Transmission capacity design and real time generation reserve management for electric power under uncertainty

This thesis proposes new models to perform transmission network expansion planning (TNEP) considering the economic dispatch, demand levels at different demand nodes and their duration and also the impacts of uncertainty. These models constitute a significant improvement over the traditional TNEP formulations that only consider one deterministic peak demand scenario at different demand nodes. The TNEP models presented in this thesis are of a non-convex mixed integer non-linear programming (MINLP) type. Two methods are proposed to solve these models: the first, through convexification of the non-linearities in the bilinear formulation of the TNEP problem, which is new; and the second, through the use of a multi-step simultaneous changes constructive heuristic algorithm (CHA). The TNEP models formulated in the thesis are solved using the proposed CHA. Most of the available CHA are limited to making one change in capacity of the transmission network at a time. The CHA developed in this thesis accelerates the search process by making multi-step simultaneous changes in transmission capacities. It also helps to avoid getting trapped in local minimas, because simultaneous changes lead the search over different neighborhoods of the set of feasible solutions.;The issue of cost recovery in the transmission sector is intimately tied up with capacity planning. This thesis proposes the use of Markov chains to allocate linearly varying transmission charges due to individual transactions by tracing the flows from the generators to the loads. The proposed method is simpler to implement than some other methods available in the literature based on the same principle.;Once the optimal capacities of generation and transmission are decided, real time fluctuations in demands are absorbed by having appropriate quantities of generation reserves. Most of the current procedures to determine generation reserves are empirical based. In this thesis a model is proposed for determining quantities of real energy and a number of generation reserves to meet random demands for a given dispatch interval.;The models presented in this thesis can have a positive impact in the power industry, since they allow decision making in an environment that is closer to reality.