A deterministic fracture mechanics analysis does not address the uncertaint
ies involved in material properties, magnitudes of loads, location and size
of the flaws, etc. However, in a real life situations such uncertainties c
an affect significantly the conclusions drawn out of a deterministic analys
is. The principles of probabilistic fracture mechanics may be used to ascer
tain the effects of such uncertainties. A computer code PARISH (Probabilist
ic Assessment of Reactor Integrity under pressurised thermal SHock) has bee
n developed based on principles of PFM for analysing a reactor vessel subje
cted to pressurised thermal shock. The code assumes a crack in the reactor
vessel of random dimension depending upon Marshall flaw depth cumulative di
stribution function. The applied SIF at the tip of this crack is computed e
ither using closed form solution or a precomputed data base. The material K
-IC is then calculated using the crack tip temperature and RTNDT The value
of RTNDT depends on the initial value of RTNDT and the increase in the valu
e of RTNDT depending upon the fluence, copper content and nickel content. A
Gaussian distribution is assumed for these parameters. If the applied SIF
is more than the material K-IC, the crack is assumed to propagate. The crac
k can be arrested only if the applied SIF is less than the material K-Ia at
that location. The material K-Ia again depends upon the RTNDT, which in tu
rn depends upon the fluence, copper content and nickel content of the mater
ial at that location. The vessel failure is assumed if the crack propagates
by the 75% of the thickness. Such procedure is repeated for large number o
f cracks (of the order of one million). Using Monte-Carlo simulation, proba
bilities of no crack initiation, crack initiation and vessel failure are ca
lculated. The present probabilities are conditional in the sense that the t
ransient is assumed to occur. The case studies are presented involving a nu
clear reactor vessel subjected to two different kinds of pressurised therma
l shocks. (C) 1999 Elsevier Science Ltd. All rights reserved.