IMPACT OF INVERTER-BASED RESOURCES ON

IMPEDANCE-BASED PROTECTION FUNCTIONS

Mohammad R Dadash Zadeh1, Prashanth Kumar Mansani2, Daniel Ting3
1, 2
ETAP, Irvine, USA
3
ETAP, Buckinghamshire, UK
*[email protected]

Keywords: IMPEDANCE PROTECTION, INVERTER, RENEWABLE RESOURCES

Abstract
Most of the inverter-based resources (IBRs) installed do not inject or inject non-conventional negative-sequence current
during faults. This paper investigates the impact of lack of negative-sequence current on impedance-based protection
functions. This paper first introduces different methods adopted by major relay vendors to implement the impedance-based
protection function. The main focus of the paper is on core impedance measurement or comparison. Therefore, other required
functionalities such as fault type identification and directionality to realize this protection function are not discussed. Finally, a
comprehensive analysis is presented comparing the impact of IBR fault current characteristics on different implementations of
impedance-based protection function as compared to a case where IBR is replaced with an equivalent conventional
synchronous machine.

1 Introduction deliberately inject non-conventional negative sequence to

minimize certain requirements on inverter components.
Growing concerns about climate change have raised
significant attention towards using renewable energy Latest grid codes mandate renewable energy resources to
resources. Most of these renewable energy resources are inject reactive power during fault named as fault ride-through
connected to the electric network through inverters. The capability similar to conventional generators. However, these
terms converter and inverter have been used interchangeably mandates mostly have been limited to positive-sequence
in many works of literature. In this paper, the term “inverter” component. Recently, new German Code has mandated
is used to convey the same. Fault characteristics of an negative sequence current injection similar to conventional
inverter-based resource (IBR) heavily depend on its control machine during unbalanced faults [1].
logic and settings. Most of the modern inverters are designed
to inject only positive-sequence current while some also Impedance-based protection function depends on fault type
inject negative sequence. The magnitude and angle of these identification and directional functions to operate correctly.
currents play a significant role in the fault current This paper does not discuss the impact of IBR fault current
characteristics. characteristics on these functions as other papers have
investigated this subject in more detail [2]. Relay vendors
In case of type 3, wind turbine generation (WTG), an inverter have utilized different methods to implement impedance-
is used in the rotor circuit while controlling the total based protection functions especially for ground fault loops.
power/current injected to the power system. In case of fault, Next section will discuss these approaches in detail. As will
WTG stator current increases and this induces more current be presented in this paper, lack of or non-conventional
in the rotor. If this current exceeds the protection level of negative sequence current injection by IBR adversely impacts
inverter switches, a crowbar bypasses the inverter. In this some of these approaches.
case, the fault current of WTG will have the same
characteristics of conventional WTG i.e. type2. However, if Several publications have investigated the impact of IBR
crowbar is not bypassed, WTG control system is capable to fault current characteristic on distance/impedance protection
control the injected power/current like type 4. [3]-[5]. However, these investigations have been limited to
the basic implementation of impedance-based protection
In case of type 4, WTG is connected to the power system function. This paper aims to take a deeper look at various
through a full AC to DC and DC to AC converter. Similarly, techniques employed by relay vendors [6] especially those
Photovoltaic (PV) plants are interconnected to power system that adopted single-ended fault location equations to
through DC to AC converter/inverter. Fault current implement impedance-based protection for ground fault loop.
characteristic of WTG type 4 or PV generation is fully
defined by its controller logic and settings. As adopted in Variation in implementing the phase-to-phase loop is limited
most of recent inverter technologies, it is more convenient for and mostly they are equivalent. Due to the space limitation,
inverters to only inject positive sequence. Some inverters impact of IBR on phase-to-phase element will be covered in
future publications by the authors.

1
2 Impedance-Based Protection Function polarization is most commonly used in modern
microprocessor-based relays. The philosophy behind using
Impedance/Distance protection function is widely used in memory voltage is that the pre-fault and fault voltage angles
transmission protection applications due to high speed and are very close. This is true for strong sources while for weak
reliable operation. In this protection function, the positive sources, there is a considerable phase angle change in relay
sequence impedance between the relay and fault location is measured voltage. This angle shift results in a dynamic
compared directly or indirectly using voltage and current expansion that increases resistive coverage of the relay in a
measurements at the relay location against a characteristic. favourable fashion. This is the case as higher fault resistance
There have been different approaches employed by relay is expected for weak sources.
vendors to implement such protection function. Within
microprocessor-based relay technology, from the The polarizing and operating signal for the mho relay use
implementation point of view, it is possible to classify phase current and the philosophy behind the operation logic
impedance-based protection function into 1- phase- is Kirchhoff’s Voltage law. Hence, lack of negative-sequence
comparator-based and 2- impedance-measurement-based current from the IBRs do not directly affect the operation of
methods. Each method and its variation are discussed briefly the mho relay. However, since IBR is a weak source as
in the following. However, the impact of IBR fault current compared to a synchronous generator, the mho characteristic
characteristic particularly lack of negative sequence current is expands more as compared to a synchronous machine source.
highlighted.
2.1.1 Quadrilateral ground characteristic
2.1 Phase-comparator-based (PC-based) Method
In a simple form, the quadrilateral ground characteristic can
In PC-based methods, first two signals called polarizing and be implemented using four comparators as shown in Table I.
operating signals are formed out of voltage and current Here, K0 is the zero-sequence compensation factor given by
measurements and relay settings. See below. Z0/Z1-1, ZRev is the reverse reach settings, ZR is right blinder
characteristic impedance and ZL is the left blinder
𝑆𝑃𝑜𝑙 = 𝐾1 𝑉 + 𝐾2 𝐼 (1) characteristic impedance. Reverse reactance comparator can
be replaced with a phase comparator to create directional
𝑆𝑂𝑝𝑟 = 𝐾3 𝑉 + 𝐾4 𝐼 (2)
quadrilateral ground function. The quadrilateral ground
element may use phase current, superimposed phase current,
where, K1, K2, K3 and K4 are multiplying coefficients derived
zero-sequence current, or negative sequence current for
from settings, and V and I are the relay measured voltages
reactance comparator polarization. Each choice has its pros
and currents.
and cons. Phase current polarization is adversely impacted by
If the angle of operating signal with respect to polarizing the load component. Zero-sequence polarization is adversely
signal falls within a specified range, the comparator output is impacted by mutual coupling.
set to true otherwise false. Multiplying coefficients can be
Negative sequence is a very common choice in modern
selected to create different types of characteristics
protective relaying. However, the negative sequence current
representing the boundary of operation of phase comparator.
is usually small in systems with high penetration of IBRs.
In order to create an impedance-based protection function,
Lack of or non-conventional negative sequence current
one or multiple phase comparators can be combined. Since
results in mal-operation of quadrilateral element based on
only the angles of phase comparator signals are used for
negative sequence polarization. Further, negative sequence
comparison, the polarizing signal can be selected in such a
currents may be used as a polarizing signal in the directional
way to have sufficient magnitude during the fault. In the
phase comparator of quadrilateral protection function.
following, basic implementation of Mho and Quadrilateral
ground functions based on the phase comparator are Table 1 Quadrilateral characteristics comparators
presented.

2.1.1 Mho ground characteristic Characteristic Polarizing signal Operating signal

The mho characteristic is a circle formed by the polarizing Reactance j I0 or j I2 I Z+K0 I0 Z-V
and operating signals as shown below. Reverse j I0 or j I2 I Z+K0 I0 ZRev –V
reactance
𝑆𝑃𝑜𝑙 = 𝑉𝑀𝑒𝑚 (3) Right blinder Ia ZR+K0 I0 ZR I ZR+K0 I0 ZR-V
Left blinder Ia ZL+K0 I0 ZL I ZL+ K0 I0 ZL-V
𝑆𝑂𝑝𝑟 = 𝐼 𝑍 + 𝐾0 𝐼0 𝑍 − 𝑉 (4)
2.2 Impedance-Measurement-based (IM-based) Methods
where, VMem is the pre-fault voltage, Z is the user reach
setting, I0 is the zero-sequence current and V and I are the In IM-based approach, first, the positive-sequence impedance
measured phase voltage and current. In fact, there is a between relay location and fault location is estimated. Then,
dedicated comparator for each phase covering three possible this impedance is compared against the specified
phase to ground fault loops. There are several choices for the characteristic defined by user settings. Impedance-
polarizing signal [7]. However, positive-sequence memory

2
measurement-based protection functions are typically used to Use of this method to estimate impedance and compare that
implement quadrilateral characteristic. to a characteristic is equivalent to using phase comparator
based on self-polarization. For the Mho element based on
For a line-to-ground fault at distance m from the relay as memory polarization, as discussed earlier, characteristic may
shown in Fig. 1, the voltage seen by the relay is given by, expand depending on system strength. For quadrilateral
element, self-polarization is used for reactance and side
𝑉 = 𝑚𝑍1 𝐼1 + 𝑚𝑍2 𝐼2 + 𝑚𝑍0 𝐼0 + 𝑅𝑓 𝐼𝑓 (5) blinders. Hence, there is no expansion. However, choice of
negative or zero sequence current for reactance line reduces
where, Z0, Z1, and Z2 are the sequence impedances of the line, sensitivity to the load component. Based on this reasoning,
I0, I1, and I2 are the sequence currents, If is the fault current, Method I performance for a case of IBR is expected to be
and Rf is the fault resistance. Substituting the zero-sequence valid for the comparator-based impedance-based protection
compensation factor K0 (i.e. Z0/Z1-1) in (5) and simplifying element where negative sequence is not used for polarization.
yields, Thus, no dedicated evaluation has been performed for PC-
based approaches in this paper.
𝑉 = 𝑚𝑍1 (𝐼1 + 𝐾0 𝐼0 ) + 𝑅𝑓 𝐼𝑓 (6)
2.2.2 Method II
In this method, equation (6) is divided with I as shown
below:

𝑉 𝑚𝑍1 (𝐼 + 𝐾0 𝐼0 ) 𝑅𝑓 𝐼𝑓 (9)
= +
𝐼 𝐼 𝐼

Ignoring the receiving end infeed (I ≈ If) in (9) results in,

𝑉 𝑚𝑍1 (𝐼 + 𝐾0 𝐼0 ) (10)
= + 𝑅𝑓
𝐼 𝐼

Equating imaginary parts of (10) yields

𝑉 𝐼0 (11)
𝐼𝑚 { } = 𝐼𝑚 {𝑚(𝑅1 + 𝑗𝑋1 ) (1 + 𝐾0 )}
Fig. 1 Sequence network diagram for line-to-ground fault 𝐼 𝐼

Apparent impedance estimation methods used by major relay Neglecting fault resistance and equating real parts of (10)
vendors can be broadly classified into four types as outlined yields
below. 𝑉 𝐼0 (12)
𝑅𝑒 { } = 𝑅𝑒 {𝑚(𝑅1 + 𝑗𝑋1 ) (1 + 𝐾0 )}
2.2.1 Method I 𝐼 𝐼
In this method, positive-sequence impedance mZ1 is
estimated by assuming Rf as zero in (6). Therefore, the The two equations (11) and (12) can be solved to find the two
estimated impedance mZ1 is given by: unknowns mR1 and mX1. The reader may refer to [8] for more
𝑉 (7) details. Assuming I ≈ 3 I0, the approximate error in the
𝑚𝑍1 = estimated resistance can be calculated by
1 + 𝐾0 𝐼0
𝑅𝑓 (13)
Since this method neglects fault resistance in the calculation, 𝑅1𝐸𝑟𝑟𝑜𝑟 =
(1 + 𝐾𝑟 )
an error is introduced in the estimated impedance. The error
introduced in this method is given by: where, Kr = (R0-R1)/(3R1).
𝑅𝑓 𝐼𝑓 (8)
𝐸𝑟𝑟𝑜𝑟 = Using the same assumption for resistance estimation, the
1 + 𝐾0 𝐼0
approximate error in the estimated reactance becomes zero
that is advantageous as compared to Method I. This is due to
Here K0 is a complex number. Therefore, the error is
the fact that Rf is only ignored in one of the two equations.
introduced in both estimated resistance and reactance.
Similar to Method I, the fault resistance estimated by this
Considering typical values for K0, the estimated fault
method is smaller as compared to the actual value and the
resistance is considerably less than the actual value. The error
desired resistive coverage is achieved by setting right blinder
is generally approximated as 3Rf/(3+K0). To compensate for
compensating the expected error defined in (13).
the error, the desired resistive coverage is multiplied with the
correction factor when setting the relay to achieve a desired 2.2.3 Method III
resistive coverage.
This method is based on a conventional single-ended fault
location algorithm for transmission line application [9]. Here

3
the effect of fault resistance Rf is eliminated from the fault comparison, reactance is shown below of the reactance line
location by multiplying (6) by If* as shown below. implemented through phase comparator.

𝑉 𝐼𝑓∗ = 𝑚𝑍1 (𝐼 + 𝐾0 𝐼0 )𝐼𝑓∗ + 𝑅𝑓 𝐼𝑓 𝐼𝑓∗ (14) 𝑉 (20)

𝑋𝑠𝑒𝑒𝑛 = 𝐼𝑚 { }
𝐼 + 𝐾0 𝐼0
Equating the imaginary parts on both sides and simplifying
yields, Since this method is a combination of other methods, the
results from this method are skipped in this paper. The
𝐼𝑚{𝑉𝐼𝑓∗ } (15) performance of this method can be inferred by observing the
m=
𝐼𝑚{(𝑅1 + 𝑗𝑋1 )(𝐼 + 𝐾0 𝐼0 )𝐼𝑓∗ } results from Methods I and III.

From (17), the seen reactance of the relay is given by, 3 Analysis of Impedance Estimation Methods
𝐼𝑚{𝑉 𝐼𝑓∗ } In this section, the performance of the impedance estimation
𝑚𝑋1 = (16) methods is analyzed. A relay model is developed in a
𝑅
𝐼𝑚 {( 1 + 𝑗) (𝐼 + 𝐾0 𝐼0 )𝐼𝑓∗ } commercial power system analysis software with the
𝑋1
previously discussed impedance estimation methods to
To calculate the fault resistance, some relays use the perform the simulations. Synchronous generators are
following approach. Equation (6) is multiplied with connected to a strong point of the grid through a transformer
Z1*(I+K0I0)* to eliminate the effect of fault location in the and 150 kM long transmission line. The synchronous
fault resistance as shown below: generators are replaced by the IBRs of same capacity to
perform the study. To study the impact of lack of negative
𝑉 𝑍1∗ (𝐼 + 𝐾0 𝐼0 )∗ = 𝑚𝑍1 (𝐼 sequence current injection on relay performance, relative
+ 𝐾0 𝐼0 )𝑍1∗ (𝐼 + 𝐾0 𝐼0 )∗ (17) resistance and reactance seen by the relay are calculated as
+ 𝑅𝑓 𝐼𝑓 𝑍1∗ (𝐼 + 𝐾0 𝐼0 )∗ follows:

Taking the imaginary parts of (17) and simplifying yields, 𝑅𝑠𝑒𝑒𝑛 = 𝑅𝑠𝑒𝑒𝑛 𝑊𝑇𝐺 − 𝑅𝑠𝑒𝑒𝑛 𝑆𝑦𝑛 (21)

𝐼𝑚{𝑉 𝑍1∗ (𝐼 + 𝐾0 𝐼0 )∗ } 𝑋𝑠𝑒𝑒𝑛 = 𝑋𝑠𝑒𝑒𝑛 𝑊𝑇𝐺 − 𝑋𝑠𝑒𝑒𝑛 𝑆𝑦𝑛 (22)

𝑅𝑓 = (18)
𝐼𝑚{𝐼𝑓 𝑍1∗ (𝐼 + 𝐾0 𝐼0 )∗ }
where, Rseen WTG, Rseen Syn are the resistances seen by the relays
where, Rf is the estimated fault resistance. The total seen connected to IBR and synchronous generator, respectively.
resistance by relay including the faulted section of line and Xseen WTG, Xseen Syn are the reactances seen by the relays. Since
fault resistance is given by, the WTG and synchronous generator are connected to the
same system, the effect of infeed should be cancelled out.
𝑚𝑋1 However, the lack of negative-sequence injection from the
𝑅𝑠𝑒𝑒𝑛 = 𝑚𝑅1 + 𝑅𝑓 = 𝑅 + 𝑅𝑓
𝑋1 1 (19) IBR creates non-homogeneity between the sending end and
receiving end thereby effecting the relay operation.
Several options have been adopted by relay vendors to
estimate If in (16) and (18) such as 1) 3I0, 2) 1.5×I0 + 1.5× I2, To properly evaluate the effect of lack of negative-sequence
and 3) 3I2. The object of this paper is to study the injection by IBR, both cases of IBR connected to strong and
performance of the relay connected to the IBR. Since there is weak systems are studied. Further, all methods are evaluated
lack of or non-conventional negative sequence current for different fault locations along the line and 5 and 15 ohms
injection from the IBRs, the choice of If = 3×I2 will not be fault resistance.
acceptable. As can be seen from (16), the estimated reactance
is independent from estimated fault current magnitude. Fig. 3 (a) and (b) shows the difference in seen resistance and
Hence, choice of 1.5×I0 + 1.5× I2 as estimated If has no reactance in all studied methods for a case of single-phase to
considerable impact on the estimated reactance as far as I2 ground fault with 5 ohm fault resistance at different fault
injected by IBR is zero or small. However, there will be an locations. The location of the fault is changed in 10 %
error on the estimated resistance as the term If is only in the increments. As it is shown in Fig. 3 (a), lack of negative-
denominator. sequence injection by IBR adversely impacts Method 3 with
choice of (1.5×I0 + 1.5× I2) as If the most and the seen
2.2.4 Method IV resistance is changed by 20.4 ohms at the end of the line.
Impact on Method 3 with choice of 3I0 as IF is the least as
This method is a combination of Methods I and III that is expected and discussed in Section 2.2.3. Further, the error in
used by at least one of the relay vendors [10][11]. Here, the fault resistance calculation increase as fault gets closer to the
seen resistance is calculated using Method III as shown in end of line.
(20) and reactance is implemented through phase comparator.
As discussed earlier in this paper, Method I is used to For a case of 15 ohms fault resistance, as shown in Fig. 4, the
represent the equivalent impedance and in this case for result trend in Rseen, Xseen is similar to previous results. However,
the change in seen impedance is increased significantly with

4
the increased fault resistance as expected. The magnitude of 4

maximum change in seen resistance is increased from 24.4 to 2

61 ohms while in seen reactance from 3.2 to 10.7 ohms. 0

Therefore, it can be concluded that the performance of some 0 20 40 60 80 100 120

ΔXseen (ohms)
-2
of the impedance-based protection functions during high
resistance faults is severely affected by the lack of negative- -4

sequence current injection. -6

-8
25
-10
20
-12
Fault Location (%)
15
ΔRseen (ohms)

M1 M2 M3 3I0 M3 1.5 I0+ 1.5 I2

10 (b)
Fig. 4 (a) Rseen as a function of fault location for a strong system
5
with Rf = 15 ohms, (b) Xseen as a function of fault location for a
0 strong system with Rf = 15 ohms
0 20 40 60 80 100 120
-5
Fault Location (%)
As shown in Fig. 5 and Fig. 6, for a case of connection to a
weak system, it can be observed that Rseen, Xseen are
M1 M2 M3 3I0 M3 1.5 I0+ 1.5 I2 relatively lower when compared to the results from strong
(a) system. A weak system contributes smaller infeed and
1 therefore, the difference in the negative-sequence current
0.5
injection from the both ends is lower resulting in lower
0
Rseen, Xseen. Finally, it can be concluded that the impact of
0 20 40 60 80 100 120 lack of negative-sequence injection affects the performance
-0.5
of the relay and impedance-based protection function with 3I0
ΔXseen (ohms)

-1
as If is the recommended choice to increase the reliability of
-1.5 the power systems with IBRs.
-2
6
-2.5
5
-3
4
-3.5
ΔRseen (ohms)

Fault Location (%)

3

M1 M2 M3 3I0 M3 1.5 I0+ 1.5 I2

2

(b) 1

Fig. 3 (a) Rseen as a function of fault location for a strong system

0
with Rf = 5 ohms, (b) Xseen as a function of fault location for a 0 20 40 60 80 100 120

strong system with Rf = 5 ohms -1

Fault Location (%)

M1 M2 M3 3I0 M3 1.5 I0+ 1.5 I2

70

60
(a)
0.2

50 0
0 20 40 60 80 100 120
ΔRseen (ohms)

40 -0.2
ΔXseen (ohms)

30 -0.4

20 -0.6

10 -0.8

0 -1
0 20 40 60 80 100 120
-10 -1.2
Fault Location (%) Fault Location (%)

M1 M2 M3 3I0 M3 1.5 I0+ 1.5 I2

M1 M2 M3 3I0 M3 1.5 I0+ 1.5 I2

(a) (b)
Fig. 5 (a) Rseen as a function of fault location for a weak system
with Rf = 5 ohms, (b) Xseen as a function of fault location for a
weak system with Rf = 5 ohms

5
 16
affects some techniques used for impedance-based protection
14
function and Method III employing 3I0 is found to be
12
recommended choice to operate the power system reliably
10
with IBRs.
ΔRseen (ohms)

6
5 References
4

2
[1] VDE-AR-N 4120: 'Technical connection rules for
0
0 20 40 60 80 100 120
high-voltage', 2018
-2
Fault Location (%) [2] 'Impact of inverter-based resources on protection
M1 M2 M3 3I0 M3 1.5 I0+ 1.5 I2
schemes based on negative sequence components'.
(EPRI, 2019)
(a) [3] Hooshyar, A., Azzouz, M. A., El-Saadany, E. F.:
0.5
'Distance protection of lines connected to induction
0
0 20 40 60 80 100 120
generator-based wind farms during balanced faults',
-0.5 IEEE Trans. on Sustainable Energy, 2014, 5, (4), pp.
1193-1203
ΔXseen (ohms)

-1

-1.5 [4] Hooshyar, A., Azzouz, M. A., El-Saadany, E. F.:

-2
'Distance protection of lines emanating from full-scale
-2.5
converter-interfaced renewable energy power plants—
part I: problem statement', IEEE Trans. on Power Del.,
-3
2015, 30, (4), pp. 1770-1780
-3.5
Fault Location (%) [5] Hooshyar, A., Azzouz, M. A., El-Saadany, E. F.:
M1 M2 M3 3I0 M3 1.5 I0+ 1.5 I2 'Distance protection of lines emanating from full-scale
(b) converter-interfaced renewable energy power plants—
part II: solution description and evaluation', IEEE Trans.
Fig. 6 (a) Rseen as a function of fault location for a weak system
on Power Del., 2015, 30, (4), pp. 1781-1791
with Rf = 15 ohms, (b) Xseen as a function of fault location for a
[6] Kaiser, S.: 'Different representation of the earth
weak system with Rf = 15 ohms
impedance matching in distance protection relays', Proc.
4 Conclusion OMICRON User Conference, Germany, 2004, pp. 11-1
[7] Fentie, Donald D.: 'Understanding the dynamic mho
Conventional inverter-based resources do not inject or inject
distance characteristic'. Proc. 69th Annual Conference
non-conventional negative-sequence current during the faults.
for protective Relay Engineers, College Station, Texas,
Different methods named as Method I to III adopted by relay
USA, April, 2016, pp. 1-15
vendors to implement impedance-based protection function
[8] Ziegler, G.: 'Mode of operation', in 'Numerical
were discussed in detail. The effect of lack of negative-
distance protection: principles and applications' (John
sequence current injection from IBRs on these methods has
Wiley & Sons, 4th edn.), pp. 104-105
been discussed first theoretically and then through simulation
[9] Schweitzer III, E.O., Roberts, J.: 'Distance relay
study.
element design'. Proc. 46th Annual Conference for
As shown by the simulation results, all methods are found to Protective Relay Engineers, College Station, Texas,
be affected by the IBRs lack of negative sequence injection. USA, April, 1993
Method 1 shows higher error in reactance measurement as [10] 'SEL-411L Advanced Line Differential Protection,
compared to resistance measurement. In Method II, a change Automation, and Control System Instruction Manual'
in seen resistance is observed in the results while the seen (SEL, 2019)
reactance is largely unaffected. A significant change in the [11] E. O. Schweitzer, III, Jeff Roberts, “Distance Relay
seen resistance is observed in the Method III employing 1.5 Element Design”, Published in the SEL Journal of
I0+1.5 I2 and in addition, it is found that lack of negative Reliable Power, Volume 1, Number 1, July 2010.
sequence current does not affect the reactance calculation.
Method III employing 3I0 is found to be least affected by the
IBR. It can be concluded that the lack of negative sequence
injection from the inverter-based resources significantly