Professional Documents
Culture Documents
Fire Research, SP Technical Research Institute of Sweden, Box 857, SE-501 15 Bors, Sweden
Department of Civil, Environmental and Natural Resources Engineering, Lule University of Technology, SE-971 87 Lule, Sweden
art ic l e i nf o
a b s t r a c t
Article history:
Received 25 June 2014
Received in revised form
12 January 2015
Accepted 1 February 2015
Available online 14 February 2015
This paper presents a model for estimating temperatures in pre-ashover res where the re enclosure
boundaries are assumed to have lumped heat capacity. That is, thermal inertia is concentrated to one
layer with uniform temperature and insulating materials are considered purely by their heat transfer
resistance. The model yields a good understanding of the heat balance in a re enclosure and was used to
predict temperatures in insulated and non-insulated steel-bounded enclosures. Comparisons were made
with full scale experiments and with other predictive methods, including CFD modeling with FDS and the
so called MQH relationship. Input parameter values to the model were then taken from well-known
literature and the heat release rates were provided from the experiments. The re temperature predictions of the model matched very well with experimental data. So did the FDS predictions while the
original MQH relationship gave unrealistic results for the problems studied. Major benets of using the
model in comparison with CFD modeling are its readiness and simplicity as well as the negligible
computation times needed. An Excel application of the presented pre-ashover re model is available on
request from the author.
& 2015 Elsevier Ltd. All rights reserved.
Keywords:
Temperature prediction
Fire temperature
Zone model
Pre-ashover
Heat transfer
Compartment re
1. Introduction
To model the many and complex effects of res, increasingly
sophisticated computer models have been developed the last few
decades; they are now used routinely both in the re research
scientic community and in applied re safety engineering. The
desire for quick approximate answers has, however, also motivated the development of simple procedures. A key value in assessments of the severity of a compartment re before ashover is
the temperature of the upper hot layer, here referred to as the re
temperature. Commonly used simple methods for pre-ashover
re temperature approximations are often based on [1] the correlation developed by McCaffrey, Quintiere, and Harkleroad [2],
the so called MQH relationship. It builds on a simplied energy
balance and regression correlation with data from numerous test
res.
Whilst the MQH relationship is simple to use it has many limitations which are seldom properly considered. It is for example
often claimed that the relationship is applicable to steady-state as
well as transient re growths [13]. However, the temperature is
solved as a function of the heat release rate at a particular time step
and takes no account to the re growth history (e.g. slow or constant). The applicability in case of a growing heat release rate is
therefore questionable. Furthermore, it has not been generally understood that the original MQH relationship is inappropriate when
surrounding boundaries are thin and highly conductive [cf. 1,3]. To
manage this shortcoming, Peatross and Beyler developed a modied expression for the heat transfer coefcient in the MQH relationship [4]. It gave improved correlation with tests in steelbounded enclosures, similar to the ones studied here, but the
general limitations of the empirical correlation remain.
This paper shows a clear and understandable enclosure re
model for estimating temperatures in pre-ashover res, derived
based on well-known re physics. The model presented here is
limited to re enclosures where the boundaries may be assumed
to have lumped heat capacity, i.e. where thermal inertia is concentrated to one layer of the boundary structure with uniform
temperature. It may be used to predict the upper layer re temperature as well as the temperatures in the boundary structure as
functions of time for given heat release rates varying with time.
2. Method
*
Corresponding author.
E-mail addresses: franz.evegren@sp.se (F. Evegren),
ulf.wickstrom@ltu.se (U. Wickstrm).
http://dx.doi.org/10.1016/j.resaf.2015.02.008
0379-7112/& 2015 Elsevier Ltd. All rights reserved.
78
Nomenclature
A
c
d
H
h
L
l
m
R
T
t
q
area [m2]
specic heat capacity
[J/(kg K)]
core thickness [m]
opening height [m]
heat transfer coefcient
[W/(m2 K)]
ame height [m]
thickness of insulation [m]
mass ow rate of gases [kg/s]
heat resistance [m2 K/W]
temperature [K]
time [s]
heat ux [W/m2]
Greek symbols
temperatures in enclosure res by simple, often analytical, methods. For ventilation controlled res it was presented [5] assuming
a one-zone model, similar to the models by Magnusson and Thelandersson [6] and the parametric curves in Eurocode 1, EN19911-2 [7]. In this paper a pre-ashover model is derived with the
same approach but based on a two-zone assumption and expressions of the heat transfer through the re enclosure boundaries.
The upper layer re temperature and temperatures throughout the
boundary structure may thereby be predicted as functions of time
for given heat release rate histories. The model presented in this
paper applies only to enclosure boundaries for which lumped heat
capacity may be assumed. Different insulation alternatives can be
considered and the heat capacity of the insulation can either be
neglected or numerically added to the core of the boundaries. To
validate the model, which was derived straight from re physics,
comparisons were made with two large scale experiments performed in a steel container with and without insulation [8].
Comparisons were also made with FDS simulations [9] and calculations based on the MQH relationship [2]. FDS simulations were
performed as they were expected to yield the most accurate theoretical predictions, in particular in comparisons with the more
approximate two-zone models.
2.1. New simple pre-ashover re model
The heat released by combustion in an enclosure is lost in
different ways, as illustrated in Fig. 1. By setting up a heat balance,
Superscripts
*
.
pre-ashover model
per unit time
per unit area
Subscripts
1
C
c
core
f
g
h
i
k
L
O
o
p
R
r
s
tot
ult
W
ambient
combustion
convection
core of structure
re
gas
boundary heat transfer
inside
conduction
air replacement
opening
outside
at constant air pressure
radiation
radiation
surface
total
ultimate
walls
the
the
the
the
79
A
1
2
= O f (T
+ T2f )(T + Tf )
R r
Atot
(2)
=
qW
Fig. 2. Electric analogy of the new pre-ashover re model assuming lumped heat.
4
opening as qR = AO f T f4 T
, where m is the mass ow of gases
out through the opening, cp is the specic heat for air, Tf is the re
(gas) temperature, T is the ambient temperature, AO is the
opening area, f is the emissivity of the elements emitting radiation inside the enclosure, and is the StefanBoltzmann constant.
The radiation out through the opening, from ames, smoke layer
and enclosure surfaces, is thereby calculated assuming a ctitious
emissivity f and a uniform internal temperature equal the re
temperature Tf . Using these expressions and assuming that the
heat is spread evenly to all surrounding surfaces allow expressing
the mean heat ux to the boundaries as in Eq. (1).
=
qW
p qC
mc
AO
4
f T
T f4 +
f
p
Atot
Atot mc
1
1
1
1
f + (ult
f ) = + (max
f )
R r
Rc
Rc
Rr
qi = s, i T 4f T s4, i +
1
(Tf Ts, i)
R c, i
(4)
(1)
where Atot is the total internal enclosure surface area (i.e. excluding the opening area) and f is the temperature rise Tf T.
When expressing the heat transfer to the surrounding
(3)
80
(Tf Ts,i). A radiative heat transfer resistance Rr,i may then be derived as Eq. (5). The boundary condition at the inside surfaces can
thereby be simplied and expressed in analogy with the heat
transfer by convection, as shown in Eq. (6). A new heat transfer
resistance for the inside surface of the boundary structure Rh, i ,
which includes both convection and radiation, can thereby be
dened as Eq. (7).
1
= s, i (T2f + Ts2, i)(Tf + Ts, i)
R r, i
qi =
1
1
(Tf Ts, i) + R (Tf Ts, i)
R r, i
c, i
1
1
1
=
+
Rh, i
R c, i
R r, i
(6)
(8)
= cd
qcore
dTcore
dt
i+1
i
Tcore
Tcore
=
(9)
t
(q " qo ")
cd i
(10)
4
qo = s, o T s4, o T
+
1
(Ts, o T)
R c, o
qo =
1
1
(Tcore T) =
core
Ro
Ro
(12)
(5)
(7)
1
1
qi = (Tf Tcore) = (f core)
Ri
Ri
(11)
Similar to above, the heat transfer by radiation can be expressed as a function of the temperature difference (Tf Ts, o) if
expressing the radiation heat transfer resistance in analogy with
Eq. (5). A total heat transfer resistance for radiation and convection
combined at the unexposed side of the core can then be dened in
analogy with Eq. (7).
If there is no insulation at the unexposed side of the core, the
outside surface temperature equals the temperature of the core,
qo =
1
R R*
Ri + Rc* + i c
R r*
i+1
i
core
= core
*
ult
R * + R * core
c
Ri + r
R r*Rc*
(13)
1
t
+
*
R R * ult
cd
Ri + Rc* + i c
R r*
1
1 i
+
core
*
*
R o
R + R r + Rc
R r*Rc*
(14)
i
where core
is the core temperature rise at the time increment i ,
t is the length of the time step, c and are the specic heat
capacity and the density of the core material, respectively and d is
the core thickness. R is heat resistance with subscripts indicating
radiative, convective, or conductive at the inside or outside of the
core. The core temperature of time increment (i 1) is hence calculated by determining the parameters on the right hand side of
Eq. (14) at time increment i. This may be determined only from the
current heat release rate and known constants, as shown in Section 2.3.
Based on the core temperature, the re temperature and temperatures throughout the boundary structure may be calculated
depending on the various thermal resistances according to the law
of proportion (see Fig. 2). The calculation procedure is demonstrated step by step in Appendix A.
Eq. (14) is expressed in terms of the ultimate temperature,
which may be the easiest for calculation purposes. Alternatively
1
1
( max
) + ( max
)0
Rc ult
Rr
(15)
i+1
i
core
= core
t
+
cd
*
R f +
Rk, o
1
i
* core
max
R R
r, i c, i + Rk, i
R r , i + R c, i
1
i
core
R R
r , o c, o
R r , o + R c, o
(16)
81
Fig. 4. Dimensions of the test enclosure and a photo of the insulated test enclosure.
82
layer height. The correlation was optimized for a ratio between the
re temperature and the ambient temperature of Tf /T = 1.7, giving the coefcient a value of 0.684, and is given in Eq. (17). As the
correlation is independent of the neutral layer height it may be
solved for the smoke layer height, z. The result was at each time
step inserted into the Heskestad re plume ow correlation, see
Eq. (18), for a less uncertain approximation of the mass ow rate
m .
z
m = AO HO 1
HO
(17)
1
L
1
+
m=
AO HO
0.0056cqC HO
(18)
The Heskestad re plume ow correlation includes the convective proportion of the total heat release rate qC . The average
convective part of the heat release rate c was measured to approximately 0.67 in the experiments, which was adopted.
2.4. FDS simulations
For comparison, the experiments were modeled in FDS, Fire
Dynamics Simulator, version 5.5.3 [9]. The heat release rate recorded in the experiments was used as input to the simulations.
Since circular objects cannot be specied in FDS, a square re
source with the same surface area was assumed, placed at the
same height as in the experiments. Heptane was used as fuel in
FDS and the soot yield was set to 0.037 [15]. The walls and roof of
corrugated were assumed plane in the FDS model. The walls, oor,
and ceiling of the construction were given material properties for
steel or steel and insulation, as presented in Table 1. The oor was
not insulated in any of the simulations, as it was never insulated in
the experiments. Neither was any consideration taken to the
concrete oor on which the container was placed. The size of the
grid cells was set to 50 mm cubes, which gives a D*/x value of 16
after about 50 s in the simulations (a value of at least 416 is recommendable to adequately resolve re plume dynamics [9]). A
mesh sensitivity study was performed, which showed grid independent results.
For comparison, output devices were specied in FDS which
predict temperatures as the 0.25 mm thick thermocouples used in
the experiments. The core (wall) temperatures were derived directly from the model.
Table 1
Input parameters to the models and correlations.
Parameter
Value
Unit
Ho
Ao
Atot
T1
cp (air)
kcore (steel)
(steel)
c (steel)
d (steel thickness)
k (insulation)
ins (insulation)
cins (insulation)
l (insulation thickness)
f
s,i
Rc,i
s,o
Rc,o
2
2
65.33
293
1012
45
7820
460
1.8
0.09
30
800
95
1
0.7
0.04
0.7
0.25
0.684
0.67
m
m2
m2
K
J/(kg K)
W/(m K)
kg/m3
J/(kg K)
mm
W/(m K)
kg/m3
J/(kg K)
mm
dimensionless
dimensionless
m2 K/W
dimensionless
m2 K/W
dimensionless
dimensionless
3. Results
Two full scale experiments were performed. The measured heat
release rates were input to the calculation models. Then the
measured core and re temperatures were compared with the
corresponding temperatures predicted by the models.
3.1. Observations and heat release rates from experiments
Flames reached the ceiling 10 s after ignition and thereafter the
res grew quickly, reaching a heat release rate of 700 kW in about
1 min. In the non-insulated enclosure the re grew slower than in
the insulated, reaching about 950 kW after 500 s. The re was
extinguished after 960 s. In the experiment with insulated
boundaries, small ames emerged at the opening already after
375 s. After 465 s the re was considered to have reached ashover and was extinguished. The walls were then glowing red.
The measured heat release rates from the experiments were
directly input to the new pre-ashover re model and to the MQH
83
measured.
Fig. 6. Heat release rate histories used as inputs to the different models.
Fig. 7. Comparison of the measured and estimated core temperatures in the insulated container.
4. Discussion
Comparisons with experiments and FDS simulations showed
Fig. 8. Comparison of the measured and estimated core temperatures in the noninsulated container.
Fig. 9. Fire temperatures estimated by the new pre-ashover re model in comparison with experiments for the insulated and non-insulated cases.
84
that the new model predicts core and re temperatures well, both
for insulated and non-insulated steel enclosures. The new model
and FDS gave similarly good predictions for the steel core temperature. Both models under predicted these temperatures only
slightly up until the end of the tests. The new model predicted the
re temperature in the insulated case well but over predicted
slightly in the non-insulated case.
When comparing temperatures calculated by the new model
with measurements and FDS predictions, arithmetic average
temperatures were used which is an uncertainty of the
comparison.
The calculations with new model include a number of simplications and assumptions of which the following are deemed the
most signicant:
rates.
As an alternative, mass ow rates were for comparison also
calculated based on the Heskestad re plume ow correlation
[13] and the Johansson and van Hees [14] simplication. A
sensitivity study showed insignicant effects on the results
from using these correlations in the studied cases.
Heat ux to the enclosure boundaries was assumed uniform to
all surrounding surfaces, including the oor. This assumption is
justied by the fact that radiation is the dominant mode of heat
transfer at high temperatures.
The compartment boundary structure was modeled assuming
lumped heat, i.e. the thermal inertia is concentrated to a core
possibly insulated on either side. The heat capacity of the insulation material is neglected and the conductivity of the core
is assumed to be innite.
Effects of the corrugation of panels were disregarded.
Input data for calculations were obtained from measurements
and where relevant with reference to Eurocodes [7,17]. The
only exceptions are the insulation properties which were provided by the manufacturer and the emissivity at the enclosure
opening which was assumed to be 1.
The great strengths of the new pre-ashover re model are its
Acknowledgments
The authors are grateful to associate professor Johan Anderson
at SP Fire Research for support in FDS simulations. Thanks are also
extended to Anna Back for managing the re tests. The tests were
nanced by the European Community's Seventh Framework Programme under Grant agreement no. 233980 and two competence
centers at SP Technical Research Institute of Sweden, Novel Designs at Sea: Shipping and Offshore and Zero Emission
Buildings.
i+1
i
core
= core
1
t
+
*
R R * ult
cd
Ri + Rc* + i c
R r*
1
1 i
+
core
*
*
R o
R + R r + Rc
R r*Rc*
(A2)
qC
p
mc
(A3)
Atot
p
mc
A
1
2
R r = O f (T
+ T2f )(T + Tf )
Atot
1
= s, i (T2f + Ts2, i)(Tf + Ts, i)
R r, i
(A7)
1
1
1
+ Rk, o
+
R o =
R r, o
R c, o
(A8)
1
= s, o (T2f + Ts2, o)(Tf + Ts, o)
R r, o
(A9)
qi =
qo =
1
R R*
Ri + Rc* + i c
R r*
*
ult
= qW
R * + R * core
r
c
Ri +
R r*Rc*
1
1
(Tcore T) =
core
Ro
Ro
(A10)
(A11)
qi =
1
(f core) f = core + Riqi
Ri
(A12)
qi =
1
(core s, o) s, o = core Rk, oqi
Rk, o
(A13)
References
Rc =
(A6)
(A1)
There are several ways to estimate the mass ow rate. The mass
ow rates in and out through the opening are assumed equal to
the plume ow and may be estimated directly from a re plume
ow correlation. An alternative is to estimate the mass ow rate by
iterating the original equations for the mass ow rates in and out
through the opening [6] and a re plume ow correlation [13]
(possible with information from the previous time step). As a third
alternative a simplied correlation between the mass ow rates
through the opening and the smoke layer height [14] may be used
instead of the original mass ow equations. For example, together
with the Heskestad re plume ow correlation the mass ow can
then be calculated from Eq. (A2).
ult
=
1
1
1
+ Rk, i
+
Ri =
R r, i
R c, i
1
L
1
+
m =
AO HO
0.0056cqC HO
85
(A4)
(A5)
The heat resistance at the inside (re exposed side) of the core
Ri depends on convection and radiation heat transfer resistances at
the surface and on conductive heat resistance of potential covering
material, as shown in Eq. (A6). The latter is zero if the inside of the
core is not covered. The radiation heat transfer resistance at the
inside surface may be calculated in a separate column according to
Eq. (A7), where the re and inside surface temperatures are taken
from the previous time step.
[1] W.D. Walton, P.H. Thomas, Estimating temperatures in compartment res, in:
P.J. DiNenno (Ed.), The SFPE Handbook of Fire Protection Engineering, third ed.,
NFPA, Quincy, MA, USA, 2002, pp. 3-1713-188.
[2] B. McCaffrey, J. Quintiere, M. Harkleroad, Estimating room temperatures and
the likelihood of ashover using re test data correlations, Fire Technol. 17 (2)
(1981) 98119.
[3] B. Karlsson, J.G. Quintiere, Enclosure Fire Dynamics, CRC Press, Boca Raton,
2000.
[4] M.J. Peatross, C.L. Beyler, Thermal environmental prediction in steel-bounded
preashover compartment res, in: Proceedings of the Fourth International
Symposium on Fire Safety Science. International Association for Fire Safety
Science, Boston, 1994, pp. 205216.
[5] U. Wickstrm, A. Bystrm, Compartment re temperature a new simple
calculation method, in: Proceedings of the Eleventh International Symposium
on Paper Presented at the Fire Safety Science. Christchurch, New Zeeland,
2014.
[6] S.E. Magnusson, S. Thelandersson, Temperature-time curves of complete
process of re development: theoretical study of wood fuel res in enclosed
spaces, Civ. Eng. Build. Constr. Ser. 65 (1970), pp. 181.
[7] CEN, EN 1991-1-2, Eurocode 1: Actions on Structures Part 12: General
Actions Actions on Structures Exposed to Fire. European Committee for
Standardisation, 2002.
[8] A. Back, Fire Development in Insulated Compartments: Effects from Improved
Thermal Insulation, Department of Fire Safety Engineering and Systems Safety,
86
Lund, 2000.
[9] K. McGrattan, B. Klein, S. Hostikka, J. Floyd, Fire Dynamics Simulator (Version
5) User Guide, National Institute of Standards and Technology, USA, 2008.
[10] O. Pettersson, K. deen, Brandteknisk Dimensionering: Principer, Underlag,
Exempel. 1978.
[11] V. Babrauskas, Heat release rates, in: P.J. DiNenno, D. Drysdale, C.L. Beyler,
et al., (Eds.), The SFPE Handbook of Fire Protection Engineering, fourth ed.,
National Fire Protection Association, Quincy, MA, USA, 2008, pp. 3-13-59.
[12] S. Welch, A. Jowsey, S. Deeny, R. Morgan, J.L. Torero, BRE large compartment
re tests-characterising post-ashover res for model validation, Fire Saf. J. 42
(8) (2007) 548567. http://dx.doi.org/10.1016/j.resaf.2007.04.002.
[13] G. Heskestad, Fire plumes, ame height, and air entrainment, in: P.J. DiNenno,
D. Drysdale, C.L. Beyler, et al., (Eds.), The SFPE Handbook of Fire Protection
Engineering, fourth ed.,National Fire Protection Association, Quincy, MA, USA,
2008, pp. 2-12-20.
[14] J. Johansson, P. Van Hees, A simplied relation between hot layer height and
opening mass ow, in: Proceedings of the Eleventh International Symposium
on Paper Presented at the Fire Safety Science. Christchurch, New Zeeland,
2014.
[15] A. Tewarson, Generation of heat and chemical compounds in res, in: P.
J. DiNenno, D. Drysdale, C.L. Beyler, et al., (Eds.), The SFPE Handbook of Fire
Protection Engineering, third ed.,National Fire Protection Association, Quincy,
MA, USA, 2002, pp. 3-8283-161.
[16] S. Weaver, A Comparison of Data Reduction Techniques for Zone Model Validation. Fire Engineering Research Report, University of Canterbury, Canterbury, 2000.
[17] CEN, EN 1993-1-2, Eurocode 3: Design of Steel Structures General Rules
Structural Fire Design. European Committee for Standardisation, 2005.