Printer Friendly

OSSB and Hybrid Methods to Prevent Cable Faults for Harmonic Containing Networks.

I. Introduction

Transmission and distribution of high voltage are made by using overhead and underground cable lines. High voltage creates insulation faults and electrical safety problems. Especially, providing of electrical safety of overhead lines is major problem in city center or crowded areas. Hence, underground cable lines are used instead of overhead lines in city center or crowded areas because high voltage underground cable (HVUC) has an insulation layer, so electrical insulation of HVUL is more than overhead line. Also, insulation layer is cowered by metallic sheath to protect insulation layer against to environmental factors. In Fig. 1, HVUC is shown [1].

Metallic sheath protects insulation material of cable against to environmental factors, also metallic sheath contributes smoothing of electric field on high voltage underground cable, but SC generates on metallic sheath. SC and harmonics increase cable temperature, so cable insulation life and performance reduce. Also the sheath voltage (SV) increases due to SC and harmonics, so electroshock risk occurs for human due to high SV. Metallic sheath of HVUC is grounded to reduce SC effects [2]-[5]. These bonding methods are indicated as single point bonding, solid bonding and cross bonding methods in IEEE 575-1988 standard. Bonding methods are shown in Fig. 2, Fig. 3 and Fig. 4.

If cable line length is short, single point bonding is used. If cable length is long, solid bonding or cross bonding are used. Namely SC value affects using of bonding method type, and SC value HVUL which will be installed as a new line should be known to select the most suitable method. Hence, cable faults and electroshock risks are prevented during electric transmission or distribution. Forecasting methods are suggested to determine SC of a new high voltage underground line in this study. ANN and hybrid ANN methods are used to forecast SC of the new line. However, inputs of forecasting methods must be determined to forecast SC. In this study, the sheath current forming factors are used as inputs. Therefore, knowing of the sheath current forming factors are very important for forecasting of the sheath current.

Three phase systems is generally accepted as balanced system, and total magnetic field around cable can be accepted as zero because vectorial sum of total magnetic field is zero in the balanced three phase system. However, total magnetic field is not zero in unbalanced three phase systems. Thus SC occurs on metallic sheath. Namely, the most important factor is unbalanced phase current for forming of SC [6]. Another factor is the cable configuration of HVUC. These cable configurations are shown in Fig. 5. The sheath current of trefoil configuration is lower than flat formation. However, flat formation is generally used [7], [8].

HVUL length (L), distance of between phases (d) and grounding resistance (Rg) of bonding method are the other effective factors for forming of SC. [9]-[12]. In this study, different cable lines are generated in PSCAD/EMTDC simulation software. These simulation results are used as input data of ANN, and different ANN network type and hybrid ANN are used to forecast SC. In literature, the results of hybrid ANN are compared with the result of feed forward back propagation network type ANN, but the other network types are not compared [13]. In this study, the results of these different ANN network types are compared with hybrid ANN methods. In literature different types bonding methods are used to prevent SC effects, and the sectional solid bonding (SSB) is newest method. However, harmonic effect is not considered in these bonding methods. In this study, SSB is used, and harmonic effect is considered. However, SSB should be optimized to find the most suitable parameter values. Hence, different optimization methods are used to optimize SSB.

II. Material and Method

Electromagnetic transient programs can be used for simulation of HVUL [14]. In this study, PSCAD/EMTDC as electromagnetic transient program are used for simulation studies of HVUL.

In simulation studies, primarily high voltage cable is modelled, the modelled cable is shown in Fig. 6. Different 59 high voltage cable lines are simulated, and SC of these lines are determined in PSCAD/EMTDC. Hence, L, d, Iub, Rg and SC of different 59 lines are used in training and forecasting studies.

A. Forecasting Methods

Working principle of ANN is based on mathematical model of human learning, and the basic element of ANN is neuron. Also ANN occurs from input layer, hidden layer and output layer [15]--[17]. The basic schema of ANN is shown in Fig. 7.

Neurons work as a transfer function. This transfer function is shown in (1), and a neuron working schema is shown in Fig. 8

[y.sub.i] = [f.sub.i]([n.summation over (j=1) [w.sub.ij] x [x.sub.j] + [b.sub.i]), (1)

where [x.sub.j] is input, [w.sub.ij] is weight, [b.sub.i] is bias, [f.sub.i] is transfer function, and [y.sub.i] is output of the neuron.

Mean square error (MSE) method is used in determination of training error. MSE is shown in (2)

E(t) = 1/n [n.summation over (i=1)][(p(i) - o(i)).sup.2], (2)

where E(t) is forecasting error at tth iteration, p(i) is the desired value for ith output, o(i) is the real value for ith output. Weights of ANN must be updated to reduce training error, to update weights in classic ANN is used (3)

[w.sub.i](t +1) = [w.sub.i](t) + [DELTA][w.sub.i](t). (3)

There are many ANN network types in literature. In this study, feedforward backpropagation (FFBP), cascade feedforward backpropagation (CFFBP), Elman backpropagation (EBP) and layer recurrent (LR) network types are used as different network types. Also hybrid ANN methods are used to obtain the most accurate results. In hybrid ANN methods, weights of ANN are optimized with optimization methods according to training and forecasting errors. In classic ANN, weights are updated by (3), but in hybrid ANN, weights are updated by optimization methods. In this study, differential evolution algorithm (DEA), particle swarm optimization (PSO), genetic algorithm (GA) and inertia weight particle swarm optimization (iPSO) are used as optimization method to determine the most suitable weights. Flow chart of hybrid ANN method is shown in Fig. 9.

B. Bonding Methods

If single-point bonding is used for long HVUL, high voltage occurs at the end of line. Solid bonding and cross bonding methods are used for long HVUL. However, SV of cross bonding method is generally more than solid bonding method and touch voltage limit (50 V rms or 70 V peak) for human, so solid bonding can be used for bonding of cable [18], [19]. However, SV of solid bonding can exceed touch voltage limit in case of increasing of unbalanced phase current and harmonic distortion. It is seen that these methods are not exactly sufficient to prevent SC effects. Thus, SSB is developed, but performance of SSB is not measured with high harmonic condition. SSB is shown in Fig. 10 [20]. In SSB, total line length is called as major part. Major part consists of minor parts, and minor part parameters should be optimized according to touch voltage limit value. Minor part parameters are the maximum minor part length (Lmax), minimum distance of between phases (dMIN) and grounding resistance (Rg). GA, PSO, iPSO and DEA are used for optimization of SSB, and the results of these methods are compared to find the most suitable parameter values for minor parts. Flow chart of the optimized SSB (OSSB) is shown in Fig. 11. Vs is the sheath voltage of high voltage underground cable. Also performance of OSSB is examined for load current with high harmonic distortion.

III. Experimental Studies

Experimental studies is occurred as forecasting studies and application of OSSB in high harmonic distortion case.

A. Forecasting Studies

In forecasting studies, 59 x 5 data matrix is used for training of ANN and hybrid ANN, and 12 x 5 data matrix is used for test of ANN. In these matrices, L, d, Iub, Rg and SC are used as variables. Values of these variables are obtained in PSCAD/EMTDC. This 12 x 5 data matrix is shown in Table I. In Table I [I.sub.SA] is defined as the real value of SC. [I.sub.SA] is obtained from simulation studies in PSCAD/EMTDC. Isa is used in training and test studies to obtain errors. In Table II, Isf is defined as the forecasted sheath current of different type ANN models. Also, Test matrix is used to forecast SC of 12 different HVUL. These results are shown in Table III.

B. Application of OSSB

Unbalanced phase current (Iub) is selected as 464 A because ampacity of the modeled cable is 464 A. Total voltage harmonic distortion is selected as 5 % because total voltage harmonic distortion limit for 31.5 kV is determined maximum 5 % in IEEE STD 519-1992 Harmonic Limits. Total cable length is selected as 1000 m because maximum length of high voltage underground cable is produced maximum 1000 m as one piece. In application of OSSB, SC of HVUL which will be installed as a new line is determined by iPSO-ANN, and optimization of SSB is made by using DEA, GA, PSO and iPSO. When DEA is used for optimization of SSB, the major part is divided as 3 parts, and each minor part length is determined as 333 m, minimum Rg is determined as 17.88 ohm, d is determined as 0.3659 m. When this line is simulated for without harmonic distortion condition in PSCAD/EMTDC, SC is determined as 3.35 A, and SV is determined as 59.99 V. These results are shown in Fig. 12. When this line is simulated for with harmonic distortion condition in PSCAD/EMTDC, SC is determined as 3.56 A, and SV is determined as 63.70 V. These results are shown in Fig. 13.

If more reducing of SC is desired, minimum Rg must be bigger than 17.88 ohm. When iPSO is used for optimization of SSB, the major part is divided as 3 parts, and two minor part lengths are determined as 360 m. Minimum Rg is determined as 11.14 ohm, and d is determined as 0.35 m for 360 m minor parts. Length of another part is determined as 280 m, and minimum Rg is determined as 36.51 ohm, and d is determined as 0.3688 m. When this line is simulated for without harmonic distortion condition in PSCAD/EMTDC, SCs are determined as 5.81 A (for Minor 1 and Minor 2) and 1.38 A (for Minor 3). SVs are determined as 64.84 V (for Minor 1 and Minor 2) and 50.71 V (for Minor 3). These results are shown in Fig. 14.

When this line is simulated for with harmonic distortion condition in PSCAD/EMTDC, SCs are determined as 6.25 A (for Minor 1 and Minor 2) and 1.49 A (for Minor 3). SVs are determined as 67.90 V (for Minor 1 and Minor 2) and 54.42 V (for Minor 3). These results are shown in Fig. 15.

If more reducing of SC is desired, minimum Rg for Minor 1 and Minor 2 must be bigger than 11.44 ohm, and minimum Rg for Minor 3 must be bigger than 36.51 ohm.

When PSO is used for optimization of SSB, the major part is divided as 2 parts, and minor part lengths are determined as 500 m, and minimum Rg is determined as 35.77 ohm, and d is determined as 0.1809 m. When this line is simulated for without harmonic distortion condition in PSCAD/EMTDC, SC is determined as 2.59 A, and SV is determined as 92.67 V. These results are shown in Fig. 16. When this line for with harmonic distortion condition is simulated in PSCAD/EMTDC, SC is determined as 2.69 A, and SV is determined as 94.76 V. These results are shown in Fig. 17. If more reducing of SC is desired, minimum Rg must be bigger than 35.77 ohm.

When GA is used for optimization of SSB, the major part is divided as 3 parts, and minor part lengths are determined as 333 m, and Rg is determined as 28.83 ohm, and d is determined as 0.3943 m. When this line is simulated for without harmonic distortion condition in PSCAD/EMTDC, SC is determined as 2.06 A, and SV is determined as 59.5 V. These results are shown in Fig. 18. When this line is simulated for with harmonic distortion condition in PSCAD/EMTDC, SC is determined as 2.18 A, and SV is determined as 62.4 V. These results are shown in Fig. 19. If more reducing of SC is desired, minimum Rg must be bigger than 28.83 ohm.

IV. Discussion

This study consists of two main sections. The first section is forecasting of SC. Different ANN networks and hybrid ANN methods are used to forecast SC of HVUL. Training and forecasting (test) errors of different ANN networks and hybrid ANN methods are compared, and these results are shown in Table II. It is seen in Table II that training and forecasting errors of hybrid ANN method are generally lower than the other errors of ANN network types. Also training and forecasting errors of hybrid iPSO-ANN are lower than the other hybrid ANN methods.

In this study, long HVUL is considered to see SC effects, and OSSB is suggested to reduce SC effects for long HVUG. In literature, solid bonding and cross bonding are suggested to reduce SC effects for long HVUL. In application studies of OSSB, solid bonding is selected to compare OSSB because SV of solid bonding is lower than cross bonding. Also high harmonic distortion is considered during simulation studies. In simulation studies, if Iub is 464 A, and cable length is 1000 m, SC of solid bonding is determined as 3.56 A, and SV of solid bonding is determined as 178.2 V. This result is shown in Fig. 20. When harmonic distortion is considered, SC of solid bonding is determined as 3.66 A, and SV of solid bonding is determined as 183.05 V. This result is shown in Fig. 21.

It is seen that if solid bonding is used to reduce SC effects, SV exceeds touch voltage limit. Also high harmonic distortion increases SV of HVUL.

When DEA, iPSO and GA are used for optimization of SSB (OSSB), SV does not exceed touch voltage limit although high harmonic distortion. However, when PSO is used for optimization of SSB, SV exceeds touch voltage limit for human. Thus DEA, iPSO and GA can be used for optimization of SSB (OSSB). If further reducing of SC is desired, Rg must be bigger than minimum Rg the determined by iPSO, GA and DEA.

V. Conclusions

In this study, OSSB method is suggested to eliminate SC of HVUL in high harmonic distortion condition, and parameters of OSSB should be optimized according to SC value. Hence SC of HVUL should be accurately determined to determine OSSB parameters in the high harmonic conditions. Thus in this study, different ANN network types and hybrid ANN methods are used to forecast correctly SC of HVUL. When iPSO-ANN is used to forecast SC of HVUL, training error of iPSO-ANN is determined as 0.1304, and forecasting error of iPSO-ANN is determined as 0.2012. Thus, iPSO-ANN that is among these methods has high accuracy level. Therefore, iPSO-ANN is used to determine OSSB parameters.

After SC is forecasted by iPSO-ANN, parameters of OSSB are optimized by using iPSO, GA and DEA methods separately, and SV is respectively measured as 63.7 V, 62.4 V and 63.7 V for iPSO, GA and DEA in PSCAD/EMTDC. Also, the results of OSSB and solid bonding methods are compared to determine the most suitable method for bonding of HVUL. When solid bonding is used in high harmonic condition, SV is measured 183.05 V in PSCAD/EMTDC. According to these results, OSSB is the most suitable method to prevent SC effects, and iPSO, GA or DEA can be used for optimization of OSSB parameters.

http://dx.doi.org/10.5755/j01.eie.24.2.20633

REFERENCES

[1] A. D. Tziouvaras, "Protection of high-voltage AC cables", in Proc. 59th Annual Conf. for Protective Relay Engineers, TX, USA pp. 4861.DOI: 10.1109/CPRE.2006. 1638691. R. Benato, "Multiconductor analysis of underground power transmission systems: EHV AC cables", Electric Power Systems Research, vol. 79, no. 1, pp. 27-38, 2009. DOI: 10.1016/j.epsr.2008.05.016.

[2] J. R. Ruiz, A. Garcia, X. A. Morera, "Circulating sheath currents in flat formation underground power lines", in Proc. 2007 International Conf. Renewable Energies and Power Quality, 2007, pp. 1-5. DOI: 10.24084/repqj05.217.

[3] L. Zhonglei, B. X. Du, L. Wang, C. Yang, H. J. Liu, "The calculation of circulating current for the single-core cables in smart grid", in Proc. 2012 IEEE Innovative Smart Grid Technologies--Asia, 2012, pp. 1-4. DOI: 10.1109/ISGT-Asia.2012.6303176.

[4] S. Czapp, K. Dobrzynski, J. Klucznik, Z. Lubosny, "Calculation of induced sheath voltages in power cables - single circuit system versus double circuit system", Journal of Information, Control and Management Systems, vol. 12, pp. 113-123, 2014.

[5] C. K. Jung, J. B. Lee, J. W. Kang, "Sheath circulating current analysis of a cross-bonded power cable systems", J Electr Eng & Technol, vol. 2, pp. 320-328, 2007. DOI: 10.5370/JEET.2007.2.3.320. O. E. Gouda, A. A. Farag, "Factors affecting the sheath losses in single-core underground power cables with two-points bonding method", Int J Electr & Comp Eng, vol. 7, pp. 7-16, 2012. DOI: 10.11591 /ij ece .v2i 1.115.

[6] J. R. Ruiz, X. A. Morera, "Effects of the circulating sheath currents in the magnetic field generated by an underground power line", in Proc. International Conf. Renewable Energies and Power Quality, 2006, Mallorca, pp. 1-5. DOI: 10.24084/repqj04.217.

[7] C. K. Jung, J. B. Lee, J. W. Kang, "Sheath circulating current analysis of a cross-bonded power cable systems", Journal of Electrical Engineering & Technology, vol. 2, pp. 320-328, 2007. DOI: 10.5370/JEET.2007.2.3.320. O. E. Gouda, A. A. Farag, "Factors affecting the sheath losses in single-core underground power cables with two-points bonding method", International Journal of Electrical and Computer Engineering, vol. 2, pp. 7-16, 2012. DOI: 10.11591 /ijece.v2i 1.115.

[8] J. R. Ruiz, X. A. Morera, "Effects of the circulating sheath currents in the magnetic field generated by an underground power line", in Proc. Int. Conf. Renewable Energies and Power Quality, pp. 1-5, 2006. [Online]. Available: http://dx.doi.org/ 10.24084/repqj04.217

[9] W. Moutassem, G. J. Anders, "Calculation of the eddy current and hysteresis losses in sheathed cables inside a steel pipe", IEEE Trans. Power Delivery, vol. 25, pp. 2054-2063, 2010. DOI: 10.1109/TPWRD.2010.2049509.

[10] B. Akbal, "Neural-network-based current forecasting on high-voltage underground cables", Electronics World, vol. 122, no. 1959, pp. 30-34, 2016. U. S. Gudmundsdottir, B. Gustavsen, C. L. Bak, W. Wiechowski, "Field test and simulation of a 400-kV cross-bonded cable system", IEEE T Power Deliver, vol. 26, pp. 1403-1410, 2011. DOI: 10.1109/TPWRD.2010.2084600.

[11] W. Charytoniuk, M. S. Chen, "Very short-term load forecasting using artificial neural networks", IEEE Trans. Power Systems, vol. 15, pp. 263-268, 2000. DOI: 10.1109/59.852131.

[12] R. Achanta, "Long term electric load forecasting using neural networks and support vector machines", International Journal of Computer Science and Technology, vol. 3, pp. 266-269, 2012. DOI: 10.1016/j.apm.2008.07.010.

[13] T. Weigerta, Q. Tianb, Q. Lianb, "State-of-charge prediction of batteries and battery-supercapacitor hybrids using artificial neural networks", Journal of Power Sources, vol. 196, pp. 4061-4066, 2011. DOI: 10.1016/j.jpowsour. 2010.10.075.

[14] K. V. Gouramanis, C. G. Kaloudas, T. A. Papadopoulos, G. K. Papagiannis, E. K. Stasinos, "Sheath voltage calculations in long medium voltage power cables", in Proc. IEEE Trondheim Power Tech, Norway, 2011, pp. 1-7. DOI: 10.1109/PTC.2011.6019234.

[15] C. K. Jung, J. B. Lee, J. W. Kang, X. H. Wang, Y. H. Song, "Sheath current characteristic and its reduction on underground power cable systems", in Proc. IEEE Power Engineering Society General Meeting, 2005, CA, USA, pp. 2562-2569. DOI: 10.1109/PES.2005.1489525.

[16] B. Akbal, "Sectional solid bonding for grounding of high voltage underground cables to reduce the sheath current effects", International Journal of Innovative Research in Engineering & Management, vol. 2, no. 2, pp. 103-107, 2016.

Bahadir Akbal

Department of Electric and Electronic Engineering, Selcuk University, Konya, Turkey

bakbal@selcuk.edu.tr

Caption: Fig. 1. High voltage underground cable.

Caption: Fig. 2. Single-point bonding.

Caption: Fig. 3. Solid bonding.

Caption: Fig. 4. Cross bonding.

Caption: Fig. 5. Trefoil and flat formations

Caption: Fig. 6. The modeled high voltage underground cable

Caption: Fig. 7. The basic schema of ANN.

Caption: Fig. 8. Neuron working schema.

Caption: Fig. 9. Hybrid ANN flow chart.

Caption: Fig. 10. The sectional solid bonding.

Caption: Fig. 11. Flow chart of OSSB.

Caption: Fig. 12. The result of DEA without harmonics condition.

Caption: Fig. 13. The result of DEA with harmonics condition.

Caption: Fig. 14. The result of iPSO without harmonics condition.

Caption: Fig. 15. The result of iPSO with harmonics condition.

Caption: Fig. 16. The result of PSO without harmonics condition.

Caption: Fig. 17. The result of PSO with harmonics condition.

Caption: Fig. 18. The result of GA without harmonics condition.

Caption: Fig. 19. The result of GA with harmonics condition.

Caption: Fig. 20. The result of solid bonding without (a) and with (b) harmonics condition.
TABLE I. TEST MATRICES FORECASTING RESULTS OF THE
SHEATH CURRENT.

Line Type    [I.sub.UB](A)     L(m)     d(m)    Rg([ohm])

  Line 1        224.630        250      0.1         6
  Line 2        224.630        500      0.1         10
  Line 3        224.630        500      0.1         40
  Line 4        224.630        500      0.5         9
  Line 5        224.630        250      0.5         6
  Line 6        464.120        250      0.5         10
  Line 7        464.120        400      0.1         35
  Line 8        464.120        700      0.1         40
  Line 9        464.120        500      0.5         35
 Line 10        464.120        500      0.5         40
 Line 11        464.120        1000     0.5         50
 Line 12        224.630        1000     0.1         50

Line Type    [I.sub.SA](A)

  Line 1         3.61
  Line 2         3.30
  Line 3         1.08
  Line 4         4.42
  Line 5         3.34
  Line 6         4.48
  Line 7         2.20
  Line 8         3.32
  Line 9         2.54
 Line 10         1.97
 Line 11         3.48
 Line 12         1.71

TABLE II. TRAINING AND FORECASTING ERRORS OF DIFFERENT
TYPE ANN MODELS.

              ANN type                 Training   Forecasting
                                        Errors       Errors

    Feedforward backpropagation         1.1493       0.7732
Cascade feedforward backpropagation     0.1689       0.2769
          Layer recurrent               0.5380       0.6877
       Elman backpropagation            0.5608       0.8863
              iPSO-ANN                  0.1307       0.1139
              PSO-ANN                   0.2285       0.3998
              DEA-ANN                   0.1653       0.2373
               GA-ANN                   0.2217       0.3154

TABLE III. THE FORECASTED SC OF DIFFERENT NETWORK TYPE
ANN AND HYBRID ANN MODELS.

L                              [I.sub.SF] (A)
      FFBP    CFFBP      LR      EBP    iPSO   PSO    DEA     GA

1     3.49     3.01     2.33    2.56    3.63   3.22   3.62   3.36
2     3.49     2.02     1.19    1.22    4.45   3.67   4.45   3.42
3     3.49     3.51     3.76    3.80    0.90   3.66   0.91   3.40
4     3.49     2.91     2.71    2.66    4.55   3.66   4.59   3.42
5     3.49     2.39     1.90    1.81    3.62   3.22   3.64   3.36
6     3.49     1.95     1.35    1.26    4.35   1.87   4.39   3.10
7     3.49     1.32     0.81    0.75    2.18   2.13   2.20   3.12
8     3.49     4.65     4.33    5.45    3.12   2.67   3.17   3.20
9     3.49     3.34     2.26    2.97    2.68   2.32   2.74   3.15
10    1.35     3.66     3.92    5.48    2.08   2.31   2.15   3.14
11    1.35     2.50     2.00    3.00    3.47   3.20   3.58   3.27
12    1.35     1.67    1.055    1.41    2.28   4.54   2.34   3.52
COPYRIGHT 2018 Kaunas University of Technology, Faculty of Telecommunications and Electronics
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2018 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:optimized sectional solid bonding
Author:Akbal, Bahadir
Publication:Elektronika ir Elektrotechnika
Article Type:Report
Date:Apr 1, 2018
Words:4094
Previous Article:Planar Inductor for Biological Experimentation in Pulsed Magnetic Fields.
Next Article:Evaluation of (1 + [alpha]) Fractional-Order Approximated Butterworth High-Pass and Band-Pass Filter Transfer Functions.
Topics:

Terms of use | Privacy policy | Copyright © 2020 Farlex, Inc. | Feedback | For webmasters