Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

CIGRE PaperC4 122

Download as pdf or txt
Download as pdf or txt
You are on page 1of 11

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/329399240

Power Quality Monitoring in Power Grids focusing on Accuracy of High


Frequency Harmonics

Conference Paper · August 2018

CITATIONS READS

0 1,740

7 authors, including:

Stefan Tenbohlen Thomas Brügger


Universität Stuttgart X
430 PUBLICATIONS 5,690 CITATIONS 8 PUBLICATIONS 12 CITATIONS

SEE PROFILE SEE PROFILE

Martin Siegel
BSS Hochspannungstechnik GmbH
51 PUBLICATIONS 555 CITATIONS

SEE PROFILE

All content following this page was uploaded by Martin Siegel on 04 December 2018.

The user has requested enhancement of the downloaded file.


21, rue d’Artois, F-75008 PARIS C4-122 CIGRE 2018
http://www.cigre.org

Power Quality Monitoring in Power Grids


focusing on Accuracy of High Frequency Harmonics

S. TENBOHLEN1, C. KATTMANN1, T. BRÜGGER2, M. SIEGEL3,


M. KONERMANN4, E. JUNGE4, J. CHRISTIAN5
1
University of Stuttgart, Germany, 2Fachkommission für Hochspannungsfragen, Switzerland,
3
BSS Hochspannungstechnik GmbH, Germany, 4Netze BW GmbH, Germany,
5
Transnet BW GmbH, Germany

SUMMARY
The assessment of power quality is an important part of the grid operators’ tasks. One emerging aspect
is the measurement of high frequency harmonics from 2 – 150 kHz, which are fed into the grid mainly
by inverters, which are built into electric vehicle chargers or PV plants. The accurate measurement of
these high frequency harmonics is a challenge, especially in high and extra high voltage grids. This
paper presents an overview of established voltage transducer technologies and their suitability for high
frequency measurements. The voltage measurement using capacitance graded bushings is presented in
detail including bandwidth measurements, temperature dependency and long-term stability. The results
indicate that this technology is particularly suited to harmonic and transient measurements.
Additionally, the necessary properties of the power quality measurement device are discussed with a
particular attention to the sample rate and the vertical resolution of the A/D-converter. Using simulations
with predefined waveforms, it can be concluded that the sample rate has a larger influence on the
accuracy of harmonic measurements than the vertical resolution.
Finally, two measurement installations are presented. One was conducted directly at an electric vehicle
charging station and shows that harmonic distortion occurs at frequencies up to at least 50 kHz. The
other was a test installation at a 380 kV transformer using the aforementioned capacitance graded
bushings, where high frequency harmonics are measurable up to 9 kHz.

KEYWORDS

Power quality, harmonics, high voltage, voltage transducer, bushings

stefan.tenbohlen@ieh.uni-stuttgart.de

1
1. INTRODUCTION
The compliance to the power quality standard EN 50160 is an objective of all grid operators. Besides
limits for the frequency, RMS voltage and flicker, one important requirement of this standard is a limit
on individual voltage harmonics up to the 25th (1.25 kHz), as well as a limit on the Total Harmonic
Distortion (THD), which is the aggregation of all harmonics up to the 40th (2 kHz). Therefore, the
measurement of harmonics up to 2 kHz in the power grid is an established procedure.
Additionally, IEC/TR 61000-3-6 contains indicative planning levels for MV, HV, and EHV grids up to
the 50th harmonic (2.5 kHz).
At the moment, there are no widely used, valid standards or planning levels for frequencies over 2.5 kHz.
In the past, higher frequencies were disregarded due to the small number of systems with a potential for
disturbances. An exact measurement is also a challenging task for voltage transducers and measurement
devices.
With the growing adoption of large power electronic devices like photovoltaic (PV) converters and
chargers of electrical vehicles, there are growing concerns about harmonic emissions in higher frequency
ranges. A higher switching frequency of power electronics generally goes along with a higher efficiency,
therefore PV and automotive inverters increasingly use switching frequencies of multiple kHz, up to
100 kHz in some cases. These high frequency voltage harmonics can disturb connected devices, increase
losses and potentially damage grid components like transformers, generators, motors, and control,
measurement and protection equipment.
Consequently, there are efforts to establish a power quality measurement standard for higher frequency
harmonics in a revision of EN 61000-4-30. As a reasonable upper bound for measurements and
investigations, 150 kHz is usually chosen because it includes the entire CENELEC-Band of frequencies
reserved for power line communication as defined in EN 50065, and starting from 150 kHz EMC-
standards like EN 55011 take over. The measurement of harmonics up to 150 kHz is challenging on
multiple levels. First, the bandwidth of the entire analogue measurement chain has to feature a sufficient
bandwidth and exhibit neither significant damping nor resonances up to the required frequency. This is
especially difficult to achieve in higher voltage levels. Chapter 2.1 discusses the suitability of currently
available options in HV and EHV grids and presents the usage of capacitance graded bushings as
coupling capacitors for high-frequency measurements.
Second, the used measurement device has to provide a sufficient sample rate to even be able to capture
the harmonics. Also, the sample rate and the vertical resolution have to be suitable to achieve a sufficient
accuracy for the high-frequency harmonics, which are usually small compared to the 50 Hz signal.
Chapter 2.2 presents a theoretic explanation and simulation results of what different sampling rates and
vertical resolutions mean for the achievable accuracy of the measurement of high-frequency harmonics.
Chapter 3 presents two test installations, one at a connection point of several electric vehicle chargers
and one at a high voltage transformer and discusses the used sensors, measurement device properties
and results.

2. ACCURACY OF HIGH-FREQUENCY HARMONICS MEASUREMENTS


2.1. Overview of Voltage Transducers
There are currently several principles of voltage dividers in use in electric grids.
Voltage Transformers or Instrument Transformers (Figure 1a) are widely used in AC grids today,
especially as part of combined voltage and current transducers. They are usually tuned to the
fundamental voltage frequency (50 or 60 Hz) and are not specified for higher frequencies. Measurements
of the frequency response for higher frequencies have been conducted by [1], [2] and [3], and the results
indicate that inductive voltage transformers are not suitable for harmonic measurements. There are even
resonant peaks in the frequency response around 4 - 5 kHz with deviations of 100% or more.
Resistive dividers (Figure 1b) are sometimes used in medium voltage grids. As the resistances used have
to be quite large (50 MΩ or more for the primary resistance), parasitic capacitances are a main concern
and limit the upper frequency these dividers are capable of. For HVDC applications, where traditional
AC voltage measurement principles fail, resistive dividers or RC dividers are the only options.
Capacitive Dividers (Figure 1c) are mainly used in extra high voltage applications. The primary
capacitance is usually formed by stacking multiple capacitive elements. They are naturally capable of

2
transmitting very high frequencies and are sometimes used to couple high frequency communication
signals. By design, they only transmit AC signals and are therefore not suitable for DC measurements.
RC Dividers (Figure 1d) are a combined capacitive and resistive divider, where the resistive branch
allows for the measurement of low frequency and DC components. They are usually used specifically
for precise power quality [4] and transient measurements [5].
Capacitive Instrument Transformers (CVTs, Figure 1e) are a combination of a capacitive divider and a
potential transformer. The additional inductance L needs to be tuned to the fundamental frequency for
an accurate measurement, and the divider ratio quickly becomes unpredictable for higher frequencies
[6].
Among those five voltage divider principles, capacitive dividers and RC dividers are the only
transducers that are applicable to high frequency harmonics measurement in high voltage grids.

Figure 1: Different principles of high voltage transducers

2.2. Voltage Measurement with Capacitance Graded HV Bushings


An elegant technique to construct a capacitive divider in an HV or EHV grid is the usage of the
capacitance graded bushings of a power transformer as the primary capacitance C1. These bushings
usually feature test taps which allow the measurement of the bushing capacitance and the dissipation
factor tan(δ), which are indicators of the overall ageing or internal damages to the bushing. The
diagnostic measurement and monitoring of these properties at the test tap is an established procedure.
By connecting a test tap adapter with an integrated capacitance C2, the bushing and the adapter form a
capacitive divider, which allows the safe measurement of voltage. The basic installation is outlined in
Figure 2.

3
Figure 2: Installation of the adapter containing the secondary capacitance C2 to the bushing

For HV/EHV bushings, the primary capacitance C1 of the bushing is usually in the range of 600 pF, so
in order to achieve a safe secondary measurement voltage of 20 V in a 420 kV grid, the necessary
secondary capacitance C2 is:
1 1
𝑈1 𝑍1 + 𝑍2 𝑗𝜔𝐶1 + 𝑗𝜔𝐶2 1 1 𝐶2
= = = 𝐶2 ( + ) = + 1
𝑈2 𝑍2 1 𝐶1 𝐶2 𝐶1
𝑗𝜔𝐶2

𝑈1 420 𝑘𝑉⁄√3
⟹ 𝐶2 = ( − 1) 𝐶1 = ( − 1) ∙ 600 𝑝𝐹 ≈ 7.3 𝜇𝐹
𝑈2 20 𝑉

This capacitance has be stable over the entire voltage and frequency range and needs to be unaffected
by temperature deviations and ageing. The choice of SMD capacitors is therefore limited to Class 1-
type capacitors (NP0), which are usually only available up to 220 nF, so around 30 of those capacitors
have to be used in parallel. Alternatively, polypropylene-film-capacitors could be used, however, they
are large in size and introduce stray inductances. While it is possible to control the properties of the
secondary capacitance C2, the primary capacitance C1 is provided by the bushing and all its properties
have to be assessed and accounted for. In the following, measurements of the bandwidth and temperature
dependency of the complete capacitive divider are discussed. Both depend mostly on the behaviour of
the bushing capacity itself. Additionally, some experiences about the long-term stability of the bushing
capacitance are presented.

2.2.1. Bandwidth Measurements


The bandwidth of the capacitive divider formed by the bushing and the adapter is measured using a
vector network analyser (VNA) in a controlled environment.

Figure 3: Test setup for the bandwidth and temperature test of the bushing

4
The result in Figure 4 shows that the divider ratio increases slightly with the frequency up to around
100 kHz and deviates by around 1% at 200 kHz. At 1 MHz, the ratio is offset around 3.5% from the
ratio at 50 Hz. An increasing ratio means a lower secondary voltage, hence voltage harmonics
measured with this method will be around 0.6% lower at around 150 kHz (3000th harmonic). In EN
61000-4-7, an accuracy of 5% is requested for the entire harmonic measurements chain. This leaves a
lot of headroom for the measurement device.

Figure 4: Bandwidth of the capacitive divider using a capacitance graded bushing at 12°C. The
deviation is given relative to the divider ratio at 50 Hz.

2.2.2. Temperature Dependency


Components in HV and EHV grids are usually large in size due to the insulation requirements in air.
This leads to a temperature influence on those electric parameters which are dependent on geometry. In
the case of the voltage measurement using capacitance graded bushings, the capacitance of the bushing
is likely the main temperature dependent factor. In order to test the influence of the ambient temperature
on the frequency response, the test setup in Figure 3 was brought into a climate controlled test chamber
and the measurements using the VNA are repeated with ambient temperatures of -20°C and 60°C. The
results are seen in Figure 5.

Figure 5: Bandwidth of the capacitive divider using a capacitance graded bushing with different
temperatures (augmented from Figure 4)

The results show that the deviation over frequency changes only a little when the temperature changes.
Notably, the frequency response is more flat at a 60°C than at 12°C, and the deviations become worse
when the temperature decreases.
The results indicate that a measurement with an accuracy of 5% as given in EN 61000-4-7 is still possible
even when the ambient temperature reaches extremely cold values for Central Europe.

2.2.3. Long-Term Stability


In order to assess the long-term stability of the bushing capacitance, a series of bushing measurements
from a total of 14 RIP-type bushings from two different manufacturers is analyzed. For six bushings,
the earliest measurements are from 1993 and span 24 years up to 2017. In this time, the total deviation
from the starting value never exceeded 2%.
5
Figure 6: Effect of ageing on the primary bushing capacitance C1 of 14 different bushings

However, there is no information about the long-term stability of the frequency behaviour of the
bushings. A shift in capacitance as outlined in Figure 6 would have no effect on the frequency response
as outlined in chapter 2.1.1, but it would affect the RMS voltage measurement.

2.3. Accuracy of Harmonic Evaluation of Power Quality Measurement Devices


As mentioned in chapter 1, there are no established standards for power quality measurement devices
for harmonic measurements above 2 kHz. One open question is the required accuracy and the optimal
means to achieve it. The accuracy that is achievable during the harmonics calculations depends mainly
on two parameters of the A/D converter inside of the power quality meter: the vertical resolution and
the sample rate. The general instructions for calculating harmonics from the sampled signal are outlined
in EN 61000-4-7, and the basic steps are shown in Figure 7.

Analogue
Sampling FFT Grouping Normalisation Harmonics
Signal

Figure 7: Basic steps of harmonics calculation

In order to assess the accuracy that is achievable using different sampling rates and vertical bits, the
entire processing chain has to be taken into account. The sampled signal is first transformed to the
frequency domain using an FFT. As power quality data is processed in 10-period-intervals (≅200 ms at
50 Hz), the result is a set of 5-Hz frequency bins. Figure 8 shows these bins for a simulated data stream
with a harmonic component of 1% (of a 230 V fundamental) at 9 kHz (180 th harmonic) at different
sample rates and vertical resolutions. As the FFT yields amplitudes, the correct value is
230 V · 1% · √(2) ≈ 3.25 V. With a sample rate of 20 kS/s, just over the Nyquist frequency, the
harmonic blurs into the neighbouring frequency bins. This effect is less pronounced at 100 kS/s and is
nearly invisible at 500 kS/s. An increased vertical resolution does not increase the accuracy at all. An
A/D-converter with 500 kS/s and 8 bit is therefore capable of greater accuracy for harmonic
measurements than one with 100 kS/s and 24 bit, as can be seen in Figure 8.
This error is the main contributor to the total error during the calculation of the harmonics. The grouping
and normalization steps alleviate the blurring problem by aggregating three frequency bins and
normalizing the harmonic to the 50 Hz amplitude, but a small deviation persists.
In order to assess the achievable accuracy considering all the steps of the harmonics calculation, a set of
different A/D-converter properties is again simulated using a 50 Hz-waveform with a superimposed
harmonic at 9 kHz with a 0.1% amplitude. Figure 9 shows the accuracy of the different setups after the
FFT, grouping, and normalization.
It is clear that a vertical resolution of 8 bit is insufficient. The reason for the larger deviation than in the
example in Figure 8 is the normalisation step, which is much more imprecise at 8 bit. Overall however,
an increased sample rate contributes much more to the accuracy of harmonics calculations than an
increased vertical resolution. A sample rate of more than 14 bits barely contributes to a greater accuracy.
6
Figure 8: Results of the FFT for different sample rates and vertical resolutions

Figure 9: Error of the harmonic calculation after sampling, FFT, grouping, and normalisation for
different sample rates and vertical resolutions

7
3. TEST INSTALLATION - RESULTS OF HIGH FREQUENCY MEASUREMENTS IN LV
AND HV GRIDS
3.1. PQ-Monitoring of a charging station for EV in the low voltage grid
In order to assess the high frequency harmonics that occurs during the charging process of electric
vehicles, a PQ measurement device was installed at the connection point of several chargers. An A/D-
converter with 500 kS/s and 14 bit vertical resolution was selected for all channels, according to the
findings in chapter 2.3. Figure 10 shows a heatmap of the measured current harmonics (value in Ampere
coded in color) of one phase over one day.

I/A

Figure 10: High frequency current harmonics of one day of electric vehicle charging. Frequency
resolution is 50 Hz; the values are given in A. In the bottom: current drawn through the phase over the
corresponding time.

Because the LED lighting is also supplied through the connection point, there is a constant harmonic
distortion over the entire day, with a component around 12 kHz. During the charging process of the
electric vehicles from around 9:00 to 18:30, the harmonics more elevated, with broadband distortion at
several instances. In general, harmonic distortion is hard to quantify, so this measurement does not allow
a deeper analysis. It is however clear that electric vehicle charging contributes to harmonic distortion at
much higher frequencies than the 2 kHz that is covered by norms and standards today.

8
3.2. PQ-Monitoring in the HV grid

The technique described in chapter 2.2 using transformer bushings was tested at several locations. Figure
11 shows two different installations. The installation of the adapters with the secondary capacitance C2
is minimally invasive and requires at most a couple of hours in which the transformer needs to be
switched off. The installation is also easy to remove.

Figure 11: Installation of bushing adapters on transformer bushings

Figure 12 shows the 95th and the 100th percentiles of the harmonic content of all three phases from a
measurement in the 380 kV-grid. The elevated levels of the 5th and 7th harmonic likely stem from
inverters in the grid.

Figure 12: 95th and 100th percentile of harmonics from a measurement using a capacitance graded
bushing. Dark shading : 95th percentile, light shading : 100th percentile

Figure 13 shows the higher frequency range of 2 - 9 kHz in 200 Hz-intervals. The 95th percentile is
barely visible. The 100th percentile, meaning the maximal values that occurred during the measurement
period, are significantly higher for the first and third phase than for the second. A peak occurs at around
8 kHz for the first and third phase. Again, the interpretation of these values is impossible without
significant context. However, it is clear that there is harmonic content in the 380 kV-grid.

9
Figure 13: High frequency harmonics measured with capacitance graded bushings

4. CONCLUSION
Harmonics measurements using capacitance graded bushings can be a viable alternative to capacitive
dividers or RC dividers, both of which require separate installation in a high voltage substation. The
installation of adapters with a secondary capacitance is fast, simple and can be reverted easily. The
measurements of the bandwidth, the temperature deviation and the data concerning long-term stability
indicate that an accuracy of 5% for the harmonics measurement, which is required by the norm EN
61000-4-7 , is possible up to 150 kHz, and the bandwidth of the bushing is sufficient for transient
measurements up to 1 MHz with only a 3% deviation of the divider ratio from 50 Hz.
The sample rate and not the vertical resolution is the main driver of the accuracy of power quality
measurement devices, especially for higher frequency harmonics.
An example installation shows the practical viability of the power quality using capacitance graded
bushings.

BIBLIOGRAPHY
[1] J. Meyer, R. Stiegler, M. Klatt, M. Elst and E. Sperling, “Accuracy of Harmonic Voltage Measurements
in the Frequency Range up to 5 kHz Using Conventional Instrument Transformers”, CIRED 2011, 21st
International Conference on Electricity Distribution, Frankfurt, 6-9 June 2011, Paper 0917.
[2] C. Buchhagen, M. Fischer, L. Hofmann and H. Däumling, "Metrological determination of the frequency
response of inductive voltage transformers up to 20 kHz," 2013 IEEE Power & Energy Society General
Meeting, Vancouver, BC, 2013, pp. 1-5. doi: 10.1109/PESMG.2013.6672835.
[3] J. Meyer, R. Stiegler and J. Kilter, "Accuracy of voltage instrument transformers for harmonic
measurements in elering's 330-kV-transmission network," 2016 Electric Power Quality and Supply
Reliability (PQ), Tallinn, 2016, pp. 85-90. doi: 10.1109/PQ.2016.7724094
[4] E. Sperling and P. Schegner, "A possibility to measure power quality with RC-divider," 22nd
International Conference and Exhibition on Electricity Distribution (CIRED 2013), Stockholm, 2013, pp.
1-4. doi: 10.1049/cp.2013.0602
[5] L. Schwalt, J. Plesch, S. Pack, W. Schulz and G. Achleitner, "Transient measurements in the Austrian
high voltage transmission system," 2017 International Symposium on Lightning Protection (XIV SIPDA),
Natal, 2017, pp. 208-211. doi: 10.1109/SIPDA.2017.8116925
[6] H. J. Vermeulen, L. R. Dann and J. van Rooijen, "Equivalent circuit modelling of a capacitive voltage
transformer for power system harmonic frequencies," in IEEE Transactions on Power Delivery, vol. 10,
no. 4, pp. 1743-1749, Oct 1995. doi: 10.1109/61.473385

10

View publication stats

You might also like