This is a peer-reviewed, author's accepted manuscript of the following research article: Alassi, A., Ahmed, K.

, Egea Alvarez, A., & Foote, C. (Accepted/In press).

Transformer inrush current mitigation techniques for grid-forming inverters dominated grids. IEEE Transactions on Power Delivery.

Transformer Inrush Current Mitigation Techniques

for Grid-Forming Inverters Dominated Grids
Abdulrahman Alassi, Member, IEEE, Khaled H. Ahmed, Senior Member, IEEE, Agusti Egea-Alvarez, Member,
IEEE, Colin Foote, Member, IEEE

Abstract— The use of inverters-based resources (IBRs) is rising peak, and b) faster transient decay [4]. However, this technique
rapidly in power networks due to increased renewable energy requires additional equipment and increases the overall system
penetration. This requires revisiting of classical network operation cost and footprint. A recent variation aims to utilize virtual
standards. For instance, high transformer energization inrush
current has been studied extensively under the classical network damping through grid-forming inverter control (GFC) to mimic
paradigm. Whereas this paper investigates transformers’ the PIR behavior for an offshore windfarm HVDC link [5].
energization techniques in the context of inverters dominated Another recent work proposes modifying the inverter voltage
grids, where inverters with limited short-circuit current are or current references for inrush mitigation. Experimental results
expected to utilize their grid-forming capabilities for black-start. using a 1 MVA battery to energize a network show significant
Common transformer energization techniques such as controlled inrush mitigation [6]. Though, the presented analysis does not
switching and soft energization are first analyzed with a new
perspective aiming to assess their feasibility when used with grid- take the residual flux (𝜙𝑟 ) impact into account. 1
forming inverters and existing network assets. Parameters Some transformer energization techniques require direct
influencing soft energization voltage ramp-up time (𝑻𝒓𝒂𝒎𝒑 ) are modification to the inverter control to be implemented such as
investigated, and a new 𝑻𝒓𝒂𝒎𝒑 estimation framework for soft energization (SE) through a voltage ramp [7], whereas
transformer energization from IBRs is introduced. Due to the other techniques require first establishing a 1 pu voltage at the
variability of available point-on-wave circuit breakers (CBs) in IBR output, followed by inrush mitigation by controlling the
distribution networks, controlled energization using single-pole circuit breaker (CB) closing instant through controlled
and three-pole CBs is investigated for various configurations and
switching (CS) [8, 9]. Soft energization from IBRs is recently
their application limits are identified. A comprehensive case study
is then presented using a test network with multiple transformers attracting more industrial and research focus since it exploits
to benchmark the performance and requirements of each the inverters voltage control flexibility. Defining a universally
technique when the network is energized from an IBR, followed suitable ramping time is a challenging task due to the varying
by a set of practical recommendations. transformer core types, used control, and operating conditions.
Index Terms— Black-start, transformers energization, inrush Several works provide different ramp-rates, ranging from less
current, controlled switching, soft energization. than a second to tens of seconds [10, 11], with limited
justifications. CS technique has also been typically approached
I. INTRODUCTION in the literature with the assumption that the CB phases can be

C
LASSICAL transformers inrush current identification controlled independently. Although such CBs are available and
and mitigation techniques have been long presented in used with dedicated relays [12], there remains a high number of
the literature, aiming to maintain system reliability and implemented three-pole (3-PL) CBs, especially in medium and
to avoid the mis-operation of protection devices, and continue low voltage networks [13]. Since many IBRs are integrated in
being studied until today [1, 2]. The rising penetration of distribution networks and capable of participating in black-
inverters-based resources (IBR) to interface renewable energy start, industrial efforts are aiming to develop and test solutions
resources and storage devices into power networks is
necessitating a paradigm shift allowing IBRs to provide
ancillary services such as black-start. Power inverters are
known to have limited overcurrent capability (up to 1.5 pu for t im t im
short durations) [3]. Identifying suitable inrush current
mitigation techniques thus becomes a necessity in the new
paradigm rather than a reliability concern. Several inrush
current mitigation techniques have been proposed in the
literature, ranging in complexity and level of required t t
software/hardware changes. Classical techniques include Pre- (a) (b)
insertion Resistors (PIR), which adds to the transformer core Fig. 1: Visualization of transformer core saturation: (a) normal
flux damping. PIR mainly contributes to: a) reducing the inrush unsaturated operation, (b) high inrush current in saturated conditions.

Manuscript received May 28, 2022. This work is supported by Iberdrola S.A. Strathclyde, Glasgow, UK, along with Khaled Ahmed and Agusti Egea-Alvarez
as part of its innovation department research activities. Its contents are solely (email: [email protected]; [email protected]). Colin Foote
the responsibility of the authors and do not necessarily represent the official was with ScottishPower Energy Networks, Glasgow, UK. He is now with The
views of Iberdrola Group. National HVDC Center, Cumbernauld, UK (email: [email protected]).
Abdulrahman Alassi is with Iberdrola Innovation Middle East, Doha, Qatar Color versions of one or more of the figures in this article are available online
(email: [email protected]). He is also affiliated to University of at http://ieeexplore.ieee.org.
 Transformer inrush current mitigation techniques for grid-forming inverters dominated grids

that integrate CS to classical CBs. Such as using dedicated 𝑑𝜙 𝑑𝑖𝑚

𝑁 + 𝑖𝑚 𝑅1 + 𝐿1 = 𝑉𝑃 sin (𝜔𝑡 + 𝛼) (1)
point-on-wave (PoW) relays to control the 3-PL CB closing at 𝑑𝑡 𝑑𝑡
instants that could generate minimum inrush across all phases
𝑅
[14]. Implementing CS also requires measuring 𝜙𝑟 across all −𝐿𝑚𝑉𝑃 𝑐𝑜𝑠(𝜔𝑡+𝛼) 𝐿𝑚 𝑉𝑃 𝑐𝑜𝑠(𝛼) − 1𝑡
𝜙≈ + (𝜙𝑟 + )𝑒 𝐿𝑇 (2)
transformer phases. Recent works propose core pre-fluxing √𝑅12 +(𝜔𝐿𝑇 )2 √𝑅12 +(𝜔𝐿𝑇 )2
before energization to avoid this requirement [15].
The objective of this paper is to present a comprehensive
approach that aims to evaluate and identify suitable where 𝜙 is the instantaneous core flux, 𝐿𝑚 is the core
inductance, and 𝐿 𝑇 = 𝐿𝑚 + 𝐿1 , 𝑁 is the transformer turns
transformers energization techniques from IBRs under different
ratio, 𝜙𝑟 is the residual flux in the transformer core, 𝑉𝑃 is the
network conditions and measurements availability. The
energizing side winding voltage amplitude, 𝜔 is the angular
contributions of this paper are summarized as follows.
frequency and 𝛼 is the energization angle on the voltage
- Analysis of CS energization techniques from single-pole waveform. This derivation neglects the parallel shunt 𝑅𝑚
(1-PL) and 3-PL CBs, with a focus on the latter type branch for simplicity. Equation (2), similar to those reported in
because of its wider availability in distribution networks. [8, 15], presents key flux influencing factors. Though, it should
- Proposing a new SE framework that aims to answer the be noted that 𝐿𝑚 here is variable between high values in the
question of how fast a voltage ramp should be set to avoid linear region and very low values during saturation.
large inrush currents. The framework is designed to be Approximations such as piecewise expansion of this equation
compatible to large networks with multiple transformers. are thus required for detailed time-simulations in a similar
- Detailed case study of a practical network, mimicking a fashion to that presented in [17]. The analogy of flux
influencing factors in (2) can be extended to three-phase
distribution network in the United Kingdom, with multiple
transformers, considering the possible influence between
transformer units. The case study investigates energizing
phases and/or the use of single/three core transformers [8, 18].
the network from an IBR through CS and SE, showcasing
the techniques capabilities and limitations, with
A. Classical (1-PL-CB) Controlled Switching
recommendations for similar industrial applications based
on the presented analysis and findings. This technique, also known classically as Point on Wave (PoW)
energization, relies on switching the transformer in the instant
The rest of this paper is organized as follows: section II presents
when the core residual flux is equal to the ‘prospective flux’ to
an overview on controlled and soft transformer energization
eliminate the transient term impact in (2) and neutralize inrush
techniques, including 3-PL controlled switching. Then, section
current. The prospective flux (𝜙𝑝𝑟𝑜𝑝𝑠 ) is estimated by
III introduces the new 𝑇𝑟𝑎𝑚𝑝 estimation framework for soft
integrating the voltage at the transformer terminals as in (3).
energization. Section IV then covers the case study testing The CB closing scheme requirement under PoW switching is
methodology and presents the simulation results with detailed represented in (4). The optimal closing angle (𝛼𝑜𝑝𝑡 ) can be
analysis. Practical implications of the presented analysis are obtained by equating the transient term in (2) to zero and
discussed in Section V, followed by concluding remarks. solving for 𝛼 at t = 0, resulting in (5).
II. TRANSFORMER ENERGIZATION TECHNIQUES
Energizing a transformer at a random instant can lead to high 𝜙𝑝𝑟𝑜𝑠𝑝 = ∫ 𝑉𝑃 sin (𝜔𝑡) 𝑑𝑡 (3)
inrush currents driven by the non-linear transformer saturation
characteristics as illustrated in Fig. 1. When the core flux (𝜙) 𝐶𝐵 = 1 ⇔ (𝜙𝑟 = 𝜙𝑝𝑟𝑜𝑠𝑝 ) (4)
surpasses the linear-region, the magnetizing current increases 𝛼𝑜𝑝𝑡 ≈ cos −1 (−𝜙𝑟(𝑝𝑢) ) (5)
sharply even against tiny flux increments. Transformer
energization techniques aim to avoid conditions leading to the
An illustration for the CS optimal CB closing instant is
transient core operation in the saturation region. A simplified
presented in Fig. 3 for inrush elimination, where 𝜙𝑟 is initially
core flux equation of a single-phase transformer can be derived
set to −0.7 𝑝𝑢. The re-energization signal is sent at 𝑡 = 0.06 𝑠,
from the electrical model equivalent illustrated in Fig. 2, with
and the CB is closed when 𝜙𝑝𝑟𝑜𝑠𝑝 = −0.7 𝑝𝑢 @ 𝛼 ≈ 45°. In
the assumption that secondary winding is open during unloaded
energization, and that 𝑋𝐿𝑇 ≫ 𝑅1 . Such mathematical this case, the flux moves along the prospective path with no DC
derivations and simplifications make it easier to infer the flux component, and the inrush is eliminated. Detailed mathematical
influencing factors as illustrated in equations (1) and (2). derivations for this technique were introduced in [19, 20], and
expanded in different research works to include different
transformer core types and configurations [21]. When extended
to three-phase transformers, the classical CS technique requires
im independent control of the CB phases to close at their
m
corresponding optimal closing instants. As an example, the
m existence of a ∆ winding at the energizing side requires
energizing two CBs first to form a winding voltage, followed
by the third CB at the optimal instant that eliminates the inrush
Fig. 2: Single-phase transformer equivalent electrical circuit model.
current in the remaining two windings.
 Transformer inrush current mitigation techniques for grid-forming inverters dominated grids

(a) (b)
Fig. 4: Three-pole closing flux errors comparison -from (7)- between (a):
𝜙𝑅 = [0.8, 0, −0.8] 𝑝𝑢, and b) 𝜙𝑅 = [0, 0, 0] 𝑝𝑢.

Fig. 3: CS of a single-phase transformer: (a) voltage energizing point, (b)

resulting inrush current, (c) 𝜙𝑟 and 𝜙 behavior during energization.

B. Three-pole Breakers Controlled Switching

The independent control of CB phases is not widely
implemented in many distribution networks. An industrial
example is presented in Distributed ReStart UK project where
the use of a 3-PL PoW relay is planned for a large transformer Fig. 5: Three-pole closing peak inrush current at different voltage angles.
energization in a black-start scenario [22]. The use of such Each dot on the graph is a distinct power simulation.
relays implies closing the three CB phases simultaneously,
while being able to control a common PoW instant. region. Though, equation (8) can serve as a first-level
Practically, the common PoW closing time should coincide approximation to define 𝜀𝑐𝑒𝑖𝑙 around the flux value resulting in
with low prospective flux to residual flux error instants as 1 𝑝𝑢 inrush current (𝜙𝑖𝑛𝑟 ) in the transformer as in (9).
defined in equation (6) across all phases to achieve good results. 𝜀𝑐𝑒𝑖𝑙(𝑝𝑢) ≈ |𝜙𝑖𝑛𝑟(𝑝𝑢) | − 1 (9)
The ‘good results’ terminology here is relative. This is because
different transformer cores have different characteristics such Two examples are presented here to illustrate two possible
as saturation curve knee-point and air-core inductance in their extremes for this technique. In the first example, the residual
saturation region. That is to say, the same error in (6) can result flux combination is set to [0.8, 0, −0.8] 𝑝𝑢. It is seen in Fig. 4a
into two different per unit inrush current magnitudes for two that a common minimum error point exists for this combination
different transformer types. Equation (7) presents a around 𝜃𝐶𝐵 = 210° with (𝜀𝜙𝑚𝑎𝑥 @ 𝛼𝐶𝐵 ) ≈ 0.065 𝑝𝑢 in phase A
mathematical condition that is derived to approximate the and C. Closing the three-pole CB at this instant should generate
optimal CB closing instant 𝛼𝐶𝐵 (defined with respect to phase a minimum inrush current in all phases. On the other hand, a
A voltage waveform), based on minimizing simultaneously the different residual flux combination (e.g., zeros in all phases -
errors in each of the three phases. An additional constraint can demagnetized core) is observed to not have a global minimum
be added to limit the application to instances where the error points when applying (7) as illustrated in Fig. 4b, and a
maximum flux error is ceilinged to avoid excessive inrush. This common angle where the three phases errors 𝜀𝜙 have near zero
ceiling is parametrized here as 𝜀𝑐𝑒𝑖𝑙 to reflect the highlighted values is missing since (𝜀𝜙𝑚𝑎𝑥 @ 𝛼𝐶𝐵 ) ≈ 0.86 𝑝𝑢. Many
variability in transformer core materials. transformers generate high inrush currents when the peak flux
𝜀𝜙 = 𝜙𝑟 − 𝜙𝑝𝑟𝑜𝑠𝑝 (6) is around 1.8 𝑝𝑢. Other residual flux combinations between the
two reported extremes will exhibit different flux error
𝛼𝐶𝐵 ≈ 𝛼(min(|𝜀𝜙𝑎 | + |𝜀𝜙𝑏 | + |𝜀𝜙𝑐 |)) ∶ max (𝜀𝜙 < 𝜀𝑐𝑒𝑖𝑙 ) (7) behaviors, influenced by the existence of common zero (or
near-zero) crossing points in all phases.
Assuming a low damping factor for the energized transformer,
Simulations are performed to confirm these theoretical
then equation (2) can be used to approximate the worst-case
predictions, using the standard MATLAB/Simulink
peak flux deviation beyond 1 pu due to the flux error. This is
transformer model with default saturation characteristics (see
done by setting the time-dependent term in (2) to
Table I) from an ideal voltage source. Here, applying (9) for the
𝜙𝑝𝑘(𝑝𝑢) (𝑖. 𝑒. , 1 𝑝𝑢), and substituting the transient term with the
transformer model in Table I results in 𝜀𝑐𝑒𝑖𝑙 estimate around
per-unit peak flux error from (6) at 𝛼𝐶𝐵 . Assuming this error is 0.52 pu (above the first case recorded maximum error, and
0.5 𝑝𝑢, then the worst-case flux peak occurs around 1.5 𝑝𝑢 below that from the demagnetized core case). The results of
according to (8). worst-case inrush current scenarios for 0 < 𝛼𝐶𝐵 < 360° are
𝜙𝑝𝑢 (𝑡𝑝𝑘 ) ≈ 𝜙𝑝𝑘(𝑝𝑢) + 𝜀𝜙(𝑝𝑢) ≈ 1 + |𝜀𝜙(𝑝𝑢) | (8) presented in Fig. 5 for both the residual flux combinations
presented here, where each dot represents a separate simulation.
Practically, the peak flux is less than the value set in (8), to a As expected, the first residual flux sequence [0.8, 0, −0.8] 𝑝𝑢
degree that depends on the circuit damping factor in saturation results into a near-elimination of inrush current around 𝛼𝐶𝐵 =
 Transformer inrush current mitigation techniques for grid-forming inverters dominated grids

210°. Whereas the second residual flux sequence [0, 0, 0]

Breakers Operation at Tx Energization Instant
results into a pattern of inrush currents that has a global
minimum of 1.63 pu, signifying a violation of the transformer PoWa 3P-PoW
rating throughout the complete 360° range for this combination
using the standard MATLAB/Simulink model. The same test
was performed on three-core and single core transformers in PoWb
wye and delta configuration with similar results. This
consistency is expected since all phases are energized PoWc
simultaneously. This demonstration serves to illustrate that
although 3-PL CS technique can be quite effective in many
Independent Three-Pole Soft Energization
cases when access to independent phase CBs is not available, it Controlled Controlled (SE)
can still lead to significant inrush currents under other residual Switching (CS) Switching
flux combinations for the whole PoW angles range. This can Grid-Forming IBR Control Requirements
cause more significant issues when combined with inverters
limited overcurrent capabilities. Pref mp
ω i
1/s
dω

Generation
B. Soft Energization Pmeas

PWM
ωref
When applying this technique, the transformer is energized Vref Options
V
PI
through a ramping voltage between 0 and 1 pu in a period equal
CS: 1 pu
to 𝑇𝑟𝑎𝑚𝑝 . Very fast ramps can still cause significant inrush Vmeas
SE: Vramp
currents, while very slow ramps extend the time during which
protection equipment may lack an ability to detect faults due to Fig. 7: Breakers and GFC requirements for each energization technique.
the voltage being too low. Selecting an appropriate 𝑇𝑟𝑎𝑚𝑝 thus
becomes an important task. The applied input voltage across which illustrates the absolute peak envelope of both flux and
ramping and steady-state stages is defined as in (10). inrush current waveforms for phase A for these different 𝑇𝑟𝑎𝑚𝑝
𝑡
values. Clearly, the peak flux is reduced as the ramping time
𝑉
𝑇𝑟𝑎𝑚𝑝 𝑃
sin (𝜔𝑡 + 𝛼) ∶ 𝑡 < 𝑇𝑟𝑎𝑚𝑝 increases due to the damping effect of the transient term while
𝑣𝑝 = { (10) the voltage takes longer time to build-up. This results in reduced
𝑉𝑃 sin (𝜔𝑡 + 𝛼) ∶ 𝑡 ≥ 𝑇𝑟𝑎𝑚𝑝
inrush current magnitude, peaking at 𝑡 = 𝑇𝑟𝑎𝑚𝑝 , before both
Numeric simulation tools such as MATLAB or PSCAD can be flux and current continue decaying until the transient term
used to visualize the ramping voltage impact on the flux and impact is vanished. Although Fig. 6 accurately demonstrates the
inrush current behavior. During the ramp, two opposite actions trends during SE, the results orders-of-magnitude cannot be
are observed: flux building-up due to the voltage ramp, and the generalized to all transformer models. However, the key
transient flux exponential decay by the damping factor. To influencing factors can be generalized. These are observed to
visually demonstrate the behavior during SE, the default be: a) residual flux, b) source-to-transformer impedance, c)
MATLAB/Simulink transformer is used with a residual flux transformer core characteristics, and d) ramping time.
combination of [0.8, 0, −0.8] 𝑝𝑢 in a 𝑌𝑔 − 𝑌𝑔 configuration.
C. Energization Techniques Benchmark for IBRs
The transformer is connected to an ideal voltage source to
measure the direct soft energization impact. The energization is The transformer energization techniques covered in this paper
repeated four times at different ramp-rates with 𝑇𝑟𝑎𝑚𝑝 = vary significantly in their operation and requirements. For that,
a brief high-level comparison for their requirements from GFC
1 𝑠, 2 𝑠, 5 𝑠 𝑎𝑛𝑑 10 𝑠. The results are summarized in Fig. 6,
and CBs points of view are presented here. The voltage control
reference for soft energization should be modified to
incorporate and track a ramping reference, whereas CS requires
establishing a 1 pu voltage at the transformer CBs input
terminals. Fig. 7 summarizes the transformer CBs control
arrangements for the different techniques, in addition to the
GFC variations using a basic droop loop as a benchmark, based
on equations (11)-(13).

𝑉𝐺𝐹𝐶 = 𝑉 ∠ 𝛿𝑖 (11)
𝐾𝑖
𝑉 = (𝑉𝑟𝑒𝑓 − 𝑉𝑚𝑒𝑎𝑠 )(𝐾𝑝 + ) (12)
𝑠
1
𝛿 = (𝜔𝑟𝑒𝑓 + 𝑚𝑝 (𝑃𝑟𝑒𝑓 − 𝑃𝑚𝑒𝑎𝑠 )) (13)
𝑠

III. NEW SOFT ENERGIZATION FRAMEWORK

The framework presented here for 𝑇𝑟𝑎𝑚𝑝 estimation is an
Fig. 6: Impact of varying the voltage ramp time on peak flux and extension to that presented in [16]. The new framework is
magnetizing current (Ph. A of a Y-Y transformer with 𝜙𝑟 = 0.8 𝑝𝑢). tailored and extended to IBR applications with larger networks
 Transformer inrush current mitigation techniques for grid-forming inverters dominated grids

consideration. It aims to answer the questions of how fast a SE network operators have network models with the key required
ramp should be to limit the peak source current and power to parameters, this is combined with the emerging network
the rated values. Moreover, the framework aims to provide a measurements digitalization trend that further increases the
flexible approach that can impact sizing the inverter if it is yet available data reliability. That said, in cases where parameters
to be installed mainly for the network energization purposes. are partially missing (e.g., residual flux measurements), then
The methodology is tailored and extended to grid-forming worst-case assumptions can be used (i.e., to the higher 𝜙𝑟(𝑝𝑢)
inverter applications with larger networks consideration that end). The framework can be used with various power network
may consist of a single or multiple transformers with various simulation tools with automation such as MATLAB/Simulink
ratings. The proposed framework should mainly be applied or PSCAD/EMTDC in combination with Python. The idea here
before the network energization to provide a suitable ramp-rate is to iteratively run the simulation while varying 𝑇𝑟𝑎𝑚𝑝 until a
recommendation. stopping criteria is achieved.
The ramping time estimation approach presented here is The new framework flowchart is presented in Fig. 8, and is
iterative and model based in nature (i.e., it requires access to the configured to accommodate networks with 𝑛 transformers. To
energized transformer and network parameters). Knowledge of implement the proposed methodology, the network model is
the energizing network configuration and the energizing source first initialized, with all the CBs in the energized routes in their
control is recommended for improved estimation accuracy, on-state. The worst-case assumptions for missing parameters, if
such as the impedance between energizing source and any, are also made at this stage. The soft energization 𝑇𝑟𝑎𝑚𝑝 is
transformer terminals (e.g., CBs and cables impedance), set to a minimal value close to zero (e.g., 0.1 𝑠) to start the
transformer saturation characteristics and residual flux, as well simulation. The peak MVA and source currents (in the three-
as the used source control. This is because flux, being the phases) are recorded and compared to the defined stopping
voltage integral, is influenced by this control. The network criteria. If this criteria is not satisfied, then 𝑇𝑟𝑎𝑚𝑝 is increased
model availability requirement is, in principle, not a limiting
and the process is repeated. Naturally, the next simulation
factor in various cases since many distribution and transmission
should produce lower peak current and inverter MVA due to the
Model energizing increased ramp-time. The iterations are repeated until the
Start stopping criteria is met, and 𝑇𝑟𝑎𝑚𝑝 is set to the first value
network*
*Tx + Energizing Source + meeting this target.
Impedance/Cables + Control
In the new framework, a threshold current 𝑖𝑠𝑡𝑜𝑝 is defined to
balance between the inverter and the transformers. Equation
Txj j =1 (14) mathematically defines 𝑖𝑠𝑡𝑜𝑝 based on both equipment
limits: a) when the inverter rating is the energizing bottleneck
(smaller MVA than the smallest energized transformer), and b)
Yes
Φr Available? Use Φr value when smaller MVA transformers than the inverter rating exists
For n in the network. For instance, if a 10 MVA inverter energizes a
No Transformers string of transformers including a 7.5 MVA unit, then a
Assume worst case preferable stopping criteria should consider the transformer
j++
Φr combination rating as well as the inverter to maintain operation within rated
boundaries for both the IBR and the transformers, since the
No
Yes small transformer is the energization bottleneck here. This way,
Initiate Tramp j > n?
the ramping time stopping criteria (𝑖𝑠𝑡𝑜𝑝 ) ensures that the
inverter and transformer limits are respected. Though, if some
transformers are much smaller than the IBR rating (e.g., 10
Record: iinv(t), vinv(t), MVAinv(t),
Run the model MVA to 500 kW), then using them to define the stopping
max(iinv), max(MVAinv)
criteria could be impractical. In this case, the small units may
either be energized separately, or using a ramp-rate that ignores
Tramp Loop
their rating (if their protection can withstand short-term inrush).
Yes max(iinv) > istop &
Tramp = Tramp + Ts 𝐹𝑠 𝑖𝑖𝑛𝑣𝑚𝑎𝑥 ∶ 𝑀𝑉𝐴𝑖𝑛𝑣 ≤ min(𝑀𝑉𝐴 𝑇𝑥 )𝑗𝑗
MVAinv > MVArated? 𝑗
𝑖𝑠𝑡𝑜𝑝 = { (14)
𝐹𝑠 𝑖 𝑇𝑥𝑖 ∶ 𝑀𝑉𝐴𝑖𝑛𝑣 > min (𝑀𝑉𝐴 𝑇𝑥 )
No
Validate Waveforms where 𝑀𝑉𝐴𝑖𝑛𝑣 refers to the source inverter rating,
Record Tramp
& Protection Limits min(𝑀𝑉𝐴 𝑇𝑥 ) refers to the minimum MVA transformer rating
in the network, 𝑖𝑖𝑛𝑣𝑚𝑎𝑥 is the maximum inverter rated current,
𝑖 𝑇𝑥𝑖 is the rated current of the smallest transformer and 𝐹𝑠 is a
Sensitivity Yes Vary parameters
Analysis? and repeat process user-defined factor between 0 and 1 that sets the desired
No stopping-criteria safety ‘head-room’. The other stopping
criteria is related to the inverter MVA power rating that should
end also be satisfied. This is because in some cases, the current has
been observed to be below its stopping threshold while the peak
Fig. 8: Flowchart of the proposed 𝑇𝑟𝑎𝑚𝑝 estimation algorithm.
MVA is slightly higher than the rating due to the non-
 Transformer inrush current mitigation techniques for grid-forming inverters dominated grids

nverter m s n

rid

i inv
m s n

oads
rid

Fig. 9: Block diagram of the network used for the energization case study.

symmetrical nature of inrush conditions and its impact on transformers combination is used to showcase the energization
voltage. The stopping criteria can be defined with respect to of multiple transformer topologies. The cables are modeled as
flux or transformer magnetizing current if only a single 𝜋 − 𝑠𝑒𝑐𝑡𝑖𝑜𝑛𝑠 with the default MATLAB/Simulink library
transformer is energized. To maintain generality, the stopping parameters. The base case network parameters are detailed in
criteria is defined here with respect to the inverter current and Table I. Given the rising penetration of DERs in power
rated power due to: a) the limited IBRs overcurrent capability, networks, assessing the feasibility of using small IBR units in
and b) accommodate the possible existence of multiple energizing large networks is particularly important for network
transformers in the energized network. A power de-rating operators looking for available units for black-start. Hence, a 15
factor, similar to 𝐹𝑠 , can also be introduced for the inverter MVA unit is selected for the base case. The current threshold is
MVA rating if high uncertainty is associated with the based on this rating with (𝐹𝑠 = 0.95) from Equation 14 for SE
transformer or network measurements or to comply with investigations. This equates to 0.787√2 × 0.95 = 1.056 (𝑘𝐴)
network protection requirements in cases where long ramp- for a 15 MVA inverter at 11 kV.
durations that are recommended by the framework for a specific
MVA rating risk late fault detection outside grid-code limits. B. Energization Techniques Testing Methodology
Sensitivity analysis has also been incorporated into the Controlled switching in its 1-PL and 3-PL CB forms is tested
framework and can be optionally coded to run automated with the assumption that residual flux and prospective flux
testing for different combinations of uncertain parameters, such measurements are available for all transformers. The delay
as different network impedances. This helps in establishing effect between sending the CB closing command and the actual
ranges that can be correlated with field results. 𝑖𝑠𝑡𝑜𝑝 can also be contact closing instant is ignored and assumed to be properly
varied to consider different inverter ratings, taking into account compensated by the algorithm, a relevant compensation
the tradeoff between using small inverters and energizing large technique example is presented in [23]. The case study assumes
transformers, which may lead to higher 𝑇𝑟𝑎𝑚𝑝 when compared that 𝐶𝐵1 is 3-PL CB type, and that 𝐶𝐵2 & 𝐶𝐵3 are 1-PL type
to larger (but more costly) inverters. CBs. The selected three transformers provide a combination of
topologies that cover various scenarios to study their
IV. NETWORK ENERGIZATION CASE STUDY energization under different assumptions. In Fig. 9, the CS
sequence in this case study is such that the grid-forming inverter
A. Test System Description control first provides 1 pu voltage at 𝐶𝐵1 input, then the CB is
The methodology to test and benchmark the presented closed at its combined least flux error point from equation (7).
energization techniques in this paper is presented here. First, the Following the first transformer energization, 𝐶𝐵2 & 𝐶𝐵3 are
network to be energized is illustrated in Fig. 9 . The network is closed independently to draw least inrush current using separate
composed of the energizing IBR, transformers with various 1-PL closing algorithms that are adjusted to accommodate their
ratings and configurations, CBs, and cables. The presented individual configuration. The next black-start step is typically
connecting load blocks and synchronizing to the HV and MV
TABLE I: TEST NETWORK PARAMETERS (FOR THE BASE CASE) grids. A relevant synchronization technique is presented in [24].
Inverter parameters In the presented case study, SE is tested under the new defined
Rating (MVA), 𝑓 (𝐻𝑧) 15, 50 framework in Section III. CBs between the inverter and the
𝑍𝑓𝑖𝑙𝑡𝑒𝑟 (𝑝𝑢) 0.075
energized transformers are initially closed, and the energization
𝑍𝑓𝑖𝑙𝑡𝑒𝑟 𝑋/𝑅 ratio 12.5
is then simulated while increasing 𝑇𝑟𝑎𝑚𝑝 after each simulation
MATLAB/Simulink Transformer pu parameters
𝑅𝑝,𝑠 (𝑝𝑢) 0.002
𝐿𝑝,𝑠 (𝑝𝑢), 𝐿0 (𝑝𝑢) 0.08, 0.5 TABLE II: NETWORK ENERGIZATION CASE STUDY – COVERED SCENARIOS
𝑅𝑚 (𝑝𝑢) 500 Technique Case # Parameters Variation from Base Case
𝑇𝑥1 𝜙𝑟(𝑎,𝑏,𝑐) (𝑝𝑢) 0.8, 0, -0.8
Controlled Case 1 Base Case
𝑇𝑥2 𝜙𝑟(𝑎,𝑏,𝑐) (𝑝𝑢) -0.4, 0.7, -0.3
Switching Case 2 𝜙𝑟 (𝑇𝑥1 ) = [0, 0, 0]
𝑇𝑥3 𝜙𝑟(𝑎,𝑏,𝑐) (𝑝𝑢) 0, 0, 0
Transformers saturation curve Case 1 Base Case
𝑖𝑚 (𝑝𝑢) [0, 0, 0.0024, 1] 𝑖 = [0, 0, 0.0024, 3]
Soft Case 2 Saturation curve: 𝑚
𝜙 (𝑝𝑢) [0, 0.85, 1.2, 1.52] 𝜙 = [0, 0.85, 1.1, 1.52]
Energization
Cable Parameters (per km) 𝜙𝑟 (𝑇𝑥1 , 𝑇𝑥2 , 𝑇𝑥3 ) = [0, 0, 0]
Case 3
(𝑅, 𝐿, 𝐶)𝑐𝑎𝑏𝑙𝑒 12.73 𝑚Ω, 0.93 mH, 12.74 nF (Demagnetized cores)
 Transformer inrush current mitigation techniques for grid-forming inverters dominated grids

until the stopping criteria is satisfied. The simulations continue independently. 𝑇𝑥1 energization is initiated at 𝑡 = 0.2 𝑠, and the
here after the stopping criteria is satisfied for improved results CB is closed at 𝛼𝐶𝐵 = 210° based on equation (7). Given this
visualization and trends identification. The total number of 𝜙𝑟 combination, the inrush current is nearly zero and the
simulations per SE scenario is 34, with 𝑇𝑟𝑎𝑚𝑝 ∈ [0.2, 30] 𝑠. energization occurs smoothly despite the simultaneous phases
closing. The two cable segments (30 km and 20 km) are
C. Test Scenarios Definition energized simultaneously with 𝑇𝑥1 . A settling period of 10
This section presents the network energization case study cycles is introduced before energizing 𝑇𝑥2 and 𝑇𝑥3 to mitigate
through multiple scenarios covering CS and SE. The scenarios any transient impact, which may be modified as needed. Then
(and their sensitivity variations) are summarized in Table II. the two remaining transformers are energized. 𝑇𝑥2 is energized
The CS sensitivity case only considers varying the residual flux from the 𝑌 side, and 𝑇𝑥3 from the ∆ side. The CBs are closed
of the first transformer (𝑇𝑥 1 ), since it is energized with a 3-PL independently at instances when the flux error is around zero,
CB. Varying 𝜙𝑟 of the other transformers does not have a producing near-zero magnetizing currents. A delay of 0.02 s is
noticeable impact since their 1-PL CB are ideally able to track added between the 1-PL CBs activation for improved
and largely mitigate the inrush impact. Whereas two sensitivity visualization. Transformers flux and magnetizing current
cases are considered for SE: a) varying the core characteristics results at the energization moments for all transformers are
of all transformers, and b) demagnetizing all transformers by summarized in Fig. 10, demonstrating negligible inrush current.
setting 𝜙𝑟 to zero across all phases. The first sensitivity case The energizing inverter measurements are presented in Fig. 11,
aims to demonstrate the impact of multiple transformer core showcasing its instantaneous output MVA and current
types with different saturation-curve knee-voltage and air-core throughout the energization process. The spikes at 𝑇𝑥1
inductance, and the second case is considered to demonstrate energization instant are due to the cables energization, since
the paramount impact of residual flux on 𝑇𝑟𝑎𝑚𝑝 selection. cables can develop their own transient inrush current [25]. In
this case study, the cables inrush is within the base inverter
D. Controlled Switching Results rating (below 15 MVA) and peak per-phase current (1.11 kA).
The results for CS testing are presented here, starting with the If the resulting current exceeds inverter rating, then SE with
base network case. 𝑇𝑥1 is connected to a 3-PL CB with 𝜙𝑟 = sufficient 𝑇𝑟𝑎𝑚𝑝 can be used for network energization to
[0.8, 0, −0.8] 𝑝𝑢, while 𝑇𝑥2 and 𝑇𝑥3 phases are energized mitigate this issue.

Fig. 10: CS base case: flux & inrush currents at energization points. Fig. 12: CS sensitivity case: flux & inrush currents at energization points.

Fig. 11: CS base case: a) IBR peak MVA, b) instantaneous current. Fig. 13: CS sensitivity case: a) IBR peak MVA, b) instantaneous current.
 Transformer inrush current mitigation techniques for grid-forming inverters dominated grids

Controlled Switching Sensitivity Analysis

The sensitivity case presented here tests the network
energization performance under a different 𝜙𝑟 combination for
𝑇𝑥1 . The chosen combination [0, 0, 0] generates high minimum
common 𝜀𝜙 and inrush current across the full 360° range is
illustrated in Fig. 5. 𝑇𝑥1 is observed here to generate very high
inrush current (~ 1 pu) due to the simultaneous phases closing
at a high error point for phase A and phase C, as summarized in
Fig. 12. The other two transformers exhibit similar behavior to
the base case. Overall, the required inrush current from the
inverter peaks at 56.8 MVA, severely exceeding the chosen
base rating of 15 MVA. The inverter MVA and current output
for this scenario are illustrated in Fig. 13. From the presented
CS analysis, 𝐶𝐵1 is the network energization bottleneck. Had it
been composed of three 1-PL CBs with accurate residual flux
measurements, then 𝑇𝑥1 would have followed a similar
behavior as the other two transformers with near-zero inrush
Fig. 14: SE base case: a) peak inverter MVA; b) peak inverter current.
current, independent of the 𝜙𝑟 combination. Different IBR sizes are also demonstrated for comparison.

E. Soft Energization Results voltage from 1.2 pu to 1.1 pu, and increased peak inrush from
SE provides more flexibility in reducing the peak inverter 1 pu to 3 pu when the flux is 1.52 pu (i.e., decreased air-core
current demand during network energization by selecting an inductance). Reducing knee-voltage means that the saturation
appropriate ramping time. The new 𝑇𝑟𝑎𝑚𝑝 estimation region is reached faster, while decreasing the air-core
inductance leads to higher inrush current per flux increment
methodology in Fig. 8 is used here. All CBs are initially closed
within the saturation region. The modified ramping time
and then the GFC ramping reference is applied. From Fig. 14,
estimation algorithm is re-applied to the network model with
the inverter output MVA and current are significantly high for
the new transformers saturation curve, and the results are
very low 𝑇𝑟𝑎𝑚𝑝 values, as expected. 𝑇𝑟𝑎𝑚𝑝 is increased
summarized in Fig. 16. Longer ramps are required for 10, 15
gradually and the process is repeated while recording the and 20 MVA inverters in the sensitivity case to avoid violating
absolute peak inverter MVA and current values. The decreasing the IBR ratings. The minimum ramp-time for 20 MVA inverter
trend continues, and the stopping criteria defined by 𝑖𝑠𝑡𝑜𝑝 = increased from 3 s in the base case to 8 s here, and from 8 s to
1.056 𝑘𝐴 combined with the peak power < 15 MVA is reached 12 s for the 15 MVA inverter. Lastly, the 10 MVA inverter
around 𝑇𝑟𝑎𝑚𝑝 = 8 𝑠. Fig. 14 also demonstrates the stopping requires a 22 seconds ramp as opposed to the 15 s for the base
criteria if 10 MVA or 20 MVA inverters are used instead of the case. the steeper saturation curve impact is more evident at
adopted 15 MVA inverter. Increasing the rating by 5 MVA can lower 𝑇𝑟𝑎𝑚𝑝 values. Comparatively, the peak inverter power at
accelerate the ramp to 3 seconds without violating the inverter 𝑇𝑟𝑎𝑚𝑝 = 0.2 𝑠 is 55.6 MVA compared to 31.7 MVA in the base
rating, whereas decreasing the inverter size to 10 MVA requires case (75% increment), whereas at 𝑇𝑟𝑎𝑚𝑝 = 26 𝑠, the difference
using a slower ramp of at least 15 seconds to avoid violating the
in peak inverter power diminishes to 18%, between 8.9 MVA
unit rating. This example shows the influence of inverter rating
for the sensitivity, and 7.5 MVA for the base case. That is, both
on the ramp-rate selection. The inverter status (existing or to be
cases are observed to approach each other as 𝑇𝑟𝑎𝑚𝑝 increases.
installed) and the desired use applications may thus be
combined with this framework to set an appropriate size under The final sensitivity case for soft transformer energization
worst-case scenario assumptions for cost optimization. investigates the impact of energizing the studied network when
Looking back at the base 15 MVA inverter rating, time-domain
simulation results are presented for 𝑇𝑟𝑎𝑚𝑝 = 8 𝑠. Around this
time, the peak inverter power is 14.5 MVA, just below the
defined base inverter rating. Time simulation results for this
scenario are summarized in Fig. 15, illustrating the ramping and
decaying behavior for both apparent power and the inverter
current. Since multiple transformers with different 𝜙𝑟 are
energized, their cores approach saturation region at different
times per phase and per transformer. The inverter MVA and
current reach their peak at 𝑡 = 𝑇𝑟𝑎𝑚𝑝 , and then continue
decaying after the ramp is concluded since the flux transient
from this point is primarily influenced by the damping factor.
Soft Energization Sensitivity Analysis
The first sensitivity scenario for SE here considers varying the
Fig. 15: SE base case instantaneous results with 𝑇𝑟𝑎𝑚𝑝 = 8 𝑠 for: a) IBR
core saturation curve of the network transformers as in Table II. power in MVA, b) IBR current for a 15 MVA inverter with 𝐹𝑠 = 0.95.
The sensitivity saturation curve is steeper, with a reduced knee-
 Transformer inrush current mitigation techniques for grid-forming inverters dominated grids

industrial project in the UK aims to assess the feasibility of

using distributed resources (including IBRs) for black-start.
Some distribution areas targeted by this project in the UK are
observed to have a concentration of 3-PL CBs with limited
PoW relays availability. Thus, the investigation covering CS
here aims to answer the question of the 3-PL technique
effectiveness against classical CS for similar networks, and to
extend the analysis into SE as a potential alternative due to the
highlighted IBRs voltage control flexibility. In light of this, key
findings and practical/industry implications based on the results
reported in this paper are summarized below.
- Classical CS with 1-PL CBs provides superior results to
energization from controlled 3-PL CBs. Proper delay
compensations for CB closing should be considered for
effective CS operation.
- 3-PL CBs can still perform well under a range of 𝜙𝑟
combinations (especially at high 𝜙𝑟 of [𝑟, 0, −𝑟] form such
Fig. 16: SE sensitivity case: a) peak inverter MVA; b) peak inverter
current. Different IBR sizes are also demonstrated for comparison.
as [0.8, 0. -0.8] pu), but can still lead to high inrush currents
under other combinations (e.g., demagnetized transformers
or with low residual fluxes).
- Pre-assessments using (7) and (8) are thus recommended
prior to 3-PL CS implementation.
- CS techniques require 𝜙𝑟 measurements, which may not be
accurately available at all energization points. Industrial
relays, such as SynchroTeq [14] may be used (added cost).
- SE can be used as an alternative technique that, in principle,
does not require additional installations or measurements.
- SE can also avoid inrush currents resulting from cables
energization without requiring additional hardware.
- The presented new framework for 𝑇𝑟𝑎𝑚𝑝 estimation
showcases a model-based methodology that can be adopted
using industrial software tools such as PSCAD/EMTDC.
- While 𝜙𝑟 availability helps achieving better 𝑇𝑟𝑎𝑚𝑝 estimates,
Fig. 17: SE 2nd sensitivity case: instantaneous results for: a) IBR power in it is not a necessity for the new framework implementation
MVA, b) IBR current when the network transformers are demagnetized.
since worst-case assumptions for 𝜙𝑟 can be made.
- A first worst-case residual flux approximation can be
all transformers are demagnetized (i.e., with zero residual flux). obtained by simultaneously assuming high 𝜙𝑟 values across
Since a transformer inrush current is a consequence of the flux all network transformers. A confirmatory grid-search may
reaching saturation region, having demagnetized cores means also be performed across the transformers, by testing high 𝜙𝑟
that the flux must first ramp up from zero to knee-point before combinations against a single 𝑇𝑟𝑎𝑚𝑝 value and choosing the
causing a noticeable inrush current. This provides a significant
aggregate combination with highest inrush current as an input
headroom that allows for faster ramps, as opposed to the base
to the ramp-rate estimation framework.
case where 𝑇𝑥1 flux in phase A, for instance, is already at 0.8
- SE can be used to accommodate the energization process to
pu. Meaning that the headroom for the base case is very limited
low IBR ratings through 𝑇𝑟𝑎𝑚𝑝 manipulation (Fig. 14), which
before saturation (between 0.8 pu and 1.2 pu, spanning 0.4 pu),
as compared to a 1.2 pu span for the demagnetized core. In this provides an economic alternative to the inverters de-rating
scenario, 𝑇𝑟𝑎𝑚𝑝 is set to 0.2 s only, which is sufficient to option, especially when existing IBRs are used.
- That said, high 𝑇𝑟𝑎𝑚𝑝 values should be inspected carefully
suppress energizing inrush current in all three transformers as
illustrated in Fig. 17. The peak inverter power in this scenario for the target network and its protection settings/limits to
is around 0.4 MVA, with a peak current of 29 A. The cables in minimize faults non-detection possibility during the ramp
this scenario are also assumed to be initially discharged. This when voltage level is below detection thresholds.
scenario clearly demonstrates the impact of residual flux on SE, - An interesting shown contrast is that soft energization of
and the positive impact of applying the SE technique on demagnetized transformers can be smoothly achieved with
demagnetized, or nearly demagnetized (low 𝜙𝑟 ), transformers. very fast ramps, whereas high inrush current is obtained when
applying 3PL-CS technique with 𝜙𝑟 = [0,0,0] 𝑝𝑢.
V. RESULTS PRACTICAL IMPLICATIONS Finally, the results presented in this paper consider a generic
This paper addresses black-start and networks energization single-loop GFC implementation. Preliminary simulations were
provision from IBRs, as part of the rising industrial interest in carried out by the authors in MATLAB/Simulink and
similar investigations. For instance, the Distributed ReStart RSCAD/RTDS software tools, incorporating inner voltage and
 Transformer inrush current mitigation techniques for grid-forming inverters dominated grids

current loops to the inverter control. Results show that for SE, [8] J. Mitra, X. Xu, and M. Benidris, "Reduction of Three-Phase Transformer
Inrush Currents Using Controlled Switching," IEEE Transactions on
the stopping criteria with inner loops can be achieved with
Industry Applications, vol. 56, no. 1, pp. 890-897, 2020.
shorter ramp durations due to the imposed current-limiting [9] H. Ni, S. Fang, and H. Lin, "A Simplified Phase-Controlled Switching
control dynamics. The results presented recently in [6] support Strategy for Inrush Current Reduction," IEEE Transactions on Power
this initial observation, and are recommended for further Delivery, vol. 36, no. 1, pp. 215-222, 2021.
detailed investigation. [10] A. Roscoe et al., "Practical Experience of Providing Enhanced Grid
Forming Services from an Onshore Wind Park", 18th Wind Integration
Workshop, DUblin, 2019.
VI. CONCLUSIONS [11] J. Glassmire, S. Cherevatskiy, G. Antonova, and A. Fretwell, "Using
A comprehensive approach has been presented in this paper to Virtual Synchronous Generators to Resolve Microgrid Protection
assess key transformer energization techniques in IBR Challenges," in 2021 74th Conference for Protective Relay Engineers
(CPRE), 22-25 March 2021 2021, pp. 1-7.
dominated power networks, in addition to identifying suitable [12] "Switchsync PWC600 Point-on-Wave Controller - Product Report,"
conditions for their application. The paper also proposed a new Hitachi Energy, 2021.
framework for soft energization. Given the large number of 3- [13] R. Cano-González, A. Bachiller-Soler, J. Rosendo-Macías, G. Álvarez-
PL CBs in many IBR-dominated distribution networks, the use Cordero, "Optimal gang-operated switching for transformer inrush
of controlled switching through 3-PL CBs is analyzed and current reduction," Electr. Power Syst. Res., vol. 131, pp. 80-86, 2016.
[14] T. Rastall, "SynchroTeq - 3-Pole Transformer Switching," Enspec, United
shown to be effective over a limited range of residual flux Kingdom, 2020.
combinations. In contrast, energization from 1-PL CBs can be [15] Y. Pan, X. Yin, Z. Zhang, B. Liu, M. Wang, and X. Yin, "Three-Phase
effective across the full range of residual flux conditions. Transformer Inrush Current Reduction Strategy Based on Prefluxing and
Though, the application of CS requires accurate 𝜙𝑟 Controlled Switching," IEEE Access, vol. 9, pp. 38961-38978, 2021.
measurements and PoW relays, which may not always be [16] A. Alassi, K. Ahmed, A. Egea-Alvarez, and C. Foote, "Soft Transformer
Energization: Ramping Time Estimation Method for Inrush Current
readily available. The new framework for estimating the soft Mitigation," in 2021 UPEC Conference, 31 Aug.-3 Sept. 2021, pp. 1-6.
energization ramp time, 𝑇𝑟𝑎𝑚𝑝 , considers inverter current [17] S. Jazebi, F. d. León, and N. Wu, "Enhanced Analytical Method for the
protection aspects and the simultaneous energization of Calculation of the Maximum Inrush Currents of Single-Phase Power
multiple transformers. Measurements availability increases the Transformers," IEEE Transactions on Power Delivery, vol. 30, no. 6, pp.
2590-2599, 2015.
new framework estimation accuracy, but unlike CS, they are not [18] D. Bejmert, M. Kereit, and K. Boehme, "Controlled energization
essential for its application. A detailed case study using a procedures of power transformers," International Journal of Electrical
network with multiple transformers is presented to benchmark Power & Energy Systems, vol. 135, p. 107555, 2022.
the performance of the different energization techniques. [19] J. H. Brunke and K. J. Frohlich, "Elimination of transformer inrush
currents by controlled switching. I. Theoretical considerations," IEEE
Controlled switching can locally eliminate transformers inrush
Transactions on Power Delivery, vol. 16, no. 2, pp. 276-280, 2001.
current, but the inverter can still be prone to inrush currents [20] J. H. Brunke and K. J. Frohlich, "Elimination of transformer inrush
from long cables if no additional measures such as installing currents by controlled switching. II. Application and performance
more PoW CBs are considered. With a sufficient 𝑇𝑟𝑎𝑚𝑝 considerations," IEEE Transactions on Power Delivery, vol. 16, no. 2, pp.
281-285, 2001.
duration, soft energization can prevent the cables energization
[21] U. Parikh and B. R. Bhalja, "Mitigation of magnetic inrush current during
inrush with no additional required hardware and to mitigate controlled energization of coupled un-loaded power transformers in
transformers inrush effectively. That said, setting very high presence of residual flux without load side voltage measurements,"
ramp-rates may also lead to delays in fault detection until a International Journal of Electrical Power & Energy Systems, vol. 76, pp.
sufficient source voltage is present. 156-164, 2016.
[22] "Distributed ReStart: Power Engineering and Trials - Demonstration of
Black Start from DERs Part 1," National Grid ESO, UK, 2021.
REFERENCES [23] G. Benmouyal, N. Fischer, D. Taylor, M. Talbott-Williams, and R.
Chowdhury, "A Unified Approach to Controlled Switching of Power
[1] J. E. Holcomb, "Distribution Transformer Magnetizing Inrush Current," Equipment," in 72nd Annual Georgia Tech Protective Relaying
Transactions of the American Institute of Electrical Engineers. Part III: Conference, USA, 2018.
Power Apparatus and Systems, vol. 80, no. 3, pp. 697-702, 1961. [24] A. Alassi, K. Ahmed, A. Egea-Alvarez, and C. Foote, "Modified Grid-
[2] P. Pachore, Y. Gupta, S. Anand, S. Sarkar, P. Mathur, and P. K. Singh, forming Converter Control for Black-Start and Grid-Synchronization
"Flux Error Function Based Controlled Switching Method for Minimizing Applications," in 2021 UPEC Conference, 31 Aug.-3 Sept. 2021, pp. 1-5.
Inrush Current in 3-Phase Transformer," IEEE Transactions on Power [25] I. Lafaia, M. T. C. de Barros, J. Mahseredjian, A. Ametani, I. Kocar, and
Delivery, 2020. Y. Fillion, "Surge and energization tests and modeling on a 225kV HVAC
[3] R. Ierna et al., "Dispatching parameters, strategies and associated cable," Electric Power Systems Research, vol. 160, pp. 273-281, 2018.
algorithm for VSM (virtual synchronous machines) and HGFC (hybrid
grid forming convertors)", 19th Wind Integration workshop, Dublin, 2019.
[4] A. Jain, O. Saborío-Romano, J. Sakamuri, and N. Cutululis, "Blackstart
from HVDC-connected offshore wind: Hard versus soft energization,"
IET Renewable Power Generation, vol. 15, no. 1, pp. 127-138, 2021.
[5] A. Jain, O. Saborio-Romano, J. Sakamuri, and N. Cutululis, "Virtual
Resistance Control for Sequential Green-start of Offshore Wind Power
Plants," IEEE Transactions on Sustainable Energy, 2022.
[6] M. Shahparasti, H. Laaksonen, K. Kauhaniemi, P. Lauttamus, S.
Strandberg, and J. Strandberg, "Inrush Current Management During
Medium Voltage Microgrid Black Start with Battery Energy Storage
System," IEEE Access, 2022.
[7] E. Bikdeli, M. R. Islam, M. M. Rahman, and K. M. Muttaqi, "State of the
Art of the Techniques for Grid Forming Inverters to Solve the Challenges
of Renewable Rich Power Grids," Energies, vol. 15, no. 5, 2022.
 Transformer inrush current mitigation techniques for grid-forming inverters dominated grids

Abdulrahman Alassi (Member, IEEE) Colin Foote (Member, IEEE) received

received his B.Sc. and M.Sc. in Electrical MEng and PhD degrees in electrical
Engineering from Qatar University in 2014 engineering from the University of
and 2016, respectively. He is currently Strathclyde, UK, in 1999 and 2006
working towards his Ph.D. degree in respectively. He is currently a Senior
Electronic and Electrical Engineering from Simulation Engineer at the National HVDC
the University of Strathclyde. He Centre, UK. Previous roles include System
previously worked as a researcher at Qatar Analysis manager at network operator SP
University between 2014 and 2017. He Energy Networks and Head of Power
then joined Iberdrola’s innovation center in Qatar where he Systems at solutions provider Smarter Grid Solutions. He also
currently leads the energy systems group. He has published worked at Mott MacDonald as a power systems analyst and at
more than 20 articles in technical journals and international the University of Strathclyde as a research fellow. He has
conferences and received the best paper award in the 2019 IEEE extensive and varied experience in consultancy, research,
SGRE Conference. His research interests include converters management, project design and delivery, solutions
control, power system studies, Power Hardware-in-the-Loop development, and training. Research interests focus on power
(PHiL) validation, and the techno-economic optimization of system modeling, simulation and analysis, transmission and
renewables and storage integration. distribution network planning, HVDC, distributed generation,
and wind power.

Khaled H. Ahmed (M’09-SM’12)

received the B.Sc. (Hons.) and M.Sc.
degrees from Alexandria University,
Egypt, in 2002 and 2004, respectively, and
the Ph.D. degree in power electronics
applications from the University of
Strathclyde, U.K., in 2008. He was
appointed as a Professor at Alexandria
University, in 2019. He is currently a
Reader in power electronics with the University of Strathclyde.
He has published more than 160 technical articles in refereed
journals and conferences as well as a published textbook titled
High Voltage Direct Current Transmission: Converters,
Systems and DC Grids, a book chapter contribution, and holds
a PCT patent PCT/GB2017/051364. His research interests
include renewable energy integration, high power converters,
offshore wind energy, dc/dc converters, HVDC, and smart
grids. He is a Senior Member of the IEEE Power Electronics
Society and the IEEE Industrial Electronics Society.

Agusti Egea-Alvarez (Member, IEEE)

received the B.Sc., M.Sc., and Ph.D.
degrees from the Technical University of
Catalonia, Barcelona, in 2008, 2010, and
2014, respectively. In 2015, he was Marie
Curie Fellow with China Electric Power
Research Institute (CEPRI). In 2016, he
joined Siemens Gamesa as a Converter
Control Engineer, working on grid forming
controllers and alternative HVDC schemes for offshore wind
farms. He has been a member of the Power Electronics, Drives
and Energy Conversion (PEDEC) Group, since 2018. He is
currently a Senior Lecturer with the Department of Electronic
and Electrical Engineering at the University of Strathclyde. His
current research interests include the control and operation of
high-voltage direct current systems, renewable generation
systems, electrical machines, and power converter control.