21, rue d’Artois, F-75008 PARIS http : //www.cigre.org

C4-102

CIGRE 2008

Measurement of Radio Frequency Interference from High Voltage Substations Backgrounds and Considerations for Future Emission Requirements L.-E. JUHLIN*, J. SKANSEN, L. KOPPARI, E. PETERSSON, P. STENUMGAARD Power Systems, ABB AB SVK STRI AB FOI Sweden

SUMMARY Radio frequency interference (RFI) from power lines is well documented in the literature and summarized in the CISPR 18 series [1]. On the contrary, very little is documented regarding RFI from high voltage substations. There are very few reports from measurements [2],[3]. The only European standard regarding RFI emission from substations is IEC 62236-2 for railway applications. Worldwide there are some standards regarding RFI from substations. However, an analysis of the texts shows that they are written under the assumption that corona is the only significant source of RFI from a HV substation and the span between the requirements in the standards is extremely wide. Besides, measured RFI from HVDC and FACTS, reported in [3], deviates significantly from RFI due to corona. Therefore, in order to determine the RFI emission levels from normal HV substations, extensive measurements of RFI have been performed in five randomly selected 400 kV substations in Sweden. The ages of the substations are felt to be representative. There have been no complaints regarding RFI from any of the substations. To measure is to know a bit more. The measurements show that the high frequency RFI is much higher than expected, especially in dry weather and old installations. The reason is considered to be gap discharge, i.e. sparking. Probably this is an aging phenomenon, such as corrosion of metallic contacts for equalization of potentials. Another surprise was that the RFI from low voltage or telephone cables was in the same order or higher than the RFI from the 400 kV connecting lines in the frequency range 9-200 kHz. Besides, the frequency characteristic is very irregular and measurements at single frequencies give uncertain results. Based on the measurement results, IEC 62236-2 is considered as a relevant standard also for HV substations, although the requirements for frequencies below 150 kHz may be unjustifiably conservative. The theoretical expression for attenuation versus distance is reported in the paper and confirmed by measurements. At distances shorter than the physical dimension of the substation, the attenuation may be low and the local variations due to interference may be significant. Consequently, it is recommended that the limits in IEC 62236-2 be recalculated to a measuring distance of 100-200 m.

KEYWORDS Radio Interference – EMC – High Voltage – Corona – Gap Discharge – HVDC – Power Electronics – Substation – Measurement – Distance Attenuation

* E-mail: [email protected]

1. INTRODUCTION Since the enforcement of the EU EMC Directive 1996 there have been extensive activities regarding development of emission and immunity standards for various products and equipment. However, emission from high voltage substations has been an exception and the available documentation of Radio Frequency Interference (RFI) from substations is very poor. In contrast RFI due to corona in high voltage power lines is well covered in the literature and in the CISPR 18 series [1]. In order to fill the documentation gap regarding expected level of RFI from high voltage substations, this paper reports results from measurements of RFI from some typical 400 kV and 130 kV substations, which are built in accordance with present engineering praxis. The selected stations are considered to be representative regarding design and in none of the cases there have been any complaints regarding RFI. The purpose of the paper is to provide relevant background information to the standardisation bodies regarding future development of emission standards for RFI from high voltage substations. The only applicable European standard for substations is IEC 62236-2 for railway applications, as noted in [2]. This standard covers RFI for converters for railway power supply. Besides, world wide there are some standards for substations. However, those standards do not harmonize. The comparison in figure 5 shows that the span in the emission requirement is 60-70 dB depending on which standard is applied. This means a factor of 1000-3000 in amplitude. Such a large span indicates that there is a lack of reliable background information, i.e. documented measuring results from typical installations. The concern is that standards, not based on realistic assumptions, may lead to unnecessary cost increase not motivated by a real risk for radio interference. Besides, the requirement must also reflect the impact zone of substations regarding RFI [3],[4]. HV substations are few and remotely located and their overall impact zone on RFI is much less than low voltage mass produced equipment located in domestic buildings. The measurement results show that the RFI from the substations was higher than expected, but in general below the levels specified in IEC 62236-2. Especially the level in the higher frequency range 1 MHz to 1 GHz was higher than expected. The reason may be that with the large amount of energised equipment in a substation it is difficult to avoid local discharge or local sparking due to bad contacts. Another unexpected observation is that low voltage lines may be a significant source of RFI, at least in the lower frequency range. One important observation is that the frequency characteristics of the RFI from a substations and HV lines are quite irregular. The variations were up to 20 dB and sometimes even more. Consequently, measurement at a single frequency is too unreliable for verification of the RFI level from a substation.

2. PERFORMED MEASUREMENTS 2.1 Selection of sites Five substations were randomly selected for the measurements. The criteria were that they should be well representative of today’s technology but still represent some diversity and not be too remote, for practical reasons. All five substations are connected to the Swedish 400 kV grid and they have been subject for normal maintenance and replacement of worn out equipment. One substation is a 400/130 kV transformer station located in the middle part of Sweden with open switch yards. The 400 kV part has been in operation since around 25 years ago. The 130 kV part is originating from the mid-1900s. The substation serves as power supply for the surrounding area, including a major steel mill. Another substation is an HVDC converter station feeding a 600 MW line commutated HVDC converter. It is built as an open switchyard with the converter valves installed in a valve hall. The substation has been in operation since the year 2000. The substation is located at a distance of about 200 m from an older 400 kV switchyard. The most modern substation is a major connection point in the southern part of the Swedish 400 kV grid. The 400 kV substation is an open switchyard, which has been in operation since 2004. The fourth

1

substation is another major connection point of the southern part of the Swedish 400 kV network, designed as an open switchyard. It has been in operation since 1974. The last substation is originally a feeder station for a mid-size city, recently extended to also being a connection point in the central part of the Swedish 400 kV grid. The 400 kV part is a GIS switchyard being in operation for about 20 years. The 220 kV, 130 kV and 40 kV switchyards are of open design being in operation for more than 60 years.

2.2 Measurement procedures Basically the measurement procedure has been in accordance with CISPR 16-1-1, although the measurements have been somewhat extended. In some locations comparison has been made between the results with four different detectors, Peak, Quasi-peak, RMS and Average. The bandwidths in accordance with CISPR 16-1-1 have been used. A Rohde&Schwarz EMC test receiver ESCS 30 has been used. For the frequency range 9 kHz to 30 MHz a loop antenna Rohde&Schwarz HFH2-Z2 has been used and for the frequency range 30 MHz to 1 GHz a Bilog antenna Shaffner CBL611C has been used. Measurement has been performed at both different directions and different locations around the substations and along connecting 400 kV lines.

2.3 Measurements with different bandwidth For verification of RFI with digital radio communication a broadband limit is needed [4],[5]. The criterion under consideration is based on an RMS detector. However, due to the difficulty to discriminate the background noise, it is considered necessary to perform in situ verification with the present bandwidths recommended in CISPR 16, but with an RMS detector and recalculate the values to the broader bandwidth. The CIGRE/CIRED JWG has proposed a method to convert measurement at a smaller bandwidth to a measurement result with a broader bandwidth [4]. The method has been confirmed by measurement with bandwidths 9 kHz, 120 kHz and 1 MHz in a frequency range without disturbing background noise. The conversion is valid for RMS detectors only. The performed measurement results with the small bandwidth are divided into n intervals with the amplitude E1i in the midpoint of the interval i with the bandwidth BWI of each interval. The measurement is performed with the bandwidth BW1. The measurement result at the midpoint of the interval is M1i. We have:

 BWI M 2 = 10 ⋅ log   BW1

n

∑ BWI = n ⋅ BWI = BW2

(1)

and

i =1

n

∑10

( M1i / 10 )

i =1

  

(2)

Where M2 is the estimated result with bandwidth BW2. If the bandwidth of the intervals BWI differs from the measurement bandwidth BW1 the correction with BWI/BW1 is needed for estimation of the power in each interval. This procedure is applied for three examples as shown in figure 3. 80

100 PK RMS AV QP

90 80

60 50

dBµV/m

dBµV/m

70 60 50

40 30

40

20

30

10

20

0

10 10 kHz

PK RMS AV QP

70

−10

100 kHz

1 MHz Frequency

10 MHz

100 MHz

1 GHz Frequency

Figure 1. Measurement in the first substation, at a distance of 45 m from closest phase conductor (400 kV), with four detectors; peak (PK), average (AV), RMS and quasi-peak (QP). QP scanned with larger frequency steps.

2

3. MEASUREMENT RESULTS 3.1 Characteristics of the emission Figure 1 shows that the frequency characteristic of the RFI is quite irregular with significant variation of the RFI level versus frequency. This is valid for the complete frequency range and for all performed measurements. Thus, measurement at single frequency does not give a reliable result. Figure 2 shows the time domain characteristic of the antenna signal for the measurement in figure 1, 0.009-30 MHz. The peaks are due to discharge activities causing broad band RFI emission, some hundreds of MHz.

3.2 Comparison between different types of detectors Figure 1 also shows a typical example of variation of levels versus type of detectors. As the span in amplitude between the peak detector and the other detectors depends on the time domain characteristic of the signal, i.e. the pulse repetition frequency, the span is not the same in all measurements. The smallest recorded span is half of the span in figure 1. Due to limited time, the quasi-peak measurement is performed with larger frequency steps. 0.1

60

0.08

55

Calculated, BW 3 MHz Calculated, BW 1 MHz Calculated, BW 120 kHz Measured, BW 1 MHz Measured, BW 120 kHz Measured, BW 9 kHz

0.06

Level (dBµV/m)

Antenna output (V)

50

0.04 0.02 0 −0.02

45 40 35

−0.04 30

−0.06 25

−0.08 −0.1 0

2

4

6

8

10 12 Time (ms)

14

16

18

20

Figure 2: The time domain output signal from the loop antenna at the measurement in figure 1.

20 46

47

48

49 Frequency (MHz)

50

51

52

Figure 3: Measurement with different bandwidths and RMS detector, compared with calculated values.

3.3 Measurement with different bandwidth In figure 1 the radio transmitters can be identified as sharp high peaks in the frequency domain. The broader peak at 300 kHz is due to PLC communication on the power line. For frequency ranges without radio transmitters, it is possible to measure with the broader bandwidths relevant for RFI with digital radio communication. However, in a frequency range close to radio transmitters it would not be possible to discriminate the background noise if a broadband receiver would be used, e.g. the frequency range 85-110 MHz in figure 1. Therefore, it is of interest to recalculate the measured value with a small bandwidth to a wider bandwidth. Figure 3 shows that the method outlined in section 2.3 is adequate. Measurement is performed with the bandwidth 9 kHz, 120 kHz, and 1 MHz. The agreement between the measured results and the calculated results shows that the conversion to a broader bandwidth works adequately. Due to limitations in the measuring equipment it was not possible to measure with the bandwidth of 3 MHz.

3.4 Attenuation versus distance In order to transform measured and theoretical values of the electric field strength from one distance to another, a model with acceptable accuracy must be used. In [6], a model that combines two classical models; the electric dipole and the two-beam model, has been shown to have acceptable accuracy compared to measured results. The model has some mathematical complexity but the results can easily be summarized in a graph, see figure 4. As this graph also takes the impact of the earth into account, the curves are more complicated than the well known attenuation curves for dipoles in an open space. It is not an easy task to measure the attenuation versus distance for such an extended source as an HV substation. At a short distance, the antenna sees only a small fraction of the source and at a larger distance the antenna sees a larger part of the source. Consequently, the attenuation is significantly

3

lower than the theoretical value, valid for a point source, when the distance to the source is shorter than the extension of the source. Besides, radiation from different parts of an extended source may interact giving local variations. However, in the cases when the radiation was caused by an identified point source, the attenuation versus distance confirms the theoretical rules. This was also the case at a larger distance, when the substation could be seen as a concentrated source, even if the background noise impacted the measurement at larger distances. Anyhow, the measurements confirm the theoretical curves in figure 4 provided that the distance is larger than the physical size of the source. Transmitter height 1 meter over "good ground" 100 kHz 1 MHz 3 MHz 10 MHz 30 MHz

−20

110

90

−40

80

1/r −60

−80

−100

3

70 60 50 40 30

2

1/r

1/r

20

−120

−140 0 10

P-OLD P-NEW P-HVDC QP-OLD QP-NEW QP-HVDC IEC 62236 AS/NZS

100

E [dBµV/m], QP

Attenuation of veritcal E−field component [dB]

0

10

1

10

2

10 Distance from source [m]

3

4

10

0 0,01

10

Figure 4: Attenuation of vertical E-field component (just above ground) for different frequencies versus distances from source

0,1

1

10

100

1000

Frequency [MHz]

Figure 5: Maximum measured levels in a new, three old and one HVDC substation, P and QP at a distance of 1525 m compared with the limits in some standards.

For the guided wave travelling along the connected line the measured attenuation was around 15 dB/km at 10 MHz, 9 dB/km at 5 MHz, and 3 dB/km at 1 MHz, versus the distance along the line to the substation. The attenuation versus the distance to the outer phase conductor follows reasonably well equation (3), where E1 and E2 is the field levels at distances r1 and r2, c is the speed of light and f is the frequency. Measurements have been performed at distances of 15 to 120 m from the line. 2

    c c  = E1 ∗ r1 ∗ 1 +   E 2 ∗ r2 * 1 +  2 ⋅ π ⋅ f ⋅ r 2 ⋅ π ⋅ f ⋅ r 1  2   

2

(3)

3.5 Field level at 15 to 25 m from a bus in substations Figure 5 gives an overview of the highest values measured at a short distance to an energized bus, 1525 m. Both the peak values and the quasi-peak values are shown, and as comparison, the quasi-peak limits in IEC62236-2 and AS/NZS 2344:1997 are shown. Both standards stipulate limits for such a short distance. In the older stations the limit stipulated in IEC 62236-2 is exceeded by 10 dB in the frequency range 30-200 MHz, probably due to gap discharge. No substation satisfies AS/NZS 2344:97. The HVDC substation has much higher content of RFI below 150 kHz than the other substations, close to the limit in IEC 62236-2. It is considered that the nearby older 400 kV switchyard contributes to the RFI level above 0.4 MHz, which is present also when the HVDC is out of operation.

3.6 Field level at 100 m from substations The field levels are significantly attenuated at a larger distance from the HV conductors, as shown in figure 6. This figure shows the highest measured peak and quasi-peak values at a distance of 100 m from the HV conductor. As comparison the limits stipulated by AS/NZS 2344:1997 are shown. It should be noted that these levels are applicable for the requirement at the location of the radio receivers. The RFI from the new substation was below the AS/NZS limits at a distance of 100 m. In fact the emission was below the background noise, except in the frequency range 30-100 MHz. The emissions from the older substations are 15-25 dB above the AS/NZS levels at 100 m due to the gap

4

discharge activities. The higher levels in the HVDC substation for frequencies above 2 MHz are due to the nearby older substation. As comparison also the measured value with a peak detector from a nearby telecommunication line is shown, measurement distance 25 m. No QP-values are available. 90 80 70

50 40

P-1.6 km P-20 km P-HVDC QP-1.6 km QP-20 km QP-HVDC AS/NZS P-TELE

90 80 70 E [dBµV/m]

60 E [dBµV/m]

100

P-OLD P-NEW P-HVDC QP-OLD QP-NEW QP-HVDC AS/NZS P-TELE

60 50 40

30

30

20 10 0,01

20

0,1

1

10

100

1000

Frequency [MHz]

Figure 6: Maximum measured levels in a new, three old and one HVDC substation, P and QP at a distance of 100 m compared with the limits in one standard.

10 0,01

0,1

1

10

Frequency [MHz]

Figure 7: Maximum measured levels at a 15 m distance from the connected lines at a distance of 1.6 km and 20 km from the substation, P and QP. HVDC measured 5 km from the substation.

3.7 Field levels at 15 m from connected lines Figure 7 shows the highest measured values at a distance of 15 m from connecting lines for the old substations and the HVDC substation. Comparison is made with the levels stipulated by AS/NZS 2344:1997. The increased level at frequencies above 0.15-5 MHz due to gap discharge activities is notable also at a distance of 1.6 km. At a distance of 20 km the levels are down to normal corona levels. In the frequency range 9 kHz to 150 kHz the presence of an HVDC converter increased the field levels, measured at a distance of 5 km from the substation. It is notable that the RFI level from the telephone line is in the same order as the RFI due to the HVDC converter in this frequency range.

3.8 Unexpected sources The measurement has revealed two unexpected sources. The first is the emission due to sparking or gap discharges, most pronounced in the older substation, but to some degree also present in a new substation in dry weather. It is likely that this is an ageing phenomenon. The metallic contact is impacted by corrosion. This source may give dominant emission in the frequency range 150 kHz to above 1 GHz. The other unexpected source was the low frequency RFI from the low voltage cables combined with telephone cables. Probably the emission is due to digital services via the telecommunication line, i.e. ADSL. It seems less likely that the emission is caused by the 400/230 V cables. Anyhow, the RFI level from the low voltage circuits is in the same order or higher than the RFI from the 400 kV lines for frequencies below about 200 kHz.

4. LESSONS LEARNED There are some lessons learned by the performed measurements. One is that the RFI levels from 400 kV lines in locations remote from substations corresponds reasonable well with the expected level of corona according to figure B8 in CISPR 18-1 -1982 [1]. However, the RFI levels from substations are significantly higher. Broadband RFI in the frequency range 1 MHz to 1 GHz was noted in all substations, indicating gap discharge or sparking activities. In some locations discharge activities were identified. This broadband RFI was significantly reduced in wet weather. The level of the broadband RFI was higher in older substations, maybe due to an ageing process.

5

In all substations the frequency characteristic of the RFI was quite irregular with fast variation versus frequency. The error when measuring at a single frequency can be in the order of 20 dB or more. Besides, it is not possible to convert the measured RFI level at one frequency to the RFI level at another frequency. The RFI level at all frequencies of interest must be explicitly measured. By chance it was noted that also low voltage cable lines emits RFI. A LV cable line, combined with a telephone cable, passed closed to a measurement position and in the frequency range 9 kHz to 150 kHz, the RFI from the LV cable line was 20-30 dB higher than the RFI from the 400 kV OH line at a distance of 15 m. The maximum level was 82 dB [µV/m] at a frequency of 16 kHz. The RFI from the LV cable line was above the background noise level up to about 0.4 MHz. At a substation with a line commutated HVDC converter of 600 MW only the RFI level below 0.4 MHz was impacted by operation of the converter. For higher frequencies other sources dominated. The attenuation of the guided wave of RFI along a connecting line increases with frequency. At a distance of 1.6 km from the substation no RFI above 10 MHz could be seen (attenuation about 25 dB) and at a distance of 20 km the attenuation at 2 MHz was more than 30 dB. For measurement with an RMS detector it is shown to be valid to recalculate the RFI level to a broader bandwidth. There was a good agreement with measured values. The measurements were made in a frequency range without disturbing radio transmitters.

5. RECOMMENDATIONS FOR FUTURE EMISSION STANDARDS The peak detector always gives higher values than the other detectors, quasi-peak, RMS and average. Thus, for saving time, it is recommended to start the measurement procedure by scanning the frequency range with a peak detector. Measurement with other detectors is needed only if the reading of the peak detector is above the limit, regardless of specified detector. In order to verify conformance with the requirement, the specified limit must be well above the general background noise level. Consequently, the measurement distance must not be too long, as the attenuation with distance is very significant. On the other side, the measurement distance must not be too short to be relevant for the interference to the surroundings. In addition, the HV switchyard is not a point source, but a distributed source of RFI and the intensity of the source in most cases varies with the location. Therefore, the measurement distance must not be shorter than the physical size of the installation. A reasonable measurement distance for a 400 kV substation is 100-200 m. As the closest part of the source is dominating, the reference point for the measurement distance should be the closest active part, i.e. closest HV equipment. In case of an installation of power electronic equipment, i.e. HVDC and FACTS, all connected bus structures and enclosed buildings are considered as active parts. For verification of the guided wave RFI via the connecting lines, the measurement is recommended to be performed 1-2 km from the substation at a distance of 25-50 m from the outer phase conductor The measurement results show that the RFI levels vary very irregularly with the frequency. The uncertainty when measuring at a single frequency is significant, 10-20 dB. Therefore, all RFI verification shall be performed as frequency sweep. For verifying the RFI level at a frequency fc the frequency sweep shall cover at least the frequency range 0.5*fc to 2*fc. Note: When verifying the Quasi-peak or RMS level of a RFI peak identified with a peak detector, a shorter sweep may be acceptable. Installations of high power electronics such as HVDC and FACTS may have peaks of RFI level at any frequency due to local resonances. Thus, for installations of power electronic equipment, the complete frequency range has to be scanned. At verification of the corona level, it is recommended to sweep the frequency range from 0.3 to 1 MHz. For verification of moderate sparking it is recommended to sweep the frequency ranges 1-5 MHz and 30-100 MHz. It must be noted that dry weather may be the worst condition for sparking due to bad contacts.

6

For verification of RFI with digital radio communication a broadband limit is needed [4],[5]. The criterion under consideration is based on an RMS detector. However, due to the difficulty to discriminate the background noise, it is recommended to perform in situ verification with the present bandwidths recommended in CISPR 16, but with an RMS detector and recalculate the values to the broader bandwidth.

6. DISCUSSION AND CONCLUSIONS The measurement of radio frequency interference from five randomly selected representative 400 kV substations shows that RFI emission from substations is much more complex than RFI emission from HV lines. The RFI from lines are more or less completely related to corona activity, which can fairly well be predicted based on the geometrical dimensions. It was found that in all substations there was an emission of broad band RFI which is considered to be due to gap discharge, or sparking. In some cases specific discharge activities were identified. The gap discharge activity was most pronounced in older substations, especially in dry weather. However, some activity was also found in a new substation in dry weather. It must be noted that there have been no RFI complaints for any of the substations. The RFI from the substation reaches the surroundings both as direct waves and as guided waves. At a distance of 1.6 km the RFI above about 5 MHz is eliminated by attenuation. At 20 km most of the RFI above 150 kHz is negligible. However, the attenuation versus line length from the substation for frequencies below 150 kHz is quite low, but the attenuation versus the distance from the line is significant. It was also found that low voltage/telecommunication cables gave RFI in the same order as the highest recorded RFI from 400 kV lines in the frequency range 9-200 kHz. The theoretical formula for attenuation of the direct wave RFI has been verified by measurement. However, this formula is only valid for distances larger than the physical extension of the source. At smaller distances the attenuation is significantly lower and there may be significant local variations. Consequently, verification of RFI shall not be performed at a distance shorter than the substation size. The standard IEC 62236-2 is considered as a relevant standard regarding maximum allowed RFI from a substation. However, the levels should be recalculated to a measuring distance of 100 to 200 m from the active parts, i.e. the HV equipment. Another aspect is that the limit for frequencies below 150 kHz seems unnecessarily conservative considering the RFI level from the LV/telephone circuits. Countermeasures for low frequency RFI are costly and the band 9-150 kHz is sparsely used [4].

BIBLIOGRAPHY [1] [2] [3] [4]

[5] [6]

Radio interference characteristics of overhead power lines and high voltage equipment. Part 1: Description of the phenomena. (CISPR Publication 18-1, 1982). D. Imeson, I. Glover, V. Puri, K. Walker and I. Kenney “Under scrutiny – Is the UK’s transmission system our biggest source of electromagnetic interference?” (IEE Power Engineer, October/November 2005, pp. 42-45). L.-E. Juhlin, T. Larsson, J. Skansen, E. Petersson “Considerations regarding RI limits for high voltage HVDC or FACTS stations” (Paper B4-207, Cigré 2006 session). CIGRE/CIRED Joint Working Group JWG C4.202 “Guide for Measurement of Radio Frequency Interference from HV and MV Substations. Disturbance propagation, characteristics of disturbance sources, measurement techniques and conversion methodologies.” (Under final preparation. Will be published as a CIGRE technical Brochure fall 2008). Peter F. Stenumgaard “A Possible Concept of How Present Radiated Emission Standards Could be Amended in Order to Protect Digital Communication Services” (IEEE Transactions on Electromagnetic Compatibility, vol. 46, November 2004, pp. 635-640). Albert A. Smith, JR., ”Electric Field Propagation in the Proximal Region, ” (IEEE Transactions on Electromagnetic Compatibility, no. 4, November 1969, pp. 151-163).

7