Accounting for Optimal Control in the Sizing of Isolated Hybrid Renewable Energy Systems Using Imitation Learning

TL;DR

This work proposes a flexible and computationally efficient sizing framework for energy storage and renewable capacity in isolated energy systems, accounting for uncertainty in the renewable generation and the optimal feedback control, and implements an imitation learning approach to stochastic neural model predictive control.

Abstract

Decarbonization of isolated or off-grid energy systems through phase-in of large shares of intermittent solar or wind generation requires co-installation of energy storage or continued use of existing fossil dispatchable power sources to balance supply and demand. The effective CO2 emission reduction depends on the relative capacity of the energy storage and renewable sources, the stochasticity of the renewable generation, and the optimal control or dispatch of the isolated energy system. While the operations of the energy storage and dispatchable sources may impact the optimal sizing of the system, it is challenging to account for the effect of finite horizon, optimal control at the stage of system sizing. Here, we present a flexible and computationally efficient sizing framework for energy storage and renewable capacity in isolated energy systems, accounting for uncertainty in the renewable generation and the optimal feedback control. To this end, we implement an imitation learning approach to stochastic neural model predictive control (MPC) which allows us to relate the battery storage and wind peak capacities to the emissions reduction and investment costs while accounting for finite horizon, optimal control. Through this approach, decision makers can evaluate the effective emission reduction and costs of different storage and wind capacities at any price point while accounting for uncertainty in the renewable generation with limited foresight. We evaluate the proposed sizing framework on a case study of an offshore energy system with a gas turbine, a wind farm and a battery energy storage system (BESS). In this case, we find a nonlinear, nontrivial relationship between the investment costs and reduction in gas usage relative to the wind and BESS capacities, emphasizing the complexity and importance of accounting for optimal control in the design of isolated energy systems.

Figures (7)

• Figure 1: Overview of the control-aware sizing framework, starting with a cost model and the range of capacities considered and ending in the optimal sizing when generator usage and price for all configurations are known.
• Figure 2: Overview of the architecture of the network used in the neural MPC. The parameters $B, C, H$ are the batch size, number of hidden channels and sequence length respectively.
• Figure 5: Power plans for standard MPC (top) and neural MPC (middle) for the same scenario. The bottom subfigure highlights the discrepancies in BESS energy, power and gas turbine power between standard and neural MPC.
• Figure 6: Gas turbine usage as a function of the risk parameter $\alpha$. The dashed red line indicates the neural MPC with exact forecast as input.
• Figure 7: Snapshot of a single time step for the neural stochastic MPC process and its view of the time horizon. In the top row is the current state and the collection of possible wind scenarios based on the forecast, together with the red line indicating the wind speed required to power the platform completely by wind. In the bottom row is a scatter plot as well as histograms of the suggested gas turbine and battery powers for the different scenarios.

Theorems & Definitions (2)

• Remark
• Remark