Microgrid Operation Control with State-of-Charge- Dependent Storage Power Constraints

TL;DR

This paper develops a model predictive control scheme for microgrids with lithium-ion storage, incorporating dynamic, state-of-charge-dependent power constraints modeled as convex polytopes, improving operational safety and reliability.

Contribution

It introduces a novel convex polytope modeling approach for dynamic storage constraints and integrates it into a control scheme for microgrid management.

Findings

Polytopic constraints prevent power limit violations.

Without constraints, violations occurred in 11.77% of steps.

With constraints, no violations were observed.

Abstract

The microgrid concept offers high flexibility and resilience due to the possibility of switching between grid-connected and stand-alone operation. This renders microgrids an auspicious solution for rural areas and critical infrastructure. In standalone or islanded mode, the main objective is cost minimization while ensuring a safe and reliable operation. Optimal operation schemes for microgrids usually assume fixed power limits for energy storage units. This, however, is not sufficient for lithiumion energy storage systems, which often come with dynamic power limits that depend on the state of charge. These limits are especially prominent when the state of charge is close to its boundaries. In this paper, dynamic constraints for energy storages are modelled using convex polytopes and fitted to experimental data acquired from an 11.6 kWh lithium-ion energy storage system. The polytopic constraints are integrated in a model predictive control scheme that was designed for a standalone microgrid composed of a fuel cell, a photovoltaic generator and a lithium-ion energy storage system. To evaluate the advantages, a case study with two configurations is performed. The model predictive controller without polytopic constraints led to constraint violations in 11.77 % of the simulation time steps with a maximum deviation of 118 % above the power limits. The configuration with polytopic constraints in contrary led to no violations over the entire simulation horizon.