Search

The worldwide development of the natural gas industry resulted in an integration process between electrical and gas sectors in several countries. In Brazil, this process has been taking place in a consistent manner, especially on account of the increase in gas consumption for industrial use and of the installation of thermoelectric plants. Due to the predominance of hydro plants in electric power generation, thermoelectric energy production is basically dependent on hydrology and, as a result, presents a wide annual variability. Consequently, the investment applied to gas production and transportation infrastructure may become under-utilized during a large part of the time; thus, it is important to find mechanisms apt to improve its utilization. In this respect, the present work investigates the creation of a flexible market for gas, where contracts for flexible gas supply would be offered to industrial users, who would receive the gas assigned to thermal power plants when the latter are not dispatched, and would resort to an alternate fuel when these plants are dispatched. The attractiveness of such a contract would depend, of course, on its price. The purpose of this work is to develop a stochastic model for pricing flexible gas supply contracts, taking into account the uncertainty associated to the supply--dependent on the dispatch of the thermal power plants, which have the priority of use of the gas--and the risk profile of potential consumers. Index Terms--Electricity-gas integration, natural gas, power system economics, risk management.

The benefits of new transmission investment significantly depend on deployment patterns of renewable electricity generation that are characterized by severe uncertainty. In this context, this paper presents a novel methodology to solve the transmission expansion planning problem under generation expansion uncertainty in a min–max regret fashion, when considering flexible network options and $n-1$ security criterion. To do so, we propose a five-level mixed integer linear programming (MILP) based model that comprises: (i) the optimal network investment plan (including phase shifters), (ii) the realization of generation expansion, (iii) the co-optimization of energy and reserves given transmission and generation expansions, (iv) the realization of system outages, and (v) the decision on optimal post-contingency corrective control. In order to solve the five-level model, we present a cutting plane algorithm that ultimately identifies the optimal min–max regret flexible transmission plan in a finite number of steps. The numerical studies carried out demonstrate: (a) the significant benefits associated with flexible network investment options to hedge transmission expansion plans against generation expansion uncertainty and system outages, (b) strategic planning-under-uncertainty uncovers the full benefit of flexible options which may remain undetected under deterministic, perfect information methods, and (c) the computational scalability of the proposed approach. [ABSTRACT FROM PUBLISHER]

This work proposes a dynamic model to devise the optimal risk-averse investment policy in a portfolio of complementary renewable sources for a generation company in the Brazilian power system. The proposed method merges a static energy-contracting model, based on a hybrid robust-and-stochastic optimization approach, with a mean reverting binomial lattice model for real-option valuation. The proposed merge extends previous works by providing support to risk-averse investment decisions in complementary renewable sources dynamically distributed over time. The most important results of the model are: how much capacity to invest or build from each renewable source, how much to sell from the energy portfolio in bilateral contracts, and the optimal timing to invest. Unlike previous reported works, our model takes into account three classes of uncertainties simultaneously: renewable production of candidate sources and prices in the spot and contract markets. A case study with realistic data from the Brazilian power system is presented to illustrate the value of our model. [ABSTRACT FROM AUTHOR]

The objective of this work is threefold. We firstly present an optimization model for a price-taker hydrothermal generation company (Genco) to devise bidding strategies in multi-item iterative auctions of long-term contracts. The bidding model calculates a bidder's best-response function, which takes into account the key issues on the auctioned contracts, such as its time horizon, the risk factors that affect the future contract outcomes, interdependence between auctioned products, and the agents' risk profile. The risk profile of the Gencos are represented as piecewise linear utility functions and a practical specification approach is proposed. We finally present a simulator of a uniform-pricing and iterative multi-item contract auction, where the set of auction rules is implemented and the optimization model developed is used to determine the Gencos's best responses at each auction round. The simulator enables auctioneers and generators to estimate the auction's competitive equilibrium price and to study the auction dynamics. A real multiproduct descending clock auction is simulated for the Brazilian power system under the proposed bidding scheme. [ABSTRACT FROM PUBLISHER]

This paper presents a new approach for the contingency-constrained single-bus unit commitment problem. The proposed model explicitly incorporates an n - K security criterion by which power balance is guaranteed under any contingency state comprising the simultaneous loss of up to K generation units. Instead of considering all possible contingency states, which would render the problem intractable, a novel method based on robust optimization is proposed. Using the notion of umbrella contingency, the robust counterpart of the original problem is formulated. The resulting model is a particular instance of bilevel programming which is solved by its transformation to an equivalent single-level mixed-integer programming problem. Unlike previously reported contingency-dependent approaches, the robust model does not depend on the size of the set of credible contingencies, thus providing a computationally efficient framework. Simulation results back up these conclusions. [ABSTRACT FROM PUBLISHER]