Search

This paper presents a novel model for optimization of investments in new power generation under uncertainty. The model can calculate optimal investment strategies under both centralized social welfare and decentralized profit objectives. The power market is represented with linear supply and demand curves. A stochastic dynamic programming algorithm is used to solve the investment problem, where uncertainty in demand is represented as a discrete Markov chain. The stochastic dynamic model allows us to evaluate investment projects in new base and peak load power generation as real options, and determine optimal timing of the investments. In a case study, we use the model to compare optimal investment strategies under centralized and decentralized decision making. A number of interesting results follow by varying the assumptions about market structure and price response on the demand side. Index Terms--Centralized and decentralized decision making, generation expansion planning, real options, restructured power systems, stochastic dynamic optimization.

In Part I of this two-part paper, we introduce several possible methods for integrating wind power, price-responsive demand and other distributed energy resources (DERs). These methods differ with respect to information exchange requirements, computational complexity, and physical implementability. A novel look-ahead interactive dispatch that internalizes inter-temporal constraints at the DERs level, and dispatches the results of distributed decisions subject to spatial security constraints, is proposed as a possible effective algorithm. This method requires only the use of today's static security-constrained economic dispatch (SCED) by the system operators. The optimization accounting for inter-temporal constraints, and ramping rates in particular, is done by the DERs while they create their own supply and demand functions. To implement this method, today's supervisory control and data acquisition (SCADA) needs to be transformed into a multi-directional, multi-layered information exchange system. [ABSTRACT FROM PUBLISHER]

In Part II of this paper, we demonstrate on a modified 24-bus IEEE Reliability Test System (RTS) that it is indeed possible to manage large-scale intermittent resources in coordination with price-responsive demand. The algorithms introduced in Part I of this paper are simulated assuming 20% and 50% wind capacity. We show that the look-ahead dispatch of the proposed method is physically implementable and, when used with elastic demand, can accommodate the integration of close to 50% wind capacity. This is contrasted with the observation that, without elastic demand, 50% wind integration is not physically implementable. From the perspective of total market surplus, the predictive dispatch of the proposed method is 3.8% suboptimal when compared to the centralized most efficient predictive economic dispatch. Further simulations are needed to test the feasibility of coordinating wind power and price-responsive demand on realistic large-scale power systems. [ABSTRACT FROM PUBLISHER]