25th International Conference on Electricity Distribution Madrid, 3-6 June 2019

Paper n° 1398

SURVEY OF NETWORK IMPEDANCE IN THE FREQUENCY RANGE 2-9 KHZ

IN PUBLIC LOW VOLTAGE NETWORKS IN AT/CH/CZ/GE
Robert STIEGLER, Jan MEYER Stefan SCHORI, Michael HÖCKEL Karl SCHEIDA
Technische Universitaet Dresden Berner Fachhochschule Österreichs E-Wirtschaft
Germany Switzerland Austria
[email protected] [email protected] [email protected]

Jiří DRÁPELA Tomáš HANŽLÍK

Brno University of Technology EGC – EnerGoConsult CB s.r.o.
Czech Republic Czech Republic
[email protected] [email protected]

regard to emission in the frequency range 2-9 kHz and/or

ABSTRACT unnecessary investments in filter equipment.
Setting realistic emission limits for distorting customer This issue is already taken into account in IEC 61000-4-7
installations is a crucial requirement for a reliable and [3]. It proposes in an informative annex a reference
disturbance-free operation of public distribution impedance for testing the compliance of appliances with
networks. Therefore only an adequate share of the total emission limits in the frequency range 2-9 kHz, which are
allowable voltage distortion (compatibility level) has to currently under development at IEC SC77A WG1.
be allocated to each customer installation. This allowable The reference impedance in IEC 61000-4-7 is based on
contribution of a single customer installation is usually the reference short circuit power according to IEC/TR
small and difficult to assess. Hence, nowadays most 60725 [4]. This value is rather conservative in order to
standards and guidelines translate the allocated voltage ensure that more than 90 % of the connection points in
distortion in a respective current distortion by using the public low voltage (LV) networks have a higher short
network impedance at the considered frequency. As circuit power.
frequency-dependent network impedance is usually not The 3rd edition of the AT/CH/CZ/GE (Austria/Switzer-
known during the planning process, assumptions are land/Czech Republic/Germany) rules for assessment of
required, which have to be realistic and not too network disturbances for customer installations is
conservative. This applies in particular to frequencies currently under development and shall be extended to the
above 2 kHz, where the usually used extrapolation based frequency range 2-9 kHz. Therefore, a realistic, reliable
on short circuit impedance can be very conservative due and widely accepted assumption for the network
to the increasing impact of connected customer impedance in this frequency range is required. As larger
equipment on the network impedance. Based on a customer installations are usually connected at points
comprehensive measurement campaign in four different with a higher short circuit power than specified in
countries this paper identifies typical ranges of the IEC 60725, the reference impedance provided in
frequency-dependent network impedance in the frequency IEC 61000-4-7 for the frequency range 2-9 kHz might
range 2-9 kHz. It analyses the impact of short circuit still be too conservative. Especially for high short circuit
power and proposes a simplified impedance estimation, powers even the “impedance line” can be significantly
which can be used to calculate realistic current emission lower than the reference impedance according to IEC
limits in the planning stage. 61000-4-7 as it is illustrated in Fig. 1.
Consequently also the dependency of the frequency
dependent network impedance in the frequency range
INTRODUCTION 2-9 kHz on the short circuit power has to be considered
Distortion is commonly divided into subharmonics, when developing a realistic approach for the calculation
harmonics, interharmonics and the frequency range above of current emission limits for larger customer
2 kHz, which is also frequently referred to as installations. At frequencies above 2 kHz the frequency-
supraharmonics. Emission limits for a new customer
installation are determined in the planning stage. In most
cases no detailed information about the frequency
dependent network impedance is available and
consequently simplifying assumptions must be applied.
In low voltage (LV) networks usually the “impedance
line” is used, which simply corresponds to the linear
extrapolation of the short circuit impedance by
multiplying it with the harmonic order. Several studies
have shown that with increasing frequency this
assumption gets less realistic, especially in the frequency
range above 2 kHz ([1], [2]). Consequently, the
calculated emission limits might be too restrictive, Fig. 1: IEC 61000-4-7 compared to impedance lines for different
resulting in a less effective utilization of networks with short-circuit power.

CIRED 2019 1/5

 25th International Conference on Electricity Distribution Madrid, 3-6 June 2019

Paper n° 1398

dependent network impedance is getting more influenced

by the input impedance of connected devices, while the
impact of the network elements decreases. Up to now no
comprehensive knowledge about the level of influence of
the short circuit power on the frequency-dependent
network impedance is available.
In order to get a representative overview about typical
ranges of the network impedance in the frequency range
2-9 kHz and its dependence on the short circuit power, a
comprehensive measurement campaign has been
conducted with the help of DSOs in the four countries
Fig. 2: Short circuit power of measurement sites by country.
AT/CH/CZ/GE. This paper presents the results of this
measurement campaign, analyses the relation between junction boxes (JB). About 75 % of all measurements
short circuit power and frequency-dependent network have been performed at junction boxes significantly
impedance and develops a simplified approach for distant from the LV busbar.
estimating the impedance for the calculation of current Fig. 2 shows the cumulative distribution function (CDF)
emission limits in the frequency range 2-9 kHz for of the short circuit power of the measurement sites for
customer installations connected to LV networks. each country. The short circuit power of the countries is
The first part of the paper describes the measurement distributed over a similar range. In Switzerland the short
campaign, the used measurement equipment and circuit power tends to be higher, this is most likely
characterises the measured sites. The second part caused by the site selection, as in Switzerland only DSOs
provides the measurement results and evaluates the in more urbanized regions participated in the survey.
impact of the short circuit power on the network
impedance characteristic. Finally, a novel approach for Measurement system
the simplified estimation of the network impedance in the The measurement system is a joint development of the
frequency range 2-9 kHz is presented. The paper closes University of Technology in Dresden and an industrial
with some conclusions and needs for future work. partner. The measurement principle is based on an earlier
developed setup, which is described in [5].
The layout of the measurement system is shown in Fig. 3,
MEASUREMENT CAMPAIGN a photo of its field application in Fig. 4. The system
Measurement sites consists of a linear amplifier operated as current source
for network excitation and a unit for voltage and current
With the help of twelve DSOs in the four countries measurement. The measurement system is controlled by a
AT/CH/CZ/GE impedance measurements were laptop. The measurement is performed as a stepwise
conducted at 81 measurement sites. Each DSO selected at discrete sweep. A current with a given frequency is
least two typical public LV networks, one representing a injected in the network and the resulting voltage is
stronger, urban network and one a weaker, rural network. measured. Based on both values the impedance of the
In each network, at least two sites were selected for
impedance measurement: one site directly at the LV Measurement System
busbar of the MV/LV substation (SS) (highest short
circuit power) and one site at the furthest accessible point Measurement Amplifier
in the network (usually a junction box (JB)), where the L A
lowest short circuit power in the particular network is
LV-Grid

Laptop

expected.
For 76 measurement sites the DSO’s provided also the V i(t)
short circuit power, which has been obtained from their
network calculation software. In total 198 (out of 228 N
theoretical possible) single loop impedance
measurements were obtained. Not at every site the loop Fig. 3: Scheme of the measurement system.
impedance of all three phases could be measured.
Table 1 presents the number of measurement sites and
measured loop impedances for each country
distinguishing measurements at substations (SS) and at
Table 1: Overview of measurements
AT CH CZ GE ∑
SS 4 9 2 7 22
Sites JB 12 11 5 26 54
∑ 16 20 7 33 76
12 11 6 21 50
SS
Single loop (26 %) (28 %) (30 %) (23 %) (25 %)
impedances JB 35 29 14 70 148
∑ 47 40 20 91 198 Fig. 4: Measurement system while measuring at a junction box.

CIRED 2019 2/5

 25th International Conference on Electricity Distribution Madrid, 3-6 June 2019

Paper n° 1398

network at the given frequency is calculated.

The output current is measured internally, while the
voltage of the network is measured at the connection
point, so that the impedance of the measurement cables
does not influence the measurement results. Both voltage
and current signals are sampled with a rate of 10 MS/s
and a vertical resolution of 16 Bit.
The magnitude of the injected current is automatically
adapted to the situation at the connection point, in order
to ensure an as small as possible excitation of the network
by not exceeding 1 V. In this way the method is less
invasive and damage of connected equipment as well as
an unwanted influence of the measured impedance itself Fig. 5: All measured loop impedances in comparison with
is avoided. For the last reason also the power supply of IEC 61000-4-7 reference impedance including selected
the system is disconnected during the measurement percentile curves.
process. impedances at the junction box being smaller than at the
The system is able to measure the frequency-dependent respective substation. This confirms the known fact that
network impedance in the range between 0.1 kHz and at frequencies higher than 2 kHz the impact of connected
150 kHz with an uncertainty of better than 5 %. customer appliances is not negligible anymore.
In order to study the impact of short circuit power on the
MEASUREMENT RESULTS impedance in the frequency range 2-9 kHz for each
measured impedance the ratio between reference
Fig. 5 presents an overview of all measurements together impedance (Z4-7) and measured impedance (Zm) is
with selected percentile curves for the whole measured calculated and the median of all ratios for each measured
frequency range. It also contains the IEC 61000-4-7 loop impedance is determined. The results are presented
reference impedance, which only applies in the frequency in Fig. 7 and show an expected dependency, which
range 2-9 kHz (solid part of the line). As this paper however gets saturated with increasing short circuit
focuses on the frequency range 2-9 kHz, only this power. The high variation for individual short circuit
frequency range is presented and discussed further on in power values indicates the impact of the impedance of
the text. connected customer appliances, which reduces with
Almost all measured impedances are below the reference increasing short circuit power. This is a simple
impedance, which proofs its feasibility. It should be noted consequence of the reduced grid-side impedance, which
that the reference impedance is based on a (three-phase) in turn increases its impact on the measured impedance in
short circuit power of 0.57 MVA (cf. IEC 60725), but the relation to the impedance of the connected appliances.
lowest short circuit power of the measured sites is still
higher than this (cf. Fig. 2). However, three
measurements (1.5 %) are still partly higher than the
reference impedance.
Fig. 6 presents all measured loop impedance separate for
each country. The countries are rather similar, except in
Switzerland the impedances are lower by almost one
decade. The most likely reason for this is the already
mentioned higher short circuit power at the Swiss sites.
The low difference between substation and junction box
measurements suggests that the networks are rather short
and the distance between substation busbars and junction
boxes are rather low. The variation in the impedance
plots suggests that in the frequency range above 2 kHz
the short circuit power seem to have still a significant Fig. 7: 50-%-percentile of ratio between reference impedance
influence on the impedance. Some of the impedances according to IEC 61000-4-7 and measured impedance in the
measured at the junction boxes have additional range 2-9 kHz depending on short-circuit power.
resonances at higher frequencies, which can result in

Fig. 6: Measured loop impedances per country (red: substation, blue: junction box, black: reference impedance acc. to IEC 61000-4-7)

CIRED 2019 3/5

 25th International Conference on Electricity Distribution Madrid, 3-6 June 2019

Paper n° 1398

PROPOSED IMPEDANCE APPROACH

As shown in Fig. 7, the measured loop impedance can
vary about two decades and can be up to factor 100
smaller than the IEC 61000-4-7 reference impedance.
Consequently, the assumption of a constant reference
impedance seems not applicable for larger customer
installations, which are usually connected to points with
higher short circuit power than the reference short circuit
power. A certain dependency of the impedance on the
short circuit power is therefore also justified at
frequencies above 2 kHz.
Comprehensive analyses of possible dependencies
Fig. 9: Impedance lines for different reduction factors (black)
between short circuit power and characteristic parameters
compared with measurement at Ssc = 8.8 MVA (red)
of the measured impedances could not identify any clear
relation. Therefore the proposed impedance approach is A reduction factor of r = 0 means, that the short circuit
based on an appropriate scaling of the IEC 61000-4-7 ratio has the highest impact on the impedance. An
reference impedance by the short circuit power at the increasing reduction factor r means a decreasing
connection point of a customer installation. influence of the short circuit power. In case of r = 1, the
impedance is independent of the short circuit power and
Linearization of reference impedance equals the reference impedance. This corresponds to the
In order to keep the approach simple the reference method of compliance assessment for single appliances.
impedance Zref for the frequency f is linearized in the Fig. 9 illustrates the impact of the reduction factor using a
considered frequency range 2-9 kHz according to (1). real measured impedance at a short circuit power of
Ssc = 8.8 MVA. The example shows that a reduction
𝑍9 kHz − 𝑍2 kHz factor of r = 0 significantly underestimates the measured
𝑍ref (𝑓) = ⋅ (𝑓 − 9 kHz) + 𝑍9 kHz (1)
9 kHz − 2 kHz impedance, while using the reference impedance (r = 1)
Using Z2 kHz =3.25 Ω and Z9 kHz =10.25 Ω results in results in a far too conservative estimate. For the example
in Fig. 9 a reduction factor of r ≥ 0.15 leads to an
𝑓 estimated impedance Zest, which is never below the
𝑍ref (𝑓) = ( + 1.25) Ω (2)
kHz measured impedance.
The error  with regard to the true reference impedance Determination of reduction factor
characteristic is in the range of -15 % to +5 %, which For each impedance measurement the reduction factor r
results in an average in the considered frequency range of is determined iteratively so that the estimated impedance
about zero (Fig. 8). Zest(f) is just not exceeded by the measured impedance for
Scaling methodology all frequency (cf. to r = 0.15 in Fig. 9).
Fig. 10 presents the cumulative distribution function
To include the dependency on the short circuit power, the (c.d.f.) of the calculated reduction factors. A reduction
linearized reference impedance is scaled with the ratio of factor of zero is obtained for about 60 % of the
reference short circuit power (Ssc ref) and the real short measurements, which means that even at highest
circuit power at the connection point (Ssc). In order to sensitivity (r = 0) the estimated impedance is still above
keep the level of dependency flexible, a reduction factor r the measured impedance. Analysing the maximum ratio
is introduced. The final equation for the calculation of the between estimated and measured impedance has shown
impedance at the connection point Zest is given by that it is less than 2.5 for about 80 % of the respective
𝑆sc ref cases. Compared to the original approach just using the
𝑍est (𝑓) = 𝑍ref (𝑓) ⋅ (𝑟 + (1 − 𝑟) ⋅ ) (3) extrapolated short circuit impedance the new approach is
𝑆sc still about 5 to 20 times better and therefore an acceptable
trade-off.

Fig. 8: Reference impedance linearization and linearization error Fig. 10: Calculated impact reduction factor for all measurements

CIRED 2019 4/5

 25th International Conference on Electricity Distribution Madrid, 3-6 June 2019

Paper n° 1398

Therefore the proposed approach can still provide rather

conservative results under specific circumstances.
Consequently, more research is needed to improve
understanding and knowledge about the network
impedance at frequencies above 2 kHz in order to
improve the impedance estimation approach presented in
this paper. This includes also frequencies above 9 kHz,
for which compatibility levels have been established just
now and suitable methods for emission limit
determination have to be developed for future editions of
the respective rules and guidelines for network
disturbance assessment.
Fig. 11: Comparison of measured impedance (dashed) and estimated
impedance (solid) for three different short-circuit powers: ACKNOWLEDGEMENTS
black: 0.64 MVA; red: 1.01 MVA; blue: 1.92 MVA
The project was partly funded by the Federal Ministry of
On the other hand, Fig. 10 shows that for about 40 % the Economics and Energy (BMWI) according to a decision
influence of the short circuit power has to be reduced in of the German Federal Parliament under the funding code
order to ensure that the estimated impedance is not ZF4272801LT6.
exceeded by the measured ones. For a small number of We like to thank Spitzenberger & Spies GmbH & Co.
cases even a factor of r = 1 (no influence of short circuit KG for providing the measurement system.
power) is not sufficient, which means that the reference We like to thank the following DSOs for their support
impedance of IEC 61000-4-7 is still exceeded by the during the measurements and for the provision of
measured impedances. However, these few cases are topology data and short circuit data for the measured
caused by specific resonances, which has to be treated sites: Aziende Industriali di Lugano (AIL) SA,
individually and should not be considered for the Bayernwerk Netz GmbH, ČEZ Distribuce, Energie
development of a general value for the reduction factor r. Service Biel/Bienne, ewz Elektrizitätswerk der Stadt
Following a similar probabilistic approach as for the Zürich, LINZ NETZ GmbH, Netze BW GmbH, Netz
determination of the reference impedance in IEC 60725, Oberösterreich GmbH, NRM Netzdienste Rhein-Main
a reduction factor of r = 0.25 is proposed. For this value GmbH, Repower AG, Vorarlberger Energienetze GmbH.
about 90 % of the measured impedances are below the
respective estimated impedance (cf. Fig. 10).
Fig. 11 presents the application of the impedance REFERENCES
approach for three different cases. For the first case [1] D. Borkowski, A. Wetula and A. Bień, "New
(black) due to a pronounced resonance the estimated method for noninvasive measurement of utility
impedance is exceeded by the measured impedance. For harmonic impedance," 2012 IEEE Power and
the second example (blue) the estimated impedance
matches the measured impedance very good, while for Energy Society General Meeting, San Diego, CA,
the third example (red) the estimated impedance is still 2012, pp. 1-8.
significantly higher than the measured one. The last two [2] R. Gretsch and M. Neubauer, “System Impedances
example cases underline the weak link between short and Background Noise in the Frequency Range 2 to
circuit power and network impedance at frequencies 9 kHz”, CIRED 1999.
above 2 kHz. Since both cases do not show resonances, [3] IEC 61000-4-7:2002, “Electromagnetic
the better match between estimated and measured compatibility (EMC) - Part 4-7: Testing and
impedance is obtained for the site with twice as high measurement techniques - General guide on
short circuit-power. harmonics and interharmonics measurements and
instrumentation, for power supply systems and
CONCLUSIONS equipment connected thereto”, 2nd edition 2002.
This paper presents a new approach to estimate the [4] IEC TR 60725:2012: “Consideration of reference
network impedance in the frequency range 2 – 9 kHz for impedances and public supply network impedances
the determination of current emission limits during the for use in determining the disturbance
planning of customer installations. The proposal is still characteristics of electrical equipment having a rated
simple, but represents a significantly better estimation current ≤75 A per phase”, 3rd edition 2012.
compared to the current practice of using the impedance [5] R. Stiegler, J. Meyer, P. Schegner and D.
line extrapolated based on the short circuit impedance.
Chakravorty, "Measurement of network harmonic
The approach has been developed based on a
comprehensive set of impedance measurements in public impedance in presence of electronic equipment,"
LV networks. 2015 IEEE International Workshop on Applied
The analysis of the impedance measurements has Measurements for Power Systems (AMPS), Aachen,
confirmed that the link between network impedance and 2015, pp. 49-54.
short circuit power at frequencies above 2 kHz becomes
more and more weak, while the influence of the input
impedance of close-by connected equipment increases.

CIRED 2019 5/5