eprintid: 727
rev_number: 8
eprint_status: archive
userid: 17
dir: disk0/00/00/07/27
datestamp: 2018-05-27 18:02:45
lastmod: 2018-05-27 18:02:45
status_changed: 2018-05-27 18:02:45
type: report
metadata_visibility: show
creators_name: Champneys, A.
creators_name: Fresneda-Portillo, Carlos
creators_name: Hewitt, Ian
creators_name: Liu, X.
creators_name: Hunt, Matthew
creators_name: Nagapetyan, Tigran
creators_name: Please, C.P.
creators_name: Shang, Xiaocheng
creators_name: Tant, Katherine
creators_name: Witelski, Tom
creators_name: Wood, David
creators_name: Zioo, Piotr
title: Estimating the Spread of Fire in Buildings
ispublished: pub
subjects: tech_dev
studygroups: esgi91
companyname: A.W.E.
full_text_status: public
abstract: AWE is developing a general fire modelling capability at a city scale and want to upgrade their current capability by incorporating suitable building-scale models into the simulation. The main question that needs answering is: given an initial state in which fires are known to be burning in a particular location or region of a building of given dimensions, is it possible to determine estimates for the rate at which fire spreads throughout this building? In particular, estimates for the temperature at the outer walls as a function of time.
This report addresses the problem in three ways. First, the state of the art in the literature on fire modelling is reviewed and a mid-complexity model (referred to as “the Kyoto model”) which treats rooms as individ- ual compartments is recommended for further study. This is then sim- plified into an ODE compartmentalised model only those processes that are believed to be essential. A fully dimensional version of the model is developed including estimates for all the physical paramaters. After nondimensionlisation, the model is simulated for a building comprising a chain of rooms separated by fire doors. Finally, a homogenised reaction- diffusion PDE model is developed in one and two space dimensions. In this model walls are treated as areas of greater porosity. Simulation results reveal the sensitivity of the fire spread to the internal layout of the building. These final two models have both been implemented as a Matlab and C-code deliverable respectively.
problem_statement: 1 Introduction
(1.1) This report gives details of the work done during the 91st European Study Group with Industry held at Bristol from 15-19th April 2013. The problem, as presented by AWE is set out in section 2 below. The key part of the remit was not to produce a high complexity computational model, but to brainstorm how to produce a simplified model that nevertheless captures the true physics of fires at the building scale.
(1.2) The rest of the report is outlined as follows. First, in section 2 a precise statement of the problem is given. Section 3 then conducts a partial liter- ature review and makes some initial assumptions. Section 4 then presents three different methods of solution. First, in 4.1 we present the bare bone details of a compartmentalised ODE-based building-scale model from the literature. Next, Section 4.2 derives a simplified version of this model and presents results simplified, but fully parametrised version of this model for the spread of the fire through a 1D chain of rooms. Appendix A.1 presents more details of the derivation of this model, including reasoning for approx- imate values for all the key parameters in the model. The final solution, in Section 4.3 involves modelling fire spread through the building via an inho- mogeneous PDE, with the details given in Appendix A.2. Finally, Section 5 draws conclusions and suggests avenues for future work.
2 Problem statement
(2.1) How fire spreads is of great importance when considering the safety of crit- ical assets. As such, AWE plays an important role in supporting UK Gov- ernment departments in assessing risks associated with the spread of fire.
(2.2) Currently AWE has a large scale, grid-based model for urban areas that ig- nores the effects of individual buildings. However, it is known that different buildings types will change the spread of fire significantly. Thus suitable building-scale models must be incorporated into the simulation to assess the risk at specific locations.
(2.3) In principle it is possible to incorporate a great deal of information about the layout and composition of buildings. But it is extremely expensive in terms of money and time to acquire and incorporate this information into a model. Therefore, it is desirable to know what can be determined with only a minimal description of the building.
(2.4) Thus our main objective is to determine estimates for the rate of fire de- velopment, spread and decay in a building with minimal description of the building. That is, given an initial state in which fires are known to be burn- ing in a particular location or region of a building of given dimensions, we
1
Estimating the Spread of Fire in Buildings ESGI91
 want to determine estimates for the rate at which fire spreads throughout this building.
date: 2013
citation:   Champneys, A. and Fresneda-Portillo, Carlos and Hewitt, Ian and Liu, X. and Hunt, Matthew and Nagapetyan, Tigran and Please, C.P. and Shang, Xiaocheng and Tant, Katherine and Witelski, Tom and Wood, David and Zioo, Piotr  (2013) Estimating the Spread of Fire in Buildings.  [Study Group Report]     
document_url: http://miis.maths.ox.ac.uk/miis/727/1/ESGI91-AWE_CaseStudy.pdf