Background/Objectives
The rise of affordable, small-scale renewable energy solutions, such as rooftop solar, is transforming global energy systems. However, traditional large-scale electric grids often exhibit inefficiencies and equity issues, disproportionately impacting marginalized communities. These communities suffer from "energy poverty," where a household's energy expenses can be up to five times the average rate. This disparity in energy access and decision-making often leads to inequitable service during disasters and unfair rate structures.
To address these challenges, our work focuses on the development and implementation of AI-powered microgrids. These smaller, smarter grid systems emphasize the role of individuals as proactive contributors and are designed to enhance energy resilience, especially during infrastructure failures or natural disasters. The AI technology employed in these microgrids enables efficient micro-climate predictions and optimizes metering decisions, prioritizing community resilience and equity.
This approach not only encourages local self-reliance and energy efficiency (due to minimized transmission distances) but also integrates seamlessly with renewable energy sources, thus promoting sustainability. Scalable microgrids have the potential to invigorate local economies through job creation and technological innovation. Most importantly, they democratize energy distribution and combat energy poverty.
Pilot-Scale Work Context: Our pilot-scale work is being conducted in low-income and ALIPOC (African, Latinx, Indigenous, People of Color) communities in Atlanta, characterized by a history of marginalization and limited access to sustainable energy solutions. Historically Black College and Universities (HBCUs) and legacy community centers are the key structures in this program. The region's climate, coupled with its socio-economic background, provides a unique context for implementing and studying the impact of AI-powered community microgrids.
Collaboration and Methodology: In tackling this challenge, we have collaborated with multiple stakeholders, including academic institutions, public and private laboratories, and grassroots non-profit organizations, all committed to community-driven microgrid development. This presentation will detail both simulation studies using AI and the on-field impact observed in the selected communities.
Approach/Activities
Project Scale and Scientific Principles:
Our project operates at both the field and community levels, employing a combination of advanced Artificial Intelligence technologies and scientific principles to optimize microgrid performance. The key scientific principles and technologies being studied include:
Hybrid Artificial Intelligence Model: Utilizing an encoder-decoder neural architecture for micro-climate prediction, this model assists in forecasting energy production on solar panels, aiding in more informed decision-making.
Reinforcement Learning for Energy Dispatch Optimization: This model focuses on optimizing the performance of the micro-grid in terms of resilience, equity, and efficiency.
Micro-grid Design Strategy: This involves various configurations of power consumption with an emphasis on optimizing towards resiliency.
Community Engagement, Urban Planning, and Economics Principles:
In addition to technological advancements, our approach incorporates significant community engagement, urban planning, and economic principles:
Energy Resiliency Design Sessions: Collaborating with diverse stakeholders - full-time residents, local businesses, state and city representatives, grassroots non-profit organizations, faculty members from HBCUs, and corporate participants - using remix principles for inclusive decision-making.
Building Partnerships and Trust: Establishing deep and synergistic engagements with the community, focusing on building trust as a cornerstone for effective collaboration.
Evaluation-Based Mapping: Implementing strategic mapping sessions and tracking social impact returns as key metrics for assessing the success of philanthropic investments in the community. This approach is vital for allocating resources and prioritizing efforts in complex multi-party engagements.
Key Questions being Studied:
Our research is twofold: a) Exploring how technologies, particularly AI, can support and empower communities in leading and owning their social and environmental resilience efforts; b) Investigating the dynamics of sustainable and mutually beneficial corporate-community coalitions.
Results/Lessons Learned
Results from Scientific Studies:
Micro-Climate Predictions: Our simulation and on-field studies for micro-climate predictions achieved an average minimum accuracy of 85%, surpassing macro forecasts from publicly available commercial weather stations.
Optimal Controller Simulation: The simulation studies on the optimal controller indicated a net positive gain for community revenue, showcasing the economic viability of the microgrid system.
Community Engagement Outcomes:
Funding Streams Identification: During the design sessions, the community identified multiple funding sources, including corporate philanthropic giving (fellowship opportunities, community grants), and power purchase agreements. These align with corporate commitments to net-zero carbon and sustainability.
Smart Home Transformation: The first installation in the community involved transforming legacy homes into smart homes. This step was crucial for studying power consumption behavior, pollutant presence, and advising on behavior change for improved resilience and benefits.
Micro-grid Design and Workforce Planning: Optimal design, equipment selection, and workforce arrangement for setting up the microgrid were determined, starting at community resilience hubs, particularly a long-standing community center.
Vision for Economic Sovereignty: Design sessions also led to vision mapping for economic sovereignty in the community through participation in infrastructure projects, like the community center solar microgrid project.
Expected Data at Presentation:
By the time of the presentation, we anticipate having the following data available:
Economic data from the community solar grid.
Behavioral and power consumption data from the community.
Anecdotal evidence regarding community members' support for coalition projects.
Pollutant data and its impact on health and energy usage behavior.
Detailed results from various simulation studies mentioned above.
