DescriptionTraditional methods for solving inventory control problems have their roots in tools such as queueing theory and dynamic programming. De- spite being powerful tools in several contexts, particular characteristics of real-world inventory control problem might render these techniques unsuitable to deal with aspects such as multiple products, perishability, variable lead times, or randomness of the input parameters. This talk will present recent developments that show how optimal control poli- cies can be obtained using stochastic programming (SP), which is a powerful framework that allows the consideration of uncertainty using mathematical programming tools.
Due to the discrete nature of the scenario-based representation used in SP, there is an inherent trade-off between the quality of the repre- sentation of the uncertainty and the size of deterministic equivalent problems. We show that, for benchmark cases to which a closed-form optimal solution can be calculated, the approximation using SP is often very close to the optimal for an adequate number of scenarios.
We present recent developments showing that several aspects of in- ventory control problems can be easily modelled using SP and that optimal solutions can be obtained very efficiently even for large-scale problems. We illustrate these ideas showing applications related to the management of blood unit inventories in a network of hospitals served by a central blood bank.
|Event title||European Conference on Operational Research|
|Degree of Recognition||International|