The current ACR core safety cases are based on two limits, 20% brick e
ffective weight loss (equivalent to a peak value of 40%) and failure o
f any component (e.g a key/keyway or core restraint). The former has b
een adopted because of the paucity of experimental observations on AGR
graphite strength from test reactors beyond that peak value. It has n
o substantiation in terms of structural integrity of the core and is t
herefore clearly arbitrary. The latter is a structural feature, but th
e consequences for the overall serviceability of the core have not bee
n addressed. It is now reasonably clear that the failure of a graphite
component does not imply a loss of core functionality. It is, therefo
re, more appropriate to develop a safety case methodology which consid
ers the impact of the irradiation-induced changes in the graphite core
components upon the requirement for certain essential operational and
safety features (e.g. control rod insertion, maintenance of cooling,
insertion/withdrawal of fuel). Nuclear Electric has been developing su
ch a methodology based upon the probabilistic analysis of core ageing
and component failure with a computer code CORSET. The code examines t
he distortion of the core under various operating conditions, allowing
for local discontinuities due to component failure. These distortions
are repeatedly assessed allowing for statistical variations in compon
ent dimensions and properties, thus constructing a probabilistic distr
ibution against which the required operational and safety features can
be confirmed In this way, it is intended to construct a safety case f
or continued operation beyond the first failure of a graphite componen
t. AEA have carried out a large programme of work aimed at verifying a
nd validating certain constituent parts of this new methodology. The n
ew methodology meant that there was a requirement for statistical data
on material properties and strength in particular. First, the availab
le data was reviewed and a database encompassing all relevant graphite
thermo-mechanical properties was produced. The graphite component fai
lure criterion was then reviewed and a new approach was developed know
n as the fractional remanent strength. The fractional remanent strengt
h criterion required critical stress values for failure of components
under different modes of failure relevant to the AGR cores (e.g key/ke
yway loading and pinching, internal stress). A series of experiments,
both slice and full size brick geometries, were conducted to determine
the failure statistics and the response of core components in a varie
ty of situations (e.g cracked bricks, simulated internal stress and pi
nched key/keyways). Finally, the effects of material variability were
included in finite element calculations leading to a probabilistic dis
tribution of remanent strength at various times through reactor life.