“…We consider a system in which: i) the first machine has a single failure mode with parameters p 1,1 = 0.01 and r 1,1 = 0.1; ii) the second machine is characterized by three failure modes that occur with probability p 2,1 = 0.01, p 2,2 = p 2,3 = 0.005, and are repaired with probability r 2,1 = 0.1, r 2,2 = 0.05 and r 2,3 = 0.01 respectively; iii) the number of Kanbans is fixed at B = 50 items. As already shown in (Colledani et al, 2014b), under the single stage Kanban control policy (top box of Figure 1) the lead time distribution in a two machine line is characterized by a bi-modal distribution that has peaks at time units 1 and B − 1. The magnitudes of these peaks depend on the difference of the efficiencies of the two machines.…”