Sensitivity based planning and operation of modern distribution systems

Abstract

The power system is undergoing numerous changes due to the rapid increase in energy demand, rising concerns of climate change, and increased engagement of consumers in the energy market. Consumers are now motivated to invest in distributed energy resources (DERs), e.g., rooftop photovoltaic systems, due to their environmental advantages. The number of electric vehicles (EVs) is also increasing due to their reliability and low carbon footprint. Despite their numerous benefits, the rapid onset of DERs and EVs introduces new technical challenges to distribution systems including (1) complex system operation due to reverse power flows, (2) voltage instability issues; and (3) increased power losses due to poor DER and EV planning as well as their temporal uncertainty. Existing methods to improve the planning and operation of distribution systems in the presence of these technologies use available data from measurement devices in the grid together with traditional load flow analysis. However, some of the major limitations of existing impact-analysis techniques include (1) inability to capture uncertainty, (2) high computational burden; and (3) lack of foresight. This dissertation addresses these research gaps by proposing computationally efficient, yet accurate, sensitivity frameworks that help simplify planning and operation of modern distribution systems. First, a novel probabilistic sensitivity framework is developed to quantify the impact of grid-edge technologies, e.g., DERs and EVs, on line losses for balanced and unbalanced distribution systems. Results show that the developed approaches offer high approximation accuracy and four-orders faster execution time when compared to classical approaches. Secondly, this dissertation develops a novel preemptive voltage monitoring approach based on low-complexity probabilistic voltage sensitivity analysis that predicts the probability distribution of node voltage magnitudes, which is then used to identify nodes that may violate the nominal operational limits with high probability. The proposed approach offers over 95% accuracy in predicting voltage violations. To address the complexity-accuracy trade-off with existing planning methods, this dissertation develops a novel spatio-temporal sensitivity approach to analyze both spatial and temporal uncertainties associated with DER injections. The spatio-temporal framework is used to quantify voltage violations for various PV penetration levels and subsequently determine the hosting capacity of the system without the need to examine a large number of scenarios. This framework is further extended for EV charging station allocation to ensure minimum active power losses and voltage deviations. Thirdly, this dissertation develops a new system voltage influencer (SVI) paradigm that identifies strategic locations in the system that have the highest influence on node voltages. The SVI nodes are ranked and used within a stochastic control setup to eliminate voltage violations. The development of SVI paradigm is essential given the increased number of behind-the-meter and utility-controlled DERs, where it is becoming difficult to select optimal control points and counter the impact of the introduced uncertainties. The developed approaches in this dissertation help system operators quickly reveal impending voltage and loss issues resulting from power changes at the grid edge.