This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JPETS.2016.2550601, IEEE Power
and Energy Technology Systems Journal
1
Harmonic Impact of a 20 MW PV Solar Farm
on a Utility Distribution Network
Rajiv K. Varma, Senior Member, IEEE, Shah Arifur Rahman, Member, IEEE,
Tim Vanderheide, Member, IEEE, and Michael Dang, Member, IEEE
Abstract – This paper presents one of the first studies of the
harmonic impact of a significantly large PV solar farm of 20 MW
in a utility distribution system. This solar farm is a constituent of
the 80 MW PV solar farm in Sarnia, Ontario, which is so far the
largest solar farm in Canada. The utility network is modeled in
detail using commercial grade PSCAD/EMTDC software which is
validated through load flow studies conducted by CYME
software and correlated with SCADA measurements. The
validated network model is used for network resonance study and
harmonics impact analysis of the solar farms under different
network conditions. The harmonics data instrumented for several
months was provided by the transmission utility at the two solar
farm units and at main feeder substation. These data were
utilized for extensive harmonic impact studies with widely
different short circuit levels and network resonance conditions.
This paper presents the detailed procedure adopted for
performing such harmonic impact studies. It is concluded that
this large solar farm may not cause any substantial voltage
distortion on the distribution network during steady state
operating conditions. However, recommendations are made for
utilities to perform such studies to ensure safe operation of
critical loads.
Index Terms— PV solar system, Solar energy, Harmonics,
Network Resonance, Distribution Systems, Frequency Scan, Total
Harmonic Distortion, Total Demand Distortion
I.
INTRODUCTION
Power electronic-based generators in the network, such as, a
PV system produce harmonics that are injected into the power
network [1], [2]. Distribution /transmission networks are often
shunt compensated by bus capacitors for voltage regulation
and power factor correction. The combination of overhead line
inductances and shunt capacitors together with cable
capacitances can resonate at certain frequencies leading to
network resonances. The network presents high impedances at
these resonant frequencies, which appear as peaks in the
The financial support from Ontario Centres of Excellence (OCE) and
Bluewater Power, Sarnia, under the grants WE-SP109-E50712-08 and CRSG30-11182-11; and that from NSERC are gratefully acknowledged.
Rajiv K. Varma and Shah. A. Rahman are with the Department of
Electrical and Computer Engineering, University of Western Ontario,
London, Ontario, N6A 5B9, Canada (e-mail: rkvarma@uwo.ca;
srahma32@uwo.ca). Tim Vanderheide is with Bluewater Power Corporation,
Sarnia, ON, N7T 7L6, Canada (email: TVanderheide@bluewaterpower.com).
Michael Dang, formerly with Hydro One Networks Inc., Toronto, is currently
with
Mohawk
College,
Hamilton,
ON,
(email:
michaeldang@mohawkcollege.ca)
frequency scan of network impedance [3]-[7]. If the injected
harmonics from power electronic converters or nonlinear loads
come close or align with any one of the frequencies of the
resonant modes, the voltages corresponding to those harmonic
frequencies can be amplified and may damage equipment or
cause protective relay failures [5], [7], [8] - [10]. A case of
parallel resonance in a distribution network system comprising
a wind farm and power factor correction capacitors is reported
in [11].. Variations in resonant modes were also reported with
changes in transmission and sub-transmission network loading
conditions [12] and short circuit levels [13]. The harmonic
impact of a small scale photovoltaic farm on the distribution
network systems is reported in [1, 14]. Harmonic amplification
due to network resonance can potentially be the limiting factor
for PV solar connectivity in distribution feeder [15], [16].
IEEE 519 Standard recommends that a general harmonic
analysis procedure be adopted as part of the system planning
process whenever known large harmonic sources exist on a
system or when significant dispersed generators are being
proposed in a utility network [17]. Bluewater Power
Distribution Company (BWP) supplies power to the city of
Sarnia which is located in southwestern Ontario, Canada. Its
80 MW photovoltaic (PV) farm is the largest solar facility in
Canada [18]. Its first 20 MW PV plant became operational in
2009. This paper reports the study undertaken on the initiative
of Bluewater Power to ensure compliance with IEEE 519
recommendation. This study has been performed with the
collaboration amongst a university, a distribution utility, a
transmission utility, a solar farm developer, and a solar farm
operator. The objectives of this study were to identify if the
harmonic injections from the solar farms under widely
different operating conditions could potentially exceed safe
operating levels in the electrical distribution utility network.
This paper is organized as follows. Section II describes the
different components of the utility system under study. The
modeling of different components in CYME and
PSCAD/EMTDC software, together with the validation of the
developed system model with SCADA measurements is
presented in Section III. The study of network resonances for
different network conditions is elucidated in Section IV. The
evaluation of harmonic distortions for different critical cases
of solar farm harmonic injections is illustrated in Section V.
Actual measurement of harmonic distortions at the solar farm
buses are compiled in Section VI. The impact of ambient
harmonics is presented in Sec. VII, while the conclusions of
this entire research are enunciated in Sec. VIII.
2332-7707 (c) 2015 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JPETS.2016.2550601, IEEE Power
and Energy Technology Systems Journal
2
II.
THE STUDY SYSTEM
A.
Sarnia PV Solar Farm
The 80 MW photovoltaic solar power plant is served by
four 27.6 kV distribution feeders (18M14, 96M23, 96M27,
96M28) as shown in Fig. 1 which is the Geographical
Information System (GIS) map of the PV solar systems and the
surrounding areas. Each one of these feeders connects to 20
MW PV solar farms which are the constituents of the 80 MW
solar farm complex. The studies presented in this paper
correspond to the 20 MW solar farm connected to the 18M14
feeder from St. Andrews TS as shown in Fig. 2.
B.
St. Andrews Transformer Station (TS)
St. Andrews TS, which is owned and operated by the
transmission utility Hydro One Networks Inc., is radially
supplied from 115 kV circuits N6C and N7C from Scott TS
and stepped down to 27.6 kV via two 50 MVA, transformers
as shown in Fig. 2. The transformer station has four 27.6 kV
feeders supplying Sarnia domestic, commercial and industrial
loads and is also connected to the 18M14 feeder having two
10 MW solar farms. St. Andrews TS is also equipped with two
20 Mvar capacitor banks (SC1 and SC2) for voltage support.
C.
The 18M14 Feeder and the Solar Farm
The 27.6 kV 18M14 feeder supplies power to two large
industrial customers besides small commercial and residential
loads. These customers are supplied from about 54 polemounted transformers with the two large individual loads of
about 10 MW and 2 MW, respectively. A set of 1.2 Mvar
power factor correction capacitors is located in the vicinity of
the 2 MW customer premises. The daily load profile on this
feeder is observed to be almost constant between 13-14 MW.
The 18M14 feeder is made up of 7.8 km of overhead
distribution line and two small segments of three phase
underground cable. There are two 10 MW solar farms, Sarnia
Solar 2 and Sarnia Solar 5, connected on this feeder and are
located about 6.2 km from St. Andrews TS. During situations
when the power generated from the solar farms exceeds the
feeder load, it flows backwards into the St. Andrews TS.
D.
Fig. 1. GIS map of 80MW Solar Farm on BWP network.
Network Data Acquisition Systems
The data utilized in this study has been obtained from
several sources. Geographical Information System (GIS)
provided the physical line data and network configuration. The
SCADA data was obtained at BWP office in real time from
remote locations. Hydro One installed Schneider ION 7650
meters at the points of common coupling (PCCs) of the two
10 MW solar plants for harmonics and transient event
measurements.
III. SYSTEM MODEL DEVELOPMENT
At first, a system model of the utility network was
constructed from the available network data. Since some
information on network components was either outdated or
unavailable
substantial
efforts
were
made
Fig. 2. One line diagram for feeder 18M14.
to acquire reliable data from visual inspections, new
measurements and reasonable extrapolations in some cases.
CYME software using CYMEdist module [19] was then used
to model BWP network utilizing system GIS and SCADA
databases, for steady state load flow analyses. The load flow
results from CYME were validated from actual SCADA
measurements and the process was iterated with updated
network models till acceptable accuracy was obtained. The
network model was then constructed in the commercial grade
electromagnetic transients software PSCAD/EMTDC. The
steady state power flow results from PSCAD/EMTDC were
then validated again from CYME results. The objective of
2332-7707 (c) 2015 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JPETS.2016.2550601, IEEE Power
and Energy Technology Systems Journal
3
developing the system model in PSCAD was to perform
harmonic studies and transients studies, which was not
possible with the CYME software. The modeling of different
system components in the two softwares is presented below.
A.
Modeling of Grid/Source and Capacitor
i) CYME Model
St. Andrews TS is modeled in CYME as an ideal voltage
source behind its short-circuit impedance at the 27.6 kV,
18M14 feeder bus breaker.
ii) PSCAD/EMTDC Model
For load flow validation, the St. Andrews TS network is
modeled in PSCAD/EMTDC software as an ideal voltage
source behind the short-circuit impedance at the 27.6 kV,
18M14 feeder bus breaker. However for harmonic studies,
St. Andrews TS is modeled as an ideal voltage source behind
the short circuit impedance at 115kV. The two 20 Mvar SC1
and SC2 capacitors are modeled at this station.
B. Modeling of Feeder
i) CYME Model
The 27.6 kV, 18M14 feeder consisting of overhead lines
and cables is modeled as R, X and B in CYMEdist on
10 MVA base.
ii) PSCAD/EMTDC Model
The overhead line and underground cables for the 18M14
feeder are modeled as nominal PI sections in
PSCAD/EMTDC. This is a multiphase coupled equivalent PI
circuit representation [20] and is used in the studies because
overhead line segments and cables for the 18M14 feeder are
less than 150/h miles and 90/h miles respectively [3], [21],
where ‘h’ represents the harmonic number set at the maximum
30th harmonic order.
C. Modeling of Loads
All loads in the distribution network are reflected on 27.6 kV
network and are modeled as constant power loads at the
fundamental frequency [3], [10], [20]. The loads supplied by
the 18M14 feeder are distinctly modeled but remaining loads
are represented as aggregated fixed loads [3].
D. Modeling of PV Solar Farms
The two 10 MW photovoltaic farms connected to the 27.6 kV,
18M14 feeder from St. Andrews TS are modeled as single
aggregated generating sources, as follows:
i) Modeling in CYME
The PV farms are modeled as active power source with unity
power factor at their Point of Common Coupling (PCC).
ii) Modeling in PSCAD/EMTDC
In PSCAD/EMTDC software, the modeling of the inverter
based PV solar farm is simplified by modeling the solar farm
as an equivalent P-Q bus for load flow validation purposes
only. However, for detailed harmonics analyses, the inverter
based PV systems are represented as a combination of
fundamental and multiple harmonic current sources [22]. The
network is considered to be balanced. The ambient harmonics
from prior field measurements are found to be less than 5%,
which validates the modeling of the PV system by harmonic
current sources [3], [10]. The line impedance between the
PCCs of two 10 MW PV systems is negligible, and hence does
not create any significant phase differences between these two
current sources containing harmonics of same order [3], [10].
E.
Validation of Network Model through Load Flow Studies
Load flow studies are first carried out using CYME
software. Subsequently steady state power flow results are
obtained from PSCAD/EMTDC simulation studies. Tables I
and II present voltages and power flows obtained from the two
software simulations for the 18M14 feeder during daytime and
nighttime, respectively. The real time data from SCADA for
the 18M4 feeder is also included in the Tables for validation
purpose. Study results summarized in the two Tables indicate
close correlation between CYME load flow results and
SCADA measurements. It was observed that power generation
mismatches between CYME results and SCADA
measurements, and that between PSCAD and CYME results
were less than 4%. Hence both the CYME and PSCAD
network models were considered as validated. The PSCAD
model was then used for further harmonic studies.
TABLE I
DAYTIME DATA OF 18M14 (TIME STAMP- 25/05/2011 AT 12:22PM)
Measurement
SCADA
CYME PSCAD/ Deviation Deviation (%)
Location
EMTDC (%) SCADA CYME vs.
vs. CYME
PSCAD
Voltage (kV)
St. Andrews 28.615(Avg.) 28.618 28.618
0.01
0
Sarnia Solar 2
28.579
28.530 28.538
0.171
0.028
Sarnia Solar 5
28.579
28.529 28.537
0.171
0.028
Load capacitor
At largest load
St. Andrews
Sarnia Solar 2
Sarnia Solar 5
St. Andrews
-
28.486 28.486
28.321 28.240
Real Power (MW)
6.821
6.825
6.82
3.023
3.136 3.023
3.252
3.136
3.25
Reactive Power (Mvar)
2.694
2.688 2.690
-
0
0.01
0.058
3.74
3.57
0.073
3.6
3.63
0.222
0.074
TABLE II
NIGHTTIME DATA OF 18M14 (TIME STAMP- 7/10/2011 AT 12:05AM)
Measurement
SCADA
CYME PSCAD/ Deviations Deviations
Location
EMTDC (%) SCADA (%) CYME
vs. CYME vs. PSCAD
Voltage (kV)
St. Andrews 28.911(Avg.) 28.914 28.916
0.01
0.007
Sarnia Solar 2
28.694
28.618 28.633
0.265
0.05
Sarnia Solar 5
28.752
28.618 28.633
0.466
0.052
Load capacitor
At largest load
St. Andrews
St. Andrews
-
28.651 28.66
28.469 28.449
Real Power (MW)
14.44
14.44
14.46
Reactive Power (Mvar)
4.417
4.42
4.42
-
0.031
0.07
0
0.138
0.068
0
2332-7707 (c) 2015 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JPETS.2016.2550601, IEEE Power
and Energy Technology Systems Journal
4
IV. NETWORK IMPEDANCE STUDIES
The frequency-domain harmonic analysis for the 18M14
feeder from St. Andrews TS was performed using validated
PSCAD/EMTDC system model. The frequency scan
calculated the Thevenin equivalent network impedances for
different frequencies at selected network locations. The plot of
positive sequence impedance magnitude versus frequency
indicated the network resonance frequencies, i.e., frequencies
at which the network exhibited high impedance magnitudes.
Frequency scan studies were carried out at the following
locations along the 18M14 feeder: St. Andrews TS, Sarnia
Solar 2 and Solar 5 Farms, SC1 and SC2 capacitor banks at St.
Andrews TS, and customer load (this load constitutes 80% of
BWP’s total load on their network systems). Studies were
performed to evaluate the influence of station capacitors and
short circuit levels on network impedance.
Case-I: Impact of Station Capacitors
Fig. 3 illustrates the frequency scans at the four locations
described above when none of the SC1 and SC2 capacitor
banks were in service. No network resonances were observed.
Fig. 4 shows the frequency scan of the case when one 20 Mvar
SC1 capacitor bank is considered connected at St. Andrews TS
and the 1.2 Mvar power factor correction capacitor bank
installed at the customer load end. This is the normal operating
configuration at St. Andrews TS and is referred as a ‘base
operating case’ scenario. Two parallel resonance peaks are
found to occur: one between the 2nd and 5th harmonic due to
resonance with the 20 Mvar SC1 capacitor, and the other
around the 25th harmonic frequency due to resonance with the
1.2 Mvar load power factor correction capacitor.
Fig. 3. Network impedance vs. harmonics plots without capacitors.
A.
B.
Fig. 4. Network impedance vs. harmonics plots for base operating case with
20 Mvar station capacitor and 1.2 Mvar load capacitor
Case-II: Impact of System Short-Circuit Levels
The short-circuit levels at St. Andrews TS varies
significantly between summer and winter time. The maximum
and minimum short circuit levels at the St. Andrews substation
are 600 MVA and 334 MVA, respectively. The X/R ratio is
27.44. Even though the SCL becomes only half in a realistic
sense, the cases of one-fifth and four-times SCL are
investigated only to examine if any extreme case adverse
impacts may be anticipated. It is observed (though not shown)
that as the short circuit level (SCL) is varied from one-fifth
nominal SCL to four-times nominal SCL, the resonant
frequency increases from 2nd to around 4th harmonic with 20
Mvar SC1 capacitor in service. Critical operating cases are
therefore chosen to result in harmonic peak at 3rd harmonic (as
there is considerable 3rd harmonic injection from solar farms).
Fig. 5. Network impedance vs. harmonics plots with one-third nominal Short
Circuit Level and 20 Mvar station capacitor (SC1) and load capacitor.
Fig. 5 and Fig. 6 show the frequency scan plots for these
cases: i) one-third of nominal SCL with 20 Mvar capacitor
(SC1), and ii) 4 times nominal SCL with 40 Mvar capacitors
(SC1+SC2), respectively. The high frequency resonance due
to the load power factor correction capacitor remains almost
unaffected by these short-circuit variations and choice of
capacitors, and is therefore not shown in Figs. 5 and 6.
Fig. 6. Network impedance vs. harmonics plots with 4 times nominal Short
Circuit Level and 40 Mvar station capacitors (SC1+SC2) and load capacitor
2332-7707 (c) 2015 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JPETS.2016.2550601, IEEE Power
and Energy Technology Systems Journal
5
V.
HARMONICS STUDIES
Harmonics
studies
were
then
performed
with
PSCAD/EMTDC software for different critical scenarios that
could potentially lead to a high harmonic distortion of the bus
voltages, and to examine if the IEEE 519 [17] criterion for
Total Harmonic Distortion (THD) was violated. These
scenarios were:
A) Scenario 1: Highest 3rd harmonic injection from solar farms
when the network is resonant at 3rd harmonic frequency (load
capacitor with SC1 or (SC1+SC2))
B) Scenario 2: High harmonic injection during a sunny day for
base operating case (load capacitor with SC1)
C) Scenario 3: High harmonic injection during a cloudy day
for base operating case (load capacitor with SC1)
D) Scenario 4: High harmonic injection with network having
no resonances (no capacitors).
In all the cases, ambient harmonics were not considered.
A. Scenario 1: High 3rd Harmonic Injection with Network
Resonance at 3rd Harmonic
As the network demonstrates a resonance at third harmonic
(Fig. 5 and 6), a harmonic study is performed for the critical
case of maximum third harmonic emanation from Solar 2 and
Solar 5 PV farms for these network conditions. Daily
harmonics data provided by Hydro One over the month of
June 2012 was analyzed and the harmonic data for a specific
day containing a high third harmonic injection from the solar
farms was selected for this study, which is shown in Fig. 7.
Fig. 7. Harmonics data for high third harmonic injection from solar farms.
Simulation results for three cases are obtained: i) base
operating case (Fig. 4), ii) one-third nominal short-circuit level
and SC1 (20 Mvar) connected along with load capacitor (Fig.
5); and iii) 4 times nominal SCL, with SC1 and SC2 (40 Mvar)
connected along with load capacitor (Fig. 6); and are depicted
in Tables III, IV and V, respectively. These Tables summarize
the power flows, bus voltages, TDDs and THDs at the PCCs
of St. Andrews TS, Solar 2 and Solar 5 PV farms. The
following observations are made while comparing Tables III,
IV and V:
1. The common feature among the study cases in three Tables
is that the network resonance is at 3rd harmonic. The
corresponding 3rd harmonic impedance is about 16.5 Ω and
this applies to all study cases. The third harmonics result in
peak voltages at this third harmonic frequency.
TABLE III
POWER, VOLTAGE AND HARMONICS FOR BASE OPERATING CASE OF FIG. 4
St. Andrew
Solar farm
Solar farm
Quantities
SS
SS2
SS5
Power, P+jQ (MVA)
-6.561+j5.141 10+j0.2831 10.02+j0.277
Fundamental Voltage, (kV/ph.)
16.6277
16.738
16.735
Fundamental Current, If (Amp)
167.88
199.52
199.84
Current TDD (%)
1.063
1.063
Voltage THD (%)
0.2053
0.66
0.653
Harmonic Distortion (kV)
0.03415
0.1104
0.1092
3rd harmonic voltage (kV)
0.0137
0.02056
0.0204
5th harmonic voltage (kV)
0.02485
0.0235
0.02336
7th harmonic voltage (kV)
0.01132
0.01715
0.01687
11th harmonic voltage (kV)
0.011
0.06
0.05067
TABLE IV
POWER, VOLTAGE AND HARMONICS FOR CASE OF FIG. 5
St. Andrew
Solar farm
Solar farm
Quantities
SS
SS2
SS5
Power, P+jQ (MVA)
-6.397+j6.239 9.927-j0.248
9.941j0.255
Fundamental Voltage, (kV/ph.)
16.554
16.61
16.608
Fundamental Current, If (Amp)
180.64
199.52
199.84
Current TDD (%)
1.063
1.063
Voltage THD (%)
0.185
0.66
0.653
Harmonic Distortion (kV)
0.03065
0.1097
0.1085
3rd harmonic voltage (kV)
0.0137
0.02056
0.0204
5th harmonic voltage (kV)
0.0206
0.017
0.0169
7th harmonic voltage (kV)
0.01
0.0178
0.01754
11th harmonic voltage (kV)
0.0105
0.061
0.0602
TABLE V
POWER, VOLTAGE AND HARMONICS FOR CASE OF FIG. 6
St. Andrew
Solar farm
Solar farm
Quantities
SS
SS2
SS5
Power, P+jQ (MVA)
-6.931+j4.736 10.19+j0.458 10.2+j0.4525
Fundamental Voltage, (kV/ph.)
16.926
17.057
17.054
Fundamental Current, If (Amp)
166.23
199.52
199.84
Current TDD (%)
1.063
1.063
Voltage THD (%)
0.1085
0.6874
0.6805
Harmonic Distortion (kV)
0.018376
0.117259
0.116056
3rd harmonic voltage (kV)
0.00816
0.018
0.0177
5th harmonic voltage (kV)
0.0129
0.015
0.0147
7th harmonic voltage (kV)
0.0056
0.02178
0.0215
11th harmonic voltage (kV)
0.0056
0.0671
0.0664
2. The fifth harmonic voltage measured at the PV farms and
shown in Tables IV and V were found to be lower than that
given in Table III. This is further depicted in frequency
scans shown in Figs. 5 and 6 where corresponding 5th
harmonic impedances (<10Ω) are substantially lower when
compared to the 5th harmonic impedance shown in Fig. 4
(>10Ω). Hence the magnitude of the 5th harmonic voltage is
obtained to be lower.
3. The 3rd, 5th, 7th and 11th harmonic voltages measured at St.
Andrews TS and summarized in Table V were found to be
lower than corresponding values given in Tables III and IV
due to lower impedances at those corresponding frequencies.
4. Finally, the THDs at St. Andrews TS as shown in Table V
were found to be lower than the other two cases given in
Tables III and IV. This is due to the relatively lower
harmonic voltages (VTHD) obtained at the 3rd, 5th, 7th and 11th
harmonics for the high value of system strength considered.
2332-7707 (c) 2015 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JPETS.2016.2550601, IEEE Power
and Energy Technology Systems Journal
6
Frequency scan analyses show that, even in the event of 3rd
harmonic resonance, there will not be any adverse impact on
the 18M14 feeder network system in terms of voltage THD.
Further, variation in the short-circuit levels to reflect seasonal
climatic condition changes at both 27.6 kV and 115 kV PCCs,
does not cause any substantial change in the voltage THD
which is well within acceptable limits. In this study, the
harmonic currents injected by Sarnia PV farms are assumed to
remain the same regardless of changes in network conditions.
C. Scenario 3: High Harmonic Injection on a Cloudy Day
The PV power generation on a highly cloudy day is depicted in
Fig. 10. The corresponding harmonic spectrum, current THD
and TDD are also calculated and included. Fig. 11 illustrates a
snapshot of maximum total demand distortion (TDD) for each
of the harmonics generated on this highly cloudy day, which
occurred at 11:50 am.
B. Scenario 2: High Harmonic Injection on a Sunny Day
Fig. 8 shows the power output from one of the PV farms on a
sunny day. The corresponding harmonic spectrum, current
THD and TDD are then derived and plotted. It is noted that the
magnitudes of injected harmonics vary based on the amount of
solar power generation. Fig. 9 depicts a snapshot of maximum
total demand distortion (TDD) for each harmonic generated on
a typical sunny day, which occurred at 8:20 am. Table VI
summarizes all the calculated harmonic parameters.
Fig. 10. 10 MW Solar farm output for a cloudy day.
Fig. 11. High harmonics dataset on a cloudy day.
Fig. 8. 10 MW Solar farm output for a sunny day.
Fig. 9. High harmonics dataset on a sunny day.
TABLE VI
POWER, VOLTAGE AND HARMONICS DISTORTION FOR HIGH HARMONICS
INJECTION ON A SUNNY DAY FOR BASE OPERATING CONDITION (FIG. 4)
St. Andrew
Solar farm
Solar farm
Quantities
SS
SS2
SS5
Power, P+jQ (MVA)
7.844+j5.445 2.719-j0.153 2.719-0.153
Fundamental Voltage, (kV/ph.) 16.672
16.564
16.564
Fundamental Current, If (Amp) 189.44
54.88
54.88
Current TDD (%)
1.474
1.474
Voltage THD (%)
0.2866
0.843
0.835
Harmonic Distortion (kV)
0.0478
0.1397
0.1382
TABLE VII
POWER, VOLTAGE AND HARMONICS DISTORTION FOR HIGH HARMONICS
INJECTION ON A CLOUDY DAY FOR BASE OPERATING CONDITION (FIG. 4)
St. Andrew
Solar farm
Solar farm
Quantities
SS
SS2
SS5
Power, P+jQ (MVA)
7.179+j5.451 3.044-j0.1624 3.044-j0.163
Fundamental Voltage, (kV/ph.) 16.68
16.58
16.58
Fundamental Current, If (Amp) 178.64
61.28
61.28
Current TDD (%)
1.506
1.506
Voltage THD (%)
0.422
1.164
1.153
Harmonic Distortion (kV)
0.07
0.193
0.191
Table VII summarizes the voltage THD and current TDD for
this case. Comparing the results in Tables VI and VII, it is
noted that the voltage THD on a cloudy day is much higher
due to a substantially increased harmonic generation than on a
sunny day. Still the voltage THD is within the acceptable
limits per IEEE Standard 519.
D. Scenario 4: High harmonic injection with network having
no resonances
The above studies were conducted for a network that was
resonant around 3rd harmonic. However, it was noted that the
network impedance at 3rd harmonic resonance (in Figs. 4, 5
and 6) is lower than the network impedance at 5th, 7th, and 11th
harmonics for the no capacitor case (Fig. 3). It was therefore
decided to perform a THD evaluation for a non-resonant
2332-7707 (c) 2015 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JPETS.2016.2550601, IEEE Power
and Energy Technology Systems Journal
7
network (with varying short circuit levels) and with a generally
very high harmonic injection in the range of frequencies
beyond 5th harmonic. An evaluation was carried out over a
one month period to determine a set of very high harmonics
injection over a broad range, which is shown in Fig. 12. Table
VIII summarizes the simulation results for the case of 1SCL.
It is further observed that the voltage THD at Solar Farm 2
increases from 1.094 to 1.129 when the system strength
decreases from 1SCL (Table VIII) to 1/3rd SCL (not shown).
day and 0.843% on sunny day). This is because ambient
harmonics were not considered in PSCAD simulation studies.
Fig. 13 (a): Sarnia Solar 2 PV Farm Power Output (MVA)
Fig. 12. Dataset of High harmonics above 5th harmonic
TABLE VIII
SIMULATED POWER, VOLTAGE AND HARMONICS CORRESPOND TO HIGHEST
HARMONICS INJECTION BY THE SOLAR FARM AT 1 SCL (FIG.3)
St. Andrew
Solar farm
Solar farm
Quantities
SS
SS2
SS5
Power, P+jQ (MVA)
7.305+j6.77 2.98-j0.1796 2.98-j0.1802
Fundamental Voltage, (kV/ph.) 16.657
16.5266
16.526
Fundamental Current, If (Amp) 197.824
60.24
60.24
Current TDD (%)
1.567
1.567
Voltage THD (%)
0.429
1.094
1.085
Harmonic Distortion (kV)
0.0715
0.181
0.17936
VI. THD AT PV SOLAR FARMS
Hydro One Networks Inc. monitored the harmonics at St.
Andrews TS and Sarnia Solar 2 and Solar 5 PV farms using
Schneider ION 7650 meters. These meters were located at the
PCCs where they were sampled at 128 samples/cycle for 14
cycles recording voltage and current waveforms.
Measurements taken at the 10 MW Sarnia Solar 2 PV farm
over the week of June 8 – 15, 2011 depicting full sunny days
and partial cloud covered days, are shown in Fig 13 (a). The
corresponding computed Total Harmonic Distortions (THDs)
are shown in Fig. 13(b). Similar data was recorded and
corresponding THD values were obtained at 10 MW Sarnia
Solar 5 PV farm (although not shown).
The dominant harmonics were seen to be the 5th, 7th, 11th
and 13th harmonics. The highest voltage THD at Sarnia Solar 2
PV farm was observed to be 2.1% on the most cloudy day
Monday. The lowest voltage THD was noted to be 1.5% on
the full sunny day Wednesday. These measurements
exemplified that the voltage THD is higher on cloudy days
owing to high harmonic injections. A similar trend of voltage
THDs was also observed in the simulation studies for sunny
and cloudy day reported in Tables VI and VII, respectively,
which correlates with actual measurements. The measured
voltage THDs are noted to be generally higher (2.1% on
cloudy day and 1.5% on sunny day) compared to the voltage
THDs obtained from simulation studies (1.164% on cloudy
Fig. 13 (b): Sarnia Solar 2 PV Farm THD Voltages
VII.
IMPACT OF AMBIENT HARMONICS
The ambient harmonics were measured at the solar farm
PCCs and the voltage THD was found to be about 1.22%
during the same period. It is reasonable to extrapolate that the
above ambient harmonics can cause the simulated voltage
THDs to increase from by 0.936% (2.1% - 1.164%) on cloudy
days and by 0.657% (1.5% - 0.843%) on sunny days. The
ambient harmonics do not directly add with the injected
harmonics due to harmonic cancellation [2]-[3],[10]. Although
the measured voltage THD is within the limits specified in
IEEE Standard 519, this could worsen in situations where
higher ambient harmonics exist. For example, the maximum
ambient voltage harmonic distortion in a distribution utility
network was measured to be 2.69% [14]. If this distortion level
were to exist at Sarnia Solar 2 (10 MW) and Solar 5 (10 MW)
PV farms, the maximum voltage THD could possibly exceed
3.5%. Although this lies within the IEEE-519 Standards limit,
it might not be acceptable for critical voltage sensitive loads
such as in hospitals, where the distortion limit should be kept
below 3% due to sensitive medical electronic equipment [23].
VIII. DISCUSSIONS
It is understood that an accurate impact of harmonics can
only be assessed based on a system model that is validated
both at fundamental frequency and harmonic frequencies.
2332-7707 (c) 2015 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JPETS.2016.2550601, IEEE Power
and Energy Technology Systems Journal
8
However, per the Summary of the paper [3] by the IEEE Task
Force on Harmonics Modeling and Simulation, “…In general,
a harmonic analysis method is a function of the problem to be
solved. The analysis method should represent a compromise
between the required accuracy of the solution and the
availability of system data”.
The study of the harmonic impact of such a large scale 20
MW PV solar farm was being performed for the first time in
the utility network of Bluewater Power. There was no prior
system model available for such a study. Several pieces of
information about network components on this 7.8 km long
line and cable segments were either outdated or even missing.
A significant effort involving two students from Western
University and several staff members from Bluewater Power
utility was launched to determine the network data. Each and
every component of the network was actually inspected, cable
sizes were physically measured and their data was obtained. In
several cases where no data was available, Westinghouse
handbook and data-books from different sources were utilized
to either obtain or extrapolate data. Several cycles of load
flows were run and matched with SCADA results. The model
was continuously refined as more and more data got acquired,
and the accuracy became improved. This continued till an
accuracy of less than 4% was obtained between the CYME
load flow results and SCADA measurements for fundamental
frequency voltage and currents, and power flows. This entire
exercise took more than nine months.
Harmonic analysis was then performed for several cases and
the trends were validated with the data acquired by the ION
meters installed by Hydro One only for this project. Two
important factors were noted: i) the voltage THD caused by
the ambient harmonics was measured to be about 1.22% , and
ii) the worst case voltage THD over a one year period was
measured to be 2.3% (including ambient harmonics), which
was well within the THD limits stipulated by IEEE Std. 519.
Considering the above two facts, the Summary of [3] stated
above, and the significant time and manpower effort required
to develop only the fundamental frequency validated model of
the network, a joint decision was taken by the project team
comprising Western University, Bluewater Power and Hydro
One, that efforts to develop an elaborate model validated for
harmonic frequencies may not be justified in this specific case.
If the worst case THD was to get close to limits specified by
[17], this detailed modeling work might have been undertaken.
IX. CONCLUSION
This paper presents one of the first harmonic studies for a
large 20 MW PV solar farm in Bluewater Power distribution
utility network in Canada. A detailed model of the 27.6 kV
network including PV solar farms is developed in CYME load
flow software and validated with SCADA measurements. The
validated network is utilized in PSCAD/EMTDC software for
extensive network resonance and harmonic studies. Harmonic
analyses were conducted for varying system short circuit levels
and solar radiation. Worst case harmonic injections were
determined from data accumulated over several months of
solar farm operation. The following conclusions are made:
1. Network resonances are caused by SC1 and SC2 capacitor
banks at St. Andrews TS and by power factor capacitor at
the load. The 20 Mvar SC1 or SC2, or the total 40 Mvar
(SC1 + SC2) capacitors at St. Andrews TS cause network
resonance around the 3rd harmonic. The 1.2 Mvar power
factor correction capacitor causes network resonance
around the 25th harmonic.
2. The PV farms typically generate 5th, 7th, 11th, and 13th
characteristic harmonics, although a considerable 3rd
harmonic is also emanated.
3. A higher voltage THD occurs at the terminals of solar farms
during cloudy days as compared to sunny days due to
higher harmonic injection during cloudy periods.
4. High voltage THD at the solar farms can occur with
excessive harmonic injection (during cloudy day) even
when the network does not have a resonance (Table 8).
5. The worst case voltage THD on solar farms was observed
for a cloudy day for a network which was resonant at 3rd
harmonic.
6. The trend of voltage THDs obtained through
PSCAD/EMTDC simulations correlated with that
obtained from actual measurements at the solar farms.
7. The actually measured voltage THDs at the solar farms
were higher than the simulated values for those cases due
to ambient harmonics which were not considered in
simulation studies.
8. Ambient harmonics can potentially increase the harmonic
impact of PV solar farms to unacceptable levels, more so
during weak system conditions.
9. In all simulations and measurements, the steady state
voltage THDs at solar farms and at St. Andrews TS were
well within the 5% limit specified by IEEE Standard 519.
The impacts of ambient harmonics and harmonics generated
during transient situations (capacitor switching etc.) have not
been considered in this simulation study. Based on the
systematic procedure described in this work, it is
recommended that utilities having PV solar farms perform
detailed harmonic studies to determine the potential voltage
THDs that can occur, especially if their networks are weak,
have shunt capacitor resonances, and if their lines have high
ambient harmonics. This study is almost essential if utilities
have critical voltage sensitive loads such as hospitals etc.
X.
ACKNOWLEDGMENT
This study has been conducted with extensive data provided
by Bluewater Power Corp., Hydro One Networks Inc., and
with the cooperation of Enbridge Inc. and First Solar, all of
whom are sincerely acknowledged. The authors thank John
Lepoutre and his measurement team from MSP (Metering
Service Provider) Division of Hydro One in Mississauga.
Thanks are also expressed to graduate students Byomakesh
Das for help in data collection, and to Vishwajitsinh Atodaria
in improvising the figures.
2332-7707 (c) 2015 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JPETS.2016.2550601, IEEE Power
and Energy Technology Systems Journal
9
XI. REFERENCES
[1]
[2]
[3]
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
[12]
[13]
[14]
[15]
[16]
[17]
[18]
[19]
[20]
[21]
[22]
Oliva, A.R.; Balda, J.C., “A PV dispersed generator: a power quality
analysis within the IEEE 519”, IEEE Transactions on Power Delivery,
Vol.18, No. 2, pp. 525 -530, 2003
Enslin, J.H.R.; Heskes, Peter J.M., “Harmonic interaction between a
large number of distributed power inverters and the distribution
network”, IEEE Transactions on Power Electronics, Vol. 19, No. 6,
pp. 1586 - 1593, 2004
Task Force on Harmonics Modeling and Simulation, “Modeling and
Simulation of the Propagation of Harmonics in Electric Power
Networks-Part I: Concepts, Models and Simulation Techniques,” IEEE
Trans. on Power Delivery, Vol. 11, No. 1, pp. 452-465, Jan. 1996.
J. Arrillaga, D.A. Bradley, P.S. Bodger, Power System Harmonics, 2nd
ed., John Wiley and Sons, New York, 1985.
G. Lemieux, “Power System Harmonic Resonance- A Documented
Case,” IEEE Trans. On Industry applications, vol. 26, no. 3, 1990.
Zhenyu
Huang; Wilsun
Xu; Dinavahi,
V.R.,
“A
practical harmonic resonance guideline
for
shunt
capacitor
applications”, IEEE Trans. on Power Delivery, vol.18, no.4 pp. 1382 1387, 2003
Chris Halliday, Frank Iannelli and Robert Barr, “Distribution Network
Capacitor Resonance – A Case Study,” Energy NSW 2007, Balancing
the New Energy Equation Conference & Trade Exhibition in Sydney.
Paul Martin Donohue and Syed Islam “The Effect of Nonsinusoidal
Current Waveforms on Electromechanical and Solid-State Overcurrent
Relay Operation”, IEEE Trans. on Industry Applications, Vol. 46, No.
6, pp. 2127-2133, Nov/Dec 2010
D. Mueller, "Case studies of harmonic problems, analysis, and solutions
on transmission systems,"
in Proc. 2007 IEEE International
Conference on Electrical Power Quality and Utilization, pp.1-6.
Task Force on Harmonics Modeling and Simulation, “Modeling And
Simulation of the Propagation of Harmonics in Electric Power
Networks-Part II: Sample System and Examples,” IEEE Transactions
on Power Delivery, vol. 11, no. 1, pp. 466-474, 1996.
D. Patel, R. K. Varma, R. Seethapathy and M. Dang, “Impact of Wind
Turbine Generators on Network Resonance and Harmonic Distortion,”
in Proc. 23rd Canadian Conference of Electrical and Computer
Engineering (CCECE-2010), pp. 1-6, 2010.
A. Khodabakhshian, R. Hooshmand and G. Isazadeh, “Harmonic
Resonance and Over-Voltage Assessment During Switching of High
Voltage Shunt Capacitors in E.R.E.C. Network,” in Proc. 2009
Canadian Conf. on Electrical and Computer Engineering, pp. 926-930.
W. Xu, "Status and future directions of power system harmonic
analysis," in Proc. 2003 IEEE Power Engineering Society General
Meeting, pp. 4 vol. 2666, 13-17 July 2003.
Christine D. Crites, Rajiv K. Varma, Vinay Sharma and Bill Milroy,
“Characterization of Harmonics in a Utility Feeder with PV Distributed
Generation,” in Proc. 2012 IEEE Electrical Power and Energy
Conference, pp. 1-6.
R. K. Varma, J. Berge, I. Axente, V. Sharma, K. Walsh, "Determination
of maximum PV solar system connectivity in a utility distribution
feeder," in Proc. 2012 IEEE Transmission and Distribution Conference
and Exposition (T&D), pp.1-8.
A. Bhowmik, A. Maitra, S. M. Halpin, and J. E. Schatz, “Determination
of allowable penetration levels of distributed generation resources based
on harmonic limit considerations,” IEEE Trans. Power Del., vol. 18, no.
2, pp. 619–624, Apr. 2003
IEEE Recommended Practices and Requirements for Harmonic Control
in Electrical Power Systems, IEEE Standard 519-1992, 1993.
http://www.theobserver.ca/2010/10/05/worlds-largest-solar-farm-opens.
CYME International (Quebec, Canada), http://www.cyme.com/.
PSCAD/EMTDC User Manual, HVDC Research Center, Manitoba,
2003.
Task Force on Harmonics Modeling and Simulation, IEEE Power
Engineering Society T & D Committee, “Impact of Aggregate Linear
Load Modeling on Harmonic Analysis: A Comparison of Common
Practice and Analytical Models,” IEEE Trans. On Power Delivery, vol.
18, no. 2, pp. 625-630, 2003.
IEEE PES Harmonic Working Group, “Characteristics and Modeling of
Harmonic Sources—Power Electronic Devices Task Force on
Harmonics Modeling and Simulation”, IEEE Transactions on Power
Delivery, Vol. 16, No. 4, pp.791, Oct. 2001
[23] M.I.
Buzdugan,
H.
Balan,
D.T. Mureşan,
“An
electrical power quality problem in an emergency unit from a hospital –
Case Study”, 2010 IEEE Int’l Symp. on Power Electronics Electrical
Drives Automation and Motion (SPEEDAM), pp. 251 – 256, 2010.
BIOGRAPHY
Rajiv K. Varma (SMIEEE 2009) obtained
B.Tech. and Ph.D. degrees in Electrical
Engineering from Indian Institute of Technology
(IIT), Kanpur, India, in 1980 and 1988,
respectively. He is currently Associate Professor
and was Hydro One Chair in Power Systems
Engineering (2012-15) at the University of
Western Ontario (UWO), Canada. Prior to this
position, he was a faculty member in the
Electrical Engineering Department at IIT Kanpur,
India, from 1989-2001. He has co-authored an IEEE Press/Wiley book on
Thyristor Based FACTS Controllers. He is active on a number of IEEE
working groups. He has co-delivered several Tutorials on SVC sponsored by
IEEE Substations Committee. His research interests include FACTS, power
systems stability, and grid integration of wind and photovoltaic solar power
systems. He is the Chair of IEEE Working Group 15.05.17 on “HVDC and
FACTS Bibliography,” since 2004, and also the Secretary of the IEEE
“HVDC and FACTS Subcommittee” since 2015.
Shah Arifur Rahman (MIEEE 2013) received
his Ph.D. degree from the University of Western
Ontario, London, ON, Canada. He is currently
working as a Post-Doctoral Fellow with the
University and performing research activities at
Bluewater Power, Sarnia. His research interests
include grid integration of inverter based
Distributed Generation (DG) sources such as
photovoltaic (PV) solar plants, wind farms,
energy storage etc., impact analysis of harmonics
and overvoltages on power systems, implementation of FACTS capability in
inverter based DGs and their coordination. He was a member of the IEEE
Working Group on “HVDC and FACTS Bibliography and Records” since
2010 and served as Secretary during 2012-14.
Tim Vanderheide holds the position of Chief
Operating Officer for both Bluewater Power
Renewable Energy Inc. and Electek Power Services
Inc. Tim is also Vice President of Strategic
Planning for Bluewater Power Distribution
Corporation. Tim is responsible for the
development and implementation of renewable
power generation projects for Bluewater Power
Renewable Energy Inc. as well as the development
of new product and service strategies designed to continuously improve
shareholder value for Bluewater Power Distribution. In his role as Chief
Operating Officer for Electek Power Services Inc., Tim is responsible for
overall operations and company growth. Prior to his current positions, Tim
was Vice-President of Client Services for Bluewater Power Distribution
Corporation. In this role, Tim was responsible for market services, energy
services, metering, billing and information technologies.
Michael D. N. Dang: obtained his B.Sc. (Hon.) in
1968, M.Sc. in 1969 and Ph.D. in 1972 all in
Electrical Engineering from the University of
Manchester Institute of Science & Technology,
England. He worked for the Central Electricity
Generating Board in London before immigrating to
Montreal, Canada, in 1981 and joining Shawinigan
Consultants Inc. He came to Toronto and joined
Ontario Hydro/Hydro One in 1988. His major study
areas included power system protection, harmonics,
power system analyses, system operations and connections of combined-cycle
and wind-turbine generation to the Grid. Dr. Dang is a registered professional
engineer in the Province of Ontario, a Fellow of Engineers Canada in 2010
and a member of the Experience Requirement Committee of Professional
Engineers Ontario.
2332-7707 (c) 2015 IEEE. Translations and content mining are permitted for academic research only. Personal use is also permitted, but republication/redistribution requires IEEE permission. See
http://www.ieee.org/publications_standards/publications/rights/index.html for more information.