Determining Realistic Loss Estimates for Rack Storage Warehouse Fires (2006)
Type of ContentTheses / Dissertations
Thesis DisciplineFire Engineering
Degree NameMaster of Engineering in Fire Engineering
PublisherUniversity of Canterbury. Civil Engineering
AuthorsPorter, Timothy Millershow all
At present there is no simple, yet scientifically robust method for calculating insurance loss estimates due to a fire. Therefore building owners and insurers can not make suitably informed decisions when selecting fire protection measures or setting premiums as they have no way of defining the true risk they face. As a consequence this research aims to investigate a number of techniques in an effort to define one as appropriate for further research. Three different methods were explored and consist of risk based analysis, deterministic hand calculations and Computational Fluid Dynamics (CFD). Extensive literature reviews were conducted in each area and the final models were based on the outcomes of this research. Rack storage warehouses were chosen for analysis as they are currently topical within the fire engineering community and are a particular concern for insurers. The risk based analysis employed statistical techniques including event tree analysis and monte carlo simulation to calculate loss distributions and sensitivity analyses. The hand calculation method was based on equations presented in the literature and incorporated the use of a zone model (BRANZFire) to calculate deterministic loss estimates. The CFD model used was Fire Dynamics Simulator and full scale warehouse fires were modelled using this programme. It was concluded that Fire Dynamics Simulator is an inappropriate tool as the capability for providing loss estimates in a timely manner is currently beyond the model's capabilities. Of the two remaining methods the statistical risk based model was selected as the most appropriate for further investigation. The primary reasons for this decision were the ability to calculate loss distributions and conduct sensitivity analyses, as well as its versatility and user friendliness. Improved statistical data was defined as imperative for future development of the model.