Master's assignment
Student: Ties Westendorp
Supervisors: Martijn Schoot Uiterkamp and Johann Hurink
Programme: Applied Mathematics - University of Twente
Finished: August 2019
With the rising penetration of high peak demand appliances (such as electric vehicles) and local production through e.g. rooftop photovoltaics, there is an increasing need for demand side management approaches; high consumption or production peaks may cause damage to the low voltage grid, which through peak shaving could have been avoided. We assume some appliances to be 'smart', in the sense that they can communicate their power needs and schedule their usage rather than ad hoc. The simultaneous optimization of multiple smart devices can not be done efficiently. Leading us to consider devices in isolation, and aggregating their schedules on a higher level to make the optimization problem tractable again. This has the benefit that each device is relatively easy to solve on it's own. Several device types have been considered in the literature (see T. van der Klauw's thesis).
Energy storage devices (buffers) are uniquely capable of cancelling out production and consumption peaks against one another. Though optimization models of them are typically kept simplified to be able to solve them optimally. In this assignment we examine methods for solving the energy storage problem where we include conversion losses. This problem turns out to be NP-complete in the general case.