Affine Policies and Principal Components Analysis for Self-Scheduling in CAES Facilities

This paper presents a novel methodology based on Principal Components Analysis (PCA) and Affine Policies (AP) for self-scheduling of a price-taker Compressed Air Energy Storage (CAES) facility operating under uncertainties. The proposed PCA-AP model is developed from the facility owner's perspective, which partakes in energy, spinning, and idle reserve markets. A methodology is proposed to select the required price uncertainty intervals from actual data based on a Box Cox technique. For a more realistic representation, the detailed thermodynamic characteristics of the CAES facility are considered, taking into account as well modern CAES facilities that may charge and discharge concurrently. To validate the proposed PCA-AP model and approach, the results obtained are compared with an existing Affine Arithmetic (AA) model, which is also based on an affine approach, and Monte Carlo Simulations (MCS), which can be considered as the benchmark for comparison purposes. The input data, forecast prices and intervals of uncertainty, are taken from the Ontario-Canada electricity market for 2015-2019. From the studies presented, it can be observed that the new PCA-AP approach provides less conservative results as compared to the AA approach, and hence can be considered an adequate methodology for day-ahead operations in systems with significant sources of uncertainty.

Automated summary

This paper presents a novel methodology for self-scheduling of a price-taker Compressed Air Energy Storage (CAES) facility based on Principal Components Analysis (PCA) and Affine Policies (AP) in the presence of uncertainties. The proposed PCA-AP model, developed from the facility owner's perspective, considers energy, spinning, and idle reserve markets. The results show that the PCA-AP approach provides less conservative results compared to the existing Affine Arithmetic (AA) model, making it a suitable method for day-ahead operations with significant uncertainties.