FIRE DESIGN OF STEEL MEMBERS - Civil and Natural Resources ...

Figures 6.2 b shows the comparison of the results from the spreadsheet method
with those from SAFIR. The curves are much closer to each other than those in
Figures 6.2a when the changing specific heat of steel is not taken into account by
the spreadsheet method. The maximum temperature reached in the modified
spreadsheet analysis is 825 °C, which is only 25 °C higher than the maximum
temperature reached from the SAFIR simulation.
The decay region of the simulation in Figures 6.2 a and b shows differences
between the results from SAFIR and the spreadsheet method. In Figures 6.2 a this
can be explained by the differences in specific heat, but Figures 6.2 b should have
a closer decay rate than that that appears on the graph. The rate of temperature
increase is very similar in Figures 6.2 b, so it is expected that the rate of decrease
should be more similar also. The SAFIR curve appears to be slightly translated to
the right from the spreadsheet method curve, which suggests that the steel from the
SAFIR simulation takes longer before it begins to heat up. The rate of decrease
from the SAFIR simulation is quite substantially slower than the decrease rate with
the spreadsheet method, so the steel is losing heat to its surroundings more slowly
than the spreadsheet method suggests.
The SAFIR programme also has variations of specific heat, density and thermal
conductivity for its insulation. This could contribute to the difference in the
results, although it does not explain why the difference occurs only in the decay
stage and not in the heating stage at temperatures below 700 °C.
6.3 REAL FIRES:
To confirm the results and conclusions found in Sections 4 and 5 of this report, the
temperatures of steel beams and fires from experimental test data has been
compared with calculations performed by the spreadsheet method, and simulations
in SAFIR with the same fire temperature data. The beam sizes used in the
unprotected and protected tests have been used in the computer analysis where
possible, and where not the closest beam size by dimensions and mass per unit
length has been used.
123