(778e) Multi-Scale Operational Planning and Optimal Sizing of Hybrid Renewable Energy System | AIChE

(778e) Multi-Scale Operational Planning and Optimal Sizing of Hybrid Renewable Energy System


H. Lee, J., Korea Advanced Institute of Science and Technology (KAIST)
Realff, M., Georgia Institute of Technology
Decentralized and hybrid renewable energy system has recently received attention for providing energy in a small and remote area with its own grid (referred to as micro-grid). In this study, operational planning and optimal sizing of a micro-grid having large wind farm and battery storage device is studied. Operational planning is generally made in two time-scales; unit commitment decisions on which source should be on-line at what time are pre-determined one day ahead; and dispatch decisions on the output of on-line sources are then decided on hourly time basis. Two stage stochastic programming (2SSP) is conventionally formulated for this one-day operation with a number of intra-day uncertainty scenarios. However, the finite horizon setting is a main factor to limit sustainable operation due to end-effects. At the same time, the yearly investment problem for sizing energy generation and storage resources is typically performed to minimize future operating costs as well as capital cost. These argue yearly operation model beyond one-day is required. Therefore in this study, 2SSP model for one-day operation is combined with daily-evolving Markov decision process (MDP). That is, the value of being in a state of current commitment and battery at the end of a day (called value function) is included in the objective function to account for longer term implications of the decisions. In the MDP formulation, inter-day variation of wind is captured, and the value function is approximated with a linear model. The coefficient vector of the linear model is recursively updated with a sampled observation estimated from the one-day operation model. With this value function capturing all future operating costs, optimal sizing of the wind farm and battery devices is determined based on a surrogate function optimization. Meanwhile, for the integration of the decision hierarchy, a wind model consistent across multiple timescales from seasonal to hourly is developed to provide both intra-day scenarios capturing hourly ramping and inter-day transition information. The results of the proposed method are compared to those of the original 2SSP model through a case study and real wind data.