You are on page 1of 6

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

net/publication/236119496

Calculation of Harmonic Current Content in PV Power Plants Based on Single


Inverter Data

Data · October 2011

CITATIONS READS

0 566

5 authors, including:

Radu Dan Lazar Joerg Dannehl


Danfoss Drives A/S Danfoss A/S
13 PUBLICATIONS   48 CITATIONS    20 PUBLICATIONS   1,775 CITATIONS   

SEE PROFILE SEE PROFILE

Søren Bækhøj Kjær Marko Ibsch


Danfoss A/S DNV GL, Renewables Advisory
24 PUBLICATIONS   4,916 CITATIONS    1 PUBLICATION   0 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

PV-Island Bornholm View project

Intelligent and Efficient Power Electronics View project

All content following this page was uploaded by Joerg Dannehl on 04 June 2014.

The user has requested enhancement of the downloaded file.


Calculation of Harmonic Current Content in PV
Power Plants Based on Single Inverter Data
Adrian Constantin1, Radu Dan Lazar1, Jörg Dannehl1, Dr. Søren Bækhøj Kjær1
Marko Ibsch2


Abstract-- PV power plants connected in parallel with the II. REQUIREMENTS
medium voltage (MV) network have to fulfil certain
The maximum amount of harmonic currents, which a
requirements for the amount of harmonic current which they
generate. Normally, one reference measurement on a single plant connected to the same point of common coupling
inverter is extrapolated to estimate the amount of harmonics (PCC) as other plants, is given as (eq. 2.4.3-2 in [1]):
from an entire power plant with several inverters. We show in SA
this paper that the normal method of extrapolating is not I vAzul  ivzul  S kv  (1)
always valid for both weak and strong networks. The SGesamt
underlying reasons why the established method fails could be where iνzul is the relative limit given in table 1 below, Skv is
that they are made for PWM inverters with low switching the short circuit power at the PCC, SA is the apparent power
frequency and control bandwidth, whereas PV inverters tends
to operate at high switching frequencies (> 10 kHz). of the plant under investigation and SGesamt is the total
apparent power for all plants. Note that if only one plant is
Index Terms-- Harmonic current, Low voltage network, connected at the PCC, then SA = SGesamt. If the MV network
Medium voltage network, Photovoltaic. contains more PCC’s, the limit for each PCC is given as (eq.
2.4.3-4 in [1]):
I. INTRODUCTION
SGesamt (2)
I vAzul  ivzul  Skv 
P hotovoltaic power plants connected in parallel with the
medium voltage (MV) network have to fulfil certain
requirements for the amount of harmonic current which they
S Nezt
where SNetz is the size of the transformer connecting the MV
inject into the network [1], [2], [3], [4]. Since PV power network with the HV network, assuming the generators to
plants are only recently being connected with the MV include a voltage source inverter controlled by means of
network, there is not much known knowledge about their pulse width modulation (PWM) and switching at minimum 1
harmonic current injection [5]. So far, results from the wind kHz.
turbine industry has been adopted for calculating the amount TABLE 1: ALLOWED HARMONIC CURRENT, WHICH MAY BE FED IN TOTAL
of harmonics the PV power plants injects into the MV INTO THE MV NETWORK [1].
network [1], [2]. The present state of the art calculation-
principles lead to a situation where most of the power plants
exceed their limits, and hereby have to lower the amount of
installed power. Reason for that could be the calculation
principles instead of real life measurements. The background
for this paper is the method still applied for wind turbines,
but with focus on the validity of the method. The method is
investigated by measuring the harmonics content of multiple
inverters (up to nine) and the sum at the point of common
coupling on the low and medium voltage side of the mains
transformer.
The content of this paper is as follows. An introduction to
the problem is given in section 2. Measurements are
presented in section 3 and finally a conclusion is given in Equation (1) assumes that the harmonic currents add
section 4. together in a constructive approach, i.e. positive
interference, whereas (2) assumes that the sum of the
harmonic currents are attenuated due to negative
interference, i.e. harmonic phase-displacement between the
1 Søren Bækhøj Kjær (sbk@danfoss.com), Adrian Constantin generators.
(adrian@danfoss.com), Radu Dan Lazar (radu@danfoss.com) and Jörg
Dannehl (dannehl@danfoss.com) are with Danfoss Solar Inverters, DK-
In PV power plants, especially in string inverter plants
6400 Sønderborg with high number of inverters, the important question is how
2 Marko Ibsch (Marko.Ibsch@gl-group.com) is with GL Garrad Hassan these harmonics sum up on the secondary side (LV network)
America, Inc
of the connecting transformer, in order to estimate the
amount of harmonics at the PCC based on one inverter data.
The method used in [1] (eq. B.2.4-3) for PWM controlled
inverters with a switching frequency above 1 kHz, is:

 I  
n (3)
I 
2
i 1 ,i

where Iμ,i is the μ harmonic from the ith inverter. For n


th

identical generating units this will result in:


I I
  ,i n (4)

The IEC 61400-21 [2], [4] devises a different approach:

 I  
 n  (5)
I  i 1 ,i

where β is the exponent according to table 2. For inverters of


equal type (4) can be reduced to:

I I  n
  ,1
(6)

TABLE 2: SPECIFICATION OF EXPONENTS ACCORDING TO IEC 61000-3-6


AND IEC 61400-21:
Harmonic order β
Μ<5 1.0
Figure 1: PV test plant diagram. Skv = 1.55 MVA at MV busbar, SGesamt =
5 ≤ μ ≤ 10 1.4 SA = 9 * 15 kVA = 0.135 MVA, SNetz = 0.2 MVA.
Μ > 10 2.0
By comparing (4) and (6) and looking at the table 2 it can
be concluded that the two calculations are identical for
harmonics above 10th.

III. MEASUREMENTS
The PV test power plant is composed by nine Danfoss
TLX 15 kVA inverters coupled to a relatively weak grid
using the star topology as shown in Figure 1. Each inverter is
connected to the feeding transformer through a cable not
longer than 100 meters and with approximate 1% loss at
nominal power. The transformer is ΔΥ type, thus harmonics
of orders odd multiple of three do not pass to the MV
network (they are circulating in the Δ windings on the
primary side). The currents are measured simultaneously Figure 2: Busenwurth 12 MW PV power plant from the air, courtesy of
with a data acquisition solution from Dewetron. Using the Möhring Energie GmbH [6].
nominal data for the test plant (see Figure 1) the results in
Additional 8 MW
(1) and (2) are IνAzul / iνzul = 1.55 and 1.27, respectively. Uk = 20 kV power plant
Sa = 154 MVA
The first results (figures 4 to 7) are obtained with a R = 0.19 Ω
XL = 2.59 Ω

version of the control algorithm known to generate excessive TR1

harmonics, whereas the second results (figures 8 and 9) are ~ 5 km


MV cable
3 * 630 RM / 35 TR2
obtained with the default control algorithm. Furthermore, in R = 47 mΩ/km
XL = 105 mΩ/km
order to normalize the values, the results were divided with
the size of the 1st harmonic. Uk = 0.4 kV
Sa = 2 MVA
The third set of results (figures 10 and 11) is obtained by ΔU = 6%
R = 0.884 mΩ
XL = 5.95 mΩ
practical measurements in the 12 MW Busenwurth PV R = 0.43 Ω
TR6 1.8 MW installed
XL = 3.12 Ω
power plant [5]. The plant is divided in seven subsections TLX inverter

each of them connected to a LV/MV transformer and with TR7

120 pieces of the TLX 15k PRO+ string inverters (nominal


apparent power is 15 kVA). These subsections are connected Figure 3: Single line schematic for the 12 MW power plant in
Busenwurth, Germany [5].
through a MV cable to the main transformer to the HV
network some 5 km away. A. Results
In the first part of the test, the harmonics are measured at
the ‘PV inverter 1’ in Figure 1, while the rest of the plant is
disconnected. The data is used to extrapolate the harmonic
level in PCC according to (4) and (6), respectively. In the
second part of the test, the whole plant is injecting power
6
and the harmonics are measured in PCC. Furthermore the Limit for one PCC in MV
Nine inverters injecting power (measured in LV busbar)
results are plotted in percentage in the figures below, in Nine inverters injecting power (measured at MV PCC)

respect to the limit according to table 1. 5


6
Limit for one PCC in MV
One inverter injecting power (measured at inv#1)
Nine inverters injecting power (estimated by eq. 4)
Nine inverters injecting power (estimated by eq. 6)
Nine inverters injecting power (measured at LV busbar)
5 4

Harmonic level [%]


4 3
Harmonic level [%]

3 2

2 1

1 0
0 5 10 15 20 25
Harmonic order
st th
Figure 6: LV busbar vs. MV PPC harmonics in the range 1 - 25 with
non-optimized control algorithm.
0
0 5 10 15 20 25
Harmonic order
st
Figure 4: Harmonics in the range 1 – 25 with non-optimized control th If one compares the LV busbar harmonics with the
algorithm, measured at the LV busbar. harmonics measured in MV at the PCC (see Figure 7) it is
obvious that the harmonics multiple of three are almost
Figure 4 shows a poor match between the estimated and disappearing from the picture (they are circulating in the Δ
measured data for most of the harmonics. For the 3rd, 5th and windings on the primary side of the transformer, generating
7th harmonic, the sum of harmonic currents from all nine losses). This is due to the ΔΥ connection of the transformer
inverters (yellow bars) is higher than the one predicted when which will suppress these harmonics. In most cases, the
only one inverter is running (blue bars). This indicates that currents on the primary side of the transformer are slightly
the inverters are disturbing each other to generate a higher attenuated, but this could also be a case of errors in the
level of harmonics. The reason for this could be the weak measurements / calibration of current probes. The higher
network. For higher order harmonics the estimation is order harmonics are more or less showing the same picture
getting closer to the measured results, but both over and in Figure 7.
under estimating the levels. However, in all cases the results
from one inverter is always higher then the sum, indicating Limit for one PCC in MV
Nine inverters injecting power (measured in LV busbar)
destructive interference. Nine inverters injecting power (measured at MV PCC)

Limit for one PCC in MV


0.25
One inverter injecting power (measured at inv#1)
Nine inverters injecting power (estimated by eq. 4)
Nine inverters injecting power (estimated by eq. 6)
Nine inverters injecting power (measured at LV busbar)
0.25

0.2
Harmonic level [%]

0.2

0.15
Harmonic level [%]

0.15

0.1

0.1

0.05

0.05

0
25 30 35 40 45 50
Harmonic order

Figure 7: LV busbar vs. MV PPC harmonics in the range 25 th - 50th with


0
25 30 35 40 45 50 non-optimized control algorithm.
Harmonic order

Figure 5: Harmonics in the range 25th – 50th with non-optimized control


algorithm, measured at the LV busbar.
The following test is similar to the previous one excepting the estimated ones. The measurements in the plant are
that the control algorithm is the default one for TLX 15kVA. indicating higher 9th and 11th than the reference inverter and
inherently bigger than the estimated.
3
Limit for one PCC in MV
One inverter injecting power (measured at inv#1)
2.5
Nine inverters injecting power (estimated by eq. 4)
Reference inverter - Wind test measurement
Nine inverters injecting power (estimated by eq. 6)
LV bus bar estimation eq.4
Nine inverters injecting power (measured at LV busbar)
LV bus bar estimation eq.6
2.5
LV bus bar measurement

2
Harmonic level [%]

1.5

Harmonic level [%]


1.5

1
1

0.5
0.5

0
0 5 10 15 20 25
Harmonic order 0
0 5 10 15 20 25
Figure 8: Harmonics in the range 1st – 25th with default control Harmonic order

algorithm, measured at the LV busbar. Figure 10: Harmonics in the range 1st – 25th measured in Busenwurth
plant at the LV busbar for block 1.
In this case (Figure 8) the estimation is better than the
previous case showing that poor control can seriously affect 0.4
the expected harmonic levels. Still the 2nd, 5th, 6th, 7th and 9th Reference inverter - Wind test measurement
LV bus bar estimation eq.4

are presenting significant differences in respect to the 0.35


LV bus bar estimation eq.6
LV bus bar measurement

measurements, which again indicates cross-talk between the


inverters in the relatively weak network. 0.3

Limit for one PCC in MV


One inverter injecting power (measured at inv#1) 0.25
Nine inverters injecting power (estimated by eq. 4)
Harmonic level [%]

Nine inverters injecting power (estimated by eq. 6)


Nine inverters injecting power (measured at LV busbar)
0.25
0.2

0.15
0.2
Harmonic level [%]

0.1

0.15

0.05

0.1
0
25 30 35 40 45 50
Harmonic order

Figure 11: Harmonics in the range 25th–50th measured in Busenwurth


0.05
plant at the LV busbar for block 1.

In the case of higher order harmonics (Figure 11) there is


0
a better fitting that in the previous cases but still not good
25 30 35 40 45 50
Harmonic order enough. All the results indicate that the estimation based on
Figure 9: Harmonics in the range 25th – 50th with default control
one inverter data together with (4) and (6) is not really
algorithm, measured at the LV busbar.
suitable for estimating the behaviour of an entire plant in
The estimation for higher order harmonics (see Figure 9) MV.
is not even close to the measured values. Almost all the B. Discussion
estimated values are bigger than the values measured at the From the presented results it seems that the method used
LV busbar. for summing up harmonics in MV PCC is not accurate in
In the following part will be presented the measurements respect to harmonic estimation in plants based on single
made in the field at the Busenwurth 12 MW PV plant [5]. inverter data. A better approach might be the one given in
The reference inverter is measured in the laboratory by GL [4], where the back-ground voltage harmonics of the
Garrad Hassan during the certification according to FGW network is also taken into consideration.
TR3 [3]. These data are used for extrapolation according to In the same time higher order harmonics have a poor
(4) and (6) and compared to LV measurements. Figure 10 is estimation due to the fact that the estimation method does
showing almost no match between the measured values and
not take into consideration self-damping phenomena and
attenuation. The first one occurs when phase shift is
introduced between the same harmonics coming from
different inverters. The phase shift is caused by the different
cable properties, according to different cable dimensions.
Attenuation phenomena are produced by the transformers,
which can act like low pass filters. There is no easy way of
“guess-timating” the impact of these types of phenomena.
Generally, measured harmonics within plants in the order
2nd to 11th is always larger then the value indicated from the
reference measurements, which is an indication of positive
feedback within the system (PV power plant). From the 12 th
to the 19th harmonic, the measured value is always lower but
still not equal to the ones estimated by (4) or (6). Equations
(4) and (6) are valid from the 20th harmonic and up to the
50th.

IV. CONCLUSION
At this moment there is no feasible general method which
can estimate the current harmonics in MV based on data
from one LV generating unit, in case of LV plants with
identical generating units coupled to MV through a
transformer.
Nevertheless such a method cannot be pure theoretical
due to complexity of the necessary calculations and
multitude of parameters. In case of a more practical
approach (experience based) there is a sum of situations
which might not be covered.
For this reason the best method until now is still the direct
measurement in MV using the before and after approach in
order to be able to subtract the background harmonics [4].
A good current control will always ensure that the
harmonic level is low, easing the compatibility with
harmonic standards not only for LV but also for MV.

V. REFERENCES
[1] Technical guideline: Generating Plants Connected to the Medium-
Voltage Network. Guideline for generating plants’ connection to and
parallel operation with the medium-voltage network. BDEW June
2008.
[2] IEC 61400-21 ed. 2. Wind turbine generator systems – Part 21:
Measurements and assessment of power quality characteristics of grid
connected wind turbines.
[3] FGW TR3 Rev. 21, Bestimmung der Elektrischen Eigenschaften von
Erzeugungseinheiten am Mittel-, Hoch- und Höchstspannungsnetz,
http://www.wind-fgw.de/
[4] N. Goldenbaum, P. Brogan, B. Andresen, “HARMONIC
DISTORTIONS FROM LARGE WIND FARMS EQUIPPED WITH
MODERN WIND,” EWEA 2011, Brussels, Belgium.
[5] S. B. Kjær, J. Dannehl, F. Mecking and J. Godbersen, “A 12
megawatt power plant with full implemented ancillary services
according to the German grid codes the first results,” in Proc. of 26th
European Photovoltaic Solar Energy Conference and Exhibition
(EUPVSEC) / 6th World Conference on Photovoltaic Energy
Conversion, Hamburg, Germany, 2011.
[6] http://www.moehring-energie.de/

View publication stats

You might also like