Online

Khurram, Adil

Electrical and Biomedical Engineering

2021

Ph. D.

The overarching goal in power systems operations is to deliver energy in an efficient, reliable, and economical manner. To achieve this objective, the traditional power system operating paradigm is for generation to follow variable demand. As electrification and decarbonization policies are pursued, the levels of variable, renewable generation will increase, which will require that power system operator think beyond supply follows demand. This means that one needs to consider the potential flexibility provided by, for instance, internet-enabled, connected, and responsive loads, which are part of the broad class of behind-the-meter distributed energy resources (DERs). The research work presented in this dissertation is concerned with coordinating large populations of distributed energy resources (DERs) for providing services to the electric grid. DERs are flexible in the sense that their power consumption can be deferred in time, because DERs store energy in some form while serving the end-use customer. For example, electric water heaters store thermal energy in the form of hot-water in the tank. Therefore, aggregate fleets of DERs are an inexpensive source of virtual energy storage that the utilities can tap into for the purpose of balancing the variability in distributed renewable generation such as solar PV, wind etc. In this work, a novel, asynchronous and randomized load coordination scheme called packetized energy management (PEM) is considered. Packetized energy management is a device-driven scheme that uses a unique request-response mechanism for coordinating diverse fleets of DERs. The aggregate dynamics of PEM are captured using state-bin transition models. Parameter heterogeneity is incorporated by grouping together relatively similar DERs. Furthermore, a notion of state of charge can be attached to the aggregate that is representative of the energy content in the fleet by means of a low order model. This low order model is of interest to the utilities and grid operators since it allows them to design control trajectories for DER aggregations depending upon grid requirements and load forecasts. Furthermore, a cyber-physical platform is developed for the validation of aggregate models and control schemes. However, PEM modifies the normal behavior of DERs and for accurate prediction of load dynamics, the underlying customer driven end-use process must be modeled to sufficient accuracy. Moreover, the modeled end-use process must be identifiable from the available data. In this work, the focus is on the uncontrollable hot-water extraction from the tank of an electric water heater. It is relevant and of interest to independent system operators (ISO) since water extraction is not usually measured and only metered interval consumption data (kWh) is collected. This is achieved by designing an estimation strategy based on a stochastic model of the end-use consumption.