Return to search

Prediction of Fire Growth on Furniture Using CFD

A fire growth calculation method has been developed that couples a computational fluid dynamics (CFD) model with bench scale cone calorimeter test data for predicting the rate of flame spread on compartment contents such as furniture. The commercial CFD code TASCflow has been applied to solve time averaged conservation equations using an algebraic multigrid solver with mass weighted skewed upstream differencing for advection. Closure models include k-epsilon for turbulence, eddy breakup for combustion following a single step irreversible reaction with Arrhenius rate constant, finite difference radiation transfer, and conjugate heat transfer. Radiation properties are determined from concentrations of soot, CO2 and H2O using the narrow band model of Grosshandler and exponential wide band curve fit model of Modak. The growth in pyrolyzing area is predicted by treating flame spread as a series of piloted ignitions based on coupled gas-fluid boundary conditions. The mass loss rate from a given surface element follows the bench scale test data for input to the combustion prediction. The fire growth model has been tested against foam-fabric mattresses and chairs burned in the furniture calorimeter. In general, agreement between model and experiment for peak heat release rate (HRR), time to peak HRR, and total energy lost is within pm 20%. Used as a proxy for the flame spread velocity, the slope of the HRR curve predicted by model agreed with experiment within pm 20% for all but one case.

Identiferoai:union.ndltd.org:wpi.edu/oai:digitalcommons.wpi.edu:etd-dissertations-1289
Date20 May 1999
CreatorsPehrson, Richard
ContributorsHoward W. Emmons, Committee Member, Nicholas A. Dembsey, Committee Member, Brian J. Savilonis, Committee Member, Jonathan R. Barnett, Advisor
PublisherDigital WPI
Source SetsWorcester Polytechnic Institute
Detected LanguageEnglish
Typetext
Formatapplication/pdf
SourceDoctoral Dissertations (All Dissertations, All Years)

Page generated in 0.002 seconds