现代电力系统可靠性的问题外文原文.doc

Modern Problems of Electric Power Systems ReliabilityG. F. Kovalev, D. S. Krupenev, and L. M. LebedevaMelentiev Energy Systems Institute, Irkutsk, Russia Received August 20, 2009AbstractWe briey characterize certain reliability problems for power engineering systems relevant at present, on the liberalized power market. We show a reliability problem setting for these conditions and its implementation in a computational complex.DOI: 10.1134/S00051179100701791. INTRODUCTION During the transition from classical market conditions, when the main technological links of electric supply systems (ESS) are vertically integrated, to the so-called liberalized market of energy and power, when the vertical integration is destroyed and reformed into horizontal integration to make the competition between both energy suppliers and consumers more erce, the following reliability problems have surfaced: providing, in a competitive environment, primary power resources to all types of power plants, primarily heat power plants working on organic fuel. In this regard, one has to account for the oscillations of water inow in a hydroelectric plants reservoir; establishing the necessity and a rational size of the so-called commercial (market) generator power reserve; decomposing an EES system reliability into basic components: the generating link reliability and the network link reliability.Solving these problems requires developing new or modifying existing computational instruments for system reliability estimation that have to work under new industrial conditions. Below we give a brief description of how the above-mentioned problems apply for the reforming Russian energy market and offer approaches to solving them on the basis of the modied computational complex(CC) AMBER.2. PROBLEM DESCRIPTION(1) Accounting for restrictions on primary resources when estimating reliability presupposes a choice of the strategy and tactics of spending these resources. The strategy should be reected in the models algorithm. However, apart from certain special cases there are no exact principles for such a choice, and in practice a multitude of strategies are employed. This diversity is caused not only by objective reasons (e.g., the actual hydroelectric plants share of a daily, weekly, season, or annual system regulation, the fuel supply for a heat power plant, a larger uncertainty in forecasting future conditions of EES operations), but also by subjective factors (personnel qualications, values of a systems leadership and regulations of the regional power committee, current prots, and waysof limiting the consumers: reducing the frequency, power, “letter” restrictions, rolling blackouts etc.).We propose to implement a simple, but, in our opinion, efficient way to account for restrictions on primary power resources by specifying their volumes for each kind and for different power plants(water in a hydroelectric plants reservoirs, fuel on a heat power plant). In general, the volumes are specied in kilowatt hours by their forecast values (expectations). During EES imitation modeling,we determine the expenditure of all kinds of primary power resources when we optimize computation modes. Upon passing from mode to mode, power resources are stored and compared to predened limits. If, for a certain kind of power resource, its expenditure reaches its limit at some point, then we drop the corresponding power from the power coverage in what follows.Knowing the practice of certain Russian power systems, we can attest that this is what is usually done in practice. The only exception are hydroelectric plants with daily regulation (their powers are small) and large hydroelectric plants that have to provide for sanitary tolerances and tail-bay on a certain level (the power of these tolerances is also relatively small), i.e., we can disregard the peculiarities of these hydroelectric plants. Constraints can be specied not for the whole computation period (a year), but over certain intervals of the year (up to monthly values of they are available). In our opinion, this simple algorithm, despite being a little articial, virtually exactly captures the integral situation with providing EES with primary power resources. In this model, we can estimate how much of a certain resource we need. We can also account for uncertainty in the forecast data on primary power resources by performing several computations for dierent values of the forecast. If we know the probability of a certain level of primary power resources supplies, we can dene the distribution function of the power coverage value as a function of how well the plants are supplied with fuel (heat power plants) and water (hydroelectric plants).(2) Consider the problem of choosing the generator power and energy reserves. On one hand,each separate power supply company should have a certain minimum level of its own reserves; on the other hand, the overall reserve level should exceed the computed levels for technological reserves (reserves for current, capital, and intermediate repairs, modernization reserve, operational reserve).The minimal self-reserving level is dened by the existing standard of power supply reliability for the consumers in case the power supply company works on its own. These standards are known 1 (for example, in the U.S. P 0.9998, where P denotes the probability of faultless system operation 2).In Russia, this standard is still under construction, and existing recommendations vary in the range of 0.99910.9997.In power supply industry, when moving to the market economy, the most important factors shift from technological risk of failing the basic functions related to equipment failures towards economic (nancial) risk, since on a competitive market each economic activity is risky since one cannot exactly forecast and account for the conditions under which the proposed activity will be carried out.The problem, thus, is to nd ways to neutralize or reduce the negative consequences of nancial risk. To prevent these consequences, we propose to create an additional reserve (in addition to the technological reserve) in an EES, called “nancial reserve.”It 3, it is shown that the commercial reserve in energy should equalwhere pr is the rate of return related to the costs, and Ed is the power demand.The commercial power reserve iswhere Ty is the number of hours during which the powers are used over a year.It is a pity that the liberalized Russian market of energy and power has not yet accepted theserecommendations.(3) Decomposing system reliability. Under a horizontally integrates EES control system, among many reliability-related problems the most important problem is to estimate how much each of the EES technological links contributes to system reliability. These links include providing the power plants with primary power resources, generating the power, transporting the energy and distributing it. Solving this problem is necessary to dene each companys responsibilities in the market environment and reconcile reliabilities of individual EES elements. Among these elements, the most important and most interesting are the links of the basic EES structure, namely the generator link and the power transport link (network link) 4. To solve this problem, we have developed a decomposition method for system reliability. This method consists of the following operations.(A) Reliability computation for the basic EES structure (system reliability) with real life data, i.e., with the data corresponding to the real EES. Find the faultless operation probability Psys, the power underdelivery expectation Esys, and the supply coecient of power consumers sys.(B) Compute the reliability for the basic EES structure under the assumption that the network link is perfectly reliable. This computation lets us estimate the share of the generator link in system reliability. We also compute the same reliability parameters as in experiment A: Pgen, Egen, gen.(C) Compute the share of network link with the following formulas:Psys = Psys/Pgen;Enet Esys Egen;net = sys gen + 1,where Pnet, Enet, net are reliability parameters for the network link.These formulas have been obtained under the assumption that the EES is a chain of sequentially connected links (generator link, network link, power consumers), and that system reliability parameters are divided based on the fundamental assumptions of reliability theory.(D) Analyzing the obtained parameters. This analysis allows for nding a less reliable link in the EES and begin solving the reconciliation problem for basic system technological links reliability.3.MODEL DESCRIPTIONCC AMBER has been developed in the Melentiev Energy Systems Institute (ISEM) SB RAS to estimate reliability, in the sense of faultless operation and repairability, for large complex EES represented by any (radial, circular) multinode computational circuit with limited throughputs for connections between the nodes. The problem is being solved within the development management and long-term operation planning on the levels of the United, integrated, and regional EES.The problem is solved with imitational modeling of an EES during the computational period (a year). The load and system equipment states are sampled with the Monte-Carlo method. Beforehand, power distribution functions of generating apparata in the nodes and maximal throughputs of the connections are sampled. As a generating distribution function for this computation, the binomial distribution is used. Optimizing computational model is done with a predened strategy of distributing power decits in the system by power nodes. The maximal parameters o a computational circuit are 100 nodes and 160 connections. A more detailed description on CC AMBER can be found in 5. Due to the new approaches to EES management in Russia, AMBER makes an attempt to account for the characteristic features of Russian EES. Said features, rst of all, signicantly change the optimization of system computational modes.The work 4 lists basic principles of EES mode optimization for analyzing and synthesizing reliability and approaches to solving certain problems. The model shown below, besides technical and economic characteristics of power production and transport industry, accounts for the features of power supply markets on the regional and federal levels, since these features signicantly inuence power decit distributions and optimizing decitless modes.Organizational principles for wholesale energy markets dier across countries and power companies. The basic difference is the way to set prices (taris) on the power, i.e., internal power plant tariffs, tariffs on the regional wholesale market of power plant unions, and taris on the federal wholesale market within the united EES of the entire country may dier.4. MATHEMATICAL PROBLEM STATEMENT AND A PROPOSED SOLUTIONThe computational EES circuit is a directed graph with M nodes (vertices) and N edges (arcs) between the nodes. The mth node is characterized with the required load level and the operable generator power level , and the nth edge is characterized by the throughputs in the forward and backward direction and respectively, and also the loss coecient Kloss n.We denote the load power provided by the coverage at the mth node by Pml , the generator power of participating in load coverage by Pmg , and the ow through the nth edge with Pn.The optimization problem functional for each (decit and non-decit) system mode looks like the following:under the following constraints:Where is the power deficit at the mth node; is the excess of generator power at the mth node;amn are elements of the connection matrix A;bmn are elements of the matrix B such thatN0m is the set of connections by which the mth node sells (buys) the power from the wholesale market of a higher level than the local (internal) market.The functional (1) has the following coefficients of technical and economical nature:cm is the price of electric power sold to the consumers at node m (roubles/kWh);ce is the price of electric power on the wholesale interregional market (roubles/kWh);dm are the unit costs of generating 1 kWh of electric power at the corresponding node (roubles/kWh);gm is the cost of fuel necessary to generate 1 kWh (roubles/kWh);fm is the unit loss denoting compensational expenses or fines due to power underdelivery (roubles/kWh).Expression (1) for the functional has a peculiar feature: the dimensionalities do not match, since unit costs per kilowatt hour are multiplied by the power (kW). The lifetime of this mode, common for all components of the functional, is left out of the expression and is determined outside this optimization unit in the CC.In the general case, finding the optimal solution is possible for real life relations between the functional coefficients: fm cm, cm dm, dm gm ,in most nodes of the system.To solve this problem, we have used the interior point method 6 which is sufficiently flexible, efficient, and possesses sufficiently rapid convergence properties for problems of this kind.In this problem statement CC AMBER lets us solve problems specified above. We have studied consumer power supply reliability both in relation to the supplies primary energy resources and together with the supplies of generating power.CC AMBER is also well suited for estimating the necessary level of power reserves in each regional EES if it works in isolation for a given P.To estimate commercial reserves, AMBER uses and and values of balancing characte-ristics:andwhere Erec is the forecasted energy demand; is the installed generating power; is the regular annual power maximum; RT is the technical reserve; is the unused generating power in the system.The paper 5 presents first results of decomposing system reliability on the example of one energy system, which prove the concept of our proposed method. Thus, this model allows one to compute reliability taking into account specific market factors. Besides, the model can be used for a comparative analysis of various methods of organizing energy markets in EES given the possible system modes and various strategies for minimizing power deficits 3.5.CONCLUSIONAll of the above leads us to the conclusion that reliability estimation in the electric powerindustry remains an important factor for managing its development and operation. Market factors impose additional restrictions on the EES reliability computation.The AMBER model, which we present for the specialists here, can be used to solv