Optimization Models And Mehtods Of Active Power Dispatching And Scheduling In Developing Power Markets

In developing power markets, both inherent technical features of power systems and new features brought by power markets should be taken into consideration in the active power scheduling and dispatching of power systems. On the one hand, the operation of power systems is still of monopolistic features in their marketing process; on the other hand, the market pattern is not robust enough, and there are not enough categories of power merchandises for being traded. Furtherly, due to their different generating costs, fuel consumption levels, and environmental effects, generating units have to be assorted into bidding units and non-bidding units. Those above-mentioned features have greatly increased the complexities of active power dispatching and scheduling, which makes it necessary to study some proper mathematic models and solving methods for the specific situation. Beginnig with that, this paper studies the problems of active power volume auction, the problems of dispatching and scheduling in the markets in which there exist both kinds of units, and the problems of controlling emission and energy consumption of generation.The bidding curves are piecewise and horizontal in the basic active power volume action of generation markets, and which limits its application a lot. In this paper, the bidding curves are extended as the piecewise and tilted form; and by comparing the auction mathematic model with that of the basic active power economic dispatching, the uniform one is set up of the basic auction and dispatching problems. Then a steepest descendent increasing-decreasing pair optimization method (SDIDO) is proposed to solve the problems, and it can get the global optimum of the problems when the bidding curves or the cost curves are monotonically non-descending. When the curves are not monotonically non-descending, SDIDO can only get the local optimum. A method is proposed to solve these problems, which combines evolutionary computation with SDIDO, and it could search the optimum globally with efficiency.The basic model has rarely been applied to practical production directly for it neglects many practical limits. Hence, more complex dispatching and auction problems are studied, grid transmission constraints and units'prohibited operating zones are considered in the study of dispatching, and grid transmission constraints and units'non-bidding zones considered in that of auction. A uniform model is built up by comparison, and a method based on GENOCOP, a kind of genetic algorithms, is proposed to solve it. It has better performance than other random searching methods and mathematic programming methods.A series of scheduling and dispatching plans are needed to carry out the results of auctions and tradings. In the medium-term scheduling of developing power markets where there exist both bidding units and non-bidding units, the active power volume of contracts is usually decomposed into each day, on which more detailed plans are based. The volume-decomposed model is set up, the objective function of which is to minimize the total production costs of non-bidding units. To make the model more compatible to subsequent tasks, each day is divided into heavy load period and low load period, and grid transmission limits are also taken into account by means of DC power flow equations.The day-scheduling objective of developing markets is also to minimize the total production costs of non-bidding units, and its model is built in this paper. Solving this model consists of two steps: unit commitment and dynamic economic dispatching. A heuristic method is used to get unit commitment, and an augmented feasible direction method is used to get the dispatching. The latter method can get the global optimum rapidly when the model is convex, which can solve multi-period dispatching problem with quadratic piecewise cost functions.More attention should be paid to effects on enviorments and fuel consumptions in developing markets. Based on the foregoing studies, not only environmentally constrained/oriented dispatching, but fuel-consuming-minimized and fuel-consuming-constrained auctions are studied. The emissions of generators with lower fuel consumption and higher efficiency could be controlled by charging their emission, and the fuel consumptions of those could be controlled by settting limits of their marginal fuel costs. The emission of genetators with higer fuel consumption levels and lower efficiency could controlled by setting limits of their marginal or average emission, and the fuel consumption could be controlled by multi-objected programming auction. This research work is supported by Heilongjiang Nature Science Foundation of China under Grant E0326.