5366 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 51, NO.

6, NOVEMBER/DECEMBER 2015

On the Design and Capacity of a Grounding

Configuration for Grid-Connected DGUs
S. A. Saleh, Senior Member, IEEE

Abstract—Established standards and industrial codes, for in-

terconnecting distributed generation units (DGUs), address volt-
age and frequency changes, levels and quality of injected power,
islanding conditions, and protection. However, these standards
and codes do not clearly address the grounding configurations
for grid-connected DGUs. One of the challenges related to the
grounding configuration for DGUs is the harmonics generated by
power electronic converters employed in DGUs. These harmonics
can raise the ground potentials and disrupt the function of ground
fault protection. This paper develops and tests a frequency-
selective grounding configuration for grid-connected DGUs. The
developed grounding configuration offers limiting ground poten-
tials and ground fault currents, while imposing negligible impacts
on the function of ground fault protective devices. The frequency-
selective grounding configuration is designed and tested for two
wind energy conversion systems and a photovoltaic system. Test
results show that the developed grounding configuration achieves
its objectives with negligible impacts on the grounded DGU and its
ground fault protection. Fig. 1. Schematic single-line diagram for a DGU farm [5].

Index Terms—Distributed generation units (DGUs), ground remedies to some operational challenges, such as the grounding
fault protection, ground potentials, grounding configurations. configurations [8]–[15].
Rising demands for renewable and sustainable electric power
N OMENCLATURE generation have pushed toward increasing the number of DGUs
DFIG Doubly fed induction generator. installed at the same location, which is called a DGU farm (e.g.,
DGU Distributed generation unit. wind farms and solar farms) [8]. This trend of operating DGUs
PCC Point of common coupling. has demonstrated encouraging performance in terms of good
PEC Power electronic converter. compliance with the grid codes [5], [6]. The operation, control,
PMG Permanent-magnet generator. and protection of DGU farms have been subjects for several
PV Photovoltaic. research works, examples of which can be found in [5]–[19]
WECS Wind energy conversion system. and the references therein. One of the recommended designs
for the electrical system in a DGU farm is the collector system,
I. I NTRODUCTION which facilitates delivering the generated power by DGUs to an
interconnection substation, as shown in Fig. 1. The employment

T HE growing interconnection of DGUs, such as WECSs,

cogeneration units, and PV systems, has created several
challenges for host utility grids [1]–[9]. These challenges in-
of the collector system has offered more integrable grounding
systems of DGU farms. One of such layouts is established by
employing Δ–Y transformers for the grid-connection and DGU
clude variations in voltages and frequency at the PCC, quality transformers. On one hand, the Y side of the grid-connection
of injected power, protection against faults, possible redistribu- transformer is grounded to offer a neutral grounding for the
tions of load currents, etc. [6]–[8]. The introduction of the IEEE collector system. On the other hand, the Y side of the DGU
Standard 1547 has identified several basic requirements for transformer offers the DGU a solid grounding and an isolation
interconnecting DGUs to utility grids. However, this standard from zero-sequence components [5]–[9].
identifies such requirements on a fundamental level with limited The majority of grounding configurations in grid-connected
DGUs are based on the solid grounding. The solid grounding
Manuscript received December 26, 2014; accepted January 13, 2015. Date of
publication January 21, 2015; date of current version November 18, 2015. Paper configuration offers eliminating ground potentials and isolating
2014-PSEC-0832, approved for publication in the IEEE TRANSACTIONS grid-connected DGUs from zero-sequence components. How-
ON INDUSTRY APPLICATIONS by the Power Systems Engineering Com- ever, this grounding configuration neither reduces the usage of
mittee of the IEEE Industry Applications Society.
The author is with the Department of Electrical and Computer Engineering, the grounding system nor does it limit ground fault currents.
University of New Brunswick, Fredericton, NB E3B 5A3, Canada (e-mail: Other grounding configurations used in grid-connected DGUs
[email protected]). include the open grounding configuration, which eliminates the
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. usage of grounding systems and blocks ground fault currents.
Digital Object Identifier 10.1109/TIA.2015.2394448 Nonetheless, the open grounding configuration can result in
0093-9994 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
SALEH: ON THE DESIGN AND CAPACITY OF A GROUNDING CONFIGURATION FOR GRID-CONNECTED DGUs 5367

high ground potentials. High ground potentials pose safety haz- minimize the hazard of electric shocks [5]. In some WECSs,
ards and can affect the operation of grounded DGUs [1]–[11]. the collector system is not connected to the grounding system,
This paper develops a frequency-selective grounding con- where the concentric neutral acts as a bonding conductor and as
figuration for grid-connected DGUs. The developed grounding a metallic shield [5]–[19].
configuration is established as a parallel R–C circuit in order
to create a low-impedance path for current harmonics and a
low-resistance path for ground fault currents. The components B. Grounding of PV Systems
of the R–C circuit are selected based on the acceptable levels The NEC Section 690.43 mandates the establishment of
of ground fault currents and ground potentials, as well as grounding systems for any exposed metallic parts, which can
ratings of PECs used in the grounded DGU. The frequency- become energized, in PV systems operating at any voltage.
selective grounding configuration is designed and tested for This NEC section applies for PV module frames, enclosures,
2-kW DFIG-based and 3.5-kW PMG-based WECSs and a metallic mountings, circuit breakers, electrical connections,
1.5-kW PV system. Experimental results demonstrate the abil- metallic conduits, and PECs. The compliance with the NEC
ity of the developed grounding configuration to effectively limit mandate is typically accomplished by bonding all metallic
ground potentials and reduce ground fault currents, without frames of PV modules to the mounting racks, which are also
disrupting the functions of ground protective devices. bonded to all metallic enclosures. Specific types of mounting
racks are considered as equipment grounding conductors and
II. OVERVIEW OF G ROUNDING S YSTEMS IN can be bonded to the grounding system [2]–[4], [12]–[14]. In
G RID -C ONNECTED DGUs addition, this mandate for grounding applies for the dc elements
(PV collecting systems and dc choppers).
Similar to other components in a power system, grid-
The NEC demands that terminals and leads of grid-side
connected DGUs have to be featured with grounding (equip-
dc–ac PECs (inverters) have to be bonded to the system ground-
ment and system), bonding, and protection against lightning.
ing. As these PECs have dc inputs and ac outputs, both sides
These features are mandated by the National Electrical Safety
should be bonded on the metallic enclosure of the dc–ac PEC. It
Code and the National Electrical Code (NEC) to protect
should be noted that the dc terminals and leads can be connected
equipment and personnel, reduce noise and interference, and
to the same grounding busbar together with ac ones, or dc
facilitate stable functions of electrical and control devices.
terminals and leads can be connected to a separate grounding
The grounding feature of grid-connected DGUs has to satisfy
busbar, which has to be connected to the ac terminal grounding
the constraints imposed on grounding any generating unit in
busbar. Due to the latest trends of operating PV systems with
a power system. These constraints mainly focus on limiting
storage units that are typically installed on the collector dc
ground potentials and ground fault currents without disrupting
buses, equipment grounding conductors should be routed with
the function of ground fault protection [1]–[5], [12], [13].
the dc conductors. This modification is made to ensure that
current harmonics (produced during the charging/discharging
A. Grounding of WECSs cycles of the storage system) do not interact with the enclosures
The fundamental structure of a WECS is composed of the of the PV system. [1]–[4], [12]–[16].
wind turbine, electric generator, power converters, and trans-
former. Most of commercial WECSs are designed to include all
components within the nacelle, which is located at the top of C. Grounding of Interconnection Substations
the wind turbine tower. Such a structure of WECSs makes the There are several standards for the design, layout, and test-
lightning protection among the critical safety and operational ing of grounding systems in substations, including substations
requirements. The lightning protection is realized by installing that interconnect DGUs. The IEEE Standard 80-2000, IEEE
lightning terminals within each blade and then connecting these Standard 142-2007, and IEEE Standard 399-1997 are widely
terminals through brushes to the turbine shaft. These brushes practiced in designing, establishing, and testing grounding grids
create a bonding of the shaft with the mounting structure of the for substations [1]–[5], [20]–[24]. The ground grid in a substa-
nacelle and other electrical components (inside the nacelle or tion is laid out as a continuous conductor, which surrounds the
outside the tower). The mounting structure is finally bonded fences of the protected substation and creates a loop around the
to the tower, which has its base connected to a ground grid substation. This loop has conductors that are made in parallel
that is equipped with ground rods [5]. The ground grid in a lines that run as close as possible to the substation equipment
WECS can function as an electrical grounding system, as well in order to have short connection between the equipment and
as a connection to earth for the lightning protection [5]–[10]. the ground grid. At junction points within the ground grid, the
Manufacturers of WECSs require the resistance of the ground conductors have to be bonded. Finally, ground rods are installed
grid to be 1–10 Ω. However, several field measurements show at the corners and junction points along the perimeter of the
that the resistance of ground grids is close to 1 Ω [1]–[13]. ground grid. It should be noted that ground rods can be installed
The bonding of a WECS to an interconnection substation close to major substation equipment, such as surge arresters.
can be established through the collector system. A copper or The standards for substation grounding require testing for the
a copper-clad ground conductor can be included along with soil resistivity, where acceptable ground resistance range be-
the main conductors of the collection system. This ground tween 1 and 5 Ω. Ground grid conductors are selected to handle
conductor can also help reducing voltage buildups in order to system ground fault currents over their durations [5], [15], [24].
5368 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 51, NO. 6, NOVEMBER/DECEMBER 2015

Fig. 2. Ground currents during steady-state operation of solidly grounded DGUs. (a) Ground current and its spectrum for a 2-kW DFIG-based WECS.
(b) Ground current and its spectrum for a 3.5-kW PMG-based WECS. (c) Ground current and its spectrum for a 1.5-kW PV system. IG scale: 1 A/Div.; |IG (f )|
scale: 0.3 A/Div.; time scale: 20 ms/Div.

Fig. 3. Ground potentials during steady-state operation of open-grounded DGUs. (a) Ground potential and its spectrum for a 2-kW DFIG-based WECS.
(b) Ground potential and its spectrum for a 3.5-kW PMG-based WECS. (c) Ground potential and its spectrum for a 1.5-kW PV system. VG scale: 25 V/Div.;
|VG (f )| scale: 10 V/Div.; time scale: 20 ms/Div.

The previous overview of the standards and codes highlights tials and their spectra are shown in Fig. 3 for open-grounded
their focus on the structures of grounding and bonding different DGUs during steady-state conditions. It is shown in Figs. 2
components in grid-connected DGUs, as well as their host and 3 that PECs used in DGUs generate harmonic components
substations. These standards and codes also address required that are continuously flowing to the ground. These steady-state
changes in the grounding and bonding to accommodate new harmonics are very dependent on the grounding configuration.
technologies utilized in grid-connected DGUs [1]–[5], [12],
[13]. Nevertheless, the standards and codes do not clearly spec-
ify the requirements for the grounding configuration in grid- III. F REQUENCY-S ELECTIVE G ROUNDING
connected DGUs [15]–[20]. The grounding configuration of C ONFIGURATION FOR G RID -C ONNECTED DGUs
any component in a power system is usually determined based
A. Grounding Configurations for Synchronous
on the impedance, i.e., ZG , of the branch connecting the neutral
Generating Units
to the grounding system. The value of and circuit structure of
ZG are selected based on the desired levels of ground fault The grounding of conventional generating units (syn-
current and ground potential. In case of a grid-connected DGU, chronous generators) represents a critical requirement for pro-
the design of ZG has to consider the following [14]–[24]: tecting these units from excessive damages during ground
faults. The solid grounding configuration is rarely used as it
• the current harmonics generated by PECs and their flow to
offers no limitations on ground fault currents, which can inflect
the grounding system;
severe damages to generating units. Recommended grounding
• the significant capacitance of the cables used for con-
configurations for synchronous generating units include low
necting DGUs to their transformers and implementing the
resistance and neutral reactance. These grounding configura-
collector system.
tions are capable of reducing ground fault currents and limiting
In order to illustrate the challenges in grounding grid- ground potentials, along with facilitating the implementation of
connected DGUs, Fig. 2 shows the ground currents and their ground fault protection. However, the design and functions of
spectra for different solidly grounded DGUs during steady-state these grounding configurations are entirely based on the system
operation (no-fault conditions). Moreover, the ground poten- nominal frequency (i.e., no harmonic components) [25], [26].
SALEH: ON THE DESIGN AND CAPACITY OF A GROUNDING CONFIGURATION FOR GRID-CONNECTED DGUs 5369

The IEEE Standard 142 specifies the effective grounding of a TABLE I

PARAMETERS OF THE 2-kW DFIG-BASED WECS
generating unit in terms of the ratios between the zero-sequence
impedance of the grounding source and the positive-sequence
impedance of the grounded generating unit. These ratios are
defined as [4]
X0 R0
0≤ ≤3 0≤ ≤1 (1)
X1 X1
where X0 and R0 are the zero-sequence reactance and re-
sistance, respectively; and X1 is the positive-sequence reac-
tance. The relationships in (1) are applicable for conventional
synchronous generators, where the ratio X/R ranges between
20 and 40. However, for DGUs, the ratio X/R is very small The voltage rating for CG is selected as the system nominal
(only small reactances for the grid-side filters and DGU trans- line-to-line voltage VL [26].
formers). For example, in PV systems, X/R ranges between
0.015 and 0.062, and for converter-interfaced WECSs, X/R IV. E XPERIMENTAL S ETUPS
ranges between 0.2 and 0.7 [27].
The performance of the frequency-selective grounding con-
figuration was evaluated for three grid-connected DGUs that
B. Frequency-Selective Grounding Configuration for DGUs were as follows:
The main challenge in grounding DGUs is the harmonics • a 2.0-kW DFIG-based WECS;
generated by PECs that are employed to facilitate the operation • a 3.5-kW PMG-based WECS;
of DGUs. These harmonics are not allowed to flow through the • a 1.5-kW PV system.
PCC in compliance with the standards and grid codes. As a The experimental setups for these DGUs are described in the
result, harmonics continuously flow through the grounding sys- following subsections.
tem. This continuous flow of harmonics through the grounding
system can raise the ground potentials and/or adversely impact
A. The DFIG-Based WECS
the responses of ground fault protective devices [17], [25]–[27].
The challenges in grounding DGUs can be overcome by using The experimental setup of the DFIG-based WECS was con-
a frequency-selective path to the grounding system. Such a structed from a 3φ 2.0-kW 208-V four-pole 60-Hz wound-rotor
path has to appear as a short circuit for harmonic components, induction machine, which had its rotor windings fed from a
whereas it has to appear as a low impedance for the fundamental 1-kW 3φ voltage source (VS) six-pulse insulated-gate bipolar
component. These constraints on the grounding configuration transistor (IGBT) inverter (generator-side PEC). The generator-
can be met by a parallel R–C circuit. The elements of such an side PEC was supplied from a 1-kW 3φ diode rectifier (grid-
R–C circuit can be selected based on the system nominal line- side PEC) through a dc link capacitor of 300 μF. The grid-side
to-neutral voltage VP and the maximum current to flow to the PEC and stator windings of the induction machine were
grounding system (IG )max . The values of VP and (IG )max can connected to the secondary side of a 5-kVA 3φ 208/360-V
be used to specify the ohmic value of the resistance RG in the 60-Hz Δ–Y transformer (DGU transformer). The parameters
parallel R–C grounding circuit, that is [17] of the 2-kW DFIG-based WECS are listed in Table I.
VP The pulsewidth modulation (PWM) switching signals for
RG = . (2) the generator-side PEC had a switching frequency of 8 kHz
(IG )max
and were produced as outputs of the controller for the DFIG-
The complete selection of RG requires its power rating, which based WECS. The DFIG controller was designed as a current
can be specified as controller and was implemented using a dSPACE ds1104 DSP
board with a time step of Ts = 150 μs [17]. The test DFIG-
PRG = ((IG )max )2 RG . (3) based WECS was driven by a wind turbine emulator that was
realized by a 2.4-kW 120-V shunt dc motor, which had its
The desired parallel R−C grounding circuit is designed as armature supplied by a controlled rectifier to facilitate a variable
a low-pass filter to reduce ground potentials across RG due to speed operation. Fig. 4 shows a schematic for the experimental
steady-state harmonics flowing to the grounding system. This setup of the DFIG-based WECS.
function can be achieved by selecting the cutoff frequency fc
as the third harmonic component (i.e., fc = 3 × 4fs , with fs
being the system nominal frequency). The capacitance CG in B. The PMG-Based WECS
the parallel grounding R−C circuit can be selected by setting The experimental setup for the test PMG-based WECS was
the value of RG to be five times higher than the impedance of constructed from a 3.5-kW 230-V permanent-magnet machine,
CG at fc , that is [17] which had its stator windings connected to a 5-kW 3φ diode
5 5 rectifier (generator-side PEC). The output of the generator-
RG = =⇒ CG = . (4)
2π × 3 × fs × CG 2π × 3 × fs × RG side PEC output was fed to a 5-kW 3φ VS six-pulse IGBT
5370 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 51, NO. 6, NOVEMBER/DECEMBER 2015

Fig. 5. Schematic for the experimental setup of the 3.5-kW PMG-based

WECS. PC and QC denote the command active and reactive power,
respectively.

Fig. 4. Schematic for the experimental setup of the 2-kW DFIG-based WECS.
PC and QC denote the command active and reactive power, respectively.

TABLE II
PARAMETERS OF THE 3.5-kW PMG-BASED WECS

Fig. 6. Schematic for the experimental setup of the 1.5-kW PV system.

PC and QC denote the command active and reactive power, respectively.
The values of the LCL filter parameters are as follows: LI = 1.5 mH,
LG = 2.4 mH, CF = 20 μF, and RD = 1.47 Ω.

voltage of 80 V and a maximum output voltage of 180 V at

4 A. The outputs of the controllers for both dc buck–boost
PECs were switching signals to operate both converters [28].
These controllers were realized by using a ds1104 board,
which fed the generated switching signals to two optocoupler
driver circuits before being applied to the gates of the IBGT
switching elements of both dc buck–boost PECs. The output
inverter (the grid-side PEC) through a dc link capacitor of C = terminals of the grid-side PEC were connected through an LCL
300 μF. The outputs of the grid-side PEC were connected to a filter to the secondary side of a 5-kVA 3φ 208/360-V 60-Hz
longitudinal conversion loss (LCL) filter that had its terminals Δ–Y transformer. The switching signals for the grid-side PEC
connected to the secondary side of a 5-kVA 3φ 208/360-V were generated as the output of the PV controller, which
60-Hz Δ−Y transformer (DGU transformer). The PMG was was designed as a current controller. The PV controller was
driven by a wind turbine emulator that was realized by a 3.6-kW implemented by using a dSPACE ds1104 DSP board with a
240-V shunt dc motor, which had its armature fed from a time step of Ts = 150 μs [17]. Fig. 6 shows a schematic for
3φ controlled rectifier to provide a variable speed operation. the test 1.5-kW PV system.
Table II lists the parameters of the 3.5-kW PMG-based WECS. For the three test DGUs, the host grid was a 3φ 208-V 60-Hz
The switching pulses of the grid-side PEC were generated as 15-A power supply that was connected to the primary sides of
PWM with a switching frequency of 8 kHz and were produced the DGU transformers. In addition, the grounding system used
as the outputs of the PMG controller. This controller was for the test DGUs had a maximum current of (IG )max = 20 A.
designed as a current controller and was implemented using a
dSPACE ds1104 DSP board with a time step Ts = 150 μs [17].
A schematic for the test 3.5-kW PMG-based WECS is shown D. Grounding Configurations for the Test DGUs
in Fig. 5. The test DGUs were grid-connected through Δ–Y trans-
formers, which had their secondary sides connected to the
C. The PV System DGUs. The solid, low-resistance, open (used in different grid-
connected DGUs), and frequency-selective grounding config-
The experimental setup for the test PV system was com- urations were tested for each DGU. The ohmic value of the
posed of two dc power supplies, each of which fed a dc low-resistance grounding RGL was calculated so that the max-
buck–boost PEC. The outputs of the two dc buck–boost PECs imum ground fault current was less than 20 A, that is
were connected to the input of a 2-kW 3φ VS six-pulse IGBT
inverter (the grid-side PEC). The two dc buck–boost PECs were VP 208
RGL ≥ =⇒ RGL ≥ =⇒ RGL ≥ 10.4 Ω. (5)
operated at a switching frequency fc = 8 kHz, for an input (IF )max 20
SALEH: ON THE DESIGN AND CAPACITY OF A GROUNDING CONFIGURATION FOR GRID-CONNECTED DGUs 5371

The value of RGL was selected to be 13 Ω. For the frequency- TABLE III
S UMMARY OF R ESULTS FOR T ESTING THE 2-kW DFIG-BASED WECS,
selective grounding configuration, the value of RG was 3.5-kW PMG-BASED WECS, AND 1.5-kW PV S YSTEM
determined as W ITH D IFFERENT G ROUNDING C ONFIGURATIONS
VP 208
RG = = = 8.32 Ω. (6)
(IG )max 25
The power rating for RG was selected as PRG = 6 kW. Finally,
the value of CG was calculated using (4) as
5 5
CG = = = 532 μF. (7)
2π × 3 × fs × RG 360π × 8.32
The voltage rating for CG was selected to be 400 V.

V. E XPERIMENTAL R ESULTS
The frequency-selective grounding configuration was ex-
perimentally tested for the 2-kW DFIG-based and 3.5-kW
PMG-based WECSs and the 1.5-kW PV system, for steady-
state changes and ground faults. The improvements in the
performance of test DGUs were highlighted by comparing
the frequency-selective grounding configuration with the solid,
low-resistance, and open grounding configurations under simi-
lar steady-state and ground fault conditions. Several tests were
carried out for the three DGUs when experiencing steady-state
transients and fault conditions. The following three cases are
presented in details, whereas Table III summarizes the results
obtained from all tests:
1) step changes in PC and QC ;
2) line-to-ground fault on the ac side of the grid-side PEC;
3) ground fault across the dc link.
The ground fault protection was established by an inverse
definite minimum time overcurrent (IDMTOC) relay, with an
8-A pick-up current and a 0.5-s time dial [17].
Case I—Step Changes in PC and QC : This test was per-
formed to investigate ground potentials and ground currents
during nonfault transients in different grid-connected DGUs.
Each of the three DGUs was tested separately, where it was
operated with PC = 50% of the rated power at a power factor
of PF = 0.93 lagging. Sudden step changes in PC and QC were
created and lasted for 1.2 s, during which PC was increased
by 40% and QC was increased by 25% for each tested DGU.
The responses of IDMTOC relay (Trip), ground potentials, and
ground currents for tested ground configurations for the 2-kW
DFIG-based and 3.5-kW PMG-based WECSs and the 1.5-kW
PV system are shown in Fig. 7.
The results in Fig. 7 demonstrate that sudden step changes
in PC and QC did not significantly affect the ground potentials Case II—Line-to-Ground Fault on the AC Side of the Grid-
VG and currents IG in all grounding configurations for each test Side PEC: This test was performed to demonstrate the ability
DGU. As VG and IG were not significantly affected by the step of the frequency-selective grounding configuration to limit
changes in PC and QC , the trip signals generated by the tested ground fault currents without affecting the function of a ground
relay did not change their status, indicating nonfault conditions. fault protective relay. The line-to-ground fault was created
It should be noted that ground currents for the DFIG-based by activating a solid-state switch that connected phase C to
WECS were higher than those for the PMG-based WECS and the ground on the ac side of the grid-side PEC. This fault
the PV system due to the 3φ ac–dc PEC, which was used as was performed for each DGU when grounded through the
the grid-side PEC in the test DFIG-based WECS. The 3φ ac–dc solid, frequency-selective, low-resistance, and open grounding
PEC created more current harmonics than those generated by configurations. During this test, each DGU was operated with
the grid-side PECs used in the PMG-based WECS and the PC = 70% of its rated power with PF = 0.88 lagging. The re-
PV system. sponses of the IDMTOC relay (ground fault protective device),
5372 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 51, NO. 6, NOVEMBER/DECEMBER 2015

Fig. 7. Grounding configurations of grid-connected DGUs during step changes in PC (PC = PC → 1.4PC → PC ) and QC (QC = QC → 1.25QC →
QC ). Ground potential VG , ground current IG , and trip signal generated by the IDMTOC relay for the 2-kW DFIG-based WECS: (a1) solid grounding,
(a2) frequency-selective grounding (RG in parallel with CG ), (a3) low-resistance grounding with RGL = 13 Ω, and (a4) open grounding. Ground potential
VG , ground current IG , and trip signal generated by the IDMTOC relay for the 3.5-kW PMG-based WECS: (b1) solid grounding, (b2) frequency-selective
grounding, (b3) low-resistance grounding with RGL = 13 Ω, and (b4) open grounding. Ground potential VG , ground current IG , and trip signal generated by
the IDMTOC relay for the 1.5-kW PV system: (c1) solid grounding, (c2) frequency-selective grounding, (c3) low-resistance grounding with RGL = 13 Ω, and
(c4) open grounding. VG scale: 10 V/Div.; IG scale: 5 A/Div.; trip signal scale: 10 V/Div.; time scale: 0.2 s/Div.

ground potentials, and ground currents for this test are shown in almost four cycles for the solid and frequency-selective
in Fig. 8. grounding configurations, as shown in Fig. 8(c1) and (c2).
The results for the line-to-ground fault on the ac side of However, the tested fault was cleared in almost 5.6 cycles
the grid-side PEC show that the IDMTOC relay cleared the for the low-resistance grounding configuration, as shown in
fault with different response times, which varied depending on Fig. 8(c3).
the grounding configuration and the type of the DGU. For the The results of the phase-C-to-ground fault demonstrated the
DFIG-based WECS (the grid-side PEC is a 3φ ac–dc PEC), ability of the frequency-selective grounding to reduce ground
the ground fault was cleared in almost 3 cycles for the solid fault currents, while imposing negligible effects on responses of
and frequency-selective grounding configurations, as shown the ground fault protective relay. This feature could be noticed
in Fig. 8(a1) and (a2). However, the line-to-ground fault was as the trip times observed for the solid grounding configuration
cleared in almost 3.5 cycles for the low-resistance grounding were very close to those observed for the frequency-selective
configuration, as shown in Fig. 8(a3). For the open grounding grounding configuration. The type of the PEC close to the fault
in all test DGUs, the phase-C-to-ground fault on the ac side of location also affected the ground fault current, and thus trip
the grid-side PEC was not detected, and the ground protective times of the ground protective relay.
relay did not change the status of its trip signal, as shown in Case III—Ground Fault Across the DC Link: The ground
Fig. 8(a4), (b4), and (c4). For the PMG-based WECS (the grid- fault across the dc link was tested to demonstrate the ability
side PEC is a 3φ dc–ac PEC), the phase-C-to-ground fault of the frequency-selective grounding configuration to reduce
was cleared in almost 3.6 cycles for the solid and frequency- ground fault currents, when dominated by dc components. This
selective grounding configurations, as shown in Fig. 8(b1) and test was created by activating a solid-state switch that connected
(b2). However, the ground fault was cleared in almost 5 cycles the positive terminal of the dc link capacitor to ground. The
for the low-resistance grounding configuration, as shown in tested fault was conducted for each DGU when grounded
Fig. 8(b3). Finally, for the PV system (the grid-side PEC is through the solid, frequency-selective, low-resistance, and open
a 3φ dc–ac PEC), the phase-C-to-ground fault was cleared grounding configurations. The ground fault across the dc link
SALEH: ON THE DESIGN AND CAPACITY OF A GROUNDING CONFIGURATION FOR GRID-CONNECTED DGUs 5373

Fig. 8. Grounding configurations of grid-connected DGUs during phase-C-to-ground fault on the ac side of the grid-side PEC. Ground potential VG , ground
current IG , and trip signal generated by the IDMTOC relay for the 2-kW DFIG-based WECS: (a1) solid grounding, (a2) frequency-selective grounding
(RG in parallel with CG ), (a3) low-resistance grounding with RGL = 13 Ω, and (a4) open grounding. Ground potential VG , ground current IG , and trip signal
generated by the IDMTOC relay for the 3.5-kW PMG-based WECS: (b1) solid grounding, (b2) frequency-selective grounding, (b3) low-resistance grounding with
RGL = 13 Ω, and (b4) open grounding. Ground potential VG , ground current IG , and trip signal generated by the IDMTOC relay for the 1.5-kW PV system:
(c1) solid grounding, (c2) frequency-selective grounding, (c3) low-resistance grounding with RGL = 13 Ω, and (c4) open grounding. VG scale: 30 V/Div.;
IG scale: 10 A/Div.; trip signal scale: 10 V/Div.; time scale: 20 ms /Div.

was created when each DGU was operated with PC = 85% of its and (c2)], whereas it was cleared in almost three cycles for the
rated power with PF = 0.80 lagging. Fig. 9 shows the responses low-resistance grounding configuration [see Fig. 9(c3)].
of the IDMTOC relay (ground fault protective device), ground The results obtained from the ground fault across the dc link
potentials, and ground currents for this test case. showed agreement with the results obtained from the phase-
Fig. 9 shows that the IDMTOC relay cleared the ground C-to-ground fault on the ac side of the grid-side PEC. In both
fault across the dc link fault with different response times test results, the frequency-selective grounding configuration
that depended on the grounding configuration and the type of managed to reduce the ground fault currents without effecting
DGU. For the DFIG-based WECS, the fault was cleared in the responses of the ground fault protective relay. Test results
almost 2 cycles for the solid and frequency-selective grounding also confirmed the effects of the grounding configuration and
configurations [see Fig. 9(a1) and (a2)], whereas the fault was the type of PEC, close to the fault location, on the ground fault
cleared in almost 2.5 cycles for the low-resistance grounding currents and trip times of the IDMTOC relay.
configuration [see Fig. 9(a3)]. When the grounding configu- Experimental results obtained from testing the 2-kW DFIG-
ration was set as an open grounding, the ground fault across based WECS, the 3.5-kW PMG-based WECS, and the 1.5-kW
the dc link was not detected for all test DGUs, and trip signals PV system are summarized in Table III. The performance of
remained unchanged, as shown in Fig. 9(a4), (b4), and (c4). For each DGU was evaluated in terms of the ground potentials
the PMG-based WECS, the ground fault across the dc link was (VG )peak. , ground currents (IG )peak. , and trip times tTrip . For
cleared in almost 2 cycles for the solid and frequency-selective each set of tests, values of (VG )peak. were used to determine an
grounding configurations [see Fig. 9(b1) and (b2)], and it was average value (ṼG )peak. for that set of tests. Furthermore, val-
cleared in almost 3 cycles for the low-resistance grounding ues of (IG )peak. obtained during each set of tests were used to
configuration [see Fig. 9(b3)]. In case of the PV system, the calculate an average value (I˜G )peak. for that set of tests. Finally,
tested fault was cleared in almost 2 cycles for the solid and the values of tTrip for each set of tests were used to determine an
frequency-selective grounding configurations [see Fig. 9(c1) average value t̃Trip. . It should be noted that the open grounding
5374 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 51, NO. 6, NOVEMBER/DECEMBER 2015

Fig. 9. Grounding configurations of grid-connected DGUs during a ground fault across the dc link. Ground potential VG , ground current IG , and trip signal
generated by the IDMTOC relay for the 2-kW DFIG-based WECS: (a1) solid grounding, (a2) frequency-selective grounding (RG in parallel with CG ),
(a3) low-resistance grounding with RGL = 13 Ω, and (a4) open grounding. Ground potential VG , ground current IG , and trip signal generated by the IDMTOC
relay for the 3.5-kW PMG-based WECS: (b1) solid grounding, (b2) frequency-selective grounding, (b3) low-resistance grounding with RGL = 13 Ω, and
(b4) open grounding. Ground potential VG , ground current IG , and trip signal generated by the IDMTOC relay for the 1.5-kW PV system: (c1) solid grounding,
(c2) frequency-selective grounding, (c3) low-resistance grounding with RGL = 13 Ω, and (c4) open grounding. VG scale: 30 V/Div.; IG scale: 10 A/Div.;
trip signal scale: 10 V/Div.; time scale: 20 ms/Div.

configuration was not included in Table III as IG = 0 A and ground potentials and currents. The solid grounding configu-
tTrip = ∞ in all tests for each DGU. The experimental testing ration produced the highest ground currents with no ground
of the three grid-connected DGUs included the following sets potentials during fault and nonfault conditions. As a result,
of tests. ground faults were cleared in the shortest times (relative to other
grounding configurations). The frequency-selective grounding
1) Nonfault transients: This set of tests included step
configuration produced grounding currents that were lower than
changes in PC and QC , step changes in the voltage at
those produced by the solid grounding configuration, whereas
PCC, and variations in the generated power by each DGU.
it produced ground potentials lower than those produced by
2) Ground faults in DGU: This set of tests included faults
the low-resistance grounding configuration. Ground faults for
within each DGU. Ground faults in the stator and rotor
the frequency-selective grounding configuration were cleared
windings were carried out for the DFIG, whereas ground
faster than those for the low-resistance grounding. The low-
faults were conducted for the PMG. For the PV system,
resistance grounding configuration yielded the highest ground
ground faults at the input side of the dc–dc PECs were
potential and the lowest ground current during fault and non-
performed.
fault conditions. The low ground currents for the low-resistance
3) Ground faults in PECs: This set of tests included ground
grounding configuration caused the ground faults to be cleared
faults in PECs used in each DGU. Ground faults in ac–dc
in the longest times (relative to other grounding configurations).
and dc–ac PECs were tested for the DFIG and the PMG,
The data in Table III, along with test results, support the
whereas ground faults in dc–dc and dc–ac PECs were
use of the frequency-selective grounding configuration in grid-
carried out for the PV system.
connected DGUs. This grounding configuration can offer re-
4) Ground faults at PCC: This set of tests included ground
ducing ground currents and limiting ground potentials with
faults for all phases at the PCC.
negligible effects on the responses of ground protective relays.
The summary of performance in Table III shows that dif- Finally, the data in Table III show that ground potentials and
ferent grounding configurations produced different values for currents during ground faults can be significantly influenced by
SALEH: ON THE DESIGN AND CAPACITY OF A GROUNDING CONFIGURATION FOR GRID-CONNECTED DGUs 5375

the type of PECs employed in a DGU. Such influences can applications,” IEEE Trans. Ind. Appl., vol. 49, no. 6, pp. 2825–2834,
be seen from the differences between ground potentials and Nov./Dec. 2013.
[12] National Electrical Safety Code, IEEE Std. C2-2012, Aug. 2012.
currents observed for DFIG and both PMG and PV systems. [13] NFPA 70: National Electrical Code, Nat. Fire Protect. Assoc., Quincy,
The main sources for such influences include the operation, MA, USA, 2011.
control, and input/output dominant harmonic components. [14] R. F. Arritt and R. C. Dugan, “Distribution system analysis and the fu-
ture smart grid,” IEEE Trans. Ind. Appl., vol. 47, no. 6, pp. 2343–2350,
Nov./Dec. 2011.
[15] E. Muljadi, N. Samaan, V. Gevorgian, J. Li, and S. Pasupulati, “Different
VI. C ONCLUSION factors affecting short circuit behavior of a wind power plant,” IEEE
Trans. Ind. Appl., vol. 49, no. 1, pp. 284–292, Jan./Feb. 2013.
This paper has presented the design and application of a [16] F. Freschi, M. Mitolo, and M. Tartaglia, “An effective semianalytical
method for simulating grounding grids,” IEEE Trans. Ind. Appl., vol. 49,
frequency-selective grounding configuration for grid-connected no. 1, pp. 256–263, Jan./Feb. 2013.
DGUs. The main objective of designing the frequency-selective [17] S. A. Saleh et al., “Impacts of grounding configurations on responses of
grounding configuration is to overcome the problems associated ground protective relays for DFIG-Based WECSs—Part I: Solid ground
faults,” in Conf. Rec. IEEE IAS 50th Annu. Meeting I&CPS, Fort Worth,
with the harmonic components generated by PECs in DGUs TX, USA, May 2014, pp. 1–8.
and continuously flow through the grounding system. The [18] J. Matas, M. Castilla, J. miret, L. D. V. Garcia, and R. Guzman, “An
proposed grounding configuration is designed as a parallel R–C adaptive prefiltering method to improve the speed/accuracy tradeoff of
voltage sequence detection methods under adverse grid conditions,” IEEE
circuit to act as a short-circuit path for harmonic components Trans. Ind. Electron., vol. 61, no. 5, pp. 2139–2151, May 2014.
(generated by PECs employed in DGUs), and thus eliminate [19] B. Badrzadeh and M. Gupta, “Practical experiences and mitigation meth-
the ground potentials due to these harmonic components. The ods of harmonics in wind power plants,” IEEE Trans. Ind. Appl., vol. 49,
no. 5, pp. 2279–2289, Sep./Oct. 2013.
parallel R–C circuit is also designed to act as a low-resistance [20] F. Blaabjerg, M. Liserre, and K. Ma, “Power electronics converters for
path for fundamental frequency components, and hence limit wind turbine systems,” IEEE Trans. Ind. Appl., vol. 48, no. 2, pp. 708–
the ground fault currents. The frequency-selective grounding 719, Mar./Apr. 2012.
[21] S. Liang, Q. Hu, and W. J. Lee, “A survey of harmonic emissions of a
configuration has been designed and tested for three different commercially operated wind farm,” IEEE Trans. Ind. Appl., vol. 48, no. 3,
DGUs during steady-state and fault conditions. Test results pp. 1115–1123, May/Jun. 2012.
show that the developed grounding configuration is able to [22] C. J. Mozina, “Impact of smart grids and green power generation on
distribution systems,” IEEE Trans. Ind. Appl., vol. 49, no. 3, pp. 1079–
reduce ground fault currents and limit ground potentials, while 1090, May/Jun. 2013.
imposing minor impacts on the responses of ground fault pro- [23] L. Wei, Z. Liu, and G. L. Skibinski, “Investigation of voltage stresses
tective devices, when compared with the solid, low-resistance, inside adjustable-speed drives,” IEEE Trans. Ind. Appl., vol. 49, no. 1,
pp. 100–108, Jan./Feb. 2013.
and open grounding configurations that are commonly used [24] A. Ackerman, P. K. Sen, and C. Oertli, “Designing safe and reliable
in DGUs. Obtained results from different DGUs have demon- grounding in ac substations with poor soil resistivity: An interpretation
strated consistent features of the frequency-selective grounding of IEEE Std. 80,” IEEE Trans. Ind. Appl., vol. 49, no. 4, pp. 1883–1889,
Jul./Aug. 2013.
configuration. Test and comparison results support the employ- [25] AIEE Committee Report on the Application Guide for the Grounding of
ment of the frequency-selective grounding in different DGUs to Synchronous Generator Systems, Jun. 1953.
achieve safe, reliable, and stable operation. [26] IEEE Guide for the Application of Neutral Grounding in Electrical Utility
Systems Part I-Introduction, ANSI/IEEE C67.92-987, 1987.
[27] “Neutral connections and effective grounding: White Paper Report for
Solar Energy,” Adv. Energy Ind., Inc., Raleigh, NC, USA, 2013.
[28] W. Libo, Z. Zhengming, and L. Jianzheng, “A single-stage three-phase
R EFERENCES grid-connected photovoltaic system with modified MPPT method and
[1] IEEE Guide for Safety in AC Substation Grounding, IEEE Std. 80-2000, reactive power compensation,” IEEE Trans. Energy Convers., vol. 22,
2000. no. 4, pp. 881–886, Dec. 2007.
[2] IEEE Application Guide for IEEE Standard for Interconnecting Dis-
tributed Resources With Electric Power Systems, IEEE Std. 1547.2, 2008.
[3] Photovoltaic Systems-Characteristics of the Utility Interface, IEC Std.
61727, 2004. S. A. Saleh (S’03–M’06–SM’12) received the
[4] Recommended Practice for Grounding of Industrial and Commercial B.Sc. degree from Bir Ziet University, West Bank,
Power Systems, IEEE Std. 142-2007, 2007. Palestine, in 1996, and the M.Eng. and Ph.D. degrees
[5] R. Hoerauf, “Considerations in wind farm grounding designs,” IEEE from Memorial University of Newfoundland, St.
Trans. Ind. Appl., vol. 50, no. 2, pp. 1348–1355, Mar./Apr. 2014. John’s, NL, Canada, in 2003 and 2007, respectively,
[6] J. C. Das and E. Perich, “13.8-kV selective high-resistance grounding all in electrical engineering. He received a schol-
system for a geothermal generating plant—A case study,” IEEE Trans. arship from the Natural Sciences and Engineering
Ind. Appl., vol. 49, no. 3, pp. 1234–1243, May/Jun. 2013. Research Council of Canada (NSERC) for his Ph.D.
[7] D. Shipp et al., “Switching transient analysis and specifications for studies.
practical hybrid high-resistance grounded generator applications—An He was with the Palestinian Technical College,
IEEE/IAS Working Group Report #2,” IEEE Trans. Ind. Appl., vol. 48, West Bank, as an Electrical Engineer for two years
no. 1, pp. 236–244, Jan./Feb. 2012. and an Instructor and a Program Coordinator for three years. In 2007, he joined
[8] C. J. Mozina, “Why upgrade the protection and grounding of generators the Marine Institute, Memorial University of Newfoundland, as a Faculty Mem-
at petroleum and chemical plants,” IEEE Trans. Ind. Appl., vol. 50, no. 1, ber and a Researcher until 2011. He is currently an Associate Professor with
pp. 155–162, Jan./Feb. 2014. the Department of Electrical and Computer Engineering, University of New
[9] L. J. Powell, “An industrial view of utility cogeneration protection re- Brunswick, Fredericton, NB, Canada. His research interests include wavelet
quirements,” IEEE Trans. Ind. Appl., vol. IA-24, no. 1, pp. 75–81, analysis, power system protection, microgrid, power electronics, modulation
Jan./Feb. 1988. techniques, renewable energy systems, and digital signal processing and its
[10] D. Paul, P. E. Sutherland, and S. A. R. Panetta, “A novel method of applications in power systems and power electronic converters. His research
measuring inherent power system charging current,” IEEE Trans. Ind. work is supported by the NSERC and New Brunswick Innovation Foundation
Appl., vol. 47, no. 6, pp. 2330–2342, Nov./Dec. 2012. Strategic Projects.
[11] B. Kroposki, P. K. Sen, and K. Malmedal, “Selection of distribution Dr. Saleh is a Registered Professional Engineer in the Provinces of
feeders for implementing distributed generation and renewable energy Newfoundland and Labrador and New Brunswick, Canada.