Audible Noise Calculation for Different Overhead

Transmission Lines
Indra Kurniawan Suwarno Oliver Pischler Uwe Schichler
[email protected] [email protected] [email protected] [email protected]

School of Electrical Engineering and Informatics Institute of High Voltage Engineering and System Performance
Bandung Institute of Technology Graz University of Technology
Bandung, Indonesia Graz, Austria

Abstract—Overhead lines are known to emit audible noise corona induced phenomenon which occurs if the electric field
during foul weather conditions. Audible noise phenomena depend intensity on OHL conductors (surface gradient) surpasses a
largely on the OHL conductor’s surface gradient which is certain limit. Various empirical equations have been developed
therefore also used in all equations for audible noise prediction. for the prediction of audible noise levels [2]. Apart from
Various methods for the calculation of OHL surface gradients
geometry data, the surface gradient is the most important input
have been developed in the past. By using a realistic Indonesian
tower design as an example, this contribution draws a comparison parameter for such those equations. Several analytical as well as
between surface gradient results obtained with several common numerical approaches with varying accuracy and complexity are
calculation methods and with a finite element calculation. The available for the calculation of OHL surface gradients. By using
investigated methods include the Maxwell Potential Coefficient a realistic Indonesian tower design as an example, this
Method, MARKT and MENGELE’s method and Charge Simulation contribution compares three calculation methods with the finite
Method. A software has been developed for the assessment of the element method. Furthermore, also the validity of audible noise
charge simulation method which is also presented here. The prediction formulas is analyzed. For this purpose, audible noise
comparisons show that even relatively simple methods yield measurements were carried out on a four-conductor bundle
reliable results. Furthermore, this paper investigates the validity
under different precipitation rates at the high voltage laboratory
of audible noise prediction equations for AC as well as DC OHL
with four-conductor bundles. For that purpose, audible noise of the Graz University of Technology. The experiments were
measurements were carried out in a high voltage laboratory under conducted under AC as well as DC stress. The gathered emission
realistic weather conditions. The results are used to calculate the values were then converted to sound power levels which are
corresponding sound power levels which are then compared to compared to the sound power levels calculated with prediction
results calculated with the well-known BPA and EPRI equations equations.
for AC and DC audible noise prediction. The approach used to
calculate the sound power levels of conductor bundles by II. REVIEW OF CONDUCTOR SURFACE POTENTIAL
measuring audible noise at fixed immission points is explained in GRADIENT CALCULATION
great detail. The investigations show good agreement between
In the past, empirical equations were developed to predict
laboratory results and AN predictions for AC OHL. However,
some differences have been noted in the case of DC. In the last part audible noise levels and thereby aid the design process of
of this contribution the compiled results are used to predict overhead lines. Apart from data regarding the bundle-geometry
audible noise levels for a European AC line and for the planned (spacing, conductor diameter etc.), the most important
HVDC-link between Sumatra and Java. parameter for such equations is the conductor surface gradient
(electric field intensity on the conductor surface). Using normal
Keywords—Corona, audible noise, surface gradient, charge voltage gradients calculated from smooth cylindrical conductor
simulation method whose diameters are equal to the outside diameters of actual
stranded conductor has became generally accepted for the
I. INTRODUCTION evaluation of corona effects [3]. While nowadays FEM
The ever-increasing demand for energy as well as the current calculations are readily available, various methods for the
change in power generation strategies necessitate a constant calculation of OHL surface gradients have been developed in the
augmentation of overall overhead line (OHL) transmission past, three of which are described below.
capacity. However, the construction of new overhead lines lacks A. Maxwell Potential Coefficient Method
approval by the general population. One of the reasons for this
Maxwell potential coefficient method (MPCM) was firstly
rejection is the well-known fact that overhead lines are prone to introduced by TEMOSHOK in 1948 [4]. The details for the
emit corona-induced audible noise (AN) under foul weather calculation procedure for transmission lines with a single
conditions (rain, snow, fog etc.) which can be a significant conductor were explained by ADAMS [5]. This method uses only
annoyance for peopling living close by [1]. Audible noise is a a single line charge to model the charge distribution on the OHL
PT. PLN (Persero), Jalan Trunojoyo Blok M I/135, Jakarta 12160, Indonesia

978-1-5386-2910-9/18/$31.00 ©2018 IEEE

conductor. This means that the non uniform surface charge an AC ripple can be observed in the DC system. The results of
distribution which occurs on the sub-conductors of bundle the comparison can be seen in Table. 1.
conductors is neglected.
B. MARKT and MENGELE‘s Method
MARKT and MENGELE were the first to suggest a method for
the calculation of the conductor surface gradient for overhead
lines with bundle conductors [6]. In this method, each conductor
bundle is replaced by an equivalent conductor with an equivalent
radius req defined by:

= (1)
Where: n is the number of sub-conductors
r is the sub-conductor radius
A is the bundle radius
With the bundles represented by the equivalent conductors,
the total charge on each of them is calculated by the Maxwell
potential coefficient method. Knowing the total charge on the
bundle qt from the preceding step, the average bundle gradient
Eav is calculated as: Fig. 1. Surface Gradient Calculation Tool using CSM

= (2) TABLE I. COMPARISON OF HYBRID SIMULATION RESULTS

(US-C TOWER DESIGN [9])
The average maximum bundle conductor gradient Em is then
Peak Electric Field (kV/cm)
obtained as:
AC + Bias AC - Bias
Line
= 1 1 (3) [9] TUG [9] TUG

In the case of MARKT and MENGELE's method, since the AC R 18.04 17.98 20.73 20.68
bundle is assumed to be isolated, the maximum bundle AC S 22.33 22.17 19.64 19.47
conductor gradient will also be equal to Em [6].
AC T 21.44 21.38 18.07 18.00
C. Charge Simulation Method
D. Comparison of Surface Gradient Calculation Methods
The Charge Simulation Method (CSM) was introduced in
1969 when ABOU-SEADA and NASSER evaluated field strengths To assess the accuracy of the surface gradient calculation
in twin cylindrical conductors [7]. Further details of CSM were methods described above, they were all applied to calculate the
published by SINGER, STEINBIGLER, and WEISS [8]. The surface gradients of Indonesian 500 kV AC tower setup (Fig. 2).
non uniformly distributed surface charges are replaced by using An additional surface gradient calculation was carried out with
discrete fiction charges. To determine the value of line charges, a finite element tool (ElecNet v. 7.7.1) which is known to
n points on the surface of the conductor are chosen and it is provide results with high accuracy.
required that at all of these points the potential resulting from the
superposition of the charges is equal to the conductor potential.
This results in n equations constructed with n unknown line
charge variables. The line charges can be calculated by matrix
inversion and thus the electric field can be also calculated.
= (4)

Where: [λ] is the line charges matrix

[P] is the Maxwell potential coefficient matrix
[U] is the potential matrix

A CSM software tool was developed which can be used to

calculate the conductor surface gradients of arbitrary tower
setups (Fig. 1). Field calculations for hybrid tower setups are
also covered by this tool. To assess the hybrid tower calculation
performance, published results from HEDTKE et al. were used for
comparison [9]. Surface gradients were calculated and the
results show that a DC offset occurs in the AC system and that
Fig. 2. Indonesian AC 500 kV Tower Setup
 Fig. 3 shows the results of the surface gradient calculation
using MPCM, MARKT and MENGELE’s method, CSM and FEM.
The results are the average of the maximum surface gradients
occurring on each sub-conductor of phase R (system 1) for each
time step. The RMS value can be obtained by dividing the
results with √2 . The deviation of MARKT and MENGELE’s
method, CSM and FEM is less than 0.2 kV/cm. However,
simulation times are considerably higher for CSM and FEM. It
can be said that all of the investigated methods have a good
accuracy and can therefore be used for surface gradient
calculations. However, further research has also shown that for
a higher number of sub-conductors MPCM and MARKT and
MENGELE’s method are no longer accurate and it is advisable to
choose CSM or FEM based approaches

Fig. 4. Test setup for acoustic noise measurements on bundle conductors

The test object for the measurements presented in this

contribution was an ACSR conductor with a diameter of
25.9 mm and with a hydrophilic surface treatment which was
arranged in a quad bundle configuration with a spacing of
400 mm.
Due to the noise’s stochastic behavior, a sufficiently long
exposure time is needed. During the step tests, each voltage level
was applied for at least 30 s. The background noise in the high
voltage laboratory is fairly low and amounts to approx.
22 dB(A). However, the rain noise etc. raises the background
level to about 28 dB(A). The step tests started from 140 kV
which results in Emax= 12.49 kV/cm for the investigated
conductors and the highest voltage level was 270 kV which has
Fig. 3 Comparison of Surface Gradient Calculation Methods Emax= 24.09 kV/cm. The voltage levels were raised in steps of
10 kV.
III. AUDIBLE NOISE MEASUREMENT The measuring equipment consisted of two class 1
One of the main goals of this paper was to compare the microphones and sound level meter according to IEC 61672 and
results of AN measurements from the laboratory with the results a partial discharge measurement according to IEC 60270
obtained from predictions equations. Prediction equations are (fC = 100 kHz, Δf = 9 kHz) as illustrated in Fig. 5.
used to calculate the sound power of a specific bundle. Via a
simple propagation model, the sound power is then used to
calculate the immission level at a certain spot. In the laboratory
measurements, however, only the sound pressure level caused
by corona noise can be determined. To be able to compare
laboratory results and predictions equations, the laboratory
results had to be converted to sound power levels as well which
is described further below.
A. Audible noise measurements in high voltage laboratory
At Graz University of Technology, such noise emission tests
are usually carried out on single conductors or conductor
bundles with a length of 10 m which are mounted at a height of
h = 3 m [1]. Constant precipitation rates from 1 – 20 mm/h are
ensured with a special artificial rainfall simulator which is
mounted above the conductors. The ends of the conductor
samples are shielded with corona rings to ensure that the
measurement setup itself is free of any discharges. The
laboratory setup can be seen in Fig. 4. Fig. 5. Test setup for acoustic noise measurements on bundle conductors
B. Results of audible noise measurements
80
To be able to compare the measurement results with
prediction values, the bundle conductor’s sound pressure levels 70
had to be converted to sound power levels. The conversion was
performed according to the methodology described by SAMES 60
and GOOSSENS and the recommendations given in ISO 3744

Lw' in dB(A)
[10,11]. The measured noises were corrected regarding 50

transformer noise and rain noise. Furthermore, an environmental 40

correction was carried out using a reverberation time method
which considers the equivalent sound absorption area of the 30
room which was derived from the SABINE reverberation time TUG
equation using the measured reverberation time: 20 BPA
EPRI
= 0.16 (5) 10
12 13 14 15 16 17 18 19 20 21 22 23 24 25
Emax,max in kV/cm
Where: V is the volume of the test room in m3
T is the reverberation time in s Fig. 7. Comparison between the sound power levels obtained from laboratory
experiments and prediction equations (AC, precipitation rate: 2 mm/h)
An averaging process was applied because the sound
pressure level was measured at two positions. From the
80
corrected sound pressure level, the sound power level emitted
by conductor can be obtained. The processes of sound power 70
calculation can be seen in Fig. 6.
60

Lw' in dB(A) 50

40

30
TUG
20 BPA
EPRI
10
12 13 14 15 16 17 18 19 20 21 22 23 24 25
Emax,max in kV/cm

Fig. 8. Comparison between the sound power levels obtained from laboratory
experiments and prediction equations (AC, precipitation rate: 5 mm/h)

80
Fig. 6. Sound Power Level Calculation Process
70
All calculations were performed for each one-third octave
band from 100 Hz to 20 kHz individually. Lw’ denotes the sound 60
power level per unit length of conductor (m). It was calculated
Lw' in dB(A)

50
from Lw which was emitted from 6 m of wet conductor. ISO
3744 uses a reference power of 1 pW. Therefore, all sound 40
power level data provided in this paper also used this reference.
30
Fig. 7 shows the sound power results at the function of
TUG
surface gradient which was obtained from laboratory 20 BPA
measurement with 2 mm/h rain rate and the predictions equation EPRI
from EPRI [3,12] and BPA [13]. The results for other rain rates 10
12 13 14 15 16 17 18 19 20 21 22 23 24 25
(5 mm/h and 8 mm/h) can be seen in Fig. 8 and Fig. 9. Emax,max Emax,max in kV/cm
denotes the peak value of the maximum surface gradient. It was
obtained from time step simulations which also takes the non- Fig. 9. Comparison between the sound power levels obtained from laboratory
experiments and prediction equations (AC, precipitation rate: 8 mm/h)
uniform distribution of surface charges into account. It can be
seen that the sound power levels obtained from laboratory
results (TUG) are slightly higher than BPA and EPRI
predictions.
 The comparison results for DC are shown Fig. 10. BPA and And for EPRI approach as follow [3]:
EPRI prediction methodologies assume that the highest levels of . .
corona noise at DC occur during fair weather conditions and = 10 log ∑ antilog (7)
suggest subtracting 6 dB(A) from fair weather results to obtain
foul weather corona noise levels. The sound power levels It has to be noted, that for the equations above, BPA uses a
derived from the laboratory measurements agree well with the reference power of 1 pW/m whereas EPRI uses 1 W/m. The
prediction values for surface gradients above 20 kV/cm. SLA reference pressure is 20 µPa.
However, for lower surface gradients, the deviations are higher The Danube 400 kV AC line tower setup can be seen in
but those are of limited relevance for realistic OHL operation. Fig. 11 and its maximum surface gradient and sound power
Furthermore, BPA and EPRI publications state that AN levels predictions (TUG) can be observed in Table 2. The predictions
emitted under negative polarity are more or less negligible. The were performed for rain rates of 5 mm/h. Maximum surface
gathered laboratory results, however, do not support this gradients on phase S1 and R2 are higher than other phases which
assumption since AN levels were almost equal under positive results in higher corona noise levels. Furthermore, BPA and
and negative test voltage polarities. EPRI were used for AN prediction. The average deviation
between laboratory prediction (TUG) and BPA is
60 approximately 4 dB(A) and 7 dB(A) to EPRI prediction. The
total A-weighted audible noise level at the immission point is
55
approx. 5 dB(A) higher for the TUG prediction than for BPA
50 and EPRI predictions.
45
Lw' in dB(A)

40
DC (+) TUG 8 mm/h
35 DC (-) TUG 8 mm/h
DC (+) TUG 5 mm/h
30
DC (-) TUG 5 mm/h
25 DC (+) TUG 2 mm/h
DC (-) TUG 2 mm/h
20 DC (+) BPA Foul Weather
DC (+) EPRI Foul Weather
15
12 13 14 15 16 17 18 19 20 21 22 23 24 25
Emax,max in kV/cm

Fig. 10. Comparison between the sound power levels obtained from laboratory
experiments and prediction equations (DC)

Regarding the fact that the sound power levels derived from
laboratory results are overall slightly higher than the EPRI and
BPA prediction results, it has to be pointed out that EPRI and
BPA equations were designed for moderately aged conductors
(2 to 3 years in operation). Generally speaking, aged conductors
are expected to be more quiet than new conductors due to their
hydrophilic surface properties [1]. The treated conductor used in
this experiments should behave like an aged conductor. Fig. 11. Danube AC 400 kV Tower Setup
However, in this case, the conductor treatment was
unsatisfactory which results in higher corona noise levels.
TABLE II. AUDIBLE NOISE PREDICTION OF DANUBE 400 KV AC
IV. AUDIBLE NOISE PREDICTIONS OF DIFFERENT TOWER SETUP
OVERHEAD TRANSMISSION LINES
Lw’ in dB(A) ref 1 pW/m
The CSM software tool can be employed to calculate the
E
maximum surface gradient for arbitrary line designs. The sound Line
kV/cm
TUG BPA EPRI
power values obtained from laboratory experiments can then be 5 mm/h
used for AN predictions. In this paper, audible noise predictions
are performed for Danube AC 400 kV and planned Indonesian R1 20.15 62.44 57.41 55.43
DC 500 kV tower setups. The sound pressure level at certain T1 17.09 54.04 48.85 43.64
point (50 m from center of tower and 2 m above ground) was
S1 21.27 65.01 60.24 58.92
also predicted. The total A-weighted sound level (SLA) for
multiple lines can be calculated using BPA equation below [13]: R2 21.14 64.74 59.93 58.54
. . T2 16.99 53.72 48.54 43.17
= 10 log ∑ antilog (6)
S2 20.20 62.56 57.55 55.60
Where: z is number of lines
SLA in dB(A)
Lw’ is A-weighted acoustic power level ref 20 µPa
44.8* 39.91 39.70
R is radial distance measured point to phase *Calculated with BPA propagation model
 Another prediction was performed for a planned Indonesian and 7 dB(A) has been noted in the case of AC. In the case of DC,
500 kV DC line (Fig. 12). Different rain rates of 2 mm/h, the deviation amounted to approx. 4 dB(A). The investigations
5 mm/h and 8 mm/h were considered. To predict DC noise in have shown that in the case of DC audible noise, it is advisable
rainy conditions, 6 dB(A) were subtracted from DC fair weather to also consider negative pole audible noise emission which is
equation of BPA and EPRI. The predictions derived from the neglected by prediction equations so far.
laboratory results (TUG) show a deviation of approx. 1-2 dB(A)
from the BPA results and approx. 4 dB(A) from EPRI results ACKNOWLEDGMENT
(Table.3). For the total A-weighted audible noise at the The authors would like to thank the PT. PLN (Persero) Indonesia
immission point, TUG also considers the sound power emitted for their support and providing tower data.
by the negative pole which results in an increase of approx.
4 dB(A) compared to positive pole immission. REFERENCES
[1] O. Pischler and U. Schichler, “Influence Of The Conductor Surface On
Over Head Line Audible Noise Under Foul Weather Conditions”, The
20th International Symposium on High Voltage Engineering, 2017.
[2] IEEE Corona and Fields Effect Subcommittee, “A Comparison of
Methods for Calculating Audible Noise of High Voltage Transmission
Lines”, IEEE Transaction on Power Apparatus and Systems, Vol. PAS-
101, No. 10, pp. 4090–4099, 1982.
[3] J. LaForest, L. Zaffanella and M. Comber, “Transmission Line
Reference Book: 345 kV and Above”, EPRI, 2nd Edition, 1982.
[4] M. Temoshok, “Relative Surface voltage Gradients of Grouped
Conductors”, Transactions of the American Institute of Electrical
Engineers, Vol. 67, No. 2, pp. 1583–1591, 1948.
[5] G. E. Adams, “Voltage Gradients on High-Voltage Transmission Lines”,
Transactions of the American Institute of Electrical Engineers, Vol. 74,
No. 3, pp. 5–11, 1955.
[6] IEEE Corona and Field Effect Subcommittee, “A Survey of Methods for
Fig. 12. Planned Indonesian DC 500 kV Tower Setup Calculating Transmission Line Conductor Surface Voltage Gradients”,
IEEE Transactions on Power Apparatus and Systems, Vol. PAS-98,
TABLE III. AUDIBLE NOISE PREDICTION OF INDONESIAN 500 KV DC
TOWER SETUP No. 6, pp. 1996–2014, 1979.

Lw’ in dB(A) ref 1 pW/m [7] M. Abou-Seada and E. Nasser, “Digital Computer Calculation of the
Line E TUG BPA EPRI Potential and Its Gradient of a Twin Cylindrical Conductor”, IEEE
kV/cm 8 5 2 Foul Foul Transactions on Power Apparatus and Systems, Vol. PAS-88, No. 12,
mm/h mm/h mm/h Weather Weather pp. 1802–1814, 1969.
DC + 25.24 52.24 52.70 53.72 51.71 56.67
[8] H. Steinbigler, H. Singer and P. Weiss, “A Charge Simulation Method
DC – -25.24 52.10 52.34 52.81 n.a n.a
for The Calculation of High Voltage Fields”, IEEE Transactions on
SLA in dB(A)
30.05* 30.38* 31.09* 25.65 32.15 Power Apparatus and Systems, Vol. PAS-93, No. 5, pp. 1660–1668,
ref 20 µPa
*Calculated with BPA propagation model 1974.
[9] S. Hedtke, M. Pfeiffer, C. M. Franck, L. Zaffanella, J. Chan, and J. Bell,
V. OUTLOOK
“Audible Noise of Hybrid AC / DC Overhead Lines: Comparison of
The unsatisfactory performance of the conductor used for the Different Prediction Methods and Conductor Arrangements”, EPRI's
presented experiments has shown that surface treatments and the High-Voltage Direct Current & Flexible AC Transmission Systems
corresponding terminology and testing procedures lack Conference, Vol. 1, pp. 1–8, 2015.
standardization which has to be worked on in the future to assure
[10] P. Sames and M. Goossens, “Messtechnische Felduntersuchungen zu
consistent quality for utilities.
Koronageräuschen”, url: https://www.hlnug.de/fileadmin/shop/files
VI. CONCLUSION /Schriften_Laerm_587.pdf.
OHLs are known to emit AN during foul weather conditions. [11] ISO3744, “Acoustics - Determination of Sound Power Levels of Noise
AN prediction equations rely on accurate surface gradient Source Using Sound Pressure”, 1994.
calculations. Comparisons using the example of an Indonesian [12] G. Johnson, B. Clairmont, L. Zaffanella and S. Zelingher, “The Effect of
500 kV AC line have shown that commonly used approaches for HVAC - HVDC Line Separation in a Hybrid Corridor”, IEEE
field calculations yield reliable results. Single phase audible Transaction on Power Delivery., Vol. 4, No. 2, pp. 1338–1350, 1989.
noise laboratory measurements were carried out on a four-
conductor bundle under AC as well as DC stress. The gathered [13] V. Chartier and R. Stearns, “Formulas for Predicting Audible Noise from
results (TUG) were used for comparisons with the well-known Overhead High Voltage AC and DC Lines”, IEEE Transactions on
BPA and EPRI AN prediction equations. A deviation between 4 Power Apparatus and System., Vol. PAS-100, No. 1, pp. 121–130, 1981.