Modeling of Cannibalization

In Chapters 2-5 we addressed the problem where repair or resupply could be initiated at any point in time or continuously; in Chapters 6 and 7 we considered the case where resupply is periodic. In the latter case we considered the impact of cannibalization (consolidation of ORU “holes” into the smallest number of systems by remove-and-replace maintenance). Now we want to return to the earlier problem where resupply is continuous and show how cannibalization can be modeled.

In most applications, spares are purchased under the assumption that cannibalization will not be practiced regularly. Of course, there will be situations where maintenance personnel cannibalize a few items to prevent a large portion of the fleet from being grounded, but this does not imply that cannibalization is standard procedure (or cost-free) to be assumed in the optimization. Even if cannibalization is not assumed in the optimization of what to buy, we believe it is desirable to evaluate the increase in availability that could be achieved with those spares if maintenance does cannibalize. The availability under cannibalization is an upper bound on what could be achieved.

We noted in Section 1.11 that a stockage policy optimized under the assumption of no cannibalization tends to produce a robust stockage policy that performs well regardless of the cannibalization policy actually practiced by maintenance; that stockage optimized under the assumption of cannibalization tends to perform rather badly when maintenance does not cannibalize. Even so we may want to optimize under the assumption of cannibalization because: (1) with newer plug-in systems cannibalization may be easy to perform; (2) it may be desirable to anticipate a wartime scenario, in which cannibalization is more likely to take place.