Professional Documents
Culture Documents
THE IMPLICATIONS OF
THE WORLD TRADE CENTER DISASTER
Corus
21/04/04
CONTENTS
1. Papers
2. Appendices
A: The Royal Society of Edinburgh
B: Acknowledgments
C: Speaker Biographies
D: Participant List
21/04/04
PAPERS
Dr Jose Torero
Reader in Fire Dynamics, University of Edinburgh
STRUCTURES IN FIRE: AN OVERVIEW OF THE BOUNDARY CONDITION *
Co-Authors: Dr Allan Jowsey, Dr Asif Usmani, Profesor Barabara Lane and Dr Susan Lamont
** Arup Fire
13 Fitzroy Street
London, W1T 4BQ
United Kingdom
Introduction
From the perspective of a Fire Safety, the design of a building can be approached in two different ways. The first is
for the building to comply with existing regulations, and the second one is to achieve certain safety goals.
Regulations have not been developed to fully specify the design of unique and complex buildings such as the
World Trade Center and even, in the event that they existed, they are of questionable effectiveness. Furthermore,
if a scenario such as the one of September 11th, 2001 needs to be considered as a possible event during the life of
the building, design on the basis of safety goals is the only path that can be followed.
The schematic presented in Figure 1 could represent the behavior of a building in the event of a fire. It could be
argued that the safety objective should be that the time to evacuation (te) at each compartment (i.e. room of
origin, floor, building) be much smaller that time necessary to reach untenable conditions in the particular
compartment (tf). Characteristic values of te and tf can be established for different levels of containment, room of
origin, floor, building. Furthermore, it is necessary for the evacuation time to be much smaller than the time when
structural integrity starts to be compromised (tS).
In summary:
te<<tf
te<<tS
It could be added to these goals that full structural collapse is an undesirable event, therefore:
tS
Although these criteria for safety times can be considered as a simplified statement, it is clear that it describes well
the main goals of fire protection.
21/04/04
Untenable
Conditions
1st Floor
Untenable
Conditions
Building
tf
The Fire
te
100%
evacuated
Detection
time
Subsequent Sprinkler
First Sprinkler
tS
Fire Service
Figure1
Schematic of the sequence of events following the onset of a fire in a multiple story building. The
thick line corresponds to the fire size, the dotted lines to the possible outcome of the different
forms of intervention (sprinkler activation, fire service). The dashed lines are the percentage of people
evacuated, with the ultimate goal of 100% represented by a horizontal dashed line. The dashed &
dotted line corresponds to the percentage of the full structural integrity of the building.
With the objective of achieving these goals a number of safety strategies are put in place. These include those
strategies that are meant to increase tf which include active systems, such as sprinklers, or the intervention of the
fire service. As shown by Figure 1 (dotted lines), success of these strategies can result in control or suppression of
the fire. Passive protection such as thermal insulation of structural elements becomes part of the design with the
purpose of increasing tS. Finally, but most important, evacuation protocols and routes are design to minimize te at
all stages of the building. It is important to note that within the estimation of te the safe operations of the fire
service need to be included.
The events following the attack on the World Trade Center showed that these safety goals were not attained. It is
therefore important to seek the best possible understanding of why this happened. For this purpose an adequate
understanding of the nature of the event and the characteristic of the structure and its safety systems is necessary.
This requires a detailed understanding of the fire conditions, the interactions between the fire and the structural
elements and the sequence of the intervention and evacuation processes. Different methodologies and tools have
been developed to study each of these aspects. This paper will provide, within the context of the September 11th,
2001 events, and overview of the methodologies used to assess the boundary condition between the structural
elements and the fire.
dT
= A S q&S
dt
T
2T
S Cp S
= kS 2
t
x
S Cp S VS
21/04/04
Where the boundary condition for both cases corresponds to the input from the fire and is given by:
(Thermally Thin Material)
T
x
Where Tg is the imposed temperature of the gas as defined by the Temperature vs. Time curves. The emissivity of
the solid surface is given by S and that of the gas by g . For simplicity heat exchanges with the outside
environment have been ignored but could be included in these expressions. For the thermally thin elements AS will
be the exposed area. The unexposed area can be ignored or treated as a loss to some ambient temperature. For
the thermally thick materials the boundary condition at the other end will be fixed based on the conditions
established for this side of the element. If a fire is present at the other side then a similar boundary condition will
be included at this end, if no fire is imposed a heat loss to an ambient temperature can be used.
A very different way of defining the boundary condition is by assuming that the surface temperature of the
structural element is that of the gas. This is a simpler boundary condition that requires the introduction of less
parameters, but currently is consistently deemed as not describing properly the physics of the heat transfer
process.
The concept of Temperature vs. Time curves implies a number of simplifications of which the main are:
The compartment fire temperature is homogeneous with no spatial differences worth considering.
The radiation field is in thermal equilibrium within the gas phase, i.e. there is no radiation exchange between
soot particles and the gas, and thus gas temperatures can be used to establish radiative heat fluxes.
The optical depth within the gas phase is much smaller than the characteristic length scales of the
compartment. Thus heat radiation can be treated as a local phenomenon.
The computation of the emissivity ( g ) is also subject to various simplifications that vary with the author. A
common assumption states that the emissivity increases exponentially with the thickness of the emitting gas and
thus Petterson et al (1976) postulates that
g = 1 exp( x g )
(1)
Where is an emission (or absorption) coefficient and x g the thickness of the emitting layer. This approach carries
the further assumptions that single emitting temperature and gas phase emissivity is sufficient to describe radiative
heat exchange. The radiative component needs to account for all sources of radiation, thus a more complete way
to describe the above boundary condition will be:
(2)
q&r, T conglomerates all radiative inputs. Radiative inputs can come from the
hot gases, soot, other surfaces or the flame, furthermore, they are attenuated by the absorption through the gas
phase. It is important to note that absorption is a function of the soot volume fraction and temperature through
the soot absorption coefficient ( ):
= Cf S T
(3)
Where C is a constant, fS the soot volume fraction and T the temperature. Thus if the soot volume fraction is high
and the distance from the flame or other hot element is large equation (1) shows that all energy from these
emitting bodies will be absorbed by the smoke before reaching the target. The assumption that the hot gases
adjacent to the structural element are the main contributors to its heating might then be appropriate and there is
no need to resort to equation (2). Furthermore, if far from the flames, thermal equilibrium between soot and gas
phase in the smoke might also be accurate.
The relevance of each of these assumptions can be evaluated for each specific scenario but to understand the
validity of these simplifications it worth briefly reviewing some basic concepts of compartment fires, this will be
done in the following section.
21/04/04
A / Ao Ho ,
control the oxygen flow through the crib. H being the height of the vertical shafts of the crib, H0 the height of the
compartment opening, A0 the area of the vertical shaft or the compartment opening and A the surface area of the
crib or the room floor. For limited oxygen the ventilation factor controls the burning rate and a constant burning
rate is observed for different vertical shaft areas. With sufficient oxygen, the exposed surface area of the sticks
controls the burning rate and therefore the burning rate increases with (HAo/A),crib.
21/04/04
1200
1000
T [ o C]
800
600
400
200
0
0
10
20
30
1/2
AT /AH
Figure 2
40
50
-1/2
[m
Time mean temperature near the ceiling. Where AT is the total area excluding floor an opening, A
the window area and H the height of the window. The fuel loads for these tests are in the range 2040 kg/m2 that is smaller but nevertheless comparable to what would be expected in a modern office.
If the burning rate can be established then, knowing the heat of combustion, the energy release rate can be
calculated and thus the temperature of the compartment. Then a correction could be made to establish the
fraction of the energy that remains within the compartment. Figure 2 represents the curve fit presented by
Thomas and Heselden (1972) that gives estimates of the temperatures that could be expected for wood cribs in
small-scale (1 m high) compartments. The actual data has some scatter which Law and OBrien (1983) suggest to
be a result of some particularily extreme experimental conditions. The results are expressed in terms of the
ventilation-factor and surface area and are hoped to be scale independent. As can be seen in Figure 2, this study
only provides a single average temperature for each condition instead of a temporal evolution of the temperature.
Despite being less information this is consistent with the assumptions of the thermal model. The extent of the
period characterized by the peak temperature can be defined as a function of the empirical burning rates and the
duration of the decay can be estimated using a simple energy balance for the compartment. Torero et al (2002)
performed this analysis for the WTC 1 & 2 Towers.
The C.I.B. work consisted of a parametric study that included more than one hundred experiments thus allows for
a reasonable level of confidence to be associated to the data. Nevertheless, the data presented is limited to
average values and does not address the spatial temperature distributions within the compartment nor the
proximity of the flames to the structural elements. Drastic temperature variations within the compartment have
been suspected for many years but very few experiments exist to demonstrate the significance of these variations.
Numerical modelling can serve to describe the significance of these variations. Figure 3 shows the simulations
corresponding to the same fire embedded in compartments with three different aspect ratios. It can be observed
that temperature variations greater than 600oC exist throughout the compartment. Furthermore, analysis of the
soot volume fractions show also well defined distributions. These observations seem to further establish that the
basic premise of a single compartment temperature might be over simplified. The obvious consequence of this is
the need to compute the local temperatures and to solve the radiative transport equation. This can only be done
using appropriate compartment fire models or through experimental characterization of the radiative fluxes to the
different surfaces. The next section will discuss the state of the art in compartment fire modelling.
(a)
(b)
21/04/04
(c)
Figure 3 Example of three FDS calculations of a compartment fire. The temperature legend is not presented
because the emphasis is on the spatial distribution of the temperatures not on the quantitative values.
The red is approximately 1000oC and the green 400oC. For all three cases the compartment cross
section is 4m x 4m x 4m and the lengths is (a) 4m, (b) 8m and (c) 16m. For all cases the heat release
rate per unit area is 1000 kW/m2 propane fire distributed throughout the surface. All surfaces concrete,
the grid size is approximately 0.3m to 0.5m in all directions. The ventilation opening is 4m width by
2.5m height.
q&r, T are those that should be expected as outputs of Compartment Fire Models. This paper will address
21/04/04
Zone Models
Zone models treat compartments as a control volume sub-divided into two smaller control volumes. It is assumed
that within the two smaller control volume all properties and conditions are homogeneous. One control volume
considers the smoke and the other the fresh air. Flow, temperature and species fields within these control volumes
are not resolved. A process of mass and energy transfer between them links different compartments. The solution
of the flow, which is the most computationally intense aspect of these calculations, is thus avoided by this simple
two-zone approach. All heat transfer related quantities within these codes are established in an empirical manner,
therefore, no general comments on the limitations of these codes will be provided at this stage. Significant
experimental validation of the principles of this methodology has been generated in the last three decades and its
limitations have been many times described. The reader is referred to Walton (2002) for detailed information.
Two-zone models are by definition limited when analysing heat transport from the gas phase to the solid phase.
They avoid the solution of the fluid mechanics equations thus allows for faster computations and more complex
scenarios. Nevertheless relies on empirical correlations at all levels of heat and mass transfer. These empirical
correlations have in general no link with the burning conditions, thus the convective heat transfer coefficients and
radiation heat transfer used for a small fire will be the same as for large fires.
Calculation of the convective coefficient (assumed to be natural convection) is via correlations for walls,
ceilings and floors (hot surface up or cold surface down) and ceilings and floors (cold surface up or hot
surface down)(Jones et al., 2000 and Cooper, 1991). The convective transfer coefficient is generally
defined in terms of the Nusselt number (i.e in CFAST (Cooper, 1991):
(4)
Nu k
h=
= C Ra nL
Ra L = Gr L Pr =
g(Ts Tg )L3
This number is based on a characteristic length, L, of the geometry. The power n is typically 1/4 and 1/3 for
laminar and turbulent flow, respectively. All properties are evaluated at the film temperature: T f = (T s+Tg)/2.
The thermal diffusivity and thermal conductivity of air are also defined as a function of the film temperature,
from data in (Atreya, 1988).
= 1.0 10 9 Tf7 4
0.0209 + 2.33 10 5 Tf
k =
1 0.000267 Tf
21/04/04
CFD Codes
The main aspect that differentiates CFD codes is the way by which turbulence is modelled. Thus, CFD codes can
be divided into three groups, Reynolds Averaged Navier-Stokes (RANS) models, Large Eddy Simulation (LES)
models and Direct Numerical Simulation (DNS) models. For the modelling of an environment such as a
10
21/04/04
compartment fire and given the computational resources currently available, DNS simulations are not feasible for a
number of reasons. DNS requires the grid resolution to be as fine as the Kolmogorov micro scale. All eddies, down
to the dissipation scale, must be simulated with accuracy. The number of DNS grid points required for the
resolution of all scales increases approximately with the cube of the Reynolds number (Re 3). Since the Reynolds
number for typical fire and smoke movement in a compartment is approximately 105, the total number of cells
necessary for solving fire and smoke movement in a room is approx 1013. Current super computers have the
capability to provide a grid resolution not greater than 108 cells. Therefore, current computing technology is still
far too small to solve such fire motions. DNS, therefore, cannot be used to simulate complicated fire spread and
smoke movement in a full compartment.
Since full resolution of the Navier-Stokes equations is not practically possible, it is necessary to model some aspects
of the flow. The choices of which aspects of the flow will be modelled and thus the approach to be followed, is
difficult and implies inevitable subjectivity. RANS solves ensemble-averaged Navier-Stokes equations by using
turbulence modelling. RANS can be further divided into turbulent viscosity models (such as the k-e model) and
Reynolds-stress models. The most widely used turbulent viscosity model is the standard k-e model. In a RANS
solution, all dynamical degrees of freedom smaller than the size of the largest (energy-containing) eddies are
averaged, so there is no dynamic information about the smaller scales.
LES, developed in the early 1960s~1970s by Smagorinsky (1963) and Deardorff (1970), assumes that turbulent
motion can be separated into large-eddies and small-eddies. The large eddies (grid scale) motion is directly
simulated and the small eddies (sub-grid scale) motion is approximated. Since LES solves time-dependent flow, it
can provide detailed information on turbulence, such as 3D instantaneous velocity.
The key step in both LES and RANS is the derivation of the underlying dynamical equations averaged over small
scales. The only difference between LES and RANS is the definition of small scales. In LES, the small length scales
are smaller than the grid size and in RANS small length scales are smaller than the largest eddies. If the grid size of
an LES simulation is taken larger and larger, self-consistency requires that LES results approach the RANS results
(Orszag et al., 1993). LES techniques always need to be 3D and must have a time step short enough to capture
most of the important turbulent motion. Because of this, LES is computationally more expensive than RANS.
Nevertheless, recent advances in computer performance and numerical methods have made LES feasible for such
fire and smoke flow problems.
Some general limitations for both RANS and LES approaches to the modelling of turbulent flows relevant to fires
can be established:
RANS codes average over time, thus all dynamic information for scales smaller than the large turbulence
scales is lost. For the calculation of the thermal response of structural elements this might not be
significant since the time scales of solids are much larger than those of gases. Nevertheless, the loss of
dynamic information can significantly affect the predictions of fire growth therefore needs to be handled
with great proficiency.
LES does not average over time so it allows modelling the time evolution of the sub-grid scales. This can
be translated in a better resolution of the time evolution of the fire. To achieve computations within
reasonable time constraints, better time resolution requires an increase in the cell size thus large grid cells
characterize LES solutions. The grid cells are much larger than the flame thickness, therefore the
temperatures of each cell represent an average of reactive and non-reactive regions. Thus, the capability
of these codes to properly model flame temperatures and thus radiative heat transfer is questionable.
Furthermore, LES modelling implies a proper definition of the grid size that is consistent with the model
parameters and with the computation constraints. A reduction of the grid not always produces an
improvement in precision. Determination of the grid requires pre acquired empirical knowledge or
independent computations (Novozhilov, 2001).
RANS relies on numerous empirical model coefficients (between 7 to 12 different coefficients) that will
describe turbulent viscosity and fluid wall interactions. These functions are well defined for high Reynolds
numbers with homogeneous turbulence but difficult to establish for transitional flows with constraint
boundaries as those to be expected close to the boundaries with structural elements. Wall functions have
been established to address these areas but their accuracy and generality is still questionable (Bilger,
1988).
Diffusion flames representative of fires are generally considered thick, thus the validity of the direct
application of RANS and LES models could be questioned (Bilger, 1988). Despite this statement, proficient
use of these models can provide adequate results.
LES can also rely on an empirical model coefficient (i.e. Smagorinsky constant) but its calibration is easier
and is independent of the Reynolds number. In fact, this model coefficient can be avoided entirely with a
dynamic sub-grid model (Orzag et al., 1993). Calibration of the model coefficients has been done for a
multiplicity of scenarios but these rarely include conditions typical of fires (Orzag et al. 2003). Dynamic
sub-grid models are beginning to appear in the fire literature but are still ongoing research and have
never been validated.
11
21/04/04
Proper combustion models are necessary to generate correct heat release rates (thus temperatures) and
species. To achieve proper temperature predictions it is also necessary to adequately establish radiative
heat transfer. For radiation to be properly modelled the most important aspects are temperature and soot
concentrations and morphology. Significant work on the development of combustion models, radiation
models, and soot models is currently underway (NIST, BRE, Cranfield, Sandia National Laboratories,
University of Utah, etc.). These models are being incorporated into numerical tools on a constant basis.
Currently, existing combustion models have been validated only in simple scenarios and with very limited
diagnostics. Common validations rely on simple comparison with temperature measurements (Grandison
et al, 2003) that in many cases are decades old (Steckler, et al. 1982). These validation exercises clearly are
not sufficient to determine the adequacy of the complex models proposed.
Time scales more relevant to structural behaviour imply in most cases fully developed fires. None of the
existing CFD codes has been properly validated under these conditions. The data available for postflashover, fully developed fires, is generally in the form of average punctual measurements of temperature
(Thomas et al., 1972, Thomas, 1972) which is more suited for the validation of Zone Models than of CFD
codes. Combustion and soot models are greatly sensitive to the burning conditions therefore the
capability of existing model to provide reasonable predictions under fully-developed fire conditions
remains un tested.
Independent of the model used all numerical tools are severely limited by an improper definition of the
fundamental properties of materials controlling fire growth. An analysis of the input variables for all flammable
materials shows a systematic dependence of simple and very approximate databases (i.e. Drysdale, 1999,
Quintiere, 1985, Tewarson, 2003). The errors that can be induced by an improper or incomplete selection of
material properties can be more important than those generated by an improper use of the parameters of the
turbulence model.
These general limitations to these codes are by no means insurmountable but improvement and confidence can
only be achieved with systematic and precise validation. In their current state, all CFD codes are research tools that
require great proficiency in their use and by far the biggest challenge is to guarantee that the users are making a
proper use of these tools.
Conclusions
A review of the different approaches used to establish the thermal boundary condition required to properly
analyse a structure in the event of a fire has been presented. A series of general comments on the validity and
limits of the different current approaches has been provided. These general comments give a guideline of areas
that need further attention. From this evaluation it seemed to emerge that the only way to properly model the
thermal boundary condition is via numerical models and that many of the assumptions embedded in current
calculations have not been fully validated. Numerical models can also be useful for this purpose.
A review of the most commonly used modelling approaches then reveals that currently these models also have
significant limitations. Many of these limitations can be circumvented by proficient and experienced use but lack of
detailed validation still remains a serious problem. The limitations together with the complexity of the models
imply that at this stage all computer-based models for compartment fires are at a level of development that
enables their use only by qualified users and should not be promoted as design tools for general use. Especial
mention has to be made of CFD based tools where improper definition of the input parameters and user variables
can result in extremely poor answers.
It is important to note that these are complex tools thus improvements in most cases will have to be seen within
the context of specific tools.
References
Atreya, A. Convection Heat Transfer, Chapters 1-4 in SFPE Handbook of Fire Protection Engineering, 1988,
National Fire Protection Association.
Bilger, R.W. The Structure of Turbulent Nonpremixed Flames, Proceedings of 22nd Int. Symp. On Combustion,
1988, The Combustion Institute.
Bressloff, N.W., Moss, J.B., Rubini, P.A. CFD Prediction of Coupled Radiation Heat Transfer and Soot Production
in Turbulent Flames, Proceedings of 26th Int. Symp. On Combustion, 1996, The Combustion Institute.
Cooper, L.Y. Fire-Plume-Generated Ceiling Jet Characteristics and Convective Heat Transfer to Ceiling and Wall
Surfaces in a Two-Layer Zone-Type Fire Environment, NISTIR 4705, 1991, NIST.
Cox, G. and Kumar, S., Modeling Enclosure Fires Using CFD, Chapter 8, 3.194-3.218, SFPE Handbook for Fire
Protection Engineering, 3rd Edition, 2002.
12
21/04/04
Deardorff, J. W., A Three-Dimensional Numerical Study of Turbulent Channel Flow at Large Reynolds Numbers,
Journal of Fluid Mechanics, 41, 1970.
Drysdale, D.D. An Introduction to Fire Dynamics, 2nd Edition, John Wiley and Sons, 1999.
Forney, G. P. Computing Radiative Heat Transfer Occurring in a Zone Fire Model, NISTIR 4709, 1991, NIST.
Friedman, R., An International Survey of Computer Models for Fire and Smoke, SFPE Journal of Fire Protection
Engineering, 4 (3), 1992, p. 81-92.
Grandison, A.J., Galea, E.R., Patel, M.K. Development of Standards for Fire Models: Report on Phase 1
Simulations, Office of the Deputy Prime Minister, FRD Publication Number 1/2003.
Grandison, A.J., Galea, E.R., Patel, M.K. Development of Standards for Fire Models: Report on SMARTFIRE Phase
2 Simulations, Office of the Deputy Prime Minister, FRD Publication Number 2/2003.
Jones, W.W., Forney, G.P., Peacock, R.D., Reneke, P.A. A Technical Reference for CFAST: Eng. Tools for
Estimating Fire Growth and Smoke Transport, 2000, NIST.
Joulain, P. Proceedings of the Combustion Institute, v.28, 1998.
Law M. and OBrien T. Fire and steel construction: Fire safety of bare external steel, The Steel Construction
Institute, 1986.
Lewis, M.J., Moss, M.B., Rubini, P.A. CFD Modelling of Combustion and Heat Transfer in Compartment Fires.
Proceedings of 5th Int. Symp. On Fire Safety Science, March 1997, IAFSS.
McCaffrey, B.J. and Rockett, J.A., Journal of Research, National Bureau of Standards, 82: 107 1977.
McGrattan, K.B., Baum, H.R., Rehm, R.G., Hamins, A., Forney, G.P., Floyd, J.E., Hostikka, S., Prasad, K. Fire
Dynamics Simulator (Version 3) Technical Reference Guide, 2002, NIST.
McGrattan, K.B., Forney, G.P., Floyd, J.E., Hostikka, S., Prasad, K. Fire Dynamics Simulator (Version 3) Users
Guide, 2002, NIST.
Musser, A., McGrattan, K., Palmer, J. Evaluation of a Fast, Simplified Computational Fluid Dynamics Model for
Solving Room Airflow Problems, 2001, NIST.
Novozhilov, V. Computational Fluid Dynamics Modelling of Compartment Fires. Progress in Energy and
Combustion Science, 2001, Elsevier Science Ltd.
Olenick, Stephen M., and Carpenter, Douglas J., "An Updated International Survey of Computer Models for Fire
and Smoke," SFPE Journal of Fire Protection Engineering, 2002.
Orszag, S., Staroselsky, I. and Yakhot V., Some Basic Challenges for Large Eddy Simulation Research, Large
Simulation of Complex Engineering and Geophysical Flows, Orszag, G. B. et al., 1993.
Peacock, R.D., Reneke, P.A., Jones, W.W., Bukowski, R.W., Forney, G.P. A Users Guide for FAST: Eng. Tools for
Estimating Fire Growth and Smoke Transport, 2000, NIST.
Petterson, N.M. Assessing the Feasibility of Reducing the Grid Resolution in FDS Field Modelling. Project Report
for Partial Fulfillment of M.E. degree at the University of Canterbury, 2002, School of Engineering, U. of
Canterbury.
Petterson, O., Magnuson, S.E. and Thor, J., Fire Engineering Design of Structures, Swedish Institute of Steel
Construction, Publication 50, 1976.
Prasad K., Li C., Kailasanath K., Ndubizu C., Ananth R., Tatem P.A. Numerical modelling of methanol liquid pool
fires. Combustion Theory and Modelling, 1999.
Quintiere, J.G., Fire and Materials, 5,2,pp.52-60, 1982.
Quintiere, J.G., Fire Behavior in Building Compartments, Proceedings of the Combustion Institute, vol. 29, 2002.
Rockett, J.A., Combustion Science and Technology, 12: 165, 1976.
Seigel, R. and Howell, J.R., Thermal Radiation Heat Transfer, Hemisphere Publishing Corporation, 1981.
Smagorinsky, J., General Circulation Experiments with Primitive Equations I: The Basic Experiment, Monthly
Weather Review, 1963.
SOFIE, Simulation of Fires in Enclosures, 2003. Web Site: www.cranfield.ac.uk/sme/sofie.
SOFIE Technical Summary. School of Mechanical Engineering, Cranfield University, 2003.
Steckler, K.D., Quintiere, J.G. and Rinkinen, W.J., Flow induced by fire in a compartment, NBSIR-82-25-20,
National Bureau of Standards, 1982.
Tewarson, A., Generation of Heat and Chemical Compounds in Fires, Chapter 3, 3.82-3.161, SFPE Handbook
for Fire Protection Engineering, 3rd Edition, 2002.
13
21/04/04
Thomas, P.H. and Heselden, A.J.M., Fully-Developed Fires in Single Compartment A Co-operative Research
Programme of the Conseil International du Batiment (CIB Report No 20), Fire Res. Statation, UK, FR Note No. 923,
Aug. 1972.
Thomas, P.H., 14th Symp. (Int.) on Combustion, Comb. Inst., 1007, 1972.
Thomas, P.H. and Nilsson, L., Fully Developed Compartment Fires: New correlations of Burning Rates, Fire
Research Sta., UK, FR Note No 979, August 1973.
Torero, J. L., Quintiere, J. G. and Steinhaus, T., Fire Safety in High-rise Buildings: Lessons Learned from the WTC,
Jahresfachtagung der Vereingung zur Forderrung des Deutschen Brandschutzez e. V., Dresden, Germany, 2002.
Usmani, A.S., Rotter, J. M., Lamont, S., Sanad, A.M. and Gillie, M., Fundamental Principles of Structural
Behaviour Under Thermal Effects Fire Safety Journal, 36, 721-744, 2001.
Walton, W.D., Zone Computer Fire Models for Enclosures, Chapter 7, 3.189-3.194, SFPE Handbook for Fire
Protection Engineering, 3rd Edition, 2002.
Welch, S., Rubini, P.A. Three-dimensional Simulation of a Fire-Resistance Furnace. Proceedings of 5th Int. Symp.
On Fire Safety Science, March 1997, IAFSS.
Woodburn, P.J., Britter, R.E. CFD Simulations of a Tunnel Fire Parts I and II. Fire Safety Journal, 1996 (26),
Elsevier Science Ltd.
14
21/04/04
1. INTRODUCTION
This paper presents a brief history of structural fire resistance design from the origins of prescriptive
recommendations to performance based engineering over the last 100 years.
The paper will introduce the reader to the key concepts of structural fire engineering and the limitations of a
prescriptive approach. It will also highlight the most significant advances in the last 10-15 years, validation and
acceptance of computer modelling in the design office to predict the whole frame structural response of buildings
to fire. Finally, the importance of a robust engineering approach to fire resistance design of tall buildings post 911, will also be discussed.
2. HISTORY
A selection of the most significant advances in structural fire engineering are listed in Table 1. The authors
recognise that this list is not exhaustive but is representative of the type of research that has been carried out in
this field in the last 100 years.
Historically, fire resistance design of structures has been based upon single element behaviour in the standard fire
resistance test. The significant differences between the standard fire heating curve and a Temperature-time (T-t)
relationship produced in a real fire have long been recognised. Compartment fire models as well as the time
equivalence concept have tried to address this shortfall for decades.
Law and OBrian (1986) considered the preferential heating experienced by external steel to allow the Pompidou
Centre in Paris to be built with an unprotected external steel frame.
The Broadgate fire (SCI 1991) in London was a catalyst for the Cardington frame fire tests in the 1990s before the
greatest incentive for robust design of tall buildings in fire; the WTC collapse, 2001 (FEMA 2002).
Historical Event
Standard furnace testing developed
Fire load concept
Compartment fire testing, research and modelling
Researcher
Ingberg
Kawagoe
Magnusson & Thelandersson
Date
1900
1928
1963
1970
Law
Pettersson
Pettersson
Law and OBrian
1971
1976
1976
1986
1990
1990
1991 onwards
Schleich, Cajot and PierrePROFIL ARBED
BRE and CORUS
1994-1999
1994-2003
September 11th 2001
15
21/04/04
These limits are based on the size of the furnace and tend to be the maximum deflection that can be recorded
without causing damage to the furnace.
Therefore in a code compliant building in the UK with all structural elements protected the floor is permitted to
deflect up to L/20. For a 7.5m beam this equates to 375mm and for an 18m beam this is 900mm.
The fire resistance of the element is taken as the time to the nearest minute, between commencement of heating
and failure. Periods of fire resistance are normally specified as hour, 1 hour, 1.5 hours, 2 hours, 3 hours and 4
hours.
Figure 2 Comparison of the standard fire curve and real fire temperature -time histories. The fire load is in kg/m2
and the ventilation is a fraction of one wall. (Drysdale 1999)
16
21/04/04
ventilation, fire load, compartment shape). The course of temperature rise within similar protected steel elements
at the same locations in standard fire tests were then examined and compared with those from the room-burn
experiments. The equivalent fire exposure is defined as the time at which in the standard test the temperature
reached the same maximum level as in the room-burn experiments.
Law (1971) developed a time equivalence relationship to include the effect of ventilation using data gathered from
a CIB (Conseil Internationale du Btiment) study of fully developed compartment fires (Thomas and Heselden
1972). Pettersson et al (1976), adopted Laws method to time-equivalent and developed a further expression using
the family of calculated T-t curves for particular compartments derived by Magnusson and Thelanderson (1970).
Petterssons time-equivalence approach takes into consideration the effect of window height and the thermal
inertia of the compartment wall lining.
Traditional office
Figure 3
2.2.2 Eurocode 1
The most recent approach to time-equivalence is Eurocode 1. The equivalent time of fire exposure is defined in the
Eurocode as:
t e , d = q f , d .m. q1. q 2 . n .k b .k c w f
(1)
where,
te,d = equivalent fire resistance (minutes).
qd = fire load density (MJ/m2 ).
m = combustion factor (default=0.8).
q1 = factor of consequence of structural failure based upon occupancy and compartment floor area.
q 2 = factor to account for probability of occurrence of a fire based on fires reported to the fire service.
10
n = ni = factor taking into account the different active fire fighting measures (see Table 2).
i =1
k b = factor applied to account for the insulation properties of the compartment enclosure.
17
21/04/04
k c =correction factor function of the material composing structural cross sections e.g. kc =1.0 for concrete.
wf = ventilation factor
The ventilation factor in the Eurocode is defined in Equation 2. Where the term v is the area of the vertical
openings in the compartment walls divided by the floor area, and is limited to a maximum of 0.25 by the
Eurocode. H is the compartment height (m).
In a building with exterior glazing this ratio is typically higher than 0.25, which results in a lower ventilation factor,
but this has been ignored in the Eurocode.
wf =
(2)
A detailed comparison of the Eurocode, Pettersson and other available ventilation factors has been undertaken for
real fire tests involving fully developed fires (Law 1997). This study demonstrated that Pettersson's formula was the
best representation of the ventilation effect in real fire tests.
Petterssons ventilation factor is defined as follows:
w f = 0.957 Af / At Av h 0. 5
0. 5
(3)
Where
Af = floor area of the compartment (m2)
At = total internal surface area of the compartment, including windows (m2)
Av = area of ventilation for the compartment (m2)
h = height of openings for ventilation (m)
Table 2 lists the factors to take account of active fire fighting measures proposed by the latest version of Eurocode
1. The assumed fire load density is multiplied by these factors to achieve a design fire load for the particular
building. The factor to take account of sprinklers, n1 is well known and has been in earlier versions of Eurocode
1. However, the validity of reducing/increasing the fire load using the other factors is questionable.
An essential part of a time equivalent study is a sensitivity analysis. This should include varying the amount of
ventilation available, floor area involved and thermal properties of the boundary wall linings. The most critical
variable is typically ventilation. Modern triple or double glazing systems do not break as readily as single glazing in
a flashover fire. The duration and peak temperature of a fire is significantly affected by available ventilation. By
conducting a sensitivity analysis a range of equivalent times can be calculated and the upper bound value taken as
a credible worst case.
Table 3 illustrates the range of equivalent fire resistance periods that can be calculated using the Eurocode 1 timeequivalent approach depending on the factors applied, the area of the compartment considered and ventilation
considered to be available. The study was carried out for an office building in London that required 2 hours fire
resistance in accordance with the Approved Document B of the Building Regulations (ODPM 2002). Fire resistance
ratings range from 15 to 90 minutes in this particular case.
18
21/04/04
n1
0.61
012
by heat
by smoke
n2
n3
n4
n5
n6
0.87
n7
0.61 or 0.78
n8
n9
0.9 or 1 or
1.5
Smoke
exhaust
system
n10
?n
Fire Load
2
1
2
3
4
5
6
7
8
9
10
11
12
1
1
1
1
0.61*
0.61
0.61
0.61
0.34**
0.34
0.34
0.34
[MJ/m ]
[m ]
[m ]
777
777
777
777
474
474
474
474
270
270
270
270
2600
2600
2106
2106
2600
2600
2106
2106
2600
2600
2106
2106
312
0
312
0
312
0
312
0
312
0
312
0
Time Equivalent
[min]
0.88
1.77
0.77
1.53
0.88
1.77
0.77
1.53
0.88
1.77
0.77
1.53
47.7
96.2
42.1
83.4
29.1
58.7
25.7
50.9
16.6
33.5
14.7
29
19
21/04/04
Figure 4
If the structural element is lightly loaded i.e. under-stressed at the fire limit state, then the failure temperature may
be higher than 550C. This is the basis for performance based design in accordance with BS 5950 Part 8. The
structural elements are considered in isolation of the rest of the building frame.
(3)
The load capacity at 20C
The factors applied to the live and dead loads at the fire limit state have traditionally been 0.8 and 1.0 respectively.
Load ratio =
20
21/04/04
The latest revision of BS 5950 Part 8 was issued in December 2003. The load factors applied to live load at the fire
limit state have been reduced to 0.5 for office buildings allowing greater limiting temperatures for this occupancy.
The Broadgate fire prompted BRE to conduct a large scale test program on an 8 storey composite steel frame at
their test facility in Cardington, UK (see Figure 6). The Cardington Frame fire tests (Kirby 2000) in the 1990s
provided a wealth of experimental evidence about how whole frame composite steel-concrete structures behave
in fire. The Cardington Frame survived a number of full scale fire tests despite having no fire protection on the
steel beams (the unprotected steel often reached temperatures in excess of 900C). The columns were generally
protected to their full height. In all tests there was considerable deflection of the composite floor slab in the region
of the fire.
The main conclusions of the tests and the subsequent research projects were that composite framed structures
possess reserves of strength by adopting large displacement configurations with catenary action in beams and
tensile membrane behaviour in the slab (Huang et al 2000, Bailey and Moore 1999, Usmani et al 2001, The
University of Edninburgh 2000), see schematic representation in Figure 7. Furthermore, for most of the duration
before runaway failure (not observed at Cardington), thermal expansion and thermal bowing of the structural
elements rather than material degradation or gravity loading govern the response to fire (Usmani et al 2001).
Large deflections were not a sign of instability and local buckling of beams helped thermal strains to move directly
21
21/04/04
into deflections rather than cause high stress states in the structure. Near failure, gravity loads and strength will
again become critical factors.
(b) Axially restrained
22
21/04/04
These phenomena can be captured in finite element models and structures designed for fire in an engineering
manner. In the case of low risk, medium rise buildings this can lead to savings in passive fire protection.
23
21/04/04
Masonry facade
Model 1
Stability of structure maintained throughout the design fire. This is primarily assessed by looking at the
rate of deflections during the fire. Runaway deflections (a rapid increase in the rate of deflection) would
indicate failure of the floor.
Horizontal compartmentation is also assessed by monitoring the rate of deflection of the composite
floor. A rapid increase in deflection in any region of the floor plate implies compartmentation failure.
Vertical compartmentation via the vertical fire fighting shafts (a fire rated shaft required in the UK in
buildings with a floor 18m above fire service access level to provide a place of relative safety on each
floor for fire fighters) is assessed by monitoring the connections at the shaft wall to ensure that they
maintain their capacity for the fire period.
24
21/04/04
Columns
acting over 2
floors
Shell elements
representing the slab
The material properties used in the model are given in Table 4, full degradation of the stress-strain curves with
temperature was allowed. Values of thermal expansion for steel and concrete were also taken from the
appropriate Eurocodes.
Table 4 The material models
Material
Grade
Model
C30
Eurocode 2
Reinforcing mesh
S460
Eurocode 2
Steel (frame)
S275
Eurocode 3
In accordance with BS 5950 Part 8 (1990), at the fire limit state, the partial factors to be applied to live and dead
load are 0.8 and 1.0 respectively. Thus the load assumed to act over the floor slab of a typical office floor in the
model is 7.85kN/m2.
The boundary conditions assumed in the model were as follows:
Columns were fixed at their base and restrained in the horizontal directions but free to deflect
downwards at the top. These boundary conditions simulate the continuity of the columns at the base of
the structure and at the top of the columns.
Symmetry boundary conditions were applied along the sides of the model parallel to the secondary
beams.
In this model the short secondary beams were assumed to be axially restrained by the masonry faade
but rotationally free. In other models not described here, it was conservatively assumed that the
masonry wall provided no restraint because the restraint stiffness and reliability of the connections to
the masonry faade was unknown.
Four noded shell elements were used to represent the slab. Two-noded beam elements were used to represent the
beams, columns and slab ribs. Each element was associated with its appropriate section properties and material
characteristics.
The columns were modelled on the fire floor and the floor above (see Figure 11). Slab shell elements are not
connected to columns because stress can flow around the column as a result of slab continuity. Slab elements
are connected to beam elements using constraint equations between the beam and slab representing full
composite action.
25
21/04/04
Prot column
Prot short secondary beam
Unp secondary beam
Prot primary beam
Prot secondary beam
Standard fire curve
1000.0
Temperature (C)
800.0
600.0
400.0
200.0
0.0
0
1000
2000
3000
4000
5000
6000
Time (sec)
Figure 12
For each structural analysis it was assumed there was no gradient through the depth or along the length of the
steel beams and columns. For structural behaviour of composite frames the most important gradient is that
between the slab and the protected and unprotected steel beams. The gradient over the depth of the beam is
much less important because it is very small in comparison. The slab was assumed to be at a uniform through
depth gradient over the whole compartment.
3.4 RESULTS:
A contour plot of the deflection at the end of heating is shown in Figure 13 for the case where the slab and
beams are axially restrained by the masonry wall. Blue and red shades highlight the greatest deflections both
negative and positive respectively. The greatest downward displacement is near the mid-span of the unprotected
secondary beams as expected. The position of the columns are clearly visible. The structure is very stiff at the
corner of the building where the short protected secondary beams make a stiff closely spaced grid. There is very
little displacement in this region.
The mid-span displacement of a typical unprotected secondary beam is shown in Figure 14. It is plotted against
unprotected secondary beam temperature. The rate of deflection is very linear similar to deflection plots from the
Cardington tests. Runaway failure (a rapid increase in the rate of deflection) is not observed.
26
21/04/04
490mm
27
21/04/04
The total strains (thermal + mechanical) in the slab at reinforcement level are plotted in Figure 16 for the
Y(=2) direction. Compressions have negative values and tensions positive values. In general the slab is in
compression or low tension.
Thermal strains will account for about 0.1-0.3% of the total strain values. There are regions of relatively high
tension (2-3%) near the core as expected. These are mainly as a result of the hogging moment at this boundary.
Any localised concrete cracking in this region would relieve hogging moments although strains will still be present
after cracking as the deflecting slab pulls on the supports. The ability of the core connection to cope with the
conditions at the fire limit state was tested by a detailed connection model the results from which will be reported
elsewhere.
~0.1%
~2%
1.1
Figure 16
Strain in the 2 (Y) direction at the level of the reinforcement in the slab after 90 minutes
of the standard fire exposure. C=compression, T=tension.
28
21/04/04
390mm
Figure 17
The deflection at mid-span of a typical secondary beam is very linear i.e. the rate of deflection is not changing (see
Figure 18). The same behaviour was shown in Figure 14 when the beams were unprotected although the
deflections were much greater. This suggests the structure is very stable. The uniform rate of deflection was also
observed in the measurements made at Cardington during the fire tests.
The strains in the 2 direction are shown in Figure 19. Tensile strains along the core edge are in the region of 2%.
The greatest tensile strains are around the column locations and at the core wall. The strains experienced in the
slab when all beams are protected are very similar to the design case with secondary beams unprotected. It could
be expected therefore that the slab would also experience local cracking in the fully protected case.
Figure 18
29
21/04/04
C
C
0.2%
T
1%
Figure 19
T=tension.
Strain in the 2 (Y) direction at the level of the reinforcement in the slab. C=compression,
30
21/04/04
rotations caused by a severe thermal gradient in a composite slab will often cause cracking in concrete adjacent to
columns and yielding of steel beam lower flanges. These local failures will not normally spread any further. In this
instance the load carrying mechanism may not greatly change and there would generally be a small redistribution
of loading.
An increased understanding of these load carrying mechanisms and failure criteria in this case study will lead to
more robust design for tall buildings.
Figure 20
Figure 21
5. CONCLUSION
This paper provides a short history of structural fire engineering from prescriptive to performance based design.
Our understanding in this field of research and development has increased significantly in the last 20 years in
particular. Until the 1990s structural fire design was essentially based on critical steel temperatures and single
element studies based on loss of material strength and stiffness. Now 3D modeling using finite element analysis is
being used in research and consultancy to engineer fire resistance.
This paper discussed a case study which provided a snapshot of information and analysis to demonstrate the
performance based passive fire protection design for an office building in London, satisfies the appropriate
functional requirements of the Approved Document B of the Building Regulations, UK. This study uses state of the
art structural fire engineering analysis to provide a cost efficient design whilst increasing safety because the true
response to fire has been modeled and is understood.
In case study 1 a detailed finite element analysis of the structure with a standard fire was carried out to determine
the deflections and forces in the structural elements. The following assumptions were made:
All primary and secondary steel beams are composite with the floor slab through shear studs.
All primary and edge beams will be protected for 90 minutes fire resistance
All columns will be protected to their full height including connections for 90 minutes fire resistance.
31
All steelwork within the core will be protected for 90 minutes fire resistance.
Reinforcing mesh within the concrete slab will be adequately lapped in all areas.
21/04/04
A direct comparison was made between the structural behaviour of the proposed design case with secondary steel
left unprotected and the structural response if all steel had been fire protected as would be the case in a
traditional prescriptive design. The results of the fully protected model were not vastly different from the proposed
design case and clearly showed that any fire protection on the secondary beams was redundant.
In particular the deflections predicted for the fully protected structure were very similar to the proposed design
solution with bare secondary steel beams.
The comparative analyses have shown that the deflection and strain patterns in the composite slab are very similar
for both protection arrangements therefore it could be assumed that the damage to the structure would be similar
in both cases.
The comparative study was invaluable in the approvals process because the fire brigade, insurers and approving
authorities could quantify the differences in response between the design they would normally approve, because it
complies with prescriptive guidance, and the performance of the proposed design with some bare steel.
This type of design should be carried out on a case by case basis and results in this paper are applicable to the
office building only.
Design of tall buildings has changed since 9/11. Clients demand and an understanding of structural behaviour in
fire to ensure an appropriate level of life safety design for tall buildings with robust structural detailing and
increased fire protection if necessary in critical areas. This could not be accomplished using simple critical
temperature calculations and single element behaviour. It can be accomplished using finite element modeling in 3
dimensions considering nonlinear geometry and the forces as a result of the heating regime.
REFERENCES
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
ABAQUS www.hks.com
Bailey C.G. and Moore D.B. The behaviour of full-scale steel framed buildings subject to compartment fires.
The Structural Engineer. 77(8), pp. 15-21, 1999.
BS 476 : Part 20 : 1987 Fire tests on building materials and structures.
BS 5950:Part 8:1990 Structural use of steelwork in building. Code of practice for fire resistant design.
BS 5950:Part 8:2003 Structural use of steelwork in building. Code of practice for fire resistant design.
BS EN 1991-1-2:2002. Eurocode 1: Actions on structures-General actions-Actions on structures exposed to
fire.
Cadorin J-F. and Franssen J-M. A tool to design steel elements submitted to compartment fires-Ozone V2. Part
1: pre and post flashover compartment fire model. Vol 38 No. 5, pp395-427, 2003.
DD ENV 1992-1-2:1996. Eurocode 2: Design of concrete structures Part 1.2 General rules-structural fire
design, 1996.
DD ENV 1993-1-2:2001. Eurocode 3 Design of steel structures Part 1.2: Fire resistance, 1995.
Drysdale D.D. An Introduction to fire dynamics, John Wiley and Sons, 2nd Edition 1999.
Federal Emergency Management Agency (FEMA). World Trade Center Building Performance Study: Data
Collection, Preliminary Observations and Recommendations. FEMA 403, May 2002.
Huang Z., Burgess I.W. and Plank R.J. Non-linear modelling of three full scale structural fire tests. In First
International Conference, Structures in Fire, Copenhagen, June 2000.
Ingberg S.H. Fire loads. Quarterly Journal of the national Fire Protection 1. Guide to the application of fire
safety engineering principles.
Kawagoe K. and Sekine T. Estimation of fire temperature-time curves rooms. Technical Report 11, Building
research Institute, Ministry of Construction, Tokyo, Japan, 1963.
Kirby B.R. British Steel data on the Cardington fire tests. Technical report, British Steel, 2000.
Law M. A relationship between fire grading and building design and contents. Technical Report 1971.
Law M. and OBrien T. Fire and steel construction: Fire safety of bare external steel, The Steel Construction
Institute, 1986.
Magnusson S.E. and Thelandersson S. temperature time curves of the complete process of fire developmenttheoretical study of wood fuel fires in enclosed space. Technical Report 65, 1970.
Office of the Deputy Prime Minister (ODPM) Fire Safety to Approved Document B, Building Regulations, 2000.
Pettersson O., Magnusson S.E. and Thor J. Fire engineering design of steel structures. Swedish Institute of
Steel Construction, Publication 50, Stockholm, 1976.
32
21/04/04
21. Quintiere J.G., di Marzo M. and Becker R. A suggested cause of the fire induced collapse of the World Trade
Towers. Fire Safety Journal Vol. 37, No. 7, pp707-716, 2002.
22. Schleich J.B., Cajot L.G. and Pierre M. Competitive steel buildings through natural fire safety concept. Final
Technical Report composed of 2 volumes. PROFIL ARBED Centre de Recherches,, March, 1999.
23. The Steel Construction Institute. Structural fire Engineering, Investigation of Broadgate Phase 8 Fire. Technical
Report, June 1991.Broadgate
24. The University of Edinburgh (2000) Final report of the DETR-PIT project: Behaviour of steel framed structures
under fire conditions. Technical report. www.civ.ed.ac.uk/research/fire/project/main.html
25. Thomas P.H. and Heselden A.J.M. Fully developed fires in single compartments a co-operative research
programme of the Counseil International du Batiment (CIB) 1972.
26. Usmani A.S., Chung Y.C. and Torero J.L. How did the WTC towers collapse: a new theory. Fire Safety Journal.
Vol 38, pp 501-533, 2003.
27. Usmani A.S., Rotter J.M., Lamont S., Sanad A.M. and Gillie M. Fundamental principles of structural behaviour
under thermal effects. Fire Safety Journal, Vol. 36, No. 8 pp 721-744, 2001.
33
21/04/04
34
21/04/04
The zonal smoke filling assumption may be fine for relatively small compartments such as domestic rooms or small
offices but can be quite misleading for much larger ones or ones of complex geometry. For these problems the
alternative modelling approach is more appropriate.
Field models, by contrast, make essentially no assumptions about the fluid dynamics of the combustion products.
They avoid resorting, as far as possible, to experimental correlations and instead return to first principles to solve
the basic laws of physics for the fluid flow. They do this on a numerical mesh of maybe millions of elementary
control volumes throughout the calculation domain.
As a consequence of this resort to first principles, this type of model is of universal applicability. It is equally
applicable to the conditions inside the rooms of a building as it is to the evolution and dispersion of the fire gases
in, for example the atmospheric boundary layer.
Their starting point is the "exact" system of coupled partial differential equations that describe the balance
between the competing influences on the transport of mass, momentum, chemical species and energy within the
fire and throughout any enclosure containing it. However, rigorous solution of these "exact" equations, resolving
fully the length and time scales that occur in the turbulent reacting flows characteristic of fire, is still beyond the
capabilities of even the largest computers currently available except of the simplest of problems. To capture the
details of the chemical reaction zone in a fire would require a characteristic mesh size below one millimetre. As a
consequence it is necessary to simplify the system of "exact" equations by some form of modelling.
35
21/04/04
readily accessible computer power. It is likely that this approach will become increasingly attractive as computer
power develops further.
There is an even more computationally demanding methodology known as the Direct Numerical Simulation (DNS)
approach that should be the most accurate. This attempts to capture all the length scales involved in the
underlying physics and chemistry without resort to any turbulence modelling at all. Models of this type are still a
very long way from practical application to fire. This is because computer power, despite the enormous progress
of the last few decades, is still inadequate to allow rigorous simulation of all the scales for domain sizes of
practical interest to the fire engineer. There is though active fire research using this approach which is assisting in
our understanding of the contributory phenomena involved [eg 3] but such models are unlikely to yield practical
tools in the near term.
An appreciation of the degree of temporal resolution that these different methodologies capture is illustrated by
the schematic Figure 1. The fluctuating gas velocity at a particular point in the fire is shown. It has a relatively
slow underlying increase from left to right as the fire increases in its heat release rate. But in addition to this
general trend there are fluctuations of different wavelength. They represent different aspects of the phenomenon
as a whole. The degree of resolution of the contributions made by the different wavelengths achievable by the
various approaches is illustrated.
In addition to uncertainties associated with the modelling of turbulent flow, others are also introduced as a result
of the numerical methods used to solve the continuous equations, no matter which strategy is adopted.
Recognition of the limitations in each of all these departures from rigour is essential for the successful practical
exploitation of CFD to solve fire problems. Best practice guidance on the use of these types of model is now
available for RANS models [4] and will be an urgent requirement for LES models as they become increasingly
attractive to design practitioners.
Fig. 1 Schematic of gas velocity at one point in a fire. The capability of the different CFD methodologies to
capture length scales is illustrated
VALIDITY OF MODELLING
The general issue of the 'validity' of simulations is something that the American Association for Aeronautics and
Astronautics has addressed in a recent guide [5]. They use the following definitions of validation and verification:
Validation
the process of determining the degree to which a model is an accurate representation of the real
world from the perspective of the intended users of the model.
Verification
the process of determining that a model implementation accurately represents the developer's
conceptual description of the model and the solution of the model
In the fire context, for example the turbulence treatment by RANS or sub grid scale model in LES, the combustion
and radiation models need to be validated to test their representation of reality. Whether the model equations are
then solved to adequate accuracy etc is then the issue of verification.
The fire literature contains many comparisons of CFD predictions with experimental data. They contain elements
of both validation and verification. Most, however, have been conducted with a prior knowledge of the
experimental results. One of the most important model validation exercises was recently conducted under the
auspices of the Fire Commission of CIB (Conseil International du Batiment). It involved a series of 'blind'
simulations of unpublished experiments conducted in the 1980s by VTT in Finland [6].
Two wooden cribs were located inside an enclosure containing a single high level slot opening (Figure 2). One of
these cribs was ignited and fire was allowed to spread from the first to the second crib. This is a particularly severe
test of the modelling methodology since the fire reaches flashover after 20 minutes or so and continues to burn
for approximately 2 hours. This is a severe challenge to both the combustion model and also to the computer
hardware which needs to compute transient predictions for such a long period. This is very relevant for the
prediction of conditions within the World Trade Centre Towers. Although the floor area and the heat release rates
are substantially larger in the Twin Towers the fire conditions will not be dissimilar.
36
21/04/04
plan view
ignition point
A
corner crib
centre crib
1.6 m
3.0 m
7.2 m
1.6 m
concrete beam
A
vertical section at A-A
0.6 m
concrete
beam
1.24 m
opening
3.6 m
0.8 m
7.2 m
Fig. 2
Other than the geometry and the thermal properties of the wall materials, the only information supplied for the
blind simulations was the measured individual mass loss rates for each crib together with an effective heat of
combustion. This aspect of prediction provides additional complications which were not to be tested in this
exercise. Mass loss rates were determined from the raw weight loss data through which a smooth curve had been
fitted and time derivatives determined. Figure 3 shows the resultant mass loss rates for the two cribs. A
measurement of the effective heat of combustion from oxygen depletion calorimetry throughout the duration of
the fire was also supplied (Figure 4).
A selection of the results from the JASMINE model is illustrated here (Figures 5-8) [7]. It will be seen that gas
temperatures and major gas species concentrations have been reasonably well reproduced. Although not perfect,
agreement can be seen to be generally acceptable for many practical purposes. The discrepancy evident in the
minimum oxygen concentrations is due to instrumentation sensitivity limitations at low concentrations.
37
21/04/04
0.3
-1
0.25
corner crib
centre crib
0.2
2.5
2
0.15
1.5
0.1
1
0.05
Fig. 3
3.5
0.5
0
Mass loss
and heat release rates of the two wooden cribs
0
20
40
60
80
0
120
100
Time (min)
-1
30
25
20
15
10
time-dependent ('true')
5
0
0
20
40
60
80
100
120
Time (min)
1250
Fig. 4
25
20
measurement
15
JASMINE
10
5
0
0
10
20
30
40
50
60
70
80
90
Time (min)
25
20
15
measurement
10
JASMINE
5
0
0
10
20
30
40
50
60
Time (min)
38
70
80
90
Fig. 6
21/04/04
Comparison of predicted and measured O2 volume fraction near the centre of the ceiling
1400
T1
Temperature (oC)
1200
T2
1000
800
600
T1
T2
T1
T2
400
200
measurement
measurement
JASMINE
JASMINE
0
0
Fig. 7
10
20
30
40
50
60
70
80
90
Time (min)
-2)
F2
F1
50
40
F1
F2
F1
F2
30
20
measurement
measurement
JASMINE
JASMINE
10
0
-10
Fig. 8
10
20
30
40
50
60
70
80
90
Time (min)
Predicted and measured surface fluxes from gas to 'normal' density concrete bores
The poorest performance of the model was in predicted total heat fluxes to the bounding surfaces (Figure 8).
Although reasonably acceptable in terms of its impact on local gas temperatures, (see Figure 7), it is not sufficient
for use in the study of heat transfer to structural elements or to new fuel about to ignite as a consequence of the
initial fire. Clearly this is particularly crucial to improve upon if these predictions are to be of use for the analysis of
structural response to fire.
The explanation is simple. A crude one-dimensional heat conduction approximation, perfectly acceptable for
smoke movement problems, had been assumed. With a simple linear approximation for the temperature gradient
within the solid it substantially underestimated actual heat fluxes at the surface.
39
21/04/04
Fig. 9 Simulation of smoke spread and human egress in design for Brussels Airport passenger terminal
These types of model are increasingly being looked to for their exploration of heat transfer to structural elements
in fire. As attention is focussed on the merits or otherwise of the fire resistance test field models have been used
to examine conditions within fire resistance furnaces.
Figure 10 shows predicted gas and surface temperature contours in addition to gas flow streamlines for a
commercial fire-resistance wall furnace powered by natural gas and following the standard ISO 834 timetemperature curve. Thermocouple temperatures by which the furnace was controlled were simulated by use of
"virtual thermocouples" to account for thermocouple heat transfer and thermal inertia.
Clearly the need to couple models of the fire and its impact on the structure of a building has been brought into
sharp focus by the wish to analyse the WTC collapse. Until recently this has only been achieved by either replacing
the gas phase simulation by a gas temperature history described by the standard temperature-time curve and
studying the structural response in detail or alternatively using CFD models for the gas phase, calculating heat
transfer into the structure but making simplified estimates of structural response.
There have been attempts to couple CFD fire models to Finite Element structural analysis models but a seamless
coupling has not yet been achieved. BRE has been
40
21/04/04
Fig. 10 Predicted gas and furnace surface temperature contours and flow streamlines
leading European consortia in attempts to deliver such a capability for use by practitioners. Members of these
consortia include VTT (Finland), Labein (Spain), Cranfield University (UK), ProfilArbed (Luxembourg), AGB
(Germany) and Cranfield University (UK), TNO (Netherlands), CTICM (France) and the University of Liege (Belgium).
Figures 11 and 12 show some illustrations of how structural
elements can be included in the overall scheme.
41
21/04/04
500
0
Concrete
Steel
I-
MODELLING IN
DISASTER
INVESTIGATION
Computer
fire simulation has seen
increasing application to fire investigation particularly for large disasters. Zone modelling was used to study the
1980 MGM Grand Fire in Las Vegas with the first dramatic contribution from field modelling [8] being provided
during the inquiry into the 1987 Kings Cross underground station fire.
Here a mechanism for the development of rapid fire spread over the wooden escalators, not considered by
investigators, was suggested by the modelling. The flames from the initial fire on the escalator were predicted not
to rise vertically as might initially be expected but to 'lay down' in the 'trench' of the escalator. It was only with a
study of sensitivity of the results to fire source conditions as well as subsequent physical testing to confirm this
mechanism that the numerical predictions were demonstrated to be correct.
The flames behaved this way only when the fire occupied the full width of the trench. The fluid mechanics of air
entrainment was locally essentially two-dimensional from above and below the fire. Earlier in the progress of the
fire when it was still confined to one side of the escalator the local flow of entrained air was three-dimensional
and the flames did indeed rise vertically) [9]. This phenomenon with hindsight was already well known from
experiences of forest fires accelerating up slopes due to flame leaning This conclusion helped explain the rapid
flame spread over the surfaces of the wooden escalators once the fire had spread across the full tread width.
Modelling is currently central to the current investigations being undertaken by NIST into the circumstances of the
World Trade Centre disaster.
Calculations of the expansion of the initial fireball from the South Tower immediately after aircraft impact have
been performed using both LES and simpler exact solutions of the conservation equations in order to understand
the role of the jet fuel in the fires. The modelling shows that only a relatively small proportion of the aviation spirit
on the aircraft was consumed in these external fireballs leaving the majority inside the building to act as an
accelerant for ignition of its contents and linings. Furthermore scoping calculations using LES have been
conducted to compare predicted and photographic records of the trajectories of the external smoke plumes
against various possibilities for the extent of internal damage and consequent heat release rates of the fires.
Clearly this work is ongoing and much more will be made available in the near future.
42
21/04/04
Similarly simulation forms part of the current deliberations of the French courts on the 1999 Mont Blanc Tunnel
fire and on the inquiry into the 1998 Gothenburg disco fire.
43
whole
44
21/04/04
FIRE STANDARDS
The emergence of a reliable fire modelling capability is beginning the influence the development of fire safety
standards both in terms of how the practice of fire safety engineering is conducted and in the development of
new fire test methods.
The deficiencies of traditional test methods have been known for many years. Not all provide the kind of
quantitative data that can be used by engineers to perform holistic assessments of fire safety allowing them to
weigh alternative fire protection strategies. Only the cone calorimeter has been devised with such a purpose in
mind.
The difficulty with many existing tests is that they only give information on the performance of the product 'in the
test' and not 'in reality'. Often they simply supply pass or fail information only. Such tests are useful to rank
products in the test, maybe for quality assurance purposes, but they do not provide quantitative information that
can be used by the engineer. Furthermore relative 'success' in the test does not necessarily ensure relative 'success'
in the 'real world' application environment.
The expectation is that with a new testing approach coupled to a predictive capability to calculate both 'test' and
'real world' exposure scenarios, then it should be possible to assess performance for a full range of practical
possibilities. Of course the 'real world' comes in too many combinations and variations for all eventualities to be
covered but as in any other form of engineering design appropriate design scenarios can be identified.
Technical Committee 92 Fire Safety of ISO, the International Organisation for Standardisation, which I chaired
until September of 2003 is well advanced in drafting new Standards for Standards documents that will
hopefully deliver a new generation of Standards that can be used for a full engineering analysis.
This will take time but an important start has been made. ISO TC/92 is working with the fire commission of the
International Council for Research and Innovation in Building and Construction, CIB W014, to deliver this strategy.
Prior to the collapse of the World Trade Centre towers this might have been expected to have emerged first from
the standards for flammability or toxic potency. However, following the tragic events of 9/11, it is the fire
resistance tests that have been propelled to the top of the agenda.
The first Fire Safety Engineering standards documents were in 1999 as an 8 part Technical Report (ISO TR 13387
Parts 1-8). These are currently being extended and developed as full standards.
CONCLUSION
The use fire modelling both in support of fire safety design and in fire investigation is growing rapidly. Such
models are no longer restricted to just the academic developers and research institutions. Many building design
offices now have access to commercial general-purpose CFD codes offering the ability to design smoke control
systems with the same tools as can be used to assess building ventilation. The possibility of simulating ambient air
movements both before and after the outbreak of fire also offers a powerful new capability for examination of
early fire detection strategies. With the increasing international trend towards performance-based fire regulation,
such simulation tools will inevitably become increasingly attractive, particularly now that they can be used on
personal computers. Relatively low license and hardware costs now ensure much greater accessibility than ever
before. The inexorable improvement in computer hardware capacity is also likely to influence the type of CFD
model that will be used. Increasingly we will see a shift towards LES type models as they can demonstrate the
levels of validation already demonstrated by RANS models
The tragedy of the World Trade Centre disaster does allow the benefit of our current fire modelling capability to
be demonstrated and will hopefully also open up the way to the development of more robust fire test standards
which, using modelling, should deliver a future built environment that is safer, more sustainable and cost effective
than at present.
21/04/04
REFERENCES
1.
2.
McGrattan KB, Baum, HR and Rehm RG, Large Eddy Simulations of Smoke Movement, Fire Safety J. 30,
161, 1998.
3.
4.
Cox G and Kumar S, Modelling Enclosure Fires using CFD, SPFE Handbook, Chapter 3, to be published,
Society of Fire Protection Engineers, Bethesda, 2001
5.
Guide for the Verification & Validation of CFD Simulations, American Institute of Aeronautics &
Astronautics, Guide G-077-1998, AIAA, Reston,VA, 1998.
6.
Hostikka S and Keski-Rahkonen O, Design and Simulation Reports of CIB W14 Round Robin For Code
Assessment, Scenario B, VTT Building Technology Internal report RTE119-IR-2/1998.
7.
Miles SD, Kumar S and Cox G, Comparison of 'Blind Predictions' of a CFD Model with Experimental Data,
Proc. 6th International Symposium on Fire Safety Science, IAFSS, Boston, 2000.
8.
Simcox S, Wilkes NS and Jones IP, Computer Simulation of the Flows of Hot Gases from the Fire at the
King's Cross Underground Station, Fire Safety J., 18, 49, 1992.
9.
Cox G, Chitty R and Kumar S, Fire Modelling and the King's Cross Fire Investigation, Fire Safety Journal,
15, 7, 1989.
46
21/04/04
47
21/04/04
In most of the cases where an arsonist is trying to intimidate or damage a specific target the arsonist has a narrow
range of potential targets, e.g. Abortion clinics. Such targets can often be specifically identified and protected, in
the same way vulnerable political figures can be protected. For the purpose of this discussion the Pentagon attack
therefore falls into a different category, more akin to military attacks such as that on the USS COLE. 1 Persons
wishing to terrorize or punish a community as opposed to a specific actor may be relatively indifferent to the
specific identity of the target, and can instead select the most vulnerable target for maximum effect. This adds the
complex problem of intentional uncertainty to more traditional types of fire analysis.i The ability of terrorists to
attack the weakest link is critical to the problem of developing an effective regulatory response to terrorist arson
arson, since terrorists can take advantage of a known weakness in our fire safety regulatory system, the
grandfathering of existing buildings.
Grandfathering
Societies often tolerate higher fire risks in older structures because the cost of reducing fire risk in older buildings
is high compared to the benefit of installing safety systems. Current fire safety regulations often tolerate
hazardous older buildings under a belief that the probability of a significant fire actually striking the more
hazardous environment is acceptably low, and social investments in safety are more cost effective when applied to
new buildings. This grand-fathering is a rational act when fire ignition is largely random. For example the World
Trade Center upgraded certain fire safety systems only when a floor on the building was renovated. However
terrorists can single out the weak targets and attack them. Any society that grandfathers hazardous existing
buildings must realize that the risk of terrorist arson corresponds to the weakest link, instead of the
average level of vulnerability.
Effect of Fire
The danger of fire is generally related to its energy output. The chemical energy release generates heat and toxic
gas. The toxic gas acts directly on the individuals in the environment, and the heat not only impacts on the
individuals but also spreads the fire and toxic gas to other materials and can causes damage or total destruction of
the structure. The building provides most of the potential energy for its own destruction. . An uncontrolled fire in
the contents was enough to bring down # 7 World Trade Center, the 47 story building that collapsed after being
ignited by burning debris. A large fire in a fuel rich environment can rapidly become uncontrollable. If an
arsonist can trigger a rapid burnout and collapse of a structure, there is the reasonable possibility of killing
everyone in the structure. Virtually no one in the World Trade center survived form above the fire. Arson can be
used to overcome the systems that are normally installed to protect occupants from the energy stored in our
buildings.
Input energy Energy supplied to the building by the arsonist, normally in the form of burning fuel.2
The input energy required to trigger the conversion can be very small relative to the energy output. The ratio
between these two can be described as leverage.
This analysis is of course limited to regulatory issues, not the political or legal issue of whether such an attack is
terrorism.
2
In the World Trade Center the kinetic energy of the aircraft can also be considered input
48
21/04/04
Leverage- The ratio in a given arson target between any given input energy and the energy output.
(both total output and rate may be relevant)
Fires have leverage because:
1)
Fires are self-sustaining chemical reactions that use fuel already in the building.
2)
The building environment and fuel package arrangement can increase the rate and total energy
release from the fire
In modern buildings the leverage of an arson attack can be many orders of magnitude. The energy output
available to the terrorist depends on the buildings fuel load, and the resistance of the structure and its fire
protection systems. The final energy output from a totally burn out of the contents and collapse of the structure is
essentially pre-defined by the target (although the peak energy output may vary). In any given building the
leverage can be initially described by analyzing the smallest input energy that will cause complete destruction of
the building or a defined portion, or the maximum casualties in an area. Leverage is a crude cost benefit
analysis using energy as the currency. All things being equal (e.g. number of occupants), the highest risk exists in
those buildings with the highest leverage at relevant levels of input.
Thresholds
Levels of input are critical because up to a certain point the energy input is simply absorbed by the building
without releasing the buildings own potential energy. The World Trade Center towers absorbed and redistributed
the forces involved in the airliner hitting the building. Buildings are designed for a variety of stresses, and the
effect of fire can simply replicate other stresses the building is designed to resist. The buildings ability to absorb
energy can be described as a threshold
Threshold quantity of input energy a building can absorb before releasing energy in a self-sustaining
reaction I.E. a quantity of input energy below which the leverage is essentially equal to zero.
Obviously the higher the threshold the larger the input required to release the buildings energy. Existence of a
threshold is a major reason why bombs may be less effective than fire in causing catastrophic injury. . For
example in the 20th century aerial attacks on enemy cities were promoted as a military tactic. But bomb explosions
were insufficient for the desired destructive effect because they could not trigger the buildings energy effectively.
Buildings had high thresholds and bombs had poor leverage. Instead of blasting the cities it turned out to be
easier to burn them. Firebombing was the most effective use of massed aerial assaults. Hamburg, Dresden and
Tokyo all preceded Hiroshima and Nagasaki, and even the atomic bomb was fundamentally simply a big ignition
source.
Fire Growth
Fire leverage is a complex analytical problem because fires normally grow over time in an often unpredictable
manner. Fire risk analysts must use a series of complex assumptions about the fire. The most important issue is
how big is the fire at any given time? And how fast is it growing? Once a fire is above a certain size the
development and growth of the fire follows fairly well understood physical and chemical laws. But the
uncertainties involved in fire growth require all current operational fire models to provide a scenario fire as an
input to the analysis.iii In other words the fire models do not really model fire growth, they model the effects of a
given scenario fire in the building. As a result fire protection engineers still do not have a technically valid
measure for the actual risk of fire development in an actual building under a terrorist assault. Regulators have to
distinguish among several key issues:
Ignition source represents the beginning of self-sustaining combustion. It can be a spark, a match,
or a plane crash
initial fire is an analytical tool designed to describe the anticipated fire growth in the structures. It is
routinely specified in terms of a fire growth curve, often described as a t-squared fire. The heat release
rate of such fires grows proportionately to the square of the time period. Fires are characterized in the
literature as ranging from slow to ultra fast iv Such fires are useful analytical tools, but rest on
assumptions.
The ignition source is thus the terrorist attack, and the initial fire is the assumed fire size and growth rate. The
connection between the two is often a matter of judgment and has high uncertainties. Many buildings are not
designed to even deal with a fast t-squared fire. But an arsonist can create an ultra fast or even faster fire that can
overwhelm a buildings defenses.
49
21/04/04
Fire Power
The burning rate and quantity of the flammable material in its environment determines the power output of the
fire.v
Ignitability refers to the ease of ignition. This is the tendency of a material to ignite easily when
exposed to a flame. It is related to the chemical makeup, thermal inertia and physical structure of the
material. Materials vary widely in their ignitability, and there exists no accurate common test method for
defining ignitability.
Fuel load is the effective heat of combustion times the mass of material. It represents the available
energy contained in the contents measured in terms of BTU/pound or kilojoules per kilogram in various
calorimeters. Ordinary cellulostic and hydrocarbon materials have a fairly narrow 2-1 range E.g Wood is
20,000 Kj/kg, coal is 30,000 and oil is 40,000.
The relationship between ignitability and fuel load is critical to terrorist arson regulation since many common
materials are allowed in buildings because of their relatively low ignitability, rather than low fuel load. In the
early stages of typical fires the slow rate of vapor generation (roughly related to ignitability) is the critical limiting
factor. In later stages of fully developed fires the fuel load become more important, and the air supply can become
the limiting factor.
A conservative course is to assume that a terrorist can create an initial fire large enough that the ignitability
characteristics of the buildings fuel load are far less important than the fuel load itself. This tends to simplify the
regulatory problem, since fuel load can be fairly easily calculated, but it simultaneously emphasizes the
vulnerability of many of our buildings, because most of the current regulation of materials and contents is based
on issues of ignitability rather than fuel load.
Dr. James Quintiere John Bryan Professor of Fire Protection Engineering at the U of Md. suggested the term
firepower in this context
In modern performance based fire codes the term scenario fire is often used to describe the growth rate of such a
normal fire.
50
21/04/04
Normal fires generally start small and often grow slowly in their initial stages. Virtually all modern fire protection
strategies rely on containing or controlling the initial normal fire before it grows to uncontrollable size. When
dealing with normal fires the root cause of fire disaster is normally the failure of the fire safety control system.vi
Fire safety control system Static and dynamic systems build into buildings to minimize the growth
and spread of fire.
If a normal fire grows beyond a certain size it will continue to spread indefinitely in two or even three dimensions
until it runs out of fuel or air. As noted above Ignition is the initiation of burning. While of potential legal
importance, source of ignition is of less importance from a regulatory perspective since it is the spread of
uncontrolled fire from the ignition point that may be critical to the hazard. Instead the initial fire is normally the
object of analytical attention. The object is to keep the initial fire from growing beyond the buildings threshold.
There are a variety of strategies used to keep a fire below a threshold. The traditional approach was to contain the
fire in the compartment of origin.
Containment is a system for keeping fires within a specified compartment by physical barriers.
The theory of containment is that a building can be designed to allow full burnout of a compartment without
catastrophic loss. In effect the building threshold is set above the fire from the fuel load in a compartment. Walls
floors doors etc are all tested to determine that they can resist the fire assumed to be in the fuel load of the
buildings. The root technical cause of disaster at the World Trade Center was the failure of containment and
fire resistance of the structure. Containment failed at the World Trade Center for a variety of reasons that are
typical of modern buildings. The failure led to catastrophic collapse.
Operational controls are management systems designed to control contents, inhabitants or related
areas of fire hazards.
It is well known that operational fire safety controls pose very special problems for buildings and the uncertainties
in effectiveness are very large.
The primary technological approach for keeping fires small is a system of automatic fire sprinklers backed up by
fire department response. The primary professional disagreement among fire engineers in dealing with normal
fires is how high to set the building threshold since the larger the fire, the more complex and expensive the fire
protection system and the greater the limitations on the use of the building. As a result designers often claim that
sprinkler systems will keep the fire below a certain size, so the threshold can be set low. But if a fire passes the
critical threshold all material and persons in the compartment or building may be destroyed. If a terrorist can start
a fire larger than the threshold the result can be catastrophic.
Finally some buildings are not designed to survive a significant fire at all. In these buildings the loss of the building
is accepted as long as the occupants can escape. The modern approach in this type of building is to use built-in
suppression and containment systems to keep the fire below the threshold at least for long enough to evacuate
the building.
Multiple Ignitions
In many modern buildings multiple exit paths are provided to ensure a safe exit from a fire. However virtually all
building codes assume a single significant ignition. There has been little or no study to date of the problem of
multiple ignitions. In particular multiple ignitions may require re-examination of the routine design concept of
phased evacuations. Phased evacuations involve planning immediate evacuation of only the persons in the fire
compartment. Person in adjacent compartments would be held back from using the exit paths. This allowed
buildings to be designed with fewer exits. But starting a fire in both compartments destroys the underpinnings for
this strategy.
Sabotage
Many fire protection systems are vulnerable to sabotage. This is particularly true of dynamic systems such as
sprinklers, alarms, exit ways, and fire doors. Techniques are often trivial. Sabotage can range from putting a
wedge under a fire door to shutting off a sprinkler system. Most modern buildings have few if any protections
against sabotage of the fire protection systems. Sabotage can affect a building in several ways. Sabotage defeats
containment, delays evacuation and can allow a fire to grow faster. All can lead to catastrophic loss.
51
21/04/04
The material for a serious arson attack is often cheaper, more easily obtained and less well tracked
than bomb components.
2)
Explosion are a well defined events which have been studied in isolation, rather than the trigger for a
poorly understood fire development process. Our data base of understanding of fire development may be
much more limited.
3)
It is relatively difficult to sabotage the explosion resisting components of a building. Some fire
protection systems can be easily sabotaged and some are vulnerable to misuse by the occupants.
4)
A building which resists an explosion can normally be evacuated since there are no additional energy
inputs. A burning building is not a tenable environment even for rescue forces. Evacuation prior to
collapse may be impossible.
52
5)
21/04/04
Managing the dynamic fire protection features of a building may be much more complex than the passive
explosion resistant features.
As a result the phenomenon of terrorist arson must be carefully examined to allow appropriate risk analysis. One
problem with the analysis is that strategies which are suitable for protecting buildings against bombs may not be
suitable or even counter productive when dealing with arson.
2)
multiple ignitions
3)
The World Trade Center showed all three of these techniques. There were not only two planes, but each plane
was large enough to ignite several compartments. The fuel load of the aircraft contributed to the early heat
output of the fire and the impact destroyed both static and dynamic fire protection systems. The air supply was
enhanced by the large impact hole .
Conclusion
Society cannot solve the problem of arson terrorism through analysis and regulation alone. However, applying
techniques that are available, and developing new ones in critical areas, we can reduce the uncertainties involved
in controlling terrorist arson. Research will be needed on the practicality of analyzing building thresholds,
controlling the available leverage, restricting the potential for sabotage. Perhaps most important, the threat of
multiple ignitions and over reliance on phased evacuation should be the subject of immediate analysis. Many
thousands of people died in the World Trade Center. We must learn as much as possible to avoid a reoccurrence
of this disaster
Acknowledgements
Many people assisted in reviewing various portions of this paper. They Include:
Drs. James Quintiere, Marino Di Marzo and Fred Mowrer, University of Maryland
Dr. Vicki Bier, University of Wisconsin and Mr. Anthony Kilpatrick Glasgow Caledonian University
53
21/04/04
References
i
ii
iii
iv
vi
vii
viii
Brannigan V. Smidts C Performance Based Fire Safety Regulation Under Intentional Uncertainty: Proceedings
of the first International Symposium Human Behavior in Fire Ulster UK 1998 411-420
Quintiere J Di Marzo M and Becker R A Suggested Cause of the Fire-Induced Collapse of the World Trade
Towers J of Fire Research
Cooper LY Compartment Fire -generated Environment and Smoke Filling 3-174-196 The SFPE Handbook of
Fire Protection Engineering, Second Edition, National Fire Protection Association, Quincy MA (1995).
NFPA. (2000). Guide for smoke management systems in malls, atria, and large areas. NFPA 92B. Quincy,
MA: National Fire Protection Association.
Babrauskas, V., Burning Rates (Section 3/Chapter 1), pp. 3-1 to 3-15 in The SFPE Handbook of Fire
Protection Engineering, Second Edition, National Fire Protection Association, Quincy MA (1995).
Watts JM System Concepts for Building Fire Safety Fire Protection Handbook, National Fire Protection
Association, Quincy, MA, 1997.134-41
Designing Terrorist-Resistant Buildings TOD RITTENHOUSE Fire Engineering
U.S. NRC, "Operational Safeguards Response Evaluation (OSRE) Inspection Manual," Inspection Procedure
81110, July 1997
54
21/04/04
APPENDIX A
The Royal Society of Edinburgh (RSE) is Scotland's National Academy of Science & Letters. An
independent body with charitable status, its multidisciplinary fellowship of 1300 men and
women of international standing represents a knowledge resource for the people of Scotland.
Committed to its Royal Charter of 1783 for the advancement of learning and useful
knowledge the Society recognises the important role it can play in todays Scotland. Working
as part of the UK and within a global context, the RSE seeks to contribute to Scotland's social,
economic and cultural wellbeing by:
organising conferences and lectures for the specialist and for the general public on
topics of national and international importance
Inspiring school children in classrooms from the Borders to the Northern Isles and
promoting their interest in science, society and culture.
55
21/04/04
APPENDIX B
56
21/04/04
APPENDIX C
SPEAKERS
Professor Farshad Alamdari PhD CEng FCIBSE
Chief Scientist, Building Research Establishment (BRE)
In 1997 Professor Farshad Alamdari joined the management team of BRE (the Building
Research Establishment), just after its privatisation, as a key business driver in the cultural
change necessary to take a Government research organisation into a commercial researchbased consultancy company. Since then he has managed various research-based businesses
involving environmental and fire safety issues.
In 2001, Farshad was promoted to Managing Director of the fire division of BRE, FRS (formerly
Fire Research Station) and recently to the BRE Chief Scientist.
Obtained his PhD from Cranfield University, Farshad is a Chartered Engineer and is a Fellow
Member of the CIBSE, Visiting Professor at the University of Ulster and Visiting Special Professor
at Nottingham University.
Dr Peter Bressington
Director, Leader Arup Fire International, Ove Arup and Partners Ltd
Peter Bressington is a senior fire engineer who has been leader of Arup Fire in East Asia (based
in Hong Kong) and is currently leader of Arup Fire International based in London. Peter has
acted as an expert witness in fire safety matters where legal proceedings or agreements have
been required. He is Co-Chairman of the Design Criteria and Loads Group on the Council on
Tall Buildings and Urban Habitat, currently he is assisting in re-writing CIBSE Guide E, Fire
Engineering Handbook. He chaired the Arup Extreme Events Task Force set up after 11th
September.
Dr Jose Torero
Reader in Fire Dynamics, University of Edinburgh
STRUCTURES IN FIRE: AN OVERVIEW OF THE BOUNDARY CONDITION
Jos L. Torero obtained his PhD from the University of California Berkeley in 1992. He is
currently Reader in Fire Dynamics at The University of Edinburgh, UK and a Researcher (en
Disponibilit) at the French National Center for Scientific Research (CNRS). He previously held
the position of Associate Professor at the Department of Fire Protection Engineering and an
Affiliate Associate Professor at the Department of Aerospace Engineering at the University of
Maryland, USA. He is a member of the British Section Committee of the Combustion Institute
and of the Board of the International Association for Fire Safety Science (IAFSS). He is also a
member of the AIAA Micro-gravity and Space Processes Technical Committee, the ASME Fire
and Combustion (K-11) Committee, the UL Foams Fire Suppression Systems committee and
NASAs Mars or Bust. He is on the editorial board of the journals Fire Technology and Fire
Safety Journal.
57
21/04/04
He specializes in fire safety engineering and his work is primarily in the general areas of fire
dynamics, smoke detection & management, fire protection and suppression systems. Recently
he has been involved in the study of fire-induced skin burns, waste incineration and the
behaviour of structures in the event of a fire. He is the author or co-author of 3 book chapters,
more than 50 Journal Publications and more than 100 other technical documents. Dr. Torero
has been the recipient of numerous research and teaching awards that include the E. Robert
Kent Outstanding Teaching Award (1998), the William M. Carey Award for the Best Paper
Presented at the Fire Suppression and Detection Research Application Symposium (2001) and
the Harry C. Bigglestone Award for the Best Paper Published in Fire Technology in 2002.
Dr Susan Lamont
Fire Engineer, Arup Fire
STRUCTURAL FIRE PROTECTION:F ROM PRESCRIPTION TO THE PERFORMANCE BASED
APPROACH
Dr Susan Lamont is currently based in London working as a Fire Engineer for Ove Arup and
Partners. Susan is a member of the structures in fire group and specialises in structural fire
design, but is responsible for many aspects of life safety design when producing a fire strategy
for a building.
Susan is actively involved in encouraging research into structures in fire. Arup Fire sponsor a
number of PhD students in the UK. Susan completed her PhD at the University of Edinburgh in
October 2001. Her PhD title was The Behaviour of multi-storey composite steel framed
structures in response to compartment fires. This was a computing-based PhD analysing the
influence of different fire scenarios on generic composite steel frame multi-storey structures.
Susan worked extensively with the research team at Edinburgh University modelling the
Cardington frame fire tests.
fire resistance design based on quantified structural behaviour.
58
21/04/04
Mr Jim Golt
Group Commander, Fire Engineering, London Fire Brigade
A FIRE SERVICE PERSPECTIVE
Jim has served in the London Fire Brigade for 26 years and has carried out various duties
including operations, training, fire safety and risk management. For the last eight years he has
worked in the field of fire investigation, fire research and fire engineering and currently heads
up a team of six dedicated fire engineers providing consultancy services to the Brigades fire
safety teams. His team are involved in all major building and civil projects across London
together with research activities at major fires. Other activities include committee work for BSI,
CACFOA together with lecturing and a keen personal interest in sailing.
59
21/04/04
APPENDIX D
PARTICIPANTS LIST
Miss K Anderson
University of Edinburgh
Dr A Beard
Heriot-Watt University
Professor N Bicanic
Professor V Brannigan
Mr P Bressington
Mr R Carvel
Heriot-Watt University
Mr B Chisholm
Professor G Cox
Dr C Davie
University of Glasgow
Miss C Dierichs
University of Edinburgh
Mr J Donald
Mr J Dowling
Mr G Flint
University of Edinburgh
Mr J Gilloulley
Mr J Golt
Mr G Goodall
Mr I Goodlet
Mr M Hoare
Mr A Howard
Mr G Hutchison
Mr W Jackson
Mr A Jowsey
University of Edinburgh
Mr D Kee
University of Edinburgh
Dr S Lamont
Mr C Mackay
Partner, Burness
Mr J Martin
Mr J Milligan
21/04/04
Miss K Murphy
University of Edinburgh
Mr G Nicoll
Mr A Orton
Mr A Phillips
University of Edinburgh
Mr J Plumb
Mr W Russell
Dr David Sanderson
Mr Jaime Santos-Reyes
Heriot-Watt University
Miss M Serpilli
University of Edinburgh
Mr B Smart
Mr D Smith
Ms S Smith
Mr C Stokes
Stirling Council
Mr P Stollard
Dr S Shyam- Sunder
Dr J Torero
Dr A Usmani
Dr S Welch
Mr S Young
Speakers, Chairmen and members of the organising committee are denoted by Italics
61