Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
1 Fast Decoupled Power Flow to Emerging Distribution Systems via Complex pu Normalization O. L. Tortelli, E. M. Lourenço, Member, IEEE, A. V. Garcia, and B. C. Pal, Fellow, IEEE Abstract-- This paper proposes a generalized approach of the per unit normalization, named complex per unit normalization (cpu), to improve the performance of fast decoupled power flow methods applied to emerging distribution networks. The proposed approach takes into account the changes envisaged and also already faced by distribution systems, such as high penetration of generation sources and more interconnection between feeders, while considering the typical characteristics of distribution systems, as the high R/X ratios. These characteristics impose difficulties on the performance of both backwardforward sweep and decoupled-based power flow methods. The cpu concept is centred on the use of a complex volt-ampere base, which overcomes the numerical problems raised by the high R/X ratios of distribution feeders. As a consequence, decoupled power flow methods can be efficiently applied to distribution system analysis. The performance of the proposed technique and the simplicity of adapting it to existing power flow programs are addressed in the paper. Different distribution network configurations and load conditions have been used to illustrate and evaluate the use of cpu. n Pcpu Qcpu RΩ Rcpu Sbase Scpu Spu SVA Vbase Vcpu Vpu XΩ Xcpu ZΩ Zbase Zcpu Zpu Index Terms— Distribution System, Complex Normalization, Decoupled Power Flow Analysis. I. NOMENCLATURE avg γavg base IA Ibase Icpu Ipu l Average R/X ratio, rad Average of the maximum and minimum R/X ratio, rad Power factor index Current angle, degree Volt-ampere power angle, degree Impedance angle, degree Base angle, degree Current, A Base current, A Complex normalized current, pu Conventional normalized current, pu Number of branches O. L. Tortelli, and E. M. Lourenço acknowledge the financial support of the Brazilian government provided through the Science without Borders – CAPES (Brazilian Post-Graduate Federal Agency). O. L. Tortelli, and E. M. Lourenço are with the Department of Electrical Engineering, Federal University of Paraná (UFPR), Curitiba, PR, Brazil (emails: [email protected], [email protected]). A. V. Garcia is an independent consultant in the area of real-time control of power systems (email: [email protected]). B. C. Pal is with the Electrical and Electronics Engineering Department of Imperial College London, London, UK (email: [email protected]). Number of buses Complex normalized active power, pu Complex normalized reactive power, pu Resistance, Ω Complex normalized resistance, pu Power base, MVA Complex normalized power, pu Conventional normalized power, pu Complex power, MVA Voltage base, kV Complex normalized voltage, pu Conventional normalized voltage, kV Reactance, Ω Complex normalized reactance, pu Impedance, Ω Impedance base, Ω Complex normalized impedance, pu Conventional normalized impedance, pu II. INTRODUCTION T HE Distribution System (DS) is widely impacted by the new technologies oriented by the Smart Grid concepts. In order to improve the system reliability, combined with the increasing addition of generation sources closer to the load (Distributed Generation concept), the distribution feeders tend to evolve towards a more meshed topology and to operate as an active network, with bidirectional power flows. Distribution System used to be operated as a passive network with radial topology, and thereby subjected to unidirectional power flows, i.e., from the feeding substation to the loads. These assumptions guided the development of analytical tools dedicated to support operational and planning studies of DS. Power flow calculation is the most frequently performed analysis in electrical power systems. The improvement of this essential tool taking into account of the evolution of distribution systems, but considering their peculiarities, such as the high R/X ratio of its lines, is relevant in the envisioned context of the emerging power system. This work is oriented by this context, and proposes a new approach based on a complex per unit normalization (cpu), which generalizes the conventional per unit normalization. The novelty lies in the definition of a complex volt-ampere basis to circumvent the problems caused by high R/X ratio 2 lines faced by distribution systems. The new technique is simple to demonstrate and to apply and provides the same state variables (complex bus voltages) as that obtained through the conventional pu normalization. Initial efforts reported in [1, 2] have stimulated the present research, aiming to consolidate the cpu approach and to demonstrate its application and advantages when applied to power flow calculations in emerging distribution systems. Simulations carried out with different test systems and distinct topologies and arrangements, including Distributed Generation (DG), are used to evaluate the performance of the proposed approach. III. DS POWER FLOW ANALYSIS – A BRIEF REVIEW In order to ensure secure operating conditions, two features are expected from a power flow analysis routine: speed and solution accuracy. Numerical methods based on NewtonRaphson (NR) [3] are the traditional tools to solve the power flow, due to its quadratic convergence characteristic that provides great accuracy with few iteration steps. However, the full NR method is computationally expensive, even when techniques to improve efficiency are applied [4]. Several techniques to overcome the computational burden of full-NR method have been proposed, usually taking advantage of the weak coupling between voltage magnitude and active power, and also between voltage phase angles and reactive power. Among these, the so-called Fast Decoupled Power Flow (FDPF) formulated by Stott and Alsac [5] became a standard, being considered the most efficient and fast method to solve the power flow problem. Though the fast-decoupled method is very effective when applied to transmission systems, it cannot be readily used to the load flow analysis of distribution systems. Decoupled load-flow algorithms face convergence problems when network branches have high R/X ratios [6], since much of the coupling between the active and reactive power in a circuit is due to the resistive component of the line impedance. Several attempts were presented in the 1980’s to overcome this limitation. An artificial series compensation method was proposed by Dy Liacco and Ramarao [7] and a parallel one by Deckmann et al. [8]. There are two main weak points in both compensation approaches: the structure of the network is altered with an expansion of the matrix size, and also the determination of a suitable compensation level to promote a consistent convergence rate is not an easy task. Rotation techniques to convert the branch impedances [9] and admittance matrix elements [10] into nearly reactive/susceptive values were also proposed as an alternative to overcome the high R/X ratios of the lines. Rajicic and Bose [11] followed by Wang et al. [12] presented a modification on the FDPF equations, but retaining the original structure of the network. The matrix of the active sub problem (B’) is modified by an empirical criterion and the equations of the reactive sub problem are also altered in order to improve the convergence. Nanda et al. [13] and Van Amerongen [14] have also reported variations in the FDPF sub matrices formulation and its consequences in the convergence properties. Results presented in [14] shown the superior performance on the socalled BX version of the FDPF for systems with high R/X ratio lines. Another approach, regarding slack bus selection to improve the convergence, was proposed by Singh et al. [15]. On the other hand and despite of the above efforts, the wellknown Backward-Forward Sweep (BFS) based algorithms, oriented by a passive and radial DS network, became the most popular approach for distribution power flow calculation. The first version of BFS was developed by Berg et al. in 1967 [16]. Since then, several variations and improvements on the method have been proposed. A solution for weakly meshed topologies where loops are converted into radial configuration by choosing appropriate break points was presented in [17]. However, in the emerging distribution systems, meshed and/or looped networks are expected even between more than one feeding bus and also with multiple DG sources, making DS an active network. In this scenario, the Backward-Forward approach is not the best option since it was not designed to solve meshed networks. This paper brings the power flow solution for networks with high R/X ratio lines into discussion under the impact of meshed topologies, DG and the increased automation level expected to DS operation. The axis rotation technique [9] is recalled to achieve this goal. A new interpretation of this technique, based on the well-known per unit normalization, is presented and discussed next. IV. COMPLEX PER UNIT NORMALIZATION Traditionally, the parameters and variables of the power system are normalized in per unit (pu) basis, where real values are chosen for the voltage and power bases. The complex normalization (cpu) proposed in this paper can be seen as an extension of the conventional one, in which a complex voltampere base is adopted, that is: S base S base .e jbase (1) The voltage bases, however, are maintained real and defined just as in the conventional pu normalization, that is, a different (but related) magnitude value is adopted for each voltage level: j0 V Vbase base Vbase .e (2) According to (1) and (2), the impedance base will be also complex, with the same base angle applied to the power base, that is: 2 Vbase Z Z base .e jbase base * S base (3) The complex normalized values of impedances can be easily determined by: .e j Z Z R jX .e j ( base) Z cpu pu jbase Z Z . e base base (4) 3 Equation (4) shows that the magnitude of the normalized impedance is the same regardless of which normalization have been adopted: conventional (pu) or complex (cpu). However, cpu also normalizes the angle of the impedance, which varies according to the base angle, as expressed in (4). The normalized values of resistances and reactances can be easily determined from (4): . cos( ) Rcpu Z pu base (5) .sin ( ) X cpu Z pu base (6) which leads to: X cpu Rcpu tan base (7) The relation stated in (7), clearly shows that R/X ratios can be adjusted by simply adopting an adequate value for the base angle, ϕbase. The technique is fully exploited to allow Newton’s decoupled methods to preserve their typical convergence rates even when applied to distribution networks. Once cpu normalization is employed, new values for network parameters and power injections are obtained in the same fashion as in conventional pu normalization. The complex normalized values of volt-ampere power can be obtained by using the complex power base, that is: S cpu S VA .e j S base .e jbase S pu .e j ( base) (9) .I* leads to (15) assuring Substituting (9) and (14) into S V that the state variables are the same regardless the normalization. V cpu S pu .e j ( base ) Vpu .e j ( ) V pu I .e j ( base ) Although cpu produces the same results of the impedance rotation technique [9], it brings the adjustment approach of R/X ratio for the new context of the DS network. It is important to emphasize that cpu consists of a generalization of the per unit normalization concept, in which the conventional pu is a particular case (when a zero base angle is adopted). Consequently, the same advantages obtained with the application of conventional per unit normalization holds when cpu is used, including the implications with transformers, such as the equivalence between the impedance seen from the high and low voltage level after normalization. This generalization also allows further application of FDPF, such as an efficient unified power flow analysis over power networks with large differences in its R/X ratios and voltage levels [18]. A. Determination of the base angle The strategy proposed in this paper is based on the evidence that, the lower the R/X ratio, the better the performance of the decoupled PF methods (since stronger is the P-QV decoupling). Therefore, the choice of the base angle should be guided by the original R/X ratio of the distribution lines. The first step of the proposed strategy is to find the average X/R ratio of the DS, as shown by (16): ∑ S cpu ( Ppu j.Q pu ).e jbase Qcpu S pu .sin base (11) V base I base .e jbase Z .e jbase base ( ⁄ ) (13) Therefore, I cpu I .e j A I .e jbase base I pu .e j ( base ) However, in some DSs, large differences are observed in the R/X characteristic. In such cases, the average value between the maximum and minimum X/R ratios of the DS can be computed by (17) and included in (18), as described next. ( ⁄ ) (12) In order to show that the solution of the power flow, that is, the bus complex voltages, are not affected by the cpu normalization, the corresponding current base is presented: I base (16) (10) so that: Pcpu S pu . cos base (15) pu (14) (17) The average of the two angles given by (16) and (17) provides a better condition to determine the base angle. As mentioned before, the determination of the base angle is guided by the perfect decoupling, which is implicit in (18). ( ) (18) Another influence in the base angle’s determination is the power factor of the loads. In order to counterbalance the effect of low power factor loads, an additional correction can be applied to the base angle, provided by (19): ∑ ( ( ⁄ )) (19) 4 Then, the base angle can be achieved by applying (16)-(19) as follows: ( ) (20) Equation (20) leads to a simple, but effective, way to determine an adequate base angle to cpu normalization. It is important to notice that for networks with low variation of R/X ratios and high power factor loads, the determination of the base angle becomes even simpler, as αavg ≈ avg and ε ≈ 0. Moreover, small deviations in the configuration (or loading) of the network usually do not produce large impacts on the base angle. Considering this, an exhaustive power flow evaluation of the network can be performed once, looking for the perfect base angle. Then, it can be kept constant as the base angle of that network. B. cpu application to power flow algorithm The proposed approach can be readily applied to existing power flow programs or even be incorporated into a dedicated power flow routine. It is important to emphasize that, as shown in Fig.1, when applied to existing programs the proposed approach does not affect the main power flow algorithm, since cpu normalization procedure just takes part in a pre- and post-processing stage. In this case, the final values of the active and reactive power flows can be easily obtained via reverse normalization, as shown in the flowchart. A. Two-bus test feeder The single line system in Fig. 2 is used here to illustrate the performance of the proposed normalization regarding the R/X ratio. With the line impedance magnitude kept constant, the impedance angle is changed in order to increment the R/X ratio. G z = 0.1 pu 1 + j0.6 pu Fig. 2. 2-Bus Test Feeder The plots in Fig. 3 compares the number of iterations verified by three different PF methods: the full NewtonRaphson (N-R), the XB version of the FDPF (FD-XB), the BX version of the FDPF (FD-BX), and the proposed FDPF with cpu approach (FD-CPU). It is straightforward to see from Fig.3 that the cpu normalization provides the FDPF method a NR similar behavior, allowing convergence even with very high R/X ratios, where the other versions of the fast decoupled method fails. All tests were performed using a 10-6 pu tolerance to power mismatches. Original line and bus data cpu normalization Fig. 3. 2-Bus Test Feeder results PF program Vcpu = Vpu (State variables) Scpu ≠ Spu (Power Flows) Reverse cpu normalization B. Base angle impact The impact of the base angle of cpu on the performance of the FDPF method (in terms of convergence) are evaluated here. The simulations, considering the BX version, are related to two different distribution systems: a small 12-bus test feeder, depicted in Fig.4 (whose data can be found in [19]) and the 95-bus feeder, shown in Fig.5, which is based on a real distribution network model of an UK utility [20]. Both test feeders considers the presence of DG sources, as indicated in the Figs. 4 and 5. Power Flows Fig. 1. Application of Complex Normalization V. SIMULATIONS AND RESULTS Several simulations were performed using different distribution networks, showing different aspects of the implemented methodology. Fig. 4. 12-Bus Test Feeder Fig. 6 illustrates the graphical evaluation of the number of iterations versus base angle of the complex normalization procedure applied to the two test systems. Apart from the base 5 case, two more R/X conditions are considered, by multiplying the resistance of the lines by a constant factor. It can be noticed that, for both cases, a significant improvement in the performance of the FDPF can be achieved with cpu, with at least 50% of reduction in the number of iterations when compared with the conventional pu normalization (zero base angle). Also, with a multiplier factor of 3, the conventional normalization is not able to reach convergence. In addition, the convergence performance is tested regarding different load conditions, as nominal load is increased by constant factors. The results, summarized graphically in Fig. 7, reinforce the positive impact of cpu on the FDPF convergence process. (b) Fig. 6. Convergence performance versus base angles considering different R/X ratios (a) 12-bus feeder; (b) 95-bus feeder 7 11 12 14 8 10 13 9 15 29 16 17 27 30 25 6 39 26 40 23 3 85 Transformer 33/11kV 84 31 32 37 38 47 42 41 43 44 45 46 48 56 58 50 55 57 52 22 20 53 21 Power System 33 34 35 36 24 5 28 1 2 4 49 51 54 19 60 18 61 DG 64 59 62 63 65 75 76 77 80 86 78 79 83 90 91 92 93 94 81 (a) 66 68 67 74 69 70 73 72 87 82 88 95 89 DG 71 Fig. 5. 95-bus UK Distribution Feeder (b) Fig. 7. Convergence performance versus base angles considering different load conditions (a) 12-bus feeder; (b) 95-bus feeder (a) Aiming to demonstrate the effectiveness of the proposed approach to base angle’s determination, presented in Section IV.A, Table I shows a comparison with the simulation results presented in Fig.6. As one can see, the 12-bus test system has low power factor loads (high ε), while the 95-bus feeder has lines with large variation of R/X ratio (αavg significantly larger than γavg). 6 TABLE I BASE ANGLE’S DETERMINATION TABLE III BRANCH IMPEDANCES 12-bus test feeder Branch 1x 2x 3x 71.2o 80.3o 83.5o 70.8o 80.1o 83.3o R (pu) 0.3254 best range 75o to 90o 95o to 105o 94.1o 110o to 120o 106.3o 110.5o 95-bus test feeder 1x 2x 3x 64.8o 76.8o 81.1o 49.0o 58.1o 64.3o 0.0315 best range Conventional Normalization 50o to 60o 65o to 75o 70o to 90o 58.7o 69.5o 75.0o The results in Table I clearly show that the base angle value obtained with (19) is within (or very close) to the best angle range indicated by the exhaustive simulation. In order to illustrate the effects of cpu application in the feeder’s data, the active and reactive power injections and the branch impedances of the 12-bus test system, along with their normalized values are shown in Tables II and III. Also, the resulting state variables, both, with conventional pu and cpu, are presented in Table III, demonstrating the absolute coherence between them, as expected. The application of cpu on a fictitious network may contain negative values of resistance and/or active loads, as shown in Tables II and III. It is inherent to the normalization procedure, but with no physical meaning. 1-2 2-3 3-4 4-5 5-6 6-7 7-8 8-9 9-10 10-11 11-12 X(pu) |R/X| Complex Normalization R (pu) X(pu) 0.038 2.37 -0.0442 0.0871 0.51 0.098 0.041 2.39 -0.0477 0.0949 0.50 0.173 0.072 2.40 -0.0839 0.1676 0.50 0.263 0.110 2.39 -0.1281 0.2547 0.50 0.090 0.038 2.37 -0.0442 0.0871 0.51 0.083 0.031 2.68 -0.0367 0.0806 0.46 0.361 0.100 3.61 -0.1249 0.3531 0.35 0.466 0.132 3.53 -0.1642 0.4557 0.36 0.239 0.068 3.51 -0.0845 0.2337 0.36 0.125 0.035 3.57 -0.0436 0.1223 0.36 0.102 0.029 3.52 -0.0360 0.0997 0.36 C. Topological considerations In order to demonstrate the robustness and efficiency of the proposed approach regarding topology aspects, additional arrangements of test feeders are considered in this subsection. In all of them, the selected feeder is replicated and the two resulting feeders are connected in three different ways. As presented schematically in Fig. 8(a), the radial arrangement consists of the series connection of the selected feeder by a transformer, emulating a two level distribution system. In turn, weakly-meshed topology is created by connecting the end of each feeder as a closed loop, as illustrated in Fig. 8(b). The meshed arrangement, in Fig. 8(c), arises from the inclusion of additional branches connecting the two feeders. (a) TABLE II ACTIVE AND REACTIVE POWER INJECTIONS Conventional Normalization Bus Complex Normalization 1 Pd (MW) 0.000 Qd(MVAr) 0.000 Pd (pu) 0.00 Qd(pu) 0.00 Pd (cpu) 0.000 Qd(cpu) 0.000 2 0.060 0.060 6.00 6.00 -6.404 5.567 3 0.040 0.030 4.00 3.00 -3.272 3.781 4 0.055 0.055 5.50 5.50 -5.870 5.103 5 0.030 0.030 3.00 3.00 -3.202 2.783 6 -0.025 -0.005 -2.50 -0.50 0.673 -2.459 7 0.055 0.055 5.50 5.50 -5.870 5.103 8 0.045 0.045 4.50 4.50 -4.803 4.175 9 0.040 0.040 4.00 4.00 -4.269 3.711 10 0.035 0.030 3.50 3.00 -3.237 3.282 11 0.040 0.030 4.00 3.00 -3.272 3.781 12 -0.023 0.005 -2.30 0.50 -0.338 -2.329 |R/X| 0.090 (b) (c) Fig. 8. Schematic arrangements applied to test feeders: (a) radial; (b) weakly meshed; (c) fully meshed 7 The performance of FDPF methods was evaluated by applying the proposed approach to the 12- and 95-bus test feeders, regarding the topological arrangements depicted in Fig. 8. The number of iterations when conventional pu normalization is applied in comparison with the adoption of cpu is presented in Tables IV and V, considering two load conditions: the nominal load and a 50% increased load. TABLE IV 12-BUS: CONVERGENCE PERFORMANCE Meshed Looped Radial pu Nominal Load Increased Load Nominal Load Increased Load Nominal Load Increased Load cpu XB BX XB BX 20 13.5 6 6 29 19.5 8 8 18 13.5 6.5 6.5 26 18.5 8.5 8.5 18 14 6.5 6.5 25 17.5 8 8 TABLE V 95-BUS: CONVERGENCE PERFORMANCE Meshed Looped Radial pu Nominal Load Increased Load Nominal Load Increased Load Nominal Load Increased Load cpu XB BX XB BX 12 12.5 7 76 18 20.5 9.5 10.5 11 11.5 6 5.5 13 15 6.5 7.5 9 10.5 5 5 13 12.5 6.5 6.5 Simulation results in Tables IV to VI clearly show that the number of iteration required by the proposed FDPF is significantly reduced when compared to conventional one (at least 50% less iterations). One can also notice that in some cases cpu is crucial to reach convergence. Moreover, unlike the conventional pu normalization, the cpu ensures a very good performance of the method when the system operates in a more stressed load condition. Such a comparison reaffirms the consistent improvement in the convergence process of the fast decoupled method by the cpu normalization. All results presented in this section demonstrates the positive impact that complex per unit normalization promoted in the FDPF, assuring a consistent improvement in its convergence process. VI. CONCLUSIONS This paper presents the concept of complex per unit normalization as a generalization of the conventional procedure, through the adoption of a complex power base. The influence of the R/X ratio and the load power factor in the power base angle is discussed and incorporated in the base angle’s determination. The resulting technique enables the application of Fast Decoupled power flow methods to distribution system analysis. The application of the proposed approach is fully explored in the likely scenario faced by distribution systems, including looped and meshed topologies, where backward-forward based algorithms are not a suitable choice. The implementation of cpu is very simple and effective, promoting a significant reduction in the number of iterations and computational effort of FDPF routines when compared with the conventional one. From the results and bearing in mind that the proposed approach can be coupled to existing power flow programs, cpu can be seen as a promising tool for active distribution system analysis, since it meets the new challenges brought by the pursuit of a smarter grid. Three additional test-feeders [18] are also considered in this simulation section. Table VI summarizes the obtained results. VII. REFERENCES [1] pu [2] 126-bus 69-bus 33-bus TABLE VI ADDITIONAL TEST FEEDERS: CONVERGENCE PERFORMANCE Nominal Load Increased Load Nominal Load Increased Load Nominal Load Increased Load cpu XB BX XB BX 14 10.5 4.5 5 27 14.5 6 6.5 14 11.5 4.5 4.5 nc 17.5 5.5 6 17.5 24 6.5 6 29.5 nc 10 9 [3] [4] [5] [6] [7] C. C. Durce, O. L. Tortelli, E. M. Lourenço, T. Loddi. “Complex Normalization to Perform Power Flow Analysis in Emerging Distribution Systems”. IEEE PES ISGT Europe, Nov. 2012. E. M. Lourenço, T. Loddi, O. L. Tortelli, “Unified Load Flow Analysis for Emerging Distribution Systems”. IEEE PES ISGT Europe, Oct. 2010. W. Tinney and C. Hart, “Power flow solution by Newton’s method,” IEEE Trans. Power App. Syst., vol. PAS-86, no. 11, pp. 1449–1460, Nov. 1967. W. Tinney and J. Walker, “Direct solutions of sparse network equations by optimally ordered triangular factorization,” Proc. IEEE, vol. 55, no. 11, pp. 1801–1809, Nov. 1967. B. Stott and O. Alsac, “Fast decoupled load flow,” IEEE Trans. Power App. Syst., vol. PAS-93, no. 3, pp. 859–869, May 1974. F. Wu, “Theoretical study of the convergence of the fast decoupled load flow,” IEEE Trans. Power App. Syst., vol. 96, no. 1, pp. 268–275, Jan. 1977. T. E. DyLiacco and K.A. Ramarao, "Discussion of reference [6]”. 8 [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] S. Deckmann, A. Pizzolante, A. Monticelli, B. Stott and O. Alsac, "Numerical Testing of power system load flow equivalents", Trans. on Power App. & Syst., Vol. PAS-99, No.6, pp. 2292-2300, 1980. A. V. Garcia, A. Monticelli. “Simulation of Distribution Network through Fast Decoupled Power Flow and Axis Rotation”. IX Nacional Conference of Power Distribution Systems - SENDI. Bahia, Sep. 1984. (In Portughese). P.H. Haley and M. Ayres, ''Super decoupled load flow with distributed slack bus" IEEE Trans. on Power App. & Syst, Vol - PAS - No.1, pp. 104-113. 1985. D. Rajicic and A. Bose, "A modification to the fast decoupled power flow for networks with high R/X ratios" IEEE Trans. On Power Systems, Vo1.3, 1988. L. Wang, N. Xiang S. Wang, B.Zhang and M. Huang, "Novel decoupled power flow" IEE Proc. Pt. C, Vo1.137, No.1, pp.1-7, 1990. J. Nanda, D. P. Kothari and S. C. Srivastava "Some important observations on fast decoupled load f low algorithm", (Letter), Proc. IEEE, Vo1.75, No.5, pp.732-733, 1987. R. A. M. Van Amerongen, "A general purpose version of the fast decoupled load flow", IEEE Trans. on Power Systems, Vo1.4, No.2, pp. 760-770, 1989. S.P. Singh, G . S . Raju and V.S. Subba Rao. “A Technique to Improve the Convergence of FDLF for Systems with High R/X Lines”. TENCON'91 - IEEE Region 10 International Conference on Energy, Computer, Communication and Control Systems. R. Berg, E.S. Hawkins, and W.W. Pleines, “Mechanized calculation of unbalanced load flow on radial distribution circuits,” IEEE Trans. On Power Apparatus and Systems, vol. 86, no 4, pp. 415-421, Apr. 1967. D. Shirmohammadi, H.W. Hong, A. Semlyen, and G.X. Luo, “A compensation based power flow method for weakly meshed distribution networks,” IEEE Trans. on Power Systems, Vol. 3, No. 2, pp. 753-762, May 1988. C. C. Durce, E. M. Lourenço, O. L. Tortelli. “Power flow analysis for interconnected T&D networks with meshed topology”. IEEE PES ISGT Europe, Dec. 2011. U. Eminoglu, T. Gözel, M. H. Hocaoglu, “Distribution Systems Power Flow Analysis Package Using Matlab Graphical User Interface (GUI)”, 2009 Wiley Periodicals Inc. R. Singh, B.C. Pal, R.A. Jabr, “Distribution System State Estimation Through Gaussian Mixture Model of the Load as Pseudo Measurement”, IET Generation, Transmission and Distribution, Vol. 4, 2010. VIII. BIOGRAPHIES Odilon L. Tortelli received the degree of Electrical Engineer in 1992 and the M.Sc. degree in 1994, both from the Federal University of Santa Catarina (UFSC), Brazil, and the Ph.D. degree in electrical engineering from the State University of Campinas (UNICAMP), Brazil, in 2010. Since 1995, he has been a Faculty member with the Department of Electrical Engineering at the Federal University of Paraná (UFPR), Curitiba, Brazil. His research interests are in the area of computer methods for power systems operation and control. Elizete M. Lourenço (M’02) received the Electrical Engineering degree as well as the M.Sc. and Ph.D. degrees in electrical engineering from the Federal University of Santa Catarina (UFSC), Florianópolis, Brazil, in 1992, 1994, and 2001, respectively. Since 1995, she has been a Faculty member with the Department of Electrical Engineering at the Federal University of Paraná (UFPR), Curitiba, Brazil. Her research interests are in the area of computer methods for power systems operation. Ariovaldo V. Garcia received the B.Sc., M.Sc., and Ph.D. degrees in electrical engineering from the State University of Campinas (UNICAMP), Brazil, in 1974, 1977, and 1981, respectively. From 1975 to September, 2010, he was with the Electrical Energy Systems Department at UNICAMP and now is an independent consultant. He has served as a consultant for a number of organizations. His general research interests are in the areas of planning and control of electrical power systems. Bikash C. Pal (F'13) received the B.E.E. (with honors) degree from Jadavpur University, Calcutta, India, the M.E. degree from the Indian Institute of Science, Bangalore, India, and the Ph.D. degree from Imperial College London, London, U.K., in 1990, 1992, and 1999, respectively, all in electrical engineering. Currently, he is a Professor in the Department of Electrical and Electronic Engineering, Imperial College London. His current research interests include state estimation, distribution system analysis, power system dynamics, and flexible ac transmission system controllers.