Conclusions - Hong and Lie (1993) defined joint reliability importance
(JRI) as a measure of how 2 components in a system interact in contri
buting to the system reliability. Their definition and theorems regard
ing JRI were limited to statistically independent component states. Th
is paper removes the statistical independence restriction by showing t
hat similar results hold in the more general case where component fail
ures can be statistically dependent; however, the calculation of actua
l values becomes more difficult, because covariance terms can appear i
n the JRI formula. Despite this, the essential determination and inter
pretation of the signs of the JRI remain unchanged. Thus analysts who
wish to use JRI (eg, as a design heuristic) can do so in working with
real systems where statistical independence is not valid. It is furthe
r shown that JRI is always non-zero for some classes of systems.