Allocation of spinning reserves for autonomous grids subject to frequency stability constraints and short-term solar power variations

TL;DR

This paper introduces dynamic frequency constraints to improve the resiliency of low-inertia, isolated power systems with high renewable integration, optimizing the sizing of solar and storage resources while ensuring frequency stability.

Contribution

It proposes a novel dynamic frequency constraint method for resilient power system planning that accounts for short-term renewable power variations.

Findings

Neglecting frequency constraints can overestimate cost and emission reductions.

The proposed method ensures frequency stability under worst-case contingencies.

Optimal sizing reduces costs and emissions while maintaining system stability.

Abstract

Low-inertia, isolated power systems face the problem of resiliency to power variations. The integration of renewable energy sources, such as wind and solar photovoltaic, pushes the boundaries of this issue further. Higher shares of renewables requires better evaluations of electrical system stability, to avoid severe safety and economic consequences. Accounting for frequency stability requirements and allocating proper spinning reserves, therefore becomes a topic of pivotal importance in the long-term planning and operational management of power systems. In this paper, dynamic frequency constraints are proposed to ensure resiliency during short-term power variations due to, for example, wind gusts or cloud passage. The use of the proposed constraints is exemplified in a case study, the constraints being integrated into a mixed-integer linear programming algorithm for sizing the optimal capacities of solar photovoltaic and battery energy storage resources in an isolated industrial plant. Outcomes of this case study show that reductions in the levelized cost of energy and carbon emissions can be overestimated by 8.0% and 10.8% respectively, where frequency constraints are neglected. The proposed optimal sizing is validated using time-domain simulations of the case study. The results indicate that this optimal system is frequency stable under the worst-case contingency.