Fragmentation and pulverisation are commonly observed problematic
phenomenon in nuclear fuels at burnups in excess of 65 GWd/tHM. The
high burnup structure that forms at the rim of pellets with very high burnups
has been shown to contribute to this pulverisation, along with other factors
such as fission gas release and irradiation temperature. Fragmentation is
unwanted as it can cause thermal conductivity to decrease through the
introduction of insulating gas-gaps in cracks and in the spaces between
fragments, and causes significant release of fission gas which will increase
the gas pressure in the fuel rod. Pulverisation during a loss of coolant accident
(LOCA) event typically leads to fuel relocation within the rod, and dispersion
outside the rod if there is a burst as a result of rod overpressure. This
project considers two approaches for determining the likelihood and amount
of fuel pulverisation, through the use of mathematical modelling based on
inputs from fuel performance code such as EDF’s ENIGMA. The first approach,
a broadly empirical model, uses a curve mapped to fragment size data
from post irradiation examination (PIE) results to predict fragment sizes if a
transient occurs at different temperatures. The curve is a good fit between
the burnups of 60 to 100 GWd/tHM against experimental data from the Halden
reactor project and Studsvik LOCA tests. The second approach, a mechanistic
model based on bubble bursting, created by K. Kulacsy, was tested and
its methodology validated against other literature, then modified through
the addition of molecular dynamics data from LAMMPS for gas pressures.
In some example bubble size distribution data for the rim region in high
burnup fuel, the model predicted a coarse estimate for 98% of bubbles
bursting, implying this region would almost completely pulverize into small
fragments in a LOCA where the terminal temperature reached 2000K. The
link between the number and sizes of bubbles overpressuring and bursting, and the amount of pulverisation occurring needs further experimental data to
prove, however current work suggests that the more bubbles that burst, the
more fine fragmentation is expected. Comparison with literature supports
this with a link between burnup and the average amount of gas in bubbles,
and the fragment size relationship with burnup identified in the first model