Flame Spread Modelling Using FDS4 CFD model
Thesis DisciplineFire Engineering
Degree GrantorUniversity of Canterbury
Degree NameMaster of Engineering in Fire Engineering
This thesis examines the prediction of opposed flow flame spread in the Fire Dynamics Simulator version 4 (FDS4) Computational Fluid Dynamics (CFD) model by adapting the Lateral Ignition Flame Transport (LIFT) test procedure. It should be noted that FDS4 was all that was available at the time of the analysis despite FDS5 is now available for beta testing. This research follows on from previous work where LIFT experiments were conducted for various New Zealand timber and timber based products; those materials include Beech, Macrocarpa, Radiata Pine, Rimu, Hardboard, Medium Density Fibreboard (MDF), Melteca faced MDF, Plywood and Particle Board. The objective of this research is to investigate the accuracy of flame spread modelling in FDS4; where the prediction of opposed flow flame spread parameters from FDS4 were directly compared with the experimental results that were obtained experimentally. The standardised procedure for determining the material ignition and flame spread properties was followed and applied to simulate the LIFT test. The LIFT test apparatus was set up in FDS4 with a domain size of 0.9 x 0.3 x 0.3 metres in the x, y and z directions respectively. From the heat flux distribution along the calibration specimen, it indicated that calibration of the LIFT apparatus can be executed in FDS4 where the percentage error is within 1.2%. This report also provides the thermal transport properties (i.e. thermal conductivity and specific heat capacity) of the tested New Zealand timber and timber based products. These were determined using a transient plane source technique and subsequently these properties were entered as the surface identifications in FDS4. The ignition tests were not performed as part of the simulated LIFT test since a direct comparison with the results was required to give a more meaningful assessment. For this reason, the ignition parameters that were obtained from the previous experiments were employed to carry out the flame spread test. Due to the concept of a preheat time required by the standard test method and FDS4 being not able to preheat specimens, the temperature immediately after the preheat time was calculated and implemented for the specimens. The heat transfer problem was solved using an explicit method; where specimens were divided into 11 different nodes. Different scenarios were investigated to see the effect that the selected combustion model has on modelling flame spread. The two analytical models tested were (1) thermoplastic fuels and (2) charring fuels model. Furthermore, the flame spread was visualised using either the Mixture Fraction or the HRRPUV model in Smokeview; where the rate of flame spread for each specimen was obtained. And lastly, three different absorption coefficients (0.6, 0.7 and 0.8) for each specimen were examined; this parameter contributed significantly to the rate of flame spread as it determines the amount of heat flux being absorbed by the specimen during the time of preheating. A study of the grid size was also performed to investigate the accuracy of the FDS4 simulations with the grid size selected. It has been found that increasing the size of the grid cell does not greatly affect the flame spread results. Moisture content and heat of vaporisation input variables were also examined. From the flame spread data, moisture content does not have a significant role in modelling flame spread. However, it was indicated that the heat of vaporisation has an effect on the output of the flame spread parameters. It was determined from the sensitivity analysis that the most appropriate solid boundary condition to be used in predicting the flame spread would be thermoplastic fuels model with an absorption coefficient of 0.8. By using this scenario as the basis, the plot of the arrival time against the distance along the specimen exhibits a similar trend of flame spread with the experimental results at first, but later on, the extinction of flame front actually occurred at a much earlier stage than the experimental results showed. In general, the analyses showed that FDS4 cannot perform the LIFT test where the prediction of flame heating parameter and minimum heat flux for spread were out by more than 20% shown by the direct comparison between experimental results. However, the prediction of minimum heat flux required for ignition seems to agree with the experimental results where the percentage error is within 20%.