Rigorous decomposition methods for planning and scheduling of energy networks

点击翻译按钮获取中文摘要

Climate change, energy security, and public awareness in sustainability are among the central factors that have motivated the research and development of multi-energy systems. Multi-energy systems are energy networks comprising energy generation, conversion, transport, storage, and consumption units, where multiple energy carriers, such as gas, electricity, and heat, are involved. The proposed research program targets planning and scheduling of energy networks, which is to achieve the best network performance by coordinating the network units in an optimal way. This is a difficult decision-making problem, in that the problem needs to include nonlinear models for accurate description of the underlying physical-chemical relationships, and that the problems needs to include integer variables for yes/no and on/off decisions on the units. Existing methods for the decision-making problem only provides approximate solutions, which may be infeasible or far from the best. The primary goal of the proposed research program is to develop rigorous optimization methods that tackle the problem with the notion of "divide and conquer". Specifically, the methods will exploit the decomposable structure of the problem, and solve the problem via solving a set of much easier subproblems. The rigorousness of the methods ensures the feasibility and optimality of the solutions. In addition, we will also develop novel modeling strategies that can balance the model accuracy and solution efficiency. The research program includes four topics for the next five years. In the first topic, we will develop nonlinear planning and scheduling models for combined gas and power networks, at a multi-region level and at a single-region level, respectively. In the second topic, we will develop different rigorous optimization methods that can decompose the problem along temporal, spatial, logical, and scenario dimensions. In the third topic, we will coordinate the modeling and algorithmic development, i.e., we will approximate parts of the model by mixed-integer linear submodels for a predefined precision, such that the algorithmic development for the partially approximated model is a lot easier. In the fourth topic, we will develop nonlinear multi-energy models for a microgrid and a regional energy network consisting of multiple mircrogrids. The results of the research program will contribute to multiple disciplines, including energy systems engineering, planning and scheduling, and operations research. The research program will yield advanced global optimization software, which will enable reliable and efficient decision-making tools for offline design or real-time operation of energy networks, helping Canada's energy and manufacturing industries to achieve efficient, sustainable, and environmentally friendly operations.