A new study published in Engineering presents a risk-aware optimal dispatch strategy for resource aggregators, aiming to address the challenges brought by renewable energy’s uncertainty and unlock building energy flexibility.

With the global push towards green energy transition and carbon neutrality, renewable energy sources like solar and wind are playing an increasingly important role. However, their intermittent and uncertain nature has put pressure on power systems. Resource aggregators, which integrate distributed generation, energy storage, and demand-response resources, are emerging. But they face difficulties in dealing with renewable energy uncertainty and setting proper incentives for demand-side flexibility.

The research team, led by Hong Tang, Zhe Chen, Hangxin Li, and Shengwei Wang, proposed a strategy that combines probabilistic renewable energy prediction and bi-level building flexibility engagement. They used the natural gradient boosting algorithm (NGBoost) to develop a probabilistic photovoltaic (PV) prediction model. NGBoost is effective in extracting uncertainty from historical data without prior knowledge of uncertain variables. This model provides prediction intervals and probability density functions, enabling more informed decision-making.

To unlock demand-side flexibility, the researchers designed an interactive flexibility engagement scheme. It uses a bi-level optimization problem to determine demand response incentives and load-reduction capacities. This takes into account building users’ willingness based on indoor comfort and load satisfaction indices. For example, it considers the impact of load reduction on heating, ventilation, and air conditioning (HVAC) systems, dimmable lighting systems, and curtailable plug-in loads. Building users can set self-defined coefficients to quantify discomfort and dissatisfaction costs.

The aggregator’s energy management is optimized through a chance-constrained risk-aware bidding strategy. By leveraging the flexibility of multiple responsive loads and storage systems, the aggregator can maximize operating profits from electricity trading between the wholesale and retail markets.

The case study, conducted in Hong Kong, tested the operation of the resource aggregator on five weekdays. The results showed the effectiveness of the proposed strategy. The interactive flexibility engagement scheme led to cost savings for both the resource aggregator (8.90%) and the building cluster (1.87%). The NGBoost-based PV prediction was accurate, and the aggregator could make better decisions considering different confidence levels. A higher confidence level reduced the risk of supply shortages but also affected economic performance. The strategy also contributed to a 3% reduction in carbon emissions.

This new strategy provides a practical approach for resource aggregators to operate more sustainably and economically in the face of renewable energy challenges. It also offers insights for future research on integrating renewable energy and demand-side flexibility.

The paper “Risk-Aware Optimal Dispatch of Resource Aggregators Integrating NGBoost-Based Probabilistic Renewable Forecasting and Bi-Level Building Flexibility Engagements,” is authored by Hong Tang, Zhe Chen, Hangxin Li, Shengwei Wang. Full text of the open access paper: https://doi.org/10.1016/j.eng.2025.02.009. For more information about Engineering, visit the website at https://www.sciencedirect.com/journal/engineering.