 |
Center for Nonlinear Studies |
We discuss solution techniques for the new class of Integrated Network Design and Scheduling problems. Motivating applications for this problem class include infrastructure restoration after an extreme event and plug-in hybrid electric vehicle (PHEV) battery charging and discharging within a smart grid. Infrastructures, such as power grids and transportation systems, can be modeled as networks. Network managers must coordinate repairs or operational decisions using limited resources in order to maximize performance. Selecting which components to repair or utilize (i.e. downed power lines) can be viewed as network design decisions. Traditional network design decisions only focus on the end performance of the design, i.e., the network operation after all components are repaired. Clearly, in infrastructure restoration the success of the efforts depend on how well the services come back online. Therefore, it is important to allocate resources, such as work groups, to implement network design decisions. This resource allocation can be viewed as scheduling decisions. This novel model incorporating the combination of decisions occurring simultaneously does increase the problem difficulty, which motivates the need for both exact and approximate solution methods. I will present complexity results on the problem class under standard network performance metrics, exact and approximate solution methods, and case studies based on real-life data sets representing the infrastructure systems of lower Manhattan and New Hanover county, NC.
Host: Feng Pan