Towards resilience evaluation of buildings when exposed to fire based on English and USA fire statistics

Abstract

This research arises from the necessity to fill the lack of data present in the international fire safety community related to pre and post-fire conditions of real buildings subjected to real fires and create probabilistic risk assessments able to quantify structural damage according to possible mitigation factors and related financial losses.

Resilience is defined as the capacity of a building to absorb, mitigate and recover from a fire incident. It includes life safety, property protection and continuity, however, continuity plans and business impact analyses are rarely developed in common design practices, unless in exceptional cases. Moving from prescriptive to performance-based design appears as a natural evolution in understanding the response of buildings and proof that a selection of design drivers could lead to the same or greater level of safety than the one indicated following prescriptive methods. Fire resilience codes are investigated to understand how this problem is addressed and to evaluate the aspects considered in terms of prevention, response, absorption, mitigation and adaption. Understanding fire safety processes, flow charts are developed to create a holistic view and optimize choices of fire safety and structural engineers.

Fire statistics are composed of data collected by fire departments in the aftermath of a fire event. In the cases of England and the USA, this is done by filling in an online form per event that it is consequently submitted into national databases. Annual reports are published to summarise the trends and allow further research. Fire statistics from the Incident Reporting System in England and the National Fire Incident Reporting System in the USA are analysed to understand causes of fire, such as item first ignited, fire location and building descriptions, as well as consequences, such as response time from fire brigades, the influence of automatic extinguishing systems on fire spread and damage. Once fire damage and total damage are quantified, financial losses can be evaluated considering the Building Valuation Data formula which multiplies damage and unit cost per ft2 according to different building construction and occupancy types. Comparisons between the two countries allow comments and reflections about the different safety measures, preparedness and similar trends where possible. The idea is to address complex problems affecting the fire safety community, to understand fundamental relationships between fire incidents and structural response to fire, to quantify structural damage scale and to estimate direct costs. The study considers also the improvement of English fire statistics to include fire safety fields not currently covered.

An international database based on the data available in the fire statistics from England and the USA is developed as guidelines able to support fire safety community. British Standards PD 7974-7:2003 (referred as PD 7974-7), is related to the probabilistic risk assessment. Its data is based on fire incidents from 1966 to 1987 need to be compared to current statistics to understand if the fields described are still representative of fires in buildings. The comparison appears necessary to understand potential improvements from the past and possible optimization for the future. Moreover, the use of fire statistics is representative of current fire safety fields and is adopted to create contemporary comparisons.

Fire frequency is obtained considering also building stock and the number of properties for each building type. Furthermore, fire frequency is plotted to building floor space according to a power law with positive, as suggested by Rutstein and adopted in the PD 7974-7, or a negative exponent or to a polynomial function of second or third order which presents the highest R2 and seems to better approximate the sample. The fire spread and damage are analysed considering the presence or absence of automatic extinguishing systems. In the USA, the trends for the fire damage present higher percentages than those of PD 7974-7 while English fire damage data are consistent with those of PD 7974-7. In English statistics, response time assumes a relevant influence on fire spread and damage especially in the first minutes after the fire notification.

The research continues evaluating probabilistic risk assessments in buildings that are based on the concept of risks, hazards and consequences and how to reduce them.

They are derived considering English fire statistics in event tree analysis to evaluate the likelihood of fire scenarios with the presence or absence of safety systems and their effects in terms of human response and structural damage. Deductive procedures are used in event tree analysis, to highlight how potential events could mitigate outcomes related to an initiating event represented by a fire in specific buildings. These methods have the aim to better understand possible fire incidents and identify realistic fire scenarios to be implemented in fire and evacuation models. Probabilistic techniques evaluate key parameters to assess factors contributing to ignition, growth and effects of fire in terms of consequences and likelihood that a given set of consequences occurs.

The impact of the research results could create fire safety tools able to automatically process the fire information collected over a time period to ensure updated data and controlled trends which will be adopted in fire risk assessment and prevention methods to absorb and recover during and after a fire incident. Furthermore, methods to identify indirect losses will be assessed. Future research could involve the investigations on how these data could inform the design of practising engineers and on how fire statistics influence the code guidelines.

Moreover, it would be important to understand how governments adopt this information for decisions about resources allocation and how the fire safety approaches change considering different authorities and locations.