P200D/Modeling of DER for Protection Studies

Introduction

This guide provides an overview of distributed energy resource (DER) short-circuit current characteristics and how these resources are typically modeled in commercial short-circuit simulation tools for protection studies. It covers how DER may impact protection sensitivity, reliability, and security. It is intended that this guide will be updated annually building on industry experience, modeling improvements, changes to standards, technology advances, and on-going EPRI research and lab-testing. For example, DER Inverters have a wide-range of possible short-circuit responses that will need to be considered and updated in the future as technology advances and standards and interconnection requirements are revised. The guide is split into five broad sections covering impact of interconnection requirements, synchronous and induction generators, small scale inverters (sub-50 kW), larger scale inverters (50+ kW), and research gaps. This work recognizes that inverter response models will continue to evolve with grid-support expectations, DER grounding practices and settings options. Figure 0 1 shows responses to a poll taken during a EPRI Distribution Protection Project webcast in May, 2019. The questions asked if participants modeled rotating or inverter-interfaced DER in their protection studies. In both cases approximately two thirds of the respondents said they did model DER. Some responses to the question added caveats such as only DER above a certain size (e.g. 100 kVA or 1MW) was modeled. Others noted that they were not confident in the accuracy of the inverter models, so they did not model them at all. Many commented that installed DER was growing on the system and they were looking into beginning to include these models.

Figure 0 1 Responses to poll during 2019 EPRI Webcast on DER Modeling for Protection Studies

Impact of Interconnection Requirements

Standards such as IEEE1547 [1] [2] and IEC61727 [3] or Grid Codes may set out over and under voltage ride-through. These requirements typically increase the time for which inverter DER remains connected after a fault occurs and hence for how long they will contribute short-circuit current. Synchronous generators have limited ability for ride-through. They are either excluded from such requirements, or in case of IEEE 1547 requirements, machines are not certified for interconnection. The meaning of not certified in case of induction or synchronous machines is that site-level protection will be accomplished using conventional relays. In addition to ride-through, evolving grounding practices with higher penetrations will also affect coordination. These DER response characteristics are important when studying the performance of grid backup protection. With a variety of responses some or all nearby DER may have disconnected between the initial fault but before the backup protection has operated. The topic is further complicated by inverter “momentary cessation” in response to deep voltage dips, which effectively means it may temporarily stop exporting current until the fault is cleared. These characteristics can have a marked impact on protection coordination performance, particularly for high DER penetration and DER connecting to grids with low short-circuit current levels. With the approval of the new IEEE1547:2018 standard, new DER requirements were introduced with respect to continued operation during voltage and frequency disturbances. The updated standard requires DER to meet under and over-voltage ride-through requirements to continue to export power during voltage sags and depressions. This also results in the DER contributing current during certain short circuit faults. Two requirements likely to impact protection are: The DER must continue to export power for at least 3.0 seconds while the terminal voltage falls below 88% of nominal voltage or for 2.0 seconds if the voltage falls below 65%. If the voltage falls below 65% the inverter may trip or ride-through as the discretion of the inverter vendor or operator. In continuation of the requirements from previous IEEE1547 editions, there are also under-voltage trip limits such that, by default, the DER must trip as follows: Within 0.16 seconds of the voltage falling below 45% of nominal Within 21.0 seconds should the voltage fall below 88% of nominal. While 21.0 seconds may seem like a very long time, it should be noted that the DER is still required to detect and trip within 2.0 seconds of an island forming. The effectiveness of on-board islanding detection within 2.0 seconds and considering ride-through and higher penetration levels has been identified as a concern by some utilities. Table 1 Voltage Trip Limits Requirements Defined by IEEE1547:2003 and IEEE1547:2014 [1] [2] IEEE1547-2003 IEEE 1547a:2014 Voltage Range (% nominal) Trip Time (s) Voltage Range (% Nominal) Minimum and Default Trip Time (s) Maximum Trip Time (s) V < 50.0 0.160 V < 45.0 0.160 0.160 50.0 ≤ V < 88.0 2.0 45.0 ≤ V < 60.0 1.0 11.0 110.0 < V < 120.0 1.0 60.0 ≤ V < 88.0 2.0 21.0 V ≥ 120.0 0.160 110.0 < V < 120. 1.0 13.0 V≥120.0 0.160 0.160

Table 2 Voltage Ride-Through and Trip Limits Defined by IEEE1547:2018 [4] IEEE 1547-2018 Voltage Range (% Nominal) Minimum Ride-through Minimum and Default Trip Time (s) Maximum Trip Time (s) V < 45.0 0 0.160 0.160 45.0 ≤ V < 65.0 0 1.0 11.0 65.0 ≤ V < 88.0 2 2.0 21.0 110.0 < V < 120. 0 1.0 13.0 V≥120.0 0 0.160 0.160

Table 3 Voltage Trip Limits Defined by IEC61727 [3] IEC61727 Voltage Range (% nominal) Trip Time (s) V < 50.0 0.10 50.0 ≤ V < 85.0 2.0 110.0 < V < 135.0 2.0 V ≥ 135.0 0.05

With these voltage ride-through and trip limits the DER may be expected to trip quickly (less than 10 cycles) for low impedance faults due to the low voltage on the faulted phase(s). For high impedance faults, however, the voltage will not immediately collapse. In this case the DER may contribute current into the fault for some time before it is tripped by under-voltage, its islanding protection, or grid-side anti-islanding such as direct transfer trip (DTT). Some utilities install reclosers at the DER’s point of connection for reasons such as to provide a point of isolation, for monitoring, or to provide an additional layer of protection. If such a recloser is configured with voltage or frequency protection, the IEEE1547 standard requires utilities to coordinate the settings with the ride-through requirements and trip limits of the standard. Utilities can configure voltage or frequency protection that conflict with the standard for special cases such as for worker safety, but this is only permitted on a temporary basis and should be agreed or coordinated with the regional reliability coordinator. Specifically, the standard states: “It is recommended that settings applied on Area EPS equipment conform to the voltage and frequency ride-through objectives of this standard whenever the Area EPS is in normal configuration. However, it is recognized that in certain cases Area EPS operators may need to occasionally and selectively use trip settings outside the ranges of allowable settings to accommodate worker safety practices or to safeguard distribution infrastructure while in an abnormal configuration, e.g., during automatic reconfiguration of a circuit section or temporary loss of direct transfer trip of mid- and large-scale DER. Area EPS operators should limit trip settings on Area EPS equipment that conflict with this standard to only affect those selective DER and Area EPS equipment and only for a limited period necessary to meet these worker safety and equipment protection goals. Area EPS operators should coordinate these practices with the regional reliability coordinator who may consider bulk power system impacts of affected aggregate DER capacity.” [4] Other notable requirements which exist in Grid Codes and Interconnection Requirements include: Requirement for negative sequence current in proportion to any rise in negative sequence voltage [5] Inverter-interfaced DER in Italy must ride-through out-of-phase reclosing of up to 180 degrees [6]

Model parameters for synchronous and induction generators in Protection studies In synchronous generators the electrical output power is synchronized to the rotational speed of the prime mover. A complex control system is required to ensure that the generator remains synchronized to the grid and to regulate the reactive power output. These control systems also enable synchronous generators to operate over a wide range of power factors and to regulate voltage and frequency during grid disturbances or when the generator become islanded. For these reasons synchronous generators are popular for providing on-site backup power and black-start capabilities. During short circuits, however, their current contribution can reach up to five per unit for a period of up to several seconds [7] [8]. Their short circuit current follows an exponentially damped characteristic. The current is greatest in the first few cycles after the fault occurs (subtransient time period), followed up a steady reduction up to about 1 second (transient period), before reaching its steady state short circuit current if the generator is still running after this time. The magnitude of the short circuit current can create challenges in coordinating upstream protection devices.

Induction generators are characterized by a prime mover whose rotational speed is not synchronized to the electrical frequency of their output power, but instead the prime mover is slightly greater than synchronous speed. The greater the rotational speed above synchronous speed the greater the generator’s power output. Induction generators tend to be less expensive than equivalent-sized synchronous generators as they do not need synchronizing equipment or controllers. Excitation is provided by an external reactive power source; thus, when the source of reactive power is removed induction generators are unable to produce power. For these reasons induction generators have gained popularity for small combined heat and power (CHP) generators where the application does not require black-start capability and the inability of the induction generator to sustain an island means many grids do not require dedicated anti-islanding protection or direct transfer trip schemes. Induction generators are also widely used in Type 1 and Type 2 wind turbine generators.

Induction generators can produce fault current magnitudes of up to five to eight per unit for a period of a few cycles [7] [8]. The magnitude of the contribution depends on the size of the machine, its internal impedance, and the electrical distance to the fault. The fault current magnitude decreases with increasing distance from the fault, while the duration of the contribution is a function of the electrical distance to the fault as well as the machine design parameters. The duration is typically one to two cycles for a three-phase fault and up to eight cycles for a non-symmetric fault.

Modeling Synchronous and Induction Generators in Protection Studies

These generators can be modeled using some basic generator nameplate data. The main data required are the rated machine MVA, nominal voltage, sub-transient reactance (X’’d), transient reactance (X’d), zero sequence reactance, and the value of any neutral grounding resistor if the generator is connected to the grid through a transformer which permits the flow of zero sequence current such as Ygyg [8]. Many utilities enter only the sub-transient reactance or transient reactance. Table 4 gives an example form listing the information required from synchronous generators in order to create a model for use in distribution short-circuit simulations. Most utilities request this information as part of generator connection requests. Some of the entries may not apply to certain grids and addition of transformer neutral grounding impedance may not be permitted on the utility side of the customer transformer. Table 4 Example Form For Data Required of Synchronous Generators for Short-Circuit and Protection Studies Parameter Value Units Energy Source Example: hydro, gas turbine, reciprocating engine, wind - Generator Rated Power kVA Generator Power Factor - Generator Nominal Voltage kV Generator Design Maximum Single-Step Load Pick-up % Synchronous Reactance, Xd Per unit Transient Reactance, X’d Per unit Subtransient Reactance, X’’d Per unit Negative Sequence Reactance, X2 Per Unit Zero Sequence Reactance, X0 Per unit Neutral Grounding Resistance (if installed) ohm Transformer Rating kVA Transformer Utility-Side Voltage kV Transformer Customer-Side Voltage kV Transformer Utility-Side Winding Example: Wye, Wye-Grounded, Wye-Impedance-Grounded, or Delta - Transformer Customer-Side Winding Example: Wye, Wye-Grounded, Wye-Impedance-Grounded, or Delta - Transformer Utility-Side Neutral Grounding Reactance (if installed) Ohm Transformer Utility-Side Neutral Grounding Resistance (if installed) Ohm Transformer Customer-Side Neutral Grounding Resistance (if installed) Ohm Transformer Customer-Side Neutral Grounding Reactance (if installed) Ohm

The subtransient reactance typically has a value in the range of 0.09 pu to 0.32 pu, with the transient reactance being larger. The subtransient reactance is typically applicable to the first few cycles after fault occurs, while the transient reactance is applicable up to hundreds of milliseconds to seconds. A third reactance - synchronous reactance - may also be defined if fault clearance times of several seconds are being studied; this value will be larger than the transient reactance. The zero-sequence reactance and neutral grounding resistor value are only required if the grid transformer permits the flow of zero sequence current through it, that is, if the transformer is wye-connected with grounded neutral on both windings. The zero-sequence reactance is typically much lower than the subtransient reactance. Figure 0 2 and Figure 0 3 illustrates the typical configuration of a synchronous generator in the Electrocon CAPE and Digsilent PowerFactory short-circuit simulation tools.

Figure 0 2 Example of Synchronous DER Model in Electrocon CAPE

Figure 0 3 Example of Synchronous Generator Model in PowerFactory The induction generator can be modeled in a similar manner to the synchronous generator, but the main parameters are the subtransient reactance and zero sequence reactance. If the induction generator locked-rotor (also known as blocked rotor) reactance is given, this can be used directly as the subtransient reactance. Alternatively, if the stator and rotor impedances are known the subtransient reactance can be calculated using Equation 1. The induction generator subtransient reactance typically has a value of around 0.2 pu [8]; The negative sequence reactance is usually identical to the subtransient reactance, while the zero sequence reactance is in the range of 0.03 pu to 0.05 pu depending on the air-gap. Table 5 gives an example of the data which would need to be collected in order to create short circuit models of a new induction generator. Table 6 illustrates some example data taken from real induction generators. X_st^=X_s+(X_m X_r )/(X_m+X_r ) (1) X’’st: Induction generator sub-transient reactance for use in short-circuit models Xs: Induction generator stator reactance Xr: Induction generator rotor reactance Xm: Induction generator magnetizing reactance

Table 5 Example Form For Data Required of Induction Generators for Short-Circuit and Protection Studies Parameter Value Units Energy Source Example: hydro, gas turbine, reciprocating engine, wind - Generator Rated Power kVA Generator Power Factor - Generator Nominal Voltage kV Generator Design Maximum Single-Step Load Pick-up % Stator Reactance, Xs Per unit Rotor Reactance, Xr Per unit Magnetizing Reactance, Xm Per unit Negative Sequence Reactance, X2 Per Unit Zero Sequence Reactance, X0 Per unit Neutral Grounding Resistance (if installed) ohm Transformer Rating kVA Transformer Utility-Side Voltage kV Transformer Customer-Side Voltage kV Transformer Utility-Side Winding Wye, Wye-Grounded, Wye-Impedance-Grounded, or Delta - Transformer Customer-Side Winding Wye, Wye-Grounded, Wye-Impedance-Grounded, or Delta - Transformer Impedance Per Unit Transformer Utility-Side Neutral Grounding Reactance (if installed) Ohm Transformer Utility-Side Neutral Grounding Resistance (if installed) Ohm Transformer Customer-Side Neutral Grounding Resistance (if installed) Ohm Transformer Customer-Side Neutral Grounding Reactance (if installed) Ohm

To reflect the fact that the fault current will decay to zero within 6-8 cycles, a transient time constant of around 0.1 second and large transient reactance value (e.g. 100 pu) can be used. Thus, if a fault is sustained beyond 6 cycles the simulation will consider the drop-off in induction generator fault current contribution. This will better reflect the real short circuit current behavior for protection coordination and circuit breaker rating studies. Figure 0 4 illustrates a basic induction generator short-circuit model in ASPEN OneLiner.

Figure 0 4 Example of Induction Generator Model in ASPEN OneLiner Table 6 Example Parameters taken from actual Induction Generators [9] Nominal Voltage (V) Rated Power (kVA) X’’st Slip (pu) 460 50 0.358 0.0222 208 65 0.262 0.0278 460 73 0.309 0.0133 480 101 0.274 0.0133 460 115 0.288 0.0144 460 180 0.257 0.0133 480 201 0.115 0.0144 460 335 0.288 0.0167 480 624 0.124 0.0156 2400 873 0.235 0.0122 2400 1255 0.212 0.0111

Modeling Sub-50 kW Inverters for Protection Studies

This class of inverters broadly includes single-phase inverters in the range of 190 W to 15 kW and three-phase inverters in the range of 10 kW to 50 kW. The 50 kW cut-off is not based on any official or standard classification, but instead is used for convenience as this is typically the largest single inverter deployed at commercial premises. Solar farms tend to use 100 kW up to 2 MW inverters. Note that in solar farms it is common practice for the total power rating (kW) of the solar panels attached to each inverter to be approximately 20-25% higher than the inverter power rating - this allows smaller, lower-cost inverters to be used at the expense of spilling some energy around peak production times. With inverter-interfaced energy sources the short circuit current will be fundamentally limited by the thermal capacity of the power electronics. The main inverter transistors have little thermal mass; as such, if excessive current flows through them they will heat up very quickly and fail. This can take as little as microseconds. In many of the discussions to follow it may be seen that inverter short-circuit current is dependent on the depth of the voltage sag, but in general the rms fault current is less than 1.5 pu. In some cases the fault current has been as high as 2.0 pu, but this tends to be limited to smaller scale devices. Transient currents with much greater magnitude may occur in the initial cycle after the fault occurs. The response can be inconsistent across inverters of the same size or technology as it is primarily determined by the control algorithm and hardware designed and implemented by the manufacturer as well as any user-configurable options that might be chosen. Broadly speaking, testing has shown that these inverters provide short circuit fault current of between 0 pu (immediate disconnection on detection of short-circuit) and 1.5pu for terminal faults and between 1pu and 1.5pu for grid faults. When an unbalanced fault occurs near a three-phase inverter it will tend to suppress negative sequence current and predominantly output positive sequence current after a few cycles. This is due to design considerations on the dc side of the inverter as any imbalance in output current can give rise to distortion and overvoltages there. Lab tests and fields measurements have illustrated a wide-variety of inverter responses to unbalanced faults. When an unbalanced fault occurs on the grid, the response of the inverter will depend on the current magnitude from the inverter will be the same on all three phases. The magnitude of the fault current will depend on the dip in positive sequence voltage at the inverter ac terminal and the inverter’s control mode. For this reason, inverters tend to produce greater current in response to three-phase faults than single phase ones for the same voltage dip on the faulted phase(s). Where multiple single-phase inverters are tied together this dc overvoltage issue does not arise as they have three separate dc busbars. In this case, the three-phase short circuit current passed to the distribution grid could include negative sequence current magnitude up to positive sequence current magnitude. The exact characteristics will depend on the inverter controller design. In recent years there have been extensive efforts to characterize inverters using lab-tests and use the results to feed into short-circuit models. One set of results from previous EPRI research are highlighted below as these provide a good representation of the range of behavior of such inverters [10] [11] [12] [13]. Inverter Response to Open Circuit Conditions Short-circuit tests of the inverters required the use of EPRI’s Sag Generator (PortoSag). It was used to study single-phase inverters and three-phase inverters operating at various percentages of their rated output power. Figure 0 5 illustrates the response of different single-phase inverters to a sudden open-circuit. The value of the initial peak overvoltage was noted to vary between inverters as well as the pre-fault real power output level and the point-on-wave at which the open circuit was applied. Maximum transient over-voltage values ranged between 1.5 and 2.0 pu across the eight inverters which were tested. The decay of the open-circuit voltage depended on the size of the inverter’s dc link capacitor. The maximum rms voltage ranged between 1.3 and 1.9 pu.

Figure 0 5 Response of Single-Phase Inverters to Sudden Open Circuit Conditions Inverter Response to Terminal Faults These tests involved the application of bolted short circuits at the inverter terminals for different pre-fault output powers. For each power level, the sag generator was triggered at different points on the voltage sinusoid. This ranged from 90 degrees to 255 degrees in steps of 15 degrees. Each test was performed three times to establish consistent results. The inverters tended to follow similar patterns with the event beginning with a very short transient overcurrent, followed by a sustained current that increases, before the inverter trips in accordance with IEEE1547 under-voltage settings as discussed earlier. The length of time that the inverters produced fundamental frequency current after the fault occurred was noted to vary from one cycle up to ten cycles. Differences were also noted to occur depending on the pre-fault power output level and the point-on-wave at which the terminal fault was initiated. The initial transients or spikes in over-currents in the lab-tests were noted to range between 1.4 pu and 7.5 pu, while the fundamental frequency currents ranged between 0.0 pu and 1.9 pu.

Figure 0 6 Single-phase Inverter and Three-phase Inverter Response to Terminal Faults

Inverter Response to Grid Faults

These tests were performed in much the same manner as the short-circuit tests, but the sag generator was configured to mimic faults further away from the inverter on the distribution grid. This was achieved by using the PortoSag to simulate voltage sags with 90%, 70%, and 60% retained voltage. The responses of the single-phase inverters are shown in Figure 0 7, Figure 0 8, and Figure 0 9. In these cases, the fault current magnitude can be seen to rise as a function of the magnitude of the voltage dip. In the first cases where the retained voltage was 90%, the output current rose by about 10%. When the retained voltage was 70%, however, the current rose by approximately 50% from its pre-fault value. Finally, for the retained voltage of 60% the fault current reached was double its pre-fault value.

Figure 0 7 Single-phase Inverter Response to a 10% voltage dip

Figure 0 8 Single-phase inverter response to a 30% voltage dip

Figure 0 9 Single-phase inverter response to a 50% voltage dip Some microinverters were also studied as part of this EPRI work. One interesting case was identified where the microinverters were found to attempt to restart exporting power while the voltage was still depressed. This can be seen in Figure 0 10. For a retained voltage of 60% the inverter would have been required to trip within 2.0 seconds according to IEEE1547a. Unlike the inverter tests discussed above, however, the current contribution is more erratic. This would make simulations more challenging as conventional models do not capture this behavior but given this is unusual performance in comparison to most inverters and the microinverters are so small in size there may not be much value to modeling this behavior for typical protection studies.

Figure 0 10 Unexpected Reconnection Attempts of Micro-inverter

Modeling Large-Scale Three-Phase Inverters for Protection Studies

Larger scale inverters tend to have more sophisticated control systems than their smaller-scale counter-parts. In some ways this can make their performance easier to characterize. Much detailed work has been performed by EPRI to develop short-circuit models of large-scale inverters; this has culminated in the revision and standardization of inverter models in several popular commercial short-circuit analysis tools [14] [15] [16] [17].

For the majority of Type 3 and 4 wind turbine generators, solar PV, and storage inverters the controllers are designed to only provide positive sequence real and reactive power to the grid and only respond to changes in positive sequence voltage. This can mean that they provide only limited reactive power response to unbalanced faults. This can become a concern for grids with a high proportion of inverter-interfaced energy resources as the lack of negative sequence fault current means that the voltage dip due to unbalanced faults affects a much wider area of the grid. Some grid operators have acted to mitigate this issue. In January 2017 a new Grid Code requirement was introduced in Germany [18]. This requires all new generators connected after that date to provide positive and negative sequence fault current in response to unbalanced faults. The requirement specifies that inverters increase reactive power in proportion to depth of voltage sag with a slope of 2. Type 3 DFIG Wind Turbine Generator

An induction generator forms the basis for Type 3 wind turbine generators. As shown in Figure 0 11 the stator windings of the induction generator are connected directly to the grid, while the rotor windings are connected via a back-to-back converter. Thus, the rotor mechanical speed can be above or below synchronous speed and the back-to-back converter can modulate the electrical frequency of the exciter current to ensure the rotor electrical output power frequency remains synchronized to the grid. This enables the generator to operate over a wider range of rotor speeds. This results in the induction generator operating with a slip (the percentage difference between prime mover rotational speed and synchronous speed) of up to 30% [16] During terminal faults or other nearby short circuits, overvoltages can develop on rotor windings. In order to prevent these overvoltages from causing damage to the rotor or rotor-side converter, devices known as crowbars are often used. Crowbars are fast-acting grounding switches, so when they are activated they short the busbar to ground and thus prevent overvoltages and any consequential damage. The crowbar can be installed at the rotor windings or on the dc bus. As the crowbar introduces a short-circuit to ground they must carry large fault currents, so to limit the thermal capacity requirements of the crowbar they may sometimes incorporate a series resistor or impedance to limit the magnitude of the current.

When the crowbar is activated the generator behaves like an induction generator during short circuit conditions; however, if the rotor is spinning with significant slip (up to 30%) the frequency of the output current can be much lower or much greater than grid frequency. This can yield a heavily distorted fault current waveform. For grid faults which result in less severe voltage dips at the generator ac terminal, the crowbar may be only activated briefly for a few cycles, activated for parts of each cycle (near ac voltage peaks), or not activated at all. In such cases the fault current contribution may closely match the behavior of a Type 4 full converter discussed later [19] [20]. As the crowbar has limited thermal capacity, it is often preferred to activate it for as short a period at a time as required; this is why it may only activate for part of each cycle. If the crowbar is activated too many times over a certain period, protection systems may trip the wind turbine to prevent thermal damage to the crowbar. This is notable as it can result in many turbines tripping should there be several grid faults in an area in quick succession.

Figure 0 11 Example layout of a Type 3 Wind Turbine Generator

Figure 0 12 Example of current produced by simulated DFIG wind turbine during nearby three-phase fault In practice, the fault current from Type 3 wind turbine generators will depend on the power control mode and whether the crowbar protection has been activated or not. If the crowbar has not been activated – and the crowbar may activate and deactivate many times during a fault - as is the case for distant grid faults, then the fault current will be limited to around 2 pu. For close-in or nearby faults the crowbar will be activated and the generator behaves more like an induction generator resulting in fault current of up to 5 per unit for a few cycles. For other faults the crowbar may activate for part of each cycle resulting in complex, distorted current output.

Type 4 Full Converter Wind Turbine Generator

The full converter wind turbine generator uses a permanent magnet synchronous generator connected to the grid through a back-to-back AC-DC-AC converter as illustrated in Figure 0 14. The average wind turbine size has increased greatly in recent years with 4 to 5 MW turbines becoming popular for onshore wind farms and 7-12 MW turbines common for new offshore wind turbines. Most inverter-interfaced energy sources include negative sequence current control loops which act to suppress negative sequence currents during unbalanced conditions such as single and two-phase faults. Such control loops do not eliminate negative sequence current, but instead they act to reduce it within 1-2 cycles after the unbalanced disturbance occurs. Thus, there may be some negative sequence current immediately after the fault occurs, but its magnitude will decrease over the subsequent cycles. Figure 0 13 showed this behavior by simulating a phase-to-phase fault on the grid near a Type IV full converter wind turbine generator. The positive and negative sequence currents in the bottom plot can be seen to rise immediately after the fault is triggered, but the negative sequence current reduces over the subsequent cycles. This time-variation in sequence current is not usually captured by inverter models in short circuit analysis tools.

Figure 0 13 Simulated Full Converter Wind Turbine Voltages (Top), Currents (Middle) and Sequence Currents (Bottom) for a nearby line-to-line fault. The short circuit characteristics are determined by the power control mode, the most common being: Constant power factor Reactive power control (Q-control) Voltage control Dynamic reactive power (provide additional reactive power in proportion to voltage dip magnitude)

When a fault occurs and the grid voltage dips, the inverter will continue to produce power in line with the configured power control mode [19]. Thus, if the inverter is configured in constant power factor mode, the inverter may increase its power output, but maintain the same percentage split between real and reactive power to meet the fixed power factor requirement. For inverters configured to provide dynamic reactive power response – essentially increasing reactive power output in proportion to the dip in grid voltage – they may reduce real power during a fault and instead produce much greater reactive power. This injection of reactive power will serve to limit the voltage dip during the fault and thus reduce the severity of the voltage dip generally. The priority of active or reactive power can also influence the short circuit characteristic as when the inverter reaches its maximum current output it will have to decide between limiting the production of real or reactive power. Active-power priority is sometimes applied to inverters on grids with high penetrations of DER in order to prevent faults on EHV grids (where the voltage dip will be experience across a wide area) resulting in large-scale reduction in active power output. Thus, it would prevent under-voltage disturbance turning into an under-frequency disturbance. In general, Type 4 wind turbine generators will produce up to 1.6pu positive sequence fault current in response to balanced or unbalanced faults [17]. The exact value will depend on the inverter manufacturer and model as well as the chosen power control method. The power control method may be determined by the distribution grid planner or left to the wind farm developer. The power control method may be chosen to achieve a grid voltage control objective such as reducing the probability of sustained islands (by ensuring imbalance in reactive power), maximizing real-power output, or helping regulate grid voltage.

Figure 0 14 Example layout of a Type 4 Wind Turbine Generator

Solar PV Inverter

Solar PV inverter systems can take several forms. One popular example is shown in Figure 0 15. The solar PV panels produce dc which typically goes through a dc/dc boost converter to raise the voltage ahead of a dc/ac inverter. The fault current contribution from solar PV inverters is typically observed to be in the range 1-1.6 pu [17], but the characteristics measured in lab tests and disturbance fault recorders in the field can differ significantly between inverter manufacturers and models. One of the main discrepancies is negative sequence current. Some solar PV inverters quickly damp out negative sequence currents, others damp them out within several cycles, others do not suppress them at all.

Figure 0 15 Example layout of a solar PV inverter system This time-variation in negative sequence current magnitude and phase angle can impact directional polarization elements, but this behavior will not be captured by phasor-domain short circuit simulation tools. Instead, EMT or RTDS simulations are required, but these can be time-consuming to configure and perform. As the issues with directional polarization usually apply when negative sequence voltage is used for polarization, one alternative proposed by utilities is to use zero sequence polarization instead. Zero sequence polarization may not provide ideal performance in every case, however, especially where there are mutually-coupled lines as these can result in misoperation.

Battery Energy Storage System Inverters

The inverters used to interface battery energy storage devices to the grid are quite similar to solar PV inverters in terms of their short-circuit characteristics. One difference, however is that some storage inverters may have a short-term overload capability. This capability may render the inverter capable of producing greater short-circuit currents than other inverter types. This is a topic of active research and further work is planned to investigate this phenomenon. EDP, the distribution grid operator in Portugal, performed detailed simulation and field-based testing of a 472 kVA/360 kWh battery energy storage system [21]. They found that their models under-estimated the fault current contribution of the system and that the inverters provided significant negative sequence current during unbalanced faults. The short-circuit response of inverters used for storage applications will depend on all the control parameters discussed above, but also on whether it is configured to transition from grid-connected mode to microgrid mode or whether it is configured to provide reactive power support. In its most basic configuration, the inverter should provide fault current like any other inverter when the battery is discharging (exporting power to the grid) but not produce short circuit current when the battery is charging (importing power from the grid).

As the battery inverter may continue to export real power during faults, there is also a possibility that the battery will become fully discharged during the fault. This is relatively unlikely, but a question that will be asked. If this should happen, the inverter could still inject reactive power into the fault, but in practice it would be more common for the inverter to simply cease operation and stop supplying any current.

Impact of Inverter Transformer

The inverter grid-side winding configuration and neutral grounding method will depend on the utility interconnection requirements. Utility preference for the grid-side connections depends on existing protection coordination schemes as well as the connection techniques generally used on their systems [22] [23] [24]. Selecting the correct transformer winding type will reduce the time and expense of the interconnection investigation, benefiting all concerned parties.

The transformer winding on the inverter-side is generally selected based on the inverter manufacturer requirements. Manufacturers typically specify the necessary grounding scheme for the inverter-side of the grid transformer. Bipolar and internally-isolated inverters typically support the grounding of the low-side neutral. Transformer-less inverters with grounded arrays require that the neutral (if available) be left floating. Regardless of the treatment of the transformer neutral, three-phase inverters generally have 3-wire outputs, lacking a neutral conductor. The lack of a solidly grounded neutral on the inverter is considered preferable to avoid ground currents and harmonic distortion [25] [26]. As such, the inverter network is operated on an isolated-neutral basis and the inverter will not provide a path for zero sequence current.

Modeling Inverter-Interfaced Energy Sources in Protection Studies

Conventional synchronous generators are modeled in short-circuit analysis tools as a voltage source behind an impedance. Inverters, however, exhibit the characteristics of a voltage-controlled current source, which is a challenge for tools as an iterative method for solving the case is required. For this reason, some tools provide the option to either: A) model inverters as synchronous generators whose current is clamped at some fixed value

B) model inverters as a constant current source like that shown in Figure 0 17

Both methods are easy to implement but have been found to yield inaccurate results. For instance, in one simple EPRI study carried out in the P173.9 “Impact of Renewables on System Protection” project; the currents and voltages on a 9-bus system were compared where inverters were modeled using the current-limited voltage source method in a commercial short-circuit analysis tool and again with detailed inverter models using EMTP-RV. While the fault currents were within 10 %, the grid voltages and current phase angle (and hence apparent impedance) differed significantly. This may not have a significant impact on protection performance where there are only a few inverters on a system, but the impact can become greater as the proportion of inverter-interfaced sources increases particularly when under-voltage protection is expected to trip the DER. Where inverters are modeled as constant current sources – which is common in distribution protection tools such as CYME and Synergi – the simulation will tend to over-estimate the fault current contribution from the inverter. This approach is conservative and should ensure that issues are identified, but because they highly overestimate the fault current from inverters they are also likely to incorrectly identify issues (false positives). With growing DER penetration this may result in unnecessary protection upgrades. For this reason, it is recommended that such basic models are used as a screening method to determine possible protection impacts of DER, but where issues are identified more accurate models should be used to confirm if the issues are credible or not to avoid unnecessary engineering and investment. In the absence of site-specific model parameters a current limit of 1.5pu can be used as a conservative estimate for three-phase inverters. To illustrate the variation in industry practices in assumptions for inverter current limits, Figure 0 16 shows the results of a EPRI Distribution Protection webcast poll taken in May, 2019 asking what current limit participants used for inverters in their protection studies. There are clear spikes in people using 1.1pu and 1.5pu, but there is also a noticeable spread. Where possible, it is preferable to use data from the manufacturer data sheets, although it can be challenging to get the data.

Figure 0 16 2019 Webcast Poll Results asking what assumptions participants used for inverter current limits in protection studies Table 7 Differences in Voltages and Currents for fault on a 9-Bus Test Case with Different Inverter Models

	Detailed Inverter Model using EMTP-RV	Current-Limited Voltage Source 

(Old Model) Faulted bus V (pu) I (pu) I (degrees) V (pu) I (pu) I (degrees) BUS4 0.65 1.0999 -39.39 0.394 1.1 4.5 BUS5 0.842 1.0484 -17.48 0.706 1.1 -5.3 BUS6 0.891 1.0226 -12.08 0.795 1.1 -7.8

Through the EPRI work in project P173.9 “Impact of Renewables on System Protection” as well as the work under the IEEE PSRC WG C24 “Modification of Commercial Fault Calculation Programs for Wind Turbine Generators” in recent years, new models for inverter interfaced energy sources have been developed and efforts are in progress with inverter manufacturers to validate the models and with short-circuit tool vendors such as Electrocon CAPE & ASPEN OneLiner to implement them in commercial tools [17] [27] [28] [29]. Figure 0 18 illustrates the layout of the new configuration options for inverter interfaced energy sources in Electrocon CAPE [30]. Similar functionality is also available with ASPEN OneLiner [31].

Figure 0 17 Short Circuit Model of Solar PV Inverter in CYME

Figure 0 18 New Electrocon CAPE Inverter Short Circuit Model Currently being Developed by Electrocon [30] Table 8 provides an example of the dataset which would need to be collected in order to model inverter-interfaced DER for short-circuit studies. Note that some of the entries in the table may not be required depending on utility practices; for example, utilities may not permit customer transformers to have a neutral grounding impedance and so this entry may be irrelevant. Two transformers are referenced – the inverter transformer and the grid transformer. Inverter transformers tend to be used in larger wind or solar farms with large collector networks; they step-up the ac voltage from the inverter (e.g. 400 V or 690 V) to the medium voltage of the collector network (e.g. 12.47 kV or 34.5 kV in North America or 20 kV or 33 kV in Europe and elsewhere). This collector network may be interconnected to the distribution grid via the grid transformer. For smaller wind or solar farms, the inverters feed directly into the collector network and a grid transformer connects the farm to the distribution grid. Thus, inverter transformer information may or may not be required depending on the grid and application.

The main items to note are the control mode: most DER will be configured in constant power-factor mode in order to maximize export of real power, but in other cases they may be required to export some fixed quantity of reactive power (Q-control) or to adjust their reactive power to control their terminal voltage (V-control). Note that if an inverter is configured in volt-var or similar mode, these are typically configured with long time delays (20-30 seconds) so they should not actively control reactive power output during faults or other transient events. In such cases it would be more appropriate to model them as being in Q-control mode with some reasonable estimate of the reactive power. If in doubt the distribution planners may have studied the topic and can provide more information on the appropriate mode for given DER.

Finally, inverter-interfaced DER may be configured with a dynamic voltage support mode. This mode is not usually applied to distribution systems in the US but is available as a feature with some inverters and may be used in certain circumstances. It is more commonly used on megawatt scale power parks or larger. While the measured terminal voltage stays inside a dead-band (typically 0.9-1.1pu) - the inverter follows its normal output power controller (constant power factor etc). Should the voltage fall outside of this dead-band, however, the inverter will increase its reactive output current in response to voltage dip or will reduce its reactive power output in response to overvoltages. It can thus help to dynamically regulate the grid voltage back into normal operating range and limit the extent of a voltage dip due to a fault. To model this dynamic reactive power characteristic, the dead-band voltage range parameters are required as well as the proportionality factor or slope which represents how much reactive power should be provided for a given voltage dip.

Table 8 Example Form Illustrating Data Required for Modeling Inverters in Short-Circuit and Protection Studies Parameter Value Units Total number of inverters - Energy Source Wind, Solar, Hydro, Diesel, Gas etc - Conversion Process Full converter, doubly-fed induction generator etc - Individual Inverter Rated Power kVA Inverter Nominal Voltage kV Inverter Filter Susceptance Per unit Control-Mode Constant Power Factor, Constant Q etc - Control Priority P-priority or Q priority - Grid-Side Inverter Current limit Per unit Negative Sequence Impedance Per unit Power Factor (if operating with fixed power factor) - Inverter Voltage Self-Protection Trip Settings Voltage Threshold (% nominal) Delay (s) Under-voltage Stage 1 Under-voltage Stage 2 Under-voltage Stage 3 Instantaneous Over-voltage Tripping e.g. 1.4pu e.g. 1ms Over-voltage Stage 1

Over-voltage Stage 2

Over-voltage Stage 3

Inverter Frequency Trip Settings Frequency Threshold (Hz) Delay (s) Under-frequency Stage 1 Under-frequency Stage 2 Under-frequency Stage 3 Over-frequency Stage 1 Over-frequency Stage 2 Over-frequency Stage 3 Inverter Active Anti-Islanding Control Category - Are Inverter Transformers Used? Yes/No - Inverter Transformer Rating kVA Inverter Transformer Winding Voltages e.g. 12.47kV/0.4kV kV Inverter Transformer Winding Configuration Yd, Yyg, Dd etc - Inverter Transformer Impedance Per Unit Grid Transformer Rating kVA Grid Transformer Winding Voltages 69kV/12.47kV kV Grid Transformer Winding Configuration YD, Yyg, DD etc - Grid Transformer Impedance Per Unit Grid Transformer Utility-Side Neutral Grounding Reactance (if installed) Ohm Grid Transformer Utility-Side Neutral Grounding Resistance (if installed) Ohm Grid Transformer Customer-Side Neutral Grounding Resistance (if installed) Ohm Grid Transformer Customer-Side Neutral Grounding Reactance (if installed) Ohm Grounding Transformer Type (is installed) Zig-zag, wye-grounded delta - Grounding Transformer Location Grid Side, DER side - Grounding transformer rating kVA Grounding transformer zero sequence resistance Per Unit Grounding transformer zero sequence reactance Per Unit

There is no standardized format to exchange data describing the response of those inverter-interfaced energy sources which do not continue to follow their power control mode after the terminal voltage falls outside its normal operating range. Field measurements and EPRI lab testing has shown that smaller-scale inverters – and solar PV inverters in particular – do not necessarily continue to follow their power control mode when the voltage falls outside of normal operating range. In some cases, they may also produce negative sequence current in response to unbalanced faults. These devices are more difficult to model using the methods discussed above and separate data collection approach is needed.

In an initial effort to address this need, Table 9 has been developed to capture the behavior of these devices. This is intended to serve as a starting point for the discussion with relevant stakeholders. The aim will be to ensure that inverter response to faults and system disturbances can be captured to a extent that enables reasonable accuracy in protection studies. The table would be populated with the magnitudes and relative phase angles which reasonably reflect the fundamental frequency rms current from the inverter in the period between the fault occurring and the inverter entering momentary cessation or its voltage protection tripping it offline. It will not capture more complex transient behavior such as rejection overvoltages and it is assumed that the inverter does not enter any positive-feedback active anti-islanding mode.

The left-side of the table would be populated with the inverter’s current magnitude and phase angle response to balanced, three-phase voltage dips at the inverter terminal, where the voltage remains purely positive sequence with no negative or zero-sequence component.

The right-side of the table would be populated with the inverter’s negative sequence current magnitude and phase angle to be captured in response to unbalanced voltage dips at the inverter terminal. The test procedure would require the positive and negative sequence voltages to have the same magnitude and phase angle. Table 9 Draft Format for Sharing Inverter Characteristics for Protection Studies Positive Sequence Voltage Positive Sequence Current Negative Sequence Voltage Negative Sequence Current Magnitude (pu) Magnitude (pu) Phase Angle (degrees) Magnitude (pu) Magnitude (pu) Phase Angle (degrees) 0.1 0.1 0.2 0.2 0.3 0.3 0.4 0.4 0.5 0.5 0.6 0.6 0.7 0.7 0.8 0.8 0.9 0.9 1.1 1.1 1.2 1.2 1.3 1.3

Simulation Convergence

Synchronous generators are modeled as constant voltage sources. This modeling approach is both accurate and numerically stable. Inverters are often modeled as voltage-controlled current sources, however, and the solution is more computational complex. An iterative approach is often used. There are differences between how programs such as CAPE and OneLiner implement their solutions, but at a simplified and high level it can be explained as follows: the initial short circuit simulation is performed and the voltage at the inverter terminal is compared against a Voltage-Current look-up table similar to those in Table 9. The inverter current is then adjusted based on the table and the simulation re-run; the change in current from the inverter impacts its terminal voltage, so the voltage may change slightly. The lookup table is consulted again and the current updated again and the simulation run again. This cycle continues on an iterative basis until the solution settles and the voltage at the inverter terminal is stable.

There are situations where a numerical solution is impossible. If this arises, the short circuit analysis tool may issue a convergence error. An example of where a simulation would not converge follows: The inverter is modeled with a constant power factor control. Constant power factor means the phase angle between voltage and current must be kept constant; If PF=1 then voltage and current must be kept in-phase with each other.

If there is just a transformer between the inverter and the fault, the equivalent model of the short circuit is the current source of the inverter in series with a reactor (ignoring any transformer resistance for simplicity). The fault current from the inverter is injected into the transformer reactance resulting in the voltage at the inverter terminal being almost 90 degrees out of phase with the current. This differs from the control requirement for fixed power factor, so the inverter output cannot meet the control requirement and simulation won’t converge. This non-convergence is a numerical issue due to the simplified way in which the inverter performance is captured. Actual inverter hardware performance in the field has many control loops, limiters, and self-protection which ensure that they remain stable.

If a simulation does not converge there are a few solutions to try and resolve it: Update to newer version of simulation tool. 2019 saw many updates to the inverter models and these can significantly improve numerical stability and resilience. Add load models if they exist. Adding load can help stabilize the solution, especially if feeder load is a significant proportion of inverter rating Increase convergence tolerance setting. Add very small fault arc resistance to the simulation. This is not preferred, but has been known to work. Start with small arc resistance, and increase until simulation converges Model the inverter as a current-limited voltage source. This will definitely allow the simulation to converge and inverter current magnitude should be reasonably accurate, but voltage and current phase angle will be inaccurate in most cases The models and associated convergences are discussed in further detail in EPRI Grid Protection with Variable Generation Project (P173.9) Report: “Short-Circuit Phasor Models of Converter-Based Renewable Energy Resources for Fault Studies”, EPRI report 3002010936, 2017.

Conclusions

This guide for protection with DER is part of a multi-year effort to document the short-circuit behavior of DER, the impacts on grid protection, and the modeling capabilities of commercial tools. The aim is to provide a reference to aid in performing protection coordination studies, fault investigations, and grid planning. A brief overview of the short circuit characteristics of different types of DER were presented. The short circuit models of synchronous and induction generators are well understood and can be derived from simple nameplate parameters. Inverter models are much more complex and depend on the inverter controller design, the configured power control mode, and the severity of the voltage dip due to the fault. Various standards for inverter responses, need for lab testing to confirm these response, evolving utility interconnection practices with higher penetration all indicate that this effort should be ongoing.