Optimal planning and management of stochastic demand and renewable energy in smart power grid

To combat global climate change, the reduction of carbon emissions in different industries, particularly the power industry, has been gradually moving towards a low-carbon profile to alleviate any irreversible damage to the planet and our future generations. Traditional fossil-fuel-based generation is slowly replaced by more renewable energy generation while it can be harnessed. However, renewables such as solar and wind are stochastic in nature and difficult to predict accurately. With the increasing content of renewables, there is also an increasing challenge to the planning and operation of the grid. With the rapid deployment of smart meters and advanced metering infrastructure (AMI), an emerging approach is to schedule controllable end-use devices to improve energy efficiency. Real-time pricing signals combined with this approach can potentially deliver more economic and environmental advantages compared with the existing common flat tariffs. Motivated by this, the thesis presents an automatic and optimal load scheduling framework to help balance intermittent renewables via the demand side. A bi-level consumer-utility optimization model is proposed to take marginal price signals and wind power into account. The impact of wind uncertainty is formulated in three different ways, namely deterministic value, scenario analysis, and cumulative distributions function, to provide a comprehensive modeling of unpredictable wind energy. To solve the problem in off-the-shelf optimization software, the proposed non-linear bi-level model is converted into an equivalent single-level mixed integer linear programming problem using the Karush-Kuhn-Tucker optimality conditions and linearization techniques. Numerical examples show that the proposed model is able to achieve the dual goals of minimizing the consumer payment as well as improving system conditions. The ultimate goal of this work is to provide a tool for utilities to consider the demand response model into their market-clearing procedure. As high penetration of distributed renewable energy resources are most likely applied to remote or stand-alone systems, planning such systems with uncertainties in both generation and demand sides is needed. As such, a three-level probabilistic sizing methodology is developed to obtain a practical sizing result for a stand-alone photovoltaic (PV) system. The first-level consists of three modules: 1) load demand, 2) renewable resources, and 3) system components, which comprise the fundamental elements of sizing the system. The second-level consists of various models, such as a Markov chain solar radiation model and a stochastic load simulator. The third-level combines reliability indices with an annualized cost of system to form a new objective function, which can simultaneously consider both system cost and reliability based on a chronological Monte Carlo simulation and particle swamp optimization approach. The simulation results are then tested and verified in a smart grid laboratory at the University of Hong Kong to demonstrate the feasibility of the proposed model. In summary, this thesis has developed a comprehensive framework of demand response on variable end-use consumptions with stochastic generation from renewables while optimizing both reliability and cost. Smart grid technologies, such as renewables, microgrid, storage, load signature, and demand response, have been extensively studied and interactively modeled to provide more intelligent planning and management for the smart grid.published_or_final_versionElectrical and Electronic EngineeringDoctoralDoctor of Philosoph