|dc.description.abstract||There are many remote communities around the world which do not have interconnection with the power grid because of technical and/or economic constraints, and thus have to manage their energy requirements independently, mainly from fossil-fuel-based and in some cases renewable-based generation, operating as isolated microgrids. The reliable and economic operation of a microgrid is handled by an Energy Management System (EMS), which includes scheduling and dispatching Distributed Energy Resources (DERs) such as Distributed Generators (DG), Energy Storage Systems (ESS), with controllable loads and demand response (DR), while maintaining appropriate reserve levels, and considering uncertainty in the forecast of renewables. Thus, this thesis focuses on developing comprehensive EMSs that consider Unit Commitment (UC), and Optimal Power Flow (OPF) constraints, smart load models for DR, and possible deviations in the forecast of renewable-based DGs.
First, a mathematical model of smart loads in DR schemes is developed, based on a centralized and integrated UC and OPF EMS for isolated microgrids, to optimally dispatch generation and smart loads. These smart loads are modeled by a neural network (NN) load estimator as a function of the ambient temperature, time of day, Time of Use (TOU) price, and a peak demand constraint that the microgrid operator may set. A novel Microgrid EMS (MEMS) approach based on a Model Predictive Control (MPC) technique to manage forecast uncertainties is formulated; this tool yields optimal dispatch decisions of DGs, and ESS, and obtains optimal peak demand constraints for smart loads, considering power flow and UC constraints simultaneously. The impact of DR on the microgrid operation with the developed MEMS is studied using a CIGRE benchmark system that includes DERs and renewable-based generation, demonstrating its feasibility and advantages over existing EMS approaches, and showing the benefits of controllable loads in microgrids.
In isolated microgrids, the network losses and voltage drops across feeders are relatively small. This feature is utilized through a novel linearization approach applied to the unbalanced power flow equations to propose practical EMSs. The proposed EMS models are Mixed Integer Quadratic Programming (MIQP) problems, requiring less computation time and thus suitable for online applications. The proposed practical EMS models are compared with a typical decoupled UC-OPF based EMS with and without consideration of system unbalancing. These EMS models, along with ``standard" EMS models, are tested and validated, using an MPC approach to account for forecast deviations, on the CIGRE medium voltage benchmark system and the real isolated microgrid of Kasabonika Lake First Nation (KLFN) in Northern Ontario, Canada. The presented results demonstrate the effectiveness, and practicability of the proposed models.
In the third stage of the thesis, the impact of Electric Thermal Storage (ETS) systems on the operation of Northern Communities' microgrids is analyzed. A mathematical model of the ETS system is developed, in collaboration with a colleague from Karlsruhe Institute of Technology, and integrated into an EMS for isolated microgrids, in which the problem is divided into UC and OPF subproblems, to dispatch fossil-fuel-based generators, ESS, and ETS charging. To account for the deviations in the forecast of renewables and demand, an MPC technique is used. The proposed ETS-EMS framework is tested and studied on a modified CIGRE medium voltage benchmark system, which comprises various kinds of DERs, and on the real KLFN isolated microgrid system. It is shown that the ETS significantly reduces operating costs, and allows for better integration of intermittent wind and solar sources.
Finally, equivalent CO2 emission models for fossil-fuel-based DG units are developed considering their individual emission characteristic and fuel consumption. These models are then integrated within a microgrid EMS model, together with constant energy, and demand shifting load models, to examine the possible impact of DR on the total system emissions and economics of a microgrid, using again an MPC approach to manage forecast uncertainties. The impact of including the developed emission models on the operation of an isolated microgrid, equivalent CO2 emissions, and costs are examined considering five different operating strategies. The proposed operating strategies are validated on a modified CIGRE medium voltage benchmark system, with the obtained results highlighting the effectiveness of the proposed EMS and also demonstrate the impact of DR on emissions and costs.||en