You are on page 1of 6

A Fast Harmonic Load Flow Method for Industrial Distribution Systems

Jen-Hao Teng, IEEE Member Chuo-Yean Chang Department of Electrical Engineering, I-Shou University, Kaohsiung, Taiwan, R. 0. C. Abstract: A fast harmonic load flow method for industrial distribution systems is presented in this paper. The proposed method is designed to save computational time and to accomplish real-time harmonic analysis. Based on the network structures of distribution systems, currentinjection formulation and the Kirchhoffs Law, the relationships between the bus voltages, branch currents and harmonic sources can be formulated, and then the fundamental voltage and harmonic voltage for each bus can all be solved by the simple forwardhackward sweep techniques. The total harmonic distortion can also be calculated easily. The proposed method is very efficient, since the computational time for the LU decomposition of the Jacobian matrix (or Admittance matrix) could be saved. Test results, which are compared with the conventional harmonic load flow method, show that the proposed method is an efficient method and is suitable for real-time harmonic analysis. Keyword: Harmonic Load Flow, Total Harmonic Distortion, Current Injection, Frequency Scan, SteadyState Analysis, Transient-State Analysis
1. Introduction

Harmonic analysis is an important application for distribution system analysis and design. It is used to quantify the distortion in voltage and current waveforms, which is caused by nonlinear loads, electronically switched loads and converters etc., at various buses for a distribution system and to determine whether dangerous resonant problem exists and how they might be mitigated. Such analysis has become more and more important since the presence of harmonic-producing equipment is increasing. As harmonics propagate through the system, they result in increased losses and possible equipment loss-of-life. Equipment might be damaged by overloads resulting from resonant amplifications [1-61.

Distortion in AC voltage and current waveforms can be studied by expressing the harmonic sources o f the waveforms as a Fourier series with the fundamental frequency equal to the power frequency. Then, utilizing harmonic analysis to investigate the generation and propagation of these components throughout distribution systems. Research in this area led to the availability of fairly general technique and software for the formulations and solutions of harmonic propagation problem. The techniques vary in terms of data requirements, modeling complexity, problem formulation and solution algorithm [7- 121. [7-91 presented the frequency-domain based steadystate method that used the frequency-scan process to calculate the frequency response of a system. [lo] presented a time-domain based transient-state analysis that utilized the EMPT program for harmonic analysis. One of the main disadvantages of the time-domain-based method is that the lack of load flow constraints (such as constant power specification at load buses) at the fundamental frequency. Recently, the idea of Wavelet Transform is integrated into harmonic analysis [ 1 1 , 121. With this idea, Wavelet based mathematics models for harmonic sources, system components and power devices need to be derived first and then apply for harmonic analysis. All of those researches will be helpful to realize power system status under harmonic pollution. With transient-state analysis, accurate solution can be obtained, but it spends more computational time. Therefore, frequency-scan based harmonic analysis is the first choice for large-scale power system analysis due to the computational economy. The conventional harmonic load flow methods use load flow program, employ the frequency-based component model, update the Jacobian matrix (or Admittance Matrix), decompose the matrix and rerun the load flow program for each harmonic order. However, the decomposition of the Jacobian matrix (or Admittance Matrix) is a time-consuming procedure and it makes the conventional methods difficultly use for real-time analysis. In order to save the computational time and to use for realtime harmonic analysis, a fast harmonic load flow method for industrial distribution systems is presented in this paper. It is based on the network structures of distribution systems, current-injection formulation and the Kirchhoff s Law, and is solved by the simple forwardhackward sweep technique [13-161. The Total Harmonic Distortion (THD) can also be calculated easily. Test results show the proposed method is an efficient method and is suitable for real-time harmonic

0-7803-6338-8/00/$10.00(~)2000 EEE

1149

analysis.
2. Admittance Matrix Based Harmonic Analysis

load flow and THD for each bus need to be output.


3. The Proposed Algorithm

The Admittance-matrix-baed harmonic load flow analysis [9] was the most widely used method for distribution systems because it was based on the frequency-scan process and the time-honored load flow method [17]. The solving procedure is as follows: (1) Input system data Including bus data, feeder data, harmonic sources, etc.. (2) Fundamental frequency load flow analysis: According to the input data, the Admittance matrix of the fundamental frequency can be built, and then bus voltages and bus angles can be solved by the load flow program [ 173.
(3) Harmonic load flow analysis

The proposed method employs the equivalent current injection transformation and the forwardhackward sweep techniques to solve the harmonic load flow problem. In this section, the development procedure will be described in detail.

Equivalent current injections have been widely used for the applications of distribution systems [13-171. For bus i, the specified complex power Si is

Z : ) And the equivalent current injection ( iteration is


I;

at the k-th

In solving the harmonic load flow problem for various harmonic frequencies, component models need to be modified to its corresponding equivalent models first. When the modified Admittance matrix and the corresponding harmonic currents is obtained, the harmonic load flow problem can be solved by (1).

=z;(qk)+Jz;(v:)=(T P, fie;). (4)


V i

V: is the voltage of bus i at the k-th iteration.

For fundamental load flow, the equivalent current injections need to be transformed at each iteration. But, the harmonic currents are obtained already for harmonic analysis; thus the transformation can be neglected.
Backward Current Sweep
Bus k

where [ Y f h , and [Z]@ are the Admittance matrix, bus voltage vector and injection current vector, respectively. Since [Y](h)and [I]@ are constant matrix and vector, the LU decomposition and forwardhackward substitution of (1) need to be done only once for the h-th harmonic order. It doesnt need any iterative process. The THD is calculated by

@ : Harmonic Currents
Fig. 1: Parts of a Distribution System is the THD of bus i. where THD,(YO)
Vi(h) is the h-th harmonic voltage of bus i.

(4)Results Output
Including the results of fundamental load flow, harmonic

Fig. 1 shows parts of a distribution system and the harmonic currents. The harmonic currents are expressed as I ? ) , I?) and I?, and B,, Bjk and Bjl are the branch currents, respectively. The relationships between branch currents and harmonic currents of Fig. 1 are

1150

From (S), it can be seen that the relationship between branch currents and harmonic currents (or injection currents) of a radial feeder can be calculated by summing the injection current from the receiving bus toward the sending bus of the feeder. The general form can be expressed as:

The fundamental load flow solution can be obtained by solving (4), (6), and (8) iteratively. Since the harmonic currents dont need to be transformed, the harmonic currents, branch currents and bus voltages can be solved.by (6) and (8) directly. It is noticed that component models are the only adjustable parts for each harmonic order in the solution procedure. It will be efficient since the timeconsuming procedure such as LU decomposition- and regenerating the Admittance matrix can be prevented. Fig. 2 shows the flowchart of the proposed fast harmonic load flow algorithm.

I Input System Data 1


where I / is the injection current of bus j at the k-th iteration and S2 is the set of branches connected to the busj. The injection currents can be transformed by (4) for the fundamental load flow, and they are the harmonic currents for the harmonic load flow analysis. Forward Voltage Sweep From Fig. 1, if the voltage of bus i is calculated, then the relationship between branch currents and bus voltages can be expressed as:
problem

Using (8) to obtain bus voltages for the h-th harmonic order I

where Z ( h )is the equivalent impedance of line section for the h-th harmonic order. It is evident for a radial distribution system that if branch currents were calculated, then from (7), the bus voltages can be solved easily. The general form can be expressed as:
Yes

Resultoutput

Fig. 2: Flowchart of the Proposed Method


4. Components Model

Equation (8) is developed by the Kirchhoffs Voltage Law (KVL) and then the bus voltages can be obtained by calculating the voltage from the sending bus forward to the receiving bus of the feeder. Zii and B i are branch

r o m bus i to bus j for the impedance and currents f fundamental load flow analysis, relatively, they are harmonic impedance and harmonic branch current for the harmonic load flow analysis.

In an industrial distribution system, the main equipment that needs to be considered in the harmonic analysis is distribution cable, transformer, capacitor and inductor etc. Instead of using the very accurate model, some practical .and approximated models for industrial use are employed in this paper [4-91. n Cable

1151

the co-generator. Some corrections need to be applied to the resistance and the internal inductance of a distribution cable at high frequency and several modifications have been developed to cope with this purpose. However, a simple model is enough for a practical harmonic analysis. If the line impedance of fundamental frequency is given by
5. Numerical Results

The proposed fast harmonic load flow method was implemented by Boland C++ language on a Windows-98 operation system based Pentium-111 (533) PC. Two methods were used to test, they are

Then, for the h-th harmonic order, the line impedance can be written as

Transformec The transformer is difficult to model at harmonic frequency due to its nonlinear magnetizing properties. In addition, it produces some harmonics due to core saturation, especially, when it is over-excited. In this paper, we assume that the transformer is operated in normal situation and has linear relation under harmonic frequency. The impedance ( Zf) for the h-th harmonic order is

For any new method, it is important to make sure that the final solution of the proposed method is the same as the existing method. A simple 10-bus industrial system is used to verify the accuracy of the proposed method. Tables 1 and 2 are the load data and harmonic current injections of the system. Table 3 is the test results of the system including the fundamental bus voltage, harmonic voltage and THD(%) for each bus and harmonic orders. The results obtained from Method 1 are the same as Table 3 and is not shown here. It means that the proposed method can be utilized for solving the harmonic load flow problem.

I
1 2

Method 1: Admittance matrix based method [9]. Method 2: The proposed method.

0.0

0.0

Capacitor a d Inductor Capacitor and inductor can be modeled similarly, and their impedance for the h-th harmonic order can be expressed by
Xl = 2dlh L

150.0

Xhh = 1/(2nhf,C)
where L and C are the inductance and capacitance of f i is the inductor and capacitor respectively. fundamental frequency. CO-generator The co-generator is operated as a voltage source and supplies power to systems in fundamental frequency, while in harmonic analysis, it is regarded as a passive element, and the harmonic impedance will be:

9 10

250.0 500.0

155.5 311.0

where X f ) is the equivalent impedance at the h-th harmonic order, and X i is the sub-transient reactance of

1152

lateral oriented database needs to be built and maintained. It would be important to use the existing database for practical applications, since it can not only save the data preparation time but also reduce the costs of the database maintenance. In order to prevent the disadvantages, sqme more considerable methods, with the data format, threephase models, weakly meshed network topology, different harmonic-producing equipment taken into accounts, will be published in the future study.
7. References

Execution Time Test The test feeders are lo-, 57- and 156-bus singlephase radial distribution system. The Normalized Execution Time (NET) of the fundamental load flow and harmonic load flow for these two methods are shown in Table 4 and 5, respectively. From Tables 4 and 5, it can be seen that Method 2 outperforms Method 1, especially, for large-scale system. For example, firom Table 5 the NET of 156-bus system are 15454 and 15.592 for Method 1 and Method 2, respectively. Method 2 is 991 times faster than Method 1. Since the time-consuming procedure, such as LU decomposition and forwardhackward substitution are not necessary for Method 2, Method 2 is, defmitely, an efficient algorithm. Method 1
1O-bus feeder 3.375

Method2
1.ooo

57-bus feeder 156-bus feeder

198.75 4375.0

2.750 8.875

1O-bus feeder 57-bus feeder 156-bus feeder

6.136 700.2 15454

1.ooo 5.320 15.592

6. Discussion and Conclusion


A fast harmonic load flow method for industrial distribution systems was presented in this paper. The proposed method extends the works of [13-161 to harmonic analysis and is designed to save computational time and to accomplish real-time harmonic analysis. Several tests have been conducted to verify the feasibility of the proposed method. The proposed method is definitely a fast method; however, one of the main disadvantages of the proposed method is that the layer-lateral data format is essential in the solution procedure. That is, in addition to the conventional bus-branch oriented database, a new layer-

F. D. Martzloff and T. M. Gruz~, Power Quality Site Survey: Facts, Fiction and Fallacies, IEEE Trans. on IA, Vol. 24, pp. 1005-1018,1988. A. Domijan, G. T. Heydt, A. P. S. Meliopoulos, S . S . Venkata and S. West, Directions in Research in Electric Power Quality, IEEE Trans. on PWRD, Vol. PWRD-8, Jan 1993, pp. 429-436. Effects of the Harmonics on Equipment, Report of the IEEE Task Force on Harmonic Impacts. IEEE Trans. on PWRD, Vol. 8, No. 2, April 1994, pp. 672680. J. Arrilliaga, D. A. Bradley, P. S . Bodger, Power System Harmonics, John Wiley and Sons, New York, 1985. Modeling and Simulation of the Propagation of Harmonics in Electric Power Network Part I: Concepts, Models and Simulation Techniques, Task Force on Harmonics Modeling and Simulation, IEEE Trans. on PWRD, Vol. 11, No. 1, Jan. 1996, pp. 452465. Modeling and Simulation of the Propagation of Harmonics in Electric Power Network Part 11: Sample Systems and Examples, Task Force on Harmonics Modeling and Simulation, IEEE Trans on Power Delivery, Vol. 11, No. 1, pp. 466-474, Jan. 1996. D.Xia and G. T. Heydt, Harmonic Power Study Part I - Formulation and Solution, IEEE Trans. on PAS, Vol. PAS 101, No. 6, June 1982, pp. 1257-1265. D.Xia and G. T. Heydt, Harmonic Power Study Part I1 - Implementation and Practical Application, IEEE Trans. on PAS, Vol. PAS 101, No. 6, June 1982, pp. 1266-1270. Y. H. Yan, C. S. Chen, C. S . Moo and C. T Hsu, Harmonic analysis for industrial customers, IEEE Trans. on IAS, Vol. 30, No. 2, March-April 1994, pp. 462 -468. R. H. Kitchin, Converter Harmonics in Power System Using State-Variable Analysis, IEE Proc. Part C, Vol. 128, NO. 4, July 1981, pp. 567-572 G. T. Heydt and A. W. Galli, Transient Power Quality Problem Analyzed Using Wavelet, IEEE Trans. on PWRD, Vol. 12, No. 2, Apr. 1997, pp. 908915. T. Zheng, E. B. Makram and A. A. Girgis, Power

1153

System Transient and Harmonic Studies Using Wavelet Transform, IEEE Trans. on PWRD, Vol. 14, NO.4, Oct. 1999, pp. 1461-1468. [I31 W. M. Lin and J. H. Teng, Phase-Decoupled Load Flow Method for Radial and Weakly-Meshed Distribution Networks, IEE Proc.-Generation, Transmission and Distribution, Vol. 143, No. 1, Jan. 1996, pp. 39-42. [14] D. Shirmohammadi, H. W.Hong, A. Semlyen, G. X. Luo, A Compensation-Based Power Flow Method for Weakly Meshed Distribution and Transmission Networks, IEEE Trans. on PWRS, Vol. 3, No. 2, May 1988, pp. 753 -762. [151 G. X. Luo and A. Semlyen, Efficient Load Flow for Large Weakly Meshed Networks, IEEE Trans. on Power Systems, Vol. 5 , No. 4, Nov. 1990, pp. 13091316. [16] C. S. Cheng, D. Shirmohammadi, A Three-phase Power Flow Method for Real-Time Distribution System Analysis, IEEE Trans on PWRS, Vol. 10,

No. 2, May 1995, pp. 671 -679. [17] T. H. Chen, M. S. Chen K.-J. Hwang, P. Kotas, and E A. Chebli, Distribution System Power Flow Analysis A Rigid approach, IEEE Trans. on PWRD, Vol. 6, NO.3, July 1991, pp. 1146-1152.

Jen-Hao Teng was born in 1969 in Tainan, Taiwan. He received his BS, MS and Ph.D. degrees in electrical engineering from the National Sun Yat-Sen University in 1991, 1993 and 1996 respectively. He has been with.the IShou University, Taiwan, since 1998. His current research interests are Energy Management System, Distribution Automation and Power Quality. Chuo-Yean Chang received his MS degree in electrical r o m the National Sun Yat-Sen University in engineering f 1987. Currently, he is pursuing his Ph.D. degree in the IShou University. His current research interests are Power Quality and Harmonic Filter Design.

1154

You might also like