Utilizing flexibility in Microgrids using Model Predictive Control

Problem and Research Question

Microgrids can combine slow economic scheduling with faster corrective control. The central question is: how can an energy-management schedule serve as a reference for a faster MPC layer while leaving that controller enough flexibility to respond to power imbalance?

Conventional controllers act on the present state and a fixed setpoint, which wastes flexibility that only becomes valuable when forecasts and constraints are considered over a horizon. MPC solves a constrained optimization over a prediction window, but the question is how to formulate that optimization so that flexibility is explicitly utilized rather than merely accommodated.

Method

This paper developed a two-layer simulation framework for utilizing microgrid flexibility: The approach combines:

• Receding-horizon optimization — solving a constrained optimal control problem at each time step over a prediction window, so the controller re-plans as new measurements arrive.
• Soft schedule tracking — using the energy-management schedule as a penalized reference rather than an inviolable command.
• Fast balancing action — allowing the lower MPC layer to depart from the schedule when needed to counter a simulated imbalance.

The broader architecture includes a stochastic energy-management formulation, but the reported numerical example focuses on the lower MPC layer with a constructed schedule and disturbance. The optimization was implemented in Python with Gurobi, not CasADi.

Personal Contribution

Frederik Banis was the first author. He developed the MPC formulation for flexibility utilization, designed and ran the simulation experiments, and wrote the manuscript. Co-authors Daniela Guericke, Henrik Madsen, and Niels Kjølstad Poulsen contributed supervision, methodological guidance, and review.

Validation and Key Results

• The simulation illustrates the intended compromise: remain near the energy-management schedule during normal operation while using available flexibility to respond to imbalance.
• It provides a concrete formulation linking slower scheduling and faster control layers.
• The paper does not report a baseline-controller comparison, quantified improvement in cost or frequency, or a benchmark using realistic wind and solar profiles.

The paper was published in the proceedings of the Mediterranean Conference on Power Generation, Transmission, Distribution and Energy Conversion (MEDPOWER 2018), a peer-reviewed conference.

Related Outputs

• Conference paper: MEDPOWER 2018. DOI: 10.1049/cp.2018.1856
• Publication catalogue entry: data/publications.yaml, id 6.
• Related project: Target-Adjusted MPC for Microgrid Frequency Control — the broader microgrid MPC project this paper contributes to.
• Related journal paper: Load Frequency Control in Microgrids using target adjusted Model Predictive Control, IET Renewable Power Generation, 2019. DOI: 10.1049/iet-rpg.2019.0487 — the journal extension of this conference work.
• PhD thesis: Efficient Operation of Energy Grids, DTU, 2020. DOI: 10.11581/DTU.00000334

Collaborators and Institutions

• DTU Compute, Technical University of Denmark — Henrik Madsen, Niels Kjølstad Poulsen (supervisors), Daniela Guericke (co-author)

Status and Next Steps

Status: Published (2018). The flexibility-utilization MPC formulation is a constituent strand of the broader microgrid MPC research program and was extended in the subsequent IET Renewable Power Generation journal paper.

Transfer to current research: The sequential-decision framework developed here — constrained optimization over a horizon, acting before full information is available, explicitly utilizing the flexibility of the controlled system — contributes to the methodological base for adaptive waveform design for biomedical sensing. The shift is from microgrid flexibility to sensing flexibility, but the receding-horizon, control architecture is shared.

See the related project page for the broader research context, the Methods page for the model predictive control methodology, and the Research page for the research program overview.