6th International PhD Seminar on Computational electromagnetics and bioeffects of electromagnetic fields – CEMBEF 2012 Jun 28-30, 2012, Novi

Sad, Serbia

Miodrag MILUTINOV1, Anamarija JUHAS2, Neda PEKARIC –NAD3
Abstract: This paper explores how the phase currents in high voltage overhead power line for known conductor geometry can be extracted from the magnetic field measurements. In the case of a single power line and a known geometry the magnetic flux density depends on 6 unknowns: three phase current amplitudes and their phases. The minimum number of the sensors that need to be used to extract these unknowns is explored in the paper. The appropriate matrix whose elements depend on the mutual positions of the sensors and the PL geometry is formed. The positions of the sensors are proved critical to achieve the best stability of the extracted results. In order to investigate stability of the solutions, the matrix condition number is evaluated. In this paper, instead of the real measurements, the magnetic flux densities are calculated in order to relate the extracted data to the magnetic flux density and thus verify the analytical results. Keywords: Inverse magnetic field problem, magnetic field sensors, overhead power lines

Starting from human exposure point of view it is the most important to determine the highest value of the magnetic flux density and to compare it to the reference levels [7], [8]. The most citied reference levels are proposed by ICNIRP in 1998 [7] and adopted by many countries. The European Council in [9] proposed the same reference levels as ICNIRP in [7]. The widely accepted reference level for the magnetic flux density at power frequency (50Hz) is 100µT [9]. In 2009 the Republic of Serbia published obligatory document [8] with the reference values more restricted than the European and International recommendations [7], [9]. According to the regulations in Serbia, the reference level for magnetic flux density at 50Hz is 40 µT. Reference levels are used to determine the general public exposure levels. Due to the linearity between the magnetic flux density and the currents of the HV PL, and according to the European standard [10], the exposure level can be obtained from the equation (1)
Bmax = Bmeas I max , I meas

Measurements of magnetic and electric field in vicinity of high voltage (HV) overhead power lines (PL) become very important nowadays, particularly in respect to human exposure (See e.g. [1]-[3]). Especially in the last decade, electromagnetic pollution has received enormous attention [4], [5]. Some studies indicate that an excess risk may exist for childhood leukaemia due to the extremely low frequency (ELF) magnetic field generated by HV PL [6]. Magnetic field in the vicinity of HV PL depends on conductors’ geometry (distance between the conductors and the measurement points, as well as conductors’ mutual positions) and currents (both amplitudes and phases). Some of these parameters vary over time and consequently magnetic flux density varies over time, as well. The greatest variation is due to the fluctuations of the amplitude of the currents. An example of the season variation of a HV PL phase current is illustrated in Fig. 1.

where Bmax is the maximum possible value of the magnetic flux density, Bmeas and I meas is the simultaneously measured values of the magnetic flux density and amplitude of the current, respectively, and I max is the maximum possible amplitude of the HV PL current. Accredited laboratories deal with the electromagnetic (EM) field measurements and calculations in order to determine possible human exposure to the EM field. Typically the equation (1) is used, under assumption that the PL currents are balanced. This is most often correct as the phase currents unbalance is low for the transmission lines [11]. But, if the currents are not ideally balanced or almost ideally balanced it is not possible to apply the equation (1). The laboratories usually obtain the data for the currents amplitudes, I meas , as well as the maximum possible amplitude, I max , from the electric power grid company which owns the HV PL. The electric power grid company in Serbia tracks the data for only one of the three phase currents per power line, and the remaining relevant information remains unknown. The goal of this paper is to proceed with the exploration of the current data extraction using magnetic field measurements started in [12].

time jan feb mar apr may

Fig. 1. - A HV PL phase current amplitude variation over several months period.

University of Novi Sad, Dept. of energetic, electronic and telecommunications, Trg Dositeja Obradovica 6, 21000 Novi Sad, SERBIA 1 e-mail: miodragm@uns.ac.rs, 2 e-mail: ajuhas@uns.ac.rs, 3 e-mail pekaric@uns.ac.rs

y2 . it is not possible to determine whether the sequence of the currents is positive or negative. and y1 to y4 . Hence. are real and depend only on the geometry of the power line and the position of the measurement point n (Fig.0. the function cond( A) is evaluated in the five-dimensional space of the parameters x1 . Hence.. POSITIONS OF THE SENSORS In the authors’ previous work [12] it was explained that the uncertainty in the measurements of the magnetic flux density changes the right hand side of the system (6). the x and y -positions of the sensors (Fig. x1 ) and ys = ( y1 . 2 = u ( I12 . 2) are assumed as (9) xs = (− x1 . − x1 . y3 . The equitation (4) can be interpreted as an linear equation in terms of the six real unknowns.   a61  a66   Au = b. I 2 . . as shown in Fig. in [12] the positions of the sensors introduced only 3 unknowns). 2012. I 3 I 3e jψ 3 . Assuming that the phase conductors are straight lines situated at the same heights of 10 m above the ground level.. 2. Using only one magnetic field sensor the magnetic flux density cannot be measured at the different points at the same time. The simultaneous measurement at several points can be done with wireless magnetic field sensors (WMFS) [13]. u6 ). [14]. 2). The WMFS are part of the SEMONT system based on the wireless sensor network technology and intended for the broadband. x =bx1 (n) I 1 + bx 2 (n) I 2 + bx 3 ( n) I 3 The system (7) has a unique solution if and only if the determinant of the system is a non-zero. The condition number is defined as (8) cond( = A) A ⋅ A−1 . Once the system is solved for the six unknowns. separated 10 m from each other. The positions defined by (9) introduce 5 unknowns into the function (8) (as a contrast. a minimum of six equations is required. we analyse the sensitivity of the solution in respect to the measurement uncertainty.. I 32 . the x component of the magnetic flux density can be expressed as a linear function of the currents (3) Bn . B62 ). y and z components.6th International PhD Seminar on Computational electromagnetics and bioeffects of electromagnetic fields – CEMBEF 2012 Jun 28-30. the measurements need to be done simultaneously. From the remaining three unknowns only the cosines of the relative phases can be determined. Due to variation of the currents. In order to determine all six unknowns. In order to keep the number of the parameters low. Fig. b T = ( B12 . . the 50Hz is the dominating component. The coefficients ani . 2e At any point n in the vicinity of the power line. (5) 2 I 2 I 3 cos(ψ 2 − ψ 3 ).Cross-section of the power line.0. x1 . the (7) u T = (u1 . i = 1. The expression for the squared resultant magnetic flux density at the measurement point n is (4) 2 2 Bn = an1I12 + an 2 I 2 + an 3 I 32 + 2an 4 I1I 2 cos(ψ 1 − ψ 2 ) + +2an 5 I 2 I 3 cos(ψ 2 − ψ 3 ) + 2an 6 I 3 I1 cos(ψ 3 − ψ 1 ). with the positions of the conductors and the sensors ( B1 − B6 ). y1 . the respective amplitudes of the currents are easily obtained from the first three unknowns. Novi Sad. y3 ). Serbia EXTRACTION METHOD Assume that the three phase currents of the same HV power line are in the form of phasors jψ 2 (2) = I 1 I1= e jψ 1 . remote. It generally includes measurements of the overall field strength in the entire frequency range of the non-ionizing radiation. As a measure of the sensitivity of the solution of the system of linear equations in respect to errors in the data a condition number [19] is used. So. 2 I1I 2 cos(ψ 1 − ψ 2 ). but for the HV PL. The corresponding magnetic flux density vector can be obtained as a vector sum of x. The six sensors located in the plane perpendicular to the power line introduce a set of 12 unknowns into the function (8).6. 2 I 3 I1 cos(ψ 3 − ψ 1 )). based on six isotropic magnetic flux density measurements. The system of linear equations can be written in the matrix form as (6) where  a11  a16   A=     . automated and permanent monitoring of the EM fields in the real-time [15]-[18]. 2. y4 . I 2 I= .

The probability density function of the current I2. Our future work will be based on the assumption that one of the currents is known (obtained by Electric Power Company) and that the currents are mutually shifted exactly by 2π/3. first we calculate the magnetic flux density magnitudes Bn and the coefficients ani .4m.02 ← Low ← Mean ← High 0. .5 cos(psi12) -0.5m ≤ y1 . 5 present probability density function of the output variables based on M = 106 trials. 4 and 5 for the values of 0. These coverage intervals are obtained from the sensors which positions are determined applying minimum condition number explained in the previous section. and y1 to y4 . the shortest 95% coverage interval of the currents is 9. In the case of the cosines of the relative phases the coverage interval is about 70% of the mean value. in steps of 0. . where the three different curves address the three different maximum values.7 -0.035 0.015 PDF 0. Serbia parameters x1 . . The simulation of the real measurements was performed introducing the uncertainty into analytically calculated values of the magnetic flux densities. The results shown in Fig. The “mean” value is equal to the arithmetic mean. Further.0m. 2012. = y2 4.02 0. The sensitivity of the extracted data was analyzed under assumption that the phase currents are I 1 = 500 e j0 A .The probability density function of the cosines of the phase shift between the currents I1 and I2.01 0.1m. 3 it can be concluded that system becomes more stable with increasing the range of the sensor heights. Please notice the marks “mean”. 0. = y1 1. using (6) we solve the system to obtain the currents and the relative phases. The values of the magnetic flux densities at all six points are calculated with the a priori known currents and the positions of the conductors. y4 ≤ 8m.005 0 -0.015 0. the output variables. y4 ≤ ymax .3 -0.1m. and 1m ≤ y1 . to the measurement uncertainty of the magnetic flux densities we assigned a random number from the interval 0 to 3%. applying Biot-Savart low. We also test the sensitivity of the proposed method in sense of the measurement uncertainty of the sensors. . using the input uncertainty of 3%. 5.1m. It is assumed that the conductors are parallel to the ground. “low” and “high” in Figs.025 0. The authors expect that under these assumptions it will be possible to increase the stability of the solution or to reduce the number of the sensors. – The minimum condition number obtained for the different values of parameter x1.005 0 420 430 440 450 460 Jacina struje [A] 470 480 490 Fig.025 0. calculated in this way is shown in Fig. 6. 6 and i = 1. Next.03 0. I 2 = 450 e − j2π 3 A and I 3 = 550 e j2π 3 A. (10) = y3 6.. 0. the corresponding x and y -positions of the sensors with the minimum condition number of 37 are = x1 8. in steps of 0.2m. Fig. 4. .01 ← Low ← Mean ← High NUMERICAL VALIDATION In order to verify the proposed procedure.2 Fig. Novi Sad.1m.6 -0. in respect to parameter x1 .6th International PhD Seminar on Computational electromagnetics and bioeffects of electromagnetic fields – CEMBEF 2012 Jun 28-30. From the results presented in Fig. take values in the range 1 < x1 ≤ 15m in steps of 1m and 0.4 -0. The minimum condition number.1m. For the example analysed in this paper.26% of the mean value for the amplitudes of the current I 2 .04 PDF Refining the search around the local minimum with parameter 7 < x1 ≤ 9m in steps of 0. 3. = y4 6. 4 and Fig. . ymax . whereas the “low” and the “high” values are bounds of the shortest 95% coverage interval for output variable. 0. at all six “measurement” points. n = 1. 3.

” Official Journal of the European Communities. Juhas. Matrix Computations. vol. ICNIRP: “Guidelines for limiting exposure to time-varying electric. Benes. 2012. pp. “Wireless sensor networks for healthcare applications”. 47: 391–396. pp. She completed her MS and PhD thesis at the University of Belgrade in 1981 and 1984 respectively. Acta Electrotehnica. Borjanovic. Elect. Vukobratović. Conf. “Guidelines for limiting exposure to time-varying electric and magnetic fields (1 Hz-100 kHz). Her research interests include the EM exposure. Đurić. She completed her MS thesis in 1994 at Faculty of Electrical Engineering. he is an Assistant at the Faculty of Technical Sciences. magnetic and electromagnetic fields (0 Hz .6th International PhD Seminar on Computational electromagnetics and bioeffects of electromagnetic fields – CEMBEF 2012 Jun 28-30. ISBN-13: 978-159693-305-7. vol. Neda Pekaric Nadj graduated from the University of Novi Sad. 2010. The positions of the sensors play important role in stability of the extracted data. M. 3rd ed. on modern Power Systems. [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] ACKNOWLEDGEMENT [18] This work was partially supported by Project TR32055 with Serbian Ministry of Science and Technology. Cluj-Napoca. Golub. T. May 2012. Anamarija Juhas graduated from the University of Novi Sad. K. Transm. Novi Sad. Baltimore. symposium on power electronics. “Serbian System for Remote Monitoring of Electromagnetic Fields“. Milutinov. [2] Miodrag M. Serbia. “Basic standard on measurement and calculation procedures for human exposure to electric. Pejovic. Milutinov was born in Zrenjanin. Serbia. N. vol. Serbia in 1976. Djuric. Romania. The Johns Hopkins University Press. J Occup Health 2005. Serbia. Prsa. Symposium on Electrical Apparatus and Tehnologies. Milutinov. April 1998. “Power line currents data extraction from magnetic field measurements”.26% of the mean value. Gen. 481–486. “Evaluation of Continuous Exposure to Magnetic Field From AC Overhead Transmission Lines Via Historical Load Databases: Common Procedures and Innovative Heuristic Formulas”. the shortest 95% coverage interval of amplitudes of the currents is 9. “Council Recommendation on the Limitation of Exposure of General Public to Electromagnetic Fields (0 to 300GHz). April 1999. “Electromagnetic fields from power lines”. 1996. Kasas-Lazetic. In the case of the cosines of the relative phases. pp. Sep. 2009. a minimum of six simultaneous magnetic flux density isotropic measurements is needed in order to determine the currents. In 2001 he graduated from the Faculty of Technical Sciences in Novi Sad and enrolled at the PhD studies at the same Faculty.300 GHz)”. “Sensor network for power lines magnetic field monitoring”. Djuric. Olsen. Jankovic and Z. 74. 1999. glasnik RS” br. The positions of 6 sensors in the plane perpendicular to the HV PL introduce 12 unknowns. Ltd. 2011. the shortest 95% of the coverage interval is about 70% of the mean value. N. 516-525. S. Artech House. REFERENCES [1] Robert G. EN 504013:2008. 2011. Van Loan. Dishongh and M. Bulgaria. 2006 T. May 17-20. Jan. The stability was tested using the condition number. Kasas-Lazetic.” Health Physics. Swanson. in 1990. “Magnetic fields from transmission lines: Comparison of calculations and measurements. Dargie and C. EMF measurements and signal analysis. 2010. She defended her PhD thesis at University of Novi Sad in 2009. M. and electromagnetic fields (up to 300 GHz). N. Int. Currently. G. Radiation Protection Dosimetry 118 (3). Dec. It cannot be determined whether the sequence of the currents is positive or negative. Inst. SIELA 2012. University of Belgrade. Distrib. [3] [4] [5] . W. in 1978. Hence. Starting from these six equations it is possible to determine the amplitudes of the currents and the cosines of relative phases. Transactions on power delivery IEEE. Therefore the task was to find the positions of the sensors while achieving the best stability. The stability of the system and the output results depends on the geometries which include mutual positions between the conductors and the sensors. B. Pekaric-Nad. M. pp. “ECG Changes in Humans Exposed to 50 Hz Magnetic Fields”. University of Novi Sad. Oct. USA. Under some restrictions explained in the section 3. “Fundamentals of Wireless Sensor Networks: Theory and Practice”. 99. 2010 N. 494-522. “Possible health effects of power line electric fields. C. John Wiley & Sons. J. K. Selected Best Papers of the PES 2011. 25. Novi Sad. “Information Network for Continuous Electromagnetic Fields Monitoring“. [19] ICNIRP.. Special Issue. 140-142. Eng.” in Proc. 2011. 1988. Villalta. School of Electrical Engineering and Computer Science Washington State University Pullman. Penick. vol. M. Serbia. magnetic. 4th Int. M. Journal of Emerging Sciences. F. Dec. Narda Safety Test Solution: Multi-band area monitor AMB 8057 – Operating Manual.. J. A. For the example analysed in this paper and the magnetic flux density measurement uncertainty of 3%. 104/2009. Serbian national rulebook on non-ionizing radiation limits (in Serbian). Her research interests are Numerical electromagnetics and Electromagnetic compatibility. Bourgas. the minimal condition number was found to be 37. 818-836. Dec. 142. Prsa. 2010. WA 99164-2752 G Mazzanti.” EE368 Electrical Power Transmission and Distribution. 16th Int. McGrath. “Sl. M. “ELF field in the proximity of complex power line configuration measurement procedures”. Serbia CONCLUSIONS For the known geometry of the three phase power line. H.” Health Physics. pp. Poellabauer. a set of six wireless magnetic field sensors (WMFS) can provide enough information to populate the column vector of the magnetic flux densities in the equations. Currently she is professor at University of Novi Sad. 17th Int. no. 1. S. Comelli and R.