Development and assessment of capacity planning and energy management approaches of Virtual Power Plants in State-controlled electricity markets: A case study in Egypt

This research intends to explore the concept of Virtual Power Plants (VPP) on new residential developments in Egypt, a country facing energy security problems and fast population growth. VPPs, aggregating conventional, renewable power plants and energy storage systems, has demonstrated effectiveness in providing more flexible control of solar and wind power, and improves the energy trading profit of the aggregated plants. VPPs are widely studied in “deregulated markets” in which power generators and suppliers are freely trading power on hourly basis. The aggregated energy systems in VPPs could have several configurations of output, accordingly the plants’ power dispatch is driven by profit maximization every hour given the hourly varying prices and energy demand. On the other hand, VPPs’ concept has not been presented in “regulated markets”, referring to energy markets that are largely under state control, and in which energy prices are dictated. Egypt is still adopting a regulated market but recently enabled an incentive to purchase power from Independent private power plants (IPPs) at their dictated price under power purchase agreement (PPA). Based on that framework, this research presents a novel VPP model for the Egyptian market and explore the economic and technical insights from that model. The proposed VPP aggregated rooftop solar PVs, gas-fired CCHP and energy storage units, covering electricity, cooling and heating demand, and exchanging power with the grid. The hourly varying power and thermal demand, as well as the existence of energy storage and grid integration, could lead to several configurations of the output. Therefore, an energy management driven by profit maximization is deemed necessary to ensure the optimal dispatch decision is taken. The optimization is solved with GA on hourly resolution for a year (8760-time steps). An alternative deep deterministic gradient policy (DDPG) model (a machine-learning method) is developed to program the model to act optimally and respond faster to the varying input data compared to GA. Eventually, an iterative method is developed to optimize both plants’ sizes and energy management, driven by investment costs minimization and profit maximization. The GA optimization yielded a payback period of 15 years in medium density (middle-income) housing (Case-1) versus 11 years in low density (high income) housing (Case-2), while the DDPG yielded 13 years and 10 years for medium and high-income housing, respectively. The simultaneous sizing and energy management yielded a payback period of 10 years and 9 years for medium and high-income housing, respectively. Energy storage systems, due to the flat pricing structure of the market, seem not to be providing any significant profit advantage that compensates for their investment and replacement costs, hence, causing economic infeasibility of the model. In all scenarios, the VPP model reduces CO2 emissions by 47-57% in Case-1 and 45-55% in Case�2, compared to the full grid dependency. The VPP model reduced dependency on the grid by 66- 78% in both case studies, helping to improve energy security. Finally, the sizing and energy management demonstrated that the scenarios with the maximum solar PVs capacities yielded the optimal lifecycle profit, highlighting the economic advantage that solar energy may bring to the market. A one-at-a-time sensitivity analysis is performed to verify the robustness of the model against electricity prices, fuel costs and energy demand. Eventually, the most influencing factor is found to be the electricity prices, yielding +/-30% profit variation with +/-20% input variation.