Art

Cost of Fire: Exploring Fire Incident Data For A Design Tool Methodology IFE Kent Group CPD Training 29 th May PDF

Description
Cost of Fire: Exploring Fire Incident Data For A Design Tool Methodology IFE Kent Group CPD Training 29 th May 2012 Chris Salter BEng (Hons), AIFireE Stephen Emmitt Dino Bouchlaghem Prof. G Ramachandran
Categories
Published
of 55
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
Share
Transcript
Cost of Fire: Exploring Fire Incident Data For A Design Tool Methodology IFE Kent Group CPD Training 29 th May 2012 Chris Salter BEng (Hons), AIFireE Stephen Emmitt Dino Bouchlaghem Prof. G Ramachandran Civil & Building Engineering Loughborough University Background IRMP Project Introduction IRMP Project Part of the Evaluation of Prevention and Protection Activities for Commercial, Public and Heritage Buildings project Full project details and this work can be viewed at 3 IRMP Project Aims How to identify, measure and mitigate the social and economic impact that fire and other emergencies can be expected to have on individuals, communities, commerce, industry, the environment and heritage 4 PhD Aims Understand the Architecture, Engineering and Construction (AEC) industries views on Fire Engineering and cost Analyse fire incident data Construct a decision support tool for fire engineering cost decisions 5 Published Work To Date C Salter, N Bouchlaghem (2011), Fire Engineering in the UK : A UK Practitioners View, International Conference on Building Resilience, Sri Lanka. C Salter, G Ramachandran, N Bouchlaghem (2011), A Cost Benefit Tool for Fire Protection Engineers : An Analysis, 2nd IRMP Conference, Glasgow University. 6 Introduction 451 fire deaths, 12,200 injuries in 2008 Department for Communities & Local Government (2010), 'Fire Statistics, United Kingdom 2008, Department for Communities and Local Government. Fires cost the UK economy 8.3 billion in 2008 Communities and Local Government. (2011), The Economic Cost of Fire: Estimates for 2008, DCLG Publications. 7 Fire Fatalities in the UK Taken from Fire Statistics 2008, Department of Communities and Local Government 8 Cost of Fire Claims Taken from Association of British Insurers (2009), 'Tackling Fire: A Call For Action. 9 Cost Reductions in Fire Fundamentally, cost reduction is the only value we have to make our engineering better. Torero, Optimal Costs 11 Taken from The Economics of Fire Protection, Ramachandran (1998) Previous Work Work undertaken on costs of fires and fire statistics use Proposed Decision Support Tool Decision support tool aimed at new builds, not existing or heritage structures Tool could potentially be used in retro fitting but costs would be different 13 UK Fire Statistics 15 Sources of Data 16 FDR 1 Data 17 FPA Large Loss Database Data Analysis Assumptions and Notes Only 2005 FDR 1 data used Restricted data, potentially volatile results due to one year of data FDR 1 Data entry - Many fields missing 18 Alarm Probability Calculated from the FDR 1 results Only 38.4% of records in 2005 had AFD present From the filtered FDR 1 data, probability of alarm activation and raising the alarm is 74.1% 19 Does Activation Affect Damage? Alarm Activated Alarm Not Activated 30 Frequency of Fires (%) Size of Fire (Grouped by m 2 ) 20 Statistical Tests Mann Whitney Statistical Test Data is not normally distributed Proves Alarm Activation does affect final damage 21 Probability of Extinction Systems Operating Only 2.8% of fires had an extinction system (2005) Sprinklers most popular extinction system (59.1% of systems) Focus on sprinklers - other groups not large enough for statistical analysis 22 Probability of Sprinklers Activating Frequency Percent Activated Activated - Extinguished Fire Activated - Controlled Fire Failed Activated - Failed to Control Fire Failed to Activate Probability of Sprinklers Activating FDR 1 records show 34.08% activation overall Previous studies state sprinklers are 95.6% effective (Rutstein and Cooke, 1983, Vaidogas and Šakėnaitė, 2011) However this figure is when sprinklers are activated, not overall. 24 Probability of Sprinklers Activating Only 43% of fires are large enough to activate fires (Rutstein and Cooke, 1983) Ramachandran states that a fire has to be 3m² before a sprinkler activates. However, records show activations over 3m² are 67.8% 16.8% of activations of sprinklers are under 3m² 25 Sprinkler Activation FDR 1 form does not differentiate between sprinkler system types Potentially could explain the smaller than 3m² activations 26 Regression Analysis Take the factors that affect the fire size and then create a predictive model Allows each variable to quantitatively show how the variable affects the data Basis behind previous work conducted by Lin et al on Taiwanese residential buildings 27 Multiple Regression Would allow a predictive output of damage However, data required to be continuous - FDR 1 damage data is non continuous 28 Logistic Regression Logistic regressions allows regression with binary values Would give output as probability of an event In this case, fire exceeding 200m² Preliminary chi squared tests showed that largest effect was DANGSUBS and IGNTDISC 29 Logistic Regression Results Observed Predicted Percentage Correct Under 200m² Over 200m² Under 200m² 32, Over 200m² 1, Overall Calculating Costs Fire Damage Costs Previous Work - Costs Work undertaken by Ramachandran ( ) Beck (1987) Wright (1998) Ashe (2006) Lin (2009) Fraser-Mitchell (2010) 32 Cost Data 4 sources identified for cost data 1. Rateable Values 2. Average from FPA Database 3. BCIS Tool 4. Xactimate Software 33 Rateable Values Measure of a properties rental value Collected by Valuation Office Agency Pros Cons Split into different counties Underestimate total costs Freely available and already calculated as /m² 34 FPA Database Taken from loss adjustors estimates Only on incidents where a fatality happened or cost was estimated as over 100,000 Pros Cons Once calculated, only needs access to FPA data periodically to update costs Based on estimates 35 BCIS Tool Database of costs of a new build - Collected by RICS Data submitted from construction companies Pros Cons Already in /m² Not free data Very detailed data 36 37 Xactimate Software Xactimate Software Cost estimation software from Xactware Pros Cons Incredibly detailed cost breakdowns Aimed at individual buildings, not across a building type Updated quarterly 38 Costs Rateable values and Xactimate discarded Tool to use FPA data FPA and BCIS data comparison first Examine the two datasets to see how they compare Concerns that lost adjustors over estimate costs 39 FPA/BCIS Comparison FPA - 18 categories 18 occupancies BCIS - 8 main categories 414 occupancies Merge BCIS occupancies to match FPA occupancies FPA data costs calculated into /m² 40 Costs - FPA Total Loss Estimate ( ) 25,000,000 20,000,000 15,000,000 10,000,000 Industrial Processing Non Residential -Misc Food And Drink Retail Warehouses Entertainment and Culture Permenant Agricultural Education Religious Sport 5,000, Area Damaged m 2 41 Costs - BCIS Total Loss Estimate ( ) 9,000,000 8,000,000 7,000,000 6,000,000 5,000,000 4,000,000 3,000,000 Non Residential Education Entertainment and Culture Food and Drink Industrial Processing Medical Permanent Agricultural Public Utilities Retail Sport Transport Warehouses 2,000,000 1,000, Area Damaged m 2 42 Average Costs Occupancy FPA Cost /m² BCIS Cost /m² Industrial Processing 4, Non Residential Misc 1, , Food and Drink 1, , Retail 1, , Warehouses 1, Entertainment and Culture 1, , Permanent Agriculture Education 1, , Religious 1, , Sport 1, , Average Costs Occupancy FPA Cost /m² BCIS Cost /m² Industrial Processing 4, Difference here seems to be significantly different Non Residential Misc 1, , Food and Drink 1, , Retail 1, , Warehouses 1, Entertainment and Culture 1, , Permanent Agriculture Education 1, , Religious 1, , Sport 1, , Statistical Tests t-test performed on the datasets to compare Datasets are the same (to 95% certainty) Either dataset can be used in the tool Differences attributed to FPA data including loss of equipment and stored items 44 FSEC Toolkit FSEC Toolkit gives a cost figure for the time for FRS to arrive Cost based on FPA publicly released figures FSEC calculates average and then doubles them (based on work from 1977) - gives significantly higher figures than FPA costs 45 Cost Model Costs taken from either FPA or BCIS can form the model Only focuses on property cost - not consequential Fire growth follows αt² model Costs should therefore follow the same model, not linear 46 Discussions, Conclusions and Further Work Conclusion - Statistics Fire data collection adequate for current CLG usage, needs changing for predictive output Statistical analysis of dataset show that predictive model is inaccurate with current data FDR 1 - IRS dataset focus on reactive statistics, not predictive 48 Conclusion - Statistics If fire statistics are to be used predictively, alternative data needs to be collected IRS data needs to be less binary Regression analysis of fire incident data would give predictive output, if the data is correct 49 Conclusion - Statistics Log linear analysis potentially could be used to analyse FDR 1 data FPA data may prove to be more beneficial once it has accumulated more records Unclear how FPA database takes into account inflation 50 Conclusions - Statistics Care needs to be taken with sprinkler statistics Using only FPA data can skew sprinkler reliability/activations Better consideration to difference between activations and successes Better reporting needed for successful activations 51 Conclusions - Costs Methods of calculating costs considered and compared BCIS or FPA data can be used in future cost calculations Methodology now in place if fire statistics improve 52 53 Ques%ons? Contact Details 54
Search
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks