Thermal Analysis of Effect of a Compartment Fire on Window Glass. September 1, 1988-August 31, 1989.
Thermal Analysis of Effect of a Compartment Fire on
Window Glass. September 1, 1988-August 31, 1989.
(707 K)
Joshi, A.; Pagni, P. J.
NIST GCR 90-579; NISTIR 4449; 23 p. June 1990.
U.S./Japan Government Cooperative Program on Natural
Resources (UJNR). Fire Research and Safety. 11th Joint
Panel Meeting. October 19-24, 1989, Berkeley, CA,
Jason, N. H.; Cramer, D. M., Editor(s)(s), 233-252 pp,
1990.
Sponsor:
National Institute of Standards and Technology,
Gaithersburg, MD
Available from:
National Technical Information Service
Order number: PB90-244468
Keywords:
glass; windows; mathematical models; radiation; thermal
stresses; vents; thermal analysis
Abstract:
Glass breaking in fires is an important practical
problem since a window acts as a wall prior to breaking
and as a vent after breaking. As Emmons explained,
windows break in fires due to thermal stress from the
differential heating of the central portion and the
shaded edge. If the depth of shading around the edge is
much greater than the glass thickness, one can assume
that the edge remains at its initial temperature. This
paper determines the surface temperature history,
[equation] of the glass. The temperature at breaking is
when [equation] where [equation] and [equation] both
give the strain at breaking in tension. The glass
coefficient of linear thermal expansion is alpha, the
glass modulus is epsilon and sigma b is its tensile
strength. Typical property values suggest the range of
50 deg C-100 deg C for the breaking T. Here the
transient, one-dimensional (into the glass normal to the
pane), inhomogenous (in-depth radiation absorption)
energy equation is solved using an innovative Laplace
Transform technique suggested by Baum. Time varying
equations are solved numerically by using the
trapezoidal rule for numerical integration and
Newton-Raphson's method for determining the roots on
non-linear equations. Results are presented for typical
values of the governing dimensionless parameters.