TL;DR
This paper introduces a novel system architecture and algorithms for residential microgrids to minimize electricity costs by intelligently sharing renewable energy and battery storage among homes. By coordinating energy flow based on real-time pricing and predicted generation/consumption, the system achieves significant cost reductions, even for homes with similar usage patterns.
What Problem Does This Paper Address?
The increasing adoption of renewable energy sources (like solar panels) and the implementation of Time-of-Use (TOU) electricity pricing present both opportunities and challenges for residential energy consumers. While these trends aim to reduce overall electricity consumption and reliance on the grid, they also introduce complexities. Specifically, the intermittency of renewables can lead to surplus energy when demand is low and deficits when demand is high. Traditional net metering policies, which allow selling excess energy back to the grid, are often limited by regulations. This paper addresses the problem of how to effectively manage and share locally generated renewable energy and stored energy within a community of homes (a microgrid) to minimize their collective electricity bills under dynamic TOU pricing, especially in scenarios where net metering is restricted.
Methods
The authors propose a three-tiered approach for energy management within a microgrid:
- System Architecture: A centralized microgrid architecture is designed, featuring home controllers and a central controller. Home controllers monitor energy generation (solar), consumption, and battery status. The central controller orchestrates energy sharing decisions. A DC distribution network is proposed for inter-home energy sharing for efficiency, while the traditional AC grid serves as the utility connection.
- Spatial Energy Sharing: This module aims to minimize AC transmission within a single time window by maximizing energy sharing among homes. It addresses transmission conflicts and efficiency losses by calculating maximum transmission speeds for suppliers and demanders and prioritizing sharing based on these speeds.
- Temporal Energy Sharing: This module leverages TOU pricing to shift energy consumption and storage decisions across time. It uses a "look forward window" to identify future low-price periods and strategically charges batteries or requests energy from the AC line to minimize costs.
- AC Allocation Optimization: This final layer refines the energy allocation by identifying scenarios where energy sharing might lead to unnecessary AC transmission costs. It aims to further reduce overall AC energy costs by optimizing the balance between inter-home sharing and grid interaction.
Datasets
- Energy Consumption Data: Collected from 40 homes, with data logged every second.
- Energy Harvesting Data: Collected from solar panels over six days.
- Battery Charging/Discharging Data: Measured in real-time.
The paper uses these empirical traces to simulate and evaluate their proposed system under different TOU models.
Key Insights and Ideas
- Community-Level Optimization is Key: The paper strongly demonstrates that coordinating energy sharing across multiple homes in a microgrid yields significantly better cost savings than individual home optimization.
- Spatial and Temporal Sharing Synergy: Combining spatial (within a window) and temporal (across windows) energy sharing strategies is crucial for maximizing cost reductions. Spatial sharing balances immediate needs, while temporal sharing leverages price fluctuations.
- The "Look Forward Window" is Powerful: This heuristic for temporal energy sharing effectively anticipates future price changes, allowing for strategic energy storage and grid interaction to minimize costs.
- Even Homes with Similar Patterns Benefit: A surprising finding is that the system can reduce costs even for homes with similar energy usage patterns, highlighting the benefits of collective optimization and resource pooling.
- DC Microgrids for Efficiency: The proposal of a DC distribution network for inter-home sharing points towards potential efficiency gains by reducing AC-DC conversion losses.
Key Results
The proposed system, referred to as GSC (Global Sharing and Charging), demonstrated significant cost reductions:
- 20% Reduction in AC Electricity Costs: The system achieved an average reduction of 20% in AC electricity costs compared to baseline scenarios.
- Outperforming Existing Methods: GSC outperformed the GES (Global Energy Sharing) algorithm by 22% in Ontario TOU and 37.9% in New England TOU. It also significantly outperformed ISC (Individual Smart Charge), highlighting the importance of community-level sharing.
- Near-Oracle Performance: GSC's performance was close to an "Oracle" scenario (which assumes perfect future knowledge), indicating the effectiveness of its predictive and optimization strategies.
- Cost-Benefit Analysis: The system is projected to have a return on investment of less than five years, primarily due to savings on AC energy transmission costs.
Limitations and Future Work
- Prediction Accuracy: While the paper uses EWMA for consumption prediction and weather forecasts for harvesting, the accuracy of these predictions can impact performance. More sophisticated prediction models could further improve results.
- Scalability: The paper focuses on a community of homes. Scaling to larger microgrids or integrating with the broader grid infrastructure might introduce new challenges.
- Transmission Conflicts: While addressed, complex transmission scenarios with many simultaneous requests could still pose challenges.
- Future Work: The authors suggest exploring more advanced prediction models, investigating the integration of electric vehicles into the microgrid energy sharing, and developing more robust algorithms for handling dynamic grid conditions and market price fluctuations.
Summary updated