LIFI:" S:IFETl' IN IIUILDJ,\'GS :IND TIUNSf'OJI1:·1TJOi\' Sl'STE:IIS

Numerical Simulations of' Some Possible Fin: Scenarios in a Closed Car Park with

RANS and LES ......................................................... .. .................... ............. ...................... .. 233

Modelling or pcdcslrian 1110\'Cillelll around 90° and I ~0° bends............... .......... ............... 243

Experimental Rcst:arch - Large-Scale.: Tunm:l Fire Tcs1s and the use or CFD

modelling to predict Thermal Behaviour....... .... .......................................................... ...... . 255

An Experimental Revic\\' or the Homogeneous TemperaLure Assumplion in Post­

Flasho\·cr Compartment Fires.. .......................................................... ... ............................... 273

Psicosociologia aplicada a Equipus de lntcrn:ncion en Situacioncs dt: Emcrgencia. ....... 295

Estabilidad antt.: cl Fuego de la Estructma Mct;'dic<~ de Cubicna dd Edilicio lntcrmodal

tk la Nucva Tcnnin:li T I ut:! Aeropuerto de Barcclon:J ..................................... .... .... ........ 307

The Regulatory Use of Advanced Firt: Engineering Tcclmiqucs....................................... 323

Human Behuvinur in case of Fire inside a Urban Tunm:l th rough Computcr Modelling ... 3-19

Comp:1rison of STEPS and FDS+EVAC Simulations tu an Evacuation of a 5 Storey

Building............ ..................... ................. ... ....... .. ............. ..... ................ .. ................ .... .. ........ 363

An lmmersivc Simulation of Fire Evacuation Based on Virtual Reality. .................. ........ 3!l5

The Smoke: Layer lmerl;1ce tluring a Fire in an Atrium: a New Method to Locate it

using a Zone Computer lVI ode I.......................................................... .... .............................. 31J7

Thermal Analysis in Fire-Resistance Furnace

Pilr11o. P..-I.G. 1, Mc!Stjllila, L./lf.R. 1

: Ale.mmfre Pereiro: 1- Applied Mech. Dep .. Po{r!eclmic Ins t. r~j"Bragan~·a. Campus Sra . .-lpoldnia.

5301-85 i Braganra. Portugal

:- Fr!lloll' research. Po{l'leclmic Inst. of Braganra. Campus Sw . ..lpo16nin,

5301-857 Bragml('a. Portugal

ABSTRACT

Fire resistance rating or building construction elements is defined under fire-resistance test

rum ace. The geometry and shape of fire-resistance furnaces is not defined by any prescriptive

document. being necessary to comply them1nlly with speci lied nominal lire curves, such as ISO :)34 or hydrocarbon [ 1,2).

This research work intends to measure temperatures inside furnnce vo lume. using sixteen

plate thermocouples to compare average temperature in four planes. Those planes are

compared with re ference thermocouple which is responsible for controlling li.1mace operation,

sec figure I. Three tests were pcrfonncd, the firs t two running with ISO S34, tluring 45

minutes and the last one running with hydrocarbon curve, during 30 minutes. Experimentnl

results demonstrate th:ll relative temperature diflcn:nccs arc smaller than 30 % in the initial

test stage, being smaller than 5 %. after 500 [s] unti l the end of the tests.

The numerical simulations were performed using f'luent CfD. using the structuretl finite

volume mesh method. The Eddy Dissipation Model (EDM) was used for chemical species

transport and reacting flow. The governing equations for nwss, momentum and energy were

solved for the three dimensional unsteady incompressible flow, with radiativc heat tr;msfcr

and turbulence model. The numerical results agree well with experimental resulls. being the

relative temperature difference smaller than 5% for each nominal test. Numericnl simulation

also revea ls the localized effect of each burner.

INTRODUCTION

The thermal perfonnam:e of firc.:-rcsi stance furnace is investigatetl. Furnace environment is

normally considered homogeneous and with uniform tcmperuture distribution. following

specified nominal temperatur~ evolution. Tests were conducted at the Polytechnic Inst itute of

Bragan~a laboratory ( LERM) using the I cubic meter fire-res istanct! fu rnace. This small

furnace is suitable for initial vn lidation of experimental temperature measurements. The

furnace has 4 propane burners wi th 90 [kW] max imum power cnch, reference Kromschroder

BIO 65 HM-I 00/35-72/8. This furnace complies with European Standard EN 1363-1.

considering th~ maximum deviation between reference temperature and the theoreticn l specified nominal curve for 1emperature.

IOJ

All\'ANCED RESEAnCII WOHKS BOP IO·I FIRE f'ROTECT/0.\' .-IND LIFE S.IFETr IN IJU/LlJ!NGS .-IND TR.·I NSI'OIIT.·I T/ON .\TSTD!S

Four planes were delined to evaluate the furnace temperature performance. Four plait:

thcnnocouplcs define eat.:h plant:. Plant: X 1.50 and XS50 an~ parallel to tlw exhaust zone whi le

plane Z !50 and Z1150 arc parallel to the transvcrsnl plane of burners.

a/ Funh;c,•mmld wuhf mw hunh'rl Ill 11• },.JI bJ Gt•om, •Jrh J'Hl\11/ I•H/iu· plat .. · cl Fuman• lt'/1/t t'i'~'IIL'll doo r uJia tlro:mt11'Ct111plc•., II!HdL•ji,.nm· t·. nm11in~: h 'H J

Fig. / . ModeljiwfirL'-r~.ri.l'fm/c~.fimwce.

Three tes ts were performed. The first two tests used ISOS34 nominnl fi re curve, defined by

equation (I). while the last one used hydrocarbon nominnl fire curve (2). In these equations.

0" represents n:fcrencc furnace temperature for time 1, during testing conditions.

0, = 20 + 3~5 x logw(Hx r + l ) [O, ("Cl: r(minl] ( IJ

(2)

The thermal performance is dt:tt:nnined by the comparison between average plane

temperntures with reference llmwct.: tt:mpcraturc_ which represents ench nominal fi re curve.

Numerical s imulat ions arc also performed. using computntionnl flui d dynamics, to evaluate

thermal performance and validate experimental tests.

2 EXPERIMENTAL TESTS

The !ire-resistance litrnace was instrumented with 16 pla te thermocouples, suspended at

insulated slcndemcss steel frame. see figure 2. Plate thermocouples fulfi l standards and were

positioned according to the vertex posit ion of'700 [mm2] cube, centcred ins ide furnace.

.....

T/IEII.IIAL .·JNAL J'S/S IN F/RE-!IESIST·Ji\"CE FURNACE

f'iloto. P.A. G .. ;1/estfllil<l. l..MJI., Paeira, ..J. 105

!~) R~·MtkL· pmilioll/or tll lfc' tllt'fiiiiK flltJliL'J M PI. Ut.' tlla ii:OWIIJ1h' ( l_l]Jt' K cmd .fUli1Jio1 .\ tcd p f11lt.' .IJO.JJ.

Fig. ~- /'late tlu .. 'nuocouple in'itrUIJJL'Illalicm.

Thermocouple control is achieved by matching the measured thennocouple temperature with

the prescribed nominal fi re curve. However, since thennocouples irradiate heat, they adjust

themselves to the temperature at which there is a balance between the convection and net

radimivc heat transfer, [3] . Plate thermocouple is a stainless steel plate with lOO [mm2] and

0.7 ± 0. I [mm] thick. The emissivity was considered greater than 0. 7. The wire thermocouple

is positioned on the back face. screwed with a small plate and insulated with ceramic fibre ,

[1).

Measurements were pcrfom1ed with multi-channel datu acquisition system, MGCplus from

!·IBM. with frequency ~:qual to 0.1 [Hz]. Results were averaged from 4 plate thermocouples

for each plane. Plane XS50 was defined by the average temperature readings on Tf I, Tf2,

BF I, BF2, Plane X I 50 was de lined by the average temperature readings on Tb I, Tb2, Bb I,

Bb2, Plane Z850 was defined by the average temperature readings on TFL, BfL. TbL, BbL

while Plane Z 150 was defined by the uvcrage temperature rcildings on TbR, TFR. BbR and

BfR.

figure 3 represents each me;Jsured temperature with plate thermocouple and the reference

temperature for furnace.

The relative di fference between each average pl<mc temperature and reference tempc:rature

from furnace is below 5%, after the initial testing phase. corresponding to 500 [s] . H ighcr

relative difTerencc is expected before. due to higher temperature variation with Lime.

1ou ADYANCED RESEARCH WORii:SHOI'

FIRE 1'/IOTECTION :IND LIFE S:IFETr IN BUILDINGS :!ND TRANSPORT.·IT/ON Sl'STE:I!S

The exhaust temperature was also measured f(Jr test l . There was a constant temperature

difference between the reference furnace temperature and the temperature measured at the

cxlwusts. This temperature was measured with an insulated welded thennocoup le without

plate an insulation protection. This explains the oscillating registry.

Trmp.:n l;:rri'CI

f HI - rrr.

1: , ~= i ..., I

•« l :., I ;, I :oo! I«

- N>I -!l!.P.

Ti~ -ur1 m: - Ufl - mH -Ibl -nb: - nL -nR -ftll'i.\([ - L\H\ l!"ll

0 ·--------------------------~----~

""

Ttr:lft,:t!turiTJ

-HI -lfH -!U:l

'""

- rr. - UJL -n~!

--IU~! - FWtS.\LT.

"'' I ~{I()

l imr ]1]

-nr. - m - nn

cJ TL'.l l J n·lfh ISO .'i].J mmrir111l curw.

l~ m?"'ull:rr]'C]

- 111 - 11: - ilil -TIH -na -IHR

t -IJ'J!___=l!t-~------ltL

"""I lunv 1

I

'""

: oo

- TII. -n-~ - IJt.l

;:m

- H L -H·~

mt

/1) T.:.s r I wit/: ISO SJ-1 mm:uw/ ntn·c.

UththcTtrnjlfUILur lllrf~rcnn·

• :> :===============:=:;

.,

""" 1 ~\JJ Tln1r ]1]

tlJ TL'.\'1 ~ h'itiJ IS() SJ.J l!fltltll1111 CUI,'t.'.

Timtl •l

J} li.:.H 3 waf1 /~nlrjiCIIrhlm 1111111111al , ·an 'l-'.

Fig. 3. Experimental rcsrt!rs for tCfiiJh!rarun· mcasrwcmcnts m;d re/aNn.-. tcmpt:rawre de1·iutions _fOr each plw1c.

THERM.·IL ..tN..!Ll'S!S IN F!RE-RES!ST..INC£ FURNACE

Pilow. P.:l. G .. M.:sl[llita. L. .11. R .. Pereim, .-I.

3 NUMERICAL SIMULATION

)[)7

Tho.: numerical simulations were performed with Fluent software. usmg the fi nite vo lume

method, [4]. The generalized Eddy Dissipation Model \\'US used to simulate the combustion of

propane-air mixture. This model is based on the assumption that chemical reaction is almost

instantaneous. when compared with the chemical species flow transportation. The combustion

was modelled using a global one-step reaction mechanism. assuming complete conversion of

the propane to the product species of equntion 3.

(3)

The reaction will be defined in terms of stoichiometric coeflicicnts, formation cnthalpies and

parameters that control the re:1ction rate. The pressured based solver was used with unsteady

and first order implicit fommlation. The viscous model "K-Epsilon" with two equations and a

standard wall function for near wall treatment was used. The surfucc to surface radiation

model was used with computed view factors. The emissivity value for the internal wall

furnace was considered equal to 0.7.

The use of constant trunsporl properties for viscosity. thermal conductivity and mass

diffusivity coefficients is acceptable because the flow is fully turbulent. sec tables I und 2,

[4]. The molecular transport properties will play u minor role compared to turbulent transport.

The assumption of temperature dependent specific heat is an important key factor to predic t

more realistic peak flame temperature.

SpeciFic heat Mixing law (J/kg K]

Conductivity 0.0454 [W/ml<]

Viscosity 1.72x10 5 (l<glms]

Tahle !. !'rup .:rii.:.ljiw J>r"J"""'-air mi.l / 111 '<', { -!}.

Specific mass 1.225

Specific heat 1006.43 [Jikg l<]

Conductivity 0.0242 [W/ml<]

Viscosity 1.7894x10.5 [Kglms]

Tahle :! .. ·lirpmpalil'.l' , at n:f~'l'<'ll<'<' 1<'/Jifh'rtl/ll rc. :!98 / K}. [.1/.

ADVANCED RESEARCH WOIH\SBOI'

IOti FIRE PROTECT/OS AND LIFE S.·fFETr IN BU/LD/i\'GS AND TRANSPOHT:IT/ON Sl"STEMS

Ttmprnlur:~fl'l

Fig. ·1. VarimioJI t~( specific heal of I he producls and rcaelwl/x f.J j.

The specitic heat considered for mixture will be higher where the propane is concentrated,

near the fuel inlet and where the temperature and combustion product concentration should be higher. see figure 4 for temperature dependence.

The governing equations for mass momentum and energy conservation are de lined hy

equations 4-6.

~ (p J;)+\l ·(p;:};)=-\1-J,·i·R, Of

(4)

\Vhere p represents the specific mass, P the velocity vector, }: is the local mass fraction that

correspond5 to the spec ies "i ' ', J, the ditTusion flux. while R, represen ts the rate o r each

species formation by chemical reaction.

For the momentum conservation equation, the stati c pressure ts defined by p , while r

represents the stress tensor.

~ (p ;:)+ \l ·{p;:;:)= - \lp+\1.¥ Cl

(5)

For the energy equation. E represents the energy value, 1<,.11

the effective value for

conductivity, T the temperature value. h1

the sensible enthalpy. while S~o represents the

energy value from chemical reaction.

~ (p E)+ \J. (;:(pE + p))= \ ' ·( k,. .. \7T -) h, ./, + (~ .. · ;:))+ S, m · 7 ~ (6)

TIIERM.·IL ..JN..JL l"SIS IN FIHE-FIESFST·INCE FUR:\".·!CE

l'i/ow, P.A. G .. ,\/cstfllila, L.M.R .. /'crdra . . -1. 10')

An ndiabatic condition was assumed in the internal furnace walls. The inle t them1al

conditions were specified nt room tempemturc. The exhaust tempemturc products were

defined with information from cxperimentalme;tsured data.

The time dependence inlet velocity for air and propane were kept in proper ratio. during each

tcsl. Those values were defined according to the manufacturer limiting values. because they

were not measured.

To solve the unsteady solution. the initial conditions were defined and an incremental time

step was specified. An iterative process was used to solve discrctized equat ions.

The numerical model was built with a structured mesh, using 115840 hcxahcdm finite

volumes, each with 0.02[m] length side, sec figure 5. Major simpl ification was introduced

into the lour burners. using the hydraulic diameter as reference value. Four inlet gas zones

were defined concentric with the same number of air inlet zones, with dimension equal to

20x20 [mm] and 60x60 [mm], respcctivt.:ly. The exhaust is well identified at the bottom of the

furnace volume, with rectangular dimensions equal to I 00 x 400 [mm].

Fi!]. 5. ,\Jrulelfurjire-r~.<iSIIII lCL' f umace.

Flame temper<Jturc depends on several l~tct ors during the combustion process and affects,

significantly, hcut transfer inside lire-resistance furnace. The rate of heat transfer increase

with tlamc temperature. Figure 6 represents temperature and velocity, during the simulation

of test I. in di fie rent defined planes. in particular X 150. XS50, Z 150, ZH50 and exhaust. Tl1e

burner 83/84 plane is also represented.

110 .-\D\'ANCED RESEARCH WORh:SIIOP

FIRE PIWTI::C710N .·!ND LIFE S:JI·L1T I.V BUILDINGS :!ND TII:J.VSI'ORT:IT/ON SJ'ST£.1/S

1 e i~·n

l .i h ·l! ! .s>~ . !1 ! .!.t~ -n

Ul~ ~ ~!

Uk· C! 12~~·n

L:!!"·{! 1.13e•11 I.CSr · !l 5 i Y. · J:' ; 11::·\?

:.~ 1: -; : E.i ~t · i} E:J !r • ':" ~-:'5t · f.:' t..!- lc •!2 1.15~ · t:t ) . ii:: •t:

ul Gmwur,\ tifh'"'l"-'rulllrt• /K/ itt 111~' {cmr fJidlh.'l 1111/t' • JOII bf.

I .H r•el 1.71t •i) t.!:~ r· n

t.!t:·n t.!,r·n ). tjr•O

I.J'!:r·~ J t ..1er ·~ 3 1.2-it -~ J u :;r •!l U~e·O

S.l~ t •E ~ 5 Ur · t ~ U!.c· E; i .SC! •C e.T~c •t; 5.fh·t:' S:'~ ·C t.~ (e ·C :'

l iSc·l?

I

w C) CmUmtr.i taJ h'ntpc:raflln•f A'/ ill llb• fimr f'llltl a:J, ltnll' •)6(1(1 / t/

~ .~Cc · ~ ( : !: c • ~' : : ~e • ! ( ~ 1 3• · CC : .Ht < : L :: !.: ·~ ! 1 .:~- ·! ~

1.5-J:·U u.:e· ;~

t.::c•t• L: ~~ •ii I.Ue·~~

t:t..e-!1 ·.;:r· C: I £::!r ·i l H~r· ll

:;;:c ·U :.~ h-1:

1.1~·C!

l H: ·H

Bfr •U :'.ll:t •: \

~.13::-· H

.:'.C;c ••! l l .!!r • ~ 1 l. i!-t · :l l.!:~e · ; ~

L~h · '! i.J~: •i Q

I.:'! ~ ·U

I. I~:·H

I .H: · !~ e. I ~~ - ; I i.Sir · ; l E.:? :~ · :l

~ tir·il i75 t ·!l ~.: ~:·tl

l.~:.r·! l

i.H:•!:

Fig. fi. Num~ricalr.·.wiTJ.for T,·sr I.

The nux of species is descendent, with small vortices zones, as can be seen in figure 6b) and

d). The numerical resul ts allow identi(ying the localized effect of each bumer. The

experiment;!) measurements corroborate this evidence.

To determine the thermal pcrfom1nnce of the numerical simulation. the mass average

temperature value was determined for each plane defined for experi ments, using equation (7).

This averuge value is also compared with the rcft!n:nce value of the li re-resistance funmce,

detennining the relative difference, sec figure 7.

f = fT pvd.4j fpPd.4 rl . I

(7}

TIJE!IMAL ANALl'SIS IN F/R£-1/ESIST..JNCE FURN.·lCE f'i /o/U, f'.A.G .. Mescfllii<J, L.M.R .. f'crcira, .·1.

llth liotlfml'"uturt Wil'trfiiU

~ • ..,mul n .Pt'l·\1.:1 I

I

u-----J

( -~•c-r•~ ~11-' r-Unn

Ill

l' ~~ ~00 IU •J I Y• l : 10) : ~011 }li.Al ]~¥1 400 1

a) ,\·mllt:ricul ,.,_.,r lllh, ll!iln;: !Wmim:l lc.'Jting nrn'l' JSO.YJ.t ( lt._'.ftli:J.

nr.ul•l

Fig. 7. Al•eroge tt!mperuuJrL' 1-'aiuL' inxide.fiwnace ond rd:llin! lemprratW't! differ~ttce.

ACKNOWLEDGMENT

The authors acknowledge the financial support from the Portuguese Science and Technology

Foundntion, under the reference project PTDC/ EME-PME/649 13/2006.

REFERENCES

I, A EN OR, Norma UNE 1363: Fire rcsistuncc test. Part I: Genernl requirements, 2000.

2. AENOR, Nonnn UNE 1363: Fire resistance test. Part 2: Altcntative and additional

procedures, October 2000.

3. S.Welch, pa Rubini . "Three dimensional simulation of a fire resistance fumace",

Proceedings of 5th International Symposium on Fire Safety Science, Melboume, Australia,

ISBN 4-9900625-5-5 International Association for Fire Safety Science (IAFSS), pp I 009-

1020, March 1997.

4. FLUENT 6.3 Documentation, users manual , 2006.

5. Giuliano Gardolinski Venson, Giiberto Augusto Amado Moreira, Jose Eduardo Mautone

Barros, Ramon Molina Valle. "Modelagem do Perfil de Tempemtura cm Camara de

Combustao Tubular Utilizando o Modelo de Combustiio Eddy Dissipation", X E11contro de

Modelagem Complllaciollal, lnstituto Politecnico/UERJ - Nova Friburgo/RJ, 2 1 a 23 de

Novembro 2007.

6. Coelho, P. J. "Numerical simulation of a mild combustion bumer", Combustion ami name.

124: 503-518, Elsevier, 200 I. 7. Welch, S: Rubini, P.: "Three dimensional simulat ion of a firc-resistnnce furnace".

Proceedings of 5th International Symposium on Fire Safety Science, Melbourne, Australia.

March 1997, International Association for Fire Safety Science. pp. I 009-1020, ISBN 4-

9900625-5-5