## ABSTRACT

The I&AB (Initiator and All Barriers) method was first introduced at ESREL 2016, as an efficient means to calculate, thanks to closed form formulae, the reliability of a very large repairable system with dependencies among components. The mathematical support of I&AB is continuous time Markov chains, and therefore it cannot be used for modeling the spent fuel pool of a nuclear power plant, because for this system, there are two kinds of deterministic delays that must be taken into account: grace times (for example, after the total loss of cooling of the pool, it takes exactly 14 hours for the water to start boiling), and deterministic failures due to the limited capacity of water tanks. In the present paper, we extend the I&AB method to account for deterministic delays. We explain how we could apply this method in the case of the fuel pool of the EPR (European Pressurized Reactor) starting from a model in the form of a BDMP (Boolean Logic Driven Markov Process), and how results and computation times compare to a Monte Carlo simulation of the same BDMP.