Practical problems of applying performance-based structural fire engineering in building design Piotr Smardz INBEPO Sp. z o.o. ul. Domeyki 9a, 53-209 Wrocław, Poland INTRODUCTION The introduction of structural Eurocodes, which form a comprehensive set of design codes covering various aspects of structural design including fire safety has made it easier for engineers in Europe to apply performance-based approach to this particular element of building design. The relevant parts of the Eurocodes cover the topic of thermal actions on structures exposed to fire as well as the mechanical response of the structure, as appropriate for the structural material in question. In many countries in Europe the design of building structures for fire resistance was until recently largely based on a prescriptive approach. Therefore the introduction of much more advanced and complex design methods requires not only changes in the regulations, but also a proper education process for both the designers as well as the approving authorities. The engineering literature available so far is usually limited to relatively simple examples, which are not applicable to reallife situations. Structural fire design often relies on complex design tools such as CFD models for assessing thermal conditions within the fire compartment and/or FEM models for heat transfer and mechanical response analysis. This makes it quite difficult for the approving authorities to verify the validity of the design approach adopted. One of the practical problems facing structural and fire safety engineers who are trying to implement a performance-based approach to structural fire design is the lack of commonly accepted design fires to be used as basis for the natural fire model. In this respect the Eurocodes provide only limited guidance. On the other hand the design fires which can be found in the literature and design standards relevant to smoke control design are not necessarily appropriate for structural fire engineering. Even for such a relatively standardized fire load array as an open car park there is no uniformly accepted design fire curve. The problem is further exacerbated by quick pace of change in the forms and materials used in building design, as well as in building content which may constitute the fire load think for example of the many changes that have occurred in the design of cars and other vehicles over the last twenty years. Unfortunately there is not enough publicly available fire test data which would allow an informed decision by the designer as regards selecting a reasonable design fire growth curve.
A similar situation exists as regards thermal properties for certain fire protection systems such as intumescent paints, for which there are no widely available data on proxy thermo-physical properties of the protective layer. The discussion in this paper is based predominantly on the experiences of the author on projects carried out in Poland, with some limited references to other European countries. However, it is believed that similar problems are experienced by fire safety engineers in other EU countries and indeed elsewhere in the world. DESIGN FIRES In order to carry out an analysis of thermal actions on structural elements exposed to fire, it is first necessary to select a design fire. The design fire defines the rate of heat release of the fire as a function of time. This is necessary as an input parameter for the advanced fire models (e.g. twozone or CFD models) which are used to predict temperature and thermal radiation levels in the vicinity of the structural element being examined. The design fire can either be described by a simple formulae (e.g. an t 2 fire) or by a set of discrete values of HRR vs. time, often presented graphically. In order to accurately describe the fire in the fire analysis, it is also necessary to describe the area involved, thus an assumption must be made as to the heat release rate of fire per unit area. The relevant part of Eurocode 1 [1] provides some guidance on fire growth rates only for a limited number of occupancy types including dwellings, hospitals, hotels, offices, school classrooms and shopping centres. A slightly more extensive set of values can be found in some National Annexes to EN-1991-2 (e.g. in the Irish NA to EN 1991-1-2). However, no data is provided for commonly encountered occupancies such as restaurants, sports halls, bus and train stations (platform areas) or car parks. In the absence of direct guidance from EN-1991-1-2 it is necessary to refer to other available guidance documents and technical papers. Data describing design fires for various occupancies can be found for example in technical literature and standards relating to the design of smoke control systems. However such data was often developed with consideration only for the initial stages of the fire (pre-flashover) when the occupants of the building evacuate, i.e. covering the first 20-30 minutes of the fire. The question arises if such design fires can be safely used as input data for structural fire engineering purposes. The problem with design parameters which are not explicitly stated in one of the recognized national standards (such as in case of Poland - the PN EN standards) is that they are often held in less esteem by the approving authority. The design parameters sourced from European Norms (EN) which often also have the status of Polish Norms (PN) are perceived as official values and are therefore least questioned. This is clearly very superficial approach, as many of the coefficients and design parameters specified in the Eurocodes are a matter of fierce debate between the scientist and professionals involved in the field.
Open car parks Let us consider as an example the issue of design fires for multi-storey open-sided car parks. The required period of fire resistance for the elements of structure comprising such car parks varies between different European countries, from no specified fire resistance to 90 minutes FR. In many cases of such open car parks more economical design solutions can be achieved if the fire resistance of the structure is assessed with reference to the actual predicted thermal conditions in case of fire rather than the standard (ISO) curve which is representative of a fully developed fire in an enclosed compartment, with large amount of fuel present. For open sided car parks most available European publications refer to the design fires developed on the basis of real scale fire tests carried out in France in 1995. The results of those tests are publicly available in technical steel research reports published by the European Commission [2,3]. A design fire curve for a single medium-sized car (referred to as class 3) is presented in figure 1. Fig. 1 HRR curve of a single burning class 3 car [2] In a real car park configuration with many vehicles parked in a row it is assumed that fire spread to subsequent cars occurs at 12 minute intervals. The HRR curve for the subsequent cars is slightly different, due to the pre-heating of those cars. The peak HRR value is the same at 8,3 MW however the period of initial burning is shorter and more intense (see figure 2). For the purpose of design calculations it is assumed that the fire spreads to involve three, four or even five vehicles, depending on the configuration of the parked cars in relation to the structural element being analysed either a beam or a column. In this respect there is no uniformly-accepted set of scenarios which should be considered by the designer.
Fig. 2 HRR curves of three burning class 3 cars [2] Figure 3 shows a cumulative HRR curve for a design fire involving three cars, which was used by the author as input for the CFD analysis for four open multi-storey steel car parks recently designed and built in Poland. An example of one such car park is presented in picture 1. Due to the period of fire resistance required under Polish regulations for such a car park, the analysis covered the time period of 30 minutes. For this reason it was decided that a design fire involving three cars is sufficiently onerous, as during the time period considered in the analysis the burning of two cars has already reached its peak. Fig. 3 Cumulative HRR curve for a fire involving three class 3 cars
Pic. 1 Example of an open-sided car park in Poland designed on the basis of natural (localised) fire It must be stressed that in the absence of design data relating to open car parks in EN 1991-1-2 it is advisable to discuss the design fire adopted for the analysis with the authority which is ultimately approving the design (e.g. the fire brigade) prior to carrying out the calculations. One issue commonly raised in the discussions concerning design fires for car parks is whether design values based on fire tests conducted over 20 years ago are applicable to modern (or indeed future) cars which tend to have more combustible material in their construction. Furthermore, it is worth noting that in the recent proposals for the upgrade programme of the Eurocodes [4] the lack of up-to-date design data (i.e. a recommended design fire curve) for opensided car parks was identified as one of the obvious shortcoming of the current version of EN 1991-1-2. Other occupancies Another type of occupancy for which selecting a design fire for structural fire resistance calculations may prove problematic is a shopping centre with many individual retail units and a large common mall. Such premises are normally protected with a sprinkler system, so a fire within a retail unit will more than likely be a localized phenomenon and should not lead to a fully developed fire (post flash-over). A fire in a small kiosk or retail stand within the mall will also be a localized fire, even in the case of a mall / atrium with large floor-to-ceiling height where the sprinklers are unlikely to activate quickly or indeed where they may be omitted in the design. In the latter case the fire size (and hence its HRR) will be governed by the physical size of the stand, which will normally be quite small relative to the overall volume of the mall. Removal of heat will be further
enhanced by a Smoke and Heat Exhaust Ventilation System (SHEVS), which is normally provided in a shopping mall. In the circumstances described above the analysis of thermal actions on structural elements (either for the main loadbearing elements or the roof) will in most cases focus on the effects of a localized fire either using as simplified model described in annex C of EN 1991-1-2 or using an advanced CFD model. Annex E of 1991-1-2 provides some guidance as to the characteristics of the fire growth rate and the rate of heat release per unit floor area (RHR f ) for retail premises. The suggested value of RHR f =250 kw/m 2 for shopping centres seems to be relatively low, which would be conservative for smoke control calculations (i.e. it would result in a larger fire perimeter for a given heat output). However, it may not necessarily be conservative for structural fire resistance calculations, as it would result in lower flame heights and hence lower temperatures above the fire. In the background paper to the UK national Annex to EN 1991-1-2 [5] the rate of heat release rate per unit area suggested for a more general occupancy category shops is 550 kw/m 2. The same value is given for shops in the Irish National Annex to EN 1991-1-2. Pic. 2 Results of CFD analysis carried out to estimate temperature in the vicinity of roof elements in a railway station The shortage of design fire data in EN 1991-1-2 is not limited to car parks or retail premises. Other examples encountered by the author include transportation hubs (i.e. bus terminals, railway stations) as well as industrial sites (e.g. a large roof structure over bulk storage of coal). INFLUCENCE OF VENTILATION CONDITIONS In any analysis of thermal actions on structural elements which is based either on simplified or advanced model, one of the most important input parameters is the amount of available ventilation. Breakage of glass in the windows or glazed facade of the building will on the one hand allow inflow of fresh air to sustain combustion, but on the other if sufficient amount of openings is created will allow effective dissipation of heat and hence lower temperatures within the fire compartment.
For calculations based on parametric temperature-time curves (annex A of EN 1991-1-2) or estimations based on the equivalent time of fire exposure (annex F of EN 1991-1-2) the code sets minimum and maximum limits for the amount of openings in the façade which can be taken into account, relative to the floor area of the compartment. However, it does not provide any guidance on the percentage of glazing that can be assumed to break and fall out during a fire. The same problem exists where the analysis is carried out using an advanced model such as a two-zone model or a CDF model. With the current high requirements for energy conservation it is typical that double and triple glazing systems are used in commercial buildings. An assumption that 100% of such glazing will be destroyed during a fire is clearly unrealistic, although EN 1991-1-2 seems to suggest that this is the case, as it simply refers to the area of windows. Assuming unrealistically high areas of ventilation openings during a fire can in most cases lead to un-safe results - either in terms of the temperature curve or the estimated equivalent time of fire exposure. Some guidance on maximum percentages of glazing which can be realistically assumed to fail under high temperature caused by a fire can be found in design recommendations and standards published by member-states of the EU. A guidance document published in 2012 in Luxemburg [6] generally relates the amount of failed glazing with the temperature reached within the compartment. It suggests two scenarios to consider: Scenario 1 is for 90% of the glazing area to be open since the beginning of the fire Scenario 2 sets the following percentages of failed glazing for various glazing types: Simple glazing: 50% open at 100 o C and 90% open at 250 o C Double glazing: 50% open at 200 o C and 90% open at 400 o C Triple glazing: 50% open at 300 o C and 90% open at 500 o C Another document with recommendations on maximum areas of glazing which can be taken into account as ventilation openings for heat in a fire situation is German standard DIN 18230-1 [7]. This document relates the maximum percentage of glazing which can be assumed to fail to the expected duration of the fire. For single glazing up to 80% can be assumed to fail during the first 15 minutes and up to 100% if the time exceeds 30 minutes. For double glazing the values are 35%, 50% and 100% for time intervals of 15 min, 30 min and above 30 min respectively. SUMMARY The Eurocodes are a comprehensive set of standards for structural design. They provide a far more detailed guidance as regards the behaviour of structures under fire conditions, than was previously provided in the National Standards of the individual member states (except maybe for those with long tradition in applying performance-based fire safety solutions such as the UK).
However - as was discussed in this paper - a designer undertaking fire engineering analysis with the scope of establishing thermal actions on structural elements may find that the Eurocodes do not yet provide a complete guidance on the various aspects of analysis which needs to be carried out. Based on the limited information available in the public domain it can be assumed that work has just commenced on a comprehensive update program of the European structural design codes, with the view of completing the process by 2020. The aim is to simplify the Eurocodes so as to enhance their use, reduce the number of Nationally Determined Parameters (NDPs) and also to provide new guidance and design methods where necessary. All the above goals are to be applauded. It is not known however if the second generation of Eurocodes will include new information on issues discussed earlier in this paper i.e. design fires or destruction of window glazing at high temperatures (in the context of its influence on ventilation conditions within fire compartment). REFERENCES 1. EN 1991-1-2:2002 Eurocode 1 Actions on structures. Actions on structures exposed to fire 2. Joyeux D., Kruppa J., Cajot L., Schleich J., de Leur P., Twilt L., Demonstration of real fire tests in car parks and high buildings, European Commission, Final report, 2002 3. Schleich, J.B., Cajot, L.G., Franssen, J.M., Kruppa, J., Joyeux, D., Twilt, L., Van Oerle, J., Aurtenetxe, G. Development of design rules for steel structures subjected to natural fires in closed car parks, CEC Agreement 7210-SA/211/318/518/620/933, Final report 1999 4. Wald F., Burgess I., Outinen J., Vila Real P., Horova K., Fire Eurocodes The Future?, COST Action TU 0904, CTU Publishing House, 2014 5. PD 6688-1-2:2007 Background paper to the UK National Annex to BS EN 1991-1-2, BSI 6. ITM-SST 1551.1 Etude de stabilité au feu à l aide d une approche performancielle, 2012 7. DIN 18230-1:2010-09 Baulicher Brandschutz im Industriebau - Teil 1: Rechnerisch erforderliche Feuerwiderstandsdauer (Structural fire protection in industrial buildings - Part 1: Analytically required fire resistance time)