This action might not be possible to undo. Are you sure you want to continue?
Report of a Conference held at The Royal Society of Edinburgh on Wednesday 21 April 2004
Fire and Structures Conference
The Royal Society of Edinburgh
1. Papers 2. Appendices A: The Royal Society of Edinburgh B: Acknowledgments C: Speaker Biographies D: Participant List
This Report reflects opinions expressed by participants in a specific event. It does not, however, necessarily represent the views of the RSE Council, nor of the Society’s Fellowship.
Fire and Structures Conference
The Royal Society of Edinburgh
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
* School of Engineering and Electronics The University of Edinburgh Edinburgh, EH9 3JL United Kingdom
** Arup Fire 13 Fitzroy Street London, W1T 4BQ United Kingdom
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.
2001 events. The dashed & dotted line corresponds to the percentage of the full structural integrity of the building.” the dotted lines to the possible outcome of the different forms of intervention (sprinkler activation. etc. % of Total Structural Integrity. or the intervention of the fire service. such as sprinklers. With the objective of achieving these goals a number of safety strategies are put in place. Passive protection such as thermal insulation of structural elements becomes part of the design with the purpose of increasing tS. The gas phase temperature is assumed to be that of the fire compartment. evacuation protocols and routes are design to minimize te at all stages of the building. within the context of the September 11th.e. % Evacuated.i. with the ultimate goal of 100% represented by a horizontal dashed line. The dashed lines are the percentage of people evacuated. As shown by Figure 1 (dotted lines).e. Untenable Conditions 1st Room Untenable Conditions 1st Floor Untenable Conditions Building The Fire tf te 100% evacuated Detection Subsequent Sprinkler Fire Service time tS First Sprinkler Figure1 Schematic of the sequence of events following the onset of a fire in a multiple story building. Then the energy equation of the structural element can be solved (Drysdale. The Boundary Condition Fire resistance calculations have been conducted in the past and are being conducted currently on the basis of a simulation of the fire by means of Temperature vs. Finally. success of these strategies can result in control or suppression of the fire. Whatever the temperature evolution is used (Petterson et al (1976). Time curves. fire service). It is important to note that within the estimation of te the safe operations of the fire service need to be included. The thick line corresponds to the “fire size. A heat flux is imposed on the structural element on the basis of a boundary condition defined by the gas phase temperature. For this purpose an adequate understanding of the nature of the event and the characteristic of the structure and its safety systems is necessary. ISO-834) the methodology is the same. The events following the attack on the World Trade Center showed that these safety goals were not attained. Different methodologies and tools have been developed to study each of these aspects. and overview of the methodologies used to assess the boundary condition between the structural elements and the fire. It is therefore important to seek the best possible understanding of why this happened. Steel) (Thermally Thick Material – i. Concrete) 4 . but most important. This paper will provide. The energy equation can be of two forms depending on the thermal thickness of the material: dT & ′ = A S q′S dt ∂T ∂ 2T ρS Cp S = kS 2 ∂t ∂x ρS Cp S VS (Thermally Thin Material. These include those strategies that are meant to increase tf which include active systems. 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. 1999).Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 Fire.
5 . 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). It is important to note that absorption is a function of the soot volume fraction and temperature through the soot absorption coefficient ( κ ): κ = Cf S T (3) Where C is a constant. 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 optical depth within the gas phase is much smaller than the characteristic length scales of the compartment. A common assumption states that the emissivity increases exponentially with the thickness of the emitting gas and thus Petterson et al (1976) postulates that ε g = 1 − exp( −κx g ) (1) Where κ is an emission (or absorption) coefficient and x g the thickness of the emitting layer. but currently is consistently deemed as not describing properly the physics of the heat transfer process. Radiative inputs can come from the hot gases. The computation of the emissivity ( ε g ) is also subject to various simplifications that vary with the author. The concept of Temperature vs.e. fS the soot volume fraction and T the temperature. For the thermally thin elements AS will be the exposed area. Thus heat radiation can be treated as a local phenomenon. if no fire is imposed a heat loss to an ambient temperature can be used. If a fire is present at the other side then a similar boundary condition will be included at this end. thermal equilibrium between soot and gas phase in the smoke might also be accurate. Time curves. h ( Tg − TS ) is the convective contribution. soot. they are attenuated by the absorption through the gas phase. other surfaces or the flame. The radiative component needs to account for all sources of radiation. i. T − ε S σTS S (2) Where the net heat input to the structural element is the surface re-radiation and the term & ′ q ′S . The emissivity of the solid surface is given by ε S and that of the gas by ε g . 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. 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. furthermore. Furthermore. 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. there is no radiation exchange between soot particles and the gas.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 Where the boundary condition for both cases corresponds to the input from the fire and is given by: 4 & q ′′ = h( Tg − TS ) + ε g σTg4 − ε S TS S (Thermally Thin Material) & q ′′ = h( Tg − TS ) + ε g σTg4 − εS TS4 = −k S S ∂T ∂x (Thermally Thick Material) x= 0 Where Tg is the imposed temperature of the gas as defined by the Temperature vs. This approach carries the further assumptions that single emitting temperature and gas phase emissivity is sufficient to describe radiative heat exchange. − ε S σTS4 is & q ′r′. thus a more complete way to describe the above boundary condition will be: 4 & & q ′′ = h( Tg − TS ) + q ′r′. 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. • The radiation field is in thermal equilibrium within the gas phase. if far from the flames. and thus gas temperatures can be used to establish radiative heat fluxes. this will be done in the following section. This is a simpler boundary condition that requires the introduction of less parameters. T conglomerates all radiative inputs. For simplicity heat exchanges with the outside environment have been ignored but could be included in these expressions.
5 m. Rockett (1976)). 1972). H being the height of the vertical shafts of the crib. but only the region of maximum temperature and the decay stage are consistent with the assumptions of the thermal model. The fully developed compartment fire is defined as the ultimate (not always maximum) state of burning and either the fuel available or the ventilation determines its characteristics. The growth of a fire is generally described through a two-zone model where the fire through a cold lower zone entrains air and products of combustion migrate to an upper layer. Fully developed fires have been studied for many years. Furthermore. In this case the flow can be modeled via a single zone and the use of the ideal gas law in conjunction with conservation of energy and mass. the area of the vertical shafts of the crib. This study used room height scales of H=0. the exposed surface area of the sticks controls the burning rate and therefore the burning rate increases with (HAo/A). control the oxygen flow through the crib. the presence of thermal insulation can result in very minor temperature changes throughout the entire fire growth period. (HAo/A). undertook one of the most comprehensive studies on the subject (Thomas and Nilsson. Quintiere (2002) presents a comprehensive review of the existing body of work thus only a brief description of the relevant concepts will be presented here. and the cribs nearly covered the entire floor. both the standard temperature time curve (ISO-834) and the parametric curves developed by Petterson et al (1976) insist on establishing a temperature evolution with time. With sufficient oxygen. Fuel generation in turn is the result of energy feedback from the flames. usually rapid.B. The “fully-developed” state is where all of the fuel available is involved to its maximum extent according to oxygen or fuel limitations. A singularity in the growth process is the event of “flashover.I. 2002) many questions of relevance to the scenario of the World Trade Center still remain with no answer. 1973.I. The fuel available is determined by the burning rate and the ventilation is generally defined by a ventilation factor that is associated to the size of the openings of the compartment. Pressure in a compartment fire is considered to be atmospheric and flows occur at vents due to hydrostatic pressure differences (McCaffrey and Rocket (1977). For wooden cribs in a compartment. Their tests and computations result in a series of temperature time curves that are intended to represent fires for different fuel loads and ventilation conditions. Furthermore. The thermal inertia of structural elements is significantly larger than that of the gas phase.B.crib. In the growth period this implies a developing fire that is inconsistent with the single layer treatment that is used to establish the heat input to the structural elements. 6 . The C.) took a different approach in their study of compartment fires.5 to 1. H0 the height of the compartment opening. Wood cribs were used as fuel and although this arrangement has particular burning characteristics the observations illustrate the main factors controlling a fully developed fire. For limited oxygen the ventilation factor controls the burning rate and a constant burning rate is observed for different vertical shaft areas. This translates to defining the characteristic length scale of equation (1) as the characteristic size of the compartment (xg). through spread and growth to its fully developed stage.crib.” Here. hot surfaces and combustion products thus the heat input to fuel surfaces can be described by an expression of the form of equation (1).Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 The Compartment Fire A fire has a significant effect on a structure but the characteristics of the compartment that encloses the flames also have an impact on the nature of the fire. A0 the area of the vertical shaft or the compartment opening and A the surface area of the crib or the room floor. Petterson et al (1976) make a significant effort to describe the different stages of the fire. in which the fire distinctly grows bigger in the compartment. thus characteristic times for temperature changes within the solids are much longer that those required for temperature changes in the gas phase. Although significant research has been done to establish the characteristics of fully developed compartment fires (Quintiere. Following “flashover” the fire becomes fully developed fire. Temperatures within the compartment and duration of the fire are defined by the supply of fuel and oxidizer as well as being affected by heat transfer through the compartment boundaries. A fire undergoes a series of processes from its inception. A / Ao Ho . “flashover” is defined as a transition. This particular interaction between solid and gas phases generally allows using time averages for the gas phase temperatures and to assume that the fire can be considered as fully developed for all thermal calculations related to the structures. The International Counsel for Buildings (C. Thomas and Heselden. This assumption eliminates the need to establish hot and cold areas and thus allows treatment of the fire simply as an homogeneous temperature throughout the compartment. A clarification needs to be made here. and the ventilation factor of the compartment.
The C. 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. (a) (b) 7 . Nevertheless.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 1200 1000 800 T [ o C] 600 400 200 0 0 10 20 AT /AH 1/2 30 [m -1/2 40 50 ] Figure 2 Time mean temperature near the ceiling. These observations seem to further establish that the basic premise of a single compartment temperature might be over simplified. Despite being less information this is consistent with the assumptions of the thermal model. 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 next section will discuss the state of the art in compartment fire modelling. Torero et al (2002) performed this analysis for the WTC 1 & 2 Towers.B. Then a correction could be made to establish the fraction of the energy that remains within the compartment. Furthermore. As can be seen in Figure 2. Drastic temperature variations within the compartment have been suspected for many years but very few experiments exist to demonstrate the significance of these variations. The obvious consequence of this is the need to compute the local temperatures and to solve the radiative transport equation. Figure 3 shows the simulations corresponding to the same fire embedded in compartments with three different aspect ratios. 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.I. The results are expressed in terms of the ventilation-factor and surface area and are hoped to be scale independent. 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. 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. the energy release rate can be calculated and thus the temperature of the compartment. This can only be done using appropriate compartment fire models or through experimental characterization of the radiative fluxes to the different surfaces. this study only provides a single average temperature for each condition instead of a temporal evolution of the temperature. The actual data has some scatter which Law and O’Brien (1983) suggest to be a result of some particularily extreme experimental conditions. analysis of the soot volume fractions show also well defined distributions. It can be observed that temperature variations greater than 600oC exist throughout the compartment. If the burning rate can be established then. Numerical modelling can serve to describe the significance of these variations. Where AT is the total area excluding floor an opening. knowing the heat of combustion.
This paper will address compartment fire models. 2001) then the results are unknown. It is of critical importance to note that extensive experimental data has been gathered on the evolution of the temperatures within a compartment but very little information exists on the evolution of the heat fluxes imposed on a compartment surface. If the objective is to integrate CFM’s with structural analysis. required assumptions and their effect on predictive results. The objective of this paper is to demonstrate the depth required in the understanding of these models to guarantee proper use. As mentioned before. this might be appropriate but if CFM’s are going to be used. Zone models require simple computations therefore they were an appropriate solution given the computational constraints of the time. this simplification is unnecessary. this paper attempts to provide a review of the applicability of current models. For all cases the heat release rate per unit area is 1000 kW/m2 propane fire distributed throughout the surface.5m height. The objective of a CFM will be to provide a much more detailed evolution of the conditions within the compartment where the fire originated and adjacent areas. the current practise is to ignore the fire growth period. Only in the late 1980’s advancement in computer technology made Computational Fluid Dynamics (CFD) or Field Models a viable alternative for fire related calculations. As a first order of magnitude approach. By dissecting the particular application of CFM’s to structural analysis the advantages of this approach will be introduced and a number of limitations of the different methodologies will be highlighted. 8 . quantitative predictions are now mostly obtained from numerical computations. Of the three terms involved in the net heat flux the convective heat flux ( h ( Tg − TS ) ) and the global S quantity & q ′r′.5m in all directions. The red is approximately 1000oC and the green 400oC. nevertheless none of the adjacent compartments will be expected to have reached fully developed conditions.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 (c) Figure 3 Example of three FDS calculations of a compartment fire. This paper will address these two terms. Compartment Fire Models The role of Compartment Fire Models (CFM’s) within this application is to allow proper prediction of the net heat & flux q ′′ . For all three cases the compartment cross section is 4m x 4m x 4m and the lengths is (a) 4m. The ventilation opening is 4m width by 2. with consideration of gaps in pertinent predictive capabilities. Under the principle that fire resistance is given directly by the temperature of the structural elements this approximation might not matter and result simply in conservative requirements for insulation. All surfaces concrete. Experimental validation should follow because little or no useful data exists. a vast array of these computer-models exists and they compete well with traditional zone models.3m to 0.” It will emphasize the evaluation of the models as it pertains the proper quantification of equation (2). growth beyond the compartment of origin will generally be within the same time scales as the heating of structural elements. Within a fire scenario it is possible that flashover might be attained within the compartment of origin before any structural element has undergone significant heating. significant effort is necessary to establish realistic timescales and characteristic conditions of fire growth beyond the compartment of origin. Computer tools available for fire modelling in the 1970’s favoured the development of zone-models. and for models it will be understood “computer models. A number of variants emerged and their use became generalized towards the end of this decade. the grid size is approximately 0. Initially the term Model referred to either analytical or empirical formulations that allowed simple calculations associated to the growth of a fire within a compartment. The temperature legend is not presented because the emphasis is on the spatial distribution of the temperatures not on the quantitative values. Since the early 1970’s a number of Compartment Fire Models (CFM’s) have flourished. Furthermore. Computer-based models rapidly followed and were developed on the framework established by these analytical expressions and experimental data. T are those that should be expected as outputs of Compartment Fire Models. If the behaviour of the structure is to be studied dynamically and in an integral manner (Usmani et al. Furthermore. input data requirements.. (b) 8m and (c) 16m. Analytical and experimental formulations are still used to gain insight on the behaviour of fires within compartments but due to the multiple variables and complexities of the problem. Currently.
A process of mass and energy transfer between them links different compartments. All properties are evaluated at the film temperature: T f = (T s+Tg)/2. Summaries of features have been published in much greater number. Two-zone models are by definition limited when analysing heat transport from the gas phase to the solid phase. The rapid increase on the usage of CFD codes for fire has prompted more detailed reviews of which the more comprehensive is that of Novozhilov (2001). As mentioned above two different fundamental methodologies can be used for prediction of compartment fire behaviour. which is the most computationally intense aspect of these calculations. Friedman (1992) has published the most comprehensive survey of this type. Both surveys provide a complete list of all existing models. its developers and application. They avoid the solution of the fluid mechanics equations thus allows for faster computations and more complex scenarios. Zone Models (ZM) and Computational Fluid Dynamics (CFD) or Field Models. These models have been traditionally divided in two groups. α = 1.0209 + 2. L. This division is still relevant and will be used here. temperature and species fields within these control volumes are not resolved. The power n is typically 1/4 and 1/3 for laminar and turbulent flow.e in CFAST (Cooper.0 × 10 −9 Tf7 4 0. All heat transfer related quantities within these codes are established in an empirical manner.000267 Tf • • In the following table the different correlations employed in CFAST are presented. the last of them being the reviews by Walton (2002) and Cox and Kumar (2002). The solution of the flow.33 × 10 −5 Tf k= 1 − 0. numerical techniques and applications. Nevertheless relies on empirical correlations at all levels of heat and mass transfer.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 General Remarks on Computer Based CFM’s Numerous reviews on computer based CFM’s have been made in the past and its not the objective of this paper to provide a new one. No critical review of the models is provided. In most cases limitations are presented only within a general context. These summaries of features expose the basic principles of the CFM’s.. Just recently Olenick and Carpenter (2003) have developed a new survey that is currently in press. surveys and summaries of features. Reviews available in the literature are of two types. 1988). respectively. no general comments on the limitations of these codes will be provided at this stage. Surveys collect data on all existing models and provide a list of them with some brief description of the code. The thermal diffusivity and thermal conductivity of air are also defined as a function of the film temperature. ceilings and floors (hot surface up or cold surface down) and ceilings and floors (cold surface up or hot surface down)(Jones et al. The first are Zone Models and the second is CFD. • Calculation of the convective coefficient (assumed to be natural convection) is via correlations for walls. 9 . their sources and applications. Flow. 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. 1991): (4) Nu k h= L Where the Rayleigh number is defined as: L = C ⋅ Ra n L Ra L = Gr L ⋅ Pr = gβ(Ts − Tg )L3 να This number is based on a characteristic length. therefore. 2000 and Cooper. It is assumed that within the two smaller control volume all properties and conditions are homogeneous. thus the convective heat transfer coefficients and radiation heat transfer used for a small fire will be the same as for large fires. of the geometry. Zone Models Zone models treat compartments as a control volume sub-divided into two smaller control volumes. 1991). One control volume considers the smoke and the other the fresh air. The reader is referred to Walton (2002) for detailed information. The convective transfer coefficient is generally defined in terms of the Nusselt number (i. These empirical correlations have in general no link with the burning conditions. is thus avoided by this simple two-zone approach. from data in (Atreya.
that solve the net radiation equation are present in Zone Models. 1991). This information is currently very limited for fully developed fires. The validity of these models depend on the applicability of two zone models and of the empirical correlations. Because of the constant properties in each zone. The thickness of the boundary layer is determined by the temperature difference between the gas zone and the wall or object being heated (Jones et al. For the modelling of an environment such as a 10 . This method shows adequate results when appropriate absorbance coefficients are applied (0. nevertheless are fully empirical. thus the user must set pyrolysis rates. Nevertheless are simple to use and robust in nature. Among the empirical correlations those corresponding to entrainment rates are critical and their validation under conditions other than free axis-symmetric or line fires is limited (Joulain. have to be calculated in a different way. Zone Models generally do not include pyrolysis models. For fully developed fires they are defined by ventilation and a very restricted set of data is available. The zones and surfaces are assumed to radiate and absorb like a grey body. all the convective heat transfer is calculated based on the temperature difference between the zone and the object. complicated geometries would be treated in the same way as less complicated ones.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 Table 1 Different heat transfer correlations employed in CFAST In two-zone models the Reynolds number cannot be calculated properly since there are no velocity fields within the two zones. The use of none of these models for very complicated geometries has not been validated and thus is questionable. Zone model assumptions have been found to break down in flashover fire scenarios (Novozhilov. Approximate pyrolysis rates for pre-flashover fires are defined by empirical heat release rates and abundant data is available in the literature. From the principles of this model it is impossible to improve this approach. The heat exchange between layers is also possible. Extensive validation is available in the literature and clear estimates of error can be generated. These coefficients represented reasonable approximations for fires with sooty upper layers and clean lower layers. the convective heat transfer. respectively). CFD Codes The main aspect that differentiates CFD codes is the way by which turbulence is modelled. In summary. So. gases and fires. derived by Siegel and Howell (1981). The limits of the two-zone approximation have not been studied extensively. and thus heat transfer and entrainment are defined by the pyrolysis rates. nevertheless validation that it is adequate to use these correlations is necessary under conditions that are relevant to the modelling of structural response to fires. Large Eddy Simulation (LES) models and Direct Numerical Simulation (DNS) models. 2000).01 for the upper and lower layers. Zone Models are inherently limited by their basic assumptions. 2001) leading to predicted heat release rates that are lower than the actual ones. Thermal radiation tends to be treated in a complex manner. Thus.. This is very important because the flame characteristics. For this reason. Nevertheless. Appropriate validation data under these conditions currently does not exist. 1998). Reynolds Averaged Navier-Stokes (RANS) models. Gas layer absorption can be calculated. For fully developed conditions these coefficients have never been validated (Forney. and the boundary layer. Methods such as the four-wall algorithm. The absence of velocity fields and lack of turbulence modelling implies that the convective heat transfer will not be affected by complicated geometries. its intrinsic limitations are clearly of great importance when addressing the application to modelling of structures exposed to fires. Equation (1) will also be defined by the soot absorption coefficient and thus by the soot yield of the fire under each specific condition. The objects that participate in the radiation exchange are walls. CFD codes can be divided into three groups.5 and 0.
LES is computationally more expensive than RANS. The number of DNS grid points required for the resolution of all scales increases approximately with the cube of the Reynolds number (Re 3). recent advances in computer performance and numerical methods have made LES feasible for such fire and smoke flow problems. 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. must be simulated with accuracy. proficient use of these models can provide adequate results. Smagorinsky constant) but its calibration is easier and is independent of the Reynolds number. self-consistency requires that LES results approach the RANS results (Orszag et al.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 compartment fire and given the computational resources currently available. 1993). assumes that turbulent motion can be separated into large-eddies and small-eddies. • LES does not average over time so it allows modelling the time evolution of the sub-grid scales. • Diffusion flames representative of fires are generally considered thick. Current super computers have the capability to provide a grid resolution not greater than 108 cells. 2001). thus all dynamic information for scales smaller than the large turbulence scales is lost. better time resolution requires an increase in the cell size thus large grid cells characterize LES solutions. 11 . RANS solves ensemble-averaged Navier-Stokes equations by using turbulence modelling. 1988). Dynamic sub-grid models are beginning to appear in the fire literature but are still ongoing research and have never been validated. the total number of cells necessary for solving fire and smoke movement in a room is approx 1013. Since LES solves time-dependent flow. Since full resolution of the Navier-Stokes equations is not practically possible. the small length scales are smaller than the grid size and in RANS small length scales are smaller than the largest eddies. DNS. This can be translated in a better resolution of the time evolution of the fire. it is necessary to model some aspects of the flow. Because of this. In LES. If the grid size of an LES simulation is taken larger and larger.e. Thus. cannot be used to simulate complicated fire spread and smoke movement in a full compartment. down to the dissipation scale. 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. LES modelling implies a proper definition of the grid size that is consistent with the model parameters and with the computation constraints. 1988). The key step in both LES and RANS is the derivation of the underlying dynamical equations averaged over small scales. DNS simulations are not feasible for a number of reasons. Nevertheless. Furthermore. The only difference between LES and RANS is the definition of small scales. all dynamical degrees of freedom smaller than the size of the largest (energy-containing) eddies are averaged. • LES can also rely on an empirical model coefficient (i. Since the Reynolds number for typical fire and smoke movement in a compartment is approximately 105. The most widely used turbulent viscosity model is the standard k-e model. The large eddies (grid scale) motion is directly simulated and the small eddies (sub-grid scale) motion is approximated. The choices of which aspects of the flow will be modelled and thus the approach to be followed. 1993). Therefore. Wall functions have been established to address these areas but their accuracy and generality is still questionable (Bilger. 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. current computing technology is still far too small to solve such fire motions. is difficult and implies inevitable subjectivity. Calibration of the model coefficients has been done for a multiplicity of scenarios but these rarely include conditions typical of fires (Orzag et al. therefore the temperatures of each cell represent an average of reactive and non-reactive regions. LES techniques always need to be 3D and must have a time step short enough to capture most of the important turbulent motion. developed in the early 1960s~1970s by Smagorinsky (1963) and Deardorff (1970). therefore. All eddies. In fact.. such as 3D instantaneous velocity. the loss of dynamic information can significantly affect the predictions of fire growth therefore needs to be handled with great proficiency. Despite this statement. 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 the validity of the direct application of RANS and LES models could be questioned (Bilger. DNS requires the grid resolution to be as fine as the Kolmogorov micro scale. RANS can be further divided into turbulent viscosity models (such as the k-e model) and Reynolds-stress models. so there is no dynamic information about the smaller scales. In a RANS solution. LES. this model coefficient can be avoided entirely with a dynamic sub-grid model (Orzag et al. The grid cells are much larger than the flame thickness. • RANS relies on numerous empirical model coefficients (between 7 to 12 different coefficients) that will describe turbulent viscosity and fluid wall interactions. the capability of these codes to properly model flame temperatures and thus radiative heat transfer is questionable. it can provide detailed information on turbulence. 2003). To achieve computations within reasonable time constraints..
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. These validation exercises clearly are not sufficient to determine the adequacy of the complex models proposed. 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.B. For radiation to be properly modelled the most important aspects are temperature and soot concentrations and morphology. 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. 1988. On Combustion. “CFD Prediction of Coupled Radiation Heat Transfer and Soot Production in Turbulent Flames”. Symp.218. 2003). On Combustion. 1996. G.e. and soot models is currently underway (NIST. N. 1972) which is more suited for the validation of Zone Models than of CFD codes. Independent of the model used all numerical tools are severely limited by an improper definition of the fundamental properties of materials controlling fire growth.W. and Kumar.194-3.. Proceedings of 22nd Int. Time scales more relevant to structural behaviour imply in most cases fully developed fires. Common validations rely on simple comparison with temperature measurements (Grandison et al.A. The Combustion Institute. R. fully developed fires. A series of general comments on the validity and limits of the different current approaches has been provided. Cooper. “Modeling Enclosure Fires Using CFD. To achieve proper temperature predictions it is also necessary to adequately establish radiative heat transfer. Numerical models can also be useful for this purpose.). 12 . University of Utah. Thomas. “The Structure of Turbulent Nonpremixed Flames”. These general comments give a guideline of areas that need further attention. None of the existing CFD codes has been properly validated under these conditions. P. 1988. “Convection Heat Transfer. existing combustion models have been validated only in simple scenarios and with very limited diagnostics.” Chapters 1-4 in SFPE Handbook of Fire Protection Engineering. NIST. S. 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. Currently. 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. NISTIR 4705..Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 • • Proper combustion models are necessary to generate correct heat release rates (thus temperatures) and species. 2003) that in many cases are decades old (Steckler. 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. 1991.. Bilger. radiation models. Cox. The data available for postflashover. A. et al. J. Symp. Cranfield. 1985. The Combustion Institute. SFPE Handbook for Fire Protection Engineering. is generally in the form of average punctual measurements of temperature (Thomas et al. L. BRE. A review of the most commonly used modelling approaches then reveals that currently these models also have significant limitations. These general limitations to these codes are by no means insurmountable but improvement and confidence can only be achieved with systematic and precise validation. Tewarson. 2002. Quintiere. Significant work on the development of combustion models. “Fire-Plume-Generated Ceiling Jet Characteristics and Convective Heat Transfer to Ceiling and Wall Surfaces in a Two-Layer Zone-Type Fire Environment”. 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. 1972. 3rd Edition.Y. References Atreya. 1982). In their current state. Drysdale. Sandia National Laboratories. These models are being incorporated into numerical tools on a constant basis. Many of these limitations can be circumvented by proficient and experienced use but lack of detailed validation still remains a serious problem. 1999. Proceedings of 26th Int. 3. Moss. 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. An analysis of the input variables for all flammable materials shows a systematic dependence of simple and very approximate databases (i..W. National Fire Protection Association. Rubini.” Chapter 8. etc. Bressloff.
R... Jones. National Bureau of Standards. Joulain. W.. P. 29.R. Prasad.. Symp.. 2nd Edition. Proceedings of the Combustion Institute. “Evaluation of a Fast. “Development of Standards for Fire Models: Report on Phase 1 Simulations. FRD Publication Number 1/2003. J.... 2002. J. “Computational Fluid Dynamics Modelling of Compartment Fires”. P. K. Palmer. 12: 165. Combustion Theory and Modelling.. Smagorinsky. "An Updated International Survey of Computer Models for Fire and Smoke. Galea. G. 2000. “Computing Radiative Heat Transfer Occurring in a Zone Fire Model”..R. Fire and steel construction: Fire safety of bare external steel... Forney.P.J. “General Circulation Experiments with Primitive Equations – I: The Basic Experiment. I. Tools for Estimating Fire Growth and Smoke Transport”. 3.B. “An International Survey of Computer Models for Fire and Smoke.B. A. Hamins.A.” Office of the Deputy Prime Minister.J.. O. Grandison. Tewarson. Petterson. M. School of Mechanical Engineering. 2003.” Hemisphere Publishing Corporation. “Generation of Heat and Chemical Compounds in Fires.. Lewis. 2002. Reneke.D.. 81-92. J. G. 41.. Rubini. Forney. K. NIST. U. S. NIST.K. J..cranfield. D. M.. “CFD Modelling of Combustion and Heat Transfer in Compartment Fires”.A. Ndubizu C. John Wiley and Sons. Novozhilov. Hostikka. of Canterbury. Peacock.E. Swedish Institute of Steel Construction. Journal of Research.P.A. J..” Journal of Fluid Mechanics. 3rd Edition.uk/sme/sofie. “Some Basic Challenges for Large Eddy Simulation Research. R. Drysdale. Simplified Computational Fluid Dynamics Model for Solving Room Airflow Problems”. Web Site: www. “A Three-Dimensional Numerical Study of Turbulent Channel Flow at Large Reynolds Numbers. R... 2002.2. NIST. P. W.. Hostikka.P. P. Moss. Prasad K.. S. Seigel. 1976. Fire Behavior in Building Compartments. “A Technical Reference for CFAST: Eng. 2002.J. 5.” Chapter 3. 1993.P. J.28.. Fire and Materials. March 1997. Cranfield University. 2001. McGrattan. Steckler." SFPE Journal of Fire Protection Engineering. 1970. Jones. W.E. K. G. p. 2003... G. NIST. R. The Steel Construction Institute. 2002. McGrattan.. H. NIST.E. Forney. “Development of Standards for Fire Models: Report on SMARTFIRE Phase 2 Simulations. K. M. Proceedings of the Combustion Institute. Project Report for Partial Fulfillment of M. E. Ananth R. M. Galea. and Rinkinen. Forney. 1986...W..J.. 1976. R.. A.82-3. and Howell. R. J. McCaffrey. Publication 50. v.E.M. Simulation of Fires in Enclosures. G. Floyd.. School of Engineering. Patel. National Bureau of Standards. Tools for Estimating Fire Growth and Smoke Transport”. B.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 Deardorff. and Carpenter.52-60.. Petterson. 1982.” NBSIR-82-25-20. “A User’s Guide for FAST: Eng.” Office of the Deputy Prime Minister. J. McGrattan. Rehm.161. K. Bukowski. P. “Fire Dynamics Simulator (Version 3) – User’s Guide”.W. 4 (3). Musser. Tatem P. Douglas J. 2000. 1999. Baum. “Numerical modelling of methanol liquid pool fires”..A.. A.R. Law M. Li C. 2002. degree at the University of Canterbury.J.B. R. FRD Publication Number 2/2003. B. 2001. Proceedings of 5th Int. Orszag.. “Flow induced by fire in a compartment. 82: 107 1977. 1963.G.ac. N.G. J. J. G. Olenick.pp. Magnuson. “Fire Dynamics Simulator (Version 3) – Technical Reference Guide”... Grandison. “Assessing the Feasibility of Reducing the Grid Resolution in FDS Field Modelling”. 13 .. V. “Thermal Radiation Heat Transfer. Orszag. S. NISTIR 4709.D... E. and Rockett. SFPE Handbook for Fire Protection Engineering. Patel. J. SOFIE Technical Summary. and Thor. Progress in Energy and Combustion Science. vol. IAFSS. et al. 1999. NIST. On Fire Safety Science.” Large Simulation of Complex Engineering and Geophysical Flows”. Reneke. 1991. S. Stephen M. A.. and Yakhot V. Quintiere.G. Prasad. A.. SOFIE.D. Peacock.A. Fire Engineering Design of Structures.K.. Quintiere.W. 1998... and O’Brien T. Friedman. An Introduction to Fire Dynamics. K. Quintiere.G. 1981. J.” SFPE Journal of Fire Protection Engineering. Forney. W. 1992.A. Staroselsky.. 1982. Rockett.D. Elsevier Science Ltd. Kailasanath K.” Monthly Weather Review. Combustion Science and Technology.. Floyd.
923.. Inst.H. P.. Fire Safety Journal. “Fire Safety in High-rise Buildings: Lessons Learned from the WTC. Sanad. Chapter 7. Proceedings of 5th Int. R. 3rd Edition.. 14th Symp. 14 . 1996 (26).J. P. Thomas. A.A.. IAFSS. UK. Quintiere.M.) on Combustion. W. Rubini.. Welch. SFPE Handbook for Fire Protection Engineering. Usmani.. “Fully-Developed Fires in Single Compartment – A Co-operative Research Programme of the Conseil International du Batiment (CIB Report No 20). Woodburn. J. Germany. Elsevier Science Ltd. Lamont. L. and Steinhaus.H. Rotter. “Three-dimensional Simulation of a Fire-Resistance Furnace”. Comb. Zone Computer Fire Models for Enclosures. S. 1007. Dresden.D.. P. UK. A. Fire Research Sta. Torero. V. Walton. “Fundamental Principles of Structural Behaviour Under Thermal Effects” Fire Safety Journal... “CFD Simulations of a Tunnel Fire – Parts I and II”.194. On Fire Safety Science. (Int.. A. March 1997. 721-744. and Gillie. T. P.E. 2002.M. Symp. M. L. P. J. August 1973. FR Note No 979.. J. FR Note No. 3. 2001. 1972.189-3. Britter.H. 36.. G. Fire Res. Aug. 1972. M...” Jahresfachtagung der Vereingung zur Forderrung des Deutschen Brandschutzez e..J.S. 2002.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 Thomas. Statation. Thomas. and Nilsson. “Fully Developed Compartment Fires: New correlations of Burning Rates”. and Heselden. S.
1 FIRE RESISTANCE Fire resistance design of structures is traditionally based on the results of the standard fire resistance test. Fire Engineer. 15 .Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 Dr Susan Lamont. thermal expansion. design fires. performance based design. 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. Historically. Finally. 2001 (FEMA 2002). Arup Fire STRUCTURAL FIRE PROTECTION-FROM PRESCRIPTION TO THE PERFORMANCE BASED APPROACH* *Co-Authors: Dr Asif Usmani. DR Jose Torero Keywords: Fire engineering. The paper will introduce the reader to the key concepts of structural fire engineering and the limitations of a prescriptive approach. The fire resistance levels recommended in regulatory documents are based on the results of this test. beams and floor slabs failure is defined at a deflection of L/20. Specifically. 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. or when the deflection exceeds L/30 failure can be defined as a rate of deflection of L 2 /9000d. Cajot and PierrePROFIL ARBED BRE and CORUS 1994-1999 1994-2003 September 11th 2001 Cardington Frame Fire Tests WTC collapse Table 1 History of structures in fire research and design 2. e. research and modelling Development of the Time-equivalence concept Structural steel design for fire based on real fire data Pompidou Centre. prescriptive design.g. fire resistance design of structures has been based upon single element behaviour in the standard fire resistance test. Dr Barbara Lane. For stability of load bearing horizontal elements of structure. structural response. 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. L=clear span of the specimen under test and d = the distance from the top of the structural section to the bottom of the design tension zone. 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. for load bearing elements and/or separating elements of construction BS 476 Part 20 defines three criteria for insulation. will also be discussed. 2. the importance of a robust engineering approach to fire resistance design of tall buildings post 911. Law and O’Brian (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. validation and acceptance of computer modelling in the design office to predict the whole frame structural response of buildings to fire.Fire Safety of bare external structural steel BS 5950 Part 8 (Load ratio) Broadgate fire Eurocodes including new time equivalent concept and design fire curves Natural fire safety concept Researcher Ingberg Kawagoe Magnusson & Thelandersson Law Pettersson Pettersson Law and O’Brian Date 1900 1928 1963 1970 1971 1976 1976 1986 1990 1990 1991 onwards Schleich. Paris. It will also highlight the most significant advances in the last 10-15 years. 1. Compartment fire models as well as the time equivalence concept have tried to address this shortfall for decades. HISTORY A selection of the most significant advances in structural fire engineering are listed in Table 1. The test determines the ability of a building element to continue to perform its function for a period of time without exceeding defined limits. the WTC collapse. integrity and stability that must be passed in order to achieve a fire resistance rating. Historical Event Standard furnace testing developed Fire load concept Compartment fire testing.
5m beam this equates to 375mm and for an 18m beam this is 900mm. between commencement of heating and failure. Figure 1: Standard Temperature-time curve The standard T-t curve bears little resemblance to a real fire T-t history. For a 7. the compartment size. This approach was inappropriate because it took no account of the ventilation to the room. 16 . Most regulatory bodies accepted Ingberg’s fire severity approach and fire resistance testing to the standard T-t curve continued. which would be “equivalent” to the exposure time in the standard test. Periods of fire resistance are normally specified as ½ hour. Ingberg (1928) was the first to propose a solution to this problem when he suggested that fire severity could be related to the fire load of a room and expressed as an area under the T-t curve. (Drysdale 1999) 2. 2 hours. shape or the properties of the boundary wall linings (Drysdale 1999) all of which dictate the severity of a compartment fire.2 EQUIVALENT FIRE EXPOSURE Since Ingberg’s early attempt at relating the severity of the standard fire to a real compartment fire.g. This approach was based on limited information from room fire tests. The time equivalent concept makes use of the fire load and ventilation data in a real compartment fire to produce a value. although it is designed to typify temperatures experienced during the post-flashover phase of most fires.5 hours. Formulating equivalent fire exposures has traditionally been achieved by gathering data from room-burn experiments where protected steel temperatures were recorded and variables relating to the fire severity were systematically changed (e. 1 hour. Therefore in a code compliant building in the UK with all structural elements protected the floor is permitted to deflect up to L/20. The fire load is in kg/m2 and the ventilation is a fraction of one wall. The standard curve has no growth or decay phase and as such does not represent a real fire. 3 hours and 4 hours. 1. The requirements for fire resistance were related to the assumed levels of fire loads in different occupancies. The fire resistance of the element is taken as the time to the nearest minute.Fire and Structures Conference The Royal Society of Edinburgh 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. The severity of two fires were equal if the area under the T-t curves are equal (above a base line of 300°C). many researchers have developed similar but more sophisticated relationships. Figure 2 compares the T-t histories of fires with varying fire load and ventilation with the standard curve. Figure 2 Comparison of the standard fire curve and real fire temperature -time histories.
2 Eurocode 1 The most recent approach to time-equivalence is Eurocode 1. m = combustion factor (default=0. fire load. δ q 2 = factor to account for probability of occurrence of a fire based on fires reported to the fire service. i =1 10 k b = factor applied to account for the insulation properties of the compartment enclosure. Pettersson et al (1976).d = equivalent fire resistance (minutes). The traditional office would require 120 minutes fire resistance to elements of structure and the time-equivalent calculation confirmed this value. The traditional building with limited glazing is typical of what was built when fire resistance ratings for the Building Regulations were agreed after the Post-War building studies. Traditional office Figure 3 Greater London Authority (GLA) building Glazed areas in traditional and modern office buildings 2. UK (ODPM 2002) to 60 minutes. (1) δ q1 = factor of consequence of structural failure based upon occupancy and compartment floor area. adopted Law’s 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).Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 ventilation. te. 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. qd = fire load density (MJ/m2 ). d = q f .k c w f where.2. time equivalence allowed an overall reduction in fire resistance rating from 120 minutes as prescribed by the Approved Document B.1 Traditional versus modern office buildings Arup Fire carried out a comparative study of the equivalent time concept in a traditional office building with relatively low area of windows in the façade and a modern office building with a fully glazed facade.δ q 2 . 17 . 2.δ n . As a result of the large amount of available ventilation in the façade of the Greater London Authority Building (see Figure 3).2. d . Building Regulations.m.8). 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. Both office buildings are shown in Figure 3. δ n = ∏δ ni = factor taking into account the different active fire fighting measures (see Table 2). Law (1971) developed a time equivalence relationship to include the effect of ventilation using data gathered from a CIB (Conseil Internationale du Bâtiment) study of fully developed compartment fires (Thomas and Heselden 1972). The equivalent time of fire exposure is defined in the Eurocode as: t e . Pettersson’s time-equivalence approach takes into consideration the effect of window height and the thermal inertia of the compartment wall lining.k b .δ q1.
Fire resistance ratings range from 15 to 90 minutes in this particular case. Pettersson and other available ventilation factors has been undertaken for real fire tests involving fully developed fires (Law 1997).25 by the Eurocode. kc =1. the area of the compartment considered and ventilation considered to be available. wf = (6 / H )0. An essential part of a time equivalent study is a sensitivity analysis.3 [0. Pettersson’s ventilation factor is defined as follows: w f = 0. the validity of reducing/increasing the fire load using the other factors is questionable. 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). The assumed fire load density is multiplied by these factors to achieve a design fire load for the particular building.0 for concrete.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 k c =correction factor function of the material composing structural cross sections e. However. By conducting a sensitivity analysis a range of equivalent times can be calculated and the upper bound value taken as a credible worst case. and is limited to a maximum of 0.957 Af / At Av h 0. δ n1 is well known and has been in earlier versions of Eurocode 1. This should include varying the amount of ventilation available. but this has been ignored in the Eurocode. 18 . The most critical variable is typically ventilation. 5 Where ( ) 0. wf = ventilation factor The ventilation factor in the Eurocode is defined in Equation 2. which results in a lower ventilation factor. 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. 5 (3) Af = floor area of the compartment (m2) At = total internal surface area of the compartment.62 + 90(0.25. The factor to take account of sprinklers. Modern triple or double glazing systems do not break as readily as single glazing in a flashover fire. floor area involved and thermal properties of the boundary wall linings. H is the compartment height (m). 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 duration and peak temperature of a fire is significantly affected by available ventilation.4 − α v )4 ] (2) A detailed comparison of the Eurocode. Where the term αv is the area of the vertical openings in the compartment walls divided by the floor area.g. In a building with exterior glazing this ratio is typically higher than 0. This study demonstrated that Pettersson's formula was the best representation of the ventilation effect in real fire tests.
61 Automatic Work fire alarm brigade transmissio n to fire brigade δ n2 δ n3 δ n4 δ n5 0.88 1.34 0.77 0.5 1.77 0.61 0.2 42. This was acknowledged by the members.88 1. The final outcome was a single database covering 40.0 ¦ 0.5 1.5 14. The main outcome of WG 1 was the one zone compartment fire model Ozone (Cadorin and Franssen 2003).88 1. This information was quantified in terms of factors on the fire load and is the basis of much of the information in Eurocode 1 Part 1.7 0. *Factor to take account of sprinklers only **Factor to take account of sprinklers+detection+Fire Brigade 2.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 δ ni Function of Active Fire Fighting Measures Automatic fire suppression Automatic water extinguishing system Automatic fire detection Manual fire suppression Off site fire brigade Safe access Fire routes fighting devices Smoke exhaust system Independen Automatic fire detection and t water alarm supplies 0¦1¦2 by heat by smoke δ n1 0. Finland and the UK over the period 1986-1995.77 1.53 0.34 0. Probabilities deduced from the database formed the basis for a risk analysis of a fire start considering the influence of active fire fighting measures (fire fighters and sprinkler systems) and occupancy type. Working group 3 was responsible for supplying data and guidance on the input for natural fire design for those with no expert knowledge of compartment fire dynamics.61 or 0.61 0.0 or 1.7 29 [m ] 312 0 312 0 312 0 312 0 312 0 312 0 0.2.7 96.61 0.87 δ n6 δ n7 δ n8 δ n9 δ n10 1.77 1. 19 .87 or 0. Working group 2 reviewed t-equivalence relationships from the literature and design codes. There were 5 working groups (WG) set up by the project each contributing to a different aspect. Netherlands and Australia.000 fires.4 29.53 0.34 Fire Load [MJ/m ] 777 777 777 777 474 474 474 474 270 270 270 270 2 Net Floor Area (Af) Ventilation (Av) wf [m ] 2600 2600 2106 2106 2600 2600 2106 2106 2600 2600 2106 2106 2 Time Equivalent [min] 47. France.3 Natural fire safety concept The research project ”Competitive steel buildings through natural fire safety concept'' was undertaken by 11 European partners co-ordinated by PROFIL-ARBED-Research (Shleich et al 1999).87 ¦ 0. A database of natural fire tests was also collated from experiments conducted in France.1 58.9 16.61* 0.0 or 1.53 2 Table 3 Fire resistance ratings calculated for an office building in London. UK.6 33. from June 1994 until December 1998.7 25.77 1. Working group 1 analysed the parametric temperature-time curve in Eurocode 1 and reviewed existing CFD and zone models. These methods were analysed at an early stage in the project but are not relevant to the Natural Fire Safety Concept because they relate to the standard T-t curve.78 0. Working groups 4 and 5 gathered statistics from real fires in Switzerland.7 50.1 83.5 Table 2 Factors proposed by Eurocode 1 to calculate an appropriate design fire load Scenario 1 2 3 4 5 6 7 8 9 10 11 12 ?n 1 1 1 1 0.34** 0.9 or 1 or 1.73 0.77 0.
The calculations consider the different fire exposures experienced by external structural members as compared to the same members in a fire compartment.4. then the upper limit of the steel temperature may be greater than 550°C. It also allows fire resistance by calculation using the Limiting Temperature Method. If this value is low i. The structural elements are considered in isolation of the rest of the building frame. It details fire resistance derived by tests. In general these are conservative.2 Load Ratio BS 5950 Part 8 is the British Standard code of practice for fire resistance design. 2.0 respectively. The code details limiting temperatures for various load ratios. A number of performance based design approaches rely on single element calculations based on critical temperatures.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 2. These are based on the external structural steel calculations originally developed by Law and O’Brien (1986). then the failure temperature may be higher than 550°C. This is the basis for performance based design in accordance with BS 5950 Part 8. Figure 4 The concept of thermal actions on external members If the structural element is lightly loaded i. Load ratio = The factored load at the fire limit state 20 . the load capacity at 20°C is high compared to the applied load at the fire limit state.e.4 Critical temperatures of single elements of structure in fire The critical temperature of most unprotected steel sections falls within the range 400-600°C.8 and 1.2. No heat build-up since the member is outside. The concept is shown in Figure 4. The engineer calculates the load ratio of the single structural element (see Equation 3).4.1 External structural steel The thermal actions for external members are outlined in Annex B of Eurocode 1 Part 1. (3) The load capacity at 20°C The factors applied to the live and dead loads at the fire limit state have traditionally been 0.e. Heating based on flame size & position of member with respect to the façade Radiative heat flux from the fire compartment Radiative and convective heat flux from the external flames through the windows Radiative and convective heat loss from the steelwork to the ambient surroundings Size and location of the structural steelwork Through draught conditions. The limiting temperature method allows the designer to compare the temperature at which the member will fail with the member temperature at the required fire resistance time. Cooling from surrounding air. under-stressed at the fire limit state. At these temperatures the section has lost 40-50% of its original ambient strength and if it is tested as a single element in a standard furnace will fail as a result of the loss in strength. The calculations account for: • • • • • • • • • The fire being within an adjacent compartment. 2. The design guide assumes a critical temperature of 550°C for steel although any critical temperature could be considered.
UK (see Figure 6). Pettersson looked at different loading and beam configurations against temperature for different heating rates. He also considered the influence of boundary conditions including the effects of restraint. or floors. it was not well understood and large scale fire tests had not been carried out.5 Broadgate fire and the Cardington tests Engineers have always recognised that whole frame structural behaviour in fire cannot be described by a test on a single element. with a severe period for about two hours. Large deflections were not a sign of instability and local buckling of beams helped thermal strains to move directly 21 . The methodology and principles outlined in the guide are still applicable today. 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 900°C). In all tests there was considerable deflection of the composite floor slab in the region of the fire. The columns were generally protected to their full height. Figure 5 Aftermath of the Broadgate Fire Figure 6 One of the Cardington frame fire tests 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. 2. primarily because it was difficult to prove. The structure of the building consisted of composite steel deck/concrete floors. However. there was no collapse of any of the columns. In June 1990 a fire developed in a large contractors hut on the first floor of the 14-storey Broadgate building. Despite some large deflections (see Figure 5). for most of the duration before runaway failure (not observed at Cardington). see schematic representation in Figure 7. The University of Edninburgh 2000). 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. The total duration of the fire was in excess of four-and-a-half hours. it is only in relatively recent years since the Broadgate Phase 8 fire in London. 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). Through experimental and theoretical studies an empirical equation for the critical deflection of beams and the corresponding critical load was derived.5 for office buildings allowing greater limiting temperatures for this occupancy. Furthermore. The load factors applied to live load at the fire limit state have been reduced to 0. beams. Pettersson developed a series of calculation methods based on structural engineering principles for steel members in fire. Bailey and Moore 1999. BS 5950 Part 8 does not.3 Swedish Design Guide Over 27 years ago Pettersson and co workers (Pettersson et al 1976) published one of the most innovative design guides for fire safety design of structures. The fire detection system had yet been fitted and smoke filled the entire building.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 The latest revision of BS 5950 Part 8 was issued in December 2003. Usmani et al 2001.4. 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 steel structure was partially unprotected at this stage of the construction. allowing for creep effects. 2. The guide advocates the use of natural fire curves and heat transfer calculations to obtain protected and unprotected steel temperatures in fire. UK and the subsequent Cardington frame fire tests that researchers have fully investigated and understood the behaviour of whole frame composite steel-concrete structures in response to fire. All of the available design methods up to 1990 ignored the benefits of whole frame behaviour.
Quintiere et al (2002) predicted that the lack of fire proofing was critical and compared failure times with the predicted level of fire protection. By considering this in the structural design robust anchoring of the slab can be detailed. A full structural engineering analysis of a tall building allows structural mechanics in fire to be used and predicted deformations and damage to be assessed.e. Some have centered around a fire induced collapse rather than as a result of the impact. seismic design. robust design of structures for fire without total reliance on passive fire protection. Thermal expansion of the many floors on fire pushed the exterior columns outwards. more thoroughly. At some stage the exterior columns were no longer restrained by the other floors and buckled out. (b) Axially restrained δ (a) 1D Catenary action in beams (b) 2D Tensile membrane action in slabs Figure 7 Catenary action in beams and tensile membrane action in slabs An indeterminate structure such as a multi-storey frame is capable of transferring load through many alternate load paths. Hypotheses have ranged from lack of fire proofing on the floor truss members to failure of the simple connections. Near failure. They postulate that the towers may have collapsed as a result of geometric changes in the structure caused by thermal expansion effects and not material degradation at high temperatures. gravity loads and strength will again become critical factors. 2.6 Collapse of the WTC towers 9-11 The events of 9-11 (see Figure 8) highlighted the inherent robustness of 3D structures in fire. The Cardington frame tests showed that the slab is key to the increased strength of composite frames in fire. This is true at ambient and high temperatures in fire. Several theories have been published in the public domain regarding the cause of collapse of the WTC towers. Collapse as a result of buckling of the exterior columns over the height of many floors has been suggested by Usmani et al (2003). The work by Usmani et al (2003) ignores the damage as a result of impact but does consider the possible structural mechanisms at impending failure as a direct result of the heating regime on the structure. Consequently the pattern of forces and stresses in an indeterminate beam (as part of a structure) are determined by the relative stiffness of the other parts of the structure as well as equilibrium and compatibility considerations. Tall building design for fire needs to be treated in a similar manner to other structural engineering problems e. tall buildings with slender floors. The towers survived the structural damage as a result of impact and flashover fires on several floors simultaneously for 56 minutes in the case of the South Tower and 1 hour 43 minutes in the case of the North Tower (FEMA 2002). In tall buildings a reduction in passive fire protection would typically not be advocated but more importantly very specific structural detailing to cope with the heating regime. Their analyses were based on critical temperatures and single element behaviour which cannot predict the complex response of a multi-storey tower with fires on multiple floors.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 into deflections rather than cause high stress states in the structure. Passive fire protection can be applied as required and increased if necessary in critical areas. The collapse mechanism is likely to be unique to the type of structure that the WTC towers represented i. If a structure has adequate ductility and stability the redundancy under fire conditions enables the structure to find different load paths and mechanisms to continue supporting additional load when its strength has been reached at a single location. The events of 9-11 have forced building designers and clients to consider.g. 22 .
This resulted in a reduction of fire resistance rating from 2 hours to 90 minutes. New York Due to its simplicity the standard fire resistance test misses vital structural phenomena found in the 3D behaviour of real buildings including. without compromising safety. The remaining two sides are steel frame with cladding. The case study. The floor slabs are of composite steel and normal weight concrete construction. is an office development in London consisting of an 11-storey office. eight storeys above ground and three below. The finite element modeling showed that all secondary steel beams could be left unprotected and that the response was the comparable with a fully protected prescriptive design.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 Figure 8 The WTC collapse 9-11. During detailed design a finite element approach was used to assess the structural behaviour.MEDIUM RISE OFFICE BUILDING A time-equivalence approach was applied to this building during the concept phase of the building design. In the case of low risk. Membrane and catenary load carrying mechanisms in slabs and beams These phenomena can be captured in finite element models and structures designed for fire in an engineering manner. CASE STUDY 1. which are also designed for fire fighting. presented here to compare and contrast performance based design and a “code compliant” structure design for fire. Two sides of the building have a load bearing stonework façade which behaves as columns at 3m centres. There is a concrete core at the centre of the building containing services and escape stairs. The building is medium rise therefore relatively low risk. In this height of building it is desirable to make cost savings where possible. The floor plate measures 40mx60m (see Figure 9). • • • Large deflections and nonlinear geometry Restrained thermal expansion and thermal bowing. 3. Secondary and primary steel beams generally span 9m although on the south side of the core secondary beams span 10m. 23 . medium rise buildings this can lead to savings in passive fire protection. Composite action is achieved by shear studs between the top flange of the beams and the concrete dovetail deck slab.
This paper describes the result of the larger model developed. 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. The primary. this is unlikely to cause structural collapse i. • • 3. Horizontal compartmentation is also assessed by monitoring the rate of deflection of the composite floor. (model 1) the extent of which is shown in Figures 9.1 Failure criteria in composite frame structures In order to assess the data provided from a finite element analysis some means of defining failure criteria must be established. stability requirements can be met.e. On this basis the following are proposed as acceptance criteria: • Stability of structure maintained throughout the design fire. This is primarily assessed by looking at the rate of deflections during the fire. although a compartment fire may lead to large deflections of main and secondary beams. A rapid increase in deflection in any region of the floor plate implies compartmentation failure. The columns and the steelwork in the fire fighting shaft and the core would be fully protected.2 The fe model Two finite element models were developed using the commercial software package ABAQUS (see Figure 9) to assess the structural behaviour in a typical floor plate. 10 and 11. 24 . The proposed protection arrangement is shown in Figure 10 in the region modelled. edge and short secondary beams highlighted in pink would be protected leaving the main secondary beams bare. The term ‘failure’ is not straightforward to define in the context of this analysis on the basis that. Runaway deflections (a rapid increase in the rate of deflection) would indicate failure of the floor.Fire and Structures Conference The Royal Society of Edinburgh Masonry facade 21/04/04 Model 1 Model 2 (beams span 10m) Figure 9 Plan of the office building 3. However for compartmentation large deflections could cause a breach of the separating function of the element.
columns and slab ribs. Slab shell elements are not connected to columns because stress can ‘flow’ around the column as a result of slab continuity. Values of thermal expansion for steel and concrete were also taken from the appropriate Eurocodes. Thus the load assumed to act over the floor slab of a typical office floor in the model is 7. In other models not described here. • • • Four noded shell elements were used to represent the slab.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 Columns acting over 2 floors Shell elements representing the slab Fire below floor slab Figure 10 Proposed protection arrangement Figure 11 The finite element mesh The material properties used in the model are given in Table 4. it was conservatively assumed that the masonry wall provided no restraint because the restraint stiffness and reliability of the connections to the masonry façade was unknown. Each element was associated with its appropriate section properties and material characteristics.8 and 1. Symmetry boundary conditions were applied along the sides of the model parallel to the secondary beams. Two-noded beam elements were used to represent the beams.85kN/m2. The columns were modelled on the fire floor and the floor above (see Figure 11).0 respectively. These boundary conditions simulate the continuity of the columns at the base of the structure and at the top of the columns. the partial factors to be applied to live and dead load are 0. full degradation of the stress-strain curves with temperature was allowed. Table 4 The material models Material Light weight Concrete (slab) Reinforcing mesh Steel (frame) Grade C30 S460 S275 Model Eurocode 2 Eurocode 2 Eurocode 3 In accordance with BS 5950 Part 8 (1990). 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. In this model the short secondary beams were assumed to be axially restrained by the masonry façade but rotationally free. Slab elements are connected to beam elements using constraint equations between the beam and slab representing full composite action. 25 . at the fire limit state. Slab and beams were fully fixed at the core wall.
Steel temperatures were calculated using the heat transfer equations in Eurocode 3 Part 1.0 Prot column Prot short secondary beam Unp secondary beam Prot primary beam Prot secondary beam Standard fire curve 1000.0 400. The rate of deflection is very linear similar to deflection plots from the Cardington tests. The slab was assumed to be at a uniform through depth gradient over the whole compartment.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 3. The position of the columns are clearly visible. Floor to floor compartmentation is provided. The mid-span displacement of a typical unprotected secondary beam is shown in Figure 14. The steel temperatures are illustrated in Figure 12.2. Runaway failure (a rapid increase in the rate of deflection) is not observed.0 800. In this paper a selection of results from a design fire equal to the standard fire only are presented. Blue and red shades highlight the greatest deflections both negative and positive respectively. A 1D heat transfer model was used to establish concrete gradients through the depth of the slab. The fire acts underneath the slab modelled.0 Temperature (C) 600. The structure is very stiff at the corner of the building where the short protected secondary beams make a stiff closely spaced grid. The greatest downward displacement is near the mid-span of the unprotected secondary beams as expected. There is very little displacement in this region. 1200.0 200. It is plotted against unprotected secondary beam temperature.4 RESULTS: Proposed structure with unprotected secondary steel beams 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. 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.3 Structure temperatures Sprinklers are conservatively ignored therefore it was assumed a credible fire would be a full flashover on any one floor of the building. 3.0 0 1000 2000 3000 Time (sec) 4000 5000 6000 Figure 12 Steel temperatures used in the fe model. 26 .0 0. The gradient over the depth of the beam is much less important because it is very small in comparison.
As expected the beam is in tension initially then the unprotected steel expands against the surrounding structure producing compressive forces very rapidly until 140°C when the steel reaches its first yield. Beyond this temperature the axial force declines in compression with an increasing loss in material strength and stiffness until at the end of heating the axial force is effectively zero.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 490mm Figure 13 Contour plot of deflection at the end of heating Figure 14 Mid-span deflection of a typical unprotected secondary beam The axial force at mid-span of a typical secondary beam is shown in Figure 15. Figure 15 Axial force at mid-span of the unprotected secondary beam 27 . At this stage the slab is carrying load in membrane action.
This is contrary to the common belief that protected structure does not deflect.3% of the total strain values. C T ~0. The maximum deflection experienced is 390mm. T=tension. In general the slab is in compression or low tension.e.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 The total strains (thermal + mechanical) in the slab at reinforcement level are plotted in Figure 16 for the Y(=2) direction. 28 . C=compression. identical structural member replacement measures after a fire of severity assumed in this model.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. There are regions of relatively high tension (2-3%) near the core as expected. Compressions have negative values and tensions positive values. 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. This can be compared to the design case with unprotected secondary beams where the maximum deflection is 490mm and the mid-span deflection of the unprotected secondary beams is about 450mm (see Figures 13 and 14).5 RESULTS:Fully protected structure A direct comparison has been made between the structural behaviour observed in the ABAQUS model of the proposed design with that which would normally be designed as a result of the recommendations in the UK Building Regulations i. all structural steel protected. Therefore in terms of damage to the structure in the context of insurance.1% C ~2% T 1. the traditional design approach and the proposed design results in. Figure 17 is a contour plot of deflections at the end of heating when all structural steel is protected. The heating regime in this analysis is based on the assumption that the protected steel will reach a maximum temperature of about 550°C at the end of 90 minutes. These are mainly as a result of the hogging moment at this boundary. Note also it is in excess of the BS476 requirement for L/30 deflection limits for beams/floors.1-0. Most secondary steel beams deflect up to 200mm. Thermal strains will account for about 0. 3.
The strains in the 2 direction are shown in Figure 19. 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. Figure 18 Deflection at mid-span of the secondary beam 29 .e. The uniform rate of deflection was also observed in the measurements made at Cardington during the fire tests. Tensile strains along the core edge are in the region of 2%. The strains experienced in the slab when all beams are protected are very similar to the design case with secondary beams unprotected. the rate of deflection is not changing (see Figure 18).Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 390mm Figure 17 Contour plot of deflection at the end of heating The deflection at mid-span of a typical secondary beam is very linear i. The greatest tensile strains are around the column locations and at the core wall. It could be expected therefore that the slab would also experience local cracking in the fully protected case.
CASE STUDY 2 HIGH RISE HIGH RISK BUILDING IN FIRE Case study 1 considered a medium rise office where it was acceptable to reduce costs by removing fire protection. Global models of this size and larger are being used to understand the global response of a typical tall building frame to fires on multiple floors. Equilibrium considerations drive propagating failures. which may never be found leading to global failure. It extends the work of Usmani et al (2003) to assess the robustness of tall building design in particular structures similar to the WTC towers with truss floor systems and relatively slender slabs. In very high rise buildings especially since 9-11.6 Significance of results 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.g. More detailed connection and truss models have been used to understand the local structural mechanisms in a fire. Limiting failure mechanisms are almost entirely driven by compatibility of displacements and rotations. structural fire engineering has been used at Arup to ensure robust design at the fire limit state.2% T 1% Figure 19 T=tension. The finite element models allowed approving authorities. This type of design process must be carried out on a case by case basis. When quoting insurance premiums insurers have traditionally had to guess the likely damage to structures in fully flashed over compartment fires because real structural behaviour is vastly different from the standard furnace test. For example excessive 30 . There are many possible failures at both local and global level in tall buildings some of which can be associated with relatively low fire temperatures. Failure mechanisms can be generally categorized in two groups as either limiting or propagating.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 C C T 0. Strain in the 2 (Y) direction at the level of the reinforcement in the slab. where the loss of load resistance locally spreads the local failure outwards from the area to find new equilibrium positions. This particular case study is a theoretical study being carried out in collaboration with Edinburgh University. The structure wants to adopt a particular displaced shape as a result of the strains and displacements caused by loads and/or thermal expansions it experiences in fires subject to the constraints of compatibility. The modelling methodology provides invaluable information for all concerned. Figure 20 shows a model of a single floor. continuous rupture (unzipping) of the reinforcing mesh in the floor slab then this will result in a huge redistribution of the loading from the area of failure to alternative load paths. C=compression. This model has been used to understand the response of a typical floor to a fully flashed over fire. 4. A finite element approach allows designers to consider any credible fire scenario including fires on multiple floors. If the failure mechanism is propagating e. insurers and clients to see the likely damage rather than relying on prescriptive guidance. It should be noted that the results of these models is for this particular building and in another structure with different spans and layout the results of a similar comparative study may not be so similar. Figure 21 shows one of the multiple floor models. In part this has been driven by client demand. 3.
These local failures will not normally spread any further. Figure 20 Finite element model of a single floor Figure 21 Finite element model of multiple floors 5. Our understanding in this field of research and development has increased significantly in the last 20 years in particular. 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. 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 . The following assumptions were made: • • • • Sprinklers were conservatively ignored All primary and secondary steel beams are composite with the floor slab through shear studs. An increased understanding of these load carrying mechanisms and failure criteria in this case study will lead to more robust design for tall buildings. 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. UK. satisfies the appropriate functional requirements of the Approved Document B of the Building Regulations. CONCLUSION This paper provides a short history of structural fire engineering from prescriptive to performance based design.Fire and Structures Conference The Royal Society of Edinburgh 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. In this instance the load carrying mechanism may not greatly change and there would generally be a small redistribution of loading. 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. Now 3D modeling using finite element analysis is being used in research and consultancy to engineer fire resistance. 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.
BS 5950:Part 8:1990 Structural use of steelwork in building. Building Regulations. In particular the deflections predicted for the fully protected structure were very similar to the proposed design solution with bare secondary steel beams. 8. Copenhagen. Drysdale D.W. 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. Design of tall buildings has changed since 9/11. 2000. British Steel data on the Cardington fire tests. 17. Guide to the application of fire safety engineering principles. BS 5950:Part 8:2003 Structural use of steelwork in building. 3. Technical report. 1976. Reinforcing mesh within the concrete slab will be adequately lapped in all areas. Cadorin J-F. 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. Vol 38 No. An Introduction to fire dynamics.D. World Trade Center Building Performance Study: Data Collection.J. The behaviour of full-scale steel framed buildings subject to compartment fires. FEMA 403. 12. DD ENV 1993-1-2:2001. pp395-427. May 2002.hks. In First International Conference.B. Fire and steel construction: Fire safety of bare external steel. 77(8). Building research Institute. Technical Report 65. British Steel. and Thor J.. Eurocode 3 Design of steel structures Part 1. Part 1: pre and post flashover compartment fire model. 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. 32 . Structures in Fire. Eurocode 1: Actions on structures-General actions-Actions on structures exposed to fire. and Thelandersson S. Magnusson S. 11. Kirby B.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 • • All steelwork within the core will be protected for 90 minutes fire resistance.com Bailey C. It can be accomplished using finite element modeling in 3 dimensions considering nonlinear geometry and the forces as a result of the heating regime.2 General rules-structural fire design. BS 476 : Part 20 : 1987 Fire tests on building materials and structures. Publication 50. and Moore D. DD ENV 1992-1-2:1996. Japan. 1996. Non-linear modelling of three full scale structural fire tests.E. John Wiley and Sons. insurers and approving authorities could quantify the differences in response between the design they would normally approve. Stockholm. 7. Federal Emergency Management Agency (FEMA). 15. 15-21. Preliminary Observations and Recommendations. 2000. Code of practice for fire resistant design. and Franssen J-M. Ministry of Construction. 18. and Plank R. 14.R. Fire engineering design of steel structures. 1970.2: Fire resistance. Quarterly Journal of the national Fire Protection 1. Ingberg S.E. 1995. The Steel Construction Institute. 20. Technical Report 11. 9. Technical Report 1971. Burgess I. Estimation of fire temperature-time curves rooms. pp. 2. 5. Office of the Deputy Prime Minister (ODPM) Fire Safety to Approved Document B. ABAQUS www. Law M. Fire loads. 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.G. 2003. 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. 1986. and Sekine T. Tokyo. The Structural Engineer. Law M. Code of practice for fire resistant design. Huang Z. A relationship between fire grading and building design and contents. June 2000. because it complies with prescriptive guidance. 1999.. 5.H. 4. temperature time curves of the complete process of fire developmenttheoretical study of wood fuel fires in enclosed space. and the performance of the proposed design with some bare steel. 1963. A tool to design steel elements submitted to compartment fires-Ozone V2. Eurocode 2: Design of concrete structures Part 1. 6. Kawagoe K. 16. 19. This could not be accomplished using simple critical temperature calculations and single element behaviour. The comparative study was invaluable in the approvals process because the fire brigade. and O’Brien T. Pettersson O. Swedish Institute of Steel Construction. 2nd Edition 1999. 13. Magnusson S. REFERENCES 1. BS EN 1991-1-2:2002. 10.
. and Becker R. di Marzo M.. Fundamental principles of structural behaviour under thermal effects.Broadgate 24. www. Quintiere J.B. The University of Edinburgh (2000) Final report of the DETR-PIT project: Behaviour of steel framed structures under fire conditions. Fire Safety Journal.C.S. 37.G.J. and Heselden A. Rotter J. Thomas P. Investigation of Broadgate Phase 8 Fire. The Steel Construction Institute. Chung Y.ac. 7..L. Vol 38.uk/research/fire/project/main. Vol. Sanad A.S. 2001. No. PROFIL ARBED Centre de Recherches. Fire Safety Journal Vol. 1999. Fire Safety Journal.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 21..M.civ. Technical Report.G.. 23.H. pp707-716. A suggested cause of the fire induced collapse of the World Trade Towers.. 22. and Torero J. 8 pp 721-744. March.html 25. Lamont S. Usmani A. 36. Cajot L. June 1991.. Schleich J. 2002. Fully developed fires in single compartments a co-operative research programme of the Counseil International du Batiment (CIB) 1972. Usmani A. Competitive steel buildings through natural fire safety concept. pp 501-533.M.M. Final Technical Report composed of 2 volumes. and Gillie M. 2003. 33 . 27. No. 26. Technical report. and Pierre M. How did the WTC towers collapse: a new theory.ed. Structural fire Engineering.
known as zone and field (or CFD for Computational Fluid Dynamics) modelling methodologies is in the way they treat the movement of the products of combustion within the enclosure and their respective reliance on empirical information. Zone models make the assumption that the products of combustion from an initiating fire will fill any enclosure from the ceiling down much as a bathtub fills up with water. They include evacuation for the evaluation of occupant escape in the event of fire and models for structural response to fire. Fire resistance furnace tests dating back to the beginning of the 20th century represent an early type of physical model of fire. However very little of our modern understanding of fire is incorporated in such models Modern physical models are often based on dimensional reasoning to exploit reduced-scale representations of “reality”. within enclosures is possible using either of two different modelling methodologies. Physical models will have a contribution to make to understanding the collapse of the World Trade Centre.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 Professor Geoff Cox Building Research Establishment. FIELD AND ZONE MODELS Computer simulation for the dynamics of the gas phase of fires. particularly. These include models of heat transfer from the fire to the key structural elements of a building and models that describe the response of those elements enabling the structural integrity of a whole building to be assessed. UK FIRE MODELLING INTRODUCTION The modelling of fire has been crucial to the modern development of fire safety science. floors or partitions are exposed to the controlled conditions of a furnace used as a surrogate for fire for the purposes of component evaluation. But the main purpose of this presentation is to consider our progress in the computer modelling of fire and its impact with particular reference to the disaster. Here building elements such as walls. on structures or on the environment. well-stirred reactor after the fire flashes over. Salt-water. But with the arrival of cheap and fast computer processors. Since such models are relatively less demanding of computer power than their CFD counterparts they provide a greater opportunity to conduct Monte Carlo simulations for use in quantitative risk assessment. numerical simulation tools have taken centre-stage for the engineering practitioner to exploit the freedoms offered to designers by performance-based regulation. hot. Both physical and theoretical models of “reality” have spearheaded our progress towards the blossoming profession of Fire Safety Engineering. One-zone post flashover models have been used for many years to assess the impact of fully developed fires on their enclosing structures whilst two zone models have enjoyed considerable success in application to smoke movement problems. The two separate cool and hot zones become essentially one single. etc. These allow systematic quantitative study of the effects of changes to fire size. characteristic. the walls etc) is then described by mathematical relationships derived from dimensional analysis supported by experimental measurement. released into a fresh water model environment to represent the influences of the buoyancy of fire gases has provided valuable insight into the movement of smoke within enclosures. Having made the assumptions as to how to break down the problem into its constituent parts. lengths etc. each individual part or zone (eg the plume. The difference between the two. the hot layer. 34 . This is particularly valuable for scoping studies and for evaluating the sensitivity of the outcome to variations in scenario. There are also “sub-models” used as part of the fire model that describe the behaviour of fire protection devices such as sprinklers and detectors and there are models for the most complex of all these and that is the detailed spread of flame over fuel assemblies. Such models are now available with varying degrees of sophistication and scope to describe every aspect of the fire problem. However at the core of all fire modelling must be a treatment for the gas phase phenomena of the fire itself for any assessment of impact on people. for example. These model types are very closely related to the reduced scale physical models referred to earlier and to earlier theoretical models not enjoying the benefits of the modern digital computer.
for example the atmospheric boundary layer. this type of model is of universal applicability. Essentially the RANS methodology views the transient signatures of local gas temperature. rigorous solution of these "exact" equations. since the LES models need finer numerical meshes than the RANS models there is a consequent increased computational demand. FIELD MODELS-RANS. RANS modelling is relatively less demanding of computer power and as a consequence has been the more widely used of the two approaches in practical application [eg 1] whilst the LES approach is becoming increasingly attractive as computer power continues to increase. These encompass the effect of turbulence influences across the whole of the turbulence spectrum from very large room-scale turbulent eddy sizes (several metres or more) down to the very smallest scales (order of a mm) associated with the viscous dissipation of energy and the chemical kinetics. newer. These models solve only for timeaveraged equations which describe the principles of mass. known as RANS (for Reynolds Averaged Navier Stokes) modelling. resolving fully the length and time scales that occur in the turbulent reacting flows characteristic of fire. the computer is the enabling technology. views the transient signatures as comprising larger eddies. momentum. As with the RANS models LES models provide predictions of time-averaged properties such as gas temperature at each grid cell but now obtained by averaging the time dependent predictions from the model as the experimenter would average his thermocouple measurements. What determines the size of eddies that are resolvable and those that are not is the fineness of the numerical mesh. Field models.Fire and Structures Conference The Royal Society of Edinburgh 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. However. For these problems the alternative modelling approach is more appropriate. The inexorable progress in the availability of computer power has spurne d two distinct branches of CFD modelling that have evolved. momentum. is still beyond the capabilities of even the largest computers currently available except of the simplest of problems. is known as LES (Large Eddy Simulation) modelling. The first of these. Indeed the LES models originally developed by Baum and Rehm  are now demonstrating their utility to the ongoing NIST investigations of the World Trade Centre disaster. 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. The LES methodology. to experimental correlations and instead return to first principles to solve the basic laws of physics for the fluid flow. 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. the technique could not have developed because it involves millions of calculations for every step forward in time that the simulation makes. This choice needs to be made with particular care since coarse meshes can give misleading results. It has only been recently that such a demand for practical problems could be met by 35 . make essentially no assumptions about the fluid dynamics of the combustion products. chemical species and energy within the fire and throughout any enclosure containing it. They do this on a numerical mesh of maybe millions of elementary “control volumes” throughout the calculation domain. has been in practical application in many branches of engineering for well over two decades. Clearly. To capture the details of the chemical reaction zone in a fire would require a characteristic mesh size below one millimetre. energy and species conservation. as far as possible. instead. Without it. They avoid resorting. The RANS models predict the temporal evolution of the time-averaged properties of the fire at millions of spatial locations throughout the enclosure of interest. As a consequence of this resort to first principles. LES and DNS For field modelling. the other. resolvable by the calculation procedure and smaller unresolvable eddies which need to be modelled in a manner similar to the “turbulence models” used in the RANS methodology. velocity or chemical species as comprising a time averaged component and a fluctuating perturbation about that average. by contrast. The two branches differ primarily in their respective treatments of the effects of turbulence on the heat and mass transfer processes and on chemical kinetics. LES models make less assumptions about the role of turbulent mixing than RANS models. The influences of the turbulent fluctuations are included by “turbulence models” which help close the system of equations. As a consequence it is necessary to simplify the system of "exact" equations by some form of modelling.
Best practice guidance on the use of these types of model is now available for RANS models  and will be an urgent requirement for LES models as they become increasingly attractive to design practitioners. 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 . no matter which strategy is adopted. The fluctuating gas velocity at a particular point in the fire is shown. 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. It has a relatively slow underlying increase from left to right as the fire increases in its heat release rate. The fire literature contains many comparisons of CFD predictions with experimental data. the combustion and radiation models need to be validated to test their representation of reality. others are also introduced as a result of the numerical methods used to solve the continuous equations. Whether the model equations are then solved to adequate accuracy etc is then the issue of verification.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 readily accessible computer power. This is very relevant for the prediction of conditions within the World Trade Centre Towers. This attempts to capture all the length scales involved in the underlying physics and chemistry without resort to any turbulence modelling at all. One of the most important model validation exercises was recently conducted under the auspices of the Fire Commission of CIB (Conseil International du Batiment). Most. Although the floor area and the heat release rates are substantially larger in the Twin Towers the fire conditions will not be dissimilar. 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. Fig. 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. The degree of resolution of the contributions made by the different wavelengths achievable by the various approaches is illustrated. It is likely that this approach will become increasingly attractive as computer power develops further. is still inadequate to allow rigorous simulation of all the scales for domain sizes of practical interest to the fire engineer. 36 . In addition to uncertainties associated with the modelling of turbulent flow. have been conducted with a prior knowledge of the experimental results. An appreciation of the degree of temporal resolution that these different methodologies capture is illustrated by the schematic Figure 1. It involved a series of 'blind' simulations of unpublished experiments conducted in the 1980’s by VTT in Finland . 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. Models of this type are still a very long way from practical application to fire. however. They represent different aspects of the phenomenon as a whole. One of these cribs was ignited and fire was allowed to spread from the first to the second crib. They contain elements of both validation and verification. 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. despite the enormous progress of the last few decades. But in addition to this general trend there are fluctuations of different wavelength. They use the following definitions of validation and verification: Validation Verification 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. There is an even more computationally demanding methodology known as the Direct Numerical Simulation (DNS) approach that should be the most accurate. 1 Schematic of gas velocity at one point in a fire. This is because computer power. for example the turbulence treatment by RANS or sub grid scale model in LES. Two wooden cribs were located inside an enclosure containing a single high level slot opening (Figure 2).
Mass loss rates were determined from the raw weight loss data through which a smooth curve had been fitted and time derivatives determined.8 m 7.0 m 1. 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.2 m Fig. agreement can be seen to be generally acceptable for many practical purposes. Figure 3 shows the resultant mass loss rates for the two cribs. It will be seen that gas temperatures and major gas species concentrations have been reasonably well reproduced.6 m 7. A selection of the results from the JASMINE model is illustrated here (Figures 5-8) . 1.24 m 3. A measurement of the effective heat of combustion from oxygen depletion calorimetry throughout the duration of the fire was also supplied (Figure 4). Although not perfect. The discrepancy evident in the minimum oxygen concentrations is due to instrumentation sensitivity limitations at low concentrations.6 m concrete beam opening 0. 2 Geometry of the experiment Other than the geometry and the thermal properties of the wall materials.6 m 37 .6 m concrete beam ignition point 1.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 plan view A corner crib centre crib 3.2 m A vertical section at A-A 0. This aspect of prediction provides additional complications which were not to be tested in this exercise.
3 0 Mass loss and heat release rates of the two wooden cribs 0 20 40 60 80 100 Time (min) 30 Effective heat of comb (MJ kg ) -1 25 20 15 10 time-dependent ('true') 5 0 0 20 40 60 80 100 120 fixed ('blind' simulation) Time (min) 1250 Fig.1 corner crib centre crib 2.15 0.5 2 1.5 0.25 0.2 0.5 0 120 Fig.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 0.05 0. 4 25 Measured heat of combustion as a function of time Volume fraction (%) 20 15 10 5 0 0 10 20 30 40 50 60 70 80 90 measurement JASMINE Time (min) Fig.3 3.5 1 0. 5 Comparison of predicted and measured CO2 volume fractions 25 Volume fraction (%) 20 15 10 5 0 0 10 20 30 40 50 60 70 80 90 measurement JASMINE Time (min) 38 Heat release rate (MW) Mass loss rate (kg s ) -1 3 .
Fire and Structures Conference
The Royal Society of Edinburgh
Comparison of predicted and measured O2 volume fraction near the centre of the ceiling
1000 800 600 400 200 0 0 10 20 30 40 50 60
T1 T2 T1 T2 measurement measurement JASMINE JASMINE
Predicted and measured gas temperatures close to the back wall
-2) F1 F2
40 30 20 10 0 -10 0 10 20 30 40 50 60 70 80 90 Time (min)
F1 F2 F1 F2 measurement measurement JASMINE JASMINE
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.
SOME APPLICATIONS OF MODELLING
Many commercial applications of models of this kind have been for the assessment of smoke control design strategies with field modelling being the method of choice in innovative designs. It is in these kinds of structure where the traditional building regulations are often not readily applicable and an engineered solution is, of necessity, required. Covered shopping malls, atrium hotels, leisure complexes, airport and railway terminals are just some examples of where the technology is finding its utility for the practising engineer. Often these structures are unique in nature but increasingly the models are being used for more routine problem types. An illustration of the use of CFD field modelling in conjunction with an evacuation model to assess the progress of smoke as it hazards a building’s occupants is provided in Figure 9. The illustration shows the extent of spread of the combustion products, three minutes from ignition of a luggage fire in the terminal building of Brussels International Airport. In this example the heat release history of the luggage fire source has been taken from experimental data and used as input to the model. The occupants are making their escape based on assumed detection of this fire at one minute from ignition.
Heat flux (kW m
Fire and Structures Conference
The Royal Society of Edinburgh
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
Fire and Structures Conference
The Royal Society of Edinburgh
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.
Here a mechanism for the development of rapid fire spread over the wooden escalators. Zone modelling was used to study the 1980 MGM Grand Fire in Las Vegas with the first dramatic contribution from field modelling  being provided during the inquiry into the 1987 Kings Cross underground station fire.temperature profile at two minutes in steel beam below a concrete slab exposed to a constant temperature of 500o C MODELLING IN DISASTER INVESTIGATION Computer fire simulation has seen increasing application to fire investigation particularly for large disasters. The fluid mechanics of air entrainment was locally essentially two-dimensional from above and below the fire. 42 . 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. 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) .Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 500 0 Concrete Figure 11 . 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. 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. not considered by investigators. was suggested by the modelling. 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.examples for specifying structural components in the fire engineering interface Steel I- Figure 12 . Clearly this work is ongoing and much more will be made available in the near future. The flames behaved this way only when the fire occupied the full width of the trench. 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. 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.
43 .Fire and Structures Conference The Royal Society of Edinburgh 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.
6(b) 0.6 MW fire across the width of channel whole Fig.5 MW fire across one third of channel width 44 . 6(a) 1.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 Fig.
but they do not provide quantitative information that can be used by the engineer. such simulation tools will inevitably become increasingly attractive. 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. more sustainable and cost effective than at present. . With the increasing international trend towards performance-based fire regulation. should deliver a future built environment that is safer. ISO TC/92 is working with the fire commission of the International Council for Research and Innovation in Building and Construction. This will take time but an important start has been made. The first Fire Safety Engineering standards documents were in 1999 as an 8 part Technical Report (ISO TR 13387 Parts 1-8). Often they simply supply pass or fail information only. Furthermore relative 'success' in the test does not necessarily ensure relative 'success' in the 'real world' application environment. 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'. CONCLUSION The use fire modelling both in support of fire safety design and in fire investigation is growing rapidly. Relatively low license and hardware costs now ensure much greater accessibility than ever before. The deficiencies of traditional test methods have been known for many years. Such tests are useful to rank products in the test. following the tragic events of 9/11. Only the cone calorimeter has been devised with such a purpose in mind. the International Organisation for Standardisation. 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. The expectation is that with a new testing approach coupled to a predictive capability to calculate both 'test' and 'real world' exposure scenarios. 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. 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. CIB W014. 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. it is the fire resistance tests that have been propelled to the top of the agenda. These are currently being extended and developed as full standards. However.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 inexorable improvement in computer hardware capacity is also likely to influence the type of CFD model that will be used. 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. to deliver this strategy. Such models are no longer restricted to just the academic developers and research institutions. Technical Committee 92 “Fire Safety” of ISO. 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. using modelling. particularly now that they can be used on personal computers. then it should be possible to assess performance for a full range of practical possibilities. maybe for quality assurance purposes.
5. AIAA. 1998.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 REFERENCES 1. 1998. Bethesda. 49. 161. Boston. 2000. London. Academic Press. 1995. Chapter 3. 28. Kumar S and Cox G. Design and Simulation Reports of CIB W14 Round Robin For Code Assessment. 1989. 8. 46 . 18. 6th International Symposium on Fire Safety Science. Comb. Reston. Luo K H and Williams JJR. Dynamic Behaviour in Reacting Plumes. 9.VA. HR and Rehm RG. to be published. Modelling Enclosure Fires using CFD. 4. 15. 6. Zhou X. Miles SD. Large Eddy Simulations of Smoke Movement. 2000 Cox G and Kumar S. 3. Fire Safety Journal. Guide G-077-1998. Scenario B. Cox G. 2. Wilkes NS and Jones IP.. 7. Hostikka S and Keski-Rahkonen O. Fire Modelling and the King's Cross Fire Investigation. Combustion Fundamentals of Fire. 1989. IAFSS. Fire Safety J. Inst. Computer Simulation of the Flows of Hot Gases from the Fire at the King's Cross Underground Station. SPFE Handbook. VTT Building Technology Internal report RTE119-IR-2/1998. Baum. Chitty R and Kumar S. Society of Fire Protection Engineers. Simcox S. 7. Proc. Cox G. 2001 Guide for the Verification & Validation of CFD Simulations. Fire Safety J. Proc. 1992. 30. American Institute of Aeronautics & Astronautics. Comparison of 'Blind Predictions' of a CFD Model with Experimental Data. McGrattan KB.
47 . “ Terrorist arson” is a premeditated attack on a building using fire to destroy the building and/or injure or kill the occupants. The arsonist is trying to use the target’s own fire characteristics to destroy itself. its contents or its occupants. nor did they have sophisticated arson tools or the ability to sabotage the building. Department of Fire Protection Engineering. which supported the temple. juvenile fire setting or even malicious burning or homicide. But in the fire safety environment it is very useful to distinguish arson from the sabotage of fire protection systems. For comparison consider the difference between “hostage taking’ and political kidnapping. since fire can be used as a weapon. University of Maryland and Visiting Professor in Fire Risk Engineering at Glasgow Caledonian University THE WORLD TRADE CENTER: TERRORIST ARSON AND THE LAW Using Buildings as Weapons The idea of using the destructive potential incorporated in a building to destroy the building itself is not new. Terrorist Arson as a Special Risk Arson (the intentional burning of structures and contents) has always been a major threat to life and property. and he braced himself against them. And Samson took hold of the two middle pillars. Unlike terrorism these events do not attempt to influence a wider target audience. Fire is a design event for buildings. generally for the purpose of influencing a wider audience as opposed to simply targeting the building or inhabitants. since we can often identify the targets for attack or intimidation. ARSON and SABOTAGE “Arson” has always been a complex technical/legal concept. Few mass murderers in the United States have killed as many people as the arsonists at Dupont Plaza (96 fatalities) or Happy Land Social Club (87 fatalities). The Bible describes the destruction by Sampson of the banquet hall and his enemies. “ Sabotage” is intentionally damaging the fire safety protection systems built into buildings to make the arson attack more effective. disabling sprinkler systems or destroying alarms. Such actions should be separately classified from the arson event. as opposed to attacks designed to influence a wider audience. with the ignition following after. by collapsing a vital column. Terrorism must be specifically distinguished from arson for attack or intimidation. Sabotage can be the first event. But terrorist arson represents an additional dimension over and above typical arson. However. For example an arson attack on a newspaper to protest its editorials or an abortion clinic to disrupt its operations are “intimidation” rather than terrorism since they are direct attacks on the target. Political kidnapping involves targeting a specific person who can be specially protected. The arsonists in these cases may not have even intended a mass murder. . “ Arson” is intentionally setting a fire designed to ignite a structure or its contents. The World Trade Center attack would fall in this category. one on his right and the other on his left… And he pushed with all his might. for the specific purpose of damaging or destroying the structure. Arson therefore implies a direct intentional interaction between the arsonist and the fire characteristics of the target.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 Professor Vince Brannigan Lecturer in Building Safety and the Law. Terrorist arson is not necessarily political but it should be distinguished from arson for revenge or excitement. To enhance the destructive effect of fire arsonists can also intentionally damage these control systems by opening fire doors. and the temple fell on the lords and all the people who were in it. arson for insurance. They are criminal arson (and “terrifying” to the victim) but are not terrorist arson. Unfortunately burning them down can be much easier. They simply took advantage of the vulnerability of the building environment itself to cause mass casualties. A structure’s vulnerability to sabotage may be the key component of arson risk. In arson ignition is not necessarily the first event. Arsonists can easily cause catastrophic injuries. The rationale for this distinction is essentially practical. Distinguishing between arson and sabotage will allow regulators to better describe the possible progression of an arson attack from beginning to end. Even a biblical strong man can bring down few modern buildings. arson has also been described as a form of sabotage. so structures normally have “built in” fire protection or control systems. Hostages can be anyone that can be abducted and those taking hostages can easily select the least protected. Sabotage and the size and location of the initial fire often distinguish arson from accidental fires. Judges Chap 16 However the small input energy available to Sampson required a special vulnerability to have the desired effect.
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 can instead select the most vulnerable target for maximum effect. normally in the form of burning fuel. This adds the complex problem of intentional uncertainty to more traditional types of fire analysis. the 47 story building that collapsed after being ignited by burning debris. and social investments in safety are more cost effective when applied to new buildings. Such targets can often be specifically identified and protected. These energies are extremely large relative to the “energy size” of most arson attacks. If the energy stored in the building can be quickly released it represents a major threat to human life. Virtually no one in the World Trade center survived form above the fire.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. in the same way vulnerable political figures can be protected. However terrorists can single out the weak targets and attack them. Any ignition source that could have started simultaneous large fires on several connected floors might have had the same effect on the building. and physical energy in the potential for fire-induced collapse. In the World Trade Center the vast majority of the thermal energy available was in the contents normally in the building. Any society that grandfathers hazardous existing buildings must realize that the risk of terrorist arson corresponds to the “weakest link”.ii The precise relationship between the level of energy release and life safety hazard in any given building will require further work. the release of the chemical energy causing the building’s collapse . For example the World Trade Center upgraded certain fire safety systems only when a floor on the building was renovated. the “grandfathering” of existing buildings. Input energy – Energy supplied to the building by the arsonist. Arson can be used to overcome the systems that are normally installed to protect occupants from the energy stored in our buildings. much the same way as a blasting cap might be used to trigger dynamite. Modern buildings contain enormous quantities of potential physical and chemical energy. more akin to military attacks such as that on the USS COLE. The jet fuel in the aircraft that hit the World Trade Center was essentially only a very effective ignition source to start a fire in the contents. Abortion clinics.Fire and Structures Conference The Royal Society of Edinburgh 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. If an arsonist can trigger a rapid burnout and collapse of a structure. 1 This analysis is of course limited to regulatory issues. 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. there is the reasonable possibility of killing everyone in the structure. . 2 In the World Trade Center the kinetic energy of the aircraft can also be considered input 48 . 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. gives at a least a very rough order of magnitude estimate of the potential hazard involved in arson. Arson Inputs and Outputs Arsonists use an ignition source to provide sufficient “input energy” to trigger the conversion of the building’s own potential to kinetic energy. not the political or legal issue of whether such an attack is terrorism. Effect of Fire The danger of fire is generally related to its energy output. 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. since terrorists can take advantage of a known weakness in our fire safety regulatory system. The toxic gas acts directly on the individuals in the environment. An uncontrolled fire in the contents was enough to bring down # 7 World Trade Center. A large fire in a “fuel rich” environment can rapidly become uncontrollable.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”. Chemical energy is stored in the flammable contents of the building. e. instead of the average level of vulnerability. The building provides most of the potential energy for its own destruction. For the purpose of this discussion the Pentagon attack therefore falls into a different category.g. However energy release can be used as a rough proxy for the creation of a lethal environment and coupled with the number of person exposed. ARSON and the release of stored energy The major distinction between arson and other forms of terrorism is that the hazard from arson comes primarily from the rapid release of the stored energy in the building itself rather than from the flammable material introduced by the arsonist. The chemical energy release generates heat and toxic gas. This “grand-fathering” is a rational act when fire ignition is largely random.
they model the effects of a given “scenario fire” in the building. 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. Hamburg. or a plane crash “initial fire ” is an analytical tool designed to describe the anticipated fire growth in the structures. 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. Thresholds Levels of input are critical because up to a certain point the energy input is simply absorbed by the building without releasing the building’s own potential energy. Obviously the higher the threshold the larger the input required to release the building’s energy. The “energy output” available to the terrorist depends on the building’s fuel load. Existence of a “threshold” is a major reason why bombs may be less effective than fire in causing catastrophic injury. But bomb explosions were insufficient for the desired destructive effect because they could not trigger the building’s energy effectively. Leverage is a crude “cost benefit analysis” using energy as the currency. and even the atomic bomb was fundamentally simply a big ignition source. Many buildings are not designed to even deal with a fast t-squared fire. The World Trade Center towers absorbed and redistributed the forces involved in the airliner hitting the building. and the effect of fire can simply replicate other stresses the building is designed to resist. 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). a quantity of input energy below which the leverage is essentially equal to zero. but rest on assumptions. Regulators have to distinguish among several key issues: “Ignition source” represents the beginning of self-sustaining combustion.g. Buildings are designed for a variety of stresses. often described as a t-squared fire. a match. 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. 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. . All things being equal (e. 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. Dresden and Tokyo all preceded Hiroshima and Nagasaki. (both total output and rate may be relevant) Fires have leverage because: 1) 2) Fires are self-sustaining chemical reactions that use fuel already in the building. Instead of blasting the cities it turned out to be easier to burn them. Firebombing was the most effective use of massed aerial assaults. or the maximum casualties in an area. But an arsonist can create an ultra fast or even faster fire that can overwhelm a buildings defenses. It can be a spark.The ratio in a given arson target between any given input energy and the energy output. and the resistance of the structure and its fire protection systems. It is routinely specified in terms of a fire growth curve.E. The connection between the two is often a matter of judgment and has high uncertainties. number of occupants). But the uncertainties involved in fire growth require all current operational fire models to provide a “scenario fire” as an input to the analysis.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 Leverage. the highest risk exists in those buildings with the highest leverage at relevant levels of input. 49 . The ignition source is thus the terrorist attack. 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.iii In other words the fire models do not really model fire growth. For example in the 20th century aerial attacks on enemy cities were promoted as a military tactic. Buildings had high thresholds and bombs had poor leverage. and the initial fire is the assumed fire size and growth rate. Fire Growth Fire “leverage” is a complex analytical problem because fires normally grow over time in an often unpredictable manner.
Fortunately. especially the size and configuration of the compartment and the effects of the products of combustion. “Normal fires” are accidental. coal is 30. or more typically in kilowatts or megawatts. Fire Hazard of Contents The major fuel available to the arsonist is contained in the building’s contents. Terrorist arson can create a large initial fire with a fast growth rate. The size of the initial fire is crucial since a building with a 5 megawatt threshold may be completely destroyed by a 20 megawatt initial fire. It represents the available energy contained in the contents measured in terms of BTU/pound or kilojoules per kilogram in various calorimeters.000 and oil is 40. or allow evacuation of a fire area before the fire transitions into a fast growth phase. The fire can quickly get beyond any building’s capability to extinguish. Fast Fire Growth The special risk of fast growing fires occurs because building designers and users typically assume relatively slow growing fires in designing building fire protection systems. contain or survive.v Firepower 3 is measured in “Watts”. Ordinary cellulostic and hydrocarbon materials have a fairly narrow 2-1 range E. Fire safety codes and standards are not typically designed to deal with large ignitions. 50 . In later stages of fully developed fires the fuel load become more important.g Wood is 20. The fire hazard of contents is not a simple quantity and is not easily described by reference to chemical characteristics separate from its physical form. most of the time this potential is not released because we use fire protection systems to control fires before they become “large & fast growing”. negligent or even intentional isolated ignitions of the easily available combustible materials present in a building. Instead they focus on what can be defined as “normal fires”.000 Kj/kg. thermal inertia and physical structure of the material. because most of the current regulation of materials and contents is based on issues of ignitability rather than fuel load.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 Fire Power The burning rate and quantity of the flammable material in its environment determines the power output of the fire. Initial fires can be described in kilowatts or megawatts. 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. Fuel load is the effective heat of combustion times the mass of material. Building and Fire codes. In such a case the “initial fire” is “further along” on anticipated the fire growth curve and poised for immediate transition to rapid growth. but it simultaneously emphasizes the vulnerability of many of our buildings. ∗ 3 Dr.000. For the purpose of regulation two very general characteristics must be defined. and there exists no accurate common test method for defining ignitability. rather than low fuel load. 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. This is the tendency of a material to ignite easily when exposed to a flame. Modern buildings are specifically designed to control or suppress growing fires. since fuel load can be fairly easily calculated. In the early stages of typical fires the slow rate of vapor generation (roughly related to ignitability) is the critical limiting factor. The WTC is estimated to be a gigawatt. “Transition “ the point on a fire growth curve where fire growth rapidly accelerates. This may lead to the dangerous assumption that fires always have a slow growth stage. Ignitability refers to the ease of ignition. Fast growing and uncontrollable terrorist fires are possible in our buildings and cities because many of the building elements and contents and procedures that we use to make our lives comfortable and productive harbor extraordinary fire potential. and the air supply can become the limiting factor. James Quintiere John Bryan Professor of Fire Protection Engineering at the U of Md. Firepower determines the range of possible hazards but cannot be converted directly into a hazard analysis without considering the fire environment. natural. A large burning Christmas tree can be a one megawatt fire. This tends to simplify the regulatory problem. It is related to the chemical makeup. A conservative course is to assume that a terrorist can create an initial fire large enough that the ignitability characteristics of the building’s fuel load are far less important than the fuel load itself. Materials vary widely in their ignitability.
Containment failed at the World Trade Center for a variety of reasons that are typical of modern buildings. All can lead to catastrophic loss. The failure led to catastrophic collapse. The theory of containment is that a building can be designed to allow full burnout of a compartment without catastrophic loss. inhabitants or related areas of fire hazards. Operational controls are management systems designed to control contents. But starting a fire in both compartments destroys the underpinnings for this strategy. either by limiting the fuel available or using dynamic fire safety systems to keep the fire “small”. depending on its configuration. 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. It is well known that operational fire safety controls pose very special problems for buildings and the uncertainties in effectiveness are very large. 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. As noted above Ignition is the initiation of burning. Instead the initial fire is normally the object of analytical attention. This is particularly true of dynamic systems such as sprinklers. Person in adjacent compartments would be held back from using the exit paths.vi Fire safety control system Static and dynamic systems build into buildings to minimize the growth and spread of fire. delays evacuation and can allow a fire to grow faster. exit ways. Multiple Ignitions In many modern buildings multiple exit paths are provided to ensure a safe exit from a fire. One approach is to control the flammable contents. The traditional approach was to contain the fire in the compartment of origin. Techniques are often trivial. If a terrorist can start a fire larger than the threshold the result can be catastrophic. But if a fire passes the critical threshold all material and persons in the compartment or building may be destroyed. 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. There has been little or no study to date of the problem of multiple ignitions. Phased evacuations involve planning immediate evacuation of only the persons in the fire compartment. the more complex and expensive the fire protection system and the greater the limitations on the use of the building. and fire doors. In particular multiple ignitions may require re-examination of the routine design concept of “phased evacuations”. 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. Fire power and building thresholds The alternative strategy to “containment” is to try to keep the fire small. Of course a building may have different sized threshold fires in different parts of the building. Sabotage defeats containment. In these buildings the loss of the building is accepted as long as the occupants can escape. This allowed buildings to be designed with fewer exits. The primary technological approach for keeping fires small is a system of automatic fire sprinklers backed up by fire department response. As a result designers often claim that sprinkler systems will keep the fire below a certain size. The root technical cause of disaster at the World Trade Center was the failure of “containment” and fire resistance of the structure. alarms. The object is to keep the initial fire from growing beyond the building’s threshold. Containment is a system for keeping fires within a specified compartment by physical barriers.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 Normal fires generally start small and often grow slowly in their initial stages. Obviously a critical design goal is to keep any fire below the threshold. Most modern buildings have few if any protections against sabotage of the fire protection systems. 51 . A fire protection strategy based on keeping the fire small can be described as keeping the fire power below the building’s defined “threshold”. There are a variety of strategies used to keep a fire below a threshold. In effect the building threshold is set above the fire from the fuel load in a compartment. However virtually all building codes assume a single significant ignition. 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. Sabotage Many fire protection systems are vulnerable to sabotage. Sabotage can affect a building in several ways. so the threshold can be set low. Sabotage can range from putting a wedge under a fire door to shutting off a sprinkler system. While of potential legal importance. Finally some buildings are not designed to survive a significant fire at all. 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.
Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 Public Perceptions and Reactions to Fire Hazard analysis problem does not just involve engineering assumptions. Many such fires lose most of their evolved heat by radiation or convection so that feedback to the fire is reduced. and detail the building to improve its response to explosions. fireplaces and other small vented fires. If a terrorist can create an initial fire larger than the design threshold. A burning building is not a tenable environment even for rescue forces. The problem of progressive collapse was identified but has not been solved. thus simplifying security systems. Evacuation prior to collapse may be impossible. While the localized damaged zone will not be in pristine condition. which gives a misleading indicator of how close an unprotected person can approach a fire. Progressive collapse occurs when a structure loses a support unit (e. there is an enormous potential for disaster This is the gap that can be exploited by terrorist arson. The rest died in the progressive collapse of the structure. In the attack on the Oklahoma City federal building. A public exposed to such images of fire may fail to react appropriately in the event of a terrorist attack and increase the rate of casualties. can create multiple fires or can sabotage the protection system. The public misperception of fire can also contribute to poor anticipation of fire or inadequate reaction to fire risks and actual fires.” vii Analyzing the “leverage” problem of terrorist arson is more complex than the risk of “explosion induced progressive collapse” because: Single bombs big enough to damage large buildings must themselves be fairly large. more easily obtained and less well “tracked” than bomb components. Bonfires and similar free burning fires in large open areas are essentially fuel limited at all times. Bombs (explosive devices) act in a different way from fires. A building which resists an explosion can normally be evacuated since there are no additional energy inputs. Our data base of understanding of fire development may be much more limited. For example jet fuel. The use of terrorist arson as a weapon involves overwhelming or destroying the building’s defenses in a variety of ways. Explosives produce a “detonation” and the shock waves from such a detonation can both directly cause injuries and shatter parts of a structure. because of an explosion) and the load transfer to other members causes additional building elements to fail one after another.g. Fire extinguisher training routinely takes place in the open air. but in some ways more complex than the problem of “explosion induced progressive collapse” identified in buildings exposed to terrorist bombs. Authors have urged the USA to do the same. no toxic gasses and little fire growth. gasoline and margarine can show roughly similar flammability characteristics in a fully developed fire. at least in the USA. resulting in partial or total collapse. Public misunderstanding of the problem of fire growth is routine. When more than 80 percent of the deaths are caused by the structure's falling on top of occupants who otherwise would have survived the blast. This is not true for highly leveraged buildings. They may be totally unprepared for the rapid increase of energy release when the fire transitions to flashover. As a result many believe that large fires are simple extensions of small fires. and is the problem for the risk analyst. Many people have experiences with matches. 52 3) 4) . But once the margarine is burning all three contribute roughly similar fuel to the fire. only about 20% of the victims were killed by the original explosion. In addition the hazard of such fires is far smaller because the toxic gasses typically can vent themselves away from any bystanders. birthday cakes. Terrorist Arson and Terrorist Bombings Terrorist arson is analogous to. It is relatively difficult to “sabotage” the explosion resisting components of a building. Even large fires can mislead to the public. if the terrorist can take advantage of materials already in the building. with little smoke. Explosion are a well defined “events” which have been studied in isolation. While many people would describe a truck full of jet fuel as a hazard. construction money would be most prudently spent to properly design. Analyzing Arson as a Terrorist Weapon All conventional design fire safety design strategies implicitly assume the ignition of single “normal” fires. Similarly movie and television fires are staged special effects. rather than the trigger for a poorly understood fire development process. The public often confuses ignitability with flammability. Some fire protection systems can be easily sabotaged and some are vulnerable to misuse by the occupants. 1) 2) The “material” for a serious arson attack is often cheaper. reinforce. it will remain safe enough to facilitate the rescue of potential victims. The UK recognized the problem of progressive collapse and changed their building standards in the 1970s. few would classify a truck full of margarine as a similar hazard.
However if the goal is to just to reduce the risk of catastrophic life loss. The World Trade Center showed all three of these techniques. Regulating The Risk Of Terrorist Arson Assessing fire safety even in the absence of terrorism is a difficult task. Assessing Vulnerability to Terrorist Arson Buildings are inviting targets for terrorists. the problem becomes more tractable. The goal of regulation must be to develop techniques to harden the fire protection system against sabotage and properly evaluate the fuel load to raise the threshold and reduce the leverage available to a terrorist. Arsonists can also rearrange the building’s own fuel or air supply to enhance the burning rate. Fire protection systems are also routinely designed to deal with only one fire at a time. This requires careful analysis of the vulnerability of our building stock to terrorist arson. Casualties can increase because person near the fire may have no idea how fast a fire can develop and spread. the threat of multiple ignitions and over reliance on phased evacuation should be the subject of immediate analysis. controlling the available leverage. James Quintiere. The controversy over the Nuclear Power OSRE program shows some of the difficulties in regulating the private response to public attacks.viii In that case the costs of mitigating terrorist assault appeared to be so great that the private parties were unwilling to commit the resources needed. Many thousands of people died in the World Trade Center. But it is clear that the risk of terrorist arson may be greatest in buildings with many occupants. 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. A determined terrorist attack normally might involve: 1) 2) 3) a very large initial fire multiple ignitions Sabotage of the protective system. By setting multiple fires the arsonist can both increase the size of the ultimate fire and block exit paths for inhabitants. we can reduce the uncertainties involved in controlling terrorist arson. However it is most important to raise the threshold in the buildings with the highest leverage. Static or dynamic fire protection systems can be sabotaged. and developing new ones in critical areas. Anthony Kilpatrick Glasgow Caledonian University 53 . Naturally precise information on vulnerability to terrorist arson risk is extremely sensitive. since that reduces the potential for loss. but each plane was large enough to ignite several compartments. Marino Di Marzo and Fred Mowrer. There were not only two planes. However. University of Maryland Dr. there is value in raising the threshold of any building. As a result the phenomenon of terrorist arson must be carefully examined to allow appropriate risk analysis.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 5) Managing the dynamic fire protection features of a building may be much more complex than the passive explosion resistant features. They Include: Drs. open public access. Raising Thresholds and reducing leverage At the current time the key relationship is the leverage in the building at the building’s threshold. Like the relationship between flammability and ignitability. Building systems that are designed to deal with normal fires are not adequate to deal with terrorist arson. limited exit paths and a reliance on vulnerable dynamic fire protection systems. Research will be needed on the practicality of analyzing building thresholds. Conclusion Society cannot solve the problem of arson terrorism through analysis and regulation alone. The air supply was enhanced by the large impact hole . large quantities of flammable materials. 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. 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. applying techniques that are available. A rapidly growing fire can exceed the building’s design threshold before systems or inhabitants have a chance to react. University of Wisconsin and Mr. Vicki Bier. restricting the potential for sabotage. Perhaps most important.
Second Edition. 3-1 to 3-15 in The SFPE Handbook of Fire Protection Engineering. Babrauskas.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 References i ii iii iv v vi vii viii Brannigan V. 1997. and large areas. 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. July 1997 54 . MA.134-41 Designing Terrorist-Resistant Buildings TOD RITTENHOUSE “Fire Engineering U. NFPA. Burning Rates (Section 3/Chapter 1). Guide for smoke management systems in malls.. NFPA 92B. Quincy. MA: National Fire Protection Association." Inspection Procedure 81110. Quincy. National Fire Protection Association. V. Second Edition. Quincy MA (1995). National Fire Protection Association. (2000). "Operational Safeguards Response Evaluation (OSRE) Inspection Manual. Watts JM System Concepts for Building Fire Safety Fire Protection Handbook. National Fire Protection Association.S. Quincy MA (1995). pp. atria. NRC.
0131 240 5000 Fax.5million annually to Scotland’s top young academics to promote research in Scotland enabling leading Scottish-based researchers to collaborate with the best of their international counterparts Inspiring school children in classrooms from the Borders to the Northern Isles and promoting their interest in science. Producing academic journals of international standing Full details: www.royalsoced. economic and cultural wellbeing by: • • • • • • organising conferences and lectures for the specialist and for the general public on topics of national and international importance providing independent.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 APPENDIX A The Royal Society of Edinburgh (RSE) is Scotland's National Academy of Science & Letters. expert advice to key decision makers in Scotland awarding over £1. 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.uk Tel.org. Working as part of the UK and within a global context. 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 today’s Scotland. the RSE seeks to contribute to Scotland's social. society and culture. 0131 240 5024 The Royal Society of Edinburgh 22-26 George Street Edinburgh EH2 2PQ 55 .
Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 APPENDIX B The Royal Society of Edinburgh would like to acknowledge the work of the Conference Organising Committee Professor Dougal Drysdale FRSE Professor of Fire Safety Engineering University of Edinburgh Mr Christopher Mackay Partner Burness Mr Bill Smart Associate Connell Mott MacDonald Professor Ian Stevenson FRSE Programme Convener The Royal Society of Edinburgh Mr Paul Stollard The Scottish Executive Dr Jose Torero Reader in Fire Dynamics University of Edinburgh Dr Asif Usmani Senior Lecturer University of Edinburgh Ms Susan Walker Events Officer The Royal Society of Edinburgh The Royal Society of Edinburgh would also like to acknowledge the support of Corus and FM Global Research Finally. the Royal Society of Edinburgh would thank to thank the speakers who gave their time so generously 56 .
Obtained his PhD from Cranfield University. currently he is assisting in re-writing CIBSE Guide E. 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. 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). Farshad was promoted to Managing Director of the fire division of BRE. He is on the editorial board of the journals Fire Technology and Fire Safety Journal. Building Research Establishment (BRE) In 1997 Professor Farshad Alamdari joined the management team of BRE (the Building Research Establishment). just after its privatisation. FRS (formerly Fire Research Station) and recently to the BRE Chief Scientist. He is also a member of the AIAA Micro-gravity and Space Processes Technical Committee. 57 . Dr Jose Torero Reader in Fire Dynamics.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 APPENDIX C SPEAKERS Professor Farshad Alamdari PhD CEng FCIBSE Chief Scientist. Visiting Professor at the University of Ulster and Visiting Special Professor at Nottingham University. USA. the ASME Fire and Combustion (K-11) Committee. as a key business driver in the cultural change necessary to take a Government research organisation into a commercial researchbased consultancy company. the UL Foams Fire Suppression Systems committee and NASA’s Mars or Bust. Since then he has managed various research-based businesses involving environmental and fire safety issues. He is Co-Chairman of the Design Criteria and Loads Group on the Council on Tall Buildings and Urban Habitat. 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. Peter has acted as an expert witness in fire safety matters where legal proceedings or agreements have been required. Fire Engineering Handbook. Torero obtained his PhD from the University of California Berkeley in 1992. Leader Arup Fire International. He chaired the Arup Extreme Events Task Force set up after 11th September. UK and a Researcher (en Disponibilité) at the French National Center for Scientific Research (CNRS). Dr Peter Bressington Director. Farshad is a Chartered Engineer and is a Fellow Member of the CIBSE. University of Edinburgh STRUCTURES IN FIRE: AN OVERVIEW OF THE BOUNDARY CONDITION José L. In 2001. He is currently Reader in Fire Dynamics at The University of Edinburgh.
Recently he has been involved in the study of fire-induced skin burns. 58 .Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 He specializes in fire safety engineering and his work is primarily in the general areas of fire dynamics. the William M. Susan is actively involved in encouraging research into structures in fire. Susan is a member of the structures in fire group and specialises in structural fire design. more than 50 Journal Publications and more than 100 other technical documents. fire protection and suppression systems. 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 completed her PhD at the University of Edinburgh in October 2001. Torero has been the recipient of numerous research and teaching awards that include the E. Bigglestone Award for the Best Paper Published in Fire Technology in 2002. Her PhD title was “The Behaviour of multi-storey composite steel framed structures in response to compartment fires. Susan worked extensively with the research team at Edinburgh University modelling the Cardington frame fire tests. Dr Susan Lamont Fire Engineer.” This was a computing-based PhD analysing the influence of different fire scenarios on generic composite steel frame multi-storey structures. Robert Kent Outstanding Teaching Award (1998). but is responsible for many aspects of life safety design when producing a fire strategy for a building. smoke detection & management. Arup Fire sponsor a number of PhD students in the UK. Dr. He is the author or co-author of 3 book chapters. Carey Award for the Best Paper Presented at the Fire Suppression and Detection Research Application Symposium (2001) and the Harry C. fire resistance design based on quantified structural behaviour. waste incineration and the behaviour of structures in the event of a fire.
59 . Professor Vince Brannigan Lecturer in Building Safety and the Law. Department of Fire Protection Engineering. Building Research Establishment (Retired) and Visiting Professor Cranfield University FIRE MODELLING Geoff Cox was. He has been very active in fire safety codes research on a world wide basis. 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. For the last eight years he has worked in the field of fire investigation. He pioneered the development of fire modelling using computational fluid dynamics and is author of over 100 scientific papers and 3 books. University of Maryland and Visiting Professor in Fire Risk Engineering at Glasgow Caledonian University THE WORLD TRADE CENTER: TERRORIST ARSON AND THE LAW Professor Vincent Brannigan teaches Building Safety and the Law in the Department of Fire Protection Engineering at the University of Maryland. Fire Engineering. Other activities include committee work for BSI. He joined FRS in 1973 and has undertaken research on many aspects of fire dynamics. His BS is in the History of Technology from The University of Maryland. CACFOA together with lecturing and a keen personal interest in sailing. fire research and fire engineering and currently heads up a team of six dedicated fire engineers providing consultancy services to the Brigade’s fire safety teams. and His JD is from Georgetown University. training. He is currently a Visiting Professor at Cranfield University and continues to contribute to fire safety science as an advisor to FRS and through his links with various universities. His team are involved in all major building and civil projects across London together with research activities at major fires. Research Director of the UK Fire Research Station (FRS) and Chairman of the International Standards Technical Committee on Fire Safety.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 Professor Geoff Cox Fire Research Station . fire safety and risk management. Mr Jim Golt Group Commander. He is also Visiting Professor in Fire Risk Engineering at Glasgow Caledonian University. He is a member the bar and a regular lecturer at the US fire academy in Emmitsburg MD. until his retirement last year.
Strathclyde Fire Brigade Group Commander. Scottish Executive 60 . University of Edinburgh University of Edinburgh Station Officer. Leader Arup Fire International. Department of Civil Engineering. Scottish Executive Station Officer. Strathclyde Fire Brigade Corus Construction Centre Group Manager. University of Maryland and Visiting Professor in Fire Risk Engineering at Glasgow Caledonian University Director. Ove Arup Partners Ltd Heriot-Watt University Mr P Bressington Mr R Carvel Mr B Chisholm Professor G Cox Dr C Davie Miss C Dierichs Mr J Donald Mr J Dowling Professor D D Drysdale FRSE Mr G Flint 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 Mr D Kee Professor C Kuo FRSE Dr S Lamont Mr C Mackay Mr J Martin Mr J Milligan Fire Research Station.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 APPENDIX D PARTICIPANTS LIST Miss K Anderson Dr A Beard Professor N Bicanic Professor V Brannigan University of Edinburgh Heriot-Watt University Head. BRE (Retired). London Fire Brigade Assistant Fire Inspector. Office of the Deputy Prime Minister Building Control Officer. Fife Council Estates Manager. Fife Council Assistant Fire Inspector (Government Premises). University of Strathclyde Fire Engineer. Burness Principal Building Control Officer. University of Glasgow Lecturer in Building Safety and the Law. National Library of Scotland University of Edinburgh University of Edinburgh Professor of Marine Technology. Ove Arup and Partners Ltd Partner. Visiting Professor Cranfield University University of Glasgow University of Edinburgh Health and Safety Executive Construction Development Manager. Corus Professor of Fire Safety Engineering.
Chairmen and members of the organising committee are denoted by Italics 61 . University of Edinburgh Senior Lecturer. The Highland Council Programme Convener. Building and Fire Research Laboratory. Fife Council Market Development Manager. University of Edinburgh Building Research Establishment Building Control Officer. Fife Council Speakers. Connell Mott MacDonald Divisional Officer. Professor of Pharmacology. University of Dundee Stirling Council The Scottish Executive Acting Deputy Director. National Library of Scotland Station Officer. The Royal Society of Edinburgh. Strathclyde Fire Brigade Scottish Universities Environmental Research Centre Heriot-Watt University University of Edinburgh Associate. Corus University of Edinburgh Deputy Estates Manager. Strathclyde Fire Brigade Senior Building Control Officer.Sunder Dr J Torero Dr A Usmani Dr S Welch Mr S Young University of Edinburgh Building Control Officer.Fire and Structures Conference The Royal Society of Edinburgh 21/04/04 Miss K Murphy Mr G Nicoll Mr A Orton Mr A Phillips Mr J Plumb Mr W Russell Dr David Sanderson Mr Jaime Santos-Reyes Miss M Serpilli Mr B Smart Mr D Smith Ms S Smith Professor I H Stevenson FRSE Mr C Stokes Mr P Stollard Dr S Shyam. National Institute of Standards and Technology Reader in Fire Dynamics.
This action might not be possible to undo. Are you sure you want to continue?
We've moved you to where you read on your other device.
Get the full title to continue reading from where you left off, or restart the preview.