Traditional optimization approaches for handling uncertainty and risk typically require severe assumptions that are often not satisfied in complex practical settings. In an effort to overcome such limitations, several methods have been developed to handle uncertainty when the data and associated real world parameters do not behave according to classical assumptions. Two of the leading and most widely used examples are scenario optimization and robust optimization, both of which seek high-quality decisions that are feasible for all scenarios. However, both of these approaches likewise succumb to deficiencies encountered by classical methods, by exhibiting serious limitations in terms of the complexity and size of the models they can handle. Simulation optimization overcomes these limitations, and its flexibility and ease of use has contributed to its popularity as a preferred optimization approach to risk management applications.
This paper explains the techniques of scenario optimization and robust optimization and provides examples of how they are customarily used in real-world settings. Following a discussion of these methodologies is an introduction of simulation optimization, describing its flexibility and breadth across many potential applications, highlighting the separation of and interaction between the simulator and optimization engine, and reviewing its use in solving an energy industry problem involving selection of optimal portfolios of projects. The paper concludes with a summary statement of the types of problems that are best solved via simulation optimization.
Simply complete the form at the top to download our 19-page white paper.