Haiyang Lin, Qiuwei Wu, Xinyu Chen, Xi Yang, Xinyang Guo, Jiajun Lv, Tianguang Lu, Shaojie Song, and Michael B. McElroy. 2021. “
Economic and Technological Feasibility of Using Power-to-Hydrogen Technology under Higher Wind Penetration in China.” Renewable Energy, 173, Pp. 569-580.
Publisher's VersionAbstractHydrogen can play a key role in facilitating the transition to a future deeply decarbonized energy system and can help accommodate higher penetrations of renewables in the power system. Arguments to justify this conclusion are supported by an analysis based on real-world data from China’s Western Inner Mongolia (WIM). The economic feasibility and decarbonization potential of renewable-based hydrogen production are discussed through an integrated power-hydrogen-emission analytical framework. The framework combines a high-resolution wind resource analysis with hourly simulation for the operation of power systems and hydrogen production considering technical and economic specifications on selection of three different types of electrolyzers and two operating modes. The results indicate that using wind power to produce hydrogen could provide a cost-competitive alternative (<2 $kg-1) to WIM’s current coal-dominated hydrogen manufacturing system, contributing at the same time to important reductions in wind curtailment and CO2 emissions. The levelized cost for hydrogen production is projected to decrease in the coming decade consistent with increases in wind power capacity and decreases in capital costs for electrolyzers. Lessons learned from the study can be applied to other regions and countries to explore possibilities for larger scale economically justified and carbon saving hydrogen production with renewables.
Tianguang Lu, Xinyu Chen, Michael B. McElroy, Chris Nielsen, Wu Qiuwei, Hongying He, and Qian Ai. 2021. “
A reinforcement learning-based decision system for electricity pricing plan selection by smart grid end users.” IEEE Transactions on Smart Grid, 1949-3061 .
Publisher's VersionAbstractWith the development of deregulated retail power markets, it is possible for end users equipped with smart meters and controllers to optimize their consumption cost portfolios by choosing various pricing plans from different retail electricity companies. This paper proposes a reinforcement learning-based decision system for assisting the selection of electricity pricing plans, which can minimize the electricity payment and consumption dissatisfaction for individual smart grid end user. The decision problem is modeled as a transition probability-free Markov decision process (MDP) with improved state framework. The proposed problem is solved using a Kernel approximator-integrated batch Q-learning algorithm, where some modifications of sampling and data representation are made to improve the computational and prediction performance. The proposed algorithm can extract the hidden features behind the time-varying pricing plans from a continuous high-dimensional state space. Case studies are based on data from real-world historical pricing plans and the optimal decision policy is learned without a priori information about the market environment. Results of several experiments demonstrate that the proposed decision model can construct a precise predictive policy for individual user, effectively reducing their cost and energy consumption dissatisfaction.