Microgrid Energy Management Optimization

An optimization-based energy management system designed to reduce exposure to wholesale electricity market volatility. This framework schedules battery dispatch in response to time-varying spot prices, enabling cost-effective energy arbitrage for grid-connected distributed energy resources (DER).

Project Overview

Wholesale electricity markets such as the NSW National Electricity Market (NEM) exhibit significant price volatility, with spot prices ranging from negative values during oversupply to $17,500/MWh during peak demand events. This price uncertainty creates both risk and opportunity for DER operators.

This project implements a linear programming-based decision-support tool that determines optimal battery charge/discharge schedules under price uncertainty. The system integrates:

• Solar PV generation forecasts
• Load demand profiles
• Real-time and forecast spot price signals
• Battery operational constraints

The optimizer outputs actionable dispatch schedules that maximize value extraction from price differentials while respecting technical and regulatory limits.

Key Performance Indicators

KPI	Description
Net Cost Reduction	Reduction in electricity procurement cost vs. no-storage baseline
Peak Import Mitigation	Avoidance of grid imports during extreme price intervals (>$1,000/MWh)
Negative Price Capture	Value extracted from grid charging during oversupply (negative pricing)
Self-Consumption Rate	Proportion of solar generation consumed on-site via storage

Industry Relevance

This project demonstrates competencies directly applicable to:

• Utilities & Retailers: DER dispatch optimization, tariff arbitrage, demand response integration
• Market Operators: Understanding of NEM price signals, dispatch scheduling, constraint management
• Renewable Integration: Solar-storage coordination, grid export management, capacity firming

The methodology scales from residential behind-the-meter systems to commercial/industrial virtual power plants (VPPs) and aggregated DER portfolios.

Technical Approach

Optimization Formulation

Objective: Minimize total operating cost over the scheduling horizon

$$\min \sum_{t=1}^{T} \left( \pi_t \cdot P_{\text{import},t} - \pi_{\text{FiT}} \cdot P_{\text{export},t} \right) \cdot \Delta t$$

Subject to:

• Power balance constraint (generation = demand + storage flows)
• Battery state-of-charge dynamics with round-trip efficiency losses
• SOC operating limits (10–90% to preserve battery health)
• Maximum charge/discharge power ratings

Technical Stack

Component	Technology
Optimization	CVXPY, HiGHS, PuLP
Data Processing	pandas, NumPy
Visualization	Matplotlib
Configuration	YAML-based parametric setup

Quick Start

# Clone and setup
git clone https://github.com/ZeuPark/MicrogridEnergyManagementOptimization.git
cd MicrogridEnergyManagementOptimization
python -m venv .venv && .venv\Scripts\activate  # Windows
pip install -r requirements.txt

# Run with real NEM data
python scripts/ingest_data.py --period 7d
python scripts/run_baseline.py
python scripts/run_optimization.py

# Or use synthetic data for benchmarking
python scripts/ingest_data.py --synthetic

Real-time NEM price data is fetched via OpenElectricity API. Synthetic profiles available for reproducible benchmarking.

Project Structure

├── src/
│   ├── optimization/    # Core LP solver and battery dispatch model
│   ├── features/        # Solar and load forecasting modules
│   ├── evaluation/      # Baseline comparison and sensitivity analysis
│   └── data/            # Data ingestion and NEM API client
├── scripts/             # CLI execution scripts
├── notebooks/           # Analysis and visualization notebooks
└── config/              # YAML configuration files

Development Roadmap

Phase	Milestone	Status
v1.0	Core LP optimization with synthetic data	Done
v1.1	Real NEM price data integration (OpenElectricity API)	Done
v1.2	Rolling-horizon Model Predictive Control (MPC)	Planned
v2.0	Stochastic optimization for forecast uncertainty	Planned
v2.1	Multi-objective: cost vs. battery degradation	Planned
v3.0	5-minute dispatch aligned with NEM settlement	Planned

Testing

pytest tests/ -v

Conclusion

This project demonstrates that optimization-based battery dispatch can materially reduce electricity procurement cost under volatile wholesale market conditions.

Validation with Real NEM Data

Using real NSW NEM price data (Dec 2024), the optimization model reduced net electricity cost by strategically avoiding extreme price spikes exceeding $5,000/MWh and exploiting negative pricing intervals during oversupply conditions.

Key outcomes:

1. Extreme Event Mitigation The economic value is driven primarily by avoidance of extreme price spike intervals. The optimizer pre-charges batteries during low-price periods to avoid forced imports during scarcity events.

2. Negative Price Exploitation The optimizer exploits negative pricing periods by charging from the grid, effectively monetizing oversupply conditions—a behavior unique to real wholesale market dynamics.

3. Operational Interpretability The resulting dispatch pattern aligns with expected economic behavior—charging during periods of solar surplus or low prices and discharging during evening peak demand—confirming model consistency with market incentives.

4. Computational Efficiency The convex formulation solves in under 0.3 seconds, supporting real-time deployment in rolling-horizon MPC architectures.

See reports/optimization_report.md for detailed analysis with real market data.

Author

Electrical Engineering Undergraduate Portfolio Project — Energy Systems Optimization

License

MIT License — see LICENSE for details.

References

• AEMO NEM wholesale market data
• Tesla Powerwall 2 specifications (reference battery parameters)
• Standard microgrid EMS formulations from IEEE literature