Fiv Toskealeg Marabutsnd is Tle Ukied Satas CrooMark sapc:(0oi-ag18.3067/s30694-0064773-2 ACriticalEvaluationofBSPD7974-7 Structural FireResponseData Based on UsA Fire Statistics M. Manes* and D. Rush* . School of Engineering. University of Edinburgh. Edinburgh UK Received: 16 March 2018/Accepted: 11 September 2018 Abstract. Probabilistic techniques deal with the randomness of variables and relia- bility of safety system but their application in fire safety engineering is limited due to the lack of data related to real structures subjected to real fires. This can be over- e by analysis of national fire statistics provided by fire departments. Fire statistics databases are a collection of data from real structures subjected to real fires and pro vides an understanding of real effectiveness of different fire safety measures (i.e. - partmentation) which influence the spread and growth of fire and ultimately their monetary consequence. The ability to understand the realistic responses of buildings in fire is the fundamental basis of British Standards PD 7974-7 which provides data to perform probabilistic risk assessments for fire. However the current data pre- sented by BS PD 7974-7:2003 (referred to as PD 7974-7 within this paper) was devel- oped between 1966 and 1987. This research has used the USA fire statistic database of 2014 to recreate the tables present in the PD 7974-7 pare the results and understand their evolution in time. The parison between PD 7974-7 and the USA fire statistics introduced in this paper shows that modern fire frequency can be up to more than 10 times smaller than presented in PD 7974-7; area damage in m? and spread of fire are linked to automatic extinguish systems effectiveness and greater in the USA fire statistics than predicted by PD 7974-7 This clearly demonstrate the need of updates to PD 7974-7 and feeds towards a better understanding of the robustness and potentially the resilience of real structures in fire. Keywords: Probabilistic risk assessment Fire statisties Fire frequency. Fire damage Fire safety sys- lems Fire financial loss Performance bused-design 1.Introduction Fire design codes and regulations are transitioning from a prescriptive basis to more of a performance-based design paradigm. This is due to perceived lack of regular updates within the codes limiting the creation of innovative and diverse design solutions [1]. The fundamentals of performance-based design (PBD) are to ‘enable improvement of structural fire safety in fire to increase the design flexibil- ity and reduce the costs of fire protection to structures’ [2]. One of the methods * Correspondence should be addressed to: M. Manes E-mail: m.mansiged.ac.uk; D. Rush E-mail: d.rush(@:ed.ac.uk Published online: 12 October 2018
Fire Technology 2018 available to be used by engineers is to adopt a probabilistic assessment of the fire and its consequences on the structure and within the UK there are codes (PD 7974 parts 7 and 8) that provide data and methods to assess the risks of fire and their impact [3 4]. However like their prescriptive counterparts data within the probabilistic codes (i.e. [3]) are often decades old and as such they add an extra level of uncertainty about their validity to be used in the current context. Governments regulations aim to ensure life safety (i.e. means of warning and escape in the Approved Document B for dwellings [5] and buildings other than dwelling houses [6] in England) and previsions of prescriptive codes (ie. [7]) assume that the intended life safety fire performance criteria are implicitly satisfied by meeting the minimum code requirements for design and detailing of structural and non-structural ponents [8]. In the USA life safety goals are covered by the fact that every new and existing building shall ply with the NFPA 1: Fire Code and NFPA 101 [9 10]. These life safety goals have been shown to be more and more well satisfied since the early 2000s; the number of primary fires (i.e. in structures rather than outdoor or vehicular) in the UK has decreased from a peak of 178 000 in 2003/2004 to 75 000 in 2016/2017 with fire related deaths in the UK reduced from around 454 in 2003/2004 to 261 in 2016/2017 [11]. A similar picture also emerges from the USA where in 2003 520 000 structure fires caused 3385 deaths pared to 475 500 structure fires with 2950 deaths in 2016 [12]. However whilst life safety goals are improving the economic cost of fire remains high with cconomic losses due to fire in the USA in 2011 being at an estimated S14.9 Bn (S13.3 Bn direct S1.6 Bn indirect losses such as business interruption) a value that has not varied greatly over the past 30 years [13]. Risk and consequence of fires have received a lot of academic inquiry with the majority interrogating the probability of failure in terms of structural collapse and/or life safety [14] however very few explicitly consider damage and thus the potential reinstatement and re-use of the property [15]. Property protection is an essential part of business continuity and forms a major part of the resilience of the business [4] where resilience is the ability of a system to absorb and recover from disruptions (fires in this context) to normal functionality. Whilst there is no unique definition yet of fire resilience there are three main principles that sit beneath it: life safety property protection and business continuity [16 17]. As in any built environment discipline there are natural variabilities and uncer- tainties which are dealt through design codes for instance Eurocode 1991-1-2 [18] provides details on the actions on structures with respect to fire effects describing how (if permitted by national authorities as is the case in the UK) the inclusion of different fire safety measures the fuel load and thus the effective worst case severity of a fire can be reduced. However due to the plexity of fires with multiple interactions occurring between different materials ventilations safety sys- tems and/or structural systems uncertainties will emerge in all aspects of reported national and international fire statistics. The concept of safety itself is one of the uncertainties there is no such thing as absolute safety [19]. It is in this uncertain context that probabilistic techniques which deal with the randomness of variables and reliability of safety system have bee available [3] however their application is limited due to the lack of contemporary data
A Critical Evaiuation of BS PD 7974-7 Structural Fire Response Data (data in BS PD 7974-7:2003 from 1966 to 1987) related to real structures sub- jected to real fires. Previsions of prescriptive codes [7] assume that the intended life safety fire performance criteria are implicitly satisfied by meeting the minimum ponents. This represents an expectation and not necessarily the reality since the plexity of actual fires and the interaction with structures is difficult to predict accurately. Structural fire engineering knowledge for instance is predominantly based on the response of single elements to a standard time temperature curve under idealised furnace conditions and assessing the response to a binary pass/- fail criterion [20]. However these idealised furnace conditions are not real fires which may or may not cause a failure but will produce some level of damage. Pre- vious experiments conducted on full structural frames and sub-frames [21] have provided valuable data for modelling and design code enhancements nevertheless due to their expense they are limited in numbers so the relevance of any data is confined to the specific experimental building characteristics. Therefore fire statis- tics when coupled with appropriate and valid modelling scenarios could provide the necessary means to assess and design real structures in real fires for different performance criteria. Performance-based design is used to explicitly demonstrate using pre-identified performance objectives that the same level of life safety can be provided with enhanced property protection. However models used for these demonstrations are rarely validated or benchmarked due to a lack of relevant experimental data and can also be limited in their ability to describe structural behaviour close to the point of failure. Probabilistic performance-based design [22] attempts to overe this however again lack of validation data for these models limits their applica- tion in practice. For this reason statistical data on existing structures in the after- math of an event can be instrumental in defining the real impact effects on structures and the quantification of damage. Therefore fire statistics can be used to understand the real relation between fire occurrences fire size damage caused by fire and interventions to extinguish the fire. This paper presents an assessment and parison of the current PD 7974-7 Probabilistic Risk Assessment for fire with fire statistics from the USA and where able with statistics from the UK. Tables within PD 7974-7 [3] document present frequency of fire and fire damage classifications according to sprinklered and un- sprinklered buildings. However at the start of analysis procedure data regarding sprinklers were not available for the UK while these were prehensively and readily available for fires within the USA and thus the focus of this paper. New UK fire statistics released in September 2017 [23] will allow further parisons between PD 7974-7 data and the UK fire statistics. This paper aims to recreate the tables (specifically Tables A.1 A.2 A.4 A.5 A.6 A.7 A.8 and A.12) presented in the PD 7974-7 using contemporary fire statistics from the USA and pare the results allowing engineers to under- stand pre-fire conditions fire growth and post-fire conditions according to data obtained on real structures subjected to real fires and reduce the uncertainties that inevitably arise in fire safety design.
Fire Technology 2018 2.HistoricalDevelopmentofPD7974-7 In 1907 von Sawitsh expressed that fire insurance claims frequency had a linear relationship with volume or value at risk. In 1937 Berge demonstrated on the basis of Swedish dwellings how the claims frequency and the fire loss ratio incrcase with the size of the house. D'Addario [24] followed in 1940 by expressing claims frequency f(s) of the Swedish statistics explained by Berge as a function of the size (s sum insured) as: f(s) = As* (1) where A and α are empirically fitted coefficients. This power law model is cur- rently in use in PD 7974-7. In 1968 Ramachandran bined national fire statistics developed by local authority fire brigades with financial loss data from British Insurance Associa- tion. Ramachandran estimated for different occupancy types the total number of fires and total cost in thousands of pounds [25] for large fires (defined as requiring five or more jets). In 1969 Ramachandran assessed large fires from 1965 to 1968 again considering the total number of fires and total cost in thousands of pounds for different occupancy types as well as fire frequency place of origin source of ignition material first ignited age of the building number of storeys spread of fire attendance time and fire protection devices [26]. In 1970 Ramachandran pro- duced fire loss indexes defined as loss per ft² of floor area or loss per hundred pounds of value at risk [27]. In 1979 Rutstein [28] affirmed that fire risk (probability of fire and its conse- quence) can only be expressed in probabilistic terms and can be estimated by examining past fire incidence data. Furthermore Rutstein determined the proba- bility of fire by paring the number of fires reported by the fire brigades divi- ded by the total amount of property at risk determined from 1977 Home Office survey data of UK manufacturing industry. The fire probability F is described with a power law according to the total area of the building A: F = a.Ab (2) where α and b are empirical constants determined according to the different occu- pancy types. In particular α defines the ratio between total number of fires and total number of buildings at risk while b defines the total number of fires divided by the building maximum floor area. The analysis interrogated only industrial buildings leaving a wide margin of uncertainties for the other occupancy types [28]. The 1977 Home Ofice survey enumerated the number and floor space of each building occupied by firms. Infor- mation relating to 6000 separate buildings was then used to estimate the total number of buildings of each size in each industry in UK. Consequently plotting the data according to building floor space in m^ (x-axis) and probability of fire (y- axis) a nonlinear function was estimated (Fig. la) where the non-linearity could be attributed to two potential reasons. The first that there may be a geruine scale
A Critical Evaination of BS PD 7974-7 Structuraf Fire Response Data 0.04 139 04 tt eT (a) (b) Figure 1.(a)Probability of fire occurring (Production building manufacturing industry) and (b) The estimated probability of fire in different industries[28]. effecr: if a building is enlarged to double its original size not all the services would double in size or in fire risk so that the risk of fire in a larger building would be less than twice than that of the original building: the second the po- management or first aid-firefighting capability. They may have lower fire risk than the small ones. Initially Rutstein estimated the probability of a fire considering all buildings within each industry together where most of them were used for production and a smaller number for storage office and other buildings. In a second instance the probability of fire was calculated for those buildings primarily used for production as showed in Fig. 1b. Rutstein's functions are always power laws with positive exponents therefore the probability of fire increases non-linearly with the increase of the dimension of the buildings. Coefficients a and b according to different industry types but solely for production buildings are those reported in Table A.1 of the PD 7974-7 [3] pletely neglecting the coefficients for the class of all buildings. The studies of D'Addario Ramachandran and Rutstein have converged into PD 7974-7 which sets out the general principles and techniques of risk analysis that can be used in fire safety outlines the circumstances where this approach is appropriate and gives example illustrating their use. It also includes data for prob- abilistic risk assessment and criteria for assessment that cover life safety and prop- erty protection in both absolute and parative terms (Absohrte terms: reliability analysis between two or more peting solutions) [3]. performance of one situation against predetermined criteria; Compararive terms: Despite being an essential guide for engineers to express a reasoned judgement based on local conditions data within PD 7974-7 range from 1966 to 1987 (except
基于美国火灾统计数据对 BS PD 7974-7 结构火灾响应数据进行严格评估.pdf
