Vous êtes sur la page 1sur 6

OPTIMIZATION METHODS IN POWER SYSTEMS APPLICATION AND PERSPECTIVES Xisto Vieira Filho Luiz Alberto S.

Pilloto Srgio Granville Eurico Salgado Sobrinho Boris G. Gorenstin Joo Carlos O. Mello Dourival S. Carvalho Jr.

CEPEL Centro de Pesquisas de Energia Eltrica Brasil

Fundao Pe. Leonel Franca Brasil

ABSTRACT - In this paper it is presented a modern Optimal Power Flow program developed at CEPEL whose optimization engine is a customized implementation of a direct interior point method based on the primal dual algorithm. The examples presented in the paper showed the effectiveness of the program in a wide variety of applications and its adequacy to deal with the new challenges represented by the restructuring of the electric utility industry. KEY WORDS - Electric Utility Restructuring, Deregulation, Optimal Power Flow, Ancillary Services, Voltage Collapse, Interior Point Methods. 1. INTRODUCTION The worldwide restructuring of the electric utility industry with the introduction of new paradigms such as competition in generation, the idea of electricity as a commodity, contracts for power exchanges, future electricity markets, ISO operators and so on [1-3], will imply in a great impact on existing network power transactions and will create an entirely new set of transactions among utilities, end users, power aggregators, independent power producers, etc. In this scenario, tools for power system optimization to increase network performance will be of great importance. In this paper it is presented the main aspect of a major tool for power system optimization which is the Optimal Power Flow. The original formulation of the OPF [4] was motivated by the desire to minimize generation cost in predominantly thermal systems or to minimize active power system losses in the network. However with the new challenges in the electric power industry a set of

entirely new applications of OPF are envisaged as will be described here. After presenting the basic aspects of a modern OPF program developed at CEPEL [5], it is described an optimization methodology which can accommodate all these aspects in a very effective way. Next a selected number of applications are shown to illustrate the power of the new tool.

2. AC OPTIMAL POWER FLOW PROGRAM

The Optimal Power Flow (OPF) problem consists in determining the state of an electric power system which optimizes a given objective function and satisfies a set of physical and operating constraints. Besides the electrical network the OPF is composed of an objective function, a list of controls than can be optimized and a set of operating constraints that should be obeyed at its optimal solution. Traditionally the most common objective functions specified in OPF applications were: Active Power Generation Cost Minimization. Active Power Deviation Minimization. Active Power Loss Minimization. Minimum Load Shedding to Correct System Operating Violations. Minimum Interchanging Deviation. Minimum Number of Changing Controls to Correct System Operating Violations.

In additions to the above a modern OPF should have the flexibility for the user to specify the new types of objective functions such as: Load Maximization at a Subset of Network Buses for voltage collapse analysis or to determine the maximum load which can be safety delivered at a given bus or set of buses to define contracts with large electricity users; Maximum Active Power Transfer Maximization Among Areas for inter-utilities transactions; Active Power Transfer Maximization Between Any Pair Of Network Buses to define maximum wheeling transaction capability; Minimum Load Shedding to avoid voltage collapse; Reactive Power Generation Cost Minimization to maximize reactive reserve for network disturbances; Active Power Injection Cost Minimization as a tool to determine the best a site to install a given generator to supply a load; Also it is well known that the installation of reactive support equipment has a great impact in network performance and the cost of these devices are low compared with active related objective functions. Then it makes sense to combine some of the operation related objective functions above with investment in reactive support equipment such as shunt devices and series capacitor. With respect to controls the following options are considered: Active Power Dispatch; Generator Voltage Levels; Reactive Power Generation; Taps of LTCs; Phase-Shifter Angles; Shunt Susceptance of Capacitor / Reactor Banks; Series Capacitor Reactance. With respect to constraints, besides bounds on control and state variables, there are the functional constraints as circuit power flow limits in MW or MVA, constraint on power factor at area interconnections or constraints on the sum of active power flow at a subset of circuits specified by the user. 3. OPTIMIZATION ALGORITHM
The optimization methodology chosen to accommodate in a very effective way all aspects of the OPF program discussed above was a customized implementation of a direct interior point method based on the primal dual algorithm. The OPF problem can be mathematically stated as:

Min

g(z)

(1)

s.t. h(z) = 0 a zb (1.1) (1.2)

The first step in the application of the primal-dual algorithm is to incorporate constraints (1.2) to a logarithmic barrier function: Min g(z) log(z j a j) log(b j z j) (2) j j

s.t. h(z) = 0
where is the barrier parameter. The basic idea of the algorithm is to solve approximately problem (2) for each value of and force go to zero; at the limit, the optimal solution of problem (1) is obtained. For each value of one iteration of the Newton-Raphson algorithm is applied to the nonlinear system of equations derived from the optimality conditions of problem (2). A crucial point in the method is the control of the primal and dual variables in its iterative process. Figure 3 shows the basic steps of the interior point method applied here; for further details see [6], [7].
Start

(2.1)

InitializePrimal and Dual Variables

Converged ?

Yes

Stop

No Compute and Solve Sistem of Linear Equations

Compute Maximum Steplenghts for the Primal and Dual Variables

Update Primal and Dual Variables

Update Barrier Parameter

Figure 1 - Primal-Dual Direct Interior Point Method

The interior point algorithm described above belongs to the class infeasible interior point methods [8] and one of its important characteristics is the fact that feasibility is attained together with optimality in the iterative process i.e. at intermediate iterates only the bound constraints are enforced by the logarithm barrier function. This means that the optimization does not require a starting point which satisfies the power balance equations. This aspect is important for voltage collapse analysis where the OPF program has to deal with unsolvable power flows.

4. EXAMPLES OF APPLICATION

The CEPEL OPF program just described has been used as successful tool for the solution and the evaluation of o broad of problems concerning power system planning and operational planning. Some of these application are presented next to shown the power of the new tool. 4.1 Thermal Generation at Fortaleza The fast load growth in the metropolitan area of the city of Fortaleza located in the northeast of Brazil, associated with the implementation of a new industrial plant in that region, led to the evaluation of local thermal generation as an alternative to the reinforcement of load supply from the main national power grid. That part of Brazil has no expressive power generation, and hydrogenerations are located hundreds of kilometers far a way and connected to it through several 230 kV transmission lines. Some of those 230 kV lines are on their thermal limit for heavy conditions. There is also a project to extend to Fortaleza the 500 kV network in a near future. A study carried out with the CEPEL OPF Program considering the scenarios from 1998 until the year of 2003, resulted with the recommendation to inject 240 MW near Fortaleza at the Pecem substation. The thermal energy cost for the new power plant was a function of the energy cost of the remainders power plant generation of the system. The study reveled the network bottlenecks, i. e. the network elements which imposed limits on the power flow. Prospective investigation also were carried out, as an comparison between different location for the thermal plant, and losses evaluation. In the study, the northeast Brazilian power system was simulated considering the power plants generation limits range, and transmission lines and transformers thermal limits. The considered objective functions for the optimal power flow were the Active Power Injection Cost Minimization and the Reactive Power Injection Cost Minimization. A range of thermal generation costs was taken into account. 4.2 Brazil-Argentina Electrical Interconnection The development of an electrical interconnection between Argentina and Brazil was studied to deliver 500 MW rated power to the Brazilian network electrical system in Rio Grande do Sul. Necessary frequency conversion will be accomplished by a HVDC Back-toBack station located on the Brazilian end of the transmission interconnection. Two alternative points of

connection to the 230 kV network, located at about 200 km from each other, were investigated in the study. CEPEL has performed an analysis on the southern Brazilian interconnection ac network considering the proposed project. For each supply alternative, evaluations considered peak and light load scenarios forecasted for the years 1999 and 2004, for a total of 8 distinct operational base case conditions. The steady state study initially focuses on preliminary investigations, in order to evaluate the impact of 500 MW additional power injection on the local 230 kV power system. From these evaluations, the necessary reinforcements to draw this power were determined for each scenario, in such a way that no violations resulted in thermal limits of the transmission lines and transformers, or in the admissible operational voltage range. In the study sequence, contingency analysis was performed, and if required converter station run-back was determined for each studied case, in order to avoid any system violations. Operating measures and system upgrades were also investigated to minimize the amount of converter run-back power required. For this study, run-back means the amount of power subtracted from the 500 MW HVDC Back-to-Back converter station, i. e., the amount of power the converter station is obliged not to convert in order to fulfill the network power system requirements. The simulations performed with the OPF program considered the objective functions Active Power Generation Cost Minimization and Reactive Power Injection Cost Minimization. The Brazilian South electrical system operational limits of voltage and transmission line/transformer branch thermal ratings were taken into account. The power flow simulations considered power injection alternatively at one of the substation at the Rio Grande do Sul network, and the power balance was achieved by allowing adjustment of generators in the Brazilian Southeast electrical system.
4.3 Voltage Collapse

Voltage Collapse is a kind of dynamic behavior but under certain conditions it can be related to the solvability of power flow balance equations. Standard power flow programs are not suitable to be used as tool for such kind of analysis because the jacobian matrix associated to the power flow balance equations is singular at the solvability frontier of the power flow.

This implies in algorithm convergence problems. Recent experiences developed at CEPEL have shown the great robustness of OPF program for this kind of analysis [7], [9], [10]. 4.4 Occurrences of April /97 Occurrences registered with the interconnected south/southeast Brazilian power system network during two days in April /97 resulted in slow voltage degradation and in the shedding of significant amount of load. A task force group established at ELETROBRS with the CEPELs support, analyzed the phenomena and considered static simulations the CEPEL OPF program. Load-flow convergence was very difficult using regular load-flow programs, due to the very severe network conditions. One of the main objective of the task force was to give a fast indication of the gravity of the situation, in terms of existing margin for load growth. This should be an element to design future system operation rules. Considering the data recovered for an situation registered minutes before the main occurrence of one of the April days, a margin for the power system entire load to growth was calculated with the OPF program. The result gave some indication of how near of a voltage collapse the system was. The analysis considered also the effect of certain contingencies in the calculated margin. For those simulations the objective functions considered was Load Maximization. 4.5 Reactive Power Cost Allocation in a Competitive Framework In the new power industry scenario with a regulated transmission system and competition in generation, it is likely that the transmission system will be primarily responsible to ensure adequate system-wide reactive support. To accomplish this task it should buy reactive power from the reactive providers and the cost incurred in acquiring this ancillary service [11] should be somehow shared by the transmission access agents as illustrated in Figure 2.
Generators ~ Transmission System ~ ~ ~ ~ Loads

Figure 2 - Reactive Support Transactions

Several cost allocation rules may be devised but in this context the Aumann-Shapley cost allocation scheme [12], [13] is the most appropriate because it is the unique cost allocation method which recovers the original costs (revenue reconciliation), is additive, weakly aggregation invariant, monotonic and it is fair in the sense that it does not imply in any cross subsidies among transmission access agents. Let n the number of power transactions in the transmission system, P = (P1, P2, ..., Pn) a vector which represents the amount of power transactions and C(P) the reactive support cost as a function of P. In the Aumann-Shapley cost allocation scheme the amount associated to the i-th transaction is: ci = Pi x
1 C ( tP ) dt t =0 Pi

(3)

This integral in (3) is computed numerically taking a discretization in the variable t in the interval [0,1]. For each value of t the OPF program is solved with the objective of computing the minimum cost to supply the reactive support for the corresponding transaction amounts, for more details see [12]. 4.6 On-Line Optimal Power Flow Power System engineers are currently facing new challenges to increase the transmission capability of existing networks due to financial and environmental restrictions. Therefore, power systems are being operated closer to their transfer limits and should be carefully monitored and controlled to avoid voltage, electromechanical and control stability problems. Control centers play an important role to guarantee a safe and stable operation of the power grid. SCADA systems were designed over the last ten years without paying attention to details of the ongoing market reform. The basic structures of SCADA systems are not expected to suffer major changes. However, generation and transmission scheduling on the new environment will be different from the current practice and will certainly impose certain requirements on the SCADA system. Control centers will have to carry out a large number of generation and transmission scheduling operations along the day and this will dictate an increase of performance requirements of actual systems.

Reactive Providers

An essential element of the SCADA/EMS environment is the optimal power-flow program. The CEPEL OPF program was designed to act as the heart of the network analysis package being incorporated to CEPELs SCADA system. The OPF program is essential for contingency analysis and for the generation of emergency control actions when connected to the State Estimation package. The new control centers being developed nowadays make use of an information structure called OASIS Open Access Same Time Information System. One of the most important informations posted by OASIS systems is the ATC Available Transmission Capacity. The calculation of ATC is an important addition to EMS environments since it will have to be based on a special version of an optimal power flow capable of calculating transmission capacity margins. The CEPEL OPF program is already being modified to perform on-line calculations of the capacity margin of the systems. The ATC calculation will have to be backed up by on-line transient (TSA) and voltage stability analyses (VSA). The on-line dynamic security assessment tools will be critical for assessing the potential impact of next worst contingency, as system operates through multiple unplanned contingencies. Another problem that will have to be addressed by the new generation of network analysis software of control centers is the ability to solve loop flows on neighboring systems caused by use of contract path scheduling. Expert software should be designed to identify the source of loop flows, and be capable of curtailing flows that threaten reliability, compensating transmission owners for the transmission assets used. Transmission access costing using OPF programs is another addition to conventional SCADA/EMS centers. These centers now are not only control centers from the energy point of view but are a vital point of the company to guarantee competitiveness on the market. The ability to calculate transmission costs in a fast and correct way for posting on OASIS systems may have a strong impact on the utility capacity to get contracts. At CEPEL, loop flow analysis and transmission access costing are two major areas being currently under development using the OPF program code as a starting point. Real-time Dynamic Security Assessment is probably one of the great enhancements that control centers will have to incorporate. On-line voltage stability analysis and transient stability analysis will be a must, at least at the ISO, since this entity will be responsible for the

reliability of the Brazilian Power System as whole. Online voltage and transient stability indices will provide the centers operator additional measures of the stability of the system, apart from those already given by conventional contingency and emergency analyses. The use of CEPEL OPF program at the voltage stability level of the dynamic security assessment is foreseen. 4.7 Other Applications of the OPF Program The CEPEL OPF program was installed at the Brazilian National Control Center (CNOS), it was distributed to other ELETROBRS Departments such as DOL, and to some Brazilian utilities such as FURNAS, CHESF, ELETRONORTE, ELETROSUL and CEMIG. At the CNOS the program is used (in an off-line environment, at the moment) by the dispatch engineers in the analysis of the system operating conditions. Besides the CNOS, the OPF program is currently applied by the Brazilian utilities in the following contexts: In DOL/ELETROBRS - operation planning for the interconnected south/southeast Brazilian system implying in great savings on the planning activities effort; In FURNAS - as tool for the analysis of difficult power flow cases; In CHESF - optimal allocation of new series compensation devices, optimal sitting of thermal new plants and maximum power transfer studies between the north and northeast Brazilian regions; In CEMIG - computation of system maximum loadability for a set of network buses to define new contracts with large electric power users; The OPF program was also successfully applied to the maximum active power transfer among network areas in a probabilistic context [14]. 5. PERSPECTIVES Optimization formulations have many advantages and possibilities, and there is an enormous perspective for its use in nowadays and in the future. Even though it is not possible to solve all problems using optimization, the practical examples presented in this article illustrates the variety of possibilities. This is specially true for a complex power system like the Brazilian, where the conventionally load-flow programs are able no more, in many situations, to converge to an solution. The OPF program can goes further and solve many of these unsolved cases. The contingency analysis of our interconnected power system endorses the differences between the two methodologies. This is specially true, also, in view of a deregulation electricity

market. The OPF methods will be challenged even more by requirements from the power system planners and operators, as an optimization problem is the problem of finding the best solution to a given question. For a large and routinely use of the methodology, however, it is necessary to overcome some sort of difficulties which includes, the quality and amount of data and the need of higher education for the personals involved. 6. CONCLUSIONS The worldwide restructuring of the electric utility industry will imply in a great impact on existing network power transactions and will create an entirely new set of transactions among utilities, end users, power aggregators, independent power producers etc. In this scenario tools for power system optimization to increase network performance such as the OPF will be of great importance. In this paper it was presented the main aspects of a modern OPF program developed at CEPEL whose optimization engine is a customized implementation of a direct interior point method based on the primal dual algorithm. The examples showed the effectiveness of the program in a wide variety of applications and its adequacy to deal with the new challenges represented by the restructuring of the electric utility industry.
7. REFERENCES [1] R.D. Tabors, Transmission System Management and Pricing: New Paradigm and International Comparisons, IEEE Transactions on Power Systems, February 1994 I.J. Perez-Arriaga, H. Rudnick, W.O. Stadlin, International Power System Trasmission Open Access Experience, IEEE PES Summer Meeting, San Francisco, CA, July 24-28, 1994

[6]

S.Granville, "Optimal Reactive Dispatch Through Interior Point Methods", IEEE Transactions on Power Systems, Vol.9 No.1 , February 1994. S.Granville, J.C.O.Mello, A.C.G.Melo, "Application of Interior Point Methods to Power Flow Unsolvability", IEEE/PES Transactions on Power Systems, Vol. 11, No. 2, May 1996. M.J.Todd, "Theory and Practice for Interior-Point Methods", Commentary in ORSA Journal on Computing, Vol. 6, No. 1, Winter 1994. Y. Mansour, E. Vaahedi, J. Tamby, S. Granville, Large Scale Voltage Stability and VAr Planning Studies Using Existing OPF/Optimal VAr Planning Tools. Paper presented in V SEPOPE, Recife, 1996.

[7]

[8]

[9]

[10] A.C.G. Melo, J.C.O. Mello, S. Granville, "The Effects of Voltage Collapse Problems in the Reliability of Composite Systems", IEEE IEEE/PES Transactions on Power Systems, Vol. 12, No. 1, February 1997. [11] D. Shirmohammadi, A.Vojdani, An Overview of Ancillary Services, paper presented at the V Symposium of Specialist in Electric Operational and Expansion Planning. Recife, Brasil, May, 1996. [12] X. Vieira F., S. Granville, M.V.F. Pereira, B.G. Gorenstin, J.C.O. Mello, A.C.G. Melo, Ancillary Services: How to Allocate Costs in an Optimization Based Framework, paper presented at the CIGR Symposium, Neptum, 1997.

[13] L. Billera, D.C. Heath, Allocation of Shared Costs: A Set of Axioms Yielding a Unique Procedure, Mathematics of Operations Research, Vol. 7, No. 1, Feb. 1982.
[14] J.C.O. Mello, A.C.G. Melo, S.Granville, Simultaneously Transfer Capability Assessment by Combining Interior Point Method and Monte Carlo Simulation, IEEE IEEE/PES Transactions on Power Systems, Vol. 12, No. 1, May 1997. .

[2]

[3]

CIGR SC-38 (Power System Analysis and Techniques), AG38.04 (Optimization Methods in Power Systems), TF 38.04.03, Methods and Tools for Transmission Costs, 1977. [4] J., Carpentier, "Contribution a l'tude du Dispatching Economique", Bulleting de la Societ Franaise de Electriciens, Ser. 8, Vol. 3, 1962. [5] S. Granville, L.A.C.Pereira, M.L.Latorre, M.L.G.Oliveira, Programa de Fluxo de Potncia timo FLUPOT, Manual de Usurio - V.4.0 - 11/97. Cepel, 1997.

Vous aimerez peut-être aussi