First Interim Report in Technical Requirements
First Interim Report in Technical Requirements
PARK MODULES
FIRST INTERIM REPORT ON TECHNICAL REQUIREMENTS
Final version | 3 May 2024
Grid forming capability of power park modules
Final version | 3 May 2024
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 2 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
TABLE OF CONTENTS
Table of Contents 3
EXECUTIVE SUMMARY 5
1 Introduction 6
5 Conclusions 33
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 3 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 4 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
EXECUTIVE SUMMARY
In the context of grid forming capability for power park modules, this report proposes a first interim
non-binding approach of detailing grid forming technical requirements for power park modules.
The report provides the basis for further discussions between the European TSOs and stakeholders
to be consolidated in the phase II. Moreover, a range of parameters has been proposed accompanied
with a proposal for compliance testing of grid forming capable power park modules.
To the best of the ENTSO-E view, and without prejudice of existing patents, the report adopts a
technology-neutral stance, providing a patent-free framework for prescribing technical
requirements on grid-forming capabilities for power park modules. This approach shall ensure
independence from specific control implementations or manufacturer patents issued before the
date of the report publication, fostering an environment conducive to innovation.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 5 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
1 Introduction
1.1 Context and background
On 19 December, 2023, the Agency for the Cooperation of Energy Regulators (ACER) submitted its
recommendations to the European Commission (EC) for amending the Network Connection Code
(CNC) related to the requirements for generators (RfG Regulation)1. This proposal is based on inputs
and feedback from different stakeholders gathered during an initial consultation process. Notably, it
includes non-exhaustive Grid-Forming (GFM) requirements for Power Park Modules (PPM) based on
the inputs from the EG on Advanced Capabilities for Grids with a High Share of PPM2.
It is worth noting that grid forming technical requirements have been recently adopted by system
operators in Great Britain (GB) and Australia. The Electricity System Operator (ESO) published the
GB GFM Best Practice Guide in April 20233 and the Guidance Notes in September 2023 following the
inclusion of GFM requirements into the GB Grid Code (GC0137 - Minimum Specification Required
for Provision of GB GFM Capability)4, and Australian Energy Market Operator (AEMO) released a
Core Requirements Test Framework in January 2024 to complement the Voluntary Specification for
Grid-forming Inverters5,6.
The technical requirements are intended to describe the GFM capabilities of PPM of type B, C and
D. If not specified further, the term ‘PPM’ refers to PPMs of type B, C and D. Any potential extension
to type A units may necessitate further analysis.
1 ACER proposes amendments to the electricity grid connection network codes | www.acer.europa.eu
2
GC-ESC EG ACPPM Report version 1.00 (windows.net)
3
download (nationalgrideso.com)
4
THE GRID CODE (nationalgrideso.com)
5
https://aemo.com.au/-/media/files/initiatives/engineering-framework/2023/grid-forming-inverters-jan
2024.pdf?la=en
6
https://aemo.com.au/-/media/files/initiatives/primary-frequency-response/2023/gfm-voluntary-spec.pdf
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 6 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
proposes their exhaustive formulation. To illustrate a possible method for verifying compliance,
Section 3 first describes the recommended test cases while Section 4 discusses the verification of
the compliance of the proposed exhaustive requirements. Finally, conclusions are drawn in Section
5.
The description of a droop-based control loop and an equivalent parametrization for a virtual
synchronous machine implementation is included in Appendix A7. The physical concept to describe
these needs and requirements including the physical response of an ideal system to disturbances is
analysed in Appendix B. Appendix C includes a discussion on measurement considerations for
selected key performance indicators (KPI). Finally, additional simulation results showing the
response of a GFM inverter to different disturbances are included in Appendix D.
1.4 Nomenclature
Relevant definitions of the parameters8 used are shown in Table 1. Table 2 presents those on the
grid side.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 7 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
𝑢
̲G Voltage amplitude and voltage angle of grid
equivalent
𝑧̲G Impedance of grid equivalent
𝑧̲SC Short circuit impedance Activated by SSC
𝑧̲Line Additional impedance between fault location and Activated by opening SSCR
unit. Also used to emulate changes of SCR and
voltage angle changes
𝑧̲Load Impedance for testing operation at very low SCR- Load activated by
values and for testing loss of last synchronous opening SL
generator. Loss of last synchronous
generator activated by
opening SG
while SL is closed
𝑧L,SS Additional load for testing small signal stability Activated by opening or
during normal operation and after loss of last closing SL,SS
synchronous generator
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 8 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
‘Where grid forming capability is specified by the relevant TSO in coordination with the relevant
system operator in accordance with NC RfG 2.0 proposal, Article Y. or defined in Articles 20, 21 and
22, a power park module shall be capable of providing grid forming capability at the connection
point, considering the sub-cycle character of the physical quantities where appropriate.’
Therefore, in NC RfG, all the connection requirements are evaluated for compliance (either by tests
or simulations) at the Point of Connection (PoC) of the PPM with the grid. For the case of GFM
capability and within the PPM’s current and energy limits, the PPM shall be capable of behaving at
the terminals of the individual unit(s) as a voltage source behind an internal impedance. Therefore,
technical requirements for the voltage source behaviour shall be defined at the terminal of the
individual unit(s), while the compliance shall be proven at the connection point.
As illustrated in Figure 1, a PPM typically consists of individual units, referred to as the Device Under
Test (DUT) in the remainder of this report (or else as generating units). It frequently includes an
internal grid, also known as the Power Collection System (PCS). As the PCS is highly project-specific
and may vary in design, defining general requirements for the voltage source behind an internal
impedance at the PoC of the PPM poses challenges.
Consequently, technical requirements of the voltage source behind an internal impedance (Thevenin
source) shall be specified by the relevant TSO, in coordination with the relevant system operators at
the terminals of the DUT, and may or may not include connection transformer(s). The terminal
voltage is then respectively denoted 𝑢 ̲DUT,MV and 𝑢 ̲DUT,LV .(see Figure 1). The facility owner needs
to ensure and prove that the PCS design maintains voltage source behaviour at the POC.
Section 2.1 first describes an equivalent circuit representation of the DUT. Subsequently, Section 2.2
elaborates on the expressions of the expected output currents. Finally, Section 2.3 proposes an
exhaustive definition of the GFM requirements in terms of an upper bound of the DUT effective
impedance.
9
ACER_Recommendation_03-2023_Annex_1. .a_NC_RfG_TC_to_original.pdf (europa.eu)
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 9 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
, ,
Figure 1: Definition of PPM PoC and unit terminals on medium and low voltage
Figure 2: DUT positive sequence controller equivalent representation for evaluating the response to grid disturbances.
For the technical requirements regarding grid forming capability, the following are considered:
1. ̲
𝑢inv represents the internal voltage amplitude, voltage phase angle and frequency of the
Thevenin source of a given DUT (known also as Generating Unit), as shown in Figure 2.
2. Converter physical parameters (𝑧̲Tr , 𝑧̲Filt ,…) are considered fixed for a given DUT.
10
This makes no assumptions on the parameters of the grid impedance 𝑧̲𝐺 which depends on local conditions and could
include impedances of the PPM collector system (PCS).
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 10 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
3. The impact of the control on the internal impedance can be represented by a controller
impedance11 𝑧̲Control . Its value can be defined at 50 Hz or in a wide band, up to 2.5 kHz, if
requested by the relevant TSO, in coordination with the relevant system operator.
4. The internal impedance of the Thevenin source is given by the sum of physical and controller
impedance (𝑧̲Control ) and will be referred to as the effective impedance 𝑧̲Eff . The effective
impedance can be defined both at 50 Hz (synchronous frequency) but also in a wide frequency
range, at least up to 2.5 kHz, if requested by the relevant TSO, in coordination with the relevant
system operator.
5. The network is also represented by a Thevenin equivalent of voltage 𝑢 ̲G and impedance 𝑧̲G .
, , , ,
, ,
,
Figure 3: DUT negative sequence controller equivalent representation for evaluating the response to grid disturbances.
𝑝DUT 𝑢𝐼𝑛𝑣
𝑖P,DUT = ≈− 𝑠𝑖𝑛(𝛿) (1)
𝑢DUT 𝑥Eff
11
𝑧̲𝐶𝑜𝑛𝑡𝑟𝑜𝑙 describes the impact of control response on the relation of voltage change at the DUT terminals to the change
of current. It does not make any assumptions on the (physical) control implementation.
12
NC RfG 2.0 describes requirements for positive sequence, as minimum requirements on the European level. For national
implementations, if negative sequence requirements are needed to be detailed, this report provides a guidance on how
this should be specified.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 11 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
𝑞DUT 1
𝑖Q,DUT = ≈ (𝑢 − 𝑢Inv ⋅ 𝑐𝑜𝑠(𝛿)) (2)
𝑢DUT 𝑥Eff DUT
with 𝛿 = 𝜑𝑈 DUT −𝜑𝑈 Inv denoting the phase difference between the terminal voltage angle 𝜑𝑈 DUT
and the internal voltage 𝜑𝑈 Inv of the DUT.
Under unbalanced conditions, the negative sequence reactive current (𝑖Q,neg ) can be approximated
by:
1
𝑖Q,neg ≈ (𝑢DUT,neg − 𝑢Inv,neg ⋅ 𝑐𝑜𝑠(𝛿neg )) (3)
𝑥Eff,neg
with 𝑢DUT,neg as the negative sequence voltage at the DUT terminal and 𝑢Inv,neg as the negative
sequence internal voltage. Typically, 𝑐𝑜𝑠(𝛿neg ) is assumed to be 1 under the assumption of no
negative sequence active current.
1. In accordance with the NC RfG proposal, both the inverter internal voltage 𝑢 ̲inv (in amplitude,
phase, and frequency) and effective impedance 𝑧̲Eff are considered constant during grid events
while current, energy or voltage limits are not reached.
2. The ratio 𝑟Eff /𝑥Eff of DUT impedance remains small13, such that the impact of the resistive part
of 𝑧̲Eff can be neglected for the description of the expected current response.
In particular, technical requirements for the following quantities should be defined with respect to
voltage amplitude or voltage phase angle changes:
13
A ratio of 𝑟𝐸𝑓𝑓 /𝑥𝐸𝑓𝑓 < 0.1 is recommended for implementation at national level.
14
Assumed to be within one cycle for the instantaneous current response to voltage angle and magnitude changes.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 12 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
This section details the non-exhaustive requirement specified in Article Y(7) (a) - (c) of the NC RfG
2.0 amendment proposal:
‘(a) Within the power park module’s current and energy limits, the power park module shall be
capable of behaving at the terminals of the individual unit(s) as a voltage source behind an internal
impedance (Thevenin source), during normal operating conditions (non-disturbed network
conditions) and upon inception of a network disturbance (including voltage, frequency, and voltage
phase angle disturbance). The Thevenin source is characterized by its internal voltage amplitude,
voltage phase angle, frequency, and internal impedance.
(b) Upon inception of a network disturbance and while the power park module capabilities and
current limits are not exceeded, the instantaneous AC voltage characteristics of the internal Thevenin
source according to paragraph (a) shall be capable of not changing its amplitude and voltage phase
angle while positive sequence voltage phase angle steps or voltage magnitude steps are occurring
at the connection point. The current exchanged between the power park module and the network
shall flow naturally according to the main generating plant and converter impedances and the
voltage difference between the internal Thevenin source and the voltage at the connection point.
(c) After inception of a network disturbance in voltage magnitude, frequency or voltage phase angle,
the following shall apply within the power park module’s capability, including current limits and
inherent energy storage capabilities of each individual unit.
(i) The relevant system operator in coordination with the TSO shall specify the temporal parameters
of the dynamic performance regarding voltage stability.
(ii) Where current limitation is necessary, the relevant system operator in coordination with the
relevant TSO may specify additional requirements regarding contribution of active and reactive
power at the point of connection.
(iii) The power park module shall be capable of stable operation when reaching the power park
module current limits, without interruption, in a continuous manner and returning to the behaviour
described in paragraph (b) as soon as the limitations are no longer active. If reaching the current
limit, the grid forming behaviour must be maintained for responses as specified in paragraph (b) for
disturbances that require the current to vary in the opposite direction of the current limitation.’
Note that, in accordance with NC RfG, Article Y, both the DUT (Generating Unit) internal voltage (in
amplitude, phase, and frequency) and the effective impedance are considered constant upon
inception of a network disturbance at the connection point and while the PPM capabilities and
current limits are not exceeded. Moreover, according to equations (1) and (2) when the
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 13 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
synchronisation mechanism15 comes into play, the output current depends on the amplitude of the
grid disturbance, namely voltage phase angle steps and/or voltage magnitude steps, which are
external variables, and the effective impedance, the only design parameter at the hand of the OEM.
For avoidance of doubt the internal Thevenin voltage source is allowed to slowly change to achieve
desired performance in terms of synchronisation and damping.
1
𝑖P,DUT,Peak ≈ − (𝑠𝑖𝑛(𝛿1 ) − 𝑠𝑖𝑛(𝛿2 )) (4)
𝑥Eff
If we consider this active current output as the minimum requirement for such an event, the
corresponding equivalent impedance can be established as an upper limit.
15
Understood as synchronising power or synchronising current according to (1) as the ability of the generating unit to
inherently exchange power with the grid as result of the variation of the terminal voltage with respect to the internal
inverter voltage.
16 More details on the justification for this value are provided in Appendix B. 5° degrees has been chosen as an example only.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 14 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
0.9
SCR=10, AngleStep=-4.99°
SCR=10, AngleStep=4.91°
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
1.9 1.95 2 2.05 2.1 2.15 2.2 2.25 2.3 2.35 2.4
Figure 4: Analytical calculation of phasor and instantaneous values of active current change with angle step by +- 5° at the grid
voltage source. See also Annex B
In practice, the variation in output current depends on the phase jump, but the proposed effective
impedance serves as the proportionality coefficient between both magnitude and angle, and
therefore, appears suitable for technical specification. It is worth noting that a similar approach was
used to define the reactive current injection requirements applied to classical Grid-Following
converter based generating units (GFL) as a function of the voltage drop.
Table 3 lists the recommended values depending on the definition of the DUT terminals to account
for transformer physical impedance, considering typical values. For comparison, the range of reactive
current gains ‘k’ currently used is included in the last column. It can be interpreted as equivalent to
1/𝑥Eff .
Table 3: Proposed range of maximal values of the effective reactance 𝒙𝑬𝒇𝒇 (at 50 Hz) for different definitions of the DUT terminals
or PPM connection point.
𝒙𝐄𝐟𝐟 values (in pu) Equivalent reactive current
boosting gain k17
Point of reference for evaluation Min value Default Max value
Low voltage terminals (Unit level) 0.17 0.25 0.27 5.9 .. 4 .. 3.7
17
This column is added for informative purposes only to relate the xeff to requirement to existing connection codes. The
reactive current gain k (di_Q/du_1) is equal to 1/x_eff.
18
The internal controller reactance should not become negative. In the case of large collector systems, higher reactance
values may be specified by the TSO.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 15 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
On the one hand, high effective impedance values may reduce the sensitivity of the active and
reactive current injection regarding the voltage angle and amplitude variation provided by the GFM
unit. On the other hand, too low effective impedance requirements may result in the high sensitivity
of the PPM output to grid disturbances, especially in strong grid conditions, where this property is
less essential. The objective of Table 3 is to propose a range of values that offer a reasonable trade-
off between these two conflicting objectives: contribution to the system strength19 and remaining in
the linear domain of the control (within the unit capability limits).
Depending on local system needs, a higher contribution from the DUT may be required. In this case,
the relevant system operator in coordination with the TSO could prescribe a lower threshold, typically
in weak grid conditions. For instance, an effective impedance limited to 0.25 pu and an active current
peak of 0.35 pu would be expected in the specific case considered in Figure 4 (a phase jump of -5° at
the DUT).
Finally, the non-exhaustive requirement specified in Article Y(7)(b) can be expressed exhaustively as
follows:
Upon inception of voltage phase angle steps or voltage magnitude steps at the PoC and while the
PPM capabilities and current limits are not exceeded, the instantaneous AC voltage characteristics of
the internal Thevenin source of individual units shall remain constant and exhibit an effective
impedance below the maximum values as define in Table 3.
1. The use of a virtual (control) impedance in the control solution is neither prescribed nor
forbidden as long as the performance-based requirements are achieved. The margin beyond
physical impedances can be understood as a tolerance, a degree of freedom left to the OEM
to optimise the overall performance of the solution.
2. This margin, in practice, also accounts for measurement processing and damping functions
(see Appendix B).
3. According to equation (2), and as an indicative example, the same requirement defines the
expected reactive current output following grid side voltage amplitude variations such that
a voltage change of Δ𝑢DUT = 5% at the DUT terminals would lead to a minimal reactive
Δ𝑢
current of 𝑥 DUT = 0.15 pu for an effective reactance 𝑥Eff = 0.33.
Eff
Regarding the dynamic response for both voltage phase angle and voltage magnitude changes:
1. A peak active current change of between 50 and 70% of the (RMS-) value calculated based
on (1) is expected as current response following a voltage angle step.
2. After a rise time (10% to 90%) of 10 ms, 90% of the expected instantaneous value of the
active / reactive current or power variation shall be reached following a grid side disturbance.
3. The decay time constant of the active power following a grid side disturbance does depend
on the effective reactance, the grid reactance and internal control settings. As demonstrated
19
Commonly referred to by the short circuit power as static definition
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 16 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
in Appendix B, it can be estimated by equation (21) In the case illustrated in Figure 4, the
decay time constant is estimated to 0.016 s. A 33% mismatch from the expected value may
be deemed acceptable as error between measured and simulated data.
4. In the event a steady state value is expected (i.e., for voltage amplitude changes) following
the disturbance, a settling time (defined as the last instant the measured value enters into a
tolerance band of +-5 %20 of the expected value) of 60 ms is expected.
A damping factor of at least 5% of the electromechanical (low frequency) active power oscillations
is recommended.
The following diagram presents an indicative illustration of the definition of rise time as duration
between 10% and 90% change from before the event to the new steady state value. The settling
time is defined as the time when the signals enter a tolerance band around the steady state value
for the last time without exiting any more.
This section details the non-exhaustive requirement as specified in Article 20 for type B:
‘5. With regard to grid forming capability type B power park modules shall fulfil the following
additional requirements in relation to grid forming capability:
20
Higher tolerances needed for DFIG (-10% + 20% in some national requirements)
21
‘synthetic inertia’ means a prescribed electrical dynamic performance provided by a PPM or an HVDC system at its
connection point with aim of emulating the equivalent dynamic effect of the inertia provided by a synchronous power-
generating module (Article 2 RfG2)
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 17 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
(a) The relevant, TSO in coordination with the relevant system operator, shall specify the contribution
to synthetic inertia. The power park module shall be capable of contributing to limiting the transient
frequency deviation under high frequency conditions. Additionally, the electricity storage module
shall be capable of contributing to limiting the transient frequency deviation under low frequency
conditions.’
‘5. With regard to grid forming capability type C power park modules shall fulfil the following
additional requirements in relation to grid forming capability:
(a) The relevant TSO, in coordination with the relevant system operator, shall specify the contribution
to synthetic inertia. The power park module shall be capable of contributing to limiting the transient
frequency deviation under high and low frequency conditions.
(b) The relevant TSO may require the provision of additional energy beyond the inherent energy
storage in coordination with the relevant.’
The change of active power due to a frequency change can be described by the ramp-up time
𝑇R,PPM .22
∆𝑃
(𝑃 ) ∆𝑝𝑝𝑢
Rated
𝑇R,PPM = = (5)
𝑑(𝑓/𝑓Rated) ) 𝑑𝑓𝑝𝑢
( ) ( )
𝑑𝑡 𝑑𝑡
The ramp-up time 𝑇R,PPM (in s), equivalent to 2H, can be used as a metric to describe the active
power change regarding to frequency changes. It is used to relate the energy which is exchanged by
the PPM at its connection points with the AC network to its rated power (Energy/Rated Power) while
the grid frequency changes.
22
𝑇𝑅,𝐷𝑈𝑇 is equivalent to the ramp-up time constant TR,SG of a conventional power plant, of the control of the DUTs,
whose effect on the inertia of the internal voltage angle of the inverter-based GFM unit corresponds to the effect of the
start-up time constant of a conventional power plant. TR,SG is the time required for a conventional power plant with the
rated power Prated to accelerate the turbine set (turbine and synchronous machine, pole pair number p) having the
moment of inertia JSG from standstill to rated speed or rated angular frequency 𝜔0 . The start-up time constant TR,SG is a
measure of the inertia moment JSG of the generating unit relative to rated power and rated frequency and it is defined
𝐽𝑆𝐺, ⋅𝜔02
as 𝑇𝑅,𝑆𝐺 =
𝑃𝑟𝑎𝑡𝑒𝑑 𝑝2
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 18 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
While the frequency changes, a PPM is expected to provide an additional active power ∆𝑃 according
to
𝑑𝑓/𝑓Rated
∆𝑃 = 𝑇R,PPM ⋅ ⋅ 𝑃Rated (6)
𝑑𝑡
As example: a 2 Hz/s df/dt and 𝑇R,PPM of 25 leads to a 1 pu active power variation from steady state
2 𝐻𝑧/𝑠
∆𝑃 = 25 ⋅ ⋅1=1
50 𝐻𝑧
𝑑𝑓/𝑓Rated
𝐸 = 𝑇R,PPM ⋅ ⋅ 𝑃Rated ⋅ Δ𝑡 (7)
𝑑𝑡
with Δ𝑡 as time window of the RoCoF evaluation. Assuming frequency limits of 47.5 Hz or 52.5 Hz
𝑑𝑓/𝑓Rated
(given in NC RfG), the term ⋅ 𝑡 is 0.05, independently of the rate of frequency change. The
𝑑𝑡
energy a PPM needs to provide is therefore:
Finally, the non-exhaustive requirement specified in Article 20(5)(a) and Article 21(5)(a) and (b) can
be expressed exhaustively as follows:
The PPM shall be capable of contributing to limiting the transient frequency deviation under high
frequency conditions by modulating active power / active current. In addition, the electricity storage
module shall be capable of contributing to limiting the transient frequency deviation under low
frequency conditions.
1. The relevant TSO, in coordination with the relevant system operator (RSO) may define
expected, or maximum and minimum values of the ramp-up time 𝑇R,PPM consistent with
Table 4 specified range.
2. PPM without intrinsic storage are required to provide negative power changes only.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 19 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
3. The active power change at the terminals of the PPM may be provided by all or a limited
number of DUT, as long as the performance criteria at the terminals of the PPM are met.
4. If the concept of a distributed virtual power plant (DVPP23) is accepted by a TSO, in
coordination with the relevant system operator, the location of the provision of the inertia
contribution may differ from the location of the PPM terminals.
1. After a rise time (10% to 90%) of 100 ms, 90% of the expected positive sequence active
current or power variation shall be reached following a grid side disturbance.
‘(ii) Where current limitation is necessary, the relevant system operator in coordination with the
relevant TSO may specify additional requirements regarding contribution of active and reactive
power at the point of connection.’
In the event the total current limits are reached, voltage source behaviour (increase of the share of
active current if the grid voltage angle decreases and increase of the share of capacitive reactive
current if the voltage magnitude decreases and vice versa) needs to be maintained.
To achieve this goal, the current magnitude should be limited without imposing a hard limit on the
underlying reactive and active current controllers. Consequently, in the event of a further voltage
drop, the ratio of reactive to active current moves toward reactive currents (more reactive current
on the expense of active current). If the voltage angle changes, the active current contribution is
increased.
23
Distributed Virtual Power Plant. Describes locally distributed producers, storage and possibly consumers with
centralised control, behaving like a single power plant.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 20 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
Using the equivalent DUT representation as shown in Figure 2, this can be achieved by increasing
the value of 𝑥Eff by modifying 𝑧̲Control . while maintaining the internal voltage 𝑢
̲Inv constant to limit
the currents within the hardware limits thereby modifying, positive sequence active (𝑖P,DUT ) and
reactive (𝑖Q,DUT ) current and negative sequence reactive current 𝑖Q,neg in a comparable manner
(current magnitude limitation).
In the event current limits are reached, the total apparent current (consisting of active and reactive
positive sequence and negative sequence reactive current shall be as close as possible to the rated
current and shall not be less than 95% of the rated current.
1. The amplitude of the inverter internal voltage 𝑢̲Inv is assumed to be constant during grid events.
The voltage angle 𝜑𝑈 Inv is only to be modified to improve damping (see also Annex A) or in the
case of voltage angle changes in the grid.
2. The negative sequence effective impedance should be equal to the positive sequence effective
impedance 𝑥Eff 24 ,but shall be within the range given by Table 3 while current limits are not
reached. The negative sequence active current 𝑖P,neg should be kept zero. If specified by the
relevant TSO, in coordination with the RSO, and for the specific case of small unbalances with a
negative sequence voltage amplitude of 𝑢DUT,neg < 3% the negative sequence reactive current
𝑖Q,neg is allowed to be zero. The RSO may specify that no negative sequence reactive current is
injected, which would be the equivalent to setting the negative sequence impedance to infinite.
Wind turbines with a doubly fed generator system may have a different negative sequence
impedance.
3. In the event the current limits of the generating unit (apparent current limit) are reached, the
generating unit should still contribute to stabilising the grid while limiting the generating unit
currents flowing to the grid. This can be achieved by increasing the effective impedance value
𝑥Eff while keeping the inverter voltage 𝑢
̲Inv of the equivalent DUT representation constant. The
actual currents may deviate from this requirement, as specified by the relevant TSO but not more
than 20 ms (one cycle) once a current limit is reached.
4. A DUT without storage may not be required to absorb active power. In the case of voltage angle
jumps, the DUT is not required to reduce active power below the minimum power limit of the
DUT.
5. For voltage variations within the range 0.85 – 1.1 pu (as defined in the NC RfG, and if the current
limit of the DUT is not reached, the active current reference 𝑖PRef is commonly calculated based
on the voltage as
𝑝Ref
𝑖PRef = (9)
𝑢DUT
24
For wind turbines with a doubly fed generator system (DFIG), the negative sequence effective reactance may be
modified differently to form the effective positive sequence reactance if this can avoid additional hardware costs.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 21 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
Modifying of (1), the active current does not have to remain constant but may be increased until
the unit current limit has been reached by modifying the inverter voltage angle.
6. After the fault clearance and both during and while leaving a current limitation mode, the DUT
shall counteract any possible overvoltage at the connection point by providing inherently the
relevant amount of reactive current.
The active and reactive currents using current magnitude limitation as a function of voltage are
shown in Figure 6. The corresponding values of effective reactance 𝑥Eff and inverter internal voltage
angle are shown in Figure 7.
1 1
Voltage at terminals uDUT in p.u.
Voltage at terminals uDUT in p.u.
0.8 0.8
1
𝑖Q ≈ (𝑢 − 𝑢Inv ⋅ 𝑐𝑜𝑠(𝛿))
𝑥Eff DUT
0.6 𝑢Inv 0.6
𝑖P ≈ − 𝑠𝑖𝑛(𝛿)
𝑥Eff
0.4 0.4
𝛿 = 𝜑𝑈DUT − 𝜑𝑈Inv
Figure 6: Active and reactive current of an inverter with Figure 7: Corresponding values of voltage angle 𝜹 (green
current magnitude limitation. line), and effective impedance 𝒙𝑬𝒇𝒇 (blue line) for an
inverter with current magnitude limitation. The internal
voltage of the inverter remains constant.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 22 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
1.2
1
Voltage at terminals uDUT in p.u.
0.8
1
𝑖Q ≈ (𝑢 − 𝑢Inv ⋅ 𝑐𝑜𝑠(𝛿))
𝑥Eff DUT
0.6 𝑢Inv
𝑖P ≈ − 𝑠𝑖𝑛(𝛿)
𝑥Eff
0.4
𝛿 = 𝜑𝑈DUT − 𝜑𝑈Inv
The requirements on current limitation (by limiting the current magnitude, not active or reactive
currents separately) are discussed in Appendix B.
Regarding the dynamic response for both voltage phase angle and voltage magnitude changes when
reaching the current capability limit:
1. The expected instantaneous reactive current in the case of a current limitation shall be within
+-10% of the value calculated as shown in Figure 8.
2. After a rise time (10% to 90%) of which shall not be more than10ms, not less than 90% of
the expected instantaneous value of the active / reactive current or power variation shall be
reached following a grid side disturbance.
3. In the event a steady state value is expected following the disturbance, a settling time
(defined as the last instant the measured or simulated value enters into a tolerance band of
+-10 % of the expected value) of 60 ms is expected.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 23 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
, ,
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 24 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
For events 1 - 5 is assumed that no current limits are reached, while for 6 - 9 the DUT may operate
at its current limit. The described test setup is supposed to reproduce the results of grid disturbances
by activating the switches, except for the frequency variation and very large angle steps. In these
cases, a variation of the grid voltage 𝑢̲G is necessary.
A practical method of organising the data related to various test cases is presented in Table 5. For
illustrative purposes it has been filled with the typical values used for compliance testing in grid
connection processes which are considered relevant for assessing GFM capability.
Note: the objective of this table is to limit the number of simulations while simultaneously a broad
range of operating conditions should be analysed to verify the performance and the compliance to
requirements. Table 5 shall be considered as the starting point for describing these tests. Below the
table, the motivation for each test case should be explained.
25
Ratio of grid short circuit power to PPM power. It can be calculated by using the reciprocal of the grid equivalent
impedance in per unit (with respect to PPM power), (𝑆𝐶𝑅 = 1/𝑧𝐺 )
26
It is recommended that this is followed by a change of network impedance (including phase angle jump) to account
for breaker opening
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 25 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
Motivation:
• Test 1.1: The underlying assumption is the loss of XX% (25% used here) of either
generation or load in the system. The initial response of the DUT should correspond to
this active power change in the grid. See also Appendix B.
• Test 1.2: In the event of reaching the current limitation, stable functioning, and seamless
transition back to voltage source mode is expected (see legal text proposal, c.iii).
• Test 1.3: In event of requested energy beyond intrinsic capabilities (asymmetrical
response to phase jumps), stable functioning and disturbance rejection is expected.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 26 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
Simulations are performed as an illustration example using a GFM controller described in Appendix
A.
This is a proposal for how to verify the requirements at the terminals of the DUT based on
simulations. The document will provide an illustration of compliance using a generic control.
For testing the compliance of the DUT response to voltage angle changes, a test setup as shown in
Figure 9 is used.
As indicated in Section 3.3, we recommend the application of +5° at the DUT terminal considering a
SCR of 10 (strong grid scenario) and two different OP. In practice, relevant TSOs may define
compliance criteria on a project-specific basis, depending on the minimal and maximal short-circuit
levels (SCL) at specific locations.
27Disclaimer: The content of this chapter aims to give recommendations for the compliance schemes of generating units. For
transmission connected PPMs a relevant discussion should take place separately
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 27 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
Figure 10. Simple grid equivalent consisting of a grid equivalent, unit MV transformer, inverter filter and control impedance for
testing voltage angle steps.
For testing the compliance of the DUT response to voltage amplitude changes, an equivalent
representation as shown in Figure 11 is recommended. Without the impedance 𝑧̲Line the short
circuit power during a fault increases significantly, which is unlikely in real grid faults.
Figure 11: Simple grid equivalent consisting of a grid equivalent, unit MV transformer, inverter filter and control impedance
for testing voltage amplitude steps in simulation or measurement.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 28 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
c) By inserting a bypass signal to the controller that changes the inverter voltage by ΔuInv .
For compliance testing, the following quantities are evaluated and compared to calculated reference
values:
For testing the compliance of the DUT response to the loss of the last synchronous generator, a
test setup as shown in Figure 12 - or an active load with constant power or constant current - can
be used.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 29 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
The loss of the last synchronous generator requires the DUT to stabilise frequency and voltage.
The loss of the last synchronous generator can be induced by opening switch SG in Figure 12.
1. Frequency;
2. Voltage.
The following test procedure is recommended:
1. Starting from a maximum active power injection of p = 0.75…1 pu the switch SG is opened
while switch SL is closed, so the DUT exclusively supplies the adjustable load.
The tests according to Table 5 are performed with an active power change of up to 50 %
and reactive power values of -0.41, 0 and -0.41. The test is considered successful if:
3. By activating SSC for 150 ms and the appropriate parametrisation of ZSC or by using an active
load, it should be evaluated if the DUT is capable of providing 60%, 80% and 100% I Reactive /
IActive,Rated.
For testing the compliance of the DUT to frequency increase, a test setup as shown in Figure 2 can
be used.
1. The active power / current before and during the frequency change;
2. The rise time and damping of the active power following the frequency change.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 30 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
For testing the compliance of the DUT response to voltage amplitude changes, a test setup as shown
in Figure 11 can be used.
1. The reactive current before and during the voltage change / fault;
2. The damping of the reactive current oscillation following the voltage change;
3. The post-fault voltage excursion following voltage recovery.
For compliance testing, the following quantities are evaluated and compared to calculated reference
values:
1. The positive sequence reactive current before and after the voltage change;
2. The negative sequence reactive current before and after the voltage change;
3. The post-fault voltage excursion following voltage recovery.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 31 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
For compliance testing, the following quantities are evaluated and compared to calculated reference
values:
1. The active power / current before and during the frequency change;
2. The rise time and damping of the active power following the frequency change.
The following tests are recommended with different parameters:
a) A voltage phase angle step followed by a frequency change. The parameters depend on the
ramp-up time 𝑇R,DUT of the inverter.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 32 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
5 Conclusions
The goal of this report is to deliver comprehensive recommendations for defining GFM requirements
for PPMs, with a focus on maintaining neutrality regarding any particular controller's
implementation. A general concept for defining GFM capabilities based on a simple voltage source
model has been used to derive exhaustive parameters for GFM capability requirements.
The results of the analytical solutions have been supported by time-domain simulations using a
generic GFM model. The task force gives examples of how the requirement can fulfilled at the level
of the DUT by using a concept of voltage source behind an effective impedance, but without
requiring any specific controller design.
The document is intended to be amended by specific test cases for the different types of grid faults.
The results of simulations based on these cases should be added to each section in section 4.
The document does not define how the performance may be assessed28.
28 The Relevant TSO, in coordination with the RSO, may decide for each test case, which of the following methods for
assessing performance may be applied.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 33 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
A relevant need of any GFM control is sufficient damping. The damping method used (phase feed-
forward damping) is well known and assumed to be free of patents.
The basic control structure of the GFM control implementation used is shown in Figure 13.
The error signal between the active power set point and the actual measured filtered active power
generates via the gain kf of the frequency droop an add-on frequency Δf, changing the voltage angle
until the error is zero under stationary conditions. The calculated power is filtered with a first order
low pass filter to give a good decoupling. To increase damping the basic frequency droop control in
expanded by an additional direct path, that acts as a feed forward term from the power error Δp
directly on the voltage phase Φinv (the circuit is highlighted in blue) via the phase feed forward
damping coefficient k_ϕ. This is generally identical to a differential action of the power error onto
the frequency.
The correspondence between droop based and VSM-based GFM control is shown in Figure 14
29 Klaes, Norbert, Nico Goldschmidt, and Jens Fortmann. 2020. "Voltage Fed Control of Distributed Power Generation Inverters
with Inherent Service to Grid Stability" Energies 13, no. 10: 2579. https://doi.org/10.3390/en13102579
30 Klaes, Norbert and Jens Fortmann. 2024. "Immunity of grid forming control without energy storage to transient changes of
grid frequency and phase" IEEE Open Access Journal of Power and Energy - Manuscript ID OAJPE-00020-2024
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 34 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
The following correspondence exists between a VSM and a droop based implementation if phase
feed forward damping is applied:
𝑇pfil = 2𝐻 ⋅ 𝑅 (10)
kf = 𝑅 (11)
𝑘ϕ = 𝐾ff ⋅ 𝑅 (12)
with 𝑘ϕ as the phase feed forward damping coefficient, 𝐾ff as the phase feed forward gain, 𝑘f as
frequency droop coefficient, 𝑇pfil as active power filter time constant, 𝑅 as damping coefficient.
The resulting control structure provides good GFM performance under the assumptions, that there
are no limitations in current and power capability. To prevent overloading and overcurrent some
additions have to be used as shown in Figure 15.
Figure 15 Additions to basic control structure for protection against overload conditions during grid.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 35 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
constant, then the active power remains constant. For this purpose, the quantities of the PLL can be
used.
It is proposed to feed forward this detected angle change onto the phase of the inverter output
voltage in the case of negative phase jumps to avoid an increase of active power. The details of
proposed addition are shown in Figure 16.
An asymmetric ramp generator within the negative phase change compensation structure is used to
allow for fast inverter phase changes in the case of negative phase changes but prohibit fast changes
in the case of positive phase changes. The rates have been limited to -500rad/s and 0.01rad/s for
negative and positive phase changes respectively. To counteract the reaction time of the PLL the
phase change is further processed by a PD transfer element with unity d.c. gain.
In the event there is an active power reserve available, the phase change detected by the PLL is
reduced by a regressive gain depending on the available active power reserve. In addition, a large
positive phase changes the nonlinear negative phase change compensation structure, which will
compensate partially for the positive angle jump to prevent excessive active power reduction and
keep the active power positive.
To prevent this behaviour, it is proposed to make the p-f-curve nonlinear. In [Klaes et al., 2020]29, it
has been shown that the optimal gain of the phase feed forward path is proportional to the droop.
Therefore, the nonlinear behaviour has been realised by recalculating the power difference Δp by a
nonlinear function to a new quantity Δp’, see Figure 15, that is progressively lower than the true
difference Δp resulting in a lower frequency response especially at high power errors.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 36 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
The non-linearity can ensure that the power corresponding to the frequency remains positive also
at the maximum limit of grid frequency. At reduced reference active power, the factor K has to be
adjusted to higher values to avoid negative power values. A suitable function K=f(p_{ref}) can be
stored in a one-dimensional table.
Under-frequency events would lead to an increase of active power due to the p-f-curve. However,
an increase of power is not available if the renewable power source is already operating at its
maximum available active power, i.e., without any derating, and has no additional energy storage.
The structure shownwill prevent overload using a fast feed forward control at negative rates of
change of frequency (RoCoF), while the GFM behaviour is retained at positive RoCoF values or at
negative values, if there is an active power reserve available.
The frequency detected by the PLL is first cleaned from line frequent components using a notch filter
(Q=1). The notch filtered frequency is further processed via of a symmetric ramp generator, which
cleans the frequency signal from transients during phase jumps by limiting the frequency change to
a maximum absolute. The function of this ramp generator is like a nonlinear glitch filter. The result
of the ramp generator is fed to a low pass filter with a switchable time constant. The long time-
constant is used, if the input is higher than the output or the active power is less than the maximum
available active power. In this case the influence on the inverter output frequency is fairly small and
the GFM behaviour is retained. In the other cases the small time-constant is used, so that the
detected frequency from the PLL is fed quiet directly onto the inverter frequency output, meaning
that the active power remains almost constant and does not increase beyond it´s critical limit.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 37 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
(𝑢G )2
𝑝≈− 𝑠𝑖𝑛(𝛿) (13)
𝑥G
with 𝛿 as voltage angle difference between 𝑉𝐺𝑟𝑖𝑑 and the voltage at the load. For small angles, a
proportional relationship between angle change and power change can be assumed (with 𝛿≈𝑠𝑖𝑛(𝛿)).
In a model representation shown in Figure 18, an opening of the switch SL2 can emulate a loss of
load and, a closing of SL2 is equivalent to a loss of generation, increasing the load for the remaining
generators.
Figure 18: Simple grid and load equivalent for estimating a voltage angle change following a load change.
The resulting – average - voltage angle change for a given change of active power is a function of the
system short circuit ratio (SCR) defined by the impedance 𝑧̲𝐺 .The actual angle power change (in %)
a specific unit experiences may be higher or lower than the average power difference experienced
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 38 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
by the grid as a whole and depends on the impedance between the unit and the ‘centre of gravity’
of the grid.
Therefore, the angle changes experienced by an individual unit following a grid event depends on
the location. DUT close to a grid event need to withstand higher voltage angle changes than more
distant ones.
Note: In the distribution system, voltage angle changes frequently result from changes in the grid
topology due to the connection or disconnection of lines and are not necessarily related to changes
of the system load or generation. The following calculations focus on events that have an impact on
the system frequency.
For a very simple grid equivalent consisting of a grid equivalent and an idealised Device Under Test
(DUT) described by an ideal voltage source (with infinite inertia H) as shown in Figure 19, the steady
state relationship between active current and voltage angle based on (1) can be approximated as
𝑖P,DUT ⋅ 𝑥G
𝛿 ≈ a cs ( ) (14)
𝑢Inv
The dynamic of the relationship between voltage angle change and power can be described as:
𝑢Inv 𝑢G −𝑡/𝜏 𝛿2 − 𝛿1 𝛿1 + 𝛿2
𝑝𝐼𝑛𝑣 (𝑡) = 𝑝𝑖𝑛𝑣2,𝑠𝑡𝑎𝑡 − 3 ⋅𝑒 ⋅ 𝑠𝑖𝑛 ( ) ⋅ 𝑠𝑖𝑛 (𝜔𝑡 + + 𝜑sc ) (15)
𝑧𝐺 2 2
Figure 19:Simple grid equivalent consisting of a grid equivalent and an idealised DUT.
With 𝛿1 and 𝛿2 as the voltage angle difference between grid and inverter voltage sources before and after the event.
For a reference parametrisation with SCR = 10 and X/R = 30, the angle change resulting from a 25% change of active
power in either positive direction (loss of load) or negative direction (loss of generation) can be calculated using (14) as
1.4°. The current change resulting from this change of power and voltage angle is shown in Figure 20.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 39 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
0.8
0.6
SCR=10, AngleStep=-4.99°
SCR=10, AngleStep=4.91°
0.4
0.2
0
1.9 1.95 2 2.05 2.1 2.15 2.2 2.25 2.3 2.35 2.4
Figure 20: Active current change by ± 25% resulting from a change of grid voltage angle.
𝑢 𝛿2 − 𝛿1
𝑖𝑑𝑖𝑛𝑣 (𝑡) = 𝑖dinv2,stat − 2 ⋅ 𝑒 −(𝑡−𝑡0 )/𝜏 ⋅ 𝑠𝑖𝑛 ( )
𝑧 2 (16)
𝛿1 + 𝛿2
⋅ 𝑠𝑖𝑛 (𝜔(𝑡 − 𝑡0 ) + + 𝜑𝑠𝑐 )
2
with a damping of
𝑟 0.0033
𝐷= ≈ = 0.033 (17)
𝑥 0.1
7.1.2 Phase angle step with grid impedance and inverter hardware
An extended simple grid equivalent consisting of a grid equivalent and DUT described by an ideal
voltage source and a unit hardware equivalent consisting of inverter filter and LV-MV transformer is
shown in Figure 21.
Figure 21: Simple grid equivalent consisting of a grid equivalent, unit MV transformer and inverter filter impedance.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 40 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
If the unit MV to LV transformer and the filter impedance are added, a change by 25% active power
results in an angle change between grid and inverter voltage sources using (14) and replacing 𝑥G by
(𝑥G + 𝑥Tr + 𝑥Filt ) of around 5° in either positive direction (loss of load) or negative direction (loss
of generation) as shown in Figure 22.
0.8
0.6
SCR=10, AngleStep=-1.44°
SCR=10, AngleStep=1.43°
0.4
0.2
0
1.9 1.95 2 2.05 2.1 2.15 2.2 2.25 2.3 2.35 2.4
Figure 22: Active current change by ± 25% resulting from a change of grid voltage angle with damping and without inverter
control.
The reference parametrisation is a SCR = 10 and X/R = 30 for the grid equivalent and a unit impedance (consisting of the
inverter filter impedance 𝑧̲Filt and the LV/MV transformer impedance 𝑧̲Tr ) of 𝑧̲DUT = 0.24/8 + j0.24, with a damping of
the oscillations defined by the grid and DUT parameters as
𝑟G + 𝑟Tr + 𝑟Filt 0.033
𝐷= = ≈ 0.1 (18)
𝑥G +𝑥Tr + 𝑥Filt 0.34
The current response shown so far is based on an ideal system with infinite inertia. The response of
a GFM-controlled unit to a voltage angle step can be described by an effective impedance 𝑧̲Eff , by
modelling the controller contribution using an additional impedance 𝑧̲Control , as shown in Figure 23.
𝑧̲Eff defines the desired response of a GFM controlled DUT to a voltage angle step (and a voltage
amplitude step as shown in the next section) in a method comparable to the response of a
synchronous generator.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 41 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
Figure 23: Simple grid equivalent consisting of a grid equivalent, unit MV transformer, inverter filter and control impedance.
Based on a GFM control implementation as described in Appendix A, the currents following a voltage
angle change will decay as shown in Figure 24.
0.9
SCR=10, AngleStep=-4.99°
0.8 SCR=10, AngleStep=4.91°
0.7
0.6
0.5
0.4
0.3
0.2
0.1
1.9 1.95 2 2.05 2.1 2.15 2.2 2.25 2.3 2.35 2.4
Figure 24: Active current change with angle step by ± 4.9° at the grid voltage source with simplified grid forming control
according to Figure 14. The yellow and pink lines show the phasor based calculation.
The decay rate, as the ‘average current’ depends on the internal control implementation. For the
control implementations shown in Figure 13 and Figure 14, the decay function (the pink and yellow
line in Figure 24) can be calculated as phasor values. Assuming 𝑢𝐺 = 𝑢𝑖𝑛𝑣 = 1, the active power
change can be described as:
with
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 42 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
1
𝛥𝑝0 = − ⋅ (𝑠𝑖𝑛 𝛿1 − 𝑠𝑖𝑛 𝛿2 ) (20)
𝑥Eff
and
𝑥Eff
𝜏= (21)
𝑘f 𝜔0
with 𝜏 as the time constant of the decay rate, 𝑥Eff as the unit effective reactance, 𝑘𝑓 as the frequency
droop (see Appendix B).
with
1
𝛥𝑖𝑃0 = − ⋅ (𝑠𝑖𝑛 𝛿1 − 𝑠𝑖𝑛 𝛿2 ) (23)
𝑥Eff
Inverter based DUTs with GFM control can provide an additional damping independently from the
value of 𝑧̲Eff by modifying the inverter voltage angle. Figure 25 shows the response of a unit with
the same angle step and the same calculated power change of 25%, but an increased damping as
described in (24).
0.9
SCR=10, AngleStep=-4.99°
0.8 SCR=10, AngleStep=4.91°
0.7
0.6
0.5
0.4
0.3
0.2
0.1
1.9 1.95 2 2.05 2.1 2.15 2.2 2.25 2.3 2.35 2.4
Figure 25: Active current change with angle step by ± 4.9° at the grid voltage source with inverter control and additional
damping.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 43 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
7.1.4 Performance criteria based on ideal and simulated response to voltage angle changes
A comparison of ideal and simulated response to voltage angle changes is shown in Figure 26. It
shows a good correlation between the simulation results and the expected response. A close-up
view directly following the angle step change is shown Figure 27 – 29.
Figure 26: Active current change following voltage angle Figure 27: Detailed view of active current change following
change by 5° at the terminals of the unit at rated power. voltage angle change by 5° at rated power
Comparison of ideal and simulated response to voltage angle
change.
Figure 28: Active current change following voltage angle Figure 29: Detailed view of active current change following
change by 5° at the terminals of the unit at 50% rated power. voltage angle change by 5° at 50% rated power.
Comparison of ideal and simulated response to voltage angle
change.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 44 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
The pink line shows the phasor calculation, the light blue the analytical solution-based instantaneous
values. The black line shows the response of the EMT-Simulation based on the control structure
shown in Appendix A.
There is a strong correlation between the peak value of the current on the damping. The EMT-
simulation shows an additional time delay and a reduced peak compared to the analytical solution.
The peak value of EMT-Simulation is at around 10 ms, which is comparable to the expected response
of a voltage source.
Recommendations
Based on the phasor calculation, an expected response (depending on grid and test system
impedance) to any given voltage angle change can be calculated.
Units without internal storage should only provide a reduction of power due to positive voltage
angle changes.
The response of a controlled voltage source to a voltage amplitude step change of 𝑢G can be
approximated by:
𝑝 𝑢𝐼𝑛𝑣
𝑖P = ≈− 𝑠𝑖𝑛(𝛿) (25)
𝑢 (𝑥G + 𝑥Eff )
and
𝑞 1
𝑖Q = ≈ (𝑢 − 𝑢Inv ⋅ 𝑐𝑜𝑠(𝛿)) (26)
𝑢 (𝑥G + 𝑥Eff ) G
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 45 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
f no current limits are reached (unconstrained operation), the reactive current 𝑖Q changes with
voltage changes (26). The active current 𝑖P does not change as a function of grid voltage (25). The
apparent current becomes very high for low voltages.
Figure 31 shows the active and reactive current response as given by (25) and (26). The slight
variation of active current above a voltage of 0.95 is the result of the active power control loop that
modifies the active current depending on the grid voltage to maintain constant active power output
(27). Some grid codes require that PPMs are capable of providing rated active power below 1 pu
rated voltage.
𝑝Ref
𝑖PRef = (27)
𝑢DUT
Figure 32 shows the corresponding voltage angle 𝛿 (green) that is modified (see (25)) to keep the
active power constant for small voltage amplitude changes.
𝛿 is controlled => p =1
for uDUT >0.95
1 1
Voltage at terminals uDUT in p.u.
Voltage at terminals uDUT in p.u.
0.8 1 0.8
𝑖Q ≈ (𝑢 − 𝑢Inv ⋅ 𝑐𝑜𝑠(𝛿))
𝑥Eff DUT
0.6 0.6
𝑢Inv
𝑖P ≈ − 𝑠𝑖𝑛(𝛿)
𝑥Eff
0.4 0.4
Figure 31: Active and reactive current of an inverter Figure 32: Corresponding values of voltage angle 𝜹 (green
behaving as voltage source if no current limits apply. line). This value is commonly adopted around rated voltage
to ensure constant active power. The effective impedance
𝒙𝑬𝒇𝒇 (blue line) and the internal inverter voltage remain
constant.
This is comparable to the behaviour of a synchronous generator. The internal voltage of the
synchronous generator stays constant, the ratio between active and reactive current changes with a
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 46 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
decrease of the grid voltage. In the event of a synchronous generator, the value of the impedance x
would decrease during the fault due to saturation effects.
If an effective reactance 𝑥Eff value of 0.33 (in pu) is assumed, the typical current limit of an inverter
(1.1 .. 1.2 pu) is exceeded once the voltage drops by more than 33% at the point of connection. For
grid following inverters, it is common practice to limit active and reactive currents independently as
a function of the remaining voltage. The resulting response to voltage drops can be highly nonlinear
and possibly unstable.
Figure 33 shows active and reactive current in the event of reactive current priority in a typical
configuration for wind turbines, the active current is not reduced to 0. Figure 34 shows the
corresponding values of voltage angle δ and the effective impedance 𝒙𝐄𝐟𝐟 of an equivalent voltage
source as shown in Figure 23. Figure 35 shows an equivalent current priority for solar parks, where
no active current is needed during voltage drops, Figure 36 the corresponding values of voltage angle
δ and the effective impedance 𝒙𝐄𝐟𝐟 respectively.
Both variants behave like a current source once the reactive current limit is reached. Any additional
voltage change no longer leads to any change of reactive current. This is a typical implementation of
grid following control today and does not meet the requirements of GFM behaviour.
1.2 1.2
1 1
Voltage at terminals uDUT in p.u.
Voltage at terminals uDUT in p.u.
0.8 1 0.8
𝑖Q ≈ (𝑢 − 𝑢Inv ⋅ 𝑐𝑜𝑠(𝛿))
𝑥Eff DUT
0.6 𝑢Inv 0.6
Operation as current source
𝑖P ≈ − 𝑠𝑖𝑛(𝛿)
𝑥Eff
Figure 33: Active and reactive current of an inverter with Figure 34: Corresponding values of voltage angle 𝜹 (green
voltage control and reactive current priority in a typical line), of an inverter with reactive current priority. The
configuration for wind turbines effective impedance 𝒙𝐄𝐟𝐟 (blue line) and the internal
inverter voltage remain constant.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 47 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
0.8 1 0.8
𝑖Q ≈ (𝑢 − 𝑢Inv ⋅ 𝑐𝑜𝑠(𝛿))
𝑥Eff DUT
0.6 0.6
𝑢Inv
0.4 𝑖P ≈ − 𝑠𝑖𝑛(𝛿) 0.4
𝑥Eff
Figure 35: Active and reactive current of an inverter with Figure 36: Corresponding values of voltage angle 𝜹 (green
voltage control and reactive current priority in a typical line), of an inverter with reactive current priority. The
configuration for solar plants. effective impedance 𝒙𝑬𝒇𝒇 (blue line) and the internal
inverter voltage remain constant.
The response of a controlled voltage source with current magnitude limitation to changes in voltage
is shown in Figure 37 for 100% rated power. Figure 38 shows the corresponding values of the voltage
angle and the effective impedance. The corresponding diagrams for 100% and 10% rated power are
shown in Figure 39 and Figure 40. This can be achieved by dynamically increasing the value of
𝑧̲Control in Figure 30, resulting in a dynamic increase of the effective reactance (𝑥Eff ) in equations
(25) and (26) respectively.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 48 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
1 1
0.8 1 0.8
𝑖Q ≈ (𝑢 − 𝑢Inv ⋅ 𝑐𝑜𝑠(𝛿))
𝑥Eff DUT
Figure 37: Active and reactive current of an inverter with Figure 38: Corresponding values of voltage angle 𝜹 (green
current magnitude limitation. line), and effective impedance 𝒙𝑬𝒇𝒇 (blue line) for an
inverter with current magnitude limitation.
1 1
Operation as voltage source
0.8 1 0.8
𝑖Q ≈ (𝑢 − 𝑢Inv ⋅ 𝑐𝑜𝑠(𝛿))
𝑥Eff DUT
0 0
-0.5 0 0.5 1 0 0.2 0.4 0.6 0.8 1
Currents in I/I active,Rated Effective reactance x Eff
Figure 39: Active and reactive current of an inverter with Figure 40: Corresponding values of voltage angle 𝜹 (green
current magnitude limitation at 100% and 10% active line), and effective impedance 𝒙𝑬𝒇𝒇 (blue line) for an
power. inverter with current magnitude limitation.
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 49 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
, , , ,
, ,
,
Under unbalanced fault conditions, the negative sequence reactive current can be described as:
1
𝑖Q,neg ≈ (𝑢 − 𝑢Inv,neg ) (28)
𝑥Eff DUT,neg
with 𝑢DUT,neg as the negative sequence voltage at the connection point and 𝑢Inv,neg as the negative
sequence inverter voltage (which is assumed to be zero). The negative sequence effective
impedance should be equal to the positive sequence impedance 𝑥Eff .
In the event that current limits are reached, 𝑥Eff . shall be increased dynamically to limit active and
reactive currents in the positive sequence and the reactive current in the negative sequence
according to (1) - (3).
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 50 of 51
Grid forming capability of power park modules
Final version | 3 May 2024
1. For types B, C and D PPM with connection points at medium or high voltage level, it is
expected that tests are performed at medium voltage level, while the unit is connected to
the medium voltage grid by a unit transformer.
2. For types B, C and D PPM, the evaluation of results may be performed at the medium voltage
connection point or at the low voltage connection point in Figure 8 at the choice of the
manufacturer.
3. The diagrams usually refer to an evaluation at the MV terminals of a PPM. If an evaluation at
the LV terminals is performed, a transformer as shown in Figure 8 upon needs to be included.
4. Results from field and test bench measurements need to be evaluated as positive and
negative sequence values according to [IEC 61400-21-1].
ENTSO-E aisbl | Rue de Spa, 8 | 1000 Brussels | [email protected] | www.entsoe.eu | @entso_e Page 51 of 51