Product flow analysis in distribution networks with a fixed time horizon

The movement of items through a product distribution network is a complex dynamic process which depends not only on the network's static topology but also on a knowledge of how each node stores, handles and forwards items. Analysing this time-dependent behaviour would normally require a simulation algorithm which maintains a globally-synchronised system state. For a certain class of problem, however, where the simulation is required to stop in a consistent state but not necessarily maintain consistency at all times, we show that an algorithm that makes localised decisions only is sufficient. As a motivating example we consider the practical problem of product recalls, in which our primary concern is the state of the distribution network at a specific time after a batch of suspect items was released, but we do not necessarily care about intermediate states leading up to the final one.