This paper presents the analysis and simulation of the performance of fault location and isolation (FLI) in an automated power distribution feeder. This system is composed of two coupled networks: a power system consisting of a distribution feeder with multiple load buses, and an unreliable communication network of the distributed intelligent agents in the system, namely, the substation automation and fault protection units separating segments of the feeder. We provide a complete specification of a distributed algorithm for FLI and an exact characterization of the time from the occurrence of a fault to its location and isolation. Both apply to a distribution feeder with an arbitrary number of buses. These models are then refined into a hybrid simulation that combines three models executing in parallel: a power system model based on dc power flow, a distributed automation system model for the intelligent agents constructed using the IEC 61499 distributed automation standard, and an abstract communication network model that unreliably links the physically distributed agents. The results demonstrate the effect of communication network reliability at two levels of design abstraction, the correspondence of results at the two levels, and the use of a modern cosimulation framework to verify the performance of distributed smart grid automation algorithms.