Optimization and Control of Smart Renewable Energy Systems
Date of Award
Ph.D. in Engineering
Department of Electrical and Computer Engineering
Advisor: Malcolm Daniels
Electric power grids are currently undergoing a major transition from large centralized power stations to distributed generation in which small and flexible facilities produce power closer to where it is needed. This move towards a decentralized delivery of energy is driven by a combination of economic, technological and environmental factors. In recent years, the cost of renewable energy in the form wind turbines and solar PV has dropped dramatically due to advances in manufacturing and material science, leading to their rapid deployment across the US. To supplement the intermittent nature of wind and solar energy, there is a growing need for small, highly controllable sources such as natural gas turbines. With the fracking boom in the US, there is currently abundant natural gas to use for this purpose. The resulting proliferation of many small energy producers creates technical problems such as voltage and frequency control that can be addressed with battery storage, whose cost is also dropping. These factors are leading to a move awayfrom large energy production facilities that require too much initial investment. Also, a distributed supply is more efficient and reliable. The threat of global climate change is creating pressure to increase the integration of distributed generation and information technology is now capable of managing a greater number of energy producers, utilizing a vast supply of information to predict supplies and demand and to determine optimal dispatching of energy.The move towards a higher percentage of renewable energy creates many interesting technical issues, many of which are due to the lack of control over the renewable resources. Energy dispatching between multiple sources, some controllable and some not, and multiple loads leads to a need for dispatching strategies that maximize the percentage of the load that is met with renewable energy. A growing aspect of this energy dispatch is a stream of information about energy demand, weather and prices that can be used to make predictions and influence dispatching decisions. This work examines these problems, considering how to efficiently satisfy energy demands with a combination of controllable and non-controllable sources.A microgrid sizing problem, which seeks to optimize the size of various components for a given load, is considered first. The microgrid components include solar panels, small wind turbines, battery storage, and a backup diesel generator, and these elements alone are used to satisfy a load profile that approximates a single mixed-use (with commercial and residential space) building. The complete microgrid is modeled dynamically to generate flows of power from the various sources to the load, and several power dispatchingalgorithms are explored. To achieve this, a typical meteorological year (TMY) data file along with models for solar panels and wind turbines are used for predicting hourly renewable energy production. Hourly energy demand is determined from a combination of historical data and models for energy consumption. An economic analysis of the complete microgrid is made to develop a lifetime cost model that includes capital investment, operation and maintenance, component replacement, and diesel fuel. This cost is a function of the numbers of solar panels, number of wind turbines, battery storage capacity, power dispatching algorithm, and diesel generator size. The genetic-algorithm is then applied to determine which combination of these variables minimize the cost, and comparisons are made to the cost that would be incurred by supplying the load with traditional power alone.Several variations of this microgrid sizing scenario are modeled. First, the building's hot-water energy consumption is separated from its electrical consumption and is supplied from a central thermal storage tank. The capacity of this tank then becomes a new variable in the sizing problem, and the dispatching algorithm is modified to include energy flows into the tank. Secondly, various constraints are added to the optimization problem, such as restricting the percentage of the load to be met with non-renewable energy or restricting the amount of curtailed renewable energy (energy that cannot be stored or used by the load). Next, the system performance is examined with various load mixtures, ranging from all commercial to all residential. The source of backup energy is also modified, swapping the diesel generator with traditional grid energy, and a 100% renewable system can be modeled by removing the backup source entirely. In addition to these variations, the sensitivity of the system performance, as measured by cost, penetration, and curtailment, is explored with respect to each sizing variable individually. Results indicate that there is a cost benefit to combining commercial and residential electrical loads, especially for the case of a completely renewable microgrid with no backup energy source. As is expected, the system size and overall performance is strongly dependent on the detailed shape of the load profile. The use of thermal storage is shown to reduce cost and increase renewable penetration, as it is less costly than battery storage. Also, utilizing the main power grid as a backup source for the microgrid is less expensive than a diesel generator.Work on the microgrid dispatching algorithm forms the second half of this dissertation. The microgrid sizing problem required a rule for selecting how energy is dispatched each hour between the various elements, while meeting constraints, and a non-predictive heuristic methodology was initially used to achieve this. However, it is possible to apply a model predictive control (MPC) method to perform real-time optimization to maximize the power delivery from a renewable supply to a building. The MPC strategy utilizes predictions of the building's electrical and hot water loads, on an hourly basis, along with predictions of the output from the renewable supply. At each time step, these predictions are used to create an optimized power dispatching strategy between the microgrid elements, to maximize renewable energy use. MPC is incorporated into the microgrid sizing problem, and its performance relative to the non-predictive strategy is studied.The final work in this dissertation considered the combination of renewable energy and information technologies on electrical energy distribution. There is new potential for trading between prosumers, entities which both consume and produce energy in small quantities. Here, the optimization of energy trading between two prosumers is considered, each of which consists of a load, renewable supply, and energy storage. The problem is described within an MPC framework, which includes a single objective function to penalize undesirable behavior such as the use of energy from a utility company. MPC integrates future predictions of supply and demand into current dispatch decisions. The control system determines energy flows between each renewable supply and load, battery usage, and transfers between the two prosumers. At each time step, future predictions are used to create an optimized power dispatch strategy between the system prosumers, maximizing renewable energy use. Modeling results indicate that this coordinated energy sharing between a pair of prosumers can improve their overall renewable penetration.
Electrical Engineering, Energy, Mechanical Engineering, Microgrid, load profile, mixed-use building, wind turbines, photovoltaics, Model predictive control, Renewable Energy Penetration, Renewable Energy Curtailment, Battery, Thermal storage, Electrical load, Thermal load, Energy Prosumers, Microgrid Network
Copyright 2019, author
Aldaouab, Ibrahim, "Optimization and Control of Smart Renewable Energy Systems" (2019). Graduate Theses and Dissertations. 6839.