Recovering Cyclic Schedules Using Dynamic Produce-Up-To Policies

We consider facilities that follow a cyclic schedule to replenish the inventory of a set of items through production by a shared resource. We introduce a dynamic produce-up-to policy that recovers the target cyclic schedule after a single disruption, and is also shown to be effective when disruptions are more frequent. Our policy is more flexible than traditional recovery policies in that our policy is able to adjust the amount of idle time observed during recovery in response to disruptions, and yet re-establish the target idle time as the schedule recovers. This results in a policy that not only saves cost and time, but provides better schedule stability than other recovery procedures. Furthermore, unlike simple produce-up-to policies, our policy is anticipatory—replenishments will be speeded up or delayed, whichever is necessary, to help avoid congestion at the shared resource. In addition, recovery is controlled by a single “knob” or parameter that can tune recovery to be fast and aggressive (frequent setups and small batches) or slow and methodical (few setups and larger batches). Finally, our policy is easy to implement, augmenting a traditional produce-up-to policy with a simple set of counters that control replenishment decisions.