Sistema de Proteccion
Sistema de Proteccion
Sistema de Proteccion
LUT ENERGY
ELECTRICAL ENGINEERING
MASTER’S THESIS
Sergei Muzalev
Relay protection in active distribution networks
Master’s thesis
2016
109 pages, 49 pictures, 5 tables and 1 appendix
This paper proposes solutions to the emerging need to ensure correct operation
of the equipment in question with least possible grid refinements, distinctively
for every type of distributed generation technology achieved its technical maturi-
ty to date and network’s protection. New generating technologies are equivalent-
ed from the perspective of representation in calculation of initial steady-state
short-circuit current used to dimension current-sensing relay protection, and
widely adopted short-circuit calculation practices, as IEC 60909 and VDE 0102.
The phenomenon of unintentional islanding, influencing auto-reclosing, is ad-
dressed, and protection schemes used to eliminate an sustained island are listed
and characterized by reliability and implementation related factors, whereas also
forming a crucial aspect of realization of the proposed protection operation re-
lieving measures.
Table of content
1 Introduction ................................................................................................... 8
2 Scope ............................................................................................................. 10
2.1. Generating technologies span ........................................................................10
2.2. Protection approaches and DG penetration levels ......................................12
References......................................................................................................... 103
Appendices:
B flux density
E,e emf
f frequency
I,i current
L inductance
n rotational speed
p pole number
P active power
Q reactive power
R,r resistance
s slip
S apparent power
T time constant
U,u voltage
w angular frequency
X,x reactance
Z,z impedance
σ leakage
Φ flux
0 zero-sequence value
1 positive-sequence value
2 negative-sequence value
0− right before the change occurrence
0 the moment of change occurrence
0+ right after the change occurrence
Acknowledgments
I want to express my sincere gratitude to the Head of Relay protection and auto-
mation of electrical power systems faculty – Moscow Power Engineering Insti-
tute, Moscow – professor Alexander Voloshin, who inspired me to evolve in the
cutting edge field of contemporary electrical power engineering as distributed
generation, given a start to far-beyond-thesis-boundaries scientific research.
Moreover, to thank the Head of Energy Technology faculty, professor Jarmo
Partanen for trailblazing through highly dispersed issues appearing when distrib-
uted generation collide with grid’s relay protection and automation.
Not least, I am thankful to all folks I met throughout 2014-2015 studies at LUT,
forced me to balance between studying and having a blast, who, in the long run,
made those days a history.
Eventually, I want to give special thanks to Iulia, Andrey Coffee, Dmitry S.,
Pavel, Dmitry Sk., Milan, Anil, Onur, Alexander for generating this special Lap-
pa atmosphere.
1 Introduction
Back to early 1900s the contemporary bulk electrical power system was called
“the largest, most complex machine ever devised by man” by Charles Steinmetz,
German-born American electrical engineer brought a significant value to the
development of alternating current. The systems of those times incorporated ex-
ceptionally 3-phase AC grids of constant voltage level with centralized AC syn-
chronous generation running at constant frequency and feeding loads through
relatively long-distance transmission and distribution lines on simultaneity of
generation and consumption basis (Kundur et al. 1994). The aforementioned
characteristic pattern, entrenched over about a century, is no longer ubiquitously
implementable nowadays when girds are expected to integrate both centralized
and dispersed generation, based on synchronously, asynchronously rotating ma-
chines; moreover, sources of direct energy conversion (DEC) of various extent
of operation mode heterogeneity and control possibilities. The absolutely distin-
guished niche of carrier as DC current becomes inherent in every part of the sys-
tem in a form of generation, transformation, transmission, distribution, and con-
sumption appliances. Necessity of production and consumption balance in every
time instance is provided besides by active and reactive power flows all-type
regulation, but also by means of energy storage.
tion installed capacity gaining its share because of climate change concerns.
(Илюшин 2014).
Distributed energy resources (DER) cause an impact toward relay protection and
automation functioning as they change overall power infeed and infeed coeffi-
cients in different parts of a distribution grid, whereas, conventionally organized
grid protection schemes are usually based on current sensors. In general, the ver-
satile influence may be expressed as follows:
i. Reverse power flow to a fault incepted within upstream zone (regarding
DG junction point);
ii. False tripping of a healthy feeder’s relay due to short-circuit current con-
tribution from DG connected to it;
iii. False tripping because of inrush current and transients induced by DG;
iv. Unintentional islanding – a part of the distribution network is isolated
from utility grid with DG/-s connected within;
v. Blinding of relay protection which can be caused in several instances,
such as high-power-capacity DG installation between two adjacent feed-
ers with sensitivity decreasing of the upper one (upstream feeder) and is-
land sustainable operation;
vi. Disturbance of AR due to contribution to arcing and preventing self-
extinguishing while isolated from the mains;
vii. Out-of-phase AR causing damage to DG units;
viii. Loss of protection device coordination.
As relay protection and vast majority of automation are meant to ensure reliable
power supply for end-customers (eliminate or reduce customer curtailment), re-
quired power quality at points of common coupling (PCCs), and useful lifetime
of primary equipment, the correct operation of both is a high priority issue.
10
2 Scope
2.1. Generating technologies span
1
Thermionic, thermoelectric converters, MHD generators.
12
− Gas turbine1.
2.1.5.7 (Small) conventional hydraulic generation:
− Hydraulic turbine.
2.1.5.8 Wave, tidal, ocean heat-based generation:
− Hydraulic turbine2;
− Air-driven turbine3;
− Steam turbine4;
− DEC.
2.1.5.9 Low grade thermal energy based generation:
− Propeller5.
2.1.5.10 Solid domestic waste based generation;
2.1.5.11 Technological industry by-products based generation.
Conventional types of generation are extensively described in special literature
from the perspective of short-circuit current calculations, so only new types
achieved its technical maturity will be equivalented below.
1
Fuel cell hybrid (combined) system with gas turbine/energy storage system.
2
Except the conventional installations, this type of prime mover is implemented in Fetkovich’s
ocean thermal converter.
3
Pneumatic wave power plant’s prime mover.
4
Single-loop open/Claude cycle ocean thermal converter, double-loop thermodynamic Rankin
cycle with intermediary working fluid.
5
Updraft tower power station’s prime mover.
6
There is also multi-microgrid concept, which implies mutual control and coordination of elec-
trically connected, but separate LV microgrids and upline DERs.
13
1
The need in AI protection implementation will be described below.
14
Grids with high-penetration level are taken for analysis in this paper. As DG
units are not tripped with smallest time delay among grid protections, the impact
they cause toward protection functioning has to be properly mitigated if needed.
passing to the generator and, in case of inverter utilizing (will be specified be-
low), vice versa.
The first type of utility-size wind turbine is a fixed speed turbine with a squirrel-
cage induction machine (Type 1 in wind-related applications), which generates
active power only in a case of rotation with speed higher than so-called synchro-
nous speed of electromagnetic grid-side (stator) flux, which determines the nega-
n1 − n
tive slip s = value in normal operating mode. A typical real power versus
n1
slip characteristics profile is shown in Fig. 3.1.
Fig. 3.1. The relation between slip and active power output of Type 1 WTG.
(Muljadi et al. 2010).
By its nature, an induction generator consumes reactive power both in the motor-
ing and generating modes with significant surge of the latter when the power
16
output increases. This phenomenon determines the view of the phasor diagram in
pre-fault regime, which reflects the underexciting mode of the machine.
The connection diagram of Type 1 WTG is shown in Fig. 3.2. The capacitor
bank is implemented in order to regulate the power factor at the unit terminals.
Due to direct interconnection with the grid the voltage drop at the generator ter-
minals during short-circuit causes electromagnetic transients in the stator and
rotor windings, while the latter is rotating almost with the same speed (here and
in further evaluations it is assumed that inertia of rotating masses is high enough
to somehow change the speed during electromagnetic transients).
On the strength of magnetic flux linkage invariability law, a portion of the stator
emf formed by the magnetic flux linkage of the rotor and stator windings (so-
called d-axis flux in Park’s model) stays unchangeable in the first moment of
low-voltage inception, whence the next equations (3.1), (3.2) may be concluded
analogically to underexcited synchronous machine low-voltage mathematical
representation (Крючков 2008).
E ''(0) = (U(0) − x ''d I (0) sin ϕ (0) )2 + (x ''d I (0) cos ϕ (0) )2 (3.1)
1
x "= x = (3.2)
*d *2 I lock.rotor
*
17
Where I -locked rotor current of the induction machine [pu], x ''d - subtran-
* lock.rotor
As this machine type has no independent excitation, the resulting magnetic flux
collapses with the delay determined by time constants (3.4), (3.5) based on the
equivalent circuit of asynchronous machine (Fig. 3.3), where the value of the
former (3.5) influences the rate of decay of the DC component, whereas, (3.6)
determines the damping of periodic one.
Lrσ Lm
L' s = Lsσ + (3.3)
Lrσ + Lm
L' s
T 's = (3.4)
Rs
Lsσ Lm
L' r = Lrσ + (3.5)
Lsσ + Lm
L' r
T 'r = (3.6)
Rr
where the second one is slip dependable, leakage reactances Lsσ and Lrσ , re-
In the light of the preceding, the SCC does not reach any steady-state value and
exponentially declines in case of no reactive power absorption – three-phase
short circuit at he machine terminals. It allows stating further protection-related
DG impact, as
ix. Variations of infeed coefficients in time domain while in fault regime
due inability of some sources to sustain short-circuit current (SCC).
So-called Type 3 WTG1 comprises variable speed turbine bound by a shaft and
gearbox with doubly fed induction generator (DFIG). The latter is a wound rotor
machine providing full modulation capability of the rotor current by decoupled
junction of its three phase rotor windings through AC-DC-AC converter with the
electrical grid, whereas the stator ones are coupled directly (Fig. 3.4.).
ws wr
= + wm (3.7)
ps pr
1
The initial consideration of the 3rd type of WEC instead of the consistent 2nd one here
is due to usage of equations derived in the following for the 2nd machine type equivalen-
tizing.
19
field angular frequency, rotor field angular frequency, stator and rotor pole num-
bers, correspondingly (Fault current contributions from wind plants 2015).
Generally, the rotor speed of the wind turbine is allowed to vary within a slip
range of +/- 0.3 (+/- 0.4). Consequently, rotor side power converter can be sized
to approximately 30% (40%) of the machine rated power, which forms the eco-
nomical feasibility of the type (Gevorgian et al. 2010), but induces necessity to
protect electronic power switches against overload likely to appear during severe
faults in the grid. Thereby, distinctive protection approaches, reflecting the
crowbar functionality, have been elaborated for this particular purpose, including
(Fault current contributions from wind plants 2015):
1. Shortening device installed between rotor winding’s terminals and rotor-side
converter, which is used to divert high-value currents from rotor windings
bypass the AC-DC-AC converter. Depending on the manufacturer, it may in-
corporate some impedance in shorting path, converting the DFIG to the in-
duction generator with relatively high rotor winding’s impedance (in compar-
ison to SCIG) (Fig. 3.4);
2. Shortening the rotor windings by switching off rotor-side converter, which
leads to machine’s representation identical to SCIG.
3. А chopper circuit on the converter’s DC bus limiting the voltage rise by dis-
sipating the excessive energy, remaining the excitation current regulation cir-
cuit in operation.
The crowbar switching sensor’s signals, threshold values and time durations of
staying in engaged position vary widely from one ACS to another, contributing
to diversity of Type 3 machine representations and making impossible elabora-
tion of the generic model. In addition to the preceding, the distance from the
faulted point to the machine come into play significantly here by changing the
representation of the machine throughout SCC routine.
20
The inherent characteristics of the DFIG, incorporating fast response of the con-
verters and laminated rotor, result in very short rotor flux time constants similar
to STATCOM, which leads to probability of observance the considerable leeway
of current operating point right after fault detection by relay protection due to
reactive power injection as a LVRT requirement and/or in attempts of control
system to sustain the predefined active power input.
Non-engaged crowbar
Symmetrical faults
The general list of controllers consists of a pitch angle controller, rotor-side and
grid-side converter’s controllers. The operation of the former has no impact to-
ward SCC infeed value due to its use in order to keep turbine rotation speed
within allowed limits, whereas, assuming that voltage of DC link stays constant
in the moment of fault inception leads to neglection of the latter one.
The rotor-side controller generally operates in a stator-flux oriented reference
frame, which leads to possibility of fully decoupled regulation of active and reac-
tive power. The controller regulates the rotor current through altering the AC
voltage of the converter with respect to the present power references (Xiong et
al. 2011). This is the focal decisive aspect having the essential impact on the way
the machine can be represented as a part of SCC calculation scheme.
Considering that the standard wound rotor induction machine is used under
DFIG abbreviation, the Park model consistent with asynchronously rotating ma-
chine is used as fundamental base for all further reasoning.
The equations are written for the directions of the currents reflecting motoring
mode of the electrical drive: stator windings – from the terminals, rotor – from
the PWM converter.
21
dψ ds
U ds = rsids + − w0ψ qs (3.8)
dt
dψ qs
U qs = rsiqs + + w0ψ ds (3.9)
dt
dψ dr
U dr = rrids + − (w0 − w)ψ qr (3.10)
dt
dψ qs
U qr = rr iqr + + (w0 − w)ψ ds (3.11)
dt
ψ ds = Lsids + Lmidr (3.12)
All the equations above are written based on the assumption that magnetic satu-
ration of the stator and rotor packs is neglected and rotation speed of the latter
stays unchangeable during transients (Крючков 2008), likewise rotor field fre-
quency. The preceding is justified due to relatively low-speed proceeding of the
electromechanical transients (in a light of rotating mass inertia) in comparison
with electromagnetic ones.
d(ψ qr + jψ dr )
Ur = (U qr + jU dr ) = rr (iqr + jidr ) + + (w0 − w)(ψ dr − jψ qr ) (3.16)
dt
dψ!
U! s = rs I!s + s + jwsψ! s (3.17)
dt
dψ!
Ur = rr I!r + r + j(ws − wr )ψ! r (3.18)
dt
ψ! s = Ls I!s + Lm I!r (3.19)
Fig. 3.5. Equivalent scheme of induction machine revealing the current directions submitted as
Park model basis.
Substituting (3.19) into (3.17) and determining ψ! s lead to the next equation.
Rs dψ! R
( + jws )ψ! s + s − (U! s + s Lm I!r ) = 0
Ls dt Ls
R
U! s + s Lm I!r
R Ls
( s + jws )ψ! s ( p) + pψ! s ( p) − − ψ! s (0) = 0
Ls p
R
U! s + s Lm I!r
Ls 1
ψ! s ( p) = − ψ! s (0) =0
Rs Rs
p( p + ( + jws )) p + ( + jws )
Ls Ls .
ψ! s = {inver.Laplace.tr} =
R
U! s + s Lm I!r R
−( s + jws )t
R
−( s + jws )t
Ls
= (1− e Ls ) + ψ! s (0)e Ls
Rs
+ jws
Ls
R R
U! s + s Lm I!r U! s + s Lm I!r −( Rs + jw )t
Ls Ls s
ψ! s = +[ψ! s (0) − ]e Ls (3.21)
Rs Rs
+ jws + jws
Ls Ls
R
U! s (0 − ) + s Lm I!r (0 − )
Ls
ψ! s (0 − ) =
Rs
+ jws
Ls
R R
U! s (0 + ) + s Lm I!r (0 + ) U! s (0 + ) + s Lm I!r (0 + ) −( Rs + jw )t
Ls Ls s
ψ! s (0 + ) = +[ψ! s (0 − ) − ]e Ls (3.22)
Rs Rs
+ jws + jws
Ls Ls
Where t=0 is the moment of short circuit occurrence in the network.
The stator current can be solved from (3.19) by substituting steady-state term of
the equation (3.22) into it.
24
R
U! s (0 + ) + s Lm I!r (0 + )
ψ! − Lm I!r Ls L I! (0 )
I!s = s = − m r +
Ls R Ls
Ls ( s + jws )
Ls (3.23)
Determining the rotor current takes almost the same mathematical path used pre-
viously: (3.19) → (3.20) → (3.18) , where the latter should be firstly covered by
Laplace transformation set, then transient part should be neglected as it was done
while conducting (3.23). All described, in the long run, results in (3.24).
U! (0 ) − I!r (0 + )rr
ψ! r (0 + ) = r +
j(ws − wr )
U! s (0 + )
U! r (0 + ) − j(ws − wr )Lm
(rs / Ls ) + jws
I!r = 2
L (r / L ) L2
j(ws − wr )[Lr + m s s − m + rr ]
(rs / Ls ) + jws Ls (3.24)
Finally, the SCC contributed by DFIG may be calculated from (3.23) by substi-
tuting the value of the rotor current, which is expressed as a function of PWM
converter voltage, which, in its turn, is calculated through closed-loop bandwidth
of rotor-side controller. In such a case, when converter has sufficient capacity
providing necessary voltage to damp transients of rotor current, neglecting
switching transients of the power electronic switches, considering much faster
response from converter in comparison with electromagnetic system of every
electrical rotating machine, the rotor current can be approximately evaluated as
reference predefined value (Xiong et al. 2011), determined by power curve.
As observed from equations (3.23) and (3.24), whether the terminal voltage is
fixed, the steady-state SCC, contributed by the machine with deactivated crow-
bar, will stay unchanged, justifying the representation of the latter as current
25
source for particular moment in time domain. As only initial value of steady-
state current is of interest in terms of relay protection parameterization, the find-
ing given above is sufficient.
Non-symmetrical faults
Due to rotor excitation current modulation capability, the negative-sequence cir-
cuit appears here as active one, where the negative-sequence machine source
regime is coupled with positive by converter variables, reflecting the AC-DC-AC
converter capacity (Fault current contributions from wind plants 2015). The lat-
ter calculations will not be conducted below.
U! s (0 + )
I!s = (3.25)
Rs + jws Ls
Taking into consideration the assumptions made above, the crowbarred DFIG
can be represented as Type 2 WTG, whereas the low-voltage behavior of the
latter is described below. Whether the shortening resistance is not included, it is
26
still accurate enough to equivalent the machine as induction one (Fault current
contributions from wind plants 2015), thus equations introduced for Type 1
WTG can be used in this case.
Fig. 3.6. Connection diagram for a Type 2 WTG (Gevorgian et al. 2010).
Fig. 3.7. The relation between slip and active power output of Type 2 WEC.
(Muljadi et al. 2010).
Fig. 3.8. Equivalent circuit for a Type 2 generator (Gevorgian et al. 2010).
resistance connected in series and bearing in mind that U! r (0 + ) is zero for isolat-
ed rotor circuits, the substitution of the rotor current to (3.23) results in lower
28
stator current values, which means less contribution to the total short-circuit cur-
rent at the point of occurrence.
Generally, mathematical equations used for Type 1 WTG do not justify this ma-
chine type due to considerable values of rotor winding resistance at high slips,
which ruins the congeniality of equivalent schemes of the synchronous rotating
machine with two amortisseur windings and stopped asynchronous machine
(s=1), making possible using the locked rotor current to calculate subtransient
reactance. More generalized approach is given by the Thevenin impedance
(3.26) and pre-fault voltage behind it (3.27) (equation are depicted for symmet-
rical fault at the generator terminals, where the current direction is consistent
with Fig. 3.8).
jX m (Rr + jXrσ )
Z Th = jX sσ +
jX m + (Rr + jXrσ )
(3.26)
R + Rext
Rr = r
s
X X
U! Th = U! s − I!s (RS + j(X sσ + m rσ )) (3.27)
X m + Xrσ
Fig. 3.9. Connection diagram for a Type 4 WTG (Gevorgian et al. 2010).
29
This design provides separation between the WTG and the network, allowing the
generator to induce voltages and currents of non-synchronous frequency and,
thus, the directly-interconnected (the designs based on utilizing asynchronous
generators may include gearbox) turbine to yield power on variable rotation
speeds, forming the efficiency of Type 4 installations. Besides, it also leads to
buffering the transients from both ends, i.e. mechanical transients from the gen-
eration unit side and dynamic ones from another do not transfer through the junc-
tion point.
In most of the cases, inverter control algorithm maintains the specified active
output power with limited variation of current infeed. Whereas, the allowed cur-
30
rent vs. time duration curve is determined by thermal withstand capability and
thermal time constant of the electronic switches, which, in its turn, are depended
on cooling and the rating of the inverter. Generally, the value of maximum
steady-state current drawn from the DC link is set up as 1.1 p.u (up to 1.5 p.u.)
and the surge is detected during 1 to 2 cycles of nominal synchronous frequency.
From this point, the maximum value of the steady-state short-circuit current can
be fixed as 1.1 p.u. (up to 1.5 p.u.), regardless the voltage deep below the critical
value resulting in current limitation, and, thereby, allows to represent this ma-
chine type as current source.
As one more influencing factor coming into play, the applicable grid code should
be considered, specifying so-called Low Voltage Ride-Through (LVRT) (Fault
Ride-Through) reactive power droop and its dead band as grid dynamic support
remedy. The requirements can vary significantly depending on the established
grid code. Hereafter, the VDE Transmission Code (Berndt et al. 2007) is taken as
the basis, determining the reactive current injection capability of at least 100% of
the rated current and control response time no more than 20 ms.
Both listed factors forms the paradigm of rough calculation of the maximum
steady-state current as an iterative process depicted below.
Assuming exceptionally inductive current nature of the current source, initially,
voltage (3.29) and current (3.28) drawn from the mains are calculated at the
point of fault occurrence, omitting the Type 4 machine contribution (Fig. 3.12).
1,1U grid1
I sc1 = (3.28)
Z external1 + Z grid1 + Z sc1
''
E DC1 − U sc1
I sc _ inver1 = = j1,1I Gnom (3.30)
Z external1 + Z inver1
'
E eq1
I sc1(1) =
Z eq1
E DC1 −U sc1(1)
I sc _ inver1(1) = (3.31)
Z external1 + Z inver1(1)
'
The routine continues till the error between current and previous iteration values
is eliminated I sc _ inver1(i−1) = I sc _ inver1(i) .
The accuracy of the described method can be risen by involving the exact reac-
tive power droop curve into the routine.
The Type 5 WTG is a generating set comprising variable-speed wind turbine and
synchronous generator mechanically decoupled through torque-control module
(Fig. 3.13) (Gevorgian et al. 2010).
33
Fig. 3.13. Connection diagram for a Type 5 WTG. (Gevorgian et al. 2010).
depending on the presence of d-axis damping winding, if the time duration con-
sidered is less or equal to 0,5 s, otherwise – transient values characterize the ma-
chine under LVRT conditions. The mentioned change in equivalent model char-
acteristics is due to dying away of leakage fluxes of the magnetically bound d-
axis excitation winding and damping winding, while magnetizing flux linkage of
considered magnetic circuit has higher time constant and, thus, maintained.
The maximum power, possible to be yield from the PV element for the given
insulation level, is the knee curve point (Fig. 3.14), denoted as Pmp, which, tak-
ing into account inherently unregulated nature of this DC power source, should
be maintained intermediary between the source and the grid connection point.
The mentioned Maximum Power Point Tracking (MPPT) is generally performed
by varying source output voltage, whereas the latter is assigned to a DC/ DC
converter preconnected before an output inverter, functionally identical to the
units used in other renewable energy converters, such as variable speed wind
turbines.
3.4. Microturbines
where the second-shaft, holding the generator rotor, run at the speed, providing
synchronous frequency of the stator electrical field with given number of poles
(Farret et al. 2006).
The first microturbine type, due to its high-frequency electrical field, intercon-
nects with a grid only by means of AC-DC-AC converter, designed as that de-
scribed for Type 4 WTG. Consequently, the representation of this machine type
simmers down to similar, as was derived for Type 4 WTG, covering all the as-
sumptions for the latter.
The second type, running at synchronous grid frequency, utilizes direct junction,
which latches the equivalent model over the exact generator type, implemented
here:
− conventional synchronous generator with the LVRT model described
above for Type 5 WTG;
− induction generator with similar LVRT behavior to Type 1 WTG.
Based on this reasoning, the generic classification branches assigning the behav-
ior and the models of electrical power sources under LVRT conditions is elabo-
rated and depicted in Table 1.1.
Generating Prime mover Generator type Interconnection Phase-to-phase fault LVRT model
source type speed regime type
Without variable resistance connected in series with
wound-rotor windings:
E ''(0) = (U(0) − x ''d I (0) sin ϕ (o) )2 + (x ''d I (0) cosϕ (0) )2
1
x "= x =
*d *2 I lock.rotor
*
Synchronous-
AC generat- Induction genera- With variable resistance connected in series with wound-
speed Direct coupling
ing source tor rotor windings1:
with a grid
X X
U! Th = U! s − I!s (RS + j(X sσ + m rσ ))
X m + Xrσ
jX m (Rr + jXrσ )
Z Th = jX sσ +
jX m + (Rr + jXrσ )
Rr + Rext
Rr =
s
1
Depicted equations refer to Thevenin’s equivalent generator.
39
x '' , x
Synchronous gen- * d *2
x' , x
* d *2
1
Summing pre-fault terminal voltage with d-axis voltage drop in internal impedance reflects the overexcited pre-fault regime of the synchronous machine, whereas
subtracting reveals the underexcited mode.
2
Summing pre-fault terminal voltage with d-axis voltage drop in internal impedance reflects the overexcited pre-fault regime of the synchronous machine, whereas
subtracting reveals the underexcited mode.
40
Partial decoupling
1
The proposed model is justified only in case of inverter protection threshold value excess, full generating current substitution by reactive consistent and presence
of exceptionally positive-sequence in infeed current (see subsection “Wind Turbine with Permanent Magnet Synchronous Generator”).
2
Reference predefined value of the rotor current reflects the operating point location on configured power curve (for instance, wind speed vs. generated power), as
a parametric part of the AC-DC-AC converter ACS. The equality is justified only when all the assumptions made in subsection “Wind Turbine with Doubly-Fed
Induction Generator” are justified.
41
1
See subsection Wind Turbine with Doubly-Fed Induction Generator.
2
As rotating electrical DC generators are out of the boundary, encircling distributed generation technologies; only DEC systems are considered here, i.e. PVs, FCs,
thermionic, thermoelectric converters, MHD generators.
3
The proposed model is justified only in case of inverter protection threshold value excess, full generating current substitution by reactive consistent and presence
of exceptionally positive-sequence in infeed current (see subsection “Wind Turbine with Permanent Magnet Synchronous Generator”).
42
Scheme layout should include upstream and relatively prolonged by at least one
more electric power element downstream part (with regard to DG junction point)
to evaluate both surplus short-circuit current influence toward selectivity opera-
tion of downstream protections and opposite impact on the boundary between
upstream and downstream bays. Meeting these requirements may lead to suffi-
ciency of coupling DG in only one point.
All the features described lead to the following one-of-several applicable scheme
(Fig. 3.1).
Under first approach, on the grounds of calculations listed in Appendix 1.1, fol-
lowing results are gained:
I
I p1 = 9,162kA, t p1
I
= 0,1s
II
I p1 = 2, 99kA, t p1
II
= 0, 7s
III
I p1 = 0, 768kA, t p1
I
= 1, 3s
I
I p2 = 6, 915kA, t p1
I
= 0,1s
II
I p2 = 2, 718kA, t p1
I
= 0, 4s
III
I p2 = 0, 475kA, t p1
I
= 1s
Where I, II and III here and in further pages are related to instantaneous (I>>>),
slightly time-delayed (I>>), and stage with significant time delay (I>), respec-
tively.
Slight time delay of instantaneous stage are needed to offset protection from “ar-
tificial” short-circuits induced by tubular arrester operation principle (if one is
installed as overvoltage protection within observed grid).
44
Higher time delay implemented to I>> stage of the first protection is determined
due to its low sensitivity while coordinated with I> stage of adjacent one and,
thus, second stage of latter protection is used for this purpose with necessary
increasing in time delay of first protection I>> stage by selective time interval.
Evaluated lengths of protected zones revealed that instantaneous relays do not
cover critical distances, which are of high importance due to the fact that every
three-phase short circuiting within this zone leads to decreasing of voltage on the
feeding substation busbar lower than 0,5 [p.u.] . The latter has crucial impact
toward transient stability of rotating motors (synchronous and asynchronous)
connected to low-voltage busbar of feeding substation. Consequently, one of
possible solutions of posed issue can be an acceleration of I>>> stage performed
by relay of minimal voltage connected though positive-sequence filter (Шабад
2012) with threshold value of U III1acceler = 0, 5U nom = 17, 5kV .
Taking into account all the previous, the network protection can be considered as
one fully met all the requirements.
from the farthest from utility grid voltage source to the closest one) in cases there
is a reason for this, i.e. short circuit.
Stages with small time delay (I>>) are not exposed to DG current contribution in
considered case, but sensitivity of next-to-integration-point protection can be
reduced, which undergoes in direct ration to infeed coefficient and, thus, to DG
rated power. The latter phenomenon is partly damped by field collapsing inher-
ent to asynchronous generators under LVRT conditions before the moment of
I>> tripping ( ≥ 0, 5s ) (based on rotor electromagnetic time constant depicted in
(Gevorgian et al. 2010), e.g. it results in current infeed declining to zero value
for thee-phase fault at the machine terminals.
4
2×10
ithrough.p1 ( x)
ithrough.p2 ( x)
iwithoutDG ( x) 4
1.5×10
II1
III1
IIII1 4
1×10
II2plot ( x)
III2plot ( x)
IIII2plot ( x)
3
5×10
Imaxoper( x)
0
0 2 4 6 8 10
x
Fig. 5.2. Impact caused by asynchronously rotating machine ((x – SC placement in km, which
limited by both cable lengths constituting 10 km in overall).
DGs are prohibited by IEEE 1547 (IEEE Std. 2003) referred in this paper as ap-
plicable grid code, only asynchronously rotating load re-acceleration current
flows have to be reevaluated in terms of exceeding respective protections’
threshold values.
The way these generators in question affects the grid protection is almost the
same as in induction machines’ case, by which it is meant that alike stages are
influenced, but in another extent with regard to the same nominal power and
point of coupling of integrated DG. As clearly seen from the Fig. 5.3 the full loss
of selectivity by I>>>stage of downstream protection took place here as well,
likewise similar findings for time-delayed stages can be stated, except the ma-
chine’s field collapsing not related to independently excited synchronous ma-
chines.
4
2×10
ithrough.p1 ( x)
ithrough.p2 ( x)
iwithoutDG ( x) 4
1.5×10
II1
III1
IIII1 4
1×10
II2plot ( x)
III2plot ( x)
IIII2plot ( x)
3
5×10
Imaxoper( x)
0
0 2 4 6 8 10
x
Fig. 5.3. Impact caused by DG with full-flexible grid coupling (x – SC placement in km, which
limited of both cable lengths constituting 10 km in overall)
47
ithrough.p1 ( x)
ithrough.p2 ( x)
iwithoutDG ( x) 4
1.5×10
II1
III1
IIII1 4
1×10
II2plot ( x)
III2plot ( x)
IIII2plot ( x)
Imaxoper( x) 3
5×10
0
0 2 4 6 8 10
x
Fig. 5.4. Impact caused by DG with full-flexible grid coupling (x – SC placement in km, which
limited of both cable lengths constituting 10 km in overall).
1
AC-DC-AC converter incorporates an inverter, thereby meets stated conditions.
48
The TCC are common decision for systems characterized by large variations of
short-circuit currents in different fault locations on the grounds of better utiliza-
tion of protected object overload capability and shorter back-up tripping times.
Here the current dependable time characteristics are evaluated as one of the pos-
sible network protection design approaches, where I >> stages are inhibited and,
consequently, only instantaneous and I> are in operation.
k ⋅ TMS
t(s) = (5.1)
I
( )α −1
Is
Table 6.1. IDMT equation coefficients in compliance with IEC 60255-3 (BS 142).
k α
Normal inverse 0,14 0,02
Very inverse 13,5 1
Extremely inverse 80 2
Long-time inverse 120 1
Every curve under this standard is applied in its particular network conditions,
i.e. very inverse is steeper in comparison with normal inverse, thus it is more
suitable when needed to achieve time selectivity between incoming and outgoing
bays of a mere short-circuit current difference, extreme inverse – when coordi-
49
Based on previous findings regarding current infeed changes when adding vari-
ous types of DG, it can be concluded that somehow notable disturbances here
appear in cases of direct connection of either asynchronous or synchronous gen-
erators.
t III1( x)
t III2( x)
t III1.asynch ( x)
t III2.asynch ( x) 1.5
t III1.synch ( x)
t III2.synch ( x)
t diff.asynch( x)
t diff.synch ( x) 1
t III3( x)
t diff23.asynch( x)
t diff23.synch( x)
1 0.5
1.3
0
0 5 10 15
x
Although, the increased tripping time can lead to low-inertia DGs going out-step
(power electronics interfaced generation could be involved as well, where crucial
aspects in question are DC-link capacitance, its nominal voltage and severity of
the fault) (Xiong et al. 2015).
Overall, the threat of selectivity loss here occurs in upline area when connecting
DG, also, the probability of latter is increasing in direct ratio with α coefficient,
i.e. when shifting from normal inverse to extremely inverse.
51
t III1( x)
t III2( x)
t III1.asynch ( x)
t III2.asynch ( x) 1.5
t III1.synch ( x)
t III2.synch ( x)
t diff12.asynch ( x)
t diff12.synch( x) 1
t III3( x)
t diff23.asynch ( x)
t diff23.synch( x)
1 0.5
1.3
0
0 5 10 15
x
Fig. 5.6. Extremely inverse curves in presence of DG (x – SC placement in km, which limited of
both cable lengths constituting 10 km in overall; 10-15 sector refers to adjacent transformer im-
pedance share).
At this stage, the additional influence toward relay protection (in scope of this
paper – also out-of-step automation (if any) is affected) should be established, as
x. Low-inertia generation transient stability loss due to change in grid pro-
tection tripping time (including power electronics interfaced generation)
(TCC, shift from instantaneous to current-dependent part of a curve in
combined tripping characteristics), violating given LVRT profile of the
interconnection node.
52
It reveals the proportionality of the operating value to the distance from the pro-
tection placement point to faulted one for all types of short-circuits (except sin-
gle-phase, where polarization is needed (5.2)) in case fault resistance is not in-
cluded ( R = 0 ).
z 0 − z1
kI 0 = I0 (5.3)
z1
Floating neutral
1
Regimes of distribution network’s neutral point vary significantly depending on the applicable
grid code and are determined by insulation level of electrical power grid’s elements, circuit
breaker’s interrupting capacity, costs of power/energy not supplied and capacitive current level.
2
Insulation monitoring device.
54
to realize earth-fault protection in grids with floating neutral point. Due to possi-
bility of capacitive leakage current increase, determined by own capacitances of
lumped and distributed elements of the interconnecting DG facility, there is a
probability of necessity to shift to the resonant grounding with an eye to fault
current limiting. The latter requires protection principles reconsideration, which
will not be conducted below.
Resonant grounding
rent path is lined almost bypass the resonant grounding branches, which, in its
turn, means no infeed change and no influence toward this earth protection type.
Protection type, measuring wave transient components of the faulted phase ca-
pacitance discharge and the charge of healthy ones, operates identically whether
the DG neutral point is resonantly grounded or not. The reason for this is high-
order frequency of the both abovementioned processes (several kHz and hun-
dreds Hz, respectively) and low-speed voltage rise around the coil (Шабад М.
2007).
Resistive grounding
There are two distinguished types of resistively earthed neutral point regimes:
high-resistive and low-resistive grounding, which determines the target protec-
tion tripping time and allows to meet requirements of electrical safety and
equipment withstand capability.
There is a disturbance of earth-fault protection functioning in every case, regard-
less the exact implemented grounding type, when the protection principles are
based on current sensing:
− zero-sequence one for high-resistive grounding, where the fault current is
maintained on the level of 2 capacitive current, providing enough sens-
ing and tripping durability and capability of postponing breaking of the
circuit to change-over loads to another source and fault localization.
− 50 Hz current fundamental harmonic component in case of low-resistant
grounding, where the fault sensing is assigned to phase-to-phase fault
protection of the grid.
If insulation-monitoring device ( 3Uo ) is used, the protection malfunctioning as-
Combined grounding
The findings made for high-resistant grounding relates to this subclause as well.
57
Table 5.1. Justification of IEEE 1547 requirements in terms of interconnecting network’s grounding regime
Grounding type Protection principle (operating values) Compliance with IEEE 1547
2.1.6 Floating Zero-sequence voltage 3U 0 Interconnection transformer’s vector group
Zero-sequence current 3I 0 has to incorporate delta/ungrounded wye, if
Fig. 5.7. Island balanced conditions attributes as a function of DG penetration level (where pene-
tration factor = 1 reflects the level of DG providing maximum active load-to-generation ratio, the
60
other factors 0,5 and 2 were set for relatively lower and higher penetration levels, respectively)
(Brundlinger et al. 2005).
Table 5.2 Voltage/frequency windows of DG allowed operation and load quality factors
(Brundlinger et al. 2005).
On the other hand, when it comes to synchronous machines, it goes without say-
ing that sustainable island occurrence has to be taken into consideration (Fig.
5.9), as well as neglecting of asynchronous ones operating without reactive pow-
er injecting sources connected within hypothetically possible-to-occur island.
Implications for the safe operation of an islanded network part are the following:
− exceeding of acceptable limits by power quality parameters, which may
cause malfunction or damage of network and end-customers equipment1;
− uncleared faults in low SCC level conditions formed by loss of substan-
tial infeed, such as mains, or unearthing of neutral points, while protec-
tion threshold values are remained unchanged;
1
Generally, this threat is overcome by DG protection schemes.
62
Fig. 5.10. Single line diagram of a distributed grid with power line carrier AI
protection (Wilsun et al. 2007).
The proposed scheme incorporates two mutually bound devices: a signal genera-
tor (SG) – entrusted to form a signal with regard to preset protocol and a signal
detector (SD) – targeted to recognize it. The reasoning for tripping is formed by
commutation devices installed between considered DG and primary substation
busbar, where signal generator is connected, by interrupting signal for certain
time duration. This particular placement extends the capabilities of this scheme
to detect also continuous voltage unavailability at busbar, which is another type
of mains’ loss. Moreover, SG’s auxiliary inputs (Fig. 5.10) imply manual con-
trol, which can be implemented in case of forming upstream (higher voltage lev-
el) islands or in other circumstances leading to such necessity.
Fig. 5.12. Floating deduction window algorithm resulting waveforms (Wilsun et al.
2007).
The method adapts several ways to recognize the continuously formed signal:
1. RMS value as operating variable - constituting in direct comparison with
preset threshold value over certain period (Fig. 5.13).
65
Fig. 5.13. Segmenting waveform with regard to thyristors’ firing angel ( 30° here)
(Wilsun et al. 2007).
The segmentation point here is placed with regard to thyristors’ firing angle. If
the latter is equal or less than 30°, so 30° degrees ahead the curve can be divided
to insure that signal consists only in II segment.
The signal flagged as present if threshold value is exceeded by vsignal RMS with-
in II segment and opposite situation takes place for the another threshold over I
sector, otherwise – protection trips. Two distinctive setup values are meant to
detect abnormal conditions by being exceeded simultaneously.
where index i refers to sampled point of real and template curves (Fig. 5.14).
Fig. 5.14. Waveform of a windowed sinusoidal template adjusted to thyristors’ firing angle of
30°
A higher index (5.4) implies a higher matching extent between the real signal
and template (Wilsun et al. 2007).
On the grounds of field test oscillation pattern based (spectrum-based and tem-
plate-oriented) algorithms are marked as more reliable due to alleviation of high
noise and low signal level influence on eventual analysis. Furthermore, no volt-
age quality deterioration was measured while implementing techno-economically
feasible signal level of 4% of nominal voltage
As a possible drawback, the impact toward AMR is not identified clearly, but is
still under spotlight and critical due to intersecting principles constituting simul-
taneous overlapping of AI signal and AMR one (Wencong et al. 2007).
Transfer-trip technique
Using telephone lines, optical cables and etc. for the purpose in question. In or-
der to achieve sufficient reliability, a continuously presented signal has to be
implemented, whereas it requires allocated data volume, which forms the main
drawbacks in addition with those listed for SCADA.
NDZ: none.
Parameterization: narrowed down to choosing the time delay
68
Passive methods
With regard to power flow content supplied from utility grid, three situations are
possible with corresponding consequences when the latter is disconnected:
1. ΔP ≠ 0 - sudden voltage jump occurs;
2. ΔQ ≠ 0 - instantaneous phase shift occurs, which makes PLL to adjust the
frequency in accordance with preset template, whereas it causes further
change in reactive power. The iterative process stops when load resonant
frequency is met, so steady state is fixed.
3. ΔP ≠ 0, ΔQ ≠ 0 - the initial steady state.
69
Regarding previous one can conduct that unless steady state occurs within al-
lowed frequency interval and voltage operating point strays inside of prescribed
limits the island is detected.
NDZ: determined by too low active power imbalance to cause OV/UV relay trip-
ping and possibility of reaching load resonant frequency within threshold
boundaries (5.7, 5.8 respectively).
U 2 ΔP U 2
( ) −1 ≤ ≤( ) −1 (5.7)
U max P U min
ΔQ
Q f (1 − ( f f )2 ) ≤ ≤ Q f (1 − ( f f )2 ) (5.8)
min P max
∑U 2
i
THD = i=2
(5.9)
U1
Generally, grid has low impedance forcing low-distorted voltage to PCC with
exceptions for so-called “weak grids”, which possess quite opposite attributes
regarding the question, but are encountered rarely. Likewise, harmonic distortion
injected by switching-mode inverters cannot exceed value of 5% in compliance
with DG interconnecting standards (IEEE 1547, EN 50438 and etc.). By summa-
tion of previous aspects neither THD nor voltage harmonics magnitude/RMS
cannot provide significant values at PCC while mains are connected, which are
laid as fundamental of detection mechanism here.
other nonlinearities, which reduces, chiefly, to the third harmonic. The latter
ends up with high THD and harmonic magnitude/RMS itself.
NDZ: the proposed detection mechanism is prone to fail while high-order har-
monics are filtrated or non-induced, thus when load content possess low-pass
characteristics, such as that high-Q loads have or when low distortion is pro-
duced initially, e.g low distortion outputs of implemented inverters.
Parameterization: suffers challenges of choosing thresholds due to convergence
of its determinants having opposite effect, i.e. the presence of high-value “weak”
grid’s impedance and filtrating characteristics of the RLC load, which can lead
to voltage harmonics of lower intensity than contains in inverter output current
with allowed 5% distortion.
Power quality deterioration: none.
Ability of implementing with multiple DGs: full.
(Bower 2002).
L and H refer to low and high pass filtering, respectively and ↓ 2 denotes sub-
sampling1.Outputs are determined in accordance with following equations (5.10),
(5.11).
+∞
a j+1[ p] = ∑ l[n − 2 p]a [n]
j (5.10)
n=−∞
+∞
d j+1[ p] = ∑ h[n − 2 p]a [n]
j (5.11)
n=−∞
Elements a j are used to declare the next step (scaling), d j – so-called wavelet
coefficients determining the output of the transform, l[n] and h[n] are coeffi-
cients of low and high pass filters, respectively.
1
The decimation in two is justified due to sampling theorem as half-band pass filter is applied.
74
Fig. 5.18. Deviations of DWT coefficients reflecting the intensity of various harmonic orders.
NDZ: none.
Parametrization: reveal islanding pattern of high-order harmonics’ intensity
change and distinguish among other electromagnetic transients.
Power quality deterioration: none.
Ability of implementing with multiple DGs: full.
Active methods
Noticeable difference in impedances of local load and utility power grid covered
by kirchhoff’s law reveals that current flow through at most grid impedance path
is fully forced to the load in case of disconnection from utility with all conse-
quents regarding impedance and power changes.
1
As an alternative paradigm can be used for enclave of difficult-to-identify islanding conditions.
76
dvPCC
If the mains are disconnected, the impedance determined as reaches sig-
diDG
nificant value, likewise dvPCC by itself. Voltage change in case of active power
dP R
imbalance can be calculated as dV = (Bower et al. 2002), whereas, in
2 P
case of precise balance, variations of CSI current is entrusted to eliminate steady-
sate conditions.
NDZ: as long as grid impedance is lower than the local load one, whereas, it is
fully met while the grid is strong, the NDZ is extremely small for single consid-
ered DG.
As integration density increases the contribution of each individual unit, keeping
up with reverse dependence, goes down. The latter results in need of mutual syn-
chronization in perturbations. This phenomenon is depicted in Fig. 5.19, where
the number of generating units is 50 times increased while stay unsynchronized
in perturbations.
Parameterization: the thresholds values are determined by grid-load impedance
window, which in some cases is far small, such as “weak” grid and extremely
high-power loads (assuming short circuit).
Power quality deterioration: significant reduction when either high-impedance
grid is connected or perturbations are synchronized.
Ability of implementing with multiple DGs: full inability due to inducing of in-
stability even when the perturbations are not synchronized and directly propor-
tional voltage quality degradation, accelerated when mentioned above synchro-
nization is used.
77
Fig. 5.19 Unsynchronized perturbations for case of 50 DGs (Bower et al. 2002).
Fig. 5.20. Phase-shift-frequency curve of an inverter and resistive load (Lopes et al.
2006).
The error occurred is entrusted to make operating value to go beyond the al-
lowed frequency window.
As it can be revealed from the figure above, the proposed method fails, when
dϕ load
rate of load frequency response excess idem of the inverter, which results
df
in steepness increasing of the load curve and, also, A and C points are contained
within the normal frequency range (Lopes et al. 2006) (Table 5.2).
2tz
cf = (5.12)
TVutility
When an island is incepted, assuming for simplicity that load has resistive na-
ture, the voltage curve is kept up with distorted current leading to zero crossing
by former in advance and, so to reduction of tz . This consequence reflects the
Fig. 5.21. Inverter current frequency shift by means of positive feedback link to voltage curve at
PCC (Bower et al. 2002).
80
1
φ = −arctg(R(wC − )) (5.13)
wL
NDZ: chopping fraction dependable, i.e. when the implemented value is small –
the same as inherent to SMS algorithm, whereas increasing in cf leads to NDZ
shifting to capacitive loads’ zone: providing precise detection for low-Q loads,
albeit high-Q loads still interfere the islanding detection even here.
Parameterization: same as in SMS.
Power quality deterioration: slight, increasing with penetration level.
Ability of implementing with multiple DGs: needs synchronization in perturba-
tions.
Sandia Frequency Shift (SFS) is an extension of AFS method, where the imple-
mented positive feedback link has voltage frequency at PCC as the reference.
Small deviations from rated one, which are likely to occur when the utility grid is
81
disconnected are amplified and used to shape inverter output current in accord-
ance with assembled equations (5.14).
Where W0 -initial length of the truncation/dead time [%] (Fig. 5.22), K F - SFS
gain [%/Hz], fPCC , fgrid - PCC and nominal frequencies respectively, Tgrid - volt-
Fig. 5.22 Opposite-directional inverter current biasing by means of frequency change (Robitaille
et al. 2005).
In case there is ideal match of generated and consumed power in every time in-
stant, the permanent perturbation f0 incorporated in (5.14) will “swing out” the
operating point to deviate from steady state. Regarding preceding the proposed
method almost eliminates the NDZ.
82
Sandia Voltage Shift (SVS) has similar pathway as SFS, but the latter is applied
to voltage magnitude at PCC as (5.15).
NDZ:
kf
Standalone SFS: depends on , where the increase of latter shifts the NDZ for-
ward to capacitive loads’ region.
Standalone SVS: has similar nature of NDZ as UV/OV relays, but the values are
marginally low.
SFS+SVS: infinitesimal NDZ, especially for Q-component.
Parameterization: similar to SMS.
Power quality deterioration: increasing with penetration level.
Ability of implementing with multiple DGs: at high penetration levels there is a
threat of causing system-level power deterioration and noticeable transients re-
sponse.
1 f [k −1]− 50Hz
θ APS [k] = ⋅ 2π ( ) + θ 0 [k] (5.16)
α 50Hz
θ 0 [k] = θ 0 [k −1]+ Δθ ⋅ sgn(ΔfSS ) (5.17)
1, ΔfSS > 0
sgn(ΔfSS ) = 0, ΔfSS = 0 (5.18)
−1, ΔfSS < 0
frequency.
84
The paradigm entirely overcomes NDZ cornerstones of SMS and AFD, which is
best understandable from the Fig. 5.23 for RLC load, where angle compensation
(opposite signs) possibility is highlighted at 55,7 Hz for load and S-shaped in-
verter response curve, as well as reaching resonant frequency by RLC-circuit at
57 Hz, which mean stabilizing and forming of NDZ.
Fig. 5.23. Comparative depiction of three frequency-based active AI algorithms from the per-
spective of their NDZs (Hung et al. 2003).
Fig. 5.24. Voltage curves for opposite power deviations (Jeong et al. 2006).
The phase-frequency mutual changes for resistive load are represented in Fig.
5.25.
86
Fig. 5.25. Phase-frequency dependence for section maintaining (a) positive Q deviation, (b)
negative Q deviation (Jeong et al. 2006).
The phase difference raises the increasing frequency deviation while operating in
islanded mode, which makes UF/OF relay to trip, but due to perturbation limita-
tion of a mere 2,5% of rated active power output the power factor is not some-
how significantly affected in grid-connected mode. (Jeong et al. 2006).
NDZ: none.
All units monitor voltage, frequency and impedance, where for accomplishing
impedance measurements small current is injected to a grid on a permanent ba-
sis. It makes possible from protection perspective to react on rapid changes of
impedance value, i.e. happening within small decrement of time.
87
When the utility having low impedance at the harmonic frequency is connected,
the injected current flows almost bypass the load directly to the grid non-causing
abnormal voltage level at PCC. Upon disconnection from utility, the harmonic
current is forced to load, which leads to high voltage level of this particular fre-
quency, which in its turn can be detected and treated as islanding.
The drawbacks of this method are voltage quality decrement due to direct de-
pendence of harmonics’ intensity on load impedance and inability of implement-
ing in multiple DGs’ case. The latter is due to possibility of false tripping while
injecting the same frequency oscillation simultaneously.
Passive methods
NDZ: widening is kept up with threshold value and isolated network rotation
masses increase; inverse relationship with load power factor.
Parameterization: threshold value calculation (5.19) and ensuring the condition
(5.18), the same must be conducted for local generation/load commutation.
t tP k
df ΔPgen f0 − τ f ΔPgen − load load
τ gen (Pload− ΔPgen )+τ load Pload
=− e =− e (5.19)
dt Pload kloadτ f τ gen (Pload − ΔPgen ) + τ load Pload
,
where f0 – nominal frequency , ΔPgen - induced active power imbalance, Pload -
summarized active power of islanded grid’s load, τ f , τ gen , τ load - time constants
of frequency change, the same separately for generation and load, respectively.
The target value should be set in accordance with (5.20).
df dfutility _ grid
>
dt dt , (5.20)
dfutility _ grid
where dt is referred to system-wide power imbalances.
Power quality deterioration: none.
Ability of implementing with multiple DGs: full whether ranged threshold values
are set for DGs interconnected in one distribution grid.
∂f
As an enhancement of described algorithm partial derivative ∂P can be used to
increase small-imbalance sensitivity (Pai et al. 2002).
dp
Rate of change of power dt (Redfern et al. 1993) was substituted by abovemen-
tioned algorithms and is not currently used.
90
Vector surge
Steady-state regime of a synchronous generator is characterized by voltage phas-
or diagram depicted in Fig. 5.27 (a), whereas every change in active power sup-
ply is reflected in vector shifting (Fig. 5.27 (b)) due to slowing/acceleration of a
rotor, caring excitation windings.
Fig. 6.35. Synchronous generator (a) steady-state and (b) change-in-supply regime (Freitas et al.
2005).
Fig. 6.36. Generator’s terminal voltage waveform deviation in case of active power supply
change (Freitas et al. 2005).
Vector surge relays compare the duration of adjacent periods, starting at zero-
crossing point with positive waveform derivative, and form tripping signal in
case of threshold value excess. Undervoltage relays’ signals are involved into
considered algorithm similarly to ROCOF protection.
91
Active methods
Active AI schemes for synchronous machines are implemented as positive feed-
back links to frequency and voltage deviations integrated into speed and excita-
tion ACS as series connected washout and first-order filters with inertialess gain
(Fig. 6.37 and Fig. 6.38).
1
Switch to higher threshold value, calculated identically to ROCOF, can be implemented for
relay-receivers in case of relay-transmitters malfunction (Bright 2001).
92
Such filter connection provides scheme operation only during transients exclud-
ing steady-state regulation and attenuating high-order frequency noise.
NDZ: none.
Parameterization: compromise K value to simultaneously ensure speed-enough
tripping under islanding conditions and infinitesimal impact under grid-
connected
Power quality deterioration: none.
Ability of implementing with multiple DGs: conceptually full but still needs vali-
dation (Ye Z. 2006).
Based on findings acquired throughout literature review, and the protection and
auto-reclosing behavior analysis, various in principle remedies are proposed here
93
Table 6.1. Incoordination elimination iterative sequences for various types of grid protection – DG groups
1
Hereafter, mentioning application of directional element is included for originally non-directional protections.
96
In compliance with the above list, one of the possible approaches, proposed in
(Muljadi et al. 2011) for wind power farms, for facilities with applied motor and
generation (if asynchronous) soft starters, the saturable core FCL can be consid-
98
ered. It consist of several coils wound across two or more ferromagnetic cores
(Fig. 6.1).
Fig. 6.1. Magnetic and electrical circuits of saturable FCL (Muljadi et al. 2011).
A coil depicted in the center is fed by direct current, which magnetizes and bias-
es a hysteresis loop with respect to the origin (Fig. 6.2), in such a way, that load-
flow currents induce fluxes oscillating within green zone. The fluxes in both
cores is lined up with implemented coil interconnection, thereby oppositely vary-
ing:
− load-flow current – within green area referring to low-impedance of the
FCL;
− SCC – with partial oscillation beyond the green zone due to current mag-
nitude increase and, consequently, proportional change in flux value and,
in the long run, in emf (6.3)
Φ = Li (6.3)
99
Fig. 6.3. FCL emf change under LVRT conditions (Muljadi et al. 2011).
Corresponding to the latter, resulting emf curves are depicted in Fig. 7.4
for load-flow and LVRT from left to right, respectively.
Fig. 6.4. Resulting emf of the FCL under load-flow and LVRT conditions ((Muljadi et
al. 2011).
2
Where iac - instantaneous value of AC-current through the FCL, I max - current of
the full saturation of the magnetic cores, corresponding to average magnetic in-
ductance Bsat , K – current interval of active change from unsaturated to saturated
state and vice versa.
The expression above is scaled in such a way that while iac = 0 the value of
magnetic induction is equal to induced biasing induction B = Bdc _ bias , which cor-
responds to virtual line O 'O '' in Fig. 6.2. Moreover, it is assumed, that maximum
instantaneous current value through the FCL does not exceed the maximum pre-
defined value I max ( iac ≤ I max ). The mentioned maximum current value is estab-
101
lished for every FCL and derived from the value of magnetic core’s coercitive
force.
∂B(iac )
L = nac Acore , (6.6)
∂iac
Where nac is a number of windings, Acore – a cross section of the magnetic core.
To sum up, the following parameters have to be evaluated under load-flow and
fault regimes in order to opt a FCL: I max , Bsat , K and Lair−core , representing air-
Conclusion
In this paper, the meaning of the term “distributed generation” is addressed from
the scope-encircling standpoint, and generating technologies encompassed by the
latter are listed as primary energy source-prime mover combinations.
102
In chapter 3, all sources appearing as new generation types are equivalented un-
der LVRT conditions, with models adapted for manual calculations of the initial
values of the steady-state short-circuit currents of symmetrical and unsymmet-
rical faults for positive and negative sequences separately (where possible). The
preceding essentially excludes a software simulation intermediary for vast major-
ity of considered sources, adding to simplicity of protection-specific routines.
Moreover, the paradigm of Type 4 ENERCONE WECs’ representation are ex-
tended and adapted for inverter-based generation in order to achieve ubiquitous
application of existing short-circuit calculation practices.
In chapter 4, the preceding, with the addition of conventional generation phase-
to-phase fault LVRT models, is classified by characteristics determining their
behavior when faulted, namely generating source type, generator type, and inter-
connection type – where the latter is dependable on prime mover (with mechani-
cal power transmission system) speed regime.
To sum up all findings made through, the issues and remedies are collected in
Chapter 6 with referring to previous sections and subordinate problem descrip-
tion of current limiting.
103
References
(Artale et al. 2013) Artale, G.; Cataliotti, A.; Cosentino, V.; Nguyen, N.; Di Cara, D.;
Tine, G., "Measurement and communication interfaces for distribut-
ed generation in smart grids," Applied Measurements for Power Sys-
tems (AMPS), 2013 IEEE International Workshop on , vol., no.,
pp.103,107, 25-27 Sept. 2013.
(Berndt et al. 2007) H. Berndt & M. Hermann, H.D. Kreye, R. Reinisch, U. Scherer, & J.
Vanzetta, "Network and system rules of the German transmission
system operators 2007". VDN, Berlin, Germany, 2007.
Available at
http://www.vde.com/de/fnn/dokumente/documents/transmissioncode
%202007_engl.pdf
(Brundlinger et al. 2005) Brundlinger, R.; Bletterie, B., "Unintentional islanding in distribu-
tion grids with a high penetration of inverter-based DG: Probability
for islanding and protection methods," in Power Tech, 2005 IEEE
Russia , vol., no., pp.1-7, 27-30 June 2005.
(CIGRE 2000) CIGRE TF38.01.10 Modeling New Forms of Generation and Stor-
age, November 2000.
Available at
http://fglongatt.org/OLD/Archivos/Archivos/SistGD/CIGRE-TF-
380110.pdf
(Farret et al. 2006) Farret, F.; Simões, M., "Microturbines," in Integration of Alternative
Sources of Energy , 1, Wiley-IEEE Press, 2006, pp.215-232.
(Fault current contribu- "Fault current contributions from wind plants," Protective Relay
tions from wind plants Engineers, 2015 68th Annual Conference for, College Station, TX,
2015) 2015, pp. 137-227.
(Freitas et al. 2005) W. Freitas, Wilsun Xu, C. M. Affonso and Zhenyu Huang, "Com-
parative analysis between ROCOF and vector surge relays for dis-
tributed generation applications," in IEEE Transactions on Power
Delivery, vol. 20, no. 2, pp. 1315-1324, April 2005.
105
(Gevorgian et al. 2010) Gevorgian, V. & Muljadi, E. 2010. Wind Power Plant Short Cir-
cuit Current Contribution for Different Fault and Wind Turbine To-
pologies [online document].
Available at http://www.nrel.gov/docs/fy11osti/49113.pdf
(Huang et al. 1998) HUANG S.J., HSIEH C.T. , HUANG C.L. : ‘Application of wave-
lets to classify power system disturbances’, Electr. Power Syst. Res.
, 1998, 47 , pp. 87–93.
(Hung et al. 2003) Guo-Kiang Hung; Chih-Chang Chang; Chen, Chern-Lin, "Automat-
ic phase-shift method for islanding detection of grid-connected pho-
tovoltaic inverters," Energy Conversion, IEEE Transactions on ,
vol.18, no.1, pp.169,173, Mar 2003.
(IEEE Std. 2003) IEEE Standard for Interconnecting Distributed Resources with Elec-
tric Power Systems," IEEE Std 1547-2003 , vol., no., pp.1,28, July
28 2003.
(Jeong et al. 2006) Jeong, J.B.; Kim, H.J., "Active anti-islanding method for pv system
using reactive power control," Electronics Letters , vol.42, no.17,
pp.1004,1005, August 17, 2006.
(Kundur et al. 1994) P. Kundur, N. Balu, and M. Lauby, Power system stability and con-
trol. McGraw-Hill Professional, 1994.
106
(Lopes et al. 2006) LOPES L.A.C. , HUILI S. : ‘Performance assessment of active fre-
quency drifting islanding detection methods’, IEEE Trans. Energy
Convers. , 2006, 21 , (1), pp. 171–180.
(Mohan et al. 1995) Mohan, N., Undeland, T. M., and Robbins, W. P., Power Electron-
ics: Converters, Applications, and Design, 2nd ed., John Wiley &
Sons, New York, 1995.
(Muljadi et al. 2011) Muljadi, E.; Gevorgian, V.; DeLaRosa, F., "Wind power plant en-
hancement with a fault current limiter," Power and Energy Society
General Meeting, 2011 IEEE , vol., no., pp.1,7, 24-29 July 2011.
(Pai et al. 2002) Fu-Sheng Pai; Shyh-Jier Huang, "A detection algorithm for island-
ing-prevention of dispersed consumer-owned storage and generating
units," Power Engineering Society Winter Meeting, 2002. IEEE ,
vol.1, no., pp.359 vol.1, 2002.
(Redfern et al. 1993) M. A. Redfern, O. Usta and G. Fielding, "Protection against loss of
utility grid supply for a dispersed storage and generation unit," in
IEEE Transactions on Power Delivery, vol. 8, no. 3, pp. 948-954,
July 1993.
(Salman et al. 2004) Salman, S.K.; King, D.J.; Weller, G., "Investigation into the devel-
opment of a new ANN-based relay for detecting loss of mains of
embedded generation," Developments in Power System Protection,
2004. Eighth IEE International Conference on , vol.2, no.,
pp.579,582 Vol.2, 5-8 April 2004.
(Singam et al. 2006) SINGAM B. , HUI L.Y. : ‘Assessing SMS and PJD schemes of anti-
islanding with varying quality factor’. IEEE Power and Energy
Conf., November 2006, pp. 196–201.
(Wencong et al. 2007) Wencong Wang; Kliber, J.; Guibin Zhang; Wilsun Xu; Howell, B.;
Palladino, T., "A Power Line Signaling Based Scheme for Anti-
Islanding Protection of Distributed Generators—Part II: Field Test
Results," in Power Delivery, IEEE Transactions on , vol.22, no.3,
pp.1767-1772, July 2007.
(Wilsun et al. 2007) Wilsun Xu; Guibin Zhang; Chun Li; Wencong Wang; Guangzhu
Wang; Kliber, J., "A Power Line Signaling Based Technique for
Anti-Islanding Protection of Distributed Generators—Part I: Scheme
and Analysis," in Power Delivery, IEEE Transactions on , vol.22,
no.3, pp.1758-1766, July 2007.
(Xiong et al. 2011) Xiong, X. & Ouyang, J. & Jin J. 2011. Calculation of steady-state
short-circuit current of wind turbines with doubly fed induction gen-
erator. Advanced Power System Automation and Protection
(APAP). 16-20 Oct. 2011. Page(s): 1782 - 1788.
(Xiong et al. 2015) L. Xiong; F. Zhuo; F. Wang; X. Liu; Y. Chen; M. Zhu; H. Yi, "Stat-
ic Synchronous Generator Model: A New Perspective to Investigate
Dynamic Characteristics and Stability Issues of Grid-tied PWM In-
verter," in IEEE Transactions on Power Electronics , vol.PP, no.99,
108
pp.1-1
Available at
http://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=7322271
&isnumber=4359240
(Ye Z. 2006) Ye, Z. 2006. Study and Development of Anti-Islanding Control for
Synchronous Machine-Based Distributed Generators [online docu-
ment].
Available at http://www.nrel.gov/docs/fy06osti/38018.pdf
Utility grid:
U nom = 110kV
SSC min = 2500MVA
SSC max = 5000MVA
Transformers:
T1:
Str1 = Snom = 80MVA
U SC 01 = U SC(0tap) = 10, 5%
U SC min1 = U SC(−9tap/−16%) = 10, 44%
U SC max1 = U SC(+9tap/+16%) = 10, 91%
U HVnom1 = 115kV
U LVnom1 = 38, 5kV
OLTC: ±16%, ±9
T2, T3:
Str 2 = Snom = 16MVA
U SC 02 = 7, 5%
U HVnom2 = 38, 5kV
U LVnom2 = 6, 3kV
Cables:
U nom = 35kV
Phases conductors made of Al 240 square mm in cross section cabled in a row
in the ground
r = 0,125Ω / km
x = 0,193Ω / km
Length between adjacent busbars, i.e. A and B< B and C
l = 5km
Loads:
Load 1 (industrial type):
Industrial generalized load approximated to include 25% synchronous, 10% in-
duction, 40% low-voltage induction motors, 10% of converters, 11% and 4%
of electrothermal and lighting loads (РД 153-34.0-20.527-98).
Appendix 1.2
The general assumptions used throughout further calculations’ routines (РД 153-
34.0-20.527-98):
1. neglecting emf phase shift of different synchronous machines and varia-
tions of their rotating speeds for fault duration less than 0,5s;
2. neglecting DC ties;
3. neglecting magnetic saturation of electrical machine’s steel;
4. neglecting magnetizing currents of transformers and autotransformers;
5. neglecting resistances of electrical power equivalent scheme’s elements if
the resulting one, determined with regard to a short-circuit point, consti-
tutes less than 30% of resulting inductance;
6. approximate consideration of electric loads connected in various nodes of
initial scheme;
7. considering electrical power elements’ resistances as DC based.
Protection design
U110 ULVnom1
EUGmin:=
3 UHVnom1⋅ ( 1 + 0.16)
U35
Eload1 := 0.865⋅
3
U35
Eload2 := 0.788⋅
3
Referring utility grid impedance to 35 kV voltage level:
2 2
⎡
U110 ULVnom1 ⎤
xUGmax:= ⎢ ⎥ = 0.384Ω
SSCmax UHVnom1⋅ ( 1 − 0.16)
⎣ ⎦
2 2
⎡
U110 ULVnom1 ⎤
xUGmin:= ⎢ ⎥ = 0.403Ω
SSCmin UHVnom1⋅ ( 1 + 0.16)
⎣ ⎦
(
u SC02 UHVnom2 )2
xtr2 := ⋅ = 6.948Ω
100 Str2
( )
zcable := rcab.per.km + xcab.per.km⋅ i ⋅ l = ( 0.625 + 0.965i) Ω
Generalized domestic and industrial load impedances (load 1 and load 2, respec-
tively):
Appendix 1.4
2
UHVnom2
zload1 := ( 0.04 + 0.54i) ⋅ = ( 3.706 + 50.026i) Ω
Str2
2
UHVnom2
zload2 := ( 0.15 + 0.43i) ⋅ = ( 13.896 + 39.835i) Ω
Str2
U35
Iself.acceler1 := = ( 125.186 − 658.446i) A
(
3⋅ zUGmax+ ztr1min + zcable + zeqload2 )
(
zload1 + ztr2 )
Iself.acceler2 := Iself.acceler1⋅ = ( 104.754 − 344.423i) A
zload2 + zcable + ztr3 + zload1 + ztr2
Iself.acceler2
IIII2 := koffset ⋅ = 412.502A ,
kreturn
Where koffset - offset coefficient taken into the equation to consider possible inac-
( )
IIII1 := max IIII1, IIII1alt1, IIII1alt2 = 767.985A
tIII2 := 1s
tIII1 := tIII2 + Δt = 1.3s
EUGmax 3
ISC2max:= = 6.287× 10 A
zUGmax+ ztr1min + 2zcable
( ) (
EUGmax⋅ zload1 + ztr2 + Eload1 ⋅ zUGmax+ ztr1min + zcable )
EeqSC2max:=
(
zload1 + ztr2 + zUGmax+ ztr1min + zcable )
( )(
zload1 + ztr2 ⋅ zUGmax+ ztr1min + zcable )
xeqSC2max:= + zcable
zload1 + ztr2 + zUGmax+ ztr1min + zcable
EeqSC2max 3
ISC2max.with.load:= = 6.426 × 10 A
xeqSC2max
Appendix 1.6
ISC2max.with.load − ISC2max
kportion.of.load.contrib := ⋅ 100 = 2.175
ISC2max.with.load
3
II2 := koffset ⋅ ISC2max = 6.915 × 10 A
EUGmax 3
ISC1max:= = 8.329× 10 A
zcable + zUGmax+ ztr1min
3
II1 := koffset ⋅ ISC1max = 9.162 × 10 A
EUGmax 3
ISC3max:= = 2.471× 10 A
2zcable + zUGmax+ ztr1min + ztr3
3
III2 := koffset ⋅ ISC3max = 2.718 × 10 A
tII2 := 0.4s
EUGmax 3
ISC4max:= = 2.716 × 10 A
zcable + zUGmax+ ztr1min + ztr2
3
III1alt := koffset ⋅ ISC4max = 2.988 × 10 A
Appendix 1.7
( )
III1 := max III1, III1alt = 7.607 × 10 A
3
tII1 := 0.4s
3 EUGmin 3
ISC1min:= ⋅ = 4.605× 10 A
2 zcable + zUGmin+ ztr1max
3
Ithrough.relay.1.min := ISC1min = 4.605 × 10 A
Ithrough.relay.1.min
ksII1 := = 0.605
III1
Ithrough.relay.1.min
ksIII1main := = 5.997
IIII1
3 EUGmin 3
ISC2min := ⋅ = 3.504 × 10 A
2 2zcable + zUGmin+ ztr1max
3
Ithrough.relay.2.min := ISC2min = 3.504 × 10 A
Ithrough.relay.2.min
ksII2 := = 1.289
III2
Ithrough.relay.2.min
ksIII2main := = 7.373
IIII2
3
Ithrough.relay.1.min := ISC2min = 3.504 × 10 A
Ithrough.relay.1.min
ksIII1reser.2 := = 4.562
IIII1
EUGmin 3
ISC4 := = 1.77 × 10 A
zcable + zUGmin+ ztr1max + ztr2
3
Ithrough.relay.1 := ISC4 = 1.77 × 10 A
Appendix 1.8
Ithrough.relay.1
ksIII1reser.1 := = 2.304
IIII1
EUGmin 3
ISC3 := = 1.612× 10 A
2zcable + zUGmin+ ztr1max+ ztr3
3
Ithrough.relay.2 := ISC3 = 1.612 × 10 A
Ithrough.relay.2
ksIII2reser.1 := = 2.099
IIII1
ksII1 = 0.605
ksIII1main = 5.997
ksIII1reser.1 = 2.304
ksIII1reser.2 = 4.562
ksII2 = 1.289
ksIII2main = 7.373
ksIII2reser.1 = 2.099
Feasibility of coordinating the I>> stage of the first protection with I>> stage of
the second one:
3
III1 := koffset ⋅ III2 = 2.99 × 10 A
Ithrough.relay.1.min
ksII1 := = 1.54
III1
Evaluation of a critical length within which all the synchronously rotating ma-
chines, assumed to be connected to feeding substation 35 kV busbar, are unac-
ceptably speeded down and on the edge of loosing their transient stability:
U35 4
Ucr.ph := 0.5⋅ = 1.01 × 10 V
3
Maximal regime:
EUGmax
Ucr.ph ⋅ a⋅ zcab.per.km
a⋅ zcab.per.km + zUGmax+ ztr1min
Minimal regime:
EUGmin
Ucr.ph ⋅ b⋅ zcab.per.km
b⋅ zcab.per.km + zUGmax+ ztr1min
( )
E''AG⋅ zcable + zUGmax+ ztrmin1 + EUGmax⋅ x''AG⋅ i
( 4 )
ETh := = 2.494 × 10 + 331.715i V
zcable + zUGmax+ ztrmin1 + x''AG⋅ i
( )
zcable + zUGmax+ ztrmin1 ⋅ x''AG⋅ i
zeq := = ( 0.23 + 2.015i) Ω
zcable + zUGmax+ ztrmin1 + x''AG⋅ i
EUGmax
ithrough.p1 ( x) := if x ≤ l
zcab.per.km⋅ x + zUGmax+ ztrmin1
ETh
EUGmax− ⋅ ⎡z ⋅ ( x − l)⎤
zeq + zcab.per.km⋅ ( x − l) ⎣ cab.per.km ⎦
otherwise
zcable + zUGmax+ ztrmin1
ETh
ithrough.p2 ( x) :=
zeq + zcab.per.km⋅ x
1
Disturbed current curves quantitative dependencies are depicted here only for asynchronous
generator among all directly-coupled generating sources due to identical mathematical pathway,
except the subtransient/transient impedance calculation, which is inherent to synchronous one.
Appendix 1.11
SDGnom
IDGnom := = 329.914A
3⋅ U35
35 4
EDC := kV = 2.021× 10 V
2
3
EUGmax
i( x) :=
zUGmax+ ztr1min + zcab.per.km⋅ x
( )
ubus3 ( x) := EUGmax − i( x) ⋅ zUGmax+ ztr1min + zcable
Ubus3 := for x ∈ 5 .. 10
B ← u bus3 ( x)
x
B
Appendix 1.12
( )
I Ubus3 := for x ∈ 5 .. 10
U ← Ubus3
x
i←0
while 1
i←i+ 1
EDC − U
Z ←
i 1.1IDGnom
(
EUGmax⋅ Z + EDC⋅ zUGmax+ ztr1min + zcable
i )
Eth ←
i (zUGmax+ ztr1min + zcable) + Zi
⎡⎣Zi⋅ ( zUGmax+ ztr1min + zcable)⎤⎦
Zth ←
i ⎡⎣Zi + ( zUGmax+ ztr1min + zcable)⎤⎦
Eth
i
Ith ←
i Zth + zcab.per.km⋅ ( x − l )
i
U ← Ith ⋅ zcab.per.km⋅ ( x − l)
i
break if Z − Z < 0.0001Ω
i i −1
Z
i
(
EUGmax⋅ Z + EDC⋅ zUGmax+ ztr1min + zcable
i )
Eitr ←
(zUGmax+ ztr1min + zcable) + Zi
⎡⎣Zi⋅ ( zUGmax+ ztr1min + zcable)⎤⎦
Zitr ←
⎡⎣Zi + ( zUGmax+ ztr1min + zcable)⎤⎦
Eitr
I ←
x Zitr + zcab.per.km⋅ ( x − l )