Decision analytics in energy systems : from thermal grids to power grids

Abstract

The rapidly evolving energy sector has thrown open a plethora of research questions that broadly deal with the harmonization of energy generation, transmission, distribution and end use for achieving different sustainability objectives. In this thesis, we discuss a suite of such problems relating to evolving smart grids of today and tomorrow. Specifically, we apply tools from decision science and analytics such as modeling methods, control theory and optimization algorithms to answer the research questions posed for smart thermal grids as well as smart power grids.

In the fourth part of this work, we evaluate the potential for revenue maximization for stacked energy storage service providers with a combination of different storage units. Specifically, we discuss the conditions under which operating a conventional grid-scale storage system (like a pumped hydro storage system) might become uneconomical, especially in markets that require fast response (like regulation services market). We also evaluate the potential of a fast-responsive storage system (such as a Battery Energy Storage system or a flywheel), stacked with the conventional storage resource (in our case a pumped hydro storage system), to enhance the total revenue from participation in the energy market as well as the ancillary services markets. We study this problem using real pricing data taken from the California electricity market.

In the third part of the work, we shift our focus from thermal grids to power grids. Heating, Ventilation and Cooling (HVAC) systems constitute a significant portion of the power grid demand in USA. Optimal management of this energy consumption in the HVAC systems is hence, of utmost importance in the present context. We consider the electricity aggregator's (load serving entity's) problem of centrally optimizing the HVAC energy usage in buildings under its purview, with an aim of reducing the energy usage cost. While doing so, we ensure that the indoor zone temperatures of energy consumers in these buildings remain within some specific user-defined comfort bounds. We propose a low complexity optimal control approach to solve this classical research problem by optimally using the building thermal mass as a demand response resource. We show that under some assumptions, it is possible to express the cost-optimal precooling times for building HVAC units in closed form. We validate our algorithm using data from the electricity markets in New York and Texas.

Note that thermal grids, in practice, are hierarchical networks, i.e. they have multiple levels at which heat exchange takes place. For example, a higher substation level heat exchanger may connect several buildings as a cluster to the main DHC network, while each individual building has local heat exchange mechanisms controlled selfishly by the corresponding consumer. Demand side management by optimal tuning of parameters at the individual building level in such hierarchical networks can become computationally intractable and resource intensive. In the second part of this work, we leverage the hierarchy of heat exchange processes in thermal grids to propose an architecture named DReAM (Demand Response Architecture for Multi-level DHC systems). DReAM lowers the DR optimization complexity by a bi-level network decomposition and subsequently working only at the level of substations, that in turn, supply heat to groups of buildings. In a specific use case, we demonstrate how the DReAM framework can be used in a scalable manner to determine the optimal heat inflow to different substations for minimizing network wide operational cost, while respecting comfort constraints of energy consumers at the individual building level.

We first consider the problem of ensuring thermal fairness in thermal grids (also known as District Heating and Cooling (DHC) networks). Thermal grids are interconnected network of buildings, where hot (or cold) water is piped from an energy source in order to meet the consumers' space heating requirements. Selfish control of energy inflow in individual consumer premises under source energy inadequacy may cause a wide dispersion in consumer comfort, thus causing potentially unfair operation. Factors such as extreme climatic conditions, leakage and transportation losses and differences in thermal properties of buildings may exacerbate such issues. We propose a demand response (DR) scheme whereby the DHC utility manager can control the energy inflow to buildings centrally to optimize carefully chosen thermal fairness metrics. We also extend our study to cover conditions where the DHC utility manager is unaware of exact thermal parameters of the network elements (buildings) under its purview.