Fire statistics
Fire statistics Ignition frequencies are used for calculating the frequency of the initial event. Fire service response times are used in risk analysis. Reliability of active systems. Fire losses in terms of deaths, injuries, direct economic losses. Insurance companies maintain statistics of other economical losses as well. References Incident database PRONTO (www.prontonet.fi) Tillander, Oksanen & Kokki 2009. Paloriskin arvioinnin tilastopohjaiset tiedot. VTT Tiedotteita 2479 Tillander, K. 2004. Utilization of statistics to assess fire risks in buildings. VTT Publications 537. (PhD thesis) Finnish rescue service statistics at pelastustoimi.fi U.S. Fire statistics at nfpa.org. U.K. Fire statistics at gov.uk. CTIF: World fire statistics http://www.ctif.org/ctif/world-fire-statistics FEU http://www.f-e-u.org/statistics.php 9/12/2017 26
Ignition frequencies Two characteristic measures: Ignitions per floor area and year (1/m 2 a) Applies to buildings where the amount of activity leading to ignition is proportional to the area, e.g. dwellings and offices, i.e. human activity. Ignitions per operational unit per year (1/a) Applies to facilities where the activity is directly related to certain process of machine, not the floor area around it. E.g. pumps of the power plant. Depends e.g. on occupancy type, building size, etc. 27
Ignition frequency Ignition frequency in Finnish buildings (VTT Tiedotteita 2479) 28
Ignition frequency Ignition frequency Ignition frequency Effect of the building size on ignitions Industrial Row houses Multi-storey apartm. Single-family houses Floor area [m 2 ] Storage Floor area [m 2 ] (VTT Tiedotteita 2479) Floor area [m 2 ] 29
Fire service response times What does the response mean? Accident happens Call Alarm Vehicle starts Arrival on site Total response time (Help delivery) Rescue operations start efficiently UK Incident Reporting System: A response time measures the minutes and part minutes taken from time of call to time of arrival at scene of the first vehicle. Call time Alarm time Start time Driving time Scout and preparation time Dispatching time Rescue service response time Having the vehicle on site does NOT help or put out fires! Accident detected Call received by emergency response centre Rescue units alarmed Rescue unit moves Rescue unit arrived on accident site Finland: Design guide for rescue operations. Ministry of Interior, 2012 30
Finnish requirement for total response time Risk area Dispatch time (min) Rescue service response time (min) Total response time (min) I 2 11 13 II 2 14 16 III 2 22 24 IV 2 < 40 31
Fire service response times (2) Finnish Rescue Services pocket statistics 2012-2016 (D2_2017) First unit start+drive: 2016 median 7:33 Rescue services response time: 2016 median 12:06 Total response time 2016 median 14:39 UK Fire incident response times Response (start+drive) to fires: 2015/16 average 8:47 32
Response time distributions Finland: First unit start+driving times 2004-2007. England: First unit start+driving times to dwelling fires (51 % in 7 minutes) Home Office. Statistical Bulletin 01/17. 33
Reliability of first extinguising VTT Tiedotteita 2479 (2001-2007) Unavailability of manual suppression equipment Used / not used Effect (suppress, limit, no effect) Effect Dwellings Business Office Healtcare Industry Storage Suppressed/limited 47 % 42 % 40 % 53 % 59 % 38 % No effect 8 % 4 % 1 % 3 % 6 % 13 % Not available/used/functional 45 % 54 % 59 % 44 % 35 % 49 % Rescue service pocket statistics 2012-2016 (D2/2017) 2012 2013 2014 2015 2016 USE OF FIRST EXTINGUISHING IN BUILDING FIRES Used 2040 1965 2005 1730 1891 34 % Not_Used 3456 3351 3563 3188 3333 59 % Unknown 472 477 471 356 367 7 % EFFECT OF FIRST EXTINGUISHING IN BUILDING FIRES Extinguished 1201 1116 1176 1038 1116 59 % Limited 608 600 599 494 536 29 % No effect 225 238 227 190 231 12 % Did not work 6 11 3 8 8 0 % 34
Reliability of automatic systems in Finland Reliability of active systems was studied in a number of TKK MSc theses in late 1990 s/early 2000s. Since then, the data collection has become difficult as the inspection services have been opened for competition and private companies. These companies have no oblication to collect such data. Sprinkler system reliability is difficult to estimate because there are very few fires in sprinlered buildings. 2012-2016, smoke alarm activated in 38 % of Finnish fires. 35
Reliability of sprinkler systems Component reliability, maintenance, inspection, NFPA (Koffel 2005): Operational reliability (does the system activate when it should) is about 90 %. Performance reliability (does the system put out fire) 1989-1998 data: 8.5 % - 33 % NFPA 2005 report (Rohr & Hall): Sprinklers operated AND were effective in 36 % of all fires (22-48 % depending on occupancy type). Operated and Not effective in 4 % of fires. Effective when operated: 96 % Moinuddin et al. (2008): Sprinkler system failure probability in high-rise buildings in range 3 % to 14 %. 36
Fire deaths Deaths / a Total Men Women Source: Geneva Association http://www.pelastustoimi.fi/turvatietoa/ehkaise-palonsyttyminen/tulipalon-vaarallisuus/palokuolemat 37
Fire deaths, UK UK, Home Office Statistical Bulletin 07/17 Decreasing trend At the same time, the number of fires is coming down as well. England population 53 M 0.57 deaths/10 5 38
Reasons behind high fatality rate in Finland Ability to function at the time of ignition has been reduced in almost all fires. Factors affecting the ability: Age Physical and mental illnesses Alcohol Drugs http://www.pelastustoimi.fi/turvatietoa/ehkaise-palonsyttyminen/tulipalon-vaarallisuus/palokuolemat 39
Societal risk and individual risk The risks associated with different situations or designs can be compared using a F-N curve. FN-curve shows the frequency of an event vs. the number of loss. Developed for process industries. Applied for cruise ships, buildings, e.g. traffic tunnels. Accurate computation of a F-N curve is very laborious - but the orders of magnitude matter. Acceptance criterion either as a political decision or from statistics. Individual risk Korhonen, T., Hostikka, S., Keski-Rahkonen, O. A proposal for the goals and new techniques of modelling pedestrian evacuation in fires. Fire Safety Science 8: 557 569, 2005. doi:10.3801/iafss.fss.8-557 9/12/2017 40
1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 Milj. Property loss Direct losses or total losses (interruption)? Rescue service statistics use the construction cost index (Haahtela index) and estimate of property loss. 2016 residential fires caused 75 142 m 2 and 142 M direct lossess. ( 1900 /m 2 ). Average loss in residential building fire (2016) was 24 m 2 and 25 000. Insurance (Finanssiala ry) 250 200 150 100 50 0 Fire Burglary Water 41
Property loss sum Property loss distribution VTT Tiedotteita 2479 (2001-2007) Few big fires cause most losses (2 % 50 % loss) Property loss distributions Cumulative fire number sum Direct damage [m 2 ] Building and property total loss [ ] 42
Reading for next week Meacham Brian J., Charters, D., Johnson, P., Salisbury, M. Building Fire Risk Analysis. SFPE Handbook (ed. 5) Chapter 75. 43
Homework tips 1. Calculate detection time from detector temperature using LOOKUP-function. 2. Switch to Monte Carlo mode (Setup sheet) 3. Define the random variables 4. Link the random variable values into the inputs of the detector model and the fire brigade arrival time model. 5. Define output quantities (detection time, intervention time) 6. Carry out simulation 7. Generate the output distributions 9/12/2017 44